Validation of PAM50 for predicting progression in active surveillance: Results from the Miami MAST prospective clinical trial.

238 Background: Active surveillance (AS) has emerged as an increasingly common management strategy for men with low- to favorable intermediate-risk prostate cancer. Although rates of prostate cancer mortality and metastasis remain low, some men experience cancer progression that necessitates treatment. Tools that improve the prediction of progression can enable more risk-adapted and timely treatment. We evaluated a PAM50 molecular classification profile for predicting cancer progression in men undergoing AS for prostate cancer. Methods: A total of 205 men enrolled in the Miami MAST trial who underwent a follow-up protocol involving serial multiparametric MRI and biopsies, including MRI-targeted and systematic biopsies. The highest-grade cores from each targeted and systematic biopsy were sent to Veracyte for genomic profiling. Patients were categorized into three groups based on the PAM50 genomic profile from the Decipher GRID. Time to progression, mutation analysis, and other prognostic signatures available on the Decipher GRID were compared across the three PAM50 classifier groups. Statistical analysis was performed using ANOVA and the rank-sum test, with significance defined as p<0.05. Results: Among the 205 patients enrolled in the trial, 128 had successful genomic profiling for baseline PAM50 classification. 46 were classified as Luminal A, 26 as Luminal B, and 56 as Basal subtypes. The Luminal B subtype demonstrated the highest risk of progression (77%), while the Basal subtype showed the lowest risk (45%) (p=0.011). Median time to progression was shorter in the Luminal B subtype (1.7 years) compared to the Luminal A and Basal subtypes (2.9 years each) (p=0.005). Decipher scores were lowest in the Luminal A group, followed by the Basal group, and highest in the Luminal B group (p=0.0015). Intra-patient variability based on subtyping of different cores within the same biopsy was observed in 37.1% of cases. Transcriptome analysis revealed distinct enrichment profiles for each PAM50 subtype, and mutation pattern analysis highlighted differences in mutation associations, with Luminal B showing a stronger association with SPOP and PTEN mutations. Conclusions: PAM50 shows promise as a molecular classification tool for predicting the risk of progression in prostate cancer. Given its stronger association with SPOP and PTEN mutations, as well as associations with higher progression risk, shorter progression times, and higher Decipher scores, the Luminal B subtype indicates a poorer prognosis compared to the Luminal A and Basal subtypes. This is the first study to validate PAM50 for predicting cancer progression in a prospective cohort of men undergoing active surveillance for prostate cancer. Clinical trial information: NCT02242773 . Progression rates among different PAM50 subtypes. PAM50 Group Non-progressor (%) Progressor (%) P VALUE BASAL 31 (55) 25 (45) 0.011 LUMINAL A 26 (57) 20 (43) LUMINAL B 6 (23) 20 (77)