Cohort Studies with Postmortem Diagnosis
| Reference and AAN level | Diagnostic standard | Study type | Subjects | Major findings |
| Silverman et al., 2001 (42), AAN level II | Autopsy confirmation and clinical follow-up | Multicenter retrospective analysis based on postmortem diagnosis | AD (n = 97); non-AD (n = 41), such as progressive supranuclear palsy, Parkinson disease, cerebrovascular disease, or mixed | AD was identified in 85/89 (sensitivity, 96%) AD-only cases and 6/8 AD-plus cases (overall sensitivity, 94%). Absence of AD was confirmed in 30/41 cases (specificity, 73%), including 23 with other neurodegenerative dementias. Absence of neurodegenerative disease was confirmed in 14/18 cases (specificity, 78%). Negative PET scan indicated that pathologic progression of cognitive impairment during mean 3-y follow-up was unlikely. |
| Jagust et al., 2007 (39), AAN level II | Autopsy confirmation | Retrospective study with 4-y clinical follow-up and 5 y until death and autopsy | Forty-four subjects with dementia, cognitive impairment, or normal cognitive functions; postmortem diagnosis included AD (n = 20), FTD, DLB, mixed, and vascular dementia | For diagnosing AD, accuracy of 18F-FDG PET (sensitivity, 84%; specificity, 74%) was better than that of initial clinical evaluation (sensitivity, 76%; specificity, 58%). 18F-FDG PET (78%) also had better NPV than did initial clinical evaluation (65%). |
| Minoshima et al., 2001 (41), AAN level II | Autopsy confirmation and clinical follow-up | Retrospective 18F-FDG PET analysis based on postmortem diagnosis, and retrospective 18F-FDG PET diagnosis based on clinical follow-up | AD (n = 10); autopsy-confirmed DLB (n = 11); additional 53 patients with clinically probable diagnosis of AD (n = 40) or DLB (n = 13) based on follow-up evaluation | 18F-FDG PET can distinguish AD from DLB with 90% sensitivity and 80% specificity. |
| Foster et al., 2007 (16), AAN level II | Autopsy confirmation | Retrospective consensus study of 6 dementia experts reviewing clinical history and 18F-FDG PET studies | AD (n = 31); FTD (n = 14); controls (n = 33) | 18F-FDG PET is significantly more accurate in distinguishing FTD from AD than clinical methods. 18F-FDG PET adds important information that appropriately increases diagnostic confidence, even among experienced dementia specialists. Mean interrater κ was 0.31–0.42 for clinical information and 0.73–0.78 for 18F-FDG PET. For AD diagnosis compared with FTD, sensitivity was 96.7% and specificity was 85.7%. |