Radial access in patients with acute coronary syndromes without persistent ST segment elevation: metaanalysis,and metaregression

Background: Despite more consistent evidence of benefit in STEACS, including reduced longterm mortality, radial access (RA) is now considered the access of choice in NSTEACS. However, no randomized trial has been conducted focused of this heterogeneous subgroup of patients, with prespecified substudies of RIVAL and MATRIX trials offering the sole randomized, albeit nonconclusive, data. Recently, however, several nonrandomized post hoc subanalyses of vascular access in trials of other interventions have been made available. Objectives: To ascertain whether RA is associated with improved outcomes also among NSTEACS patients, using metaanalytic techniques and pooling randomized and nonrandomized data sources. Methods: A systematic review of all the available evidence was performed. Classic metaanalysis techniques were employed. We focused on identifying and exploring the sources of heterogeneity, using a combination of fixed, random, and mixedeffects modelling. Controlweighted metaregression modelling was used to identify studylevel covariates that may explain heterogeneity. Results: Eleven studies (including ACUITY, OASIS5, EARLYACS, RIVAL, MATRIX, CHAMPION PHOENIX, ACCOAST trials) met inclusion criteria. The pooled analysis included 131.332 patients, 46.447 receiving RA and 84.885 receiving femoral access (FA). Thirtyday mortality was 20% lower in the RA group (p<0.001 with fixedand mixedeffects model, but p=NS with randomeffects model, I2=46%), and this result was critically dependent on the inclusion of one large observational nationwide database with propensity matching from United Kingdom, in which overall major bleeding rates were unexpectedly low (<1%). Thirtyday MACE rate was 10% lower in RA group (p≤0.001 with fixedand mixedeffects model, but p=NS with randomeffects model, I2=64%), and similarly influenced by this single study. Thirtyday major bleeding rates were consistently reduced by 50% with RA (p<0.001, I2=42%), albeit an inverse relationship between the effect size of RA in reducing major bleeding and the proportion of patients in each study receiving FA and experiencing major bleeding after the index procedure was evident (p=0.012). The association of RA with 25% reduced longterm mortality was of borderline significance (p=0.054 with randomeffects model, p=0.001 with fixedeffect model, I2=60%), was strongly influenced by a single posthoc study, and also depended on the proportion of FA patients in each study experiencing major bleeding (p=0.04). Conclusions: Overall, RA is associated with better outcomes as compared with FA in NSTEACS. However, large heterogeneity exists between and within subgroups of included studies. In particular, the benefit on shortterm mortality and MACE is influenced by one observational study, whereas the effect on major bleeding, and consequently on longterm mortality, appears to be related to the overall bleeding risk of the patients.