This computer-enhanced image shows interacting chromatin chains – the densely packed structure of DNA. New research from UTSW shows that a machine learning program can diagnose subtypes of heart disease by analyzing how DNA molecules inside heart cells are organized into chromatin. Credit: Horng Ou, Sebastien Phan, Mark Ellisman, Clodagh O’Shea, Salk Institute, La Jolla, CA (via NIH).
DALLAS – September 19, 2022 – The human heart is a complex and intricate organ and, like a car that begins to sputter, its function deteriorates for all sorts of reasons. Cardiomyopathy – any disease of the heart muscle that prevents it from pumping blood efficiently – can be caused by, among other things, blockages, thickened muscles or enlarged heart chambers. For clinicians, differentiating and correctly diagnosing these types of diseases can be challenging.
Nikhil Munshi, MD, Ph.D., Associate Professor of Internal Medicine
Now, an interdisciplinary team of clinicians and researchers at UT Southwestern has shown that certain subtypes of cardiomyopathy can be accurately diagnosed by analyzing how DNA molecules inside heart cells are organized into chromatin – the densely packed structure of DNA. Chromatin changes impact active genes in a cell and can therefore affect heart function.
“Being able to better diagnose cardiomyopathies makes a difference not only in guiding treatment, but also in informing patients of their prognosis,” said cardiologist Nikhil Munshi, MD, Ph.D., associate professor of internal medicine at UT Southwestern and lead co-author of the new book, published in Traffic.
When a person has symptoms of heart disease, such as shortness of breath, dizziness, or swollen legs and feet, doctors usually scan their heart using tools like echocardiograms to assess muscle function. and heart valves. If they suspect a blockage, they will seek coronary angiography, in which a dye helps visualize the flow of blood through the heart. The results help determine what type of treatment — from drugs to heart stents — is best for a given patient.
About a quarter of the time, however, cardiologists cannot identify any particular underlying cause of cardiomyopathy. Additionally, autopsies of heart patients have revealed that doctors often misdiagnose subtypes of cardiomyopathy.
Dr. Munshi, in collaboration with UT Southwest surgeons, transplant cardiologists, and Gary Hon, Ph.D., associate professor at the Cecil H. and Ida Green Center for Reproductive Biology Sciences, analyzed cells from the left ventricles of 15 patients with cardiomyopathy as well as six healthy people. Since heart biopsies aren’t routinely done for cardiomyopathy, the researchers used samples taken from patients undergoing heart transplants or myectomy (surgical removal of muscle tissue) and from healthy donor hearts.
They used a machine learning approach to analyze thousands of sections of chromatin in each patient’s cells and identify what differed between patients with three subtypes of cardiomyopathy – hypertrophic cardiomyopathy (caused by thickening of the walls left ventricle), ischemic cardiomyopathy (caused by blockages in coronary arteries), and non-ischemic cardiomyopathy (caused by an enlarged left ventricle with no underlying structural changes). The machine learning program was able to recognize different chromatin signatures in each patient group.
“Chromatin is like a very unique fingerprint of a cell’s state,” Dr. Hon said. “This was a proof-of-principle study to show that we can indeed train an algorithm to differentiate these fingerprints between patient groups.”
To test the effectiveness of the program, the researchers used it on three new samples of patients who had not been included in the initial sample. The program correctly identified each patient’s cardiomyopathy type and showed that chromatin patterns changed after cardiomyopathy treatment.
The researchers said that since cardiac biopsies are not currently the standard of care for patients with cardiomyopathy, there is no immediate path to using the new data in the clinic. However, if chromatin patterns allow drastically improved diagnosis of cardiomyopathy, this may encourage the use of biopsies.
“Heart biopsies have become very safe, so if we can prove there is a very good reason to start doing more routine biopsies to guide treatment, they could become more routine,” Dr Munshi said.
Other UTSW researchers who contributed to this study include Samadrita Bhattacharyya, Jialei Duan, Ryan Vela, Minoti Bhakta, and Pradeep Mammen. Pietro Bajona also contributed.
This work was supported by the American Heart Association (17PRE33670730), the National Institutes of Health (HL136604, HL133642 and HL135217 to NVM; DP2GM128203 to GCH; UM1HG011996 to NVM and GCH; R01HL102478 and P50HD087351 to PPAM), the Burroughs Wellcome Fund (8 to NVM; 1019804 to GCH), March of Dimes Foundation (#5-FY13-203 to NVM), Department of Defense (PR172060 to NVM and GCH), Cancer Prevention and Research Institute of Texas (GCH) and the Green Center for Reproductive Biology (GCH).
About UT Southwestern Medical Center
UT Southwestern, one of the nation’s leading academic medical centers, integrates pioneering biomedical research with exceptional clinical care and education. The institution’s faculty has received six Nobel Prizes and includes 26 members of the National Academy of Sciences, 17 members of the National Academy of Medicine, and 14 researchers from the Howard Hughes Medical Institute. Full-time faculty of more than 2,900 are responsible for groundbreaking medical advances and committed to rapidly translating scientific research into new clinical treatments. UT Southwestern physicians provide care in more than 80 specialties to more than 100,000 inpatients, more than 360,000 emergency room cases, and oversee nearly 4 million outpatient visits annually.
#UTSW #researchers #DNA #analysis #diagnose #heart #disease #subtypes