Phenotype-Specific Screening Model for Deep Brain Stimulation in Parkinson's Disease.

Area of Science:

• Neurology
• Medical Technology
• Biostatistics

Background:

• Deep brain stimulation (DBS) is a key Parkinson's disease (PD) therapy, but many eligible patients face delayed or absent referrals.
• Current DBS candidacy assessments, including OFF-ON medication testing and neuropsychological evaluations, are resource-intensive and not always accessible.
• Existing metrics may not fully capture the diverse motor phenotypes in PD, impacting accurate DBS candidacy determination.

Purpose of the Study:

• To evaluate a phenotype-specific, longitudinal screening strategy for earlier identification of PD patients suitable for DBS referral.
• To improve the timeliness and appropriateness of DBS referrals in Parkinson's disease management.

Main Methods:

• Utilized retrospective longitudinal data from the Parkinson's Progression Markers Initiative (PPMI) database.
• Developed supervised logistic regression models to predict DBS referral probability, stratified by PD motor phenotypes (bradykinesia/rigidity-predominant and tremor-predominant) and a combined cohort.
• Applied strict inclusion criteria for clinically relevant screening windows (age <75, Hoehn & Yahr stage 2-3, MoCA ≥26) and assessed model performance using discrimination, calibration, and interpretability metrics.

Main Results:

• Phenotype-specific and combined models showed moderate predictive accuracy (AUC 0.80-0.82) for DBS implantation.
• In bradykinesia/rigidity-predominant PD, functional impairment and sleepiness increased DBS likelihood, while motor severity decreased it.
• In tremor-predominant PD, better cognition and higher motor severity correlated with increased referral likelihood, whereas impulse control issues reduced it.

Conclusions:

• A phenotype-specific, longitudinal screening approach effectively stratifies DBS referral likelihood in Parkinson's disease.
• This method facilitates earlier and more precise DBS referrals, complementing established evaluation protocols.
• The findings support a more personalized approach to identifying DBS candidates in heterogeneous PD populations.