Multigene predictors in early-stage breast cancer: moving in or moving out?

Whole genome profiling has greatly expanded our knowledge of the genes and genetic pathways associated with the development and progression of breast cancer [1–4]. As a result, numerous commercialized multigene prognostic and predictive tests have entered the complex and expanding landscape of breast cancer diagnostics (Table 1)[5]. These multigene predictors utilize a variety of test platforms including immunohistochemistry (IHC), FISH, quantitative real-time PCR (RT-PCR) and genomic microarrays and gene chips. Both decentralized reagent sales as diagnostic kits and centralized commercial laboratory assays are now available with integrated proprietary statistical data analysis tools and algorithms designed to calculate the test results [5–7].

Several of the tests are performed on readily available formalin-fixed paraffinembedded (FFPE) samples whereas others require fresh tissue stored in an RNA-preserving solution [5]. Several organizations have sought and achieved regulatory approval for their tests, others have not and offer their tests as ‘homebrew’ assays. Although a number of the multigene assays have been compared against conventional diagnostic and prognostic tests in a variety of multivariate analysis models with the results reported in peer-reviewed, well respected scientific journals [5], several multigene predictor kits that are now on the market have essentially only been tested ‘in house’ and do not have any accompanying medical literature to support their developer’s claims of clinical utility and value.

Molecular classification & grading of breast cancer

The most widely accepted molecular classification of breast cancer is the ‘molecular portrait’ database, which divided the tumors into the luminal, normal, HER2 and basal-like subtypes of invasive breast cancer [1]. This classification of breast cancer into distinct portraits has mostly been validated using other profiling platforms and databases, which has confirmed that the subgroups are heavily driven by the estrogen receptor (ER) and HER2 status and by proliferate activity of the tumor, explaining why, despite having a greater response to chemotherapy, tumors in the basal-like subgroup continue to have a relatively poor prognosis (if they fall into the chemotherapy-insensitive category) reflecting their higher histologic grade and inability to be impacted by hormonal targeted therapies [8,9]. Another important approach to the genomic classification of breast cancer was the development of a series of 97 genes that reclassified the traditional three histologic grades of breast cancer into only two distinct molecular grades [4]. These data revealed that the histologic grade 2 tumors appeared to be similar to the grade 1 or grade 3 tumors in their molecular profiles, but lacked a distinctive profile of their own. One practical ramification of this might be the possibility that if, after surgical removal or biopsy, a grade 3 tumor sits at room temperature for a protracted length of time before it is immersed in formalin fixative, this would allow for the mitotic figures that would have been observed to disappear as the tumor cells were able to complete their cell division. This tumor could then be falsely graded by the pathologist as grade 2, but would cluster with the grade 3 tumors when evaluated at the mRNA expression level. The Genomic Grade Index has been licensed to Ipsogen Inc., (Marseille, France) for commercial development.

Immunohistochemistry-based multigene predictors

There are two IHC-based multigene assays for the prediction of outcome for breast cancer, one of which is in development [10] and the other is now on the market [11].

The ProExBr^™ (Tripath Oncology, Durham, NC, USA) is a five antibody/five separate slide IHC assay that utilizes an image analysis system-based slide scoring system [10]. Overexpression of two or more of these markers (ProExBr score of 2 or higher) has been associated with disease relapse in both lymph node-negative and -positive patient cohorts. The test has not been directly linked to response to a specific therapy. The Mammostrat^™ (Applied Genomics Inc., Huntsville, AL, USA) assay has been fully commercialized and is currently available via centralized testing at The Molecular Profiling Institute (Molecular Profiling Institute, Inc., Phoenix, AZ, USA). This standard IHC format test uses five antibodies with routine slide scoring to divide cases of ER-positive, lymph node-negative tumors treated with tamoxifen alone into low-, moderate- or high-risk groups. In a multivariable analysis study, the calculated risk of recurrence for Mammostrat was independent of stage, grade, and lymph node status [11].

FISH-based predictors

The eXagenBC^™ test (eXagen Diagnostics, Inc., Albuquerque, NM) has been developed to predict breast cancer outcome in node-positive or node-negative patients. Performed on FFPE surgical pathology slides, this assay employs fluorescent-labeled DNA probes to measure the copy numbers of three genes for ER-positive tumors with a proprietary algorithm used to predict the prognostic index [12].

RT-PCR-based multigene predictors

The Oncotype DX^™ test (Genomic Health, Inc., Redwood City, CA, USA) is a 21-gene multiplex RT-PCR assay performed on primary FFPE breast cancer samples by a central laboratory. The original 16 informative genes that calculate the recurrence score (RS) were discovered on archived frozen samples by transcriptional profiling and then converted to the FFPE RT-PCR assay [13]. Oncotype Dx determines the 10-year risk of disease recurrence in ER-positive, lymph node-negative tumors utilizing a continuous variable algorithm and assigning a tripartite RS (≤ 17: low risk; 18–30: intermediate risk; > 30: high risk). Of the multiple pathways assessed by the assay, the proliferation and ER pathways are the most influential on the RS calculation followed by the HER2 pathway. The other 14 informative mRNA levels play their greatest roles in determining the RS in tumors with intermediate ER and Ki-67 mRNA levels. After validation in lymph node-negative patients, the assay has now been studied in lymph node-positive patients with promising results, which has led the developer to expand clinical trials for the test in this setting [14]. To date, Oncotype Dx, which is approved by the California State Licensing Agency for Laboratories, has not been submitted to the US FDA for formal approval. As a centralized ‘homebrew’ assay it is currently exempt from the standard review the FDA requires for diagnostic kits. The original discovery and validation of the test were conducted on archived National Surgical Adjuvant Breast and Bowel Project (NSABP) clinical samples [13] and the test has been successfully marketed without classic prospective validation. According to the Genomic Health website, since January 2004, more than 6000 physicians have ordered the Oncotype Dx test for over 33,000 patients. The test is accepted by a wide variety of third-party payers including the Centers for Medicare and Medicaid Services and is endorsed by the most recent American Society of Clinical Oncology tumor marker guidelines [15]. The TAILORx clinical trial has been launched by the National Cancer Institute to test prospectively whether the Oncotype Dx can successfully guide treatment selection of breast cancer in the adjuvant setting. Known as the Trial Assigning Individualized Options for Treatment (Rx), the accrual of patients began in May 2006 [16]. This trial is being conducted by the North American Breast Cancer Intergroup, which includes all of the major National Cancer Institute-funded cooperative groups in the USA and Canada and is coordinated by the Eastern Cooperative Oncology Group. The TAILORx trial plans to enroll at least 10,000 women with ER- or PR-positive, HER2-negative, lymph node-negative breast cancer at 900 sites in North America. The main goal of the trial is to determine whether ER-positive patients with intermediate RS benefit from chemotherapy or not. Of note is the fact that the RS criteria have been changed for the TAILORx trial with the 18–30 intermediate RS range of the current clinically available assay changed to 11–25 for the trial [17]. Patients with a RS of 10 or less receive hormonal therapy alone, the patients with a RS of 26 or higher receive hormonal and chemotherapy and the patients with a RS between 11 and 25 are randomized into either hormonal therapy alone or to hormonal therapy plus chemotherapy. Accrual to this trial appears to be progressing at a rapid rate, but the trial results will not be known for a number of years and likely not until at least 2013.

Other quantitative RT-PCR assays in this space include the Breast Cancer Two Gene Expression Ratio H/I^™ (Aviara Dx, Inc., Carlsbad, CA, USA) [18], the Celera Metastasis Score^™ (Celera, Inc., Rockville, MD, USA) [19] and the Breast BioClassifier (Associates in Regional and University Pathologists, Salt Lake City, UT, USA) [20]. These assays serve as either stand-alone prognostic tests or predictors of hormonal therapy response. They have not, to date, reached the level of commercial development associated with the Oncotype Dx assay.

Genomic microarray-based multigene predictors

The MammaPrint^™ assay (Agendia BV, Amsterdam, The Netherlands) is the first fully commercialized microarray-based multigene assay for breast cancer. This test has received 510(k) clearance from the FDA and is offered as a prognostic test for women under the age of 61 years with either ER-positive or ER-negative, lymph node-negative breast cancer. The test was also the first assay to be approved by the FDA’s new ^{In Vitro} Diagnostic Multivariate Index Assay (IVDMIA) classification. Unlike Oncotype Dx, this test cannot currently be performed on FFPE tissues and requires freshly prepared tissues collected into an RNA preservative solution. The commercial significance of the requirement for fresh or stored frozen tissue on the general acceptance of the test is not currently known. The test is not yet marketed in the USA. The 70 genes that comprise the MammaPrint assay are focused primarily on proliferation, with additional genes associated with invasion, metastasis, stromal integrity and angiogenesis. The MammaPrint assay is at its best when identifying cases at the extremes of the spectrum of disease outcome: the identification of patients with very good or very poor prognosis. It has not yet been studied if the assay can also predict sensitivity to various treatment modalities. The Microarray In Node-negative Disease may Avoid ChemoTherapy (MINDACT) Trial, is sponsored by the European Organization for Research and Treatment of Cancer (EORTC), involves the assessment of patients in the adjuvant treatment setting by the standard clinicopathologic prognostic factors included on Adjuvant! Online and by the 70-gene MammaPrint assay [21]. If both the traditional and molecular assays predict high-risk status, the patient receives adjuvant cytotoxic chemotherapy and also hormonal therapy if ER-positive. If both assays indicate a low risk, no chemotherapy is given and ER-positive patients are given adjuvant hormonal therapy only. When there is a discordance between the traditional clinicopathologic prognostic factor prediction of risk and MammaPrint prediction of risk, the patients are randomized to receive treatment either based on the genomic or by the clinical prediction results.

Comparison of Oncotype Dx^™ & MammaPrint^™

At first glance, given that the two assays share only one overlapping gene, it would appear that the two assays have little in common in terms of genomic content. However, the priority gene lists are consistent in that three biologic pathways are emphasized in both: proliferation, ER and HER2. In that virtually all newly diagnosed breast cancers in the USA and Europe are fully processed for microscopic examination it provides an ease of use for FFPE-based assays. However, the larger gene number of the MammaPrint assay could conceivably provide an increased opportunity to assess additional pathways and potentially provide additional pharmacogenomic information than the 21-gene Oncotype Dx assay can provide. The MammaPrint test also has a wider indication than Oncotype Dx by including both ER-positive and -negative patients, which also allows for inclusion of a greater number of younger patients. Based on currently published studies, the Oncotype Dx test has been validated as a stand-alone prognostic test and has been interpreted as a predictive test for response to tamoxifen and to the cytoxan, methotrexate and fluorouracil adjuvant chemotherapy regimen (although concurrent with tamoxifen). On the other hand, the MammaPrint assay is validated as a prognostic test only and has not been formally tested as a predictive test for specific endocrine or cytotoxic therapy regimens. Thus, in the current marketplace, it would appear that each test has current competitive advantages and disadvantages. The MammaPrint assay has received 510(k) clearance by the FDA whereas Oncotype Dx has been exempt. The Genomic Health Inc.’s central testing laboratory has been approved by the US Clinical Laboratory Improvement Act to offer the Oncotype Dx test as a homebrew assay. More recently, Oncotype Dx has been designated as ‘recommended for use’ by the American Society of Clinical Oncology Breast Cancer Tumor Markers Update Committee whereas the MammaPrint assay was classified by the group as ‘under investigation’ [15].

Other microarray-based multigene predictors

The Rotterdam Signature, also known as the 76-gene assay, was developed at the Erasmus University Cancer Center in Rotterdam, The Netherlands, and is being commercially developed by the Veridex Corp. (Veridex LLC, Warren, NJ, USA). Sharing no genes in common with either Oncotype Dx or MammaPrint, this stand-alone prognostic test is validated for lymph node-negative patients independent of hormone receptor status [22,23]. The gene list for this assay is heavily weighted towards proliferation genes. This assay requires fresh/frozen extracted mRNA and, similar to MammaPrint, has not been validated for use on FFPE tissues or core biopsies. The Invasiveness Signature^™ (OncoMed Pharmaceuticals, Redwood City, CA, USA) consists of 186 genes and is designed for both node-negative and -positive and ER-negative and -positive breast cancers and also appears to predict prognosis for lung and prostate cancers as well as medulloblastoma [24]. The NuvoSelect^™ (Nuvera Bioscience, Inc., Woburn, MA, USA) assay uses a 30-gene set to predict complete response to preoperative paclitaxel (Taxol^®), 5-fluouracil, doxorubicin (Adriamycin^™) and cyclophosphamide (TFAC) chemotherapy and a 200-gene set to predict tumor response after 5 years of endocrine therapy [25,26]. This program also identified the overexpression of mRNA of the biomarker microtubule-associated protein (MAP)-tau as a major predictor of resistance to the TFAC regimen but at the same time a predictor of increased sensitivity to endocrine therapy among the ER-positive patients [27].

Another microarray-based test for breast cancer is the CytoChrome p450 CYP2D6 Genotyping Test (Roche Diagnostics Inc., Indianapolis, IN, USA). Evidence suggests that, for patients deficient in CYP2D6 who have been diagnosed with ER-positive breast cancers, aromatase inhibitors may be the preferred hormonal therapeutics rather than tamoxifen [28–30]. This assay has been approved by the FDA and, recently, the Laboratory Corporation of America (LCA; Burlington, NC, USA) has announced a partnership with Medco Health Solutions (Medco Health Solutions Inc., Franklin Lakes, NJ, USA), designed to offer CYP2D6 testing on a large scale. A number of smaller laboratories have announced the availability of homebrew assays for this biomarker. It is too early to know whether this approach towards personalizing the selection of hormonal therapy will become widely used.

Summary & conclusions

The rapid adoption of these complex and expensive multigene predictor assays before they have been proven to be fully validated and clinically effective in prospective clinical trials involving large numbers of patients has caused serious concern among clinicians and healthcare economists. In this extremely competitive landscape, concerns continue to arise over the scientific validity, true clinical utility and ultimate cost/benefit ratios of the assays. It remains to be seen whether these new approaches to breast cancer management will hold up over time as more patients are tested. These novel diagnostics have the propensity to be misused including employing the test in the wrong clinical settings and ending up with misleading reassurance regarding the test-driven decisions [31]. Although the prospective clinical trials that are currently underway, including the TAILORx and the MINDACT trials in the USA and Europe respectively, could provide significant answers regarding the clinical value of these multigene predictors, many questions may well remain unanswered for a number of years.

Table 1. Summary of breast cancer multigene classifiers and predictors.

Test	Manufacturer	Method	Sample conditions	Indication	Stand-alone prognostic role established	Guide to specific therapy	Commercial status	US FDA status	Cost	Ref.
Classification
Sorlie-Perou Classifier	Not commercialized	Microarray	Fresh/frozen	All cases of invasive breast cancer	Partially	No	In development	Unknown	Unknown
Arup Bioclassifier	In development	RT-PCR	FFPE	ER positive ER negative LN negative	Yes	No	In development	Submitted for 510(k) clearance	Unknown	[101]
Grading
Sotiriou Molecular Grading	Ipsogen, Inc.	Microarray	Fresh/frozen	All cases of invasive breast cancer	Yes	No	In development	Not submitted		[102]
Prognosis only
Pro-Ex Br^®	Becton-Dickinson/ Tripath Oncology	IHC	Formalin fixed	LN negative LN positive	Yes	No	In development	Not submitted	Unknown	[103]
Mammostrat^®	Applied genomics	IHC	FFPE	ER positive LN negative	No	Yes, tamoxifen	On the market centralized testing	Exempt (Homebrew assay)	Billed as 5 88342 Global CPT codes (approximately US$600)	[104]
eXagenBC^™	Exagen Diagnostics	FISH	FFPE	LN negative ER positive or negative	No	No	In development	Submitted for 510k clearance (4/07)	Billed as a multiplex FISH test (US $700-1000 expected reimbursement)	[105]
MammaPrint^®	Agendia BV	Microarray	Fresh/frozen preserved mRNA	ER positive ER negative LN negative	Yes	No	On the market centralized testing	Approved (510k)	Unknown	[106]
Invasiveness Signature^™	OncoMed Pharmaceuticals	Microarray	Fresh/frozen	ER positive and negative LN positive and negative	Yes	No	In development	Not submitted	Unknown	[107]
Prognosis and response to therapy
Oncotype Dx^®	Genomic Health	RT-PCR	FFPE	ER positive LN negative	Yes	Yes, tamoxifen, adjuvant CMF	On the market centralized testing	Exempt	US$3460	[108]
Two-Gene Ratio	AviaraDx/Quest	RT-PCR	FFPE	ER positive	No	Yes, tamoxifen	In development	Not submitted	US$1400	[109]
Celera Metastasis Test	Celera/Labcorp	RT-PCR	FFPE	ER positive LN negative	No	Yes, tamoxifen	In development	Not submitted	Unknown	[110]
Rotterdam Signature	Veridex	Microarray	Fresh/frozen	ER positive and negative LN positive and negative	Yes	Possible, (tamoxifen)	In development	Not submitted	Unknown	[111]
NuvoSelect (TFAC/ER)	Nuvera Biosciences	Microarray	Fresh/frozen	Neoadjuvant predictor of TFAC response Prediction of response to hormonal therapy	No	Yes, neoadjuvant TFAC, tamoxifen	In development	Not submitted	Unknown	[112]
Roche AmpliChip^® P450 Cytochrome P450 CYP2D6	Roche Diagnostics/Merck Medco/Lab Corp	Microarray	Fresh/frozen	ER positive ER negative LN positive LN negative	No	Yes, tamoxifen CypChip^™	On the market	Approved		[113]

CMF: Cytoxan, methotrexate, 5-fluorouracil; CPT: Current procedural terminology; ER: Estrogen receptor; FFPE: Formalin-fixed paraffin-embedded; IHC: Immunohistochemistry; LN: Lymph node; RT-PCR: Real-time PCR; TFAC: 5-fluorouracil, doxorubicin (Adriamycin^™) and cyclophosphamide.

Financial & competing interests disclosure

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.