Motor progression phenotypes in early-stage Parkinson’s Disease: A clinical prediction model and the role of glymphatic system imaging biomarkers

Abstract

BackgroundSubstantial heterogeneity of motor symptoms in Parkinson’s disease (PD) poses a challenge to disease prediction. ObjectivesThe aim of this study was to construct a nomogram model that can distinguish different longitudinal trajectories of motor symptom changes in early-stage PD patients. MethodsData on 90 patients with 5-years of follow-up were collected from the Parkinson’s Progression Marker Initiative (PPMI) cohort. We used a latent class mixed modeling (LCMM) to identify distinct progression patterns of motor symptoms, and backward stepwise logistic regression with baseline information was conducted to identify the potential predictors for motor trajectory and to develop a nomogram. The performance of the nomogram model was then evaluated using the optimism-corrected C-index for internal validation, the area under the curve (AUC) of the receiver operating characteristic (ROC) curve for discrimination, the calibration curve for predictive accuracy, and decision curve analysis (DCA) for its clinical value. ResultsWe identified two trajectories for motor progression patterns. The first, Class 1 (Motor deteriorated group), was characterized by sustained, continuously worsening motor symptoms, and the second, Class 2 (Motor stable group), had stable motor symptoms throughout the follow-up period. The best combination of 7 baseline variables was identified and assembled into the nomogram: Scopa-AUT [odds ratio (OR), 1.11; p = 0.091], Letter number sequencing (LNS) (OR, 0.76; p = 0.068), the asymmetry index of putamen (OR, 0.95; p = 0.034), mean caudate uptake (OR, 0.14; p = 0.086), CSF pTau/α-synuclein (OR, 0.00; p = 0.011), CSF tTau/Aβ (OR, 25434806; p = 0.025), and the index for diffusion tensor image analysis along the perivascular space (ALPS-index) (OR, 0.02; p = 0.030). The nomogram achieved good discrimination, with an original AUC of 0.901 (95% CI, 0.813–0.989), and the bias-corrected concordance index (C-index) with 1,000 bootstraps was 0.834. The calibration curve and DCA also suggested both the high accuracy and clinical usefulness of the nomogram, respectively. ConclusionsThis study proposes an effective nomogram to predict different motor progression patterns in early-stage PD. Furthermore, the imaging biomarker indicating glymphatic function could be an independent predictive factor for PD motor progression.