Routine CT-derived pulmonary artery dimensions independently predict mortality after surgical aortic valve replacement
E Boxhammer, L Brandstaetter, N Clodi, C Knapitsch, N Schoerghofer, K Gansterer, J Steindl, M Lichtenauer, M Hammerer, U Hoppe, K Hergan, C DingesAbstract
Background
Risk stratification after surgical aortic valve replacement (SAVR) relies mainly on clinical scores, while imaging-based markers are underutilized. Pulmonary artery dimensions can be easily obtained from routine preoperative computed tomography (CT) and may reflect right heart burden. We investigated their prognostic value for long-term mortality after SAVR.
Methods
In this single-center cohort, 476 patients undergoing isolated SAVR with available preoperative CT were analyzed. Truncus pulmonalis diameter (TP), TP indexed to body surface area (TP/BSA), and the TP-to-ascending aorta ratio (TP/AA) were assessed. The primary endpoint was all-cause mortality during follow-up of up to 7 years. Cut-offs were derived using receiver operating characteristic (ROC) analysis. Survival was analyzed by Kaplan–Meier estimates. Associations were evaluated using Cox regression and restricted cubic spline analyses. Multivariable models were adjusted for clinical variables (age, sex, eGFR, COPD, NYHA) and alternatively for EuroSCORE II. Incremental prognostic value was assessed by changes in C-index and likelihood ratio testing (LRT).
Results
Using ROC-derived cut-offs (TP ≥ 29 mm, TP/BSA ≥ 15.5 mm/m², TP/AA > 0.75), higher values were associated with significantly reduced survival (all log-rank p < 0.001). ROC analysis demonstrated good discrimination for 1-year mortality (AUC: TP 0.782, TP/BSA 0.808, TP/AA 0.746). In clinical adjusted Cox regression models, all parameters were independently associated with mortality (per 1 SD: TP HR 1.26 [1.08–1.47], p = 0.003; TP/BSA HR 1.28 [1.08–1.52], p = 0.005; TP/AA HR 1.24 [1.05–1.45], p = 0.009). Restricted cubic splines demonstrated a continuous increase in risk with increasing values, with the strongest association for TP/BSA (overall p < 0.01). Addition of pulmonary artery parameters to the clinical model improved discrimination (C-index 0.647 vs. 0.674 for TP) and significantly improved model fit (all LRT p ≤ 0.011). Similarly, pulmonary artery parameters provided incremental prognostic value beyond EuroSCORE II (all LRT p < 0.001), with only minor changes in discrimination.
Conclusion
CT-derived pulmonary artery dimensions are simple, reproducible markers that independently predict short- and long-term mortality after SAVR and provide incremental prognostic information beyond established clinical risk models. These findings support their integration into preoperative risk stratification.Kaplan-Meier CurvesFor image description, please refer to the figure legend and surrounding text.Restricted Cubic SplinesFor image description, please refer to the figure legend and surrounding text.