Temporal validation of a prognostic model for mortality of patients with AAA treated with EVAR

Abstract

Abstract Objective The role of EVAR in patients with asymptomatic AAA not suitable for open surgical repair has been questioned. The aim of this study was to temporally validate a previously published predictive model for survival after EVAR in asymptomatic AAA. Methods This retrospective observational cohort study included all consecutive patients treated with standard EVAR for asymptomatic AAA at a tertiary center between 2001 and 2020. A predictive Cox model for survival after EVAR in patients treated between 2001 and 2012 (Cohort A) has been published but not yet implemented. Older age, lower eGFR, and COPD were found to be independent predictors for worse survival after EVAR. This Cox model was temporally validated using patients treated at the same institution between 2013 and 2020 (Cohort B) and then updated using the overall cohort. A risk score with four different risk groups was built to predict the 5-year survival and internally validated with resampling methods. Results A total of 558 patients (91.2% males, mean age 74.9 years) were treated for asymptomatic AAA with EVAR. At 5 years, mortality was 35.9% in Cohort A, and 36.2% in Cohort B, p=.102 (log-rank). The original model based on Cohort A was tested on Cohort B and showed moderate to good discrimination ability (Harrell's C: 0.68, 95%-CI: 0.62–0.75). Older age, lower eGFR, and COPD were confirmed as independent predictors for worse survival after EVAR. The updated model showed good discrimination ability (C=0.70, 95%-CI: 0.66–0.75) and was robust in bootstrap validation. The risk score was built using the three variables and split into four risk groups. Patients in the “high risk” groups are expected to have a 5-year survival probability of only 40% (95%-CI: 32–50%), whereas patients in the “low risk” group and have an excellent 5-year survival probability of 89% (95%-CI: 84–95%). Patients in the “low-to-moderate” and “moderate-to-high” risk group have 5-year survival probabilities of 83% (95%-CI: 76–90%) and 68% (95%-CI: 60–79%), respectively. Conclusion Age, eGFR and COPD are independent predictors of survival after EVAR. This predictive model could identify low-risk patients with a favourable life expectancy who would likely benefit from EVAR and high-risk patients with a short life expectancy who may not benefit from EVAR at the current diameter threshold.