The report is a product of the VA/HSR&D Evidence Synthesis Program.
Prognostic Value of Genomic Classifier Testing for Prostate Cancer:
A Systematic Review
Takeaway: Genomic classifier tests for prostate cancer may offer some additional prognostic benefit over existing clinically-based risk schemas and may be helpful when management decisions are uncertain. However, definitive evidence of the prognostic ability of these tests is still needed from current management-era data.
Prostate cancer is the most common malignancy in men, with an estimated 268,490 new cases in the United States in 2022, and 12,500 new diagnoses annually within the VA healthcare system. A major challenge for prostate cancer management is identifying patients who would benefit from treatment and tailoring the intensity of that treatment to personalized risk assessments. Individualized prognosis – beyond clinically-based risk stratification schemas – could inform patient-physician decision-making, reduce unnecessary overtreatment, and improve patient outcomes. A relatively recent advancement in prostate cancer risk stratification is the development of commercially available, tissue-based genomic classifiers.
The VA Evidence Synthesis Program (ESP) Center located at the Durham VA Medical Center conducted a systematic review to evaluate the impact of genomic classifier tests on risk reclassification, treatment recommendations, and key clinical outcomes among patients with prostate cancer at the time of diagnosis and after definitive initial treatment. The genomic classifier tests studied included: Decipher, Oncotype DX GPS (now named Genomic Prostate Score but referred to in this report as Oncotype), and Prolaris. Investigators searched MEDLINE (via Ovid), Embase (via Elsevier), and Web of Science (via Clarivate) databases from 2010 to the present. From more than 1,500 potentially relevant articles, 59 articles (55 unique studies) were included in this review. The 55 studies consisted of 1 randomized controlled trial, 1 secondary analysis of a randomized controlled trial, 1 individual patient-level meta-analysis, 2 case-control studies, and 50 observational cohorts (8 prospective, 42 retrospective). Of the 55 studies, 4 were conducted solely within VA and 7 additional studies included a VA cohort.
Summary of Findings
Overall, investigators found that three genomic classifier tests (Decipher, Oncotype DX GPS, and Prolaris) seemed to provide modest additional prognostic information over existing clinical risk prediction schemas with respect to biochemical recurrence, development of metastatic disease, and prostate-cancer-specific mortality. More specifically:
- Eleven studies (6,953 patients) reported risk reclassification (e.g., low risk to intermediate risk) either as a direct comparison or through integration of the genomic classifier test results with a clinical feature-based risk assessment.
- While there was a wide range of impacts of these tests on risk reclassification across studies, no change was found in risk classification for a majority of patients across test types: Decipher (21% to 51%; 3 studies), Oncotype (37% to 81%; 6 studies), and Prolaris (58%; 1 study).
- Across 14 observational studies (2,561 patient cases) at the time of diagnosis and after prostatectomy, treatment recommendations often changed after receipt of genomic classifier tests. However, in the single randomized trial (191 patients) evaluating the impact of Oncotype test results into treatment decisions, there was no statistically significant effect.
- Prolaris or Oncotype test results received prior to definitive treatment led to a change in management 16% to 65% of the time, while studies using Decipher tests noted that a higher-risk score increased the odds of having a change in treatment post-prostatectomy compared to getting a lower-risk score.
- Thirty-nine studies, including more than 10,000 patients, demonstrated small improvements in prognostic ability when incorporating genomic classifiers into clinical risk stratification schemes despite variability in the risk classification schema used, outcome of interest, and treatment time frame.
Significant limitations of the existing literature include that it was largely drawn from patients diagnosed and treated prior to the current era of prostate cancer management (prior to 2000). Investigators were unable to compare across genomic test types, as no identified literature provided a direct comparison. In addition, no study reported harms due to genomic classifier testing.
Implications for VA
Across all included studies, the patient populations were similar to VA patients based on patient characteristics (e.g., age, comorbidities). Several of the identified studies included VA-sourced data. These findings are expected to be generalizable to the VA clinical setting.
To strengthen the body of evidence outlined in this report, investigators suggest the following considerations for future research.
- Randomized trials would be preferable to determine if these tests can predict response to treatment.
- Additional studies to evaluate Oncotype and Prolaris in patients after definitive treatment – and to evaluate Decipher in patients prior to definitive treatment are needed.
- Studies offering direct comparisons across these tests could inform comparative value.
- Harms from the use of these tests should be reported as an important outcome. Any future studies should provide explicit descriptions of the source of cohort data, outline attrition rates from cohort populations due to inadequate tissue samples or test results, and employ a standardized set of core confounding measures for analytic adjustment.
Boyer MJ, Carpenter D, Gingrich JR, et al. Prognostic Value of Genomic Classifier Testing for Prostate Cancer: A Systematic Review. Washington, DC: Evidence Synthesis Program, Health Services Research and Development Service, Office of Research and Development, Department of Veterans Affairs. VA ESP Project #09-010; 2023.
To view the full report, go to https://vaww.hsrd.research.va.gov/publications/esp/genomics-master.cfm (Intranet only).
ESP is currently soliciting review topics from the broader VA community. Nominations will be accepted electronically using the online Topic Submission Form. If your topic is selected for a synthesis, you will be contacted by an ESP Center to refine the questions and determine a timeline for the report.
This Management e-Brief is provided to inform you about recent HSR&D findings that may be of
interest. The views expressed in this article are those of the authors and do not necessarily
reflect the position or policy of the Department of Veterans Affairs. If you have any questions or
comments about this Brief, please email CIDER.
The Center for Information Dissemination and Education Resources (CIDER) is a VA HSR&D Resource
Center charged with disseminating important HSR&D findings and information to policy makers, managers,
clinicians, and researchers working to improve the health and care of Veterans.