Received: 21 September 2022

- -Revised: 9 November 2022

https://doi.org/10.1016/j.rpth.2023.100052

BRIEF REPORT
Accepted: 30 December 2022

Clinically relevant increases in the international normalized

ratio and model of end-stage liver disease score by therapeutic
doses of direct oral anticoagulants in patients with cirrhosis

Ton Lisman PhD1,2 | William Bernal MD, PhD3 | Jelle Adelmeijer BSci1 |
Pieter-Willem Kamphuisen MD, PhD4 | Sarah Bos MD, PhD5 | Robert J. Porte MD, PhD2

1
Surgical Research Laboratory, Department
of Surgery, University of Groningen, Abstract
University Medical Center Groningen,
Background: Patients with cirrhosis are increasingly using direct oral anticoagulants
Groningen, the Netherlands
2 (DOACs) in therapeutic doses for the treatment of portal vein thrombosis or for
Section of Hepatobiliary Surgery and Liver
Transplantation, Department of Surgery, concomitant atrial fibrillation. DOACs may affect routine diagnostic tests of coagula-
University of Groningen, University Medical
tion including the international normalized ratio (INR). The INR is a part of the model of
Center Groningen, Groningen, the
Netherlands end-stage liver disease (MELD) score, a validated score that predicts the mortality risk
3
Liver Intensive Therapy Unit, Institute of in patients with cirrhosis and is used to prioritize patients for liver transplantation.
Liver Studies, Kings College Hospital,
Denmark Hill, London SE5 9RS, UK
DOAC–induced increases in the INR may thus lead to artificial inflation of the MELD
4
Department of Internal Medicine, Tergooi score.
Hospital, Hilversum, the Netherlands and Objective: We studied the effect of DOACs on INR prolongation in patients with
Department of Vascular Medicine,
Amsterdam Cardiovascular Science, cirrhosis.
Amsterdam University Medical Centres, Methods: We spiked plasma from 20 healthy individuals and 20 patients at the start of
University of Amsterdam, Amsterdam, the
Netherlands liver transplantation with DOACs in concentrations representing peak therapeutic
5
Department of Gastroenterology, Treant levels. In addition, we studied INR increases in healthy controls and patients with mild
Hospital, Emmen, the Netherlands cirrhosis who received the DOAC edoxaban for 1 week for study purposes.
Correspondence Results: In controls and patients, the INR increased by an ex vivo addition of a DOAC,
Ton Lisman, Department of Surgery, and the INR increase in patients was proportional to the baseline INR values. The in-
University Medical Center Groningen,
BA44, Hanzeplein 1, 9713 GZ Groningen, crease in INR translated to a median increase of between 3 and 10 MELD points,
the Netherlands. depending on the DOAC used. In controls and patients alike, the INR increased on the
Email: [email protected]
ingestion of edoxaban, which translated to an increase in 5 MELD points.
Funding information Conclusions: Taken together, DOACs result in an INR increase that translates to
This study was funded by the departmental
funds of T.L. clinically meaningful increases in MELD points in patients with cirrhosis, and pre-
cautions to avoid artificial inflation of the MELD score in these patients are warranted.
Handling Editor: N Zakai

KEYWORDS
anticoagulants, cirrhosis, end-stage liver disease, international normalized ratio, liver
transplantation

-
© 2023 The Author(s). Published by Elsevier Inc. on behalf of International Society on Thrombosis and Haemostasis. This is an open access article under the CC BY
license (http://creativecommons.org/licenses/by/4.0/).

Res Pract Thromb Haemost. 2023;7:e100052 www.rpthjournal.org 1 of 5

https://doi.org/10.1016/j.rpth.2023.100052
2 of 5
- LISMAN ET AL.

Essentials
• Patients with chronic liver disease increasingly use direct oral anticoagulants (DOACs) to prevent or treat thrombosis.
• DOACs increase the international normalized ratio, a laboratory test used to prioritize patients for a donor liver.
• The model of end-stage liver disease score used for this prioritization process is substantially affected by DOACs.
• Measures to fairly and adequately prioritize waitlisted patients using DOACs are needed.

1 | INTRODUCTION patient sample as the INR may be normal in the presence of thera-
peutic drug doses. Conversely, the INR may also be substantially
Patients with cirrhosis may develop substantial alterations in their prolonged in the presence of (sub)therapeutic drug doses. It has not
hemostatic system that paradoxically may be associated with both been studied to what extent DOACs increase the INR in patients with
bleeding and thrombosis [1]. There is an increasing awareness of the cirrhosis, who may already have INR prolongations at baseline.
thrombotic risk in patients with cirrhosis, notably the risk for venous Here, we studied the effect of DOACs on INR prolongation in
thromboembolism (deep vein thrombosis and pulmonary embolism) patients with cirrhosis. We used samples from patients undergoing
[2] and the risk for portal vein thrombosis [3]. Patients may, therefore, liver transplantation to which we added DOACs in clinically relevant
be treated with anticoagulant drugs for the prevention or treatment concentrations. Specifically, we used median peak DOAC values that
of venous thromboembolism or portal vein thrombosis, and in addi- have been reported in patients without underlying liver disease. In
tion, patients may receive anticoagulant drugs to prevent thrombotic addition, we studied INR changes in patients with cirrhosis who
complications related to atrial fibrillation. received one specific DOAC (edoxaban) for study purposes. We
Long-term anticoagulation in patients with cirrhosis may be calculated the effects of INR prolongation by the various DOACs on
accomplished using vitamin K antagonists, low-molecular weight the MELD score.
heparin, or direct oral anticoagulants (DOACs). Because vitamin K
antagonists are difficult to dose in patients with cirrhosis and an
elevated baseline international normalized ratio (INR), and daily in- 2 | MATERIALS AND METHODS
jections with low-molecular weight heparin are a substantial burden
for the patient, DOACs are gaining popularity in the cirrhotic patients, 2.1 | Patients
although data on safety and efficacy are still scarce [4]. Despite this
lack of firm evidence, recent clinical guidance documents state that We studied samples from 20 adult patients undergoing liver trans-
DOACs are a reasonable treatment option in patients with Child A or plantation at King’s College Hospital, London, between September
B cirrhosis [5,6]. 2017 and December 2017, and 20 healthy individuals recruited in the
In patients with cirrhosis that become liver transplant candidates, same study. Patient characteristics have been described previously [10].
the model for end-stage liver disease (MELD) score is used to priori- Only samples taken after the induction of anesthesia were studied. The
tize patients on the waiting list. The MELD score is calculated using study was approved by the NRES Committee London—Westminster,
the following formula: 9.57 × loge (creatinine) + 3.78 × loge (total Study Number 17/LO/0527. In addition, samples from 16 adult patients
bilirubin) + 11.2 × loge (INR) + 6.43. Because the MELD score includes with cirrhosis and 16 healthy volunteers recruited in the University
the INR, the score is artificially inflated in patients receiving vitamin K Medical Center Groningen, the Netherlands, who received edoxaban
antagonists. It has been previously proposed to use a simplified MELD (60 mg once daily, administered for 7 consecutive days) were studied.
score that does not include the INR to prioritize patients who use Characteristics of patients and controls have been outlined previously
vitamin K antagonists [7]. Within Eurotransplant, the MELD score for [11]. Samples were taken twice at day 1 (baseline and 2 h after the first
waitlisted patients on vitamin K antagonists has to be calculated using dose) and once on day 3 and day 7, 2 hours after ingestion of edoxaban.
the last value before starting vitamin K antagonists, or the vitamin K The study protocol was approved by the local medical ethical committee
antagonists have to be stopped for at least 2 weeks to determine the (METc 2016/226) and was registered at the Netherlands Trial Register
current INR (https://www.eurotransplant.org/wp-content/uploads/2 (NTR6397). Both studies were executed in accordance with both the
022/03/H5-ELAS-MELD-March-2022.pdf). Declarations of Helsinki and Istanbul, and written informed consent was
In the general population, DOACs may also lead to prolongations obtained from each subject before inclusion in both studies.
in the INR [8,9]. The sensitivity of the INR for DOAC depends on the
type of DOAC (with rivaroxaban having the most and apixaban the
least profound effect) and the reagent and coagulation analyzer used. 2.2 | INR determination
The prolongation of the INR has a poor correlation with the plasma
concentration of the drug. The INR, therefore, cannot be used to INRs were measured on an automated coagulation analyzer (STA-
ascertain whether a clinically relevant DOAC dose is present within a Compact 3, Stago) with the use of reagents and protocols from the
LISMAN ET AL.
- 3 of 5

T A B L E 1 The effect of ex vivo addition of DOACs on the INR in healthy individuals and patients undergoing liver transplantation sampled
after the induction of anesthesia.

INR healthy individuals, INR patients with cirrhosis, MELD score increase in
median (range); fold increase median (range); fold increase patients by the DOAC

- 1.06 (0.93-1.15) 1.48 (1.06-3.06)

Dabigatran (175 ng/mL) 1.35 (1.22-1.50); 1.3 2.30 (1.36-7.62); 1.5 4.6 (2.8-10.2)

Rivaroxaban (250 ng/mL) 1.99 (1.74-2.47); 1.9 3.71 (2.08-12.00); 2.5 10.2 (7.3-15.3)

Apixaban (150 ng/mL) 1.16 (1.07-1.31); 1.1 1.92 (1.15-6.41); 1.3 2.9 (0.9-8.3)

Edoxaban (200 ng/mL) 1.59 (1.46-1.81); 1.5 2.79 (1.52-8.04); 1.9 7.0 (4.0-10.8)

The fold increase in INR is significantly higher in patients for all DOACs (all P < .001 by the Mann–Whitney U-test).
DOAC, direct oral anticoagulant; INR, international normalized ratio.

manufacturer. Plasma samples from the liver transplant recipients 3 | RESULTS AND DISCUSSION
were spiked with dabigatran (175 ng/mL), rivaroxaban (250 ng/mL),
apixaban (150 ng/mL), or edoxaban (200 ng/mL). All drugs were pur- In 20 patients with cirrhosis undergoing liver transplantation, the INR
chased from Alsachim (Illkirch-Graffenstaden, Germany). was 1.48 [1.06-3.76] (median [range]), whereas the INR was 1.06
[0.93-1.15] in healthy volunteers. We spiked plasma samples from
these patients with DOACs with a concentration found at peak level in
2.3 | Contribution of INR increase to the MELD
patients by using these drugs in a therapeutic regimen [8]. The INR
score
increased after an ex vivo addition of dabigatran, rivaroxaban, apix-
aban, and edoxaban in both patients and controls. However, the in-
We calculated the contribution of the INR to the MELD score as
crease in patients was more pronounced than that in controls
11.2 × loge (INR). The contribution of the INR to the MELD score was
(Table 1). The increase in INR in patients resulted in a clinically rele-
determined in absence and presence of a DOAC, and the increase in
vant increase in MELD points for all drugs, with the largest increase
MELD points by the DOAC is reported.
after the addition of rivaroxaban and the smallest increase after the
addition of apixaban. In patients, but not in healthy controls, the
2.4 | Statistical analyses baseline INR positively correlated to the absolute INR increase
(Figure).
Data are reported as median with ranges. The relative increase in INR We next measured INRs in samples taken before and at 2 hours, 3
after the addition of a DOAC was compared between patients and days, and 7 days after the start of a once daily regimen of 60 mg of
controls using a Mann–Whitney U-test. Relationship between baseline edoxaban. In both patients and controls, the INR increased on edox-
INR and INR increase after the addition of a DOAC was tested using aban (Table 2). The increase in patients was comparable with that in
simple linear regression, and R2 values are reported. P values <.05 controls and resulted in a median increase of approximately 5 MELD
were considered statistically significant. points.

A B
Increase in INR after addition of DOAC
Increase in INR after addition of DOAC

3 4
Dabigatran Dabigatran, r2 = 0.83
Rivaroxaban Rivaroxaban, r2 = 0.76
2 Apixaban 3 Apixaban, r2 = 0.76
Edoxaban Edoxaban, r2 = 0.70

1 2

0 1
75
00
25
50
75
00
25
50
75
00
25
50
75

00

25

0.
1.
1.
1.
1.
2.
2.
2.
2.
3.
3.
3.
0.

1.

1.

Baseline INR Baseline INR

F I G U R E Correlation of baseline international normalized ratio values to the absolute increase in the international normalized ratio values
after the addition of direct oral anticoagulants in healthy controls (A) and in patients undergoing liver transplantation (B). R2 values are
indicated for patients (all P < .0001), and in controls, the correlations were not significant.
4 of 5
- LISMAN ET AL.

T A B L E 2 INR changes in healthy individuals and patients with clinical practice, DOACs may not be stopped immediately following
cirrhosis before and during 1-week treatment with edoxaban.
worsening of disease, and extra care should be taken in interpreting
INR healthy INR patients MELD score the MELD score in such patients.
individuals, with cirrhosis, increase in Taken together, we have demonstrated that DOACs at peak
median (range); median (range); patients by
plasma levels may result in profound increases in the MELD score in
fold increase fold increase the DOAC
patients with cirrhosis owing to their effects on the INR. These in-
- 0.97 (0.89-1.19) 1.11 (1.03-1.77)
creases may still be present at trough levels in patients taking dabi-
Day 1 1.43 (1.08-2.34); 1.4 1.83 (1.38-2.90); 1.6 5.6 (2.8-9.4) gatran or apixaban because plasma levels of these drugs are only 1.5-
Day 3 1.43 (1.18-2.41); 1.5 1.75 (1.06-2.68); 1.5 4.7 (0.2-8.5) fold to 3-fold lower than the peak levels. Given the clinical relevance
of our findings, confirmation of our findings by independent labora-
Day 7 1.54 (1.18-2.22); 1.6 1.79 (1.46-3.01); 1.6 5.6 (2.9-9.2)
tories is warranted. Ideally, such studies would measure INRs at
DOAC, direct oral anticoagulant; INR, international normalized ratio.
various time points after DOAC ingestion in larger number of patients
taken all clinically available DOACs. In addition, studies assessing the
Here, we demonstrate that ex vivo addition of DOACs at doses strategies to determine DOAC-independent INR and MELD scores in
found at therapeutic peak levels in the general population to plasma patients on DOACs, for example, by using DOAC stop, are indicated.
from liver transplant candidates increases the INR to a greater extent Given the increasing number of sicker patients who use DOACs that
compared with healthy individuals. The doses studied result in an INR may continue to use these drugs while on the waiting list or even up to
prolongation that translate to a median of 3 to 10 MELD points, transplantation [13–15], it will be important to take DOAC use into
depending on the DOAC, and the effect is larger with a higher base- account in the current organ allocation procedures in order to avoid
line INR. In patients with mild cirrhosis who received a therapeutic an unwanted favoring of DOAC users.
dose of edoxaban, the INR prolongation at peak levels was similar
between patients and controls and resulted in an additional 5 MELD
FUNDI NG
points. The clinically relevant inflation of the MELD score by peak
This study was funded by departmental funds of T.L.
therapeutic plasma levels call for action to fairly and adequately pri-
oritize patients who are using DOACs while on the transplant waiting
AUTHOR C ONTRIBUTIONS
list.
T.L. conceived the study, analyzed data, and wrote the manuscript;
A simple solution to assess INR values in patients who are using
W.B. supervised sample collection, interpreted data, and revised the
DOACs is to remove the DOAC from the blood sample before the INR
manuscript; J.A. performed analyses, interpreted data, and revised the
determination. Commercially available agents, such as DOAC stop
manuscript; P.W.K. supervised sample collection, interpreted data, and
(Haematex), that use activated charcoal to remove DOACs from the
revised the manuscript; S.B. collected samples, interpreted data, and
sample have been shown to effectively neutralize DOAC effects on
revised the manuscript; R.J.P. conceived the study, interpreted data,
coagulation tests including the INR [12]. However, not all clinical
and revised the manuscript.
laboratories offer the use of DOAC removal. Other options include
the use of a simplified MELD score, as previously proposed for pa-
tients that use vitamin K antagonists while on the waiting list [7], or REL ATI O NS HI P D I SC LOS U RE
sampling of patients at the trough DOAC level. We have not quanti- There are no competing interests to disclose.
fied the INR effect at trough levels, but because there is a linear
relation between INR and DOAC levels [8], these effects can be
DA TA AVA ILAB IL IT Y
estimated. Because in the general population, therapeutic trough
The data that support the findings of this study are available from the
plasma levels are 5 to 10 times lower than peak levels for rivaroxaban
corresponding author upon reasonable request.
and edoxaban, the MELD inflation is largely gone at trough levels,
assuming comparable pharmacokinetics in patients with cirrhosis and
REFER ENC ES
individuals with a healthy liver. However, trough levels for dabigatran [1] Lisman T, Porte RJ. Pathogenesis, prevention, and management of
and apixaban are only 1.5-fold to 3-fold lower than the peak levels, bleeding and thrombosis in patients with liver diseases. Res Pract
which means that MELD inflation at the trough levels would still be Thromb Haemost. 2017;1:150–61.
significant. In these patients, the INR ideally could be measured after a [2] Ambrosino P, Tarantino L, Di Minno G, Paternoster M, Graziano V,
Petitto M, et al. The risk of venous thromboembolism in patients
deliberately missed dose, but this may not be feasible and safe in all
with cirrhosis: a systematic review and meta-analysis. Thromb Hae-
patients. Importantly, because DOACs are metabolized by liver and most. 2017;117:139–48.
kidney in patients with advancing cirrhosis who may be complicated [3] Loffredo L, Pastori D, Farcomeni A, Violi F. Effects of anticoagulants
by renal failure, DOAC metabolism may be compromised, leading to in patients with cirrhosis and portal vein thrombosis: a systematic
review and meta-analysis. Gastroenterology. 2017;153:480–7.e1.
elevated peak and trough levels with accompanying additional ele-
[4] Lisman T, Kamphuisen PW, Northup PG, Porte RJ. Established and
vations in INR and further inflation of the MELD score. Although new-generation antithrombotic drugs in patients with cirrhosis—
DOACs are contraindicated for the sickest patients with cirrhosis, in possibilities and caveats. J Hepatol. 2013;59:358–66.
LISMAN ET AL.
- 5 of 5

[5] Northup PG, Garcia-Pagan JC, Garcia-Tsao G, Intagliata NM, [10] Bos S, van den Boom B, Ow TW, Prachalias A, Adelmeijer J,
Superina RA, Roberts LN, et al. Vascular liver disorders, portal vein Phoolchund A, et al. Efficacy of pro- and anticoagulant strategies in
thrombosis, and procedural bleeding in patients with liver disease: plasma of patients undergoing hepatobiliary surgery. J Thromb
2020 practice guidance by the American Association for the Study of Haemost. 2020;18:2840–51.
Liver Diseases. Hepatology. 2021;73:366–413. [11] Bos S, Schreuder T, Blokzijl H, Adelmeijer J, Lisman T,
[6] European Association for the Study of the Liver. EASL Kamphuisen PW. Anticoagulant activity of edoxaban in patients with
Clinical Practice Guidelines on prevention and management of cirrhosis. Blood. 2020;136:1561–4.
bleeding and thrombosis in patients with cirrhosis. J Hepatol. [12] Exner T, Michalopoulos N, Pearce J, Xavier R, Ahuja M. Simple
2022;76:1151–84. method for removing DOACs from plasma samples. Thromb Res.
[7] Heuman DM, Mihas AA, Habib A, Gilles HS, Stravitz RT, Sanyal AJ, 2018;163:117–22.
et al. MELD-XI: a rational approach to “sickest first” liver trans- [13] Semmler G, Pomej K, Bauer DJM, Balcar L, Simbrunner B, Binter T,
plantation in cirrhotic patients requiring anticoagulant therapy. Liver et al. Safety of direct oral anticoagulants in patients with advanced
Transpl. 2007;13:30–7. liver disease. Liver Int. 2021;41:2159–70.
[8] Douxfils J, Ageno W, Samama CM, Lessire S, Ten Cate H, [14] Mort JF, Davis JPE, Mahoro G, Stotts MJ, Intagliata NM,
Verhamme P, et al. Laboratory testing in patients treated with direct Northup PG. Rates of bleeding and discontinuation of direct oral
oral anticoagulants: a practical guide for clinicians. J Thromb Haemost. anticoagulants in patients with decompensated cirrhosis. Clin Gas-
2018;16:209–19. troenterol Hepatol. 2021;19:1436–42.
[9] Samuelson BT, Cuker A, Siegal DM, Crowther M, Garcia DA. Labo- [15] Intagliata NM, Maitland H, Pellitier S, Caldwell SH. Reversal of direct
ratory assessment of the anticoagulant activity of direct oral anti- oral anticoagulants for liver transplantation in cirrhosis: a step for-
coagulants: a systematic review. Chest. 2017;151:127–38. ward. Liver Transpl. 2017;23:396–7.