Skip to main content
Premium Trial:

Request an Annual Quote

Study Suggests EndoPredict Clinical Score Outperforms Oncotype DX for Assessing Late Recurrence


SAN ANTONIO, TX – A study presented at a major medical conference this week showed that Sividon Diagnostics' EndoPredict breast cancer recurrence test, which combines gene expression analysis with nodal status and tumor size, identifies a subset of patients who have particularly good prognosis in the decade after diagnosis. 

Although this finding needs further validation, researchers led by Mitch Dowsett from the Institute of Cancer Research in the UK presented an analysis at the San Antonio Breast Cancer Symposium suggesting that the EndoPredict clinical score may be more accurate than Genomic Health's Oncotype DX in certain scenarios.

EndoPredict is a reverse transcription qPCR test that analyzes the expression of 12 genes. The gene expression score is combined with a breast cancer patient's tumor size and nodal status to yield an "EPclin" score that estimates the risk of recurrence over 10 years. According to their EPclin score, patients are delineated into high- and low-risk categories, meaning their recurrence risk over 10 years is above or below 10 percent. The test is intended to be used for women with hormone receptor-positive, early-stage breast cancer with or without lymph node infiltration.

In the latest analysis, Dowsett and colleagues used samples from around 930 patients previously enrolled in the prospective TransATAC study, who received treatment for five years with anastrozole or tamoxifen. They retrospectively tested samples with EndoPredict and compared how accurately the EPclin score and Oncotype DX recurrence scores gauged disease risk. Oncotype DX was validated on TransATAC samples, and so patients already had recurrence scores.

Overall, close to 6 percent of patients deemed to be low risk by EPclin went on to experience distant recurrence, compared to 10 percent of patients with low risk Oncotype DX recurrence scores. In node-positive patients, 5 percent of patients deemed low risk by EPclin had a recurrence over 10 years, versus 25 percent with Oncotype DX low recurrence scores.

A difficulty with comparing Oncotype DX recurrence score and EPclin is that the former deems patients to be at high, intermediate, or low risk of recurrence, and the latter classifies patients into high and low risk categories. One way the researchers adjusted for this was by dividing the risk classifications from both tests into three groups, or terciles. Looked at this way, out of 227 node-negative patients in the lowest risk tercile by EPclin, only one patient, or 0.5 percent, had distant recurrence over 10 years. Comparatively, 7 percent of 226 patients deemed to be in the lowest risk tercile by Oncotype DX experienced late recurrences.

"That obviously requires confirmation to see if [the EPclin score] is just as good as that," Dowsett said at the meeting, adding that this was the most striking data coming out of this analysis.

According to Dowsett, the EPclin score, partly but not entirely due to the pathological features considered in the algorithm, provided more accurate prognostic information than the Oncotype DX recurrence score. "The data highlight the importance of including clinical pathological parameters, including type of endocrine treatment, when estimating residual risk and distant recurrence," Dowsett said at the meeting.

Daniel Hayes from the University of Michigan Comprehensive Cancer Center, who has been an investigator in a number of studies involving Oncotype DX, including the recently published prospective TAILORx study, had a different interpretation of the data on EndoPredict. He noted that researchers needed to validate these findings further and said that the data seem to be in line with what many experts in the field believe, "that EndoPredict and Oncotype DX tell us pretty much, but not identically, the same thing."

Treatment benefit

Genomic Health has validated Oncotype DX in node-negative and node-positive patients. While there is broad coverage for testing node-negative patients, payors selectively cover testing in the node-positive subset. The company also markets the test as a way to gauge the risk of 10-year distant recurrence. However, there is data suggesting that while Oncotype DX is an accurate predictor of recurrence in node-negative patients at five years, other tests may be more accurate in gauging distant recurrence, after five years of endocrine therapy.

For example, in the abstract on the latest analysis, Dowsett and colleagues reported that the EPclin score provided substantially more prognostic information than Oncotype DX across all patient subgroups and time periods (in the first five years, between five and 10 years, and 10 years overall), except for node-negative patients in the initial five-year period.

Based on currently available data, Genomic Health claims that Oncotype DX is the only test validated to predict whether a patient will benefit from chemotherapy in addition to endocrine therapy. Still, other labs marketing competing tests are doing or have completed studies to suggest their tests can also be used to guide treatment strategies. Dowsett said at SABCS that "EPClin did identify a group of low-risk patients who may be spared chemotherapy."

Hayes agreed. "I think both of these assays [Oncotype DX and EndoPredict] are good [at] identifying patients with ER-positive, node-negative breast cancer whose prognosis is so good they don't need chemotherapy," he said.

The latest study follows the recent publication of the prospective TAILORx study in the New England Journal of Medicine. The data showed that among the more than 1,600 hormone receptor-positive, node-negative, early-stage breast cancer patients deemed low-risk by Oncotype DX and treated with endocrine therapy, the rate of invasive disease-free survival at five years was 99.3 percent. 

In the NEJM paper, researchers wrote that Oncotype DX predicts benefit from adjuvant chemotherapy, which has shown to prevent early breast cancer recurrences within five years of diagnosis. They acknowledged that more recurrences will happen the longer one follows patients, but the data are unclear whether early chemotherapy treatment would have staved off distant recurrence. They suggested that for some patients, adjuvant endocrine treatment could be given beyond five years to try to prevent late recurrences.

"Whether [EndoPredict or Oncotype DX] or any other assays can be used to identify patients who are disease free after five years of tamoxifen (or an aromatase inhibitor) and do not need extended endocrine therapy is still an open question," said Hayes, one of the author of the TAILORx study, over email. "Many of us [researchers], including Dr. Dowsett and myself, are trying to address this question. It is very complicated."

TAILORx was designed to provide definitive insights on whether patients deemed to be at intermediate risk of recurrence by Oncotype DX should receive chemotherapy. The lack of clarity as to how to treat intermediate risk patients is something competing labs have pointed to as a downside of the test. These data haven't yet been captured in TAILORx because patients are doing so well, and there haven't been enough recurrences or other events to complete analysis.

With longer follow-up, TAILORx will provide more definitive guidance on how accurate OncotypeDX is beyond five years, Hayes indicated.

The evidence is also not yet clear, in Hayes' view, whether breast cancer recurrence assays accurately predict chemotherapy benefit for node-positive patients. The RxPonder trial, which has completed accruing study subjects, is assessing whether node-positive breast cancer patients with low-to-intermediate Oncotype DX recurrence scores benefit from chemotherapy in addition to hormonal therapy following surgery.

Market push

After more than a decade on the market, the Oncotype DX breast cancer recurrence test dominates the US invasive breast cancer space with 90 percent market share, according to Genomic Health's own data. As such, Oncotype DX is the one to beat for competitors Agendia's MammaPrint, NanoString's Prosigna, and BioTheranostics' Breast Cancer Index which like EndoPredict have been compared to the older standard.

Sividon is certainly building the evidence around its test to compete with Oncotype DX, which has broad adoption in the US and in international markets. For example, Oncotype DX has been evaluated by the UK's National Institute for Health and Care Excellence (NICE) and is offered by the National Health Service (NHS) for a subset of patients. Sividon's test has not been evaluated by NICE to determine if it will be cost effective for the NHS to provide the test as an option in the UK and Wales. As such, EndoPredict is currently only available privately in the UK. 

"Whilst it’s possible that more evidence to support the use of EndoPredict will be required before it is considered for routine use in the UK, research like this is essential if we are to constantly improve the accuracy with which patients are treated," Katherine Woods, senior research communications manager at Breast Cancer Now, a breast cancer charity in the UK, said in a statement . Breast Cancer Now partly funded the study comparing EndoPredict and Oncotype DX.

In the US, Sividon is aiming to offer EndoPredict as a kit with 510(k) clearance from the US Food and Drug Administration. The firm believes that selling its test as a kit has advantages in the way of shorter turnaround times and delivery of test results by local experts, compared to lab-developed tests that have to be performed by a single lab.

To date, breast cancer recurrence testing has been offered in the US through lab-developed tests, such as Oncotype DX, Prosigna, Breast Cancer Index, and MammaPrint, which also has 510(k) clearance from the FDA. The companies offering these tests are continuing to build the evidence around their products and presented new data at the SABCS.

BioTheranostics, for example, used samples from 219 patients in the TransATAC trial aimed at developing a new Breast Cancer Index prognostic test for estrogen receptor-positive, node- positive disease. By combining tumor size and grade with Breast Cancer Index into a comprehensive risk score, researchers showed that 55 women deemed to be at low risk had no distant recurrences within 10 years. Researchers noted that these women could potentially forgo adjuvant chemotherapy or receive extended endocrine therapy. 

NanoString presented analysis on approximately 2,700 patients samples from a Danish database, showing that Prosigna predicted local recurrence over standard variables. The study validated NanoString's pre-specified cutoffs for determining patients at high and low risk of local recurrence. 

Agendia described the design of IMPACT, a prospective registry study to measure how MammaPrint impacts treatment decisions in breast cancer patients. 

Meanwhile, Genomic Health continued to raise the bar for competitors by presenting a number of prospective studies at the symposium. In one study, Genomic Health collaborated with the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) registry program to gauge five-year breast cancer specific survival in more than 38,000 node-negative, hormone receptor-positive patients.