The performance of connected speech is correlated with cognitive tests in very early Alzheimer disease

Abstract

AbstractBackgroundLanguage impairment is one of the most important clinical features of Alzheimer disease. However, it is very difficult to detect at a very early stage. In this study, we try to analyze the language performance via a set of language tests. At the same time, various biomarkers were also checked and compared.MethodFrom 2020 to 2021, we recruited 11 aMCI and 24 normal controls from the memory clinic of Cardinal Tian Hospital in Taiwan. The clinical diagnosis of aMCI is based on clinical history (with short‐term memory impairment over 6 months), brain MRI, amyloid PET, and various neuropsychological tests. All the participants received the Picture Description Test (PDT), Taiwanese Boston naming test (T‐BNT, 30 items) and semantic fluency test. In PDT, the participants were requested to describe three Taiwanese‐culture‐based pictures to evaluate their performance of connected speech. The responses were recorded, transcribed, and subsequently analyzed by the Chinese tokenize system. Linguistic variables such as the number of total words, content words, grammatically intact sentences, long‐pauses (more than 2s), and fillers were recorded and analyzed.ResultIn our cohort, their baseline MMSE scores are lower (24.1 ± 3.1 V.S. 28.9 ± 1.6, p<0.001). The amyloid positive rate is significantly higher in the aMCI group (54.8% V.S. 29.1%, p = 0.05). The scores of T‐BNT (19.9 ± 4.7 V.S. 24.9 ± 2.3, p<0.001) and semantic fluency test (9.7 ± 3.8 V.S. 17.3 ± 4.0, p<0.001) were significantly lower in the aMCI group. The multiple regression model showed that the variable of non‐repeated words and numbers of long‐pauses were significant in the final model and explained of 32% the total variance of the MMSE (R2 = 0.32; p<0.01).ConclusionEven in the MCI stage, the language‐based examinations can be very different from the normal controls. More non‐repeated words and less long‐pauses in connected speech predict better cognitive performance. Combining with biomarkers, these linguistic variables may detect AD at the earlier stage.