Accepted Article

Article type : Article

Pharmacogenetics of bleeding and thromboembolic events in direct

oral anticoagulant users

Jaakko Lähteenmäki, VTT Technical Research Centre of Finland

Anna-Leena Vuorinen, VTT Technical Research Centre of Finland
Juha Pajula, VTT Technical Research Centre of Finland
Kari Harno, Department of Health and Social Management, University of Eastern Finland, Kuopio
Mika Lehto, Helsinki University Hospital, Heart and Lung Center; and University of Helsinki
Mikko Niemi, Department of Clinical Pharmacology and Individualized Drug Therapy Research Program,
University of Helsinki and Helsinki University Hospital
Mark van Gils, VTT Technical Research Centre of Finland; Tampere University

Correspondence: Jaakko Lähteenmäki, VTT Technical Research Centre of Finland, PO Box 1000, 02044
VTT, Finland. Tel: +358405149869, E-mail: jaakko.lahteenmaki@vtt.fi
Conflict of Interest statement:
Mika Lehto: Consultant: BMS-Pfizer-alliance, Bayer, Boehringer-Ingelheim, and MSD; Speaker: BMS-Pfizer-
alliance, Bayer, Boehringer-Ingelheim, MSD, Terve Media and Orion Pharma.

All other authors declared no competing interests for this work.

Funding
This study was funded by Business Finland, VTT Technical Research Centre of Finland, Ltd., Karl Fazer AB,
Novartis Finland Oy, Pfizer Oy, Roche Diagnostics Oy, Avaintec Oy, Crown CRO Oy, Mediconsult Oy and
Biobank Cooperative Finland.

This article has been accepted for publication and undergone full peer review but has not been
through the copyediting, typesetting, pagination and proofreading process, which may lead to
differences between this version and the Version of Record. Please cite this article as doi:
10.1002/CPT.2316
This article is protected by copyright. All rights reserved
 Keywords: Adverse Drug Reactions, Thrombosis, Cardiovascular Disease, Pharmacogenetics, Precision
Accepted Article
medicine

Abstract

This study aimed to analyse associations between genetic variants and the occurrence of clinical
outcomes in dabigatran, apixaban, and rivaroxaban users. This was a retrospective real-world study
linking genotype data of three Finnish biobanks with national register data on drug dispensations and
healthcare encounters. We investigated several single nucleotide variants (SNVs) in the ABCG2, ABCB1,
CES1 and CYP3A5 genes potentially associated with bleeding or thromboembolic events in direct oral
anticoagulant (DOAC) users based on earlier research. We used Cox regression models to compare the
incidence of clinical outcomes between carriers and non-carriers of the SNVs or haplotypes. In total 1,806
patients on apixaban, dabigatran or rivaroxaban were studied. The ABCB1 c.3435C>T (p.Ile1145=,
rs1045642) SNV (HR 0.42, 95% CI 0.18-0.98, p=0.044) and 1236T-2677T-3435T (rs1128503-rs2032582-
rs1045642) haplotype (HR 0.44, 95% CI 0.20-0.95, p=0.036) were associated with a reduced risk for
thromboembolic outcomes, and the 1236C-2677G-3435C (HR 2.55, 95% CI 1.03-6.36, p=0.044) and 1236T-
2677G-3435C (HR 5.88, 95% CI 2.35-14.72, p<0.001) haplotypes with an increased risk for
thromboembolic outcomes in rivaroxaban users. The ABCB1 c.2482-2236G>A (rs4148738) SNV associated
with a lower risk for bleeding events (HR 0.37, 95% CI 0.16-0.89, p=0.025) in apixaban users. ABCB1
variants are potential factors affecting thromboembolic events in rivaroxaban users and bleeding events
in apixaban users. Studies with larger numbers of patients are warranted for comprehensive assessment
of the pharmacogenetic associations of DOACs and their relevance for clinical practice.

Introduction
Direct Oral Anticoagulants (DOACs) are increasingly used in the context of various clinical conditions and
procedures, including atrial fibrillation (ischemic stroke prevention), deep vein thrombosis, and
pulmonary embolism. DOACs have favourable pharmacodynamic and pharmacokinetic properties and do

This article is protected by copyright. All rights reserved

 not require continuous therapeutic monitoring except in specific cases.1–3 However, serious bleeding
Accepted Article
events and thromboembolic events of DOAC users are not uncommon.4 Notable interindividual variation
exists in the plasma levels of DOACs.5,6

The effect of genetic factors on the pharmacokinetics of DOACs have been investigated in several
studies.7–9 Indications of association between genotypes and drug plasma levels have been found in some
studies10–17, but also contradictory findings have been reported.5,18,19

In particular, the genome-wide association substudy of the RE-LY trial13,20, shows associations of ABCB1
and CES1 variants with dabigatran plasma concentration and an association of one CES1 variant with risk
of bleeding. The role of genetic variants in excessive drug exposure and resulting bleeding episodes has
been proposed also in some case reports.21,22

The aim of this retrospective real-world study was to provide more insight on the effect of genetic factors
on endpoints associated with the use of DOACs. We have linked genotype data of three Finnish biobanks
with national register data on drug dispensations and healthcare encounters to explore potential
associations of genetic variants with responses to dabigatran, rivaroxaban and apixaban, which are the
three most used DOACs in Finland.23

Methods

Study design
Pharmacogenomics of antithrombotic drugs (PreMed study) is a retrospective cohort study linking health
data from multiple sources. The data for the study consists of three subcohorts provided by Auria
Biobank, Helsinki Biobank and THL Biobank. Each biobank first identified subjects fulfilling the inclusion
criteria of the study. Then, the biobanks formed subcohorts by linking their genomic and demographic
data with healthcare encounter data from the Finnish Institute of Health and Welfare, with drug
dispensation data from the Social Insurance Institution of Finland and with patient record and laboratory
data from Finnish hospital districts and municipalities. The three pseudonymized subcohorts provided by
the biobanks were then merged by VTT Technical Research Centre of Finland to form the PreMed study
cohort24. In this article we investigate the genetics-response associations of DOACs. Results on warfarin
pharmacogenetics have been reported earlier.25

This article is protected by copyright. All rights reserved

 Patients
Accepted Article
Individuals of at least 18 years of age at the time of the first purchase of a DOAC, with one of the inclusion
diagnoses (Table S1) and with at least one drug dispensation as a new user of dabigatran, rivaroxaban or
apixaban in the time frame of January 2007 – June 2018 were included in the analysis. Additionally, it was
required that the patient’s genetic variant data were available from the biobanks. We analysed genetic
variants which have been identified in earlier research to be potentially associated with bleeding or
thromboembolic events in DOAC users. 7,8

Subjects with purchases of antithrombotic drugs during the period January 1st 2005 - December 31st 2006
were excluded to ensure a wash-out period of at least two years for all patients.

Genotyping
Genotyping of patients was carried out in the framework of the on-going FinnGen project26 and past
research projects of the Finnish Institute of Health and Welfare.27,28 The following SNVs were included:
ABCG2 c.421C>A (p.Gln141Lys, rs2231142), ABCB1 c.3435C>T (p.Ile1145=, rs1045642), ABCB1 c.2677G>T
(p.Ala893Ser, rs2032582), ABCB1 c.2677G>A (p.Ala893Thr, rs2032582), ABCB1 c.2482-2236G>A
(rs4148738, intron 18), ABCB1 c.1236C>T (p.Gly412=, rs1128503), CYP3A5 g.6981A>G (splice defect,
CYP3A5*3, rs776746), CES1 c.1168-33C>T (rs2244613, intron 10) and CES1 c.257+885T>C (rs8192935,
intron 2).

We verified the quality of the genotype data by comparing the allele frequencies of the cohort with other
genotyping projects.29 In addition, we verified that all SNVs were in Hardy-Weinberg equilibrium and
confirmed that the imputation quality was sufficient (info score > 0.85).30

Drug exposure

We estimated the drug exposure based on the data on drug dispensations including package size, product
strength and the date of dispensation. The exposure was considered to have started on the date of the
first drug dispensation and was deemed to be continuous until the earliest of the following: 1) all
dispensed packages of the drug were consumed and a new package was not dispensed within 30 days, 2)
another anticoagulant (Table S2) was purchased, or 3) the study follow-up ended (December 31st 2018). If
the subject had used more than one DOAC during the study period, we chose the first DOAC used for

This article is protected by copyright. All rights reserved

 investigation to ensure that all patients were naïve to DOAC treatment at baseline. Subsequent periods of
Accepted Article
using other DOACs were excluded from the analysis.

Outcome events
Bleeding and thromboembolic events were observed as outcomes. We considered an outpatient or
inpatient healthcare episode to be eligible as an outcome event if it occurred for the first time during the
drug exposure period and if one of the defined outcome diagnosis codes listed in Table S1 was
documented for the visit. Specific consideration was applied to recurrent thromboembolic events,
referring to thromboembolic events which had occurred before the drug use started and occurred again
during the drug exposure and for which the same diagnosis code had been documented. We considered a
recurrent thromboembolic event as an outcome only if it involved a hospitalization and if it was
documented as the primary diagnosis at the admission. These conditions were deemed necessary because
a diagnosis code of a past event is often documented also for a subsequent outpatient visit even when a
new outcome event has not occurred.

Covariates
Participant’s sex and age were available in the demographic data provided by the biobanks. Drug
indication was based on the inclusion diagnosis documented for the patient closest before the initiation of
the drug use. We categorised the patient to have cancer in case any cancer diagnosis (ICD-10 code group
C) was documented for an outpatient or inpatient visit during the exposure time or up to two years before
drug exposure started. Renal function was evaluated based on the glomerular filtration rate (GFR)
computed from the laboratory data (creatinine test). We considered the renal function to be reduced if
the GFR value was below 60 ml/min/1.73 m2.31

The patient was defined to be on dose reduction if a reduced drug dose was prescribed in 50% or more of
the dispensations. The daily doses were for dabigatran 150 mg bid (nominal) and 110 mg bid (reduced),
for rivaroxaban 20 mg qd (nominal) and 2.5/10/15 mg qd (reduced) and for apixaban 5 mg bid (nominal)
and 2.5 mg bid (reduced).

We defined that the patient was a user of an interacting drug if the patient purchased a potentially
interacting drug listed in Table S2 at least once during the DOAC exposure time. The list of potentially
interacting drugs was based on the Inxbase drug-drug interaction database.32 We included 46 drugs
potentially increasing the bleeding risk and three drugs potentially increasing the risk for thromboembolic

This article is protected by copyright. All rights reserved

 events. The drugs potentially increasing the bleeding risk were further divided into antiplatelet and statin
Accepted Article
groups for the analysis of confounders (Table S2). Dispensations of acetylsalicylic acid, as an over-the-
counter drug, could not be extracted from the drug purchase register, which only includes prescribed
medication. Instead, we extracted information of acetylsalicylic acid use from the patient records (data
entered by the physician based on information given by the patient).33

Statistical analyses
We analysed associations between carriers and non-carriers of SNVs or haplotypes and outcome events
for each drug user group (dabigatran, apixaban, rivaroxaban) separately.

For the SNV association analysis, we primarily used the dominant genetic model34 comparing the non-
carriers (no variant alleles) with carriers (one or two variant alleles). We repeated the analysis by using
the recessive genetic model comparing homozygous subjects having two variant alleles with subjects
having only one variant allele or no variant alleles. We also carried out an association analysis for the
ABCB1 haplotypes of c.1236C>T, c.2677G>T and c.3435C>T. The haplotypes were estimated from the
genotype data by using the PHASE (v2.1.1) software package.35 For the haplotype analysis, the dominant
and recessive models were used to compare non-carriers with carriers analogously to the SNV analysis.

The occurrence of outcome events was reported as incidence rates (events per 100 patient years). We
used the Cox proportional hazards regression models to investigate the association of the SNV and
haplotype groups with the occurrence of outcome events. The results were reported using hazard ratios
(HRs), 95% confidence intervals (CI) and p-values. We investigated potential confounders using univariate
Cox regression analysis. Based on the analysis we used antiplatelet medication in the adjusted analysis of
bleeding events, and we used atrial fibrillation as drug indication together with sex in the adjusted
analysis of thromboembolic events. We used the Schoenfeld residuals to test the proportional hazards
assumption.36 In all analyses the follow-up time was restricted to 730 days.

We used the Chi-square test to check if the allele frequencies of variants were in Hardy–Weinberg
equilibrium. We conducted all analyses with R Studio version 1.1.456 using R version 3.5.1 (2018-07-02).
Conclusions of statistical significance were based on two-sided tests with a significance level of 0.05.

This article is protected by copyright. All rights reserved

 Ethical aspects
Accepted Article
This study was conducted in accordance with the Declaration of Helsinki. The study was approved by the
ethics committee of the Hospital District of Helsinki and Uusimaa (HUS/513/2019) and carried out under
contracts with three biobanks Auria Biobank (study number: AB19-9833), Helsinki Biobank (study number:
HBP20190038) and THL Biobank (study number: BB2019_6). After data linkage carried out by the
biobanks, all data were pseudonymized. The informed consents of patients were obtained through
biobanks as defined in the Finnish Biobank Act 2013 (688/2012).37

Results
The full PreMed cohort consists of 7,005 patients. Of those, 25.8% (n=1,806) fulfilled the eligibility criteria
for the current study by having made at least one DOAC purchase (Figure 1). The number of study
patients was highest with rivaroxaban, and atrial fibrillation was the most prevalent indication for an oral
anticoagulant treatment. Table 1 shows key characteristics of the study patients. More complete baseline
data stratified by genetic variants are given in Table S3.

A summary of the outcome events is presented in Table S4. Hematuria (R31) and cerebral infarction
(I63.1, I63.3, I63.4 and I63.9) were the most common bleeding and thromboembolic outcome events
respectively. The median follow-up times were 217, 128 and 290 days for the occurrence of bleeding
events of dabigatran, rivaroxaban and apixaban users, respectively. Correspondingly, the median follow-
up times for thromboembolic events were 210, 128 and 298 days. During the follow-up time the incidence
of bleeding events in patients treated with dabigatran, rivaroxaban and apixaban was 4.4 (95% CI 2.6-7.0),
5.9 (95% CI 4.4-7.7) and 4.7 (95% CI 3.1-6.7) per 100 patient years respectively. Correspondingly, the
incidence of thromboembolic events for the three drugs was 2.4 (95% CI 1.1-4.4), 3.9 (95% CI 2.7-5.3) and
3.3 (95% CI 2.0-5.0) per 100 patient years.

The results of comparing non-carriers with carriers of the SNVs are shown in Table 2 and Figure 2. The
ABCB1 c.3435C>T SNV was associated with a reduced risk for thromboembolic events in rivaroxaban users
(HR 0.42, 95% CI 0.18-0.98, p=0.044). A similar trend was seen for the ABCB1 c.2677G>T SNV (HR 0.50,
95% CI 0.23-1.08, p=0.077). The ABCB1 c.2482-2236G>A SNV showed an association with a reduced risk
for bleeding events (HR 0.37, 95% CI 0.16-0.89, p=0.025) in apixaban treated patients.

Results of haplotype-based analysis are shown in Table 3 and Figure 3. The 1236T-2677T-3435T haplotype
was associated with a reduced risk of thromboembolic events while the 1236C-2677G-3435C and 1236T-
2677G-3435C haplotypes showed associations with increased risk of thromboembolic events in

This article is protected by copyright. All rights reserved

 rivaroxaban users. The hazard ratios were 0.44 (95% CI 0.20-0.95, p=0.036), 2.55 (95% CI 1.03-6.36,
Accepted Article
p=0.044) and 5.88 (95% CI 2.35-14.72, p<0.001) respectively for the TTT, CGC and TGC haplotypes.
Statistically significant associations were not found for the investigated haplotypes in dabigatran or
apixaban users.

When repeating the analyses using the recessive genetic model the association of ABCB1 c.3435C>T SNV
with thromboembolic events was still found (HR 0.34, 95% CI 0.12-1.00, p=0.05). The results for ABCB1
c.2677G>T and c.2482-2236G>A SNVs as well as for 1236T-2677T-3435T and 1236C-2677G-3435C
haplotypes showed consistent trend but were not statistically significant when analysed using the
recessive genetic model. The 1236T-2677G-3435C haplotype could not be analysed using the recessive
model due to the low number of homozygote carriers.

Results on the univariate analyses for potential confounders are listed in Table S5. Concomitant use of
acetylsalicylic acid with a DOAC was associated with increased risk for bleeding when all three DOAC drugs
were analysed together (HR 2.27, 95% CI 1.25-4.14, p=0.007). The results show a similar trend for
antiplatelets in general (HR 1.65, 95% CI 0.92-2.96, p=0.091). The analysis covering all potentially
interacting drugs did not show statistically significant association with bleeding events. We also did not
find association of concomitant use of statins with bleeding events.

We were not able to analyse the effect of interacting medication on the risk for thromboembolic events
due to low number of users of the potentially interacting drugs. Instead, we found that female sex was
associated with decreased risk of thromboembolic events when all three drugs were analysed together.
Furthermore, patients on a DOAC and with atrial fibrillation as the drug indication were found to be in a
lower risk for thromboembolic outcomes as compared to patients with other indications (HR 0.53, 95% CI
0.29-0.96, p=0.035).

The other investigated covariates – age, cancer, renal function, and reduced dose - did not show
statistically significant association with either of the outcome events. A considerable risk of
thromboembolic events in dabigatran users with reduced dose was seen, however, the result should be
interpreted with caution due to the small total number of thromboembolic outcome events (seven
events) in the dabigatran user group. For the analysis on kidney function, 73% of patients (those with
creatinine laboratory test available) were included.

Based on the results of the univariate analyses on potential confounders, we selected the use of
antiplatelet drugs to be adjusted in the analyses on bleeding events and atrial fibrillation as drug
indication and sex to be adjusted in the analyses on thromboembolic events. The results of adjusted SNV

This article is protected by copyright. All rights reserved

 and haplotype analysis are presented in Tables S6 and S7. The adjusted results were comparable with the
Accepted Article
crude estimates presented in Tables 2 and 3. Reduced risk for thromboembolic events in rivaroxaban
users was also seen in the adjusted analyses for ABCB1 c.3435C>T (HR 0.39, 95% CI 0.17-0.91, p=0.028),
and a similar trend for c.2677G>T (HR 0.48, 95% CI 0.22-1.04, p=0.062) SNVs as well as for 1236T-2677T-
3435T (HR 0.42, 95% CI 0.20-0.92, p=0.029), 1236C-2677G-3435C (HR 2.68, 95% CI 1.07-6.68, p=0.035)
and 1236T-2677G-3435C (HR 5.80, 95% CI 2.31-14.58, p<0.001) haplotypes.

Discussion
Association between genetic variants and clinical outcomes of DOAC users have earlier been investigated
only in a few studies carried out in prospective trial settings. For our study, we adopted the real-world
data approach which enabled us to use a longer follow-up time than has been the case in the earlier
studies.38,39

We found the ABCB1 c.3435C>T SNV and 1236T-2677T-3435T haplotype to be associated with a lower risk
and 1236C-2677G-3435C and 1236T-2677G-3435C haplotypes to be associated with a higher risk for
thromboembolic outcomes in rivaroxaban users. The ABCB1 c.3435C>T SNV has previously been identified
as potentially affecting rivaroxaban and dabigatran pharmacokinetics by modulating the P-glycoprotein
activity and thereby leading to increased plasma drug levels.12,22,40 Our results on the thromboembolic
outcomes are in line with this anticipated effect although we could not see a corresponding association
with bleeding risk.
We also found an association of the ABCB1 c.2482-2236G>A SNV with a decreased bleeding risk in
apixaban users. This variant has been found to be associated with bleeding in one earlier study10 while
other studies have not been able to confirm the association.19,34

So far, the most extensive investigation of the pharmacogenomics of DOACs has been the genome-wide
association study of Paré et al. focusing on dabigatran users.13 The study found association of the CES1
rs2244613 SNV with a decreased risk of bleeding. We could not confirm this association with bleeding in
dabigatran patients.

Among the other variants investigated in our study ABCG2 c.421C>A and CES1 c.257+885T>C suggested
possible association with bleeding events in apixaban and dabigatran users, respectively, although
statistical significance was not reached. Our results on the ABCG2 c.421C>A SNV are in line with the
results of earlier prospective studies on plasma concentration and drug clearance 11,15,41 as well as with a

This article is protected by copyright. All rights reserved

 case report.42 Our results on CES1 c.257+885T>C SNV are in line with studies reporting association of the
Accepted Article
variant with drug plasma concentration.13,16,43

The present study revealed once again the important role of antiplatelets as interacting drugs.
Antiplatelets are in many cases used in parallel with a DOAC to increase protection towards
thromboembolic events. As shown in this study and others, concomitant use of antiplatelets and a DOAC
increases the risk of bleeding.44–46 One strength of the present study is that we were able to access
information on acetylsalicylic acid usage via the data obtained from patient records.33 Sex was the most
significant confounder for thromboembolic events, which is in line with current knowledge. Sex
differences have been reported in platelet function and coagulation factor activities as well as in the rate
of cardiovascular events.47

The main limitation of the study is the relatively small number of participants. As DOACs are relatively
new drugs, the accumulated clinical data of the drug users are still limited and, furthermore, only
individuals with already existing genotype data in the biobanks could be included in the study. The small
number of subjects guided us to study patients as a single group regardless of the indication for use of
DOAC. The analysis of confounders revealed that the risk of atrial fibrillation patients for thromboembolic
events was lower, and therefore we controlled atrial fibrillation as the drug indication in the adjusted
results. The number of patients enabled us to detect only major differences in the clinical outcomes of
the genotype or haplotype groups. Furthermore, due to the limited sample size we may have missed to
find the effect of some of the potential confounders. For example, the results reflected the anticipated
effect of the renal function48, but statistical significance was not achieved. Furthermore, we could not
apply stratified analysis by dividing the bleeding events into minor and major bleedings as has been done
in larger studies with warfarin users.49

Another potential limitation is the risk of inaccuracy due to missing data. In a real-world setting the
completeness of data cannot be fully guaranteed, although the Finnish registries have proven to be of
high quality, especially in the registration of cardiovascular diagnoses.50

Conclusions
In our real-world study, we investigated ABCB1, CES1 and ABCG2 variants potentially associated with the
risk of bleeding or thromboembolic events in DOAC users. We found ABCB1 c.3435C>T SNV and 1236T-
2677T-3435T haplotype to be associated with a lower risk and the 1236C-2677G-3435C and 1236T-2677G-
3435C haplotypes to be associated with a higher risk for thromboembolic outcomes in rivaroxaban users.

This article is protected by copyright. All rights reserved

 We also found an association of the ABCB1 c.2482-2236G>A SNV with a decreased bleeding risk in
Accepted Article
apixaban users. This is one of the first real-world studies evaluating potential associations of genetic
variants with clinical endpoints in patients on DOACs. Further studies with larger sample sizes are
warranted to verify the findings.

This article is protected by copyright. All rights reserved

Accepted Article
Study Highlights
What is the current knowledge on the topic?
Earlier studies have found associations between genetic variability and plasma levels of direct oral
anticoagulants (DOAC), but it is unclear if these associations translate into clinical outcomes.

What question did this study address?

Are variants in the ABCG2, ABCB1, CES1 or CYP3A5 genes associated with bleeding or thromboembolic
events in DOAC users?

What does this study add to our knowledge?

We found ABCB1 single nucleotide variants and haplotypes to be associated with clinical outcomes in
rivaroxaban and apixaban users.

How might this change clinical pharmacology or translational science?

This knowledge might aid in identifying individuals with suboptimal response to rivaroxaban or apixaban.
Further research is warranted on the clinical impact and cost-effects of pharmacogenetic variability in
DOAC response.

This article is protected by copyright. All rights reserved

Accepted Article
Acknowledgements
The data used for the research were obtained from Auria Biobank (study number: AB19-9833), Helsinki
Biobank (study number: HBP20190038) and THL Biobank (study number: BB2019_6). We thank all study
participants for their generous participation in biobank research. We also thank Merja Perälä, Perttu
Terho and Mikko Tukiainen from Auria Biobank, Theresa Knopp, Miika Koskinen and Otto Manninen from
Helsinki Biobank and Niina Eklund, Anni Joensuu and Katariina Peltonen from THL Biobank for their
contribution in defining the subcohorts for the study and Francesco De Pretis from VTT for valuable
comments to the manuscript.

Author contributions
J.L. and A.V. wrote the manuscript; J.L., A.V., J.P., K.H., M.L., M.N. and M.v.G. designed the research; J.L.,
A.V. and J.P. performed the research; J.L., A.V. and J.P. analyzed the data.

References

1. Patel, J. P., Byrne, R. A., Patel, R. K. & Arya, R. Progress in the monitoring of direct oral
anticoagulant therapy. Br. J. Haematol. 184, 912–924 (2019).
https://doi.org/10.1111/bjh.15756
2. Sikorska, J. & Uprichard, J. Direct oral anticoagulants: A quick guide. Eur. Cardiol. Rev.
12, 40–45 (2017). https://doi.org/10.15420/ecr.2017:11:2
3. Lentz, S. R. Genetic testing to guide warfarin dosing: Impact of direct oral anticoagulants.
Clin. Pharmacol. Ther. 100, 128–130 (2016). https://doi.org/10.1002/cpt.395
4. Shehab, N. et al. US emergency department visits for outpatient adverse drug events, 2013-
2014. JAMA - J. Am. Med. Assoc. 316, 2115–2125 (2016).
https://doi.org/10.1001/jama.2016.16201
5. Gouin‐Thibault, I. et al. Interindividual variability in dabigatran and rivaroxaban exposure:
contribution of ABCB1 genetic polymorphisms and interaction with clarithromycin. J.

This article is protected by copyright. All rights reserved

Accepted ArticleThromb. Haemost. 15, 273–283 (2017). https://doi.org/10.1111/jth.13577
6. Testa, S. et al. Plasma levels of direct oral anticoagulants in real life patients with atrial
fibrillation: Results observed in four anticoagulation clinics. Thromb. Res. 137, 178–183
(2016). https://doi.org/10.1016/j.thromres.2015.12.001
7. Tseng, A. S. et al. Clinical Review of the Pharmacogenomics of Direct Oral
Anticoagulants. Cardiovasc. Drugs Ther. 32, 121–126 (2018).
https://doi.org/10.1007/s10557-018-6774-1
8. Kanuri, S. H. & Kreutz, R. P. Pharmacogenomics of Novel Direct Oral Anticoagulants:
Newly Identified Genes and Genetic Variants. J. Pers. Med. 9, (2019).
https://doi.org/10.3390/jpm9010007
9. Raymond, J. et al. Pharmacogenetics of Direct Oral Anticoagulants: A Systematic Review.
J. Pers. Med. 11, 37 (2021). https://doi.org/10.3390/jpm11010037
10. Dimatteo, C. et al. ABCB1 SNP rs4148738 modulation of apixaban interindividual
variability. Thromb. Res. 145, 24–26 (2016). https://doi.org/10.1016/j.thromres.2016.07.005
11. Ueshima, S. et al. Population pharmacokinetics and pharmacogenomics of apixaban in
Japanese adult patients with atrial fibrillation. Br. J. Clin. Pharmacol. 84, 1301–1312
(2018). https://doi.org/10.1111/bcp.13561
12. Sennesael, A. L. et al. Rivaroxaban plasma levels in patients admitted for bleeding events:
Insights from a prospective study. Thromb. J. 16, (2018). https://doi.org/10.1186/s12959-
018-0183-3
13. Paré, G. et al. Genetic determinants of dabigatran plasma levels and their relation to
bleeding. Circulation 127, 1404–1412 (2013).
https://doi.org/10.1161/CIRCULATIONAHA.112.001233
14. Sychev, D. A. et al. The impact of ABCB1 (rs1045642 and rs4148738) and CES1
(rs2244613) gene polymorphisms on dabigatran equilibrium peak concentration in patients
after total knee arthroplasty. Pharmgenomics. Pers. Med. 11, 127–137 (2018).
https://doi.org/10.2147/PGPM.S169277
15. Gulilat, M. et al. Drug interactions and pharmacogenetic factors contribute to variation in
apixaban concentration in atrial fibrillation patients in routine care. J. Thromb.
Thrombolysis 49, 294–303 (2020). https://doi.org/10.1007/s11239-019-01962-2
16. Ji, Q. et al. The impact of ABCB1 and CES1 polymorphisms on dabigatran
pharmacokinetics and pharmacodynamics in patients with atrial fibrillation. Br. J. Clin.

This article is protected by copyright. All rights reserved

Accepted ArticlePharmacol. (2020). https://doi.org/10.1111/bcp.14646
17. Sychev, D. et al. Effect of CES1 and ABCB1 genotypes on the pharmacokinetics and
clinical outcomes of dabigatran etexilate in patients with atrial fibrillation and chronic
kidney disease. Drug Metab. Pers. Ther. 35, (2020). https://doi.org/10.1515/dmpt-2019-
0029
18. Kryukov, A. V. et al. Influence of ABCB1 and CYP3A5 gene polymorphisms on
pharmacokinetics of apixaban in patients with atrial fibrillation and acute stroke.
Pharmgenomics. Pers. Med. 11, 43–49 (2018). https://doi.org/10.2147/PGPM.S157111
19. Roşian, A. N. et al. Interindividual variability of apixaban plasma concentrations: Influence
of clinical and genetic factors in a real-life cohort of atrial fibrillation patients. Genes
(Basel). 11, (2020). https://doi.org/10.3390/genes11040438
20. Reilly, P. A. et al. The effect of dabigatran plasma concentrations and patient characteristics
on the frequency of ischemic stroke and major bleeding in atrial fibrillation patients: The
RE-LY trial (Randomized Evaluation of Long-Term Anticoagulation Therapy). J. Am. Coll.
Cardiol. 63, 321–328 (2014). https://doi.org/10.1016/j.jacc.2013.07.104
21. Huppertz, A. et al. Unexpected excessive apixaban exposure: Case report of a patient with
polymorphisms of multiple apixaban elimination pathways. BMC Pharmacol. Toxicol. 20,
(2019). https://doi.org/10.1186/s40360-019-0331-9
22. Lorenzini, K. I., Daali, Y., Fontana, P., Desmeules, J. & Samer, C. Rivaroxaban-induced
hemorrhage associated with ABCB1 genetic defect. Front. Pharmacol. 7, 494 (2016).
https://doi.org/10.3389/fphar.2016.00494
23. Helin, T., Joutsi-Korhonen, L. & Lassila, R. Clinical use and laboratory testing of oral
anticoagulation therapy: experience from Finland. Ann. Blood 4, 17–17 (2019).
https://doi.org/10.21037/aob.2019.07.01
24. Lähteenmäki, J. et al. Integrating data from multiple Finnish biobanks and national health-
care registers for retrospective studies: Practical experiences. Scand. J. Public Health
140349482110044 (2021). https://doi.org/10.1177/14034948211004421
25. Vuorinen, A.-L. et al. Pharmacogenetics of Anticoagulation and Clinical Events in
Warfarin-Treated Patients: A Register-Based Cohort Study with Biobank Data and National
Health Registries in Finland. Clin. Epidemiol. Volume 13, 183–195 (2021).
https://doi.org/10.2147/CLEP.S289031
26. FinnGen research project <https://www.finngen.fi/en> (2017). Accessed 5 May 2021.

This article is protected by copyright. All rights reserved

 27.
Accepted ArticleBorodulin, K. et al. Forty-year trends in cardiovascular risk factors in Finland. Eur. J.
Public Health 25, 539–546 (2015). https://doi.org/10.1093/eurpub/cku174
28. Aromaa, Arpo; Koskinen, S. Health and functional capacity in Finland: Baseline results of
the Health 2000 health examination survey. National Public Health Institute Finland; 2004.
https://www.julkari.fi/handle/10024/78534.
29. National Center for Biotechnology Information - dbSNP. at
<https://www.ncbi.nlm.nih.gov/snp/>. Accessed 5 May 2021.
30. Coleman, J. R. I. et al. Quality control, imputation and analysis of genome-wide genotyping
data from the Illumina HumanCoreExome microarray. Brief. Funct. Genomics 15, 298–304
(2016). https://doi.org/10.1093/bfgp/elv037
31. Levey, A. S. et al. A new equation to estimate glomerular filtration rate. Ann. Intern. Med.
150, 604–612 (2009). https://doi.org/10.7326/0003-4819-150-9-200905050-00006
32. INXBASE – MedBase <https://www.medbase.fi/en/professionals/inxbase>. Accessed 5
May 2021.
33. Huybrechts, K. F. et al. Claims Data Studies of Direct Oral Anticoagulants Can Achieve
Balance in Important Clinical Parameters Only Observable in Electronic Health Records.
Clin. Pharmacol. Ther. 105, 979 (2018). https://doi.org/10.1002/cpt.1256
34. Roşian, A. N. et al. An exploratory association analysis of ABCB1 rs1045642 and ABCB1
rs4148738 with non-major bleeding risk in atrial fibrillation patients treated with dabigatran
or apixaban. J. Pers. Med. 10, 1–14 (2020). https://doi.org/10.3390/jpm10030133
35. Stephens, M. & Scheet, P. Accounting for Decay of Linkage Disequilibrium in Haplotype
Inference and Missing-Data Imputation. Am. J. Hum. Genet 76, (2005).
36. Therneau, T. M. & Grambsch, P. M. Modeling Survival Data: Extending the Cox Model.
(Springer New York, New York, NY, 2000). https://doi.org/10.1007/978-1-4757-3294-8
37. Finnish Biobank Act (2012)
<https://www.finlex.fi/en/laki/kaannokset/2012/en20120688.pdf>. Accessed 5 May 2021.
38. Arlett, P., Kurz, X., Soltys, K. & Blum, M. D. International Collaboration in Real‐World
Evidence Generation for Direct Acting Oral Anti‐Coagulants. Clin. Pharmacol. Ther. 109,
299–301 (2021). https://doi.org/10.1002/cpt.1999
39. Ntaios, G. et al. Real-World Setting Comparison of Nonvitamin-K Antagonist Oral
Anticoagulants Versus Vitamin-K Antagonists for Stroke Prevention in Atrial Fibrillation:
A Systematic Review and Meta-Analysis. Stroke 48, 2494–2503 (2017).

This article is protected by copyright. All rights reserved

Accepted Articlehttps://doi.org/10.1161/STROKEAHA.117.017549
40. Gouin-Thibault, I. et al. Interindividual variability in dabigatran and rivaroxaban exposure:
contribution of ABCB1 genetic polymorphisms and interaction with clarithromycin. J.
Thromb. Haemost. 15, 273–283 (2017). https://doi.org/10.1111/jth.13577
41. Ueshima, S. et al. Impact of ABCB1, ABCG2, and CYP3A5 polymorphisms on plasma
trough concentrations of apixaban in Japanese patients with atrial fibrillation.
Pharmacogenet. Genomics 27, 329–336 (2017).
https://doi.org/10.1097/FPC.0000000000000294
42. Huppertz, A. et al. Unexpected excessive apixaban exposure: Case report of a patient with
polymorphisms of multiple apixaban elimination pathways. BMC Pharmacol. Toxicol. 20,
53 (2019). https://doi.org/10.1186/s40360-019-0331-9
43. Dimatteo, C. et al. Pharmacogenetics of dabigatran etexilate interindividual variability.
Thromb. Res. 144, 1–5 (2016). https://doi.org/10.1016/j.thromres.2016.05.025
44. Schaefer, J. K. et al. Impact of Adding Aspirin to Direct Oral Anticoagulant Therapy
without an Apparent Indication. Blood 134, 787–787 (2019). https://doi.org/10.1182/blood-
2019-122526
45. Said, A. et al. Concomitant use of direct oral anticoagulants and aspirin versus direct oral
anticoagulants alone in atrial fibrillation and flutter: A retrospective cohort. BMC
Cardiovasc. Disord. 20, (2020). https://doi.org/10.1186/s12872-020-01509-x
46. Zhang, Y. et al. Risk of major bleeding among users of direct oral anticoagulants combined
with interacting drugs: A population-based nested case–control study. Br. J. Clin.
Pharmacol. 86, 1150–1164 (2020). https://doi.org/10.1111/bcp.14227
47. Nordstrom, S. M. & Weiss, E. J. Sex differences in thrombosis. Expert Rev. Hematol. 1, 3–
8 (2008). https://doi.org/10.1586/17474086.1.1.3
48. Limdi, N. A. et al. Thromboembolic and Hemorrhagic Outcomes in the Direct Oral
Anticoagulant Trials Across the Spectrum of Kidney Function. Clin. Pharmacol. Ther.
cpt.2131 (2021). https://doi.org/10.1002/cpt.2131
49. Mega, J. L. et al. Genetics and the clinical response to warfarin and edoxaban: findings
from the randomised, double-blind ENGAGE AF-TIMI 48 trial. Lancet 385, 2280–2287
(2015). https://doi.org/10.1016/S0140-6736(14)61994-2
50. Sund, R. Quality of the Finnish Hospital Discharge Register: A systematic review. Scand. J.
Public Health 40, 505–515 (2012). https://doi.org/10.1177/1403494812456637

This article is protected by copyright. All rights reserved

Accepted Article
Figure legends

Figure 1. Cohort formation.

Figure 2. Incidence of bleeding and thromboembolic events for SNVs ABCB1 c.2482-2236G>A (a), ABCB1
c.3435C>T (b) and ABCB1 c.2677G>T (c). Abbreviations: HR, hazard ratio; CI, confidence interval

Figure 3. Incidence of bleeding and thromboembolic events for haplotypes ABCB1 1236T-2677T-3435T (a),
1236C-2677G-3435C (b) and 1236T-2677G-3435C (c). Abbreviations: HR, hazard ratio; CI, confidence
interval

Supplementary files
1. Table S1
2. Table S2
3. Table S3
4. Table S4
5. Table S5
6. Table S6
7. Table S7

This article is protected by copyright. All rights reserved

 Table 1. Characteristics of the study patients.
Accepted Article dabigatran rivaroxaban apixaban

Number of patients 340 999 467

Sex, female, n (%) 162 (47.6%) 498 (49.8%) 239 (51.2%)

Age, years, mean (SD) 69.8(8.7) 69.6(9.8) 72.3(9.5)

Exposure time, days, median (IQR) 217 (60-671) 137 (57-391) 300 (202-561)

Indication for antithrombotic drug use

Atrial fibrillation, n (%) 268 (78.8%) 537 (53.8%) 337 (72.2%)

Vascular disease, n (%) 32 (9.4%) 230 (23.0%) 44 (9.4%)

Pulmonary embolism, n (%) 10 (2.9%) 66 (6.6%) 34 (7.3%)

Stroke, cerebral infarction, atherosclerosis, n (%) 16 (4.7%) 44 (4.4%) 27 (5.8%)

Venous thrombosis, n (%) 14 (4.1%) 122 (12.2%) 25 (5.4%)

Abbreviations: IQR, interquartile range; SD, standard deviation

This article is protected by copyright. All rights reserved

 Table 2. Results of SNV association analysis. Non-carriers (individuals with two reference alleles)
Accepted Article
were compared to carriers (heterozygous or homozygous individuals) for the variant allele.

Number of patients
Bleeding event risk Thromboembolic event risk
(non-carrier /
SNV hazard ratio hazard ratio
heterozygote/
(95% CI), p-value (95% CI), p-value
homozygote)

Dabigatran
ABCB1 c.3435C>T 64/148/128 1.08 (0.24-4.87), p=0.923 1.16 (0.14-9.70), p=0.888
ABCB1 c.2677G>T 91/172/77 2.00 (0.44-9.02), p=0.368 0.88 (0.17-4.56), p=0.883
ABCB1 c.2677G>A 317/23/0 n/a 2.17 (0.26-18.10), p=0.473
ABCB1 c.1236C>T 93/170/77 1.27 (0.35-4.61), p=0.719 0.95 (0.18-4.87), p=0.947
ABCB1 c.2482-2236G>A 83/170/87 1.83 (0.41-8.25), p=0.432 0.83 (0.16-4.26), p=0.820
CES1 c.1168-33C>T 229/101/10 0.77 (0.24-2.51), p=0.665 1.31 (0.29-5.89), p=0.721
CES1 c.257+885T>C 168/150/22 0.49 (0.16-1.52), p=0.219 1.09 (0.24-4.88), p=0.913

Rivaroxaban
ABCB1 c.3435C>T 170/492/337 0.84 (0.37-1.91), p=0.682 0.42 (0.18-0.98), p=0.044
ABCB1 c.2677G>T 282/495/222 0.88 (0.44-1.78), p=0.727 0.50 (0.23-1.08), p=0.077
ABCB1 c.2677G>A 928/69/2 1.06 (0.33-3.43), p=0.926 n/a

ABCB1 c.1236C>T 274/509/216 1.03 (0.50-2.12), p=0.929 0.82 (0.36-1.89), p=0.648

Apixaban
ABCG2 c.421C>A 400/65/2 2.30 (0.89-5.95), p=0.085 0.79 (0.18-3.51), p=0.755
ABCB1 c.345C>T 75/245/147 0.81 (0.27-2.40), p=0.699 2.69 (0.35-20.48), p=0.339
ABCB1 c.2677G>T 118/256/93 1.13 (0.41-3.08), p=0.815 1.48 (0.42-5.25), p=0.545

ABCB1 c.2677G>A 440/27/0 n/a 1.28 (0.17-9.79), p=0.810

ABCB1 c.2482-2236G>A 107/257/103 0.37 (0.16-0.89), p=0.025 0.58 (0.20-1.69), p=0.314
ABCB1 c.1236C>T 113/256/98 1.46 (0.49-4.35), p=0.494 1.44 (0.41-5.13), p=0.571

CYP3A5 g.6981A>G1 7/54/406 1.49 (0.35-6.41), p=0.591 0.63 (0.18-2.23), p=0.471

Abbreviations: CI, confidence interval; SNV, single nucleotide variant

1
Patients with zero or one variant allele compared to homozygotes to enable sufficiently large
patient groups.

This article is protected by copyright. All rights reserved

 Table 3. Results of haplotype-based association analysis. Non-carriers (individuals without the
Accepted Article
haplotype) were compared to carriers (heterozygous or homozygous individuals) for the haplotype.

Number of patients
Bleeding event risk Thromboembolic event risk
(non-carrier/
hazard ratio hazard ratio
ABCB1 haplotype heterozygote/
(95% CI), p-value (95% CI), p-value
homozygote)

Dabigatran
1236T-2677T-3435T 108/167/65 1.58 (0.44-5.76), p=0.484 1.18 (0.23-6.08), p=0.844
1236C-2677G-3435C 155/143/42 2.02 (0.62-6.55), p=0.244 2.23 (0.43-11.50), p=0.338
1236C-2677G-3435T 262/73/5 0.51 (0.11-2.33), p=0.388 0.49 (0.06-4.06), p=0.507
1236T-2677G-3435C 319/21/0 1.15 (0.15-8.83), p=0.896 n/a

Rivaroxaban
1236T-2677T-3435T 318/500/181 1.09 (0.54-2.19), p=0.808 0.44 (0.20-0.95), p=0.036
1236C-2677G-3435C 431/445/123 1.37 (0.71-2.63), p=0.347 2.55 (1.03-6.36), p=0.044
1236C-2677G-3435T 778/207/14 1.07 (0.52-2.20), p=0.847 0.78 (0.29-2.06), p=0.611

1236T-2677G-3435C 941/56/2 0.51 (0.07-3.73), p=0.508 5.88 (2.35-14.72), p<0.001

Apixaban
1236T-2677T-3435T 142/247/78 1.50 (0.55-4.10), p=0.428 1.99 (0.56-7.07), p=0.287

1236C-2677G-3435C 194/224/49 1.08 (0.45-2.60), p=0.866 0.75 (0.27-2.06), p=0.570

1236C-2677G-3435T 367/94/6 0.36 (0.08-1.54), p=0.169 0.81 (0.23-2.88), p=0.745
1236T-2677G-3435C 431/36/0 0.69 (0.09-5.13), p=0.714 1.06 (0.14-8.13), p=0.952

Abbreviations: CI, confidence interval

This article is protected by copyright. All rights reserved

Accepted Article cpt_2316_f1.pdf

Auria Biobank Helsinki Biobank THL Biobank

839 patients 1,521 patients 4,645 patients

Original PreMed study cohort of 7,005 patients

Excluded patients
with no DOAC purchases
(n=5,199)

1,806 users of dabigatran, rivaroxaban or apixaban

This article is protected by copyright. All rights reserved
Accepted Article

cpt_2316_f2.tif

This article is protected by copyright. All rights reserved

Accepted Article

cpt_2316_f3.tif

This article is protected by copyright. All rights reserved