Background Better understanding and prediction of PD progression could improve disease management and clinical trial design. We aimed to use longitudinal clinical, molecular, and genetic data to develop predictive models, compare potential biomarkers, and identify novel predictors for motor progression in PD. We also sought to assess the use of these models in the design of treatment trials in PD. Methods A Bayesian multivariate predictive inference platform was applied to data from the Parkinson’s Progression Markers Initiative (PPMI) study (NCT01141023). We used genetic data and baseline molecular and clinical variables from PD patients and healthy controls to construct an ensemble of models to predict the annualised rate of the Movement Disorder Society-Unified Parkinson’s Disease Rating Scale parts II and III combined. We tested our overall explanatory power, as assessed by the coefficient of determination (R2), and replicated novel findings in an independent clinical cohort of PD patients from the Longitudinal and Biomarker Study in PD (LABS-PD; NCT00605163). The potential utility of these models for clinical trial design was quantified by comparing simulated randomized placebo-controlled trials within the out-of sample LABS-PD cohort. Findings A total of 117 controls and 312 PD cases were available for analysis. Our model ensemble exhibited strong performance in-cohort (5-fold cross-validated R2=41%, 95% CI: 35% – 47%) and significant, though reduced, performance out-of-cohort (R2=9%, 95% CI: 4% – 16%). Individual predictive features identified from PPMI data were confirmed in the LABS-PD cohort of 317 PD patients. These included significant replication of higher baseline motor score, male sex, and increased age, as well as a novel PD-specific epistatic interaction all indicative of faster motor progression. Genetic variation was the most useful predictive marker of motor progression (2.9%, 95%CI: 1.5–4.3%). CSF biomarkers at baseline showed a more modest (0.3%; 95%CI: 0.1–0.5%), but still significant effect on motor progression prediction. The simulations (n=5000) showed that incorporating the predicted rates of motor progression into the final models of treatment effect reduced the variability in the study outcome allowing significant differences to be detected at sample sizes up to 20% smaller than in naïve trials. Interpretation Our model ensemble confirmed established and identified novel predictors of PD motor progression. Improving existing prognostic models through machine learning approaches should benefit trial design and evaluation, as well as clinical disease monitoring and treatment. Funding Michael J. Fox Foundation for Parkinson’s Research and National Institute of Neurological Disorders and Stroke (1P20NS092529-01).

中文翻译：

---

新诊断帕金森病患者运动进展的临床和遗传预测因子的大规模鉴定：纵向队列研究和验证

背景 更好地了解和预测 PD 进展可以改善疾病管理和临床试验设计。我们旨在使用纵向临床、分子和遗传数据来开发预测模型，比较潜在的生物标志物，并确定 PD 运动进展的新预测因子。我们还试图评估这些模型在 PD 治疗试验设计中的使用。方法 将贝叶斯多变量预测推理平台应用于帕金森病进展标志物倡议 (PPMI) 研究 (NCT01141023) 的数据。我们使用来自 PD 患者和健康对照的遗传数据和基线分子和临床变量来构建模型集合，以预测运动障碍协会-统一帕金森病评定量表第二部分和第三部分的年化率。我们测试了我们的整体解释力，如通过决定系数 (R2) 评估的，并在来自 PD 纵向和生物标志物研究 (LABS-PD；NCT00605163) 的独立临床 PD 患者队列中复制了新发现。通过比较样本外 LABS-PD 队列中的模拟随机安慰剂对照试验，量化了这些模型在临床试验设计中的潜在效用。结果 共有 117 个对照和 312 个 PD 病例可供分析。我们的模型集合在队列中表现出强大的性能（5 倍交叉验证的 R2=41%，95% CI：35% – 47%）和显着但降低的队列外性能（R2=9%, 95 % CI：4% – 16%）。从 PPMI 数据中确定的个体预测特征在 317 名 PD 患者的 LABS-PD 队列中得到证实。这些包括较高基线运动评分的显着复制、男性和年龄增加，以及新的 PD 特异性上位相互作用，所有这些都表明运动进展更快。遗传变异是运动进展最有用的预测标志物（2.9%，95%CI：1.5-4.3%）。基线时的 CSF 生物标志物显示出更温和的（0.3%；95%CI：0.1–0.5%），但仍对运动进展预测有显着影响。模拟 (n=5000) 表明，将运动进展的预测速率纳入治疗效果的最终模型中，可以降低研究结果的可变性，从而可以在比初始试验小 20% 的样本量下检测到显着差异。解释我们的模型集合证实了已建立和确定的 PD 运动进展的新预测因子。通过机器学习方法改进现有的预后模型应该有利于试验设计和评估，以及临床疾病监测和治疗。资助 Michael J. Fox 帕金森研究基金会和国家神经疾病和中风研究所 (1P20NS092529-01)。