RT Journal Article SR Electronic T1 A novel approach for estimation of amyloid burden from florbetapir images acquired 0-20 minutes after injection JF Journal of Nuclear Medicine JO J Nucl Med FD Society of Nuclear Medicine SP 315 OP 315 VO 55 IS supplement 1 A1 Abhinay Joshi A1 Michael Devous A1 Michael Pontecorvo A1 Michael Navitsky A1 Ian Kennedy A1 Mark Mintun A1 Daniel Skovronsky YR 2014 UL http://jnm.snmjournals.org/content/55/supplement_1/315.abstract AB 315 Objectives Florbetapir binding is typically quantified by calculating a SUVr using images acquired after an uptake period of >40 min. We investigated an empirical method to evaluate amyloid binding of florbetapir using an acquisition only 0-20 min after injection Methods Using a learning set of florbetapir PET scans from early phase development (Wong 2010), an empirical approach was developed that used the dynamic 0-20 min acquisition to calculate a florbetapir amyloid binding estimate (FABE). We then evaluated the performance of this method with data from the ADNI2 database for 70 subjects (6 AD, 15 MCI, 31 subjective memory complaints and 18 cognitively normal). Subjects had a dynamic acquisition from 0-20min, which was followed by a 10 min acquisition at 50 min after injection of florbetapir. FABE value was estimated using equation 1. SUVr value from 50 min data was calculated as ratio of cortex to cerebellum.Performance of FABE in classifying cases relative to SUVr was evaluated using statistical analyses. Results FABE ranged from -3733 to 11697. Correlation between SUVr and FABE was r=0.79 (p<0.001). A pre-specified threshold (SUVr>1.10) was used to define florbetapir positivity. The optimal threshold for FABE was determined as FABE>210. With this threshold, 96% (67/70) of cases were classified into the same category (positive or negative) as the standard SUVR method. Kappa between the two measures was 0.91 (95% CI: 0.80 to 1.00). Conclusions An empirical method of processing florbetapir PET data from 0-20 min after injection correlates with conventional SUVr values and agrees with SUVr threshold-determined positive and negative classifications. However, to determine an optimal approach to extracting binding information from the early tracer behavior, it is likely that additional methods may need to be examined. Research Support Avid Radiopharmaceuticals Inc.