The risk for serious upper gastrointestinal (GI) bleeding is higher with some oral anticoagulants (OAC) than others, but protection offered by coadministration of a proton-pump inhibitor (PPI) varies among the different OACs, especially among patients at highest risk for the complication.
The findings came from a retrospective cohort analysis that used a newly developed scoring system for predicting the risk specifically of hospitalization for upper GI bleeding that incorporates dozens of patient risk factors.
The scoring system could potentially guide choice of OAC and whether to coprescribe a PPI in individual patients according to their propensity to develop GI bleeds, the researchers say.
“For patients at lowest risk, the need for PPI cotherapy and to think about which anticoagulant to use has far fewer clinical consequences, whereas for patients with the highest risk scores, these are really critical questions,” Wayne A. Ray, PhD, Vanderbilt University School of Medicine, Nashville, Tennessee, told theheart.org | Medscape Cardiology.
Although HAS-BLED bleeding risk scores can be applied in the same setting, they predict bleeding risk more broadly and don’t account for some risk factors for upper GI bleeding in particular, said Ray, who is lead author on the report published in the December 4 issue of the JAMA.
In practice, intracranial hemorrhage as a possible complication deservedly gets a lot of attention when a patient is considered for one oral anticoagulant or another, whereas the risk for upper GI bleeding may often be overlooked, Ray observed.
“But the risk of major gastrointestinal bleeding does need to be kept in mind, and for patients who are at high risk of that side effect, the choice of both PPI therapy and the specific anticoagulant can make a big difference.”
The new upper GI-bleed risk-stratification system was derived from a Medicare cohort of more than 1.6 million patients about to start an OAC primarily for nonvalvular atrial fibrillation (AF), although non-AF indications, such as prevention of deep-venous thrombosis, were also included.
The study then compared warfarin and direct oral anticoagulants (DOAC) to show how well PPI cotherapy protected against GI-bleed-hospitalization in patients with higher- and lower-risk scores.
In patients on an OAC without PPI cotherapy, the incidence of GI-bleed-hospitalization was highest for rivaroxaban (Xarelto, Bayer/Janssen Pharmaceuticals) and lowest for apixaban (Eliquis, Bristol-Myers Squibb/Pfizer), regardless of risk score. Incidences were intermediate for dabigatran (Pradaxa, Boehringer Ingelheim) and warfarin.
Also regardless of risk score, the incidence of upper GI-bleed hospitalization was lower for those on PPI cotherapy than for those not taking PPIs for each of the OACs. But PPI cotherapy protection was most pronounced in patients who started on dabigatran and least pronounced in patients prescribed rivaroxaban.
“We would suggest that for patients with elevated GI risk — for example with a history of GI bleed or peptic ulcer disease, or concurrent use of P2Y12 inhibitor or nonsteroidal anti-inflammatory agents — practitioners should consider a PPI when starting apixaban or rivaroxaban.”
In contrast to the abundance of randomized-trial data comparing DOACs with warfarin, Ray noted, there are none pitting any of the DOACs against each other. “Thus, selection of the individual anticoagulant is not obvious, and may be made on the basis of pharmacokinetic data, perceived adherence, or other factors.”
But for patients scoring the highest for upper GI-bleeding, he said, clinicians should consider that risk when choosing the anticoagulant. For example, although PPI cotherapy best protected against such bleeding in patients on dabigatran, “there is some concern that PPIs interfere with dabigatran’s anticoagulant effects; thus, apixaban plus PPI might be better choice for high-GI-bleed-risk patients.”
The study used Medicare data on 1,643,123 patients, accounting for 1,713,183 new episodes of OAC therapy from the start of 2011 through September 2015. It excluded data on use of edoxaban (Savaysa/Lixiana, Daiichi Sankyo), the last DOAC to be approved in the United States and little used during those cohort years.
During 754,389 person-years in the absence of PPI cotherapy, the adjusted incidence of hospitalization for upper GI bleeding was 115 per 10,000 person-years.
|All Risk Quartiles: Incidence Rate Ratio (IRR) for Upper GI Bleeding Hospitalization on OACs Without PPI Cotherapy, Rivaroxaban vs Other OACs*|
|OAC Comparison||IRR (95% CI)|
|*754,389 person-years on OAC without PPI cotherapy. Upper GI bleeding end points were primarily bleeding related to esophagitis, peptic ulcer disease, and gastritis.|
For patients on rivaroxaban but not on PPI cotherapy, the incidence was 144 per 10,000 person-years, significantly greater than the 120, 113, and 73 per 10,000 person-years incidences for dabigatran, warfarin, and apixaban, respectively. The incidence for apixaban was significantly lower than that for either dabigatran or warfarin.
|All Risk Quartiles: Incidence Rate Ratio (IRR) for Upper GI Bleeding Hospitalization on OACs, On vs Off PPI Cotherapy*|
|Oral Anticoagulant||IRR (95% CI)|
|*Upper GI bleeding end points were primarily bleeding related to esophagitis, peptic ulcer disease, and gastritis.|
The incidence of upper GI bleeding hospitalization was consistently lower with than without PPI cotherapy across all risk-score quartiles, and also within the top risk quartile. Within that top quartile, the effect of PPI cotherapy on incidence was strongest for dabigatran and weakest for rivaroxaban.
|Highest-Risk Quartile: Incidence Rate Ratio (IRR) for Upper GI Bleeding Hospitalization on OACs, On vs Off PPI Cotherapy*|
|OAC||IRR (95% CI)|
|Upper GI bleeding endpoints were primarily bleeding related to esophagitis, peptic ulcer disease, and gastritis.|
The effects of PPI cotherapy on incidence of upper GI bleeding for the different DOACs in relation to the new risk scoring system are novel observations. But, Ray observed, it’s well recognized that rivaroxaban, more so than (or in contrast to) other DOACS, is associated with an increased risk of bleeding complications compared with warfarin.
And that insight has apparently worked its way well into clinical practice. “We found that patients who were actually at highest risk were being channeled away from rivaroxaban. So I think the folks out in the field already have a sense of this. What that says about our findings is that, if anything, we might be underestimating the difference between rivaroxaban” and the other DOACs.
The study was supported in part by a grant from the National Heart, Lung, and Blood Institute, and by grants to a coauthor from the National Institute of Arthritis and Musculoskeletal and Skin Diseases and the Rheumatology Research Foundation Career Development K Supplement. Neither Ray nor the other authors had disclosures.
JAMA. 2018 Dec 4;320(21):2221-2230. Full text