Advertisement

VALUE OF AMYLOID IMAGING FOR PREDICTING CONVERSION TO DEMENTIA IN MCI SUBJECTS WITH INITIALLY INDETERMINATE FDG-PET SCANS

      Background: Amyloid and FDG-PET have been previously independently tested for their abilities to predict future cognitive and functional decline. A subset of subjects who undergo FDG-PET have scans that are ambiguous for predicting subsequent cognitive course. We recently reported that for most of such subjects, this ambiguity could be resolved by repeating FDG-PET one year later (JNM 2013; 54:142). In the present analyses, we tested an alternate strategy: namely, analyzing amyloid PET data obtained contemporaneously with the initial FDG-PET scan.
      Methods: A total of 108 subjects with mild cognitive impairment (MCI) (98 early MCI, 10 late MCI) from the Alzheimer's Disease Neuroimaging Initiative undergoing amyloid and FDG-PET within one month of each other and who had at least 2yrs of clinical follow-up were assessed. Conversion was defined by clinical dementia rating increasing from 0.5 to ≥ 1 within 2yrs. Using commercially available software FDA-cleared for clinical use, both a metabolic dementia prognosis index(DPI), optimized for FDG predictive value, and an amyloid tracer retention index(TRI), based on recently published literature and optimized for amyloid imaging diagnostic value, were automatically calculated and used to assess accuracies of predicting conversion to dementia. FDG criteria defined in an independent ADNI-1 data set were applied to classify ADNI-2 subjects as having high, low, or statistically indeterminate risk of progression to dementia.
      Results: Of 108 subjects, 6 (6%) converted to dementia, 12 (11%) reverted to normal cognition, and 90 (83%) remained cognitively stable. By baseline metabolic DPI, 11 (10%) were predicted to show imminent cognitive progression, 71 (66%) had low risk for progression, and 26 (24%) were in the indeterminate range. Repeating FDG for ambiguous scans one year later had yielded a prognostic accuracy of 78%. By amyloid TRI, 19/26(73%) were correctly classified, while applying the DPI to baseline amyloid scans yielded correct classifications in 81% of cases.
      Conclusions: A strategy of following up an indeterminate FDG-PET scan with an amyloid scan obtained within a month had comparable prognostic value for predicting likelihood of conversion of MCI patients to dementia as did repeating FDG-PET scans in such patients one year after the initial FDG-PET.