Identification Of MiRNAs Biomarker For Early Detection And Prognosis Prediction Of Lung Cancer

Lung cancer is the most commonly diagnosed malignancy and the leading cause of cancer mortality in the world. In China, the morbidity and mortality of lung cancer are still increasing. Because effective biomarkers for early detection for lung cancer patients are still lack, most patients are diagnosed at advanced stage. Despite decades-long improvements in treatment, survival of advanced stage patients remains poor.Thus, there is an urgent need to develop novel effective biomarkers for early diagnosis and outcome prediction for lung cancer patients.MicroRNAs (miRNAs) are small, highly conserved non-coding RNAs. By binding to the 3' or 5' untranslated region (UTR) of target mRNA, miRNAs can modulate genes expression through translational repression or cleavage of mRNA. MiRNAs can function as either tumor suppressors or oncogenes by regulating genes involved in tumorigenesis. Moreover, miRNAs have been identified in the circulating samples of cancer patients and have been demonstrated to be remarkable stable. This suggests that miRNAs can be potential effective, noninvasive biomarkers for cancer.In the first two part of this study, we aim to identify plasma miRNAs with diagnostic value for early stage lung adenocarcinoma (LAD) and lung squamous cell carcinoma (LSCC) patients, which are the most common types of lung cancer. In the first part, we profiled the expression of plasma miRNAs of early stage LAD patients and cancer-free controls using TaqMan low density array(TLDA). 20 most significantly altered miRNAs identified with TLDA were selected for further analysis.After validation in three independent cohorts, miR-425-3p, miR-628-3p and miR-532 were selected. Then, 3-miRNA risk score model was developed, and ROC analysis showed that the 3-miRNA biomarker could distinguish LAD from healthy controls with high sensitivity (91.5%) and specificity (97.8%). We also compared the plasma expression level of the three miRNAs in LAD with LSCC, lung large cell cancer and lung small cell cancer. These three miRNAs could also distinguish LAD from lung benign diseases and other subtypes of lung cancer. In addition, the expression pattern of the 3 mRNAs in the plasma of gastric cancer, pancreatic cancer, thyroid cancer and colorectal cancer were different from the in LAD. Compared with early stage samples,we found that the expression level of miR-628-3p and miR-425- 3p in advanced stage were obviously higher, but miR-532 were lower in patients with LAD in ? or ? stage.In the second part, we profiled the expression of plasma miRNAs of stage I LSCC patients and cancer-free controls using TLDA and 20 most significantly altered miRNAs were selected for further analysis. With validation in three independent cohorts, we found miR-324-3p was significantly up-regulated and mir-1285 was significantly down-regulated in plasma of stage I LSCC patients compared with healthy controls. Then, by combining the 2 miRNAs, risk score model for LSCC patients was developed. ROC curve analysis showed that the sensitivity and specificity of this model for early stage LSCC were 85.4% and 81.8%, respectively. By comparing the expression level of miR-324-3p and miR-1285 in LSCC with that in LAD, lung large cell cancer, lung small cell cancer and lung benign diseases, we found that the aberrant expression of the 2 miRNAs were only occurred in plasma of LSCC patients. In addition, we assessed expression level of miR-324-3p and miR-1285 in plasma of pancreatic cancer, esophageal squamous cell cancionma, gastric cancer, thyroid cancer and breast cancer. The results showed that the expression pattern of both miRNAs in LSCC were different from other cancer types. Moreover, expression level of plasma miR-324-3p was significantly upregulated in advance stage of LSCC, but the level of plasma miR-1285 in early stage patients was not significantly altered in advance stage patients.In the third part, a double blind prospective lung cancer high-risk population screening study was carried out for evaluating the combination performance of low dose computered tomography (LDCT) and plasma miRNA risk model which developed in the first two parts. A total of 364 high-risk individuals were enrolled.LDCT alone identified 31 patients with suspicious nodules. 6 patients were I stage LAD after surgery validation, the other 25 were benign diseases. Although the sensitivity of LDCT alone was 100%, but the false positive rate was 80.6%. When double-positive participants (miRNA risk model and LDCT positive) were considered as positive patient, the false positive rate was significantly reduced to 42.8%. Therefore, the plasma miRNAs risk model was an effective complementary tool for LDCT in lung cancer screening.In the fourth part, TCGA LSCC cohort were used to identify miRNA with prognosis valve for LSCC patients. Initially, the miRNA expression profiles of 45 pairs of LSCC tumor tissues with normal lung tissues were analyzed and 133 significant differentially expressed miRNAs were identified. Then, we conducted univariate Cox regression assays to identify common miRNAs correlated with overall survival (OS)within each subclass of the following clinical parameters: T, N, M stage. MiRNAs were selected if they were significantly correlated with OS in at least two subclasses. Twelve miRNAs were identified in this analysis. The patients were separated into the training set and testing set randomly. By using the supervised principal component (SPC)method, seven (hsa-mir-139, hsa-mir-326, miR-101-2, miR-182, miR-183, miR-190,hsa-miR-944) of the 12 miRNAs identified were selected. Next, we developed a miRNA prognostic model using the 7 miRNAs expression levels. The 447 patients were separated into high-risk group or low-risk group using the optimum cutoff point of miRNA scores, Kaplan-Meier analysis revealed that patients in low-risk group are associated with better OS. Moreover, the 7-miRNA signature could effectively predicted outcome of early stage patients. Multivariate Cox regression analysis showed that the 7- miRNA signature was an independent prognostic factors related with OS.In summary, the current study identified 2 plasma miRNA signatures for early detection for LAD and LSCC, respectively. A new model for high-risk lung cancer population screening was established by using LDCT and miRNA signatures. And the false positive rate of LDCT could be obviously reduced. A 7-miRNA signature with prognosis value for LSCC was also identified. After validation of future studies using independent cohorts of large sample size from multiple institutions, it may be applied into clinical practice as a novel biomarker for outcome prediction for LSCC patients.