Optimisation of warfarin-dosing algorithms for Han Chinese patients with CYP2C9*13 variants

Existing pharmacogenetic algorithms cannot fully explain warfarin dose variability in all patients. CYP2C9*13 is an important allelic variant in the Han Chinese population. However, adjustment of warfarin dosing in CYP2C9*13 variant carriers remains unclear. To the best of our knowledge, this study is the first to assess the effects of adjusting warfarin dosages in Han Chinese patients harbouring CYP2C9*13 variants. In total, 971 warfarin-treated Han Chinese patients with atrial fibrillation were enrolled in this study. Clinical data were collected, and CYP2C9*2, *3, *13 and VKORC1-1639 G > A variants were genotyped. We quantitatively analysed the effect of CYP2C9*13 on warfarin maintenance dose and provided multiplicative adjustments for CYP2C9*13 using validated pharmacogenetic algorithms. Approximately 0.6% of the Han Chinese population carried CYP2C9*13 variant, and the genotype frequency was between those of CYP2C9*2 and CYP2C9*3. The warfarin maintenance doses were significantly reduced in CYP2C9*13 carriers. When CYP2C9*13 variants were not considered, the pharmacogenetic algorithms overestimated warfarin maintenance doses by 1.03–1.16 mg/d on average. The actual warfarin dose in CYP2C9*13 variant carriers was approximately 40% lower than the algorithm-predicted dose. Adjusting the warfarin-dosing algorithm according to the CYP2C9*13 allele could reduce the dose prediction error. Our study showed that the algorithm-predicted doses should be lowered for CYP2C9*13 carriers. Inclusion of the CYP2C9*13 variant in the warfarin-dosing algorithm tends to predict the warfarin maintenance dose more accurately and improves the efficacy and safety of warfarin administration in Han Chinese patients.


Introduction
Warfarin is the most widely used oral anticoagulant, with ample clinical evidence and an affordable price [1,2].However, the clinical application of warfarin has been significantly limited by its narrow therapeutic window and large interindividual variation.Additionally, inappropriate doses may lead to critical bleeding events [3].Hence, the international normalised ratio (INR) must be frequently monitored to adjust warfarin doses for optimal therapeutic effects [4].Many genetic and clinical factors have been associated with warfarin dosage [5].Numerous studies have shown that gene polymorphisms, including cytochrome P450 2C9 (CYP2C9) and vitamin K epoxide reductase complex 1 (VKORC1), play important roles in warfarin dose variation [6][7][8][9].Multiple pharmacogenetic algorithms have been established to Dongxu Wang and Hualan Wu contributed equally to this work.evaluate the influence of genetic factors on stable warfarin doses [10,11].For CYP2C9, most current algorithms only sequence and analyse the CYP2C9*2 and *3 alleles, which are common in European populations but relatively rare in Asian and African populations [10].In African populations, the rare variants CYP2C9*5, *6, *8, and *11 have also been integrated into algorithms [11,12].However, studies on the contributions of other rare CYP2C9 allele variants to warfarin dosing algorithms are limited.Due to differences in pharmacogenetic backgrounds, warfarin-dosing algorithms based on data from other populations may not be effectively applied to the Han Chinese population [10,13].
CYP2C9*13 was first identified in the Han Chinese in 2004 [14] and has been found almost exclusively in East Asian populations [15].In the Chinese population, CYP2C9*13 is a relatively common CYP2C9 variant other than CYP2C9*3, with a gene frequency of 0.16-0.7%[16][17][18][19].Several in vivo and in vitro studies have demonstrated that CYP2C9*13 is associated with a lower drug metabolic activity than CYP2C9*1 [15,20,21].Furthermore, it has been found that warfarin dose requirements were lower than expected in individuals with CYP2C9*13 [22][23][24].Therefore, previous models overestimated the warfarin dose requirements in CYP2C9*13 carriers.These patients may have an increased risk of bleeding with usual doses.Thus, the present study aimed to assess the effects of warfarin dosing in CYP2C9*13 variant carriers in Han Chinese.We hope that the inclusion of CYP2C9*13 variants will optimise the predictive power of existing models and guide warfarin dosing in CYP2C9*13 variant carriers.

Study participants
A total of 971 Han Chinese patients with atrial fibrillation admitted to Beijing Hospital between January 2013 and October 2019 were enrolled in this study.The inclusion criteria were as follows: age ≥ 18 years old, meeting the indicators for anticoagulant therapy, and agreeing to receive warfarin anticoagulation therapy for ≥ 3 months.The exclusion criteria were as follows: a blood pressure ≥ 170/110 mmHg, active bleeding, abnormal coagulation function, malignant tumour, and pregnancy.Demographic and clinical data including age, sex, height, weight, smoking status, serum creatinine level, combined medications (statins and amiodarone), and warfarin maintenance doses were recorded in detail.The warfarin maintenance dose was defined as the mean daily warfarin dose (weekly doses divided by 7 days) for patients under warfarin treatment after at least two consecutive (7-14 days apart) INRs within the target range (2.0-3.0), while the warfarin dose was consistent [5].This study was approved by the Beijing Hospital Ethics Committee.Informed consent was obtained from all patients.

Gene sequencing
Peripheral venous blood (2 mL) was collected from each subject, anticoagulated with EDTA, and genomic DNA was extracted from the leukocytes by precipitation.PCR amplification and purification of the sequenced products were performed using our previously reported method [17,25].The purified sequencing products were placed in a CEQ™ 8000 DNA sequencer (Beckman Coulter, CA, USA), and the sequencing results were analysed using LaserGene (DNASTAR, Madison, WI, USA) and Chromas software (Technelysium, South Brisbane, Australia).According to CYP2C9 and VKORC1 polymorphisms, and known genome sequences, primers for PCR amplification and sequencing were provided by Tianyihuiyuan Bioengineering Co., Ltd.(Beijing, China) (Table 1).CYP2C9*2, *3, and *13, and VKORC1-1639 G > A genotypes were sequenced.

Statistical analyses
All data were analysed by SPSS 25.0 software and P < 0.05 was considered statistically significant.The measurement data were expressed as the mean and standard deviation (x̄ ± s), and the count data were expressed as percentages (%).The χ 2 test was used to determine whether CYP2C9 and VKORC1 were in Hardy-Weinberg genetic equilibrium.Variance analyses and the χ 2 test were used to compare differences between various groups.The mean warfarin doses between groups were compared using an independent sample t-test.The patient's genetic and non-genetic information were inputted into two validated pharmacogenetic algorithms, International Warfarin Pharmacogenetics Consortium (IWPC) algorithm and Gage algorithm, to calculate the predicted dose [5,26].Based on the method reported by Lindley et al. [27], we calculated the algorithm prediction error (the difference between the predicted and actual warfarin doses) and the adjustment for the existing algorithms.In order to get an adjustment, CYP2C9*13 genotype was coded as 0, 1, or 2 for the number of variant alleles present.We then regress CYP2C9*13 onto the residual of the predicted doses according to the pharmacogenetic algorithm.Therefore, the multiplicative adjustment could be reflected by the ratio of the actual dose to the predicted dose.Finally, the prediction error and the mean absolute error of the adjusted algorithm including the CYP2C9*13 variant was calculated.

Results
Table The mean warfarin dose was 1.63 ± 0.31 mg/d in CYP2C9*13 variant carriers and was significantly lower than in non-carriers (3.33 ± 1.46 mg/d) (P = 0.004).When the patient's clinical and genetic data were put into the Gage algorithm, the results showed that failure to consider CYP2C9*13 led to an overestimation of the warfarin dose by an average of 1.16 ± 0.45 mg/d in variant carriers (Table 3).In contrast, the Gage algorithm had satisfactory predictive power for non-carriers, with a mean prediction error of 0.16 mg/d.The prediction error was significantly higher in *13 carriers than in the non-carriers (P = 0.018).
Similar results were obtained using the IWPC algorithm (Table 4).Here, the prediction error increased more significantly in CYP2C9*13 carriers than in non-carriers (P = 0.022).This indicated that traditional algorithms failed to accurately predict stable warfarin doses in CYP2C9*13 carriers.
Table 5 provides the multiplicative adjustment of the Gage and IWPC algorithms for the CYP2C9*13 variant.Carrying the CYP2C9*13 variant (CYP2C9*1/*13) was associated with approximately a 40% (38-41%) reduction in the warfarin dose requirement.For example, if the stable daily warfarin dose is 3.00 mg/d in CYP2C9*13 carriers according to the Gage algorithm, the physician should adjust the warfarin dose requirement to 3.00 mg × 0.60 = 1.80 mg.We adjusted both warfarin-dosing algorithms adding the CYP2C9*13 adjustment coefficient (Table 6).The prediction error of CYP2C9*13 carriers was significantly reduced in the adjusted algorithms.And the mean absolute error was

Discussion
The cytochrome P450 enzyme encoded by CYP2C9 is the key enzyme involved in S-warfarin metabolism [8].Currently, more than 70 CYP2C9 alleles have been detected and their distribution varies significantly among different populations [10].CYP2C9*2 and *3 are two common CYP2C9 allelic variants, and the frequency of these two alleles is relatively lower in Asian and African populations than in European populations [10,28].CYP2C9*3 is the most common variant in the Han Chinese population with an allele frequency of 4-5%, while CYP2C9*2 is extremely rare [10,13,17].In recent years, the influence of other CYP2C9 alleles on stable warfarin doses has received considerable attention.CYP2C9*5, *6, *8, and *11 have been confirmed as important alleles in individuals of African descent [12,27,29], and carriers of these alleles have significantly reduced warfarin maintenance doses.The Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines also recommend that warfarin dosing be adjusted based on the CYP2C9 genotype of the patient [13].In recent years, a series of studies have identified a variety of rare CYP2C9 variants in the Han Chinese population [17,30].However, studies on the effects of these variants on warfarin maintenance doses are limited.The CYP2C9*13 allele was first reported in a Chinese population [14] and it occurs almost exclusively in East Asian populations [15].In addition to CYP2C9*3, CYP2C9*13 is considered a relatively common allelic variant in East Asian populations, with an allele frequency of 0.16-0.7%[16][17][18][19].In this study, the allele frequency of CYP2C9*13 was intermediate between CYP2C9*2 and CYP2C9*3.Many in vivo and in vitro studies have demonstrated that CYP2C9*13 is associated with a lower drug metabolic activity than the wild type [20,21,31,32].CYP2C9*13 reduced intrinsic clearance by 92%-significantly better than the wild-type CYP2C9*1, in vitro [31].In an in vivo pharmacokinetic study, Hu et al. found that CYP2C9*1/*13 genotype carriers had significantly lower intrinsic clearance than those with the wild-type, and an even lower clearance than those with the most common variant genotype, CYP2C9*1/*3.Studies have also shown that carrying two allelic variants, *3 and *13 (CYP2C9*3/*13), can further reduce the intrinsic clearance, which is comparable to that of CYP2C9*3/*3 [32].Previous studies on warfarin dosing in CYP2C9*13 carriers were single-patient case reports [23,24].Kwon et al. reported a patient with a CYP2C9*3/*13 genotype who achieved the target INR range while taking an extremely low daily warfarin dose (0.57 mg/d) [23], which was significantly lower than the usual dose for CYP2C9*1/*3 carriers.This indicates that CYP2C9*13 may play a synergistic role with CYP2C9*3, resulting in further reduction in warfarin maintenance doses.
Our results are consistent with those of previous in vitro and in vivo studies on CYP2C9*13.Patients with the CYP2C9*13 variant required approximately 40% lower warfarin doses than predicted.This effect significantly exceeded the 19% reduction observed in CYP2C9*2 allele carriers and was slightly higher than the 33% reduction observed in CYP2C9*3 allele carriers [8].Previous studies have confirmed that CYP2C9*2 and *3 carriers have a significantly higher risk of over-anticoagulation and bleeding during warfarin treatment [33].Thus, CYP2C9*13 may also have a similar clinical significance, and carriers may have lower stable warfarin doses and a higher risk of bleeding.
Currently, the Gage algorithm [26] and the International Warfarin Pharmacogenetics Consortium (IWPC) algorithm are two of the most used warfarin-dosing algorithms [5].Both have been used in several large-scale clinical trials [34][35][36].Existing warfarin-dosing algorithms fail to consider the effect of the CYP2C9*13 variant [5,11,25,26].This may result in lower accuracy in predicting warfarin dose requirements for the Han Chinese population.We found that the maintenance doses were significantly reduced in carriers of the CYP2C9*13 variant.When two validated pharmacogenetic dosing algorithms were used to adjust for clinical and other genetic factors affecting warfarin dosing, we found that the actual and predicted doses in patients with CYP2C9*13 variants were remarkably different and that the algorithm significantly overestimated the warfarin dose requirement for CYP2C9*13 carriers.Therefore, these patients are at a higher risk of over-anticoagulation and bleeding events when receiving the usual warfarin doses.Lindley et al. provided a dose-adjustment method for a rare CYP2C9 allele, CYP2C9*5, using a warfarin dosing algorithm [27].Similarly, the prediction error of CYP2C9*13 carriers was significantly reduced by adding the CYP2C9*13 adjustment coefficient to the conventional model, and results were identical for clinical administration of warfarin.
In conclusion, patients carrying the CYP2C9*13 allele required significantly reduced warfarin doses.Higher doses may predispose them to bleeding complications during warfarin administration.In this study, based on the characteristics of CYP2C9 gene polymorphisms in the Han Chinese population, the pharmacogenetic algorithm was improved by incorporating the influence of the CYP2C9*13 variant.This was helpful in improving the accuracy and safety of warfarin therapy in Han Chinese patients.
A limitation of this study was the small sample size of patients with the CYP2C9*13 variant.In future, the accuracy of the conclusions should be verified in a larger population.Secondly, no CYP2C9*13 homozygotes were identified and it is unclear how to further adjust the predicted dose for homozygotes.Finally, this was a retrospective study; future prospective studies are needed to determine whether CYP2C9*13 is associated with an increased risk of adverse events.

Table 1
Primers used for amplification and sequencing of CYP2C9 and VKORC1 genes F forward, R reverse, S sequence, bp base pair F: AGA CGC CAG AGG AAG AGA GT R: TTT GCG CTT ACC CTA TGC CA S: AGA CGC CAG AGG AAG AGA GT

Table 2
Baseline characteristics of the study population

Table 4
Actual and predicted warfarin doses with predictions based on the IWPC Algorithm

Table 5
Multiplicative adjustment for the CYP2C9*13 variant

Table 6
Actual and predicted warfarin doses with predictions based on the adjusted Gage and IWPC algorithms