Thirty-six cognitively normal (CDR 0) subjects underwent neuropsychological assessment and PiB PET imaging. The Memory Function Questionnaire “Frequency of Forgetting” (FoF) variable was used as a measure of selfawareness of memory functioning, and actual memory performance was measured with the 12-word Selective Reminding Test “Delayed Recall” (DR) variable. We calculated an Anosognosia Ratio (AR) for each subject as [self-assessment – actual performance] divided by [self-assessment + actual performance] (range -1.0 to 1.0). More positive scores indicate an overestimation of memory performance, similar to anosognosia, while more negative scores indicate an underestimation of memory performance. A score of 0 is an accurate estimation of performance.

Higher AR score was associated voxel-wise with greater PiB retention (DVR) in the inferior temporal lobe (Z=4.33, voxel-level corrected p=0.03; cluster-level corrected p=0.02). AAL ROI analyses confirmed that subjects who overestimated performance (n=20) had greater PiB retention in medial and lateral temporal cortex, hippocampus, fusiform and amygdala compared to those who did not overestimate performance (n=16; p<0.05). Lower DR score was also associated voxel-wise with greater PiB retention in the inferior temporal lobe (Z=4.24, voxel-level corrected p=0.04; cluster-level corrected p=0.10).

We conclude that clinically normal older adults who overestimated their own memory performance had greater amyloid burden in multiple temporal regions. In addition, amyloid was related to mild memory impairment in the full sample. Clinical follow-up will be required to determine whether these individuals develop memory impairment, signs of anosognosia, or AD.