Integrated Multimodal Brain Signatures Predict Sex-Specific Obesity Status
Ravi R. Bhatt*2, Newton Peng*1, Soumya Ravichandran1, Priten Vora1, Jean Stains1, Bruce Naliboff1, Emeran A. Mayer1, Arpana Gupta1 *Authors share co-first authorship
1G. Oppenheimer Center for Neurobiology of Stress and Resilience, Vatche and Tamar Manoukian Division of Digestive Disease, David Geffen School of Medicine at UCLA; 2Imaging Genetics Center, Mark and Mary Stevens Neuroimaging and Informatics Institute, Keck School of Medicine at USC, University of Southern California
Background: Neuroimaging studies in obesity have individually utilized structural MRI, functional resting state, and diffusion tensor imaging to uncover mechanisms causing altered ingestive behaviors associated with diminished dietary restraint and increased cravings. However, few, if any, studies have integrated data from multimodal brain imaging to predict sex-specific brain signatures present in individuals with obesity. We aimed to investigate if a multimodal MRI – clinical signature could predict people with obesity, dependent on sex-related differences. Methods: 183 participants (Female=118; Male=65; Obese=78; Non-obese=105) underwent multimodal MRI scans (structural, functional resting state, and diffusion tensor imaging). A Data Integration Analysis using Latent Components (DIABLO) was conducted using training and test sets to determine whether clinical features, resting-state functional connectivity, anatomical connectivity and brain morphometry could accurately differentiate participants stratified by obesity and sex. Significant brain regions across all modalities and clinical variables were correlated. Result: The derived models were tested on holdout testing data differentiating obese against nonobese participants, obese males against obese females obtaining accuracies of 74% and 74%. Kappa of the respective models were 46%, and 40% while BER of the respective models were 26% and 32%. Pertinent features were then extracted with the absolute loadings on each component of the models. Discussion: The results indicate that differences in morphometry and anatomical connectivity within the default mode network, and resting-state functional connectivity between the default mode network and orbital gyrus are able to distinguish people with obesity. Greater psychological resilience was also correlated with lowered structural integrity in the default mode network as the inability to handle large cognitive loads could lead to lowered cognitive restraint and anxiety causing overeating behaviors seen in obese individuals. Alterations in the extended reward network, in regions like the orbital gyrus, could lead to dopaminergic dysregulation causing unregulated food consumption and obesity. Females with obesity had greater mean curvature in the postcentral gyrus and lateral temporal cortex and greater RSFC from the cerebellum to the SMN which were associated with greater likelihood of early life trauma and depression. The distinct sex-specific brain signatures identified through multimodal brain imaging has important clinical implications for the prevention and treatment of obesity with regards to employing more tailored diet interventions.
Breakout Room: Peng, Newton
View Poster: https://uclacns.org/symposium2021/17-Peng-Newton.pdf