Circulation September 6, 2022 Issue

This week, please join author Keith Channon as he discusses the article "Risk of Myocarditis After Sequential Doses of COVID-19 Vaccine and SARS-CoV-2 Infection by Age and Sex."

Dr. Carolyn Lam:

Welcome to Circulation on the Run, your weekly podcast summary and backstage pass to the journal and its editors. We're your co-hosts, I'm Dr. Carolyn Lam, associate editor from the National Heart Center and Duke-National University of Singapore.

Dr. Greg Hundley:

And I'm Dr. Greg Hundley, associate editor, director of the Pauley Heart Center at VCU Health in Richmond, Virginia.

Dr. Carolyn Lam:

Oh, Greg, today's feature paper, something that's really been discussed a lot in the press and in lay public as well, the risk of myocarditis following sequential doses of the COVID-19 vaccine and SARS-CoV-2 infection by age and sex. Everyone's going to want to tune into that one. But before we get there, shall we go through some of the key papers in today's issue?

Dr. Greg Hundley:

You bet, Carolyn. How about if I go first?

Dr. Carolyn Lam:

Please.

Dr. Greg Hundley:

So Carolyn, this first manuscript involves the world of machine learning and ECG interpretation. And as you know, novel targeted treatments increase the need for prompt hypertrophic cardiomyopathy detection; however, it's low prevalence, 0.5%, and resemblance to common diseases really present challenges. So Carolyn, these authors, led by Dr. Rahul Deo from Brigham and Women's Hospital, sought to develop machine learning models to detect hypertrophic cardiomyopathy and differentiate it from other cardiac conditions using EKGs and echocardiograms with a robust generalizability across multiple cohorts.

So Carolyn, what did they do? They used single-institution hypertrophic cardiomyopathy EKG models that were then trained and validated on data from three academic medical centers in the United States and Japan using a federated learning approach, which enables training on distributed data without data sharing. Models were validated on held out test sets for each institution and from a fourth academic medical center and were further evaluated for discrimination of hypertrophic cardiomyopathy from aortic stenosis, long-standing hypertension, and cardiac amyloidosis. And then finally, automated detection was compared to manual interpretation by three cardiologists on a data set with a realistic hypertrophic cardiomyopathy prevalence.

Dr. Carolyn Lam:

Wow, incredible. So what were the results?

Dr. Greg Hundley:

Right, Carolyn. So the authors identified 74,476 EKGs for 56,129 patients and 8,392 echocardiograms for 6,825 patients across the four academic medical centers. Now, while ECG models trained on data from each institution displayed excellent discrimination of hypertrophic cardiomyopathy on internal test data, the generalizability was limited, most notably for a model trained in Japan and then subsequently tested in the United States. Now, however, when trained in a federated manner, discrimination of hypertrophic cardiomyopathy was excellent across all institutions, including for phenotypic subgroups. The models further discriminated hypertrophic cardiomyopathy from hypertension, aortic stenosis, and cardiac amyloid. Analysis of ECG and echocardiography paired data from 11,823 patients from an external institution indicated a higher sensitivity of automated HCM detection at a given positive predictive value compared with cardiologists.

So Carolyn, in conclusion, federated learning improved the generalizability of models that use EKGs and echocardiograms to detect and differentiate hypertrophic cardiomyopathy from other causes of left ventricular hypertrophy compared to training within a single institution. It will be really interesting to see the future applicability of these methods.

Dr. Carolyn Lam:

Oh, I'm such a fan of this work. Awesome. Thank you, Greg. My paper, it's a preclinical paper that uncovers a novel mechanism through which GATA4 mutations can lead to heart disease.

Dr. Greg Hundley:

All right, Carolyn, no quiz this time, I'm just coming right out. I'm reversing the question on the teacher. Tell me, what is GATA4?

Dr. Carolyn Lam:

I'm glad you asked, Greg. GATA4 is a zinc finger-containing DNA binding transcription factor essential for normal cardiac development and homeostasis in mice and humans, and mutations in this gene have been reported in human heart defects. Now, in today's paper, authors led by Dr. Srivastava from Gladstone Institutes in San Francisco, California, showed that GATA4 regulated cell-type-specific splicing through direct interaction with RNA and the spliceosome in human-induced pluripotent stem cell-derived cardiac progenitors.

An unbiased search for GATA4 interacting proteins in these human iPS cells revealed interaction with many members of the sp...