Review Article

The Safety of Anticoagulants and Antiplatelet Agents in Patients With Cirrhosis

Jiayi Ma; Naga P. Chalasani; Linus Schwantes-An; Einar Stefán Björnsson

Aliment Pharmacol Ther. 2023;57(1):52-71.

Abstract and Introduction

Abstract

Patients with cirrhosis were long thought to be coagulopathic. However, this paradigm has changed in recent years and currently, cirrhosis is recognised as a prothrombotic state. Due to the increasing incidence of cirrhosis from nonalcoholic steatohepatitis which is closely associated with cardiac disease, patients with cirrhosis increasingly require therapy with anticoagulants and antiplatelet agents. However, their potential for causing catastrophic and life-threatening bleeding in patients with cirrhosis leads to hesitancy about their use in patients with cirrhosis. Overall, traditional anticoagulation is safe for all Child-Pugh classes while newer direct oral anticoagulants (DOACs) are mostly safe in Child-Pugh class A/B and contraindicated in severe hepatic impairment. For different indications, published data to date suggest that anticoagulation is overall safe for patients with cirrhosis who have venous thromboembolism, atrial fibrillation and portal vein thrombosis, and does not increase the risk of variceal bleeding. Moreover, DOACs appear to have similar safety profiles as traditional anticoagulants. Finally, most studies suggest that antiplatelet agents are also safe to use in patients with cirrhosis although they are mostly contraindicated in severe hepatic impairment. For both anticoagulants and antiplatelet agents, severe thrombocytopaenia presents a relative contraindication to their use. More prospective trials and large cohort studies are needed to advance our understanding of the safety and nuances of DOACs and antiplatelet agents in patients with advanced cirrhosis.

Introduction

The alterations in the coagulation of patients with chronic liver disease and cirrhosis are complex. In early literature, these patients were considered auto-anticoagulated due to their prolonged prothrombin time, elevated International Normalised Ratio (INR) and thrombocytopenia. However, in recent years, the paradigm of haemostasis in cirrhosis has undergone many changes. Currently, cirrhosis is no longer considered just a state of coagulopathy, but also a prothrombotic state. In fact, studies have shown that patients with chronic liver disease and particularly those with cirrhosis are prone to develop thrombotic complications such as venous thromboembolism (VTE)^[1] and portal vein thrombosis (PVT).^[2] Furthermore, with the recent increase in the incidence and prevalence of nonalcoholic steatohepatitis (NASH) and its associated metabolic risk factors, many patients with cirrhosis develop atrial fibrillation (AF).^[3] Thus, many patients with chronic liver disease and cirrhosis have important indications for anticoagulation.

Similarly, antiplatelet (AP) medications are more frequently used in patients with cirrhosis. With the increasing prevalence of nonalcoholic fatty liver disease (NAFLD) in the general population, the burden of cardiovascular comorbidities is also increasing in patients with chronic liver disease.^[4] Whilst patients with cirrhosis were initially thought to be protected from coronary artery disease (CAD), recent studies have shown that patients with cirrhosis have a higher prevalence of CAD versus the general population.^[5] In fact, underlying CAD is a major contributor to morbidity and mortality of liver transplant recipients.^[6]

However, the potential for causing catastrophic and life-threatening bleeding in patients with cirrhosis leads to hesitancy about using anticoagulants including DOACs and AP agents in this population. Thus, it is of major importance to assess the safety profile of these drugs in patients who otherwise have appropriate clinical indications for these classes of medicines. In the following sections, we review current prescribing recommendations for different classes of anticoagulants and APs with a focus on hepatic impairment. The Child-Pugh score uses a combination of biochemical and clinical data to assess the severity of liver disease summarised as class A, B and C representing mild, moderate and severe hepatic impairment, respectively. We will also review the literature evaluating the safety of anticoagulants and AP agents in patients with cirrhosis for different clinical indications.

1 2 3 4

Next Section

Table 1. Direct oral anticoagulants versus traditional anticoagulants for the treatment of atrial fibrillation, portal vein thrombosis and venous thromboembolism
Study publication year	Patient population	Liver disease severity	Anticoagulation (n)	Bleeding	Comments
Semmler (2021)	191 patients	CP A/B/C DOACs: 78/47/8 VKA/LMWH: 35/21/2	DOACs: 75 edoxaban, 24 apixaban, 24 rivaroxaban, 3 dabigatran, 7 sequential treatment Traditional: 30 VKA, 28 LMWH	Procedure-related bleedings DOACs vs traditional: 11.7% vs 12.8% (p = 0.764) Spontaneous bleeding DOACs vs traditional: 28.6% vs 19% (p = 0.162)	Child-Pugh stage independently associated with major bleeding. Presence of varices or DOACs use were not
Jones et al. (2020)	79 patients	CP A^a/B^a/C DOACs: 34/8/0 VKA: 16/19/2	DOACs: 29 apixaban, 9 rivaroxaban, 4 dabigatran	DOACs vs VKA All-cause bleeding: 16.7% vs 21.6%, p = 0.77 Major bleeding: 2.4% vs 5.4% p = 0.60 Nonmajor bleeding: 14.3% vs 16.2% p = 0.99	No significant risk factors identified for all-cause bleeding including CP score and presence of varices
Davis et al. (2020)	167 patients	CP A/B/C VKA 43/53/7 DOACs 35/11/3	VKA 110 (29% DVT, 16% PE, 17% PVT^a, 25% AF^a) DOACs 57 (18% DVT, 16% PE, 5% PVT^a, 40% AF^a)	VKA vs DOACs Major bleeding at 90 days: 9.1% vs 5.2% p = 0.381 GI bleed 6% vs 4% p = 0.706	Rate of all-cause mortality not significantly different between VKA and DOACs groups (25.5% vs 26.3% p = 0.904)
Hum et al. (2016)	45 patients	CP A/B/C VKA/LMWH 7/9/2 DOACs 11/12/4	VKA 15, LMWH 3 (50% AF, 44% DVT, 17% PVT) DOACs 27 (56% AF, 44% DVT, 15% PVT)	VKA/LMWH vs DOACs Total bleeds 56% vs 30% p = 0.12 Major bleeds 28% vs 4% p = 0.03 GI bleeds 22% vs 18% p = 1.00	Four GI bleeds in the traditional group versus five GI bleeds in DOACs group (p = 1.00)
Intagliata et al. (2016)	39 patients	CP A/B VKA/LMWH 9/10 DOACs 9/11	VKA 13, LMWH 6 (63% VTE, 32% PVT, 5% AF) apixaban 11, rivaroxaban 9 (20% VTE, 60% PVT, 20% AF)	VKA/LMWH vs DOACs Any bleeding 16% vs 20% p = 0.99 Major bleeding 11% vs 5% p = 0.6	One patient (5%) discontinued DOACs due to bleeding complications versus 3 patients (16%) discontinued VKA or LMWH

Abbreviations: AF, atrial fibrillation; CP, Child-Pugh; DOACs, direct oral anticoagulant; DVT, deep vein thrombosis; LMWH, low molecular weight heparin, PE, pulmonary embolism; PVT, portal vein thrombosis; VKA, vitamin K antagonist; VTE, venous thromboembolism.
^aStatistically significant.

Table 2. Anticoagulation for atrial fibrillation
Study publication year	Patient population	Liver disease severity	Anticoagulation (n)	Bleeding	Comments
Serper et al. (2021)	2400 patients	VKA-matched cohort No treatment: 73% CP A VKA: 70% CP A DOACs-matched cohort No treatment: 90% CP A DOACs: 91% CP A	VKA n = 614 DOACs n = 201	Any bleeding VKA vs no treatment HR: 1.50 (1.10–2.06) DOACs vs no treatment HR: 0.77 (0.40–1.48) DOACs vs VKA HR: 0.49 (0.26–0.94)	Effect on HD VKA vs no treatment HR 0.73 (0.54–1.02) DOACs vs no treatment HR 0.71 (0.40–1.28)
Lee et al. (2019)	2428 patients	CP scores not available	VKA n = 990 DOACs n = 1438 DOACs (171 apixaban, 535 dabigatran, 732 rivaroxaban)	DOACs vs VKA GI bleeding: HR 0.51 (0.32–0.79, p = 0.0030) Major bleeding: HR 0.51 (0.32–0.74, p = 0.0003)	Compensated cirrhosis, DOACs vs VKA GI bleeding: HR 0.45 (0.26–0.78, p = 0.0045) Major bleeding: HR 0.51 (0.33–0.79, p = 0.0027)
Lee et al. (2019)	4942 patients	CP scores not available	VKA n = 1827 DOACs n = 3115	DOACs vs VKA Hospitalisation for major bleeding HR 0.622, 0.442–0.870, p = 0.005 Hospitalisation for GI bleeding HR 0.764, 0.496–1.168, p = 0.212	Patients with cirrhosis (n = 768) DOACs vs VKA Hospitalisation for major bleeding HR 0.591 (0.304–1.120) Hospitalisation for GI bleeding HR 0.790 (0.349–1.749)
Goriacko and Veltri (2018)	233 patients	CP^a A/B/C% DOACs: 64/35/1 VKA: 35/59/6	VKA n = 158 DOACs n = 75 (11 apixaban, 35 dabigatran, 29 rivaroxaban)	All-cause bleeding DOACs vs VKA: 8.4 vs 8.8% (HR 0.9, 0.4–1.8) Major bleeding DOACs vs VKA: 3.3 vs 3.9% (HR 0.8, 0.3–2.5)	MELD score and previous bleeding independent predictors of bleeding on anticoagulation
Pastori et al. (2018)	129 patients	LF as defined by FIB-4 > 3.25	VKA n = 77 DOACs n = 52	VKA bleeding LF vs no LF: 27.3% vs 19.7% p = 0.027 VKA major bleeding LF vs no LF: 14.3% vs 5.6% p < 0.001 DOACs bleeding LF vs no LF: 5.8% vs. 9.5% p = 0.374	FIB-4 independently associated with major bleeding in VKA-treated patients (HR 3.075, 1.626–5.818, p = 0.001)
Choi et al. (2017)	465 patients	Mean CP VKA: 6.1^a No treatment: 7.6^a	VKA n = 113	Incidence of major bleeding VKA versus no treatment: 5.9% vs 2.6%, p < 0.05	VKA use independently associated with bleeding events HR 2.40 (1.32–5.12)
Lee SJ et al. (2015)	321 patients	Mean CP VKA: 7.5 No treatment: 7	VKA n = 148	Major bleeding VKA versus no treatment: aHR 1.87 (1.13–3.09)	Major bleeding VKA versus no treatment CP A: aHR 0.89 (0.48–1.66) CP B/C: aHR 2.98 (1.23–7.19)

Note: Warfarin versus no anticoagulation: OR 2.09 (95% CI 1.17–3.7, p = 0.0121). Warfarin versus DOAC: OR 2.13 (95% CI 1–4.54, p = 0.049).
Abbreviations: CP, Child-Pugh, DOACs, direct oral anticoagulant, LF, liver fibrosis; HD, hepatic decompensation; GI, gastrointestinal; MELD, Model for End Stage Liver Disease; VKA, vitamin K antagonist.
^aStatistically significant.

Table 3. Anticoagulation for portal vein thrombosis
Study publication year	PVT type	Patient population	Liver disease severity	Anticoagulation (n)	Bleeding	Comments
Naymagon et al. (2021)	Acute Non-neoplastic	214 patients	CP A/B/C% Anticoagulation: 24/49/27 No treatment: 24/45/31	86/214, 42 LMWH, 26 VKA, 18 DOACs	Anticoagulation vs none Major bleeding 20% vs 17% p = 0.5207 Variceal bleeding: 9% vs 10%	Anticoagulant use not independent predictor of major bleeding HR 1.29 (0.68–2.46, p = 0.4423)
Zhou et al. (2020)	Not specified	64 patients randomised 1:1	Intervention: CP 6.51 Control: CP 6.81	32/64, 1-month LMWH +5 months VKA	Intervention: 1 hematemesis Control: 0	CP score significantly improved in the intervention group at 6 months
Ai et al. (2020)	>30 days from diagnosis Non-neoplastic	80 patients	Intervention: CP 7.2 Control CP 7.4	40/80, 26 rivaroxaban, 14 dabigatran	DOACs: 3 episodes (1 hematuria, 1 hemoptysis, 1 melena) Control: 1 episode (melena)	No significant difference in bleeding rates between the two groups. Two patients with melena were hypocoagulable on thromboelastography
Pettinari et al. (2019)	Non-neoplastic	182 patients	CP A/B/C% Anticoagulation: 53/41/6 No treatment: 37/44/19	81/182, 56 LMWH, 15 fondaparinux, 10 VKA	Anticoagulation: 16/81, 4 variceal No treatment: 22/101, 12 variceal p = 0.855	Anticoagulation independent predictor of improved survival HR 0.263 (0.105–0.657, p = 0.004)
Hanafy et al. (2019)	Acute non-neoplastic	80 patients randomised 1:1	Rivaroxaban: CP 6.4 VKA: CP 6.2	40 rivaroxaban, 40 VKA	GI bleeding Rivaroxaban: 0/40 VKA: 17/40	CP and MELD score improved significantly in the rivaroxaban group
Senzolo et al. (2018)	Within 6 months of diagnosis Non-neoplastic	149 patients	Median CP Anticoagulation: 7 No treatment: 7	92/149, 84 LMWH or fondaparinux, 32 VKA, 6 UFH	Major bleeding anticoagulant vs control 10.2 vs 12.3 per 100 patient-years	Only time on anticoagulation (HR 0.83, 0.69–0.99) independently predicted the risk of bleeding episodes
Scheiner et al. (2018)	Non-neoplastic	51 patients	CP A/B/C% 28/37/35	16/51, all LMWH + VKA	Patients with bleeding, anticoagulation versus no treatment 0% vs 23%, p = 0.045	Anticoagulated patients had significantly improved albumin, +9% vs −2%, p = 0.04
La Mura (2018)	Non-neoplastic	63 patients with LC and PVT, 160 patients without LC with VTE, 139 with LC without PVT	LC with PVT Mean CP 6.06* LC without PVT Mean CP 7.07*	63/63 PVT, 160/160 VTE LMWH + VKA	Major bleedinga* LC PVT: 24% No LC VTE: 7% Minor bleeding* LC PVT: 29% No LC VTE: 19%	1. No difference in rates of UGIB comparing LC with PVT on anticoagulation vs LC without PVT not on anticoagulation 2. Significantly better cumulative survival for 2 years in LC patients anticoagulated vs not
Nagaoki et al. (2018)	Non-neoplastic	50 patients	CP A/B/C Edoxaban: 15/5/0 VKA: 15/10/5	20 edoxaban 30 VKA	GI bleeding Edoxaban 15% VKA 7%	No patient in either group developed liver dysfunction
Cui et al. (2015)	Acute non-neoplastic	65 patients	Mean CP 7 in both groups	LMWH 1.5 mg/kg daily 34/65 1 mg/kg bid 31/65	Variceal bleeding 0 in both arms	No significant difference in new or worsening ascites, SBP, HCC

Abbreviations: CP, Child-Pugh; DOACs, direct oral anticoagulant; HCC, hepatocellular carcinoma; LC, liver cirrhosis; LMWH, low molecular weight heparin; MELD, Model for End Stage Liver Disease; PVT, portal vein thrombosis; SBP, spontaneous bacterial peritonitis; VKA, Vitamin K antagonist; VTE, venous thromboembolism.
*indicates statistically significant values.

Table 4. Traditional anticoagulants for portal vein thrombosis and impact on variceal bleeding
Study publication year	Patient population	Liver disease severity	Anticoagulation (n)	Bleeding	Comments
Wang et al. (2016)	64 patients	CP A/B/C Anticoagulation: 12/17/2 Control: 12/15/6	31/64, VKA	One episode of variceal bleeding in each arm	No significant difference in rates of overall GI bleeding, HE, or death
Cerini et al. (2015)	156 patients	CP A/B/C% Treated: 8/65/27 Untreated: 6/64/30	52/156, 38 VKA, 14 LMWH	Anticoagulation use not independently associated with 5-day treatment failure	Anticoagulation use not independently associated with 6-week mortality
Chung et al. (2014)	28 patients	CP A/B/C Anticoagulation: 3/10/1 Control: 7/6/1	14/28, VKA	1 episode of variceal bleeding in the control group	Two deaths in the VKA group, not related to anticoagulation
Cai et al. (2013)	11 patients	CP A/B 6/5	5/11, VKA	0 variceal bleeds in anticoagulated patients	4/5 patients treated CP A, 1 CP B
Senzolo et al. (2012)	56 patients	CP A/B/C Treated: 11/16/8 Untreated: 5/9/7	35/56 LMWH	Treated: 1 episode variceal bleeding Untreated: 5 episodes variceal bleeding	Non-significant difference in rates of variceal bleeding (p = 0.09). One death due to variceal bleed in an untreated arm
Amitrano et al. (2012)	53 patients	Mean CP^a Treated: 7 Untreated: 9	28/53, LMWH	1-year variceal rebleeding Treated: 3.6% vs Untreated: 8% p = 0.597	1-year mortality significantly lower in treated group 0 vs 16%, p = 0.043

Abbreviations: CP, Child-Pugh, HE, hepatic encephalopathy; GI, gastrointestinal; LMWH, low molecular weight heparin; VKA, vitamin K antagonist.
^aStatistically significant.

Table 5. Safety of antiplatelet agents in patients with cirrhosis and coronary artery disease and their effect on gastrointestinal bleeding
Study publication year	Patient population	Liver disease severity	Antiplatelet (n)	Bleeding	Comments
Wu et al. (2019)	914 patients with LC and MI 3656 patients without LC and MI	CP not available	4570/4570 (aspirin + clopidogrel)	LC vs no LC Major bleeding: 3.7% vs 2.9% (HR 1.23, 0.87–1.73) GI bleeding: 28% vs 20.2% (HR 1.49, 1.31–1.70)	No significant difference in rates of major and GI bleeding between single versus DAPT
Patel et al. (2018)	78 patients with decompensated cirrhosis undergoing transplant workup	No ASA n = 48 Mean MELD 16 ASA n = 30 Mean MELD 17	30/78 (aspirin)	Aspirin use not associated with overt GI or acute variceal bleeding	Aspirin users had a significant drop in mean platelet count after 6 months
Krill et al. (2017)	148 patients with LC and CAD	Mean CP Stent: 6.2 Medical management: 6.1	Aspirin^a Stent: 67/68 Medical: 64/80 DAPT^a Stent: 67/68 Medical: 4/80	GI bleeding Stent vs Medical 1 year: 22.1% vs 5% p = 0.003 2 years: 27.9% vs 5% p = 0.0002	Use of DAPT not independent predictor of GI bleeding
Russo et al. (2012)	48 patients with LC undergoing transplant workup, 16 stented on AP	Mean MELD 14.8 in both groups	13/16 aspirin, 1/16 clopidogrel	2 fatal variceal bleeds in stenting + AP group 2 non-fatal variceal bleeds in no stent + AP group (p = 0.86)	No significant difference in transfusion requirements during liver transplant between groups
Luo et al. (2012)	19,819 patients, 4013 with cirrhosis	CP not available	Amongst cirrhosis patients n = 283 aspirin, n = 42 clopidogrel or ticlopidine	Risk of peptic ulcer bleed Aspirin HR 0.87 (0.51–1.48, p = 0.600) Clopidogrel or Ticlopidine HR 0.30 (0.04–2.18, p = 0.236)	VKA use was not independently associated with the risk of peptic ulcer bleed
De Lédinghen et al. (1999)	200 patients with cirrhosis, 125 cases, 75 controls	CP A/B/C Cases: 24/39/52 Controls: 9/31/35	Aspirin alone or with other NSAID^a Cases 21/125 (17%) 3/75 (4%)	Risk of portal hypertensive bleeding Aspirin vs none OR 4.7, 1.3–17.3, p = 0.020	In patients with grade 1 varices, aspirin did not significantly increase the odds of bleeding

Abbreviations: AP, antiplatelet; CAD, coronary artery disease; CP, Child-Pugh; DAPT, dual antiplatelet therapy; GI, gastrointestinal; LC, liver cirrhosis; MELD, model for end stage liver disease; MI, myocardial infarction; NSAID, nonsteroidal anti-inflammatory drug; VKA, vitamin K antagonist.
^aStatistically significant.