Related Experiment Video
Updated: Jun 12, 2025
Microarray-based Identification of Individual HERV Loci Expression: Application to Biomarker Discovery in Prostate Cancer
Published on: November 2, 2013
Cross-Comparison Individual Patient-Level Analysis of Three Gene Expression Signatures in Localized Prostate in Over
Angela Y Jia1, Yilun Sun2, Maitry Patel1
1Department of Radiation Oncology, University Hospitals Seidman Cancer Center and Case Western Reserve University, Cleveland, OH.
Purpose:
The 22-gene Decipher genomic classifier (GC) (22-gene GC) is the only gene expression test with National Comprehensive Cancer Network (NCCN) level 1 evidence for localized prostate cancer (PCa) treatment decision making. It is unclear whether other commercial signatures-Genomic Prostate Score (GPS) or Prolaris (cell cycle progression [CCP])-correlate sufficiently to explain differences in evidence strength. This study assesses correlation between these three classifiers on a per-patient basis by performing a cross-comparison of these signatures in a large cohort of patients diagnosed with PCa.
Materials And Methods:
Primary PCa biopsy samples underwent whole-transcriptome gene expression microarray analysis. 22-gene GC scores were calculated using the commercially locked model. To reduce bias, GPS and CCP signatures were retrained for prediction of metastasis to harmonize end points. Pearson correlations and linear regressions (univariable/multivariable) adjusting for age, grade group, prostate-specific antigen (PSA), and T stage were calculated between signatures.
Results:
Among 50,881 patients (median age 68 years, median PSA 6.2 ng/mL), 60% were NCCN intermediate-risk. The GPS-derived and CCP-derived models had poor goodness-of-fit to 22-gene GC (R2 = 0.36 and 0.32, respectively). Multivariable analysis adjusting for clinical factors showed similar results. Given the many variables that may contribute to or correlate with the 22-gene GC, a multivariable R2 analysis was performed to assess how much of the variation could be attributed to known factors. A variance analysis revealed approximately 60% of 22-gene GC variation remained unexplained. GPS-derived and CCP-derived models accounted for 24.7% and 22.7% of variance, respectively, with additional contributions from Gleason score.
Conclusion:
Correlation between 22-gene GC and either GPS-derived or CCP-derived signatures is minimal to moderate. These tests are not interchangeable, and their use should be guided by the specific evidence supporting each signature.
