The Personalized Medicine for Prostate Cancer (PMPC) Study: The role of race, genomics, and patient complexity in treatment outcomes.

Abstract

302 Background: Matching of patients with optimal treatment for localized prostate cancer (PCa) requires accurate estimates of risk for progression or recurrence. Traditional risk assessment methods may underestimate risk for certain patient subgroups, (e.g. minorities). In the PMPC study, we replicated an expanded risk model, adding genomic data to improve risk prediction and optimize treatment assignment for PCa. Methods: A prospective cohort of 660 patients with early PCa was identified from 3 teaching hospitals and 2 VA Hospitals in Southern California. Risk model data were collected from electronic medical records, patient questionnaires and tissue-based genomic data provided by GenomeDx. Complexity variables included demographic characteristics, clinical markers (PSA levels, Gleason score), patient-reported health status measures (e.g. comorbidity burden, depression, stress) and Decipher score. We compared individual components of the complexity score for 425 Non-Hispanic White (NHW) and 104 African American (AA) men using t-tests. Composite complexity scores were derived from weights from general linear models. Mean complexity scores were compared using t-tests. Results: Compared with NHW men, AA men had statistically significantly more comorbidity, poorer health ratings, poorer physical functioning, more depression, less energy, more fatigue, more stress and less resilience. More AA men had Gleason scores >6 (60.0%) than NHW men (46.6%), p=.0134. Fewer AA men (41.5%) had ‘low risk’ Decipher scores compared to NHW men (53.8%). More AA men had high complexity scores compared with NHW men (63.4% vs. 45.1%, p=.008), however, there were no statistically significant differences in the proportion of NHW vs. AA men on active surveillance, 46.0% vs. 40.7%, p=.384. Using clinical variables alone there were no significant differences in low risk between NHW and AA men. Conclusions: The addition of genomic data improved the complexity model developed earlier. AA men may be at higher risk for suboptimal treatment than predicted by clinical variables alone. Data from the longitudinal cohort will test accuracy of model prediction for disease progression or recurrence. Clinical trial information: NCT03770351.