About half of the patients with dementia have AD, but there are also various cases of non-AD dementia, such as dementia with Lewy bodies (DLB), frontotemporal dementia (FTD), and vascular dementia (VaD), which must be differentiated from AD. Differentiating between non-AD dementia and AD is important for deciding on the treatment approach, estimating future symptoms, and determining prognosis. There are also a number of reports about differentiation between AD and non-AD dementia with FDG-PET [26, 27, 29, 38, 45–52], but according to a meta-analysis, the sensitivity and specificity of FDG-PET are 93% and 70%, respectively; the specificity is somewhat low [43]. Even if only reports with histopathological confirmation are selected, specificity still tends to be low [48, 50–52]. Among non-AD dementia cases, there are many instances of false positive FDG-PET scans which show hypometabolic patterns similar to those seen in AD.

The frequency of DLB is high among non-AD dementia cases. Differential diagnosis of DLB and AD is important for accurate prognostication and appropriate treatment; however, it is not easy because there is clinically overlapping symptoms and hypometabolism observed in DLB is similar to that observed in AD (Fig. 9.2). In addition to a decline in glucose metabolism that is similar to that found in AD, hypometabolism in the occipital lobe including visual cortex is characteristic of DLB. Low uptake in occipital lobe on SPECT and/or PET is one of the supportive features suggested as a clinical diagnostic criterion for DLB [53], although specificity is not high [45–47]. Also, regarding differentiation between AD and FTD, which are covered by the US Medicare system insurance, FDG-PET has been shown to have 99% sensitivity and 65% specificity in a study of a large number of cases [38]. In reports confirmed by histopathological diagnosis, sensitivity was high at 97% and specificity was 71% [48]. These results suggest that it is not uncommon for DLB and FTD to be diagnosed as AD.

MCI was proposed and revised by Petersen et al. as a summary of cognitive function status expressing normal and dementia statuses [54]. Subtypes include amnestic type and non-amnestic type; single-domain types, in which disability forms in the single higher brain function area; and multi-domain type, in which the disability manifests in many areas. The cause of MCI is heterogeneous and involves a number of conditions; besides AD, degenerating dementia, such as DLB and FTD; cerebrovascular disease, including VaD; and psychiatric conditions, such as depression and external injury-type changes; and normal aging [54, 55]. Among these, amnestic-type MCI patients convert to AD at a rate of about 12–15% per year [54]. If disease-modifying treatment is developed in the future, the MCI stage would be considered to be the appropriate period to begin treatment, so there is an especially high necessity for early diagnosis at this stage. Early diagnosis at the MCI stage using FDG-PET allows prediction of conversion from MCI to AD. In past reports with 1–2-year follow-up periods, the accuracy of predicting conversion from MCI to AD was high, at 80% or greater [56–58], and in a meta-analysis conducted by Yuan et al., sensitivity and specificity both tended to be high, at 88.8% and 84.9%, respectively [59]. However, recent reports using ADNI data suggested low diagnostic ability, with a sensitivity of 57% and a specificity of 67% [60]. This lack of consistency in diagnostic ability may be due to differences in backgrounds of the registered MCI groups, discrepancies in analysis or evaluation methods, or differences in follow-up periods.

In Japan, a prospective multicenter study targeting amnestic MCI, the Study on Diagnosis of early Alzheimer’s disease-Japan, SEAD-J, was conducted; this report contains the results of 3-yr. observations of progress [61]. According to this study, the diagnostic ability of FDG-PET using visual evaluation has a sensitivity of 98%, specificity of 41%, and an accurate diagnosis rate of 71%. This trend coincides with the results of a similar multicenter study using cerebral blood flow SPECT, the Japan Cooperative SPECT Study on Assessment of Mild Impairment of Cognitive Function, J-COSMIC [62], but the overall diagnostic ability of FGD-PET is higher. In the SEAD-J extension, which included a 5-year follow-up of SEAD-J cases, there were cases of MCI that converted to AD in the fourth or fifth year (Fig. 9.3); hence, it may be possible that the low specificity is caused by being a slow converter. Compared with slow converters, rapid converters have clear tendencies for AD-type changes on baseline images (Fig. 9.3). In fact, on assessments by mathematical indicators using AD t-sum, diagnostic ability was highest at 2 years when sensitivity was 70%, specificity was 90%, and accuracy was 83%, whereas at 3 years, sensitivity was 60%, specificity was 91%, and accuracy was 77% [61]. These results indicate that if there is no hypometabolism that suggests AD, the probability of conversion from MCI to AD is low and that selecting cases with clear-cut AD changes using semi-quantitative indicators could make it possible to select rapid converters up to the second year.

In addition to FDG-PET evaluation, concomitant use of the ApoE genotype has been reported to improve accuracy [56, 57]. In any case, in MCI research, histopathological confirmation of the diagnosis is difficult, and results are based on clinical diagnosis using follow-up examinations.

There are a limited number of reports on the use of FDG-PET in the asymptomatic preclinical stage. In the middle age and older groups at high risk for AD due to family history and positive ApoE ε4, hypometabolism of the parietotemporal association cortex has been reported [63, 64]; this abnormality is also seen in their 20s and 30s [65]. In asymptomatic middle aged and older cases with positive ApoE ε4, parietotemporal association cortex and posterior cingulate glucose metabolism declines at an annual rate of 2% [66]. In elderly cases with normal cognitive function progressing to MCI within 3 years of observation, abnormalities were found in the hippocampus, suggesting that abnormal glucose metabolism may start in this area [67]. It is difficult to predict progression to MCI or AD on an individual level because changes in the association cortex and inner part of the temporal lobe during the preclinical stage are normally extremely subtle; but in cases of preventive intervention for asymptomatic cases, FDG-PET may determine efficacy of medical and non-pharmacologic treatment.

In recent years, a substantial number of cognitively normal, elderly individuals without proven deposition of Aβ on amyloid imaging were reported to have at least one significant marker of neurodegeneration including FDG metabolism [68–70]. Cases like these are placed in a separate category from preclinical AD and are designated as “suspected non-AD pathology, SNAP.” It may be affected by the presence of conditions such as cerebrovascular disease, tauopathy, or synucleinopathy, but further verification is required.