High-dimensional Analyses of MicroRNAs in Lung Neuroendocrine Neoplasms

Abstract

Lung neuroendocrine neoplasms (NENs) are a diverse group of enigmatic cancers. Lung NENs consist of two tumor families: carcinoids, consisting of tumor types typical carcinoids (TC) and atypical carcinoids (AC); and neuroendocrine carcinomas (NECs), consisting of small cell lung carcinomas (SCLC) and large cell neuroendocrine carcinomas (LCNEC). Due to their diverse biology and unspecific symptomatology, they are difficult to diagnose and whose molecular mechanisms are incompletely understood. MicroRNAs (miRNAs) are small, RNA molecules that post-transcriptionally regulate gene expression. They regulate many important pathways in cancers and are valuable biomarkers in many neoplastic diseases. To assess the diagnostic utility of miRNAs in lung NENs and evaluate their roles in regulation of lung NEN genetic pathways, we generated miRNA expression profiles from archived lung NEN tumor samples and representative lung NEN cell lines and investigated them using machine learning and data mining techniques. We first developed a novel feature selection algorithm that uses an ensemble method to identify discriminatory -omics biomarkers and used it to discover miRNA biomarkers of lung NENs. Finally, we compared lung NEN cell lines to non-NEN control lines to identify important miRNAs in lung neuroendocrine biology, and predicted their regulatory targets and their related pathways. Through these analyses we discovered two miRNAs which were able to discriminate carcinoids from NECs with high accuracy. We also identified two sets of three miRNAs as candidate markers for discriminating TC from AC, and SCLC from LCNEC, although these findings require further validation due to limited samples size. Finally, we found five miRNAs that regulate cell development pathways in lung NENs.