Development and validation of nomograms based on clinical characteristics and CT reports for the preoperative prediction of precise lymph node dissection in lung cancer

Abstract

ObjectiveTo develop and validate nomograms for preoperative prediction of precision lymph node (LN) dissection in lung cancer. Patients and methodsThe prediction models of each group LNs (LNx) were developed in a primary cohort that consisted of 1380 patients with clinicopathologically confirmed lung cancer. Clinical characteristics and CT reports were extracted. Patients with LNx dissection were divided into training cohort and testing cohort. Nomograms were built through univariate and multivariate regression analysis in the training cohort and internally verified in the testing cohort. The accuracy of the models was verified by constructing survival analysis in patients without LNx dissection. ResultsDue to the lack of sufficient patients for LN1, 8, 13, a total of 10 nomograms were constructed in this study, including LN-2 ∼ 7, 9 ∼ 12. According to the nomogram of each group LN, the most common independent risk factors predicting LN status were CT-reported lymphadenectasis, tumor diameter and location, and the others include age, gender, and whether there were multiple nodules, etc. All models showed good discrimination, with the average C-index of 0.738 in the training cohort and 0.707 in the testing cohort. Survival analysis in patients without LNx dissection all showed the high accuracy of each nomogram to predict LN metastasis status and TNM staging. ConclusionWe constructed nomograms to predict the metastasis status of each group of lymph nodes based on clinical characteristics and CT reports. Surgeons can accurately determine the extent of lymph node dissection in patients with lung cancer based on our nomogram models before surgery.