728PD - Impact of AR-V7 protein localization in the prediction of therapeutic benefit of taxanes over androgen receptor signaling inhibitors (ARSi) in meta...

Date 09 October 2016
Event ESMO 2016 Congress
Session Genitourinary tumours, prostate
Topics Prostate Cancer
Presenter Howard Scher
Citation Annals of Oncology (2016) 27 (6): 243-265. 10.1093/annonc/mdw372
Authors H. Scher1, R. Graf2, N. Schreiber1, D. Lu2, J. Louw2, H.A. Vargas Alvarez1, R. Bambury1, D. Danila1, B. McLaughlin1, G. Heller3, M. Fleisher1, R. Dittamore2
  • 1Medicine, Memorial Sloan-Kettering Cancer Center, 10065 - New York, NY/US
  • 2Translational Research, Epic Sciences, Inc, 92121 - San Diego/US
  • 3Epidemiology & Biostatistic, Memorial Sloan Kettering Cancer Center, 10065 - New York/US



AR-V7 is a truncated form of the AR lacking the ligand binding domain that activates AR signaling in the absence of androgens. The presence of AR-V7 mRNA in CTCs predicts a poor outcome on the ARSi, abiraterone and enzalutamide. Localization of AR-V7 in the nucleus of CTCs also predicts poor ARSi outcomes, and a treatment-specific interaction that predicted a reduced risk of death if Taxanes vs. ARSi (HR = 0.242, p = 0.0350) are prescribed. Overall, the incidence of nuclear AR-V7 protein was lower than that of AR-V7 mRNA, but more predictive of ARSi outcomes. In a post-hoc analysis of outcomes of samples evaluated with AR-V7 CTC protein we asked if the presence of cytoplasmic AR-V7 protein could explain the difference in prognostic ability of the two tests, and separately, explored the association between non-nuclear AR-V7 and benefit from taxanes over ARSi.


191 blood samples (n = 128 pre-ARSi, n = 63 pre-Taxane) from 161 patients were processed utilizing the Epic Sciences CTC nuclear AR-V7 protein test. Patients were followed up to 29.5 months (range 1.3 to 29.5). Samples were rescored by readers, blinded to outcome to determine the frequency of and outcome with cytoplasmic AR-V7.


Inclusion of non-nuclear localized AR-V7 protein as positive scoring criteria increased the incidence of detection in all lines of therapy, an equivalent increase in false positives (PSA responders), and loss of the significance of the treatment-specific survival interaction in multivariate model.

AR-V7 Localization Detection Rate by Line of Therapy (1st, 2nd, 3rd +) HR of OS on ARSi Therapy Interaction (Taxanes vs. ARSi)
Nuclear (alone) 3%, 18%, 31% 11.5, p 


Including cytoplasmic AR-V7 in the “positive” test definition reduces the prognostic power of the assay and negates the treatment predictive value of AR-V7. Not all AR-V7 signal is equivalent. It remains to be seen if non-nuclear localized AR-V7 protein samples would test positive via mRNA methods.

Clinical trial identification

Legal entity responsible for the study

Memorial Sloan Kettering Cancer Center


Epic Sciences


H. Scher: Consulting: Astellas, AstraZeneca, BIND Therapeutics, Blue Earth, Bristol-Myers Squibb, Chugai, Endocyte, Ferring, Genentech, Janssen, Med IQ, Medivation, OncologySTAT, Palmetto GBA, Pfizer, Sanofi, Takeda, Ventana, WIRB-Copernicus Group. R. Graf, D. Lu, J. Louw, R. Dittamore: Epic Sciences Employee. All other authors have declared no conflicts of interest.