Article

Prognosis of CKD Patients Receiving Outpatient Nephrology Care in Italy

Luca De Nicola,* Paolo Chiodini,† Carmine Zoccali,‡ Silvio Borrelli,* Bruno Cianciaruso,§ Biagio Di Iorio, Domenico Santoro,¶ Vincenzo Giancaspro,** Cataldo Abaterusso,†† Ciro Gallo,† Giuseppe Conte,* and Roberto Minutolo,* for the SIN-TABLE CKD Study Group‡‡

Summary *Nephrology Division Background and objectives Prognosis in nondialysis chronic kidney disease (CKD) patients under regular and †Unit of Medical nephrology care is rarely investigated. Statistics at the Second University of Naples, Naples, Italy; Design, setting, participants, & measurements We prospectively followed from 2003 to death or June 2010 a ‡ Ն Nephrology Division, cohort of 1248 patients with CKD stages 3 to 5 and previous nephrology care 1 year in 25 Italian outpa- Center of National tient nephrology clinics. Cumulative incidence of ESRD or death before ESRD were estimated using the Research-Institute of competing-risk approach. Biomedicine and Molecular Immunology Hospital, Reggio Results Estimated rates (per 100 patient-years) of ESRD and death 8.3 (95% confidence interval [CI], 7.4 to Calabria, Italy; 9.2) and 5.9 (95% CI 5.2 to 6.6), respectively. Risk of ESRD and death increased progressively from stages 3 §Nephrology Division, to 5. ESRD was more frequent than death in stage 4 and 5 CKD, whereas the opposite was true in stage 3 University Federico II, Naples, Italy; CKD. Younger age, lower body mass index, proteinuria, and high phosphate predicted ESRD, whereas ʈ Nephrology Division, older age, diabetes, previous cardiovascular disease, ESRD, proteinuria, high uric acid, and anemia pre- County Hospital, dicted death (P Ͻ 0.05 for all). Among modifiable risk factors, proteinuria accounted for the greatest contri- Solofra, Italy; bution to the model fit for either outcome. ¶Nephrology Division, University of Messina, Messina, Italy; Conclusions In patients receiving continuity of care in Italian nephrology clinics, ESRD was a more frequent **Nephrology Division, outcome than death in stage 4 and 5 CKD, but the opposite was true in stage 3. Outcomes were predicted Di Venere Hospital, by modifiable risk factors specific to CKD. Proteinuria used in conjunction with estimated GFR refined risk Bari, Italy; ††Division of stratification. These findings provide information, specific to CKD patients under regular outpatient ne- Nephrology, University of Verona, Italy; and phrology care, for risk stratification that complement recent observations in the general population. the ‡‡Italian Society of Clin J Am Soc Nephrol 6: 2421–2428, 2011. doi: 10.2215/CJN.01180211 Nephrology Study Group, “Target BP LEvels in CKD” (see Appendix for the Introduction age and comorbidities modifying the predictive role complete list of The knowledge on the competing risk of the two main of main factors, BP in primis (10–13). Third, intensity Investigators) outcomes of chronic kidney disease (CKD), that is, of nephrology care modifies survival (14). ESRD and death, and on the risk factors underlying Previous studies in referred patients have shown Correspondence: Dr. these outcomes is of paramount importance to put in ESRD rates similar or higher than mortality (10,11,15– Luca De Nicola, Cattedra di Nefrologia - place effective prevention strategies. Community 20); however, the definition of risk factors for these Dip. Gerontologia, studies and analyses made on large health insurance outcomes still remains uncertain. Indeed, in most Geriatria, Mal. databases reported mortality rates remarkably larger studies, information was retrospectively collected, Metabolismo, Seconda than ESRD rates (1–5). However, information on co- and the duration of nephrology care and of CKD Universita`di Napoli, Piazza Miraglia, 80131 horts referred to renal clinics, and particularly in pa- diagnosis, which are main modifiers of the competing Napoli, Italia. Phone/ tients under continuous nephrology care, is scarce. risk of ERSD versus death (14,21), was fairly short or Fax: 39-081-2549409; Specific information on the prognosis and risk fac- unspecified. E-mail: luca.denicola@ unina2.it tors responsible for CKD progression and death in In 2003, we designed the TArget BP LEvels (TABLE) CKD patients followed in the setting of tertiary ne- multicenter cohort study, aimed at identifying risk phrology care is of major relevance for three reasons. factors for ESRD, death, and CV complications in First, these patients represent a selected population adult stage 3 to 5 CKD patients, attending Italian renal with peculiar clinical characteristics with respect to clinics for at least 1 year before the study. The cross- unreferred patients, including younger age, more ad- sectional evaluation revealed a high prevalence of vanced disease, higher burden of cardiovascular (CV) patients not achieving main therapeutic goals (22). In comorbidities, and higher BP (6–9). Second, ESRD this study, we report the prospective follow-up re- and death are predicted by different risk factors, with sults to estimate the competing risks of ESRD and www.cjasn.org Vol 6 October, 2011 Copyright © 2011 by the American Society of Nephrology 2421 2422 Clinical Journal of the American Society of Nephrology

death and to assess the main determinants of these out- ing to their distribution, as assessed by the Shapiro–Wilk comes. The results can be helpful in refining the global risk test. Categorical variables were reported as percentages. profile in CKD patients receiving continuity of care in a Differences in characteristics of patients among the three nephrology clinic. CKD stages were tested by means of one-way ANOVA or Kruskal–Wallis (according to their distribution) and Pear- Study Population and Methods son chi-squared test for continuous and categorical vari- This is a prospective observational study performed in ables, respectively. Cochran–Armitage trend test was used 25 Italian outpatient nephrology clinics exclusively dedi- to compare prevalence of modifiable risk factors across cated to the conservative care of CKD and with predefined stages. Possible heterogeneity of target prevalence among clinical and laboratory protocols. centers was investigated by means of intracluster correla- tion coefficient (31). Median follow-up was estimated by the inverse Kaplan–Meier approach (32). Patients Eligible subjects were all of the consecutive patients To assess prognosis of CKD patients according to stage 3 attending the centers during a 9-month period of 2003 with to 5, we used ESRD and death before ESRD as outcomes. A diagnosis of CKD (low estimated GFR [eGFR] and/or pro- further composite end point included ESRD and death, teinuria persisting for Ն3 months), eGFR Ͻ60 ml/min per whichever occurred first. Because ESRD and death before 1.73 m2 (no substitutive treatment), and a first visit at the ESRD are mutually exclusive events (i.e., the occurrence of nephrology clinic dating back at least 1 year before the either one prevents the occurrence of the other), Kaplan– study visit. Patients with acute kidney injury during the 6 Meier estimates of time to ESRD or death before ESRD are months preceding the study visit were excluded. All of the biased; we therefore calculated the cumulative incidence of patients gave informed consent to the protocol, which was ESRD or death before ESRD using the competing-risk ap- approved by the local ethical committee. proach (33), and stages were compared with the Gray test (34). Incidence of the composite outcome was estimated by standard Kaplan–Meier approach. Data Collection and Definitions To assess the predictive role of the uncontrolled modi- The study visit in 2003 was the starting date of the fiable risk factors, we used ESRD and overall (before and follow-up study. The data were collected by participating after ESRD) death as outcomes. Multivariable Cox propor- nephrologists in anonymous case report forms, filled in at tional-hazards models, stratified for CKD stage and center, each center, and then sent back to the coordinating center were used to estimate event-specific hazard ratios (HRs) for quality checks and analyses. At the study visit, infor- and 95% confidence intervals (CIs). When evaluating over- mation was collected on demographic, clinical, and labo- all survival, ESRD was included as time-dependent cova- ratory data and medical history, including any previous riate; when evaluating time to ESRD, dead subjects were CV event, defined as any event among coronary artery censored at the date of death. For each modifiable risk disease, congestive heart failure, and cerebrovascular and factor, the heterogeneity of predictive role among CKD peripheral vascular disease. Twenty-four–hour urine col- stages was assessed by likelihood ratio test of two CKD lection was repeated if the value of the measured creati- stage-stratified models: one with CKD stage-specific esti- nine excretion rate was outside the 60% to 140% range of mates and one with overall risk factor estimate. Under the the value calculated according to Dwyer and Kenler (23). null hypothesis of no heterogeneity, this statistic follows GFR was estimated by the four-variable Modification of approximately a chi-squared distribution on J-1 (i.e., 3to1) Diet in Renal Disease (MDRD) study equation. degrees of freedom (35). The contribution of each covariate At the study visit, we also collected information on main to the model fit was estimated as percentage reduction of modifiable risk factors that were defined as uncontrolled R2 value of the model resulting, from omitting each vari- according to predefined thresholds (24–30): uncontrolled able in turn from the full model (36). We calculated R2 hypertension (BP, Ն130/80 mmHg), anemia (hemoglobin, values according to the work of Nagelkerke (37). Ͻ11 g/dl), high phosphate (serum phosphate, Ͼ4.6 mg/dl Nonlinear association of the covariates with the end in CKD stage 3 to 4 or Ͼ5.5 mg/dl in CKD stage 5), points was evaluated by restricted cubic splines and as- proteinuria (protein excretion, Ͼ0.5 g/24 h), high choles- sessed by likelihood ratio test (36). Only for proteinuria terol (total cholesterol, Ͼ190 mg/dl), high uric acid (serum was there evidence of a nonlinear association for either uric acid, Ͼ6 in women and Ͼ7 mg/dl in men), and outcome, and a restricted cubic spline was used with four smoking habit (smoking in the last 6 months). All of the knots placed a priori at clinically relevant values (0, 0.5, 1, thresholds were accepted and shared by participating and 3 g/24 h of proteinuria). nephrologists in dedicated meetings of the study group. Because the thresholds we used to categorize risk factors The outcome measures were death and ESRD (either may not be universally accepted, we repeated the analyses dialysis or renal transplant). The follow-up expiration date by using continuous variables in the multivariable Cox was June 30, 2010. Periodic updates were planned at 18, 24, models. The data were analyzed using SAS version 9.2 30, and 36 months after the study visit and yearly thereaf- (SAS Inc., Cary, NC). ter to collect information on outcome, BP, and eGFR. In- formation on other variables was not mandatory. Results Statistical Analyses Characteristics of the Cohort at Study Visit Continuous variables were reported as either the means We studied 1248 out of 1492 patients; 244 patients were and SD or median and interquartile ranges (IQRs) accord- excluded because of acute kidney injury (n ϭ 189) or lack Clin J Am Soc Nephrol 6: 2421–2428, October, 2011 CKD Patients under Regular Nephrology Care, De Nicola et al. 2423

of follow-up information (n ϭ 55). All of the patients were Survival Analysis Caucasian. As reported in Table 1, the study cohort was Median follow-up was 60 months (IQR, 55 to 64). Overall characterized by advanced age (67 Ϯ 14 years), high prev- 363 ESRD events and 307 all-cause deaths (229 before and alence of diabetes (28%), and previous CV disease (32%). 78 after ESRD) were observed. Estimated rates (per 100 Median follow-up in the clinic before the study visit was patient-years) were 8.3 (95% CI, 7.4 to 9.2) for ESRD and 5.9 2.6 years (IQR, 1.4 to 5.6). Nutritional status was generally (95% CI, 5.2 to 6.6) for all-cause death. Figure 2 shows the adequate, as documented by mean body mass index and cumulative incidence of ESRD, death before ESRD, and the mean serum albumin levels. composite outcome by CKD stage. Risk of ESRD or death Therapy at study visit is reported in Table 2. Only one increased progressively from stages 3 to 5 (P Ͻ 0.0001 for Ͻ patient out of five had a salt intake 100 mEq/24 h, either outcome). ESRD was a more frequent outcome than Ͻ whereas a protein intake 0.8 g/kg per 24 h was detected death in stages 4 and 5 CKD, whereas the opposite was in 52% of patients. Inhibition of renin angiotensin system true in stage 3 CKD. (RAS) was a mainstay of pharmacologic intervention, but Multivariable Cox analysis is reported in Table 3. In- only a small fraction of patients were under dual RAS creased proteinuria significantly interacted with CKD blockade. As to diuretic drugs, use of thiazides was spo- stage in the prediction of ESRD (P ϭ 0.02), but not death, radic (6.8% overall), whereas furosemide was given to with HR progressively reducing from stage 3 to 5. No more than one-third of patients with increasing frequency interaction with CKD stage was found for other modifiable and dose from stages 3 to 5. Erythropoiesis-stimulating risk factors for either outcome. Younger age, lower body agents were generally used at low doses: median doses of mass index and high phosphate, as well as proteinuria, darbepoetin and ␣ or ␤ epoetin were equal to 20 ␮g/wk (IQR, 15 to 30) and 4.000 IU/wk (IQR, 4000 to 6000), predicted ESRD, whereas older age, diabetes, CV disease, respectively. Phosphate binders were prescribed to 7.9% ESRD, proteinuria, high uric acid, and anemia predicted patients only. death. When estimating the hierarchy of prognostic factors Figure 1 shows the distribution of the uncontrolled mod- (Table 3), age was the main predictor of death, and among ifiable risk factors by CKD stage. The prevalence of pro- modifiable risk factors, proteinuria accounted for the great- teinuria and anemia rose from stages 3 to 5, whereas poor est contribution to the model fit. Additional analyses did hypertension control represented the main complication, not evidence significant interactions between high BP and with only 12.6% (95% CI, 10.8 to 14.5) of patients showing proteinuria in the prediction of either outcome (data not BP of Ͻ130/80 mmHg. There was no heterogeneity among shown). Adding the use of RAS inhibitors in the Cox the 25 participant clinics in the prevalence of patients with models did not change HRs estimates (data not shown), uncontrolled factors (intracluster correlation coefficient and the use of these drugs did not contribute to explain the ranged from 0.036 to 0.066 for the seven factors examined). variability in ESRD (P ϭ 0.17) or death (P ϭ 0.11).

Table 1. Demographics and clinical characteristics of patients at study visit by CKD stage

Stage 3 Stage 4 Stage 5 P (n ϭ 609) (n ϭ 449) (n ϭ 190) Age, years 66 (14) 67 (13) 67 (14) 0.19 Male gender, % 65.1 51.7 46.3 Ͻ0.01 Previous FU, years 2.3 (1.2 to 5.3) 2.7 (1.5 to 5.2) 3.2 (1.8 to 6.0) Ͻ0.01 BMI, kg/m2 27.4 (4.3) 27.3 (4.9) 27.0 (4.9) 0.52 sAlbumin, g/dl 3.98 (0.47) 3.86 (0.51) 3.90 (0.50) 0.01 Diabetes, % 24.0 31.4 31.1 0.01 Previous CV event, % 28.2 35.9 33.7 0.03 Renal disease, % 0.01 diabetes 11 19 15 hypertension 26 23 20 GN/IN/PKD 30 27 34 other 32 32 31 Systolic BP, mmHg 138 (18) 141 (19) 142 (18) 0.01 Diastolic BP, mmHg 81 (11) 81 (11) 81 (9) 0.86 eGFR, ml/min per 1.73 m2 43.1 (8.8) 22.3 (3.9) 11.8 (2.8) Ͻ0.01 Calcium, mg/dl 9.4 (0.7) 9.2 (0.6) 9.1 (0.7) Ͻ0.01 Phosphate, mg/dl 3.6 (0.7) 4.0 (0.8) 4.5 (0.9) Ͻ0.01 Hemoglobin, g/dl 13.6 (1.6) 12.0 (1.7) 11.4 (1.5) Ͻ0.01 Uric acid, mg/dl 6.1 (1.5) 6.2 (1.8) 6.3 (1.9) 0.35 Cholesterol, mg/dl 199 (36) 201 (43) 191 (43) 0.01 Proteinuria, g/24 h 0.34 (0.10 to 0.90) 0.70 (0.27 to 1.42) 1.0 (0.56 to 2.0) Ͻ0.01

Values are means (SD), medians (interquartile range), or percentages. The P values refer to the trend across stages. Previous FU, follow-up in the clinic before the study visit; BMI, body mass index; CV, cardiovascular; GN, glomerulonephritis; IN, interstitial nephritis; PKD, polycystic kidney disease; eGFR, GFR estimated by the four-variable Modification of Diet in Renal Disease equation; sAlbumin, serum albumin. 2424 Clinical Journal of the American Society of Nephrology

Table 2. Therapeutic regimens in patients at study visit by CKD stage

Stage 3 Stage 4 Stage 5 P (n ϭ 609) (n ϭ 449) (n ϭ 190) LSD, % 17.9 23.5 23.3 0.13 LPD, % 46.6 55.5 65.7 0.01 BP-lowering drugs, n 2.15 (0.98) 2.39 (1.11) 2.38 (1.05) Ͻ0.01 ACEi or ARB, % 74.0 76.0 56.0 Ͻ0.01 ACEi ϩ ARB, % 6.2 5.1 4.2 0.51 Furosemide use, % 25.9 47.7 50.0 Ͻ0.01 Furosemide dose, mg/24 h 25 (12.5 to 50) 50 (25 to 125) 50 (25 to 125) Ͻ0.01 Statin, % 23.3 22.1 17.4 0.22 ESA, % 2.6 16.7 33.7 Ͻ0.01

The values are the means (SD), medians (interquartile range), or percentages. The P values refer to the trend across stages. LSD, low salt diet (Ͻ100 mEq/24 h); LPD, low protein intake (Ͻ0.8 g/kg per 24 h); ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; ESA, erythropoiesis-stimulating agents.

Stage 3 Stage 4 Stage 5

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

Figure 1. | Prevalence of uncontrolled risk factors (as defined in the Study Population and Methods section) in CKD stages 3 (white), 4 (light gray), and 5 (dark gray). The bars show 95% confidence interval values. The P values for trends across stages are as follows: P ϭ 0.69 for hypertension, P Ͻ 0.001 for proteinuria, P ϭ 0.27 for high cholesterol, P ϭ 0.05 for hyperuricemia, P Ͻ 0.001 for anemia, P ϭ 0.12 for high phosphate, and P ϭ 0.12 for smoking.

The adjusted HRs of proteinuria as continuous variable prognostic role of dichotomized variables was mainly de- for ESRD (by CKD stages) and death in the whole cohort pendent on the choice of the cut off level. are shown in Figure 3. With mild to moderate proteinuria, a marked increase in the risk of ESRD was detected in CKD Discussion stage 3 and 4, whereas no predictive role of proteinuria on The first main finding of this study is that in CKD ESRD was detected in stage 5 even in the presence of patients under regular nephrology care, the incidence rate severe proteinuria. On the other hand, a significant non- of ESRD outweighs that of death. This marks a fundamen- linear association between proteinuria level and mortality tal difference in comparison with studies based on the was found in all of the three CKD stages (P ϭ 0.02), the risk general population (1–5). General healthcare and adminis- of death showing no further increment at levels greater trative databases do not provide information on the type of than 0.5 g/24 h. Similar results were found when categor- care, which is relevant information because mortality in ical variables were replaced by continuous variables (Sup- CKD patients treated by nephrologists is lower than that of plemental Table), thus excluding the possibility that the those treated by other specialists (14). In addition, our Clin J Am Soc Nephrol 6: 2421–2428, October, 2011 CKD Patients under Regular Nephrology Care, De Nicola et al. 2425

S egat 3 St ega 4 Sta eg 5 .0 .0 .0 1 ESRD or death 1 1

ESRD

Death before ESRD

0.6 0.8 0.6 0.8 0.6 0.8 ncidence

0.2 0.4 0.2 0.4 0.2 0.4

Cumulative inC

0.0 0.0 0.0

Years 01234567 01234567 01234567 Pts atikt risk 609 584 535 473 379 243 37 5 449 412 327 256 189 87 8 1 190 135 65 28 14 8 1 1

Figure 2. | Cumulative incidence of ESRD, death before ESRD, and the composite outcome (first occurrence of ESRD or death) stratified by chronic kidney disease stage. A competing risk approach was used to evaluate the extent to which the cumulative probabilities of the overall composite end point reflected ESRD or death during follow-up. The number of patients at risk refers to the composite end point.

Table 3. Multivariable Cox model of determinants of ESRD and death with estimated contribution of each determinant to model fit

ESRD Death

2 R2 P R P HR (95% CI) Reduction (%) HR (95% CI) Reduction (%) Age, 5 years 0.94 (0.90 to 0.98) Ͻ0.01 11.2 1.56 (1.46 to 1.66) Ͻ0.01 77.7 Male gender 1.18 (0.94 to 1.48) 0.16 2.4 1.29 (0.99 to 1.67) 0.05 1.2 Body mass index, kg/m2 0.97 (0.95 to 0.99) 0.08 8.7 0.99 (0.97 to 1.02) 0.61 0.1 Diabetes mellitus 1.19 (0.93 to 1.52) 0.17 2.3 1.31 (1.01 to 1.69) Ͻ0.05 1.3 Previous CV event 1.27 (1.00 to 1.62) 0.05 4.4 1.35 (1.06 to 1.71) 0.02 1.8 Smoking 1.01 (0.71 to 1.43) 0.97 Ͻ0.01 1.18 (0.80 to 1.74) 0.41 0.2 BP Ն130/80 mmHg 1.10 (0.77 to 1.55) 0.60 0.30 0.72 (0.50 to 1.04) 0.08 0.9 Hemoglobin Ͻ11 g/dl 1.27 (0.99 to 1.63) 0.05 4.0 1.43 (1.08 to 1.89) 0.01 1.9 Phosphate Ͼ4.6 to 5.5 1.59 (1.18 to 2.15) 0.02 10.2 1.35 (0.93 to 1.95) 0.12 0.7 mg/dl Cholesterol Ͼ190 mg/dl 0.90 (0.72 to 1.12) 0.34 1.1 0.95 (0.75 to 1.20) 0.67 0.1 Uric Acid Ͼ6 to 7 mg/dl 0.88 (0.70 to 1.11) 0.27 1.5 1.30 (1.02 to 1.65) 0.04 1.3 Proteinuria Ն0.5 g/24 h 39.2 1.41 (1.10 to 1.82) 0.07 2.3 stage 3 3.17 (1.76 to 5.72) Ͻ0.01 stage 4 2.02 (1.41 to 2.88) 0.01 stage 5 1.13 (0.73 to 1.76) 0.59 ESRD 1.51 (1.07 to 2.13) 0.02 1.7

The model is stratified for center and CKD stage. CV, cardiovascular; HR, hazard ratio; CI, confidence interval. patients had a CKD diagnosis dating from at least 1 year identical to that recorded in the TABLE cohort (8.3 per 100 before the study visit. This stringent criterion consistently person-years). The greater risk of ESRD was also found in a identifies patients with true CKD. Indeed, prolonging the secondary analysis of another major clinical trial in referred chronicity criterion for CKD definition from 3 to 12 months African-American patients, the African American Study of reduces the number of patients with CKD, by 16% to 51% Kidney Disease and Hypertension (AASK) study (19); this according to the stage, and it is associated with higher rates result suggests that racial differences should not play a major of ESRD and lower mortality (21). role on the competing risk of ESRD versus death in referred In the MDRD study cohort, a large group of referred pa- patients. In contrast with MDRD, with AASK, and with tients who took part in a landmark clinical trial that started in TABLE, three other studies with fairly short nephrology care the early 1990s (18), ESRD was the most common outcome reported a higher incidence of death than of ESRD in patients with a low competing risk of death, a relationship indepen- with moderate to severe CKD (10,15,20). dent of demographic factors, etiology of CKD, level of basal Remarkably, we found that the CKD stage is a relevant eGFR, and proteinuria. Notably, the incidence rate of ESRD prognostic criterion for predicting the occurrence of ESRD (8.4 per 100 person-years) in the MDRD study was almost as an event competing with the risk of death. Indeed, ESRD 2426 Clinical Journal of the American Society of Nephrology

ESRD risk-Stage 3 ESRD risk-Stage 4 p 00.0< 01 0.0

024680 2 4 6 8 0 2 4 68 ProteinuriaProteinuria (g/day)(g/24h) ProteinuriaProteinuria (g/day)(g/24h) ESRD risk-Stage 5 Death risk-All stages p=0. 68 2 0=p 0. 61 0.0 10.0 0.0 1 5.0 2.0 5.0 atio 2.0 5.0 Ratio Hazard R Hazard R 0.5 1.0 0.5 1.0 0.2 0.2 0.2

024680 2 4 6 8 0 2 4 6 8 Proteinuria (g/24h) Proteinuria (g/24h)

Figure 3. | Plots of adjusted hazard ratios and 95% confidence intervals (as indicated by the curvilinear dash lines; the horizontal dash lines represent HR1) for ESRD in the three CKD stages and death for all stages by level of proteinuria as continuous variable (reference level, 0.5 g/24 h). risk outweighed mortality in CKD stages 4 and 5, whereas modifiable risk factors and the two outcomes of ESRD the opposite occurred in CKD stage 3. This finding in our and death. This issue has never been tested in a cohort contemporary CKD cohort differs from observations in the study of patients with a sufficiently long observation MDRD study. In the context of that trial, at all stages of period in nephrology. A relevant finding of our study is CKD and independently of proteinuria, the risk of ESRD that some modifiable determinants, more specific to was uniformly higher than that of death. MDRD study was CKD, are strong predictors of the risk of progression to on the basis of a relatively young population (average age ESRD and also contribute to modify the risk of death 50 years) of nondiabetic patients with a low burden of even though in a less relevant way. These CKD-specific cardiovascular comorbidities (8%), whereas our cohort, risk factors were different for the two outcomes; in which reflects the population now commonly seen in ne- particular, proteinuria and hyperphosphatemia pre- phrology clinics in Europe (38), had an average age of 67 dicted the risk of ESRD, whereas proteinuria, anemia, years and included large fractions of diabetics (28%) and of and high uric acid levels predicted death. Notably, no patients with CV disease (32%). Therefore, the TABLE predictive role on either outcome was observed for tra- study, on the basis of patients reflecting today’s CKD ditional modifiable determinants, such as uncontrolled population on stable nephrology care, confirms that ESRD hypertension, high cholesterol levels, and smoking. This rather than death is the dominant clinical outcome in pa- finding supports the concept that CKD-specific modifi- tients on long term follow-up in nephrology offices. able factors progressively take stage as renal function Whether this finding depends on nephrology care or in- deteriorates, whereas traditional risk factors are domi- trinsic features of the “prevalent” patients in nephrology nant in triggering the initial renal and CV damage (39). clinics remains to be elucidated. The lack of a predictive role for high BP is only appar- The second main objective of our study was to provide ently surprising when considering the growing evidence a global evaluation of the relationship between the main of a J-shaped relationship between BP and outcomes Clin J Am Soc Nephrol 6: 2421–2428, October, 2011 CKD Patients under Regular Nephrology Care, De Nicola et al. 2427

(12,13,40), as well as of the superior predictive role of duria-Hospital); F. Petrarulo, V. Giancaspro (Bari-Di Ve- ambulatory BP measurements versus office BP (41). nere Hospital); M. Strippoli (Bari-University Medical In our study, proteinuria strongly predicted ESRD at School); E. Laraia (Bari-S. Rita Hospital); M. Gallucci, B. stages 3 and 4 but not at stage 5. Conversely, in the recent Gigante (Galatina-Hospital); C. Lodeserto, D. Santese landmark collaborative meta-analysis, no such interaction (Taranto-Hospital); A. Montanaro, R. Giordano (Martina was observed in the general populations cohorts (42). CKD Franca-Hospital); A. Caglioti, G. Fuiano (Catanzaro- patients under regular nephrology care are a highly se- University Medical School); C. Zoccali, G. Caridi, M. Pos- lected, population and the reduced role of proteinuria torino (Reggio Calabria-CNR); V. Savica, P. Monardo when renal function declines might rather suggest that (Messina-Hospital); G. Bellinghieri, D. Santoro (Messina- mechanisms triggered by renal function loss per se, like the University Medical School); and P. Castellino, F. Rap- accumulation of uremic toxins (43), may be of particular isarda, P. Fatuzzo, A. Messina, (Catania-University Medi- relevance at advanced stages of renal disease. Conversely, cal School). we did not find a greater mortality risk in patients with severe proteinuria (Figure 3). This finding is possibly ac- Acknowledgments counted for by the low number of patients with elevated This work was endorsed by the Italian Society of Nephrology proteinuria and the predominant weight of age on mortal- (Gruppo di Studio sul Trattamento Conservativo della Insuffi- ity that blunts the role of the other factors, including pro- cienza Renale Cronica). It was partially supported by a govern- teinuria. In addition to proteinuria, other nontraditional, ment grant from Ministero della Istruzione, Universita`, e Ricerca); modifiable factors such as anemia, hyperphosphatemia, Rome, Italy, to G.C. in 2007. Partial results have been presented in and hyperuricemia played minor but still relevant prog- abstract form at the 2010 ERA-EDTA Congress in Munich (Ger- nostic roles (Table 3), which is in keeping with previous many). observations (24,44–46). Our study has limitations. First, the TABLE cohort was Disclosures formed only by Caucasian patients. Second, CKD patients None. under regular nephrology care represent a selected popu- lation with survival and referral biases; thus, results cannot References be extrapolated to patients not followed in nephrology 1. Go AS, Chertow GM, Fan D, McCulloch CE, Hsu CY: clinics, which unfortunately is the rule rather than the Chronic kidney disease and the risks of death, cardiovascular exception (47). Third, even though we adjusted our anal- events, and hospitalization. N Engl J Med 351: 1296–1305, yses for several potential confounders, the possibility of 2004 residual confounding, including socioeconomic status and 2. Keith DS, Nichols GA, Gullion CM, Brown JB, Smith DH: Longitudinal follow-up and outcomes among a population glycemic control, cannot be excluded. Fourth, analysis of with chronic kidney disease in a large managed care organi- risk factors was based on a single data collection, and we zation. Arch Intern Med 164: 659–663, 2004 cannot exclude effects caused by changes over time that we 3. O’Hare AM, Choi AI, Bertenthal D, Bacchetti P, Garg AX, were unable to assess. Finally, the observational nature Kaufman JS, Walter LC, Mehta KM, Steinman MA, Allon M, McClellan WM, Landefeld CS: Age affects outcomes in does not allow interpretation of results in causal terms. chronic kidney disease. J Am Soc Nephrol 18: 2758–2765, In conclusion, this prospective study points out that in 2007 CKD patients under regular nephrology care in Italy: (1) 4. Hallan S, Astor B, Romundstad S, Aasarød K, Kvenild K, ESRD is more frequent than death in stage 4 and 5 CKD, Coresh J: Association of kidney function and albuminuria but the opposite is true in stage 3; (2) among the main with cardiovascular mortality in older vs younger individuals: The HUNT II study. Arch Intern Med 167: 2490–2496, 2007 modifiable risk factors, proteinuria and high phosphate 5. Foley R, Murray A, Herzog C, McBean A, Eggers P, Collins predict ESRD, whereas proteinuria, high uric acid, and A: Chronic kidney disease and the risk for cardiovascular dis- anemia predict death; and (3) proteinuria must be consid- ease, renal replacement, and death in the United States ered in conjunction with eGFR to refine risk stratification. Medicare population, 1998 to 1999. J Am Soc Nephrol 16: 489–495, 2005 6. John R, Webb M, Young A, Stevens PE: Unreferred chronic kidney disease: A longitudinal study. Am J Kidney Dis 43: Appendix 825–835, 2004 TArget BP LEvels in CKD (TABLE-CKD) Study Group: 7. Patel UD, Young EW, Ojo AO, Hayward RA: CKD progres- L. De Nicola, R. Minutolo, P. Zamboli, F., C. Iodice, S. sion and mortality among older patients with diabetes. Am J Borrelli, P. Chiodini, S. Signoriello, C. Gallo, G. Conte Kidney Dis 46: 406–414, 2005 8. Peralta CA, Shlipak MG, Fan D, Ordon˜ez J, Lash JP, Chertow (Napoli-2° University Medical School); T. Materiale, B. GM, Go AS: Risks for end-stage renal disease, cardiovascular Minale, C. Paglionico (Napoli-ASL NA 1); B. Cianciaruso, events, and death in Hispanic versus non-Hispanic white A. Pota (Napoli-University Medical School Federico II); F. adults with chronic kidney disease. J Am Soc Nephrol 17: Nappi, F. Avella (Nola-Hospital); B.R. Di Iorio, V. Bellizzi 2892–2899, 2006 (Solofra-Hospital); R. Cestaro (Sapri-Hospital); V. 9. Minutolo R, De Nicola L, Zamboli P, Chiodini P, Signoriello G, Toderico C, Arfe`G, Boschi G, Brancati C, Iaccarino P, Martignetti, L. Morrone (Benevento-Hospital); A. Lupo, C. Conte G: Management of hypertension in patients with CKD: Abaterusso (Verona-University Medical School); C. Dona- Differences between primary and tertiary care settings. Am J dio (Pisa-University Medical School); M. Bonomini, V. Kidney Dis 46: 18–25, 2005 Sirolli (Chieti-University Medical School); F. Casino, T. 10. Conway B, Webster A, Ramsay G, Morgan N, Neary J, Whit- worth C, Harty J: Predicting mortality and uptake of renal Lopez (Matera-Hospital); F. Detomaso, M. Giannattasio replacement therapy in patients with stage 4 chronic kidney (Putignano-Hospital); M. Virgilio, G. Tarantino (Molfetta- disease. Nephrol Dial Transplant 24: 1930–1937, 2009 Hospital); C. Cristofano, S. Tuccillo, S. Chimienti (Man- 11. Obi Y, Kimura T, Nagasawa Y, Yamamoto R, Yasuda K, Sa- 2428 Clinical Journal of the American Society of Nephrology

saki K, Kitamura H, Imai E, Rakugi H, Isaka Y, Hayashi T: 29. Expert Panel on Detection, Evaluation, and Treatment of Impact of age and overt proteinuria on outcomes of stage 3 High Blood Cholesterol in Adults: Executive Summary of the to 5 chronic kidney disease in a referred cohort. Clin J Am Third Report of the National Cholesterol Education Program Soc Nephrol 9: 1558–1565, 2010 (NCEP) Expert Panel on Detection, Evaluation, and Treatment 12. Kovesdy CP, Trivedi BK, Kalantar-Zadeh K, Anderson JE: As- of High Blood Cholesterol In Adults (Adult Treatment Panel sociation of low blood pressure with increased mortality in III). JAMA 285: 2486–2497, 2001 patients with moderate to severe chronic kidney disease. 30. Cianciaruso B, Italian Society of Nephrology: Conservative Nephrol Dial Transplant 21: 1257–1262, 2006 therapy guidelines for chronic renal failure. G Ital Nefrol 20 13. Agarwal R: Blood pressure components and the risk for end- [Suppl 24]: 48–60, 2003 stage renal disease and death in chronic kidney disease. Clin 31. McGraw KO, Wong SP: Forming inferences about some in- J Am Soc Nephrol 4: 830–837, 2009 traclass correlation coefficients. Psychol Methods l: 30–46, 14. Tseng CL, Kern EF, Miller DR, Tiwari A, Maney M, Rajan M, 1996 Pogach L: Survival benefit of nephrologic care in patients 32. Schemper M, Smith TL: A note on quantifying follow-up in with diabetes mellitus and chronic kidney disease. Arch In- studies of failure time. Control Clin Trials 17: 343–346, 1996 tern Med 168: 55–62, 2008 33. Kalbfleisch JD, Prentice RL: The statistical analysis of failure 15. Johnson ES, Thorp ML, Yang X, Charansonney OL, Smith time data, New York: John Wiley, 1980 DH: Predicting renal replacement therapy and mortality in 34. Gray RJ: A class of K-sample tests for comparing the cumula- CKD. Am J Kidney Dis 50: 559–565, 2007 tive incidence of a competing risk. Ann Stat 16: 1141–1154, 16. Agarwal R, Bunaye Z, Bekele DM, Light RP: Competing risk 1988 factor analysis of end-stage renal disease and mortality in 35. Smith CT, Williamson PR, Marson AG: Investigating hetero- chronic kidney disease. Am J Nephrol 28: 569–575, 2008 geneity in an individual patient data meta-analysis of time to 17. Levin A, Djurdjev O, Beaulieu M, Er L: Variability and risk factors for kidney disease progression and death following event outcomes. Stat Med 24: 1307–1319, 2005 attainment of stage 4 CKD in a referred cohort. Am J Kidney 36. Harrell F: Regression modelling strategies with applications to Dis 52: 661–671, 2008 linear models, logistic regression, and survival analysis, New 18. Menon V, Wang X, Sarnak MJ, Hunsicker LH, Madero M, York: Spinger-Varlag, 2001 Beck GJ, Collins AJ, Kusek JW, Levey AS, Greene T: Long- 37. Nagelkerke NJD: A note on a general definition of the coeffi- term outcomes in nondiabetic chronic kidney disease. Kidney cient of determination. Biometrika 78: 691–692, 1991 Int 73: 1310–1315, 2008 38. Sims RJ, Cassidy MJ, Masud T: The increasing number of 19. Alves TP, Wang X, Wright JT, Appel LJ, Greene T, Norris K, older patients with renal disease. BMJ 327: 463–464, 2003 Lewis J for the AASK Collaborative Research Group: Rate of 39. Zoccali C: Traditional and emerging cardiovascular and renal ESRD exceeds mortality among African Americans with hy- risk factors: An epidemiologic perspective. Kidney Int 70: pertensive nephrosclerosis. J Am Soc Nephrol 21:1361–1369, 26–33, 2006 2010 40. Lewis JB: Blood pressure control in chronic kidney disease: Is 20. Hoefield RA, Kalra PA, Baker P, Lane B, New JP, less really more? J Am Soc Nephrol 21: 1086–1092, 2010 O’Donoghue DJ, Foley RN, Middleton RJ: Factors associated 41. Agarwal R, Andersen MJ: Prognostic importance of ambula- with kidney disease progression and mortality in a referred tory blood pressure recordings in patients with chronic kid- CKD population. Am J Kidney Dis 56: 1072–1081, 2010 ney disease. Kidney Int 69: 1175–1180, 2006 21. Eriksen BO, Ingebretsen OC: In chronic kidney disease stag- 42. Levey AS, de Jong PE, Coresh J, Nahas ME, Astor BC, Matsu- ing the use of the chronicity criterion affects prognosis and shita K, Gansevoort RT, Kasiske BL, Eckardt KU: The defini- the rate of progression. Kidney Int 72: 1242–1248, 2007 tion, classification and prognosis of chronic kidney disease: A 22. De Nicola L, Minutolo R, Chiodini P, Zamboli P, Zoccali C, KDIGO Controversies Conference report. Kidney Int 80: Castellino P, Donadio C, Strippoli M, Casino F, Giannattasio 17–28, 2011 M, Petrarulo F, Virgilio M, Laraia E, Di Iorio BR, Savica V. 43. Vanholder R, Baurmeister U, Brunet P, Cohen G, Glorieux G, Conte G on behalf of the TArget Blood Pressure LEvels in Jankowski J: European Uremic Toxin Work Group: A bench Chronic Kidney Disease (TABLE in CKD) Study Group: to bedside view of uremic toxins. J Am Soc Nephrol 19: 863– Global approach to cardiovascular risk in chronic kidney dis- 870, 2008 ease: Reality and opportunities for intervention. Kidney Int 44. Schwarz S, Trivedi BK, Kalantar-Zadeh K, Kovesdy CP: Asso- 69:538–545, 2006 ciation of disorders in mineral metabolism with progression 23. Dwyer J, Kenler SR. Assessment of nutritional status in renal of chronic kidney disease. Clin J Am Soc Nephrol 1: 825– disease. In: Nutrition and the Kidney, edited by Mitch WE, 831, 2006 Klahr S, 2nd Ed., Boston: Little, Brown and Company, 1993, 45. Madero M, Sarnak MJ, Wang X, Greene T, Beck GJ, Kusek pp 61–95 JW, Collins AJ, Levey AS, Menon V: Uric acid and long-term 24. Messerli FH, Frohlich ED, Dreslinski GR, Suarez DH, Aris- outcomes in CKD. Am J Kidney Dis 53: 796–803, 2009 timuno GG: Serum uric acid in essential hypertension: An 46. Levin A, Djurdjev O, Duncan J, Rosenbaum D, Werb R: Hae- indicator of renal vascular involvement. Ann Intern Med 93: moglobin at time of referral prior to dialysis predicts survival: 817–821, 1980 An association of haemoglobin with long-term outcomes. 25. K/DOQI clinical practice guidelines on hypertension and an- Nephrol Dial Transplant 21: 370–377, 2006 tihypertensive agents in chronic kidney disease. Am J Kidney Dis 43[Suppl 1]: S1–S290, 2004 47. Szczech LA: Where Is Walter? If He Isn’t in My Office, Is He 26. National Kidney Foundation: K/DOQI clinical practice guide- Really My Responsibility? The 2011 National Kidney Founda- lines and clinical practice recommendations for anemia in tion Presidential Address. Am J Kidney Dis 57: 529–531, chronic kidney disease. Am J Kidney Dis 47[Suppl 3]: S17– 2011 S27, 2006 27. K/DOQI Clinical Practice Guidelines for Bone Metabolism, & Received: February 7, 2011 Accepted: June 7, 2011 Disease in Chronic Kidney Disease. Am J Kidney Dis 42[Suppl 4]: S1–S201, 2003 Published online ahead of print. Publication date available at 28. Ruggenenti P, Perna A, Remuzzi G. GISEN Group Investiga- www.cjasn.org. tors. Retarding progression of chronic renal disease: The ne- glected issue of residual proteinuria. Kidney Int 63: 2254– Supplemental information for this article is available online at 2261, 2003 www.cjasn.org/. Supplementary Table. Multivariable Cox model (continuous variables) on ESRD and death

ESRD Death P P HR (95% CI) HR (95% CI) Age, 5 yrs 0.94 (0.90-0.98) 0.006 1.56 (1.47-1.67) <0.001 Male gender 1.21 (0.97-1.53) 0.09 1.21 (0.94-1.56) 0.14 Body Mass Index, kg/m2 0.97 (0.95-0.99) 0.01 1.00 (0.97-1.02) 0.78 Diabetes Mellitus 1.14 (0.88-1.46) 0.32 1.29 (1.00-1.67) <0.05 Previous CV event 1.26 (0.99-1.61) 0.06 1.35 (1.06-1.72) 0.02 Smoking 1.02 (0.72-1.45) 0.90 1.18 (0.80-1.74) 0.39 Systolic BP, 5 mmHg 1.00 (0.97-1.03) 0.99 0.97 (0.93-1.00) 0.06 Hemoglobin, g/dL 0.92 (0.86-0.99) 0.04 0.92 (0.85-0.99) 0.03 Phosphate, mg/dL 1.17 (1.03-1.34) 0.02 1.01 (0.86-1.19) 0.88 Cholesterol, mg/dL 1.00 (1.00-1.00) 0.41 1.00 (1.00-1.00) 0.59 Uric Acid, mg/dL 0.95 (0.89-1.01) 0.09 1.09 (1.02-1.16) 0.01 ESRD - - 1.50 (1.07-2.12) 0.02 Proteinuria, g/24h ^ 0.02 0.0 vs 0.5 - - 0.66 (0.46-0.95) 1.0 vs 0.5 - - 1.12 (0.96-1.31) 3.0 vs 0.5 - - 1.19 (0.82-1.73) Proteinuria – Stage 3, g/24h ^ <0.001 0.0 vs 0.5 0.56(0.23-1.39) 1.0 vs 0.5 1.56(1.14-2.13) 3.0 vs 0.5 3.03(1.39-6.59) Proteinuria – Stage 4, g/24h ^ <0.001 0.0 vs 0.5 0.57 (0.31-1.06) 1.0 vs 0.5 1.25 (1.04-1.50) 3.0 vs 0.5 1.83 (1.16-2.88) Proteinuria – Stage 5, g/24h ^ 0.96 0.0 vs 0.5 1.09 (0.55-2.16) 1.0 vs 0.5 1.05 (0.86-1.28) 3.0 vs 0.5 1.15 (0.71-1.85)

Model is stratified for center and CKD stage. ^HRs for proteinuria are derived from restricted cubic splines and for specific values.