Authors' Affiliations: Departments of Medical Oncology, Clinical Chemistry, and Trials and Statistics, Erasmus University Medical Center, Erasmus MC Cancer Institute, Rotterdam, the Netherlands; Department of Pharmaceutical Sciences, St Jude Children's Research Hospital, Memphis, Tennessee; and Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden.
Clin Cancer Res. 2013 Sep 15;19(18):5210-7. doi: 10.1158/1078-0432.CCR-13-0487. Epub 2013 Aug 5.
Paclitaxel is used in the treatment of solid tumors and displays high interindividual variation in exposure. Low paclitaxel clearance could lead to increased toxicity during treatment. We present a genetic prediction model identifying patients with low paclitaxel clearance, based on the drug-metabolizing enzyme and transporter (DMET)-platform, capable of detecting 1,936 genetic variants in 225 metabolizing enzyme and drug transporter genes.
In 270 paclitaxel-treated patients, unbound plasma concentrations were determined and pharmacokinetic parameters were estimated from a previously developed population pharmacokinetic model (NONMEM). Patients were divided into a training- and validation set. Genetic variants determined by the DMET platform were selected from the training set to be included in the prediction model when they were associated with low paclitaxel clearance (1 SD below mean clearance) and subsequently tested in the validation set.
A genetic prediction model including 14 single-nucleotide polymorphisms (SNP) was developed on the training set. In the validation set, this model yielded a sensitivity of 95%, identifying most patients with low paclitaxel clearance correctly. The positive predictive value of the model was only 22%. The model remained associated with low clearance after multivariate analysis, correcting for age, gender, and hemoglobin levels at baseline (P = 0.02).
In this first large-sized application of the DMET-platform for paclitaxel, we identified a 14 SNP model with high sensitivity to identify patients with low paclitaxel clearance. However, due to the low positive predictive value we conclude that genetic variability encoded in the DMET-chip alone does not sufficiently explain paclitaxel clearance.
紫杉醇用于治疗实体瘤，其暴露量在个体间存在高度差异。紫杉醇清除率低可能导致治疗期间毒性增加。我们基于药物代谢酶和转运体（DMET）平台，提出了一种基因预测模型，能够检测 225 个代谢酶和药物转运体基因中的 1936 个遗传变异，从而识别低紫杉醇清除率的患者。
在 270 名接受紫杉醇治疗的患者中，测定了未结合的血浆浓度，并从先前开发的群体药代动力学模型（NONMEM）中估算了药代动力学参数。患者被分为训练集和验证集。DMET 平台确定的遗传变异在训练集中与低紫杉醇清除率（低于平均清除率 1 个标准差）相关时，被选择用于预测模型，并在验证集中进行测试。
在训练集上建立了一个包含 14 个单核苷酸多态性（SNP）的基因预测模型。在验证集中，该模型的敏感性为 95%，正确识别了大多数低紫杉醇清除率的患者。该模型的阳性预测值仅为 22%。在多元分析中，该模型仍与低清除率相关，校正了基线时的年龄、性别和血红蛋白水平（P=0.02）。
在 DMET 平台首次用于紫杉醇的大型应用中，我们确定了一个 14 SNP 模型，具有高敏感性来识别低紫杉醇清除率的患者。然而，由于阳性预测值较低，我们得出结论，DMET 芯片编码的遗传变异不能充分解释紫杉醇清除率。
