Radiology
Beyond BMI with AI: muscle fat flags cardiometabolic risk
Clinical takeaway: Consider muscle composition alongside BMI when assessing cardiometabolic risk, particularly in patients who appear metabolically healthy by conventional measures.
Cardiometabolic risk often takes shape in body composition long before BMI catches it, yet most risk scores still rely on height, weight, and waist circumference. Two recent studies used deep learning to extract granular body composition measures from MRI scans across the UK Biobank and German National Cohort, examining how those measures relate to cardiometabolic outcomes.
Both studies converge on the same clinical message: muscle quality matters as much as muscle quantity, and fat distribution matters more than total adiposity.
In the larger of the two analyses of more than 66,000 adults, high visceral fat was associated with a 2.26-fold higher risk of incident diabetes, high intramuscular fat with a 1.54-fold higher risk of major cardiovascular events, and low skeletal muscle with a 1.44-fold higher all-cause mortality, all independent of traditional risk factors over follow-up that was a median of 4.2 years.
The second study, in more than 11,000 individuals without any pre-existing condition, focused on the paraspinal muscles. The paraspinal muscles were chosen because they appear consistently on routine spinal, chest, and abdominal scans, making them the most practical target for opportunistic AI analysis without requiring dedicated whole-body imaging.
The analysis identified previously undiagnosed hypertension in 16%, dysglycemia in 9%, and atherogenic dyslipidemia in 46%. Higher intermuscular fat was associated with substantially higher odds of all three conditions in both sexes. Higher lean muscle mass was protective, but only in men, a sex difference the authors attribute partly to menopausal changes in muscle composition that emerge after age 40 to 50.
“We focused on a healthy population with no known prior disease, and yet we found quite substantial cardiometabolic risk factors in these participants,” said lead researcher Sebastian Ziegelmayer, MD, associate professor and attending radiologist at Technical University of Munich. “We found that the higher the intermuscular fat and the lower the muscle mass, the greater the cardiometabolic risk factors.”
Together, the studies sketch a coherent picture. Muscle is metabolically active tissue, and fat infiltrating muscle behaves differently from subcutaneous fat. Patients who look healthy by BMI may carry meaningful intermuscular fat burden and reduced muscle quality, both of which appear to track with insulin resistance, inflammation, and cardiovascular risk.
The authors of the first study also released an open-source z-score calculator that adjusts body composition measures for age, sex, and height, intended to support opportunistic use of routine CT and MRI scans already being performed. Both teams emphasize that prospective validation is needed before muscle composition metrics can be incorporated into clinical risk scores.
“Adjusting for confounding factors is critical for improving screening accuracy and tailoring treatment decisions,” said Jakob Weiss, MD, PhD, senior author of the larger study and radiologist at University Medical Center Freiburg in Germany. “This tool has the potential to identify whether an individual’s body composition puts them at greater risk for metabolic disease compared to their age-matched peers.”
“This tool can allow clinicians to use routine imaging opportunistically,” he continued. “A dedicated whole-body MRI is not necessarily required. If a routine CT or MRI body scan already exists, the information can be extracted for benchmarking against the reference values.”
The findings have practical implications for the GLP-1 era. As patients lose substantial weight on these medications, distinguishing desirable fat loss from undesirable muscle loss becomes clinically important, but BMI cannot enable that distinction.
Sources: Jung M. Radiology. 2026 May. Body composition in the general population: whole-body MRI–derived reference curves from over 66,000 individuals. Ziegelmayer S. Radiology. 2026 May. Associations of MRI-derived paraspinal IMAT and LMM with cardiometabolic risk factors: results from a German cohort