Paul L. Nguyen, MD
Overall, 23.4% of those classified as high-risk by the test developed metastasis. The PCSM was 9.4% at 5 years. Additionally, the findings for Decipher held up regardless of the frontline therapy utilized. The next steps in the research will be to conduct a clinical trial using the test to determine when intensification of therapy is warranted, according to lead author Paul L. Nguyen, MD.
"The Decipher classifier obtained from biopsy samples was associated with distant metastases and PCSM after radical prostatectomy or radiation therapy and androgen deprivation therapy," said Nguyen, from the Brigham and Women’s Hospital and the Dana-Farber Cancer Institute. "Further validation is ongoing in larger cohorts. Work is planned to study the test in completed randomized trials to determine predictive value."
For the study, diagnostic biopsy samples were obtained from 175 patients from the Cleveland Clinic, Brigham and Women’s Hospital, and Johns Hopkins. Testing with Decipher was used on the highest-grade core from the biopsy. Overall, by NCCN classification, 87% of patients had intermediate (50.9%) or high-risk (33.7%) prostate cancer. Forty-three percent of patients had received frontline radical prostatectomy and 57% were treated with frontline radiation therapy (RT) plus androgen deprivation therapy (ADT).
After 6 years of follow-up, 32 patients had developed metastases and 11 had died from prostate cancer. Of those classified as intermediate- and low-risk by Decipher, 9.3% and 5.0% developed metastasis at 5 years, respectively. By NCCN alone, 33.1% of those with high-risk disease developed metastasis, as did 11% and 1.2% of those with intermediate- and low-risk, respectively.
Use of Decipher improved the overall c-index for metastasis compared with risk levels alone. The 5-year post-biopsy c-index was 0.74 for Decipher alone and 0.75 for Decipher and NCCN combined compared with 0.66 with NCCN risk classification only. "[Decipher] adds to what we already know and enhances our ability to decide which patients are going to develop metastases," said Nguyen.
By univariate and multivariate analysis, the only statistically significant variable for predicting metastases was Decipher (univariate HR, 1.52; 95% CI, 1.24-1.88; P <.001; multivariate HR, 1.39; 95% CI, 1.09-1.77; P = .0069). The investigators also assessed Gleason score, clinical stage, and type of frontline treatment, but none were statistically significant.
The use of Decipher also improved upon NCCN staging for PCSM. Those listed as being genomic high-risk had a PCSM of 9.4%. Those in the intermediate- and low-risk groups had a PCSM of 0%. For every 10% increase in Decipher score, the risk of PCSM increased by 57% (HR, 1.57; 95% CI, 1.07-2.40; P = .02).
The low number of overall mortality prevented a multivariate analysis of survival. However, "by univariate analysis, Decipher was the only factor associated with PCSM," said Nguyen.
The Decipher test utilizes the expression levels of 22 RNA biomarkers. The test has been explored on samples from over 2000 patients in clinical studies following surgery for men with prostate cancer. In these studies, 60% of those classified as intermediate and high-risk were reclassified as low-risk. Of those reclassified, 98.5% had not developed metastasis after 5 years of assessment post radical prostatectomy. The c-index for Decipher across trials remained around 0.75.
Nguyen PL, Haddad Z, Lam LLC, et al. Evaluation of the Decipher prostate cancer classifier to predict metastasis and disease-specific mortality from genomic analysis of diagnostic prostate needle biopsy specimens. J Clin Oncol. 2017;35 (suppl 6S; abstract 4).