Efficacy of an RNA-based multigene assay with core needle biopsy samples for risk evaluation in hormone-positive early breast cancer

Breast cancer is a major health problem in women worldwide. It is the most frequent malignancy in women, with nearly 1.4 million new cases and 0.5 million breast cancer-related deaths [1]. Moreover, the incidence of breast cancer has been increasing gradually, in South Korea, with more than 20% increase in 2012 compared to 2008 [2]. Among newly diagnosed breast cancer patients, about 80% of have hormone-positive breast cancer with early-staged tumors, and can therefore expect a low risk of recurrence, which means that they may not need adjuvant chemotherapy or radiotherapy [3]. Commercialized multigene kits, such as Oncotype DX®, EndoPredict®, Mammaprint®, have been approved to estimate the risk of local relapse or distant metastasis [4‐6].

Gene expression profiling, which can be used to guide treatment, has emphasized the importance of tumor proliferation as a prognostic factor for hormone receptor-positive breast cancer [7, 8]. Numerous studies have examined multigene signatures in order to investigate whether the major genes related to tumor proliferation and apoptosis are associated with prognosis [9‐13]. This work is ongoing; indeed, the companies which produce commercial multigene kits are still searching for additional key factors with greater relevance to breast cancer. Meanwhile, a 9-gene test, GenesWell BCT (Gencurix Inc., Seoul, South Korea), developed domestically in South Korea, has been granted limited approval for estimating risk of distant metastasis within the first 10 years of early, hormone receptor-positive, HER2-negative breast cancer with a prognostic risk score [14]. The present study focused on this kit.

To be useful for clinical practice, gene expression profiling should be available before induction of treatment [15]. However, since multigene kits are usually developed with surgical specimens, and commercial multigene tests have not been extensively studied with needle biopsy samples, they are not comprehensively validated for such use [16‐18]. Even if the core needle biopsy (CNB) shows highly accurate results in determining estrogen receptor (ER), progesterone receptor (PR), and HER2 status [19‐21], there is a degree of uncertainty in predicting prognosis using multigene test kits with needle biopsy samples prior to initial treatment.

In the present study, we compared the degree of conformity between gene-expression signatures from CNB and surgical specimen samples from the same patient. We used a newly developed multigene test kit (GenesWell BCT) to evaluate the reliability of gene-expression profiling from needle biopsy for determining treatment plan in early, hormone-positive breast cancer prior to surgery.

Although the incidence of breast cancer is increasing, its mortality rate is declining as a result of more widespread screening allowing earlier diagnosis, and advances in treatment [23, 24]. However, the higher long-term survival rate of breast cancer may increase the probability of locoregional recurrence or distant metastasis. The treatment options for breast cancer include not only surgery but also other adjuvant treatments, including invasive treatment such as chemotherapy or radiotherapy, and less invasive treatment such as hormone or target therapy. Although the invasive treatments are unnecessary for several subgroups of breast cancer which show better prognosis, it is hard to distinguish those subgroups with conventional diagnostic methods [25, 26].

Genomic profiling technologies have become an effective diagnostic method to predict clinical course of breast cancer and are helpful for making decisions relating to the tailoring of its treatment cancer [15, 21, 22, 27]. Based on multigene assays, patients with better prognosis may avoid invasive treatments such as chemotherapy or radiotherapy and their associated side effects. Several commercialized multigene predictors aimed at providing genetic information to enhance prognostic medical decisions in breast cancer have been developed recently cancer [28, 29]. One example is the microarray-based 70-gene assay, MammaPrint®, launched in 2002 by the Netherlands Cancer Institute in Amsterdam. Another is the a 21-gene multiplex prognostic and predictive RT-PCR assay oncotype DX™, which was introduced in 2004 with the Trial Assigning Individualized Options for Treatment (TAILORx) sponsored by the National Cancer Institute. Later, in 2011, the first RNA-based prognostic test, EndoPredict®, was introduced to predict the distant recurrence of ER-positive, HER2-negative breast cancer [4, 28]. These commercialized multigene assay kits were developed using FFPE or freshly frozen tissue obtained from surgery. Therefore, their development was limited by a lack of genetic information obtained prior to treatment. To overcome this limitation, studies have been conducted to confirm a satisfactory level of correlation between results obtained from needle biopsy tissue and surgical specimens [15, 19‐21].

Adding further complexity, previous gene expression profiling studies have shown discordant results between different ethnicities [30‐32]. GenesWell BCT was developed using six prognostic genes including five proliferative genes (UBE2C, TOP2A, RRM2, FOXM1, and MKI67), one gene involved in the immune system (BTN3A2) and three reference genes (CTBP1, CUL1, UBQLN1). These genes were selected from public microarray gene expression data, and the reliability of the BCT score produced by GenesWell BCT was validated in an independent cohort of Koreans [22].

Because this product was developed especially for ER+, HER2- breast cancer patients, who show little benefit from chemotherapy, chemotherapy can be omitted for this group when the result indicates low risk. However, the genetic information should be obtained before treatment is started.

Recently, most breast cancer is detected by screening image and diagnosed at an early stage [33, 34]. To confirm diagnosis, ultrasound-guided needle biopsy is more popular than excision. Usually three or four cores are obtained using a needle biopsy kit and tumor characteristics are analyzed with immunohistochemical staining as well as examined visually for presence of cancer cells. However, breast cancer is known to be particularly heterogeneous within the tumor, and results can differ between operators. Furthermore, 10–20% of immunohistochemical staining results disagree between needle biopsy samples and surgical specimens in assessment of molecular subtype [17, 35]. Therefore, when the breast cancer is multifocal lesion or containing heterogenous materials, including microcalcifications, the tissue concordance would be decreased between needle biopsy samples and surgical specimens.

In the present study, we investigated whether the GenesWell BCT multigene test is suitable for CNB samples in addition to surgical specimens. The samples were obtained from areas in which at least 30% of total tissue volume was tumorous, if possible. The overall concordance rate that we observed between CNB and surgical specimens was 83.9%, which was similar or lower than other commercialized multigene kits. However, when the breast tumor showed an imaging finding of solitary nodule without microcalcification, the concordance rates were higher as 92.0 and 95.8%, respectively.

The current study aimed at validating the use of the GenesWell BCT kit to improve prognostic prediction in hormone-positive breast cancer, in order to establish treatment plans prior to initial treatment. With this aim in mind, two limitations should be noted. First, the small number of paired samples cautions strong conclusions, in particular inference to other subtypes of breast cancer. Even if we conducted the study with thirty-one paired samples of hormone positive breast cancer, it is still a small study. Therefore, the further study would be necessary to validate the results and provide the stronger evidences with reproduction of results. However, it is notable that in our thirty-one paired samples of hormone-positive breast cancer, we observed a very high concordance rate between needle biopsy samples and surgical specimens when the breast cancer is not multifocal, without containing microcalcifications. The second limitation relates to the oncologic conditions of gene profiling using GenesWell BCT. Although the actual BCT scores were developed with pathologic tumor size and nodal status, in the present study we used clinical tumor size and nodal status for obtaining BCT scores from biopsy samples. However, based on preoperative breast magnetic resonance (MR) images showing quite a high concordance rate between clinical and pathologic tumor size in early breast cancer [36, 37], it is reasonable to apply clinical tumor size for assessing BCT score before treatment. Because we also evaluated tumor size with preoperative breast MR, the clinical and pathologic staging we recorded were identical in most cases. Although there was one case in which sentinel lymph node biopsy was omitted in order to shorten the operative time because of the patient’s severe underlying disease, she has not experienced locoregional recurrence or distant metastasis during the follow-up period. Therefore, the authors regarded this case as clinically and pathologically N0 stage.

The correlations of risk classification obtained from the GenesWell BCT multigene kit between CNB samples and surgical specimens in hormone positive breast cancer was relatively low, but not much inferior to other proven commercialized genetic profiling products. However, the multigene tests were performed in cases of solitary breast cancer without microcalcification, the concordance rate was much higher. This data supports the possibility of using the risk scores obtained to help establish a treatment plan before starting initial breast cancer treatment, especially for a solitary nodule of breast cancer without microcalcifications.