Development of tRNA-based biomarkers for breast cancer: Insight from machine learning

Introduction Breast cancer (BC) is one of the most frequent cancers in women globally. Research on tRNA-related biomarkers for predicting BC survival remains notably lacking. In this study, bioinformatics analysis was utilized to identify tRNA-related gene targets. Material and methods We obtained closely related mRNAs by screening BC prognosis-associated tRNAs from the OncotRF database. Next, we selected prognostically important mRNAs further using the Bruta algorithm. We developed a risk model based on these significant genes by using a variety of machine learning techniques and validated the expression with experiment. Data from the TCGA, GEO, and IMvigor210 datasets were used to validate the predictive efficacy of the t-mRNA characteristics. We also obtained the single-cell RNA sequencing (scRNA-Seq) data from the TISCH2 database and the RNA-Seq data from the UCSC Xena database for pan-cancer analysis. Results We created a prognostic model with 12 t-mRNAs associated with BC. Strong predictive performance of this model was demonstrated by nomogram, ROC and survival analyses. Functional enrichment analysis revealed differences between the low-risk and high-risk groups in immunological-related biological processes. The high-risk group showed reduced immunotherapy efficiency and greater M2 macrophage infiltration, according to the analysis of immune infiltration and immunotherapy responsiveness. Furthermore, the pan-cancer investigation revealed that high-risk tumors typically exhibit more aggressive features. We also found the express difference of model genes between normal and cancer cells. Conclusions We created an t-mRNA feature that may accurately predict the prognosis of BC patients and promote the development of precision medicine for cancer.