Prediction of CRT response using a novel CRT responder score in an Asian patient population

Abstract

Abstract Introduction Cardiac Resynchronization Therapy (CRT) guidelines, derived from trials among U.S. and European patients, assign weaker recommendations to those with midrange QRS duration (<150ms). Patient height and body size may modulate CRT efficacy. Moreover, these factors may affect CRT utilization and efficacy in Asia. Aim The purpose of this analysis was to determine the prediction scoring system model for CRT responders in a Japanese patient population. Methods We analyzed the database of the Multicenter Prospective Pilot Study to Test LV Intracardiac Conduction Time as a Predictor of CRT Response (BIO|SELECT) which was a single-arm study conducted at 28 hospitals in Japan. This study enrolled patients eligible for a de-novo CRT-defibrillator (CRT-D) based on current guideline indications, who had no recent or planned cardiac surgery and were planned to receive a Sentus ProMRI OTW quadripolar LV lead and a BIOTRONIK CRT-D with multipole pacing feature. Patients were followed for 7 months after implantation. Four accepted parameters to define CRT response [LVESV (cut-off -15%), LVEF (+5%), BNP (-25%), and NYHA class (one class improvement)] were combined into a composite benefit index (CBI) by forming the normalized sum, with CBI = 4.0 meaning response. The CBI was found to be more robust than the individual 4 parameters due to the reduced influence of noise and measurement errors in single parameters. Several factors were studied to create the scoring tool. The CRT responder score was the sum of these individual scores (range 0-100). To construct a predictive scoring system, logistic regression analysis was performed with a backward selection procedure. In addition, the Akaike information criterion (AIC) was applied to determine the best model among those tested. The score was then assigned employing the β coefficient of each variable divided by the reference value. The model's calibration was assessed using the Hosmer-Lemeshow goodness-of-fit test. Results A total of 196 patients were enrolled in this study (mean age 69 years, mean LVEF 30%, left bundle-branch block [LBBB] 58%). The Ooccurrence of CRT response during the 7-month follow-up was 54%. Future CRT responders were significantly predicted by low age (< 70 years)(18 points), female (19 points), prior heart failure hospitalization (16 points), body mass index (<20) (17 points), QRS/height (<0.858)(30 points). This score predicted CRT response well (area under curve 0.72; 95% CI 0.643-0.798). When the cut-off of 63 points or more was applied the sensitivity was 0.56, the specificity was 0.72, and the negative predictive value was 0.58. Conclusions The CRT responder score is a novel approach for the prediction of CRT response, but should be validated and further improved to increase its predictive value. This approach highlights the role of preventing the development of severe heart failure and could help to identify individuals who might benefit from CRT.