Abstract 15862: Risk Prediction Model for Early Detection of Left Bundle Branch Associated Nonischemic Cardiomyopathy: Insights From the UK Biobank

Abstract

Introduction: Ventricular dyssynchrony resulting from left bundle branch block (LBBB) can lead to development of nonischemic cardiomyopathy (NICM) in some individuals. Increasing evidence suggests early treatment with cardiac resynchronization therapy yields a greater treatment effect while a delayed diagnosis of LBBB NICM may fail to capture the critical period to reverse the progressive myocardial damage. Early identification of LBBB patients most likely to develop NICM would be valuable to clinical management. Methods: International classification of diseases (ICD) codes from the linked health data in the UK Biobank were used to identify individuals with LBBB NICM along with baseline demographics and comorbidities. Logistic regression models adjusted for covariates in baseline characteristics were used to identify early predictive indicators based on electrocardiogram (ECG) parameters and physical activity data obtained at time of enrollment into the UK Biobank Results: Among 3,766 individuals with LBBB, 481 developed LBBB NICM. The adjusted logistic regression analyses demonstrated a one standard deviation increase in p-axis was associated with a 128% higher incidence of LBBB NICM (OR 2.28, CI 1.10-4.73, p=0.01) and an increase of one day per week of vigorous physical activity was associated with an 8% lower incidence of LBBB NICM (OR 0.92, CI 0.87-0.98, p=0.006). A risk prediction model using the predictive indicators of p-axis and vigorous activity per week yielded an area under the receiver operating characteristic curve of 0.87. Conclusions: A risk prediction model using ECG parameters and physical activity was developed which provides personalized estimation of the likelihood of development of NICM in individuals with LBBB using readily accessible clinical parameters. Further investigations are necessary to refine risk stratification strategies, validate this in a larger population, and optimize patient management based on these findings.