Predicting cortical thickness with [¹⁸F]flortaucipir PET: A longitudinal MRI study

Abstract Background The association between tau pathology and neurodegeneration (atrophy) in Alzheimer’s disease (AD) is not fully understood. This study aimed to investigate whether tau pathology is associated with concurrent and longitudinal cortical thickness in individuals along the AD continuum. Method We included 92 amyloid‐positive cognitively impaired individuals (age 65±7, 48% female, MMSE 23±4) with AD dementia (n = 73) or mild cognitive impairment due to AD (MCI, n = 19) and 61 cognitively normal individuals (CN; age 66±8, 48% female, MMSE 29±1) with subjective cognitive decline (SCD, n = 55, 19 amyloid‐positive) or healthy controls (HC, n = 6, amyloid‐status unknown). All participants underwent dynamic [ 18 F]flortaucipir PET and structural 3T‐MRI at baseline. A subset underwent two‐year follow‐up structural 3T‐MRI (CN: n = 35, time‐interval 24±8 months; AD: n = 26; time‐interval 28±4 months). Parametric [ 18 F]flortaucipir binding potential (BP ND ) images were generated using receptor parametric mapping (cerebellar gray matter reference region) to quantify tau pathology. Cortical thickness (over time) was measured using Freesurfer (version 6.0.1). Regional associations between [ 18 F]flortaucipir BP ND and cortical thickness were assessed using linear mixed models in three commonly used a priori defined regions‐of‐interest, including entorhinal, middle‐and inferior temporal, and neocortical areas. Models were adjusted for age, sex and amyloid status (in case of CN individuals). Result For cognitively impaired individuals, higher [ 18 F]flortaucipir BP ND (in all three regions) was associated with lower concurrent cortical thickness in the entorhinal cortex [Table‐1&Fig‐1]. Higher middle‐ and inferior temporal and neocortical BP ND was associated with cortical thinning over time within the same region (intra‐regionally). Higher middle‐ and inferior temporal BP ND additionally showed associations with cortical thinning over time in the neocortex (inter‐regionally). For cognitively normal individuals, we found no associations between [ 18 F]flortaucipir BP ND and cortical thickness (over time) [Table‐1&Figure‐1]. Conclusion Tau pathology is associated with both concurrent and longitudinal atrophy in a region‐specific manner in cognitively impaired individuals (MCI/AD), but not in cognitively normal individuals (HC/SCD). The regional‐specificity of these associations might suggest that there are similarities between tau‐related progression patterns and patterns of neurodegeneration. Taking into account a temporal delay between the development of region‐specific tau pathology and neurodegeneration, this potentially suggests that tau pathology is a (partial) driver of neurodegeneration in AD.