Effect of Angiotensin-Converting Enzyme Inhibitors on the Progression of Nondiabetic Renal Disease

Methods

Literature Search

The meta-analysis was performed by using methods described elsewhere [18, 19]. We searched the MEDLINE database for English-language studies on the effect of ACE inhibitors on renal disease in humans that were published between 1977 (the year in which the studies of ACE inhibitors in humans were first published) and May 1996. We also searched references, review articles, and abstracts from recent U.S. and international congresses. In addition, we sent letters to investigators who had experience in conducting trials that studied the effect of antihypertensive agents on the progression of renal disease to ask about other published or unpublished reports.

Selection Criteria

We included data from published or unpublished reports of randomized, controlled studies in which most patients had nondiabetic renal disease. The included studies had a planned length of follow-up of at least 1 year and reported the number of patients who developed end-stage renal disease (measured by initiation of dialysis or transplantation), died, and dropped out. If a series of papers was published by the same authors, all data were retrieved from the most recent report. Studies of patients who predominantly had diabetic renal disease or who had had renal transplantation were excluded. We did not require that patients have hypertension or renal insufficiency at baseline.

Data Extraction

Two authors extracted the data. We defined four outcome measures: end-stage renal disease, death, the combined outcome of end-stage renal disease or death, and drop out. The four outcome measures were summarized for each randomly assigned group in each study. This summary was sent to the principal investigator or co-investigator with a request to verify the data; provide any missing data; and, when applicable, update the results of ongoing studies. We also requested individual patient data for all studies.

The mean systolic and diastolic blood pressures during follow-up were computed as the weighted mean of the blood pressure results reported in published reports or from results provided to us by the investigators. Mean baseline and follow-up blood pressures of the ACE inhibitor and control groups were weighted by the number of patients in each group. We then computed the decline in weighted mean blood pressures from baseline to follow-up in both groups and the difference between groups in the weighted mean decreases in blood pressure.

Statistical Analysis

A two-sided P value less than 0.05 was used to indicate statistical significance. No adjustment was made for multiple comparisons.

Comparison of Randomly Assigned Groups

For each study, we computed the relative risk for each outcome in the ACE inhibitor group compared with the control group. The pooled relative risk for each outcome was computed by using the random-effects model described by DerSimonian and Laird [19]. The random-effects model incorporates both between-study and within-study variability. Heterogeneity in relative risk among studies was assessed by use of the chi-square test. The data are presented as relative risks with 95% CIs.

We also compared the changes in weighted mean blood pressure during follow-up. Some studies did not provide data on the variability of follow-up blood pressures or individual patient data. Thus, we could not compute estimates of variance or CIs for the differences in the decline in blood pressure between groups.

Meta-Regression Analysis

Meta-regression analysis is a statistical method used to determine whether specific factors (covariates) influence the magnitude of the point estimate of the treatment effect across studies [20]. The results are generally reported as slope coefficients and CIs. For the analyses reported here, characteristics of the sample in each study were related to the relative risk for end-stage renal disease in that study. We used univariate linear regression analysis to examine the effect of selected baseline variables and the difference in blood pressure between the randomly assigned groups on the relative risk expressed on a logarithmic scale. These analyses may be less sensitive for detecting associations than are multiple regression analysis of individual patient data in a pooled analysis.

Data Synthesis

Selection and Characteristics of the Studies

Ten published randomized trials with at least 1 year of follow-up were identified [6-17]. Four unpublished studies that met these criteria were identified; three were completed and one was ongoing (Brenner BM, Toto R, van Essen GG, Aurell MA. Personal communications). No data on end-stage renal disease, death, or drop out were available for four studies (three published and one unpublished); thus, these studies were not included. After exclusions, seven published [6-14] and three unpublished studies were included, with a total of 1594 patients. Six studies-three published [6, 12-14] and the three unpublished-were blinded, and four studies were not blinded [7-11]. Investigators from all 10 studies confirmed the number of randomly assigned patients and the number of patients who reached each defined outcome.

Characteristics of the studies and the patient samples are listed in Table 1, Table 2, Table 3, Table 4. The planned length of follow-up ranged from 12 to 48 months (Table 1). Most patients were men (range, 48% to 77%), and the mean age ranged from 44 to 66 years. Mean baseline impairment of renal function was mild in three studies (mean serum creatinine level, 1.0 to 1.8 mg/dL), moderate in five studies (mean serum creatinine level, 2.1 to 3.0 mg/dL), and severe in two studies (mean serum creatinine level, 4.2 to 4.4 mg/dL). In most studies, the proportion of patients with hypertensive nephrosclerosis, glomerular and interstitial diseases, and polycystic kidney disease was similar to that of patients with nondiabetic end-stage renal disease (Table 2) [21]. One study [11], however, included only patients with IgA nephropathy, and another study [14] included only patients with essential hypertension. Two studies included patients with diabetic nephropathy: 13 of 70 patients (19%) in one study [9] and 21 of 583 patients (3.6%) in the other [6]. These patients are included in our analysis.

Table 3 lists blood pressures and data on antihypertensive medications. Enalapril was used in seven studies, and captopril, cilazapril, and benazepril were used in one study each. The control groups received placebo in four studies, β-adrenergic blockers in three studies, calcium-channel blockers in two studies, and an unspecified combination of antihypertensive agents in one study. The same blood pressure target was defined for both ACE inhibitor and control groups in each study, and various medications other than ACE inhibitors, including β-adrenergic blockers, calcium-channel blockers, diuretics, peripheral α-adrenergic blockers, central α-adrenergic agonists, and vasodilators, were added to both groups as needed in an effort to reach the target. Variation in baseline blood pressure was due, in part, to the timing of blood pressure measurements dictated by study protocol (for example, measurement while patients received medications compared with measurement while patients did not receive medications). Weighted mean baseline systolic blood pressures were 150.1 mm Hg in the ACE inhibitor group and 151.0 mm Hg in the control group. Weighted mean baseline diastolic blood pressures were 91.7 mm Hg and 91.9 mm Hg, respectively. In most studies, blood pressure declined from baseline to follow-up. Weighted mean follow-up systolic blood pressures were 138.3 mm Hg in the ACE inhibitor group and 143.6 mm Hg in the control group. Weighted mean follow-up diastolic blood pressures were 84.2 mm Hg and 85.6 mm Hg, respectively. Therefore, the respective declines in weighted mean systolic and diastolic blood pressure were 4.9 and 1.2 mm Hg greater in the ACE inhibitor group than in the control group.

Decline in renal function was estimated from the serum creatinine level in seven studies, creatinine clearance in four studies, and glomerular filtration rate in nine studies (Table 4). The methods for measuring glomerular filtration rate included renal clearance of inulin in one study, plasma clearance of technetium-99m pentetate in two studies, plasma clearance of chromium 51 EDTA in three studies, and renal clearance of iodine¹²⁵ iothalamate in three studies. None of the studies was designed to detect a difference in end-stage renal disease or death. Five studies concluded that ACE inhibitors were more effective than other antihypertensive agents in slowing the decline in renal function. Five studies (including the three unpublished studies) did not find that ACE inhibitors were more effective.

Comparisons of Randomly Assigned Groups

The number of randomly assigned patients in each group who reached the defined outcomes is shown in Table 5. Because of the slow rate of progression of renal disease and the relatively short duration of follow-up, few patients in any study developed renal failure or died.

End-Stage Renal Disease

The relative risk for the onset of end-stage renal disease in the ACE inhibitor group compared with the control group is shown in the left side of the (Figure 1). The reported relative risk for end-stage renal disease ranged from 0.51 to 3.36 and was not significant in any single study. However, the pooled relative risk was 0.70 (CI, 0.51 to 0.97), indicating a statistically significantly lower risk for development of end-stage renal disease in the ACE inhibitor group. The result of the test for heterogeneity of relative risk among studies was not significant (P < 0.75 and >0.5), indicating that treatment effect did not significantly differ among studies. Restricting the analysis to the seven published studies gave similar results (relative risk, 0.65 [CI, 0.45 to 0.94]). In the seven studies that used enalapril, the pooled relative risk for end-stage renal disease was 0.74 (CI, 0.52 to 1.05).

Death

The relative risk for death is shown in the right side of the (Figure 1). The pooled relative risk was not significant (1.24 [CI, 0.55 to 2.83]), and the result of the test for heterogeneity among studies was not significant (P > 0.2). In Maschio and colleagues' study [6], a trend toward an increased risk for death was seen in the ACE inhibitor group (relative risk, 7.55 [CI, 0.95 to 60.0]). In the other studies, the pooled relative risk was 0.89 (CI, 0.36 to 2.17).

Combined Outcome (End-Stage Renal Disease or Death) and Drop Outs

No significant difference was seen in the risk for the combined outcome of end-stage renal disease or death (relative risk, 0.80 [CI, 0.55 to 1.17]). If Maschio and colleagues' study [6] is omitted from the analysis, the risk for end-stage renal disease or death is significantly lower in the ACE inhibitor group (relative risk, 0.70 [CI, 0.52 to 0.94]). The number of drop outs did not significantly differ among studies (relative risk, 1.16 [CI, 0.91 to 1.47]).

Meta-Regression Analysis

We used meta-regression analysis to examine the associations of baseline factors and blood pressure during follow-up with the effect of ACE inhibitors on end-stage renal disease.

Baseline Factors

We related the relative risk for end-stage renal disease (on a logarithmic scale) in each study to the following characteristics of that study: the percentage of male patients; mean patient age; percentage of patients with nephrosclerosis, glomerular diseases, interstitial diseases, polycystic kidney disease, and other or unknown conditions; mean serum creatinine level; mean glomerular filtration rate; mean systolic and diastolic blood pressures; and planned duration of follow-up. It is important to note, however, that with the exception of baseline renal function, these analyses had limited power because the variability in these factors across studies was not great (Table 1, Table 2, Table 3).

Baseline glomerular filtration rate was reported in 8 of 10 studies (Table 1). Using the mean baseline creatinine clearance and serum creatinine level in the remaining two studies, we estimated that the mean baseline glomerular filtration rate was 26 mL/min in Zucchelli and colleagues' studies [7, 8] and 38 mL/min in Maschio and colleagues' study [6]. The regression coefficient of baseline glomerular filtration rate on the logarithm of relative risk for end-stage renal disease was 0.017 (CI, −0.0063 to 0.041). Serum creatinine levels were measured in all 10 studies. The regression coefficient for the association between mean baseline serum creatinine level and the logarithm of relative risk was −0.096(CI, −0.48 to 0.29). These results do not suggest a relation between the mean baseline level of renal function and the treatment effect.

Follow-up Blood Pressure

We next related the relative risk for end-stage renal disease in the ACE inhibitor group in each study (on a logarithmic scale) to the observed difference between groups in the decline in systolic and diastolic blood pressures in that study. For decrease in systolic blood pressure, the regression coefficient was −0.007(CI, −0.13 to 0.11). For decrease in diastolic blood pressure, the regression coefficient was −0.004(CI, −0.21 to 0.20). These analyses do not reveal a statistically significant association between mean blood pressure reduction and the beneficial effect of ACE inhibitors; however, the CIs are too wide to rule out a clinically significant association.

Discussion

Diabetes is the single largest cause of end-stage renal disease in the United States, accounting for approximately 30% of new cases [21]. Over the past decade, many clinical trials have shown that ACE inhibitors are effective in slowing the progression of diabetic renal disease [1-4]. Most patients with end-stage renal disease have other renal diseases; for several reasons, however, progress in identifying effective treatments for these other diseases has been slower. First, the prevalence of other individual renal diseases is lower than that of diabetic renal disease, and the inclusion of patients with various diseases in clinical trials creates a heterogeneous study sample that may not uniformly respond to treatment. Second, unlike diabetes, most renal diseases have no clearly defined stages other than the progressive decline in renal function, which is often used as a surrogate end point in clinical trials. However, differences in methods of measuring renal function can lead to discrepancies in the interpretation of results of clinical trials [22-24]. Third, in most diseases other than diabetes, the mean rate of decline in renal function is slow. Within the limited duration of most clinical trials (2 to 4 years), few patients reach a “hard” end point, such as the onset of end-stage renal disease or death. Thus, most clinical trials did not have sufficient statistical power to determine the effect of the intervention on end-stage renal disease or death.

The Modification of Diet in Renal Disease (MDRD) Study, the largest clinical trial to date in patients with nondiabetic renal disease, recently showed a beneficial effect of a lower-than-usual blood pressure goal in patients with renal insufficiency and proteinuria [25, 26]. In that study, ACE inhibitors were the treatment of choice in both the usual and low blood pressure groups; thus, the effect of ACE inhibitors themselves was not studied. Many clinical trials comparing ACE inhibitors with other antihypertensive agents have now been completed, but the results are not uniform. In the largest clinical trial conducted thus far [6], neither the magnitude of the benefit nor the safety of ACE inhibitor therapy was established conclusively.

We therefore combined data from randomized trials in nondiabetic renal disease and found a significantly lower risk for end-stage renal disease in patients treated with ACE inhibitors. In animals, ACE inhibition slows rather than arrests the progression of renal disease [27]. It is therefore likely that the lower risk that we observed reflects a delay in the onset of end-stage renal disease; this delay is consistent with a slowing of progression rather than full prevention of that outcome. Furthermore, other clinical trials suggest that changes in renal function predict the time to onset of end-stage renal disease [4, 28]. Thus, we infer that the beneficial effect of ACE inhibitors on the development of end-stage renal disease we found in our analysis reflects a slowing in the decline in renal function. Unfortunately, the decline in renal function was not expressed in a similar way in all studies. A pooled analysis of individual patient data could provide an estimate for the magnitude of the slowing of the rate of progression.

We used meta-regression analysis to explore possible associations between the treatment effect and clinical or demographic characteristics. Maschio and colleagues [6] suggested that men and patients with proteinuria benefited most from ACE inhibition. We could not confirm or refute these findings because variability across studies in these characteristics was not great. Other studies have shown that lack of variability may limit the power of meta-regression analysis compared with analysis of individual patient data [20]. Regression analysis that uses individual patient data rather than group data would allow more sensitive evaluation of the magnitude of benefit in subgroups. The mean level of baseline renal function varied widely among studies, but we did not observe a significant relation between mean baseline renal function and relative risk for end-stage renal disease with ACE inhibitor treatment. As in diabetic nephropathy, we suspect that treatment with ACE inhibitors is beneficial in patients who have nondiabetic renal diseases in which the degree of impaired renal function varies widely [1, 2, 4]. In practice, ACE inhibitors could be prescribed early and their use continued throughout the course of chronic renal disease.

We found no significant difference in the risk for death between the ACE inhibitor and control groups. This finding is consistent with Maschio and colleagues' conclusion [6] that the higher mortality rate in their ACE inhibitor group was a chance event rather than a detrimental effect of the treatment. Nonetheless, several side effects and risks of ACE-inhibitor therapy are well described in patients with chronic renal disease. These risks include hyperkalemia, cough, mild reduction in glomerular filtration rate in patients with parenchymal renal disease, and acute renal failure in patients with bilateral renal artery stenosis or volume depletion [29-31]. Patients receiving ACE inhibitors should have regular measurement of blood pressure, renal function, and serum electrolyte levels, especially during intercurrent illness.

We observed a greater decline in blood pressure in patients who received ACE inhibitors than in those receiving other antihypertensive agents, even though the target blood pressure was the same in both the ACE inhibitor and control groups. The magnitude of the difference in decline in systolic blood pressure (4.9 mm Hg) is large enough to have a clinically significant beneficial effect on the progression of renal disease [25, 26]. In meta-regression analysis, no statistically significant relation was seen in the difference between randomly assigned groups in blood pressure decline and the beneficial effect of ACE inhibition. However, the CIs for the regression coefficients were too wide to exclude a clinically significant association. Thus, we could not conclusively determine whether the beneficial effect of ACE inhibition was due to the greater decline in blood pressure.

Angiotensin-converting enzyme inhibition slows the progression of renal disease in animals by numerous mechanisms. In addition to decreasing systemic blood pressure, ACE inhibition decreases glomerular capillary pressure, reduces proteinuria, and suppresses mediators of glomerular and tubular hypertrophy and fibrosis [31, 32]. These effects seem to be shared among ACE inhibitors and among angiotensin-II-receptor antagonists [33]. No studies directly comparing different ACE inhibitors were available, and we found no apparent differences among clinical trials of different ACE inhibitors. However, the number of studies of agents other than enalapril was too small to allow meaningful comparisons. Reports of clinical trials of angiotensin-II-receptor antagonists on slowing the progression of renal disease have not yet been published.

Like all meta-analyses, our analysis is limited by differences among the clinical trials, particularly in patients' baseline characteristics. However, the result of the test for heterogenicity among studies was negative, indicating that ACE inhibition has a consistently beneficial effect despite these differences. The analysis is also limited by the lack of uniform data on other outcomes, such as rates of decline in renal function, and by lack of individual patient data. Hence, as discussed above, we could not determine the magnitude of the slowing of the decline in renal function or determine whether ACE inhibitor therapy was more or less beneficial in subgroups of patients defined by clinical or demographic characteristics. It is also possible that the patients who developed end-stage renal disease during the short follow-up were not representative of the majority of patients enrolled in these studies, but we believe that this is unlikely. If these patients were atypical, the beneficial effect of ACE inhibitors on the development of end-stage renal disease may not accurately reflect their effect on the decline in renal function. Finally, as discussed above, we could not determine whether the beneficial effect was due to the greater decline in blood pressure or to other effects of ACE inhibition.

In summary, the findings from our meta-analysis of randomized trials suggests that ACE inhibitors have a substantial beneficial effect in delaying the onset of end-stage renal disease and do not increase mortality. Together with the recent results of the largest study thus far to show a beneficial effect on the decline in renal function [6], our meta-analysis supports the conclusion that ACE inhibitors may be more effective than other antihypertensive agents in slowing the progression of chronic renal disease.

Appendix

The members of the Angiotensin-Converting-Enzyme Inhibition and Progressive Renal Disease Study Group who contributed data include Drs. Pietro C. Zucchelli, Anne-Lise Kamper, Svend Strandgaard, Paul P. Leyssac, Robert D. Toto, Barry M. Brenner, Nicolaos E. Madias, Barbara G. Delano, Shahnaz Shahinfar, Gabe G. van Essen, Alfred J. Apperloo, Dick de Zeeuw, Paul E. de Jong, Paul Landais, Jean-Pierre Grunfeld, Kym M. Bannister, Lennart Hansson, Anders Himmelmann, Gavin J. Becker, Benno U. Ihle, and Giuseppe Maschio.

Dr. Lau: Division of Clinical Care Research, New England Medical Center, 750 Washington Street, Boston, MA 02111.