Performance of a [18F]Flortaucipir PET Visual Read Method Across the Alzheimer Disease Continuum and in Dementia With Lewy Bodies

BACKGROUND AND OBJECTIVES: Recently, the US Food and Drug Administration approved the tau-binding radiotracer [ F]flortaucipir and an accompanying visual read method to support the diagnostic process in cognitively impaired patients assessed for Alzheimer's disease (AD). Studies evaluating this visual read method are limited. Here, we evaluated the performance of the visual read method in participants along the AD continuum and dementia with Lewy Bodies (DLB) by determining its reliability, accordance with semi-quantitative analyses and associations with clinically relevant variables. METHODS: We included participants who underwent tau-PET at Amsterdam University Medical Center. A subset underwent follow-up tau-PET. Two trained nuclear medicine physicians visually assessed all scans. Inter-reader agreement was calculated using Cohen's kappa (?). To examine the concordance of visual read tau-positivity with semi-quantification, we defined standardized uptake value ratio (SUVr)-positivity using different threshold approaches. To evaluate the prognostic value of tau-PET visual read, we performed linear mixed models with longitudinal Mini-Mental State Examination (MMSE). RESULTS: We included 263 participants (mean age: 68.5, 45.6% female), including 146 cognitively unimpaired (CU), 97 amyloid-positive mild cognitive impairment or AD-dementia (AD), and 19 DLB. The visual read inter-reader agreement was excellent (?=0.95 [CI: 0.91-0.99]). None of the amyloid-negative CU (0/92 [0%]) and 1 amyloid-negative DLB (1/12 [8.3%]) were tau-positive. Among amyloid-positive participants, 13 CU (13/52 [25.0%]), 85 AD (85/97 [87.6%]) and 3 DLB (3/7 [42.9%]) were tau-positive. Two-year follow-up visual read status was identical to baseline. Tau-PET visual read corresponded strongly to SUVr status, with up to 90.4% concordance. Visual read tau-positivity was associated with a decline on the MMSE in CU (ß=-0.52 [CI: -0.74, -0.30], <0.001) and AD (ß=-0.30 [CI: -0.58, -0.02], =0.04). DISCUSSION: The excellent inter-reader agreement, strong correspondence with SUVr, and longitudinal stability indicate that the visual read method is reliable and robust, supporting clinical application. Furthermore, visual read tau-positivity was associated with prospective cognitive decline, highlighting its additional prognostic potential. Future studies in unselected cohorts are needed for a better generalizability to the clinical population. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that [ F]flortaucipir visual read accurately distinguishes patients with low tau-tracer binding from those with high tau-tracer binding, and is associated with amyloid-positivity and cognitive decline.