NEW YORK (GenomeWeb) – Most women with poorly differentiated breast cancer tumors and low risk scores from Genomic Health's Oncotype DX did not die from their disease five years after diagnosis, even though most of them were spared chemotherapy after receiving test results, data presented at the San Antonio Breast Cancer Symposium last week showed.
Using information collected in the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) registries, Genomic Health evaluated clinical outcomes in 9,200 patients with node-positive and -negative disease and grade 3 tumors — a characterization of how "organized" tumor tissue looks under a microscope and an indicator of how quickly the tumor will spread. The analysis showed that women between the ages of 40 and 84 who had hormone receptor-positive disease and Oncotype DX low risk scores had five-year, breast cancer-specific survival of greater than 99 percent, regardless of their nodal status and tumor size.
In reviewing the study at SABCS, Mayo Clinic's Matthew Goetz said that based on prior data on Oncotype DX, he thought there may not be much need for the test in patients with poorly differentiated, or grade 3, tumors, which are generally considered to be at high risk of recurrence. But the present analysis, he acknowledged, demonstrates that "even in the poorly differentiated tumors, [the recurrence score] is still prognostic" and that the test might be useful in reducing variability between pathologists' assessments of such tumors.
The 21-gene expression test provides a low, intermediate, and high score that characterizes the chances of a woman's early-stage breast cancer coming back. Researchers led by Steve Shak presented data on patient outcomes in patients with low risk scores (less than 18), according to nodal status and tumor size (smaller or bigger than 2 cm).
In the node-negative group, five-year breast cancer-specific survival was 99.5 percent in patients with smaller, poorly differentiated tumors who had low Oncotype DX scores, even though 9 percent received chemotherapy. "With larger tumors, the risk is somewhat increased," Shak said, and around 14 percent received chemotherapy, but five-year breast cancer-specific survival was still high at 99.1 percent. With higher recurrence scores, chemotherapy rates also increased in node-negative patients with poorly differentiated tumors, the data show.
In patients with poorly differentiated node-positive disease, five-year breast cancer-specific survival was also greater than 99 percent for those with low-risk Oncotype DX scores, though 27 percent of those with small tumors and 22 percent of those with large tumors received chemo.
Although a minority of low-risk patients received chemotherapy, Shak said the data show that the Oncotype DX low risk score is a strong predictor of five-year, breast cancer related survival outcomes for those with grade 3 tumors.
In his review of the data at SABCS, Goetz said the study demonstrates that "not all high-grade tumors are created equal."
"This is really an incredible effort to link the data within the NCI's databases and the recurrence scores," Goetz added. "We're really continuing to move toward precision medicine."
The SEER program, which collects cancer statistics across several regions throughout the US, decided in 2010 to start gathering breast cancer multi-gene test results. Initially, registry staffers went to hospital sites to collect the recurrence scores manually, but ran into difficulties since these types of results don't typically reside in patients' surgical charts.
For example, Genomic Health sends test reports to oncologists, who may or may not work at a hospital where their patients got surgery, so the Oncotype DX score may not end up in patients' hospital charts. At the same time, Genomic Health is increasingly asked to show data on how test results translate into patient outcomes. Did patients who are at low risk for breast cancer recurrence by Oncotype DX, for example, really do well without chemotherapy or did they recur? And how accurately does the recurrence score predict outcomes in different breast cancer subpopulations, such as node-positive and -negative patients?
Traditional research models aren't well suited to answering these questions quickly. Moreover, outcomes data are often siloed at different institutions and protected by various informed consent and privacy rules. SEER registry staffers estimated that they were missing 40 percent of recurrence scores when they tried to collect the data manually.
But after SEER required the collection of results from breast cancer multigene tests, Genomic Health partnered with the program and last year electronically linked its database with its registries so that recurrence scores could be more easily studied alongside patients' outcomes within the registries.
Shak characterized the partnership as a new model for industry/government collaboration. "This is why everyone is talking about the MoonShot, big data, and data sharing," he told GenomeWeb.
Through the collaboration with SEER, Genomic Health is investigating the use and impact of Oncotype DX in the real world. According to SEER data on women diagnosed between 2004 and 2011 with hormone receptor-positive, early-stage breast cancer, Oncotype DX was performed in around 22 percent of 40,000 patients with node-negative disease, and on 8 percent of nearly 4,700 women with micrometastases or one to three positive nodes.
Earlier this year, researchers from NCI, SEER, and Genomic Health published results of a prospective evaluation involving around 38,000 breast cancer patients, which showed that among node-negative patients, 0.4 percent of patients with low risk scores, 1.4 percent with intermediate risk scores (18 to 30), and 4.4 percent of patients with high risk scores (above 30) died from breast cancer five years after diagnosis. In node-positive patients, five-year breast cancer-specific mortality was 1 percent in low-risk patients, 2.3 percent in intermediate-risk patients, and 14.3 percent in high-risk patients.
Noting that 14 percent of high-risk patients died despite receiving chemotherapy, Shak said that this is an example of how Oncotype DX can identify patients who cancers are wolves in sheep's clothing. "They might look like they are going to do well, because they have favorable [characteristics like] estrogen receptor positivity," he said. "But with a high recurrence score, they are a wolf."
According to Genomic Health, the prospectively collected SEER data also backs up what Oncotype DX validation studies using archival samples, and prospective investigations, such as the TAILORx and the West German Study Group "Plan B" trial have shown: that node-negative and node-positive breast cancer patient with low risk scores do well without chemotherapy.
Moreover, the collaboration with SEER has also identified some important disparities regarding the use of Oncotype DX, Shak said, for example, that the test is ordered much less frequently in people 70 years or older. Within SEER, only 8 percent of node-negative patients in this age group had a recurrence score, while nearly 28 percent of patients younger than 70 had Oncotype DX testing.
In more than 4,500 women in the registry who were older than 70 years, those with low recurrence scores had good outcomes generally, but women with intermediate and high scores had much worse outcomes than younger women. This finding was surprising, Shak said, because the conventional wisdom in estrogen receptor-positive breast cancer is that older women tend to have favorable outcomes.
He highlighted that this real-world data corroborates a large randomized trial published in the Journal of the American Medical Association in 2012, in which breast-cancer related mortality was higher in older women compared to younger women. The SEER data is also in line with publications that have reported that older women don't always receive the necessary surgical and treatment interventions, and Shak said these disparities need to be better understood to bring personalized medicine to everyone.
In more than 1,400 women younger than 40 years, close to 700 of them got low risk scores from Oncotype DX and 18 percent got chemotherapy, but less than 1 percent of them died from breast cancer at five years. It's scary for a young woman, Shak said, when she's diagnosed with breast cancer and is told based the average patient experience, that she needs to be treated aggressively. "But guess what … all tumors are not the same, and not everyone who has breast cancer, who's young, is doomed," he said.