Animal fluency improvements and confrontation naming declines over 6 months are associated with differential conversion from MCI to AD in ADNI

Abstract Background Not everyone with mild cognitive impairment (MCI) progresses to Alzheimer’s dementia (AD). Predicting who will decline is important. We previously reported that 6‐months change scores for language, but not for memory, executive functioning, or visuospatial functions, identified a high‐risk subgroup. Here we consider more specific language processes and consider AD fluid and imaging biomarkers. Methods We used longitudinal data from 764 ADNI participants enrolled with MCI and still had MCI at 6 months, which was our baseline for Cox proportional hazards modeling of conversion to AD. We derived scores for four aspects of language: confrontation naming, animal fluency, letter fluency tests, and other language tests. We evaluated changes from enrollment to month 6 for each of these scores. We also adjusted models for APOE genotype, entorhinal thickness and hippocampal volume, and CSF amyloid and tau biomarkers. Results The sample’s mean age was 73.2 years, and 59% were male. From over 2918 total follow‐up years (mean 3.8), there were 287 conversions to AD. Cox models controlling for demographic characteristics, showed improvements in animal fluency scores associated with lower risk (HR=0.58, p =.005), but declines in confrontation naming were associated with higher risk (HR=1.74, p <.001). The animal fluency finding was somewhat attenuated when further controlling the Cox model for hippocampal volume (HR=0.64, p =0.10) and CSF biomarker findings (HR=0.61, p =0.12). Confrontation naming was not impacted by any of the imaging or fluid biomarkers we considered. Conclusions In MCI, the association between six‐month changes in language, and the AD risk appears to be driven by animal fluency improvements and confrontation naming declines. Improvement in animal fluency may reflect still functional access to semantic networks. The attenuation of this signal when controlling for hippocampal volume and CSF biomarkers suggests inability to improve animal fluency may reflect processes associated with hippocampal volumes and CSF biomarkers. Declines in confrontation naming may represent impaired lexical access. The independence of this signal from hippocampal volumes and CSF biomarkers suggests declines in this network may act independently of these biomarkers. These findings suggest intriguing lines of research to further understand the importance of language abilities among people with MCI.