Raman, Betty

Importance Current risk prediction guidelines for hypertrophic cardiomyopathy predict only sudden cardiac death and are imperfect, leading to avoidable deaths and unnecessary implantable cardioverter defibrillators. Objective To combine prospectively collected clinical history, imaging, genetic, and biomarker data to improve risk prediction of adverse events in hypertrophic cardiomyopathy. Design, Setting, and Participants A total of 2750 patients with hypertrophic cardiomyopathy were prospectively enrolled in the registry-based study from 44 sites in North America and Europe with expertise in hypertrophic cardiomyopathy and cardiac magnetic resonance (CMR) imaging. Participants were enrolled from April 1, 2014, to April 7, 2017. Exposures Patients underwent a health history questionnaire, blood sampling for biomarkers and genotyping, and contrast-enhanced CMR. Patients were followed up yearly by telephone and through records review regarding event documentation. Main Outcomes and Measures The predefined composite adjudicated primary end point was time to first event for hypertrophic cardiomyopathy–related deaths; nonfatal sustained ventricular arrhythmias (VAs) requiring cardioversion or defibrillation; and left ventricular (LV) assist device implant or heart transplant. A secondary end point was a composite of sudden cardiac death and nonfatal VA events. The elastic-net method identified the most important predictors. Cox proportional hazards regression assessed associations with time to the first end point. Results Of the 2750 prospectively enrolled patients, 2698 (98%) had analyzable data after 9 were excluded because they had hypertrophic cardiomyopathy phenocopies and 43 withdrew. Of these remaining patients, 1919 (71%) were male, mean age was 50 years (SD, 11 years), and 423 (16%) were from underrepresented racial and minority groups. The mean follow-up was 6.9 years (SD, 2.1 years). The primary event model in 104 patients included LV scar as a percentage of LV mass by late gadolinium enhancement (LGE%; hazard ratio [HR], 1.86; 95% CI, 1.58-2.20; P < .001), LV mass index (HR, 1.09; 95% CI, 1.01-1.17; P = .03), LV end-systolic volume index (HR, 1.28; 95% CI, 1.12-1.46; P < .001 ), all per 10-unit increase, history of heart failure at study entry (HR, 2.89; 95% CI, 1.75-4.77; P < .001), and log N-terminal pro–B-type natriuretic peptide (NT-proBNP; HR, 1.41; 95% CI, 1.17-1.70; P < .001) level per log unit, (C index for all, 0.77). An LGE percentage of the LV mass of 9% or higher substantially increased the primary composite event rate ( P = .001). The secondary sudden cardiac death and VA risk factor model (in 69 patients) included LGE%, LV mass index, LV ejection fraction, and log(NT-proBNP) (C index, 0.76). Conclusions and Relevance These results provide prospective evidence for incorporating cardiac magnetic resonance and NT-proBNP in the evaluation of patients with hypertrophic cardiomyopathy. Trial Registration ClinicalTrials.gov Identifier: NCT01915615

Abstract Aims Recently, deep learning artificial intelligence (AI) models have been trained to detect cardiovascular conditions, including hypertrophic cardiomyopathy (HCM), from the 12-lead electrocardiogram (ECG). In this external validation study, we sought to assess the performance of an AI-ECG algorithm for detecting HCM in diverse international cohorts. Methods and results A convolutional neural network-based AI-ECG algorithm was developed previously in a single-centre North American HCM cohort (Mayo Clinic). This algorithm was applied to the raw 12-lead ECG data of patients with HCM and non-HCM controls from three external cohorts (Bern, Switzerland; Oxford, UK; and Seoul, South Korea). The algorithm’s ability to distinguish HCM vs. non-HCM status from the ECG alone was examined. A total of 773 patients with HCM and 3867 non-HCM controls were included across three sites in the merged external validation cohort. The HCM study sample comprised 54.6% East Asian, 43.2% White, and 2.2% Black patients. Median AI-ECG probabilities of HCM were 85% for patients with HCM and 0.3% for controls (P < 0.001). Overall, the AI-ECG algorithm had an area under the receiver operating characteristic curve (AUC) of 0.922 [95% confidence interval (CI) 0.910–0.934], with diagnostic accuracy 86.9%, sensitivity 82.8%, and specificity 87.7% for HCM detection. In age- and sex-matched analysis (case–control ratio 1:2), the AUC was 0.921 (95% CI 0.909–0.934) with accuracy 88.5%, sensitivity 82.8%, and specificity 90.4%. Conclusion The AI-ECG algorithm determined HCM status from the 12-lead ECG with high accuracy in diverse international cohorts, providing evidence for external validity. The value of this algorithm in improving HCM detection in clinical practice and screening settings requires prospective evaluation.