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.