ABSTRACT: The development of tau-PET allows paired helical filament tau pathology to be visualized in vivo. Increased knowledge about conditions affecting the rate of tau accumulation could guide the development of therapies halting the progression of Alzheimer's disease. However, the factors modifying the rate of tau accumulation over time in Alzheimer's disease are still largely unknown. Large-scale longitudinal cohort studies, adjusting for baseline tau load, are needed to establish such risk factors. In the present longitudinal study, 419 participants from four cohorts in the USA (Avid 05e, n?=?157; Expedition-3, n?=?82; ADNI, n?=?123) and Sweden (BioFINDER, n?=?57) were scanned repeatedly with tau-PET. The study participants were cognitively unimpaired (n?=?153), or patients with mild cognitive impairment (n?=?139) or Alzheimer's disease dementia (n?=?127). Participants underwent two to four tau-PET (18F-flortaucipir) scans with a mean (± standard deviation) of 537 (±163) days between the first and last scan. The change in tau-PET signal was estimated in temporal meta- and neocortical regions of interest. Subject specific tau-PET slopes were predicted simultaneously by age, sex, amyloid status (determined by amyloid-? PET), APOE ?4 genotype, study cohort, diagnosis and baseline tau load. We found that accelerated increase in tau-PET signal was observed in amyloid-?-positive mild cognitive impairment (3.0?±?5.3%) and Alzheimer's disease dementia (2.9?±?5.7%), respectively, when compared to either amyloid-?-negative cognitively unimpaired (0.4?±?2.7%), amyloid-?-negative mild cognitive impairment (-0.4?±?2.3%) or amyloid-?-positive cognitively unimpaired (1.2?±?2.8%). Tau-PET uptake was accelerated in females (temporal region of interest: t?=?2.86, P?=?0.005; neocortical region of interest: t?=?2.90, P?=?0.004), younger individuals (temporal region of interest: t = -2.49, P?=?0.013), and individuals with higher baseline tau-PET signal (temporal region of interest: t?=?3.83, P?
