An invasive strategy including percutaneous coronary intervention (PCI) improves clinical outcomes in patients with ST-segment elevation myocardial infarction (STEMI) and in high-risk patients with non–ST-segment elevation acute coronary syndromes (NSTEACS) ((1),2). However, because these patients are also treated with multiple antithrombotic and antiplatelet therapies, they are at increased risk of bleeding complications. Bleeding has been linked with higher mortality in several large observational studies ((3),(4),5). Randomized trials have suggested that antithrombotic treatments with fewer bleeding complications may lead to improved longer-term clinical outcomes, including mortality ((6),(7),8). In observational studies of patients undergoing PCI, radial artery access reduced access site–related bleeding compared with femoral artery access ((9),(10),(11),12), and in some studies, this benefit was strongly associated with improvements in mortality ((13),(14),(15),16). However, these observational studies are limited by unmeasured confounding and selection bias; there have been very few randomized trials to adequately evaluate this relationship.

RIVAL (A Trial of Trans-radial Versus Trans-femoral Percutaneous Coronary Intervention (PCI) Access Site Approach in Patients With Unstable Angina or Myocardial Infarction Managed With an Invasive Strategy) was a multinational randomized trial involving 7,021 patients with either NSTEACS or STEMI that tested the hypothesis that a radial artery approach would reduce bleeding and major cardiovascular events compared with a femoral approach (17). There was no significant difference in the primary composite outcome of death, myocardial infarction (MI), stroke or non–coronary artery bypass graft (CABG) –related major bleeding, although there was a substantial reduction in major vascular access site complications favoring the radial approach. However, patients with STEMI differ from those with NSTEACS because they are exposed to more potent antithrombotic therapies, have higher risk-adjusted rate of bleeding complications (18), and undergo PCI with much higher frequency than patients with NSTEACS. We therefore hypothesized that there might be differences in treatment effect of radial versus femoral artery access site in patients with STEMI and NSTEACS enrolled in RIVAL.

RIVAL was a randomized, parallel-group, multicenter trial of radial versus femoral access site intervention (17). The study protocol was previously reported ((19),20). Briefly, patients were included if they had either STEMI or NSTEACS and a planned invasive approach, and the interventional cardiologist was willing to proceed with either radial or femoral access (and had expertise in both, including at least 50 radial procedures within the previous year). Patients were required to have intact dual circulation of the hand as assessed by an Allen's test. Patients were ineligible if they had cardiogenic shock, severe peripheral vascular disease precluding a femoral approach, or previous coronary bypass surgery with an internal mammary artery graft. The study was approved by all appropriate national regulatory authorities and the ethics committees of participating centers. All patients provided written informed consent to participate before enrollment.

Patients were randomly assigned to radial or femoral access by a 24-h computerized, central automated voice response system. This pre-specified analysis of the trial evaluates outcomes separately in patients with a pre-randomization diagnosis of STEMI or NSTEACS. STEMI patients were defined as those presenting with ischemic symptoms >20 min with ST-segment elevation of >2 mm in 2 contiguous pre-cordial leads or >1 mm in 2 contiguous limb leads or new left bundle branch block. Patients with NSTEACS were required to have unstable ischemic symptoms and electrocardiographic changes compatible with new ischemia or increased cardiac biomarkers. Patients 60 years of age and younger with normal cardiac biomarkers were also eligible if they had documented evidence of coronary artery disease. The primary efficacy outcome of RIVAL was the occurrence of death, myocardial infarction, stroke, or non-CABG–related major bleeding within 30 days. Key secondary outcomes were: 1) death, myocardial infarction, or stroke; and 2) non-CABG–related major bleeding at 30 days. Other secondary outcomes included components of the primary outcome and major vascular access site complications.

Detailed outcome definitions were previously reported ((19),20). In brief, major bleeding was defined as bleeding that 1) was fatal; 2) resulted in transfusion of ≥2 units of blood; 3) caused substantial hypotension with the need for inotropes; 4) needed surgical intervention; 5) caused severely disabling sequelae; 6) was intracranial and symptomatic or intraocular and led to significant visual loss; or 7) led to a decrease in hemoglobin of at least 50 g/l. ACUITY (Acute Catheterization and Urgent Intervention strategy) non-CABG–related major bleeding was defined as RIVAL major bleeding, large hematomas (greater than what would be normally be expected), and pseudoaneurysms requiring intervention. Minor bleeding was defined as bleeding events that did not meet the criteria for a major bleed and required transfusion of 1 unit of blood or modification of the drug regimen (i.e., cessation of antiplatelet or antithrombotic therapy). Major vascular access site complications included a pseudoaneurysm needing closure, a large hematoma, an arteriovenous fistula, or an ischemic limb needing surgery. These complications were classified as a major bleeding event only they also met the above definition of major bleeding.

Categorical variables in each cohort and between cohorts were compared by the chi-square test and continuous variables by the Student t test (for normally distributed variables) or nonparametric Wilcoxon sum rank test (for non-normally distributed variables). All analyses were by intention to treat (assignment to either radial or femoral access). Outcomes of patients randomized to radial versus femoral artery access were stratified according to pre-randomization diagnosis STEMI or NSTEACS. Cumulative event rates were determined from time-to-event data and are displayed through the use of Kaplan-Meier plots. Comparisons between groups were made using of the log-rank test. Tests for interaction were performed to determine whether there was heterogeneity in treatment effect of radial versus femoral access site with pre-randomization diagnosis (STEMI vs. NSTEACS). A multivariable analysis using a Cox proportional hazards model was performed to examine whether the statistical interaction between treatment and pre-randomization diagnosis on mortality was independent of baseline variables, operator, and center volume.

We enrolled a total of 7,021 patients with ACS; 1,958 with a pre-randomization diagnosis of STEMI and 5,063 patients with NSTEACS. Compared with patients with NSTEACS, patients with STEMI were younger and more likely to be male and to smoke (Table 1). They also had lower rates of previous diabetes, MI, and PCI. STEMI patients had significantly greater use of glycoprotein IIb/IIIa inhibitors, clopidogrel, and heparin, whereas fondaparinux and bivalirudin were used more commonly in patients with NSTEACS. Fibrinolytic therapy was used in 12% of patients with STEMI and in no patients with NSTEACS. Characteristics of the randomized treatment groups were well matched in each cohort with the exception of unfractionated heparin, which was used more commonly in STEMI patients allocated to radial access, and bivalirudin, which was used more commonly in STEMI patients to allocated femoral access (Table 1).

STEMI patients had much higher rates of PCI compared with NSTEACS patients (85% vs. 59%, p < 0.0001) (Table 2). In the STEMI group, operators had significantly more experience in radial artery access (median, 400 radial procedures/year) than operators in the NSTEACS group (median of 326 radial procedures/year, p < 0.0001). By contrast, femoral artery experience was higher among operators performing procedures in patients with NSTEACS cohort (median, 428 procedures/year) compared with operators in the STEMI patients (300 procedures/year, p < 0.0001). Total PCI volume was similar between STEMI and NSTEACS operators (median, 300 vs. 300 procedures/year). Drug-eluting stents were used almost twice as frequently in NSTEACS patients (44%) compared with STEMI patients (23%).

Procedural characteristics were well matched between the radial versus femoral artery randomized groups with no significant differences except for sheath sizes, which were smaller in the radial group compared with the femoral group and the use of drug-eluting stents in the NSTEACS population, which was slightly more common among patients randomized to the radial approach (Table 2). There were no significant differences in operator procedure volume between the radial and femoral groups.

For the primary outcome of death, MI, stroke or non-CABG–related major bleeding, there was a significant interaction (p = 0.025) between randomized treatment (radial or femoral artery access) and pre-randomization diagnosis (STEMI or NSTEACS) ((Figure 50_gr1), Table 3). We also found significant interactions for the secondary composite of death, MI, or stroke (p = 0.011) and for mortality alone (p = 0.001).

Among STEMI patients, the primary outcome occurred in 3.1% of patients randomized to radial access compared with 5.2% of patients randomized to femoral access (HR: 0.60; 95% CI: 0.38 to 0.94; p = 0.026) ((Figure 50_gr1)A, Table 3). Compared with femoral access, radial access also reduced the first secondary composite outcome of death, MI, or stroke (2.7% vs. 4.6%; HR: 0.59; 95% CI: 0.36 to 0.95; p = 0.031). This benefit was driven mainly by a reduction in mortality with radial artery access (1.3% vs. 3.2%; HR: 0.39; 95% CI: 0.20 to 0.76; p = 0.006), with similar rates of MI and stroke ((Figure 50_gr2)A, Table 3). The other secondary outcome of non-CABG major bleeding occurred infrequently and was not significantly different between the groups (Table 3).

Among patients with NSTEACS, the primary outcome occurred in 3.8% randomized to radial artery intervention compared with 3.5% randomized to femoral artery intervention (HR: 1.11; 95% CI: 0.83 to 1.48; p = 0.49) ((Figure 50_gr1)B, Table 3). There were no reductions in either of the 2 secondary outcomes or in any of the individual components of the primary outcome (Table 3). Moreover, among NSTEACS patients who underwent PCI, we found no significant difference in the primary outcome between the radial (3.63%, n = 1,486) and femoral (3.38%, n = 1,507) groups (HR: 1.07; 95% CI: 0.73 to 1.57; p = 0.73). In patients with unstable angina (n = 1,903; PCI rate: 53.9%), the primary outcome was 2.40% in the radial group and 2.65% in the femoral group (HR: 0.90; 95% CI: 0.51 to 1.59), and there was a reduction in ACUITY major bleeding (1.20% vs. 3.87%; HR: 0.31: 95% CI: 0.16 to 0.60). Similarly, in patients with non–ST-segment elevation myocardial infarction (n = 3,160; PCI rate: 62.2%) , the primary outcome was similar in the radial and femoral artery groups (4.76% radial vs. 3.92% femoral; HR: 1.21; 95% CI: 0.87 to 1.70) with a reduction in ACUITY major bleeding (2.26% vs. 4.80%; HR: 0.47; 95% CI: 0.31 to 0.70).

PCI success rates were similar in the radial and femoral groups in both STEMI and NSTEACS patients. ACUITY major bleeding (defined as RIVAL major bleeding + large hematomas + pseudoaneurysms requiring closure) occurred less frequently in patients with STEMI allocated to radial artery access (HR: 0.49; 95% CI: 0.28 to 0.84; p = 0.009) as well as in patients with NSTEACS (HR: 0.41; 95% CI: 0.29 to 0.58; p < 0.0001), without significant heterogeneity (Table 3). Similarly, major vascular access site complications alone were reduced with radial access in both STEMI and NSTEACS, with no significant heterogeneity (Table 3). Access site crossover was higher in the radial group compared with the femoral group, and this was consistent in both STEMI and NSTEACS cohorts.

Among patients with STEMI undergoing PCI, 1,451 patients (74%) received a primary PCI and 507 patients (26%) received a secondary PCI (i.e., 3% facilitated, 12% rescue, or 11% routine adjunctive). For the primary outcome, there was no heterogeneity in treatment effect in patients receiving a primary versus secondary PCI (interaction p = 0.79). Among patients undergoing primary PCI, 30-day mortality occurred in 1.4% of patients randomized to radial access versus 3.07% randomized to femoral access (HR: 0.46; 95% CI: 0.22 to 0.97; p = 0.041). Major vascular access site complications (1.4% vs. 4.0%; HR: 0.35; p = 0.005) and ACUITY-defined major bleeding (1.86% vs. 4.68%; HR: 0.39; p = 0.004) were substantially lower with radial access in primary PCI patients. Overall time from hospital presentation to PCI start was not significantly different between the radial and femoral access groups (Table 2). However, time from randomization to the end of PCI was 5 min longer in the radial artery group (58 min vs. 53 min, p = 0.0009).

In a multivariable model for death, the interaction between pre-randomization diagnosis (STEMI vs. NSTEACS) and randomized treatment (radial vs. femoral access) remained highly significant (p = 0.0001), after adjustment for baseline variables, center radial volume, and operator radial experience (Table 4). Analyses focusing on bivalirudin revealed no significant interaction in the primary outcome between access site allocation and a bivalirudin-based anticoagulation strategy and one that did not include bivalirudin in the overall cohort of STEMI plus NSTEACS patients (p for interaction = 0.2823). Similarly, among patients with STEMI, there was also no significant interaction between a bivalirudin-based strategy and access site allocation (p for interaction = 0.5519).

Other independent predictors of death included older age, diabetes, current smoking, and PCI performed at a low- (compared with a high-) volume radial center.

Among patients with STEMI, those who died by day 30 were more likely to have experienced a major bleeding event compared with those who survived (11.0% vs. 1.0%, p < 0.0001) (Table 5). Other factors significantly associated with death included femoral artery access site allocation, older age, female sex, CABG surgery after randomization, unsuccessful PCI, stent thrombosis, and new MI (Table 5).

RIVAL is the largest randomized comparison of radial and femoral artery access in patients with STEMI as well as in those with NSTEACS. In patients with STEMI, radial artery access significantly reduced the primary outcome and the secondary outcome of death, MI, or stroke as well as all-cause mortality. In patients presenting with NSTEACS, we found no significant differences in any of these outcomes. In both STEMI and NSTEACS patients, radial access reduced major vascular access site complications and major bleeding as defined by the ACUITY definition.

In STEMI patients, the reduction in the primary and secondary composite outcomes was driven mainly by a reduction in mortality with a directionally consistent reduction in MI. No such benefit was observed in patients with NSTEACS. In the multivariable analysis, this interaction between pre-randomization ACS type and access-site allocation remained highly significant, even after adjustment for baseline variables, operator radial experience, and center radial volume. There are several plausible reasons why a differential response to radial versus femoral access on mortality might occur. First, within the first 30 days, STEMI patients are at higher risk of mortality compared with NSTEACS patients (3.19% vs. 0.76%, respectively in the femoral access site groups), in whom nonfatal ischemic events (i.e., new MI) are more common ((21),22). Therefore, if a reduction in bleeding-related complications was associated with lower mortality, it might most likely be detected in the STEMI group of patients. Second, STEMI patients generally undergo a much higher rate of PCIs (>90%) compared with NSTEACS patients (50% to 60%), exposing them to a higher frequency of access site complications. Third, STEMI patients are often treated with more potent initial and subsequent antiplatelet and antithrombotic therapies (as well as fibrinolytic therapy) compared with patients with NSTEACS. Therefore, the risk-adjusted rate of bleeding (particularly access-site bleeding) is higher, making the association between bleeding and mortality more readily detectable in this population.

In RIVAL, we observed substantially lower rates of major vascular access site complications and ACUITY-defined bleeding with radial access in both the STEMI and NSTEACS cohorts. The rate of bleeding using the more conservative RIVAL study definition, which excluded major vascular access site complications (unless they led to death, hemoglobin decrease of 5 g/dl, blood transfusion of >2 units, or surgery), was very low (<1%), and consequently no difference was found between the 2 groups. Despite this, we found much higher 30-day mortality rates among those patients who had a non-CABG–related major bleeding event compared with those who did not in both the STEMI and NSTEACS populations, irrespective of whether the RIVAL or ACUITY bleeding definition was used. One possibility is that in the STEMI population, operators were significantly more experienced, as were centers in the performance of transradial intervention, compared with the NSTEACS population. There is clearly a learning curve for radial artery intervention, and there may be a threshold before significant benefits with this procedure are observed. Despite this, in a multivariable model of predictors of mortality in RIVAL, the interaction between pre-randomization ACS type and access site allocation remained highly significant, even after adjustment for baseline variables, operator and center experience, indicating that it was independent of operator and center radial access experience.

Our study has several strengths. It is the largest randomized comparison of radial and femoral access site approaches in patients with STEMI. The trial consisted of experienced operators and high-volume radial access site centers. Because RIVAL included both STEMI and NSTEACS patients in large numbers, it allowed us to compare and contrast the relative benefits and risks of radial intervention in these patients and explore reasons for a possible differential response.

Limitations of our analysis also need to be considered. The overall result of RIVAL on the primary outcome was neutral, so it may not be appropriate to look at subgroups because the overall result of the trial may be the most reliable ((23),24). Replication in independent randomized trials would strengthen the conclusions of our analysis. In the RIFLE-ACS (Radial Versus Femoral Investigation in ST Elevation Acute Coronary Syndrome) trial of 1,001 patients with STEMI randomized to radial or femoral intervention, rates of bleeding due to access site complications (12.2% vs. 7.8%, p = 0.026) and subsequent mortality (9.2% vs. 5.2%, p = 0.020) were lower using the radial artery approach (27). Data from this independent study are consistent with the results of the larger STEMI population in RIVAL. Second, there were exceedingly few RIVAL-defined major bleeding events in this trial (<1%). Although we demonstrated a much higher mortality rate in patients who experienced a major bleed compared with those who did not, the low frequency of events may have impeded our ability to determine whether a reduction in major bleeding could have affected longer-term mortality. Bleeding is an outcome that is definition dependent ((25),26). Using the ACUITY definition of major bleeding, there were substantial reductions in bleeding in both the STEMI and NSTEACS cohorts in RIVAL. Third, because centers participating in RIVAL were highly experienced in the radial technique, similar outcomes may not apply in centers performing lower volumes. Fourth, the ACUITY definition of major bleeding used in RIVAL included large hematoma (defined as large if it prolonged hospitalization), which differs slightly from the ACUITY definition of hematoma (>5 cm). Finally, unfractionated heparin was the anticoagulation strategy most commonly used in RIVAL, with few patients receiving bivalirudin. However, in the HORIZON-AMI (Harmonizing Outcomes With Revascularization and Stents in Acute Myocardial Infarction Trial), there was a consistent benefit of the radial technique in patients receiving bivalirudin as well as those receiving heparin plus glycoprotein IIb/IIIa inhibitors (28).

In this large randomized comparison, radial artery access reduced the primary outcome and mortality in patients presenting with STEMI. There was no such benefit in patients presenting with NSTEACS. This interaction between ACS type and access site allocation was independent of operator radial experience and center radial volume. These data suggest that radial artery access might be the preferred option in patients with STEMI.