AI measures Afib risk, how the BCG vaccine works, insights into tumor metabolism, and more
Research Roundup: November 19, 2021
Welcome to the November 19, 2021 installment of Research Roundup, a recurring snapshot of recent studies published by scientists at the Broad Institute and their collaborators.
AI got the beat
Artificial intelligence (AI) models can tease information out of standard clinical tests that human experts might not see. Using ECGs from more than 120,000 people from MGH, Brigham and Women's Hospital, and UK Biobank, Shaan Khurshid, Cardiovascular Disease Initiative associate member Steven Lubitz, and colleagues (including several from the Machine Learning for Health project) have developed ECG-AI, an AI-based method capable of inferring a patient's five-year risk for developing atrial fibrillation. They report that ECG-AI's predictions compare well with those of a clinical model called CHARGE-AF, and that together the two models provide greater predictive accuracy than either individually. Learn more in Circulation, an MGH/Broad press release, and a tweetorial by Khurshid.
Lessons from returning actionable genomic results
In the American Journal of Human Genetics, a team led by Carrie Zawatsky, Nidhi Shah, Kalotina Machini, Matthew Lebo, and associate member Robert Green of the Program in Medical and Population Genomics describes the results of sharing genomic findings with participants in the Mass General Brigham Biobank. The investigators analyzed genomic results from 36,417 participants, focusing on pathogenic and likely pathogenic variants in 59 genes. The work provides detailed data on consent processes, initial analysis and verification, recontact efforts, confirmation of research findings, disclosure, and clinical referral among participants, as well as the costs required to carry out such a protocol. Read more in a press release from Brigham and Women's Hospital.
Uncovering the basis of the BCG vaccine immune response
The Bacillus Calmette-Guérin (BCG) vaccine, among the most commonly used vaccines in the world, is used to prevent tuberculosis, but also confers some non-specific protection against other pathogens through a yet unknown mechanism. In Genome Biology and Cell Reports, Martin Stražar, Lingjia Kong, and core institute member Ramnik Xavier, director of the Klarman Cell Observatory and Immunology Program and co-director of the Infectious Disease and Microbiome Program, together with colleagues at the Radboud University Medical Center (the Netherlands) use metagenomics of the gut microbiome and single-cell transcriptomics of monocytes to study the immune response to BCG. They found the microbiome is strongly associated with the adaptive and innate responses to BCG. They also showed that BCG reduces systemic inflammation, and highlighted 75 genes that may contribute to non-specific protection and trained immunity.
Exploring an enzyme’s effects on tumor growth
Inhibiting the methionine synthase enzyme restrains tumor growth, but it’s unclear how. Mark Sullivan and Alicia Darnell (Koch/MIT), institute member Matthew Vander Heiden in the Cancer Program, and others found the enzyme is necessary to utilize the physiological folate vitamin 5-methyl tetrahydrofolate and maintain intracellular nucleotide synthesis, but it was not required with a different folate vitamin source. The extracellular environment can therefore alter whether methionine synthase is essential to cancer cell proliferation, suggesting that experimental models with physiological folates are critical to defining tumor types most responsive to methionine synthase inhibition and understanding the response of cancers to other drugs that target folate metabolism. Read more in Nature Metabolism.