Individualized transfusion decisions to minimize adverse cardiovascular outcomes in patients with acute myocardial infarction and anemia.

Background: Risk-benefit tradeoffs between restrictive versus liberal red blood cell transfusion strategies may vary across individuals. This exploratory analysis aimed to derive and evaluate individualized treatment effects of defined transfusion strategies in patients with acute MI and anemia with the goal of minimizing adverse cardiovascular outcomes.

Methods: This study analyzed 3,447 (98.4%) patients randomized in the MINT (Myocardial Ischemia and Transfusion) trial between April 2017 to April 2023. Outcomes for this analysis included 30-day death or recurrent MI, death, and major adverse cardiovascular events (MACE, a composite of death, MI, stroke, and ischemia-driven unscheduled revascularization). Machine learning methods were used to identify baseline patient characteristics that informed the individualized treatment effect of a restrictive versus liberal transfusion strategy for each patient. The expected population risk of an outcome under a scenario in which patients received their optimal treatment, as indicated by the individualized treatment effect, was contrasted with expected risks for universally applying a restrictive strategy or a liberal strategy to all patients.

Results: Baseline characteristics did not inform individualized treatment effects on 30-day death and death or MI, suggesting minimal heterogeneity in treatment effect on these outcomes. An algorithm for estimating the individualized treatment effect on 30-day MACE included 12 baseline factors. If all patients received the optimal treatment as indicated by their estimated individualized treatment effect, the predicted risk of 30-day MACE in the sample population was 15.2% (95% CI 14.2% to 16.2%). This corresponded to 4.0 (difference: -4.0%, 95% CI -5.8, -2.1) and 2.3 (difference: -2.3%, 95% CI -3.7, -0.9) percentage point risk reductions compared to applying a restrictive or liberal strategy to everyone respectively.

Conclusions: The MINT trial average treatment effect, favoring a liberal strategy, may be optimal to minimize risk of 30-day death and death or MI for acute MI patients with anemia represented in the MINT sample as no individualized treatment effects were estimated on these outcomes. However, individualized transfusion strategy decisions have potential to reduce risk of 30-day MACE. External validation of the MACE algorithm is required before clinical use.

Trial Registration: ClinicalTrials.gov, NCT02981407, https://clinicaltrials.gov/study/NCT02981407.