Correlations of CSF Biomarkers of Alzheimer’s Disease with Cognitive Measures in MCI and AD Dementia: A Cross-Sectional Analysis

Background and Objectives: We examined the relationships between cerebrospinal fluid (CSF) amyloid-β, tau, p-tau, and cognition in an Alzheimer’s Disease Neuroimaging Initiative (ADNI) dataset. Studies have examined the longitudinal relationships between CSF biomarkers for Alzheimer’s disease (AD) and cognition, but there is discordance in the strength of associations between CSF amyloid-β, t-tau, or p-tau with cognition at different disease stages. Methods: The study included 665 patients from the combined ADNI dataset: 128 cognitively normal (CN), 175 with mild cognitive impairment (MCI), and 362 with AD dementia. All patients were amyloid-β–positive according to established CSF amyloid-β cut-off values. Cognitive assessments, diagnoses, and specimen collection/processing were conducted via published standardized protocols. We examined cross-sectional baseline data and performed correlational and regression analyses to evaluate the associations between CSF biomarkers and assessments of learning, memory, executive function, language, attention, visuospatial skills, and activities of daily living. Results: In the MCI cohort, significant negative correlations were observed between CSF amyloid-β and ADAS-Cog11 (r=-0.164, p=0.02, ADAS-Cog13 (r=-0.181, p=0.01), Trails B (r=0.11, p=0.04), and FAQ (r=-0.131, p=0.01). There were positive correlations between amyloid-β and MMSE (r=0.149, p=0.004) and amyloid-β and WMS-delayed recall (r=0.116, p=0.03). Statistically significant correlations were observed between CSF t-tau and p-tau and CDRSB (t-tau: r=0.105, p=0.047), ADAS-Cog11 (t-tau: r=0.114, p=0.03; p-tau-181: r=0.114, p=0.03), ADAS-Cog13 (t-tau: r=0.165, p=0.002; p-tau-181: r=0.162, p=0.002), RAVLT-forgetting (t-tau: r=0.173, p=0.001; p-tau-181: r=0.167, p=0.001), and WMS-delayed recall (t-tau: r=-0.237, p<0.001 and p-tau-181: r=-0.235, p<0.001). In the AD cohort, no statistically significant relationships were observed between CSF amyloid-β₁₋₄₂, t-tau, p-tau-181, and any of the cognitive scores. Discussion: The findings suggest that the relationship between CSF biomarkers and cognitive performance is strongest in MCI. The lack of significant correlations in the AD cohort may indicate other pathophysiological changes dominating cognitive dysfunction at this disease stage. Both tau and amyloid showed similar utility in reflecting cognitive impairment, in contrast to some reports in the literature. Further research is warranted to explore biomarker longitudinal impacts and their predictive values across the spectrum of cognitive impairment.