ABSTRACT
Artificial intelligence-enhanced electrocardiography (AI-ECG) interpretation can identify hypertrophic cardiomyopathy (HCM) on 12-lead ECGs and is a novel way to monitor response to treatment. While initial therapies were procedural – i.e. the surgical or percutaneous reduction of the interventricular septum (SRT), medical therapy with mavacamten offers an oral alternative. To evaluate biological response to SRT, we applied an AI-ECG model for HCM detection to ECG images obtained from the same patients before and after SRT across three sites: Yale New Haven Health System (YNHHS), Cleveland Clinic Foundation (CCF), and Atlantic Health System (AHS). We also applied the model in ECGs from patients who had received mavacamten at YNHHS over the past year since its approval. A total of 70 patients underwent SRT at YNHHS, 100 patients at CCF, and 145 at AHS. Among patients at YNHHS, there was no significant change in the AI-ECG score before versus after SRT (pre-SRT: 0.55 [IQR 0.24–0.77] vs post-SRT: 0.59 [IQR 0.40–0.75], p = 0.42 by Wilcoxon signed-rank test). The AI-ECG HCM scores were higher post SRT at CCF (0.61 [IQR 0.32–0.79] vs 0.69 [IQR 0.52–0.79], p = 0.003) and AHS (0.52 [IQR 0.35– 0.69] vs 0.61 [IQR 0.49–0.70], p < 0.001). At YNHHS, among 36 patients undergoing treatment with Mavacamten, the median AI-ECG score prior to starting mavacamten was 0.46 (IQR 0.21–0.72). At the end of a median follow-up period of 237 days (IQR 140–338), the median AI-ECG score decreased significantly to 0.28 (IQR 0.11–0.50, p = 0.002). Our results demonstrate a measurable difference in the myocardial effects of SRT and mavacamten detectable on ECGs, which can reflect different mechanisms for modification of the HCM substrate. Our multicenter design also demonstrates the feasibility of an image-based AI-ECG approach for disease monitoring in HCM at the point-of-care.
BACKGROUND
Artificial intelligence-enhanced electrocardiography (AI-ECG) interpretation can identify hypertrophic cardiomyopathy (HCM) on 12-lead ECGs and is a novel way to monitor response to treatment. In a phase-2 clinical trial of obstructive HCM (oHCM), the AI-ECG phenotype for HCM improved on serial ECGs among patients receiving mavacamten.1 Mavacamten is the first cardiac myosin inhibitor to have shown efficacy in relieving hemodynamic obstruction and improving health status.2 Before the emergence of mavacamten, septal reduction therapy (SRT), which includes alcohol septal ablation and ventricular myectomy, was the key treatment for patients experiencing symptomatic limitations despite optimal medical therapy.3 We leveraged AI-ECG to evaluate disease modification by SRT for oHCM. For this, we deployed a model that detects HCM from images of 12-lead ECGs and evaluated ECGs from cohorts of individuals who underwent SRT at three distinct institutions. We used mavacamten recipients at one of these institutions as a positive control.
METHODS
The AI-ECG model was developed at the Yale-New Haven Health System (YNHHS) for the detection of HCM on ECG images.4 Briefly, the model was developed and validated in a demographically diverse cohort using ECG images from patients with cardiac magnetic resonance imaging-confirmed HCM who had not received mavacamten or undergone SRT. This model had an AUROC of 0.95 (95% CI 0.93–0.97), with 92% sensitivity and 88% specificity for identifying patients with HCM. It is publicly available for research use at https://www.cards-lab.org/ecgvision-hcm.
We applied the AI-ECG model to ECG images obtained from the same patients before and after SRT across three sites: (1) YNHHS (New Haven, CT), (2) Cleveland Clinic Foundation (CCF; Cleveland, OH), and (3) Atlantic Health System (AHS; Morristown, NJ). For each patient across sites, the last ECG before SRT and the most recent available ECG were used to determine the AI-ECG HCM probability before and after SRT.
RESULTS
A total of 70 patients (n=32 [46%] women, n=14 [20%] non-White) underwent SRT at YNHHS with a median age of 70 years (IQR 59-75). At the external sites, a total of 100 patients underwent SRT at CCF (n=48 [48%] women, n=10 [10%] non-White, median age 62 [IQR 52-72] years), and 145 at AHS (n=63 [43%] women, n=20 [13.8%] non-White, median age 60 [IQR 48–67] years).
Among patients at YNHHS, there was no significant change in the AI-ECG score before versus after SRT (pre-SRT: 0.55 [IQR 0.24–0.77] vs post-SRT: 0.59 [IQR 0.40–0.75], p = 0.42 by Wilcoxon signed-rank test). The AI-ECG HCM scores were higher post SRT at CCF (0.61 [IQR 0.32–0.79] vs 0.69 [IQR 0.52–0.79], p = 0.003) and AHS (0.52 [IQR 0.35– 0.69] vs 0.61 [IQR 0.49–0.70], p < 0.001) (Figure, Panel A).
In our positive control experiment, we replicated the observed effect of mavacamten on the AI-ECG HCM score in patients who had received mavacamten at YNHHS since its approval. We assessed the AI-ECG scores of these patients before and after the initiation of mavacamten, comparing the mean of all ECGs up to 3 years before therapy versus the most recent ECG after mavacamten initiation (Figure, Panel B). At YNHHS, 36 patients (n=15 [42%] women, n=6 [17%] non-White, median age 65 [IQR 59–69] years) undergoing treatment with mavacamten who had ECGs before and after starting the therapy were studied. The median AI-ECG score prior to starting mavacamten was 0.46 (IQR 0.21–0.72). At the end of a median follow-up period of 237 days (IQR 140–338), the median AI-ECG score decreased significantly to 0.28 (IQR 0.11–0.50, p = 0.002).
DISCUSSION
The contrasting effects of SRT and cardiac myosin inhibition on AI-ECG HCM score may reflect different mechanisms for modification of the HCM substrate. While SRT relieves obstructive physiology, an effect also observed with mavacamten, it does so by removing the anatomical obstruction without modifying the pathophysiology of diseased sarcomeres in other parts of the myocardium. In this respect, a significant decrease in the AI-ECG score among patients treated with mavacamten is concordant with its known effects on myocardial remodeling and energetics.5
Our study’s multicenter design also demonstrates the feasibility of an image-based AI-ECG approach for disease monitoring in HCM. This approach can be used across health systems regardless of underlying data formats and also by clinicians who may only have access to ECGs as printouts or digital images. Our primary limitation is the low number of patients undergoing treatment with mavacamten, which is a result of the relatively recent approval of the drug. Nonetheless, our findings are consistent with the effect of mavacamten observed in a previous report that used raw ECG signal data from a phase-2 clinical trial.1
In conclusion, our study demonstrates a measurable difference in the myocardial effects of SRT and cardiac myosin inhibition detectable on ECG images and suggests a potential strategy for serial monitoring of pathophysiological improvement in HCM at the point-of-care.
Sources of Funding
The study was supported in part by Bristol Myers Squibb (BMS) and the National Heart, Lung, and Blood Institute of the National Institutes of Health (under award K23HL153775 to Dr. Khera, and award 1F32HL170592-01 to Dr. Oikonomou). The SRT patient cohort from YNHHS (Yale New Haven Health System) is part of this analysis. Data from this cohort were presented at the 2023 AHA Scientific sessions. The funders had no role in the design and conduct of the protocol; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclosures
Mr. Sangha and Dr. Khera are the coinventors of U.S. Provisional Patent Application No. 63/346,610, “Articles and methods for format-independent detection of hidden cardiovascular disease from printed electrocardiographic images using deep learning”. Dr. Khera receives support from the National Heart, Lung, and Blood Institute of the National Institutes of Health (under award K23HL153775) and the Doris Duke Charitable Foundation (under award 2022060). He receives support from the Blavatnik Foundation through the Blavatnik fund for Innovation at Yale. He also receives research support, through Yale, from Bristol-Myers Squibb, Novo Nordisk, and BridgeBio. He is an Associate Editor at JAMA. In addition to 63/346,610, Dr. Khera is a coinventor of U.S. Provisional Patent Applications 63/177,117, 63/428,569, and 63/484,426. Dr. Khera and Dr. Oikonomou are co-founders of Evidence2Health, a precision health platform to improve evidence-based cardiovascular care. Dr. Oikonomou is a co-inventor of the U.S. Patent Applications 63/508,315 & 63/177,117 and has served as a consultant to Caristo Diagnostics Ltd (all outside the current work). Dr. Krumholz works under contract with the Centers for Medicare & Medicaid Services to support quality measurement programs, was a recipient of a research grant from Johnson & Johnson, through Yale University, to support clinical trial data sharing; was a recipient of a research agreement, through Yale University, from the Shenzhen Center for Health Information for work to advance intelligent disease prevention and health promotion; collaborates with the National Center for Cardiovascular Diseases in Beijing; receives payment from the Arnold & Porter Law Firm for work related to the Sanofi clopidogrel litigation, from the Martin Baughman Law Firm for work related to the Cook Celect IVC filter litigation, and from the Siegfried and Jensen Law Firm for work related to Vioxx litigation; chairs a Cardiac Scientific Advisory Board for UnitedHealth; was a member of the IBM Watson Health Life Sciences Board; is a member of the Advisory Board for Element Science, the Advisory Board for Facebook, and the Physician Advisory Board for Aetna; and is the co-founder of Hugo Health, a personal health information platform, and co-founder of Refactor Health, a healthcare AI-augmented data management company. Mr. Sangha, Dr. Krumholz, and Dr. Khera are cofounders of Ensight-AI, Inc., a Yale-backed startup focusing on cardiovascular diagnostics. Dr. Desai acknowledges The Haslam family endowed chair in cardiovascular medicine and receives philanthropic support from the Holekamp and Stinson family.
Data Availability Statement
The dataset cannot be made publicly available because they are electronic health records. Sharing this data externally without proper consent could compromise patient privacy and would violate the Institutional Review Board’s approval for the study. The code for the study is available from the authors upon request.
Ethical Approval
The Yale institutional review board reviewed the study, approved the study protocol, and waived the need for informed consent, as the study represents a secondary analysis of existing data.
(A) Box and whisker plots for artificial intelligence-enabled electrocardiogram hypertrophic cardiomyopathy scores (AI-ECG HCM score) for ECGs spanning septal reduction therapy. (B) Box and whisker plot for AI-ECG HCM scores for ECGs recorded pre- and on-mavacamten therapy, and a fourth order generalized additive model curve for AI-ECG HCM scores for on-mavacamten ECGs. Abbreviations: AHS, Atlantic Health System; CCF, Cleveland Clinic Foundation; ECG, electrocardiogram; SRT, Septal Reduction Therapy; YNHHS, Yale New Haven Health System.
ACKNOWLEDGMENTS
Footnotes
The manuscript includes results from an updated cohort of 36 patients on mavacamten. The figure has been updated to reflect the same. Funding source 1F32HL170592-01 has been added.