Software-based analysis of T-wave morphology: identifying the electrocardiogram signature of high-risk long QT syndrome.

AIMS: Despite T-wave morphology abnormalities being well-known distinctive ECG features in patients with long QT syndrome (LQTS), the subjectivity of qualitative 'eyeballing' in T-wave characterization still hampers its integration into diagnostic/prognostic criteria. We herein evaluated whether our quantitative software-based analysis of T-wave morphology (AnTwM) applied to digital ECGs may identify predictors of cardiac events (CEs) in our cohort of LQTS patients. METHODS AND RESULTS: We enrolled LQT1, LQT2, and LQT3 patients having at least one digital ECG from our cohort of genotype-confirmed LQTS patients. Automated AnTwM analysis, using Glasgow and Bravo algorithms embedded in the CalECG software (AMPS-IIc, USA), provided scalar descriptors of ventricular repolarization. Cox regression analyses identified potential predictors of CEs (i.e. syncope, sudden cardiac death, resuscitated cardiac arrest, or appropriate shock delivered by implantable cardioverter defibrillators). A total of 467 (58% female) patients were followed up for 15 ± 9 years, including 253 (54.2%) LQT1, 182 (39%) LQT2, and 32 (6.8%) LQT3 patients. Corrected QT interval predicted CEs in the whole population (1 ms QTc increase: HR = 1.01, 95% CI: 1.0-1.01, P = 0.03) but not across genotyped subpopulations. Genotype-specific ECG markers associated with a greater risk of CEs were (i) those expressing a delayed accumulation of the mid-late T-wave area (decreased t25 and increased t50) among LQT1 patients and (ii) those expressing T-wave flattening/widening (decreased T-wave ascending/descending slopes) among LQT2 patients. CONCLUSION: The software-based AnTwM on digital ECGs represented a reliable tool in clinical practice and identified unique ECG T-wave 'fingerprints' that allowed prediction of CEs in a genotype-specific manner.