We screened datasets from Gene Expression Omnibus (GEO), and The Cancer Genome Atlas to explore . In this article, we selected datasets of localized samples and CRPC samples from the GEO database. Finally, GSE32269, GSE32982, and GSE70770 met our inclusion criteria. Details of these three datasets are listed in the Supplementary Table. We used the SVA package in R language to combine the data into a database containing 276 samples, of which 45 were CRPC samples and 231 were localized samples.
Screening and identification of the differentially expressed genes (DEGs)
We compared the protein expression data of the localized samples with those of metastatic samples using limma package of R (3.48.3) to identify the significant DEGs (|Log2FC | > 1.3, adjusted p value < 0.05) . Volcano plots and heatmaps were used to characterize the DEGs, and the KEGG pathway database was used to choose the signal pathways in DEGs enrichment [18, 19].
At first, we selected DEGs between localized and metastatic samples from the combined set of GSE32269, GSE32982, and GSE70770. Thereafter, we executed the weighted gene co-expression network analysis (WGCNA) (1.70-3) on the basis of DEGs in R. The DEGs were hierarchically clustered into eight gene modules when the β value was defined as 6.
We downloaded the data from GSE32269 dataset and analyzed the ROC curve in SPSS to identify the specificity and sensitivity. Survival analysis was performed on the GSE16560 dataset.
Continuous data were described using mean ± standard deviation or median (interquartile range). The t-test was used to compare normal data while the Mann–Whitney U test was used to compare non-normal data. Spearman correlation analysis was used to analyze the correlations in the cross-sectional study. The receiver operating characteristic (ROC) curve and area under the curve were used to calculate the best predictive cut-off value; p values < 0.05 were considered statistically significant. Statistical analyses of the data were performed using GraphPad Prism version 8.0 and SPSS version 25.0.
Serum sCD206 levels of patients with PC
All serum samples were routinely collected from patients before administering treatments, during hospitalization, and stored at −80 °C. The concentrations of sCD206 were measured using commercial enzyme-linked immunosorbent assay kits (Human MMR ELISA Kits, RayBiotech, Norcross, GA). First, 100 μl of standard solutions or samples was added to each and incubated for 2.5 h. After four washes, 100 μl of prepared biotin antibodies was added to each well. After 1-h incubation, 100 μl of prepared streptavidin solution was added and incubated for 45 min. The mixture was washed four times, and 100 μl of TMB one-step substrate reagent was added to each well and incubated for 30 min, followed by another four rounds of washing. Finally, 50 μl of stop was added to each well.
Prostate tissue specimens, used in this study, were surgical specimens from patients with PC haying complete clinicopathological data. ADPC specimens were acquired by radical prostatectomy, and BPH/CRPC specimens were acquired by transurethral resection of the prostate. These samples were paraffin-embedded and subjected to IHC with standard DAB staining protocols. All tissue samples were obtained from patients with PC, who had undergone surgical operation in the Second Affiliated Hospital of Tianjin Medical University (Tianjin, China), and were inspected by three qualified pathologists to acquire accurate grades. The main site of metastasis was bone tissues in patients with mPC.
Benign prostatic hyperplasia cells (BPH) and various PC cell lines (22Rv1, C4-2, PC3, LNCaP, and DU145) were acquired from ATCC. The cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin in a humidified environment containing 5% CO2 at 37 °C.
Western blot (WB)
Proteins from BPH, 22Rv1, C4-2, PC3, LNCaP, DU145, and different carcinomas were extracted using PMSF and RIPA. BCA kit was used to detect the concentration of different proteins. Protein samples were separated by SDS-polyacrylamide gel electrophoresis in 10% acrylamide gel and then transferred to polyvinylidene fluoride membrane. Next, the was blocked with skimmed milk powder (5%) and incubated with primary antibodies [GAPDH (1:1000) (ab9485), AR-V7 (1:1000) (ab198394), vimentin (1:1000) (ab92547), β-catenin (1:1000) (ab32572), β-tubulin (1:1000) (ab78078), E-cadherin (1:1000) (ab40772), claudin-1 (1:1000) (ab211737), slug (1:1000) (ab180714), twist (1:1000) (ab175430), snail (1:1000) (ab180714)] at 4 °C overnight. The membrane was washed twice with PBS, and incubated with anti-mouse IgG/anti-rabbit IgG at normal temperature for 1 h. The membrane was washed twice with PBS once again. Finally, wb automatic chemiluminescence imaging system was used to detect the bands.
For flow cytometry analysis, the tumor mass was dissociated into single cells. Prior to antibody staining, red blood cells were removed with ammonium chloride-potassium lysis buffer for 3 min at room temperature, followed by incubation or staining with cell surface antibody [APC anti-mouse CD206 (BioLegend)] for 30 min on ice. The cells were then washed twice and re-suspended in FACS buffer. Flow cytometry was performed using a CytoFLEX flow cytometer (Beckman Coulter), and the resulting data were analyzed using CytExpert software.
After clinical and animal laboratory surgery, we obtained human tissue specimens and mouse tumor specimens, which were then fixed with formalin. We prepared pathological sections of the specimens by freezing, paraffin fixing, and sectioning. Next, we put the pathological sections into the oven at 60 °C for 60 min, dewaxed the slices in xylene, and used graded alcohol for rehydration. Thereafter, we used PBS to wash the pathological sections twice, and used citric acid buffer to recover the pathological sections (7 min on high fire and 10 min on medium fire). The pathological sections were washed twice with PBS, and endogenous peroxidase was added to sections for 20 min. Finally, primary antibody was added to the pathological section and left in the refrigerator at 4 °C overnight. The next day, we uesd secondary antibody to detect after washing the pathological sections with PBS twice. DAB chromogen was used for detection, and hematoxylin was used for redyeing it after washing with tap water. After dehydration, the plates were sealed with neutral gum and photographed under a microscope.
We seeded PC cells into 6-well plates. Next, we drew a straight line in the middle of the plates with the tip of the 10-μl microsphere after transfecting negative control siRNA, negative control siRNA + IL-17A, CTSK siRNA, and CTSK siRNA + IL-17A into PC cells (after 24 h). Thereafter, we washed the plates twice with PBS and cultured the cells in a cell incubator for 24–72 h. Photographs were taken under a microscope at 0, 24, and 72 h, respectively.
Clone formation assay
We seeded digestive cells (2.0 × 103 DU145 cells, 2.0 × 103 LNCaP cells) into a 6-well plate. Then, we transfected negative control siRNA, negative control siRNA + IL-17A, CTSK siRNA, or CTSK siRNA + IL-17A into the cells after 24 h. After 1–2 weeks of culturing the PC cells, the plates were washed twice with PBS. Next, we fixed the cells with paraformaldehyde and and used PBS to wash the plates again twice. Finally, the cells were stained with crystal violet for 0.5 h, and the plates were washed twice with PBS and dried thereafter.
We transfected CTSK-siRNA or negative control siRNA into DU145 or LNCaP cells, respectively. We added 2 × 104 cells and 1640 (10% FBS) to the top of transwell insert, and 1640 (10% FBS) to the bottom chamber. Then, we cultivated the cells for 48 h at 37 °C, and used PBS to wash the chambers twice. Next, we fixed the cells with paraformaldehyde and used PBS to wash the chambers twice. Finally, the cells were stained with crystal violet for 1 h.
Tumor xenografts mouse model
Male mice were injected with 2 × 106 PC3 cells, suspended in 150 μl of Matrigel and 1640 medium, under the skin of the abdomen in control, control + IL-17A, shCTSK, and shCTSK + IL-17A groups. Tumor volume data were collected for at least 2 weeks, being measured at the same time every day. Finally, the mice were sacrificed and weight of the tumor were measured with precision. Parts of the fresh specimens were examined by flow cytometry to verify the immune-related indicators. Rest of the mouse tumors were fixated with paraformaldehyde, and immunohistochemical staining was performed for markers of CTSK, β-catenin, vimentin and E-cadherin. All procedures involving mice were approved by the University Committee on Use and Care of Animals at the Tianjin Medical University and met all regulatory standards.
This study was approved by the ethics committee of Fujian Provincial Hospital (reference number, K2019-10-017), and all methods were carried out following relevant guidelines and regulations. This study was carried out in compliance with the Declaration of Helsinki. Informed consent was obtained from all participants and/or their legal guardians.
Sample size calculation
We conducted a prospective observational study to analyse the difference between 99mTc-PSMA SPECT/CT and 99mTc-MDP SPECT/CT in the detection of bone metastasis in PCa. In this study, the sensitivity and specificity of 99mTc-PSMA and 99mTc-MDP scans were assumed to be greater than 50% (H0 = 50%). Referring to similar published literature on 68 Ga-PSMA and 99mTc-MDP scans78, an H1 = 80% was assumed. PASS 11 software (power analysis and sample size, NCSS, LLC) was used to estimate the required sample size. Assuming α = 0.05 (unilateral), β = 0.1, and a 1:1 ratio between groups, the calculations indicated that at least 46 patients needed to be included in the study. Ultimately, 74 individuals were enrolled in the study.
A total of 74 men were enrolled in this study from October 2019 to November 2021. The inclusion criteria were as follows: (1) PCa confirmed by surgical or puncture histopathology; (2) completion of initial treatment [such as radical prostatectomy (RP), external-beam radiotherapy (EBRT), endocrine therapy, etc.]; (3) biochemical recurrence (defined as: (i) a PSA level ≥ 0.2 μg/L on two consecutive measurements after RP; (ii) after EBRT, any PSA increase > 2 ng/mL higher than the PSA nadir value, regardless of the serum concentration of the nadir)9; and (4) complete medical records, control data and clinical follow-up results. The exclusion criteria were as follows: (1) the presence of sever syndromes that were difficult to manage; (2) active or upcoming participation in other clinical drug trials; (3) lack of regular review or follow-up results; and (4) inability to obtain relevant contrast imaging and clinical data. All eligible patients underwent both 99mTc-PSMA SPECT/CT and 99mTc-MDP SPECT/CT at an average interval of 12.1 days (1–14 days). None of the patients received any antineoplastic therapy between the two scans. The characteristics of the patients are given in Table 1.
A PSMA lyophilized kit was provided by Shanghai Engineering Research Center of Molecular Imaging Probes. Before each use, a bottle of lyophilized reagent was selected, and after 5 min, 4 mL 0.9% NaCl solution was added to dissolve the reagent, followed by approximately 5 mL 3.7–4.44 GBq 99mTcO–4 solution. The solution was then mixed well and heated in a 100 °C water bath for 10 min. The radiochemical purity (RCP) of the 99mTc-PSMA was assessed by analytical high-performance liquid chromatography (HPLC) on an Agilent 1200 system6. The 99mTc-PSMA was discarded if the RCP was lower than 95%. The 99mTc-MDP was provided by Guangdong CI Pharmaceutical Co., LTD. Fuzhou Branch. Quality control (QC) of 99mTc-MDP was carried out by the manufacturer.
For the 99mTc-PSMA scan, all patients were injected intravenously with a dose of 0.74 GBq (20 mCi) 99mTc-PSMA. Whole-body planar imaging and regional (neck-pelvic) SPECT/CT were performed 2 h after injection on a Discovery NM/CT 670Pro (GE, USA) with low energy high resolution collimators. The image acquisition protocol was as follows: (1) Planar imaging: peak energy 140 keV (99mTc) and scan velocity 15 cm/min in a 1025 × 256 matrix. (2) Regional SPECT/CT: camera matrix size 128 × 128, zoom 1.0, rotation 360°, and 30 s/frame for a total of 60 frames. For CT, low-dose CT (130 keV; 60 mAs) was used.
For the 99mTc-MDP scan, a dose of 0.74 GBq (20 mCi) 99mTc-MDP was injected intravenously, and imaging was performed following a delay of 3 to 5 h. The imaging instrument and acquisition protocol were the same as those of the 99mTc-PSMA scan.
Image processing was performed on workstations (Xeleris, General Electric, Waukesha, WI). All images were anonymized and analysed by 3 senior nuclear medicine physicians. On SPECT/CT, areas with higher imaging agent uptake than normal tissue after excluding physiological uptake and traumatic fracture were considered “imaging positive bone lesions”. Areas with abnormal SPECT/CT findings but no imaging agent uptake on the corresponding site of SPECT were considered negative lesions. The flow diagram is shown in Fig. 1.
Interpretation of the degree of uptake of the positive lesion imaging agent by semiquantitative evaluation
(1) For the 99mTc-PSMA scan, the region of interest (ROI) of the lesions was delineated on whole-body plane imaging, and a mirror ROI was delineated on the liver, avoiding the gallbladder to the greatest extent possible. Then, the ratio of focus to liver (F/L) was calculated; if F/L > 1, the lesion was regarded as having high uptake (score: 3); if F/L = 1, it was regarded as having moderate uptake (score: 2); and if F/L < 1, it was regarded as having mild uptake (score: 1). (2) For the 99mTc-MDP scan, the ROI of the lesions was delineated on whole-body plane imaging, and the mirror ROI was delineated on the contralateral part (if the lesion was located in the spine, the adjacent vertebrae were delineated), and the ratio of focus to contralateral (F/CL) was calculated. If F/CL ≥ 2, lesion uptake was regarded as high (score 3); if F/L was 1.5–2, it was regarded as moderate (score: 2); and if F/L was 1–1.5, it was regarded as mild (score: 1).
Diagnostic criteria for “typical metastasis” and “equivocal metastasis”
(1) Diagnostic criteria for “typical metastasis” (i) typical metastatic changes (increased bone density and/or destruction of bone structure, with or without surrounding soft tissue mass) on CT (fused with SPECT) of the lesion and an imaging score of 1–3. (ii) lack of typical benign or metastatic changes on CT (fused with SPECT) and an imaging score of 2–3. (2) Diagnostic criteria for “equivocal metastasis”: (iii) no typical benign or metastatic changes on CT (fused with SPECT) and an imaging score of 1.
Validation criteria for bone metastases
The pathological criteria for bone metastasis are difficult to obtain; thus, the clinical diagnosis method reported in a previous study was used7. All patients were followed up for at least 6 months (or until death). Serum PSA was reviewed every 3 months for all patients. Subsequent therapeutic schedule options depended on the patient’s condition, including radical prostatectomy, local radiation therapy, chemotherapy, abiraterone, etc. 99mTc-PSMA and 99mTc-MDP imaging were routinely performed every 6 months to describe changes in activity on bone lesions. Future imaging modalities (CT, magnetic resonance (MR), positron emission tomography (PET)/CT, PET/MR, etc.) were selected according to their respective clinical needs and were not bound by a specific protocol. For patients with BR, 18F-FDG PET/CT was performed annually. One of these patients was found left supraclavicular fossa lymph node metastases both by 18F-FDG PET/CT and 99mTc-PSMA SPECT/CT at follow-up one year later.The material was assessed by the specialists involved in the study (nuclear medicine physicians and urologists) to determine the affected bone regions and overall metastatic status. Patients who met at least two of the following conditions were clinically diagnosed with bone metastasis: (1) two or more imaging scans suggestive of bone metastases; (2) symptoms of bone pain and imaging examination suggesting bone metastasis at the site of bone pain, which was relieved after antitumor therapy; (3) a reduction in size or activity for positive metastases after antitumor therapy on imaging examination; and (4) PSA ≥ 100 ng/mL, suggesting distant metastasis10.
Data analysis was performed using SPSS 19.0 software (statistical product and service solutions, Chicago, IL). The sensitivity and specificity of the two imaging methods were calculated, and using receiver operating characteristic (ROC) curve analysis, the area under the ROC curve (AUC) was calculated, compared and analysed between the two methods. The Wilcoxon signed-rank test was used to analyse the difference between the proportion of “typical metastasis” versus “equivocal metastasis” as detected by the two imaging methods. The Wilcoxon rank-sum test was used to analyse the difference in the number of bone metastatic lesions detected by the two imaging methods. Binary logistic regression analysis and ROC curve analysis were used to calculate the predictors and optimal critical values of the positive results of the two imaging methods. P < 0.05 was considered statistically significant.
Therapeutic resistance in prostate cancer can be driven by lineage plasticity, but the mechanisms behind this are unclear, and therapies to prevent or reverse the process are nonexistent. A new study reveals the JAK/STAT signaling axis as a driver of lineage plasticity with tremendous therapeutic potential.
Over the past decade, advances in the mechanistic understanding of prostate cancer biology have led to significant improvements in patient survival. Although nearly all patients with advanced prostate cancer initially respond to therapies targeting androgen receptor (AR) signaling, castration-resistant prostate cancer (CRPC) resistant to these therapies frequently develops. Most resistant tumors remain dependent on the AR signaling pathway through a variety of genomic and epigenetic mechanisms, including AR gene amplification, mutations and splice variants. However, nearly 40% of CRPC tumors display evidence of transitioning away from an AR-dependent phenotype toward alternative lineage programs1. These AR-independent tumors are further subcategorized into distinct phenotypes, including neuroendocrine prostate cancer (NEPC), double-negative prostate cancer (DNPC) and intermediate phenotypes. Shared genomic alterations between NEPC and prostate adenocarcinoma indicate that NEPC tumors arise clonally from adenocarcinoma during treatment and that significant epigenetic deregulation occurs during the transition process2,3,4. With the increased use of more potent AR-targeted therapies in the past decade, the emergence of AR-negative tumors during the course of treatment has more than tripled (from 11.7% to 36.6%)1. Many studies have identified potential drivers of this transition to androgen independence, including loss of CHD15,6, concurrent loss of TP53 and RB17, and upregulation of MYCN8,9, EZH210, BRN211 and SOX212. Although these have been nominated as drivers of lineage plasticity and NEPC, there is still much to learn about the molecular and genomic changes that promote the acquisition of the AR-indifferent phenotypes, the progression of prostate adenocarcinoma cells to androgen independence following AR-targeted therapies, and the hierarchical and temporal relationships among different cell populations.
Neural lineage network enriched in enzalutamide-resistant prostate cancer cells
It has been well established that the alteration of the neural-associated molecular landscape in castration-resistant tumors may contribute to the androgen and antiandrogens indifferences and the neuroendocrine progression after the AR-targeted treatment5,7,13. However, the overall neural lineage network underlying the end-stage emergences of these neuroendocrine markers remains unknown. We have previously generated an enzalutamide-resistant cell subline named C4-2B MDVR from C4-2B cell line through a long-term culture of C4-2B cells in the presence of increasing doses of enzalutamide22. Recent studies have shown multiple characteristics presented in C4-2B MDVR cells including overexpression of AR/AR-V7, Wnt signaling activation such as Wnt5a and WLS, and increased expression of markers of neuroendocrine such as NSE and CHGA23,24, suggesting that C4-2B MDVR cells may represent multiple cellular lineages including neuroendocrine.
To dissect the molecular changes associated with neuroendocrine differentiation in C4-2B MDVR cells, we first analyzed several well-known NED gene signatures in the transcriptome of C4-2B MDVR cells. Genes encoding several of the classic NE markers such as ENO2, CHGA, and SYP were upregulated in C4-2B MDVR cells compared to the parental C4-2B cells (Fig. 1a). Since neuroendocrine cells may derive from neural crest cells and embryonic stem cells, we analyzed the transcriptome of C4-2B MDVR cells for the pathways relative to neural stem cell proliferation and neuron differentiation and projection pathways (Supplementary Table 1) in order to determine if neural lineage genes are enriched in C4-2B MDVR cells. Among these pathways, we found that eight pathways are significantly enriched (NES <−1.4, FDR p value <0.05) in the transcriptome of C4-2B MDVR cells compared to that of parental C4-2B cells (Fig. 1b, c). The eight enriched pathways fall into four major categories, including neural stem cell differentiation, neural precursor cell proliferation, embryonic stem cell pluripotency, and neuron differentiation/projection (Fig. 1b, c). Our data suggest that neural lineages may emerge from enzalutamide-resistant cells, which may adopt a cell plasticity of early or intermediate stage of neuroendocrine differentiation.
Neural lineage pathways and gene signatures in clinical neuroendocrine prostate cancer
To examine if the neural lineage pathways identified in C4-2B MDVR cells are presented in patients with NEPC, we performed GSEA pathway enrichment analyses on small-cell and NE prostate cancer cohorts from two multi-institutional prospective studies, mCRPC13, and treatment-emergent small-cell neuroendocrine prostate cancer (t-SCNC)5. The analyses were based on a total of 49 mCRPC cases, including 34 CRPC and 15 NEPC samples in Beltran’s study, and a total of 119 samples including 15 pure small cells, 6 mixed small cells, and 98 adenocarcinoma samples from Aggarwal’s study. We found that the eight neural lineage pathways enriched in C4-2B MDVR cells were also enriched in these NEPC patient datasets, including Neural Stem Cell Differentiation Pathways and Lineage-Specific Markers, Neural Precursor Cell Proliferation, Central Nervous System Neuron Differentiation, and Neuron Projection Development from Gene Ontology (Fig. 2a, b and Supplementary Fig. 1). The results from patient datasets suggest that the neural lineage pathways identified in C4-2B MDVR cells are also enriched in NEPC patients.
Having demonstrated that neural lineage programs are enriched in NEPC, we next analyzed the signature panel of neural lineage genes that are upregulated in the neural lineage programs. Based on the enriched neural lineage pathways aforementioned, we performed the Wilcox rank-sum test analyses and found that 239 genes were differentially upregulated in t-SCNC groups in the Aggarwal study and 165 genes in CRPC-NE cohorts in the Beltran study (p < 0.05). In parallel, we found 1060 genes upregulated in C4-2B MDVR cells (|log 2-fold change|>1, FPKM value >1). Venn diagram (Fig. 2c, displayed the commonly upregulated genes from the two patient datasets and C4-2B MDVR cells, and a collective 95 unique genes were identified as the gene panel of NLS (Fig. 2c and Supplementary Data 1). The neural lineage 95-gene panel was comprised of commonly upregulated genes in both Aggarwal and Beltran study (57 genes) and shared 7 genes and upregulated genes in C4-2B MDVR overlapped with either of the patient cohorts (13 and 18 genes for overlapped in MDVR vs. Aggarwal and MDVR vs. Beltran). Unsupervised hierarchical cluster analysis of 95 differentially expressed NLS genes (95 NLS genes) in Beltran CRPC-NE and Aggarwal t-SCNC patients were presented in Fig. 3a, b, which markedly clustered in the small-cell neuroendocrine groups. A correlation plot revealed that 95 neural lineage genes were positively correlated with these 29 NE markers and inversely correlated with AR and classical AR-targeted genes in both Beltran CRPC-NE and Aggarwal t-SCNC patients (Supplementary Fig. 2a, b). GSEA analysis also showed a significant enrichment of 95-gene NLSs in NEPC groups of these two patient datasets (Fig. 3c, d), which is consistent with the enrichment of 29 genes from the defined NEPC classifier13. Collectively, our analyses suggested that the 95 NLS genes (95 NLS) provide a molecular background giving rise to neuroendocrine differentiation in enzalutamide-resistant prostate cancer.
The 95 neural lineage gene signatures stratified NEPC from CRPC
We next determined if the 95 neural lineage gene signatures (NLS) could be used to stratify NEPC from CRPC in two advanced CRPC databases19,25. A hierarchical clustering heatmap demonstrated that NLS genes significantly clustered in the group of small cell or adenocarcinoma with NE features in the Abida-Wassim cohort (Fig. 4a, b), which align with upregulation patterns of the aforementioned Beltran NEPC classifier genes and downregulation of AR related genes (Supplementary Fig. 3a). We also analyzed the NLS in the Labrecque study including refractory metastatic CRPC specimens (AR-positive tumors (ARPC, n = 59), AR-low tumors (ARLPC, n = 9), amphicrine tumors expressing both AR and NE markers (AMPC, n = 11), double-negative tumors (DNPC, n = 7), and tumors with small cell or NE makers without AR features (SCNPC, n = 10))19. The heatmap of NLS genes demonstrated a clear distinction between the subtypes of mCRPC specimens with GSEA enrichment significance (NES = 2.44, FDR q value <0.01) (Fig. 4c, d), which was also consistent with the expression pattern of classic NED markers and AR target genes (Supplementary Fig. 3b). We also examined our NLS genes in an experimental model, TLT331R NEPC tumor established after castration and relapsed after 24–32 weeks26. As shown in Supplementary Fig. 4, top upregulated NLS genes were displayed in the unsupervised hierarchical clustering heatmap, which aligned well with NE markers and negatively correlated AR target genes. GSEA enrichment analyses also showed that NLS genes were significantly enriched in TLT331R NEPC tumor groups compared with prostate adenocarcinoma groups. In summary, these data indicated that these differentially expressed NLS genes could stratify prostate cancer with neuroendocrine differentiation from prostate adenocarcinoma.
Higher levels of ARHGEF2, EPHB2, and LHX2 expression correlate with poor survival in castration-resistant prostate cancer
Among the genes from the 95 NLS, we further analyzed 7 of the neural lineage genes (ARHGEF2, EPHB2, LXH2, DPYSL3, EPHB2, FYN, and GNG4) shared among C4-2B MDVR cells, Beltran, Aggarwal, Abida-Wassim and Labrecque datasets5,13,19,25. Our data showed that all the seven genes were upregulated in CRPC-NE/small-cell groups compared to the CRPC-adeno group across the four databases (Fig. 5a). To determine whether the seven NLS genes were associated with survival in prostate cancer patients, we further conducted the Kaplan–Meier survival analysis and log-rank test to determine the correlation of expression of the neural lineage genes to the overall survival in prostate cancer patients. In 75 out of 266 prostate cancer patients with overall survival of the first-line AR-targeted inhibitors treatment in the Abida-Wassim study25, higher expression of ARHGEF2 (p = 0.041), LHX2 (p = 0.0091), and EPHB2 (p = 0.15) showed correlation with shortened overall survival time (Fig. 5b), while DPYSL3, EPHB2, FYN, and GNG4 did not positively correlate with shortened overall survival. We also performed the Kaplan–Meier survival analysis on 148 patients with both disease-free survival information and RNA sequencing data available from the MSKCC study, which included 131 primary tumors and 9 metastases27. The data revealed that higher expression of ARHGEF2 (p = 0.011), LHX2 (p = 0.0091), and EPHB2 (p = 0.0019) correlate with shorter disease-free time, respectively (Fig. 5c), while DPYSL3, EPHB2, FYN, and GNG4 did not reach statistical significance (data not shown). Collectively, these results suggest the potential of ARHGEF2, LHX2, and EPHB2 as indicators of poor survival for advanced prostate cancer.
Downregulation of ARHGF2 expression suppresses viability and neuroendocrine markers of C4-2B MDVR and H660 cells
We analyzed ARHGEF2, LHX2, and EPHB2 gene expression in the RNA sequencing data from GSE154576 (DeLucia et al., 2021). As shown in Fig. 6a, gene expression of ARHGEF2, LHX2, and EPHB2 were significantly upregulated in NEPC MSKCC-EF1 and H660 compared to 22RV1 and LNCaP95 cell lines. Quantification of ARHGEF2, LHX2, and EPHB2 mRNA levels verified that these three genes increased in enzalutamide-resistant C4-2B MDVR compared with C4-2B parental cells, and further increased in H660 cells (Fig. 6b). We next focused on the effect of ARHGEF2 on cell growth and neuroendocrine differentiation by knocking down ARHGEF2 expression using siRNA in C4-2B MDVR and H660 cells. Knocking down of ARHGEF2 expression downregulates CHGA, NSE, and SYP protein expression (Fig. 6c) and inhibits cell viability (Fig. 6d) in both C4-2B MDVR and H660 cells. Furthermore, we analyzed LuCaP49 neuroendocrine PDX tumors and LuCaP35CR castration-resistant PDX tumors for their expression of AR-targeted genes and NED markers and found that LuCaP49 tumors express higher levels of NE markers such as CHGA, NSE, and SYP than LuCaP35CR tumors (Fig. 6e), consistent with the characteristics of NED for LuCaP49 and CRPC for LuCaP35CR28. We also found that ARHGEF2 mRNA levels were much higher in LuCaP49 tumors than LuCaP35CR tumors (Fig. 6e). Knocking down ARHGEF2 expression through siRNA significantly inhibits the viability and growth of organoids derived from LuCaP49 PDX tumors (Fig. 6f). Collectively, these data suggest that ARHGEF2 could serve as a potential therapeutic target for NEPC.
Greater PSA decline from baseline in men with metastatic castration-resistant prostate cancer (mCRPC) treated with the radioligand therapy 177lutetium-PSMA-617 (177Lu-PSMA-617) predicts better outcomes, investigators reported at the European Society for Medical Oncology’s 2022 Congress (ESMO 2022) in Paris, France.
The finding is from a post hoc analysis of the phase 3 VISION trial, which previously demonstrated that men with PSMA-positive mCRPC treated with 177Lu-PSMA-617 had prolonged radiographic progression-free survival (rPFS) and overall survival (OS).
For the post hoc analysis, Andrew J. Armstrong, MD, of Duke Cancer Institute Center for Prostate and Urologic Cancers at Duke University in Durham, North Carolina, and colleagues classified patients into 4 subgroups by magnitude of confirmed best PSA decline from baseline: no decline; 50% or less; more than 50% and up to 90%; and more than 90%.
Compared with patients who had no PSA decline, those with declines of 50% or less, greater than 50% and up to 90%, and more than 90% from baseline had a significant 60%, 80%, and 96% reduced risk for radiographic disease progression, respectively, and 42%, 58%, and 90% lower risk for death, respectively.
The median rPFS and OS were 3.0 and 8.4 months, respectively, for the patients who did not have a PSA decline. By comparison, the patients who had a PSA decline of 50% or less, more than 50% and up to 90%, and more than 90% had a median rPFS of 6.0, 8.8, and 19.7 months, respectively. Patients who had a PSA decline of 50% or less and more than 50% and up to 90% had a median OS of 12.0 and 15.0 months, respectively. The median OS was not estimable for those with a PSA decline greater than 90%.
In addition, study findings show that PSA declines delayed worsening of health-related quality of life, according to the investigators.
“These findings suggest that PSA decline is of prognostic importance for clinical outcomes during radioligand therapy with 177Lu-PSMA-617 in patients with PSMA-positive mCRPC,” Dr Armstrong’s team reported.
Disclosure: This research was supported by Advanced Accelerator Applications, a Novartis Company. Please see the original reference for a full list of disclosures.
ZIC5 is overexpressed in human PCa specimens and cell lines
To explore the potential impact of ZIC5 on PCa, we first extracted gene expression profiles from TCGA using the UCSC Xena platform. As shown in Fig. 1A, ZIC5 expression was markedly elevated in PCa tissues compared with normal ones. To investigate the prognostic value of ZIC5 in PCa, we performed survival analysis and log-rank tests on the above TCGA-PCa dataset. Results showed that higher expression of ZIC5 correlated with worse overall survival in PCa patients (Fig. 1B). A previous study reported that ZIC5 overexpression promotes melanoma aggressiveness and metastatic spread . However, whether ZIC5 overexpression contributes to tumor metastasis in PCa remains unclear. Thus, PCa GEO datasets were included in our analysis. Based on data from GSE6919, we found that the expression levels of ZIC5 were notably higher in metastatic PCa than in localized carcinomas (Fig. 1C). In addition, analysis of the GSE3325 dataset also revealed the same trend (Supplementary Fig. S1C). To further validate these findings, four sets of clinical samples were collected, including benign prostatic hyperplasia (BPH) tissue samples, localized PCa and adjacent non-tumor samples, and metastatic PCa tumor samples. Immunohistochemical staining showed that the expression of ZIC5 was predominately located in the nucleus, and the number of ZIC5-positive cells increased along with disease aggressiveness. In particular, the most intense ZIC5 staining was found in metastatic tumor tissues (Fig. 1D). Furthermore, reinforcing a potential contribution of ZIC5 to PCa metastasis, RT-qPCR and western blot data indicated that ZIC5 levels were obviously higher in metastatic lesions than in localized tumors (Fig. 1E, F).
Next, we assessed ZIC5 expression levels in five PCa cell lines (PC3, DU145, C4-2B, LNCAP, and 22RV1) and in normal human prostate epithelial RWPE1 cells. Similar to results from human PCa tissues, ZIC5 expression was significantly upregulated in the PCa cell lines compared to RWPE1 cells (Fig. 1G). These results indicated that overexpression of ZIC5 correlates with poor prognosis in PCa patients and is probably involved in the progression and metastasis of PCa.
ZIC5 promotes EMT progression in PCa cell lines
We further explored the biological function of ZIC5 in PCa cells. Because C4-2B and 22RV1 cells exhibited higher ZIC5 expression levels than the other three PCa cell lines examined (Fig. 1G), those two cell lines were selected for subsequent analyses. We then applied RNA silencing to suppress ZIC5 expression and used a lentiviral plasmid to overexpress ZIC5 in both C4-2B and 22RV1 cells. High transfection efficiency was confirmed by RT-qPCR and western blotting assays (Supplementary Fig. 1A, B). Then, wound healing, Transwell-Matrigel, and colony formation assays were performed to measure the migration and invasion abilities and colony formation capacities of C4-2B and 22RV1 cells. Knockdown of ZIC5 expression markedly attenuated migration and invasion and colony formation potential, while forced ZIC5 expression conferred stronger migratory, invasive, and colony formation abilities in both cell lines (Supplementary Fig. 2A, B and Supplementary Fig. 1D, E). In addition, we also evaluated the effect of ZIC5 on the metastasis in PC3 cells, an AR-negative PCa cell line that is widely used in prostate cancer research. The results showed that ZIC5 inhibition could barely affect PC3 cell invasion and migration, whereas restoration of ZIC5 slightly induced metastasis of PC3 cells (Supplementary Fig. 1F, G), which might be due to the moderate expression of ZIC5 and AR in PC3 cells.
Given that EMT is a major step in the process of cancer cell metastasis , and ZIC5 was reported to modulate the expression of EMT genes . we investigated whether ZIC5 promotes EMT in C4-2B and 22RV1 cells. Analysis of the association between ZIC5 and EMT-related markers in TCGA-PCa patient data using the ENCORI platform revealed that ZIC5 expression correlated positively with TWIST1 and CDH2 (N-cadherin) expression in PCa specimens (Fig. 2A). Furthermore, after ZIC5 silencing, both RT-qPCR and western blotting showed significantly increased levels of E-cadherin, a protein responsible for epithelial adherens junction formation, and a remarkable decline in the levels of mesenchymal-associated proteins, namely N-cadherin, TWIST1, and Snail1. In contrast, exogenous expression of ZIC5 upregulated the expression of EMT markers in both C4-2B and 22RV1 cells (Fig. 2B, C).
Next, we explored the potential mechanisms of how ZIC5 is involved in the progression of EMT in C4-2B and 22RV1 cells. Based on the above findings, we conducted studies to verify transcriptional activation of TWIST1, a critical activator of the EMT process , by ZIC5. Bioinformatics prediction was carried out and identified four potential ZIC5-binding sites on the promoter region of TWIST1 (Fig. 2D). We then applied luciferase reporter assays to determine whether ZIC5 directly binds to the TWIST1 gene promoter. Results revealed that ZIC5 silencing reduced, whereas its overexpression drastically increased, the activity of the wild-type TWIST1 promoter in both C4-2B and 22RV1 cells. In contrast, neither silencing nor upregulation of ZIC5 altered the activity of a mutant TWIST1 promoter construct (Fig. 2E, F). Subsequently, we verified the interaction between ZIC5 and the TWIST1 promoter through ChIP assays. Four potential binding sites, namely B1 (−427 to −442 bp), B2 (−516 to −531 bp), B3 (−655 to −670 bp) and B4 (−842 to −857 bp), were included in our study. A strong enhancement in the recruitment of ZIC5 was found only on the B3 region of the TWIST1 promoter, indicating that ZIC5 binds to a region located 655 to 670 bp upstream of the transcription start site (TSS) (Fig. 2G, H). Next, to determine whether TWIST1 expression mediates ZIC5-induced motility and metastasis of PCa cells, siRNA-mediated TWIST1 silencing was induced in C4-2B and 22RV1 cells. Wound healing and Transwell-Matrigel assays showed that depletion of TWIST1 significantly restricted ZIC5-induced migration and invasion of PCa cells (Supplementary Fig. 2C, D). Collectively, our data proved that ZIC5 promotes EMT via enhancing TWIST1 transcription, thus facilitating metastasis of PCa cells.
ZIC5 regulates Wnt/β-catenin signaling in vitro
Aberrant activation of the Wnt/β-catenin pathway is closely associated with cancer progression and metastasis . GEPIA analyses showed a potential link between ZIC5 and Wnt/β-catenin signaling genes, namely CTNNB1 (β-catenin) and GSK3B (GSK-3β), in the TCGA-PCa dataset (Supplementary Fig. 3A). We therefore surmised that ZIC5 might regulate the Wnt/β-catenin pathway to support PCa metastasis. Indeed, compared to control cells, a markedly increased expression of Wnt/β-catenin downstream genes, including c-Myc, MMP2, and MMP7, was noted in ZIC5-overexpressing PCa cells (Fig. 3A, B). In contrast, ZIC5 silencing was associated with significant repression of the above genes (Fig. 3A, B). Of note, the former effect could be blunted by application of LiCl, which enhances β-catenin activity by inhibiting GSK-3β (Fig. 3C, D). In addition, stimulation of Wnt/β-catenin signaling via LiCl markedly abrogated the inhibitory effect of ZIC5 knockdown on migration and invasion of C4-2B and 22RV1 cells (Supplementary Fig. 3B, C). These data suggest that ZIC5 promotes PCa cell metastasis through Wnt/β-catenin pathway activation.
To assess the above hypothesis, the effect of ZIC5 silencing and overexpression on Wnt/β-catenin activation was examined using a TCF/LEF luciferase reporter assay. Supporting our assumptions, ZIC5 knockdown drastically reduced, while ZIC5 overexpression significantly augmented, the luciferase activity of the TCF/LEF-responsive reporter in PCa cells (Fig. 3E). Since the translocation of β-catenin into the nucleus is a critical step in the transduction of WNT signals , we then assessed the relationship between ZIC5 and β-catenin expression. Unexpectedly, neither silencing nor overexpression of ZIC5 had an obvious effect on the expression levels of β-catenin, either in whole cells or in cell nuclei lysates (Fig. 3F, G). Consistent with these findings, immunofluorescence assays showed that ZIC5 overexpression barely altered the nuclear localization of β-catenin (Fig. 3H). These data indicated that ZIC5 induces Wnt/β-catenin signaling without affecting the nuclear translocation of β-catenin.
Subsequently, we addressed the mechanism by which ZIC5 regulates the transduction of Wnt signaling. Nuclear β-catenin binds to transcription factor 4 (TCF4) to form a β-catenin/TCF4 complex, which then activates the transcription of specific target genes . To assess whether ZIC5 influences β-catenin/TCF4 complex formation, Co-IP assays were performed in C4-2B and 22RV1 cells. Results confirmed that ZIC5 co-immunoprecipitated with both β-catenin and TCF4 (Fig. 3I). Moreover, exogenous expression of β-catenin and TCF4 in 293T cells could be interact with ZIC5, respectively. (Supplementary Fig. 3D). To detect whether the β-catenin/TCF4 complex could be affected by ZIC5. Our results of Co-IP showed that ZIC5 strengthened β-catenin-TCF4 association in 293T (Fig. 3J, K). Collectively, these findings strongly suggest that ZIC5 promotes PCa metastasis by activating Wnt/β-catenin signaling via potentiating β-catenin/TCF4 complex formation.
AR enhances ZIC5 expression through miR-27b-3p downregulation
To explore the mechanism responsible for ZIC5 upregulation in PCa, various pathways potentially involved were investigated. Given the key role of AR in PCa progression and its positive correlation with ZIC5 (Supplementary Fig. 4A). Further suggesting a possible link between AR and ZIC5 expression in PCa, we noticed that ZIC5 levels were higher in AR-positive than in AR-negative PCa cells (Fig. 1G). Next, androgen-sensitive LNCaP cells were cultured in charcoal-stripped serum medium for 3 days and then administered various doses of dihydrotestosterone (DHT) to stimulate AR signaling. Western blotting showed a strong upregulation of ZIC5 expression following stimulation with 1 nmol/L DHT (Fig. 4A). We then performed AR knockdown in C4-2B cells and induced DHT-mediated AR expression in 22RV1 cells to determine the influence of AR on ZIC5 expression. Consistent with the above findings, ZIC5 protein expression was reduced in C4-2B cells but was elevated instead in 22RV1 cells (Fig. 4B). Moreover, co-treatment with enzalutamide (Enz), a second-generation AR pathway antagonist, inhibited DHT-mediated ZIC5 expression in both C4-2B and 22RV1 cells (Fig. 4C).
Then, we focused on possible mechanisms underlying AR-dependent ZIC5 expression. Because ZIC5 mRNA levels were clearly altered by AR at 48 h, but not at 12 or 24 h compared to controls (Fig. 4D, E and Supplementary Fig. 4B). we speculated that AR modulates ZIC5 expression through a post-transcriptional mechanism. Considering the critical role of miRNAs in post-transcriptional regulation, an Ago2 antibody was used to pull down endogenous miRNA-ZIC5 complexes. Suggesting that AR-induced ZIC5 expression is indeed modulated by miRNA-ZIC5 interactions, assay results showed that ZIC5 mRNA levels in the Ago2 complex were reduced in DHT-treated 22RV1 cells but increased instead in AR-silenced C4-2B cells (Fig. 4F).
We next searched for potential ZIC5-binding miRNAs in multiple databases, including miRanda, PicTar, TargetScan, and PITA, accessed through the ENCORI platform. Search results consistently indicated that miR-27b-3p was a main predicted candidate. Based on this prediction, we interrogated TCGA data in the ENCORI platform and found that miR-27b-3p expression was downregulated in PCa, and its levels were inversely correlated with those of AR and ZIC5 (Supplementary Fig. 4D–F). Subsequently, we conducted RT-qPCR assays that showed that miR-27b-3p levels were elevated in AR-knockdown C4-2B cells and reduced instead in AR-stimulated 22RV1 cells (Supplementary Fig. 4C). Importantly, western blot analysis demonstrated that ZIC5 expression levels were notably reduced following transfection with miR-27b-3p mimics and increased, in turn, after miR-27b-3p inhibition (Fig. 4G).
Since miRNAs characteristically repress protein expression by binding to the 3′UTR of target mRNAs . we next assayed a luciferase reporter vector containing putative miR-27b-3p binding sites in the 3′UTR of ZIC5. As shown in Fig. 4H–J, deletion of miR-27b-3p in 22RV1 cells led to upregulation of wild-type ZIC5-3′UTR luciferase activity, while the activity of a mutant ZIC5-3′UTR luciferase reporter was not altered. Conversely, transfection of miR-27b-3p mimics significantly reduced luciferase activity in C4-2B cells transfected with the wild-type, but not with the mutant, ZIC5-3′UTR reporter. Furthermore, western blot assays revealed that the introduction of miR-27b-3p mimics markedly diminished AR activation-induced upregulation of ZIC5 in 22RV1 cells, whereas miR-27b-3p inhibition reversed AR-reduced ZIC5 expression in C4-2B cells (Fig. 4K). The above data indicate that AR activation inhibits miR-27b-3p expression, resulting in enhanced translation of ZIC5 mRNA in PCa.
AR association with SRC-3 modulates the transcription of miR-27b-3p
Since previous evidence implied that AR exerts transcriptional regulation of microRNAs in PCa [32, 33], we hypothesized that AR might bind to the promoter region of the miR-27b-3p gene to regulate its transcription. Bioinformatics analysis revealed five potential androgen-response-elements (AREs) on the promoter region of miR-27b-3p (Fig. 5A). Thus, those five AREs were selected for ChIP assay. We found obvious enrichment of AR in the ARE4 of the miR-27b-3p promoter (1847 to 1861 bp upstream of the TSS) but not on the other AREs (Fig. 5B). In addition, luciferase reporter assays in C4-2B and 22RV1 cells showed that siRNA-mediated AR inhibition drastically increased, whereas DHT-induced AR stimulation markedly inhibited, luciferase activity of the miR-27b-3p promoter. In contrast, neither inhibition nor stimulation of AR altered the activity of a mutant miR-27b-3p promoter in the above cell lines (Supplementary Fig. 5A). These results indicated that AR binds to the promoter of miR-27b-3p to repress its expression in PCa cells. The steroid receptor coactivator family (SRC-1, SRC-2, and SRC-3) has been well documented to interact with AR and regulate gene expression [34, 35]. However, whether this mechanism involves AR-mediated microRNA regulation remains uncertain. To verify the potential impact of SRCs on miR-27b-3p transcription in PCa cells, we first evaluated the SRCs expression in PCa via the GSE6919 and GSE3325 datasets. It was found that SRC-3 expression levels were obviously upregulated in metastatic PCa relative to localized carcinomas (Supplementary Fig. 5B). Similarly, analysis of TCGA-PCa dataset through GEPIA platform showed a positive correlation between AR and SRC-3 (NCOA3) (Supplementary Fig. 5C).
Next, ChIP assays using SRC-1, SRC-2, and SRC-3 antibodies were performed, and the results showed that only SRC-3 was notably enriched in the promoter of miR-27b-3p compared with IgG group (Fig. 5C). Furthermore, its occupancy could be strongly declined by AR knockdown (Fig. 5C). Likewise, SRC-3 depletion resulted in the erasure of AR binding on the miR-27b-3p promoter (Fig. 5D), indicating that binding of the coregulator stabilized the AR-complex on miR-27b-3p. In addition, neither AR nor SRC-3 showed obvious occupancy of the ZIC5 promoter, suggesting indirect modulation of ZIC5 by the AR (Supplementary Fig. 5D, E). To further ascertain the contribution of SRC-3 to AR-mediated miR-27b-3p transcriptional repression, SRCs were either inhibited, using bufalin (a pharmaceutical agent that selectively degrades SRC-1 and SRC-3)  or silenced, via specific siRNAs. Luciferase assay revealed that application of SRC-3 siRNA or bufalin, but not SRC-1 and SRC-2 depletion, was able to reduce AR activation-elicited repression of miR-27b-3p promoter activity in C4-2B and 22RV1 cells (Fig. 5E, Supplementary Fig. 5F). In parallel experiments, RT-qPCR confirmed that SRC-3 inhibition or depletion could increase the expression of miR-27b-3p. Moreover, the suppressive effect of AR activation on miR-27b-3p expression was relieved upon bufalin treatment or SRC-3 knockdown (Fig. 5F, Supplementary Fig. 5G). These findings demonstrated that SRC-3 associates with AR to prevent miR-27b-3p transcription in PCa cells.
Previous evidence revealed that AR or SRCs are able to recruit histone deacetylase families to exert gene repression functions [34, 37]. Moreover, our ChIP analysis with H3K9Ac (active histone mark) and H3K9Me2 (inactive histone mark) antibodies disclosed a strong occupancy of H3K9Ac at the miR-27b-3p promoter, while a significant decrease of H3K9Me2 at the miR-27b-3p promoter, after AR knockdown (Fig. 5G, Supplementary Fig. 5H). Thus, the impacts of pan-HDAC inhibitor vorinostat (VST) on miR-27b-3p expression were evaluated by RT- qPCR, and the results displayed that VST obviously rescued AR activation-mediated repression of miR-27b-3p, and this effect was further enhanced after combined treatment with bufalin (Fig. 5H). Our data revealed that AR/SRC-3 complex dependent transcriptional modulation may be achieved through the recruitment of HDACs to the miR-27b-3p promoter.
Subsequently, we assessed whether AR-mediated miR-27b-3p modulate ZIC5 levels to influence metastasis potential in PCa. We found that AR silencing repressed cell migration and invasion of C4-2B cells, and either application of miR-27b-3p inhibitors or ZIC5 overexpression reversed this effect (Supplementary Fig. 6A, B). Contrarily, AR stimulation increased the migration and invasion potential of 22RV1 cells, and this effect was abolished by miR-27b-3p mimics or ZIC5 inhibition (Supplementary Fig. 6C, D). Moreover, miR-27b-3p-elicited suppression of migration and invasion could be ameliorated by ZIC5 overexpression. Collectively, these findings showed that AR represses the transcription of miR-27b-3p to sustain ZIC5 expression, facilitating metastasis of PCa cells.
ZIC5 elevates AR expression and potentiates resistance to enzalutamide in PCa cells
AR modulates the expression of many androgen-response gene products , several of which may in turn influence AR expression and activation of AR signaling [39, 40]. In our study, we found that AR could augment ZIC5 levels in PCa cells. However, whether AR could be altered by ZIC5 is still uncertain. To test our hypothesis, three AR-positive cell lines (LNCAP, C4-2B and 22RV1) were used in our analysis. Genetic overexpression or inhibition of ZIC5 in LNCAP, C4-2B and 22RV1 cells caused a notably increase or decrease in the mRNA levels of AR target genes including PSA and TMPRSS2, respectively (Fig. 6A and Supplementary Fig. 7A, B). Nevertheless, ZIC5 had no significant effect on the mRNA levels of AR and AR-V7 (Fig. 6A and Supplementary Fig. 7A, B). Next, we performed western blot assays to determine whether ZIC5 levels influence the expression of AR and AR-splice variant 7 (AR-V7) protein in PCa cells. Suggesting a stimulatory effect of ZIC5 on AR expression and signaling in PCa cells, our results confirmed a decline in AR protein levels upon ZIC5 depletion, as well as downregulation of AR-V7 (Fig. 6B).
Compelling evidence indicates that sustained AR activity is one of the essential causes of PCa resistance to enzalutamide (Enz) . We thus posited that ZIC5-induced AR expression might contribute to Enz resistance in PCa. Cell proliferation assays on C4-2B and 22RV1 cells treated with various doses of Enz revealed that ZIC5 silencing compromised Enz resistance by reducing viability in 22RV1 cells, while ZIC5 overexpression alleviated Enz-mediated growth suppression in C4-2B cells (Fig. 6C–F). Importantly, we found that Enz application alone or in combination with ZIC5 depletion dramatically impaired the colony formation capacity of C4-2B cells, and this effect was reduced by AR overexpression (Supplementary Fig. 7C, D). Similarly, EdU assays showed that forced AR expression notably weakened the inhibitory effect of combined Enz treatment and ZIC5 knockdown on the proliferation of 22RV1 cells (Supplementary Fig. 8A, B). In turn, ZIC5-overexpressing C4-2B cells showed less apoptosis in response to Enz, an effect reversed by AR inhibition (Fig. 6G). Conversely, ZIC5 silencing increased apoptosis in 22RV1 cells treated with Enz, and this effect was diminished upon AR overexpression (Fig. 6H). These results cumulatively suggest that ZIC5 induces Enz resistance in PCa cells by enhancing AR expression.
ZIC5 inhibition increases the sensitivity of PCa to enzalutamide in mice
To recapitulate the findings of the cell experiments described above, the impact of ZIC5 expression was examined using PCa xenografts. 22RV1 cells (Enz-insensitive PCa cell line) transfected with ZIC5-targeted shRNA or control shRNA were injected subcutaneously into nude mice, divided into four groups to receive Enz or saline (control). As shown in Supplementary Fig. 9A, there was no significant difference in 22RV1 tumor size between control and Enz-treated mice. However, ZIC5 depletion led to a reduction in tumor growth, and this effect was enhanced by the combination of Enz treatment and ZIC5 knockdown. Parallelly, the tumor weight and tumor volume revealed a similar trend (Supplementary Fig. 9B, C). Consistent with these findings, IHC staining revealed lower Ki-67 expression in tumors from ZIC5-inhibited mice, and the combination of Enz treatment and ZIC5 knockdown strengthened this antiproliferative effect (Supplementary Fig. 9D, E). These data indicate that ZIC5 inhibition increases the efficacy of Enz against PCa growth in vivo. Finally, a schematic model depicting the proposed mechanism responsible for AR-ZIC5 axis-mediated metastasis and resistance to Enz in PCa is shown in Fig. 7.
Treatment with 177lutetium-PSMA-617 (177Lu-PSMA) is safe and effective in patients with metastatic castration-resistant prostate cancer (mCRPC) who have previously received 223radium (223Ra), according to study findings presented at the European Society for Medical Oncology’s 2022 Congress (ESMO 2022) in Paris, France.
In a study of 133 patients who participated in the RALU trial, Kambiz Rahbar, MD, of the University of Münster Medical Center in Münster, Germany, and colleagues found that the median overall survival from the start of 177Lu-PSMA was 13.2 months overall, 12.0 months for patients who received 223Ra followed by chemotherapy and then 177Lu-PSMA, and 14.0 months for those who received chemotherapy followed by 223Ra and then 177Lu-PSMA, respectively.
During treatment with 177Lu-PSMA, investigators observed a 50% or greater decline in PSA in 42% of patients overall, 46% of those who received 223Ra followed by chemotherapy and then 177Lu-PSMA, and 36% of those who received chemotherapy followed by 223Ra and then 177Lu-PSMA.
“In patients for whom 223Ra had been used as part of routine disease management, subsequent 177Lu-PSMA therapy was clinically feasible and well tolerated, with acceptable myelosuppression rates,” Dr Rahbar’s team concluded in a poster presentation.
The effectiveness of 177Lu-PSMA in patients previously treated with 223Ra was similar to other findings with 177Lu-PSMA, indicating no cross-resistance, according to Dr Rahbar and colleagues.
The study population had a median age of 73 years, and 56% had received 4 or more life-prolonging therapies, including abiraterone, enzalutamide, docetaxel, and cabazitaxel. All patients had bone metastases, and 27% had visceral metastases.
The latest findings echo those of a recently published real-world study led by Oliver Sartor, MD, of Tulane University School of Medicine in New Orleans, Louisiana. He and his collaborators examined the use of 177Lu-PSMA following 223Ra for bone-metastatic CRPC. The median time between the treatments was 8 months. The median overall survival was 28 months from the start of 223Ra and 13.2 months from the start of 177Lu-PSMA, the investigators reported in the Journal ofNuclear Medicine. The authors acknowledged that their small sample size precludes definitive conclusions, but noted that their data, especially related to the duration of 177Lu-PSMA, suggest that the use of this medication after 223Ra is feasible in a real-world setting.
Investigators have found an increasing probability of upstaging on prostate-specific membrane antigen positron emission tomography (PSMA PET) among patients with clinically localized high-risk prostate cancer.
Using 2010-2017 data from the National Cancer Database, investigators identified 45,772 men with high-risk disease who underwent radical prostatectomy (median age 64 years). The median PSA level was 8.8 ng/mL and percentage of positive cores was 50%. Gleason grade group 4 and 5 disease affected 46.5% and 30.0% of the cohort, respectively.
Investigators calculated the likelihood of upstaging on Gallium [Ga68] PSMA PET using a University of California, Los Angeles, nomogram.
The median risk of PSMA PET upstaging overall was 16.3%, significantly increasing from 13.0% in 2010 to 17.6% in 2017, Jonathan E. Shoag, MD, of Cleveland Medical Center, Case Western Reserve University School of Medicine in Ohio, and colleagues reported in JAMA Network Open. From 2010 to 2017, the median risk of nodal upstaging significantly increased from 11.7% to 15.4% and distant metastatic upstaging from 3.6% to 4.7%.
Investigators found an increasing proportion of Gleason grade group 4 and 5 cancers after surgery. PSA, T stage, and the percentage of positive cores did not change.
With respect to accuracy, the nomogram had an area under the receiver operative curve of 0.74.
“While upstaging may be secondary to nodal or distant metastatic findings on PSMA PET, the risk of nodal upstaging in particular may affect the decision between surgery and radiation,” Dr Shoag’s team stated. “Therefore, with increasing risk of nodal upstaging, the immediate impact of stage migration from PSMA PET may be to alter treatment decisions in a substantial number of patients with high-risk prostate cancer.”
Disclosure: Some study authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original reference for a full list of authors’ disclosures.
Heterogeneous response of prostate tumor cells to castration
We have previously demonstrated that ADT overall reduces the expression of non-homologous end joining repair proteins, such as Ku70 and P-DNA-PKcs, and thereby hampers the DNA repair capacity of PCa cells, which agrees with increased response to radiation therapy54. However, we noticed an extensive variability in the levels of these repair proteins in the biopsies, both within and between patients.
To further investigate this, we analyzed a set of formalin-fixed needle biopsies pre- and post-ADT from five patients with Gleason scores ranging from six to nine (GG 1–5) and analyzed repair protein expression patterns along with nuclear AR, using immunohistochemistry (IHC). Epithelial nuclei were distinguished by an in-house-developed software (Fig. 1a), and epithelial nuclear AR intensities were measured from biopsies pre- and post-ADT, showing the presence of nuclear AR also post-ADT in a fraction of the epithelial glands, in all five patients, which indicates treatment unresponsiveness (Fig. 1b, c).
We found evidence that high levels of nuclear AR predict high levels of said repair proteins and a correlation between levels of nuclear AR and serum PSA post-ADT (Supplementary Fig. 1; Supplementary Table 1). Non-responsiveness, was thus strongly linked to the expression of nuclear AR post-ADT. ADT resistance may also be the result of the activation of other genes or oncogenic pathways. To gain a broader molecular understanding of implicated genes and pathways contributing to ADT resistance in situ we continued with an exploratory transcriptome analysis.
Design of spatial gene expression experiments
To chart the molecular landscape of non-responsive cell clusters, fresh core needle biopsies pre- and post-ADT were collected from the prostates of three patients. The biopsies were snap frozen to facilitate gene expression analysis. We investigated the spatial gene expression profiles of each biopsy using ST50, a transcriptome-wide methodology capturing mRNA from tissue sections using spatially barcoded spots on microscopic slides, followed by a data driven analysis that identifies gene expression patterns across the tissue sections (overviewed in Supplementary Fig. 2).
In total eight core needle biopsies per patient were analyzed; four biopsies pre-ADT and four biopsies eight weeks after gonadotropin-releasing hormone (GnRH)-analogue treatment (post-ADT). Clinical data during this period were collected (Supplementary Table 2). Sections from each biopsy were hematoxylin and eosin (H&E) stained, scanned, independently annotated by two pathologists, and analyzed with ST. Adjacent tissue sections were immunostained for AR to enable a comparison to in situ gene expression patterns. Spatially variable high levels of nuclear AR were observed in the new set of fresh frozen biopsies, validating previous observations (exemplified in Fig. 2a).
The histological annotations of the tissue sections were conducted using the Gleason grading system. Information regarding immune response/inflammation and high-grade prostatic intraepithelial neoplasia (HGPIN) was also notated (Supplementary Fig. 3). In addition, the cells covering the spatially barcoded spots were further categorized into four tissue types: (i) stroma, (ii) 1–10% epithelium, (iii) 11–50% epithelium, (iv) 51–100% epithelium. An overview of the collected data, annotations, and performed analyses is illustrated in Supplementary Fig. 4.
Each patient was assigned to a representative clinical response group regarding ADT-treatment: responder (patient 1), moderate responder (patient 2), and non-responder (patient 3), based on the clinical data (Table 1, Supplementary Table 2), such as PSA nadir, PSA progress, and metastasis status.
Gene expression analysis in situ before and after ADT treatment
Gene expression analysis in tissue sections was achieved using barcoded arrays with spots on the surface, each with a known x- and y-coordinate. Each spot had a diameter of 100 micrometers, capturing the transcriptome from around 10–50 cells. The sample handling protocol for prostate tissue55 was adjusted to be compatible with the limited amount of tissue provided by core needle biopsies (Supplementary Figs. 5–6). On average, we detected approximately 4000 expressed genes per spot for all biopsies (Supplementary Fig. 6c).
First, to obtain an overview of patients and their biopsies, we bulked all spots per biopsy into individual pseudo-bulk samples and performed a principal component analysis (PCA). Most of the biopsies separated patient-wise (Fig. 2b) as has previously been observed in patients with PCa56. For patient 1 (responder), biopsies separated pre- and post-ADT. Patient 3 (non-responder) exhibited only a small separation, while patient 2 (moderate responder) showed more extensive spread between the biopsies.
To further assess the overall gene expression differences, we next conducted differential gene expression (DGE) analysis between histological areas comparing pre- and post-ADT samples. By merging gene expression data from epithelial and stromal spots, respectively, followed by DGE analysis, we could compare the temporal changes between the two histological entities. AR-regulated genes, such as KLK3, KLK2, and NKX3-157, were downregulated post-ADT in epithelial spots, for the three patients (Fig. 2c, Supplementary Fig. 7a–d). In line with clinical responsiveness, patient 1 had more differentially expressed AR-regulated genes compared to the other patients, indicative of successful and persistent downregulation of AR by ADT.
To identify biological pathways associated with differentially expressed genes (DEGs, q < 0.01), we performed functional enrichment analysis, querying against the KEGG database58 (Fig. 2d, Supplementary Fig. 7e–g). Among the three patients, pre-ADT, pathways related to AR (e.g., PPAR signaling, steroid hormone biosynthesis, sphingolipid signaling) and protein processing (e.g., protein processing in endoplasmic reticulum, lysosome, phagosome) were activated. Pathways related to cell migration (regulation of actin cytoskeleton, focal adhesion) were activated post-ADT for all patients. Subsequently, the same analysis procedure was performed for stroma demonstration and an upregulation of immune response genes post-ADT in all patients was observed (Supplementary Fig. 8).
Spatially resolved transcriptomes of patient biopsies
Next, we investigated the spatially resolved transcriptomes for the study cases. In total, we generated spatial and transcriptome-wide data for more than 4000 barcoded spots from 48 core needle biopsy sections including two consecutive tissue sections per biopsy. ST data from patients were analyzed individually by applying Spatial Transcriptome Decomposition (STD)52 (Supplementary data files 1–6; schematically overviewed in Supplementary Fig. 2). STD is a probabilistic model that factorizes the observed transcript data into latent gene expression factors (Methods). The factors characterize distinct metagenes, groups of genes that are likely to be co-expressed, and their spatial expression patterns.
The patient-specific factor analysis provided 13, 16, and 10 gene expression factors for patient 1, 2, and 3, respectively (Methods), and a control was made to ensure the independence between the factors (Supplementary Fig. 9). The ranked list of marker genes within each factor was used for molecular annotation. We broadly categorized factors into three entities; stroma, immune enriched stroma, and tumor activity (Supplementary Figs. 10–12). Factors representing normal epithelium were not identified. Overall, the molecular annotation of the factors overlapped with the histological annotations, but the factor analysis provided an increased resolution in the analysis. For example, multiple tumor factors were identified enabling the temporal subdivision into responding or non-responding factors. These factors were then compared with AR staining in adjacent sections.
Hierarchical clustering of the factors was performed for the patients (Supplementary Figs. 13–16) providing a summary of the factors, the corresponding histological entities and the molecular annotations for the tumor factors. The overview also depicts the annotation of temporal differences in AR staining.
The UMAP visualization of factor activities and its spatial position in the core needle biopsies is shown in Fig. 3. The two-dimensional embedding (UMAP) presents the distinct factors that correspond to histological features such as stroma, and tumor. In addition to the temporal differences observed for the tumor factors, we identified stroma factors that also displayed temporal differences.
The molecular heterogeneity as determined by the number of tumor factors and spread in UMAP space (left panels) appeared to be most prominent for patient 3. In patient 3, at least eight distinct tumor factors could be identified while patient 1 displayed five tumor factors. The UMAP results for patient 2 points to a broader representation, represented by eleven tumor factors, in line with the initial PCA data on pseudo-bulk data on tissue sections, which indicate a higher degree of tissue heterogeneity.
The spatial position of the tumor factors along the needle biopsies outlines the temporal differences after eight weeks of ADT, raising certain observations from our analysis; Patient 1, the treatment responder, has a reduction of tumor factors post-ADT, where they largely are replaced by stroma factors (right panel). Interestingly, even though this patient responded clinically, as determined by a persistent low PSA-value, we can still observe some tumor factors post-ADT. We cannot determine whether the post-ADT tumor cells that express these factors are in a transient mode of disappearing, still capable of having AR-activity, or represent pre-existing resistant cells.
In patient 2, the moderate responder with high initial and low PSA levels at week eight, we observe four responding and seven non-responding factors. However, we observe a major change in tissue makeup post-ADT, similar to Patient 1. Most of the tissue after treatment is composed of stroma and immune response genes.
Patient 3, the non-responder with moderate PSA levels before and after eight weeks, had no responding factor but eight non-responding factors. The spatial distribution and amount of the tumor factors were not changed for Patient 3, over the therapy period. Although patient 3 only displayed non-responding factors, we detected a down-regulation of AR-regulated genes when comparing epithelial spots pre- and post-ADT (Supplementary Fig. 7c). This suggests that spots designated to non-responding factors also contain a fraction of cells that do respond to ADT.
Overall, the ratio of responding and non-responding tumor factors after eight weeks of treatment agrees with the clinical outcome of ADT (Supplementary Fig. 13). In all three patients, the tissue area corresponding to identified tumor factors post-ADT, and thus, the number of cancer cells, decreased substantially due to the apoptotic effect of ADT59. Responding factors were overall present in larger groups of spatially confined spots pre-ADT than non-responding factors (pre-ADT) (Supplementary Figs. 10–12). This is also true for the non-responding tumor factors, which are generally present in larger areas pre-ADT compared to a more dispersed pattern post-ADT, except for patient 3 (Fig. 3, Supplementary Fig. 17).
Further, we observed in the PCA analysis (Fig. 2b) that biopsy 1 (pre-ADT) for patient 2, overlaps with non-responding factors in patient 3, which might indicate presence of a resistant phenotype in patient 2.
Neuroendocrine differentiation (NED) can occur in prostate cancer. Prostatic adenocarcinomas that have undergone NED are resistant to ADT. To investigate if there was an enrichment of cells that had undergone NED in the areas expressing resistant factors, we stained all biopsies in patient 2, before and after ADT, with the neuroendocrine marker chromogranin A (CgA). Areas expressing a given factor were mapped against the corresponding area in the CgA stained section, and the ratio of CgA positive cells for the areas was quantified. The criteria for the area depicted was that at least five spots with a given factor should be clustered together in both ST-replicates. No difference in ratio of CgA positive cells was found in areas expressing resistant factors compared to areas expressing non-resistant factors. (Supplementary Figs. 18–19, Supplementary Tables 3–4).
To interpret the spatial RNA data obtained from the ST method it is of importance to know how well it correlates with the corresponding protein levels in the tissue. In a large study, including expression data from 60 genes in several different tissues and cell-lines, the correlation between the number of RNA transcripts and number of protein molecules was good within each gene, but less accurate when comparing the number of transcripts with the number of protein molecules in-between different genes60. In accordance with this result, we showed in a previous article, that there is a good concordance between RNA expression detected with ST technology and protein levels detected with immunocytochemistry for all 7 proteins tested in prostate tissue applying similar ST protocol used herein55. Here we show that this relationship between RNA expression and protein levels also is valid for the androgen receptor (Supplementary Fig. 20).
Processes modified in non-responsive tumor factors
The factor annotation across the biopsy sections served as a tool to improve our investigation of the processes that underlie non-responsiveness. We, therefore, undertook a DGE analysis by first bulking the gene counts of the responding and non-responding factors, respectively, for patients 1 and 2. Here, spot selections were based on a stringent scheme: for example, spots with tumor factors above a defined factor intensity threshold were counted as non-responding spots if they consisted of histologically annotated tumor cells (see Methods for details, Supplementary Fig. 21a, b). Importantly, we only used information collected from spots pre-ADT to exclude ADT as a confounder in the DGE analysis.
The selection process resulted in an approximately equal number of responding and non-responding spots for patient 1 (71 and 90 spots, respectively, i.e., 56% non-responding spots from a total of 161 spots), while for patient 2 the fraction of non-responding spots was higher (64% non-responding spots from a total of 254 spots). The top differentially expressed genes (DEGs, q-value < 0.05 and logFC > 0.3 for patient 1 and > 0.5 for patient 2) between responder and non-responder spots were visualized as a gene expression heatmap (Fig. 4a, b). Only four DEGs were identified in patient 1, while patient 2 had more than 50 DEGs, potentially reflecting the clinical responsiveness of the patients.
To check that the DE-results of patient 1, which are immune related genes, are belonging to the non-responding factors and not simply is the result of non-factor-related cell types (since we are doing the analysis on all transcripts in the selected spots), we plotted all factors in patient 1 for their log2 fold changes for these specific genes (Supplementary Fig. 22). The result shows that the non-responsive factor 11 has a FC between 2 and 3 for all these four genes.
For patient 2, spots from factor 5, 9, 10 and 14 were excluded from the DGE analysis, because of not fulfilling the specified criteria, such as presence of factor activity in more than one biopsy (see Methods for details). The DEGs in patient 2 could be separated into two groups – one group of genes that are predominantly expressed in the non-responding spots and another group of genes with less distinct separation between responding and non-responding spots. The non-responding group of genes (including DHCR24, SNHG25, TRPM8, IFI6, FKBP2, PPDPF, HLA-DRA, TAGLN, TIMP1, AEBP1, IGFBP7, MGP, A2M, and CD74) has previously been described to correlate with e.g., androgen-independence, cell migration, resistance, PCa, cancer, tumor stroma, immune response, and inflammation61,62,63,64,65,66,67,68,69,70,71,72,73,74 (see Supplementary Tables 5–6 for complete list including downregulated genes). To spatially analyze the observations, the gene expressions of DHCR24 and SNHG25 were plotted onto the biopsy sections (Fig. 4c). Both genes are present over investigated time but more spatially prevalent before treatment onset.
To evaluate the set of predominantly non-responding genes, we took advantage of patient 3, the non-responder. Indeed, we confirmed that many of the marker genes (DHCR24, TRPM8, IFI6, H2AFJ) in non-responding areas in patient 2 were also expressed in the non-responder spots of patient 3 (Supplementary Fig. 23a). To ascertain that predicted areas with non-responding gene profiles are indeed non-responding we took advantage of the AR staining in the adjacent tissue sections, demonstrating nuclear staining for tumor cells in all areas post-ADT that had been annotated as non-responding. Some of these areas were crowded with eAR(+) nuclei and some had only sparse nuclei of this type. Non-responsive prostate glands and scattered cells, with nuclear AR, were found in areas where DHCR24 and SNHG25 were expressed, and a representative image of corresponding AR-stained tissue region shows a small AR-active prostate gland (Fig. 4d). Representative areas of AR-staining across the tissue sections post-ADT are shown in Supplementary Fig. 23b.
We continued by performing pathway analysis on all significant DEGs found in non-responding versus responding spots in patient 2 (Fig. 4e). Analysis revealed pathways associated with migration75,76,77, survival75, metastasis75,76,78,79 and androgen-independence75, proliferation80, and angiogenesis80 (focal adhesion, regulation of actin cytoskeleton, cell adhesion molecules (CAMs), proteoglycans in cancer, and platelet activation). Furthermore, pathways such as ECM-receptor interaction, MAPK signaling, RAS signaling, Rap1 signaling, PI3K-Akt signaling, and immune-related leukocyte transendothelial migration were more prominent in the non-responding spots.
The underlying non-responding genes give further insight into molecular processes. For example it has been demonstrated that primary tumor cells can stimulate platelets to get activated81 which leads to the release of a wide range of growth factors and cytokines, such as TGFβ182 which has been shown to induce metastasis83,84,85,86.
Platelet activation can also activate intracellular signaling cascades, such as p42 MAPK, which can stimulate proliferation, survival, adhesion and chemotaxis of hematopoietic cells81. Further, MAPK- and the PI3K-Akt pathways play a key role in apoptosis and bone metastasis87. MAPK-mediated phosphorylation of the nuclear receptor co-activator 1 (NCoA1, or SRC-1) may increase the coactivators affinity for AR, contributing to disease recurrence and CRPC88.
The identified RAS/MAPK-pathway in the non-responding regions has been shown to contribute to PCa progression and metastasis89, and is activated in 43% of primary tumors while in 90% of metastatic tissues90. RAS signaling has, in cell lines, shown to decrease androgen dependence and promote metastasis91. Several studies suggest that the PI3K-Akt pathway is involved in androgen-independent growth of PCa92,93,94,95,96,97, and genetic alterations of PI3K-Akt pathway occur in 100% of metastatic PCa which suggest a key role in the progression to CRPC12. The PI3K-Akt signaling pathway is also known to induce stem-like properties, proliferation, migration, angiogenesis, regulation of cellular growth and survival98, as well as having a potential correlation with PTEN-loss99.
The enrichment of ECM-receptor interaction in non-responding areas has previously been shown to play a key role in metastasis since it needs to allow for a CAM-mediated coordinated balance between adhesion and detachments of tumor cells100,101,102. Akt signaling is a master regulator when it comes to inducing EMT and cancer stem cell phenotype by the ECM, and this can be mediated by various focal adhesion proteins and lead to activation of e.g., NF-κB103,104. Focal adhesion formations transduce ECM signaling into the tumor cells and activate the PI3K-Akt pathway105.
Furthermore, the pinpointing of Rap1 signaling in non-responding areas is interesting as this has shown to induce cancer cell proliferation and disease progression in several cancer types106,107,108, and particularly in PCa its activation affects integrins important in migration, invasion, and bone metastasis109,110. Increased Rap1 activity correlated with high metastatic potential in both PCa cell lines and in vivo, implicating Rap1 could be of therapeutic importance for curing PCa110. The leukocyte transendothelial migration is an important step in the initiation of an inflammatory immune response and chronic inflammation, suggested to serve as an anti-cancer therapy111. Further, for invasive cervical cancer, KEGG pathway enrichment analysis has revealed activated pathways such as ‘focal adhesion’, ‘ECM-receptor interaction’ and ‘platelet activation’112.
We also observe that these resistant tumor areas have a lower cell cycle activity as compared to non-resistant areas before treatment onset, when comparing to a 71 cell cycle gene signature113 (Methods, Supplementary Fig. 24).
The tumor microenvironment in prostate cancer
Previous studies have shown that lack or low levels of nuclear AR in stromal cells (sAR(-)), adjacent to tumor cells, are observed in the context of high Gleason scores, and metastasis114,115. In contrast, normal stroma expresses nuclear AR. Across our tissue sections, we observed multiple regions of sAR(-) cells (Fig. 5a) that was compared to our factor analysis. Hereby, we could investigate the stroma of responding and non-responding tumor factors, respectively.
The proportions of stromal AR-staining in stromal nuclei is illustrated as a pie chart per factor for each of the patient in Supplementary Figs. 14–16. We hypothesized that a responding tumor factor would have a higher extent of surrounding stromal AR-positive (sAR(+)) cells, while non-responding tumor factors would be encircled by sAR(−) cells.
All stromal tissue spots were annotated by their AR-content using a defined threshold where tissue signals below the cutoff were treated as AR(−) stroma. In short, spots, by visual detection, composed solely of nuclei that lacked AR were counted as sAR(−), and spots with nuclei with at least 50% nuclear AR, were counted as sAR(+) (described in detail in Methods). Areas with a mix of sAR(−) and sAR(+) spots were annotated to sAR(mix) (Supplementary Fig. 21c).
For patient 1, 72% of the 287 epithelial-containing spots with attributed non-responding tumor factors were associated with sAR(−), while 15% were associated with sAR(+) (the remaining spots contained a mixture of AR positive and AR negative cells). 27% of the 79 epithelial-containing spots with attributed responding tumor factors were associated with sAR(−), while 60% were associated with sAR(+) (the remaining spots contained a mixture).
For patient 2, >99% of the 424 epithelial-containing spots with attributed non-responding tumor factors were associated with sAR(−). 53% of the 61 epithelial spots with attributed responding tumor factors were associated with sAR(−), and 47% contained a mixture.
For patient 3, 81% of the 399 epithelial-containing spots with attributed non-responding tumor factors were associated with sAR(−), and 9% were associated with sAR(+). No responding factors were found in patient 3.
We continued to investigate sAR(−) areas in patient 2 and 3 in more detail. We sought to compare gene expression levels between the stromal tissue regions sAR(+) and sAR(−) located next to tumor areas, independent of responsiveness of potential nearby tumor factors. DGE analysis was performed on the expression data between spots belonging to each tissue type (Fig. 5b, Supplementary Fig. 21d). The fraction of sAR(−) spots was 42% of a total of 172 spots. 26 DEGs were identified of which 21 was upregulated in sAR(-) spots (Supplementary Table 7), for example the epithelial- mesenchymal transition (EMT)-associated genes COL1A1, COL1A2, COL3A1, BGN, POSTN, SPARC, and AEBP1116,117,118. BGN is also associated with poor prognosis and PTEN deletion119 and upregulation during tumor angiogenesis120, and SPARC promotes bone metastasis in prostate cancer121. Both SPARC and POSTN are glycoproteins important for the structural network in the ECM. POSTN has been shown, using in vitro models, to be upregulated in advanced stages of cancer stroma and in bone metastases, however not in advanced stages of tumor cells122, in line with our observations (Fig. 5b, Fig. 4b). Further, SFRP4 is a marker for aggressive PCa and also a post-surgery recurrent marker123, and TIMP1 expression has been shown to be elevated in PCa stroma, to stimulate cancer associated fibroblasts, and to promote tumor progression124. The genes AEBP1 and TIMP1 are plotted onto the tissue section of biopsy 2 from patient 2 (pre-ADT) (Fig. 5c), and a comparison with histology reveals that these stromal compartments are located adjacent to PCa cells annotated as GG5.
Finally, in sAR(-) regions, we found an upregulation of several interesting pathways (Fig. 5d), of which the four most prominent ones were also upregulated in the non-responding tumor factors (ECM-receptor interaction, focal adhesion, PI3K-AKT, and Proteoglycans in cancer; Fig. 4b), together with TGF-β. The ECM-receptor interaction pathway has previously been shown to correlate with high reactive stroma content in PCa125. Also, changes in proteoglycans in the tumor microenvironment occur during tumor progression and affect e.g. cell signaling, chemokines, growth factors, and apoptosis126. The fact that the TGF-β pathway was upregulated in the sAR(-) regions is of particular interest since this can be induced by platelet activation which was activated in the non-responding tumor areas. TGF-β signaling promotes tumor initiation, progression, metastasis, EMT, stroma-tumor crosstalk, inflammation, immune-response, and angiogenesis81,82,83,84,85,86,127,128. Also, upon dissemination to the bones, tumor cells activate osteoclasts to degrade the bone matrix and release the stored TGF-β, which in turn leads to enhanced tumor cell malignancy129.
In summary, albeit sparseness of the material in the needle biopsies, we observed that a majority of the stroma in proximity to non-responding cancer cells pre-ADT lacks AR to a higher extent, in all patients. We noted that patient 1, who clinically had the best response to the treatment, displayed more sAR(+) surrounding the cancer, independent of responsiveness, while patient 2 and 3, who developed CRPC, displayed a higher frequency of sAR(−) areas in proximity to the non-responding epithelial spots. Future validation of these findings is important to reveal biomarkers and drug targets connected to stromal changes during the development of CRPC.
Adding apalutamide to androgen deprivation therapy (ADT) prolongs prostate-specific antigen (PSA) progression-free survival (PFS) in patients with biochemically recurrent prostate cancer, according to results from the phase 3 PRESTO trial.
The results also suggest that adding abiraterone acetate, prednisone, and apalutamide to ADT prolongs PSA PFS. However, it isn’t clear if this 4-drug combination provides an additional benefit over the 2-drug combination, as the trial was not powered to compare these 2 interventions.
These findings were presented at ESMO Congress 2022 by Rahul Aggarwal, MD, of the University of California, San Francisco Helen Diller Family Comprehensive Cancer Center.
The PRESTO study (ClinicalTrials.gov Identifier: NCT03009981) enrolled 503 patients who had undergone radical prostatectomy and had biochemical recurrence with PSA levels greater than 0.5 ng/mL and a doubling time of 9 months or less. Patients had no metastases on conventional imaging, and their last dose of ADT was more than 9 months prior to study entry.
At baseline, the patients had a median age of 66.7 years, and 83.7% were White. The median PSA level was 1.77 ng/mL, and 74.2% of patients had a PSA doubling time of 3-9 months.
The patients were randomly assigned to receive 1 of the following treatments for 52 weeks or until disease progression:
Luteinizing hormone-releasing hormone (LHRH) analog monotherapy (n=166)
LHRH analog plus apalutamide (n=168)
LHRH analog, apalutamide, abiraterone acetate, and prednisone (n=169).
Results showed that apalutamide improved PSA PFS. At a median follow-up of 21.5 months, the median PSA PFS was 20.3 months in the ADT monotherapy arm and 24.9 months in the apalutamide-ADT arm (hazard ratio [HR], 0.52; 95% CI, 0.35-0.77; P =.00047).
Likewise, adding apalutamide, abiraterone acetate, and prednisone to ADT prolonged PSA PFS. At a median follow-up of 21.3 months, the median PSA PFS was 20.0 months in the ADT monotherapy arm and 26.0 months in the 4-drug arm (HR, 0.48; 95% CI, 0.32-0.71; P =.00008).
A subgroup analysis showed a benefit with both experimental arms in patients with a PSA doubling time less than 3 months and in those with a PSA doubling time of 3-9 months.
In the testosterone-evaluable population, prolonged PSA PFS was observed in the apalutamide-ADT arm (HR, 0.53; 95% CI, 0.34-0.82; P =.00197) and the 4-drug arm (HR, 0.60; 95% CI, 0.39-0.92; P =.00851) compared with the ADT-alone arm. There were no significant differences between the arms in time to testosterone recovery (>50 ng/dL).
Adverse events (AEs) occurred in 90.6% of patients in the monotherapy arm, 90.8% in the 2-drug arm, and 96.3% in the 4-drug arm. Grade 3-4 AEs occurred in 18.8%, 25.2%, and 37.9%, respectively.
The most common grade 3 or higher AE was hypertension, which occurred in 8% of the monotherapy recipients, 7% of the dual therapy recipients, and 19% of patients who received the 4-drug combination.
Follow-up for this trial is ongoing, and the researchers aim to determine the impact of ADT plus androgen receptor pathway inhibition on patient-reported outcomes, time to castration resistance, and metastasis-free survival.
However, Dr Aggarwal noted that treatment decisions in biochemically recurrent prostate cancer are often predicated on PSA kinetics alone. Therefore, ADT plus apalutamide could be considered for high-risk patients with a short PSA doubling time.
Disclosures: This research was supported by Alliance Foundation Trials, LLC, and Janssen Research & Development, LLC. Some of the study authors declared affiliations with biotech, pharmaceutical, and/or device companies. Please see the original references for a full list of disclosures.