The Cancer Genome Atlas Breast Research Articles

SNRPB2 promotes triple-negative breast cancer progression by controlling alternative splicing of MDM4 pre-mRNA.

Alternative splicing generates cancer-specific transcripts and is now recognized as a hallmark of cancer. However, the critical oncogenic spliceosome-related proteins involved in triple-negative breast cancer (TNBC) remain elusive. Here, we explored the expression pattern of spliceosome-related proteins in TNBC, non-TNBC, and normal breast tissues from The Cancer Genome Atlas breast cancer (TCGA-BRCA) cohort, revealing higher expression of nearly half of spliceosome-related proteins in TNBC than their counterparts. Among these TNBC-specific spliceosome-related proteins, the expression of SNRPB2 was associated with poor prognosis in patients with TNBC. In TNBC cells, the knockdown of SNRPB2 strongly suppressed cell proliferation and invasion and induced cell cycle arrest. Mechanistically, transcriptome data showed that SNRPB2 knockdown inactivated E2F1 signaling, which regulated the cell cycle. We further validated the downregulation of several cell cycle genes in SNRPB2 knockdown cells. Moreover, the analysis showed that SNRPB2 knockdown triggered the alteration of many alternative splicing events, most of which were skipping of exon. In TNBC cells, it was found that SNRPB2 knockdown led to the skipping of exon 6 in MDM4 pre-mRNA, generating MDM4-S transcript and downregulating MDM4 protein expression. More importantly, downregulation of MDM4 decreased retinoblastoma 1 (Rb1) protein expression, which is a target of MDM4 and a regulator of E2F1 signaling. In summary, the current study revealed an SNRPB2/MDM4/Rb axis in promoting the progression of TNBC, providing novel insights and novel targets for combating TNBC.

Abstract PO2-01-08: Differential HOXB13 gene expression and promoter methylation analysis in breast cancer

Abstract Background: The Breast Cancer Index (BCI) is a gene expression-based signature comprising two functional biomarker panels: the Molecular Grade Index (MGI) and BCI (H/I), which is a ratio of the HOXB13 and IL17BR gene expression. Several studies have shown that BCI (H/I) is a predictive biomarker for extended adjuvant endocrine therapy benefit in hormone receptor-positive (HR+) early-stage breast cancer. Molecular mechanisms underlying endocrine responsiveness remain to be fully explored. The objective of this analysis is to evaluate the correlation between gene expression and methylation in breast cancer patients displaying different HOXB13 mRNA expression levels from The Cancer Genome Atlas (TCGA) project. Methods: Data from methylation microarrays and mRNA sequencing were examined in combination with clinical metadata from 1,095 TCGA breast cancer patients. After removing outliers, HR+ samples were ordered based on their HOXB13 expression levels and the top 10% of samples (HOXB13-high; n=95) and bottom 10% (HOXB13-low; n=95) were selected for further analysis. Differentially expressed genes (DEGs) were determined by comparing all protein-coding genes between the HOXB13-high and HOXB13-low sets, based on an absolute log-fold change above 2 and a Benjamini-Hochberg adjusted p-value threshold of 0.01. Among the subset of samples with matching methylation data (n=60 for the HOXB13-high group and n=48 for the HOXB13-low group), hypo-methylated and hyper-methylated differentially methylated probes (DMPs) were identified based on at least 10% change of β (i.e., average promoter methylation) and a Benjamini-Hochberg adjusted p-value threshold of 0.01. Motif and transcription factor (TF) enrichment analyses were identified at a minimum incidence of 10 and a lower boundary of 1.5 for the 95% confidence interval of the odds ratio. Results: Our analysis identified a total of 613 DEGs between the HOXB13-high and HOXB13-low groups. Gene ontology analysis of DEGs revealed a statistically significant enrichment of genes associated with PTK6 regulated cell cycle, oncogene-induced cellular senescence, and regulation of T cell activation. Analysis of protein-protein interactions of differentially expressed genes revealed differences in key biological processes including changes in cell cycle regulation, p53 pathway, and IL-17 signaling. A total of 5,295 hyper-methylated and 580 hypo-methylated DMPs were discovered when comparing HOXB13-high and HOXB13-low samples. Changes in global methylation patterns enabled classification of breast cancer samples with high or low HOXB13 expression with an accuracy of 0.90. Reconstruction of gene regulatory networks from DNA methylation and transcriptome profiles elucidated significant pairs of DMPs and differentially expressed genes. Specifically, a total of 197 differentially expressed genes were attributed to hyper-methylated DMPs, including genes involved in response to estrogen, activation of HOX genes during differentiation, and drug-mediated inhibition of CDK4/6 activity. Additionally, 8 differentially expressed genes were attributed to hypo-methylated DMPs and were associated with proteasomal protein catabolic processes. Lastly, motif analysis of transcriptional factors for the hyper-methylated DMPs revealed enrichment of binding motifs attributed to FOXA1, ESR1, XBR1, and FOXP1. Conclusion: A comprehensive and comparative analysis of mRNA expression levels and methylation patterns between TCGA breast cancer samples with high and low expression of HOXB13 revealed key signaling pathways and biological processes that may provide insights on the molecular mechanism of HOXB13 in the regulation of response to endocrine therapy in HR+ breast cancer. Citation Format: Natalia Siuliukina, Yi Zhang, Kai Treuner, Mark Pegram. Differential HOXB13 gene expression and promoter methylation analysis in breast cancer [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO2-01-08.

Abstract PO5-03-12: Genomic differences of primary triple negative breast cancer in patients younger than 45 years vs. patients older than 45 years of age

Abstract Background: Triple negative breast cancer (TNBC) is often associated with an aggressive clinical course, especially in younger women. We previously demonstrated that there are differences in the genomic spectrum of metastatic TNBC in patients who are ≤45 years and those who are >45 years of age (Vidula et al, SABCS, 2020). Here we explore genomic differences of primary TNBC in patients ≤45 years and patients >45 years to understand variations in the overall number of non-silent mutations and proportion of patients with specific gene mutations. Methods: We obtained filtered mutation calls generated by the Multi-Center Mutation Calling in Multiple Cancers (MC3) working group (Ellrott et al, Cell Syst, 2018), from whole exome sequencing data of 9079 The Cancer Genome Atlas (TCGA) samples across 33 cancer types that had been used for the comprehensive characterization of cancer driver genes and mutations (Bailey et al, Cell, 2018). Receptor subtype and age at diagnosis of 1247 TCGA breast cancer samples were obtained from the UCSC Xena browser (Goldman et al, Nat Biotechnol, 2020). These datasets were combined to identify 119 patients with primary TNBC with harmonized high confidence mutation calls for analysis. We compared the overall number of non-silent mutations between patients ≤45 years vs. >45 years using the Wilcoxon rank sum test. We also compared the proportion of patients with non-silent mutations in genes mutated at >10% frequency between age groups using the Fisher Exact test. Results: Altogether, 25 patients ≤45 years and 94 patients >45 years were identified with primary TNBC. No significant difference in the number of non-silent mutations among patients ≤45 years vs. >45 years was seen (median [range], ≤45 years: 65 [5-228] vs. >45 years: 74 [2-742], Wilcoxon rank sum p=0.57). In total, 35 unique genes were mutated at >10% frequency in either patients ≤45 years (30 genes) or >45 years (7 genes); notably, TP53 mutations were commonly seen in both age groups (≤45 years: 76%; >45 years: 79%, p=0.79) and TTN mutations were also frequently mutated in both cohorts (≤45 years: 20%; >45 years: 27%, p=0.61). Among the 35 genes identified, 12 genes were mutated at significantly different frequencies among patients ≤45 years vs. those >45 years, as depicted in Table 1. Conclusions: Relatively high frequencies of TP53 and TTN mutations were seen in both patients ≤45 years and those >45 years with primary TNBC. However, many genes associated with varying cellular functions including motility, signaling receptors, and growth were more frequently mutated in patients ≤45 years. These results suggest that there are genomic differences of primary TNBC in patients ≤45 years vs. those >45 years. Further research is needed to validate these findings in a larger cohort of primary TNBC since genomic differences may affect tumor biology and clinical behavior. Table 1. Significant Gene Mutation Differences in Primary TNBC Based on Age. Citation Format: Neelima Vidula, Leif Ellisen, Aditya Bardia, Christina Yau. Genomic differences of primary triple negative breast cancer in patients younger than 45 years vs. patients older than 45 years of age [abstract]. In: Proceedings of the 2023 San Antonio Breast Cancer Symposium; 2023 Dec 5-9; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2024;84(9 Suppl):Abstract nr PO5-03-12.

Abstract 6548: Leveraging compact feature sets for TCGA-based molecular subtype classification on new samples

Abstract The NCI's The Cancer Genome Atlas (TCGA) project profiled over 10,000 tumor samples over the course of 10 years. As different tissue-specific working groups reviewed all of the available data, these patient samples were separated into distinct molecular subtypes, and these clusters were reported in various marker papers. While these assignments provided invaluable information about the common patterns of molecular characteristics in different types of cancer there was no consistent methodology for assigning new samples to these defined molecular subtypes.The NCI's Tumor Molecular Pathology group was formulated to create machine learning-based models that could be applied to non-TCGA samples and determine their TCGA mapped subtypes. Five modeling systems, JADBio, SKGrid by the Oregon Health and Science University, CloudForest by the Institute of Systems Biology, AKLIMATE by University of California Santa Cruz and subSCOPE by BC Cancer’s Genome Sciences Centre, were trained to recognize TCGA subtypes using multi-omic measurements from gene expression, DNA methylation, miRNA expression, copy number, and somatic mutation calls. While the TCGA samples were profiled using multi-omic technologies, single platform and/or compact feature set models also were assessed for their ability to assign these classifications. Each machine learning system created predictive models for 106 subtypes from 26 cancer types using as few features as possible, with a maximum of 100 features allowed for scored models. A set of 411,706 models was developed, composed of results of each of the learning methods across the various omic platforms. Top models, both multi-omic and single platform, were selected for each cancer type. On average, models were able to achieve an overall weighted F1 score of 0.895 with 42 features. While the top models for each cancer type had an overall weighted F1 mean performance of 0.936 with a mean of 29 features, in 20 of the 26 cancer types models using only gene expression provided the best performance. Analysis of features selected by the models showed some known onco-drivers were selected by many models, but many times different models would utilize features of different genes with similar levels of performance. Network-level analysis revealed that many genes of these selected features operated within the same pathways.Transferability of these models to external datasets was tested, taking TCGA breast cancer trained models and applying them to AURORA and METABRIC datasets. Interestingly, despite the data platform difference between TCGA (RNAseq) and METABRIC (microarray), model performance saw only minimal degradation of F1 values in transfer. This set of models and the training dataset will provide new opportunities for researchers and translational scientists to connect new tumors to the subtypes seen in the TCGA cohorts. Citation Format: Kyle Ellrott, Chris K. Wong, Christina Yau, Mauro A. Castro, Jordan Lee, Brian Karlberg, Jasleen K. Grewal, Vincenzo Lagani, Bahar Tercan, Verena Friedl, Toshinori Hinoue, Vladislav Uzunangelov, Lindsay Westlake, Xavier Loinaz, Ina Felau, Peggy Wang, Anab Kemal, Samantha J. Caesar-Johnson, Ilya Shmulevich, Alexander J. Lazar, Ioannis Tsamardinos, Katherine A. Hoadley, The Cancer Genome Atlas Analysis Network, Gordon A. Robertson, Theo A. Knijnenburg, Christopher C. Benz, Joshua M. Stuart, Jean C. Zenklusen, Andrew D. Cherniack, Peter W. Laird. Leveraging compact feature sets for TCGA-based molecular subtype classification on new samples [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6548.

MicroRNA Expression Profile in Early-Stage Breast Cancers.

Breast cancer is one of the leading causes of cancer deaths in women. Early diagnosis offers the best hope for a cure. Ductal carcinoma in situ is considered a precursor of invasive ductal carcinoma of the breast. In this study, we carried out microRNA sequencing from 7 ductal carcinoma in situ (DCIS), 6 infiltrating ductal carcinomas (IDC Stage IIA) with paired normal, and 5 unpaired normal breast tissue samples. We have deployed miRge for microRNA analysis, DESeq for differential expression analysis, and Cytoscape for competing endogenous RNA network investigation. Here, we identified 76 miRNAs that were differentially expressed in DCIS and IDC. Additionally, we provide preliminary evidence of miR-365b-3p and miR-7-1-3p being overexpressed, and miR-6507-5p, miR-487b-3p, and miR-654-3p being downregulated in DCIS relative to normal breast tissue. We also identified a miRNA miR-766-3p that was overexpressed in earlystage IDCs. The overexpression of miR-301a-3p in DCIS and IDC was confirmed in 32 independent breast cancer tissue samples. Higher expression of miR-301a-3p is associated with poor overall survival in The Cancer Genome Atlas Breast Cancer (TCGA-BRCA) dataset, indicating that it may be associated with DCIS at high risk of progressing to IDC and warrants deeper investigation.

A multimodal graph neural network framework for cancer molecular subtype classification

BackgroundThe recent development of high-throughput sequencing has created a large collection of multi-omics data, which enables researchers to better investigate cancer molecular profiles and cancer taxonomy based on molecular subtypes. Integrating multi-omics data has been proven to be effective for building more precise classification models. Most current multi-omics integrative models use either an early fusion in the form of concatenation or late fusion with a separate feature extractor for each omic, which are mainly based on deep neural networks. Due to the nature of biological systems, graphs are a better structural representation of bio-medical data. Although few graph neural network (GNN) based multi-omics integrative methods have been proposed, they suffer from three common disadvantages. One is most of them use only one type of connection, either inter-omics or intra-omic connection; second, they only consider one kind of GNN layer, either graph convolution network (GCN) or graph attention network (GAT); and third, most of these methods have not been tested on a more complex classification task, such as cancer molecular subtypes.ResultsIn this study, we propose a novel end-to-end multi-omics GNN framework for accurate and robust cancer subtype classification. The proposed model utilizes multi-omics data in the form of heterogeneous multi-layer graphs, which combine both inter-omics and intra-omic connections from established biological knowledge. The proposed model incorporates learned graph features and global genome features for accurate classification. We tested the proposed model on the Cancer Genome Atlas (TCGA) Pan-cancer dataset and TCGA breast invasive carcinoma (BRCA) dataset for molecular subtype and cancer subtype classification, respectively. The proposed model shows superior performance compared to four current state-of-the-art baseline models in terms of accuracy, F1 score, precision, and recall. The comparative analysis of GAT-based models and GCN-based models reveals that GAT-based models are preferred for smaller graphs with less information and GCN-based models are preferred for larger graphs with extra information.

Abstract B012: Pedigree and multi-omics analyses implicate nucleotide excision repair genes, XPA, ERCC6, and ERCC8, in susceptibility to breast cancer

Abstract Background: The majority of low- and moderate-risk [odds ratios (OR) ≤ 3] susceptibility genes for peri- and post-menopausal breast cancer (≥50 years of age), particularly those with minor allele frequency (MAF) of <30%, have not yet been identified and/or validated. We used a hypothesis-driven integrative approach to identify breast cancer susceptibility genes by analyzing pedigree and multi-omics (genomics, transcriptomics, mutation landscape) data. Methods: We conducted pedigree analysis in 31 families with DNA repair disorders, xeroderma pigmentosum (XP), trichothiodystrophy (TTD), Cockayne syndrome (CS), XP/TTD and XP/CS. These disorders can be caused by mutations in genes involved in the nucleotide excision repair (NER) pathway such as XPA-XPG, ERCC6 and ERCC8. We estimated the frequency of hereditary breast cancer, defined as the presence of at least two cases of premenopausal breast cancer (diagnosed <50 years of age) or one case of premenopausal breast cancer and one case of ovarian cancer in one side of the family, in our study. Furthermore, we conducted individual analysis of four genome-wide association studies (GWAS) datasets (totaling 2961 cases and 4335 controls), individual and meta-analysis of two gene expression microarray datasets (98 cases and 163 controls), and targeted analysis of The Cancer Genome Atlas (TCGA) (522 breast cancers). All analyses involved invasive ductal breast carcinoma in women diagnosed at ≥50 years of age. Results: In an XP family with causative mutations in XPA, there were four cases of premenopausal breast cancer (including heterozygote mutation carriers), one case of ovarian cancer, and one case of prostate cancer. Comprehensive sequencing failed to identify a BRCA1 or BRCA2 mutation in this family. Overall, 7 of 31 (22.6%) families with DNA repair disorders had a family history that was compatible with hereditary breast cancer. Hence, we selected a panel of NER genes as candidates for association studies with breast cancer. Our candidate-gene association analysis of GWAS datasets suggested an increased risk of breast cancer with ERCC6 (main effect:1.29 ≤ OR ≤ 2.91, 0.005 ≤ p ≤ 0.04, 11.8 ≤ MAF ≤ 29.7%) and implicated its interaction with ERCC8 (joint effect:3.03 ≤ OR ≤ 5.31, 0.01 ≤ pinteraction ≤ 0.03). Our transcriptomics analysis identified significant upregulation of ERCC6 (p=7.95 × 10–6) and ERCC8 (p=4.67 × 10–6) in breast cancer. Our analysis of TCGA breast tumor data showed similar mutation frequency of ERCC6 (1.8%) and ERCC8 (0.3%) as known breast cancer susceptibility genes such as BLM (1.9%) and LSP1 (0.3%). Conclusion: Our integrative analyses suggest that ERCC6 may be a previously unreported low- to moderate-risk breast cancer susceptibility gene, which may also interact with ERCC8. ERCC6 and ERCC8 code for the main components of the transcription-coupled (TC) repair sub-pathway of the NER. TC-NER protects against genotoxic agents, such as free radicals and reactive oxygen species. Our results provide a deeper biologic insight and leads for prevention of common breast cancer. Citation Format: Roxana Moslehi, Amiran Dzutsev. Pedigree and multi-omics analyses implicate nucleotide excision repair genes, XPA, ERCC6, and ERCC8, in susceptibility to breast cancer [abstract]. In: Proceedings of the AACR Special Conference in Cancer Research: DNA Damage Repair: From Basic Science to Future Clinical Application; 2024 Jan 9-11; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2024;84(1 Suppl):Abstract nr B012.

Potential biomarkers as a predictive factor of response to primary chemotherapy in breast cancer patients.

In this study, we identified miRNAs and their potential mRNA targets that are intricately linked to primary chemotherapy response in patients with invasive ductal carcinomas. A cohort of individuals diagnosed with advanced invasive breast ductal carcinoma who underwent primary chemotherapy served as the cornerstone of our study. We conducted a comparative analysis of microRNA expression among patients who either responded or did not respond to primary systemic therapy. To analyze the correlation between the expression of the whole transcriptome and the 24 differentially expressed (DE) miRNAs, we harnessed the extensive repository of The Cancer Genome Atlas (TCGA) database. We mapped molecular mechanisms associated with these miRNAs and their targets from TCGA breast carcinomas. The resultant expression profile of the 24 DE miRNAs emerged as a potent and promising predictive model, offering insights into the intricate dynamics of chemotherapy responsiveness of advanced breast tumors. The discriminative analysis based on the principal component analysis identified the most representative miRNAs across breast cancer samples (miR-210, miR-197, miR-328, miR-519a, and miR-628). Moreover, the consensus clustering generated four possible clusters of TCGA patients. Further studies should be conducted to advance these findings.

Epigenetic Dysregulation and Its Correlation with the Steroidogenic Machinery Impacting Breast Pathogenesis: Data Mining and Molecular Insights into Therapeutics.

Breast cancer (BC) is a heterogeneous condition and comprises molecularly distinct subtypes. An imbalance in the levels of epigenetic histone deacetylases (HDACs), modulating estrogen accumulation, especially 17β-estradiol (E2), promotes breast tumorigenesis. In the present study, analyses of The Cancer Genome Atlas (TCGA) pan-cancer normalized RNA-Seq datasets revealed the dysregulation of 16 epigenetic enzymes (among a total of 18 members) in luminal BC subtypes, in comparison to their non-cancerous counterparts. Explicitly, genomic profiling of these epigenetic enzymes displayed increases in HDAC1, 2, 8, 10, 11, and Sirtuins (SIRTs) 6 and 7, and decreases in HDAC4-7, -9, and SIRT1-4 levels, respectively, in TCGA breast tumors. Kaplan-Meier plot analyses showed that these HDACs, with the exception of HDAC2 and SIRT2, were not correlated with the overall survival of BC patients. Additionally, disruption of the epigenetic signaling in TCGA BC subtypes, as assessed using both heatmaps and boxplots, was associated with the genomic expression of factors that are instrumental for cholesterol trafficking/utilization for accelerating estrogen/E2 levels, in which steroidogenic acute regulatory protein (STAR) mediates the rate-limiting step in steroid biosynthesis. TCGA breast samples showed diverse expression patterns of a variety of key steroidogenic markers and hormone receptors, including LIPE, CYP27A1, STAR, STARD3, CYP11A1, CYP19A1, ER, PGR, and ERBB2. Moreover, regulation of STAR-governed steroidogenic machinery was found to be influenced by various transcription factors, i.e., CREB1, CREM, SF1, NR4A1, CEBPB, SREBF1, SREBF2, SP1, FOS, JUN, NR0B1, and YY1. Along these lines, ingenuity pathway analysis (IPA) recognized a number of new targets and downstream effectors influencing BCs. Of note, genomic, epigenomic, transcriptional, and hormonal anomalies observed in human primary breast tumors were qualitatively similar in pertinent BC cell lines. These findings identify the functional correlation between dysregulated epigenetic enzymes and estrogen/E2 accumulation in human breast tumors, providing the molecular insights into more targeted therapeutic approaches involving the inhibition of HDACs for combating this life-threatening disease.

Dissecting tumor antigens and immune subtypes for mRNA vaccine development in breast cancer

PurposesCancer mRNA vaccines are a promising strategy and a hot topic in cancer immunotherapy. However, mRNA vaccines for breast cancer (BRCA) remain undeveloped. This study aimed to identify potential tumor antigens for mRNA vaccine development and a population with BRCA suitable for vaccination.MethodsGene expression profiles and the clinical information of the TCGA-BRCA (the Cancer Genome Atlas Breast Cancer) and METABRIC (Molecular Taxonomy of Breast Cancer International Consortium) cohorts were downloaded from the TCGA and cBioPortal databases, respectively. cBioPortal was used to identify mutant genes. DEG (differentially expressed gene) identification and survival analysis were performed with the GEPIA2 tool. ssGSEA (single-sample gene set enrichment analysis) was applied to estimate abundances of 28 immune cells for each sample. An unsupervised consensus clustering algorithm was used to identify ISs (immune subtypes). A graph learning-based dimensionality reduction analysis algorithm was utilized to construct an immune landscape. WGCNA (weighted correlation network analysis) was performed to identify immune gene modules.ResultsFour potential tumor antigens, i.e., SLC7A5, CHPF, CCNE1, and CENPW, associated with poor prognosis and APCs (antigen-presenting cells) among overexpressed and mutated genes were identified in BRCA. Two ISs (IS1-2) characterized by distinct clinical, immune cell infiltration, and molecular features were observed in both the TCGA-BRCA and METABRIC cohorts. BRCA patients with IS2 tumors related to poor prognosis had an immune "hot" phenotype, while those patients with IS1 tumors related to superior prognosis had an immune "cold" phenotype. Distinct IS tumors were observed in different ICD (immunogenic cell death modulator) and ICP (immune checkpoint) expression profiles. The immune landscape showed an immune distribution in BRCA patients. Additionally, we identified 2 immune gene modules with different biological functions.ConclusionsSLC7A5, CHPF, CCNE1, and CENPW are the potential tumor antigens for mRNA vaccine development with BRCA. Patients with IS2 tumors are a suitable population for mRNA vaccination. This study provides a new insight into mRNA vaccine development, population selection for vaccination, and prognosis prediction.

Machine learning- and WGCNA-mediated double analysis based on genes associated with disulfidptosis, cuproptosis and ferroptosis for the construction and validation of the prognostic model for breast cancer.

Disulfidptosis, a recently discovered cellular death mechanism, has not been extensively studied in relation to breast cancer (BC). Specifically, no previous research has integrated disulfidptosis-related genes (DRGs), cuproptosis-related genes (CRGs), and ferroptosis-related genes (FRGs) to construct a prognostic signature for BC. DRGs, CRGs and FRGs with prognostic potential were identified through Cox regression analysis. A predictive model was constructed by intersecting the core genes obtained from unsupervised cluster analysis and weighted correlation network analysis (WGCNA). Differences in chemotherapy drug sensitivity, immune checkpoint levels were analyzed according to different risk score groups. The expression of the core disulfidptosis gene, SLC7A11, was analyzed using immunofluorescence. Single-cell RNA sequencing analysis revealed differential expression of DRGs in the BC tumor microenvironment. We developed a prognostic model, consisting of six genes, based on machine learning which included unsupervised cluster analysis and Lasso-Cox analysis. An internal training set and a validation set, both derived from the Cancer Genome Atlas-Breast Cancer (TCGA-BRCA) database, GSE20685 and GSE42568 as external validation sets all verified the model's validity. The low-risk group exhibited increased sensitivity to paclitaxel. Additionally, the high-risk group demonstrated significantly higher expression of tumor mutation burden and microsatellite instability compared to the low-risk group. A nomogram confirmed that the risk score can be an independent risk factor for BC. Notably, our findings highlighted the impact of SLC7A11 on the BC tumor microenvironment. Immunofluorescence analysis revealed significantly higher expression of SLC7A11 in BC tissues compared to paracancerous tissues. Multiplex analysis based on DRGs, CRGs and FRGs correlated strongly with BC, providing new insights for developing clinical prognostic tools and designing immunotherapy regimens for BC patients.

Clinical implications and immune features of CENPN in breast cancer

BackgroundA number of human diseases have been associated with Centromere protein N (CENPN), but its role in breast cancer is unclear.MethodsA pan-cancer database of Genotype Tissue Expression (GTEx) and the Cancer Genome Atlas (TCGA) were used to examine the expression of CENPN. Using TCGA clinical survival data and breast cancer specimens from our center for validation, the relationship between CENPN expression, breast cancer prognosis, and clinicopathological characteristics of patients was examined. Bioinformatics was utilized to conduct an enrichment study of CENPN. Additionally, the potential of CENPN as a predictive biomarker for immunotherapy success was confirmed by analyzing the co-expression of CENPN with immune-checkpoint related genes, reviewing the TCGA database, and evaluating the correlation between CENPN expression and immune cell infiltration. Using the CCK8 test and colony formation assay, CENPN was evaluated for its ability to inhibit breast cancer cell proliferation. Transwell assays and scratch tests were used to assess the impact of CENPN on breast cancer cell migration.ResultsCENPN is found in a wide range of tumors, including breast cancer. Additional investigation revealed that CENPN was co-expressed with the majority of immune checkpoint-related genes, had the potential to serve as a predictive biomarker for immunotherapy effectiveness, and that high CENPN expression was linked to high Tregs and low CD8 + T cells and NK cells. Breast cancer cells' malignant characteristics, such as migration and cell proliferation, were inhibited by CENPN knockdown.ConclusionsAccording to our findings, CENPN may be an oncogene in breast cancer, as well as a new therapeutic target for immune checkpoint inhibitors.

THE SCALABLE BIRTH-DEATH MCMC ALGORITHM FOR MIXED GRAPHICAL MODEL LEARNING WITH APPLICATION TO GENOMIC DATA INTEGRATION.

Recent advances in biological research have seen the emergence of high-throughput technologies with numerous applications that allow the study of biological mechanisms at an unprecedented depth and scale. A large amount of genomic data is now distributed through consortia like The Cancer Genome Atlas (TCGA), where specific types of biological information on specific type of tissue or cell are available. In cancer research, the challenge is now to perform integrative analyses of high-dimensional multi-omic data with the goal to better understand genomic processes that correlate with cancer outcomes, e.g. elucidate gene networks that discriminate a specific cancer subgroups (cancer sub-typing) or discovering gene networks that overlap across different cancer types (pan-cancer studies). In this paper, we propose a novel mixed graphical model approach to analyze multi-omic data of different types (continuous, discrete and count) and perform model selection by extending the Birth-Death MCMC (BDMCMC) algorithm initially proposed by Stephens (2000) and later developed by Mohammadi and Wit (2015). We compare the performance of our method to the LASSO method and the standard BDMCMC method using simulations and find that our method is superior in terms of both computational efficiency and the accuracy of the model selection results. Finally, an application to the TCGA breast cancer data shows that integrating genomic information at different levels (mutation and expression data) leads to better subtyping of breast cancers.

Abstract 5696: A prognostic machine learning model for early breast cancer which combines clinical and genetic data in patients treated with neo/adjuvant chemotherapy

Abstract Background: Accurate modelling of the impact of patient-specific features and cancer treatments on survival allows the assignment of targeted therapy. There has not been any effort to build a multi-source model for the survival analysis of breast cancer. We show in this study a prognostic model, which integrates genetic (DNA), clinical and therapy inputs to predict survival for early breast cancer (stages 1-3) for all breast cancer subtypes. Methods: We used a data-driven Random Survival Forest approach, a statistical non-parametric ensemble learning method, that incorporates censor and time-to-event data. The learning is performed by creating numerous decision trees and selecting the model based on the correct responses in unseen data. We used The Cancer Genome Atlas Breast Cancer (TCGA) dataset and observed improvements in the accuracy when more sources of data were used, in line with the previous research. Integrating the impact of non-silent somatic tumor mutations (whole exome) and gene copy number variation (CNV) were analyzed on all mutations and per particular mutation. Results: Data from 1096 women with stage 1-3 early breast cancer were inputs to the model n=437 ER+ve HER2-ve, n=123 HER2+ve ER+ve, n=40 HER2+ ER-ve and n=126 TNBC. Pathological stage 1, n=183; stage 2, n=620; and stage 3, n=249. The following chemotherapy and hormonal treatments were used in the analysis: anthracycline, taxanes, platinum, alkylating and anti-metabolite agents, anti-oestrogen, aromatase inhibitors, ovarian suppression and HER2 antibody treatment. The model accuracy for predicting survival for early breast cancer using only clinical data was 0.78 for Area Under Curve (AUC) and c-index. The predictive accuracy improved stepwise by adding hormone, genetic and treatment data to AUC of 0.86 and c-index to 0.85. We observed the same trend if the proportion of test data increased from 0.25 to 0.75. Changes in median genes FGFR2 and CDKN2A copy number were strongly prognostic with p=0.0001 and p=0.002, and weaker signals for CBFB p=0.05, HRAS p=0.06, AKT p=0.07. Conclusion: Using public datasets, we developed a predictive survival model for an individual with early breast cancer up to 5 years from diagnosis using multi-source and patient-specific data. We show that using this approach for survival analysis yields good accuracy. Citation Format: Aidan (Amanzhol) Kubeyev, Andrea Giorni, Prabu Siva, Luiz Silva, Jordan Laurie, Matthew Foster, Matthew Griffiths, Uzma Asghar. A prognostic machine learning model for early breast cancer which combines clinical and genetic data in patients treated with neo/adjuvant chemotherapy. [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2023; Part 1 (Regular and Invited Abstracts); 2023 Apr 14-19; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2023;83(7_Suppl):Abstract nr 5696.

PHF6 recruits BPTF to promote HIF-dependent pathway and progression in YAP-high breast cancer

BackgroundAberrant epigenetic remodeling events contribute to progression and metastasis of breast cancer (Bca). The specific mechanims that epigenetic factors rely on to mediate tumor aggressiveness remain unclear. We aimed to elucidate the roles of epigenetic protein PHF6 in breast tumorigenesis.MethodsPublished datasets and tissue samples with PHF6 staining were used to investigate the clinical relevance of PHF6 in Bca. CCK-8, clony formation assays were used to assess cell growth capacity. Cell migration and invasion abilities were measured by Transwell assay. The gene mRNA and protein levels were measured by quantitative real-time PCR and western blot. Chromatin immunoprecipitation (ChIP)-qPCR assays were used to investigate transcriptional relationships among genes. The Co-immunoprecipitation (Co-IP) assay was used to validate interactions between proteins. The CRISPR/Cas9 editing technology was used to construct double HIF knockout (HIF-DKO) cells. The subcutaneous xenograft model and orthotopic implantation tumor model were used to asess in vivo tumor growth.ResultsIn this study, we utilized MTT assay to screen that PHF6 is required for Bca growth. PHF6 promotes Bca proliferation and migration. By analyzing The Cancer Genome Atlas breast cancer (TCGA-Bca) cohort, we found that PHF6 was significantly higher in tumor versus normal tissues. Mechanistically, PHF6 physically interacts with HIF-1α and HIF-2α to potentiate HIF-driven transcriptional events to initiate breast tumorigenesis. HIF-DKO abolished PHF6-mediated breast tumor growth, and PHF6 deficiency in turn impaired HIF transcriptional effects. Besides, hypoxia could also rely on YAP activation, but not HIF, to sustain PHF6 expressions in Bca cells. In addition, PHF6 recuits BPTF to mediate epigenetic remodeling to augment HIF transcriptional activity. Targeting PHF6 or BPTF inhibitor (AU1) is effective in mice models. Lastly, PHF6 correlated with HIF target gene expression in human breast tumors, which is an independent prognostic regulator.ConclusionsCollectively, this study identified PHF6 as a prognostic epigenetic regulator for Bca, functioning as a HIF coactivator. The fundamental mechanisms underlying YAP/PHF6/HIF axis in breast tumors endowed novel epigenegtic targets for Bca treatment.

Development and validation of a combined hypoxia and ferroptosis prognostic signature for breast cancer.

Hypoxia is involved in tumor biological processes and disease progression. Ferroptosis, as a newly discovered programmed cell death process, is closely related to breast cancer (BC) occurrence and development. However, reliable prognostic signatures based on a combination of hypoxia and ferroptosis in BC have not been developed. We set The Cancer Genome Atlas (TCGA) breast cancer cohort as training set and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) BC cohort as the validation set. Least Absolute Shrinkage and Selection Operator (LASSO) and COX regression approaches were used to construct ferroptosis-related genes (FRGs) and hypoxia-related genes (HRGs) prognostic signature (HFRS). The CIBERSORT algorithm and ESTIMATE score were used to explore the relationship between HFRS and tumor immune microenvironment. Immunohistochemical staining was used to detect protein expression in tissue samples. A nomogram was developed to advance the clinical application of HFRS signature. Ten ferroptosis-related genes and hypoxia-related genes were screened to construct the HFRS prognostic signature in TCGA BC cohort, and the predictive capacity was verified in METABRIC BC cohort. BC patients with high-HFRS had shorter survival time, higher tumor stage, and a higher rate of positive lymph node. Moreover, high HFRS was associated with high hypoxia, ferroptosis, and immunosuppression status. A nomogram that was constructed with age, stage, and HFRS signature showed a strong prognostic capability to predict overall survival (OS) for BC patients. We developed a novel prognostic model with hypoxia and ferroptosis-related genes to predict OS, and characterize the immune microenvironment of BC patients, which might provide new cures for clinical decision-making and individual treatment of BC patients.

Construction and validation of chemoresistance-associated tumor- infiltrating exhausted-like CD8+ T cell signature in breast cancer: cr-TILCD8TSig.

Accumulating evidence has revealed that CD8+ T cell exhaustion (Tex) results in worse immunotherapy outcomes. However, the molecular functions and mechanisms of action of Tex in chemoresistance needed to be elucidated. The populations of tumor-infiltrating CD8+ T cells (TILCD8Ts) in chemoresistant and chemosensitive groups of the GSE25066 dataset were calculated using CIBERSORT. Differentially expressed genes (DEGs) between TILCD8Ts and other immune cells were explored by integrating 16 immune cell datasets downloaded from the gene expression omnibus (GEO) database. Gene ontology (GO)/Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment, univariate and multivariate Cox regression, and least absolute shrinkage and selection operator (LASSO) regression of TILCD8T-specific upregulated genes were used to construct a chemoresistant TILCD8T signature (cr-TILCD8TSig). Clinical prognostic data, genomic alterations, chemotherapy response, and immunotherapy response were compared between the different cr-TILCD8TSig subgroups in the GSE25066 and the cancer genome atlas breast cancer (TCGA-BRCA) cohorts. A cr-TILCD8TSig with exhausted features was identified, consisting of seven genes (TCF7, RARRES3, ARL4C, ITK, CDH3, GZMB, and KLRD1), which were identified from 104 TILCD8Ts-specific DEGs. Our results showed that compared to the cr-TILCD8TSig-low subgroup, the -high subgroup had a poorer distant relapse-free survival (DRFS) in the GSE25066 cohort and worse progression-free survival (PFS) in the TCGA-BRCA cohort. Univariate and multivariate Cox regression analyses also demonstrated that cr-TILCD8TSig was an independent prognostic factor in the two independent cohorts. Furthermore, cr-TILCD8TSig-low patients benefited more from chemotherapy and immunotherapy than cr-TILCD8TSig-high patients. Besides, we found cell transmembrane signal transduction and the ECM may provide the molecular basis for resistance to antitumor agents in the cr-TILCD8Sig-high subgroup. For genomic alterations, we revealed that mutations in PIK3CA, DMD, and APOB were more common in the cr-TILCD8Sig-high subgroup than in the cr-TILCD8Sig-low subgroup. A nomogram was finally constructed with good discrimination and calibration. cr-TILCD8TSig is a useful tool to independently predict prognosis, chemotherapy response, and immunotherapy outcomes in patients with breast cancer.

Artificial intelligence-based digital scores of stromal tumour-infiltrating lymphocytes and tumour-associated stroma predict disease-specific survival in triple-negative breast cancer.

Triple-negative breast cancer (TNBC) is known to have a relatively poor outcome with variable prognoses, raising the need for more informative risk stratification. We investigated a set of digital, artificial intelligence (AI)-based spatial tumour microenvironment (sTME) features and explored their prognostic value in TNBC. After performing tissue classification on digitised haematoxylin and eosin (H&E) slides of TNBC cases, we employed a deep learning-based algorithm to segment tissue regions into tumour, stroma, and lymphocytes in order to compute quantitative features concerning the spatial relationship of tumour with lymphocytes and stroma. The prognostic value of the digital features was explored using survival analysis with Cox proportional hazard models in a cross-validation setting on two independent international multi-centric TNBC cohorts: The Australian Breast Cancer Tissue Bank (AUBC) cohort (n= 318) and The Cancer Genome Atlas Breast Cancer (TCGA) cohort (n= 111). The proposed digital stromal tumour-infiltrating lymphocytes (Digi-sTILs) score and the digital tumour-associated stroma (Digi-TAS) score were found to carry strong prognostic value for disease-specific survival, with the Digi-sTILs and Digi-TAS scores giving C-index values of 0.65 (p= 0.0189) and 0.60 (p= 0.0437), respectively, on the TCGA cohort as a validation set. Combining the Digi-sTILs feature with the patient's positivity status for axillary lymph nodes yielded a C-index of 0.76 on unseen validation cohorts. We surmise that the proposed digital features could potentially be used for better risk stratification and management of TNBC patients. © 2023 The Authors. The Journal of Pathology published by John Wiley & Sons Ltd on behalf of The Pathological Society of Great Britain and Ireland.

Chromatin profile-based identification of a novel ER-positive breast cancer subgroup with reduced ER-responsive element accessibility

BackgroundOestrogen receptor (ER) signalling-dependent cancer cell growth is one of the major features of ER-positive breast cancer (BC). Inhibition of ER function is a standard and effective treatment for ER-positive tumours; however, ~20% of patients with ER-positive BC experience early or late recurrence. In this study, we examined intertumour heterogeneity from an epigenetic perspective based on the hypothesis that the intrinsic difference in epigenetic states around ER signalling pathway underlies endocrine therapy resistance.MethodsWe performed transposase-accessible chromatin sequencing (ATAC-seq) analysis of 42 BC samples, including 35 ER-positive(+) human epidermal growth factor receptor 2 (HER2)-negative(−) and 7 triple-negative tumours. We also reanalysed ATAC-seq data of 45 ER + /HER2 − tumours in the Cancer Genome Atlas (TCGA) BC cohort to validate our observations.ResultsWe conducted a comprehensive analysis of cis-regulatory elements (CREs) using ATAC-seq, identifying three subgroups based on chromatin accessibility profiles. We identified a subgroup of ER-positive BCs with a distinctive chromatin accessibility pattern including reduced accessibility to ER-responsive elements (EREs). The same subgroup was also observed in TCGA BC cohort. Despite the reduced accessibility to EREs, the expression of ER and potential ER target genes were not decreased in these tumours.ConclusionOur findings highlight the existence of a subset of ER-positive BCs with unchanged ER expression but reduced EREs accessibility that cannot be distinguished by conventional immunostaining for ER. Future studies should determine whether these tumours are associated with resistance to endocrine therapy.

Genetic determinants and absence of breast cancer in Xavante Indians in Sangradouro Reserve, Brazil

Genetic compositions of distinct human populations are different. How genomic variants influence many common and rare genetic diseases is always of great medical and anthropological interest, and understanding of genetic architectures of population groups in relation to diseases can advance our knowledge of medicine. Here, we have studied the genomic architecture of a group of Xavante Indians, an indigenous population in Brazil, and compared them with normal populations from the 1000 Genomes Projects. Principal component analysis (PCA) indicates that the Xavante Indians are genetically distinctive when compared to other ethnic groups. No incidence of breast cancer cases has ever been reported in the population, and polygenic risk analysis indicates extremely low breast cancer risk in this population when compared with germline TCGA (The Cancer Genome Atlas) breast cancer normal control samples. Low germinal mutation burden among this population is also observed. Our findings will help to deepen the understanding of breast cancer and might also provide new approaches to study the disease.