SAN FRANCISCO — There are “notable differences” in the oncologic outcomes predicted by the three leading prostate cancer genomics tests, according to a small retrospective study of Oncotype Dx, Prolaris, and Decipher clinical usage.
“You can do a Prolaris test on one man and tell him you think he should undergo active surveillance, and then do an Oncotype on the same man and tell him he may need treatment,” senior study author Joseph Wagner, MD, from the Hartford Healthcare Medical Group in Connecticut, told an audience here at the American Urological Association 2018 Annual Meeting.
The comparative study also “highlights the difficulty of interpreting genomic tests for prostate cancer,” Wagner and his colleagues write in their meeting abstract.
“I need a lot of help figuring out how to use these tests,” Wagner acknowledged during a meeting press conference.
Traditionally, clinicians have used Gleason score, prostate-specific antigen (PSA) level, clinical staging, the number of positive needle cores, and related measures to risk-stratify patients with biopsy-confirmed disease and to help with subsequent treatment and management decisions, including active surveillance (AS), he explained.
But it is now becoming increasingly common for clinicians to use genomics tests, also known as molecular tests, to predict outcomes and guide treatment for favorable-risk prostate cancers (including very low, low, and intermediate [GS 3+4] risk).
Genomics tests incorporate both traditional measures and genetic data. However, discrepancies can exist among genomics tests in terms of their recommendations for treatment or for AS, said Wagner.
In their retrospective chart review, the research team identified 22 prostate cancer patients who had undergone at least two of the three genomic tests at Hartford Hospital from 2014 to 2017. “Multiple testing was by patient choice, not physician recommendation,” Wagner emphasized.
Most patients undergo only one test, but multiple tests facilitated the aims of the study, he noted.
Study participants were first assessed to determine whether they qualified for AS on the basis of National Comprehensive Cancer Network (NCCN) guidelines (i.e., traditional measures). The results of each of their genomics tests were then compared with the NCCN determination to see if they were in agreement.
Of the 22 participants, 21 met the NCCN criteria for AS.
There was 75% agreement between the 20 Prolaris tests and NCCN recommendations for AS, and the k score, which indicates whether a measure is above or below chance, was 0.21 (P = .117). In other words, Prolaris and NCCN agreed on active surveillance 75% of the time for these patients, which is the highest degree of conformity among the three tests.
For the 15 Decipher tests, there was 60% agreement and the k score was 0.15 (P = .268). And for the 10 Oncotype DX tests, there was 50% agreement but, because there were too few patients, the k score could not be calculated.
The results “don’t mean Prolaris is a better test than Decipher,” Wagner cautioned.
Instead, the team concluded that the majority of patients (21 of 22) met the NCCN criteria for AS and that the analysis “suggests Polaris is most apt to confirm this [NCCN] recommendation while Oncotype DX is more likely to go against it.”
The findings were not statistically significant, and the k scores indicate that the results might be the result of chance, said Wagner. However, they do show that there are potential differences in the test outcomes, he noted.
He acknowledged that the study was not scientifically rigorous and was conducted to more to provoke questions about the utility of the tests.
There’s no clear evidence of a superior test.
Because there are no head-to-head studies of these genomics tests, “there has been very little reason necessarily to choose one over another,” said Stacy Loeb, MD, from New York University in New York City, who moderated the press conference. “There’s no clear evidence of a superior test.”
Clinicians tend to use the test that is preferred at their institution, if they use one at all, Loeb explained.
Mark Stovsky, MD, from the Cleveland Clinic, who attended the press conference, said that his institution predominantly uses the Oncotype DX test, which was tested at the Ohio clinic.
The three tests provide reports that have largely converged in their analyses, said Loeb. But they still have some differences.
For Decipher and Prolaris, a “favorable” clinical prediction (indicating suitability for AS) is defined as a 3% or less likelihood of 10-year prostate cancer mortality. For Oncotype DX, it is defined as a more than 70% likelihood of organ-confined, grade group 1 or 2 disease at surgery (determined on pathology).
In everyday practice, using more than one test with a patient can become tricky and confusing, Wagner noted.
He described a recent patient with an elevated PSA (7 ng/mL), a favorable intermediate-risk Gleason score (7; i.e., 3+4), and only one positive needle core out of 12. Prolaris indicated a projected 10-year prostate cancer mortality rate of 2.4%, which is considered favorable and thus appropriate for AS. However the man pushed for another test, the Oncotype DX, which indicated a 39% risk for unfavorable pathology.
“He sees that Oncotype [score] and goes, ‘Holy crap, I’ve got to get my prostate out’,” said Wagner. “That’s what can happen when you give more than one test.”
The tests cost $5000 to $6000, most of which is covered by Medicare and private insurance. However, when asked by a reporter if these costly tests have ever been shown to actually improve outcomes, such as disease-specific or overall survival, Loeb answered no.
Dr Wagner reports financial ties to Genomic Health, the maker Oncotype DX. Dr Loeb reports ties to Lilly, MDx Health, GenomeDx, General Electric, Astellas, Sanofi, Minomic, and Boehringer Ingelheim.
American Urological Association (AUA) 2018 Annual Meeting: Abstract PD06-09. Presented May 18, 2018.
Follow Medscape senior journalist Nick Mulcahy on Twitter: @MulcahyNick
For more from Medscape Oncology, follow us on Twitter: @MedscapeOnc
Tidak ada komentar:
Posting Komentar