Introduction: OncotypeDx, Prolaris, and Decipher have each been validated to predict outcomes and guide treatment for patients with clinically localized prostate cancer, but they have yet to be compared to one another. Here we assess the correspondence between the results of each.
Materials and methods: We performed a retrospective chart review to identify patients who underwent at least two of the three genomic tests at Hartford Hospital between 2014 and 2017. We used test-specific definitions of a favorable prediction for each to compare the percent agreement between each pair. Results were also compared to treatment recommendations based on current National Comprehensive Cancer Network (NCCN) guidelines. We compared pair-wise agreement using Cohen's kappa (K).
Results: Twenty-two patients received at least two different genomic tests. For 12 patients who received both the Decipher and Prolaris, % agreement and K were 66.7 and 0.31 (p = .276), respectively. For 8 patients who received both Prolaris and Oncotype DX, % agreement and K were 75 and 0.39 (p = .168), respectively. Two patients received both Decipher and Oncotype DX, yielding 50% agreement and an incalculable K. For Prolaris versus NCCN, % agreement and K were 75 and .21, respectively (p = .117; n = 20). For Decipher versus NCCN, % agreement and K were 60 and .15, respectively (p = .268; n = 15). For Oncotype DX versus NCCN (n = 10), agreement was 50%, K was incalculable.
Conclusions: Notable differences exist in prognostic outcomes obtained from OncotypeDx, Prolaris, and Decipher.