Development of risk groups in metastatic castration-resistant prostate cancer (mCRPC) to facilitate the identification of active chemotherapy regimens.

5137 Background: Our aim was to develop and validate clinically applicable predictive factors for a >30% PSA decline 3 months following chemotherapy initiation, and to assess the performance of a risk-group based classification in predicting PSA declines and overall survival (OS) in men receiving chemotherapy for mCRPC. METHODS: In TAX327, 1006 men with mCRPC were randomized to receive docetaxel (D) in two schedules, or mitoxantrone (M), each with prednisone: 989 provided data on PSA decline at 3 months. Predictive factors for a >30% 3-month decline in PSA levels were identified using multivariate logistic regression in D treated men (n = 656) and validated in M treated men (n = 333). Risk factors were combined to form risk groups to predict PSA declines, OS, tumor, and pain responses. Prostate Cancer Working Group (PCWG2) disease states were evaluated for these outcomes in docetaxel treated men. RESULTS: In multivariate analysis, four independent risk factors predicted for absence of >30% decline in 3-month PSA: significant baseline pain (OR 0.63 p = 0.02), visceral metastases (OR 0.66, p = 0.03), anemia (hemoglobin <13 g/dl, OR 0.72 p = 0.07), and bone scan progression at baseline (OR 0.60 p = 0.009). Predictive accuracy was moderate with a concordance index (c-index) of 0.61. Risk groups (good, intermediate, poor) were developed with median OS of 25.7 (95% CI 23.3-28.6), 18.7 (17.3-19.7), and 12.8 (11.5-14.6) months, respectively (p < 0.0001), and >30% PSA decline in 78, 66, and 58 percent of men (p < 0.001). In the validation M cohort, similar trends for PSA declines and OS were noted across risk groups (OS 22.5, 16.0, 11.8 mo, p < 0.001). PCWG2 subtypes (node only, bone metastatic, and visceral disease), were also highly prognostic and predictive but did not predict OS as well as the TAX327 risk groups (c-index 0.56). CONCLUSIONS: Risk groups have been identified and internally validated that predict PSA decline and OS in men with mCRPC. This classification may facilitate evaluation of new regimens of systemic therapy that warrant definitive testing in comparison to docetaxel and prednisone in phase III trials. Prospective validation of these risk groups is needed. [Table: see text].

Duke Scholars

Author Andrew John Armstrong Medicine, Medical Oncology

Author Susan Halabi Biostatistics & Bioinformatics, Division of Biostatistics