Department of Quantitative Health Sciences, Lerner Research Institute, Cleveland Clinic Foundation, Cleveland, OH, USA.
Center for Neurological Restoration, Neurological Institute, Cleveland Clinic Foundation, Cleveland, OH, USA.
Parkinsonism Relat Disord. 2018 Nov;56:70-75. doi: 10.1016/j.parkreldis.2018.06.031. Epub 2018 Jun 19.
To create a multivariable model to predict early cognitive decline among de novo patients with Parkinson's disease, using brief, inexpensive assessments that are easily incorporated into clinical flow.
Data for 351 drug-naïve patients diagnosed with idiopathic Parkinson's disease were obtained from the Parkinson's Progression Markers Initiative. Baseline demographic, disease history, motor, and non-motor features were considered as candidate predictors. Best subsets selection was used to determine the multivariable baseline symptom profile that most accurately predicted individual cognitive decline within three years.
Eleven per cent of the sample experienced cognitive decline. The final logistic regression model predicting decline included five baseline variables: verbal memory retention, right-sided bradykinesia, years of education, subjective report of cognitive impairment, and REM behavior disorder. Model discrimination was good (optimism-adjusted concordance index = .749). The associated nomogram provides a tool to determine individual patient risk of meaningful cognitive change in the early stages of the disease.
Through the consideration of easily-implemented or routinely-gathered assessments, we have identified a multidimensional baseline profile and created a convenient, inexpensive tool to predict cognitive decline in the earliest stages of Parkinson's disease. The use of this tool would generate prediction at the individual level, allowing clinicians to tailor medical management for each patient and identify at-risk patients for clinical trials aimed at disease modifying therapies.
利用简短、廉价且易于纳入临床流程的评估方法，为初诊帕金森病患者创建一个可预测早期认知能力下降的多变量模型。
Parkinson's Progression Markers Initiative 收集了 351 名未经药物治疗的特发性帕金森病患者的数据。将基线人口统计学、疾病史、运动和非运动特征作为候选预测因素。采用最佳子集选择确定多变量基线症状特征，以最准确地预测三年内个体认知下降。
11%的样本经历了认知能力下降。预测下降的最终逻辑回归模型包括五个基线变量：言语记忆保留、右侧运动迟缓、受教育年限、主观认知障碍报告和 REM 行为障碍。模型的区分度较好（乐观调整后的一致性指数=0.749）。相关的列线图提供了一种工具，可用于确定疾病早期个体患者出现有意义认知变化的风险。
通过考虑易于实施或常规收集的评估方法，我们确定了一个多维基线特征，并创建了一个方便、廉价的工具，以预测帕金森病的早期认知下降。该工具的使用将在个体水平上进行预测，使临床医生能够根据每位患者的情况调整医疗管理，并识别有风险的患者，以便进行针对疾病修饰疗法的临床试验。
