Skip to main content
Premium Trial:

Request an Annual Quote

One Year Post-TAILORx: Attitudes Shift and Questions Linger in Breast Cancer Recurrence Testing


CHICAGO (Precision Oncology News) – Nearly a year has passed since the presentation and publication of mature results from the TAILORx trial last June, which reiterated the chemopredictive and prognostic ability of Genomic Health's Oncotype DX breast cancer test and clarified how clinicians should interpret the test's mid-range recurrence score results.

Based on the data, clinicians who already embraced Oncotype DX have been able to further cement their confidence in genetics-informed treatment decisions. Genomic Health has also attributed growth in its US and international sales to new adopters that it believes have been persuaded by the trial.

Others don't view the impact of TAILORx as quite so monolithic. Public financial reports from other companies in the space reflect both that the field remains incompletely penetrated — some doctors continue to rely on clinical risk factors alone — and that various commercial tests continue to find their own customers both in the US and abroad.

Importantly, field leaders recognize that there remain unanswered questions about the utility of molecular assays to inform other treatment decisions, for example, to reliably guide the care of patients with lymph node involvement, or inform the best use of extended endocrine therapy regimens.

One of the main impacts Genomic Health has highlighted as emerging from last year's TAILORx publication has been the clarification of the test's ability to predict chemotherapy benefit. Prior to the updated trial readout, clinicians didn't have definitive evidence of whether to give chemotherapy to women with intermediate recurrence scores — between 11 and 25.

The new data showed clearly that women, at least older women, with scores up to 25 have very little to no benefit from added chemo, while those with scores above 25 have what Genomic Health Chief Scientific Officer Steve Shak called a "clear and important benefit."

Shak said in an interview this month that there is no question in his mind that there has been significant impact from TAILORx on use of Oncotype DX not just in the US but in countries around the world over the last year.

"What we know in oncology is that large randomized trials define the standard and define … what should be done in clinical practice," he said.

Shak highlighted the fact that the trial results led directly to updates in professional guidelines, including changes to recommendations by the National Comprehensive Cancer Network (NCCN), which recategorized the Oncotype DX Breast Recurrence Score as a "preferred" test for chemotherapy treatment decision making in patients with node-negative early-stage breast cancer.

Shak added that he hears clinicians speak about the TAILORx results and the use of the test "with a confidence and certainty that just didn't exist before."

Some of that is directly due to the clarification of intermediate scores. "We knew from market research that [this] was a barrier," Emily Faucette, Genomic Health's vice president of corporate communications and investor relations, added.

A breast cancer surgeon at the OhioHealth Physician Group, Deepa Halaharvi said in an email that she is an example of a physician who already viewed Oncotype DX as holding superior evidence for both its prognostic and chemotherapy-predictive impact, choosing it in lieu of competitors even before TAILORx. But, prior to the trial publication last year, she was more unsure of what to do for patients with an intermediate score.

"It's now clear who benefits from chemotherapy," she said.

But alongside doctors like Halaharvi, there are some investigators in the field who disagree that chemoprediction, as demonstrated in TAILORx, is unique to Oncotype DX, or that this is even the most important aspect of breast cancer recurrence assays.

Ivana Sestak, a statistician working at the Queen Mary University of London's Centre for Cancer Prevention, has been involved with numerous trials evaluating molecular breast cancer tests.

"The landscape of genomic assays [for breast cancer] is obviously a very competitive area. For a long time, Genomic Health [was] the only provider, [but] in the last five to 10 years, second-generation tests have been developed," she wrote in an email.

Importantly, Sestak argued, there is strong evidence that some of these newer tests provide better prognostic information than Oncotype. "One important study was our [test] comparison … in TransATAC, which showed that Prosigna and EndoPredict showed superior prognostic power, especially for prediction of late metastasis and in node-positive disease." Prosigna was developed and is sold by NanoString Technologies and EndoPredict is sold by Myriad Genetics.

"Prognostic power of a gene expression test is the most important factor to identify low-risk patients who can safely forgo chemotherapy," Sestak argued. However, she wrote, "It might be important for some clinical situations that clinicians know the expected chemotherapy benefit for high-risk patients.

In contrast to how Genomic Health views the TAILORx trial as defining Oncotype's chemopredictive ability above and beyond other assays, Sestak said that she sees compelling evidence that the same is true for other tests like Agendia's Mammaprint and Myriad's EndoPredict.

Sestak and her coauthors published a study last month in Breast Cancer Research and Treatment in which they showed this for EndoPredict by retrospectively analyzing patient outcomes from five phase III clinical trials and concluding that the test could predict which patients benefit from the addition of chemotherapy.

Genomic Health's Steve Shak argued that the study design of Sestak and colleague's EndoPredict study falls far short of the evidence for chemoprediction in TAILORx. According to Shak, faced with the question of whether to treat with chemotherapy or not, clinicians should be seeking an answer based on randomization.

Because it combined trials of chemotherapy-treated patients with results in endocrine-treated patients from a separate historical control, he argued that the conclusions of the study are "not supported by the data."

"Everyone knows from looking at drug effects that that's [unreliable]," he added. "You can't talk about treatment benefit by looking at historical controls because they might have done better just because they were different."

But, Sestak disagreed, raising the fact that clinicians used Oncotype DX, and did so enthusiastically, prior to definitive evidence from TAILORx, based on its prognostic performance and on the suggestion of chemoprediction from prospective-retrospective studies.

"Without accepting the prospective-retrospective evidence, we wouldn't have had the innovation of gene expression tests for the benefit of thousands of patients with breast cancer," she wrote.

"We now have second-generation tests with an improved prognostic power as shown in TransATAC, [so] the question is whether additional prospective trials with 10-year follow-up are still necessary before bringing innovations into the clinics for the benefit of patients," she added.

"Both TAILORx and our study confirm that separating low-risk from high-risk patients at the 10 percent risk cut-off might be appropriate for clinical decision making in favor for or against chemotherapy. The MINDACT trial (evaluating Agendia's Mammaprint) has also shown in a prospective study that patients at clinical high risk but genomic low risk can safely forego chemotherapy. With the current evidence, it is clear that there are several genomic assays that can predict chemotherapy benefit," Sestak wrote.

Shak and Sestak also disagreed about the differences in how randomized trial data may be viewed in the US and Europe. According to Shak, the growth Genomic Health is seeing in international markets reflects a desire among European oncologists for the kind of randomized evidence TAILORx provided.

As an example, the company highlighted late last year that the German Institute for Quality and Efficiency in Health Care had updated its assessment of breast cancer gene expression profiling tests, concluding that, based on TAILORx, Oncotype DX can support the decision for or against chemotherapy.

Sestak, in contrast, said that she believes the significance of tests is not viewed as resting solely on predictive ability in Europe as much as it is in the US.

According to Sestak, the genesis for her and her colleagues' recent EndoPredict study was the demand from clinicians, mainly from the US, to show a chemopredictive ability for EndoPredict in addition to prognostic prediction.

"They believe that this additional information would improve their clinical decision making, especially in the case of high-risk patients so that they can [weigh] the individual expected benefit from chemotherapy against expected side effects," she wrote.

But in Europe, she said, "It is generally accepted that the relative chemotherapy benefit is approximately 30 percent, independent from risk," she wrote. "This implies low absolute chemotherapy benefit in low-risk patients and a high absolute chemotherapy benefit in high-risk patients."

Mark Robson, medical oncologist and chief of the breast medicine service at Memorial Sloan Kettering, communicated a slightly different take on the question of TAILORx and its impact on the field, echoing Sestak's view that Oncotype DX was "pretty heavily adopted" even before TAILORx, based on its putative predictive ability, but recognizing the importance of the randomized trial.

"There are lots of prognostic assays, and it's challenging to choose between those to be honest, but predicting chemotherapy benefit is really the differentiator," he said in an interview.

"Could other tests do that?  Yes. But they didn't design and execute the big trials to really prove it," he added. "It’s an incredibly important trial, but I suspect that most people comfortable with the assay in that setting were already using it. We'll [have to] find out from Genomic Health's financial results how much" of an impact it had.

Lajos Pusztai, director of breast cancer translational research at the Yale Cancer Center, said much the same. "There were early adopters … but some people were waiting for more data. For them, TAILORx nailed it," he said.

In Pusztai's view, the major tests on the market — Oncotype DX, Prosigna, EndoPredict, and MammaPrint — may accomplish more or less the same thing, but "the most randomized data has been for Oncotype, so they have a valid claim to say their test is the best-validated," he added.

Asked why clinicians might embrace tests that don't rise to this "best-validated" level, Pusztai said that one shouldn't ignore the role that sales and marketing has on the medical field. "It comes down to why buy a Honda when [your local guy sells] Chevys and Fords]," he said. On that front Myriad has said it is focused on using its large sales force to further penetrate the women's health market with its tests. 

Open questions

As questions about chemoprediction for node-negative, early-stage HER2-negative breast cancers have become more solidly resolved, uncertainties still remain for other uses of molecular testing, primarily the extension of prognostic and predictive ability for these assays into node-positive disease, and how they might play a role in guiding regimes of extended endocrine therapy.

One gray area that TAILORx didn't clarify as persuasively was the predictive ability of Oncotype DX in younger women. "These very young patients are a gray area," Pusztai said. "Even if they have a low score, clinicians may be skeptical if that's valid," to make the decision against chemo.

Genomic Health's Steve Shak agreed. In the company's development of Oncotype DX, the test has always been seen as a continuous measure, reflecting a biological gradient as opposed to a binary cutoff. Trials like TailoRx offer clinicians a way to confidently translate that gradient into certain binary decisions. But, that doesn't mean that every clinical decision can be made that way.

"There are still areas where doctors just have to be doctors the way they always were," one of them being young women with Oncotype recurrence scores below 25, Shak said.

According to Robson, another area where the field continues to see "more heterogeneity of comfort" is in the extension of molecular testing into node-positive patients.

For its part Genomic Health has set itself up, via the ongoing RxPONDER trial, to be able to offer the same level of evidence in node-positive disease as it did for node-negative disease via TAILORx.

"The question I get now all the time now is, 'When do we get [those] results?,'" Shak said.

Pending results of RxPonder, Robson said the field is in largely the same place with node-positive disease as it was with node-negative disease five or so  years ago. "The default setting is to treat with chemo, but particularly because of the neoadjuvant experience in HR-positive disease there is at least [some] reason to believe that there are groups that are not going to benefit much from chemo."

"Observational evidence, preliminary real world evidence … suggests Oncotype may perform just as well in that setting," he added

According to Shak, even though trial results for node-positive patients are still pending, Genomic Health is seeing doctors' increased confidence in the test's performance inin node-negative disease based on TAILORx affecting their willingness to use of Oncotype DX in node-positive patients. That has been coupled with new data, including a study presented late last year at the San Antonio Breast Cancer Symposium, from an analysis of the SEER database, showing that Oncotype DX's predictions are reliable in patients with up to three positive nodes.

But with RxPonder still ongoing, other assays have tried to position themselves as more prognostic in the node-positive setting, Robson said.

Sestak said results from studies like her and her colleagues' TransATAC comparison may sway clinicians concerned about performance in node-positive women to choose a second-generation molecular test like Prosigna or EndoPredict.

"Clinical guidelines are different regarding genomic tests for women with node-positive disease," than they are for node-negative patients, she wrote. "NCCN guidelines, for instance, recommend [tests] for node-negative and node-positive disease, while ASCO guidelines recommend only Mammaprint for node-positive disease (based on the MINDACT trial)."

Open questions also remain about how various assays might help inform decisions about who might benefit from extended endocrine therapy.

"I think anything people do with that right now is extrapolation," Robson said. "But it is definitely something we perhaps didn't appreciate as being as important as it is, so people are trying to retrofit studies now to get at that, [though] it's a tough nut to crack."

Genomic Health's Shak highlighted an aspect of the TAILORx results that speaks to lingering challenges in this vein.

"In these node-negative cases, we saw that despite getting chemo, [the high-risk women] with scores from 26 to 100 still had a 13 percent rate of distant recurrence at nine years," he said.

Shak pointed to new studies looking at treatments like immunotherapy and CD4K6 inhibitors that could address that remaining risk, as well as other regimens, like endocrine therapy plus a biologic therapy instead of chemotherapy.

Regarding extended endocrine therapy, Shak said that there are studies that have explored it, and ASCO has updated its guidelines to say that there are pathological features that can guide extended regimens, but no biomarkers have been well validated to predict who benefits.

Genomic Health is continuing to look at that question. He noted, for example, that there had beenobservations years ago that certain parts of the Oncotype assay, like quantitative ER, may predict benefit.

Both Robson and Pusztai said that the future for the field may also involve the use of liquid biopsy technologies — detection of either circulating tumor cells or cell-free tumor DNA — to try to define the lingering risk that prognostic molecular tests can't seem to capture.