Frontal T-wave axis deviation and risk of sudden cardiac death in coronary syndromes.

Abstract

Frontal T-wave axis (TA) and QRS-T angle (QRSTA) are automated ECG measures that reflect depolarisation-repolarisation abnormalities-and have been linked to mortality. We tested whether increasing deviation in TA/QRSTA predicts adjudicated sudden cardiac death (SCD) events in patients undergoing coronary angiography-elective (chronic coronary syndrome) and acute (acute coronary syndrome)-and whether signals differ when ECGs are recorded post-acute. Retrospective consecutive cohort at Tampere University Hospital, 2007-2018. Exposures were automatically measured frontal TA and QRSTA (Marquette 12SL). Primary endpoint was SCD event (true SCD, resuscitated SCA or implantable cardioverter defibrillator-terminated VF/very fast ventricular tachycardia) adjudicated to guideline definitions, with Fine-Gray competing-risk models adjusted for clinical covariates. Follow-up through 31 December 2022. We analysed 18 828 patients (elective n=10 303; acute n=8525; median follow-up 9.0 and 7.8 years). In the elective cohort, SCD hazard increased monotonically with TA/QRSTA deviation; severely abnormal TA (<-75° or >165°) HR 4.06 (95% CI 2.94 to 5.61) and severely abnormal QRSTA (≥151°) HR 3.16 (2.35-4.24). Associations persisted after excluding left ventricular ejection fraction (LVEF) ≤35%. In the acute cohort, effects were smaller and non-monotonic (severely abnormal TA HR 2.01 (1.42-2.86); QRSTA HR 1.73 (1.24-2.39)). In a sensitivity cohort with ECGs 30-365 days postangiography (acute 30, n=5529), the TA-SCD association resembled the elective setting (severely abnormal TA HR 2.69 (1.70-4.24)); QRSTA was not significant. Severity of frontal TA deviation is a robust, graded marker of SCD risk in angiography patients, strongest in elective and postacute settings; QRSTA is supportive but weaker. These automated, low-cost metrics could augment SCD risk stratification beyond LVEF, warranting external validation and clinical utility testing.