Data-driven AD-Related Pathology Phenotypes Differentiate Slow Gait and Memory Impairment in Mild Cognitive Impairment

Gait impairment is increasingly recognized as a motor marker associated with Alzheimer’s disease (AD) pathology. However, it remains unclear how distinct AD pathology phenotypes relate to gait performance and whether gait impairment reflects AD-related changes independent of memory decline. This secondary analysis used baseline data from older adults with mild cognitive impairment (MCI) enrolled in the J-MINT study (n=280, mean age 73.6 years, SD 4.9; 41.1% female). Plasma amyloid-β composite markers, phosphorylated tau 217, and hippocampal volume were used to derive AD pathology phenotypes through data-driven clustering. Gait speed, memory, and executive function were assessed, and their associations with pathology phenotypes were examined using regression analyses. Three phenotypes were identified: low-pathology profile (LPP), hippocampal atrophy–dominant (HA), and AD‑typical biomarker pattern (ADP). Compared with LPP, both HA and ADP exhibited significantly slower gait speed and poorer memory performance. Slower gait speed and lower memory performance were independently associated with pathology cluster membership; however, gait speed and memory performance were not significantly associated with each other. Executive function did not differ across phenotypes but was associated with gait speed at the individual level. Distinct AD pathology phenotypes are associated with both gait slowing and memory impairment in MCI. The absence of a direct association between gait and memory suggests that gait slowing reflects AD-related functional changes that are partly independent of memory dysfunction, supporting the notion that AD pathology may affect gait and memory through partially distinct mechanisms.