From the Journals

Dabigatran, rivaroxaban linked to slight increase in GI bleeding risk

Novel oral anticoagulants (NOACs) receive a lot of press now. In randomized controlled trials (RCTs) comparing NOACs to warfarin for prevention of strokes and thromboembolism in atrial fibrillation (AF) and venous thromboembolism (VTE), fewer thromboembolisms are reported, but risks of gastrointestinal bleeding vary. To expand analyses for gastrointestinal bleeding, several systematic reviews and meta-analyses are reported, including this one by CS Miller et al. Their goals were to delineate risks of gastrointestinal bleeding for different NOACs compared with warfarin. What can GI clinicians now recommend about gastrointestinal bleeding for patients requiring anticoagulants? While we lack RCTs to give the highest quality of evidence about GIB as a primary outcome, conclusions now depend on the weight of evidence from recent secondary data analyses and I have some recommendations. First, although there may be differences among NOACs in risks of bleeding, all are likely to increase the risk of GI bleeding, comparable with warfarin. Some report that dabigatran and rivaroxaban have a higher risk of GI bleeding than other NOACs or warfarin, but differences are small. Second, some patients who need NOACs/warfarin have increased risks of ulcer bleeds including elderly patients and those with a history of upper GI bleeding, renal or hepatic impairment, low body weight, and concomitant antiplatelet agents. Such high-risk patients warrant treatment with a proton pump inhibitor or histamine2-receptor agonists for primary prevention while on anticoagulants. Finally, for patients with severe ulcer bleeding who require anticoagulation, warfarin or NOACs should be restarted after successful endoscopic hemostasis and proton pump inhibitors, usually within 3-5 days.

Dr. Jensen is professor of medicine at the University of California, Los Angeles; associate director of the CURE: DDRC, where he directs the Human Studies Core; a full-time staff physician in the UCLA division of digestive diseases; and a part-time staff physician in the GI section of the VA Greater Los Angeles Healthcare Center.



Compared with conventional anticoagulants, both dabigatran and rivaroxaban conferred small but statistically significant increases in the risk of major gastrointestinal bleeding in a systematic review and meta-analysis of randomized trials reported in the November issue of Clinical Gastroenterology and Hepatology. (doi: 10.1016/j.cgh.2017.04.031)

But other novel oral anticoagulants (NOACs) showed no such effect compared with warfarin, aspirin, or placebo, reported Corey S. Miller, MD, of McGill University, Montreal, and his associates. “The potentially increased risk of GI bleeding associated with dabigatran and rivaroxaban observed in some of our subgroup analyses merits further consideration,” they wrote.

The NOACs (also known as non–vitamin K antagonist oral anticoagulants) help prevent stroke in patients with atrial fibrillation and prevent and treat venous thromboembolism. However, large AF trials have linked all except apixaban to an increased risk of major GI bleeding compared with warfarin. Dabigatran currently is the only NOAC with an approved reversal agent, “making the question of GI bleeding risk even more consequential,” the authors wrote.

They searched the MEDLINE, EMBASE, Cochrane, and ISI Web of Knowledge databases for reports of randomized trials of NOACs for approved indications published between 1980 and January 2016, which identified 43 trials of 166,289 patients. Most used warfarin as the comparator, but one study compared apixaban with aspirin and six studies compared apixaban, rivaroxaban, or dabigatran with placebo. Fifteen trials failed to specify bleeding sources and therefore could not be evaluated for the primary endpoint, the reviewers noted.

In the remaining 28 trials, 1.5% of NOAC recipients developed major GI bleeding, compared with 1.3% of recipients of conventional anticoagulants (odds ratio, 0.98; 95% confidence interval, 0.80-1.21). Five trials of dabigatran showed a 2% risk of major GI bleeding, compared with 1.4% with conventional anticoagulation, a slight but significant increase (OR, 1.27; 95% CI, 1.04-1.55). Eight trials of rivaroxaban showed a similar trend (bleeding risk, 1.7% vs. 1.3%; OR, 1.40; 95% CI, 1.15-1.70). In contrast, subgroup analyses of apixaban and edoxaban found no difference in risk of major GI bleeding versus conventional treatment.

Subgroup analyses by region found no differences except in Asia, where NOACs were associated with a significantly lower odds of major GI bleeding (0.5% and 1.2%, respectively; OR, 0.45; 95% CI, 0.22-0.91).

Most studies did not report minor or nonsevere bleeds or specify bleeding location within the GI tract, the reviewers noted. Given those caveats, NOACs and conventional anticoagulants conferred similar risks of clinically relevant nonmajor bleeding (0.6% and 0.6%, respectively), upper GI bleeding (1.5% and 1.6%), and lower GI bleeding (1.0% and 1.0%).

A post hoc analysis using a random-effects model found no significant difference in risk of major GI bleeding between either rivaroxaban or dabigatran and conventional therapy, the reviewers said. In addition, the increased risk of bleeding with dabigatran was confined to the RELY and ROCKET trials of AF, both of which exposed patients to longer treatment periods. Dabigatran is coated with tartaric acid, which might have a “direct caustic effect on the intestinal lumen,” they wrote. Also, NOACs are incompletely absorbed across the GI mucosa and therefore have some anticoagulant activity in the GI lumen, unlike warfarin or parenteral anticoagulants.

The reviewers disclosed no funding sources. Dr. Miller and another author reported having no conflicts of interest. One author received research grants and speaker honoraria from Boehringer Ingelheim Canada, Bayer Canada, Daiichi Sankyo, Bristol Myers Squibb, and Pfizer Canada; another author disclosed serving as a consultant to Pendopharm, Boston Scientific, and Cook.

Next Article: