Development And Application Of Nomogram Model For Multiple Pulmonary Nodules

Objective:The diagnosis of benign and malignant multiple lung nodules and choice the treatment strategies are clinically recognized problems in thoracic surgery.In this paper,univariate and multivariate analysis were used to screen the independent risk factors of benign and malignant multiple nodules in the lungs.Nomogram was used to construct a mathematical model for predicting the benign and malignant of multiple pulmonary nodules,and verify its accuracy.It provides a valuable reference for the diagnosis and treatment strategy selection of multiple pulmonary nodules.Methods:A total of 75 patients with multiple Pulmonary nodules of the Northern Theater General Hospital from Januay 2015 to December 2018 were enrolled.Underwent minimally invasive surgery of Robot-assisted Thoracic Surgery(RATS)or Video-assisted Thoracic Surgery(VATS)at same time and postoperative obtained definite pathological diagnosis,33 cases of men,42 cases of women and a total of 155 nodes.Collection of patient data:clinical data including patient age,gender,symptoms,duration of disease,smoking history,smoking index(daily smoking count × years of smoking),history of previous tuberculosis,history of previous lung disease(asthma.emphysema).occupation contact history(asbestos,dust),history of cancer,family history of cancer,a total of 11 indicators:Serological tumor markers include 4 indicators:Carcinoembryonic Antigen(CEA).Neuron Specific Enolase(NSE),Squamous Cell Carcinoma Antigen(SCCA),and Cytokeratin 19 Fragments.(CYFRA21-1);Collect imaging data for each nodule including nodule location,nodular type(ground-glass nodules,part-solid nodules,solid nodules),maximum diameter(lung window),calcification,whether the shape is regular and the boundary is clear,spiculation.lobulation,bronchial aeration,vascular bundle,pleural traction and cavitation,a total of 12 indicators.Meanwhile,the presence or absence of PET/CT and postoperative pathological diagnosis of each nodule was collected,Through univariate and multivariate analysis,screening the independent influencing factors related to the benign and malignant of multiple pulmonary nodules and using the Nomogram build mathematical model for predicting the probability of malignant tumors of each nodule.Finally,the model predictive performance was evaluated using the calibration curve and the Area Under the Curve(AUC).Results:A total of 75 patients with multiple pulmonary nodules were retrospectively analyzed.Of these.36 cases underwent surgical resection by Robot-assisted Thoracic Surgery(RATS)and 39 cases by Video-assisted Thoracic Surgery(VATS).A total of 155 nodules were removed,including 85(62.6%)malignant nodules and 70(37.4%)benign nodules.In the univariate analysis,the patient's age(P=0.004).serum CEA(P=0.019),nodule type(P<0.001),tumor diameter(P=0.005),calcification(P<0.001).shape(P<0.001),border(P<0.001).lobulation(P<0.001).spiculation(P<0.001),vascular bundle(P<0.001),bronchial aeration(P<0.001)and pleural traction(P<0.001).a total of 12 indicators were significantly associated with the degree of malignancy in the multiple lung nodules.Multivariate analysis showed that patients' age(P=0.011),serum CEA(P=0.006),nodule type(P=0.001).calcification(P=0.03),lobulation(P=0.022),spiculation(P=0.01),vascular bundle(P=0.038),bronchial aeration(P=0.019).and pleural traction(P=0.033),a total of 9 indicators were the independent risk factors of multiple pulmonary nodules malignancy.The above independent risk factors were included in the Nomogram to construct a mathematical model for predicting the probability of multiple nodular malignancies.The Nomogram model shows better discrimination and consistency.The area under the curve(AUC)of the receiver operating characteristic curve(ROC)is 0.945(95%CI 0.913-0.977).When T=0.758 is intercepted,the Youden index is the largest,with a sensitivity of 84.7%and a specificity of 90.00.The calibration curve consistency index(C-index)=0.824.and the probability of malignant tumors predicting multiple nodules in the lungs is substantially parallel to the probability of actual malignant tumors,with a slope of approximately 45°.Conclusion:We developed a new.easy-to-use.objective,and accurate Nomogram model to predict the probability of multiple nodular malignancies in the lungs,and verified the model with good accuracy through internal validation.Bv having a highly accurate,well-calibrated mathematically predictive model,clinicians and surgeons can predict the malignant risk of each nodule in multiple lung nodules based on the patient's preoperative data,then effectively making a personalized and reasonable treatment plan,which is targeted to treat multiple pulmonary nodules.