Construction of a multiple-class classifier based on mRNAs and lncRNA FAM66A and PSORS1C3 for predicting distant metastasis in lung adenocarcinoma.

Background: There are several mechanisms believed to be essential for the development of distant metastasis in lung adenocarcinoma (LUAD), but the prediction of distant metastasis is still a challenge. The purpose of the present study was to examine the specific changes in RNA expression, including long non-coding RNAs (lncRNAs) in distant metastasis patients. Methods: We compared differentially expressed genes involved in distant metastasis from otherwise non-metastasis and healthy adults using a gene expression profile. We first ranked gene sets (or gene signatures) that identify each class. An advanced multiple-class classifier was built based on the gene sets. Our classifier consisted of 282 genes and could predict cancer and distant metastasis with error rates of approximately 0.01 and 0.2, respectively. Then, gene networks were built to undermine gene relations to each class. Results: Cytochrome P450 family 4 subfamily F member 12 (CYP4F12) was the first gene in the ranking of the distant metastasis case. Down syndrome cell adhesion molecule (DSCAM) was the top gene in the rank list of the non-metastasis case. Solute carrier family 6 member 4 (SLC6A4) was associated with normal tissues. LncRNA family with sequence similarity 66 member A (FAM66A) and lncRNA PSORS1C3 were found to be associated with tumor metastasis. Conclusions: Our classifier could successfully predict distant metastasis in LUAD patients. LncRNA FAM66A and lncRNA PSORS1C3 in our model could play a role in cancer development.