Partial volume effect correction impairs the diagnostic utility of [F]-THK-5351 PET in nonfluent-agrammatic variant primary progressive aphasia.

Objectives: Partial volume effects in positron emission tomography occur frequently in neurodegenerative diseases due to increasing cortical atrophy during the disease course, and fronto-temporal dementia is often characterized by severe atrophy. The aim of this study was to challenge partial volume effect correction (PVEC) in patients with nonfluent-agrammatic variant primary progressive aphasia (nfv-PPA) imaged with [F]-THK-5351 PET a marker of reactive neuroinflammatory astrogliosis as well as tau-binding.

Methods: Patients with nfv-PPA (n = 20) were imaged with [F]-THK-5351 PET accompanied by structural magnetic resonance tomography imaging (MRI). Region specific cortical grey matter volumes and standard uptake value ratios (SUVr) of the Hammers atlas were compared with eight healthy control (HC) (n = 8) data before and after performing region-based voxel-wise PVEC. We evaluated regional coefficients of variance (CoV) and the number of regions with significant [F]-THK-5351 PET signal differences between nfv-PPA and controls before and after PVEC. Additionally, a blinded visual read was performed by three nuclear medicine physicians (consensus) before and after PVEC.

Results: Prior to PVEC, [F]-THK-5351 tracer uptake was significantly higher in the bilateral frontal cortex of patients with nfv-PPA when compared to HC (left > right), despite significant grey matter atrophy in the same brain regions in patients with nfv-PPA. SUVr differences between nfv-PPA and HC were further increased by PVEC in frontal brain regions, but group level variance increased in parallel and reduced the number of significant differences between SUVr of nfv-PPA and HC (uncorrected: 10 significant regions, CoV[nfv-PPA]: 20.8 % ± 4.7 %, CoV[HC]: 7.9 % ± 2.4 %/PVEC: 3 significant regions, CoV[nfv-PPA]: 28.4 % ± 8.9 %, CoV[HC]: 9.8 % ± 2.5 %). Sensitivity/specificity of the visual read for detection of nfv-PPA was 0.85/1.00 without PVEC and 0.85/0.75 with PVEC.

Conclusions: [F]-THK-5351 PET facilitates detection of pathological alterations in patients with nfvPPA with severe atrophy. PVEC increases quantitative SUVr differences between patients with nfv-PPA and HC but introduces a parallel increase of variance at the group level. Visual assessment of [F]-THK-5351 images in patients with nfv-PPA is impaired by PVEC due to loss of specificity and does not support the use of PVEC even in patients with severe atrophy.