Advertisement

CABA-V7: a prospective biomarker selected trial of cabazitaxel treatment in AR-V7 positive prostate cancer patients

Open AccessPublished:October 28, 2022DOI:https://doi.org/10.1016/j.ejca.2022.09.032

      Highlights

      • AR-V7 is observed to be an independent prognostic marker in mCRPC.
      • No association of AR-V7 status to mCRPC-associated alterations.
      • PSA response rate of 12%, primary end-point therefore not met.
      • Cabazitaxel may still improve prognosis in AR-V7 positive patient population.

      Abstract

      Background

      Metastatic castration-resistant prostate cancer (mCRPC) patients with positive AR-V7 expression in their circulating tumour cells (CTCs) rarely derive benefit from abiraterone and enzalutamide.

      Design

      We performed a prospective, multicenter, single arm phase II clinical trial (CABA-V7) in mCRPC patients previously treated with docetaxel and androgen deprivation therapy.

      Objective

      In this trial, we investigated whether cabazitaxel treatment resulted in clinically meaningful PSA response rates in patients with positive CTC-based AR-V7 expression and collected liquid biopsies for genomic profiling.

      Results

      Cabazitaxel was found to be modestly effective, with only 12% of these patients obtaining a PSA response. Genomic profiling revealed that CTC-based AR-V7 expression was not associated with other known mCRPC-associated alterations. CTC-based AR-V7 status and dichotomised CTC counts were observed as independent prognostic markers at baseline.

      Conclusions

      AR-V7 positivity predicted poor overall survival (OS). However, cabazitaxel-treated AR-V7 positive patients and those lacking AR-V7 positivity, who received cabazitaxel as standard of care, appeared to have similar OS. Therefore, despite the low response rate, cabazitaxel may still be an effective treatment in this poor prognosis, AR-V7 positive patient population.

      Keywords

      1. Introduction

      The androgen receptor splice variant 7 (AR-V7) is reported as the most frequently expressed and biologically-significant AR-V; formed as the consequence of alternative splicing of the AR-transcript. AR-V7 lacks the ligand-binding domain and thereby functions as a constitutionally-active transcription factor in the nucleus [
      • Antonarakis E.S.
      • et al.
      Androgen receptor variant-driven prostate cancer: clinical implications and therapeutic targeting.
      ,
      • Tagawa S.T.
      • et al.
      Expression of AR-V7 and ARv(567es) in circulating tumor cells correlates with outcomes to taxane therapy in men with metastatic prostate cancer treated in TAXYNERGY.
      ].
      The role of AR-V7 and its detection in circulating tumour cells (CTCs) has been extensively explored due to its association with primary resistance to abiraterone or enzalutamide therapy in men with metastatic castration-resistant prostate cancer (mCRPC) [
      • Antonarakis E.S.
      • et al.
      AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer.
      ]. These findings make AR-V7 a potential biomarker in predicting the outcome or response following androgen-targeted therapy (ART) [
      • Antonarakis E.S.
      • et al.
      Androgen receptor variant-driven prostate cancer: clinical implications and therapeutic targeting.
      ,
      • Antonarakis E.S.
      • et al.
      AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer.
      ,
      • Del Re M.
      • et al.
      The detection of androgen receptor splice variant 7 in plasma-derived exosomal RNA strongly predicts resistance to hormonal therapy in metastatic prostate cancer patients.
      ,
      • Armstrong A.J.
      • et al.
      Prospective multicenter validation of androgen receptor splice variant 7 and hormone therapy resistance in high-risk castration-resistant prostate cancer: the PROPHECY study.
      ]. Despite the lack of response on androgen receptor targeted agents (ARTA), AR-V7 presence in CTCs from mCRPC patients (AR-V7Pos.) was not associated with primary resistance to taxane-based chemotherapy. Thus, taxane-based regimens could potentially be an effective treatment-option in pre-treated AR-V7Pos. mCRPC patients [
      • Smith T.J.
      • et al.
      Recommendations for the use of WBC growth factors: American society of clinical oncology clinical practice guideline update.
      ,
      • Sieuwerts A.M.
      • et al.
      An in-depth evaluation of the validity and logistics surrounding the testing of AR-V7 mRNA expression in circulating tumor cells.
      ,
      • van Dessel L.F.
      • et al.
      Application of circulating tumor DNA in prospective clinical oncology trials - standardization of preanalytical conditions.
      ].
      Various studies have been performed on biomarker-guided therapy in mCRPC patients [
      • Conteduca V.
      • et al.
      Plasma androgen receptor and response to adapted and standard docetaxel regimen in castration-resistant prostate cancer: a multicenter biomarker study.
      ,
      • van der Doelen M.J.
      • et al.
      Impact of DNA damage repair defects on response to radium-223 and overall survival in metastatic castration-resistant prostate cancer.
      ,
      • Belderbos B.P.S.
      • et al.
      Associations between AR-V7 status in circulating tumour cells, circulating tumour cell count and survival in men with metastatic castration-resistant prostate cancer.
      ], but it currently remains to be confirmed that pre-treated AR-V7Pos. mCRPC patients benefit from cabazitaxel treatment, and prospective trials are still lacking. We postulate that the presence of AR-V7 in CTCs from patients who have progressed to docetaxel does not affect potential response to cabazitaxel; thereby establishing taxane-based treatment as the preferred first choice for AR-V7Pos. mCRPC patients.
      The CABA-V7 trial (registered at ClinicalTrial.gov under NCT03050866) investigated whether cabazitaxel would be a viable alternative for AR-V7Pos. mCRPC patients who had previously been treated with docetaxel and androgen deprivation therapy (ADT).

      2. Methods

      2.1 Patient inclusion and clinical parameters

      From February 2017 until November 2019, mCRPC patients who progressed after docetaxel were included in this prospective, multicenter (8 sites), single arm phase II clinical trial, the CABA-V7 trial (Cabazitaxel in mCRPC Patients With AR-V7 Positive Circulating Tumour Cells (CTCs) (CABA-V7). Patients should have continued with ADT either by luteinising hormone-releasing hormone (LHRH) agonist/antagonist or orchiectomy, and they should be in reasonable to good physical condition (defined as WHO performance status ≤2). This trial was conducted in accordance with the Declaration of Helsinki and approved by the independent Dutch medical ethical committee (BEBO) (MEC 16–703). All patients provided written informed consent before any study procedure was performed. Patients were screened for the presence of CTCs and AR-V7 status. For additional information on patient inclusion and clinical parameters, we refer to Suppl. File 1. Due to the single-arm design of this trial, a post-hoc analysis was performed to identify patients whom were also treated with cabazitaxel outside of the study context.

      2.2 Enrichment of CTC and respective detection of AR-V7

      Blood samples for CTC enumeration and mRNA profiling were processed within 24–96 h after withdrawal, using the CellSearch® system (Menarini Silicon Biosystems, Castel Maggiore, Bologna, Italy) at the Laboratory of Translational Cancer Genomics and Proteomics of the Department of Medical Oncology, Erasmus MC Cancer Institute in Rotterdam. For additional information, we refer to Suppl. File 1. During subsequent analysis, we utilised the following dichotomisation for CTCs (per 7.5 mL) at baseline: <5 versus ≥5 CTCs.

      2.3 Blood sampling and cfDNA extraction

      Per patient, ten mL blood was centrifuged twice to obtain plasma as detailed previously by van Dessel et al. [
      • van Dessel L.F.
      • et al.
      Application of circulating tumor DNA in prospective clinical oncology trials - standardization of preanalytical conditions.
      ]. Next, cfDNA was isolated using the QIAamp® Circulating Nucleic Acid kit (QIAGEN, Venlo, The Netherlands) according to the manufacturer's instructions from an average of 1726.7 μL plasma (range: 350––2000 μL), on average yielding 88.5 ng cfDNA (range: 13.132–844.2 ng) as measured on the Qubit fluorometer (dsDNA High Sensitivity assay by Thermo Fisher Scientific, Waltham, MA) [
      • van Dessel L.F.
      • et al.
      Application of circulating tumor DNA in prospective clinical oncology trials - standardization of preanalytical conditions.
      ].
      Obtained cfDNA was evaluated by Next-Generation Sequencing (NGS) using a targeted mutational profiling by the QIAseq platform, as detailed below.

      2.4 Mutational screening of cfDNA using panel-based sequencing using the QIAseq platform

      We employed a customised QIAseq platform using a minimum of 10 ng cfDNA to detect tumour-specific mutations within an a priori set of 57 known prostate-specific genes described by Annala et al. [
      • Annala M.
      • et al.
      Circulating tumor DNA genomics correlate with resistance to abiraterone and enzalutamide in prostate cancer.
      ] and Robinson et al. [
      • Robinson D.
      • et al.
      Integrative clinical genomics of advanced prostate cancer.
      ] (Suppl. Table 1). For additional information on the selection of somatic variants for downstream analysis, we refer to Suppl. File 1.
      Table 1Baseline characteristics of included patients within CABA-V7.
      AR-V7 Status:AR-V7Pos.AR-V7 - Neg.AR-V7 - Und.
      CharacteristicTotal (N = 137)Main study (N = 25)Other (N = 15)Other (N = 63)Other (N = 34)Missing (no.)
      Age at registration.
       Median (range) - yr70 (46–89)68 (46–78)71 (62–78)71 (52–89)69.5 (53–84)
      WHO PS at registration - no. (%).
      • -
        0
      52 (38%)10 (40%)4 (26.7%)28 (44.4%)10 (29.4%)
      • -
        1
      76 (55.5%)15 (60%)9 (60%)30 (47.6%)22 (64.7%)
      • -
        2
      9 (6.6%)0 (0%)2 (13.3%)5 (7.9%)2 (5.9%)
      Initial PSA, μg/L19
       Mean ± SD366.1 ± 912258.8 ± 428.6598.3 ± 1140.6346.5 ± 481.4371.6 ± 1439
       Median (range)80.8 (0.23–8185)104 (4.19–2000)180.50 (7.50–4098.9)104 (0.23–1820)32.5 (4–8185)
      Absolute neutrophil count19
       Mean ± SD5.3 ± 2.35.7 ± 2.75 ± 2.15.3 ± 2.25.2 ± 2.5
       Median (range)4.8 (0.7–11.8)5 (0.7–11.8)4.4 (2.6–9.2)4.8 (2.1–11)4.7 (2.1–11.4)
      Haemoglobin – g/L19
       Mean ± SD7.7 ± 0.97.2 ± 0.77.4 ± 1.27.9 ± 0.97.9 ± 0.8
       Median (range)7.7 (4.5–10.3)7.2 (5.5–8.2)7.5 (4.5–9.4)8.1 (5.7–10.3)7.8 (6.5–9.1)
      Alkaline phosphatase - IU/L10
       Mean ± SD171 ± 174.5207.8 ± 164.5245.6 ± 398.4167.5 ± 111.2114.1 ± 85.7
       Median (range)122 (42–1608)129 (69–691)140 (51–1608)135 (42–645)87 (51–417)
      Lactate dehydrogenase - IU/L10
       Mean ± SD332.8 ± 293.6495 ± 471.8470.1 ± 474.4268.6 ± 101.2258 ± 151.7
       Median (range)239 (52–1770)287 (154–1770)258.5 (176–1619)253 (144–666)218 (52–828)
      CTC count (7.5 mL).
       Mean ± SD114.7 ± 467.8335.9 ± 1016.1245.3 ± 391.552.9 ± 107.58.79 ± 25.6
       Median (range)10 (0–5105)62 (5–5105)28 (0–1380)11 (0–678)1 (0–129)
      Disease history ----
      Gleason score by diagnosis
       Median (range)8 (4–10)8 (7–10)8 (4–10)9 (7–10)8 (6–10)6
      M-stage at diagnosis.
      • -
        M0
      48 (35%)13 (52%)4 (26.7%)19 (30.2%)12 (35.3%)
      • -
        M1a, M1b, M1c
      64 (46.7%)7 (28%)8 (53.3%)32 (50.8%)17 (50%)
      • -
        Mx
      25 (18.2%)8 (5)%3 (20%)12 (19%)5 (14.7%)
      Type of post therapy - no. (%).
      • -
        Cabazitaxel
      95 (69.3%)25 (100%)9 (60%)43 (68.3%)18 (52.9%)
      • -
        Enzalutamide
      14 (10.2%).1 (6.7%)7 (11.1%)6 (17.6%)
      • -
        Abiraterone
      11 (8%)..7 (11.1%)4 (11.8%)
      • -
        Other
      12 (8.8%).2 (13.3%)6 (9.5%)4 (11.8%)
      • -
        Unknown
      5 (3.6%).3 (20%).2 (5.9%)
      Type of prior therapy - no. (%)
      • -
        Local therapy
      .
      • -
        Yes
      63 (46%)13 (52%)6 (40%)26 (41.3%)18 (52.9%)
      • -
        No
      74 (54%)12 (48%)9 (60%)37 (58.7%)16 (47.1%)
      • -
        Prior hormonal therapy
      .
      • -
        Yes
      137 (100%)25 (100%)15 (100%)63 (100%)34 (100%)
      • -
        No
      .....
      • -
        Prior ARTA
      73
      • -
        Abiraterone
      7 (5.1%)1 (4%)2 (13.3%)3 (4.8%)1 (2.9%)
      • -
        Enzalutamide
      56 (40.9%)16 (64%)7 (46.7%)26 (41.3%)7 (20.6%)
      • -
        Both
      1 (0.7%)0 (0%)0 (0%)0 (0%)1 (2.9%)
      • -
        Prior chemotherapy
      .
      • -
        Yes (Docetaxel)
      137 (100%)25 (100%)15 (100%)63 (100%)34 (100%)
      • -
        No
      .....
      Pos. = Positive, Neg. = Negative, range = min. - max., WHO PS = World Health Organization Performance Score, SD = Standard Deviation, PSA = prostate-specific antigen, ARTA = androgen receptor targeted agent(s).

      2.5 Determining enrichment of (coding) mutations within patients stratified on AR-V7 status

      Per gene, we compared the number of patients harbouring coding mutations with the number of patients without that coding mutation between AR-V7Pos. and AR-V7Neg. groups. We only performed this on genes harbouring coding mutations in at least five percent of either group. We performed a two-sided Fisher's Exact Test with correction for multiple-testing (Benjamini-Hochberg).

      2.6 Pharmacokinetic measurements for exposure–response relationship

      For pharmacokinetic (PK) measurements in patients treated with cabazitaxel, blood samples were drawn during the first cycle prior to infusion, 5 min before end of infusion, and 2 and 6 h after end of infusion, based on a published limited sampling strategy of docetaxel [
      • Baille P.
      • et al.
      Optimal sampling strategies for bayesian estimation of docetaxel (Taxotere) clearance.
      ]. Cabazitaxel plasma concentrations were measured using a validated liquid chromatography with tandem mass spectrometry method [
      • de Bruijn P.
      • et al.
      Quantification of cabazitaxel in human plasma by liquid chromatography/triple-quadrupole mass spectrometry: a practical solution for non-specific binding.
      ]. For additional information on pharmacokinetic measurements, we refer to Suppl. File 1.

      2.7 Statistical analysis

      We aimed to identify whether cabazitaxel treatment had a clinically meaningful PSA response rate of 25% (p1). If PSA response would be <5% (p0), we considered cabazitaxel insufficiently active, and not worth further investigating in this patient population. We used the A'Hern single-stage design with α = 0.05 (one-sided) and β = 0.10 [
      • A'Hern R.P.
      Sample size tables for exact single-stage phase II designs.
      ]. The required sample size was 25. If ≥ 4 PSA responses are observed among these 25 patients, we conclude that the regimen is sufficiently active and does warrant further investigation in this patient population. For additional information on statistical analysis, we refer to Suppl. File 1.

      3. Results

      3.1 Baseline and treatment characteristics

      A total of 137 patients were enrolled in the CABA-V7 study. From all enrolled patients (Fig. 1), we collected blood at baseline for CTC enumeration and analysis of AR-V7 mRNA expression in CTCs. We detected AR-V7 expression in CTCs (at baseline) for 40 patients (AR-V7Pos.) whilst 63 patients lacked expression thereof (AR-V7Neg.). In the remaining 34 patients, AR-V7 expression could not be determined (AR-V7Und.), either due to lack of CTCs, insufficient mRNA quality or insufficient epithelial signal in the AR-V7 RNA profile, as detailed by Siewerts et al. [
      • Sieuwerts A.M.
      • et al.
      An in-depth evaluation of the validity and logistics surrounding the testing of AR-V7 mRNA expression in circulating tumor cells.
      ]. Baseline characteristics of the included patients (n = 137) are described in Table 1. The median age of the patients was 70 years and metastases were present at diagnosis in 46.7% of the patients.
      Fig. 1
      Fig. 1Schematic overview of mCRPC-patient inclusion within the CABA-V7 study. Overview of mCRPC-patient enrolment into the CABA-V7 study and subsequent genotyping efforts. In total, 137 patients were enrolled and for whom AR-V7 status was determined within captured CTCs. Subsequently, captured cfDNA was interrogated for the presence of somatic mutations within known mCRPC-associated genes at baseline. A selection of patients eligible for further trial-enrolments (AR-V7Pos.) received multiple cycles of cabazitaxel.
      All AR-V7Pos. patients with ≥3 CTCs (n = 40) were eligible for further enrolment into the treatment phase of the CABA-V7 trial, in which they were to receive cabazitaxel treatment. Out of these 40 patients, 15 patients declined further enrolment (for unrecorded reasons); thus, in total 25 patients with ≥3 CTCs and AR-V7 expression within their CTCs were treated with cabazitaxel (Fig. 1). The incidence of adverse events of grade 3 or higher that occurred within this group was 16% and the incidence of any serious adverse events was also 16%. These adverse events have been further detailed in.
      Post-hoc analysis revealed that an additional 9 AR-V7Pos. (out of the 15 that declined further enrolment), 43 AR-V7Neg. and 18 AR-V7Und. patients were also treated with cabazitaxel outside of the trial context; therefore, we also utilised these cabazitaxel-treated AR-V7Pos. and AR-V7Neg. patients within the OS analysis. The cabazitaxel-treated AR-V7Neg. patients which were used as a surrogate control arm within this study harboured no significant differences within their clinical characteristics compared to the cabazitaxel-treated AR-V7Pos. patients.

      3.2 PSA response in cabazitaxel treated AR-V7 positive patients

      As primary outcome, we evaluated PSA response within the 25 AR-V7Pos. patients treated with cabazitaxel on-study (Fig. 1). A PSA response according to PCWG2.0 criteria [
      • Scher H.I.
      • et al.
      Trial design and objectives for castration-resistant prostate cancer: updated recommendations from the prostate cancer clinical trials working group 3.
      ] was observed in 3/25 patients (12%; 95% confidence interval (CI): 2.5–31.2%) (Fig. 2). Thus, we were not able to reject our null hypothesis and our predefined criterion to consider cabazitaxel an active treatment in AR-V7Pos. patients warranting further investigation was not met.
      Fig. 2
      Fig. 2Clinical response of CABA-V7 enrolled (AR-V7Pos.) mCRPC-patients following cabazitaxel-treatment. In total, 25 (AR-V7Pos.) mCRPC-patients were included into the final CABA-V7 trial to undergo cabazitaxel-treatment. Depicted mCRPC-patients are sorted (decreasing) on the total length within the CABA-V7 study and thereby receiving cabazitaxel treatment. a) Overview of ΔPSA-measurements during cabazitaxel treatment, centred on their PSA-level prior to pre-cabazitaxel treatment (grey data-point). The number of days within the CABA-V7 study (plus 3 weeks; dotted blue lines) is shown on the x-axis whilst the y-axis depicts the relative ΔPSA (capped to ±100% ΔPSA). b) Overview of clinical responses (PSA, CTC and CTC-decline as detailed in the methodology; formal responses in black). Missing data due to insufficient input material (CTCs or cfDNA) are depicted with crosses (X). c) In addition, the availability of one or multiple CT and/or bone scan for each patient at progression is shown (by presence of a circle). If any of these respective scans revealed progression, it is coloured grey. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)
      As PSA response has its limitations, we also determined response to cabazitaxel treatment as measured by CTC response and CTC decline. We detected CTCs within 20/25 patients after three cabazitaxel cycles and observed a formal CTC response and decline in, respectively, three and ten of these patients (Fig. 2b). For the remaining five patients, a second CTC count was unavailable as they stopped cabazitaxel before the third cycle.
      In addition, no significant difference of PSA nor CTC responses could be observed based on pre-treatment between abiraterone and enzalutamide (χ2-test).

      3.3 Pharmacokinetic measurements and AR-V7 status

      Of the 25 AR-V7Pos. patients treated with cabazitaxel; 20 patients were evaluable for pharmacokinetic analysis of cabazitaxel exposure. Four patients were excluded due to missing samples and a single patient was excluded as the measured concentration did not fit the pharmacokinetic model; which would have resulted in an unreliable Area Under the Curve for the duration of 24 h (AUC024h).
      Geometric mean (dose-corrected) cabazitaxel AUC024h within the 20 included AR-V7Pos. patients was 218 ng∗h/mL (CV: 37.1). We also compared this to a reference cohort from 51 distinct patients (n = 69). The geometric mean (dose-corrected) cabazitaxel AUC024h within the reference cohort was 175 ng∗h/mL (CV: 27.4). We observed a higher dose-corrected cabazitaxel AUC024h within the AR-V7Pos. patients compared to the reference cohort (p = 0.016; Fig. 3).
      Fig. 3
      Fig. 3Cabazitaxel pharmacokinetics in AR-V7Pos. patients compared to a reference cohort. Boxplots representing the geometric means and coefficient of variation (CV) of dose corrected cabazitaxel systemic exposure (AUC024h in ng∗h/mL) in AR-V7Pos. patients (n = 20) and in a reference cohort (n = 69; n = 51 distinct patients). Underlying data-points are shown and coloured per cohort, p-value is shown on top.

      3.4 Relation between AR-V7 expression in CTCs and targeted screening of coding mutations

      We next investigated whether AR-V7 status is independent of other known mCRPC-specific somatic coding aberrations using a targeted screening approach within cfDNA (Fig. 4). We detected 360 somatic coding aberrations over 44 of the captured genes, with a median of 3 coding mutations per sample. The most-frequently mutated genes were found to be AR (n = 45 samples), KMT2C (n = 44), TP53 (n = 22), NBN (n = 19) and ZFHX3 (n = 16). Twenty-six patients revealed no coding mutation within our targeted genes. No significant differences in the frequency of mutant genes between AR-V7Pos. versus AR-V7Neg. patients was observed.
      Fig. 4
      Fig. 4Mutational genotyping reveals AR-V7 as an independent biomarker. Mutational genotyping of cfDNA using targeted sequencing of mCRPC-associated genes (QIAseq; n = 57 genes) for CABA-V7 mCRPC-patients (excluding six samples with insufficient input-material). Each sample is represented by a single column and samples are sorted on mutual-exclusivity of detected coding mutations (shown in f). a) Total number of coding mutation(s). b) Maximum variant allele frequency (VAF) of all sample-respective coding mutations. c) Number of captured CTCs at baseline (per 7.5 mL), in log10-scale. d) Yield of cfDNA (in ng) at baseline, in log10-scale. e) Overview of detected coding mutations, coloured on mutational category. Multiple sample-respective coding mutations within the same gene have been categorised as ‘Multiple mutations’. f) Gene-wise overview of the total number of mCRPC-patients which harboured coding mutations and the respective distribution of AR-V7Pos., AR-V7Neg. and AR-V7Und. samples. The total number of distinct mutant samples (regardless of AR-V7 status) is shown beside each element. g) AR-V7 determination at baseline. h) Formal PSA response status for enrolled CABA-V7 patients. i) Formal CTC response status for enrolled CABA-V7 patients. j) Formal CTC-decline status for enrolled CABA-V7 patients.

      3.5 Prognostic value of CTC, AR-V7 status, and ctDNA-derived biomarkers

      We determined whether age at registration, total Gleason-score, cabazitaxel treatment, AR-V7 status, dichotomised CTC scores and additional ctDNA-derived metrics at baseline (n = 137) were prognostic for overall survival (OS). To this end, we identified relevant variables (p < 0.1) using univariate analysis with backward selection (Suppl. File 1). From this analysis, AR-V7 status, AR-V7 status coupled with cabazitaxel treatment and dichotomised CTC scores remained possible prognostic markers. Using multivariate Cox regression analysis of these parameters (Fig. 5a), we observed that AR-V7 status and dichotomised CTC counts both served as independent prognostic markers (q = 0.01 and q < 0.001, respectively). In univariate analysis in all 137 patients, median OS when stratified by AR-V7 status, was determined as 6.87 months (AR-V7Pos.), 10 months (AR-V7Neg.) and 15.87 months (AR-V7Und.) (Fig. 5b). Stratification on dichotomised CTC counts revealed a median OS of 15.47 months and 7.33 months for CTC counts <5 and ≥ 5, respectively (Fig. 5c). Dichotomised WHO status and whether patients had received cabazitaxel treatment yielded no significant differences in OS (Suppl. Fig. 1a-b).
      Fig. 5
      Fig. 5Multivariate Cox regression analysis of overall survival at baseline. a) Multivariate Cox regression analysis for all included patients with complete observations for CTC-derived AR-V7 status, cabazitaxel treatment status and dichotomised CTC counts at baseline (n = 137). The total number of observations (N), events (death), forest plot, hazard ratio (HR), 95% confidence interval, p and q-values are shown. b) Survival probability (OS) using univariate analysis of all baseline-included patients (n = 137; y-axis), stratified and coloured by AR-V7 status at baseline (AR-V7Pos. vs. AR-V7Neg. and AR-V7Und.), depicted in months (x-axis); censoring is shown by crosses (+). The bottom table represents the total number of remaining cases per depicted time-point. Median OS per stratified group (sorted on decreasing OS) is shown on the right-hand side. c) Similar to b), however stratified for dichotomised CTC counts.
      We investigated whether AR-V7 status and dichotomised CTC counts would retain their clinical relevance within only AR-V7Pos. and AR-V7Neg. cabazitaxel-treated patients; n = 34 and 43, respectively (Fig. 6). The median OS for these patients, stratified on AR-V7 status, was 8.33 and 9.54 months (Fig. 6b, Suppl. Fig. 1c). This revealed that no significant difference in OS was seen for AR-V7 status.
      Fig. 6
      Fig. 6Multivariate Cox regression analysis of overall survival within cabazitaxel-treated patients. a) Multivariate Cox regression analysis for all cabazitaxel-treated AR-V7Pos. and AR-V7Neg. patients with complete observations for CTC-derived AR-V7 status, cabazitaxel treatment status and dichotomised CTC counts at baseline (n = 77). The total number of observations (N), events (death), forest plot, hazard ratio (HR), 95% confidence interval, p and q-values are shown. b) Survival probability (OS) using univariate analysis for all cabazitaxel-treated AR-V7Pos. and AR-V7Neg. patients (n = 77; y-axis), stratified and coloured by AR-V7 status at baseline (AR-V7Pos. vs. AR-V7Neg.), depicted in months (x-axis); censoring is shown by crosses (+). The bottom table represents the total number of remaining cases per depicted time-point. Median OS per stratified group (sorted on decreasing OS) is shown on the right-hand side. c) Similar to b), however stratified for dichotomised CTC counts.
      Stratification on dichotomised CTC counts in these patients revealed a significant difference in OS (p < 0.01), with a median OS of 15.77 months and 7.79 months for CTC counts <5 and ≥ 5, respectively (Fig. 6c).

      3.6 Dynamic changes of AR-V7 expression in cabazitaxel-treated AR-V7 positive patients

      After 9 weeks of treatment, AR-V7 status was again determined for patients included into the treatment phase of the CABA-V7 trial. Of these 25 patients, 19 patients had a second AR-V7 status determination; for the remaining 6 patients, AR-V7 status could not be determined due to lack of captured CTCs or unavailable blood samples due to premature end of treatment (<9 weeks). This second determination showed that 10/19 patients (53%) converted from being AR-V7Pos. at baseline to AR-V7Neg. following cabazitaxel treatment (AR-V7Conv.). In addition, two out of the three patients with previously-detailed PSA responses were found to be converted from AR-V7Pos. Tto AR-V7Neg. status.

      4. Discussion

      Many studies have been performed on cabazitaxel treatment and the role of liquid biopsies in treating mCRPC [
      • Aldea M.
      • et al.
      Cabazitaxel activity in men with metastatic castration-resistant prostate cancer with and without DNA damage repair defects.
      ,
      • León-Mateos L.
      • et al.
      Improving circulating tumor cells enumeration and characterization to predict outcome in first line chemotherapy mCRPC patients.
      ,
      • Gurioli G.
      • et al.
      Circulating tumor cell gene expression and plasma AR gene copy number as biomarkers for castration-resistant prostate cancer patients treated with cabazitaxel.
      ,
      • Mehra N.
      • et al.
      Plasma cell-free DNA concentration and outcomes from taxane therapy in metastatic castration-resistant prostate cancer from two phase III trials (FIRSTANA and PROSELICA).
      ,
      • Sumanasuriya S.
      • et al.
      Elucidating prostate cancer behaviour during treatment via low-pass whole-genome sequencing of circulating tumour DNA.
      ]. It currently remains to be confirmed whether pre-treated AR-V7Pos. mCRPC patients benefit from cabazitaxel treatment, and prospective trials are still lacking. This prospective non-randomised study enrolled metastatic castration-resistant (mCRPC) patients, who had previously been treated with docetaxel and ADT, for CTC enumeration and characterisation, and explored whether subsequent cabazitaxel treatment is a viable treatment option for AR-V7Pos. mCRPC patients. Due to our inclusion criteria (more than 2 CTCs) and highly pretreated population, we captured a cohort of mCRPC patients with high incidences of AR-V7 positivity. We observed a clinically meaningful PSA response in only 12% of included AR-V7Pos. patients and therefore did not meet our primary end-point (≥25%). This modest efficacy could not be attributed due to pharmacokinetic parameters.
      Our secondary end-point regarding the utility of CTC-based AR-V7 expression as potential biomarker, revealed the prognostic value thereof. This prognostic value was independent of other previously reported prognostic markers including clinical data, dichotomised CTC counts and ctDNA-based mutational profiling. Although we previously reported that CTC-based AR-V7 status does not predict OS [
      • Gurioli G.
      • et al.
      Circulating tumor cell gene expression and plasma AR gene copy number as biomarkers for castration-resistant prostate cancer patients treated with cabazitaxel.
      ], we currently report an association between AR-V7 status at baseline and OS. We confirm the prognostic value of dichotomised CTC counts and through mutational screening showed that AR-V7 status had no significant association to known mCRPC-associated coding alterations. The absence of known genetic proxies to predict AR-V7 expression could present an opportunity for ctDNA analysis to better capture this phenomenon in retrospective studies. This would allow for a greater understanding of the incidence and role of AR-V7 within the landscape of prostate cancer and castration resistance.
      Whether CTC-based AR-V7 status has predictive value remains to be determined. In our study, AR-V7Pos. patients had a significantly worse OS compared to AR-V7Neg. and AR-V7Und. patients, regardless of dichotomised CTC counts. The detection-rate of AR-V7, and thereby its predictive value, is dependent on the chosen detection assay. We utilised the CellSearch system, as this is the only FDA-approved CTC detection method. However, we cannot rule out that utilising other alternative systems might affect the outcome [
      • Bastos D.A.
      • Antonarakis E.S.
      AR-V7 and treatment selection in advanced prostate cancer: are we there yet?.
      ]. We did not collect prospective clinical data for non-enrolled patients, however retrospective data collection was successful for a substantial proportion of these patients. This allowed us to use the cabazitaxel-treated AR-V7Neg. patients as a surrogate control group. When comparing the OS between only the cabazitaxel-treated AR-V7Pos. and AR-V7Neg. patients, no significant difference in OS remained. This suggests that CTC-based AR-V7 status is not a predictive marker for benefit from cabazitaxel treatment. However, the observed conversion of AR-V7Pos. to AR-V7Neg. within 53% of cabazitaxel-treated patients warrants additional investigation, regardless of PSA response, as these patients might be re-sensitised to additional ARTA.
      Furthermore, it is crucial to realise that outside of CABA-V7, the choice to administer cabazitaxel versus an antihormonal therapy was at the discretion of the physician. Therefore, these data indicate that clinical judgement identified a subgroup of AR-V7Neg. patients who were treated with cabazitaxel which had an overall poor survival, comparable to the AR-V7Pos. patients treated with cabazitaxel.
      Despite these significant observations, this presented study has several limitations. As this study was designed as a single arm phase II study with an already limited sample size, we also lacked a randomised control arm. However, the retrospective usage of cabazitaxel-treated AR-V7Neg. patient as surrogate control arm did allow for explorative analysis of OS. Our pool of clinical variables could have been too limited to explore other meaningful clinical associations (e.g. pain, performance, extent of disease) in relation to cabazitaxel. Moreover, the final inclusion of only 25 out of 40 AR-V7Pos. patients may have biased our results. The observed modest efficacy of cabazitaxel may partly be due to utilising PSA as a response marker, as the limitations and dynamic range of using PSA in clinical practice have been extensively reported [
      • Scher H.I.
      • et al.
      Association of AR-V7 on circulating tumor cells as a treatment-specific biomarker with outcomes and survival in castration-resistant prostate cancer.
      ]. However, other response related end-points such as the well-validated CTC response were also assessed and showed similar data [
      • Pantel K.
      • Hille C.
      • Scher H.I.
      Circulating tumor cells in prostate cancer: from discovery to clinical utility.
      ]. We currently only investigated somatic coding mutations within our a priori selected target-genes, which may be underestimating the full somatic landscape present within these malignancies.
      High CTC count (5) is known to identify a poor prognosis subgroup, including a worse response to cabazitaxel [
      • León-Mateos L.
      • et al.
      Improving circulating tumor cells enumeration and characterization to predict outcome in first line chemotherapy mCRPC patients.
      ]. The prognostic value of CTC-based AR-V7 expression seems to be lost in regards to cabazitaxel treatment. However, within this trial we cannot exclude that the cabazitaxel-treated AR-V7Neg. patients received this off-study treatment only if they presented an overall poor prognosis and therefore represented a population with overall poor response. As is, the presented observations point towards a potential role for cabazitaxel in at least compensating for the poor prognosis of AR-V7Pos. patients in regards to its AR-V7Neg. counterpart and allowing for conversion of AR-V7 status. Therefore, despite the low PSA response rate, cabazitaxel may still have a role within these difficult to treat and overall poor prognosis AR-V7Pos. patients. However, PSMA-Lu should be considered as an alternative treatment. This would however need to be confirmed in prospectively randomised trials.

      Funding/support and role of the sponsor

      This work was supported by a grant from Sanofi. Sanofi had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

      Author contributions

      Khrystany T. Isebia, Job van Riet and Martijn P. Lolkema had full access to all the data in the study and took responsibility for the integrity of the data and the accuracy of the data analysis.
      Study concept and design: Lolkema, Mostert, de Wit, Sleijfer.
      Acquisition of data: All authors.
      Analysis and interpretation of data: Isebia, van Riet, Lolkema, Mostert, Helmijr, Jansen, Oomen-de Hoop, Buck, Mathijssen, Martens.
      Drafting of the manuscript: Isebia, van Riet, Buck, de Wit, Martens, Mostert, Lolkema.
      Critical revision of the manuscript for important intellectual content: All authors.
      Statistical analysis: Isebia, van Riet, Buck, Mathijssen, Oomen-de Hoop.
      Obtaining funding: Lolkema, Sleijfer, de Wit, Mostert.
      Administrative, technical, or material support: Isebia, van Riet, Belderbos, Buck, Mathijssen, Sieuwerts, Beaufort, Helmijr, Kraan, Van, Brouwer, van IJcken, van den Hout-van Vroonhoven, Oole, Martens.
      Supervision: Lolkema, Martens supervised the lab.
      Other: None.

      Conflict of interest statement

      The authors declare the following financial interests/personal relationships which may be considered as potential competing interests:
      Ronald de Wit has acted in a consulting or advisory role for Sanofi, Merck, Astellas, Bayer, Orion; and has received research funding from Bayer, Sanofi, Travel support from Sanofi, Bayer.
      Stefan Sleijfer has acted in a consulting or advisory role for the Center for Personalized Cancer Treatment (chair), Dutch Science Agenda Personalized Medicine (chair) and Supervisory board of SkylineDx; and holds an interest in patent #10555926 (Use of cabazitaxel in the treatment of prostate cancer) held by Erasmus University Medical Center Rotterdam.
      Martijn P. Lolkema has acted in a consulting or advisory role for Sanofi, Johnson & Johnson, Merck, Astellas, Incyte, Amgen, Janssen Cilag, Bayer, Servier and Pfizer.
      John. W. M. Martens has acted in a consulting or advisory role for Novartis.
      Paul Hamberg has acted in a consulting or advisory role for Astellas, MSD, Pfizer AstraZeneca, BMS, Ipsen.
      Bianca Mostert has acted in consulting or advisory role for Servier, BMS and Lilly; and has received research funding from Sanofi, BMS.
      Ron H.J. Mathijssen acted in a consulting or advisory role Servier; and has received research funding for Sanofi, Pamgene, Astellas, ISI, Bayer, Servier, Roche, Novartis, Cristal Therapeutics, and Pfizer.
      All remaining authors have declared no conflicts of interest.

      Acknowledgements

      The authors would like to thank the included patients and their families for their willingness to participate in this study. We also would like to thank all the medical oncologists, research nurses, and laboratory technicians involved in this study for their assistance in patient inclusion, blood sampling, processing, and data analysis. Furthermore, we would like to acknowledge the Clinical Trial Center for their support in data and trial management and also like to courteously acknowledge the Erasmus MC Cancer Computational Biology Center (CCBC) for allowing us to use their computational infrastructure and expertise. The study protocol was written during the EORTC-ESMO-AACR Workshop on Methods in Clinical Cancer Research at Flims, Austria in 2017.

      Appendix A. Supplementary data

      The following are the Supplementary data to this article:
      • Suppl. Fig. 1

        Loss of AR-V7 expression (in CTCs) within a small majority of post-cabazitaxel patients shows survival benefit. a) Survival probability (OS) using univariate analysis for all patients (n = 137; y-axis), stratified and coloured by pooled WHO status at baseline, depicted in months (x-axis); censoring is shown by crosses (+). The bottom table represents the total number of remaining cases per depicted time-point. Median OS per stratified group (sorted on decreasing OS) is shown on the right-hand side. The log-rank p-value and respective hazard ratio with 95% CI is shown on top. b) Same as a), but stratified on whether patients received cabazitaxel treatment. c) Same as a), but stratified and coloured by AR-V7 status and whether patients received cabazitaxel treatment. The hazard ratio and the respective 95% CI are not shown due to multiple groups.

      References

        • Antonarakis E.S.
        • et al.
        Androgen receptor variant-driven prostate cancer: clinical implications and therapeutic targeting.
        Prostate Cancer Prostatic Dis. 2016; 19: 231-241
        • Tagawa S.T.
        • et al.
        Expression of AR-V7 and ARv(567es) in circulating tumor cells correlates with outcomes to taxane therapy in men with metastatic prostate cancer treated in TAXYNERGY.
        Clin Cancer Res. 2019; 25: 1880-1888
        • Antonarakis E.S.
        • et al.
        AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer.
        N Engl J Med. 2014; 371: 1028-1038
        • Del Re M.
        • et al.
        The detection of androgen receptor splice variant 7 in plasma-derived exosomal RNA strongly predicts resistance to hormonal therapy in metastatic prostate cancer patients.
        Eur Urol. 2017; 71: 680-687
        • Armstrong A.J.
        • et al.
        Prospective multicenter validation of androgen receptor splice variant 7 and hormone therapy resistance in high-risk castration-resistant prostate cancer: the PROPHECY study.
        J Clin Oncol. 2019; 37: 1120-1129
        • Smith T.J.
        • et al.
        Recommendations for the use of WBC growth factors: American society of clinical oncology clinical practice guideline update.
        J Clin Oncol. 2015; 33: 3199-3212
        • Sieuwerts A.M.
        • et al.
        An in-depth evaluation of the validity and logistics surrounding the testing of AR-V7 mRNA expression in circulating tumor cells.
        J Mol Diagn. 2018; 20: 316-325
        • van Dessel L.F.
        • et al.
        Application of circulating tumor DNA in prospective clinical oncology trials - standardization of preanalytical conditions.
        Mol Oncol. 2017; 11: 295-304
        • Conteduca V.
        • et al.
        Plasma androgen receptor and response to adapted and standard docetaxel regimen in castration-resistant prostate cancer: a multicenter biomarker study.
        Eur J Cancer. 2021; 152: 49-59
        • van der Doelen M.J.
        • et al.
        Impact of DNA damage repair defects on response to radium-223 and overall survival in metastatic castration-resistant prostate cancer.
        Eur J Cancer. 2020; 136: 16-24
        • Belderbos B.P.S.
        • et al.
        Associations between AR-V7 status in circulating tumour cells, circulating tumour cell count and survival in men with metastatic castration-resistant prostate cancer.
        Eur J Cancer. 2019; 121: 48-54
        • Annala M.
        • et al.
        Circulating tumor DNA genomics correlate with resistance to abiraterone and enzalutamide in prostate cancer.
        Cancer Discov. 2018; 8: 444-457
        • Robinson D.
        • et al.
        Integrative clinical genomics of advanced prostate cancer.
        Cell. 2015; 162: 454
        • Baille P.
        • et al.
        Optimal sampling strategies for bayesian estimation of docetaxel (Taxotere) clearance.
        Clin Cancer Res. 1997; 3: 1535-1538
        • de Bruijn P.
        • et al.
        Quantification of cabazitaxel in human plasma by liquid chromatography/triple-quadrupole mass spectrometry: a practical solution for non-specific binding.
        J Pharm Biomed Anal. 2012; 59: 117-122
        • A'Hern R.P.
        Sample size tables for exact single-stage phase II designs.
        Stat Med. 2001; 20: 859-866
        • Scher H.I.
        • et al.
        Trial design and objectives for castration-resistant prostate cancer: updated recommendations from the prostate cancer clinical trials working group 3.
        J Clin Oncol. 2016; 34: 1402-1418
        • Aldea M.
        • et al.
        Cabazitaxel activity in men with metastatic castration-resistant prostate cancer with and without DNA damage repair defects.
        Eur J Cancer. 2021; 159: 87-97
        • León-Mateos L.
        • et al.
        Improving circulating tumor cells enumeration and characterization to predict outcome in first line chemotherapy mCRPC patients.
        Oncotarget. 2017; 8: 54708-54721
        • Gurioli G.
        • et al.
        Circulating tumor cell gene expression and plasma AR gene copy number as biomarkers for castration-resistant prostate cancer patients treated with cabazitaxel.
        BMC Med. 2022; 20: 48
        • Mehra N.
        • et al.
        Plasma cell-free DNA concentration and outcomes from taxane therapy in metastatic castration-resistant prostate cancer from two phase III trials (FIRSTANA and PROSELICA).
        Eur Urol. 2018; 74: 283-291
        • Sumanasuriya S.
        • et al.
        Elucidating prostate cancer behaviour during treatment via low-pass whole-genome sequencing of circulating tumour DNA.
        Eur Urol. 2021; 80: 243-253
        • Bastos D.A.
        • Antonarakis E.S.
        AR-V7 and treatment selection in advanced prostate cancer: are we there yet?.
        Precis Cancer Med. 2018; 1
        • Scher H.I.
        • et al.
        Association of AR-V7 on circulating tumor cells as a treatment-specific biomarker with outcomes and survival in castration-resistant prostate cancer.
        JAMA Oncol. 2016; 2: 1441-1449
        • Pantel K.
        • Hille C.
        • Scher H.I.
        Circulating tumor cells in prostate cancer: from discovery to clinical utility.
        Clin Chem. 2019; 65: 87-99