In 2013 the Amyloid Imaging Task Force developed the Appropriate Use Criteria (Auc) for clinical use of amyloid PET imaging. The AUC utility remains to be empirically tested.
Fifty-three patients underwent F18-Florbetapir scanning at UCLA as part of their individualized diagnostic work up. 50 were evaluated by dementia experts. 3 presented with lobar hemorrhages suggestive of cerebral amyloid angiopathy and were evaluated by stroke experts. Clinical amyloid PET interpretations were performed, and F18-Florbetapir data were further subjected to automated quantitative analysis. Mean standardized uptake volume ratios (SUVR’s) were obtained using the Clark method with whole cerebellum as reference. Images were dichotomized as positive or negative using the recently proposed cut-off of SUVR=1.17. Analyses (T-test and Chi square statistics) were first done in the full sample and then in the dementia-expert sample only.
Subjects were classified based on age of onset (cut-off=65 years) as early onset (EO, N=23) and late onset (LO, N=30), and as Auc-congruent (Auc+, N=39) and incongruent (Auc-, N=14). Compared to Auc- subjects, Auc+ were significantly younger (67 vs. 76 years, p<0.0001). There were no differences in sex, education or disease duration between Auc+ and Auc- or EO and LO. Compared to LO, EO were more likely to be amyloid positive (91% vs. 60%, p=0.01) but showed comparable mean SUVR (1.45 vs. 1.33, p=0.12). Auc+ were as likely to be amyloid positive (74% vs. 71%, p=0.83) and had similar mean SUVR (1.39 vs. 1.36, p=0.43) as Auc-. There was no significant difference in rate of diagnostic (21% vs. 33%, p=0.41) or treatment change (66% vs. 77%, p=0.38) between Auc+ and Auc-. We observed significantly greater rate of diagnostic change in LO compared to EO (43% vs. 13%, p=0.017) but no difference in treatment change (62% vs. 78%, p=0.21). These results remained unchanged in the dementia-expert sample only.
In our preliminary retrospective series we no difference in pre/post-scan diagnosis in Auc- vs. Auc+ and treatment changes in both comparisons. Changes in diagnosis were significantly more common in LO relative to EO suggesting that greater emphasis on scanning LO might be appropriate.
© 2015 Published by Elsevier Inc.