Prediction of survival following first-line chemotherapy in men with castration-resistant metastatic prostate cancer.

Purpose: We sought to evaluate predictors of overall survival following progression after systemic chemotherapy in men with metastatic castration-resistant prostate cancer.

Experimental Design: For our study population, we used the TAX327 multicenter randomized phase III trial comparing administration of docetaxel and prednisone every 3 weeks, weekly administration of docetaxel and prednisone, and administration of mitoxantrone and prednisone every 3 weeks. Progression was defined as the earliest of prostate-specific antigen (PSA), tumor, or pain progression. We analyzed predictors of postprogression survival according to both prechemotherapy and postchemotherapy variables with adjustment for potential confounders.

Results: Among 1,006 men, 640 had evaluable information on protocol-defined progression leading to further therapy. Median postprogression survival was 14.5 months. In the multivariable analysis, several pretreatment factors were associated with postprogression survival: pain, performance status, alkaline phosphatase, number of sites of metastatic disease, liver metastases, hemoglobin, PSA, and time since diagnosis. In addition, we found that the number of progression factors (PSA, pain, and tumor size), the duration of first-line chemotherapy, and whether progression occurred during chemotherapy independently predicted postprogression survival. We found evidence for the benefit of continuation of chemotherapy beyond progression only for men who had isolated worsening of pain. A nomogram was constructed and internally validated with a concordance index of 0.70.

Conclusions: An internally validated model to predict postchemotherapy survival was developed. Evaluation of men in the postdocetaxel setting should consider the type of progression, duration of therapy, and known pretreatment prognostic factors. Definitions of progression in castration-resistant prostate cancer that include pain should also consider composite measures of tumor or PSA progression. External validation is planned.