A B S T R A C T PurposePersistent androgen signaling is implicated in castrate-resistant prostate cancer (CRPC) progression. This study aimed to evaluate androgen signaling in bone marrow-infiltrating cancer and testosterone in blood and bone marrow and to correlate with clinical observations. Patients and MethodsThis was an open-label, observational study of 57 patients with bone-metastatic CRPC who underwent transiliac bone marrow biopsy between October 2007 and March 2010. Patients received oral abiraterone acetate (1 g) once daily and prednisone (5 mg) twice daily. Androgen receptor (AR) and CYP17 expression were assessed by immunohistochemistry, testosterone concentration by mass spectrometry, AR copy number by polymerase chain reaction, and TMPRSS2-ERG status by fluorescent in situ hybridization in available tissues. ResultsMedian overall survival was 555 days (95% CI, 440 to 965ϩ days). Maximal prostate-specific antigen decline Ն 50% occurred in 28 (50%) of 56 patients. Homogeneous, intense nuclear expression of AR, combined with Ն 10% CYP17 tumor expression, was correlated with longer time to treatment discontinuation (Ͼ 4 months) in 25 patients with tumor-infiltrated bone marrow samples. Pretreatment CYP17 tumor expression Ն 10% was correlated with increased bone marrow aspirate testosterone. Blood and bone marrow aspirate testosterone concentrations declined to less than picograms-per-milliliter levels and remained suppressed at progression. ConclusionThe observed pretreatment androgen-signaling signature is consistent with persistent androgen signaling in CRPC bone metastases. This is the first evidence that abiraterone acetate achieves sustained suppression of testosterone in both blood and bone marrow aspirate to less than picograms-per-milliliter levels. Potential admixture of blood with bone marrow aspirate limits our ability to determine the origin of measured testosterone. J Clin

Purpose Small-cell prostate carcinoma (SCPC) morphology predicts for a distinct clinical behavior, resistance to androgen ablation, and frequent but short responses to chemotherapy. We sought to develop model systems that reflect human SCPC and can improve our understanding of its biology. Experimental Design We developed a set of CRPC xenografts and examined their fidelity to their human tumors of origin. We compared the expression and genomic profiles of SCPC and large cell neuroendocrine carcinoma (LCNEC) xenografts to those of typical prostate adenocarcinoma xenografts. Results were validated immunohistochemically in a panel of 60 human tumors. Results The reported SCPC and LCNEC xenografts retain high fidelity to their human tumors of origin and are characterized by a marked upregulation of UBE2C and other mitotic genes in the absence of AR, retinoblastoma (RB1) and cyclin D1 (CCND1) expression. We confirmed these findings in a panel of CRPC patients' samples. In addition, array comparative genomic hybridization of the xenografts showed that the SCPC/LCNEC tumors display more copy number variations than the adenocarcinoma counterparts. Amplification of the UBE2C locus and microdeletions of RB1 were present in a subset, but none displayed AR nor CCND1 deletions. The AR, RB1, and CCND1 promoters showed no CpG methylation in the SCPC xenografts. Conclusion Modeling human prostate carcinoma with xenografts allows in-depth and detailed studies of its underlying biology. The detailed clinical annotation of the donor tumors enables associations of anticipated relevance to be made. Futures studies in the xenografts will address the functional significance of the findings.