A quintet of studies presented at the American Society of Clinical Oncology annual meeting last week, could help support NanoString Technologies' diagnostic ambitions.
During the meeting, members of the Bioclassifier consortium, which comprises cancer experts from four leading research institutions, and their collaborators presented findings that demonstrated how a gene-expression signature called PAM50 can sub-classify and yield a prognostic score based on an individual’s breast tumor.
Seattle-based NanoString announced in December 2010 that it had secured from Bioclassifier an exclusive worldwide license for the PAM50 gene signature to develop breast cancer-related in vitro diagnostic and research products for its nCounter Analysis System.
The company subsequently made the signature available for research purposes with an eye toward submitting the test, called the Breast Cancer Intrinsic Subtyping Assay, for US Food and Drug Administration approval within the following few years (BAN 12/7/2010).
"These studies have extended the body of evidence demonstrating clinical validity of the PAM50 gene signature," CEO Brad Gray told BioArray News this week. Furthermore, he said, the studies "strongly suggest that NanoString will be successful in demonstrating the clinical utility" of the BCIS test, which runs on the nCounter.
Additionally, he said, the company has experienced a "groundswell of interest" from breast cancer researchers who are interested in accessing the assay for their own projects.
"These researchers are eager to collaborate with NanoString and incorporate the assay into their research activities," Gray said, without naming any of the researchers. "In the months ahead, we will begin collaborating with several of these researchers to further demonstrate the clinical utility of the PAM50 gene signature."
Gray added that while "some of these studies will be used to support future regulatory submissions," NanoString does not expect to get a regulatory green light to develop a clinical version of the assay "any time soon."
Bioclassifier is a partnership of four breast cancer experts: Matthew Ellis, director of the breast cancer program at the Washington University School of Medicine in St. Louis, Charles Perou of the University of North Carolina at Chapel Hil, Torsten Nielsen at the British Columbia Cancer Agency, and Philip Bernard of the University of Utah.
Using arrays, the researchers created and validated the PAM50 gene signature, which essentially is a panel of 50 genes that can subtype breast cancer samples into luminal A, luminal B, HER2-enriched, basal-like, and normal-like. It can also provide a continuous risk-of-recurrence score by comparing individual samples with prototypic subtypes.
As Ellis told BioArray News last December, Bioclassifier opted to license the assay to NanoString for a number of reasons (BAN 12/7/2010).
"The problem with microarrays is the manufacturing," Ellis said at the time. "It would require custom chips, but most custom platforms rely on high-quality RNA. We wanted an assay that could work on formalin-fixed, paraffin-embedded tissue that was a decade old."
NanoString says its nCounter is designed to enable researchers to measure gene expression in a multiplexed fashion using color-coded molecular barcodes. It also uses single-molecule imaging to detect and count hundreds of unique transcripts in a single reaction.
While all assays that run on the system are intended for research purposes, at the time the signature was licensed Gray said the firm's long-term vision is to "deliver a series of gene-expression assays for solid tumor oncology to hospital and pathology labs."
[pagebreak]
The Five Studies
The five ASCO studies were performed by Bioclassifier members and external collaborators.
The University of Utah's Bernard, a co-author on all five studies, said in a statement that each project suggested the PAM50 signature "reproducibly identified prognostic and predictive breast cancer subtypes across multiple cohorts, platforms, and methods of procurement."
Although additional validation trials with other teams are ongoing, Bernard said Bioclassifier's data show that the PAM50 assay "provides independent prognostic and predictive information over the standard of care for risk stratification and treatment decision-making in breast cancer."
The first ASCO study, a head-to-head comparison, was called "Using the PAM50 breast cancer intrinsic classifier to assess risk in ER+ breast cancers: A direct comparison to Oncotype DX." In it, researchers used the PAM50 Breast Cancer Intrinsic Classifier to re-analyze151 estrogen receptor-positive breast cancer samples that Genomic Health originally tested on its Oncotype Dx for the University of Texas MD Anderson Cancer Center.
After quality control, the team identified 119 invasive breast cancers, or 78 percent of the total, that both platforms successfully assayed. The presenters determined that while the PAM50 Breast Cancer Intrinsic Classifier and Oncotype Dx use different gene sets and algorithms, there is a "very large and significant overlap in the ability of these clinical tests to determine risk in ER+ breast cancers."
The second study, which was randomized, was entitled "The responsiveness of intrinsic subtypes to adjuvant anthracyclines versus nonanthracyclines in NCIC.CTG MA.5 randomized trial."
Using PAM50 to see how individuals respond to adjuvant anthracycline chemotherapy, the authors determined that HER2-E assignment "strongly predicted" anthracycline sensitivity, and argued that the benefit of anthracyclines is "greatest in tumors that are both clinically Her2+ and Her2-E."
The third study was called "Concordance among gene-expression-based predictors for ER-positive breast cancer treated with adjuvant tamoxifen." Researchers evaluated the prognostic ability and biologic significance of several independent gene-expression signatures in patients with ER+ tumors treated only with adjuvant tamoxifen.
Using four public microarray data sets and added clinical data, they evaluated Oncotype Dx, Agendia's MammaPrint, PAM50, and other signatures. They determined that, from a clinical perspective, the signatures should "specifically aid in identifying patients with node-negative luminal A breast cancers that might be considered for adjuvant endocrine therapies alone."
The fourth study was entitled "PAM50 breast cancer intrinsic classifier: Clinical validation of a multianalyte laboratory developed test." In it, the presenters showed the clinical validation of the PAM50 signature in 171 breast samples, comparing the results with previously published microarray data.
The cross-validation of the clinical subtype predictor ultimately showed 91.6-percent concordance, the presenters reported.
The fifth study was entitled "Determining agreement between immunohistochemistry and RT-qPCR for standard biomarkers in breast cancer: Validation on GEICAM 9906 clinical trial."
The presenters noted that the common practice for comparing different technologies has been to use receiver-operator characteristic curves to measure the same biomarker to establish cut-offs, and to maximize sensitivity and specificity. The researchers introduced a new method for establishing RT-qPCR cut-off points using gene expression across biologic subtypes and compared it to standard methods.
Specifically, the researchers used the PAM50 gene-expression signature to establish an RT-qPCR training set to identify biologic subtypes of breast cancer in 155 invasive cases and 16 controls.
They found that calling cut-off points based on RT-qPCR expression across subtypes is "reproducible across datasets and has good agreement with expression by immunohistochemistry for clinically used biomarkers."
Have topics you'd like to see covered in BioArray News? Contact the editor at jpetrone [at] genomeweb [.] com