Simplified quantification of Pittsburgh Compound B amyloid imaging PET studies: a comparative analysis.

PET studies have been performed using the amyloid binding radiotracer Pittsburgh Compound B (PIB). Previous quantitative analyses using arterial blood showed that the Logan graphical analysis using 90 min of emission data (ART90) provided a reliable measure of PIB retention. This work reports on simplified methods of analysis for human PIB imaging. PIB PET scans were conducted in 24 subjects (6 Alzheimer's disease [AD], 10 mild cognitive impairment [MCI], 8 controls) with arterial blood sampling. Retest scans were performed on 8 subjects (3 AD, 1 MCI, 4 controls) within 28 d. Data were analyzed over 60 and 90 min using the Logan analysis and (a) metabolite-corrected input functions based on arterial plasma (ART60, ART90), (b) carotid artery time-activity data with a population average metabolite correction (CAR60, CAR90); and (c) cerebellar reference tissue (CER60, CER90). Data also were analyzed using the simplified reference tissue method (SRTM60, SRTM90) and a single-scan method based on late-scan ratios of standardized uptake values (SUVR60, SUVR90). All methods of analysis examined effectively discerned regional differences between AD and control subjects in amyloid-laden cortical regions, although the performance of the simplified methods varied in terms of bias, test-retest variability, intersubject variability, and effect size. CAR90 best agreed with ART90 distribution volume ratio (DVR) measures across brain regions and subject groups and demonstrated satisfactory test-retest variability (+/-7.1% across regions). CER90 and CER60 showed negative biases relative to ART90 in high-DVR subjects but had the lowest test-retest variability. The single-scan SUV-based methods showed the largest effect sizes for AD and control group differences and performed well in terms of intersubject and test-retest variability. Of the simplified methods for PIB analysis examined, CAR90 provided DVR measures that were most comparable to ART90; CER90 was the most reproducible and SUVR90 produced the largest effect size. All simplified methods were effective at distinguishing AD and control differences and may be effectively used in the analysis of PIB. SUVR60 data can be obtained with as little as 20 min of PET emission data collection. The relative strengths and limitations of each method must be considered for each experimental design.