Deep Learning Derived Visceral Abdominal Fat Predicts Brain Atrophy at Midlife in 10,001 Individuals

Abstract

AbstractBackgroundVisceral abdominal fat can predict health outcomes with increasing interest in applications on brain health.MethodA total of 10,001 healthy participants from four sites were scanned on 1.5T MR machines with a short whole‐body MR imaging protocol. Core whole body sequences included coronal T1, STIR from vertex to feet, whole‐body axial DWI from vertex to proximal‐thighs and axial T2 TSE without fat suppression from skull base to pelvis. Brain sequences were T1 MPRAGE and 2D FLAIR. Deep learning with FastSurfer trained on 134 participants aged 27‐66 and segmented 96 brain regions. A separate deep learning model was trained to segment abdominal visceral fat from the whole‐body coronal T1. Partial correlation analysis of visceral abdominal fat volume and brain volumes were evaluated, controlling for age, sex, and total intracranial volume. Benjamini‐Hochberg False Discovery Rate of 5% controlled for multiple comparisons. Logistic regression models determined risk of brain total gray and white matter atrophy based upon the highest quartile of visceral fat and lowest quartile of these brain volumes.ResultThis cohort had an average age of 52.97 ± 13.04 years with a range of 11‐97 years with 52.8% men and 47.2% women. Mean visceral abdominal fat was 3417.81 ± 514.88 ml for overweight persons (BMI ³ 25) and 5355.24 ± 505.04 for obese persons (BMI ³ 30). Even accounting for co‐variates and multiple comparisons, deep learning segmented visceral abdominal fat predicted atrophy in multiple brain regions including: total gray matter volume (Partial R = ‐.09, p = 1.17e‐19), total white matter volume (Partial R = ‐.09, p = 2.25e‐21), hippocampus (Partial R = ‐.02, p = .005), frontal cortex (Partial R = ‐.09, p = 1.79e‐19), temporal lobes (Partial R = ‐.12, p = 2.6e‐30), parietal lobes (Partial R = ‐.04, p = .0004), occipital lobe (Partial R = ‐.03, p = .002), orbital frontal cortex (Partial R = ‐.09, p = 3.76e‐21). Visceral fat predicted increased risk for lower total gray matter (age 20‐39: OR = 5.9; age 40‐59, OR = 5.4; 60‐80, OR = 5.1) and white matter atrophy: (age 20‐39: OR = 3.78; age 40‐59, OR = 4.4; 60‐80, OR = 5.1).ConclusionDeep learning determined increased visceral abdominal fat volume predicts brain volume loss and may represent a novel modifiable factor in determining brain health.