In the PROPHECY study, men with mCRPC and AR-V7–positive CTCs had only a 0% to 11% probability of benefit from abiraterone or enzalutamide.
Detection of circulating tumor cells (CTCs) positive for the nuclear-specific AR-V7 protein was an independent predictor of shortened progression-free survival (PFS) and overall survival (OS) when treating metastatic castration-resistant prostate cancer (mCRPC) with abiraterone or enzalutamide, according to results of the PROPHECY study (abstract 5004). The findings were presented at the 2018 Annual Meeting of the American Society of Clinical Oncology (ASCO), held June 1–5 in Chicago.
“Men with AR-V7–positive CTCs have a very low probability of benefit from abiraterone or enzalutamide, ranging from 0% to 11%,” said Andrew J. Armstrong, MD, of Duke Cancer Institute. “However, a lack of AR-V7 detection does not guarantee response or benefit” where these therapies are concerned, he added.
Abiraterone and enzalutamide improve survival in the vast majority of men, reflecting that the androgen receptor (AR) stays activated in most men through either AR amplification or persistent autocrine or adrenal androgen synthesis. However, 10% to 20% of patients remain refractory to these therapies, and acquired resistant is fairly universal, Armstrong said.
There is an unmet need for predictive medicine reflecting the cross resistance of abiraterone and enzalutamide.
One of the first indicators of cross resistance discovered was CTC AR-V7. There are currently two assays for measuring AR-V7: Johns Hopkins modified-AdnaTest CTC AR-V7 mRNA assay (JHU AR-V7) and the Epic Sciences CTC nuclear AR-V7 protein assay (Epic AR-V7). Two single-institution studies of both assays showed CTC AR-V7 positivity to be predictive of resistance to enzalutamide and abiraterone.
In PROPHECY, researchers enrolled 120 men with metastatic CRPC who were about to start treatment with enzalutamide or abiraterone. About one-third of patients had received prior abiraterone/enzalutamide. Patients were followed with serial liquid biopsies over time, and with taxane therapy until second progression. Included patients had to have two or more high-risk features.
The primary endpoint was association of baseline AR-V7 with radiographic/clinical PFS. Median follow-up was 19.6 months.
The two assays produced different results. Prevalence of a positive baseline CTC AR-V7 test using prespecified criteria was 24% with the JHU testing and 10% with the Epic test. However, when patients progressed on the two drugs, the prevalence of a positive biomarker increased. With the JHU test, 41% of patients were positive at baseline, increasing to 44% at progression; with the Epic test, 12% were positive at baseline, increasing to 20% at progression.
AR-V7 differed by known baseline prognostic features. Patients with low lactate dehydrogenase, low alkaline phosphatase, and few CTCs had very rare detection of AR-V7. Patients with multiple poor prognostic features were more likely to harbor AR-V7 positive CTCs.
After adjusting for CTC count and established clinical factors, AR-V7 detection by the JHU AR-V7 assay and the Epic AR-V7 assay were independently associated with worse PFS and OS. There was 82% concordance between the two assays.
For the JHU assay, median radiographic PFS was 3.1 months for AR-V7 positive disease compared with 6.9 months for AR-V7–negative disease (HR, 2.4; 95% CI, 1.5–3.7). Median OS was 10.8 months for AR-V7–positive disease compared with 27.2 months for AR-V7–negative disease (HR, 3.9; 95% CI, 2.2–6.9).
For Epic, the median radiographic PFS was 3.1 months for AR-V7–positive disease, compared with 6.1 months for AR-V7–negative disease (HR, 2.5; 95% CI, 1.3–4.7). Median OS was 8.4 months for AR-V7–positive disease compared with 20.3 months for AR-V7–negative disease (HR, 4.5; 95% CI, 2.1–9.8).
“AR-V7 in CTCs should be interpreted with caution in the context of both phenotypic and genomic heterogeneity associated with aggressive disease, suggesting that targeting patients and targeting AR-V7 will require a broad approach,” Armstrong concluded.