Evidence indicates that breast cancer patients with brain metastasis (BM) have a very poor prognosis and a short survival time following BM . However, much progress has been made regarding the diagnosis of the condition, especially with the application of PET-CT and MRI . Despite these, the progress towards effective therapies has been slow, mainly because the standard treatment options for BM, such as surgical operation, chemotherapy, and radiotherapy are ineffective . As such, there is a need to identify sensitive markers to improve the prognosis of BC with BM.
First, we screened the differential expressed genes which may be associated with BM in BC in our study. GO term analysis revealed that most of the DEGs participate in tumorigenesis through regulating cell proliferation, differentiation, apoptosis adhesion, and migration, as well as angiogenesis. Also, the KEGG pathway analysis indicated that most of the DEGs had a strong relation to the development of tumours through the key signalling pathways, including the PI3K-AKT pathway, HIF-1 signalling pathway, pathways in cancer and the metabolic pathway [12–14]. Subsequently, the PPI network was constructed to explore the interactions of the DEGs. The top ten hub genes were significantly related to the pathway of Focal adhesion, ECM-receptor interaction, Amoebiasis, PI3K-Akt signalling pathway, Proteoglycans in cancer, Protein absorption and digestion, Platelet activation, Bladder cancer and Pathways in cancer (Table 1). FN1 and VEGFA were the most connected nodes in the module, in which FN1 (fibronectin 1 protein) regulates the migratory and adhesion ability of cells. Upregulation of FN1 has been observed BC metastases . VEGFA has a vital and dual function in angiogenesis and tumour metastasis and has been implicated in the pathogenesis of breast cancer . Lastly, we performed survival analysis to determine the prognostic value of ten hub genes, we found that the three hub genes (FN1, VEGFA and DCN) were correlated with worse OS (Fig. 4A), and the seven hub genes (FN1, VEGFA, COL1A1, POSTN, DCN, BGN and LOX) were correlated with worse DMFS (Fig. 4B). Of the 10 genes, five genes (FN1, VEGFA, COL1A1, BGN and LOX) have been implicated in tumour progression, as well as in predicting disease outcomes. For example, COL1A1 expression increases the invasion and metastatic capacity of gastric cancer cells . BGN up-regulation has been demonstrated in many malignancies . Also, LOX is upregulated in many cancers, such as BC, nasopharyngeal carcinoma, hepatocellular carcinoma, head and neck tumours, colorectal cancer and is related to poor prognosis. These indicating that our big data analysis above has prognostic values in the BC cohort of GEO database. Moreover, POSTN and DCN genes were identified herein for the first time as possible markers for BC prognosis.
Herein, we identified seven possible biomarkers FN1, VEGFA, COL1A1, POSTN, DCN, BGN and LOX. However, their role in the formation and progression of cancers should be investigated further. Several studies have explored the molecular dynamics that occur during the formation and development of breast cancer[20, 21]. As a result, many BC prognostic markers have been identified. Most of these studies were conducted using animal models, in vitro cell experiments, and clinical tumour samples involving small sample size[22–24]. However, the complex interplay of BC with BM and the delicate nature of the condition requires that we conduct comprehensive studies that involve large sample sizes. Luckily, substantial progress has been made so far in bioinformatics, including the establishment of high-throughput tumour databases like GEO and TCGA. The databases collect and manage data which are made public for use by researchers and has so far assisted in big data analysis consisting of large-scale cohorts of BC with BM [26, 27].
The current study is unique because, other than focusing on the mechanisms by which the activation of tumour intrinsic gene affect BC pathogenesis, we focused on gene signatures in BC with BM and their effect on BC prognosis. The findings of the present study could offer new insights into the understanding of the complex interactions between BC and BC with BM. However, this study had a few limitations. Firstly, the clinical data in the GEO database was not complete, which will prevent us from performing a comprehensive survival analysis. Secondly, we only studied one data set, which may lead to biases in our results. Thirdly, our results still need further verification. For example, Western blot and real-time PCR can be employed to verify the selected hub genes expression in the basic experiments.