Multimodal Integration of Gait Dysfunction, Amyloid PET, and Plasma Biomarkers for Differentiating Etiological Subtypes in Mild Cognitive Impairment.

OBJECTIVE

To investigate gait characteristics and plasma biomarkers in individuals with mild cognitive impairment (MCI) stratified by amyloid-β (Aβ) positivity on positron emission tomography (PET), and to evaluate the predictive value for Alzheimer's disease (AD)-related MCI.

METHODS

A total of 168 participants were enrolled, including 50 amyloid PET-negative MCI (MCI-), 51 amyloid PET-positive MCI (MCI+), and 61 cognitively normal (CN) individuals. Gait assessments were conducted using a multi-sensor motion analysis system during dual-task paradigms (cognitive load superimposed on locomotion). Plasma samples were analyzed for neurofilament light chain (NfL), glial fibrillary acidic protein (GFAP), and phosphorylated tau at threonine 217 (p-tau217) via ultra-sensitive immunoassays.

RESULTS

Gait analyses identified 125 features, with 69 distinguishing MCI+ from CN and 36 differentiating MCI+ from MCI-. Receiver operating characteristic (ROC) analyses showed that dual-task gait under the countdown and animal-naming conditions (gait-countdown [GCD] and gait-animal naming [GAN]) discriminated MCI+ from CN with an AUC of 0.850. The combined models integrating GCD and GAN with plasma GFAP or p-tau217 yielded AUCs of 0.919 and 0.951, respectively. Similarly, GCD&GAN features demonstrated an AUC of 0.862 for distinguishing MCI+ from MCI-, with GFAP (AUC = 0.933) and p-tau217 (AUC = 0.986) enhancing predictive performance.

CONCLUSION

This study provides evidence that dual-task gait assessments, when combined with plasma biomarkers such as GFAP and p-tau217, may improve the discriminatory power for identifying AD-related MCI. The integration of biomechanical and molecular markers holds promise for advancing early detection strategies and therapeutic monitoring in AD.