A risk score to predict postoperative complications in patients with resectable non-small cell lung cancer

Pulmonary resection for non-small cell lung cancer (NSCLC) is associated with a high risk of postoperative morbidity. We developed a system to predict the risk of major postoperative complications in patients with resectable NSCLC as there is no universally accepted system for predicting such a risk.

Data from a single-center cohort of 300 patients with resected NSCLC, stages 0 to IIIB [tumor-node-metastasis (TNM) Classification 8th edition], were retrospectively collected between 2017 and 2020. Postoperative complications classified as grade IIIa or higher according to the Clavien-Dindo classification system were considered "major complications" and the primary endpoint. The secondary endpoint was in-hospital mortality. We created a logistic regression model using selected parameters with regard to confounding, redundancy, and overfitting criteria to predict the risk of major complications.

Of 300 patients, 84 (28%) experienced major complications, and 15 (5%) experienced fatal complications. The diffusing capacity of the lung for carbon monoxide (DLCO) was significantly lower in the group with major complications than in the group with no or minor complications (63.9% vs. 73.6% of the predicted value, P<0.001). In the final logistic regression model, extended tumor resection, DLCO, and thoracotomy were identified as main predictors of major complications. The risk prediction model had a prediction accuracy of 72% (c statistic).

Major postoperative morbidity and mortality was most frequently circumvented for patients with borderline pulmonary function, who underwent minimally invasive surgery, had relatively high preoperative DLCO values, and did not need extended tumor resection.