Impact of partial volume effect correction on cerebral β-amyloid imaging in APP-Swe mice using [(18)F]-florbetaben PET

We previously investigated the progression of β-amyloid deposition in brain of mice over-expressing amyloid-precursor protein (APP-Swe), a model of Alzheimer's disease (AD), in a longitudinal PET study with the novel β-amyloid tracer [(18)F]-florbetaben. There were certain discrepancies between PET and autoradiographic findings, which seemed to arise from partial volume effects (PVE). Since this phenomenon can lead to bias, most especially in the quantitation of brain microPET studies of mice, we aimed in the present study to investigate the magnitude of PVE on [(18)F]-florbetaben quantitation in murine brain, and to establish and validate a useful correction method (PVEC). Phantom studies with solutions of known radioactivity concentration were performed to measure the full-width-at-half-maximum (FWHM) resolution of the Siemens Inveon DPET and to validate a volume-of-interest (VOI)-based PVEC algorithm. Several VOI-brain-masks were applied to perform in vivo PVEC on [(18)F]-florbetaben data from C57BL/6(N=6) mice, while uncorrected and PVE-corrected data were cross-validated with gamma counting and autoradiography. Next, PVEC was performed on longitudinal PET data set consisting of 43 PET scans in APP-Swe (13-20months) and age-matched wild-type (WT) mice using the previously defined masks. VOI-based cortex-to-cerebellum ratios (SUVR) were compared for uncorrected and PVE-corrected results. Brains from a subset of transgenic mice were ultimately examined by autoradiography ex vivo and histochemistry in vitro as gold standard assessments, and compared to VOI-based PET results. The phantom study indicated a FWHM of 1.72mm. Applying a VOI-brain-mask including extracerebral regions gave robust PVEC, with increased precision of the SUVR results. Cortical SUVR increased with age in APP-Swe mice compared to baseline measurements (16months: +5.5%, p<0.005; 20months: +15.5%, p<0.05) with uncorrected data, and to a substantially greater extent with PVEC (16months: +12.2% p<0.005; 20months: +36.4% p<0.05). WT animals showed no binding changes, irrespective of PVEC. Relative to autoradiographic results, the error [%] for uncorrected cortical SUVR was 18.9% for native PET data, and declined to 4.8% upon PVEC, in high correlation with histochemistry results. We calculate that PVEC increases by 10% statistical power for detecting altered [(18)F]-florbetaben uptake in aging APP-Swe mice in planned studies of disease modifying treatments on amyloidogenesis.