Background Renin–angiotensin system inhibitors, both angiotensin-converting enzyme inhibitors and angiotensin receptor blockers, slow progression of mild and moderate chronic kidney disease. However, some evidence suggests that discontinuation of renin–angiotensin system inhibitors in patients with advanced chronic kidney disease might increase estimated glomerular filtration rate or slow its decline. Objective To test the hypothesis that stopping angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, or a combination of both, compared with continuing these treatments, improves or stabilises kidney function in patients with progressive stages 4 or 5 chronic kidney disease based on assessment of kidney function using the modification of diet in renal disease four-variable estimated glomerular filtration rate at 3 years, follow-up. Setting Thirty-seven UK hospitals with kidney services. Design An investigator-led multicentre open-label, randomised controlled trial of 411 participants with advanced (stage 4 or 5) progressive chronic kidney disease. Participants Adult patients with advanced (estimated glomerular filtration rate < 30 ml/minute/1.73 m2) and progressive chronic kidney disease who were receiving either angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, or a combination of both. Interventions Participants were randomised in a 1 : 1 ratio using a centralised internet-based system to either discontinue renin–angiotensin system inhibitors (n = 206) or continue renin–angiotensin system inhibitors (n = 205). Main outcome measures The primary outcome was the estimated glomerular filtration rate at 3 years; measurements of estimated glomerular filtration rate made after commencing kidney replacement therapy were excluded. Secondary outcomes included development of end-stage kidney disease or starting kidney replacement therapy, a composite of either a > 50% decline in estimated glomerular filtration rate or commencement of kidney replacement therapy (including end-stage kidney disease), cystatin C, hospitalisations, blood pressure, exercise capacity and quality of life. Cardiovascular events, death and safety were recorded. Results At 3 years, the least-squares mean (± standard error) estimated glomerular filtration rate was 12.6 ± 0.7 ml/minute/1.73 m2 in the discontinuation group and 13.3 ± 0.6 ml/minute/1.73 m2 in the continuation group [difference −0.7, 95% confidence interval (−2.5 to 1.0; p = 0.42)] with a negative value favouring the continuation group. The treatment effect did not differ (heterogeneity) when data were analysed by the pre-specified subgroups. End-stage kidney disease or kidney replacement therapy occurred in 128 (62%) and 115 (56%) participants randomised to the discontinue and continue renin–angiotensin system inhibitor groups, respectively (hazard ratio 1.28, 95% confidence interval 0.99 to 1.65). The numbers of cardiovascular events and deaths observed were similar for those randomised to discontinue (108 events and 20 deaths) or continue (88 events and 22 deaths) renin–angiotensin system inhibitors. Limitations Non-white ethnic backgrounds were poorly represented, limiting the generalisability of our findings. The open-label nature of the trial may have affected clinical care and subjective end points, such as quality of life and exercise capacity. We only included patients who were receiving renin–angiotensin system inhibitors at the time of randomisation, thus excluding those who had already discontinued these agents. Conclusions Discontinuing renin–angiotensin system inhibitors in advanced and progressive chronic kidney disease does not cause a clinically relevant change in estimated glomerular filtration rate or difference in its long-term decline. Future work Future work should focus on updating clinical guidelines. Further analyses, in addition to the prespecified analyses, may be undertaken if new estimated glomerular filtration rate equations are introduced into clinical practice. Subgroup analysis by kidney disease aetiology and gender may be undertaken to look for potential differences in outcome in specific groups. Trial registration This trial is registered as STOP ACEi EudraCT Number, 2013-003798-82; ISRCTN62869767. Funding This award was funded by the Efficacy and Mechanism Evaluation (EME) programme (NIHR award ref: 11/30/07), a Medical Research Council (MRC) and National Institute for Health and Care Research (NIHR) partnership. This is published in full in Efficacy and Mechanism Evaluation; Vol. 11, No. 5. See the NIHR Funding and Awards website for further award information. Plain language summary Drugs called angiotensin-converting enzyme inhibitors or angiotensin receptor blockers, together known as renin–angiotensin system inhibitors, are used to treat high blood pressure, slow worsening kidney function and lower the risk of kidney failure (known as end-stage kidney disease which requires treatment with dialysis or kidney transplantation) in patients with early chronic kidney disease. However, we did not know if patients treated with renin–angiotensin system inhibitors and who have progressed to more advanced chronic kidney disease (stage 4 or 5) should stop or continue renin–angiotensin system inhibitors. To determine whether stopping renin–angiotensin system inhibitors in people with advanced chronic kidney disease leads to an improvement or stabilisation of kidney function required a study comparing the outcomes of people who had had these drugs stopped with a group who continued these drugs (the STOP-angiotensin-converting enzyme inhibitors trial). We recruited 411 participants with advanced chronic kidney disease who were receiving renin–angiotensin system inhibitors from 37 kidney units in the UK, and randomly (like flipping a coin) allocated them to either stop or continue renin–angiotensin system inhibitors. We then compared kidney function between the two groups at 3 years. We also assessed whether stopping or continuing renin–angiotensin system inhibitors had an influence on the development of end-stage kidney disease or need for kidney replacement therapy, the number of hospitalisations, blood pressure, quality of life and physical function. We collected data on safety outcomes including death and heart-related events (such as heart attacks). The results of the trial showed no difference in kidney function at 3 years. The number of participants requiring dialysis, or a kidney transplant was also similar, as was the quality of life and physical function between the groups. Deaths and the number of heart events were similar in both groups. This research suggests that there is no benefit in stopping renin–angiotensin system inhibitors in patients with advanced chronic kidney disease. Scientific summary Background Renin–angiotensin system (RAS) inhibitors, both angiotensin-converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARB), slow the progression of mild and moderate chronic kidney disease (CKD). However, some evidence suggests that discontinuation of RAS inhibitors in patients with advanced CKD might increase estimated glomerular filtration rate (eGFR) or slow its decline. Primary objective To test the hypothesis that discontinuing ACEi or ARB treatment, or a combination of both, compared with continuing on these treatments, improves or stabilises kidney function in patients with progressive stage 4 or stage 5 CKD based on assessment of kidney function using the modification of diet in renal disease (MDRD) four-variable eGFR at 3 years follow-up. Secondary objectives To test whether in each of the randomised groups. Clinical outcomes The number of participants starting kidney replacement therapy (KRT) (dialysis or transplantation) or sustaining a > 50% decline in eGFR differs. There is a difference in the time taken to reach end-stage kidney disease (ESKD) or need for KRT. Hospitalisation rates from any cause are different. Participant quality of life and well-being [measured using the kidney disease quality of life (KDQoL)-SF™ v1.3 questionnaire] differs. Participant physical function (measured using the 6-minute walk test) differs. Withdrawal of these treatments does not cause excess harm [e.g. increased cardiovascular (CV) events such as heart failure, hypertension, myocardial infarction, stroke] and is not associated with an increase in adverse effects. Participant survival in each group is similar. Blood pressure (BP) control is the same. Cystatin-C levels differ. Mechanistic outcomes There is a change in urine protein excretion [urinary protein-to-creatinine ratio (uPCR)]. Discontinuation of ACEi/ARB affects haemoglobin concentration. Discontinuation of ACEi/ARB affects the requirement for erythropoietin stimulating agents. Methods An investigator-initiated, multicentre, open-label randomised trial where people with advanced and progressive CKD (eGFR < 30 ml/minute/1.73 m2) were randomised to either discontinue or continue RAS inhibitors, and then followed up every 3 months for 3 years. Patients underwent screening at 39 centres in the UK. Adults (≥ 18 years of age) with stage 4 or stage 5 CKD (eGFR, < 30 ml/minute/1.73 m2 of body-surface area) were eligible to participate in the trial if they were not receiving dialysis and had not undergone kidney transplantation. All eligible patients were required to have had a decrease of more than 2 ml/minute/1.73 m2 per year in the eGFR during the previous 2 years and to have been receiving treatment with an ACEi, an ARB, or both for more than 6 months. We calculated the eGFR using the four-variable equation used in the MDRD study, as updated in 2005 (MDRD175). Exclusion criteria included uncontrolled hypertension or a history of myocardial infarction or stroke within the previous 3 months. All the patients provided written informed consent. Patients were randomly assigned in a 1 : 1 ratio to either discontinue or continue RAS inhibitors. Randomisation used a centralised internet-based system with minimisation to ensure balance between groups for the following variables: age (< 65 or ≥ 65 years), eGFR (< 15 or ≥ 15 ml/minute/1.73 m2), diabetes (type 1, type 2, or none), mean arterial pressure (< 100 or ≥ 100 mmHg) and proteinuria [protein-to-creatinine ratio (PCR), < 100 or ≥ 100 mg/mmol]. In the group that discontinued RAS inhibitor, any guideline-recommended antihypertensive agent other than a RAS inhibitor could be used to control BP. In the group that continued RAS inhibitors, the responsible clinician chose the agent and dose of the RAS inhibitor and could combine it with any other guideline-recommended antihypertensive agent. The primary outcome was the eGFR at 3 years as calculated according to the MDRD175 four-variable equation. Secondary outcome measures included the time until the development of ESKD or initiation of KRT; a composite of a decrease of more than 50% in the eGFR, the development of ESKD, or the initiation of KRT; hospitalisation for any cause; measures of cystatin C and BP; quality of life (as measured on the KDQoL 36-Item Short Form Survey, version 1.3); exercise capacity (as assessed by the 6-minute walk test); and CV events and death. At the time of this report, the transfer and processing of samples for cystatin C measurement had not yet occurred, so the results are not provided here. Secondary mechanistic outcomes included measures of haemoglobin and urinary protein excretion (PCR). The trial aimed to recruit 410 patients (205 patients in each trial group) which would provide 80% power to determine a minimum relevant between-group difference in the eGFR of 5 ml/minute/1.73 m2 (alpha level of 0.05), assuming an attrition rate of 20%. This difference represents an effect size of 0.31, with a standard deviation of 16 ml/minute/1.73 m2. The analyses were based on the intention-to-treat principle and were adjusted for the minimisation variables and baseline values. A repeated-measures, mixed-effects linear regression model was used to estimate the between-group difference in eGFR at 3 years. Any measurements of eGFR that were made after patients had initiated dialysis or undergone kidney transplantation were excluded from the primary analysis. 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. We also repeated analyses for the primary outcome with the use of two other four-variable equations for the eGFR calculation: the Chronic Kidney Disease Epidemiology Collaboration 2009 equation and the MDRD186 equation. Continuously distributed secondary outcomes, such as BP, were analysed using the same methods as per the primary analysis, but data were not censored at the time of initiation of KRT. Categorical (dichotomous) secondary outcomes were analysed with the use of a Poisson regression model with robust standard errors (SEs) to estimate the relative risk (RR) and 95% confidence interval (CI). A Cox proportional-hazards model was used to calculate hazard ratios (HRs) and 95% CIs for time-to-event outcomes, such as the development of ESKD or the initiation of KRT. Categorical safety outcome measures (i.e. hospitalisation and serious adverse events) were summarised as the percentage of patients with these events. Data collection for kidney outcomes did not distinguish between ESKD and kidney-replacement outcomes (i.e. both outcomes used the same end-point code), although investigators could note the specific outcome. Prespecified subgroup analyses were performed only for the primary outcome according to the minimisation variables. Time and subgroup were included in the model to allow for the possibility of differential changes over time within subgroups, time according to subgroup and the three-way interaction among the variables of treatment, time and subgroup. Although all data were included in the regression models for the subgroup analyses, only estimates of differences at 3 years are presented. All analyses were performed with the use of SAS software, version 9.4 (SAS Institute), and Stata software, version 17 (StataCorp). Results At 3 years, among the 411 patients who underwent randomisation, the least-squares mean (LS-Mean) (±SE) eGFR was 12.6 ± 0.7 ml/minute/1.73 m2 in the discontinuation group and 13.3 ± 0.6 ml/minute/1.73 m2 in the continuation group (difference −0.7, 95% (CI −2.5 to 1.0; p = 0.42) with a negative value favouring the continuation group. End-stage kidney disease or the initiation of KRT occurred in 128 patients (62%) in the discontinuation group and in 115 patients (56%) in the continuation group (HR 1.28, 95% CI 0.99 to 1.65). The number of patients with > 50% decline in eGFR or need to start KRT (including ESKD) was 140/206 (68%) in the discontinue RAS inhibitor group compared to 127/202 (63%) in the continue RAS inhibitor group; RR 1.07, 95% CI 0.94 to 1.22. The number of hospitalisations were similar between the groups; 414 in the stop RAS inhibitor group versus 413 in the continue RAS inhibitor group. The difference in LS-Mean at 3 years for systolic BP was 0 mmHg, 95% CI −4 to 5 mmHg. The results were similar for diastolic BP; 0 mmHg, 95% CI −2 to 3 mmHg. Adverse events were similar in both the discontinuation group and continuation group with respect to CV events (108 vs. 88) and deaths (20 vs. 22). Conclusions Our STOP-ACEi trial showed that discontinuing RAS inhibitors for patients with advanced and progressive CKD does not lead to a clinically relevant change in eGFR or difference in the rate of long-term decline in eGFR, overall or in pre-specified subgroups by age, severity of CKD, diabetes, proteinuria or BP. Numerically more patients who discontinued RAS inhibitors had progression to ESKD or need for KRT, so a larger trial might have shown an advantage to continuing with RAS inhibition. The rate of CV events and death was similar. Systolic and diastolic BP and proteinuria were greater over the first year of follow-up in those randomised to discontinue RAS inhibitors but there was little difference, thereafter, reflecting initiation of antihypertensive agents other than RAS inhibitors. No differences in quality of life or exercise capacity were observed for those who discontinued or continued RAS inhibitors. Our trial lacked sufficient power to investigate the effect of withdrawing RAS inhibitors on CV events or mortality. However, because our trial suggests that there is no advantage in discontinuing RAS inhibitors from the perspective of kidney function, there is little rationale to conduct a larger randomised trial to investigate CV safety. Future work Future work should initially focus on updating clinical guidelines in the UK and potentially worldwide. Further analyses, in addition to the prespecified analyses, may be undertaken if new eGFR equations are introduced into routine clinical practice such as the National Institute for Health and Care Excellence recommended removal of black ethnicity correction factor from the eGFR equation. Consideration of subgroup analysis by aetiology of kidney disease and gender will be considered to look for any potential differences in outcome in specific groups which might warrant future studies. Trial registration This trial is registered as STOP ACEi EudraCT Number, 2013-003798-82; ISTRCTN62869767. Funding This award was funded by the Efficacy and Mechanism Evaluation (EME) programme (NIHR award ref: 11/30/07), a Medical Research Council (MRC) and National Institute for Health and Care Research (NIHR) partnership. This is published in full in Efficacy and Mechanism Evaluation; Vol. 11, No. 5. See the NIHR Funding and Awards website for further award information.