Route efficiency in spatial navigation as an early indicator of cognitive decline in amnestic mild cognitive...

The hippocampus is one of the brain regions most affected by neurodegeneration in Alzheimer's disease (AD). This structure and its neural circuits are critically involved in spatial learning and memory. Poor spatial navigation performance in virtual environments such as the Virtual Morris Water Navigation Task (VMWNT) may precede other clinical findings in amnestic mild cognitive impairment (aMCI), a pre-dementia stage of AD. To explore this idea, we correlated aMCI and cognitively healthy control (HC) performance in the VMWNT with Montreal Cognitive Assessment (MoCA) and Memory Index Score (MoCA-MIS), testing whether navigation-derived metrics are associated with cognitive performance within a framework of shared medial temporal lobe vulnerability.

Thirty-eight participants (18 aMCI, 20 HC) were assessed for neurologic evaluation and VMWNT performance. Neuropsychological tests, including MoCA, MoCA-MIS, Clinical Dementia Rating (CDR), CDR sum of boxes (CDR-SOB) and AD8, were performed. Group differences were assessed using Mann-Whitney U tests on participant-level means and complemented with mixed-effects models to account for the repeated-measures structure of trial-level data. All behavioral parameters obtained in the VMWNT were reduced to a single variable, "Route Efficiency," through Principal Component Analysis (PCA) on five navigation variables, with the aim to correlate spatial memory performance with clinical findings.

Significant differences in VMWNT performance were observed. aMCI patients displayed longer path lengths (p = 0.006), more quadrant crossings (p = 0.010), reduced time in target quadrant (p < 0.001), and fewer target crossings (p = 0.006), with group differences most evident in Stage 2; Stage 3 showed no significant differences for most variables. Route Efficiency was significantly lower in aMCI than HC (p < 0.001); correlations with MoCA and MoCA-MIS were observed across the full sample but were nonsignificant within groups. Stage 3 allocentric navigation did not reveal consistent group differences. Using nested cross-validation, the AUC for aMCI detection was 0.78 (95% CI: 0.63-0.91).

Group differences in Route Efficiency and correlations with cognitive measures were observed, though these associations primarily reflected between-group rather than individual variation. Allocentric significant findings were not observed. These preliminary results support further exploration of virtual navigation paradigms in clinical settings.