Prognostic value of surgical risk assessment scales in patients with centrally located malignant lung tumors

Objective. To evaluate predictive value of current operative risk scales in non-small cell lung cancer (NSCLC) depending on tumor localization and identify independent predictors of 30-day mortality. Material and methods. A retrospective single-center cohort study enrolled 511 NSCLC patients who underwent anatomical lung resection (central tumors n=148, peripheral tumors n=363). Eurolung I/II, Modified Eurolung II, Thoracoscore, SABCIP, and ESOS scales were assessed by ROC analysis using a single Youden-derived cut-off applied uniformly across subgroups. Complications were classified as minor (Clavien-Dindo I–IIIa) or major (IIIb–V). Risk factors were identified by logistic regression. Results. Thirty-day mortality was 2.3% (12/511). In centrally located tumors, AUC fell to 0.581–0.729 versus 0.690–0.918 in peripheral tumors. Specificity dropped to 38–62%. Central localization showed a stepwise increase in prognostic impact: non-significant for minor complications (OR=1.249; p=0.289), significant for major complications (OR=2.604; 95% CI 1.209–5.609; p=0.014), and the strongest for mortality (OR=5.129; 95% CI 1.520–17.302; p=0.008). Multivariable analysis identified central localization (OR=3.593; p=0.046) and FEV1 <65% (OR=7.286; p=0.001) as independent predictors of mortality. The model delineates four risk groups with mortality from 0.71% to 15.8%. Conclusion. Current risk scales systematically underperform in central lung tumors. A two-factor model based on localization and FEV1 provides preoperative risk stratification and requires external validation.