Advanced diffusion‐weighted MRI sensitively detects age and sex effects in 34,423 adults

Abstract

AbstractBackgroundAging and Alzheimer’s disease (AD) are associated with alterations in the brain’s white matter (WM) microstructure. AD also exhibits significant sex differences and, similar to age, sex has also been associated with WM microstructural variability. A comprehensive characterization of how aging and sex impact WM microstructure may thus improve our understanding of processes involved in AD. Here we examined age and sex effects on traditional and advanced measures of WM microstructure in middle‐ to older‐aged adults.MethodCross‐sectional diffusion‐weighted MRI (dMRI) scans were analyzed from 34,423 UK Biobank participants (aged 45‐80 years; 53% female) using single‐shell models (DTI, TDF), and advanced multi‐shell models (NODDI, MAP). Metrics were projected to a standard WM skeleton using publicly available ENIGMA protocols, based on TBSS (FSL). Mean values for each dMRI metric were extracted from the whole WM skeleton, and age was modeled using fractional polynomial regressions, co‐varying for years of education, waist/hip ratio, population structure, and the Townsend index. Sex‐stratified centile curves were created for each dMRI metric to provide normative reference charts.ResultWM microstructure across the brain was significantly associated with participants’ age and sex for all dMRI models (main effects, ps<.001). Increasing age in our cross‐sectional sample was associated with lower anisotropy (DTI‐FA, TDF‐FA), neurite density (NODDI‐ICVF, MAP‐RTOP, MAPMRI‐RTAP), and restriction (MAP‐RTPP) as well as higher diffusivity (DTI‐AxD, DTI‐MD, DTI‐RD), free water (NODDI‐ISOVF), and white matter dispersion (NODDI‐OD) for both males and females (Figures 1 and 2). Some metrics showed a gentle decline until around age 60, followed by a precipitous decline thereafter (e.g., TDF‐FA, NODDI‐ICVF, MAP‐RTAP, Figures 1 and 2). Significant sex differences in aging were most sensitively detected by the advanced dMRI multi‐shell models NODDI and MAPMRI (age‐by‐sex interaction; males showing a steeper trajectory for NODDI‐OD, p<.001, females showing a steeper trajectory for MAPMRI‐RTPP, p<.01, with age).ConclusionParticipant age and sex are significantly associated with the brain’s WM microstructure; advanced dMRI models displayed the greatest sensitivity, strenghtening previous findings in a smaller UKB sample (15k). Along with our calculated centile curves, we offer an important normative reference to guide future studies of Alzheimer’s disease.