Identifying Risk Factors and Constructing Predictive Models for Wearing-Off and Dyskinesia in Chinese Patients With Parkinson's Disease on Long-Term Levodopa Therapy

Aims

This study aimed to investigate the incidence and risk factors of motor complications including wearing-off (WO) and dyskinesia during long-term levodopa (LD) therapy in Chinese patients with Parkinson's disease (PD), and develop corresponding predictive models, thereby providing a basis for personalized treatment strategies.

Methods

This cross-sectional study included 208 consecutive PD patients who were recruited. The presence of WO and dyskinesia was assessed by a 9-item wearing-off questionnaire and the Unified Parkinson's Disease Rating Scale part IV. Univariate and multivariate logistic regression analyses were used to predict the risk factors of WO and dyskinesia. Predictive models for WO and dyskinesia were then constructed, and their diagnostic performance was evaluated using the area under the curve (AUC).

Results

The overall prevalence rate of motor complications was 46.2% (96/208), with a prevalence of 45.7% (95/208) for WO, 22.1% (46/208) for dyskinesia, and 21.6% (45/208) for the simultaneous occurrence of WO and dyskinesia. Younger age at onset (OR 0.92, p < 0.001), higher levodopa-equivalent daily dose (LEDD) (OR 1.00, p < 0.001), and higher Hoehn-Yahr stage (OR 3.41, p < 0.001) were independent risk factors for WO. A predictive model for WO constructed using these three variables demonstrated high diagnostic efficacy with an AUC of 0.887 (95% CI 0.842–0.932), a sensitivity of 84%, and a specificity of 83%. The independent risk factors for dyskinesia included younger age at onset (OR 0.94, p < 0.001), akinetic-rigid type (OR 2.42, p = 0.034), and higher LEDD (OR 1.01, p < 0.001). A predictive model for dyskinesia constructed using these three variables yielded an AUC value of 0.829 (95% CI 0.767–0.897), with a sensitivity of 67% and a specificity of 89%. The two models were both well calibrated and had a high net clinical benefit.

Conclusion

Our findings suggest that the prevalence of motor complications during long-term LD treatment is relatively high among PD patients in China, with WO occurring more commonly than dyskinesia. Younger age at PD onset, higher LEDD, more severe disease, and akinetic-rigid subtype are key predictors of motor complications. The predictive models developed in this study could serve as a potential tool to assist clinicians in identifying patients at higher risk for WO and dyskinesia, and may support personalized treatment optimization.