Primary Investigation On Cinical Value Of Detecting VOCs Of Exhaled Breath In Patients With Lung Cancer

Background: Lung cancer (LC)is the leading cause of cancer death.According to data from WHO, there are more than 1.69 million people worldwide who died from LC in 2015, accouting for 19.2% of total cancer death. Late diagnosis contributes to high mortality of lung cancer, and the five-year survival rate of LC in China is only 15.6%.Therefore, it is crucialto diagnose LC early in order to prolong survival time. So far,many efforts have been made to develop proper screening methods for an early diagnosis of LC, including radiation imaging, bronchialscopy, biopsy and so on.However, regardless of all these diagnose methods and evolving therapy strategies,early diagnosis andprognosis of LC still remain poor.Recently, an emerging approach, briefly called breath test, might become a new way for assessment of pulmonary dieases.It is totally non-invasive, convenient and low-cost, which relies on volatile organic compounds (VOCs) of human breath. Among all the different detection methods and instrument, PTR-MS is anextremely convienient and fast method to detect breath gas, which needs no pretreatment of sample gas.Objectives: We use PTR-MS to analyse the exhaled breath of lung cancer patients (LC),benign pulmonary diseases patients (BPD) and health controls (HC). By comparing the different VOCsand clinical data among the three groups, we attempt to set up a diagnose model for lung cancer.Methods: We used Tedlar bags to collect breath gas of 78 LC patients, 70 BPD patients,108 HC and 50 air samples,then detect them by PTR-QMS 500 highsensitive instrument,in order to find different VOCs among different groups. We assemble LC and BPD into a illness group(IG), BPD group and HC into non-cancerous group(NG).By combining with the clinical data, we set up a model to diagnose LC. Then we used ROC and discriminant analysis to evaluate the predictive ability of the model. We analyzed all the data using IBM SPSS 22.0 and Medcal 15.8.Results: 1. There was differentiation in VOCs concerntration among different groups .2. Patients of different LC stages had significantly variation in VOCs concentration,while patients of different pathological type, the site of tumor and smoking status did not show any significant difference in VOCs converntration3. The model of distinguishing IG and HC exhibited a sensitivity of 80.41%, specificity of 94.44%,cut-off value of prediction probablity is 0.71; the model distinguishing LC and NG showed a sensitivity 66.10%, specificity 91.14%, cut-off value of prediction probablity is 0.19; the model distinguishing LC and BPD showed a sensitivity of 62.03%, specificity 82.61%, cut-off value of prediction probablity is 0.54.4. Discriminant analyisis showed an overall accuracy of 68.0%.Conclusion: PTR-MS detection of VOCs is effective to predict lung cancer probability.But it is less effective in distinguishing between lung cancer and benign pulmonary disease patients. Different stage of lung cancer is an impotant factor affecting VOCs concerntrations.