Complete revascularization is associated with higher mortality in patients with ST-elevation myocardial infarction, multi-vessel disease and shock defined by hyperlactataemia: results from the Harefield Shock Registry incorporating explainable machine learning

Abstract

Aims: Revascularization strategy for patients with ST-elevation myocardial infarction (STEMI) and multi-vessel disease varies according to the patient's cardiogenic shock status, but assessing shock acutely can be difficult. This article examines the link between cardiogenic shock defined solely by a lactate of ≥2 mmol/L and mortality from complete vs. culprit-only revascularization in this cohort.

Methods and results: Patients presenting with STEMI, multi-vessel disease without severe left main stem stenosis and a lactate ≥2 mmol/L between 2011 and 2021 were included. The primary endpoint was mortality at 30 days by revascularization strategy for shocked patients. Secondary endpoints were mortality at 1 year and over a median follow-up of 30 months. Four hundred and eight patients presented in shock. Mortality in the shock cohort was 27.5% at 30 days. Complete revascularization (CR) was associated with higher mortality at 30 days [odds ratio (OR) 2.1 (1.02-4.2), P = 0.043], 1 year [OR 2.4 (1.2-4.9), P = 0.01], and over 30 months follow-up [hazard ratio (HR) 2.2 (1.4-3.4), P < 0.001] compared with culprit lesion-only percutaneous coronary intervention (CLOP). Mortality was again higher in the CR group after propensity matching (P = 0.018) and inverse probability treatment weighting [HR 2.0 (1.3-3.0), P = 0.001]. Furthermore, explainable machine learning demonstrated that CR was behind only blood gas parameters and creatinine levels in importance for predicting 30-day mortality.

Conclusion: In patients presenting with STEMI and multi-vessel disease in shock defined solely by a lactate of ≥2 mmol/L, CR is associated with higher mortality than CLOP.

Keywords: Cardiogenic shock; Machine learning; Revascularisation strategy; STEMI; XGBoost.