A recent comparison of Oncotype DX and PAM50 found that the two genomic tests for breast cancer recurrence do not agree in their classification of intermediate-risk patients, suggesting that there may be room for adjustment when it comes to Oncotype Dx's cutoffs, according to a researcher involved in the project.
Catherine Kelly, a Medical Oncologist at University College in Dublin, Ireland, presented the results of the comparison at the 2011 IMPAKT Breast Cancer Conference this May 5 through 7. Testing the assays on 119 breast cancer specimens, Kelly found that there was "reasonably good agreement" between the two for both high and low prognostic risk, but the tests diverged from each other in the categorization of intermediate-risk patients.
Of those with an intermediate Oncotype DX recurrence score, 51 percent were classified as being a low-risk luminal A type by PAM50, while the other half were split between luminal B and two other types.
"For patients with early-stage breast cancer, particularly estrogen receptor positive, HER2-negative, and lymph node negative, we often have a discussion about the magnitude of benefits of chemotherapy, and sometimes it can be quite uncertain using traditional clinical and pathological variables," Kelly told PGx Reporter.
Several genomic tests have been introduced that help resolve this uncertainty, said Kelly, including Oncotype DX from Genomic Health and PAM50, which was developed by researchers at several academic institutions and is currently offered as a laboratory-developed test by ARUP Laboratories. However, "we are still uncertain how to treat some of the groups identified by these tests," specifically patients with an Oncotype DX intermediate recurrence score," Kelly said.
"If a patient is in the [Oncotype DX] low recurrence score group, we think hormonal therapy is adequate, and if a person is in the higher-risk group we treat that person with chemo followed by hormone therapy. But if a person is stratified to the intermediate-risk score group, we're not certain as to the magnitude of benefit she will derive from chemo," she said.
According to Kelly, the comparison study followed on the heels of earlier research by her group on how Oncotype DX scores patients with stage-one or stage-two ER-positive, HER2-negative, lymph node negative cancer.
"We wanted to see if we perform [PAM50] on a group of tumors already tested with Oncotype DX, how it would stratify the patients," she said. "What would the concordance be between the two tests?"
Oncotype DX is a 21-gene RT-PCR assay commercially available since 2004, which scores women with HER2-negative, node-negative breast cancer on a recurrence risk scale of one to 100.
PAM50 is also a multi-gene RT-PCR test, which 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. In January, ARUP Labs of Salt Lake City, Utah, launched PAM50 as an LDT. Several of Kelly's collaborators in the comparative study are employees of ARUP.
Bioclassifier, a company formed by PAM50's developers to commercialize the test, has also licensed the gene signature to NanoString, which plans to commercialize a PAM50 test for its nCounter multiplex gene expression platform.
In her study, Kelly performed PAM50 on a set of samples from patients who had already been given a risk score by Oncotype DX, and compared the two stratifications.
She reported that all patients with a high recurrence score according to OncotypeDX were classified as luminal B or "basal like" by PAM50, and the majority of low risk score specimens, about 83 percent, were designated luminal A. "This is what you would expect," Kelly said.
In the intermediate-score group, 53 specimens (about half) were classified luminal A, the subtype considered to have a lower risk of recurrence, 13 were classified luminal B, and another nine fell into two other groups, HER2-enriched and normal-like.
"This could potentially be helpful in that there might be further stratification in the Oncotype intermediate recurrence score groups by [PAM50] – it might give you additional information on those patients," said Kelly. "For the patients who had an intermediate recurrence score and were luminal A, that means those patients are probably the ones that are likely to do very well, and endocrine therapy might be adequate."
In an e-mailed response, Phillip Bernard, the medical director of the Solid Tumor Molecular Diagnostics lab at ARUP Laboratories and a co-inventor of PAM50, said his team was "pleased to see that there were no luminal A tumors that were high [recurrence score] and that those with an intermediate [recurrence score] were all in the lower half of that class, suggesting perhaps a better cut-point, which is being sought in the TAILORx trial."
In the TAILORx trial, Genomic Health is evaluating the effect of chemotherapy for those with a midrange recurrence score. The company said that the trial has enrolled more than 10,000 patients.
Bernard added that "the fact that there was such high agreement for ER+ tumors was reassuring for both tests. Subtypes other than luminal A are likely to be higher risk regardless of the Oncotype DX score. A larger study with longer follow-up will help answer this question."
Kelly said the results of the comparison do not suggest that PAM50 is necessarily a more accurate test, even though it broke the intermediate-score group into several classifications.
"In terms of which you'd use in the clinic today, Oncotype has been validated extensively, whereas PAM50 is still undergoing validation," Kelly said.
"All you can say is that it is interesting that PAM50 further stratifies the intermediate-risk group. But ultimately we have to wait on outcome," she said.
Steve Shak, chief medical officer at Genomic Health, told PGx Reporter that he views the study in much the same light.
"This would appear to be an early study," he said. "Clearly, with regard to deciding whether a patient needs chemotherapy or not, there needs to be evidence related to the ability of the test to predict recurrence risk, clinical outcome data, and there needs to be evidence with regard to the prediction of chemotherapy benefit."
He also said that while the two tests did show some congruity in the high- and low-risk scores, he wouldn't call their performance "similar," considering Oncotype's recurrence scores and PAM50's classifications both represent more of a gradient than a series of cutoff points.
Shak added that the TAILORx trial should shed further light on the test's predictive capabilities. "The very large number of patients that are going to be randomized to hormonal therapy versus hormonal plus chemotherapy will allow us to do two things. One is to really evaluate with greater detail, with contemporary regimens, the effect of chemo in that group. And it may be that the effect in the whole group is very low.
"The second thing would be that it might help us to refine how the effect of chemo might depend on the individual score within that range. It's going to address both of those," he said.
He also said Genomic Health is examining other ways to refine recurrence scores to help guide treatments for patients in the intermediate group.
"We've done some work that looks at the development of a refinement of recurrence risk, which is performed by doing a more formal integration of the risk score… but looks also at the contribution of tumor size, and grade, and age. And we have presented at meetings results that suggest that integration of those factors can lead to the identification of more patients at lower risk," he said.
Another study, RxPONDER, is testing Oncotype DX's utility in lymph node-positive patients, examining whether chemotherapy benefits node-positive patients who have low to intermediate Oncotype DX scores. As part of RxPONDER, researchers are also evaluating PAM50.
Mammaprint, Agendia's established test for breast cancer recurrence risk, is not part of the RxPONDER trial, according to Agendia and was not examined in Kelly's comparison. Kelly said her team might have looked at other tests like Mammaprint as well, but decided to focus in this case on OncotypeDX and PAM50.
Have topics you'd like to see covered in Pharmacogenomics Reporter? Contact the editor at mashford [at] genomeweb [.] com.