Comparative effectiveness and safety of proton pump inhibitors after myocardial infarction on dual antiplatelet therapy: Insights from a target trial emulation.

BACKGROUND

Current guidelines recommend co-prescribing proton pump inhibitors (PPIs) with dual antiplatelet therapy (DAPT) after acute myocardial infarction (MI) to prevent gastrointestinal bleeding. Emerging evidence indicates that PPIs have distinct pharmacokinetic profiles, and their cardiovascular and bleeding effects may not represent a class effect.

METHODS

We used electronic health record data from the Tianjin Health and Medical Data Platform to emulate target trials comparing effectiveness and safety of three commonly used PPIs in clinical practice. The effectiveness outcome was gastrointestinal bleeding, and the safety outcome was composite cardiovascular events. Cox proportional hazards model was used to estimate hazard ratios (HRs) and 95% confidence intervals (CI) between individual PPIs. We further applied XGBoost-based prognostic models to assess treatment effect heterogeneity across strata of predicted baseline risk.

RESULTS

Of 51 072 MI patients on DAPT, 22 671 were prescribed rabeprazole, 18 081 pantoprazole and 10 320 lansoprazole. During 1 year of follow-up, both pantoprazole (HR, 1.05 [95% CI, 0.97-1.13]) and lansoprazole (HR, 0.99 [0.91-1.09]) showed comparable cardiovascular risk to rabeprazole, but pantoprazole was associated with an increased risk of recurrent MI (HR, 1.20 [1.07-1.35]). For gastrointestinal bleeding risk, lansoprazole showed lower risk (HR, 0.81 [0.70-0.93]) compared with rabeprazole. Results were robust in multiple sensitivity analyses, and treatment effect heterogeneity showed gastrointestinal benefits of lansoprazole increase with the rising baseline bleeding risk.

CONCLUSIONS

In patients with acute MI receiving DAPT, the three PPIs showed similar cardiovascular safety, while lansoprazole was associated with a lower risk of gastrointestinal bleeding. External validation is warranted to examine these findings.