Abstract
PURPOSE: We aim to identify biomarkers of progression in an ongoing pilot trial using adaptive dosing in metastatic castrate sensitive prostate cancer (mCSPC). PATIENTS AND METHODS: Men with mCSPC were given combined androgen deprivation therapy with an androgen receptor signaling inhibitor followed by a treatment break after achieving >75% PSA decline. This was followed by evolution-informed drug cycling to prevent resistance outgrowth. Just 6 of 16 patients have progressed at month 55, and we wish to identify predictive features that would help identify these patients. We compare clinical features, testosterone metrics, and PSA metrics over the trial cohort, and we fit a mathematical model to the dynamic patient-specific data to gain more mechanistic insight as to what might be driving early progression. RESULTS: Significant clinical risk factors predicting progression included the risk status, the number of bone metastases, and the total number of metastases. However, analyses of the dynamical changes in PSA and testosterone during the initial cycle of treatment identified several useful features to discriminate early, late, and non-progressors. Specifically, a higher PSA at the end of the induction phase, a lower percent change in PSA over the induction period, and a higher average PSA/Testosterone ratio over the first cycle significantly predicted progressors. Thresholds for these metrics are able to distinguish the earliest progressors. A simple mathematical model coupling the PSA and the testosterone dynamics suggests a higher fraction of resistant cells exist prior to therapy for those that progress earlier. CONCLUSION: PSA/Testosterone ratio is a useful feature to couple drug response to tumor burden dynamics and predict early progression from the first cycle of adaptive therapy.