Author + information
- Mark A. Crowther, MD, MSc∗ ( and )
- Wendy Lim, MD, MSc
- ↵∗Address for correspondence:
Dr. Mark A. Crowther, Department of Medicine, McMaster University, 1280 Main Street West, HSC 4V32, Hamilton Ontario L8S 4K1, Canada.
Anticoagulant agents are highly effective for the prevention and treatment of arterial and venous thrombosis. The aging population has resulted in increasing numbers of patients eligible for long-term anticoagulation for indications such as nonvalvular atrial fibrillation and venous thrombosis. Oral anticoagulant therapy started with the introduction of vitamin K antagonists (warfarin in North America, and others, including acenocoumerol and phenprocoumon, outside North America). Oral vitamin K antagonists are highly effective when optimally used, typically in the context of clinical trials and specialized anticoagulant clinics. Outside of these settings, anticoagulant control, as measured by time in the therapeutic range, is generally suboptimal. Even in warfarin-experienced patients, the mean time in the therapeutic range is <70% (1). Further, oral vitamin K antagonists are associated with major and life-threatening bleeding, including intracerebral bleeding. The limitations of oral vitamin K antagonists, coupled with the substantial population of patients who require such drugs for secondary prevention of stroke and recurrent venous thromboembolism, presented an opportunity for pharmaceutical manufacturers. As a result, the last 10 years have witnessed a revolution in oral anticoagulant therapy with the introduction of rationally designed drugs that are highly specific inhibitors of either thrombin or factor Xa. These drugs are collectively referred to by a variety of names including NOACs (novel or non-vitamin K oral anticoagulants) and DOACs (direct oral anticoagulants), the latter being the preferred term (2). Generally, DOACs have short half-lives, few clinically important interactions with food and drugs and are administered in fixed doses without dose adjustment. Currently, there are 5 such agents approved in the United States: dabigatran, rivaroxaban, apixaban, edoxaban, and betrixaban.
Unfortunately, the initial foray into rationally designed inhibitors of the coagulation cascade was unsuccessful as a result of hepatotoxicity, mirroring findings from other drug development programs (3). Ximelagatran was evaluated in the early 2000s for prevention of deep vein thrombosis (DVT) post-total knee arthroplasty, secondary prevention of DVT following treatment of acute DVT, and for stroke prevention in patients with nonvalvular atrial fibrillation (4). Ximelagatran was the oral prodrug of melagatran, a specific inhibitor of thrombin. It was the first orally available anticoagulant with predictable pharmacokinetics obviating routine laboratory monitoring and dose adjustment. Traditional safety models that were evaluated in the development program of ximelagatran failed to demonstrate evidence of hepatotoxicity—as did short-term exposure in orthopedic surgery prophylaxis studies (5). Longer-term exposure, however, demonstrated evidence of hepatotoxicity. After exposures ranging from 1 to 6 months, approximately 8% of patients treated with ximelagatran and 1% of patients treated with a comparator anticoagulant developed an elevation of the alanine aminotransferase (ALT) to >3 times the upper limit of normal (6). A 35-day extended prophylaxis post-orthopedic surgery study was stopped early due to a hepatotoxicity signal; in this study, 27 ximelagatran-treated patients developed an ALT >3 times the upper limit of normal with an additional 11 patients demonstrating an ALT increase after the end of study treatment. Of the patients with evidence of hepatic injury, 3 developed clinical symptoms of liver injury, and all 3 recovered (7). On the basis of these results, further development of ximelagatran ceased.
Although unexpected hepatotoxicity is an ongoing concern with any new drug, the experience with ximelagatran heightened concern about hepatotoxicity with the DOACs. This concern is magnified by the large number of patients eligible for extended-duration oral anticoagulant therapy; idiosyncratic hepatotoxicity occurring in even a tiny proportion of patients has the potential to have an impact on significant numbers at a population level, given the widespread use of extended-duration oral anticoagulant therapy. For example, it is estimated that up to 6.1 million Americans have atrial fibrillation, with the majority being candidates for oral anticoagulant therapy (8). Case reports of hepatotoxicity with DOACs are noteworthy because these may represent an initial signal of an unexpected, rare, and/or toxic reaction.
When examining adverse reactions to oral anticoagulants, there is a significant and pervasive bias that influences results. Warfarin has been used for decades. It has numerous drug and food interactions, has a narrow therapeutic window, and its response varies widely between patients and within an individual patient over time. Further, its major toxicity—bleeding—occurs with sufficient frequency in the population that clinicians have become accustomed to the complications of warfarin therapy. Reporting every hemorrhage or drug interaction occurring in a warfarin-treated patient would quickly overwhelm the regulators, pharmacists, and physicians. Consequently, adverse reactions to warfarin are rarely reported. By contrast, there is an understandable propensity to report complications in DOAC-treated patients in the interest of post-marketing surveillance and detection of rare side effects. Unless accounted for, this bias for reporting will result in apparent higher rates of adverse reactions with DOACs compared with warfarin.
In this context, the work of Douros et al. (9) in this issue of the Journal is of importance. Using a large and comprehensive administrative database from the province of Quebec, these authors examine the rates of serious liver injury in patients receiving either oral vitamin K antagonists or one of the DOACs. Of the patients in the study, just over 6% had a history of prior liver disease. The authors accumulated almost 69,000 person-years of follow-up, over which time 585 patients experienced a serious liver injury. The hazard ratio for serious liver injury in users of DOACs compared with users of vitamin K antagonists was 0.99 (95% confidence interval: 0.68 to 1.45); in patients with prior liver disease the hazard ratio was 0.68 (95% confidence interval: 0.33 to 1.37).
This study has important limitations. First, patients were only eligible if they were being treated for nonvalvular atrial fibrillation. Although this represents the largest group of patients who would be exposed to long-term oral anticoagulant therapy, there may be important differences in patients requiring anticoagulation for other indications. Second, allocation to therapy was not randomized, and allocation bias could account for the lack of observed hepatotoxicity. Third, the results are derived from a closely controlled, provincially run pharmaceutical program where there is oversight of drug prescription; whether these results can be extrapolated to a broader population with less stringent prescription criteria remains to be seen. Finally, the outcome of interest was determined using codes derived from electronic medical records and is thus subject to potential bias and errors implicit in such analyses.
The analysis had a rigorous statistical approach. The authors adjusted for known acetaminophen use, given both its prominent role in drug-induced liver injury and the high frequency of its use in the general population. A series of secondary analyses were used to support the overall hypothesis, including an analysis based on duration of therapy, individual type of DOAC, and accounting for the effect of age and sex. Eight additional sensitivity analyses were performed testing the overall hypothesis.
The development of the DOACs is a remarkable success story. These agents are highly effective, safe and are preferred by patients compared with oral vitamin K antagonists (10). However, vigilance is warranted, given the large number of patients who are treated with these agents and the extended duration of such treatment. Further, DOACs are not a panacea, and careful evaluation of their efficacy and safety for each major indication is required before their use. For example, DOACs are inferior to warfarin for the prevention of thrombotic complications of mechanical heart valves and are contraindicated in this setting (11).
Clinicians should be reassured by the results of this analysis. Outside of bleeding, which is a predictable complication of any anticoagulant, DOACs appear highly safe and have become the preferred agents for the prevention of thrombotic complications in many patients who have, or are at risk of, this complication.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Dr. Crowther has participated in data safety monitoring boards for Bayer and Daiichi; has received personal fees and/or has participated in advisory boards for Shionogi, Octapharma, and BMS Canada; has received institutional funding from Bayer and Leo Pharma; has received fees for educational material preparation and/or presentations for Bayer, Pfizer, and Alexion; and has individual stock ownership in Alynlam. Dr. Lim has served as a consultant and/or on advisory boards and received honoraria (for CME) from Leo Pharma, Pfizer, and Portola.
- 2018 American College of Cardiology Foundation
- Keisu M.,
- Andersson T.B.
- January C.T.,
- Wann L.S.,
- Alpert J.S.,
- et al.
- Douros A.,
- Azoulay L.,
- Yin H.,
- Suissa S.,
- Renoux C.
- Brekelmans M.P.,
- Kappelhof M.,
- Nieuwkerk P.T.,
- Nierman M.,
- Buller H.R.,
- Coppens M.