68-5

61/120

DLB, and the VOI2 is the area that remains after the VOI1 is subtracted from the SVA-A, the volume of interest of AD. Independently, these indices reflect the quantity of blood flow decrease in the respective areas. Significant difference was observed, suggesting the progression from MCI to AD or DLB. However, the CIScore, the ratio of these indices, did not show significant difference. The MCI group includes “due to AD” and “due to DLB”, thus implying the possibility that character-istics of both diseases start developing from the MCI stage. Further research needs to be conducted to compare patients with healthy people and longi-tudinally evaluate indices. In the case of poor blood flow reduction, it is assumed that the above results were obtained because the relative difference increased by taking the ratio.The CIScore was highly effective in differenti-ating early-onset AD and DLB. In a previous study, Imabayashi et al. used the CIScore for differenti-ating late-onset AD and DLB13) (AUC0.882). The present study achieved a higher differentiating capability. This may reflect the fact that compared with late-onset dementia, blood flow decrease is more prominent in the VOI than in the area of brain atrophies in morphological imagery in early-onset dementia., and higher values tend to appear in the SVA16). Therefore, this suggests that it may be useful for differentiation in early-onset dementia. However, the fact that the clinical diagnosis also referred to the cerebral blood flow SPECT imaging findings may be the cause of the particularly high sensitivity and specificity levels.In the MCI due to DLB group, the CIScore tended to decline over time in a single case. In other words, the CIS findings increased. Iizuka et al. have reported that the CIS findings increased as the MMSE score decreased down to around 22. However, below the score of 22, the CIS findings disappeared17). In the present study, where the final MMSE score was 27, the results were consis-tent with the tendency reported by them. There-fore, in the early-onset DLB group, the MMSE was predicted to correlate with the CIScore reflecting changes in blood flow. However, no apparent correlation was observed as a whole. Iizuka et al. have argued that the disappearance of the CIS find-ings is related to the progression of AD pathology, thus explaining the changes in the CIS findings by the chronological changes of the posterior cingulate cortex and precuneus plus cuneus17). In general, the neuropathological findings of DLB include numerous Lewy bodies in the cerebral cortex and AD pathological findings such as senile plaques and degraded neurofibrils (common form). In contrast, early-onset cases often exhibit only Lewy lesions without showing any AD lesions (pure form)18). Moreover, just like atrophies and tissue denatur-ation in the cerebral cortex differ between AD and SDAT, the characteristics of DLB were maintained in early-onset cases from MCI to more advanced conditions. Consequently, no correlation between the MMSE and CIScore was observed.For differentiating early-onset dementia cases, eZIS was used. When severity was used for differ-entiating MCI from AD and DLB, the AUC values were 0.960 and 0.911, respectively. The CIScore was used for differentiating AD from DLB, and the threshold value was set at 0.225. The resultant AUC was 0.941, and the accuracy, sensitivity, and specificity were 90.6%, 87.5%, and 93.7%, respec-tively. The SVA-A was suggested to be useful for evaluating the conversion of MCI into both early-onset AD and DLB. Furthermore, the CIScore was suggested to be highly effective in differentiating AD from DLB not only in late-onset dementia but also early-onset dementia cases. It is important to understand the characteristics of each index and apply it to clinical practice.Not applicable.No funding was received.TO and HA were the doctors in charge of these cases and collected the data. MN analyzed and interpreted the data and wrote the manuscript. TT, KK, NS and YI provided guidance on manu-script writing. All authors have read and approved the final manuscript.The authors declare that there are no conflicts of interest.511AcknowledgmentsFundingAuthor contributionsConflicts of interest statement