Author + information
- Received April 14, 2009
- Revision received July 8, 2009
- Accepted July 12, 2009
- Published online September 29, 2009.
- Eugenia Nikolsky, MD, PhD*,* (, )
- Gregg W. Stone, MD*,
- Ajay J. Kirtane, MD, SM*,
- George D. Dangas, MD, PhD*,
- Alexandra J. Lansky, MD*,
- Brent McLaurin, MD‡,
- A. Michael Lincoff, MD§,
- Frederick Feit, MD†,
- Jeffrey W. Moses, MD*,
- Martin Fahy, MSc*,
- Steven V. Manoukian, MD∥,
- Harvey D. White, MD**,
- E. Magnus Ohman, MD¶,
- Michel E. Bertrand, MD††,
- David A. Cox, MD# and
- Roxana Mehran, MD*
- ↵*Reprint requests and correspondence:
Dr. Eugenia Nikolsky, Director, Academic Affairs, Cardiovascular Research Foundation, 111 East 59th Street, 11th Floor, New York, New York 10022
Objectives We assessed the incidence, predictors, and outcomes of gastrointestinal bleeding (GIB) in patients with acute coronary syndromes (ACS).
Background GIB is a potential hemorrhagic complication in patients with ACS treated with antithrombotic and/or antiplatelet medications. The clinical outcomes associated with GIB in this setting have not been systematically studied.
Methods In the ACUITY (Acute Catheterization and Urgent Intervention Triage Strategy) trial, 13,819 patients with moderate- and high-risk ACS, enrolled at 450 centers in 17 countries between August 2003 and December 2005, were randomized to the open-label use of 1 of 3 antithrombin regimens (heparin plus a glycoprotein IIb/IIIa inhibitor, bivalirudin plus a glycoprotein IIb/IIIa inhibitor, or bivalirudin monotherapy).
Results GIB within 30 days occurred in 178 patients (1.3%). Older age, baseline anemia, longer duration of study drug administration before angiogram, smoking, ST-segment deviation ≥1 mm, and diabetes were identified as independent predictors of GIB. On multivariable analysis, GIB was strongly associated with 30-day all-cause mortality (hazard ratio [HR]: 4.87 [interquartile range (IQR) 2.61 to 9.08], p < 0.0001), cardiac mortality (HR: 5.35 [IQR 2.71 to 10.59], p < 0.0001), and composite ischemia (HR: 1.94 [IQR 1.14 to 3.30], p = 0.014), as well as with 1-year all-cause mortality (HR: 3.97 [IQR 2.64 to 5.99], p < 0.0001), cardiac mortality (HR: 3.77 [IQR 2.14 to 6.63], p < 0.0001), myocardial infarction (HR: 1.74 [IQR 1.01 to 3.02], p = 0.047), and composite ischemia (HR: 1.90 [IQR 1.37 to 2.64], p = 0.0001). Patients who experienced GIB had significantly higher rates of stent thrombosis compared with patients without GIB (5.8% vs. 2.4%, p = 0.009).
Conclusions GIB is a serious condition in the scenario of ACS and is independently associated with mortality and ischemic complications.
Gastrointestinal bleeding (GIB) is one of the common sources of hemorrhage in conservatively treated patients with acute coronary syndromes (ACS) and in patients undergoing elective or urgent percutaneous coronary intervention (PCI) (1,2). Even in nonselected populations, GIB results in high rates of morbidity and mortality (3,4). The data are scarce, however, regarding the specific implications of GIB in patients with ACS whose contemporary management typically includes a combination of antithrombotic and antiplatelet therapy (5,6).
We, therefore, examined the database from the large, multicenter, randomized ACUITY (Acute Catheterization and Urgent Intervention Triage Strategy) trial to analyze the incidence, predictors and outcomes of GIB occurring within 30 days after randomization in moderate- to high-risk patients with ACS.
Design, inclusion and exclusion criteria, and principal results of the ACUITY trial have been reported in detail elsewhere (7–9). In brief, eligible patients were randomized to the open-label use of 1 of 3 antithrombin regimens started prior to angiography: heparin plus a glycoprotein IIb/IIIa inhibitor (GPI), bivalirudin plus a GPI, or bivalirudin monotherapy. Patients assigned to heparin plus a GPI or bivalirudin plus a GPI were randomized again in a 2 × 2 factorial design to routine upstream GPI started in all patients immediately after randomization versus deferred selective GPI initiation in the catheterization laboratory only after diagnostic angiography identified which patients would undergo immediate PCI (7).
After mandatory angiography performed within 72 h of randomization, the patients were triaged according to physician discretion to management by PCI, coronary artery bypass grafting (CABG), or medical therapy. Aspirin 300 to 325 mg orally (PO) or 250 to 500 mg intravenously (IV) was administered daily during the index hospitalization, and 75 to 325 mg daily was prescribed indefinitely after discharge. The initial dosing and timing of clopidogrel were left to physician discretion per local standards, although a 300-mg or greater loading dose was required in all cases within 2 h of PCI. Clopidogrel 75 mg daily was recommended for 1 year in all patients with coronary artery disease.
Clinical end points
The primary end point in the ACUITY trial was composite ischemia (death from any cause, myocardial infarction [MI], or unplanned revascularization for ischemia), major bleeding not related to CABG, and net clinical outcomes (composite ischemia or major bleeding) (7). The 30-day end points were measured as cumulative events occurring within 25 to 35 days after randomization. Major bleeding was intracranial or intraocular hemorrhage, hematoma ≥5 cm in diameter, access site hemorrhage requiring intervention, reoperation for bleeding, clinically overt bleeding resulting in a decrease in hemoglobin by ≥3 g/dl, reduction in hemoglobin concentration of ≥4 g/dl without an overt source of bleeding, or the use of any blood product transfusion (7). The definitions of other individual primary end points have been described (7–9). Bleeding events were also stratified using Thrombolysis In Myocardial Infarction (TIMI) classification (10).
Gastrointestinal bleeding was defined as an episode of coffee ground emesis, hematemesis, melena, or red blood per rectum occurring within 25 to 35 days after randomization and documented by a treating physician in the patient's original source documents. Anemia was defined as a baseline hematocrit value <39% for men and <36% for women (11). Chronic renal insufficiency was defined as creatinine clearance <60 ml/min (12). Acquired thrombocytopenia was defined as a nadir in-hospital platelet count ≤150 × 109/l in patients whose admission platelet count was >150 × 109/l and was further subdivided into mild, moderate, severe, and profound when nadir platelet counts were 100 to 150, 50 to 100, 20 to 50, and <20 × 109/l, respectively. Stent thrombosis was adjudicated based on Academic Research Consortium (ARC) criteria (13).
A clinical events committee blinded to treatment assignment adjudicated all end point events using original source documents. The study was conducted according to the Declaration of Helsinki and was approved by the institutional review board or ethics committee at each participating center; all patients provided written informed consent.
Categorical variables were compared with chi-square tests or the Fisher exact test where appropriate. Continuous variables are presented as medians with interquartile ranges (IQRs) and were compared using the Kruskal-Wallis test. The 30-day and 1-year outcomes are presented as Kaplan-Meier percentages and compared using log-rank tests. Multivariable logistic regression was used to identify independent predictors of GIB. The independent association of the occurrence of GIB and subsequent mortality, cardiac mortality, MI, composite ischemia, and stent thrombosis within the 1-year follow-up period was examined using Cox proportional hazards models in which GIB was treated as a time-dependent covariate. All multivariable models controlled for the following baseline covariates: age, sex, diabetic status, hypertension, current smoking, prior MI, prior CABG, prior PCI, baseline cardiac biomarker elevation, ST-segment deviation, chronic renal insufficiency, baseline anemia, treatment with heparin before randomization, randomization to 1 of the antithrombotic regimens, triage to PCI versus CABG versus medical management, and time from study drug administration to angiogram. A forward stepwise algorithm was used to select significant covariates from this list.
Among a total of 13,819 patients with ACS enrolled at 450 sites in 17 countries in the ACUITY trial, GIB within 30 days of randomization occurred in 178 patients (1.3%), including 107 of 7,789 (1.4%) patients triaged to PCI, 28 of 1,539 (1.8%) patients triaged to CABG, and 43 of 4,491 (1.0%) patients triaged to medical management. Rates of GIB were the highest in patients randomized to heparin plus a GPI (1.5%) followed by bivalirudin plus a GPI (1.4%) and bivalirudin monotherapy (0.9%) (p = 0.019). The incidence of GIB increased with age from 0.57% in patients ≤50 years of age to 0.95%, 1.13%, 1.87%, and 3.54% in patients >50 to ≤60, >60 to ≤70, >70 to ≤80, and >80 years of age, respectively.
A total of 48.2% of patients with GIB versus 4.1% of patients without GIB met ACUITY protocol criteria for major bleeding not related to CABG (p < 0.0001), whereas TIMI major bleeding criteria were met in 19.8% versus 1.2% of the patients, respectively (p < 0.0001).
Clinical and procedural characteristics of patients who developed GIB versus patients without GIB are shown in Table 1.Patients who developed GIB were older and more frequently female, and had higher rates of diabetes mellitus, including a higher prevalence of insulin-treated diabetes, worse renal function, anemia, and higher TIMI risk scores. The occurrence of GIB was associated with higher rates of low molecular weight heparin administration before coronary angiography and GPI before angiography and during PCI, longer duration of study drug administration before performance of angiogram, and with triage to revascularization with either PCI or CABG (75.8% vs. 67.4%, respectively, p = 0.016) rather than medical therapy.
Independent predictors of GIB
As shown in Table 2,multivariable predictors of GIB in the entire study population included older age, baseline anemia, longer duration of study drug administration before angiogram, smoking, ST-segment deviation ≥1 mm, and diabetes. In patients triaged to PCI, multivariable predictors of GIB included older age, baseline anemia, diabetes, smoking, and longer duration of study drug administration before angiogram. Randomization to bivalirudin monotherapy compared with randomization to heparin plus a GPI showed a trend toward lower risk of GIB in the entire study population, especially in the PCI cohort. Older age was the only independent predictor of GIB in patients triaged to CABG. In patients triaged to medical management, time from study drug administration to angiogram, chronic renal insufficiency, and baseline anemia were independent predictors of GIB.
Outcomes of patients with GIB
Patients with GIB compared with patients without GIB had a higher frequency of acquired thrombocytopenia (21.6% vs. 10.5%, p < 0.0001) due to a higher incidence of mild (14.7% vs. 8.4%, p = 0.003) and moderate thrombocytopenia (6.3% vs. 1.8%, p < 0.0001), but not severe or profound thrombocytopenia (0.6% vs. 0.3%, p = 0.61). A hemoglobin drop of ≥3 g/dl during index hospitalization occurred in 30.4% versus 1.3% of the patients with and without GIB, respectively (p < 0.0001). As expected, rates of blood product transfusion not related to CABG were significantly higher in patients with versus without GIB (39.2% vs. 1.8%, p < 0.0001). The index hospitalization was significantly longer in patients who experienced GIB compared with those who did not (6.0 days [IQR 4.0 to 10.0 days] vs. 3.0 days [IQR 2.0 to 6.0 days], p < 0.0001).
As shown in Figures 1 and 2,⇓⇓in the entire study population, at 30-day and 1-year follow-up, GIB was associated with significantly increased rates of all-cause mortality, cardiac mortality, and nonfatal MI resulting in notably higher rates of composite ischemia. Rates of urgent revascularization were not influenced by GIB.
By multivariable analysis, GIB was an independent predictor of 30-day all-cause mortality (hazard ratio [HR]: 4.87 [IQR 2.61 to 9.08], p < 0.0001), cardiac mortality (HR: 5.35 [IQR 2.71 to 10.59], p < 0.0001), and composite ischemia (HR: 1.94 [IQR 1.14 to 3.30], p = 0.014). At 1-year follow-up, this relationship continued to be significant, and GIB was also a predictor of myocardial infarction (Fig. 3).
Outcomes of patients with GIB in relation to triaged treatment
The negative impact of GIB on rates of death, MI, and composite ischemia was consistent across patients triaged to PCI and medical treatment (Table 3).In patients triaged to CABG, the occurrence of GIB also resulted in higher all-cause and cardiac mortality at 30 days and 1 year, whereas rates of nonfatal MI were unrelated to GIB.
Among 7,789 patients triaged to PCI, stent thrombosis (definite, probable, or possible) developed within 1 year in 6 of 107 (5.8%) patients with GIB and 161 of 7,682 (2.4%) patients without GIB (p = 0.009). All 6 cases of stent thrombosis (1 definite and 5 probable) in the GIB group occurred within 35 days of randomization. Five of these cases were fatal, including 3 patients who died during the index hospitalization (no information is available concerning whether antiplatelet therapy was continued after GIB episodes in these patients) and 2 patients who died after discharge (both were prescribed dual antiplatelet therapy at discharge). The only patient with GIB who survived an episode of stent thrombosis was prescribed a thienopyridine, but not aspirin, at discharge.
Medications at admission did not differ significantly between patients with and without GIB (Table 4).At discharge and 30-day and 1-year follow-up, patients with GIB were significantly less likely to be treated with aspirin than those without GIB. Daily dosages of aspirin at discharge were similar between the 2 groups. Treatment with thienopyridines at discharge, 30 days, and 1 year was also less likely in patients with versus without GIB. However, the difference was significant only at discharge. Other medications at follow-up, including beta-blockers, statins, and angiotensin-converting enzyme inhibitors or receptor blockers, did not differ significantly between the 2 groups.
The major results of this analysis are: 1) GIB within 30 days of randomization occurred in 1.3% of moderate- and high-risk patients with ACS, and was associated with strikingly increased rates of early and late mortality, nonfatal MI, stent thrombosis, acquired mild or moderate thrombocytopenia, need for blood product transfusions, and prolonged in-hospital stay; 2) older age, baseline anemia, longer duration of study drug administration before angiogram, smoking, ST-segment deviation ≥1 mm, and diabetes were strongly related to GIB within 30 days of randomization; and 3) aspirin and/or thienopyridine withdrawal was frequent among patients experiencing GIB, resulting in one-fourth of patients not receiving aspirin and almost one-half of patients not receiving thienopyridines at discharge.
In the ACUITY trial, GIB was the second most frequent source of non–CABG-related bleeding after access site bleeding in the entire study population, and was the most common source of bleeding among patients triaged to medical management. As patients with hemorrhagic diatheses and recent bleeding were excluded from the ACUITY trial, the true incidence of GIB in a nonselected ACS population is likely to be higher given the increasing incidence of GIB as the population ages. In 1 retrospective analysis of consecutive patients with ACS treated with a combination of aspirin, clopidogrel, and enoxaparin, rates of GIB were 2.7% (5).
Outcomes of patients with ACS experiencing GIB are typically poor. In this analysis, index hospitalization was twice as long in patients who experienced GIB compared with those who did not. Patients who developed GIB had extremely high rates of mortality and MI. Specifically, 9.6% of the patients who had GIB died within 35 days, with the majority of these patients (77%) dying from cardiac causes. Five of 6 patients who had GIB and probable/definite stent thrombosis died. At 1 year, around one-fifth of patients with GIB had died, one-half from a cardiac cause; cardiac mortality in patients with GIB nearly doubled; and more than one-third had died or experienced a nonfatal MI or an unplanned revascularization. In this analysis, 1-year mortality in patients experiencing GIB was almost 2-fold higher compared with a combined unstable angina/ACS population of 5,673 patients undergoing PCI from a large case-controlled Australian study (6). In the same study, patients with ACS had an approximately 5-fold increase in the odds of GIB compared with patients with stable coronary syndromes.
The mechanisms behind the high rates of mortality and nonfatal MI in patients with ACS experiencing GIB are multifactorial. These patients had a more unfavorable baseline clinical profile that included older age, higher prevalence of diabetes, anemia, and chronic renal insufficiency, all factors known to worsen the prognosis of patients with ACS (14–17). Bleeding-related hemodynamic instability may provoke or aggravate ischemia, resulting in adverse clinical outcomes. Patients with GIB are commonly treated with blood product transfusion, as happened with more than one-third of the patients in this analysis. Although no randomized trial has prospectively assessed the impact of blood transfusion on outcomes in patients with ACS, analyses from large retrospective series have provided evidence that transfusion may be harmful (18,19). In this analysis, occurrence of GIB correlated with higher incidence of thrombocytopenia, which is also known to worsen the prognosis of patients with ACS (20). The retrospective nature of this study does not allow us to identify the exact mechanisms of thrombocytopenia that might be related to the higher rates of treatment with GPI in patients who developed GIB or to administration of heparin resulting in heparin-induced thrombocytopenia with/without thrombosis. Gastrointestinal bleeding is a well-known cause of premature cessation of antiplatelet therapy, which poses additional risk for ischemic outcomes in patients with ACS (1,21,22). In this analysis, one-fifth (20.8%) of the patients with GIB were not taking either aspirin or thienopyridines at discharge. The percentage of patients not on aspirin or thienopyridines was highest among patients triaged to medical management (40.0%), followed by patients triaged to CABG (27.3%) and to PCI (13.3%). Of note, in the ACUITY trial, aspirin was recommended at a relatively high daily dosage (300 to 325 mg PO or 250 to 500 mg IV) during index hospitalization. Although aspirin at any dosage is known to be ulcerogenic, an escalating dose is related to a higher incidence of GIB (23). The optimal dosage of aspirin in ACS is still not known and is being investigated in an ongoing trial (24).
Consistent with previous reports, in this analysis, occurrence of GIB was strongly associated with older age, anemia, and smoking (25–27). A novel and an important practical finding of this study is that the duration of antithrombotic therapy before angiography was more than doubled in patients with GIB compared with patients without GIB. By multivariable analysis, there was a 12% increment in odds of GIB per 10-h increase in duration of antithrombotic medications. To reduce the incidence of GIB and perhaps other bleeding complications, this time should be shortened in ACS patients at high risk for bleeding.
Concern is warranted regarding treatment with heparin plus a GPI, which by multivariable analysis was associated with a trend toward a higher incidence of GIB compared with bivalirudin monotherapy across the entire ACUITY population, especially in patients triaged to PCI. Although GPI are known to increase access site–related bleeding in patients treated with PCI, an increased incidence of GIB has not been reported in any of the pivotal trials assessing safety and efficacy of different GPIs in the setting of PCI (28–30) or conservative treatment (31,32). Still, in the randomized REPLACE-2 (Randomized Evaluation of PCI Linking Angiomax to Reduced Clinical Events) trial, GIB occurred significantly more often in patients randomized to heparin plus a GPI compared with patients randomized to bivalirudin (0.6% vs. 0.1%, p = 0.003) (2). It is also possible that rates of GIB may vary with different types of GPI. In a large regional registry of contemporary PCI, treatment with abciximab as opposed to eptifibatide was associated with a higher incidence of GIB in patients treated with primary PCI (4.8% vs. 2.8%, p = 0.01) (33). The results of this analysis coupled with the data from the REPLACE-2 trial support the use of bivalirudin rather than heparin plus a GPI in patients with ACS at high risk for GIB.
The ACUITY trial was not specifically powered to evaluate the predictors and clinical outcomes of GIB. Nonetheless, GIB was rigorously defined in the protocol, and all bleeding and end point data were prospectively collected and adjudicated by an independent Clinical Event Committee that was blinded to randomized treatment assignment. Given the post hoc nature of this analysis, data are lacking on the specific causes and management of GIB. Even by multivariable analysis, GIB was strongly associated with mortality and the composite ischemic end point; however, no definite conclusion may be drawn with regard to GIB as a direct cause of these events. Information on other drugs known to increase the risk of GIB, including warfarin, nonsteroidal anti-inflammatory drugs, corticosteroids, selective serotonin reuptake inhibitors, and spironolactone, was not collected. This analysis also lacks data on treatment with proton pump inhibitors. Although their efficacy for the secondary prevention of upper GIB was confirmed in randomized controlled trials (34,35), no study has prospectively assessed the safety and efficacy of this therapy in the primary prevention of upper GIB. Moreover, recent evidence suggests that proton pump inhibitors may decrease clopidogrel's inhibitory effect on platelet P2Y12 receptors and are associated with a higher risk of adverse outcomes after ACS (36,37). Additional clinical measurements not included in this analysis (e.g., certain comorbidities) may also have interacted with GIB. By study protocol, bleeding events occurring after 30-day follow-up were not collected, nor was the timing of discontinuation of aspirin and/or thienopyridines.
GIB, even at a level that does not meet the definition of major bleeding, is a devastating condition in the setting of ACS and is associated with high rates of mortality, nonfatal MI, stent thrombosis, and prolonged hospitalization. Physicians should be aware of GIB in high-risk populations, including the elderly, smokers, and patients with anemia. Given the adverse prognostic significance of GIB, it should be reported in any trial assessing the safety of new antithrombotic agents and regimens.
Dr. Nikolsky has received lecture fees from Abbott. Drs. Stone and Lincoff have received research support from The Medicines Company. Dr. Kirtane has received honoraria from The Medicines Company and is a consultant for Medicure. Dr. Dangas is a speaker for The Medicines Company. Dr. Lansky has received unrestricted research support (modest) from The Medicines Company. Dr. Feit is a consultant for The Medicines Company and is a shareholder for Johnson & Johnson and The Medicines Company. Dr. Manoukian is a consultant for The Medicines Company (significant) and for Bristol-Myers Squibb (modest), and has received an educational grant from Sanofi-Aventis (significant). Dr. White has received a research grant from The Medicines Company. Dr. Ohman is a consultant for Abiomed, CV Therapeutics, Datascope, Inovise, Liposcience, Response Biomedical, Northpoint Domain, and Pozen Inc.; has received research grants from Bristol-Myers Squibb, CV Therapeutics, Inc., Daiichi Sankyo, Datascope, Eli Lilly & Co, Sanofi-Aventis, Schering-Plough Corporation, and The Medicines Company; and has equity in Inovoise. Dr. Cox is a consultant for Abbott Vascular and for Boston Scientific and is on the Speakers' Bureau for The Medicines Company. Dr. Mehran is a consultant for and has received honoraria from The Medicines Company, Daichi Sankyo, Sanofi-Aventis, Cordis, Medtronic, Abbott, and Boston Scientific.
- Abbreviations and Acronyms
- acute coronary syndrome
- coronary artery bypass grafting
- gastrointestinal bleeding
- glycoprotein IIb/IIIa inhibitor
- interquartile range
- myocardial infarction
- percutaneous coronary intervention
- Thrombolysis In Myocardial Infarction
- Received April 14, 2009.
- Revision received July 8, 2009.
- Accepted July 12, 2009.
- American College of Cardiology Foundation
- Lincoff A.M.,
- Kleiman N.S.,
- Kereiakes D.J.,
- et al.,
- REPLACE-2 Investigators
- Barkun A.,
- Sabbah S.,
- Enns R.,
- et al.,
- RUGBE Investigators
- Stone G.W.,
- White H.D.,
- Ohman E.M.,
- et al.,
- Acute Catheterization and Urgent Intervention Triage strategy (ACUITY) Trial Investigators
- Bovill E.G.,
- Terrin M.L.,
- Stump D.C.,
- et al.
- ↵(1968) Nutritional Anemias: Report of a WHO Scientific Group (World Health Organization, Geneva).
- Cutlip D.E.,
- Windecker S.,
- Mehran R.,
- et al.,
- Academic Research Consortium
- Nikolsky E.,
- Aymong E.D.,
- Halkin A.,
- et al.
- Feit F.,
- Manoukian S.V.,
- Ebrahimi R.,
- et al.
- Yang X.,
- Alexander K.P.,
- Chen A.Y.,
- et al.,
- CRUSADE Investigators
- Grines C.L.,
- Bonow R.O.,
- Casey D.E. Jr..,
- et al.
- Ferrari E.,
- Benhamou M.,
- Cerboni P.,
- Marcel B.
- Lanas A.,
- García-Rodríguez L.A.,
- Arroyo M.T.,
- et al.,
- Asociación Española de Gastroenterología
- Mehta S.R.,
- Bassand J.P.,
- Chrolavicius S.,
- et al.,
- CURRENT-OASIS 7 Steering Committee
- Gurm H.S.,
- Smith D.E.,
- Collins J.S.,
- et al.
- Gilard M.,
- Arnaud B.,
- Cornily J.C.,
- et al.