Acute liver failure after changing oral anticoagulant from apixaban to rivaroxaban
- 1 Department of General Medicine, Monash Medical Centre, Clayton, Melbourne, Victoria, Australia
- 2 Department of Nephrology, Monash Medical Centre, Clayton, Melbourne, Victoria, Australia
- Correspondence to Dr Vikram Rao; vikramaditya.rao@monashhealth.org
Abstract
Rivaroxaban is a commonly used anticoagulant agent for treatment and prevention of thromboembolism. There are case reports demonstrating an association between its use and drug-induced liver injury. However, this has not been reported in a patient who previously tolerated apixaban. An 88-year-old man presented to hospital with worsening lethargy, jaundice and vomiting. He had severely elevated liver transaminases, an abnormal coagulation profile and elevated bilirubin in keeping with acute liver injury. This is in the context of having had his anticoagulation medication switched from apixaban to rivaroxaban 2 weeks prior. The patient recovered well after cessation of rivaroxaban, suggesting that it was the likely offending agent. The mechanism of rivaroxaban-induced liver injury remains to be investigated. Drug-induced liver injury should be discussed and monitored for as a potential adverse reaction when commencing rivaroxaban, even if a patient has previously tolerated a drug of the same class.
Background
Rivaroxaban is a direct oral anticoagulant (DOAC) drug, used for treatment and prevention of deep venous and pulmonary thromboembolism.1 It is also commonly used to reduce the incidence of stroke in patients with atrial fibrillation. Randomised control trials have shown rivaroxaban to be non-inferior in efficacy and safety for the above indications when compared with the stalwart choice of warfarin.2 3 Though there is no significant difference in bleeding risk between the two, rivaroxaban can be administered without the need for ongoing laboratory and dose monitoring.1 3 This is due to its well understood pharmacological profile. It works by directly inhibiting factor Xa, an enzyme involved in the coagulation pathway. It is rapidly absorbed and has as a high bioavailability.4 Rivaroxaban has a dual elimination pathway, with approximately two-thirds of the drug being metabolised by hepatic cytochrome P450 enzymes; and the remaining one-third being eliminated unchanged in the urine.4
Bleeding is the most documented side-effect of rivaroxaban therapy. However, emerging postmarketing reports have increasingly shown an increased incidence of drug-induced liver injury (DILI) and acute liver failure (ALF). A 2018 systematic review reported 26 cases of acute or subacute liver damage due to rivaroxaban between the years 2006–2018, with a median time of onset of 15 days after initiation of the drug.5 In 2015, after 2 years of postmarketing surveillance of the US Food and Drug Administration Adverse Event Report System, 146 reports of DILI had been submitted for rivaroxaban, but only one had been submitted for apixaban.6 It appears that rivaroxaban is associated with a greater risk of DILI compared with apixaban; however, ongoing comparative postmarketing surveillance is needed to see whether this trend is preserved over a longer time-period.
Below, we present a case in which a patient developed ALF shortly following initiation of rivaroxaban, having previously tolerated apixaban without similar incident.
Case presentation
An 88-year-old man was initially admitted to hospital for a probable transient ischaemic attack (TIA) with acute-onset right arm and leg weakness. This was in the setting of subtherapeutically treated atrial fibrillation with apixaban and a 70% left carotid bulb stenosis, as demonstrated on a previous perfusion scan.
CT perfusion and angiography scans showed no evidence of acute ischaemic or haemorrhagic stroke. The patient’s limb weakness subsequently improved and his lower limb neurological examination was unremarkable. During this workup, it was established that the prescribed dose of apixaban of 2.5 mg two times per day, was likely subtherapeutic. The rationale for this dosing was unclear. Thus, the impression for the presentation was a possible TIA in the setting of inadequate anticoagulation. The patient also described a 7 kg weight loss in the preceding 4 months and was found to be anaemic with a haemoglobin of 84 g/L. A CT scan of the patient’s abdomen and pelvis demonstrated mild prominence of the gastro-oesophageal junction and suspected mural thickening of the rectum. A gastroscopy found moderate gastritis and oesophagitis, but no active bleeding was identified. A haemolytic screen was negative. The patient’s anticoagulant was then changed to 15 mg of rivaroxaban daily, although the reason for this change as well as the use of an adjusted dose is unclear on review of documentation. The patient was discharged 8 days later with a plan for an outpatient colonoscopy. At the time of discharge, the patient’s transaminases were normal (alanine aminotransferase (ALT) 23 U/L, GGT 46 U/L, alkaline phosphatase (ALP) 307 U/L, bilirubin 28 μmol/L, international normalised ratio (INR) 1.9, Ferritin 740 μg/L) he was not thrombocytopenic (161×109/L) and had stable kidney function (estimated glomerular filtration rate 70–80 mL/min, creatinine 80–90 mcmol/L). He continued his usual medications of atorvastatin and verapamil, which were both longstanding.
Two weeks later, the patient represented with nausea, vomiting and worsening lethargy. These symptoms had progressively worsened over the course of the week leading up to his presentation. He was afebrile and haemodynamically stable on presentation. He was jaundiced but without evidence of hepatic flap or confusion.
Investigations
Laboratory investigation revealed a significant mixed hyperbilirubinaemia (total bilirubin 1903 μmol/L, unconjugated 115 bilirubin μmol/L/L), transaminase rise (ALT 1940 U/L, AST 2304 U/L, ALP 360 U/L), elevated ferritin (26 218 μg/L) as well as thrombocytopenia and significantly elevated INR of 14 and APTT of 57. A rivaroxaban level taken on admission was found to be 117.1 ng/mL. An admission COVID-19 swab was found to be negative. The patient was once again found to be anaemic, with a haemoglobin of 67 g/L. Again, there was no evidence of acute bleeding. On further questioning, there was no prior history of chronic liver disease, alcohol abuse or illicit drug abuse. Blood alcohol, paracetamol levels and viral hepatitis serology panel were negative. Screening for autoimmune liver disease was also negative. A liver ultrasound showed moderate hepatic steatosis, which was also noted on his prior CT abdomen, with no evidence of cirrhosis or portal hypertension and no focal liver lesions.
Treatment
All the patient’s medications were with held, and he was treated with vitamin K, multiple units of fresh frozen plasma and packed red blood cells. The patient was also given an n-acetylcysteine infusion based on previous study findings showing a mortality benefit and reduced hospital stay when used in non-paracetamol DILI.7
Outcome and follow-up
By day 2, the patient’s liver enzymes began to downtrend, his INR reduced to 1.4, APTT improved to 38 and his haemoglobin incremented to 116 g/L. He also reported feeling symptomatically improved. As illustrated in figures 1 and 2, there was a downward trend in liver enzymes and bilirubin over the following week (1-week postrivaroxaban cessation: ALT 397 U/L, total bilirubin 82 μmol/L, AST 142 U/L, ALP 282 U/L, ferritin 3470 μg/L).
Trend in liver function tests during patient admission. rivaroxaban likely ceased 1–2 days prior and definitively ceased on admission. there was no previous aspartate level prior to this admission, ALT on commencement of rivaroxaban was 23 U/L
Trend in total bilirubin during patient admission. rivaroxaban likely ceased 1–2 days prior and definitively ceased on admission.
Despite the resolution of the patient’s ALF, he had an ongoing thrombocytopenia and a progressive anaemia. A repeat haemolytic screen was negative (haptoglobin 0.9, nil urinary haemosiderin, retics 1.5, DAT negative). A blood film during his admission showed a leucoerythroblastic shift. A bone marrow aspirate demonstrated markedly fibrotic marrow with distorted architecture as well as a marked increase in distorted megakaryocytes. In addition, his aforementioned CT abdomen demonstrated an enlarged spleen of 12.2 cm. Finally, he was found to have a positive JAK2 mutation in V617F. Thus, the patient was given an additional diagnosis of primary myelofibrosis. In retrospect, this diagnosis may have contributed to the patient’s initial TIA presentation, despite being anticoagulated. In terms of anticoagulation for his atrial fibrillation, the patient was discharged on long-term low-dose enoxaparin rather than reverting to apixaban.
Discussion
DOACs are increasingly becoming the oral anticoagulant of choice for the treatment and prevention of thromboembolic events in high-risk patients. However, emerging case report evidence suggests an increased risk of DILI associated with the use of DOACs, particularly rivaroxaban.8 9 These have been further substantiated by reports from worldwide pharmacovigilance and postmarketing databases. In addition to the US study mentioned above, a 2016 report by European pharmacovigilance database, Eudravigilance, described 763 cases of varying hepatotoxicity associated with rivaroxaban.10 As well, a 2017 prospective study following 113 717 patients newly commenced on either warfarin or various DOACs over a 12-month period found that, compared with warfarin, the risk of DILI was higher for rivaroxban (0.88, 95% CI 0.75 to 1.03) than apixaban (0.70, 95% CI 0.50 to 0.97).11 Our report presents the case of a patient presenting with DILI and ALF after commencing rivaroxaban, while having previously tolerated apixaban.
In the above case, the patient presented to hospital with hepatocellular injury 15 days after being discharged with a new medication, rivaroxaban. The Rousself Uclaf Causality Assessment Method was used for causality risk assessment of DILI in this case (table 1).12 The patient had a score of 9 out of 14, which indicates that it is ‘highly probable’ that rivaroxaban was the causative agent of the patient’s liver injury and ALF.12 As well, the timeline of liver injury shortly after initiation and that the patient rapidly improved on cessation of the drug lends further support to the diagnosis of rivaroxaban-induced liver injury.
RUCAM assessment of the case patient
RUCAM criteria | Patient’s Score |
Pattern of Injury | R>5, hepatocellular injury |
Time to onset | 5–90 days after initiation, 2 points |
Biochemical course | Alanine aminotransferase (ALT) fall by >50% within 8 days after cessation of drug, 3 points, highly suggestive |
Risk factors | >55 years old, occasional EtoH use, 1 point |
Concomitant hepatotoxins | Verapamil, however, incompatible time to onset (patient tolerating for many years), 0 points |
Exclusion of other causes | 2 points |
Previous information on drug’s hepatotoxicity | Not stated on drug’s label but previously documented, 1 point |
Response to readministration of drug | Not observed, 0 points |
Total Score, out of 14 | 9 points, highly Likely |
-
RUCAM criteria reprinted from Journal of Clinical Epidemiology, Vol 46, G.Danan, C.Benichou, Causality assessment of adverse reactions to drugs—I. A novel method based on the conclusions of international consensus meetings: Application to drug-induced liver injuries, 1323–1330, Copyright (1993), with permission from Elsevier.12
-
RUCAM, Rousself Uclaf Causality Assessment Method.
DILI was a diagnosis of exclusion in the above case. Viral hepatitis was ruled out by a negative viral hepatitis serology as well as negative Epstein-Barr and cytomegalovirus serologies. Paracetamol-induced hepatic injury was ruled out as the patient had a negative paracetamol level on admission. Mechanical causes, such as biliary obstruction or portal vein thrombosis was ruled out by a normal abdominal ultrasound. Rarer causes such as Wilson’s disease were ruled out with a negative immunological and vasculitic screen. The patient was haemodynamically stable throughout his admission, making the diagnosis of ischaemic hepatitis unlikely. The patient’s admission rivaroxaban level of 117.1 ng/mL falls in the range consistent with a plasma trough concentration (18–136 ng/mL) which is defined as a sample collected 20–28 hours after dosing.1 Given rivaroxaban’s half life of 11–13 hours in older adults,1 at least two half lives would have passed between the patient’s last dose and hospital admission. Thus, the DILI likely occurred a few days prior to presentation and the patient may have missed or not tolerated 1–2 doses of his rivaroxaban due to worsening nausea and vomiting. The timeline of liver injury shortly after initiation and that the patient improved on cessation of the drug lends further support to the diagnosis of rivaroxaban-induced liver injury.
Considering that the patient had previously been taking apixaban without issue, this suggests an alternate mechanism of injury that is independent of rivaroxaban’s action as a DOAC. The precise nature of this mechanism is currently unknown. A 2015 review of four past similar case reports, in which liver biopsies were obtained, showed a recurring trend of eosinophilic hepatic infiltrates.13 Similarly, a case report described a patient who developed a mild pruritic and purpuric rash of the trunk 1 week after commencing rivaroxaban, with an associated rise of AST and ALT to 209 and 115 U/L.14 The rash subsided a week after ceasing rivaroxaban, and the liver function test had normalised during this time as well. A skin biopsy demonstrated perivascular lymphocytes and eosinophilic infiltrates.14 This may fit with a proposed model for DILI: which posits that an offending drug acts as antigen as well as causes increased oxidative stress within the liver, triggering an exaggerated adaptive immune response which then leads to liver damage.15 The patient’s significantly worsened ferritin (a non-specific marker of inflammation) on admission appears to be in keeping with this model. However, this remains conjecture and requires further study and elaboration in both a laboratory and clinical setting.
The patient’s pre-existing non-alcoholic fatty liver disease (NAFLD) may have been a risk factor predisposing him to an increased risk of a DILI with rivaroxaban therapy. In the previously mentioned 2017 prospective study by Alonso et al, pre-existing liver disease was the strongest predictor of DILI hospitalisation in patients taking all DOACs.11 The reason for this increased risk is unclear but may be associated with the overproduction of reactive oxygen species (ROS) from increased lipolysis of fatty acids in the liver.16 This then leads to an increasingly pro-inflammatory environment in the liver, which in conjunction with the potential antigenic properties of rivaroxaban increases the risk of a subsequent immune response and DILI. Note that the patient had a pre-existing elevated ferritin of 741 two weeks prior to admission, which may be in keeping with this proinflammatory state. As well, increased ferritin production has previously been posited to be a protective response against oxidant stress and overexpression of ferritin has been shown to reduce accumulation of ROS in vitro.17 Once again, this model requires further investigation and elucidation.
The patient’s previously undiagnosed myelofibrosis may also have been a contributor to his subsequent liver injury. JAK2 is a tyrosine kinase that plays an important role in the cytokine signalling and production. An activating JAK2 mutation is associated with hyperactive cytokine signalling and a systemic proinflammatory state, which has been theorised to be a driving factor for the development of myeloproliferative disorders.18 This chronic proinflammatory state would also increase the risk of an immune reaction to the previously posited antigenic properties of rivaroxaban, resulting in a DILI. As well, myelofibrotic patients have been shown to have elevated levels of ROS, likely as a result of this chronic inflammatory state.19 Given that the liver is an organ that is known to be susceptible to ROS-related damage, this may have then exacerbated our case patient’s pre-existing NAFLD, increasing the risk of a subsequent DILI.
Learning points
-
This case, as well as other previously documented case reports, demonstrates a highly likely causal relationship between rivaroxaban and acute liver injury.
-
Patients with pre-existing liver disease appear to be at higher risk, and care should be taken when commencing these patients on rivaroxaban.
-
Liver function tests should be monitored when initially commencing on the drug, even if patients have tolerated other direct oral anticoagulants in the past.
-
Further studies are required to elucidate the idiosyncratic mechanism by which rivaroxaban causes liver injury.
Acknowledgments
RUCAM Criteria adapted from: Journal of Clinical Epidemiology, Vol 46, G.Danan,C.Benichou, Causality assessment of adverse reactions to drugs—I. A novel method based on the conclusions of international consensus meetings: Application to drug-induced liver injuries, 1323-1330., Copyright (1993), with permission from Elsevier.
Footnotes
-
Contributors VR is the main author of this case report and contributed to conception, planning, composition of the case report and also involved in reviewing the literature and finalising the manuscript. AM contributed to supervising and planning the composition of the case report and was also involved in reviewing the literature. She was involved in supervising and finalising the manuscript.
-
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
-
Competing interests None declared.
-
Provenance and peer review Not commissioned; externally peer reviewed.
- © BMJ Publishing Group Limited 2021. No commercial re-use. See rights and permissions. Published by BMJ.
References
Use of this content is subject to our disclaimer