Establishment Of An Independent Predictor And Predictive Model For Malignant Subsolid Pulmonary Nodules

Background: According to a large randomized controlled lung cancer screening study dominated by the National Lung Cancer screening trial in 2011,compared with X-ray chest radiographs,the utilization of low-dose computed tomography(LDCT)for the screening of lung cancer among the risky population can reduce the mortality by 20%.Based on the beneficial results from NLST,numerous authoritative medical organizations in United States have initiated and updated the new guidelines since 2012,recommending the usage for lung cancer screening among high-risky groups.With the extended application of thoracic LDCT,the number of pulmonary nodules found in clinic shows an increasing trend.While the diagnosis and treatment for pulmonary nodules have always been a difficult part and hot spot,owing to the complexity of the etiology,the small size,the lack of specificity,and a higher possibility of false negative or positive diagnosis.Subsolid nodules(SSN)can be defined as the nodules containing the part of ground-glass density.The subsolid nodules,whose detection rate has increased significantly due to the utilization of LDCT screening for lung cancer,are closely related to pulmonary adenocarcinoma.This article makes a retrospective analysis on the characteristics of diseases related to patients with subacute pulmonary nodules in our teaching hospital and its group hospitals.Objective: The purpose of this study is to analyze the high-risk factors and imaging characteristics of pulmonary nodules found in the patients by using the LDCT for lung cancer screen.Seeking to establish an ideal,stable and accurate diagnostic prediction model to assist in the evaluation of benign and malignant subsolid pulmonary nodules.At the same time,it is expected to establish a stable and accurate prediction model for diagnosis in order to assist the identification of benign and malignant subsolid pulmonary nodules in the clinical practice.Methods: From January 2017 to February 2019,345 patients with subsolid pulmonary nodules underwent thoracic surgery in Zhongda Hospital affiliated to Southeast University,Jiangbei Hospital affiliated to Zhongda Hospital,and Wuxi Xishan Hospital(Group Hospital affiliated to Zhongda Hospital)and obtained definite pathological results.All the cased were selected in strict accordance with the following criteria.Inclusion criteria: 1.Chest CT showed a single round or oval subsolid pulmonary nodule with a diameter of 5-30 mm.2.Nodules with the thickness less than 1.5mm on continuous thin layered CT scan was reserved.3.The clinical data and surgical and pathological results must be definite and accurate.Exclusion criteria: 1.The nodules are accompanied by hilar and mediastinal lymphadenopathy.2.Imaging examination revealed atelectasis,pleural effusion and obstructive pneumonia.3.Auxiliary examination indicates remote metastasis.Clinical and imaging data of the subjects were collected.According to the surgical and pathological results,the patients were divided into benign group and malignant group for single-factor and multi-factor analysis.Results: Multivariate Logistic regression analysis showed that smoking history,nodular diameter,vascular bundle sign,calcification,vacuoles and the character of the margin were independent factors for identification of benign and malignant diseases.The established mathematical equations for predicting benign and malignant diseases are: P= ex/(1+ ex),x=-1.879+(0.980× smoking history)+(0.078× nodule diameter)+(1.589× vascular bundle sign)+(1.389× vacuole)+(1.521× rule)+(-3.229× calcification),e is natural logarithm.The area under the ROC curve(AUC)of this model was calculated to be 0.72(0.596-0.786).The sensitivity is 0.887 and the specificity is 0.585 when the appropriate critical point is 0.472.Conclusions: 1.This risk prediction model of pulmonary nodules can make a preliminary assessment of uncertain pulmonary nodules and assist clinical decision-making.2.In this study,clinical and imaging data related to the malignant probability of subsolid pulmonary nodules were screened out by binary Logistic regression analysis,and a mathematical prediction model for benign and malignant subsolid pulmonary nodules was established.Through data verification,we can conclude that the model proposed in this study has a high accuracy and sensitivity.