Computational identification of DNA damage-relevant lncRNAs for predicting therapeutic efficacy and clinical outcomes in cancer

The role of long non-coding RNAs (lncRNAs) in cancer treatment, particularly in modulating DNA repair programs, is an emerging field that warrants systematic exploration. This study aimed to systematically identify the lncRNA regulators that potentially regulate DNA damage response (DDR). Using genome-wide mRNA and lncRNA expression profiles of the same tumor patients, we proposed a novel computational framework. This framework performed Gene Set Variation Analysis to calculate DDR pathway enrichment score, which relies on weighting by tumor purity to obtain DDR activity score for each patient. Then, an in-depth differential expression profiling was conducted to identify pathway activity lncRNAs between high- and low-activity groups, utilizing a bootstrap-based randomization method. We comprehensively charted the landscape of DDR-relevant lncRNAs across 23 epithelial-based cancer types. Its effectiveness was validated by assessing the consistency of these lncRNAs within various datasets of the same cancer type (hypergeometric test P < 0.001), examining the expression perturbation of these lncRNAs in response to treatment and demonstrating its application in prioritizing clinically-related lncRNAs. Furthermore, leveraging 820 epithelial ovarian cancer patients from four public datasets, we applied these lncRNAs identified by DDRLnc to establish and validate a risk stratification model to evaluate the benefits of platinum-based adjuvant chemotherapy for the improvement of clinical treatment outcomes. Comprehensive pan-cancer analysis illustrates the utility of computational framework in identifying potentially lncRNAs involved in DDR, thereby offering novel insights into the complex realm of cancer research, and it will become a valuable tool for identifying potential therapeutic targets for cancer.