Improved microRNA biomarkers for pathological stages in lung adenocarcinoma via clustering of dysregulated microRNA-target associations.

Most of cancer-related deaths are due to lung cancer, and there is a need for reliable prognosis biomarkers to predict stages in lung adenocarcinoma cases. Recently, microRNAs are found to have potential as both biomarkers and therapeutic targets for lung cancer. As evidence suggests microRNA dysregulations are implicated in many cancer malignancies, it is important to consider the changes in miRNA-target associations among different lung cancer biological states. We proposed a novel clustering strategy to identify groups of miRNAs with similar dysregulated targets. Then, we incorporated the learned clusters of miRNA as prior knowledge to a Sparse Group Lasso classifier to improve classification results, thereby leading to more relevant selection of microRNA biomarkers. We apply the method to the TCGA Lung Adenocarcinoma dataset. In cross-validation tests, the AUC rate for each stages is 1.0, 0.71, 0.68, 0.64, and 0.90 for normal, Stage I, Stage II, Stage III, and Stage IV, respectively. Among the candidate miRNAs selected in the model, 87% are reported to be related to Lung Adenocarcinoma. Further result demonstrates that clustering miRNAs by considering the dysregulation between miRNAs and mRNA targets leads to biomarkers with higher precision and recall rate to known lung adenocarcinoma miRNAs.