Body composition measures from magnetic resonance imaging scans may be associated with adverse health outcomes

1. In this prospective cohort study, an artificial intelligence tool was used to segment magnetic resonance imaging scans into body composition compartments including subcutaneous adipose tissue, visceral adipose tissue, and skeletal muscle.

2. Visceral adiposity and skeletal muscle fat fraction were found to be associated with risk of diabetes mellitus and major adverse cardiovascular events.

Evidence Rating Level: 2 (Good)

Study Rundown: Current definitions of obesity are based on body mass index (BMI), but this measure overlooks the distribution of body fat, which is an important factor in determining cardiometabolic risk. Although high-resolution, cross-sectional imaging scans such as magnetic resonance imaging (MRI) may have utility for obesity assessment, these modalities are not routinely used to derive body composition (BC) measures in clinical settings. This study aimed to determine the association between artificial intelligence (AI) estimates of relative BC measures and future cardiometabolic disease risk. In females, the largest BC compartment was subcutaneous adipose tissue (SAT); females had larger SAT volumes compared with men at all age ranges. An increase in SAT was seen as individuals of both sexes aged. In males, the largest compartment was skeletal muscle (SM), which peaked in midlife before declining thereafter. After final statistical adjustment, it was found that higher visceral adipose tissue (VAT) proportion and skeletal muscle fat fraction (SMFF) were associated with higher incidence of diabetes mellitus (DM) and major adverse cardiovascular events (MACE) in both sexes. In males, lower SM proportion was also associated with higher risk of DM. The generalizability of this study is limited by its study population of non-Hispanic White adults older than 45 years and by potential residual confounding from a lack of information on lifestyle. Nevertheless, this study suggests that AI-derived, MRI-based BC measures are accurate and may be predictive of cardiometabolic risks.

Click to read this study in AIM

Relevant Reading: Deep learning-based body composition analysis from whole-body magnetic resonance imaging to predict all-cause mortality in a large western population

In-Depth [prospective cohort]: This study aimed to derive relative body compartment proportions from an AI tool for MRI scans and evaluate associations between these volumes and risk of cardiometabolic outcomes. Participants were derived from the UK Biobank. A validated deep-learning model was used to measure SAT, VAT, SM, and SMFF from whole-body MRI. The model was trained on 150 random participants from the German National Cohort. The study included 33,432 people, 52.8% (n = 17,657) of whom were female. The mean age was 65.0 years and the mean BMI was 25.8 kg/m² (standard deviation [SD], 4.2). In females of all age groups, SAT was the largest compartment, making up 58.1% of total BC at age 40 to 49 years and increasing to 60.1% above 70 years. In males, SAT made up 40.0% of total BC at age 40 to 49 years and 41.5% above 70 years. Skeletal muscle was the largest compartment in males of all ages and peaked for both sexes at age 40 to 49 years (45.3% in males vs. 35.5% in females) before decreasing thereafter. Higher VAT proportion and SMFF were also seen with increased age in both sexes. Throughout a median follow-up of 4.2 years (interquartile range, 3.3 to 5.6 years), 187 (1.1%) of 17,657 females and 344 (2.2%) of 15,775 males were diagnosed with incident DM, and 177 (1.0%) females and 365 (2.3%) males reported MACE. A negative association between relative SAT and diabetes risk in females was found, with females in the bottom 20th percentile of relative SAT (<54.4%) having an adjusted hazard ratio (aHR) of 1.46 (95% CI, 1.23 to 1.72) while females in the top 20th percentile (>64.6%) had an aHR of 0.71 (95% CI, 0.59 to 0.84). High VAT proportions and high SMFF were associated with higher risk for incident DM in both sexes. The top fifth percentile for high VAT proportions and high SMFF had aHRs for DM of 2.16 (95% CI, 1.59 to 2.94) and 1.27 (95% CI, 0.89 to 1.80) in females and 1.84 (95% CI, 1.48 to 2.27) and 1.84 (95% CI, 1.43 to 2.37) in males, respectively. For MACE, aHRs were 1.37 (95% CI, 1.00 to 1.88) and 1.72 (95% CI, 1.23 to 2.41) in females and 1.22 (95% CI, 0.99 to 1.50) and 1.25 (95% CI, 0.98 to 1.60) in males, respectively. In males, low SM proportions were also associated with increased risk for DM (aHR for bottom fifth percentile, 1.96 [95% CI, 1.45 to 2.65]) and MACE (aHR for bottom fifth percentile, 1.55 [95% CI, 1.15 to 2.09]). Overall, this study suggests that AI-derived measurements from MRI scans are independently associated with cardiometabolic risk after adjustment for BMI, waist circumference, and other traditional risk factors.

Image: PD

©2025 2 Minute Medicine, Inc. All rights reserved. No works may be reproduced without expressed written consent from 2 Minute Medicine, Inc. Inquire about licensing here. No article should be construed as medical advice and is not intended as such by the authors or by 2 Minute Medicine, Inc.