Optimal Reference Region to Measure Longitudinal Amyloid-β Change with 18F-Florbetaben PET

Accurate measurement of changes in amyloid-b (A beta) deposition over time is important in longitudinal studies, particularly in anti-A beta therapeutic trials. To achieve this, the optimal reference region (RR) must be selected to reduce variance of A beta PET measurements, allowing early detection of treatment efficacy. The aim of this study was to determine the RR that allows earlier detection of subtle A beta changes using F-18-florbetaben PET. Methods: Forty-five patients with mild cognitive impairment (mean age +/- SD, 72.69 +/- 6.54 y; 29 men/16 women) who underwent up to 3 F-18-florbetaben scans were included. Baseline scans were visually classified as high (A beta+) or low (A beta-) amyloid. Six cortical regions were quantified using a standardized region-ofinterest atlas applied to the spatially normalized gray matter image obtained from segmentation of the baseline T1-weighted volumetric MRI. Four RRs (cerebellar gray matter [CGM], whole cerebellum [WCER], pons, and subcortical white matter [SWM]) were studied. The SUV ratio (SUVR) for each RR was calculated by dividing cortex activity by RR activity, with a composite SUVR averaged over +/- cortical regions. SUVR increase from baseline to 1 and 2 y, and percentage A beta deposition per year, were assessed across A beta+ and A beta- groups. Results: SUVs for any RR were not significantly different over time. Percentage A beta accumulation per year derived from composite SUVR was 0.10 +/- 1.72 (A beta-) and 1.36 +/- 1.98 (A beta+) (P = 0.02) for CGM and 0.13 +/- 1.47 and 1.32 +/- 1.75 (P = 0.01), respectively, for WCER. Compared with baseline, the composite SUVR increase in A beta+ scans was significantly larger than in A beta- scans at 1 y (P = 0.04 [CGM]; P = 0.03 [WCER]) and 2 y (P = 0.02 [CGM]; P = 0.01 [WCER]) using these 2 RRs. Significant SUVR changes using the pons as the RR were detected only at 2 y (P = 0.46 [1 y], P = 0.001 [2 y]). SUVR using the SWM as the RR showed no significant differences at either followup (P = 0.39 [1 y], P = 0.09 [2 y]). Conclusion: RR selection influences reliable early measurement of A beta changes over time. Compared with SWM and pons, which do not fulfil the RR requirements and have limited sensitivity to detect A beta changes, cerebellar RRs are recommended for F-18-florbetaben PET because they allow earlier detection of A beta accumulation.