Establishment of a mathematical predictive model of mediastinal lymph node metastasis in early or mid-term non-small cell lung cancer with diameter ≤5 cm

Objective To establish a mathematical prediction model of mediastinal lymph node metastasis in early or mid-term non-small cell lung cancer (NSCLC) with diameter ≤5 cm and improve the accuracy of preoperative staging of lung cancer. Methods A retrospective analysis of 608 patients with NSCLC meeting the inclusion criteria in the Department of Thoracic Surgery, Clinical Medical College of Yangzhou University from January 2012 to August 2017 was randomly divided into modeling and validation groups according to a 3∶1 ratio by SPSS random number generator method.Using the data of the model group, the independent risk factors of mediastinal lymph node metastasis were screened by single factor and multivariate analysis to establish the mathematical prediction model.External validation of the model was performed using validation group case data and compared with previous models. Results Multivariate tumor size, tumor location (central or peripheral), pathological type and pleural traction were independent risk factors for mediastinal lymph node metastasis.The mathematical predictive model established was P=ex/(1+ ex), x=-2.831+ (0.825×tumor diameter)+ (1.53×central type)+ (0.779×pleural traction sign)+ (1.883×pathological type)-(0.06×age). The Hosmer-Lemeshow test showed no significant difference between the predicted and observed values.The area under the curve for the receiver operating characteristic curve was 0.763(95%CI: 0.697~0.829). External verification results show that compared with the VA model and the Fudan model, the proposed model is applicable to a wider range and higher accuracy. Conclusions The mathematical model established in this study has high sensitivity and specificity for the diagnosis of mediastinal lymph node metastasis of NSCLC ≤5 cm in diameter, and its prediction ability and accuracy are higher than other similar models.This model allows for more rational clinical decisons on whether to perform further mediastinal lymph node. Key words: Carcinoma, non-small-cell lung; Lymphatic metastasis; Logistic models; Clinical verification