ASTRO 2021: Validation of a 22-Gene Genomic Classifier in the NRG Oncology/RTOG 9202, 9413, and 9902 Phase III Randomized Trials: A Biopsy-Based Individual Patient Meta-Analysis in High Risk Prostate Cancer

( The 2021 American Society for Radiation Oncology (ASTRO) Annual Meeting’s included a session on biomarkers and salvage radiotherapy and discussion by Dr. Paul Nguyen regarding the validation of a 22-gene genomic classifier in the NRG Oncology/RTOG 9202, 9413, and 9902 phase III randomized trials.

High-risk prostate cancer accounts for 2/3 of the deaths from localized prostate cancer. Current radiation is a “one-size fits all” scheme of radiotherapy + long-term ADT, however the use of genomic testing to stratify patients into cohorts with higher and lower risk of metastases could allow for personalization of therapy. The purpose of the study presented by Dr. Nguyen and colleagues at ASTRO presented results validating the performance of the Decipher 22-gene genomic classifier in pre-treatment biopsy samples collected in three randomized phase III high-risk definitive radiotherapy trials: NRG/RTOG 9202, 9413, and 9902. Of note, these tissue samples were collected up to 29 years ago.

NRG/RTOG 9202 is a phase 3 trial of the use of long term total androgen suppression following neoadjuvant hormonal cytoreduction and radiotherapy in locally advanced carcinoma of the prostate. Patients were included if they were T2c-T4, PSA < 150 ng/mL, N0 (N1 only below the iliac lines), M0. Total androgen suppression included flutamide and zoladex, and radiotherapy included a total of 65-70 Gy for stage T2c (regional lymphatics only 44-46 Gy with 20-29 boost to the prostate) and a total of 67.5-70 Gy for stage T3 and T4. NRG/RTOG 9902 was a phase 3 protocol of androgen suppression and radiation therapy versus androgen suppression and radiotherapy followed by chemotherapy with paclitaxel, estramustine, and etoposide for localized, high-risk, prostate cancer. Patients were included if they had a PSA of 20-100 ng/mL and Gleason Score > 7 (any T stage) or clinical stage > T2 and Gleason Score > 8 (PSA <= 100) (M0, Nx or N0). Total androgen suppression included LHRH + bicalutamide or flutamide for 8 weeks then only LHRH post-radiotherapy, and radiotherapy was 70.2 Gy (regional lymphatics only 46.8 Gy with 23.4 Gy boost to the prostate). NRG/RTOG 9413 was a phase 3 trial comparing whole pelvic irradiation followed by a conedown boost to boost irradiation only and comparing neoadjuvant to adjuvant total androgen suppression. Patients were included if they were T2c-T4, Gleason Score 6+, PSA < 100 ng/mL, >15% risk of lymph node positivity (based on the Partin’s table), M0, and N1 must be below the pelvis proper. Total androgen suppression included flutamide and Zoladex or Lupron, and radiotherapy included 70.2 Gy (Arm 1 & 3: with whole pelvic irradiation; Arm 2 & 4: just the prostate only).

 For this study, after central review, the highest-grade tumors were profiled on clinical-grade whole transcriptome arrays and genomic classifier scores were obtained. Cox multivariable analyses were used to validate the independent prognostic ability of the genomic classifier for distant metastases, prostate cancer-specific mortality, and OS.

Baseline characteristics between the three trials were comparable. The distribution of genomic classifier scores in this study is as follows:



On univariable analysis, the genomic classifier was prognostic for distant metastasis (HR 1.30, 95% CI 1.18-1.43), prostate-cancer specific mortality (HR 1.31, 95% CI 1.17-1.43), and overall survival (HR 1.16, 95% 1.08-1.22). On multivariable analysis, the genomic classifier was still prognostic for distant metastasis (HR 1.24, 95% CI 1.11-1.39), prostate-cancer specific mortality (HR 1.27, 95% CI 1.13-1.43), and overall survival (HR 1.12, 95% 1.05-1.20). Additionally, the Forest plot demonstrates consistent findings for the genomic classifier across all endpoints and analysis cohorts. Stratified by low (genomic classifier <0.23), intermediate (genomic classifier 0.23-0.32), and high (genomic classifier > 0.32) risk groups, the cumulative incidences of outcomes by these genomic classifier risk groups was statistically significant for distant metastases (p < 0.001), prostate cancer specific mortality (p < 0.001), and OS (p = 0.002). The NRG GU-009/PREDICT-RT trial is assessing personalization by Decipher score:


The PREDICT-RT trial has a sample size goal of 2,478, which correlates to enrollment of 38 patients per month needed.

Dr. Nguyen concluded his presentation with the following take-home messages:

  • This is the first validation of any gene expression biomarker on pre-treatment biopsy samples from prospective randomized trials and demonstrates an independent association of the genomic classifier score with distant metastasis, prostate cancer specific mortality, and OS
  • High-risk prostate cancer is a heterogeneous disease stage and the genomic classifier can improve risk stratification to help personalize shared decision-making
  • The NRG/GU 009/PREDICT-RT (NCT04513717) will further determine the optimal therapy based on the genomic classifier score


Presented by: Paul Nguyen, MD, Department of Radiation Oncology, Dana Farber Cancer Institute, Boston, MA

Written by: Zachary Klaassen, MD, MSc – Urologic Oncologist, Assistant Professor of Urology, Georgia Cancer Center, Augusta University/Medical College of Georgia, @zklaassen_md on Twitter during the 2021 American Society for Radiation Oncology (ASTRO) Hybrid Annual Meeting, Sat, Oct 23 – Wed, Oct 27, 2021.