Renal Papillary Necrosis-A Sixteen-Year Clinical 1
Matthew D. Griffin, Erik J. Bergstralh, and Timothy S. Larson2
emphasizes that continued recognition of papillary M.D. Griffin, IS. Larson. Division of Nephrology and necrosis will necessitate an awareness of its clinical Internal Medicine. Mayo Clinic and Mayo Foundation. spectrum in the context of current diagnostic and Rochester, MN therapeutic practice.
E.J. Bergstralh. Section of Biostatistics, Mayo Clinic and Key Words: Diabetes mellitus. analgesic abuse, urinary tract Mayo Foundation. Rochester. MN infection, urinan, tract obstruction, sickle cell disease (J. Am. Soc. Nephrol. 1995; 6:248-256) R enal papillary necrosis (RPN) is a term used to describe a particular form of renal damage in ABSTRACT which any or all of the papillae undergo selective This study sought to characterize patients with renal necrosis in a manner that can be clearly demonstrated papillary necrosis seen at one tertiary referral center radiologically or histologically. The clinical syndromes by reviewing medical records of patients with a con- that have been described as accompanying this pro- cess include acute fulminant renal failure, obstruc- firmed diagnosis between January 1 , 1976 and Sep- tion by a sloughed papilla, hematuria, and loin pain, tember 1 , 1992. One hundred sixty-five cases were passage of tissue in the urine, recurrent or severe identified. The mean age at diagnosis was 57 yr (SD urinary tract infection (UT!), and chronic renal failure 15). The female-to-male ratio was 1 .1 :1 .0. Ninety-two (1-8). percent of patients were white. Seventy-seven percent From its first clear description by Friedreich in 1877 of cases were unsuspected before diagnosis, and in association with benign prostatic obstruction (9), 16% were diagnosed at autopsy. The most common accepted relationships have been established between associated conditions were urinary tract infection, RPN and diabetes mellitus, excessive use of analgesic analgesic abuse, urinary tract obstruction, diabetes mixtures, sickle cell disease, and urinary obstruction (6,9-12). The role ofUTI as an independent risk factor mellitus, and sickle cell disease. There was consider- remains unclear, but It is undoubtedly a frequent able overlap in the presence ofthese conditions, with accompaniment ( 1 ,8, 13, 14). Individual reports and two or more identified in 36% of patients. In addition, small collections of cases have suggested associations 1 1 % of patients had none of these well-recognized between RPN and a number of other conditions (15- risks. Other diagnoses in this group included lupus 19). nephritis, Wegener’s granulomatosis, and renal artery The currently available literature on the clinical stenosis. A decline in case numbers of approximately spectrum of RPN draws heavily from autopsy studies 50% was demonstrated over the last 10 yr studied. This (20-24), small groups ofcases from single institutions period was associated with a 57% reduction in the (2,3,25,26), and studies confined to radiologically or number of excretory urograms carried out, suggest- surgically diagnosed cases (27-29). Only a few series ing that changes in diagnostic imaging preference have previously included large numbers of patients or a full clinical spectrum ( 1 ,8,30). Moreover, there are may have contributed. Vital status and renal out- no recent studies of RPN to define the relevance of this come data after diagnosis were obtained in 93% of lesion in modern medical practice, the prognosis over cases. Of those diagnosed while living, survival was a prolonged period from the time of diagnosis, and the lowest among diabetic patients. Ten-year survival for trends that may have occurred in the past two de- nondiabetics was not significantly different from the cades. We therefore sought to described the charac- expected survival of an age- and sex-matched co- teristics of RPN at a single, large, tertiary referral hort. The overall risk for requiring renal replacement center over the past 1 6 yr and observe trends and therapy after the diagnosis of renal papillary necrosis outcome during this period of time. in surviving patients was low (7% of 136 patients at METHODS risk). Diabetes was the only factor associated with an increased risk of end-stage renal failure. This study Chart Review By use ofthe Mayo Clinic’s unique medical records system, a computerized search was made of all patient visits regis- r Received November 22, 1994. Accepted February 8, 1995. tened between January 1976 and September 1992 for diag- 2 Correspondence to Dr. IS. Larson, Division of Nephrology and Internal Med noses of “renal papillary necrosis” on equivalent terms. The clne. Mayo Clinic. 200 FIrst Street, SW, Rochester, MN 55905. chart of each patient identified was reviewed. A radiologic 1046.6673/0602-0248$03.00/0 Journal of the American Society of Nephrology diagnosis was accepted if excretory urognaphy (EXU), netno- CopyrIght C 1995 by the American society of Nephrology grade unognaphy, or both were reported by a consultant
248 Volume 6 - Number 2 ‘ 1995 Griffin et a)
radiologist at our institution as showing “changes of renal Follow-Up papillary necrosis” or “definite renal papillary necrosis. “ Re- An attempt was made to obtain outcome data through ports of ultrasound and computed tomognaphic (CT) scans March 1993 on all patients. Questionnaires were sent out to were also reviewed when available-in no case was a diagno- those without adequate chart information on or after this sis of RPN made for the first time by these imaging modali- date, and phone contact was attempted with nonnesponders ties. Cases with reports stating “possible” on “probable renal after 2 months. papillary necrosis” were not included in the absence of other definite diagnostic evidence. Pathologic diagnosis was ac- Statistical Analysis cepted if examination of passed tissue, a surgical specimen, on autopsy by a consultant pathologist at our institution was Data were analyzed with the SAS software package (SAS reported as showing RPN. Cases with pathologic or radiologic Institute, Cary, NC). Survival curves were constructed by use diagnosis from other Institutions without subsequent review of the Kaplan-Meien method for the overall group and for of diagnostic material at Mayo Clinic were also excluded. Of selected subgroups. An expected survival curve was con- 229 charts reviewed, 165 met the above criteria. Along with structed on the basis of an age- and sex-matched cohort basic patient characteristics, details of diagnostic proce- drawn from the 1980 Minnesota life tables. Survival free of dunes, relevant laboratory studies, additional medical condi- end-stage renal failure was also estimated by the Kaplan- lions, nadiologic and histologic reports, and available fol- Meier method. Differences in survival between groups were low-up were recorded. compared by use ofthe log-rank test. Characteristics of those
Data for analgesic abuse were accepted if a clear history of with Type I versus Type II diabetes were compared with prolonged heavy use of analgesic mixtures, aspirin on phen- Wilcoxon rank sum on tests as appropriate. All tests were acetin alone, on nonstenoidal anti-inflammatory drugs was two sided with a significance level of 0.05. documented with a compatible clinical picture. Obstruction was accepted if structural or functional urinary tract ob- RESULTS stnuction (not due to a sloughed papilla) was clearly docu- The characteristics of the 165 patients who met our mented. UTI was based on the presence of positive urine criteria for RPN are shown in Table 1 . The patients culture at on within 1 month of diagnosis. Details regarding prior single or recurrent pyelonephnitis on simple urinary were predominantly of the white race (92. 1%), with a infections were also recorded when available. Sickle cell small female predominance overall. The diagnosis was disease was accepted with hemoglobin S levels diagnostic of suspected before radiographic or histologic diagnosis the hetenozygous on homozygous state. of RPN in less than one-quarter (23.0%) of patients
TABLE 1. Characteristics of overall group with RPN as well as subgroups with major associated conditions and a group with rare associations or no identified association#{176}
Urinary Sickle Overall Analgesic Diabetes Other Characteristic UTI Tract Cell Group Abuse Mellitus Cases Obstruction Disease
N (% Total Cases) 165 67 (40.6) 59 (35.7) 48 (29.1) 37 (22.4) 5 (3.0) 18 (10.9) Age (yr) (mean ± SD) 57 ± 15.2 60 ± 15.5 55 ± 13.8 58 ± 16.0 56 ± 16.0 53 ± 13.7 51 ± 16.4 F:M 1.1:1 1.9:1 1.6:1 1:1.3 1.5:1 1:1.5 1:1.3 White 92 94 97 94 89 0 90 Suspicion Before 23.0 19.4 37.2 8.3 27.0 40.0 11.1 Diagnosis Diagnostic EXU 57.6 46.3 84.7 33.3 48.6 100 61.1 Diagnostic RU 15.8 11.9 16.9 16.7 27.0 0 11.1 Tissue Passed 4.2 13.4 3.5 0 16.2 0 0 Surgical Specimen 12.7 13.4 5.1 33.3 8.1 0 16.7 Autopsy Diagnosis 15.8 29.9 0 20.8 13.5 0 16.7 Bilateral RPN 63.0 68.7 79.7 45.8 51.4 100 33.3 Sepsis 15.8 26.9 1.7 18.8 21.6 0 16.7 Hypertension 43.0 38.8 49.2 33.3 51 .4 60.0 38.9 History of TCC 6. 1 3.0 1.7 14.6 5.4 0 11.1 History of Nephrolithiasis 27.2 22.4 32.2 45.8 27.2 0 27.7 Serum Creatinine at 230 ± 274 248 ± 221 195 ± 230 230 ± 204 266 ± 266 106 ± 18 230 ± 504 Diagnosis (mean ± SD In mol/Lb) Normal Serum 24.8 16.4 25.4 12.5 24.3 20.0 55.6 Creatinine at Diagnosis a Note: Subgroups for the five major risk factors are not mutually exclusive because combinations were common. Values given are percentages except where otherwise specified. Abbreviations: l?U. retrograde urogram; icc. transitional cell carcinoma. b Conversion factor to milligrams per deciliter = 0.0113.
Journal of the American Society of Nephrology 249 Renal Papillary Necrosis
and was made at the time of autopsy in 26 (15.8%). A Diabetes MeIIitus (n = 37) Analgesic Abuse (n = 59) history of hypertension or nephrolithiasis was fre- quently present (43.0 and 27.3% overall). Bilateral disease was demonstrated in 63.0% of patients. Of those with unilateral disease, involvement of the right kidney was marginally more common (35 versus 26 patients). Mean serum creatinine at diagnosis was 230 mol/L (median, 124; range. 53 to 2,177; SD 274). Forty-one patients (24.8%) had normal serum Urinary Tract Infection (n = 67) Urinary Obstruction (n = 48) creatinine values at the time of diagnosis (upper limits of normal: 106 j mol/L for men, 80 j. mol/L for wom- en). Seventy-three (44.2%) had abnormally high val- ues that were less than twice the upper limit of normal, and the remaining 51 (30.9%) had serum 0 No other risk factor creatinine levels higher than this. The most common 0 i other risk factor diagnostic test was EXU (57.6% of patients), with the . 2 other risk factors diagnosis being made by retrograde urography in 15.8%. Passage of tissue in the urine was recorded Figure 1.Proportions of single or combined risk factors for RPN among patients with the four most common associated predominantly in those with diabetes mellitus and conditions observed. There were 52 cases encountered with UT!. Transitional cell carcinoma of the bladder or two risk factors: OBS+UTI (N = 16), DM+UTI (N = 13), AA+UTI collecting system was diagnosed at the time of diag- (N = 10), DM+OBS (N = 4), AA+OBS (N = 4), DM+AA (N = 4), nosis of RPN or during the follow-up period In 10 UTI+SCD (N = 1), and 8 cases encountered with three risk patients (6.1%). factors: DM+OBS+UTI (N = 3), DM+AA+UTI (N = 2),
DM+SCD+UTI (N = 1), DM+AA+OBS (N = 1), AA+OBS+UTI
Subgroups (N = 1). Abbreviations: DM, diabetes mellitus; AA, analgesic abuse; OBS, urinary tract obstruction; SCD, sickle cell dis- Patients were divided into categories of recognized ease. associated factors as well as a group (N = 18) in which none were clearly demonstrated (Table 1 ). UT! was treated as an Independent risk factor because there (23.8%) had a history of recurrent UT!, 5 (23.8%) had were 2 1 patients in whom no other major factor could a history of nephrolithiasis, and 9 (42.9%) had addi- be clearly identified. There was considerable overlap tional renal conditions that may have contributed to between the subgroups with 52 (3 1 .5%) having two RPN (renal artery stenosis [N = 2], acute tubular major risk factors and 8 (4.8%) having three. Com- necrosis [N = 2], congenital anomalies [N = 2], vascu- bined risks were more prevalent among those with litis [N = 1 1 glomerulonephritis EN = 1 1, interstitial diabetes, UTI, and urinary obstruction (Figure 1). nephritis [N = 1 1’ and surgery for acquired cyst [N = Only analgesic abuse and sickle cell disease were 1 1). Only two patients had, clinically, a single, uncom- present more commonly alone than in the presence of plicated lower UTI. The most common organism spe- another factor. cies cultured were Escherlchla colt, Candlda, and Kieb- Patients With Rare or Uncertain Associations. This siella. group is labeled as “Other Cases” in Table 1 . Addi- Patients With Diabetes Mellitus and RPN. Table 2 tional diagnoses in this group included renal artery shows additional data on the subgroup with diabetes stenosis (N = 1 ), Wegener’s granulomatosis (N = 1), mellitus. As expected. those with Type I were younger systemic lupus erythematosus (N = 1 ), probable im- with a longer duration of diabetes (mean, 29.6 yr munoglobulin A nephropathy (N = 1 ), and cirrhosis in versus 13.2 yr duration) compared with Type II diabe- a patient with a single kidney (N = 1 ). In addition, tes mellitus patients. !n both types, current and/or there was suspicion of analgesic abuse in four pa- past UTI were common as were other risk factors. In tients, a history of nephrolithiasis in three, and a all, only six (three of each type) had no other possible history of prior, but not current, UT! in three. contributing factor identified. Of these, one was on Patients With UT!. Of the 67 patients in whom UT! dialysis at the time of diagnosis and one was ap- was present by urine culture at the time of diagnosis proaching ESRD. Retinopathy and neuropathy were ofRPN, 32 (47.8%) had acute pyelonephritls (clinically commonly present at diagnosis (43.2 and 48.6% over- and/or pathologically), 18 (29.9%) had a history of all, respectively), although more so in Type I (58.3 and recurrent UT!, 46 (68.7%) had one or more additional 66.7%, respectively). This difference did not reach clear predisposing factors for RPN, and 2 1 (47.8%) statistical significance. The presence or absence of had sepsis at the time of presentation. Twenty-one underlying diabetic glomerulosclerosis could not be patients could not be clearly identified as having one accurately determined in this retrospective study be- of the other recognized major associated conditions cause dipstick protelnuria was a common feature of (Figure 1). Nevertheless, of these patients, 8 (31.3%) RPN of any cause and more formal assessment of had acute pyelonephritis, 9 (42.9%) had sepsis, 5 protein or albumin excretion rates were seldom per-
250 Volume 6 ‘ Number 2 ‘ 1995 Griffin et a)
TABLE 2. Additional characteristics of diabetic patients-Type I, Type II, and overall groupa
Characteristic Type I Type II Overall
Number (% Total) 12 (32.4) 25 (67.6) 37
Duration of Diabetes (yr), mean ± SD 29.6 ± 9,9b 3.2 ± 8.0 18.5 ± 11.6
Age at Diagnosis of RPN (yr), mean ± SD 40.3 ± 131b 63.8 ± 10.8 56.2 ± 16.0 Hypertension 25.Oc ,7 52.8 UTI At diagnosis 33.3 56.0 48.6 Pyelonephritis 8.3 24.0 18.9 Past recurrent 41 .7 24.0 29.7 Past single 25.0 32.0 29.7 No history of UTI 33.3 24.0 27.0 Other Major Ri5kd 41 .7 40.0 40.5 No Other Major Risk 25.0 12.0 16.2 Complications Retinopathy 58.3 36.0 43.2 Neuropathy 66.7 40.0 48.6 Diabetic ulcers 16.7 15.0 16.2 Macrovascular Disease 16.7 24.0 21.6 Serum Creatinine at Diagnosis (. mol/L), 319 ± 398 239 ± 177 266 ± 266 mean ± SD % Glycosylated Hemoglobin at Diagnosis No. recorded 6 8 14 Mean ± SD 12.4 ± 2.3 9.7 ± 3.7 10.9 ± 3.3 a Values given are percentages except where otherwise indicated. b p < 0.0001 for Type I versus Type II. C p 0.018 for Type I versus Type II. d Other major risks are analgesic abuse, obstruction, and sickle cell disease.
Obstruction was unilateral in 26 patients (54.2%) and formed. The glycosylated hemoglobin was available in 1 4 cases and was higher in patients with Type I bilateral in 22 (45.8%). diabetes compared with Type II, suggesting poorer control among Type I diabetics (mean, 12.4 versus Trends Over Time 9.7%). Because of the small numbers, this difference Figures 2 and 3 show diagnostic trends from 1976 did not reach statistical significance (P = 0.14). to 1992 by 4-yr time periods. The final period extends Among the 18 patients with diabetes mellitus and UT! to September 1 , 1 992. After an initial rise from the at the time of diagnosis, the most common organisms first to second period, there was a steady decline in the cultured were E. colt (N = 8) and Candida species number of cases diagnosed in the last 10 yr. The
(N = 5). proportion diagnosed at autopsy has remained rela- Group With Analgesic Abuse. Of the 59 patients tively stable. Figure 3 shows that the numbers have with documented heavy use of analgesics, 28 (47.5%) fallen in each of the major subgroups In a similar had used aspirin/phenacetin combinations, 5 (8.5%) fashion. The mean ages at diagnosis for each time had used nonsteroidal anti-inflammatory drugs period were 55.9, 54.6, 59.8, and 63.2 yr. EXU was (NSA!D) as well as aspirln/phenacetin, 6 ( 10.2%) had used in the majority of patients to establish a diagno- used NSAID and aspirin, and 5 (8.5%) had used sis of RPN. In seeking to determine whether a fall In NSAID alone. In eight ( 1 3.6%), the nature of analgesic the number of EXU being ordered was associated with abuse was not clearly documented. In only two pa- the reduction in cases, we found that the number of tients was acetaminophen possibly implicated in com- EXU carried out each year at our institution has also bination with other analgesics. dropped steadily from 10, 155 in 1980 to 4,394 in Group With Urinary Obstruction. Urinary calculi 1993. In contrast, the overall number of registered were the most common cause of obstruction (20 patient visits per year (luring this same period rose [41 .7%J patients). Malignancies accounted for 1 1 pa- from 270,064 to 316,809. !n no patient was ultra- tients (22.9%) (prostate, 3; bladder or ureteral, 6; sound or CT scanning the primary mode of diagnosis. female genital, 2). Other causes included benign pros- Survival and Follow-Up tatic hypertrophy (N = 4), neurogenic bladder (N = 4), structural anomalies (N = 6), chronic benign bladder Vital status information was available by chart re- disease (N = 2), and retroperitoneal fibrosis (N = 1). view up to March 1 ,1993, for 75 patients. Information
Journal of the American Society of Nephrology 251 Renal Papillary Necrosis
70
60 U UVING
U AUTOPSY
50
z 0 40 U. 0 Cl) LU 30
C.)
20
10
I 976-79 I 980-83 1984-87 I 988-92
Figure 2. Total number of cases of RPN diagnosed during life and at autopsy during each 4-yr time period. (The final period extends to September 1992.)
was obtained by questionnaire in 78 of the remaining analysis for sex, age at the time ofdlagnosls (<60 yr or 90 patients, giving an overall follow-up of 92.7%. Of >60 yr), and the presence or absence of each of the the 1 2 for whom Information was not up to date, 9 major diagnostic subgroups showed a significant ef- could not be located and 3 elected not to take part. fect on renal outcome for diabetes only. Ten-year Chart data were available on these 1 2 patients for a survival free of dialysis was 84% for diabetic patients mean of 1 .8 (range, 0 to 9) yr from the time of compared with 95% for nondiabetics (Figure 5). Of the diagnosis. 1 1 patients who received long-term dialysis, either Mortality. Figure 4 shows overall survival from the before or after the diagnosis of RPN, 6 had diabetes time of diagnosis (excluding 26 diagnosed at the time mellitus and 5 had a history ofanalgesic abuse. Two of ofautopsy), expected survival for the whole group, and these patients also had lupus nephritis. survival for diabetic and nondiabetic patients. Ob- served 5- and 10-yr survival rates overall were 86 and DISCUSSION 69% compared with the expected survivals of 93 and 83.5%, respectIvely. The 32 diabetic patients moni- This study provides a description of the clinical tored fared considerably worse, with overall 10-yr features, significant disease associations, and sur- survival of 44%, compared with 77% for 107 patients vival characteristics of a large group of patients found without diabetes. The survival difference from ex- to have RPN at a single institution over a 16-yr period. pected was statistically significant for diabetics but As such, its prime aim is to provide insight into the not for nondiabetics. relevance of this form of renal damage in modern Renal Outcome. Three patients were receiving he- practice. The historical path of RPN, from Its first modialysis at the time of the diagnosis of RPN. An description in 1877, is punctuated by landmark ob- additional 9 (6.6%) of the remaining 136 living pa- servations that have defined our current approach to tients went on to receive chronic hemodialysis during the condition (9-12). Nonetheless, there have been the follow-up period, and 1 received a renal trans- few new data added over the last two decades that plant. Another patient for whom follow-up information might serve to clarify the role that RPN has occupied in was not available was close to requiring dialysis (se- more recent clinical settings. Much of the existing rum creatinine, 850 mol/L) at the time of diagnosis. data is based on autopsy studies, smaller groups of Freedom from end-stage renal failure for patients cases, or surgical or radiologic reviews from the 1 940s surviving 10 yr after diagnosis was 93%. Univariate to 1970s (2,3,7,21-29). In this series, the diagnosis
252 Volume 6 ‘ Number 2 ‘ 1995 Griffin et al
30
I UTI