NEW YORK – A new study this week has added valuable cross-signature and cross-population data on how currently used gene expression tests for prostate cancer recurrence risk perform and compare in African American versus caucasian men.
Appearing online today in Cancer Epidemiology, Biomarkers & Prevention, the report concluded that the racial differences observed across three gene panels — Genomic Health's Oncotype DX Prostate, Myriad Genetics' Prolaris, and Decipher Biosciences' Decipher — suggest that more caution may be warranted when using the tests to guide clinical decision-making for men of African ancestry than is currently employed.
As genomic profiling and precision medicine move further into the mainstream of oncology, concerns have also come to the fore over how to make sure that the promise of these new paradigms can be realized across different ethnic or racial populations.
One hindrance to this is that things like genomic databases, the prognostic or predictive test signatures gleaned from them, the biological resources like cell lines used to validate and quality assure them, and the trial subjects in which they have been clinically evaluated have historically skewed toward populations of predominantly European ancestry.
Describing their impetus for their study this week, the authors from Tampa, Florida's Moffitt Cancer Center cited the fact that gene signatures that form the backbones for the tests in question were developed and validated in cohorts of predominantly European American men, with limited research so far on how patterns of expression in these gene lists may differ in different racial populations, and how the resulting predictive value of the tests might vary.
Travis Gerke, an assistant member at Moffitt and senior author of the study, said in an interview that because the analysis didn't compare actual clinical test results from the companies that offer these assays, or track cancer patients' outcomes, it shouldn't be taken as concrete evidence for a lack of utility in African-American men.
"We definitely do not provide evidence that the tests don't work," he said. Rather, the results highlight that this question may need to continue to be researched — to confirm conclusively if and how the tests perform in this population.
All three of the tests studied are included in the recommendations issued by the National Comprehensive Cancer Network to predict outcomes in men with low- or intermediate-risk prostate cancer. But NCCN doesn't describe the assays as required or necessary for treating patients.
"To my knowledge I don't think there is any race-specific guidance toward clinical application of these tests, and as a result I'm not sure if [these questions] would be at the front of many clinicians minds," he said.
In their study this week, Gerke and colleagues used a gene expression analysis platform, NanoString's nCounter, to analyze tumor samples of 327 patients — 95 African American patients and 232 European American patients — measuring across the 60 genes included in the three commercial tests.
The group wrote that recapitulating the gene signatures on the NanoString device, rather than having the three companies run samples using their own internal protocols, is certainly a caveat to the results. But Gerke described the choice as both a strength and limitation.
"Because the these commercial tests are run on different platforms we think they are likely to get better signal for the genes they have chosen. But we wanted to put them on the same footing, to agnostically ask the question how do these patterns differ by race group," he explained.
When they looked at the gene expression results, the Moffitt investigators saw that about half of the 60 genes were expressed at different levels among African-American men in the cohort compared to European American men, although the expression-level differences were not especially large in magnitude.
The team then translated the patterns into continuous risk prediction scores. According to the authors, the distribution of these estimated risk scores for the Prolaris signature and the Decipher signature were not significantly different for African American versus European American men.
For the Oncotype DX-based scoring though, risk predictions for the African-American men in the cohort were lower, on average, than those for the rest of the study subjects.
The issue this raises, Gerke and his colleague wrote, is that in both cases, the result seems to run contrary to what has been observed in the clinic, which is that African-American men have worse outcomes and are much more likely to die of prostate cancer than men with predominantly European ancestry.
Gerke said the study can't answer the question of whether this means that tests might be falsely returning low-risk genetic results to men who have a high risk of recurrence and aggressive disease. But considering the fact that these various gene signatures were developed, and validated, in predominantly European-ancestry cohorts, it hopefully adds to the rationale for studying this further.
“Until the risk prediction accuracy of the tests has been thoroughly studied in African American patients, we urge caution," he said in a statement. "We hope that our report escalates interest" in further population-specific validation, he added.
Similar questions in the breast cancer space have led to studies helping to more definitively confirm genetic test accuracy and utility across different racial and ethnic populations. For the breast cancer version of its Oncotype DX test, for example, Genomic Health, was recently able to confirm that both risk and therapy prediction held true for women of a variety of races and ethnicities, using prospective long-term outcome data from its TAILORx trial.
And companies have done similar work for prostate cancer. Myriad Genetics has shared data in the past breaking down Prolaris results specific to African-American men. A company spokesperson said prognostic performance of the test was demonstrated to be independent of race in a validation against long-term clinical outcomes in a large clinical cohort, nearly 40 percent of whom were African American.
Similarly to the Moffitt results, research using the actual Prolaris test found that the distribution of scores doesn't seem to differ by race. And when comparing test results to recorded 10-year outcomes, the predictive power of the test held just as true for African Americans in that cohort as it did for the study group as a whole, investigators wrote.
Genomic Health has also highlighted data showing the predictive value of its test to be similar in African Americans and Caucasians. In an emailed comment responding to the study today, a company representative specifically highlighted results published in the European Journal of Eurology in 2015 — in which researchers found that Oncotype DX was similarly predictive of outcomes in both groups.
According to the company, "drawing conclusions about the performance of Oncotype DX without actually using the commercialized assay to generate a validated GPS," has clear limitations that "cannot be ignored."
Finally, Decipher Biosciences (fomerly GenomeDx) has also recently presented data along these lines, including evidence that it said shows its test may predict poor outcomes even more accurately in African-American men than in Europeans.
According to Gerke though, emerging evidence of genetic and biological differences and the fact of racial outcome disparities both suggest the field needs to further study this including collecting prospective evidence and comparing platforms head to head more comprehensively.
"There is increasing evidence that African-American men have biologically and genomically distinct tumors in the prostate. And if that's true, it's important to know if these signatures that were developed, again, mainly in European american men ... translate over to this biologically distinct setting," Gerke said.
"It may well be that they work," he added, "but without long-term follow up ... we are kind of at the point that we're just not sure."
The Moffitt team is not alone in advocating for a deeper look. In a review piece published last year, for example, University of Minnesota researchers wrote that despite an absence of evidence that gene expression tests perform differently in African-American men, evidence for biological differences in cancers in this population "maintains the possibility that [such] tests may be affected."
Interestingly, some of the new data on genetic differences between prostate cancers of African-American and European men is emerging from Decipher Biosciences' own database program.
For example, a presentation last year at the annual meeting of the American Urological Association from a study conducted using Decipher's GRID repository showed that African American patients tended to have higher genetic-risk scores in low-grade tumors but lower risk scores in higher grade tumors, as well as higher expression of certain genes and immune-related pathways, and lower expression of things like androgen receptor and androgen receptor downstream targets.
Gerke said he and his colleagues, alongside other institutions, are "continuing to follow large diverse cohorts."
Among ongoing efforts at Moffitt, the cancer center has had a prospective observational study specific to the Decipher test running since 2016, with the goal of measuring accuracy in African Americans compared to non-African Americans.
Moffitt researchers have also worked with Decipher to establish gene expression biomarkers specific to the African American population, which could improve prognostic and recurrence risk evaluation.