Analysis of AR-FL and AR-V1 in whole blood of patients with castration resistant prostate cancer as a tool for predicting response to abiraterone acetate

Introduction

Prostate cancer (PCa) is one of the most common malignancies worldwide, and the second leading cause of cancer-related death among men in Europe and North America,^1,2 while the absolute majority of mortality is attributed to castration-resistant PCa (CRPC). Over the last decade overall survival and quality of life for CRPC patients has improved with the introduction of new pharmacological therapies.³ Therapeutic armamentarium has been enriched with new taxanes (docetaxel and cabazitaxel), radionuclide radium-223, immunotherapy (Sipuleucel-T) and new generation androgen receptor (AR)-targeted drugs – AR inhibitors enzalutamide (EN), apalutamide and androgen synthesis inhibitor abiraterone acetate (AA).⁴ However, because of biologic heterogeneity of CRPC, 20–40% of the patients have primary resistance to these novel compounds.^5,6 Therefore more advances are required in the identification of predictive biomarkers for CRPC treatment personalization initially, and at the time of progression.

There is now ample evidence that the AR axis remains active and plays the main role in the progression to CRPC. This happens, at least in part, due to the aberrantly expressed full length androgen receptor (AR-FL) and androgen receptor splice variants (AR-Vs) that lack ligand-binding domain and therefore are constitutively active. Since this discovery, AR-Vs have been strongly related to development of AR-directed treatment resistance.^7–10 To date, there are over 20 known AR-Vs;¹¹ however, AR-V7, the most abundant form of AR-V detected in RNA and protein level in clinical specimens,¹² remains the predominant topic of AR-Vs studies. Detection of circulating AR-V7, encouraged initiation an array of translational studies and development of the commercial blood tests, but the number of wide-range studies on the AR transcript and its variants in whole blood is still insufficient for a comprehensive understanding of biological heterogeneity of CRPC.^13–15

In this study, we performed a comprehensive analysis of AR transcripts (AR-FL, AR-V1, -V3, and -V7) in prostate tissue specimens as well as in whole blood, collected from CRPC patients prior and during treatment with AA to evaluate the association with treatment response and survival outcomes.

Materials and Methods

Results

Discussion

Primary and acquired resistance to a new-generation AR-targeted therapies is a common phenomenon in CRPC treatment and approximately a quarter of the patients do not properly respond to AA or EN.¹⁹ There is increasing evidence that resistance to AA or EN therapies may be, in part, due to the expression of AR-Vs that lead to the formation of constitutively active AR proteins that drive CRPC progression even in the absence of androgens. Multiple AR-Vs have been characterized and linked to CRPC development; however, the broad clinical value of these markers remains controversial,^20,21 and wide-spectrum AR-V analyses are still lacking. Therefore only thorough analysis of known AR-Vs, can aid in the development of biomarkers for primary and secondary AA resistance identification and improve management of CRPC.

In this study, after transcript preselection in a pilot examination, the expression of AR-FL, AR-V1, AR-V3, and AR-V7 was investigated in PCa and non-cancerous prostatic tissues, as well as in the blood collected from CRPC patients at baseline and during AA treatment. In line with previous research,^9,22 expression analysis of AR transcripts in hormone-naïve PCa tissue revealed a significantly higher expression of AR-V1 and AR-V7 in cancerous tissue and its correlation with BCR. The data suggests that already at the time of diagnosis, prostatic carcinoma confers the predisposition for the development of eventual resistance to androgen deprivation therapy (ADT),²³ and, according to the AR-Vs expression level, might be stratified into the subgroups for the most appropriate treatment.

Quantitative reverse transcription PCR and custom-designed assays were used for AR transcripts detection in whole blood. Out of the 185 whole blood samples analyzed in our study, 81% were positive for at least one AR-V transcript and at least one AR-V transcript was identified at a high level in 44% of the samples. A high level of AR-FL and/or AR-V1 was identified as a potential biomarker of predicting the response to AA treatment. Particularly, patients having higher levels of AR-FL and AR-V1 in blood samples collected at baseline showed primary resistance to AA and experienced a rapid progression of the disease. A higher AR-FL and AR-V1 level was also associated with shorter OS. AR-FL is the main therapeutic target in PCa, and the upregulated expression of this key biomarker has been reported in numerous cell line and clinical specimen, especially CRPC, studies.²⁴ AR-V1 has been found upregulated in CRPC in comparison to hormone-naïve PCa,⁹ but to the best of our knowledge this study is the first to demonstrate the link between associated AR-FL and AR-V1 in whole blood and response to AA therapy.

Although the most common AR variant in our and other studies was AR-V7 and the strongest association with the resistance to treatment was previously attributed to AR-V7,²⁵ the clinical utility of this biomarker remains controversial. In many reports, no or weak association has been reported between AR-V7 and PSA decline,^20,21,26 PSA-PFS,²¹ and OS in EN-treated cohort.²⁷ In our study, the prognostic power of AR-V7 was weak, but the combination with other markers (AR-FL and all tested AR transcripts) helped to improve it. In addition to AR-V7, the capacity to predict treatment outcomes from the whole blood has been reported for blood-circulating AR-V12²⁸ and AR-V9²¹, showing contradictory results. Analyzed together with AR-V7, AR-V12 was associated with treatment resistance,²⁸ while a combination of AR-V7 and -V9 failed to predict treatment outcomes as patients with AR-Vs still benefited from AA or EN treatment.²¹ We tested several AR-Vs in a pilot study, but the detection rates for AR-V9 and -V12 in the blood were too low for further analysis. Of note, in our study AR-FL was detected in all samples, and the level of AR-FL correlated with the level of all AR-Vs, which is in concordance with a recent research on PCa metastasis.²⁹ The pressure of AR-directed therapies (e.g. ADT) was shown to cause the increase in transcriptional and splicing rates of AR and raise the overall AR-FL transcript level, eventually triggering formation of various AR-Vs.³⁰ Therefore, a concomitant detection of various AR-Vs highlights that analysis and inhibition of several AR-Vs might be clinically more significant than exclusive focus on AR-V7.

Regardless of high sensitivity and careful technical validation of the method, this study has several drawbacks. Since the level of transcripts of interest in blood sample is low, precautions should be taken with sample collection, storage, and preparation to ensure the quality of RNA. In this study, PAXgene Blood RNA tubes were used, preserving the sample and ensuring high quality of RNA for at least a decade. Regarding the low transcript level, to avoid any background expression from peripheral blood mononuclear cells (PBMCs), selected genes must be highly PCa-specific. Despite the consideration that AR-V transcripts are highly PCa-specific, they were detected in healthy control blood samples in this study similarly as have been reported by other research teams.^20,25,26 This demonstrates non-specific signals from PBMCs that might still have interfered with the results. Despite the fact that asymptomatic subjects were age-matched and their samples were processed in parallel with those of the study patients, the size of the control group was relatively small. Because of unknown outcomes and due to the prospective nature of the study, it was difficult to maintain the uniformity and proportions of the cohort; therefore, some patient subgroups were smaller and that may have affected the statistics. Prospectively collected samples also had shorter follow-up time compared to well-characterized retrospective cohorts with longer follow-up periods.

Conclusions

This is the first report of a thorough analysis of full-length AR and AR-V1, -V3, and -V7 in prostate tissues and whole blood from prospectively collected CRPC cases treated with AA. The data suggests that AR transcripts can be used for patients’ selection for AA treatment, and the whole blood-based assay might be more technically eligible and acceptable for AR-Vs detection in clinical settings.

Supplementary Material