Skip to main content
Premium Trial:

Request an Annual Quote

Epic Sciences AR-V7 Study Raises Questions for Upcoming Head-to-Head Assay Comparison


NEW YORK (GenomeWeb) – Epic Sciences and its academic collaborators published new data last week that suggest that assessing AR-V7 expression only in the nuclei of circulating tumor cells might be a better way to predict therapeutic benefit in prostate cancer patients than a less-specific, overall measurement of the protein or its associated RNA transcript.

Epic made a deal last year giving Genomic Health exclusive right to market its CTC-based AR-V7 test in the US, though it would still run the assay out of its own CLIA lab.

In the new study, which appeared in European Urology on Friday, a research team led by Memorial Sloan Kettering's Howard Scher — who has been working with Epic throughout development of the test — followed up on results from an earlier investigation, published last year in JAMA.

In that study, the team validated the Epic AR-V7 protein assay (with a nuclear-specific scoring criteria) to predict response to anti-androgen drugs versus taxane chemotherapy in men with progressing metastatic castration-resistant prostate cancer.

At the time, they observed that the rate of AR-V7 detection in CTCs using their nuclear-specific criteria was lower than was seen with more general mRNA-based detection methods in comparable patient cohorts — 18 percent versus up to 55 percent of patients, respectively.

Because it was possible that strict nuclear-specific AR-V7 protein scoring might be too stringent, potentially missing some patients who could be better served by chemo than anti-androgen treatment, the team set out to compare the initial nuclear-specific approach to scoring criteria that included both nuclear and cytoplasmic AR-V7. Overall, they analyzed 191 mCRPC patient samples in the new study.

The results suggested that the original nucleus-only criteria could discriminate completely between androgen inhibitor responders and non-responders. None of the patients with nuclear localized AR-V7 protein expressed in CTCs showed a response to anti-androgen therapy.

In contrast, while patients identified using the less-restrictive scoring criteria showed generally poor outcomes with androgen inhibition on average, several of the individual patients did respond to the drugs.

"When the criteria was expanded to nuclear-agnostic, six of the additional 16 samples classified as AR-V7-positive showed sensitive PTPC, which represents a false-positive result, relative to the nuclear-specific criterion, that could potentially deny patients a minimally toxic, safe, life-prolonging therapy," the authors wrote.

According to Epic, the change in scoring also significantly affected prediction of therapeutic benefit in switching from hormonal to chemotherapy.

Only the nuclear scoring criteria, not the unselected method, demonstrated extension of life through switching therapies, Epic said.

Promoting the new data, Epic argued that the results imply that its own technology — which allows specific assessment of nuclear AR-V7 protein via a platform for arraying and examining individual CTCs — may better predict which patients should be treated with chemotherapy over androgen-inhibiting drugs than other assays.

"This study shows that using Epic Sciences' proprietary CTC platform to determine whether AR-V7 protein is sequestered in the cytoplasm or translocated to the nucleus can provide an important clinical difference compared to other technologies with less specific AR-V7 detection techniques," Epic vice president of translational research and clinical affairs Ryan Dittamore said in a statement.

However, despite this conclusion on the company's part, the question of how these tests actually measure up in terms of their clinical validity and utility is still very much open, and will require direct comparison of Epic's technology to other assays, others in the field said.

"The jury is still out on which assay will prove to be the best and moreover whether knowing AR-V7 status will have [sufficient] clinical utility," MSKCC chairman of medicine Philip Kantoff said in an email.

Kantoff has also been working on harnessing AR-V7 to help guide treatment for mCRPR. In his previous position at Dana Farber, he and colleagues completed a retrospective study showing that they could predict outcomes of patients on androgen inhibitors using droplet-based PCR measurement of both PSA and AR-V7 transcripts in blood samples, without the need to isolate CTCs.

Emmanuel Antonarakis, leader of a team at Johns Hopkins that has been working to develop an RT-PCR-based mRNA AR-V7 assay, said that while the discussion by the Epic and MSKCC authors makes strong claims for the potential superiority of Epic's test, the actual study data don't necessarily back up that conclusion.

The fact that Epic did not cement its nuclear-specific versus general AR-V7 scoring criteria in advance to allow a prospective evaluation in the cohort also limits the implications of the results, Antonarakis said. If one defines the cutoff post-hoc to give the data the best statistical discrimination, the results should at best be considered exploratory, he argued.

Dittamore agreed in an interview last week that the ultimate word on which is the best methodology will require a head-to-head comparison. However, he still maintained that the study results provide evidence for the potential superiority of Epic's approach.

"The important thing that we learned [from] this publication is that the importance of nuclear localization could be a key differentiator of our test over others," he said.

Moreover, he added, the study provided evidence that Epic's test can predict not only that patients are unlikely to benefit from androgen inhibition, but that they will benefit from using chemotherapy. "This a real-world question," he said.

Choosing taxane chemotherapy over androgen-inhibition is a crucial decision. "Most patients prefer the [latter] given the ease of administration and the more favorable safety profile. The decision to choose a cytotoxic drug is therefore not taken lightly," the study authors wrote.

"Thus, it is essential that the false positive rate of any therapy-guiding predictive biomarker at this decision point in mCRPC be as low as possible."

Meanwhile, a head-to-head comparison of three AR-V7 assays is already underway — involving a prospective trial led by Andrew Armstrong at Duke University comparing the Epic test, the RT-PCR test developed by Antonarakis and colleagues at Johns Hopkins, and a third test that has not yet been published, which is a multiplex assay looking at AR-V7, wild-type AR, and another splice variant called AR-V567es.

The study recently completed recruitment of a planned 120 patients, Antonarakis said, all of whom are being followed prospectively, with each of the three assays performed at every time point.

Another lingering issue is that even if an assay shows high specificity for predicting resistance to anti-androgen drugs, AR-V7, regardless of the particular assay, appears to be fundamentally limited — unable to catch all patients with resistance.

In that vein, Scher and colleagues have been researching other capabilities of Epic's platform that may help further stratify patients, including measuring the overall genomic and phenotypic heterogeneity of CTCs as a predictor of response to anti-androgen drugs.

In previous research he and his team found that patients whose CTC population showed a high degree of heterogeneity — either in their phenotypes or their genetic clonality — did not respond as well to hormone therapy as those with less heterogeneity.