The relationship between cortical microstructural changes and in vivo amyloid‐β and tau in aging and preclinical Alzheimer's disease

Background A novel technique for the analysis of diffusion‐weighted imaging (DWI) has allowed quantifying brain microstructural properties by means of cortical mean diffusivity (cMD) in the grey matter. Using this approach a biphasic trajectory of the cMD signal has been revealed, characterized by an early decline followed by an increase across preclinical stages of Alzheimer's disease (AD), primarily in temporo‐parietal brain regions (Montal et al .,Alz&Dem,2018). The initial cMD reduction has been hypothesized to be due to glial activation (Vilaplana et al .,Neurology,2020), while increased cMD in later stages is thought to reflect breakdown of cortical tissue microstructure and neurodegeneration. However, the pathological underpinnings of the cMD signal, specifically its relationship to in vivo amyloid‐β and tau, remain unknown. Methods 84 cognitively‐normal participants from the Harvard Aging Brain Study underwent cross‐sectional DWI, T1‐weighted‐MRI, 11 C‐PiB‐PET, and 18 F‐Flortaucipir (FTP)‐PET imaging (Table 1). 11 C‐PiB‐PET was quantified using non‐PVC‐corrected Logan DVR, and individual amyloid‐β burden was extracted from a cortical composite including frontal‐lateral‐retrosplenial (FLR) regions. 18 F‐FTP‐PET was quantified using PVC‐corrected SUVR and individual tau burden was extracted from the entorhinal cortex. All PET data were normalized to cerebellar grey matter. First, a whole‐brain surface‐based approach was used to investigate the linear and quadratic (correcting for the linear term) vertex‐wise relationships of cMD to amyloid‐β and tau. Second, cMD signal was extracted from significant regions‐of‐interest to visualize its relationships with amyloid‐β and tau. Results Cortical MD was not significantly associated to amyloid‐β using either the linear or quadratic term. The relationship between cMD and tau was best described by a quadratic model across the whole cohort, as seen vertex‐wise (Fig.1) and regionally (Fig.2). Specifically, cMD signal decreased and then increased as both amyloid‐β and tau accumulated. Conclusions The novel imaging technique of cMD is a promising approach to measure cortical microstructural changes. Our findings of a non‐linear trajectory of cortical mean diffusivity, particularly in the temporal regions, may indicate an initial phase of reduced cMD due to glial activation followed by increased cMD signal due to breakdown of cortical tissue microstructure as amyloid‐β and tau accumulate. Ongoing longitudinal studies will help elucidate the temporal dynamics of cortical microstructure and protein accumulation.