Diagnostic Accuracy of Amyloid and FDG PET in Pathologically-Confirmed Dementia (S8.005)

OBJECTIVE: To assess the diagnostic accuracy of amyloid and FDG PET compared to pathology in a consecutive series of patients evaluated at an academic dementia clinic. BACKGROUND: Pathological validation studies of amyloid imaging have been conducted largely in end-of-life subjects that may not represent clinical practice. Few studies have compared amyloid PET with other diagnostic modalities. DESIGN/METHODS: 37 patients (mean age 66.5±8.8, MMSE 21.1±6.8) underwent PIB (all) and FDG (N=34) PET. Clinical diagnoses included Alzheimer’s disease (AD, N=10), frontotemporal dementia (FTD, N=24), mild cognitive impairment (N=2) and prion disease (N=1). PET scans were visually interpreted by two raters blinded to clinical diagnosis (PIB: positive or negative; FDG: AD or FTD). Pathologic diagnoses (2.9±1.8 years after PET; 36 autopsies and one biopsy) were based on standard research criteria. RESULTS: Pathologic diagnoses included: high-likelihood AD (N=9), FTD (N=24) and prion disease (N=1). Three patients met criteria for both high-likelihood AD and FTD - these cases were considered AD in evaluating PET performance. PIB reads had a sensitivity of 100% (both raters) and specificity of 88%-92% for AD. Mixed AD/FTD cases were all PIB+. Three of four PIB+ patients (by at least one rater) who did not meet high-likelihood AD criteria had frequent neuritic plaques (CERAD) and one had sparse neuritic but frequent diffuse plaques. FDG reads had a sensitivity of 83%-92% and specificity of 77% for AD. Mixed AD/FTD cases were usually read as AD (Rater 1 2/3, Rater 2 3/3). Positive predictive values: PIB 80%-86%, FDG 67%-69%; negative predictive values: PIB 100%, FDG 89%-94%. Overall accuracy (excluding mixed AD/FTD): PIB 91%-94%, FDG 81% (p>0.35). CONCLUSIONS: Amyloid PET had high sensitivity and negative predictive value for the pathologic diagnosis of AD. Positive test interpretation can be complicated by incidental amyloid co-pathology. Amyloid PET performed at least as well as FDG in predicting underlying neuropathology. Study Supported by: NIH/NIA, Alzheimer9s Association, John Douglas French Alzheimer9s Foundation, Hellman Family Foundation. Disclosure: Dr. Rabinovici has received personal compensation for activities with Eli Lilly & Co. as a consultant. Dr. Rabinovici has received research support from Avid Radiopharmaceuticals, and Eli Lilly & Co. Dr. Lehmann has nothing to disclose. Dr. Rosen has nothing to disclose. Dr. Ghosh has nothing to disclose. Dr. Cohn-Sheehy has nothing to disclose. Dr. Trojanowksi has received personal compensation for activities with Pfizer Inc., Johnson & Johnson, MetLife, and Bristol-Myers Squibb Co. as a consultant. Dr. Trojanowski has received royalty payments through Penn licenses. Dr. Trojanowksi has received research support from AstraZeneca and Bristol-Myers Squibb Co. Dr. Vinters has nothing to disclose. Dr. Dickson has received personal compensation for activities with Neotope, Inc. as a consultant. Dr. Gorno-Tempini has nothing to disclose. Dr. Boxer has received personal compensation for activities with Archer, Envivo, Grifols and Iperian as a consultant. Dr. Boxer has received research support from Allon, Forest, Genentech Inc., Bristol-Myers Squibb Co., Janssen Pharmaceutica, and Pfizer Inc. Dr. Kramer has nothing to disclose. Dr. Miller has received personal compensation in an editorial capacity for Cambridge University Press, Guilford Publications, Inc., and Neurocase. Dr. Grinberg has nothing to disclose. Dr. Seeley has nothing to disclose. Dr. Jagust has received personal compensation for activities with Genentech Inc., Synarc, Janssen Alzheimer Immunotherapy, TauRx, and Siemens. Dr. Jagust has received research support from the National Institutes of Health.