Redenšek Sara, Jenko Bizjan Barbara, Trošt Maja, Dolžan Vita
Pharmacogenetics Laboratory, Institute of Biochemistry, Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia.
Department of Neurology, University Medical Centre Ljubljana, Ljubljana, Slovenia.
Front Genet. 2019 May 16;10:461. doi: 10.3389/fgene.2019.00461. eCollection 2019.
The response to dopaminergic treatment in Parkinson's disease depends on many clinical and genetic factors. The very common motor fluctuations (MF) and dyskinesia affect approximately half of patients after 5 years of treatment with levodopa. We did an evaluation of a combined effect of 16 clinical parameters and 34 single nucleotide polymorphisms to build clinical and clinical-pharmacogenetic models for prediction of time to occurrence of motor complications and to compare their predictive abilities. In total, 220 Parkinson's disease patients were included in the analysis. Their demographic, clinical, and genotype data were obtained. The combined effect of clinical and genetic factors was assessed using The Least Absolute Shrinkage and Selection Operator penalized regression in the Cox proportional hazards model. Clinical and clinical-pharmacogenetic models were constructed. The predictive capacity of the models was evaluated with the cross-validated area under time-dependent receiver operating characteristic curve. Clinical-pharmacogenetic model included age at diagnosis (HR = 0.99), time from diagnosis to initiation of levodopa treatment (HR = 1.24), rs165815 (HR = 0.90), rs6280 (HR = 1.03), and rs9904341 (HR = 0.95) as predictive factors for time to occurrence of MF. Furthermore, clinical-pharmacogenetic model for prediction of time to occurrence of dyskinesia included female sex (HR = 1.07), age at diagnosis (HR = 0.97), tremor-predominant Parkinson's disease (HR = 0.88), beta-blockers (HR = 0.95), alcohol consumption (HR = 0.99), time from diagnosis to initiation of levodopa treatment (HR = 1.15), rs1001179 (HR = 1.27), rs4880 (HR = 0.95), rs2293054 (HR = 0.99), rs165815 (HR = 0.92), and rs628031 (HR = 0.80). Areas under the curves for clinical and clinical-pharmacogenetic models for MF after 5 years of levodopa treatment were 0.68 and 0.70, respectively. Areas under the curves for clinical and clinical-pharmacogenetic models for dyskinesia after 5 years of levodopa treatment were 0.71 and 0.68, respectively. These results show that clinical-pharmacogenetic models do not have better ability to predict time to occurrence of motor complications in comparison to the clinical ones despite the significance of several polymorphisms. Models could be improved by a larger sample size and by additional polymorphisms, epigenetic predictors or serum biomarkers.
帕金森病对多巴胺能治疗的反应取决于许多临床和遗传因素。非常常见的运动波动（MF）和异动症在左旋多巴治疗5年后影响约一半的患者。我们评估了16个临床参数和34个单核苷酸多态性的联合效应，以建立临床和临床-药物遗传学模型，用于预测运动并发症发生的时间，并比较它们的预测能力。总共220例帕金森病患者纳入分析。获取了他们的人口统计学、临床和基因型数据。在Cox比例风险模型中使用最小绝对收缩和选择算子惩罚回归评估临床和遗传因素的联合效应。构建了临床和临床-药物遗传学模型。通过时间依赖性受试者工作特征曲线下的交叉验证面积评估模型的预测能力。临床-药物遗传学模型包括诊断时年龄（HR = 0.99）、从诊断到开始左旋多巴治疗的时间（HR = 1.24）、rs165815（HR = 0.90）、rs6280（HR = 1.03）和rs9904341（HR = 0.95）作为MF发生时间的预测因素。此外，预测异动症发生时间的临床-药物遗传学模型包括女性性别（HR = 1.07）、诊断时年龄（HR = 0.97）、震颤为主型帕金森病（HR = 0.88）、β受体阻滞剂（HR = 0.95）、饮酒（HR = 0.99）、从诊断到开始左旋多巴治疗的时间（HR = 1.15）、rs1001179（HR = 1.27）、rs4880（HR = 0.95）、rs2293054（HR = 0.99）、rs165815（HR = 0.92）和rs628031（HR = 0.80）。左旋多巴治疗5年后MF的临床和临床-药物遗传学模型的曲线下面积分别为0.68和0.70。左旋多巴治疗5年后异动症的临床和临床-药物遗传学模型的曲线下面积分别为0.71和0.68。这些结果表明，尽管有几种多态性具有显著性，但临床-药物遗传学模型在预测运动并发症发生时间方面并不比临床模型具有更好的能力。通过更大的样本量以及额外的多态性、表观遗传预测因子或血清生物标志物可以改进模型。
