AI-ECGs may help evaluate treatment response in obstructive hypertrophic cardiomyopathy
Artificial intelligence to assess changes in ECGs may represent a novel method to evaluate disease status and treatment response to mavacamten in patients with obstructive hypertrophic cardiomyopathy, researchers reported.
“Although hypertrophic cardiomyopathy (HCM) causes significant morbidity and is a leading cause of sudden death in adolescents, initial detection remains difficult,” Geoff Tison, MD, MPH, cardiologist and assistant professor in the division of cardiology at the University of California, San Francisco (UCSF), and colleagues wrote. “Although echocardiography is an important diagnostic modality for HCM, the ECG is more widely accessible. AI-based analysis of standard 12-lead ECG (AI-ECG) has achieved an accurate fully automated diagnosis of HCM. However, it is unknown whether AI-ECGs can track disease status, including cardiac structural and hemodynamic changes over time, to inform disease-specific clinical decision-making.”
For the present study, published in the Journal of the American College of Cardiology, researchers evaluated data from the PIONEER-OLE trial using two AI-ECG algorithms trained to assess pretreatment and on-treatment ECGs of patients with obstructive HCM assigned mavacamten (Bristol Myers Squibb).
The two algorithms were independently developed and trained at UCSF and Mayo Clinic, according to the study.
As Healio previously reported, 1 year of mavacamten improved left ventricular outflow tract gradients, symptoms and N-terminal pro-B-type natriuretic peptide levels in patients with obstructive HCM.
For the PIONEER-OLE trial, participants (mean age at baseline, 58 years; 69% men; 92.3% NYHA class II) had 12-lead ECGs, echocardiography, drug plasma and NT-proBNP measurements taken at pretreatment and on-treatment at weeks 4, 8, 12 and every 12 weeks thereafter, according to the study. Median follow-up was 79 weeks.
Researchers reported a sensitivity and specificity of 84.6% and 96.3%, respectively, for the UCSF AI-ECG algorithm and 92.3% and 94.1%, respectively, for the Mayo Clinic algorithm.
When the AI algorithms were applied to the PIONEER-OLE participant ECGs, both demonstrated decreased mean HCM scores across each time point during the on-treatment period, with a mean reduction of 43% observed with the UCSF AI-ECG algorithm and a reduction of 56% with the Mayo Clinic algorithm.
Moreover, the decreasing HCM score trends mirrored the trends over time for LV outflow tract gradient with Valsalva and NT-proBNP measurements, according to the study.
“The key finding from this study was that AI-ECG HCM scores correlated with disease status as measured by decreases over time in left ventricular outflow tract gradients and NT-proBNP levels in patients with obstructive HCM on mavacamten treatment,” the researchers wrote. “Therefore, AI-ECGs might hold promise as a potential tool for monitoring disease status, cardiac hemodynamics, and drug therapeutic response.
“These significant longitudinal associations of the AI-ECG HCM score likely reflect changes in the raw ECG waveform detectable by AI-ECGs that correlate with HCM disease pathophysiology and severity,” the researchers wrote.
Collapse