Abstract P5-18-08: Defining molecular signatures to personalise management of patients with early breast cancer

Background

A review of breast screening highlighted the need to reduce overdiagnosis. Ductal Carcinoma In-Situ (DCIS) contributes significantly to this overdiagnosis. Epithelial cells in DCIS are as genetically advanced as those in invasive disease, focusing attention on the tumour microenvironment (ME). A key components of the ME in DCIS is the myoepithelial cell(MEC). These cells lie at the interface of the epithelial and stromal compartments, regulating cell function. We previously have identified changes in the MEC that contribute to tumour progression. Here we investigate the functional and clinical significance of a novel change in MEC phenotype: loss of Galectin-7 (Gal-7) expression. Gal-7 is proposed to play a role in apoptosis. We hypothesise that changes in MEC phenotype in DCIS alter the ME towards a pro-invasive phenotype, and hypothesise that loss of Gal-7 modifies the ME, destabilizes the MEC interface and ultimately may lead to loss of the MEC population through apoptosis.

Methods

Gal-7 expression and function was investigated in clinical samples and in-vitro model systems, respectively.

Gal-7 expression was assessed in a series of pure DCIS samples (low risk model) and DCIS with co-existant invasion (high risk model). Tissue sections were stained for Gal-7 and MEC expression scored on a duct-by-duct basis as positive, heterogeneous or negative.

An in-vitro model of normal primary myoepithelial cells isolated from reduction mammoplasty was used to investigate the functional impact of loss of Gal-7. These cells have high endogenous levels of Gal-7. Gal-7 was knocked down using siRNA and apoptosis assessed using cleaved caspase-3. The effect of Gal-7 on MEC layer integrity was assessed using immunofluorescence and adhesion assays.

The global impact of loss of Gal-7 was investigated using RNA sequencing.

Results

In the tissue analysis 1926 DCIS ducts were scored for MEC expression of Gal-7. Significantly more ducts showed loss of Gal-7 in DCIS with co-existant invasion, with pure DCIS showing 388 ducts positive and DCIS with invasion 144 DCIS ducts positive (p=0.0014). Pure DCIS and DCIS with invasion had 99 and 646 negative DCIS ducts respectively (p=0.0002).

In model systems of primary MEC, knockdown of Gal-7 resulted in increased expression of cleaved caspase-3, suggesting lower levels of Gal-7 increases apoptosis. In functional assays silencing Gal-7 reduces adhesion to both fibronectin and laminin extracellular matrices (p-value 0.005 and 0.001 respectively)

RNA sequencing indicates silencing Gal-7 increases LOX expression - a key regulator of the collagen matrix of the microenvironment.

Conclusion

Normal MEC strongly express Gal-7. Expression is lost in DCIS, with significantly more frequent loss in DCIS with co-existant invasion, suggesting that loss is associated with a more advanced phenotype. Functional assays indicate that loss of MEC Gal-7 enhances MEC apoptosis, which may be one mechanism by which this interface is lost during progression. Gal-7 negative MEC also show impaired adhesion to matrix proteins and lead to up-regulation of LOX, an enzyme key in promoting tumourigenesis. The incorporation of Gal-7 expression into a risk stratification algorithm has functional evidence and is currently being investigated.

Citation Format: Allen N, Allen M, Ahmed K, Gomm J, Nelan R, Nagano A, Chelala C, Gadaleta E, Thorat M, Cuzick J, Jones LJ. Defining molecular signatures to personalise management of patients with early breast cancer [abstract]. In: Proceedings of the 2018 San Antonio Breast Cancer Symposium; 2018 Dec 4-8; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2019;79(4 Suppl):Abstract nr P5-18-08.

Abstract PD04-08: Cell cycle algorithm correlates with grade of DCIS and p53 status, allows elimination of ‘intermediate grade’ disease and gives clinically meanignful information

We have previously shown that multi-parameter analysis of the DNA replication initiation machinery (Mcm2-7, geminin) and mitotic proteins (Plk1, Aurora A, H3S10ph), expressed during various phases of the cell division cycle, provides a method to accurately assess the proliferative state of dynamic tumour cell populations. We have shown that this novel form of cell cycle biomarker analysis allows separation of breast cancers into three discrete cell cycle phenotypes of major prognostic significance including (i) an out-of-cycle state, (ii) a G1-delayed/arrested state and (iii) an actively cycling state. Importantly, we identified groups of patients with apparently good grade cancers on routine clinicopathological criteria but who nevertheless exhibited the poor prognostic actively cycling phenotype (iii) and might therefore benefit from adjuvant chemotherapy. Additionally, we identified patients with typically poor prognosis tumours but which were actually cycling slowly (phenotype i or ii) and therefore likely not to have benefited from chemotherapy. Notably, when this cell cycle algorithm was used in a multi-variate analysis, the effect of Ki-67 disappeared.

We have now applied the algorithm to 72 cases of DCIS (all of which were re-graded by one pathologist for consistency) and also to a TMA of 136 cases with known outcome from the UK DICS trial. There was excellent correlation between tumour grade and p53 status. The high grade DCIS was positively associated with a high cell cycle score (actively cycling phenotype iii; 28 out of 34 tumours) and low grade DCIS was associated with a low score (namely phenotypes i and ii; 13 out of 19). Interestingly, 7 of the 10 intermediate grade DCIS could be classified as phenotype, i or iii suggesting that the algorithm may be useful in stratifying this group of patients.

A clinical correlation between phenotype III and recurrence and/or progression to invasive disease was evident. Further analysis of this from data derived from the TMAs generated from the UK DCIS trial will be presented. If the early findings are confirmed we may be able to solve the dilemma of which cases of DCIS need treatment and which we currently overtreat.

Citation Information: Cancer Res 2012;72(24 Suppl):Abstract nr PD04-08.

Correlation of FOXA1 expression with Oncotype Dx recurrence scores

11058

Background: Oncotype Dx assay (ODX) is a diagnostic assay that predicts the benefit of chemotherapy and assesses the likelihood of breast cancer (BC) recurrence in patients with estrogen receptor (ER) positive BC. The forkhead-box A1 (FOXA1) transcription factor is linked to ER regulated gene transcription and its expression is associated with ER positive luminal A type BC. Since higher FOXA1 expression is associated with ER positive luminal A BC, we investigated whether FOXA1 expression levels show a correlation with ODX recurrence scores in BC patients. Methods: 120 cases of BC from IU and affiliated hospitals had ODX performed on tumor blocks. 78 patient tumor blocks with ODX recurrence score have been analyzed for FOXA1 using IHC. FOXA1 expression was measured using intensity X percentage scoring. The correlation of FOXA1 scores and ODX recurrence scores with each other and with other variables was examined. Other variables included age, progesterone receptor (PR) status, tumor size, grade, histological type, and race. All p values are two-sided and have been adjusted for multiple comparisons using Bonferroni-Holm's method. Results: Twenty-one of 78 cases analyzed were < 50 years. ODX recurrence scores were low, intermediate, and high in 55.1%, 32.1%, and 12.8% of cases, respectively. Ductal carcinomas accounted for 73.1% of tumors, while 11.5% were lobular. Fifty percent of tumors were histologic grade II, 37% and 13% were grade I and III, respectively. PR was negative in 9 cases. Caucasians accounted for 78.2% of cases vs 7.7% African-Americans. FOXA1 expression correlated negatively with ODX recurrence score (p=.003), and histologic type (p=.024). ODX recurrence score also correlated negatively with PR (p=.034) with 78% of the PR negative cases having a high or intermediate ODX score. The correlation between FOXA1 expression and ODX recurrence score remained significant after adjusting for multiple comparisons and controlling for confounders such as histological type, grade, and PR. Conclusions: We demonstrate that FOXA1 expression has an inverse correlation with ODX recurrence scores. FOXA1 expression can potentially be used as a prognostic marker in low risk ER positive BC patients who lack access to ODX assays. This observation needs to be confirmed with a larger sample size.

No significant financial relationships to disclose.

Molecular characteristics of matched brain metastasis (BM) versus the primary breast cancer (PBC)

Abstract #2028

Background: Brain is increasingly a site of relapse in breast cancer patients, however the molecular patho-physiology of this process is not well understood. Using a novel assay allowing high-throughput analysis of gene expression from formalin-fixed paraffin-embedded (FFPET) tumor samples, we compared molecular characteristics of BM with those of the PBC in a series of breast cancer patients who underwent excision of brain metastasis.&#x2028; Methods: In each patient we performed molecular analysis of paired archived FFPET specimens from both the PBC and excised BM. Of the 37 pairs of PBC/BM samples, 24 pairs had sufficient pathology material for molecular analysis. Patient characteristics: mean age at PBC diagnosis was 46.7 years (range 27-67 years); 7 (29%) ER+, 17 (71%) ER-, 12 (50%) HER2-positive(3+). We performed cDNA-mediated annealing, selection, extension and ligation (DASL) assay (Illumina Corp) for expression of 502 known cancer genes, using 200 ng RNA. Statistical analysis for microarrays (SAM) accounting for the pairing of the primary and metastasized tumors was used to identify differentially expressed genes while controlling the false discovery rate (FDR &lt;0.01). BeadStudio™ Absolute correlation clustering was used to cluster samples based on their expression profiles.&#x2028; Results: In only 7 (29%) cases matched PBC and BM pair clustered together on unsupervised hierarchical clustering, the pairs in remaining cases clustered apart. Comparison of the two groups (PBC and BM) showed that 41 genes were up regulated in BM, including proliferation genes (CDC2, CDC25a, CCNA2 and E2F family), anti-apoptotic (BIRC5), DNA repair (RAD51, -54b, XRCC2, BRCA2, BARD1, TOP1), angiogenesis (VEGF), and development of drug resistance (ABCB1, -G2). In contrast, the 43 genes that were down regulated in BM included those associated with invasion (MMP2, -3, -14), cellular motility, and epithelial to mesenchymal transformation (CDH11). Additional analysis to validate these trends and to identify potential therapeutic targets is underway.&#x2028; Conclusions: Although some BM retain remarkable similarity to the PBC, majority exhibit considerable deviation in their gene expression profile. These “adaptive” changes include greater resistance to drug therapy, increased DNA repair, a reversal back to the epithelial phenotype and decreased capacity for cell motility and invasion.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 2028.

Control of luminal type A intrinsic subtype enriched transcription factor network by insulin: implications of diabetes on breast cancer classification

Abstract #3029

Background: Luminal type A and type B represent estrogen receptor alpha (ERalpha)-positive breast cancers with luminal type A expressing higher levels of ERalpha and is associated with better prognosis. Recent studies have identified a specific functional transcription factor network comprising GATA-3, FOXA1 and ERalpha in normal luminal cells as well as in luminal type A breast cancer that dictates their hormone dependence. Signaling molecules that may disrupt this network and force these cells to acquire hormone-independence are not known. T-bet (Tbx21) has been described as a major negative regulator of GATA-3 activity. As the expression and/or activity of some of the above factors are controlled by insulin, the objective of this study was to investigate whether elevated level of insulin, as evidenced in type II diabetes, alters gene expression pattern in luminal type A breast cancers by interrupting GATA-3:FOXA1:ERalpha network and thus forcing these cancers to acquire non-luminal phenotype and/or hormone-independence.&#x2028; Methodologies: The effect of insulin on the expression of ERalpha, FOXA1, GATA-3 and T-bet was measured in ERalpha-positive MCF-7 cells by Western blot analysis. The effect of T-bet on estrogen-regulated gene expression was measured by stable overexpression of T-bet in MCF-7 cells and subsequent qRT-PCR analysis. Publicly available Oncomine database was used to determine the expression pattern of T-bet and its relation to ERalpha status in primary breast cancers. A proliferation assay was used to determine sensitivity of T-bet overexpressing cells to tamoxifen in the presence and absence of insulin.&#x2028; Results: Insulin induced the expression of T-bet, which was partially reversed by estrogen. ERalpha and GATA-3 levels were reduced in MCF-7 cells stably overexpressing T-bet suggesting that T-bet reduces GATA-3-dependent ERalpha expression. Estrogen-inducible expression of estrogen target genes GREB-1 and Myb was lower in T-bet overexpressing cells compared to parental cells, although basal expression was elevated in T-bet overexpressing cells. Treatment of T-bet overexpressing cells with insulin decreased tamoxifen sensitivity. Although T-bet expression was generally higher in ERalpha-negative breast cancers compared to ERalpha-positive breast cancers, a subpopulation of ER-positive breast cancers express elevated levels of T-bet.&#x2028; Conclusions: Insulin may change the gene expression pattern through T-bet-mediated disruption of master cell-type-specific transcriptional network including GATA-3, ERalpha and FOXA1 that dictates the phenotype of hormone-dependent luminal type A breast cancer. T-bet may serve as a marker to identify ERalpha-positive breast cancers that express low levels of GATA-3 and have progressed to hormone-independence.

Citation Information: Cancer Res 2009;69(2 Suppl):Abstract nr 3029.