Pharmacogenetic variants in pharmacokinetic and pharmacodynamic pathways and clinical factors associated with bleeding risk in Thai patients receiving rivaroxaban: a case

Abstract

Background:

While bleeding is an inherent risk of all anticoagulant therapies, individual susceptibility varies considerably. Rivaroxaban, a direct oral anticoagulant widely used for thromboembolic prevention, is associated with substantial interindividual variability in bleeding risk, posing an ongoing clinical challenge.

Objectives:

To investigate associations between pharmacogenetic variants in five pharmacokinetic (PK) and nine pharmacodynamic (PD) genes, along with clinical factors, and bleeding outcomes in Thai patients receiving rivaroxaban.

Design:

A retrospective case–control study was conducted among 91 patients with atrial fibrillation, including 27 who experienced bleeding events.

Methods:

Genetic variants in PK and PD genes were genotyped using the MassARRAY^® system, and associations between pharmacogenetic variants, clinical factors, and bleeding risk were evaluated using logistic regression analysis.

Results:

The CT genotype of ABCB1 rs10276036 (C>T) was associated with an increased risk of bleeding. Clinical factors, including prior bleeding history, were also significantly associated with increased bleeding risk. In addition, allele frequency patterns suggested population-specific genetic variation in this Thai cohort. These findings suggest that genetic and clinical factors may jointly contribute to bleeding susceptibility.

Conclusion:

Bleeding risk in Thai patients receiving rivaroxaban appeared to be influenced by both pharmacogenetic variants, particularly the ABCB1 rs10276036, and established clinical factors. These findings are exploratory and highlight the potential value of integrating genetic and clinical information for risk stratification, pending validation in larger cohorts.

Trial registration:

As a retrospective non-interventional study, this work was not registered in a clinical trial registry. Ethical approval was obtained (COA No. 1657/2022).

Plain language summary

Genetic differences and personal health factors related with the risk of bleeding in Thai patients who take rivaroxaban

Why was the study done?

Blood thinner agent (rivaroxaban) can increase the risk of bleeding side effects. However, different people have different risks. One possible reason is that people process medicines differently because of their genes. This study wanted to find out whether specific genetic differences, combined with clinical factors (such as kidney function or concurrent medicine use), could help identify which patients are more likely to bleed while taking rivaroxaban.

What did the researchers do?

The researchers reviewed medical records of 91 Thai patients with atrial fibrillation who were taking rivaroxaban. They compared the genetic profiles of patients who had bleeding events with those who did not. They tested several genes known to influence how the body handles rivaroxaban or responds to it. They applied statistical analyses to evaluate the association between genetic variants and clinical factors with bleeding events.

What did the researchers find?

They found that the variant in gene involved with drug transport, ABCB1 rs10276036 CT genotype showed the association with increased bleeding risk. Clinical factors also mattered: people with previous bleeding had a higher risk. These genetic associations should be interpreted in conjunction with clinical factors.

What do the findings mean?

These results suggest that doctors could better consider which patients are likely to bleed on rivaroxaban by looking at both genetic and clinical information. This could eventually help personalize care—choosing the safest dose or even a different medication for high-risk individuals.

Keywords

bleeding pharmacodynamic pharmacogenetic pharmacokinetic rivaroxaban

Introduction

Anticoagulants are essential for preventing and treating venous thromboembolism and reducing thromboembolic risk in patients with atrial fibrillation (AF) and valvular heart disease.^1,2 Rivaroxaban, a direct oral anticoagulant (DOAC) that selectively inhibits factor Xa, is widely used for stroke prevention in AF but remains associated with significant bleeding risk.³ In non-valvular AF (NVAF), bleeding occurs in approximately 7.3% of patients, with 2.7% classified as major, predominantly gastrointestinal.³

Rivaroxaban elimination involves both hepatic metabolism through cytochrome P450 (CYP) enzymes and transport via P-glycoprotein (P-gp), with approximately 65% cleared hepatically and 35% renally.⁴ CYP3A4 and CYP3A5 mediate roughly 18% of metabolism, whereas CYP2J2 contributes 14%⁴. These pharmacokinetic (PK) pathways are regulated by transporter genes (ABCB1, ABCG2) and drug-metabolizing enzymes.^4,5 Single nucleotide polymorphisms (SNPs) in these genes have been studied for their influence on rivaroxaban PK and clinical outcomes.^6–8 For instance, ABCB1 rs4148738 affects plasma concentrations but not bleeding risk in Chinese patients with NVAF.⁶ Similarly, ABCB1 rs1045642 may enhance thromboembolic prevention, although neither rs1045642 nor rs2032582 showed bleeding risk differences in Caucasians.⁷ Variants in CYP3A5 rs776746 and CYP3A4 (rs2242480, rs4646437) were also not linked to bleeding risk or trough concentrations,⁶ whereas CYP2J2 rs11572325 has been associated with increased bleeding risk, including epistaxis, within 4 months of therapy.⁸

Pharmacodynamic (PD) genes have likewise been implicated in rivaroxaban response variability. An exome-wide study found that SUSD3 rs76292544 was significantly associated with higher bleeding risk within 12 months in Chinese patients with NVAF.⁹ In the same cohort, NCMAP rs4553122 correlated with increased anti-factor Xa levels, whereas PRF1 rs885821, PRKAG2 rs12703159 and rs13224758, and POU2F3 rs2298579 were associated with decreased activity.⁹ Furthermore, AKR7A3 rs1738023 and rs1738025 and ABCA6 rs7212506 were linked to bleeding susceptibility.¹⁰

This study extends previous findings by evaluating a broader range of PK and PD variants, marking the first comprehensive pharmacogenetic assessment of rivaroxaban in the Thai population. Unlike prior studies limited to specific SNPs in Chinese and Caucasian cohorts,^6,7,9 this research investigates genetic contributions from PK and PD pathways, integrating with clinical factors to better characterize their individual associations with bleeding risk. This approach aims to refine the identification of genetic determinants of bleeding risk and inform personalized rivaroxaban therapy.

Materials and methods

The reporting of this study conforms to the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.¹¹

Study design

This retrospective case–control study evaluated genetic risk factors for bleeding in Thai patients with AF receiving rivaroxaban at King Chulalongkorn Memorial Hospital, Thai Red Cross Society, Thailand and was approved by the Institutional Review Board of Faculty of Medicine, Chulalongkorn University (COA No. 1657/2022, IRB No. 0720/65) on December 6, 2022. The study adhered to the Declaration of Helsinki, and all participants provided written informed consent before enrollment.

Study participants

Eligible participants were Thai adults (⩾18 years) diagnosed with AF (ICD-10-CM code I48), treated continuously with rivaroxaban for at least 2 weeks, and who provided blood samples. Patients were excluded if they were receiving other anticoagulant agents within 1 week prior to enrollment, had known coagulation disorders (hemophilia A/B or von Willebrand disease) or clinically significant thrombocytopenia (platelet count < 90,000 cells/µL, immune thrombocytopenia or secondary to chemotherapy with bone marrow suppression), experienced postoperative bleeding within 1 week, or did not provide informed consent.

Relevant demographic and clinical data were extracted from electronic medical records to reduce recall bias. Bleeding events were classified by the International Society on Thrombosis and Haemostasis criteria,^12,13 and assessed from October 1, 2013, to September 30, 2022. Cases were defined as AF patients who experienced bleeding after rivaroxaban initiation. To minimize selection bias, controls were AF patients using rivaroxaban who did not experience a bleeding event and visited on the same day or the closest day to the cases group (2:1 ratio) during at least 2-year follow-up period after initiation of rivaroxaban. The sample size was calculated for an unmatched case–control study, following the previous study.¹⁰ The expected exposure proportion was 0.95 among cases and 0.70 among controls, with a control-to-case ratio of 2:1. A two-sided significance level of α = 0.05 and a statistical power of 80% were assumed. The minimum required sample size was 28 cases and 56 controls, a total of 84 participants.

DNA extraction

Whole blood samples were collected in ethylenediaminetetraacetic acid (EDTA) tubes and centrifuged at 4000 rpm for 10 min at 4°C to separate the buffy coat. Genomic DNA was extracted from the buffy coat using the FlexiGene DNA Kit (Qiagen, Hilden, Germany). DNA concentration and purity were assessed using the NanoDrop One spectrophotometer (Thermo-Scientific, MA, USA). Samples were required to have concentrations above 20 ng/μL, an optical density (OD) 260/280 nm ratio between 1.5 and 2.2, and OD 260/230 nm ratio greater than 1.5.

Genotyping

Genetic variants in PK genes (ABCB1, ABCG2, CYP2J2, CYP3A4, and CYP3A5) and PD genes (ABCA6, AKR7A3, APOB, APOE, F13A1, POU2F3, PRF1, PRKAG2, and SUSD3) associated with rivaroxaban response were genotyped using the MassARRAY^® system (Agena Bioscience, CA, USA). CYP3A4 and CYP3A5 haplotypes and CYP3A5 metabolizer phenotypes were classified based on the Clinical Pharmacogenetics Implementation Consortium (CPIC) guidelines.¹⁴ Hardy–Weinberg equilibrium (HWE) was evaluated using the exact test in the HardyWeinberg R package (R version 4.3.1) following established recommendations for genetic association studies.¹⁵ Allele and genotype frequencies for bleeding and non-bleeding groups, as well as HWE results, are presented in Table S1. Population-based allele frequency data for global, East Asian (EAS), and South Asian (SAS) populations were retrieved from the 1000 Genomes Project Phase III through the Ensembl database (Tables 2 and S2). Variants with a call rate ⩾ 90% and a minor allele frequency (MAF) ⩾ 1% were included for association analyses. The reference and alternative alleles were defined according to the NCBI dbSNP database.

Statistical analysis

Baseline characteristics were summarized as means ± standard deviations for continuous variables and as frequencies with percentages for categorical variables. Associations between genetic or clinical factors and bleeding events were assessed using Stata Statistical Software: Release 19.5 (StataCorp LLC, College Station, TX, USA). Differences between bleeding and non-bleeding groups were evaluated using independent t-tests for continuous variables and Fisher’s exact test for categorical variables. Multiple logistic regression analyses were conducted separately for PK and PD genes. To account for confounding, established clinical factors were included as adjustment variables in both models to estimate the independent associations of genetic variants. Factors with p ⩽ 0.10 and clinically significant factors, along with those approaching the predefined threshold in univariate analyses, were included in the respective multivariable models.

Results

Patient demographic and clinical characteristics

A total of 91 patients with AF receiving rivaroxaban were included, comprising 27 patients with bleeding events and 64 without (Figure 1). Among bleeding cases, 48.15% were classified as major bleeding, 37.04% as clinically relevant non-major bleeding, and 14.81% as minor bleeding. Most common events were upper gastrointestinal bleeding (n = 8) and lower gastrointestinal bleeding (n = 7), followed by gross hematuria (n = 4), epistaxis and intramuscular hematoma (n = 2 each), and one case each of bleeding gum, hemoptysis, intracranial hemorrhage, and subconjunctival hemorrhage. Most participants were elderly (aged > 70 years) and received dose adjustments based on renal function. Although most baseline characteristics were similar between groups, prior bleeding history was substantially more frequent in patients who experienced bleeding and represents an important clinical confounder that should be considered when interpreting the results (Table 1). Variables such as age, sex, weight, HAS-BLED score, and comorbidities did not differ significantly. However, a prior history of bleeding was significantly more frequent in patients who experienced bleeding during the study (44.44% vs 6.25%; p < 0.001; Table 1).

Figure 1.

Flow diagram of study participants.

Table 1.

Baseline demographic and clinical characteristics of patients treated with rivaroxaban.

Characteristics	Non-bleeding (n = 64)	Bleeding (n = 27)	p-Value
Age (years), mean ± SD	73.84 ± 8.92	73.56 ± 12.03	0.900^a
Gender (n, %)			0.645
Female	36 (56.25%)	17 (62.96%)
Male	28 (43.75%)	10 (37.04%)
Weight (kg), mean ± SD	65.27 ± 14.80	63.27 ± 11.09	0.532^a
HAS-BLED score, mean ± SD	2.50 ± 0.94	2.93 ± 0.92	0.050^a
Prior history of bleeding (n, %)			<0.001*
No	60 (93.75%)	15 (55.56%)
Yes	4 (6.25%)	12 (44.44%)
Creatinine clearance (mL/min), mean ± SD	57.65 ± 20.93	56.66 ± 23.62	0.843^a
Creatinine clearance (mL/min)^# (n, %)			0.235
⩾50	37 (57.81%)	16 (59.26%)
30−49	24 (37.50%)	7 (25.93%)
<30	3 (4.69%)	4 (14.81%)
Albumin (g/dL), mean ± SD	3.82 ± 0.42	3.63 ± 0.60	0.086^a
Chronic heart failure (n, %)			0.770
No	52 (81.25%)	23 (85.19%)
Yes	12 (18.75%)	4 (14.81%)
Diabetes (n, %)			0.086
No	39 (60.94%)	22 (81.49%)
Yes	25 (39.06%)	5 (18.51%)
Anemia (n, %)			1.000
No	39 (60.94%)	17 (62.96%)
Yes	25 (39.06%)	10 (37.04%)
Rivaroxaban dose (n, %)			0.084
10 mg	1 (1.56%)	3 (11.11%)
15 mg	33 (51.56%)	10 (37.04%)
20 mg	30 (46.88%)	14 (51.85%)
Rivaroxaban dosage too high (n, %)			0.669
No	58 (90.62%)	26 (96.30%)
Yes	6 (9.38%)	1 (3.70%)
Concurrent NSAIDs or anti-platelets^b (n, %)			1.000
No	55 (85.94%)	23 (85.19%)
Yes	9 (14.06%)	4 (14.81%)
Concurrent CYP and P-gp inhibitors^c (n, %)			0.579
No	62 (96.88%)	25 (92.59%)
Yes	2 (3.12%)	2 (7.41%)
Concurrent SSRI/SNRIs^d (n, %)			1.000
No	61 (95.31%)	26 (96.29%)
Yes	3 (4.69%)	1 (3.71%)

p-Value calculated using t-test; all others by Fisher’s exact test.

Nonsteroidal anti-inflammatory drugs (NSAIDs)/antiplatelet: aspirin, clopidogrel, naproxen, celecoxib, etoricoxib.

CYP and P-gp inhibitors: verapamil, amiodarone, clarithromycin.

Selective serotonin reuptake inhibitor/serotonin-norepinephrine reuptake inhibitors (SSRI/SNRIs): sertraline.

p < 0.05.

Creatinine clearance from the Cockcroft-Gault equation, was categorized into three groups based on the standard dosing criteria for rivaroxaban.

Variants with distinct allele frequencies in the Thai population

Allele frequencies in this study were compared with those of global and Asian populations.¹⁶ Overall, most SNPs showed patterns broadly consistent with EAS populations (Table S2), reflecting shared regional genetic backgrounds. However, several variants demonstrated distinct frequency distributions that may be particularly relevant in the Thai cohort (Table 2). The alternative allele frequency of ABCB1 rs10276036 (C>T) was lower in this cohort (26%) compared with global (57%), EAS (37%), and SAS (41%) populations. Similarly, ABCB1 rs10280101 (A>C) showed a reduced frequency (5%) relative to global (14%), EAS (7%), and SAS (18%) populations. In contrast, ABCG2 rs2622624 (T>C) exhibited a higher allele frequency in this study, comparable to EAS but higher than global and SAS populations. The frequency of ABCA6 rs7212506 (C>T) was also high, consistent with EAS and SAS populations but exceeding the global estimate. Notably, APOB rs693 (G>A) showed a lower frequency in this study compared with global and SAS populations, aligning more closely with EAS. In contrast, POU2F3 rs2298579 (T>C) demonstrated a high frequency, similar to EAS but markedly higher than global and SAS populations. Additionally, variants such as CYP2J2 rs11572325 (A>T) and F13A1 rs5985 (C>A) were observed at low frequencies, consistent with their rarity in Asian populations. The remaining SNPs are presented in Table S2.

Table 2.

Variants with notable differences in alternative allele frequency relative to global and regional populations.

Gene	SNPs (Ref>Alt alleles)^a	Global (%)^b	EAS (%)^b	SAS (%)^b	This study (%)
ABCB1	rs10276036 (C>T)	57	37	41	26
ABCB1	rs10280101 (A>C)	14	7	18	5
ABCG2	rs2622624 (T>C)	43	64	47	57
CYP2J2	rs11572325 (A>T)	11	8	8	4
ABCA6	rs7212506 (C>T)	75	88	93	82
APOB	rs693 (G>A)	25	6	22	9
F13A1	rs5985 (C>A)	15	0	9	2
POU2F3	rs2298579 (T>C)	18	64	25	58

Reference and alternative alleles following the NCBI dbSNP database.

Alternative allele frequency data retrieved from 1000 Genomes Project Phase III in the Ensembl database.

EAS, East Asian; SAS, South Asian.

Genetic distribution in bleeding and non-bleeding groups

Genetic variants from five PK and nine PD genes were genotyped using the MassARRAY^® system. Thirty-two variants passed the inclusion thresholds (call rate ⩾ 90% and MAF ⩾ 1%). No variation was observed for ABCB1 rs3213619, CYP2J2 rs537207272, CYP3A4*22 rs35599367, CYP3A5*6 rs10264272, and CYP3A5*7 rs76293380 (Table S1). Most SNPs met HWE (p > 0.05), except for ABCA6 rs7212506, POU2F3 rs2298579, and PRF1 rs885821, which were deviated from HWE in the overall study population. Notably, despite this deviation, the allele frequencies of these variants were comparable to those reported in EAS populations (Table S2), indicating that the observed distributions are not inconsistent with regional genetic patterns. Among the 27 polymorphic variants, ABCB1 rs2032582 (A>C/T) showed a higher frequency of the AA genotype in patients with bleeding than those without (22.22% vs 14.06%, p = 0.048). Similarly, the AA genotype of SUSD3 rs76292544 (A>T) was more frequent in the bleeding group than in the non-bleeding group (88.89% vs 75.00%, p = 0.038).

Association of genetic and clinical factors with bleeding events

Twenty-five variants from 12 genes, along with the CYP3A5 metabolizer phenotype and clinical factors known to influence rivaroxaban-related bleeding, including prior bleeding history, creatinine clearance, and albumin level,^17–19 were analyzed using univariate and multivariable logistic regression. Variables approaching the significance threshold (p ⩽ 0.10) in univariate analyses were entered into multivariable logistic regression models, conducted separately for PK and PD genes, with adjustment for significant clinical factors to estimate genetic associations. In the univariate analysis, none of the genetic polymorphisms demonstrated a significant association with bleeding. Only a prior history of bleeding remained a strong independent risk factor (OR 12.00, 95% CI 3.39–42.52, p < 0.001; Table 3).

Table 3.

Associations of genetic polymorphisms with bleeding events in rivaroxaban-treated patients.

Genetic/clinical factors	SNPs rsID	Group	Univariate analysis
Genetic/clinical factors	SNPs rsID	Group	OR (95% CI)	p-Value
Efflux transporter genes
ABCB1	rs10276036	CC	Ref
		CT	2.198 (0.852–5.674)	0.104
		TT	0.864 (0.087–8.543)	0.900
ABCB1	rs10280101	AA	Ref
		AC	1.309 (0.299–5.720)	0.720
ABCB1	rs1045642	AA	Ref
		AG	0.350 (0.090–1.364)	0.130
		GG	0.867 (0.226–3.309)	0.834
ABCB1	rs2032582	AA	Ref
		AC or AT	0.333 (0.092–1.206)	0.094
		CC or CT	1.083 (0.309–3.802)	0.901
ABCB1	rs4148738	CC	Ref	–
		CT	0.611 (0.172–2.172)	0.447
		TT	1.553 (0.428–5.640)	0.504
ABCG2	rs12505410	TT	Ref
		TG	1.222 (0.474–3.151)	0.678
ABCG2	rs2231137	CC	Ref
		CT	0.911 (0.359–2.314)	0.845
ABCG2	rs2231142	GG	Ref
		GT	1.538 (0.614–3.855)	0.358
ABCG2	rs2622624	TT	Ref
		TC	1.517 (0.418–5.497)	0.526
		CC	1.750 (0.456–6.722)	0.415
Drug-metabolizing enzyme genes
CYP2J2	rs11572325	AA	Ref
		AT	1.844 (0.383–8.865)	0.445
CYP2J2	rs3820538	GG	Ref
		GA	1.208 (0.279–5.230)	0.800
CYP3A41*	rs274057	AA	Ref
		AG	0.782 (0.077–7.873)	0.835
CYP3A51*	rs15524	AA	Ref
		AG	1.800 (0.699–4.648)	0.222
		GG	1.030 (0.181–5.863)	0.973
CYP3A53*	rs776746	TT	Ref
		TC	1.500 (0.262–8.579)	0.649
		CC	1.147 (0.205–6.426)	0.876
CYP3A59*	rs28383479	CC	Ref
		CT	2.480 (0.331–18.587)	0.377
CYP3A5 metabolizer phenotype
	1/1	NM	Ref
	1/3	IM	1.500 (0.262–8.579)	0.649
	3/3	PM	1.030 (0.180–5.880)	0.972
	3/9	Indeterminate	3.000 (0.239–37.672)	0.395
Pharmacodynamic-related genes
ABCA6	rs7212506	CC	Ref
		CT	0.667 (0.099–4.478)	0.677
		TT	1.000 (0.227–4.400)	1.000
AKR7A3	rs1738023	TT	Ref
		TC	0.123 (0.011–1.386)	0.090
		CC	0.129 (0.012–1.325)	0.085
AKR7A3	rs1738025	CC	Ref^‡
		CT	1.100 (0.422–2.870)	0.846
APOB	rs1367117	GG	Ref
		GA	1.958 (0.735–5.222)	0.179
		AA	1.468 (0.125–17.305)	0.760
APOB	rs2163204	TT	Ref
		TG	0.782 (0.078–7.873)	0.835
APOB	rs693	GG	Ref
		GA	0.822 (0.236–2.856)	0.758
POU2F3	rs2298579	TT	Ref
		TC	1.600 (0.512–4.996)	0.418
		CC	0.452 (0.139–1.470)	0.187
PRF1	rs885821	GG	Ref
		GA	0.487 (0.144–1.640)	0.246
PRKAG2	rs13224758	GG	Ref
		GA	0.770 (0.280–2.114)	0.612
SUSD3	rs76292544	AA	Ref
		AT	0.250 (0.053–1.177)	0.080
Prior history of bleeding		No	Ref
		Yes	12.000 (3.386–42.523)	<0.001*
Creatinine clearance (mL/min)		⩾ 50	Ref
		30–49	0.674 (0.242–1.882)	0.452
		< 30	3.083 (0.618–15.390)	0.170
Albumin level (g/dL)			0.443 (0.173–1.134)	0.090

‡

Because no homozygous reference allele was observed, the homozygous alternative allele was used as the reference.

p < 0.05.

In the multivariable model for PK genes, ABCB1 rs10276036 and rs2032582 were included, with adjustment for prior bleeding history and albumin level (Table 4). ABCB1 rs10276036 was the only PK variant significantly associated with bleeding, with the CT genotype showing an increased risk compared with the CC genotype (Adjusted OR 3.88, 95% CI 1.04–14.55). Using the same analytical approach, AKR7A3 rs1738023 and SUSD3 rs76292544 were included in the PD model with adjustment for the same clinical factors; however, no PD gene variants were significantly associated with bleeding (Table S3). These results describe clinical and genetic factors associated with bleeding risk in Thai patients with AF treated with rivaroxaban, with ABCB1 rs10276036 identified as a variant of potential interest for further investigation.

Table 4.

Associations of pharmacokinetic (PK) gene polymorphisms with bleeding events in rivaroxaban-treated patients.

Genetic/clinical factors	SNPs rsID	Group	Multivariable analysis
Genetic/clinical factors	SNPs rsID	Group	Adjusted OR (95% CI)^a	p-Value
ABCB1	rs10276036	CC	Ref
		CT	3.880 (1.035–14.547)	0.044*
		TT	0.117 (0.006–2.452)	0.167
ABCB1	rs2032582	AA	Ref
		AC or AT	0.224 (0.035–1.430)	0.114
		CC or CT	1.584 (0.257–9.780)	0.620
Prior history of bleeding		No	Ref
		Yes	25.421 (4.977–129.834)	<0.001*
Albumin level (g/dL)			0.537 (0.164–1.762)	0.305

Adjusted odds ratios were estimated for PK gene variants and clinical factors with a p-value ⩽ 0.10 in univariate analysis.

p < 0.05.

Discussion

This study revealed the association between ABCB1 rs10276036 (C>T) with increased bleeding risk, with the CT genotype conferring an approximately fourfold higher bleeding risk compared with the CC genotype in patients with AF receiving rivaroxaban (Table 4). To our knowledge, this is the first study to evaluate this variant in relation to rivaroxaban safety outcomes. This intronic variant has not been functionally characterized, but data from the Genotype-Tissue Expression (GTEx) Portal show lower ABCB1 mRNA expression among carriers of the T allele in testis and heart tissues.²⁰ Reduced ABCB1 expression may impair rivaroxaban efflux, leading to higher plasma concentrations and increased bleeding risk. Furthermore, this variant has been linked to olanzapine PK, cancer survival, nephrotic syndrome, and seizure outcomes.^21–24 Despite the lower T-allele frequency of rs10276036 in this Thai cohort (26%) compared with other populations (37–57%, Table S2), the association with bleeding risk remained evident. While the population differences may contribute to variability in bleeding incidence across ethnic groups,^25–27 this interpretation should be made cautiously and requires further confirmation. For transporter polymorphisms, no significant associations were found for other ABCB1 and ABCG2 variants (Table 4), consistent with previous studies. Although ABCB1 rs1045642, rs2032582, and rs4148738 have been associated with rivaroxaban PK parameters,^28,29 none showed clinical relevance to bleeding risk.^7,28–31 Similarly, ABCG2 rs2231142 was not linked to bleeding among rivaroxaban users.^28,31

For enzymatic genes involved in rivaroxaban metabolism, no significant associations were observed between bleeding risk and the common non-functional CYP3A5*3 or CYP3A5 metabolizer phenotypes defined by *1 and *3 diplotypes (Table 3), consistent with findings in Pakistani and Chinese populations.^29,31 Similarly, the decreased-function variants CYP3A5*6 and *7 were not detected in this study (Table S1), in line with their low frequencies in Asian populations¹⁴ (Table S2). Although the CYP3A5*9 is classified by CPIC as a variant with uncertain function,¹⁴ in vitro data have demonstrated reduced enzymatic activity using testosterone and nifedipine as substrates.³² However, this variant is observed at low frequency, with a reported T-allele frequency of 0% globally³³ and 2% in this cohort (Table S2). Therefore, the limited number of carriers in this study may have constrained the ability to robustly evaluate its association with bleeding risk. No significant associations were identified for variants in CYP2J2 or CYP3A4 (Table 3). Nonetheless, several variants with uncertain function, including ABCB1 rs10280101, ABCG2 rs2622624, and CYP2J2 rs11572325, demonstrated distinct allele frequencies in Thai and broader Asian populations (Table S2), suggesting potential population-specific differences in allele distribution, the clinical relevance of which remains to be determined.

We also investigated variants in pharmacodynamic genes implicated in rivaroxaban response.^9,10,28 These variants were selected based on previous studies with anti-factor Xa activity or bleeding outcomes^9,10,28 and were analyzed separately from PK variants to improve model robustness and reduce sparse data bias. Overall, no significant associations between PD variants and bleeding risk were identified, consistent with previous studies in Chinese using rivaroxaban (e.g., ABCA6, AKR7A3,²⁸ PRF1, and PRKAG2⁹ and in Korean using DOAC for APOE variants.³⁴ The rs693 (G>A) in apolipoprotein B (APOB), a key gene in lipid transport and metabolism,³⁵ has been associated with bleeding risk in DOAC-treated Korean patients, where the CC genotype was linked to an increased risk in univariate analysis.³⁴ For SUSD3 rs76292544 (A>T), a higher frequency of the AA genotype was observed in bleeding cases (Table S1). While a prior study reported an association between the T allele and a higher incidence of rivaroxaban-related hemorrhage,²⁸ our study was underpowered to confirm this relationship due to the limited number of carriers. Similarly, POU2F3 rs2298579 (T>C), previously linked to lower peak anti-factor Xa levels but not to bleeding events,⁹ was consistent with the findings of this study (Table 3). Notably, the C allele is highly prevalent in Thai and EAS populations (58–64%, Table S2), reflecting population-specific allele distribution and warranting further investigation to clarify any clinical significance. Finally, the missense variant in the Factor XIII gene, F13A1 rs5985 (C>A), which has been associated with rapid fibrin clot formation³⁶ was observed at low frequency in this cohort, consistent with its distribution in EAS populations (Table S2).

Among clinical factors previously reported to influence rivaroxaban-related bleeding,¹⁸ over 40% of patients who experienced bleeding had a prior history of bleeding (Table 1). These patients exhibited an approximately 12-fold higher risk of bleeding compared with those without such a history (Table 3), consistent with prior evidence identifying previous bleeding as an important clinical risk factor.^18,37 Although renal impairment has been associated with increased bleeding risk in patients receiving rivaroxaban and other DOACs,^17–19 no significant difference was observed between case and control groups in this study. In contrast, higher serum albumin levels showed a trend toward reduced odds of bleeding (Table 1). Given that rivaroxaban is highly protein-bound, lower albumin levels may increase the free drug fraction, potentially contributing to an elevated bleeding risk.³⁸

Concurrent use of CYP and P-gp inhibitors was considered; however, this variable was excluded from the final multivariable model for methodological and interpretational reasons. First, the number of patients exposed to CYP/P-gp inhibitors was very small (Table 1), precluding meaningful subgroup analyses to isolate the effect of individual drugs. Second, the exposed agents—predominantly verapamil (CYP1A2 and CYP3A4 and P-gp inhibitor)³⁹ and amiodarone (CYP2C9, CYP2D6, CYP3A4, and P-gp inhibitor)⁴⁰—exhibit overlapping and non-selective inhibitory profiles, making it impossible to disentangle specific drug–gene or drug–drug–gene interactions related to bleeding risk. Moreover, none of the exposed agents belonged to drug classes for which dose adjustment or avoidance is currently recommended (e.g., combined P-gp and strong CYP3A inhibitors, combined P-gp and moderate CYP3A inhibitors, or strong CYP3A inducers).⁴¹ While drug–drug interaction effects are inherently complex, our genetic findings may provide a more consistent mechanistic perspective on rivaroxaban disposition. Our results highlight the importance of the ABCB1 transporter in modulating rivaroxaban-related bleeding risk. Because the kidney is a primary route of rivaroxaban elimination, ABCB1 expression in renal tubular cells likely contributes to drug clearance and systemic exposure. This finding aligns with previous reports demonstrating that co-administration of CYP and P-gp inhibitors increases bleeding risk in patients treated with rivaroxaban and other DOACs.^17,18 The identified genetic associations should therefore be interpreted as exploratory signals that require confirmation in larger, independent cohorts.

The strength of this study lies in its integrated assessment of genetic and clinical determinants of rivaroxaban-related bleeding by exploring PK and PD variants with clinical covariates. However, the relatively small number of bleeding cases limits the power to detect weaker associations. Larger, multi-center studies are needed to validate these findings and assess their clinical utility in guiding personalized anticoagulant therapy across diverse populations.

Limitation

This study has several important limitations that warrant careful interpretation of the findings. First, the planned number of outcome events was not fully reached. There are some odds ratio estimates that were derived from reference genotype groups with very small sample sizes, which may result in unstable or inflated effect estimates. These associations should therefore be considered exploratory and hypothesis-generating, consistent with the discovery-phase nature of this study and highlighting the need for further validation. Second, as variants with a genotype call rate of ⩾ 90% were included, the number of patients with available genotype results in Table S1 may be slightly less than the total study population for some variants. Future studies should include larger sample sizes to adequately capture variants with low minor allele frequencies and incorporate clinically relevant drug–drug interaction exposures, such as combined CYP3A4/5 and P-glycoprotein inhibitors, to better characterize their impact on bleeding risk. Finally, the modest sample size limited our ability to robustly evaluate rare variants and gene–drug or drug–drug interactions. Future studies with larger cohorts should incorporate clinically relevant covariates, including exposure to combined CYP3A4/5 and P-glycoprotein inhibitors, to better characterize their contribution to rivaroxaban-related bleeding risk. Given the number of genetic variants evaluated, the possibility of false-positive findings due to multiple testing cannot be excluded and should be considered when interpreting the results.

Conclusion

This study identified an association between ABCB1 rs10276036 and rivaroxaban-related bleeding in Thai patients with AF, alongside clinical factors, including prior bleeding. Allele frequency analysis in this study also highlighted several variants, including ABCB1 rs10276036, as population-specific genetic features of interest. Given the limited sample size, these findings require validation in larger, multi-center cohorts, as well as functional studies to clarify the role of ABCB1 rs10276036 in transporter activity. Overall, the results support a cumulative risk framework, in which genetic variation contributes to bleeding risk in conjunction with established clinical factors, rather than acting independently.

Supplemental Material

sj-docx-1-taw-10.1177_20420986261452545 – Supplemental material for Pharmacogenetic variants in pharmacokinetic and pharmacodynamic pathways and clinical factors associated with bleeding risk in Thai patients receiving rivaroxaban: a case–control study

Supplemental material, sj-docx-1-taw-10.1177_20420986261452545 for Pharmacogenetic variants in pharmacokinetic and pharmacodynamic pathways and clinical factors associated with bleeding risk in Thai patients receiving rivaroxaban: a case–control study by Varalee Yodsurang, Krit Wattanathum, Thanate Srimatimanon, Krittin Pitinanon, Andro Ranti, Pattaraporn Siripattarachai, Piyapha Hirunpatrawong, Voravut Rungpradubvong, Seiichi Sakamoto and Alisara Sangviroon Sujarit in Therapeutic Advances in Drug Safety

Footnotes

Acknowledgements

We would like to express our gratitude to all the patients who participated in this study. We also deeply appreciated the physicians and nurses at King Chulalongkorn Memorial Hospital for their excellent cooperation and support throughout the study. We are grateful to The Pharmaceutical Research Instrument Center of Faculty of Pharmaceutical Sciences at Chulalongkorn University for providing the research facilities.

Declarations

ORCID iDs

Varalee Yodsurang

Krit Wattanathum

Thanate Srimatimanon

Krittin Pitinanon

Andro Ranti

Pattaraporn Siripattarachai

Piyapha Hirunpatrawong

Voravut Rungpradubvong

Seiichi Sakamoto

Alisara Sangviroon Sujarit

Supplemental material

Supplemental material for this article is available online.

References

Boonpattharatthiti

Vichitkajee

Dhippayom

, et al. Efficacy and safety of new direct oral anticoagulants for the treatment of venous thromboembolism. Thai J Clin Pharm 2021; 31: 196–207.

Subcommittee on Guideline for Management of Bleeding in Patients Taking Direct Oral Anticoagulants, The Thai Society of Hematology. Guideline for Management of Bleeding in Patients Taking Direct Oral Anticoagulants [Internet]. Bangkok: The Thai Society of Hematology https://tsh.or.th (2022, accessed 1 October 2025).

Akhtar

Fratti

JDC

Mattumpuram

, et al. Factors associated with bleeding events in patients on rivaroxaban for non-valvular atrial fibrillation: a real-world experience. Int J Cardiol 2020; 320: 78–82.

Shnayder

Petrova

Shesternya

, et al. Using pharmacogenetics of direct oral anticoagulants to predict changes in their pharmacokinetics and the risk of adverse drug reactions. Biomedicines 2021; 9: 451.

Zhao

Chen

Wang

, et al. Identifying the dominant contribution of human cytochrome P450 2J2 to the metabolism of rivaroxaban, an oral anticoagulant. Cardiovasc Drugs Ther 2022; 36: 121–129.

, et al. The impact of ABCB1, CYP3A4/5 and ABCG2 gene polymorphisms on rivaroxaban trough concentrations and bleeding events in patients with non-valvular atrial fibrillation. Hum Genomics 2023; 17: 59.

Lähteenmäki

Vuorinen

A-L

Pajula

, et al. Pharmacogenetics of bleeding and thromboembolic events in direct oral anticoagulant users. Clin Pharmacol Ther 2021; 110: 768–776.

Šimicˇević

Slišković

Kirhmajer

, et al. Risk factors for rivaroxaban-related bleeding events-possible role of pharmacogenetics: case series. Pharmacy (Basel) 2023; 11: 20230205.

Xiang

Wang

, et al. Genetic variants influenced the risk of bleeding and pharmacodynamics of rivaroxaban in patients with nonvalvular atrial fibrillation: a multicentre prospective cohort study. Clin Transl Med 2023; 13: e1263.

10.

Zhao

Zhang

Wang

, et al. Non-synonymous alterations in AKR7A3 and ABCA6 correlate with bleeding in aged patients treated with rivaroxaban. Clin Transl Sci 2022; 15: 923–929.

11.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: guidelines for reporting observational studies. PLoS Med 2007; 4: e296.

12.

Schulman

Kearon

Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. J Thromb Haemost 2005; 3: 692–694.

13.

Kaatz

Ahmad

Spyropoulos

, et al. Definition of clinically relevant non-major bleeding in studies of anticoagulants in atrial fibrillation and venous thromboembolic disease in non-surgical patients: communication from the SSC of the ISTH. J Thromb Haemost 2015; 13: 2119–2126.

14.

Birdwell

Decker

Barbarino

, et al. Clinical pharmacogenetics implementation consortium (CPIC) guidelines for CYP3A5 genotype and tacrolimus dosing. Clin Pharmacol Ther 2015; 98: 19–24.

15.

Wigginton

Cutler

Abecasis

GR.

A note on exact tests of Hardy-Weinberg equilibrium. Am J Hum Genet 2005; 76: 887–893.

16.

Dyer

Austine-Orimoloye

Azov

, et al. Ensembl 2025. Nucleic Acids Res 2025; 53: D948–d957.

17.

Sholzberg

Pavenski

Shehata

, et al. Bleeding complications from the direct oral anticoagulants. BMC Hematol 2015; 15: 1–5.

18.

Siripattarachai

Sujarit

Rungpradubvong

, et al. Factors affecting bleeding in patients with atrial fibrillation receiving rivaroxaban: a multicenter matched case-control study Thai J Pharm Pract 2023; 15: 728–740.

19.

Jackevicius

Ghaznavi

, et al. Bleeding risk of direct oral anticoagulants in patients with heart failure and atrial fibrillation. Circ Cardiovasc Qual Outcomes 2021; 14: 155–168.

20.

GTEx Portal. rs10276036 [Internet], https://www.gtexportal.org/home/snp/rs10276036 (2025, accessed 1 October 2025).

21.

Aly Hassan

Mohamed Elsaid

Mahmoud El-Refaey

, et al. Association of ABCB1(Rs10276036, C/T) gene, IL-18, and TNFα as risk factors for nephrotic syndrome incidence. Rep Biochem Mol Biol 2024; 13: 67–78.

22.

Caronia

Patiño-Garcia

Peréz-Martínez

, et al. Effect of ABCB1 and ABCC3 polymorphisms on osteosarcoma survival after chemotherapy: a pharmacogenetic study. PLoS One 2011; 6: e26091.

23.

Louis

Busch

Lal

, et al. Genetic and molecular features of seizure-freedom following surgical resections for focal epilepsy: a pilot study. Front Neurol 2022; 13: 942643.

24.

Zubiaur

Soria-Chacartegui

Koller

, et al. Impact of polymorphisms in transporter and metabolizing enzyme genes on olanzapine pharmacokinetics and safety in healthy volunteers. Biomed Pharmacother 2021; 133: 111087.

25.

Mitsuntisuk

Nathisuwan

Junpanichjaroen

, et al. Real-world comparative effectiveness and safety of non-vitamin K antagonist oral anticoagulants vs. warfarin in a developing Country. Clin Pharmacol Ther 2021; 109: 1282–1292.

26.

Evans

Davies

Osborne

, et al. Evaluation of the incidence of bleeding in patients prescribed rivaroxaban for the treatment and prevention of deep vein thrombosis and pulmonary embolism in UK secondary care: an observational cohort study. BMJ Open 2020; 10: e038102.

27.

Roupe

Rosengren

Pivodic

, et al. Bleeding risk on rivaroxaban versus apixaban in older patients with venous thromboembolism: a nationwide observational register-based study. Eur J Int Med 2025; 142: 106438.

28.

Guan

, et al. Association of genetic variants with rivaroxaban pharmacokinetics and pharmacodynamics and hemorrhage risk factors in patients with venous thromboembolism. J Am Heart Assoc 2025; 14: e040698.

29.

Ain

Ali

Ullah

, et al. Bleeding events associated with rivaroxaban therapy in naive patients with nonvalvular atrial fibrillation: a longitudinal study from a genetic perspective with INR follow-up. Medicina (Kaunas) 2024; 60: 20241018.

30.

Mardi

Abbasi

Shafiee

, et al. Pharmacogenetic approach for the prevention of rivaroxaban’s ADRs: a systematic review and meta-analysis. Genet Res (Camb) 2023; 2023: 6105320.

31.

Wang

Zhou

, et al. Relevance of selected pharmacogenetic polymorphisms to bleeding and thromboembolic risks in Chinese patients taking direct-acting oral anticoagulants. Br J Clin Pharmacol 2025; 91: 2736–2744.

32.

Lee

S-J

Usmani

Chanas

, et al. Genetic findings and functional studies of human CYP3A5 single nucleotide polymorphisms in different ethnic groups*. Pharmacogenet Genom 2003; 13: 461–472.

33.

Sherry

Ward

Kholodov

, et al. dbSNP: the NCBI database of genetic variation. Nucleic Acids Res 2001; 29: 308–311.

34.

Yoon

H-Y

Song

T-J

Yee

, et al. Association between genetic polymorphisms and bleeding in patients on direct oral anticoagulants. Pharmaceutics 2022; 14: 1889.

35.

Basavaraju

Balasubramani

Kathiresan

, et al. Genetic regulatory networks of apolipoproteins and associated medical risks. Front Cardiovasc Med 2021; 8: 788852.

36.

Kreutz

Bitar

Owens

, et al. Factor XIII Val34Leu polymorphism and recurrent myocardial infarction in patients with coronary artery disease. J Thromb Thrombolysis 2014; 38: 380–387.

37.

Kanaan

Malloy

Knowles

Evaluation of patient characteristics linked to major bleeding events in patients prescribed direct oral anticoagulants. Clin Appl Thromb Hemost 2023; 29: 1–7.

38.

Wojakowski

Cheruvil

Hassan

, et al. Albumin and bleed risk in rivaroxaban treated patients. J Thromb Thrombolysis 2020; 50: 1004–1011.

39.

Verapamil. Lexi-Drugs, https://online.lexi.com (accessed 16 January 2026).

40.

Amiodarone. Lexi-Drugs, https://online.lexi.com (accessed 16 January 2026).

41.

Janssen Pharmaceuticals I. XARELTO (rivaroxaban): US prescribing information. 2021. Silver Spring, MD: U.S. Food and Drug Administration.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.05 MB

0.00 MB