Androgen receptor splice variant 7 (AR-V7) and AR full-length (AR-FL) as predictive biomarkers of therapeutic resistance: partners in crime?

The prostate cancer treatment armamentarium has expanded over the last decade to include taxane-based chemotherapies (docetaxel, cabazitaxel), sipulecel-T, radium-233, and newer androgen receptor signaling (ARS) inhibitors (abiraterone, enzalutamide, apalutamide). Despite these improvements, persistent androgen receptor (AR) signaling remains a key driver of prostate cancer progression after androgen deprivation therapy (ADT), transition to castrate-resistant disease, and even after resistance to ARS inhibitors. Cross-resistance between ARS inhibitors is common. Predictive biomarkers are therefore needed to optimize treatment selection. Mechanisms of resistance have been attributed to genomic heterogeneity, molecular alterations to the AR and/or upregulation of bypass mechanisms that drive AR activation, including expression of AR splice variants lacking the ligand-binding domain. AR splice variant 7 (AR-V7), the most abundant AR splice variant, has been implicated in abiraterone and enzalutamide resistance and poor patient outcomes. While knowledge of AR-V7 status may guide treatment decisions, AR-V7 alone cannot sufficiently predict response; detection of other variants (AR^v567es) or partners, such as AR full length (AR-FL), might improve prediction.

In this issue of BJUI, Del Re et al. evaluated the expression of AR-V7 and AR-FL in exosomal RNA as combined predictive biomarkers of resistance to ARS therapy (1). AR-FL was detected in all 73 patients (22% were AR-V7+) and AR-FL expression was significantly higher in AR-V7+ vs AR-V7-patients (p<0.0001). These findings that AR-V7 detection has the higher impact on response to therapy confirmed several previous studies; however, the authors took a novel approach to refine the predictive value by stratifying the patient pool into AR-V7 positive and negative populations, and then into tertiles based on AR-FL expression. Analysis of patient outcomes, both in terms of OS and PFS, in these six groups reveals a more nuanced potential treatment strategy. Though AR-V7 expression better predicts OS and PFS to ARS therapy than does AR-FL expression, patients with discordant AR-FL expression relative to their AR-V7 expression may also benefit from treatment different from what their AR-V7 status would suggest. For example, patients positive for AR-V7 but in the bottom tertile of AR-FL expression may be effectively treated with anti-androgen therapies; a breakthrough for patients ineligible for chemotherapy. Additionally, patients negative for AR-V7 but in the top tertile of AR-FL expression may respond better to front-line taxane chemotherapy. Thus, the addition of AR-FL to AR-V7 may aid in better treatment selection.

Recently, the PROPHECY trial prospectively validated the clinical utility of AR-V7 by demonstrating that detection of AR-V7 in circulating tumor cells (by two blood-based assays) is predictive of whether CRPC patients have become resistant to ARSIs, thereby reducing future benefit from further ARSI therapy. Although the PROPHECY study found a strong association between positive AR-V7 and anti-androgen therapy resistance, some AR-V7 negative men did still exhibit resistance to anti-androgen therapy, demonstrating that a second predictive marker (like AR-FL) would be helpful to further guide patient selection (2). It would be interesting to see whether the approach described in Del Re et al. could be replicated using the PROPHECY trial data. Moreover, a recent study established that AR-V7 is very rarely expressed in primary tissue, with expression emerging in response to primary ADT (and in CRPC progression) and further enhanced in resistance to ARS inhibitors (3). AR-V7 was shown to associate with AR-FL expression and copy number in CRPC with many cases of high AR-FL expression having undetectable/low AR-V7 expression, indicating that mRNA splicing remains crucial for AR-V7 generation. Although AR-V7 negative tumors responded to ARS therapy as expected, some AR-V7 positive tumors also responded, suggesting that AR-V7 detection does not preclude response to ARS therapy and providing further evidence that a second marker could be useful as a predictive tool. Finally, AR-V7 status in determining taxane response/resistance remains in conflict with studies either showing that taxanes retain activity in patients with positive AR-V7 or that the absence of AR-splice variants (AR-V7 and AR^v567es) may be associated with superior response to taxane treatment, leading to the hypothesis that AR-FL would be most sensitive to taxane treatment, followed by AR^v567es and AR-V7 (4, 5).

While several studies have shown that the AR-V7/AR-FL ratio tends to be elevated in CRPC tissues, the role of AF-FL as a predictive biomarker for AR-targeted therapy remains controversial. One study found that positive AR-V7, but not higher AR-FL, was associated with worse prognosis (6). As detection methodologies in liquid biopsies and AR data analysis improve over time, the interplay between AR-FL and AR-V7 as well other AR variants warrants further study, which should shed light on whether AR-V7 and/or AR-FL (either as homodimers or possibly heterodimers with AR-V7) are driving resistance to ARS inhibitors. Are they equal partners or is one the dominant driver of the crime?