Breast cancer is the most common malignancy in women, with the frequency rising with increasing age (Smith et al., 2003; Van et al., 2010). An increase in the incidence of breast cancer over the past several decades is thought to relate to a greater number of women who delay child-bearing (Andersson et al., 2009; Raphael, Trudeau and Chan 2015). Conventionally, pregnancy-associated breast cancer (PABC) is a rare type of breast cancer that is diagnosed during pregnancy or within the following year (Lee, Mayer and Partridge 2017). The incidence of PABC is approximately 15–35 per 100,000 deliveries, with more cases diagnosed during the first postpartum year (Smith et al., 2003; Woo, Yu and Hurd 2003). The remodeling of the mammary gland to its pre-pregnant state after pregnancy might cause the mammary microenvironment to become tumor-promoting (Pepper 2006). Furthermore, diagnosis of PABC is complicated by physiological changes that accompany pregnancy. PABC generally presents as a painless palpable mass, skin changes including thickening, or bloody nipple discharge, which are often mistaken for pregnancy symptoms. Lack of detection, hesitation to proceed with medical tests, and limitations of imaging during pregnancy often result in delays in diagnosis of PABC (O’Neill et al., 2016). Therefore, it is important to improve detection and therapy of PABC.

The immune system can prevent tumor progression through immune surveillance mechanisms; however, immunogenic phenotypes in breast cancer that promote tumor growth may arise (Dunn et al., 2002). Besides, CD88⁻CD1c+CD163+ DCs (called DC3s) infiltrated luminal breast cancer primary tumors in vivo and DC3s was regarded to have strong potential to regulate tumor immunity (Bourdely et al., 2020). Thus, infiltrating immune cells may be both prognostic and predictive of response to breast cancer therapy (Cimino-Mathews, Foote and Emens 2015). Breast cancer can be subdivided into intrinsic molecular subtypes based on differential gene expression profiles (Perou et al., 2012; Sorlie et al., 2001; Sorlie et al., 2003; Nielsen et al., 2004). Furthermore, recent studies suggest that PABC may have a unique genomic signature associated with increased hormone levels in pregnancy (Harvell et al., 2013). Therefore, understanding the regulatory pathways and pathogenesis of breast cancer may provide a blueprint for effective immunotherapy. Nevertheless, there have been few studies that elucidate the role of the immune system in PABC, making it a challenge to design effective therapeutic strategies.

In this study, we applied microarray technology and comprehensive bioinformatics analysis to characterize gene expression profiles of PABC and compared the results with non-pregnancy-associated breast cancer (NPABC). The results characterize a distinctive gene expression pattern in PABC, which might be used in clarifying the pathogenesis of PABC and exploring a possible prognostic biomarker for early diagnosis of PABC.

Numerous studies have attempted to characterize the expression profile in breast cancer to identify biomarkers for diagnosis, prognosis or personalized therapy (Liu et al., 2017; Zhang et al., 2018). Furthermore, PABC has been recognizes as a distinct form of breast cancer with unique properties (Pepper, 2006; Andersson et al., 2009; Lee et al., 2017). A 2013 study evaluated PABC transcriptomic signatures from the GSE31192 dataset (Harvell et al., 2013), and a more recent study integrated two microarray profile datasets (GSE31192 and GSE53031) to identify core genes and clinical roles in PABC through integrated bioinformatic analysis (Zhang et al., 2019). The GSE31192 and GSE53031 datasets were from different sample sources, including cancer epithelial tissues, cancer stroma tissues, normal epithelial tissues, and normal stroma tissues. Therefore, data were grouped according to the sample sources. However, in this study, we compared PABC with both NPABC samples and normal samples regardless of the tissue source, which may give us a more holistic assessment of signature features that are specific to PABC regardless of the epithelial versus stromal sample source.

We identified 871 DEGs for the PABC versus NPABC group comparison and 917 DEGs for the PABC versus normal group comparison. Among the most highly dysregulated genes, the MAGE family oncogenes, including MAGE A6/A3, A2B/A2 and A12, were found to be expressed at elevated levels in PABC. Given the association of MAGE proteins with extremely aggressive tumor types (Li et al., 2021), these findings may underscore the aggressive PABC phenotype. We also determined that CXCL2 is upregulated and CXCL10 and 11 are downregulated in PABC versus adjacent normal tissues. These proteins belong to a family of chemokines that are associated with proliferation and angiogenesis in breast cancer and have been explored as potential therapeutic targets (Palacios-Arreola et al., 2014).

To further understand the patterns of differential expression, we generated heatmaps of the most highly dysregulated genes, which demonstrated that samples of the PABC group were clearly separated from those of the NPABC and normal groups. Gene ontology and enrichment analyses further indicated that the DEGs in PABC versus NPABC were mainly enriched in pathways in cancer, the PI3K-Akt signaling pathway, focal adhesion, the Rap1 signaling pathway, proteoglycans in cancer, microRNAs in cancer, prostate cancer, the p53 signaling pathway, small cell lung cancer, and melanoma and glioma pathways. Many studies have demonstrated dysregulation in the PI3K-AKT pathway in cervical cancer, breast cancer, malignant glioma, and other cancers (Lee et al., 2015; Guerrero-Zotano, 2016; Mayer and Arteaga; Li et al., 2016), thus suggesting that the PI3K-AKT pathway may be an important contributor to the oncogenic PABC phenotype.

For more in-depth analysis of functional pathways associated with DEGs in PABC, we also performed PPI network analysis and clustering (Berkhin 2006). Interestingly, the classic breast cancer marker genes, BRCA1 (Semmler et al., 2019) and CD44 (Chen et al., 2018) were overexpressed in the PABC PPI network, suggesting that PABC may be characterized by higher activity of proteins that contribute to NPABC. Our results also emphasize the role for immune and inflammatory pathways in PABC. Notably, the well-established immune pathway activators PTPRC/CD45 and IL1B (Sigismund et al., 2018; Al Barashdi et al., 2021) were among the proteins with the highest node degree value in the corresponding PPI network. According to the clustering analysis, six clusters of proteins were dysregulated, and the largest cluster was associated with the MHC class II protein complex (36.36%), type I interferon signaling pathway (36.36%), regulation of α-β T-cell proliferation (18.18%) and T-cell apoptotic process (9.09%). MHC class II is known to be downregulated in many metastatic tumors, including human breast cancer cells (Shi et al., 2006). On the other hand, expression of MHC-II in melanoma is associated with an intensified response to PD-1-targeted immunotherapy (Johnson et al., 2016), which has been subsequently validated in classic Hodgkin lymphoma (Roemer et al., 2018) and with combination immunotherapy (Rodig et al., 2018). MHC-II expression can characterize T-cell-inflamed or immune-responsive subset of tumors (Balko et al. 2018). The association of type I interferon within the PPI may be indicative of a complex relationship between stress and immune function that has been reported in breast cancer (Mundy-Bosse et al., 2011). Type I interferon is the prototype member of a class of antiviral immunomodulatory cytokines involved in tumor initiation and progression and can act on tumor cells directly by inhibiting cell growth or indirectly by activating immune cells to mount antitumor responses (Zitvogel et al., 2015). Metastatic progression is the major cause of breast cancer-related mortality, and in multiple models of breast cancer, dysregulated immunity due to the loss of host type-I IFN signaling may drive metastasis (Rautela et al., 2015). A role for T-cells in the PABC phenotype, as indicated by our clustering analysis, is also consistent with mechanisms that are known to impact breast cancer progression. Higher levels of T cells at early stages of differentiation have been detected in breast cancer patients (p < 0.05) (Speigl et al., 2017), and CD14⁺ myeloid cells from breast cancer patients have been shown to have enhanced ability to suppress autologous T-cell proliferation (Speigl et al., 2017). Furthermore, interleukin 10 and interleukin two synergistically function to promote cytotoxicity of CD8⁺ T-cell, which is inhibited by regulatory T cells in breast cancer (Li et al., 2017). Thus, increased understanding of functional pathways and molecular mechanisms will provide a blueprint for effective immunotherapy for breast cancer, and for PABC in particular.

Our PPI network analysis of PABC versus normal breast tissue samples also identified a single cluster for which EGFR, IGF1, PTGS2, FGF1, CAV1 and PLCB1 were verified to be differentially expressed in an additional independent cohort of patients’ PABC tissue compared with normal tissue. EGFR has been demonstrated to regulate epithelial tissue development and homeostasis in a variety of types of cancer, including breast cancer (Sigismund et al., 2018), with approximately half of triple-negative breast cancer and inflammatory breast cancer cases overexpressing EGFR (Masuda et al., 2012). Furthermore, the IGF1 signaling pathway has previously been shown to be critical for normal cell division and to be modulated in PABC (Loddo et al., 2014). Targeted therapies have been developed for suppressing the post-partum pro-tumorigenic extracellular matrix via inhibition of PTGS2 (O'Brien et al., 2011), and the potential benefit of targeting the FGF1 pathway in breast cancer has been considered (Francavilla and O’Brien, 2022). On the other hand, CAV1 expression, which is down-regulated during tumorigenesis, has been shown to be suppressed during lactation (Park et al., 2001). Therefore, the identification of these proteins within a PPI cluster is consistent with known mechanisms in breast cancer that may be modulated during the pregnancy and postpartum periods. Notably, three of these genes, including IGF1, PTGS2, and FGF1, were significantly related to breast cancer patient outcomes, thus further verifying their potential roles as prognostic biomarkers.

In this study, we screened differentially expressed genes from microarray datasets of PABC by using an integrated bioinformatics analysis approach. A distinctive pattern of cancer and immune-related gene expression revealed a landscape of PABC-associated genes and an interaction network of their protein products, suggesting that these genes may have a distinct biological nature in PABC. Furthermore, EGFR, IGF1, PTGS2, FGF1, CAV1, and PLCB1, were verified to be expressed differentially in PABC compared with normal tissues, and IGF1, PTGS2, and FGF1 were demonstrated to be significantly related to breast cancer patient prognosis, suggesting that the three genes might serve as biomarkers for precision diagnosis and treatment of PABC.