Abstract
Despite prostate MRI, image-guided biopsies, and other improvements in Active Surveillance, pathologic reclassification still occurs in 11% to 32% of confirmatory biopsies in contemporary series. Therefore, we evaluated if the Decipher Genomic Classifier (GC) and other genomic signatures in biopsy cores with Gleason Group (GG) 1 cancer can predict coexisting higher-grade disease elsewhere in the prostate among participants of the Miami Active Surveillance Trial (MAST).
style="margin-bottom: 12pt; text-align: justify; line-height: 150%">MAST enrolled 205 men with low and favorable intermediate-risk prostate cancer undergoing AS. Participants underwent a multiparametric MRI and confirmatory biopsy at enrollment, and annually thereafter for three years. Positive biopsy cores underwent GC version 1.2 profiling, and Genomic Prostate Score (GPS) and Cell Cycle Progression (CCP) signatures were also derived based on the normalized (to a 0.0 to 1.0 scale) expression of the specific target genes included in these tests. The primary objective was to compare GC scores in GG1 cores from prostates with (positive) vs. without (negative) coexisting GG2+ cancer elsewhere. Similar comparisons were made using the derived CCP and GPS. As secondary objectives, the Decipher GC, derived GPS, and derived CCP were also compared regarding 1) coexisting GG3+ cancer elsewhere and 2) the highest-volume GG1 core from each biopsy session. Neither patients, providers, nor investigators had access to genomic data that could affect clinical decisions throughout the trial.
style="margin-bottom: 12pt; text-align: justify; line-height: 150%">141 men contributed with 324 cores harboring GG1 cancer with successful genomic analysis. Men with coexisting GG2+ in at least one biopsy were older (median 69 vs. 62 years, p=0.003), had higher PSA density (median 0.19 vs. 0.10, p=0.008), and higher baseline 4K scores (median 47.0% vs. 20.5%, p=0.001). There was no significant difference between groups in GC or CCP scores regarding the primary objective, whereas median (IQR) GPS was 0.14 (0.08–0.24) in positive vs. 0.10 (0.07–0.16) in negative groups (p=0.018) – Figure 1. Figure 2 shows genomic scores distribution for each core regarding coexisting GG2 cancer elsewhere in the gland. For coexisting GG3+ elsewhere, GC or CCP scores again showed no difference, while GPS was 0.24 (0.19–0.29) vs. 0.10 (0.07–0.16) in positive and negative groups, respectively (p=0.005). The three tests did not differ when considering only the highest-volume GG1 core for expression profiling.
In a prospective AS trial, the genomic classifiers – Decipher GC and the derived CCP and GPS signatures – showed no clinically relevant differences between GG1 biopsy cores with and without coexisting higher-grade cancer elsewhere in the prostate. The slightly higher values observed in the GPS signature are insufficient to support any clinical application but merit further investigation. Findings suggest that prostate cancer foci are heterogeneous and genomically independent, and results from any particular core are more reflective of the region sampled, rather than informing on the risk of missed higher-grade cancer elsewhere in the gland