American Journal of Kidney Diseases (AJKD)

American Journal Of Kidney Diseases (AJKD)

September 2005 (Vol. 46, Issue 3)

Review Malnutrition in Hemodialysis Patients: What Therapy?, 29 July 2005 Bossola M, Muscaritoli M, Tazza L, Giungi S, Tortorelli A, Fanelli FR, Luciani G pages 371-386

Review Multinutrient Oral Supplements and Tube Feeding in Maintenance Dialysis: A Systematic

Review and Meta-Analysis, 1 August 2005 Stratton RJ, Bircher G, Fouque D, Stenvinkel P, Mutsert Rd, Engfer M, Elia M pages 387-405

Original investigation CKD Progression and Mortality Among Older Patients With Diabetes, 22 July 2005 Patel UD, Young EW, Ojo AO, Hayward RA pages 406-414

Original investigation

Earlier Diagnosis of Autosomal Dominant Polycystic Kidney Disease: Importance of Family History and Implications for Cardiovascular and Renal Complications, 22 July 2005 Taylor M, Johnson AM, Tison M, Fain P, Schrier RW pages 415-423

Original investigation Proteinuria and Hemoglobin Levels in Patients With Primary Glomerular Disease, 29 July 2005 Mähr N, Neyer U, Prischl F, Kramar R, Mayer G, Kronenberg F, Lhotta K pages 424-431

Original investigation Host and Bacterial Virulence Factors Predisposing to Emphysematous Pyelonephritis, 14 July 2005 Tseng CC, Wu JJ, Wang MC, Hor LI, Ko YH, Huang JJ pages 432-439

Original investigation Absorptive Hyperoxaluria Leads to an Increased Risk for Urolithiasis or Nephrocalcinosis

in Cystic Fibrosis, 29 July 2005 Hoppe B, von Unruh GE, Blank G, Rietschel E, Sidhu H, Laube N, Hesse A pages 440-445

Original investigation The Course of Anticardiolipin Antibody Levels Under Immunoadsorption Therapy,

8 August 2005 Hauser AC, Hauser L, Pabinger-Fasching I, Quehenberger P, Derfler K, Hörl WH pages 446-454

Relationship of Phosphorus and Calcium-Phosphorus Product With Mortality in CKD,

29 July 2005 Menon V, Greene T, Pereira AA, Wang X, Beck GJ, Kusek JW, Collins AJ, Levey AS,

Sarnak MJ pages 455-463

Original investigation Changes in Cardiovascular Calcification After Parathyroidectomy in Patients With ESRD,

14 July 2005 Bleyer AJ, Burkart J, Piazza M, Russell G, Rohr M, Carr J J pages 464-469

The Hemodynamic Effect of Calcium Ion Concentration in the Infusate During

Predilution Hemofiltration in Chronic Renal Failure, 18 July 2005 Karamperis N, Sloth E, Jensen JD pages 470-480

Original investigation Epoetin Alfa Use in Patients With ESRD: An Analysis of Recent US Prescribing

Patterns and Hemoglobin Outcomes, 14 July 2005 Collins AJ, Brenner RM, Ofman JJ, Chi EM, Stuccio-White N, Krishnan M, Solid C,

Ofsthun NJ, Lazarus J M pages 481-488

Original investigation Association of Morbid Obesity and Weight Change Over Time With Cardiovascular

Survival in Hemodialysis Population, 18 July 2005 Kalantar-Zadeh K, Kopple JD, Kilpatrick RD, McAllister CJ, Shinaberger CS,

Gjertson DW, Greenland S pages 489-500

Original investigation Tunneled Catheters in Hemodialysis Patients: Reasons and Subsequent Outcomes,

22 July 2005 Lee T, Barker J, Allon M pages 501-508

Validation of a Method to Predict Required Dialysis Time for Cases of Methanol and

Ethylene Glycol Poisoning, 18 July 2005 Youssef GM, Hirsch DJ pages 509-511

Original investigation Time Profiles of Peritoneal and Renal Clearances of Different Uremic Solutes in Incident

Peritoneal Dialysis Patients, 18 July 2005 Bammens B, Evenepoel P, Verbeke K, Vanrenterghem Y pages 512-519

History of Cardiovascular Disease Is Associated With Endothelial Progenitor Cells in

Peritoneal Dialysis Patients, 22 July 2005 Steiner S, Schaller G, Puttinger H, Födinger M, Kopp CW, Seidinger D, Grisar J, Hörl

WH, Minar E, Vychytil A, Wolzt M, Sunder-Plassmann G pages 520-528

Original investigation Cardiovascular Risk and Renal Transplantation: Post Hoc Analyses of the Assessment of Lescol in Renal Transplantation (ALERT) Study, 18 July 2005 Jardine AG, Fellström B, Logan JO, Cole E, Nyberg G, Grönhagen-Riska C, Madsen S,

Neumayer HH, Maes B, Ambühl P, Olsson AG, Pedersen T, Holdaas H pages 529-536

The Role of Pretransplantation Renal Replacement Therapy Modality in Kidney

Allograft and Recipient Survival, 14 July 2005 Goldfarb-Rumyantzev AS, Hurdle JF, Scandling JD, Baird BC, Cheung AK pages 537-549

Case report Nephrotic Syndrome and Renal Failure After Allogeneic Stem Cell Transplantation:

Novel Molecular Diagnostic Tools for a Challenging Differential Diagnosis, 22 July 2005 Romagnani P, Lazzeri E, Mazzinghi B, Lasagni L, Guidi S, Bosi A, Cirami C, Salvadori M pages 550-556

Editorial Hypertension in Individuals at Risk for Autosomal Dominant Polycystic Kidney Disease:

To Screen or Not to Screen?, 1 August 2005 Perrone RD, Miskulin DC pages 557-559

Tubulointerstitial Diseases, 29 July 2005 Braden GL, O’Shea MH, Mulhern JG pages 560-572

Kidney biopsy teaching case Membranoproliferative Injury Pattern in a Renal Allograft, 29 July 2005 Gökden N, Rossini M, Kumar J, Fogo AB pages 573-576

AJKD Electronic Page

Quiz Page Answers September 2005, 8 August 2005 pages e31-e33

AJKD Electronic Page Tethered Hemodialysis Catheter With Retained Portions in Central Vein and

Right Atrium on Attempted Removal, 29 July 2005 Thein H, Ratanjee SK pages e35-e39

Tubulointerstitial Nephritis and Fanconi Syndrome in Primary Biliary Cirrhosis,

29 July 2005 Lino M, Binaut R, Noël LH, Patey N, Rustin P, Daniel L, Serpaggi J, Varaut A,

Vanhille P, Knebelmann B, Grünfeld JP, Fakhouri F pages e41-e46

Masthead page A2

Editorial Board page A8

Contents pages A13-A16

October Highlights 1 page A24

October Highlights 2 page A32

October Highlights 3 page A37

October Highlights 4 page A43

Information for Contributors pages A49-A52

Announcements page A55

AJKD Quiz Page

Quiz Page September 2005, 8 August 2005 page A61

The Official Journal of the National Kidney Foundation VOL 46, NO 3, SEPTEMBER 2005

American Journal of AJKD Kidney Diseases REVIEWS Malnutrition in Hemodialysis Patients: What Therapy?

Maurizio Bossola, MD, Maurizio Muscaritoli, MD, Luigi Tazza, MD, Stefania Giungi, MD, Antonio Tortorelli, MD, Filippo Rossi Fanelli, MD, and Giovanna Luciani, MD

● Malnutrition is common in hemodialysis patients and is a powerful predictor of morbidity and mortality. Although much progress has been made in recent years in identifying the causes and pathogenesis of malnutrition in hemodialysis patients, as well as recognizing the link between malnutrition and morbidity and mortality, no consensus has been reached concerning its management. Along with such conventional interventions as nutri- tional counseling, oral nutritional supplements, and intradialytic parenteral nutrition, novel preventive and therapeu- tic strategies have been tested, such as appetite stimulants, growth hormone, androgenic anabolic steroids, and anti-inflammatory drugs, with contradictory and nonconclusive results. Malnutrition still remains a great challenge for nephrologists in the third millennium. Am J Kidney Dis 46:371-386. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Hemodialysis (HD); malnutrition; prevention; therapy.

ROTEIN-ENERGY malnutrition (PEM) is ing hemodialytic treatment; the prevalence var- P common in patients with end-stage renal ies between 23% and 73%.2-4 disease (ESRD) on maintenance hemodialysis (HD) therapy and is a powerful predictor of PATHOGENESIS morbidity and mortality.1,2 Although much The pathogenesis of PEM in HD patients is progress has been made in recent years in identi- multifactorial. fying the causes and pathogenesis of malnutri- tion in HD patients, as well as in recognizing the Predialysis Restrictive Diets link between malnutrition, inflammation, and Before starting maintenance dialysis therapy, mortality, no consensus has been reached yet patients with ESRD often are prescribed low- concerning the therapeutic strategies aimed at improving nutritional status and nutrition-related clinical outcome in such patients.1,2 From the Istituto di Clinica Chirurgica, Università Cat- The aim of this review is to briefly define the tolica del Sacro Cuore; and the Dipartimento di Medicina prevalence, causes, and effects of PEM in pa- Clinica, Università La Sapienza, Rome, Italy. Received January 13, 2005; accepted in revised form May tients on maintenance HD therapy and discuss in 31, 2005. depth its prevention and treatment. Originally published online as doi:10.1053/j.ajkd.2005.05.031 on August 1, 2005. PREVALENCE M.B. and M.M. contributed equally to the writing and editing of this review. The prevalence of PEM in predialysis patients Address reprint requests to Maurizio Bossola, MD, Isti- ranges from 20% to 80%, depending on the tuto di Clinica Chirurgica, Università Cattolica del Sacro choice of nutritional markers and the population Cuore, Largo A. Gemelli, 8, 00168 Roma, Italia. E-mail: 2-4 [email protected] studied. Although nutritional status tends to © 2005 by the National Kidney Foundation, Inc. improve after initiation of maintenance HD 0272-6386/05/4603-0001$30.00/0 therapy, PEM remains common in patients receiv- doi:10.1053/j.ajkd.2005.05.031

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 371-386 371 372 BOSSOLA ET AL protein and low-phosphate diets. These diets leading to a net amino acid loss of 10 to 13 may be hypocaloric1,2 and therefore may be g/dialysis treatment. In fasting patients using deleterious to the patient’s nutritional status if high-flux dialyzers, approximately8goffree not adequately monitored by skilled dieticians. amino acids are removed during a routine HD Recent evidence suggests that PEM often begins treatment.14 During HD with glucose-free dialy- incipiently when glomerular filtration rate is ap- sate, an average of 20 to 30 g of glucose is lost proximately 28 to 35 mL/min/1.73 m2 (0.47 to into the dialysate.15,16 0.58 mL/s) or even greater and continues to decrease gradually as glomerular filtration rate Alterations in Protein Metabolism decreases to less than these values.5 Thus, it is It is a common belief that both uremia and common that patients with ESRD arrive to HD dialysis are associated with increased protein therapy in poor nutritional condition.1,2 catabolism as a consequence of metabolic acido- sis and exposure of blood to the HD membrane, Inadequate Nutritional Intake with a consequent release of proinflammatory Both protein and energy requirements are cytokines.1,2,17 However, nitrogen balance stud- greater in HD patients than in healthy persons1,2 ies and whole-body amino acid turnover kinetics because of dialytic losses of amino acids, meta- studies indicate that uremia-related protein ca- bolic acidosis, and comorbid conditions.6 Thus, tabolism is not so marked.17-24 Moreover, stud- Dialysis Outcomes Quality Initiative guidelines ies that investigated the involvement of proteo- for nutrition have proposed a daily energy intake lytic pathways (calcium-activated proteases, of 35 kcal/kg of body weight for adults younger adenosine triphosphate–dependent ubiquitin- than 60 years and 30 to 35 kcal/kg of body proteasome) at the muscular level clearly have weight for individuals 60 years or older and shown that these systems are not overexpressed protein intake of 1.2 g/kg body weight per day.6 or hyperactivated in HD patients,25,26 unlike those However, actual energy and protein intakes with other wasting diseases.27 are often inadequate in HD patients: energy in- Whether HD per se is a catabolic procedure take has been reported to be 24 to 27 kcal/kg/d, has long been a matter of debate.21-24 Lim et al,21 whereas mean protein intake varies from approxi- by measuring whole-body leucine flux continu- mately 0.94 to 1 g/kg/d of protein.7-9 ously, showed that protein degradation compared Causes of inadequate nutritional intake in pa- with baseline data obtained before HD remained tients with uremia are taste abnormalities (acuity, constant during and for 4 hours after HD. How- metal flavor, and dry mouth), gastropathy and ever, 2 recent studies seem to conclusively show enteropathy, accumulation of anorectic factors, that HD is a protein catabolic event. Ikizler et inflammation and/or infection, medications, psy- al23 studied 11 patients 2 hours before, during, chosocial factors (depression, poverty, and alco- and 32 hours after HD treatment by means of a hol and/or drug abuse), and HD-related factors primed constant infusion of L-[1-13C]leucine (inadequate Kt/V,postdialysis fatigue, and cardio- and L-[ring-2H5] phenylalanine. As a result, HD vascular instability).8 led to increased whole-body (10%) and muscle Anorexia is present in approximately one third protein (133%) proteolysis with increased net of patients receiving HD and is the consequence whole-body protein loss (96%) and net forearm of multiple complex and only partially defined protein loss (164%).23 In a subsequent study, disturbances involving inflammation and altered Veeneman et al24 reported similar findings. hormonal and amino acid patterns.10,11 Acidosis Dialysis-Related Nutrient Losses Metabolic acidosis, a common event in HD During a standard HD treatment using low- patients, has been implicated as a potential cause flux cuprophane membranes, 4 to9goffree of PEM.28,29 Studies in vitro in cultured myo- amino acids are lost during the procedure through cytes and in vivo showed that acidosis increases the dialyzer during fasting and 8 to 10 g if whole-body protein degradation, skeletal muscle patients are eating.13 Peptides also are removed protein breakdown, and oxidation of branched in a range of 2 to 3 g/dialysis treatment,13 thus chain amino acids (BCAAs); decreases albumin THERAPY OF MALNUTRITION IN HEMODIALYSIS 373 synthesis; reduces the expression of insulin-like quently, the same investigators, in a study of growth factor (IGF) and growth hormone (GH) 17,185 patients on HD therapy, showed that receptors; and determines nutritional abnormali- serum albumin level was the strongest predictor ties.28-31 of mortality.40 Appetite rate also seems to predict outcome. Inflammation Kalantar-Zadeh et al41 recently showed that mor- In recent years, it was postulated that PEM in tality risk in patients with a poor appetite was 4 HD patients may be a consequence of chronic to 5 times greater than that in patients with a inflammatory processes that are common in pa- normal appetite, suggesting that a patient’s sub- tients with ESRD.32,33 According to Stenvinkel jective answer to a simple question about appe- et al,33 at least 2 types of PEM may occur in tite could have a high magnitude of predicting dialysis patients. The first type is associated with clinical outcome in HD patients. the uremic syndrome per se or factors associated PEM also significantly affects quality of life. with uremia (such as physical inactivity, under- Through the survey of 1,387 HD patients en- dialysis, dietary restrictions, and psychosocial rolled in the Hemodialysis Study, Dwyer et al42 factors) and is characterized by a modest reduc- observed that such nutritional status indicators as tion in serum albumin levels, absence of signifi- dietary energy intake, equilibrated normalized cant comorbidity, normal proinflammatory cyto- protein catabolic rate (nPCR), serum albumin kine levels, and low protein-energy intake because level, body mass index (BMI), calf circumfer- of uremic anorexia. The other type, defined as ence, and appetite were associated with health- malnutrition-inflammation complex syndrome, is related quality of life, even after controlling for characterized by marked hypoalbuminemia, comorbidities and HD dose. PEM-related deple- greater resting energy expenditure, markedly in- tion of muscle mass and muscle weakness also creased oxidative stress, increased protein catabo- impinge on quality of life. It is well known that lism, high C-reactive protein and proinflamma- HD patients are weaker than healthy subjects,43,44 tory cytokine levels, and the presence of comorbid with weakness including proximal and distal conditions. muscle groups. Johansen et al43 showed that ankle dorsiflexor muscles of dialysis patients are CONSEQUENCES weaker than those of healthy sedentary control Does PEM have an impact on the prognosis of subjects, and atrophy, not central activation fail- HD patients? To answer this question, numerous ure, is the primary mechanism for this weakness. reports focused on the association between hy- Together, the available evidence thus suggests poalbuminemia and/or poor nutritional status as- that PEM in HD patients acts as an independent sessed by means of anthropometry, subjective causative factor for mortality and suggests that global nutritional assessment (SGA), or total- preventive and therapeutic strategies should be body nitrogen and clinical outcome, providing adopted to improve prognosis and quality of life support to the hypothesis that malnutrition may for HD patients. cause or contribute to mortality.34-38 Qureshi et al38 clearly showed that PEM, NUTRITIONAL SCREENING AND ASSESSMENT assessed by means of SGA and anthropometric Similarly to all other chronic disorders, ESRD and biochemical measurements, was more preva- and HD therapy bear a significant risk for nutri- lent at the start of the study in patients who tional deterioration. Nephrologists should be- subsequently died than in survivors. Moreover, come progressively aware of this and routinely Kaplan-Meier survival at 36 months was signifi- screen their patients for the risk for or presence cantly lower in patients shown to be mildly of malnutrition. (40%) or moderately to severely (30%) malnour- Aims of nutritional screening are summarized ished at SGA than in those with normal nutri- as follows: (1) avoid time-consuming assess- tional status (75%). Lowrie et al39 reported in a ment of nutritional status in large patient popula- study of 12,000 patients on HD therapy that the tions, (2) detect all or nearly all patients at adjusted risk ratio for mortality increased progres- nutritional risk, and (3) select only patients in sively as serum albumin level decreased. Subse- need of more accurate evaluation of nutritional 374 BOSSOLA ET AL status. In addition to SGA (discussed next), more has been reported. However, concentrations of recently, other tools for screening of malnutrition this visceral protein with a 21-day half-life are have been proposed and validated, such as the influenced critically by dilution and inflamma- Nutritional Risk Screening 2002.45 It is strongly tion, which should always be taken into account suggested that 1 of these tools should be used when assessing malnutrition in HD patients.46 routinely by HD units at the time the patient is Transthyretin level was reported to be a reliable enrolled. marker of nutritional status, showing significant Aims of nutritional assessment may be summa- relationships with energy and protein intake, as rized as follows: (1) confirm the presence of well as with fat stores and lean body mass in both malnutrition after nutritional screening has been HD and peritoneal dialysis patients. Serum accomplished, (2) define the degree of malnutri- transthyretin levels less than 30 mg/dL (Ͻ300 tion, (3) predict and quantify the risk for compli- mg/L) are associated with increased risk for cations deriving from impaired nutritional status, morbidity and mortality, with the predictive value and (4) monitor the adequacy of nutritional treat- independent of serum albumin level. Transthyre- ment. tin level appears to be dependent on dilution and Nutritional assessment is a complex procedure inflammation, to a lesser extent.47 Interestingly, a including anthropometric, biochemical, and func- low percentage of lymphocytes recently was tional measurements. However, it must be recog- shown to be a predictor of mortality and hospital- nized that none of the variables currently used ization in HD patients and therefore appears to for nutritional assessment is a pure indicator of be a simple, universally available good nutri- nutritional status because both hydration and the tional marker.48 inflammatory response may significantly alter During recent years, SGA has been used for nutritional indices irrespective of changes in true both nutritional screening and assessment.49 SGA nutritional status. Conversely, it now is widely involves the judgment of malnutrition by several recognized that disease-related malnutrition is markers, but overall is dependent on body weight crucially related to the inflammatory response, and gastrointestinal symptoms, not serum albu- which, in turn, determines energy and substrate min level. Subjective measures include the pa- metabolism abnormalities and accelerates deple- tient’s history of weight loss, anorexia, and vom- tion of lean body mass. This renders the assess- iting and the physician’s estimate of muscle ment of nutritional status particularly difficult in wasting, edema, and loss of subcutaneous fat. Its HD patients, in whom a condition of chronic validity has been established by using objective inflammation frequently is present.46-51 measures in continuous ambulatory peritoneal Among anthropometric indices, measurement dialysis patients and HD patients. It is inexpen- of skinfold thickness requires particular skill, is sive, can be performed rapidly, and requires only subject to a 20% interobserver error, and may be brief training. However, it is largely subjective unreliable because of the influence of fluid reten- and therefore its sensitivity, precision, and repro- tion, thus providing a falsely high estimation of ducibility may be insufficient to detect the pres- body fat.46 Arm muscle circumference is a reli- ence of PEM or small changes in nutritional able index of lean body mass, but also may be status over time in HD patients. Moreover, SGA influenced by hyperhydration. Therefore, routine consists of only 3 nutritional levels (normal, use of these measurements cannot be recom- mildly to moderately malnourished, and severely mended. Conversely, there is consensus that BMI malnourished).46,49 values less than 18.5 and recent undesired weight Using components of conventional SGA, a loss greater than 10% of usual body weight are fully quantitative scoring system, the Dialysis indicators of malnutrition.46 Malnutrition Score recently was developed.50 Among biochemical indices, serum albumin The score consists of 7 features: weight change, concentrations have long been used to detect dietary intake, gastrointestinal symptoms, func- malnutrition in HD patients, with levels less than tional capacity, comorbidity, subcutaneous fat, 3.5 g/dL (Ͻ35 g/L) considered indicators of and signs of muscle wasting. Considering that malnutrition. Moreover, for every 1-g/L decrease each component has a score from 1 (normal) to 5 in albumin level, a 10% increase in mortality risk (very severe), malnutrition score is a number THERAPY OF MALNUTRITION IN HEMODIALYSIS 375 between 7 (normal) and 35 (severely malnour- teracted, and drugs that reduce appetite have ished). The Dialysis Malnutrition Score has been avoided.2 shown a significant correlation with age, years of Initiation of maintenance HD therapy has been dialysis therapy, and the combination of midarm reported to improve not only quality of life, but muscle circumference, BMI, serum albumin con- also nutritional status.46 However, it recently centration, and total iron-binding capacity.50 It was shown in an elegant study that amelioration was suggested that the Dialysis Malnutrition of nutritional status during the first year of HD Score may be superior to the SGA because it therapy essentially is caused by an increase in extends its reliability and precision and better body fat mass.52 defines PEM severity.50 Increasing the dose or using high-flux dialysis Subsequently, by adding 3 elements to the does not seem to improve nutritional status. Dialysis Malnutrition Score (BMI, serum albu- Rocco et al53 analyzed nutritional data from the min level, and total iron-binding capacity), a new Hemodialysis Study, a multicenter, prospective, comprehensive Malnutrition-Inflammation Score randomized, 2 ϫ 2 factorial, clinical trial that (MIS) has been developed.51 The MIS has 10 evaluated the efficacy of standard and high deliv- components, each with 4 levels of severity from ered dose of dialysis and low- and high-flux 0 (normal) to 3 (very severe). MIS score was dialysis. The investigators showed that neither associated significantly with prospective hospital- mean serum albumin levels which declined by ization and mortality, as well as measures of 0.21 g/dL (2.1 g/L), nor mean postdialysis weight, nutrition, inflammation, and anemia.51 Among which declined by 2.7 kg, was affected signifi- currently available tools, the MIS appears to be cantly by either study intervention. Moreover, the most reliable for grading PEM in HD pa- all anthropometric measures, appetite scores, tients. and mean equilibrated PCR declined during follow-up in all treatment groups. PREVENTION AND TREATMENT The increase in number of dialysis sessions Setting a boundary between prevention and through daily HD or 6 HD sessions weekly has treatment of malnutrition in HD patients is diffi- been shown to improve appetite and food in- cult and probably not clinically advantageous. take.54-61 This probably is caused by a general The nutritional approach to HD patients should feeling of well-being; increased physical activ- be dynamic in nature and continuously adapted ity; fewer dietetic restrictions; decreased dose of to the changing metabolic and nutritional needs such medications as phosphate, potassium bind- of every single patient throughout his or her ers, and antihypertensive drugs; body composi- dialytic history. tion; and nutritional status. It also was suggested Although nutritional screening, dietary coun- that daily HD increases the clearance of potential seling, and dialytic adequacy have a critical role anorexic factors.58,59 Moreover, it seems that in malnutrition prevention, other measures, such daily and nocturnal HD significantly increase as pharmacological intervention, acidosis correc- serum albumin levels.54-65 Galland et al59 showed tion, artificial nutrition, or oral supplement admin- that the switch from standard 4-hour thrice- istration may be effective in both preventing or weekly HD to 2-hour daily HD in a 1-year correcting malnutrition, depending on the phase program increases protein and energy intake, at which they are used. However, increasing the improves well-being, and decreases the use of awareness of physicians about the nutritional sodium polystyrene sulfonate (Kayexalate; risk of HD patients still appears to be the most Sanofi-Synthelabo, Milan, Italy) and phosphate valuable tool for preventing the onset of deterio- binders, drugs known to decrease appetite. The ration of nutritional status. most recent report from the London Daily/ Nocturnal Study62 showed that serum albumin Dialytic Adequacy levels and arm muscle area increased in patients An adequate nutritional program can lead to undergoing daily HD, but not those receiving improvement in nutritional status only when op- nocturnal HD. timal dialytic dose has been established, the Conversely, hemodiafiltration with on-line re- catabolic stimulus possibly present has been coun- generation of the ultrafiltrate, a dialytic tech- 376 BOSSOLA ET AL nique in which the ultrafiltrate passes through a tite through stimulation of neuropeptide Y in the cartridge containing uncoated charcoal and is hypothalamus; modulation of calcium channels reinfused into the blood stream, which has been in the ventromedial hypothalamus, a well-known claimed to improve nutritional status or prevent satiety center; and inhibition of the activity of its deterioration while decreasing inflammatory such proinflammatory cytokines as interleukin 1 response, failed to improve serum albumin and (IL-1), IL-6, and tumor necrosis factor ␣ (TNF- cholesterol levels long term or decrease levels of ␣). In several clinical trials, megestrol acetate inflammatory parameters.66 improved appetite, caloric intake, and nutritional status in patients with cancer.69 Dietary Counseling Few studies have been conducted in HD pa- A comprehensive nutritional, diet, and appe- tients.70,71 Burrowes et al,70 after administration tite assessment to identify whether nutritional of a moderate dose of megestrol acetate (Ն320 status is too low is mandatory, as well as defini- mg/d) to a long-term HD patient, observed a fat tion of problems related to self-feeding, access to mass increase (by 163%) and fat-free mass de- food, gastrointestinal distress, and, eventually, crease (by 10.6%) as a consequence of improve- identification of active psychic, social, medical, ment in reported appetite and increases in energy dialytic, or medicinal-related issues that could and protein intake. In the study by Boccanfuso et affect food intake.1,2,46 Dietary counseling to al,71 in which megestrol acetate was adminis- correct decreased or unhealthy nutrient intake, tered long term, many side effects, such as head- performed by a nutritionist, has been useful. aches, dizziness, confusion, diarrhea, and hyper- Akpele and Bailey67 recently showed that the glycemia, were observed. It is well known that rate of change in serum albumin levels in stable megestrol acetate also can induce thromboem- HD patients was significantly greater in patients bolic phenomena, breakthrough uterine bleeding, randomly assigned to receive intensive nutri- peripheral edema, hyperglycemia, hypertension, ad- tional counseling than those administered oral renal suppression, and adrenal insufficiency.70,71 supplements. Similarly, Leon et al68 showed that On the basis of these considerations, it seems tailored nutritional intervention to patient-spe- that the use of megestrol acetate in clinical cific barriers to adequate protein nutrition re- practice cannot be recommended. Large, ran- sulted in increased serum albumin concentra- domized, controlled trials are warranted to tions in 52 HD patients, even those with high define the exact role of megestrol acetate in serum C-reactive protein levels. preventing and treating anorexia in malnour- Consumption of a protein- and energy- ished HD patients. enriched meal also may be useful. Veeneman et al24 showed that patients who consumed a meal Oral Nutritional Supplements that gave 0.62 Ϯ 0.09 g/kg of protein and 15 Ϯ 2 Oral nutritional supplements have been de- kcal/kg during the HD session had a positive signed specifically for HD patients and are avail- whole-body protein balance to the same extent as able as energy or protein sources or a combina- on a nondialysis day. Moreover, whole-body tion of both. Supplements are in the form of solid protein breakdown was decreased to approxi- food, powders, or liquid formulations. mately two thirds the rate observed during fast- Earlier nonrandomized uncontrolled studies ing, protein synthesis increased slightly to 125%, led to contradictory and nondefinitive results. and oxidation increased strongly to 205%. Ac- Moreover, interpretation of these results is com- cording to the investigators, this nutritional strat- plicated by the omission of relevant clinical egy, abolishing the negative effect of dialysis on information and the different supplements used whole protein balance, offers a possibility of (Table 1).72-78 preventing PEM. Recently, a few randomized controlled trials were conducted (Table 1).79-84 Eustace et al79 Appetite Stimulants showed that oral amino acid supplements in Megestrol acetate is a synthetic orally active hypoalbuminemic HD patients resulted in a sig- derivate of the naturally occurring hormone pro- nificant improvement in serum albumin levels, gesterone. Megestrol acetate may induce appe- grip strength, and Short-Form Health Survey THERAPY OF MALNUTRITION IN HEMODIALYSIS 377

Table 1. Effects of Oral Nutritional Supplements on Nutritional Status in HD Patients

No. of Duration Protein Intake Reference Patients Type of Oral Supplementation (mo) (g/d) Effects

Phillips et al,72 1978 20 EAA ϩ histidine 3 60 Increase in serum albumin Acchiardo et al,73 1982 15 EAA ϩ histidine 3.5 1/kg Increase in total protein, albumin, transferrin Allman et al,74 1990 21 Glucose polymer 6 1.0-1.2/kg Weight gain Allman et al,75 1992 10 Glucose polymer 6 1.0-1.2/kg Weight gain, increase in body fat Mastroiacovo et al,76 1993 36 Mixture of EAA and NEAA 1 ? Increase in EAA/NEAA ratio Cuppari et al,77 1994 14 Energy and protein 4 1.2/kg Increase in body fat supplementation Cockram et al,78 1998 ? Energy and protein 3 wk 1.25/kg Improvement of serum phosphorus and calcium-phosphorus product Eustace et al,79 2000 47 EAA 3 10.8 Increase in serum albumin Hiroshige et al,80 2001 28 BCAA 12 12 Increase in serum albumin and anthropometric indices Wilson et al,81 2001 46 Energy and protein 4 ? Improvement of nutritional status in malnourished patients Caglar et al,83 2002 85 Energy and protein 6 16.6 Increase in serum albumin, prealbumin, SGA Sharma et al,82 2002 Control group: dietary recall 15 Both groups, improvement in dry and counseling; Treatment weight and BMI; treatment group: oral group, increase in serum supplementation albumin Steiber et al,84 2003 26 26, treatment group; 91, 3 ? Decrease in prognostic nutrition control group index score and patient risk for hospitalization

Abbreviations: EAA, essential amino acid; NEAA, nonessential amino acid.

Mental Health score, but not serum amino acid al,82 adult HD patients without diabetes with a levels, Short-Form Health Survey Physical Health BMI less than 20 and serum albumin level less score, or anthropometric measures. than 4.0 g/dL (Ͻ40 g/L) were randomized into a Hiroshige et al80 randomly assigned 28 mal- control group that received appropriate monitor- nourished patients to the administration of BCAA ing, including dietary recall and counseling for (12 g/d) supplementation or placebo. After 6 the prescribed diet (protein intake, 1.2 g/kg/d; months, anthropometric indices and serum albu- energy intake, 35 to 40 kcal/kg/d) and a treat- min levels showed a statistically significant in- ment group that, in addition, received a post-HD crease in the BCAA group. After exchanging nutritional supplement providing 500 kcal and BCAA for placebo, spontaneous oral food intake 15 g of protein for 1 month. Both groups showed decreased, although the favorable nutritional sta- an improvement in dry weight and BMI, whereas tus persisted for the next 6 months. According to the supplemented group showed a significant the investigators, these data suggest that normal- increase in serum albumin levels and functional ization of plasma BCAA levels can decrease scoring on a 10-point Karnofsky scale. Recently, anorexia and improve energy and protein intake Caglar et al83 followed up 85 HD patients for a in HD patients. 3-month baseline period during which they re- Wilson et al81 showed that nutritional reple- ceived conventional nutritional counseling and tion occurred more quickly and was maintained then switched to an intervention period during for a longer period in HD patients with mild which an oral supplement (475 calories and hypoalbuminemia who were randomized to re- 16.6 g of protein) was administered for 6 months ceive diet counseling associated with oral supple- during the HD session to ensure adequate compli- mentation, with respect to patients who received ance. Unfortunately, 17 patients (20%) refused to diet counseling only. In the study by Sharma et ingest the oral supplement and 10 patients 378 BOSSOLA ET AL

(11.7%) showed decreased compliance, ingest- Pupim et al85 showed that IDPN promotes a ing less than 75% of the prescribed nutritional large increase in whole-body protein synthesis supplementation. At the end of the study, a signifi- and a significant decrease in whole-body proteol- cant increase in serum albumin and prealbumin ysis, along with a significant increase in forearm levels and SGA was observed. Potential limita- muscle protein synthesis. The net result is a tions of the study are the lack of a randomized or change from an essentially catabolic state to a placebo-controlled design and the high incidence highly positive protein balance. The same inves- of noncompliance or reduced compliance. These tigators also showed that IDPN improves the results further confirm that poor compliance is a fractional synthetic rate of albumin during HD in key limiting factor for oral nutritional supple- parallel with significant improvements in ment efficacy and suggest that improvements in whole-body protein synthesis.86 Pupim et al87 palatability might considerably improve patient also showed that the acute anabolic effects of acceptance of these kinds of nutritional supple- IDPN are significantly augmented by exercise mentation. during HD. Artificial Nutrition However, studies that examined potential ben- Intradialytic parenteral nutrition (IDPN) has efits of IDPN in terms of correction of PEM and 88-97 the advantage of providing calories and proteins improvement of clinical outcome (Table 2) during HD without the need for establishing a are characterized by various and numerous bi- central venous catheter. Nutrients are infused ases: absence of randomized control groups; di- into the blood returning from the dialyzer to the etary intake of food not controlled or monitored; patient. the size of study population too small for a valid

Table 2. Effects of IDPN on Nutritional Status in HD Patients

No. of Duration Reference Patients Protein-Calorie Intake/d (mo) Effects

Snyder et al,88 1991 6 Glucose (125 g) ϩ lipid (50 g) 3-6 None ϩ amino acids (42.5) Schulman et al,89 1993 7 18 kcal/kg ϩ 0.069 g/protein/ 1.5 Increase in serum albumin kg ϩ GH Capelli et al,90 1994 50 9 Better survival, increase in serum albumin Chertow et al,91 1994 Better survival in hypoalbuminemic patients; increased mortality in patients with normal albumin levels Hiroshige et al,92 1998 10 Glucose 50% (200 mL) ϩ 12 Increase in serum albumin, transferrin, essential amino acids 7% lymphocyte count, anthropometric (200 mL) ϩ lipid emulsion parameters 20% (200 mL) Mortelmans et al,93 1999 16 Glucose 50% (250 mL) ϩ 9 Increase in body weight, serum lipids (250 mL) ϩ amino prealbumin, transferrin acids 7% (250 mL) Berneis et al,94 1999 7 Glucose 15% (37.5 g/h) ϩ fat 3 Increase in lean and fat body mass (12.5 g/h) Pupim et al,85 2002 7 188 cal/h Large increase in whole-body protein synthesis, significant decrease in whole-body proteolysis Krause et al,95 2002 4 Children Glucose ϩ amino acids ϩ fat 1-3 Increase in oral caloric intake, BMI, total 20% lymphocyte count Cherry and Shalansky,96 24 Dextrose 50% (250 mL) ϩ fat 4.3 Increase in body weight, serum albumin 2002 emulsion 20% (250 mL) ϩ amino acids 10% (250 or 500 mL) Czekalski and 97 Amino acids 10% (500 mL) Ն6 SGA improvement, increase in serum Hozejowski,97 2004 albumin, small increase in midarm circumference THERAPY OF MALNUTRITION IN HEMODIALYSIS 379 statistical analysis; data not adjusted for comor- binding proteins is common in patients with bid conditions of patients; measurement of only ESRD receiving long-term HD therapy.99 Follow- nutritional outcomes, with morbidity and mortal- ing studies on the positive effects of recombinant ity rarely evaluated; and energy and protein in- GH substitution therapy on body composition take often inappropriate. Overall, it seems that and quality of life in adult patients with GH no clear clinical benefits secondary to IDPN deficiency,100 recent short-term and long-term have been shown, and the lack of adequate trials prospectively evaluated the role of recom- prospective randomized studies has hampered binant GH in reversing malnutrition and, in par- definitive conclusions. ticular, muscle wasting, in HD patients (Table Enteral nutrition (EN) is a safe and effective 3).101-107 It seems that although GH increases way to provide nutritional support in patients nitrogen balance, dietary protein use, and serum with a functioning gastrointestinal tract in whom and IGF-1 levels, a significant long-term (12- sufficient oral intake is precluded. Depending on month) improvement in anthropometric and nu- the anticipated duration of EN treatment, a naso- tritional parameters usually is not observed, as gastric tube or percutaneous endoscopic gastros- well as an increase in muscle mass and function. tomy may be placed. The latter approach is Moreover, it remains to be clarified through large indicated when periods longer than 30 to 40 days randomized trials whether GH administration of EN are required. HD-related PEM does not reduces the nutrition-related morbidity and mor- represent in itself an indication for EN. However, tality of HD patients and what the long-term it is possible that patients on long-term HD effects are on other tissues and organs (eg, bone). therapy may develop concurrent diseases (eg, Finally, it must be emphasized that GH has the head/neck or upper gastrointestinal cancer, sur- potential to increase serum glucose levels and gery, and critical illness) as an indication to EN exacerbate hyperglycemia in patients with diabe- treatment. In this case, disease-specific EN formu- tes. las are administered to prevent electrolyte and/or No data currently are available on the efficacy metabolic disturbances. In pediatric HD patients, of IGF-I administration in HD patients, whereas EN through a nasogastric tube or percutaneous effects on muscle mass in humans with wasting endoscopic gastrostomy often is indicated to not diseases other than uremia are controversial.108 only prevent/correct malnutrition, but particu- In the clinical setting, systemic IGF-I administra- larly to prevent growth deficits.98 tion poses a number of concerns: its glucose- lowering effect restricts the dose that can be Anabolic Hormones: GH, IGF-I used, exogenously administered IGF-I has a short Resistance to the actions of GH, insulin, and half-life with consequent limited bioavailability IGF-I secondary to chronic metabolic acidosis, and scarce potential of single intravenous injec- inflammation, and high concentrations of IGF- tions, and it is likely that IGF-I effects also could

Table 3. Effects of GH on Nutritional Status in HD Patients

No. of Duration of GH Reference Patients GH Dose Treatment Effects

Ziegler et al,101 1991 5 5 or 10 mg 3 wk Reduction in urea generation and PCR Schulman et al,102 1993 7 5 mg/each dialysis 6 wk Increase in serum albumin, transferrin, IGF-I Iglesias et al,103 1998 17 0.2 IU/kg/d 4 wk Increase in body weight, IGF-I, transferrin Johansson et al,104 1999 20 66.7 ␮g/kg every 2 d 6 mo Increase in serum albumin, handgrip strength Hansen et al,105 2000 20 4 IU/m2/d 6 mo Increase in lean body mass, decrease in fat mass Garibotto et al,106 2000 6 5 mg/d 6 wk Increase in muscle protein synthesis, lean body mass, resting energy expenditure; decrease in fat mass Kotzmann et al,107 2003 19 0.125 IU/kg 3 times/wk, 12 mo No significant changes in nutritional then 0.250 IU/kg 3 parameters times/wk 380 BOSSOLA ET AL

Table 4. Effects of Nandrolone Decanoate on Nutritional Status in HD Patients

Duration of No. of Dose treatment with Reference Patients (mg/wk) nandrolone Effects

Gascon et al,111 1999 14 200 6 mo Increase in body weight, muscle mass, hemoglobin Johansen et al,112 1999 14 100 6 mo Increase in body weight, creatinine; decrease in fatigue Pai et al,113 2002 9 100 Ͼ30 d Increase in serum albumin and hematocrit take place in tissues other than skeletal muscle, nandrolone decanoate on lean body mass, func- thus promoting stimulation and survival of rap- tional status, and quality of life in HD patients. idly dividing cells (eg, cancer cells). It must also The study included 29 patients undergoing HD be taken into account that jaw pain, nausea, for at least 3 months; 14 of these patients were hypoglycemia, occasional altered mentation, and administered 100 mg of nandrolone decanoate cardiac arrhythmias are not uncommon after intramuscularly once a week for 6 months and 15 IGF-I administration.109 patients were administered a placebo. Subjects administered nandrolone decanoate gained an Androgenic Anabolic Steroids average of 2.5 kg more lean body mass than Stimulation of protein anabolism in muscle those administered placebo, had a concomitant may be attempted through the use of anabolic increase in serum creatinine levels (suggesting androgenic steroids (AASs), a large family of that nandrolone caused increased muscle mass), testosterone-related hormones that vary in terms and experienced a significant reduction in fatigue of chemical structure, mode of action, anabolic 109,110 and time required for walking and stair climb- effects, and risk for undesired side effects. ing.112 Pai et al113 reviewed medical records of The most commonly used AASs in clinical prac- long-term HD patients administered nandrolone tice are nandrolone decanoate and oxandrolone. decanoate for longer than 30 days. Of 9 patients AAS administration induces increases in messen- evaluated, 2 patients were administered 25 mg of ger RNA expression of skeletal muscle androgen nandrolone decanoate intramuscularly every week receptor, increases intracellular use of amino and 7 patients were administered 100 mg every 2 acids derived by protein degradation, and stimu- weeks. Increases in serum albumin levels (from lates net muscle protein synthesis.109,110 In addi- 2.9 Ϯ 0.6 to 3.3 Ϯ 0.4 g/dL [29 Ϯ 6to33Ϯ 4 tion, AASs would induce inhibition of protein g/L]), dry weight (64.4 Ϯ 11.7 to 66 Ϯ 10.9 kg), catabolic processes through interaction with the and hematocrit (from 28.2% Ϯ 4.5% to 33.2% Ϯ glucocorticoid receptor.109,110 5.1%) were observed.113 Although positive effects have been docu- These data warrant additional controlled clini- mented with nandrolone decanoate and oxan- cal trials evaluating both long-term efficacy and drolone in patients with burns, cancer, human risk for dangerous side effects.109 To this regard, immunodeficiency virus infection, and chronic an important and significant warning comes from obstructive pulmonary disease, data for HD pa- tients, although undoubtedly promising, are lim- the report of complications associated with AAS ited (Table 4).111-113 use among athletes, such as reduction in high- Gascon et al111 administered nandrolone de- density lipoprotein levels, irregular menses, viril- canoate (200 mg/wk intramuscularly for 6 ization and hirsutism in women, testicular atro- months) to 14 elderly HD patients not adminis- phy and infertility in men, hypercoagulation, cardiomyopathy, psychiatric disorders, hepatocel- tered erythropoietin and compared them with 19 109-110 patients administered erythropoietin. In the nan- lular carcinoma, and sudden death. drolone group, but not the erythropoietin group, significant increases in body weight, muscle mass, Correction of Acidosis and hemoglobin level were observed. Johansen Five prospective clinical studies evaluated the et al112 conducted a randomized, double-blind, effects of acidosis correction on nutritional status placebo-controlled trial to assess the effects of in HD patients (Table 5). THERAPY OF MALNUTRITION IN HEMODIALYSIS 381

Table 5. Effects of Correction of Acidosis on Nutritional Status in HD Patients

No of Duration Reference Patients Treatment (wk) Effects

Williams et al,115 1997 46 40 v 30 mEq/L of bicarbonate 32 Small significant increase in triceps skinfold in dialysate thickness Brady et al,114 1998 640v 35 mEq/L of bicarbonate 16 No difference in serum albumin, weight in dialysate gain, lymphocyte count, or nPCR Movilli et al,117 1998 12 Oral sodium bicarbonate 12 Significant increase in serum albumin supplementation Lin et al,30 2002 17 Dialysate bicarbonate 24 No changes in BMI, midarm muscle concentration increased circumference, triceps skinfold thickness, from 35 to 38-40 mEq/L serum albumin, transferrin, cholesterol Blair et al,116 2003 35 39 v 35 mEq/L bicarbonate in 24 No differences in serum albumin, SGA; dialysate decrease in nPCR

NOTE. To convert bicarbonate in mEq/L to mmol/L, multiply by 1.

Brady and Hasbargan114 showed that serum (from 3.49 Ϯ 0.21 to 3.79 Ϯ 0.29 g/dL [34.9 Ϯ albumin levels, interdialytic weight gain, total 21 to 37.9 Ϯ 29 g/L]), whereas nPCR decreased lymphocyte count, and nPCR were similar in significantly (from 1.11 Ϯ 0.17 to 1.03 Ϯ 0.17 patients randomly assigned to high-bicarbonate g/kg/d). Thus, continuous correction by means of (40 mEq/L [40 mmol/L]) or standard (35 mEq/L sodium bicarbonate would appear to be more [35 mmol/L]) dialysate after 16 weeks. In the effective than intradialytic bicarbonate adminis- study by Williams et al,115 in which 46 patients tration in preventing and correcting acidosis- with nearly normal serum and bicarbonate levels related malnutrition. Long-term oral administration were hemodialyzed with low- (30 mEq/L [30 of sodium bicarbonate also improved nutritional mmol/L]) or high-bicarbonate (40 mEq/L [40 status and decreased the duration of hospitalization mmol/L]) dialysate, a small significant increase in peritoneal dialysis patients.118,119 in triceps skinfold thickness was documented with high-bicarbonate dialysate, whereas no dif- Anti-Inflammatory Drugs ferences in midarm muscle circumference, urea Although the prevalence of inflammation in generation rate, or nPCR were observed. In the HD patients is high, there are as yet no valid study by Blair et al,116 who evaluated the effects recommendations on how chronic inflammation of increasing delivered bicarbonate dose in dialy- should be handled.33 Possible strategies for re- sate from 35 to 39 mEq/L (35 to 39 mmol/L) for versing the cytokine-dependent inflammatory pro- 6 months, nPCR decreased significantly with cess in HD patients include anti-cytokine antibod- respect to baseline, whereas serum albumin level ies and such drugs as steroids, pentoxifylline, and SGA did not differ. Lin et al30 evaluated the statins, and angiotensin-converting enzyme in- effect on nutritional status of increasing dialysate hibitors.33 bicarbonate concentration from 35 to 38 mEq/L Anti-cytokine antibodies (eg, anti–TNF-␣ an- (35 to 38 mmol/L) for 1 month and then 38 to 40 tibodies, soluble TNF-␣ receptors, IL-1, and IL-6 mEq/L (38 to 40 mmol/L) for 5 months in 17 receptor antagonists) have improved clinical find- acidotic HD patients. There was no significant ings and inflammatory parameters in patients change in interdialysis body weight gain, BMI, with such wasting diseases as rheumatoid arthri- triceps skinfold thickness, midarm muscle cir- tis and chronic heart failure.108 Unfortunately, cumference, or serum albumin, transferrin, and data for HD patients are still lacking. cholesterol levels. Interestingly, when Movilli et Pentoxifylline, a xanthine-derived phosphodi- al117 supplemented 12 patients with ESRD on esterase inhibitor used to clinically improve cir- maintenance HD therapy with relatively normal culation in patients with occlusive diseases of the serum albumin levels with oral sodium bicarbon- arteries in the lower limbs, inhibits monocyte ate for 3 months, with a mean dose of 2.7 Ϯ 0.94 and T-cell production of TNF-␣ and interferon ␥ g/d, serum albumin levels increased significantly and has been shown to improve the clinical 382 BOSSOLA ET AL outcome of patients affected by idiopathic di- contribute significantly to reduce the gap exist- lated cardiomyopathy, childhood type 1 diabetes, ing between recommended and actual nutritional systemic vasculitis, and rheumatoid arthritis.108 intakes, the latter frequently being inadequate to No study has been conducted in HD patients. meet the increased nutritional needs of HD pa- However, Biolo et al120 showed in patients with tients. Patients also should be screened regularly chronic uremia not receiving dialysis that intrave- for the presence of anorexia by using validated nous infusion of pentoxifylline significantly de- questionnaires. creased whole-body proteolysis in either the post- A cost-effective approach to these aims would absorptive state or during amino acid infusion. appear to include an expert dietitian on the staff With regard to statins, Chang et al121 showed of every HD unit. that simvastatin administration (20 mg/d for 8 If food intake is decreased because of an- weeks) significantly decreased total cholesterol orexia, BCAA supplements (12 g/d) may be used and C-reactive protein levels and increased se- safely without the risk for adverse effects or poor rum albumin concentration with respect to a adherence to treatment. Conversely, routine use control group. of megestrol acetate to improve anorexia in HD Angiotensin-converting enzyme inhibitors also patients cannot be recommended in clinical prac- have shown anti-inflammatory properties in both tice until confirmatory prospective clinical trials the general population and patients with ESRD122 are performed. Long-term use of oral nutritional and are possible future candidates for the treat- supplements may be limited by poor palatability. ment of malnutrition in the HD population. The use of IDPN should be limited to severely All these promising results suggest that pro- malnourished patients in whom oral intake is spective, randomized, controlled studies are insufficient to meet nutritional needs until defi- needed to elucidate whether anti-inflammatory nite data on its risk-benefit and cost-efficacy therapies are safe and may have a beneficial ratios are available. effect on nutritional status, cardiovascular risk, Acidosis should be accurately prevented. Suf- and mortality in patients with ESRD with chronic ficient evidence exists to recommend long-term inflammation on maintenance HD therapy. oral administration of low-dose sodium bicarbon- ate (2 to 3 g/d) to prevent metabolic acidosis, CONCLUSIONS AND PRACTICAL thus improving nutritional status, possibly by RECOMMENDATIONS decreasing the rate of protein catabolism. PEM is a major issue in the long-term manage- AASs certainly are among the most effective ment of HD patients, adversely affecting morbid- and promising agents for improvement of muscle ity, mortality, functional activity, and quality of mass in patients with ESRD and other wasting life. Given its multifactorial pathogenesis, the conditions. However, their routine use in HD approach to this disease-related and therapy- patients cannot be recommended until additional related life-threatening syndrome must imply a evidence is available on their long-term efficacy multidisciplinary effort that includes nutritional, and safety. metabolic, and pharmacological interventions. Accurate monitoring of inflammatory response Routine nutritional screening and assessment also is advisable because it might show the at the start of dialysis therapy should be per- presence of malnutrition-inflammation complex formed by using such rapid and user-friendly syndrome. Considering the negative role of in- tools as the SGA or Nutritional Risk Screening flammation on protein metabolism, prospective 2002, which can be carried out easily, even by trials should evaluate the risks and benefit of busy admitting nursing and medical staff. long-term anti-inflammatory drug therapy, at least Nutritional counseling should become part of in patients showing altered levels of inflamma- the holistic approach to HD patients. A personal- tion indices (C-reactive protein, fibrinogen, and ized nutrition plan for every patient should be ferritin).12,122 worked out. Nutritional follow-up of HD pa- tients is of pivotal importance in monitoring REFERENCES adequacy of food intake and nutritional status 1. Pupim LB, Ikizler TA: Uremic malnutrition: New throughout the years of HD treatment. This might insights into an old problem. Semin Dial 16:224-232, 2003 THERAPY OF MALNUTRITION IN HEMODIALYSIS 383

2. Locatelli F, Fouque D, Heimburger O, et al: Nutritional 21. Lim VS, Yarasheski KE, Flanigan M: The effect of status in dialysis patients: A European consensus. Nephrol uraemia, acidosis, and dialysis treatment on protein metabo- Dial Transplant 17:563-572, 2002 lism: A longitudinal leucine kinetic study. Nephrol Dial 3. Bergstrom J, Lindholm B: Nutrition and adequacy of Transplant 13:1723-1730, 1998 dialysis. How do hemodialysis and CAPD compare? Kidney 22. Berkelhammer CH, Baker JP, Letter LA, et al: Whole- Int Suppl 43:S39-S50, 1993 body protein turnover in adult hemodialysis patients as 4. Aparicio M, Cano N, Chauveau P, et al: Nutritional measured by 13C-leucine. Am J Clin Nutr 46:778-783, 1987 status of hemodialysis patients: A French National Coopera- 23. Ilkizer TA, Pupim LB, Brouillette JR, et al: Hemodi- tive Study. Nephrol Dial Transplant 14:1679-1686, 1999 alysis simulates muscle and whole body protein loss and 5. Ikizler TA, Greene JH, Wingard RL, Parker RA, Hakim alters substrate oxidation. Am J Physiol Endocrinol Metab RM: Spontaneous dietary intake during progression of 282:E107-E116, 2002 chronic renal failure. J Am Soc Nephrol 6:1386-1391, 1995 24. Veeneman JM, Kingma HA, Boer ST, et al: Protein 6. National Kidney Foundation: K/DOQI Clinical Prac- intake during hemodialysis maintains a positive whole body tice Guidelines for Nutrition in Chronic Renal Failure. Am J protein balance in chronic hemodialysis patients. Am J Kidney Dis 37:S66-S70, 2001 (suppl 2) Physiol Endocrinol Metab 284:E954-E965, 2003 7. Bossola M, Muscaritoli M, Tazza L, et al: Variables 25. Bossola M, Muscaritoli M, Costelli P, et al: Muscle associated with reduced dietary intake in hemodialysis pa- ubiquitin-mRNA levels in patients with end-stage renal tients. J Ren Nutr 15:244-252, 2005 disease on maintenance hemodialysis. J Nephrol 15:552- 8. Riella M: Malnutrition in dialysis: Malnourishment or 557, 2002 uremic inflammatoy response? Kidney Int 57:1211-1232, 26. Raj DS, Shah H, Shah VO, et al: Markers of inflam- 2000 mation, proteolysis, and apoptosis in ESRD. Am J Kidney 9. Ahmed KR, Kopple JD: Nutrition in maintenance Dis 42:1212-1220, 2003 hemodialysis patients, in Kopple JD, Massry SG (eds): 27. Micth WE, Goldberg AL: Mechanisms of muscle Nutritional Management of Renal Disease, chap 8. Balti- wasting. The role of the ubiquitin-proteasome system. N Engl more, MD, Williams & Wilkins, 1997, pp 563-575 J Med 335:1897-1905, 1996 10. Aguilera A, Codoceo R, Bajo MA, et al: Eating 28. Mehrotra R, Kopple JD, Wolfson M: Metabolic acido- disorders in uremia: A question of balance in appetite regula- sis in maintenance dialysis patients: Clinical considerations. tion. Semin Dial 17:44-52, 2004 Kidney Int Suppl 88:S13-S25, 2003 11. Bossola M, Muscaritoli M, Valenza V, et al: Anorexia 29. Szeto CC, Chow KM: Metabolic acidosis and malnu- and serum leptin levels in hemodialysis patients. Nephron trition in dialysis patients. Semin Dial 17:371-375, 2004 Clin Pract 97:C76-C82, 2004 30. Lin SH, Lin YF, Chin HM, Wu CC: Must metabolic 12. Anderstam B, Mamoun AH, Sodersten P, Bergstrom acidosis be associated with malnutrition in hemodialysed J: Middle sized molecule fractions isolated from uremic patients? Nephrol Dial Transplant 17:2006-2010, 2002 ultrafiltrate and normal urine inhibit ingestive behavior in 31. Chauveau P, Fouque D, Combe C, et al: Acidosis and the rat. J Am Soc Nephrol 7:2453-2460, 1996 nutritional status in hemodialyzed patients. French Study 13. Kopple JD, Swendseid ME, Shinaberger JH, Umezawa CY: The free and bound aminoacid removed by hemodialy- Group for Nutrition in Dialysis. Semin Dial 13:241-246, sis. Tran Am Soc Artif Intern Organs 19:309-313, 1973 2001 14. Ikizler TA, Flakoll PJ, Parker RA, Hakim RM: Ami- 32. Kalantar-Zadeh K, Ikizler TA, Block G, Avram MM, noacid and albumin losses during hemodialysis. Kidney Int Kopple JD: Malnutrition-inflammation complex syndrome 46:830-837, 1994 in dialysis patients: Causes and consequences. Am J Kidney 15. Wathen RL, Keshaviah P, Hommeyer P, et al: The Dis 42:864-881, 2003 metabolic effect of hemodialysis with and without glucose 33. Stevinkel P, Heimburger O, Lindholm B, Kaysen GA, in the dialysate. Am J Clin Nutr 312:1870-1875, 1978 Bergstrom J: Are there two types of malnutrition in chronic 16. Gutierrez A, Bergstrom J, Alvestrand A: Hemodialy- renal failure? Evidence for relationship between malnutri- sis-associated protein catabolism with and without glucose tion, inflammation and atherosclerosis (MIA syndrome). in the dialysis fluid. Kidney Int 46:814-822, 1994 Nephrol Dial Transplant 15:953-960, 2000 17. Lim VS, Kopple JD: Protein metabolism in patients 34. Owen WF, Lew NL,Liu Y, Lowrie EG, Lazarus JM: with chronic renal failure: Role of uremia and dialysis. The urea reduction ratio and serum albumin concentrations Kidney Int 58:1-10, 2000 as predictors of mortality in patients undergoing hemodialy- 18. Pupim L, Flakoll PJ, Ikizler EA: Protein homeostasis sis. N Engl J Med 329:1001-1006, 1993 in chronic hemodialysis patients. Curr Opin Clin Nutr Metab 35. Goldwasser P, Mittman N, Antignani A, et al: Predic- Care 7:89-95, 2004 tors of mortality in hemodialysis patients. J Am Soc Nephrol 19. Lim VS, Bier DM, Flanigan MJ, Sum-Ping ST: The 3:1616-1622, 1993 effect of hemodialysis on protein metabolism. J Clin Invest 36. Avram MM, Mittman N, Bonomini L, Chatto- 91:2429-2436, 1993 padhyay J, Fein P: Markers of survival in dialysis: A 20. Goodship THJ, Mitch WE, Hoerr RA, Wagner DA, seven-year prospective study. Am J Kidney Dis 26:209-219, Steinman TI, Young VR: Adaption to low protein diets in 1995 renal failure: Leucine turnover and nitrogen balance. J Am 37. Bergstrom J: Nutrition and mortality in hemodialysis. Soc Nephrol 1:66-75, 1990 J Am Soc Nephrol 6:1329-1341, 1995 384 BOSSOLA ET AL

38. Qureshi AR, Alvestrand A, Danielsson A, et al: Fac- 55. Kooistra MP, Vos PF, Koomans HA, Vos PF: Daily tors predicting malnutrition in hemodialysis patients: A home hemodialysis: Towards a more physiological treat- cross-sectional study. Kidney Int 53:773-782, 1998 ment of patients with ESRD. Semin Dial 12:424-430, 1999 39. Lowrie EG, Lew NL: Death risk in hemodialysis 56. Spanner E, Suri R, Heidenheim AP, Lindsay RM: The patients: The predictive value of commonly measured vari- impact of quotidian hemodialysis on nutrition. Am J Kidney ables and an evaluation of death rate differences between Dis 42:S30-S35, 2003 (suppl 1) facilities. Am J Kidney Dis 15:458-482, 1990 57. Schulman G: Nutrition in daily hemodialysis. Am J 40. Lowrie EG, Lew NL, Huang WH: Race and diabetes Kidney Dis 41:S112-S115, 2003 (suppl 1) as risk predictors in hemodialysis patients. Kidney Int Suppl 58. Galland R, Traeger J, Arkouke W, Delawari E, Fou- 38:S22-S31, 1992 que D: Short daily hemodialysis and nutritional status. Am J 41. Kalantar-Zadeh K, Block G, McAllister CJ, Hum- Kidney Dis 37:S95-S98, 2001 (suppl 2) phreys MH, Kopple JD: Appetite and inflammation, nutri- 59. Galland R, Traeger J, Arkouche W, Cleaud C, Dela- tion, anemia, and clinical outcome in hemodialysis patients. wari D, Fouque D: Short daily hemodialysis rapidly im- Am J Clin Nutr 80:299-307, 2004 proves nutritional status in hemodialysis patients. Kidney Int 42. Dwyer JT, Larive B, Leung J, et al, for the Hemodialy- 60:1555-1560, 2001 sis Study Group: Nutritional status affects quality of life in 60. Spanner E, Suri R, Heidenheim AP, Lindsay RM: The Hemodialysis (HEMO) Study patients at baseline. J Ren impact of quotidian hemodialysis on nutrition. Am J Kidney Nutr 12:213-223, 2002 Dis 42:S30-S35, 2003 (suppl 1) 43. Johansen KL, Shubert T, Doyle J, Soher B, Sakkas 61. Schulman G: Nutrition in daily hemodialysis. Am J GK, Kent-Braun A: Muscle atrophy in patients receiving Kidney Dis 41:S112-S115, 2003 (suppl 1) hemodialysis: Effects on muscle strength, muscle quality, 62. Blagg CR: Quotidian Dialysis: The London, Ontario and physical function. Kidney Int 63:291-297, 2003 Experience. Am J Kidney Dis 42:S1-S2, 2003 (suppl 1) 44. Spindler A, Paz S, Bernman A: Muscular strength and 63. Pierratos A, Ouwendyk M, Francoeur R, et al: Noctur- bone mineral density in hemodialysis patients. Nephrol Dial nal hemodialysis: Three-year experience. J Am Soc Nephrol Transplant 12:128-132, 1997 9:859-868, 1998 45. Kondrup J, Rasmussen HH, Hamberg O, Stanga Z, 64. McPhatter LL, Lockridge RS, Albert J, et al: Nightly for the Ad Hoc ESPEN Working Group: Nutritional Risk home hemodialysis: Improvement in nutrition and quality of Screening (NRS 2002): A new method based on an analysis life. Adv Ren Replace Ther 6:358-365, 1999 of controlled clinical trials. Clin Nutr 22:321-336, 2003 65. O’Sullivan DA, MsCharty JT, Kumar R, Williams 46. Locatelli F, Fouque D, Heimburger O, et al: Nutri- AW: Improved biochemical variables, nutrient intake, and tional status in dialysis patients: A European consensus. hormonal factors in slow nocturnal hemodialysis: A pilot Nephrol Dial Transplant 17:563-572, 2002 study. Mayo Clin Proc 73:1035-1045, 1998 47. Mittman N, Avram MM, Oo KK, Chattopadhyay J: 66. Bossola M, Muscaritoli M, Panocchia N, et al: Switch Serum prealbumin predicts survival in hemodialysis and from bicarbonate hemodialysis to hemodiafiltration with peritoneal dialysis: 10 Years of prospective observation. online regeneration of the ultrafiltrate (HFR): Effect on Am J Kidney Dis 38:1358-1364, 2001 nutritional and inflammatory status. Artif Organs 29:259- 48. Kuwae N, Kopple JD, Kalantar-Zadeh K: A low 263, 2005 lymphocyte percentage is a predictor of mortality and hospi- 67. Akpele L, Bailey JL: Nutrition counseling impacts talization in hemodialysis patients. Clin Nephrol 63:22-34, serum albumin levels. J Ren Nutr 14:143-148, 2004 2005 68. Leon JB, Majerle AD, Soinski JA, Kushner I, Ohri- 49. Enia G, Sicuso C, Alati G, Zoccali C: Subjective Vachaspati P, Sehgal AR: Can a nutrition intervention im- Global Assessment of nutrition in dialysis patients. Nephrol prove albumin levels among hemodialysis patients? A pilot Dial Transplant 8:1094-1098, 1993 study. J Ren Nutr 11:9-15, 2001 50. Kalantar-Zadeh K, Kleiner M, Dunne E, Lee GH, 69. Inui A: Cancer anorexia-cachexia syndrome: Current Luft FC: A modified quantitative Subjective Global Assess- issues in research and management. CA Cancer J Clin ment of nutrition for dialysis patients. Nephrol Dial Trans- 52:72-91, 2002 plant 14:1732-1738, 1999 70. Burrowes JD, Bluestone PA, Wang J, Pierson RN: 51. Kalantar-Zadeh K, Kopple JD, Block G, Humphreys The effects of moderate doses of megestrol acetate on MH: A malnutrition-inflammation score is correlated with nutritional status and body composition in a hemodialysis morbidity and mortality in maintenance hemodialysis pa- patient. J Ren Nutr 9:89-94, 1999 tients. Am J Kidney Dis 38:1251-1263, 2001 71. Boccanfuso JA, Hutton M, McAllister B: The effects 52. Ishimura E, Okuno S, Kim M, et al: Increasing body of megestrol acetate on nutritional parameters in a dialysis fat mass in the first year of hemodialysis. J Am Soc Nephrol population. J Ren Nutr 10:36-43, 2000 12:1921-1926, 2001 72. Philips ME, Havard J, Howard J: Oral essential 53. Rocco MV, Dwyer JT, Larive B, et al, for the HEMO aminoacids supplementation in patients on maintenance Study Group: The effect of dialysis dose and membrane flux hemodialysis. Clin Nephrol 9:241-248, 1978 on nutritional parameters in hemodialysis patients: Results 73. Acchiardo S, Moore L, Cockrell S: Effect of essential of the HEMO Study. Kidney Int 65:2321-2334, 2004 aminoacids (EAA) on chronic hemodialysis (CHD) patients. 54. Woods JD, Port FK, Orzol S, et al: Clinical and Trans Am Soc Artif Intern Organs 28:608-613, 1982 biochemical correlates of starting daily hemodialysis. Kid- 74. Allman M, Stewart P, Tiller D, Horvath JS, Duggin ney Int 55:2467-2476, 1999 GG, Truswell AS: Energy supplementation and the nutri- THERAPY OF MALNUTRITION IN HEMODIALYSIS 385 tional status of hemodialysis patients. Am J Clin Nutr 90. Capelli JP, Kushner H, Camiscioli TC, Chen SM, 51:558-562, 1990 Torres MA: Effect of intradialytic parenteral nutrition on 75. Allman M, Yau D, Tiller D, et al: The effect of dietary mortality rates in end-stage renal disease care. Am J Kidney glucose polymer supplementation on the nutrition and plasma Dis 23:808-816, 1994 aminoacids of hemodialysis patients. J Ren Nutr 2:59-66, 91. Chertow GM, Ling J, Lew NL, Lazarus JM, Lowrie 1992 EG: The association of intradialytic parenteral nutrition 76. Mastroiacovo P, Pace V, Sagliaschi G: Aminoacids administration with survival in hemodialysis patients. Am J for dialysis patients. Clin Ther 15:698-704, 1993 Kidney Dis 24:912-920, 1994 77. Cuppari L, Medeiros FA, Papini HF, et al: Effective- 92. Hiroshige K, Iwamoto M, Kabashima N, Mutoh Y, ness of oral energy-protein supplementation in severely Yuu K, Ohtani A: Prolonged use of intradialysis parenteral malnourished hemodialysis patients. J Ren Nutr 4:127-135, nutrition in elderly malnourished chronic hemodialysis pa- 1994 tients. Nephrol Dial Transplant 13:2081-2087, 1998 78. Cockram DB, Hensley MK, Rodriguez M, et al: 93. Mortelmans AK, Duym P, Vandenbroucke J, et al: Safety and tolerance of medical nutritional products as sole Intradialytic parenteral nutrition in malnourished hemodialy- sources of nutrition in people on hemodialysis. J Ren Nutr sis patients: A prospective long-term study. J Parenter En- 8:25-33, 1998 teral Nutr 23:90-95, 1999 79. Eustace JA, Coresh J, Kutchey C, et al: Random- 94. Berneis K, Iseli-Scahub J, Garbani E, Meier R, Kiss ized double-blind trial of oral essential amino acids for D: Effects of intradialytic parenteral nutrition in chronic dialysis-associated hypoalbuminemia. Kidney Int 57:2527- hemodialysis patients with malnutrition: A pilot study. Wien 2538, 2000 Klin Wochenschr 12:876-881, 1999 80. Hiroshige K, Sonta T, Suda T, Kanegae K, Ohtani A: 95. Krause I, Shamir R, Davidovits M, et al: Intradialytic Oral supplementation of branched-chain amino acid im- parenteral nutrition in malnourished children treated with proves nutritional status in elderly patients on chronic hemo- hemodialysis. J Ren Nutr 12:55-59, 2002 dialysis. Nephrol Dial Transplant 16:1856-1862, 2001 96. Cherry N, Shalansky K: Efficacy of intradialytic 81. Wilson B, Fernandez-Madrid A, Hayes A, Hermann parenteral nutrition in malnourished hemodialysis patients. K, Smith J, Wassell A: Comparison of the effects of two Am J Health Syst Pharm 15:1736-1741, 2002 97. Czekalski S, Hozejowski R: Intradialytic aminoacids early intervention strategies on the health outcomes of supplementation in hemodialysis patients with malnutrition: malnourished hemodialysis patients. J Ren Nutr 11:166-171, Results of a multicenter cohort study. J Ren Nutr 14:82-88, 2001 2004 82. Sharma M, Rao M, Jacob S, Jacob CK: A controlled 98. Coleman JE, Watson AR, Rance CH, Moore E: Gas- trial of intermittent enteral nutrient supplementation in main- trostomy buttons for nutritional support on chronic dialysis. tenance hemodialysis patients. J Ren Nutr 12:229-237, 2002 Nephrol Dial Transplant 13:2041-2046, 1998 83. Caglar K, Fedje L, Dimmitt R, Hakim RM, Shyr Y, 99. Mak RHK, Pak Y: End organ resistance to growth Ikizler TA: Therapeutic effects of oral nutritional supple- hormone and IGF-I in epiphyseal chondrocyte of rats with mentation during hemodialysis. Kidney Int 62:1054- chronic renal failure. Kidney Int 50:400-406, 1996 1059, 2002 100. Blum WF, Ranke MB, Kietzmann K: Growth hor- 84. Steiber AL, Handu DJ, Cataline DR, Deighton TR, mone resistance and inhibition of somatomedin activity by Weatherspoon LJ: The impact of nutrition intervention on a excess of insulin-like growth factor binding protein in ure- reliable morbidity and mortality indicator: The hemodialysis- mia. Pediatr Nephrol 5:539-544, 1991 prognostic nutrition index. J Ren Nutr 13:186-190, 2003 101. Ziegler TR, Lazarus JM, Young LS, Hakim R, 85. Pupim L, Flakoll PJ, Broillette JR, Levenhagen DK, Wilmore DW: Effects of recombinant human growth hor- Hakim RM, Ikizler TA: Intradialytic parenteral nutrition mone in adults receiving maintenance hemodialysis. J Am improves protein and energy homeostasis in chronic hemodi- Soc Nephrol 52:726-730, 1991 alysis patients. J Clin Invest 110:483-492, 2002 102. Schulman G, Wingard RL, Hutchison RL, Lawrence 86. Pupim L, Flakoll PJ, Ikizler TA: Nutritional supple- P, Hakim RM: The effects of recombinant human growth mentation acutely increases albumin fractional synthetic rate hormone and intradialytic parenteral nutrition in malnour- in chronic hemodialysis patients. J Am Soc Nephrol 15:1920- ished hemodialysis patients. Am J Kidney Dis 21:527-534, 1926, 2004 1993 87. Pupim L, Flakoll PJ, Levenhagen DK, Ikizler TA: 103. Iglesias P, Diez JJ, Fernandez-Reyes MJ, et al: Exercise augments the acute anabolic effects of intradialytic Recombinant human growth hormone therapy in malnour- parenteral nutrition in chronic hemodialysis patients. Am J ished dialysis patients: A randomized controlled study. Am J Physiol Endocrinol Metab 286:E589-E597, 2004 Kidney Dis 32:454-463, 1998 88. Snyder S, Bergen C, Sigler MH, Teehan BP: Intradia- 104. Johansson G, Bengtsson BA, Ahlmen J: Double- lytic parenteral nutrition in chronic hemodialysis patients. blind, placebo-controlled study of growth hormone treat- ASAIO Trans 37:M373-M375, 1991 ment in elderly patients undergoing chronic hemodialysis: 89. Schulman G, Wingard RL, Hutchinson RL, Lawrence Anabolic effect and functional improvement. Am J Kidney P, Hakim RM: The effects of recombinant human growth Dis 33:709-717, 1999 hormone and intradialytic parenteral nutrition in malnour- 105. Hansen TB, Gram J, Jensen PB, et al: Influence of ished hemodialysis patients. Am J Kidney Dis 21:527-534, growth hormone on whole body and regional soft tissue 1993 composition in adult patients on hemodialysis. A double- 386 BOSSOLA ET AL blind, randomized, placebo-controlled study. Clin Nephrol 114. Brady JP, Hasbargen JA: Correction of metabolic 53:99-107, 2000 acidosis and its effect on albumin in chronic hemodialysis 106. Garibotto G, Barreca A, Sofia A, et al: Effects of patients. Am J Kidney Dis 31:35-40, 1998 growth hormone on leptin metabolism and energy expendi- 115. Williams AJ, Dittmer ID, McArley A, Clarke J: ture in hemodialysis patients with protein-calorie malnutri- High bicarbonate dialysate in hemodialysis patients: Effects tion. J Am Soc Nephrol 11:2106-2113, 2000 on acidosis and nutritional status. Nephrol Dial Transplant 107. Kotzmann H, Schmidtt A, Lercher P, et al: One-year 12:2633-2637, 1997 growth hormone therapy improves granulocyte function 116. Blair D, Bigelow C, Sweet SJ: Nutritional effects of without major effects on nutritional and anthropometric delivered bicarbonate dose in maintenance hemodialysis parameters in malnourished hemodialysis patients. Nephron patients. J Ren Nutr 13:205-211, 2003 Clin Pract 93:C75-C82, 2003 117. Movilli E, Zani R, Carli O, et al: Correction of 108. Muscaritoli M, Bossola M, Bellantone R, Rossi metabolic acidosis increases serum albumin concentration Fanelli F: Therapy of muscle wasting in cancer: What is the and decreases kinetically evaluated protein intake in hemodi- alysis patients: A prospective study. Nephrol Dial Transplant future? Curr Opin Clin Nutr Metab Care 7:459-466, 2004 13:1719-1722, 1998 109. Orr R, Fiatarone Singh M: The anabolic androgenic 118. Stein A, Moorhouse J, Iles-Smith H, et al: Role of an steroid oxandrolone in the treatment of wasting and cata- improvement in acid-base status and nutrition in CAPD bolic disorders: Review of efficacy and safety. Drugs 64:725- patients. Kidney Int 52:1089-1095, 1997 750, 2004 119. Szeto CC, Wong TY, Chow KM, Leung CB, Li PK: 110. Foque D, Guebre-Egziabher F, Laville M: Advances Oral sodium bicarbonate for the treatment of metabolic in anabolic interventions for malnourished dialysis patients. acidosis in peritoneal dialysis patients: A randomized placebo- J Ren Nutr 13:161-165, 2003 control trial. J Am Soc Nephrol 14:2119-2126, 2003 111. Gascon A, Belvis JJ, Derisa F, Iglesias E: Nan- 120. Biolo G, Ciocchi B, Bosutti A, Situlin R, Toigo G, drolone decantate is a good alternative for the treatment of Guarnieri G: Pentoxifylline acutely reduces protein catabo- anemia in elderly male patients on hemodialysis. Geriatr lism in chronically uremic patients. Am J Kidney Dis 40: Nephrol Urol 9:67-72, 1999 1162-1172, 2002 112. Johansen KL, Mulligan K, Schambelan M: Anabolic 121. Chang JW, Yang WS, Min WK, Lee SK, Park JS, effects of nandrolone decanoate in patients receiving dialy- Kim SB: Effects of simvastatin on high-sensitivity C- sis. A randomized controlled trial. JAMA 281:1275-1281, reactive protein and serum albumin in hemodialysis patients. 1999 Am J Kidney Dis 39:1213-1217, 2002 113. Pai B, Chretien C, Lau AH: The effects of nan- 122. Stenvinkel P, Andersson P, Wang T: Do ACE-inhibi- drolone decanoate on nutritional parameters in hemodialysis tors suppress tumor necrosis factor-alpha production in ad- patients. Clin Nephrol 58:38-46, 2002 vanced chronic renal failure? J Intern Med 246:503-507, 1999 Multinutrient Oral Supplements and Tube Feeding in Maintenance Dialysis: A Systematic Review and Meta-Analysis

Rebecca J. Stratton, PhD, Gemma Bircher, MSc, Denis Fouque, MD, PhD, Peter Stenvinkel, MD, PhD, Renée de Mutsert, MSc, Meike Engfer, PhD, and Marinos Elia, MD

● Background: This systematic review aims to determine the potential benefits of enteral multinutrient support (oral or tube) in patients with chronic kidney disease (CKD) receiving maintenance dialysis. Methods: Studies of multinutrient oral supplements and enteral tube feeding that involved comparisons of nutritional support versus routine care (ie, usual diet), disease-specific formulae (with adapted macronutrient and micronutrient composition for use in maintenance dialysis patients) versus standard formulae, and enteral tube feeding versus parenteral nutrition are included in this review. The outcome measures sought were clinical (quality of life, complications, and mortality), biochemical (albumin and electrolyte levels), and nutritional (dietary intake and anthropometry). Meta- analyses were performed when possible. Results: This review of 18 studies (5 randomized controlled trials [RCTs], 13 non-RCTs) suggests that enteral nutritional support increased total (energy and protein) intake and increased serum albumin concentration by 0.23 g/dL (2.3 g/L; 95% confidence interval, 0.037 to 0.418 g/dL [0.37 to 4.18 g/L ]; 1 RCT, 2 non-RCTs), with little effect on electrolyte status (serum phosphate and potassium). Few studies reported clinical outcome, and there was insufficient information to compare disease-specific versus standard formulae or enteral versus parenteral nutrition. Conclusion: This systematic review suggests that enteral multinutrient support significantly increases serum albumin concentrations and improves total dietary intake. This may improve clinical outcome, especially in malnourished patients, but insufficient published data exist to examine this. Additional research is required to investigate clinical, economic, and nutritional consequences of using oral supplements and tube feeding (using standard or disease-specific feeds) in patients with CKD receiving maintenance dialysis. Am J Kidney Dis 46:387-405. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Meta-analysis; supplement; enteral; formula; chronic renal failure; dialysis; nutrition; nephrology.

ALNUTRITION IS COMMON in pa- has been shown to reduce mortality,13,14 al- M tients with chronic kidney disease (CKD), though randomized controlled studies are lack- with prevalences ranging from 28% to 48% in ing.3,10,15 Oral supplementation of malnourished predialysis patients1,2 and 9% to 72% in dialysis patients undergoing hemodialysis16-18 and perito- patients.3,4 Protein-energy malnutrition and mi- neal dialysis19,20 may improve nutritional param- cronutrient deficiencies typically are caused by eters. Similarly, enteral tube feeding has been inadequate dietary intake associated with severe used successfully in pilot studies of adults21 and anorexia, nausea and vomiting, impaired taste children.22-24 and smell, and the need for patients to adhere to unpalatable and restricted dietary regimens.3-5 Hormonal derangements, comorbidities, inflam- From the Institute of Human Nutrition, University of mation, inadequate dialysis, poor control of aci- Southampton; Leicester General Hospital, Leicester, UK; dosis, loss of nutrients in dialysate, and long- University Claude Bernard Lyon 1, Lyon, France; Karolin- term dialysis also may adversely affect appetite ska University Hospital at Huddinge, Stockholm, Sweden; and nutritional intake.3,4 Royal Numico, Numico Research and Clinical Nutrition Division, Schiphol, The Netherlands. Malnutrition has been associated with in- Received February 23, 2005; accepted in revised form creased morbidity, greater health care require- April 27, 2005. ments, and reduced functional ability in patients Originally published online as doi:10.1053/j.ajkd.2005.04.036 with CKD.4 Similarly, wasting at the start of on August 1, 2005. dialysis treatment is an independent predictor of This study was conducted using an educational grant 6,7 supplied by Numico. cardiovascular mortality. Improving the nutri- Address reprint requests to Rebecca J. Stratton, PhD, tional status of patients with CKD by providing Institute of Human Nutrition, School of Medicine, University nutritional support is expected to decrease mor- of Southampton, MP 113 Southampton General Hospital, bidity and mortality.8-12 Certainly in other pa- Tremona Rd, Southampton, SO16 6YD UK. E-mail: [email protected] tient groups, the use of nutritional support im- © 2005 by the National Kidney Foundation, Inc. proves recovery and reduces complication rates 0272-6386/05/4603-0002$30.00/0 and mortality.4 Intradialytic parenteral nutrition doi:10.1053/j.ajkd.2005.04.036

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 387-405 387 388 STRATTON ET AL

Expert guidelines for the nutritional treatment QUORUM guidelines.31 A flow chart (Fig 1) shows the of patients with CKD receiving maintenance principle stages and processes undertaken. dialysis advise diets rich in protein of high bio- Identification and Retrieval of Studies logical value, but low in phosphate.25 For the Potentially relevant studies were identified by searching treatment of malnourished patients with CKD, electronic databases. These included PubMed (http:// an increase in dietary energy and protein intake www.ncbi.nlm.nih.gov/entrez/query.fcgi) accessed August 9, often is recommended, but energy and protein- 2004, Cochrane (http://www.nelh.nhs.uk/cochrane.asp) ac- rich diets and standard formulae usually also cessed July 27, 2004, Turning Research Into Practice (www. provide a substantial amount of liquid and micro- tripdatabase.com) accessed August 26, 2004, Clinical Evi- dence (www.clinicalevidence.com) accessed August 26, nutrients (especially electrolytes). Consequently, 2004, National Electronic Library for Health guidelines disease-specific formulae (for use as oral nutri- finder (http://rms.nelh.nhs.uk/guidelinesfinder/) accessed Au- tional supplements or enteral tube feeding) with gust 26, 2004, and National Service Frameworks (http://www. greater energy density (up to 2 kcal/mL); lower dh.gov.uk/PolicyAndGuidance/HealthAndSocialCareTopics/ ϭ potassium, phosphate, and sodium concentra- HealthAndSocialCareArticle/fs/en?CONTENT_ID 4070951& chkϭW3ar/W) accessed August 26, 2004. Search terms tions; and adjusted fat-soluble vitamin levels included chronic renal failure, dialysis, haemodialysis, hemodi- have been developed. Because of the limited alysis, nutrition*, nutrie*, enteral*, oral*, supplement*, sip, micronutrient content of these formulae, they feed, liquid, tube, nasogastric, nasojejunal, nasoduodenal, gas- should be used in addition to patient food intake trostomy, jejunostomy, and formula*. Cross-referencing of bib- to provide extra energy and protein, without liographies and consultation with experts were undertaken to obtain any additional studies. extra uremic toxins, thereby allowing the patient more freedom in their food choices.26 Study Selection Criteria No consensus about the value of standard or Studies were deemed eligible for inclusion in the review if disease-specific multinutrient oral supplements they conformed to predetermined inclusion and exclusion and tube feeding has been reached, partly be- criteria (Table 1). Inclusion criteria specified subjects, inter- cause of a lack of quantitative synthesis of the vention, outcome measures, and study design. Subjects Ͼ evidence. Although a systematic review of nutri- eligible for inclusion were adults ( 18 years) with CKD receiving maintenance dialysis (any technique admissible), tional support for patients with CKD was pub- of any nutritional status (well nourished, malnourished), and 27 lished in 2001, very few studies of enteral based in any setting (eg, hospital, outpatient, or home). nutritional support were included because of the Eligible interventions were feeds that were given enterally, stringent selection criteria used. Furthermore, as either oral nutritional supplements or tube feeding, that this review,27 along with other reviews,8,9,11,12 contained at least 2 macronutrients (carbohydrate, fat, and/or protein), as well as micronutrients (vitamins, minerals, and was descriptive, rather than quantitative. trace elements). The intervention could provide either a Therefore, a systematic review and meta- portion of or the complete daily requirement for energy and analysis of studies of patients with CKD receiv- could be nutritionally complete or incomplete. Studies using ing maintenance dialysis was undertaken with concurrent parenteral nutrition or dietary advice were admis- sible, but those using only parenteral nutrition or only dietary the following aims: (1) to examine the impact of counseling were excluded. The main outcome measures sought enteral nutritional support (multinutrient oral were clinical outcomes (eg, quality of life, complications, and supplements and tube feeding) versus routine mortality). Other outcome data included dietary intake, nutri- care on clinical outcome and nutritional status; tional status (anthropometry), and biochemical endpoints (elec- (2) to investigate whether specifically designed trolyte and albumin levels). No other restrictions were placed on studies with regard to type of comparator (eg, no nutritional multinutrient enteral formulae (named disease- support, dietary advice, or parenteral nutrition), year of publica- specific formulae in this review) are superior to tion, language (providing an English translation of the report or standard multinutrient formulae; and (3) to com- its abstract was available), and source. Randomized controlled pare the efficacy of enteral tube feeding versus trials (RCTs) were the priority, although nonrandomized con- parenteral nutrition in this patient group. trolled clinical trials (CCTs), “before-after” clinical trials (CTs), and observational study designs (eg, cohort or case study) also were admissible, ensuring that all available information was METHODS captured. The review was planned, conducted, and reported fol- lowing published guidelines. These include those issued Data Extraction by the Cochrane Collaboration,28 the UK National Health After identification of potentially relevant studies based Service Centre for Reviews and Dissemination,29,30 and the on titles and abstracts, full reports were obtained and evalu- NUTRITIONAL SUPPORT FOR DIALYSIS PATIENTS 389

Fig 1. Summary of stages and processes in the systematic review of the efficacy of enteral nutritional support for patients with CKD. ated by 1 researcher; a second assessor verified inclusion/ Quality Assessment exclusion decisions. A predetermined data extraction table was designed to capture study characteristics and all outcome data The quality of individual studies was assessed by 1 and allow assimilation of data from differing study designs. researcher and verified by a second assessor by using 2 Outcome measures were recorded based on definitions pro- scales.32,33 The first method was a 6-point scale adapted vided by the original investigators of each study. from the Quality of Evidence Quality Assessment scale

Table 1. Summary of Inclusion and Exclusion Criteria Applied During Evaluation of Studies for Systematic Review

Selection Criterion Inclusion Criteria Exclusion Criteria

Population Adult studies Animal data Nutritional status either well nourished or malnourished Patients undergoing dialysis (of any type) Intervention All studies using oral nutritional supplements and/or Feeds given orally or by tube that contain tube feeding (all routes/methods), including those Ͻ 2 macronutrients (with or without simultaneously using dietary counseling and/or micronutrients) or only micronutrients parenteral nutrition and/or simultaneous standard Dietary counseling only diet Parenteral nutrition only Setting in hospital or community (eg, outpatient, home) Outcome measures Dietary intake Anthropometry Serum albumin level Serum prealbumin level Clinical outcomes (quality of life, mortality, complications) Electrolytes (serum potassium and phosphate) 390 STRATTON ET AL

(Agency for Health Care and Policy Research).32 The high- statistical analyses were conducted using SAS, version 8.2 est grade (score, 1) was given to RCTs, whereas lower (SAS Institute Inc, Cary, NC). All data are presented as grades were given to well-controlled trials without random- mean Ϯ SD unless otherwise stated. ization (score, 2); well-designed cohort or case-control ana- lytical studies (score, 3); multiple time-series or dramatic RESULTS results or uncontrolled experiments (score, 4); opinions of respected authorities based on clinical evidence, descrip- Overall Search Findings tive studies, or reports of expert committees (score, 5); A total of 9,157 studies were identified by the and evidence inadequate owing to problems of method or search strategy (Fig 1). After evaluation of the conflicts of evidence (score, 6). The second assessment system was that used by Jadad et al33 and based on a title and abstract, 46 reports were deemed poten- highest score of 5, achieved by only well-conducted tially relevant and obtained in full. On reading double-blind RCTs with adequate description of dropouts, the full text of these 46 studies, 18 complied with as well as the use of appropriate methods of randomiza- inclusion criteria and a total of 541 patients were tion and blinding. included in the review (Table 2).16,18-21,38-50 The other 28 studies were rejected because of not Data Synthesis and Statistical Methods being an original experimental study, not using a After data extraction, when appropriate and feasible, nutritional intervention, using an ineligible nutri- results of similar groups of trials were combined and meta- tional intervention, using ineligible subjects, or it analysis was undertaken on relevant outcome measures. The was not possible to source the document or an main comparisons to be addressed in the analysis were multinutrient oral supplements or enteral tube feeding ver- English translation of it. sus routine care, disease-specific formulae versus standard formulae, and enteral tube feeding versus parenteral nutri- Description of Included Studies tion. Separate analyses were intended for studies of differing Of 18 studies included in the review, 5 were duration (short-term versus long-term feeding) and accord- RCTs20,21,39,41,45, scoring the highest grade of 1 ing to nutritional status of patients (malnourished versus 32 well nourished). according to the quality-of-evidence scale. Hedges’ unbiased estimator of the standardized mean However, methods of individual RCTs often were difference for relevant treatment comparisons was calcu- poorly described (with regard to methods of lated.34 The mean treatment difference was considered statis- randomization, blinding, and recording number tically significant if the 95% confidence interval did not span of dropouts). Therefore, none attained the high- the value 0. Heterogeneity was investigated from the Q test 33 35 est score of 5 on the Jadad method scale ; most of heterogeneity derived by the Mantel-Haenszel method. 21,39 20,41,45 Because of the small number of studies included in the scored 3 (n ϭ 2 )or2(nϭ 3 ). Thir- meta-analysis, it was deemed inappropriate to investigate teen trials were nonrandomized, with 6 publication bias through the use of funnel plots.36 A fixed- CCTs16,19,38,43,48,50 scoring 2, 1 cohort study effects model was used to combine treatment estimates, which scoring 3,44 and 6 before-after clinical trials assumes no heterogeneity between study results. Meta-analysis scoring 418,40,42,47,49 on the quality-of-evidence estimate of treatment-effect size was calculated as a weighted 32 sum of the effect size for each study, in which weight was scale. calculated as the reciprocal of Hedges’ estimated variance of Length of follow-up generally was long term the effect size for each individual study. (1 to 11 months; Table 2), with 1 study lasting 14 All comparisons were interventions versus comparator. days and 1 study incorporating single-meal data. Selection of data used for such analyses was dependent on whether baseline data were available for all comparisons; The populations studied consisted mainly of pa- when these were available, the analysis was conducted by tients undergoing hemodialysis (n ϭ 14 studies), comparing change from baseline measurements in the inter- with fewer studies of patients undergoing perito- vention versus comparator arms. When not provided by neal dialysis (n ϭ 4). individual investigators, changes from baseline data were There were 14 studies comparing nutritional calculated by subtracting baseline from postintervention 37 support with routine care (Table 2), including 2 values, and the SD of the difference was calculated. If1or 20,21 16,19,39,43,50 18,40,42,47,48 more studies in the meta-analysis failed to report baseline RCTs, 5 CCTs, 5CTs data, postintervention data were compared. The correlation using oral nutritional supplements, and 1 CT49 between baseline and postintervention data in individual and 1 cohort study44 that involved enteral tube ϭ trials was assumed to be 0 (r 0), which is the most feeding. Of these, 4 trials were published only as conservative method of analysis because it assumes no abstracts.20,38,47,50 Two RCTs compared disease- correlation between groups at baseline. The sensitivity of 21,41 this assumption was tested by additionally conducting analy- specific with standard formulae (Table 2); 1 ses using the less conservative assumption of r ϭ 0.5. All of these21 provided data for comparison of nutri- URTOA UPR O ILSSPTET 391 PATIENTS DIALYSIS FOR SUPPORT NUTRITIONAL

Table 2. Studies Included in the Review After Evaluation of the Full Text

Population Length of Reference Baseline Status Description Intervention Regimen Route Intervention Main Outcome Measures Studies of nutritional support v routine care (oral nutritional supplements) RCTs Han et al,20 1997 Group A: mean BW, 52.4 23 Peritoneal Standard supplement (800 Supplement to provide Oral 3 mo Anthropometry (BW, BMI, body RCT, parallel kg; BMI, 18.7 kg/m2 dialysis kcal/d, 12 En% protein), 800 kcal/d in fat, lean mass, triceps groups Group B: not stated patients n ϭ 12 addition to usual skinfold, arm muscle area, Abstract only Routine care (usual diet), diet midarm circumference, body n ϭ 11 muscle), dietary intake Sharma et al,21 Group A: dialysis duration, 40 Nondiabetic Disease-specific Supplement Oral 1 mo Serum biochemistry (albumin, 2002 2 mo; mean age, 29.6 y; adult supplement (Renocare administered urea, K, P), nPCR, dietary RCT, parallel BW, 47.3 kg; BMI, 17.2 maintenance II [Criticare Ltd, Mumbai, postdialysis intake, anthropometry (BW, groups kg/m2; serum albumin, hemodialysis India], 2 sachets ϭ 170 BMI), appetite score, 3.4 g/dL; urea, 1,010 mg/ patients mL ϭ 500 kcal, 14 En% Karnofsky Index L; nPNA, 0.82 g/kg protein, 187.5 mg Na, Group B: dialysis duration, 4.8 mEq K, 172 mg P) 2.25 mo; mean age, plus dietary counseling, 32.7 y; BW, 48.7 kg; BMI, n ϭ 10 17.9 kg/m2; serum Standard homemade albumin, 3.4 g/dL; urea, supplement (150 mL ϭ 1,090 mg/L; nPNA, 0.85 497 kcal, 13 En% g/kg protein, 83.2 mg Na, 4.5 Group C: dialysis duration, mEq K, 181.3 mg P) 1.5 mo; mean age, plus dietary counseling, 31.9 y; BW, 48.5 kg; BMI, n ϭ 16 17.1 kg/m2; serum Routine care (usual diet, albumin, 3.4 g/dL; urea, 1.2 g protein/kg BW/d 1,130 mg/L; nPNA, 0.84 and 35-45 kcal/kg BW/d, g/kg 60 mEq K/L, fluid restricted to 500-1,500 mL, 2-4 g/d Na, 1 g/d P plus dietary counseling), n ϭ 14 (Continued) 392

Table 2 (Cont’d). Studies Included in the Review After Evaluation of the Full Text

Population Length of Reference Baseline Status Description Intervention Regimen Route Intervention Main Outcome Measures Non-RCTs Acchiardo & Group A v group B: mean 108 Hemodialysis Standard supplement 1 can/d, no further Oral Not stated Dietary intake, serum Riley,38 2002 age, 54.5 v 54.2 y; BW, patients (Ensure Plus [Abbott information provided biochemistry (albumin), CCT, parallel 70.2 v 79.8 kg; serum Laboratories, nPCR, mortality groups albumin, 3.49 v 4.0 g/dL; Columbus, OH], no Abstract only nPCR, 0.84 v 1.05 g/kg/ composition provided), n d; time on dialysis, 110 v ϭ 54 120 mo; Kt/V, 1.37 v Routine care (usual diet), 1.30; creatinine, 11.3 v n ϭ 54 12.2 mg/dL Included some patients with diabetes Beutler et al,16 Group A/B: mean time on 11 Hemodialysis Disease-specific Consume 1 or 2 100-g Oral 4 mo Dietary intake, anthropometry 1997 dialysis, 39/52 mo; age, patients supplement (ReNeph servings/d either as (BW, lean mass, triceps CCT, parallel 66.3/66.8 y; albumin, 3.2 [Nutrabalance, frozen dessert or skinfold), serum groups g/dL/32 g/L; PCR, 0.75/ Indianapolis, IN]; 100 g thawed to liquid biochemistry (albumin, 0.77 g/kg/d; Kt/V, 7.95/ ϭ 240 kcal, 13 En% consistency, in creatinine, K, P), Kt/V 6.56 protein, 25 mg K, 36 mg addition to usual No BMI provided P, 75 mg Na) plus diet Muscle wasting (triceps dietary counseling, n ϭ skinfold Ͻ 5th percentile) 6 n ϭ 3 group A v n ϭ 2 Routine care (usual diet group B plus dietary counseling), All patients 81%-154% IBW n ϭ 5 Heaf et al,43 Group A: dialysis duration, 42 Peritoneal Standard supplement 2 Portions/d (total, 400 Oral 10 wk Dietary intake, anthropometry 1999 28.9 mo; BW, 65.2 kg; dialysis (Fortimel [Nutricia, The kcal) (BW, body fat), Kt/V, serum CCT, parallel Kt/V, 1.8 patients Netherlands], 400 kcal/ biochemistry (albumin, groups Group B: not stated d, 40 En% protein, 400 prealbumin, creatinine, All subjects previously mg P), n ϭ 12 transferrin, urea, protein,

prescribed protein-rich Routine care (usual diet), lipoproteins, nPNA), AL ET STRATTON P-depleted diet n ϭ 30 symptom score Included some patients with heart disease, diabetes, SLE, COPD URTOA UPR O ILSSPTET 393 PATIENTS DIALYSIS FOR SUPPORT NUTRITIONAL

Patel & Group A/B: mean age, 66/ 22 Peritoneal Standard supplements 1 ϫ 200-mL carton/d Oral 8 wk Dietary intake Raftery,19 52 y; BMI, 22/26 kg/m2; dialysis (Protein Forte 1997 serum albumin, 3.3/3.2 patients [Fresenius Kabi, CCT, parallel g/dL; Kt/V, 1.96/2.05; Runcorn, UK], Ensure groups nPCR, 0.9/0.8 g/kg/d; Plus [Abbott dialysis duration, 29/34 Laboratories, mo Maidenhead, UK], 200 Significantly different age mL ϭ 200-300 kcal, 17- and BMI at baseline 40 En% protein, 200- 212 mg P, 300-367 mg K, 179-239 mg Na), n ϭ 10 Routine care (usual diet), n ϭ 12 Shah et al, 50 Mean age, 54.5 y 88 Hemodialysis Disease-specific 1 can each day of Oral 3 mo Serum biochemistry (albumin), 1999 Serum albumin: group A, patients supplement (Nepro dialysis (3 times/wk) quality of life CCT, parallel 3.68 g/dL; group B, 3.93 [Abbott Laboratories, groups g/dL Columbus, OH], for Abstract only Baseline albumin levels composition, see Caglar significantly different et al), n ϭ 44 Routine care (usual diet), n ϭ 44 Caglar et al,40 Mean age, 62.3 y; BMI, 39 Long-term Disease-specific One 237-mL can Oral 6 mo Serum biochemistry (albumin, 2002 25.8 kg/m2; SGA, 4.94; hemodialysis supplement (Nepro during each dialysis prealbumin, transferrin), CT, single group, serum albumin, 3.33 patients [Abbott Laboratories, session dietary intake, before-after g/dL; 46.3% had Columbus, OH], 237 mL anthropometry (BW, BMI), diabetes mellitus ϭ 475 kcal, 14 En% SGA protein, 42 En% fat, 44 En% carbohydrate), n ϭ 39 Cuppari et al,18 Mean age, 40 y; BW, 48.9 10 Uremic Standard supplement 2 Packages/d diluted Oral 90 d Anthropometry (BW, BMI, body 1997 kg, BMI, 17.1kg/m2; time hemodialysis (Liotecnica [Sao Paulo, in 300-400 mL water fat, triceps skinfold, arm CT, single group, on dialysis, 39.4 mo; patients Brazil], 85-g package muscle area, midarm before-after serum albumin, 3.32 g/ contains 400 kcal, circumference, lean mass), dL; BUN, 77 mg/dL; 10 En% protein, dietary intake, Kt/V, creatinine, 11.4 mg/dL; 35 En% fat, 55 En% hematology and serum triceps skinfold, 6.5 mm carbohydrate, 5 mg biochemistry (albumin, zinc) plus usual diet (35 creatinine, urea, kcal/kg/d, including 1.2 lymphocytes, K, P, g/kg/d protein), n ϭ 10 hemoglobin, protein, cholesterol, triglycerides, Ca)

(Continued) 394

Table 2 (Cont’d). Studies Included in the Review After Evaluation of the Full Text

Population Length of Reference Baseline Status Description Intervention Regimen Route Intervention Main Outcome Measures

Mareckova et Mean serum albumin, 2.5 8 Hemodialysis Disease-specific To provide 20% of Oral 3 wk Dietary intake, serum al,47 2000 g/dL; Whitehead patients supplement (Nutrilac energy biochemistry (albumin, K, CT, single group, quotient, 1.8; BMI, 21.8 Renal, 1,000 kcal, 40 g requirements (400 urea, amino acids, vitamin before-after kg/m2 protein/L), n ϭ 8 mL/d) A), Whitehead quotient English abstract only Dare et al,42 Hemodialysis patients: 3 Hemodialysis, 2 Disease-specific Supplement to replace Oral 3 mo Serum biochemistry (albumin, 1997 mean serum albumin, 4.1 peritoneal supplement (Renamil cows milk in diet bicarbonate, creatinine, K, P, CT, single group, g/dL; creatinine, 1,127 dialysis [Kimal Scientific Hemodialysis patients urea, total protein, before-after ␮mol/L; urea, 28.7 patients Products Ltd, UK]; 100 also given the cholesterol, triglycerides, mmol/L mL ϭ 200 kcal, 4 En% supplement in liver enzymes, Ca, Na), Peritoneal dialysis patients: protein, 36 En% fat, 60 addition to food medication requirements mean serum albumin, En% carbohydrate, 12 3.64 g/dL; creatinine, 930 mg K, 10 mg P, 200 mg ␮mol/L; urea, 20.75 Ca, 40 mg Na), n ϭ 5 mmol/L Predialysis patients All patients 86%-109% IBW excluded Patel et al,48 Dialysis duration, 28.5 mo; 17 Stable Standard supplement Individualized: liquid Oral 2 mo Dietary intake , serum 2000 mean age, 61.2 y; BMI of hemodialysis (Ensure Plus [Abbott and dessert biochemistry (albumin, CT, single group, all patients, Ͼ20 kg/m2; patients Laboratories, supplements added bicarbonate, creatinine, before-after Kt/V, 1.14; serum Maidenhead, UK], to diet to produce urea), nPCR, anthropometry albumin, 4.2 g/dL; nPCR, Fortipudding [Nutricia total dietary protein (BW, BMI), Kt/V 0.95 g/kg/d; urea, 22 Clinical Care, intake of 1.2 g/kg mmol/L Trowbridge, UK], BW/d Protein Forte [Fresenius Kabi], 17-40 En% protein), n ϭ 17 AL ET STRATTON URTOA UPR O ILSSPTET 395 PATIENTS DIALYSIS FOR SUPPORT NUTRITIONAL

Studies of nutritional support v routine care (enteral tube feeding) All non-RCTs Sayce et al,49 Median time on dialysis, 62 8 Hemodialysis Standard and disease- Home gastrostomy, as Tube 3 mo Anthropometry (BW, BMI, 2000 mo; age, 68 y; BW, 43 patients specific supplements required, some triceps skinfold, mid–upper CT, single group, kg; BMI, 16.4 kg/m2; (variety, Nepro [Abbott pump, some bolus arm and arm muscle before-after serum albumin, 2.95 g/ Laboratories, circumferences), serum dL; triceps skinfold, 7.3 Maidenhead, UK], biochemistry (albumin) mm TwoCal HN [Abbott Multiple comorbidities Laboratories, Maidenhead, UK], Ensure Plus [Abbott Laboratories, Maidenhead, UK], Entera Fibre Plus [Fresenius Kabi], composition not provided), providing 474-1,722 kcal/d), n ϭ 8 Holley & Kirk,44 Median BW, 73.2 kg; serum 10 Long-term Disease-specific formulae Provided some or all Tube 11 mo Serum biochemistry (albumin, 2002 albumin, 2.8 g/dL; hemodialysis (1,800-2,000 kcal/L, 14- energy creatinine, K, P, urea Cohort, creatinine, 5.1 mg/dL; patients 16 En% protein, 1,057 requirements Route: reduction rate, Ca, intact retrospective urea reduction rate, 74%; mg/L K, 600-800 mg/L PEG or NG parathyroid hormone), BW, and Kt/V, 1.6; nPCR, 1.4 P, 1,373 mg/L Ca), n ϭ mortality prospective, g/kg; 2 Patients already 9 single group on enteral formulae at One patient received study start standard formula (1,800-2,000 kcal/L, 13- 14 En% protein, 1,060 mg/L P, 1,820 mg/L K, 1,056 mg/L Ca) (Continued) 396

Table 2 (Cont’d). Studies Included in the Review After Evaluation of the Full Text

Population Length of Reference Baseline Status Description Intervention Regimen Route Intervention Main Outcome Measures Studies of disease-specific v standard formulae (oral nutritional supplements) RCTs Cockram et al,41 Group A: mean age, 53 y; 79 Patients on Standard formula Sole source of Oral 14 d Serum biochemistry (albumin, 1998 time on dialysis, 8 y; hemodialysis (Magnacal [Sherwood nutrition (4.2-5 cans/ urea, K, P, transthyretin, Ca, RCT, single serum albumin, 3.9 g/dL; Medical, St. Louis, MO], d) Ca-P, Na, carbon dioxide, blind, parallel nPCR, 1.09 g/kg/d, BUN 1 can ϭ 475 kcal, 14 triglycerides, cholesterol), groups predialysis, 63 mg/dL En% protein, 36 En% nPCR, gastrointestinal Group B: mean age, 53 y; fat, 50 En% complications, feces quality time on dialysis, 6 y; carbohydrate, 237 mg and output serum albumin, 4.0 g/dL; P, 296 mg K), n ϭ 27 nPCR, 0.99 g/kg/d; BUN Disease-specific formula predialysis, 60 mg/dL (Nepro [Abbott Group C: mean age, 46 y; Laboratories, time on dialysis, 7 y; Columbus, OH], 1 can ϭ serum albumin, 4.0 g/dL; 475 kcal, 14 En% nPCR, 1.02 g/kg/d; BUN protein, 43 En% fat, 43 predialysis, 56 mg/dL En% carbohydrate, 163 All subjects had at least 1 mg P, 165 mg K), n ϭ chronic disease n ϭ 17 26 patients with diabetes Disease-specific formula (Nepro [Abbott Laboratories, Columbus, OH], 1 can ϭ 475 kcal, 14 En% protein, 42 En% fat, 44 En% carbohydrate [including 3.7 g FOS], 750 IU beta-carotene, 165 mg P, 250 mg K), n AL ET STRATTON ϭ 26 Sharma et al,21 2002 see above URTOA UPR O ILSSPTET 397 PATIENTS DIALYSIS FOR SUPPORT NUTRITIONAL

Other studies Boudville et al,39 Mean age, 58.4 y; BMI, 13 Peritoneal Oral supplement (Nepro Patients did 3 Oral Single meal Dietary intake 2003 27.5 kg/m2; time on dialysis [Abbott Laboratories, separate visits; no RCT, single dialysis, 2.3 y; Kt/V, 2.56; patients Columbus, OH]; 237 mL alcohol night before, blind, creatinine clearance, ϭ 475 kcal, 14 En% normal breakfast, crossover 73.8 L/wk; nPCR, 0.86 protein, 42 En% fat, 44 then no food until g/kg/d; serum albumin, En% carbohydrate) on visit; took 3.48 g/dL; midarm test meal (self-select supplement or circumference, 29.5 cm; buffet) intake 30 min placebo, then at triceps skinfold, 14.1 later, n ϭ 12 either 30 or 120 min, mm; midarm muscle Oral supplement (Nepro had test lunch, circumference, 25.1 cm; [Abbott Laboratories, which was weighed 46% malnourished Columbus, OH]; 237mL (SGA), 39% ϭ 475 kcal, 14 En% malnourished (albumin Ͻ protein, 42 En% fat, 44 0.35 g/dL). Various En% carbohydrate) on concurrent illness, test meal (self-select including diabetes, heart buffet) intake 120 min disease later, n ϭ 12 Placebo (calorie-free cordial) on test meal (self-select buffet) intake 30 min later, n ϭ 13 Kuhlmann et Malnourished according to 8 Hemodialysis Disease-specific Supplement to usual Oral 3 mo Serum biochemistry (albumin, al,45 1997 SGA classification B or C patients supplement (to increase diet prealbumin), dietary intake, RCT, parallel and serum biochemistry intake by 25%, BW groups (2 low of albumin, composition not Abstract only prealbumin, transferrin, provided), n ϭ 4 cholesterol) Disease-specific No further information supplement (to increase provided intake by 10%, composition not provided), n ϭ 4 (Continued) 398

Table 2 (Cont’d). Studies Included in the Review After Evaluation of the Full Text

Population Length of Reference Baseline Status Description Intervention Regimen Route Intervention Main Outcome Measures

Kuhlmann et Group A: mean age, 61.6 y; 18 Maintenance Disease-specific Tailored to meet Oral 3 mo Serum biochemistry (albumin, al,46 1999 BW, 52.9 kg; BMI, 17.6 hemodialysis supplements individual prealbumin), dietary intake, CCT, parallel kg/m2; serum albumin, patients (combination of Renamil requirements BW groups 4.14 g/dL and Renapro [Renacare Group B: mean age, 55.7 y; Nephromed Bartz BW, 64.1 kg; BMI, 21.7 GmbH, Germany]) plus kg/m2; serum albumin, usual diet (overall high- 4.08 g/dL protein: target, 1.5 g Group C: mean age, 63 y; protein/kg/d and 45 kcal/ BW, 60.6 kg; BMI, 22.2 kg/d), n ϭ 5 kg/m2; serum albumin, Disease-specific 4.19 g/dL supplements (combination of Renamil and Renapro [Renacare]) plus usual diet (overall standard protein: target, 1.2 g protein/kg/d and 35 kcal/ kg/d), n ϭ 5 Spontaneous intake supplemented with 10% of mean protein and energy intake, n ϭ 8

NOTE. To convert albumin in g/dL to g/L, multiply by 10; blood urea nitrogen in mg/dL to mmol/L, multiply by 0.357; creatinine in mg/dL to ␮mol/L, multiply by 88.4. Abbreviations: BMI, body mass index; P, phosphate; K, potassium, Ca, calcium; Na, sodium; BW, body weight; IBW, ideal body weight; PCR, protein catabolic rate; nPCR, normalized PCR; SLE, systemic lupus erythematosus; COPD, chronic obstructive pulmonary disease; SGA, Subjective Global Assessment; BUN, blood urea nitrogen; NG, AL ET STRATTON nasogastric; PEG, percutaneous endoscopic gastrostomy; FOS, fructooligosaccharides; nPNA, normalized protein nitrogen appearance; En%, percent energy from. NUTRITIONAL SUPPORT FOR DIALYSIS PATIENTS 399 tional support versus routine care, as well as review. One small study (n ϭ 8) of home gastros- disease-specific versus standard formulae. tomy feeding suggested that complications were Three studies (2 RCTs39,45 and 1 CCT46) were few and the average number of hospital admis- identified that incorporated comparisons of differ- sions was 1.59/y; however, no control group was ent enteral formulations, including both standard provided for comparison.49 and disease-specific formulae. No comparison of Total nutritional intake. There were insuffi- enteral tube feeding with parenteral nutrition was cient comparable data to allow meta-analysis of identified. the effect of nutritional support versus routine Although separate analyses of studies accord- care on total energy or protein intake. Two RCTs, ing to the duration of nutritional support were including 1 abstract report of malnourished peri- planned, there were insufficient data to allow this toneal dialysis patients20 and 1 of hemodialysis comparison. In addition, nutritional status of patients,21 showed marked increases in total (food patients in the majority of studies was unclear. plus supplement) energy and protein intakes with 1 to 3 months of oral nutritional supplementa- Outcomes tion, with little change in voluntary food intake Nutritional support (standard and disease or appetite. These findings, confirmed by results specific) versus routine care. There were 14 of other studies, suggest that total energy in- studies comparing nutritional support with takes18,19,47,48 and total protein intakes18,19,48 routine care (Table 2), including 2 typically can be increased by 20% to 50% with RCTs,20,21 5 CCTs,16,19,38,43,50 and 5 before- nutritional support. Similarly, in a randomized after CTs18,40,42,47,48 using oral nutritional supple- single-meal study of peritoneal dialysis patients, ments, in addition to 1 before-after CT55 and 1 administration of an oral nutritional supplement, cohort study44 that used enteral tube feeding. either 30 minutes or 2 hours before a self- The majority compared disease-specific formu- selected lunch, did not significantly reduce volun- lae with routine care (n ϭ 816,21,40,42,44,47,49,50). tary intake compared with a placebo and signifi- Clinical outcomes. There were insufficient cantly increased short-term total energy (85%) comparable data to allow meta-analysis of and protein (36%) intakes.39 quality of life, mortality, or complications. In Studies of tube feeding44,49 did not assess total the 1 RCT that assessed clinical outcome,21 1 energy or protein intakes or the impact of tube month of oral supplementation significantly feeding on oral intake. Furthermore, no study increased functional status (Karnofsky index) included in the review assessed intake of other during the course of the study (8.0 to 8.4 in nutrients (eg, vitamins or minerals), although in supplemented versus 8.1 to 8.0 in control some studies, patients also were administered group). In an abstract report of a CCT,50 there supplements of water-soluble vitamins.16 was no significant difference in change in Anthropometry. There were insufficient com- Kidney Disease Quality of Life Questionnaire parable data to allow meta-analysis of the effect scores between patients administered an oral of nutritional support versus routine care on nutritional supplement and those who received body weight. One abstract report of an RCT20 routine care, although the physical domain documented significant increases in weight (ϩ1.8 increased significantly in the intervention kg; 3.3%), muscle, and fat mass after 3 months group. An abstract report of 1 CCT38 sug- of an oral nutritional supplement. Similar changes gested that annual mortality rates in patients were not seen in the control group, but no statis- administered oral nutritional supplements were tical comparisons between groups were pre- not significantly different from those of the sented. One non-RCT18 of 4 months’ duration control group given routine care. No study reported significant increases in weight (ϩ1.5 reported comprehensive data on infective com- kg; 3%) and fat mass, but no changes in muscle plications, and there were insufficient data on mass were observed in this uncontrolled study. the effect of nutritional support on the use of One CT of malnourished hemodialysis patients medication, including phosphate binders. showed significant increases in dry weight by There were no quality of life data in the 12% (ϩ5.3 kg) after 3 months’ intervention with studies of enteral tube feeding included in the enteral tube feeding.49 Other studies reported no 400 STRATTON ET AL

Fig 2. The effect of enteral nutritional support on serum albumin concentration: a meta-analysis.16,21,50 Abbrevia- tion: ONS, oral nutritional supplement. significant changes in weight or measures of g/dL [34 to 39 g/L]) after 1 month of oral muscle or fat mass with oral supplementation, supplementation and were significantly greater including 1 RCT21 and 4 non-RCTs,16,40,43,48 than those in patients given routine care (from although most studies lacked a control arm. After 3.4 to 3.5 g/dL [34 to 35 g/L]; Table 321). Most 3 months of enteral tube feeding (4 different other studies reported increases in serum albu- formulae, mainly disease-specific) in malnour- min levels after nutritional intervention (vary- ished hemodialysis patients, 1 CT showed an ing from 0.05 to 0.5 g/dL [0.5 to 5 g/L]), overall improvement in nutritional status mea- but were not meta-analyzable because of in- sured as significant increases in mid–upper arm complete data presentation43,44,49 (including circumference, mid–upper arm muscle circumfer- 2 abstracts38,47) or incompatible study de- ence, and triceps skinfold thickness compared 18,40-42,45,46,48 49 sign. Among these, 2 small un- with baseline. controlled studies using enteral tube feeding Biochemical outcomes: Serum albumin, potas- indicated that significant improvements (ϩ0.5 sium, and phosphate. A meta-analysis of com- g/dL [5 g/L]) in serum albumin levels were parable data from 1 RCT21 and 2 CCTs,16,50 seen after 349 and 8 months44 of feeding. including 1 abstract,50 showed that compared One CT reported a significant increase of 18% with routine care, nutritional support (using dis- in serum prealbumin levels after nutritional sup- ease-specific formulae) resulted in significantly 40 greater serum albumin concentrations (0.227 g/dL port compared with routine care, whereas 1 CCT reported no significant effect compared [2.27 g/L]; 95% confidence interval, 0.037 to 43 0.418 g/dL [0.37 to 4.18 g/L]; Fig 2; Table 3). with routine care. This analysis was conducted including all 3 Generally, there was little change in electro- studies (n ϭ 123) and using the conservative lyte concentrations, including serum potas- assumption that there was no correlation be- sium and phosphate. When a low-electrolyte tween groups at baseline (r ϭ 0). By relaxing the formula was used as the main source of nutri- correlation to r ϭ 0.5, an additional increase in tion,41,44,49 this occasionally was associated this outcome was seen. with problems of hypokalemia, hyponatremia, Individually, in the 1 RCT, albumin concen- and hypophosphatemia.44,49 However, when trations significantly increased (from 3.4 to 3.9 used as supplementary feeding (1 to 2 servings/ NUTRITIONAL SUPPORT FOR DIALYSIS PATIENTS 401

d), no adverse effects on electrolyte status were reported.16,42,47

Baseline Disease-Specific Versus Standard Formulae Change From Only 2 studies compared disease-specific with standard formulae; both were RCTs using oral nutritional supplements21,41 (Table 2). However, the definition of disease-specific formula was inconsistent, and although manufacturers labeled their products as such, the compositions varied greatly (Table 2). In the study of Sharma et al,21 composition of disease-specific and standard for- mulae apparently were very close, although it is possible that variation existed in other nutrients

Baseline Postintervention (possibly vitamins), which could have supported

Mean SD Mean SD Mean SD the manufacturer’s definition. Clinical outcomes. There were insufficient comparable data to allow meta-analysis of qual- ity of life, mortality, or complications. In an RCT

No. of 21 Patients of hemodialysis patients, there was a signifi- cant increase in Karnofsky Index scores in both the disease-specific and standard-supplement groups compared with respective baseline values (8.0 to 8.4 in the disease-specific group versus 8.3 to 8.7 in the standard supplement group), but no difference between groups. The only other clinical outcome reported was the proportion of hemodialysis patients with gastrointestinal com- plications (eg, diarrhea, constipation) after 14 days of an oral nutritional supplement as the sole Disease-specific oral supplement 44 36.8 3.3 37.5 4 NR NR Disease-specific oral supplement 6 32 0.8* 33.2 0.8* NR NR Disease-specific oral supplement 10 34 4 39source 3 of nutrition, NR NR which was similar in patients Table 3. Data Used in Meta-Analyses receiving disease-specific (6 of 26 patients) and standard (8 of 27 patients) formulae in an RCT.41 Total nutritional intake. There were insuffi- cient data to allow meta-analysis of disease- CCT Routine care 44 39.3 3.4 38.1 3.7 NR NR CCT Routine care 5 32 0.7* 31.6 6.7* NR NR RCT Routine carespecific 14 34 versus 5 standard 35 formulae 3 on NR total NR (food plus supplement) energy or protein intake. In the RCT of hemodialysis patients,21 there was no 2002

1997 significant difference in energy or protein intakes 21 1999 16 in patients receiving disease-specific versus stan- 50 dard oral nutritional supplements during dialysis sessions. Anthropometry. There were insufficient com- Shah et al, Beutler et al, parable data to allow meta-analysis of the effect of disease-specific versus standard formulae on anthropometric outcomes. Only 1 RCT using oral nutritional supplements for 1 month in hemo- dialysis patients reported body weight.21 This study reported a significant increase in body Parameter Reference Study Typeweight Intervention (from 48.7 to 50.5 kg) in patients receiv- NOTE. To convert albumin in g/LAbbreviation: to NR, g/dL, not divide reported. by 10. *SEM.

Serum albumin (g/L) Sharma et al, ing a standard formula, whereas those receiving 402 STRATTON ET AL a disease-specific formula experienced little ings that a decrease in serum albumin concentra- change in weight (47.3 to 47.7 kg).21 tion from 4.0 to 3.5 g/dL (40 to 35 g/L) doubles Biochemical outcomes: Serum albumin, potas- mortality risk, and relative death risk is 5 times sium, and phosphate. Both RCTs reported bio- greater if the albumin concentration decreases to chemical endpoints.21,41 After 1 month of an oral 3.0 to 3.5 g/dL (30 to 35 g/L).54-56 Therefore, an nutritional supplement, serum albumin concentra- increase in serum albumin level of 0.2 to 0.3 tions significantly increased in both the disease- g/dL (2 to 3 g/L), as seen here, may have prognos- specific (from 3.4 to 3.9 g/dL [34 to 39 g/L]) and tic relevance in patients with CKD. standard (from 3.4 to 4.0 g/dL [34 to 40 g/L]) A recent survey of approximately 1,300 hemo- groups.21 Use of an oral supplement as the sole dialysis patients showed that nutritional status is source of nutrition for 14 days showed increased associated strongly with health-related quality of albumin concentrations, although these failed to life, even after controlling for comorbidities and reach significance.41 Meta-analysis of albumin dialysis dose.57 Similarly, because malnutrition data for these 2 studies was not conducted be- and wasting in dialysis patients are associated cause of differences in study design; 1 study with increased morbidity and mortality,4,6,7 im- compared a formula (electrolyte concentrations provement in body function and clinical outcome not reduced) with a homemade feed,21 whereas might be expected to occur from alleviating the other compared a formula with adapted micro- malnutrition. nutrient concentrations with a standard commer- In the reviewed studies, increased total energy cial formula.41 and protein intake were reported in both RCTs and non-RCTs of oral nutritional supplements, Enteral Tube Feeding Versus Parenteral with apparently little suppression of voluntary Nutrition intake. Nevertheless, only a few studies reported No study comparing enteral with parenteral a significant improvement in nutritional status nutritional support was identified. (body weight, fat, and/or muscle mass18,20,49). Optimal timing of oral supplementation in dialy- DISCUSSION sis patients also is unknown, although 1 RCT This systematic review suggests that multinu- reported no difference between provision of the trient enteral support may improve dietary intake feed 0.5 or 2 hours before mealtime.39 and serum albumin concentrations. There were The effect of nutritional support on clinical insufficient data to assess effects on clinical outcomes was difficult to evaluate in the current outcome. review because these outcomes were reported Meta-analysis of 1 RCT and 2 CCTs showed a infrequently. Mortality was only recorded in 2 significant increase of 0.23 g/dL (23 g/L) in non-RCTs of hemodialysis patients,38,44 report- circulating albumin concentrations for patients ing either no difference between the control and receiving disease-specific nutritional support intervention groups or no existing control group compared with those receiving routine care (ie, for comparison. Similarly, quality of life was usual diet without nutritional support). The in- reported in only 2 studies,21,50 including 1 pub- crease in serum albumin levels in those given lished as an abstract.50 In 1 RCT, an oral nutri- nutritional support may be caused by associated tional supplement significantly improved Karnof- changes in disease activity (inflammation) and sky index scores in hemodialysis patients, hydration status,51 but there was insufficient in- although statistical comparisons with the control formation to assess this in the reviewed trials. group were lacking,21 whereas in a CCT (ab- Albumin level often is regarded as a prognostic stract), there was no overall improvement in clinical indicator for patients with CKD, and quality of life (significant improvements in the although it does not correlate with malnutrition physical domain were noted).50 There was a per se,52 serum albumin level strongly correlates notable lack of information on the effects of oral with mortality. In a recent international multi- supplements and enteral tube feeding on infec- center study, albumin concentrations less than tive, cardiovascular, and skeletal complications; 3.5 g/dL (Ͻ35 g/L) were associated with a 1.38 physiological function; and economic outcomes greater risk for death,53 supporting previous find- in dialysis patients. NUTRITIONAL SUPPORT FOR DIALYSIS PATIENTS 403

The most malnourished patients might be ex- etary intake compared with routine care (usual pected to benefit most from nutritional support, diet). This may improve clinical outcome; how- but it was not possible to determine this in the ever, insufficient published data exist to assess current review. Although 2 RCTs of oral nutri- this. Additional research is required to investi- tional supplements involved malnourished pa- gate the clinical, economic, and nutritional conse- tients,20,21 it was difficult to evaluate nutritional quences of using oral supplements and tube status in other studies (eg,16,38,43,50). Some inves- feeding (with either standard or disease-specific tigations included both well-nourished and mal- feeds) in patients with CKD receiving mainte- nourished patients (eg,39), but in the absence of nance dialysis. subgroup analysis, it is not possible to assess which patients benefited from treatment. In addi- ACKNOWLEDGMENT tion, the definition of malnutrition varied widely The authors thank Abacus Int, UK, for research support (Table 2), making it difficult to synthesize the and Statwood, UK, for statistical support. available data. REFERENCES Results of the reviewed studies suggest little 1. Heimbürger O, Qureshi AR, Blaner WS, Berglund L, effect of supplementary feeding (standard or Stenvinkel P: Hand-grip muscle strength, lean body mass disease specific; typically contributing Ͻ 25% of and plasma proteins as markers of nutritional status in the total intake) on electrolyte status. Use of patients with chronic renal failure close to the start of dialysis therapy. Am J Kidney Dis 36:1213-1225, 2000 formulae with low electrolyte concentrations as 2. Lawson JA, Lazarus R, Kelly JJ: Prevalence and the sole or predominant source of nutrition (often prognostic significance of malnutrition in chronic renal as tube feeding) may not always be optimal for insufficiency. J Ren Nutr 11:16-22, 2001 patients receiving dialysis because hypokalemia 3. Kopple JD: McCollum Award Lecture, 1996: Protein- and hypophosphatemia may occur. In general, energy malnutrition in maintenance dialysis patients. Am J Clin Nutr 65:1544-1557, 1997 regular monitoring of nutritional and electrolyte 4. Stratton RJ, Green CJ, Elia M: Prevalence of disease- status is required in patients with CKD receiving related malnutrition, in Stratton RJ, Green CJ, Elia M (eds): nutritional support to achieve optimal condi- Disease-Related Malnutrition: An Evidence-Based Ap- tions. proach to Treatment. Wallingford, UK, CABI, 2003, pp A major finding of this systematic review is 35-92 5. Kalantar-Zadeh K, Kopple JD: Trace elements and the scarcity of data comparing disease-specific vitamins in maintenance dialysis patients. Adv Ren Replace with standard formulae. Furthermore, varying Ther 10:170-182, 2003 composition of the disease-specific formulae 6. Stenvinkel P, Lindholm B, Heimbürger O: Novel ap- made comparisons difficult, and guidelines on proaches in an integrated therapy of inflammatory-associ- the composition of supplementary and complete ated wasting in end-stage renal disease. Semin Dial 17:505- 515, 2004 formulae for CKD are warranted. Overall, most 7. Stenvinkel P, Heimbürger O, Lindholm B: Wasting, studies compared nutritional support with rou- but not malnutrition, predicts cardiovascular mortality in tine care, the majority had small samples sizes end stage renal disease. Nephrol Dial Transplant 19:2181- and lacked randomization, and a number were 2183, 2004 published only as abstracts (Table 2). Additional 8. Wolfson M: Effectiveness of nutrition interventions in the management of malnourished patients treated with main- research clearly is indicated to examine the clini- tenance dialysis. J Ren Nutr 9:126-128, 1999 cal, physiological, metabolic, and economic ef- 9. Kopple JD: Therapeutic approaches to malnutrition in fects of nutritional support in patients with CKD chronic dialysis patients: The different modalities of nutri- undergoing dialysis. There also is a need to tional support. Am J Kidney Dis 33:180-185, 1999 consider the effects of interventions in patients 10. Akner G, Cederholm T: Treatment of protein-energy malnutrition in chronic nonmalignant disorders. Am J Clin with different degrees of malnutrition and the Nutr 74:6-24, 2001 effect of inflammation and distinguish between 11. Fedje L, Moore L, McNeely M: A role for oral the type of feed (multinutrient, disease-specific, nutrition supplements in the malnutrition of renal disease. J modular) and dialysis provided. Ren Nutr 6:198-202, 1996 In conclusion, this systematic review suggests 12. Bergstrom J: Nutrition in renal failure—The role of enteral feeding. Nestle Nutrition Workshop Series Clinical that enteral multinutrient support with disease- and Performance Program 3:247-254, 2000 specific formulae significantly increases serum 13. Capelli JP, Kushner H, Camiscioli TC, Chen SM, albumin concentrations and improves total di- Torres MA: Effect of intradialytic parenteral nutrition on 404 STRATTON ET AL mortality rates in end stage renal disease care. Am J Kidney vice Centre for Reviews and Dissemination. Report no. 4 (ed Dis 23:808-816, 1994 2). York, UK, University of York, 2001 14. Chertow GM, Ling J, Lew NL, Lazarus JM, Lowrie 31. Moher D, Cook DJ, Eastwood S, et al: Improving the EG: The association of intradialytic parenteral nutrition quality of reports of meta-analyses of randomised controlled administration with survival in hemodialysis patients. Am J trials. The QUORUM statement. Quality of reporting meta- Kidney Dis 24:912-920, 1994 analyses. Lancet 354:1896-1900, 1999 15. Cano N: Intradialytic parenteral nutrition: Where do 32. Agency for Health Care Policy and Research: Acute we go from here? J Ren Nutr 14:3-5, 2004 pain management: Operative or medical procedures and 16. Beutler KT, Park GK, Wilkowski MJ: Effect of oral trauma. Clin Pract Guidel 92: 1992 (Appendix B). Available supplementation on nutrition indicators in haemodialysis at: http://www.ncbi.nlm.nih.gov/books/bv.fcgi?ridϭhstat6. patients. J Ren Nutr 7:77-82, 1997 chapter.8991. Accessed July 18, 2005. 17. Clement JD, Hasbargen JA: Effect of oral supplemen- 33. Jadad A, Moore R, Carroll D, et al: Assessing the tation on nutritional indices in chronic haemodialysis pa- quality of reports of randomized clinical trials: Is blinding tients. J Am Soc Nephrol 9:189A, 1997 (abstr A0883) necessary? Control Clin Trials 17:1-12, 1996 18. Cuppari L, Medeiros FAM, Papini HF, et al: Effective- 34. Morgan WC, Bushman BJ: Integrating Results ness of oral energy-protein supplementation in severely Through Meta-Analytic Review Using SAS Software. Cary, malnourished haemodialysis patients. J Ren Nutr 4:127-135, NC, SAS Institute Inc, 1999 1994 35. Mantel N, Haenszel W: Statistical aspects of the 19. Patel MG, Raftery MJ: The use of dietary supple- analysis of data from retrospective studies of disease. J Natl ments in continuous ambulatory peritoneal dialysis patients. Cancer Inst 22:719-748, 1959 J Ren Nutr 7:129-133, 1997 36. Egger M, Davey Smith G, Schneider M, Minder C: 20. Han DS, Lee IH, Noh SK, et al: The effect of oral Bias in meta-analysis detected by a simple, graphical test. nutritional supplement in malnourished CAPD patients. J Am BMJ 315:629-634, 1997 Soc Nephrol 9:263A, 1997 (abstr A1212) 37. Mosteller F, Rourke REK, Thomas GB: Probability 21. Sharma M, Rao M, Jacob S, Jacob CK: A controlled With Statistical Applications (ed 2). London, England, Addi- trial of intermittent enteral nutrient supplementation in main- son-Wesley, 1970 tenance hemodialysis patients. J Ren Nutr 12:229-237, 2002 38. Acchiardo SR, Riley ML: Five year’s experience 22. O’Reagan S, Garel L: Percutaneous gastrojejunos- with oral nutritional supplements (ONS) in haemodialysis tomy for caloric supplementation in children on peritoneal (HD) patients (pts). Are they useful? J Am Soc Nephrol dialysis. Adv Perit Dial 6:273-275, 1990 13:415A, 2002 (abstr SA-P0742) 23. Warady B, Weis L, Johnson L: Nasogastric tube 39. Boudville N, Rangan A, Moody H: Oral nutritional feeding in infants on peritoneal dialysis. Perit Dial Int supplementation increases caloric and protein intake in peri- 16:S521-S525, 1996 (suppl 1) toneal dialysis patients. Am J Kidney Dis 41:658-663, 2003 24. Warady B, Kriley M, Belden B, Hellerstein S: Nutri- 40. Caglar K, Fedje L, Dimmitt R, et al: Therapeutic tional and behavioural aspects of nasogastric tube feeding in effects of oral nutritional supplementation during hemodialy- infants receiving chronic peritoneal dialysis. Adv Perit Dial sis. Kidney Int 62:1054-1059, 2002 6:265-268, 1990 41. Cockram DB, Hensley MK, Rodriguez M, et al: 25. Toigo G, Aparicio M, Attman PO, et al: Expert Safety and tolerance of medical nutritional products as sole Working Group report on nutrition in adult patients with sources of nutrition in people on hemodialysis. J Ren Nutr renal insufficiency (Part 2 of 2). Clin Nutr 19:281-291, 2000 8:25-33, 1998 26. Wolfson M: Management of protein and energy in- 42. Dare BK, Whiteside EJ, Wilson DE: An evaluation take in dialysis patients. J Am Soc Nephrol 10:2244-2247, study to assess the potential use of Renamil (formerly named 1999 Ren-o-mil) for patients with chronic renal failure. J Hum 27. Zarazaga A, Garcia-De-Lorenzo L, Garcia-Luna PP, Nutr Diet 10:25-36, 1997 et al: Nutritional support in chronic renal failure: Systematic 43. Heaf JG, Honore K, Valeur D, Randlov A: The effect review. Clin Nutr 20:291-299, 2001 of oral protein supplements on the nutritional status of 28. Cochrane: Cochrane Reviewers Handbook 4.2.2 malnourished CAPD patients. Perit Dial Int 19:78-81, 1999 (updated March 2004). Available at: www.cochrane.org/ 44. Holley JL, Kirk J: Enteral tube feeding in a cohort of resources/handbook/hbook.htm. Accessed: July 22, 2004 chronic hemodialysis patients. J Ren Nutr 12:177-182, 2002 29. UK National Health Service Centre for Reviews and 45. Kuhlmann MK, Schmidt F, Kohler H: Oral nutri- Dissemination: Finding studies for systematic reviews: A tional support in malnourished HD patients: Preliminary basic checklist for researchers, in National Health Service results of a randomised controlled study. J Am Soc Nephrol Centre for Reviews and Dissemination. Centre for Re- 8:199A, 1997 (abstr A0924) views and Dissemination Publications Office. York, 46. Kuhlmann MK, Schmidt F, Kohler H: High protein/ UK, University of York. Available at: www.york. energy vs standard protein/energy nutritional regimen in the ac.uk/inst/crd/revs.htm.2000. Accessed: July 22, 2004 treatment of malnourished hemodialysis patients. Miner 30. UK National Health Service Centre for Reviews and Electrolyte Metab 25:306-310, 1999 Dissemination: Undertaking Systematic Reviews of Re- 47. Mareckova O, Schuck O, Teplan V, et al: A nutrition- search on Effectiveness: CRD’s Guidance for Those Carry- ally defined liquid diet for haemodialized patients. Cas Lek ing Out or Commissioning Reviews. National Health Ser- Cesk 134:77-79, 1995 NUTRITIONAL SUPPORT FOR DIALYSIS PATIENTS 405

48. Patel MG, Kitchen S, Miligan PJ: The effect of 53. Combe C, McCullough KP, Asano Y, et al: Kidney dietary supplements on the nPCR in stable hemodialysis Disease Outcomes Quality Initiative (K/DOQI) and the patients. J Ren Nutr 10:69-75, 2000 Dialysis Outcomes and Practice Patterns Study (DOPPS): 49. Sayce HA, Rowe PA, McGonigle RJS: Percutaneous Nutrition guidelines, indicators and practices. Am J Kidney endoscopic gastrostomy feeding in haemodialysis outpa- Dis 44:S39-S46, 2004 (suppl 2) tients. J Hum Nutr Dietet 13:333-341, 2000 54. Lowrie EG, Lew NL: Death risk in haemodialysis 50. Shah NA, Mueller BA, Thomas J, Kraus MA, Scott patients: The predictive value of commonly measured vari- MK: Effects of supplemental enteral nutrition on nutri- ables and an evaluation of death rate differences between tional status and quality of life in ESRD patients receiv- facilities. Am J Kidney Dis 15:458-482, 1990 ing hemodialysis. J Am Soc Nephrol 10:303A, 1999 55. Owen WF, Lew NL, Liu Y, Lowrie EG, Lazarus JM: (abstr A1535) The urea reduction ratio and serum albumin concentration as 51. Shenkin A, Cederblad G, Elia M, Isaksson B: Interna- predictors of mortality in patients undergoing hemodialysis. tional Federation of Clinical Chemists. Laboratory assess- N Engl J Med 329:1001-1006, 1993 ment of protein energy status. Clin Chim Acta 253:S5-S59, 56. Lowrie EG, Huang WH, Lew NL: Death risk predictors 1996 (suppl 1) among peritoneal dialysis and hemodialysis patients: A prelimi- 52. dos Santos NSJ, Draibe SA, Kamimura MA, et al: Is nary comparison. Am J Kidney Dis 26:220-228, 1995 serum albumin a marker of nutritional status in hemodialysis 57. Dwyer JT, Larive B, Leung J, et al: Nutritional status patients without evidence of inflammation? Artif Organs affects quality of life in Hemodialysis (HEMO) Study pa- 27:681-686, 2003 tients at baseline. J Ren Nutr 12:213-223, 2002 ORIGINAL INVESTIGATIONS Pathogenesis and Treatment of Kidney Disease and Hypertension CKD Progression and Mortality Among Older Patients With Diabetes

Uptal D. Patel, MD, Eric W. Young, MD, MS, Akinlolu O. Ojo, MD, PhD, and Rodney A. Hayward, MD

● Background: Chronic kidney disease (CKD) is clearly associated with an increased risk for adverse outcomes; however, the cumulative impact of renal and cardiac complications in high-risk populations is not known. In addition, little is known about patterns of nephrology care in patients with CKD. Methods: We conducted a retrospective longitudinal cohort study assessing CKD prevalence and progression, associations with all-cause mortality, and variations in patterns of nephrology consultation in older patients with diabetes in a vertically integrated health care system. Results: A total of 12,570 patients within a 7–Veterans Affairs hospital service network in 1998 to 1999 were identified by means of computerized records. Nearly half (48%) were affected with CKD; most had mild to moderate CKD. After an observation period of 3 years, mortality rates in those unaffected with CKD were high (4.7 deaths/100 person-years) and increased substantially with progressive CKD (eg, 20.1 deaths/100 person-years with an estimated glomerular filtration rate [GFR] of 15 to 29 mL/min/1.73 m2 [0.25 to 0.48 mL/s/1.73 m2]). Only 7.2% of patients with CKD had a nephrology visit during the entire 5-year study period. Although visits increased with more advanced CKD, only 32% of patients with an estimated GFR of 15 to 29 mL/min/1.73 m2 had been seen in a nephrology clinic. We also found that nephrology referrals were driven preferentially by elevations in serum creatinine levels, rather than low GFRs. Conclusion: Many in this cohort of older patients with diabetes are affected with CKD. Mortality rates are high, and mortality risks associated with CKD amplify those of other risk factors. Nephrology visits are low and may represent an unexploited resource for improving CKD management. Underrecognition of CKD likely is related to overestimation of kidney function by relying on serum creatinine level in elderly patients. Am J Kidney Dis 46:406-414. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Chronic kidney disease (CKD); mortality; progression; cardiovascular disease; nephrology referral.

IXTEEN MILLION Americans currently are 2050.1 Largely because of the increasing preva- S afflicted with diabetes mellitus, and nearly lence of diabetes, the incidence and prevalence 29 million are projected to have diabetes by of chronic kidney disease (CKD) and end-stage renal disease (ESRD) are projected to increase dramatically during the next few decades.2 Dia- From the Veterans Affairs Health Services Research and betic nephropathy already is the leading cause of Development Center of Excellence, VA Ann Arbor Health- care System; and Departments of Internal Medicine and chronic kidney failure in the United States, ac- Pediatrics and Communicable Diseases, University of Michi- counting for nearly half of all incident cases each gan Health System, Ann Arbor, MI. year,2 and these increases in CKD and ESRD Received April 8, 2005; accepted in revised form May 26, rates will result in a marked increase in patient 2005. Originally published online as doi:10.1053/j.ajkd.2005.05.027 morbidity and mortality related to underlying on July 25, 2005. kidney disease, as well as a significant increase Supported in part by grant no. QUERI DIB 98-001 from in health care costs. The Office of Academic Affiliations and Health Services The natural history of diabetes and its renal Research and Development, Department of Veterans Affairs; complications have been studied extensively.3 The Robert Wood Johnson Foundation; and grant no. P60 4,5 DK-20572 from the National Institute of Diabetes and Diges- Recent studies focusing on CKD also have tive and Kidney Diseases, The National Institutes of Health. included patients with diabetes. However, these Presented in part at the 37th Annual Meeting of the observational studies have focused on relatively American Society of Nephrology, St Louis, MO, October healthy populations, not high-risk populations 29-November 1, 2004. Address reprint requests to Uptal D. Patel, MD, Robert with multiple cardiovascular risk factors. Veter- Wood Johnson Clinical Scholars Program, University of ans Affairs (VA) facilities serve a population that Michigan, 6312 Medical Science Bldg I, 1150 West Medical tends to be sicker and poorer than the general Center Dr, Ann Arbor, MI 48109-0604. E-mail: patelu@ population, leading to a disproportionately high umich.edu 6-8 © 2005 by the National Kidney Foundation, Inc. burden of comorbid diseases, allowing for a 0272-6386/05/4603-0003$30.00/0 detailed evaluation of those at greatest risk for doi:10.1053/j.ajkd.2005.05.027 CKD complications.

406 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 406-414 CKD PROGRESSION AND MORTALITY 407

Because CKD is associated with increased available in the electronic database at each study period by using the abbreviated Modification of Diet in Renal Disease all-cause mortality, efforts to improve adherence 9 to the proven and effective treatments outlined in (MDRD) Study formula, defined as follows : Estimated GFR (mL/min/1.73 m2) ϭ 186.3 ϫ [serum current guidelines have the potential to dramati- creatinine]–1.154 ϫ age–0.203 ϫ [0.742 if female] ϫ [1.21 if cally impact on such important outcomes as black]. 9 death and progression of CKD. Early referral of If the race variable was reported as unknown, estimation patients with CKD to nephrology care is one of GFR was performed with white race by default (20% of strategy believed to improve health care out- cases). comes.3,9,10 Although care for patients with CKD We then staged the severity of CKD by using modified 11–13 definitions from the National Kidney Foundation (NKF)– by nephrologists is not always optimal, re- Kidney Disease Outcomes Quality Initiative (K/DOQI) clini- cent efforts involving comprehensive manage- cal practice guidelines.9 Consistent with the K/DOQI defini- ment with nephrology consultation appear prom- tion of CKD, individuals with stages 1 and 2 disease are ising.14–16 To better understand the impact of those with a GFR of 60 mL/min/1.73 m2 or greater (Ն1.00 2 CKD on a high-risk patient population with mL/s/1.73 m ) who also have evidence of kidney damage. For our study, we define kidney damage as the presence of diabetes, we conducted a longitudinal cohort documented proteinuria defined by 1 of the following crite- study assessing CKD prevalence and progres- ria: (1) urine dipstick analysis result of 1ϩ or greater, (2) sion, associations with all-cause mortality, and urine protein level of 30 mg/dL or greater, or (3) urine variations in patterns of nephrology consultation microalbumin-creatinine ratio of 30 ␮g/g or greater. If urine in a vertically integrated health care system. protein data were not available (26% of cases), the default categorization was no documented proteinuria. Patients were classified according to the following groups (corresponding METHODS K/DOQI CKD stage in parentheses): GFR of 60 mL/min/ Study Population 1.73 m2 or greater (Ն1.00 mL/s/1.73 m2) without docu- mented proteinuria (stage 0, at risk), GFR of 60 mL/min/ We used automated clinical data from 7 VA medical 1.73 m2 or greater (Ն1.00 mL/s/1.73 m2) with documented centers located in a single Veterans’ Integrated Service proteinuria (stages 1 and 2), GFR of 45 to 59 mL/min/1.73 Network to identify a cohort of patients who had diabetes. m2 (0.75 to 0.98 mL/s/1.73 m2; stage 3, early), GFR of 30 to To build our retrospective longitudinal cohort, electronic 44 mL/min/1.73 m2 (0.50 to 0.73 mL/s/1.73 m2; stage 3, pharmacy, laboratory, and clinical encounter information for 2 17,18 late), GFR of 15 to 29 mL/min/1.73 m (0.25 to 0.48 fiscal years 1998 and 1999 were used to identify patients mL/s/1.73 m2; stage 4), and GFR less than 15 mL/min/1.73 who met at least one of the following criteria within the past m2 (Ͻ0.25 mL/s/1.73 m2; stage 5). 12 months: (1) 1 hospitalization with a diabetes-related Predictor variables. We identified potential predictors International Classification of Diseases, Ninth Revision (ICD- of CKD progression and complications from the regional VA 9), code (250.x, 357.2, 362.0, or 366.41), (2) 2 outpatient outpatient and inpatient databases, including age, sex, race, visits with a diabetes-related ICD-9 code, or (3) at least 1 blood pressure, and laboratory data (hemoglobin A1c, hemo- prescription for an oral hypoglycemic agent, insulin, or globin, and potassium levels). Using ICD-9 diagnostic codes, blood glucose monitoring supplies. We identified 29,863 we identified 6 comorbidities shown to independently pre- potential study subjects. dict mortality (hypertension, coronary artery disease, con- To restrict all analyses to patients who were actively using gestive heart failure, peripheral vascular disease, hyperlip- VA facilities during the study period, final eligibility criteria idemia, and stroke) and 2 additional diabetes-related included: (1) 2 outpatient visits of any kind in 1998 (8,327 complications (neuropathy and retinopathy) that account patients excluded); (2) at least one visit of any kind in each for the burden of microvascular disease. The frequency of subsequent year through 2002, if alive (5,867 patients ex- outpatient primary care and subspecialty (cardiology, cluded); (3) survival through the cohort identification period hypertension, diabetes, and nephrology) visits during the (1,638 patients excluded because they died before the end of 2-year cohort development period were identified by fiscal year 1999); and (4) a baseline serum creatinine mea- using VA encounter codes. Hyperkalemia is defined as surement during fiscal years 1998 to 1999 (1,450 patients having any serum potassium level greater than 5.5 mEq/L excluded). An additional 8 patients were excluded because (mmol/L) from a nonhemolyzed specimen. of missing demographic variables, and 3 patients were Outcomes. Death and CKD progression outcomes were excluded for age younger than 18 years. Baseline character- recorded if the event occurred by the end of the 3-year istics for the final cohort of 12,570 patients were developed observation period. Mortality data were obtained from cen- from information from fiscal years 1998 and 1999. This tralized VA records, which have been shown to be accurate, cohort was observed for 3 years from the beginning of fiscal but do not contain information on disease-specific mortal- year 2000 through fiscal year 2002. The VA Ann Arbor ity.19,20 Any progression of CKD is defined as a change to Institutional Review Board approved the study. any disease group category with a lower GFR and/or pres- ence of proteinuria with at least a 20% decrease in estimated Study Data GFR (to avoid trivial changes in CKD classifications). Estimation of kidney function. We estimated glomerular Progression to ESRD is defined as a change to an estimated filtration rates (GFRs) from the last serum creatinine value GFR less than 15 mL/min/1.73 m2 (Ͻ0.25 mL/s/1.73 m2; 408 PATEL ET AL stage 5) because this level of kidney function in older as level of kidney function declined. However, patients with diabetes represents a reasonable threshold for 9 even in participants without documented CKD dialysis therapy. Small proportions of patients (1% to 5%) (stage 0), such comorbid conditions as hyperten- had improvements in their level of kidney function and were included in the “no-change” group. Follow-up creatinine sion, coronary artery disease, and hyperlipid- values were available for 91% of cohort survivors after 3 emia were prevalent (Table 1). Mean hemoglo- years. bin A1c levels were greatest in patients with early To assess variations in patterns of nephrology care, we CKD (8.1% for those with stage 1 to 2 disease), examined receipt of nephrology visits. A nephrology visit which is the group in which intensive glycemic was recorded if any outpatient clinic visit occurred during the cohort period or the 3-year observation period. Because control probably would have had the most ben- 21 an outpatient visit to a nephrology clinic in the VA system efit. Mean systolic blood pressures were greater requires a referral by one of the patients’ providers, nephrol- than the current recommended target of 130 mm ogy visits are likely to be a reasonable surrogate for nephrol- Hg3,9,22 and were greater with lower levels of ogy referral patterns. kidney function. Overall, the proportion of study Statistical Analysis participants with a mean systolic blood pressure greater than goal was high (72%, Ͼ130 mm Hg; We used linear regression for continuous measures and Ͼ logistic regression for categorical measures to compare 52%, 140 mm Hg). Those with stage 3 CKD or independent effects of baseline predictor variables with greater were much more likely to have a history outcomes. We used dummy variables for each CKD group of hyperkalemia and anemia. Unfortunately, VA with the no-CKD group as reference. Chi-square testing was administrative data did not fully capture data on used for bivariate comparisons for categorical variables. Unadjusted event rates are presented as number of events race (missing for 20%); however, among those per 100 person-years, with 95% confidence intervals. for whom racial data were available, African To evaluate the relative influence of GFR, serum creati- Americans comprised 28% of those with stage 5 nine level, and age on decisions regarding nephrology refer- CKD despite making up only 15% of the entire ral, we used logistic regression models. These analyses were cohort (risk ratio, 2.1; 95% confidence interval, limited to patients with a serum creatinine level less than 2.5 mg/dL (Ͻ191 ␮mol/L; n ϭ 10,022) to examine study 1.5 to 3.0). subjects without severe CKD (ie, to examine early referrals). In general, 61% of this cohort had no change Only survivors throughout the study period were included in in their status during the 3-year follow-up period these analyses to limit analyses to those who were eligible (no death or progression of CKD), but this varied for referral for the full 3-year study period. Predictive dramatically by baseline CKD stage, ranging performance of logistic models was evaluated with areas under the receiver operator characteristic curve (C statistic). from 69% for those with no baseline CKD to All statistical analyses were performed using STATA 8.2 33% for those with stages 4 and 5 CKD. Overall, (2004; Stata Corp, College Station, TX). 21% died during the 3-year study period, and rates of death were dramatically higher in pa- RESULTS tients with more advanced CKD (Fig 1A). Al- Baseline characteristics by CKD status are though mortality rates were relatively high even listed in Table 1. Study participants were predomi- in those without CKD (4.7 deaths/100 person- nantly older white men with a mean age of 66 years), these rates were more than 4-fold greater years. The distribution of estimated GFRs shows for those with stage 4 CKD (20.1 deaths/100 that half (52%) had no evidence of any kidney person-years). In addition, these rates of death disease at baseline (stage 0 CKD). Overall, 18% were also much greater than those in patients had early CKD with an estimated GFR of 60 with similar degrees of CKD in a community- mL/min/1.73 m2 or greater (Ն1.00 mL/s/1.73 based population4 (Fig 2). Finally, even in those m2) with proteinuria (stages 1 and 2 CKD), and unaffected with CKD, rates of death were much 26% of patients had moderate CKD (stage 3 greater than in members of the general popula- CKD: 69% had a GFR of 45 to 59 mL/min/1.73 tion of similar age (1.0 death/100 person-years m2 [0.75 to 0.98 mL/s/1.73 m2] and 31% had a for ages 55 to 64 years, 2.3 deaths/100 person- GFR of 30 to 44 mL/min/1.73 m2 [0.50 to 0.73 years for ages 65 to 74 years, and 5.6 deaths/100 mL/s/1.73 m2]). Overall, few participants were person-years for ages 75 to 84 years).23 found to have advanced CKD (3%, stage 4; 1%, Rates of any progression to a lower GFR stage 5 CKD). group or new development of proteinuria also As expected, age and presence of diabetic were graded (Fig 1B), increasing from 3.2/100 complications and other comorbidities increased person-years in patients without CKD (stage 0) K RGESO N MORTALITY AND PROGRESSION CKD

Table 1. Characteristics of Eligible Study Participants

CKD Stage 0 CKD Stages1&2 CKDStage 3 CKD Stage 4 CKD Stage 5

GFR Ն 60 mL/min/1.73 GFR Ն 60 mL/min/1.73 m2, No Documented m2, With Documented GFR 45-59 mL/ GFR 30-44 mL/ GFR 15-29 mL/ GFR Ͻ 15 mL/ Proteinuria Proteinuria min/1.73 m2 min/1.73 m2 min/1.73 m2 min/1.73 m2 All

No. of patients (%) 6,494 (52) 2,304 (18) 2,274 (18) 978 (8) 365 (3) 155 (1) 12,570 Mean age (y) 64 Ϯ 11 64 Ϯ 11 71 Ϯ 8* 72 Ϯ 9* 70 Ϯ 10* 66 Ϯ 10* 66 Ϯ 11 Male sex (%) 97.4 97.9 97.3 97.0 97.8 98.1 97.4 Race/ethnicity (%) White 61 60 71* 70* 69* 54 64 African-American 14 25* 8* 11* 16 28* 15 Other 1 1 1 1 1 2 1 Unknown 24 14* 20* 18* 13* 16* 20 Mean hemoglobin A1c (%) (n ϭ 11,977) 7.5 Ϯ 1.6 8.1 Ϯ 1.9* 7.3 Ϯ 1.4* 7.3 Ϯ 1.4* 7.3 Ϯ 1.5* 7.3 Ϯ 1.6* 7.6 Ϯ 1.6 Proteinuria (%) (n ϭ 9,318) 0 100* 49* 66* 87* 91* 42 Mean blood pressure (mm Hg) (n ϭ 5,969) Systolic 140 Ϯ 18 144 Ϯ 20* 143 Ϯ 20* 143 Ϯ 21* 148 Ϯ 21* 148 Ϯ 21* 142 Ϯ 19 Diastolic 75 Ϯ 11 76 Ϯ 12* 73 Ϯ 11* 71 Ϯ 12* 71 Ϯ 12* 72 Ϯ 13* 74 Ϯ 11 Potassium Ͼ 5.5 mEq/L (%) 3 4* 8* 16* 27* 34* 6 Mean hemoglobin (g/dL) (n ϭ 8,602) 14.2 Ϯ 1.6 13.9 Ϯ 1.7* 13.7 Ϯ 1.7* 12.7 Ϯ 1.7* 11.8 Ϯ 1.7* 10.6 Ϯ 1.4* 13.8 Ϯ 1.8 Diabetes-related complications (%) Neuropathy 8 14* 13* 15* 20* 23* 11 Retinopathy 17 26* 21* 30* 41* 52* 22 Comorbidities (%) 78 85* 87* 91* 95* 94* 82 Hypertension Coronary artery disease 45 53* 60* 72* 69* 67* 52 Congestive heart failure 16 27* 33* 48* 51* 53* 25 Peripheral vascular occlusive disease 17 29* 31* 37* 44* 41* 24 Hyperlipidemia 48 54* 56* 55* 58* 38* 51 Cerebrovascular accident 11 16* 18* 25* 22* 17* 15 Mean primary care visits in last year 5 Ϯ 46Ϯ 4* 6 Ϯ 4* 7 Ϯ 5* 6 Ϯ 4* 6 Ϯ 56Ϯ 4 Mean subspecialty visits in last year 0.5 Ϯ 1 0.4 Ϯ 1 0.5 Ϯ 1* 1 Ϯ 2* 2 Ϯ 3* 2 Ϯ 3* 1 Ϯ 1

NOTE. Values expressed as number (percent) or mean Ϯ SD. To convert GFR in mL/min to mL/s, multiply by 0.01667; serum potassium in mEq/L to mmol/L, multiply by 1; hemoglobin in g/dL to g/L, multiply by 10. *P Ͻ 0.05 versus CKD stage 0 group (no documented proteinuria with GFR Ն 60 mL/min/1.73 m2). 409 410 PATEL ET AL

Fig 1. Unadjusted event rates by estimated GFR (error bars represent 95% confidence intervals). Rates of (A) death and (B) any progression were graded with decreasing levels of kidney function. (C) Rates of progression to ESRD were dominated by rates of death at all baseline levels of kidney function and only increased significantly with advanced CKD. *Rates for those with a GFR less than 15 mL/min/1.73 m2 are 0 per 100 person-years because their kidney function could progress no further. †ESRD presumed equivalent to an estimated GFR less than 15 mL/min/1.73 m2. To convert GFR in mL/min to mL/s, multiply by 0.01667. to 15.1/100 person-years in those with advanced patients surviving the full 3 years of the study [n CKD (stage 4). Rates of progression to ESRD ϭ 10,252]). Overall, only 7.2% of patients with (Fig 1C) were substantially lower than rates of CKD had a nephrology visit during the study. death at all baseline CKD groups (eg, a patient Whereas those with more advanced CKD were with a baseline GFR of 30 to 44 mL/min/1.73 m2 likely to see a nephrologist, we still found that [0.50 to 0.73 mL/s/1.73 m2] was nearly 9 times less than one third of those with stage 4 CKD had more likely to die in the next 3 years than to been seen in a nephrology clinic. There also was develop ESRD). considerable variability by clinical site (C statis- We then examined nephrology visits for differ- tic, 0.60; chi-square ϭ 44.7; P Ͻ 0.0001). ent CKD groups (Table 2; analysis limited to In unadjusted analyses in patients without ad- vanced CKD (Table 3;nϭ 10,022), older age was associated with an increased odds of nephrol- ogy referral, but this was caused by elderly persons having a much greater prevalence of CKD. Given similar degrees of renal function, older patients were much less likely to be seen by a nephrologist (odds ratio, 0.75/decade of age; 95% confidence interval, 0.60 to 0.94). For ex- ample, in patients with stage 3 CKD, a 50-year- old had a 10.5% probability of seeing a nephrolo- gist compared with a 5.6% probability for a Fig 2. Comparison of rates of death among mem- 75-year-old. bers of Kaiser Permanente and the VA, by estimated GFR. To convert GFR in mL/min to mL/s, multiply by Part of the reason for this age effect may be 0.01667. that providers were keying into serum creatinine CKD PROGRESSION AND MORTALITY 411

Table 3. Unadjusted and Adjusted Odds Ratios and 0.67 25.1 18.8 Confidence Intervals for Probability of Having a Ϯ Ϯ Ϯ All Nephrology Visit

Odds Ratio (95% confidence interval)

Characteristic Unadjusted Adjusted* 2 43.0† 3.7 2.77† 1.21 3.1† 76.1 mol/L, multiply by 88.4. 15 mL/min/ Ϯ Ϯ Ϯ Age (/10 y) 1.05 (0.94-1.18) 0.75 (0.60-0.94) ␮ 1.73 m Ͻ Creatinine (/1 mg/dL) 14.8 (10.3-21.2) 10.0 (4.2-24.2) GFR (/10 mL/min/ GFR 1.73 m2) 0.59 (0.54-0.66) 0.88 (0.71-1.08)

NOTE. Analyses were limited to patients with complete

2 follow-up and baseline creatinine level less than 2.5 mg/dL 46.7† 23.9 4.2† 9.9 0.59† 7.14 (191 ␮mol/L; n ϭ 10,022). To convert serum creatinine in Ϯ Ϯ Ϯ

1.73 m mg/dL to ␮mol/L, multiply by 88.4; GFR in mL/min to mL/s, multiply by 0.01667.

GFR 15-29 mL/min/ *Results adjusted for other variables shown in the table by using logistic regression analysis. 2 36.5† 31.8 4.1† 23.8 0.24† 3.02 level, rather than calculating GFR. Regression Ϯ Ϯ Ϯ

1.73 m analyses found that after controlling for serum creatinine level, GFR was no longer significant, GFR 30-44 mL/min/ whereas serum creatinine levels remained a very strong predictor even after controlling for GFR ). 2 ). 2 (Table 3). The proportion of study participants at 2 4.2† 38.9 0.13† 1.88 22.1† 15.8 risk for underrecognition of decreased GFR was Ϯ Ϯ Ϯ

1.73 m substantial: 76% of study participants with late stage 3 CKD (GFR, 30 to 44 mL/min/1.73 m2 GFR 45-59 mL/min/ [0.50 to 0.73 mL/s/1.73 m2]) had a creatinine Յ ␮

60 mL/min/1.73 m level of 2 mg/dL or less ( 153 mol/L), and 60 mL/min/1.73 m Ն 2 62% of study participants with stage 4 CKD Ն 21.7* 53.6 0.19* 1.42 17.5† 5.2 2 60 mL/min/ Ϯ Ϯ Ϯ (GFR, 15 to 30 mL/min/1.73 m [0.25 to 0.50

1.73 m 2 Ն

Proteinuria mL/s/1.73 m ]) had a creatinine level of 3 mg/dL Յ ␮ With Documented

GFR or less ( 229 mol/L). These rates were much SD. To convert GFR in mL/min to mL/s, multiply by 0.01667; serum creatinine in mg/dL to

Ϯ greater for elderly individuals, thereby putting them at particularly high risk for having underrec- 2 19.2 87.1 0.17 1.01 9.0 3.2 ognized marked CKD. Table 2. Creatinine, GFR, and Nephrology Visits by Estimated GFR Ϯ Ϯ Ϯ 60 mL/min/ 1.73 m Ն DISCUSSION Proteinuria 86.1 No Documented

GFR This study is one of the largest and most )

2 comprehensive longitudinal studies examining the progression and mortality risk associated with patients with CKD with diabetes. We found that in a cohort of patients with diabetes cared for in the VA health care system, nearly half are affected with CKD. Progressive levels of CKD less than an estimated GFR of 60 mL/min/1.73 m2 (1.00 mL/s/1.73 m2) were associated with large increases in the relative risk for mortality, 0.05 versus CKD stage 0 (no documented proteinuria with GFR 0.001 versus CKD stage 0 (no documented proteinuria with GFR similar to the pattern observed in a recent large Ͻ Ͻ 4 P P community-based cohort. As in that study, we NOTE. Values expressed as number (percent) or mean * † Mean estimated GFR (mL/min/1.73 m Mean with any nephrology visit (%) 0.8 No. of patients (%)Mean creatinine (mg/dL) 1.00 5,647 (55)also 1,931 (19) observed 1,781 (17) a substantial 625 (6) increase 201 (2) in the rate 67 (1) of 10,252 death in subjects with an estimated GFR less 412 PATEL ET AL than 45 mL/min/1.73 m2 (Ͻ0.75 mL/s/1.73 m2), and 2-fold greater for those with a GFR less than suggesting that the NKF staging system may be 30 mL/min/1.73 m2 (Ͻ0.50 mL/s/1.73 m2). No- improved by including early and late subdivi- tably, members of the Kaiser integrated health sions within stage 3 CKD (early, 45 to 59 mL/min/ care system had fewer comorbidities, but similar 1.73 m2 [0.75 to 0.98 mL/s/1.73 m2]; late, 30 to ages, in CKD groups compared with members of 44 mL/min/1.73 m2 [0.50 to 0.73 mL/s/1.73 our VA study. Given such high overall death m2]). However, despite the increasing risk for rates, targeted interventions in this high-risk group mortality with decline in kidney function, ne- have the potential to yield profound mortality phrology visits remained low. Although the preva- reductions with even modest decrements in rela- lence of nephrology visits in patients without tive death rates. advanced CKD has not been described previ- Several studies have documented that late refer- ously, our finding of underutilization of nephrol- ral to nephrology is associated with poorer pre- ogy referrals is consistent with several prior ESRD care11,24,25 and greater mortality after initiat- reports.11,24-26 ing renal replacement therapy for ESRD.24,25,27 However, our study also highlights several While kidney function declines, implementing findings that have not been well addressed in past therapies to slow the progression of CKD and treat research. The proportion of patients affected with the adverse effects of declining kidney function CKD in our study is greater than that observed in (eg, anemia, hyperparathyroidism, and acidosis) is adults with diabetes in the Third National Health vital to improving outcomes. Initiating erythropoi- and Nutrition Examination Survey.26 Patients in etin therapy, managing calcium and phosphorous our cohort were older and had more comorbidi- levels, and planning renal replacement therapy are ties overall, suggesting that causes other than within the nephrologist’s purview. However, all classic diabetic glomerulosclerosis may be con- these effective interventions require an initial refer- tributing to the high proportion of CKD.26 In ral for nephrology consultation. However, even in addition, although prior studies noted a graded the VAsystem, where access to subspecialty consul- association between lower GFR and risk for tation is universal and each person in this cohort death,4,5 most studies that observed associations was receiving care regularly for diabetes, overall between CKD and cardiovascular disease or death use of nephrology visits was low and remained low included relatively healthy populations; the mag- even with stages 3 and 4 CKD. nitude of increased risk in a high-risk population Although nephrology referrals are often late of patients with diabetes has not been well char- and underutilized at the initiation of ESRD, po- acterized. Our study suggests that the substantial tential explanatory factors in patients without mortality relative risk increases seen in healthier advanced CKD have not been investigated previ- populations remains a high multiplier of risk in ously. We found that nephrology referrals for populations with high comorbidities. Although early and moderate CKD in elderly patients were this may seem expected, risk factors do not driven preferentially by elevation in serum creat- always follow this pattern. For example, relative inine level, rather than a low estimated GFR. risk increases for low-density lipoprotein choles- Serum creatinine level alone is a particularly terol level and diabetes are both attenuated with insensitive indicator of deteriorating renal func- advancing age. The persistence of a high relative tion in older patients.28,29 Thus, our results sug- risk increase shown in this study highlights that gest that primary care providers often may fail to mortality risks associated with CKD amplify recognize diminished GFR in the setting of mild those of other risk factors. For example, the high elevations in serum creatinine levels. Whereas baseline risk of this patient population resulted in other potential mechanisms probably are contrib- absolute excess mortality risks from CKD that uting to underreferral, such as “therapeutic nihil- were dramatically greater than a more heteroge- ism”30 or poor outcome expectancy,31 our study neous population in the Kaiser Permanente health shows that at least 1 problem includes under- care system4 (Fig 2): 7-fold greater for those recognition of decreased kidney function based with an estimated GFR greater than 45 mL/min/ on serum creatinine level alone. 1.73 m2 (Ͼ0.75 mL/s/1.73 m2), 3-fold greater One potential solution with widespread appeal for those with an estimated GFR of 30 to 44 involves automated reporting of estimated GFR mL/min/1.73 m2 (0.50 to 0.73 mL/s/1.73 m2), by laboratories, an explicit recommendation by CKD PROGRESSION AND MORTALITY 413 the NKF to improve identification of patients bias our results toward the null; therefore, the with unrecognized CKD.9 The modified MDRD robustness of our findings suggests that this was Study equation can be used by linking to demo- not a major problem. A population of relatively graphic information in the patient record. With compliant patients may have been selected by an advanced electronic medical record platform the rigorous criteria used to ensure that study across each of its facilities, the VA health care participants used the VA as their regular source system has tremendous potential to maximize of health care because more than half of the detection of patients with CKD. Automated labo- original potential sample was excluded. How- ratory reporting of GFR recently has been imple- ever, this potential selection bias would only mented system wide and is being activated at understate the full impact of CKD on mortality. individual centers.32 Although other vertically Race is not measured well in VA administrative integrated health care systems have initiated au- data; however, the overall classification by esti- tomated reporting and even automatic nephrol- mated GFR did not change significantly when ogy referrals,15,32,33 the impact on patient out- unknown race was assigned to all black versus comes has not yet been determined. Although the all white default. Finally, visits to nephrology automated laboratory GFR reporting may pro- clinics may be underestimated if patients had vide improved recognition of CKD, additional nephrology providers outside the VAsystem. efforts to improve the recognition of CKD, such In conclusion, we found that nearly half the as educational programs directed at primary care patients with diabetes being cared for within the physicians, probably will be required to optimize VA health care system are affected with CKD, CKD care.34 An additional barrier to optimal and CKD was associated with dramatic increases reporting in the VA system (and in many other in mortality risk, providing a critically important settings) is inadequate electronically available target group for quality improvement interven- data on race. More complete racial data would tions. In this vertically integrated health care allow reporting of estimated GFRs without the system, nephrology visits are underutilized and need for further adjustments by individual provid- represent a strategic resource that may improve ers for African-American patients9; however, an the management of CKD, a hypothesis that should alternative approach would be for the laboratory be investigated further. Our findings shed light to routinely report GFR estimates for both white on one of the pathways for this underutilization: and black race. Finally, an unanticipated conse- underrecognition of CKD related to overestima- quence of automated GFR reporting may be that tion of kidney function by relying on serum it leads to a significant increase in nephrology creatinine measures in elderly patients. Continu- referrals, potentially overwhelming current capac- ous quality improvement efforts focusing on ity. Efficient use of resources will require appro- identifying patients with CKD through auto- priately balancing nephrology and preventive mated GFR reporting by laboratories may facili- cardiology inputs. Nephrology consultations tate more complete and timely nephrology refer- should be used to address management issues rals and improve the management of CKD. specific to uncertain diagnoses, delaying progres- sion, and preparing for renal replacement therapy. ACKNOWLEDGMENT Meanwhile, preventive cardiology consultations The authors thank Jennifer Davis and Sonya DeMonner should focus on primary or secondary cardiovas- for outstanding database management and technical support. cular preventive strategies. In addition to these results being limited to a REFERENCES high-risk male population receiving care within a 1. Narayan KM, Boyle JP, Thompson TJ, Sorensen SW, Williamson DF: Lifetime risk for diabetes mellitus in the single hospital network, this study has several United States. JAMA 290:1884-1890, 2003 other important limitations. First, variability in 2. US Renal Data System: USRDS 2002 Annual Data creatinine and proteinuria measurements and Report. The National Institutes of Health, National Institute known imprecision of the MDRD formula with of Diabetes and Digestive and Kidney Diseases; Bethesda, mild renal impairment may have led to some MD, 2002 3. Molitch ME, DeFronzo RA, Franz MJ, et al: Nephrop- misclassification of our baseline CKD groups, 35,36 athy in diabetes. Diabetes Care 27:S79-S83, 2004 (suppl 1) particularly for those with early CKD. How- 4. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY: ever, such measurement error most likely would Chronic kidney disease and the risks of death, cardiovascu- 414 PATEL ET AL lar events, and hospitalization. N Engl J Med 351:1296-1305, bases: Comparison with state death certificates. Ann 2004 Epidemiol 11:286-291, 2001 5. Keith DS, Nichols GA, Gullion CM, Brown JB, Smith 21. Vijan S, Hofer TP, Hayward RA: Estimated benefits DH: Longitudinal follow-up and outcomes among a popula- of glycemic control in microvascular complications in type 2 tion with chronic kidney disease in a large managed care diabetes. Ann Intern Med 127:788-795, 1997 organization. Arch Intern Med 164:659-663, 2004 22. Chobanian AV, Bakris GL, Black HR, et al: Seventh 6. Petersen LA, Normand S-L, Daley J, McNeil BJ: Report of the Joint National Committee on Prevention, Outcome of myocardial infarction in Veterans Health Admin- Detection, Evaluation, and Treatment of High Blood Pres- istration patients as compared with Medicare patients. N Engl sure. Hypertension 42:1206-1252, 2003 J Med 343:1934-1941, 2000 23. Kochanek KD, Smith BL: Deaths: Preliminary data 7. Fink JC, Salmanullah M, Blahut SA, et al: The inevita- for 2002. Natl Vital Stat Rep 52:1-47, 2004 bility of renal function loss in patients with hypercreatine- 24. Cleveland DR, Jindal KK, Hirsch DJ, Kiberd BA: mia. Am J Nephrol 21:386-389, 2001 Quality of prereferral care in patients with chronic renal 8. Young BA, Maynard C, Boyko EJ: Racial differences insufficiency. Am J Kidney Dis 40:30-36, 2002 in diabetic nephropathy, cardiovascular disease, and mortal- 25. Roderick P, Jones C, Drey N, et al: Late referral for ity in a national population of veterans. Diabetes Care end-stage renal disease: A region-wide survey in the south 26:2392-2399, 2003 west of England. Nephrol Dial Transplant 17:1252-1259, 9. National Kidney Foundation: K/DOQI Clinical Prac- 2002 tice Guidelines for Chronic Kidney Disease: Evaluation, 26. Kramer HJ, Nguyen QD, Curhan G, Hsu CY: Renal classification, and stratification. Am J Kidney Dis 39:S1- insufficiency in the absence of albuminuria and retinopathy S246, 2002 (suppl 2) among adults with type 2 diabetes mellitus. JAMA 289:3273- 10. Mendelssohn DC, Barrett BJ, Brownscombe LM, et 3277, 2003 al: Elevated levels of serum creatinine: Recommendations 27. Huisman RM: The deadly risk of late referral. Neph- for management and referral. CMAJ 161:413-417, 1999 rol Dial Transplant 19:2175-2180, 2004 28. Duncan L, Heathcote J, Djurdjev O, Levin A: Screen- 11. Arora P, Obrador GT, Ruthazer R, et al: Prevalence, ing for renal disease using serum creatinine: Who are we predictors, and consequences of late nephrology referral at a missing? Nephrol Dial Transplant 16:1042-1046, 2001 tertiary care center. J Am Soc Nephrol 10:1281-1286, 1999 29. Swedko PJ, Clark HD, Paramsothy K, Akbari A: Serum 12. Harris LE, Luft FC, Rudy DW, Kesterson JG, Tierney creatinine is an inadequate screening test for renal failure in WM: Effects of multidisciplinary case management in patients elderly patients. Arch Intern Med 163:356-360, 2003 with chronic renal insufficiency. Am J Med 105:464-471, 1998 30. Levin A, Stevens L, McCullough PA: Cardiovascular 13. Kausz AT, Khan SS, Abichandani R, et al: Management disease and the kidney. Tracking a killer in chronic kidney of patients with chronic renal insufficiency in the northeastern disease. Postgrad Med 111:53-60, 2002 United States. J Am Soc Nephrol 12:1501-1507, 2001 31. Tonelli M, Gill J, Pandeya S, Bohm C, Levin A, Kiberd 14. Ghossein C, Serrano A, Rammohan M, Batlle D: The BA: Barriers to blood pressure control and angiotensin enzyme role of comprehensive renal clinic in chronic kidney disease inhibitor use in Canadian patients with chronic renal insuffi- stabilization and management: The northwestern experi- ciency. Nephrol Dial Transplant 17:1426-1433, 2002 ence. Semin Nephrol 22:526-532, 2002 32. Stevens LA, Levey AS: Clinical implications of esti- 15. Lee TH: Proving and improving the value of consul- mating equations for glomerular filtration rate. Ann Intern tations. Am J Med 113:527-528, 2002 Med 141:959-961, 2004 16. Levin A: Identification of patients and risk factors in 33. Yeoh HH, Rasgon S, Rutkowski M, Crooks P, Idroos chronic kidney disease—Evaluating risk factors and thera- M, Diesto J: Management of patients with chronic kidney peutic strategies. Nephrol Dial Transplant 16:57-60, 2001 disease at Kaiser Permanente. Nephrol News Issues 17:25- 17. Pogach LM, Hawley G, Weinstock R, et al: Diabetes 27, 30, 2003 prevalence and hospital and pharmacy use in the Veterans 34. Akbari A, Swedko PJ, Clark HD, et al: Detection of Health Administration (1994). Use of an ambulatory care chronic kidney disease with laboratory reporting of esti- pharmacy-derived database. Diabetes Care 21:368-373, 1998 mated glomerular filtration rate and an educational program. 18. Heisler M, Bouknight RR, Hayward RA, Smith DM, Arch Intern Med 164:1788-1792, 2004 Kerr EA: The relative importance of physician communica- 35. Coresh J, Astor BC, McQuillan G, et al: Calibration tion, participatory decision making, and patient understand- and random variation of the serum creatinine assay as ing in diabetes self-management. J Gen Intern Med 17:243- critical elements of using equations to estimate glomerular 252, 2002 filtration rate. Am J Kidney Dis 39:920-929, 2002 19. Fisher SG, Weber L, Goldberg J, Davis F: Mortality 36. Rule AD, Larson TS, Bergstralh EJ, Slezak JM, ascertainment in the veteran population: Alternatives to the Jacobsen SJ, Cosio FG: Using serum creatinine to esti- National Death Index. Am J Epidemiol 141:242-250, 1995 mate glomerular filtration rate: Accuracy in good health 20. Dominitz JA, Maynard C, Boyko EJ: Assessment of and in chronic kidney disease. Ann Intern Med 141:929- vital status in Department of Veterans Affairs national data- 937, 2004 Earlier Diagnosis of Autosomal Dominant Polycystic Kidney Disease: Importance of Family History and Implications for Cardiovascular and Renal Complications

Matthew Taylor, MD, PhD, Ann M. Johnson, MS, Maryellyn Tison, NP, Pamela Fain, PhD, and Robert W. Schrier, MD

● Background: Autosomal dominant polycystic kidney disease (ADPKD) is a common and serious cause of hereditary renal disease. The emerging possibilities to intervene early in the disease course elevate the importance of both accurate and early diagnosis of ADPKD. Family history analysis is a simple and inexpensive approach to identifying individuals at risk for ADPKD. We hypothesized that advances in knowledge of and potential interven- tions for ADPKD have led to increased use of family history screening. Methods: We distributed surveys to 1,527 subjects from our ADPKD research database to determine the extent to which examination of family history was used to diagnose ADPKD, by birth cohort. Results: Six hundred thirty-seven subjects with ADPKD (42%) completed and returned surveys. Family history analysis led to the initial ADPKD diagnosis in 49% of all subjects overall. In the birth-cohort analysis, ADPKD was more likely to have been diagnosed in individuals born between 1951 and 1974 because of family history (55% versus 38%; P < 0.0002) and patients were younger at diagnosis (27 versus 39 years; P < 0.0001) than individuals born before 1951. Conclusion: In a large cohort of subjects with ADPKD, we found increased use of family history analysis as a tool for diagnosing ADPKD and earlier age of diagnosis in the more recent birth cohort. This trend may reflect increased overall awareness of ADPKD by physicians, as well as encouraging hypertension and proteinuria treatment outcome data that may depend on intervening early in the course of disease. Am J Kidney Dis 46:415-423. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Autosomal dominant polycystic kidney disease (ADPKD); diagnosis; hypertension.

uria, older age, hematuria, mutations in PKD1 Editorial, p. 557 (in comparison to mutations in PKD2), hyperten- sion, and increased left ventricular mass index UTOSOMAL DOMINANT polycystic kid- (LVMI).6,7 The opportunity to intervene in the A ney disease (ADPKD) is a common heredi- natural history of the disease by targeting modifi- tary renal disease occurring in between 1 in 400 able aspects of these indicators, such as antihyper- and 1,000 persons.1,2 Mutations in either the poly- tensive therapy for hypertension, which occurs cystic kidney disease (PKD) type 1 (PKD1) gene earlier in patients with ADPKD than in those on chromosome 16 or the PKD2 gene on chromo- with essential hypertension, or blockade of the some 4 account for approximately 85% and 15% of renin-angiotensin-aldosterone system to reduce affected families, respectively.ADPKD is character- LVMI,8 argues for the need to diagnose ADPKD ized by multiple fluid-filled cysts and subsequent early in its course. Epidemiological studies found interstitial fibrosis developing in both kidneys, lead- ing to renal failure in 50% of affected persons by 3 the age of 60 years and accounting for 2.3% of From the University of Colorado Health Sciences Center, end-stage renal disease in the United States be- Denver, CO. tween 1997 and 2001.4 In addition to renal cystic Received March 31, 2005; accepted in revised form May disease, hepatic cysts, mitral valve prolapse, left 31, 2005. Originally published online as doi:10.1053/j.ajkd.2005.05.029 ventricular hypertrophy (LVH), and cerebral aneu- on July 25, 2005. rysms also may occur.5 Supported in part by grant no. P01 DK34039 from the Despite the serious nature of the condition, National Institute of Diabetes and Digestive and Kidney only limited ADPKD research was performed Diseases; grants no. MO1 RR00051 and M01 RR00069 from the General Clinical Research Centers Program of the historically. However, the past 20 years have National Center for Research Resources, The National Insti- witnessed a substantial increase in ADPKD stud- tutes of Health; and the Zell Family Foundation. ies, including several large, prospective, and on- Address reprint requests to Robert W. Schrier, MD, PKD going family-based studies around the world. Research, UCHSC, C283, 4200 East 9th Ave, Denver, CO 80262. E-mail: [email protected] Prognostic indicators for a more rapid decline in © 2005 by the National Kidney Foundation, Inc. renal function now have been identified, includ- 0272-6386/05/4603-0004$30.00/0 ing male sex, African-American race, protein- doi:10.1053/j.ajkd.2005.05.029

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 415-423 415 416 TAYLOR ET AL significant associations between slowed progres- their family history and would be diagnosed at an sion of renal disease and lower mean arterial earlier age than in persons from earlier cohorts. pressure and increased use of angiotensin- converting enzyme (ACE) inhibitors.7 One pro- spective 7-year randomized study showed that METHODS rigorous control of hypertension in hypertensive For 20 years, the PKD Research group at the University of subjects with ADPKD (blood pressure goal, Colorado (Denver, CO) has been studying affected patients with ADPKD and their families. Our database contains clinical 125/75 mm Hg) decreased LVMI to a greater 9 information from more than 400 participating ADPKD fami- extent than standard therapy (Ͻ140/90 mm Hg). lies. There are an additional 443 families with ADPKD that A positive family history can readily identify have requested to be on our mailing list; clinical data for these family members at risk for ADPKD. However, patients are limited. there is concern in some settings (predominantly We developed a 6-page survey that was distributed to 1,527 study subjects listed as having ADPKD in our data- cancer genetics) that many physicians, both gen- base. The survey asked 8 basic demographic questions and eralists and subspecialists, underutilize family specific questions related to ADPKD diagnosis, symptoms, history screening.10–16 The clinical application renal function, and resources used by patients to learn about of family history analysis is not well described in ADPKD (Table 1). The survey was sent in a single mailing patients with ADPKD, and it is unknown whether (November 2002) with instructions to return the survey using a provided return envelope. Statistical analysis was increases in basic knowledge of ADPKD corre- performed using SAS, version 7 (SAS Institute, Cary, NC). late with increased use of family history screen- Categorical variables were compared between groups by ing. We hypothesized that advances in knowl- using chi-square analysis; continuous variables, by using edge of ADPKD and an understanding of treatable t-tests or analysis of variance. Survey subjects were divided prognostic factors (eg, hypertension) would lead into 3 categories based on year of birth: 1950 or earlier, 1951 to 1974, and 1975 or later. These groups were chosen so that to increased use of family history analysis in the each category would span approximately 25 years (the identification and early diagnosis of ADPKD in earliest birth cohort spans a longer interval than the later 2 individuals. Furthermore, the recent availability cohorts). of clinical molecular genetic testing means such testing may be applied to asymptomatic at-risk RESULTS relatives of subjects with ADPKD. The direct Six hundred thirty-seven subjects with ADPKD sequencing method (Athena Diagnostics Inc, (42%) completed and returned surveys. Of 890 Worcester, MA), which typically requires testing surveys not completed, we learned that 14 sub- of an initial affected family member before test- jects had died and 116 surveys were returned by ing of relatives, is expected to elevate further the the post office because the subject no longer importance of family history as a risk assessment resided at the address in our database. Thus, our tool in ADPKD families. Specifically, during the overall response rate for the 1,397 surveys that previous 3 decades, we expected to see an in- presumably could have reached the intended crease in use of family history screening, which study subject was approximately 46%. Relatives would identify patients with ADPKD earlier in filled out 73 surveys: 68 for a minor child, 3 for the course of their disease. To study this hypoth- an incapable adult, and 2 for a deceased family esis, we distributed a survey to more than 1,500 member. Women accounted for 66% of respon- subjects with ADPKD to determine the role of dents, and 84% of respondents identified them- family history information in the diagnosis of selves as non-Hispanic whites. Fifty-nine sur- ADPKD in each subject. veys were excluded from analysis because the Our principal hypothesis for this study is that respondents could not recall their year of diagno- increased awareness of ADPKD and the genetic sis or most of the survey was not completed. This nature of the condition in recent years has led to left 578 surveys for analysis. the diagnosis of ADPKD in more subjects through Results for those respondents are listed in evaluations prompted by family history, rather Table 2. Respondents were predominantly white than delayed diagnoses prompted by the onset of (84%) females (66%). Mean age of subjects specific symptoms. We predicted that ADPKD was 46 years, with a mean age of diagnosis of would be diagnosed more frequently in persons ADPKD of 28 years. Family history analysis from more recent birth cohorts by exploration of led to the initial ADPKD diagnosis in 49% of EARLIER DIAGNOSIS OF ADPKD 417

Table 1. ADPKD Survey Questions

Questions Possible Responses (partial text of responses)

Diagnosis of PKD When did a physician diagnose you with PKD? (Respondent indicated month and year, if known) Why were you diagnosed with PKD? Saw a doctor because of PKD symptoms Screened for PKD because of family history of PKD An abnormality (laboratory/physical examination) was identified during a regular physician visit Other Were you the first person in your family to be Yes/no diagnosed with PKD? First PKD symptom Check the box that best corresponds to the (List of symptoms provided, plus “Other”) first PKD symptom you had First PKD symptom leading to physician visit (List of symptoms provided, plus “Other”) PKD symptoms (ever had) The following is list of PKD symptoms. Please (List of symptoms provided, plus “Other”) check all symptoms that apply to you Time to diagnosis of PKD after occurence of symptoms Who diagnosed you with PKD? Family physician (internal medicine/family practice) Kidney specialist (nephrologist) Other specialist (indicate what specialty) Diagnosed with PKD as part of a research study Dialysis and transplantation Are you currently receiving dialysis treatment? Yes/no Have you received a kidney transplant(s)? Yes/no Resources used by patients What resources have you found useful for My family physician (primary care) learning about PKD? My specialist (example: nephrologist/kidney physician or other specialist) Books Magazines Internet Other patients or family members PKD support groups None of the above Other (describe) What single resource have you found most (Same list as above) useful for learning about PKD? subjects, symptoms or signs recognized by the 39%; P ϭ 0.004). A greater percentage of male patient were reported in 36% of cases, and the than female patients with ADPKD reported hyper- remaining diagnoses were because of abnor- tension as a reason for first seeing a physician, malities noted by a physician (15%). Hyperten- but this did not reach statistical significance sion was the most commonly reported initial (36% versus 28%; P ϭ 0.077). ADPKD tended sign of ADPKD and also the most common to be diagnosed in males on the basis of family reason for the initial physician visit that led to history less often than in females (44% versus the ADPKD diagnosis. Flank or back pain was 52%; P ϭ 0.069). There were no significant sex the most common initial symptom reported by differences in age of diagnosis or age at onset of patients. first ADPKD symptom(s). Males were more Sex likely to report ever having blood in their urine Males (n ϭ 194) were more likely than fe- as an ADPKD symptom (32% versus 24%; P ϭ males (n ϭ 384) to report hypertension as one of 0.024), but reported less flank pain (40% versus their first ADPKD symptoms/signs (52% versus 51%; P ϭ 0.019), nausea (11% versus 20%; P ϭ 418 TAYLOR ET AL

Table 2. Characteristics of 578 Survey Respondents Table 2 (Cont’d). Characteristics of 578 Survey With ADPKD Respondents With ADPKD

Variable Result Variable Result

Sex Ever had symptom/sign Male 194 (34) Hypertension 413 (71) Female 384 (66) Flank pain 272 (47) Race Kidney infection 187 (32) Non-Hispanic white 485 (84) Back pain 165 (29) Other 37 (6) Blood in urine 154 (27) No response 46 (10) Mass 124 (21) Age at time of survey (y) 46 Ϯ 16 Microscopic blood in urine 108 (19) Year of birth Abdominal pain 107 (19) Յ1950 200 (35) Urine problem 49 (8) 1951-1974 295 (51) Do not remember 4 (1) Ն1975 83 (14) No symptoms 47 (8) Age at ADPKD diagnosis (y) 28 Ϯ 14 Others (%) Age at first ADPKD symptom (y) (n ϭ 450) 28 Ϯ 13 Headache 19 Reason for diagnosis Nausea 17 Family history 284 (49) Heart murmur 16 Symptoms/sign noted by patient 207 (36) Shortness of breath 16 Other abnormality/sign noted by Loss of appetite 13 physician 87 (15) Kidney stone(s) 13 First person diagnosed with ADPKD in Ovarian cysts 12 family? Uterine prolapse 5 Yes 95 (16) Stroke 2 No 469 (81) Seizure(s) 1 No answer or unknown 14 (2) First ADPKD symptom/sign* NOTE. Values expressed as number of patients (per- Hypertension 250 (43) cent) or mean Ϯ SD, unless noted otherwise. Flank pain 105 (18) *Twenty percent of subjects selected more than 1 re- Back pain 58 (10) sponse. Kidney infection 57 (10) †Thirteen percent of subjects selected more than 1 Blood in urine 47 (8) response. Abdominal pain 40 (7) Mass 28 (5) Urine problem 22 (4) 0.013), kidney infections (18% versus 34%; P Ͻ Microscopic blood in urine 21 (4) 0.0001), or noting an abdominal mass (16% Do not remember first symptom 22 (4) versus 24%; P ϭ 0.039). No symptom 52 (9) Others: loss of appetite, nausea, weight Signs and Symptoms loss, diarrhea, kidney stone(s), proteinuria, ruptured cyst(s), other Whereas hypertension was the most com- pain monly reported first sign of ADPKD (43%), First symptom/sign leading to physician flank or side pain was reported as the most visit† common first symptom (18%; Table 2). Eight Hypertension Flank pain 178 (31) percent of respondents had not experienced a Blood in urine 69 (12) PKD symptom or sign, 16% experienced a single Kidney infection 53 (9) symptom or sign, 37% reported 2 to 4 symptoms Back pain 44 (8) or signs, 21% reported 5 to 7 symptoms or signs, Abdominal pain 32 (6) and 17% reported experiencing more than 7 Urine problem 32 (6) Mass 24 (4) symptoms or signs. Seventy-one percent re- Microscopic blood in urine 23 (4) ported having ever been given a diagnosis of Do not remember 13 (2) hypertension. No symptom leading to physician visit 28 (5) Others: as above, plus intracranial Care and Resources aneurysm 147 (25) Family physicians (33%) and nephrologists (Continued) (28%) were more likely to make the diagnosis of EARLIER DIAGNOSIS OF ADPKD 419

ADPKD than other specialists (17%) or research- 25% of those at risk in the most recent cohort ers (11%). The most useful resource for learning (instead of the expected 50%), this cohort cannot about ADPKD, chosen by 34% of respondents, be directly compared with earlier cohorts be- was their specialist (usually their nephrologist). cause not all expected subjects with ADPKD are Other useful resources were other patients or accounted for. family members (20%), the Internet (14%), sup- However, statistical comparisons can be made port groups (11%), books (9%), their family between the 2 older cohorts (Table 3), for which physician (5%), and magazines (3%). Seventeen ascertainment of affected subjects with ADPKD percent of subjects listed more than 1 resource as is more complete. Respondents born before 1951 most useful, 20% listed another resource not on were older at the time of diagnosis, were less the list, 2.7% did not select a resource, and 2.5% likely to have been diagnosed because of family of respondents indicated that they had not found history, and more frequently reported hyperten- a useful resource for information about PKD. sion as the first indication of their disease than respondents born between 1951 and 1974. Fam- Birth Cohorts ily history led to the diagnosis of ADPKD more Survey respondents were divided into 3 birth frequently in the 1951 to 1974 cohort (55% cohorts (date of birth: Յ1950; 1951 to 1974; and versus 38%), a finding consistent with our hypoth- Ն1975). The sensitivity of ultrasound for the esis of increased use of family history screening diagnosis of ADPKD approaches 100% after the for the more recent birth cohort (Table 3). age of 30 years.17 Most individuals at risk for We further predicted that increased attention ADPKD born after 1974 are younger than 30 to family history would lead to the diagnosis of years, and their affected status cannot be reliably ADPKD in more individuals who are asymptom- scored as negative with a normal ultrasound atic or in an early stage of their disease. Figure 1 result because they could still develop cystic shows survival curves to diagnosis for the 2 disease before the age of 30 years. In addition, cohorts. There was a significantly older age at many at-risk individuals may not have been diagnosis for the earlier cohort (Fig 1A; P Ͻ screened for ADPKD. To explore the possibility 0.0001). Because the disease has autosomal domi- that our survey approach may be surveying an nant inheritance and had been diagnosed in 57% incomplete ADPKD population in the most re- and 43% of at-risk relatives in our largest fami- cent cohort (because ADPKD is not yet diag- lies in the pre-1951 and 1951 to 1974 birth nosed in some younger individuals), we ana- cohorts, we believe we have complete or nearly lyzed the proportion of ADPKD diagnoses in complete ascertainment of affected individuals, at-risk individuals in 233 of our families, account- respectively. To exclude the possibility that the ing for more than 2,400 reported at-risk family lack of any respondent older than 50 years in the members born after 1930. The 233 families had 1951 to 1974 cohort had biased our findings, we at least 1 family member returning the survey generated a set of hypothetical respondents for and complete pedigree information on diagnosed this cohort. Using age of distribution for diagno- and nondiagnosed family members. Four hun- sis of the pre-1951 cohort, we created 65 hypo- dred six survey respondents were from these 233 thetical respondents (ie, subjects for whom families; 173 respondents were not. Compared ADPKD was diagnosed after the age of 50 years) with the expected rate of diagnosis of 50% in for the 1951 to 1974 cohort. The observed differ- these at-risk relatives, we found 57%, 43%, and ence in age of diagnosis between the cohorts 25% affected for the 3 birth cohorts of 1930 to persisted in this second analysis (Fig 1B). 1950, 1951 to 1974, and 1975 or later, respec- tively. This suggests that ascertainment of infor- DISCUSSION mation about ADPKD diagnosis may be incom- Our primary hypothesis is that ADPKD would plete in the 2 more recent cohorts; some of this be diagnosed in more recent birth cohorts at a effect presumably is present because of at-risk younger age based on family history. Our study members in the more recent cohorts for whom is the first to show that in individuals with ADPKD will be diagnosed in years to come. ADPKD born between 1951 and 1974, ADPKD Because ADPKD has been diagnosed in only was more likely to have been diagnosed because 420 TAYLOR ET AL

Table 3. Characteristics of 495 Survey Respondents With ADPKD Born Before 1975, by Birth Cohort

Variable Born Յ1950 (n ϭ 200) Born 1951-1974 (n ϭ 295) P

Demographics Male (%) 32 31 0.6978 Non-Hispanic white (%) 88 84 0.2879 Age at diagnosis of ADPKD (y) 39 Ϯ 13 27 Ϯ 9 Ͻ0.0001 Age at first sign and/or symptom (y) 36 Ϯ 12 (N ϭ 167) 27 Ϯ 9(Nϭ 235) Ͻ0.0001 Reason for diagnosis (%) Family history 38 55 Ͻ0.0002 (2 ϫ 3 table) Symptom(s) 39 36 Abnormality/sign 23 9 First sign/symptom (%) Hypertension (alone) 41 35 0.0209 (2 ϫ 5 table) Hypertension ϩ others 14 8 Other than hypertension 40 45 None 3 8 Do not remember 2 4 Hypertension one of first symptoms (%) 55 43 0.0112 Ever reported hypertension (%) 82 75 0.1010

NOTE. Values expressed as mean Ϯ SD or percent. of family history (55% versus 38%; P Ͻ 0.0002) pends heavily on the completeness of ultrasound and individuals were younger at diagnosis (27 screening of relatives. In our survey, flank or side versus 39 years; P Ͻ 0.0001) than those born pain was a presenting symptom in 18% of respon- before 1951. These results are compatible with dents. There are multiple causes of pain in pa- the improved recognition by physicians and pa- tients with ADPKD, most often attributed to the tients of the autosomal dominant inheritance of presence of renal cystic disease,21,23,24 which ADPKD. These data also support the notion that likely is preceded by a presymptomatic or less advances in awareness by clinicians and patients symptomatic stage when cysts are present (and of the genetic basis of ADPKD has led to in- detectable by ultrasound), but have not caused creased use of family history to identify individu- sufficient pain symptoms to lead to diagnosis. als at risk for ADPKD from affected families. As The inverse correlation in our study between a result, ADPKD is more likely to be diagnosed increased use of family history in the initial in individuals from more recent birth cohorts diagnosis of ADPKD and younger age of diagno- when they are asymptomatic. As with most renal sis in the 1951 to 1974 cohort may indicate that diseases, early diagnosis with implementation of clinicians are becoming more adept at diagnos- effective interventions (eg, antihypertensive ing the condition earlier in its course. therapy) has the best chance for preventing or With the availability of treatment for end- slowing renal progression and cardiac complica- stage renal disease, cardiovascular deaths have tions in patients with ADPKD. emerged as the major cause of mortality in pa- A detailed analysis of family history is an tients with ADPKD.25 It now is known that obvious and inexpensive diagnostic approach to hypertension, a major cardiovascular risk factor, identifying relatives at risk for ADPKD, but occurs much earlier in adults with ADPKD than trends in the use of this approach have not been those with essential hypertension26; thus, a fam- well studied in patients with ADPKD. Pain symp- ily history of ADPKD should prompt earlier toms reportedly are the most common diagnostic hypertension screening. Many children with indicators of ADPKD, ranging in frequency from ADPKD have blood pressures greater than the 21% to 37%.18-22 For patients with ADPKD, a 95th percentile for children of their age.27 In positive family history of other affected relatives both adults28 and children,29 an increase in LVMI with ADPKD is present in approximately two has been associated with increased blood pres- thirds of patients,1 although this estimate de- sures. In 1 study, 41% of patients with ADPKD EARLIER DIAGNOSIS OF ADPKD 421

at a mean age of 40.7 years had LVH.28 The initial evaluation of newly discovered hyperten- sion in a young patient at risk for ADPKD often involves ultrasonography and can lead to the diagnosis of polycystic disease at an early stage. In these instances, the risk of early diagnosis and potential for health insurance discrimination (a poorly quantified risk at the moment, given only limited evidence to suggest discrimination is widespread30) increasingly appears to be offset by the benefit of initiating antihypertensive thera- pies. Novel ADPKD-specific therapies, includ- ing anti–epithelial growth factor receptor inhibi- tors,31 vasopressin receptor antagonists,32,33 rapamycin therapy,34 and caspase inhibitory methods,35 are being developed. These future applications are likely to further motivate clini- cians to identify ADPKD in its early stages, possibly even before hypertension develops. LVH has been well established as a major risk factor for cardiovascular morbidity and mortality secondary to systolic and diastolic ventricular dysfunction, arrhythmias, sudden deaths, and myocardial ischemia.36 In a 7-year prospective randomized study of 75 patients with ADPKD with hypertension and LVH, a blood pressure goal of 125/75 mm Hg was shown to have a significantly greater decrease in LVMI than the standard goal of less than 140/90 mm Hg.9 More- over, treatment with an ACE inhibitor, enalapril, was more effective than antihypertensive therapy with a calcium channel blocker, amlodipine.9 Early intervention in diagnosing and aggres- sively treating blood pressure in patients with Fig 1. (A) Kaplan-Meier survival curves of age at ADPKD therefore has the potential of prevent- diagnosis of ADPKD in survey respondents born in 1950 ing LVH, cardiovascular complications, and .(295 ؍ and between 1951 and 1974 (n (200 ؍ or earlier (n Curves show the percentage of subjects for whom mortality. The present survey indicates that ADPKD was not yet diagnosed at a given age. Curves early diagnosis of ADPKD by family history is were significantly different (P < 0.0001). (B) Kaplan- Meier survival curves of age at diagnosis of ADPKD for occurring more often in the 1951 to 1974 birth subjects in different birth cohorts. Curves show the cohort compared with the earlier cohort. The percentage of subjects for whom ADPKD was not yet availability of renal ultrasound to document diagnosed at a given age. The curve for respondents born in 1950 or earlier is unchanged from (A) and shows the presence of ADPKD in individuals with a data from 200 survey respondents. The curve for respon- family history of disease can be diagnostic as dents born between 1951 and 1974 shows age at diagno- early as in utero.37 sis data for 295 survey respondents plus 65 “dummy” respondents representing the worse-case scenario of Another advantage of earlier diagnosis of 18% as yet undiagnosed cohort members. The 65 dummy ADPKD is the ability to initiate inhibition of the respondents were assumed to have been given a diagno- renin-angiotensin-aldosterone system. It is clear sis of ADPKD after the age of 50 years, with a distribution similar to subjects born in 1950 or earlier who were given that patients with ADPKD show increased activ- a diagnosis of ADPKD after the age of 50 years. Curves ity of the renin-angiotensin-aldosterone system are significantly different (P < 0.0001). Median ages at compared with patients with essential hyperten- diagnosis were 39 years in those born in 1950 or earlier and 29 years in those born between 1951 and 1974. sion matched for sex, age, renal function, and 422 TAYLOR ET AL degree of hypertension.38 In addition to the ef- 7. Schrier RW, McFann KK, Johnson AM: Epidemiologi- fect of ACE inhibitors on LVH in patients with cal study of kidney survival in autosomal dominant polycys- ADPKD,9 there also is accumulating evidence tic kidney disease. Kidney Int 63:678-685, 2003 8. Ecder T, Schrier RW: Hypertension and left ventricular relative to a beneficial effect on the kidney. A hypertrophy in autosomal dominant polycystic kidney dis- harbinger of progression of renal disease in pa- ease. Expert Rev Cardiovasc Ther 2:369-374, 2004 tients with most chronic kidney diseases is an 9. Schrier R, McFann K, Johnson A, et al: Cardiac and increase in proteinuria.39 In this regard, a prospec- renal effects of standard versus rigorous blood pressure tive randomized 5-year study showed preven- control in autosomal-dominant polycystic kidney disease: Results of a seven-year prospective randomized study. J Am tion of increased albuminuria in patients with Soc Nephrol 13:1733-1739, 2002 ADPKD randomized to ACE-inhibitor therapy 10. Burke W: Genetic testing in primary care. Annu Rev compared with a progressive increase in protein- Genomics Hum Genet 5:1-14, 2004 uria in those treated with a calcium channel 11. Rich EC, Burke W, Heaton CJ, et al: Revitalizing the blocker.40 Also, patients with ADPKD treated family history in primary care. J Gen Intern Med 19:273- 280, 2004 with ACE inhibitors without diuretics who were 12. Rose PW, Watson E, Yudkin P, et al: Referral of followed up retrospectively showed a slower loss patients with a family history of breast/ovarian cancer— of creatinine clearance compared with patients GPs’ knowledge and expectations. Fam Pract 18:487-490, with ADPKD treated with diuretics without an 2001 ACE inhibitor (2.7 versus 5.3 mL/min/y; P Ͻ 13. Schroy PC III, Barrison AF, Ling BS, Wilson S, 0.05).41 Geller AC: Family history and colorectal cancer screening: A survey of physician knowledge and practice patterns. In summary, early diagnosis of ADPKD by us- Am J Gastroenterol 7:1031-1036, 2002 ing family history and the genetic knowledge of 14. Suther S, Goodson P: Barriers to the provision of ADPKD appears to be occurring and has important genetic services by primary care physicians: A systematic implications for altering the renal and cardiovascu- review of the literature. Genet Med 5:70-76, 2003 lar complications of the disease. In that regard, it 15. Crouch MA, Thiedke CC: Documentation of family health history in the outpatient medical record. J Fam Pract has been found that the age of end-stage renal 22:169-174, 1986 disease in both men and women with ADPKD has 16. Medalie JH, Zyzanski SJ, Goodwin MA, et al: Two increased in recent years, and this effect has been physician styles of focusing on the family. J Fam Pract associated with lower blood pressures and in- 49:209-215, 2000 creased therapy with ACE inhibitors.7 17. Nicolau C, Torra R, Badenas C, et al: Autosomal dominant polycystic kidney disease types 1 and 2: Assess- ment of US sensitivity for diagnosis. Radiology 213:273- REFERENCES 276, 1999 1. Dalgaard OZ: Bilateral polycystic disease of the kid- 18. Segal AJ, Spataro RF, Barbaric ZL: Adult polycystic neys: A follow-up of two hundred and eighty-four patients kidney disease: A review of 100 cases. J Urol 118:711-713, and their families. Acta Med Scand 158:S1-S255, 1957 1977 (suppl 328) 19. Delaney VB, Adler S, Bruns FJ, Licinia M, Segel DP, 2. Iglesias CG, Torres VE, Offord K, Holley KE, Beard Fraley DS: Autosomal dominant polycystic kidney disease: CM, Kurland LT: Epidemiology of adult polycystic kidney Presentation, complications, and prognosis. Am J Kidney disease, Olmsted County, Minnesota: 1935-1980. Am J Dis 5:104-111, 1985 Kidney Dis 2:630-639, 1983 20. Bourquia A: Societe marocaine des maladies renales 3. Ravine D, Walker RG, Gibson RN, et al: Phenotype [Autosomal dominant polycystic kidney disease in Mo- and genotype heterogeneity in autosomal dominant polycys- rocco. Multicenter study of about 308 families]. Nephrolo- tic kidney disease. Lancet 340:1330-1333, 1992 gie 23:93-96, 2002 4. US Renal Data System: USRDS 2001 Annual Data 21. Bajwa ZH, Sial KA, Malik AB, Steinman TI: Pain Report. The National Institutes of Health, National Institute patterns in patients with polycystic kidney disease. Kidney of Diabetes and Digestive and Kidney Diseases, Bethesda, Int 66:1561-1569, 2004 MD, 2001 22. Kheder A, Hamida B, el Younsi F, et al: [Dominant 5. Fick-Brosnahan G, Ecder T, Schrier RW: Polycystic polycystic renal disease. Study of 105 cases]. Rev Med kidney disease, in Schrier RW (ed): Diseases of the Kidney Interne 13:345-348, 1992 and Urinary Tract. Philadelphia, PA, Lippincott Williams & 23. Bajwa ZH, Gupta S, Warfield CA, Steinman TI: Pain Wilkins, 2001, pp 547–548 management in polycystic kidney disease. Kidney Int 60: 6. Gabow PA, Johnson AM, Kaehny WD, et al: Factors 1631-1644, 2001 affecting the progression of renal disease in autosomal- 24. Thanos A, Farmakis A, Davillas N: Spontaneous dominant polycystic kidney disease. Kidney Int 41:1311- rupture of the kidney: A cause of acute abdominal pain. Case 1319, 1992 report. Scand J Urol Nephrol 23:313-314, 1989 EARLIER DIAGNOSIS OF ADPKD 423

25. Fick GM, Johnson AM, Hammond WS, et al: Causes 34. Tao Y, Kim J, Schrier RW, Edelstein CL: Rapamycin of death in autosomal dominant polycystic kidney disease. markedly slows disease progression in a rat model of autoso- J Am Soc Nephrol 5:2048-2056, 1995 mal dominant polycystic kidney disease. J Am Soc Nephrol 26. Kelleher C, McFann K, Johnson A, Schrier R: Char- 16:46-51, 2005 acteristics of young adults with autosomal dominant polycys- 35. Tao Y, Kim J, Faubel S, et al: Caspase inhibition tic kidney disease compared with the general US population. reduces tubular apoptosis and proliferation and slows dis- Am J Hypertens 17:1029-1034, 2004 ease progression in polycystic kidney disease. Proc Natl 27. Fick GM, Duley IT, Johnson AM, et al: The spectrum Acad Sci U S A 102:6954-6959, 2005 of autosomal dominant polycystic kidney disease in chil- 36. Benjamin EJ, Levy D: Why is left ventricular hyper- dren. J Am Soc Nephrol 4:1654-1660, 1994 trophy so predictive of morbidity and mortality? Am J Med 28. Chapman AB, Johnson AM, Rainguet S, Hossack K, Sci 317:168-175, 1999 Gabow PA, Schrier RW: Left ventricular hypertrophy in 37. Pretorius DH, Lee ME, Manco-Johnson ML, Wein- autosomal dominant polycystic kidney disease. J Am Soc gast GR, Sedman AB, Gabow PA: Diagnosis of autosomal Nephrol 8:1292-1297, 1997 dominant polycystic disease in utero and in the young infant. 29. Ivy DD, Shaffer EM, Johnson AM, Kimberling WJ, J Ultrasound Med 6:249-255;1987 Dobin A, Gabow PA: Cardiovascular abnormalities in chil- 38. Chapman A, Johnson A, Gabow P, et al: The renin- dren with autosomal dominant polycystic kidney disease. angiotensin-aldosterone system and autosomal dominant J Am Soc Nephrol 5:2032-2036, 1995 polycystic kidney disease. N Engl J Med 323:1091-1096, 30. Golin CO, Johnson AM, Fick G, Gabow PA: Insur- ance for autosomal dominant polycystic kidney disease in 1990 patients prior to end-stage renal disease. Am J Kidney Dis 39. Venkat KK: Proteinuria and microalbuminuria in 27:220-223, 1996 adults: Significance, evaluation, and treatment. South Med J 31. Sweeney WES Jr, Chen Y, Nakanishi K, Frost P, 97:969-979, 2004 Avner ED: Treatment of polycystic kidney disease with a 40. Ecder T, Chapman AB, Brosnahan GM, et al: Effect novel tyrosine kinase inhibitor. Kidney Int 57:33-40, 2000 of antihypertensive therapy on renal function and urinary 32. Gattone VH, Wang X, Harris PC, Torres VE: Inhibition of albumin excretion in hypertensive patients with autosomal renal cystic disease development and progression by a vasopressin dominant polycystic kidney disease. Am J Kidney Dis 35: V2 receptor antagonist. Nat Med 9:1323-1326, 2003 266-271, 2000 33. Torres VE, Wang X, Qian Q, Somlo S, Harris PC, 41. Ecder T, Edelstein CL, Fick-Brosnahan G, et al: Gattone VH: Effective treatment of an orthologous model of Diuretics versus angiotensin-converting enzyme inhibitors autosomal dominant polycystic kidney disease. Nat Med in autosomal dominant polycystic kidney disease. Am J 10:363-364, 2004 Nephrol 21:98-103, 2001 Proteinuria and Hemoglobin Levels in Patients With Primary Glomerular Disease

Nina Mähr, MD, Ulrich Neyer, MD, Friedrich Prischl, MD, Reinhard Kramar, MD, Gert Mayer, MD, Florian Kronenberg, MD, and Karl Lhotta, MD

● Background: Anemia in association with nephrotic syndrome has been described in small patient series and case reports. Whether nephrotic patients develop anemia has not been formally investigated. Methods: We undertook a retrospective cross-sectional study of patients with biopsy-proven primary glomerular disease, various degrees of proteinuria, and creatinine levels less than 2 mg/dL (<177 ␮mol/L). In addition to proteinuria, values for hemoglobin (Hb), age, body mass index (BMI), serum albumin and protein, and estimated glomerular filtration rate (eGFR) were derived from patient charts. Results: We studied 297 patients, 187 men and 110 women, aged between 16 and 81 years. Univariate analysis showed no correlation between Hb level and proteinuria in either sex. Stratification of women and men into quartiles according to proteinuria showed no differences in Hb levels among the 4 groups. Three of 52 non-nephrotic women (6%) were anemic (Hb < 12 g/dL [<120 g/L]) compared with 11 of 58 nephrotic Multiple regression analysis of all patients showed Hb level to have a correlation with sex .(0.047 ؍ women (19%; P and borderline ;(0.028 ؍ negative correlation with age (P ;(0.005 ؍ P < 0.001), BMI (P < 0.001), and eGFR (P) ؍ In women, BMI showed a positive correlation with Hb level (P .(0.054 ؍ negative correlation with proteinuria (P (0.013 ؍ and eGFR (P (0.001 ؍ In men, BMI (P .(0.093 ؍ Proteinuria did not reach statistical significance (P .(0.030 was associated negatively with Hb level. Conclusion: These results (0.031 ؍ were associated positively and age (P indicate that nephrotic syndrome is not associated with anemia in men, but with a tendency to decrease Hb levels in women. Am J Kidney Dis 46:424-431. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Anemia; nephrotic syndrome; proteinuria; glomerulonephritis; hemoglobin.

EVERE PROTEINURIA exceeding 3.5 g/d thromboembolic events and infection.1 In addi- S of protein causes a number of symptoms tion, it was proposed that nephrotic syndrome and metabolic changes commonly called ne- may cause anemia.2,3 As the main mechanisms, phrotic syndrome. Patients with nephrotic syn- inappropriately low renal erythropoietin synthe- drome show hypoalbuminemia and hypoproteine- sis and urinary loss of erythropoietin have been 4 mia. Their cholesterol and triglyceride levels are postulated. In addition, renal loss of transferrin 5,6 markedly elevated. Other consequences are so- and iron might cause anemia. In patients with dium retention and edema and increased risk for diabetic nephropathy, which is the most common cause of nephrotic syndrome, anemia develops early in the course of the disease.7,8 The occur- rence of anemia in patients with nephrotic pri- From the Division of Nephrology and Department of mary glomerular disease has been described in Medical Genetics, Molecular and Clinical Pharmacology, Division of Genetic Epidemiology, Innsbruck Medical Uni- only small groups of patients or case reports, but versity, Innsbruck; Department of Nephrology and Dialysis, it is not yet clear whether anemia is a common Feldkirch Hospital, Feldkirch; and Third Department of feature of nephrotic syndrome.2,9-12 Therefore, Internal Medicine/Nephrology, Klinikum Kreuzschwestern to formally study a possible association between Wels, Wels, Austria. urinary protein loss and anemia, we undertook a Received February 24, 2005; accepted in revised form June 7, 2005. retrospective cross-sectional study of a larger Originally published online as doi:10.1053/j.ajkd.2005.06.002 number of patients with biopsy-proven primary on August 1, 2005. glomerular disease, well-preserved renal func- Supported in part by Vorarlberg Institute of Vascular tion, and various levels of proteinuria. Investigation and Treatment (N.M.); Herwig Wallmann (N.M.); a grant from Project 9331 from the Austrian Na- tional Bank (F.K.); and the Austrian Heart Fund (F.K.). METHODS Address reprint requests to Karl Lhotta, MD, Klinische Abteilung für Nephrologie, Medizinische Universität Patients Innsbruck, Anichstrasse 35, A-6020 Innsbruck, Austria. Patients were recruited at the 3 nephrology departments of E-mail: [email protected] Innsbruck Medical University, Feldkirch District Hospital, © 2005 by the National Kidney Foundation, Inc. and Klinikum Kreuzschwestern Wels. All patients older than 0272-6386/05/4603-0005$30.00/0 16 years who had undergone a renal biopsy within the last 20 doi:10.1053/j.ajkd.2005.06.002 years and had a serum creatinine level less than 2 mg/dL

424 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 424-431 PROTEINURIA AND HEMOGLOBIN LEVELS 425

(Ͻ177 ␮mol/L) were enrolled. Indications for biopsy were 0.90 0.38 0.94 0.66 1.05 1.35 0.60 0.84 0.69 0.38 1.26 proteinuria with or without hematuria and, in some patients, 0.88 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ isolated glomerular hematuria without significant protein Ϯ (g/dL) excretion. Only patients for whom a primary glomerular Albumin disease was diagnosed were included in the study. Exclusion criteria were such systemic diseases as diabetes, lupus, vasculitis, or amyloidosis. Patients with other conditions 1.30 3.69 0.92 3.38 0.97 2.99 0.84 4.15 1.18 3.63 1.45 2.78 1.01 3.74 1.05 3.43 0.98 2.80 0.71 4.13 1.36 3.30 that can cause anemia, such as malignancy, infectious dis- 1.15 2.87 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ ease, or recent blood loss, also were excluded from the Ϯ study. The following parameters were retrieved from patient charts: age, sex, body mass index (BMI), serum creatinine 3.00 6.79 2.11 6.03 3.92 5.63 2.56 7.03 5.99 6.55 4.98 5.87 3.45 6.94 4.30 6.44 4.24 5.63 1.42 7.04 5.14 6.54 level, hemoglobin (Hb) level, hematocrit, serum protein 3.65 5.77 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ level, and serum albumin level. Estimated glomerular filtra- Ϯ (g/d) Protein (g/dL) tion rate (eGFR/1.73 m2) was calculated by using the Modi- Proteinuria fication of Diet in Renal Disease 2 formula (eGFR ϭ 183 ϫ creatinine–1.154 ϫ age–0.203 [ϫ 0.742 if female]).13 Of ini- tially 516 patients eligible for the study, 219 patients were 1.51 3.65 1.89 5.13 1.65 5.91 1.43 2.21 1.92 6.43 1.33 6.93 1.7 2.85 1.09 4.30 1.53 6.54 1.51 1.97 0.94 5.14 excluded because of incomplete data (mainly missing body 1.53 5.23 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ weight, Hb level, or quantification of proteinuria). Of ex- Ϯ cluded patients, 75 were women and 144 were men. Mean Hb (g/dL) age was 35 years, and renal function was similar to that of included individuals. Proteinuria was either derived from a 24-hour urine collection (281 patients) or a protein- ) 2 28.53 14.55 17.75 13.30 26.07 14.69 22.84 14.84 25.85 14.86 17.80 15.55 18.38 13.05 24.67 13.32 22.81 13.05 24.10 13.63 24.45 13.64 creatinine ratio (16 patients), in milligrams per deciliter per 20.87 13.74 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ milligrams per deciliter, was calculated from a single urine Ϯ sample. For a subgroup of 90 patients, levels of serum iron, 1.73 m transferrin, and ferritin were available. eGFR (mL/min/ )

Statistical Analysis 2 3.74 67.91 8.05 69.23 3.64 81.30 4.31 72.27 4.76 75.63 4.22 88.23 2.93 68.26 1.44 58.96 3.59 75.74 4.70 78.44 6.22 70.47 4.41 83.38 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Statistical analysis was conducted using Statistical Pack- Ϯ age for the Social Sciences (SPSS Inc, Chicago, IL) for Windows 12.0. Because reference values for Hb differ between the sexes, most calculations were performed sepa- rately for women and men. Differences in Hb levels between the diagnostic groups were calculated by means of analysis 15.22 26.64 15.68 29.43 17.29 26.15 14.78 26.06 5.11 26.10 13.15 25.87 11.54 22.63 16.29 23.04 14.34 25.34 11.10 25.10 11.39 25.28 of variance. We calculated Spearman correlation coefficients 13.77 24.30 Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ Ϯ between Hb levels and various other variables. In addition, Ϯ to investigate nonlinear associations between proteinuria and Hb level, patients were stratified into 4 groups of almost equal size according to proteinuria per 24 hours: protein less than 1 g, 1 to 3.5 g, greater than 3.5 to 7 g, and greater than 7 g. Hb levels of those groups were compared by means of No. of Table 1. Patient Demographic Characteristics and Laboratory Values analysis of variance. The frequency of anemic individuals Patients Age (y) BMI (kg/m with and without nephrotic proteinuria was compared by using Fisher exact test. Multiple regression analysis was used to test the influence of the variables eGFR, BMI, age, serum albumin level, and proteinuria on Hb values.

RESULTS Hb and Histological Diagnosis A total of 297 patients, 110 women and 187 men, were included in the analysis. Mean age Diagnosis was 39.5 years (range, 16 to 81 years). Patients were divided into 6 groups according to histolog- ical diagnosis: minimal change nephropathy, fo- cal segmental glomerulosclerosis, immunoglobu- Others 18 41.72 Membranoproliferative glomerulonephritis 8 38.50 Membranous nephropathy 35 49.09 NOTE. To convert eGFR in mL/min to mL/s, multiply by 0.01667; Hb, protein, and albumin in g/dL to g/L, multiply by 10. Immunoglobulin A nephropathy 68 38.32 Focal segmental glomerulosclerosis 38 41.13 Minimal change nephropathy 20 33.35 Others 15 36.80 Membranoproliferative glomerulonephritis 5 47.40 Membranous nephropathy 24 43.46 Immunoglobulin A nephropathy 22 35.32 Focal segmental glomerulosclerosis 18 37.72 Minimal change nephropathy 26 36.19 Men 187 lin A nephropathy, membranous nephropathy, Women 110 426 MA¨ HR ET AL

Table 2. Bivariate Correlation of Hb Level With Other As the next step, we stratified patients into 4 Variables groups according to proteinuria (Tables 3 and 4).

Correlation Women Men If nephrotic syndrome is defined as proteinuria with protein greater than 3.5 g/d, 58 women and Age 86 men were nephrotic. Numbers of patients in r Ϫ0.066 Ϫ0.161 each group were similar, and groups did not P 0.490 0.027 BMI differ in eGFR levels. Comparison of Hb values r 0.179 0.067 in the 4 male proteinuria groups showed no P 0.082 0.379 significant differences (P ϭ 0.74). In women, Hb eGFR levels tended to be lower in nephrotic patients r 0.132 0.196 compared with those without severe proteinuria, P 0.166 0.007 ϭ Proteinuria but this difference was not significant (P r Ϫ0.154 Ϫ0.041 0.28). Three of 52 women (6%) without ne- P 0.106 0.578 phrotic proteinuria were anemic (Hb Ͻ 12 g/dL Protein [Ͻ120 g/L] according to the World Health Orga- r 0.124 0.103 nization definition) compared with 11 of 58 P 0.203 0.163 ϭ Albumin women (19%) with nephrotic syndrome (P r 0.071 0.106 0.047). Mean eGFRs were identical in anemic 2 P 0.484 0.169 and nonanemic women (75 mL/min/1.73 m [1.25 mL/s/1.73 m2]). In men, 10 of 101 non- NOTE. r ϭ Spearman correlation coefficient. Parame- nephrotic (10%) and 10 of 86 nephrotic subjects ters with statistical significance are indicated in bold. (12%) were anemic (Hb Ͻ 13 g/dL [Ͻ130 g/L]; P ϭ not significant). Mean eGFR of anemic men membranoproliferative glomerulonephritis, and was 64 mL/min/1.73 m2 (1.07 mL/s/1.73 m2) other. The last category mainly included patients compared with 78 mL/min/1.73 m2 (1.30 mL/s/ with non–immunoglobulin A mesangial glomer- 1.73 m2) in nonanemic men. These data confirm ulonephritis and Alport syndrome. Main demo- graphic data and laboratory values according to sex and diagnostic group are listed in Table 1. In women, Hb levels did not differ among diagnostic groups (P ϭ 0.4). In men, a signifi- cant difference among groups was observed be- cause of the lower Hb levels in patients with membranoproliferative glomerulonephritis (P ϭ 0.03). However, the number of patients in that group was small. Univariate Correlation Between Hb and Other Variables Correlation analysis between Hb levels and other variables is listed in Table 2. In women, no significant correlations were observed. BMI and eGFR showed a borderline positive correlation, and proteinuria, a borderline negative correla- tion, with Hb level. The same analysis in men showed a negative correlation between Hb level and age (r ϭ –0.161; P ϭ 0.027) and a positive correlation with eGFR (r ϭ 0.196; P ϭ 0.007). Figure 1 shows correlations between protein- Fig 1. Correlation of Hb level and proteinuria in (A) uria and Hb values in detail for female and male women and (B) men. To convert Hb in g/dL to g/L, study subjects. multiply by 10. PROTEINURIA AND HEMOGLOBIN LEVELS 427

Table 3. eGFR and Hb Level According to Quartiles of Proteinuria for Women

Proteinuria eGFR Group (g/d) No. of Patients (mL/min/1.73 m2) Hb (g/dL) Anemic Patients

Ͻ1 23 74.29 Ϯ 20.57 13.63 Ϯ 0.92 2 (8) 1-3.5 29 75.21 Ϯ 25.22 13.75 Ϯ 1.42 1 (3) 3.5-7 33 77.60 Ϯ 23.89 13.28 Ϯ 1.75 6 (18) Ͼ7 25 74.11 Ϯ 23.46 13.12 Ϯ 1.45 5 (20) Total 110 75.27 Ϯ 23.45 13.42 Ϯ 1.46 14 (13)

NOTE. Values expressed as number of patients, mean Ϯ SD, or number (percent). Women were considered anemic when Hb level was less than 12 g/dL (World Health Organization definition). To convert eGFR in mL/min to mL/s, multiply by 0.01667; Hb in g/dL to g/L, multiply by 10. our results of univariate (as discussed) and multi- between women and men, we decided to perform variate (discussed next) analysis that Hb level is regression analysis stratified for sex. In women, associated with eGFR in men, but not women. BMI showed a significant influence on Hb level Nineteen patients were receiving treatment (P ϭ 0.03). Proteinuria did not reach statistical with either an angiotensin-converting enzyme significance (P ϭ 0.093). In men, Hb values inhibitor or angiotensin receptor blocker, whereas were associated significantly with BMI (P ϭ 137 patients were definitely not treated. In the 0.001), eGFR (P ϭ 0.013), and age (P ϭ 0.031), remaining individuals, no exact information was but not protein excretion (P ϭ 0.26). available. Hb levels did not differ between pa- tients with or without such treatment (14.4 Ϯ 2.1 Proteinuria and Iron Metabolism g/dL [144 Ϯ 21 g/L] versus 14.4 Ϯ 1.6 g/dL [144 Parameters for iron metabolism, namely, se- Ϯ16 g/L]). In addition, no nephrotic subject was rum iron, transferrin, transferrin saturation, treated with immunosuppressive drugs that may and ferritin, were available for 44 female and cause anemia, such as, cyclophosphamide or myco- 46 male patients. Serum iron level was not phenolate. associated with proteinuria. However, there was a strong negative correlation between se- Multiple Regression Analysis rum transferrin level and proteinuria (P ϭ 0.03 Results of multiple regression analysis are for women, P ϭ 0.006 for men). Transferrin listed in Table 5. Primarily, we analyzed the level also was associated closely with serum entire patient group and observed sex as the most albumin level (P Ͻ 0.001 in either sex). In important variable related to Hb level (P Ͻ women, ferritin levels did not correlate with 0.001). In addition, BMI and GFR showed a posi- proteinuria. However, in men, serum ferritin tive association with Hb level, and age, a negative levels increased slightly with proteinuria (P ϭ correlation. The correlation between proteinuria 0.02). These results are shown in Fig 2. How- and Hb level was borderline (P ϭ 0.054). Be- ever, Hb levels did not correlate with iron, cause of these major differences in Hb levels transferrin, or ferritin levels in either sex, as

Table 4. eGFR and Hb According to Quartiles of Proteinuria for Men

Proteinuria eGFR Anemic Group (g/d) No. of Patients (mL/min/1.73 m2) Hb (g/dL) Patients

Ͻ1 43 79.80 Ϯ 27.14 14.78 Ϯ 1.42 3 (7) 1-3.5 58 76.96 Ϯ 21.91 14.96 Ϯ 1.56 7 (12) 3.5-7 43 73.99 Ϯ 25.75 14.59 Ϯ 1.85 5 (12) Ͼ7 43 74.47 Ϯ 25.88 14.80 Ϯ 1.75 5 (12) Total 187 76.26 Ϯ 24.93 14.80 Ϯ 1.64 20 (11)

NOTE. Values expressed as number of patients, mean Ϯ SD, or number (percent). Men were considered anemic when Hb level was less than 13 g/dL (World Health Organization definition). To convert eGFR in mL/min to mL/s, multiply by 0.01667; Hb in g/dL to g/L, multiply by 10. 428 MA¨ HR ET AL

Table 5. Results of Multiple Linear Regression Analysis With Hb Level as Dependent Variable

Variables Coefficient SE Significance (P) Change in R2

Total group (women and men) Sex (0 ϭ women, 1 ϭ men) 1.129 0.197 0.001 0.137 BMI 0.103 0.023 0.001 0.041 eGFR 0.013 0.004 0.005 0.028 Age –0.017 0.008 0.028 0.015 Proteinuria –0.041 0.021 0.054 0.011 Women BMI 0.072 0.033 0.030 0.053 Proteinuria –0.058 0.034 0.093 0.028 Men BMI 0.109 0.039 0.001 0.048 eGFR 0.015 0.006 0.013 0.041 Age –0.020 0.009 0.031 0.025

NOTE. Variables not in the model for women and men combined and women: age (P ϭ 0.20) and GFR (P ϭ 0.10). Variables not in the model for men: proteinuria (P ϭ 0.26). listed in Table 6. Transferrin saturation corre- DISCUSSION lated positively with Hb level in men (P ϭ 0.045). Of 11 nephrotic women with anemia, Results of our investigation suggest that the serum ferritin levels were available for 7 association between nephrotic proteinuria and women. All were in the normal range, exclud- the development of anemia might be less strong ing iron deficiency as the cause of anemia than previously thought.2,3 We found that, with (mean, 45.6 ␮g/L; range, 18 to 76 ␮g/L; nor- the possible exception of the small group of 8 mal range for women, 8 to 128 ␮g/L). men with membranoproliferative glomerulone-

Fig 2. Parameters of iron metabolism, proteinuria, and serum albumin. To convert transferrin in mg/dL to g/L, multiply by 0.01; albumin in g/dL to g/L, multiply by 10. PROTEINURIA AND HEMOGLOBIN LEVELS 429

Table 6. Spearman Correlation Coefficients of Iron ever, to detect a true influence of proteinuria, Metabolism Parameters Versus Proteinuria and nephrotic patients must be compared with pa- Versus Hb level tients who have glomerular disease, but without Proteinuria Hb severe proteinuria, as in this study. Otherwise, observed decreases in Hb levels might be caused Women not only by proteinuria, but also by kidney dis- Iron Ϫ Ϫ ease per se. Furthermore, some patients in the r 0.101 0.019 2 P 0.510 0.904 study by Vaziri et al had markedly elevated Transferrin serum creatinine levels and a considerable propor- r Ϫ0.327 Ϫ0.122 tion of them had diabetic nephropathy. P 0.030 0.447 In contrast to our findings for primary glomer- Ferritin Ϫ Ϫ ular disease, anemia is found frequently in pa- r 0.066 0.019 7,8,14,15 P 0.672 0.910 tients with diabetic nephropathy. Bosman 7 Transferrin saturation et al reported anemia in 13 of 27 patients with r 0.074 0.020 type 1 diabetes compared with 0 of 26 patients P 0.634 0.901 with glomerulonephritis and similar levels of Men Iron renal function and proteinuria. Likewise, anemia r Ϫ0.057 0.225 also occurs in the early stages of nephropathy in 8,15 P 0.697 0.116 patients with type 2 diabetes. Similar to our Transferrin observations, female sex was a risk factor for the r Ϫ0.396 Ϫ0.051 development of anemia. There is compelling P 0.006 0.739 Ferritin evidence that autonomic neuropathy causing sym- r 0.335 0.184 pathetic denervation of the kidneys has an impor- 7 P 0.023 0.220 tant role in early anemia development. Auto- Transferrin saturation nomic neuropathy interferes with sympathetic r 0.157 0.297 stimulation of renal erythropoietin production P 0.299 0.045 and therefore induces anemia. This mechanism NOTE. Parameters with statistical significance are indi- obviously is not active in patients with primary cated in bold. renal disease. Another possible explanation is that interstitial damage, which occurs early in phritis, for whom Hb levels were lower than patients with diabetic nephropathy, reduces eryth- those in other diagnostic groups, there is virtu- ropoietin production.14 ally no association in male patients. In women, It has been suggested that urinary loss of univariate and multivariate regression analysis erythropoietin may cause anemia in patients with showed a trend toward lower Hb values with nephrotic syndrome.2,4,12 Erythropoietin has a increasing proteinuria, which did not reach sig- molecular weight of 30.4 kd and therefore should nificance. Multivariate analysis showed that the undergo glomerular filtration to some extent. influence of proteinuria was weak (R2 ϭ 0.028, However, its molecule is highly glycosylated, ie, ϳ3% of Hb levels can be explained by amount and the side chains with negatively charged sialic of proteinuria). In addition, the frequency of acid residues protect it from glomerular filtra- nephrotic women with anemia (Hb Ͻ 12 g/dL tion.16 Increased urinary erythropoietin excre- [Ͻ120 g/L]) was significantly greater than the tion has been described in nephrotic rats4 and 2 frequency of anemic women without nephrotic studies of humans.2,11 However, 2 other groups proteinuria. Together, these results imply that in failed to detect erythropoietin in urine of patients women, nephrotic syndrome shows only a small with proteinuria.9,17 In addition, the amount of association with lower Hb levels and anemia. erythropoietin in urine described by Vaziri et al2 No other studies of similar size have been (1.000 to 3.000 mU/g creatinine) and Bayes et conducted on the association between nephrotic al12 (8 to 22 mU/mL), although probably not syndrome and anemia. In addition, reports pub- reflecting the total amount of filtered hormone, is lished to date compared Hb values of nephrotic low in relation to the amount present in extracel- patients with those of healthy controls.2,9 How- lular fluid (normal serum levels, 10 to 20 mU/ 430 MA¨ HR ET AL mL). Moreover, such urine losses should be and men is unknown. A separate analysis of compensated easily by renal erythropoietin pro- female nephrotic patients who were anemic duction.2,12 Furthermore, the prompt and effec- showed that they were relatively young, with a tive response to recombinant human erythropoi- mean age of 34 years. It therefore is possible that etin in nephrotic patients with anemia argues menstrual blood loss contributed to the develop- against the influence of urinary erythropoietin ment of anemia. However, all had normal ferritin losses.9-11,16 levels, which excludes iron deficiency as the Several investigators described low erythropoi- primary cause of anemia. etin levels despite marked anemia in nephrotic The question thus arises of whether these patients and suggested that nephrotic syndrome women really were anemic or their low Hb levels somehow interferes with erythropoietin synthe- could have been caused by hemodilution. The sis in the kidneys.2,10,12,17 Unfortunately, serum combination of estradiol and progesterone de- samples obtained from our patients at the time of creases renal sodium excretion by stimulating renal biopsy were not available. Therefore, we the renin-angiotensin-aldosterone system or di- were unable to measure erythropoietin levels or rectly affecting renal tubules.19 Thus, nephrotic clarify whether anemia is associated with inad- women could be more prone to extracellular equately low erythropoietin levels. A report of 32 fluid accumulation. More importantly, estrogen children with nephrotic syndrome described nor- has been shown to increase plasma volume in mal Hb levels in steroid-responsive patients and relation to total extracellular fluid volume.19 This low levels in steroid-resistant patients. Anemia phenomenon is caused by a reduction in transcap- developed on average 4 months after the onset of illary escape of albumin, thus causing increased disease.17 It therefore is possible that anemia intravascular oncotic pressure and fluid reten- develops as the nephrotic state persists over time. tion.19 Plasma volume expansion results in he- Unfortunately, we do not have reliable data for modilution and lower Hb levels despite normal the duration of renal disease before biopsy in our red blood cell mass. Obviously, red blood cell patients. However, it is unlikely that all our mass in such patients must be measured to an- patients underwent biopsy at the immediate on- swer these questions. It is surprising that ne- set of disease. Most likely, many of them were phrotic patients who frequently experience so- proteinuric for weeks, months, or even years dium retention and edema do not more frequently before biopsy. If duration of nephrotic state is an show signs of hemodilution with lower Hb lev- important denominator in anemia development, els. This implies that nephrotic patients are able a significant proportion of our patients can be to keep their plasma volume constant and excess expected to be anemic. sodium and water are present exclusively in the Our iron metabolism results show a strong extravascular space. Geers et al20 found blood negative correlation between serum transferrin and plasma volume to be normal in patients with level and proteinuria and a very tight correlation nephrotic syndrome. However, loss of intravascu- with serum albumin level. Urinary losses of lar fluid into the interstitium because of low transferrin do not cause iron depletion, indicated oncotic pressure, which would cause hypovole- by normal ferritin levels in our patients. Ne- mia and greater Hb levels, is balanced by a phrotic men had greater ferritin levels compared decrease in interstitial oncotic pressure.21 with non-nephrotic patients, a phenomenon also Our results show an association of Hb level described by Branten et al.18 The reason for these with BMI. This association also was found in the elevated ferritin levels in patients with nephrotic general population22 and patients after kidney syndrome is unclear at present. None of the transplantation.23 A plausible explanation for this parameters of iron metabolism, except for trans- effect is lacking. ferrin saturation in men, correlated with Hb lev- Our study has several limitations. Diagnosis els, showing that changes induced by nephrotic and quantification of proteinuria relied on a single syndrome do not have an important impact on urine sample. We do not have data on duration of erythropoiesis. proteinuria and nephrotic syndrome, which may The reason the correlation between protein- be an important determinant of anemia. Other uria and Hb level differs slightly between women causes of anemia, such as deficiency of vitamin PROTEINURIA AND HEMOGLOBIN LEVELS 431

B12 or folic acid, were not investigated, and our 10. Gansevoort RT, Vaziri ND, de Jong PE: Treatment of data for iron metabolism and use of angiotensin- anemia of nephrotic syndrome with recombinant erythropoi- etin. Am J Kidney Dis 28:274-277, 1996 converting enzyme inhibitors and angiotensin 11. Ishimitsu T, Ono H, Sugiyama M, et al: Successful receptor blockers are incomplete. Although we erythropoietin treatment for severe anemia in nephrotic excluded patients with systemic and infectious syndrome without renal dysfunction. Nephron 74:607-610, disease, bleeding, or malignancy, it still is pos- 1996 sible that some patients had other inflammatory 12. Bayes B, Serra A, Junca J, Lauzurica R: Successful treatment of anaemia of nephrotic syndrome with recombi- diseases or occult blood loss. As stated, we do nant human erythropoietin. Nephrol Dial Transplant 13:1894- not have data for serum and urinary erythropoi- 1895, 1998 etin levels. Finally, eGFR calculated by means of 13. Levey ASGT, Kusek JW, Beck GJ, for the MDRD the Modification of Diet in Renal Disease for- Study Group: A simplified equation to predict glomerular filtration rate from serum creatinine. J Am Soc Nephrol mula is not very precise in the high GFR range of 11:0828A, 2000 (abstr) our patients, in which the formula tends to over- 14. Dikow R, Schwenger V, Schomig M, Ritz E: How estimate true GFR.24,25 should we manage anaemia in patients with diabetes? Neph- In conclusion, our study of a large group of rol Dial Transplant 17:S67-S72, 2002 (suppl 1) 15. Thomas MC, MacIsaac RJ, Tsalamandris C, Power patients with primary renal disease and near- D, Jerums G: Unrecognized anemia in patients with diabe- normal renal function shows that Hb levels do tes: A cross-sectional survey. Diabetes Care 26:1164-1169, not correlate with proteinuria in men. However, 2003 women with nephrotic syndrome showed a ten- 16. Misaizu T, Matsuki S, Strickland TW, Takeuchi M, Kobata A, Takasaki S: Role of antennary structure of N- dency toward lower Hb levels. Whether this linked sugar chains in renal handling of recombinant human phenomenon reflects a true reduction in red blood erythropoietin. Blood 86:4097-4104, 1995 cell mass or hormone-induced hemodilution re- 17. Feinstein S, Becker-Cohen R, Algur N, et al: Erythro- mains to be investigated. poietin deficiency causes anemia in nephrotic children with normal kidney function. Am J Kidney Dis 37:736-742, 2001 18. Branten AJ, Swinkels DW, Klasen IS, Wetzels JF: REFERENCES Serum ferritin levels are increased in patients with glomeru- 1. Orth SR, Ritz E: The nephrotic syndrome. N Engl lar diseases and proteinuria. Nephrol Dial Transplant 19: J Med 338:1202-1211, 1998 2754-2760, 2004 2. Vaziri ND, Kaupke CJ, Barton CH, Gonzales E: Plasma 19. Stachenfeld NS, Keefe DL, Taylor HS: Responses to concentration and urinary excretion of erythropoietin in a saline load in GnRH antagonist-pretreated premenopausal adult nephrotic syndrome. Am J Med 92:35-40, 1992 women on progesterone or estradiol-progesterone therapy. 3. Crew RJ, Radhakrishnan J, Appel G: Complications of J Clin Endocrinol Metab 90:386-394, 2005 the nephrotic syndrome and their treatment. Clin Nephrol 20. Geers AB, Koomans HA, Boer P, Dorhout Mees EJ: 62:245-259, 2004 Plasma and blood volumes in patients with the nephrotic 4. Zhou XJ, Vaziri ND: Erythropoietin metabolism and syndrome. Nephron 38:170-173, 1984 21. Koomans HA: Pathophysiology of oedema in idio- pharmacokinetics in experimental nephrosis. Am J Physiol pathic nephrotic syndrome. Nephrol Dial Transplant 18: 263:F812-F815, 1992 Svi30-Svi32, 2003 (suppl 6) 5. Prinsen BH, de Sain-van der Velden MG, Kaysen GA, 22. Cirillo M, Laurenzi M, Trevisan M, Stamler J: Hemat- et al: Transferrin synthesis is increased in nephrotic patients ocrit, blood pressure, and hypertension. The Gubbio Popula- insufficiently to replace urinary losses. J Am Soc Nephrol tion Study. Hypertension 20:319-326, 1992 12:1017-1025, 2001 23. Lorenz M, Kletzmayr J, Perschl A, Furrer A, Horl 6. Vaziri ND: Erythropoietin and transferrin metabolism WH, Sunder-Plassmann G: Anemia and iron deficiencies in nephrotic syndrome. Am J Kidney Dis 38:1-8, 2001 among long-term renal transplant recipients. J Am Soc 7. Bosman DR, Winkler AS, Marsden JT, Macdougall IC, Nephrol 13:794-797, 2002 Watkins PJ: Anemia with erythropoietin deficiency occurs 24. Bostom AG, Kronenberg F, Ritz E: Predictive perfor- early in diabetic nephropathy. Diabetes Care 24:495-499, mance of renal function equations for patients with chronic 2001 kidney disease and normal serum creatinine levels. J Am Soc 8. Thomas S, Rampersad M: Anaemia in diabetes. Acta Nephrol 13:2140-2144, 2002 Diabetol 41:S13-S17, 2004 (suppl 1) 25. Rule AD, Larson TS, Bergstralh EJ, Slezak JM, 9. Shibasaki T, Misawa T, Matsumoto H, et al: Character- Jacobsen SJ, Cosio FG: Using serum creatinine to estimate istics of anemia in patients with nephrotic syndrome. Nip- glomerular filtration rate: Accuracy in good health and in pon Jinzo Gakkai Shi 36:896-901, 1994 chronic kidney disease. Ann Intern Med 141:929-937, 2004 Host and Bacterial Virulence Factors Predisposing to Emphysematous Pyelonephritis

Chin-Chung Tseng, MD, PhD, Jiunn-Jong Wu, PhD, Ming-Cheng Wang, MD, Lien-I Hor, PhD, Yen-Hsiu Ko, BS, and Jeng-Jong Huang, MD

● Background: Emphysematous pyelonephritis (EPN) is a rare, severe, gas-forming infection of the kidney, and its precise pathogenesis remains obscure. Methods: To investigate the roles of host and bacterial virulence factors in with those in patients with (47 ؍ the pathogenesis of EPN, we compared: (1) host factors in patients with EPN (n the prevalence of virulence gene in causative (2) ,(79 ؍ acute renal infections without gas formation (non-EPN; n Escherichia coli strains from 16 of the 47 EPN cases with all 79 non-EPN cases by means of polymerase chain reaction analysis, and (3) gas volumes produced by EPN and non-EPN strains cultured in broths at 3 glucose concentrations (100, 180, and 250 mg/dL [5.6, 10.0, and 13.9 mmol/L]). Results: Diabetes mellitus (DM) with poor glycemic control (ie, hemoglobin A1c level > 11%) and urinary tract obstruction were more prevalent in the EPN group. However, DM with poor glycemic control was the only host factor independently associated with EPN (odds .EPN strains had a greater prevalence of the uropathogenic-specific protein (usp) genes .(0.018 ؍ ratio, 4.9; P Multivariate analyses also showed the association between usp and EPN with borderline significance (odds ratio, There was no significant difference in gas production by E coli isolated from patients with or without .(0.057 ؍ P ;8.4 EPN. Conclusion: DM with poor glycemic control and urinary tract obstruction are host factors predisposing to EPN. There was no difference in gas production between EPN and non-EPN E coli strains. The distribution of E coli virulence genes was remarkably similar between the 2 groups. However, the PapG II adhesin (papG II) gene is significantly decreased and the usp gene is increased with borderline significance in EPN E coli strains. Am J Kidney Dis 46:432-439. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Emphysematous pyelonephritis; diabetes mellitus; urinary tract obstruction; Escherichia coli; virulence factors.

MPHYSEMATOUS PYELONEPHRITIS The most common causative organism for E (EPN) is a severe necrotizing infection of EPN is Escherichia coli, which accounts for the renal parenchyma and its surroundings that approximately 60% to 70% of cases.1-7 Although results in the accumulation of gas in the collect- various theories have been proposed to account ing system, renal parenchyma, or perinephric for the development of EPN,2,3,8-11 no study tissue.1-7 EPN deserves special attention because concurrently investigated host and bacterial fac- of its life-threatening potential (reported mortal- tors that predispose to EPN, and its pathogenesis ity rate varies widely [7% to 75%]) and mortality remains obscure. EPN occurs almost exclusively related to its treatment.1 in patients with diabetes mellitus (DM). How- ever, patients with DM with acute renal infec- tions rarely develop EPN, and predisposing risk From the Department of Internal Medicine, Division of factors for EPN are still unclear. Previous studies Nephrology, National Cheng Kung University Hospital; have shown that mixed acid fermentation of Departments of Medical Laboratory Science and Biotechnol- glucose by Enterobacteriaceae (eg, E coli, Kleb- ogy and Microbiology and Immunology, Institute of Clinical siella pneumoniae, and Proteus species) under Medicine, College of Medicine, National Cheng Kung Uni- the relatively anaerobic conditions is the major versity, Tainan, Taiwan. 2,10 Received January 12, 2005; accepted in revised form May pathway for gas production in EPN. Because 19, 2005. an interrelationship may exist between gas vol- Originally published online as doi:10.1053/j.ajkd.2005.05.019 ume produced by different E coli strains and on July 13, 2005. glucose level of kidney tissue in the development Supported in part by grant no. NSC 90-2314-B-006-097 from the National Science Council, Taiwan. of EPN, a controlled study is needed to confirm Address reprint requests to Jeng-Jong Huang, MD, Divi- the proposed interactions. sion of Nephrology, Department of Internal Medicine, Na- Another interesting issue is virulence factors tional Cheng Kung University Hospital, 138 Shing-Li Rd, of the E coli strains causing EPN compared with Tainan, 70428, Taiwan. E-mail: [email protected] © 2005 by the National Kidney Foundation, Inc. those of strains causing acute renal infections 0272-6386/05/4603-0006$30.00/0 without gas formation. The roles of E coli viru- doi:10.1053/j.ajkd.2005.05.019 lence factors, including P-fimbriae, type 1 fim-

432 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 432-439 VIRULENCE FACTORS IN EMPHYSEMATOUS PYELONEPHRITIS 433 briae, S-fimbriae, F1C-fimbriae, afimbrial ad- positive E coli culture results in blood and/or pus also were hesins, hemolysin, cytotoxic necrotizing factor included in this study. For patients with underlying compli- 1, the iron uptake system, and uropathogenic- cating factors, such as functional or structural abnormality of the urinary tract, or a severe clinical manifestation, such specific protein, in the pathogenesis of acute as unstable hemodynamics (systolic blood pressure Ͻ 90 12-15 renal infections have been studied widely. mm Hg) and/or organ failure, renal ultrasonography and However, no such investigation has been con- abdominal computed tomography (CT) were performed for ducted for EPN. the diagnosis of EPN or other acute renal infections. Diagnos- In the present study, we sought to identify tic criteria for EPN included evidence of gas accumulation in the collecting system, renal parenchyma, perinephric, or which host and bacterial virulence factors predis- pararenal spaces proved by abdominal CT scan, without pose to the development of EPN through a com- fistula between the urinary tract and bowel, and no trauma- parison of gas-forming renal infections and analo- or urinary catheterization–related gas accumulation. Among gous non–gas-forming renal infections. We also the 79 patients with non-EPN, 67 patients had acute pyelone- compared gas volumes produced by EPN and phritis without clinical or imaging evidence of renal abscess, acute bacterial nephritis, or EPN; 3 patients had a renal non-EPN E coli strains cultured in broths (con- abscess (included because these represented non–gas- taining different glucose concentrations, but no forming infections); and 9 patients had acute bacterial nephri- other carbohydrates)2,10,16 that were designed to tis proved by enhanced abdominal CT scan. mimic human renal abscess to determine the role of bacterial gas-producing capacity in the devel- Clinical Features and Laboratory Data at opment of EPN. Presentation Shock is defined as systolic blood pressure less than 90 METHODS mm Hg. Disturbance of consciousness includes confusion, delirium, stupor, and coma. Acute renal function impairment Patient Selection is defined as additional elevation of more than 1 mg/dL From February 1997 to March 2001, a total of 95 consecu- (Ͼ88 ␮mol/L) in serum creatinine levels if the baseline level tive patients with definite acute renal infections caused by E was greater than or equal to 3 mg/dL (Ն265 ␮mol/L) and coli at our institution were enrolled in the present study. more than 0.5 mg/dL (Ͼ44 ␮mol/L) if the baseline was less Availability of a stock of the causative E coli strain was a than 3 mg/dL (Ͻ265 ␮mol/L). White blood cell count, prerequisite for inclusion in the study. Among these cases, C-reactive protein level, and serum albumin level at presen- 16 patients fulfilled diagnostic criteria for EPN (the EPN tation were checked. Thrombocytopenia is defined as a group) and the other 79 patients fulfilled diagnostic criteria platelet count less than 120 ϫ 103/␮L(ϫ 109/L), and for acute renal infections without gas accumulation within bacteremia and bacteriuria are defined as a positive blood the renal parenchyma or perirenal space (the non-EPN and urine culture result for E coli during an episode of acute group). Because EPN is a rare disease, the number of EPN renal infection, respectively. cases collected from January 1997 to March 2001 was relatively low for statistical analyses. Therefore, we in- Definition and Classification of Host Factors cluded an additional 31 consecutive patients with EPN Host factors included age, sex, DM, immunosuppression, caused by E coli who were admitted to our hospital between urinary tract obstruction (UTO), chronic renal insufficiency, August 1989 and January 1997. These 31 patients met all and recent instrumentation of the urinary tract (eg, use of an inclusion criteria, except for the stock of E coli, which was indwelling urinary catheter before the onset of acute renal unavailable at that time. Clinicoradiological manifestations 2 infection). For patients with DM, poor glycemic control in these 31 patients have been analyzed previously. In the before the episode of acute renal infection is defined as present study, predisposing host factors in these 31 patients blood glycosylated hemoglobin A1c (HbA1c) level greater plus the 16 new patients collected from January 1997 to than 11%. Diabetic retinopathy was screened and diagnosed March 2001 were evaluated and compared with those of the by ophthalmologists in cooperative patients who could toler- 79 non-EPN cases. ate the examination. Immunosuppression was diagnosed in patients who presented with any of the following diseases or Diagnostic Criteria conditions: neutropenia caused by hematologic disease, im- Diagnostic criteria for acute renal infections included munosuppressive therapy, active systemic lupus erythemato- fever of 38.3°C or greater and flank pain and/or tenderness at sus, liver cirrhosis, and advanced malignancy. UTO is de- the costovertebral angle with or without disturbed micturi- fined as stasis of urine flow stemming from any underlying tion. A positive urine culture result is defined as 105 or urinary tract abnormality and/or evidence of hydronephrosis greater colony-forming units of E coli isolated from a and/or hydroureter, identified first by means of renal or freshly voided midstream urine sample, and any E coli bladder ultrasonography, then confirmed by means of intra- cultured from aspirated urine or pus samples was considered venous urography, retrograde or antegrade pyelography, a positive isolation. Some patients with negative urine cul- and/or CT. Causes of UTO included ureteral stones or tumor, ture results, but fulfilling clinical and imaging criteria for ureterocele, ureteral stricture, and uterine myoma with ure- EPN (n ϭ 27; 57%) or non-EPN (n ϭ 11; 14%) and with teral stretching and urethral stricture. Chronic renal insuffi- 434 TSENG ET AL ciency is defined as a serum creatinine level greater than 1.4 EPN.2,10,21 To examine bacterial gas-producing capacity, a mg/dL (Ͼ124 ␮mol/L) and persisting for more than 3 culture environment that mimicked human renal abscess months. was designed on the basis of this gas-forming mechanism. Bacteria were cultured in broths at 3 glucose concentrations, Bacterial Isolates and Detection of E coli including 100, 180, and 250 mg/dL (5.6, 10.0, and 13.9 Virulence Factors mmol/L), representing optimal, suboptimal, and poor glyce- mic control in patients with DM, respectively. The glucose E coli strains were identified by using the standard 17 broth consisted of distilled water and nutrient-broth powder method. A single colony of E coli from each positive containing beef extract, 3 g/L; peptone, 5 g/L (Difco; Becton isolate was selected arbitrarily from the initial culture plate Dickinson, Frankin Lakes, NJ), and no other carbohydrate and stored in 20% glycerol at Ϫ70°C for use in all subse- 18 except glucose. pH was adjusted to 6.8, and a layer of quent analyses. Causative E coli strains were available for mineral oil was added to cover the broth and establish 16 of 47 patients with EPN and all 79 non-EPN analogues. relatively anaerobic conditions. For gas collection, a small Genetic determinants of E coli virulence factors were veri- inverted tube (diameter, 5 mm; surface area, 19.6 mm2) was fied by using polymerase chain reaction (PCR), with a total inserted into each tube of glucose broth. E coli (1.8 ϫ 106 of 14 genes detected. Primer pairs specifically discriminat- colony-forming units) was added to each glucose-broth tube, ing for the 3 PapG adhesin (papG) genes (class I to III) of followed by a thick layer of mineral oil, and the tube was P-fimbriae and genes for type l fimbrial adhesins (fimH), sealed to prevent gas losses during incubation at 37°C. The S-/F1C-fimbriae (sfa/foc), afimbrial adhesins (afa), hemoly- height of the collected gas in the top of the small inverted sin (hlyA), cytotoxic necrotizing factor 1 (cnf1), aerobactin tube was recorded at 24 and 48 hours after incubation, and receptor (iutA), catecholate siderophore receptor (iroN), iron gas volume (V) was estimated using the following formula: regulated gene A homologue adhesin (iha), uropathogenic- V(␮L) ϭ 19.62 (mm2) ϫ height (mm). This experiment was specific protein (usp), and outer membrane protease T (ompT) 12,15,18-20 repeated 3 times, and average volumes (in microliters) of gas have been described previously. produced by the EPN and non-EPN E coli strains were To isolate the DNA to be amplified, bacteria were har- compared. vested from Luria-Bertani agar, suspended in 200 ␮Lof sterile water, incubated at 100°C for 10 minutes, and centri- Statistical Analysis fuged at 13,000g. The supernatant was used as the DNA template. Amplification was performed in a 50-␮L reaction For univariate analysis, chi-square or Fisher exact tests mixture containing 5 ␮L of prepared template DNA, 50 were used for categorical variables, whereas Wilcoxon-Mann- pmol of each primer, 0.2 mmol each of the 4 deoxyribo- Whitney test was used for continuous variables. P less than nucleoside triphosphates, and1UofDynaZyme II DNA 0.1 is considered statistically significant for univariate anal- polymerase (Finnzymes, Oy, Finland) in 1 ϫ PCR buffer II ysis. Multivariate analysis was performed on all statistically (Finnzymes). The thermal cycle protocol was as follows: significant univariate variables, with multiple logistic regres- denaturation at 95°C for 7 minutes, followed by 30 cycles of sion used to determine host and E coli virulence factors denaturation at 94°C for 1 minute, annealing for 1 minute (2 independently associated with EPN. P less than 0.05 is minutes for papG I to III multiplex PCR), and extension at considered statistically significant. All statistical analyses 72°C for 1 minute (4 minutes for papG I to III multiplex were performed using JMP software (version 3.2.6; SAS PCR), followed by a final extension (72°C for 10 minutes). Institute Inc, Cary, NC). Annealing temperatures were 52°C for fimH and hlyA; 56°C for cnf1 and aer; 62°C for ompT, iha, usp, iroN, afa, sfa, and RESULTS foc; and 68°C for papG I to III. PCR was performed using a Clinical features and laboratory data at presen- System 9600 thermal cycler (Perkin-Elmer, Foster, CA). tation for the EPN and non-EPN groups are listed PCR results were duplicated, and positive and negative control strains for the traits of interest were included in each in Table 1. Patients with EPN had a significantly assay to confirm that the PCR was functioning correctly and greater blood white blood cell count and serum exclude contamination. The nucleotide sequence for the C-reactive protein level, but lower serum albu- PCR products for each virulence gene was determined by min level in comparison to their non-EPN coun- using an autosequencing system (3100 Genetic Analyzer; terparts. The EPN group had a significantly ABI Prism, Foster, CA) to confirm that amplification prod- ucts obtained from the PCR truly represented the expected greater incidence of thrombocytopenia, shock, sequences. and mortality despite a lower prevalence of bac- teriuria. Comparison of Gas Production Host factors for the 2 groups are listed in Table 2. To investigate bacterial gas-production capacity, 16 E coli The prevalence of DM was significantly greater for strains from patients with EPN were compared with a the EPN group (94% versus 34%; P Ͻ 0.0001). In 28-strain arbitrary sampling of non-EPN E coli strains. patients with DM, poor glycemic control was more There is no international standard method for evaluation of ϭ E coli gas-producing capacity. However, mixed acid fermen- prevalent in the EPN group (56% versus 20%; P tation of glucose by E coli under relatively anaerobic condi- 0.01). Also, mean HbA1c level in the EPN group tions is the major pathway for gas formation in patients with was significantly greater (10.9% versus 9.3%; P ϭ VIRULENCE FACTORS IN EMPHYSEMATOUS PYELONEPHRITIS 435

Table 1. Clinical Features and Laboratory Data for the EPN and Non-EPN Groups at Presentation

Parameter EPN (n ϭ 47) Non-EPN (n ϭ 79) P

High fever (Ͼ38.4°C) 30 (64) 59 (75) NS Mean white blood cells (1,000/␮L) 17.5 14.6 Ͻ0.05 C-reactive protein level (mg/L) 223 (n ϭ 34) 146 (n ϭ 75) 0.007 Serum albumin (g/dL) 2.3 (n ϭ 44) 3.0 (n ϭ 50) Ͻ0.0001 Thrombocytopenia (platelets Ͻ120,000/␮L) 29 (61) 17 (24) Ͻ0.0001 Bacteriuria 20 (43) 68 (86) Ͻ0.0001 Bacteremia 31 (66) 54 (68) NS Acute renal impairment 21 (45) 23 (29) NS Consciousness disturbance 11 (23) 12 (15) NS Shock 22 (47) 23 (29) 0.045 Mortality 6 (13) 1 (1) 0.01

NOTE. Values expressed as number (percent) unless noted otherwise. To convert serum albumin in g/dL to g/L, multiply by 10. Abbreviation: NS, not significant.

0.04; not shown in the table). Of 71 patients with 3). The papG II gene was identified less com- DM, 40 patients could be evaluated for diabetic monly in the EPN strains (P ϭ 0.04). There was retinopathy (11 patients, EPN group; 29 patients, no intergroup difference in prevalence of genetic non-EPN group). No significant difference was determinants for other fimbriae or adhesin, hemo- shown between the 2 groups. The prevalence of lysin, and cytotoxic necrotizing factor 1. With UTO was significantly greater in patients with EPN regard to iron uptake–related genes, the iroN (32% versus 11%; P ϭ 0.005), whereas urinary gene was more prevalent in the EPN group (69% tract instrumentation–related infection was less versus 46%; P ϭ 0.09). The genetic determinant prevalent (0% versus 6%; P ϭ 0.09). When signifi- of uropathogenic-specific protein, usp, also was cant host factors were tested concurrently by using significantly more prevalent in EPN strains (P ϭ multiple logistic regression analysis, DM with poor 0.04). Concurrent testing of significant genetic glycemic control was the only independent factor factors by using multiple logistic regression anal- associated with EPN (odds ratio [OR], 4.9; ysis showed that usp was associated with EPN P ϭ 0.018). with borderline significance. However, papG II The 2 groups were compared in terms of was associated negatively with EPN (OR, 0.2; prevalence of E coli urovirulence genes (Table P ϭ 0.01).

Table 2. Host Factors in the EPN and Non-EPN Groups

Multivariate Univariate Analysis Analysis

Host Factors EPN (n ϭ 47) Non-EPN (n ϭ 79) OR P OR P

Mean age (y) 60 59 NS Female 41 (87) 63 (79) 2.2 NS DM 44 (94) 27 (34) 28.2 Ͻ0.0001 NA NA Poor glycemic control* 19/34 (56) 4/20 (20) 5.1 0.01 4.9 0.02 Retinopathy 5/11 (46) 16/29 (55) 0.7 NS Immunosuppression 7 (15) 15 (19) 0.8 NS UTO 15 (32) 9 (11) 3.7 0.005 3.3 NS Chronic renal insufficiency 10 (21) 18 (23) 0.9 NS Urinary tract instrumentation 0 (0) 5 (6) 0.1 0.09 0.1 NS

NOTE. Values expressed as number (percent) unless noted otherwise. Abbreviations: NS, not significant; NA, not available.

*Poor glycemic control defined as an HbA1c level greater than 11%. 436 TSENG ET AL

Table 3. E coli Virulence Genes in Strains Isolated From the EPN and Non-EPN Groups

Univariate Analysis Multivariate Analysis

Virulence Genes EPN (n ϭ 16) Non-EPN (n ϭ 79) OR P OR P

papG I 0 (0) 0 (0) NS papG II 9 (56) 66 (84) 0.3 0.04 0.2 0.01 papG III 0 (0) 5 (6) 0.4 NS fimH 13 (81) 74 (94) 0.3 NS sfa 1 (6) 1 (1) 5.2 NS foc 5 (31) 10 (13) 3.1 NS afa 0 (0) 1 (1) 1.6 NS iha 4 (25) 36 (46) 0.4 NS hlyA 7 (44) 34 (43) 1.0 NS cnf1 6 (38) 31 (40) 0.9 NS iutA 13 (81) 62 (79) 1.2 NS iroN 11 (69) 36 (46) 2.6 0.09 1.8 NS usp 15 (94) 53 (67) 7.4 0.04 8.4 0.057 ompT 14 (88) 71 (90) 0.8 NS

NOTE. Values expressed as number (percent) unless noted otherwise. Abbreviation: NS, not significant.

Mean gas volumes produced by the E coli DISCUSSION strains are listed in Table 4. No significant differ- It has been postulated that the development of ences were shown comparing the 2 groups in EPN requires the presence of gas-forming bacte- terms of mean E coli gas production at 24 or 48 ria, high tissue glucose level, impaired tissue hours after incubation in broths at different glu- perfusion and gas transport, and defective immu- cose concentrations. However, mean 24- and nity to speed bacterial growth.2,3,8-11 However, 48-hour gas volumes were significantly in- the precise pathogenesis remains obscure. Our creased for both groups, comparing culture broths study shows a significant association between at the lower glucose concentration with those at DM and EPN, and DM apparently is a risk factor the higher 250-mg/dL (13.9-mmol/L) concentra- for the development of EPN. We also found that tion (Fig 1). Ͼ DM with poor glycemic control (ie, HbA1c 11%) was the only independent host factor asso- Table 4. Mean Gas Volumes for E coli Cultured in ciated with EPN. These findings suggest that Different Glucose Broths for 24 and 48 Hours poor glycemic control (ie, high renal glucose concentration) in patients with DM, rather than Gas Volume (␮L) the DM itself, is a more important factor predis- EPN Strains Non-EPN Strains* posing patients with diabetes to EPN. Glucose (mg/dL) (n ϭ 16) (n ϭ 28) P A second important factor is impaired gas 100 transport caused by poor local circulation, which 2,10 24 h 132 Ϯ 14 116 Ϯ 29 NS may be associated with more severe UTO. 48 h 145 Ϯ 20 141 Ϯ 35 NS The unrelieved hydronephrosis may increase pel- 180 vicaliceal pressure and compromise renal circula- Ϯ Ϯ 24 h 130 16 126 33 NS tion, resulting in impaired gas transport and 48 h 155 Ϯ 16 157 Ϯ 39 NS 250 subsequent creation of gas pockets in patients 2,10 24 h 155 Ϯ 18 139 Ϯ 39 NS with EPN. Our study supports that impaired 48 h 192 Ϯ 18 171 Ϯ 39 NS tissue circulation caused by hydronephrosis or other vascular disease is a risk factor for EPN. NOTE. To convert glucose in mg/dL to mmol/L, multiply by 0.05551. In addition, we found that diabetic retinopa- Abbreviation: NS, not significant. thy, a microvascular complication of DM, was *Randomly sampled from 79 non-EPN strains. not associated with EPN. This might imply that VIRULENCE FACTORS IN EMPHYSEMATOUS PYELONEPHRITIS 437

Fig 1. Mean gas volume (␮L) produced by E coli incubated in different concentrations of glucose broth for 24 and 48 hours. EPN versus non-EPN group, all P > 0.05. *Significant differences (P < 0.05) comparing 100- or 180-mg/dL with the 250-mg/dL glucose broths. #Significant differences (P < 0.05) comparing the 100-mg/dL with the 250-mg/dL glucose broths. To convert glucose in mg/dL to mmol/L, multiply by 0.05551. minor ischemia in the kidney is not sufficient to Of the known and putative E coli virulence cause EPN. Because the number of patients ex- genes examined in this study, we found that only amined for diabetic retinopathy was low in the the genetic determinant of uropathogenic-spe- present study, we could not conclude that dia- cific protein (usp, a putative bacteriocin gene) betic retinopathy is not a predictor for develop- was significantly increased in EPN strains. The ment of EPN. Our study also shows that instru- reported usp gene prevalence in pyelonephritis mentation of the urinary tract is associated isolates was 93%, and in stool sample E coli negatively with EPN. Although this instrumenta- isolates from healthy individuals, 24%.25 The tion may increase the risk for urinary tract infec- contribution of the usp gene to the virulence of E tion,22 it may relieve the UTO and prevent the coli has been assessed by using an experimental development of EPN. mouse model, with enhanced infectivity shown Hazouard et al23 (1999) reported an E coli with carriage of the gene.26 Uropathogenic- strain that can produce a high gas level from specific protein is homologous to bacteriocins of necrotic tissue and glucose in vitro. These inves- Pseudomonas aeruginosa and Photorhabdus lu- tigators suggested that E coli strains that cause minescens.27,28 Bacteriocins are proteins pro- EPN produce greater gas volumes. However, to duced by host bacteria to destroy competing date, this hypothesis remains unconfirmed. A bacteria in their ecological niche, thereby enhanc- previous analysis of gas produced by 6 EPN E ing their own infectivity.27 On the basis of our coli strains found that 5 strains produced hydro- findings, we hypothesize that bacteriocin from gen and all produced carbon dioxide.2 This is in EPN strains might help eliminate competitive line with the findings of Müller24 and supported bacteria colonizing stone materials or necrotic the hypothesis that mixed acid fermentation of tissue. Because there is no evidence that the usp glucose by Enterobacteriaceae (eg, E coli, K gene is expressed by E coli strains causing EPN, pneumoniae, and Proteus species) under anaero- additional study is needed to examine the role of bic conditions is the major pathway for gas uropathogenic-specific protein in the pathogene- production in patients with EPN.2,10,21 There is sis of EPN. The iroN gene is associated signifi- no standardized method to study the gas-produc- cantly with the usp gene in this study (P ϭ 0.001; ing capacity of E coli. Our method, providing data not shown). The mildly increased preva- conditions conducive to glucose fermentation lence of the iroN gene might be caused by its through mixed acid fermentation, found that in- copresence with the usp gene in strains isolated creasing the glucose concentration from 100 to from patients with EPN. A greater proportion of 250 mg/dL (5.6 to 13.9 mmol/L) can result in a EPN (7 of 16 strains; 44%) than non-EPN strains significant increase in mean gas volume for both (10 of 79 strains; 13%) was obtained from blood. groups. These results imply that tissue glucose However, virulence characteristics of E coli be- level is more important than strain-specific varia- tween the strains isolated from urine and blood tion in gas production for the development of were not different, except the prevalence of the EPN. papG II gene in blood isolates was increased 438 TSENG ET AL

(P ϭ 0.01; data not shown). However, the papG imaging findings and clinical outcome. Radiology II gene was less predominant in EPN strains. 198:433-438, 1996 This finding is consistent with previously re- 7. Pappas S, Peppas ThA, Sotiropoulos A, Katsadoros D: Emphysematous pyelonephritis: A case report and review of ported findings of decreased predominance of the literature. Diabet Med 10:574-576, 1993 the papG class II gene in E coli strains isolated 8. Ireland GW, Javadpour N, Cass AS: Renal emphysema from patients with urinary tract infection with and retention of renal function. J Urol 106:463-466, 1971 underlying host factors, such as DM and 9. Clifford NJ, Katz I: Subcutaneous emphysema compli- UTO.13,29 In such conditions, the pathogenic cating renal infection by gas-forming coliform bacteria. role of bacterial adherence to uroepithelial cells Nord Hyg Tidskr 266:437-439, 1962 10. Huang JJ, Chen KW, Ruaan MK: Mixed acid fermen- by P-fimbrial adhesin might be less important tation of glucose as a mechanism of emphysematous urinary than that of the host factors. Notably, the pres- tract infection. J Urol 146:148-151, 1991 ence of specific genes does not necessarily indi- 11. Schainuck LI, Fouty R, Cutler RE: Emphysematous cate that microorganisms are expressing them. In pyelonephritis. A new case and review of previous observa- addition, one limitation of this study is that strain tions. Am J Med 44:134-139, 1968 12. Tseng CC, Wu JJ, Liu HL, Sung JM, Huang JJ: The number was too low to draw conclusions about roles of host and bacterial virulence factors in the develop- the role of bacterial virulence factors in patients ment of upper urinary tract infection caused by Escherichia with EPN. coli. Am J Kidney Dis 39:744-752, 2002 In conclusion, the present study shows that 13. Hagberg L, Jodal U, Korhonen TK, Lidin-Janson G, DM with poor glycemic control and UTO with Lindberg U, Svanborg-Eden C: Adhesion, hemagglutina- impaired renal circulation are the important tion, and virulence of Escherichia coli causing urinary tract infections. Infect Immun 31:564-570, 1981 host factors predisposing patients to the devel- 14. Otto G, Sandberg T, Marklund BI, Ulleryd P, Svan- opment of EPN. There was no significant dif- borg C: Virulence factors and pap genotype in Escherichia ference in gas production between EPN and coli isolates from women with acute pyelonephritis, with or non-EPN E coli isolates. E coli virulence genes without bacteremia. Clin Infect Dis 17:448-456, 1996 in the 2 groups of strains examined in this 15. Kanamaru S, Kurazono H, Ishitoya S, et al: Distribu- study were remarkably similar, except for a tion and genetic association of putative uropathogenic viru- lence factors iroN, iha, kpsMT, ompT and usp in Escherichia decreased prevalence of papG II and increased coli isolated from urinary tract infection in Japan. J Urol prevalence of the usp gene, with borderline 170:2490-2493, 2003 significance, in the EPN group. Expression of 16. Yang WH, Shen NC: Gas-forming infection of the virulence genes during the infectious process urinary tract: An investigation of fermentation as a mecha- will be needed to confirm their roles in the nism. J Urol 143:960-964, 1990 17. Ferraro MJ, Gilligan PH, Saubolle MA, Weissfeld pathogenesis of EPN. AS: Bacteriology, in Murray PR, Baron EJ, Pfaller MA, Tenover FC, Yolken RH (eds): Manual of Clinical Microbi- ACKNOWLEDGMENT ology (ed 6). Washington, DC, American Society for Micro- The authors thank Shu-Chen Yang for skillful secretarial biology, 1995, pp 246-262 work. 18. Johnson JR, Stapleton AE, Russo TA, Scheutz F, Brown JJ, Maslow JN: Characteristics and prevalence within REFERENCES serogroup O4 of a J96-like clonal group of uropathogenic 1. Michaeli J, Mogle P, Perlberg S, Hemiman S, Caine M: Escherichia coli O4:H5 containing the class I and class III Emphysematous pyelonephritis. J Urol 131:203-208, 1987 alleles of papG. Infect Immun 65:2153-2159, 1997 2. Huang JJ, Tseng CC: Emphysematous pyelonephritis: 19. Mitsumori K, Terai A, Yamamoto S, Yoshida O: Clinico-radiological classification, management, prognosis, Identification of S, F1C and three PapG fimbial adhesions in and pathogenesis. Arch Intern Med 160:791-805, 2000 uropathogenic Escherichia coli by polymerase chain reac- 3. Goyzueta JD, Katz R, Dumitrescu O, Choi HS, Kahn tion. FEMS Immunol Med Microbiol 21:261-268, 1998 T: The disappearing kidney. A case of emphysematous 20. Blanco M, Blanco FE, Alonso MP, et al: Detection of pyelonephritis. Arch Intern Med 154:2613-2615, 1994 pap, sfa and afa adhesion-encoding operons in uropatho- 4. Pontin AR, Barnes RD, Joffe J, Kahn D: Emphysema- genic Escherichia coli strains: Relationship with expression tous pyelonephritis in diabetic patients. Br J Urol 75:71-74, of adhesions and production of toxins. Res Microbiol 148: 1995 745-755, 1997 5. Shokeir AA, El-Azab M, Mohsen T, El-Diasty T: 21. Davis BD: Bacterial physiology and bacterial genet- Emphysematous pyelonephritis: A 15-year experience with ics: Nutrition, energy, membrane transport, chemotaxis, in 20 cases. Urology 49:343-346, 1997 Davis BD, Dulbecco R, Eisen HN, Ginsberg HS (eds): 6. Wan YL, Lee TY, Bullard MJ, Tsai CC: Acute gas- Microbiology (ed 4). Philadelphia, PA, Harper & Row, producing bacterial renal infection: Correlation between 1990, pp 65-90 VIRULENCE FACTORS IN EMPHYSEMATOUS PYELONEPHRITIS 439

22. Warren JW, Muncie HL Jr, Hebel JR, Hall-Craggs M: genic Escherichia coli is associated with urinary tract Long-term urethral catheterization increases risk of chronic infection in an experimental mouse model. J Urol pyelonephritis and renal inflammation. J Am Geriatr Soc 165:1347-1351, 2001 42:1286-1290, 1994 27. Parret AH, De Mot R: Escherichia coli’s uropatho- 23. Hazouard E, Barat D, Lanotte P, et al: Emphysema- genic-specific protein: A bacteriocin promoting infectivity? tous pyelonephritis related to specific gas-forming Esche- Microbiology 148:1604-1605, 2002 richia coli without diabetes mellitus. Intensive Care Med 28. Sharma S, Waterfield N, Bowen D, et al: The lumi- 25:628-630, 1999 cins: Novel bacteriocins from Photorhabdus luminescens 24. Müller F: Berlin Klin Wchnschr 26:889, 1889. with similarity to the uropathogenic-specific protein (USP) 25. Kurazono H, Yamamoto S, Nakano M, et al: Charac- from uropathogenic Escherichia coli. FEMS Microbiol Lett terization of a putative virulence island in the chromosome 214:241-249, 2002 of uropathogenic Escherichia coli possessing a gene encod- 29. Tseng CC, Huang JJ, Ko WC, Yan JJ, Wu JJ: ing a uropathogenic-specific protein. Microb Pathog 28:183- Decreased predominance of papG class II allele in Esche- 189, 2000 richia coli strains isolated from adults with acute pyelone- 26. Yamamoto S, Nakano M, Terai A, et al: The presence phritis and urinary tract abnormalities. J Urol 166:1643- of the virulence island containing the usp gene in uropatho- 1646, 2001 Absorptive Hyperoxaluria Leads to an Increased Risk for Urolithiasis or Nephrocalcinosis in Cystic Fibrosis

Bernd Hoppe, MD, Gerd E. von Unruh, PhD, Gesa Blank, MD, Ernst Rietschel, MD, Harmeet Sidhu, PhD, Norbert Laube, PhD, and Albrecht Hesse, PhD

● Background: Hyperoxaluria has been incriminated to account for the increased incidence of urolithiasis or nephrocalcinosis in patients with cystic fibrosis (CF). Hyperoxaluria presumably is caused by fat malabsorption and the absence of such intestinal oxalate-degrading bacteria as Oxalobacter formigenes. To better elucidate its pathophysiological characteristics, we prospectively studied patients with CF by determining these parameters and performing renal ultrasonography twice yearly. Methods: In addition to routine tests in urine (lithogenic and stone-inhibitory substances), the presence of O formigenes was tested in stool, plasma oxalate was measured, and 13 a[ C2]oxalate absorption test was performed in 37 patients with CF aged 5 to 37 years (15 females, 22 males) who were constantly hyperoxaluric before the study. Results: Hyperoxaluria (oxalate, 46 to 141 mg/1.73 m2/24 h [0.51 to 1.57 mmol/1.73 m2/24 h]; normal, <45 mg/1.73 m2/24h[<0.5 mmol/1.73 m2/24 h]) was now found in 24 patients .(Plasma oxalate levels were elevated in 6 patients (7.92 to 19.5 ␮mol/L; normal, 6.3 ؎ 1.1 ␮mol/L .(64.8%) Oxalobacter species were detected in only 1 patient. Intestinal oxalate absorption was elevated (11.4% to 28.5%; normal, <10%) in 23 patients. Hypocitraturia was present in 17 patients (citrate, 0.35 to 2.8 g/1.73 m2/24 h [0.2 to 1.1 mmol/1.73 m2/24 h]; normal female, >2.8 mg/1.73 m2/24h[>1.6 mmol/1.73 m2/24 h]; male, >3.3 mg/1.73 m2/24 h [>1.9 mmol/1.73 m2/24 h]). Urine calcium oxalate saturation was elevated in 17 patients (5.62 to 28.9 relative units; -or nephrocalci (2 ؍ normal female, <5.5 relative units; male, <6.3 relative units). In 16% of patients, urolithiasis (n was diagnosed ultrasonographically. Conclusion: Absorptive hyperoxaluria and hypocitraturia are the (4 ؍ nosis (n main culprits for the increased incidence of urolithiasis and nephrocalcinosis in patients with CF. We advocate high fluid intake, low-oxalate/high-calcium diet, and alkali citrate medication, if necessary. Additional studies are necessary to determine the influence of Oxalobacter species or other oxalate-degrading bacteria on oxalate handling in patients with CF. Am J Kidney Dis 46:440-445. © 2005 by the National Kidney Foundation, Inc.

13 INDEX WORDS: [ C2]oxalate absorption test; Oxalobacter formigenes; absorptive hyperoxaluria; urolithiasis; nephrocalcinosis; cystic fibrosis.

ATIENTSWITH CYSTIC fibrosis (CF) have urinary calcium oxalate saturation.2 Preliminary P an increased risk of developing urolithiasis data suggested that intestinal absorption of ox- and nephrocalcinosis.1-9 Since 1993, we have alate is increased as a result of the malabsorption measured urinary excretion of lithogenic and typically seen in patients with CF and the ab- stone-inhibitory substances and performed regu- sence of such intestinal oxalate-degrading bacte- lar renal ultrasonography in our patients with CF. ria as Oxalobacter formigenes secondary to con- We have shown that the increased risk primarily tinuous antibiotic treatment.1,2,10 To test this is caused by hyperoxaluria, with elevation of hypothesis, we started a prospective study of enteric/absorptive hyperoxaluria and performed an oxalate absorption test in 37 constantly hyper- From the University Children’s Hospital Cologne; Depart- oxaluric patients with CF from the CF center in ments of Internal Medicine I and Urology, University of Cologne, Germany. Bonn; Germany; and Ixion Biotechnology Inc, Alachua, FL. Received February 2, 2005; accepted in revised form METHODS June 6, 2005. Originally published online as doi:10.1053/j.ajkd.2005.06.003 Eighty-two patients with CF (33 females, 49 males) with on August 1, 2005. different grades of clinical severity currently are under our Supported in part by grants Un 91/3 and He 1132/11-4 care. Of 63 initial patients with CF followed up since 1993,2 from the Deutsche Forschungsgemeinschaft. H.S. is affili- 6 patients died of worsening pulmonary function. Of the ated with Ixion Biotechnology Inc, Alachua, FL, which surviving 57 patients, 2 patients underwent successful lung performs specific studies in Oxalobacter research. transplantation and 1 patient developed a testicular tumor. Address reprint requests to Bernd Hoppe, MD, University An additional 25 patients were included in our follow-up Children’s Hospital, Division of Pediatric Nephrology, Ker- procedure since 1998. In all our patients, urine analysis for penerstrasse 62, D-50924 Cologne, Germany. E-mail: lithogenic and stone-inhibitory parameters, stool analysis for [email protected] the presence of Oxalobacter species, plasma oxalate determi- © 2005 by the National Kidney Foundation, Inc. nation, and renal ultrasound examination were performed 0272-6386/05/4603-0007$30.00/0 twice yearly. Between January 2001 and June 2004, we doi:10.1053/j.ajkd.2005.06.003 found 37 patients with CF (15 females, 22 males; age, 5 to

440 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 440-445 ABSORPTIVE HYPEROXALURIA IN CYSTIC FIBROSIS 441

37 years; mean age, 21.1 Ϯ 9.83 years) to be continuously urine sample was collected into 2-L bottles containing 30 hyperoxaluric, defined as urinary oxalate excretion greater mL of 6 mol/L of HCl.16 Oxalate absorption was measured than 0.5 mmol/1.73 m2/24h(Ͼ45 mg/1.73 m2/24 h) for the by means of a gas chromatography-mass spectrometry last 12 months before study entry. Those patients were method, as previously published.16,17 Patients had to adhere studied further, with 21 of 37 patients followed since 1993 to a fluid schedule (1,000 to 1,500 mL/d) and dietary and hence being included in work published previously.2,10 recommendations (eg, 800 mg/d of calcium intake) pre- The 37 patients included in this study were administered scribed by a certified nutritionist before the start of the pancreatic enzymes, vitamins, and mucolytic agents accord- test.16,17 ing to the recommended dosages. Stool fat excretion was For statistical analysis, Mann-Whitney and Spearman determined qualitatively, and pancreatic enzyme administra- tests were performed using SPSS software (SPSS Inc, Chi- tion was adjusted per gram of oral fat intake (6,890 Ϯ 4,155 cago, IL), and P less than 0.05 is judged significant. The U/g of fat intake). Adequate enzyme substitution was consid- study was approved by the ethical committees of the facul- ered in all patients when low or absent stool fat excretion, no ties of medicine of the Universities of Bonn and Cologne, abdominal pain, and low stool frequency were observed. No especially with regard to the use of labeled oxalate. In- vitamin C was administered, but vitamins A (2 to 4,000 formed consent was obtained from patients or their parents units), E (330 Ϯ 250 units), and D (400 units) were adminis- for plasma and stool analysis and the oxalate absorption test. tered daily. Oral or intravenous antibiotic treatment was administered only at acute exacerbation of pulmonary symp- RESULTS toms and included 1 or more of the following antibiotics: tobramycin, azithromycin, trimethoprim-sulfamethoxazole, Hyperoxaluria was found in 24 patients (10 ceftazidime, meropenem, imipenem, and ciprofloxacin, with females, 14 males; oxalate range, 46 to 141 only 5 patients not routinely administered prophylactic anti- mg/1.73 m2/24 h [0.51 to 1.57 mmol/1.73 m2/24 biotics. Patients ingested a high-energy/high-protein diet, Ͻ 2 Ͻ but oxalate-rich food (eg, spinach, rhubarb, and chocolate) h]; normal, 45 mg/1.73 m /24h[ 0.5 mmol/ 2 18 was avoided based on previous studies.2,10 1.73 m /24 h] ). There was no specific differ- In addition to current urine analysis for lithogenic and ence in oxalate excretion between sex or age stone-inhibitory parameters, stool analysis for the presence groups. Intestinal O formigenes was detected in of Oxalobacter species, plasma oxalate determination, and only 1 patient. renal ultrasound examination, an oxalate absorption test was performed in all 37 patients. All data points reported here Plasma oxalate levels were elevated in 6 pa- were derived from these measurements only. Urine collected tients with hyperoxaluria (7.92 to 19.5 ␮mol/L; during 24 hours was analyzed for urinary lithogenic (cal- normal, Ͻ6.3 Ϯ 1.1 ␮mol/L12). All patients had cium, oxalate, phosphate, and uric acid) and stone-inhibitory normal kidney function, with an estimated glo- substances (citrate and magnesium), and urinary saturation merular filtration rate in the high-normal range of calcium oxalate and brushite was calculated as previously Ϯ ϫ 2 Ϯ described.2,10,11 Separation and quantification of urinary (156.7 54.5 mL/min 1.73 m [2.61 0.91 oxalate were performed using a Dionex Series Dx500 gradi- mL/s]). ent ion chromatography system (Dionex, Sunnyvale, CA), The oxalate absorption test showed marked adopting a method for plasma oxalate determination.12 Uri- hyperabsorption in 23 of 37 patients. Overall nary concentrations of calcium and magnesium were deter- mean oxalate absorption was 13.4% Ϯ 6.85%; mined by means of atomic absorption spectroscopy; phospho- rus and sulfate, by means of ion chromatography. Citric and values for hyperabsorbers ranged from 11.4% to 16,17 uric acid were analyzed enzymatically by using the citrate 28.5% (normal, 5% to 10% ). Although uri- lyase and uricase methods, respectively. Renal ultrasonogra- nary oxalate excretion did not correlate with phy was performed by the same radiologist (G. Benz-Bohm) intestinal oxalate absorption (r ϭ 0.28; P ϭ not using a real-time sector scanner with a 5- or 7.5-MHz transducer (128XP/10; Accuson Computed Sonography, significant [NS]), 19 of 24 patients with hyperox- Mountain View, CA). Nephrocalcinosis was categorized aluria also were hyperabsorbers (Fig 1). according to the classification of Dick et al.13 Stool speci- Urinary calcium excretion was elevated in 8 mens were analyzed for intestinal colonization of O formi- patients (5 females, 3 males), with a range of 4.3 genes by direct culture or polymerase chain reaction to 14 to 7.8 mg/kg body weight/24 h (0.11 to 0.2 detect bacterial messenger RNA. Plasma oxalate was mea- Ͻ sured by means of ion chromatography, as described.12 mmol/kg body weight/24 h (normal, 4 mg/kg Glomerular filtration rate was estimated by using the body weight/24 h [0.1 mmol/kg body weight/24 Schwartz formula.15 h]). Patients with hypercalciuria were distributed 13 The oxalate absorption test using [ C2]oxalate was per- equally throughout all age groups. formed as follows: after 24-hour urine collection (baseline), Hypocitraturia was found in 17 patients (4 a capsule containing 50 mg (patients Ͼ 30 kg of body Ͻ females, 13 males; citrate range, 0.35 to 2.8 weight) or 25 mg (patients 30 kg) of sodium oxalate 2 2 13 g/1.73 m /24 h [0.2 to 1.1 mmol/1.73 m /24 h]; (33.82 or 16.925 mg of [ C2]oxalate, respectively) was administered 1 hour before breakfast, and a second 24-hour normal for females, Ͼ2.8 g/1.73 m2/24h[Ͼ1.6 442 HOPPE ET AL

؍ P ;0.28 ؍ Fig 1. No correlation was seen between urinary oxalate excretion and intestinal oxalate absorption (r NS). Dashed-line, normal range of intestinal oxalate absorption (<10% [absorption]); solid-line, normal urinary oxalate excretion (<45 mg/1.73 m2/24 h); 19 patients were both hyperoxaluric and hyperabsorbers. mmol/1.73 m2/24 h]; males, Ͼ3.3 g/1.73 m2/24 hypocitraturia was seen in 3 other patients. Hy- h[Ͼ1.9 mmol/1.73 m2/24 h]) compared with percalciuria as an additional risk factor was ob- normal values obtained in previous studies.19,20 served in 2 patients with hyperoxaluria. Urinary Hypocitraturia was associated with hyperoxalu- calcium oxalate saturation was elevated in all ria in 7 patients and elevated calcium oxalate except 1 patient, and brushite supersaturation saturation in 6 patients. was seen in 3 patients with nephrocalcinosis. Urinary calcium oxalate saturation was el- However, there was no significant difference in evated in 17 patients (8 females, 9 males; range, mean urinary excretion parameters or saturation 5.62 to 28.9 relative units, normal: females, Ͻ5.5 levels between patients with urolithiasis or relative units; males, Ͻ6.3 relative units19,21), all nephrocalcinosis and those with a normal renal of whom had hyperoxaluria. Patients with cal- ultrasound (Table 1). No patient with urolithiasis cium oxalate supersaturation were distributed or nephrocalcinosis had Oxalobacter species. equally through the age and sex groups. Urinary brushite saturation was elevated in 10 patients (4 DISCUSSION females, 6 males; range, 1.49 to 5.16 relative Patients with CF have an increased risk for units; normal, Ͻ1 relative unit). urolithiasis or nephrocalcinosis.1-9 We and oth- At the time of the study, urolithiasis was ers suggested that this is caused primarily by present in 2 patients, and 2 other patients had a secondary hyperoxaluria resulting from fat mal- history of kidney stones. Stone analysis per- absorption and the lack of such intestinal oxalate- formed in 1 patient showed mixed calcium ox- degrading bacteria as O formigenes.2,3,10 We alate monohydrate/-dihydrate. Nephrocalcinosis now show that the secondary hyperoxaluria is was detected by means of ultrasonography in 4 caused by increased intestinal absorption of ox- patients, of whom 2 had marked medullary alate in the majority of patients with hyperoxalu- nephrocalcinosis. Hyperoxaluria caused by el- ria (79%). In comparison to previous data, we evated intestinal oxalate absorption was found in now observe that patients with CF lose intestinal 5 patients with stones or nephrocalcinosis, and colonization with oxalate-degrading bacteria over ABSORPTIVE HYPEROXALURIA IN CYSTIC FIBROSIS 443

time, most likely because of antibiotic therapy, 8.92 6.37 which increases their risk for secondary hyperox- ts Ϯ Ϯ aluria.10 Oxalate What are the mechanisms for the increased Absorption (%) intestinal absorption of oxalate? First, the steator- rhea present in patients with CF leads to intesti- nal binding of calcium to fatty acids instead of 3.42 15.46 3.97 12.9 1.1 5-10 Ϯ Ϯ Ϯ binding to oxalate; hence, more soluble oxalate mol/L)

␮

Plasma 22 Oxalate ( is available for absorption in the colon. Al- though stool fat excretion was low or even absent in our patients, thus correlating with their clinical well-being, a malabsorption state is not necessar- 1.7 4.12 1.22 5.11

1 6.3 ily excluded because other parameters also are Ϯ Ϯ Ͻ

Brushite important. For example, malabsorbed bile acids Saturation

(relative units) also have a major role, eg, for intestinal oxalate absorption. All patients with CF still have a persistent and, in the long run, deteriorating state 2.14 1.54 5.65 0.96 of malabsorption, also supported by the fact that Ϯ Ϯ they need ongoing supplementation of vitamins 6.3, male Saturation 5.5, female; and trace elements. Second, the lack of intestinal Ͻ (relative units) Calcium Oxide /24 h, multiply by 0.571. Ͻ 2 oxalate-degrading bacteria, presumably because of long-term administration of antibiotics in the majority of patients, leads to an increase in 0.68 6.88 0.83 6.88 oxalate absorption. Normally, up to 70% of daily Ϯ Ϯ

Urine oxalate intake is degraded in the intestine by Volume (L) such oxalate-degrading bacteria as O formi-

/24 h to mmol/1.73 m genes, Enterobacter faecalis, or lactic acid bacte- 2 ria.10,23,24 A lack of such bacteria therefore leaves /d) 2 2.47 1.81 2.91 1.48 considerably more oxalate for intestinal absorp- Ϯ Ϯ tion, which, in turn, results in hyperoxaluria. 3.3, male 2.8, female;

Urine Citrate The lack of correlation between intestinal ox- Ͼ (mg/1.73 m Ͼ alate absorption and urinary oxalate excretion (r ϭ 0.28; P ϭ NS; Fig 1) was unexpected. However,

/d) intestinal binding of oxalate with calcium is not 2 37.8 3.13 37.8 3.94 constant, thus leading to differences in the absorp- 45 Ϯ Ϯ

Ͻ tion pattern, although we kept oral intake of Ͻ Urine Oxalate

(mg/1.73 m calcium (800 mg) and oxalate ( 100 mg) stable on both days of the oxalate absorption test.

SD (range). To convert urine calcium in mg/kg body weight/24 h to mmol/kg body weight/24 h, multiply by 0.025; urine oxalate in Furthermore, daily oxalate intake was not stan-

Ϯ dardized; only food with high oxalate content 1.21 59.4 1.96 65.7 4 Ϯ Ϯ NSϽ NS NS NS NS NS NS NS

/d, multiply by 0.0111; urine citrate in mg/1.73 m was avoided to keep oxalate intake less than 100 2 25 (mg/kg/d)

2.4 mg/d. In addition, intestinal transit time has an (0.46-7.83) (22.1-125.1) (0.35-12.57) (0.35-3.86) (1.62-16.64) (0.13-4.7) (0.6-19.5) (3.4-24.5) (0.97-4.32) (32.4-140.4) (0.39-7.21) (1.0-3.08) (3.1-9.6) (0.15-5.16) (0.87-10.7) (4.6-28.5) 2.62 Urine Calcium additional effect on intestinal oxalate absorption and varies from day to day.26 Hence, a certain 13 percentage of [ C2]oxalate will only be excreted 6) 31) 24 to 48 hours after ingestion, especially in ϭ ϭ patients with CF. Therefore, prolongation of the /24 h to mmol/1.73 m 2 collection period or repeated determination of intestinal oxalate absorption probably would show Table 1. Mean Urinary Excretion and Saturation in Patients With CF With Abnormal (urolithiasis, nephrocalcinosis) or Normal Renal Ultrasound Resul better correlation between oxalate absorption and ultrasound (n ultrasound (n NOTE. Values expressed as mean P Normal values Patients with abnormal Patients with normal mg/1.73 m excretion. Still, 56% of all patients were both 444 HOPPE ET AL hyperoxaluric and hyperabsorbers (Fig 1). In led to normalization of urinary citrate excretion addition, because patients with CF are a very in all our patients so treated. heterogeneous group of patients with specific Additional studies are necessary to determine mutations leading to differences in disease expres- the influence of oxalate-degrading bacteria on sion, differences in malabsorption profiles and urinary oxalate excretion in patients with CF or hence intestinal oxalate absorption also might be other malabsorption syndromes. Oral application expected. of such bacteria as a novel therapeutic approach Although Oxalobacter species were absent in already was studied in patients with primary nearly all patients, not all patients had hyperoxalu- hyperoxaluria, leading to a significant reduction ria. We do not think pancreatic enzyme replace- in urinary oxalate excretion with an O formi- ment is responsible for differences in urinary ox- genes preparation (authors’ experiences). How- alate excretion because stool fat examination gave ever, routine antibiotic treatment in patients with evidence of an adequate dosage in all patients. In CF may be an obstacle. Whether successful treat- addition, intestinal bacteria other than Ox- ment is possible and has the desired effect needs alobacter species may be capable of degrading to be tested prospectively. oxalate.23,24 However, we did not specifically However, because there was no significant analyze stool samples, eg, for lactic acid bacteria difference in urinary excretion parameters or or E faecalis.23,24 Nevertheless, we think that saturation levels between patients with urolithia- such oxalate-degrading bacteria as Oxalobacter sis or nephrocalcinosis and those with normal species are major players with regard to intesti- renal ultrasonography results (Table 1), other, nal oxalate “metabolism” in patients with second- probably environmental, factors have an impor- ary hyperoxaluria. Recolonization with O formi- tant role in the pathogenesis of urolithiasis or genes in humans leads to a significant decrease in nephrocalcinosis in patients with CF. Although urinary oxalate excretion without influencing the urolithiasis or nephrocalcinosis is not the pri- absorption pattern (Hoppe et al, unpublished mary problem of patients with CF, it nevertheless data, 2005). Hence, we speculate that in the is important to avoid those additional, but prevent- presence of Oxalobacter species, oxalate elimina- able, complications. tion might occur through transepithelial oxalate efflux into the colonic lumen, as suggested from ACKNOWLEDGMENT 27,28 experiments in rats. The authors thank Ernst Leumann, Zurich, Switzerland, Hypocitraturia is a risk factor for stone disease for editorial advice; Gabriele Benz-Bohm for renal ultra- in children and adults. Because we and others sound examinations; S. van Koningsbruggen for her support found diminished citrate excretion in patients in patient recruiting; nurses of the CF centers of the Univer- sity of Cologne for obtaining urine and stool samples; and A. with CF, they are at an increased risk for nephro- Gradehand (University of Cologne), B. Bär, and M. Klöck- lithiasis and nephrocalcinosis, particularly if hy- ner (University of Bonn) for laboratory assistance. peroxaluria also is present.2,10,29 However, we can only speculate on the reasons for this finding. REFERENCES Worsening of the pulmonary situation with respi- 1. Böhles H, Gebhardt B, Beeg T, Sewell AC, Solem E, ratory acidosis or certain dietary habits (high Posselt G: Antibiotic treatment-induced tubular dysfunction calorie intake, acid load) probably could be the as a risk factor for renal stone formation in cystic fibrosis. culprit. However, we did not see overt acidosis in J Pediatr 140:103-109, 2002 our patients. In addition, 8 of our patients with 2. Hoppe B, Hesse A, Brömme S, Rietschel E, Michalk D: Urinary excretion substances in cystic fibrosis: Risk of CF had hypercalciuria as an isolated or associ- urolithiasis? Pediatr Nephrol 12:275-279, 1998 ated risk factor. 3. Turner MA, Goldwater D, David TJ: Oxalate and We therefore suggest keeping the diet in pa- calcium excretion in cystic fibrosis. Arch Dis Child 83:244- tients with CF low in oxalate, yet high in cal- 247, 2000 cium. In addition, high fluid intake is necessary 4. Strandvik B, Hjelte L: Nephrolithiasis in cystic fibro- sis. Acta Paediatr 82:306-307, 1993 to increase urine volume and solubility product. 5. Böhles H, Michalk D: Is there a risk for kidney stone In patients with persistent hypocitraturia, admin- formation in cystic fibrosis? Helv Paediatr Acta 37:267-272, istration of alkali citrate is advocated, which has 1982 ABSORPTIVE HYPEROXALURIA IN CYSTIC FIBROSIS 445

6. Chidekel AS, Dolan TF Jr: Cystic fibrosis and calcium 18. Leumann EP, Dietl A, Matasovic A: Urinary oxalate oxalate nephrolithiasis. Am J Respir Crit Care Med 151: and glycolate excretion in healthy infants and children. 742A, 1995 (abstr) Pediatr Nephrol 4:493-497, 1990 7. Parries G, Wohl ME, Khaw KT, Lapey A, Bauer SB: 19. Hesse A, Classen A, Knoll M, Timmermann F, Vahl- Urolithiasis: A late complication of CF. Ped Pulmonol Suppl ensieck W: Dependence of urine composition on the age and 12:393, 1995 (abstr) sex of healthy subjects. Clin Chim Acta 160:79-86, 1986 8. Matthews LA, Doershuk CF, Stern RC, Resnick MI: 20. Bach D, Hesse A, Bernal-Sprekelsen MS, Nemat S: Urolithiasis in cystic fibrosis. J Urol 155:1563-1564, 1996 Normal values of lithogenic and inhibitory substances in the 9. Katz S, Krueger LJ, Falkner B: Microscopic nephrocal- urine of healthy children, in Walker VR, Sutton RAL, cinosis in cystic fibrosis. N Engl J Med 319:263-266, 1988 Cameron ECB, Pak CYC, Robertson WG (eds): Urolithia- 10. Sidhu H, Hoppe B, Hesse A, et al: Antibiotic induced sis. New York, NY, Plenum, 1989, pp 775-777 loss of the gut associated bacterium Oxalobacter formi- 21. Hoppe B, Jahnen A, Bach D, Hesse A: Urinary genes: A risk factor for hyperoxaluria in cystic fibrosis calcium-oxalate saturation in healthy infants and children. patients. Lancet 352:1026-1030, 1998 J Urol 158:557-559, 1997 11. Werness PG, Brown CM, Smith LH, Finlayson B: 22. Williams HE, Wandzilak TR: Oxalate synthesis, trans- EQUIL 2: A basic computer program for the calculation of port and the hyperoxaluric syndromes. J Urol 141:742-747, urinary saturation. J Urol 134:1242-1244, 1985 1989 12. Hoppe B, Kemper MJ, Hvizd MG, Sailer DE, Lang- 23. Campieri C, Campieri M, Bertuzzi V, et al: Reduction of oxaluria after an oral course of lactic acid bacteria at high man CB: Simultaneous determination of oxalate, citrate and concentration. Kidney Int 60:1097-1105, 2001 sulfate in children’s plasma with ion-chromatography. Kid- 24. Hokama S, Honma Y, Toma C, Ogawa Y: Oxalate ney Int 53:1348-1352, 1998 degrading Enterococcus faecalis. Microbiol Immunol 44:235- 13. Dick PT, Shuckett BM, Tang B, Daneman A, Kooh 240, 2000 SW: Observer reliability in grading nephrocalcinosis on 25. Holmes RP, Goodman HO, Assimos DG: Contribu- ultrasound examinations in children. Pediatr Radiol 29:68- tion of dietary oxalate to urinary oxalate excretion. Kidney 72, 1999 Int 59:270-276, 2001 14. Sidhu H, Holmes RP, Allison MJ, Peck AB: Direct 26. Read NW: Transit of meal through the stomach, small quantification of the enteric bacterium Oxalobacter formi- intestine and colon in normal subjects and its role in the genes in human fecal samples by quantitative competitive- pathogenesis of diarrhea. Gastroenterology 79:1276-1282, template PCR. J Clin Microbiol 37:1503-1509, 1999 1980 15. Schwartz GJ, Haycock GB, Edelmann CM Jr, Spritzer 27. Hatch M, Freel RW: Renal and intestinal handling of A: A simple estimate of glomerular filtration rate in children oxalate following oxalate loading in rats. Am J Nephrol derived from body length and plasma creatinine. Pediatrics 23:18-26, 2003 58:259-263, 1976 28. Sidhu H, Allison MJ, Chow JM, Clark A, Peck AB: 16. Hesse A, Schneeberger W, Engfeld S, Von Unruh GE, Rapid reversal of hyperoxaluria in a rat model after probiotic Sauerbruch T: Intestinal hyperabsorption of oxalate in cal- administration of Oxalobacter formigenes. J Urol 166:1487- cium oxalate stone formers: Application of a new test with 1491, 2001 13 [ C2]oxalate. J Am Soc Nephrol 10:S329-S333, 1999 (suppl 29. Sidhu H, Schmidt ME, Cornelius JG, et al: Direct 14) correlation between hyperoxaluria/oxalate stone disease and 17. Von Unruh GE, Langer MAW, Paar DW, Hesse A: the absence of the gastrointestinal bacterium Oxalobacter Mass spectrometric ion monitoring assay for an oxalate formigenes: Possible prevention by gut recolonisation or 13 absorption test applying [ C2]oxalate. J Chromatogr B enzyme replacement therapy. J Am Soc Nephrol 10:S334- 716:343-349, 1998 S340, 1999 (suppl 14) The Course of Anticardiolipin Antibody Levels Under Immunoadsorption Therapy

Anna-Christine Hauser, MD, Lorenz Hauser, PhD, Ingrid Pabinger-Fasching, MD, Peter Quehenberger, MD, Kurt Derfler, MD, and Walter Hermann Hörl, MD, PhD

● Background: The emergence of anticardiolipin antibodies in patients with systemic lupus erythematosus is a serious occurrence in regard to a high risk for thrombosis and thromboembolic complications, fetal loss, and renal insufficiency. In an observational analysis, we studied anticardiolipin antibodies during immunoadsorption therapy. Methods: We analyzed the magnitude and time course of serum concentrations of the immunoglobulin G (IgG) and IgM subtypes of anticardiolipin antibodies (CIgG and CIgM) along with IgG and IgM, antinuclear antibodies, and antibodies to double-stranded DNA before and after single immunoadsorption sessions and their long-term course led to a (842 ؍ in 11 patients with systemic lupus erythematosus. Results: Single immunoadsorption sessions (n rapid decline in CIgG and CIgM levels by 62.94% ؎ 21.60% and 42.02% ؎ 22.14%, respectively (P < 0.0001), along -with a corresponding decline in serum levels of antinuclear antibodies (65.04% ؎ 18.83%), antibodies to double stranded DNA (64.67% ؎ 21.20%), IgG (58.11% ؎ 16.84%), and IgM (32.15% ؎ 15.58%). Reduction rates of CIgG and CIgM levels were greater when high initial concentrations (P < 0.0001) and low IgG levels (P < 0.0001) were present. Mean reductions in pretreatment values of CIgG and CIgM during 6 months of immunoadsorption therapy were and for the 1-year period were (21.55 ؍ and 29.39% ؎ 70.41% (mean number of sessions/patient 39.94% ؎ 42.85% Conclusion: We observed that .(30.46 ؍ and 58.05% ؎ 40.16% (mean number of sessions/patient 22.49% ؎ 63.20% immunoadsorption therapy is an effective method to reduce anticardiolipin antibody levels rapidly and keep them at a low level in the long term. Am J Kidney Dis 46:446-454. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Anticardiolipin antibodies; immunoadsorption; systemic lupus erythematosus (SLE); fetal loss; renal insufficiency.

NTIPHOSPHOLIPID ANTIBODIES are a In patients with systemic lupus erythematosus A major part of antiphospholipid antibody (SLE), the prevalence of antiphospholipid anti- syndrome, which is characterized by vascular bodies is particularly high, ranging from 12% to thrombosis and/or repeated fetal loss in associa- 30% of affected patients.1 After 20 years of tion with antibodies to certain plasma proteins follow-up, 50% to 70% of patients with SLE that are bound to anionic phospholipids. These may have developed antiphospholipid antibodies antibodies can occur in patients without clinical and antiphospholipid syndrome.1 Venous throm- evidence of another autoimmune disease, as in bosis is the most common antiphospholipid syn- up to 5% of apparently healthy subjects, or in drome manifestation, occurring in 29% to 55% association with autoimmune or other diseases.1-3 of affected patients during an average follow-up of less than 6 years.1 Pulmonary emboli occur in up to half these patients.1 Stroke and transient From the Department of Medicine III, Division of Nephrol- ischemic attacks, often resulting from emboli ogy and Dialysis; Department of Medicine I, Division of from valve vegetations, as well as coronary occlu- Hematology; Institute of Medical and Chemical Laboratory sions, are additional clinical consequences.4-8 Diagnostics, Medical University of Vienna, Austria; and School of Aquatic and Fishery Sciences, University of Wash- Anticardiolipin antibodies also seem to have a ington, Seattle, WA. negative impact on the renal outcome of patients Received March 8, 2005; accepted in revised form May 9, with lupus nephritis.9 Thrombotic microangiopa- 2005. thy may lead to hemolytic uremic-like syndrome Originally published online as doi:10.1053/j.ajkd.2005.05.023 and renal failure. The severity of the disease is on August 4, 2005. Address reprint requests to Anna-Christine Hauser, MD, often varying and can lead to a devastating Department of Medicine III, Division of Nephrology and Dialy- syndrome with organ manifestation of the kid- sis, Medical University of Vienna, Währinger Gürtel 18-20, ney, lung, central nervous system, heart, and skin A-1090 Vienna, Austria. E-mail: anna-christine.hauser@ and disseminated intravascular coagulation, called meduniwien.ac.at 1,10 © 2005 by the National Kidney Foundation, Inc. catastrophic antiphospholipid syndrome. The 0272-6386/05/4603-0008$30.00/0 optimal therapy for such events is not yet estab- doi:10.1053/j.ajkd.2005.05.023 lished. Anticoagulants, steroids, plasmapheresis,

446 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 446-454 ANTICARDIOLIPIN ANTIBODIES AND IMMUNOADSORPTION 447 or intravenous immune globulins commonly are The observational study comprised 842 immunoadsorp- applied.1,10 Immunoadsorption therapy, now an tion sessions. Mean numbers of immunoadsorption sessions Ϯ established therapy in patients with various auto- per patient were 22.4 12.5 (range, 8 to 41 sessions) during the first 6 months and 42.1 Ϯ 18.4 (range, 19 to 69 sessions; immune diseases and which seems to be superior consecutive quarter years, 13.2, 11.7, 7.5, and 8.6 sessions/ to plasmapheresis in general, has not been stud- quarter year/patient) during the first year. In 6 patients, ied to date in patients with SLE and/or antiphos- immunoadsorption therapy was continued for more than 1 pholipid syndrome.11-19 To gain insight into the year. During the long-term treatment period of more than 1 capacity of immunoadsorption therapy to elimi- year, the quarter-year frequency of immunoadsorption ses- sions per patient was 5.9 (range, 3.0 to 7.3). In 1 patient, nate anticardiolipin antibodies, we studied the immunoadsorption treatment was continued for more than elimination of anticardiolipin antibodies in single 80.4 months. sessions and their course during long-term immu- Ig-Therasorb (Baxter, Munich, Germany) immunoadsorp- noadsorption therapy in a cohort of patients with tion was performed on Sepharose-coupled polyclonal sheep antibodies (Baxter, Munich, Germany) against human immu- SLE. noglobulins, as described previously, in processing 2.5 calcu- lated plasma volumes by each session of plasma per treat- METHODS ment.11-14 In detail, for all treatments, blood was drawn from an antecubital vein through a 15-G dialysis needle at a flow The primary aim of our study is to assess the dynamics of rate of 70 to 100 mL/min. anticardiolipin antibodies during immunoadsorption therapy. Coagulation was prevented with heparin (input rate, 20 For this, in an observational study of 11 patients with SLE U/min; not exceeding 5,000 U/treatment) and citrate as and anticardiolipin antibodies, we analyzed the behavior of anticoagulant citrate dextrose, formula A (ACD-A; Baxter, serum anticardiolipin antibody concentrations (immuno- Munich, Germany). The ratio of citrate to whole-blood flow globulin G [IgG] and IgM subtypes: CIgG, CIgM), along was kept at 1:20. with serum levels of IgG and IgM, antinuclear antibodies For primary plasma separation, an Autopheresis-C Thera- (ANAs), and antibodies to double-stranded DNA (anti- peutic Plasma System (TPS; Baxter, Deerfield, IL) was used, dsDNAs) before and after single immunoadsorption ses- as recently described.13,14,16 With this method, a sole periph- sions and in the long term (up to 80.4 months). Furthermore, eral venovenous approach is possible.14,16 we studied the behavior of proteinuria (n ϭ 10) and preg- Immunoglobulin apheresis was performed in an auto- nancy outcome (n ϭ 4) during immunoadsorption treatment. mated double-needle continuous-flow operation in which The 11 patients with SLE with anticardiolipin antibodies the TPS is connected to an adsorption-desorption-automate were from a cohort of 19 patients with SLE who were treated (ADA; Baxter, Munich, Germany) controlling the flow of with immunoadsorption within a 9.5-year observation pe- plasma and regeneration solutions. riod (8 of 19 patients did not have anticardiolipin antibod- For immunoadsorption, 2 reusable columns, each contain- ies). All 11 patients had SLE fulfilling the American College ing 150 mL of Sepharose coupled with polyclonal sheep of Rheumatology criteria for SLE.20,21 Immunoadsorption antibodies to human IgG heavy and light chains (Ig- therapy was applied in patients with SLE when the disease Therasorb) were used. Each column had an immunoglobulin- was resistant to conventional medical treatment with cyclo- 22 binding capacity of approximately 4.0 g. In each adsorption phosphamide or this substance was contraindicated. In 1 cycle (lasting ϳ15 minutes), 400 mL of plasma was loaded patient, immunoadsorption treatment was terminated after on 1 column with a plasma flow rate of 25 to 40 mL/min. only 2 sessions because of the occurrence of hemolysis and This system uses 2 columns, one of which is loading while thrombocytopenia accounted to hemolytic uremic syn- the other one is desorbing by protein elution with glycin drome. In this patient, immunoadsorption treatment was buffer at pH 2.8, followed by washing cycles with phosphate- replaced by plasma exchange therapy. The dynamics of buffered saline and 9 g/L of isotonic sodium chloride solu- anticardiolipin antibodies, but not the course of proteinuria, tion. Mean duration of extracorporeal treatment was approxi- was assessed in this case. All patients were on prednisolone mately 4.0 hours. therapy before and during immunoadsorption treatment. Cyclophosphamide dosage was reduced during immunoad- sorption treatment. Ten of 11 patients were on cyclophospha- Laboratory Measurements mide therapy at onset and no patient was on this therapy at Anticardiolipin antibodies were detected by means of a the end of immunoadsorption treatment. Seven of 11 pa- commercially available indirect noncompetitive enzyme im- tients received anticoagulation therapy (phenprocoumon or munoassay (Varelisa; Pharmacia Diagnostics, Vienna, Aus- low-molecular-weight heparin) before immunoadsorption tria) and expressed as IgG (GPLU) and IgM phospholipid treatment, 9 of 11 patients during the observation period of units (MPLU). ANAs were detected by means of immunoflu- immunoadsorption treatment, and all 5 patients who finished orescence on HEp-2 cells (Kallestad HEp-2 cell line slide; immunoadsorption treatment. Bio-Rad, Redmond, WA). For measurement of anti-dsDNA All patients gave informed consent to both treatment of in human serum, a commercially available radioimmunoas- SLE and analysis of data obtained in the course of long-term say was used (Anti-dsDNA Kit; Trinity Biotech, Dublin, clinical care. Antiphospholipid syndrome was assumed when Ireland). Quantitative determination of human IgG and IgM both clinical manifestations and serum antibodies were was performed by means of nephelometry, using N antisera present.1,2 Patient characteristics are listed in Table 1. to human IgG and IgM (Dade Behring, Marburg, Germany). 448 HAUSER ET AL

Table 1. Patient Characteristics

No. of patients with SLE with anticardiolipin antibodies 11 Age (y) 34.2 Ϯ 9.9 (19-47) Female (%) 9/11 (81.8) Body mass index (kg/m2) 22.79 Ϯ 4.1 (18.3-30.5) History of catastrophic APS before onset of IAS treatment 5/11 Before onset of IAS treatment 63.1 patient-years Cerebral (stroke or transient ischemic attack) 7 events/6 patients Fetal loss 8 events/4 patients At onset of IAS treatment Disease duration (y) 5.74 Ϯ 6.33 (0.2-17) Erythrocytes (1012/L) 3.58 Ϯ 0.63 (2.77-4.70) Leukocytes (109/L) 8.22 Ϯ 4.15 (2.90-19.40) Platelets (109/L) 142.81 Ϯ 104.12 (5-285,000) Creatinine clearance (mL/min) 90.73 Ϯ 29.21 (49.80-136.40) Renal (CKD stages 1-3) 11 Proteinuria (g/24 h) 3.48 Ϯ 4.23 (0.5-12) Observation period (patient-years) 27.45 Of IAS treatment (mo) 29.87 Ϯ 31.87 (0-80.4) No. of IAS sessions (total) 842 No. of patients 11 Mean no. of IAS sessions/patient 76 Ϯ 71 (2-210) Cerebral (stroke or transient ischemic attack) 4 events/3 patients Fetal loss 3 events/4 patients Successful pregnancy outcome 1 (twins) At end of observation period Creatinine clearance (mL/min) 94.11 Ϯ 26.82 (42.48-125.34)* Proteinuria (g/24 h) 0.97 Ϯ 1.81 (0.04-5.7)† No. of patients finishing IAS treatment 5 No. of patients continuing IAS treatment after observation period 6 After ending the IAS treatment Disease duration after IAS therapy to end of observation period (patient-years) 34.75 Cerebral (stroke or transient ischemic attack) 13 events/5 patients

NOTE. Values expressed as mean Ϯ SD (range) or number (percent) unless noted otherwise. Creatinine clearance was estimated by means of the Cockcroft-Gault equation.23 Stage of chronic kidney disease was defined according to the National Kidney Foundation–Kidney Disease Outcomes Quality Initiative Clinical Practice Guidelines.41 To convert creati- nine clearance in mL/min to mL/s, multiply by 0.01667. Abbreviations: APS, antiphospholipid antibody syndrome; IAS, immunoadsorption; CKD, chronic kidney disease. *Creatinine clearance at the onset of immunoadsorption therapy versus at the end of the observation period (P ϭ 0.4951). †Proteinuria at onset of immunoadsorption therapy versus at the end of the observation period (P ϭ 0.0152).

Creatinine clearance was estimated using the Cockcroft- square tests were applied. P less than 0.05 is considered Gault equation.23 Proteinuria was assessed in the absence of significant. urinary tract infections (negative Uricult [Orion Dianostica, Espoo, Finland] test result) using a commercially available RESULTS colorimetric assay (U/CSF Protein; Roche Diagnostics, Vi- enna, Austria). Changes in Parameters During Single Immunoadsorption Sessions Statistical Analysis Single immunoadsorption sessions led to a Data are presented as mean Ϯ SD, absolute number, or percentage unless otherwise indicated. For comparisons of rapid decline in anticardiolipin antibody levels. means, 2-sided paired and unpaired t-tests or Wilcoxon Mean CIgG level decreased from 16.94 Ϯ signed rank test and Mann-Whitney U test as nonparametric 18.28 to 5.25 Ϯ 6.29 GPLU/mL (P Ͻ 0.0001; Fig tests were applied, as appropriate. For multiple comparisons, 1), representing a mean reduction of 62.94% Ϯ analysis of variance or Kruskal-Wallis test was used. Corre- 21.60% (Fig 2). On 60 occasions (7.13%), serum lation testing was performed by using simple regression or Spearman rank correlation analysis. Calculations of mean CIgG posttreatment values decreased to less than reduction rates were based on measurable concentrations. the detection limit (Ͻ1 GPLU/mL). The reduction For comparisons of groups for categorical variables, chi- rate was dependent on CIgG level before immuno- ANTICARDIOLIPIN ANTIBODIES AND IMMUNOADSORPTION 449

corresponding reduction rates for CIgG and IgG (P Ͻ 0.0001; Fig 2). Percentages of reduction rates were dependent on the quartile of pretreatment values. The reduc- tion in IgG levels was similar for quartiles 1 and 4, but greater in quartiles 2 and 3 (Table 2). The reduction rate of IgM also increased with higher quartile serum levels (Table 2). Reduction rates of CIgG, CIgM, ANA, and anti- dsDNA were influenced by serum IgG levels; the greater the serum IgG concentration, the lower the reduction rate (P Ͻ 0.0001; Figs 3 and 4). Fig 1. Serum concentrations of anticardiolipin anti- Changes in Parameters During Short-Term and bodies (IgG subtype CIgG and IgM subtype CIgM) before and after single immunoadsorption sessions (mean ؎ Long-Term Immunoadsorption Treatment SD), showing a statistically significant decrease in both Immunoadsorption sessions led to a rapid de- subtypes. CIgG pretreatment value versus CIgG post- ϭ treatment value,*P < 0.0001. CIgM pretreatment value cline in both serum CIgG (P 0.0073) and versus CIgM posttreatment value, ؉P < 0.0001. CIgM (P ϭ 0.0186) levels within a 30-day period (74 immunoadsorption sessions). A signifi- adsorption sessions. The reduction was less in the cant decrease in serum levels during 30 days also lower quartile of CIgG pretreatment concentrations was found for IgG (P ϭ 0.0226), but not for and increased continuously with the higher quar- ANA (P ϭ 0.8520), anti-dsDNA (P ϭ 0.9657), tiles of this antibody (P Ͻ 0.0001; Table 2). and IgM (P ϭ 0.9661). The behavior of CIgG CIgM levels showed similar behavior. Mean pretreatment and posttreatment values in the 180- serum CIgM value decreased from 4.68 Ϯ 5.81 day treatment period is shown in Fig 5 (P Ͻ to 2.57 Ϯ 2.59 MPLU/mL during single immuno- 0.0001). Mean reductions in CIgG and CIgM adsorption sessions (P Ͻ 0.0001; Fig 1), repre- levels were 42.85% Ϯ 39.94% and 29.39% Ϯ senting a mean reduction of 42.04% Ϯ 22.18% 70.41% during 6 months (P ϭ 0.0007) and (Fig 2). On 123 occasions (16.04%), posttreat- 63.20% Ϯ 22.49% and 58.05% Ϯ 40.16% dur- ment serum levels were less than the detection ing the 1-year period (P ϭ 0.0005), respectively. limit (Ͻ1 MPLU/mL). The percentage of CIgM After 1 year on immunoadsorption treatment, a level reduction increased significantly with pre- treatment values from the lowest to the highest quartile of this antibody level (P Ͻ 0.0001; Table 2). In 77.71% (CIgG) and 45.76% (CIgM) of sessions, serum values decreased by more than 50% or to levels less than the detection limit (P Ͻ 0.0001). Corresponding mean reduction rates of ANA titers and anti-dsDNA, IgG, and IgM serum levels were 65.04% Ϯ 18.83%, 64.67% Ϯ 21.20%, 58.11% Ϯ 16.84%, and 31.15% Ϯ 15.58%, respectively (Fig 2). IgG and IgM post- session serum levels were less than the detection limit in 15.10% and 15.24% of overall immuno- adsorption sessions (detection limits: IgG, Ͻ195 mg/dL [Ͻ1.95 g/L]; IgM, Ͻ30 mg/dL [Ͻ0.30 Fig 2. Percentage of reduction (mean ؎ SD) in serum levels of anticardiolipin antibodies (IgG sub- g/L], respectively). In 74.70% (IgG) and 25.35% type CIgG and IgM subtype CIgM), IgG, IgM, ANAs, and (IgM) of sessions, serum values decreased by anti-dsDNAs during single immunoadsorption ses- more than 50% or to levels less than the detec- sions. Reduction rates on CIgM and IgM levels were Ͻ statistically significantly lower than those for CIgG tion limit (P 0.0001). Reduction rates for and IgG. CIgM versus CIgG, *P < 0.0001. IgM versus CIgM and IgM were significantly less than the IgG, #P < 0.0001. 450 HAUSER ET AL

Table 2. Percentage of Reduction in Parameters During Single Immunoadsorption Sessions

P (analysis Total Quartile 1 Quartile 2 Quartile 3 Quartile 4 of variance) CIgG (GPLU/mL) (n ϭ 195) (n ϭ 196) (n ϭ 195) (n ϭ 196) Ͻ5.40 5.40-10.40 10.41-22.40 Ͼ22.40 % Reduction mean 62.94 54.69 59.99 64.15 72.89 Ͻ0.0001 SD 21.60 24.00 21.52 19.70 16.43 CIgM (MPLU/mL) (n ϭ 161) (n ϭ 161) (n ϭ 161) (n ϭ 161) Ͻ2.21 2.21-3.30 3.31-5.80 Ͼ5.80 % Reduction mean 42.02 30.86 39.44 42.07 55.75 Ͻ0.0001 SD 22.18 21.49 17.96 18.76 22.95 ANA (titre) (n ϭ 197) (n ϭ 196) (n ϭ 197) (n ϭ 197) Ͻ320 320-640 641-1,280 Ͼ1,280 % Reduction mean 65.04 61.18 58.94 73.03 66.84 0.0002 SD 18.83 16.35 15.03 16.10 23.65 Anti-dsDNA (IU/mL) (n ϭ 69) (n ϭ 70) (n ϭ 69) (n ϭ 70) Ͻ8.70 8.70-21.15 21.16-53.70 Ͼ53.70 % Reduction mean 64.67 43.01 56.07 77.20 82.28 Ͻ0.0001 SD 21.20 13.43 13.48 13.37 15.90 IgG (mg/dL) (n ϭ 177) (n ϭ 177) (n ϭ 177) (n ϭ 177)* Ͻ519.50 519.50-766.50 766.51-959.50 Ͼ959.50 % Reduction mean 58.11 51.07 64.03 65.20 52.15 Ͻ0.0001 SD 16.84 12.55 13.72 19.75 15.21 IgM (mg/dL) (n ϭ 107) (n ϭ 107) (n ϭ 107) (n ϭ 107)† Ͻ46.00 46.00-62.00 62.01-81.00 Ͼ81.00 % Reduction mean 32.15 15.30 33.32 40.20 39.79 Ͻ0.0001 SD 15.58 13.42 10.80 10.21 12.84

NOTE. Values expressed as mean Ϯ SD of the pretreatment value of anticardiolipin antibodies (IgG subtype CIgG and IgM subtype CIgM), ANAs, anti-dsDNAs, IgG, and IgM during single immunoadsorption sessions. Data are presented for total and quartile serum concentrations. To convert IgG and IgM in mg/dL to g/L, multiply by 0.01. *Reduction rates of quartile 4 versus those of quartile 3 and quartile 2, P ϭ 0.001, P ϭ 0.001. †Reduction rates of quartile 4 versus quartile 3, P ϭ 0.8034. significant decline could be observed for pretreat- Clinical Results ϭ ment immunoadsorption values for ANA (P Proteinuria and renal function under immuno- Ͻ 0.0284) and anti-dsDNA (P 0.0001), as well adsorption treatment. Proteinuria decreased as IgM (P ϭ 0.0060). The decrease in IgG from protein of 3.48 Ϯ 4.23 g/24 h (median, levels reached statistical significance for the 1.10 g/24 h; range, 0.5 to 12 g/24 h) at the 180-day period (P ϭ 0.0139), but not the ϭ onset of immunoadsorption treatment to 0.97 365-day period (P 0.1047). Ϯ Five of 6 patients who were maintained on 1.81 g/24 h (median, 0.24 g/24 h; range, immunoadsorption therapy for more than 1 0.04 to 5.7 g/24 h) at the end of treatment or year showed a transient increase in CIgG val- end of the treatment observation period, when ues. During this period, mean frequency of immunoadsorption therapy was still continued ϭ immunoadsorption sessions (sessions per pa- (P 0.0152). In 3 of 10 patients, proteinuria tient per quarter year) remained unchanged subsided, and in an additional 3 patients, reduc- (P ϭ 0.2095). CIgG levels showed a positive tion in proteinuria exceeded 50%. In 2 pa- relationship with CIgM (P Ͻ 0.0001) and tients, reduction in proteinuria was 20% or anti-dsDNA levels (P Ͻ 0.0032), but not ANA greater and less than 50%. In 2 patients, no serum titers (P ϭ 0.868). response was obtained. Glomerular filtration ANTICARDIOLIPIN ANTIBODIES AND IMMUNOADSORPTION 451

Fig 3. Relationship between the reduction (per- Fig 4. Relationship between the reduction (per- cent) in serum levels of anticardiolipin antibodies (IgG cent) in serum anti-dsDNA levels during single immu- subtype CIgG) during single immunoadsorption ses- noadsorption sessions and serum IgG pretreatment sions and serum IgG pretreatment values (milligrams values (milligrams per deciliter). Greater serum IgG per deciliter). Greater serum IgG levels are associated levels are associated with lower reduction rates (per- with lower reduction rates in CIgG levels: correlation cent) in anti-dsDNA levels: correlation coefficient (r2), coefficient (r2), 0.157; P < 0.0001. To convert IgG in 0.448; P < 0.0001. To convert IgG in mg/dL to g/L, mg/dL to g/L, multiply by 0.01. multiply by 0.01. rate remained stable. Creatinine clearance was Side Effects 90.73 Ϯ 29.21 mL/min (range, 49.80 to 136.40 No serious side effects occurred during all 842 mL/min [1.51 Ϯ 0.49 mL/s; range, 0.83 to 2.27 immunoadsorption sessions. mL/s]) at the initiation of immunoadsorption treatment and 94.11 Ϯ 26.82 mL/min (range, DISCUSSION 42.48 to 125.34 mL/min [1.57 Ϯ 0.45 mL/s; Anticardiolipin antibodies are associated with range, 0.71 to 2.09 mL/s]) at the end of the a major risk for thrombosis and thromboembolic observation period (P ϭ 0.4951). events. Moreover, they seem to have an adverse Pregnancy outcome. Four patients received impact on the outcome of lupus nephritis.9,24 immunoadsorption therapy during pregnancy. Although anticoagulant treatment has proven effi- All had repeated fetal loss before immunoad- cacious for secondary prevention of antiphospho- sorption treatment; 3 of 4 patients experienced lipid syndrome–related thrombosis, the usefulness 2 or more fetal losses. In 3 patients, immunoad- of such therapy in patients with other microvascu- sorption treatment was initiated between gesta- lar complications, in particular, antiphospholipid tion weeks 8 and 14, and 1 patient gave birth to healthy twins at gestation week 36. This pa- tient with high activity SLE had 2 fetal losses before the initiation of immunoadsorption treat- ment. In the remaining 2 patients, fetal loss occurred at gestation weeks 22 and 24. One patient became pregnant when already treated with immunoadsorption for serious SLE for 2 years. Pregnancy loss occurred at gestation week 22 despite continuation of immunoad- sorption therapy.

Thrombotic and Thromboembolic Complications Our study covered 27.45 patient-years. In Fig 5. Bivariate scattergram (supersmoother) show- this period, 4 cerebral events (stroke or tran- ing the course of pretreatment (upper line) and post- treatment values (lower line) for CIgG of anticardio- sient ischemic attack) occurred in 3 patients lipin antibodies in 6 patients during their 180-day (Table 1). treatment period. 452 HAUSER ET AL syndrome nephropathy, remains unknown.2,25 Par- Six patients were treated for more than 1 year. ticularly in patients with severe antiphospholipid In 5 of these patients, a transient increase in antibody syndrome, prompt elimination of antiphos- CIgG levels was noted thereafter. We believe this pholipid antibodies is targeted.1,10,26 occurrence was caused by varying disease activ- In our study, we observed that immunoadsorp- ity because overall CIgG serum level showed a tion therapy is an excellent technique to reduce positive relationship with anti-dsDNA serum con- anticardiolipin antibody levels. Immunoadsorption centration, and the frequency of immunoadsorp- sessions overall lead to rapid declines in both CIgG tion sessions remained unchanged after 1 year of and CIgM serum levels. Thereby, mean reductions treatment. Moreover, many publications reported in CIgG levels were 62.94% Ϯ 21.60%, and in that anticardiolipin antibodies developed in the CIgM levels, 42.04% Ϯ 22.14%. Percentages for long-term course of SLE.1-3 However, one can- serum ANA titer and anti-dsDNA, IgG, and IgM not exclude that long-term immunoadsorption serum levels were similar. Percentages of reduction therapy per se might have induced the develop- in anticardiolipin antibody levels, as well as ANA ment of anticardiolipin antibodies. Such a phe- and anti-dsDNA levels, were influenced by the nomenon was observed in long-term hemodialy- pretreatment concentrations. The upper quartiles of sis patients, for whom the frequency of antibodies concentrations showed greater reduction rates than seemed to depend on the duration of hemodialy- lower quartiles. This phenomenon also applies to sis treatment.1,31,32 It is assumed that oxidative IgM, but not IgG, for which greater IgG levels were stress during extracorporeal therapy might cause associated with a decline in reduction rate. More- this occurrence.31,32 over, reductions in CIgG, CIgM, ANA, and anti- Alternatively, our findings suggest that the fre- dsDNA levels were dependent on serum IgG lev- quency of approximately 6 immunoadsorption ses- els. Thereby, greater serum IgG levels were sions per patient per quarter year in the long-term associated with decreasing reduction rates. We be- course (in contrast to ϳ13 immunoadsorption ses- lieve these phenomena can be accounted for by the sions per patient per quarter year at the beginning) competition of compounds for protein A binding was too few to keep CIgG and CIgM in the major- sites, differences in their affinity to these binding ity of patients at a low level equal to those at the sites, and limits in the immunoadsorption capacity end of the 180-day treatment period, which was of the columns, as in the case of IgG.27 As a less than 7.5 GPLU/mL on average. therapeutic consequence, replacement of IgG dur- In previous experimental animal studies, it was ing therapy, which seems to have no clinical ben- shown that antibody depletion is not followed by efit, can diminish the removal of anticardiolipin increased antibody synthesis.12 To assess whether antibodies.28 these results are similar to antibody depletion in Overall, our data show that immunoadsorption humans, free light chains, as markers of current therapy is a feasible technique to decrease el- antibody synthesis, in 8 patients treated with immu- evated serum CIgG and CIgM levels within a noadsorption were analyzed by us in a previous very short period. This finding is of considerable study.12 Specific and bulk immunoglobulin levels impact because other types of treatment fail to were obtained simultaneously. Mean serum free decrease serum antiphospholipid antibody levels light chain concentration increased to the preimmu- effectively.29 The immunoadsorption technique noadsorption value within 1 day and remained especially favors the removal of CIgG by 2 unchanged thereafter. Total IgG and specific anti- mechanisms: first, by better elimination of the body concentrations increased to pretreatment val- IgG subtype in general, and second, by the greater ues in 88% and 43% of patients, respectively; they efficacy of immunoadsorption in the setting of remained less than the original values in the others. greater blood levels, as in the case of CIgG. This Thus, the lack of increased free light chain synthe- phenomenon is important because several stud- sis after immunoadsorption confirms the absence ies showed that CIgG led to a greater risk for of a feedback mechanism regulating antibody syn- thromboembolism than CIgM.1,30 thesis. The restoration of serum IgG levels after Mean serum CIgG and CIgM levels decreased immunoadsorption therefore does not result from further over a 1-year treatment period. A statisti- increased antibody synthesis, but probably is re- cally significant decrease also could be noted for lated to changes in catabolism and immunoglobu- serum ANA titers and anti-dsDNA levels. lin backflow.12 ANTICARDIOLIPIN ANTIBODIES AND IMMUNOADSORPTION 453

Removal of anticardiolipin antibodies is an pregnancies. Full-term delivery was achieved in additional advantage of immunoadsorption only 26% of women with high-activity SLE and therapy in patients with severe SLE, for whom it 61% of patients with mild SLE activity.38 The is an efficacious procedure to eliminate ANAs presence of anticardiolipin antibodies further de- and anti-dsDNAs in principle, in addition to teriorates pregnancy outcome, increasing the rate plasmapheresis.1,33 Recently, we showed in a of fetal loss from 15% to 40% in 1 study, as well negatively selected cohort of patients with SLE as the frequency of preterm delivery.38-40 In our that immunoadsorption treatment led to a signifi- study, only 1 of 4 pregnancies was successful. cant response in terms of reduction in protein- Thus, immunoadsorption treatment does not seem uria, improvement in global disease activity, de- to be effective in avoiding pregnancy losses in creased anti-dsDNA levels, and lower general. However, immunoadsorption treatment cyclophosphamide and glucocorticoid dosages.22 seems to be favorable in selected cases, as the Immunoadsorption therefore can be a rescue successful outcome shows in 1 woman with high therapy in patients with severe SLE because it is activity SLE who had 2 fetal losses before immu- acceptably safe with regard to severe bacterial noadsorption treatment. and viral infections, considering that particularly An important issue is the cost efficacy of long-term cyclophosphamide treatment in pa- immunoadsorption therapy, which depends on tients with SLE is accompanied by a high risk for the frequency of use in single patients. Because a infections.34-37 In a previous investigation com- pair of columns (priced at ϳ$16,700 US) can be paring immunoadsorption treatment with conven- used up to 50 times in an individual patient, the tional therapy with cyclophosphamide and glu- cost of a single treatment declines rapidly with cocorticoid, a strikingly lower rate of infectious increasing duration of therapy. complications was shown in patients on long- In conclusion, we observed that immunoad- term immunoadsorption therapy.35 Furthermore, sorption is a technique to eliminate both CIgG the cumulative probability of hospitalization for and CIgM rapidly, particularly in the presence infection was 10-fold less in the immunoadsorp- of low IgG serum levels. Immunoadsorption tion group compared with conventionally treated was especially effective in reducing the level patients with SLE.35 Plasma exchange, which is of CIgG, which increases the risk for thrombo- a crude approach to autoantibody and immune embolic events in particular.30 In the long-term complex removal, generally has not been effec- course, the appropriate frequency of immuno- tive in prospective trials. When combined with adsorption sessions has to be adjusted to indi- intravenous pulse cyclophosphamide therapy, an vidual disease activity. increase in fatal viral and bacterial infections 37 was encountered. REFERENCES Immunoadsorption treatment also diminished 1. Levine JS, Branch DW, Rauch J: The antiphospholipid 22 proteinuria in the present study. In3of10 syndrome. N Engl J Med 346:752-763, 2002 patients, proteinuria subsided, and in an addi- 2. Fakhouri F, Noel LH, Zuber J, et al: The expanding tional 3 patients, reduction in proteinuria was spectrum of renal diseases associated with antiphospholipid more than 50%. Because sequential biopsies syndrome. Am J Kidney Dis 41:1205-1211, 2003 3. Petri M: Epidemiology of the antiphospholipid anti- were not performed, it is uncertain whether this body syndrome. J Autoimmun 15:145-151, 2000 finding can be accounted to the removal of anti- 4. Levine SR, Brey RL, Tilley BC, et al, for the APASS cardiolipin antibodies or anti-dsDNAs alone or Investigators: Antiphospholipid antibodies and subsequent both of them. Moreover, the interpretation of thrombo-occlusive events in patients with ischemic stroke. renal biopsies may be inconclusive because an- JAMA 291:576-584, 2004 5. Chapman J, Rand JH, Brey RL, et al: Non-stroke tiphospholipid antibodies can be associated with neurological syndromes associated with antiphospholipid renal pathological states other than the typical antibodies: Evaluation of clinical and experimental studies. finding of thrombotic microangiopathy and can Lupus 12:514-517, 2003 show various types of glomerulonephritis.2 6. Brey RL, Chapman J, Levine SR, et al: Stroke and the SLE has an adverse effect on pregnancy out- antiphospholipid syndrome: Consensus meeting Taormina 2002. Lupus 12:508-513, 2003 come, depending on disease activity. In an analy- 7. Petri M: The lupus anticoagulant is a risk factor for sis of 266 female patients with SLE, Clowse et myocardial infarction (but not atherosclerosis): Hopkins al38 found fetal loss in approximately 15% of Lupus Cohort. Thromb Res 114:593-595, 2004 454 HAUSER ET AL

8. Lockshin M, Tenedios F, Petri M, et al: Cardiac disease antiphospholipid-antibody syndrome. N Engl J Med in the antiphospholipid syndrome: Recommendations for treat- 332:993-997, 1995 ment. Committee consensus report. Lupus 12:518-523, 2003 26. Asherson RA, Cervera R, de Groot PG, et al, for the 9. Moroni G, Ventura D, Riva P, et al: Antiphospholipid Catastrophic Antiphospholipid Syndrome Registry Project antibodies are associated with an increased risk for chronic Group: Catastrophic antiphospholipid syndrome: Interna- renal insufficiency in patients with lupus nephritis. Am J tional consensus statement on classification criteria and Kidney Dis 43:28-36, 2004 treatment guidelines. Lupus 12:530-534, 2003 10. Asherson RA, Cervera R, Piette JC, et al: Cata- 27. Braun N, Gutenberger S, Erley CM, Risler T: Immu- strophic antiphospholipid syndrome: Clues to the pathogen- noglobulin and circulating immune complex kinetics during esis from a series of 80 patients. Medicine (Baltimore) immunoadsorption onto protein A Sepharose. Transfus Sci 80:355-377, 2001 19:S25-S31, 1998 (suppl 1) 11. Haas M, Mayr N, Zeitlhofer J, Goldammer A, Derfler 28. Schmaldienst S, Muellner M, Goldammer A, et al: K: Long-term treatment of myasthenia gravis with immuno- Intravenous immunoglobulin application following immuno- adsorption. J Clin Apheresis 17:84-87, 2002 adsorption: Benefit or risk in patients with autoimmune 12. Goldammer A, Derfler K, Herkner K, Bradwell AR, diseases? Rheumatology 40:513-521, 2001 Hörl WH, Haas M: Influence of plasma immunoglobulin 29. Joseph RE, Radhakrishnan J, Appel GB: Antiphospho- level on antibody synthesis. Blood 100:353-355, 2002 lipid antibody syndrome and renal disease. Curr Opin Neph- 13. Schmaldienst S, Jansen M, Hollenstein U, et al: rol Hypertens 10:175-181, 2001 Treatment of systemic lupus erythematosus by immunoad- 30. Galli M, Luciani D, Bertolini G, Barbui T: Anti-beta sorption in a patient suffering from tuberculosis. Am J 2-glycoprotein I, antiprothrombin antibodies, and the risk of Kidney Dis 39:415-418, 2002 thrombosis in the antiphospholipid syndrome. Blood 102: 14. Schmaldienst S, Goldammer A, Spitzauer S, Derfler 2717-2723, 2003 K, Hörl WH, Knöbl P: Local anticoagulation of the extracor- 31. Hauser AC, Pabinger-Fasching I, Quehenberger P, poreal circuit with heparin and subsequent neutralization Kettenbach J, Hörl WH: A patient with sudden abdominal with protamine during immunoadsorption. Am J Kidney Dis pain 10 years after successful renal transplantation. Nephrol Dial Transplant 18:1021-1025, 2003 36:490-497, 2000 32. Molino D, De Santo NG, Marotta R, Anastasio P, 15. Laczika K, Knapp S, Derfler K, Soleiman A, Hörl Mosavat M, De Lucia D: Plasma levels of plasminogen WH, Druml W: Immunoadsorption in Goodpasture’s syn- activator inhibitor type 1, factor VIII, prothrombin activa- drome. Am J Kidney Dis 36:392-395, 2000 tion fragment 1ϩ2, anticardiolipin, and antiprothrombin 16. Jansen M, Schmaldienst S, Banyai S, et al: Treatment antibodies are risk factors for thrombosis in hemodialysis of coagulation inhibitors with extracorporeal immunoadsorp- patients. Semin Nephrol 24:495-501, 2004 tion (Ig-Therasorb). Br J Haematol 112:91-97, 2001 33. Lewis EJ, Hunsicker LG, Lan SP, Rohde RD, Lachin 17. Staudt A, Böhm M, Knebel F, et al: Potential role of JM: A controlled trial of plasmapheresis therapy in severe autoantibodies belonging to the immunoglobulin G-3 sub- lupus nephritis. The Lupus Nephritis Collaborative Study class in cardiac dysfunction among patients with dilated Group. N Engl J Med 326:1373-1379, 1992 cardiomyopathy. Circulation 106:2448-2453, 2002 34. Balow JE, Austin HA III: Maintenance therapy for 18. Böhm M, Dörner T, Knebel F, Bruns A, Jochmann N, lupus nephritis—Something old, something new. N Engl Baumann G: Longlasting effects of immunoadsorption in se- J Med 350:1044-1046, 2004 vere Sjogren’s syndrome. Ann Rheum Dis 63:214-215, 2004 35. Stummvoll GH, Aringer M, Jansen M, Smolen JS, 19. Rabitsch W, Knoebl P, Greinix H, et al: Removal of Derfler K, Graninger WB: Immunoadsorption (IAS) as a persisting isohaemagglutinins with Ig-Therasorb immunoad- rescue therapy in SLE: Consideration on safety and efficacy. sorption after major ABO-incompatible non-myeloablative Wien Klin Wochenschr 116:716-724, 2004 allogeneic haematopoietic stem cell transplantation. Neph- 36. Contreras G, Pardo V, Leclercq B, et al: Sequential rol Dial Transplant 18:2405-2408, 2003 therapies for proliferative lupus nephritis. N Engl J Med 20. Tan EM, Cohen AS, Fries JF, et al: The 1982 revised 350:971-980, 2004 criteria for the classification of systemic lupus erythemato- 37. Aringer M, Smolen JS, Graninger WB: Severe infec- sus. Arthritis Rheum 25:1271-1277, 1982 tions in plasmapheresis-treated systemic lupus erythemato- 21. Petri M, Magder L: Classification criteria for systemic sus. Arthritis Rheum 41:414-420, 1998 lupus erythematosus: A review. Lupus 13:829-837, 2004 38. Clowse ME, Magder LS, Witter F, Petri M: The 22. Stummvoll GH, Aringer M, Smolen JS, et al: IgG impact of increased lupus activity on obstetric outcomes. immunoadsorption (IAS) reduces SLE activity and protein- Arthritis Rheum 52:514-521, 2005 uria—A long term observational study. Ann Rheum Dis 39. McNeil HP, Chesterman CN, Krilis SA: Immunology 64:1015-1021, 2005 and clinical importance of antiphospholipid antibodies. Adv 23. Cockcroft DW, Gault MH: Prediction of creatinine Immunol 49:193-280, 1991 clearance from serum creatinine. Nephron 16:31-41, 1976 40. Clark CA, Spitzer KA, Nadler JN, Laskin CA: Pre- 24. Alba P, Bento L, Cuadrado MJ, et al: Anti-dsDNA, term deliveries in women with systemic lupus erythemato- anti-Sm antibodies, and the lupus anticoagulant: Significant sus. J Rheumatol 30:2127-2132, 2003 factors associated with lupus nephritis. Ann Rheum Dis 41. National Kidney Foundation: K/DOQI Clinical Prac- 62:556-560, 2003 tice Guidelines for Chronic Kidney Disease: Evaluation, 25. Khamashta MA, Cuadrado MJ, Mujic F, Taub NA, classification and stratification. Am J Kidney Dis 39:S46- Hunt BJ, Hughes GR: The management of thrombosis in the S47, 2002 (suppl 1) Relationship of Phosphorus and Calcium-Phosphorus Product With Mortality in CKD

Vandana Menon, MD, Tom Greene, PhD, Arema A. Pereira, MBBS, Xuelei Wang, MS, Gerald J. Beck, PhD, John W. Kusek, PhD, Alan J. Collins, MD, Andrew S. Levey, MD, and Mark J. Sarnak, MD

● Background: Abnormalities of mineral metabolism are prevalent in patients with kidney failure and are associ- ated with increased risk for cardiovascular events. There are limited data investigating relationships of phosphorus and calcium-phosphorus product with outcomes in patients with chronic kidney disease (CKD) stages 3 to 4. Methods: Serum phosphorus and calcium were measured at baseline in 840 participants from the randomized cohort of the Modification of Diet in Renal Disease Study. Survival status until December 31, 2000, was obtained from the National Death Index. Cox models were performed to assess the relationship of serum phosphorus level and calcium-phosphorus product with all-cause and cardiovascular disease (CVD) mortality. Results: Mean serum ,phosphorus level was 3.8 ؎ 0.7 mg/dL (1.23 ؎ 0.23 mmol/L), calcium-phosphorus product was 34.7 ؎ 6.3 mg2/dL2 and glomerular filtration rate (GFR) was 33 ؎ 12 mL/min/1.73 m2 (0.55 ؎ 0.20 mL/s/1.73 m2). All-cause and CVD mortality rates were 25% and 15%. Serum phosphorus level was not related to all-cause mortality in multivariable In unadjusted analysis, serum phosphorus level was associated with (hazard ratio [HR] per 1 .(0.46 ؍ models (P increased risk for CVD mortality, but this (0.02 ؍ mg/dL increase, 1.34; 95% confidence interval [CI], 1.05 to 1.71; P association was partly attenuated and not statistically significant after adjustment for GFR and other confounders Calcium-phosphorus product was not associated with all-cause mortality .(0.12 ؍ HR, 1.27; 95% CI, 0.94 to 1.73; P) Calcium-phosphorus product was related to CVD .(0.35 ؍ or multivariate analysis (P (0.23 ؍ in unadjusted (P analysis, but this (0.04 ؍ mortality in unadjusted (HR per 10 mg2/dL2 increase, 1.30; 95% CI, 1.01 to 1.69; P association was not statistically significant after adjustment for GFR and other confounders (HR, 1.22; 95% CI, 0.89 Conclusion: In the Modification of Diet in Renal Disease Study cohort, serum phosphorus level .(0.23 ؍ to 1.66; P and calcium-phosphorus product were not statistically associated with all-cause or CVD mortality after adjustment for GFR; however larger studies with additional statistical power are needed to evaluate these relationships, especially in the context of current practice patterns in patients with CKD. Am J Kidney Dis 46:455-463. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Calcium-phosphorus product; cardiovascular disease; chronic kidney disease (CKD); mineral metabolism; phosphorus.

BNORMALITIES IN phosphorus metabo- was predominantly male and older, with CKD A lism appear to be an early event in the stage 3 and a high prevalence of heart disease course of chronic kidney disease (CKD), with and diabetes. increases in serum phosphorus levels apparent at creatinine clearances of 50 to 60 mL/min (0.83 to 1.00 mL/s).1 A growing body of evidence impli- cates hyperphosphatemia and elevated calcium- From the Department of Medicine, Division of Nephrol- phosphorus product as contributors to the excess ogy, Tufts-New England Medical Center, Boston, MA; De- cardiovascular disease (CVD) risk in patients partment of Biostatistics and Epidemiology, Cleveland Clinic with kidney failure.2-5 Although the exact mecha- Foundation, Cleveland, OH; The National Institutes of nism underlying this relationship is still unde- Health, Bethesda, MD; and the Division of Nephrology, Hennepin County Medical Center, Minneapolis, MN. fined, potential pathways include increased large- Received April 5, 2005; accepted in revised form May 26, vessel calcification with its associated effects on 2005. arterial stiffening, increased pulse pressure, and Originally published online as doi:10.1053/j.ajkd.2005.05.025 left ventricular hypertrophy.6-9 on July 28, 2005. Supported in part by grants no. 1 K23 DK67303, 1 K23 There are limited data evaluating the relation- DK02904, and UO1 DK 35073 from the National Institute of ships of serum phosphorus levels and calcium- Diabetes and Digestive and Kidney Diseases. phosphorus product with all-cause and CVD Address reprint requests to Mark J. Sarnak, MD, Division mortality in patients with earlier stages of CKD. of Nephrology, Department of Medicine, 750 Washington St, In a recent analysis of a cohort of US veterans NEMC #391, Boston, MA 02111. E-mail: msarnak@tufts- nemc.org with CKD, serum phosphorus levels greater than © 2005 by the National Kidney Foundation, Inc. 3.5 mg/dL (Ͼ1.13 mmol/L) were independent 0272-6386/05/4603-0009$30.00/0 predictors of all-cause mortality.10 This group doi:10.1053/j.ajkd.2005.05.025

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 455-463 455 456 MENON ET AL

Fig 1. Distribution of base- line serum phosphorus levels in the randomized cohort of the MDRD Study. To convert phosphorus in mg/dL to mmol/L, multiply by 0.3229.

The objectives of the present study were to filtration rate (GFR) of 25 to 55 mL/min/1.73 m2 (0.37 to investigate whether serum levels of phosphorus 0.92 mL/s/1.73 m2) were randomized into study A, and 255 2 and calcium-phosphorus product in patients with patients with a baseline GFR of 13 to 24 mL/min/1.73 m (0.22 to 0.40 mL/s/1.73 m2) were randomized into study B. CKD stages 3 to 4 were associated with future Patients in study A and study B were combined for the risk for all-cause or CVD mortality. We evalu- current analyses. Serum phosphorus and calcium levels were ated this hypothesis in the randomized cohort of measured at baseline. Phosphorus was measured at the the Modification of Diet in Renal Disease (MDRD) central laboratory, and calcium was measured at local labora- Study, a group with predominantly nondiabetic tories in each center. kidney disease, precise measurements of kidney Survival status and date and cause of death were ascertained from the National Death Index. A death was function, representation of women and different ascribed to CVD if the primary cause of death was age groups, and a low prevalence of CVD at International Classification of Diseases, Ninth Revision, baseline. codes 390 to 459 (n ϭ 102) or kidney disease was listed as the primary cause of death and CVD (International Classi- METHODS fication of Diseases, Ninth Revision, codes 390 to 459) was the secondary cause (n ϭ 25). Survival time is Study Population defined as time from randomization to death or end of Details of the MDRD Study have been published previ- follow-up (December 31, 2000). The Cleveland Clinic ously.11 In brief, 585 patients with a baseline glomerular (Cleveland, OH) and Tufts-New England Medical Center PHOSPHORUS AND MORTALITY IN CKD 457

Fig 2. Distribution of base- line serum calcium levels in the randomized cohort of the MDRD Study. To convert cal- cium in mg/dL to mmol/L, mul- tiply by 0.2495.

(Boston, MA) Institutional Review Boards approved data adjustment and subsequently adjusting for several covari- collection procedures. ates. The first multivariable model, model 1, included serum phosphorus level or calcium-phosphorus product with age, Statistical Analyses race, sex, and randomization assignments to different blood The distribution and normality of variables of interest pressure goals and protein diets. Model 2 adjusted for were evaluated by using box plots and histograms. Baseline traditional CVD risk factors: smoking status, diabetes, his- characteristics were compared between quartiles of serum tory of coronary artery disease, body mass index, systolic phosphorus levels. Summary statistics for the groups are blood pressure, and low-density lipoprotein and high- presented as percentages for categorical data, mean Ϯ SD density lipoprotein cholesterol levels, in addition to vari- for approximately normally distributed continuous vari- ables in model 1. Model 3 adjusted for variables in model 2 ables, and median and interquartile range for skewed continu- and cause of kidney disease and proteinuria. ous variables. Differences between groups were tested by using chi-square test for categorical variables, 1-way analy- Additional Analyses sis of variance for approximately normally distributed con- Because stratifying by study may not fully adjust for level tinuous variables, and Kruskall-Wallis test for skewed con- of kidney function, we repeated the multivariable Cox model tinuous variables. for phosphorus level and calcium-phosphorus product, adjust- Cox proportional hazards models, stratified by study, were ing for baseline GFR as a continuous variable instead of used to evaluate the relationship of serum phosphorus level stratifying by study (model 4). Because using a single and calcium-phosphorus product, treated as continuous vari- measurement of phosphorus may potentially introduce mea- ables, with all-cause and CVD mortality, initially without surement bias, we repeated this model using an average of 2 458 MENON ET AL

Fig 3. Distribution of base- line serum calcium phospho- rus product levels in the ran- domized cohort of the MDRD Study. serum phosphorus levels measured before the run-in phase dependent diabetes was 5%. Minimum and maxi- and after a 3-month run-in phase and before randomization. mum GFRs were 12 to 55 mL/min/1.73 m2 (0.20 The final Cox model (model 4) for phosphorus level and to 0.92 mL/s/1.73 m2), with a mean GFR of 33 Ϯ all-cause and CVD mortality was repeated after the addition 2 Ϯ 2 of C-reactive protein level. We also evaluated interactions 12 mL/min/1.73 m (0.55 0.20 mL/s/1.73 m ). between serum phosphorus level and study. Seven percent (n ϭ 56) of participants were Hazard ratios (HRs) are presented per 1-mg/dL increase in administered oral calcium at baseline, and 1% serum phosphorus levels and 10-mg2/dL2 increase in calcium- (n ϭ 11) were administered vitamin D (cal- phosphorus product; 95% confidence intervals (CIs) were cifedol, calcitriol [Rocaltrol; Roche Laboratories calculated for HRs. Proportional hazards assumptions were tested by using log minus log survival plots and plots of Inc, Nutley, NJ], and ergocalciferol). Schoenfeld residuals versus survival time. Distribution of Serum Phosphorus Levels RESULTS Figures 1, 2, and 3 show distributions of Mean age of the study cohort was 52 Ϯ 12 serum phosphorus, calcium, and calcium- (SD) years. Sixty-one percent were men, 85% phosphorus product values in the randomized were white, and the prevalence of non–insulin- cohort of the MDRD Study, respectively. Mean PHOSPHORUS AND MORTALITY IN CKD 459

Table 1. Baseline Characteristics According to Serum Phosphorus Levels

Quartile 1 Quartile 2 Quartile 3 Quartile 4 (2.1-3.3 mg/dL; (3.4-3.7 mg/dL; (3.8-4.2 mg/dL; (4.3-6.3 mg/dL; n ϭ 201) n ϭ 209) n ϭ 233) n ϭ 196) P

Serum phosphorus (mg/dL)* 3.0 Ϯ 0.2 3.5 Ϯ 0.1 4.0 Ϯ 0.1 4.7 Ϯ 0.4 — Age (y)* 53.0 Ϯ 11.1 52.6 Ϯ 12.7 51.7 Ϯ 11.5 49.5 Ϯ 13.9 0.02 Male (%) 77 61 58 46 Ͻ0.001 White (%) 92 87 77 86 0.001 Diabetes (%) 43760.24 Current smoker (%) 10 9 10 11 0.95 Cardiovascular risk factors History of coronary artery disease (%) 6 11 10 12 0.12 Body mass index (kg/m2)* 27.6 Ϯ 4.0 27.4 Ϯ 4.5 27.4 Ϯ 4.6 26.1 Ϯ 4.6 0.002 Systolic blood pressure (mm Hg) 131.5 Ϯ 17.2 131.6 Ϯ 16.1 132.0 Ϯ 18.0 132.6 Ϯ 19.1 0.93 Diastolic blood pressure (mm Hg) 82.3 Ϯ 9.8 80.7 Ϯ 9.7 80.4 Ϯ 10.7 80.6 Ϯ 10.1 0.22 Total cholesterol (mg/dL)* 212.4 Ϯ 44.3 219.7 Ϯ 46.6 217.8 Ϯ 45.1 216.8 Ϯ 44.8 0.41 Low-density lipoprotein cholesterol (mg/dL)* 143.0 Ϯ 41.8 150.0 Ϯ 42.3 147.6 Ϯ 40.1 149.0 Ϯ 41.0 0.34 High-density lipoprotein cholesterol (mg/dL)* 37.1 Ϯ 13.2 40.3 Ϯ 14.0 40.3 Ϯ 14.3 41.8 Ϯ 15.1 0.01 C-Reactive protein (mg/dL)† 0.2 (0.5) 0.2 (0.6) 0.3 (0.6) 0.2 (0.5) 0.17 Kidney disease factors Kidney disease category (%) Polycystic kidney disease 21 24 25 25 Glomerular disease 29 28 31 38 0.22 Other 50 48 44 37 Albumin (g/dL)* 4.1 Ϯ 0.3 4.0 Ϯ 0.4 4.0 Ϯ 0.3 4.0 Ϯ 0.4 0.06 Proteinuria (g/d)† 0.15 (0.93) 0.28 (1.1) 0.29 (1.5) 0.67 (2.2) Ͻ0.001 Serum creatinine (mg/dL)* 2.0 Ϯ 0.7 2.1 Ϯ 0.7 2.4 Ϯ 0.9 3.1 Ϯ 1.1 Ͻ0.001 GFR (mL/min/1.73 m2)* 38.9 Ϯ 10.4 35.8 Ϯ 10.1 31.9 Ϯ 11.7 23.2 Ϯ 9.8 Ͻ0.001

NOTE. To convert phosphorus in mg/dL to mmol/L, multiply by 0.3229; total, low-density lipoprotein, and high-density lipoprotein cholesterol in mg/dL to mmol/L, multiply by 0.02586; albumin in g/dL to g/L, multiply by 10; creatinine in mg/dL to ␮mol/L, multiply by 88.4; GFR in mL/min to mL/s, multiply by 0.01667. *Mean Ϯ SD. †Median (interquartile range). serum phosphorus level was 3.8 Ϯ 0.7 mg/dL Serum Phosphorus Levels and Outcomes Ϯ (1.23 0.23 mmol/L), with a range of 2.1 to Median follow-up was 123 months. The all- 6.3 mg/dL (0.68 to 2.03 mmol/L); calcium cause mortality rate was 25% (n ϭ 208), and Ϯ Ϯ level was 9.1 0.5 mg/dL (2.25 0.12 CVD mortality rate was 15% (n ϭ 127). Seventy mmol/L), with a range of 6.4 to 11.9 mg/dL individuals died of any cause, and 38 individuals (1.60 to 2.97 mmol/L); and calcium-phospho- died of CVD before kidney failure. All analyses Ϯ 2 2 rus product was 34.7 6.3 mg /dL , with a included deaths that occurred both before and 2 2 range of 18.7 to 58.6 mg /dL . after kidney failure. One hundred twenty-nine individuals (22%) died of any cause in study A, Baseline Characteristics of Sample by Serum and 79 individuals (31%), in study B. Seventy- Phosphorus Level eight deaths (13%) in study A and 49 deaths (19%) Patients in the highest quartile of serum phos- in study B were caused by CVD. Patients from phorus level were younger and more likely to be study A and B were combined for the present women and nonwhite (Table 1). Higher serum analyses, and Cox models were stratified by study. phosphorus levels were not associated with a Serum phosphorus level was not associated worse CVD risk profile, but were associated with with all-cause mortality in unadjusted analysis, lower GFR and higher proteinuria. but was significantly associated with mortality 460 MENON ET AL

Table 2. Relationship of Serum Phosphorus Level With All-Cause and Cardiovascular Mortality

All-Cause Mortality Cardiovascular Mortality

Hazard Ratio Hazard Ratio (95% confidence interval)* P (95% confidence interval) P

Unadjusted 1.16 (0.96-1.42) 0.13 1.34 (1.05-1.71) 0.02 Adjusted Model 1† 1.37 (1.10-1.72) 0.006 1.62 (1.23-2.13) 0.001 Model 2‡ 1.26 (1.00-1.59) 0.05 1.43 (1.08-1.91) 0.01 Model 3§ 1.18 (0.93-1.50) 0.18 1.33 (0.99-1.79) 0.06 Model 4 1.10 (0.86-1.40) 0.46 1.27 (0.94-1.73) 0.12

*Hazard ratio per 1-mg/dL increase in phosphorus level. †Stratified by study and adjusted for age, sex, race, blood pressure, and protein diet randomization assignment. ‡Adjusted for model 1 plus current smoking, history of coronary artery disease or diabetes, low-density and high-density lipoprotein cholesterol, body mass index, and systolic blood pressure. §Adjusted for model 2 plus proteinuria and cause of kidney disease. With the addition of GFR to model 3 instead of stratifying by study. after adjustment for demographic factors in model there was only a trend toward statistical signifi- 1(Table 2). This finding probably is accounted cance. for by the lower age and fewer men seen in the groups with greater phosphorus levels (Table 1). Calcium-Phosphorus Product and Outcomes Adjustment for CVD risk factors weakened this relationship, which was attenuated further by After adjusting for the younger age and fewer adjustment for kidney disease factors. men in the group with high phosphorus levels, There was a positive association between calcium-phosphorus product was significantly as- serum phosphorus level and CVD mortality in sociated with all-cause mortality; however, this unadjusted analysis, and a 1-mg/dL increase in relationship did not persist after adjustment for serum phosphorus level resulted in a 34% CVD and kidney disease factors (Table 3). There increase in risk for CVD mortality. This asso- was a significant association between calcium- ciation persisted after adjusting for randomiza- phosphorus product and CVD mortality in unad- tion assignments and demographic and CVD justed analysis and after adjustment for demo- risk factors, but after the addition of protein- graphic and CVD factors. However, the addition uria and cause of kidney disease to the model, of kidney disease factors (proteinuria and cause

Table 3. Relationship of Calcium-Phosphorus Product With All-Cause and Cardiovascular Mortality

All-Cause Mortality Cardiovascular Mortality

Hazard Ratio Hazard Ratio (95% confidence interval)* P (95% confidence interval) P

Unadjusted 1.14 (0.92-1.40) 0.23 1.30 (1.01-1.69) 0.04 Adjusted Model 1† 1.28 (1.02-1.62) 0.04 1.53 (1.14-2.03) 0.004 Model 2‡ 1.18 (0.92-1.50) 0.19 1.35 (1.00-1.82) 0.05 Model 3§ 1.13 (0.88-1.44) 0.35 1.26 (0.93-1.71) 0.14 Model 4 1.07 (0.83-1.37) 0.62 1.22 (0.89-1.66) 0.23

*Hazard ratio per 10-mg2/dL2 increase in calcium-phosphorus product. †Stratified by study and adjusted for age, sex, race, blood pressure, and protein diet randomization assignment. ‡Adjusted for model 1 plus current smoking, history of coronary artery disease or diabetes, low-density and high-density lipoprotein cholesterol levels, body mass index, and systolic blood pressure. §Adjusted for model 2 plus proteinuria and cause of kidney disease. With the addition of GFR to model 3 instead of stratifying by study. PHOSPHORUS AND MORTALITY IN CKD 461 of kidney disease) to the model resulted in a loss eral metabolism that leads to hyperphosphatemia, of statistical significance. the treatment of which may lead to increased calcium levels and elevated calcium-phosphorus Additional Analyses product. Model 3 was repeated, adjusting for baseline The kidney is the primary site of phosphorus GFR as a continuous variable rather than stratify- and calcium excretion, and abnormalities of phos- ing by study (model 4; Tables 2 and 3). This phorus and calcium homeostasis are apparent in further attenuated the association of serum phos- the early stages of kidney disease.1 Several stud- phorus level and calcium-phosphorus product ies implicated divalent ion abnormalities as a with all-cause and CVD mortality. potential mechanism in the pathophysiology un- The Cox model with adjustment for GFR was derlying the excess CVD seen in patients with repeated using an average of 2 baseline measure- kidney failure.9 Data from the US Renal Data ments of serum phosphorus. Mean phosphorus System showed an increased risk for mortality level using 2 baseline measurements was 3.8 Ϯ with serum phosphorus level greater than 6.5 0.6 mg/dL (1.23 Ϯ 0.19 mmol/L). HRs did not mg/dL (Ͼ2.10 mmol/L) and calcium-phospho- differ appreciably from analyses using a single rus product greater than 72 mg2/dL2.2 In a large baseline measurement (all-cause mortality HR, cohort of maintenance dialysis patients, there 1.19; 95% CI, 0.90 to 1.58; P ϭ 0.23; CVD was a significant association of all-cause mortal- mortality HR, 1.34; 95% CI, 0.93 to 1.92; P ϭ ity with greater serum phosphorus concentra- 0.12). tions.3 More recently, using data from 2 large Model 4 for serum phosphorus level and all- national databases of long-term hemodialysis cause and CVD mortality was repeated after the patients, Ganesh et al4 showed that serum phos- addition of C-reactive protein level. HRs for phorus level greater than 6.5 mg/dL (Ͼ2.10 all-cause and CVD mortality for serum phospho- mmol/L) and elevated calcium-phosphorus prod- rus level were unchanged by the addition of uct were independent risk factors for CVD mor- C-reactive protein level (data not shown). Fi- tality. Serum phosphorus level greater than 6.5 nally, in models examining all-cause (P ϭ 0.29) mg/dL (Ͼ2.10 mmol/L) and calcium-phosphate and CVD (P ϭ 0.57) mortality, interaction terms product greater than 52 mg2/dL2 also were asso- between study and serum phosphorus level were ciated with increased risk for CVD mortality in a not significant. cohort of European hemodialysis patients.18 The cause of the excess CVD risk associated DISCUSSION with high phosphorus levels remains unclear, In the MDRD Study cohort, neither serum although growing evidence suggests a role for phosphorus level nor calcium-phosphorus prod- vascular calcification as an intermediary in this uct were independently related to all-cause mor- pathway. In prevalent hemodialysis patients, high tality. There was a trend towards statistical signifi- serum calcium and phosphorus concentrations cance in the associations of serum phosphorus appear to be associated with more extensive level and calcium-phosphorus product with CVD aortic and coronary calcification.8,19 Elevations mortality. in phosphorus, calcium, and calcium-phosphorus CVD accounts for more than half the mortality product values induced by therapy with calcium- in patients with kidney failure and is a major containing phosphorus binders in hemodialysis cause of morbidity and mortality in patients in patients are thought to be associated with acceler- the earlier stages of kidney disease.12-15 The ated vascular calcification.6 In vitro data showed excess risk for CVD in patients with CKD is only induction of vascular smooth muscle cell calcifi- partly attributable to a high prevalence of tradi- cation by elevated calcium and phosphorus lev- tional risk factors, such as diabetes, hyperten- els.20 In a rat model of kidney failure, hyperphos- sion, and hyperlipidemia.16,17 Thus, recent inves- phatemia induced by a high-phosphorus diet tigations have focused on identifying additional produced cardiac fibrosis and arterial wall thick- uremia-related factors that may contribute to the ening.21 The vascular calcification resulting from increased risk for CVD in patients with CKD. disordered mineral metabolism in patients with One such putative risk factor is disordered min- kidney disease can have several adverse CVD 462 MENON ET AL effects, including arterial stiffening, increased m2 (0.53 mL/s/1.73 m2). No information is pro- pulse pressure, decreased coronary perfusion, vided on level of proteinuria in the veterans and left ventricular hypertrophy.22,23 cohort, and proteinuria was not included in the Despite the abundance of data in dialysis pa- multivariable Cox models. The veterans cohort tients, there are few studies to date examining the were predominantly men (97%), older (mean relationship between abnormalities of mineral age, 70 years), and had greater rates of prevalent metabolism and CVD risk in patients in the ischemic heart disease (32%) and history of earlier stages of kidney disease. In a recent study diabetes (23%) compared with the MDRD Study of a cohort of US veterans with CKD, elevated cohort. Thus, it is possible that the significant serum phosphorus levels were independent pre- relationship seen between serum phosphorus level dictors of risk for mortality and a composite end and all-cause mortality in this cohort of older point of death and myocardial infarction.10 men with a history of CVD is driven mainly by In the MDRD Study cohort of patients with CVD mortality. In support of this hypothesis, the CKD stages 3 to 4, the relationship between HR for our adjusted model (model 3; Table 2) for serum phosphorus level and all-cause mortality CVD mortality (1.33) was identical to that re- was attenuated by adjustment for kidney disease ported by Kestenbaum et al10 for all-cause mor- factors, including proteinuria, cause of kidney tality. disease, and GFR. In unadjusted analysis, serum To our knowledge, the association between phosphorus level was associated with an in- calcium-phosphorus product and mortality has creased risk for CVD mortality. The strength of not been assessed in patients with CKD before this association was attenuated by adjustment for the development of kidney failure. In the MDRD kidney disease factors. We acknowledge that one Study cohort of patients with CKD stages 3 to 4, explanation for the lack of a definitive result may calcium-phosphorus product was associated with be limited statistical power. The relatively wide CVD mortality, but not all-cause mortality, in CIs support this contention. However, as sug- unadjusted analysis. After adjustment for covari- gested by Chertow and Moe,24 it also is possible ates, calcium-phosphorus product was no longer that the relationship between serum phosphorus statistically significant as an independent predic- level and mortality is reflective of the severity of tor of CVD mortality. It is possible that this lack CKD. In the study by Kestenbaum et al,10 esti- of significance reflects inadequate statistical mated creatinine clearance was used as a mea- power, rather than a true negative result. Alter- sure of kidney function, and this may have led to nately, the effect of calcium-phosphorus product residual confounding. In the MDRD Study, use on mortality may reflect the severity of CKD and of iothalamate GFR, a more precise marker of therefore be accounted for by level of kidney kidney function, as well as additional adjustment function. for proteinuria, may have better accounted for A major strength of this study is the inclusion severity of kidney disease. However, the point by of a large cohort of patients with precise measure- Chertow and Moe24 is well taken because the HR ment of GFR and detailed ascertainment of such for CVD mortality decreased from 1.43 to 1.27 confounding variables as proteinuria, body mass with adjustment for GFR, proteinuria, and cause index, and blood pressure. Findings of this study of kidney disease. are generalizable to relatively healthy, predomi- Any potential differences in results between nantly white patients with CKD without diabe- our study (HR, 1.18; 95% CI, 0.93 to 1.50) and tes. The low prevalence of diabetes, malnutri- those reported by Kestenbaum et al10 (HR, 1.33; tion, and CVD decreases the confounding effect 95% CI, 1.15 to 1.54) with regard to all-cause of these comorbid conditions on the relationship mortality may be attributable to differences in of serum levels of phosphorus and calcium- study populations in demographic characteristics phosphorus product with outcomes. The primary and severity of kidney disease. In the latter study, limitations are the possibility of restricted statis- 68% of participants had CKD stage 3, and me- tical power and the narrow range of serum phos- dian estimated creatinine clearance was 45.1 phorus and serum calcium-phosphorus product mL/min (0.75 mL/s), whereas median GFR in values in the MDRD Study cohort. We also the MDRD Study cohort was 32 mL/min/1.73 acknowledge that compared with the MDRD PHOSPHORUS AND MORTALITY IN CKD 463

Study era, practice patterns may have changed, 10. Kestenbaum B, Sampson JN, Rudser KD, et al: with increased use of calcium and vitamin D in Serum phosphate levels and mortality risk among people patients in the earlier stages of CKD. We were with chronic kidney disease. J Am Soc Nephrol 16:520-528, 2005 unable to examine any potential modifying effect 11. Greene T, Bourgoignie JJ, Habwe V, et al: Baseline of parathyroid hormone levels on the relation- characteristics in the Modification of Diet in Renal Disease ships studied because parathyroid hormone lev- Study. J Am Soc Nephrol 4:1221-1236, 1993 els were not measured in the MDRD Study. 12. Anavekar NS, McMurray JJ, Velazquez EJ, et al: In conclusion, levels of serum phosphorus and Relation between renal dysfunction and cardiovascular out- calcium-phosphorus product measured in pa- comes after myocardial infarction. N Engl J Med 351:1285- 1295, 2004 tients with CKD stages 3 to 4 may be predictive 13. Fried LF, Shlipak MG, Crump C, et al: Renal insuffi- of CVD mortality; however, neither was related ciency as a predictor of cardiovascular outcomes and mortal- to all-cause mortality. Larger studies with more ity in elderly individuals. J Am Coll Cardiol 41:1364-1372, statistical power are necessary to further explore 2003 these results, especially in the context of current 14. Manjunath G, Tighiouart H, Ibrahim H, et al: Level of kidney function as a risk factor for atherosclerotic cardiovas- practice patterns in the management of mineral cular outcomes in the community. J Am Coll Cardiol 41:47- metabolism in patients with CKD, and to deter- 55, 2003 mine whether interventions to normalize serum 15. Levey AS, Beto JA, Coronado BE, et al: Controlling phosphorus levels in this population lead to im- the epidemic of cardiovascular disease in chronic renal proved CVD outcomes. disease: What do we know? What do we need to learn? Where do we go from here? Special report from the National REFERENCES Kidney Foundation Task Force on Cardiovascular Disease. Am J Kidney Dis 32:853-906, 1998 1. Hsu CY, Chertow GM: Elevations of serum phospho- 16. Sarnak MJ, Coronado BE, Greene T, et al: Cardiovas- rus and potassium in mild to moderate chronic renal insuffi- cular disease risk factors in chronic renal insufficiency. Clin ciency. Nephrol Dial Transplant 17:1419-1425, 2002 Nephrol 57:327-335, 2002 2. Block GA, Hulbert-Shearon TE, Levin NW, Port FK: 17. Kasiske BL, Chakkera HA, Roel J: Explained and Association of serum phosphorus and calcium ϫ phosphate unexplained ischemic heart disease risk after renal transplan- product with mortality risk in chronic hemodialysis patients: tation. J Am Soc Nephrol 11:1735-1743, 2000 A national study. Am J Kidney Dis 31:607-617, 1998 18. Marco MP, Craver L, Betriu A, et al: Higher impact of 3. Block GA, Klassen PS, Lazarus JM, et al: Mineral metabolism, mortality, and morbidity in maintenance hemo- mineral metabolism on cardiovascular mortality in a Euro- dialysis. J Am Soc Nephrol 15:2208-2218, 2004 pean hemodialysis population. Kidney Int Suppl 85:S111- 4. Ganesh SK, Stack AG, Levin NW, Hulbert-Shearon T, S114, 2003 Port FK: Association of elevated serum PO(4), Ca ϫ PO(4) 19. Goodman WG, Goldin J, Kuizon BD, et al: Coronary- product, and parathyroid hormone with cardiac mortality artery calcification in young adults with end-stage renal risk in chronic hemodialysis patients. J Am Soc Nephrol disease who are undergoing dialysis. N Engl J Med 342:1478- 12:2131-2138, 2001 1483, 2000 5. Stevens LA, Djurdjev O, Cardew S, Cameron EC, 20. Reynolds JL, Joannides AJ, Skepper JN, et al: Human Levin A: Calcium, phosphate, and parathyroid hormone vascular smooth muscle cells undergo vesicle-mediated cal- levels in combination and as a function of dialysis duration cification in response to changes in extracellular calcium and predict mortality: Evidence for the complexity of the associa- phosphate concentrations: A potential mechanism for accel- tion between mineral metabolism and outcomes. J Am Soc erated vascular calcification in ESRD. J Am Soc Nephrol Nephrol 15:770-779, 2004 15:2857-2867, 2004 6. Chertow GM, Raggi P, Chasan-Taber S, et al: Determi- 21. Amann K, Tornig J, Kugel B, et al: Hyperphos- nants of progressive vascular calcification in haemodialysis phatemia aggravates cardiac fibrosis and microvascular dis- patients. Nephrol Dial Transplant 19:1489-1496, 2004 ease in experimental uremia. Kidney Int 63:1296-1301, 7. Yildiz A, Tepe S, Oflaz H, et al: Carotid atherosclerosis 2003 is a predictor of coronary calcification in chronic haemodi- 22. London GM: Cardiovascular calcifications in uremic alysis patients. Nephrol Dial Transplant 19:885-891, 2004 patients: Clinical impact on cardiovascular function. J Am 8. Raggi P, Boulay A, Chasan-Taber S, et al: Cardiac Soc Nephrol 14:S305-S309, 2003 (suppl 4) calcification in adult hemodialysis patients. A link between 23. London GM, Guerin AP, Marchais SJ, et al: Arterial end-stage renal disease and cardiovascular disease? J Am media calcification in end-stage renal disease: Impact on Coll Cardiol 39:695-701, 2002 all-cause and cardiovascular mortality. Nephrol Dial Trans- 9. Goodman WG: Importance of hyperphosphataemia in plant 18:1731-1740, 2003 the cardio-renal axis. Nephrol Dial Transplant 19:S4-S8, 24. Chertow GM, Moe SM: Calcification or classifica- 2004 (suppl 1) tion? J Am Soc Nephrol 16:293-295, 2005 Dialysis Therapies Changes in Cardiovascular Calcification After Parathyroidectomy in Patients With ESRD

Anthony J. Bleyer, MD, John Burkart, MD, Michael Piazza, MD, Gregory Russell, MS, Michael Rohr, MD, and J. Jeffrey Carr, MD

● Background: The effect of parathyroidectomy on vascular calcification in patients with end-stage renal disease has been a subject of interest for many years, although studies in this area have not been definitive. The purpose of this investigation is to determine changes in vascular calcification after subtotal parathyroidectomy by using fast-gated helical computed axial tomographic imaging to measure coronary and carotid artery calcification. Methods: Computed tomographic imaging was performed at baseline and in follow-up on 10 patients who had undergone subtotal parathyroidectomy and 10 reference patients who had not undergone parathyroidectomy. ؎ Results: Patients who underwent subtotal parathyroidectomy had a mean change in coronary calcification of ؊92.3 The 2 parathyroidectomy patients with .(0.03 ؍ y, and reference patients had a mean change of ؉479 ؎ 630/y (P/469 the highest baseline scores had significant declines in both coronary and carotid calcification. Conclusion: In this study, subtotal parathyroidectomy is associated with a significant decrease in vascular calcification in 2 of 10 dialysis patients with high coronary artery calcium scores and stabilization in 7 of 10 patients with low baseline scores. Am J Kidney Dis 46:464-469. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Vascular calcification; coronary; carotid; end-stage renal disease (ESRD); hemodialysis (HD); peritoneal dialysis (PD); cardiovascular.

HE EFFECT OF parathyroidectomy on vas- mately 42% of dialysis patient deaths in the US T cular calcification in patients with end- Renal Data System recorded as cardiovascular in stage renal disease has been a subject of interest origin.6 Therapeutic interventions that stabilize for many years,1-3 although studies in this area or potentially reverse calcification may be of have not been definitive. This issue is of increas- great value to patients with end-stage renal dis- ing importance with the identification of cardio- ease with vascular disease. vascular calcification as a risk factor for cardiac Prior studies investigating the relationship be- death.4 Hemodialysis patients have a very high tween vascular calcification and parathyroidec- prevalence of vascular calcification; 83% of pa- tomy were limited by the use of radiological tients had coronary calcification in a cross- techniques that could only grossly quantify calci- 1,3 sectional study of 205 hemodialysis patients.5 In fication. Recent advances in computed tomo- addition, dialysis patients disproportionately ex- graphic (CT) scanning of coronary arteries now perience cardiovascular death, with approxi- allow us to more accurately quantitate vascular calcium content.4 In the present study, we report on a group of patients who underwent subtotal From the Department of Internal Medicine, Section on parathyroidectomy and gated CT scanning of the Nephrology; Department of Public Health Sciences, Section coronary and carotid arteries before and after on Biostatistics; and Departments of General Surgery and subtotal parathyroidectomy. Radiology; Wake Forest University School of Medicine, Winston-Salem, NC. Received November 8, 2004; accepted in revised form METHODS April 27, 2005. This study was approved by the Wake Forest University Originally published online as doi:10.1053/j.ajkd.2005.04.035 School of Medicine (Winston Salem, NC) Institutional Re- on July 12, 2005. view Board. Supported in part by a grant from the National Kidney Foundation of North Carolina, Inc. Population Address reprint requests to Anthony J. Bleyer, MD, Sec- tion on Nephrology, Department of Internal Medicine, Wake Individuals undergoing peritoneal dialysis or hemodialy- Forest University School of Medicine, Winston-Salem, NC sis at dialysis centers in the southeastern United States 27157. Email: [email protected] operated by Wake Forest University School of Medicine © 2005 by the National Kidney Foundation, Inc. who were to undergo subtotal parathyroidectomy for el- 0272-6386/05/4603-0010$30.00/0 evated intact parathyroid hormone (iPTH) levels refractory doi:10.1053/j.ajkd.2005.04.035 to medical therapy were invited to participate in the study. A

464 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 464-469 VASCULAR CALCIFICATION AFTER PARATHYROIDECTOMY 465

Table 1. Baseline Characteristics of Subtotal Parathyroidectomy and Reference Patients

Characteristic Parathyroidectomy (n ϭ 10) Reference (n ϭ 10) P

Age (y) 44.8 Ϯ 11.4 54.4 Ϯ 16.8 0.15 Race (% white) 2 (20.0) 6 (60.0) 0.17 Sex (% male) 5 (50.0) 4 (40.0) 0.99 Vintage (y on hemodialysis) 6.7 Ϯ 5.2 4.5 Ϯ 3.8 0.28 Modality (% hemodialysis) 9 (90.0) 7 (70.0) 0.58 Baseline iPTH (pg/mL) 926.2 Ϯ 628.5 151.2 Ϯ 80.8 0.0036 Baseline adjusted serum calcium (mg/dL) 9.0 Ϯ 1.2 10.1 Ϯ 1.5 0.064 Baseline serum phosphorus (mg/dL) 6.7 Ϯ 1.6 4.8 Ϯ 1.4 0.024 Baseline serum calcium ϫ phosphorus product (mg2/dL2) 56.7 Ϯ 17.0 49.2 Ϯ 17.0 0.33

NOTE. To convert iPTH in pg/mL to ng/L, multiply by 1; serum calcium in mg/dL to mmol/L, multiply by 0.2495; serum phosphorus in mg/dL to mmol/L, multiply by 0.3229; calcium phosphorus product in mg2/dL2 to mmol2/L2, multiply by 0.0808. reference group of dialysis patients who had not undergone Statistical Analysis parathyroidectomy also was included. Descriptive statistics, including means and SDs for con- Coronary artery calcification (CAC) was measured by tinuous data and frequencies and proportions for categorical using fast-gated helical CT as previously described.7 Scans measures, were generated for both study groups. Indepen- were performed on 2 single-slice subsecond helical CT dent t-tests were used to test for differences in continuous scanners equipped for retrospective cardiac gating and ca- data, and for dichotomous data, chi-squared tests were used pable of 500-millisecond temporal resolution (HiSpeed LX to assess differences between groups. In comparing coro- with the SmartScore Cardiac scan package; General Electric nary calcification, several methods were used. First, the Medical Systems, Milwaukee, WI). Scanning of the entire difference between the modified Agatston score at baseline heart and carotid arteries was performed during suspended and follow-up was determined for each patient and adjusted respiration at end-inspiration. The scan was obtained at 50% for the time between scans. t-Test comparison of adjusted of the RR interval. Scanning of cardiac vessels was per- Agatston scores then was performed. Square-root transforma- formed twice, and results were averaged. CAC score was tion also was performed and compared between the 2 groups. determined by using a modified Agatston method with the A volumetric score also was obtained and compared be- traditional 130-Henry U threshold and a minimum lesion tween the 2 groups, as was square-root transformation of the definition of 0.52 mm2. This method has an extremely high volumetric score. For carotid artery scores, a similar method correlation with electron-beam CT–derived CAC scores (r ϭ was used with comparison of baseline and follow-up scans 0.98) based on a previous investigation we performed.8 adjusted for time, using both scores and square-root transfor- Carotid scans were performed measuring calcified plaque mation of the scores. burden in the common carotid, internal carotid, external Variability between the modified Agatston scores was carotid, and bulb. These measurements were averaged for determined by obtaining the difference between the 2 CAC the left and right sides.9 scores obtained and dividing by mean CAC score. Baseline fast-gated helical CT scans were performed at the time of subtotal parathyroidectomy in 10 individuals. Follow-up scans were performed during a 2-year period. RESULTS Timing of follow-up scans was determined by the desire for Table 1 lists baseline demographic data for the increased length of follow-up. Scans were performed in a similar manner for the reference group. Mean carotid score 2 groups. Parathyroidectomy patients were more was the average of the right and left carotid scores. likely to have elevated iPTH levels and greater Medical records and dialysis records for each patient were serum phosphorus values. reviewed to determine demographics, phosphate-binder use, Table 2 lists time-averaged laboratory values, and monthly laboratory values between scans. Calcium binder time-averaged phosphate-binder dosage (in grams dosage was converted to the amount of elemental calcium for each preparation. Vitamin D dosage was converted to the of elemental calcium per day), and time-aver- equivalent amount of calcitriol, assuming a 3:1 conversion for aged calcitriol-equivalent intake (in micrograms doxercalciferol or paricalcitol to calcitriol. Chemistry values per week). Patients who underwent subtotal para- and medication dosages were time-averaged over the duration thyroidectomy had lower serum calcium and between the 2 scans. higher serum phosphorus values, whereas calcit- PTH was measured as iPTH by using a solid-phase 2-site chemiluminescent enzyme-labeled immunometric assay on the riol-equivalent dosages and oral calcium dosages Immulite 2000 (Diagnostic Products Corp, Los Angeles, CA). were much greater. Of interest, iPTH levels were The reference range for this assay is 11 to 67 pg/mL (ng/L). similar in follow-up between the 2 groups. 466 BLEYER ET AL

Table 2. Characteristics of Patients During the Interval Between Scans

Characteristic Parathyroidectomy (n ϭ 10) Reference (n ϭ 10) P

Time between scans (y) 1.66 Ϯ 0.58 2.14 Ϯ 0.35 0.009 Time-averaged calcitriol-equivalent dosage (␮g/wk) 8.2 Ϯ 5.5 2.3 Ϯ 1.4 0.007 Time-averaged calcium intake (g/d) 2.0 Ϯ 1.1 0.91 Ϯ 0.61 0.01 Time-averaged adjusted serum calcium (mg/dL) 8.3 Ϯ 0.8 9.9 Ϯ 0.33 0.004 Time-averaged serum phosphorus (mg/dL) 6.6 Ϯ 1.1 5.4 Ϯ 1.0 0.014 Time-averaged iPTH (pg/mL) 197 Ϯ 144 232 Ϯ 155 0.66 Time-averaged serum calcium ϫ phosphorus product (mg2/dL2) 53.7 Ϯ 9.3 53.6 Ϯ 10.7 0.98

NOTE. To convert iPTH in pg/mL to ng/L, multiply by 1; serum calcium in mg/dL to mmol/L, multiply by 0.2495; serum phosphorus in mg/dL to mmol/L, multiply by 0.3229; calcium phosphorus product in mg2/dL2 to mmol2/L2, multiply by 0.0808.

Mean changes in CAC scores were Ϫ92.3 Ϯ 66.9 in reference patients (P ϭ 0.055). With 469 per year in patients undergoing parathyroid- reciprocal transformation, P was not statistically ectomy and ϩ479 Ϯ 630 per year in reference significant (P ϭ 0.17). patients (P ϭ 0.03). Table 3 lists results for Variability between CAC scores was 13.1% Ϯ individual patients. Square-root transformation 19.6% for all 20 scans combined. also was performed, and t-test of transformed Figure 1 shows changes in CAC scores accord- scores again was statistically significant (P ϭ ing to baseline scores. Despite having uncon- 0.019). Mean changes per year in volumetric trolled hyperparathyroidism, a number of pa- data for coronary scores were 922.35 Ϯ 1,123 for tients undergoing subtotal parathyroidectomy had controls and Ϫ431.5 Ϯ 978.6 for parathyroidec- low baseline calcification scores. Patients under- tomy patients (P ϭ 0.01). With reciprocal trans- going parathyroidectomy had stabilization of formation of volumetric scores, changes re- CAC scores, with the exception of 1 patient, for mained statistically significant (P ϭ 0.019). whom the score increased from 885 to 1,767. Mean changes in carotid scores were Ϫ15.3 Ϯ The 2 parathyroidectomy patients with the high- 78.3 in parathyroidectomy patients and ϩ51.7 Ϯ est CAC scores had decreases in their CAC

Table 3. Characteristics of Individual Patients

Patient Baseline CAC Final CAC Change in Baseline Carotid Final Carotid Change in Carotid No. Parathyroidectomy Score Score CAC/y Score Score Score/y

1 Yes 0 0 0 0 0 0 2 Yes 7.0 0 Ϫ6.8 3.8 7.5 3.7 3 Yes 38.5 141 75.0 5.3 89 61.5 4 Yes 158 186 15.7 396 445 27.7 5 Yes 244 265 8.25 143 174 11.6 6 Yes 495 902 168 45 65 8.0 7 Yes 886 1,767 549 1,294 928 Ϫ228 8 Yes 1,810 1,871 28.6 16 0 Ϫ7.3 9 Yes 2,763 1,710 Ϫ1,180 151 131 Ϫ22.4 10 Yes 2,845 2,004 Ϫ581 170.3 160 Ϫ7.1 Reference Patients 1 No 0 0 0 0 6.75 2.78 2No0 00 0 0 0 3 No 0 22 9.4 0 0 0 4 No 126 227 60.8 130 183 32 5 No 852 1,060 128 481 658 109 6 No 1,011 2,640 678 14 35 8.9 7 No 1,081 2,542 607 117 122 2.1 8 No 1,939 2,279 207 459 523 39 9 No 2,680 5,545 1,261 428 786 157 10 No 4,241 8,640 1,838 582 982 167 VASCULAR CALCIFICATION AFTER PARATHYROIDECTOMY 467

Fig 1. Changes in mean coronary scores according to baseline scores. (ࡗ), PTX; (□), Reference. scores. These patients, both white, also had simi- patient who underwent scanning at 1 and 2 years lar decreases in their carotid artery scores, de- after parathyroidectomy, CAC scores were 61.5 creasing from 150.5 to 130.5 and 170.3 to 160. and 95.5, respectively. Time-averaged serum calcium levels after para- Conversely, there was a general increase in thyroidectomy in these patients were 9.57 mg/dL cardiovascular calcification in the reference group (2.40 mmol/L) and 8.7 mg/dL (2.18 mmol/L). (Fig 1). Individuals with the highest scores had Time-averaged serum phosphorus levels were the greatest increases in scores over time. 5.44 mg/dL (1.76 mmol/L) and 8.0 mg/dL (2.58 Figure 2 shows yearly change in carotid calci- mmol/L), and time-averaged iPTH values were fication scores according to baseline scores. Ref- 178.8 and 373.5 pg/mL (ng/L). Average calcium erence patients with elevated baseline scores intakes over time for these 2 patients were 1,837 showed larger increases in scores than patients and 1,211 mg/d. These patients were adminis- who had undergone subtotal parathyroidectomy. tered an average of 9.7 and 8.4 ␮g/wk of calcit- riol-equivalent replacement. In 1 parathyroidec- DISCUSSION tomy patient with a baseline score of 1,810, there The present study shows a significant decline was essential stabilization of CAC scores, with in cardiovascular calcification in several patients an increase of only 28 per year (1.5%). In the with high baseline CAC scores and stabilization

Fig 2. Changes in carotid scores ac- cording to baseline scores. (ࡗ), PTX; (□), Reference. 468 BLEYER ET AL in 1 patient with a high baseline score. These arterial calcifications in 56% of patients. Meema results were in contrast to those found in a et al1 also noted an increase in cardiac calcifica- reference group of 10 other dialysis patients, for tion after parathyroidectomy in 4 of 6 patients by whom CAC scores increased over time in patients using skeletal x-rays. DiLeo et al15 noted a with high baseline scores. Other studies noted an decrease in cardiac and pulmonary calcification approximate annual 25% increase in CAC scores in 4 years after parathyroidectomy. We believe the patients treated with calcium-based binders,10 which prior studies were limited because of the inabil- was significantly greater than the change in cardio- ity to quantitate vascular calcification through vascular scores noted in our population who under- such techniques as CT scanning. went subtotal parathyroidectomy. These results Despite the presence of severe hyperparathy- suggest that subtotal parathyroidectomy may be roidism, a number of patients did not have in- beneficial in preventing progression of cardiovas- creased CAC scores at baseline. Others noted cular disease in patients with elevated iPTH that PTH level is not indicative of the level of levels who have significant cardiovascular calci- vascular calcification.16 Subtotal parathyroidec- fication, although additional studies need to be tomy may have prevented progression in these performed. Patients with end-stage renal disease individuals with low vascular calcium scores, who have undergone amputation11 or those who although it appears that some individuals are have experienced myocardial infarction12 have “immune” from cardiac calcification.10 Previous been noted to have poor survival. Subtotal para- studies noted a genetic component to cardiovas- thyroidectomy potentially could benefit such in- cular calcification, which may be of importance.7 dividuals who have high iPTH levels. In a cross- A weakness of this study is the small num- sectional study using ultrasound, London et al13 ber of patients participating. This is caused in noted a correlation between high arterial calcifi- large part by decreased indications for parathy- cation scores and low iPTH levels and specu- roidectomy given newer therapies to decrease lated that parathyroidectomy could lead to wors- iPTH levels.17 However, although these thera- ening of cardiovascular calcification. Results from pies may decrease iPTH levels, it is unclear our longitudinal study suggest that vascular cal- what effect they may have on vascular calcifi- cification actually decreased with subtotal para- cation. Additional studies need to be per- thyroidectomy. That patients underwent subtotal formed to see whether these patients would parathyroidectomy and iPTH levels were not benefit more from parathyroidectomy, treat- severely depressed postoperatively may be par- ment with calcimimetic agents, or treatment tially responsible for the beneficial effect. Previ- with vitamin D or its analogues. ous studies documented the benefit of parathy- Another weakness of this study is the lack of roidectomy in patients with calciphylaxis,14 in an appropriate control group. The appropriate which a similar decline in vascular calcium con- control group would have included individuals tent may occur. who required parathyroidectomy, but did not Patients undergoing parathyroidectomy had undergo this procedure. The only purpose the greater baseline iPTH levels than the reference reference group in this study serves is to show group of patients because they were chosen for that cardiovascular progression shown by CT subtotal parathyroidectomy based on elevated scanning occurs in our hemodialysis population iPTH levels. Patients in the reference group were at rates similar to those reported in other stud- selected randomly and therefore did not have ies.10 The groups were different with regards to markedly elevated iPTH levels. Patients who baseline characteristics and length of follow-up. underwent subtotal parathyroidectomy had a re- The current study showed stabilization in 7 of maining remnant of parathyroid tissue. Between 10 patients and reversal of CAC in 2 of 10 scans, over time, their iPTH levels increased patients undergoing subtotal parathyroidectomy. slowly, such that their iPTH levels were similar Patients with the highest CAC scores had the in postoperative follow-up to the reference group. most benefit. Subtotal parathyroidectomy in ap- Prior studies also looked at changes in vascu- propriate patients may prevent or reverse progres- lar calcium content by using skeletal x-rays. De sion of vascular calcification. Additional studies Francisco et al3 noted an increase in peripheral should be performed to confirm these findings. VASCULAR CALCIFICATION AFTER PARATHYROIDECTOMY 469

REFERENCES calcification and carotid intimal medial thickness. Stroke 1. Meema H, Oreopoulos D, deVeber G: Arterial calcifi- 35:E97-E99, 2004 cations in severe chronic renal disease and their relationship 10. Chertow G, Burke S, Raggi P: Sevelamer attenuates to dialysis treatment, renal transplant, and parathyroidec- the progression of coronary and aortic calcification in hemo- tomy. Radiology 121:315-321, 1976 dialysis patients. Kidney Int 62:245-252, 2002 11. O’Hare A, Feinglass J, Reiber G, et al: Postoperative 2. Parfitt A: Soft-tissue calcification in uremia. Arch mortality after nontraumatic lower extremity amputation in Intern Med 124:544-556, 2004 patients with renal insufficiency. J Am Soc Nephrol 15:427- 3. DeFrancisco A, Ellis H, Owen J, et al: Parathyroidec- 434, 2004 tomy in chronic renal failure. Q J Med 218:289-315, 1985 12. Herzog C, Ma J, Collins A: Poor long-term survival 4. Wexler L, Brundage B, Crouse J, et al: Coronary artery after acute myocardial infarction among patients on long- calcification: Pathophysiology, epidemiology, imaging meth- term dialysis. N Engl J Med 339:799-805, 1998 ods, and clinical implications. Circulation 94:1175-1192, 13. London GM, Marty C, Marchais SJ, Guerin AP, 2004 Metivier F, de Vernejoul MC: Arterial calcifications and 5. Raggi P, Boulay A, Chasan-Taber S, et al: Cardiac bone histomorphometry in end-stage renal disease. J Am calcification in adult hemodialysis patients. J Am Coll Car- Soc Nephrol 15:1943-1951, 2004 diol 39:695-701, 2005 14. Girotto J, Harmon J, Ratner L, Nicol T, Wong L, 6. Bleyer AJ, Russell GB, Satko SG: Sudden and cardiac Chen H: Parathyroidectomy promotes wound healing and death rates in hemodialysis patients. Kidney Int 55:1553- prolongs survival in patients with calciphylaxis from second- 1559, 1999 ary hyperparathyroidism. Surgery 130:645-651, 2001 7. Wagenknecht L, Bowden D, Carr J, Langefeld C, 15. DiLeo C, Gallieni M, Bestetti A, et al: Cardiac and Freedman B, Rich S: Familial aggregation of coronary pulmonary calcification in a hemodialysis patient: Partial artery calcium in families with type 2 diabetes. Diabetes regression 4 years after parathyroidectomy. Clin Nephrol 50:861-866, 2001 59:59-63, 2003 8. Carr J, Crouse J, Goff D, D’Agostino R Jr, Peterson M, 16. Goodman W, Goldin J, Kuizon B, et al: Coronary Burke G: Evaluation of subsecond gated helical CT for artery calcification in young adults with end-stage renal quantification of coronary artery calcium and comparison disease who are undergoing dialysis. N Engl J Med 342:1478- with electron beam CT. AJR Am J Roentgenol 174:915-921, 1482, 2000 2000 17. Kestenbaum B, Seliger S, Gillen D, et al: Parathyroid- 9. Wagenknecht L, Langefeld C, Carr J, et al: Race- ectomy rates among United States dialysis patients: 1990- specific relationships between coronary and carotid artery 1999. Kidney Int 65:282-288, 2004 The Hemodynamic Effect of Calcium Ion Concentration in the Infusate During Predilution Hemofiltration in Chronic Renal Failure

Nikolaos Karamperis, MD, Erik Sloth, MD, PhD, DMSc, and Jens Dam Jensen, MD, PhD

● Background: It is the prevailing view that convective dialysis techniques stabilize blood pressure. Calcium concentration in the substitution fluid may be important in this respect. The aim of this study is to investigate the influence of calcium ion concentration in the substitution fluid on hemodynamic stability during predilution hemofiltration (HF). Methods: We conducted a randomized, crossover, blinded, controlled trial with 12 stable long-term hemodialysis patients without diabetes. Each patient was randomly assigned to substitution fluid with a calcium ion (iCa) concentration of 2.5 mEq/L (1.25 mmol/L; low-calcium session [L-HF]) or 3.5 mEq/L (1.75 mmol/L; high-calcium session [H-HF]) during 4.5 hours of predilution HF with a volume of 1.24 ؎ 0.09 L/kg dry body weight and a temperature of 37°C. Ultrafiltration was kept constant in each patient. Blood pressure (mean, systolic [SBP], and diastolic blood pressure [DBP]), pulse rate, arterial and venous temperature, energy transfer, and relative blood volume were measured at 15-minute intervals. Cardiac output, total peripheral resistance, stroke volume, and iCa were measured hourly. The 2 treatments were matched with the exception of iCa concentration. Results: A significant intratreatment reduction in cardiac output and stroke volume was shown to the same extent for both groups. Intertreatment comparisons showed a significantly lower mean arterial pressure, SBP, DBP, and total peripheral resistance in the L-HF compared with the H-HF group. Conclusion: iCa concentration of 3.5 versus 2.5. mEq/L (1.75 versus 1.25 mmol/L) in the infusate during predilution HF stabilized blood pressure, possibly because of greater peripheral resistance rather than through changes in cardiac performance. Am J Kidney Dis 46:470-480. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Hemodynamics; predilution hemofiltration; calcium concentration.

IRCULATORY INSTABILITY occurs fre- changes have not been fully elucidated. It is C quently during dialysis treatment and still attributed, among other things, to improved remains a significant cause of patient morbidity tolerance to volume removal,16,17 better preser- and mortality.1-4 As the numbers of elderly, dia- vation of blood volume,10 increased arterial betic, and cardiovascular-compromised patients peripheral resistances,8,17,18 preserved myocar- on dialysis treatment are increasing, the clinical dial contractility,10 and an advantageous ther- significance of the problem is growing.5-7 mal balance profile.19-22 Despite these advan- It has been the prevailing view for many tages, the use of hemofiltration (HF) has been years that convective techniques show im- unfavorable for years because of the limited proved intradialytic hemodynamics8-12 in terms removal of low-molecular-weight substances, of better cardiovascular stability9,13 and fewer technical problems, high costs, and lengthi- blood pressure–related symptoms14,15 com- ness of the treatment.5,23,24 During the last pared with classic diffusive treatments. The years, there is growing interest, especially in specific mechanisms responsible for the supe- some European countries, in on-line high-dose riority of convective transport principles in predilution HF as technological advances have preventing dialysis-induced hemodynamic helped us overcome most of the practical drawbacks of traditional low-dose postdilu- tion HF.5,25,26 From the Departments of Renal Medicine C and Anesthe- It is well recognized that calcium ions (iCas) siology and Intensive Care I, Skejby, Aarhus University Hospital, Aarhus, Denmark. have a pivotal role in the contractile process of Received February 2, 2005; accepted in revised form May both vascular smooth muscle cells and cardiac 31, 2005. myocytes.27-33 The importance of iCas in intra- Originally published online as doi:10.1053/j.ajkd.2005.05.022 dialytic hemodynamics has been shown during on July 18, 2005. 29,30,34-36 Address reprint requests to Nikolaos Karamperis, MD, acetate hemodialysis (HD) and bicarbon- Department of Renal Medicine C, Skejby, Aarhus University ate HD.37-39 Conversely, detailed experimental Hospital, Aarhus N, 8200, Denmark. E-mail: karamperis@ data on HF are lacking. stofanet.dk The aim of this study is to investigate the acute © 2005 by the National Kidney Foundation, Inc. 0272-6386/05/4603-0011$30.00/0 influence of low and high iCa concentrations in doi:10.1053/j.ajkd.2005.05.022 substitution fluid on hemodynamics during predi-

470 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 470-480 HIGH VERSUS LOW CALCIUM DURING HEMOFILTRATION 471

Table 1. Inclusion and Exclusion Criteria dialysis console was a Fresenius 4008H (Fresenius Medical Care, Bad Homburg, Germany). Filters were high-flux Inclusion criteria HdF100 (Fresenius Medical Care). Composition of the infu- Body dry weight Ͻ 95 kg sate (substitution fluid) was as follows: sodium, close to the Age Ͼ 18 y patient’s usual serum sodium level at dialysis start; potas- Absence of severe heart failure (New York Heart sium, 2 mEq/L (2 mmol/L); chloride, 106 mEq/L (106 Association classes 0-II) mmol/L); bicarbonate, 36 mEq/L (36 mmol/L); acetate, 30.6 Regular dialysis therapy for Ͼ 3mo mg/dL (3 mmol/L); magnesium, 1.2 mg/dL (0.5 mmol/L); Possibility to ultrafiltrate 3% of body weight glucose, 108 mg/dL (6 mmol/L); and iCa, 2.5 mEq/L (1.25 Hematocrit Ͼ 30% mmol/L) or 3.5 mEq/L (1.75 mmol/L). Ultrafiltration rate Stable arteriovenous fistula and sodium concentration were fixed during each session. Exclusion criteria The infusate was prepared on-line by the dialysis machine. Intradialytic hypotensive episodes within 4 wk prior Patients were instructed to achieve a predialysis weight inclusion close to 3% greater than their usual end-of-dialysis weight Hypertension (predialysis DBP Ͼ 110 mm Hg) (EDW) by adjustments of fluid intake. EDW was determined Diabetes mellitus before the study based on the usual clinical standards of the Acute myocardial infarction within 3 mo department. Ultrafiltration per study session was adjusted to Angina pectoris reach the defined EDW. Extra ultrafiltration of 300 mL was Severe heart failure (New York Heart Association added to compensate for saline injections used for cardiac classes III-IV) output (CO) measurements (described next). Effective blood Cardiac arrhythmia flow was adjusted individually to 1.3 ϫ EDW (kg) * Cerebrovascular incident within 3 mo 1,000/270 mL/min in all sessions, with a maximum of 350 Predialysis plasma iCa Ͻ 2.1 mEq/L (Ͻ1.05 mmol) or mL/min. Infusate flow rate was equal to dialyzer blood flow Ͼ2.8 mEq/L (Ͼ1.40 mmol) rate. Patients were not allowed to eat, drink, or sleep during Pregnancy the study sessions. All medications were continued and Severe illness, such as malignancy, infection, administered as usual throughout the study. gastrointestinal hemorrhage Alcohol or drug abuse Measurements Noncompliance or unwillingness to follow the protocol Blood pressure, pulse rate, and CO. After insertion of the dialysis needles, patients were allowed to rest for 10 minutes. After basal blood pressure and pulse recordings, lution HF in a randomized, crossover, blinded, heparin was administered and H-HF or L-HF was started controlled trial. according to randomization. Blood pressure and pulse rate subsequently were recorded at 15-minute intervals during METHODS dialysis and 10 minutes after dialysis by use of an automatic oscillometric blood pressure monitor mounted as part of the Study Design dialysis console (BPM; Fresenius Medical Care). CO was A blinded, randomized, controlled, crossover design was measured by means of the ultrasound velocity dilution applied, with each patient serving as his or her own control, method using a Transonic HD01 monitor (Transonic Sys- including 1 session of high-calcium HF (H-HF) and 1 tems Inc, Ithaca, NY). Before the initial measurements, the session of low-calcium HF (L-HF). Patients were blinded absence of local fistula recirculation was confirmed by use with regard to the calcium concentration of the treatment. of the Transonic device. A bolus of 30 mL of 37°C saline The study was conducted in 2 dialysis sessions, separated solution was injected into the venous cannula within 5 with a 1-week washout period. The order of treatments was seconds at each measurement. Ultrasound sensors were randomized. positioned on the arterial and venous tubing. After passing through the cardiovascular circuit, the ultrasound sensor on Patients the arterial tubing records the dilution curve resulting from the saline bolus. CO was calculated from the area under the Patients were recruited from the long-term HD population curve.38,40-42 Measurements were performed in duplicate from the Department of Renal Medicine, Aarhus University and if results deviated by more than 15%, a third measure- Hospital, Denmark. Inclusion and exclusion criteria are ment was performed. The mean of the 2 closest recordings listed in Table 1. was noted as the result. CO was measured at 15, 75, 135, The study was performed in accordance with the Helsinki 195, and 255 minutes. II Declaration and with approval of the local ethics commit- Total peripheral resistance (TPR) was calculated as mean tee. arterial pressure (MAP)/CO, and stroke volume, as CO/ pulse rate. Total cardiopulmonary recirculation was mea- Study Protocol sured by using the blood temperature monitor mounted as Each patient was randomly assigned to substitution fluid part of the dialysis console (BTM; Fresenius Medical Care) with an iCa concentration of 2.5 mEq/L (1.25 mmol/L; within the first and last hour of each session by use of the L-HF) or 3.5 mEq/L (1.75 mmol/L; H-HF) during 4.5 hours thermodilution method. All hemodynamic measurements of predilution HF. Infusate temperature was set at 37°C. The were performed with patients in the supine position with the 472 KARAMPERIS, SLOTH, AND JENSEN

Table 2. Hemodynamic and Thermal Variables

P (L-HF start P (H-HF start P (L-HF v L-HF Start L-HF End v L-HF end)* H-HF Start H-HF End v H-HF end)* H-HF)†

SBP (mm Hg) 141 Ϯ 4.6 124 Ϯ 4.4 Ͻ0.01 132 Ϯ 4.6 132 Ϯ 5.5 NS Ͻ0.01 DBP (mm Hg) 82 Ϯ 2.9 71 Ϯ 3.9 Ͻ0.01 77 Ϯ 2.1 78 Ϯ 3.3 NS Ͻ0.01 MAP (mm Hg) 103 Ϯ 490Ϯ 3 Ͻ0.05 101 Ϯ 399Ϯ 3NSϽ0.05 Pulse rate (beats/min) 74 Ϯ 381Ϯ 4 Ͻ0.05 77 Ϯ 383Ϯ 5NSNS CO (L/min) 6.5 Ϯ 0.3 5.1 Ϯ 0.2 Ͻ0.01 6.6 Ϯ 0.2 5.2 Ϯ 0.3 Ͻ0.01 NS Stroke volume (mL) 89.5 Ϯ 5.5 66.1 Ϯ 5.0 Ͻ0.01 90.9 Ϯ 4.0 70.2 Ϯ 5.9 Ͻ0.01 NS TPR (mm Hg * min/L) 15.4 Ϯ 07 17.7 Ϯ 0.8 Ͻ0.01 14.9 Ϯ 0.7 19.3 Ϯ 1.0 Ͻ0.01 Ͻ0.05 Relative blood volume (%) 88 Ϯ 1.2 89 Ϯ 0.9 NS Ultrafiltration (mL) 2,850 Ϯ 159 2,808 Ϯ 124 NS Arterial temperature (°C) 36.14 Ϯ 0.14 36.61 Ϯ 0.09 Ͻ0.01 36.18 Ϯ 0.11 36.67 Ϯ 0.08 Ͻ0.01 NS Venous temperature (°C) 35.26 Ϯ 0.13 35.84 Ϯ 0.09 Ͻ0.01 35.22 Ϯ 0.22 35.89 Ϯ 0.05 Ͻ0.01 NS Total energy transfer (kJ) 204 Ϯ 17.5 223 Ϯ 16.8 NS

NOTE. Values expressed as mean Ϯ SEM. *Within-treatment paired Student t-test start versus end. †Between-treatment paired Student t-test of individual (start versus end) delta values. head elevated approximately 30°. A standard cardiac echo 52 Ϯ 15 years (range, 25 to 70 years), there were scan was obtained on a separate occasion to ascertain that 5 women and 7 men, body weight was 68 Ϯ 12 there was no serious left ventricular disorder or hemodynami- kg (range, 48 to 90.5 kg), height was 170 Ϯ 10 cally significant valvular disease. Temperature and energy transfer. Temperature and en- cm, and time on renal replacement therapy was ergy transfer in the arterial and venous blood tubing were 6 Ϯ 6 years (range, 0.6 to 16.2 years). Six recorded continuously by the BTM module mounted on the patients were administered antihypertensive console. medication: a calcium channel blocker (n ϭ 1), Relative blood volume and ultrafiltration. Changes in ␤-blocker (n ϭ 3), and angiotensin II antagonists relative blood volume during treatment were recorded by the ϭ module integrated in the dialysis machine using continuous (n 2). Original renal diseases were chronic on-line measurement of the total blood protein concentration glomerulonephritis (n ϭ 3), polycystic kidney (blood volume monitor, Fresenius Medical Care). Cumula- disease (n ϭ 2), congenital renal disease (n ϭ 1), tive ultrafiltration volume was noted at 15-minute intervals. obstructive nephropathy (n ϭ 2), undetermined Biochemical parameters and dialysis dose. Before and cause of end-stage renal disease (n ϭ 1), bilateral 10 minutes after dialysis treatments, blood was drawn from ϭ the arterial cannula for determination of hemoglobin, hemat- nephrectomy (n 2), and analgesic nephropathy ocrit, sodium, potassium, total carbon dioxide, albumin, (n ϭ 1). Treatment modality before inclusion in creatinine, urea, and phosphorus levels. Additionally, serum the study was HF predilution (n ϭ 3), postdilu- iCa level was determined every 30 minutes. The urea reduc- tion hemodiafiltration (n ϭ 1), and low-flux HD tion ratio was calculated, and dialytic Kt/V urea was esti- (n ϭ 8). Mean final substitution fluid volume mated by using the Daugirdas formula.43 was 1.24 Ϯ 0.09 L/kg body weight per session Statistical Analysis (total volume, 84 Ϯ 9 L). Within-treatment changes in hemodynamic parameters Hemodynamic and thermal variables are listed were calculated as differences (delta values) between start in Table 2. and end values in each individual session. Between- intervention comparisons (L-HF versus H-HF) were per- Blood Pressure and Pulse Rate formed by analyzing the calculated delta values (start versus end) between H-HF and L-HF treatments, with no hemody- MAP decreased significantly during L-HF namic differences between treatments as the null hypothesis. throughout the treatment from (start versus Statistical significance (␣ level) was set to 0.05. Data were end) 103 Ϯ 4.4 to 90 Ϯ 3.5 mm Hg (P Ͻ 0.05). analyzed by use of paired or unpaired 2-tailed Student This was not the case in H-HF, for which a t-tests, as appropriate, using SPSS, version 10.0.5 (SPSS Inc, Chicago, IL). Data are expressed as mean Ϯ SEM. small decline in MAP failed to reach signifi- cance levels (corresponding values were 101 RESULTS Ϯ 3.4 to 99 Ϯ 3.5 mm Hg; Fig 1). Systolic After informed consent was obtained, 12 pa- (SBP) and diastolic blood pressure (DBP) de- tients were enrolled in the study. Mean age was creased significantly (P Ͻ 0.01) during L-HF HIGH VERSUS LOW CALCIUM DURING HEMOFILTRATION 473

Fig 1. Mean blood pressure change and pulse rate (mean ؎ SEM) during predilution HF with high (3.5 mEq/L) and low (2.5 mEq/L) iCa concentrations in infusate. Between-treatment comparison, P < 0.05 for mean blood pressure .NS for pulse rate. To convert iCa in mEq/L to mmol/L, multiply by 0.5 ؍ and P

(SBP, from 141 Ϯ 4.6 to 124 Ϯ 4.4 mm Hg; 89.5 Ϯ 5.5 to 66.1 Ϯ 5.0 mL in L-HF (P Ͻ 0.01) DBP, from 82 Ϯ 2.9 to 71 Ϯ 3.9 mm Hg), but versus 90.9 Ϯ 4.0 to 70.2 Ϯ 5.9 mL in H-HF remained stable (P ϭ not significant [NS]) (P Ͻ 0.01; Fig 2). There were no significant during H-HF (between-treatment comparison, differences between sessions at any time regard- P Ͻ 0.01). ing CO and stroke volume. As shown in Fig 1, heart rate was greater As shown in Fig 3, TPR increased linearly by during the second half of both treatments, but approximately 30% during H-HF from 14.9 Ϯ reached significant levels for only L-HF. Pulse 0.7 to 19.3 Ϯ 1.0 mm Hg * mL/min (P Ͻ 0.01). rate showed minor deviations in H-HF from 77 This also was the case for L-HF until the middle Ϯ 3.2 to 83 Ϯ 4.7 beats/min (P ϭ NS), but of the session; up to this point, TPR increased increased significantly in L-HF from 74 Ϯ 2.8 to linearly by approximately 15%, but then re- 81 Ϯ 4.3 beats/min (P Ͻ 0.05). No significant mained stable at this plateau until the end of the differences were observed between H-HF and treatment (start, 15.4 Ϯ 0.7 to end, 17.7 Ϯ 0.8 L-HF. mm Hg * mL/min; P Ͻ 0.01). Between-treat- ment comparison reached significant levels (P Ͻ CO, Stroke Volume, and Peripheral Resistance 0.05). CO declined significantly (P Ͻ 0.01) in L-HF (from 6.5 Ϯ 0.3 to 5.1 Ϯ 0.2 L/min) and H-HF Relative Blood Volume and Ultrafiltration (from 6.6 Ϯ 0.2 to 5.2 Ϯ 0.3 L/min; P Ͻ 0.01; Mean relative blood volume declined linearly Fig 2). Stroke volume declined similarly from during both treatments (Fig 4) and was reduced 474 KARAMPERIS, SLOTH, AND JENSEN

Fig 2. CO and stroke volume (mean ؎ SEM) during predilution HF with high (3.5 mEq/L) and low (2.5 mEq/L) iCa NS for both CO and stroke volume. To convert iCa ؍ concentrations in infusate. Between-treatment comparison, P in mEq/L to mmol/L, multiply by 0.5. to 88% Ϯ 1.2% in L-HF versus 89% Ϯ 0.9% in ultrafiltration volume was 2,808 Ϯ 124 mL in H-HF. Ultrafiltration rates did not differ signifi- H-HF versus 2,850 Ϯ 159 mL in L-HF (P ϭ NS; cantly between groups at any time. Cumulative Fig 4).

Fig 3. Peripheral resistance (mean ؎ SEM) during predilution HF with high (3.5 mEq/L) and low (2.5 mEq/L) iCa concentrations in infusate. Between-treatment comparison, P < 0.05. To convert iCa in mEq/L to mmol/L, multiply by 0.5. HIGH VERSUS LOW CALCIUM DURING HEMOFILTRATION 475

(Fig 4. Relative blood volume and ultrafiltration volume (mean ؎ SEM) during predilution HF with high (3.5 mEq/L NS for both relative blood ؍ and low (2.5 mEq/L) iCa concentrations in infusate. Between-treatment comparison, P volume and ultrafiltration volume. To convert iCa in mEq/L to mmol/L, multiply by 0.5.

Energy Transfer and Temperatures 2.5 Ϯ 0.1 mg/dL (1.2 Ϯ 0.05 mmol/L) to 2.3 Ϯ 0.1 mg/dL (1.16 Ϯ 0.06 mmol/L) in L-HF (Fig As shown in Fig 5, arterial temperatures in- 6). The rest of the biochemical parameter values creased in both sessions from 36.14°C Ϯ 0.14°C followed the same patterns in both treatment to 36.61°C Ϯ 0.09°C in L-HF (P Ͻ 0.01) and modalities. With the exception of iCa level, be- 36.18°C Ϯ 0.11°C to 36.67°C Ϯ 0.08°C in H-HF tween-treatment comparison did not show statis- (P Ͻ 0.01). Venous temperature increased signifi- tically significant differences. Within-treatment cantly from 35.22°C Ϯ 0.22°C to 35.89°C Ϯ comparisons were statistically significant (P Ͻ 0.05°C (P Ͻ 0.01) in H-HF and 35.26°C Ϯ 0.01) for all parameters with the exception of 0.13°C to 35.84°C Ϯ 0.09°C (P Ͻ 0.01) in L-HF. sodium level (P ϭ NS). Dialysis dose in both No significant differences were observed be- groups was similar according to Kt/V urea tween H-HF and L-HF. Total energy transfer (1.2 Ϯ 0.2). from patient to dialyzer did not differ signifi- cantly between treatments at any time (204 Ϯ DISCUSSION 17.5 kJ in L-HF and 223 Ϯ 16.8 kJ in H-HF; The present study assesses the hemodynamic Fig 5). influence of high and low calcium concentrations in substitution fluid during on-line high-dose Biochemical Parameters and Dialysis Dose predilution HF. This study shows that HF with a Data are listed in Table 3. iCa concentration high calcium concentration has a stabilizing influ- increased significantly from 2.4 Ϯ 0.1 mg/dL ence on blood pressure compared with a low (1.22 Ϯ 0.07 mmol/L) to 2.8 Ϯ 0.1 mg/dL (1.41 calcium concentration. The latter apparently was Ϯ 0.07 mmol/L) in H-HF (P Ͻ 0.01). Con- accomplished because of an iCa-induced in- versely, it declined significantly (P Ͻ 0.01) from crease in TPR, rather than through changes in 476 KARAMPERIS, SLOTH, AND JENSEN

Fig 5. Energy transfer and temperature (mean ؎ SEM) during predilution HF with high (3.5 mEq/L) and low (2.5 NS for both energy transfer and ؍ mEq/L) iCa concentrations in infusate. Between-treatment comparison; P temperature. Abbreviations: ATP, arterial temperature; VTP, venous temperature. To convert iCa in mEq/L to mmol/L, multiply by 0.5. cardiac performance. Because all other parame- both animal models31,45,46 and human subjects ters were matched between H-HF and L-HF, they (healthy or uremic),27,31-33,47,48 also showed the could not account for the differences noted in effect of iCas on arterial smooth muscle cells, hemodynamic responses. which resulted in increased peripheral vascular Several studies have shown the direct influ- resistance. Additionally, it was shown that iCas ence of iCas in cardiac muscle27,29,30,44 and its have a critical role in the release of cat- positive inotropic effect.28 Various studies, in echolamines from adrenergic nerves,49 and

Table 3. Clinical Chemistry

L-HF Start L-HF End H-HF Start H-HF End

Urea (mg/dL) 57 Ϯ 14 22 Ϯ 659Ϯ 14 22 Ϯ 6 Creatinine (mg/dL ) 10.7 Ϯ 2.9 4.7 Ϯ 1.6 10.6 Ϯ 2.7 4.6 Ϯ 1.4 Sodium (mEq/L) 136 Ϯ 5 138 Ϯ 3 137 Ϯ 3 138 Ϯ 1 Potassium (mEq/L) 4.8 Ϯ 0.7 3.6 Ϯ 0.4 4.6 Ϯ 0.5 3.7 Ϯ 0.4 Hematocrit (%) 38 Ϯ 341Ϯ 337Ϯ 240Ϯ 2 Carbon dioxide (mEq/L) 25.8 Ϯ 1.9 31.7 Ϯ 1.6 26.8 Ϯ 2.7 32.1 Ϯ 2.1 Phosphorus (mg/dL) 5 Ϯ 1.5 2.6 Ϯ 0.7 4.9 Ϯ 1.5 3.0 Ϯ 0.8 iCa (mEq/L) 2.5 Ϯ 0.1 2.3 Ϯ 0.1 2.4 Ϯ 0.1 2.8 Ϯ 0.1

NOTE. Values expressed as mean Ϯ SD. All P for between-treatment comparisons ϭ NS, with the exception of iCa (P Ͻ 0.01). All P for within-treatment comparisons are statistically significant, with the exception of sodium (P ϭ NS). To convert urea nitrogen in mg/dL to mmol/L, multiply by 0.357; creatinine in mg/dL to ␮mol/L, multiply by 88.4; sodium, potassium, and carbon dioxide in mEq/L to mmol/L, multiply by 1; phosphorus in mg/dL to mmol/L, multiply by 0.3229; iCa in mEq/L to mmol/L, multiply by 0.5. HIGH VERSUS LOW CALCIUM DURING HEMOFILTRATION 477

-Fig 6. iCa levels (mean ؎ SEM) during predilution HF with high (3.5 mEq/L) and low (2.5 mEq/L) iCa concentra tions in infusate. Between-treatment comparison, P < 0.05. To convert iCa in mEq/L to mmol/L, multiply by 0.5. greater calcium concentrations resulted in in- ations in peripheral resistance. We speculate that creased catecholamine levels and subsequently this difference could be attributed to the unique more pronounced vasoconstriction.32,47 Earlier solute transport characteristics and thermal prop- studies suggested that the sympathetic response erties of the convective techniques that, in con- can differ among various end-stage renal disease trast to diffusive techniques, show a more physi- replacement therapies17-19,50; likewise, convec- ological response to fluid removal and volume tive techniques showed better sympathetic re- depletion.10,19 sponse to fluid removal in terms of increased The effect of temperature on peripheral resis- catecholamine levels,17,50,51 increased or pre- tance during HF, as well as the thermal profile of served TPR,11,17,18 and preserved venous tone22 convective techniques, is well documented.1,20,22 compared with conventional HD. However, a lot During HF, blood has been shown to “cool down” of conflicting data still exist regarding the sympa- during extracorporeal circulation. Thermal en- thetic response shown by different renal replace- ergy losses and cooling of the blood in the ment methods.1,9,22 extracorporeal circuit cause vasoconstriction, A search of the literature was unable to show which increases vascular resistance. In our study, hemodynamic studies of HF using low and high total energy transfer for both sessions was at the iCa concentrations. Nevertheless, iCa influence same level (Table 3) and could not contribute to on intradialytic blood pressure stability during observed differences in peripheral resistance. We conventional low-flux HD has been described in noticed that during the first half of the treatment, earlier studies. During HD with a low dialysate during which patients in the low-calcium group calcium concentration, a significant decrease in were still normocalcemic (Fig 6), peripheral re- blood pressure occurred compared with a high sistance increased linearly and to the same extent calcium concentration in dialysis patients with as in the high-calcium group (Fig 3). In the normal cardiac function39,52 and cardiac compro- second half of the treatment, as patients in the mise.38 Conversely, a high dialysis calcium con- low-calcium group were becoming slightly hy- centration improved intradialytic hemodynam- pocalcemic, TPR decreased (but was still signifi- ics.35-39,53 This response, contrary to our findings, cantly greater compared with predialysis levels) was mediated mainly by changes in myocardial and remained at that level until the end of the contractility,29,30,34,38,39,53 not through alter- treatment. In the high-calcium group, patients 478 KARAMPERIS, SLOTH, AND JENSEN reached hypercalcemic levels as early as intradia- physiological reference” for future studies includ- lytic minute 75, and TPR continued to increase ing hypotensive-prone patients. Considering the linearly throughout the entire treatment, possibly previous mentioned design limitation, extrapola- because of iCa-induced vasoconstriction. tion of results to other dialysis subpopulations This observation is in agreement with the should be done with caution. In theory, if patients results of Drop and Scheidegger,27 who reported with cardiac comorbidity had been included, the that different iCa levels showed distinguishable difference between H-HF and L-HF would be physiological influence. Although an increase more pronounced. Conversely, in this group of from less than normal to near-normal iCa levels patients, increased TPR could make them more induces predominantly cardiac changes, an addi- unstable by increasing their afterload and thus tional increase in iCa levels to greater than the further decreasing stroke volume, precipitating physiological range (from normal to greater than hypotension. Therefore, we cannot exclude a normal) mainly affected systemic vascular resis- differentiated response if a hypotensive-prone tance.27 The additional increase in peripheral population was chosen. resistance caused by hypercalcemia reduces cuta- We are aware of the ongoing discussion regard- neous thermal loss, which consequently in- ing the potential drawbacks of a positive calcium creases blood temperature and results in exces- balance during dialysis (eg, vascular classifica- sive thermal loss from the extracorporeal circuit. tion) and the trend toward minimization of total This is not in conflict with the observed greater intradialytic calcium load.54-60 This study cannot total energy transfer during H-HF compared with directly address this specific issue because it was L-HF (223 Ϯ 16.8 versus 204 Ϯ 17.5), although designed merely to focus on the pathophysiologi- it failed to reach significance. It should be stressed cal/hemodynamic mechanism under different cal- at that point that we chose a fixed infusate cium-loading conditions. However, it can indi- temperature (37°C), which represents a standard cate some possible clinical implications. A high procedure in most dialysis units, although it is calcium concentration in the substitution fluid well documented that it results in an increase in can be an alternative option for seriously hypoten- body temperature and vasodilation. We cannot sive patients during convective treatments who exclude that a euthermic treatment may have a do not respond to the usual routine clinical proce- different impact on hemodynamics. A small in- dures. We showed a time difference between crease in sodium concentration in both treat- changes in iCa levels and TPR “response.” iCa ments was seen (Table 3). We speculate that it concentrations between L-HF and H-HF diverge can be attributed to the intravenous physiological very early during the HF process (as soon as 60 saline injections necessary for CO measurements minutes; Fig 6). Conversely, TPR started to dif- (ϳ300 mL for every patient). ferentiate between the 2 treatments after only CO and stroke volume followed parallel reduc- minute 150 (Fig 3). Presumably, iCa augments tion profiles in both treatments. There was a TPR only in the setting of progressive volume tendency for greater stroke volume and CO dur- depletion as ultrafiltration progresses. Taking the ing H-HF compared with L-HF, although it failed “calcium loading” issue under consideration, an to reach significant levels. This tendency was antihypotension strategy can be designed based more prominent during the second half of the on this observation. Patients could start their sessions as the difference in iCa levels between treatment at low calcium levels and switch to the 2 treatments was increasing. We presume that high levels after the first half. In that way, we for H-HF, the possible positive cardiac inotropic “provide” iCas at that point at which patients are effect of hypercalcemia was overwhelmed by the more prone to develop hypotension because of reduction in blood volume caused by fluid re- “failure” of TPR to increase. moval. However, how can these observations fit with One limitation of this study is that it was the group of patients experiencing a hypotensive designed as an acute study of stable HD patients. episode at the first part of the dialysis treatment? A hemodynamically unstable population would Our stable dialysis patients showed an equal be a more rational choice, but our aim is to increase in TPR at the first part of the treatment, investigate and use this group as a “basic patho- although the iCa concentration difference be- HIGH VERSUS LOW CALCIUM DURING HEMOFILTRATION 479 tween L-HF and H-HF already was marked. We 9. Fox SD, Henderson LW: Cardiovascular response dur- assume that another underlying mechanism ex- ing hemodialysis and hemofiltration: Thermal, membrane, and catecholamine influences. Blood Purif 11:224-236, 1993 ists for hypotensive episodes during the first part 10. Santoro A, Mancini E, Zucchelli P: The impact of of the dialysis treatment. Additional studies need haemofiltration on the systemic cardiovascular response. to be performed to clarify these speculations. Nephrol Dial Transplant 15:S49-S54, 2000 (suppl 2) We believe our findings are important not only 11. Shaldon S, Deschodt G, Beau MC, Claret G, Mion H, because they are unique in the literature, but also Mion C: Vascular stability during high flux haemofiltration (HF). Proc Eur Dial Transplant Assoc 16:695-697, 1979 because they contribute toward better understand- 12. Vagge R, Cavatorta F, Queirolo C, Rosselli P, Gentile ing of the role of iCas in such a controversial A, Bertulla A: Hemodynamic changes during acetate dialy- issue as “dialysis hemodynamics.” sis, bicarbonate dialysis and hemofiltration. Blood Purif In summary, a significant intratreatment reduc- 6:43-50, 1988 13. Quellhorst E: Long-term follow up in chronic hemo- tion in CO and stroke volume was shown to the filtration. Int J Artif Organs 6:115-120, 1983 same extent for both H-HF and L-HF. Relative 14. Altieri P, Sorba G, Bolasco P, et al: Pre-dilution blood volume, ultrafiltration volume, energy haemofiltration—The Sardinian multicentre studies: Present transfer, and temperature were matched between and future. The Sardinian Collaborative Study Group on the 2 treatments and therefore could not account Haemofiltration On-Line. Nephrol Dial Transplant 15:S55- S59, 2000 (suppl 2) for the observed differences. Intertreatment com- 15. Altieri P, Sorba G, Bolasco P, et al: Predilution parisons showed significantly lower MAP, SBP, haemofiltration—The Second Sardinian Multicentre Study: DBP, and TPR in the L-HF group compared with Comparisons between haemofiltration and haemodialysis the H-HF group. In conclusion, we show that iCa during identical Kt/V and session times in a long-term cross-over study. Nephrol Dial Transplant 16:1207-1213, concentration of 3.5 versus 2.5 mEq/L (1.75 2001 versus 1.25 mmol/L) in infusate during predilu- 16. Baldamus CA, Knobloch M, Schoeppe, Koch KM: tion HF stabilizes blood pressure, possibly be- Hemodialysis/hemofiltration: A report of a controlled cross- cause of greater peripheral resistance, rather than over study. Int J Artif Organs 3:211-214, 1980 through changes in cardiac performance. 17. Baldamus CA, Ernst W, Lysaght MJ, Shaldon S, Koch KM: Hemodynamics in hemofiltration. Int J Artif Organs 6:27-31, 1983 REFERENCES 18. Hampl H, Paeprer H, Unger V, Fischer C, Resa I, 1. Daugirdas JT: Dialysis hypotension: A hemodynamic Kessel M: Hemodynamic changes during hemodialysis, se- analysis. Kidney Int 39:233-246, 1991 quential ultrafiltration, and hemofiltration. Kidney Int Suppl 2. Henrich WL: Hemodynamic instability during hemodi- 10:S83-S88, 1980 alysis. Kidney Int 30:605-612, 1986 19. Hendrikus W, van Kuijk M, Leunissen KM: Hemody- namic stability during different forms of dialysis therapy: A 3. van der Sande FM, Kooman JP, Leunissen KM: Intra- pathogenetic analysis. Blood Purif 14:405-420, 1996 dialytic hypotension—New concepts on an old problem. 20. Maggiore Q, Pizzarelli F, Sisca S, et al: Blood tempera- Nephrol Dial Transplant 15:1746-1748, 2000 ture and vascular stability during hemodialysis and hemofil- 4. Zager PG, Nikolic J, Brown RH, et al: “U” Curve tration. Trans Am Soc Artif Intern Organs 28:523-527, 1982 association of blood pressure and mortality in hemodialysis 21. Maggiore Q, Pizzarelli F, Dattolo P, Maggiore U, patients. Medical Directors of Dialysis Clinic, Inc. Kidney Cerrai T: Cardiovascular stability during haemodialysis, Int 54:561-569, 1998 haemofiltration and haemodiafiltration. Nephrol Dial Trans- 5. McCarthy JT, Moran J, Posen G, et al: A time for plant 15:S68-S73, 2000 (suppl 1) rediscovery: Chronic hemofiltration for end-stage renal dis- 22. van Kuijk WH, Hillion D, Savoiu C, Leunissen KM: ease. Semin Dial 16:199-207, 2003 Critical role of the extracorporeal blood temperature in the 6. Berthoux F, Gellert R, Jones E, et al: Epidemiology hemodynamic response during hemofiltration. J Am Soc and demography of treated end-stage renal failure in the Nephrol 8:949-955, 1997 elderly: From the European Renal Association (ERA- 23. Ledebo I: On-line hemofiltration. Old concept—New EDTA) Registry. Nephrol Dial Transplant 13:S65-S68, 1998 approach. Contrib Nephrol 137:221-226, 2002 (suppl 7) 24. Ledebo I: Principles and practice of hemofiltration 7. Berthoux F, Bernheim J, Gellert R, et al: The project of and hemodiafiltration. Artif Organs 22:20-25, 1998 the European Renal Association (ERA-EDTA) for a Euro- 25. David S, Bostrom M, Cambi V: Predilution hemofil- pean nephrological network. ERA Registry and the ERA tration. Clinical experience and removal of small molecular Council. Nephrol Dial Transplant 13:S30-S33, 1998 (suppl 1) weight solutes. Int J Artif Organs 18:743-750, 1995 8. Baldamus CA, Ernst W, Frei U, Koch KM: Sympa- 26. ERA-EDTARegistry: ERA-EDTARegistry 2003 An- thetic and hemodynamic response to volume removal during nual Report. Amsterdam, The Netherlands, Academic Medi- different forms of renal replacement therapy. Nephron 31: cal Center. Available at: www.era-edta-reg.org. Accessed: 324-332, 1982 June 7, 2005 480 KARAMPERIS, SLOTH, AND JENSEN

27. Drop LJ, Scheidegger D: Plasma ionized calcium 45. Overbeck HW, Molnar JI, Haddy FJ: Resistance to concentration: Important determinant of the hemodynamic blood flow through the vascular bed of the dog forelimb. response to calcium infusion. J Thorac Cardiovasc Surg Local effects of sodium, potassium, calcium, magnesium, 79:425-431, 1980 acetate, hypertonicity and hypotonicity. Am J Cardiol 8:533- 28. Feinberg H, Boyd E, Katz LN: Calcium effect on 541, 1961 performance of the heart. Am J Physiol 202:643-648, 1962 46. Berl T, Levi M, Ellis M, Chaimovitz C: Mechanism 29. Henrich WL, Hunt JM, Nixon JV: Increased ionized of acute hypercalcemic hypertension in the conscious rat. calcium and left ventricular contractility during hemodialy- Hypertension 7:923-930, 1985 sis. N Engl J Med 310:19-23, 1984 47. Marone C, Beretta-Piccoli C, Weidmann P: Acute 30. Lang RM, Fellner SK, Neumann A, Bushinsky DA, hypercalcemic hypertension in man: Role of hemodynamics, Borow KM: Left ventricular contractility varies directly catecholamines, and renin. Kidney Int 20:92-96, 1981 with blood ionized calcium. Ann Intern Med 108:524-529, 48. Overbeck HW, Pamnani MB, Derifield RS: Similar 1988 vasoconstrictor responses to calcium in normotensive and 31. Lifschitz MD: Serum calcium and peripheral resis- essential hypertensive men. Proc Soc Exp Biol Med 149:519- tance. Ann Intern Med 77:482, 1972 525, 1975 32. Vlachakis ND, Frederics R, Valasquez M, Alexander 49. Rubin RP: The role of calcium in the release of N, Singer F, Maronde RF: Sympathetic system function and neurotransmitter substances and hormones. Pharmacol Rev vascular reactivity in hypercalcemic patients. Hypertension 4:452-458, 1982 22:389-428, 1970 33. Weidmann P, Massry SG, Coburn JW, Maxwell MH, 50. Baldamus CA, Ernst W, Fassbinder W, Koch KM: Atleson J, Kleeman CR: Blood pressure effects of acute Differing haemodynamic stability due to differing sympa- hypercalcemia. Studies in patients with chronic renal failure. thetic response: Comparison of ultrafiltration, haemodialysis Ann Intern Med 76:741-745, 1972 and haemofiltration. Proc Eur Dial Transplant Assoc 17:205- 34. Fellner SK, Lang RM, Neumann A, Spencer KT, 212, 1980 Bushinsky DA, Borow KM: Physiological mechanisms for 51. Zucchelli P, Santoro A, Sturani A, Degli EE, Chiarini calcium-induced changes in systemic arterial pressure in C, Zuccala A: Effects of hemodialysis and hemofiltration on stable dialysis patients. Hypertension 13:213-218, 1989 the autonomic control of circulation. Trans Am Soc Artif 35. Maynard JC, Cruz C, Kleerekoper M, Levin NW: Intern Organs 30:163-167, 1984 Blood pressure response to changes in serum ionized cal- 52. Kyriazis J, Stamatiadis D, Mamouna A: Intradialytic cium during hemodialysis. Ann Intern Med 104:358-361, and interdialytic effects of treatment with 1.25 and. 1.75 1986 mmol/L of calcium dialysate on arterial compliance in 36. Sherman RA, Bialy GB, Gazinski B, Bernholc AS, patients on hemodialysis. Am J Kidney Dis 35:1096-1103, Eisinger RP: The effect of dialysate calcium levels on blood 2000 pressure during hemodialysis. Am J Kidney Dis 8:244-247, 53. Kyriazis J, Glotsos J, Bilirakis L, et al: Dialysate 1986 calcium profiling during hemodialysis: Use and clinical 37. Leunissen KM, van den Berg BW, van Hooff JP: implications. Kidney Int 61:276-287, 2002 Ionized calcium plays a pivotal role in controlling blood 54. Block GA, Port FK: Re-evaluation of risks associated pressure during haemodialysis. Blood Purif 7:233-239, 1989 with hyperphosphatemia and hyperparathyroidism in dialy- 38. van der Sande FM, Cheriex EC, van Kuijk WH, sis patients: Recommendations for a change in management. Leunissen KM: Effect of dialysate calcium concentrations Am J Kidney Dis 35:1226-1237, 2000 on intradialytic blood pressure course in cardiac-compro- 55. Goodman WG, Goldin J, Kuizon BD, et al: Coronary- mised patients. Am J Kidney Dis 32:125-131, 1998 artery calcification in young adults with end-stage renal 39. van Kuijk WH, Mulder AW, Peels CH, Harff GA, disease who are undergoing dialysis. N Engl J Med 342:1478- Leunissen KM: Influence of changes in ionized calcium on 1483, 2000 cardiovascular reactivity during hemodialysis. Clin Nephrol 56. Goodman WG, London G, Amann K, et al: Vascular 47:190-196, 1997 calcification in chronic kidney disease. Am J Kidney Dis 40. Krivitski NM: Novel method to measure access flow 43:572-579, 2004 during hemodialysis by ultrasound velocity dilution tech- nique. ASAIO J 41:M741-M745, 1995 57. Guerin AP, London GM, Marchais SJ, Metivier F: 41. Kisloukhine VV, Dean DA: Validation of a novel Arterial stiffening and vascular calcifications in end-stage ultrasound dilution method to measure cardiac output during renal disease. Nephrol Dial Transplant 15:1014-1021, 2000 hemodialysis. ASAIO J 42:M906-M907, 1996 58. Salusky IB, Goodman WG: Cardiovascular calcifica- 42. Nikiforov YV, Kisluchine VV, Chaus NI: Validation tion in end-stage renal disease. Nephrol Dial Transplant of a new method to measure cardiac output during extracor- 17:336-339, 2002 poreal detoxification. ASAIO J 42:M903-M905, 1996 59. Coladonato JA, Szczech LA, Friedman EA, Owen 43. Daugirdas JT: Second generation logarithmic esti- WF Jr: Does calcium kill ESRD patients—The skeptic’s mates of single-pool variable volume Kt/V: An analysis of perspective. Nephrol Dial Transplant 17:229-232, 2002 error. J Am Soc Nephrol 4:1205-1213, 1993 60. Block GA, Hulbert-Shearon TE, Levin NW, Port FK: 44. Mori K: The effects of infusion of calcium and Association of serum phosphorus and calcium ϫ phosphate magnesium ions on the cardiovascular system in man. Jpn product with mortality risk in chronic hemodialysis patients: Heart J 19:226-235, 1978 A national study. Am J Kidney Dis 31:607-617, 1998 Epoetin Alfa Use in Patients With ESRD: An Analysis of Recent US Prescribing Patterns and Hemoglobin Outcomes

Allan J. Collins, MD, Robert M. Brenner, MD, Joshua J. Ofman, MD, Eric M. Chi, PhD, Nina Stuccio-White, DO, Mahesh Krishnan, MD, Craig Solid, MS, Norma J. Ofsthun, PhD, and J. Michael Lazarus, MD

● Background: It is unknown to what degree physicians adjust erythropoietin doses to achieve hemoglobin levels (11.0 to 12.0 g/dL [110 to 120 g/L]) recommended by the National Kidney Foundation–Kidney Disease Outcomes Quality Initiative (NKF-K/DOQI) for patients with end-stage renal disease receiving hemodialysis. Our objective is to examine epoetin alfa prescribing patterns for achieving the target hemoglobin level range in this population. Methods: Monthly hemoglobin levels and epoetin alfa doses from 2 large databases were retrospectively analyzed. One data set comprised 31,267 patients from the Fresenius Medical Care–North America (FMC-NA) database, and the other comprised 128,761 patients based on claims for Medicare services. Results: Longitudinal evaluation of the FMC-NA data set showed that hemoglobin levels in patients administered epoetin alfa cycled in and out of the NKF-K/DOQI hemoglobin target range, and doses were decreased in 98.8% of patients with persistent hemoglobin levels greater than 12.0 g/dL (>120 g/L). Hemoglobin levels in patients from the Medicare data set that initially were outside the target range migrated into the range with epoetin alfa dose titration. FMC-NA patients with a 3-month average hemoglobin level less than 11.0 g/dL (<110 g/L) were administered significantly greater epoetin alfa doses than those with average hemoglobin levels greater than 12.0 g/dL (>120 g/L; 21,838 versus 13,503 U/wk; P < 0.0001). Less than 0.4% of patients administered epoetin alfa were persistently anemic (hemoglobin < 11.0 g/dL [<110 g/L]) and were administered persistently high doses (>30,000 U/wk), but failed to respond with a 0.5-g/dL or greater (>5-g/L) increase in hemoglobin levels. Conclusion: In these analyses, few hemodialysis patients experienced persistent anemia while being administered high epoetin alfa doses. Physicians appeared to appropriately adjust doses to achieve hemoglobin levels recommended by the NKF-K/DOQI guidelines. Am J Kidney Dis 46:481-488. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Anemia; hemodialysis (HD); end-stage renal disease (ESRD); epoetin alfa; erythropoietin; hemoglo- bin; Medicare; reimbursement.

NEMIA DEVELOPS during the course of agement has been a central part of ESRD care to A chronic kidney disease (CKD), primarily improve patient well-being. as a result of insufficient erythropoietin produc- The introduction of recombinant human eryth- tion by the diseased kidneys. The severity of ropoietin (epoetin alfa) in 1989 marked a signifi- anemia is related directly to the severity of cant advance in anemia treatment in hemodialy- CKD.1,2 Even before progression to end-stage sis patients, of whom more than 90% in the renal disease (ESRD), nearly 40% of patients United States were being administered epoetin with CKD have hemoglobin levels less than 11.0 alfa therapy as of 2004.21 Mean monthly hemo- g/dL (Ͻ110 g/L).2,3 Anemia is a nearly ubiqui- tous comorbidity in patients with ESRD. As of December 2002, 308,910 patients with ESRD in the United States were undergoing hemodialysis.4 From the Chronic Disease Research Group, Minneapolis The negative consequences of anemia in pa- Medical Research Foundation and University of Minnesota, tients with ESRD are well documented and in- Minneapolis, MN; Amgen Inc, Thousand Oaks, CA; and Fresenius Medical Care–North America, Lexington, MA. clude decreased oxygen delivery and utiliza- Received December 17, 2004; accepted in revised form 5-7 tion, impaired cognition and decreased mental May 4, 2005. acuity,8 and worsened cardiovascular function.9-12 Originally published online as doi:10.1053/j.ajkd.2005.05.018 In addition, anemia adversely affects long-term on July 12, 2005. 13-15 13-16 Funding for the analysis described in this manuscript was survival and increases hospitalizations. provided by Amgen Inc, Thousand Oaks, CA. Patients with ESRD who achieve target hemoglo- Address reprint requests to J. Michael Lazarus, MD, bin levels experience improvement in health- Fresenius Medical Care–North America, 2 Ledgemont Cen- related quality-of-life measures, including en- ter, 95 Hayden Ave, Lexington, MA 02420. Email: [email protected] ergy and activity levels, sleeping and eating © 2005 by the National Kidney Foundation, Inc. behaviors, sexual function, and health and life 0272-6386/05/4603-0012$30.00/0 satisfaction.17-20 For these reasons, anemia man- doi:10.1053/j.ajkd.2005.05.018

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 481-488 481 482 COLLINS ET AL globin levels in hemodialysis patients increased METHODS from 9.7 g/dL (97 g/L) in 1991 to 11.7 g/dL (117 This is a 3-part retrospective analysis. The first part is a g/L) in 2002. Increased hemoglobin levels were cross-sectional analysis of the distribution of hemoglobin accompanied by an increase in mean weekly levels in a population of patients with ESRD from the epoetin alfa dose from 6,000 to 17,000 units.4 Fresenius Medical Care–North America (FMC-NA) data- base who were receiving hemodialysis. In the second part of During this same period, the percentage of preva- the analysis, we examine the longitudinal relationship be- lent hemodialysis patients with hemoglobin lev- tween hemoglobin levels and epoetin alfa dosing patterns by els less than 11.0 g/dL (Ͻ110 g/L) decreased using data from records of claims submitted to the CMS for from approximately 85% to 25%.4 services provided to patients with ESRD at the Minneapolis A hemoglobin value of 11.0 g/dL (110 g/L) is Medical Research Foundation. In the third part, we again use the FMC-NA database to examine persistency rates for low a significant benchmark because the National hemoglobin levels and high epoetin alfa doses. Kidney Foundation–Kidney Disease Outcomes Quality Initiative (NKF-K/DOQI) Clinical Prac- Distribution of Hemoglobin Levels: FMC-NA tice Guidelines for Anemia of CKD recommend Data Set a target hemoglobin level between 11.0 and 12.0 Data for 31,267 patients from the FMC-NA database were g/dL (110 and 120 g/L) in hemodialysis patients analyzed for May through December 2001 to determine the administered epoetin alfa.22 The Centers for distribution of mean hemoglobin values in prevalent hemo- Medicare and Medicaid Services (CMS) has in- dialysis patients. Rolling averages were calculated for each corporated the NKF-K/DOQI guidelines for tar- of six 3-month periods, ending in July, August, September, October, November, and December 2001. Several monthly get hemoglobin level into its epoetin alfa reim- records of predialysis hemoglobin levels and epoetin alfa bursement policy and its ESRD Clinical doses were obtained for each patient. For consistency with Performance Measures (CPM) project. The CPM data captured on Medicare claims, the hemoglobin value project provides comparative data to providers to used for purposes of this analysis was the last value of the month. Missing hemoglobin values were imputed by the assist them in assessing and improving the care last-observation-carried-forward method. A weekly average of patients with ESRD. epoetin alfa dose was calculated for each patient by dividing In addition to recommending target hemoglobin total dose for the month by number of days in the month and levels, the practice guidelines also address the means multiplying by 7. Doses missed because of hospitalization or of achieving the recommended levels, including missed hemodialysis treatments were not captured in the analysis; therefore, actual epoetin alfa doses administered appropriate adjustments in epoetin alfa dosing, man- may have been underestimated. Epoetin alfa doses were agement of iron deficiency, and reduced use of compared by using a simple t-test with log-transformation. catheters for vascular access. The guidelines appear A monthly cross-sectional analysis of the distribution of to be viewed as a positive development by physi- FMC-NA patients in the hemoglobin level categories also cians and reimbursement agencies. However, it is was conducted during the 2-year period from July 2000 through June 2002. unknown how successful dialysis facilities, acting under the order of physicians, are in achieving the Relationship Between Hemoglobin Levels and recommended hemoglobin levels and associated Epoetin Alfa Doses: CMS Data Set dosing of epoetin alfa. An examination of this The prevalent cohort in the analysis of hemoglobin levels question is important because of the ongoing scru- and epoetin alfa doses consisted of all hemodialysis patients tiny of Medicare reimbursement for erythropoietin who were alive for the 6-month period from April through therapy in hemodialysis patients. The goal of the September 2001 and had Medicare as primary payor. A total retrospective analyses reported here is to examine of 128,761 patients representing 3,516 dialysis facilities physician practice patterns and understand the rela- were included in this analysis. Facilities were grouped according to mean hemoglobin levels of their patients in tionship among hemoglobin levels, epoetin alfa April 2001 (Ͻ11.0 g/dL [Ͻ110 g/L], 11.0 to 12.5 g/dL [110 dose, and epoetin alfa dosing decisions. We per- to 125 g/L], and Ͼ12.5 g/dL [Ͼ125 g/L]). The upper limit of formed cross-sectional and longitudinal evalua- 12.5 g/dL (125 g/L), measured during a 3-month period, was tions of 2 large patient databases. These databases selected because it is the audit trigger used by the CMS for reimbursement decisions. The CMS data set very likely offer a unique opportunity to understand variability included patients from the FMC-NA data set. As with the in epoetin alfa dosing and hemoglobin levels in FMC-NA data set, missed doses may have resulted in large cohorts of hemodialysis patients. underestimation of the actual epoetin alfa dose administered. EPOETIN ALFA PRESCRIBING PATTERNS IN ESRD 483

Persistency Analysis: FMC-NA Data Set In the persistency analysis of the FMC-NA data set, we evaluated clinical response in 31,267 patients with persis- tently low hemoglobin levels and persistently high epoetin alfa doses for the 6-month period from July through Decem- ber 2001. We define clinical response as a 0.5-g/dL-or- greater (Ն5-g/L) increase in hemoglobin level. We define low hemoglobin level as a last hemoglobin level of the month less than 11.0 g/dL (Ͻ110 g/L) for 6 consecutive months and high epoetin alfa dose as more than 30,000 U/wk for 6 consecutive months. Because we are not aware of a standard definition of high-dose epoetin alfa therapy, we used an arbitrary definition of 30,000 U/wk, which is approxi- mately the 85th percentile of weekly epoetin alfa doses in the FMC-NA data set. Fig 1. Distribution of 3-month rolling average hemo- RESULTS globin levels in hemodialysis patients (FMC-NA data set). To convert hemoglobin in g/dL to g/L, multiply by Distribution of Hemoglobin Levels: FMC-NA 10. Data Set A cross-sectional analysis of the FMC-NA g/L], and Ͼ12.0 g/dL [Ͼ120 g/L]) remained data set showed that, during the period from May relatively stable over time (Fig 2). Notably, there through July 2001, the 3-month rolling average was very little fluctuation in the proportion of hemoglobin level of the patient population was patients with hemoglobin levels between 11.0 11.6 Ϯ 1.4 g/dL (116 Ϯ 114 g/L). The distribu- and 12.0 g/dL (110 and 120 g/L). A reciprocal tion of hemoglobin values approximated a nor- relationship existed between the 2 groups outside mal distribution, with 27% of patients (n ϭ the NKF-K/DOQI target range; an increase in the 8,432) having a 3-month rolling average hemo- proportion of patients in 1 group was accompa- globin level less than 11.0 g/dL (Ͻ110 g/L), 39% nied by a decrease in the proportion of patients in (n ϭ 12,113) having an average hemoglobin the other group. Over time, mean hemoglobin level within the NKF-K/DOQI target range of values for the entire patient population remained 11.0 to 12.0 g/dL (110 to 120 g/L), and the relatively constant. remaining 34% (n ϭ 10,722) having an average hemoglobin level greater than 12.0 g/dL (Ͼ120 Relationship Between Hemoglobin Levels and g/L; Fig 1). Epoetin Alfa Doses: CMS Data Set Monthly analysis of the FMC-NA database The CMS data analysis examined the relation- during a 24-month period from July 2000 through ship between hemoglobin levels and epoetin alfa June 2002 showed that the proportion of patients dosing patterns over time. Figure 3 shows a in each of 3 hemoglobin level categories (Ͻ11.0 6-month (April through September 2001) longi- g/dL [Ͻ110 g/L], 11.0 to 12.0 g/dL [110 to 120 tudinal follow-up of patients in dialysis facilities

Fig 2. Temporal trend during 2 years in the distribu- tion of hemodialysis patients in hemoglobin level catego- ries (FMC-NA data set). To convert hemoglobin in g/dL to g/L, multiply by 10. 484 COLLINS ET AL

Fig 3. Mean monthly he- moglobin levels and weekly epoetin alfa doses for hemo- dialysis patients at facilities with a mean hemoglobin level of 11.0 to 12.5 g/dL (110 to 125 g/L) in April 2001 (CMS data set). To convert hemo- globin in g/dL to g/L, multi- ply by 10. who had a mean hemoglobin level within the mean weekly epoetin alfa dose and 3-month target range defined by the NKF-K/DOQI guide- rolling average hemoglobin level (Fig 4). Pa- lines (lower limit of 11.0 g/dL [110 g/L]) and the tients with mean hemoglobin values less than audit trigger of the Hematocrit Measurement 11.0 g/dL (Ͻ110 g/L) and greater than 12.0 g/dL Audit–Program Memorandum (upper limit of (Ͼ120 g/L) were administered mean weekly 12.5 g/dL [125 g/L]). Patients with a mean hemo- epoetin alfa doses of 21,838 and 13,503 units, globin level less than 11.0 g/dL (Ͻ110 g/L) in respectively (P Ͻ 0.0001), a 38% difference. Of April 2001 showed a noticeable increase in mean patients who began the 6-month analysis period weekly epoetin alfa dose, as well as an increase being administered epoetin alfa doses greater in mean hemoglobin level. Patients with a mean than 30,000 U/wk and completed the analysis Ͼ hemoglobin level greater than 12.5 g/dL ( 125 period, 30% (1,173 of 3,955 patients) persis- g/L) in April 2001 showed a decrease in mean tently were administered these doses. Of patients weekly epoetin alfa dose, as well as a decline in who began the analysis period being adminis- mean hemoglobin level. Patients with a mean tered doses greater than 40,000 and greater than hemoglobin level between 11.0 and 12.5 g/dL 60,000 U/wk and completed the analysis period, (110 and 125 g/L) in April 2001 showed consis- tent epoetin alfa dosing while maintaining a 25% (145 of 574 patients) and 27% (576 of mean hemoglobin level within the target range. 2,123 patients) persistently were administered Thus, in this analysis, physicians appeared to these doses, respectively. adjust epoetin alfa doses to alter (or maintain) In addition to evaluating persistent epoetin hemoglobin levels within the NKF-K/DOQI alfa use, we determined the extent to which guidelines. patients persistently remained in the same hemo- globin level category (Ͻ11.0 g/dL [Ͻ110 g/L], Persistency Analysis: FMC-NA Data Set 11.0 to 12.0 g/dL [110 to 120 g/L], and Ͼ12.0 In a persistency analysis of the FMC-NA data g/dL [Ͼ120 g/L]). We found that only 5% of set, we observed an inverse relationship between patients (n ϭ 10,518) who began the 6-month

Fig 4. The distribution of epoetin alfa doses by hemo- globin level category (FMC-NA data set). To convert hemo- globin in g/dL to g/L, multiply by 10. EPOETIN ALFA PRESCRIBING PATTERNS IN ESRD 485

Fig 5. Hemoglobin level persistency rates in hemodi- alysis patients (FMC-NA data set). To convert hemoglobin in g/dL to g/L, multiply by 10. period with a mean hemoglobin value within the small percentage of patients with hemoglobin NKF-K/DOQI target range of 11.0 to 12.0 g/dL levels persistently greater than the target range (110 to 120 g/L) remained within that range for 6 did not have epoetin alfa dose adjustments in an consecutive months (Fig 5). Greater persistency attempt to bring them into the range. rates (10% and 17%, respectively) were observed for patients in the groups with a hemoglobin level DISCUSSION less than 11.0 g/dL (Ͻ110 g/L; n ϭ 9,161) and We performed retrospective cross-sectional and greater than 12.0 g/dL (Ͼ120 g/L; n ϭ 10,821). longitudinal evaluations of 2 large patient data- By combining persistency analyses for both bases. Our goal is to examine physician practice high-dose epoetin alfa administration (Ͼ30,000 patterns in relationship to hemoglobin levels and U/wk) and mean hemoglobin levels, we deter- epoetin alfa doses. Because these were retrospec- mined that only 0.6% (n ϭ 187) of 30,500 tive analyses, it is acknowledged that a causal hemodialysis patients administered epoetin alfa relationship between epoetin alfa dosing patterns between July and December 2001 had mean and hemoglobin levels cannot be ensured. hemoglobin levels persistently less than 11.0 We found that 27% of FMC-NA patients had g/dL (Ͻ110 g/L) and were administered persis- average hemoglobin levels less than 11.0 g/dL tently high doses of epoetin alfa. Of these 187 (Ͻ110 g/L), a level less than the lower boundary patients, 113 patients (Ͻ0.4% of the total) failed of the NKF-K/DOQI–recommended hemoglobin to respond with at least a 0.5-g/dL (5-g/L) in- target range. Similar findings were described in crease in mean hemoglobin levels after 6 months. the CMS 2003 Annual Report of the ESRD CPM A longitudinal cohort analysis of the FMC-NA Project. According to this report, the percentage data set was performed to examine practice pat- of adult in-center hemodialysis patients with terns in a subset patient cohort with persistent mean 3-month hemoglobin values less than 11.0 hemoglobin levels greater than 12.0 g/dL (Ͼ120 g/dL (110 g/L) decreased from 57% to 21% g/L; n ϭ 1,660). Action was taken to decrease between 1997 and 2002.23 During this same the epoetin alfa dose in 1,640 patients (98.8%); period, the CMS reported that mean hemoglobin approximately 52.3% (n ϭ 868) had at least 1 values increased from 10.7 to 11.9 g/dL (107 to dose withheld, 46.5% (n ϭ 772) had at least 1 119 g/L), a value also in agreement with our dose reduction of 10% or greater without a findings. withheld dose, and 1.2% (n ϭ 20) had no action Several factors may contribute to the high taken. percentage of patients with an average hemoglo- Thus, only a very small percentage of patients bin level less than 11.0 g/dL (Ͻ110 g/L). One with hemoglobin values persistently less than the factor is that incident hemodialysis patients have NKF-K/DOQI target range did not respond to pretreatment hemoglobin levels that are low and high epoetin alfa doses. In addition, only a very skew the distribution in that direction. An addi- 486 COLLINS ET AL tional factor is heterogeneity in the quality of than and greater than the NKF-K/DOQI target care at US dialysis facilities. The CMS reported range when measured at any given point in time. that the percentage of adult in-center hemodialy- Our analysis of serial hemoglobin values us- sis patients with mean hemoglobin values of 11.0 ing the FMC-NA database shows that only a g/dL or greater (Ն110 g/L) ranged from 73% to relatively small percentage of hemoglobin mea- 81% across dialysis networks during October to surements were consistently within the NKF-K/ December 2001.23 In the more complete Medi- DOQI target range. Patients cycled in and out of care sample, heterogeneity of care also was iden- the 11.0-to-12.0-g/dL (110-to-120-g/L) hemoglo- tified; 67% to 81% of patients achieved the lower bin range, perhaps as a consequence of the defi- boundary of the NKF-K/DOQI target range.24 nition of the range, as well as biological variabil- Although 78% of dialysis facilities, acting under ity. As hemoglobin level increased, epoetin alfa a physician’s prescription, achieve a mean facil- dose was decreased; and as hemoglobin level ity-level hemoglobin value between 11.0 and decreased, epoetin alfa dose was increased. Thus, 12.5 g/dL (110 and 125 g/L), only 5% of facili- at least for FMC-NA patients, epoetin alfa doses ties have 80% of patients within this range, and appear to be adjusted to maintain hemoglobin 77% of facilities have less than 50% of patients levels within the NKF-K/DOQI target range. It within this range (A. Collins, personal communi- may not be possible to extrapolate from the cation, December 2004). A 2003 General Ac- FMC-NA results to the practice patterns of other counting Office report concluded that approxi- providers in managing epoetin alfa dosing in mately half the dialysis facilities (assumed, but response to hemoglobin levels. not stated as occurring under a physician’s direc- It is noteworthy that 17% of patients from the tion) failed to meet the goal for anemia manage- FMC-NA data set had persistent hemoglobin ment (hemoglobin levels of at least 11.0 g/dL levels greater than 12.0 g/dL (Ͼ120 g/L). An [110 g/L] for Ն 80% of patients).25 examination of physician practice patterns in a Another important factor influencing the distri- subset of those patients showed that epoetin alfa bution of hemoglobin levels is intrapatient hemo- doses were being adjusted appropriately in 98.8% globin level variability. Hemoglobin level vari- of patients. Maintenance of hemoglobin levels ability is a function of numerous factors, including greater than 12.0 g/dL (Ͼ120 g/L) may be medi- appropriateness of physician orders for a dose cally justified. For example, some patients with change in response to a hemoglobin level, biologi- ischemic heart disease may experience angina cal diversity within the patient population, and pectoris with lower hemoglobin levels. variability in patient responsiveness to epoetin Our longitudinal cohort analysis of the CMS alfa, including individual differences in patient data set suggested that epoetin alfa dosing deci- pharmacokinetic response to treatment, ad- sions were based on the NKF-K/DOQI target equacy of iron stores, use of catheters for vascu- range. Patients with hemoglobin levels greater lar access, inflammatory and infectious disease than the 12.5-g/dL (125-g/L) upper boundary processes, hemodialysis adequacy, and gastroin- when the 6-month follow-up period began were testinal bleeding.26 Interestingly, despite the in- brought into range by the end of 6 months, as crease in mean hemoglobin levels during the past were patients with hemoglobin levels that ini- several years, variability around the mean, ex- tially were less than the 11.0-g/dL (110-g/L) pressed by the SD, did not change between 1997 lower boundary. During this 6-month period, and 2001.27 Because of this inherent intrapatient epoetin alfa doses appeared to be appropriately variability, Lacson et al28 reported that a typical titrated to manage patient hemoglobin levels. patient with a mean 3-month rolling average Although practice guidelines have an important hemoglobin level of 11.5 g/dL (115 g/L) and an influence on dosing decisions, it must be acknowl- SD within the median could be expected to have edged that physicians use their own individual 42% of individual 3-month rolling average hemo- treatment algorithms, which appear to result in globin values outside the NKF-K/DOQI target appropriate management of anemia. range. Therefore, it is not surprising that cross- As expected, we found an inverse relationship sectional analyses show a relatively high percent- between mean weekly epoetin alfa dose and age of patients with hemoglobin levels both less hemoglobin level when comparing the lowest EPOETIN ALFA PRESCRIBING PATTERNS IN ESRD 487 and highest hemoglobin level categories (Ͻ11.0 In our analyses, less than 0.4% of patients g/dL [Ͻ110 g/L] and Ͼ12.0 g/dL [Ͼ120 g/L], administered epoetin alfa persistently had hemo- respectively). A similar inverse relationship was globin levels less than 11.0 g/dL (Ͻ110 g/L) and reported previously.15,29 Patients who are se- persistently were administered high-dose epoetin verely anemic are treated with increases in epo- alfa without achieving at least a 0.5-g/dL (5-g/L) etin alfa doses to achieve the target hemoglobin increase in hemoglobin levels at the end of the level, which partially explains this relationship. 6-month period. Persistency of high-dose epo- Another contributing factor is that patients with etin alfa treatment and its relationship to clinical greater hemoglobin levels often have doses that outcomes is an important topic in the nephrology are decreased or withheld. A final factor is that community. An index that correlates persistency patients who are hyporesponsive to epoetin alfa with outcomes would be a very useful research and consequently have lower hemoglobin levels tool. are administered greater epoetin alfa doses to Longitudinal analysis of patient data appears enhance their ability to attain the NKF-K/DOQI to provide a more dynamic assessment of the target range. management of anemia than the cross-sectional Epoetin alfa hyporesponsiveness is defined in approach. Our longitudinal analyses of the rela- the NKF-K/DOQI guidelines as failure to achieve tionship between hemoglobin levels, epoetin alfa target hemoglobin level in the presence of ad- dosing patterns, and epoetin alfa doses show that equate iron stores. In addition to inadequate physicians in these facilities appropriately man- maintenance of iron stores, causes of epoetin alfa aged anemia in patients with ESRD. Hemoglo- hyporesponsiveness include infection and inflam- bin outcomes are a key determinant of epoetin mation, chronic blood loss, and malnutrition. alfa use. Persistency rate analysis showed that Inflammatory mediators associated with infec- only a very small percentage of patients was tion suppress erythropoietin production and re- administered high-dose epoetin alfa therapy with- duce sensitivity to epoetin alfa.22,30,31 A modifi- out improvement in hemoglobin levels. Thus, in able treatment factor that increases the risk for following the practice guidelines, these physi- infection-related hospitalization and has been cians increased the consistency and quality of associated with high epoetin alfa doses is the use care given to patients with ESRD, thus acting in of tunneled catheters for vascular access instead the public interest. It is important to better under- of arteriovenous fistulae.32,33 Between 1998 and stand the underlying causes of epoetin alfa hypo- 2001, catheter use in the United States in hemo- responsiveness in this patient population to maxi- dialysis patients increased from 19% to 26%.23 mize the benefits of patient care and make the Catheter use was found in 35% of patients admin- most efficient use of epoetin alfa. istered high doses of epoetin alfa (Ͼ36,000 U/wk) in a prevalent cross-section of the US Dialysis ACKNOWLEDGMENT Outcomes and Practice Patterns Study II sample The authors thank Bill Karbon, PhD, and Michele (based mainly on data for 2002; P.J. Held, per- Vivirito, Amgen Inc, Thousand Oaks, CA, for editorial sonal communication, December 2004). It should assistance. be noted that patients with comorbid illnesses, including those with severe atherosclerotic car- REFERENCES diovascular disease, may be poor candidates for 1. Astor BC, Muntner P, Levin A, Eusatce JA, Coresh J: arteriovenous fistulae. These patients may be Association of kidney function with anemia: The Third hyporesponsive to epoetin alfa because of their National Health and Nutrition Examination Survey (1988- 1994). Arch Intern Med 162:1401-1408, 2002 illness and not because of catheter use. It also 2. National Kidney Foundation: K/DOQI Clinical Prac- should be noted that many of the factors involved tice Guidelines for Chronic Kidney Disease: Evaluation, in hyporesponsiveness to epoetin alfa, once modi- classification, and stratification. Am J Kidney Dis 39:S1- fied, become tools in anemia management. Main- S266, 2002 (suppl 1) tenance of adequate iron stores, treatment of 3. Levin A, Thompson CR, Ethier J, et al: Left ventricular mass index in early renal disease: Impact of decline in comorbidities, and use of fistulae for vascular hemoglobin. Am J Kidney Dis 34:125-134, 1999 access instead of catheters can all be expected to 4. US Renal Data System: USRDS 2004 Annual Data improve hemoglobin levels. Report. The National Institutes of Health, National Institute 488 COLLINS ET AL of Diabetes and Digestive and Kidney Diseases, Bethesda, dialysis and dialysis patients. Nephrol Dial Transplant MD, 2004 18:353-361, 2003 5. Braumann KM, Nonnast-Daniel B, Böning D, Böcker 19. Muirhead N, Laupacis A, Wong C: Erythropoietin for A, Frei U: Improved physical performance after treatment of anaemia in haemodialysis patients: Results of a maintenance renal anemia with recombinant human erythropoietin. study. Nephrol Dial Transplant 7:811-816, 1992 Nephron 58:129-134, 1991 20. Eschbach JW, Abdulhadi MH, Browne JK, et al: 6. Horina JH, Schwaberger G, Brussee H, Sauseng- Recombinant human erythropoietin in anemic patients with Fellegger G, Holzer H, Krejs GJ: Increased red cell 2,3- end-stage renal disease. Ann Intern Med 111:992-1000, diphosphoglycerate levels in haemodialysis patients treated 1989 with erythropoietin. Nephrol Dial Transplant 8:1219-1222, 21. Pisoni RL, Bragg-Gresham JL, Young EW, et al: 1993 Anemia management outcomes from 12 countries in the 7. Robertson HT, Haley NR, Guthrie M, Cardenas D, Dialysis Outcomes and Practice Patterns Study (DOPPS). Eschbach JW, Adamson JW: Recombinant erythropoietin Am J Kidney Dis 44:94-111, 2004 improves exercise capacity in anemic hemodialysis patients. 22. National Kidney Foundation: K/DOQI Clinical Practice Am J Kidney Dis 15:325-332, 1990 Guidelines for Anemia of Chronic Kidney Disease: Update 8. Marsh JT, Brown WS, Wolcott D, et al: rHuEPO 2000. Am J Kidney Dis 37:S182-S238, 2001 (suppl 1) treatment improves brain and cognitive function of anemic 23. Centers for Medicare and Medicaid Services: 2003 dialysis patients. Kidney Int 39:155-163, 1991 Annual Report, End Stage Renal Disease Clinical Perfor- 9. Harnett JD, Parfrey PS: Cardiac disease in uremia. mance Measures Project. Baltimore, MD, Department of Semin Nephrol 14:245-252, 1994 Health and Human Services, Centers for Medicare and 10. Macdougall IC, Lewis NP, Saunders MJ, et al: Long- Medicaid Services, Center for Beneficiary Choices, 2003 term cardiorespiratory effects of amelioration of renal anae- 24. Collins AJ, Roberts TL, St Peter WL, Chen S-C, Ebben J, Constantini E: United States Renal Data System mia by erythropoietin. Lancet 335:489-493, 1990 assessment of the impact of the National Kidney Foundation– 11. Pascual J, Teruel JL, Moya JL, Liaño F, Jiménez- Dialysis Outcomes Quality Initiative guidelines. Am J Kid- Mena M, Ortuño J: Regression of left ventricular hypertro- ney Dis 39:784-795, 2002 phy after partial correction of anemia with erythropoietin in 25. US General Accounting Office: Dialysis Facilities: patients on hemodialysis: A prospective study. Clin Nephrol Problems Remain in Ensuring Compliance With Medicare 35:280-287, 1991 Quality. Washington, DC, October GAO-04-63:1-57, 2003 12. Wizemann V, Schäfer R, Kramer W: Follow-up of 26. Berns JS, Elzein H, Lynn RI, Fishbane S, Meisels IS, cardiac changes induced by anemia compensation in normo- DeOreo PB: Hemoglobin variability in epoetin-treated hemo- tensive hemodialysis patients with left-ventricular hypertro- dialysis patients. Kidney Int 64:1514-1521, 2003 phy. Nephron 64:202-206, 1993 27. Wish JB: Can evidence drive the development of a 13. Collins AJ, Li S, St Peter W, et al: Death, hospitaliza- sound national EPO reimbursement policy? Am J Kidney tion, and economic associations among incident hemodialy- Dis 41:254-258, 2003 sis patients with hematocrit values of 36% to 39%. J Am Soc 28. Lacson E, Ofsthun N, Lazarus JM: Effect of variabil- Nephrol 12:2465-2473, 2001 ity in anemia management on hemoglobin outcomes in 14. Ma JZ, Ebben J, Xia H, Collins AJ: Hematocrit level ESRD. Am J Kidney Dis 41:111-124, 2003 and associated mortality in hemodialysis patients. J Am Soc 29. Reddan DN, Frankenfield DL, Klassen PS, et al: Nephrol 10:610-619, 1999 Regional variability in anaemia management and haemoglo- 15. Ofsthun N, LaBrecque J, Lacson E, Keen M, Lazarus bin in the US. Nephrol Dial Transplant 18:147-152, 2003 JM: The effects of higher hemoglobin levels on mortality 30. Eckardt K-U: Anaemia of critical illness—Implica- and hospitalization in hemodialysis patients. Kidney Int tions for understanding and treating rHuEPO resistance. 63:1908-1914, 2003 Nephrol Dial Transplant 17:S48-S55, 2002 (suppl 5) 16. Xia H, Ebben J, Ma JZ, Collins AJ: Hematocrit levels 31. Macdougall IC, Cooper AC: Erythropoietin resis- and hospitalization risks in hemodialysis patients. J Am Soc tance: The role of inflammation and pro-inflammatory cyto- Nephrol 10:1309-1316, 1999 kines. Nephrol Dial Transplant 17:S39-S43, 2002 (suppl 11) 17. Evans RW, Rader B, Manninen DL, et al: The quality 32. Pisoni RL, Young EW, Dykstra DM, et al: Vascular of life of hemodialysis recipients treated with recombinant access use in Europe and the United States: Results from the human erythropoietin. JAMA 263:825-830, 1990 DOPPS. Kidney Int 61:305-316, 2002 18. Furuland H, Linde T, Ahlmén J, Christensson A, 33. Greenwood RN, Ronco C, Gastaldon F, et al: Erythro- Strömbom U, Danielson BG: A randomized controlled trial poietin dose variation in different countries and its relation- of haemoglobin normalization with epoetin alfa in pre- ship to drug resistance. Kidney Int Suppl 87:S78-S86, 2003 Association of Morbid Obesity and Weight Change Over Time With Cardiovascular Survival in Hemodialysis Population

Kamyar Kalantar-Zadeh, MD, PhD, MPH, Joel D. Kopple, MD, Ryan D. Kilpatrick, MS, Charles J. McAllister, MD, Christian S. Shinaberger, MPH, David W. Gjertson, PhD, and Sander Greenland, DrPH

● Background: In maintenance hemodialysis (MHD) outpatients, a reverse epidemiology is described, ie, baseline obesity appears paradoxically associated with improved survival. However, the association between changes in weight over time and prospective mortality is not known. Methods: Using time-dependent Cox models and adjusting for changes in laboratory values over time, the relation of quarterly-varying 3-month averaged body mass index (BMI) to all-cause and cardiovascular mortality was examined in a 2-year cohort of 54,535 MHD patients from virtually all DaVita dialysis clinics in the United States. Results: Patients, aged 61.7 ؎ 15.5 (SD) years, included 54% men and 45% with diabetes. Time-dependent unadjusted and multivariate-adjusted models, based on quarterly- averaged BMI controlled for case-mix and available time-varying laboratory surrogates of nutritional status, were calculated in 11 categories of BMI. Obesity, including morbid obesity, was associated with better survival and reduced cardiovascular death, even after accounting for changes in BMI and laboratory values over time. Survival advantages of obesity were maintained for dichotomized BMI cutoff values of 25, 30, and 35 kg/m2 across almost all strata of age, race, sex, dialysis dose, protein intake, and serum albumin level. Examining the regression slope of change in weight over time, progressively worsening weight loss was associated with poor survival, whereas weight gain showed a tendency toward decreased cardiovascular death. Conclusion: Weight gain and both baseline and time-varying obesity may be associated with reduced cardiovascular mortality in MHD patients independent of laboratory surrogates of nutritional status and their changes over time. Morbidly obese patients have the lowest mortality. Clinical trials need to verify these observational findings. Am J Kidney Dis 46:489-500. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Hemodialysis (HD); obesity; weight loss; cardiovascular death; malnutrition; inflammation; time- varying model.

F MORE THAN 300,000 maintenance he- patients randomly assigned to either atorvastatin, O modialysis (MHD) patients in the United 20 mg, or placebo did not show a significant States, at least 60,000 patients, or 20%, die every advantage of using statins in improving survival year, mostly because of cardiovascular disease in dialysis patients with diabetes.3 In addition, (CVD).1,2 Despite decades of ongoing efforts on several other recent multicenter clinical trials, treating such conventional cardiovascular risk including the Hemodialysis (HEMO) Study,4 factors in MHD patients as obesity, hyperten- failed to show a survival advantage of increasing sion, hypercholesterolemia, and hyperhomocys- dialysis dose or using different dialysis mem- teinemia, their survival has not improved substan- branes in MHD patients. tially.2 The recently completed Die Deutsche A number of epidemiological studies with Diabetes Dialyse Studie (4D) of 1,255 dialysis large sample sizes indicated paradoxically in-

From the Division of Nephrology and Hypertension, Los R.D.K. contributed to analysis of the data and reviewed and Angeles Biomedical Institute at Harbor-UCLA Medical Cen- approved the final manuscript; C.J.M. contributed to the ter, Torrance; School of Public Health and Department of design of the study, provision of data, and final review and Statistics, University of California, Los Angeles; and DaVita, approval of the manuscript; C.S.S. contributed to analysis of Inc, El Segundo, CA. the data and reviewed and approved the final manuscript; Received March 8, 2005; accepted in revised form May 9, D.W.G. contributed to the design of the study and analysis of 2005. data and reviewed and approved the final manuscript; S.G. Originally published online as doi:10.1053/j.ajkd.2005.05.020 contributed to the design of the study, provided advice and on July 15, 2005. consultation on the study design, analysis, and writing of the Supported in part by a Young Investigator Award from the manuscript, and final review and approval. National Kidney Foundation and grant no. DK61162 (K.K.-Z.) Address reprint requests to Kamyar Kalantar-Zadeh, MD, from the National Institute of Diabetes, and Digestive and PhD, MPH, Associate Professor of Medicine and Pediatrics, Kidney Diseases. C.J.M. is an employee of DaVita, Inc. Division of Nephrology and Hypertension, Harbor-UCLA K.K.-Z. contributed to the design of the study, collation Medical Center, Harbor C1-Annex, 1000 W Carson St, and analysis of data, and writing of the manuscript and its Torrance, CA 90509-2910. E-mail: [email protected] revisions; J.D.K. contributed to the design and analysis of © 2005 by the National Kidney Foundation, Inc. the study and provided advice and consultation on the 0272-6386/05/4603-0013$30.00/0 writing of the manuscript and final review and approval; doi:10.1053/j.ajkd.2005.05.020

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 489-500 489 490 KALANTAR-ZADEH ET AL verse associations between classic risk factors Body Size Measures for CVD and death in MHD patients. For ex- Eleven categories of BMI9 (weight/height2) were used, ample, a high mortality rate has been observed in less than 18 kg/m2, 45 kg/m2 or greater, and 9 categories MHD patients who have low, rather than high, from 18 to 44.99 kg/m2. Selected category boundaries in- 5,6 7 cluded the cutoff levels defined by the World Health Organi- body mass index (BMI), blood pressure, and 13 8 zation classifications, including overweight (25 to 29.99 serum concentrations of cholesterol and homo- 2 2 9 kg/m ), obese (30 to 39.99 kg/m ), and morbidly obese cysteine. High values of these risk factors para- (Ͼ40 kg/m2 without any obesity sequelae or Ͼ35 kg/m2 and doxically are protective and associated with im- experiencing a severe negative health effect of obesity). In proved survival.10,11 The inverse relationship our study, underweight is defined as a BMI less than 20 2 between such indices of body size as BMI and kg/m , and normal weight range, as a BMI between 20 and 24.99 kg/m2. Patients with missing weight or height values clinical outcome in MHD patients appears to be in all 8 quarters or with values less than 10 kg/m2 or greater relatively consistent and has been studied exten- than 60 kg/m2 (corresponding to ϳ0.25th and 99.75th per- sively.6 Moreover, other patient populations, in- centile levels) were excluded. cluding patients with heart failure, appear to have this so-called “obesity paradox.”12 How- Cohort Time, Vintage, and Mortality ever, most investigators studied the effect of Cohort time includes the number of days the patient baseline BMI at the start of the cohort on subse- participated in the cohort and was a number between 1 and quent mortality. Virtually no study has examined 731 days. Left-censorship events include death, renal trans- the association between change in weight over plantation, or moving to a non-DaVita dialysis facility, among others. Change to peritoneal dialysis therapy is not time on cardiovascular mortality in the entire considered a left-censorship event based on the intent-to- cohort or its subpopulations. No study has con- treat principle. Dialysis vintage is defined as time between trolled for time-varying laboratory measures, in- the first day of dialysis treatment and the first day the patient cluding the available surrogates of nutritional entered the cohort. Four categories of vintage were formed: status, when exploring the obesity paradox. More- (1) first 6 months, (2) between 6 and 24 months, (3) between 2 and 5 years, and (4) greater than 5 years. The entry quarter over, it is not clear whether morbid obesity has is defined as the first quarter in which a patient’s dialysis the same survival advantages as moderate obe- vintage was greater than 3 months for at least half the sity in MHD patients. We examined whether duration of the quarter. By implementing this criterion, any survival advantages of obesity, including morbid patient who was not maintained in the cohort beyond the first obesity, maintain even with changes in weight 3 months of MHD therapy was excluded. The computerized causes of death, reflecting, but not restricted to, the reported over time and whether these associations are information in the Cause of Death form (Form 2746), were independent of other nutritional measures and obtained and summarized into 6 main categories: cardiovas- are observed uniformly across all subgroups of cular, infectious, gastrointestinal, cancer related, others, and MHD patients. We also examined whether a unspecified/unknown. decrease in weight over time is associated with increased cardiovascular death, and whether Laboratory Data and Indicators of weight gain confers improved survival, indepen- Malnutrition-Inflammation Complex Syndrome dent of other nutritional factors. Most blood samples were predialysis, with the exception of postdialysis serum urea nitrogen to calculate urea kinet- METHODS ics. Blood samples were drawn by using uniform techniques in all dialysis clinics across the nation and transported to the Database Creation Central DaVita Laboratory in Deland, FL, within 24 hours. Database creation has been described elsewhere.7 In sum- All laboratory values were measured by using automated mary, the data warehouse of DaVita, Inc, the then second and standardized methods in DaVita Laboratory. Most labo- largest dialysis care provider in the United States, had ratory values, including complete blood cell counts; serum information on approximately 40,000 patients at any given levels of urea nitrogen, albumin, creatinine, and ferritin; and time and was accessed to obtain all relevant data between total iron-binding capacity (TIBC), were measured monthly. July 1, 2001, and June 30, 2003. All repeated measures of Serum ferritin was measured quarterly. Hemoglobin was every relevant variable for each patient within a given measured weekly to biweekly in most patients. Kt/V to quarter (13-week interval) were averaged to obtain 1 quar- reflect dialysis dose and normalized protein equivalent of terly mean value for that variable. Hence, 8 independent sets total nitrogen appearance (nPNA), also known as protein of quarterly database were created to include the entire catabolic rate, an estimation of daily protein intake, were 2-year observation period. The study was approved by the measured monthly. Instructional Review Committees of Harbor-UCLA and Eight quarterly-varying 13-week averaged laboratory mea- DaVita, Inc. sures with known associations with prospective mortality WEIGHT CHANGE AND SURVIVAL IN DIALYSIS 491 were selected, some of which are considered potential surro- of existing values. The rate of weight change over time for gates of nutritional status, with some having a possible each patient was calculated as the slope (␤) of the regression association with inflammation14: (1) serum albumin15; (2) model that included up to 8 quarterly BMI values. For this nPNA as a marker of daily protein intake16; (3) serum TIBC, model, patients with fewer than 2 BMI values were ex- known to have a strong association with subjective global cluded. All descriptive and multivariate statistics were car- assessment of nutrition17; (4) serum ferritin, a possible ried out using SAS, version 8.02 (SAS Institute Inc, Cary, 18 inflammatory marker ; (5) serum creatinine, a marker of NC); most models were repeated using Stata, version 7.0 19 muscle mass ; (6) peripheral white blood cell count (WBC), (Stata Corp, College Station, TX) for consistency verifica- 20 known to correlate with serum C-reactive protein level ; (7) tion. lymphocyte percentage, a known nutritional marker21,22; and (8) blood hemoglobin level.23 RESULTS Epidemiological and Statistical Methods The original 2-year national database for all Because the dialysis population is a dynamic cohort with MHD patients included 69,819 subjects. After a high turnover rate, a nonconcurrent cohort was formed to include all existing MHD patients of the first quarter (q1) implementing the mentioned selection criteria, and all new MHD patients of the subsequent quarters (q2 including deleting patients who were not main- through q8). Hence, 8 quarterly data sets were merged by tained beyond 3 months of MHD therapy or who using unique patient identifiers. For patients not linked to the had inadequate or overtly missing data, the result- database initially, additional merging methods using the ing cohort included 58,058 MHD patients, of initial letters of the patient’s last and first names combined with his or her date of birth were performed. In addition to 8 whom 37,049 patients (64%) originated from the quarterly values for every variable, a baseline value also was q1 data set, and the rest, from the subsequent created for each measure by left-truncating the first available quarters (q2 through q8). Table 1 lists baseline value of the entry quarter for each patient. demographic, clinical, and laboratory character- Because of the very large numbers involved, all associa- tions we discuss have very low values for P. In addition to istics of the cohort. Almost 45% of patients had standard descriptive statistics, both regular and time- diabetes mellitus, and 40% had been on dialysis dependent Cox proportional hazard regression analyses for therapy for less than 6 months on entry to the 24 truncated and censored data were performed to determine cohort. Cardiovascular death comprised more whether 2-year survival was associated with baseline and time-varying categories of BMI. The reference BMI cat- than half of all known causes of death. egory for all analyses was 23 to 24.99 kg/m2. This category BMI was missing or an acceptable BMI value was chosen as the reference because it is within the normal was not available in 3,523 patients (6.1%). Table range of BMI set forth by the World Health Organization,13 2 lists the 11 BMI categories for the remaining was adjacent to the modal category and had almost the same 54,535 MHD patients. Only 13.1% of patients sample size, and had the highest number of death cases and allowed for the most precise comparison with higher BMI were underweight at baseline, whereas more categories. Five race/ethnic groups were generated: (1) Cau- than a third (37.1%) were within normal range. casians (including non-Hispanic whites and Middle Eastern- Overweight, obese, and morbidly obese MHD ers), (2) blacks (including African Americans and other patients were 28.9%, 18.5%, and 7.7%, respec- Africans), (3) Asians (including Pacific Islanders), (4) Ameri- tively; morbid obesity is defined as a BMI greater can Indians, (5) Hispanics, and (6) others. 2 For each analysis, 3 models were examined based on the than 40 kg/m . Both all-cause and cardiovascular level of multivariate adjustment: (1) the unadjusted model mortality showed almost strictly decreasing rates included only BMI categories and mortality data; (2) case- across increasing BMI categories, ie, morbidly mix–adjusted models included age, sex, race and ethnicity, obese MHD patients had the greatest survival diabetes mellitus, vintage categories, entry quarter, primary insurance (Medicare, Medicaid, private, and others), mar- rates. riage status (married, single, divorced, widowed, and others), Figures 1 and 2 show hazard ratios for all- standardized mortality ratio of the dialysis clinic during the cause and cardiovascular death for different time- entry quarter, Kt/V (single pool), and residual renal function varying BMI categories, respectively. Obesity, during the entry quarter); and (3) case-mix and laboratory– adjusted models included all mentioned covariates, as well including morbid obesity, was associated with as 8 surrogates of nutritional status, including nPNA, serum improved survival and decreased cardiovascular albumin level, TIBC, ferritin level, creatinine level, WBC mortality, even after exhaustive adjustment for count, lymphocyte percentage, and hemoglobin level. time-varying laboratory markers. These associa- In time-dependent models, in addition to time-varying quarterly BMI categories, 8 laboratory measures and Kt/V tions were independent of changes in BMI over also were entered as quarterly time-varying variables. Miss- time. Non–time-dependent Cox regression mod- ing covariate data (Ͻ5%) were input by the mean or median els also were performed by using baseline BMI 492 KALANTAR-ZADEH ET AL

Table 1. Baseline Data for the Nonconcurrent death in all patients and in different strata of (left-truncated) Cohort of 58,058 MHD Patients, diabetes status, sex, age (Ͼ65 years versus Including 37,049 From q1 and 21,009 From q2 to q8 younger), dialysis vintage, serum albumin levels Variable (Ͼ3.8 g/dL [Ͼ38 g/L] versus lower), and dietary protein intake (Ͼ1.0 g/kg/d versus lower). Ex- Age (y) 60.7 Ϯ 15.5 cept for the adjusted hazard ratio in Asian pa- Age Ͼ 65 y (%) 43.8 Sex (% women) 46.2 tients, high BMI was associated with decreased Diabetes mellitus (%) 44.7 all-cause and cardiovascular mortality. Non–time- Race and ethnicity: dependent Cox models using baseline BMI gave Caucasians (%) 36.8 essentially the same results (not shown). Blacks (%) 32.0 In 47,629 MHD patients who had at least 2 Asians (%) 4.3 American Indians (%) 2.2 quarterly BMI values, the rate of weight change Hispanics (%) 15.2 per annual quarter was estimated as a proportion Vintage (time on dialysis): of baseline weight and based on the slope of the 3-6 mo (%) 40.1 linear regression equation for each patient. Table 6-24 mo (%) 21.9 3 lists the selected weight gain categories for this 2-5 y (%) 24.1 ϭ Ͼ5 y (%) 13.8 analysis. Almost half of all patients (n 22,784) Primary insurance had stable weight, ie, a weight gain or loss less Medicare (%) 69.9 than 1% per quarter (reference group). Five Medicaid (%) 6.4 weight-gain and 5 weight-loss categories were Private provider (%) 23.7 created based on increments of 1% weight change Causes of death: Cardiovascular (%) 51.6 rate per quarter. The stable (reference) group had Infectious (%) 10.9 one of the lowest unadjusted mortality rates and Cancer (%) 2.9 the highest mean serum albumin concentration. Gastrointestinal (%) 1.6 Mean baseline BMI was the lowest in MHD Others (%) 8.2 patients with the greatest rate of weight gain; Unknown/unspecified (%) 24.8 Standardized mortality ratio 0.81 Ϯ 0.31 they also had a greater mortality rate than the Cohort time (mo) 14.8 Ϯ 8.5 stable group, although still substantially less than Posthemodialysis weight (kg) 73.8 Ϯ 19.8 that in the group with the greatest rate of weight BMI (kg/m2) 26.1 Ϯ 6.3 loss. Kt/V (single pool) 1.54 Ϯ 0.32 To examine whether the differences in base- nPNA (g/kg/d) 0.99 Ϯ 0.25 Serum albumin (g/dL) 3.74 Ϯ 0.41 line BMI and serum albumin levels could ac- Serum creatinine (mg/dL) 9.0 Ϯ 3.4 count for greater mortality rates across the weight- Serum ferritin (ng/mL) 609 Ϯ 521 gain groups, Cox regression models with different TIBC (mg/dL) 202 Ϯ 32 multivariate adjustments were studied. Table 4 Blood hemoglobin (g/dL) 11.98 Ϯ 1.28 and Figs 4 and 5 show increases in the adjusted WBCs (per fl) 7,326 Ϯ 2,429 Lymphocytes (% of total WBCs) 20.9 Ϯ 7.9 hazard ratio of death across decrements of weight- loss groups, whereas a tendency toward better NOTE. Values expressed as mean Ϯ SD or percent. To survival and reduced cardiovascular death is ob- convert serum creatinine in mg/dL to ␮mol/L, multiply by served across increments of weight-gain groups. 88.4; albumin and hemoglobin in g/dL to g/L, multiply by Among relevant interactions examined, the statis- 10; ferritin in ng/mL to ␮g/L, multiply by 1. Abbreviation: nPNA, normalized protein catabolic rate. tical interaction between baseline BMI and weight change did not have significant effect on mortal- categories at the start of the cohort. Results were ity (P ϭ 0.67). similar (not shown). To explore the impact of several relevant cut- DISCUSSION off levels of increased body weight on survival in In a national database cohort that included different subgroups of patients, continuous BMI all eligible MHD patients of a major dialysis data were dichotomized at BMI levels of 25, 30, care provider with uniform patient manage- and 35 kg/m2. Figure 3 shows unadjusted and ment practices and standardized laboratory val- fully adjusted hazard ratios for cardiovascular ues, all measured in 1 laboratory, we found WEIGHT CHANGE AND SURVIVAL IN DIALYSIS 493

Table 2. BMI Groups and 2-Year Mortality Census Among Them

World Health Organization All-Cause Cardiovascular Category BMI Range (kg/m2) Sample Size Death Death

Ͻ18 2,605 (4.8) 1,023 (39.3) 384 (15.9) Underweight 18-19.99 4,515 (8.3) 1,498 (32.2) 601 (14.1) 20-21.49 5,402 (9.9) 1,627 (30.1) 685 (13.4) Normal range 21.5-22.99 6,452 (11.8) 1,753 (27.2) 728 (11.8) 23-24.99 (reference) 8,383 (15.4) 2,095 (25.0) 926 (11.5) 25-27.49 8,860 (16.2) 2,045 (23.1) 911 (10.7) Overweight 27.5-29.99 6,385 (11.7) 1,326 (20.8) 580 (9.4) 30-34.99 7,028 (12.9) 1,399 (19.9) 657 (9.7) Obese 35-39.99 3,047 (5.6) 545 (17.9) 269 (9.1) 40-44.99 1,171 (2.1) 186 (15.9) 85 (7.5) Morbidly obese Ͼ45 687 (1.3) 112 (16.3) 47 (7.0) All patients 54,535 (100) 13,609 (25.0) 5,873 (11.2)

NOTE. Values expressed as number (percent) unless indicated otherwise. The denominator for all-cause mortality is the total sample of MHD patients with at least 1 baseline BMI value (n ϭ 54,535), whereas the denominator for cardiovascular death is 52,300 because in 2,235 MHD patients, the cause of death was not registered in the database. that obesity, including morbid obesity, is asso- trend toward improved survival and reduced ciated with improved survival and reduced cardiovascular mortality. These findings are in cardiovascular death. These associations were sharp contrast to the conventional understand- independent of changes in BMI during the ing of the epidemiology of obesity, especially 2-year study period and held even after adjust- morbid obesity, in the general population.25 ment for time-varying laboratory surrogates of In the United States, as well as in most indus- nutritional status. With the possible exception trialized nations, mean body weight is increas- of Asian MHD patients, survival advantages of ing.26 Many epidemiological studies have shown overweight, obesity, and morbid obesity held a strong association between obesity and de- in different patient subgroups. Weight loss was creased survival, especially because of an in- associated with increased cardiovascular and creased risk for CVD,27,28 although recent stud- all-cause death, whereas weight gain showed a ies have found a more reduced relative risk for

Fig 1. Time-dependent association between BMI and 2-year all-cause mortality in 54,535 MHD patients (95% confidence interval bars are not shown for the case-mix– adjusted group to enable bet- ter distinction of confidence intervals for other 2 groups). 494 KALANTAR-ZADEH ET AL

Fig 2. Time-dependent association between BMI and 2-year cardiovascular mortal- ity in 54,535 MHD patients (95% confidence interval bars are not shown for the case-mix–adjusted group to enable better distinction of the confidence intervals for other 2 groups). death associated with obesity than previously tients. Finally, Johansen et al39 recently showed reported.29,30 There is a direct relationship be- that markers of body fat or muscle mass are tween BMI and body weight, as well as body fat. associated with improved survival in obese MHD In some studies of healthy adults, a “J” curve patients, although this was not found by Beddhu effect was observed in which individuals with a et al,43 which may be caused by method flaws in low BMI also showed increased mortality, al- the latter study, discussed elsewhere.6,11 A few though not as high as obese individuals.28,31 This studies with much smaller sample sizes did not “J” shape may disappear when data are adjusted detect improved survival in obese MHD pa- for smoking status.28 tients,44,45 including a study by Kutner and In contrast to trends seen in the general popu- Zhang45 reporting that the association of race lation, in MHD patients, lower BMI at baseline and BMI with mortality differed by sex. How- consistently is found to be a strong predictor of ever, these studies did not examine changes in elevated mortality.6,32-38 Furthermore, greater weight over time or the rate of change and its BMI generally has not been associated with an relation to survival and cardiovascular death. increase in mortality risk in MHD patients, ex- The reverse epidemiology of obesity is not cept in Asian-American MHD patients.36,39 The unique to dialysis populations. Patients with Diaphane Collaborative Study Group32 was one chronic heart failure,12,46 geriatric populations,47 of the first to report on the paradoxical observa- hospitalized patients,48 and patients with malig- tion of low mortality with high baseline BMI in nancy49 or acquired immunodeficiency syn- 1,453 younger, mostly nondiabetic, French MHD drome50 also show a risk-factor reversal. Hence, patients followed up between 1972 and 1978.32 a better understanding of the phenomenon of Leavey et al,34 Fleischmann et al,35 Wolf et al,40 reverse epidemiology in dialysis patients, espe- and Port et al41 reported similar findings in larger cially as it pertains to obesity and body size, may MHD cohorts. Kopple et al42 found the same help improve the poor outcome in this and other trend for weight-for-height percentiles. The Di- similar, but distinct, populations and disease alysis Outcomes and Practice Patterns Study33 states, altogether more than 20 million Ameri- found similar associations among western Euro- cans.11 Moreover, the reverse epidemiology phe- pean MHD patients. Glanton et al38 found that nomenon is not restricted to obesity. Blood pres- the inverse relationship between BMI and mortal- sure and serum cholesterol, creatinine, and ity was stronger in African-American MHD pa- homocysteine levels also appear to have an in- WEIGHT CHANGE AND SURVIVAL IN DIALYSIS 495

Fig 3. Impact of (left) overweight (BMI > 25 kg/m2), (middle) obesity (BMI > 30 kg/m2), and (right) morbid obesity (BMI > 35 kg/m2) on 2-year survival in 54,535 MHD patients and their subgroups according to race, diabetes, sex, age, dialysis vintage, serum albumin level, and dietary protein intake. Analyses are based on time-dependent Cox regression models. Open circles, unadjusted hazard ratios; filled circles, multivariate-adjusted hazard ratios. To convert albumin in g/dL to g/L, multiply by 10. verse association with survival in these popula- versus undernutrition),10 and the overwhelming tions.7,9,51 effect of malnutrition inflammation complex on The concept of reverse epidemiology may traditional cardiovascular risks.14 Because most appear counterintuitive, especially because obe- MHD patients die within 5 years of starting sity, especially morbid obesity, is an established dialysis treatment,1 long-term effects of conven- risk factor for CVD and poor outcome in the tional risk factors on future mortality must be general population. Nonetheless, given the con- overwhelmed by the short-term effects of these sistency of the observations, there must be condi- or other risk factors intrinsic to dialysis popula- tions in these populations that render them more tions, such as undernutrition and inflammation. It susceptible to a poor outcome when low BMI is may be that dialysis patients do not live long present. Several explanations have been sug- enough to die of the consequences of overnutri- gested,6,52 including a more stable hemodynamic tion. status in obese individuals,53 greater concentra- Our study lacked data for history of cardiovas- tion of tumor necrosis factor ␣ receptors in cular comorbidity and smoking. However, diabe- obesity,54 neurohormonal alterations of obe- tes data were available and adjusted for in all sity,55 endotoxin-lipoprotein interaction,56 re- multivariate models. Moreover, many other co- verse causation,57 survival bias,10 time discrepan- variates included in the models have strong asso- cies among competitive risk factors (overnutrition ciations with comorbid conditions. Hence, we 496 KALANTAR-ZADEH ET AL

Table 3. Categories of Weight Change Percentage Over Time in 47,629 MHD Patients With at Least 2 Quarterly BMI

Weight Change All-Cause Mortality Baseline BMI Baseline Serum Category (%) Group Population (%) (k/m2) Albumin (g/dL)

Ͼϩ5 (weight gain) 1,038 (2.2) 29.2 22.3 Ϯ 6.2 3.61 Ϯ 0.44 ϩ4toϩ5 625 (1.3) 24.0 24.2 Ϯ 5.5 3.68 Ϯ 0.39 ϩ3toϩ4 1,319 (2.8) 21.6 24.5 Ϯ 5.7 3.70 Ϯ 0.39 ϩ2toϩ3 2,680 (5.6) 19.3 25.2 Ϯ 6.0 3.73 Ϯ 0.37 ϩ1toϩ2 5,510 (11.6) 17.4 25.7 Ϯ 5.9 3.78 Ϯ 0.36 –1 to ϩ1 (no change) 22,784 (47.8) 17.5 26.5 Ϯ 6.2 3.85 Ϯ 0.34 –2 to –1 6,189 (13.0) 27.1 26.8 Ϯ 6.3 3.78 Ϯ 0.37 –3 to –2 3,102 (6.5) 33.7 26.9 Ϯ 6.4 3.73 Ϯ 0.38 –4 to –2 1,649 (3.5) 43.2 27.2 Ϯ 6.4 3.66 Ϯ 0.40 –5 to –4 909 (1.9) 48.1 26.8 Ϯ 6.3 3.58 Ϯ 0.41 Ͻ–5 (weight loss) 1,824 (3.8) 53.7 26.4 Ϯ 6.7 3.49 Ϯ 0.47

NOTE. Values expressed as number (percent) or mean Ϯ SD unless noted otherwise. To convert albumin in g/dL to g/L, multiply by 10. suspect that the associations would not have dialysis initiation form (Form 2728), in which been very different if additional adjustments had comorbid conditions are significantly underre- been made for other comorbidity. Our adjust- ported,58 and that is outdated for prevalent pa- ments are not too different from those of past tients with higher vintage periods. Another limi- studies because the limited comorbidity data tation of our study is the lack of explicit laboratory used in those studies usually originated from the markers of inflammation, such as C-reactive pro-

Table 4. Hazard Ratio of Death for Categories of Weight Change Percentage Over Time

Unadjusted Case-Mix Adjusted Case-Mix & MICS Adjusted Weight Change Category (%) HR (95% CI) P HR (95% CI) P HR (95% CI) P

All-cause mortality Ͼϩ5 (weight gain) 1.28 (1.14-1.44) Ͻ0.001 1.01 (0.89-1.14) 0.897 0.89 (0.79-1.01) 0.068 ϩ4toϩ5 1.10 (0.93-1.30) 0.255 1.03 (0.87-1.21) 0.734 0.96 (0.81-1.13) 0.605 ϩ3toϩ4 1.04 (0.93-1.18) 0.478 0.98 (0.87-1.11) 0.750 0.90 (0.80-1.02) 0.094 ϩ2toϩ3 1.07 (0.97-1.17) 0.175 1.01 (0.92-1.11) 0.788 0.94 (0.85-1.03) 0.169 ϩ1toϩ2 0.97 (0.91-1.04) 0.421 0.98 (0.91-1.05) 0.521 0.95 (0.88-1.02) 0.124 –1 to ϩ1 (no change) 1.00 NA 1.00 NA 1.00 NA –2 to –1 1.35 (1.28-1.43) Ͻ0.001 1.28 (1.21-1.36) Ͻ0.001 1.21 (1.15-1.29) Ͻ0.001 –3 to –2 1.52 (1.42-1.63) Ͻ0.001 1.39 (1.30-1.49) Ͻ0.001 1.30 (1.22-1.40) Ͻ0.001 –4 to –2 1.77 (1.63-1.91) Ͻ0.001 1.59 (1.46-1.72) Ͻ0.001 1.39 (1.28-1.51) Ͻ0.001 –5 to –4 1.84 (1.67-2.04) Ͻ0.001 1.62 (1.47-1.79) Ͻ0.001 1.38 (1.24-1.52) Ͻ0.001 Ͻ–5 (weight loss) 2.06 (1.91-2.21) Ͻ0.001 1.81 (1.68-1.95) Ͻ0.001 1.50 (1.39-1.62) Ͻ0.001 Cardiovascular mortality Ͼϩ5 (weight gain) 1.20 (1.01-1.44) 0.043 0.96 (0.80-1.15) 0.645 0.86 (0.71-1.03) 0.101 ϩ4toϩ5 1.05 (0.81-1.34) 0.728 0.99 (0.77-1.28) 0.955 0.93 (0.72-1.19) 0.562 ϩ3toϩ4 0.95 (0.78-1.14) 0.558 0.90 (0.75-1.09) 0.289 0.84 (0.69-1.01) 0.064 ϩ2toϩ3 1.10 (0.97-1.26) 0.143 1.06 (0.93-1.21) 0.409 0.99 (0.87-1.13) 0.878 ϩ1toϩ2 0.97 (0.87-1.08) 0.556 0.98 (0.89-1.09) 0.766 0.95 (0.86-1.06) 0.361 –1 to ϩ1 (no change) 1.00 NA 1.00 NA 1.00 NA –2 to –1 1.29 (1.19-1.41) Ͻ0.001 1.23 (1.13-1.34) Ͻ0.001 1.18 (1.08-1.29) Ͻ0.001 –3 to –2 1.41 (1.27-1.57) Ͻ0.001 1.28 (1.15-1.42) Ͻ0.001 1.22 (1.10-1.36) Ͻ0.001 –4 to –2 1.51 (1.33-1.71) Ͻ0.001 1.34 (1.17-1.52) Ͻ0.001 1.21 (1.06-1.38) 0.004 –5 to –4 1.68 (1.44-1.96) Ͻ0.001 1.49 (1.28-1.75) Ͻ0.001 1.33 (1.14-1.56) Ͻ0.001 Ͻ–5 (weight loss) 1.90 (1.70-2.13) Ͻ0.001 1.64 (1.46-1.84) Ͻ0.001 1.44 (1.28-1.62) Ͻ0.001

Abbreviations: MICS, malnutrition-inflammation complex syndrome; HR, hazard ratio; CI, confidence interval; NA, not applicable. WEIGHT CHANGE AND SURVIVAL IN DIALYSIS 497

Fig 4. Association be- tween the rate of weight change over time and subse- quent all-cause mortality in 46,629 MHD patients (95% confidence interval bars are not shown for the case-mix– adjusted group to enable bet- ter distinction of confidence intervals for other 2 groups). tein level. However, we used data for serum dependent analysis is that it is based on 2-year albumin and ferritin, TIBC, and WBCs, which cross-sections of the cohort, rather than longer may have associations with inflammation. longitudinal follow-up over many years, and Among the strengths of our study is the use of thus may not apply to long-term survival of time-dependent models to examine the associa- individuals. Nonetheless, the narrow time win- tion between quarterly-varying weight on cardio- dow of our study ensures that confounding by vascular mortality while controlling for the time- changes in practice or technology is minimal. varying effect of laboratory surrogates of Our time-dependent findings are supported by nutritional status. A limitation of our time- the observed relations of different rates of weight

Fig 5. Association be- tween the rate of weight change over time and subse- quent cardiovascular mortal- ity in 46,629 MHD patients (95% confidence interval bars are not shown for the case-mix–adjusted group to enable better distinction of the confidence intervals for other 2 groups). 498 KALANTAR-ZADEH ET AL gain and weight loss over time with cardiovascu- tion: The 58th annual fall conference and scientific sessions. lar survival. Data originate from 1 dialysis care Hypertension 45:811-817, 2005 provider that has uniform patient management 8. Nishizawa Y, Shoji T, Ishimura E, Inaba M, Morii H: Paradox of risk factors for cardiovascular mortality in ure- practices; all laboratory measurements are per- mia: Is a higher cholesterol level better for atherosclerosis in formed in 1 facility, and most data are means of uremia? Am J Kidney Dis 38:S4-S7, 2001 (suppl 1) several measures. Hence, measurement variabil- 9. Kalantar-Zadeh K, Block G, Humphreys MH, McAllis- ity is minimized. ter CJ, Kopple JD: A low, rather than a high, total plasma homocysteine is an indicator of poor outcome in hemodialy- If causal, the inverse association of obesity sis patients. J Am Soc Nephrol 15:442-453, 2004 with survival would have major clinical and 10. Kalantar-Zadeh K, Block G, Humphreys MH, public health implications. Hence, it may be time Kopple JD: Reverse epidemiology of cardiovascular risk for clinical trials of weight-increasing interven- factors in maintenance dialysis patients. Kidney Int 63:793- tions in populations that show reverse epidemiol- 808, 2003 11. Kalantar-Zadeh K, Kilpatrick RD, Kuwae N, Wu DY: ogy. Currently, more than 60% of dialysis pa- Reverse epidemiology: A spurious hypothesis or a hardcore tients die within 5 years of starting dialysis reality? Blood Purif 23:57-63, 2005 treatment; thus, long-term consequences of such 12. Curtis JP, Selter JG, Wang Y, et al: The obesity interventions may be of secondary concern. How- paradox: Body mass index and outcomes in patients with ever, as more effective treatments for dialysis heart failure. Arch Intern Med 165:55-61, 2005 13. Joint WHO/FAO Expert Consultation on Diet Nutri- patents become available, these patients may live tion and the Prevention of Chronic Diseases: Diet, Nutrition long enough for the long-term adverse effects of and the Prevention of Chronic Diseases: Report of a Joint obesity to come back into play, as found in WHO/FAO Expert Consultation. Geneva, Switzerland, World kidney transplant recipients59; hence, so-called Health Organization, 2003 “reversal of the reverse epidemiology” or “back 14. Kalantar-Zadeh K, Ikizler TA, Block G, Avram MM, Kopple JD: Malnutrition-inflammation complex syndrome to normal.” However, because the causality of in dialysis patients: Causes and consequences. Am J Kidney our reported associations is not known, extreme Dis 42:864-881, 2003 caution needs to be practiced. Therefore, initial 15. Owen WF Jr, Lew NL, Liu Y, Lowrie EG, Lazarus trials would probably best be limited to patients JM: The urea reduction ratio and serum albumin concentra- with low initial BMIs (eg, BMI Ͻ 23 kg/m2), and tion as predictors of mortality in patients undergoing hemo- dialysis. N Engl J Med 329:1001-1006, 1993 even if weight gain is found to be beneficial for 16. Kalantar-Zadeh K, Supasyndh O, Lehn RS, McAllis- short-term survival, periodic reevaluation of its ter CJ, Kopple JD: Normalized protein nitrogen appearance net long-term effects will be needed. is correlated with hospitalization and mortality in hemodialy- sis patients with Kt/V greater than 1.20. J Ren Nutr 13:15- 25, 2003 REFERENCES 17. Kalantar-Zadeh K, Kleiner M, Dunne E, et al: Total 1. Foley RN, Parfrey PS, Sarnak MJ: Clinical epidemiol- iron-binding capacity–estimated transferrin correlates with ogy of cardiovascular disease in chronic renal disease. Am J the nutritional subjective global assessment in hemodialysis Kidney Dis 32:S112-S119, 1998 (suppl 3) patients. Am J Kidney Dis 31:263-272, 1998 2. US Renal Data System: Excerpts from the USRDS 18. Kalantar-Zadeh K, Rodriguez RA, Humphreys MH: 2004 Annual Data Report. Am J Kidney Dis 45:S1-S280, Association between serum ferritin and measures of inflam- 2005 (suppl 1) mation, nutrition and iron in haemodialysis patients. Neph- 3. Wanner C, Krane V, Marz W, et al: Randomized rol Dial Transplant 19:141-149, 2004 controlled trial on the efficacy and safety of atorvastatin in 19. Beddhu S, Samore MH, Roberts MS, Stoddard GJ, patients with type 2 diabetes on hemodialysis (4D study): Pappas LM, Cheung AK: Creatinine production, nutrition, Demographic and baseline characteristics. Kidney Blood and glomerular filtration rate estimation. J Am Soc Nephrol Press Res 27:259-266, 2004 14:1000-1005, 2003 4. Eknoyan G, Beck GJ, Cheung AK, et al: Effect of 20. Reddan DN, Klassen PS, Szczech LA, et al: White dialysis dose and membrane flux in maintenance hemodialy- blood cells as a novel mortality predictor in haemodialy- sis. N Engl J Med 347:2010-2019, 2002 sis patients. Nephrol Dial Transplant 18:1167-1173, 2003 5. Salahudeen AK: Obesity and survival on dialysis. 21. Wolfson M, Strong CJ, Minturn D, Gray DK, Kopple Am J Kidney Dis 41:925-932, 2003 JD: Nutritional status and lymphocyte function in mainte- 6. Kalantar-Zadeh K, Abbott KC, Salahudeen AK, Kil- nance hemodialysis patients. Am J Clin Nutr 39:547-555, patrick RD, Horwich TB: Survival advantages of obesity in 1984 dialysis patients. Am J Clin Nutr 81:543-554, 2005 22. Kuwae N, Kopple JD, Kalantar-Zadeh K: A low 7. Kalantar-Zadeh K, Kilpatrick RD, McAllister CJ, lymphocyte percentage is a predictor of mortality and hospi- Greenland S, Kopple JD: Reverse epidemiology of hyperten- talization in hemodialysis patients. Clin Nephrol 63:22-34, sion and cardiovascular death in the hemodialysis popula- 2005 WEIGHT CHANGE AND SURVIVAL IN DIALYSIS 499

23. Lowrie EG: Acute-phase inflammatory process con- 40. Wolfe RA, Ashby VB, Daugirdas JT, Agodoa LY, tributes to malnutrition, anemia, and possibly other abnor- Jones CA, Port FK: Body size, dose of hemodialysis, and malities in dialysis patients. Am J Kidney Dis 32:S105- mortality. Am J Kidney Dis 35:80-88, 2000 S112, 1998 (suppl 1) 41. Port FK, Ashby VB, Dhingra RK, Roys EC, Wolfe 24. Kalbfleisch JD, Prentice RL: The Statistical Analy- RA: Dialysis dose and body mass index are strongly associ- sis of Failure-Time Data (ed 2). New York, NY, Wiley, ated with survival in hemodialysis patients. J Am Soc 2002 Nephrol 13:1061-1066, 2002 25. Calle EE, Thun MJ, Petrelli JM, Rodriguez C, 42. Kopple JD, Zhu X, Lew NL, Lowrie EG: Body Heath CW Jr: Body-mass index and mortality in a prospec- weight-for-height relationships predict mortality in main- tive cohort of US adults. N Engl J Med 341:1097-1105, tenance hemodialysis patients. Kidney Int 56:1136-1148, 1999 1999 26. Kuczmarski RJ, Flegal KM, Campbell SM, Johnson 43. Beddhu S, Pappas LM, Ramkumar N, Samore M: CL: Increasing prevalence of overweight among US adults. Effects of body size and body composition on survival in The National Health and Nutrition Examination Surveys, hemodialysis patients. J Am Soc Nephrol 14:2366-2372, 1960 to 1991. JAMA 272:205-211, 1994 2003 27. Byers T: Body weight and mortality. N Engl J Med 44. Kaizu Y, Tsunega Y, Yoneyama T, et al: Overweight 333:723-724, 1995 as another nutritional risk factor for the long-term survival 28. Manson JE, Willett WC, Stampfer MJ, et al: Body of non-diabetic hemodialysis patients. Clin Nephrol 50:44- weight and mortality among women. N Engl J Med 333:677- 50, 1998 685, 1995 45. Kutner NG, Zhang R: Body mass index as a predictor 29. Mark DH: Deaths attributable to obesity. JAMA of continued survival in older chronic dialysis patients. Int 293:1918-1919, 2005 Urol Nephrol 32:441-448, 2001 30. Flegal KM, Graubard BI, Williamson DF, Gail MH: 46. Kalantar-Zadeh K, Block G, Horwich T, Fonarow Excess deaths associated with underweight, overweight, and GC: Reverse epidemiology of conventional cardiovascular obesity. JAMA 293:1861-1867, 2005 risk factors in patients with chronic heart failure. J Am Coll 31. Kushner RF: Body weight and mortality. Nutr Rev Cardiol 43:1439-1444, 2004 51:127-136, 1993 47. Stevens J, Cai J, Pamuk ER, Williamson DF, Thun 32. Degoulet P, Legrain M, Reach I, et al: Mortality risk MJ, Wood JL: The effect of age on the association factors in patients treated by chronic hemodialysis. Report between body-mass index and mortality. N Engl J Med of the Diaphane Collaborative Study. Nephron 31:103-110, 338:1-7, 1998 1982 48. Landi F, Onder G, Gambassi G, Pedone C, Carbo- 33. Leavey SF, McCullough K, Hecking E, Goodkin D, nin P, Bernabei R: Body mass index and mortality among Port FK, Young EW: Body mass index and mortality in hospitalized patients. Arch Intern Med 160:2641-2644, ‘healthier’ as compared with ‘sicker’ haemodialysis pa- 2000 tients: Results from the Dialysis Outcomes and Practice 49. Yeh S, Wu SY, Levine DM, et al: Quality of life and Patterns Study (DOPPS). Nephrol Dial Transplant 16:2386- stimulation of weight gain after treatment with megestrol 2394, 2001 acetate: Correlation between cytokine levels and nutritional 34. Leavey SF, Strawderman RL, Jones CA, Port FK, status, appetite in geriatric patients with wasting syndrome. J Held PJ: Simple nutritional indicators as independent predic- Nutr Health Aging 4:246-251, 2000 tors of mortality in hemodialysis patients. Am J Kidney Dis 50. Chlebowski RT, Grosvenor M, Lillington L, Sayre J, 31:997-1006, 1998 Beall G: Dietary intake and counseling, weight mainte- 35. Fleischmann E, Teal N, Dudley J, May W, Bower JD, nance, and the course of HIV infection. J Am Diet Assoc Salahudeen AK: Influence of excess weight on mortality and 95:428-432; quiz, 433-435, 1995 hospital stay in 1346 hemodialysis patients. Kidney Int 51. Liu Y, Coresh J, Eustace JA, et al: Association between 55:1560-1567, 1999 cholesterol level and mortality in dialysis patients: Role of 36. Wong JS, Port FK, Hulbert-Shearon TE, et al: Sur- inflammation and malnutrition. JAMA 291:451-459, vival advantage in Asian American end-stage renal disease 2004 patients. Kidney Int 55:2515-2523, 1999 52. Kalantar-Zadeh K: Causes and consequences of the 37. Abbott KC, Glanton CW, Trespalacios FC, et al: reverse epidemiology of body mass index in dialysis pa- Body mass index, dialysis modality, and survival: Analy- tients. J Ren Nutr 15:142-147, 2005 sis of the United States Renal Data System Dialysis 53. Horwich TB, Fonarow GC, Hamilton MA, MacLel- Morbidity and Mortality Wave II Study. Kidney Int 65:597- lan WR, Woo MA, Tillisch JH: The relationship between 605, 2004 obesity and mortality in patients with heart failure. J Am 38. Glanton CW, Hypolite IO, Hshieh PB, Agodoa LY, Coll Cardiol 38:789-795, 2001 Yuan CM, Abbott KC: Factors associated with improved 54. Mohamed-Ali V, Goodrick S, Bulmer K, Holly JM, short term survival in obese end stage renal disease patients. Yudkin JS, Coppack SW: Production of soluble tumor necro- Ann Epidemiol 13:136-143, 2003 sis factor receptors by human subcutaneous adipose tissue in 39. Johansen KL, Young B, Kaysen GA, Chertow GM: vivo. Am J Physiol 277:E971-E975, 1999 Association of body size with outcomes among patients 55. Weber MA, Neutel JM, Smith DH: Contrasting beginning dialysis. Am J Clin Nutr 80:324-332, 2004 clinical properties and exercise responses in obese and 500 KALANTAR-ZADEH ET AL lean hypertensive patients. J Am Coll Cardiol 37:169-174, 58. Longenecker JC, Coresh J, Klag MJ, et al: Validation 2001 of comorbid conditions on the end-stage renal disease medi- 56. Rauchhaus M, Coats AJ, Anker SD: The endotoxin- cal evidence report: The CHOICE study. Choices for Healthy lipoprotein hypothesis. Lancet 356:930-933, 2000 Outcomes in Caring for ESRD. J Am Soc Nephrol 11:520- 57. Macleod J, Davey Smith G: Psychosocial factors and 529, 2000 public health: A suitable case for treatment? J Epidemiol 59. Pischon T, Sharma AM: Obesity as a risk factor in renal Community Health 57:565-570, 2003 transplant patients. Nephrol Dial Transplant 16:14-17, 2001 Tunneled Catheters in Hemodialysis Patients: Reasons and Subsequent Outcomes

Timmy Lee, MD, Jill Barker, PhD, and Michael Allon, MD

● Background: Reducing the use of tunneled catheters in hemodialysis patients requires concerted efforts to convert them to a usable permanent vascular access. The goal of this study is to evaluate the reasons for tunneled catheter use in our prevalent hemodialysis population and the success in converting them to a permanent vascular access. Methods: We identified all catheter-dependent hemodialysis patients at our center on a single date. These patients were followed up prospectively during a 1-year period to evaluate access procedures and conversion to permanent access use. Results: Of 458 prevalent hemodialysis patients, 108 patients (23.6%) were dialyzing through cuffed tunneled catheters: 18.5% had no further options for creation of a permanent vascular access, 28.7% had an immature access, 43.5% had access placement pending, and 9.2% had repeatedly refused access surgery. For 78 catheter-dependent patients (excluding patients with no access options and those who refused permanent access surgery), the likelihood of using a permanent access was 53% by 6 months and 80% by 1 year. In patients with an immature access, 50% were using a permanent access at 3 months, and 80%, at 6 months. Of patients with access surgery pending, 45% had access surgery performed within 3 months, and 70%, at 6 months. Finally, of all patients, the likelihood of catheter-related bacteremia was 48% at 6 months. On multivariable analysis, only duration of catheter dependence predicted subsequent use of a permanent access (hazard ratio, 3.11; 95% .for catheter dependence less than versus greater than 6 months (0.0002 ؍ confidence interval, 1.70 to 5.68; P Conclusion: Almost one quarter of our hemodialysis population is catheter dependent. Despite concerted efforts, there remain very long delays in achieving a usable permanent access, attributable to delays in both surgical access placement and access maturation. In the interim, this patient population developed a high frequency of catheter-related bacteremia. Am J Kidney Dis 46:501-508. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Vascular access; catheter; fistula; infection.

HE NATIONAL KIDNEY Foundation– manent vascular access poses similar challenges. T Kidney Disease Outcomes Quality Initia- Very little has been reported on this patient tive Clinical Practice Guidelines for Vascular population pertaining to reasons for tunneled- Access recommend minimizing the use of tun- catheter dependence and obstacles to achieving a neled catheters for hemodialysis.1 In part, achiev- usable permanent access. Identifying specific rea- ing this goal requires aggressive efforts in plac- sons for catheter dependence and targeting spe- ing arteriovenous fistulae in predialysis patients. cific obstacles for obtaining a usable permanent Helpful measures include preoperative vascular access in this subgroup of patients may help mapping, creation of new fistulae, subsequent interventions to convert immature fistulae to ones that can be cannulated successfully for dialysis, and acquisition of access cannulation skills by 2 From the Division of Nephrology, University of Alabama dialysis staff. Despite implementing such mea- at Birmingham, AL; and Statistical Consulting of Montana, sures, approximately 60% of incident US hemo- Bozeman, MT. dialysis patients use catheters.3,4 Additional pa- Received March 1, 2005; accepted in revised form May tients may become catheter dependent when their 31, 2005. Originally published online as doi:10.1053/j.ajkd.2005.05.024 permanent vascular access fails. on July 25, 2005. A recent study evaluated factors affecting pre- Supported in part by grant no. 1 K24 DK59818-01 from dialysis vascular access management in predialy- the National Institute of Diabetes and Digestive and Kidney sis patients at our institution.5 We observed 2 Diseases (M.A.). Presented in abstract form at the 14th Annual Meeting of striking hurdles to increasing the rate of success- the National Kidney Foundation, Washington, DC, May 4-8, ful fistulae: (1) the rate of predialysis access 2005. surgery performed was substantially less than Address reprint requests to Michael Allon, MD, Division that of surgical referral, and (2) many fistulae of Nephrology, 728 Richard Arrington Blvd, Birmingham, AL 35294. E-mail: [email protected] placed months in advance were still not usable © 2005 by the National Kidney Foundation, Inc. during the first hemodialysis session. For catheter- 0272-6386/05/4603-0014$30.00/0 dependent hemodialysis patients, achieving a per- doi:10.1053/j.ajkd.2005.05.024

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 501-508 501 502 LEE, BARKER, AND ALLON develop future strategies to improve the preva- Table 1. Categories of Patients With Tunneled lence of fistulae in hemodialysis patients. Catheters

The goal of the current study is to evaluate Category reasons for dialysis catheter use at a single insti- tution and the success in converting this popula- 1: No options 20 (18.5) tion to permanent vascular access after 1 year. 2: Immature access, without revision 27 (25.0) 3: Immature access, with revision 4 (3.7) METHODS 4: Access surgery scheduled 15 (13.9) 5: Access appointment missed 11 (10.2) Study Population 6: Access appointment not yet scheduled 21 (19.4) The University of Alabama at Birmingham provides long- 7: Refuses permanent access 10 (9.2) term dialysis treatment to approximately 550 patients, of Total 108 (100.0) whom 85% receive in-center hemodialysis. At the onset of NOTE. Values expressed as number of patients (per- this study (March 15, 2004), a total of 458 patients were cent). receiving hemodialysis at 1 of 5 dialysis units in the metro- politan Birmingham area. Of 458 patients, 108 were dialyz- ing by means of a cuffed tunneled catheter. Ten clinical attempted because of poor success with this intervention at nephrologists, all full-time university faculty in the Division our institution. Arteriovenous grafts typically were allowed of Nephrology, provide medical care to these patients. to mature for 2 to 3 weeks before cannulation for dialysis. Vascular Access Management Data Analysis Dialysis-dependent patients with cuffed tunneled cath- We obtained approval from the University of Alabama at eters were referred by their nephrologists to the Division of Birmingham Institutional Review Board to review patient Transplant Surgery for access evaluation and surgery by 1 of medical records for research purposes. Using a prospective 3 experienced surgeons. Preoperative sonographic vascular computerized database of access procedures, we identified mapping was performed routinely to assess vessel diameter all patients who were dialyzing with a tunneled catheter as of and exclude vein stenosis or thrombosis. These results were 6,7 March 15, 2004. Patient records were reviewed to deter- used by surgeons for optimal planning of vascular surgery. mine: (1) duration of tunneled catheter use; (2) number of Fistulae were placed preferentially to grafts when preopera- previous permanent accesses (fistula or graft) before March tive vascular mapping indicated suitable vessels. 15, 2004; and (3) specific reasons why patients were dialyz- ing with a catheter. Seven mutually exclusive categories Vascular Access Scheduling were developed to explain patients’ tunneled catheter depen- All access appointments were scheduled by 2 full-time dence (Table 1). The categories took into consideration access coordinators who maintained a prospective computer- whether a permanent access was already in place, there were ized database of all previous and current access clinic options for placing a new access, and the patient was willing appointments and procedures.8 All patients were provided to undergo access surgery. In addition, we obtained informa- with typed instructions about their access appointments, tion about sex, age, race, diabetes, peripheral vascular dis- followed by telephone reminders. Each patient was sched- ease, and date of initiation of dialysis therapy. uled for a preoperative vascular mapping appointment in the Patients were followed up prospectively during 1 year Radiology Clinic, a preoperative appointment in the Surgery until March 15, 2005, to determine vascular access out- Clinic, and a preoperative assessment by Anesthesia. If the comes. Their medical records were reviewed to determine: patient attended all scheduled appointments, access surgery (1) whether a permanent access was in use by March 15, typically was performed within 1 month of the initial referral 2005; (2) number of access procedures; and (3) number of by the nephrologist. catheter-associated infections during this 1-year interval. A For any patient who missed his or her scheduled appoint- patient was deemed to be using a permanent access when our ment, access coordinators rescheduled all new appoint- computerized database of access procedures documented ments. Of note, because many of our patients are of low removal of the patient’s tunneled catheter, typically after socioeconomic status and have transportation hardships, we successful cannulation of the access on 3 consecutive dialy- were able to provide taxi vouchers for access-related appoint- sis sessions. Tunneled catheter–related bacteremia was diag- ments with a grant obtained from the University of Alabama nosed in catheter-dependent patients with fever or chills, at Birmingham Health Services Foundation. positive blood culture results, and no evidence of an alter- nate source of infection.11,12 Postoperative Management of Vascular Access Fistulae typically were not cannulated for 6 to 8 weeks. In Statistical Analysis addition, a postoperative ultrasound was scheduled routinely Survival analysis techniques were used to model time to to assess fistula maturation.9 Patients with an access deter- use of permanent vascular access for hemodialysis, time to mined to be immature by clinical or sonographic criteria at 6 access surgery of patients without permanent access, time to to 8 weeks were referred for salvage procedures by Radiol- first scheduled preoperative access appointment, time to first ogy or Surgery.10 If a fistula clotted, thrombectomy was not actual access appointment, time to access placement, and REASONS FOR DIALYSIS CATHETERS 503

Table 2. Prevalence of Tunneled Catheters by dialysis therapy for 6.2 Ϯ 1.2 years, had 3.9 Ϯ Hemodialysis Unit 0.5 previous failed fistulae or grafts, and had Ϯ No. of No. of Patients Percent With been continuously catheter dependent for 1.5 Dialysis Unit Patients With Catheter Catheter 0.3 years. They included 14 patients with no more sites for permanent access in any extremity, A 116 26 22.4 4 patients with severe peripheral vascular disease B 95 27 28.4 C 127 28 22.0 precluding placement of a thigh graft, 1 patient D 70 15 21.4 with severe arterial calcification precluding place- E 50 12 24.0 ment of a thigh graft, and 1 patient with recurrent Total 458 108 23.6 access infections in whom the surgeon declined to place another access. After assembling the list of catheter-depen- time to first episode of catheter-related bacteremia. Observa- tion time for all survival analyses was started from March dent patients and distributing it to nephrologists 15, 2004. Patient follow-up was censored at the time of and dialysis staff, we followed up patients pro- death or transfer to an outside dialysis facility. Univariate spectively during 1 year to determine subsequent Cox proportional hazards models were fit. Multivariable access outcomes. Of 78 catheter-dependent pa- Cox proportional hazards models allowed for the evaluation tients (excluding patients with no access options of several independent variables in the presence of each other. Hazard ratios and their associated 95% confidence [category 1] and those who refused permanent intervals were computed. access surgery [category 7]), the likelihood of using a permanent access was 53% at 6 months RESULTS and 80% at 1 year (Fig 1). In the subset of At the initial evaluation period (March 15, 2004), 108 patients were dialyzing with a tun- Table 3. Clinical Features of Patients With neled catheter at 5 University of Alabama at Tunneled Catheters

Birmingham in-center hemodialysis units. Aver- Parameter age prevalence of catheters among all dialysis units was 23.6% and was similar for units of Age (y) different size (Table 2). Demographic and clini- Ն65 34 (31.5) Ͻ65 74 (68.5) cal characteristics of the catheter-dependent he- Sex modialysis population are listed in Table 3. Nearly Female 58 (53.7) one third were older than 65 years. There was a Male 50 (46.3) slight preponderance of female patients. The Race majority of our patients were black, reflecting the Black 94 (87.0) White 14 (13.0) demographics of our hemodialysis population. Diabetes More than half the patients had diabetes. Almost Yes 66 (61.1) half the patients had been on hemodialysis therapy No 42 (38.9) for less than 1 year at the time of evaluation, Peripheral vascular disease reflecting new patients starting dialysis therapy Yes 22 (20.4) No 86 (79.6) without a mature permanent access. The remain- Years on hemodialysis* der had been on dialysis therapy from 1 to 24 Ͻ1 45 (42.0) years, reflecting patients with a previous perma- 1-2 10 (9.3) nent access that had failed. 2-3 12 (11.2) Of 108 catheter-dependent hemodialysis pa- 3-4 11 (10.3) 4-5 5 (4.7) tients, 18.5% had exhausted all options for cre- 5-6 5 (4.7) ation of a permanent vascular access, 28.7% had 6-7 4 (3.7) immature accesses (with or without revisions), 7-8 4 (3.7) 43.5% had access surgery pending (either sched- 8-10 1 (1.0) uled, missed, or not yet scheduled), and 9.2% Ͼ10 10 (9.3) had repeatedly refused permanent access surgery NOTE. Values expressed as number of patients (per- (Table 1). The 20 patients with no further options cent). for permanent access (category 1) had been on *Missing hemodialysis duration for 1 patient. 504 LEE, BARKER, AND ALLON

Fig 1. The cumulative likelihood of using a perma- This Fig 3. The cumulative likelihood of having access .(78 ؍ nent access after 1 year of follow-up (n survival curve excludes patients with no options for a surgery performed after 1 year of follow-up in patients permanent access (category 1) and those who persis- with pending access surgery at baseline (categories 4 .(47 ؍ tently refused access surgery (category 7). to 6; n patients with an immature access at the initial appointment scheduled, first access appointment evaluation (categories 2 to 3; n ϭ 31), the like- seen, first access surgery performed, and first use lihood of using a permanent access was 50% at of a permanent access were 18, 31, 154, and 360 3 months, 80% at 6 months, and 92% at 1 year days, respectively. Thus, median time from sur- (Fig 2). gery appointment to access surgery was approxi- Of patients for whom access surgery was pend- mately 4 months, and median time from access ing (categories 4 to 6; n ϭ 47), 45% had access surgery to access use was 7 months. surgery performed within 3 months, 70% at 6 The 108 catheter-dependent patients enrolled months, and 85% at 1 year (Fig 3). Finally, of in this study experienced numerous episodes of patients who did not yet have a surgery date catheter-related bacteremia. Specifically, during scheduled on initial evaluation (categories 5 to 6; the 1-year follow-up, there were a total of 92 n ϭ 32), 96% had access appointments sched- infections. Adjusted for the 20,104 catheter-days uled, 85% were seen by an access surgeon, 62% of follow-up, this translates to 4.6 episodes of had a permanent access placed, and only 16% catheter-related bacteremia per 1,000 days. The had a permanent access in use after a 6-month likelihood of catheter-associated bacteremia was follow-up (Fig 4). Median times to first access 35% at 3 months and 48% at 6 months and

Fig 4. The cumulative likelihood of having a sched- uled preoperative visit (thin dotted line), actually being seen by the surgeon (thick dotted line), having access surgery performed (thin solid line), and using the per- Fig 2. The cumulative likelihood of using a perma- manent access (thick solid line) in patients who did nent access after 1 year of follow-up in patients with an not yet have an access surgery date at baseline (cat- .(32 ؍ egories 5 to 6; n .(31 ؍ immature access at baseline (categories 2 to 3; n REASONS FOR DIALYSIS CATHETERS 505

catheter dependent for less than 6 months were significantly more likely to use a permanent access compared with patients who had been catheter dependent for greater than 6 months (hazard ratio, 3.11; 95% confidence interval, 1.70 to 5.68; P ϭ 0.0002).

DISCUSSION Between 1998 and 2002, the proportion of US hemodialysis patients dialyzing with a fistula increased from 27% to 33%, whereas the propor- tion with catheters increased from 19% to 27%.13 More fistulae are being placed, as recommended Fig 5. The cumulative likelihood of having an epi- by the Kidney Disease Outcomes Quality Initia- sode of catheter-related bacteremia during 1 year of tive guidelines.1 However, compared with grafts, .(108 ؍ follow-up in all catheter-dependent patients (n fistulae have a greater primary failure rate, take longer to mature adequately for successful cannu- appeared to increase in a linear fashion up to 9 lation, and require more salvage procedures to months of follow-up (Fig 5). We also monitored achieve maturation.2 In patients initiating dialy- other aspects of catheter function. Instillation of sis therapy with a tunneled catheter, this trans- tissue plasminogen activator for catheter malfunc- lates to prolonged catheter dependence. tion was required at a frequency of 3.0 episodes/ The process leading from identifying the need 1,000 catheter-days, and catheter exchange be- for vascular access to having a mature fistula re- cause of malfunction occurred at a frequency of quires multiple steps and a multidisciplinary ap- 1.1 episodes/1,000 catheter-days. Adequacy of proach involving an access coordinator, nephrolo- dialysis also was compared in March 2004 be- gists, access surgeon, and interventional radiologist tween catheter-dependent patients and those us- (or interventional nephrologists) to achieve optimal ing a permanent access (fistula or graft). Mean access outcomes.2 Failure to achieve any of the Kt/V was significantly lower in catheter-depen- intermediate steps results in progressive delays in Ϯ Ϯ dent patients (1.38 0.27 versus 1.52 0.26; obtaining a mature fistula. The stakes in streamlin- Ͻ P 0.0001). The proportion of patients with a ing the process are considerably greater after the Kt/V less than 1.20 was significantly greater in patient has started maintenance hemodialysis catheter-dependent patients (25.2% versus 9.7%; therapy. As documented in the present study, cath- Ͻ P 0.0001). eter dependence translated into a high likelihood of Finally, we evaluated clinical factors predic- tive of subsequent use of a permanent access in catheter-dependent patients. On univariate analy- Table 4. Association Between Baseline Clinical sis, prolonged catheter dependence and presence Parameters and Likelihood of Subsequent Permanent Access Use of peripheral vascular disease each were associ- ated with a lower likelihood of using a perma- 95% nent access during follow-up (Table 4). In addi- Hazard Confidence tion, female sex tended to be associated with a Variable P Ratio Interval lower likelihood of subsequent permanent access Age (/y) 0.18 0.99 0.97-1.01 use (P ϭ 0.07). Conversely, patient age, race, Race (black v white) 0.17 0.57 0.256-1.27 diabetic status, and duration of dialysis therapy Sex (male v female) 0.07 1.68 0.96-2.91 were not associated significantly with the likeli- Diabetes (yes v no) 0.98 1.01 0.58-1.76 Peripheral vascular hood of future permanent access use. Using disease (yes v no) 0.019 3.42 1.23-9.52 multivariable stepwise proportional hazard re- Duration of dialysis (/y) 0.41 0.97 0.90-1.04 gression analysis, only duration of catheter depen- Duration of catheter dence was associated with future permanent ac- dependence (Ͻ6or Ͼ cess use. Specifically, patients who had been 6 mo) 0.0002 3.11 1.70-5.68 506 LEE, BARKER, AND ALLON catheter-related bacteremia (Fig 5). Other problems cess surgery performed within the 6-month win- intrinsic to catheters include frequent thrombosis dow (Fig 3). This disappointing outcome was and suboptimal dialysis blood flows, which jeopar- obtained despite heroic efforts to streamline the dize dialysis adequacy.14 process, optimize communication, and overcome We documented that 23.4% of our hemodialy- perceived financial barriers. A major obstacle to sis patients were catheter dependent (Table 1), increasing timely access surgery is shown in Fig similar to the 27% rate reported for a random 4. Whereas 96% of patients had a preoperative national sample reported by the End-Stage Renal access surgery appointment scheduled within 6 Disease Clinical Performance Measures Project.13 months, only 62% actually had an access placed. The distribution of reasons for catheter depen- We previously documented the high rate of dence in our hemodialysis population (Table 3) missed preoperative appointments in predialysis also was reasonably similar to that described in patients, with 60% of patients missing at least 1 the Clinical Performance Measures Project: no scheduled appointment and 39% missing 3 or options for permanent access (18% versus 12%), more appointments.5 Our multiple variable anal- immature access (29% versus 27%), access sur- ysis found that duration of catheter dependence gery pending (44% versus 31%), and patient was the single factor predicting likelihood of refusal (9% versus 13%). subsequent use of permanent access. The longer After excluding patients with no options for a the duration of catheter dependence, the lower permanent vascular access and those who consis- the likelihood that the patient will convert to tently refused access surgery, we found that only permanent access use in the subsequent year. 53% were using a permanent access after a This observation highlights the critical impor- 6-month follow-up, and 80%, at 1 year (Fig 1). tance of streamlining the steps required to achieve Previous publications tracked access outcomes a usable access in catheter-dependent hemodialy- only among incident patients who were catheter sis patients. dependent. Stehman-Breen et al3 reported that In theory, achieving compliance with surgery 46% of patients initiating dialysis therapy with a appointments should be easier once the patient catheter were still using a catheter for dialysis has initiated dialysis therapy. Because dialysis is after 60 days. The Clinical Performance Mea- now a reality, rather than a hypothetical possibil- sures Project reported that 67% of patients initi- ity, one would expect the patient to be more ating dialysis therapy with a catheter remained highly motivated to get a permanent access. catheter dependent after 90 days. Moreover, hemodialysis patients are a “captive Conversely, the present investigation included audience,” and the staff has an opportunity thrice all catheter-dependent hemodialysis patients. weekly to educate them of the importance of Nearly 60% had been on dialysis therapy for at permanent vascular access and remind them of least 1 year and had become catheter dependent their appointments. However, many of these pa- after their previous permanent access had failed. tients have had failed accesses in the past and are It is notable that almost one fifth of catheter- hesitant to undergo further access surgery, know- dependent patients had no options for placement ing the possibility that the access may fail in the of a fistula or graft, which means they were future. Others prefer the convenience of dialyz- consigned to permanent catheter dependence. ing with a catheter and avoiding multiple needle This subset of patients represented more than 4% sticks. Furthermore, many of our dialysis popula- of our total hemodialysis population. After includ- tion have comorbid medical problems, and their ing patients who adamantly refused permanent surgery may be postponed when they are hospi- access, this means that more than one quarter of talized with an acute medical problem. Similarly, our catheter-dependent population will never have the occurrence of catheter-related bacteremia of- a fistula or graft placed. ten leads to surgery postponement until the infec- Six months would appear more than adequate tion has been resolved. time to ensure that a catheter-dependent hemodi- Other potential scenarios leading to delays in alysis patient has a permanent vascular access access surgery included patients not coming for placed. Only 70% of patients with access surgery their scheduled surgery appointment, patients pending at our study initiation actually had ac- coming to their scheduled appointments in a REASONS FOR DIALYSIS CATHETERS 507 nonfasting state, patient surgery postponed be- specifically focuses on prevalent catheter-depen- cause of a requirement for preoperative cardiac dent hemodialysis patients and dissects the spe- evaluation, surgery rescheduled by the surgeon cific obstacles to achieving a usable permanent because of an emergent surgical procedure (kid- vascular access. Although some of the hurdles ney transplantation), or surgery canceled be- encountered may be unique to our institution, it cause the patient’s preoperative laboratory tests is likely that many of them would generalize to showed hyperkalemia. Finally, we only classified other dialysis centers. patients as refusing permanent access if they We observed a high frequency of catheter- repeatedly stated this position despite intense related bacteremia in our hemodialysis popula- efforts to change their minds. It is possible that tion. The frequency in the current study (4.6 some patients with multiple missed surgery ap- episodes/1,000 catheter-days) is within the range pointments have no intention of proceeding with reported in several large series (2.4 to 5.5/1,000 access surgery, but are unwilling to explicitly catheter-days19). These infections contribute sub- state their position. stantially to the morbidity of these patients. Hos- The second major hurdle to converting cath- pitalizations caused by access-related infections eter-dependent patients to usable fistulae is evi- occur more commonly in catheter-dependent pa- dent from Fig 2. In the subset of patients who had tients than those using a permanent access; more- an immature access at study initiation, only 50% over, the risk for infection-related death is greater were actually using the permanent access after a in catheter-dependent patients.20 Although not 3-month follow-up, and 80%, at 6 months. It has specifically assessed in the present study, it is been well documented that fistulae may fail to likely that catheter-related infections have a nega- mature because of either a stenotic lesion or the tive impact on patient quality of life. Prophylac- presence of tributary veins. An aggressive ap- tic measures, such as use of antimicrobial lock proach to correcting these abnormalities can suc- solutions or exit-site antibiotic ointments, may cessfully promote the maturation of these fistu- reduce the frequency of catheter-related bactere- lae.15-17 Moreover, some fistulae require surgical mia.19 In addition to infectious problems with superficialization to be safely cannulated be- catheter use, we also documented problems with cause they are too deep in their native state.18 It frequent loss of catheter patency and inadequate usually is evident within 4 to 6 weeks of their dialysis dose. construction which fistulae may require salvage In conclusion, in our prevalent hemodialysis procedures to ensure their maturation.9 How- patients, nearly one quarter dialyze with a tunneled ever, as in the case of the primary access proce- catheter. Even more troubling is that after 6 months dure, multiple missed appointments result in of follow-up, the likelihood of dialyzing with a long delays in achieving the corrective action. functional access was only 53%. We identified 2 The Dialysis Outcomes and Practice Patterns major obstacles to explain the low rate of perma- Study reported that fistula failure was greater in nent access in our dialysis population; namely, long incident patients initiating dialysis with a cath- delays in obtaining access surgery and a high rate eter compared with patients using a fistula on of immature fistulae. Additional investigation is 4 their first dialysis treatment. Thus, the extended required to determine the specific factors respon- catheter use in the prevalent patients enrolled in sible for delays in access surgery and also under- the current study may have contributed to their stand the factors responsible for immature accesses high fistula failure rate. that are unusable for dialysis. Otherwise, it has Recent multicenter observational studies have been well shown that the longer the patient contin- reported on predialysis vascular access manage- ues to dialyze with a catheter, the greater the ment and outcomes of incident hemodialysis likelihood of catheter-related bacteremia and its 3,4 patients with catheters. However, they did not associated complications. address the outcomes of prevalent patients with catheters or provide detailed analysis of the spe- REFERENCES cific hurdles to converting such patients to a 1. National Kidney Foundation: DOQI Clinical Practice permanent vascular access. The present investi- Guidelines for Vascular Access: Update 2000. Am J Kidney gation is a single-center observational study that Dis 37:S137-S181, 2001 (suppl 1) 508 LEE, BARKER, AND ALLON

2. Allon M, Robbin ML: Increasing arteriovenous fistu- 12. Poole CV, Carlton D, Bimbo L, Allon M: Treatment las in hemodialysis patients: Problems and solutions. Kidney of catheter-related bacteremia with an antibiotic lock proto- Int 62:1109-1124, 2002 col: Effect of bacterial pathogen. Nephrol Dial Transplant 3. Stehman-Breen CO, Sherrard DJ, Gillen D, Caps M: 19:1237-1244, 2004 Determinants of type and timing of initial permanent hemo- 13. Centers for Medicare and Medicaid Services: 2003 dialysis vascular access. Kidney Int 57:639-645, 2000 Annual Report: End-Stage Renal Disease Clinical Perfor- 4. Pisoni RL, Young EW, Dykstra DM, et al: Vascular mance Measures Project. Am J Kidney Dis 44:S1-S92, 2004 access use in Europe and in the United States: Results from (suppl 2) the DOPPS. Kidney Int 61:305-316, 2002 14. Schwab SJ, Beathard GA: The hemodialysis catheter 5. Lee T, Barker J, Allon M: Associations with predialy- conundrum: Hate living with them, but can’t live without sis vascular access management. Am J Kidney Dis 43:1008- them. Kidney Int 56:1-17, 1999 1013, 2004 15. Beathard GA, Settle SM, Shields MW: Salvage of the 6. Robbin ML, Gallichio ML, Deierhoi MH, Young CJ, nonfunctioning arteriovenous fistula. Am J Kidney Dis 33: Weber TM, Allon M: US vascular mapping before hemodi- 910-916, 1999 alysis access placement. Radiology 217:83-88, 2000 16. Turmel-Rodrigues L, Mouton A, Birmele B, et al: 7. Allon M, Lockhart ME, Lilly RZ, et al: Effect of Salvage of immature forearm fistulas for haemodialysis by preoperative sonographic mapping on vascular access out- interventional radiology. Nephrol Dial Transplant 16:2365- comes in hemodialysis patients. Kidney Int 60:2013-2020, 2371, 2002 2001 8. Allon M, Bailey R, Ballard R, et al: A multidisci- 17. Beathard GA, Arnold P, Jackson J, Litchfield T, for plinary approach to hemodialysis access: Prospective evalu- the Physician Operators Forum of RMS Lifeline: Aggressive ation. Kidney Int 53:473-479, 1998 treatment of early fistula failure. Kidney Int 64:1487-1494, 9. Robbin ML, Chamberlain NE, Lockhart ME, et al: 2003 Hemodialysis arteriovenous fistula maturity: US evaluation. 18. Zielinski CmV, Mittal SK, Anderson P, et al: Delayed Radiology 225:59-64, 2002 superficialization of brachiobasilic fistula: Technique and 10. Miller CD, Robbin ML, Allon M: Gender differences initial experience. Arch Surg 136:929-932, 2001 in outcomes of arteriovenous fistulas in hemodialysis pa- 19. Allon M: Dialysis catheter-related bacteremia: Treat- tients. Kidney Int 63:346-352, 2003 ment and prophylaxis. Am J Kidney Dis 44:779-791, 2004 11. Krishnasami Z, Carlton D, Bimbo L, et al: Manage- 20. Allon M, Depner TA, Radeva M, et al: Impact of ment of hemodialysis catheter related bacteremia with an dialysis dose and membrane on infection-related hospitaliza- adjunctive antibiotic lock solution. Kidney Int 61:1136- tion and death: Results of the HEMO Study. J Am Soc 1142, 2002 Nephrol 14:1863-1870, 2003 Validation of a Method to Predict Required Dialysis Time for Cases of Methanol and Ethylene Glycol Poisoning

George M. Youssef, MBBCh, and David J. Hirsch, MD

● Background: Traditional dialysis management of ethylene glycol and methanol poisoning includes frequent intradialytic measurements of concentrations of the involved alcohol and its metabolite. A simple formula to predict the required dialysis time in advance by using patient age, sex, weight, height, dialyzer specifications, and initial toxin level was proposed and tested by us previously in 5 cases. To reach a 5-mmol/L-or-less toxin concentration target, required hemodialysis time, in hours, would be [–V ln (5/A)/0.06k], where V is the Watson estimate of total-body water in liters, A is the initial toxin concentration in mmol/L, and k is 80% of the manufacturer-specified dialyzer urea clearance in milliliters per minute at the initial observed blood flow rate. Methods: We further assessed the accuracy of this formula by reviewing all dialyzed new patients with methanol or ethylene glycol poisoning from Results: There were no clinically or statistically significant differences between .(13 ؍ March 2001 to March 2004 (N mean predicted (8.7 ؎ 3.4 [SD] hours) and required (8.4 ؎ 3.2 hours) dialysis time. No rebound increase in toxin levels occurred. Conclusion: The proposed formula is a simple, yet accurate, method to predict dialysis time for patients with methanol and ethylene glycol toxicity, confirmed by validation on an independent data set. Only initial, 2 hours before termination of dialysis, and 1 to 2 hours postdialysis measurements of toxin levels are required to ensure adequate dialysis therapy. Am J Kidney Dis 46:509-511. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Alcohol poisoning; methanol poisoning; ethylene glycol poisoning; dialysis.

SIMPLE FORMULA to predict the re- and k is the 80% of the manufacturer-specified A quired dialysis time in advance, using pa- dialyzer urea clearance in milliliters per minute tient age, sex, weight, height, dialyzer specifica- at the initial observed blood flow rate. tions, and initial toxin level in mmol/L, was This formula has been the standard protocol proposed and tested by us previously in 5 adult for dialysis of patients with methanol and ethyl- patients.1 Usually, dialysis therapy targets toxin ene glycol overdose in our institution since intro- concentrations of less than 25.6 mg/dL (8 mmol/ duced in 2001. The accuracy of the formula was L), 36.8 mg/dL (8 mmol/L), 37.2 mg/dL (6 tested previously in 5 adult patients between mmol/L), and 45.5 mg/dL (6 mmol/L) of metha- June 2000 and March 2001. The aim of this nol, its metabolite formic acid, ethylene glycol, quality assurance study is to validate that for- and its metabolite glycolic acid, respectively. mula on an independent second data set. Our formula targets a universal lower level of 5 mmol/L or less for both toxins and their metabo- METHODS lites to ensure a satisfactory safety margin. As- Patient Population suming that: (1) renal clearance and metabolism Fourteen dialyzed patients with methanol or ethylene of the toxic alcohols are negligible compared glycol poisoning from March 2001 to March 2004 were with dialysis clearance, (2) toxic alcohols would reviewed retrospectively. After excluding 1 methanol poison- have a dialysis clearance similar to urea, (3) ing case in which the protocol formula was not used, 3 methanol and 10 ethylene glycol cases were included in the toxin dialysis clearance would be a constant 80% of the manufacturer’s stated urea clearance for the dialyzer at the observed blood pump speed at From the Division of Nephrology, Dalhousie University, the initiation of hemodialysis therapy, and (4) Halifax, NS, Canada. volume of distribution of toxins was total-body Received January 7, 2005; accepted in revised form May water as determined by the Watson formula,2 the 19, 2005. Originally published online as doi:10.1053/j.ajkd.2005.05.017 predicted serum toxin concentration P at time T on July 15, 2005. –(Kt/V) will equal Ae and subsequently, dialysis Address reprint requests to David J. Hirsch, MD, Queen time in hours required to reach the therapeutic Elizabeth II Health Sciences Centre, 5820 University Ave, target toxin concentration of 5 mmol/L or less of Dickson Bldg, 5th Floor, Halifax, Nova Scotia B3H 1V8, Canada. E-mail: [email protected] toxin, would be [–V ln (5/A)/0.06k], where V is © 2005 by the National Kidney Foundation, Inc. the Watson estimate of total-body water in liters, 0272-6386/05/4603-0015$30.00/0 A is the initial toxin concentration in mmol/L, doi:10.1053/j.ajkd.2005.05.017

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 509-511 509 510 YOUSSEF AND HIRSCH

Table 1. Demographic and Laboratory Data and Predicted and Required Dialysis Time in Patients With Methanol (patients 1 to 3) and Ethylene Glycol Poisoning (patients 4 to 13)

Required Patient Age Initial Serum Initial Toxin Initial Metabolite Predicted Dialysis Dialysis No. (y) Sex Weight (kg) Creatinine (mg/dL) Level (mg/dL) Level (mg/dL) Time* (h) Time (h)

125M ϳ73 1.06 144.2 87.4† 12 10.25 225M ϳ73 1.2 67.3 55.2† 5.5 6 335M ϳ75 1.78 291.6 88.8† 13 12.25 4 19 F Unavailable 1.14 25 113.6‡ 4.5 2.5 525M ϳ70 2.3 18.6 98.5‡ 5 5 6 54 M Unavailable 0.8 267 0.0‡ 9 9 737M ϳ157 1.26 68.3 75.8‡ 5 6 822M ϳ62 0.7 167.6 22.7‡ 9.5 7.5 942M ϳ59 1.92 515.2 181.8‡ 9.5 10.5 10 42 M Unavailable 5.81 18.6 151.5‡ 7.5 8 11 35 M ϳ113 1.16 93.1 91.0‡ 7 6.75 12 38 M ϳ100 1.16 285.5 0.0‡ 11 12.75 13 36 M ϳ100 1.24 471.8 0.0‡ 15 12.75

NOTE. To convert serum creatinine in mg/dL to ␮mol/L, multiply by 88.4; methanol in mg/dL to mmol/L, multiply by 0.3121; formic acid in mg/dL to mmol/L, multiply by 0.2173; ethylene glycol in mg/dL to mmol/L, multiply by 0.1611; and glycolic acid in mg/dL to mmol/L, multiply by 0.132. *P Ͼ 0.05 compared with required dialysis time (not statistically significant). †Metabolite is formic acid. ‡Metabolite is glycolic acid. study. All patients except for 1 patient with ethylene glycol less than 0.05 for statistical significance was used for com- .poisoning were men. Age ranged from 19 to 54 years. parison, and results are expressed as mean ؎ SD

Protocol RESULTS Predicted dialysis time targeting a toxin level of 5 mmol/L or Predicted and required dialysis times are listed less, based on a computerized form of the mentioned formula, in Table 1. Patients 1 to 3 had methanol poison- was noted clearly by the treating nephrologists. Initial level, in millimolar, of the toxin (methanol or ethylene glycol) or its ing, whereas patients 4 to 13 had ethylene glycol metabolite (formic or glycolic acid, respectively), whichever poisoning. There were no clinically or statisti- was greater, was used to predict required dialysis time. All 13 cally significant differences between predicted patients were administered a 10% ethanol infusion with a target and required dialysis times, as listed in Table 2 ϳ ethanol blood level of 100 to 200 mg/dL ( 22 to 45 mmol/L). (P Ͻ 0.05). There was no marked variability in In addition, absolute ethanol was added to the dialysate fluid to achieve a dialysate ethanol concentration of approximately 150 dialysis pump speed during dialysis requiring mg/dL (33 mmol/L). A Fresenius F8 dialyzer (Fresenius AG, recalculation of predicted dialysis time. All pa- Germany) was used for all patients. Measurement of the level tients except for number 10, who had significant of the toxin or its metabolite, in addition to the initial level, was renal failure and profound acidosis at presenta- performed 2 hours before the end of the predicted dialysis time tion, survived. No rebound increase in toxin and 1 to 2 hours after dialysis in all 13 patients. A default dialysate flow rate of 500 mL/min and pump speed (blood flow rate) of 250 mL/min were used for both the initiation of dialysis Table 2. t-Test of Predicted and Required Dialysis therapy and calculation of dialysis time in all patients. Time in Patients With Methanol and Ethylene Actual required dialysis time was the observed duration of Glycol Poisoning hemodialysis needed to decrease the blood level of metha- nol, formic acid, ethylene glycol, or glycolic acid to less than Predicted Required 25.6 mg/dL (8 mmol/L), 36.8 mg/dL (8 mmol/L), 37.2 No. of Dialysis Dialysis Patients Time (h) Time (h) mg/dL (6 mmol/L), and 45.5 mg/dL (6 mmol/L), respec- tively. Methanol 3 10.2 Ϯ 4.1 9.5 Ϯ 3.2 Ϯ Ϯ Analysis Ethylene glycol 10 8.3 3.2 7.1 3.2 All 13 8.7 Ϯ 3.4 8.4 Ϯ 3.2 For each treated patient, the time estimate was compared with required dialysis time. Paired t-test with an accepted P NOTE. Values expressed as mean Ϯ SD. PREDICTING DIALYSIS TIME IN ALCOHOL POISONING 511 levels occurred in any of the 13 patients. All hours before the end of estimated dialysis) would patients, including the patient who died, required detect any underestimation of required dialysis only 1 dialysis session for toxin removal. Patient time. 5 required additional dialysis for oliguric acute In summary, only 3 toxin measurements are renal failure. needed: (1) initial level to predict required dialy- sis time, (2) 2 hours before the end of predicted DISCUSSION dialysis time to ensure adequate toxin removal Validation of our previously reported method and indicate whether longer dialysis is needed, of estimating required dialysis time (formula) on and (3) 1 to 2 hours after dialysis to detect a a second independent data set further ensures rebound increase in toxin levels. These 3 toxin that this formula provides a useful tool for predict- level measurements, along with patient demo- ing required dialysis time in patients with metha- graphic data, pump speed, and dialyzer informa- nol and ethylene glycol poisoning. tion, are all that is needed to plan adequate Termination of dialysis at the point at which dialysis with no risk for inadequate toxin re- the formula predicts that blood toxin concentra- moval or rebound increase in toxin levels. tion would have reached 5 mmol/L would have This approach, in our experience, has led to: achieved the aforementioned desired serum toxin (1) elimination of unnecessary frequent intradia- concentration (Յ8 mmol/L for methanol or its lytic toxin level measurements, (2) better dialy- metabolite, Յ6 mmol/L for ethylene glycol or its sis planning, and (3) cost reduction by avoiding metabolite). Conversion of toxin level to millimo- unnecessary frequent blood work and prolonged lar units allows the use of a single formula in all dialysis therapy. cases. Occasional small differences between pre- All patients on our study were dialyzed using dicted and required dialysis times could be attrib- Fresenius F8 dialyzers. Jindal and Goldstein3 uted to lack of accuracy of weight and height found that using 80% of the manufacturer’s assessment. In our studied cases, weights and specified urea clearance as an estimate of in vivo heights used to calculate dialysis times were urea clearance was valid for 2 other dialyzers; based on measurements on admission, previous therefore, it is likely that this method of estimat- clinical visits, or an approximate estimate by the ing dialysis time for poisonings is applicable to medical team. For practicality and safety pur- other dialyzers. We invite any dialysis center poses and because accurate weight and height planning to use our formula to validate this are not always readily available at time of initial estimate approach for their preferred dialyzers. assessment, particularly with critically ill, uncon- In addition, this formula has not been validated scious, or ventilated patients, a toxin level mea- in a pediatric population and therefore we cannot surement 2 hours before the end of the predicted recommend its use in children. dialysis time would ensure adequate toxin re- REFERENCES moval and indicate if longer dialysis was needed. 1. Hirsch D, Jindal K, Wong P, et al: A simple method to Extreme weight situations, especially marked estimate the required dialysis time for cases of alcohol obesity, affect the accuracy of the Watson esti- poisoning. Kidney Int 60:2021-2024, 2001 mate of total-body water and, subsequently, the 2. Watson P, Watson I, Batt R, et al: Total body water accuracy of our formula.2 However, our formula volumes for adult males and females estimated from simple is still safe to use because there would be a anthropometric measurements. Am J Clin Nutr 33:27-39, 1980 tendency to overestimate dialysis time in case of 3. Jindal K, Goldstein MB: Urea kinetic modeling in marked obesity. Conversely, in underweight pa- chronic hemodialysis; Benefits, problems, and practical solu- tients, the second toxin level measurement (2 tions. Semin Dial 1:82-85, 1988 Time Profiles of Peritoneal and Renal Clearances of Different Uremic Solutes in Incident Peritoneal Dialysis Patients

Bert Bammens, MD, Pieter Evenepoel, MD, PhD, Kristin Verbeke, Pharm, PhD, and Yves Vanrenterghem, MD, PhD

● Background: Residual renal function (RRF) contributes substantially to the adequacy of peritoneal dialysis (PD). In the presence of RRF, maintenance of an adequate fluid balance is facilitated, level of systemic inflammation is lower, and renal endocrine functions are preserved. The beneficial impact of RRF also may be related to the preservation of specific renal elimination mechanisms, such as tubular metabolism or secretion, which are crucial for the removal of some uremic retention solutes. Methods: Time profiles of peritoneal and renal clearances of urea ␤ ␤ nitrogen (60 d), creatinine (113 d), phosphate (96 d), the middle molecule 2-microglobulin ( 2M; 11.8 kd), and the protein-bound solute p-cresol (108 d) were investigated prospectively in 24 incident PD patients. Data were analyzed by using the linear mixed models procedure. Results: During a median follow-up of 7.2 months (range, 5.6 declined significantly. Twenty-four–hour (0.004 ؍ and 24-hour urine volume (P (0.001 ؍ to 8.6 months), RRF (P ؍ creatinine (P ,(0.0002 ؍ peritoneal drainage volume increased (P < 0.0001). Renal clearances of urea nitrogen (P > P ,0.002 ؍ decreased. Peritoneal clearances of these solutes increased (P (0.001 ؍ and phosphate (P ,(0.001 ؍ ␤ 0.0001, and P < 0.0001, respectively). There was a decline in renal clearances of 2M(P 0.0004) and p-cresol (P < .(respectively ,0.559 ؍ and P 0.188 ؍ No change in peritoneal clearances of these solutes was noted (P .(0.0001 Conclusion: Increasing PD dose may compensate for deteriorating RRF with respect to the elimination of ␤ water-soluble solutes. This is not the case for the middle molecule 2M and the protein-bound solute p-cresol. Am J Kidney Dis 46:512-519. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: p-Cresol; protein-bound solutes; peritoneal dialysis (PD); clearances; residual renal function.

URING THE PAST decade, it has been The importance of RRF as a predictor of D recognized that residual renal function survival has been attributed to its role in the (RRF) contributes substantially to the adequacy maintenance of an adequate fluid balance.8-10 of peritoneal dialysis (PD). A reanalysis of the This is emphasized by the Canada USA reanaly- Canada USA study was first to describe a 12% sis, in which urinary volume per se had a stron- survival advantage for every 5-L/wk/1.73 m2 ger predictive power than the mean of renal urea 1 increase in RRF.1 Several well-controlled studies nitrogen and creatinine clearances. On the other have confirmed this finding.2-7 hand, several observations showed that patients with lower RRF have greater circulating proin- flammatory cytokine levels, making them more vulnerable to wasting and atherosclerosis.3,11,12 Furthermore, preservation of endocrine func- From the Department of Medicine, Division of Nephrol- ogy, and Laboratory of Digestion and Absorption, Univer- tions of the kidney also may contribute to the sity Hospital Gasthuisberg, Leuven, Belgium. beneficial impact of RRF on survival. The produc- Received March 22, 2005; accepted in revised form May tion of erythropoietin and 1-␣-hydroxylation of 18, 2005. vitamin D have been suggested to be better Originally published online as doi:10.1053/j.ajkd.2005.05.016 preserved in dialysis patients with than without on July 15, 2005. 13-16 Supported in part by governmental grant no. 1127602N substantial RRF. from the Fonds voor Wetenschappelijk Onderzoek. However, the importance of renal clearance Presented in part at the 24th Annual Dialysis Conference, also might be explained by the finding that the San Antonio, TX, February 9-11, 2004, and the First Joint elimination of some uremic retention solutes Congress of the International Society for Peritoneal Dialysis and the European Peritoneal Dialysis Meeting, Amsterdam, depends largely on renal metabolism and/or se- The Netherlands, August 28-31, 2004. cretion at the tubular level, which cannot be Address reprint requests to Pieter Evenepoel, MD, PhD, equaled by peritoneal transport. This is the case Department of Medicine, Division of Nephrology, University ␤ ␤ for the middle molecule 2-microglobulin ( 2M), Hospital Gasthuisberg, Herestraat 49, B-3000 Leuven, Bel- as well as for the group of protein-bound sol- gium. E-mail: [email protected] 17 ␤ © 2005 by the National Kidney Foundation, Inc. utes. 2M is filtered in glomeruli and reab- 0272-6386/05/4603-0016$30.00/0 sorbed and degraded in proximal tubular cells. doi:10.1053/j.ajkd.2005.05.016 Its renal catabolism in a healthy subject was

512 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 512-519 TIME PROFILES OF CLEARANCES IN PD PATIENTS 513 found to be 150 to 220 mg/d.18 Protein-bound Treatment solutes are excreted mainly by tubular secretion According to usual clinical practice, all patients were through the organic anion transport system of started on a continuous ambulatory PD regimen with 4 proximal tubular cells.19 In a recent observa- exchanges of conventional lactate-buffered glucose solu- tional study including 30 prevalent PD patients, tions (Dianeal 1.36%; Baxter, Lessines, Belgium). Auto- ␤ mated PD using an electronic cycler was started later, total clearances of 2M and the protein-bound depending on patient preference. After an initial inhospital solute p-cresol were not unexpectedly less than training period of approximately 1 week, patients were seen those of the small water-soluble solutes urea at the outpatient clinic at 6- to 8-week intervals. At each nitrogen, creatinine, and phosphate. However, visit, clinical parameters and clearance data for urea nitro- the contribution of renal to total clearances was gen and creatinine were taken into account for therapy ␤ adjustments (adjustment of dry weight, medication, and considerably greater for 2M and p-cresol than dialysis prescription). Fluid balance was maintained by the 20 for the water-soluble molecules. Moreover, peri- use of high glucose concentrated solutions (Dianeal 3.86%; toneal clearance values of urea nitrogen, creati- Baxter) and/or the polyglucose icodextrin during 1 long nine, and phosphate correlated inversely with dwell per day (Extraneal; Baxter), as judged appropriate by their respective renal clearances, which was not the treating physician. ␤ the case for 2M and p-cresol. The findings Study Design generated the hypothesis that the well-accepted At outpatient visits 1 to 5, sampling was performed for the practice of adapting PD prescription to compen- ␤ calculation of urea nitrogen, creatinine, phosphate, 2M, and sate for the gradual loss of renal function may p-cresol clearances, as well as for the calculation of weight- succeed in increasing peritoneal clearances of normalized protein nitrogen appearance (nPNA). A midday small water-soluble solutes, but not middle mol- blood sample was obtained, and total amounts of urine and ecules and protein-bound solutes. However, the peritoneal drainage were collected during the preceding 24-hour period, weighed, and sampled. All samples were cross-sectional design of the study precluded stored at –80°C until analysis. Blood pressure and dry body firm conclusions on this issue. weight were measured at each occasion. The study was The present study aims to confirm the hypoth- approved by the Ethical Committee of the University Hospi- esis by prospectively investigating the time pro- tal Leuven. files of peritoneal, renal, and total clearances of Analytical Methods different types of uremic retention solutes in Urea nitrogen, creatinine, and phosphate were measured incident patients treated with PD. ␤ by means of standard laboratory techniques. 2M was quan- titated by means of rate nephelometry using an Immage METHODS Instrument (Beckman Coulter, Brea, CA). Total p-cresol (ie, combined free and protein-bound fraction) was analyzed by Patients means of gas chromatography-mass spectrometry (GC-MS) 20 We performed a single-center, longitudinal, observational technology, as described earlier. In summary, after depro- study. All patients with end-stage renal disease starting PD teinization (acid and heat), addition of the internal standard treatment at the Leuven University Hospital (Belgium) be- (2,6-dimethylphenol), and extraction (ethyl acetate), samples tween July 1, 2002, and December 1, 2003 (N ϭ 30), were were transferred to the GC-MS (Trace GC-MS; Thermofinni- informed about the study and asked for their consent. Four gan, San José, CA) for injection and separation on a column, of these patients refused to participate. Two of the remaining followed by identification of p-cresol by means of mass 26 patients were excluded from the final analysis because spectrometry. Quantitative results were obtained by means of the internal standard method and calculated as concentra- PD-related complications urged early transfer to hemodialy- 2 sis therapy, which resulted in an insufficient number of study tions. R of the calibration line was 0.998. The method has visits to enable proper evaluation. The present report de- low intra-assay and interassay variabilities (coefficients of variation, 3.33% and 5.30%, respectively). The detection scribes data for the 24 remaining patients. limit for measurement of p-cresol is 0.15 mg/L. Extraction Causes of end-stage renal disease were diabetic nephrop- efficiency is 91.5%. athy (n ϭ 1), polycystic kidney disease (n ϭ 1), glomeru- lar disease (n ϭ 8), tubulointerstitial disease (n ϭ 7), vascular disease (n ϭ 4), and unknown (n ϭ 3). Demo- Calculations graphic (age at start of PD therapy, sex, length) and Peritoneal, renal, and total clearances normalized to clinical data (medication and comorbidity at start of PD 1.73 m2 of body surface area (liters per week per 1.73 m2) therapy) were collected by reviewing medical records. were calculated for all solutes by means of direct determina- Comorbidity at the start of PD therapy was scored accord- tion from dialysis fluids, urine, and midday serum solute ing to Davies et al21 and reported as low, medium, or high concentrations. According to National Kidney Foundation– grade. Kidney Disease Outcomes Quality Initiative guidelines,22 514 BAMMENS ET AL

Table 1. Patient Characteristics as medians (and interquartile ranges) in the figures. The SAS version 8.02 (SAS Institute, Cary, NC) software No. of patients 24 program was used for statistical analysis. Age at start of PD (y) 49.4 Ϯ 3.8 (50, 21-79) RESULTS Men/women 12/12 RRF at first visit (L/wk/1.73 m2) 59.9 Ϯ 7.9 Patient Characteristics (52.3, 13.2-171.2) Patient characteristics are listed in Table 1.At Comorbidity at start 7/14/3 the start of PD therapy, age was 49.4 Ϯ 3.8 years (low/medium/high) (median, 50 years; minimum to maximum, 21 to Former renal replacement 3/6/15 Ϯ 2 therapy (hemodialysis/ 79 years) and RRF was 59.9 7.9 L/wk/1.73 m 2 transplantation/none) (median, 52.3 L/wk/1.73 m ; minimum to maxi- Switch to automated PD at some 22/2 mum, 13.2 to 171.2 L/wk/1.73 m2). In 15 pa- point (yes/no) tients, PD was the first mode of renal replace- ϭ Timing of switch to automated PD (n 22) ment therapy. Most patients (n ϭ 22) switched to (d from start of PD) 20.4 Ϯ 6.2 (6, 0-98) automated PD therapy after a short period on Use of icodextrin at some point 15/9 continuous ambulatory PD therapy. The use of (yes/no) icodextrin was started at some point during the Timing of icodextrin use (n ϭ 15) observation period in 15 patients. (d from start of PD) 114.7 Ϯ 26.1 (87, 13-340) Time Profiles of Volumes, Weight, and

NOTE. Data expressed as mean Ϯ SEM (median, mini- Blood Pressure mum to maximum). The first visit took place at 27.6 Ϯ 2.0 Follow-up time amounted to 7.1 Ϯ 0.2 months days (median, 24; minimum to maximum, 17 to 45) after (median, 7.2 months; minimum to maximum, 5.6 the start of PD. to 8.6 months). Like 24-hour urine volume, RRF (Fig 1) declined significantly with time (P ϭ RRF was estimated by calculating the arithmetic mean of 0.004 and P ϭ 0.001 for the LMM with log- renal urea nitrogen and creatinine clearances and expressed transformed values, respectively). There was a in liters per week per 1.73 m2. Peritoneal, renal, and total significant increase in 24-hour peritoneal drain- Kt/V were calculated only for urea nitrogen (Kt/V ). Body UN age volume (LMM P Ͻ 0.0001), reflecting both surface area was estimated by using the Du Bois and Du Bois method.23 The distribution volume of urea nitrogen the greater dialysate volume prescribed (LMM P (V) was assessed by using the formula of Watson et al24 ϭ 0.045) and increasing 24-hour ultrafiltration for total-body water. nPNA was calculated according to volume (LMM P ϭ 0.014). Dry body weight Bergström et al.25 increased significantly, whereas there were no Statistics Data are expressed as mean Ϯ SEM, median, and minimum to maximum. The linear mixed-model (LMM) approach was used to model time courses of clearances of the solutes studied. The LMM has the advantage that one can use observations from all subjects regardless of the exact timing of the observations and even if some data are missing. The time variable, defined as number of days from the start of PD treatment, was introduced as a fixed effect. By using random effects, the model allowed for interindividual variation of the intercept and slope of time courses. The linearity assumption, an essential condition for the use of the LMM, was fulfilled for all variables studied, except for those related to RRF and for total clearances of ␤ M and p-cresol. The latter showed an Fig 1. Time course of RRF, defined as the arith- 2 metic mean of renal urea nitrogen and creatinine clear- ؍ exponential, rather than a linear, time profile. For that 2 ␤ ance (liters per week per 1.73 m ) from visits 1 to 5 (N reason, all RRF-related data and total clearances of 2M 24). For display purposes, data were grouped per visit. and p-cresol were log-transformed for the LMM analysis. Median values and interquartile ranges are shown. P less than 0.05 is considered significant. Although exact Positions along the X-axis are defined as the corre- time observations were used in statistical analyses, for sponding visit’s median time after the start of PD display purposes, data were grouped per visit and shown treatment. TIME PROFILES OF CLEARANCES IN PD PATIENTS 515 significant changes in systolic or diastolic blood soluble solutes urea nitrogen, creatinine, and pressure. The intercept and slope of the time phosphate. However, peritoneal clearances of ␤ curves obtained by the LMM analysis for each of the middle molecule 2M and the protein- the variables are listed in Table 2. bound solute p-cresol did not change signifi- cantly. For the latter molecule, the resulting Time Profiles of Solute Clearances decline in total weekly clearance was statisti- Figures 2 to 6 show time courses of renal, cally significant. peritoneal, and total clearances of urea nitro- ␤ gen, creatinine, phosphate, 2M, and p-cresol Time Profiles of Kt/V and nPNA for all 24 patients from visits 1 to 5. In Table 2, UN corresponding data from the LMM analysis are The last section of Table 2 lists results of the listed. As expected, renal clearances of each LMM analysis for renal, peritoneal, and total solute studied declined significantly with dete- Kt/VUN and nPNA. As expected, Kt/VUN val- rioration in RRF. There was a compensatory ues followed exactly the same pattern as weekly increase in peritoneal clearances of the water- clearances of urea nitrogen. The slope of the

Table 2. Time Profiles

Variable Intercept Slope of Time Course LMM P

24-h urine volume (mL; log-transformed) 2.94 Ϫ0.0023 0.004 RRF (L/wk/1.73 m2; log-transformed) 1.74 Ϫ0.0025 0.001 24-h peritoneal drainage (mL) 10,618 ϩ11.74 Ͻ0.0001 24-h prescribed dialysate volume (mL) 9,877 ϩ3.76 0.045 24-h ultrafiltration volume (mL) 534 ϩ1.56 0.014 Dry body weight (kg) 67.2 ϩ0.01 0.025 Systolic blood pressure (mm Hg) 127 ϩ0.04 0.228 Diastolic blood pressure (mm Hg) 82 Ϫ0.002 0.880 Urea nitrogen (MW, 60 d) Renal clearance (L/wk/1.73 m2; log-transformed) 1.48 Ϫ0.0019 0.0002 Peritoneal clearance (L/wk/1.73 m2) 50.84 ϩ0.04 0.002 Total clearance (L/wk/1.73 m2) 85.25 Ϫ0.01 0.821 Creatinine (MW, 113 d) Renal clearance (L/wk/1.73 m2; log-transformed) 1.89 Ϫ0.0025 0.001 Peritoneal clearance (L/wk/1.73 m2) 30.94 ϩ0.04 Ͻ0.0001 Total clearance (L/wk/1.73 m2) 116.8 Ϫ0.12 0.023 Phosphate (MW, 96 d) Renal clearance (L/wk/1.73 m2; log-transformed) 1.52 Ϫ0.0024 0.001 Peritoneal clearance (L/wk/1.73 m2) 21.20 ϩ0.05 Ͻ0.0001 Total clearance (L/wk/1.73 m2) 57.46 ϩ0.02 0.457 ␤ 2M (MW, 11,815 d) Renal clearance (L/wk/1.73 m2; log-transformed) 0.88 Ϫ0.0047 0.0004 Peritoneal clearance (L/wk/1.73 m2) 5.99 ϩ0.01 0.188 Total clearance (L/wk/1.73 m2; log-transformed) 1.22 Ϫ0.0005 0.133 p-Cresol (MW, 108 d) Renal clearance (L/wk/1.73 m2; log-transformed) 1.03 Ϫ0.0022 Ͻ0.0001 Peritoneal clearance (L/wk/1.73 m2) 3.55 ϩ0.003 0.559 Total clearance (L/wk/1.73 m2; log-transformed) 1.15 Ϫ0.0008 0.007 Ϫ Ϫ Renal Kt/VUN (log-transformed) 0.07 0.0019 0.0002 ϩ Peritoneal Kt/VUN 1.46 0.001 0.005 Ϫ Total Kt/VUN 2.42 0.0001 0.867 nPNA (g/kg/d) 1.66 ϩ0.001 0.029

NOTE. N ϭ 24. Follow-up: 7.1 Ϯ 0.2 months (median, 7.2 months; minimum to maximum, 5.6 to 8.6 months). Intercept and slope of the time course in the 24 patients when modeled by means of LMM. The time variable is defined as number of days from the start of PD treatment. To avoid violation of the linearity assumption, some data had to be log-transformed, as indicated. Abbreviation: MW, molecular weight. 516 BAMMENS ET AL

Fig 2. Time course of renal (gray bars), peritoneal Fig 4. Time course of renal (gray bars), peritoneal (white bars), and total clearances (; liters per week (white bars), and total clearances (; liters per week .(24 ؍ per 1.73 m2) of phosphate from visits 1 to 5 (N .(24 ؍ per 1.73 m2) of urea nitrogen from visits 1 to 5 (N For display purposes, data were grouped per visit. For display purposes, data were grouped per visit. Median values are shown. Median values are shown. time course of nPNA was significantly posi- 2; Figs 2 to 4). Conversely, no relevant changes tive. in peritoneal clearances of the middle molecule ␤ 2M and the protein-bound solute p-cresol were DISCUSSION seen (Table 2; Figs 5 and 6). This longitudinal observational study includ- The increasing peritoneal clearances of the ing 24 incident PD patients describes time pro- small water-soluble molecules can be explained files of renal, peritoneal, and total weekly clear- by the synergism of 3 different mechanisms. ances of different types of uremic retention First, 24-hour prescribed dialysate volume in- solutes. As expected from the loss of RRF, renal creased significantly during the study period, clearances of all solutes showed a negative time reflecting treating physicians’ attempts to com- trend during the first 7 months of PD treatment pensate for the loss of dialysis adequacy second- (Table 2; Figs 2 to 6). However, the evolution of ary to the decline in RRF (Table 2). As a conse- peritoneal clearances was not identical for all quence, the concentration gradient is maintained solutes studied. Dialytic clearance values of the during a greater portion of the day, enhancing water-soluble molecules urea nitrogen, creati- peritoneal elimination of the small water-soluble nine, and phosphate increased significantly (Table molecules according to Fick’s first law of diffu-

Fig 3. Time course of renal (gray bars), peritoneal Fig 5. Time course of renal (gray bars), peritoneal (white bars), and total clearances (; liters per week (white bars), and total clearances (; liters per week ؍ ␤ 2 ؍ 2 per 1.73 m ) of creatinine from visits 1 to 5 (N 24). For per 1.73 m )of 2M from visits 1 to 5 (N 24). For display purposes, data were grouped per visit. Median display purposes, data were grouped per visit. Median values are shown. values are shown. TIME PROFILES OF CLEARANCES IN PD PATIENTS 517

bound solute p-cresol. It was shown earlier that the peritoneal membrane restriction coefficient for macromolecules increases with time on PD therapy.26,28,33,34 This may partly explain the ␤ lack of a relevant increase in peritoneal 2M clearance because it counterbalances the men- tioned potentially beneficial effects. Moreover, ␤ in a normal kidney, 2M elimination is accom- plished mainly by an active degradation process, which cannot be equaled by the peritoneum when RRF is lost.18 The same holds true for p-cresol and other protein-bound solutes, for Fig 6. Time course of renal (gray bars), peritoneal which the single most important elimination (white bars), and total clearances (; liters per week For mechanism in normal circumstances is active .(24 ؍ per 1.73 m2)ofp-cresol from visits 1 to 5 (N display purposes, data were grouped per visit. Median renal tubular secretion. Organic anion transport- values are shown. ers 1 and 3, located at the basolateral membrane of the proximal tubular cells, have been impli- sion. Second, as described earlier by several cated in the elimination of several representa- studies, structural changes in the peritoneal mem- tives of the protein-bound solutes.19,35 The brane caused by the exposition to dialysis fluid present data reaffirm the findings of our earlier components lead to increased transport of small 26-30 cross-sectional study showing that the contribu- solutes. These changes have been shown to tion of renal to total clearances was of consider- occur early in the course of PD treatment and ably greater importance for elimination of ␤ M thus may have contributed to the changes in 2 and p-cresol than for water-soluble molecules.20 dialytic small-solute clearances observed in the We acknowledge that the follow-up time of present study.30 Intuitively, one would expect the the present study was relatively short and our latter to cause a loss of ultrafiltration because of results do not necessarily reflect changes in clear- more rapid dissipation of the glucose-dependent 31,32 ances in the long term. Based on a report by osmotic driving force for fluid transport. 36 However, in the present study, there was a signifi- McKane et al, it can be expected that after an cant increase in 24-hour ultrafiltration volume initial abrupt decline in RRF, deterioration con- with time (Table 2). This paradox can be ex- tinues at a rate similar to predialysis initiation. plained because by adjustments of the dialysis The exponential time profiles of RRF and related regimen (time schedule, use of icodextrin), 24- clearances in the present study are in agreement hour ultrafiltration volume may have been kept with that finding. Analysis of 1-year data, avail- in line with the demands of fluid balance despite able for 11 patients of the present population, the membrane’s loss of ultrafiltration capacity. showed essentially the same results as the shorter On the other hand, a recent prospective survey follow-up data for the entire group of 24 patients by Davies30 showed that the increase in small- (data not shown). Nevertheless, information on solute transport is not associated with a decrease the evolution of clearances beyond the first treat- in ultrafiltration capacity during the first year of ment year would be of value, and a study to treatment. Whatever the reason for the increasing address this issue currently is being undertaken. 24-hour ultrafiltration volume, it is clear that the Detailed data on the use of nephrotoxic medi- concomitant convection may have contributed as cations and technical procedures during the a third mechanism to enhance the peritoneal study’s observation period were not available removal of small water-soluble solutes. and could not be taken into account in the statis- Although the mechanisms of enhanced diffu- tical analysis. However, although this may have sion and convection are relevant to the small influenced individual time courses of RRF, it water-soluble solutes, they apparently do not would not have biased our conclusions because influence time profiles of the peritoneal removal clearances of different solutes would have been ␤ of the middle molecule 2M and the protein- affected uniformly. 518 BAMMENS ET AL

Several recent studies indicated the value of 6. Davies SJ, Phillips L, Russell GI: Peritoneal solute RRF as a predictor of survival in PD patients.1-7 transport predicts survival on CAPD independently of re- This has been explained by its role in the mainte- sidual renal function. Nephrol Dial Transplant 13:962-968, 1998 nance of adequate fluid balance, its relationship 7. Paniagua R, Amato D, Vonesh E, et al: Effects of with inflammation and comorbidity, and its endo- increased peritoneal clearances on mortality rates in perito- crine functions.3,8-16 Considering existing and neal dialysis: ADEMEX, a prospective, randomized, con- emerging data on the toxicity of several middle trolled trial. J Am Soc Nephrol 13:1307-1320, 2002 molecules and protein-bound solutes,20,37,38 the 8. Wang T, Lindholm B: Beyond CANUSA, DOQI, findings of the present prospective study and our ADEMEX: What’s next? Perit Dial Int 22:555-562, 2002 9. Konings CJAM, Kooman JP, Schonck M, et al: Fluid former cross-sectional data may point to the preser- status in CAPD patients is related to peritoneal transport and vation of renal elimination mechanisms other than residual renal function: Evidence from a longitudinal study. glomerular filtration as an additional explanation Nephrol Dial Transplant 18:797-803, 2003 for the important role of RRF in PD patients. 10. Pecoits-Filho R, Goncalves S, Barberato SH, et al: In summary, our prospective data show that Impact of residual renal function on volume status in chronic renal failure. Blood Purif 22:285-292, 2004 attempts to increase the peritoneal clearance of a 11. Pecoits-Filho R, Heimbürger O, Barany P, et al: uremic solute to compensate for the gradual loss Associations between circulating inflammatory markers and of RRF can be achieved much easier for water- residual renal function in CRF patients. Am J Kidney Dis soluble solutes than for the middle molecule 41:1212-1218, 2003 ␤ 12. Wang AY, Wang M, Woo J, et al: Inflammation, 2M and the protein-bound solute p-cresol. These findings point to the preservation of renal elimi- residual kidney function, and cardiac hypertrophy are inter- related and combine adversely to enhance mortality and nation mechanisms other than glomerular filtra- cardiovascular death risk of peritoneal dialysis patients. tion as an additional explanation for the impor- J Am Soc Nephrol 15:2186-2194, 2004 tant role of RRF in PD patients and confirm our 13. Zappacosta AR, Caro J, Erslev A: Normalization of earlier cross-sectional data. hematocrit in patients with end-stage renal disease on con- tinuous ambulatory peritoneal dialysis: The role of erythro- poietin. Am J Med 72:53-57, 1982 ACKNOWLEDGMENT 14. Besarab A, Caro J, Jarrell BE, Francos G, Erslev AJ: The authors thank M. Dekens and H. de Loor for excellent Dynamics of erythropoiesis following renal transplantation. technical assistance and all participating patients and the Kidney Int 32:526-536, 1987 renal nurses and physicians taking care of them. 15. Erkan E, Moritz M, Kaskel F: Impact of residual renal function in children on hemodialysis. Pediatr Nephrol REFERENCES 16:858-861, 2001 1. Bargman JM, Thorpe KE, Churchill DN, for the 16. Lund B, Clausen E, Friedberg M, et al: Serum 1,25- CANUSA PD Study Group: Relative contribution of re- dihydroxycholecalciferol in anephric, haemodialyzed and sidual renal function and peritoneal clearance to adequacy of kidney-transplanted patients. Effect of vitamin D3 supple- dialysis: A reanalysis of the CANUSA study. J Am Soc ment. Nephron 25:30-33, 1980 Nephrol 12:2158-2162, 2001 17. Vanholder R, De Smet R, Glorieux R, et al: Review 2. Ates K, Nergizoglu G, Keven K, et al: Effect of fluid on uremic toxins: Classification, concentration, and interin- and sodium removal on mortality in peritoneal dialysis dividual variability. Kidney Int 63:1934-1943, 2003 ␤ patients. Kidney Int 60:767-776, 2001 18. Karlsson FA: Physical-chemical properties of 2- 3. Chung SH, Heimbürger O, Stenvinkel P, Qureshi AR, microglobulin. Immunochemistry 11:111-117, 1974 Lindholm B: Association between residual renal function, 19. Enomoto A, Takeda M, Tojo A, et al: Role of organic inflammation and patient survival in new peritoneal dialysis anion transporters in the tubular transport of indoxyl sulfate patients. Nephrol Dial Transplant 18:590-597, 2003 and the induction of its nephrotoxicity. J Am Soc Nephrol 4. Szeto CC, Wong TYH, Chow KM, Leung CB, Li PKT: 13:1711-1720, 2002 Are peritoneal dialysis patients with and without residual 20. Bammens B, Evenepoel P, Verbeke K, Vanrenterghem renal function equivalent for survival study? Insight from a Y: Removal of middle molecules and protein-bound solutes retrospective review of the cause of death. Nephrol Dial by peritoneal dialysis and relation with uremic symptoms. Transplant 18:977-982, 2003 Kidney Int 64:2238-2243, 2003 5. Termorshuizen F, Korevaar JC, Dekker FW, van Manen 21. Davies SJ, Phillips L, Naish PF, Russell GI: Quantify- JG, Boeschoten EW, Krediet RT for the NECOSAD Study ing comorbidity in peritoneal dialysis patients and its relation- Group: The relative importance of residual renal function ship to other predictors of survival. Nephrol Dial Transplant compared with peritoneal clearance for patient survival and 17:1085-1092, 2002 quality of life: An analysis of the Netherlands Cooperative 22. National Kidney Foundation: K/DOQI Clinical Prac- Study on the Adequacy of Dialysis (NECOSAD)-2. Am J tice Guidelines for Peritoneal Dialysis Adequacy: Update Kidney Dis 41:1293-1302, 2003 2000. Am J Kidney Dis 37:S65-S136, 2001 (suppl 1) TIME PROFILES OF CLEARANCES IN PD PATIENTS 519

23. Du Bois D, Du Bois EF: A formula to estimate the 31. Heimburger O, Waniewski J: Ultrafiltration failure in approximate surface area if height and weight be known. peritoneal dialysis patients. Perit Dial Int 24:506-508, 2004 Arch Intern Med 17:863-971, 1916 32. Rippe B, Venturoli D, Simonsen O, de Arteaga J: 24. Watson PE, Watson ID, Batt RD: Total body water Fluid and electrolyte transport across the peritoneal mem- volumes for adult males and females estimated from simple brane during CAPD according to the three-pore model. Perit anthropometric measurements. Am J Clin Nutr 33:27-39, 1980 Dial Int 24:10-27, 2004 25. Bergström J, Heimburger O, Lindholm B: Calcula- 33. Bouts AHM, Davin JC, Groothoff JW, Ploos van tion of the protein equivalent of total nitrogen appearance Amstel S, Zweers MM, Krediet RT: Standard peritoneal from urea appearance. Which formulas should be used? Perit permeability analysis in children. J Am Soc Nephrol 11:943- Dial Int 18:467-473, 1998 950, 2000 26. Struijk DG, Krediet RT, Koomen GC, et al: Func- 34. Ho-Dac-Pannekeet MM, Koopmans JG, Struijk DG, tional characteristics of the peritoneal membrane in long- Krediet RT: Restriction coefficients of low molecular weight term continuous ambulatory peritoneal dialysis. Nephron solutes and macromolecules during peritoneal dialysis. Adv 59:213-220, 1991 Perit Dial 13:72-76, 1997 27. Selgas R, Fernandez-Reyes MJ, Bosque E, et al: 35. Deguchi T, Kusuhara H, Takadate A, et al: Character- Functional longevity of the human peritoneum: How long is continuous peritoneal dialysis possible? Results of a prospec- ization of uremic toxin transport by organic anion transport- tive medium long-term study. Am J Kidney Dis 23:64-73, ers in the kidney. Kidney Int 65:162-174, 2004 1994 36. McKane W, Chandna SM, Tattershall JE, Greenwood 28. Struijk DG, Krediet RT, Koomen GC, Boeschoten RN, Farrington K: Identical decline of residual renal func- EW, Hoek FJ, Arisz L: A prospective study of peritoneal tion in high-flux biocompatible hemodialysis and CAPD. transport in CAPD patients. Kidney Int 45:1739-1744, 1994 Kidney Int 61:256-265, 2002 29. Davies SJ, Phillips L, Griffiths AM, Russell LH, Naish 37. Vanholder R, Argilés A, Baurmeister U, et al: Uremic PF, Russell GI: What really happens to people on long-term toxicity: Present state of the art. Int J Artif Organs 24:695- peritoneal dialysis? Kidney Int 54:2207-2217, 1998 725, 2002 30. Davies SJ: Longitudinal relationship between solute 38. De Smet R, Van Kaer J, Van Vlem B, et al: Toxicity of transport and ultrafiltration capacity in peritoneal dialysis free p-cresol: A prospective and cross-sectional analysis. patients. Kidney Int 66:2437-2445, 2004 Clin Chem 49:470-478, 2003 History of Cardiovascular Disease Is Associated With Endothelial Progenitor Cells in Peritoneal Dialysis Patients

Sabine Steiner, MD, Georg Schaller, MD, Heidi Puttinger, MD, Manuela Födinger, MD, Christoph W. Kopp, MD, Daniela Seidinger, Johannes Grisar, MD, Walter H. Hörl, MD, PhD, Erich Minar, MD, Andreas Vychytil, MD, Michael Wolzt, MD, and Gere Sunder-Plassmann, MD

● Background: It is unknown whether traditional cardiovascular disease risk factors influence the number of endothelial progenitor cells (EPCs) and whether numbers of EPCs correlate with endothelial function in patients with end-stage renal disease. Methods: In a cross-sectional study of 38 peritoneal dialysis patients, we examined numbers of circulating CD34؉/KDR؉/CD133؉ cells, CD34؉ hematopoietic stem cells, and EPCs cultured from peripheral blood. We also assessed conventional cardiovascular disease risk factors, such as history of vascular disease, diabetes, hypercholesterolemia, hypertension, and smoking. We determined endothelial function by measurement of endothelium-dependent and endothelium-independent reactivity of forearm resistance arteries by using strain-gauge plethysmography. Results: Numbers of EPCs cultured from peripheral blood and forearm blood flow reactivity did not differ between erythropoietin-treated peritoneal dialysis patients and healthy individuals. A history of vascular disease was associated with number of cultured EPCs, but other cardiovascular disease risk factors showed no association. Furthermore, there was no association of endothelial-dependent and endothelial- independent forearm blood flow reactivity with EPCs in peritoneal dialysis patients. Conclusion: In this first study of EPCs in peritoneal dialysis patients, we found an association between history of vascular disease and EPCs, but no association of EPCs with endothelial function or other cardiovascular disease risk factors. Am J Kidney Dis 46:520-528. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Endothelial progenitor cells; end-stage renal disease (ESRD); endothelial function; cardiovascular disease risk factors; peritoneal dialysis (PD).

ARDIOVASCULAR DISEASE morbidity with the general population.1 Major causes for C and mortality are greatly increased in pa- this excess risk for cardiovascular disease out- tients with end-stage renal disease compared comes include hypertension, diabetes, hyperlip- idemia, and smoking,2 as well as other nontradi- tional risk factors, such as increased C-reactive From the Department of Medicine II, Division of Angiol- protein and total homocysteine concentrations ogy; Department of Clinical Pharmacology; Department of and disturbances in calcium-phosphate homeosta- Medicine III, Division of Nephrology and Dialysis; Depart- 3,4 ment of Medicine III, Division of Rheumatology; and Clini- sis. In addition, endothelial function, an impor- cal Institute of Medical and Chemical Laboratory Diagnos- tant predictor of death in patients with atheroscle- tics, Medical University Vienna, Austria. rosis,5 also was impaired in patients with renal Received March 14, 2005; accepted in revised form May disease.6 9, 2005. A substantial body of evidence suggests that Originally published online as doi:10.1053/j.ajkd.2005.05.015 on July 22, 2005. bone marrow–derived endothelial progenitor cells Supported in part by a research grant from the Major of (EPCs) circulate in the blood, and EPCs are Vienna and institutional revenue. important players in the repair or formation of S.S., A.V., M.W., and G.S.-P. created the concept and design blood vessels.7,8 Number and migratory activity of this study; S.S., J.G., and C.W.K. performed endothelial of circulating EPCs correlated inversely with progenitor cell measurements; G.S. performed forearm blood 9 flow measurements and statistical analyses; H.P. and M.F. were risk factors for coronary artery disease. Further- responsible for collection of patient and laboratory data; S.S., more, number of EPCs was postulated to be a G.S., M.F., M.W., and G.S.-P. took part in analysis and interpre- surrogate biological marker for vascular func- tation of data; S.S. and G.S.-P. contributed to drafting of the tion, measured by means of flow-mediated bra- manuscript; and S.S., G.S., M.F.,A.V., M.W., E.M., W.H.H., and chial artery reactivity and cumulative cardiovas- G.S.-P. performed critical revision of the manuscript for impor- 10 11 tant intellectual content. cular risk. Recently, de Groot et al showed Address reprint requests to Gere Sunder-Plassmann, MD, that the number of EPCs was decreased in pa- Division of Nephrology and Dialysis, Department of Medi- tients with renal failure, with normalization after cine III, Medical University Vienna, Währinger Gürtel 18- initiation of renal replacement therapy,11 possi- 20, A-1090 Vienna, Austria. E-mail: gere.sunder-plassmann@ bly indicating a novel mechanism for the in- meduniwien.ac.at © 2005 by the National Kidney Foundation, Inc. crease in cardiovascular disease burden in pa- 0272-6386/05/4603-0017$30.00/0 tients with chronic kidney disease. Furthermore, doi:10.1053/j.ajkd.2005.05.015 therapy with erythropoiesis-stimulating agents

520 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 520-528 ENDOTHELIAL PROGENITOR CELLS IN ESRD 521 has the potential to increase the number of EPCs EPC culture assay. Immediately after isolation, mono- in patients with end-stage renal disease.12-14 nuclear cells (4 ϫ 106) were resuspended in 1 mL of Associations between cardiovascular disease endothelial basal medium (EBM-2; Clonetics, Cambrex, East Rutherford, NJ) supplemented with EGM-2-MV- risk factors and EPCs in patients with renal SingleQuots (Clonetics) and 10% fetal calf serum (Gibco, failure currently are unknown. The aim of the Life Technologies, Carlsbad, CA) and plated on 24-well present study is to test the hypothesis that num- culture dishes (TTP; Trasadingen, Schweiz) coated with ber of circulating EPCs is related to cardiovascu- human fibronectin (Sigma, St Louis, MO). After 3 days of lar disease risk factors and endothelial function culture, nonadherent cells were removed. Day 7, adherent cells underwent cytochemical analysis after a thorough wash- in patients with end-stage renal disease. ing with phosphate-buffered saline. To confirm the endothe- lial cell phenotype, cells were incubated with 2.4 ␮g/mL of METHODS 1,1’-dioctadecyl-3,3,3’,3’-tetramethylindo-carbocyanine per- Design and Patients chlorate–labeled acetylated low-density lipoprotein (Bio- medical Technologies, Stoughton, MA) in fresh medium at We performed a cross-sectional study of 38 consecutive 37°C for 3 hours. After fixation in 2% paraformaldehyde, peritoneal dialysis patients at the University Hospital of cells were counterstained with fluorescein isothiocyanate– Vienna, Austria. The Ethical Review Board at the University conjugated Ulex europaeus agglutinin-1 lectin (10 ␮g/mL; of Vienna approved the study. All patients provided written Sigma) for 1 hour at room temperature. Samples were informed consent. examined with an inverted fluorescent microscope, and Participants had to be clinically stable, older than 18 adherent cells staining positive for both 1,1’-dioctadecyl- years, treated with peritoneal dialysis for at least 3 months, 3,3,3’,3’-tetramethylindo-carbocyanine perchlorate–labeled and free from peritonitis within 2 months before the study acetylated low-density lipoprotein and lectin were consid- began. Exclusion criteria were overtreatment with erythro- ered EPCs.9 Number of EPCs was determined by averaging poietin (serum hemoglobin Ͼ 14.0 g/dL [Ͼ140 g/L]), over- EPC counts from 3 random high-power fields per subject by treatment with intravenous or oral iron (serum ferritin Ͼ 500 2 independent investigators. ng/mL [␮g/L]), inclusion in another study within 4 weeks Flow cytometry analysis. A total of 2 ϫ 106 mono- before study enrollment, poor general condition of health or nuclear cells was incubated for 30 minutes in the dark at 4°C infection, acute vascular disease (history of stroke, myocar- with saturating concentrations of the following monoclonal dial infarction, or peripheral occlusive disease within 3 antibodies (mAbs): phycoerythrin-labeled anti-CD133 mAb months before the day of study), and pregnancy. Coronary (clone AC133; Miltenyi Biotec, Auburn, CA), allophycocya- artery disease is defined as a history of percutaneous or nin-labeled anti-CD34 mAb (clone 581, Instrumentation surgical revascularization and/or myocardial infarction. Pe- Laboratories, Marseille, France), biotinylated anti-KDR mAb ripheral arterial disease is defined as a history of percutane- (clone KDR-2; Sigma), and corresponding isotype controls ous or surgical revascularization and/or gangrene, and cere- (mouse immunoglobulin G1-biotinylated, clone MOPC21, brovascular arterial disease is defined as a history of cerebral mouse immunoglobulin G1-phycoerythrin, clone 679.1Mc7, stroke. In the majority of patients, coronary angiography or mouse immunoglobulin G1-allophycocyanin, clone stress tests, duplex ultrasound of carotid and lower limb 679.1Mc7, all from Instrumentation Laboratories). After a arteries, and ankle-brachial index tests were performed to single wash step with phosphate-buffered saline, samples examine vascular disease burden. Recruitment of patients were incubated for 15 minutes with phycoerythrin-Cy5– lasted 4 months in 2003 and 2004. conjugated streptavidin (Instrumentation Laboratories) at room temperature. Then cells were washed with phosphate- Measurements buffered saline and fixed in 2% paraformaldehyde. Patients were asked to fast for 10 or more hours before Quantitative analysis was performed on a FACS-Calibur measurement of vascular function. Smokers were requested flow cytometer (Becton Dickinson, San Jose, CA) measur- to refrain from smoking for 12 hours before the study. ing 200,000 cells/sample (Fig 1). Data were analyzed using Studies were conducted in a quiet room with an ambient Cellquest software (Becton Dickinson) by side scatter– temperature of 22°C with complete resuscitation facilities. fluorescence dot plot analysis. Number of EPCs is defined as Blood was drawn for blood chemistry and EPC assays while events triple-positive for CD34, CD133, and KDR with low ϩ participants were still in a fasting state on the same day cytoplasmatic granularity (low sideward scatter). CD34 immediately after measurement of endothelial function. cells are defined as hematopoietic stem cells. Endothelial progenitor cells. This study of EPCs in- cluded: (1) analysis of EPCs grown from peripheral-blood Forearm Blood Flow Measurements mononuclear cells within 7 days in cell culture, and (2) flow Forearm blood flow was measured in both arms as de- cytometry analysis of peripheral-blood CD34ϩ cells, CD34ϩ/ scribed previously.15,16 Briefly, strain gauges, placed on the KDRϩ cells, and CD34ϩ/KDRϩ/CD133ϩ cells. forearms, were connected to plethysmographs (EC-6; D.E. Isolation of mononuclear cells. Mononuclear cells were Hokanson, Bellevue, WA) to measure changes in forearm isolated by means of density-gradient centrifugation with volume in response to inflation of venous congesting cuffs. Ficoll-Paque Plus (Amersham Bioscience, Buckingham- Drug effects are expressed as the ratio of blood flow in the shire, UK) from EDTA-anticoagulated blood (5 ϫ 10–2 intervention to the control arm,16,17 for which predose ratio mol/L final concentration). is defined as 100%. Wrist cuffs were inflated to suprasystolic 522 STEINER ET AL

Fig 1. Gating strategies to detect EPCs by means of flow cytometry; a representative plot is shown. CD34؉ cells were determined by means of sideward scatter–fluorescence dot plot analysis within the mononuclear cell population. Acquisition was stopped after 200,000 events (A; the circle defines the gate with CD34؉ cells). These CD34؉ cells were characterized further by coexpression of CD133 and KDR (B; right upper quadrant; quadrants were set on the basis of isotype controls). pressures during each measurement to exclude circulation to residual renal function were calculated as described previ- the hands. Flow measurements were recorded for 9 seconds ously.18 at 30-second intervals during drug infusions. A fine-bore needle (27-G needle; Sterican, B. Braun, Statistical Methods Melsungen, Germany) was inserted into the brachial artery Continuous data are summarized as medians and interquar- of the left arm for administration of vasodilators. After a tile ranges (IQRs), and counts and percentages are given for 10-minute resting period, forearm blood flow during infu- categorical variables. Forearm blood flow measurements sion of 0.9% saline was measured for 5 minutes, followed by were calculated as mL·min–1·100 mL–1 forearm volume and measurements of forearm blood flow response to incremen- expressed as percentage of change from predose ratio. Asso- tal doses of the endothelium-dependent dilator acetylcholine ciations of cardiovascular disease risk factors and endothe- (Clinalfa; Läufelfingen, Switzerland; 25, 50, and 100 lial function with EPC counts were assessed by means of –1 nmol·min for 3 min/dose level). After a 15-minute wash- simple linear regression analysis. out period with intra-arterial saline infusions to allow resto- The Statistica software package (release 6.0; StatSoft Inc, ration of predose blood flow, the response to the endothelium- Tulsa, OK) was used for all analyses. independent dilator glyceryl trinitrate was measured (Perlinganit, Gebro Pharma, Fieberbrunn, Austria; 4, 8, 16 RESULTS mmol·min–1 for 3 min/dose level). After another 15-minute washout period, this was repeated for the endothelium- Patients and Cardiovascular Disease dependent vasoconstrictor and nitric oxide synthase–inhibi- Risk Factors ␮ –1 tor L-N-monomethyl-arginine (Clinalfa; 1, 2, 4 mol·min Important demographic and clinical character- for 3 min/dose level). Subsequently, a plastic cannula was inserted into an istics, as well as biochemical data for the 38 antecubital or dorsal hand vein for blood draw. Arterial patients, are listed in Table 1. Four patients had a blood pressure and pulse rate were recorded during measure- failed kidney graft. Thirty-six patients were ment of forearm blood flow. treated for arterial hypertension (8 patients, angio- Biochemical Methods tensin II receptor blockers; 16 patients, angioten- sin-converting enzyme inhibitors; 6 patients, both Complete blood counts, as well as blood and dialysate chemistry, were performed by using standard procedures in classes of antihypertensives), 21 patients had an International Standards Organization 9001/2000–certi- uncontrolled hypertension (systolic and/or dia- fied laboratory at the Medical University Vienna. Kt/V and stolic blood pressure Ն 130 mm Hg and/or Ն 80 ENDOTHELIAL PROGENITOR CELLS IN ESRD 523

Table 1. Characteristics of 38 Peritoneal showed no or not significant stenoses (Ͻ50% Dialysis Patients narrowing of the vessel lumen) of coronary arter- Age (y) 56 (46-68) ies. Eleven patients had a negative cardiac radio- Men 28 (74) nuclide imaging study result; 3 patients had a Body mass index (kg/m2) 24.9 (22.3-28.1) negative exercise test result. In 8 other asymptom- Native kidney disease atic patients, invasive or noninvasive testing for Glomerulonephritis 12 (32) Polycystic kidney disease 5 (13) coronary artery disease was not performed. Diabetic nephropathy 4 (11) Cerebrovascular disease. Two patients had a Various others 5 (13) history of ischemic stroke, and 1 patient had Unknown 12 (32) undergone a carotid endarterectomy (6 to 12 Duration of dialysis treatment (mo) 18 (7-23) CAPD/APD 16 (42)/22 (58) years before the study). Among the other 35 Weekly Kt/V 2.47 (2.17-3.04) patients, 26 patients showed a normal finding or Residual renal function (mL/min) 5.70 (3.71-8.14) nonsignificant narrowing of the extracranial arter- Systolic blood pressure (mm Hg) 128 (120-135) ies by means of Doppler ultrasonography. Nine Diastolic blood pressure (mm Hg) 78 (74-80) Diabetes mellitus 7 (18) other asymptomatic patients did not undergo History of vascular disease 11 (29) imaging studies of the extracranial arteries. Current smoking 8 (21) Peripheral artery disease. No patient pre- Total cholesterol (mg/dL) 190 (168-215) sented with clinically relevant peripheral artery HDL cholesterol (mg/dL) 44 (35-55) disease (no intermittent claudication, no is- LDL cholesterol (mg/dL) 114 (93-139) Triglycerides (mg/dL) 146 (99-197) chemic nonhealing ulcers, no marked decrease in C-Reactive protein Ͼ 0.5 mg/dL 16 (42) temperature of the lower leg). Twenty-three pa- Hemoglobin (g/dL) 12.1 (11.5-12.8) tients had an ankle-brachial index test result Serum bicarbonate (mmol/L) 23.5 (22.4-24.7) within normal ranges19 or a Doppler ultrasonog- Calcium ϫ phosphorus ion product 4.34 (3.90-4.75) raphy study of the lower limb that showed no relevant stenosis. Three patients showed slight NOTE. Values expressed as median (IQR) or count impairment of the ankle-brachial index, and 12 (percent). To convert residual renal function in mL/min to patients had no examination of the peripheral mL/s, multiply by 0.01667; total, HDL, and LDL cholesterol in mg/dL to mmol/L, multiply by 0.02586; triglycerides in arteries. None of the 38 patients was classified to mg/dL to mmol/L, multiply by 0.01129; hemoglobin in g/dL have relevant peripheral artery disease. to g/L, multiply by 10. Abbreviations: CAPD, continuous ambulatory peritoneal Endothelial Progenitor Cells dialysis; APD, automated peritoneal dialysis. Numbers of cultured EPCs, circulating CD34ϩ/ CD133ϩ/KDRϩ cells, and peripheral hematopoi- mm Hg), and 35 patients were administered etic stem cells (CD34ϩ) of peritoneal dialysis erythropoiesis-stimulating agents. Nineteen pa- patients are listed in Table 2. Healthy individuals tients were administered statins, and 7 patients (n ϭ 37; 19 women, 18 men; median age, 49.3 were administered vitamin supplements. years; IQR, 40.0 to 59.7) had a similar number of History of Vascular Disease cultured EPCs (median, 52.0 cells/high-power field; IQR, 25.2 to 64.0; not significant). A history of vascular disease was present in 11 study participants (29%). Coronary artery disease. At the time of the Table 2. EPCs of Peritoneal Dialysis Patients study, none of 38 patients reported angina. Of 8 Median (IQR) patients with a history of myocardial infarction (3 to 18 years before the study), 4 patients EPCs (cells/high-power field) 54.7 (22.7-132.3) previously underwent coronary artery bypass graft CD34ϩ (cells/2 ϫ 105 MNCs) 112 (71-178) ϩ ϩ surgery; 3 patients, a percutaneous coronary inter- CD34 /KDR (cells/2 ϫ 105 MNCs) 2.0 (0.0-9.5) CD34ϩ/KDRϩ/CD133ϩ (cells/2 ϫ 105 vention; and 1 patient, both procedures. Thirty MNCs) 1.5 (0.0-8.5) patients were judged to be free of coronary artery disease: in 8 patients, coronary angiography Abbreviation: MNCs, mononuclear cells. 524 STEINER ET AL

Table 3. Univariate Associations of Cardiovascular Disease Risk Factors With EPCs of Peritoneal Dialysis Patients

Circulating EPCs Cultured EPCs (CD34ϩ/CD133ϩ/KDRϩ)

rPrP

Age 0.1008 0.564 0.1744 0.309 Sex 0.1453 0.405 0.0905 0.599 Body mass index Ϫ0.2329 0.178 Ϫ0.0507 0.769 Dialysis vintage Ϫ0.1655 0.342 Ϫ0.2374 0.163 History of cardiovascular disease Ϫ0.3401 0.046 0.0497 0.773 Systolic blood pressure 0.0152 0.931 0.1941 0.257 Diastolic blood pressure Ϫ0.1535 0.379 0.2417 0.156 Uncontrolled hypertension Ϫ0.1374 0.431 0.2368 0.164 Diabetes mellitus 0.0279 0.874 Ϫ0.1401 0.415 Smoking 0.0542 0.757 Ϫ0.0548 0.751 Serum cholesterol Ϫ0.0531 0.762 0.0628 0.716 HDL cholesterol Ϫ0.1459 0.403 0.2534 0.136 LDL cholesterol Ϫ0.0859 0.624 0.0601 0.728 Triglycerides 0.3295 0.053 Ϫ0.1762 0.304 C-Reactive protein Ͼ normal 0.0205 0.907 Ϫ0.1779 0.299 Statin use Ϫ0.2415 0.162 Ϫ0.0695 0.687 Erythropoietin use 0.1590 0.362 0.0806 0.640 Angiotensin II receptor blocker use 0.0452 0.797 0.0660 0.702 Angiotensin-converting enzyme inhibitor use Ϫ0.2650 0.124 0.0038 0.982 Forearm blood flow, acetylcholine Ϫ0.2286 0.201 Ϫ0.1441 0.424 Forearm blood flow, glyceryl trinitrate Ϫ0.0113 0.950 Ϫ0.0182 0.920 Forearm blood flow, L-N-monomethyl-arginine 0.0739 0.683 Ϫ0.2824 0.111

Endothelial Function Basal forearm blood flow was 3.22 mL·min–1· 100 mL–1 (IQR, 2.74 to 3.93). Acetylcholine caused a dose-dependent increase in forearm blood flow to a maximum of 228% (IQR, 180 to 332). The endothelium-independent vasodilator glyceryl trinitrate increased resting forearm blood flow to 210% (IQR, 169 to 277), and the nitric oxide synthase–inhibitor L-N-monomethyl-argi- nine decreased resting forearm blood flow to 70% (IQR, 63 to 81). No clinically relevant adverse event occurred during measurement of endothelial function.

Association of Cardiovascular Disease Risk Factors and Endothelial Function With EPCs Results of simple linear regression analyses are listed in Table 3. A history of cardiovascular dis- ease was associated with EPC count (r ϭ –0.3401; P ϭ 0.046; Fig 2). All other variables (age, sex, body mass index, statin use, erythropoiesis-stimu- lating agent use, angiotensin II receptor blocker Fig 2. Association of history of vascular disease ؍ use, angiotensin-converting enzyme inhibitor use, with EPC counts in 35 peritoneal dialysis patients (P time since initiation of peritoneal dialysis therapy, 0.016 by means of t-test). ENDOTHELIAL PROGENITOR CELLS IN ESRD 525

tients with end-stage renal disease, we found an inverse association of history of vascular disease with EPC number in erythropoietin-treated peri- toneal dialysis patients, whereas this was not the case for other conventional cardiovascular dis- ease risk factors. We observed no association of endothelial-dependent and endothelial-indepen- dent forearm blood flow reactivity with either circulating or cultured EPCs. EPCs are bone marrow–derived cells that are able to mature into the cell type that constitutes the lining for blood vessels in the body.7 It is the current understanding that EPCs have a crucial role in rapid re-endothelialization after injury and in neovascularization.8,20 The number or functional activity of EPCs may depend on the presence or absence of several cardiovascular disease risk factors, including smoking, hyperten- sion, diabetes, or hyperlipidemia in patients with and without coronary artery disease9,10 and on overall kidney function.11 Furthermore, statin 21 12–14 Fig 3. Lack of association of endothelium-dependent use or erythropoietic therapy was shown forearm blood flow with EPC counts in 35 peritoneal to mobilize EPCs from bone marrow and en- dialysis patients. hance the functional capabilities of these cells. In our analysis, history of vascular disease and blood pressure, presence of uncontrolled hyperten- duration of dialysis treatment showed an inverse sion, diabetes mellitus, smoking status, residual association with cultured EPCs and CD34ϩ he- renal function, Kt/V, and serum levels of triglycer- matopoietic stem cells, respectively. Conversely, ides, cholesterol, low-density lipoprotein [LDL] important cardiovascular disease risk factors (eg, cholesterol, high-density lipoprotein [HDL] choles- smoking, serum cholesterol level, diabetes, or terol, and C-reactive protein level) showed no asso- blood pressure) showed no such associations. ciation with EPCs. Furthermore, we observed no association of car- Duration of dialysis treatment was associated diovascular disease risk factors with circulating with number of peripheral CD34ϩ cells (r ϭ CD34ϩ/KDRϩ/CD133ϩ cells. –0.4087; P ϭ 0.013), but not with circulating Our finding of no association of serum lipid and cultured EPCs. There was no association of fraction levels with number of EPCs is in line EPCs with number of peripheral CD34ϩ cells with the lack of association of LDL cholesterol (r ϭ 0.2149; P ϭ 0.215). There also was no levels with EPCs in patients with stable coronary association of the aforementioned variables with artery disease or acute coronary syndromes/ numbers of peripheral CD34ϩ/CD133ϩ/KDRϩ myocardial infarction.9 Conversely, serum cho- cells or CD34ϩ cells. lesterol level was associated inversely with EPCs We also observed no association of cultured in men without a history of cardiovascular dis- EPCs, CD34ϩ cells, or CD34ϩ/CD133ϩ/KDRϩ ease.10 cells with endothelial-dependent and endothelial- Dimmeler et al22 previously showed that stat- independent forearm vasodilatation. The lack of ins increase the number of differentiated EPCs in correlation of endothelial-dependent forearm vitro. These cells also proliferate, migrate, and blood flow with EPC count is shown in Fig 3. acquire resistance to apoptotic cell death during statin treatment.23 Statin therapy also can in- DISCUSSION crease the number and migratory capacity of In this first study of the association of EPCs circulating EPCs in patients with stable coronary with cardiovascular disease risk factors in pa- artery disease21 and improves the differentiation 526 STEINER ET AL of EPCs into cardiomyogenic cells in such pa- vascular disease risk factors with EPCs cannot tients.24 In our study, we observed no association be extended to erythropoietin-naive and/or under- of statin use with number of EPCs. This finding dialyzed patients. may be related to the cross-sectional design of In contrast to a previous study of healthy our study. Therefore, randomized controlled tri- individuals,10 we did not observe an association als are needed to further clarify this important of EPC number with endothelial function in issue. peritoneal dialysis patients, a finding that may fit An interesting study of individuals without a into the concept of a reverse epidemiology of history of cardiovascular disease showed that cardiovascular disease risk factors in patients levels of colony-forming units of circulating EPCs with end-stage renal disease.35,36 are a better predictor of vascular function (mea- We did not find an association of EPCs with sured by means of brachial artery reactivity) than other important cardiovascular risk factors in our the presence or absence of conventional risk patients with end-stage renal disease. For ex- factors.10 In view of the predictive role of endo- ample, smoking, which showed a strong associa- thelial function on cardiovascular disease out- tion with numbers of EPCs and circulating comes,5,25 the investigators suggested that endo- CD34ϩ/KDRϩ cells in patients with coronary thelial injury in the absence of sufficient artery disease,9 showed no correlation with EPCs circulating EPCs might affect the progression of in our study. The lack of association with fasting cardiovascular disease. In patients with renal plasma glucose level in our study is in line with failure, vascular function frequently is im- findings of others showing that hyperglycemia paired26,27 and does not improve in response to showed no correlation with EPCs in patients various pharmacological interventions, such as with coronary artery disease.9 However, a mod- 28–30 31 folic acid or L-arginine, but responds to est association of blood glucose level or presence administration of vitamin C.32 of diabetes with EPCs also was reported in There was no difference in basal forearm individuals without cardiovascular disease.10 blood flow in peritoneal dialysis patients (3.46 Although recombinant C-reactive protein lev- mL·min–1·100 mL–1) compared with healthy els diminished the functional activity of EPCs in individuals from one of our previous studies vitro,37 our study and the work of others failed to (3.9 mL·min–1·100 mL–1),33 and there also was find an association of serum C-reactive protein no impairment of endothelial-dependent fore- levels with EPCs.11,12 Although serum C-reac- arm vasodilation in our patients compared with tive protein is an independent predictor of mor- healthy individuals from our previous study.33 bidity and mortality in patients with end-stage This well-preserved endothelial function of the renal disease,38 it appears that modest C-reactive patients included in our study most likely is protein level elevations have no major effect on related to the young age, lack of a history of EPCs in these patients. vascular disease in several patients, and statin We did not observe an association of blood use in a significant proportion of study partici- pressure or uncontrolled hypertension with EPCs pants. In this context, levels of asymmetrical in our study, a finding in line with observations dimethylarginine, a potent inhibitor of endothe- of others who did not find a reduction in number lial nitric oxide synthase, were normal in peri- of EPCs in the presence of hypertension in 45 toneal dialysis patients.34 patients with coronary artery disease.9 In another Most of our patients were administered eryth- study of men without a history of cardiovascular ropoiesis-stimulating agents shown previously to disease, hypertension also showed no correlation improve the number and function of EPCs in with EPCs in multivariate modeling.10 Con- patients with renal disease.12–14 Thus, the num- versely, EPC migration was diminished by the ber of EPCs of our patients was similar to that of presence of hypertension, shown in multivariate healthy individuals examined in our laboratory analysis including important risk factors in pa- (median, 52.0 cells/high-power field; IQR, 25.2 tients with coronary artery disease.9 However, to 64.0). This finding also might be related to the EPC migration was not examined in our study. high dose of dialysis delivered to our patients. The use of angiotensin II receptor blockers re- Therefore, our finding of no association of cardio- cently was shown to increase the number of ENDOTHELIAL PROGENITOR CELLS IN ESRD 527

EPCs in patients with diabetes mellitus type 2.39 6. van Guldener C, Lambert J, Janssen MJ, Donker AJ, In our study of peritoneal dialysis patients, we Stehouwer CD: Endothelium-dependent vasodilatation and distensibility of large arteries in chronic haemodialysis pa- did not observe such an association. tients. Nephrol Dial Transplant 12:S14-S18, 1997 (suppl 2) Limitations to our study include the cross- 7. Rosenzweig A: Endothelial progenitor cells. N Engl sectional design, which may have introduced J Med 348:581-582, 2003 time-related biases. The small set of 38 patients 8. Urbich C, Dimmeler S: Endothelial progenitor cells. included in this study may have resulted in Characterization and role in vascular biology. Circ Res 95:343-353, 2004 limited power to discriminate between catego- 9. Vasa M, Fichtlscherer S, Aicher A, et al: Number and ries and levels of exposure and be the reason for migratory activity of circulating endothelial progenitor cells the finding of no associations. It was shown that inversely correlate with risk factors for coronary artery EPC number or function is related to cumulative disease. Circ Res 89:E1-E7, 2001 cardiovascular risk, estimated by means of the 10. Hill JM, Zalos G, Halcox JPJ, et al: Circulating 10 endothelial progenitor cells, vascular function, and cardiovas- Framingham risk score. We did not calculate cular risk. N Engl J Med 348:593-600, 2003 this score because it underestimates cardiovascu- 11. de Groot K, Bahlmann FH, Sowa J, et al: Uremia lar disease risk in patients with end-stage renal causes endothelial progenitor cell deficiency. Kidney Int disease.40 At present, mechanisms by which car- 66:641-646, 2004 12. Heeschen C, Aicher A, Lehmann R, et al: Erythropoi- diovascular disease risk factors reduce EPC num- etin is a potent physiologic stimulus for endothelial progeni- 20 ber and function remain to be elucidated. Our tor cell mobilization. Blood 102:1340-1346, 2003 study also does not provide an explanation for 13. Bahlmann FH, de Groot K, Duckert T, et al: Endothe- the lack of association of risk factors with num- lial progenitor cell proliferation and differentiation is regu- lated by erythropoietin. Kidney Int 64:1648-1652, 2003 bers of EPCs in peritoneal dialysis patients. How- 14. Bahlmann FH, de Groot K, Spandau JM, et al: Eryth- ever, it was speculated that increased apoptosis ropoietin regulates endothelial progenitor cells. Blood 103: or interference with pathways regulating differen- 921-926, 2004 tiation and mobilization of EPCs might be opera- 15. Hokanson DE, Sumner DS, Strandness DE Jr: An elec- tional in patients without advanced kidney insuf- trically calibrated plethysmograph for direct measurement of 9 limb blood flow. IEEE Trans Biomed Eng 22:25-29, 1975 ficiency. 16. Pleiner J, Heere-Ress E, Langenberger H, et al: Adre- In summary, our study provides first evidence noceptor hyporeactivity is responsible for Escherichia coli that a history of vascular disease is associated endotoxin-induced acute vascular dysfunction in humans. with EPCs in erythropoietin-treated patients with Arterioscler Thromb Vasc Biol 22:95-100, 2002 17. Petrie JR, Ueda S, Morris AD, Murray LS, Elliott HL, end-stage renal disease and for a lack of associa- Connell JM: How reproducible is bilateral forearm plethys- tion of conventional vascular disease risk factors mography? Br J Clin Pharmacol 45:131-139, 1998 and endothelial function with EPCs in these 18. Vychytil A, Födinger M, Pleiner J, et al: Acute effect patients. Future studies should explore the asso- of amino acid peritoneal dialysis solution on vascular func- ciation of EPCs with cardiovascular disease out- tion. Am J Clin Nutr 78:1039-1045, 2003 19. Rutherford RB, Baker JD, Ernst C, et al: Recom- comes, as well as the effect of drug therapy on mended standards for reports dealing with lower extremity EPCs in patients with end-stage renal disease. ischemia: Revised version. J Vasc Surg 26:517-538, 1997 20. Szmitko PE, Fedak PWM, Weisel RD, Stewart DJ, REFERENCES Kutryk MJB, Verma S: Endothelial progenitor cells: New hope for a broken heart. Circulation 107:3093-3100, 2003 1. Foley RN, Parfrey PS, Sarnak MJ: Epidemiology of 21. Vasa M, Fichtlscherer S, Adler K, et al: Increase in cardiovascular disease in chronic renal disease. J Am Soc circulating endothelial progenitor cells by statin therapy in Nephrol 9:S16-S23, 1998 (suppl 2) patients with stable coronary artery disease. Circulation 2. Khot UN, Khot MB, Bajzer CT, et al: Prevalence of 103:2885-2890, 2001 conventional risk factors in patients with coronary heart 22. Dimmeler S, Aicher A, Vasa M, et al: HMG-CoA disease. JAMA 290:898-904, 2003 reductase inhibitors (statins) increase endothelial progenitor 3. Hackam DG, Anand SS: Emerging risk factors for cells via the PI 3-kinase/Akt pathway. J Clin Invest 108:391- atherosclerotic vascular disease: A critical review of the 397, 2001 evidence. JAMA 290:932-940, 2003 23. Llevadot J, Murasawa S, Kureishi Y, et al: HMG- 4. Zoccali C, Mallamaci F, Tripepi G: Novel cardiovascu- CoA reductase inhibitor mobilizes bone marrow-derived lar risk factors in end-stage renal disease. J Am Soc Nephrol endothelial progenitor cells. J Clin Invest 108:399-405, 15:S77-S80, 2004 (suppl 1) 2001 5. Vita JA, Keaney JF Jr: Endothelial function: A barom- 24. Rupp S, Badorff C, Koyanagi M, et al: Statin therapy eter for cardiovascular risk? Circulation 106:640-642, 2002 in patients with coronary artery disease improves the im- 528 STEINER ET AL paired endothelial progenitor cell differentiation into cardio- 33. Pleiner J, Mittermayer F, Schaller G, MacAllister RJ, myogenic cells. Basic Res Cardiol 99:61-68, 2004 Wolzt M: High doses of vitamin C reverse Escherichia coli 25. Endemann DH, Schiffrin EL: Endothelial dysfunc- endotoxin-induced hyporeactivity to acetylcholine in the tion. J Am Soc Nephrol 15:1983-1992, 2004 human forearm. Circulation 106:1460-1464, 2002 26. van Guldener C, Janssen MJ, Lambert J, Steyn M, 34. Kielstein JT, Boger RH, Bode-Boger SM, et al: Asym- Donker AJM, Stehouwer CDA: Endothelium-dependent va- metric dimethylarginine plasma concentrations differ in pa- sodilatation is impaired in peritoneal dialysis patients. Neph- tients with end-stage renal disease: Relationship to treatment rol Dial Transplant 13:1782-1786, 1998 method and atherosclerotic disease. J Am Soc Nephrol 27. Morris ST, McMurray JJ, Spiers A, Jardine AG: 10:594-600, 1999 Impaired endothelial function in isolated human uremic 35. Kalantar-Zadeh K, Block G, Humphreys MH, resistance arteries. Kidney Int 60:1077-1082, 2001 Kopple JD: Reverse epidemiology of cardiovascular risk 28. van Guldener C, Janssen MJFM, Lambert J, ter Wee factors in maintenance dialysis patients. Kidney Int 63:793- PM, Donker AJM, Stehouwer CDA: Folic acid treatment of 808, 2003 hyperhomocysteinemia in peritoneal dialysis patients: No 36. Liu Y, Coresh J, Eustace JA, et al: Association be- tween cholesterol level and mortality in dialysis patients: change in endothelial function after long-term therapy. Perit Role of inflammation and malnutrition. JAMA 291:451-459, Dial Int 18:282-289, 1998 2004 29. Bennett-Richards K, Kattenhorn M, Donald A, et al: 37. Verma S, Kuliszewski MA, Li SH, et al: C-Reactive Does oral folic acid lower total homocysteine levels and protein attenuates endothelial progenitor cell survival, differ- improve endothelial function in children with chronic renal entiation, and function: Further evidence of a mechanistic failure? Circulation 105:1810-1815, 2002 link between C-reactive protein and cardiovascular disease. 30. Thambyrajah J, Landray MJ, McGlynn FJ, Jones HJ, Circulation 109:2058-2067, 2004 Wheeler DC, Townend JN: Does folic acid decrease plasma 38. Zimmermann J, Herrlinger S, Pruy A, Metzger T, homocysteine and improve endothelial function in patients Wanner C: Inflammation enhances cardiovascular risk and with predialysis renal failure? Circulation 102:871-875, 2000 mortality in hemodialysis patients. Kidney Int 55:648-658, 31. Bennett-Richards KJ, Kattenhorn M, Donald AE, et 1999 al: Oral L-arginine does not improve endothelial dysfunction 39. Bahlmann FH, de Groot K, Mueller O, Hertel B, in children with chronic renal failure. Kidney Int 62:1372- Haller H, Fliser D: Stimulation of endothelial progenitor 1378, 2002 cells: A new putative therapeutic effect of angiotensin II 32. Cross JM, Donald AE, Nuttall SL, Deanfield JE, receptor antagonists. Hypertension 45:526-529, 2005 Woolfson RG, Macallister RJ: Vitamin C improves resis- 40. Longenecker JC, Coresh J, Powe NR, et al: Tradi- tance but not conduit artery endothelial function in pa- tional cardiovascular disease risk factors in dialysis patients tients with chronic renal failure. Kidney Int 63:1433- compared with the general population: The CHOICE Study. 1442, 2003 J Am Soc Nephrol 13:1918-1927, 2002 Transplantation Cardiovascular Risk and Renal Transplantation: Post Hoc Analyses of the Assessment of Lescol in Renal Transplantation (ALERT) Study

Alan G. Jardine, MD, FRCP, Bengt Fellström, MD, PhD, John O. Logan, MSc, Edward Cole, MD, Gudrun Nyberg, MD, PhD, Carola Grönhagen-Riska, MD, PhD, Søren Madsen, MD, Hans-Hellmut Neumayer, MD, PhD, Bart Maes, MD, PhD, Patrice Ambühl, MD, PhD, Anders G. Olsson, MD, PhD, Terje Pedersen, MD, and Hallvard Holdaas, MD, PhD

● Background: Renal transplantation is associated with an increased risk for premature cardiovascular disease. We analyzed the data in the placebo arm of Assessment of Lescol in Renal Transplantation (ALERT) to improve our understanding of the relationship between cardiovascular risk factors and outcomes in this unique population. Methods: We performed Cox survival analysis for myocardial infarction, cardiac death, and noncardiac death in 1,052 patients recruited to the placebo arm of ALERT. These subjects were aged 30 to 75 years, had stable graft function at least 6 months after transplantation, had a serum total cholesterol level between 155 and 348 mg/dL (4 and 9 mmol/L), and were receiving cyclosporine-based immunosuppression. Results: The results confirm previous studies. In multivariate analysis, preexisting coronary heart disease (hazard ratio [HR], 3.69; P < 0.001), total were (0.0023 ؍ and prior acute rejection (HR, 2.36; P ,(0.0045 ؍ cholesterol level (HR, 1.55 per 50 mg/dL; P ؍ independent risk factors. Conversely, independent risk factors for cardiac death were age (HR, 1.58 per decade; P and serum creatinine level ,(0.0004 ؍ ST-T changes on the ECG (HR, 3.17; P ,(0.0002 ؍ diabetes (HR, 3.35; P ,(0.0033 (HR, 2.65 per milligram per deciliter; P < 0.0001). Conclusion: This analysis confirms that renal transplant recipients share risk factors for myocardial infarction and cardiac death with the general population. However, the pattern of risk factors and their relationship with outcomes is atypical, highlighting the unique nature of cardiovascular risk in transplant recipients. Am J Kidney Dis 46:529-536. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Cardiovascular risk; renal disease; transplantation.

ATIENTS WITH end-stage renal failure, Assessment of Lescol in Renal Transplantation P including those treated by means of dialy- (ALERT) is the largest published interventional sis and transplantation, have an increased inci- trial in renal transplantation and the first large-scale dence of premature cardiac disease and a high clinical trial to address CV complications in this prevalence of conventional cardiovascular (CV) population.11,12 The study examined the effects of risk factors.1-3 However, the relationship be- fluvastatin on cardiac outcomes in 2,102 renal tween these CV risk factors and outcome is incompletely understood and complicated by the fact that renal transplant recipients are at in- From the University of Glasgow, UK; University Hospi- creased risk for premature mortality from other tal, Uppsala; Sahlgrenska University Hospital, Göteborg; causes, specifically, malignant disease and infec- University Hospital, Linköping, Sweden; Novartis, Basel University Hospital, Zürich, Switzerland; Toronto General tion. Furthermore, because graft failure is associ- Hospital, Toronto, Canada; University Hospital, Helsinki, ated with an increase in cardiac (and all-cause) Finland; Skejby Hospital, Aarhus, Denmark; Univer- mortality,4 the determinants of graft dysfunction sitätsklinikum Charité, Berlin, Germany; University Hospi- and failure impact on cardiac risk. tal, Leuven, Belgium; Preventive Medicine Clinic, Ullevaal University Hospital; and Rikshospitalet University Hospital, Our understanding of the relationships between Oslo, Norway. CV risk factors and outcome is reliant on registry Received November 3, 2004; accepted in revised form reports and the studies by Kasiske et al3,5,6 and May 18, 2005. others.7-10 From these studies, it is clear that conven- Originally published online as doi:10.1053/j.ajkd.2005.05.014 on July 15, 2005. tional CV risk factors, such as hypertension and Address reprint requests to Alan Jardine, MD, FRCP, hyperlipidemia, are common in transplant recipi- Reader in Nephrology, Division of Cardiovascular and Medi- ents. Although early reports5 suggested that these cal Sciences, University of Glasgow, Gardiner Institute, risk factors were not associated with cardiac events, Western Infirmary, Glasgow G11 6NT, UK. E-mail: more recent studies suggested that the greatly in- [email protected] © 2005 by the National Kidney Foundation, Inc. creased CV risk may be explained by the presence 0272-6386/05/4603-0018$30.00/0 of multiple risk factors.3,7-10 doi:10.1053/j.ajkd.2005.05.014

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 529-536 529 530 JARDINE ET AL transplant recipients followed up for 5 to 6 years. In Table 1. Study Population Characteristics view of the size of this study, the placebo arm (with Demographic and Clinical high-quality follow-up and outcome data) provides Characteristics an opportunity to study the relationship between cardiac risk factors and outcomes.1,3,5 Age (y) 50.0 Ϯ 11.0 Sex (male) 686 (65.2) Diabetes 199 (18.9) METHODS BMI (kg/m2) 25.8 Ϯ 4.6 The study design, main outcome findings, and subgroup Donor age (y) 40.9 Ϯ 15.4 analyses11,12 have been published. Briefly, this was an inves- Smoker (current/previous/ 185 (17.6)/358 (34.0)/ tigator-led, randomized, double-blind, parallel-group study never) 509 (48.4) that compared fluvastatin (Lescol; Novartis Pharma AG, History of CHD 101 (9.6) Basel, Switzerland; 40 to 80 mg/d) with placebo in 2,102 Hypertension 777 (73.9) patients followed up for a minimum of 5 and a maximum of Total cholesterol (mg/dL) 249.6 Ϯ 42.8 6 years. Recorded end points included death (specifically, LDL cholesterol (mg/dL) 159.9 Ϯ 38.9 cardiac death) and nonfatal myocardial infarction (MI). All HDL cholesterol (mg/dL) 51.8 Ϯ 17.2 events were adjudicated by a critical end point committee Triglycerides (mg/dL) 196.5 Ϯ 133.9 Ϯ blinded to treatment allocation.11,12 The diagnosis of MI was SBP (mm Hg) 144 19 Ϯ based on the presence of 2 of the following: clinical history, DBP (mm Hg) 86 10 Ϯ electrocardiogram (ECG) changes, and cardiac enzyme level Pulse pressure (mm Hg) 58 10 increases (based on local ranges). Cardiac deaths included ECG LVH 156 (14.8) sudden death, death caused by MI, and death caused by heart ECG ST-T changes 199 (18.9) Ϯ failure. Noncardiac deaths included those caused by cancer Proteinuria (g/d) 0.38 0.72 Ϯ and infection. Causes of death are listed in the main study Time on dialysis therapy (y) 2.3 3.6 publication.11 Time on renal replacement Ϯ Patients recruited to ALERT were adult renal transplant therapy (y) 7.4 4.9 recipients from northern Europe and Canada who had re- Transplantation (cadaver/ ceived a transplant at least 6 months previously and whose living donor) 823 (78.2)/229 (21.8) Ϯ graft function was stable. All patients received cyclosporine- Serum creatinine (mg/dL) 1.7 0.6 based immunosuppression, and no patient was administered No. of transplants (1/2/3) 900 (85.6)/126 (12.0)/ statins before recruitment. Entry criteria included fasting 26 (2.4) total cholesterol level in the range 155 to 348 mg/dL (4 to 9 Rejection (yes) 448 (42.6) mmol/L), with a lower limit of 155 to 271 mg/dL (4 to 7 Ϯ mmol/L) for patients who had previously experienced an NOTE. Data expressed as mean SD or number (per- MI. Patients recruited to the study were of low risk, reflected cent of total). To convert cholesterol in mg/dL to mmol/L, in a low event rate.11,12 multiply by 0.02586; triglycerides in mg/dL to mmol/L, The statistical analysis plan of the main study has been multiply by 0.01229. described previously. Baseline data used in analyses were based on single measurements obtained at the initiation visit. Hazard ratios (HRs) with 95% confidence intervals and P Blood pressure was measured by using a semiautomated are presented for a large set of independent risk factors by device with patients in the seated position. The ECG was using a stepwise model-building approach. For closely depen- recorded and interpreted at the local hospital for the pres- dent variables (eg, diastolic [DBP] and systolic blood pres- ence of ST-T changes (ischemia or “strain”) and left ventric- sure [SBP] or total and low-density lipoprotein [LDL] cho- ular hypertrophy (LVH). Plasma lipids were measured after lesterol level), single parameters were inserted to identify an overnight fast by a central laboratory.11 In this post hoc the best fit in the multivariate model. Cigarette smoking was analysis, we assessed end points, including MI, cardiac categorized (0, 1, and 2) for nonsmokers, reformed smokers, death, and noncardiac death, in the placebo arm of the study. and current smokers, respectively. A large number of poten- For each end point, we assessed potential risk factors by tial risk factors were analyzed for each outcome before using Cox survival analysis methods. model building. In the Cox multivariate model, P for inclu- Risk factor analysis was performed in the placebo arm of sion and exclusion were 0.01 and 0.1, respectively. Martin- the study. Risk factors included in the analysis (Table 1) gale residual plots were generated to validate the assumption were age, sex, donor type and age, total time spent on of linearity for each of the variables in the Cox model. All dialysis and all forms of renal replacement therapy, number analyses were performed using the SAS statistics package of transplants, history of acute rejection, serum creatinine (SAS Institute, Cary, NC). level, proteinuria, diabetes, body mass index (BMI), smok- ing history, history of coronary heart disease (CHD) or RESULTS hypertension, plasma lipid subfraction levels, blood pres- sure, and ECG abnormalities. Each potential risk factor Between June 1996 and October 1997, a total (Tables 1 to 4) was related to risk for outcome in a univariate of 2,102 patients were recruited to ALERT, of manner by using a Cox proportional hazards survival model. whom 1,052 were randomly assigned to placebo CARDIOVASCULAR RISK AND TRANSPLANT SURVIVAL 531

Table 2. Risk Factors for Nonfatal MI: Univariate and Multivariate Analyses

Univariate HR (95% confidence Multivariate HR (95% confidence Variable intervals for HR) P intervals for HR) P

Age (decade) 1.27 (1.02-1.59) 0.035 Sex (female v male) 0.76 (0.45-1.29) 0.313 Diabetes 2.41 (1.44-4.02) 0.001 BMI (kg/m2) 1.04 (0.99-1.09) 0.103 Donor age (y) 1.01 (0.99-1.03) 0.211 Smoking 1.17 (0.86-1.60) 0.314 CHD 3.22 (1.81-5.73) Ͻ0.001 3.69 (1.98-6.90) Ͻ0.001 Hypertension 1.23 (0.69-2.18) 0.488 Total cholesterol (50 mg/dL) 1.54 (1.19-2.00) 0.001 1.55 (1.15-2.10) 0.005 LDL cholesterol (50 mg/dL) 1.56 (1.15-2.10) 0.004 HDL cholesterol (50 mg/dL) 0.65 (0.30-1.40) 0.271 Triglycerides (50 mg/dL) 1.06 (1.01-1.12) 0.012 SBP (5 mm Hg) 1.03 (0.96-1.09) 0.419 DBP (5 mm Hg) 0.93 (0.82-1.05) 0.222 Pulse pressure (5 mm Hg) 1.06 (0.99-1.13) 0.100 LVH 0.89 (0.42-1.87) 0.763 ST-T changes 1.27 (0.70-2.30) 0.427 Proteinuria (1 g/d) 1.08 (0.80-1.47) 0.610 Dialysis (y) 1.01 (0.95-1.08) 0.762 Renal replacement therapy (y) 1.02 (0.97-1.07) 0.398 Serum creatinine (1 mg/dL) 1.11 (0.72-1.70) 0.643 No. of transplants 1.38 (0.85-2.24) 0.191 Transplantation (cadaver v living donor) 1.34 (0.72-2.51) 0.362 Rejection (yes v no) 1.97 (1.20-3.21) 0.007 2.36 (1.36-4.11) 0.002

NOTE. To convert cholesterol in mg/dL to mmol/L, multiply by 0.02586; triglycerides in mg/dL to mmol/L, multiply by 0.01229. therapy. All patients received cyclosporine-based years. There was a relatively low prevalence of immunosuppressive therapy, in combination with preexisting or symptomatic CV disease: 7.3% of prednisolone in 80.6%, azathioprine in 64.6%, patients had symptomatic angina pectoris, 7.4% and mycophenolate mofetil in 15.1% of subjects. had peripheral vascular disease, and 5.7% had Major causes of chronic renal failure were glo- cerebrovascular disease. The majority had “hyper- merulonephritis in 34.6%, polycystic kidney dis- tension.” A total of 17.6% admitted to smoking ease in 17.4%, diabetic nephropathy in 13.1%, cigarettes, and 18.9% had diabetes (including and chronic pyelonephritis or interstitial nephri- diabetic nephropathy in 13.1%). Average fasting tis in 12.8%. These patients were followed up for total cholesterol level was 251 mg/dL (6.5 mmol/ a mean of 65.3 months. During the course of the L); LDL cholesterol, 159 mg/dL (4.1 mmol/L); study, 14% of patients in the placebo arm of the high-density lipoprotein (HDL) cholesterol, 54 trial received “drop-in” therapy with lipid- mg/dL (1.4 mmol/L); and triglycerides, 195 lowering agents, principally statins.10,11 Patients mg/dL (2.2 mmol/L). We analyzed 3 major end in this arm of the study experienced 54 deaths points: MI, cardiac death, and noncardiac death. from cardiac causes; cardiac deaths represented These analyses are listed in Tables 2 to 4. 9.4 events/1,000 patient-years. There were 66 definite nonfatal MIs (12.1/1,000 patient-years). Myocardial Infarction In addition, 65 patients died of non-CV causes Determinants of nonfatal MI are listed in (11.4/1,000 patient-years). Table 2. The major univariate risk factors are The distribution of risk factors examined in preexisting CHD (HR, 3.69), diabetes (HR, the survival analysis is listed in Table 1. Patients 2.41), cigarette smoking (current versus previ- were an average of 50 years old and had been on ous or previous versus never; HR, 1.17), and renal replacement therapy for just more than 7 age (HR, 1.27 per decade). There were univar- 532 JARDINE ET AL

Table 3. Risk Factors for Cardiac Death: Univariate and Multivariate Analyses

Univariate HR (95% confidence Multivariate HR (95% Variable intervals for HR) P confidence intervals for HR) P

Age (decade) 1.65 (1.28-2.14) Ͻ0.001 1.58 (1.116-2.14) 0.003 Sex (female v male) 0.74 (0.41-1.33) 0.311 Diabetes 2.82 (1.62-4.91) Ͻ0.001 3.35 (1.77-6.33) Ͻ0.001 BMI (kg/m2) 1.02 (0.96-1.08) 0.543 Donor age (y) 1.00 (0.98-1.02) 0.990 Smoking 1.20 (0.85-1.70) 0.290 CHD 3.60 (1.96-6.63) Ͻ0.001 Hypertension 0.75 (0.42-1.32) 0.315 Total cholesterol (50 mg/dL) 1.20 (0.88-1.63) 0.241 LDL cholesterol (50 mg/dL) 1.37 (0.99-1.92) 0.061 HDL cholesterol (50 mg/dL) 0.39 (0.16-0.95) 0.037 Triglycerides (50 mg/dL) 1.04 (0.98-1.12) 0.204 SBP (5 mm Hg) 1.07 (1.00-1.14) 0.050 DBP (5 mm Hg) 0.92 (0.81-1.06) 0.249 Pulse pressure (5 mm Hg) 1.11 (1.03-1.19) 0.003 LVH 2.08 (1.49-2.89) 0.023 ST-T changes 3.59 (2.07-6.21) Ͻ0.001 3.17 (1.68-5.97) Ͻ0.001 Proteinuria (1 g/d) 1.11 (0.81-1.54) 0.513 Dialysis (y) 1.01 (0.94-1.08) 0.829 Renal replacement therapy (y) 1.03 (0.98-1.08) 0.249 Serum creatinine (1 mg/dL) 2.08 (1.49-2.89) Ͻ0.001 2.65 (1.84-3.80) Ͻ0.001 No. of transplants 1.85 (1.17-2.92) 0.008 Transplantation (cadaver v living donor) 0.79 (0.44-1.45) 0.452 Rejection (yes v no) 1.15 (0.67-1.96) 0.611

NOTE. To convert cholesterol in mg/dL to mmol/L, multiply by 0.02586; triglycerides in mg/dL to mmol/L, multiply by 0.01229. iate associations with total cholesterol level changes (consistent with ischemia or strain) on (HR, 1.54 per 50 mg/dL), LDL cholesterol the ECG (HR, 3.59), and serum creatinine (HR, 1.56 per 50 mg/dL), and triglycerides level (HR, 2.08 per milligram per deciliter). (HR, 1.06 per 50 mg/dL), but not HDL choles- There was a univariate association with HDL terol (HR, 0.65 per 50 mg/dL). There were no cholesterol level (HR, 0.39), but no significant significant associations with blood pressure association with other cholesterol subfraction parameters or ECG abnormalities, but an unex- levels, although LDL cholesterol level was pected association with a history of treated close to significance (HR, 1.37). In contrast to acute rejection after the current transplantation nonfatal MI, there were significant univariate (HR, 2.36). In multivariate analysis of this end associations with blood pressure; SBP (HR, point, there were 3 significant determinants of 1.07 per 5 mm Hg) and pulse pressure (HR, outcome: total cholesterol level (HR, 1.55 per 1.11 per 5 mm Hg). There also was a signifi- 50 mg/dL), prior CHD (HR, 3.69), and previ- cant association with number of transplants ous acute rejection (HR, 2.36). (HR, 1.85 per transplant). In multivariate anal- ysis of this end point, main determinants of Cardiac Death outcome were age (HR, 1.58 per decade), Determinants of cardiac death are listed in diabetes (HR, 3.35), ST-T changes on the ECG Table 3. This end point includes patients who (HR, 3.17), and serum creatinine level (HR, died suddenly of presumed primary arrhyth- 2.65 per milligram per deciliter). mias and those who died quickly after MI. Major univariate risk factors are age (HR, 1.65 Noncardiac Death per decade), diabetes (HR, 2.82), preexisting Identification of risk factors for non-CV CHD (HR, 3.60), LVH (HR, 2.08) or ST-T death (Table 4) is of great importance given CARDIOVASCULAR RISK AND TRANSPLANT SURVIVAL 533

Table 4. Risk Factors for Noncardiac Death: Univariate and Multivariate Analyses

Univariate HR (95% confidence Multivariate HR (95% Variable intervals for HR) P confidence intervals for HR) P

Age (decade) 1.95 (1.53-2.48) Ͻ0.001 2.30 (1.73-3.06) Ͻ0.001 Sex (female v male) 0.98 (0.58-1.63) 0.926 Diabetes 1.53 (0.87-2.70) 0.138 BMI (kg/m2) 0.99 (0.93-1.04) 0.621 Donor age (y) 1.02 (1.00-1.04) 0.027 Smoking 1.39 (1.02-1.90) 0.035 CHD 1.90 (0.97-3.73) 0.062 Hypertension 0.69 (0.41-1.16) 0.159 Total cholesterol (50 mg/dL) 1.04 (0.78-1.39) 0.773 LDL cholesterol (50 mg/dL) 0.98 (0.71-1.36) 0.912 HDL cholesterol (50 mg/dL) 1.15 (0.56-2.35) 0.701 Triglycerides (50 mg/dL) 1.00 (0.92-1.10) 0.965 SBP (5 mm Hg) 1.07 (1.00-1.13) 0.035 DBP (5 mm Hg) 0.98 (0.87-1.11) 0.738 Pulse pressure (5 mm Hg) 1.09 (1.02-1.16) 0.009 LVH 1.90 (1.06-3.38) 0.030 ST-T change 2.23 (1.32-3.76) 0.003 Proteinuria (1 g/d) 1.14 (0.86-1.51) 0.362 Dialysis (y) 1.08 (1.03-1.13) 0.002 1.10 (1.04-1.16) 0.001 Renal replacement therapy (y) 1.06 (1.02-1.11) 0.004 Serum creatinine (1 mg/dL) 1.80 (1.29-2.51) Ͻ0.001 2.05 (1.44-2.93) Ͻ0.001 No. of transplants 1.47 (0.92-2.35) 0.104 Transplantation (cadaver v living donor) 1.14 (0.62-2.10) 0.677 Rejection (yes v no) 1.33 (0.82-2.17) 0.253

NOTE. To convert cholesterol in mg/dL to mmol/L, multiply by 0.02586; triglycerides in mg/dL to mmol/L, multiply by 0.01229. the increased risk for premature noncardiac risk.11,12 Although patients with preexisting death in renal transplant recipients.11 There coronary disease were not excluded (in the were univariate associations with age (HR, absence of evidence in 1996), few were in- 1.95 per decade) and the related donor age cluded, reflecting the adoption of data from with serum creatinine level (HR, 1.80 per other populations.12-14 Thus, the study was a milligram per deciliter), time on dialysis primary prevention study in a relatively homo- therapy (HR, 1.08 per year), total time on all geneous population of patients with stable allo- forms of renal replacement therapy (HR, 1.06), graft function receiving cyclosporine-based im- and cigarette smoking (HR, 1.39). There were munosuppression who previously had not been no associations between lipid subfraction lev- administered statins. Although the core study els and non-CV death, although SBP (HR, 1.07 did not achieve a statistically significant reduc- per 5 mm Hg), pulse pressure (HR, 1.09 per 5 tion in the primary composite end point of mm Hg), LVH (HR, 1.90), and ST-T changes major adverse cardiac events, statin use was on the ECG (HR, 2.23) were significant. In associated with a reduced incidence of cardiac multivariate analysis of this end point, there death and MI,11,12 the magnitude of which was were 3 independent associations with death: 12-14 age (HR, 2.30 per decade), serum creatinine similar to other statin outcome trials. level (HR, 2.05 per milligram per deciliter), Strengths of the study are the independent and time on dialysis therapy (HR, 1.10 per verification of individual CV events and end annum). points. This may offset the limitations of study size compared with registry data, and overall, DISCUSSION the placebo population in ALERT should pro- ALERT recruited a population of patients vide information on the relationship between with stable graft function and low overall CV risk factors and a spectrum of outcomes in 534 JARDINE ET AL stable, low-risk, long-term renal transplant re- points and different populations studied; specifi- cipients.1,3,5,7-10 cally, recruitment of longstanding transplant re- In this post hoc analysis, we chose to assess cipients with stable function and the lack of the impact of various classic and potential CV nonwhite subjects in our study.9 Notably, Abbott risk factors on a number of important outcomes et al9 also observed a time-dependent increase in by using a Cox proportional hazards analysis (as events that was lower in patients with good graft used by others5,6). Although risk for CV mortal- function, and the selection of patients with stable ity is increased in patients with end-stage renal graft function in ALERT11,12 may reduce the failure, including transplant recipients, the pat- ability of the study to detect relationships with tern differs from that in the general population, graft function and in the early posttransplanta- with a predominance of sudden (presumed ar- tion period. rhythmic) cardiac death rather than classic MI.1 Another factor that may relate to the study The explanation for this observation may lie in design is the reported change in event rate with the high prevalence of uremic cardiomyopa- transplant era.8 Although we cannot exclude this, thy1,15,16 and resultant predisposition to sponta- all patients were on cyclosporine therapy and neous or ischemia-induced arrhythmia.1,16 Previ- thus are likely to be of similar vintage. Abbot et ous studies did not distinguish between MI and al9 also studied determinants of hospital admis- cardiac death (Tables 2 to 4), but focused on sions for heart failure in transplant recipients. composite coronary end points, acute coronary Independent risk factors associated with this end syndrome, or heart failure.3,5,8-10 point included pretransplantation dialysis therapy MI is the result of occlusion of atherosclerotic for more than 1 year, renal dysfunction, African- coronary vessel plaque caused by thrombosis. In American race, age, diabetes, BMI, and prior the analysis of nonfatal MI, independent determi- heart failure. We could not find an association nants were preexisting CHD, total cholesterol with duration of dialysis therapy and MI or level, and prior rejection. Univariate associations cardiac death. However, this relationship may be also were seen with age, diabetes, and other lipid linked to the development of uremic cardiomyop- subfraction levels. Conversely, independent risk athy,15 as are ECG abnormalities,10,16 a risk factors associated with cardiac death were age, factor in our analysis. The absence of ECG, diabetes, ST-T changes on the ECG, and serum blood pressure, and lipid measurements in the creatinine level. Univariate associations were study by Abbot et al9 also limits comparison. seen with preexisting CHD, HDL cholesterol Rigatto et al10 reported that ECG evidence of level, SBP and pulse pressure, and number of LVH is associated with worse survival in trans- transplants. plant recipients, but did not investigate the cause Overall, these risk factors are broadly similar of death or the importance of other ECG changes. to those reported by other studies. The HR for Additional studies that examined the role of age is virtually identical to that reported else- CV risk factors include those by Kasiske et al3,5 where3,5,7-9 of approximately 5% per annum for and others.7-10 These showed the impact of ciga- the development of ischemic heart disease. Ab- rette smoking,3,6 with approximately 2-fold in- bott et al9 studied risk factors associated with creased CV risk for current smokers, similar to hospitalization for acute coronary syndrome in our findings and those of Abbott et al.9 Kasiske et the US Renal Data System registry. They identi- al3 also studied the impact of blood pressure and fied renal dysfunction, age, diabetes, preexisting lipid levels on the development of ischemic heart CHD, and smoking as the main risk factors. disease. The relative risk associated with an Risks associated with diabetes (HR, 1.84) and increase in total cholesterol level of approxi- smoking (HR, 1.94) are similar to relative risks mately 100 mg/dL (2.59 mmol/L) was approxi- associated with MI and cardiac death in our mately 2-fold and similar to that estimated from study. Although we did not observe a significant the Framingham equations. In an earlier publica- relationship between serum creatinine level and tion, Kasiske et al5 also showed a relationship MI, but only for cardiac death, Abbot et al9 between HDL cholesterol level and ischemic observed a relationship with acute coronary syn- heart disease. The lack of consistency between drome. This may reflect the overlap between end such risk factors as lipid levels and outcomes in CARDIOVASCULAR RISK AND TRANSPLANT SURVIVAL 535 these and other studies3,5,7-10 may reflect the use tensive therapy and use of statin therapy.11,12,20 of composite cardiac end points. Similarly, blood We did not specify whether diabetes was the pressure is not invariably associated with isch- cause of renal failure or had developed posttrans- emic heart disease in transplant populations.1 We plantation. However, the latter (new-onset diabe- failed to show a relationship with MI, but showed tes after transplantation) is now associated with a relationship with cardiac death. However, the reduced life expectancy,21 and as part of a wider relationship was complex, with a positive HR for metabolic syndrome in transplant recipients, is a SBP and negative relationship with DBP. Pulse potentially remediable risk factor. pressure showed the strongest relationship. The Although we recognize the importance of pre- simplest explanation for our findings is that blood mature CV disease as a cause of death in renal pressure (specifically SBP and pulse pressure, transplant recipients, they are equally at risk for which are linked to vascular stiffness) are the premature death from infection and malignant main determinants of LVH, the development of disease.22 In the analysis of determinants of which is linked to sudden death and death caused noncardiac death, we identified a pattern of risk by heart failure.10,15-18 factors, some common to determinants of car- Of the nonclassic risk factors associated with diac death and MI. These include age (a stronger cardiac events, acute rejection has been linked to risk factor for noncardiac than cardiac death), the development of CHD in other studies.5,7 diabetes, and serum creatinine level. In contrast Although this is not a consistent finding, it may to CV end points, time spent on dialysis therapy reflect the close relationship between rejection, before transplantation is a significant risk factor graft loss, and the major impact of graft loss on for premature death from non-CV causes, and in all-cause mortality.8,19 An additional mechanism multivariate analysis, this was independent of may be the existence of continuing inflammation patient age at the time of transplantation. A in patients who have experienced a rejection history of cigarette smoking6 was an indepen- episode or have ongoing low-grade rejection. dent risk factor for non-CV death compared with Alternatively, the increased risk may be caused nonsmokers, presumably through the develop- by the increased use of steroids in this patient ment of pulmonary disease and malignancies. group,20 reflecting the reluctance of clinicians to There were a number of factors that were signifi- withdraw steroid therapy after rejection epi- cant in the univariate analysis, but that were not sodes. Regardless of the underlying mechanism, significant independent risk factors in the subse- rejection appears to be a preventable risk factor quent multivariate analysis. It is likely that these for MI that merits further investigation. reflect the frequent association of comorbid dis- Our data suggest that individual CV risk fac- eases (eg, peripheral vascular disease, cardiac tors may not have the same impact on specific disease, chronic respiratory diseases, and risk for CV events. Thus, elevated lipid levels are risk cancer) and the relationship between age and factors for nonfatal MI, but not cardiac death, many of these univariate risk factors. Finally, whereas renal dysfunction and ECG abnormali- renal dysfunction (increased serum creatinine ties are risk factors for cardiac death, but not level) again was a very strong independent risk nonfatal MI (Tables 2 and 3). From a clinical factor for non-CV deaths and thus is the major viewpoint, age and preexisting CHD are irreme- potentially remediable risk factor for all-cause diable risk factors. Potentially modifiable risk mortality. The most likely explanation is the factors include hyperlipidemia,11,12 improved strong association8,19,23 between increased se- blood pressure control (specifically SBP and pulse rum creatinine level and graft failure and the pressure), reversal or regression of LVH (and association between graft failure and subsequent associated ischemic ST-T changes), prevention mortality.8 However, there may be independent of acute rejection, preservation of renal function, mechanisms involved because even mild renal and prevention of posttransplantation diabetes impairment in other populations is associated (new-onset diabetes after transplantation21). with increased mortality and CV risk.24 This Modification of these risk factors may require a relationship requires further investigation. combined approach involving changes in immu- Although this study identifies and confirms nosuppression, together with improved antihyper- risk factors in transplant recipients, there are 536 JARDINE ET AL problems in the application of this knowledge 8. Abbott KC, Bucci JR, Cruess D, Taylor AJ, Agodas that may require additional interventional tri- LYC: Graft loss and acute coronary syndromes after renal als. Many of the risk factors associated with transplantation in the United States. J Am Soc Nephrol 13:2560-2569, 2002 non-CV mortality are irremediable; the most 9. Abbott KC, Yuan CM, Taylor AJ, Cruess DF, Agodoa important of these is age. Although time on LYC: Early renal insufficiency after renal transplantation in dialysis therapy before transplantation theoreti- the era of modern immunosuppression. J Am Soc Nephrol cally is variable, in many countries, it is lim- 14:2358-2365, 2003 ited by the availability of cadaver organs. We 10. Rigatto C, Foley R, Jefrey J, Negrin C, Tribula C, did not differentiate between diabetic nephrop- Parfrey P: Electrocardiographic left ventricular hypertrophy in renal transplant recipients: Prognostic value and impact of athy and posttransplantation diabetes, al- blood pressure and anaemia. J Am Soc Nephrol 14:462-468, though one may argue that the impact on 2003 outcomes is similar in long-standing transplant 11. Holdaas H, Fellström B, Jardine AG, et al: Effect of recipients.21 Nevertheless, strategies that limit fluvastatin on cardiac outcomes in renal transplant recipi- the development of diabetes after organ trans- ents: A multicentre, randomised, placebo controlled trial. Lancet 14:2024-2031, 2003 plantation may reduce the impact of diabetes 12. Jardine AG, Holdaas H, Fellström B, et al: Fluva- on outcome. Cigarette smoking is a potentially statin reduces myocardial infarction and cardiac death follow- remediable risk factor for both premature graft ing renal transplantation; Sub-group and post-hoc analyses failure and mortality that has received insuffi- of the ALERT study. Am J Transplant 4:988-995, 2004 cient attention, and it is not clear whether 13. Shepherd J, Cobbe SM, Ford I, et al for the West of smoking cessation in transplant recipients will Scotland Coronary Prevention Study Group: Prevention of coronary heart disease with pravastatin in men with hyper- confer survival benefits, although the im- cholesterolemia. N Engl J Med 333:1301-1307, 1995 proved survival of exsmokers in the present 14. Scandinavian Simvastatin Survival Study Group: Ran- study suggests that it will. Hypertension, spe- domised trial of cholesterol lowering in 4444 patients with cifically elevated SBP, is an obvious target for coronary heart disease: The Scandinavian Simvastatin Sur- an interventional trial. Overall, data from vival Study (4S). Lancet 344:1383-1389, 1994 ALERT complement existing data from regis- 15. Parfrey PS, Foley RN, Harnett JD: Outcome and risk 1,3,5,8,9 factors for left ventricular disorders in chronic uraemia. try reports and single-center studies on Nephrol Dial Transplant 11:1277-1285, 1996 the determinants of premature CV disease in 16. Stewart GA, Gansevoort RT, Mark PB, et al: QT transplant recipients and reinforce the need for dispersal, arrhythmias and uraemic cardiomyopathy. Kidney additional clinical trials in this population. Int 67:217-226, 2005 17. Woo YM, Jardine AG, Clark AF, et al: The influence REFERENCES of early graft function on patient survival following renal transplantation. Kidney Int 55:692-699, 1999 1. Baigent C, Burbury K, Wheeler D: Premature cardio- vascular disease in chronic renal failure. Lancet 356:147- 18. McCulloch PA, Soman S: Cardiovascular calcifica- 152, 2000 tion in patients with chronic renal failure: Are we on target 2. Lindholm A, Albrechtsen D, Frödin L, Tufveson G, with this risk factor. Kidney Int Suppl 66:S18-S24, 2004 Persson NH, Lundgren G: Ischemic heart disease—A major 19. Guerin AP, London GP, Marchais SJ, Metivier F: cause of death and graft loss after renal transplantation in Arterial stiffening and vascular calcifications in end-stage Scandinavia. Transplantation 60:451-457, 1995 renal disease. Nephrol Dial Transplant 15:1014-1021, 2000 3. Kasiske BL, Chakkera HA, Roel J: Explained and 20. Miller LW: Cardiovascular toxicities of immunosup- unexplained ischemic heart disease risk after renal transplan- pressive agents. Am J Transplant 2:807-818, 2002 tation. J Am Soc Nephrol 11:1735-1743, 2000 21. Davidson J, Wilkinson A, Dantal J, et al: New-onset 4. Ojo AO, Hanson JA, Wolfe RA, Leichtman AB, Ag- diabetes after transplantation: 2003 International consensus odoa LY, Port FK: Long-term survival in renal transplant guidelines. Transplantation 75:S1-S24, 2003 (suppl 10) recipients with graft function. Kidney Int 57:307-313, 2000 22. Chapman JR, Webster AC: Cancer after renal trans- 5. Kasiske BL, Guijarro C, Massy ZA, Wiederkehr MR, plantation: The next challenge. Am J Transplant 4:841-842, Ma JZ: Cardiovascular disease after renal transplantation. 2004 J Am Soc Nephrol 7:158-165, 1996 23. Fellstrom B, Holdaas H, Jardine AG, et al: Risk 6. Kasiske BL, Klinger D: Cigarette smoking in renal factors for reaching renal endpoints in the Assessment of transplant recipients. J Am Soc Nephrol 11:753-759, 2000 Lescol in Renal Transplantation (ALERT) trial. Transplanta- 7. Rigatto C, Parfrey P, Foley R, Negrijn C, Tribula C, tion 79:205-212, 2005 Jeffrey J: Congestive heart failure in renal transplant recipi- 24. Mann JF, Gerstein HC, Dulau-Florea I, Lonn E: ents: Risk factors, outcomes, and relationship with ischemic Cardiovascular risk in patients with mild renal insufficiency. heart disease. J Am Soc Nephrol 13:1084-1090, 2002 Kidney Int Suppl 84:S192-S196, 2003 The Role of Pretransplantation Renal Replacement Therapy Modality in Kidney Allograft and Recipient Survival

Alexander S. Goldfarb-Rumyantzev, MD, PhD, John F. Hurdle, MD, PhD, John D. Scandling, MD, Bradley C. Baird, MS, MStat, and Alfred K. Cheung, MD

● Background: The effect of pretransplantation renal replacement therapy (RRT) modality on allograft and recipient survival outcome is not well understood. Methods: We studied allograft and recipient survival by using US Renal Data System records from January 1, 1990, to December 31, 1999, with a follow-up period through December 31, males; 70% white; 23% black). Pretransplantation and predominant RRT modality during the %60 ;92,844 ؍ n) 2000 end-stage renal disease (ESRD) period and number and specific combinations of RRT modalities were evaluated. Results: Compared with hemodialysis (HD), a Cox model showed that peritoneal dialysis (PD) immediately before transplantation predicts a 3% lower risk for graft failure (P < 0.05) and 6% lower risk for recipient death (P < 0.001). When predominant RRT modality was analyzed (modality used for > 50% of the ESRD time), PD (hazard ratio [HR], 0.97; P < 0.05) had a protective effect for graft survival compared with HD. Better recipient survival also was associated with PD (HR, 0.96; P < 0.05). Increased number of RRT modalities during the ESRD course was associated with increased risk for graft failure (HR, 1.04 per additional modality used; P < 0.005) and recipient death HR, 1.11 per additional modality used; P < 0.001). Any combination or any single modality (except for PD ؉ HD for) graft survival and PD ؉ HD and PD ؉ HD ؉ transplantation for recipient survival) had protective effects on graft and recipient survival compared with HD. Conclusion: Our results suggest that compared with PD, HD as an RRT modality immediately before transplantation or as a predominant RRT modality during the ESRD course, used alone or in combination with other RRT modalities, is associated with increased risks for graft failure and recipient death. Increased number of RRT modalities used during the ESRD course is associated with worsening of graft and recipient survival. Am J Kidney Dis 46:537-549. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Kidney transplant; outcome; hemodialysis (HD); peritoneal dialysis (PD); graft failure; prediction.

LTHOUGH KIDNEY transplantation im- that long-term graft survival was affected by the A proves survival in transplant recipients modality of dialysis treatment.19,20 What is estab- compared with patients with end-stage renal dis- lished is that increased time on dialysis therapy is ease (ESRD) remaining on the waiting list,1 loss associated with decreased survival of transplant of the allograft terminates the patient survival recipients.20 benefit.2 Recent improvement in immunosuppres- sion has reduced the incidence of acute rejection, but the effect on chronic allograft nephropathy and late graft loss has not been well shown in the From the University of Utah School of Medicine; The literature.3 Geriatric Research, Education, and Clinical Center, Veter- Predictive factors of graft survival have been ans Affairs Salt Lake City Healthcare System, Salt Lake City, studied extensively in adults4,5 and children6,7 UT; and Division of Nephrology, Kidney and Pancreas Transplant Program, Stanford University Medical Center, based on data from the United Network for Stanford, CA. Organ Sharing and North American Pediatric Received February 15, 2005; accepted in revised form Renal Transplant Cooperative Study. May 10, 2005. Donor and recipient age,8 preexisting donor Originally published online as doi:10.1053/j.ajkd.2005.05.013 hypertension and diabetes,9 non–heart-beating on July 12, 2005. 10 11 Supported in part by the Dialysis Research Foundation. donor, prolonged cold storage time, retrans- The data reported here were originally supplied by the US plantation,12 multiple blood transfusions,13 and Renal Data System. The interpretation and reporting of body mass index of donor and recipient,14 along these data are the responsibility of the authors and in no way with other factors, have important roles in graft should be seen as official policy or interpretation of the US government. outcome. However, the ESRD course itself (eg, Address reprint requests to Alexander S. Goldfarb- the modality of renal replacement therapy [RRT], Rumyantzev, MD, PhD, Division of Nephrology and Hyper- alone or in combination) as a predictor of graft tension, University of Utah Health Sciences Center, 85 and patient outcomes has not been well studied. North Medical Dr, East Rm 201, Salt Lake City, UT 84112. E-mail: [email protected] Several studies suggested that pretransplantation © 2005 by the National Kidney Foundation, Inc. dialysis modality had an impact on patient out- 0272-6386/05/4603-0019$30.00/0 come.15-18 However, some reports did not show doi:10.1053/j.ajkd.2005.05.013

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 537-549 537 538 GOLDFARB-RUMYANTZEV ET AL

The goal of the present project is to perform a “60-day rule,” the convention adopted by the USRDS retrospective analysis of US Renal Data System stating that a dialysis modality must continue for at least (USRDS) records to evaluate the role of renal 60 days to be considered stable and therefore constitute a replacement modalities, number of modalities change in modality. The final decision about which covariates to include into used, and their combinations in allograft and final models was based primarily on known associations recipient survival. between variables that could cause confounding of the primary variables of interest; therefore, even variables with METHODS no significant association with the outcome that were not Data Set selected by the stepwise regression, but were deemed to be clinically significant or represent potential confounding, Using the USRDS database, we collected data for all were included in the final model. Results of the stepwise kidney allograft recipients (both pediatric and adults) who analysis were used only as a supportive tool. Covariates underwent kidney or kidney-pancreas transplantation from included a recipient comorbidity score that used cardiovascu- January 1, 1990, through December 31, 1999. Follow-up lar disease, symptomatic peripheral vascular disease, diabe- data were collected through December 31, 2000. tes mellitus, and hypertension (similar to the Davis comor- For recipients of multiple transplants, the most recent one bidity index,21 but limited to the comorbid conditions listed was considered the target transplantation (transplant of inter- that were collected at the time of transplantation); recipient est). Patient records with missing information regarding variables (age, sex, race, height, weight, history of hyperten- graft or patient survival were excluded from the study. A sion, diabetes, history of prior transplant, total duration of total of 92,844 patients with a kidney transplant were identi- ESRD, total number of transplants, panel reactive antibody fied. Records of patients with prior kidney transplants (n ϭ 11,714) also were identified and analyzed separately. levels [mean and peak], education level, primary source of renal care payment, and citizenship); donor variables (heart- Outcome beating donor or not, age, sex, race, height, weight, number of matched HLA antigens, and citizenship); and transplanta- There are 2 outcomes in this study. The first outcome is tion procedure variables (day of the week for the procedure, time between the most recent kidney transplantation and season and year of the transplantation, and cold storage failure of the graft. The second outcome is time between the time). most recent kidney transplantation and patient death. Both Other variables, delayed graft function, episodes of acute outcomes were modeled by using continuous survival time rejection, and type of immunosuppressive medications were variables. The graft failure definition did not include patient death not included in the models. Delayed graft function and acute with a functioning graft, the latter determined in the USRDS rejection may represent intermediate outcome, rather than as a single binary variable. In case the value of this variable the confounding factor, and therefore we speculated that was missing and the patient’s death date was equal to the adjusting for it might yield false-negative results (type 2 graft failure date, we assumed the patient died with a error: failure to reject null hypothesis). functioning graft unless the cause of death was coded as one Patients with a prior history of kidney transplantation of the following: 3200 (graft failure: primary failure), 3201 were analyzed separately and the following variables were (graft failure: rejection), 3202 (graft failure: technical), 3299 added to the analysis: donor type for the transplantation (graft failure: other), or 3903 (miscellaneous: renal failure). immediately before the current transplantation, age at first Allograft outcome was censored at the earliest of the transplantation, age at first graft failure, age at transplanta- following events: loss to follow-up, patient death, or study tion immediately before the current transplantation and at completion date (December 31, 2000) and analyzed as days graft failure immediately before the current transplantation, to graft failure or censor. Patient follow-up was censored at and time between last transplant failure and current transplan- the earliest of loss to follow-up or study completion date and tation. To reduce lead time bias, the models also were analyzed as day to recipient death or censor. adjusted for total duration of ESRD. Independent Variables Statistical Analysis The primary variables of interest were those pertinent to RRT from the USRDS database: RRT modality imme- Categorical variables in the subgroups were compared diately before the current transplantation, predominant by using cross-tabulation. Continuous variables were sum- RRT modality during the ESRD course (defined as modal- marized by using means and SDs. Kaplan-Meier graphs ity used for Ͼ50% of the ESRD period; if none of the and Cox regression models were used for survival analy- modalities were used for Ͼ50%, the predominant modal- sis. To avoid collinearity between the primary variables of ity was labeled “none”), number of different RRTs used, interest, we analyzed them in separate Cox models. SAS combination of RRT modalities used (eg, peritoneal dialy- (SAS Institute, Cary, NC) was used for survival analysis sis [PD] and hemodialysis [HD] and transplantation), and (Kaplan-Meier and Cox proportional hazards models), time course during the pretransplantation period that the whereas S-Plus (Insightful, Seattle, WA) was used for patient was treated with a specific RRT modality. We descriptive statistics and tree-based modeling for data defined the use of a specific dialysis modality by using the imputation. RRT MODALITY AND TRANSPLANT OUTCOME 539

RESULTS of patients (Table 3). The protective effect of PD therapy is not significant in this patient subgroup. Baseline Characteristics In the analysis of recipient survival in the Cox The data set consisted of 92,844 records of model using HD as a reference, both prior trans- patients receiving kidney or kidney-pancreas plantation (HR, 0.80; P Ͻ 0.005) and PD (HR, transplants starting January 1, 1990, and through 0.94; P Ͻ 0.001) had a protective effect on December 31, 1999. Study population character- recipient survival compared with HD. This asso- istics are listed in Table 1. Average age was 43.3 ciation was confirmed again in the subgroup of years, and 60% were male. The graft failed patients who had prior transplants, although the during the 11 years of the study period in 34.9% difference between PD and reference (HD) pa- of patients. We compared some baseline charac- tients was not statistically significant (Table 3). teristics in patients on HD and PD therapy as a predominant pretransplantation modality. Aver- Predominant RRT Modality age ages were 44.8 Ϯ 13.7 years in HD patients Predominant RRT modality, defined as RRT and 41.1 Ϯ 15.1 years in PD patients (P Ͻ modality used for more than 50% of the entire 0.001), comorbidity scores were 0.86 Ϯ 0.81 in ESRD period, was analyzed in a Cox model in HD patients and 0.82 Ϯ 0.78 years in PD patients relation to graft survival; both PD (HR, 0.97; (P Ͻ 0.001), pretransplantation durations of P Ͻ 0.05) and transplantation (HR, 0.86; P Ͻ ESRD were 2.63 Ϯ 2.86 years in HD patients 0.001) had a protective effect for graft survival and 2.11 Ϯ 1.96 years in PD patients (P Ͻ compared with HD. Absence of the predominant 0.001), total numbers of transplants before the modality (each modality was used for Ͻ 50% of current one were 1.08 Ϯ 0.31 in HD patients and the duration of ESRD or no RRT was used) also 1.05 Ϯ 0.25 in PD patients (P Ͻ 0.001), and was associated with lower risk for graft failure Ͻ peak and mean panel reactive antibody levels (HR, 0.90; P 0.001; Table 2). These results were 12.01 Ϯ 21.89 and 5.15 Ϯ 14.83 for HD were illustrated by Kaplan-Meier plots (Fig 1A). patients and 9.93 Ϯ 19.47 and 4.15 Ϯ 12.98 for In the Cox model, better recipient survival also Ͻ PD patients, respectively (P Ͻ 0.001 for both). was associated with both PD (HR, 0.96; P Ͻ Numbers of matched HLA antibodies between 0.05) and transplantation (HR, 0.82; P 0.001) recipient and donor were 1.80 Ϯ 1.52 in HD as predominant pretransplantation RRT modali- patients and 1.84 Ϯ 1.54 in PD patients (P Ͻ ties. Patients who had no predominant modality 0.001). Finally, donor ages were 34.72 Ϯ 15.64 during the ESRD course also had better survival years in HD patients and 33.81 Ϯ 15.41 years in compared with HD, although the difference was not statistically significant (HR, 0.921; P ϭ PD patients (P Ͻ 0.001). 0.063). The worst patient outcome associated with HD is illustrated by Kaplan-Meier plots RRT Modality Immediately (Fig 1B). The same trends for graft and recipient Before Transplantation survival were found in the subgroup of patients The Cox model using HD as a reference showed with prior transplants (Table 3). the following results. Having a transplant immedi- ately before the transplantation of interest without Number of Different Modalities Used dialysis therapy in between was associated with We calculated the number of different RRT increased risk for graft failure (hazard ratio [HR], modalities that a patient was exposed to during 1.65; P Ͻ 0.001). PD as a modality immediately the ESRD course by using the 60-day rule, in before transplantation predicts better graft outcome which the change in dialysis technique is con- compared with HD (HR, 0.97; P Ͻ 0.05; Table 2). sidered stable if the patient remained on a new A similar association was found in the subgroup of modality for 60 or more days (the 60-day rule patients with a previous history of kidney transplan- does not apply to transplantation). We ana- tation: having the transplantation as an RRT modal- lyzed the Cox model and showed that number ity before the last transplantation without going on of RRT modalities is a significant predictor of dialysis therapy in between poses a greater risk for graft failure (HR, 1.04 per additional modality; graft failure (HR, 1.99; P Ͻ 0.001) in this subgroup P Ͻ 0.001) and recipient death (HR, 1.11 per 540 GOLDFARB-RUMYANTZEV ET AL

Table 1. Baseline Characteristics of Kidney Transplant Recipients at the Time of the Most Recent Transplantation

Recipient characteristics Age (y) 43.3 Ϯ 14.2 Sex (% men) 60.3 Race (white/African American/Asian/Native American; %) 70.2/23.0/3.4/0.9 Weight (kg) 72.6 Ϯ 17.2 Height (cm) 169.0 Ϯ 13.7 Primary cause of ESRD (%) Diabetes mellitus 25.2 Hypertension 17.2 Glomerulonephritis 25.8 Cystic disease 7.6 Other 24.2 Comorbidity score* 0.8 Ϯ 0.8 History of diabetes (%) 27.2 History of hypertension (%) 52.5 Total duration of ESRD (y) 3.1Ϯ3.6 Percent of ESRD time on PD† 22.8 Ϯ 38.0 Percent of ESRD time on HD† 67.3 Ϯ 41.5 Percent of total ESRD duration with transplant† 6.1 Ϯ 20.1 RRT modality immediately before transplantation (%) HD 71.3 PD 21.8 Transplantation (dialysis-free retransplantation) 1.1 Unknown 5.8 Predominant RRT modality‡ (%) HD 67.3 PD 22.6 Transplantation 6.4 None 3.6 Total no. of transplants (including the current one) 1.2 Ϯ 0.4 Time on the transplant list (y) 1.3 Ϯ 1.1 Peak reactive antibody level (%) 12.1 Ϯ 21.5 Mean reactive antibody level (%) 5.3 Ϯ 14.7 No. of matched HLA antibodies 1.8 Ϯ 1.5 Cold ischemia time (h) 15.5 Ϯ 8.7 Transplant day of the week§ 4.0 Ϯ 1.8 History of previous kidney transplant(s) (%) 12.6 Donor characteristics Age (y) 34.4 Ϯ 15.5 Sex (% men) 56.2 Race (white/African American/Asian/Native American; %) 82.5/11.5/1.3/0.4 Weight (kg) 72.8 Ϯ 19.0 Height (cm) 164.3 Ϯ 21.9 Terminal serum creatinine level (mg/dL) 0.9 Ϯ 0.3 Terminal blood urea nitrogen level (mg/dL) 12.1 Ϯ 6.1 Living donors (%) 24.8 NOTE. N ϭ 92,844. Continuous variables presented as mean Ϯ SD. To convert serum creatinine in mg/dL to ␮mol/L, multiply by 88.4; urea nitrogen in mg/dL to mmol/L, multiply by 0.357. *The comorbidity score used in our study was calculated based on the following coexisting conditions, each contributing 1 point: cardiovascular disease (defined in the USRDS as symptomatic cardiovascular disease or angina/coronary artery disease), symptomatic peripheral vascular disease, diabetes mellitus, and hypertension. †Information obtained from USRDS RXHIST file; because of missing/unknown data and the 60-day rule convention adopted by the USRDS (see text), the total is less than 100%. ‡Defined as modality used for greater than 50% of the duration of ESRD. §Transplant day of the week expressed in numbers starting with Sunday (1 ϭ Sunday, 2 ϭ Monday, etc). RRT MODALITY AND TRANSPLANT OUTCOME 541

Table 2. Results of Cox Proportional Hazard Model Analyzing Allograft and Patient Survival in the Entire Patient Population

Graft Survival Recipient Survival

95% 95% Confidence Confidence HR Interval P HR Interval P

RRT modality immediately before current transplantation* PD 0.97 0.94-1 Ͻ0.05 0.94 0.91-0.97 Ͻ0.001 Transplantation 1.65 1.51-1.8 Ͻ0.001 0.8 0.68-0.93 Ͻ0.005 Unknown 0.92 0.85-1 Ͻ0.05 0.87 0.77-0.97 Ͻ0.05 Lost to follow-up 1.07 1-1.15 0.069 0.9 0.81-0.99 Ͻ0.05 Predominant RRT modality* PD 0.97 0.94-1 Ͻ0.05 0.96 0.92-0.99 Ͻ0.05 Transplantation 0.86 0.81-0.9 Ͻ0.001 0.82 0.76-0.89 Ͻ0.001 None 0.9 0.84-0.95 Ͻ0.001 0.92 0.85-1 0.063 Time spent on HD (y) 1.02 1.01-1.02 Ͻ0.001 1.05 1.04-1.05 Ͻ0.001 Ͼ0-1 1.05 1.01-1.09 Ͻ0.05 1.18 1.12-1.24 Ͻ0.001 Ͼ1-3 1.18 1.13-1.23 Ͻ0.001 1.42 1.34-1.5 Ͻ0.001 Ͼ3-10 1.18 1.12-1.23 Ͻ0.001 1.59 1.5-1.7 Ͻ0.001 Ͼ10-33 1.27 1.16-1.39 Ͻ0.001 1.77 1.57-2 Ͻ0.001 Time spent on PD (y) 1.02 1.01-1.03 Ͻ0.005 1.04 1.03-1.06 Ͻ0.001 Ͼ0-1 1.04 1.01-1.08 Ͻ0.05 1.12 1.07-1.17 Ͻ0.001 Ͼ1-3 1.08 1.04-1.12 Ͻ0.001 1.21 1.16-1.27 Ͻ0.001 Ͼ3-10 1.13 1.06-1.2 Ͻ0.001 1.33 1.23-1.44 Ͻ0.001 Ͼ10-33 1.28 0.99-1.64 0.057 1.43 1-2.04 0.053 Time spent with prior transplant (y) 0.98 0.97-0.99 Ͻ0.001 1 0.99-1.01 0.525 Ͼ0-1 0.84 0.78-0.9 Ͻ0.001 1.06 0.95-1.19 0.319 Ͼ1-3 0.82 0.75-0.9 Ͻ0.001 1.08 0.94-1.23 0.291 Ͼ3-10 0.72 0.67-0.78 Ͻ0.001 1.08 0.95-1.22 0.24 Ͼ10-33 0.67 0.6-0.75 Ͻ0.001 1.12 0.94-1.33 0.217 No. of different RRT modalities† 1.04 1.02-1.07 Ͻ0.005 1.11 1.08-1.15 Ͻ0.001 Combinations of RRT modalities* PD only 0.93 0.9-0.96 Ͻ0.001 0.9 0.86-0.94 Ͻ0.001 PD ϩ transplantation 0.87 0.78-0.97 Ͻ0.05 0.98 0.83-1.17 0.86 PD ϩ HD 1.09 1.05-1.12 Ͻ0.001 1.1 1.06-1.15 Ͻ0.001 HD ϩ transplantation 0.74 0.69-0.8 Ͻ0.001 0.96 0.86-1.08 0.508 Transplantation only 0.94 0.85-1.05 0.269 0.89 0.75-1.06 0.196 PD ϩ HD ϩ transplantation 0.73 0.67-0.8 Ͻ0.001 1.11 0.98-1.27 0.106 None 0.75 0.69-0.81 Ͻ0.001 0.81 0.73-0.89 Ͻ0.001

NOTE. The Cox model represents multivariate analysis of graft and recipient survival. To avoid colinearity between the primary variables of interest, they were analyzed in separate Cox models. Only primary variables of interest are listed in the table. All models also were adjusted for the following covariates: recipient variables: age, sex, race, height, weight, history of hypertension, diabetes, comorbidity score, history of prior transplant, total duration of ESRD, total number of transplants, panel reactive antibody levels (mean and peak), education level, primary source of renal care payment, and citizenship; donor variables: heart-beating donor or not, age, sex, race, height, weight, number of matched HLA antigens, and citizenship; and transplant procedure variables: day of the week for the procedure, season and year of the transplantation, and cold storage time. *HD is a reference. †The 60-day rule is applied. additional modality; P Ͻ 0.001; Table 2). In Ͻ 0.005; Table 3). These associations are the separate model in the subgroup of patients shown by Kaplan-Meier plots: increased num- with prior transplants, number of modalities ber of modalities is associated with worsening was not a significant risk for graft failure, of graft survival (Fig 2A), and best graft sur- whereas it was for recipient death (HR, 1.09; P vival is associated with 0 pretransplantation 542 GOLDFARB-RUMYANTZEV ET AL

Table 3. Results of Cox Proportional Hazard Model Analyzing Allograft and Patient Survival in the Subgroup of Patients With a Prior History of Kidney Transplantation

Graft Survival Recipient Survival

95% Confidence 95% Confidence HR Interval P HR Interval P

RRT modality before transplantation* PD 0.94 0.86-1.03 0.179 0.97 0.86-1.09 0.603 Transplantation 1.99 1.8-2.21 Ͻ0.001 0.74 0.62-0.9 Ͻ0.005 Unknown 0.89 0.81-0.98 Ͻ0.05 0.79 0.68-0.91 Ͻ0.005 Lost to follow-up 0.96 0.75-1.23 0.729 1.05 0.74-1.49 0.772 Predominant RRT modality* PD 0.87 0.77-0.97 Ͻ0.05 0.92 0.78-1.08 0.285 Transplantation 0.86 0.8-0.92 Ͻ0.001 0.8 0.73-0.89 Ͻ0.001 None 0.82 0.75-0.9 Ͻ0.001 0.84 0.74-0.96 Ͻ0.05 Time spent on HD (y) 1.02 1.01-1.03 Ͻ0.001 1.04 1.03-1.06 Ͻ0.001 Ͼ0-1 0.88 0.8-0.98 Ͻ0.05 1.18 1-1.39 0.053 Ͼ1-3 0.94 0.85-1.04 0.217 1.36 1.15-1.6 Ͻ0.001 Ͼ3-10 0.96 0.87-1.07 0.479 1.54 1.3-1.82 Ͻ0.001 Ͼ10-33 1.15 0.98-1.35 0.087 1.83 1.45-2.3 Ͻ0.001 Time spent on PD (y) 1.01 0.99-1.03 0.299 1.05 1.02-1.08 Ͻ0.005 Ͼ0-1 1.02 0.94-1.1 0.716 1.13 1.01-1.27 Ͻ0.05 Ͼ1-3 0.99 0.91-1.08 0.823 1.24 1.09-1.4 Ͻ0.001 Ͼ3-10 1.05 0.93-1.18 0.48 1.29 1.09-1.54 Ͻ0.005 Ͼ10-33 1.13 0.75-1.72 0.555 1.02 0.51-2.06 0.957 Time spent with prior transplant (y) 0.99 0.98-0.99 Ͻ0.001 1 0.99-1.01 0.768 Ͼ0-1 0.89 0.77-1.02 0.092 0.8 0.66-0.96 Ͻ0.05 Ͼ1-3 0.9 0.77-1.05 0.176 0.81 0.66-1 Ͻ0.05 Ͼ3-10 0.8 0.69-0.93 Ͻ0.005 0.81 0.67-0.99 Ͻ0.05 Ͼ10-33 0.74 0.63-0.88 Ͻ0.001 0.82 0.64-1.04 0.104 No. of different RRT modalities† 0.99 0.95-1.04 0.688 1.09 1.02-1.16 Ͻ0.05 Combinations of RRT modalities* PD only 0.6 0.38-0.95 Ͻ0.05 0.73 0.42-1.3 0.289 PD ϩ transplantation 0.84 0.7-1.01 0.07 0.73 0.56-0.94 Ͻ0.05 PD ϩ HD 1.12 0.84-1.49 0.446 0.94 0.65-1.36 0.724 HD ϩ transplantation 0.79 0.68-0.93 Ͻ0.005 0.74 0.6-0.91 Ͻ0.005 Transplantation only 0.86 0.71-1.03 0.098 0.62 0.48-0.8 Ͻ0.001 PD ϩ HD ϩ transplantation 0.79 0.67-0.93 Ͻ0.005 0.86 0.69-1.06 0.16 None 0.92 0.4-2.08 0.838 1.1 0.41-2.99 0.85

NOTE. The Cox model represents multivariate analysis of graft and recipient survival. To avoid colinearity between the primary variables of interest, they were analyzed in separate Cox models. Only primary variables of interest are listed in the table. All models also were adjusted for the following covariates: recipient variables: age, sex, race, height, weight, history of hypertension, diabetes, comorbidity score, total duration of ESRD, total number of transplants, panel reactive antibody levels (mean and peak), education level, primary source of renal care payment, and citizenship; donor variables: heart-beating donor or not, age, sex, race, height, weight, number of matched HLA antigens, and citizenship; transplant procedure variables: day of the week for the procedure, season and year of the transplantation, and cold storage time; parameters of prior transplant(s): donor type for the transplant immediately before the current transplant, age at first transplantation, age at first graft failure, age at transplantation immediately before the current transplantation and at graft failure immediately before the current transplantation, and time between last transplant failure and current transplantation. *HD is a reference. †The 60-day rule is applied.

RRT modality. Similarly, 0 modality used be- Combination of Different RRT Modalities fore transplantation was associated with the We considered 8 different combinations of best recipient survival, but increased number RRT modalities during the ESRD course indepen- of modalities greater than 1 does not affect dent of the sequence and number of times a recipient survival (Fig 2B). patient would return to a particular modality: PD RRT MODALITY AND TRANSPLANT OUTCOME 543

Fig 1. Predominant RRT modality and (A) graft and (B) recipient survival. The worst graft and recipient out- come is associated with HD. only, HD only, transplantation only, PD plus HD, pared with HD alone, PD alone was associated PD plus transplantation, HD plus transplanta- with lower risk for graft failure (HR, 0.60; P Ͻ tion, all 3 modalities, and none. We define combi- 0.05); PD plus HD plus transplantation also was nations of RRT modalities by using the 60-day beneficial (HR, 0.79; P Ͻ 0.005), as well as HD rule described. In the Cox model (HD only was plus transplantation (HR, 0.791; P Ͻ 0.005; used as a reference group), any combination or Table 3). When recipient survival was used as an single modality (except for transplantation only, outcome, PD plus transplantation (HR, 0.73; P Ͻ PD plus HD, and none) was better than HD only 0.05), HD plus transplantation (HR, 0.74; P Ͻ (Table 2). In particular, PD only was associated 0.005), and transplantation only (HR, 0.62; P Ͻ with an HR of 0.93 (P Ͻ 0.001). PD plus HD and 0.0005) were associated with lower mortality none were associated with a not statistically risk. significant greater risk. When patient survival was evaluated, modality combinations showing Number of Years on Specific Dialysis Modality the significant difference with the reference group The Cox model discussed in this section was (HD only) were PD only (HR, 0.90; P Ͻ 0.001) not adjusted for total duration of ESRD to avoid and none (HR, 0.81; P Ͻ 0.001) and also PD plus colinearity with the primary variables of interest. HD (HR, 1.10; P Ͻ 0.001). When patients with The duration of both dialysis modalities was prior transplants were analyzed separately com- associated with greater risk for graft failure. 544 GOLDFARB-RUMYANTZEV ET AL

Fig 2. Number of differ- ent RRT modalities and (A) graft and (B) recipient sur- vival. Increased number of modalities is associated with worsening of graft sur- vival, and the best graft and recipient survival are asso- ciated with 0 RRT modali- ties (preemptive kidney transplantation).

Each year of PD therapy is associated with an HD therapy (HR, 1.04; P Ͻ 0.001) was associ- HR of 1.02 (P Ͻ 0.005), and each year of HD ated with greater risk for recipient mortality, therapy is associated with an HR of 1.02 (P Ͻ whereas number of years with a functioning graft 0.001). Conversely, number of years with a func- was not associated with a significant change tioning graft in the past had a protective effect on (Table 2). current graft survival (HR, 0.98; P Ͻ 0.001). In the subgroup of patients with a prior transplant, DISCUSSION the same trend is true, but the association be- Prediction of renal transplant recipient and graft tween duration of PD therapy and graft failure is survival presents an important clinical tool, espe- not statistically significant (Table 2). We per- cially in view of the growing shortage of renal formed the same analysis for recipient survival transplants. Almost half the transplant recipients and showed similar associations. The longer the die with a functioning graft. In the rest, the allograft patient was on HD (HR, 1.05; P Ͻ 0.0001) or PD fails because of chronic renal allograft dysfunction. therapy (HR, 1.04; P Ͻ 0.005), the greater the Although many risk factors are known (as de- risk for dying, whereas number of years with a scribed), the role of pretransplantation RRT modal- prior transplant did not make a significant differ- ity in both graft and patient survival was not clear ence. Recipient survival in the subgroup of pa- from the literature. In general, prior studies compar- tients with a prior transplant also was analyzed. ing different RRT modalities in relation to transplan- Number of years on PD (HR, 1.05; P Ͻ 0.005) or tation outcome were based on small data sets16,22; RRT MODALITY AND TRANSPLANT OUTCOME 545

Table 4. Cross-Tabulation Between RRT Modality Immediately Before the Most Recent Transplantation and Predominant RRT Modality During the ESRD Course

Predominant RRT RRT Modality Before the Most Recent Transplantation Modality During the ESRD Course HD PD Transplantation Unknown Total

HD 58,389 1,880 72 2,179 62,520 PD 3,310 16,942 21 711 20,984 Transplantation 2,960 677 952 1,367 5,956 None 1,564 705 20 1,095 3,384 Total 66,223 20,204 1,065 5,352 92,844

NOTE. Predominant RRT modality is defined as a modality used for greater than 50% of the duration of ESRD. evaluated short-term, rather than long-term, out- with PD compared with HD.27,32-35 A greater come16,18,23,24; did not study combination of RRT rate of graft thrombosis in PD patients might modalities or were performed 10 to 20 years ago, occur because hypercoagulable states are not before significant changes in immunosuppressive detected as readily in PD patients as in HD regimens22-26; and most studies examined graft patients. A greater rate of acute rejection was outcome only, not patient survival. In addition, associated with PD therapy16; however, in other most investigators evaluated the role of RRT modal- reports, acute rejection rates were not different ity immediately before transplantation, rather than between patients on PD versus HD therapy be- predominant modality, as a primary variable of fore transplantation.18,26,29,35 interest. Patients who were on PD therapy for a PD was associated with a greater rate of post- number of years and were switched to HD therapy transplantation infection compared with patients immediately before transplantation would be classi- on HD therapy in some reports,36 but in others, fied as HD in these studies, which introduced a infection rates between PD and HD patients were significant degree of misclassification bias.As listed similar16,37 or lower in patients with pretransplan- in Table 4, there is significant discrepancy between tation PD therapy.38 RRT modality before transplantation and predomi- When long-term outcome (1 and 5 years) was nant RRT modality in the USRDS data set. studied, no difference between dialysis modali- Previous studies of short-term transplantation ties was reported.26 Similarly, PD and HD showed outcome have yielded somewhat conflicting re- similar 1-year outcome in a report by Donnelly et sults. A greater rate of early graft failure was al24 in the mid-1980s. In a small case-control associated with PD therapy, shown by27. In the analysis, 1-year transplantation outcome in pa- report by Bleyer et al,18 the investigators evalu- tients on continuous ambulatory PD therapy was ated delayed graft function after deceased donor not significantly different from that in HD pa- transplantation based on dialysis modality imme- tients with similar clinical characteristics.22 Simi- diately before and found an association between lar long-term graft and patient survival was HD therapy and delayed graft function. Similar achieved independent of dialysis modality be- results were shown by other investiga- fore transplantation in a retrospective analysis of tors.16,17,27–29 Vanholder et al16 also showed the the first cadaveric graft. Graft and patient sur- advantage of PD therapy for short-term outcome: vival cases were identical in the HD and continu- patients on PD therapy had a reduced incidence ous ambulatory PD groups (5-year graft survival: of acute renal failure. PD therapy is recom- continuous ambulatory PD, 67%; HD, 66%; mended to be chosen as an initial modality in 5-year patient survival: continuous ambulatory patients who plan kidney transplantation within PD, 88%; HD, 87%).25 2 to 3 years.30 No difference in long-term outcome between Although there was no association between patients treated with PD compared with HD was dialysis modality and graft thrombosis in 1 re- shown in other studies.23,29,39,40 Snyder et al27 port,31 several other studies reported an in- compared long-term transplant outcome between creased incidence of graft thrombosis associated PD and HD patients and showed that, compared 546 GOLDFARB-RUMYANTZEV ET AL with pretransplantation HD, death-censored long- ceased donor (HR, 0.68; P Ͻ 0.001; HR, 0.67; term graft failure was 15% greater in patients on P Ͻ 0.001 for graft and recipient survival, respec- PD therapy and short-term graft failure was 33% tively), donor age (HR, 1.01; P Ͻ 0.001; HR, greater. Their analysis was based on 22,776 Medi- 1.01; P Ͻ 0.001 per year of life for graft and care beneficiaries with a kidney transplant. Com- recipient survival, respectively), and number of pared with our analysis, the study by Snyder et HLA-matched antigens (HR, 0.94; P Ͻ 0.001; al27 used a smaller number of patients and shorter HR, 0.96; P Ͻ 0.001 per antigen matched for follow-up (3 years), and some baseline character- graft and recipient survival, respectively). istics of the study population were different from Number of RRT modalities used during the ours (pediatric patients and those with prior ESRD course is a significant predictor of graft history of transplantation were excluded). Those failure and recipient death. Almost any single investigators studied pretransplantation dialysis RRT modality or their combination was associ- modality (based on a United Network for Organ ated with better graft and recipient outcome than Sharing form) adjusted for dialysis modality HD only. change as a binary variable, whereas we evalu- The mechanism of the outcome observed in ated the role of both pretransplantation and pre- our analysis is not completely clear. Residual dominant dialysis modality, as well as number of renal function that might be better preserved in modalities used and their combinations. patients on PD therapy may contribute to better In our study, compared with HD, PD therapy preservation of kidney function after transplanta- immediately before transplantation has a protec- tion.41 One can hypothesize that the degree of tive effect on graft and recipient survival. Al- residual renal function is more important for though the effect size associated with PD therapy graft and recipient outcome than either PD or HD is modest, we disprove previous reports claiming modality. Unfortunately, we did not have informa- a greater long-term risk associated with PD tion about residual renal function for the entire therapy. We evaluated the role of predominant study population. Answering this question could RRT modality during the ESRD course. PD, be a subject of another research project. Further- transplantation, and preemptive or very short- more, body mass index and degree of hypervol- course dialysis therapy had a protective effect for emia might be different in PD and HD patients graft and recipient survival. This approach again and therefore confound results. The rate of post- confirmed an advantage for allograft and recipi- transplantation infections associated with HD ent survival of PD over HD as a modality imme- might be greater compared with PD.38 In addi- diately before transplantation and as a predomi- tion, there are some indications that HD mem- nant modality during the ESRD course. Although branes and vascular access might cause sensitiza- statistically significant, the effect size of dialysis tion in transplant candidates.42 It has been shown modality is limited (HRs of 0.97 and 0.94 for that HD patients have elevated levels of natural graft and recipient survival for PD as an RRT killer cells43 and production of cytokines.44 Other modality immediately before transplantation and immunologic differences might exist between HRs of 0.97 and 0.96 for graft and recipient HD and PD patients. It was postulated that PD survival for PD as a predominant RRT modality modifies the population of T-helper (TH) cells, compared with HD, respectively). For compari- with an increase in percentage of TH subtype 2 45 son, the effect size of other predictors of graft (TH2) cells and normal percentage of TH1 cells. and recipient survival evaluated in our analysis TH2 cells produce interleukin 4 and interleukin were as follows: recipient age (HR, 1.01; P Ͻ 10, which inhibit interferon ␥ secretion and cell Ͻ 46 0.001; HR, 1.04; P 0.001 per year of life for immunity, and increased TH2 cell fraction may graft and recipient survival, respectively), recipi- provide additional immunosuppression. ent history of diabetes (HR, 0.96; P ϭ 0.48; HR, Potential selection bias should be considered 1.11; P ϭ 0.107 for graft and recipient survival, when interpreting results of this study. We specu- respectively), recipient comorbidity score (HR, lated that the decision regarding dialysis modal- 1.1; P Ͻ 0.001; HR, 1.26; P Ͻ 0.001 per unit ity was made based on patient age, diabetic increase in score for graft and recipient survival, status, comorbidity, ability to learn the tech- respectively), living donor compared with de- nique, prior history of abdominal surgeries, dis- RRT MODALITY AND TRANSPLANT OUTCOME 547 tance to the dialysis center, and status of the survival. PD is a reasonable choice of RRT and vascular access. Adjusting our multivariate mod- should not be avoided in transplantation candi- els for recipient age, diabetic status, comorbidity dates. index, socioeconomic status (indicated by educa- tional level, primary source of pay for renal care, ACKNOWLEDGMENT and citizenship), and duration of ESRD should The authors thank Zhi Wang and Greg Stoddard for considerably reduce the selection bias. We recog- statistical assistance. nize that some factors not included in the models REFERENCES (eg, exhausted vascular access) might force selec- 1. Ojo AO, Hanson JA, Meier-Kriesche HU, et al: Sur- tion of the dialysis modality and confound re- vival in recipients of marginal cadaveric donor kidneys sults. Although our models were adjusted care- compared with other recipients and wait-listed transplant fully for pertinent covariates, the HD and PD candidates. J Am Soc Nephrol 12:589-597, 2001 populations are different in our study, as well as 2. Ojo AO, Wolfe RA, Agodoa LY, et al: Prognosis after primary renal transplant failure and the beneficial effects of in other reports. For example, it has been shown repeat transplantation. Multivariate analyses from the United that PD patients are more likely to undergo States Renal Data System. Transplantation 66:1651-1659, transplantation than HD patients in the group of 1998 adults27 and, to a lesser extent, group of pediatric 3. Pascual M, Theruvath T, Kawai T, Tolkoff-Rubin N, patients.47 One can speculate regarding the causes Cosimi B: Strategies to improve long-term outcomes after renal transplantation. N Engl J Med 346:580-590, 2002 of this discrepancy, eg, demographic characteris- 4. Gjertson DW: Determinants of long-term survival of tics and potentially the more assertive personal- adult kidney transplants: A 1999 UNOS update, in Clin ity of PD patients versus HD patients might Transplants 1999. Los Angeles, CA, UCLA Immunogenet- make them more aggressive in pursuing trans- ics Center, 2000, pp 341-352 plantation. In our study population, PD was a 5. Goldfarb-Rumyantzev AS, Scandling JD, Pappas L, Smout RJ, Horn S: Prediction of 3-yr cadaveric graft sur- pretransplantation RRT modality in 21.8% and vival based on pre-transplant variables in a large national predominant RRT modality in 22.6% of patients, dataset. Clin Transplant 17:485-497, 2003 whereas in the dialysis population, PD patients 6. Gjertson DW, Cecka JM: Determinants of long-term comprise less than 15%.48 To explain this phe- survival of pediatric kidney grafts reported to the United nomenon, Snyder et al27 proposed that there is a Network for Organ Sharing kidney transplant registry. Pedi- atr Transplant 5:5-15, 2001 perception among physicians that PD patients 16,30 7. Tejani A, Stablein DM, Donaldson L, et al: Steady may be better candidates for transplantation improvement in short-term graft survival of pediatric renal and therefore selection bias, in which potential transplants: The NAPRTCS experience, in Clin Transplants transplantation candidates are more likely to be 1999. Los Angeles, CA, UCLA Immunogenetics Center, placed on PD therapy than HD. Other potential 2000, pp 95-110 8. Alexander JW, Bennett LE, Breen TJ: Effect of donor shortcomings should be considered in interpret- age on outcome of kidney transplantation. Transplantation ing results of this retrospective data analysis. 57:871-876, 1994 Retrospective analysis of a data registry shows 9. Ojo AO, Leichtman AB, Punch JD, et al: Impact of the association (but not necessarily causative pre-existing donor hypertension and diabetes mellitus on relationships) between the primary variables of cadaveric renal transplant outcomes. Am J Kidney Dis 36:153-159, 2000 interest and outcome. The sequence of PD and 10. Sánchez-Fructuoso AI, Prats D, Torrente J, et al: HD for patients receiving both has not been Renal transplantation from non-heart beating donors: A evaluated in this study and might be a subject of promising alternative to enlarge the donor pool. J Am Soc future research. Nephrol 11:350-358, 2000 In conclusion, our results suggest that com- 11. Jordan ML, Shapiro R, Vivas CA, et al: High-risk donors: Expanding donor criteria. Transplant Proc 31:1401- pared with PD, HD as an RRT modality immedi- 1403, 1999 ately before transplantation or as a predominant 12. Mouquet C, Benalia H, Chartier-Kastler E, et al: RRT modality during the ESRD course, used Renal retransplantation in adults. Comparative prognostic alone or in combination with other RRT modali- study. Prog Urol 9:239-243, 1999 ties, is associated with increased risk for graft 13. Chavers BM, Sullivan EK, Tejani A, Harmon WE: Pre-transplant blood transfusion and renal allograft out- failure and recipient death. Increased number of come: A report of the North American Pediatric Renal RRT modalities used during the ESRD course is Transplant Cooperative Study. Pediatric Transplantation 1:22- associated with worsening of graft and recipient 28, 1997 548 GOLDFARB-RUMYANTZEV ET AL

14. Remuzzi G, Grinyo J, Ruggenenti P, et al: Early 30. Thodis E, Passadakis P, Vargemezis V, Oreopoulos experience with dual kidney transplantation in adults using DG: Peritoneal dialysis: Better than, equal to, or worse than expanded donor criteria. J Am Soc Nephrol 10:2591-2598, hemodialysis? Data worth knowing before choosing a dialy- 1999 sis modality. Perit Dial Int 21:25-35, 2001 15. VanBiesen W, Vanholder R, Lameire N: Impact of 31. Perez Fontan M, Rodriguez-Carmona A, Garcia Fal- pretransplantation dialysis modality on patient outcome af- con T, Tresancos C, Bouza P, Valdes F: Peritoneal dialysis is ter renal transplantation: The role of peritoneal dialysis not a risk factor for primary vascular graft thrombosis after revisited. Perit Dial Int 19:103-106, 1999 renal transplantation. Perit Dial Int 18:311-316, 1998 16. Vanholder R, Heering P, Loo AV, et al: Reduced 32. Murphy BG, Hill CM, Middleton D, et al: Increased incidence of acute renal graft failure in patients treated with renal allograft thrombosis in CAPD patients. Nephrol Dial peritoneal dialysis compared with hemodialysis. Am J Kid- Transplant 9:1166-1169, 1994 ney Dis 33:934-940, 1999 33. van der Vliet JA, Barendregt WB, Hoitsma AJ, Bus- 17. Van Biesen W, Vanholder R, Van Loo A, Van Der kens FG: Increased incidence of renal allograft thrombosis Vennet M, Lameire N: Peritoneal dialysis favorably influ- after continuous ambulatory peritoneal dialysis. Clin Trans- ences early graft function after renal transplantation com- plant 10:51-54, 1996 pared to hemodialysis. Transplantation 69:508-514, 2000 34. Ojo AO, Hanson JA, Wolfe RA, et al: Dialysis modal- 18. Bleyer AJ, Burkart JM, Russell GB, Adams PL: ity and the risk of allograft thrombosis in adult renal trans- Dialysis modality and delayed graft function after cadaveric plant recipients. Kidney Int 55:1952-1960, 1999 renal transplantation. J Am Soc Nephrol 10:154-159, 1999 35. Vats AN, Donaldson L, Fine RN, Chavers BM: Pre- 19. Cacciarelli TV, Sumrani NB, DiBenedetto A, Hong transplant dialysis status and outcome of renal transplanta- JH, Sommer BG: The influence of mode of dialysis pretrans- tion in North American children: A NAPRTCS Study. North plantation on long-term renal allograft outcome. Ren Fail American Pediatric Renal Transplant Cooperative Study. 15:545-550, 1993 Transplantation 69:1414-1419, 2000 20. Cosio FG, Alamir A, Yim S, et al: Patient survival 36. Passalacqua JA, Wiland AM, Fink JC, Bartlett ST, after renal transplantation: I. The impact of dialysis pre- Evans DA, Keay S: Increased incidence of postoperative transplant. Kidney Int 53:767-772, 1998 infections associated with peritoneal dialysis in renal trans- 21. Davies SJ, Russell L, Bryan J, Phillips L, Russell GI: plant recipients. Transplantation 68:535-540, 1999 Comorbidity, urea kinetics, and appetite in continuous ambu- 37. Kang Z, Fang G, Chen W: A comparative study of the latory peritoneal dialysis patients: Their interrelationship outcome of renal transplantation in peritoneal dialysis and and prediction of survival. Am J Kidney Dis 26:353-361, hemodialysis patients. Chin Med Sci J 7:49-52, 1992 1995 38. Miemois-Foley J, Paunio M, Lyytikainen O, Salmela 22. Triolo G, Segoloni GP, Salomone M, et al: Compari- K: Bacteremia among kidney transplant recipients: A case- son between two dialytic populations undergoing renal trans- control study of risk factors and short-term outcomes. Scand plantation. Adv Perit Dial 6:72-75, 1990 J Infect Dis 32:69-73, 2000 23. Evangelista JB Jr, Bennett-Jones D, Cameron JS, et 39. Rubin J, Kirchner KA, Raju S, Krueger RP, Bower al: Renal transplantation in patients treated with haemodialy- sis and short term and long term continuous ambulatory JD: CAPD patients as renal transplant patients. Am J Med peritoneal dialysis. Br Med J 291:1004-1007, 1985 Sci 294:175-180, 1987 24. Donnelly PK, Lennard TW, Proud G, et al: Continu- 40. Cosio FG, Alamir A, Yim S, et al: Patient survival ous ambulatory peritoneal dialysis and renal transplantation: after renal transplantation: I. The impact of dialysis pre- A five year experience. Br Med J 291:1001-1004, 1985 transplant. Kidney Int 53:767-772, 1998 25. O’Donoghue D, Manos J, Pearson R, et al: Continu- 41. Misra M, Vonesh E, Van Stone JC, Moore HL, ous ambulatory peritoneal dialysis and renal transplantation: Prowant B, Nolph KD: Effect of cause and time of dropout A ten-year experience in a single center. Perit Dial Int on the residual GFR: A comparative analysis of the decline 12:242, 245-249, 1992 of GFR on dialysis. Kidney Int 59:754-763, 2001 26. Cacciarelli TV, Sumrani NB, DiBenedetto A, Hong 42. Lopez-Cepero M, Sanders CE, Buggs J, Bowers V: JH, Sommer BG: The influence of mode of dialysis pretrans- Sensitization of renal transplant candidates by cryopre- plantation on long-term renal allograft outcome. Ren Fail served cadaveric venous or arterial allografts. Transplanta- 15:545-550, 1993 tion 73:817-819, 2002 27. Snyder JJ, Kasiske BL, Gilbertson DT, Collins AJ: A 43. Nishimoto A, Matsumoto Y: Increase of peripheral comparison of transplant outcomes in peritoneal and hemo- natural killer T cells in hemodialysis patients. Clin Nephrol dialysis patients. Kidney Int 62:1423-1430, 2002 55:121-126, 2001 28. Perez Fontan M, Rodriguez-Carmona A, Bouza P, et 44. Cooper AC, Mikhail A, Lethbridge MW, Kemeny al: Delayed graft function after renal transplantation in DM, Macdougall IC: Increased expression of erythropoiesis patients undergoing peritoneal dialysis and hemodialysis. inhibiting cytokines (IFN-gamma, TNF-alpha, IL-10, and Adv Perit Dial 12:101-104, 1996 IL-13) by T cells in patients exhibiting a poor response to 29. Binaut R, Hazzan M, Pruvot FR, Dracon M, Lelievre erythropoietin therapy. J Am Soc Nephrol 14:1776-1784, G, Noel C: Comparative study of chronic ambulatory perito- 2003 neal dialysis versus hemodialysis patients after kidney trans- 45. Yokoyama T, Nitta K, Futatsuyama K, et al: Identifi- plantation: Clinical and financial assessment. Transplant cation of T helper cell subsets in continuous ambulatory Proc 29:2428, 1997 peritoneal dialysis patients. Nephron 89:215-218, 2001 RRT MODALITY AND TRANSPLANT OUTCOME 549

46. Libetta C, Rampino T, Dal Canton A: Polarization of tudinal study, in Clin Transplants 1991, Los Angeles, CA, T-helper lymphocytes toward the Th2 phenotype in uremic UCLA Immunogenetics Center, 1992, pp 71-85 patients. Am J Kidney Dis 38:286-295, 2001 48. US Renal Data System: USRDS 2002 Annual Data 47. Held PJ, Turenne MN, Liska DW, et al: Treatment Report. The National Institutes of Health, National Institute modality patterns and transplantation among the United of Diabetes and Digestive and Kidney Diseases, Bethesda, States pediatric end-stage renal disease population: A longi- MD, 2002 CASE REPORT Nephrotic Syndrome and Renal Failure After Allogeneic Stem Cell Transplantation: Novel Molecular Diagnostic Tools for a Challenging Differential Diagnosis

Paola Romagnani, MD, PhD, Elena Lazzeri, PhD, Benedetta Mazzinghi, PhD, Laura Lasagni, PhD, Stefano Guidi, MD, Alberto Bosi, MD, Calogero Cirami, MD, and Maurizio Salvadori, MD

● Background: Sudden onset of nephrotic syndrome after allogeneic stem cell transplantation is rare and has been associated mostly with membranous glomerulonephritis related to chronic graft-versus-host disease (cGVHD). We report a case of nephrotic syndrome and rapidly progressive renal failure occurring in a young woman 3 years after allogeneic stem cell transplantation from her HLA-identical brother. In the renal biopsy, a diffuse mononuclear cell infiltrate was observed. Furthermore, histological analysis, immunofluorescence, and electron microscopy of the kidney specimen defined the diagnosis as minimal change disease, a T-cell–mediated glomerulopathy associated with lymphoproliferative disorders, but that has never been described as an isolated manifestation of cGVHD. Methods: The differential diagnosis was performed by using immunohistochemistry and laser capture microdissec- tion combined with Taq-Man quantitative polymerase chain reaction. Results: Infiltrating mononuclear cells in renal tissue consisted of T cells expressing DNA levels of a Y chromosome–specific gene quantitatively similar to those observed in a male subject, showing that these cells derived from the transplant donor and definitely excluding leukemia relapse. However, the large number of infiltrating T cells allowed the possibility that in this patient, minimal change disease could be related to an atypical form of GVHD. Conclusion: This is the first study to use molecular techniques to show the differential diagnosis of nephrotic syndrome after allogeneic stem cell transplan- tation. This novel method approach might represent a key tool to characterize kidney infiltrate after allogeneic stem cell transplantation. Am J Kidney Dis 46:550-556. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Laser microdissection; minimal change disease; graft-versus-host disease; real-time polymerase chain reaction; Taq-Man.

HE APPEARANCE OF an isolated mono- ease (GVHD). Leukemic relapse can present as T nuclear cell (MNC) infiltrate in a solid either hematopoietic (involving BM and periph- organ is a frequent event in a bone marrow (BM) eral blood) or extramedullary (involving other transplant recipient. The differential diagnosis sites).1 Extramedullary relapses of acute leuke- usually involves 2 important pathological condi- mia after BM transplantation (BMT) are as fre- 2 tions: leukemic relapse and graft-versus-host dis- quent as 50% of all relapses. Approximately 40% of extramedullary relapses occur without concomitant hematopoietic relapse.3 The central From the Interdepartmental Laboratory of Cellular and nervous system and testes, so-called sanctuary Molecular Nephrology, Department of Clinical Pathophysi- sites, are the predominant sites of isolated ex- ology, Center for Research, Transfer and High Education tramedullary disease. However, relapses can in- DE NOVO therapies; University of Florence, Italy. volve other isolated sites, including the kid- Received March 25, 2005; accepted in revised form May 1-5 26, 2005. ney. Generally, treatment of extramedullary Originally published online as doi:10.1053/j.ajkd.2005.05.026 relapses consists of local radiotherapy and/or on July 25, 2005. chemotherapy,1-5 but the outcome is closely re- Supported in part by funds from the Associazione Italiana lated to prompt treatment initiation. per la Ricerca sul cancro; Azienda Ospedaliera of Florence; Chronic GVHD (cGVHD) develops at least 2 Italian Ministry of Health; and Italian Ministry of University and Scientific Research. to 3 months after BMT, presents as a systemic Address reprint requests to Paola Romagnani, MD, PhD, autoimmune disorder, and occurs in approxi- Interdepartmental Laboratory of Cellular and Molecular mately 50% of long-term survivors.6 Symptoms Nephrology, University of Florence, Viale Pieraccini 6, and signs involve skin, liver, mouth, eyes, and 50139, Firenze, Italy. E-mail: [email protected]fi.it; gastrointestinal or upper respiratory system. Al- [email protected] © 2005 by the National Kidney Foundation, Inc. though liver and skin involvement is very fre- 0272-6386/05/4603-0020$30.00/0 quent, selective renal involvement is rare. When doi:10.1053/j.ajkd.2005.05.026 other characteristic clinical signs are lacking, the

550 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 550-556 MOLECULAR DIAGNOSIS OF NEPHROTIC SYNDROME 551 isolated presence of a mononuclear infiltrate in a tures and infiltrating MNCs were dissected separately from 2 solid organ represents a very challenging diagno- areas of identical proportions on the same sections by means sis based on currently available diagnostic as- of an ultraviolet laser that performs circumferential dissec- tion of selected tissue areas, precisely following a drawn says. incision path. By this cold ablation, the material to be We report the case of a patient who developed extracted is never directly exposed to the laser. Microdis- nephrotic syndrome 3 years after allotransplanta- sected tissue areas were measured, documented, and col- tion performed as treatment for acute leukemia. lected on the cap of the eppendorf (0.5 mL) containing the Renal biopsy was performed, and results, ob- Qiagen (Hilden, Germany) lysis buffer for nucleic acid extraction. For each cell population, the procedure was tained by means of light microscopy, immunoflu- repeated 3 times in each case. Specificity of selected cells orescence, and electron microscopy analysis, were was evaluated by means of direct microscopic visualization. compatible with minimal change disease (MCD), a glomerulopathy usually occurring as nephrotic Molecular Biology Studies syndrome, that has been associated strongly with DNA from laser microdissected tissue fractions was ex- lymphoproliferative disorders, especially Hodgkin tracted and isolated by using the QIAamp DNA Micro kit lymphoma and acute lymphoblastic leukemia.7,8 (Qiagen) according to instructions given by the manufac- turer. Conversely, only 1 case of electron microscopy– Samples were immediately used for PCR analysis accord- proven MCD has been reported in the literature, in ing to previously described protocols.11 Appropriate positive a patient with a severe systemic form of GVHD.9 and negative controls were added to each reaction. In addi- In this study, laser capture microdissection (LCM) tion, to ensure that nonspecific transfer of cells to the cap combined with Taq-Man quantitative polymer- could occur, the extraction procedure was performed with 2 caps that were placed on the tissue sample without laser ase chain reaction (PCR) were applied for the activation and 2 caps with a cut nude area adjacent to the first time to show the differential diagnosis of stained tissue. Primers used for PCR amplification (Assays nephrotic syndrome after allogeneic stem cell on Demand; Applied Biosystems; Warrington, UK) targeted transplantation. The results unequivocally ex- DNA specific for the sex-determining region Y (SRY) tran- cluded leukemia relapse and suggested that in scription factor gene, localized on chromosome Y (Yp11.3). As controls, pooled DNA extracted from tissue samples this case, MCD might represent an atypical form (25 ng each) obtained from 30 male and 30 female subjects of GVHD. were used. Taq-Man real-time quantitative PCR was per- formed as described elsewhere by using the ABI Prism METHODS 7900HT Sequence Detection System (Applied BioSys- tems).11 Immunohistologic Studies Renal biopsy samples were stained with hematoxylin- CASE REPORT eosin or periodic acid–Schiff and examined by means of A 31-year-old white woman underwent allogeneic stem light microscopy. They also were stained with fluorescein cell transplantation from her HLA-identical brother because isothiocyanate–labeled goat antihuman immunoglobulin A she had acute leukemia. The leukemia was FAB L1 peroxi- (IgA), IgG, IgM, C1q, properdin, C3, C3d, and fibrinogen dase negative, and approximately 60% of cells in the BM antibody. Histological diagnosis was implemented by the ϩ ϩ ϩ ϩ ϩ were CD34 , CD13 , CD33 , CD15 , CD4 , and HLA- characterization of inflammatory infiltrates, performed using ϩ DR blasts. Conventional cytogenetic analysis was unsuc- the following antibodies: CD3 (1:20), CD4 (1:40), CD8 cessful, and BCR/ABL detection by means of fluorescence (1:100), CD20 (1:100), CD68 (1:50; all from Dako, Copen- in situ hybridization was negative. After remission with 1 hagen, Denmark), and CD34 (1:100; BD Biosciences, San course of induction therapy, the patient completed 1 cycle of Diego, CA). Immunohistochemical studies were performed consolidation with cytarabine and mitoxantrone. During according to previously published protocols.10 consolidation therapy, acute respiratory distress syndrome was diagnosed that required intensive treatment. Cyclospor- Laser Capture Microdissection ine A and methotrexate were administered as prophylaxis for Frozen renal biopsy samples were subjected to LCM with GVHD. The posttransplantation period was uneventful. Peri- the Palm-Axiovert 200 System (Zeiss, Jena, Germany), odic controls always yielded negative results. Two years according to suggestions given by the manufacturer’s proto- after allogeneic stem cell transplantation, at a routine hema- cols. Briefly, serial 4-␮m–thick cut sections were mounted tologic control, all hematochemic parameters were normal, on a polyethylene naphthalate membrane slide (Palm Mem- but 2 weeks later, the patient began to experience general- braneSlides); immunostained with CD3, CD4, CD8, or CD34 ized edema, mainly periorbital and pretibial, that resulted in antibodies by means of the avidin-biotin-peroxidase method; weight gain. Oral diuretic therapy was initiated. Blood and counterstained for 1 minute with hematoxylin. Selection analyses showed an increase in serum creatinine level to 1.7 of cells for LCM was guided by immunohistochemical mg/dL (150 ␮mol/L), and urinalysis showed 25 to 50 red detection of T lymphocytes, marked by CD3. Tubular struc- blood cells/high-power field and 2ϩ proteinuria, even though 552 ROMAGNANI ET AL renal function was normal only 2 weeks before. At this point, the patient was admitted to our hospital. On physical examination, the patient’s weight was 54 kg, height was 1.65 m, blood pressure was 150/90 mm Hg, heart rate was 90 beats/min, and she was afebrile. Head, neck, chest, and heart examination findings were negative. No abnormalities were shown by abdominal, neurological, and musculoskeletal examinations. White blood cell count was 5.19 ϫ 103/␮L (5.19 ϫ 109/L), with a balanced differential white blood cell count; hematocrit was 35%; platelet count was 290,000 ϫ 103/␮L (290,000 ϫ 109/L); international normalized ratio was 0.9; activated partial thromboplastin time was 28 seconds; and fibrinogen level was 480 mg/dL (14.1 ␮mol/L). All viral markers were negative. Antistrepto- lysin antibody titer was normal. Serum levels of electrolytes, aminotransferases, and total bilirubin were normal. Total protein level was 4.1 g/dL (41 g/L), albumin level was 1.19 ␣ ␥ g/dL (11.9 g/L), 2-globulin level was increased, and -globu- lin level was decreased. Serum IgG level was 389 mg/dL (3.89 g/L); IgA, IgM, and complement levels were normal; and total cholesterol level was 390 mg/dL (10.09 mmol/L). Serum creatinine level was 1.8 mg/dL (159 ␮mol/L), and blood urea nitrogen level was 59 mg/dL (21.1 mmol/L). A 24-hour urine collection had 12 g of protein. Therefore, nephrotic syndrome was diagnosed. Conventional Analysis of the Renal Biopsy and BM Aspirate A renal biopsy was performed. Light microscopy showed an infiltrate of MNCs scattered in the interstitium (Fig 1A) and periglomerular areas. However, glomeruli were normal, without crescents (which are pathognomonic for rapidly progressive glomerulonephritis), areas of segmental glomer- ulosclerosis (thus excluding focal segmental glomeruloscle- rosis), or membrane thickening (suggestive of membranous glomerulonephritis), also confirmed by means of periodic acid–Schiff reaction (Fig 1B). Immunofluorescence did not show the presence of IgM, IgG, IgA, C3, or fibrinogen. Electron microscopy showed effacement of podocyte foot processes and confirmed the absence of dense deposits (Fig 1C). Glomerular analysis was consistent with MCD; the most relevant finding was the infiltration of MNCs (Fig 1). The occurrence of an isolated MNC infiltrate in a solid organ is a frequent feature of cGVHD, but given the absence of other symptoms and the strong association of MCD with 7,8 lymphoproliferative disorders, we first had to rule out the Fig 1. Renal biopsy specimen showing (A) diffuse possibility of leukemia relapse. BM aspirate excluded dis- MNC infiltrate (hematoxylin and eosin; original magni- -ease relapse at the medullary level. Meanwhile, the patient’s fication ؋100), (B) normal glomerular structures (peri ,(proteinuria reached protein of 28 g/d, and serum creatinine odic acid–Schiff reaction; original magnification ؋400 levels increased to 3.7 mg/dL (327 ␮mol/L). and (C) effacement of podocyte foot processes at electron microscopy (arrow). Molecular Analysis of Infiltrating MNCs Despite the negativity of the BM aspirate, leukemia re- An anti-CD34 antibody positively stained small vessels, lapse could not be unequivocally excluded because approxi- but did not show reactivity with infiltrating MNCs (Fig 2D), mately 40% of extramedullary relapses can occur in the suggesting that these cells did not belong to the original presence of persistent remission at the BM level. Immunohis- leukemic population. To provide direct evidence, MNCs tochemistry performed on cryostat sections obtained from were recovered by using LCM, microextraction of DNA was the renal biopsy specimen showed that infiltrating MNCs performed, and the presence of the SRY gene (a transcription were CD3ϩ (T lymphocytes), prevalently CD8ϩ, and the factor selectively localized on the Y chromosome) was other was CD4ϩ (Fig 2A to C). assessed by means of Taq-Man quantitative PCR. As con- MOLECULAR DIAGNOSIS OF NEPHROTIC SYNDROME 553

Fig 2. Immunohistochemistry on renal biopsy specimen showing that infiltrating MNCs are (A, B) CD8؉ (red) and (.C) CD4؉ (red), but (D) not CD34؉ (red). (Original magnification: [A, C, D] ؋100; [B] ؋400) trols, samples of identical dimensions were obtained from serum creatinine levels were 0.9 mg/dL (80 ␮mol/L). Nine the kidney in an adjacent area of the same biopsy specimen months after steroid therapy withdrawal, the patient is still in which MNCs were absent, as well as from another biopsy healthy. specimen obtained from the kidney of a male subject (Fig 3A and B). Because the patient was female and the donor DISCUSSION was male, cells resulting from an extramedullary leukemia relapse should show an XX karyotype, whereas an immune The ever-growing number of BMTs and the reaction caused by donor-derived lymphocytes should be increasing short-term and long-term survival of provoked by cells bearing a Y chromosome. Taq-Man quan- patients undergoing BMT have considerably in- titative PCR performed on DNA extracted from these samples creased the prevalence of renal disorders. Cur- showed that infiltrating MNCs showed a number of copies rently, 5% to 15% of all patients undergoing of the SRY gene, similar to that observed in the kidney of a male subject (Fig 3C). Conversely, no copies of SRY DNA BMT may develop acute renal failure, and 5% to were found in the adjacent area from the kidney biopsy 20% of long-term BMT survivors will develop specimen of the patient, in which no infiltrating MNCs were chronic renal failure.12 The typical acute renal detectable (Fig 3C). These findings unequivocally show that failure seen in the first 30 days after BMT has infiltrating MNCs in the kidney of the patient are T cells been associated with sepsis, hypotension, use of derived from the transplant donor and definitely exclude the nephrotoxic antibiotics, and concurrent liver possibility of a leukemia relapse. We have not proved that 12 GVHD caused the MCD, but the large number of infiltrating disease. Chronic renal failure after BMT was T cells and the observation that they were mainly CD8ϩ recognized more than 10 years ago as being allows us to suggest this possibility.6 related to the total-body irradiation used for Finally, the patient was treated with oral prednisone in a conditioning,12 and its incidence has increased in daily dosage of 1 mg/kg of body weight for 12 weeks, which then was tapered slowly in an attempt to reduce the likeli- the last years. The usual time of occurrence of hood of relapse. After 3 months of treatment, proteinuria radiation nephropathy is 8 to 12 months after disappeared and renal function was completely restored; BMT. The clinical presentation is azotemia, hy- 554 ROMAGNANI ET AL

Fig 3. (A, B) Infiltrating MNCs were recovered from the renal biopsy specimen by using LCM on an area of 21,652 ␮m2 (A, B; red area) and an identical area also was recovered from a zone free of infiltrating MNCs (A, B; blue area). (C) (Left) Quantitative assessment of DNA levels shows that infiltrating cells (red area), but not renal cells (blue area), in our patient show levels of the Y chromosome similar to those of a male subject. (Right) Representative amplification plot of the microdissected areas are shown. pertension, and, disproportionately, severe ane- cases of clinically significant GVHD-associated mia that, in the absence of appropriate treatment, renal damage other than membranous nephropa- leads to renal failure requiring dialysis or kidney thy have been recorded in the English literature. transplantation.13,14 Finally, acute and chronic All these patients presented with heavy protein- renal failure also may be caused by the nephro- uria/nephrotic syndrome. Focal segmental glo- toxicity of the calcineurin inhibitors cyclospor- merulosclerosis, diagnosed in 3 of these patients, ine A or tacrolimus, which are administered in was the most common form of glomerular injury, the first few months after allogeneic BMT to followed by 2 cases of IgA nephropathy and 2 prevent GVHD. cases of crescentic glomerulonephritis. Only 1 However, after BMT, sudden onset of ne- case of electron microscopy–proved MCD was phrotic syndrome and renal failure are very rare, reported, which appeared in the context of a and our patient had normal urinalysis results and severe systemic form of GVHD.9,15-18 renal function only 2 weeks before the onset of MCD is a disorder usually occurring as ne- edema. Nephrotic syndrome caused by membra- phrotic syndrome, which is related to injuries to nous glomerulonephritis has been well docu- the glomerular epithelial foot processes caused mented as one of the rare clinical manifestations by the activity of cytokines released by activated of cGVHD. To the best of our knowledge, only 8 T lymphocytes. This, in turn, leads to decreased MOLECULAR DIAGNOSIS OF NEPHROTIC SYNDROME 555 synthesis of polyanions, which constitute the REFERENCES normal charge barrier to the filtration of macro- 1. Cormier MG, Armin A, Daneshgari F, Castelli M: molecules, and albumin leakage occurs. MCD Unusual extramedullary relapse of acute lymphoblastic leu- can be idiopathic, but also can be related to drugs kaemia in a bone marrow transplant patient. J Surg Oncol 36:290-294, 1987 (nonsteroidal anti-inflammatory drugs, rifampin, 2. Lee KH, Lee JH, Kim S, Lee JS, Kim SH, Kim WK: interferon, penicillin, and trimethadione), toxic High frequency of extramedullary relapse of acute leukemia agents (mercury, lithium, and bee stings), infec- after allogeneic bone marrow transplantation. Bone Marrow tions (mononucleosis, human immunodeficiency Transplant 26:147-152, 2000 virus, and immunization), or tumors (most com- 3. Au WY, Lie AKW, Liang R, Kwong YL: Isolated monly Hodgkin lymphoma and lymphoblastic extramedullary relapse of acute lymphoblastic leukaemia after allogeneic bone marrow transplantation. Bone Marrow leukemia; rarely carcinomas). The occurrence of Transplant 24:1137-1140, 1999 nephrotic syndrome and renal failure during the 4. Chong G, Byrnes G, Szer J, Grigg A: Extramedullary course of a lymphoproliferative disorder is fre- relapse after allogeneic bone marrow transplantation for quent.7,8 Glomerulonephritis may precede, coex- haematological malignancy. Bone Marrow Transplant 26: ist, or follow the diagnosis of lymphatic malig- 1011-1015, 2000 5. Kebaili K, Manel AM, Chapelon C, Taylor P, Philippe nancy by several years. At autopsy, up to 90% of N, Bertrand Y: Renal enlargement as presentation of isolated cases (range, 6% to 90%) have kidney infiltra- renal relapse in childhood leukemia. J Pediatr Hematol tion.19-21 Thus, given the strong association of Oncol 22:454-456, 2000 MCD with lymphoproliferative disorders and the 6. Atkinson K: Chronic graft-versus-host disease. Bone high frequency of extramedullary relapses after Marrow Transplant 5:69-82, 1990 BMT,3 in our patient, leukemia relapse could be 7. Da’as N, Polliack A, Cohen Y, et al: Kidney involve- ment and renal manifestations in non-Hodgkin’s lymphoma strongly suspected. and lymphocytic leukemia: A retrospective study in 700 Nephrotic syndrome in an adult subject repre- patients. Eur J Haematol 67:58-64, 2001 sents an absolute indication for renal biopsy, 8. Dabbs DJ, Striker LM, Mignon F, Striker G: Glomeru- which usually allows one to settle the question. lar lesions in lymphomas and leukemias. Am J Med 80:63- However, in a patient with a BMT-treated leuke- 70, 1986 9. Spalding EM, Watkins S, Warwicker P, Walker JV, mia, involvement of a single organ by an MNC Morich D, Anasetti C: Minimal-change nephrotic syndrome infiltrate frequently occurs, and it always raises after cyclosporine withdrawal in a marrow transplant recipi- the problem of a differential diagnosis between ent. Am J Kidney Dis 26:532-534, 1995 extramedullary leukemia relapse or cGVHD. Al- 10. Romagnani P, Annunziato F, Lasagni L, et al: Cell though the clinical context usually helps achieve cycle-dependent expression of CXC chemokine receptor 3 by endothelial cells mediates angiostatic activity. J Clin the right diagnosis, in some cases, the absence of Invest 107:53-63, 2001 other symptoms makes the differential diagnosis 11. Lasagni L, Francalanci M, Annunziato F, et al: An impossible on the basis of currently available alternatively spliced variant of CXCR3 mediates IP-10, Mig diagnostic assays. In our patient, LCM combined and I-TAC induced-inhibition of endothelial cell growth and with Taq-Man quantitative PCR was used as a acts as functional receptor for PF-4. J Exp Med 197:1537- rapid highly specific diagnostic tool. This is the 1549, 2003 12. Cohen EP: Renal failure after bone-marrow transplan- first study to use molecular techniques to show tation. Lancet 357:6-7, 2001 the differential diagnosis of nephrotic syndrome 13. Parikh CR, McSweeney PA, Korular D, et al: Renal after allogeneic stem cell transplantation. Such a dysfunction in allogeneic hematopoietic cell transplantation. novel approach might be suitable for different Kidney Int 62:566-573, 2002 clinical conditions characterized by infiltration 14. Hamawi K, de Magalhaes-Silverman M, Bertolatus JA: Outcomes of renal transplantation following bone mar- of the kidney and other organs by MNCs in BM row transplantation. Am J Transplant 3:301-305, 2003 transplant recipients. Instead of the SRY gene, 15. Hisse C, Goldschmidt E, Santelli G, Charpentier B, HLA genes can be used if the recipient and donor Machover D, Fries D: Membranous nephropathy in a bone are the same sex. marrow transplant recipient. Am J Kidney Dis 11:188-191, 1988 16. Nergizoglu G, Keven K, Ates K, et al: Chronic ACKNOWLEDGMENT graft-versus-host disease complicated by membranous glo- The authors thank Mario Serio, Florence, Italy, for critical merulonephritis. Nephrol Dial Transplant 14:2461-2463, review of the manuscript. 1999 556 ROMAGNANI ET AL

17. Akar H, Keven K, Çelebi H, et al: Nephrotic syn- causing nephrotic syndrome after bone marrow trans- drome after allogeneic peripheral blood stem cell transplan- plantation. Histopathology 45:648-651, 2004 tation. J Nephrol 15:79-82, 2002 20. Phillips JK, Bass PS, Majumdar G, Davies DR, Jones 18. Kimura S, Horie A, Hiki Y, et al: Nephrotic syndrome NF, Pearson TC: Renal failure caused by leukaemic infiltra- with crescent formation and massive IgA deposition follow- tion in chronic lymphocytic leukaemia. J Clin Pathol 46:1131- ing allogeneic bone marrow transplantation for natural killer 1133, 1993 cell leukemia/lymphoma. Blood 101:4219-4221, 2003 21. Barcos M, Lane W, Gomez GA, et al: An autopsy 19. Chan GS-W, Lam MF, Au WY, et al: IgA nephropathy study of 1206 acute and chronic leukemias. Cancer 60:827- complicating graft-versus-host disease, another nephropathy 837, 1987 EDITORIAL Hypertension in Individuals at Risk for Autosomal Dominant Polycystic Kidney Disease: To Screen or Not to Screen?

detection and management of hypertension, pro- Related Article, p. 415 gressive kidney failure, and its complications; (3) provision of more detailed and specific infor- UTOSOMAL DOMINANT polycystic kidney mation regarding prognosis and risk for compli- A disease (ADPKD) is one of the most com- cations; (4) the ability to make more informed mon genetic disorders. Recent advances in the reproductive choices; (5) motivation for en- understanding of the molecular and cell biology of hanced compliance and medical follow-up; and cystic kidney disease, including discoveries of the (6) the ability to share otherwise blood-group– likely participation of the primary cilium in the incompatible kidneys across extended ADPKD pathogenesis of cystogenesis and the role of cyclic families to maximize the number of living donor adenosine monophosphate in the enhancement of kidney transplantations. Unfortunately, at this growth and fluid secretion of cystic epithelium, time, specific therapy with proven efficacy tar- provide a basis for great excitement and hope geted at renal progression is not available. regarding prospects for the development of thera- The overall and cardiovascular benefits of peutic interventions. The recent demonstration of treating hypertension in patients with ADPKD the marked inhibition of cyst growth in multiple are well summarized in the article by Taylor et al.2 animal models of cystic kidney disease by the The Seventh Report of the Joint National Com- vasopressin V2 receptor antagonists OPC31260 mittee on Prevention, Detection, Evaluation, and and OPC41061 has led to substantial interest in the Treatment of Blood Pressure recommends a use of these agents in clinical investigations of lower-than-usual blood pressure goal of less than human ADPKD.1 Pending proof of efficacy to 130/80 mm Hg for individuals with proteinuria delay the progression of cystic kidney disease by (atypical of ADPKD) and/or those with a glomer- these agents or others (ie, angiotensin-converting ular filtration rate less than 60 mL/min/1.73 m2 enzyme inhibitors and/or angiotensin receptor (Ͻ1.00 mL/s),3 in recognition of the high cardio- blockers), it is reasonable to ask whether the at-risk vascular risk of this population. There are several individual in an ADPKD family should undergo compelling reasons why individuals with AD- diagnostic testing in the absence of signs or symp- PKD, even with normal kidney function, also toms of illness. should be treated to this lower blood pressure Taylor et al2 report in the present issue of the goal. Specifically, hypertension occurs earlier in American Journal of Kidney Diseases that indi- the course of disease than with other forms of viduals born between 1951 and 1974 were diag- CKD, and cardiovascular disease also has been nosed with ADPKD at an earlier age than a shown to be prevalent at relatively early stages of cohort born before 1951. It is unclear whether the disease. In addition, the progressive loss of earlier diagnosis in the younger cohort reflects kidney function, which further increases the risk greater understanding of the genetic nature of for cardiovascular disease, is a predictable out- ADPKD or the more widespread use of kidney come of ADPKD. An analysis of long-term fol- ultrasound. The benefits of making an earlier low-up of the Modification of Diet in Renal diagnosis need to be considered, along with the Disease Study cohort showed that a relatively potential for anxiety and discrimination that may short interval (median treatment time, 2.2 years) accompany the knowledge that one carries a of low target blood pressure control (mean arte- chronic and, ultimately, fatal disease. rial pressure, Ͻ92 mm Hg) decreased the relative The many advantages of knowing one’s dis- ease status at an early and asymptomatic stage include: (1) the ability to institute specific therapy Originally published online as doi:10.1053/j.ajkd.2005.07.016 on August 2, 2005. targeted at delaying the progression of kidney © 2005 by the National Kidney Foundation, Inc. disease (although, at present, a specific therapy 0272-6386/05/4603-0021$30.00/0 to reduce cystogenesis does not exist); (2) earlier doi:10.1053/j.ajkd.2005.07.016

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 557-559 557 558 PERRONE AND MISKULIN risk for kidney failure after a median of approxi- infection, nephrolithiasis, or pain in an at-risk mately 6 years by one third compared with that individual should mandate diagnostic testing with in individuals treated with usual target blood kidney imaging. In addition, manifestation of pressure control (mean arterial pressure, Ͻ107 any of the extrarenal complications of ADPKD mm Hg).4 This effect was similar irrespective of should be investigated fully with all appropriate the cause of kidney disease, including ADPKD. diagnostic modalities. The earlier detection of Whether aggressive blood pressure lowering has hypertension, given the young age of onset and even greater benefits if applied at an early stage high cardiovascular risk of this population, ar- of disease (glomerular filtration rate Ͼ 60 mL/ gues in favor of making an earlier diagnosis. The min/1.73 m2 [Ͼ1.00 mL/s/1.73 m2]) will be choice of antihypertensive agents is not necessar- assessed as part of the National Institutes of ily affected by precise knowledge of ADPKD, Health–sponsored Halt Progression in Autoso- but the level of blood pressure targeted possibly mal Dominant Polycystic Kidney Disease (HALT- should be lower for an individual with ADPKD, PKD) Study, scheduled to start in the fall of even with normal kidney function, than for treat- 2005. In terms of an earlier diagnosis affecting ment of patients with essential hypertension. the choice of antihypertensive agent, angiotensin Increased awareness of ADPKD as an inher- blockade already is used commonly for treat- ited disorder has led to the possibility of earlier ment of hypertension in patients with ADPKD.5 diagnostic screening and potentially earlier imple- Recent guidelines6 suggest no preference for an mentation of medical interventions that could agent in the absence of proteinuria, so early impact on disease progression and extrarenal diagnosis would not alter this aspect of hyperten- manifestations. We believe that the potential sion management. medical benefits of screening for ADPKD are Given the actual or potential medical benefit of evident. However, concerns about establishing a making a diagnosis of ADPKD, what is the down- “preexisting condition” for insurance purposes side of testing asymptomatic members of an may rightfully raise concerns about testing for ADPKD family? For some individuals, the emo- ADPKD and thereby prevent initiation of inten- tional burden and anxiety of knowing their disease sive antihypertensive therapy and other appropri- status prevents them from seeking testing for ate interventions. It also is possible that some ADPKD. However, for others, the major concerns family members in ADPKD families avoid medi- about presymptomatic testing relate to concerns cal encounters entirely because of such concerns, about insurance- or employment-based discrimina- but there is only anecdotal information regarding tion because of their risk for needing expensive this possibility. Access to clinical research trials medical care for kidney failure and the need for for individuals at earlier stages of disease, espe- dialysis or transplantation. The extent to which this cially those with well-preserved kidney function, is a realistic concern has not been defined. In 1996, would be precluded should such individuals re- Golin et al7 surveyed individuals with ADPKD and main undiagnosed. Therefore, it is unfortunate found that 57% made employment choices based and potentially tragic that some individuals af- on the availability of employer-provided health fected by ADPKD may not avail themselves of insurance, 37% stayed on the job because of health appropriate medical interventions or access to insurance, and 30% had previously been denied therapeutic clinical trials because of concerns health insurance. A recent commentary suggests about health insurance or other discrimination. that genetic discrimination is increasingly com- Genetic nondiscrimination should be the law of mon.8 Concerns about this issue have risen to the the land. federal level. The Senate has unanimously passed legislation addressing genetic discrimination (S. Ronald D. Perrone, MD 306, the Genetic Information Nondiscrimination Dana C. Miskulin, MD Act of 2005). A similar version of the legislation Department of Medicine (HR 1227) awaits action in the House of Represen- Division of Nephrology tatives. Tufts–New England Medical Center Few would disagree that the presence of such Tufts University School of Medicine signs or symptoms as hematuria, urinary tract Boston, Massachusetts EDITORIAL 559

REFERENCES 4. Sarnak M, Greene T, Wang X, et al: The effect of a 1. Wang X, Gattone V, Harris PC, Torres VE: Effective- lower target blood pressure on the progression of kidney ness of vasopressin V2 receptor antagonists OPC-31260 and disease: Long-term follow-up of the Modification of Diet in OPC-41061 on polycystic kidney disease development in Renal Disease Study. Ann Intern Med 142:342-351, 2005 the PCK rat. J Am Soc Nephrol 16:846-851, 2005 5. Chapman A, Guay-Woodford L, Grantham J, et al: 2. Taylor M, Johnson AM, Tison M, Fain P, Schrier RW: Renal structure in early autosomal-dominant polycystic kidney Earlier diagnosis of autosomal dominant polycystic kidney disease (ADPKD): The Consortium for Radiologic Imaging disease: Importance of family history and implications for Studies of Polycystic Kidney Disease (CRISP) cohort. Kidney cardiovascular and renal complications. Am J Kidney Dis Int 64:1035-1045, 2003 46:415-423, 2005 6. National Kidney Foundation: K/DOQI Clinical Prac- 3. Chobanian A, Bakris G, Black H, et al, for the National tice Guidelines on Hypertension and Antihypertensive Agents Heart, Lung, and Blood Institute Joint National Committee in Chronic Kidney Disease. Am J Kidney Dis 43:S1-S290, on Prevention, Detection, Evaluation, and Treatment of 2004 (suppl 1) High Blood Pressure; National High Blood Pressure Educa- 7. Golin C, Johnson A, Fick G, Gabow P: Insurance for tion Program Coordinating Committee: Seventh Report of autosomal dominant polycystic kidney disease patients prior the Joint National Committee on Prevention, Detection, to end-stage renal disease. Am J Kidney Dis 27:220-223, Evaluation, and Treatment of High Blood Pressure: The JNC 1996 7 Report. JAMA 289:2560-2572, 2003 (erratum in JAMA 8. Billings P: Genetic nondiscrimination. Nat Genet 37: 290:197, 2003) 559-560, 2005 CORE CURRICULUM IN NEPHROLOGY Tubulointerstitial Diseases

Gregory L. Braden, MD, Michael H. O’Shea, MD, and Jeffrey G. Mulhern, MD

Type 2 interstitial cells may act as den- INTRODUCTION dritic cells, which are capable of antigen presentation Definitions ● Space between interstitial cells contains Tubulointerstitial nephritis (TIN) may be ei- types 1 and 3 collagen ther acute or chronic: Medullary interstitium Acute TIN ● Contains 3 types of cells: ● Associated with acute renal failure (ARF), Type 1 cells do not produce erythropoi- which develops over period of days to several etin, but may produce prostaglandins via cyclo-oxygenase 2 (COX-2) weeks due to either acute infection of kidneys or delayed hypersensitivity reaction to medi- Type 2 cells resemble lymphocytes; func- cation (reviewed in another section) tion unknown Type 3 cells located near the vasa recti; function unknown Chronic interstitial nephritis (CIN) ● ● Extracellular matrix of types 1 and 3 colla- Develops over months or years from causes gen lies between cells in Table 1 ● Is associated with progressive loss of glo- Mechanisms of Tubular Interstitial Injury merular filtration rate (GFR) over time and ● Tubulointerstitial response to injury from characterized by many syndromes of renal Table 1 diseases is associated with tubular tubular dysfunction cell proliferation and tubular dilatation and ● Primary CIN is associated with chronic renal cast formation followed by atrophy and/or tubular infection with Epstein-Barr virus apoptosis and fibrosis ● Secondary CIN is due to renal tubular damage from wide variety of causes Pathology of CIN (Table 1) ● Interstitial filtration with lymphocytes, monocytes, macrophages; depending on the Composition of Normal Interstitium etiology, neutrophils, eosinophils, or plasma Cortical interstitium cells accumulate in renal interstitium ● ● Contains 2 types of cells: Tubular atrophy, flattened epithelial cells, Type 1 interstitial cells resemble fibro- tubular dilation occur, also tubular base- blasts and produce erythropoietin ment membrane thickening ● Glomerulosclerosis: loss of glomeruli can occur indirectly from severe tubular dam- From the Department of Medicine, Baystate Medical age in nephron segments of glomerulus or Center, Springfield, MA, and Tufts University School of due to periglomerular fibrosis and segmen- Medicine, Boston, MA. tal sclerosis eventually leading to global Received May 26, 2004; accepted in revised form March glomerular sclerosis 28, 2005. ● Originally published online as doi:10.1053/j.ajkd.2005.03.024 Medullary microcysts from hypokalemia on August 1, 2005. ● Cast formation (thyroidization) occurs, par- Address reprint requests to Gregory L. Braden, MD, ticularly in myeloma, but can be seen in Chief, Renal Division, Baystate Medical Center, 759 Chest- idiopathic CIN nut Street, Springfield, MA 01199. E-mail: Gregory. ● Interstitial fibrosis [email protected] ● © 2005 by the National Kidney Foundation, Inc. Immunofluorescence microscopy is nega- 0272-6386/05/4603-0022$30.00/0 tive except in Sjögren syndrome, lupus, and doi:10.1053/j.ajkd.2005.03.024 myeloma

560 American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 560-572 CORE CURRICULUM IN NEPHROLOGY 561

Table 1. Secondary Causes of CIN (TIN) Daily protein excretion usually Ͻ1.5 g Urinary sediment is bland with a few Category Causes white and red blood cells and, rarely, Infections Chronic pyelonephritis casts VUR Anemia disproportionately severe at same Drugs Analgesic nephropathy GFR due to damage to erythropoietin- Lithium producing cells Nucleoside inhibitors Sodium wasting occurs, but usually mild (cidofovir, tenofovir) Non–anion gap metabolic acidosis re- Calcineurin inhibitors (cyclosporine, tacrolimus) sults from proximal renal tubular acido- Aristolochic acid sis (RTA) with or without Fanconi syn- (Chinese herbs) drome and from types 1 and 4 distal RTA Chemotherapy Renal papillary necrosis is associated (cisplatin, ifosfamide) with analgesics or acute pyelonephritis Toxins Lead nephropathy Kidney stones are associated with meta- Heavy metals (cadmium) bolic or inherited disorders Hematologic/neoplastic Multiple myeloma Nephrogenic diabetes insipidus (NDI) diseases Lymphoproliferative occurs from drugs, metabolic or genetic disorders disorders Light-chain disease Sickle-cell disease Immune-mediated Sarcoidosis PRIMARY (IDIOPATHIC) CIN disease Sjögren syndrome Systemic lupus Pathogenesis erythematosus TIN with uveitis ● May be mediated by Epstein-Barr virus TIN with hypocomplementemia ADDITIONAL READING Metabolic disorders Hypokalemia 1. Becker JL, Miller F, Nuovo GJ, Josepovitz C, Schubach Hypercalcemia WH, Nord EP: Epstein-Barr virus infection of renal proxi- Urate nephropathy mal tubule cells: Possible role in chronic interstitial nephri- Genetic disorders Cystinosis tis. J Clin Invest 104:1673-1681, 1999 Dent disease Primary hyperoxaluria Adenine phosphoribosyl SECONDARY CIN transferase deficiency Primary hyperoxaluria Chronic Pyelonephritis and Reflux Nephropathy Autosomal dominant hypoparathyroidism Overview Karyomegalic interstitial ● nephropathy Chronic pyelonephritis is term used for infection-related CIN without vesicoureteral Miscellaneous Balkan endemic reflux (VUR) nephropathy ● Radiation nephritis Reflux nephropathy with secondary focal Papillary necrosis glomerulosclerosis associated with VUR Inflammatory bowel disease accounts for overwhelming majority of Post acute tubular necrosis cases of CIN associated with bacterial infections of urinary tract Clinical Features and Course of CIN ● Table 2 illustrates diverse abnormalities of Pathogenesis of VUR renal tubular function ● Occurs due to congenital anomalies in ● Compared with chronic glomerulonephritis vesicoureteral junction leading to incompe- in CIN: tent vesicoureteral valves upon bladder Hypertension is less common contraction 562 BRADEN, O’SHEA, AND MULHERN

Table 2. Clinical Features of Chronic ● Staged 1 through 5 based on voiding cys- Tubulointerstitial Disease tourethrogram ● Electrolyte/Acid-Base Disorders Diagnosed in 20% to 35% of infants and Proximal RTA or Fanconi syndrome children after first urinary tract infection Multiple myeloma ● Up to 35% to 45% of asymptomatic sib- Dent disease lings have VUR Cystinosis Sjögren syndrome Unique pathologic features Distal RTA Bacterial pyelonephritis ● Chronic inflammation can lead to xan- VUR thogranulomatous degeneration Lithium ● Reflux nephropathy is characterized by Lead focal and segmental glomerulosclerosis Myeloma Light-chain disease leading to nephrotic-range proteinuria Sjögren disease Systemic lupus Clinical and laboratory features Hypercalcemia ● Recurrent urinary tract infections and acute Hyperkalemic type IV RTA VUR pyelonephritis are common Lead ● Proteinuria can occasionally become Systemic lupus nephrotic Sickle-cell disease Sodium wasting Treatment and outcome Any disorder ● Trials of surgical versus medical therapy Clinical Syndromes show surgery reduces new episodes of Kidney stones Hypercalcemia acute pyelonephritis, but does not influence Hyperoxaluria progressive renal insufficiency or new scar Urate nephropathy formation Dent disease Sarcoidosis ADDITIONAL READING Inflammatory bowel disease Adenosine transferase deficiency 1. Dillon MJ, Goonasekera CD: Reflux nephropathy. J Am NDI Soc Nephrol 9:2377-2383, 1998 Lithium 2. Wheeler D, Vimalachandra D, Hodson EM, Roy LP, Cisplatin Smith G, Craig JC: Antibiotics and surgery for vesi- Hypokalemia coureteric reflux: A meta-analysis of randomized controlled Hypercalcemia trials. Arch Dis Child 88:688-694, 2003 Dent disease ARF Drugs Pyelonephritis Analgesics Analgesic nephropathy Lithium Epidemiology. Calcineurin inhibitors ● Strong association between analgesic ne- Cisplatin phropathy and long-term analgesic use with Nucleoside inhibitors Myeloma analgesic combination medications that con- Lymphoma tain aspirin, phenacetin, and caffeine Systemic lupus ● Association between acetaminophen and Hypercalcemia the metabolite of phenacetin, either alone Uric acid or in combination with aspirin and caffeine, Radiation Papillary necrosis is suggestive, but not definitive Acute pyelonephritis ● Aspirin alone is associated with acute GFR Analgesic nephropathy decreases, particularly in patients on low- sodium diet and elderly patients, but long- term use alone is not associated with analge- sic nephropathy CORE CURRICULUM IN NEPHROLOGY 563

● Long-term nonsteroidal anti-inflammatory Lithium-induced renal diseases drug (NSAID) use has been associated with Lithium is reabsorbed by the renal tubules, CIN in smaller number of patients similar to sodium at nephron sites where sodium Pathogenesis. is reabsorbed. ● Renal damage from analgesics predomi- Pathogenesis. nately affects renal medulla ● NDI primarily is due to inhibition of adenyl- ● Acetaminophen undergoes oxidative metab- ate cyclase (ADH)–dependent aspects of olism via prostaglandin H synthase path- water conservation way, which utilizes glutathione; NSAIDs ● Mechanism of CIN unknown and aspirin deplete cortex and medulla of Unique pathologic features. glutathione, allowing reactive acetamino- ● Unique tubular lesion consisting of microcyst phen metabolites to induce lipid peroxides formation due to cystic dilation of distal and oxygen-free and hydroxyl radicals, tubules lined with columnar epithelium which are toxic to renal tissue proteins ● ● Lithium rarely causes nephrotic syndrome NSAIDs and aspirin inhibit renal prostaglan- Clinical features. din production, which induces medullary ● NDI vasoconstriction with consequent ischemic Polydipsia in 40% and polyuria in up to injury and papillary necrosis 20% of patients Clinical and laboratory features. ● ● RTA More common in women, typically with a Incomplete distal RTA in up to 50% of history of chronic pain and analgesic use ● patients Decreased urinary concentrating ability, ● Chronic lithium nephropathy acidification defects, papillary necrosis, ster- CIN is the most common pathologic ile pyuria, low-grade proteinuria, hyperten- finding sion, and anemia ● Episodes of lithium intoxication and Characteristic findings on noncontrast com- lithium-induced NDI may predispose to puted tomography (CT) include papillary CIN development calcifications, decreased renal volume, and Treatment and outcome. bumpy renal contours ● Withdrawal of lithium may be associated ● Papillary necrosis can be seen on intrave- with gradual improvement in NDI and GFR nous pyelography ● May progress to end-stage renal failure, ● Increased risk for transitional cell cancers particularly if serum creatinine level Ն2.5 of uroepithelium mg/dL (Ն221 ␮mol/L) at time of diagnosis Treatment and outcome. ● While patients with focal segmental glomeru- ● No specific treatment, but if drug stopped losclerosis and serum creatinine Ͼ2.0 mg/dL early, there may be renal function stabiliza- (Ͼ177 ␮mol/L) at diagnosis can progress to tion or even improvement end-stage renal failure, patients with minimal change nephropathy experience complete re- ADDITIONAL READING mission upon drug withdrawal 1. Elseviers MM, DeSchepper A, Corthouts R, et al: High ● Amiloride blocks distal tubular reabsorp- diagnostic performance of CT scan for analgesic nephropa- thy in patients with incipient to severe renal failure. Kidney tion of lithium and attenuates NDI Int 48:1316-1323, 1995 2. McLaughlin JK, Lipworth L, Chow WH, Blot WJ: Analgesic use and chronic renal failure: A critical review ADDITIONAL READING of the epidemiologic literature. Kidney Int 54:679-686, 1. Braden GL: Lithium-induced renal disease, in Green- 1998 berg A (ed): Primer on Kidney Diseases (ed 2), chap 49. San 3. Ad Hoc Committee of the International Study Group on Diego, CA, Academic Press, 1998, pp 332-334 Analgesics and Nephropathy: Relationship between nonphen- 2. Presne C, Fakhouri F, Noel LH, et al: Lithium-induced acetin combined analgesics and nephropathy: A review. nephropathy: Rate of progression and prognostic factors. Kidney Int 58:2259-2264, 2000 Kidney Int 64:585-592, 2003 564 BRADEN, O’SHEA, AND MULHERN

Acyclic nucleoside inhibitors ● Patients receiving both tenofovir and ritona- Cidofovir, adefovir, and tenofovir have been vir should have frequent measurements of renal function, electrolytes, and phosphorus utilized to treat resistant cytomegalovirus infec- ● tion, hepatitis B, and human immunodeficiency Incomplete recovery after drug withdrawal virus, respectively. All 3 drugs are cleared at may occur from any of these agents rates greater than the GFR, indicating significant ADDITIONAL READING drug secretion by the proximal tubule. 1. Izzedine H, Launay-Vacher V, Isnard-Bagnis C, DeRay Pathogenesis. G: Drug-induced Fanconi’s syndrome. Am J Kidney Dis ● Drugs are transported into proximal tubules 41:292-309, 2003 by ornithine aminotransferase 1 trans- Chronic calcineurin inhibitor toxicity porter; cidofovir and adefovir induce proxi- Acute calcineurin inhibitor toxicity is associ- mal tubule cell (PTC) damage by causing ated with ARF, which is reversible upon dose mitochondrial damage reduction or cessation of therapy. Chronic cal- ● Mechanism of tenofovir toxicity is due to a cineurin inhibitor use is associated with CIN in drug interaction with ritonavir; tenofovir is renal transplant recipients and in patients with secreted into urine by multidrug resistance– autoimmune disorders treated with these drugs. associated protein 2 transporter, which is Pathogenesis. inhibited by ritonavir, leading to very high ● Chronic afferent arteriolar vasoconstriction PTC concentrations of tenofovir causes glomerular ischemia and scarring Unique pathologic features. ● Upregulation of renin angiotensin system, ● Adefovir and cidofovir cause acute tubu- transforming growth factor ␤, and osteopon- lar necrosis with enlarged proximal tu- tin stimulate interstitial fibrosis bule mitochondria, which are dysmorphic ● CIN occurs in 20% of patients with nonre- and have lost their cristae on electron nal transplants and can lead to end-stage microscopy renal disease (ESRD) ● Tenofovir induces PTC necrosis with en- ● Toxicity is more typically associated with larged dystrophic proximal epithelial cell chronic high-dose therapy, but can occur nuclei with low-dose therapy Clinical and laboratory features. Unique pathologic features. ● ● Cidofovir Characteristic findings include striped inter- Probenecid, which inhibits ornithine ami- stitial fibrosis in cortex and medulla, affer- notransferase 1 transporter, may mini- ent arteriolar hyalinosis, vacuolization of mize cidofovir nephrotoxicity tubular epithelium, and tubular atrophy ● Associated with glomerular thrombotic Proteinuria occurs in 12%, metabolic microangiopathy acidosis in 16%, and Fanconi syndrome Clinical and laboratory features. may occur ● Toxicity manifests as insidious develop- ● Adefovir ment of decrease in GFR and increased ARF and Fanconi syndrome occur with blood pressure Ͼ doses 60 mg/d ● Tubular abnormalities include metabolic Up to 10% of patients have CIN acidosis, hyperkalemia, hypercalciuria, ● Tenofovir hypophosphatemia, hyperuricemia, and Fanconi syndrome and ARF due to acute hypomagnesemia tubular necrosis occur between 1.5 weeks Treatment and outcome. and 2 years of therapy ● Reducing dose or stopping the drug alto- Treatment and outcome. gether may be beneficial ● Cidofovir should be administered with con- ● Replace calcineurin inhibitors with myco- comitant oral probenecid phenolate mofetil or sirolimus ● Patients receiving adefovir should have ● Angiotensin-converting enzyme (ACE) in- frequent determinations of renal function hibitors may lessen interstitial fibrosis CORE CURRICULUM IN NEPHROLOGY 565

● Fish oil, pentoxifylline, and calcium chan- have been shown to cause glutathione deple- nel blockers have not been shown to slow tion and lipid peroxidation progressive renal function loss Clinical and laboratory features. ● Associated with total ifosfamide dose, age, ADDITIONAL READING prior or concurrent treatment with cisplatin, 1. Burdmann EA, Andoh TF, Yu L, Bennett WM: and unilateral nephrectomy Cyclosporine nephrotoxicity. Semin Nephrol 23:465-476, ● 2003 Proximal tubular dysfunction leads to meta- 2. Schlitt HF, Barkmann A, Boker H, et al: Replacement bolic acidosis, hypophosphatemia, amino- of calcineurin inhibitors with mycophenolate mofetil in liver aciduria, and hypokalemia; severe renal transplant patients with renal dysfunction: A randomized failure also has been reported controlled study. Lancet 357:587-591, 2001 Treatment and outcome. 3. Ojo AO, Held PF, Port FK, et al: Chronic renal failure ● after transplantation of a non-renal organ. N Engl J Med Nephrotoxicity may be mild, acute, and 349:931-940, 2003 reversible or chronic and lead to long-term metabolic abnormalities and/or renal failure Aristolochic acid–associated nephropathy Pathogenesis. ADDITIONAL READING ● Substitution of Aristolochia fangchi for the 1. Skinner R, Cotterill SJ, Stevens MC: Risk factors for Chinese herb Stephania tetranda in pills nephrotoxicity after ifosfamide treatment in children: A UKCCSG Late Effects Group study. United Kingdom used for weight reduction exposed patients Children’s Cancer Study Group. Br J Cancer 82:1636-1645, to high doses of the nephrotoxic and carci- 2000 nogenic aristolochic acids Unique pathologic features. Toxins ● Extensive, hypocellular cortical interstitial fi- Lead nephropathy brosis, and upper tract urothelial tumors in up Pathogenesis. to 50% of patients with ESRD from this cause ● Early accumulation of filtered lead, particu-

Clinical and laboratory features. larly by S3 segment of proximal tubule, ● Initial presentation of anemia, tubular pro- likely leads to direct tubulotoxic effects and teinuria, and normotension in over half the subsequent interstitial fibrosis; subsequent patients hypertension and hyperuricemia also may Treatment and outcome. contribute to further renal compromise ● Prednisolone therapy in patients with mod- Unique pathologic findings. erate renal insufficiency (serum creatinine, ● Acid-fast intranuclear inclusions of PTCs 1.8-3.9 mg/dL [159-345 ␮mol/L]) may slow are characteristic of acute lead intoxication; rate of renal failure progression in chronic nephropathy, focal tubular atro- ● Left untreated, aristolochic acid–associated phy, interstitial fibrosis, and minimal cellu- nephropathy leads to rapid progression to lar infiltrates predominate ESRD in most patients Clinical and laboratory features. ● Decreased urate excretion, proximal tubu- ADDITIONAL READING lar dysfunction, and hyporeninemic hypoal- 1. Vanherweghem JL, Abramowicz D, Tielemans C, dosteronism are early renal manifestations Depierreux M: Effects of steroids on the progression of renal failure in chronic interstitial renal fibrosis: A pilot study in of lead intoxication; late findings of progres- Chinese herbs nephropathy. Am J Kidney Dis 27:209-215, sive renal failure, hypertension, and recur- 1996 rent episodes of gout (saturnine) are typical 2. Reginster F, Jadoul M, van Ypersele de Strihou C: ● Diagnosis of lead nephropathy dependent Chinese herbs nephropathy presentation, natural history and on recognition of patients with an appropri- fate after transplantation. Nephrol Dial Transplant 12:81-86, 1997 ate lead exposure history, chronic renal failure, hypertension, and gout (saturnine) Ifosfamide nephrotoxicity ● Because Ͼ90% of total body lead resides in Pathogenesis. bone, serum lead levels generally are un- ● May be related to the ifosfamide metabo- helpful in diagnosis of chronic lead expo- lites chloracetaldehyde or acrolein, which sure 566 BRADEN, O’SHEA, AND MULHERN

● Ethylenediaminetetraacetic acid (EDTA) ● Acute and chronic myeloma kidney results mobilization test with resultant urinary lead from light-chain toxicity; light chains are neph- excretion (in patients with renal insuffi- rotoxic due to either direct tubular toxicity or ciency, a 72-hour collection is necessary) intrarenal obstruction from cast formation greater than Ͼ600 ␮g is diagnostic of ● Multiple factors predispose patients with elevated total body lead burden multiple myeloma to renal disease: Treatment and outcome. Volume depletion ● Chelation therapy with EDTA leads to Hypercalcemia reversal of early tubular dysfunction, im- Hyperuricemia proves GFR in patients with mild to moder- Contrast media ate renal failure, and decreases frequency Other nephrotoxins of gouty flares; however, EDTA chelation is ● Characteristics thought to increase light- ineffective in reversing advanced renal fail- chain toxicity include: ure secondary to lead nephropathy Light-chain concentration and isoelectric point ADDITIONAL READING Acidic intraluminal pH 1. Lin JL, Lin-Tan DT, Hsu KH, Yu CC: Environmental Tubular flow rate lead exposure and progression of chronic renal diseases in Presence of intact Tamm-Horsfall pro- patients without diabetes. N Engl J Med 348:277-286, 2003 tein 2. Brewster UC, Perazella MM: A review of chronic lead intoxication: An unrecognized cause of chronic kidney Tubular concentration of calcium and disease. Am J Med Sci 327:341-347, 2004 sodium ● Acute myeloma kidney is ARF due to intratu- Cadmium bular myeloma light-chain deposition as tubu- Pathogenesis. lar casts; a more chronic process of tubular ● Renal toxicity is associated with prolonged obstruction occurs over months and years in low-level exposure, principally from contami- chronic myeloma kidney nated food, cigarettes, and workplace exposure Unique pathologic features. ● Cadmium is taken up by PTCs via pinocyto- ● Tubular casts surrounded by multinucleated sis; inside cell, complex is degraded by giant cells, interstitial infiltrates of plasma lysozymes and free cadmium causes cellu- cells, and mononuclear cells; chronic my- lar toxicity eloma kidney refers to aforementioned ab- Clinical and laboratory features. normalities, plus interstitial fibrosis and ● Risk for injury increases with age tubular atrophy ● Irreversible proximal tubular dysfunction, ● Renal amyloidosis occurs in some patients hypercalciuria, and nephrolithiasis with light-chain deposition in glomeruli ● May be associated with bone disease, lung leading to nephrotic-range proteinuria; char- injury, and cancer acteristic fibrillary changes seen on elec- Treatment and outcome. tron microscopy and Congo red staining ● Minimize exposure Clinical and laboratory features. ● No role for chelating agents ● Monoclonal light chains in serum and urine are found as M spikes on protein electro- ADDITIONAL READING phoresis or as ␬ or ␭ light chains on 1. Hellstrom L, Elinder CG, Dahlberg B, et al: Cadmium immunofixation studies exposure and end-stage disease. Am J Kidney Dis 38:1001- ● Urinalysis shows a bland sediment, normal 1008, 2001 kidney size, and Fanconi syndrome, or low Hematologic Neoplastic Diseases anion gap due to cationic immunoglobulin G (IgG) or IgM paraproteins, anemia, and Multiple myeloma hypercalcemia Pathogenesis. Treatment and outcome. ● Renal dysfunction occurs in Ͼ50% of pa- ● Volume repletion, correction of hypercalce- tients mia/hyperuricemia, discontinuation of neph- CORE CURRICULUM IN NEPHROLOGY 567

rotoxic medications, appropriate chemo- interleukin 2, interferon ␥, and other therapy, and dialysis as needed cytokines Unique pathologic features. ADDITIONAL READING ● Although noncaseating granulomatous inter- 1. Winearls CG: Acute myeloma kidney. Kidney Int stitial nephritis is classic lesion in sarcoid- 48:1347-1361, 1995 osis, it is uncommon finding 2. Pozzi C, D’Amico M, Fogazzi GB, et al: Light chain ● Renal lesions include mesangiocapillary deposition disease with renal involvement: Clinical charac- teristics and prognostic factors. Am J Kidney Dis 42:1154- and mesangial proliferative glomerulone- 1163, 2003 phritis, IgA nephropathy, membranous glomerulonephritis, and crescentic glo- Lymphoproliferative disorders merulonephritis Lymphomatous or leukemic cell infiltration of Clinical and laboratory features. the kidneys is commonplace in non-Hodgkin ● Hypercalcemia occurs in up to 20% of lymphoma and lymphocytic leukemias. How- cases of sarcoidosis, particularly in summer ever, clinically identifiable renal disease is rela- ● Calcium oxalate nephrolithiasis occurs in tively rare. up to 14% of patients with sarcoidosis Pathogenesis. Treatment and outcomes. ● ARF may develop as a result of rapid ● Corticosteroid therapy inhibits macrophage increases in interstitial pressure from cell activity and suppresses calcitriol synthesis infiltration; chronically, tubular atrophy and ● Nephrocalcinosis may be responsible for necrosis predominate CIN in up to 50% of patients Unique pathologic features. ● Corticosteroid treatment for up to 6 months ● Diffuse lymphocytic infiltration of the inter- leads to improved renal function stitium with dense monomorphic lymphoid ● Incomplete renal recovery often occurs due cells with preserved glomerular architec- to irreversible nephrosclerosis ture occurs ADDITIONAL READING Clinical and laboratory features. ● ARF is associated with non–nephrotic- 1. Gobel U, Kettritz R, Schneider W, Luft FC: The protean face of renal sarcoidosis. J Am Soc Nephrol 12:616-623, range proteinuria and bilateral enlarged, 2001 nodular kidneys noted on imaging Treatment and outcomes. Primary Sjögren syndrome ● Treatment with systemic chemotherapy Sjögren syndrome is a disease with lympho- and/or radiation therapy leads to rapid renal cytic infiltration of the epithelial ducts of sali- function improvement and decrease in kid- vary and lacrimal glands. It accompanies B-cell ney size; prognosis dependent on response hyperactivity with antinuclear antibodies and cir- of malignancy to treatment culation immune complexes. Pathogenesis. ADDITIONAL READING ● Unknown 1. Tornroth T, Heiro M, Marcussen N, Franssila K: Unique pathologic features. Lymphomas diagnosed by percutaneous renal biopsy. Am J ● Most patients have CIN with predominance Kidney Dis 42:960-971, 2003 of T lymphocytes and, to a lesser degree, B Immune Disorders cells, monocytes, and plasma cells ● Nephrotic syndrome can occur, with most Sarcoidosis common glomerular lesion being mem- Pathogenesis. branoproliferative glomerulonephritis, mes- ● Increased production of 1-␣ hydroxylase angial proliferative glomerulonephritis, and, from activated mononuclear cells leads to rarely, membranous nephropathy

increased 1,25-dihydroxyvitamin D3 levels Clinical and laboratory features. and enhanced intestinal calcium absorption ● Distal RTA is most common RTA and ● Tissue infiltration with activated CD4–T- occurs in up to 5% lymphocytes of the T-helper 1 type produce ● NDI occurs in up to 13% 568 BRADEN, O’SHEA, AND MULHERN

● Renal potassium wasting with severe ● Uveitis often requires systemic corticoste- hypokalemia roids and often has relapsing course Treatment and outcome. ● CIN can improve with corticosteroids if ADDITIONAL READING started early; RTA rarely responds to corti- 1. Takemura T, Okada M, Hino S, et al: Course and costeroid therapy outcome of tubulointerstitial nephritis and uveitis syndrome. ● CIN occurs early within first 2 to 4 years Am J Kidney Dis 34:1016-1021, 1999 ● Glomerulonephritis develops in patients af- ter 8 to 10 years; value of immunosuppres- Metabolic Disorders sive therapy for glomerular lesions is uncertain Hypokalemic nephropathy Hypokalemia can be associated with func- ADDITIONAL READING tional renal disturbances, particularly NDI, as 1. Bossini N, Savoldi S, Franceschini F, et al: Clinical and well as renal cyst formation and irreversible morphological features of kidney involvement in primary Sjogren’s syndrome. Nephrol Dial Transplant 16:2328- CIN. 2336, 2001 Pathogenesis. ● Major cause of NDI is tubular resistance to TIN with uveitis ADH due to impaired generation of cyclic This disorder presents in adolescence and adenosine monophosphate from adenylate young adults, particularly females, often as ARF. cyclase, impaired ADH- and cyclic adeno- The uveitis can develop prior to, concurrently, or sine monophosphate–mediated water flow, after the TIN. and downregulation of aquaporin-2 water Pathogenesis. channels in cortex and medulla ● Peripheral blood shows increased numbers ● Increased ammoniagenesis from potassium of B cells without abnormalities in T cells depletion may induce renal tubular injury ● Associated with Epstein-Barr virus, antineu- by interstitial complement activity trophil cytoplasmic antibody, and chlamydia ● Hypokalemia can stimulate insulin-like Unique pathologic features. growth factor 1 and transforming growth ● Renal tissue has a predominance of CD4 T factor ␤, leading to chemotaxis of inflamma- lymphocytes, CD8 T lymphocytes, and tory cells and fibrosis monocytes and macrophages Unique pathologic features. Clinical and laboratory features. ● Any disorder producing chronic hypokale- ● Often presents with signs of fever, anemia, mia may be associated with proximal tubu- and asthenia lar lesion, interstitial fibrosis, tubular atro- ● Uveitis of anterior chamber is most com- phy, and medullary cysts mon Clinical and laboratory features. ● Blood testing includes peripheral eosino- ● NDI occurs with serum potassium Ͻ3.0 philia, anemia, and elevated erythrocyte mEq/L (mmol/L) sedimentation rate; serologic testing for Treatment and outcome. systemic immunologic disease, such as sar- ● Morphological changes of chronic hypoka- coidosis, Sjögren syndrome, Wegener lemia are reversible within first few months granulomatosis, Behçet disease, as well as of potassium repletion, but irreversible CIN infectious diseases, are negative can occur ● May be associated with Fanconi syndrome, ● Renal cysts can decrease after resection of distal RTA, and NDI adrenal adenoma or potassium therapy in ● In adolescents and young adults, renal primary hyperaldosteronism disease spontaneously remits over 1 year without corticosteroid therapy ADDITIONAL READING Treatment and outcome. 1. Torres VE, Young WF Jr, Offord KP, Hattery RR: ● In adults, corticosteroid therapy associated Association of hypokalemia, aldosteronism and renal cysts. with improved renal function N Engl J Med 322:345-351, 1990 CORE CURRICULUM IN NEPHROLOGY 569

Hypercalcemic nephropathy Clinical and laboratory features. ● Hypercalcemia is associated with NDI, RTA, Acute urate nephropathy presents with kidney stones, ARF, and CIN. abrupt oliguria or anuria, and an elevated Pathogenesis. uric acid (typically Ͼ15 mg/dL [Ͼ892 ␮ ● NDI is due to decreased medullary solute mol/L]) and a urinary uric acid to creati- Ͼ gradient and predominantly due to im- nine ratio 1 ● paired hydro-osmotic effect of ADH Chronic urate nephropathy presents with Unique pathologic features. hypertension, mild renal dysfunction, mild ● Chronic hypercalcemia leads to interstitial proteinuria, decreased urinary concentrat- calcification, tubular cell necrosis, tubular ing ability, and bland urine sediment atrophy, and interstitial fibrosis, predomi- Treatment and outcome. ● nantly in the medulla Allopurinol is used to lower serum uric acid ● Nephrocalcinosis often is present on plain to prevent acute urate nephropathy, but its film, but CT is more sensitive efficacy in slowing progressive chronic Clinical and laboratory features. urate nephropathy is unproven ● NDI occurs in up to 20% of patients with chronic hypercalcemia ADDITIONAL READING ● Large increases in serum calcium Ͼ12 1. Johnson RJ, Kivlighn SD, Kim YG, Suga S, Fogo AB: Ͼ Reappraisal of the pathogenesis and consequences of hyper- mg/dL ( 2.99 mmol/L) can cause ARF due uricemia in hypertension, cardiovascular disease, and renal to renal arterial vasoconstriction and vol- disease. Am J Kidney Dis 33:225-234, 1999 ume contraction from natriuresis 2. Nickeleit V, Mihatsch MJ: Uric acid nephropathy and ● CIN is associated with polyuria, salt wast- end-stage renal disease—review of a non-disease. Nephrol ing, calcium oxalate stones, and distal RTA Dial Transplant 12:1832-1838, 1997 ● Most patients with CIN have chronic Cystinosis hypercalcemia Treatment and outcome. Pathogenesis. ● ● Early correction of hypercalcemia may lead Autosomal recessive disorder, 1 per to recovery of renal function or slower 100,000-200,000 ● progression of CIN Caused by mutation of the CTNS gene on chromosome 17p13, which encodes protein cystinosin ADDITIONAL READING ● 1. Braden GL, Singer I, Cox M: Nephrogenic diabetes Abnormal cystinosin impairs cystine trans- insipidus, in Gonick HC, Buckalew VM (eds): Renal port from lysosomes Tubular Disorders. New York, NY, Marcel Dekker, 1985, pp Unique pathologic features. 431-494 ● Hexagonal birefringent cystine crystals on polarized microscopy are present in urine, Urate nephropathy cornea, liver, spleen, lymph nodes, kidneys, Pathogenesis. thyroid, intestines, brain, and bone marrow ● Acute urate nephropathy is typically seen in ● Renal tubules have swan-neck deformity in setting of tumor lysis syndrome proximal renal tubule and later develop ● Although chronic toxicity from uric acid is CIN controversial, mechanism is thought to in- Clinical and laboratory features. volve deposition of urate crystals in medul- ● Diagnosis is made by measuring cystine lary interstitium with consequent secondary content in peripheral blood leukocytes chronic inflammatory response, leading to ● Fanconi syndrome develops between 6-12 interstitial fibrosis months of age and is associated with hy- Unique pathologic features. pophosphatemic rickets and polyuria due to ● Birefringent uric acid crystal deposition in obligate solute excretion tubules and interstitium Treatment and outcome. ● Occasionally, medullary renal tophi are ● Cysteamine binds to cystine in lysosomes found on gross anatomic dissection and transports it through a lysine trans- 570 BRADEN, O’SHEA, AND MULHERN

porter; 4 oral doses per day are given along 2. Hoopes RR, Hueber PA, Reid RJ, et al: CLCN5 with eye drops to prevent corneal blindness chloride-channel mutations in six new North American ● Renal transplantation is therapy of choice, families with X-linked nephrolithiasis. Kidney Int 54:698- 705, 1998 but extrarenal manifestations of cystinosis require continued cysteamine therapy Primary hyperoxaluria ADDITIONAL READING Pathogenesis. 1. Gahl WA, Thoene JG, Schneider JA: Cystinosis. ● 2 forms: primary hyperoxaluria types 1 and N Engl J Med 347:111-121, 2002 2 (PH1 and PH2) ● PH1 is more common and due to deficiency Dent disease of the hepatic peroxisomal enzyme alanine X-Linked recessive disorder of the proximal glyoxylate aminotransferase (AGT), which tubule characterized by Fanconi syndrome, kid- leads to increased urinary oxalate and glyox- ney stones, nephrocalcinosis, rickets, and progres- alate; mutations of the AGT genes on sive renal insufficiency. chromosome 2 Q36-37 lead to decreased Pathogenesis. function of AGT ● Originally named X-linked recessive neph- ● PH2 is due to deficiency of the liver rolithiasis cytosolic enzyme hydroxypyruvate reduc- ● Caused by mutations in CLC5 channel tase, which leads to increased urinary ox- protein gene at Xp11.22 alate and glycerate excretion ● Female carriers rarely develop manifesta- ● PH1 can be associated with severe calcium tions of the disease oxalate deposition in kidney interstitium ● Proximal tubular endosomal function is with nephrocalcinosis and, once GFR Ͻ25 inhibited, leading to Fanconi syndrome mL/min (Ͻ0.42 mL/s), there is diffuse Unique pathologic features. systemic oxalate deposition ● Nephrocalcinosis occurs at young age in Unique pathologic features. 75% of patients ● Early medullary nephrocalcinosis progress- Clinical and laboratory features. ing to diffuse nephrocalcinosis ● Diagnosis can be made by gene testing for ● Extensive calcium oxalate deposition on defect on chromosome Xp11.22 tissue biopsy ● ␤ Increased excretion of 2 microglobulin Clinical and laboratory features. and retinal 2–binding protein in carriers and ● Nephrolithiasis usually occurs before age patients ● of 5 years, but adults can present with either Hypophosphatemic rickets occurs in 25% stones or progressive renal insufficiency of males ● from nephrocalcinosis Hypercalciuria occurs at early age with ● Once GFR Ͻ25 mL/min (Ͻ0.42 mL/s), kidney stones and nephrocalcinosis ● cardiac conduction defects, distal gangrene, CIN with nephrocalcinosis occurs in two arthropathy, and retinal macular disease thirds of affected males, leading to end- develop stage renal failure between ages 30-40 ● PH1 diagnosed by increased urinary ox- years alate and glyoxalate and by demonstration Treatment and outcome. ● on liver biopsies of decreased AGT activity Hypercalciuria can improve with low- ● PH2 diagnosed by increased urinary ox- sodium diet and thiazide diuretics ● alate and glycerate and decreased hydroxy- Oral phosphate and carefully dosed vitamin pyruvate reductase on liver biopsy D can improve bone disease Treatment and outcome. ● Renal transplantation is definitive therapy ● General measures to decrease urinary super- ADDITIONAL READING saturation include increased fluid intake, 1. Scheinman SJ: X-Linked hypercalciuric nephrolithia- pyridoxine (3-7 mg/kg/d), orthophosphate sis: Clinical syndromes and chloride channel mutations. (30-40 mg/kd/d), and citrate (3-4 mEq/d) Kidney Int 53:3-17, 1998 and magnesium (400-1600 mg/d) CORE CURRICULUM IN NEPHROLOGY 571

● High-dose orthophosphate and pyridoxine Radiation Nephritis have been shown to preserve renal function Pathogenesis and decrease nephrolithiasis ● Ͼ ● Combined liver and renal transplants are In patients receiving 1,500-2,500 rads to required once GFR Ͻ20 mL/min (Ͻ0.33 kidney, there is endothelial cell injury and swelling, with eventual vascular occlusion mL/s) to decrease systemic oxalate and and chronic ischemic injury renal oxalate deposition ● Direct tubular epithelial cell injury occurs ● Orthophosphate, citrate, and magnesium from the radiation needed in posttransplantation period to ● Certain chemotherapy also may potentiate lessen systemic and renal oxalosis effects of radiation on kidney

ADDITIONAL READING Unique pathologic features 1. Milliner DS, Wilson DM, Smith LH: Phenotypic ● Glomerular capillary endothelial injury with expression of primary hyperoxaluria: Comparative features of types I and II. Kidney Int 59:31-36, 2001 swelling and basement membrane splitting 2. Millan MT, Berquist WE, So SK, et al: One hundred and occasionally thrombotic microangiopa- percent patient and kidney allograft survival with simulta- thy, especially in children neous liver and kidney transplantation in infants with primary hyperoxaluria: A single-center experience. Trans- Clinical and laboratory features plantation 78:1458-1463, 2003 ● Acute radiation nephritis occurs 6-12 Balkan endemic nephropathy months after exposure, characterized by progressive renal insufficiency accompa- Pathogenesis. ● nied by proteinuria, accelerated renin- Although no specific causative agent has dependent hypertension, edema, and occa- been identified, most data support environ- sional intravascular hemolysis mental factors as leading cause of Balkan ● Onset of chronic radiation nephritis occurs nephropathy more than 18 months after exposure and is Unique pathologic features. characterized by proteinuria, progressive ● No characteristic renal pathology has been renal insufficiency, and hypertension identified ● Chronic radiation damage may present years Clinical and laboratory features. later with significant proteinuria with pre- ● Slowly progressive TIN served renal function or with just hyperten- ● Normotension predominates and anemia sion and mild proteinuria out of proportion to renal failure is common- place Treatment and outcome ● Urothelial tumors occur up to 100 times ● Prevention is only specific measure, usually more frequently in endemic areas and can with kidney shielding and/or a fractionated be bilateral in up to 14% of affected patients radiation dose Treatment and outcome. ● Aggressive treatment of hypertension and ● Specific therapy is lacking the use of ACE inhibition also may be ● Balkan endemic nephropathy accounts for helpful up to 10% of all causes of ESRD in some Balkan regions ADDITIONAL READING 1. Cassady JR: Clinical radiation nephropathy. Int J ADDITIONAL READING Radiat Oncol Biol Phys 31:1249-1256, 1995 1. Stefanovic V, Polenakovic MH: Balkan nephropathy. Papillary Necrosis Kidney disease beyond the Balkans? Am J Nephrol 11:1-11, 1991 Pathogenesis 2. Petronic VJ, Bukurov NS, Djokic MR, et al: Balkan ● endemic nephropathy and papillary transitional cell tumors Conditions that compromise papillary blood of the renal pelvis and ureters. Kidney Int Suppl 34:S77- flow, either structurally or hormonally, can S79, 1991 result in ischemic necrosis (eg, diabetes 572 BRADEN, O’SHEA, AND MULHERN

mellitus, NSAIDs, urinary tract obstruc- Pathogenesis tion, and sickle cell disease) ● Acute interstitial nephritis is associated ● Nephrotoxic agents, such a analgesics, can with aminosalicylic acid (ASA) therapy be heavily concentrated in papilla, increas- ● Enteric hyperoxaluria is possible cause of ing their toxicity CIN, but rarely described ● More than 1 clinical condition predisposing ● CIN can occur in Crohn disease without to papillary necrosis is present prior exposure to salicylates Unique pathologic features ● ASA-induced CIN occurs at expected fre- quency of 1/500 patients ● Coagulative necrosis in inner medulla and papilla is characteristic Unique pathologic features ● Overlying cortical changes of CIN can coexist with papillary necrosis ● CIN without hypokalemic changes or cal- cium oxalate deposition Clinical and laboratory features ● Proteinuria and sterile pyuria are common; Clinical and laboratory features gross or microscopic hematuria can be ● Most patients with CIN have recurrent seen, particularly with sloughing of papil- episodes of ARF associated with volume lae depletion and prerenal azotemia and occa- ● Polyuria and nocturia are early findings ● sionally acute interstitial nephritis or acute Sloughed papillae can cause ureteral colic tubular necrosis or serve as a nidus for infection ● Ͻ ● Proteinuria usually 2.0 g/mg and urinaly- Diagnosis can be made by finding sloughed ses are bland without casts tissue in urine or radiographically with intravenous pyelography or retrograde stud- Treatment and outcome ies; ultrasound and CT are less sensitive ● CIN from ASA may stabilize and improve Treatment and outcome upon ASA withdrawal ● Course can be variable, ranging from ● Most cases with CIN are unrelated to ASA asymptomatic disease to recurrent episodes and progress to end-stage renal failure of urinary tract infection, renal colic, and requiring dialysis and transplantation progressive renal insufficiency ● Variety of forms of glomerulonephritis, ● Although no specific treatment exists, under- particularly membranoproliferative glomer- lying condition or risk factor should be ulonephritis and amyloid, may occur in addressed; blood pressure control and ACE inflammatory bowel disease inhibition may be helpful ADDITIONAL READING ADDITIONAL READING 1. De Broe ME, Stolear JC, Nouwen EJ, Elseviers MM: 1. Griffin MD, Bergstralhn EJ, Larson TS: Renal papillary 5-Aminosalicylic acid (5-ASA) and chronic tubulointersti- necrosis—A sixteen-year clinical experience. J Am Soc tial nephritis in patients with chronic inflammatory bowel Nephrol 6:248-256, 1995 disease: Is there a link? Nephrol Dial Transplant 12:1839- Inflammatory Bowel Disease 1841, 1997 2. Pardi DS, Tremaine WJ, Sandborn WJ, McCarthy JT: Acute and chronic interstitial nephritis may Renal and urologic complications of inflammatory bowel occur in inflammatory bowel disease. disease. Am J Gastroenterol 93:504-514, 1998 KIDNEY BIOPSY TEACHING CASE Membranoproliferative Injury Pattern in a Renal Allograft

Neriman Gökden, MD, Michele Rossini, MD, Jayant Kumar, MD, and Agnes B. Fogo, MD

INDEX WORDS: Membranoproliferative glomerulonephritis; acute transplant glomerulitis; acute vascular rejection.

EMBRANOPROLIFERATIVE glomeru- workup, she was noted to have persistently elevated protein- M lonephritis (MPGN), caused by subendo- uria on dipstick (protein, 300 mg/dL). Spot urine protein- thelial immune-complex deposits, can occur as a creatinine ratio was 1.95. Proteinuria persisted despite treat- ment of an episode of urinary tract infection with an de novo or recurrent disease in the transplanta- appropriate antibiotic. She underwent a diagnostic transplant tion setting. When an MPGN-like injury pattern kidney biopsy. with endocapillary proliferation and double con- tours of the capillary wall is seen by means of Renal Biopsy light microscopy in the allograft biopsy, the The biopsy specimen consisted of 2 pieces of cortex and differential diagnosis includes recurrent MPGN, medulla containing 15 glomeruli, 1 of which was globally de novo MPGN, acute transplant glomerulitis, sclerosed. There was diffuse glomerulitis with endocapillary transplant glomerulopathy, and chronic throm- proliferation and many mononuclear cells and rare polymor- botic microangiopathy, among others. Careful phonuclear leukocytes in the capillary lumens (Fig 1). Glo- meruli showed an increase in mesangial matrix and cellular- integration of immunofluorescence and electron ity without mesangiolysis. Podocytes were unremarkable. microscopic findings is essential for accurate Glomerular basement membranes did not have spikes, holes, diagnosis. We present the case of a renal trans- or corrugation, but there was frequent splitting. There were plant recipient with MPGN-like injury pattern no crescents, adhesions, or fibrin thrombi in Bowman space. detected by means of light microscopy. With There was mild interstitial fibrosis involving 10% of the biopsy tissue, with proportional tubular atrophy. There was careful correlation of immunofluorescence and minimal lymphocytic infiltrate involving 3 tubules/high- electron microscopic findings, a specific diagno- power field (Fig 2). There were several interlobular arteries sis was made and appropriate treatment was showing activated endothelium, 1 showing adherent lympho- rendered. cyte and 1 showing endothelialitis, diagnostic of acute vascular rejection Collaborative Clinical Trials in Transplan- 1 CASE REPORT tation (CCTT) type 2 (Fig 3). Immunofluorescence microscopy showed trace mesangial Clinical History staining for immunoglobulin M and trace mesangial and trace arteriolar staining for C3. Immunoglobulin G, immuno- A 59-year-old African-American woman presented for globulin A, C1q, and polyvalent antisera were negative. C4d cadaveric kidney transplantation. Her past medical history staining showed diffuse 2ϩ peritubular capillary staining was significant for polycystic kidney disease diagnosed in (Fig 4). The control worked appropriately. 1986, type 2 diabetes mellitus for 12 years, and hyperten- sion. She also had hepatic cysts associated with polycystic Electron microscopy showed endocapillary proliferation kidney disease and chronic cholecystitis. She had been on with polymorphonuclear leukocytes and mononuclear cells dialysis therapy for 6 years. On admission, blood urea in the capillary lumen and swelling of endothelial cells, but nitrogen level was 36 mg/dL (12.9 mmol/L), and serum no reticular aggregates in endothelial cells. The glomerular creatinine level was 1.9 mg/dL (168 ␮mol/L). Other labora- tory tests showed normal sodium and potassium levels, with a glucose level of 186 mg/dL (10.3 mmol/L). From the Department of Pathology, Vanderbilt University The patient underwent successful cadaveric transplanta- Medical Center, Nashville, TN; Department of Pathology, tion, along with bilateral native nephrectomies and cholecys- and Department of Medicine, Division of Nephrology, Uni- tectomy. Immunosuppression consisted of prednisone, myco- versity of Arkansas for Medical Sciences, Little Rock, AR. phenolate mofetil, FK506, and induction with basiliximab Received February 15, 2005; accepted as submitted pretransplantation and on day 4. Her postoperative course March 18, 2005. was complicated by delayed graft function, with urine out- Originally published online as doi:10.1053/j.ajkd.2005.03.026 put of 1 to 2 L/d and slowly improving creatinine levels. She on August 2, 2005. required 3 sessions of hemofiltration on days 3, 4, and 5 for Address reprint requests to Agnes B. Fogo, MD, Depart- volume overload. Serum creatinine level improved to 2.3 ment of Pathology, C3310, Medical Center North, Vander- mg/dL (203 ␮mol/L) by day 14 without hemodialysis. bilt University Medical Center, Nashville, TN 37232-2561. The patient was maintained on prednisone, mycopheno- E-mail: [email protected] late mofetil, and FK506 therapy as an outpatient. Her serum © 2005 by the National Kidney Foundation, Inc. creatinine level stabilized at 1.8 mg/dL (159 ␮mol/L) at 6 0272-6386/05/4603-0023$30.00/0 weeks posttransplantation. During a routine laboratory doi:10.1053/j.ajkd.2005.03.026

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: pp 573-576 573 574 GÖKDEN ET AL

Fig 1. A glomerulus shows global endocapillary proliferation with numerous mononuclear cells in the capillary lumens (Jones silver stain; original magnifi- .(cation ؋200

basement membrane showed normal thickness. There were no subendothelial or subepithelial deposits. There was 50% to 75% foot-process effacement. Visceral epithelial cells showed cytoplasmic vacuolization. The mesangium showed normal matrix and cellularity, no mesangial immune- complex deposits, and no mesangial interposition. There was interstitial edema and infiltrate. Tubules were unremark- able. Final Diagnosis The final diagnosis was acute rejection, CCTT type 2, with glomerulitis and C4d peritubular capillary staining, Fig 3. (A) An arteriole shows activated endothelium suggestive of a possible antibody-mediated component.1 with an adherent lymphocyte (Jones silver stain; origi- nal magnification ؋1,000). (B) An interlobular artery Clinical Follow-Up shows 2 lymphocytes underneath the endothelial layer (ie, endothelialitis) (Jones silver stain; original magni- .(On the day of the biopsy, the patient was treated with 1 fication ؋400 dose of intravenous methylprednisolone sodium succinate, 250 mg. After biopsy results were known, she was treated with 7 daily doses of antithymocyte globulin (rabbit), 1.5 DISCUSSION mg/kg/d, in addition to intravenous ganciclovir. She toler- ated the therapy well and was discharged home on therapy Our case represents a patient 6 weeks posttrans- with prednisone, FK506, and mycophenolate mofetil, with plantation with endothelialitis and diffuse acute oral valganciclovir for cytomegalovirus prophylaxis. Her transplant glomerulitis characterized by an in- serum creatinine level has improved to 1.0 mg/dL (88 creased number of mononuclear cells and rare ␮ mol/L) and proteinuria has decreased, with a spot urinary polymorphonuclear leukocytes in glomerular cap- protein-creatinine ratio less than 0.25.

Fig 2. A tubular profile shows 2 lymphocytes within Fig 4. Immunofluorescence for C4d shows diffuse the tubular epithelium (ie, tubulitis) (Jones silver stain; 2؉ positivity along the peritubular capillaries (anti-C4d .(original magnification ؋400). antibody; original magnification ؋400 MPGN PATTERN IN ALLOGRAFT 575 illary lumina, resulting in diminished capillary and endothelial cells appear activated. The infil- patency. In addition, there was endocapillary trating mononuclear cells also have an activated proliferation, increase in mesangial matrix, and appearance. Crescents are not found, and thrombi cellularity with diffuse splitting of glomerular are rare.7 Acute transplant glomerulitis can be basement membranes reminiscent of MPGN. The scored according to criteria of the Banff system. initial differential diagnosis based on light micros- All glomeruli are affected in a global manner in copy in this patient included acute transplant the severe form (G3), but the lesion can be glomerulitis or de novo MPGN, in addition to the segmental and focal (G1). G2 is defined as seg- acute rejection. Immunofluorescence micros- mental or global glomerulitis in 25% to 75% of copy did not show evidence of immune complex– glomeruli.14 mediated glomerulonephritis, excluding MPGN. The evolution of morphologically distinct Immunofluorescence for C4d was strongly posi- changes in acute transplant glomerulitis has been tive in peritubular capillaries, which, along with described. The earliest stage is called evolving the acute glomerulitis, suggests an antibody- glomerulitis and is recognizable by swelling of mediated component of the acute rejection.2 endothelial and mesangial cells. The intermedi- Acute transplant glomerulitis has been vari- ate stage is characterized by enlarged glomeruli ously called transplant glomerulitis,3 endocapil- with lobular simplification, spongy matrix, and lary glomerulitis,4 acute allograft glomerulopa- glomerular basement membrane deformities. In thy,5 and early allograft glomerulitis.6 The the advanced stage, the glomerular basement incidence is reported to be 4% to 10% of renal membrane shows reduplication and interposition allograft biopsies. In approximately 2% to 4% of with monocytes.15 Thus, our patient showed pre- biopsies, a severe diffuse form of glomerular dominantly advanced stage lesions. injury is evident and dominates the histological The importance of acute transplant glomeruli- pattern.7 The first detailed description of acute tis in the prognosis of renal allograft outcome is glomerulitis was made in 1981 by Richardson et controversial. An association with poor graft al,8 who found an association with cytomegalovi- survival was reported in earlier series.5,10,15 There rus viremia. However, several other studies could was some correlation with conventional type 1 not confirm this correlation between glomerulitis acute rejection, but 40% of all biopsy specimens and active cytomegalovirus infection.9-12 Glomer- with glomerulitis showed no rejection by stan- ular endothelial cell hypercellularity has been dard criteria, and 53% of all biopsy specimens described at an early stage after transplantation with rejection showed no glomerulitis.4 In a in the early renal allograft7,8 and has been pro- retrospective cohort study, acute transplant glo- posed as a particular pattern of glomerular rejec- merulitis was associated significantly with a trend tion.9 However, the cause of acute transplant toward earlier rejection episodes, greater serum glomerulitis remains uncertain. creatinine level at the time of the index biopsy, Acute transplant glomerulitis is characterized greater prevalence of vascular rejection, and less morphologically by enlargement of the glomeru- improvement in mean reciprocal serum creati- lar tuft, with swollen endothelial cells and accu- nine level at 1 to 2 weeks after biopsy, but was mulation of mononuclear cells in glomerular not an independent predictor of graft survival.15 capillaries, resulting in closure of the capillary The association of acute transplant glomeruli- lumina. A significant increase in the relative tis with a greater prevalence of variable degrees number of mesangial cells also has been reported of endothelialitis/“vascular rejection” has been in acute transplant glomerulitis.13 Visceral epithe- documented.14,15 In our case, acute transplant lial cell proliferation is observed rarely, and glomerulitis was associated with acute “vascu- when present, was only mild. Polymorpho- lar” rejection (CCTT type 2) and also strong C4d nuclear cells in glomeruli in acute transplant peritubular staining, a finding linked to humoral glomerulitis are rare. A fine web of periodic mechanisms of rejection. Conversely, there was acid–Schiff–positive fibrils has been described in only minimal lymphocytic infiltrate involving the capillary lumina between intracapillary cells less than 5% of biopsy tissue, with occasional and in areas of mesangiolysis. Occasional endo- tubulitis, not diagnostic of acute “cellular” rejec- thelial cell necrosis in glomeruli can be found, tion (CCTT type 1). However, earlier studies 576 GÖKDEN ET AL showed that acute transplant glomerulitis is not 4. Olsen S, Spencer E, Cockfield S, Marcussen N, Solez an independent predictor of graft survival. This K: Endocapillary glomerulitis in the renal allograft. Trans- graft lesion is not included in the Banff or CCTT plantation 59:1421-1425, 1995 5. Tuazon TV, Schneeberger EE, Bhan AK, et al: Mono- classification systems as a criterion for the pres- nuclear cells in acute transplant glomerulopathy. Am J ence or grading of rejection. Pathol 129:119-132, 1987 In summary, acute transplant glomerulitis may 6. Racusen LC, Solez K, Colvin RB, et al: The Banff 97 be a peculiar pattern of rejection with a pathogen- working classification of renal allograft pathology. Kidney esis different from that of conventional rejection, Int 55:713-723, 1999 7. Colvin RB: Renal transplant pathology, in Jennette JC, but present data do not show independent ad- Olson JL, Schwartz MM, Silva FG (eds): Heptinstall’s 4,11 verse effects on graft function or prognosis. Pathology of the Kidney, vol 2 (ed 5). Philadelphia, PA, Differentiation from other causes of MPGN-like Lippincott-Raven, 1998, pp 1417-1418 pattern injury is important. This case illustrates 8. Richardson WP, Colvin RB, Cheeseman SH, et al: the utility of immunofluorescence and electron Glomerulopathy associated with cytomegalovirus viremia in renal allografts. N Engl J Med 305:57-63, 1981 microscopic study for workup in transplant recipi- 9. Axelsen RA, Seymour AE, Mathew TH, Canny A, ents with unusual light microscopic findings. The Pascoe V: Glomerular transplant rejection: A distinctive association between vascular rejection and acute pattern of early graft damage. Clin Nephrol 23:1-11, 1985 transplant glomerulitis should direct a very care- 10. Herrera AG, Alexander RV, Cooley CF, et al: Cyto- ful examination of sampled arteries in biopsy megalovirus glomerulopathy: A controversial lesion. Kidney Int 29:725-733, 1986 specimens with glomerulitis to ensure that endo- 11. Messias NC, Eustace JA, Zachary AA, Tucker PC, thelialitis, which may be focal, is detected when Charney D, Racusen LC: Cohort study of the prognostic present and timely appropriate aggressive therapy significance of acute transplant glomerulitis in acutely reject- is initiated. ing renal allografts. Transplantation 72:655-660, 2001 12. Boyce N, Hayes K, Gee D, et al: Cytomegalovirus infection complicating renal transplantation and its relation- REFERENCES ship to acute transplant glomerulopathy. Transplantation 1. Colvin RB, Cohen AH, Saiontz C, et al: Evaluation of 45:706-709, 1988 pathologic criteria for acute renal allograft rejection: Repro- 13. Hanberg FS: A quantitative study of the renal cor- ducibility, sensitivity, and clinical correlation. J Am Soc puscles in acute renal allograft rejection. Acta Pathol Micro- Nephrol 8:1930-1941, 1997 biol Scand (A) 85:367-372, 1977 2. Racusen LC, Colvin RB, Solez K, et al: Antibody- 14. Solez K, Axelsen R, Benediktsson H, et al: Interna- mediated rejection criteria—An addition to the Banff 97 tional standardization of criteria for the histologic diagnosis classification of renal allograft rejection. Am J Transplant of renal allograft rejection: The Banff working classification 3:708-714, 2003 of kidney transplant pathology. Kidney Int 44:411-422, 1993 3. Marcussen N, Solez K, Spencer E, Cockfield S, Olsen 15. Maryniak RK, First MR, Weiss MA: Transplant glo- S: Early transplant glomerulitis; Glomerular size and ultra- merulopathy: Evolution of morphologically distinct changes. structure. Transplant Proc 28:468-469, 1996 Kidney Int 27:799-806, 1985 QUIZ PAGE ANSWERS AJKD SEPTEMBER 2005 A 44-year-old African-American man with a history of hypertension presented with kidney failure. He had been on a crack cocaine binge 3 days before admission. Physical examination findings were unremarkable. His blood urea nitrogen level was 46 mg/dL (16 mmol/L), and serum creatinine level was 9.9 mg/dL (875 ␮mol/L). Serum sodium level was 138 mEq/L (mmol/L); potassium, 4.3 mEq/L (mmol/L); chloride, 103 mEq/L (mmol/L); and total carbon dioxide, 17 mEq/L (mmol/L). Urinalysis showed 100 mg/dL (0.1 g/L) of protein, 100 mg/dL (5.6 mmol/L) of glucose, 5 white blood cells/high-power field, and 1 to 4 red blood cells/high-power field. Serological workup was negative. Renal ultrasound showed no evidence of obstruction. Serum and urine toxicology screens were positive for cocaine only. A kidney biopsy was performed.

Figure 43A. Low-power view of a glomerulus, tubules, and interstitium. Note crystals occluding the tubular lumen. QUIZ PAGE ANSWERS AJKD (continued)

Figure 43B. High-power view of a tubule with calcium oxalate crystals.

Figure 43C. High-power view of a tubule under polarized light showing calcium oxalate crystals.

What is your diagnosis? After the biopsy was performed, the patient admitted to ingesting 3 small cups of antifreeze during his crack cocaine binge 3 days before admission. Toxicology screens were negative because the ethylene glycol had already been metabolized by the liver through alcohol dehydrogenase. The metabolites of ethylene glycol, glycolic acid and oxalic acid, are responsible for the kidney failure seen with antifreeze ingestion. Although glycolic acid is directly toxic to the tubules, oxalic acid can precipitate in the tubules as calcium oxalate crystals and cause tubular obstruction and resultant QUIZ PAGE ANSWERS AJKD (continued) renal failure. The biopsy specimen showed clear evidence of acute tubular necrosis and calcium oxalate crystals. The patient required dialysis therapy for several days, but he recovered function. His most recent blood urea nitrogen and creatinine levels were 16 mg/dL (5.7 mmol/L) and 2.3 mg/dL (203 ␮mol/L), respectively. Final diagnosis: Ethylene glycol toxicity. Case provided by Steve Keiran, MD, If you have an interesting case you would Bhagwan Bhimani, MD, and Ashwin Dixit, like to submit for consideration, please go to MD, University of Louisville, Louisville, KY. http://ajkd.edmgr.com to do so. CASE REPORT Tethered Hemodialysis Catheter With Retained Portions in Central Vein and Right Atrium on Attempted Removal

Hla Thein, MBBS, and Sharad K. Ratanjee, MBBS

● A case of tunneled Tesio (Medcomp, Harrisburg, PA) twin hemodialysis catheters (silicone) extensively tethered subcutaneously and intravascularly is reported. Attempted removal resulted in significant portions of both catheters being retained in the superior vena cava and right atrium. Am J Kidney Dis 46:E35-E39. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Tethered dialysis catheter; fracture; retained portions; central veins; right atrium.

EMODIALYSIS (HD) treatment requires was stopped and changed to HD therapy in January 2001 H reliable access to the circulation. An arte- because of recurrent continuous ambulatory peritoneal dialy- riovenous fistula or graft is recommended for sis peritonitis. HD therapy was started through Tesio twin cuffed tunneled venous catheters (10 Fr) that were inserted treatment of patients with end-stage renal failure. January 26, 2001, through the left internal jugular vein. She However, only approximately 30% of patients had very poor limb vessels, with failed attempts at brachio- have a permanent functioning vascular access at cephalic fistula/graft and no further suitable native vascular the start of dialysis therapy.1 At present, tunneled access available in the upper limbs (she was reluctant to cuffed central venous catheters have an increas- have permanent vascular access in the lower limbs). Her other medical problems were quiescent rheumatoid arthritis, ingly important role in the delivery of HD, either hypertension, ischemic heart disease (previous coronary as interim vascular access or in patients with angiography showed triple-vessel disease), duodenal ulcer, difficult vascular access.2 At our unit, 27% of and osteoporosis. prevalent HD patients have a catheter for HD In November 2003, the patient presented with fever, access. This compares with 23% for New Zea- chills, and rigors during HD. There was no obvious source of infection, and the catheter exit site looked dry and clean. land, 9% for Australia (Australia and New Zea- There were no previous catheter-related infections. Blood land Dialysis and Transplant Registry), and 19% flow rates of 250 mL/min were being achieved, and Kt/V in the United States (US Renal Data System). (urea single pool) in the previous week was 1.34. A complete The majority of catheter use in New Zealand is blood count showed the following values: hemoglobin, 10.6 as a bridge to the creation and/or maturation of a g/dL (106 g/L); white blood cells, 10 ϫ 103/␮L (10 ϫ 109/L) ϫ 3 ␮ ϫ native arteriovenous fistula. Two major complica- with neutrophilia; and platelets, 273 10 / L (273 109/L). Sodium level was 134 mEq/L (134 mmol/L); potas- tions of HD catheters are catheter and/or central sium, 4.8 mEq/L (4.8 mmol/L); chloride, 98 mEq/L (98 venous thrombosis or catheter-related infections. mmol/L); urea, 22 mg/dL (8 mmol/L); creatinine, 4.8 mg/dL With the growing use of catheters, it is antici- (420 ␮mol/L); calcium, 9.2 mg/dL (2.3 mmol/L); phosphate, pated that more complications, including short- 3.7 mg/dL (1.18 mmol/L); albumin, 2.9 g/dL (29 g/L); and ␮ term and long-term problems, increasingly will ferritin, 785 ng/m (785 g/L). A chest radiograph did not show consolidation, and the catheter position remained be observed. satisfactory (Fig 1). Blood cultures grew nonfermenting We report a rare case of Tesio (Medcomp, gram-negative bacilli sensitive to the usual antibiotics. She Harrisburg, PA) silicone twin cuffed tunneled was treated with intravenous gentamicin and flucloxacillin. catheters extensively tied down by adhesion/ fibrous tissue in the subcutaneous tunnel and also tethered in the central vein and possibly the right From the Department of Renal Medicine, Middlemore atrium in a HD patient. Attempts at catheter Hospital, Counties Manukau District Health, Auckland, NZ. Received December 28, 2004; accepted in revised form removal resulted in significant portions of the May 31, 2005. catheters being retained in the superior vena cava Originally published online as doi:10.1053/j.ajkd.2005.05.030 and right atrium. on August 1, 2005. Address reprint requests to Hla Thein, MBBS, Depart- ment of Renal Medicine, Middlemore Hospital, Counties CASE REPORT Manukau District Health, Auckland 1006, NZ. E-mail: A 53-year-old woman was referred to our dialysis unit in [email protected] 2000 for treatment of end-stage renal failure secondary to © 2005 by the National Kidney Foundation, Inc. interstitial nephritis and amyloidosis. She initially was on 0272-6386/05/4603-0025$30.00/0 continuous ambulatory peritoneal dialysis therapy, which doi:10.1053/j.ajkd.2005.05.030

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: E35-E39 e35 e36 THEIN AND RATANJEE

Fig 1. Chest radiographs obtained (left) January 27, 2001 (the day after insertion of the Tesio twin catheter), and (right) November 2, 2003 (on admission with catheter-related sepsis); both show satisfactory positions of the Tesio twin catheters.

Her temperature settled for a day, but recurred during the addition, the fragments were unlikely to migrate for the next dialysis treatment. A diagnosis of catheter-related per- same reason. sistent infection was considered, and the decision was made Unfortunately, the patient developed pain in the front of to remove the Tesio twin catheters. the neck associated with tachycardia with a heart rate of 104 Initial removal of the catheters was attempted by the beats/min and new ST segment depression, as well as T nephrologist in the ward procedure room. Extensive adhe- wave inversion in the lateral leads on an electrocardiogram sions of subcutaneous fibrous tissue to the cuffs, as well as overnight. Cardiac troponin I levels also were increased to 7 the catheters (both proximal and distal to the cuff), were ng/mL (7 ␮g/L; normal range, 0 to 2 ng/mL) as a peak, noted. A circumferential scoring (ie, cutting) of the subcuta- which was consistent with troponin positive non–ST eleva- neous fibrous sheath proximal and distal to the cuffs of both tion myocardial infarction. Cardiology review was requested. catheters did not release them, as one would expect. The skin An echocardiogram showed the tips of the catheter in the right excision was extended to dissect the tightly adherent fibrous atrium without thrombus or vegetations. Cardiac surgeons were sheath up to almost the point of entry into the left internal consulted for an additional opinion. Because of the patient’s jugular vein. However, both catheters failed to come out. comorbidities and recent non–ST elevation myocardial infarc- Neither catheter was broken or fractured at this stage. A vascular surgeon was consulted, and the patient was taken to the operating room. The neck wound was explored under general anesthesia. The venous catheter was dissected down to entry into the vein. The catheters appeared to be tethered within the vessels and/or right atrium. On traction, the venous catheter broke, with the break several centimeters in the left internal jugular vein. The arterial catheter also was dissected down, and the vein was mobilized and dissected from the left vagus nerve and carotid artery. It also appeared to be tethered more centrally in the vein and right atrium. On an attempt at extraction with traction, it broke off many centimeters in the central vein. The hole in the left internal jugular vein then was closed. A subsequent chest radiograph (Fig 2) did not show any distal movement or embolization of the retained fragments. The patient remained hemodynami- cally stable during the operation and returned to the ward. The feasibility of removing the retained parts of the catheters was discussed with the interventional radiologist and surgeons. It was believed that the catheters were likely Fig 2. Chest radiograph showing retained portions to be tightly adhered to the vessel wall and/or atrium, and of the Tesio twin catheters after attempted removal percutaneous attempts were unlikely to be successful. In (arrows). TETHERED HEMODIALYSIS CATHETER e37 tion and the nature of the operation, it was decided to treat her percutaneous transvenous snare technique was conservatively at this stage with intravenous antibiotics. Her used to retrieve dislocated fragments.14 temperature settled and she was able to be discharged with a In our case, the extravascular tunneled por- temporary right internal jugular catheter for dialysis. Treatment was continued with intravenous gentamicin postdialysis. A plan tions of the catheters were firmly attached to the was made to create a polytetrafluoroethylene graft in the lower subcutaneous tissue. The catheters were sur- limbs by the vascular surgeons. rounded by fibrous tissue throughout the subcuta- A week later, the patient re-presented with fever with neous length (ie, parts distal and proximal to the severe chest pain that was not typical of angina. More blood Dacron cuff (Medcomp), as well as the cuff cultures were obtained, and ciprofloxacin was added to her itself). The catheters could not be extracted de- therapy. She underwent coronary angiography in preparation for surgical removal of the retained portions of the catheters. spite careful circumferential dissection of the The coronary angiogram showed severe triple-vessel disease fibrous tissue sheath down to the venous inser- unchanged from the previous study (which was done June 8, tion site. The catheters fractured several centime- 2002), with no lesions suitable for percutaneous angioplasty. ters deep to the venous insertion site when trac- However, she developed an acute myocardial infarction tion on the catheter was applied by the vascular complicated by cardiac arrest the same night. Cardiopulmo- surgeon. This most likely is caused by the cath- nary resuscitation was unsuccessful. Postmortem examina- tion was not performed for social and cultural reasons. eters being firmly adhered to the central veins The removed portions of the catheters were sent for addi- and possibly also the right atrial wall. tional analyses. These were compared with a catheter removed In our case, because of suspected extensive from another patient who also had the same type of Tesio adherence in the superior vena cava and right catheters in for only a few months. Histological and electron atrium, surgical removal was considered, al- microscopic examination did not show significant differences between the catheters or significant defects or cellular in- though it was postponed initially because of the growths (fingerlike projections) into the catheter material. patient’s comorbidities and recent non–ST eleva- tion myocardial infarction. DISCUSSION Circumferential fibrocellular sheaths surround- ing the intravascular parts of the catheters are well To our knowledge, this is the first report of a known to occur when catheters are left in situ for tethered central venous catheter leaving behind prolonged periods.15-18 Forauer and Theoharis16 significant intravascular and intracardiac frag- noted in an autopsy study that in the short term ments after attempted removal in adults. There is (Ͻ14 days after catheter placement), the access 1 case report of a spontaneous hemodialysis vein and superior vena cava adjacent to the central catheter fracture, thought to be caused by a 3 venous catheters showed foci of local intimal in- manufacturing fault. jury, endothelial denudation, and a layer of adher- Spontaneous catheter fractures with remaining ent thrombus. In longer term catheters (Ͼ90 days), intravascular fragments have been reported there was smooth muscle cell proliferation leading mainly in patients who had catheters inserted for to vein wall thickening. In the focal area of catheter 4-7 reasons other than HD therapy. The exact attachment to the vein wall, composed of thrombi incidence of catheter fracture is not known. Biffi in various stages of organization, collagen and 8 et al reported an incidence of 0.68% in 1,320 endothelial cells also were observed. Several re- subclavian port placements. Mainly, such frag- ports also described the circumferential sheath that mentation concerned catheters inserted in the forms around the intravascular part of the cath- subclavian vein, with the fracture at the place of eter.16,17 Some reported that this was fibrin,16 passage under the clavicle, so-called “pinch-off” whereas others17 reported that it was a stable cellu- syndrome.9,10 Catheter fragments may cause such lar collagen tissue formed mainly by smooth muscle severe complications as cardiac arrest, vascular cells migrating from injured vein wall into the early perforation, and pulmonary embolism.11 Two pericatheter thrombus covered by endothelium. This cases of spontaneous breakdown of an HD cath- encapsulating sheath may become tightly adhered eter have been reported; 1 caused extensive recir- to the vessel wall. culation,12 and the other caused leakage, but no In our case, it is feasible that such a process free fragment developed.13 may have occurred in the central veins and right The management of retained portions of cath- atrium. Ingrowth of fibrocelluar tissue and/or eters is not well established. In the literature, a thrombus into holes of the distal ends of the e38 THEIN AND RATANJEE catheters, thus tethering the distal catheters tightly The optimal duration of tunneled central ve- to the right atrial wall, also is a possibility, but nous catheter use is unknown. Perhaps regular less likely to be an extensive process in view of replacement at set intervals (eg, 18 to 24 months) the good blood flow rates of at least 250 mL/min may be required. However, this can only be that were achieved. Silicone catheters are prone judged after more long-term experience with to fracture when directional stress is applied. In these catheters is reported. our case, the catheters probably fractured when traction was applied by the vascular surgeon in REFERENCES the operating room. 1. Stehman-Breen CO, Sherrard DJ, Gillen D, Caps M: There are reports of perforation of the superior Determinants of type and timing of initial permanent hemo- vena cava leading to hydrothorax, pleural effu- dialysis vascular access. Kidney Int 57:639-645, 2000 sion, chemical mediastinitis, and right phrenic 2. Schwab SJ, Buller GL, McCann RL, Stickel DL: Prospec- nerve palsy.19 However, these relate mainly to tive evaluation of a Dacron cuffed hemodialysis catheter for prolonged use. Am J Kidney Dis 11:166-169, 1988 catheters used for purposes other than dialysis, 3. Weijmer MC, Kars SM, ter Wee SMPM: A scanning mainly in patients receiving chemotherapy. The electron microscopy analysis of spontaneous hemodialysis majority of these complications occurred when a catheter fracture. Am J Kidney Dis 38:858–861, 2001 subclavian approach (either left or right) was 4. Richardson JD, Grover FL, Trinkle JK: Intravenous taken. This may increase the chance of central catheter emboli: Experience with twenty cases and collec- vein endothelial damage and increased sheath tive review. Am J Surg 128:722-727, 1974 5. Biffi R, Corrado F, de Braud F, et al: Long-term, totally formation. Our patient had no evidence of central implantable central venous access ports connected to a vein perforation. Secondary distal migration also Groshong catheter for chemotherapy of solid tumors. Eur J was considered as a possible cause, but the Cancer 33:1190-1194, 1997 postinsertion chest radiograph (January 27, 2001) 6. Kock HJ, Pietsch M, Krause U, Wilke H, Eigler FW: and chest radiograph during the current admis- Implantable vascular access systems. World J Surg 22:12- sion (November 2, 2003; Fig 1) did not show 16, 1998 20 7. Inoue Y, Nezu R, Nakai S, Takagi Y,OkadaA: Spontaneous distal migration during a 33-month period. partial fracture of the catheter of a totally implantable subcutane- Unfortunately our patient died of a cardiac event, ous infusion port. J Parenter Enteral Nutr 16:75-77, 1992 and no postmortem examination was available to 8. Biffi R, Pozzi S, Orsi F, et al: Catheter rupture and study the retained portions of the catheters by distal embolisation: A rare complication of venous ports. electron microscopy. In the previously reported Vasc Access 1:19-22, 2000 case of spontaneous fracture, electron microscopy 9. Hou WY, Sun WZ, Chen YA, Wu SM, Lin SY: Case report: “Pinch off sign” and spontaneous fracture of im- and energy-dispersive X-ray spectral analysis planted central venous catheter. J Formos Med Assoc 93:S65- showed that barium sulfate particles, used to visual- S69, 1994 (suppl 1) ize the catheter on fluoroscopy, were not properly 10. Ramsden WH, Cohen AT, Blanshard KS: Case report: mixed in silicone matrix, suggesting an error in the Central venous catheter fracture due to compression be- manufacturing process.3 Confocal and electron mi- tween the clavicle and first rib. Clin Radiol 50:59-60, 1995 11. Uflacker R, Lima S, Melichar AC: Intravenous for- croscopy of the removed part of the silicone Tesio eign bodies: Percutaneous retrieval. Radiology 160:731- twin catheters from our patient (who had catheters 735, 1986 for 33 months) was compared with the same type 12. Sarnak MJ, Halin N, King AJ: Severe access recircu- of silicone Tesio twin catheters successfully re- lation secondary to free flow between the lumens of a moved after only a few months in another patient. dual-lumen dialysis catheter. Am J Kidney Dis 33:1168- No significant differences were noted, suggesting 1170, 1999 13. Verhage AH, van Bommel EFH: Catheter frac- that material and manufacturing problems are un- ture—An under recognized and serious condition in hemodi- likely. alysis. Nephrol Dial Transplant 15:901-903, 2000 Thus, our case likely is related to fibrous 14. Amesur NB: Vascular Access. Available at: http:// adhesions developing in a catheter in place for www.emedicine.com/radio. Accessed: November 24, 2004 almost 3 years. Other possible explanations for 15. Suojanen JN, Brophy DP, Nasser I: Thrombus on extensive adhesions are unrecognized previous indwelling central venous catheters: The histology of “fibrin sheath.” Cardiovasc Intervent Radiol 23:194-197, 2000 infections; imperfect calcium, phosphate, and 16. Forauer AR, Theoharis C: Histological changes in parathyroid hormone control; and the material the human vein wall adjacent to indwelling central venous used for the catheter. catheters. J Vasc Intervent Radiol 14:1163-1168, 2003 TETHERED HEMODIALYSIS CATHETER e39

17. Hoshal VL, Ause RG, Hoskins PA, Mich AA: Fibrin right phrenic nerve palsy during 5-fluorouracil continuous sleeve formation on indwelling venous catheters. Arch Surg infusion plus cisplatin and vinorelbine in breast cancer 102:353-358, 1971 patients. J Natl Cancer Inst 92:755-756, 2000 18. Xiang DZ, Verbeken EK, Van Lommel ATL, Stas 20. Petersen J, Delaney JH, Brakstad MT, Rowbotham M, De Wever EK: Composition and formation of the RK, Bagley CM Jr: Silicone venous access devices sleeve enveloping a central venous catheter. J Vasc Surg positioned with their tips high in the superior vena cava 28:260-271, 1998 are more likely to malfunction. Am J Surg 178:38-41, 19. Munzone E, Nole F, Orlando L, et al: Unexpected 1999 CASE REPORT Tubulointerstitial Nephritis and Fanconi Syndrome in Primary Biliary Cirrhosis

Marie Lino, MD, Raynald Binaut, MD, Laure-Hélène Noël, MD, Natacha Patey, MD, Pierre Rustin, MD, Laurent Daniel, MD, Jeanne Serpaggi, MD, Anne Varaut, MD, Philippe Vanhille, MD, Bertrand Knebelmann, MD, PhD, Jean-Pierre Grünfeld, MD, and Fadi Fakhouri, MD

● Primary biliary cirrhosis is a chronic cholestatic liver disease of unknown cause that predominantly affects middle-aged women. Distal tubular acidosis is the main renal complication of primary biliary cirrhosis. Tubulointer- stitial nephritis and Fanconi syndrome have been reported more rarely. We report on 2 patients with primary biliary cirrhosis who presented with tubulointerstitial nephritis and Fanconi syndrome and review similar cases published previously. Serum from 1 patient exerted an inhibitory effect on pyruvate dehydrogenase and ␣-ketoglutarate dehydrogenase, 2 mitochondrial enzymes that are the main targets of antimitochondrial antibodies in primary biliary cirrhosis. Antimitochondrial antibodies may have a role in the genesis of tubulointerstitial nephritis and Fanconi syndrome, 2 typical renal features of mitochondrial cytopathies. Tubulointerstitial nephritis and Fanconi syndrome have to be added to the spectrum of renal diseases associated with primary biliary cirrhosis. Am J Kidney Dis 46:E41-E46. © 2005 by the National Kidney Foundation, Inc.

INDEX WORDS: Primary biliary cirrhosis; Fanconi syndrome; tubulointerstitial nephritis; antimitochondrial antibodies.

RIMARY BILIARY cirrhosis (PBC) is a with microscopic polyangiitis. Only the 2 patients with TIN P chronic cholestatic liver disease of un- and Fanconi syndrome are presented here. known cause that predominantly affects middle- Fanconi syndrome is defined by the coexistence of hypo- aged women.1 The presence of antimitochon- kalemia, hypophosphatemia with a low renal fractional tubular reabsorption of phosphate, metabolic acidosis, drial antibodies (AMAs) is a hallmark of the normoglycemic glycosuria, and generalized aminoaciduria. disease. Distal tubular acidosis (DTA) is the Aminoaciduria was measured by using urinary amino acid main feature of renal involvement in patients chromatography. Glycosuria was detected by using dipsticks with PBC.2 It is found in up to 33% of patients, and quantified by means of the enzymatic method. Frac- but usually is without clinical consequence. More tional tubular reabsorption of phosphate was estimated as rarely, tubulointerstitial nephritis (TIN) has been 1 – (U/P) phosphate/(U/P) creatinine, where U and P are reported in patients with PBC.3-5 We report 2 urinary and plasma concentrations, respectively. cases of TIN associated with PBC and present histopathologic kidney and liver data. Further- more, the inhibitory effect of circulating AMAs on mitochondrial enzymes in shown. Finally, we compare our findings with the scarce data previ- From the Service d=Hépatologie AP-HP, Hôpital Necker; ously published. Service de néphrologie, Unité INSERM U507, and Labora- toire d=anatomie pathologique, AP-HP, Hôpital Necker- Enfants Malades; Université Paris V-René Descartes; METHODS INSERM U393, Paris; Service de médecine interne-néphrolo- We reviewed medical records and pathological data for gie, Hôpital de Valenciennes, Valenciennes; and Laboratoire patients referred from 1990 to 2004 to the Department of d’anatomie pathologique, Hôpital de La Timone, Marseille, Nephrology at Hôpital Necker-Enfants Malades (Paris, France. France) and Department of Internal Medicine and Nephrol- Received March 4, 2005; accepted in revised form May 9, ogy at Hôpital de Valenciennes (France) to identify patients 2005. with PBC and renal disease. PBC diagnosis was made based Originally published online as doi:10.1053/j.ajkd.2005.05.021 on the presence of at least 3 of the following criteria: on August 1, 2005. alkaline phosphatase or ␥-glutamyltransferase level greater Address reprint requests to Fadi Fakhouri, MD, Service than the upper limit of normal; positive AMAs at a titer of de Néphrologie, Hôpital Necker-Enfants Malades, 149 rue 1:20; increased immunoglobulin M (IgM) level; absence of de Sèvres, 75015 Paris, France. E-mail: fadi.fakhouri@ biliary obstruction by means of ultrasonography, computed nck.ap-hop-paris.fr tomography, or cholangiography; or compatible liver bi- © 2005 by the National Kidney Foundation, Inc. opsy. Five patients were identified: 4 patients with TIN 0272-6386/05/4603-0026$30.00/0 (Fanconi syndrome was noted in 2 patients) and 1 patient doi:10.1053/j.ajkd.2005.05.021

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: E41-E46 e41 e42 LINO ET AL

Enzyme Studies Table 1. Relevant Biological Data at the Time of Diagnosis of TIN in 2 Patients With PBC We studied the effect of patient 1 plasma (patient 2 plasma was not available) and total IgG on 2 mitochondrial enzymes Patient 1 Patient 2 (pyruvate dehydrogenase [PDH] and ␣-ketoglutarate dehy- ␣ drogenase [ -KDH]). These 2 enzymes are major targets of SCr (mg/dL) 1.77 1.3 ␣ AMAs in patients with PBC. PDH and -KDH activity was CrCl (mL/min) 32 41 measured spectrophotometrically by following the reduction Serum potassium (mEq/L) 3.9 3.3 of the oxidized form of nicotinamide-adenine dinucleotide ϩ Serum phosphate (mg/dL) 1.17 2.41 (NAD ) at 340 nm in a medium consisting of 0.3 mol/L of Total carbon dioxide (mEq/L) 21.5 21.8 mannitol, 5 mmol/L of MgCl2, 10 mmol/L of KCl, 0.1% Uricemia (mg/dL) 2.01 2.85 Triton X-100, 1 mmol/L of freshly prepared cysteine, 240 Total bilirubin (mg/dL) 0.35 0.29 ␮mol/L of thiamine pyrophosphate, 200 ␮mol/L of coen- ϩ Alkaline phosphatase (U/L) 107 276 zyme A, 0.5 mmol/L of NAD , and 10 mmol/L of KH2PO4 ␥-Glutamyltranferase (U/L) 53 51 (pH 7.4). Each enzyme measurement was started by adding Aspartate aminotransferase (U/L) 41 25 its organic acid substrate, ie, 0.4 mmol/L of pyruvate and 0.5 Alanine aminotransferase (U/L) 67 34 ␣ ␣ mmol/L of -ketoglutarate for the PDH and -KDH assays, Proteinuria (g/d) 1.3 1.47 ␣ respectively. Measurements of -KDH activity were per- Hematuria (red blood cells/␮L) 25 25 formed by using mouse liver mitochondria prepared as ␤ -Microglobulinuria* (mg/24 h) 25 57 5 2 previously described. The PDH activity assay was per- Fractional tubular reabsorption of 42 16 formed on PDH purified from porcine heart (Sigma Chemi- phosphate† (%) cal Comp, Lyon, France). Protein content was estimated by Hyperaminoaciduria ϩϩ using the Bradford standard assay. Urinary pH 6.5 6 Glycosuria ϩϩ CASE REPORTS NOTE. To convert SCr in mg/dL to ␮mol/L, multiply by Patient 1 88.4; CrCl in mL/min to mL/s, multiply by 0.01667; potas- A 51-year-old woman was referred for the evaluation of sium and total carbon dioxide in mEq/L to mmol/L, multiply chronic renal failure diagnosed 3 years earlier. At that time, by 1; uric acid in mg/dL to ␮mol/L, multiply by 59.48; serum creatinine (SCr) level was 1.4 mg/dL (121 ␮mol/L; bilirubin in mg/dL to ␮mol/L, multiply by 17.1. creatinine clearance [CrCl], 40 mL/min [0.67 mL/s]). Hy- *Normal, less than 0.35 mg/24 h. pophosphatemia (phosphate, 1.9 mg/dL [0.62 mmol/L]), †Normal, greater than 0.86 mg/24 h. hypouricemia (uric acid, 1.74 mg/dL [104 ␮mol/L]), and mild proteinuria (protein, 1 g/d) were noted. Hematuria was absent. A renal biopsy performed at that time showed TIN present. Xerostomia was noted, and mild lymphocytic infil- with a marked lymphocytic (CD3ϩ) infiltrate. Two years trate was present in a salivary gland biopsy. One year later, later, the patient reported bone pain affecting mainly the ribs she presented with bone fractures. Hypophosphatemia and and hips. Rheumatological workup, including bone biopsy, hypocalcemia were noted. Osteomalacia was suspected, but led to the diagnosis of osteomalacia complicating Fanconi could not be formally confirmed on bone biopsy. At that syndrome. Relevant biological data at referral are listed in time, SCr level was 1 mg/dL (90 ␮mol/L) and rapidly Table 1. Fanconi syndrome was diagnosed based on hypouri- increased to 1.35 mg/dL (120 ␮mol/L) 4 months later. cemia, hypophosphatemia, generalized aminoaciduria, and On admission, blood pressure was 135/74 mm Hg. Physi- normoglycemic glycosuria. A salivary gland biopsy was cal examination was unremarkable. Relevant laboratory test unremarkable. Immunologic test results were negative ex- results are listed in Table 1. Fanconi syndrome was diag- cept for the presence of antinuclear antibodies (1/160) and nosed based on hypokalemia, hypouricemia, hypophos- high-titer (Ͼ1/800) type 2 AMAs. Test results for hepatitis B phatemia, generalized aminoaciduria, and normoglycemic and C virus were negative. Kidney and liver biopsies were glycosuria. Test results for type 2 AMAs were positive performed (Fig 1A-D). (1/640), with low-titer antinuclear antibodies (1/40) in the The patient was started on calcitriol therapy (0.25 ␮g for 2 absence of antiextractable nuclear antigen antibodies, includ- days), and phosphatemia normalized. She subsequently was ing anti-SSA and anti-SSB antibodies. Serological test re- administered ursodeoxycholic acid and a course of steroid sults for hepatitis B and C were negative. A renal biopsy was therapy (0.5 mg/kg/d, with progressive tapering over 6 performed (Fig 1F). The patient was started on therapy with months), which led to a transient improvement in renal sodium bicarbonate calcitriol and potassium chloride oral function (SCr, 1.35 mg/dL [120 ␮mol/L]). At last follow-up, supplementation. Despite a course of steroids (0.5 mg/d with 1 year after referral, SCr level was 1.58 mgdL (140 ␮mol/L; gradual tapering), renal failure progressed, and at last follow- CrCl, 36 mL/min [0.6 mL/s]). up, SCr level was 2.29 mg/dL (203 ␮mol/L). Patient 2 Histopathologic Studies This 68-year-old woman had been given a diagnosis of Kidney biopsy findings. In both patients, renal biopsies PBC (stage 2 on liver biopsy) 2 years before her referral to showed features of severe TIN characterized by: (1) marked the nephrologist. High-titer (1/640) type 2 AMAs were interstitial cellular infiltrate (Fig 1A, F, and G); cellular PRIMARY BILIARY CIRRHOSIS AND KIDNEY e43

Fig 1. (A-D) Patient 1. Kidney biopsy specimen; light microscopy shows (A) diffuse interstitial cellular infiltration and the presence of atrophic tubules with intratubular casts (periodic acid–Schiff staining; original magnification ؋250) and (B) glomerular cyst with retraction of the glomerular tuft, enlargement of the urinary space, and moderate thickening of tubular basement membranes (Jones staining; original magnification ؋250). (C) Liver biopsy specimen; light microscopy shows the presence of a periportal cellular infiltrate. (Hematoxylin-eosin staining; -original magnification ؋100.) (D) Kidney biopsy specimen; electron microscopy shows increased size of mitochon dria with rarification of invaginations and accumulation of a granulous material (#). Patient 2. (E) Kidney biopsy specimen; light microscopy shows CD3؉ cells infiltrating the renal interstitium, marked interstitial fibrosis and tubular atrophy, and the presence of a sclerotic glomerulus (# #). (Periodic acid–Schiff staining; original magnifica- (.tion ؋250 infiltrate was characterized in cases 1 and 2 and composed of patient IgG (Fig 3, trace c) led to a progressive reduction mainly of CD4ϩ, CD3ϩ, and CD20ϩ lymphocytes, respec- in NADϩ, allowing approximate titration of the inhibition tively; and (2) diffuse interstitial fibrosis and tubular atrophy effect exerted by serum (10 ␮L of IgG neutralizing as much (Fig 1B and F). Sclerotic glomeruli were 2 of 15 and 4 of 15 as8to9␮UI of the purified PDH enzyme). in patients 1 and 2, respectively. Immunofluorescence study results were negative in both cases. DISCUSSION Liver biopsy findings. Liver biopsy was available in both patients and showed typical features of PBC, including DTA classically is the main feature of renal a marked reduction in number of biliary ducts and periportal involvement in patients with PBC. It occurs in cellular infiltrate (Fig 1). one third of patients with an advanced stage of No amyloid deposits were found in kidney and liver the disease, but usually is without clinical conse- biopsy specimens. quence. The pathogenesis of renal DTA in pa- tients with PBC remains speculative. More rarely, Enzyme Studies membranous nephropathy6 and microscopic poly- The target of AMAs in patient 1 was studied by means of angiitis7 have been reported in patients with Western blot analysis using patient serum purified mitochon- dria and purified PDH and ␣-KDH (Fig 2). AMAs were PBC. directed against both PDH and ␣-KDH. We describe 2 cases of severe TIN and Fan- Patient 1 serum (50 ␮L) exerted a significant inhibitory coni syndrome in patients with PBC and review effect (Ͼ75%) on ␣-KDH activity in purified mouse liver 2 similar previously published cases3,4 (Table 2). mitochondria. Noticeably, a similar inhibitory effect was In contrast to DTA, this type of nephropathy observed on ␣-KDH purified from human cultured skin fibroblasts (data not shown). Patient total IgG (10 ␮L) also rarely occurs in the course of PBC. However, we strongly inhibited purified porcine heart PDH (Fig 3, traces a are unable to estimate the incidence of TIN/ and c). Successive additions of purified PDH in the presence Fanconi syndrome in patients with PBC because e44 LINO ET AL

Fig 2. Western blot analysis shows reactivity of patient 1 serum with purified mitochondria and enzymes. Patient serum (1/500 dilution) was used to hybridize the membrane in which purified PDH (1 and 5 mg; lanes 1, 2, respectively), ␣-KDH (1 and 5 mg; lanes 3 and 4, respectively), human brain mitochondria (10 and 100 mg; lanes 5 and 6), and mouse liver mitochondria (10 and 100 mg; lanes 7 and 8) were deposited. Molecular weight markers are indicated on the left. The patient’s serum clearly recognizes the PDH-E2 at 74 kd, protein X at 56 kd, and a 45-kd subunit of PDH (lane 2). The serum also recognizes ␣-KDH E2 at 52 kd and the PDH-E2 band at 74 kd (lane 4) because of a known 5% contamination of PDH in the ␣-KDH preparation, indicated by the manufacturer.

Fig 3. Effect of serum and purified IgG from patient 1 on ␣-KDH and PDH activity. (A) ␣-KDH activity measured in mouse liver mitochondria in the presence of (a) control and (b) patient 1 serum (50 ␮L); percent indicates re- sidual ␣-KDH activity mea- sured in the presence of pa- tient serum compared with control serum. (B) Purified PDH activity from porcine heart measured in the (a) ab- sence or presence of either (b) control or (c) patient total IgG (10 ␮L). Successive addi- tions of purified PDH in the presence of (trace c) patient IgG led to a progressive re- duction of NAD؉, allowing approximate titration of the inhibition exerted by serum (10 ␮L of IgG neutralizing as much as 8 to 9 ␮UI of the purified PDH enzyme). Num- bers along the trace are nanomoles per minute. PRIMARY BILIARY CIRRHOSIS AND KIDNEY e45

Table 2. Previously Published Cases of TIN and Fanconi Syndrome in Patients With PBC

Age SCr CrCl Proteinuria Fanconi Renal Hepatic Reference (y)/Sex (mg/dL) (mL/min) (g/d) Syndrome DTA AMA Test Results Treatment

3 36/F 0.8 84 0.2 ϩ 1/640 Normal None 4 58/F 2.03 25 NS ϩϩNS Elevated ALP, Steroids; no AST, and improvement in ALT levels renal function

NOTE. To convert SCr in mg/dL to ␮mol/L, multiply by 88.4; CrCl in mL/min to mL/s, multiply by 0.01667. Abbreviations: ALP, alkaline phosphatase; AST, aspartate aminotransferase; ALT, alanine aminotransferase; Nal, . our patients were identified through nephrology Moreover, Fanconi syndrome related to such departments, not hepatology departments. antiviral agents as cidofovir and adefovir is These patients share several clinical and bio- thought to be caused by drug-induced mitochon- logical features. First, in all patients, renal dis- drial toxicity.11 Type 2 AMAs in patients with ease occurred during the early phase of PBC in PBC are directed against 3 mitochondrial en- the absence of marked hepatic abnormalities. zymes: PDH, ␣-KDH, and branched-chain keto- This observation precludes an implication of acid decarboxylase. Previous in vitro studies12,13 hepatic dysfunction in the genesis of TIN, as and our assays using total plasma and IgG of previously suggested for DTA. Second, proximal patient 1 (Fig 2) show that AMAs significantly tubulopathy led to the diagnosis of PBC in 2 decrease the activity of these enzymes. More- cases, underlying that renal disease may show over, in vitro studies suggest that IgA purified from PBC. Of note, decreased renal tubular uric acid patients with PBC can translocate into renal tubular reabsorption, a stigma of proximal tubule dys- cells, probably through endocytosis, and colocalize function, has been documented previously in 14 8 with PDH complex. One cannot exclude that patients with PBC. Third, renal failure was these antibodies may in some instances interfere significant in 3 of 4 patients. with intrarenal mitochondrial machinery, leading Extrahepatic autoimmune disorders are en- to Fanconi syndrome and TIN. countered frequently in patients with PBC. They There are 2 other intriguing observations. (1) include Sjögren syndrome, scleroderma/calcino- Why do some patients with PBC develop DTA sis, Raynaud phenomenon, esophageal motility and others, more rarely, develop proximal tubu- disorders, sclerodactyly, and telangiectasia syn- 1 lopathy? In patients with Sjögren syndrome, DTA drome, and autoimmune thyroiditis. TIN and ϩ probably is related to the absence of the H - Fanconi syndrome may represent additional auto- immune diseases coexistent to PBC, as previ- transporting adenosine triphosphatase synthase pump from the apical membrane of collecting ously reported in patients with Sjögren syn- 15 drome, although proximal tubulopathy is very duct cells. The absence of this pump possibly is rare in the setting of the latter. However, one caused by autoantibodies that interfere with its cannot exclude that TIN may be pathogenically targeting to the apical membrane. It is unknown related to PBC. First, hepatic cellular infiltrate whether such abnormalities occur in PBC- composed of CD4ϩ and CD8ϩ lymphocytes is a associated DTA, and whether AMAs have a common pathological feature in patients with pathogenic role in this setting. (2) In the case that PBC. Autoreactive T lymphocytes may be driven AMAs have a pathogenic role in the genesis of by abnormal antigen expression in hepatocytes, PBC-related TIN/Fanconi syndrome, the rarity such as mitochondrial antigen (ie, PDH-E2) on of this type of nephropathy contrasts with the the hepatocyte membrane surface.9 Such abnor- high frequency of high-titer AMAs in patients mal antigen expression may occur in renal tubu- with PBC. This discrepancy may be caused by lar cells and lead to infiltration of the renal peculiar physicochemical characteristics restricted interstitium by autoreactive T cells. to rare subsets of AMA or variable susceptibility of Second, Fanconi syndrome and TIN are typi- mitochondrial enzymes to the inhibitory effect of cal renal features of mitochondrial cytopathies.10 AMAs. Similarly, proximal tubulopathy is not e46 LINO ET AL constant in patients receiving treatments with 4. Kamouchi M, Tsuji H, Hirakata H, et al: Tubulointersti- well-established mitochondrial toxicity. tial disorders in the kidney associated with primary biliary cirrhosis (PBC). Clin Nephrol 35:134-135, 1991 Symptomatic treatment of patients with Fan- 5. Lecoeur H, Langonné A, Baux L, et al: Real-time flow coni syndrome associated with PBC consists cytometry analysis of permeability transition in isolated mainly of the administration of calcitriol to avoid mitochondria. Exp Cell Res 294:106-117, 2004 the occurrence of osteomalacia and potassium, 6. Vlassopoulos D, Divari E, Savva S, et al: Membranous phosphate, and sodium bicarbonate supplementa- glomerulonephritis associated with primary biliary cirrhosis. Nephrol Dial Transplant 13:459-461, 1998 tion. Steroid administration has been effective in 7. Iannone F, Falappone P, Pannarale G, et al: Micro- some patients with TIN with intense cellular scopic polyangiitis associated with primary biliary cirrhosis. infiltrate. A short course of prednisone, which J Rheumatol 30:2710-2712, 2003 usually has no impact on the evolution of liver 8. Izumi N, Hasumura Y, Takeuchi J: Hypouricemia and hyperuricosuria as expressions of renal tubular damage in disease, had a slight transitory beneficial effect primary biliary cirrhosis. Hepatology 3:719-723, 1983 on TIN and renal failure in both patients pre- 9. Poupon R, Poupon RE: Retrovirus infection as a trig- sented here. It is unknown whether treatment for ger for primary biliary cirrhosis? Lancet 363:260-261, 2004 PBC, mainly ursodeoxycholic acid, may alter the 10. Niaudet P, Rotig A: The kidney in mitochondrial cytopathies. Kidney Int 51:1000-1007, 1997 course of renal disease. 11. Izzedine H, Launay-Vacher V, Isnard-Bagnis C, et al: In summary, TIN and Fanconi syndrome are to Drug-induced Fanconi’s syndrome. Am J Kidney Dis 41:292- be added to the spectrum of renal diseases associ- 309, 2003 ated with PBC. Our data suggest a role of AMAs 12. Fregeau DR, Roche TE, Davis PA, et al: Primary biliary cirrhosis. Inhibition of pyruvate dehydrogenase com- in the pathogenesis of renal disease during PBC, plex activity by autoantibodies specific for E1 alpha, a but this remains to be confirmed. non-lipoic acid containing mitochondrial enzyme. J Immu- nol 144:1671-1676, 1990 REFERENCES 13. Van de Water J, Fregeau D, Davis P, et al: Autoanti- bodies of primary biliary cirrhosis recognize dihydrolipo- 1. Talwalkar JA, Lindor KD: Primary biliary cirrhosis. amide acetyltransferase and inhibit enzyme function. J Im- Lancet 362:53-61, 2003 munol 141:2321-2324, 1988 2. Pares A, Rimola A, Bruguera M, et al: Renal tubular 14. Malmborg AC, Shultz DB, Luton F, et al: Penetration acidosis in primary biliary cirrhosis. Gastroenterology 80: and co-localization in MDCK cell mitochondria of IgA 681-686, 1981 derived from patients with primary biliary cirrhosis. J Auto- 3. Kodama T, Imai H, Wakui H, et al: Tubulointerstitial immun 11:573-580, 1998 nephritis with renal tubular acidosis and asymptomatic 15. Cohen EP, Bastani B, Cohen MR, et al: Absence of primary biliary cirrhosis accompanied by antibody to a H(ϩ)-ATPase in cortical collecting tubules of a patient with 52-kDa mitochondrial protein alone. Clin Nephrol 45:401- Sjogren’s syndrome and distal renal tubular acidosis. J Am 405, 1996 Soc Nephrol 3:264-271, 1992 The Official Journal of the National Kidney Foundation

American Journal of Kidney Diseases AJKDAmerican Journal of Kidney Diseases (ISSN 0272- elsevier.com. Requests may also be completed on-line 6386) is published monthly by Elsevier. (For Post Office via the Elsevier homepage (www.elsevier.com/locate/ use only: Volume 46 issue 3 of 6) Corporate Offices: 170 permissions). S Independence Mall W 300 E, Philadelphia, PA 19106- The appearance of the code at the bottom of the first 3399. Accounting and Circulation Offices: Elsevier, 6277 page of an article in this journal indicates the copyright Sea Harbor Dr, Orlando, FL 32887-4800. Periodicals owner’s consent that copies of the article may be made for postage paid at Orlando, FL 32862, and at additional personal or internal use, or for the personal or in- mailing offices. ternal use of specific clients. This consent is given on the POSTMASTER: Send change of address to American condition, however, that the copier pay the stated per-copy Journal of Kidney Diseases (ISSN 0272-6386), Elsevier, fee through the Copyright Clearance Center, Inc (222 Periodicals Department, 6277 Sea Harbor Dr, Orlando, Rosewood Dr, Danvers, MA 01923; (978) 750-8400) for FL 32887-4800. copying beyond that permitted by Sections 107 or 108 of Editorial correspondence should be addressed to Ber- the US Copyright Law. This consent does not extend to tram L. Kasiske, MD, Editor-in-Chief, AJKD, Hennepin other kinds of copying, such as copying for general distri- Faculty Associates, 600 HFA Building, Room D508, 914 bution, for advertising or promotional purposes, for creat- South 8th Street, Minneapolis, MN 55404. Telephone: ing new collective works, or for resale. Absence of the code (612) 347-7770 indicates that the material may not be processed through Business correspondence (subscriptions, change of the Copyright Clearance Center, Inc. Reprints of single address) should be addressed to the Publisher, Elsevier, articles available online may be obtained by purchasing Periodicals Department, 6277 Sea Harbor Dr, Orlando, Pay-Per-View access for $25 per article on the journal web FL 32887-4800. E-mail: [email protected] site, www.ajkd.org Change of address notices, including both the old and Reprints: To order 100 or more reprints for educational, new addresses of the subscriber, should be sent at least one commercial, or promotional use, contact the Commercial month in advance. Reprints Department, Elsevier Inc., 360 Park Avenue Customer Service: 1-800-654-2452; outside the United South, New York, NY 10010-1710. Fax: 212-462-1935; States and Canada, 1-407-345-4000. e-mail: [email protected]. Yearly subscription rates: United States and posses- Advertising representative: Cunningham Associates, sions: individual, $374.00; institution, $642.00; single 180 Old Tappan Rd, Old Tappan, NJ 07675. Telephone issue, $58.00. All other countries: individual, $551.00; 1-201-767-4170; fax 1-201-767-8065. institution, $694.00; single issue, $58.00. For all areas The ideas and opinions expressed in the American outside the United States and possessions, there is no Journal of Kidney Diseases do not necessarily reflect additional charge for surface delivery. Student and resi- those of the National Kidney Foundation, the Editor, or dent: United States and possessions, $190.00; all other the Publisher. Publication of an advertisement or other countries, $250.00. To receive student/resident rate, product mentioned in the American Journal of Kidney orders must be accompanied by name of affiliated insti- Diseases should not be construed as an endorsement of tution, date of term, and the signature of program/ the product or the manufacturer’s claims. Readers are residency coordinator on institution letterhead. Orders encouraged to contact the manufacturer with any ques- will be billed at individual rate until proof of status is tions about the features or limitations of the products received. Single issues, both current and back, exist in mentioned. The National Kidney Foundation and the limited quantities and are offered for sale subject to Publisher do not assume any responsibility for any availability. Current prices are in effect for back vol- injury and/or damage to persons or property arising out umes and back issues. Back issues sold in conjunction of or related to any use of the material contained in this with a subscription are on a prorated basis. Prices are periodical. The reader is advised to check the appropri- subject to change without notice. Checks should be ate medical literature and the product information cur- made payable to Elsevier and sent to American Journal of rently provided by the manufacturer of each drug to be Kidney Diseases, W.B. Saunders, Periodicals Depart- administered to verify the dosage, the method and dura- ment, PO Box 628239, Orlando, FL 32862-8239. tion of administration, or contraindications. It is the Copyright © 2005, National Kidney Foundation. All responsibility of the treating physician or other health rights reserved. No part of this publication may be repro- care professional, relying on independent experience duced or transmitted in any form or by any means now and knowledge of the patient, to determine drug dos- or hereafter known, electronic or mechanical, including ages and the best treatment for the patient. photocopy, recording, or any information storage and The contents of the American Journal of Kidney Dis- retrieval system, without permission in writing from the eases are indexed in Index Medicus and MEDLINE, Cur- publisher. Printed in the United States of America. rent Contents/Clinical Medicine, Science Citation In- Permissions may be sought directly from Elsevier’s dex, EMBASE/Excerpta Medica, ISI, and CINAHL. Rights Department in Philadelphia, PA, USA: phone 215- The American Journal of Kidney Diseases home page is 239-3804, fax 215-239-3805, e-mail healthpermissions@ located at www.ajkd.org. The Official Journal of the National Kidney Foundation

EDITORIAL BOARD American Journal of AJKD Kidney Diseases Editor-in-Chief Bertram L. Kasiske, MD Minneapolis, MN

Co-Editors Blanche Chavers, MD Robert Foley, MB Minneapolis, MN Minneapolis, MN Mark Rosenberg, MD Suzanne Swan, MD Minneapolis, MN Minneapolis, MN

Education Editor Pathology Editor Stuart Linas, MD Agnes Fogo, MD Denver, CO Nashville, TN

Associate Editors Sharon Adler, MD Kunitoshi Iseki, MD Yoav Segal, MD Torrance, CA Okinawa, Japan Minneapolis, MN Sharon Andreoli, MD Michel Jadoul, MD Der-Cherng Tarng, MD Indianapolis, IN Brussels, Belgium Taipei, Taiwan Rashad Barsoum, MD Bryce Kiberd, MD Armando Torres, MD Cairo, Egypt Halifax, Canada Tenerife, Spain Gavin Becker, MD Kar Neng Lai, MD Pablo Uren˜a,MD Melbourne, Australia Hong Kong, China Aubervilliers, France Jeffrey Berns, MD Norbert Lameire, MD Stefan Vı´tko, MD Philadelphia, PA Gent, Belgium Prague, Czech Republic Edward Cole, MD Mark Perazella, MD Matthew Weir, MD Toronto, Canada New Haven, CT Baltimore, MD Bengt Fellstro¨m,MD Bernardo Rodrı´guez-Iturbe, MD Andrzej Wiecek, MD Uppsala, Sweden Maracaibo, Venezuela Katowice, Poland Ulrich Frei, MD Gerald Schulman, MD Carmine Zoccali, MD Berlin, Germany Nashville, TN Reggio Calabria, Italy Ronald Hogg, MBChB Dallas, TX

Statistical Reviewers

David Gilbertson, PhD Jon Snyder, MS Minneapolis, MN Minneapolis, MN

Editorial Office Minneapolis, MN

Managing Editor Assistant Managing Editor Editorial Assistant Deanna Gunderson Anna Gillette Benjamin Jackson The Official Journal of the National Kidney Foundation VOL 46, NO 3, SEPTEMBER 2005

CONTENTS American Journal of AJKD Kidney Diseases

ANNOUNCEMENTS ...... lv

QUIZ PAGE SEPTEMBER 2005 Steve Keiran, Bhagwan Bhimani, and Ashwin Dixit ...... lxi

REVIEWS

Malnutrition in Hemodialysis Patients: What Therapy? Maurizio Bossola, Maurizio Muscaritoli, Luigi Tazza, Stefania Giungi, Antonio Tortorelli, Filippo Rossi Fanelli, and Giovanna Luciani ...... 371

Multinutrient Oral Supplements and Tube Feeding in Maintenance Dialysis: A Systematic Review and Meta-Analysis Rebecca J. Stratton, Gemma Bircher, Denis Fouque, Peter Stenvinkel, Rene´e de Mutsert, Meike Engfer, and Marinos Elia ...... 387

ORIGINAL INVESTIGATIONS

Pathogenesis and Treatment of Kidney Disease and Hypertension

CKD Progression and Mortality Among Older Patients With Diabetes Uptal D. Patel, Eric W. Young, Akinlolu O. Ojo, and Rodney A. Hayward ...... 406

Earlier Diagnosis of Autosomal Dominant Polycystic Kidney Disease: Importance of Family History and Implications for Cardiovascular and Renal Complications Editorial, p. 557 Matthew Taylor, Ann M. Johnson, Maryellyn Tison, Pamela Fain, and Robert W. Schrier ...... 415

Proteinuria and Hemoglobin Levels in Patients With Primary Glomerular Disease Nina Ma¨hr, Ulrich Neyer, Friedrich Prischl, Reinhard Kramar, Gert Mayer, Florian Kronenberg, and Karl Lhotta ...... 424

Host and Bacterial Virulence Factors Predisposing to Emphysematous Pyelonephritis Chin-Chung Tseng, Jiunn-Jong Wu, Ming-Cheng Wang, Lien-I Hor, Yen-Hsiu Ko, and Jeng-Jong Huang ...... 432

Cover Illustration: Monthly Feature—AJKD Quiz Page, p. lxi Absorptive Hyperoxaluria Leads to an Increased Risk of Urolithiasis or Nephrocalcinosis in Cystic Fibrosis Bernd Hoppe, Gerd E. von Unruh, Gesa Blank, Ernst Rietschel, Harmeet Sidhu, Norbert Laube, and Albrecht Hesse ...... 440

The Course of Anticardiolipin Antibody Levels Under Immunoadsorption Therapy Anna-Christine Hauser, Lorenz Hauser, Ingrid Pabinger-Fasching, Peter Quehenberger, Kurt Derfler, and Walter Hermann Ho¨rl ...... 446

Relationship of Phosphorus and Calcium-Phosphorus Product With Mortality in CKD Vandana Menon, Tom Greene, Arema A. Pereira, Xuelei Wang, Gerald J. Beck, John W. Kusek, Alan J. Collins, Andrew S. Levey, and Mark J. Sarnak ...... 455

Dialysis Therapies

Changes in Cardiovascular Calcification After Parathyroidectomy in Patients With ESRD Anthony J. Bleyer, John Burkart, Michael Piazza, Gregory Russell, Michael Rohr, and J. Jeffrey Carr ...... 464

The Hemodynamic Effect of Calcium Ion Concentration in the Infusate During Predilution Hemofiltration in Chronic Renal Failure Nikolaos Karamperis, Erik Sloth, and Jens Dam Jensen ...... 470

Epoetin Alfa Use in Patients With ESRD: An Analysis of Recent US Prescribing Patterns and Hemoglobin Outcomes Allan J. Collins, Robert M. Brenner, Joshua J. Ofman, Eric M. Chi, Nina Stuccio-White, Mahesh Krishnan, Craig Solid, Norma J. Ofsthun, and J. Michael Lazarus ...... 481

Association of Morbid Obesity and Weight Change Over Time With Cardiovascular Survival in Hemodialysis Population Kamyar Kalantar-Zadeh, Joel D. Kopple, Ryan D. Kilpatrick, Charles J. McAllister, Christian S. Shinaberger, David W. Gjertson, and Sander Greenland ...... 489

Tunneled Catheters in Hemodialysis Patients: Reasons and Subsequent Outcomes Timmy Lee, Jill Barker, and Michael Allon ...... 501

Validation of a Method to Predict Required Dialysis Time for Cases of Methanol and Ethylene Glycol Poisoning George M. Youssef and David J. Hirsch ...... 509

Time Profiles of Peritoneal and Renal Clearances of Different Uremic Solutes in Incident Peritoneal Dialysis Patients Bert Bammens, Pieter Evenepoel, Kristin Verbeke, and Yves Vanrenterghem ...... 512

History of Cardiovascular Disease Is Associated With Endothelial Progenitor Cells in Peritoneal Dialysis Patients Sabine Steiner, Georg Schaller, Heidi Puttinger, Manuela Fo¨dinger, Christoph W. Kopp, Daniela Seidinger, Johannes Grisar, Walter H. Ho¨rl, Erich Minar, Andreas Vychytil, Michael Wolzt, and Gere Sunder-Plassmann ...... 520 Transplantation

Cardiovascular Risk and Renal Transplantation: Post Hoc Analyses of the Assessment of Lescol in Renal Transplantation (ALERT) Study Alan G. Jardine, Bengt Fellstro¨m, John O. Logan, Edward Cole, Gudrun Nyberg, Carola Gro¨nhagen-Riska, Søren Madsen, Hans-Hellmut Neumayer, Bart Maes, Patrice Ambu¨hl, Anders G. Olsson, Terje Pedersen, and Hallvard Holdaas ...... 529

The Role of Pretransplantation Renal Replacement Therapy Modality in Kidney Allograft and Recipient Survival Alexander S. Goldfarb-Rumyantzev, John F. Hurdle, John D. Scandling, Bradley C. Baird, and Alfred K. Cheung ...... 537

CASE REPORT

Nephrotic Syndrome and Renal Failure After Allogeneic Stem Cell Transplantation: Novel Molecular Diagnostic Tools for a Challenging Differential Diagnosis Paola Romagnani, Elena Lazzeri, Benedetta Mazzinghi, Laura Lasagni, Stefano Guidi, Alberto Bosi, Calogero Cirami, and Maurizio Salvadori ...... 550

EDITORIAL

Hypertension in Individuals at Risk for Autosomal Dominant Polycystic Kidney Disease: To Screen or Not to Screen? Related Article, p. 415 Ronald D. Perrone and Dana C. Miskulin ...... 557

CORE CURRICULUM IN NEPHROLOGY

Tubulointerstitial Diseases Gregory L. Braden, Michael H. O’Shea, and Jeffrey G. Mulhern ...... 560

KIDNEY BIOPSY TEACHING CASE

Membranoproliferative Injury Pattern in a Renal Allograft Neriman Go¨kden, Michele Rossini, Jayant Kumar, and Agnes B. Fogo ...... 573

AJKD ELECTRONIC PAGES (Available only online at www.ajkd.org)

QUIZ PAGE ANSWERS SEPTEMBER 2005 Steve Keiran, Bhagwan Bhimani, and Ashwin Dixit ...... e31 CASE REPORTS

Tethered Hemodialysis Catheter With Retained Portions in Central Vein and Right Atrium on Attempted Removal Hla Thein and Sharad K. Ratanjee ...... e35

Tubulointerstitial Nephritis and Fanconi Syndrome in Primary Biliary Cirrhosis Marie Lino, Raynald Binaut, Laure-He´le`ne Noe¨l, Natacha Patey, Pierre Rustin, Laurent Daniel, Jeanne Serpaggi, Anne Varaut, Philippe Vanhille, Bertrand Knebelmann, Jean-Pierre Gru¨nfeld, and Fadi Fakhouri ...... e41

New Feature: Articles in Press

Effective December 1, 2004, American Journal of Kidney Diseases initiated a new feature. Articles appear online several weeks before appearing in the print journal. After the final version of a manuscript is accepted for publication and authors have approved proofs, the article is posted on the journal’s website (www.ajkd.org) under “Articles in Press.” During this period, the article is identified by a DOI (eg, 10.XXXX/j.ajkd.2004.04.025) and can be searched and cited by this unique identifier. When the article is assigned to an issue of the journal, it moves from the “Articles in Press” section to the table of contents of that issue. OCTOBER AJKD HIGHLIGHTS

ANNOUNCEMENTS

QUIZ PAGE OCTOBER 2005 Mohammad Vaseemuddin and Mark A. Kraus

ORIGINAL INVESTIGATIONS Pathogenesis and Treatment of Kidney Disease and Hypertension

Associations of Metabolic Syndrome With Inflammation in CKD: Results From the Third National Health and Nutrition Examination Survey (NHANES III) Srinivasan Beddhu, Paul L. Kimmel, Nirupama Ramkumar, and Alfred K. Cheung

Obesity and Prevalent and Incident CKD: The Hypertension Detection and Follow-Up Program Holly Kramer, Amy Luke, Anil Bidani, Guichan Cao, Richard Cooper, and Dan McGee

Delay in Time to Receipt of Thrombolytic Medication Among Medicare Patients With Kidney Disease Britt B. Newsome, David G. Warnock, Catarina I. Kiefe, Norman W. Weissman, Thomas K. Houston, Robert M. Centor, Sharina D. Person, William M. McClellan, and Jeroan J. Allison

Correlation Between the Resistive Index by Doppler Ultrasound and Kidney Function and Histology Ryota Ikee, Shuzo Kobayashi, Noriaki Hemmi, Toshihiko Imakiire, Yuichi Kikuchi, Hidekazu Moriya, Shigenobu Suzuki, and Soichiro Miura

B-Type Natriuretic Peptide (BNP) and Amino-Terminal proBNP in Patients With CKD: Relationship to Renal Function and Left Ventricular Hypertrophy Susan Vickery, Christopher P. Price, R. Ian John, Nasir A. Abbas, Michelle C. Webb, Michelle E. Kempson, and Edmund J. Lamb

Dialysis Therapies

Changes in Medicare Reimbursement and Patient-Nephrologist Visits, Quality of Care, and Health-Related Quality of Life Evelyn K. Mentari, Peter B. DeOreo, Andrew S. O’Connor, Thomas E. Love, Edmond S. Ricanati, and Ashwini R. Sehgal (continued) Body Fat Mass and Serum Leptin Levels Influence Epoetin Sensitivity in Patients With ESRD Jonas Axelsson, Abdul Rashid Qureshi, Olof Heimbu¨rger, Bengt Lindholm, Peter Stenvinkel, and Peter Ba´ra´ny

Quality of Life and Its Determinants of Hemodialysis Patients in Taiwan Measured With WHOQOL-BREF(TW) Shu-Chang Yang, Pei-Wen Kuo, Jung-Der Wang, Ming-I Lin, and Syi Su

Physician-Diagnosed Depression as a Correlate of Hospitalizations in Patients Receiving Long-Term Hemodialysis S. Susan Hedayati, Steven C. Grambow, Lynda A. Szczech, Karen M. Stechuchak, Andrew S. Allen, and Hayden B. Bosworth

Trends in Intravenous Iron Use Among Dialysis Patients in the United States (1994-2002) Wendy L. St. Peter, Gregorio T. Obrador, Tricia L. Roberts, and Allan J. Collins

Reuse-Associated Mortality in Incident Hemodialysis Patients in the United States, 2000 to 2001 Qiao Fan, Jiannong Liu, James P. Ebben, and Allan J. Collins

Medication-Related Problems in Ambulatory Hemodialysis Patients: A Pooled Analysis Harold J. Manley, Carrie A. Cannella, George R. Bailie, and Wendy L. St. Peter

Adequacy of a Vancomycin Dosing Regimen in Patients Receiving High-Flux Hemodialysis Robert E. Ariano, Adrian Fine, Daniel S. Sitar, Stacey Rexrode, and Sheryl A. Zelenitsky

Impact of Brachial-Ankle Pulse Wave Velocity and Ankle-Brachial Blood Pressure Index on Mortality in Hemodialysis Patients Tokuyuki Kitahara, Kumeo Ono, Akiyasu Tsuchida, Hironobu Kawai, Masatoshi Shinohara, Yoshitaka Ishii, Hikaru Koyanagi, Toshiharu Noguchi, Takayuki Matsumoto, Tetsuo Sekihara, Yukiyasu Watanabe, Hideo Kanai, Hideki Ishida, and Yoshihisa Nojima

Effect of Oral Glucose Administration on Serum Potassium Concentration in Hemodialysis Patients Shigeaki Muto, Kiyoko Sebata, Hitomi Watanabe, Fumie Shoji, Yoshie Yamamoto, Michie Ohashi, Takako Yamada, Hisako Matsumoto, Tomoko Mukouyama, Takekazu Yonekura, Sachiko Namiki, and Eiji Kusano

The Value of Serum Antilipoarabinomannan Antibody Detection in the Diagnosis of Latent Tuberculosis in Hemodialysis Patients Theodoros Eleftheriadis, Pashalia Tsiaga, Georgia Antoniadi, Vassilis Liakopoulos, Alexandros Kortsaris, Evagelos Giannatos, Konstantinos Barbutis, Ioannis Stefanidis, and Vassilis Vargemezis

Rosiglitazone Reduces Insulin Requirement and C-Reactive Protein Levels in Type 2 Diabetic Patients on Peritoneal Dialysis Teresa Yuk-Hwa Wong, Cheuk-Chun Szeto, Kai-Ming Chow, Chi-Bong Leung, Christopher Wai-Kei Lam, and Philip Kam-Tao Li (continued) Transplantation

De Novo Congestive Heart Failure After Kidney Transplantation: A Common Condition With Poor Prognostic Implications Krista L. Lentine, Mark A. Schnitzler, Kevin C. Abbott, Leiming Li, Thomas E. Burroughs, William Irish, and Daniel C. Brennan

Rates of Completion of the Medical Evaluation for Renal Transplantation Francis L. Weng, Marshall M. Joffe, Harold I. Feldman, and Kevin C. Mange

CASE REPORTS

Pseudohyperkalemia due to Pneumatic Tube Transport in a Leukemic Patient Paul S. Kellerman and James M. Thornbery

Manganese-Induced Parkinsonism in a Patient Undergoing Maintenance Hemodialysis Takayasu Ohtake, Kousuke Negishi, Kouji Okamoto, Machiko Oka, Kyoko Maesato, Hidekazu Moriya, and Shuzo Kobayashi

Nephrogenic Fibrosing Dermopathy/Nephrogenic Systemic Fibrosis: Report of a New Case With Literature Review Sumanth R. Daram, Cherise M. Cortese, and Bahar Bastani

EDITORIALS

CORE CURRICULUM IN NEPHROLOGY

Diabetes and the Kidney Andrew S. O’Connor and Jeffrey R. Schelling

COCHRANE RENAL GROUP REPORT

KIDNEY BIOPSY TEACHING CASE

An Unusual Finding in the Renal Medulla Lawrence Tsao, David S. Siegel, Susan Zeveloff, Govind Bhagat, Vivette D. D’Agati, and Glen S. Markowitz

CORRESPONDENCE

(continued) AJKD ELECTRONIC PAGES (Available only online at www.ajkd.org)

QUIZ PAGE ANSWERS OCTOBER 2005 Mohammad Vaseemuddin and Mark A. Kraus

CASE REPORTS

Renal Hypersensitivity Vasculitis Associated With Dapsone Edson N. Alves-Rodrigues, Luciano C. Ribeiro, Margareth D. Silva, Arley Takiuchi, Osvaldo C. Rabel-Filho, Dino Martini-Filho, and Cor J.F. Fontes

Glomerular Tip Lesion Associated With Nonsteroidal Anti-Inflammatory Drug–Induced Nephrotic Syndrome Inderpreet Sekhon, Sandeep Munjal, Byron Croker, Richard J. Johnson, and A. Ahsan Ejaz

An Adult With Acute Poststreptococcal Glomerulonephritis Complicated by Hemolytic Uremic Syndrome and Nephrotic Syndrome Tomoko Izumi, Toshitake Hyodo, Yuichi Kikuchi, Toshihiko Imakiire, Tatsuyoshi Ikenoue, Shigenobu Suzuki, Nobuyuki Yoshizawa, and Soichiro Miura

Hepatitis C Virus-Associated Glomerulonephritis Without Hepatitis C Virus in the Blood Hideaki Yamabe, Norio Nakamura, Masayuki Nakamura, Michiko Shimada, Ryuichiro Kumasaka, Takeshi Fujita, Hiroshi Osawa, Ken Okumura, Naoto Oyama, and Akishi Momose INFORMATION FOR AJKD CONTRIBUTORS

he American Journal of Kidney Diseases (AJKD) Authors must submit manuscripts to AJKD elec- T is primarily written by and for practitioners tronically. If an author experiences difficulty submit- who care for patients with kidney disease or hyperten- ting a manuscript, he/she must contact the editorial sion. We welcome articles in the following categories: office for technical assistance (612-347-5809 or (1) Original Investigations that deal with one of [email protected]). DO NOT MAIL three subcategories: pathogenesis and treatment of A DISK AND HARD COPIES OFA MANUSCRIPT kidney disease and hypertension, dialysis therapies, TO THE EDITORIAL OFFICE UNLESS RE- and kidney transplantation. We only accept Original QUESTED. If an author does not receive an e-mail Investigations that focus on clinical research. Studies confirmation of submission within 5 business days, that focus on laboratory measurements are acceptable this means the manuscript has not been received by only if they are directly linked to measurements or the editorial office. It is the responsibility of the outcomes in human subjects. author to notify the editorial office when they do not (2) Editorials about any subject of interest to receive a confirmation of submission. practitioners.* To begin, go to http://ajkd.edmgr.com. First-time (3) Reviews about any clinical or basic science users must click “Register” on the red navigation topic of interest to practitioners.* menu at the top of the screen. EM will send an (4) Systematic Reviews and Meta-Analyses, automatic e-mail with a username and password. which differ from Reviews only in that the literature Detailed guidelines for authors are available at the review must be systematic. A quantitative synthesis of EM site. the results of the systematic review, or meta-analysis, In lieu of a cover letter, authors are required to is preferred whenever possible. complete the Mandatory Checklist, located on the (5) Case Reports, which should be succinct and home page of the EM site, and attach it to their online original and have a single, well-defined message. submission. Authors are also required to fill in the (6) Kidney Biopsy Teaching Cases, which are case Potential Reviewer form, located on the home page of reports that include kidney biopsy material. They should the EM site, and attach it to the online manuscript provide a clearly delineated educational message. submission. (7) Letters to the Editor, which must not exceed Note that no submitted materials, including photomi- 250 words, not including references. See www.ajkd. crographs, will be returned. org for guidelines.* (8) Special Articles covering any topic of interest PREPARING MANUSCRIPT to practitioners that is not included in the aforemen- The AJKD editorial staff prefers that manuscripts tioned categories.* be submitted according to the style requirements (9) Quiz Page, see www.ajkd.org for guidelines. detailed below. However, manuscripts meeting the *Since Editorials and Reviews require selection Uniform Requirements for Manuscripts Submitted to and interpretation of the medical literature, AJKD Biomedical Journals (see www.icmje.org) will be generally requires that authors of these articles, considered for publication; authors may be asked to and in some cases, Letters to the Editor and Special make AJKD-specific style changes should their manu- Articles, not have any financial interest in a com- script be accepted. pany (or its competitor) that makes a product Manuscripts must be double-spaced using 12-pt. discussed in the article. type (preferably Times New Roman) and unjustified All manuscripts submitted to AJKD will first be margins. Pages must be numbered starting with the reviewed by two Editors. Some manuscripts may be title page. rejected after this stage if the papers are deemed unacceptable for publication in AJKD. We will subject Title Page the others to further review by peers. Authors may The title page should include the following: (1) title check the status of their submitted manuscripts online (concise and descriptive); (2) authors’ first and last at www.ajkd.org. names and highest degree; (3) institution where the work was done; (4) position, institution, and location SUBMITTING MANUSCRIPT of all authors; (5) acknowledgment of research sup- port; (6) corresponding author’s name, address, tele- Manuscript submissions are processed via the phone number, fax number, and e-mail; and (7) short internet system Editorial ManagerTM (EM). Go to title (not to exceed 45 characters including spaces), to http://ajkd.edmgr.com for more detailed guidelines. be used as a running head. (continued) Information for Contributors (continued)

Abstract manuscript to be published and provide written in- Manuscripts must include a brief, structured (Back- formed consent before publication. ground, Methods, Results, Conclusion) abstract of not Illustrations and Tables more than 250 words followed by a list of index Authors are responsible for applying for permission words. Abstracts for Reviews, Case Reports, and from a publisher for both print and electronic rights to Special Articles may be unstructured. all borrowed material and are responsible for paying Abbreviations any fees related to applying for these permissions. In Use only standard abbreviations, and expand all addition to sending the signed permission statement(s) abbreviations at first mention. Avoid using abbrevia- to the AJKD editorial office, authors should include tions in titles, abstracts, and running heads. notation of the permission as an addition to the foot- note/legend and must provide complete information Generic Names as to the source. Photographs of identifiable persons Use only generic names for drugs whenever pos- must be accompanied by a signed release that indi- sible; if not possible, include a proprietary name and cates informed consent. the name and location of the drug manufacturer in Figures and tables should be cited in numerical parentheses at first mention. order in the text using Arabic numbering. Each figure Units of Measurement should have a legend, but legends should not appear Throughout the text, most units of measurement on the same page with the figure. Legends should be should be expressed in conventional units with Syste`me typed double-spaced and grouped under the heading International (SI) units given in parenthesis. In figures “Legends” on a separate page at the end of the and tables, use conventional units, with conversion manuscript. factors given in legends or footnotes. Manuscripts Each table should be typed on a separate page, with containing only conventional or SI units will not be the table number and title included above the table on returned, but the editorial office will request that the same page. Include any additional information in authors insert the other type of units if the manuscript footnotes below the table. Use spaces, not vertical is accepted. See the Systeme International (SI) Con- rules, to separate columns. Do not include grid lines. verter under Attention Authors at www.ajkd.org. The figure legends and tables should be included in the manuscript file. Figures should be attached as Examples separate files, in TIFF or EPS format, not embedded in In text: . . .serum creatinine at 3 months was 9.62 the manuscript file. ␮ mg/dL (850 mol/L). . . Acknowledgments In figure legends: If any, acknowledgments should be inserted after To convert hemoglobin in g/dL to g/L, multiply by 10. the manuscript text and before the reference list. Authors are responsible for informing all named indi- In tables: viduals/parties that they are being mentioned in their Serum Component Patient 1 Patient 2 submitted manuscript. References Creatinine 0.6 mg/dL 1.2 mg/dL References should be compiled at the end of the Urea nitrogen 8 mg/dL 18 mg/dL article according to the order of citation in the text, not alphabetically. They should be typed, double-spaced NOTE. To convert serum creatinine in mg/dL to ␮mol/L, multiply by 88.4; urea nitrogen in mg/dL to mmol/L, multiply under the heading REFERENCES. Abbreviations for by 0.357. titles of medical periodicals should conform to those in the latest edition of Index Medicus and should not Human Investigation Studies include periods. Authors must include first and last For all studies in humans, appropriate safeguards page numbers and expand the numbers (131-139, not should be included in the Methods section regardless 131-9). If the article has six or fewer authors, list all; if of country of origin (eg, local Institutional Review more than six, list the first three and then “et al.” Board, Ministry of Health approval). Examples Informed Consent Journal article (list all authors): Avoid any information identifying individual study 1. Richardson D, Barlett C, Will EJ: Optimizing erythropoi- subjects. If identifying information is necessary for etin therapy in hemodialysis patients. Am J Kidney Dis 38:109- scientific purposes, the patient must be shown the 117, 2001 (continued) Information for Contributors (continued)

Journal article (more than six authors): of the page charges. This will be only an estimate 2. Hotta O, Miyazaki M, Furuta T, et al: Tonsillectomy and and the actual invoice for page charges will be steroid pulse therapy significantly impact on clinical remission forwarded after the manuscript is published. in patients with IgA nephropathy. Am J Kidney Dis 38:736-743, 2001 The editorial office has a fiduciary responsibility to the publisher and the National Kidney Founda- Complete book: tion and cannot waive page charges. If authors feel 3. Ahmad S: Manual of Clinical Dialysis. London, England, Science Press Ltd, 1999 that they will not be able to pay page charges, they must notify the editorial office at the time of submis- Chapter of book: sion. 4. Batlle D: Metabolic acidosis, in Greenberg A (ed): Primer on Kidney Diseases (ed 2), chap 9. San Diego, CA, Academic Press, 1998, pp 71-79 Color Reproduction Charges Authors must bear all costs connected with printed Chapter of book that is part of published meeting: 5. Natig JB, Kunkel HG, Gedde-Dahl T Jr: Chain subgroups color illustrations. When a manuscript with color of G globulin, in Killander J (ed): Gamma-Globulins, Proceed- illustrations is accepted, the authors will be contacted ings of the Third Nobel Symposium. New York, NY, Wiley, prior to publication, provided a cost estimate, and 1967, pp 37-54 given a choice of publication in color or black and white. Color reproduction costs and page charges are billed separately. AFTER ACCEPTANCE Proofreading If a manuscript is accepted for publication, the Contributors are provided with proofs via the inter- author may be asked to send the following to the net and are asked to proofread them for typesetting editorial office: errors. Important changes in data are allowed, but 1. A file containing the final version of the manu- authors will be charged for excessive alterations to script (without any highlights or track changes). proofs. Corrections must be returned within 48 hours. 2. Two sets of figures in hard copy, or electronic TIFF or EPS files. The figures must be high resolution Reprints (300 dpi for color or grey scale, 1,200 dpi for black Reprints of articles can be ordered before or after and white line art). Detailed figure/graphic guidelines publication. An order form showing costs of reprints are available at www.ajkd.org. will be e-mailed by the publisher prior to publication. 3. Copyright form signed by at least one of the Individuals wishing to obtain reprints of an article that authors. appears in AJKD can do so by contacting the author at the address given in the Journal. Do not contact the Copyright Transfer editorial office regarding reprints. The copyright will be assigned exclusively to the National Kidney Foundation, including the right to Publication on AJKD Electronic Pages reproduce the article in all forms and media. The Because AJKD receives many more meritorious Publisher will not refuse any reasonable request by papers than can be published in the print edition, some the author for permission to reproduce any of his or case reports may be accepted for publication solely on her contributions following publication in AJKD. Per- the AJKD web site (www.ajkd.org). When this occurs, mission requests must be sent to: General Permissions the editorial office will specify potential electronic Department, W.B. Saunders/Elsevier Science, 625 Wal- publication in the initial decision letter e-mailed to the nut Street, Suite 300, Philadelphia, PA 19106. Tele- authors. These articles will be listed in the printed phone: 215-238-7869; Fax: 215-238-2239, e-mail: table of contents and the author and subject indexes of [email protected]. the printed version. They also will be included in Index Medicus and MEDLINE. Manuscripts pub- Page Charges lished exclusively on the web edition will incur no Manuscripts will be assessed a $75.00 charge per page charges or charges for color figures. They will be page for each printed page exceeding four. One printed subject to the same copyright laws as the printed text page is equivalent to about 2.5 double-spaced edition. A sample citation for electronic edition ar- manuscript pages (12-pt. type), 35 references, or 2 ticles is as follows: tables/figures. Page charges and color reproduction costs are billed separately. The letter of acceptance Smith AB, Jones CD: Article title. Am J Kidney Dis 31(3):E3, e-mailed to the author will provide an estimate 1998 (available www.ajkd.org) (continued) Information for Contributors (continued)

ADDITIONAL INFORMATION: EDITORIAL OFFICE CONTACT INFORMATION: The AJKD web site (www.ajkd.org) contains infor- Bertram L. Kasiske, MD mation about the following: Editor-in-Chief, AJKD Hennepin Faculty Associates Announcement Guidelines 600 HFA Building, Room D508 Conflict of Interest Policy 914 S. 8th Street Cost of Color Reproduction Minneapolis, MN 55404, USA Excess Page Charge Policy Deanna Gunderson, Managing Editor Figure Instructions Anna Gillette, Assistant Managing Editor Letter to Editor Instructions Benjamin Jackson, Technical Support Quiz Page Instructions Telephone: 612-347-7770 Supplement Guidelines Fax: 612-347-4321 Syste`me International (SI) Converter E-mail: [email protected] AJKD ANNOUNCEMENTS Annual Scientific Symposium of the National Kidney submission can be obtained from the EuroPD web site at Foundation of Southern California www.europd.com. The abstract deadline is May 13, 2005. The annual scientific symposium of the National Kid- Contact information: Congress organizers and abstract sub- ney Foundation of Southern California will be held on mission: EuroPD Congress Secretariat, c/o In-Conference Saturday, September 24, 2005, in the Pacific Palms Ltd, 10b Broughton Street Lane, Edinburgh EH1 3LY, Conference Resort located in the City of Industry, 35 United Kingdom; tel: 44 (0) 131 556 9245; fax: 44 (0) 131 miles East of Los Angeles International Airport. This 556 9638; e-mail: [email protected] year’s symposium will address current and future issues dation Bureau: Czech-In, Prague Congress Centre, 5 Kvetna in nephrology and dialysis and will include an afternoon 65, CZ-140 21 Prague 4, Czech Republic; tel: 00420 261 174 plenary session, in which the chief medical officers of 305; fax: 00420 261 174 307; e-mail: [email protected]. Fresenius and DaVita will speak. Participation is encour- 3rd Cachexia Conference aged for all nephrology and allied health care profession- The 3rd Cachexia Conference will be held Decem- als. For more information, contact Dr. Kamyar Kalantar- ber 8-10, 2005 in Rome, Italy. Over 60 internationally Zadeh at [email protected] or visit www.kidneysocal.org. renowned speakers will discuss both basic science and Transplant Immunosuppression 2005: clinical topics with emphasis on management of pa- Improving Recipient Outcome tients with wasting syndrome and state-of-the-art and future potential therapies. This year’s symposium will The congress will be held September 28-October 1, also address issues pertaining to malnutrition, inflam- 2005, at the Radisson Hotel Metrodome on the campus of mation and wasting in chronic kidney disease. The the University of Minnesota. Arthur J. Matas, MD, is the deadline for abstract submission is September 15, chairman for this event. The congress will focus on immu- 2005. For more information, contact Prof. F. Rossi nosuppressive protocols used at different times posttransplan- Fanelli at fi[email protected] or Prof. K. tation, and on efforts to improve posttransplantation patient Kalantar-Zadeh at [email protected] or visit www. and graft survival and to decrease posttransplantation mor- cachexia.or. bidity. To be added to the mailing list contact Continuing Medical Education, University of Minnesota, 190 Mc- 15th International Vicenza Course on Namara Alumni Center, 200 Oak Street SE, Minneapolis, Peritoneal Dialysis MN; or call 612-626-7600; toll free 1-800-776-8636; fax: The 15th International Vicenza Course on Peritoneal 612-626-7766; CME web site: www.cme.umn.edu. Dialysis will be held May 30-June 2, 2006, at the Congress Center Ente Fiera, Vicenza, Italy. The course 7th European Peritoneal Dialysis Meeting on Peritoneal Dialysis is in strict collaboration with the The 7th European Peritoneal Dialysis Meeting International Society of Nephrology, the International (EuroPD7) will be held October 15-18, 2005 at the Prague Society of Peritoneal Dialysis and the industry of the Congress Centre, Prague, Czech Republic. This year’s meet- field. A series of lectures and panel discussions will be ing is being held in association with the Czech Society of held during the course. The deadline for registration is Nephrology and under the patronage of the Czech Ministry May 10, 2006. After May 10th, registration will be of Health. The meeting will comprise symposia, clinical subject to space availability. For more information please practice sessions, free communications and posters, as well contact: Dr. Anna Saccardo, Secretariat of the 15th Inter- as a pre-congress educational course and industry spon- national Vicenza Course on Peritoneal Dialysis; e-mail: sored symposia. Further details and information on abstract [email protected] or tel/fax ϩ39 0 444 753949.

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: p lv lv QUIZ PAGE AJKD SEPTEMBER 2005 A44-year-oldAfrican-American man with a history of hypertension presented with kidney failure. He had been on a crack cocaine binge 3 days before admission. Physical examination findings were unremarkable. His blood urea nitrogen level was 46 mg/dL (16 mmol/L), and serum creatinine level was 9.9 mg/dL (875 ␮mol/L). Serum sodium level was 138 mEq/L (mmol/L); potassium, 4.3 mEq/L (mmol/L); chloride, 103 mEq/L (mmol/L); and total carbon dioxide, 17 mEq/L (mmol/L). Urinalysis showed 100 mg/dL (0.1 g/L) of protein, 100 mg/dL (5.6 mmol/L) of glucose, 5 white blood cells/high-power field, and 1 to 4 red blood cells/high-power field. Serological workup was negative. Renal ultrasound showed no evidence of obstruction. Serum and urine toxicology screens were positive for cocaine only. A kidney biopsy was performed.

Figure 43A. Low-power view of a glomerulus, tu- bules, and interstitium.

Figure 43B. High-power view of a tubule. Figure 43C. High-power view of a tubule under po- larized light.

What is your diagnosis? For the answer, go to www.ajkd.org

Case provided by Steve Keiran, MD, Bhagwan Bhimani, MD, and Ashwin Dixit, MD, University of Louisville, Louis- Originally published online as doi:10.1053/j.ajkd.2005.03.025 on August 3, 2005. ville, KY. © 2005 by the National Kidney Foundation, Inc. If you have an interesting case you would like to submit for doi:10.1053/j.ajkd.2005.03.025 consideration, please go to http://ajkd.edmgr.com to do so.

American Journal of Kidney Diseases, Vol 46, No 3 (September), 2005: p lxi lxi