Propensity-matched cohort validates findings of the VALOR trial.

INTRODUCTION: The Evaluation of the Medtronic Vascular Talent Thoracic Stent Graft System for the Treatment of Thoracic Aortic Aneurysms (VALOR) trial findings noted superior 30-day and 1-year outcomes of the Talent thoracic endograft (Medtronic Vascular, Santa Rosa, Calif) compared with surgical repair of descending thoracic aneurysms (DTAs). Data from 195 prospective thoracic endovascular aneurysm repair (TEVAR) patients treated with the Talent device and 189 retrospective controls undergoing open surgical repair (OSR) from three centers of excellence were included in the trial after completion of TEVAR enrollment and compared. Such comparisons are biased by baseline differences among TEVAR vs OSR, however, propensity score (PS) analysis can reduce bias and validate such comparisons.

METHODS: Logistic regression was used to generate a PS (range, 0-1) to identify baseline characteristics more likely in TEVAR. The PS estimated the probability that any patient would undergo TEVAR (eg, a PS of 0.99 represents a 99% chance a patient belongs to TEVAR). PSs were then generated for all patients, and TEVAR and OSR patients were divided into tertiles based on the PS to reduce up to 80% of inherent bias. Outcomes from the middle tertile (T2), patients equally likely (midrange PS) to be in TEVAR or OSR and therefore best matched, were compared using regression analysis and were also compared with the outcomes in the overall trial group.

RESULTS: Correlates of membership in TEVAR were smaller aneurysm (P < .001), anticoagulants (P < .01), no previous abdominal aortic aneurysm (AAA) repair (P < .01), no peripheral vascular disease (P = .001), statin use (P = .002), aspirin use (P = .002), older age (P = .028), race (P = .007), male gender (P = .02), and heart failure (P = .035). T2 included 68 TEVAR (PS, 0.58 ± 0.2) and 67 OSR patients (PS, 0.46 ± 0.2). VALOR overall reported differences in aneurysm size (56 mm TEVAR vs 69 mm open) and prior AAA repair (19% TEVAR vs 37% open), and this adjusted to no differences in T2 patients. In the well-matched T2 cohort, TEVAR patients had similar 30-day mortality (0% vs 3% OSR; P = .2) and improved 1-year aneurysm-related mortality rates (0% TEVAR vs 8% OSR; P = .05) compared with the OSR patients. This finding was in concurrence with the VALOR trial reporting similar benefit in TEVAR patients. The all-cause 1-year mortality showed a favorable trend for TEVAR in the VALOR trial; however, in T2 patients, 1-year all-cause mortality was similar in both groups of patients (17% TEVAR vs 15% OSR; P = .8). Age (P = .01), history of cerebrovascular accident (P < .05), antiarrhythmia medication (P = .04), and renal disease (P < .03) independently predicted all-cause and aneurysm-related mortality by regression analysis.

CONCLUSIONS: PS analysis is an important tool for elimination of bias inherent when retrospective controls are used. Its application to VALOR validates the long-term benefit in aneurysm-related mortality conferred by TEVAR in patients undergoing endovascular DTA repair.