Evaluation of the stopping angiotensin converting enzyme inhibitor compared to angiotensin receptor blocker (STOP ACEi trial) in advanced and progressive chronic kidney disease

In the STOP-ACEi trial, the outcome was similar whether or not renin–angiotensin system inhibitors (RASi) were discontinued. We now investigate whether the effect of withdrawing angiotensin converting enzyme inhibitors (ACEi) or angiotensin-receptor blockers (ARBs) differed. In this open label trial patients with estimated glomerular filtration rates (eGFR) under 30ml/min per 1.73 m2 and progressive chronic kidney disease (CKD) were randomized to stop or continue RASi. The primary outcome was eGFR at three years. The composite of kidney failure, over 50% fall in eGFR, or kidney replacement therapy (KRT) was also assessed. Of patients randomized, 99 stopped and 123 patients continued ACEi while 104 stopped and 77 continued ARB at baseline. At three years, the eGFR was similar whether or not patients were withdrawn from ACEi or from ARB. Kidney failure or initiation of KRT occurred in 65% of those stopping and 54% continuing ACEi (hazard ratio if stopped, 1.52; 95% Confidence Interval, 1.07 to 2.16) and in 60% on an ARB regardless of randomized group (hazard ratio if stopped, 1.23; 0.83 to 1.81). Kidney failure/Initiation of KRT with over 50% decline in eGFR occurred in 71% of those stopping and 59% continuing ACEi (relative risk if stopped, 1.19; 95% CI, 1.00 to 1.41) and in 65% stopping and 69% continuing ARB (relative risk if stopped, 0.96; 0.79 to 1.16). Thus, neither discontinuing ACEi nor ARB slowed the rate of decline in eGFR. Although discontinuation of ACEi appeared to have more unfavorable effects on kidney outcomes than stopping ARB, the trial was neither designed nor powered to show differences between agents. In the STOP-ACEi trial, the outcome was similar whether or not renin–angiotensin system inhibitors (RASi) were discontinued. We now investigate whether the effect of withdrawing angiotensin converting enzyme inhibitors (ACEi) or angiotensin-receptor blockers (ARBs) differed. In this open label trial patients with estimated glomerular filtration rates (eGFR) under 30ml/min per 1.73 m2 and progressive chronic kidney disease (CKD) were randomized to stop or continue RASi. The primary outcome was eGFR at three years. The composite of kidney failure, over 50% fall in eGFR, or kidney replacement therapy (KRT) was also assessed. Of patients randomized, 99 stopped and 123 patients continued ACEi while 104 stopped and 77 continued ARB at baseline. At three years, the eGFR was similar whether or not patients were withdrawn from ACEi or from ARB. Kidney failure or initiation of KRT occurred in 65% of those stopping and 54% continuing ACEi (hazard ratio if stopped, 1.52; 95% Confidence Interval, 1.07 to 2.16) and in 60% on an ARB regardless of randomized group (hazard ratio if stopped, 1.23; 0.83 to 1.81). Kidney failure/Initiation of KRT with over 50% decline in eGFR occurred in 71% of those stopping and 59% continuing ACEi (relative risk if stopped, 1.19; 95% CI, 1.00 to 1.41) and in 65% stopping and 69% continuing ARB (relative risk if stopped, 0.96; 0.79 to 1.16). Thus, neither discontinuing ACEi nor ARB slowed the rate of decline in eGFR. Although discontinuation of ACEi appeared to have more unfavorable effects on kidney outcomes than stopping ARB, the trial was neither designed nor powered to show differences between agents. Lay SummaryThe STOP ACEi randomized trial (Multi-center Randomized Controlled Trial of Angiotensin Converting Enzyme inhibitor [ACEi]/Angiotensin Receptor Blocker [ARB] withdrawal in advanced renal disease) found that stopping compared with continuing renin-angiotensin system inhibitors (RASis) did not influence the decline in estimated glomerular filtration rate (eGFR) over the following 3 years or the number of kidney events in patients with advanced and progressive chronic kidney disease. Further analysis now shows that the decline in eGFR with RASi discontinuation was similar for patients who were on angiotensin-converting enzyme inhibitor (ACEi) or angiotensin receptor blocker (ARB) at baseline. Discontinuation of ACEi compared with discontinuation of ARB was associated with a higher incidence of progression to end-stage kidney disease or kidney replacement therapy. A similar trend was not observed for ARB, but this may reflect the small sample size and the play of chance. The trial was neither designed nor powered to show differences between agents. These data further support continuing RASis, especially ACEi, in advanced and progressive chronic kidney disease. The STOP ACEi randomized trial (Multi-center Randomized Controlled Trial of Angiotensin Converting Enzyme inhibitor [ACEi]/Angiotensin Receptor Blocker [ARB] withdrawal in advanced renal disease) found that stopping compared with continuing renin-angiotensin system inhibitors (RASis) did not influence the decline in estimated glomerular filtration rate (eGFR) over the following 3 years or the number of kidney events in patients with advanced and progressive chronic kidney disease. Further analysis now shows that the decline in eGFR with RASi discontinuation was similar for patients who were on angiotensin-converting enzyme inhibitor (ACEi) or angiotensin receptor blocker (ARB) at baseline. Discontinuation of ACEi compared with discontinuation of ARB was associated with a higher incidence of progression to end-stage kidney disease or kidney replacement therapy. A similar trend was not observed for ARB, but this may reflect the small sample size and the play of chance. The trial was neither designed nor powered to show differences between agents. These data further support continuing RASis, especially ACEi, in advanced and progressive chronic kidney disease. Renin-angiotensin system inhibitors (RASis), including angiotensin-converting enzyme inhibitors (ACEis) and angiotensin receptor blockers (ARBs), slow the decline in renal function in patients with mild to moderate chronic kidney disease (CKD), but patients with advanced CKD have been excluded from randomized trials.1Xie X. Liu Y. Perkovic V. et al.Renin-angiotensin system inhibitors and kidney and cardiovascular outcomes in patients with CKD: a Bayesian network meta-analysis of randomised clinical trials.Am J Kidney Dis. 2016; 67: 728-741Abstract Full Text Full Text PDF PubMed Scopus (256) Google Scholar,2Taler S.J. Agarwal R. Bakris G.L. et al.KDOQI US commentary on the 2012 KDIGO clinical practice guideline for management of blood pressure in CKD.Am J Kidney Dis. 2013; 62: 201-213Abstract Full Text Full Text PDF PubMed Scopus (168) Google Scholar The STOP-ACEi trial showed that, for patients with advanced and progressive CKD, neither the decline in estimated glomerular filtration rate (eGFR) over 3 years nor the development of end-stage kidney disease (ESKD) and/or initiation of kidney replacement therapy (KRT) differed according to whether RASis were continued or stopped.3Bhandari S. Mehta S. Khwaja A. et al.Renin–angiotensin system inhibition in advanced chronic kidney disease.N Engl J Med. 2022; 387: 2021-2032Crossref PubMed Scopus (84) Google Scholar However, it is possible that discontinuing ACEis may have different effects from discontinuing ARBs for these end points as they have different mechanisms of action: ACEis reduce the production of angiotensin II and prevent degradation of the vasoactive peptide bradykinin; ARBs bind to angiotensin-1 receptors.4Goodfriend T. Elliott M. Catt K. Angiotensin receptors and their antagonist.N Engl J Med. 1996; 334: 1649-1654Crossref PubMed Google Scholar One randomized controlled trial in patients with type 2 diabetes and early diabetic kidney disease showed no difference in the rate of decline of kidney function based on eGFR between enalapril or telmisartan.5Barnett A.H. Bain S.C. Bouter P. et al.Diabetics Exposed to Telmisartan and Enalapril Study Group. Angiotensin-receptor blockade versus converting-enzyme inhibition in type 2 diabetes and nephropathy.N Engl J Med. 2004; 351: 1952-1961Crossref PubMed Scopus (807) Google Scholar In a Bayesian network meta-analysis of randomized controlled trials that assessed kidney and mortality outcomes that included 4 trials of ACEis and ARBs, among other antihypertensive agents, there was no difference in doubling of serum creatinine, 50% decrease in eGFR, or development ESKD between ACEis and ARBs, but these trials did not include patients with advanced CKD.1Xie X. Liu Y. Perkovic V. et al.Renin-angiotensin system inhibitors and kidney and cardiovascular outcomes in patients with CKD: a Bayesian network meta-analysis of randomised clinical trials.Am J Kidney Dis. 2016; 67: 728-741Abstract Full Text Full Text PDF PubMed Scopus (256) Google Scholar A previous Cochrane Review from Strippoli et al. of 49 studies with 12,067 people at all stages of kidney disease examined the use of RASis in retarding the progression of kidney disease in people with diabetes.6Strippoli G.F. Bonifati C. Craig M. et al.Angiotensin converting enzyme inhibitors and angiotensin II receptor antagonists for preventing the progression of diabetic kidney disease.Cochrane Database Syst Rev. 2006; 2006: CD006257Google Scholar The authors found that both ACEis and ARBs improved kidney outcomes (ESKD, delayed doubling of creatinine, prevented progression of microalbuminuria to macroalbuminuria, and increased resolution of microalbuminuria). Furthermore, when compared with placebo, use of ACEis at maximum tolerated doses (n = 2034 participants) appeared to reduce mortality in people with diabetic kidney disease (relative risk [RR], 0.78; 95% confidence interval [CI], 0.61–0.98), which was not the case for ARBs (n = 3049 participants) (RR, 0.99; 95% CI, 0.85–1.17). A meta-analysis of 9 randomized trials (n = 9797 participants) of at least 6 months' duration in adults with diabetes and non–dialysis-dependent CKD stage 3 to 5 investigated the effects of a single RASi versus placebo or an alternative antihypertensive agent.7Nistor I. De Sutter J. Drechsler C. et al.Effect of renin–angiotensin–aldosterone system blockade in adults with diabetes mellitus and advanced chronic kidney disease not on dialysis: a systematic review and meta-analysis.Nephrol Dial Transplant. 2018; 33: 12-22Crossref PubMed Scopus (22) Google Scholar The authors found no difference between the RASi group and control group regarding all-cause mortality (RR, 0.97; 95% CI, 0.85–1.1), cardiovascular mortality (RR, 1.03; 95% CI, 0.75–1.41), or adverse events (RR, 1.05; 95% CI, 0.89–1.25). There was, however, a trend toward a lower risk of the composite end point of need for KRT/doubling of serum creatinine (RR, 0.81; 95% CI, 0.70–0.92) in the RASi group versus the control group, but this was dependent on the selected outcome measure. In the STOP-ACEi trial, we did not try to balance the use of ACEi and/or ARB at the point of randomization (i.e., not included as a minimization variable in the randomization system); these differences in medication may have an impact on the outcomes. Robert Speth, in a recent letter, highlighted important different biological effects of ACEi therapy and ARB therapy.8Bhandari S. Ives N. Brettell E.A. et al.Multicentre randomized controlled trial of angiotensin-converting enzyme inhibitor/angiotensin receptor blocker withdrawal in advanced renal disease: the STOP-ACEi trial.Nephrol Dial Transplant. 2016; 31: 255-261Crossref PubMed Google Scholar,9Speth R.C. Renin–angiotensin system inhibition in advanced CKD.N Engl J Med. 2023; 388: 1436-1437Google Scholar Therefore, we now report separately the effect of withdrawing ACEi or ARB in the STOP ACEi randomized trial.3Bhandari S. Mehta S. Khwaja A. et al.Renin–angiotensin system inhibition in advanced chronic kidney disease.N Engl J Med. 2022; 387: 2021-2032Crossref PubMed Scopus (84) Google Scholar The STOP-ACEi trial was an open-label (nonblinded), multicenter trial that randomized patients with advanced and progressive CKD to stop or continue RASi.3Bhandari S. Mehta S. Khwaja A. et al.Renin–angiotensin system inhibition in advanced chronic kidney disease.N Engl J Med. 2022; 387: 2021-2032Crossref PubMed Scopus (84) Google Scholar,10Bhandari S. Cockwell P. Renin–angiotensin system inhibition in advanced CKD: reply.N Engl J Med. 2023; 388: 1438Google Scholar The primary outcome was eGFR at 3 years. Secondary outcomes included clinically important kidney, cardiovascular, and safety events. Patients aged ≥18 years, with an eGFR <30 ml/min per 1.73 m2 and a decrease of ≥2 ml/min per 1.73 m2 per year in eGFR over the previous 2-years, and receiving an ACEi, an ARB, or both for >6 months were included. Exclusion criteria included uncontrolled hypertension, a history of myocardial infarction or stroke within 6 months, and kidney disease requiring active immunosuppression or KRT, including dialysis or transplantation. Outcomes were assessed separately for patients taking ACEi or ARB at baseline. The allocation arms for patient randomized to stop or continue were kept the same; however, the subgroup of whether the patient had ACEi or ARB therapy was not controlled for at randomization. The primary outcome for each subgroup was eGFR at 3 years, calculated using the 175 Modification of Diet in Renal Disease (MDRD175) equation as in the main article.11Al-Maqbali S.R.S. Mula-Abed W.-A.S. Comparison between three different equations for the estimation of glomerular filtration rate in Omani patients with type 2 diabetes mellitus.Sultan Qaboos Univ Med J. 2014; 14: e197-e203PubMed Google Scholar Data for the primary analysis were censored at the initiation of KRT. Secondary outcome measures included the time until the development of ESKD (as defined by the local investigator according to criteria that included terminal palliative care or KRT); a composite of a decrease of >50% in the eGFR, the development of ESKD, or the initiation of KRT; hospitalization for any cause; cardiovascular events (these included hospitalization for heart failure, myocardial infarction, stroke, heart failure events, angina, coronary intervention, hypertension, atrial arrhythmias, venous thromboembolism, peripheral vascular disease, and other cardiac conditions); and mortality. Analyses were based on the intention-to-treat principle and were adjusted for the prespecified minimization variables in the main trial (age, diabetes, mean arterial blood pressure, proteinuria, and eGFR) and baseline values (where available, for all those collected). The intention-to-treat population included all the patients who had undergone randomization, regardless of what treatment (if any) they had received, except for those on both ACEi and ARB therapy (n = 6) and 2 patients with no data for whether they were on an ACEi or an ARB at randomization. These 8 patients were included in the original study but not in this analysis, where knowledge of whether a patient was receiving an ACEi or ARB was required. Therefore, these 8 patients were excluded from all subgroup analysis. All available data for patients who had been lost to follow-up, had withdrawn from the trial, or had died before trial completion were included in the analysis. The continue RASi group was used as the reference category in all model-based analyses, so values >0 for continuous outcomes and <1 for time to event and binary outcomes indicated better outcome for the discontinuation group. We used a repeated-measures, mixed-effects linear regression random slope model to estimate the between-group difference in eGFR at 3 years and effects on eGFR slope. A compound symmetry covariance structure was assumed in the model, and robust SEs were used. Time and subgroup (i.e., ACEi or ARB) were included to allow for the possibility of differential changes over time within subgroup, time according to subgroup, and the 3-way interaction among the variables of treatment, time, and subgroup. Any measurements of eGFR that were made after patients had initiated dialysis or undergone kidney transplantation were excluded. To examine the effect of data that were not missing at random, we performed sensitivity analyses by fitting pattern mixture and joint models for the primary outcome. A Cox proportional hazards model (which included a term for the interaction of the subgroup with treatment group) was used to calculate hazard ratios and 95% CIs for time-to-event outcomes, such as the development of ESKD or the initiation of KRT. Poisson regression model with robust SEs (which included a term for the interaction of the subgroup with treatment group) was used to calculate the RRs and 95% CIs for binary outcomes, such as composite of a decrease of >50% in the eGFR, the development of ESKD, or the initiation of KRT. A total of 411 patients underwent randomization to stop or continue ACEi/ARB, and 403 were eligible for this subanalysis; 99 patients receiving an ACEi and 104 patients receiving an ARB were assigned to discontinue RASi, and 123 patients receiving an ACEi and 77 patients receiving an ARB were assigned to continue RASi. The characteristics of the 4 patient groups at baseline are shown in Table 1. Median age (range, 61.4–64.1 years), mean serum creatinine (range, 312–322 μmol/L [3.53–3.64 mg/dl]), and mean eGFR (range, 17.7–17.9 ml/min per 1.73 m2) were similar between the 4 groups, as were other baseline variables.Table 1Baseline data by ACEi and ARBBaseline characteristicsACEiARBStop (N = 99)Continue (N = 123)Stop (N = 104)Continue (N = 77)Age, yr, median [IQR]64.1 [53.3 to 72.3]61.6 [50.6 to 72.1]61.4 [54.6 to 74.1]64 [49.4 to 71.1]Age group, yr <6551 (51.5)68 (55.3)63 (60.6)39 (50.6) ≥6548 (48.5)55 (44.7)41 (39.4)38 (49.4)Sex Men72 (72.7)83 (67.5)65 (62.5)54 (70.1) Women27 (27.3)40 (32.5)39 (37.5)23 (29.9)BMI, kg/m2, median [IQR]29.5 [25.1 to 33.8]29.5 [25.5 to 33.3]29.4 [25.6 to 33.1]27.9 [24.8 to 31.9]Hemoglobin, g/l, median [IQR]116 [107 to 127]115 [107 to 126]116 [108 to 127]115 [106 to 122.5]eGFR, ml/min per 1.73 m2, median [IQR]17 [13 to 21]18 [14 to 21]18 [14 to 22]18 [14 to 22]eGFR group, ml/min per 1.73 m2 <1528 (28.3)34 (27.6)29 (27.9)24 (31.2) ≥1571 (71.7)89 (72.4)75 (72.1)53 (68.8)Rate of decline of eGFR, ml/min per 1.73 m2, median [IQR]–4.7 [–7.8 to –3.2]–4.7 [–7.3 to –3.6]–5.1 [–7.4 to –3.4]–4.8 [–7.7 to –3.2]Albumin, g/L, median [IQR]41 [37 to 43]40 [38 to 43]40 [37 to 43]40 [38 to 42]C-reactive protein, mg/l, median [IQR]4.4 [2 to 7.5]4.9 [2.3 to 9]4 [2 to 6.8]5 [2 to 6]Potassium, mmol/L Mean (SD)5 (0.5)5 (0.6)4.8 (0.5)5 (0.6) Median [IQR]5.1 [4.7 to 5.4]5.1 [4.7 to 5.4]4.9 [4.5 to 5.2]5 [4.6 to 5.4] Min–Max3.5 to 6.33.7 to 6.42.9 to 6.23.3 to 6.6Proteinuria, mg/mmol Mean (SD)198 (246.6)178.1 (224.9)158 (167.4)165.2 (156) Median [IQR]119 [22 to 236.3]90 [22 to 255]100.6 [31.2 to 233.5]119 [46.5 to 243.3] Min–Max1.8 to 12092.1 to 11370.3 to 790.51 to 753Proteinuria group, mg/mmol <10045 (45.5)63 (51.2)52 (50)34 (44.2) ≥10054 (54.5)60 (48.8)52 (50)43 (55.8)Systolic blood pressure, mm Hg Mean (SD)134.8 (14.2)134.9 (13.3)138.1 (11.7)138.3 (14.1) Median [IQR]135 [124 to 146]136 [125 to 146]137 [130 to 147.5]140 [130 to 149] Min–Max99 to 15999 to 160106 to 16098 to 158Diastolic blood pressure, mm Hg Mean (SD)74.7 (9)75.2 (8.6)76.5 (9.4)77.4 (9.4) Median [IQR]75 [68 to 81]77 [69 to 81]78 [70 to 84]78 [73 to 85] Min–Max50 to 9053 to 9040 to 9045 to 90Mean arterial pressure, mm Hg Mean (SD)94.7 (8.3)95.1 (7.8)97 (7.8)97.7 (9.4) Median [IQR]96 [89.7 to 100.7]95.3 [91.3 to 100.7]97.5 [91.2 to 102]100 [92.7 to 104] Min–Max71.3 to 111.371 to 112.376 to 113.370.7 to 112.7Mean arterial pressure group, mm Hg <10071 (71.7)89 (72.4)59 (56.7)38 (49.4) ≥10028 (28.3)34 (27.6)45 (43.3)39 (50.6)Diabetic Type 17 (7.1)4 (3.3)2 (1.9)6 (7.8) Type 232 (32.3)44 (35.8)32 (30.8)22 (28.6) Nondiabetic60 (60.6)75 (61)70 (67.3)49 (63.6) Ethnicity White82 (82.8)109 (88.6)87 (83.7)66 (85.7) Black11 (11.1)4 (3.3)5 (4.8)3 (3.9) Asian5 (5.1)10 (8.1)8 (7.7)6 (7.8) Other1 (1)0 (0)4 (3.8)2 (2.6)ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; eGFR, estimated glomerular filtration rate; IQR, interquartile range; Max, maximum; Min, minimum.Data are given as n (%) of each group unless otherwise indicated.A total of 6 patients were taking both ACEi and ARB, and 2 patients withdrew from the study before baseline data could be obtained; therefore, these 8 patients are not included in the baseline table. Open table in a new tab ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; BMI, body mass index; eGFR, estimated glomerular filtration rate; IQR, interquartile range; Max, maximum; Min, minimum. Data are given as n (%) of each group unless otherwise indicated. A total of 6 patients were taking both ACEi and ARB, and 2 patients withdrew from the study before baseline data could be obtained; therefore, these 8 patients are not included in the baseline table. At 3 years, among the 222 patients receiving ACEi at randomization, the least-squares mean ±SE eGFR was 10.4 ± 1.1 ml/min per 1.73 m2 in the discontinuation group and 12.0 ± 0.8 ml/min per 1.73 m2 in the continuation group (difference, –1.6; 95% CI, –4.2 to 1.1, with negative values favoring continuation) (Figure 1 and Table 2). At 3 years, among 181 patients receiving ARB at randomization, the least-squares mean ± SE eGFR was 9.7 ± 0.7 ml/min per 1.73 m2 in the discontinuation group and 10.2 ± 1.0 ml/min per 1.73 m2 for ARB in the continuation group (difference, –0.5; 95% CI, –3.0 to 2.0; Figure 1 and Table 2). No heterogeneity in outcome by subgroup was observed (P = 0.77). The results of sensitivity analyses that were performed with the pattern-mixture model and joint model were similar to the primary outcome results (Table 3).Table 2Primary outcome summary (eGFR calculated using 4-MDRD175 equation) by ACEi and ARB over 3 yearsACEiARBeGFR valuesTime point, moSummary statisticStopContinueStopContinueeGFR,ml/min per 1.73 m2BaselineN9912310477Empirical mean (SD)17.7 (5.3)17.9 (5.2)17.9 (5.4)17.9 (4.7)3N901139474Empirical mean (SD)17.1 (6.6)17.2 (6.3)17.3 (6.2)16.8 (5)6N79998470Empirical mean (SD)16.3 (6.9)17.3 (7.1)16.9 (5.9)16.1 (5.7)9N69897666Empirical mean (SD)16.2 (7)17.6 (6.7)17.1 (5.7)16 (5.6)12N60826858Empirical mean (SD)16.1 (7.2)17.5 (6.9)16.8 (5.5)15.4 (5.9)15N54706152Empirical mean (SD)16.2 (7.4)16.6 (6.4)16.5 (5.4)16 (6.1)18N44635544Empirical mean (SD)16.5 (8)17.8 (6.4)17 (5.7)16.8 (5.9)21N44615138Empirical mean (SD)16.5 (7.9)17.8 (6.4)16.5 (5.8)16.2 (6.4)24N40594735Empirical mean (SD)15.8 (8.8)16.9 (5.7)16.7 (5.9)16.2 (5.6)27N36544132Empirical mean (SD)16.5 (8.2)16.4 (5.6)15.8 (6)15 (5.6)30N31483731Empirical mean (SD)16.7 (8.8)16.3 (7)15.9 (6.1)15.8 (7.2)33N25423323Empirical mean (SD)16.4 (10.3)17.5 (6.6)16.1 (5.3)14.2 (5.9)36N22423326Empirical mean (SD)18.4 (10.7)17.5 (6.6)16.1 (5.5)15.3 (7.7)ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; eGFR, estimated glomerular filtration rate; 4-MDRD175, 4 variable revised 175 Modification of Diet in Renal Disease Equation.eGFR values after commencing kidney replacement therapy (i.e., dialysis/transplant) are not included. Open table in a new tab Table 3Primary outcome and sensitivity analysis (eGFR calculated using 4-MDRD175 equation) by ACEi and ARBPrimary outcomeStop ACEiContinue ACEiMean difference (95% CI)Stop ARBContinue ARBMean difference (95% CI)Primary outcome: LS value at 3 yr Estimated eGFR MDRD175 4 variable10.4 ± 1.112.0 ± 0.8–1.6 (–4.2 to 1.1)P = 0.259.7 ± 0.710.2 ± 1.0–0.5 (–3.0 to –2.0)P = 0.69Primary outcome sensitivity analysis: pattern mixture modelaPattern mixture model: this method is a 2-stage procedure where in the first stage any missing data are first imputed using multiple imputation assuming data are "missing at random." A total of 50 imputations were generated for any missing data for the primary outcome. The second stage involves changing the imputed values for those participants from the time that they commenced dialysis or had a kidney transplant. For the second stage, the following approach was used: • Any imputed values for those participants from the time that they commenced dialysis or had a kidney transplant to an eGFR was replaced with a value of 5. • This value was agreed with the clinicians and an appropriate average value for eGFR that is likely observed before patients commence dialysis or have a kidney transplant. Imputed results will be combined using the Rubin rule, which in Stata is the "mi estimate" command. Estimated eGFR MDRD175 4 variable with: Flat value 5 imputation for MNAR eGFR values9.6 ± 0.710.5 ± 0.6–0.9 (–2.6 to 0.9)P = 0.349.2 ± 0.69.2 ± 0.70.0 (–1.8 to 1.8)P = 0.98Primary outcome sensitivity analysis: joint modelbJoint model: the primary outcome eGFR over time and time to end-stage kidney disease/dialysis/transplant were jointly modeled using the joint model method. Estimated eGFR MDRD175 4 variable13.2 ± 0.713.8 ± 0.5–0.6 (–2.1 to 1.0)P = 0.4711.5 ± 0.612.1 ± 0.6–0.6 (–2.2 to 1.1)P = 0.48ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; CI, confidence interval; eGFR, estimated glomerular filtration rate; LS, least squares; 4-MDRD175, 4 variable revised 175 Modification of Diet in Renal Disease Equation; MNAR, missing not at random.Data are given as mean ± SE unless otherwise indicated.a Pattern mixture model: this method is a 2-stage procedure where in the first stage any missing data are first imputed using multiple imputation assuming data are "missing at random." A total of 50 imputations were generated for any missing data for the primary outcome. The second stage involves changing the imputed values for those participants from the time that they commenced dialysis or had a kidney transplant. For the second stage, the following approach was used:• Any imputed values for those participants from the time that they commenced dialysis or had a kidney transplant to an eGFR was replaced with a value of 5.• This value was agreed with the clinicians and an appropriate average value for eGFR that is likely observed before patients commence dialysis or have a kidney transplant.Imputed results will be combined using the Rubin rule, which in Stata is the "mi estimate" command.b Joint model: the primary outcome eGFR over time and time to end-stage kidney disease/dialysis/transplant were jointly modeled using the joint model method. Open table in a new tab ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; eGFR, estimated glomerular filtration rate; 4-MDRD175, 4 variable revised 175 Modification of Diet in Renal Disease Equation. eGFR values after commencing kidney replacement therapy (i.e., dialysis/transplant) are not included. ACEi, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; CI, confidence interval; eGFR, estimated glomerular filtration rate; LS, least squares; 4-MDRD175, 4 variable revised 175 Modification of Diet in Renal Disease Equation; MNAR, missing not at random. Data are given as mean ± SE unless otherwise indicated. The eGFR of those randomized to discontinuation of RASi, censored for death, dialysis, or transplantation, declined by –9.1 ml/min per 1.73 m2 (95% CI, –10.3 to –8.0 ml/min per 1.73 m2) over 3 years. A similar decline in eGFR occurred in those randomized to continue RASi: –7.6 ml/min per 1.73 m2 (95% CI, –8.8 to –6.5 ml/min per 1.73 m2) over 3 years. Therefore, this equated to an average difference of–0.5 ml/min per 1.73 m2 per year between RASi discontinuation compared with RASi continuation. For the subgroup analysis of ACEi and ARB for the 4 groups, these slopes were also similar. In those receiving ACEi at randomization, the secondary composite outcome of ESKD or initiation of KRT occurred in 64 patients (65% of ACEi-treated subgroup) assigned to stop ACEi and 67 patients (54% of ACEi-treated subgroup) assigned to continue ACEi (hazard ratio if stopped, 1.52; 95% CI, 1.07–2.16) (Figure 2). In those receiving ARB at randomization, ESKD or the initiation of KRT occurred in 62 patients (60% of ARB-treated subgroup) randomized to stop ARB and in 46 patients (60% of ARB-treated subgroup) randomized to continue ARB (hazard ratio if stopped, 1.23; 95% CI, 0.83–1.81) (Figure 2). In those receiving ACEi at randomization, a composite of ESKD, initiation of KRT, or >50% decrease in eGFR occurred in 70 patients (71% of ACEi-treated subgroup) randomized to stop ACEi and in 72 patients (59% of ACEi-treated subgroup) randomized to continue ACEi (relative risk if stopped, 1.19; 95% CI, 1.00–1.41). In those receiving ARB at randomization, the composite occurred in 68 patients (65% of ARB-treated subgroup) assigned to stop ARB and in 53 patients (69% of ARB-treated subgroup) assigned to continue ARB (relative risk, 0.96; 95% CI, 0.79–1.16). However, there was no heterogeneity in outcome according to treatment at baseline (P = 0.41 and P = 0.11, respectively, after adjustment for minimization variable and treatment by ACEi or ARB (Figure 3). There were 46 (22 participants) cardiovascular events among those assigned to stop ACEi and 50 (29 participants) events in those assigned to continue ACEi in the ACEi-treated subgroup compared with 62 events (33 participants) and 36 events (19 participants) among those assigned to stop or continue ARB, respectively, in the ARB-treated subgroup (Table 4). Various cardiovascular events were similar in the 4 subgroups, except for hypertension, which was more common in those participants in the STOP ACEi or ARB subgroups (Table 5).Table 4Secondary outcomes by ACEi and ARB subgroupsACEiARBStop (N = 99)Continue (N = 123)Stop (N = 104)Continue (N = 77)Clinical variableTotal eventsPatients with an event, n (%)Total eventsPatients with an event, n (%)Total eventsPatients with an event n (%)Total eventsPatients with an event, n (%)Cardiovascular eventsaCa