216-73: QT/RR regression correlation coefficient as non-invasive risk stratification tool Free

Introduction: An increased beat-to-beat variability of repolarization as measure of decreased cellular repolarization reserve has been linked to arrhythmia. We hypothesize that the linear regression correlation coefficient of a 24h QT/RR plot might approach the same phenomenon and could be used as a predictor of arrhythmic death.

Methods: All primary prevention ICD recipients with ischemic cardiomyopathy at the University Hospitals of Leuven with a 2 lead 24h holter (ELA, Sorin) available prior to ICD implant were included in the analysis. Patients with paroxysmal or persistent AF on holter were excluded. The QT/RR linear regression analysis was performed (Synescope, Sorin) after precise manual beat classification and template correction with calculation of slopes and correlation coefficients (QT/RRcorr). Multivariate Cox regression modelling was performed with clinical and other non-invasive risk parameters as deceleration capacity, HRT, HRV.

Results: In total, 89 patients were included (age 57 ± 13y, male 90%, LVEF 27 ± 9%) with a mean follow-up of 4.0 ± 2.9 y. Of these 23 (25.8%) died and 29 (32.6%) received a first appropriate shock. Of these 12 (13.5%) patients were excluded because of limited data quality.

ROC analysis of QT/RRcorr showed an AUC of 0.680 for predicting AS within 1y after ICD implant (n = 11, 12.4%), for the complete follow-up AUC was 0.537. Dichotomization was performed at a value of r = 0.80 with 68% sensitivity and 73% specificity for predicting AS within 1y after implant, for all follow-up this was 66% and 41% respectively.

After univariate Cox regression analysis QT/RRcorr ≤ 0.8 was a univariate predictor of AS within 1y after implant (p = 0.024, HR 4.62, 95% CI 1.22-17.43), not for mortality either within 1y (p = 0.876) or overall (p = 0.358).

Multivariate analysis identified QT/RRcorr ≤ 0.8 as an independent predictor of AS within 1y after implant (p = 0.020, HR 8.08, 95% CI 1.40-46.5), together with age, presence of non-sustained VT or VT on holter and LFnu.

Conclusion: After meticulous manual correction of 24h holter recordings a low QT/RR linear correlation coefficient, as a possible measure of higher beat-to-beat QT variability, discriminated a high risk of arrhythmia from a risk of dying within ischemic cardiomyopathy ICD recipients.

Disclosure: The research leading to these results has received funding from the European Community's Seventh Framework Program FP7: EU-CERT-ICD (grant agreement no. HEALTH-F2-2013-602299).