BACKGROUND: Pharmacogenetics-based algorithms would be especially desirable for patients undergoing heart valve replacement (HVR), who are particularly sensitive to warfarin during the initial treatment phase following surgery. We aimed to derive a warfarin dosing algorithm from data of Chinese patients undergoing HVR, and to compare it with previously published dosing algorithms as applied to our HVR patients. METHODS: 641 Chinese HVR patients on stable maintenance dose of warfarin were enrolled from a single clinic site. Data of 321 patients were used to derive a warfarin dosing algorithm using stepwise multiple linear regression analysis. Previously published algorithms were selected from Pubmed database for comparison. The performance of all the algorithms was characterized according to mean absolute error (MAE) and percentage of predicted doses falling within +/- 20% of clinically observed doses (percentage of ideal prediction) in the other 320 patients. RESULTS: The newly developed algorithm included eight factors: VKORC1-1639G > A, CYP2C9*3, BSA, age, number of increasing INR drugs, smoking habit, preoperative stroke history and hypertension. Our algorithm accounted for 56.4% of variations in the inter-patient warfarin stable doses. All the algorithms showed better performance in a medium-dose (1.88-4.38 mg/day) and high-dose (> or = 4.38 mg/day) groupings than in a low-dose (< or = 1.88 mg/day) grouping. Compared with the 14 previously published algorithms, our algorithm had the lowest MAE (-0.07 mg/day) and the highest percentage of ideal prediction (62.8%) in the total validation cohort. CONCLUSIONS: Our warfarin dosing algorithm is potentially useful for patients whose population profiles are similar to those of our patients.
BACKGROUND: Pharmacogenetics-based algorithms would be especially desirable for patients undergoing heart valve replacement (HVR), who are particularly sensitive to warfarin during the initial treatment phase following surgery. We aimed to derive a warfarin dosing algorithm from data of Chinese patients undergoing HVR, and to compare it with previously published dosing algorithms as applied to our HVR patients. METHODS: 641 Chinese HVR patients on stable maintenance dose of warfarin were enrolled from a single clinic site. Data of 321 patients were used to derive a warfarin dosing algorithm using stepwise multiple linear regression analysis. Previously published algorithms were selected from Pubmed database for comparison. The performance of all the algorithms was characterized according to mean absolute error (MAE) and percentage of predicted doses falling within +/- 20% of clinically observed doses (percentage of ideal prediction) in the other 320 patients. RESULTS: The newly developed algorithm included eight factors: VKORC1-1639G > A, CYP2C9*3, BSA, age, number of increasing INR drugs, smoking habit, preoperative stroke history and hypertension. Our algorithm accounted for 56.4% of variations in the inter-patientwarfarin stable doses. All the algorithms showed better performance in a medium-dose (1.88-4.38 mg/day) and high-dose (> or = 4.38 mg/day) groupings than in a low-dose (< or = 1.88 mg/day) grouping. Compared with the 14 previously published algorithms, our algorithm had the lowest MAE (-0.07 mg/day) and the highest percentage of ideal prediction (62.8%) in the total validation cohort. CONCLUSIONS: Our warfarin dosing algorithm is potentially useful for patients whose population profiles are similar to those of our patients.
Authors: J Liu; H H Jiang; D K Wu; Y X Zhou; H M Ye; X Li; Z Y Luo; Z Guo; Y L Zhang; Y C Wang; W Zhang; H H Zhou; L S Wang Journal: Pharmacogenomics J Date: 2015-12-08 Impact factor: 3.550
Authors: Li Zhao; Chunxia Chen; Bei Li; Li Dong; Yingqiang Guo; Xijun Xiao; Eryong Zhang; Li Qin Journal: PLoS One Date: 2014-04-11 Impact factor: 3.240