Introduction

Breast cancer is the most frequently diagnosed cancer and the leading cause of death due to a malignancy among women [1]. Hereditary genetic factors are thought to account for approximately 5 to 10% of breast cancers due to germ-line variants in genes that increase the risk for breast cancer, such as BRCA1 and BRCA2 [2]. Although a number of germ-line variants have been linked to increased risk of breast cancer [2], less is known about the prognostic significance of hereditary variants.

ErbB4 is a member of the EGF receptor (EGFR) subfamily of receptor tyrosine kinases including EGFR (ErbB1), ErbB2 (HER2, neu), ErbB3 (HER3), and ErbB4 (HER4) [3]. Despite of active research on ErbB4 biology in normal mammary tissue and breast cancer, significance of ErbB4 for breast carcinogenesis is still poorly understood. ErbB4 expression is typically associated with ER- and PR-positivity, ErbB2 receptor-negativity, well-differentiated phenotype and favorable outcome [4]–[7]. On the other hand, ErbB4 overexpression has been associated with shorter relapse-free survival in early, node-negative tumors [8] and with decreased survival in patients with node-positive tumors [9]. Treatment with ErbB4-targeted monoclonal antibody suppresses the growth of breast cancer cells [10], suggesting a possible oncogenic role of ErbB4 in breast cancer.

The role of ERBB4 gene variation in breast cancer has been less extensively studied. We have previously shown that a single nucleotide polymorphism (SNP) −782 G>T in the promoter region of ERBB4 gene is a novel risk variant for breast cancer in German population [11]. Two recent studies have also discovered variants of ERBB4 gene that are associated with increased risk for breast cancer. ERBB4 SNP rs13393577 was implicated as a new risk variant in a genome-wide association study (GWAS) conducted in Korean population [12], and three ERBB4 risk variants (rs905883, rs7564590, and rs7558615) were identified in a family-based GWAS in patients of the Framingham heart study [13]. However, no studies have addressed the possible prognostic or predictive value of ERBB4 variants.

According to the Catalogue Of Somatic Mutations In Cancer (COSMIC), somatic ERBB4 mutations in breast cancer are rare, as only 1.4% of breast cancers harbor ERBB4 missense mutations (17 out of 1200 patients) [14]. Although the functional consequences of ERBB4 breast cancer mutations have not been studied, one ERBB4 kinase domain mutation (E872K) initially found in breast cancer [15] has later been shown to be functionally active in metastatic melanoma [16].

Here, we analyzed the frequencies and prognostic value of two ERBB4 promoter variants, −782G>T and −815A>T in a large phase III clinical trial data set of high-risk early breast cancer patients (n = 1010). Frequency of nine specific ERBB4 kinase domain mutations [15] was also analyzed. The results indicate that the ERBB4 variant −815A>T was significantly associated with poor distant disease-free survival, indicating for the first time a possible prognostic significance for a genetic variant of ERBB4 in cancer. The frequency of the analyzed ERBB4 kinase domain mutations was low (0.2%). However, both somatic mutations had previously been shown to be functional and promote cancer cell growth in vitro [16], [17], suggesting a presence of rare but oncogenic ERBB4 mutations in breast cancer.

Materials and Methods

Results

Associations of ERBB4 variants with clinicopathological features and ErbB4 protein expression

When the ERBB4 promoter region SNP status was compared with clinicopathological characteristics, the −782G>T variant was associated with well-differentiated cancer (P = 0.018; Table 1). Neither of the SNPs was significantly associated with primary tumor diameter, axilliary nodal status, histology, tumor grade, ER or PR expression, or ERBB2 amplification (Table 1). Sixteen (69.6%) out of the 23 cancers with the −782G>T variant were ER-positive and ERBB2-negative in immunohistochemical stainings (an approximation for the luminal A biological subtype) as compared with 548 (62.8%) out of the 872 cancers that did not harbor this variant (P = 0.510; one of the seven remaining −782G>T cases was ER+/ERBB2+, three ER−/ERBB2+, and three ER−/ERBB2−). Nine (75.0%) of the 12 cancers with the −815A>T variant were ER+ and ERBB2− (the remaining three were ER–/ERBB2–) as compared with 553 (63.2%) out of the 874 cancers that did not harbor −815A>T (P = 0.551). All 8 tumors with the −815A>T variant stained by immunohistochemistry for the p53 protein expression stained negative, whereas 4 (20.0%) out of the 20 cases with the −782G>T variant stained positively for p53 [20].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Clinicopathological features of patients harboring ERBB4 variants −782G>T or −815A>T.

https://doi.org/10.1371/journal.pone.0102388.t001

To address whether the two SNPs regulated ErbB4 expression levels in primary tumors, tumor sections from all the 1010 patients were immunohistochemically stained with a monoclonal antibody recognizing the C-terminus of ErbB4 (HFR-1). However, no associations were found between ERBB4 SNP status and ErbB4 staining intensity (Table 1). Also, no statistically significant associations were found when cytoplasmic and nuclear ErbB4 staining intensities were separately scored and compared with the ERBB4 SNP status (data not shown). ErbB4 protein expression did not correlate with patient survival (P = 0.826, n = 926), but was strongly associated with ER-positivity (P = 0.003) (Supplementary table 1).

Prognostic significance of ERBB4 variants

The prognostic significance of the ERBB4 variants was assessed by analyzing the association of ERBB4 SNP status with distant disease-free survival (DDFS). No association between ERBB4 −782G>T and DDFS was found (Figure 1A). In contrast, the ERBB4 variant −815A>T was significantly associated with poor prognosis (HR = 2.86 [95% CI 1.15-6-67], P = 0.017; Figure 1B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Prognostic associations of the ERBB4 variants.

Kaplan-Meier plots of distant disease-free survival of patients harboring wild-type ERBB4 or ERBB4 variants −782G>T (A) or −815A>T (B).

https://doi.org/10.1371/journal.pone.0102388.g001

Discussion

Clinical studies on association of ErbB4 expression with breast cancer patient survival are contradictory [4]–[9], [21], despite in vitro as well as in vivo mouse xenograft data suggesting an oncogenic role for ErbB4 in breast cancer [10], [22], [23]. However, the prognostic or predictive role of germ-line or somatic ERBB4 mutations in breast cancer has not been addressed. Here we analyzed the frequencies and prognostic significance of two ERBB4 genetic variants, −782G>T and −815A>T [11] in a cohort of 1010 patients with high-risk early breast cancer. The frequencies of nine specific ERBB4 kinase domain mutations [15] was also investigated.

The frequencies of the two ERBB4 variants were 2.5% (23 out of 936 patients) and 1.3% (12 out of 932 patients) for 782G>T and −815A>T, respectively. In our previous study using samples from German GENICA breast cancer colletion, the frequencies were 5.3% and 1% for 782G>T and −815A>T, respectively [11]. The variant −815A>T was significantly associated with poor prognosis. Interestingly, the variant −782G>T, which was implicated as a risk factor for breast cancer in our previous study [11], was not associated with distant disease-free survival, but with well-differentiated cancer. These data suggest that the heterozygous genotype ERBB4 −815A/T could be a prognostic marker in high-risk early breast cancer. This is the first indication of prognostic significance for a genetic variant of ERBB4 in cancer. However, these findings should be confirmed in an independent large patient cohort. The association between the ERBB4 −815A/T polymorphism and clinical outcome serves the hypothesis that ErbB4-targeted therapy could be beneficial for a subgroup of breast cancer patients in the adjuvant setting.

Immunohistochemical analysis of ErbB4 protein expression levels in the tumors demonstrated that neither of the ERBB4 variants induced significant changes in ErbB4 expression or subcellular localization in the primary tumors. Total ErbB4 expression also did not associate with DDFS of the patients, but correlated with ER-positivity. This is in accordance with reports associating ErbB4 protein expression with markers of favorable prognosis [4]–[7].

Our analysis of specific ERBB4 kinase domain mutations revealed two patients harboring somatic ErbB4 mutations G802dup and E872K, respectively. Interestingly, E872K is a mutation initially found in breast cancer [15] that was later also detected in melanoma [16]. The other mutation, G802dup, has previously been reported in non-small cell lung cancer [15]. Both mutations have been shown to be functional and promote cancer cell/tumor growth in vitro [16], [17], suggesting the presence of rare but potentially oncogenic ERBB4 mutations in breast cancer. Although the observed ERBB4 mutation frequency (0.2%) is low, it corresponds to the frequency of ERBB2 kinase domain mutations (0.5%) in the same patient cohort [24]. Rare ERBB2 kinase domain mutations have recently been suggested to serve as predictive markers for ErbB2-targeted therapy in the absence of ERBB2 amplification [25], indicating that rare mutations may have clinical relevance in high-incidence cancers such as breast cancer.

Taken together, this study presents a genetic ERBB4 variant as a novel prognostic marker in high-risk early breast cancer and indicates the presence of rare but potentially oncogenic ERBB4 mutations in breast cancer.

Supporting Information

Acknowledgments

We thank Maria Tuominen and Minna Santanen for excellent technical assistance.