Skip to main content
Premium Trial:

Request an Annual Quote

Landscape Still Unclear for Prostate Cancer Therapy Prediction Tests as New Approaches Emerge


NEW YORK (GenomeWeb) – With one test for guiding the choice of therapies for advanced prostate cancer slated to launch this year, research continues to paint a complex picture of the molecular determinants of drug response in this setting, raising questions for how the commercial testing landscape might evolve in the future.

New data published last month supports the potential for a simple droplet PCR-based method that interrogates a different marker than most tests have so far — androgen receptor copy number variation rather than expression of the AR-v7 splice variant.

In the study, which appeared in Annals of Oncology in May, a team led by researchers at the Institute of Cancer Research, London showed that AR copy number changes in the circulating cell-free DNA of men with advanced prostate cancer were significantly associated with outcome for those treated with the hormonal therapies abiraterone and enzalutamide.

In a first cohort, the researchers found that plasma AR copy number gain was associated with shorter overall outcome as well as with shorter progression-free survival in both patients who had undergone chemotherapy as well as those who had not undergone chemotherapy. Men with multiple copies of the AR gene were nearly four times more likely to die during the course of the study.

They then replicated this finding in a separate cohort of 94 men from whom plasma was obtained prior to enzalutamide treatment, and, again, AR gain was significantly associated with shorter overall survival.

According to Gerhardt Attard, team leader in the Centre for Evolution and Cancer at the ICR and a senior author of the study, tests that can help determine which men will respond to androgen-directed drugs and which will not are greatly needed, offering an attractive target for commercial diagnostics development.

The two drugs work for some men for years, but in others they barely seem to make a dent, allowing cancer to return and spread almost immediately. And though there is currently no clinical test available, several groups have been pursuing various molecular strategies actively over the last several years, and Epic Sciences is poised to launch its test this year through a partnership with MDx firm Genomic Health.

Qiagen is also advancing an AR-V7 assay using technology it acquired from AdnaGen and licensed from Johns Hopkins University.

Attard said in an interview this week that one of the cohorts in which he and his colleagues demonstrated their AR copy number approach was from a study called PREMIER, in which both AdnaGen-based AR-V7 testing and AR gene testing have been evaluated.

Researchers presented data in an abstract at the ESMO 2016 Annual Meeting that showed that AR-V7 was not predictive of therapy response in the PREMIER cohort.

In those same patients, Attard and colleagues found that AR copy number was strongly predictive.

However, he said this week, that doesn't necessarily mean that AR-V7 testing will be a bust. It could just reflect an issue with the testing done in that study, since there has been demonstrated variability in results when different labs have used the AdnaGen-based method.

Meanwhile, Epic Sciences, which uses a unique cell analysis platform to measure nuclear-specific AR-V7 in isolated circulating tumor cells, has collected significant data that shows that its approach does indeed allow prediction of therapy outcome using AR-V7.

Importantly, Attard said that an advantage to his team's approach is that measuring AR copy number gain in a blood sample is relatively simple and inexpensive. The BioRad droplet digital PCR platform the researchers employed is increasingly available in labs, and the researchers estimated that their method costs only about $64 per test to perform.

Whether the ICR team's AR copy number test can be brought into the clinic depends on further validation studies that the group is now working on. This includes a study further validating the findings reported in Annals of Oncology, but using prospectively collected samples and comparing the predictive ability of AR copy number compared to AR-V7 expression in CTCs.

Investigators hope to tease out exactly what the overlap is between AR copy number and AR-V7-based prediction of outcome.  Do both methods identify the same non-responders or might there be an added value in combining both biomarkers?

From a biology perspective, AR gene copy gains are directly related to functional expression of AR, a subset of which would be represented by the AR-V7 splice variant, so a connection between the different biomarkers is logical. However, it's not clear yet how robust that relationship might be, Attard said.

"In the next six to nine months we will have data from concurrent testing with both methods," Attard said.

Considering the advantages in terms of ease and cost, he added that a clear additive benefit for AR-V7 may be necessary to merit its use over ddPCR analysis of AR copy number. Based only on the ESMO data from PREMIER, that doesn't look likely at least for the AdnaGen-based assay, he said. However, there could be technical reasons for that, so the question is still open.

Attard has worked with Epic Sciences before, and he said he and colleagues have also been discussing with the company how to include its platform in these types of comparative studies, since the results might be different for its technology compared to Qiagen's.

Meanwhile, a head-to-head comparison of three AR-V7 assays is also underway — led by Andrew Armstrong at Duke University — comparing the Epic test, the AdnaGen-based RT-PCR test developed by Johns Hopkins and now licensed by Qiagen, and a third multiplex assay looking at AR-V7, wild-type AR, and another splice variant called AR-V567es.

Research led by a group from Dana-Farber Cancer Institute also showed last year that it's possible to predict outcomes of patients on hormonal prostate cancer therapy based on ddPCR measurement of PSA and AR-V7 transcripts in blood samples without the need to isolate CTCs.

Epic Sciences Vice President of Translational Research and Clinical Affairs Ryan Dittamore said this week that Epic Sciences has seen some hints of what the interaction and relationship between AR gene copies and the expression of particular splice variants like AR-V7 might look like in its own data.

For example, in a study published last year in JAMA Oncology — which validated Epic's AR-V7 assay in predicting response to anti-androgen drugs versus taxane chemotherapy — researchers also looked at AR N terminal overexpression.

Though not a direct measure of AR copies, Dittamore argued that because N terminal expression should be the functional product of copy number gain, it's at least a close correlate.

Based on the company's data, "[hazard ratio and survival performance [for] AR N terminal overexpression in CTCs has very good performance at predicting poor [overall survival with anti-androgen drugs] as well as good sensitivity in the context for use."

In comparison, "our AR-V7 assay … calls out less patients but has a more dramatic specificity and HR … which leads to the predicted therapy interaction with chemotherapy. While AR N terminal [was] close to achieving the therapy interaction, in this cohort it failed to hit statistical significance," he added.

According to Attard, the future of commercial clinical tests is also complicated by the fact that the landscape of treatments may change over the next few years – beyond just the question of abiraterone and enzalutamide versus chemotherapy. "We know we need better drugs to treat AR aberrant patients," for example, he said.

"However, once the space for using [therapy-predictive tests] has been defined, that will come with important commercial opportunities."