Ahead of a planned launch of its prostate cancer prognostic test, Decipher, later this year, GenomeDx presented data from four studies at the American Society of Clinical Oncology annual Genitourinary Cancers Symposium last week.
Three of the studies assessed the test's clinical validity and utility in predicting which patients are at risk for recurrence and metastasis after prostate cancer surgery. The fourth looked at how physicians changed their treatment decisions in cases where the results of the test differed from standard clinical risk assessment tools and algorithms.
Overall, the data from the first three studies suggested that Decipher can offer a more accurate prediction of metastasis than conventional risk assessment tools such as PSA, tumor stage and grade. The fourth study found that the test led to a change in treatment recommendation 43 percent of the time and a change from more aggressive treatment to observation in 31 percent of the cases evaluated where treatment was initially recommended. That study did not assess long-term outcomes, though the company plans to do so in future studies.
Doug Dolginow, GenomeDx's CEO, told Pharmacogenomics Reporter this week that the company is excited to demonstrate the test's usefulness through studies focused on different aspects of clinical decision making in prostate cancer after initial surgery.
"There are many specific clinical questions in managing patients with prostate cancer," he said. "And we think that all of those clinical questions have to be validated. That's why we have something like 22 studies with 4,000 patients underway.
"We are happy [with these first studies] to see that the technology is able to answer some of these specific clinical claims."
According to Elai Davicioni, GenomeDx's chief science officer, the three validation studies presented at the meeting were based on analysis of the same patient population, a random sampling of 219 subjects from a larger cohort of clinically high-risk patients treated at the Mayo clinic.
"By current guidelines, these are the guys that are candidates for secondary therapy," Davicioni explained. "So they're all at risk. But we know some will never actually develop potentially lethal cancer. We need a way to better identify among that population who actually needs secondary therapy after surgery."
"No matter how you look at pathology, or PSA, or other scoring methods, half of patients are misclassified [currently], so actual risk is poorly assessed," Dolginow added. In light of this, GenomeDx is hoping Decipher can be a useful tool for reducing the number of men misclassified as high risk, thereby avoiding unnecessary and costly treatment.
The company is currently preparing to launch the test, which measures the expression of 22 genes using microarrays, later this year. Dolginow said that early users are also already using the test in clinical practice in the context of a "utilization field trial."
GenomeDx is also conducting several studies to support an application to Palmetto GBA's MolDx program to receive Medicare coverage.
The company's main presentation at the ASCO symposium included prospective validation data from a study comparing Decipher's prediction of later metastatic disease progression to risk predictions using other individual clinical variables. In an abstract for the study, the company reported that Decipher showed a c-index of 0.79 in anticipating later disease progression, which was significantly better than any single clinical variable.
According to the authors, "cumulative incidence curves" in the cohort showed that for the 72 percent of patients with low-risk scores according to Decipher, only three percent and six percent suffered a progression to metastatic disease at five and 10 years post prostatectomy, respectively. For the 28 percent of patients with high-risk scores, meantime, "the cumulative incidence was 17 percent and 25 percent at five and 10 years post [surgery]," the group wrote.
A second study — spearheaded by co-authors at Johns Hopkins University interested in looking specifically at patients with biochemical signs of recurrence after surgery — examined Decipher's prognostic ability against clinical tools such as PSA doubling time and Gleason score in a subset of 110 patients from the same cohort that developed this biochemical recurrence.
"They wanted to look in a very defined population — [at] these men who had surgery and then had a PSA failure — to see if the test could predict metastasis better than an algorithm that measures PSA doubling time and other variables," Davicioni said.
That analysis found that the test did better at predicting metastatic progression than the clinical measures, with a c-index for predicting metastasis three years after a biochemical recurrence of 0.82.
In a third study, GenomeDx worked with collaborators at the University of California, San Francisco, to test Decipher against a relatively new clinical decision-making algorithm developed by researchers at the university, called CAPRA-S, short for "Cancer of the Prostate Risk Assessment Post-Surgical."
"This is recognized, at least in the literature, as the most powerful nomogram or clinical prediction tool that has been developed," Davicioni said. "So they wanted to look specifically at whether the genomic model can add information independently or does it need the clinical variables to get enough power to get a good result."
In this analysis the researchers found that both CAPRA-S and Decipher were significant predictors of later disease progression after surgery, but that the genomic test was able to "down-risk" more than fifty percent of men stratified to a high-risk category using CAPRA-S alone.
In the fourth, final study presented at the meeting, GenomeDx shared data on how physicians changed their treatment recommendations based on genomic testing results using Decipher. In the study, the researchers presented 12 cases — from a group of 240 pathologically high-risk post surgery prostate cancer patients — to each of twenty urologic oncologists from 18 institutions.
The researchers recorded the physicians' treatment decisions before and after access to Decipher testing results, and found that treatment recommendations changed overall in 43 percent of all cases, and from more to less treatment 31 percent of the time.
"We were specifically interested in that because high-risk patients typically might be given very expensive radiation that costs up to $60,000 or $70,000 per patient but with this information, physicians could choose to defer that treatment," Dolginow said.
He also said that the company, as it prepares to launch Decipher as a commercial prognostic tool, is also researching whether the gene expression signature at the heart of the test could also be used to predict whether patients are likely to respond to specific therapies like radiation or hormonal treatments.
The company faces competition from Myriad Genetics, which already markets its Prolaris gene expression test for prostate cancer recurrence, as well as Genomic Health, which plans to launch its own such test this year. Unlike Myriad's test, which the company has studied in both pre-surgery patients and post-prostatectomy patients, Decipher is targeted specifically at those deemed high-risk by clinical measures after prostate surgery.