Electrogastroenterographic profiling of Parkinson's disease: A data-driven taxonomy for delineating patient subtypes...

Parkinson's disease is a neurodegenerative disorder characterized by high clinical heterogeneity, involving both motor and non-motor symptoms. Gastrointestinal dysfunction is prevalent among patients with Parkinson's disease and may be associated with disease subtypes. Cutaneous electrogastroenterogram, a non-invasive technique for quantifying gastrointestinal electrical rhythms, has not been thoroughly investigated in the context of Parkinson's disease subtype analysis.

This cross-sectional study enrolled 71 patients with Parkinson's disease and employed electrogastroenterograms to assess gastrointestinal electrical rhythms. Cluster analysis and machine learning methods were applied to classify patients into subtypes. Spearman's rank correlation analysis was conducted to evaluate the relationship between electrogastroenterograms and clinical variables. Machine learning models were further explored to assess whether the combination of electrogastroenterogram indicators and clinical covariates could reproduce the cluster assignments.

Cluster analysis categorized patients into two subgroups, G1 and G2. The G2 group exhibited significantly higher 9-item Wearing-off Questionnaire scores than the G1 group. Correlation analyses revealed trending associations between preprandial intestinal electrogastroenterograms indicators and the 9-item Wearing-off Questionnaire scores. Machine learning models were able to reproduce the cluster assignments with moderate to good performance in a hold-out test set, and repeated cross-validation yielded consistent estimates.

This study demonstrates that electrogastroenterograms combined with clinical covariates can identify patients with Parkinson's disease subgroups with distinct intestinal dysrhythmia patterns that are associated with wearing-off severity.