Circulating Tumor Cell Counts Are Prognostic of Overall Survival in SWOG S0421: A Phase III Trial of Docetaxel With or Without Atrasentan for Metastatic Castration-Resistant Prostate Cancer

Abstract

PURPOSE: Circulating tumor cell (CTC) enumeration has not been prospectively validated in standard first-line docetaxel treatment for metastatic castration-resistant prostate cancer. We assessed the prognostic value of CTCs for overall survival (OS) and disease response in S0421, a phase III trial of docetaxel plus prednisone with or without atrasentan. PATIENTS AND METHODS: CTCs were enumerated at baseline (day 0) and before cycle two (day 21) using CellSearch. Baseline counts and changes in counts from day 0 to 21 were evaluated for association with OS, prostate-specific antigen (PSA), and RECIST response using Cox regression as well as receiver operator characteristic (ROC) curves, integrated discrimination improvement (IDI) analysis, and regression trees. RESULTS: Median day-0 CTC count was five cells per 7.5 mL, and CTCs < versus ≥ five per 7.5 mL were significantly associated with baseline PSA, bone pain, liver disease, hemoglobin, alkaline phosphatase, and subsequent PSA and RECIST response. Median OS was 26 months for < five versus 13 months for ≥ five CTCs per 7.5 mL at day 0 (hazard ratio [HR], 2.74 [adjusting for covariates]). ROC curves had higher areas under the curve for day-0 CTCs than for PSA, and IDI analysis showed that adding day-0 CTCs to baseline PSA and other covariates increased predictive accuracy for survival by 8% to 10%. Regression trees yielded new prognostic subgroups, and rising CTC count from day 0 to 21 was associated with shorter OS (HR, 2.55). CONCLUSION: These data validate the prognostic utility of CTC enumeration in a large docetaxel-based prospective cohort. Baseline CTC counts were prognostic, and rising CTCs at 3 weeks heralded significantly worse OS, potentially serving as an early metric to help redirect and optimize therapy in this clinical setting.