Introduction:

Tissue based prognostic signatures have been used in various stages of prostate cancer, including biopsy tissue from men who are deciding between active surveillance versus treatment. Recent studies have suggested that these tests may be affected by tumor heterogeneity impacting their performance for appropriate risk stratification. We evaluated the performance of tissue based prognostic signatures on biopsy tissue from two ongoing prospective trials in men with prostate caner at the University of Miami.

Methods:

The Miami Active Surveillance Trial (MAST) enrolls low–intermediate risk men who undergo an MRI and annual biopsy for three years, with positive cancers cores sent for genomic sequencing. The BlastEM trial includes intermediate-high-risk men who undergo targeted biopsy at the time of fiducial marker placement to assess for dose escalation during radiotherapy, in which both diagnostic and targeted biopsies are sent for genomic sequencing. This study reports on preliminary results from both trials on men whose biopsy cores were sequenced at GenomeDx, and evaluated for three commercially available gene signatures (Genome Dx Decipher Genomic Classifier, Oncotype DX Genomic Prostate Score, and Myriad Prolaris Cell Cycle Progression score). The biopsy cores from both trials represent either the diagnostic biopsy or a subsequent biopsy done within 12 months of diagnosis. 

Results:

The current cohort consists of 78 men from the MAST (n=46) and BlastEM (n=32) trials whose 231 biopsy cores were sent to GenomeDX for genomic sequencing. Among the 78 patients, 40 had Grade Group 1, 15 had grade Group 2, 10 had Grade Group 3 and 13 had Grade group 4 and 5 prostate cancer.  We found that for each signature there was a trend toward higher scores with higher grade groups, however each signature displayed a significant degree of variation within each grade group (p<0.001). (Figure 1). When assessing genomic scores from different cores we found significant variability (Figure 2), with the level of genomic risk within each biopsy session changing by  25-62% depending on which core was sequenced and which signature was used. When restricting to MAST, which includes active surveillance patients whose biopsy tissue these tests are currently being used, we found that level of genomic risk changed in 10-57% of cases depending on which core was sequenced and which signature was used.

Conclusion:

We assessed the performance of three commercially available genomic prognostic markers in men enrolled on two ongoing prospective trials whose biopsy tissue was sequenced at GenomeDx and found a significant degree of genomic heterogeneity between biopsy cores This may have implications when considering the reliability of these signatures in biopsy tissue. 

Funding: N/A