Deterioration and predictive values of semantic networks in mild cognitive impairment

Recent study has suggested semantic memory deterioration may be the earliest cognitive changes in Alzheimer's disease (AD). Few previous researchers have investigated specific changes in the semantic structures in the memory of patients with amnestic mild cognitive impairment (aMCI). This study examined the clustering performance in semantic fluency among 160 participants in various MCI subgroups (aMCI single domain, aMCI-sd, n = 30; aMCI multiple domain, aMCI-md, n = 30; non-aMCI multiple domain, naMCI-md, n = 10) as well as a group of mildly impaired individuals with dementia of AD type (DAT, n = 20), and a group of healthy controls (HC, n = 70). Compared with HC group, DAT patients presented deficient clustering in each semantic category related to living things. aMCI-sd group presented defective clustering when dealing with the clustering of items that may be more strongly associated with praxis and perceptual information in the categories that included inanimate living things. aMCI-md group displayed defective patterns similar to those in the aMCI-sd group; however, they displayed more profound deficits in clustering that may require perceptual information. Patients with naMCI-md preserved their ability to perform clustering on all of the categories. The poor clustering of items that may be more strongly associated with praxis could be used as a means of predicting conversion from aMCI-sd to DAT, whereas performance on items that may require perceptual information could be used to predict conversion among aMCI-md patients. These findings demonstrate the degree to which the semantic structures in memory can be used for the assessment of aMCI patients and prediction of conversion to DAT. •Semantic memory deficits may be earliest cognitive changes in Alzheimer's disease.•Structural changes in semantic memory in mild cognitive impairment are unclear.•Clustering performance discriminated mild cognitive impairment from normal aging.•Contents of semantic fluency are valuable in predicting dementia.