Pharmacogenetics of warfarin dosing in Chinese adults with nonvalvular atrial fibrillation

The guide for the use of genotype-guided warfarin dosing in patients for the treatment of non-valvular atrial fibrillation (AF) is still lacking. We aimed to evaluate whether genotype-guided warfarin dosing is superior to conventional clinical dosing for the outcomes of interest in Chinese patients. Our study consisted of 508 newly recruited and 471 existing Chinese AF patients. Among the total 979 patients, 585 patients received their dose of warfarin determined by a genetic and clinical factor (gene group), while the remaining 394 patients whose dosing was determined empirically in control group. We incorporated CYP2C9 and VKORC1 genotypes into the gene group. The international normalized ratio (INR) measurement and standard protocols were used for further dose adjustment in both groups. The primary outcomes were the percentage of time in the therapeutic range (%TTR) and INR during 12-month follow-up. Secondary safety outcome included bleeding and thrombotic events. Compared with the control group, the average TTR of the gene group was higher [68.4 ± 20.6% vs 48.5 ± 21.6%, P < 0.001]. The average INR monitoring times to reach the therapeutic time in the gene group was lower (P < 0.001). The risk ratios (RR) for cumulative incidence of total bleeding events, minor bleeding events, gastrointestinal bleeding, and intracerebral bleeding events were not significantly different between the two groups (P > 0.05). Comparing to the analysis using existing 471 patients, the analysis using total 979 patients showed that the gene group experienced a lower (RR 0.4 (95% CI 0.2 to 0.8), P = 0.008) incidence of cumulative ischemic stroke. Genotype-guided warfarin administration increases the average TTR, reaches higher TTR levels in the early anticoagulant phase, and significantly reduces the risk of ischemic stroke events.

Email: liujia85912@163.comBackground -Warfarin is an effective treatment of thromboembolic disease but has a narrow therapeutic index.Optimal anticoagulation dosage can differ tremendously between individuals.We aimed to evaluate whether genotype-guided warfarin dosing is superior to conventional clinical dosing for the outcomes of interest in Chinese patients.
Methods -All patients with nonvalvular AF (atrial fibrillation) were randomly divided into two groups having their warfarin dose determined by an algorithm using genetic and clinical factors (genetic group) or the same algorithm using clinical factors only (control group).We included genotypes for CYP2C9 and VKORC1 in the gene group.Genomic DNA was extracted from peripheral blood samples for patients from the genetic group using standard protocols.In the control group, doctors and pharmacists used the warfarin dosing algorithm to determine patients' initial dose.The international normalized ratio (INR) measurement and standard protocols were used for further dose adjustment in both groups.The primary outcome measure was the percentage of time in the therapeutic range (%TTR) of the INR during follow up after initiation of warfarin treatment.

Results
The average TTR was (68.36 ± 20.57) % vs (48.52 ± 21.56) %, P <0.001) in the gene group compared with the control group.At the end of follow-up, the genetic group had a significant lower risk of cumulative incidences of ischemic stroke events in the adjusted model [relative risk (RR) 0.38 (95% CI 0.18 to 0.80), log-rank test P =0.008] than control group.There was no significant difference in the risk ratios (RR) for cumulative incidence of total bleeding events, minor bleeding events, gastrointestinal bleeding and intracerebral bleeding events between the two groups(P >0.05).
Conclusion Genotype-guided dosing could improve the average TTR, improve the safety of treatment, achieve a higher level of TTR in the early anticoagulation period and reduce the risk of ischemic stroke events significantly.

Key words : atrial fibrillation, warfarin, CYP2C9, VKOCR1, time in therapeutic range
Warfarin is an oral anticoagulant, which has been widely used in the treatment and prevention of thromboembolic events including atrial fibrillation (AF) and ischemic stroke [1].The dosage of warfarin varies across individuals and populations, often leading toinadequate or excessive anticoagulation [2].These concerns, coupled with a lack of optimal warfarin dosing recommendations contribute to a very low usage rate of anticoagulants (only 6.16% among eligible patients) in China.To ensure its efficacy and minimize the risk of adverse reactions, warfarin dosage needs to be adjusted based on rapid and simple monitoring methods.The quality of vitamin K antagonist (VKA) management in the clinic is usually measured and reported using 'Time in Therapeutic Range' (TTR) [3].Patients with shorter TTR are more likely to experience adverse cardiovascular events such as venous thromboembolism (VTE) and hemorrhage [4].
More than 30 genes have been reported to be associated with warfarin metabolic pathways.Among them, the warfarin metabolic enzyme CYP2C9 and warfarin target vitamin K epoxide reductase complex 1 (VKORC1) gene polymorphisms are the most significant ones resulting in individual differences, which could account for 30 to 40% of the warfarin dose variation [5].However, previous studies have shown that individual differences in warfarin outcomes are closely related to genetic factors such as CYP2C9 and VKORC1, but the effects of these genetic factors vary by race.The mutation rate of CYP2C9*2 in Caucasians exceeds 10%; CYP2C9*3 in Caucasians and Asians is about 7.5%˜10% and 3%, respectively.CYP2C9*2 mutation is very rare in Asians [6].Observational research shows that CYP2C9 rs1057910, and VKORC1 rs9923231 are related to warfarin dose and bleeding risk in Chinese populations [7].Given what is already known of warfarin, there are still many questions, such as how medical indication may affect warfarin dose requirement in addition to genotypes [6,8].Despite extensive studies, warfarin dosing and its relationship with genetic polymorphisms remain an unresolved problem in China.In particular, the influence of other medical indications on warfarin dose requirement in addition to genotypes needs to be explored.Therefore, our research assessed two genes (CYP2C9*3 and VKORC1c.1639G> A) which may influence warfarin drug responseand evaluate whether genotype-guided warfarin dosing is superior to routine clinical use in Chinese patients with nonvalvular AF.

Study population
We conducted a study of Chinese patients with AF who were selected from the Department of Cardiovascular Medicine in Northern Jiangsu People's Hospital (Yangzhou, China) between April 2016 and January 2021.Patients met the following criteria for inclusion in the study:(a) continuous warfarin therapy for at least 3 months; (b) the monitoring of international normalized ratio (INR) of outpatient during follow-up had been carried out [?]5 times.
The criteria for exclusion were as follows: (a) heart failure (NYHA class 3 or class 4); (b) hepatic dysfunction, (c) cancer, (d) renal failure, (f) thyroid disease, (g) liver cirrhosis.The protocol was carried out in accordance with the Helsinki Declaration and approved by the ethics committee of Northern Jiangsu People's Hospital.Informed Consent in writing was provided by all participants.

Study design
The patients were randomly assigned to two groups to receive their dose of warfarin determined by a genetic and clinical factor (genetic group) or the same clinical factors alone (control group).In the genetic group, we included genotypes for CYP2C9 and VKORC1.Clinical information of each patient was recorded.For genotypes in the genetic group, algorithm estimates included clinical factors to calculate the warfarin dose (predicted dose)and warfarin stable dose.The personalized dose administered on day 1 and the subsequent maintenance dose (from day 2 to day 6) were calculated using the warfarin model predicting doses as previously described [9].The warfarin algorithm was used by doctors and pharmacists to determine the initial warfarin doses in the control group.Further dose adjustments were made in both groups using INR measurements and standard schemes, which are based on consensus management strategies from the American College of Cardiology (ACC)and the American Heart Association (AHA) [10].

Human genomic DNA samples
Using the standard scheme, genomic DNA was extracted from peripheral blood samples of patients in the gene group [11].The primers (one of which was labeled with biotin) were synthesized by Generay Biotechnology Co., LTD.(Shanghai, China).Primers for PCR amplification and pyrosequencing of polymorphic loci of warfarin personalized medication related genesis shown in Table 1.In terms of detection method, each sample requires only one PCR run for surveying of relevant subtypes of VKORC1 (1639G > A) and CYP2C9*3.
The pyrosequencing reaction was analyzed on PyroMark Q24 software.20 μL PCR template was immobilized with 2 μL sepharose beads and 40 μL binding buffer by incubator.For sequence primer annealing, 24 μL 1×annealing buffer and 1 μL sequencing primer were incorporated into each well of 24-well pyrosequencing plate.For strand separation, all liquid was removed by the Vacuum Prep Workstation and then the bead containing immobilized PCR template was captured.The captured beads on probes were transferred to 70% ethanol for 5 s, denaturing solution for 5 s and 1×wash buffer for10 s, respectively.All liquid was totally drained from the filter probes.The sepharose beads with the biotinylated single-stranded templates attached were released into 24-well pyrosequencing plate containing sequence primer annealing.The process was analyzed on QiagenPyroMark Q24 software and compared with the updated consensus standard gene sequence [12].

Evaluation of anticoagulation quality
Patients on warfarin with low-intensity anticoagulation achieved an INR target of 1.8-3.0.The quality of warfarin anticoagulant management for both groups is measured by TTR as reported previously [13].Clinicians responsible for warfarin initiation and on-going dose adjustment were blinded to genotype data.The frequency of visits depended primarily on whether the INRs were in the therapeutic range and the need for dose adjustments of warfarin.INR monitoring is at least once in 3 months for the patients with stable INR reaching the maintenance dose.

Follow-up measures
Data collection and 12-month follow-up were performed by patient interview at each visit to the anticoagulation clinic and through medical records.Primary outcomes of the follow-up were as follows: (a) the onset of AF, (b) the number of anticoagulation visits and TTR, (c) frequency of INR monitoring, (d) thrombotic and hemorrhagic events.
The criteria for exclusion were as follows: (a) INR was not routinely monitored; (b) INR measurements were obtained within the first 7 days and the detection time was more than three months between consecutive INR measurements; (c) INR value was not obtained.Hemorrhagic events were categorized as major (serious/life threatening) or clinically significant non-major bleeding according to previously defined by the 2005 International Society on Thrombosis and Hemostasis criteria [14].Major hemorrhagic events included any bleeding requiring hospitalization or transfusion (usually gastrointestinal bleeding or intracranial hemorrhage).Minor hemorrhagic events were any bleeding events not requiring hospitalization or transfusion.Thrombotic events were deep venous thromboses, pulmonary embolism or thromboembolic strokes that developed or progressed after treatment with warfarin.

Statistical analysis
Continuous variables were expressed as mean ± standard deviation (SD).Clinical characteristics of the two groups were compared by Chi-squared test or t test.Hardy-Weinberg equilibrium for the target SNPs was analyzed by Chi-squared test.CYP2C9 metabolizer status and VKORC1 activity status were determined according to the Clinical Pharmacogenetics Implementation Consortium guideline (https://www.pharmgkb.org).
To explore the factors associated with anticoagulation after warfarin initiation, the correlation analysis of age, gender, height, weight, two gene polymorphisms and stable warfarin dose was evaluated by the Spearman test.Fisher's exact test and Chi-squared test were used to estimate the risk for the outcomes between two groups.Kaplan-Meier curves were plotted by the cumulative event.A P value of less than 0.05 was considered statistically significant.All statistical analyses were performed by using SPSS 19.0 software.

Participants
Between April 2016 and January 2021, 998 eligible patients were invited to participate in the study.A total of 979 patients agreed to join and were randomly assigned to either the genetic group or the control group.The flow of study participants in the genetic group versus control group is shown in Fig. 1.There were 585 patients in the genetic group with an average age of 71.51±6.96(range 35-89 years).There were 394 patients in the control group with an average age of 70.63±6.21(range37-85 years).Demographic and clinical characteristics, including age, gender, CHA2DS2-VASc score, HAS-BLED score and risk factors are summarized in Table 2.There was no statistically significant difference in demographic and clinical characteristics between the two groups (P >0.05).

Genotypic analysis
Pyrosequencing results of CYP2C9*3 and VKORC1 genotype are shown in Figure 2. (a) CYP2C9 AA genotype, (b) AC genotype, (c) CC genotype.The result was detected by forward sequence method.(d) VKORC1 AA genotype, (e) AG genotype, (f) GG genotype.The result was detected by reverse sequence method, so the actual CC genotype detected was GG genotype, while the CT and TT genotypes were GA and AA genotypes, respectively.The genotypic distribution conformed with Hardy-Weinberg equilibrium, which demonstrated that the samples enrolled were from the same Mendelian population.

Comparison of clinical data and stable warfarin dose between patients with different genotypes
Predicted dose was collected for patients using warfarin.Comparison of demographic variables and stable warfarin dose between patients with different genotypes is shown in Table 4.The predicted daily warfarin dose was (2.79 +-0.65) mg, (1.25 +-0.43) mg and (1.56+-1.77)mg in patients with the CYP2C9 AA, AC and CC genotype respectively.The stable warfarin maintenance dose was (2.53 +-0.81) mg, (1.52 +-0.31) mg and(1.41+-1.56)mg,respectively.The dosage of AC genotype was significantly lower than that in patients with AA genotype (P < 0.05), and the dose in patients with CC genotype was also lower than that in patients with AA genotype.
The predicted daily warfarin dose was (2.58 +-0.52) mg, (3.72 +-0.60) mg, (4.22 +-1.62) mg in patients with the VKORC1 AA, GA and GG genotype.The stable warfarin maintenance dose was (2.28 +-0.67) mg, (3.22 +-1.19) mg and (3.59 +-0.23) mg, respectively.The dose of GA patients was significantly higher than that in patients with AA genotype at follows (P <0.01).The stable dose of warfarin in patients with GG type was higher than that in patients with AA type.However, the patients with GG genotype were only two, there was no significant difference in warfarin stable dose between them (P > 0.05).

Primary outcome measure
TTR% in both groups increased with time; Since the INR peaks at around 2 weeks and then plateaus, we included 651 participants with at least 2 weeks of INR data in the analysis.Of the 585 patients in the gene group, TTR could be calculated for 350 patients, 235 were lost to follow-up, with a follow-up rate of 59.83%.Of the 394 patients in the control group, 236 patients could be calculated as TTR and 158 was lost to follow-up, with a follow-up rate of 59.90%.The average TTR was (68.36 +-20.57)% vs (48.52 +-21.56)%, P <0.001) in the genetic group compared with the control group.This represents a difference of 22.42% (95% CI, 16.84 to 28.00, P < 0.001) after adjustment for clinical characteristics.

Secondary outcome measures
INR increased rapidly during the first 2 weeks, then declined slowly after 2 weeks where it stayed within the therapeutic range.The median time for the genome to reach treatment for INR was shorter than for the control group (P <0.001).Among all patients, the frequency of INR testing was at most 32 times.The frequency of the average INR monitoring times in the genetic group was (7.96+-5.63),which was significantly higher [INR difference 2.28 95% CI (1.11 to 3.45), P =0.016] than in the control group (5.69 +-4.85).The frequency of the average INR monitoring times to reach the therapeutic INR in the gene group was (5.95+-4.87),which was significantly higher [INR difference 2.61 95% CI (1.63 to 3.59),P < 0.001] than in the control group (3.34 +-3.89).The results of INR between the two groups are shown in Table 5.

Adverse event analysis
Incidence of cumulative bleeding events and cumulative cardiovascular events between two groups are shown in Table 6.A total of 586 participants were enrolled in adverse event analysis: 350 participants from the gene group and 236 participants from the control group.A total of 61 bleeding events were reported (33 in the gene group and 28 in the control group).The risk ratios (RR) for cumulative incidence of total bleeding events, minor bleeding events, gastrointestinal bleeding and intracerebral bleeding events was not significantly different between the two groups(P >0.05).The incidence of cardiovascular and cerebrovascular events showed that 10 (2.86%) of 350 patients in the genetic group had stroke events, while 18 (7.63%) of 236 patients in the control group had stroke events.At the end of follow-up, the genetic group had a significant lower risk than control group of cumulative incidence of ischemic stroke events in the adjusted model [RR 0.38 (95% CI 0.18 to 0.80), P =0.008].
The incidence of cumulative ischemic stroke in the two groups shown in the Figure 4. Kaplan-Meier curve showed that incidence of cumulative ischemic stroke in the genetic group was lower than in the control group and the log-rank test showed there was a significant difference between the two groups (P = 0.008).

Discussion
To our knowledge, our study is by far the largest randomized trial in a single center on genotype-guided dosing of warfarin in a Chinese population.There sults showed that genotype-guided warfarin dose increased the average TTR, improved the safety of treatment, and the TTR reached a higher level in the early stage of anticoagulation, which significantly reduced the risk of ischemic stroke.
In the search for better dose prediction algorithms of warfarin, polymorphisms of VKORC1 and CYP2C9 genes were studied by many researchers.However, there are some controversies about the effectiveness of CYP2C9 and VKORC1 pharmacogenetic tests related to warfarin therapy dose.COAG study indicated that genotype-guided warfarin dose did not improve anticoagulation control in the first 4 weeks of treatment, but subgroup results suggested that genotype-guided algorithms performed better on predicting the maintenance dose among non-Black patients than predicting among Black patients [15].Shah RR et al found that dose of warfarin guided by CYP2C9/VKORC1 genotypes may be more accurate than empirical dosing in predicting the dose [16].However, the clinical benefit of this is still uncertain.
In this study, impact of different genotypes on warfarin dose and the relationship between warfarin stable dose and model predicted dose were analyzed.For CYP2C9*3 genotyping, there were 540 subjects with wild homozygous AA genotype (92.31%), 43 cases with heterozygous AC genotype (7.35%) and 2 subjects with mutational homozygote CC genotype (0.34%).For VKORC1 genotyping, there were 495 subjects with homozygous AA genotype (84.62%), 88 cases with heterozygous GA genotype(15.04%)and 2 cases with homozygous GG genotype (0.34%).In addition, TTR of anticoagulation and clinical outcomes were compared between the gene group and control group.Therefore, the dosage of warfarin should be reduced for patients with CYP2C9 and VKORC1 mutations.Consistent with previous studies, our result confirmed that patients with AA genotype in VKORC1 required lower doses of warfarin than those with AG or GG genotype [17].
Warfarin anticoagulation results showed that the frequency of the average INR monitoring times was significantly higher in the genetic group than in the control group.In relation to the quality of anticoagulant control in patients treated with warfarin, typically a TTR of 65%-70% is recommended.In Asia, a relatively short three-month follow-up for subjects is less likely to achieve a higher TTR target.Patients with good anticoagulation control within three months may be affected by various factors during subsequent treatment.It is necessary to repeatedly assess the TTR during long-term VKA therapy.The genetic group could achieve the effective TTR, while control group was lower than the effective TTR.Moreover, the average %TTR of 68.36% in our study was higher than the reported 63.8% observed in the EU-PACT study [18], which estimated that gene-guided dosing was relevant with a higher TTR than standard dosing during the initiation of warfarin treatment.Firstly, patients in the genetic group received more detailed medication education because we have unique follow-up system in our center, their compliance was higher, and they were more likely to accept follow-up of more than three months.A previous study reported that the anticoagulation or an AF outpatient service can improve the TTR [19].In addition, the AF out-patient service in our hospital may also improve the value of TTR.The frequency of the average INR monitoring times to reach the therapeutic INR in the genetic group was (5.95+-4.87),which was higher than in the control group significantly.Reduction of frequency of dose titrations using genotype-based algorithms is highly desirable in China, especially long distances from suburban or rural areas to healthcare facilities poses a barrier to optimal anticoagulation treatment [20].
In this study, the follow-up results showed that there was no significant difference in the risk ratios for cumulative incidence of bleeding events between the two groups(P >0.05).The reported rate of total bleeding events observed in our study (9.43%) was much lower than the bleeding events (18%) reported for warfarin-treated Chinese patients in a community-based hospital [21].Unlike previous studies, the incidence of ischemic stroke events was lower than the control group.Our study findings provide assurance that with standard clinical monitoring of INR, warfarin can be safely used in Chinese populations, which demonstrated the utility of genotype-guided dosing of warfarin to optimize individual warfarin dose.In addition, it is necessary to add more genetic polymorphisms to guide the treatment of warfarin, such as CYP4F2 [22].

Limitations
Our research had some limitations.First, although our study is by far the largest single-center study in a Chinese population, future studies with even larger sample sizes and different centers are needed.The observed high discontinuation rate of warfarin therapy in China may be related to a lack of knowledge and overestimation of bleeding risk associated with warfarin therapy.Second, our study only included participants from Jiangsu Province, China.However, the ethnic distribution of the population in Jiangsu Province is very similar to that of the nationwide ethnic distribution in China, therefore, the participants included in our study were likely to be a good representation of the Chinese population.Although the same genes could be used to determine dose requirements in different ethnic groups, the non-genetic related ethnic factors also played an independent and important role in predicting warfarin dose.Therefore, more work remains to be done to develop a robust ethnic-and genotype-guided algorithm for improving warfarin dosing in all ethnic groups in China.Finally, it remains a possibility that additional unknown candidate genes affecting warfarin treatment response may exist.Their discovery awaits future studies.

Conclusions
In conclusion, genotype-guided dosing could improve the average TTR, improve the safety of treatment, achieve a higher level of TTR in the early anticoagulation period, and reduce the incidence of ischemic stroke events significantly.