Abstract
NF1 germline mutation predisposes to breast cancer. NF1 mutations have also been proposed as oncogenic drivers in sporadic breast cancers. To understand the genomic and histologic characteristics of these breast cancers, we analyzed the tumors with NF1 germline mutations and also examined the genomic and proteomic profiles of unselected tumors. Among 14 breast cancer specimens from 13 women affected with neurofibromatosis type 1 (NF1), 9 samples (NF + BrCa) underwent genomic copy number (CN) and targeted sequencing analysis. Mutations of NF1 were identified in two samples and TP53 were in three. No mutation was detected in ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CHEK2, NBN, PALB2, PTEN, RAD50, and STK11. HER2 (ErbB2) overexpression was detected by IHC in 69.2% (9/13) of the tumors. CN gain/amplification of ERBB2 was detected in 4 of 9 with DNA analysis. By evaluating HER2 expression and NF1 alterations in unselected invasive breast cancers in TCGA datasets, we discovered that among samples with ERBB2 CN gain/amplification, the HER2 mRNA and protein expression were much more pronounced in NF1-mutated/deleted samples in comparison with NF1-unaltered samples. This finding suggests a synergistic interplay between these two genes, potentially driving the development of breast cancer harboring NF1 mutation and ERBB2 CN gain/amplification. NF1 gene loss of heterozygosity was observed in 4 of 9 NF + BrCa samples. CDK4 appeared to have more CN gain in NF + BrCa and exhibited increased mRNA expression in TCGA NF1--altered samples. Cancer Prev Res; 11(10); 655–64. ©2018 AACR.
Introduction
Neurofibromatosis type 1 (NF1) is an autosomal dominant hereditary condition caused by germline mutation or genomic deletion of the NF1 gene. It is rare with a population birth incidence of 1:2–3,000. NF1 is known to be a hereditary neoplasm predisposition syndrome with characteristic features including multiple dermal neurofibromas, café-au-lait macules on the skin, freckles on the intertriginous area, and internal and external large plexiform neurofibromas. Occurrence of several types of benign and malignant neoplasms is increased in these individuals. Women affected with NF1 have moderately increased risk of breast cancer (1–8). Breast cancer lifetime risk for these women is approximately 18%. One study also suggested worsened breast cancer survival in NF1 women compared with age-matched controls (4).
Meanwhile, somatic NF1 deep deletion and putative driver mutations are found in approximately 2% to 3% of the sporadic breast cancers [The Cancer Genome Atlas Network (TCGA) 2012, TCGA data bank; available from http://www.cbioportal.org; ref. 9; Catalogue of Somatic Mutations in Cancer (COSMIC) database available from http://cancer.sanger.ac.uk/cancergenome/projects/cosmic/; ref. 10]. The association between NF1 deletion and mammary adenocarcinoma was also demonstrated in Chaos3 mice (11). However, it is not known whether NF1 biallelic mutation/deletion is common in breast cancers occurring in individuals with NF1. A wide range of mechanisms and pathways are known to contribute to mammary oncogenesis and tumor development. Somatic mutations in hereditary breast cancer predisposition genes, such as BRCA1, are only found in a small fraction of sporadic breast cancers despite their prominent role in highly penetrant hereditary predisposition to breast cancer. A germline mutation is the first mutation in tumor cells, whereas a somatic mutation develops later and exists in a specific population of cells. Comparing the tumors harboring certain mutations in germline versus somatic (postzygotic) lines may reveal the order of oncogenic events and the dynamics of molecular biological changes. Analyzing the breast cancer specimens with a constitutionally defective NF1 (i.e., a mutation in germline) as well as sporadic breast cancers harboring an altered NF1 gene may offer insight into the roles of the NF1 gene and its encoded protein, neurofibromin, in the landscape of breast cancers.
Neurofibromin is a negative regulator for Ras action by converting GTP-Ras to GDP-Ras in cytoplasm. Two of the downstream signaling pathways activated by GTP-Ras are the PI3K/Akt and Raf/MEK/ERK pathways. Our report characterizes NF1 breast cancer by histology, IHC stain of key proteins in signaling pathways, somatic genomic analysis of tumors with germline NF1 mutations, as well as TCGA genomic and expression profiles in sporadic tumors with somatic NF1 alterations.
Materials and Methods
Recruitment
Women with NF1 and a history of breast cancer were recruited by three Children's Tumor Foundation (CTF)-affiliated neurofibromatosis clinics at Henry Ford Health System, Johns Hopkins University (Baltimore, MD), and Children's National Medical Center (Washington, DC). Additional recruitment advertisements were distributed by CTF newsletters as well as among NF-patient support groups. Participants' archived, formalin-fixed paraffin-embedded (FFPE) breast cancer tumor specimens were collected with informed consent.
IHC assay for archived breast tumor samples from women with NF1 (NF + BrCa)
Hematoxylin and eosin–stained slides were reviewed to select tumor-containing paraffin block with more than 90% tumor. Cases with tumor percentage less than 50% were marked for microdissection to enrich tumor DNA. A tissue microarray was constructed using standard IHC protocol.
All antibodies were obtained from Cell Signaling Technology. Validation process is described in Supplementary S1. IHC staining for the following proteins was performed on the basis of the manufacturer's protocol: pMEK (i.e., phospho-MEK1/2, Ser221, the phosphorylated and activated form of MEK), pErk [i.e., phospo-p44/42 MAPK (Erk1/2; Thr202/Tyr204), the phosphorylated and activated ERK], Erk (i.e., P44/42 MAPK, Erk1/2, the nonphosphorylated ERK), AKT (pan), mTOR, p53, PTEN, and HER2. Staining intensity was graded in 3 tiers, from low to high as 1, 2, and 3. Staining score is the product of the intensity (1, 2, and 3) and the percentage of the cells stained.
Specifically for HER2, immunostained slides were reviewed by light microscopy and the membrane staining intensity pattern and percentage of immunoreactive cells were semiquantitatively assessed and scored as per the ASCO-CAP Guidelines established in 2013 (12). Results were scored as negative (either 0 or 1+ immunostaining), positive (3+ immunostaining), or equivocal (2+ immunostaining).
DNA extraction
Manual DNA extraction protocol (QIAamp DNA Mini Kit, Qiagen) was used for isolation of genomic DNA from the archived breast tumor specimens. DNA yield and purity were evaluated by NanoDrop and TapeStation.
Affymetrix OncoScan FFPE assay
The genomic DNA samples extracted from NF + BrCa specimens were subjected to Affymetrix OncoScan FFPE Assay (Affymetrix) to investigate copy number (CN) variations. Allelic status of individual genes, such as CN gain, loss, or loss of heterozygosity was assigned by Chromosome Analysis Suite (ChAS) Software 3.1 (Affymetrix). Gain is determined by OncoScan as CN > 2; loss is CN < 2; amplification is CN ≥ 4. At the single gene level, we obtained the CN of 77 genes known to be amplified in various cancer types (9, 13).
Targeted gene mutation analysis by OneSeq for NF + BrCa samples
The genomic DNA extracted from NF + BrCa specimens was also subjected to tests using OneSeq Constitutional Research Panel developed by Agilent Technologies. Genes known to be associated with hereditary diseases were sequenced using the Agilent SureSelect Focused Exome Panel and analyzed with Agilent SureCall software. The Agilent SureSelect Focused Exome Panel Catalog Kit includes baits for a combination of a CNV backbone and all content from the Focused Exome Panel, targeting hereditary disease-associated genes and regions previously annotated within the HGMD, OMIM, and ClinVar databases. Genes commonly recognized as cancer drivers but not known to be associated with hereditary diseases were not included in this Agilent panel. After the initial variant call, variants were further categorized and curated for pathogenicity based on the following categories: missense, frameshift, nonsense, splice-site altering, variant flagged as “Pathogenic” in dbSNP database, or reported as “hotspot mutation” in cancer somatic genomic database.
Genomic, mRNA, and proteomic dataset analysis on TCGA unselected invasive breast cancers
To examine the effects of somatic NF1 mutations and deletions, the clinical information as well as somatic mutation, CN, and RNA sequencing V2 RSEM data from TCGA breast invasive carcinoma dataset were downloaded from the cBioPortal (9) and combined with reverse-phase protein array data from The Cancer Proteome Atlas (TCPA; ref. 14). The data were classified by the status of the NF1 gene: tumors with NF1 mutation or deletion (i.e., NF1 altered) were compared with tumors without NF1 mutation or deletion (labeled as NF1 nonaltered). The groups were compared using the Mann–Whitney U test.
Human research subjects' protection
These studies were conducted in accordance with the U.S. Common Rule and performed after approval by the Institutional Review Board of each participating institution. The investigators obtained informed written consent from the subjects (wherever necessary).
Results
IHC assay
Fourteen archived breast cancer tumor specimens were collected from 13 women with NF1. One sample was suspected of poor sample processing and was excluded. Among the remaining 13 specimens from 12 women, 9 (69.2%) stained strongly positive for HER2 (ErbB2; Table 1). The sample histology, hormone receptor status, and IHC staining results for all proteins, that is, pMEK (i.e., phospho-MEK1/2, Ser221, the phosphorylated and activated form of MEK), pErk [i.e., phospo-p44/42 MAPK (Erk1/2; Thr202/Tyr204), the phosphorylated and activated ERK], Erk (i.e., P44/42 MAPK, Erk1/2, the nonphosphorylated ERK), AKT (pan), mTOR, p53, PTEN, and HER2, are provided in Table 1.
Histology, hormone receptors, HER2 status, and IHC for other proteins
. | . | . | . | Phospho-MEK1 . | Phospho-MAPK (Erk 1/2) P44/42 . | MAPK (Erk 1/2) . | PTEN . | AKT . | mTOR . | p53 . | HER2/ErbB2 . | |||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ID . | Histology . | ER . | PR . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Score . | Interpretation . |
1 | IDC + DCIS | POS | NEG | 1 | 50 | 50 | 3 | 5 | 15 | 1 | 95 | 95 | 1 | 100 | 100 | 1 | 80 | 80 | 1 | 100 | 100 | 3 | 95 | 285 | 3 | POS |
3 | IDC | NA | NA | 1 | 50 | 50 | 0 | 0 | 0 | 1 | 70 | 70 | NA | NA | 1 | 100 | 100 | 1 | 100 | 100 | 2 | 85 | 170 | 3 | POS | |
6 | DCIS | NA | NA | 1 | 100 | 100 | 3 | 3 | 9 | 2 | 100 | 200 | 1 | 100 | 100 | 2 | 95 | 190 | 2 | 100 | 200 | 1 | 30 | 30 | 2–3 | POS |
8 | IDC + DCIS | POS | NEG | 1 | 100 | 100 | 3 | 2 | 6 | 1 | 100 | 100 | 1 | 100 | 100 | 2 | 100 | 200 | 1 | 100 | 100 | 3 | 95 | 285 | 3 | POS |
9 | IDC + DCIS | POS | POS | 1 | 100 | 100 | 0 | 0 | 0 | 2 | 90 | 180 | 1 | 100 | 100 | 1 | 80 | 80 | 1 | 100 | 100 | 1 | 5 | 5 | 1 | NEG |
2 | IDC + DCIS | POS | POS | 1 | 100 | 100 | 3 | 1 | 3 | 1 | 95 | 95 | 1 | 100 | 100 | 1 | 100 | 100 | 1 | 90 | 90 | 1 | 5 | 5 | 1 | NEG |
10a | IDC+ DCIS | POS | POS | 0 | 0 | 0 | 2 | 30 | 60 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
4 | IDC | POS | POS | 1 | 100 | 100 | 3 | 20 | 60 | 2 | 65 | 130 | 0 | 0 | 0 | 2 | 100 | 200 | 3 | 40 | 120 | 2 | 80 | 160 | 3 | POS |
7 | IDC | NA | NA | 2 | 100 | 200 | 3 | 30 | 90 | 1 | 40 | 40 | 1 | 100 | 100 | 2 | 100 | 200 | 3 | 90 | 270 | 1 | 2 | 2 | 3 | POS |
5 | IDC | POS | POS | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 90 | 90 | 1 | 100 | 100 | 1 | 50 | 50 | 0 | 0 | 0 | 1 | 20 | 20 | 0 | NEG |
11 | DCIS | NEG | NEG | 1 | 80 | 80 | 3 | 20 | 60 | 1 | 30 | 30 | 1 | 100 | 100 | 1 | 50 | 50 | 0 | 0 | 0 | 3 | 90 | 270 | 3 | POS |
12 | IDC | NA | NA | 2 | 100 | 200 | 3 | 40 | 120 | 2 | 100 | 200 | 1 | 100 | 100 | 1 | 100 | 100 | 1 | 100 | 100 | 3 | 95 | 285 | 3 | POS |
13 | DCIS | NA | NA | 2 | 100 | 200 | 3 | 70 | 210 | 2 | 100 | 200 | 1 | 100 | 100 | 1 | 95 | 95 | 2 | 90 | 180 | 3 | 95 | 285 | 3 | POS |
14 | IDC + DCIS | POS | POS | 1 | 100 | 100 | 0 | 0 | 0 | 1 | 100 | 100 | 1 | 100 | 100 | 2 | 100 | 200 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | NEG |
. | . | . | . | Phospho-MEK1 . | Phospho-MAPK (Erk 1/2) P44/42 . | MAPK (Erk 1/2) . | PTEN . | AKT . | mTOR . | p53 . | HER2/ErbB2 . | |||||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ID . | Histology . | ER . | PR . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Inten-sity . | % . | Score . | Score . | Interpretation . |
1 | IDC + DCIS | POS | NEG | 1 | 50 | 50 | 3 | 5 | 15 | 1 | 95 | 95 | 1 | 100 | 100 | 1 | 80 | 80 | 1 | 100 | 100 | 3 | 95 | 285 | 3 | POS |
3 | IDC | NA | NA | 1 | 50 | 50 | 0 | 0 | 0 | 1 | 70 | 70 | NA | NA | 1 | 100 | 100 | 1 | 100 | 100 | 2 | 85 | 170 | 3 | POS | |
6 | DCIS | NA | NA | 1 | 100 | 100 | 3 | 3 | 9 | 2 | 100 | 200 | 1 | 100 | 100 | 2 | 95 | 190 | 2 | 100 | 200 | 1 | 30 | 30 | 2–3 | POS |
8 | IDC + DCIS | POS | NEG | 1 | 100 | 100 | 3 | 2 | 6 | 1 | 100 | 100 | 1 | 100 | 100 | 2 | 100 | 200 | 1 | 100 | 100 | 3 | 95 | 285 | 3 | POS |
9 | IDC + DCIS | POS | POS | 1 | 100 | 100 | 0 | 0 | 0 | 2 | 90 | 180 | 1 | 100 | 100 | 1 | 80 | 80 | 1 | 100 | 100 | 1 | 5 | 5 | 1 | NEG |
2 | IDC + DCIS | POS | POS | 1 | 100 | 100 | 3 | 1 | 3 | 1 | 95 | 95 | 1 | 100 | 100 | 1 | 100 | 100 | 1 | 90 | 90 | 1 | 5 | 5 | 1 | NEG |
10a | IDC+ DCIS | POS | POS | 0 | 0 | 0 | 2 | 30 | 60 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | 0 | |
4 | IDC | POS | POS | 1 | 100 | 100 | 3 | 20 | 60 | 2 | 65 | 130 | 0 | 0 | 0 | 2 | 100 | 200 | 3 | 40 | 120 | 2 | 80 | 160 | 3 | POS |
7 | IDC | NA | NA | 2 | 100 | 200 | 3 | 30 | 90 | 1 | 40 | 40 | 1 | 100 | 100 | 2 | 100 | 200 | 3 | 90 | 270 | 1 | 2 | 2 | 3 | POS |
5 | IDC | POS | POS | 0 | 0 | 0 | 0 | 0 | 0 | 1 | 90 | 90 | 1 | 100 | 100 | 1 | 50 | 50 | 0 | 0 | 0 | 1 | 20 | 20 | 0 | NEG |
11 | DCIS | NEG | NEG | 1 | 80 | 80 | 3 | 20 | 60 | 1 | 30 | 30 | 1 | 100 | 100 | 1 | 50 | 50 | 0 | 0 | 0 | 3 | 90 | 270 | 3 | POS |
12 | IDC | NA | NA | 2 | 100 | 200 | 3 | 40 | 120 | 2 | 100 | 200 | 1 | 100 | 100 | 1 | 100 | 100 | 1 | 100 | 100 | 3 | 95 | 285 | 3 | POS |
13 | DCIS | NA | NA | 2 | 100 | 200 | 3 | 70 | 210 | 2 | 100 | 200 | 1 | 100 | 100 | 1 | 95 | 95 | 2 | 90 | 180 | 3 | 95 | 285 | 3 | POS |
14 | IDC + DCIS | POS | POS | 1 | 100 | 100 | 0 | 0 | 0 | 1 | 100 | 100 | 1 | 100 | 100 | 2 | 100 | 200 | 0 | 0 | 0 | 0 | 0 | 0 | 1 | NEG |
Abbreviations: DCIS, ductal carcinoma in situ; ER, estrogen receptor status; IDC, invasive ductal carcinoma; NA, not available; PR, progesterone receptor status.
aSample excluded because staining pattern suggests poor sample processing leading to false negative results.
CN analysis by Affymetrix OncoScan FFPE assay
Sufficient genomic DNA was obtained from 11 breast cancer specimens from 11 NF1 women. OncoScan FFPE Assay applied to these samples generated 9 datasets that were considered to be of sufficient quality to be analyzed. CN gain on 1q, 8q, and CN loss on 8p were observed as prominent features of these samples, which coincide with the landscape features of general breast cancers from TCGA and arrayMap datasets (15, 16; Supplementary Fig. S2).
By manually examining each sample in NF + BrCa using Affymetrix ChAS software, we identified ERBB2 CN gain in 2 samples and ERBB2 CN amplification in 2 samples (Table 2). Only one sample has the CN gain involving the entire 17q arm (CN = 2.33). NF1 loss of heterozygosity (LOH) was observed in 4 of 9 samples. Many genes have been found to be amplified or overexpressed in various cancers (13). CN status of 77 such genes was examined using ChAS software. CN gain/amplification of the following genes appeared to be much more prominent in NF + BrCa than those in TCGA METABRIC datasets: YEATS4, MDM2, and DYRK2, 55% (5/9) versus 10% (251/2,509); CCND2, 44% (4/9) versus 10% (251/2,509); CDK4 55% (5/9) versus 7% (176/2,509), and KRAS, 44% (4/9) versus 8% (201/2,509; Table 2). ChAS chromosome view showed that the CN gain of these genes is mostly a result of CN gain of the entire chromosome 12 or the long arm of chromosome 12. Statistical analysis is not applied for the comparison because of the small sample size of NF + BrCa and the different methods from which the TCGA CN datasets were generated.
CN, allelic status, and sequencing variants of targeted genes in NF + BrCa samples
. | . | . | . | . | . | . | HER2/NEU/ERBB2 . | NF1 . | TP53 . | 12q15 YEATS4, MDM2, DYRK2 . | 12q14.1 CDK4 . | 12p13.32 CCND2 . | 12p12.1 KRAS . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ID . | Quality . | Total CN Aberr. . | LOH % . | Genome changed % . | Aberr. cells % . | Ploidy . | IHC . | CN . | LOH . | CN . | LOH . | Tumor V . | Germline V . | CN . | LOH . | Tumor V . | CN . | LOH . | CN . | CN . | CN . |
1 | 0.0799 | 75 | 0.93 | 15 | Homo | 2 | POS | Het, G>T, chr17:29662050 | Het, c.5943+1G>T | 1.5 | |||||||||||
3 | 0.0899 | 197 | 23.3 | 57.6 | NA | NaN | POS | 8.67 | 1.67 | LOH | ND | NA | 1.67 | LOH | Het TrnV C>T, R to Ter; rs121913344 | 1.67 | LOH | 2.67 | |||
6 | 0.102 | 128 | 17.4 | 21.7 | 30 | 2 | POS | ND | NA | 1.5 | |||||||||||
8 | 0.157 | 93 | 13.3 | 20.6 | Homo | 2 | POS | LOH | 1.5 | LOH | ND | Het c.1733T>G (p.Leu578Arg) | 1.5 | LOH | Hom, T>C, H to R, rs786201838 confirmed somatic | 2.5 | 2.5 | 2.5 | |||
9 | 0.206 | 82 | 16.7 | 60.6 | NA | NaN | NEG | 2.33 | LOH | 2.33 | LOH | NS | Het Del exon 37 and 39 | LOH | 2.33 | 2.33 | 2.33 | 2.33 | |||
2 | 0.0863 | 99 | 28.8 | 40.9 | NA | NaN | NEG | 2.33 | NS | Het SpliceV, c.1260+1604A>G | 2.33 | 2.33 | 2.33 | 2.33 | |||||||
4 | 0.0922 | 36 | 9.55 | 13.3 | 60 | 2 | POS | ND | Het TrnV/SpliceV in exon 21 (c.2621_2634dupAGGGTTCTATGATT) | ||||||||||||
7 | 0.104 | 199 | 17.3 | 75.8 | 40 | 3 | POS | 25 | LOH | Hom, TrnV, Chr17:29657433TC>T | NA | LOH | Het TrnV, TA>T, Chr17:7578266 confirmed somatic | 3 | 3 | ||||||
5 | 0.0984 | 29 | 1.95 | 10.8 | Homo | 2 | NEG | ND | Het TrnV c.1541_1542delAG | 2.5 | 2.5 | 2.5 | 2.5 |
. | . | . | . | . | . | . | HER2/NEU/ERBB2 . | NF1 . | TP53 . | 12q15 YEATS4, MDM2, DYRK2 . | 12q14.1 CDK4 . | 12p13.32 CCND2 . | 12p12.1 KRAS . | ||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
ID . | Quality . | Total CN Aberr. . | LOH % . | Genome changed % . | Aberr. cells % . | Ploidy . | IHC . | CN . | LOH . | CN . | LOH . | Tumor V . | Germline V . | CN . | LOH . | Tumor V . | CN . | LOH . | CN . | CN . | CN . |
1 | 0.0799 | 75 | 0.93 | 15 | Homo | 2 | POS | Het, G>T, chr17:29662050 | Het, c.5943+1G>T | 1.5 | |||||||||||
3 | 0.0899 | 197 | 23.3 | 57.6 | NA | NaN | POS | 8.67 | 1.67 | LOH | ND | NA | 1.67 | LOH | Het TrnV C>T, R to Ter; rs121913344 | 1.67 | LOH | 2.67 | |||
6 | 0.102 | 128 | 17.4 | 21.7 | 30 | 2 | POS | ND | NA | 1.5 | |||||||||||
8 | 0.157 | 93 | 13.3 | 20.6 | Homo | 2 | POS | LOH | 1.5 | LOH | ND | Het c.1733T>G (p.Leu578Arg) | 1.5 | LOH | Hom, T>C, H to R, rs786201838 confirmed somatic | 2.5 | 2.5 | 2.5 | |||
9 | 0.206 | 82 | 16.7 | 60.6 | NA | NaN | NEG | 2.33 | LOH | 2.33 | LOH | NS | Het Del exon 37 and 39 | LOH | 2.33 | 2.33 | 2.33 | 2.33 | |||
2 | 0.0863 | 99 | 28.8 | 40.9 | NA | NaN | NEG | 2.33 | NS | Het SpliceV, c.1260+1604A>G | 2.33 | 2.33 | 2.33 | 2.33 | |||||||
4 | 0.0922 | 36 | 9.55 | 13.3 | 60 | 2 | POS | ND | Het TrnV/SpliceV in exon 21 (c.2621_2634dupAGGGTTCTATGATT) | ||||||||||||
7 | 0.104 | 199 | 17.3 | 75.8 | 40 | 3 | POS | 25 | LOH | Hom, TrnV, Chr17:29657433TC>T | NA | LOH | Het TrnV, TA>T, Chr17:7578266 confirmed somatic | 3 | 3 | ||||||
5 | 0.0984 | 29 | 1.95 | 10.8 | Homo | 2 | NEG | ND | Het TrnV c.1541_1542delAG | 2.5 | 2.5 | 2.5 | 2.5 |
NOTE: Quality, the quality and general assessment of the genome in cells given by OncoScan; Aberr., aberrations; NA, not available; Homo, homogeneous when aberrant cells are 100%; ploidy, the most likely ploidy state and is assigned as the median CN state of all markers; NaN, the percentage of aberrant CNs cannot be determined; POS, positive; NEG, negative; CN, copy number called by OncoScan; LOH, loss of heterozygosity called by OncoScan; Tumor V, the variant was identified by tumor tissue sequencing using OneSeq; Germline V, the variant was identified in the lymphocytes of the subject; these samples with germline mutations were published by Wang and colleagues, 2017.
Variant description, described in terms of changes in cDNA nucleotide, amino acid, GRCh37/hg19 genomic coordinates, and its reference SNP ID number “rs#”; Hom, homozygous; Het, hyeterozygous; TrnV, nonsense or frameshift variant; SpliceV, splicing variant.
ND, not detected; and NS, variant detected is not significant as it is not expected to affect function.
Targeted gene sequencing analysis
OneSeq Constitutional Research Panel (targeted at hereditary diseases) sequencing generated limited data from the DNA extracted from archived tumor specimens. Identifiable pathogenic variants of NF1 and TP53 gene are included in Table 2. This OneSeq panel did not identify any pathogenic variant of other genes related to hereditary risk for breast cancers (ATM, BARD1, BRCA1, BRCA2, BRIP1, CDH1, CHEK2, NBN, PALB2, PTEN, RAD50, and STK11). Many known somatic tumor-driver gene mutations are not covered by this panel. Variants without convincing evidence of pathogenicity for cancer were filtered out and are not included in this report.
TCGA data analysis of unselected breast cancer specimens
HER2 status.
In the TCGA and TCPA breast invasive carcinoma dataset, the HER2 expression, measured by RNA-seq and protein array (RPPA), was significantly lower in tumors with NF1 mutation/deletion (categorized as “NF1 altered”) compared with tumors without such alteration (P < 0.001). In this analysis, “NF1 altered” is defined as the samples with deleterious mutations or heterozygote CN loss or homozygous CN loss; “NF1 unaltered” is defined as CN neutral or with CN gain/amplification. However, when we separated the samples with HER2 positive from those with HER2-negative status, we discovered a divergent pattern based on NF1 alteration status. Among tumors with ERBB2 amplification reported by FISH, the ERBB2 mRNA and protein were significantly overexpressed in NF1 mutated/deleted tumors compared with the NF1 unaltered tumors, P = 0.005 and P = 0.0402, respectively. Among ERBB2 nonamplified tumors identified by FISH, HER2 mRNA, and protein expression were significantly lower in NF1-mutated/deleted tumors, P < 0.001 and P < 0.001, respectively (Fig. 1; Supplementary Tables S1 and S2).
Expression of ERBB2 mRNA and its protein product HER2 in TCGA breast-invasive carcinoma dataset. The samples are stratified by their NF1 gene status and the presence of HER2 amplification.
Expression of ERBB2 mRNA and its protein product HER2 in TCGA breast-invasive carcinoma dataset. The samples are stratified by their NF1 gene status and the presence of HER2 amplification.
Only 375 of 960 TCGA samples with complete sequencing data have HER2 FISH status. To further verify the above divergent ERBB2 expression pattern, we repeated this analysis on 960 samples with ERBB2 CN data generated by tumor next-generation sequencing. We define the “CN gain” as log2 (ratio = CN in tumor/CN in normal tissue) > 1 and ≤ 2, that is equivalent to 2 < CN ≤ 4; “CN amplification” as log2 (ratio) > 2, that is equivalent to CN > 4. Once again, in tumors with ERBB2 CN gain or amplification, ERBB2 mRNA and protein were significantly overexpressed in NF1 mutated/deleted tumors compared with unaltered tumors, P < 0.001 and P < 0.001, respectively. Among tumors without ERBB2 gain or amplification, ERBB2 mRNA and protein expression were significantly lower in NF1-mutated/deleted tumors, P < 0.001 and P < 0.001, respectively (Supplementary Tables S1 and S2).
However, after removing ERBB2 CN gained, this ERBB2 overexpression phenomenon is no longer significant on ERBB2-amplified samples at mRNA or protein level, P = 0.0645 and P = 0.1552, respectively. The effect of altered NF1 in tumors with ERBB2 CN not gained or nonamplified remains the same, P < 0.001 (Supplementary Tables S1 and S2).
Genes on chromosome 12.
Because several genes on chromosome 12 (YEATS4, MDM2, DYRK2, CCND2, CDK4, and KRAS) appeared to have more CN gain in NF + BrCa samples, we examined them in TCGA datasets. In TCGA samples, mRNA expression of DYRK2 and CDK4 on the long arm of chromosome 12, that is, 12q, was both elevated in NF1 mutated/deleted samples (P < 0.001, multiple test P < 0.001; Supplementary Table S1). mRNA of YEATS4 and MDM2; however, were decreased (P = 0.045 and P < 0.001, respectively, with multiple test P = 0.805 and P < 0.001, respectively). KRAS, the gene on the short arm, 12p12.1, showed preferential CN gain/amplification (33% in NF1 altered vs. 23% in NF1 nonaltered; P < 0.001). The mRNA of KRAS showed overexpression (P = 0.0097) in TCGA NF1 altered samples; however, it failed the multiple testing (0.174; Supplementary Table S1).
Akt pathway.
We examined a Ras responder, Akt in TCGA samples. The NF1 altered samples showed increased level of Akt proteins (P = 0.0047, P = 0.0424 after correction for multiple testing; Supplementary Table S2), which appeared to be primarily contributed by higher mRNA expression of AKT1 only (P < 0.001; Supplementary Table S1). However, the levels of Akt phosphorylation at residues T308 and S473 did not appear to show significant difference by NF1 status (Fig. 2; Supplementary Table S2).
Protein and mRNA levels of selected signaling molecules in the TCGA breast-invasive carcinoma dataset. The samples are stratified by their NF1 gene status.
Protein and mRNA levels of selected signaling molecules in the TCGA breast-invasive carcinoma dataset. The samples are stratified by their NF1 gene status.
MEK–ERK pathway.
MEK–ERK pathway is another major responder to Ras signaling. The NF1 altered did not show a preferentially elevated phosphorylation of the MEK1 protein. Furthermore, the MEK/ERK pathway downstream effector, Erk1/2 protein residues T202/Y204 were slightly less phosphorylated in NF1 altered samples (P = 0.0098). However, the association is not significant after correcting for multiple testing (P = 0.0879; Fig. 2; Supplementary Table S2).
Expression of other selected genes and proteins.
In comparison with NF1 unaltered samples, total protein level of mTOR is decreased (P < 0.001, P = 0.0031 after correction for multiple testing); p53 protein level is slightly increased (P < 0.001; Fig. 2; Supplementary Table S2).
Discussion
To our knowledge, this is the first study exploring the genomic and histologic features of breast cancer harboring germline defects in the NF1 gene. By interrogating the histology, genomic and proteomic profiles of tumor samples with germline versus somatic NF1 mutations, we aimed at uncovering previously uncharacterized consequences of NF1 deficiency in breast cancer.
Our analysis revealed an unusual pattern of HER2 expression both in the breast tumor samples with germline NF1 mutations and with somatic NF1 mutations. Our small cohort of breast cancer with germline NF1 mutations (NF + BrCa) demonstrated HER2 overexpression by IHC in a majority of the samples (69%, 9/13). The ERBB2 CN amplification (n ≥ 4, generated by Affymetrix OncoScan) alone is found in 22% (2/9) and CN low-level gain (4 > n > 2) is found in 22% (2/9), which echoes the findings in the Finnish study showing HER2 amplification by FISH analysis in 31% (8/26) of the tumors from NF1 patients (3). However, our 9 tumor samples' IHC HER2 positivity was discordant with the ERBB2 CN that 4 of the 6 IHC strong positive samples were with no detectable ERBB2 gene gain or amplification. Chromosome 17 polysomy did not appear to be responsible for the ERBB2 CN gain or amplification in the NF + BrCa samples. One sample had CN gain involving the entire 17q arm. FISH was not performed on these samples. This discordance between IHC and CN amplification may be due to the heterogeneity of the tissue, that is, the CN gain being present only in a small percentage of cells and consequently below the limit of detection in next-generation sequencing and microarray analyses that use genomic DNA extracted from homogenized tumor tissue. The overexpression observed by HER2 IHC staining may also be a result of enhanced transcription or translation rather than genomic CN gain or amplification. In comparison, IHC and FISH results' discordance has been seen in a much smaller scale in TCGA unselected breast cancer samples. Approximately 80% of the HER2 overexpression is due to ERBB2 gene amplification detected by FISH (9, 17, 18). Our NF + BrCa data suggest that HER2 may be preferentially overexpressed in NF1 deficient breast tumor via some mechanisms in addition to HER2 CN gain or amplification.
Upon further dissection of the sporadic breast cancer TCGA genome, RNA expression and protein expression data, we discovered an intriguing phenomenon that when ERBB2 CN gain or amplification presents with defective or deficient NF1, the HER2 expression is much more pronounced compared with normal NF1. On the contrary, when ERBB2 is without CN gain, the NF1 deficiency or defect surprisingly associates with relatively lower HER2 expression. Our analysis was based on correlating NF1 mutations or deletions with HER2 expression. Because NF1 and ERBB2 genes are located in close proximity on chromosome 17, deletions of the NF1 gene may also span the ERBB2 gene. Indeed, the samples with NF1 deletion and no gain or amplification of ERBB2 mostly contained loss of ERBB2 CN, which is likely to explain the decrease in HER2 expression. In contrast, when ERBB2 amplification is detected by FISH or sequencing, the deletion of the NF1 gene does not extend to ERBB2, allowing higher expression of HER2. It therefore seems that the defective NF1 associates with either upregulation of the HER2 expression or a survival benefit for HER2 overexpression at least when excessive ERBB2 copies are present. This may also be the case in the absence of ERBB2 gain/amplification but the overlapping chromosomal deletions mask the effect.
Both germline and somatic alterations of the NF1 gene associate with an increase in HER2 expression in breast cancer. Also, ERBB2 amplifications are overrepresented in breast tumors of NF1 patients, as demonstrated in this study and also described previously (3). It therefore seems that NF1 deficiency and HER2 overexpression act synergistically to provide the cancer cells an evolutionary benefit. Further molecular and biochemical studies at the cellular level are needed to reveal the nature of NF1–HER2 interaction.
Several mechanisms other than gene amplification have been proposed to regulate HER2 expression, including transcription factors, catecholamine receptor, and nuclear receptor coactivator (19–22). Kannan and Tainsky described Ras activating mutation that led to cell transformation by activation of transcription factor AP-2α (23). Later, Kannan discovered that anomalous abundance of AP-2α is accompanied by elevated levels of HER2 protein in a mammary epithelial cell line (24). HER2-positive breast cancer has long been recognized as a unique category with well-established targeted treatment options (25, 26). However, resistant tumors still exist. Understanding the mechanism of enhanced HER2 transcription may offer novel therapeutic approaches.
NF1 LOH was not consistently observed in this small cohort of breast cancer with a germline heterozygous NF1 mutation. This is in contrast to the most common malignant neoplasm in NF1, malignant peripheral nerve sheath tumor (MPNST), where biallelic NF1 loss is a common feature. Although NF1 LOH may not be essential to NF + BrCa, observing it in as many as 4 of 9 tumors suggests an oncogenic role of NF1 LOH in breast cancer development. Next-generation whole-exome sequencing of the tumor sample may not be comprehensive enough to identify all NF1 mutations, partly because of the heterogeneity of the tumor tissue, the wide spectrum of the NF1 alterations (including deep-intron splice variants and rearrangements), the large gene size and the complex gene structure (27). Because of these limitations, it is possible that some biallelic NF1 defects in the sample were not detected. In addition, because NF1 is associated with an approximately 2-fold lifetime risk of breast cancer, half of the breast cancer cases would occur even without the NF1 mutation. Therefore, it is not surprising that no biallelic loss of NF1 was observed in 5 of 9 tumors. These NF + BrCa samples analyzed can be “sporadic” in nature without being driven by NF1 germline mutation. Some of the NF + BrCa specimens do not have sequencing data from a matched sample representing germline tissue. Therefore, the somatic versus germline origin of the mutation identified in tumor tissue cannot be determined in all samples.
It is known that various NF1-deficient tumor cells and cell lines in human or animals, such as MPNST, hematologic malignancies (AML, CMML, and JMML), melanoma, or optic glioma, demonstrate mostly activated Raf/MEK/ERK pathway and sometimes PI3K/Akt pathways (28–35). Clinically, NF1 plexiform neurofibroma has shown favorable responses to selumetinib, an MEK inhibitor (36). Investigating these pathways in breast cancer may yield useful information for future treatment. Our small cohort study was retrospective in nature and only archived FFPE tumor sample was available to be analyzed. It is known that formalin cross-links peptides may mask epitopes and impair the sensitivity of IHC assay. The age of storage may also cause the tissue section to lose its antigen reactivity. In addition, the protein phosphorylation is a dynamic process where prolonged time between resection and fixation can change the phosphorylation status in the tissue. IHC analysis of the archived NF + BrCa samples did not yield any signs of MEK–ERK or Akt pathway activation. We proceeded with analyzing samples in TCGA database. The expression and phosphorylation patterns of unselected breast cancers did not support an activated MEK–ERK pathway either. However, Akt expression was increased in NF1 altered samples, contributed by AKT1, but the Akt phosphorylation showed no difference.
The percentage of mutation or shallow CN loss of the TP53 gene in NF + BrCa was not deviated from the sporadic cancer in TCGA. No additional mutations in the hereditary risk genes were found. Many common breast cancer gene mutations in sporadic breast cancers were not tested in this small cohort.
The inherent limitation of this study is secondary to the small sample size of NF + BrCa, which was deemed exploratory in nature. More samples are needed to characterize the mutations and CN variations in contrast to sporadic breast tumor samples using comparable sequencing and array methodology. Limited tumor material and funding restricted the opportunity to a thorough molecular investigation with FISH or in-depth somatic tumor exome sequencing with attention to cancer driver genes. Targeted tumor sequencing results could not be confirmed with Sanger sequencing at this time, either.
This study demonstrates that both germline and somatic NF1 mutations associate with an increase in HER2 expression, resulting from either gene amplification or transcriptional upregulation. Furthermore, this is the first genetic characterization of breast cancer in NF1, providing basis for future mechanistic studies on breast cancer and NF1.
Disclosure of Potential Conflicts of Interest
J.O. Blakeley is a consultant/advisory board member for Abbvie. No potential conflicts of interest were disclosed by the other authors.
Authors' Contributions
Conception and design: M.A. Tainsky, X. Wang
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R.N. Tousignant, S.J. Yoder, J.O. Blakeley, M.T. Acosta, B.R. Korf, L.M. Messiaen, X. Wang
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): R.A. Kallionpää, P.R. Gonzales, D.A. Chitale, J.P. Crowley, Z. Chen, J.O. Blakeley, L.M. Messiaen, X. Wang
Writing, review, and/or revision of the manuscript: R.A. Kallionpää, P.R. Gonzales, D.A. Chitale, R.N. Tousignant, J.O. Blakeley, M.T. Acosta, B.R. Korf, L.M. Messiaen, M.A. Tainsky, X. Wang
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): J.P. Crowley, L.M. Messiaen, X. Wang
Study supervision: M.T. Acosta, X. Wang
Acknowledgments
This study was supported by the New Investigator Award of U.S. Army Medical Research and Material Command (W81XWH-11-1-0671), the Department of Individualized Cancer Care Management in Moffitt Cancer Center and NF Michigan. The authors would like to thank Jeannette Schmidt, PhD, Affymetrix (Thermo Fisher Scientific), for her assistance on OncoScan data analysis. The authors also wish to especially thank Dr. Howard L. McLeod, for his generous support to complete this project.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.