Abstract 2449: Clinical Outcomes in Women with Stable Coronary Artery Disease Randomized to Optimal Medical Therapy With or Without PCI: Pre-Specified Subset Analysis of the COURAGE Trial
Background: The main results of the COURAGE Trial showed no significant differences in the primary outcome of all-cause mortality or non-fatal MI, or in major secondary outcomes (the composites of death, MI or stroke, and hospitalization for ACS) in the 2,287 patients with stable coronary artery disease randomized to optimal medical therapy (OMT) with or without percutaneous coronary intervention (PCI). In a pre-specified subset analysis, women who received an initial strategy of PCI during a median 4.6 year follow-up appeared to benefit for death or MI, but the interaction P value for this comparison (0.03) exceeded the nominal pre-specified value for all trial subgroup interactions (P<0.01) .
Methods: We explored various pre-specified clinical endpoints in the 338 women (15%) of the COURAGE cohort who were randomized to PCI + OMT vs. OMT alone. Chi-Square analyses were conducted for the following outcomes between groups (time to first event): death or MI; death; MI; the composite of death, MI or stroke; death, MI or hospitalization (Hosp) for ACS; and Hosp for CHF.
Results: There were no significant differences between treatment arms for major pre-specified cardiovascular events in men, but there was apparent benefit for many clinical outcomes in women randomized to PCI + OMT vs. OMT alone, although in all but one comparison (i.e., CHF), the P values exceeded the pre-specified value for interaction (P < 0.01) during long-term follow-up (Table⇓):
Conclusions: Although the primary endpoint and death alone were not significantly different between groups, women randomized to PCI appeared to fare better than those randomized to OMT alone for selected clinical outcomes. However, because women comprised only 15% of the COURAGE population, it is unclear whether these findings indicative true clinical benefit for PCI in women, or alternatively may represent a statistical play of chance or type II error. These important findings may warrant additional investigation and analysis.