The online version of this article (https://doi.org/10.1186/s13058-018-0933-y) contains supplementary material, which is available to authorized users.
Approximately one in five breast cancers are driven by amplification and overexpression of the human epidermal growth factor receptor 2 (HER2) receptor kinase, and HER2-enriched (HER2E) is one of four major transcriptional subtypes of breast cancer. We set out to understand the genomics of HER2 amplification independent of subtype, and the underlying drivers and biology of HER2E tumors.
We investigated published genomic data from 3155 breast tumors and 5391 non-breast tumors.
HER2 amplification is a distinct driver event seen in all breast cancer subtypes, rather than a subtype marker, with major characteristics restricted to amplification and overexpression of HER2 and neighboring genes. The HER2E subtype has a distinctive transcriptional landscape independent of HER2A that reflects androgen receptor signaling as replacement for estrogen receptor (ER)-driven tumorigenesis. HER2 amplification is also an event in 1.8% of non-breast tumors.
These discoveries reveal therapeutic opportunities for combining anti-HER2 therapy with anti-androgen agents in breast cancer, and highlight the potential for broader therapeutic use of HER2 inhibitors.
Additional file 1: Flow chart summarizing sample size and available data for the TCGA, Metabric and USO1062 cohort; Clinical metadata, subtype information, HER2A status, AR-ness score, HER2 copy number, and expression levels of ERBB2, ER, PR and AR for the tumors in these three cohorts. (XLSX 669 kb)13058_2018_933_MOESM1_ESM.xlsx
Additional file 2: Robust development of the CN/ploidy definition, and comparison with two alternative HER2A measures. (PDF 115 kb)13058_2018_933_MOESM2_ESM.pdf
Additional file 3: Definition of HER2-amplified breast cancer. (A) HER2 overexpression in 864 TCGA tumors is defined as log2 (nRPKM + 1) ≥8.2 (normal mixture modeling). (B) Abundant HER2 protein in 367 TCGA tumors is defined as log2 (RPPA) ≥0.92. (C) Abundant phosphorylated HER2 (Tyr1248) is defined as log2 (RPPA) ≥0.605. (D) Concordance of three HER2A classification schemes with HER2 expression, HER2 protein, phosphorylated HER2, clinical HER2 status and PAM50 subtype in TCGA (Additional file 2). In each case, concordance to HER2 gene expression drops with the alternative measures, from 96.5% to 94.7% for total copies (McNemar test p = 0.002), and to 86.1% for centromere-corrected copies ( p = 2e-19). Concordance with HER2 protein levels drops from 94.1% to 92.8% ( p = 0.18) and 86.6% ( p = 7e-6), respectively. Concordance with pHER2 protein levels drops from 93.6% to 91.7% ( p = 0.05) and 85% ( p = 2e-7). Concordance with clinical HER2 status drops from 94.9% to 94.0% (p = 0.11) and 86.7% ( p = 2e-10). Ploidy-corrected HER2A captures a larger fraction of the PAM50 HER2E subtype (90.4% concordance for HER2E vs. other subtypes) than either total (89.0%, p = 0.025) or centromere-corrected HER2A status (80.0%, p = 2e-19). (E) HER2 overexpression in 1107 Metabric tumors is defined as log2 expression ratio ≥12.5. (F) Concordance of three HER2A classification schemes with HER2 expression, clinical HER2 status and PAM50 subtype in Metabric. The concordance to HER2 gene expression drops from 97.7% to 95.1% for total copies (McNemar test p = 1e-5), and to 90.9% for centromere-corrected copies ( p = 5e-17). Concordance with clinical HER2 status drops from 94.1% to 91.5% ( p = 3e-3) and 90.9% ( p = 2e-6), respectively. Overlap with PAM50 HER2E drops from 88.6% to 86.8% ( p = 2e-3) and 82.3% ( p = 1e-14). (G) HER2 overexpression in 987 USO1062 tumors is defined as normalized log10 counts ≥1.02. (H) Concordance of HER2A status with HER2 overexpression, clinical HER2 status and PAM50 subtype in the USO1062 cohort. (PDF 147 kb)
Additional file 4: HER2 amplification is a discrete event on top of a luminal or basal state, with minor consistent correlation with gene expression. (A) Row-scaled expression of 50 PAM50 genes in 864 TCGA breast tumors, labeled on top by HER2A status and PAM50 subtype. (B) PAM50 scores for TCGA breast tumors categorized by PAM50 subtype and HER2A status. HER2A tumors, in red, are confidently classified as luminal A, luminal B, basal-like, or HER2E, with PAM50 scores within 3.3–4.3% of the PAM50 scores of non-HER2A tumors of the same subtype. (C) Shown are the odds that a genomic alteration in gene A will occur in an HER2A tumor of subtype X compared to the odds of gene A being altered in a non-HER2A tumor of subtype X. Each dot represents the enrichment of alterations in a gene in HER2A compared to non-HER2A tumors of a particular subtype, colored by PAM50, with mutations shown as diamond and copy number alterations as circle. Fisher’s exact p values were corrected for multiple testing per PAM50 subtype, and separately for the set of 21 genes for which we assessed mutation status (Additional file 5A-B) and 28 genes for which we assessed copy number alterations (Additional file 5C-D). Significance is defined as FDR p value <0.1. Enrichments are based on the combined TCGA and Metabric cohorts. (D) Gene-gene expression correlation in TCGA breast tumors for 43 genes differentially expressed between HER2A and non-HER2A tumors when accounting for PAM50 subtype and chromosomal instability. Two sets of 3 or more highly correlated genes (gene-gene correlation >0.6) are highlighted on the right: 28 genes on 17q12-21 near HER2, and 3 SCGB genes at 11q13 (Additional file 6A). (E) Two genes outside of 17q12-21 are more highly expressed in HER2A than non-HER2A TCGA tumors with FC >4: SCGB2A2 (FC 7.7) and SCGB1D2 (FC 4.4). (PDF 390 kb)
Additional file 5: Mutation and copy number profiles of breast tumors. (A) Mutation profile for 23 genes previously shown to be significantly mutated in breast cancer [ 13] in 679 TCGA breast tumors with exome-seq data. (B) Mutation profile for 21 genes (FOXA1 and MLL3 are not available) in 1039 Metabric tumors with targeted sequence data. (C) Copy number profile for 28 genes previously shown to have subtype-specific copy number alterations [ 13], in 864 TCGA breast tumors. (D) Copy number profile for 28 genes in 1107 Metabric tumors. PAM50 subtype and HER2A status are provided. (XLSX 417 kb)
Additional file 6: Protein-coding genes associated with HER2 amplification. (A) List of 43 protein-coding genes with subtype-independent expression differences between HER2A and non-HER2A breast tumors, after accounting for PAM50 subtype and differences in chromosomal instability, in the TCGA and Metabric cohort. These genes fulfilled the following criteria in at least one cohort: adjusted p value <0.05, and fold change >2 (linear model). (B) List of 14 protein-coding genes with cancer-independent expression differences between HER2A and non-HER2A tumors in a panel of 2838 non-breast TCGA tumors, after accounting for cancer and chromosomal instability. These genes fulfilled the following criteria: adjusted p value <0.001, and fold change >2. (XLSX 24 kb)13058_2018_933_MOESM6_ESM.xlsx
Additional file 7: Characterization of HER2E tumors and AR-ness signature. (A) Overview of genes differentially expressed between HER2E and non-HER2E TCGA tumors, accounting for ER expression, PR expression, and PAM50 proliferation score, and omitting genes on chromosome 17. (B) Gene set enrichment results for the C2 gene set collection from MSigDB [ 22] (KEGG gene sets were removed), for the comparison of HER2E versus non-HER2E TCGA tumors. (C) 45-gene AR-ness signature. (XLSX 6449 kb)
Additional file 8: Identification of ER- tumors that are AR-driven. (A) Top 10 C2 gene sets from MSigDB [ 22] enriched among genes differentially expressed between HER2E and other breast tumors. See Additional file 7B for the full list of C2 gene sets. (B) Overlap in number of genes between the top 10 C2 gene sets and androgen responsive gene sets (from Additional file 7B). (C) AR-ness score across PAM50 subtypes for 205 breast tumors from Metabric that are ER- by IHC. AR-ness score is calculated as the difference in average z-scored expression of 14 positive signature genes and average z-scored expression of 31 negative signature genes. Tumors are colored by HER2A status. (D) AR protein levels in 90 ER- TCGA tumors by AR-ness score. (E) Kaplan-Meier curve of overall survival (OS) in 977 Metabric tumors with a median follow up of 7.3 years, divided into 5 groups based on ER IHC status, PAM50 subtype, and AR activity (positive vs. negative AR-ness score). OS was truncated to 17 years of follow up. (F) Left, Kaplan-Meier curve of OS in 57 ER- HER2A tumors from Metabric, divided by AR-ness score. Right, Kaplan-Meier curve of OS in 121 ER- non-HER2A tumors from Metabric, divided by AR-ness score. OS was truncated to 17 years of follow up. (PDF 167 kb)
Additional file 9: Characterization of the HER2 amplicon and HER2 co-amplification in breast cancer. (A) Chromosomal instability, shown as the number of breaks per kb per autosome, is higher in HER2A than non-HER2A TCGA tumors, with one-sided t test p values per chromosome ranging from 0.37 (chr 5) to 5e-26 (chr 17), and <0.05 for 17/22 autosomes. (B) Copy number levels in 106 HER2A TCGA breast tumors, for genes on chromosome 17 from 35 Mb to 40 Mb (ordered by genomic location). The core HER2 amplicon on top is shown in green, the broad HER2 amplicon in yellow, and genes outside of the broad HER2 amplicon on 17q in magenta. Shown in green on top is correlation between copy number and expression (log2 nRPKM + 1) for each gene across the 864 tumors. (C) HER2 amplicon profile in 133 HER2A Metabric tumors. See Fig. 2a legend for details. (D) Four HER2A TCGA breast tumors show 17q arm-level amplification without additional HER2 focality. Shown are total copy number levels for genes on 17q from 35 Mb to 40 Mb (colored as per panel (B)). (E) Fisher’s exact FDR p values for co-amplification of genes with HER2 in 1971 tumors from TCGA and Metabric cohorts. Amplification was defined as 4 or more ploidy-corrected copies. Chromosomes are colored alternatingly in black and gold. (F, G) We detected two regions with germline micro-deletions or micro-gains in normal breast tissue from TCGA: 34.4–34.6 Mb and 44.1–44.8 Mb. These regions are defined as loci with copy number levels either >2.4 or <1.6 in at least 5% of normal breast samples, and were removed from Fig. 2b for visual purposes. Shown are copy number levels in 765 matched normal breast samples, for all genes on chromosome 17, before (F) and after (G) removal of those two regions. (PDF 969 kb)
Additional file 10: HER2 co-amplification events: 14 regions with 931 genes are significantly co-amplified with HER2 in TCGA and Metabric combined cohorts (Fisher’s exact FDR p value <0.05). Percentage of amplification in the HER2A vs. non-HER2A tumors, Fisher’s exact FDR p value, and gene annotation are shown. Genes in the core or broad HER2 amplicon are listed, and known cancer genes as reported previously [ 44, 45] are highlighted. (XLSX 104 kb)
Additional file 11: HER2 amplification and mutation in other cancers. (A-C) For each cancer, tumors are grouped by HER2 amplification and mutation status. Tumors without HER2 amplification or mutation are shown in black, HER2A tumors (regardless of HER2 mutation) in red, unamplified tumors with an activating HER2 mutation (HER2MUT act) in green, and unamplified tumors with an untested or functionally inactive HER2 mutation (HER2MUT unk) in gold. (A) Phospho-ERBB3 (Tyr1289) levels are significantly higher in HER2A breast (p = 1e-7), endometrium (p = 0.003), and lung squamous cell carcinoma (p = 0.009) compared to unaltered tumors. (B) Phospho-AKT (pan-AKT Ser473) levels do not differ by HER2 status (p = 0.41, 0.13, and 0.45 for HER2A tumors, tumors with an activating HER2 mutation, or a non-functional HER2 mutation, respectively, in comparison to unaltered tumors). (C) Phospho-p44/42 MAPK (ERK1/2) (Thr202/Tyr204) levels do not differ by HER2 status (p = 0.26, 0.99, and 0.30, respectively, as shown in (B)). (D) Prevalence of HER2 mutations varies from 0.4 to 8.6%. (PDF 214 kb)13058_2018_933_MOESM11_ESM.pdf
Additional file 12: HER2A tumors in other cancers share certain similarities with breast cancer. (A) Copy number of genes on chromosome 17 are shown in 98 HER2A non-breast tumors. Three distinct groups of HER2A tumors are labeled on the right: tumors with 17q arm-level amplification, defined as 5 or more copies for at least 80% of genes (cyan); tumors with 17q gain (copy number between 2.5 and 5 for 80% or more genes) (magenta); and tumors that are mainly 17q diploid with copy number <2.5 for the majority of 17q genes (green)", with "as per Fig. 2b". Chromosome 17 annotation is indicated on top. Regions 34.4–34.6 Mb and 44.1–44.8 Mb with germline micro-deletions or micro-gains were removed for visual purposes (see Additional file 9F-G). (B) HER2 amplification is a discrete event across multiple molecular subtypes, for cancers with well-established subtypes. HER2-amplified gastric tumors are either chromosomal instable or EBV-positive. (C) Shown on the left axis are the percentage of 98 HER2A non-breast tumors with gene amplification (indicated in solid red) and the percentage of 106 HER2A breast tumors with gene amplification (dashed red), for genes on chromosome 17 from 35 Mb to 40 Mb (ordered by genomic location). Shown on the right axis are the average copy number level in HER2A non-breast tumors (indicated in solid blue) and in HER2A breast tumors (dashed blue) with gene amplification. Shown at the bottom are core pan-cancer HER2 amplicon (6) and broad amplicon (19) genes. (D) 17q12-21 genes that are significantly associated with HER2A in breast cancer (Additional file 4D) show consistent increased expression in HER2A tumors, regardless of cancer ( p = 1e-235). (E) The three SCGB genes on 11q13 trend towards increased expression in cervix and ovarian cancer, but lack consistent pan-cancer association with HER2A ( p = 0.72). (D, E) Fold change values were calculated as the difference between log2 average expression, with significance defined as t test p < 0.05. P values were obtained from multivariate linear models, predicting gene set expression by HER2A and cancer in 2838 non-breast tumors. (PDF 271 kb)
Additional file 13: Coordinated expression of HER2-neighboring genes in the absence of amplification. (A-H) Expression of genes in the core HER2 amplicon (PGAP3, ERBB2, MIEN1, GRB7), representative genes in the broad HER2 amplicon (MED1, CDK12, NR1D1), and TOP2A (more telomeric on 17q), in HER2A tumors (red), non-HER2A tumors without HER2 overexpression (black), and non-HER2A tumors with HER2 overexpression (o/e, log2 nRPKM + 1 ≥ 8.2; green). (A) Seven non-HER2A, o/e breast tumors. (B) Six non-HER2A, o/e gastric tumors. (C) Two non-HER2A, o/e endometrial tumors. (D) Two non-HER2A, o/e cervix tumors. (E) Two non-HER2A, o/e bladder tumors. (F) Two non-HER2A, o/e lung squamous cell carcinoma tumors. (G) One non-HER2A, o/e lung adenocarcinoma tumor. (H) One non-HER2A, o/e ovarian tumor. (I) Relative copy number levels for broad HER2 amplicon genes in 23 non-HER2A o/e tumors. Relative copy number levels exceed 2 (or are borderline at 1.9) in 6 out of 23 tumors. Tumors are colored by cancer, and genes in the broad HER2 amplicon are colored as per Fig. 2a. (J-M) Average log2 ratio of methylated to unmethylated intensity of CpG probes near HER2 and its closest neighbors PGAP3, MIEN1 and GRB7 (in the gene body or maximum 2 kb upstream of the transcription start site) with Kruskal-Wallis test FDR p value below the indicated value for a four-group comparison: HER2A, non-o/e; HER2A, o/e; non-HER2A, non-o/e; non-HER2A, o/e. (J) For breast cancer, included are 27 methylation probes with Kruskal-Wallis FDR p < 1e-15. (K) For gastric cancer, included are 37 methylation probes with p < 1e-5. (L) For cervix cancer, included are 28 methylation probes with p < 0.01. (M) For bladder cancer, included are 21 methylation probes with p < 0.01. (PDF 571 kb)
Parker JS, Mullins M, Cheang MCU, Leung S, Voduc D, Vickery T, et al. Supervised risk predictor of breast cancer based on intrinsic subtypes. J Clin Oncol Off J Am Soc Clin Oncol. 2009;27:1160–7. CrossRef
Carey LA, Berry DA, Cirrincione CT, Barry WT, Pitcher BN, Harris LN, et al. Molecular heterogeneity and response to neoadjuvant human epidermal growth factor receptor 2 targeting in CALGB 40601, a randomized phase III trial of paclitaxel plus trastuzumab with or without lapatinib. J Clin Oncol Off J Am Soc Clin Oncol. 2016;34:542–9. CrossRef
Llombart-Cussac A, Cortés J, Paré L, Galván P, Bermejo B, Martínez N, et al. HER2-enriched subtype as a predictor of pathological complete response following trastuzumab and lapatinib without chemotherapy in early-stage HER2-positive breast cancer (PAMELA): an open-label, single-group, multicentre, phase 2 trial. Lancet Oncol. 2017;18:545–54. CrossRefPubMed
Fumagalli D, Venet D, Ignatiadis M, Azim HA, Maetens M, Rothé F, et al. RNA sequencing to predict response to neoadjuvant anti-HER2 Therapy: a secondary analysis of the NeoALTTO randomized clinical trial. JAMA Oncol. 2017;3:227–34. CrossRef
Perez EA, Ballman KV, Mashadi-Hossein A, Tenner KS, Kachergus JM, Norton N, et al. Intrinsic subtype and therapeutic response among HER2-positive breast tumors from the NCCTG (Alliance) N9831 trial. J Natl Cancer Inst. 2017;109. https://doi.org/10.1093/jnci/djw207.
Prat A, Carey LA, Adamo B, Vidal M, Tabernero J, Cortés J, et al. Molecular features and survival outcomes of the intrinsic subtypes within HER2-positive breast cancer. J Natl Cancer Inst. 2014;106. https://doi.org/10.1093/jnci/dju152.
Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Ser B Methodol. 1995;57:289–300.
The Cancer Genome Atlas Research Network. Comprehensive molecular characterization of gastric adenocarcinoma. Nature. 2014;513:202–9. CrossRef
Zafrakas M, Petschke B, Donner A, Fritzsche F, Kristiansen G, Knüchel R, et al. Expression analysis of mammaglobin A (SCGB2A2) and lipophilin B (SCGB1D2) in more than 300 human tumors and matching normal tissues reveals their co-expression in gynecologic malignancies. BMC Cancer. 2006;6:88. CrossRefPubMedPubMedCentral
Rahim B, O’Regan R. AR Signaling in breast cancer. Cancers. 2017;9:21.
Zhao H-B, Zhang X-F, Wang H-B, Zhang M-Z. Migration and invasion enhancer 1 (MIEN1) is overexpressed in breast cancer and is a potential new therapeutic molecular target. Genet Mol Res GMR. 2017;16:gmr16019380.
Ruggieri V, Agriesti F, Tataranni T, Perris R, Mangieri D. Paving the path for invasion: the polyedric role of LASP1 in cancer. Tumor Biol. 2017;39:1010428317705757. CrossRef
Bang Y-J, Van Cutsem E, Feyereislova A, Chung HC, Shen L, Sawaki A, et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet Lond Engl. 2010;376:687–97. CrossRef
American Cancer Society. Cancer facts & figures 2015. Atlanta: American Cancer Society; 2015.
Cortes J, Crown J, Awada A, Schmid P, Gianni L, Garcia-Estevez L, et al. Overall survival (OS) from the phase 2 study of enzalutamide (ENZA), an androgen receptor (AR) signaling inhibitor, in AR+ advanced triple-negative breast cancer (aTNBC) [abstract]. EJC. 2015;51(Suppl 3):S265.
Burris HA, Hurwitz H, Perez EA, Spigel D, Swanton C, Hainsworth JD, et al. MyPathway: An open-label phase IIa study of trastuzumab/pertuzumab, erlotinib, vemurafenib, and vismodegib in patients who have advanced solid tumors with mutations or gene expression abnormalities targeted by these agents. J Clin Oncol. 2015;33 suppl; abstr TPS11111. http://meetinglibrary.asco.org/content/150017-156. Accessed 27 Sept 2016.
- HER2 is not a cancer subtype but rather a pan-cancer event and is highly enriched in AR-driven breast tumors
- BioMed Central
Neu im Fachgebiet Onkologie
Mail Icon II