Development and validation of a predictive model for lymph node metastases in peripheral non-small cell lung cancer with a tumor diameter ≤ 2.0 cm and a consolidation-to-tumor ratio > 0.5

Background Precisely predicting lymph node metastasis (LNM) status is critical for the treatment of early non-small5-cell lung cancer (NSCLC). In this study, we developed a LNM prediction tool for peripheral NSCLC with a tumor diameter ≤ 2.0 cm and consolidation-to-tumor ratio (CTR) > 0.5 to identify patients where segmentectomy could be applied. Methods Clinical characteristics were retrospectively collected from 435 patients with NSCLC. Logistic regression analysis of the clinical characteristics of this development cohort was used to estimate independent LNM predictors. A prediction model was then developed and externally validated using a validation cohort at another institution. Results Four independent predictors (tumor size, CTR, pleural indentation, and carcinoembryonic antigen (CEA) values) were identified and entered into the model. The model showed good calibration (Hosmer–Lemeshow (HL) P value = 0.680) with an area under the receiver operating characteristic curve (AUC) = 0.890 (95% confidence interval (CI): 0.808–0.972) in the validation cohort. Conclusions We developed and validated a novel and effective model that predicted the probability of LNM for individual patients with peripheral NSCLC who had a tumor diameter ≤ 2.0 cm and CTR > 0.5. This model could help clinicians make individualized clinical decisions.