By Turna Ray
Researchers presented data at a scientific conference last week suggesting that NanoString's PAM50 prognostic breast cancer recurrence test may provide more actionable information about patients' risk of recurrence than Genomic Health's Oncotype DX test.
Based on the study results presented at the San Antonio Breast Cancer Symposium, NanoString believes that the PAM50 test could be an effective second-generation diagnostic for assessing whether breast cancer patients will experience recurrence and could provide advantages over Oncotype DX, currently the market leader in this space.
"Oncotype DX is a very good test developed by a pioneering company. It has helped thousands of breast cancer patients make tough decisions about their treatment," NanoString CEO Brad Gray told PGx Reporter this week. "However, we believe that a second-generation test can be even better."
At SABCS, NanoString presented data from a study in which researchers led by Mitch Dowsett of the Royal Marsden Hospital analyzed samples from more than 1,000 patients in TransATAC, a translational study group using tissue and data from the Arimidex, Tamoxifen, Alone or in Combination, or ATAC, trial.
The researchers found that the PAM50 prognostic score assigned 21 percent fewer estrogen-receptor-positive, node-negative breast cancer patients with estimates of recurrence risk that put them in the intermediate risk group than did Oncotype DX, and deemed a portion of these patients to be at high risk of recurrence. "This means that PAM50 provided ... more patients [with] actionable information about their risk than did Oncotype DX," Gray said.
PAM50 is an RT-PCR test that measures 50 classifier genes and five control genes and categorizes patients into five intrinsic breast cancer subtypes that confer prognostic information: luminal A, luminal B, HER2-enriched, basal-like, and normal-like.
NanoString is running the PAM50 test on its nCounter Analysis System, and plans to gain approval for the test from the US Food and Drug Administration, though it has not yet disclosed a timeframe for doing so. Currently, the nCounter platform and PAM50 assay are available for research use only. In January, ARUP Laboratories of Salt Lake City launched a laboratory-developed version of the test.
Meanwhile, Oncotype DX, a 21-gene RT-PCR assay, has been on the market as an LDT since 2004. The gene-expression test provides a score between one and 100 indicating the recurrence risk over 10 years for women with HER2-negative, estrogen receptor-positive, early-stage breast cancer that hasn't spread to the lymph nodes. Studies have shown that post-menopausal women who have hormone receptor-positive breast cancer that has spread to the lymph nodes may also derive benefit from testing with Oncotype DX.
An Oncotype DX recurrence score below 18 indicates that patients will respond well to the hormonal therapy tamoxifen, but will gain little benefit from treatment with chemotherapy. Those who receive a recurrence score higher than 31 will likely benefit from chemotherapy in addition to tamoxifen treatment. For patients in the intermediate recurrence risk category — between 18 and 31 — it is unclear whether chemotherapy is necessary and whether patients should be treated with tamoxifen alone.
Unlike PAM50, Oncotype DX does not classify patients' recurrence risk by intrinsic subtypes. Furthermore, while the recurrence score is associated with whether patients are more likely to derive benefit from chemotherapy treatment, Oncotype DX does not predict whether patients are likely to respond to any particular type of chemotherapy.
Although the current TransATAC analysis on PAM50 doesn't validate the test's ability to predict whether patients will derive benefit from specific chemotherapies, Gray believes that by placing more patients in the high- and low-recurrence risk categories, PAM50 will ultimately enable physicians to make better personalized treatment decisions.
Genomic Health's Chief Medical Officer Steve Shak, meanwhile, believes the use of intrinsic subtyping to gauge breast cancer recurrence and treatment benefit risk is still very much in the investigational phase. Furthermore, he told PGx Reporter that Genomic Health is conducting research to identify genes that can predict whether patients will benefit from specific treatments. Such markers will likely be incorporated into updated versions of the Oncotype DX test.
Wall Street reaction to the PAM50 TransATAC data was positive. Recognizing the growing significance of intrinsic subtyping in the breast cancer community, Tycho Peterson of JP Morgan wrote in a research note last week that "PAM50 does look like the first real competitor for Oncotype DX in breast cancer, though its launch is likely some time off and it is too early to determine how much [Oncotype DX's] market share could be at risk."
Study Design
In their analysis of more than 1,000 samples from the TransATAC study, Dowsett and colleagues assessed whether the PAM50 test provided prognostic information over clinical factors in gauging disease recurrence in patients treated with the aromatase inhibitor anastrozole or tamoxifen. The secondary outcome measures of the study were to investigate the relationship between the patients’ intrinsic subtype and their 10-year risk of distant recurrence. The researchers also compared the PAM50 assay's ability to provide information about breast cancer prognosis to that of Oncotype DX and the so-called IHC4 test developed by Dowsett.
IHC4 is an immunohistochemistry test that derives a recurrence score based on quantifying ER, progesterone receptor, Ki67, and HER2. In an earlier analysis of the TransATAC samples, the IHC4 test was found to provide prognostic information on par with Oncotype DX. In that paper, published in the Journal of Clinical Oncology in October, Cuzick et al. suggested that additional studies were needed to determine the "general applicability of the IHC4 score."
Even before the IHC4 analysis, the TransATAC samples were also used to validate the ability of Oncotype DX to gauge the risk of distant recurrence in node-negative and node-positive post-menopausal breast cancer patients.
For the latest TransATAC study comparing PAM50, Oncotype DX, and IHC4, the researchers ran the PAM50 test on the same samples previously analyzed by Oncotype DX. "The RNA that was run in our study was the exact same RNA that was run in the Genomic Health study, so there's no chance of bias there," Gray said.
The RNA used in the PAM50 TransATAC analysis was extracted by Genomic Health, not by NanoString. Still, Gray said he was "confident that the results presented [at SABCS] are representative of the performance of our PAM50-based assay" since NanoString has had consistent results when it has tested RNA extracted from formalin-fixed, paraffin-embedded tissue using various extraction protocols with the PAM50 assay.
Comparing to Clinical Measures, Other Tests
To determine whether PAM50 added prognostic information about distant recurrence-free survival over clinical measures — which factored in patients' nodal status, tumor grade, size tumor, age, and treatment with anastrozole or tamoxifen — the researchers conducted statistical analysis applying the so-called likelihood ratio test.
In total evaluable patients, the measure of prognostic information calculated by the likelihood ratio test was 33.9 points, which led the researchers to conclude that the PAM50 test provides information on disease recurrence beyond clinical measures at statistically significant levels. The additional prognostic data the PAM50 test provided over clinical factors was 24.6 points in node-negative patients, 9.3 points in node-positive patients, and 28 points in HER-negative patients by the likelihood ratio. All these measures were statistically significant.
This analysis shows that "PAM50 provides prognostic information that you cannot get with clinical pathological variables alone," Gray said.
Next, Dowsett and colleagues compared the prognostic information provided by PAM50 and Oncotype DX in node-positive and node-negative patients when clinical measures were factored in. They also compared the prognostic data gleaned from the two tests in node-negative patients when clinical factors were included and excluded.
"The results were the same in all [these] analyses: PAM50 added a statistically significant amount of prognostic information to Oncotype DX, but Oncotype DX did not add a statistically significant amount of prognostic information to PAM50," Gray said.
For example, researchers tested how much prognostic information was gathered when the PAM50 test was performed first on node-negative women and then analyzed how much new prognostic data was garnered when Oncotype DX test was administered. Then they reversed the order, considering the amount of information provided when Oncotype DX was performed first, followed by PAM50.
When PAM50 was performed first, study investigators calculated 60.9 points of prognostic information, while follow-on testing with Oncotype DX provided 1.4 points of information not gathered with initial PAM50 testing. "The researchers asked whether that 1.4 points of additional prognostic information from Oncotype DX was statistically significant," Gray said, "and the answer was no."
In the reverse scenario, Oncotype DX yielded 28.2 points of information, and follow-on testing with PAM50 contributed 31.1 points of additional prognostic data. This information gleaned from follow-on testing with PAM50 was found to be statistically significant.
"This is the analysis that allows you to conclude definitely that in this population PAM50 was yielding more information about disease prognosis than Oncotype DX," Gray said. "The Oncotype DX recurrence score adds no information to the PAM50 prognostic score but PAM50 adds information to Oncotype DX."
Dowsett and colleagues similarly compared the prognostic data provided by PAM50 and IHC4 in node-positive and node-negative patients for whom clinical classifiers were known, and found that when PAM50 was performed first it provided 19.1 points of information and follow-on testing with IHC4 yielded 9.2 points of information. When IHC4 was performed first, it provided 22.5 points of prognostic data and then testing with PAM50 added 5.8 points of information.
However, Gray noted that because IHC4's algorithm was developed with samples from TransATAC, the statistical analysis provided an "unadjusted score" that is not an "apple-to-apples comparison with [Oncotype DX's] recurrence score or [PAM50's] prognostic score" since those tests were developed on different sample sets.
The Importance of Subtyping
The study investigators also attempted to compare recurrence risk predictions with the PAM50 test to recurrence risk scores from Oncotype DX.
Among node-negative patients, when clinical factors were not considered, the PAM50 test and Oncotype DX identified similar numbers of patients at low risk of disease recurrence — 428 patients and 434 patients, respectively.
PAM50 assigned more women to the high-risk group and fewer women to the intermediate-risk group than did Oncotype DX, however. The PAM50 test found 192 patients to have intermediate risk, while Oncotype DX identified 243 patients in this category. In terms of high risk, PAM50 identified 119 patients, whereas Oncotype DX found 62 patients.
"One of the big frustrations with Oncotype DX from physicians and patients is that a relatively large amount of women fall in this intermediate category and don't get a definitive, helpful result back from Oncotype DX because their risk is not low enough to forgo chemotherapy, nor is it high enough that they feel confident that they need chemotherapy," Gray said. PAM50's ability to place fewer women in the intermediate category is "helpful because … women who are getting helpful results are either high or low [risk]."
[ pagebreak ]
The study investigators concluded that the PAM50 test gave 74 percent out of 740 node-negative women a high- or low-risk score, while Oncotype DX placed 67 percent in one of these two categories. Given this finding, PAM50 is "an improvement" over Oncotype DX, Gray said.
Additionally, study investigators found that node-positive and node-negative patients have different outcomes on endocrine therapy when stratified by tumor subtypes via PAM50. For example, patients with tumors subtyped luminal A had a low risk of recurrence and those with luminal B subtypes had higher risk of recurrence.
"The study found that intrinsic subtype had a statistically significant relationship to prognosis in each patient sub-population tested," Gray said. "This really shows that this intrinsic subytping biology is real, that the subtypes are a powerful way to parse and stratify breast cancer."
An earlier study led by Catherine Kelly of University College Dublin comparing Oncotype DX and PAM50 reported similar results. In that study, researchers assayed 119 breast cancer specimens and found "reasonably good agreement" between the two tests in terms of low and high prognostic risk. However, of the patients that fell in the intermediate risk group by Oncotype DX, approximately half were classified luminal A by PAM50 and considered to have lower risk of recurrence, while 13 percent of the samples in the intermediate risk category were found by PAM50 to be luminal B subtype, or at high risk of recurrence (PGx Reporter 5/18/2011).
Genomic Health's Shak noted that intrinsic subtyping — and how best to use it to gauge disease prognosis and chemotherapy benefit — is an ongoing area of research. "There is a lot of controversy and uncertainty about how to use intrinsic subtypes in clinical practice," Shak said. "Many different groups have different methodologies for defining luminal A and luminal B."
While Gray agreed that there is a need for a single standard for determining intrinsic subtype based on gene expression, he disagreed that there is "significant controversy or confusion" regarding how to use intrinsic subtypes in practice.
The study led by Kelly, as well as the TransATAC analysis by Dowsett and colleagues involving PAM50, add to a growing body of knowledge supporting the use of intrinsic subtyping in breast cancer. Earlier this year, an expert panel at the St. Gallen International Breast Cancer Conference issued recommendations focusing on how molecular subtyping may be used to guide treatment strategies.
"It is no longer tenable to consider breast cancer as a single disease. Subtypes can be defined by genetic array testing or approximations to this classification using immunohistochemistry," the St. Gallen's expert panel wrote in recommendations published in the Annals of Oncology. "These subtypes have different epidemiological risk factors, different natural histories, and different responses to systemic and local therapies."
Agendia, which markets MammaPrint, a 70-gene microarray-based breast cancer recurrence test, and BluePrint, an 80-gene breast cancer subtyping test, reported data at SABCS from a study that found that these two tests can "detect specific groups … at high risk of recurrence," the these risk classifications "suggest who would have a higher likelihood to benefit from chemotherapy" (see related story, in this issue).
Chemotherapy Benefit
The St. Gallen's panel also recommended the use of Oncotype DX to predict which breast cancer patients should receive chemotherapy. Additionally, Oncotype DX is the only gene expression test recommended for patients with node-negative breast cancer who are estrogen-receptor positive or progesterone-receptor positive in treatment guidelines from the American Society of Clinical Oncology and the National Comprehensive Cancer Network.
Genomic Health does not market Oncotype DX to predict whether patients are likely to respond to a particular type of chemotherapy, but studies are underway that may in the future allow the company to make such a claim.
In an earlier trial, the company used the test to analyze samples from the National Surgical Adjuvant Breast and Bowl Project B20 trial, and found that it "predicted the magnitude of chemotherapy benefit" in patients. In the original B20 study, node-negative breast cancer patients were treated with a chemo regimen containing cyclophosphamide/methotrexate/fluorouracil or just methotrexate/fluorouracil. However, these regimens are no longer readily used in medical practice.
Shak noted that Genomic Health is exploring new genes associated with response to specific chemotherapies that could be added to the Oncotype DX test.
For example, in 2008, the company and collaborators from Sanofi-Aventis and elsewhere presented data at the American Society of Clinical Oncology annual meeting on a study that evaluated whether PR expression and a 371-gene expression signature (which included the 21-gene Oncotype DX recurrence score) could predict benefit from adjuvant doxorubicin plus cyclophosphamide or docetaxel (marketed by Sanofi as Taxotere).
The study, which analyzed 734 samples from the E2197 trial, found that the genomic classifier "strongly predicted" docetaxel benefit in HR-positive patients with an Oncotype DX recurrence score higher than 18. This signature, "if externally validated, might be useful in defining differential benefit in patients with HR-positive disease," the authors concluded.
In addition, the National Cancer Institute is conducting the TAILORx trial to investigate whether node-negative, ER-positive breast cancer patients deemed to be at intermediate risk of recurrence by Oncotype DX will benefit from chemotherapy. The trial has enrolled more than 10,000 patients and is slated for completion in 2015. Also, the Southwest Oncology Group is conducting the RxPONDER trial to study whether node-positive breast cancer patients with low to intermediate risk scores will have good outcomes on chemotherapy.
With regard to the chemopredictive abilities of the PAM50 test, this latest analysis by Dowsett et al. does not validate whether the test can be used to guide treatment with specific chemotherapy treatments. Such an indication will be important for NanoString if it intends to present the PAM50 test as an improvement over Oncotype DX.
"It is important to note … that the TransATAC study design did not allow for evaluation of chemotherapy benefit, which is one of the most important selling points for [Oncotype DX]," JP Morgan's Peterson wrote. "There have only been a very small number of cancer studies with designs that would allow evaluation of chemotherapy benefit, and our understanding is that many of the samples have been used up, so how NanoString chooses to address this issue will be important for investors to watch."
Regulatory and Commercialization Plans
Although the current study yielded favorable results for the PAM50 test, NanoString is planning to conduct a second validation study to confirm its assay’s prognostic performance for inclusion in treatment guidelines and to establish the clinical value of the test for payors. NanoString is also conducting a large-scale analytical validation trial as part of the data it will submit to regulatory authorities.
The design of this second validation study will be similar in design to the TransATAC study, but Gray could not provide additional design details or project a timeframe for when NanoString might submit PAM50 for regulatory approval with the FDA. However, by launching an FDA-approved diagnostic kit for breast cancer recurrence, NanoString is hoping to gain a market advantage over Genomic Health, which must analyze patient samples through a single laboratory since Oncotype DX is an LDT.
"We believe that the most efficient and cost effective way to provide 21st century medicine is through the laboratory infrastructure that already exists in pathology laboratories and medical centers worldwide. Upon regulatory approval, we plan to distribute our breast cancer assay as an in vitro diagnostic product," Gray said. "This is a sharp contrast to Genomic Health, which offers its test only as a service provided by a single laboratory in California … There are many regions of the world in which physicians are unable or unwilling to ship tissue to North America for testing."
According to Genomic Health, Oncotype DX is performed by physicians in 60 countries who have ordered nearly 250,000 Oncotype DX tests to date. The international market comprises around 10 percent of the company's sales and revenue from Oncotype DX. Genomic Health has identified international expansion as a key priority over the coming years.
NanoString hasn't yet announced how it intends to price the PAM50 test. ARUP Labs charges $3,000 for the LDT version. The list price for Oncotype DX, meanwhile, is $4,175. "Given this price, it is highly unlikely that this test and Oncotype DX will both be ordered, and therefore [the two tests] should be seen as competing assays," JP Morgan's Peterson wrote.
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at tray [at] genomeweb [.] com.