Abstract:

Objective

To compare the preoperative clinical and pathological characteristics of stage I non-small cell lung cancer （NSCLC） with spread through air spaces （STAS） using dual-energy CT （DECT）imaging parameters and construct a nomogram model for predicting STAS.

Methods

A total of 269 patients with pathologically confirmed stage I NSCLC from January 2016 to September 2024 were enrolled，including 100 STAS-positive and 169 STAS-negative cases. Clinical data，preoperative DECT imaging parameters，and pathological features were analyzed. LASSO-logistic regression was used to screen predictors and develop the nomogram. Model performance was evaluated via receiver operating characteristic （ROC） curve analysis （area under the curve，AUC），calibration plots，decision curve analysis （DCA），and clinical impact curve （CIC）.Propensity score matching （PSM，1∶1） adjusted for age，sex，smoking history，diabetes，hypertension，tumor location，and TNM stage yielded 86 STAS-positive and 86 STAS-negative cases for postoperative pathological analysis.

Results

Univariate analysis revealed significant differences in smoking history，emphysema，pleural retraction，spiculation，nodule type，consolidation-to-tumor ratio （CTR），RECIST diameter，small nodule volume，iodine ratio，and mean CT value between STAS-positive and-negative groups （P＜0.05）. Multivariate analysis identified emphysema background，spiculation，CTR，iodine ratio，and CT value as independent predictors of STAS. The nomogram model achieved an AUC of 0.800 （95%CI： 0.744 ～0.854），accuracy=75.1%，sensitivity=78%，specificity=73.4%，positive predictive value=63.4%，negative predictive value=84.9%，and F1-score=0.699. Calibration curves showed good agreement with ideal predictions，while DCA and CIC demonstrated clinical utility. Post-PSM analysis indicated that STAS correlated with aggressive pathological features，including vascular invasion （7.0% vs. 0.0%，P=0.029），intravascular tumor thrombus （11.6% vs.1.2%，P=0.013），elevated Ki-67 levels，and higher IASLC grading （P＜0.05）.

Conclusion

The clinical prediction model based on preoperative DECT parameters （emphysema background，spiculation，CTR，iodine ratio，and mean CT value） provides diagnostic value for preoperative STAS assessment in stage I NSCLC when combined with three-dimensional imaging features. STAS is associated with abundant blood flow and highly invasive pathological characteristics.

Key words: Non-small cell lung cance, Spread through air spaces, Dual-energy CT, Prediction model