Author + information
- Received June 19, 2015
- Accepted June 23, 2015
- Published online September 1, 2015.
- Michel R. Le May, MD∗ (, )
- Kuljit Singh, MD and
- George A. Wells, PhD
- ↵∗Reprint requests and correspondence:
Dr. Michel R. Le May, Ottawa Heart Institute, 40 Ruskin Street, Ottawa, Ontario K1Y 4W7, Canada.
In the recently published MATRIX (Minimizing Adverse Haemorrhagic Events by TRansradial Access Site and Systemic Implementation of angioX) trial, the use of transradial access (TRA) compared to transfemoral access (TFA) during percutaneous coronary intervention (PCI) for acute coronary syndrome (ACS) was associated with a reduction in net adverse cardiovascular events. However, the results of MATRIX must be interpreted with caution due to several limitations including the strong modulating effect of operator/center experience on the relative efficacy of TRA and the inclusion of 2 distinct patient populations (ST-segment elevation and non-ST-segment elevation ACS). Therefore, although important, the results of MATRIX have strong limitations and are not sufficient to definitively identify an approach of choice during PCI for ACS. Further research is needed before strong, evidence-based recommendations regarding the approach of choice during PCI for ACS can be made.
A recently published, large, randomized study, MATRIX (Minimizing Adverse Haemorrhagic Events by TRansradial Access Site and Systemic Implementation of angioX) (1), compared transradial access (TRA) to transfemoral access (TFA) in patients presenting with acute coronary syndromes (ACS) who were referred for percutaneous coronary intervention (PCI). The results of this trial have been interpreted to suggest that TRA is superior to TFA in reducing net adverse clinical events (NACE) through a reduction of bleeding and mortality. This conclusion could significantly affect our practice guidelines and lead to a strong recommendation that the approach of choice for PCI in ACS is radial rather than femoral. Hence, this trial has significant implications for both PCI centers and interventionalists, and it could have an effect on medical practice and education. However, the MATRIX trial has serious shortcomings that need to be considered.
The MATRIX trial randomly assigned 8,404 ACS patients to TRA (n = 4,197) or TFA (n = 4,207) to compare clinical outcomes in patients referred for coronary angiography and PCI (1). The study was designed with 2 30-day coprimary endpoints: 1) major adverse cardiovascular events (MACE), defined as all-cause mortality, myocardial infarction, or stroke; and 2) NACE, defined as major bleeding unrelated to coronary artery bypass graft surgery or major adverse cardiovascular events. Major bleeding was classified according to the Bleeding Academic Research Consortium (2). Because of multiple comparisons, the 2-sided α was pre-specified at 0.025 for each primary endpoint. MACE was recorded in 8.8% of patients assigned to TRA and in 10.3% of patients assigned to TFA (p = 0.03); this was interpreted as nonsignificant. However, the rate of NACE was significantly lower in patients assigned to TRA compared to TFA (9.7% vs. 11.7%, respectively; p = 0.009); a difference said to be driven by major bleeding (1.6% vs. 2.3%; p = 0.013) and all-cause mortality (1.6% vs. 2.2%; p = 0.045). The authors suggest that the benefits associated with implementation of TRA for the treatment of ACS “might be especially relevant for countries such as the USA where use of the radial approach is currently uncommon” (3). However, a critical appraisal of the MATRIX trial’s results will cast a word of caution before accepting the authors’ conclusions.
First, it is quite clear that the outcomes were dependent upon the center’s experience at performing PCI. A center’s experience is determined by its access preference (i.e., the proportion of TRA vs. TFA) and by its annual PCI volume; in addition, the experience levels of the operator, the catheterization team, as well as the team monitoring the patient after the PCI all contribute to the overall experience of the center. The MATRIX study divided patients into 3 groups based on the participating center’s proportion of radial PCIs: “low” (14.9% to 64.4%), intermediate (65.4% to 79.0%), and high (80.0% to 98.0%). The results of this stratified analysis are shown in Figure 1. As noted by the authors, and illustrated in the figure, there is a strong interaction between the randomized mode of access and the center’s proportion of radial procedures for both MACE (p = 0.0048) and NACE (p = 0.0048). This interaction is so strong that to compare TRA and TFA without taking the center’s experience into consideration would be an oversight.
In fact, the only time TRA is significantly better than TFA occurs when the results are considered only for centers with a high proportion (80.0% to 98.0%) of PCIs done using TRA. There is no difference even when the proportion of TRA is as high as 79% (14.9% to 79.0%). Although the labels “low,” “intermediate,” and “high” are used, more appropriate labels would be “intermediate,” “high,” and “very high,” respectively, given the percentages that they represent. It is only in the “very high” group that there is a difference favoring TRA, and it occurs in centers with essentially no or very limited TFA experience. One could argue that operators in these centers have optimal TRA skills that enable the benefit of TRA to be more evident; however, it remains unexplained why the absolute rates of MACE and NACE in the TRA group were unexpectedly higher in centers with a “high” proportion of radial procedures compared with those in the “low” and “intermediate” centers. Furthermore, in the “high” radial proportion centers, the rates of MACE and NACE in the TFA group were excessive, 15.5% and 17.1%, respectively, compared with the rates reported for centers with a “low” or “intermediate” proportion of radial procedures. Notably, these results did not appear to be linked to the overall annual PCI volume.
An alternative explanation is that centers performing PCI almost exclusively by TRA have limited contemporary experience with TFA and consequently have more complications. Randomizing patients to receive TFA in centers with very little experience would obviously be detrimental for outcomes in the TFA group. With such an interaction, one needs to take center experience into consideration either by interpreting the results within levels of center experience or by statistically adjusting the results by including center experience as a covariate in any modeling. Hence, the analysis of this trial may have actually assessed center experience rather than the use of the access site itself, and this may very well account for the measured differences in clinical outcomes.
Second, the MATRIX trial had a complex design that attempted to resolve many questions by incorporating multiple comparisons: 1) TRA versus TFA; 2) bivalirudin monotherapy versus unfractionated heparin plus provisional glycoprotein platelet inhibitors (GPIs); and 3) short- versus long-term administration of bivalirudin. This approach likely introduced multiple interventions that could potentially distort the interpretation of the results.
In light of the multiple comparisons, the alpha for significance was set at 0.025 for the 2 primary outcomes (4). However, the p value was reset at 0.05 for the individual components of MACE and NACE. One could argue that, given the 4 components of NACE, a Bonferroni-corrected alpha of 0.0125 should have been used. This argument coincides with many authors’ recommendations, most recently by Rauch et al. (5), that strategies such as the Bonferroni-Holm’s approach should be used when evaluating the components of a composite outcome. In so doing, the difference reported for mortality in the MATRIX trial would not have been statistically significant.
Third, the MATRIX trial enrolled patients presenting with ST-segment elevation myocardial infarction (STEMI) and non–ST-segment elevation acute coronary syndrome (NSTE-ACS). Of note, the MATRIX trial did not stratify STEMI and NSTE-ACS in the randomization process (4). Without including stratification into the study design, one needs to exercise caution in the interpretation of the results in these subgroups as the 2 clinical entities differ considerably in pathophysiology and management options, thus possibly skewing the results. Notable differences between both patient populations include: 1) the acuity level; 2) the importance of time to reperfusion; 3) the choice and duration of anticoagulant therapy; and 4) the likelihood of bleeding (6).
The MATRIX trial enrolled a total of 4,010 of 8,404 (47.7%) STEMI patients. A subgroup analysis of these STEMI patients found no difference in MACE (121 of 2,001 [6.0%] in the TRA group vs. 126 of 2,009 [6.3%] in the TFA group; p = 0.77), and no difference in NACE (142 of 2,001 [7.1%] in the TRA group vs. 165 of 2,009 [8.2%] in the TFA group, p = 0.19).
Interestingly, a similar study, the RIVAL (RadIal Vs femorAL access for coronary intervention) trial (7), compared TRA with TFA in 7,021 patients presenting with ACS, of which 1,958 presented with STEMI (8). The primary outcome, like in the MATRIX trial, was defined as NACE, a composite of death, myocardial infarction, stroke, or non-coronary artery bypass graft surgery–related major bleeding within 30 days; the primary outcome, NACE, was measured in 3.7% of patients in the TRA group and in 4.0% of patients in the TFA group (p = 0.50). Hence, RIVAL was a negative study. In addition, the RIVAL study showed no significant difference in mortality between TRA and TFA and no significant difference in reinfarction, stroke, or major bleeding. Despite these negative results, the RIVAL study claimed a significant reduction in mortality in the subgroup of patients with STEMI treated with TRA (1.3% vs. 3.2%; p = 0.006) (8). However, amongst NSTE-ACS patients, there was a troublesome trend toward a higher mortality in the TRA group (1.2% vs. 0.8%; p = 0.08). With respect to NACE, NSTE-ACS patients benefitted less from TRA than those with STEMI (interaction p = 0.026); in contrast, the MATRIX trial identified a trend indicating that TRA benefited NSTE-ACS patients (relative risk: 0.84; 95% confidence interval: 0.72 to 0.97) rather than STEMI patients.
It remains unclear why differences in clinical outcomes in STEMI and NSTE-ACS populations were discordant between the 2 trials. Because many confounding factors could affect the results, it may not be appropriate to include these 2 populations in a single trial assessing the access site for PCI, unless stratification is included in the study design.
Fourth, MATRIX involved drugs such as GPIs, which have been shown to increase bleeding and mortality (9,10). In a pooled analysis of patients enrolled in PCI randomized trials comparing bivalirudin to heparin plus GPIs, non–access site bleeding represented two-thirds of all bleeding events and was associated with a 4-fold increase in 1-year mortality; access site bleeding represented one-third of the bleeds and was associated with a 2-fold increase in 1-year mortality (11). Even if TRA could reduce most access site bleeds, it would likely not reduce non–access site bleeds, which not only constitute a significant proportion of total bleeds but are associated with a greater risk of mortality. Furthermore, an important reduction in total bleeding events would be expected if GPIs were avoided altogether.
Fifth, MATRIX did not report on the use of vascular closure devices (VCDs). These devices have been reported to reduce femoral bleeding. Amongst 12,937 patients undergoing angiography/PCI, VCDs were associated with a nearly 50% propensity-adjusted reduction in vascular complications (12). In the ACUITY (Acute Catheterization and Urgent Intervention Triage Strategy) trial, the use of a VCD (Angio-Seal, St. Jude Medical, St. Paul, Minnesota), bivalirudin monotherapy, or both minimized rates of major access site bleed, 2.5% in VCD versus 3.3% with no VCD (relative risk: 0.76; 95% confidence interval: 0.61 to 0.94); bleeding was lowest in patients treated with both bivalirudin monotherapy and a VCD (0.7%) (13). In the RADIAMI II (Radial Versus Femoral Approach for Percutaneous Coronary Interventions in Patients With Acute Myocardial Infarction) trial, the incidence of bleeding complications with primary PCI were not significantly different between TRA and TFA plus a VCD (StarClose, Abbott Laboratories, Abbott Park, Illinois) (14). Similarly, in the RIVAL trial, 25.6% of patients assigned to TFA received a VCD: bleeding was 0.7% with TRA and 0.7% with TFA with a VCD (vs. 1% for TFA without a VCD) (7). Furthermore, the use of VCD improves patient comfort by eliminating the need for a femoral clamp, and has become standard practice in many centers.
Finally, cumulative radiation exposure is a potential risk for the interventionalist and should therefore be considered when evaluating the efficacy of both access sites. In the MATRIX study, radiation exposure was not reported. In the RIVAL study, TRA was associated with longer fluoroscopy time: 9.8 min (5.8 to 15.0 min) vs. 8.0 min (4.5 to 13.0 min); p < 0.0001 (7). This observation is of concern in view of a recent report on excessive malignancies found in the left cerebral hemisphere amongst interventionalists (15). This is an important secondary outcome that should be of great interest to the interventional community, which houses high-volume operators who are progressively accommodating TRA as the preferred choice. Cumulative small differences in radiation exposure may become significant throughout the career of high-volume operators. Because interventional cardiologists have the highest radiation exposure amongst health professionals, studies “further delineating occupational risks are essential” (15).
In summary, the conclusion reached by the authors of the MATRIX trial, stating that TRA reduces net adverse clinical outcomes as compared with TFA in patients with ACS undergoing invasive assessment, needs to be interpreted with caution. Although, the MATRIX trial was designed to resolve clinically relevant questions, the trial had several shortcomings. Primarily, the analyses of the results clearly demonstrate that operator/center experience is a key variable influencing the measured outcomes. Moreover, the trial introduced too many confounders in an attempt to resolve multiple questions. Thus, the interpretation of the culminated results may be distorted, given the multiple interactions and the inclusion of 2 distinct patient populations. Finally, the MATRIX trial did not apply pharmaceutical strategies or report on the usage of vascular closing devices that may reduce bleeding. The MATRIX trial also did not report on cumulative radiation exposure, which should be an important risk factor incorporated into the efficacy evaluation of both access sites.
A change in guidelines favoring TRA may not be appropriate for all patients, institutions, and cardiologists. It would be undesirable for skilled and well-established interventionalists who have been highly successful with TFA to change practice without the results of a trial specifically designed to resolve the debate. It would also be undesirable for interventionalists who practice in lower-volume centers to feel compelled to change practice based on the current evidence as this may not benefit patients. Furthermore, in some clinical situations, such as cardiogenic shock, proficiency at performing PCI via the TFA could ensure better outcomes. Therefore, it is important that interventional programs aim to ensure competency at performing the 2 approaches. Additional trials, particularly in the STEMI population, are needed to resolve the debate on access site looking at mortality as a primary outcome and should be conducted in centers where there is a reasonable balance in the centers’ experience at performing PCI via either access mode. From these future trials, the question of whether it is the “operator’s experience or the operation itself that matters,” can finally be answered.
The authors thank Christina Osborne for her assistance in editing the manuscript.
The authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- acute coronary syndromes(s)
- major adverse cardiovascular events(s)
- net adverse clinical events(s)
- non–ST-segment elevation acute coronary syndrome
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
- transfemoral access
- transradial access
- Received June 19, 2015.
- Accepted June 23, 2015.
- American College of Cardiology Foundation
- Mehran R.,
- Rao S.V.,
- Bhatt D.L.,
- et al.
- Bradley S.M.,
- Rao S.V.,
- Curtis J.P.,
- et al.
- Mehran R.,
- Pocock S.J.,
- Nikolsky E.,
- et al.
- Mehta S.R.,
- Jolly S.S.,
- Cairns J.,
- et al.
- Stone G.W.,
- Witzenbichler B.,
- Guagliumi G.,
- et al.
- Sanborn T.A.,
- Ebrahimi R.,
- Manoukian S.V.,
- et al.
- Chodor P.,
- Kurek T.,
- Kowalczuk A.,
- et al.