Author + information
- Received June 8, 2013
- Revision received September 8, 2013
- Accepted September 16, 2013
- Published online January 28, 2014.
- Payal Kohli, MD∗,†,
- Jacob A. Udell, MD, MPH†,
- Sabina A. Murphy, MPH†,
- Christopher P. Cannon, MD†,
- Elliott M. Antman, MD†,
- Eugene Braunwald, MD† and
- Stephen D. Wiviott, MD†∗ ()
- ∗University of California San Francisco, Division of Cardiology, San Francisco, California
- †TIMI Study Group Division of Cardiology, Department of Medicine and Brigham and Women’s Hospital and Harvard Medical School, Boston, Massachusetts
- ↵∗Reprint requests and correspondence:
Dr. Stephen D. Wiviott, TIMI Study Group/Cardiovascular Division, Brigham and Women’s Hospital, 350 Longwood Avenue, 1st Floor Office, Boston, Massachusetts 02115.
Objectives The goal of this study was to determine whether there is a relationship between aspirin dose and the potent antiplatelet agent prasugrel in the TRITON–TIMI 38 (Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis In Myocardial Infarction 38) study.
Background Optimal aspirin dosing after acute coronary syndromes remains uncertain. Previous studies have raised questions regarding an interaction between high-dose aspirin and the potent antiplatelet agent ticagrelor.
Methods In TRITON–TIMI 38, we classified 12,674 patients into low-dose (<150 mg) or high-dose (≥150 mg) aspirin groups based on discharge dose. We identified independent correlates of dose selection and studied the impact of aspirin dose on the clinical effects of prasugrel.
Results There was significant geographical variation in aspirin dosing, with North American patients receiving high-dose aspirin more frequently than other countries (66% vs. 28%; p < 0.001). Clinical factors correlating with high-dose aspirin included previous percutaneous coronary intervention and use of aspirin before randomization. Characteristics associated with the use of low-dose aspirin included age ≥75 years, white race, and use of bivalirudin or a glycoprotein IIb/IIIa inhibitor during coronary intervention. Regardless of low- or high-dose aspirin use, prasugrel had lower rates of the primary efficacy endpoint (cardiovascular death, myocardial infarction, or stroke [CVD/MI/stroke]) (hazard ratio [HR]CVD/MI/stroke = 0.78 [95% confidence interval (CI) 0.64 to 0.95] and HRCVD/MI/stroke = 0.87 [95% CI 0.69 to 1.10], respectively; p value for interaction = 0.48) and higher rates of the primary safety endpoint (HR TIMI major bleeding = 1.40 [95% CI 0.81 to 2.42] and TIMImajor bleeding = 1.30 [95% CI 0.63 to 2.68], respectively; p value for interaction = 0.84) compared with clopidogrel.
Conclusions In TRITON–TIMI 38, the safety and efficacy outcomes of prasugrel compared with those of clopidogrel were directionally consistent regardless of aspirin dose, although only the primary efficacy endpoint achieved statistical significance. There was no clinically meaningful interaction of aspirin with prasugrel, suggesting that previous observations with potent antiplatelet agents indicating differential results are not universal. (A Comparison of Prasugrel [CS-747] and Clopidogrel in Acute Coronary Syndrome Subjects Who Are to Undergo Percutaneous Coronary Intervention; NCT00097591)
The use of aspirin in the secondary prevention of patients with acute coronary syndromes (ACS) traces a long history that began with the landmark ISIS-2 (Second International Study of Infarct Survival) study in 1988, which reported that 160 mg/day of aspirin within 24 h of presentation significantly reduced re-infarction and nonfatal stroke without a significant increase in cerebral hemorrhage (1). Demonstration of the efficacy of aspirin was subsequently extended across the entire spectrum of ACS, from ST-segment elevation myocardial infarction (MI) to unstable angina/non–ST-segment elevation myocardial infarction (NSTEMI) (2–4). The dose of aspirin used in these preliminary studies, however, was highly variable and ranged from 75 to 325 mg, leaving many unanswered questions about what the optimal daily dose of aspirin should be after ACS (5).
The PLATO (Platelet Inhibition and Patient Outcomes) trial randomly assigned 18,624 patients with ACS to receive either the potent P2Y12 receptor antagonist ticagrelor or the first-generation thienopyridine clopidogrel, in addition to aspirin, with the dose of aspirin left to the discretion of the local investigator. This trial demonstrated a benefit of ticagrelor compared with clopidogrel in the reduction of recurrent cardiovascular events (6). Within the trial, however, an apparent decreased efficacy of ticagrelor was noted in North American subjects (7). Subsequent analysis of this finding suggested that it might have been related to the reduced efficacy of ticagrelor in combination with the higher doses of aspirin more commonly prescribed in North America (7). Based on these findings, the US Food and Drug Administration recommended that ticagrelor be used only in combination with low doses of aspirin (i.e., 75 to 100 mg/day) (8).
Like ticagrelor, prasugrel is a more potent P2Y12 receptor antagonist than clopidogrel. The goal of the current study was to determine whether the dose of aspirin affected the comparison between prasugrel and aspirin in the TRITON-TIMI 38 (Trial to Assess Improvement in Therapeutic Outcomes by Optimizing Platelet Inhibition With Prasugrel–Thrombolysis In Myocardial Infarction 38) trial (9).
Patient population and treatments
TRITON–TIMI 38 was a multicenter, double-blind trial of patients with ACS, including high-risk unstable angina/NSTEMI and ST-segment elevation MI with planned percutaneous coronary intervention (PCI), who were randomized to receive either prasugrel or clopidogrel. All patients randomized into TRITON–TIMI 38 with previously published inclusion and exclusion criteria (10) who were still alive and free from MI, stroke, coronary artery bypass grafting (CABG), or Thrombolysis In Myocardial Infarction (TIMI) major bleed at the time of discharge and had information on discharge aspirin dose were included in our substudy.
All patients in this study were treated with aspirin plus a thienopyridine. The thienopyridine was randomly assigned as either prasugrel (60-mg loading dose, 10 mg daily) or clopidogrel (300-mg loading dose, 75 mg daily) for up to 15 months in a double-blind fashion; study drug could be administered on presentation with ST-segment elevation MI for primary PCI and at the time coronary anatomy was known to be suitable for PCI in all other subjects. The dose of aspirin selected at the time of PCI, during hospitalization, and at discharge was left to the discretion of the treating physician, with a loading dose suggested for those not previously taking aspirin and a recommended (but not mandated) maintenance dose of 75 to 100 mg. Dose adjustments were permitted for aspirin during the study period.
In this analysis, we report the relationship of randomized treatment arm (prasugrel vs. clopidogrel) and discharge aspirin dose to the primary efficacy and safety outcomes of the overall trial up to a follow-up time of 450 days, including the primary composite endpoint of the trial (cardiovascular death, MI, or stroke [CVD/MI/stroke]). Definite or probable stent thrombosis, as defined by the Academic Research Consortium, was also examined in the trial as a secondary outcome. Key safety endpoints included non–CABG-related TIMI major and minor bleeding as well as non-CABG gastrointestinal bleeding. All efficacy and safety outcomes were adjudicated by a blinded clinical events committee by using prespecified definitions (10).
Substudy population and classification
To minimize confounding with respect to aspirin dose at discharge, we excluded patients from the overall study cohort who experienced any component of the primary endpoint (CVD/MI/stroke), CABG, or bleeding event (n = 873) during the index hospitalization; those with definite/probable stent thrombosis (n = 8) during index hospitalization; and those missing aspirin dose information before or at hospital discharge (n = 53). Based on the discharge dose of aspirin, all remaining patients (n = 12,674) were classified as receiving low-dose (<150 mg) or high-dose (≥150 mg) aspirin. As a sensitivity analysis, we also report the relationship between prasugrel versus clopidogrel and aspirin dose and efficacy and safety outcomes, with low-dose aspirin defined as ≤100 mg and high-dose defined as >100 mg, as previously defined (5).
The baseline characteristics of low- and high-dose aspirin groups were compared by using the Wilcoxon rank sum test for continuous variables and the chi-square test for categorical variables. Using a univariate Cox proportional hazards regression model, we calculated hazard ratios (HRs) and 95% confidence intervals (CIs) to compare the efficacy and safety of prasugrel versus clopidogrel within the low-dose and high-dose aspirin groups and assessed the interaction between aspirin dose group and clinical efficacy of prasugrel. In these analyses, an interaction was considered statistically significant if the p value was <0.05. Multivariable logistic regression was derived to study the association between baseline characteristics and maintenance aspirin dose, with the low-dose aspirin groups serving as the reference group. Kaplan-Meier rates and curves were reported through 450 days of follow-up. To adjust for baseline differences, a propensity score was calculated by using forward selection, adding in all baseline characteristics with p < 0.20, and propensity score–adjusted Cox models were generated. Randomization group was included as a covariate in these propensity score–adjusted Cox models, and we assessed the association of high-dose versus low-dose aspirin with clinical outcomes. Therefore, the relationship between aspirin dose and outcomes was adjusted for by using both the propensity score and the randomization group. All analyses reported in the current paper were investigator-driven and performed independently by the TIMI Study Group by using STATA/SE version 12.1 (Stata Corp, College Station, Texas).
Of the 12,674 patients eligible for our analysis, 7,606 (60%) received low-dose aspirin (<150 mg) and 5,068 (40%) received high-dose aspirin (≥150 mg) at time of discharge. Within both groups, individuals were evenly balanced between the randomized treatment arms: in the low-dose group, 3,831 (50.4%) were randomized to receive prasugrel and 3,775 (49.6%) were randomized to receive clopidogrel; in the high-dose group, 2,549 (50.3%) were randomized to receive prasugrel and 2,519 (49.7%) were randomized to receive clopidogrel (p = 0.94). When assessed over time, the majority of patients remained within the same category (low or high) as their discharge dose during the remainder of follow-up (30 days 94%; 90 days 92%; 180 days 90%; 270 days 89%; 360 days 87%).
There were several differences in baseline characteristics between aspirin dose groups (Table 1). Notably, in the TRITON–TIMI 38 trial, high-dose aspirin was used more frequently in younger patients, those with unstable angina/NSTEMI as an index event, those with previous MI, and in patients who had a drug-eluting stent placed during PCI. In addition, there was marked geographical variation with respect to aspirin dosing, with 66% of patients in North America receiving high-dose aspirin compared with only 28% in other countries (p < 0.001) (Fig. 1). Within each stratum of aspirin dosing, baseline characteristics were generally well balanced between prasugrel versus clopidogrel (Online Table 1).
Independent correlates of discharge aspirin dose selection
After adjustment, several clinical factors emerged as independent correlates of aspirin dosing. As depicted in Figure 2, the strongest correlate of high-dose aspirin use was geographical region, with an odds ratio (OR) of 5.19 (95% CI: 4.72 to 5.70; p < 0.001) for North America versus other countries worldwide. Additional independent characteristics associated with high-dose aspirin use included previous PCI (OR: 1.18 [95% CI: 1.03 to 1.35], p = 0.015), systolic blood pressure (OR: 1.03 per 10-mm Hg increase [95% CI: 1.02 to 1.05], p < 0.001), and use of aspirin at randomization (OR 1.37 [95% CI: 1.07 to 1.76], p = 0.013). Factors associated with use of low-dose aspirin were age ≥75 years (OR: 0.87 [95% CI: 0.77 to 0.98], p = 0.023), white race (OR: 0.76 [95% CI: 0.65 to 0.88], p < 0.001), higher baseline heart rate (OR: 0.96 per 10 beats/min increase in heart rate [95% CI: 0.93 to 0.99], p = 0.004), and use of a glycoprotein IIb/IIIa inhibitor (OR: 0.87 [95% CI: 0.80 to 0.95], p = 0.002), bivalirudin (OR: 0.73 [95% CI: 0.58 to 0.91], p = 0.006), or another nonheparin antithrombin agent (OR: 0.90 [95% CI: 0.82 to 1.00], p = 0.04) during PCI.
Prasugrel versus clopidogrel stratified according to aspirin dose group: efficacy and safety endpoints
There was no modification of the clinical effect of prasugrel versus clopidogrel based on discharge aspirin dose with respect to the primary efficacy endpoint (HRCVD/MI/stroke = 0.78 [95% CI: 0.64 to 0.95] for aspirin <150 mg; HRCVD/MI/stroke= 0.87 [95% CI: 0.69 to 1.10] for aspirin ≥150 mg; p value for interaction = 0.48) (Fig. 3A) or the primary safety endpoint (HRTIMI major bleeding = 1.47 [95% CI 0.97 to 2.21] for aspirin <150 mg; HRTIMI major bleeding = 1.58 [95% CI: 0.92 to 2.69] for aspirin ≥150 mg; p value for interaction = 0.84) (Fig. 3B). Although the only comparison that achieved statistical significance for the superiority of prasugrel over clopidogrel was the primary efficacy endpoint of CVD/MI/stroke in the low-dose aspirin group, the point estimates remain directionally consistent with the results of the overall TRITON–TIMI 38 trial. Similarly, for both the high- or low-dose aspirin groups, the pattern of prasugrel effects compared with clopidogrel (fewer primary and secondary endpoint events and higher bleeding) with directionally preserved point estimates were similar to the parent trial, although not all comparisons achieved statistical significance (Fig. 4). Patients randomized to receive prasugrel had higher bleeding rates than patients randomized to receive clopidogrel, regardless of the dose of aspirin used. These results were also consistent in sensitivity analyses when the aspirin dose cutoff of ≤100 mg was used to define the low-dose group or when a cutoff of <150 mg was used and in-hospital endpoint events were included in the analysis. In addition, when analyzed within aspirin dose groups, the safety and efficacy relationship of prasugrel vs. clopidogrel was maintained (Online Table 2).
In this analysis of the TRITON–TIMI 38 trial of ACS patients, we made several observations. First, we added to the previous body of literature that there is significant global variation with respect to discharge aspirin dose. In TRITON–TIMI 38, higher-dose aspirin was used much more frequently in North America compared with the rest of the world, which more frequently prescribed lower-dose aspirin. Second, our major novel finding was that there seems to be no significant interaction between discharge aspirin dose and the clinical effect of prasugrel versus clopidogrel. Prasugrel was statistically superior to clopidogrel in reducing the composite primary endpoint of CVD/MI/stroke and directionally lower for stent thrombosis in the low dose aspirin group, with a directionally consistent improvement that did not achieve statistical significance in the high-dose aspirin group for the primary endpoint and for stent thrombosis. Similarly, prasugrel resulted in a higher bleeding rate, including TIMI major non-CABG bleeding and TIMI major/minor non-CABG bleeding, compared with clopidogrel, regardless of discharge aspirin dose.
The geographical variation of aspirin use after ACS in our study confirms a practice pattern similar to previous studies, including the PLATO study, which also demonstrated significantly more high-dose aspirin use in North America (7). Even after adjustment for baseline differences in patient characteristics and risk profiles, geographical region persisted as an independent correlate of aspirin dosing. The explanation for this disparity is unclear, but it is likely a result of historical prescribing practices and guidelines that differ according to country (5,7,11,12).
Until recently (including during the conduct of TRITON–TIMI 38), the American College of Cardiology/American Heart Association ACS guidelines had maintained recommendations for higher-dose aspirin in the setting of recent coronary stents, particularly drug-eluting stents (13). In 2011, an update to these guidelines for the management of ACS revised the recommendation to change the discharge aspirin dose from high dose to low dose (14).
Aspirin exhibits dose-dependent pharmacology. At low daily doses (∼100 mg), it selectively acetylates cyclooxygenase (COX)-1, leading to irreversible inhibition of this enzyme and decreased production of its downstream mediators, which are implicated in platelet aggregation and vascular reactivity and vasoconstriction; therefore, inhibiting the upstream COX-1 enzyme results in an antiplatelet and vasodilatory effect (15). At doses typically much higher than those used in ACS (650 mg to 4 g), however, aspirin can also inhibit COX-2, another isoform of this enzyme that is inducible and involved in pathological processes, including inflammation and pain, and potentially more complete inhibition of COX-1 (16,17). However, there remains a continuum and dose-response relationship between these lower and higher doses of aspirin. Therefore, higher doses may improve efficacy but may also lead to increased bleeding (18) by depleting gastroprotective prostacyclins. In addition, aspirin impairs adenosine diphosphate (ADP) release and ADP-dependent platelet aggregation, and its effects on platelets are dose dependent and irreversible (18). Therefore, higher doses of aspirin also have the potential for increased drug–drug interactions with respect to the ADP receptor antagonists (clopidogrel, prasugrel, and ticagrelor).
Our results do not support any interaction (favorable or unfavorable) between prasugrel and aspirin dose. There are many possible explanations for our findings in the context of the PLATO study, in which ticagrelor was associated with worse outcomes when given concurrently with high-dose aspirin (7). Although both ticagrelor and prasugrel act via the potent inhibition of the P2Y12 ADP receptor, the mechanism of action of the 2 drugs is slightly different. Ticagrelor binds reversibly, allowing for the potential for greater drug–drug interaction with the irreversible action of aspirin on platelets. However, it is also possible that there might not be a true pharmacological interaction between ticagrelor and aspirin dose, and the interaction previously observed with ticagrelor and high-dose aspirin may have occurred by chance due to a small patient subgroup (19).
Our findings are relevant with respect to the possible interaction between the first-generation thienopyridine clopidogrel and aspirin. In the CURRENT OASIS-7 (Clopidogrel and Aspirin Optimal Dose Usage to Reduce Recurrent Events–Seventh Organization to Assess Strategies in Ischemic Syndromes) trial, 25,086 ACS patients undergoing PCI were randomly assigned in a 2 × 2 factorial design to receive either a lower dose (75 to 100 mg) or a higher dose (300 to 325 mg) of aspirin with either a standard or a double dose of clopidogrel for 7 days (20). There was a significant interaction between high-dose aspirin and high-dose clopidogrel, suggesting that high-dose clopidogrel was more effective than standard-dose clopidogrel in the setting of high-dose aspirin. Based on the PLATO and CURRENT-OASIS 7 findings, controversy exists as to whether more intensive antiplatelet therapy (including ticagrelor and high-dose clopidogrel) has altered efficacy in the setting of high-dose aspirin. In TRITON-TIMI 38, prasugrel remains more effective than standard-dose clopidogrel at all doses of aspirin, and there is no interaction (in either direction) with aspirin and prasugrel for the primary or secondary efficacy endpoint.
The safety findings in our analysis also reflect no effect modification of discharge aspirin dose on the clinical bleeding observed with prasugrel versus clopidogrel. This finding is consistent across key bleeding endpoints, including TIMI major non-CABG bleeding, TIMI minor non-CABG bleeding, and TIMI major non-CABG bleeding (although only TIMI major/minor bleeding achieved statistical significance in both groups). In the CURE (Clopidogrel in Unstable Angina to Prevent Recurrent Events) study, which randomized patients with ACS to receive aspirin plus placebo versus aspirin plus clopidogrel, there was a higher bleeding risk observed with higher-dose aspirin, with or without clopidogrel, suggesting that bleeding may be additive when high-dose aspirin and clopidogrel are used simultaneously (21). In the CURRENT-OASIS 7 trial, there were no significant differences with respect to TIMI major bleeding or interaction between the aspirin dose and the higher dose of clopidogrel with respect to bleeding. Our study also found no such interaction, and prasugrel continued to confer a higher risk of bleeding than clopidogrel, regardless of whether low- or high-dose aspirin was used.
Aspirin dose at discharge is a postrandomization variable and therefore subject to confounding by indication, selection, and survival bias. As a result, we focused on the results of the randomized prasugrel versus clopidogrel comparison for our analysis in relation to discharge aspirin doses rather than comparing the efficacy and safety of aspirin doses directly. Because we analyzed subgroups based on discharge aspirin doses, because we excluded from our analysis those patients who had an in-hospital endpoint or bleeding event, and because there may be differences in in-hospital treatment, we cannot exclude the possibility of an early interaction or that (although unlikely given directional consistency) an interaction could have been observed had the individual subgroups been larger. In addition, there may have been some variation in aspirin dosing during the study due to adverse effects or changes in a patient’s clinical status. To address this limitation, we were able to demonstrate that a very high percentage of patients remained within the same aspirin dose group throughout the duration of the study as at the time of discharge. Finally, although we have accounted for baseline differences in our observational study through multivariable adjustment and propensity matching, residual unmeasured confounding may still exist.
These results provide contemporary evidence within the context of a trial of patients with moderate- to high-risk ACS treated with potent dual antiplatelet therapy that discharge aspirin dose does not modulate clinical outcomes of patients treated with prasugrel compared with clopidogrel. Notably, unlike the potential effect reported with high-dose aspirin and ticagrelor, there seems to be no clinically meaningful interaction of maintenance dose of aspirin with prasugrel.
For supplemental tables, please see the online version of this article.
The TRITON-TIMI 38 trial was supported by research grants from Daiichi Sankyo and Eli Lilly and Company. Ms. Murphy and Drs. Antman, Braunwald, and Wiviott have received research grants from Eli Lilly and Company, Daiichi Sankyo, and Sanofi-Aventis. Dr. Braunwald has received consulting fees or paid advisory board fees from Sanofi-Aventis; and lecture fees from Eli Lilly and Daiichi Sankyo. Dr. Wiviott is a consultant for Angelmed, Eisai, Aegerion, Janssen Pharmaceuticals, St. Jude Medical, and Xoma; and has received research funding from Merck & Co., Eisai, AstraZeneca, and Eli Lilly and Company/Daiichi Sankyo. Dr. Cannon has received research grants/support from Accumetrics, AstraZeneca, CSL Behring, Essentialis, Inc., GlaxoSmithKline, Merck & Co., Inc., Regeneron Pharmaceuticals, Inc., Sanofi, and Takeda; he is also on the advisory board (with funds donated to charity) of Alnylam, Bristol-Myers Squibb, Lipimedix, and Pfizer Inc.; and is a clinical advisor with equity in Automedics Medical Systems, Inc. Dr. Kohli has received consulting fees from Daiichi Sankyo. Dr. Udell has reported that he have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronymns
- acute coronary syndrome(s)
- adenosine diphosphate
- coronary artery bypass grafting
- confidence interval
- cardiovascular death
- hazard ratio
- non–ST-segment elevation myocardial infarction
- odds ratio
- percutaneous coronary intervention
- Thrombolysis in Myocardial Infarction
- Received June 8, 2013.
- Revision received September 8, 2013.
- Accepted September 16, 2013.
- American College of Cardiology Foundation
- Levine G.N.,
- Bates E.R.,
- Blankenship J.C.,
- et al.
- Mahaffey K.W.,
- Wojdyla D.M.,
- Carroll K.,
- et al.
- 8.↵Services USDoHH. U.S. Food and Drug Administration. Available at:http://www.fda.gov/NewsEvents/Newsroom/PressAnnouncements/ucm263964.htm. Accessed October 16, 2013.
- Wiviott S.D.,
- Antman E.M.,
- Gibson C.M.,
- et al.
- Silber S.,
- Albertsson P.,
- Aviles F.F.,
- et al.
- King S.B. 3rd.,
- Smith S.C. Jr..,
- Hirshfeld J.W. Jr..,
- et al.
- Anderson J.L.,
- Adams C.D.,
- Antman E.M.,
- et al.
- Wright R.S.,
- Anderson J.L.,
- Adams C.D.,
- et al.
- Capone M.L.,
- Tacconelli S.,
- Sciulli M.G.,
- et al.
- Pillinger M.H.,
- Capodici C.,
- Rosenthal P.,
- et al.
- Patrono C.,
- Ciabattoni G.,
- Patrignani P.,
- et al.
- Pocock S.,
- Calvo G.,
- Marrugat J.,
- Prasad K.,
- Tavazzi L.,
- Wallentin L.,
- et al.
- Peters R.J.,
- Mehta S.R.,
- Fox K.A.,
- et al.