基因指导新疆维吾尔族患者华法林日稳定剂量预测模型的建立及验证

目的：评价新疆维吾尔族患者临床特征及基因型分布，建立华法林日稳定剂量预测模型并验证准确性。方法：按照纳入排除标准选取接受华法林抗凝治疗并达稳定剂量的患者120例，分为建模组（90例）和验证组（30例），应用PCR直接测序法检测VKORC1、CYP2C9基因多态性，记录华法林日稳定剂量、人口学资料、个人史、合并用药、合并疾病，采用多元线性回归方法建立华法林稳定剂量预测模型并进行准确性验证。结果：新疆维吾尔族患者华法林日稳定剂量预测公式为：Dose=3.642+0.910×VKORC1-1 173TC+1.774×VKORC1-1 173CC-2.660×CYP2C9CC+0.991×吸烟×0.02年龄（yr）；该预测模型在验证组患者中的预测剂量与实际剂量无统计学差异（P>0.05）；IWPC预测模型在验证组患者中预测剂量低于实际剂量比例为46.7%；Huang等预测模型在验证组患者中的预测剂量与实际剂量无统计学差异（P>0.05），其中40%的患者为理想预测剂量；本预测模型在心脏瓣膜置换术后患者中预测剂量为理想预测剂量的百分比为70%。结论：本预测模型可预测新疆维吾尔族患者的华法林日稳定剂量，但样本量较少，该预测模型需要进一步的验证。

Establishment and validation of maintenance dose algorithm through genotypes in Xinjiang Uygur population

OBJECTIVE To evaluate clinical features and distribution of genotypes in Uygur patients in Xinjiang and establish an algorithm for predicting daily dose of warfarin.METHODS A total of 120 patients were selected, who received warfarin anticoagulant therapy and reached stable doses, and divided into derivation group (n=90) and validation group (n=30). PCR direct sequencing method was used to detect VKORC1 and CYP2C9 gene polymorphisms to record stable doses, population data, personal history, combined medication, concurrent diseases, multiple linear regression method was used to establish stable maintenance warfarin dose algorithm, of which the accuracy was validated.RESULTS The algorithm of this study:dose=3.642+0.910×VKORC1-1 173TC+1.774×VKORC1-1 173CC-2.660×CYP2C9CC+0.991×0.02 smoking age (yr). No significant difference was observed between predicted dose in validation group and actual dose (P>0.05). Predicted dose by IWPC algorithm in validation group was lower than actual dose in 46.7% cases. No significant difference was observed between Huang algorithm predicted dose in validation group and actual dose (P>0.05), of which 40% were ideal predicted dose. Percentage of ideal predicted dose was 70% in patients after cardiac valve replacement.CONCLUSION Algorithm of this study can calculate maintenance warfarin dose for Xinjiang Uygur patients. However, this algorithm needs to be further verified due to the small sample size.