Clinical Features And Preoperative Risk Assessment Of Nodal Metastasis In Early Lung Cancer

Background:Mediastinal lymph node status of cTl stage lung cancer patients can largely determine the prognosis afterwards.Preoperative neoadjuvant therapies are commonly recommended for early-staged lung cancer patients with mediastinal lymph nodes metastasis.Thus,lymph nodes evaluation before surgery is capable of providing evidence for further clinical decision making.However,nodal evaluation based on CT images or biopsies from trans-bronchial needle aspiration(TBNA)still lack the sensitivity and specificity.A predictive model based on clinical features to evaluate preoperative status of mediastinal lymph nodes might subserve the interpretation and selection of examinations,assisting in lymph node staging.Purpose:This study analyzed the clinical features of early lung cancer patients with lymph node metastasis.We aimed to select independent risk factors predicting mediastinal lymph node metastasis and establish a model for preoperative assessment of mediastinal lymph node status.Methods:We retrospectively reviewed newly-diagnosed lung cancer patients from Jan.2011 to Dec.2016 in Second Affiliated Hospital of Zhejiang University School of Medicine,whose tumor sizes were within 3 cm.Among them,924 patients diagnosed from Jan.2011 to Dec.2015 were enrolled in development group and 380 patients diagnosed from Jan.2016 to Dec.2016 were enrolled in validation group.Univariate analysis was performed to compare the clinical features of patients with or without nodal metastasis,and of patients with different stages of nodal metastasis.In polytomous logistic regression,clinical variables was further analyzed to estimate different odds ratios of nodal metastasis stratified by TNM staging.In multivariate analysis,a model predicting mediastinal metastasis was established based on development data,externally validated on validation data and compared with previous models.The chnicopathological information was obtained from the hospital information system(HIS)and the lymph node staging was referred to the 8th edition of TNM staging.Results:One hundred patients(10.82%)with early lung cancer were diagnosed as nodal metastasis.Univariate analysis showed that nodal metastasis was more prone to occur in patients with male gender,smoking history,respiratory or cancer-related symptoms,chronic pulmonary diseases,central-located tumors,non-adenocarcinoma histology,larger tumor nodules,enlarged lymph nodes and elevated CEA level.Among patients with different stages of nodal metastasis,factors associated with tumor burden such as consolidation tumor size and CEA level appeared no significant difference.Patients with single station metastasis were more likely to possess normal-sized lymph nodes.Progression-free survival time was decreased in patients with higher metastatic lymph node burden.Nodal metastasis stratified by TNM staging differed in risk factors and odds ratios using polytomous logistic analysis.In multivariate analysis,we identified five independent predictors by dichotomous logistic regression:central location(OR=0.469,95%CI:0.239-0.924),consolidation tumor size(OR=6.824,95%CI:4.095-11.373),enlarged lymph nodes on CT(OR=8.546,95%CI:4.491-16.262),smoking history(OR=0.345,95%CI:0.169-0.704)and level of serum CEA(OR=1.066,95%CI:1.031-1.102).The model showed good calibration and discrimination ability,with reasonable Hosmer-Lemeshow test(p=0.839)and area under ROC in validation group being 0.906(95%CI:0.857-0.956),The AUC of our model was better than other models.The model showed great validated specificity of 90.3%.Conclusion:Our study analyzed the clinical features of early lung cancer patients with nodal metastasis.We built a model with great specificity for evaluating mediastinal lymph node metastasis of clinical T1 stage lung cancers.Our models can help selecting patients with low probability of N2 metastasis and assist in clinical decision making.