Predictive factors of global cognitive impairment in de novo Parkinson disease with normal cognition at baseline: a 10-year cohort study.

objectiveCognitive impairment (CI) constitutes a heavy burden to patients with Parkinson disease (PD) and their care partners. This study was to investigate the predictors of cognitive decline in de novo PD patients during the 10-year follow-up period.

methodsPD patients from the Parkinson’s Progression Marker Initiative (PPMI) database with normal baseline global cognition measured by Montreal Cognitive Assessment (MoCA) were included and stratified by their 10-year global cognitive performance. A total of 59 baseline clinical, biological, and radiological variables were compared separately between the patient groups with and without CI (combing mild cognitive impairment and dementia). A predictive model was constructed, followed by the evaluation of model performance using the areas under the receiver operating characteristic curve (AUC).

resultsA total of 155 subjects with normal global cognition at baseline were included, in which 31 patients developed CI during the 10-year follow-up period. After feature selection, a total of 6 features including age were significantly associated with the development of CI. These features were included into the final predictive model. The predictive accuracy of our model was obviously higher than the model using age alone (AUC, 0.868 [95%, 0.800-0.936] vs. 0.734 [0.646–0.822], p ˂ 0.01).

conclusionsUsing our model, risks of CI during the 10-year follow-up period could be predicted with high accuracy. Our study indicated the possibilities and necessities of earlier identification of individuals with higher risks of cognitive decline in the long-term. Further studies are needed before the established application of our model into clinical practice.