Author(s): Qi Peng aff1 , Shujun Huang [*] aff2 , Xiaoying Chen aff3 , Yayan Yuan aff2 , Yajun Yu aff2 , Liang Tao aff1 , Zhenlu Zhang aff4 , Mousheng Xu aff2
algorithm; Chinese; clinical accuracy; dose; pharmacogenetics; warfarin
Warfarin is a widely used coumarin anticoagulant with a narrow therapeutic range and a large interindividual variability for requirement of doses [ 1 ]. The inappropriate therapeutic dose could cause hemorrhage or thrombosis. Therefore, warfarin is still among the top ten medicines causing severe side effects [2 ]. The factors affecting the warfarin dose requirement among different races and individuals include age, height, weight, diet, coexistent disease, coexistent medication and so on [3 ]. Among all the factors, the genetics accounts for up to 40% of the interindividual and interracial variability [4 ]. The US FDA updated the pharmaceutical direction of warfarin again in 2010 to provide recommended doses based on genotypes of VKORC1 and CYP2C9 .
To reduce the side effects of warfarin and promote the personalized warfarin therapy, researchers have developed numerous warfarin dosing algorithms based on the clinical and genetic information of patients [4-22 ]. Although these algorithms have large significance in the clinical treatment and healthcare decision-making, validating the accuracy of these algorithms with independent populations is important before applying them. Actually, only a few studies have focused on the independent testing work of these algorithms [17,20,23-28 ]. Three studies have compared the performance of these algorithms in Chinese populations [20,26,28 ]. However, the study of Li et al . was limited to the comparison of the predictive abilities of linear and nonlinear models [28 ]; the study of Liu et al . tested several algorithms only in patients under low-intensity warfarin anticoagulation [26 ]; and the study of Tan et al . validated these algorithms in a relatively small cohort of patients receiving heart valve replacement (HVR) surgery (n = 320) [20 ]. Heart valve replacement patients are the major intakers of warfarin and are more sensitive immediately after surgery [29 ]. Therefore, to find the most appropriate warfarin-dosing algorithms for Chinese HVR patients would be beneficial and meaningful. In this study, we collected the clinical information of Han Chinese HVR patients receiving warfarin treatment and genotyped the variants of VKORC1 1639G[greater than]A, CYP2C9*2 and CYP2C9*3 . The clinical accuracy of the nine published algorithms was tested with our cohort.
Materials & methods
A total of 586 patients with HVR were collected from February 2012 to June 2014 in Wuhan Asia Heart Hospital in China. Each patient had received the observed stable dose. In the first week, the international normalized ratio (INR) was measured every day to get the initial stable dose. The initial stable dose was defined as the constant dose under which three successive INRs were within the target INR range of 2.0-3.0. The warfarin dose was decreased by half when the INR was above the target INR range; while the dose was increased by 0.75 mg/day when the INR was below the target INR range. After getting the initial stable dose, the INR was measured...