Development of Renal Disease in People at High Cardiovascular Risk: Results of the HOPE Randomized Study

J Am Soc Nephrol 14: 641 647, 2003 Development of Renal Disease in People at High Cardiovascular Risk: Results of the HOPE Randomized Study JOHANNES F. E. MANN, HERTZEL C. GERSTEIN, QI-LONG YI, EVA M. LONN, BYRON J. HOOGWERF,* ANDREW RASHKOW,* and SALIM YUSUF, on behalf of the HOPE investigators *Department of Cardiology and Population Health Research Institute, McMaster University, Hamilton, Canada; Schwabing General Hospital, Ludwig Maximilians University Munich, Germany; and German Institute for High Blood Pressure Research, Heidelberg, Germany. Abstract. In people with diabetes, renal disease tends to progress from microalbuminuria to clinical proteinuria to renal insufficiency. Little evidence has been published for the nondiabetic population. This study retrospectively analyzed changes of proteinuria over 4.5 yr in the HOPE (Heart Outcomes and Prevention Evaluation) study, which compared ramipril s effects to placebo in 9297 participants, including 3577 with diabetes and 1956 with microalbuminuria. This report is restricted to 7674 participants with albuminuria data at baseline and at follow-up. Inclusion criteria were known vascular disease or diabetes plus one other cardiovascular risk factor, exclusion criteria included heart failure or known impaired left ventricular function, dipstick-positive proteinuria ( 1 ), and serum creatinine 2.3 mg/dl (200 M). Baseline microalbuminuria predicted subsequent clinical proteinuria for the study participants overall (adjusted odds ratio [OR], 17.5; 95% confidence interval [CI], 12.6 to 24.4), in participants without diabetes (OR, 16.7; 95% CI, 8.6 to 32.4), and in participants with diabetes (OR, 18.2; 95% CI, 12.4 to 26.7). Any progression of albuminuria (defined as new microalbuminuria or new clinical proteinuria) occurred in 1859 participants; 1542 developed new microalbuminuria, and 317 participants developed clinical proteinuria. Ramipril reduced the risk for any progression (OR, 0.87; 95% CI, 0.78 to 0.97; P 0.0146). People without and with diabetes who are at high risk for cardiovascular disease are also at risk for a progressive rise in albuminuria. Microalbuminuria itself predicts clinical proteinuria in nondiabetic and in diabetic people. Ramipril prevents or delays the progression of albuminuria. Microalbuminuria is a known risk factor for overt nephropathy and subsequent renal insufficiency in people with diabetes. Estimates of the magnitude of this risk vary widely in people with type 2 diabetes compared with the variation of estimates in people with type 1 diabetes (1,2). The significance of microalbuminuria for renal outcomes in nondiabetic people is even less clear. Indeed, in the latter group, a sequence of events progressing from microalbuminuria to clinical proteinuria and then to subsequent renal insufficiency has not been established. Despite these uncertainties, both nondiabetic and diabetic people at high risk for cardiovascular disease are also at risk for renal insufficiency and renal replacement therapy (3,4). The number of patients with generalized atherosclerosis as well as renal failure is increasing steadily in developed countries (5). We therefore hypothesized that individuals at high risk for cardiovascular disease would also be at high risk for the development of renal disease. Received March 20, 2002. Accepted November 13, 2002. Correspondence to: HOPE Office, McMaster University, Hamilton General Hospital, 237 Barton Street East, Hamilton, ONT L8L 2X2 Canada. Phone: 905-527-4322; Fax: 905-521-1166; E-mail: johannes.mann@kms.mhn.de 1046-6673/1403-0641 Journal of the American Society of Nephrology Copyright 2003 by the American Society of Nephrology DOI: 10.1097/01.ASN.0000051594.21922.99 The HOPE (Heart Outcomes and Prevention Evaluation) study was a prospective trial including a broad spectrum of people with and without a history of diabetes who were at risk for cardiovascular events (6 9). People with a serum-creatinine 2.3 mg/dl (200 mol/l), who had no evidence of clinical proteinuria, were included. The urine albumin/creatinine ratio was measured at baseline and follow-up in all participants. This article examines the role of microalbuminuria as a predictor of clinical proteinuria, the development of new microalbuminuria or proteinuria in people without and with diabetes, and the effects of ramipril on these renal outcomes. Materials and Methods Patients The design, baseline clinical characteristics, and primary outcomes of the HOPE and MICRO-HOPE (Microalbuminuria and Renal Outcomes) studies have been described elsewhere (6 9). In brief, men and women 55 yr of age from 267 centers were included if they had (a) objective evidence of vascular disease or (b) diabetes plus at least one other cardiovascular risk factor (hypertension, dyslipidemia, smoking, microalbuminuria) or evidence of vascular disease. The main exclusion criteria were heart failure, uncontrolled hypertension, intolerance to angiotensin-converting enzyme (ACE) inhibitors or to vitamin E, a serum creatinine level above 200 M (2.3 mg/dl), and dipstick-positive proteinuria ( 1 ). The mean age of included participants was 66 yr, 27% were women, 80% had a history of coronary heart disease, 53% of myocardial infarction, 11% of stroke, 44% of peripheral vascular disease, 38% of diabetes, and 47% of hyperten-

642 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 641 647, 2003 Table 1. Baseline characteristics a Baseline Characteristics Without Baseline MA (n 6055) With Baseline MA (n 1619) P Age (yr) 65.2 (6.4) 66.7 (6.8).0001 Systolic BP (mmhg) 137 (19) 147 (21).0001 Diastolic BP (mmhg) 78 (10) 82 (12).0001 Heart rate 1/min 68 (11) 72 (12).0001 Body mass index 27.7 (4.3) 28.5 (4.7).0001 Female gender, n (%) 1570 (26) 501 (31).0001 History of CAD n (%) 4920 (81) 1140 (70).0001 stroke n (%) 369 (6.1) 156 (9.6).0001 PVD n (%) 2485 (41) 814 (50).0001 hypertension n (%) 2579 (43) 997 (62).0001 diabetes n (%) 2210 (37) 1013 (63).0001 Elevated cholesterol n (%) 4086 (67) 1040 (64) 0.0138 Low HDL n (%) 1095 (18) 324 (20) 0.0759 Current smoking n (%) 772 (13) 269 (17).0001 Medications beta-bl n (%) 2503 (41) 541 (33).0001 aspirin n (%) 4658 (77) 1104 (68).0001 lipid lowering n (%) 1851 (31) 388 (24).0001 diuretics n (%) 784 (13) 340 (21).0001 calcium antagonist n (%) 2703 (45) 844 (52).0001 a Data are given as mean SD or as absolute numbers and percentages in brackets. MA, microalbuminuria; BP, blood pressure; CAD, coronary heart disease; PVD, peripheral vascular disease; HDL high density lipoprotein; Beta-Bl, betablocking agent. P values were calculated by 2 test for categorical variables and by t test for continuous variables. sion. Microalbuminuria was present in 21%. Participants were extensively treated with cardiovascular drugs, 76% received antiplatelet agents, 39% beta-blockers, 47% calcium-channel blockers, 15% diuretics, and 29% lipid-lowering agents. Before randomization, all eligible patients (n 10,576) participated in a run-in phase in which they received 2.5 mg ramipril daily for 7 to 10 d. During run-in, 13 participants were excluded because of a rise in serum-creatinine. After run-in, participants were randomized to treatment with ramipril, vitamin E, or placebo in a double-blind, 2 2 factorial design. Follow-up was 3.5 to 5.5 yr (median 4.5 yr), and the primary outcome measure was the composite of cardiovascular death, myocardial infarction, or stroke. Secondary outcome measures included total mortality, hospitalization for heart failure and revascularizations. Urine albumin and creatinine were measured once at randomization and at the last but one study visit in all participants and at 1 yr in all diabetic participants in four central laboratories. The urine albumin/ creatinine ratio (ACR) was calculated, and a value 2 mg/mmol was defined as microalbuminuria. Participants whose albumin/creatinine ratio was higher than 36 mg/mmol after randomization were asked to provide a 24-h urine, which was assayed in local laboratories for protein and albumin; assays were chosen according to availability at each clinical site. Clinical proteinuria was diagnosed if (a) the urine albumin excretion was 300 mg/d or (b) the urine protein was 500 mg/d or (c) the ACR was 36 mg/mmol and no 24-h urine was available. Individuals with clinical proteinuria and a history of diabetes were classified as having overt nephropathy. All measurements were sent to the project office where the development of nephropathy was centrally adjudicated. Statistical Analyses Data of urine ACR at baseline and at study end was available in 7674 participants. Those data are analyzed in the present report. Of the 9297 subjects randomized to 10 mg/d ramipril or placebo, baseline urine ACR values were missing in 254, and follow-up urines in 1369 (in 698 due to death and 671 due to drop-out). Effects of ramipril were analyzed by intention-to-treat (5). Analyses were performed in all participants and in participants without and with a history of diabetes mellitus separately. The heterogeneity across the strata defined by diabetes was examined by Breslow-day test. In the multivariate model, this was tested further by adding an interaction term in the model. We first analyzed the effects of baseline microalbuminuria as a risk for the development of clinical proteinuria. We then analyzed risk factors for the progression of proteinuria, which we defined as development of new microalbuminuria and new clinical proteinuria (considered to represent overt nephropathy in people with diabetes). Based on the urine ACR, people were categorized into two groups with and without microalbuminuria (9,10). Baseline characteristics of the participants were compared between those with and without microalbuminuria at baseline using 2 test for categorical and t test for continuous variables. Age, waist-to-hip ratio, glycated hemoglobin, and BP were treated as continuous variables. Odds ratios (OR) and their 95% confidence interval (CI) were calculated to measure the effect of each risk factor on the outcome. Logistic regression analyses were performed to calculated adjusted OR and to identify independent risk factors. All baseline characteristics (Table 1), treatment allocation to ramipril, and BP change during the study were controlled in the statistical models. To identify independent risk factors, the stepwise

J Am Soc Nephrol 14: 641 647, 2003 Renal Disease Progression in People at Cardiovascular Risk 643 selection approach was applied to construct the final model in which only the variables with a P-value less than 0.05 were retained. We could only analyze participants with baseline and follow-up urinalysis, and results may be biased; therefore, we did two sensitivity analyses with extreme assumptions. In one analysis, participants with no follow-up urinalysis were considered as having developed clinical proteinuria; in the second analysis, participants with no follow-up urinalysis were considered as having not developed clinical proteinuria. Results of these sensitivity analyses were basically not different from the main analysis and are therefore not shown. All analyses were done using SAS 6.12 for UNIX. Results Microalbuminuria as a Predictor of Clinical Proteinuria or Overt Nephropathy Table 1 outlines the demographic characteristics of people with and without microalbuminuria at baseline. People with microalbuminuria were more likely to be older, have a higher BP, have diabetes, have less vascular disease, and be treated with diuretics and calcium anatgonists (and less likely be treated with beta-blockers and aspirin). As noted in Table 2, microalbuminuria predicted clinical proteinuria in trial participants regardless of a history of diabetes. This risk was independent of other cardiovascular risk factors, including hypertension. Approximately 5% of microalbuminuric participants with no history of diabetes and 20% of microalbuminuric participants with previous diabetes developed clinical proteinuria or overt nephropathy. In participants with microalbuminuria, the risk for clinical proteinuria/ overt nephropathy was about 15- to 20-fold higher than in participants without microalbuminuria. In two sensitivity analyses, assuming that all participants with no follow-up urinalysis did or did not develop clinical proteinuria, microalbuminuria was still a significant predictor of clinical proteinuria, but its impact became less important when all missing values were taken as a positive outcome (data not shown). When urine albumin excretion rates were divided into quartiles (Table 3), there was an increasing risk for clinical proteinuria with increasing urine albumin even below the currently accepted definition of microalbuminuria. The risk for clinical proteinuria increased dramatically in the highest quartile, which included people with the currently accepted definition of microalbuminuria. After removal from the analysis of participants with baseline microalbuminuria, the trend was still significant and was of borderline significance after removal of those with baseline microalbuminuria and controlling for multiple baseline variables. Progression of Proteinuria (Development of New Microalbuminuria and New Clinical Proteinuria) Any progression of proteinuria occurred in 1859 participants (24%); this includes progression from normal urine albumin excretion to either microalbuminuria or to clinical proteinuria or the progression from microalbuminuria to clinical proteinuria. The rate of progression was higher in diabetic than in nondiabetc participants (34% versus 17%; P 0.001). New microalbuminuria developed in 1542 of 6055 participants (25.5%, comprising 38.2% of the diabetic and 18.1% of the nondiabetic participants). New clinical proteinuria developed in 317 of 7674 participants (4.1%, comprising 8% of the diabetic and 1% of the nondiabetic participants). The association of several covariates with the risk of progression of proteinuria was tested (Table 4). Many of the traditional cardiovascular risk factors were, in this univariate analysis, associated with progression of proteinuria with the exception of diastolic BP, history of stroke, and changes in serum cholesterol. Multivariate analysis (Table 5) indicated that diabetes was most strongly associated with progression of proteinuria, whereas the association with current smoking, hypertension, male gender, and peripheral vascular disease was less strong. Renal disease develops over many months to years, so the longer follow-up the higher the chance to develop progression of proteinuria. There was heterogeneity across risk strata defined by diabetes. A Breslow-day test yielded a positive interaction between albuminuria and a history of diabetes (P 0.0398). After adjustment for the covariates (Table 6) the P-value was marginally significant (P 0.0907). Table 2. Risk of clinical proteinuria in people with microalbuminuria at baseline (MA ) a OR (95% CI) MA MA Unadjusted Adjusted 1 Adjusted 2 (95% CI) b (95% CI) c All participants, n 7674 271/1619 46/6055 26.3 26.2 17.5 (16.7%) (0.8%) (19.1 to 36.1) (19.1 to 36.1) (12.6 to 24.4) DM history n 3223 231/1013 33/2210 19.5 19.4 18.2 (22.8%) (1.5%) (13.4 to 28.3) (13.4 to 28.2) (12.4 to 26.7) No DM history n 4451 40/606 13/3845 20.8 20.9 16.7 (6.60%) (0.34%) (11.1 to 39.2) (11.1 to 39.4) (8.6 to 32.4) a OR, odds ratio; CI, confidence interval; DM, diabetes mellitus; MA, microalbuminuria. b Adjusted for randomization to ramipril. c Adjusted for age, gender, smoking, hypertension, dyslipidemia, DM, abdominal obesity, and renal insufficiency in all participants; of DM duration, use of oral hypoglycaemic agents or insulin, and for glycated hemoglobin level in people with a history of DM. Adjusted and unadjusted risks were significantly higher (P 0.001) in those with microalbuminuria as compared to those without.

644 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 641 647, 2003 Table 3. Risk of clinical proteinuria by degree of albuminuria a Quartile 1 2 3 4 P-trend 1* P-trend 2* P-trend 3* Alb/Creat 0.21 0.21, 0.54 0.54, 1.57 1.57 All participants, 1.00 1.00 1.91 32.36 0.0001 0.0031 0.0837 n 7674 (0.42 to 2.41) (0.89 to 4.12) (17.16 to 61.02) DM history n 3223 No DM history n 4451 1.00 0.80 1.42 20.32 0.0001 0.1023 0.2362 (0.27 to 2.41) (0.57 to 3.50) (9.52 to 43.42) 1.00 1.37 1.72 26.76 0.0001 0.1153 0.2672 (0.31 to 6.14) (0.38 to 7.68) (8.27 to 86.63) a Urinary albumin/creatinine ratios (Alb/Creat) at baseline were subdivided into quartiles. Relative risks of outcomes with reference to the first quartile are tabulated as odds ratios (OR). The fourth quartile includes participants with microalbuminuria (Alb/Creat 2 mg/ mmol) as well as slightly lesser degrees of albuminuria; dipstick positive proteinuria was an exclusion criteria. P for trend 1*: after adjusting for randomization to ramipril. P for trend 2*: after removing people with microalbuminuria at baseline. P for trend 3*: after removing people with microalbuminuria at baseline, with multiple baseline variables controlled for (see Materials and Methods). Table 4. Association between covariates and risk of progression of proteinuria (new clinical proteinuria new microalbuminuria) a Odds Ratio (95 Confidence Interval) All Participants n 1859/7674 DM History n 1109/3223 No DM History n 750/4451 Age (yr) d 1.12 (1.06 to 1.17) c 1.07 (0.99 to 1.15) 1.22 (1.13 to 1.32) c Systolic BP d 1.08 (1.06 to 1.11) c 1.04 (1.01 to 1.08) b 1.06 (1.01 to 1.120) c Diastolic BP d 1.03 (0.98 to 1.08) 0.99 (0.92 to 1.06) 1.02 (0.95 to 1.11) Heart rate 1/min d 1.16 (1.10 to 1.21) c 1.03 (0.96 to 1.10) 1.05 (0.98 to 1.13) Body mass index d 1.42 (1.26 to 1.59) c 1.09 (0.93 to 1.26) 1.32 (1.09 to 1.61) c Female gender 0.86 (0.76 to 0.96) c 1.09 (0.94 to 1.27) 0.98 (0.81 to 1.20) History of CAD 0.66 (0.58 to 0.74) c 1.09 (0.94 to 1.26) 0.85 (0.63 to 1.15) stroke 1.15 (0.94 to 1.40) 1.17 (0.88 to 1.57) 1.21 (0.90 to 1.61) PVD 1.33 (1.20 to 1.48) c 1.16 (1.01 to 1.35) b 1.36 (1.16 to 1.59) c hypertension 1.48 (1.33 to 1.64) c 1.21 (1.05 to 1.41) c 1.40 (1.19 to 1.64) c Elevated cholesterol 0.94 (0.84 to 1.05) 0.97 (0.83 to 1.13) 0.93 (0.79 to 1.10) Low HDL 1.10 (0.96 to 1.25) 1.07 (0.90 to 1.29) 1.06 (0.87 to 1.30) Current smoking 1.19 (1.03 to 1.38) b 1.06 (0.88 to 1.30) 1.24 (0.99 to 1.56) Medications beta-blocker 0.89 (0.80 to 0.99) b 1.20 (1.03 to 1.41) b 0.98 (0.91 to 1.06) aspirin 0.64 (0.57 to 0.72) c 0.96 (0.83 to 1.11) 0.82 (0.65 to 1.04) lipid lowering 0.87 (0.77 to 0.97) c 1.15 (0.96 to 1.36) 0.86 (0.73 to 1.02) diuretics 1.46 (1.27 to 1.68) c 1.21 (1.01 to 1.45) b 1.41 (1.12 to 1.77) c calcium antagonist 1.02 (0.92 to 1.13) 1.00 (0.86 to 1.16) 1.13 (0.96 to 1.32) diabetes % 2.59 (2.33 to 2.88) c a 1859/7674 participants experienced progression of proteinuria. The association of covariates was tested by logistic regression. b P 0.05. c P 0.01. d Defined as increase by 10 (yr, mmhg, beats/min or kg/m 2 ). Further explanations see legend to Table 1.

J Am Soc Nephrol 14: 641 647, 2003 Renal Disease Progression in People at Cardiovascular Risk 645 Table 5. Multivariate analysis of factors that independently and significantly (P 0.05) predict progression of proteinuria (new microalbuminuria and new clinical proteinuria) a Odds Ratio Lower 95% CI Upper 95% CI PH diabetes 2.45 2.18 2.75 Current smoking 1.20 1.03 1.41 PH hypertension 1.20 1.06 1.35 PVD 1.18 1.05 1.32 Male gender 1.17 1.03 1.32 Age (1 yr) 1.02 1.01 1.03 BMI (10 kg/m 2 ) 1.10 1.06 1.14 Systolic BP at baseline 1.20 1.06 1.36 (10 mmhg) Lowering of systolic BP 0.86 0.82 0.90 ( 10 mmhg) Ramipril 0.87 0.78 0.97 Follow-up time (1 yr) 1.18 1.06 1.33 a Multivariate analysis was done by logistic regression, including all variables with a significant result in univariate analysis (Table 4). In addition, the mean lowering of systolic and diastolic BP during the trial and the date of entry (defining follow-up time) was included in the multivariate analysis. The 3 BP values taken during the trial were divided by 3 and subtracted from the baseline value. A change in diastolic BP was not significantly related to progression of proteinuria. PH, past history; PVD, peripheral vascular disease; BMI, body mass index; BP, blood pressure. When progression and death were taken as the outcome, there was no substantial change of the data in Table 5 (data not shown, see Sensitivity Analysis at the end of Materials and Methods). Progression of proteinuria was defined as new microalbuminuria and new clinical proteinuria; therefore, people with microalbuminuria at baseline obviously had less progression than those without microalbuminuria (16.7% versus 26.2%; n 1619 and 6055) because they were already microalbuminuric. Because of this fact and the positive interaction term of diabetes with albuminuria (see above), we repeated the multivariate analysis, excluding people with microalbuminuria at baseline (n 1619) and analyzing those with and without a history of diabetes separately (Table 6). It is obvious from Table 6 that the level of albuminuria below the accepted range of microalbuminuria is a very powerful predictor of progression of proteinuria. The table also suggests that the variables tested have differential effects on progression in people with and without a history of diabetes. However, the formation of subgroups reduces the statistical power to detect differences. Ramipril effectively reduced the risk of any progression of proteinuria by 13% (P 0.0146), the risk of new microalbuminuria by 10% (P 0.046), and the risk of new clinical proteinuria by 22% after adjustment for baseline characteristics (P 0.0495). The beneficial effect of ramipril on progression of proteinuria was independent of BP lowering. The multivariate analysis (Table 5) suggests that the drug provided a 13% reduction in risk of progression. The mean change of systolic BP was also related to progression. The 3.3 mmhg decrease in systolic BP that was observed in the HOPE study (5) may reduce progression of proteinuria by 5%, according to the statistical analysis (Table 5). The mean change of diastolic BP was not related to progression of proteinuria. Discussion The results show that microalbuminuria is a strong risk factor for the development of clinical proteinuria in nondiabetic people at high risk for cardiovascular events. We also confirm the predictive value of microalbuminuria for overt nephropathy in type 2 diabetes. We do not have further nephrologic details of the participants; in some, microalbuminuria may be a sign of underlying endothelial damage, of diabetic nephropathy, or of another undiagnosed renal disease. However, on the basis the high likelihood of generalized atherosclerosis in most participants of the HOPE trial and the exclusion of proteinuric participants, we can assume that the progression from microalbuminuria to clinical proteinuria in the nondiabetic people is suggestive evidence of progressive nephrosclerosis. We do not know, however, whether this progression heralds future renal failure, as is the case in diabetes. Considering the high number of patients with terminal renal failure due to nephrosclerosis (5), the latter point deserves further study. Levels of albuminuria below the typical cutoff values for microalbuminuria also predicted the development of clinical proteinuria/overt nephropathy in both diabetic and nondiabetic people. This is similar to the evidence that albuminuria below the microalbuminuric cutoff predicts cardiovascular events (10). However, the relationship between albumin excretion and cardiovascular risk appeared to be linear (10), whereas there is a much steeper increase in the risk of clinical proteinuria (i.e., renal risk) at albuminuria levels above the microalbuminburia cutoff. This difference is likely related to the fact that the microalbuminuria cutoffs were originally defined on the basis of the risk of nephropathy in people with diabetes (as opposed to the risk for cardiovascular events). Our data therefore confirm the appropriateness of these cutoffs for identifying people at risk for renal disease. Moreover, they suggest that the progression from microalbuminuria to overt nephropathy to renal insufficiency, which is well established for diabetes, may also relevant for people without diabetes. Within the follow-up period of 4 to 6 yr, however, very few cases of progressive renal failure and dialysis occurred (11). Therefore, we cannot exclude that the sequence from microalbuminuria to renal failure is so slow in people without diabetes that clinically significant renal failure will only rarely develop during their limited life expectancy. Progression of albuminuria, defined as a new microalbuminuria or new clinical proteinuria, was found in a substantial number of participants in the HOPE study. In other words, this well-treated population at high risk for cardiovascular disease was also prone to develop signs of renal disease. Progression of albuminuria was found in one of five participants in 4.5 yr, in one of three participants with diabetes, and in one of seven

646 Journal of the American Society of Nephrology J Am Soc Nephrol 14: 641 647, 2003 Table 6. Multivariate analysis of factors that independently and significantly (P 0.05) predict progression of proteinuria (new microalbuminuria and new clinical proteinuria/overt nephropathy) in people without microalbuminuria at baseline (n 6055), divided into those with and without diabetes a Variables Overall OR (95% CI) Diabetes OR (95% CI) No Diabetes OR (95% CI) ACR* Q2 versus Q1 1.30 (1.10 to 1.54) 1.08 (0.84 to 1.39) 1.56 (1.23 to 1.96) Q3 versus Q1 3.26 (2.78 to 3.82) 2.80 (2.23 to 3.53) 3.70 (2.96 to 4.64) Q4 versus Q1 5.57 (4.13 to 7.52) 4.24 (2.80 to 6.41) 7.38 (4.83 to 11.28) PH diabetes 2.66 (2.31 to 3.06) Current smoking 1.34 (1.04 to 1.75) PH hypertension 1.26 (1.11 to 1.44) 1.32 (1.10 to 1.58) 1.31 (1.09 to 1.57) PH CAD 1.35 (1.14 to 1.60) 1.37 (1.14 to 1.65) PVD 1.20 (1.06 to 1.37) 1.23 (1.04 to 1.48) Age (1 yr) 1.03 (1.02 to 1.04) 1.02 (1.01 to 1.04) 1.04 (1.02 to 1.05) BMI (10 kg/m 2 ) 1.18 (1.03 to 1.37) 1.36 (1.09 to 1.69) Systolic BP at baseline (10 mmhg) 1.07 (1.03 to 1.12) 1.07 (1.01 to 1.14) Lowering of systolic BP ( 10 mmhg) 0.88 (0.83 to 0.93) 0.86 (0.79 to 0.92) Ramipril 0.85 (0.75 to 0.97) 0.83 (0.70 to 1.00) 0.84 (0.70 to 1.00) Follow-up time (1 yr) 1.31 (1.14 to 1.50) 1.26 (1.07 to 1.47) 1.40 (1.10 to 1.78) a ACR, urine albumin/creatinine ratio; participants without microalbuminuria at baseline were subdivided into quartiles of ACR at baseline. OR was calculated relative to quartile 1, defined as an OR of 1. CAD, coronary artery disease; PVD, peripheral vascular disease. Blank space indicates no significant result for the respective variable. See legend to Table 5 for further details. without diabetes. In particular, one of three participants with diabetes developed new microalbuminuria, and one of five diabetic participants with microalbuminuria developed overt nephropathy (9). We know relatively little about the development of new microalbuminuria in type 2 diabetes; these data indicate that it develops at a surprisingly high rate. Assuming a constant rate of appearance of new overt nephropathy over time, the HOPE study also confirms that approximately 50% of microalbuminuric people with type 2 diabetes will develop overt nephropathy in 10 yr (9). This finding is of note because our large sample was extensively treated by antihypertensive drugs, aspirin, and beta-blockers, and many received lipidlowering drugs (6 9). Also, only participants with controlled hypertension could be included, and most were normotensive at baseline (7,9). ACE inhibition effectively reduced the progression of albuminuria in all participants as well in the subgroups without and with diabetes mellitus. Interventions that further decrease the risk or even reverse renal disease are clearly needed. In people with more advanced nephropathy, others had shown that the reduction in urinary protein excretion is associated with improved renal function (12,13). Ramipril treatment lead toa2to 3 mmhg lower BP as compared with placebo treatment (7,9). The multivariate analysis suggests that for the prevention of progression of proteinuria, both the lowering of BP and ACE inhibition play a substantial and independent role. This secondary analysis has a number of limitations. Urines were analyzed once at the beginning and at the end of the study. In people with diabetes, there was a further measurement after 1 yr (9) because only in this pre-specified subgroup was albuminuria a secondary outcome measure. A substantial number of participants could not provide follow-up urines, mainly because they had died. It is conceivable or even likely that participants that died during the trial developed new microalbuminuria or new proteinuria more frequently than those who did not die. Increased albuminuria is associated with increased mortality (10). Therefore the present data may underestimate the true rate of progression of proteinuria in a population at high cardiovascular risk. The multivariate analysis is also hampered by the missing urine data. Death and drop-out preempted urine analysis in about 11% of the HOPE trial participants. As a sensitivity analysis, we also investigated the predictive value of microalbuminuria, assuming that all missing values would either be positive or negative for clinical proteinuria. Under these conditions, microalbuminuria was still a significant predictor. This sensitivity analysis indicates some robustness of the data. Urine albumin was measured only once at each time point. Albuminuria is variable and most organizations recommend that two of three measurements need to be positive to establish the presence of microalbuminuria (1,2). Nevertheless, the large sample size and the fact that the ACR was assayed centrally were able to compensate for the variance associated with only one measurement. Moreover, the fact that the regression dilution bias introduced by a single measurement of a risk factor is an underestimate and not an overestimate of the importance of that risk factor provides further indication of the robustness of these observations. In summary, these data show that people at risk for cardiovascular diseases are also at risk for increasing albuminuria. They also show that people both without and with a history of diabetes who have microalbuminuria are at substantial risk of developing clinical proteinuria and that the risk of clinical

J Am Soc Nephrol 14: 641 647, 2003 Renal Disease Progression in People at Cardiovascular Risk 647 proteinuria is not totally confined to levels of albuminuria above the microalbuminuria threshold. Further studies of the prognostic significance of microalbuminuria in a nondiabetic population are clearly warranted, as are studies of additional strategies to prevent or reverse renal damage in people both without and with diabetes. Acknowledgments This paper is dedicated to Eberhard Ritz, MD, on the occasion of his 65 th birthday. This study was supported by the Medical Research Council of Canada grants MT12790 and UI12362, the Ontario Heart Foundation, Aventis, Astra-Zeneca, NEGMA, Natural Source Vitamin E Producers Association, and King Pharmaceuticals. References 1. Dinneen SF, Gerstein HC: The association of microalbuminuria and mortality in non-insulin dependent diabetes mellitus: A systematic overview of the literature. Arch Int Med 157: 1413 1418, 1997 2. Mogensen CE, Keane WF, Bennett PH, Jerums G, Parving HH, Passa P, Steffes MW, Striker GE, Viberti GC: Prevention of diabetic renal disease with special reference to microalbuminuria. Lancet 346: 1080 1084, 1995 3. Kasiske BL: Relationship between vascular disease and ageassociated changes in the human kidney. Kidney Int 31: 1153 1159, 1987 4. Hall WD: Abnormalities of kidney function as a cause and a consequence of cardiovascular disease (Review). Am J Med Sci 317: 176 182, 1999 5. Parfrey PS, Foley RN: The clinical epidemiology of cardiac disease in chronic uremia. J Am Soc Nephrol 10: 1053 1058, 1999 6. The HOPE investigators: The HOPE (Heart Outcomes and Prevention Evaluation) study: the design of a large, simple randomized trial of an ACE inhibitor (ramipril) and vitamin E in patients at high risk of cardiovascular events. Can J Cardiol 12: 127 137, 1996 7. The HOPE investigators: Effects of an ACE inhibitor, ramipril, on cardiovascular events in high risk patients. New Engl J Med 342: 145 153, 2000 8. The HOPE investigators: Vitamin E supplementation and cardiovascular events in high risk patients. New Engl J Med 342: 154 160, 2000 9. The HOPE investigators: Effects of ramipril on cardiovascular and microvascular outcomes in people with diabetes mellitus: Results of the HOPE study and MicroHOPE substudy. Lancet 355: 253 259, 2000 10. Gerstein HC, Mann JFE, Qilong Y, Zinman B, Dineen SF, Hoogwerf B, Halle JP, Young J, Rashkow A, Joyce C, Nawaz S, Yusuf S: Albuminuria and cardiovascular events, death and heart failure in diabetic and non-diabetic individuals. J Am Med Ass 286: 421-426, 2001 11. Mann JFE, Gerstein HC, Pogue J, Bosch J, Yusuf S: Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: The HOPE randomized trial. Ann Int Med 134: 629 636, 2001 12. Maschio G, Alberti D, Janin G, Locatelli F, Mann JFE, Ponticelli C, Ritz E, Zucchelli P: Effects of the ACE inhibitor benazepril on the progression of chronic renal insufficiency. New Engl J Med 334: 939 945, 1996 13. Ruggenenti P, Perna P, Gherardi G, Gaspari F, Benini F, Remuzzi G: Renal function and requirement for dialysis in chronic nephropathy patients on long-term ramipril: REIN follow-up trial. Lancet 352: 1252 1256, 1998 Access to UpToDate on-line is available for additional clinical information at http://www.jasn.org/