Search

Background Triple-negative breast cancer (TNBC) is a heterogeneous disease and we have previously shown that rapid relapse of TNBC is associated with distinct sociodemographic features. We hypothesized that rapid versus late relapse in TNBC is also defined by distinct clinical and genomic features of primary tumors. Methods Using three publicly-available datasets, we identified 453 patients diagnosed with primary TNBC with adequate follow-up to be characterized as ‘rapid relapse’ (rrTNBC; distant relapse or death ≤2 years of diagnosis), ‘late relapse’ (lrTNBC; > 2 years) or ‘no relapse’ (nrTNBC: > 5 years no relapse/death). We explored basic clinical and primary tumor multi-omic data, including whole transcriptome (n = 453), and whole genome copy number and mutation data for 171 cancer-related genes (n = 317). Association of rapid relapse with clinical and genomic features were assessed using Pearson chi-squared tests, t-tests, ANOVA, and Fisher exact tests. We evaluated logistic regression models of clinical features with subtype versus two models that integrated significant genomic features. Results Relative to nrTNBC, both rrTNBC and lrTNBC had significantly lower immune signatures and immune signatures were highly correlated to anti-tumor CD8 T-cell, M1 macrophage, and gamma-delta T-cell CIBERSORT inferred immune subsets. Intriguingly, lrTNBCs were enriched for luminal signatures. There was no difference in tumor mutation burden or percent genome altered across groups. Logistic regression mModels that incorporate genomic features significantly outperformed standard clinical/subtype models in training (n = 63 patients), testing (n = 63) and independent validation (n = 34) cohorts, although performance of all models were overall modest. Conclusions We identify clinical and genomic features associated with rapid relapse TNBC for further study of this aggressive TNBC subset.

Platelet-derived growth factor receptor-beta (PDGFRβ) is a receptor tyrosine kinase found in cells of mesenchymal origin such as fibroblasts and pericytes. Activation of this receptor is dependent on paracrine ligand induction, and its preferred ligand PDGFB is released by neighboring epithelial and endothelial cells. While expression of both PDGFRβ and PDGFB has been noted in patient breast tumors for decades, how PDGFB-to-PDGFRβ tumor–stroma signaling mediates breast cancer initiation, progression, and metastasis remains unclear. Here we demonstrate this paracrine signaling pathway that mediates both primary tumor growth and metastasis, specifically, metastasis to the brain. Elevated levels of PDGFB accelerated orthotopic tumor growth and intracranial growth of mammary tumor cells, while mesenchymal-specific expression of an activating mutant PDGFRβ (PDGFRβD849V) exerted proproliferative signals on adjacent mammary tumor cells. Stromal expression of PDGFRβD849V also promoted brain metastases of mammary tumor cells expressing high PDGFB when injected intravenously. In the brain, expression of PDGFRβD849V was observed within a subset of astrocytes, and aged mice expressing PDGFRβD849V exhibited reactive gliosis. Importantly, the PDGFR-specific inhibitor crenolanib significantly reduced intracranial growth of mammary tumor cells. In a tissue microarray comprised of 363 primary human breast tumors, high PDGFB protein expression was prognostic for brain metastases, but not metastases to other sites. Our results advocate the use of mice expressing PDGFRβD849V in their stromal cells as a preclinical model of breast cancer–associated brain metastases and support continued investigation into the clinical prognostic and therapeutic use of PDGFB-to-PDGFRβ signaling in women with breast cancer. Significance: These studies reveal a previously unknown role for PDGFB-to-PDGFRβ paracrine signaling in the promotion of breast cancer brain metastases and support the prognostic and therapeutic clinical utility of this pathway for patients. See related article by Wyss and colleagues, p. 594

Copy number aberrations (CNAs) are gains and losses of large genomic segments present across most cancer types and are a hallmark of cancer genomic alterations. However, the processes underlying CNAs and characteristic patterns of CNAs are poorly understood. Using single nucleotide variant (SNV) data, bioinformatic advances have identified underlying mutational signatures resulting from distinct mutational processes. Mutational signatures have led to a variety of discoveries, several of which are being investigated in clinical management of cancer. The development of algorithms able to uncover similar signatures for CNAs, rather than SNVs, is still in its infancy. Here we present an analysis package for the R programming language called CNSigs that allows for the robust and reproducible derivation of copy number signatures. Based on a list of extracted copy number features previously verified in ovarian cancer, we utilize mixed model approaches and non-negative matrix factorization to derive CNA signatures across cancer types. The development of a package to derive signatures from copy number data allows further investigation of underlying processes that may be responsible for these CNA fingerprints. To verify the reproducibility of the signatures, we derived signatures from two independent breast cancer datasets that use distinct copy number segmentation approaches. From these independent datasets we were able to recover the same signatures with high accuracy. We identified five signatures that are distinct from known breast cancer receptor-based or expression-based subtypes, yet reveal unique associations with underlying mutations, mutational processes, and transcriptional programs. To validate robustness, we applied the pipeline to 11 different cancer types from the TCGA dataset and showed that we were able to derive signatures from all of these cancer types of varying sample sizes. We identified certain signatures that cut across tumor types, while others are distinct to individual cancers. This work lays the groundwork for further analysis into the underlying molecular processes leading to these copy number signatures seen across cancer types. The CNSigs package also allows researchers to easily analyze their own samples to look for signatures in their copy number profiles and to compare these to signatures previously derived for their cancer type. Citation Format: David Tallman, Sinclair Stockard, Zachary T. Weber, Sarah Asad, Katharine Collier, Elizabeth Adams, Jerome Bey, Daniel G. Stover. CNSigs: An R package for the identification of copy number signatures [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 5471.