Related Experiment Video
Updated: Jul 13, 2026
miRNA Expression Analyses in Prostate Cancer Clinical Tissues
Published on: September 8, 2015
Prediction Analysis of Microarray of 50 genes (PAM50) classifier validated for predicting prostate cancer progression
Ankur Malpani1, Jonathan T Ryan2, Hui Yu3
1Sylvester Comprehensive Cancer Center, Miller School of Medicine, Desai Sethi Urology Institute, University of Miami, Miami, FL, USA.
Objectives:
To evaluate basal-luminal cell of origin subtyping using Prediction Analysis of Microarray of 50 genes (PAM50) genomic classification profiles for predicting disease progression in men undergoing active surveillance (AS) for prostate cancer (PCa).
Patients And Methods:
In the prospective Miami Active Surveillance Trial (MAST) trial, 205 men undergoing AS received serial multiparametric magnetic resonance imaging (MRI) and biopsies, including MRI-targeted and systematic sampling. The highest-grade core from each biopsy was sent for expression profiling using Decipher, a clinical-grade transcriptome assay (Veracyte Inc., San Diego, CA, USA). Basal-luminal subtyping was evaluated using PAM50 molecular subtype models. PCa grade progression was compared across subtypes, as were gene mutation signatures, prognostic indices, and pathway activities. Kaplan-Meier curves, log-rank test, and multivariable Cox regression were used to assess association between PAM50 and grade progression. Heatmaps and volcano plots were rendered to illustrate potential mechanistical differences between PAM50 subtypes.
Results:
Of the 205 patients, 128 had transcriptome data for baseline basal-luminal classification. PAM50 identified 46 Luminal A (LA), 26 Luminal B (LB), and 56 Basal subtypes. Decipher scores were lowest in LA, followed by Basal, and highest in LB. Grade progression-free survival was worse in patients with the LB subtype (median 1.7 years) compared to those with LA and Basal subtypes (median 2.9 years; log-rank P = 0.005); LB patients had grade progression-free survival of 34% by 24 months of AS, compared to 63% for Basal or 68% for LA. Transcriptome analysis showed distinct enrichment profiles for each subtype, with LB strongly associated with SPOP and CHD1 mutations. Limitations include small sample size and single-institution setting.
Conclusion:
The PAM50 basal-luminal subtyping shows promise as a molecular classification tool for predicting progression risk in PCa. This is one of the only prospective studies evaluating PAM50 subtyping for predicting cancer progression in a cohort of men undergoing AS for PCa.
