Effects of ACE Inhibitors, Angiotensin Receptor Blockers and Angiotensin Receptor-Neprilysin Inhibitors on Survival Free from Gastrointestinal Bleeding in HeartMate 3 Patients - A Marginal Structural Modeling Analysis

Abstract

<h3>Purpose</h3> The HeartMate 3 (HM3) is superior to HMII in survival free from disabling stroke and pump exchange, however, rates of GIB remain high. Prior studies show a protective effect of digoxin and ACEi/ARB on risk for GIB, whereas phosphodiesterase 5 inhibitors (PDE5i) have opposite effect in HMII using Cox Proportional Hazard (PH) models with inverse probability weighting (IPW). Compared to this approach, marginal structural modeling (MSM) accounts for time-dependent confounding thereby producing results closer to true causal effect. Targeted Maximum Likelihood Estimation (TMLE) can help estimate MSM effects making minimal assumptions on data distribution. We studied the impact of outpt medications on survival free from GIB in HM3 pts using TMLE-MSM. <h3>Methods</h3> Data were retrospectively collected in HM3 pts up to 18 mo after discharge from implant and summarized in 3-mo time intervals. Medication status was defined as receiving vs. not ACEi/ARB/ARNi, PDE5i and digoxin at each time-node. Medication use was calculated by summarizing duration in days within each time node. Creatinine, potassium, INR and mean arterial pressure of each node were averaged. We fit TMLE-MSMs and compared results to IPW Cox PH models using both time variant and invariant covariates. Endpoint was survival free from GIB. <h3>Results</h3> 128 HM3 pts (60±13 y, 16% F) were studied. 5 deaths and 19 first GIB events occurred in 22 (17.2%) pts during follow up. 60 pts were exposed to ACEi/ARB/ARNi, 43 to digoxin and 39 to PDEi5. There was no effect of ACEi/ARB/ARNi, PDE5i or digoxin on the endpoint when using IPW Cox PH model. When using TMLE-MSM a statistically significant effect between ACEi/ARB/ARNi and the endpoint was observed (Table). <h3>Conclusion</h3> ACEi/ARB/ARNi improves survival free from GIB in HM3 pts using TMLE-MSM. Prospective randomized studies are needed to further test the impact of pharmacotherapy on outcomes in HM3 pts.