Effects of angiotensin-converting enzyme inhibitors and angiotensin receptor blockers on cardiovascular events and residual renal function in dialysis patients: a meta-analysis of randomised controlled trials

We undertook a systematic review of the literature according to the approach recommended by the statement for the conducting of meta-analysis of intervention studies [12]. Relevant studies were identified by searching the following data sources: MEDLINE (OVID) (from 1950 to December 2016), Embase (from 1970 to December 2016), the Cochrane Library database (Cochrane Central Register of Active controlled Trials; no date restriction), and Wanfang database. We used the MeSH headings and text words of all spellings of known ACE inhibitors and ARBs, dialysis, cardiovascular events, and kidney failure (see Additional files 1). Trials were limited to randomized controlled trials (RCTs) without language restriction. Reference lists from identified trials and review articles were searched manually to identify any other relevant studies. We also searched the Clinical Trials.​gov website for randomized trials that were registered as completed but not yet published. All completed RCTs that assessed the effects of ACE-Is or ARBs compared with placebo or other antihypertensive drugs in dialysis patients, and which reported cardiovascular, renal or adverse outcomes, were eligible for inclusion.

Published reports were obtained for each eligible trial, and relevant information extracted into a spreadsheet. The data sought included dialysis modality, number of patients, country in which the study was performed, patient age, mean baseline systolic and diastolic blood pressure values, residual GFR, Kt/v, mean duration on dialysis, follow-up duration, change in blood pressure, outcome events (including CVEs, all-cause death, and RRF). Major cardiovascular events were defined as a composite of fatal or non-fatal myocardial infarction, fatal or non-fatal stroke, heart failure, or comparable definitions used by individual authors or cardiovascular mortality. Residual renal function was measured by GFR or endogenous creatinine clearance (CrCl), or urine volume, and drug-related adverse events if sufficient data were available. The literature were searched and identified by two investigators (LYX and MXX) independently. Data extraction and quality assessment (Grading of Recommendations Assessment, Development and Evaluation system) [13] was undertaken independently by two investigators (ZJ and MXX) using a standardized approach. Any disagreement between the two investigators in the abstracted data was adjudicated by a third reviewer (JJY).

The relative risk (RR) and 95% confidence interval (CI) for each outcome was calculated before pooling by the random-effects model. For the continuous measurement of change of GFR, blood pressure and urine volume, we used the weighted mean difference between groups. Heterogeneity across the included studies was analyzed using the I² to describe the percentage of variability. We made graphic representations of potential publication bias using Begg Funnel plots of the natural logarithm of the RR versus its standard error (SE) and assessed them visually. A 2-sided p value less than 0.05 was considered statistically significant, and statistical analyses were performed using STATA, version 12.0 and Review Manager 5.1.

Seven studies reported 455 cardiovascular events [8‐11, 14, 19, 20]. Of the 828 patients treated with ACEI/ARB there were 218 cardiovascular events (26.3%) and 237 events occurred in 826 patients treated with placebo or active agents (28.7%). Overall, ACE-Is and ARBs did not reduce cardiovascular events versus placebo or other antihypertensive agents (RR 0.92, 95% CI 0.79 to 1.08, Fig. 2). There was evidence of significant heterogeneity for effect of CVEs across included studies (I² = 71.6%, p = 0.002). Subgroup analysis indicated that the presence of heterogeneity was due to the different RASI category (ACEI or ARB), shown in Fig. 3. Indirect comparison suggested that ARB seemed to provided a higher probability of being beneficial for CVEs in dialysis patients (0.77, 0.63 to 0.94), while ACEI did not (1.24, 0.96 to 1.61). Subgroup analysis detected no significant difference between the two groups with regard to different control group (placebo or active agents), the dialysis mode, follow-up year, sample size and patient age. Data for heart failure events were available from 4 trials including 1115 patients in whom 132 events were recorded [8, 10, 19, 20]. ACEI/ARB therapy in dialysis patients reduced the risk of heart failure events by 33% (0.67, 0.47 to 0.93) with extensive heterogeneity in the results of individual trials (I² = 74.6%, p = 0.008, Fig. 4). In order to diminish the heterogeneity, a subgroup analysis was performed based on different type of RASI for comparison which led to a nearly 70% decrease of I² while did not affect the association of ACEI/ARB with lower risk of heart failure events (0.22, 0.38 to 0.78; p = 0.001; I² = 0.0%, p = 0.51). There were no significant differences between ACEI/ARB and placebo or active agents therapy on the outcomes of myocardial infarction (1.0, 0.45 to 2.22; I² = 0%, p = 0.71), stroke (1.16, 0.69 to 1.96; I² = 0%, p = 0.60) and cardiovascular death (0.89, 0.64 to 1.26; I² = 0%, p = 0.81) (Fig. 5).

Eight studies reported 122 deaths in 873 patients with ACEI/ARB treatment (14.0%) and 143 deaths of the 873 patients with placebo or active agents therapy (16.4%) [8‐11, 14, 18‐20]. Overall, ACEI/ARB therapy did not reduce all-cause mortality of dialysis patients (0.94, 0.75–1.17) (Fig. 5). Subgroup analyses showed that the association between ACEI/ARB therapy and risk of all-cause mortality was not modified by different control group, RASI category, dialysis mode, follow-up year, sample size and patient age (all p for heterogeneity >0.05, Additional file 1: Figure S1).

Data regarding the effects of ACEI/ARB on renal function were available from 5 trials [11, 14‐17], including 1 trial (n = 82) conducted in hemodialysis patients and 4 in peritoneal dialysis patients (n = 170). The average residual GFR declined by 1.44 ml/min per 1.73 m² in the ACEI/ARB group versus 2.37 ml/min per 1.73 m² in the placebo or active control group. The average decline in residual GFR was 0.93 ml/min per 1.73 m² (95% CI, 0.38 to 1.47 ml/min per 1.73 m²) less in patients receiving ACEI/ARB than in placebo or active control group patients (p < 0.001) with no evidence of heterogeneity (I² = 0%, p = 0.65) (Fig. 6).

Three studies including 158 participants reported the changes in urine volume between ACEI/ARB and placebo or active control therapy [11, 17, 18], and found ACEI/ARB treatment was a borderline significant factor in delaying the decline in urine volume: MD 167 ml, 95% CI 21 ml to 357 ml; p = 0.08) (Additional file 2: Figure S2).

Data on adverse events potentially associated with treatment were collected from these studies but were inconsistently reported (Table 3). Overall, ten trials reported at least 1 adverse event. Compared with control, ACE-I/ARBs therapy did not clearly increase the risk of hyperkalemia (1.29, 0.76 to 2.17), hypotension (1.03, 0.73 to 1.45) or cough (2.63, 0.00 to 39,507).

Formal statistical testing showed no evidence of publication bias for major cardiovascular events (Begg’s test p = 0.87), which was displayed in Additional file 1: Figure S3.