Feature Story

Getting personal: Cancer assay could provide individual risk of recurrence

21-gene panel

February 2004
Karen Lusky

Sometimes it seems the more things change on the cancer research front, the more they stay the same in practice. That's been especially true in the quest for breast cancer prognostication methods over the last decade-perhaps until now.

A new individualized multi-gene breast cancer assay that has just become available could shed light on the clinical dilemma that currently calls for a crystal ball. That is, which patients with early-stage breast cancer will be the ones to develop distant metastasis? And, of course, the critical corollary question: Which ostensibly low-risk patients need more aggressive treatment to improve their odds of survival over the long term?

"We know that 20 to 30 percent of patients with early-stage breast cancers go on to develop distant metastasis, as do 50 to 60 percent of more advanced cases," says Joseph Crowe, MD, director of the Cleveland Clinic Breast Clinic. "What we don't know is who those patients will turn out to be."

Genomic Health's Oncotype DX could provide information to better individualize those odds for the 70 percent of newly diagnosed, node-negative breast cancer patients each year, according to a recent clinical validation study performed by the National Surgical Adjuvant Breast and Bowel Project, or NSABP. The study found that the 21-gene assay did a better job than existing prognostic markers, such as patient age, tumor size, and tumor grade, in foretelling recurrence over 10 years in patients whose disease is newly diagnosed, node negative, stage I or II, and estrogen receptor positive and treated with tamoxifen.

If it pans out, the test could also usher in an era of personalized medicine that extends way beyond the use of Herceptin for patients with HER2-driven metastatic breast cancer.

Genomic Health, of Redwood City, Calif., has received CLIA certification as a reference laboratory to offer the assay as a high-complexity test, says the company's co-founder and chief medical officer, Steven Shak, MD, who formerly led the clinical team at Genentech that gained approval for Herceptin. "In the more distant future, it may be possible to convert the Oncotype DX test into kits," Dr. Shak says. The biotechnology company has applied for patents to claim all of the inventive components of its work.

The Oncotype DX assay quantifies the RNA expressed in thin sections from standard diagnostic pathology specimens (formalin-fixed paraffin-embedded tissue). "That's what is so special about this assay-it can work from the standard paraffin block," Dr. Shak says (Cronin, et al. Am J Pathol. 2004;164:35). "Previously, lab tests were not designed to quantify the RNA that's present in paraffin-fixed tissue."

In addition, most laboratory tests are now configured to look at just one gene at a time. By contrast, the Oncotype DX analyzes the RNA expression of a panel of 21 genes (16 of them predictive for cancer recurrence and five reference genes). "Five of the genes are related to proliferation, such as Ki-67, and four genes are related to estrogen and progesterone receptor status of the tumor," says Soonmyung Paik, MD, director of pathology at NSABP. In addition, "the panel includes two HER2-related genes, one of which is GRB7-and two invasion-related genes." (See box, below.)

Researchers culled the assay's 16 cancer-related genes from 250 candidate genes assayed individually with real-time polymerase chain reaction on 447 patients in three prior clinical trials. As Dr. Shak explains, "For real-time RT-PCR, we synthesize forward and reverse primers and a probe for each gene of interest, which can quantify . . . how many fragments of that gene's RNA are present in the tumor tissue."

The NSABP and Genomic Health then went on to clinically validate the 21-gene panel in a study on 668 patients with node-negative, ER+ breast cancer who'd been treated for five years with tamoxifen. The women were enrolled in the NSABP B-14 clinical trial from 1982 to 1988. Various NSABP clinical sites had tracked the women's clinical outcomes. Dr. Paik presented the study findings at the 26th San Antonio Breast Cancer Symposium in December 2003. (The presentation can be viewed at www.smdna.net/ smdna/ sabcs2003.)

Researchers used the Oncotype DX breast cancer assay to determine each woman's likelihood of recurrence based on a score ranging from 0 to 100. Based on their scores, women in the study population were assigned into high-, intermediate-, and low-risk groups.

The patient's age and tumor size were only marginally significant in the population studied, Dr. Paik says. "And [those risk factors] get blown away when you compare them to the recurrence score in the Oncotype DX." In addition, HER2 turned out to be "not as important as we thought it was in this particular population."

However, tumor expression of GRB7, an adaptor molecule for HER2, is associated with a poor prognosis, says Melody Cobleigh, MD, director of the Comprehensive Breast Cancer Center and Molecular Diagnostic Clinic at Rush University Medical Center, Chicago, and a participant in earlier studies that led to the development of the 21-gene set.

In Dr. Paik's view, the breast cancer assay is promising and useful clinically because it provides a "very reproducible and continuous predictor"-one that doesn't lump patients into three large risk categories of low, intermediate, and high based on tumor grade.

"Even if you look at the low-risk group defined for the sake of the NSABP validation study for the Oncotype DX assay, the average risk of 10-year recurrence runs about six percent," Dr. Paik says. Yet the individual risk in that group actually ranges from three to 12 percent. "So it's not fair to tell every woman she has a six percent risk of recurrence," he says. "The value of the Oncotype DX assay is that it provides individual risk rather than a group average risk."

Patients who get a score of 30 on the Oncotype DX assay have a 20 percent chance of distant metastasis in 10 years. "The recurrence risk is linear until the score goes above 50, but it's highly unstable after 50 due to the small number of patients in that [high-risk] group," Dr. Paik says. "It doesn't matter, however, as patients who score over 30 will generally want chemotherapy" in addition to tamoxifen to improve their odds of survival. "Up to a score of 50," he says, "we can provide fairly accurate individualized assessment of risk of failure in 10 years."

The jury's still technically out on whether the test can help clinicians better target chemotherapy to the patients who will benefit from it. Genomic Health is collaborating with the NSABP to look at the NSABP B20 trial where patients with hormone receptor-positive, node-negative breast cancer were randomly assigned to tamoxifen or tamoxifen and chemotherapy. "The followup study, expected to be conducted this spring, should provide more definitive information on the assay's benefit in helping women decide whether to add chemotherapy to their treatment program," Dr. Cobleigh says.

Dr. Paik is confident the study will show that the high-risk group identified by the Oncotype DX test is the one that will benefit from chemotherapy. "We know that the high-risk group is the one with the high level of proliferation and the low level of estrogen receptors. So our study is more a validation of what's already been suggested in other studies."

The main aim of the markers is to identify the population that can benefit maximally from chemotherapy, such as dose-dense regimens, Dr. Paik says. "Dose-dense regimens have been shown to be much more effective than the conventional way of delivering the drug, as the cancer cells don't have the time to recover during the chemo-free intervals."

The assay could also help prevent unnecessary chemotherapy. Using current clinical protocols, "women with node-negative, ER-positive breast tumors under 1 cm are usually treated with adjuvant therapy consisting of tamoxifen only following surgery, while those with tumors of 1 cm or larger usually receive chemotherapy and tamoxifen," Dr. Cobleigh says.

Yet chemotherapy offers only a 25 percent relative risk reduction for these women. And "relative" is the catchword that most patients don't catch. For example, the NSABP study population showed just a 15 percent risk of recurrence rate among the women in 10.9 years of followup. "So 85 percent of the patients did just fine with five years of tamoxifen only" over that period, Dr. Cobleigh says. Adding chemotherapy to tamoxifen treatment following initial diagnosis reduces that 15 percent risk of recurrence by 25 percent, which translates into a four percent absolute risk reduction. "In other words, four out of 100 women who take chemotherapy won't relapse-and 11 will," Dr. Cobleigh explains.

With the Oncotype DX score the individual woman will know if she has a lot higher odds of recurrence and would therefore benefit from undergoing chemotherapy. "Conversely, the test would also identify women at much lower risk than average. These women may decide to forego chemotherapy," Dr. Cobleigh says.

Could the genomic tumor assay help women with ER-positive breast cancer make an informed decision about taking the aromatase inhibitor, letrozole, after the now recommended five-year course of tamoxifen therapy? A recent clinical trial showed that letrozole dramatically cut the risk of breast cancer recurrence among a large group of postmenopausal women who took the drug for just two and a half years after finishing the standard five-years of tamoxifen. At this point, Dr. Cobleigh says it would require a "leap of faith" to use the assay in such a way, pending further validation studies.

While the genomic assay's predictive powers have been validated only on the one group of women with early-stage breast cancer, Drs. Shak and Paik believe further studies will show the test works on other breast tumors as well.

Yet a pilot study at the University of Texas M.D. Anderson Cancer Center casts doubt on that prediction. Researchers there tested the assay's predictive abilities on 149 women who had received only surgery at the center (either lumpectomy with radiation or mastectomy) from 1978 to 1995. In that small study, the 21-gene set assay could not foretell which patients were likely to develop distant metastasis.

"The study can be criticized because it involved a highly selected group of patients, in that most patients at M.D. Anderson receive adjuvant therapy, such as tamoxifen or chemotherapy," says the study's lead researcher, Francisco J. Esteva, MD, PhD, associate professor of medicine. The study also showed that women with high-grade tumors had better outcome-an unexpected finding that the researchers can't explain completely. "Most likely, that finding is due to patient selection that's not representative of the general breast cancer population," Dr. Esteva says.

In spite of the M.D. Anderson study's small size and limitations, Dr. Esteva believes it shows there's room for caution. "Thus far, there are only two validation tests of the assay-one was positive and the other wasn't," he says. "If we had 10 studies-and only one was not predictive-then you could discount that one. But we only have two of them."

The M.D. Anderson study does raise a question about the test, agrees G. Thomas Budd, MD, director of the medical oncology breast cancer program at the Cleveland Clinic. "Yet the NSABP data are probably more relevant because the NSABP is a national multicenter group with representation of community practice, and it more accurately reflects the general breast cancer population."

Dr. Paik agrees that people would be much more comfortable with the Oncotype DX test if it were validated in a large prospective study. "And NCI is considering a very large study of about 10,000 patients to prospectively validate it by assigning no treatment or specific treatment to various risk groups, and also as a way of further refining the test," he says. "But it would take 10 years to get the results."

Meanwhile, Dr. Paik believes that clinicians will use the assay for about five years before microarray tests catch up. "The microarray assay will not only provide prognostic information but also information to predict patients' response to specific treatment regimens and toxicity," he predicts.

Like any breakthrough technology, the Onco- type DX assay raises myriad other questions for pathologists, oncologists, and patients.

For example, what's the emotional impact on women who receive a high score on the gene index? Could the foreshadowing of a dire outcome affect the women's prognosis? Oncologists will have to learn to individualize how they impart that information to each patient, Dr. Budd says. "Yet, the impact of the knowledge on psychological status and immune functioning would likely be outweighed by being able to have information to optimize therapy," he says.

Another pressing question: Who will pay for the test? Hospital labs have found themselves holding the bag for expensive genetic tests that insurers cover marginally or not at all.

"Oncotype DX is a highly complex 21-gene RT-PCR assay, and the price has been set at $3,460," says a Genomic Health spokesperson. In pricing the test, Genomic Health looked at "the value the test will bring," she says, and considered the pricing of single-gene tests on the market, which range from a few hundred dollars for the FISH test to $3,000 for the BRAC1/ BRAC2 test. Genomic will not generally bill labs; it will bill patients and payers directly, the company's spokesperson says.

Genomic Health also expects to provide insurance companies the information they require to justify reimbursement. "Some women and their physicians will have more precise and accurate information concerning the likelihood of recurrence and will be more confident in their treatment plan to take or not to take chemotherapy," Dr. Shak says.

The assay's quality control and assurance is also a concern of pathologists, some of whom question whether a test involving 21 different genes really qualifies as a homebrew or lab-developed test exempt from Food and Drug Administration review under the analyte-specific reagent exception.

Genomic Health is doing quality control for each gene measured, Dr. Shak says. "We performed validation studies for measurement of each of the 21 genes used for the recurrence score. We also include controls and calibrators in every step. Also, every reagent used is qualified for the assay."

The analytical validation studies were done with tremendous rigor, Dr. Shak adds. "The assay methods used in the clinical trials and those used when we provide the reference lab service will be the same," he says. "We have tried to make this as easy as possible to fit within the normal procedures of most labs. We do not require the tissue block to be sent to us so pathologists don't have to worry about doing that."

Genomic Health will request one routinely prepared H&E slide for every case (for internal QC) and will ask for three 10-micron sections from the tumor block packed into one tube and another three 10-micron sections in a second tube.

What about the preanalytical phase in terms of RNA preservation? Dr. Shak says Genomic Health developed the methods for the assay to be minimally affected by variations in block preparation and storage. "The assay is insensitive to the variations in preparation of the tissue," he says. "It remains robust and works even if the tissue is set in formalin a little longer or shorter or sits on the table a longer or shorter time."

Announcement of the multi-gene assay by a biotechnology company also sounded a warning knell for the pathology community in terms of its role as promulgators of such breakthrough technology.

"The future of much of pathology, especially tumor pathology, is molecular pathology," says Raymond Tubbs, DO, chair of the clinical pathology department at the Cleveland Clinic. "Our future-or a large component of our future-lies in translational research, development, analytical and clinical test validation, implementation, and quality assurance of these molecular pathology assays. And simply sending tumor tissue to a company for genomic signature testing leaves the academic pathologist in the dark ages of conventional morphology. Pathologists cannot become simply the Wal-Mart purveyors of breast tumor tissue for private companies."

David Hicks, MD, director of surgical pathology at Cleveland Clinic, agrees: "If the pathology community does not get involved and participate in the development of this new molecular diagnostic technology, someone else will."

Yet pathologists can take advantage of the technological leap of using quantitative RT-PCR on formalin-fixed specimens, says Craig Allred, MD, professor of pathology at Baylor College of Medicine and head of Baylor's Breast Pathology, Breast Care Center. "Genomic Health and another leading-edge company, Arcturus, have given us a way to evaluate tissue using semi-high throughput technology on routine formalin-fixed, paraffin-embedded tissue," he says.

"Moving forward, pathologists have to make sure to stay involved by being aware of and taking advantage of innovations such as this when they come along," Dr. Allred says. "Pathologists can use this new tool now to develop genetic signatures that help us understand how breast cancer and other types of cancers respond to particular types of drugs and so on."

In applying this technology to personalize cancer treatment, Dr. Cobleigh says, "the sky is the limit."

Karen Lusky is a writer in Brentwood, Tenn.