Pei Lin, Tian Xiaoyi, Long Yan, Nan Wenhui, Jia Mei, Qiao Rui, Zhang Jie
The Department of Laboratory Medicine, Peking University Third Hospital The Department of Clinical Laboratory Center, Beijing Children's Hospital, Capital Medical University, National Center for Children's Health The Department of Laboratory Medicine, Peking University People's Hospital, Beijing, PR China.
Medicine (Baltimore). 2018 Sep;97(36):e12178. doi: 10.1097/MD.0000000000012178.
Warfarin is the most common oral anticoagulant. Because of a narrow therapeutic range, interindividual differences in drug responses, and the risk of bleeding, there are many challenges in using warfarin. We need to predict the warfarin maintenance dose. However, ethnic-specific algorithms may be required, and some Chinese algorithms do not perform adequately. Therefore, we aimed to establish a Han Chinese appropriate algorithm.We recruited a study group consisting of 361 Han Chinese patients receiving warfarin treatment who had heart valve replacements. Genotyping of 38 single nucleotide polymorphisms (SNPs) in 13 candidate genes was carried out using the MassARRAY. In the derivation cohort, a multiple linear regression model was constructed to predict the warfarin dosage. We evaluated the accuracy of our algorithm in the validation cohort and compared it with the other 5 algorithms based on Han Chinese and other races.We established a Han Chinese-specific pharmacogenetic-guided warfarin dosing algorithm. Warfarin maintenance dosage (mg/day) = 1.787 - 0.023 × (Age) + 1.151 × (BSA [m]) + 0.917 × (VKORC1 AG) + 4.619 × (VKORC1 GG) + 0.595 × (CYP4F2 TT) + 0.707 × (CYP2C19 CC). It explained 58.3% of the variance in warfarin doses in Han Chinese patients and was superior to the other 5 algorithms. The ability of the 6 algorithms which estimate the required dose correctly was tested. Our model had a mean absolute error of 0.74 mg/day, the other 5 models have mean absolute error of 0.81 mg/day,1.05 mg/day, 1.24 mg/day, 1.18 mg/day, and 0.85 mg/day, respectively. Our model had a mean percentage error of 26.9%, the other 5 models have the mean percentage error of 27.7%, 27.2%, 52.3%, 45.7%, and 29.3%, respectively.Physicians can not adopt algorithm from other race directly to predict warfarin dose in patients with heart valve replacements, they should establish a new algorithm or adjust another algorithm to fit their patients. The algorithm established in this study has the potential to assist physicians in determining warfarin doses that are close to the appropriate doses.
华法林是最常用的口服抗凝剂。由于其治疗窗狭窄、个体间药物反应差异以及出血风险，使用华法林存在诸多挑战。我们需要预测华法林维持剂量。然而，可能需要特定种族的算法，而一些针对中国人的算法效果并不理想。因此，我们旨在建立一种适合汉族人群的算法。我们招募了一个研究组，其中包括361名接受华法林治疗且进行了心脏瓣膜置换的汉族患者。使用MassARRAY对13个候选基因中的38个单核苷酸多态性（SNP）进行基因分型。在推导队列中，构建了一个多元线性回归模型来预测华法林剂量。我们在验证队列中评估了我们算法的准确性，并将其与其他5种基于汉族人群和其他种族的算法进行比较。我们建立了一种汉族特异性的药物遗传学指导的华法林给药算法。华法林维持剂量（mg/天）=1.787 - 0.023×（年龄）+1.151×（体表面积[m]）+0.917×（VKORC1 AG）+4.619×（VKORC1 GG）+0.595×（CYP4F2 TT）+0.707×（CYP2C19 CC）。该算法解释了汉族患者华法林剂量变异的58.3%，优于其他5种算法。测试了6种算法正确估计所需剂量的能力。我们的模型平均绝对误差为0.74mg/天，其他5种模型的平均绝对误差分别为0.81mg/天、1.05mg/天、1.24mg/天、1.18mg/天和0.85mg/天。我们的模型平均百分比误差为26.9%，其他5种模型的平均百分比误差分别为27.7%、27.2%、52.3%、45.7%和29.3%。医生不能直接采用其他种族的算法来预测心脏瓣膜置换患者的华法林剂量，他们应该建立新的算法或调整其他算法以适用于自己的患者。本研究建立的算法有潜力帮助医生确定接近合适剂量的华法林剂量。
