Abstract
Long non-coding RNAs (lncRNAs) could alter the tumor immune microenvironment and regulate the expression of immune checkpoints (ICPs) by regulating target genes in tumors. However, only a few lncRNAs have precise functions in immunity and potential for predicting ICP inhibitors (ICI) response. Here, we developed a computational multi-step framework that leverages interaction network-based analysis to identify cancer- and immune-context ICP-related lncRNAs (NetLnc). Based on bulk and single-cell RNA sequencing data, these lncRNAs were significantly correlated with immune cell infiltration and immune expression signature. Specific hub ICP-related lncRNAs such as BANCR, MIAT, and SNHG15 could predict three- and five-year prognosis of melanoma in independent datasets. We also validated that some NetLnc-based predictions could better effectively predict ICI response compared to single molecules using three kinds of machine learning algorithms following independent datasets. Taken together, this study presents the use of a network-based framework to efficiently select ICP-related lncRNAs, which contributes to a comprehensive understanding of lncRNA functions and accelerates the discovery of lncRNA-based biomarkers in ICI treatment.