Sequence-based detection of mutations in cadherin 1 to determine the prevalence of germline mutations in patients with invasive lobular carcinoma of the breast

Breast cancer represents a group of diseases, with different histological, molecular and clinical characteristics, including tumor growth patterns, with at least 17 different histological types currently described [1]. Invasive lobular carcinoma (ILCA), the second most common histological type, accounts for 8-14% of all breast cancers [2, 3]. ILCA is histologically characterized by a loss of cellular adhesion, resulting in growth of single cells or cells arranged in a single file or “Indian lines” [3, 4]. ILCA is often bilateral, multifocal and multicentric, and is more difficult to detect by palpation or mammography [5, 6]. Compared to invasive ductal carcinoma (IDCA), ILCA is more often diagnosed at a later age, more likely to be hormone receptor positive, larger, diploid, and HER2, p53 and EGFR negative. ILCA has a distinctive pattern of metastasis, with a propensity to metastasize to the peritoneum, gastrointestinal tract and ovaries. Despite its favorable characteristics, disease-free and overall survival did not differ between ILCA and IDCA [7].

One of the hallmarks of ILCA is the loss or altered expression of cadherin 1 (CDH1, GenBank: AB025105.1). CDH1 is a member of the classical/type I cadherin family that is involved in maintaining cellular adhesion. CDH1 interacts with members of the catenin family; this complex then links with actin molecules to provide cellular adhesion [8]. Altered expression of CHD1 explains the discohesive phenotype characteristic of not only ILCA but also hereditary diffuse gastric cancer (HDGC). In fact, ILCA is a frequent finding in HDGC families, with a 39% lifetime risk of developing ILCA in HDGC families with CDH1 germline mutations [9].

The frequency of CDH1 germline mutations in families with HDGC is 25-30% [10]. Determination of the frequency of CDH1 mutations in patients with ILCA is crucial; clinically, the diffuse nature of ILCA makes early detection more challenging [5] and increased risk for bilateral disease impacts the choice of surgical approaches. In addition, presence of hereditary CDH1 mutations will impact risk assessment, family counseling and screening, for both breast and stomach malignancies. To this end, CDH1 was screened for germline mutations in a population of patients with ILCA.

All patients included in this study were enrolled in the Clinical Breast Care Project (CBCP) between 2001 and 2010. Blood samples were collected with approval from the Walter Reed National Military Medical Center Human Use Committee and Institutional Review Board. All subjects voluntarily agreed to participate and gave written informed consent. Clinical information, including family cancer history, was obtained for all CBCP participants using questionnaires designed by and administered under the auspices of the CBCP.

The CBCP database was queried to identify all female patients diagnosed with ILCA. Family history of cancers was self-described by the patients and included data for both primary and secondary relatives. Diagnosis of every specimen was performed by a dedicated breast pathologist from hematoxylin and eosin (H&E) stained slides; staging was performed using guidelines defined by the AJCC Cancer Staging Manual seventh edition [11]. In addition, to determine the frequency of the A617T variant in African American women, all female African Americans with invasive breast cancers such as IDCA and tumors with mixed IDCA and ILCA features were identified.

Genomic DNA was isolated from blood clots using the GentraClotspin and Puregene DNA purification kits (Qiagen, Valencia, CA). Polymerase chain reaction (PCR) primers were designed to flank CDH1 exons 1–15 and the coding region of exon 16 (Table 1). PCR reactions were performed on 25 ng genomic DNA using Lightscanner Master Mix (BioFire Defense, Salt Lake City, UT) and analyzed on the LightScanner System following manufacturer’s protocols. For those samples with a variant detected by the LightScanner, PCR products were purified using ExoSAP-IT For PCR Product Clean-up (Affymetrix, Santa Clara, CA), and sequenced using BigDye terminator v 3.1 cycle sequencing kits (Life Technologies, Inc, Carlsbad, CA). The resulting products were sequenced on a 3730xl DNA Analyzer (Life Technologies, Inc, Carlsbad, CA) and sequence files were analyzed using Sequencher 4.10.1 (Gene Codes Corporation, Ann Arbor, MI).

The A617T variant was investigated using TaqMan SNP genotyping assay C__32307311_10 (Life Technologies, Inc, Carlsbad, CA) using 10 ng of DNA; reactions were amplified in duplicate and genotypes (TT, TC or CC) were determined using the ABI PRISM^® 7000 Sequence Detection System software (Applied Biosystems, Foster City, CA).

Identification of patients with hereditary predisposition to breast cancer is critical for optimum clinical management. Identification of carriers of causative mutations can assist in risk assessment and risk reduction, as well as in family counseling. In addition, hereditary cancers may have different phenotypes, requiring different treatment regimens. For example, patients with BRCA1 mutations are at risk for both bilateral breast disease as well as ovarian carcinoma; thus, women with BRCA1-associated cancers may be offered bilateral mastectomy and/or oophorectomy and may derive more benefit from PARP inhibitors compared to patients with sporadic breast cancer [12]. Patients with both hereditary and sporadic forms of ILCA may benefit from targeted therapeutics such as the use of FGFR1 inhibitors that block survival pathways and reverse tamoxifen resistance [13], however, those with hereditary ILCA may, as those with BRCA1 and BRCA2 mutations, be subjected to increased screening for bilateral disease, offered risk reduction strategies, as well encouraged to undergo prophylactic gastrectomy [14].

The frequency of potentially causative mutations in this study was low, with no instances of splice site or nonsense mutations detected. Only 5% (5/100) of patients had missense mutations. Under current guidelines in which genetic testing for BRCA1 and BRCA2 mutation is recommended for women with estimated probabilities of having a germline mutation ≥10% [15], genetic testing for CDH1 mutations in patients with ILCA would not be warranted. In addition, the clinical importance of the variants detected is unclear. To determine the pathogenicity of these variants, the pubmed database (http://​www.​ncbi.​nlm.​nih.​gov/​pubmed/​) was searched for any citations that included these variants. In addition, the Catalog of Somatic Mutations in Cancer (COSMIC) database (http://​cancer.​sanger.​ac.​uk/​cancergenome/​projects/​cosmic/​) was searched. To our knowledge, variants P825L and G879S have not been documented as causative mutations in the literature, although a single instance of the P825L variant in a lung tumor was reported in the COSMIC database. Functional assessment of A592T suggests this variant is likely tolerable and non-pathogenic [16]. The A617T variant has been detected in a patient with endometrial cancer [17] and in two female African American patients with diffuse gastric cancer [18]. Transfection of CHO cells with CDH1 carrying the A617T variant resulted in an epithelial phenotype with extensive cell-to-cell contacts with membranous E-cadherin and β-catenin staining similar to that in wild-type cells [19]. Data from our study supports A617T as being a non-pathogenic variant; 10/165 African American women with IDCA and one with mixed IDCA and ILCA were carriers of the minor allele. In addition, one 71 year-old woman with IDCA was homozygous for the minor allele. The 6.7% minor allele frequency for A617T detected in our patient samples is identical to that reported in dbSNP for African American populations; thus this variant likely represents a non-pathogenic SNP specific to individuals with African ancestry. In addition, we performed a SIFT (S orting I ntolerant F rom T olerant) [20] analysis for the four non-synonymous variants. A592T, A617T and G879S were classified as tolerable variants, while P825L was predicted to be intolerable.

Our ability to detect mutations may be limited by the technical approach employed. Large deletions may not be detected by high-resolution melting analysis or direct sequencing of amplified products, however, the reported frequency of large-scale deletions in HDGC families is low (3.8%) [21], thus it is not likely that many of the ILCA patients have an undetected deletion. In addition, genomic DNA samples from family members were not collected, thus, it was not possible in this study to determine if the variants tracked with disease status. Finally, identification of patients with a family history of gastric cancer was based on self-reporting of cancer within first- and second-degree family members by the patient and ages at diagnosis were not provided. Of the 100 patients with ILCA, six had a single self-reported family member with stomach or gastric cancer, thus, the majority of these patients may not have a family history of HDGC.

In our data of patients with ILCA unselected for age at diagnosis or family history of breast cancer, the mutation frequency was low – four missense mutations, only one of which (P825L) has been predicted to be pathogenic - in five individuals. These patients were recruited from a breast, rather than gastroenterology clinic, and thus the mutational frequency and spectrum differs based on presence or absence of a significant family history of HDGC. A previous study of 318 women with ILCA diagnosed <45 years or with a family history of breast cancer found no deletions or truncating mutations and of the six detected non-synonymous mutations, only four were found to be pathogenic, for a mutation frequency of 1.3% [22]. Together, these data suggest that the frequency of germline mutations in patients with ILCA is low, and screening for CDH1 mutations in this population is not warranted.