Epigenome erosion and SOX10 drive neural crest phenotypic mimicry in triple-negative breast cancer.

Intratumoral heterogeneity is caused by genomic instability and phenotypic plasticity, but how these features co-evolve remains unclear. SOX10 is a neural crest stem cell (NCSC) specifier and candidate mediator of phenotypic plasticity in cancer. We investigated its relevance in breast cancer by immunophenotyping 21 normal breast and 1860 tumour samples. Nuclear SOX10 was detected in normal mammary luminal progenitor cells, the histogenic origin of most TNBCs. In tumours, nuclear SOX10 was almost exclusive to TNBC, and predicted poorer outcome amongst cross-sectional (p = 0.0015, hazard ratio 2.02, n = 224) and metaplastic (p = 0.04, n = 66) cases. To understand SOX10's influence over the transcriptome during the transition from normal to malignant states, we performed a systems-level analysis of co-expression data, de-noising the networks with an eigen-decomposition method. This identified a core module in SOX10's normal mammary epithelial network that becomes rewired to NCSC genes in TNBC. Crucially, this reprogramming was proportional to genome-wide promoter methylation loss, particularly at lineage-specifying CpG-island shores. We propose that the progressive, genome-wide methylation loss in TNBC simulates more primitive epigenome architecture, making cells vulnerable to SOX10-driven reprogramming. This study demonstrates potential utility for SOX10 as a prognostic biomarker in TNBC and provides new insights about developmental phenotypic mimicry-a major contributor to intratumoral heterogeneity.

Characterization of Immune Cell Subsets of Tumor Infiltrating Lymphocytes in Brain Metastases.

The heterogeneity of tumor infiltrating lymphocytes (TILs) is not well characterized in brain metastasis. To address this, we performed a targeted analysis of immune-cell subsets in brain metastasis tissues to test immunosuppressive routes involved in brain metastasis. We performed multiplex immunofluorescence (mIF), using commercially available validated antibodies on formalin-fixed paraffin embedded whole sections. We quantitated the subsets of immune-cells utilizing a targeted panel of proteins including PanCK, CD8, CD4, VISTA and IBA-1, and analyzed an average of 15,000 cells per sample. Classifying tumors as either high (>30%) or low (<30%) TILs, we found that increased TILs density correlated with survival. Phenotyping these TILs we found tumors with low TILs had significantly higher expression of the immune-checkpoint molecule VISTA in tumor cells (p < 0.01) as well as in their microenvironment (p < 0.001). Contrastingly, the tumors with high TILs displayed higher levels of microglia, as measured by IBA-1 expression. Low TILs-tumors displayed CD8+ T-cells that co-express VISTA (p < 0.01) significantly more compared to high TILs group, where CD8+cells significantly co-express IBA-11 (p < 0.05). These results were supported by RNA analysis of a publicly available, independent cohort. Our work contributes to a growing understanding of the immune surveillance escape routes active in brain metastasis.

Clinicopathologic significance of nuclear HER4 and phospho-YAP(S127) in human breast cancers and matching brain metastases.

BACKGROUND: Human epidermal growth factor receptor-4 (HER4) and yes-associated protein-1 (YAP) are candidate therapeutic targets in oncology. YAP's transcriptional coactivation function is modulated by the HER4 intracellular domain (HER4-ICD) in vitro, but the clinical relevance of this has not been established. This study investigated the potential for targeting the HER4-YAP pathway in brain metastatic breast cancer. METHODS: We performed immuno-phenotypic profiling of pathway markers in a consecutive breast cancer series with 25 years of clinical follow up (n = 371), and patient-matched breast and metastatic brain tumours (n = 91; 30 pairs). RESULTS: Membrane localisation of phospho-HER4 [pHER4(Y1162)] was infrequent in primary breast cancer, but very frequent in brain metastases (5.9% versus 75% positive), where it was usually co-expressed with pHER3(Y1289) (p < 0.05). The presence of YAP in tumour cell nuclei was associated directly with nuclear pERK5(T218/Y210) (p = 0.003). However, relationships with disease-specific survival depended on oestrogen receptor (ER) status. Nuclear pYAP(S127) was associated with smaller, good prognostic ER+ breast tumours (log-rank hazard-ratio 0.53; p = 9.6E-03), but larger, poor prognostic triple-negative cancers (log-rank hazard-ratio 2.78; p = 1.7E-02), particularly when co-expressed with nuclear HER4-ICD (p = 0.02). This phenotype was associated with stemness and mitotic instability markers (vimentin, SOX9, ID1, SPAG5, TTK, geminin; p < 0.05). YAP expression in brain metastases was higher than matched primary tumours; specifically, nuclear pYAP(S127) in ER-negative cases (p < 0.05). Nuclear YAP was detected in ~70% of ER-negative, HER4-activated brain metastases. DISCUSSION: Our findings suggest that the canonical-mechanism where Hippo pathway-mediated phosphorylation of YAP ostensibly excludes it from the nucleus is dysfunctional in breast cancer. The data are consistent with pYAP(S127) having independent transcriptional functions, which may include transducing neuregulin signals in brain metastases. Consistent with mechanistic studies implicating it as an ER co-factor, nuclear pYAP(S127) associations with breast cancer clinical outcomes were dependent on ER status. CONCLUSION: Preclinical studies investigating HER4 and nuclear YAP combination therapy strategies are warranted.

Using whole-genome sequencing data to derive the homologous recombination deficiency scores.

The homologous recombination deficiency (HRD) score was developed using whole-genome copy number data derived from arrays as a way to infer deficiency in the homologous recombination DNA damage repair pathway (in particular BRCA1 or BRCA2 deficiency) in breast cancer samples. The score has utility in understanding tumour biology and may be indicative of response to certain therapeutic strategies. Studies have used whole-exome sequencing to derive the HRD score, however, with increasing use of whole-genome sequencing (WGS) to characterise tumour genomes, there has yet to be a comprehensive comparison between HRD scores derived by array versus WGS. Here we demonstrate that there is both a high correlation and a good agreement between array- and WGS-derived HRD scores and between the scores derived from WGS and downsampled WGS to represent shallow WGS. For samples with an HRD score close to threshold for stratifying HR proficiency or deficiency there was however some disagreement in the HR status between array and WGS data, highlighting the importance of not relying on a single method of ascertaining the homologous recombination status of a tumour.

LobSig is a multigene predictor of outcome in invasive lobular carcinoma.

Invasive lobular carcinoma (ILC) is the most common special type of breast cancer, and is characterized by functional loss of E-cadherin, resulting in cellular adhesion defects. ILC typically present as estrogen receptor positive, grade 2 breast cancers, with a good short-term prognosis. Several large-scale molecular profiling studies have now dissected the unique genomics of ILC. We have undertaken an integrative analysis of gene expression and DNA copy number to identify novel drivers and prognostic biomarkers, using in-house (n = 25), METABRIC (n = 125) and TCGA (n = 146) samples. Using in silico integrative analyses, a 194-gene set was derived that is highly prognostic in ILC (P = 1.20 × 10-5)-we named this metagene 'LobSig'. Assessing a 10-year follow-up period, LobSig outperformed the Nottingham Prognostic Index, PAM50 risk-of-recurrence (Prosigna), OncotypeDx, and Genomic Grade Index (MapQuantDx) in a stepwise, multivariate Cox proportional hazards model, particularly in grade 2 ILC cases (χ 2, P = 9.0 × 10-6), which are difficult to prognosticate clinically. Importantly, LobSig status predicted outcome with 94.6% accuracy amongst cases classified as 'moderate-risk' according to Nottingham Prognostic Index in the METABRIC cohort. Network analysis identified few candidate pathways, though genesets related to proliferation were identified, and a LobSig-high phenotype was associated with the TCGA proliferative subtype (χ 2, P < 8.86 × 10-4). ILC with a poor outcome as predicted by LobSig were enriched with mutations in ERBB2, ERBB3, TP53, AKT1 and ROS1. LobSig has the potential to be a clinically relevant prognostic signature and warrants further development.

Molecular signatures in breast cancer.

The use of molecular signatures to add value to standard clinical and pathological parameters has impacted clinical practice in many cancer types, but perhaps most notably in the breast cancer field. This is, in part, due to the considerable complexity of the disease at the clinical, morphological and molecular levels. The adoption of molecular profiling of DNA, RNA and protein continues to reveal important differences in the intrinsic biology between molecular subtypes and has begun to impact the way patients are managed. Several bioinformatic tools have been developed using DNA or RNA-based signatures to stratify the disease into biologically and/or clinically meaningful subgroups. Here, we review the approaches that have been used to develop gene expression signatures into currently available diagnostic assays (e.g., OncotypeDX® and Mammaprint®), plus we describe the latest work on genome sequencing, the methodologies used in the discovery process of mutational signatures, and the potential of these signatures to impact the clinic.