Comparative Efficacy of Angiotensin Converting Enzyme Inhibitors and Angiotensin Receptor Blockers after Coronary Artery Bypass Grafting

Although angiotensin receptor blockers (ARBs) are considered as an alternative for those with angiotensin converting enzyme inhibitors (ACEi) intolerance, the comparative effectiveness of ARBs and ACEi remains controversial in patients who underwent coronary artery bypass grafting (CABG). We aimed to compare the clinical effects of the two types of renin-angiotensin-aldosterone system (RAAS) inhibitors in patients who underwent CABG. From January 2001 to January 2015, among the 5456 patients, data from 1198 (20.1%) patients who used a RAAS inhibitor at discharge were analyzed. These 1198 patients were classified into ACEi (N = 900) and ARB (N = 298) groups. The primary outcome was major adverse cardiovascular and cerebrovascular events (MACCE) during a median follow-up period of 48 months. Propensity-matched analysis revealed that the incidence of MACCE over a 48 month follow-up period did not differ between the groups (HR, 0.65; 95% CI, 0.36–1.21; p = 0.17), but it was significantly lower in the ARB group during the 12 month follow-up period (HR, 0.46; 95% CI, 0.22–0.96; p = 0.04). In conclusion, ARBs may have comparable protective effects to ACEi and be a reasonable alternative for intolerant patients after CABG. The beneficial effects of ARBs depending on follow-up period require further investigation.

shows the cumulative incidences of clinical outcomes of the entire population. Multivariate Cox's proportion hazard regression analysis revealed that the incidence of major adverse cardiovascular and cerebrovascular event (MACCE) during 48 months of follow-up did not differ between the two groups (hazard ratio [HR], 0.69; 95% confidence interval [CI], 0.44-1.06; p = 0.09), but the incidence was significantly lower in the ARB group at the 12 month follow-up (HR, 0.49; 95% CI, 0.27-0.91; p = 0.02). Similar results were obtained in the propensity-matched analysis (HR, 0.65; 95% CI, 0.36-1.21; p = 0. 17  www.nature.com/scientificreports www.nature.com/scientificreports/ CI 95% 0.22-0.96; p = 0.04 for the 12 month follow-up) ( Table 3). In the analysis after excluding patients who discontinued or changed the class of RAAS inhibitors during the first year after CABG, the clinical outcomes did not differ between the two groups (Supplementary Table 1). The numbers of adverse events such as MACCE and cancer per 1000 months were also provided in Supplementary Table 2.
Subgroup analysis showed a significant interaction between sex and the use of RAAS inhibitors on the primary outcome. The use of ARB was beneficial only in female (HR, 0.23; 95% CI, 0.07-0.77, p = 0.02). Except for sex, there was no interaction between the various covariates and the use of RAAS inhibitors. Results of the subgroup analysis are shown in the hazard-ratio forest plots in Fig. 2.

Discussion
The main findings of the present study were as follows: (1) postoperative use of ARB after CABG showed a comparable clinical effect to ACEi during median follow-up period of 48 months; (2) the use of ARBs resulted in a significant reduction in adverse events in terms of MACCE during 12 months of follow-up; (3) ARBs showed improved clinical outcomes limited to female patients. These findings suggest that in patients who underwent CABG, the effects of the two discrete types of RAAS inhibitors may be comparable, but can differ according to follow-up period or in particular subgroups.
ACEi is the most widely prescribed type of RAAS inhibitors because of well-established evidence of their cardioprotective effects by suppressing of renin-angiotensin-aldosterone system activity, resulting in reduced mortality 2 . On the other hand, the clinical effects of ARBs for secondary prevention of ischemic heart disease are relatively unclear especially in particular subgroups of patients 6,12,13 . The recent CABG guidelines also recommend ACEi as the primary choice of RAAS inhibitor, especially when MI, LV dysfunction, diabetes mellitus, or chronic kidney disease is present 1 . However, up to 20% of patients reportedly show adverse reactions towards ACEi, such as the development of cough and angioedema, and according to the guidelines, ARBs should be considered as alternatives to ACEi in these patients [3][4][5]14 . Moreover, in Asian populations, the prevalence of cough induced by ACEi is higher compared to Caucasian populations 15 . Therefore, the withdrawal rate of ACEi is high, and the prescription of ARBs is relatively more common in Asian populations 15 . Our data also showed 11.2% of withdrawal rate. In this context, understanding the comparative effects of the two discrete RAAS inhibitors would be helpful for clinicians.
In this study, postoperative use of ARBs did not result in a different clinical outcome in patients who had undergone CABG relative to those prescribed ACEi during a median follow-up period of 48 months. This result is consistent with that of a recent meta-analysis of randomized trials comparing the effects of ARBs to ACEi in terms of MACCE in patients with heart failure (HF) 7 . However, previous results were inconsistent in other subgroups of patients with higher risk. While the OPTIMAAL (Optimal Trial in Myocardial Infarction with the Angiotensin II Antagonist Losartan) trial could not prove either 'superiority' or 'non-inferiority' of ARBs relative to ACEi 16 , a more recent study showed that the use of ARBs lowered the rates of cardiovascular events compared to ACEi 17 . The largest randomized controlled trial conducted to date also reported no significant difference in cardiovascular outcomes between the two types of RAAS inhibitors in patients at high risk for vascular events 18 .
Intriguingly, our results showed that ARBs may be more beneficial than ACEi under particular conditions. The incidence of MACCE was reduced in the ARB group during 12 months of follow-up, and based on numeric data, this reduction seemed to be mainly driven by the stroke prevention effect of ARB. Experimental evidence has shown that ARBs can prevent stroke by blocking the local angiotensin II type 1 receptor and simultaneously allowing angiotensin II to stimulate unoccupied angiotensin II type 2 receptors, increasing local blood flow to the brain and preventing the death of injured cells in ischemic areas 19 . In addition to experimental evidence, a www.nature.com/scientificreports www.nature.com/scientificreports/ recent meta-analysis also demonstrated a meaningful stroke prevention effect of ARBs in comparison to ACEi 20 . However, the benefit of ARBs over ACEi vanished to the level of insignificance during 48 months of follow-up. This could be explained by 'aldosterone-breakthrough' , which refers to the phenomenon whereby when RAAS inhibitors are used long-term, aldosterone levels are restored to baseline levels or reach even higher levels after an initial decline. It has been also reported that suppression of the RAAS is reversed in 10-53% of patients after 1 year of administration 21 . Considering that aldosterone is the end product of the RAAS and that it promotes tissue inflammation and injury in the cardiovascular and renal systems, this phenomenon explains the disappearance of the beneficial effect of ARBs compared to ACEi after a certain period of drug administration. Subgroup analysis revealed that the beneficial effects of ARBs persisted during 48-months of follow-up in female patients. This may be related to sex-based differences in the effectiveness of RAAS inhibitors, but needed further investigations 22,23 . In addition, the two types of RAAS inhibitors have shown discrete effects on other health issues such as cancer or infection 24,25 . However, these effects did not show a significant difference in our analysis.
Based on the results of this study, ARBs have a comparable effect to ACEi for secondary prevention after CABG, and therefore prescription of an ARB as an alternative to an ACEi seems reasonable for intolerant patients. In addition, ARBs may be considered as a primary choice for patients with a risk of stroke during the initial period after CABG. Our results also suggest that the gender of the patient might also alter the effects of RAAS inhibitor. However, well-designed further studies are needed to confirm our findings.
This study has several limitations. First, as a single-center retrospective study, unmeasured factors might have affected the results despite our efforts to adjust for all confounding factors by propensity score matching and multivariate logistic regression analysis. Second, the complexity of the procedure and post-operative care might have impacted clinical outcomes. Although we followed our institutional protocols, they have been updated during the course of our long study period. Lastly, types of ACEi or ARBs were not specified. Different types of RAAS inhibitors might have resulted in different outcomes. In addition, the dose and duration of RAAS inhibitor treatment and RAAS inhibitor side effects were not recorded. Although we performed an additional analysis after patients with discontinuation and changes of RAAS inhibitors during the first 12 months after CABG, prescription of RAAS inhibitors from outside the clinic may have been missed. Despite these limitations, our results provide valuable information for clinicians regarding the use of RAAS inhibitors for secondary prevention after CABG.

Conclusions
In CABG patients, ARBs may have comparable protective effects to ACEi and be a reasonable alternative for intolerant patients. The beneficial effects of ARBs depending on sex and follow-up period require further investigation.

Study population. The study protocol was approved by the Institutional Review Board of Samsung Medical
Center (IRB No. 2018-10-105) and conducted according to the guidelines of the Declaration of Helsinki. Considering that this was a retrospective study and risk to patients was therefore minimal, the need for individual consent was waived by the Institutional Review Board. From January 2001 to January 2015, 5453 consecutive adult patients who underwent CABG at Samsung Medical Center (Seoul, Korea) were initially enrolled. Inclusion criteria were patients who were prescribed for ACEi or ARBs at discharge after CABG, and the patients were classified into the ACEi or ARB group. The patients who were prescribed for alternative types of medication afterward were analyzed according to the intention-to-treat approach.
Data collection. We used Clinical Data Warehouse Darwin-C, which is an electric system designed to search for and collect data from electronic medical record systems, to extract all adult CABG patients and their prescriptions at discharge. After finalizing the list of the patients for the study, prescriptions at follow-up visits to the outpatient department were collected in the same manner to identify discontinuation of RAAS inhibitors, conversion to other types of RAAS inhibitors or replacement by other antihypertensives. Baseline characteristics were extracted automatically from electronic medical records with the aid of the institutional medical information department and were organized using a standardized form and protocol by a single investigator who was blinded to the prescription at discharge. Death of patients was determined by searching the national database. Other postoperative clinical outcomes and causes of death were collected through the manual review of each case by other investigators who were blinded to baseline characteristics.
Study outcomes and definitions. Clinical outcomes were defined as stated in a report on cardiovascular events in clinical trials by the ACC/AHA (American College of Cardiology Foundation/American Heart Association task force) 26 . The primary outcome was a MACCE. It is the composite of all-cause death, cardiac death, MI, repeat revascularization, and stroke during a median follow-up period of 48 months. Secondary outcomes included each composite of MACCE, admission for HF, and graft failure. Incidences of outcomes within 12 months were also compared. MI was defined according to the fourth universal definition 27 . Stroke was defined as a cerebrovascular accident which included both ischemic and hemorrhagic events with neurologic symptoms lasting at least 24 hours. Admission for HF was defined as a need for in-hospital treatment for at least 24 hours because of symptoms or objective evidence of new or worsening HF.
Statistical analysis. The significance of differences in continuous variables between groups was assessed using the Mann-Whitney's U-test if applicable. The Shapiro-Wilk test was used as a normality test. Chi-square or Fisher's exact tests were used for categorical variables. We selected covariates with a p-value under 0.05 and those that seemed to be clinically relevant. We included these covariates into multivariate Cox's regression analysis to adjust for the effects of confounding factors. The followings were covariates for adjustment: age, sex, hypertension, diabetes mellitus, chronic kidney disease, peripheral artery occlusive disease, left ventricular ejection fraction under 40%, off-pump CABG, history of old MI, and beta-blocker prescription. We also estimated HRs and   We used propensity score matching to generate a matched population to minimize the effects of potential confounding factors and selection bias. We assumed that covariates were well balanced when the absolute standardized mean difference between the matched groups was less than 0.1. In the propensity-matched population, we compared continuous variables with a paired t-test and categorical variables with a stratified chi-square test. We compared HRs for outcomes using univariate Cox's proportion hazard regression models in the matched population with or without Firth's penalized likelihood approach.
In addition to intention-to-treat analysis, we performed analyses in the population after excluding patients with changes of RAAS inhibitor use during the 12 months after CABG. We also performed a subgroup analysis using Cox's regression model to find any hidden interactions between risk factors and RAAS inhibitor use. All statistical analyses were performed with SAS 9.4 (SAS Institute Inc., Cary, NC, USA) and SPSS version 20 (SPSS, Inc., Chicago, IL, USA). All tests were two-tailed and assumed to be statistically significant if the p-value was less than 0.05. Continuous variables are presented as means ± standard deviations and categorical variables are expressed as percentages.