Authors:
Madhurima Datta¹, Eti Ben Simon², Saurabh Garg¹, Nasrin Akbari¹, Ahmed Gouda¹, Matt Walker², Alex Exuzides¹
Affiliations:
¹ Prenuvo Inc, Redwood City, United States
² Sleep Innovation Labs, the Centre for BrainHealth, UT Dallas, Richardson, United States
Introduction
Sleep duration and sleep efficiency have both been linked to neurodegeneration and metabolic dysfunction, but their independent associations with brain and metabolic imaging biomarkers remain unclear. We compared their associations with structural brain MRI measures and visceral adipose tissue (VAT) in a large preventive health cohort.
Methods
This cross-sectional study included 14,031 adults (mean age 50.1 ± 12.6 years; 58.4% male; mean body mass index 28.7 ± 5.5) who underwent whole-body MRI at Prenuvo. Sleep duration and sleep efficiency during the previous month were assessed using Pittsburgh Sleep Quality Index (PSQI)-based questionnaire responses. Regional brain volumes were quantified from three-dimensional T1-weighted MRI using the FastSurfer deep-learning pipeline, and VAT was segmented from abdominal T1-weighted MRI. Outcomes included total gray matter, frontal, temporal, parietal, and occipital gray matter, hippocampal and amygdala volumes, an Alzheimer’s disease signature volume, and VAT. Standardized linear regression models adjusted for age, sex, body mass index, scanner location, and intracranial volume (brain outcomes). Nested models evaluated the independent effects of both sleep measures. Standardized β coefficients, partial R² values, and false discovery rate (FDR)-adjusted p-values were reported.
Results
Higher sleep efficiency was independently associated with greater parietal gray matter volume (β = 0.025, FDR-adjusted p < 0.001), whereas longer sleep duration was associated with smaller hippocampal volume (β = -0.032, FDR-adjusted p < 0.001). Both measures were independently associated with VAT in opposite directions: longer sleep duration with greater VAT (β = 0.048, FDR-adjusted p < 0.001) and higher sleep efficiency with lower VAT (β = -0.038, FDR-adjusted p < 0.001). Additional cortical and Alzheimer’s disease-related regions showed nominal associations with sleep efficiency but not after multiple-comparison correction. Partial R² analyses showed sleep efficiency explained more unique variance in cortical outcomes, whereas sleep duration explained more variance in hippocampal volume and VAT.
Conclusions
Sleep duration and sleep efficiency were independently associated with distinct brain and metabolic imaging biomarkers. Sleep efficiency showed the strongest association with parietal cortical structure, whereas longer sleep duration was most strongly associated with smaller hippocampal volume. Both measures independently predicted VAT in opposite directions.
