Background: Overcoming anoikis is a necessity during the metastasis and invasion of tumors. Recently, anoikis has been reported to be involved in tumor immunity and has been used to construct prognosis prediction models. However, the roles of anoikis in regulating tumor immunity and drug sensitivity in breast cancer are still not clear and therefore worth uncovering.
Methods: TCGA and GEO data are the source of gene expression profiles, which are used to identify anoikis-related-gene (ARG)-based subtypes. R4.2 is used for data analysis.
Results: Breast cancer is divided into three subgroups, amongst which shows prognosis differences in pan-cancer cohort, ACC, BLCA, BRCA, LUAD, MESO, PAAD, and SKCM. In breast cancer, it shows significant differences in clinical features, immune cell infiltration and drug sensitivity. Machine learning constructs prognosis prediction model, which is useful to perform chemotherapy sensitivity stratification. Following, TJP3 is identified and verified as the key ARG, up-regulation of which increases tolerance of paclitaxel-induced cell toxicity, accompanied with increased expression of caspas3 and cleaved-caspase3. In addition, Down-regulation of TJP3 weakens the cell migration, which accompanied with increased expression of E-cad and decreased expression of vimentin, twist1, zeb1, and MMP7. Furthermore, the expression level of PD-L1 is negative correlated with TJP3.
Conclusion: ARGs-based subgroup stratification is useful to recognize chemotherapy sensitive cohort, and also is useful to predict clinical outcome. TJP3 promotes chemoresistance, tumor metastasis and potential immunotherapy escape in breast cancer.
Keywords: anoikis-related genes (ARGs); breast cancer; drug resistance; immunity; machine learning.