Circulating excitation-contraction coupling proteins and incident cardiovascular outcomes: associations and incremental predictive performance in the UK Biobank.

INTRODUCTION

Excitation-contraction coupling is central to cardiovascular physiology, but the prospective relevance of circulating proteins involved in this pathway remains uncertain.

METHODS

We evaluated a prespecified seven-protein excitation-contraction coupling panel (EC7: NEXN, SLMAP, ITPR1, CACNB1, HRC, RYR1, and CACNA1C) in the UK Biobank proteomics cohort in relation to incident stroke, heart failure, acute myocardial infarction, and angina recorded during the available observation period. Outcome-specific logistic-regression models estimated odds ratios per 1-standard-deviation higher relative protein level after adjustment for demographic and clinical factors, with false-discovery-rate correction across 28 primary tests. Predictive analyses used locked training/test partitions, screening of 13 candidate algorithms, nested cross-validation, frozen calibration and thresholds, and comparisons with clinical features and five established cardiovascular proteins.

RESULTS

Among 52,995 participants in the analysis source proteomics cohort, 39,691 had complete EC7 measurements. Higher ITPR1 (odds ratio, 1.088; 95% confidence interval, 1.033-1.146; q = 0.022) and HRC (odds ratio, 1.246; 95% confidence interval, 1.191-1.305; q < 0.001) were associated with incident heart failure; no other primary association survived correction. Logistic regression was selected for all outcomes. Adding EC7 proteins to clinical features did not improve locked-test predictive performance after multiplicity correction; for heart failure, changes were +0.0054 in ROC-AUC (q = 0.112) and +0.0084 in average precision (q = 0.217). Established comparator proteins consistently outperformed EC7 in biomarker-only comparisons.

DISCUSSION: Circulating excitation-contraction coupling proteins showed outcome-specific associations, particularly with heart failure, but limited incremental predictive performance. These findings support further mechanistic investigation and validation in independent cohorts.