Abstract
555
Objectives: [18F]florbetaben (FBB) is approved in the EU, the USA, Japan and South Korea for clinical use to measure beta-amyloid plaque density in adult patients with cognitive impairment, and is currently employed in the IDEAS trial and other research studies. Depending on the marketing authorizations, different PET acquisition time windows are possible. In a phase 2A study, the accuracy of an expert visual read for earlier acquired PET images (45-60min p.i.) did not differ compared to later acquired PET images (90-110min p.i. and 110-130min p.i.). This present project aimed at further investigating the diagnostic properties of earlier FBB images as analysed in a relative quantitative fashion.
Methods: The FBB Phase 2A study was carried out in 18 centres on 3 continents. The present analysis examined the FBB PET data of 77 patients with probable Alzheimer’s dementia (AD, age 55 years or older, MMSE=18-26, CDR=0.5-2.0) and of 69 age-matched healthy controls (HCs, MMSE 蠅28, CDR=0). Volumes of interest (modified AAL template, PMOD) were defined at the imaging core lab to calculate neocortical composite SUV ratios (cSUVRs) using the cerebellar cortex as the reference region. Clinical diagnosis with its known limitations despite thorough neuropsychiatric examination served as the standard of truth given that histopathological assessment was not available for these subjects.
Results: cSUVRs were significantly higher for the AD patients vs. HCs in all three time-windows (p<0.0001, Cohen’s effect sizes=1.54, 1.37, and 1.44). Receiver-operating characteristics curves resulted in cSUVR thresholds which separated the subject groups with sensitivities/specificities of 79%/93%, 73%/96%, and 75%/91% for the three time-windows. For the AD patients, the binary cSUVR readouts at 45-60min p.i. showed 86% and 88% concordance with those of the 90-110min p.i. and 110-130min p.i. images, respectively. For the HCs, the concordance was 99% and 97%, respectively.
Conclusion: Diagnostic accuracy is not compromised when analysing FBB PET data at the 45-60min time-window in a relative quantitative manner. Further research will address the diagnostic accuracy at this time-window when the PET data are visually analysed by trained non-experts. Research Support: Piramal Imaging, Berlin, Germany