Sulfiredoxin as a Potential Therapeutic Target for Advanced and Metastatic Prostate Cancer

The incidence of prostate cancer (PCa) is increasing, and it is currently the second most frequent cause of death by cancer in men. Despite advancements in cancer therapies, new therapeutic approaches are still needed for treatment-refractory advanced metastatic PCa. Cross-species analysis presents a robust strategy for the discovery of new potential therapeutic targets. This strategy involves the integration of genomic data from genetically engineered mouse models (GEMMs) and human PCa datasets. Considering the role of antioxidant pathways in tumor initiation and progression, we searched oxidative stress-related genes for a potential therapeutic target for PCa. First, we analyzed RNA-sequencing data from Pb-Cre4; Ptenf/f mice and discovered an increase in sulfiredoxin (Srxn1) mRNA expression in high-grade prostatic intraepithelial neoplasia (PIN), well-differentiated adenocarcinoma (medium-stage tumors), and poor-differentiated adenocarcinoma (advanced-stage prostate tumors). The increase of SRXN1 protein expression was confirmed by immunohistochemistry in mouse prostate tumor paraffin samples. Analyses of human databases and prostate tissue microarrays demonstrated that SRXN1 is overexpressed in a subset of high-grade prostate tumors and correlates with aggressive PCa with worse prognosis and decreased survival. Analyses in vitro showed that SRXN1 expression is also higher in most PCa cell lines compared to normal cell lines. Furthermore, siRNA-mediated downregulation of SRXN1 led to decreased viability of PCa cells LNCaP. In conclusion, we identified the antioxidant enzyme SRXN1 as a potential therapeutic target for PCa. Our results suggest that the use of specific SRXN1 inhibitors may be an effective strategy for the adjuvant treatment of castration-resistant PCa with SRXN1 overexpression.


Introduction
The incidence of prostate cancer (PCa) has progressively increased in the western world, representing the second most prevalent cancer with the second highest mortality rate in men [1][2][3]. Androgen receptor (AR) and circulating androgen are essential for normal prostate development [4], and AR is the main oncogenic driver of PCa initiation and progression. Therefore, therapeutic strategies against this type of tumor are usually aimed at inhibiting AR activity [5,6]. If detected early, the chances of curing PCa are high, but more advanced PCa develops resistance to androgen deprivation therapies [7,8]. These tumors are referred to as "castration-resistant PCa," are highly heterogeneous in their molecular alterations [9][10][11], and are resistant to available therapies [12][13][14]. It is therefore crucial to identify new therapeutic targets and additional approaches to cure or at least increase the survival of patients with advanced PCa [15,16].
Emerging technologies have allowed a deeper understanding of the cancer genome and the differential expression of genes involved in tumor development [17][18][19]. One recent example of the success of modern precision medicine for the treatment of cancer is the development of PARP1 inhibitors (PARPi). This specific adjuvant treatment benefits patients with defective DNA-damage repair such as the BRCA1 and BRCA2 mutations that frequently occur in breast and ovarian cancers [20,21]. Studies involving PARPi for the treatment of PCa are already underway [22,23]. In this context, one strategy for the discovery of new cancer biomarkers and/or potential therapeutic targets is to integrate genomic data from genetically engineered mouse models (GEMMs) and human cancer patients [24,25]. Data generated by gene expression analysis of GEMM tumors help to search through genomewide expression datasets generated from human prostate tumors. GEMMs are also indispensable in preclinical studies to test new drugs in immunocompetent animals [26]. Thus, cross-species analyses provide a powerful tool to pinpoint genes conserved across both species that are master regulators of tumor development [27][28][29].
Cancer is a complex disease involving many molecular variables. For PCa, oxidative stress is one of the main ageassociated factors that influences the risk of developing this tumor [30,31], such as alterations in GSTP1 expression by hypermethylation [32]. Several studies suggest that the prostate is exceptionally vulnerable to elevated reactive oxygen species (ROS) and oxidative stress [33,34]. In normal cells, elevated ROS causes cumulative damage in lipids, proteins, and DNA, which may result in mutations and cancer initiation [35][36][37]. However, oxidative stress is also directly involved in cancer progression and metastasis [38][39][40]. Antioxidant pathways play an important cytoprotective role in tumors by preventing treatment-induced apoptosis and conferring chemoresistance [41][42][43][44][45]. Since these tumor cells are highly dependent on antioxidant mechanisms [38,40], we aimed to identify genes involved in oxidative stress homeostasis that can be therapeutically targeted for the treatment of PCa. Using cross-species analyses of GEMMs and human data, we have investigated antioxidant genes with altered expression in different stages of PCa progression. In this study, we identified the antioxidant enzyme sulfiredoxin (SRXN1) as a potential target, and we validated its relevance in advanced PCa.

Genetically Engineered Mouse Model (GEMM).
We used RNAseq data and prostate samples from GEMM Pten f/f , wild type (WT), and from the Pb-Cre4; Pten f/f , which contains a deletion in both alleles of Pten, a tumor suppressor gene, exclusively in the prostate epithelium (conditional knockout). This model shows stages of tumor progression similar to human PCa, such as prostatic intraepithelial neoplasia (PIN), microinvasive and invasive well-differentiated adenocarcinoma (medium-stage tumors, MT), and fully invasive poorly differentiated adenocarcinoma (advancedstage tumors, AT), with slow progression pace. Moreover, PTEN loss has been consistently associated with more aggressive disease features and worse prognosis, since PTEN loss range from less than 20% in clinically localized prostate tumors to more than 40% in metastatic castration-resistant PCa [46]. Although there are other interesting GEMMs for PCa, few studies have combined the stages of tumor progression with all prostatic lobes (anterior, AP; ventral, VP; lateral, LP; and dorsal prostate, DP) in a deep RNAsequencing experiment. Additional details about this conditional knockout mice, histopathological analysis, and transcriptome data have been previously described [29].
RNA-sequencing data from samples of all prostatic lobes were accessed through the NCBI Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) platform, reference number GSE94574. Briefly, 72 samples were submitted to RNAseq analysis, including 20 WT prostatic lobes, 16 PIN, 20 MT, and 16 AT. At least four samples for each prostatic lobe and pathological condition were submitted to RNAseq analysis. First, we looked for the differentially expressed genes in each lobe and stages of tumor progression with adjusted p value < 0.05. We filtered all the deregulated genes using a known list of 84 genes involved in oxidative stress response [47,48] and selected those which were altered in at least three prostatic lobes in PIN, MT, and/or AT. To select a potential therapeutic target, the consensus list of the upregulated genes involved in oxidative stress pathways was evaluated for clinical relevance in PCa using the cBioPortal for Cancer Genomics [49,50], Cambridge Carcinoma of the Prostate App (CamcAPP) [51], and SurvExpress [52] databases.
The animal experiments used in this study were approved by the CRUK Institute Ethics Committee of the Cambridge University, UK, under design license 80/2435 and by the Ethics Committee on Animal Experimentation from the Institute of Biosciences of Botucatu, UNESP, Brazil, under protocol CEEA 613/2014.

Immunohistochemistry (IHC).
Paraffin blocks of all prostatic lobes containing WT and tumor samples (PIN, MT, and AT) from GEMMs were obtained by donation from David Neal's Uro-Oncology Group at the CRUK Cambridge Institute (University of Cambridge, UK). Histological sections (5 μm) of WT and tumor-bearing prostate lobes from the GEMMs (n = 5 per group) were deparaffinized, rehydrated, boiled for 30 min in 10 mM sodium citrate solution (pH 6.0) for antigen retrieval, and quenched in 3% H 2 O 2 methanol solution. Prostate sections were blocked in 5% nonfat milk in phosphate-buffered saline (PBS) and incubated overnight at 4°C with a specific primary antibody against SRXN1 (Abcam, ab92298, 1 : 100), our chosen gene. Next, sections were incubated with a secondary peroxidaseconjugated antibody (Santa Cruz Biotechnology, 1 : 200), which was developed using diaminobenzidine (Sigma-Aldrich) as the chromogen. Slides were counterstained with Harris's hematoxylin. The negative control was obtained by excluding the primary antibody incubation step. The sections were visualized using a Leica DMLB 80 microscope. According to glandular histoarchitecture and histopathological staging, tumor samples from human TMAs were classified into five prognostic categories (1-5, from more differentiated to less differentiated), according to the Gleason score [53,54] and the International Society of Urological Pathology (ISUP) grade [55,56]. The prostate tumor classifications from TMAs are presented in Table 1.

Human Database
SRXN1 protein expression in human TMAs was detected by immunohistochemistry following the previously described protocol. The results were quantified and evaluated as negative (no staining, score 0) or positive (staining present, score 1), independent of the staining intensities. This analysis was performed by two independent observers without access to clinical data. SRXN1 staining scores were associated with the clinical and pathological characteristics (Gleason score, prognosis category, and survival time). Details regarding the clinical data and the SRXN1 IHC score are available in Supplementary Table S1, and the association between Gleason score/prognosis and survival rates in Supplementary Figure S1.

Cell
Lines and Culture Conditions. The following three prostate cell lines were used for the analyses: RWPE-1 (normal), LNCaP (tumor, androgen sensitive), and PC-3 (tumor, castration-resistant). The cells were obtained from the American Type Cell Culture (Manassas, Virginia, USA). LNCaP and PC-3 cell lines were cultured using RPMI 1640 medium supplemented with 2 mM L-glutamine, 10% fetal bovine serum, 50 μg/mL penicillin, 50 μg/mL streptomycin, and 0.5 μg/mL amphotericin B (GIBCO/Invitrogen). RWPE-1 cells were cultured with a Keratinocyte Serum Free Medium Kit supplemented with bovine pituitary extract and recombinant human epidermal growth factor (GIBCO/Invitrogen).
The medium was changed twice per week, and the cells were monitored daily under an inverted microscope (Zeiss Axiovert). Cells were grown at 37°C and 5% CO 2 . For passaging, cells were detached with 0.25% trypsin (GIBCO/Invitrogen) for 5 min at 37°C, resuspended in a growth medium, and reseeded. Cell lines were authenticated by Short Tandem Repeat (STR) DNA profiling by the Biorepository Facility of the Institute of Biosciences, UNESP, Brazil. Mycoplasma testing was carried out at regular intervals throughout the experiments, and the results were negative.
2.6. RNA Extraction and RT-qPCR. Total RNA from prostate cell lines RWPE-1, LNCaP, and PC-3 was extracted using the RNeasy mini kit (QIAGEN) according to the manufacturer's instructions. RNA quantification was determined by a Nanovue Spectrophotometer (GE Healthcare). cDNA was synthesized using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems). qRT-PCR reactions were performed using the QuantStudio 12K Flex Real-Time PCR system (Applied Biosystems). Relative gene expression was calculated using the 2 -ΔΔCT method [57]. The gene target detected was SRXN1, and β-actin (ACTB) was used as a housekeeping gene. Details of primers used are given in Table 2.

Sulfiredoxin Knockdown In Vitro.
To analyze the effects of SRNX1 suppression in vitro, we chose PCa cell line LNCaP as our model due to its elevated SRXN1 mRNA expression. LNCaP cells (1 × 10 5 ) were seeded in 6-well plates using a complete RPMI medium. The transfection was performed using the Lipofectamine RNAi MAX Transfection Reagent (Invitrogen) according to the manufacturer's instructions. Lipofectamine and siRNA targeting the mRNA of SRXN1 (MISSION esiRNA, Sigma-Aldrich) were individually diluted in an Opti-MEM medium (GIBCO/Invitrogen), mixed, and incubated for 5 min. Next, siRNA-lipid complex was added to cells and incubated for 48 h. The final siRNA concentration was 25 and 50 nM. In addition, siRNA 3 Oxidative Medicine and Cellular Longevity targeting eGFP (MISSION esiRNA, Sigma-Aldrich) was used as negative control.
2.8. Cell Viability Assays. LNCaP cells (6 × 10 4 ) were seeded in 24-well plates. Once cells became 60% confluent, siRNA against the mRNA of SRXN1 was added as previously described. After 48, 72, and 96 h of transfection, cell viability was determined by the MTT (Thiazolyl Blue Tetrazolium Bromide, Sigma-Aldrich) reduction method according to the manufacturer's instructions [58,59]. The reaction was transferred to a 96-well plate, and the absorbance (550 nm) was read by a spectrophotometer (ASYS HITECH GmbH, Eugendorf) to determine the percentage of cell viability relative to control cells.

Statistical Analysis. For parametric data, Student's t-test
with Welch's Correction Factor or Factorial ANOVA was used. For nonparametric data, the Mann-Whitney or Kruskal-Wallis test was used. For association analyses, the Chi Square Contingency test and Kaplan-Meier/Log-Rank test were used. Differences were considered statistically significant when p < 0:05. Statistical analyses were performed using the GraphPad Prism program (version 5.0).

Oxidative Stress Response Genes Are Deregulated during
Prostate Tumor Progression in Pb-Cre4; Pten f/f Mice. RNAseq data from GEMMs showed on average 2,000 genes differently expressed in each prostatic lobe (AP, VP, LP, and DP) at the tumor stages of PIN, MT, and AT. We filtered all these deregulated genes using a known list of 84 genes involved in oxidative stress response [47,48]. We selected those genes which were changed in at least three prostatic lobes in PIN, MT, and/or AT, resulting in a final list with the top 19 deregulated oxidative stress-related genes, out of which 11 upregulated and 8 downregulated (Figure 1). We investigated the prognosis significance of the upregulated and druggable genes. Excluding those genes with no clinical relevance, and those which have already been associated with PCa, such as Ctsb, Gpx2, Idh1, and Nos2 [60-66], we selected the antioxidant enzyme Srxn1 that had a strong correlation with patient outcome and has no previous related functional studies in PCa.

SRXN1 Expression Is Increased in Prostate Tumors from
Pb-Cre4; Pten f/f Mice. Prostate transcriptome data from Pb-Cre4; Pten f/f mice comparing WT to tumor samples (PIN, MT, and AT) showed that the relative mRNA expression of Srxn1 increases progressively and significantly during tumor progression in all prostatic lobes, with p < 0:001 (Figure 2(a)). To evaluate the pattern of SRXN1 protein expression in PCa, IHC analysis was performed, and we observed increased SRXN1 protein expression (higher intensity immunostaining) in prostate tumors compared to WT tissue (Figure 2(b)).

SRXN1 Is Overexpressed in High-Grade Human Prostate
Tumors and Is Associated with Cancer Aggressiveness. Published and available data (database GEO profiles) on global gene expression in human PCa demonstrated increased expression of SRNX1 in patients with advanced tumors (Gleason scores 8 and 9) relative to control (Supplementary Figure S2a). Similarly, studies available in the SurvExpress database revealed that increased expression of SRXN1 correlates with a high-risk/worse prognosis of patients with PCa (Figure 3(a)). The same association was observed using   and patients with high-risk/worse prognosis (red). Data and analyses were cataloged using the SurvExpress database [52] from a MSKCC study [18]. The difference between boxplots is statistically significant with p = 3:03 −32 . (b) Kaplan-Meier curve displaying disease-/progression-free survival of PCa patients with (red) or without (blue) SRXN1 alterations, cataloged using the cBioPortal human database [49,50] from a provisional study of TCGA (https://cancergenome.nih.gov/). Curves are significantly different with p = 2:104 −4 . 5 Oxidative Medicine and Cellular Longevity lower disease-free/progression-free survival (Figure 3(b)). The same effect was observed using the CamcAPP dataset (Supplementary Figure S2f).

SRXN1 Protein Expression Is Associated with Poor
Prognosis and Lower Survival of PCa Patients. To further validate the association of SRXN1 with PCa aggressiveness, we performed IHC analyses on SRXN1 in TMAs of human prostate samples (benign and cancerous tissue). Protein expression was increased (positive immunostaining) in more than 70% of high-grade tumors (Figure 4(a)), representing patients with high Gleason score and worse prognosis.
Additionally, the negative and positive immunostaining of SRNX1 in TMAs was associated with clinical data. It was observed that among the 48 patients with better outcome (prognosis categories 1 and 2), 39.5% expressed SRXN1. In comparison, among the 56 patients with worse prognosis (categories 3, 4, and 5), 71.4% expressed SRXN1, which differed significantly (p = 0:0015) (Figure 4(b)). Analysis of survival time (n = 69) revealed an association between SRXN1 protein expression and decreased survival (Figure 4(c)). Interestingly, the expression pattern of SRXN1 stratified patients as well as the prognosis score according to the ISUP classification, since patients with expression of SRXN1 had decreased survival similarly to patients with worse prognosis (Supplementary Figure S1).

PCa Cell Lines Express Higher Levels of SRXN1 Compared
to Normal Cell Lines, and Inhibition of SRXN1 mRNA in the PCa Cell Line LNCaP Decreases Viability. Quantitative PCR showed that prostate tumor cell lines LNCaP and PC-3 express higher levels of SRXN1 than the normal prostate cell line RWPE-1 (Figure 5(a)). Among the prostate tumor cell lines, LNCaP has increased SRXN1 gene expression   Oxidative Medicine and Cellular Longevity compared to PC-3 ( Figure 5(a)). Database analysis of other studies from GEO profiles also showed increased expression of SRXN1 in LNCaP and PC-3, and in most prostate tumor cells (Supplementary Figure S2b).
To analyze the effects of decreased SRXN1 expression in vitro, we performed siRNA-mediated knockdown of SRXN1 mRNA followed by cell viability assays at three different time points after transfection (48, 72, and 96 h). PCa cell 7 Oxidative Medicine and Cellular Longevity line LNCaP was chosen because of its high expression of SRXN1. After transfection, we observed a >80% reduction of the SRXN1 mRNA ( Figure 5(b)). The silencing analyses demonstrated that the viability of LNCaP cells with decreased SRXN1 mRNA expression was lower compared to cells that expressed SRXN1 (eGFP siRNA, negative control). The viability of PCa cells significantly decreased in all analyzed time points after SRXN1 knockdown (Figure 5(c)) (p < 0:05 in all cases).

Discussion
Cross-species analyses from GEMMs and human data revealed that SRXN1 is overexpressed in PCa tissue and cell lines, with a remarked increase in advanced tumors. Cancer cells are more dependent on antioxidant mechanisms and more vulnerable to ROS-induced damage than normal cells [38,40]. SRXN1 is an antioxidant enzyme induced by NRF2 and AP-1 and plays an important role in oxidative stress balance [67][68][69]. Specifically, SRXN1 acts in peroxiredoxin (I-IV) reactivation [70]. The peroxiredoxins are a group of peroxidase enzymes responsible for the reduction of ROS, such as hydrogen peroxide and organic peroxides [71], protecting the cell from high levels of ROS-induced oxidative stress. By reducing the hyperoxidized peroxiredoxins, SRXN1 protects them from degradation [72].
Several studies have demonstrated the protective role of SRXN1 against oxidative injury [73,74]. The increased expression of SRXN1 has been observed especially in solid tumors, as we observed for PCa in this study. Wei et al. showed elevated SRXN1 protein expression in lung tumor samples by IHC analyses [75]. Another study from Wei et al. demonstrated high SRXN1 protein expression in human colon carcinoma and showed that Srxn1 knockout animals were resistant to carcinogenic induction [76]. In addition, SRXN1 expression is increased in different human skin malignancies [77,78] and gastric cancer [79].
In addition to the overexpression of SRXN1 in PCa high-grade tumors, the current study also showed a close relationship between SRXN1 expression and both cancer aggressiveness and patient outcome. Researchers have already demonstrated that SRXN1 is essential for cancer cell proliferation [80], and its depletion decreases cell viability [74], suppresses cell migration, and inhibits tumor growth [75,77], increasing the aggressiveness of cancers with SRXN1 upregulation. Polymorphisms in the SRXN1 gene have been shown to promote breast cancer development and influence patient survival [81]. SRXN1 expression is also associated with poor survival of patients with pancreatic adenocarcinoma [82]. Raatikainen et al. observed an inverse correlation between SRXN1 expression and worse PCa prognosis [83], in contrast to our results and the results of other studies.
Considering the pathogenic role of SRXN1 in human cancer, it has already been suggested as a new potential therapeutic target for different tumors, but not yet for PCa. In our study, we observed that decreasing SRXN1 mRNA in PCa cell line LNCaP decreased cell viability, reinforcing the importance of this antioxidant enzyme in PCa cells and suggesting that SRXN1 activity supports tumor cell survival and growth.
In malignant human skin tumors, Wei et al. showed that SRXN1 inhibition (by the AP-1 pathway) may be a novel strategy for skin cancer prevention and treatment [77]. Kim et al. and Kim et al. demonstrated that SRXN1 inhibition by synthetic inhibitors (J14 and K27) selectively promotes the death of A549 pulmonary tumor cells [84,85]. All these studies suggest that when SRXN1 activity is inhibited, the high ROS levels exceed cell antioxidant capacity, resulting in accumulated damage that selectively leads to tumor cell death. Thus, our work proposes that SRXN1 can be an interesting therapeutic target for further preclinical in vivo tests using an immunocompetent and clinically relevant GEMM. Patients with advanced PCa presenting SRXN1 overexpression may benefit from SRXN1 inhibition therapy, providing cancer patients with more personalized treatment.

Conclusions
SRXN1 is increased in a subset of PCa patients with highgrade tumors (advanced stage) and correlates with poor prognosis and worse survival. Thus, our cross-species analyses pinpoint SRXN1 as a potential therapeutic target for PCa, which plays an important role in protection of prostate tumor cells against oxidative stress. We hypothesize that the use of selective SRXN1 inhibitors can be an effective adjuvant treatment strategy for metastatic PCa with SRXN1 overexpression.

Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper. Oxidative Medicine and Cellular Longevity providing formalin-fixed and paraffin-embedded prostate samples of the GEEMs and the TCGA Research Network for the results published here that are in part based upon data from their database: https://www.cancer.gov/tcga. We also thank the financial support received from the National Council for Scientific and Technological Development (CNPq), Brazil (grants #310805/2018-0 and #132849/2017-8) and São Paulo Research Foundation (FAPESP), Brazil (grants #2015/26097-6, #2016/09532-3, and #2016/25945-6). The present work was also carried out with the support of the "Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES)" -Finance Code 001. This article represents part of a master thesis developed by CNB at the Institute of Biology at Campinas State University (UNICAMP), Brazil. Table S1: description of the patient clinical data used in the preparation of the TMAs, such as Gleason score, prognostic category, survival time, and patient outcome. Figure S1: patients with PCa grouped by prognosis category (1)(2)(3)(4)(5) show expected survival curves. (A) Global survival curve of patients with PCa associated to prognosis category (1-5, from more differentiated to less differentiated), according to the ISUP grade. Clinical data are from those patients which prostate samples were used to construct TMAs. Kaplan-Meier curves are statistically different with p < 0:0001. (B) Additional analysis of the same patients of Figure S1a, dividing them into groups with good prognosis (categories 1 and 2, black curve) and worse prognosis (categories 3, 4, and 5, red curve), according to the ISUP grade. Kaplan-Meier curves are significantly different with p < 0:0001. Figure S2: SRXN1 expression is increased in advanced PCa and in most prostate tumor cell lines, and its overexpression is associated with poor prognosis and lower disease-/progression-free survival. (A) Levels of SRXN1 expression in wild-type prostate (control) and advanced PCa samples (Gleason scores 8 and 9) from a study available on the GEO profile human database (reference series GSE5016) [1]. (B) SRXN1 gene expression in different prostate cell lines (androgen sensitive and castration-resistant) obtained from a study available on the GEO profile human database (reference series GSE4016) [2]. (C) Expression of SRXN1 (median) in five PCa iClusters generated by the Cambridge Carcinoma of the Prostate App (cam-cAPP dataset) [3] from an integrative study [4]. iClusters 1 (red), 3 (green), and 5 (orange) represent groups of patients with worse prognosis, while iClusters 2 (blue) and 4 (purple) represent groups with better prognosis. Boxplots are significantly different, with p = 5:7833 −9 . (D) Expression of SRXN1 (median) in five PCa iClusters generated by the Cambridge Carcinoma of the Prostate App (camcAPP dataset) [3] from an integrative study [4]. iClusters 1 (red), 3 (green), and 5 (orange) represent groups of patients with worse prognosis, while iClusters 2 (blue) and 4 (purple) represent groups with better prognosis. Boxplots are significantly different with p = 0:034473. (E) Expression of SRXN1 (median) in six PCa iClusters generated by the Cambridge Carcinoma of the Prostate App (camcAPP dataset) [3] from an integrative study [5]. iClusters 1 (salmon), 2 (dark yellow), 3 (green), and 4 (turquoise) are groups of patients with more favorable prognosis with minimal copy number alterations (CNA), while iClusters 5 (light blue) and 6 (lilac) include most of the metastatic tumors with substantial CNA. Boxplots are significantly different, with p = 3:42 −6 . (F) Kaplan-Meier curve displaying the probability of freedom from biochemical recurrence of PCa with (red) or without (blue) SRNX1 overexpression, cataloged by the Cambridge Carcinoma of the Prostate App (camcAPP dataset) [3] from an integrative study [5]. Curves are statistically different with p = 0:0079. (Supplementary Materials)