Hepatic Findings in Long-Term Clinical Trials of Ximelagatran

Objective: In clinical trials, the efficacy and safety of the oral direct thrombin inhibitor ximelagatran have been evaluated in the prevention or treatment of thromboembolic conditions known to have high morbidity and mortality. In these studies, raised aminotransferase levels were observed during long-term use (>35 days). The aim of this analysis is to review the data regarding these hepatic findings in the long-term trials of ximelagatran.

Patients and methods: The prospective analysis included 6948 patients randomised to ximelagatran and 6230 patients randomised to comparator (warfarin, low-molecular weight heparin followed by warfarin or placebo). Of these, 6931 patients received ximelagatran for a mean of 357 days and 6216 patients received comparator for a mean of 389 days. An algorithm was developed for frequent testing of hepatic enzyme levels. A panel of four hepatologists analysed all cases of potential concern with regard to causal relation to ximelagatran treatment using an established evaluation tool (Roussel Uclaf Causality Assessment Method [RUCAM]).

Results: An elevated alanine aminotransferase (ALT) level of >3 × the upper limit of normal (ULN) was found in 7.9% of patients in the ximelagatran group versus 1.2% in the comparator group. The increase in ALT level occurred 1–6 months after initiation of therapy and data were available to confirm recovery of the ALT level to <2 × ULN in 96% of patients, whether they continued to receive ximelagatran or not. There was some variability in the incidence of ALT level elevation between indications, those with simultaneous acute illnesses (acute myocardial infarction or venous thromboembolism) having higher incidences. Combined elevations of ALT level of >3 × ULN and total bilirubin level of >2 × ULN (within 1 month of the ALT elevation), regardless of aetiology, were infrequent, occurring in 37 patients (0.5%) treated with ximelagatran, of whom one sustained a severe hepatic illness that appeared to be resolving when the patient died from a gastrointestinal haemorrhage. No death was observed directly related to hepatic failure caused by ximelagatran.

Conclusion: Treatment with ximelagatran has been associated with mainly asymptomatic elevation of ALT levels in a mean of 7.9% of patients in the long-term clinical trial programme and nearly all of the cases occurred within the first 6 months of therapy. Rare symptomatic cases have been observed. An algorithm has been developed for testing ALT to ensure appropriate management of patients with elevated ALT levels. Regular ALT testing should allow the clinical benefits of ximelagatran to reach the widest population of patients while minimising the risk of hepatic adverse effects.

No sources of funding were used to assist in the preparation of this study. W.M. Lee; D. Larrey. R. Olsson and J.H. Lewis are consultants to AstraZeneca and comprise the Data and Safety Monitoring Board for Exanta®. M. Keisu; L. Auclert and S. Sheth are employed by AstraZeneca. Writing assistance was provided by PAREXEL MMS.

Authors and Affiliations

1. University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Blvd, Dallas, TX 75390-9151, USA

   William M. Lee

2. Service Hépato-Gastroentérologie, CHU Montpellier, Hôpital Saint Eloi and INSERM U632, Montpellier, France

   Dominique Larrey

3. Sahlgrenska University Hospital, Gothenburg, Sweden

   Rolf Olsson

4. Georgetown University Medical Center, Washington, District of Columbia, USA

   James H. Lewis

5. AstraZeneca R&D Mölndal, Mölndal, Sweden

   Marianne Keisu & Laurent Auclert

6. AstraZeneca LP, Wilmington, Delaware, USA

   Sunita Sheth

Corresponding author

Correspondence to William M. Lee.

Appendix

Patient 1

An 80-year-old man with AF began treatment with ximelagatran 36mg twice daily on 11 June 2001. His medical history included hyperlipidaemia, hydronephrosis and urinary retention, fibromyalgia previously treated with prednisone, coronary artery disease treated with bypass grafting and right colon cancer not in evolution. Concomitant medications included metoprolol, digoxin and tamsulosin, which were all taken for several months prior to the start of ximelagatran therapy.

An 80-year-old man with AF began treatment with ximelagatran 36mg twice daily on 11 June 2001. His medical history included hyperlipidaemia, hydronephrosis and urinary retention, fibromyalgia previously treated with prednisone, coronary artery disease treated with bypass grafting and right colon cancer not in evolution. Concomitant medications included metoprolol, digoxin and tamsulosin, which were all taken for several months prior to the start of ximelagatran therapy.

On day 100, the patient’s ALT level was at 1440 IU/L (30 x ULN), ALP level was just above normal and total bilirubin level was nearly 2 × ULN (mainly unconjugated). The search for common causes of hepatic disease was negative (tested by viral serology, immunological markers, and imaging of the liver and abdomen). A liver biopsy on day 108 demonstrated “severe active hepatitis with hepatocyte necrosis, areas of collapse and marked bile ductular proliferation consistent with acute submassive necrosis”. Transaminase levels peaked on day 108 and then decreased on day 115. ALP levels peaked at 198 IU/L (1.5 × ULN) on day 115. Total bilirubin level was 8 × ULN on day 114 and remained around this level for a month. On day 112, the patient’s synthetic liver function started to deteriorate, as was shown by an increase in the prothrombin time/INR in the absence of anticoagulant therapy and decreased serum albumin levels. The investigator labelled the adverse event as life-threatening hepatic failure and readmitted the patient to hospital on day 113, on which day the patient was jaundiced with no other symptoms and with a normal neurological examination. The patient began treatment with glucocorticoids, vitamin K and ranitidine.

At a visit on day 140, the patient complained of increasing fatigue. Liver enzyme levels continued to improve. The dosage of prednisone was decreased to 15mg daily. Profound fatigue worsened, with no evidence of encephalopathy. However, in the next couple of days the patient’s condition quickly deteriorated and he developed ascites, significant lower-extremity oedema and oliguria, and was found dead in his home on day 145.

The main finding at autopsy was a large duodenal ulcer (2.5cm) with erosion into the pancreas. The liver was small, friable and diffusely mottled, which was suggestive of severe diffuse hepatic necrosis. Microscopically, there was extensive liver necrosis with hepatocyte dropout and bile duct proliferation, which was similar to that seen in the previous biopsy. A significant amount of hepatic parenchyma remained. Tissue architecture showed early signs of resolution of the inflammation in comparison with the previous biopsy. The cause of death was an acute GI bleed from a duodenal ulcer, with a coagulopathic state from hepatic injury contributing to death.

Comment: The literature on the use of corticosteroids in the setting of suspected liver failure is difficult to interpret, although most hepatologists do not regard corticosteroids as being overtly beneficial compared with the risks of opportunistic infection developing and the risk of corticosteroid-induced GI bleeding (hence, the use of ranitidine).

This individual was not described as ever having developed encephalopathy and his INR following the discontinuation of ximelagatran peaked at approximately 2.0, but started decreasing after he received fresh frozen plasma. Therefore, he did not fit a strict definition of having acute liver failure.

At the time of his fatal GI bleed, the patient’s INR had actually declined further, along with his liver enzyme levels (last values ALT 219 IU/L [4 × ULN], AST 175 IU/L [6 × ULN], INR 1.5). The histology of the liver at the time of necropsy indicated that there was some regeneration and less inflammation than had been seen on the initial liver biopsy weeks earlier.

It is not known if this patient had an underlying ulcer history, but it is reasonable to conclude that corticosteroid therapy may have contributed to bleeding from a duodenal ulcer, which was the fatal event. Whether or not the patient would have eventually recovered from his hepatic necrosis remains unknown, although there are clinical suggestions that he may have recovered since he never fit the true definition of acute fulminant liver failure (given the absence of encephalopathy and coagulopathy). As ximelagatran has a short half-life, its effects on prothrombin time should have been minimal after just a few days. The peak INR of 2.0 is likely to reflect the underlying hepatic injury, but in cases of irreversible liver failure, the prothrombin time and INR continue to rise despite efforts at reversing the coagulopathy.

Patient 2

A 77-year-old man with AF began treatment with ximelagatran 36mg twice daily on 13 August 2001. His medical history included a cholecystectomy 2 years earlier, duodenal ulcer, a bleeding bladder (in 2000), psychosis, gout, alcohol abuse, osteoarthritis, pulmonary insufficiency, abdominal aortic aneurysm repair (April 2001), arterial hypertension, cardiomyopathy and coronary disease. Concomitant medications included carvedilol and ramipril.

His baseline ALT level was normal, but at day 63 was 216 IU/L (>4.5 × ULN). Bilirubin level was normal. Requests to have him return for repeat laboratory tests 3 and 7 days later went unheeded by the patient. He took his last dose of ximelagatran on day 80. However on day 81 he developed abdominal pain and haematochezia and was admitted for massive upper GI bleeding and severe hypotension (76/45mm Hg) that was determined to be from an anastomotic ulcer from his previous Billroth II anastomosis for ulcer disease.

On admission, his haemoglobin level was 7 g/dL, ALT 569 IU/L (>11 × ULN), AST 629 IU/L (>15 × ULN), bilirubin 1.4 × ULN and he had an elevated INR of 3.4. The patient received a massive transfusion of blood products given in a resuscitation effort, including 19 units of packed red cells, 30 units of cryoprecipitate and 15 units of fresh frozen plasma and vitamin K. Within 24 hours of this (day 82) his coagulopathy had resolved temporarily, ALT level had decreased to 97 IU/L (1.9 × ULN), but total bilirubin level was 9.4 × ULN (50% unconjugated).

The patient developed respiratory failure and pulmonary oedema that did not respond well to furosemide. Vasopressors and fluids were given to sustain blood pressure and diltiazem to reduce a rapid heart rate. Synchronised cardioversion failed four times to restore sinus rhythm and shock persisted despite resuscitation efforts with fluids, red blood cells, fresh frozen plasma and platelets. Profound coagulopathy occurred, the abdomen became rigid, gastric suction yielded blood and support was withdrawn. The patient was judged to have died on day 82 as a result of massive haemorrhage. No autopsy was conducted.

Comment: The massive transfusion of blood products can explain the indirect hyperbilirubinaemia, which is related to red cell breakdown with massive overproduction of bilirubin. Evidence in support of haemolysis includes a nearly absent haptoglobin level, elevated lactate dehydrogenase level and the exclusion of disseminated intravascular coagulation. This patient is listed as a ‘Hy’s rule’ patient because of this elevation in bilirubin level, which occurred within 30 days of the ALT level rising. It is possible that the bleeding was precipitated by the use of ximelagatran in a patient with a silent ulcer and prior ulcer history, but it is not at all clear whether the patient had irreversible hepatic injury from the drug.

The case has been adjudicated as,possibly related, to the study drug because we do not have enough information to exclude the possibility that the drug was involved in the initial rise in ALT level. No biopsy or autopsy was ever performed. However, the elevated AST >ALT levels on admission, in the setting of marked hypotension from bleeding, suggests shock liver was present (possible liver ischaemia), although the initial elevation in ALT levels may have been drug-related. The patient died of multisystem organ failure following the massive haemorrhage. The validity of invoking Hy’s rule in this case appears tenuous given the indirect hyper-bilirubinaemia that most likely resulted from massive transfusions. As a result, its predictive value for signalling acute liver failure is also highly confounded.

Patient 3

A 73-year-old man with VTE began treatment with ximelagatran 36mg twice daily on 10 January 2002. He had a history of lupus (anti-nuclear autoantibodies 1: 5120) and had underlying chronic hepatitis B infection (hepatitis B surface antigen positive and hepatitis B e-antigen positive) with elevated ALT levels of 93 IU/L (1.94 × ULN) and AST levels of 123 IU/L (2.24 × ULN) at baseline, prior to initiating ximelagatran therapy. He also had a history of gastric ulcer and diabetes mellitus. This patient was receiving immunosuppressive therapy (azathioprine and prednisone) for his underlying lupus and was not receiving treatment or prophylaxis for his hepatitis B.

On day 9 the patient had mild transaminase level elevation. On day 18 he was hospitalised based on positive serology for hepatitis B. His ALT level was 327 IU/L (7.7 × ULN) and his AST level was 481 IU/L (8.7 × ULN.) On day 24, ximelagatran was discontinued and the patient,s condition was good with no sign of liver failure. However, the patient appeared to develop progressive hepatitis by day 26: ALT 518 IU/L (14.8 × ULN), AST 593 IU/L (16.9 × ULN) and total bilirubin 40 mg/L (3.6 × ULN), rising to 268 mg/L (24.4 × ULN). The patient’s condition subsequently deteriorated and he developed ascites, which was followed by signs of encephalopathy. He had multiple gastric ulcers detected on endoscopy for abdominal pain and melena. Rapid deterioration led to coma, support was withdrawn and he died 2 days later. No autopsy was performed.

Comment: Under the current proposed management algorithm, this patient should not have received ximelagatran because of his hepatic disease. The development of progressive hepatitis at day 26 is likely to be the result of acute reactivation of underlying hepatitis B infection in a patient receiving corticosteroid/azathioprine immunosuppression. The patient did have evidence of active viral replication with hepatitis B e-antigen being positive. This was not acute hepatitis B, as the hepatitis B core IgM antibody was negative and the total core was positive. There is no information about whether this could have been a superinfection with Delta virus. There is no reason to suspect ximelagatran in the presence of immunosuppressive therapy.

Reprints and permissions