Preoperative prediction of mortality within 1 year after elective thoracic endovascular aortic aneurysm repair

Abstract

Objective: Thoracic endovascular aortic repair (TEVAR) is known to have a survival benefit over open repair in patients with descending thoracic aneurysms and has become a mainstay of therapy. Because death before 1 year after TEVAR likely indicates an ineffective therapy, we have created a predictive model for death within 1 year using factors available in the preoperative setting.

Methods: A registry of 526 TEVARs performed at the University of Florida between September 2000 and November 2010 was queried for patients with degenerative descending thoracic aneurysm as their primary pathology. Procedures with emergent or urgent indications were excluded. Preoperatively available variables, such as baseline comorbidities, anatomic-, and procedure-specific planning details, were recorded. Univariate predictors of death were analyzed with multivariable Cox proportional hazards to identify independent predictors of 30-day (death within 30 days) and 1-year mortality (death within 1 year) after TEVAR.

Results: A total of 224 patients were identified and evaluated. The 30-day mortality rate was 3% (n = 7) and the 1-year mortality rate was 15% (n = 33). Multivariable predictors of 1-year mortality (hazard ratios [95% confidence interval]) included: age >70 years (5.8 [2.1-16.0]; P = .001), adjunctive intraoperative procedures (eg, brachiocephalic or visceral stents, or both, concomitant arch debranching procedures; 4.5 [1.9-10.8]; P = .001), peripheral arterial disease (3.0 [1.4-6.7]; P = .006), coronary artery disease (2.4 [1.1-4.9]; P = .02), and chronic obstructive pulmonary disease (1.9 [1.0-3.9]; P = .06). A diagnosis of hyperlipidemia was protective (0.4 [0.2-0.7]; P = .006). When patients were grouped into those with one, two, three, or four or more of these risk factors, the predicted 1-year mortality was 1%, 3%, 10%, 27%, and 54%, respectively.

Conclusions: Factors are available in the preoperative setting that are predictive of death within 1 year after TEVAR and can guide clinical decision making regarding the timing of repair. Patients with multiple risk factors, such as age ≥ 70 years, coronary artery disease, chronic obstructive pulmonary disease, and a need for an extensive procedure involving adjunctive therapies, have a high predicted mortality within 1 year and may be best served by waiting for a larger aneurysm size to justify the risk of intervention.

Figure 1

Kaplan-Meier curve with life table for all patients after TEVAR for descending thoracic aneurysm. All displayed intervals have less than 10% standard error of the mean.

Figure 2

Thirty-day versus 1-year mortality stratified by risk factor group. The majority of deaths within one year occurred outside of the first 30 days after TEVAR.

Figure 3

Kaplan-Meier curve for 1-year survival after TEVAR for descending thoracic aneurysm stratified by risk factor group. *The defined risk factor groups were compared with the log-rank test.

Figure 4

Predicted and actual mortality within 1-year after TEVAR stratified by risk factor group. The C index of the multivariable model for mortality within one year of TEVAR was satisfactory at 0.77 and calibration of the model was assessed by determination of the slope of a plot of predicted versus observed events and found to be acceptable with a slope of 0.94 and an intercept of 0.6% (slope of 1 and intercept of 0 representing perfect calibration).

Figure 5

The mean aneurysm diameter at repair demonstrated a linear upward trend when patients were stratified by risk factor group, suggesting a higher threshold for repair in higher risk patients.

Figure 6

TAA diameter at the time of repair (as a surrogate for rupture/dissection risk) (grey bars) compared to the predicted 1-year mortality after TEVAR stratified by risk factor group (black bars). This graph demonstrates that the estimated yearly risk of aortic complications may not justify repair in higher risk patients until larger aneurysm diameters.