Background Early prediction of cardiovascular (CV) events in the general population remains an important issue. The T-wave morphology restitution (TMR), an ECG marker quantifying ventricular repolarization dynamics, is strongly associated with CV mortality in heart failure patients. Our objective was to evaluate the CV prognostic value of TMR in the general population and identify any genetic contribution.

Methods ECG recordings from 56,780 healthy individuals undergoing exercise stress testing in the UK Biobank study (EST-UKB) were analyzed. TMR was computed for exercise (TMRex) and recovery from exercise (TMRrec). The primary endpoint was CV death or hospitalizations for CV reasons. The secondary and tertiary endpoints were (1) all-cause mortality or hospitalizations for CV reasons and (2) arrhythmic mortality or hospitalizations for arrhythmic reasons. The median follow-up time was 70.7 months. Genome-wide association studies for TMRex and TMRrec were also performed and genetic risk scores (GRSs) were derived and tested for association with endpoints in the full cohort (FULL-UKB; N=402,746, median follow-up time of 85.3 months).

Results 1,727 (3.0%) individuals met the primary endpoint in EST-UKB, and 2,326 (4.1%) and 120 (0.2%) met the secondary and tertiary endpoints, respectively. TMRrec was significantly associated with the primary endpoint (hazard ratio (HR) 1.15, P=2 × 10–10, table 1), and both secondary and tertiary endpoints (HR 1.13, P=2 × 10–11 and HR 1.28, P=1 × 10–4, respectively) independent of resting QTc, and resting and recovery heart rate. Despite relatively low heritability (∼5%), we identified 12 genetic loci associated with TMRex and TMRrec, of which nine had been previously associated with other ECG markers. Individuals meeting the primary and secondary endpoints in FULL-UKB (21,328, 5.3% and 28,536, 7.1%, respectively) had higher GRS for TMRrec than unaffected individuals (P=0.026 for both endpoints). No association was found between TMRex or TMRrec and the tertiary endpoint. Individuals in the top 20% of the GRS had significantly higher risk of meeting the primary and secondary endpoints than those in the bottom 20% (HR 1.04, P=0.048, HR 1.04, P=0.024, respectively).

Conclusion TMR and TMR GRSs are significantly associated with CV events and all-cause mortality, supporting the hypothesis that increased spatio-temporal heterogeneity of ventricular repolarization identifies a high-risk substrate for CV events. This work provides support for inclusion of clinical variables, non-invasive markers and GRSs to improve personalised treatment decisions.

Conflict of Interest None