Accepted Manuscript Major Gastrointestinal Bleeding is Often Caused by Occult Malignancy in Patients Receiving Warfarin or Dabigatran to Prevent Stroke and Systemic Embolism From Atrial Fibrillation Kathryn F. Flack, MD, Desai Jay, MD, Jennifer M. Kolb, MD, Chatterjee Prapti, BA, Lars C. Wallentin, MD, PhD, Ezekowitz Michael, MD, PhD, Yusuf Salim, MD, DPhil, Connolly Stuart, MD, Reilly Paul, PhD, Brueckmann Martina, MD, Ilgenfritz John, MS, Aisenberg James, MD PII: DOI: Reference:
S1542-3565(16)30928-4 10.1016/j.cgh.2016.10.011 YJCGH 54957
To appear in: Clinical Gastroenterology and Hepatology Accepted Date: 1 October 2016 Please cite this article as: Flack KF, Jay D, Kolb JM, Prapti C, Wallentin LC, Michael E, Salim Y, Stuart C, Paul R, Martina B, John I, James A, Major Gastrointestinal Bleeding is Often Caused by Occult Malignancy in Patients Receiving Warfarin or Dabigatran to Prevent Stroke and Systemic Embolism From Atrial Fibrillation, Clinical Gastroenterology and Hepatology (2016), doi: 10.1016/ j.cgh.2016.10.011. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Major Gastrointestinal Bleeding is Often Caused by Occult Malignancy in Patients Receiving Warfarin or Dabigatran to Prevent Stroke and Systemic Embolism From Atrial Fibrillation Short Title: Anticoagulation and GI bleeding from cancer
RI PT
Authors: Flack, Kathryn F., MD1; Desai, Jay, MD1; Kolb, Jennifer M., MD 1; Chatterjee, Prapti, BA1; Wallentin, Lars C., MD, PhD2; Ezekowitz, Michael, MD, PhD3; Yusuf, Salim, MD, DPhil4; Connolly, Stuart, MD4; Reilly, Paul, PhD5; Brueckmann, Martina, MD 6,7; Ilgenfritz, John, MS8; Aisenberg, James, MD1
M AN U
SC
Affiliations and Addresses: 1. Department of Internal Medicine, Icahn School of Medicine at Mount Sinai, New York, NY. 2. Uppsala Clinical Research Center, Uppsala University, Uppsala, Sweden. 3. Sidney Kimmel Medical College, Thomas Jefferson University, Philadelphia, PA. 4. Population Health Research Institute, McMaster University and Hamilton Health Sciences, Hamilton, ON, Canada. 5. Boehringer Ingelheim Pharmaceuticals, Ridgefield, CT. 6. Boehringer Ingelheim GmbH&CoKG, Ingelheim, Germany. 7. Faculty of Medicine Mannheim,University of Heidelberg, Mannheim, Germany. 8. Ilgenfritz Consulting LLC, Conshohocken,PA.
TE D
Funding/Support: This study was supported by the Digestive Disease Research Foundation. The funding source was not involved in the study design, data acquisition or interpretation, or manuscript preparation and submission.
EP
Boehringer Ingleheim Inc (the sponsor of the RE-LY trial) provided the source documents to the investigators. The data analysis was performed independent of the sponsor. PR and MB are employees of Boehringer Ingleheim and reviewed and commented on the manuscript, but editorial control was maintained by the independent authors.
AC C
Abbreviations: Atrial fibrillation (AF); Colorectal cancer (CRC); Creatinine (Cr); Direct oral anticoagulant (DOAC); Fecal occult blood test (FOBT); Gastrointestinal (GI); Hemogobin (Hgb); Intensive care unit (ICU); International Society on Thrombosis and Hemostasis (ISTH); Major gastrointestinal bleeding (MGIB); Non-steroidal anti-inflammatory drugs (NSAIDs); Oral anticoagulant (OAC); Proton-pump inhibitor (PPI); Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial; Vitamin K antagonist (VKA). Correspondence: Kathryn Friedman Flack, MD; Address: 311 East 79th St. #2a New York, NY 10075; Email:
[email protected]; Telephone Number: (917)-453-5131; (212)-996-6633; Fax Number: (212)-996-6677
1
ACCEPTED MANUSCRIPT
Writing Assistance: None
M AN U
SC
RI PT
Disclosures: KF: No conflicts of interest to disclose; JD: Dr. Desai reports nonfinancial support from Boehringer Ingelheim, and he has participated in advisory boards for Boehringer Ingelheim, Bristol-Myers Squibb, and Daiichi Sankyo; JMK: No conflicts of interest to disclose; PC: No conflicts of interest to disclose; LCW: Dr. Wallentin has received grant support/honoraria/consulting fees from Boehringer Ingelheim, Regado Biosciences, Athera Biosciences, AstraZeneca, GlaxoSmithKline, Eli Lilly, Schering-Plough, and Bristol-Myers Squibb; MDE: Dr Ezekowitz has served as a consultant for AstraZeneca,Eisai, Pozen Inc., Boehringer Ingelheim, ARYx Therapeutics, Pfizer, Sanofi,Bristol-Myers Squibb, Portola, Daiichi Sankyo, Medtronic, Merck, Johnson& Johnson, Gilead, Janssen Scientific Affairs, and Armetheon. He has received research grant support from Boehringer Ingelheim, Bayer, DaiichiSankyo, Pfizer, and Bristol-Myers Squibb; SY: Dr Yusuf has received grant support/honoraria/consulting fees from Boehringer Ingelheim; SC: Dr Connolly has received grant support/ honoraria/consulting fees from Boehringer-Ingelheim, Bristol-Myers Squibb, Sanofi-aventis, and Portola; PR: Dr Reilly is a full-time employee of Boehringer Ingelheim; MB: Dr Brueckmann is an employee of Boehringer Ingelheim Pharma GmbH & Co. KG; JI: No conflicts of interest to disclose; JA: Dr Aisenberg reports he has participated in advisory boards and consulting for Boehringer Ingelheim and Portola.
AC C
EP
TE D
Authors’ Contributions: Study concept and design: Flack, Kathryn F, Desai, Jay, Kolb, Jennifer M, Wallentin, Lars C, Ezekowitz, Michael, Yusuf, Salim, Connolly, Stuart, Reilly, Paul, Aisenberg, James; Acquisition of data: Flack, Kathryn F, Desai, Jay, Kolb, Jennifer M, Chatterjee, Prapti, Aisenberg, James; Analysis and Interpretation of Data: Flack, Kathryn F, Desai, Jay, Kolb, Jennifer M, Chatterjee, Prapti, Ilgenfritz, John, Aisenberg, James; Drafting of the manuscript: Flack, Kathryn F, Aisenberg, James; Critical revision of the manuscript for important intellectual content: Flack, Kathryn F, Desai, Jay, Kolb, Jennifer M, Chatterjee, Prapti, Wallentin, Lars C, Ezekowitz, Michael, Yusuf, Salim, Connolly, Stuart, Reilly, Paul, Brueckmann, Martina, Ilgenfritz, John, Aisenberg, James; Statistical analysis: Ilgenfritz, John; Obtained funding: Flack, Kathryn F; Kolb, Jennifer M; Chatterjee, Prapti; Administrative, technical, or material support: Flack, Kathryn F, Kolb, Jennifer M, Chatterjee, Prapti, Connolly, Stuart, Reilly, Paul, Brueckmann, Martina; Study supervision: Desai, Jay, Wallentin, Lars C, Ezekowitz, Michael, Yusuf, Salim, Aisenberg, James.
Abstract:
2
ACCEPTED MANUSCRIPT
Background & Aims: Gastrointestinal (GI) bleeding in patients receiving anticoagulation agents can be caused by occult malignancies. We investigated the proportions and features of major GI bleeding (MGIB) events related to occult GI cancers in patients receiving anticoagulation therapy.
SC
RI PT
Methods: We analyzed data from the Randomized Evaluation of Long Term Anticoagulant Therapy study (conducted between December 2005 and March 2009 in 951 clinical centers in 44 countries worldwide), which compared the abilities of dabigatran vs warfarin to prevent stroke and systemic embolism in 18,113 patients with atrial fibrillation. Two blinded gastroenterologists independently reviewed source documents of MGIB events (n=595) that occurred during the study period. We collected data on MGIB events caused by previously unidentified GI malignancies, and compared characteristics of MGIB events in patients who received dabigatran vs warfarin (primary endpoint), and in patients with bleeding from cancer, vs patients bleeding from a nonmalignant or unidentified source.
TE D
M AN U
Results: Of 546 unique MGIB events, 44 (8.1%) were found to be from GI cancers (34/398 MGIB events in dabigatran users and 10/148 MGIB events in warfarin users; P=.60). Colorectal cancer accounted for 35/44 of all cancers identified. There were more colorectal cancer-associated MGIB events in the dabigatran group (30/34) than in the warfarin group (5/10) (P=.02), but more gastric cancer-associated MGIB events in the warfarin group (5/10) than in the dabigatran group (1/34) (P=.001). There were no differences in the short-term outcomes of cancer-related MGIB events in the dabigatran vs the warfarin group, but 75% of all cancer-related MGIB events required at least 1 blood transfusion and the mean hospital stay was 10.1 days. Compared to MGIB events from a non-malignant or unidentified source, MGIB from cancer occurred sooner (343.0 days vs 223.1 days, P=.003), but the bleeding was more likely to be chronic (for more than 7 days) (27.3% vs 63.6%, P<.001).
EP
Conclusion: In evaluating data from a study of the effects of anticoagulation therapy, we found approximately 1 of every 12 MGIB events to be related to an occult cancer. Approximately two-thirds of patients with cancer-related MGIB present with chronic bleeding, and morbidity and resource utilization is high.
AC C
KEY WORDS: RE-LY; DOAC; stomach cancer; colon cancer
3
ACCEPTED MANUSCRIPT
RI PT
Introduction:
Anticoagulation with a vitamin K antagonist (VKA) or a direct oral anticoagulant
(DOAC) reduces the incidence of stroke and systemic embolism in patients with
SC
atrial fibrillation (AF).1–5 Anticoagulation is associated with major bleeding,
including major gastrointestinal bleeding (MGIB), which occurs at a rate of
M AN U
approximately one to two percent per year.6 Most studies suggest that MGIB is more common with DOACs than VKAs, which may reflect topical anticoagulant action by intraluminal DOAC.6–10 Despite the prevalence of MGIB in AF patients receiving anticoagulation therapy, there is a paucity of evidence to guide
TE D
diagnosis and management.
In a subset of patients receiving anticoagulation therapy, MGIB represents the
EP
first presentation of occult malignancy. It has been hypothesized that initiating an anticoagulant may represent a stress test which can function to reveal an occult
AC C
cancer.9–17 Unlike warfarin, dabigatran oral ingestion results in the presence of active intraluminal anticoagulant within the gastrointestinal tract, and this might further promote bleeding from friable gastrointestinal cancers; thus dabigatran might represent a more intensive stress test than the warfarin. Indeed, a previous analysis based on database search terms suggested that cancer-
4
ACCEPTED MANUSCRIPT
related bleeding with three DOACs (rivaroxaban, apixaban, and dabigatran) may appear sooner after treatment initiation than with VKAs.17
RI PT
Dabigatran etexilate is a direct thrombin inhibitor that has a favorable safety and efficacy profile in comparison to warfarin, although dabigatran 150 mg twice daily is associated with increased MGIB.2 The Randomized Evaluation of Long-Term
SC
Anticoagulation Therapy (RE-LY) trial (N=18,113) provides the largest
comprehensive database related to MGIB in patients taking dabigatran versus
M AN U
warfarin. In the current study, we utilize individual chart review by experienced gastroenterologists in order to analyze the relationship between MGIB and underlying undiagnosed gastrointestinal (GI) tract malignancies. This is the first large cohort study to use primary source documents to analyze in detail the
TE D
characteristics of individual MGIB events related to previously undiagnosed occult malignancy in patients on DOACs and VKAs.
EP
Methods:
AC C
Adjudication of Major Gastrointestinal Bleeding Events during RE-LY
The RE-LY trial was a randomized, prospective trial comparing two blinded doses of dabigatran (110 mg and 150 mg twice daily) to open-label adjusteddose warfarin for stroke and systemic embolus prevention in 18,113 patients with non-valvular AF. Approval of the study was obtained from local regulatory
5
ACCEPTED MANUSCRIPT
authorities and ethics committees, and all patients provided written informed consent. During RE-LY, which was conducted in 951 centers across 44 countries, an adjudication committee, composed of 3 cardiologists, 3 neurologists, one
RI PT
internist, and one sponsor representative assessed all cases of major bleeding
that occurred from the start of treatment to the end of the observation period by reviewing case report forms and medical records created at the study sites.
SC
Cases which, in the judgment of the adjudicators, met International Society on Thrombosis and Hemostasis (ISTH) criteria for major bleeding (fatal bleeding,
M AN U
and/or decrease in hemogobin (Hgb) > 2 g/dL, and/or transfusion of > 2 units of blood, and/or symptomatic bleeding into a critical organ) were categorized as major bleeding events. Each MGIB event was later sub-classified to the specific organ system where the bleeding occurred, or, if this could not be determined, to
TE D
an unspecified site category.18,19
EP
Re-evaluation of Major Gastrointestinal Bleeding Events by Gastroenterologists
For the current study, the sponsor of RE-LY (Boehringer Ingelheim, Inc.)
AC C
provided the investigators with documents (case report forms, adjudication committee documents, and primary source documents) related to patients who experienced a MGIB event during RE-LY. This population included (a) 451 patients who were adjudicated during RE-LY to have experienced a major gastrointestinal bleeding event,20 plus (b) 82 patients who were re-categorized as having most likely bled from a gastrointestinal source, rather than (as originally
6
ACCEPTED MANUSCRIPT
adjudicated during RE-LY) from an unspecified site during a post hoc analysis of all MGIB in RE-LY.8 Thus, the total pool of probable MGIB patients in our study
RI PT
numbered 533. The trial sponsor did not otherwise participate in the current study.
Two experienced gastroenterologists (JD, JA) who were blinded to drug
treatment independently re-evaluated the primary source documents related to
SC
each MGIB event. The documents for three patients were not available, and thus these patients were eliminated prior to gastroenterologist review. In the interest
M AN U
of capturing the undiluted characteristics of cancer-related MGIB, stricter interpretations of the definition of a GI source were used during this re-evaluation. For example, bleeding events due to potentially non-luminal GI sources were excluded, and only first time bleeding events from a given source were included.
TE D
A second MGIB event in an individual patient was included only if it arose from a new unique lesion. The discrepancy between 546 events described in our study, and 533 patients with MGIB in the post-hoc analysis of RE-LY, is attributable to
EP
a) the inclusion of patients (N=35) who bled from more than one unique gastrointestinal source, and b) the exclusion of patients (N=23) for one or more of
AC C
the reasons enumerated above.
Data Collection and Analysis
Data regarding clinical presentation, diagnosis, and treatment of MGIB events were extracted from the primary source documents by the two treatment-blinded
7
ACCEPTED MANUSCRIPT
gastroenterologists. Non-blinded investigators extracted objective baseline characteristics and short- term outcomes from primary source documents and adjudication documents. Data related to follow-up visits or testing after trial
RI PT
completion were not available. In this analysis, patients were considered oral anticoagulant (OAC) naïve if they had never received OAC prior to their enrollment in RE-LY. Patients were considered to be taking a baseline
SC
concomitant medication [e.g. aspirin, clopidogrel, non-steroidal anti-inflammatory drugs (NSAIDs), and proton-pump inhibitor (PPI)] if they were taking this
M AN U
medication at randomization. Patients’ cancer screening histories prior to their enrollment in RE-LY were largely unavailable.
MGIBs were categorized as acute (less than two days), sub-acute (two to seven
TE D
days), or chronic (greater than seven days). If a patient had multiple presenting signs, the following hierarchy was used to determine the primary presenting sign: hematemesis>hematochezia>melena>occult bleeding. Occult bleeding was
EP
defined as a composite of the following symptoms in the absence of overt bleeding: (1) a positive fecal occult blood test (FOBT), (2) constitutional signs or
AC C
symptoms, (3) unexplained drop in Hgb, and (4) constitutional signs or symptoms with a positive FOBT or associated with a drop in Hgb. The time-to-bleed was defined as the number of days from the date of the first study drug dose to the date of the bleed. In order to be categorized as a fatal GIB, concordance between the gastroenterologists was required. Bleeding was classified as related to a cancer based on accepted histological, endoscopic, and/or surgical criteria.
8
ACCEPTED MANUSCRIPT
In all cases, the MGIB preceded the cancer diagnosis. If a patient had multiple GI bleeds, the case was assessed based on the bleed that resulted in the cancer
RI PT
diagnosis.
In order to compare the characteristics of MGIB due to cancer and MGIB due to other causes, a comparator group of events was created by pooling all MGIB
SC
events where a luminal GI malignancy was not diagnosed (n=502 MGIB events). This comparator group is comprised of MGIB events where either a non-
M AN U
malignant bleeding source was identified (n=269 events, 53.6%) or no bleeding source was identified, either because the workup was non-diagnostic (n= 138 events, 27.5%) or because the site investigator decided not to pursue a workup (e.g. because the likely source of bleeding had been determined during an
TE D
endoscopy that was performed prior to the time of the RE-LY study) (N=95 events, 18.9%). In designing this study’s protocol, we acknowledged that a fraction of the 95 uninvestigated MGIB events might reflect undiagnosed luminal
EP
GI cancer, but concluded that this fraction was acceptably small. Specifically, we estimated that < 2.0% of the total events in the comparator group could reflect a
AC C
missed cancer (cancer-related MGIB events = < 10.0% of 95 total uninvestigated MGIB events; <9.5 cancer MGIB events/502 total MGIB events = < 1.9%).
Statistical analyses
Statistical analyses were performed using Statistical Analysis System (SAS)
9
ACCEPTED MANUSCRIPT
version 9.2. Fisher’s Exact test was used for categorical data and t-test for continuous variables (two tailed, alpha = 0.05 without adjustment for multiple comparison). Kaplan-Meier curves were used to graphically display the time to
RI PT
MGIB. Comparisons were made between the study treatment groups and also
between those with GI cancer related MGIB and MGIB due to other causes. For
Results:
MGIB event case recruitment
M AN U
daily were analyzed as one group.
SC
the purpose of this analysis, patients on dabigatran 110 mg and 150 mg twice
TE D
Of the 18,113 patients randomized in RE-LY to receive drug treatment, 533 patients were determined by RE-LY investigators to have experienced a MGIB (Figure 1).8 During the gastroenterologists’ source document-based re-evaluation
EP
of 595 MGIB events performed for the current study, 49 events did not meet the current study criteria for unique MGIB events; 22 such events were removed
AC C
from the current analysis because they were considered repeat events with bleeding from the same source as the initial MGIB, and 27 events were removed from the analysis because the anatomic site was not considered to be intraluminal GI by the two expert gastroenterologists. Thus, the analysis for this study was performed on 546 unique MGIB events.
10
ACCEPTED MANUSCRIPT
Major Gastrointestinal Bleeding due to Cancer
Of all MGIB in RE-LY, 44/546 (8.1%) were due to previously undiagnosed
RI PT
luminal GI tract cancer. Cancer was responsible for 34/398 (8.5%) MGIB in
patients receiving dabigatran and 10/148 (6.8%) MGIB in patients receiving
warfarin (P=.60). Of the 34 dabigatran users experiencing cancer-related MGIB,
M AN U
dabigatran 150 mg twice daily treatment group.
SC
14 were in the dabigatran 110 mg twice daily treatment group and 20 were in the
Classification of Gastrointestinal Tract Cancers
Colorectal cancer (CRC) accounted for 35/44 (79.5%) malignancies diagnosed
TE D
after MGIB. CRC accounted for 88.2% of MGIB due to luminal GI malignancy in the dabigatran group (colonic, n=25, 73.5%; rectal, n=5, 14.7%) and 50.0% in the warfarin group (colonic, n=4, 40.0%; rectal n=1, 10.0%) (P=.02). Gastric cancer
EP
accounted for 50.0% of MGIB due to luminal GI malignancy in the warfarin group (5/10) compared to 2.9% in the dabigatran group (1/34) (P=.001). CRC and
AC C
gastric cancer accounted for all cancer-related MGIB with the exception of one ampullary cancer, and one each of renal cell carcinoma and melanoma that were metastatic to the luminal GI tract.
Characteristics of patients with major GI bleeding due to cancer
11
ACCEPTED MANUSCRIPT
The mean age of patients with cancer-related MGIB (n=44) was 76.8 years, and 65.9% were men (Table 1). At the time of randomization, 63.6% of patients with MGIB due to cancer had previously taken an OAC, 18.2% were taking a PPI, and
RI PT
45.5 % were taking aspirin. In the cancer-related MGIB group, there were no
significant differences in clinical characteristics between the two drug treatment
SC
groups.
Patients with cancer-related MGIB were less likely than patients in the non-
M AN U
cancer MGIB comparator group (n=502) to be taking NSAIDs at baseline (0/44, 0% vs. 51/502, 10.2%, P=.03), but the baseline clinical demographics and medication utilization in cancer-related MGIB and in the non-cancer comparator
TE D
group MGIB were otherwise similar.
Time to Cancer-Related Major Gastrointestinal Bleeding
EP
Cancer-related MGIB events occurred across the study period (Figure 2). The mean time interval between study drug initiation and MGIB due to cancer did not
AC C
differ between dabigatran users and warfarin users (p=.12). In dabigatrantreated patients, 19/34 (55.9%) cancer-related MGIB events occurred during the four months immediately after study drug initiation, while in warfarin-treated patients, 1/10 (10%) occurred during this time period. In both the cancer group and the comparator group, there were no significant differences between OAC naïve and OAC experienced patients in time to MGIB.
12
ACCEPTED MANUSCRIPT
The mean time interval between anticoagulant initiation and MGIB was shorter in the cancer-related bleeding group than the comparator group (223.1 vs. 343.0
RI PT
days, P=.003). This difference reflected a difference in patients receiving
dabigatran (195.9 vs. 335.9, P= .003) (Figure 2A), but not those receiving
SC
warfarin (317.9 vs. 361.8, P=.59) (Figure 2B).
M AN U
Presentation of Cancer-Related Major Gastrointestinal Bleeding
MGIB due to cancer presented as chronic bleeding in 63.6% of patients and acute/sub-acute bleeding in 27.3% (11.4% sub-acute and 15.9% acute) (Table 2). Cancer-related MGIB presented as overt bleeding in 65.9% of patients
TE D
(hematemesis 2.3%, hematochezia 40.9%, and melena 22.7%) and as occult in 34.1% (FOBT+ 4.5%, Drop in Hgb 20.5%, constitutional symptoms and drop in Hgb 6.8%, and constitutional symptoms and FOBT+ 2.3%). Cancer-related MGIB
EP
was more likely to present as hematochezia in patients taking dabigatran compared to warfarin (50.0% vs 10.0%, P= .03), and there was a trend toward
AC C
this cancer-related MGIB presenting more commonly as chronic bleeding in patients taking warfarin compared to dabigatran (90.0% vs 55.9%, P=.07).
MGIB in the cancer cohort was more likely to be chronic than MGIB in the comparator cohort in the dabigatran group, warfarin group, and overall (55.9% versus 29.1%, P=.003, 90.0% versus 22.6%, P<.001, and 63.6% versus 27.3%,
13
ACCEPTED MANUSCRIPT
P<.001, respectively). Conversely, MGIB in the comparator cohort was more likely to be acute or sub-acute in comparison to MGIB in the cancer cohort in patients taking dabigatran(58.2% versus 32.4%, P=.006), warfarin (65.2% versus
SC
Outcomes of Major Gastrointestinal Bleeding
RI PT
10.0%, P<.001), and overall (60.2% versus 27.3%, P<.001).
M AN U
Patients with cancer-related MGIB (n=44) were hospitalized in 81.4% (n=35) of cases and 25.6% (n=11) required admission to the intensive care unit (ICU) (Table 3). The average length of stay was 10.1 nights. Two (4.5%) of the MGIBs required emergent endoscopic treatment, one (2.3%) required emergent surgical
TE D
treatment, and 33 (75.0%) required at least one red blood cell transfusion. No patients died as a result of cancer-related MGIB.
EP
There were no significant differences between dabigatran- and warfarin-treated patients in short-term outcomes including: hospitalization rate, mean nights in
AC C
hospital, ICU requirement, transfusion requirement, and need for endoscopic and surgical intervention. Likewise, hospitalization rate, mean nights in hospital, ICU requirement, transfusion requirement, and need for endoscopic and surgical intervention were similar in the cancer and comparator cohorts. A total of 12 patients died as a result of bleeding in 502 MGIB events in the comparator group (2.4%) versus 0 patients in 44 cancer-related bleeds.
14
ACCEPTED MANUSCRIPT
Discussion:
RI PT
The current study represents to our knowledge the first large cohort study to use primary source documents to verify and characterize MGIB due to cancer in
patients receiving DOACs and VKAs. Most prior studies on cancer bleeding have
SC
been case reports and case series in patients receiving warfarin.13–15,21 Our study is relevant because of the increasing prevalence of AF and anticoagulation in the
M AN U
aging global population,22,23 the increasing prescription of DOACs, and the morbidity, mortality, and complex decision-making associated with MGIB and especially cancer-related MGIB in patients receiving anticoagulation therapy.24,25
TE D
The pivotal, global studies of the DOACs, such as RE-LY, upon which the approval of dabigatran in chronic AF was based, provide a novel opportunity to understand anticoagulation-related GI hemorrhage. It has been previously
EP
reported that MGIB overall is higher with dabigatran 150 twice daily (but not 110 twice daily) than with warfarin, that the highest GI bleeding rate is soon after
AC C
initiation of anticoagulation, 2, 7-9 and that a higher proportion of dabigatran- than warfarin-associated bleeds are from a lower gastrointestinal source.8,9
The principal findings of our study are: (1) approximately one in twelve MGIBs in patients receiving anticoagulation therapy reflects an underlying GI tract malignancy, most commonly CRC; (2) cancer-related MGIB in patients taking
15
ACCEPTED MANUSCRIPT
anticoagulants presents with chronic bleeding in approximately two-thirds of cases, as opposed to MGIB in patients in whom cancer is not diagnosed, where approximately two-thirds of cases present as recent-onset bleeding; (3) MGIB in
RI PT
dabigatran users due to gastrointestinal cancer presents sooner after initiation of anticoagulation than MGIB in dabigatran users in whom gastrointestinal cancer has not been diagnosed; and (4) cancer-related MGIB in patients receiving
SC
anticoagulation therapy results in significant morbidity and resource consumption.
M AN U
Our observations suggest several recommendations. First, as suggested in prior studies, MGIB, in particular chronic MGIB, in the patient receiving anticoagulation therapy requires investigation and is not dismissible as an incidental manifestation of anticoagulation.13,15,16,21 Second, because cancer-related MGIB
TE D
in AF patients receiving anticoagulation is common and results in significant morbidity and resource consumption, proactive diagnostic strategies may be worthwhile. For example, guideline-based GI cancer screening prior to initiating
EP
anticoagulation or periodic non-invasive screening for occult GI bleeding during anticoagulation treatment (e.g. by programmatic stool or blood testing) may allow
AC C
earlier cancer detection and thereby decrease morbidity and resource utilization.12,17 Prospective studies of these strategies are needed. Thus far, most studies have shown that anticoagulant use has no deleterious effect on the positive predictive value of FOBT for detecting gastrointestinal malignancy.11,26 Another study has shown no cross-reactivity between FOBTs and dabigatran.27
16
ACCEPTED MANUSCRIPT
It is uncertain why MGIB appears sooner in dabigatran users with a malignant bleeding source than in those in whom cancer is not diagnosed. This observation is of scientific rather than clinical interest, since bleeding should always be
RI PT
investigated methodically. We hypothesize that the observed difference may reflect the dissimilar anticoagulant pharmacology. The oral bioavailability of warfarin is essentially 100%,28 whereas that of dabigatran is approximately
SC
7%.29,30 Dabigatran, ingested as an inactive prodrug, is converted to the active
form by esterases as it passes through the GI tract.30 This results in progressively
M AN U
higher active dabigatran concentrations in the distal GI tract, which may act as a topical anticoagulant.8–10,30 Cancers such as CRC are highly vascular and friable, and typically bleed slowly but persistently.31 As such, they may be particularly susceptible to the topical and systemic effects of dabigatran. Earlier detection
TE D
could result in better cancer-related outcomes.32,33 Topical dabigatran activity may also underlie the significant increase in MGIB from CRC in the dabigatran
EP
group compared to those on warfarin.
Previous studies suggest that warfarin-related GI bleeding results in significant
AC C
morbidity and cost.25,34 Our results confirm this observation and suggest that dabigatran treatment is associated with similar bleeding-related outcomes to warfarin treatment in patients bleeding from a cancer source and those in whom cancer was not diagnosed. Patients bleeding from cancer required a mean of approximately ten nights in the hospital, and approximately one-fourth required intensive care, but 0/44 died as a direct result of the bleeding. It is hoped that the
17
ACCEPTED MANUSCRIPT
specific dabigatran reversal agent idarucizumab will improve bleeding outcomes in dabigatran treated patients.35
RI PT
In this retrospective review, we are limited by the quality of data collected during RE-LY. For example, as discussed in the Methods section, it is possible that a small fraction of patients in the comparator group bled from an undiagnosed
SC
cancer. Likewise, data regarding NSAID use at the time of MGIB events was not reliably available. We minimized the impact of these limitations by having two
M AN U
experienced gastroenterologists evaluate the source documents, and the high rate of concordance between these gastroenterologists and the RE-LY committee is reassuring. Additionally, although there were 546 unique MGIB in this study, there were only 34 cancer-related MGIB events with dabigatran and
limited.
TE D
10 with warfarin, so our power to detect differences between these groups was
EP
In conclusion, our results suggest that approximately one in twelve MGIBs in AF patients taking warfarin or dabigatran is due to an underlying cancer. This
AC C
bleeding results in significant morbidity and resource consumption, which is similar in the two drug treatment groups. These findings have implications related to diagnosis and clinical management of patients receiving anticoagulation treatment.
18
ACCEPTED MANUSCRIPT
Figure Legends
Figure 1: Flow chart of major gastrointestinal bleeding (MGIB) events in
RI PT
the Randomized Evaluation of Long-Term Anticoagulation Therapy (RE-LY) trial (N= 18,113) and those included in the current trial. The distribution of events bleeding from gastrointestinal (GI) cancers
SC
and those bleeding from non-cancer or undiagnosed lesions.
M AN U
Figure 2: Time from treatment initiation to major gastrointestinal bleeding event (MGIB) in patients on dabigatran (Panel A) and warfarin (Panel B). Comparisons between MGIB from cancer (solid line) versus MGIB
AC C
EP
TE D
from a non-cancer or an unidentified source (dashed line).
19
ACCEPTED MANUSCRIPT
References
1. Petersen P, Boysen G, Godtfredsen J, et al. Placebo-controlled,
RI PT
randomised trial of warfarin and aspirin for prevention of thromboembolic complications in chronic atrial fibrillation. The Copenhagen AFASAK study.
SC
Lancet 1989;1:175-179.
2. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Dabigatran versus warfarin in
M AN U
patients with atrial fibrillation. N Engl J Med 2009;361:1139-1151.
3. Giugliano RP, Ruff CT, Braunwald E, et al. Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med 2013;369:2093-2104.
TE D
4. Patel M, Mahaffey K, Garg J, et al. Rivaroxaban versus warfarin in
EP
nonvavular atrial fibrillation. N Engl J Med 2011;365:883-891.
5. Granger CB, Alexander JH, McMurray JJV, et al. Apixaban versus
AC C
warfarin in patients with atrial fibrillation. N Engl J Med 2011;365:981-992.
6. Ruff CT, Giugliano RP, Braunwald E, et al. Comparison of the efficacy and safety of new oral anticoagulants with warfarin in patients with atrial fibrillation: A meta-analysis of randomised trials. Lancet 2014;383:955-962.
20
ACCEPTED MANUSCRIPT
7. Graham DJ, Reichman ME, Wernecke M, et al. Cardiovascular, bleeding, and mortality risks in elderly Medicare patients treated with dabigatran or
RI PT
warfarin for nonvalvular atrial fibrillation. Circulation 2015;131:157-164.
8. Eikelboom JW, Wallentin L, Connolly SJ, et al. Risk of bleeding with 2
doses of dabigatran compared with warfarin in older and younger patients
SC
with atrial fibrillation: An analysis of the randomized evaluation of long-
M AN U
term anticoagulant therapy (RE-LY) Trial. Circulation 2011;123:2363-2372.
9. Sherid M, Sifuentes H, Sulaiman S, et al. Gastrointestinal bleeding with dabigatran, a comparative study with warfarin: A multicenter experience. Korean J Gastroenterol 2015;65:205-214.
TE D
10. Desai J, Granger CB, Weitz JI, et al. Novel oral anticoagulants in gastroenterology practice. Gastrointest Endosc 2013;78:227-239.
EP
11. Bini EJ, Rajapaksa RC, Weinshel EH. Positive predictive value of fecal occult blood testing in persons taking warfarin. Am J Gastroenterol
AC C
2005;100:1586-1592.
12. Johannsdottir GA, Onundarson PT, Gudmundsdottir BR, et al. Screening for anemia in patients on warfarin facilitates diagnosis of gastrointestinal malignancies and pre-malignant lesions. Thromb Res 2012;130:e20-e25.
21
ACCEPTED MANUSCRIPT
13. McCall KL, MacLaughlin EJ. Warfarin-associated bleeding complication saved life. Pharmacotherapy 2002;22:265-268.
Circulation 1962;25:804-806.
RI PT
14. Michaels MM. Bleeding from occult tumors during anticoagulant therapy.
SC
15. Norton S, Armstrong C. Lower gastrointestinal bleeding during
1997;79:38-39.
M AN U
anticoagulant therapy: A life-saving complication? Ann R Coll Surg Engl
16. Jaffin BW, Bliss CM, LaMont JT. Significance of occult gastrointestinal
TE D
bleeding during anticoagulation therapy. Am J Med 1987;83:269-272.
17. Clemens A, Strack A, Noack H, et al. Anticoagulant-related gastrointestinal bleeding—could this facilitate early detection of benign or
EP
malignant gastrointestinal lesions? Ann Med 2014;46:672-678.
AC C
18. Ezekowitz MD, Connolly S, Parekh A, et al. Rationale and design of RELY: randomized evaluation of long-term anticoagulant therapy, warfarin,
compared with dabigatran. Am Heart J 2009;157:805-810, 810 e1-e2.
19. Schulman S, Kearon C. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical
22
ACCEPTED MANUSCRIPT
patients. J Thromb Haemost 2005;3:692-694.
RE-LY trial. N Engl J Med 2010;363:1875-1876.
RI PT
20. Connolly SJ, Ezekowitz MD, Yusuf S, et al. Newly identified events in the
21. Hashash JG, Shamseddeen W, Skoury A, et al. Gross lower
SC
gastrointestinal bleeding in patients on anticoagulant and/or antiplatelet
therapy: Endoscopic findings, management, and clinical outcomes. J Clin
M AN U
Gastroenterol 2009;43:36-42.
22. Mozaffarian D, Benjamin EJ, Go AS, et al. Heart disease and stroke
TE D
statistics—2016 update. Circulation 2015;132:1-324.
23. January CT, Wann LS, Alpert JS, et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: A report of the American
EP
College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society. Circulation 2014;130:2071-
AC C
2104.
24. Manatsathit W, Al-Hamid H, Leelasinjaroen P, et al. Management of gastrointestinal bleeding in patients anticoagulated with dabigatran compared with warfarin: A retrospective, comparative case review.
23
ACCEPTED MANUSCRIPT
Cardiovasc Diagn Ther 2014;4:224-231.
25. Ghate SR, Biskupiak J, Ye X, et al. All-cause and bleeding-related health
RI PT
care costs in warfarin-treated patients with atrial fibrillation. J Manag Care Pharm 2011;17:672-684.
SC
26. Mandelli G, Radaelli F, Paggi S, et al. Anticoagulant or aspirin treatment does not affect the positive predictive value of an immunological fecal
M AN U
occult blood test in patients undergoing colorectal cancer screening: Results from a nested in a cohort case-control study. Eur J Gastroenterol Hepatol 2011;23:323-326.
TE D
27. Goss AM, van Ryn J, Schurer J, et al. Assessment of cross-reactivity in three different fecal occult blood test systems with dabigatran and dabigatran etexilate: Identification of useful test methods. Blood 2012;
EP
120: Abstr 4360.
AC C
28. COUMADIN ® (warfarin sodium). Full Prescribing Information. BristolMyers Squibb Pharma, Princeton, NJ, USA. 2015. http://packageinserts.bms.com/pi/pi_coumadin.pdf. Published December 2016. Accessed January 12, 2016.
24
ACCEPTED MANUSCRIPT
29. PRADAXA® (dabigatran etexilate mesylate). Full Prescribing Information. Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, CT, USA. 2015. http://docs.boehringer-
RI PT
ingelheim.com/Prescribing%20Information/PIs/Pradaxa/Pradaxa.pdf. Published November 2015. Accessed January 12, 2016.
SC
30. Blech S, Ebner T, Ludwig-Schwellinger E, et al. The metabolism and
disposition of the oral direct thrombin inhibitor, dabigatran, in humans.
M AN U
Drug Metab Dispos 2008;36:386-399.
31. James MW, Chen CM, Goddard WP, et al. Risk factors for gastrointestinal malignancy in patients with iron-deficiency anaemia. Eur J Gastroenterol
TE D
Hepatol 2005;17:1197-1203.
32. Mandel JS, Church TR, Ederer F, et al. Colorectal cancer mortality:
EP
Effectiveness of biennial screening for fecal occult blood. J Natl Cancer
AC C
Inst 1999;91:434-437.
33. Mandel JS, Bond JH, Church TR, et al. Reducing mortality from colorectal cancer by screening for fecal occult blood. N Engl J Med 1993;328:13651371.
25
ACCEPTED MANUSCRIPT
34. Guerrouij M, Uppal CS, Alklabi A, et al. The clinical impact of bleeding during oral anticoagulant therapy: Assessment of morbidity, mortality and
2011;31:419-423.
RI PT
post-bleed anticoagulant management. J Thromb Thrombolysis
35. Pollack Jr CV, Reilly PA, Eikelboom J, et al. Idarucizumab for dabigatran
AC C
EP
TE D
M AN U
SC
reversal. N Engl J Med 2015;373: 511-520.
26
ACCEPTED MANUSCRIPT
h
Table 1: Clinical characteristics of patients on dabigatran and warfarin with MGIB events during the RE-LY trial f Bleeding from cancer Bleeding from non-cancer or from undiagnosed g lesion Warfarin [N=10]
Total [N=502]
Dabigatran a [N=364]
Warfarin [N=138]
29 (65.9)
22 (64.7)
7 (70.0)
329 (65.5)
234 (64.3)
95 (68.8)
76.8 (6.7)
76.4 (6.3)
78.0 (8.4)
76.0 (7.2)
76.8 (6.6)
74.1 (8.3)
29.7 (6.2)
29.8 (5.5)
29.3 (7.5)
28.7 (6.0)
29.0 (6.3)
1.2 (0.3)
1.2 (0.3)
1.1 (0.4)b
1.2 (0.4)
1.3 (0.4)c
1.2 (0.4)d
35 (79.5)
27 (79.4)
8 (80.0)
377 (75.2)
269 (74.1)
16 (36.4)
13 (38.2)
3 (30.0)
141 (28.1)
103 (28.3)
38 (27.5)
20 (45.5)
15 (44.1)
5 (50.0)
261 (52.0)
191 (52.5)
70 (50.7)
0 (0.0)
53 (10.6)
36 (9.9)
17 (12.3)
2 (5.9)
SC 28.8 (6.1)
M AN U
TE D
EP
2 (4.5)
RI PT
Dabigatran a [N=34]
AC C
Male sex, no. (%) Age, mean (SD) Body Mass Index, mean (SD) Baseline Crmg/dL, mean (SD) Receiving treatment at time of event, no. (%) OAC Naïve, no. (%) Aspirin at Baseline, no. (%) Clopidogrel at Baseline, no. (%) NSAID at Baseline, no. (%) PPI at Baseline, no. (%)
Total [N=44]
e
108 (78.3)
0 (0.0)
0 (0.0)
0 (0.0)
51 (10.2)
38 (10.4)
13 (9.4)
8 (18.2)
7 (20.6)
1 (10.0)
93 (18.5)
64 (17.6)
29 (21.0)
a
A pooled dabigatran group including both those on dabigatran 110mg twice daily and dabigatran 150mg twice daily. N=9. One patient had a missing baseline Cr. N=359. Five patients had a missing baseline Cr. d N=135. Three patients had a missing baseline Cr. b c
27
ACCEPTED MANUSCRIPT
Table 1: Clinical characteristics of patients on dabigatran and warfarin with MGIB events during the RE-LY trial, Continued e
N=363. One patient had missing data on randomized treatment use at time of event. All P-values, D vs W, are non-significant. All P-values, D vs W, are non-significant, except for age, mean (SD) D vs W, P <.001 h For all comparisons of cancer vs non-cancer/undiagnosed lesions, P-values are non-significant, except for NSAID at baseline, Total vs Total, P=.0 f
AC C
EP
TE D
M AN U
SC
RI PT
g
28
ACCEPTED MANUSCRIPT
Table 2: Presentation and diagnosis of MGIB events during the RE-LY trial
Total [N=44]
Dabigatran a [N=34]
Warfarin [N=10]
Total [N=502]
Dabigatran a [N=364]
Warfarin [N=138]
Chronic
28 (63.6)
19 (55.9)
9 (90.0)
137 (27.3)
106 (29.1)
31 (22.5)
Acute/ Subacute
12 (27.3)
11 (32.4)
1 (10.0)
302 (60.2)
212 (58.2)
Unknown
4 (9.1)
4 (11.8)
0 (0.0)
63 (12.5)
46 (12.6)
Hematemesis
1 (2.3)
0 (0.0)
1 (10.0)
38 (7.6)
Hematochezia
18 (40.9)
17 (50.0)
1 (10.0)
Melena
10 (22.7)
7 (20.6)
c
15 (34.1)
Unknown
0.003
<.001
90 (65.2)
<.001
0.006
<.001
17 (12.3)
0.006
0.02
0.61
24 (6.6)
14 (10.1)
0.35
0.25
>0.99
181 (36.1)
132 (36.3)
49 (35.5)
0.52
0.14
0.17
3 (30.0)
170 (33.9)
127 (34.9)
43 (31.2)
0.18
0.13
>0.99
10 (29.4)
5 (50.0)
108 (21.5)
77 (21.2)
31 (22.5)
0.06
0.28
0.06
0 (0.0)
0 (0.0)
0 (0.0)
5 (1.0)
4 (1.1)
1 (0.7)
0.003
0.003
>0.99
Surgery
1 (2.3)
0 (0.0)
1 (10.0)
4 (0.8)
4 (1.1)
0 (0.0)
0.34
>0.99
0.07
Endoscopy
43 (97.7)
34 (100.0)
9 (90.0)
257 (51.2)
185 (50.8)
72 (52.2)
<.001
<.001
0.02
Radiology
4 (9.1)
1 (2.9)
21 (4.2)
17 (4.7)
4 (2.9)
0.13
>0.99
0.007
Mode of source diagnosisd
a
M AN U
TE D
Occult
3 (30.0)
SC
<.001
EP
Primary presenting sign
Bleeding from cancer versus from noncancer/ undiagnosed lesion Total vs D vs D W vs W Total P-Value P-Value P-Value
AC C
Time course of b bleeding
Bleeding from non-cancer or from f undiagnosed lesion
e
RI PT
Bleeding from cancer
A pooled dabigatran group including both those on dabigatran 110mg twice daily and dabigatran 150mg twice daily. Chronic was defined as >7 days, sub-acute as 2-7 days, and acute as <2 days Occult was defined as a composite of a positive fecal occult blood test (FOBT), constitutional signs or symptoms, a drop in hemoglobin (Hgb), constitutional signs or symptoms with a drop in Hgb, and constitutional signs or symptoms with a positive FOBT, without overt bleeding. d Patients could be counted in more than one category. e All P-values, D vs W, are non-significant except for hematochezia, P=0.03 and radiology P=0.03 f All P-values, D vs W, are non-significant. b c
29
ACCEPTED MANUSCRIPT
Table 3: Outcomes of MGIB events in patients receiving warfarin or dabigatran during the RE-LY trialg
Warfarin [N=10]
Total [N=502]
Dabigatran a [N=364]
Warfarin [N=138]
29 (85.3)
6 (66.7)b
441 (87.8)
321 (88.2)
10.1 (10.4)
11.3 (11.0)
5.4 (6.4)b
8.0 (17.8)
8.1 (19.5)
7.6 (11.8)
11 (25.6)
8 (23.5)
33 (75.0)
26 (76.5)
2 (4.5)
2 (5.9)
1 (2.9)
c
120 (87.0) d
3 (33.3)b
99 (19.7)
71 (19.5)
28 (20.3)
7 (70.0)
347 (69.1)
259 (71.2)
88 (63.8)
0 (0.0)
69 (13.7)
54 (14.8)
15 (10.9)
0 (0.0)
7 (1.4)
7 (1.9)
0 (0.0)
TE D
EP
1 (2.3)
M AN U
35 (81.4)
AC C
Hospitalization, no. (%) Nights in hospital, Mean (SD) Required ICU admission, no. (%) Required transfusion, no. (%) Required emergent endoscopic treatment of bleeding site, no. (%) Required emergent surgical treatment of bleeding site, no. (%) Death due to MGIB, no. (%) Death, possibly due to MGIB, no. (%)
Dabigatran a [N=34]
RI PT
Total [N=44]
Bleeding from non-cancer or from f undiagnosed lesion
e
SC
Bleeding from cancer
0 (0.0)
0 (0.0)
0 (0.0)
12 (2.4)
9 (2.5)
3 (2.2)
0 (0.0)
0 (0.0)
0 (0.0)
4 (0.8)
3 (0.8)
1 (0.7)
a
A pooled dabigatran group including both those on dabigatran 110mg twice daily and dabigatran 150mg twice daily. N=9. One patient missing data on hospitalization, nights in hospital, and ICU admission N=363. One patient missing data on nights in hospital d N=137. One patient missing data on nights in hospital b
c
30
ACCEPTED MANUSCRIPT
Table 3: Outcomes of MGIB events in patients receiving warfarin or dabigatran during the RE-LY trial, Continued e
All P-values, D vs W, are non-significant All P-values, D vs W, are non-significant g For all comparisons of cancer vs non-cancer/undiagnosed lesions, P-values are non-significant
AC C
EP
TE D
M AN U
SC
RI PT
f
31
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT
AC C
EP
TE D
M AN U
SC
RI PT
ACCEPTED MANUSCRIPT