The renoprotective effect of esaxerenone independent of blood pressure lowering: a post hoc mediation analysis of the ESAX-DN trial

The reporting of this post hoc mediation analysis is in accordance with the consensus-based guidance for the reporting of mediation analyses of randomized trials and observational studies (A Guideline for Reporting Mediation Analyses; AGReMA Statement) [22].

Study design and population

This was a post hoc analysis of a multicenter, randomized, double-blind, placebo-controlled phase 3 trial (JapicCTI-173695), the ESAX-DN study [15]. The study protocol of the ESAX-DN study was approved by the local institutional review board at each participating site and conducted in accordance with the principles of the Declaration of Helsinki and the ICH E6 Guideline for Good Clinical Practice (CPMP/ICH/135/95). All the participants in the ESAX-DN study provided written informed consent prior to enrollment.

In the ESAX-DN study, patients were randomized to either esaxerenone or placebo treatment for 52 weeks after a 4-week run-in period, continuing their treatment with RAS inhibitors at a constant dosage throughout the study. To minimize the risk of increasing serum potassium (K+) levels, esaxerenone or its placebo treatment was started at a dosage of 1.25 mg/d and then titrated to 2.5 mg/d if the serum K+ level of each patient was acceptable. Other details of the study population, such as the inclusion and exclusion criteria, have been previously published [15].

Outcomes

The primary endpoint of the ESAX-DN study was the proportion of patients with UACR remission at the end of treatment, and the key secondary endpoint was the percentage change in the UACR from baseline at the end of treatment. The changes in BP and creatinine levels and the rate of transition to overt albuminuria were also assessed. Safety endpoints included adverse events, serum K+ levels, percentage changes in eGFR from baseline to the end of treatment and time-course changes in eGFR.

Of these outcomes, data for the percentage changes in the UACR from baseline to the end of treatment, changes in BP, and changes in eGFR were used in this post hoc analysis.

Assessments

The UACR, calculated from first morning urine samples, was measured during the run-in period and every 4 weeks up to week 52 during the subsequent treatment period. BP and eGFR were monitored every 2 weeks up to week 8 and every 4 weeks from week 12 to week 52 during the treatment period. The eGFR with the modification in diet in renal disease was calculated using the formula modified by the Japanese Society of Nephrology [23]. Other details of the study visits and assessments have been previously reported [15].

Clinical assumptions

Several mechanisms may explain the renoprotective effect of esaxerenone. First, esaxerenone primarily reduces BP, which then leads to a reduction in glomerular pressure, and it can also reduce urinary albumin excretion. Second, esaxerenone may directly affect renal function and may also affect urinary albumin excretion. In addition to these mechanisms, esaxerenone may have other effects on albuminuria suppression.

Based on these clinical assumptions, we considered SBP and eGFR as mediators that reflect the changes in BP and renal function and assumed the following causal relationships between the variables in this analysis (Fig. 1a). There are four possible pathways through which treatment affects UACR changes: pathways through SBP or eGFR (i.e., treatment-SBP-UACR, treatment-SBP-eGFR-UACR, treatment-eGFR-UACR) and a pathway through neither SBP nor eGFR (i.e., treatment-UACR). The former are indirect effects, and the latter the direct effect (our primary interest in this analysis). When SBP or eGFR are considered separately, the causal relationships are simplified to only two pathways: an indirect effect through SBP or eGFR and a direct effect not through SBP or eGFR (Fig. 1b, c).

Fig. 1figure 1

Causal relationships between variables. The graph displays the assumptions about causal relationships between variables. If an arrow points away from X and toward Y, it indicates that X causally affects Y

The values of SBP and eGFR change longitudinally during treatment. In addition, SBP and eGFR may affect UACR changes in different manners. Therefore, to incorporate the changes in SBP and eGFR during treatment, two types of mediator variables were considered: the cumulative average value of change from baseline up to the end of treatment and the achieved value of change from baseline to just before the end of treatment. The former assumes a cumulative effect, and the latter assumes an acute effect of mediator variables on the outcome.

Statistical analysis

To quantitatively investigate direct and indirect effects, we used the concepts of the “natural direct” effect and the “indirect effect” [20]. The natural direct effect is defined as the between-treatment comparison of the effect on the outcome if the mediator levels were set to what they would have been if the control treatment (e.g., placebo) was initiated. The natural indirect effect is defined as the comparison of the mediator effect on the outcome between mediator levels that would have been observed under the experimental or control treatment while the treatment was set to the experimental treatment (e.g., esaxerenone). The sum of natural direct and indirect effects is the total effect defined as the treatment effect comparison between when all subjects had experimental treatment and when all subjects had control treatment, the so-called intention-to-treat (ITT) effect in randomized trials. Further detailed explanations based on the causal inference framework are included in Supplementary Text 1.

Some strong assumptions are required to identify natural direct and indirect effects: (i) no unmeasured confounding of the exposure-outcome relationship; (ii) no unmeasured confounding of the outcome-mediator relationship; (iii) no unmeasured confounding of the exposure-mediator relationship; and (iv) no mediator-outcome confounder that is itself affected by the exposure. In randomized studies such as the ESAX-DN study, assumptions (i) and (iii) would hold because of the randomization of the treatment, but (ii) and (iv) would not necessarily hold and cannot be confirmed by the data.

To overcome this problem with the assumptions, a methodology with multiple mediators was considered in this analysis. The confounder for the mediator-outcome relationship was considered the mediator, and the joint indirect effect of all mediators was estimated [20]. Another advantage of this method is that we do not need to specify the causal relationships between mediators (i.e., BP and eGFR in this analysis).

To estimate the natural direct and indirect effects, we used the regression-based approach for single mediator and multiple mediator settings (details are shown in Supplementary Text 2). In the multivariate linear regression model, we included the treatment, mediator(s), interaction between treatment and mediator(s), and baseline covariates (i.e., baseline values of the UACR and mediators) as confounders.

In addition to the direct and indirect effects, the proportion of the mediated effect was calculated as (indirect effect/total effect) × 100 (%), which expresses how much of the total effect is mediated through mediators. The CIs for estimates were constructed based on the bootstrap method with 1000 replications.

As in the ESAX-DN study, the log-transformed UACR was used in the analyses and back-transformed and expressed in the original scale in the results (i.e., the effect on UACR reduction was expressed as the geometric mean ratio to baseline, compared with placebo). All statistical analyses were performed using SAS System Release 9.4 (SAS Institute Japan Ltd., Tokyo, Japan).

留言 (0)

沒有登入
gif