AUTHOR=Liu Jie , Zhu Jing , Wang Xin , Zhou Zhisheng , Liu Haiyan , Zhu Dajiang TITLE=A Novel YTHDF3-Based Model to Predict Prognosis and Therapeutic Response in Breast Cancer JOURNAL=Frontiers in Molecular Biosciences VOLUME=9 YEAR=2022 URL=https://www.frontiersin.org/journals/molecular-biosciences/articles/10.3389/fmolb.2022.874532 DOI=10.3389/fmolb.2022.874532 ISSN=2296-889X ABSTRACT=

Background: Due to high tumor heterogeneity, breast cancer (BC) patients still suffer poor survival outcomes. YTHDF3 plays a critical role in the prognosis of BC patients. Hence, we aimed to construct a YTHDF3-based model for the prediction of the overall survival (OS) and the sensitivity of therapeutic agents in BC patients.

Methods: Based on The Cancer Genome Atlas (TCGA, https://portal.gdc.cancer.gov/) database, we obtained BC patients’ data (n = 999) with YTHDF3 expression profiles. The association between YTHDF3 expression and 5-year OS was determined via Cox proportional hazards regression (CPHR) analysis. By integrating the variables, we established a prognostic nomogram. The model was estimated via discrimination, calibration ability, and decision curve analysis (DCA). The performance of the model was compared with the TNM stage system through receiver operating characteristic (ROC) curves and DCA. By means of the Genomics of Drug Sensitivity in Cancer (GDSC) database (https://www.cancerrxgene.org/), the therapeutic agents’ response was estimated. Gene set enrichment analysis (GSEA) demonstrated possible biological mechanisms related to YTHDF3. TIMER and CIBERSORTx were employed to analyze the association between YTHDF3 and tumor-infiltrating immune cells.

Results: The high YTHDF3 expression was significantly correlated with poor 5-year OS in BC patients. Through multivariate CPHR, four independent prognostic variables (age, TNM stage, YTHDF3 expression, and molecular subtype) were determined. On the basis of the four factors, a YTHDF3-based nomogram was built. The area under the curve (AUC) of the ROC curve for the model surpassed that of the TNM stage system (0.72 vs. 0.63, p = 0.00028). The model predictions showed close consistency with the actual observations via the calibration plot. Therapeutic response prediction was conducted in high- and low-risk groups and compared with each other. The BC patients with higher risk scores showed more therapeutic resistance than those with a lower risk score.

Conclusion: YTHDF3 was verified as a prognostic biomarker of BC, and a novel YTHDF3-based model was constructed to predict the 5-year OS of BC patients. Our model could be applied to effectively predict the therapeutic response of commonly used agents for BC patients.