A novel nomogram incorporating LASSO and Cox regression analyses for predicting survival in early-stage non-small cell lung cancer patients following sublobectomy.

[BACKGROUND] Sublobectomy remains a key surgical approach for stage I non-small cell lung cancer (NSCLC) patients, yet prognostic tools for postoperative outcomes remain scarce. This study developed a nomogram to predict lung-cancer-specific survival (LCSS) following sublobectomy.

[METHODS] We retrospectively analyzed 8786 early-stage NSCLC patients who underwent sublobectomy between 2005 and 2018 from the SEER database, dividing them into training and validation cohorts. Using LASSO regression to minimize overfitting and address multicollinearity, followed by Cox regression in the training set, we selected final predictors for the nomogram. Model performance was assessed via the C-index, ROC analysis, calibration curves, and DCA, with comparisons made against the 8th TNM staging system and a risk stratification framework.

[RESULTS] The training set comprised 4184 patients, while validation sets A and B included 1794 and 2808 patients, respectively. The nomogram incorporated age, sex, tumor size, grade, surgical approach, and total tumor number. C-index values were 0.70 (training), 0.69 (validation A), and 0.61 (validation B), with 8-year AUCs of 0.698, 0.732, and 0.682, respectively. Calibration curves confirmed strong predictive accuracy, and the nomogram outperformed TNM staging in time-dependent AUC and net benefit on DCA.

[CONCLUSION] A novel nomogram integrating LASSO and Cox regression reliably predicts long-term LCSS in early-stage sublobectomy patients.