Analysis of biomarkers in plasma-derived
exosomal RNA in prostate cancer and application
in combinatorial treatments targeting signal
transduction of androgen receptor
PhD Program Director
Prof. Stefano Del Prato
Tutor:
Chiar.mo Prof. Romano Danesi
PhD Candidate:
Dott.ssa Stefania Crucitta
Table of contents
PREFACE ... 1
SECTION 1
“Detection of small RNAs associated with prostate cancer in liquid biopsies” ... 3
SECTION 2
“Monitoring androgen receptor splice variants in circulating nucleic acids and
treatment resistance of prostate cancer” ...31
SECTION 3
“Co-targeting AR and mTOR pathways is effective in AR-V7 and PTEN positive
PCa cell line” ... 59
1
PREFACE
Prostate cancer (PCa) is the third male malignancy diagnosed and a major cause of disease and mortality among men [1, 2]. In the minority of patients whose cancers are aggressive or advanced, therapeutic options include prostatectomy, radiation therapy and androgen‐deprivation therapy (ADT) [3]. Castration‐resistant prostate cancer (CRPC) is an advanced form of prostate cancer characterized by disease progression following surgical or pharmaceutical castration. The process by which prostate cancer cells become castrate resistant is unclear, but several studies proposed that androgen ablation provides a selective advantage to androgen‐independent cells, which grow and eventually repopulate the tumour [4, 5]. CRPC patients have a poor prognosis and lower overall survival [4] in part because of the heterogeneity of the tumour that leads to resistance to hormonal therapies (abiraterone, enzalutamide). Recently, extracellular vesicles have emerged as a promising source of biomarkers in several diseases [6] and implementation of modern methods for EVs nucleic acids isolation and characterization are needed.
The aim of the present study was to address the following critical questions in this field of research:
1. to set up and optimize a technique for the detection of miRNA, snoRNA, sdRNA and tRFs in plasma (EVs) samples from (metastatic) prostate cancer patients and evaluate their diagnostic and prognostic potential.
2
2. to investigate the role of AR-V7 and AR-FL to predict resistance to hormonal therapy and to develop a new methodological approach based on plasma-derived exosomal RNA to assess this marker reliably.
3. to evaluate the combination of hormonal therapy (abiraterone) and everolimus in prostate cancer cell lines expressing the AR-V7.
3
SECTION 1
“Detection of small RNAs associated with prostate cancer in
liquid biopsies”
INTRODUCTION
MiRNAs are the focus of much attention because of their function as modulators of gene expression and their strong biomarker properties [7] (Figure 1) . Recently, miRNA expression during PCa progression was analyzed in 102 PCa tissue samples by microarrays and sncRNA sequencing [8]. Differential expression analysis identified and independently validated a diagnostic classifier (from 54 miRNAs) that detects PCa with 95% accuracy. The diagnostic and prognostic performance of this miRNA subset was further validated in independent patient cohorts. The combination of the 4 best performing miRNAs called miRNA index quote (miQ) predicts PCa with high accuracy ranging from 0.77 to 0.95 in 5 different cohorts. miQ is a robust marker successfully discriminating PCa from no-PCa even in specimens containing <25% tumour cells. Importantly, miQ is an independent predictor of aggressiveness which outperforms PSA and has clear potential to be used as a clinical tool for PCa diagnosis and as a prognostic marker of disease progression [9].
4
Figure 1. Many dysregulated microRNAs (miRNAs) affect the hallmarks of prostate cancer [10]
Over recent years, deep sequencing technologies targeting the miRNA transcriptome revealed the existence of many different RNA fragments derived from small noncoding RNA (sncRNAs) species other than miRNA. Although initially discarded as RNA turnover artefacts, accumulating evidence suggests that smaller RNAs derived from small nucleolar RNA (snoRNA) and transfer RNA (tRNA) are not just random degradation products but rather stable entities, which have functional activity in the normal cell and are deregulated in cancer [10] (Figure 1).
5
It was demonstrated that the small RNA transcriptome of PCa is dominated by sncRNAs other than miRNAs. Many of these sncRNAs originate from snoRNAs and tRNAs and are referred to as snoRNA-derived RNAs (sdRNAs) and tRNA-derived fragments (tRFs), respectively. Detailed analysis of deep-sequencing data on PCa specimens from radical prostatectomies shows that sdRNAs and tRFs are upregulated in malignant tissue compared with normal prostate or benign prostate hyperplasia tissue and are candidate diagnostic biomarkers. Validation qPCR results show that snoRNA and sdRNA expression is significantly increased in cancerous compared to normal adjacent prostate further supporting these sncRNAs as novel diagnostic markers. Furthermore, the expression of 2 snoRNAs and derived from them sdRNAs is further increased already at the time of radical prostatectomy in a specific subset of PCa patients that developed aggressive metastatic cancer years after surgery [11]. Besides snoRNA and sdRNA, the expression levels of two tRFs with opposing expression patterns (tRF-544, derived from tRNAPheGAA and tRF-315, derived from tRNALysCTT) are associated with high grade, recurrent disease. The calculated expression ratio tRF-315/tRF-544 in two separate cohorts (from EMC, n=50 and from the University of Tampere, Finland, n=104) significantly discriminates high from low-grade PCa (Gleason score < 7 vs. Gleason score ≥ 7). Moreover, high expression ratio is significantly associated with poorer progression-free survival and a shorter period to disease relapse PSA relapse after radical prostatectomy) [12].
6
AIM
The aim of this research have been:
1) to establish the best small RNA isolation method from healthy donor cohort. A preliminary test was conducted to test different RNA isolation method using plasma sample from PCa patients.
2) to set up and optimize a technique for the detection of miRNA, snoRNA, sdRNA and tRFs in plasma (EVs) samples from (metastatic) prostate cancer patients and evaluate their diagnostic prognostic, predictive, and monitoring potential.
MATERIALS AND METHODS
A total of 5 healthy donors were enrolled for the first aim of this study (Cohort 1). Plasma samples from 3 PCa patients were used for the preliminary test. A total of 11 patients were enrolled for the second aim of the study (6 PCa and 5 no-PCa patients) (Cohort 2).
Plasma collection and RNA isolation
10 ml of blood were collected in Cell safe and EDTA tubes and were centrifuged at 1600 g for 10 min at RT. Plasma samples were then centrifuged at 12000 g for 10 min at 4°C to remove cellular debris. Plasma was stored at - 80 °C until analysis.
RNA (EVs RNA and total RNA) was extracted with different kit: ExoQuick (System Biosciences, LCC), ExoRNeasy Serum Plasma Midi (Qiagen, Valencia, CA, USA) - with
7
and without EV column, part 1 - and automatic system using Maxwell machine – custom kit - (Promega, Madison, WI, USA) from 200 µl of plasma following the manufacturer’s protocol and the RNA was eluted in 20 μl and 50 μl of H2O RNeasy
free, for EVs RNA and total RNA respectively.
Analysis of EVs and total RNA
RNA extracted was reverse transcribed using TaqMan™ Advanced miRNA cDNA Synthesis Kit (Thermofisher Scientific, Waltham, Massachusetts, USA). In cohort 2, athmiRNA159a a synthetic, non-human miRNA was used as an exogenous control. PCR reactions were assembled as per manufacturer’s protocols. miRNA, snoRNA, sdRNA and tRFs expression levels were evaluated as previously reported using custom-made assay and TaqMan assay (Thermofisher Scientific, Waltham, Massachusetts, USA) [9, 11]. Quantitative real-time PCR (qPCR) was performed on an Applied Biosystems ABI 7900 thermocycler (Applied Biosystems, Waltham, Massachusetts, USA) and on StepOne Real-Time PCR System (Thermofisher Scientific, Waltham, Massachusetts, USA).
8
RESULTS
PART 1 – plasma samples from PCa patients to perform different RNA
isolation methods
The qPCR analysis of miR-21 and miR-451 revealed the efficiency of ExoRNeasy Serum Plasma Midi kit for EVs RNA and Maxwell machine (automatic system) for total RNA. Instead, the RNA samples isolated with ExoQuick method were excluded for the following analysis.
PART 2 – plasma samples from PCa patients to test the Hemolysis ratio (Ct miR-23 – CtmiR-451)
Blondal et al. (2013) [13] suggested that the ratio of red blood cell-enriched miR-451a to miR-23a, the latter microRNA being unaffected by hemolysis, can be used as a surrogate indicator of hemolysis. The delta Cq (miR-23a-miR-451) of <5, 5–8 and >8 are, respectively, indicative of samples at low, moderate or severe risk of hemolysis.
9
Figure 2. Bar graph of real-time RT-qPCR data of miRNA expression in PCa patients. Data are mean ± SEM.
10
All the 3 samples from ExoRNeasy isolation kit were under the threshold of 5, suggesting a low risk of hemolysis in that samples. Interestingly, sample #1 was under the threshold of 5 with all the 3 different isolation methods. Instead, the other samples (#2,#3) isolated with the Maxwell machine and without the first part of isolation method with ExoRNeasy Midi kit, are over the threshold of 5, suggesting a possible erythrocytes contaminations (Figure 2,3).
COHORT 1 – Healthy donor plasma samples
5 plasma samples from 3 female and 2 male healthy donors were collected in Cell Safe and EDTA tube. 200 µl of each plasma sample was used for RNA isolation using ExoRNeasy Serum Plasma Midi Kit (Qiagen, Valencia, CA, USA) and automatic system with Maxwell machine (Promega, Madison, WI, USA).
Test 1 – ExoRNeasy Midi Kit
1. Hemolysis ratio (CtmiR-23 – CtmiR-451)
All the samples were under the threshold of 5, suggesting the lower risk of erythrocyte contamination. The use of Cell safe or EDTA tubes didn’t influence the RNA isolation and the qPCR analysis. Only the samples F1, from EDTA tubes, didn’t reach an amplification of miR-451, and consequently a level of hemolysis ratio (Figure 4,5).
11
Figure 4. Bar graph of real-time RT-qPCR data of miRNA expression in the healthy donor. Data are mean ± SEM.
12
2. snoRNA and tREG
The qPCR analysis of the expression of SNORD78, SNORD44, sd78, sd44, Lys_CTT and Phe_GAA (tREG) resulted differently between samples (woman and man) and tubes (Cell-Safe and EDTA). In particular, there was no amplification for sd78 in all the samples tested. No amplification for sample F1 [(Figure 6)
14
Figure 6. Bar graph of real-time RT-qPCR data of snoRNA and tRFs expression in healthy donor. Data are mean ± SEM.
3. miQ (miR96 x miR186 / miR145 x miR221)
The qPCR analysis of the expression of miQ miRNAs resulted similarly between samples (woman and man) and tubes (Cell-Safe and EDTA). In particular, there was no amplification for miR183 in all the samples tested, excepted for sample M1 (Figure 7).
15
Figure 7. Bar graph of real-time RT-qPCR data of miRNA expression (miQ) in the healthy donor. Data are mean ± SEM.
16
4. miRNA analysis
The qPCR analysis of the expression of miRNA21, miR214 and miR378 resulted similar between samples (woman and man) and tubes (Cell-Safe and EDTA). In particular, there was no amplification of miR375 for all the samples tested.
Test 2 – Maxwell machine (automatic system)
1. Hemolysis ratio
Six samples were over the threshold of 8, suggesting the high risk of hemolysis. Only 2 samples (from Cell Safe tube) were under the threshold of 5, suggesting the low risk of erythrocyte contamination. The use of Cell safe or EDTA tubes didn’t influence the RNA isolation and the qPCR analysis (Figure 8,9).
17
Figure 8. Bar graph of real-time RT-qPCR data of miRNA expression in the healthy donor. Data are mean ± SEM.
18
2. snoRNA and tREG
The qPCR analysis of the expression of SNORD78, SNORD44, sd78, sd44, Lys_CTT and Phe_GAA (tREG) resulted differently between samples (woman and man) and tubes (Cell-Safe and EDTA).
In particular, there was no amplification for sd78 in all the samples tested (Figure 10).
20
Figure 10. Bar graph of real-time RT-qPCR data of snoRNA and tRFs expression in healthy donor. Data are mean ± SEM.
3. miQ (miR96 x miR186 / miR145 x miR221)
The qPCR analysis of the expression of miQ miRNAs resulted similarly between samples (woman and man) and tubes (Cell-Safe and EDTA). Interestingly, there was an amplification of miR183, compared to previous results (Figure 11).
21
Figure 11. Bar graph of real-time RT-qPCR data of miRNA expression (miQ) in the healthy donor. Data are mean ± SEM.
22
4. miRNA analysis
The qPCR analysis about the expression of miRNA21, miR214 and miR375 resulted similar between samples (woman and man) and tubes (Cell-Safe and EDTA). Interestingly, there was an amplification of miR375, compared to previous results (Figure 12).
23
Figure 12. Bar graph of real-time RT-qPCR data of miRNA expression (miQ) in the healthy donor. Data are mean ± SEM.
COHORT 2 –Plasma samples from cancer patients
11 plasma samples from PCa, CRC and melanoma patients were collected in Cell Safe tube. 200 µl of each plasma sample was used for RNA isolation using ExiRNeasy Midi Kit (Qiagen, Valencia, CA, USA).
1. Hemolysis ratio (CtmiR-23 – CtmiR-451)
No one samples were under the threshold of 5, suggesting the higher risk of erythrocyte contamination. 8 samples were under the threshold of 8, suggesting a possible erythrocyte contamination (Figure 13,14).
24
Figure 13. Bar graph of real-time RT-qPCR data of miRNA expression in cancer patients. Data are mean ± SEM.
Figure 14. Hemolysis ratio in cancer patients
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma Ct
miR-451
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma CtmiR-23
0 5 10 15 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma (C tmi R -23 – Ctmi R -451 )25
2. miQ (miR96 x miR186 / miR145 x miR221)
The qPCR analysis of the expression of miQ miRNAs resulted differently between samples (PCa and no-PCa) (Figure 15).
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma Ct
miRNA 96
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma CtmiRNA 183
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma CtmiRNA 145
26
Figure 15. Bar graph of real-time RT-qPCR data of miRNA expression (miQ) in cancer patients. Data are mean ± SEM.
3. miRNA analysis
The expression of miRNA 375 and miRNA 214 was present only in one sample of PCa, instead, miRNA 21 was expressed in all the samples analysed (Figure 15). 0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma
miRNA 221
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC MelanomamiRNA 375
27
Figure 15. Bar graph of real-time RT-qPCR data of miRNA in cancer patients. Data are mean ± SEM.
4. Spike-in control (ath-miR159a)
The analysis of Spike-in (ath-miR159a) demonstrated the different efficiency in the cDNA reaction. As reported in the graph, only 2 samples reached the CT value expected (in the amount of =20). This results suggested the presence of contaminants in the cDNA reactions (Figure 17).
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC Melanoma
miRNA 214
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 PCa CRC MelanomamiRNA 21
28
Figure 17. Bar graph of real-time RT-qPCR data of miRNA expression in cancer patients. Data are mean ± SEM.
DISCUSSION
This methodological study evaluated the efficiency and feasibility of small RNA isolation method in plasma samples.
The results on cohort 1 (healthy donor) highlighted the efficiency of ExoRNeasy protocol for the isolation of EVs RNA (starting from 200ul of plasma), avoiding erythrocyte contamination. The use of two different blood collection tubes (Cell safe and EDTA) didn’t interfere with the qPCR analysis. In particular, RNA isolated from Cell Safe tubes was in lower amount but better quality compared to EDTA tubes. Hemolysis ratio analysis (CtmiR-23 – CtmiR-451) resulted under the cut-off of 5 delta Ct, both for Cell Safe and EDTA tubes when RNA was isolated with ExoRNeasy MIDI kit. miRNA expression analysis resulted differently comparing female and male healthy donor (both Cell Safe and EDTA tubes). The qPCR analysis of Hemolysis ratio (CtmiR-23 – CtmiR-451) analysis on RNA isolated with the automatic system (Maxwell
0 10 20 30 40 1 2 3 4 5 6 1 2 1 2 3 -RT H2O PCa CRC Melanoma Ct
ath-miR159a
29
machine), revealed the low quality of RNA, as all the samples resulted over the threshold of 5 and 8, suggesting erythrocyte contamination and severe risk of hemolysis.
Recently, several studies have shown that the break of red blood cells (RBCs) occurring most often during blood collection or sample processing, affecting the levels of certain microRNAs detectable in plasma [14-16]. These studies only investigated the effect of hemolysis on a very limited number of cell-free microRNAs, in particular, miR-16 and miR-451, but served as a first indication and warning that hemolysis can significantly alter the levels of microRNAs in plasma [17]. Then, the expression levels of miRNA can be influenced by hemolysis [18] and variations in miRNA present in plasma sample can arise from contamination occurring due to the release of miRNA from cells.
Concerning the analysis of cohort 2 (cancer patients), all the samples were over the threshold of 5 or 8, suggesting the higher risk of erythrocyte contamination and sever risk of hemolysis. This result may influence by the qPCR analysis on miQ and miRNA expression levels. In addition, the expression variability in the different diseases (prostate, colorectal and melanoma) could be influenced by the intra-individual variation, as suggested in a recent publication [18]. In agreement with these results, a previous publication on EVs urine RNA demonstrated the less expression of miR-21 and miR-375 in PCa patients compared to control men [19].
In conclusion, the measured levels of miRNAs in plasma from patients varied in the presence of hemolysis, and since hemolysis and other factors affected miRNA
30
expression, it is important to consider these confounders while developing miRNA-based diagnostic assays.
This methodologic study demonstrates the feasibility of small RNA isolation method in plasma samples and detection by qPCR. However, due to the small population, the evaluation of the diagnostic and prognostic role of these miRNA on plasma samples from cancer patients should be confirmed.
31
SECTION 2
“Monitoring androgen receptor splice variants in circulating
nucleic acids and treatment resistance of prostate cancer”
Data presented in this chapter were published as an original article in:
Del Re M, Biasco E, Crucitta S, Derosa L, Rofi E, Orlandini C, Miccoli M, Galli L, Falcone A, Jenster GW, van Schaik RH, Danesi R. 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 Apr;71(4):680-687
Data presented in this chapter were reviewed in:
Del Re M, Crucitta S, Restante G, Rofi E, Arrigoni E, Biasco E, Sbrana A, Coppi E, Galli L, Bracarda S, Santini D, Danesi R. Pharmacogenetics of androgen signalling in prostate cancer: focus on castration resistance and predictive biomarkers of response to treatment. Critical Reviews in Oncology / Hematology, 2018;(12)5:51 - 59
32
INTRODUCTION
Castration-resistant prostate cancer (CRPC) lacks of a validated biomarker to predict resistance to hormonal therapy (HT). The key parameters for therapeutic intervention are still the clinical condition of patients and radiographic or symptomatic progression [20], whereas prostate-specific antigen (PSA) is a poor predictor of clinical response to HT or chemotherapy.
Management of CRPC is characterised by taxane-based chemotherapy (docetaxel, cabazitaxel), anti-AR therapies (abiraterone acetate, enzalutamide), immunotherapies (sipuleucel-T6), and radium-223.
The androgen receptor (AR) is an important driver of prostate cancer growth and progression, and several modifications (i.e. amplification, point mutations and splice variants) may conduct to a constant AR signalling activation [21, 22] (Figure 18).
33
In the last years, several studies suggested that AR splice variant 7 (AR-V7) may be a biomarker of resistance to hormonal therapy. AR-V7 is a truncated receptor isoform lacking the C-terminal ligand-binding domain (LBD) that is a key regulator region of the full-length AR (AR-FL) (Figure 19).
Figure 19. Structure of the AR gene and the AR full length and AR-V7 transcripts and proteins.
Modified from Lamb et al. 2017 [24].
The LBD is responsible for androgen-dependent receptor activity and the target of flutamide, bicalutamide, and enzalutamide [25, 26]. Therefore, LBD deletion results in loss of the antiandrogen binding site and constitutive activation of AR-V7 [26].
AR-V7 may be detected in tumour tissue [27], CTCs [28-30] , or in messenger RNA (mRNA) extracted from whole blood [31] (Figure 20).
34
Figure 20. Comparison of current methods for detecting AR-V7 in CTCs, blood and tissue.
Modified from Lamb et al. 2017 [24].
Unfortunately, these methods have substantial limitations, such as costs, difficult and long procedures or low sensitivity. Recently, extracellular vesicles (EVs) have emerged as a promising source of biomarkers in a number of diseases because the molecular content (i.e. nucleic acids) of EVs reflects the composition of the cell of origin [6]. In addition, it was demonstrated that cancer patients present an increased number of circulating EVs. Preliminary data in different tumours highlighted the potential application of EVs as biomarkers to identify patients that respond to therapy [32]. Recently, also the AR-full length has been hypothesized as a good predictive biomarker of response since the overexpression of the AR-FL was shown to convert prostate cancer growth from a castration-sensitive to a castration-resistant stage [33-35].
35
AIM
The aims of this research have been:
1) To confirm the role of AR-V7 to predict resistance to HT and to develop a new methodological approach based on digital droplet polymerase chain reaction (ddPCR) to assess this marker reliably.
2) To investigate whether AR-FL is also a predictive biomarker of resistance to hormonal therapy on exosomal-RNA, in addition to AR-V7.
Plasma-derived exosomal RNA was used as the source material, and the study provided new data to address the correlation between exosomal AR-V7 and AR-FL and therapy resistance, given that the translation of available data on CTCs to exosomes is not obvious.
MATERIALS AND METHODS
Patient selection
A total of 88 CRPC patients were enrolled in this study and divided into two cohorts of 36 (Cohort I) and 52 (Cohort II) patients, respectively analyzed for the two aims of the project.
The study was submitted and approved by the Ethics Committee of Pisa University Hospital and conducted in accordance to the principles of the Declaration of Helsinki;
36
all patients gave their signed informed consent before blood collection and RNA analysis.
Cohort I included 36 metastatic CRPC patients treated with enzalutamide or abiraterone as per approved label. Patients were required to have histologically confirmed prostate adenocarcinoma, progressive disease despite castration levels of serum testosterone (<500 ng/l) while on stable androgen-deprivation therapy, and documented metastases, confirmed by computed tomography or technetium-99 bone scans. Patients must have had at least three increasing serum PSA values taken at least 2 weeks before the last value of at least 2.0 ng/ml, consistent with the Prostate Cancer Working Group-2 guidelines. Prior taxane-based chemotherapy was permitted.
Cohort II included 52 metastatic CRPC, treated in first or second-line with anti-androgen therapy (enzalutamide or abiraterone) as per approved label. Inclusion criteria were the same as cohort 1.
Plasma collection and RNA isolation
Six ml of blood were collected in EDTA tubes before the start of abiraterone or enzalutamide (baseline) and centrifuged at 1900 g for 10 min at 4 °C within 2 h after drawing. Plasma was stored at -80 °C until analysis. Plasma samples were then centrifuged again at 1900 g for 15 min to remove cellular debris. Exosome isolation from plasma was performed using the exoRNeasy kit (Qiagen, Valencia, CA, USA) on
37
exoEasy spin columns, and RNA was extracted from the vesicles bound to the silica membrane using the QIAzol phenol/guanidine-based lysis solution. Chloroform was then added to QIAzol-samples and centrifuged at 12 000 g for 15 min at 4 °C. The aqueous phase containing RNA was recovered and applied to the RNeasy MinElute spin column, in which the RNA binds to the membrane and contaminants are discarded. The RNA was finally eluted in 20 μl of the elution buffer.
Analysis of AR-V7 and AR-FL on plasma-derived exosomal RNA
The analysis of AR-FL and AR-V7 in RNA was performed by ddPCR using the One-Step RTddPCR kit (ddPCR Bio-Rad), as per manufacturer instructions. VCaP cells (ATCC CRL-2876) were used to set up the ddPCR method while they are known carriers of AR-V7; VCaP RNA was used to spike clinical samples and blank specimens as the control. Results were reported as copies of mutant allele per millilitre of plasma. The
ddPCR QuantaSoft (ddPCR Bio-Rad) software determined the absolute target
concentration as copies/ml in the samples.
Data analysis
Statistical analyses were performed separately in the two cohorts. Categorical
variables, as ECOG performance status, tumor stage at diagnosis, Gleason score at diagnosis, type of primary treatment, first-line treatment, second-line treatment, presence of bone metastases, presence of lymph node metastases and presence of
38
visceral metastases, were described by absolute and relative frequencies, whereas quantitative factors as time from diagnosis to start chemotherapy and hormonal therapy and baseline total PSA level by median and range. To evaluate the normality of the quantitative data distributions, the Kolmogorov-Smirnov test was performed. PSA response rates (RRs) were compared with the Fisher exact test. Time-to-event outcomes (i.e, clinical or radiographic progression-free survival [PFS] and overall survival [OS]) were evaluated by the Kaplan-Meier method, and survival-time differences were compared by the log-rank test. The receiver operating characteristic curve (ROC curve) was created by plotting the true positive rate against the false-positive rate at various threshold settings. Quantitative variables assessment as AR-FL and AR-V7 was analyzed with Mann Whitney test (two-tailed). For AR-FL analysis patients were grouped by tertiles, to ensure approximately 15 events per group. Differences were considered significant at p<0.05. All statistical analyses, descriptive and inferential, were performed with SPSS version 24 (SPSS Inc. SPSS® Chicago, IL, USA).
39
RESULTS
COHORT I
The characteristics of patients of cohort I are summarized in Table 1.
Overall, 36 patients were enrolled, of whom 26 received abiraterone and 10 enzalutamide; median follow-up time was 9 mo (range: 2.0–31.0).
Baseline Characteristic All patients n = 36 AR-V7 negative n = 22 AR-V7 positive n = 14 Age, median (range), y 66 (51-81) 67 (51-81) 66 (57-73)
Race, No. (%) - White - Nonwhite 36 (100%) 0 (0%) 22 (100%) 0 (0%) 14 (100%) 0 (0%) ECOG performance status, No. (%)
- 0 - 1 or 2 26 (72%) 10 (28%) 17 (77%) 5 (23%) 9 (64%) 5 (36%)
Time since diagnosis, median (range), y
4.86 (0,32-17,03) 5.64 (0.91-17.03) 3.60 (0.32-12.80) Tumour stage at diagnosis, No. (%)
- T1/2 - T3/4 - unknown 3 (8%) 13 (36%) 20 (56%) 2 (9%) 9 (41%) 11 (50%) 1(7%) 4 (29%) 9 (64%)
40
Gleason Sum at diagnosis, No. (%) - ≤7 - ≥8 - unknown 16 (44%) 19 (5%) 1 (3%) 12 (55%) 9 (41%) 1 (4%) 4 (29%) 10 (71%) 0 (0%) Type of local treatment, No. (%)
- Surgery - Radiation therapy - None 17 (47%) 10 (28%) 9 (25%) 11 (50%) 8 (36%) 3 (14%) 6 (43%) 2 (14%) 6 (43%) No. of prior hormonal therapies, median 2 2 2
Current treatment - Abiraterone - Enzalutamide - Docetaxel - Cabazitaxel 26 (72%) 10 (28%) 0 (0%) 0 (0%) 18 (82%) 4 (18%) 0 (0%) 0 (0%) 8 (57%) 6 (43%) 0 (0%) 0 (0%) Prior use of abiraterone, No (%)
- Yes - No 4 (11%) 32 (89%) 0 (0%) 22 (100%) 4 (28%) 10 (72%) Prior use of enzalutamide, No (%)
- Yes - No 0 (0%) 36 (100%) 0 (0%) 22 (100%) 0 (0%) 14 (100%) Prior use of docetaxel, No (%)
- Yes - No 24 (67%) 12 (33%) 11 (50%) 11 (50%) 13 (93%) 1(7%)
41
Prior use of cabazitaxel, No (%) - Yes - No 5 (14%) 31 (86%) 1 (5%) 21 (95%) 4 (29%) 10 (71%) Presence of bone metastases, No. (%)
- Yes - No 30 (83%) 6 (17%) 16 (73%) 6 (27%) 14 (100%) 0 (0%) Presence of lymph node metastases, No. (%)
- Yes - No 23 (64%) 13 (36%) 13 (59%) 9(41%) 10 (71%) 4 (29%) Presence of visceral metastases, No. (%)
- Yes - No 7 (19%) 29(81%) 1 (5%) 21 (95%) 6 (43%) 8 (57%) Baseline PSA level, median (range), ng/mL
26.3 (0.63-4581) 22.3 (0.78-4581) 99.6 (0.63-521) Baseline Alkaline phosphatase level, median (range),
U/l 180 (49-917) 152 (49-917) 258 (53-575) Baseline lactic dehydrogenase level, median (range),
U/l 220 (110-1723) 220 (110-1723) 266 (150-1720) Baseline Hgb level, median (range), g/dl
12.3 (7.9-14.9) 13.25 (9.9-14.9) 10 (7.9-12) Use of opioid, No. (%)
- Yes - No 14 (39%) 22 (61%) 5(23%) 17(77%) 9 (75%) 5 (25%)
42
A total of 14 of 36 patients (38.8%) were AR-V7+ (100–2400 copies) (Fig. 21) before the start of HT. Seven of 14 AR-V7+ patients had a plasma sample at baseline (median: 500 copies/ml) and at progression (median: 887 copies/ml); although there was an increase, the difference was not statistically significant (paired t-test, p = 0.25) (Table 2). The AR-FL was used as an internal control to validate the extraction process; AR-V7− samples were true negative because the AR-FL signal was detectable. The median number of total droplets generated was 11 000. The ROC curve is reported in Figure 21; with the curve very close to the upper left corner, the overall accuracy of the test was very high.
Figure 21. Androgen receptor splice variant 7 (AR-V7) amplification by digital droplet polymerase chain reaction. The blue dots represent positives droplets for the AR-V7 amplification. The green dots represent positive droplets for the AR wild-type amplification. The grey dots are empty droplets or droplets containing primers dimers. The insert represents the receiver operating characteristic curve generated by the data of the present study.
43
Sample Baseline AR-V7, % Baseline AR-V7, copies/ml PD AR-V7, % PD AR-V7, copies/ml
1 3.4 360 NA NA 2 5.4 390 NA NA 3 6.5 530 NA NA 4 2.1 100 NA NA 7 6 600 3 410 9 8 700 3.5 600 10 8 500 3 500 11 6 500 4.1 1200 12 2.2 300 8 2400 13 10 600 NA NA 14 5 500 6 600 15 8 700 NA NA 16 2.8 400 3 500 17 1.8 900 NA NA 5 0 0 NA NA 6 0 0 NA NA 8 0 0 NA NA 18 0 0 NA NA 19 0 0 NA NA 20 0 0 NA NA 21 0 0 NA NA 22 0 0 NA NA 23 0 0 NA NA 24 0 0 NA NA 25 0 0 NA NA 26 0 0 NA NA 27 0 0 NA NA 28 0 0 NA NA 29 0 0 NA NA
44 30 0 0 NA NA 31 0 0 NA NA 32 0 0 NA NA 33 0 0 NA NA 34 0 0 NA NA 35 0 0 NA NA 36 0 0 NA NA
Table 2. Fractional abundance (percentage) and numbers of copies per milliliter in androgen receptor splice variant 7 samples
Clinical outcomes according to the AR-V7 analysis
Table 1 reports the characteristics of patients and their AR-V7 status. The median clinic or radiographic PFS was significantly longer (20 mo) in AR-V7− patients compared with AR-V7+ patients (3 mo; 95% CI, 1.526–14.474; p < 0.001 log-rank test) (Fig. 22a). The OS was shorter in AR-V7+ patients at baseline compared with AR-V7− patients (median 8 mo vs not reached) (95% CI, 3.751–30.285; p < 0.001 log-rank test) (Fig. 22b), making AR-V7 detection in exosomes a valuable marker of resistance to HT. The overall proportion of patients who achieved a PSA response during HT was 42% (15 of 36). In the AR-V7+ patients, the PSA RR was 7% (1 of 14 men); in the AR-V7− patients, the RR was 64% (14 of 22 men). PSA responses are shown in Figure 23. The AR-V7+ participants were more likely to be younger, with a Gleason score of at least 8, visceral metastases, higher PSA levels, and prior docetaxel treatment than AR-V7− patients.
45
Figure 22 – (a) Progression-free survival of androgen receptor splice variant 7–positive (AR-V7+) versus AR-V7-negative (AR-V7S) patients; (b) overall survival of AR-V7+ versus AR-V7S patients. AR-V7 = androgen receptor splice variant 7.
46
Figure 23. Waterfall plot depicting prostate-specific antigen (PSA) responses, according to androgen receptor splice variant 7 status. The dotted line shows the threshold for defining a PSA response (≥50% reduction in PSA level from baseline). Asterisks indicate an increase of >100% in PSA response.
Twenty-six of 36 patients were treated with taxanes before HT, of whom 11 patients were AR-V7− and 13 were AR-V7+. Clinical or radiographic PFS did not differ significantly depending on AR-V7 status. Median PFS was 11 mo in AR-V7+ patients (95% CI, 7.3–14.3 mo, p = 0.6) and 11 mo in AR-V7− patients (95% CI, 9.9–12.0 mo, p = 0.6) (Fig. 24). Univariate and multivariable Cox proportional hazard ratios were used to assess the effect of AR-V7 status on the prediction of time-to-event outcomes. In the univariate model, we analyzed known risk factors for progression such as serum alkaline phosphatase level (<30 vs ≥130 U/l), lactate hydrogenase (<250 vs ≥250 U/l), and haemoglobin (Hb) levels (<12 vs ≥12 g/dl). This latter was the only one that correlated with a better PFS (25 mo vs 3 mo; p < 0.05). Owing to the small sample size
47
and the limited number of events, the multivariable model included only AR-V7 status, AR-V7 expression levels, prior use of taxanes, and Hb levels. However, probably due to the small population, the multivariable model did not confirm the role of AR-V7 as an independent prognostic factor (hazard ratio: 3.6; p = 0.18).
Figure 24. Kaplan-Meier curves showing clinical and/or radiographic progression-free survival in taxane-treated patients, according to androgen receptor splice variant 7 status.
COHORT II
Detailed clinical characteristics of patients are reported in Table 3.
Overall, 52 patients were enrolled, of whom 36 received abiraterone and 16 enzalutamide.
48
Baseline characteristics Patients treated with hormonal therapy as first-line treatment
Patients treated with hormonal therapy as second-line treatment ECOG performance status, number (%)
0 1 2 18 (66.7%) 9 (33.3%) 0 19 (76%) 5 (20%) 1 (4%) Time from diagnosis to start of abiraterone or
enzalutamide, median (range), months
53 (10-258) 0
Time from diagnosis to start of first-line chemotherapy, median (range), months
0 40 (4-153)
Time from diagnosis to start of second-line hormonal therapy, median (range), months
0 74 (15-165)
Tumour stage at diagnosis, number (%) T1/2 N0 M0 T3/4 N0 M0 any T N1 M0 any T any N M1 Unknown 5 (18.5%) 2 (7.4%) 7 (25.9%) 9 (33.3%) 4 (14.8%) 2 (8%) 6 (24%) 4 (16%) 11 (44%) 2 (8%) Gleason score at diagnosis, number (%)
≤ 7 ≥ 8 Unknown 12 (44.4%) 12 (44.4%) 3 (11.1%) 13 (52%) 10 (40%) 2 (8%) Type of primary treatment
Surgery Radiotherapy None 16 (59.3%) 4 (14.8%) 7 (25.9%) 13 (52%) 5 (20%) 7 (28%) First-line treatment Abiraterone Enzalutamide Docetaxel 20 (74.1%) 7 (25.9%) 0 0 0 25 (100%) Second-line treatment Abiraterone Enzalutamide 0 0 16 (64%) 9 (36%) Presence of bone metastases
Yes No 18 (66.7%) 9 (33.3%) 22 (88%) 3 (12%) Presence of lymph node metastases
Yes No 16 (59.3%) 11 (40.7%) 21 (84%) 4 (16% Presence of visceral metastases
Yes No 1 (3.7%) 26 (96.3%) 3 (12%) 22 (88%)
Baseline total PSA level (ng/mL), median (range) 10.01 (0.63-106) 22.90 (1.33-569.30
49
Overall, AR-FL was detected in all patients (range: 90 – 21,500 copies/ml, median: 700 copies/ml); 15 patients (29%) were AR-V7+ (range: 80 – 700 copies/ml, median: 310 copies/ml) before the start of abiraterone or enzalutamide (HT). Considering the cohort of patients receiving HT as first line, AR-FL range was 90 – 8700 copies/ml (median: 570 copies/ml) and AR-V7 range was 80 – 500 copies/ml (median: 175 copies/ml). In the cohort of patients receiving HT as second line, AR-FL range was 100 – 21,500 copies/ml (median: 700 copies/ml) and AR-V7 range was 90 – 700 copies/ml (median: 360 copies/ml). Comparing the copies/ml of AR-FL in the overall population stratified as AR-V7+ vs AR-V7-, there was a significantly higher level of AR-FL in patients AR-V7+ vs AR-V7- (6,700 vs 490 copies/ml, p<0.0001; Fig.25A). Analyzing the copies/ml of AR-FL in patients AR-V7+ vs AR-V7- divided as per line of therapy (first and second line), the amount of AR-FL was borderline significant in patients treated as first line (1600 vs 490 copies/ml, p=0.047; Fig.25B), however, there was a significant statistical increase in AR-FL in patients AR-V7+ treated as second line (8,700 vs 445 copies/ml, p<0.0001; Fig.25C). Even if AR-V7+ patients treated as second-line HT had overall a higher amount of AR-FL compared to the first line, this was not statistically significant (8,700 vs 1,600 copies/ml, p=0.058; Fig. 25D). The scatter plot depicting the linear correlation between increased AR-FL expression and AR-V7 positivity is shown in Figure 26 (r=0.741; p<0.0001).
50
Figure 25. Box-plot graph according to androgen receptor full length (AR-FL) status and androgen receptor splice variant 7 (AR-V7) status in all prostate cancer patients (A), in first-line hormonal therapy-treated patients (B), in second-line hormonal therapy-treated patients (C), in first-line vs second-line hormonal therapy-treated patients (D). Graphs are represented as medians (lines), 25th percentile to the 75th percentile (boxes) and ranges (whiskers) for all samples.
51
Figure 26. Scatter plot depicting the linear correlation between the AR-FL increased expression and AR-V7 positivity.
Clinical outcomes analysis according to the AR status
The analysis of the overall median PFS accordingly with AR-V7 status was significantly longer in AR-V7- patients compared with AR-V7+ patients (median 25 vs 4 mo, p<0.0001; Fig. 27A). The OS was shorter in AR-V7+ patients at baseline compared with AR-V7- patients (median 9 vs 38 mo, p<0.0001; Fig. 27B).
52
Figure 27. Progression-free survival (PFS) of androgen receptor splice variant 7 positive (AR-V7+) versus AR-V7 negative (AR-V7-) patients (A); overall survival of AR-V7+ versus AR-V7- patients (B).
53
The overall proportion of patients who achieved a PSA response (PSA RR) during HT was 57.7%. In AR-V7+ patients, the PSA RR was 20%, while in the AR-V7- PSA RR was 73%. In order to evaluate the role of the AR-FL, independently of AR-V7 status, patients were divided by tertiles based on their AR-FL expression, and we found that PFS significantly correlated (p=0.014), being 30 months in patients within the lower tertile (≤400 copies/ml), 18 months in patients with intermediate AR-FL expression (401-899 copies/ml) and 5 months in patients with the higher AR-FL expression (≥900 copies/ml) (Fig. 28A). With respect to OS, the tertiles with AR-FL ≤400 and 401-899 copies/ml had an OS of 38 and not reached, respectively, vs 9 months for patients in the lower tertile (≥900 copies/ml) (p<0.0001; Fig. 28B).
54
Figure 28. Progression-free survival (PFS) of patients with androgen receptor full length (AR-FL) divided as per tertiles (A); overall survival of of patients with androgen receptor full length (AR-FL) divided as per tertiles (B). ≤400: n=17; 401-899: n=16; ≥900: n=19; AR-V7+ patients per group: ≤400: n=1 401-899: n=2 ≥900: n=12.
55
The PSA RR in the group of patients with AR-FL ≤400 copies/ml was 82.3%, in patients with AR-FL 401-899 copies/ml was 63.2%, while in patients with AR-FL ≥900 copies/ml PFS RR was 25%. To assess the effect of AR-V7 and AR-FL status on the prediction of time-to-event outcomes univariate Cox proportional hazard ratio was used. In the univariate model, known risk factors for progression such as Gleason score 7 (≤7 vs >7), age, presence of metastasis, metastasis localization, LDH were analysed. None of these variables was correlated with worse PFS. No differences in terms of PFS were seen stratifying patients treated with taxanes both for AR-V7 (median 10 vs 8 mo in AR-V7- vs AR-V7+, p=0.9) or AR-FL (median 10 vs 11 vs 8 mo in AR-FL tertiles, p=0.8).
56
DISCUSSION
Extracellular vesicles released by cancer cells carrying nucleic acids and proteins [36, 37] and reflect the tumour heterogeneity. Otherwise, CTCs released from primary tumour lesions and metastatic sites into the bloodstream, may be used to interrogate tumour biology [38-40]; however, their isolations and processing is considerably more expensive and labour intensive than exosomes. Our study demonstrated that plasma-derived exosomes is a good source of RNA to detect AR-V7 in CRPC patients, comparing the CTCs-based studies [28, 29]. In particular, PFS measured in these studies was different: 2.2 versus 6.2 months for CTCs and 3 versus 20 months for exosomes. A hypothetical reason could be the detection of false negatives in the cohort of patients analysed by RNA derived from CTCs. As an alternative, RNA may also be isolated from whole blood [31], but this approach may be adversely affected by the presence of RNA-free unstable in blood and the large amount of leucocyte’s RNA. This could explain why the analysis of AR-V7 mRNA in whole blood failed to predict the PSA response/resistance in patients treated with HT [31]. In a recent study on AR-V7 analysis on CTCs, it was shown that the line of therapy could increase the amount of AR-V7, reconfirming the role as an acquired mechanism of resistance to therapy [41].
In addition, the predictive role of this biomarker on survival parameters in CRPC patients was demonstrated in a recent publication [39] and an immunohistochemical assay was validated for AR-V7 detection in tissue biopsy [27]. Although promising, the clinical role of the AR-V7 as a predictive biomarker of resistance to HT in CRPC and its
57
inclusion in clinical practice is still being debated because of the small number of enrolled patients in published studies. Recently, a study evaluated the usefulness of the AR-V7 test in clinical practice. Overall, patients who had a change in management on the basis of AR-V7 testing were significantly more likely to benefit (as a physician-reported 50% decline in prostate-specific antigen response) on next-line therapy than those who did not change treatment (54% v 31%; P =0.015) [42]. In addition, a recent report suggested the development of blood-based tests for treatment selection in CRPC patients [43].
The analysis on cohort 2 confirmed that extracellular vesicles expression of AR-V7 is associated with CRPC patients response to abiraterone or enzalutamide. In addition, we demonstrated the association between AR-FL and clinical response to HT. It is well known that AR signalling is a major driver in CRPC and persists despite treatments with HT. Several studies evaluated the correlation between AR expression and to hormonal therapy resistance in CRPC [44-46]. Studies conducted with immunohistochemical or flow cytometry techniques denoted that AR is expressed in both hormone naïve and refractory tumours [47, 48]. However, data are discussed about the level of expression of AR [49]. A recent study showed that CRPC patients with AR amplification have a good response to hormonal therapy after first-line androgen deprivation therapy [50]. On the contrary, other published data demonstrated that AR amplification analysed in CTCs is associated with resistance to enzalutamide and abiraterone [51, 52]. Our results confirmed the heterogeneous expression of AR, however only in the AR-V7+ patient's cohort. Similarly, Djusberg et al., shown that AR amplification in CRPC bone metastases is specifically associated
58
with increased expression of AR, AR-V7 presence and poor survival [53]. In addition, Hörnberg et al. demonstrated that AR-V7 transcript levels seem to be related with high nuclear AR immunostaining scores, troubled cell cycle control and particularly poor prognosis in CRPC patients [21]. Our results suggest that AR-FL plays an important role to stratify patients likely to respond to abiraterone/enzalutamide. Interestingly, based on the different expression levels of the AR-FL it seems it is possible to identify responders and not responders patients, lower and higher tertiles respectively, but it is also possible to identify an intermediate population of patients who apparently would benefit much more from a chemotherapy regimen. However, if these patients are chemotherapy unfit, the hormonal treatment has not to be excluded, since they may have a moderate response (PFS: 18 months). On the other hand, since the majority of AR-FL higher-expressant were AR-V7+ patients, AR-FL over-expression may lead to hormone resistance just because of the presence of the constitutively active AR-V7. This study first demonstrates that plasma-derived exosomes provide a viable source of RNA for AR-V7 analysis and confirms its role as a strong biomarker of resistance to HT in CRPC patients. Furthermore, our data suggested that resistance to ADT is better predicted by the availability of both AR-FL and AR-V7 status, since AR-FL expression may identify a category of patients borderline for their response to ADT; therefore, if validated in prospective trials, both biomarkers may support a clinical decision.
59
SECTION 3
“Co-targeting AR and mTOR pathways is effective in AR-V7 and
PTEN positive PCa cell line”
INTRODUCTION
Metastatic castration-resistant prostate cancer (mCRPC) is strongly related to androgen receptor (AR) signalling pathway and its growth is dependent to the hormonal signal. For this reason, the standard regimen for patients affected by mCRPC is hormonal therapy (i.e. abiraterone, enzalutamide) or taxane-based chemotherapy (i.e. docetaxel, cabazitaxel). Unfortunately, after an initial response to hormonal therapy, tumours inevitably progress [54]. Several mechanisms were able to induce resistance to hormonal therapy, i.e. AR mutations and splicing variants, AR overexpression and activation of AR-alternative signalling pathways [22, 45, 55, 56]. The androgen receptor splice variant 7 (AR-V7) is a truncated isoform of the androgen receptor, lacking the ligand binding domain (LBD), constitutively active and capable to promote activation of target genes. AR-V7 is present in benign and malignant prostate tissue but mostly enriched in metastatic disease [57]. Relevant studies suggested that the presence of AR-V7 in tumour tissue is predictive of CRPC development and shorter PFS and survival of CRPC patients to abiraterone/enzalutamide [27-29, 58-61]. All these studies have generated promising
60
data supporting further development of AR-V7 as a treatment selection marker for CRPC patients.
The PI3K/AKT/mTOR pathway is frequently activated in prostate cancer and its constitutive activation is strongly implicated in prostate cancer progression through its interaction with other cell signalling pathways important for cellular survival, growth, and differentiation [62]. One of the key regulators of the PI3K/AKT/mTOR pathway is PTEN, which negatively control the intracellular levels of PI3K, causing an over-activation of downstream effectors (serine/threonine-protein kinase AKT and mammalian target of rapamycin mTOR) [63]. Aberrations in PTEN, such as PTEN loss, has been correlated with a poor prognosis and occurs in 50% of metastatic diseases [64]. Moreover, a lower expression of PTEN was associated with negative clinical parameters (i.e. high Gleason scores and shorter time to metastasis). It was also demonstrated that in abiraterone-treated patients with PTEN loss, the duration of treatment was shorter [64]. Consequently, the loss of PTEN leads to a constitutive activation of the PI3K pathway in CRPC [65, 66]. Preclinical studies evaluated the efficacy of PI3K-AKT-mTOR inhibitors in prostate cancer models and confirmed the importance of PTEN for the activation of the pathway and the tumour growth [67] (Figure 29).
61
Figure 29. The interaction between PI3K-AKT-mTOR and AR signalling pathway [68].
In breast cancer cell lines, everolimus (mTOR inhibitor) promotes AR transcriptional activity, increasing expression of AR target genes and the simultaneous treatment with AR and mTOR antagonists in breast cancer cell lines lead to inhibition of cell proliferation [69]. Since one of the mechanisms of resistance to hormonal therapy in CRPC is the presence of the AR-V7, the hypothesis of the present study is to evaluate
62
the combination of hormonal therapy (abiraterone) and everolimus in prostate cancer cell lines expressing the AR-V7.
MATERIALS AND METHODS
Cell lines
LNCaP, VCaP and PC3 cell lines were used for this study to reproduce three different models: LNCaP cells are PTEN-loss and were used as an androgen-dependent model, since they are AR wild-type and AR-V7 positive, VCaP cell line expresses wild-type AR and AR-V7 splice variant but wild-type PTEN and it is able to grow in an androgen-independent manner [70]. Indeed, LNCaP and VCaP cells express differently AR-V7 expression: LNCaP presents AR-V7 splice variant in a lower manner [71-73] in contrast to VCaP cells that express high expression levels. We used these cell lines to investigate the potential role of AR-V7 splice variant in treatment response, because of its involvement in the resistance to abiraterone. Instead, PC3 cell line is a hormone
insensitive and without AR.
LNCaP and VCaP cell lines were purchased from Sigma-Aldrich (Milan, Italy) and PC3 cell lines were purchased from ATCC (Manassas, VA, USA). LNCaP were cultured in RPMI 1640, VCaP in Dulbecco's Modified Eagle Medium (DMEM) and PC3 in DMEM-F12K (Sigma-Aldrich, Milan, Italy) supplemented with 10% fetal bovine serum (FBS) Aldrich, Milan, Italy), 100 U/ml penicillin, and 100 μg/ml streptomycin (Sigma-Aldrich, Milan, Italy) at 37 °C in a humidified atmosphere containing 5% CO2.
63
Drug formulations and treatment
Abiraterone acetate (cat.15148) and everolimus (RAD001, cat. 11597) were purchased from Cayman Chemicals (Ann Arbor, Michigan, USA). All drugs were diluted in dimethyl sulfoxide (DMSO) (Sigma Aldrich, Milan, Italy) and aliquots were stored at − 80°C. Experiments were designed such that none of the inhibitors or inhibitor combinations would completely kill all the cells in any given experiments.
Cell viability assay
Cell viability was measured using the CellTiter 96® AQueous One Solution Cell Proliferation Assay (MTS) following manufacturer’s instructions (Promega, Madison, WI, USA).
Cell lines were seeded at a density of 5x10^4 cells/well in a 96-well plate in 10% FBS medium. After 24 hours, the complete medium was substituted with different concentration of compounds with 1% FBS medium. Abiraterone was tested in a concentration range of 0.1-100 uM, whereas everolimus in a range of 0.01-100 uM; the association of compounds were tested at everolimus 10uM with abiraterone in a concentration range of 0.1-100uM. All treatments were performed for 24 hours.
After treatment, MTT reagent was added and cells incubated for 2 hours. At the end of the incubation period, the absorbance was measured at 540 nm using the Infinite M200 NanoQuant instrument (Tecan, Salzburg, Austria). Optical density values from vehicle-treated cells were considered 100% cell viability.
64
Internucleosomal DNA fragmentation
The apoptosis was evaluated using The Cell Death Detection ELISA Kit (Ref. 11774452001, Roche, Mannheim, Germany), according to the manufacturer's instruction. Briefly, VCaP and LNCaP cells were treated with single compounds and their association. After 24 hours, the lysates obtained by 104 cells of each sample were loaded in a streptavidin-coated plate. Into each well, a mixture of anti-histone-biotin and anti-DNA-POD was added. After 2 hours’ incubation at room temperature, the amount of nucleosomes in the immune-complex was quantified by photometrically with ABTS (2,2’-azinobis-3-ethyl-benzothiazoline-6-sulfonic acid) as substrate using Infinite M200 NanoQuant instrument (Tecan, Salzburg, Austria) microplate reader at a wavelength 405 nm.
Statistical Analysis
All experiments were performed as triplicate and the results analyzed by GraphPad Prism 7 (GraphPad Software, San 165 Diego, CA, USA). Data were shown as mean values ± standard deviation (SD) obtained from at least three separate experiments. The level of statistical significance was p-value < 0.05 and was calculated by one-way ANOVA test.
65
RESULTS
VCaP and LNCaP cell proliferation is reduced by AR and mTOR inhibition
Abiraterone treatment decreased in a concentration-dependent manner the cell viability of LNCaP (IC50 10 ± 1.01 M) and VCaP (IC50 25 ± 1.02 M) cell lines and less PC3 (IC50 35 ± 1.02 M). Everolimus treatment showed to affect cell viability in LNCaP and PC3 in a concentration-dependent manner (IC50 30± 1.05 and ± 1.02M, respectively). Instead, in VCaP cells even the maximum concentration tested (100M) showed not able to reduce cell viability more than 20% compared to control (Fig. 30). Interestingly, the association treatment induced a concentration-dependent cytotoxicity in LNCaP and VCap cell lines. The drug combination caused a significant cytotoxicity in both the LNCaP and VCaP cells compared to the single compounds (in LNCap cells: p-value < 0.05, compared to abiraterone and p<0.0001 compared to everolimus; in VCap cells p<0.001, compared to abiraterone and p<0.0001 compared to everolimus). In PC3 cells, the association didn’t improve the cytotoxicity compared to single treatments.
66
Figure 30. Concentration-response curves in human cutaneous prostate cancer cell lines after treatment with abiraterone (ABI) and everolimus (EVE), both alone and in association (ABI+ EVE) for 24 h.
Combined treatment induces apoptosis in PCa cell lines
Internucleosomal DNA fragmentation induced by abiraterone-everolimus association in VCaP and in LNCaP cells was 5- and 2.5-fold greater, respectively, than that observed in control cell line PC3 (Fig. 31). Apoptosis resulted significantly improved in VCaP and LNCaP cells treated with the association compared to abiraterone compared to control (in VCaP p<0.0001; in LNCaP p<0.0005) and everolimus compared to control (in VCaP p<0.0001; in LNCaP p<0.0071) alone. Furthermore, the association treatment in VCaP cell lines enhances significantly more the
67
internucleosomal fragmentation compared to the association effect in LNCaP cell lines (p=0.0061). These data were in line with cytotoxicity previously observed, highlighting that VCaP seems the most sensitive cell line to the tested association.
Figure 31. Fragmentation of internucleosomal DNA in VCaP and LNCaP cell lines. VCaP and LNCaP cells after treatment abiraterone and everolimus both alone and in association, at corresponding IC50. Values were expressed as mean ± standard deviation (SD) from three separate experiments. **p<0.01, ***p<0.001, as compared to abiraterone treatment; ## p <0.01 as compared to association treatment in LNCaP cells (one-way ANOVA followed by Dunnetts’s multiple comparison test).
68
DISCUSSION AND FUTURE PROSPECTIVES
Androgens and AR activity are strongly related not only to development and progression of prostate cancer but they are also the main cause of castration-resistant disease [54, 74]. Several mechanisms are involved in the resistance onset as well as AR-independent pathways, as PI3K-AKT-mTOR. These preliminary results highlighted the involvement of PI3K/AKT/mTOR pathway in pharmacological resistance to AR inhibition. Indeed, in the androgen receptor positive cell line, the combined treatment with abiraterone and everolimus induced a significant increase of concentration-dependent cytotoxicity and apoptosis compared to single treatment strategy. This trend could be in agreement with the ability of these pathways to regulate each other by reciprocal negative feedback: the inhibition of only AR pathway can induce the activation of PI3K/AKT/mTOR causing treatment resistance and vice versa [67]. Instead, the co-targeting approach seems able to overcome the over-activation of intracellular signalling that can make cancer cells resistant to single treatment.
Interestingly, in AR-V7 positive and wild-type PTEN cell line (VCaP) the improvement of IC50 and of apoptosis in the association compared to single treatment was more prominent than in AR-dependent but PTEN-loss cell line, as LNCaP cells, and in the hormone-insensitive and PTEN-loss cell line, as PC3. The different behaviour of two androgen receptor positive cell lines can be strictly linked to two different features: their different status of PTEN, a physiological inhibitor of PI3K/AKT/mTOR pathway and their different AR-V7 mRNA expression levels. In conclusion, our findings demonstrate that co-targeting mTOR and AR pathway showed antitumor efficacy in
69
AR-V7 and PTEN positive CRPC cell lines. As AR-LBD–targeting drugs may have limited or no effect on AR-Vs, this novel approach may provide a therapeutic advantage for CRPC patients that are resistant to hormonal therapy by a mechanism involving the expression of AR-Vs. Future investigation will be performed using qPCR to test mRNA expression of AR pathway, and western blot analysis will be performed to analyse the PI3K, AKT, mTOR pathway. In addition, treatment with chemotherapy (taxanes) in combination with hormonal therapy (abiraterone) and mTOR inhibitor (everolimus) will be tested.
70
REFERENCES
1. Pernar, C.H., et al., The Epidemiology of Prostate Cancer. Cold Spring Harb Perspect Med, 2018.
2. Bray, F., et al., Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin, 2018. 3. Heidenreich, A., et al., EAU guidelines on prostate cancer. Eur Urol, 2008. 53(1):
p. 68-80.
4. Kirby, M., C. Hirst, and E.D. Crawford, Characterising the castration-resistant prostate cancer population: a systematic review. Int J Clin Pract, 2011. 65(11): p. 1180-92.
5. Chandrasekar, T., et al., Mechanisms of resistance in castration-resistant prostate cancer (CRPC). Transl Androl Urol, 2015. 4(3): p. 365-80.
6. Lane, R.E., et al., Extracellular vesicles as circulating cancer biomarkers: opportunities and challenges. Clin Transl Med, 2018. 7(1): p. 14.
7. Fabris, L., et al., The Potential of MicroRNAs as Prostate Cancer Biomarkers. Eur Urol, 2016. 70(2): p. 312-22.
8. Martens-Uzunova, E.S., et al., Diagnostic and prognostic signatures from the small non-coding RNA transcriptome in prostate cancer. Oncogene, 2012. 31(8): p. 978-91.
9. Larne, O., et al., miQ--a novel microRNA based diagnostic and prognostic tool for prostate cancer. Int J Cancer, 2013. 132(12): p. 2867-75.
10. Bijnsdorp, I.V., et al., The Non-Coding Transcriptome of Prostate Cancer: Implications for Clinical Practice. Mol Diagn Ther, 2017. 21(4): p. 385-400.
11. Martens-Uzunova, E.S., et al., C/D-box snoRNA-derived RNA production is associated with malignant transformation and metastatic progression in prostate cancer. Oncotarget, 2015. 6(19): p. 17430-44.
12. Olvedy, M., et al., A comprehensive repertoire of tRNA-derived fragments in prostate cancer. Oncotarget, 2016. 7(17): p. 24766-77.
71
13. Blondal, T., et al., Assessing sample and miRNA profile quality in serum and plasma or other biofluids. Methods, 2013. 59(1): p. S1-6.
14. Kirschner, M.B., et al., Haemolysis during sample preparation alters microRNA content of plasma. PLoS One, 2011. 6(9): p. e24145.
15. McDonald, J.S., et al., Analysis of circulating microRNA: preanalytical and analytical challenges. Clin Chem, 2011. 57(6): p. 833-40.
16. Pritchard, C.C., et al., Blood cell origin of circulating microRNAs: a cautionary note for cancer biomarker studies. Cancer Prev Res (Phila), 2012. 5(3): p. 492-497.
17. Kirschner, M.B., et al., The Impact of Hemolysis on Cell-Free microRNA Biomarkers. Front Genet, 2013. 4: p. 94.
18. Wu, J., et al., Intra-individual variation of miRNA expression levels in human plasma samples. Biomarkers, 2018. 23(4): p. 339-346.
19. Koppers-Lalic, D., et al., Noninvasive prostate cancer detection by measuring miRNA variants (isomiRs) in urine extracellular vesicles. Oncotarget, 2016. 7(16): p. 22566-78.
20. Crawford, E.D., K. Ventii, and N.D. Shore, New biomarkers in prostate cancer. Oncology (Williston Park), 2014. 28(2): p. 135-42.
21. Hornberg, E., et al., Expression of androgen receptor splice variants in prostate cancer bone metastases is associated with castration-resistance and short survival. PLoS One, 2011. 6(4): p. e19059.
22. Del Re, M., et al., Pharmacogenetics of androgen signaling in prostate cancer: Focus on castration resistance and predictive biomarkers of response to treatment. Crit Rev Oncol Hematol, 2018. 125: p. 51-59.
23. Feldman, B.J. and D. Feldman, The development of androgen-independent prostate cancer. Nat Rev Cancer, 2001. 1(1): p. 34-45.
24. Lamb, A.D., M.G. Lawrence, and S. Sandhu, Practical Polling for Prostate Cancer: AR-V7-based Treatment Selection. Eur Urol, 2017. 71(6): p. 883-885.
72
25. Dehm, S.M. and D.J. Tindall, Alternatively spliced androgen receptor variants. Endocr Relat Cancer, 2011. 18(5): p. R183-96.
26. Hu, R., et al., Ligand-independent androgen receptor variants derived from splicing of cryptic exons signify hormone-refractory prostate cancer. Cancer Res, 2009. 69(1): p. 16-22.
27. Welti, J., et al., Analytical Validation and Clinical Qualification of a New Immunohistochemical Assay for Androgen Receptor Splice Variant-7 Protein Expression in Metastatic Castration-resistant Prostate Cancer. Eur Urol, 2016. 70(4): p. 599-608.
28. Antonarakis, E.S., et al., AR-V7 and resistance to enzalutamide and abiraterone in prostate cancer. N Engl J Med, 2014. 371(11): p. 1028-38.
29. Antonarakis, E.S., et al., Clinical Significance of Androgen Receptor Splice Variant-7 mRNA Detection in Circulating Tumor Cells of Men With Metastatic Castration-Resistant Prostate Cancer Treated With First- and Second-Line Abiraterone and Enzalutamide. J Clin Oncol, 2017. 35(19): p. 2149-2156.
30. Onstenk, W., et al., Efficacy of Cabazitaxel in Castration-resistant Prostate Cancer Is Independent of the Presence of AR-V7 in Circulating Tumor Cells. Eur Urol, 2015. 68(6): p. 939-45.
31. Takeuchi, T., et al., Detection of AR-V7 mRNA in whole blood may not predict the effectiveness of novel endocrine drugs for castration-resistant prostate cancer. Res Rep Urol, 2016. 8: p. 21-5.
32. Minciacchi, V.R., et al., Extracellular vesicles for liquid biopsy in prostate cancer: where are we and where are we headed? Prostate Cancer Prostatic Dis, 2017. 20(3): p. 251-258.
33. Chen, C.D., et al., Molecular determinants of resistance to antiandrogen therapy. Nat Med, 2004. 10(1): p. 33-9.
34. Efstathiou, E., et al., Effects of abiraterone acetate on androgen signaling in castrate-resistant prostate cancer in bone. J Clin Oncol, 2012. 30(6): p. 637-43.
