Review of the ATLAS ACS 2-TIMI 51 trial
Anti-Xa Therapy to Lower Cardiovascular Events in Addition to Standard Therapy in Subjects with Acute Coronary Syndrome-Thrombolysis in Myocardial Infarction 51
Background Following an acute coronary syndrome, individuals have a substantially elevated risk of recurrent events, compared to a similar individual who did not experience ACS, despite the use of appropriate risk reducing therapies. This is termed “residual risk” and the concept has been mentioned in prior reviews. The quest to lower residual risk continues to be a major driver of new products and therapeutic concepts in cardiovascular medicine.
Rivaroxaban is an oral anticoagulant that directly and selectively inhibits factor Xa. Factor Xa initiates the final common pathway of the coagulation cascade resulting in the formation of thrombin. Thrombin promotes platelet aggregation.
At the time the trial was undertaken, both aspirin and thienopyridines were established agents that worked via different mechanisms, downstream of thrombin formation, to stop platelets from sticking together. Could the addition of an agent that inhibits thrombin formation, upstream of platelet activation and aggregation, reduce risk further without causing a prohibitive increase in bleeding risk?
A meta-analysis of small trials involving the use of warfarin, in addition to aspirin, suggested that warfarin, an indirect inhibitor of thrombin, could improve cardiovascular outcomes. And a phase 2 dose finding trial with rivaroxaban provided further support. Thus, the ATLAS ACS-TIMI 51 trial sought to test the hypothesis that adding rivaroxaban to standard therapies, at 2.5 or 5 mg twice daily, would reduce cardiovascular events compared to placebo.
Patients Adults who had presented with an acute coronary syndrome (STEMI, NSTEMI or unstable angina). Patients under 55 years of age had to have either diabetes or a history of a previous MI. Patients were excluded if they had a platelet count <90k, hemoglobin <10g/dl, creatinine clearance <30 ml/min, clinically significant GI bleed within 12 months, previous intracranial hemorrhage or previous stroke or TIA.
Baseline characteristics Patients averaged 62 years of age with 36% greater than or equal to 65 and only 9% greater than or equal to 75. The majority of patients were either white (73%) or Asian (21%). Existing medical conditions were typical for an ACS population. The index diagnosis was a STEMI in 50% with the other half split fairly evenly between NSTEMI and UA. The majority of patients enrolled were from Eastern Europe (39%) and Asia (21%). Nearly all patients were on aspirin (99%) and a thienopyridine (93%) and 83% were on a statin drug.
Procedures Patients were enrolled within 7 days after hospital admission for an ACS. They had to be stable at the time of enrollment with the initial revascularization strategies completed. They were randomly assigned in a 1:1:1 fashion to twice daily rivaroxaban 2.5 mg, rivaroxaban 5 mg or placebo with a maximum follow-up of 31 months. All patients were to receive standard medical therapy. Patient follow-up occurred at 4 weeks, 12 weeks and every 12 weeks thereafter.
Endpoints The primary endpoint was a composite of death from cardiovascular causes, myocardial infarction, or stroke (ischemic, hemorrhagic, or stroke of uncertain cause). The primary safety endpoint was TIMI major bleeding not related to CABG surgery.
As prespecified by the investigators, the efficacy analyses were performed with the use of a modified intention-to-treat principle; however, they performed sensitivity efficacy analyses using a standard intention-to-treat approach. *Since the statistical significance of the main findings did not meaningfully change with either approach, we report the p-values based on the standard approach since using a modified approach is unconventional.
The investigators determined they would need 983 primary endpoint events to provide a power of 96% to detect a 22.5% relative reduction between the combined-dose group receiving rivaroxaban and the placebo group with a 2-sided alpha of 0.05. This would also yield a power of 90% to determine a relative risk reduction of 22.5% for each rivaroxaban dose group compared to placebo separately.
Results There were 15,526 patients who underwent randomization (5,174 at 2.5 mg rivaroxaban, 5,176 at 5 mg rivaroxaban, and 5,176 with placebo). The median time from index event to randomization was 4.7 days.
Compared to placebo, patients receiving either dose of rivaroxaban experienced a significant reduction in the primary composite endpoint (8.9% vs 10.7%; HR 0.84; 95% CI 0.74-0.96; p=0.002). This was driven by reductions in cardiovascular death (3.3% vs 4.1%; p=0.05) and nonfatal MI (5.5% vs 6.6%; p=0.01). The difference in stroke numerically favored the placebo group but was not statistically significant (1.6% vs 1.2%; p=0.19). The difference in all-cause death numerically favored the combined rivaroxaban group but was not statistically significant (3.7% vs 4.5%; p=0.08).
TIMI major bleeding not associated with CABG was significantly increased in the combined rivaroxaban group (2.1% vs 0.7%; HR 3.96; 95% CI 2.46-6.38; p<0.001). All components of bleeding were significantly increased in the combined rivaroxaban group with the exception of fatal bleeding (0.3% vs 0.2%; p=0.66).
When specifically looking at the 2.5 mg or 5 mg rivaroxaban dose compared to placebo, no significant differences were noted in the primary composite endpoint or primary safety endpoint. However, the endpoints of death from CV causes (2.7% vs 4.1%; p=0.005) and all-cause death (2.9% vs 4.5%; p=0.02) favored the 2.5 mg dose of rivaroxaban vs placebo compared to the 5 mg dose of rivaroxaban vs placebo ([4.0% vs 4.1%; p=0.57 for CV death] & [4.4% vs 4.5%; p=0.89 for all-cause dearth]). Non-fatal MI, on the other hand, favored the 5 mg dose of rivaroxaban vs placebo (4.9% vs 6.6%; p=0.008) compared to the 2.5 mg dose (6.1% vs 6.6%; p=0.09).
In the manuscript abstract, the authors highlighted the outsized reductions in CV death and all-cause death observed in the 2.5 mg rivaroxaban dose compared to placebo; however, this was likely due to the play of chance from low statistical power as there was not a commensurate difference in fatal bleeding between the 2.5 mg and 5 mg rivaroxaban dose to account for these differences and non-fatal MI favored the 5 mg rivaroxaban dose.
Inspection of individual subgroups does not show any statistically significant interactions; however, possible treatment effect heterogeneity is evident for patients with a previous history of stroke or TIA (this represents a very small group) and for patients with diabetes.
Conclusions Compared to placebo, 2.5 or 5 mg of rivaroxaban twice daily significantly reduced the primary composite endpoint of cardiovascular death, non-fatal MI and stroke with a NNT of approximately 50 patients. It also significantly increased TIMI major bleeding not associated with CABG with a NNH of approximately 70 patients. No statistically significant difference was observed for all cause death.
In our opinion, the authors introduced spin in the manuscript abstract by highlighting outsized differences in CV death and all cause death in the 2.5 mg rivaroxaban dose vs placebo group that were more likely false positive findings from low power than real treatment response variation based on the difference in rivaroxaban dose.
Finally, it is unlikely that results from this trial can be generalized to the average patient with recurrent ACS events. Such patients are generally older with significant comorbidities like reduced hemoglobin levels, chronic kidney disease or cerebrovascular disease. These patients were either explicitly excluded or severely underrepresented in the trial.