Construction And Clinical Application Of The Pharmacogenetics-based Warfarin Dosing Algorithm For Chinese Patients

Background:Even with the availability of novel oral anticoagulants, warfarin remains the commonly recommended oral anticoagulant widely used in clinical practice. However, due to the narrow therapeutic range and the large inter-individual variability, establishment of optimal warfarin dose for anticoagulation is difficult and the risk of adverse reactions is high. The effect of warfarin anticoagulant therapy could be improved using pharmacogenetic (PG) dosing algorithm, which integrated genetic variants (CYP2C9 and VKORC1) and non-genetic factors, accounting for about 50%-60% of the variability in the individual daily dose requirement. To date, most of the published algorithms were based on the standard warfarin anticoagulation in Western populations or multi-ethnic groups. Previous studies found that the effect of the pharmacogenetics-guided warfarin dosing could be affected by the differences of race and anticoagulation intensity. The present study aimed to construct and validate a PG refinement algorithm based on a cohort of Chinese patients, and carry out a clinical trial to explore the clinical application of the constructed refined warfarin pharmacogenetics dosing algorithm.Part 1. Construction and validation of the refined pharmacogenetics-based warfarin dosing algorithm in Chinese patientsObjective:According to the clinical characteristics of warfarin anticoagulation and the related genetic variation factors in Chinese patients, this study aimed to develop and validate a pharmacogenetics-based warfarin dosing algorithm (PG algorithm) in a cohort of Chinese patients.Methods:Consecutive Chinese-Han patients (n=310) under stable warfarin treatment were randomized into derivation (n=207) and validation cohort (n=103) prospectively. Sanger sequencing method was applied to detect the polymorphisms of CYP2C9*3、 VKORC1-1639 A/G and CYP4F2 rs2108622 TT. The international normalized ratio (INR) values, warfarin dose adjustment and maintenance were recorded. In the derivation group, factors associated with warfarin stable dose were screened by multivariate logistic regression. In the validation cohort, warfarin doses predicted by various algorithms [including the algorithm established by International Warfarin Pharmacogenetics Consortium (IWPC)] were compared with the actual stable dose. In the validation cohort, the mean absolute error (MAE) was used to evaluate the difference between the predicted dose by different PG algorithms and the actual maintenance dose.Results:In the derivation group we found that age, body surface area (BSA), target INR, INR on day 4 of therapy, CYP2C9*3、VKORC1-1639 A/G and CYP4F2 rs2108622 TT were the independent predictors of the stable dose of warfarin. Conventional PG algorithm was constructed on the basis of age, BSA, CYP2C9 and VKORC1 genotypes. After integrating additional covariates of INR values (INR on day 4 of therapy and target INR) and genotype of CYP4F2 (rs2108622), a refined PG algorithm was established. Conventional and refined PG algorithms could explain 38% and 54% of the warfarin dose variability in the derivation cohort respectively. We found that warfarin dose prediction was more accurate (P< 0.01) with the PG refinement algorithm (0.66 ± 0.49 mg/d) than with the PG conventional algorithm (0.73 ± 0.56 mg/d) and the fixed dose approach (0.82 ± 0.76 mg/d). Compared with IWPC algorithm, the PG refinement algorithm performed better in patients with lower dose requirement (< 2 mg/day group: 0.97±0.45 vs.1.32±0.52, P=0.01;> 2 mg/day to< 4 mg/day group:0.47±0.39 vs. 0.65±0.49, P=0.01), carried at least one of the genetic variants (CYP2C9*3, VKORC1-1639 A/G, or CYP4F2 rs2108622 TT) (0.69±0.62 vs.0.96±0.54, P=0.01) and patients under low intensity anticoagulation (0.62±0.45 vs.0.71±0.51, P=0.01).Conclusion:PG refinement algorithm integrated with early INR response (INR on day 4), target INR and the genotypes of CYP2C9*3, VKORC1-1639A/G, CYP4F2 rs2108622 could improve the accuracy of warfarin dose prediction in Chinese patients.Part 2. Clinical application of the refined pharmacogenetics algorithm for warfarin dosingObjective:To compare the performance of the refined pharmacogenetics algorithm guided warfarin dosing and the standard warfarin therapy in a prospective designed clinical trial.Methods:Consecutive patients initiated on warfarin treatment were recruited and allocated to PG group and STD group, respectively. All patients received the standard doses of warfarin (2.5 or 3.0mg/d) during the first 3 days, and then patients in the PG group were prescribed according to the refined PG algorithm that included age, BSA, INR on day 4 of therapy, target INR and genotypes data (CYP2C9*3, VKORC1 1639A/G and CYP4F2 rs2108622 TT). Patients in STD group were guided according to the standard warfarin anticoagulation. Both of the two groups received the standard INR monitoring principle and the same monitoring period. After 3 months follow up, we compared the percentage of time in therapeutic range (%TTR), percentage of time under therapeutic rang, time to reach stable dose and dose adjustments per patient differences between PG group and STD group. Propensity Score Matching (PSM) analysis was applied to match the baseline and clinical characteristics in the two groups.Results:A total of 114 patients recruited in the PG group, and 122 patients in STD group, with 100 and 106 completing the 3 months follow-up, respectively. The %TTR in the PG group was significantly higher than that in the STD group [(69.47±21.12)% vs. (60.16±20.54)%, P=0.002]. The percentage of time lower than therapeutic range was 21.29% in the PG group, which was significantly lower as compared with 28.05% in the STD group (P=0.016). The mean time to stable maintenance dose in PG group was significantly shorter than in the STD group [(20.91±15.75) vs. (27.44±20.56), P=0.012]. Fewer dosing changes during the first month of therapy were found in PG group than in STD group (1.95±1.20 vs.2.71±1.299, P<0.001). After PSM analysis, the mean %TTR in the PG group was significantly higher than in the STD group (77.18% vs.55.80%, P<0.001). Compared with the STD group, the percentage of time lower than therapeutic range in PG group was less [(14.32±15.74)% vs. (32.20±21.14)%, P<0.001]). Besides, the days to reach therapeutic INR [inter quartile range:11.50(10.00-20.75) vs. 17.00(12.00-27.75), P=0.004], days to reach stable dose [(38.80±13.75) vs. (49.45±18.85), P=0.005] and dosing changes during the first month of therapy [(2.15±1.05) vs. (2.83±1.36), P=0.015] in PG group were significantly less than the STD group.Conclusions:The refined warfarin pharmacogenetics dosing algorithm could improve the effectiveness of the warfarin anticoagulation, and it may be preferable for clinical application.