EURURO Vol. 72 No. 5

846 860

1. Introduction
Clinical risk stratification drives first-line therapy decisions
for patients with prostate cancer (PCa). However, risk
stratification based only on clinicopathologic features such
as preoperative prostate-specific antigen (PSA), clinical
stage, and Gleason score may not capture the full metastatic
potential of a patient’s tumor . Consequently, there
remains a clinical need to improve patient selection for
active surveillance, localized therapy (radical prostatec-
tomy [RP] or radiation therapy [RT]), and/or multimodality
therapy. In recent years, molecular tests have played an
increasing role in clinical decision making
Decipher is a 22-RNA feature biomarker that was
developed and validated to predict the risk of metastasis
. It has also been validated to predict the likelihood of
prostate cancer-specific mortality (PCSM) after RP
. Testing the Decipher genomic classifier on prostate
needle biopsy tissue is feasible and has been
evaluated in a small cohort of 57 primarily National
Comprehensive Cancer Network (NCCN) low- and interme-
diate-risk patients from Cleveland Clinic . In the current
study, we aimed to validate the performance of biopsy
Decipher to predict both metastasis and PCSM in a multi-
institutional cohort of men with NCCN low-, intermediate-,
and high-risk disease who received either first-line RP or
RT androgen deprivation therapy (ADT).
2. Materials and methods
2.1. Study cohort
We identified 235 patients treated with either first-line RP
or first-line RT ADT at the discretion of their treating
physician from 1987 to 2014 with available genomic
expression profiles generated from diagnostic biopsy speci-
mens from seven tertiary referral centers: Dana-Farber/
Brigham and Women’s Cancer Center, Cleveland Clinic, Johns
Hopkins, MD Anderson Cancer Center, University of California
San Diego, University of Michigan, and University of Miami.
Entry into the RT ADT cohort required NCCN intermediate-
or high-risk disease. The RP cohort was drawn from a bank of
patients who had adverse pathology at surgery (defined as
either preoperative PSA > 20 ng/ml, stage pT3 or margin
positive, or RP Grade Group 4), the vast majority of
whom had presented with intermediate- or high-risk PCa.
Patient clinical and genomic data were deposited into the
GenomeDx PCa genomic resource information database
(GRID); institutional review boards at the participating
institutions approved the research protocol under which
the data were collected .
Our initial cohort included 909 patients who met the
eligibility criteria of our study. However, 674 patients did
not have available biopsy tissue or had insufficient RNA
available for analysis. Thus, we compared the clinical
characteristics of patients with genomic profiles versus
those without. Analysis demonstrated comparable pre-
treatment PSA levels and clinical stage between the
genomic and nongenomic patient cohorts, but showed that
patients with available genomic profiles had significantly
higher biopsy Grade Groups (Grade Group 4–5, p < 0.001)
and thus significantly higher NCCN risk groups ( p < 0.001;
Supplementary Table 1).
The primary endpoint of the study was incidence of
clinical metastasis (regional or distant) documented radio-
graphically on computed tomography or bone scan. The
secondary endpoint of the study was PCSM.
2.2. Specimen collection and handling
Specimen selection and processing has been described
previously Following microarray quality control using
the Affymetrix Power Tools packages probeset
summarization and normalization was performed utilizing
the single channel array normalization algorithm . None
of these samples were used in the development of the
Decipher genomic classifier .
2.3. Calculation of NCCN, Cancer of the Prostate Risk
Assessment, and Decipher
NCCN risk groups were calculated as described previously
. Cancer of the Prostate Risk Assessment (CAPRA) was
calculated using age at diagnosis, pretreatment PSA, biopsy
Grade Group, clinical stage, and percent of positive cores
. Decipher score was obtained based on a previously
locked 22-biomarker random forest algorithm Of note,
the Decipher score is based purely on the RNA analysis and
does not incorporate clinical data. Previously established
cut-points (3 and 5 for Cancer of the Prostate Risk
Assessment and 0.45 and 0.60 for Decipher) were used
for categorical analysis of the risk models .
2.4. Statistical analysis
The significance level was set at 0.05 for all statistical tests
and analyses were performed in R version 3.3 (R Foundation
(hazard ratio: 1.57 per 10% increase in score, 95% CI: 1.03–2.48, p = 0.037), with a 5-yr PCSM
rate of 0%, 0%, and 9.4% for Decipher low, intermediate, and high, respectively.
Conclusions: Biopsy Decipher predicted metastasis and PCSM from diagnostic biopsy
specimens of primarily intermediate- and high-risk men treated with first-line RT or RP.
Patient summary: Biopsy Decipher predicted metastasis and prostate cancer-specific
mortality risk from diagnostic biopsy specimens.
# 2017 European Association of Urology. Published by Elsevier B.V. This is an open access
article under the CC BY-NC-ND license .
E U R O P E A N U R O L O G Y 7 2 ( 2 0 1 7 ) 8 4 5 – 8 5 2 846