BMP2 alterations in mucinous cystadenocarcinoma of the breast: insights from whole-exome sequencing

Tissue collection

This study was conducted in accordance with the Declaration of Helsinki and approved by the ethics committee of Chongqing University Cancer Hospital (approval number: CZLS2023062-A). The ethics committee granted a waiver of informed consent for this retrospective study, while biological sample storage informed consent was maintained. Cases were selected based on the following criteria: histopathological diagnosis of breast mucinous cystadenocarcinoma, tumor tissue containing >50% tumor cells, and availability of complete clinical data. Cases with insufficient tissue material, poor DNA quality, or incomplete clinical information were excluded.

Based on these criteria, we retrospectively identified three female patients diagnosed with breast MCA who were treated at Chongqing University Cancer Hospital between June 2020 and October 2022. Formalin-fixed, paraffin-embedded (FFPE) tissue samples were collected from these patients. Relevant clinical and pathological data were extracted from medical records and are summarized in Table 1.

Hematoxylin and eosin (H&E) staining and whole-slide image acquisition

The representative FFPE tissue blocks of breast tumor from the patients were sectioned at 5 μm for H&E staining (VENTANA HE 600 system; Roche Diagnostics, Indianapolis, IN, USA). All the stained slides were scanned at 20× magnification, with an image resolution of 0.25 mm/pixel scanner (MAGSCANNER, KF-PRO-005-HI; KF Biopathology, Ningbo City, China).

Whole exome sequencing and bioinformatics analysis

DNA extraction and whole-exome Next-Generation sequencing were performed in the laboratory of Beijing Tsingke Biotech Co., Ltd. according to standard procedures. Briefly, genomic DNA (gDNA) was extracted from tissues using a standard DNA extraction kit (FirePureTM FFPE gDNA Extraction Kit #FG0323; FireGen, Haidian, China), as previously described (Yang et al., 2023). DNA concentration was determined using Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA), with a minimum concentration requirement of 20 ng/μL. DNA integrity was assessed by agarose gel electrophoresis to ensure fragment size distribution above 500 bp. Exome capture libraries were prepared using the Hieff NGS Ultima Pro DNA Library Prep Kit for Illumina and SSELXT CRE V4 capture kit, following the manufacturer’s instructions. Whole-exome sequencing was performed on the DNBSEQ-T7 platform (MGI Tech, Shenzhen, China), generating paired-end 150 bp reads at >100× mean coverage.

Raw sequencing data were filtered using fastp to remove adapters and low-quality reads. Only data with Q30 (Quality Score 30) ≥85% were retained for analysis. Clean reads were aligned to the GRCh38 (Genome Reference Consortium Human Build 38) human reference genome using sentieon bwa mem (default parameters), and polymerase chain reaction (PCR)/optical duplicates were marked with the Genome Analysis Toolkit (GATK) MarkDuplicates (parameters: --OPTICAL_DUPLICATE_PIXEL_DISTANCE 2500, --ASSUME_SORT_ORDER “coordinate”, --CLEAR_DT false, --CREATE_MD5_FILE true). Variants (Single Nucleotide Variants (SNVs) and indels) were identified using GATK with parameter-T to set threshold values.

MuSiC2 software was used to analyze significantly mutated genes, and at least two genes with a false discovery rate (FDR) less than 0.05 were identified as having significant mutations (Dees et al., 2012).

Downstream analyses included the identification of copy number variations (CNVs), gene rearrangements, and other genomic alterations, following previous studies (Lorente-Bermúdez et al., 2024; Wu et al., 2024).

Immunohistochemistry

Immunohistochemical staining was performed using a Ventana Benchmark System (BenchMark ULTRA; Ventana Medical Systems, Tuscon, AZ, USA) with an UltraView Universal DAB (diaminobenzidine) Detection Kit for visualization. The primary antibody (BMP2, polyclonal, 1:200 dilution) was obtained from Boster Biological Technology Co., Ltd. (Wuhan, China).

Bioinformatics for significant mutational gene in breast cancer patients

The significantly mutated gene BMP2 was analyzed using the cBioportal website (https://www.cbioportal.org/), including mutation mapper, pathological characteristics, correlation analysis, methylation information, survival analysis, mRNA (messenger ribonucleic acid) expression, mutation count, immunocyte infiltration level, Venn diagram, and enrichment analysis.

Statistical analysis

All data were statistically analyzed using SPSS (Statistical Package for the Social Sciences) 17.0 software for non-parametric tests. The Wilcoxon test was used to compare two groups and the Kruskal test was used for the global comparison of multiple groups. Statistical significance was set at P < 0.05.

Clinical characteristics and pathological findings of patients with MCA

The clinicopathological characteristics of the three cases of breast MCA are presented in Table 1. The median age at diagnosis was 55 years (range 53–66 years). All three patients were post-menopausal. In case 1, computed tomography (CT) and magnetic resonance imaging (MRI) revealed a large mass (11.6 cm) with its greatest diameter in the outer region of the right breast, with unevenly increased intensity, which had invaded the chest wall and skin. A mass in the left breast was detected simultaneously after admission to our hospital (Supplemental S1A–S1C), whereas pelvic MRI indicated no lesions (Supplemental S1D). However, in cases 2 and 3, a solid cystic mass with unevenly increased intensity was detected by MRI in the upper outer quadrant (UOQ) of the right breast (3.5 cm in its greatest diameter) and upper inner quadrant (UIQ) of the left breast (3.0 cm in its greatest diameter), respectively, without a contralateral tumor identified simultaneously (Supplemental S2A, S2B). Enlarged lymph nodes >1 cm were found in the right auxiliary region of case 1, which were absent in cases 2 and 3 at the time of initial diagnosis. In case 1, the patient received neoadjuvant chemotherapy with seven cycles of docetaxel, doxorubicin (adriamycin), and cyclophosphamide (TAC) and three cycles of Navelbine with cisplatin (NP) before modified radical mastectomy of the left and right breasts, respectively. Pathological examination confirmed the diagnosis of bilateral breast cancers: infiltrating ductal carcinoma, not otherwise specified (IDC, NOS), grade 3 (Nottingham Grade), G1 (Miller-payne Grade) in the left breast and breast MCA, grade 2 (Nottingham Grade), and G1 (Miller-payne Grade) in the right breast with metastasis to the skin of the right chest wall. In cases 2 and 3, breast MCA was detected in the unilateral breast (right and left breast, respectively), and the patients underwent radical mastectomy with sentinel lymph node biopsy in the ipsilateral armpit plus post-operative chemotherapy; both were in pT2N0M0 stage IIA. After 21 and 12 months of follow-up respectively, the two patients showed no evidence of relapse or metastasis. However, in case 1, the patient experienced a dismal clinical course and received neoadjuvant chemotherapy (NACT) with seven cycles of TAC and three cycles of NP prior to surgery because of the extensive tumor burden of bilateral breast cancers. This was followed by an extended radical mastectomy for both the left and right breast cancers, sequentially. Postoperative treatment included seven cycles of navelbine (vinorelbine) plus xeloda (capecitabine) (NX) chemotherapy, radiotherapy (intensity-modulated radiation therapy (IMRT), 46 Gy), and oral capecitabine (1.5g * 2 Bid) as maintenance therapy. Ultimately, the patient was diagnosed with ypT4N2M1 and rT4N2M1 stage IV breast cancers on both sides in case 1. Within 2 years of surgery and systemic treatment, she developed multiple metastases in the right chest wall and was subsequently lost to follow-up.

On gross examination, median maximal diameter of breast cancer in the three cases was 3.75 cm (range = 3.0–15.0 cm), mostly with a well-circumscribed solid and cystic mass. Microscopically, different proportions of MCA components were observed: case 3 exhibited pure breast MCA, grade 3 (Nottingham Grade); case 2 featured a mix with IDC, NOS, grade 2 (Nottingham Grade); and case 1 presented synchronous bilateral breast cancers: IDC, NOS, grade 3 (Nottingham Grade), G1 (Miller-payne Grade) in the left breast, and breast MCA, grade 2 (Nottingham Grade), G1 (Miller-payne Grade) in the right breast with metastasis to the skin of the right chest wall). These MCA components displayed characteristic histological features, such as a single layer or pseudostratified tall columnar epithelial cells of low to intermediate nuclear grade and complex branched mucinous papillary structures, closely resembling those found in their ovarian or pancreatic counterparts (Figs. 1, 2). However, tumor tumor-infiltrating lymphocytes (TILs) in the stroma between the areas of carcinoma cells, including all mononuclear cells (lymphocytes and plasma cells), were less than 5% in all cases. Macro-metastasis in eight lymph nodes was identified in 26 right auxiliary lymph nodes, and no metastasis was found in the left in 26 right auxiliary lymph nodes (metastasis-positive lymph node number vs. total lymph node number: 0/33) of case 1, while no metastasis was found in the lymph nodes of cases 2 and 3 (metastasis-positive lymph node number vs. total lymph node number: 0/7, 0/12 respectively). In cases 2 and 3, high-grade ductal carcinoma in situ (DCIS) components were observed adjacent to the invasive tumor, corroborating the diagnosis of primary breast MCA. However, because the bilateral breast tumors in case 1 lacked precursor components, metastasis exclusion was crucial for accurate diagnosis. Immunostaining showed that the epithelial cells in the breast MCA were positive for cytokeratin 5/6 (CK5/6) (focal) (Fig. 3A), cytokeratin 7 (CK7) (Fig. 3B), and epidermal growth factor receptor (EGFR) (Fig. 3C), but negative for ER, PR, HER-2, CK20 (cytokeratin 20) (Fig. 3D), Caudal Type Homeobox 2 (CDX2) (Fig. 3E), and PAX8 (Paired Box Gene 8) (Fig. 3F), which was in accordance with primary breast MCA. In addition, a positron emission tomography/computed tomography (PET/CT) scan performed before surgery in case 1 also indicated a primary bilateral breast mass with extensive auxiliary lymph node metastasis and no signs of visceral organ lesions. Cases 2 and 3 also demonstrated a triple-negative (hormone receptor of ER, PR negative, HER2-negative) immunophenotype, and a high Ki-67 index was observed in three cases (40–60%, with a median value of 40%). The metastatic origin was also ruled out by additional imaging examinations in cases 2 and 3.

Gene alteration and significant mutant genes in MCA patients by WES

To determine the repertoire of genetic alterations in MCA, we subjected these three cases to WES. However, WES analysis revealed a scarcity of CNVs and somatic mutations in the corresponding chromosomes (Figs. 4A, 4B); the most commonly mutated genes were AKT Serine/Threonine Kinase 1 (AKT1), CUB and Sushi Multiple Domains 2 (CSMD2) TP53, Ubiquitin Specific Peptidase 42 (USP42), WD Repeat and FYVE Domain Containing 4 (WDFY4) (each altered in two out of the three cases). The tumor mutation burden in the three cases were all below 10 mutations/megabase (Mb), which was 6.21, 4.35, 1.92 mutations/Mb respectively. All three cases were in a microsatellite-stable status. Notably, in case 1, BMP2 was identified by MuSiC2 (false discovery rate (FDR) = 0.031) as a significant mutant gene (SMG) from the WES data of three patients, although its significance remains unclear. To explore whether BMP2 could serve as a substrate for the development of breast MCA, we compared the expression level of BMP2 in breast MCA with that in triple-negative breast cancer (TNBC) and the histology of IDC by immunohistochemistry. As shown in Fig. 5, BMP2 staining was positive across all MCA cases in our study (Figs. 5A, 5B); however, TNBC patients with IDC histology exhibited a faint staining pattern (Figs. 5C, 5D).

BMP2 is associated with tumor progression in breast cancer

To further explore the function of BMP2 in breast cancer, we used the cBioPortal online tool to analyze functional genomics. In The Cancer Genome Atlas (TCGA) and the Molecular Taxonomy of Breast Cancer International Consortium (METABRIC) databases, the frequencies of BMP2 mutations were 1.2% and 2.1%, respectively, with amplification and deep deletion being the main types of gene alterations (Fig. 6A). The location of BMP2 alteration is K297N on transforming growth factor-beta (TGF-b) propetide in METABRIC database (Fig. 6B). As depicted in Fig. 7A, BMP2 alteration was associated with tumor metastasis status and metastatic tumor indicators in TCGA database and correlated with Prediction Analysis of Microarray 50 (PAM50)-based molecular subtype and tumor stage in the METABRIC database. A positive correlation between Fraction genome altered and mutation count of BMP2 was identified (Fig. 8A). Additionally, methylation status negatively correlated with the mRNA expression of BMP2 (Fig. 8B).

Next, the frequency scatter of enrichments in the alteration of BMP2 in breast cancer patients was higher than that of the unaltered ones in the TCGA and METABRIC databases (Figs. 9A, 9B). One hundred and forty-seven related genes of BMP2 were identified in both the TCGA and METABRIC databases (Fig. 10A). The network of 147 related genes is depicted in Fig. 11A. The enrichment analysis of BMP2 and its 147 related genes is shown in Fig. 12A, encompassing the Reactome, Kyoto Encyclopedia of Genes and Genomes, cellular components, molecular functions, and biological processes.

Finally, we examined the effect of BMP2 expression on the survival rate of patients with breast cancer. Patients with genetic alterations in BMP2 showed reduced survival, as demonstrated by overall survival analysis in the TCGA database (Fig. 13A) and relapse-free status analysis in the METABRIC database (Fig. 13C). Although the overall survival of the BMP2-altered group was lower than that of the BMP2-unaltered group in the METABRIC database, this trend was not statistically significant (Fig. 13B). Analysis of the correlation between mRNA expression and CNV from the Genomic Identification of Significant Targets in Cancer revealed that the mRNA expression of BMP2 was associated with shallow deletion, diploidy, and gain in the TCGA database (Fig. 14A). Similarly, the mutation count and CNV from DNAcopy demonstrated that the BMP2 mutation was associated with shallow deletion, diploidy, and gain in the METABRIC database (Fig. 14B). It should be pointed out that genetic alterations of BMP2 were also linked to immune cells, including B cells, CD8 (Cluster of Differentiation 8)+ T cells, CD4 (Cluster of Differentiation 4)+ T cells, macrophages and dendritic cells, particularly in cases of arm-level deletion of BMP2 alterations (Fig. 15A).