Over the past 20 years, we observed that > 700 genes had changed and were enriched in “PI3K-Akt signaling”, “ECM-receptor interaction” and “Toll-like receptor signaling”, etc. In different molecular disease groups, enriched pathways containing up- and down-regulated genes were different. For example, “PI3K-Akt signaling” in Luminal A, “phagosome” in Luminal B, “Ras signaling” in basal-like, and “AMPK signaling” in the HER2-enriched group. In addition, 11 genes were > 2-fold altered, were associated with a degree of survival prognosis (p<0.05), and potentially functioned as therapeutic targets.
Precision medicine has become an essential part of cancer treatment. Targeted molecular therapies and immunotherapies are rapidly moving toward an era of bespoke, precision medicine. Endocrine therapy for ER-positive patients in the 1980s32,33 and trastuzumab treatment for HER2-positive patients at the start of this century34,35 inaugurated targeted therapies for solid tumors. Surgery, radiotherapy, endocrine therapy, chemotherapy, and/or targeted therapies based on molecular subtyping have also paved the way for "precision medicine" for breast cancer. Additionally, risk prediction models, including the 21-gene assay (Oncotype DX Recurrence Score) and 70-gene assay (commercially known as Mammaprint) have become familiar in clinical settings to provide guidelines for systemic chemotherapy efficacy, and also endocrine therapy which may de-escalate chemotherapy 36–38. Similarly, the inception of gene profiling and next generation sequencing has meant precision medicine is now closer to clinical practice. Specifically, for ER-positive patients with endocrine therapy resistance, omics-data studies have uncovered mechanisms underpinning “CDK4/6 signaling” and “PI3K-Akt signaling” implicated in tumorigenesis and drug resistance. Similarly, prospective clinical trials also confirmed that the CDK4/6 inhibitor, palbociclib, the mTOR inhibitor, everolimus, and the PI3K inhibitor, buparlisib may improve progression free survival (PFS) in patients with advanced breast cancer39–42. Also, “MAPK signaling” and “PI3K-Akt signaling” activation are closely associated with tumor cell proliferation in HER2-positive patients. Several clinical trials have explored the efficacy of PI3K inhibitors and tyrosine kinase inhibitors in overcoming resistance to anti-HER2 therapy43,44. In triple negative breast cancer, immune checkpoint inhibitors may have clinical applications due to “Ras signaling” activation and the elevated expression of immune-related genes such as PD-1, PD-L1, and CTLA-445,46. Also, the Poly ADP-ribose polymerase (PARP) inhibitors, olaparib and talazaparib can prolong PFS and improve patient quality of life in metastatic breast cancer caused by germline BRCA mutations47–49.
We observed that USP50, GPR174, HADHAP2, NACA2, and IGFBPL1 showed large expression changes, significant Kaplan-Meier P values, and increasing proportions in the population. We propose these molecules may serve as potential breast cancer therapeutic targets in the future. Aressy et al. proposed that USP50 repressed activation of DNA damage checkpoints via a HSP90-dependent mechanism, leading to tumors50. Smith et al. reported that IGFBP-rP1 and IGFBPL1 expression was regulated by aberrant hypermethylation in breast cancer pathogenesis, and that these genes may be beneficial in clinical practice51. In the Luminal A group, CST1 exhibited the largest log2 fold expression increase (3.1); a previous study suggested CST1 may function as a significant prognostic indicator and breast cancer therapeutic target52. Also, ERBB4 expression exhibited a log2 fold-decrease of 3.7 in the HER2-enriched group, therefore ERBB4 overexpression could have biological and prognostic significance for breast cancer53.
Interestingly, 113 lincRNAs (21.6%) were up-regulated and 30 (14.0%) down-regulated in our study. Previous research indicated that lincRNAs regulate gene expression at epigenetic and transcription levels, and when expression is altered, they promote cancer initiation and metastasis. Currently, several lincRNAs are significantly correlated with cancer diagnosis, prognosis, and the therapeutic development of multitype cancers54–56. Our data indicated that several lincRNAs could function as potential prognostic biomarkers and have important clinical value, e.g., RFPL1S, ADAMTS9-AS2, IBA57-AS1, and MYOSLID are up-regulated lincRNAs57–62 and MORF4L2-AS1, LINC01278, and LINC0056263–67 are down-regulated. Importantly, all are related to the occurrence and development of several tumors by modulating “PI3K-Akt signaling”, “interferon type II signaling” and the expression of particular genes.
We also identified considerable changes in the “Staphylococcus aureus”, “Salmonella” and, “pathogenic Escherichia coli ” infection pathways, which we suspect may be related to antibiotics overuse. Recent studies reported associations between antibiotic use and breast cancer risk via effects on inflammation, immune function, and estrogen and phytochemical metabolism68. Friedman et al. reported that in 2.1 million women followed-up for 9 years, the use of any antibiotic was related to a slightly increased risk of developing breast cancer [Hazard ratio = 1.14; 95% confidence interval: 1.10–1.18] 69. However, Basso et al. reported that ansamycin may be a beneficial HER2-positive breast cancer treatment by inhibiting the “Akt dependent pathway” and cyclin D expression70. We observed that the “focal adhesion pathway” changed considerably and was enriched by ITGB1, ITGA4, and nine other up-regulated genes. Strelnikov et al. claimed a strong association between abnormal ITGA4 and ITGB1 hypermethylation and HER2-positive tumors71. Previous studies indicated that microenvironment-related pathways, such as “focal adhesion”, “ECM-receptor interaction”, and “complement and coagulation cascades” identified in this study are closely related to tumor initiation, disease progression, and metastasis, which are important future research directions72,73. In addition, we identified significant changes in metabolism-related pathways, especially in the Luminal B group, such as “glycosaminoglycan biosynthesis”, “proteoglycans in cancer”, “tryptophan metabolism”, and “β-alanine metabolism”. We hypothesize these pathways are associated with dietary intake and improvements in living standards74; encouraging results from animal studies and clinical trials revealed the clinical relevance of these pathways and the benefit of targeted drugs for cancer75–78. Interestingly, Budczies et al. reported that β-alanine accumulated in breast cancer tissues, especially in the ER-negative subtype, in agreement with our results79.
Notably, we observed eight up-regulated genes in “AMPK signaling” in HER2-enriched patients, whose activity may retard the growth of several cancers. Jhaveri et al. showed that AMPK regulated HER2 activity in HER2-enriched breast cancer cells, therefore AMPK activation may elicit a therapeutic benefit for such cancers80,81. The “alcoholism pathway”, enriched in Luminal A, suggested an elevated risk for breast cancer. Recent evidence suggested that every alcohol unit/day enhanced the possibility of breast cancer by 7–11%, and this process was mechanistically underpinned by increased estrogen levels, acetaldehyde, and oxidative stress82,83. Research also showed that disulfiram, an anti-alcoholism drug used in the clinic, induced apoptosis in in vitro breast cancer cells and showed potential therapeutic candidacy84. Beyond that, hormone dependence is a concerning issue; menopause hormone therapy and plasticizers used in daily life are closely associated with ER-pathway activation, potentially contributing to breast cancer85–87. The HABITS trial reported that estrogen and progestogen doses may be associated with breast cancer recurrence88.
In this study, we determined that genes and molecular pathways are constantly changing, suggesting molecular typing technologies must keep pace with this dynamic situation. Therefore, new biomarkers or pathways must be explored based on traditional molecular types. Our study had many limitations; a small sample size and short time span. Also, our analyses may not have fully reflected influences from the environment, time, habits, and other factors. Similarly, our study was an exploratory, retrospective analysis and lacked external validation using other methods. Thus, to some extent, the effectiveness and representation of TCGA database is limited. Nonetheless, ours is the first study to investigate tumor genomic changes from a historical perspective. Although limited, our work provides new research directions and instils debate on this key issue. The observation of dynamic tumor genomic changes has the potential to support and reinforce existing cancer prevention strategies, drug development research programs, and prognostic predictions.