Objective: To characterize distinct Parkinson’s Disease (PD) oculomotor phenotypes and evaluate their correlation with motor severity, cognitive function, and non-motor symptom (NMS) profiles using unsupervised machine learning.

Background: Oculomotor impairment, including saccadic slowing, pursuit deficits, and square-wave jerks, represents a significant, yet under-investigated, aspect of PD heterogeneity. These signs, often conflated with atypical parkinsonism, may offer critical diagnostic insight into early disease progression and neural system involvement.

Method: We conducted a cross-sectional analysis of a clinical cohort including PD patients. The oculomotor examination comprised assessments of saccades, smooth pursuit, gaze limitation, and square-wave jerks. Using K-means clustering, we aggregated patients into distinct phenotypes based on these oculomotor features. Subsequently, we performed correlation analyses between these clusters and key clinical indicators: age of onset, disease duration, motor phenotypes (Tremor-Dominant vs. PIGD), non-motor symptom burden (total NMSS and specific sub-domains), cognitive assessments (MMSE, BREF, Clock Drawing Test), autonomic function (SCOPA-Aut), and daytime sleepiness (Epworth).

Results: Our clustering of 58 PD-patients analysis revealed three distinct phenotypic profiles. Cluster 1 (Oculomotor-Preserved): Patients exhibited normal oculomotor function, showing a strong association with younger age, shorter disease duration, and a Tremor-Dominant phenotype. Cluster 2 (Pursuit-Deficient): Characterized by impaired smooth pursuit, this cluster correlated with a moderate NMS burden and early signs of cognitive dysfunction. Cluster 3 (Saccadic-Dysfunctional/Complex): Defined by slow saccades, square-wave jerks, and vertical gaze limitations, this group was highly correlated with the PIGD motor phenotype, advanced disease duration, significant dysautonomia (SCOPA-Aut), and cognitive impairment.

Conclusion: Oculomotor clustering effectively maps onto a gradient of PD severity. The identification of a Saccadic-Dysfunctional phenotype (Cluster 3), characterized by significant cognitive and autonomic dysfunction, suggests that oculomotor markers are robust, non-invasive indicators of disease progression, warranting integration into routine clinical assessments to identify patients at higher risk for rapid functional decline.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/unveiling-oculomotor-phenotypes-in-parkinsons-disease-a-multidimensional-clustering-analysis/