Effective Functional Pattern Between Prefrontal fNIRS Activity and Cognitive Performance in Patients With Parkinson's Disease: A Data-Driven PLS Approach

Abstract

Background:

Current clinical assessments often fail to detect early prefrontal dysfunction in Parkinson's disease (PD), highlighting the need for objective neuroimaging biomarkers to quantify causal connectivity patterns for early impairment detection.

Methods:

We enrolled 91 PD patients (71.08 ± 5.53 years) who underwent functional near-infrared spectroscopy (fNIRS) recording during both resting-state and verbal fluency task (VFT). Transfer entropy (TE) analysis was performed to quantify effective connectivity (EC) between prefrontal regions. Partial least squares (PLS) was then applied to examine multivariate relationships between task design, MoCA subscales, and VFT performance (verbal output).

Results:

PLS analysis revealed distinct EC patterns associated with task-states (resting state vs. VFT) and cognitive performance. Specifically, EC within the prefrontal network was significantly enhanced during VFT compared to the resting state. These PLS-derived patterns showed consistency with conventional paired t-test results, reinforcing the robustness of findings. Stronger top-down connectivity to the FPA-R (the right frontopolar area) in the resting state was correlated with better language performance on the MoCA. Conversely, enhanced top-down connectivity to the FPA-L (the left frontopolar area) during VFT was correlated with poorer MoCA subscales in attention, abstraction, delayed recall, and orientation, as well as with reduced verbal output.

Conclusion:

PD patients with poorer cognitive performance exhibited enhanced top-down modulation of the FPA-L by other prefrontal regions, suggesting compensatory network adaptation. These findings identify FPA-L as a critical hub for cognitive reserve in PD.