Lei Cheng, Chao Zhao, Haowei Wang, Caicun Zhou, Aiwu Li, Xuefei Li
Pulmonary nodules are common. It is important to identify novel noninvasive biomarkers for the diagnosis and prognosis of malignant lung nodules. We retrospectively collected preoperative plasma samples at Shanghai Pulmonary Hospital and measured the 24 cytokines involved in the development of lung cancer using electrochemiluminescence multiplex assays. Univariable and multivariable analysis was utilized to estimate relationships between the cytokines with malignant nodules. The model was constructed by logistic regression analysis. The cytokines associated with postoperative recurrence were analyzed and validated in an independent cohort. Receiver operating characteristic (ROC) curve analysis was applied to evaluate the performance. The counts of circulating tumor cells identified by ligand-targeted PCR in the preoperative period failed to distinguish benign from malignant nodules. After adjustment for age, gender and smoking status, IL-4 and GM-CSF were associated with decreased odds of malignant nodules (OR 0.27, 95% CI 0.07-1.02; OR 0.20, 95% CI 0.05-0.75, respectively). The plasma cytokine panel (IL-17A, IL-4 and GM-CSF) was constructed for the preoperative diagnosis of malignant nodules with an AUC of 0.73. TSLP showed significantly differential expression between recurrence patients and recurrence-free patients (p < 0.05). The corresponding ROC curve yielded an AUC of 0.844 in the screening set and 0.847 in the validation set. Taken together, we constructed a plasma cytokine panel (IL-17A, IL-4 and GM-CSF) for the preoperative diagnosis of malignant nodules, and found that TSLP could be as a recurrence predictor in resected NSCLC patients.