Association of transcriptomic biomarkers with adverse pathologic features at radical prostatectomy among active surveillance eligible men.

Abstract

338 Background: Active surveillance (AS) is the preferred management for men with NCCN very low or low risk prostate cancer (PCa) and may be considered for select favorable intermediate risk PCa. AS can be associated with significant grade reclassification and progression to treatment, adverse pathology at radical prostatectomy (RP), and need for salvage therapy. Genomic transcriptomic biomarkers have been shown to be prognostic and potentially predictive of treatment response in the locally advanced and metastatic setting. The potential of transcriptomic biomarkers to identify durable AS candidates and elucidate a potential role for molecular systemic therapies remains unclear. We sought to define transcriptomic markers associated with adverse pathologic features among AS-eligible patients who underwent RP. Methods: We retrospectively identified AS-eligible patients at our institution from February 2012 to March 2023 who underwent RP and completed Decipher genomic classifier testing (n=112). AS-eligible was defined as NCCN very low risk to favorable intermediate risk (no more than one intermediate risk factor, GG1 or 2, and <50% cores positive). The primary outcome was adverse pathologic features (APF) at RP, defined as GG3-5, pT3b, or pN1 disease. Associations between APF and previously defined transcriptomic signatures of interest derived from the Decipher GRID (i.e., PAM50, PSC, PTEN, and eleven prognostic signatures) were tested using Mann-Whitney U and Fisher’s exact tests and multivariable logistic regressions adjusted for log2(PSA) and use of MRI. Statistical significance was defined as p<0.05. Results: Of 112 men, 27% (30/112) had very low or low risk PCa and 73.2% (82/112) had favorable intermediate risk PCa. At RP, 28% (31/112) had APF, the majority due to upgrading to GG3-5 (28/31, 90%) and 10% (3/31) due to upgrading and T3b disease (none had pN1 at RP). Men with APF had higher baseline PSA (6.42 ng/mL vs 4.9 ng/mL, p=0.02) and were less likely to have undergone an MRI fusion biopsy for diagnosis (71% vs 90%, p=0.02). NCCN risk group, PSA density, and core positivity were not significantly associated with APF at RP. Seven of eleven previously characterized prognostic gene-based signatures were associated with APF on multivariable analysis. PTEN loss and increased activated tumoral CD4 activity were also significantly associated with APF. Conclusions: Despite similar clinical features at diagnosis, AS-eligible patients with similar clinical variables have significantly different outcomes and risk of APF due to molecular heterogeneity. Prognostic gene expression signatures such as the Decipher genomic classifier as well as PTEN loss and activated CD4 activity measured at diagnosis are associated with APF among AS eligible patients. Clinical trials within AS populations should consider utility of genomic signatures in patient selection.