Creation of a gene expression classifier for predicting Parkinson's disease rate of progression.

Abstract

Parkinson's disease (PD) etiology is heterogeneous, genetic, and multi-factorial, resulting in a varied disease from a mild slow progression to a more severe rapid progression. Prognostic information on the nature of the patient's disease at diagnosis aids the physician in counseling patients on treatment options and life planning. In a cohort of PD patients from the PPMI study, the relative gene expression levels of SKP1A, UBE2K, ALDH1A1, PSMC4, HSPA8 and LAMB2 were measured in baseline blood samples by real-time quantitative PCR. At baseline PD patients were up to 2years from diagnosis, H&Y scale ≤ 2 and PD treatment naïve. PD-Prediction algorithm comprised of ALDH1A1, LAMB2, UBE2K, SKP1A and age was created by logistic regression for predicting progression to ≤ 70% Modified Schwab and England Activities of Daily Living (S&E-ADL). In relation to patients negative for PD-Prediction (n = 180), patients positive (n = 30) for Cutoff-1 (at 82% specificity, 80.0% sensitivity) had positive hazard ratio (HR+) of 10.6 (95% CI, 2.2-50.1), and positive (n = 23) for Cutoff-2 (at 93% specificity, 47% sensitivity) had HR+ of 17.1 (95% CI, 3.2-89.9) to progress to ≤ 70% S&E-ADL within 3years (P value < 0.0001). Likewise, patients positive for PD-Prediction Cutoff-1 (n = 49) had HR+ 4.3 (95% CI, 1.6-11.6) for faster time to H&Y 3 in relation to patients negative (n = 170) for PD-Prediction (P value = 0.0002). Our findings show an algorithm that seems to predict fast PD progression and may potentially be used as a tool to assist the physician in choosing an optimal treatment plan, improving the patient's quality of life and overall health outcome.