(616g) Multivariate Model of an Engineered Niche Delineates Metastatic Potential of Breast Cancer

Background: Approximately 4% of all women in the United States will develop metastatic breast cancer in their lifetime. Currently, metastatic breast cancer has a 10-year survival rate of less than 15%. One of the major inhibitory factors towards improving these outcomes is the inability to detect metastatic disease prior to macroscopic secondary tumor formation, at which point millions of tumor cells have colonized distal sites. Development of metastatic tumors requires the cancer cells and local microenvironment to undergo a complex series of events referred to as the metastatic cascade. Identification of systemic markers correlating with these events could be the key to the early detection, leading to the earlier treatment of metastatic breast cancer and improvements in survival outcomes.

We have designed a porous, polycaprolactone scaffold that acts as a synthetic metastatic niche in mouse models of breast and pancreatic cancer. The scaffolds, when implanted subcutaneously, recruit aggressive populations of tumor cells prior to their detection at native metastatic sites (e.g., lungs and livers). Analysis of the microenvironment at the scaffold can more successfully predict disease progression and recurrence than either the blood or primary tumor. Herein, we employed this scaffold technology to analyze changes in the microenvironment that can delineate metastatic cancer from non-metastatic disease using cell lines derived from a single spontaneous breast tumor in a BALB/cfC3H mouse. Each cell line correlates with a different stage of the metastatic cascade: 67NR is non-metastatic, 4T07 is micro-metastatic, and 4T1 is highly metastatic. By utilizing microenvironmental changes at the scaffold, we developed a 9-gene signature that can effectively identify the metastatic potential of triple-negative breast cancer throughout disease progression.

Methods: Porous, polycaprolactone (PCL) scaffolds were implanted subcutaneously into the backs of female BALB/c mice. After two weeks, mice received an orthotopic inoculation of either the 67NR, 4T07, or 4T1 triple-negative cell line in the fourth, right mammary fat pad. The tumors were allowed to progress for 7, 14, or 21 days before we collected the scaffolds, lungs, primary tumors, and either the spleen or blood (as a systemic control). These time points correlate with the pre-metastatic, micro-metastatic, and metastatic niche in the 4T1 model. At 14 days post-inoculation, tissues were collected for RNA-sequencing analysis using a NovaSeq (S1). The raw counts were normalized by DESeq2 and the normalized tables were used for signaling pathway enrichment (GSEA) and determination of a gene set that isolates the 4T1 samples from the other conditions (mixOmics). Tissues were collected at each time point to investigate immune cell distributions and validate the gene signature with PCR.

Results: We first analyzed the primary tumor to measure its ability to differentiate metastatic potential. Clinically, biopsy of the primary tumor is often the only indicator for invasive cancers if secondary tumors cannot be detected through imaging. The immune cell populations in the 4T07 micro-metastatic and 4T1 metastatic tumors were highly comparable. Of the nine populations tested, only the Gr1+CD11b+ neutrophils were significantly different between the two tumors, which only varied by 28%. In contrast, the 67NR non-metastatic tumors had around 3 times as many F4/80+CD11b+ macrophages, 11-fold less CD11c+F4/80- dendritic cells, 3-fold less Gr1+Cd11b+ neutrophils, and 2-fold less CD4+ T cells. Furthermore, gene expression at the primary tumor was unable to differentiate the 4T07 and 4T1 cell lines. Combined, these data indicate that analysis of the primary tumors cannot identify metastatic potential of invasive cancers.

We next investigated the scaffolds to assess their ability to identify metastatic cancer cell lines. Immune cell populations were unable to distinguish the 4T07 and 4T1 cell lines in either the scaffold or the spleen for the pre- or micro-metastatic conditions, suggesting that immune cell numbers are unable to predict metastasis locally (primary tumor) or systemically (scaffolds and spleen). Interestingly, gene expression at the scaffold was successfully able to predict metastatic potential. Angiogenesis, response to wound healing, monocyte chemotaxis, T cell-mediated cytotoxicity, and epithelial-to-mesenchymal transition were among the cancer-related signaling pathways that were enriched in the 4T1 samples in the scaffold and the lung but were not predicted in the spleen. The gene expression data was then utilized to develop a 9-gene signature that differentiated the 4T1 samples from the healthy (control), 67NR, and 4T07 samples in the scaffold. This signature consisted of Dhx9, Dusp12, Fhl1, Ifitm1, Ndufs1, Pja2, Slc1a3, Soga1, and Spon2 and, upon validation with PCR, was able to isolate the 4T1 samples at day 7, day 14, and day 21 (Figure 1).

Conclusions: Changes in gene expression at the scaffolds identified the enrichment of cancer-related pathways in response to the inoculation of 4T1 (metastatic) tumors relative to the healthy controls, 67NR (non-metastatic), and 4T07 (micro-metastatic) tumors. Signaling changes at the scaffold mimicked the response at the lung, demonstrating the potential of the scaffold to act as a surrogate tissue for native metastatic sites. We established a 9-gene signature from the scaffold that can identify the metastatic potential of breast cancer prior to the development of macroscopic secondary tumors. Such a signature could classify patients in need of systemic therapy at earlier stages in disease progression, which could improve outcomes.