Visceral fat measurement on CT imaging may predict CV events

Using a fully-automated deep learning analysis of an abdominal computed tomography (CT) image to measure visceral fat area may predict future major cardiovascular (CV) outcomes, particularly heart attack and stroke, according to a study presented at RSNA 2020.

“[This study showed] that precise measures of body muscle and fat compartments achieved through [abdominal] CT outperform traditional biomarkers for predicting risk for CV outcomes,” said study author Dr Kirti Magudia from Brigham and Women’s Hospital in Boston, Massachusetts, US.

To determine body composition metrics such as subcutaneous fat, skeletal muscle, and visceral fat areas based on abdominal CT scans, a multidisciplinary team comprising radiologists, a data scientist, and a biostatistician developed a fully automated body composition analysis using deep learning, a type of artificial intelligence (AI), noted Magudia.

Using data from the Partners Healthcare database in Boston, the researchers retrospectively analysed 12,128 outpatients (mean age 52.4 years, 57.1 percent female) who had an abdominal CT examination and did not have any major CV or cancer diagnoses at the time of imaging. Participants were divided into four quartiles for each body composition parameter (Q1 as the lowest and Q4 as the highest amount of visceral fat area). [RSNA 2020, abstract SSGI04-06]

Over the 5-year follow-up period, 1,560 patients had a heart attack and 938 had a stroke.

Five years following the abdominal CT scan, those in the highest quartile (Q4) of visceral fat area were more likely to develop a heart attack (hazard ratio [HR], 1.31; p=0.04) or stroke (HR, 1.46; p=0.04) than those in the lowest quartile (Q1). 

Notably, patients in the lowest quartile with the least visceral fat area showed that they are even protected against having future stroke after the abdominal CT exam, Magudia noted.

There was no association found between body mass index (BMI) and heart attack or stroke. “Established CV risk models rely on factors like weight and BMI that are crude surrogates of body composition,” said Magudia. “It’s well established that people with the same BMI can have markedly different proportions of muscle and fat. These differences are important for a variety of health outcomes.”

“In conclusion, fully automated body composition analysis of outpatient abdominal CT exams enable prediction of major CV events better than weight, height, and BMI,” the researchers noted. “[This] analysis can [now] be applied to large scale research projects with a low failure rate of 2.5 percent,” Magudia suggested.

“This work shows the promise of AI systems to add value to clinical care by extracting new information from existing imaging data,” Magudia noted. “The deployment of AI systems would allow radiologists, cardiologists, and primary care doctors to provide better care to patients at minimal incremental cost to the health care system.”