Establishment Of Predictive Model For Malignant Probability Of Solitary Pulmonary Nodules And Micronodules

BACKGROUND AND OBJECTIVE:Primary lung cancer is one of the most common malignant tumors at present.The 5-year survival rate of early lung cancer can reach 70%-90%,while the 5-year survival rate of advanced lung cancer is only 4.2%[1-3].Solitary pulmonary nodules(SPN)are focal,spheroid-like dense shadows of diameter≤3 cm,completely surrounded by inflated lung tissue,without hilar and mediastinal lymphadenopathy,atelectasis,pleural effusion and other performance.According to the size of the nodules,the pulmonary nodules<5 mm in diameter are defined as pulmonary micro-nodules,and the pulmonary nodules with a diameter of 5-10 mm are defined as pulmonary small-nodules.With the development of modern low-dose high-resolution spiral CT and the gradual improvement of people’s health awareness,the detection rate of SPN,especially early lung cancer,is getting higher and higher,and the importance of benign and malignant identification is self-evident.The benign and malignant judgments of previous pulmonary nodules are mainly from the subjective experience of doctors,and there are few reports on the objective prediction data of malignant probabilities of pulmonary small-nodules and pulmonary micro-nodules.This study was designed to investigate the clinical risk factors of pulmonary small-nodules and pulmonary micro-nodules,and to establish a clinical prediction data model for the malignant probability of solitary pulmonary nodules and pulmonary micro-nodules.MATERIALS AND METHODS:1.We conducted a retrospective research method.We collected patients who underwent chest low-dose spiral CT examination at the First Affiliated Hospital of Soochow University from January 2018 to November 2018.And those who were initially screened for SPN and had a diameter of ≤10 mm and finally obtained a pathological report by surgery were telephoned to improve the collection of clinical risk factors.Excluding those who lost the link and the information is not perfect,and finally obtained 301 cases.Two prospective respiratory physicians performed independent readings in random order to record CT imaging risk factors for pulmonary nodules.If there is any objection to the results,the third person should be evaluated.2.Statistical analysis was performed using spss21.0.The measurement data were expressed by(x±s).The t-test was used for comparison between independent groups.The count data was expressed by n(%),and the chi-square analysis was used for comparison between groups.If the results of the univariate analysis are meaningful,a two-class logistic regression analysis is used for multivariate analysis to establish a predictive model of the malignant probability of pulmonary nodules.The Hosmer-Lemeshow goodness-of-fit test is used to evaluate the calibration ability of the predictive model.The predicted model is compared to the traditional Mayo model by calculating the area under the ROC curve.The difference was statistically significant at the test level p<0.05.RESULTS:1.In this study,301 patients with pulmonary small-nodules and pulmonary micro-nodules,including 209 cases of malignant SPN and 92 cases of benign SPN.The positive rate of surgery was 69.4%,and the negative rate was 30.6%.2.Risk factors associated with higher malignant probability of pulmonary small-nodules and pulmonary micro-nodules:Univariate analysis showed gender(P=0.001),smoking(P<0.001),history of malignant tumors(P<0.05),Burr sign(P<0.05),location(P<0.05),environmental or high-risk occupational exposure history(P<0.001),family malignant tumor history(P<0.001),nodule component(P<0.001),diameter(P<0.05),volume(P<0.001)was associated with a higher probability of malignant small-pulmonary nodules and pulmonary micro-nodules.Multivariate analysis showed that smoking(OR=4.522),environmental or high-risk occupational exposure history(OR value 5.818),family malignant tumor history(OR value 51.836),nodule component was partially solid(OR value 46.694),and the nodular component is ground glass(OR value is 7.486),the larger the volume(the OR value of 100-300mm3 is 12.090,the OR value of>300mm3 is 16.025),are independent risk factors related to malignant probability of small-pulmonary nodules and pulmonary micro-nodules.(P<0.05).3.Through variable screening,the final model includes five variables,including smoking,environmental or high-risk occupational exposure history,family malignant tumor history,nodule composition,and size,to establish a logistic regression prediction model of the malignant probability of pulmonary small-nodules and pulmonarymicro-nodules.The calibration ability of the predictive model was evaluated by the Hosmer-Lemeshow goodness of fit test.The results showed that in the Hosmer-Lemeshow goodness-of-fit test,x 2=6.813,P=0.557>0.05,suggesting that this prediction model has better calibration ability.4.The information of each patient was brought into this prediction model to compare with the Mayo model,and the ROC curve was compared.The area under the AUC curve of the Mayo model is 0.581,the cutoff value is 0.054,the sensitivity is 62.2%,the specificity is 53.3%,the area under the AUC curve of our prediction model is 0.977,the cutoff value is 0.812,the sensitivity is 89.5%,and the specificity is 97.8%.The area under the AUC curve of our prediction model is larger than the area under the AUC curve of the Mayo model,indicating that our prediction model has a higher diagnostic value for the malignant probability of pulmonary small-nodules and pulmonary micro-nodules.CONCLUSIONS:1.Smoking,environmental or high-risk occupational exposure history,family history of malignant tumors,nodular components are partially solid or purely ground glass,and the larger the volume are independent risk factors for the high probability of malignant pulmonary small-nodules and pulmonary micro-nodule.2.The constructed binary logistic regression prediction model can predict the SPN malignant probability of CT primary screening results in patients with pulmonary small-nodules and pulmonary micro-nodules,and contribute to the benign and malignant identification of it to provides auxiliary diagnostic assistance.3.The constructed binary logistic regression prediction model has higher diagnostic value than the traditional Mayo model in diagnosing the malignant probability of pulmonary small-nodules and pulmonary micro-nodules in SPN.