Consideration of Optimal Time Window for Pittsburgh Compound B PET Summed Uptake Measurements

Pittsburgh compound B (PiB) is a 11C-labeled thioflavin-T derivative that has become widely used (1–3) in imaging amyloid-β–protein deposits with PET. Amyloid imaging has proven to be useful in the study of Alzheimer disease (AD) (4–11). In the proof-of-concept human studies (4), PiB retention was assessed using the semiquantitative standardized uptake value ratio (SUVR; summed tissue uptake over 40–60 min), and these results were subsequently verified by fully quantitative studies performed in controls, patients with mild cognitive impairment (MCI), and patients with AD (5,6). These early studies consistently showed that PiB uptake in AD patients was nearly twice that in controls in specific cortical areas (P < 0.002) but similar in amyloid-spared areas (subcortical white matter [SWM], cerebellum). PiB retention in MCI patients appeared either AD-like or control-like. Quantitative studies require dynamic sampling of the radiotracer kinetics in brain and blood at many times throughout the study (e.g., 30–40 points over ≥90 min), analysis of arterial plasma to determine the metabolite-corrected arterial input function, and compartmental modeling analyses to obtain regional binding measurements (e.g., distribution volume ratio [DVR] = total distribution volume [VT]/nondisplaceable distribution volume [VND]). The quantitative arterial input–based DVR measurements generally serve as gold-standard outcomes for subsequent evaluations directed toward the identification of valid simpler methods that do not require blood sampling and allow for shorter data-collection times. Lopresti et al. (6) analyzed simplified PiB PET data over 60- and 90-min intervals using the linear Logan graphical analysis with different input functions that were either arterial, image-derived carotid, or image-derived cerebellar data. Data also were analyzed with a simplified reference tissue method and the summed SUVR. Although results obtained using data over the full 90-min scan interval were less variable than were data over the shorter 60-min interval, the 60-min data provided useful results. Of all compared methods, the SUVR approach was the most robust in terms of test–retest variability and maximum effect size between AD patients and controls, although the SUVR method required the shortest scan time (e.g., 40–60 min or 40–90 min). The SUVR is easy to compute, requires no arterial input function, and produces data with a large dynamic range, thereby making it a potential candidate for routine clinical studies (6). SUVR can also serve as an approximation of DVR, although bias (overestimation relative to the true DVR) can arise as a result of plasma clearance (12). The present work examined the PiB SUVR to determine an optimal short time window for the SUVR data collection across the control, MCI, and AD groups. A systematic evaluation was performed for several 20- and 30-min SUVR time windows, as these are the most feasible for routine clinical settings. This evaluation considered several factors, including correlation of the SUVR to Logan DVR (arterial and cerebellar input), the effective contrast of the SUVR for AD versus control groups, and the temporal dynamics of the theoretic SUVR curve.