Integrated Analysis Reveals Critical Genomic Regions in Prostate Tumor Microenvironment Associated with Clinicopathologic Phenotypes

Published OnlineFirst January 24, 2012; DOI: 10.1158/1078-0432.CCR-11-2535

Clinical Cancer Human Cancer Biology Research

Shingo Ashida1, Mohammed S. Orloff1,3, Gurkan Bebek1,5,7, Li Zhang1,2, Pan Zheng8,9, Donna M. Peehl10, and Charis Eng1,3,4,6,7

Abstract Purpose: Recent studies suggest that tumor microenvironment (stroma) is important in carcinogenesis and progression. We sought to integrate global genomic structural and expressional alterations in prostate cancer epithelium and stroma and their association with clinicopathologic features. Experimental Design: We conducted a genome-wide LOH/allelic imbalance (AI) scan of DNA from epithelium and stroma of 116 prostate cancers. LOH/AI hot or cold spots were defined as the markers with significantly higher or lower LOH/AI frequencies compared with the average frequency for markers along the same chromosome. These data were then integrated with publicly available transcriptome data sets and our experimentally derived data. Immunohistochemistry on an independent series was used for validation. Results: Overall, we identified 43 LOH/AI hot/cold spots, 17 in epithelium and stroma (P < 0.001), 18 only in epithelium (P < 0.001), and eight only in stroma (P < 0.001). Hierarchical clustering of expression data supervised by genes within LOH/AI hot/cold spots in both epithelium and stroma accurately separated samples into normal epithelium, primary cancer, and metastatic cancer groups, which could not be achieved with data from only epithelium. Importantly, our experimental expression data of the genes within the LOH/AI hot/cold spots in stroma accurately clustered normal stroma from cancer stroma. We also identified 15 LOH/AI markers that were associated with Gleason score, which were validated functionally in each compartment by transcriptome data. Independent immunohistochemical validation of STIM2 within a stromal significant LOH marker (identified as associated with Gleason grade) confirmed its downregulation in the transition from moderate to high Gleason grade. Conclusions: Compartment-specific genomic and transcriptomic alterations accurately distinguish clinical and pathologic outcomes, suggesting new biomarkers for prognosis and targeted therapeutics. Clin Cancer Res; 18(6); 1578–87. 2012 AACR.

Introduction tumor microenvironment as equally important in carcino- Previous studies have mainly focused on changes in the genesis and progression. Genetic changes in tumor stroma epithelium, but recent investigations have highlighted the have been reported by several independent groups working on a range of solid tumors (1–19), although typically, only selected markers have been used in most of these studies to Authors' Afﬁliations: 1Genomic Medicine Institute, 2Section of Statistical assess LOH/allelic imbalance (AI). Genetics and Bioinformatics, Department of Quantitative Health Sciences, Lerner Research Institute; 3Taussig Cancer Institute; 4Stanley Shalom In prostate cancer, it has also been suggested that tumor Zielony Institute for Nursing Excellence, Cleveland Clinic; 5Center for microenvironment plays an important role in the progres- 6 7 Proteomics and Bioinformatics, Department of Genetics, CASE Com- sion, acquisition of androgen independence, and distant prehensive Cancer Center, Case Western Reserve University School of Medicine, Cleveland; 8Department of Pathology, The Ohio State University, metastases (20–22). Tumor–microenvironment interac- Columbus, Ohio; 9Division of Immunotherapy, Department of Surgery, and tions through diffusible soluble factors such as TGF-b,as Program of Molecular Medicine, Comprehensive Cancer Center, University of Michigan Medical Center, Ann Arbor, Michigan; and 10Department of well as the extracellular matrix, likely lead to the develop- Urology, Stanford University School of Medicine, Stanford, California ment of metastasis (20). Nonetheless, little is known about Note: Supplementary data for this article are available at Clinical Cancer the genetic basis of the human prostate tumor microenvi- Research Online (http://clincancerres.aacrjournals.org/). ronment; one report described genotypic heterogeneity in Corresponding Author: Charis Eng, Cleveland Clinic Genomic Medicine mesenchymal cells with a small number of markers (9) and Institute, 9500 Euclid Avenue, Mailstop NE-50, Cleveland, OH 44195. another showed allelic losses in the stromal component Phone: 216-444-3440; Fax: 216-636-0655; E-mail: [email protected] with a small number of samples and markers (23). These doi: 10.1158/1078-0432.CCR-11-2535 ﬁndings suggest that genetic alterations in prostate cancer 2012 American Association for Cancer Research. stroma exist and may be biologically relevant.

1578 Clin Cancer Res; 18(6) March 15, 2012

Integrated Genome-Wide Analyses for Prostate Cancer Stroma

we used standard and well-established protocols to mini- Translational Relevance mize this, as reported previously (8, 18, 24, 25). Corre- Genome-wide LOH/allelic imbalance (AI) analysis of sponding normal DNA for each case was procured from DNA from epithelium and stroma of 116 prostate can- normal tissue, located a distance away from the tumor or cers revealed 43 LOH/AI hot/cold spots, 17 in both from a different tissue (formalin-fixed, paraffin-embedded) epithelium and stroma (P < 0.001), 18 only in epithe- block containing only normal tissue (latter being the first lium (P < 0.001), and eight only in stroma (P < 0.001). choice). Genomic DNA was extracted as previously Integration of experimentally derived transcriptome described, with the exception that incubation in proteinase data of genes within the significant LOH/AI hot/cold K was done at 65C for 2 days (7). Primer sets for multiplex spots in stroma revealed 15 LOH/AI markers that were PCR defined 381 microsatellite markers in 72 multiplex associated with Gleason score. Independent immuno- panels (Research Genetics, Invitrogen). Genotyping, anal- histochemical validation of one of these identified yses, and scoring of LOH/AI were done as reported STIM2 genes within a stromal significant LOH marker (24, 26, 27). The methodologic veracity of LOH/AI scan (identified as associated with Gleason grade) confirmed using multiplex PCR on archived templates was extensively its downregulation in the transition from moderate to validated as published, that is, by a series of control experi- high Gleason grade. Stromal genomic and transcrip- ments using chromogenic in situ hybridization (CISH), tomic alterations may be helpful in pinpointing aggres- quantitative PCR, to eliminate the possibility of epitheli- sive prostate cancers. al–stromal cell cross-contamination and to exclude any possibility of PCR artifact (24, 25).

Data analysis The data set contains LOH/AI status at 381 microsatellite To attain a global perspective of epithelial and stromal markers for 116 different samples, each with neoplastic specific LOH/AI changes for prostate cancer and further epithelium, tumor stroma, and normal tissue. The same 10 refine our understanding of the LOH/AI regions, we con- markers were never informative in at least one compartment ducted a genome-wide LOH/AI scan using 381microsatel- and so excluded from further analyses. First, we sought to lite markers and subsequently integrated these structural define regional LOH/AI hot/cold spots as the markers with data with publicly available and experimental expressional significantly higher or lower frequency of LOH/AI com- array analyses. We also sought to determine whether compared with the average frequency of the markers along the partment-specific LOH/AI could be correlated with clinico- same chromosome. We analyzed hot and cold spots of pathologic features. Finally, we conducted two types of LOH/AI together with the concept that the genetic endpoint validation: functional genomic validation and immunohis- may be similar, for example, loss of a tumor suppressor gene tochemistry in an independent series of prostate cancer. and gain (cold spot) of an oncogenic event. Toward those ends, for each informative marker, the statistical signifi- Materials and Methods cance of overall (across all samples) LOH/AI frequency Patients and samples compared with the respective chromosome average was We obtained 116 prostate cancer samples at any tumor analyzed using the exact binomial test (R base package classification with or without lymph node metastasis and binom.test; http://www.r-project.org). Second, the associa- analyzed them in accordance with the respective Institu- tions of LOH/AI across all informative markers (not just hot/ tion’s Human Subjects Protection Committees. All slides cold spot markers) in epithelial and stromal samples with were rereviewed by a single genitourinary pathologist (P. presenting clinicopathologic features (i.e., Gleason score, Zheng). Clinical information such as Gleason score, tumor tumor volume, and the status of lymph node metastasis) volume, and the status of lymph node metastasis was noted were analyzed using a logistic model with LOH frequency as (Supplementary Table S1). An independent validation an outcome and each clinicopathologic feature as a predic- series of prostate cancer is described under Immunohisto- tor (28, 29). Multiple testing adjustment was applied by chemical Analysis. controlling false-positive report probability (FPRP; ref. 30) with a prior probability of 0.05 or 0.01. The FPRP indicates Laser capture microdissection and genome-wide LOH/ the probability that a statistically significant finding is a false AI scan positive by considering 3 factors: the P value magnitude, the Laser capture microdissection of nonneoplastic normal statistical power, and the prior probability of true associa- tissue, epithelial carcinoma cells, and tumor-associated tions. Only those with P values <0.05 and estimated FPRP stroma was conducted from formalin-fixed, paraffin- values less than 50%, indicating a small probability of being embedded tissue for all 116 cases, as previously described a false positive, are reported as statistically significant find- (7, 8, 17, 24, 25) and showed at the link http://www.nejm. ings. An FPRP value with a prior probability of 0.05 and less org/doi/full/10.1056/NEJMoa071825. Stromal fibroblasts than 50% is denoted as FPRP0.05<0.5. We used more strict resided in between aggregations of epithelial tumor cells or criteria, FPRP0.01<0.1, for LOH/AI hot/cold spots expecting no more than 0.5 cm distant from tumor nodule. While higher significant regions. For the association of LOH/AI epithelial–stromal cell cross-contamination is a possibility, with clinicopathologic features, we applied FPRP0.05<0.5.

www.aacrjournals.org Clin Cancer Res; 18(6) March 15, 2012 1579

Ashida et al.

Table 1. LOH/AI hot/cold spots in prostate cancer

Markers Loci Epithelium Stroma Genes (within 250 kb of markers) miRNAs (within 2Mb)

In epithelium and stroma D1S1653 1q23.1 <0.001 <0.001 FCRL1, FCRL2, CD5L, KIRREL, CD1D mir-9-1 TPO 2p25.3 <0.001 <0.001 SNTG2, TPO, PXDN, ZBBX, SERPINI2, WDR49, PDCD10 D3S1763 3q26.1 <0.001 <0.001 SERPINI1, DGKG, CRYGS, TBCCD1, DNAJB11 D3S1262 3q27.3 <0.001 <0.001 AHSG, FETUB, HRG, KNG1, PRKAA1, RPL37, SNORD72, CARD6 D5S1457 5p13.1 <0.001 <0.001 C6, C7, HEATR7B2, ARSB, DMGDH, BHMT, BHMT2, JMY D5S1501 5q14.1 <0.001 <0.001 HOMER1, ARPC5L, GOLGA1, C9orf126, PPP6C D9S1825 9q33.3 <0.001 <0.001 RABEPK, HSPA5, GAPVD1 mir-181a, mir-181b-2 D12S269 12p13.1 <0.001 <0.001 EMP1, C12orf36, GRIN2B, CCDC60, LOC387890, PRKAB1, CIT D12S395 12q24.23 <0.001 <0.001 CCDC64 D12S2078 12q24.32 <0.001 <0.001 ESTs D13S285 13q34 <0.001 <0.001 SOX1, C13orf28 D14S588 14q24.1 <0.001 <0.001 KIAA0247, SFRS5, SLC10A1, SMOC1 D15S816 15q26.2 <0.001 <0.001 MCTP2 D16S764 16p13.11 <0.001 <0.001 LOC339047, SKAP1, HOXB1 through HOXB9 D17S2180 17q21.32 <0.001 <0.001 HOXB13, PRAC2, C17orf92, TTLL6, mir-10a, mir-196-1, CALCOCO2, ZNF554, ZNF555, ZNF556, mir-152 ZNF57 D19S591 19p13.3 <0.001 <0.001 ZNF77, TLE2, TLE6, AES, GNA11, mir-7-3 GNA15, S1PR4, NCLN, BRUNOL5, SOX12, NRSN2, TRIB3, RBCK1, D20S103 20p13 <0.001 <0.001 TBC1D20, CSNK2A1, TCF15, SRXN1 SCRT2, C20orf54 In epithelium only D1S1612 1p36.23 <0.001 PER3, UTS2, TNFRSF9, PARK7, ERRFI1 mir-34a D1S2134 1p33 <0.001 ESTs, ZRANB3, R3HDM1, UBXN4, LCT, MCM6 D2S1334 2q21.3 <0.001 DARS mir-128a D3S1286 3p24.3 <0.001 HACL1, BTD, ANKRD28 D3S1746 3q25.1 <0.001 AADAC, SUCNR1 D3S2427 3q26.31 <0.001 ESTs D3S2418 3q28 <0.001 FGF12, C3orf59 D4S2417 4q34.3 <0.001 ESTs D6S437 6q25.3 <0.001 SYNJ2, SERAC1, GTF2H5, TULP4, TMEM181 D6S1277 6q26 <0.001 ESTs D9S922 9q21.31 <0.001 ESTs D10S2470 10q23.31 <0.001 HTR7, ACVR1B, GRASP, NR4A1, mir-107 C12orf44, KRT80 through KRT86 D12S297 12q13.13 <0.001 KRT7, KRT75, KRT6B, KRT6C mir-196-2 D13S800 13q22.1 <0.001 KLF5, JPH3, LOC100129637, KLHDC4, SLC7A5 (Continued on the following page)

1580 Clin Cancer Res; 18(6) March 15, 2012 Clinical Cancer Research

Integrated Genome-Wide Analyses for Prostate Cancer Stroma

Table 1. LOH/AI hot/cold spots in prostate cancer (Cont'd ) Pa

Markers Loci Epithelium Stroma Genes (within 250 kb of markers) miRNAs (within 2 Mb)

D16S2621 16q24.2 <0.001 CA5A, BANP D17S1294 17q11.2 <0.001 SSH2, EFCAB5, CCDC55, SLC6A4, mir-193, mir-144 BLMH, GSTT2, GSTT2B, DDTL, DDT, GSTTP1 D22S345 22q11.23 <0.001 GSTT1, GSTTP2 CABIN1, SUSD2, GGT5, POM121L9P, SPECC1L D22S683 22q12.3 <0.001 RBM9, APOL1 through APOL4, MYH9 In stroma only D1S1594 1q43 <0.001 FMN2, GREM2, RGS7 D2S1400 2p25.1 <0.001 ROCK2, E2F6, GREB1, NTSR2 D2S434 2q35 <0.001 DIRC3, TNS1, GRAMD1B, SCN3B, mir-26b, mir-153-1 ZNF202, OR6X1 D11S4464 11q24.1 <0.001 OR6M1, PMP22CD, OR8D4, OR4D5, mir-100, let-7a-2, OR6T1, OR10S1 mir-125b-1 D12S1045 12q24.33 <0.001 TMEM132D, LOC100190940, FZD10 D14S606 14q31.1 <0.001 ESTs, STX4, ZNF668, ZNF646, POL3S, VKORC1 D16S753 16p11.2 <0.001 BCKDK, MYST1, PRSS8, PRSS36, FUS, PYCARD, TRIM72, PYDC1, ITGAM, ITGAX, ITGAD, COX6A2, ZNF843, ARMC5, TGFb1I1, SLC5A2, C16orf58, YWHAB, PABPC1L, TOMM34, STK4, D20S481 20q13.12 <0.001 KCNS1, WFDC5, WFDC12, PI3, SEMG1, SEMG2, SLPI, MATN4, RBPJL, SDC4, SYS1, DBNDD2, TP53TG5

a Multiple testing adjustment is based on FPRP0.01<0.1.

For the positive correlation of LOH/AI with Gleason score, P regions of interest (significant hot/cold spots), that have value is obtained by one-side trend test (R, prop.trend.test) significant expression values, and that are included in each with false discovery rate (FDR) being controlled for correct- microarray data set, for hierarchical clustering analyses. All ing multiple testing. hierarchical clustering analyses in this study were conducted using Cluster and TreeView softwares (http://rana.lbl.gov/ Integration of public microarray data with LOH/AI data EisenSoftware.htm; ref. 33). Genes belonging to hot/cold Raw data from Gene Expression Omnibus (GEO) series spots in both epithelium and stroma or in only epithelium GSE3325 at GEO platform GPL570 (Affymetrix HG- were selected and hierarchical clustering was conducted for U133_Plus_2) were downloaded (31) and normalized both genes and samples. using Robust Multi-array Averaging (RMA; ref. 32). The GSE3325 data included 13 individual benign, primary, and Generation of prostatic stroma microarray and metastatic prostate cancers and 6 pooled samples from integration with LOH/AI data benign, primary, or metastatic prostate cancers. These speci- We generated stromal microarray expression data for mens were grossly dissected maintaining at least 90% of the functional validation of our LOH/AI hot/cold spots in tissue of interest, based on examination of the frozen stroma because we could not find appropriate public micro- sections of each tissue block. We integrated microarray data array data sets for prostatic stroma. Stromal cells cultured of the genes residing within the regions that showed sig- from normal peripheral zone tissues (F-PZ-64, F-PZ-79, F- nificant LOH/AI with our genetic data. We sought to use the PZ-82, F-PZ-102, F-PZ-105) and from tumors (F-CA-31, F- expression microarray data to validate our LOH/AI hot/cold CA-39, F-CA-52, F-CA-67, F-CA-93) were established and spots and to identify the significant genes and pathways in grown as previously described (34). Total RNAs were prostate carcinogenesis or progression. Throughout this extracted from semiconfluent cells (passages 4–5) one day study, we consistently selected all genes, that are located after feeding fresh medium using RNeasy Mini Kit (Qiagen) within 250 kb upstream and downstream of the LOH/AI according to the manufacturer’s instructions and treated

www.aacrjournals.org Clin Cancer Res; 18(6) March 15, 2012 1581

Ashida et al.

with TURBO DNA-free kit (Ambion). Hybridizations were contained at 7 loci (D1S2134, D3S2427, D4S2417, conducted according to Illumina protocols by the Genomic D6S1277, D9S922, D12S2078, and D14S606), however, Core, Lerner Research Institute, Cleveland Clinic, Cleve- we found expressed sequence tags (EST) at these loci land, OH. All samples were hybridized to the Illumina (Table 1). Genes belonging to hot/cold spots common to Sentrix Human-6_v3 Expression BeadChip, which contains both epithelium and stroma include HOXB1-HOXB9 clus- 48,000 distinct oligonucleotide probes. These transcrip- ter, HOXB13, ZNF554, ZNF555, ZNF556, ZNF57, and tome data are deposited in GEO [GSE34312 (not publicly ZNF77, as well as 7 miRNA (Table 1). FGF12, ACVRIB, the released yet)]. After normalization using average normali- KRT cluster, and APOL1-APOL4, as well as 6 miRNAs, were zation algorithm of BeadStudio Gene Expression Module included in the genes lying in hot/cold spots of LOH/AI in v3.4 (Illumina), we selected all the genes within 250 kb of epithelium only (Table 1), and TGFb1I1, WFDC5, LOH/AI hot/cold spots in only stroma and applied a hier- WFDC12, and TP53TG5, as well as 5 noncoding RNAs, archical clustering method to both genes and samples. notably miR-125b-1, were among the genes of hot/cold spots in stroma only (Table 1). Integration of microarray data with LOH/AI regions associated with clinicopathologic features Association of LOH/AI with presenting We selected all the genes lying within and 250 kb of the clinicopathologic features significant LOH/AI markers associated with Gleason score LOH/AI markers in association with clinicopathologic or tumor volume in both epithelium and stroma. With the features (Supplementary Table S1) may reveal clues to genes associated with Gleason score or tumor volume, a clinical behavior and biologic diversity and/or serve as hierarchical clustering was conducted for epithelium using biomarkers of prognosis or therapeutic options. Using public microarray data, GSE3325, and for stroma using our logistic regression, we analyzed our genome-wide LOH/ stromal microarray data (31). AI scan across all informative markers to identify compartment-specific loci that were significantly associated with Immunohistochemical analysis aggressiveness of disease as reflected by Gleason score, Formalin-fixed and paraffin-embedded prostate cancer tumor volume, and regional nodal status. We identified sections from 49 independent cases were immunostained 15 LOH/AI regions, 10 in epithelium and 5 in stroma, with a rabbit anti-STIM2 polyclonal antibody (Abcam Inc) associated with Gleason score (Table 2), 11 (8 in epithe- for STIM2 expression in prostate cancer stroma. Both Glea- lium and 3 in stroma) of which showed a positive corre- son grade III (G3) and Gleason grade IV and/or V (GIV/V) lation with increasing Gleason score (one-sided trend test, < lesions were examined for each case. Deparaffinized tissue FDR 0.05; Supplementary Table S2). Among these 11 sections were placed in 10 mmol/L citrate buffer (pH ¼ 6.0) markers, 8 (6 in epithelium and 2 in stroma) remained and were heated to 125C in a Decloaking Chamber (Bio- significantly positively correlated with Gleason score even Core Medical) for 30 seconds for antigen retrieval. These when a more rigorous method was applied (Bonferroni- P < sections were then incubated with a 1:300 dilution of adjusted 0.05). In association with tumor volume, 40 primary antibody overnight at 4C, secondary antibody, markers were identified in epithelium and 35 in stroma and the avidin-biotin complex (VECTASTAIN Elite ABC Kit; (Supplementary Table S3). Interestingly, there were no Vector Laboratories, Inc.) and developed with diaminoben- significant LOH/AI markers correlated with regional nodal zidine. Sections were counterstained with hematoxylin. For metastases. negative controls, primary antibody was omitted. Stromal expression of STIM2 was determined by counting sum of Expression profiling of genes within LOH/AI hot/cold intensities of positive stromal cells divided by area of spot regions using publicly available neoplastic stroma, using Image-Pro Plus 7.0 (Media Cybernetics, Inc.). epithelial microarray data Five random fields were analyzed with a magnification of We focused on genes of LOH/AI hot/cold spot regions 40. STIM2 expression in tumor stroma was divided by and analyzed their expression profiles using publicly avail- those in normal stroma for normalization. The statistical able microarray data (see Methods). This served 2 purposes. significance of STIM2 expression levels between GIII and First, this acts as a functional (transcript expressional) GIV/V lesions was determined with Wilcoxon sign-rank test. validation of the genetic data. Second, integration of the genetic and expressional data would lend clues to the compartment-specific functions of the genes within the Results hot/cold spot regions as they relate to progression and LOH/AI hot/cold spots in prostate cancer prognosis. Hierarchical cluster analysis supervised by 87 Overall, 371 markers across all chromosomes, ranging genes found in LOH/AI hot/cold spot regions 250 kb from 7 on chromosome 22 to 31 on chromosome 1, were resulted in 3 distinct groups: normal epithelium, cancer, analyzed: 38,460 PCR reactions (19,639 for epithelium, and metastatic cancer when considering genetic informa- 18,821 stroma) were informative for LOH/AI evaluation. tion from both epithelium and stroma (Fig. 1A). When we We identified a total of 43 hot/cold spots, 17 occurring in used 61 genes residing within the LOH/AI hot/cold spot both epithelium and stroma, 18 only in the epithelium, and 250 kb regions occurring specifically in epithelium, only 2 8 only in the stroma (Table 1). No known genes were groups could be clearly distinguished, namely, normal

1582 Clin Cancer Res; 18(6) March 15, 2012 Clinical Cancer Research

Integrated Genome-Wide Analyses for Prostate Cancer Stroma

Table 2. LOH/AI in epithelium and stroma associated with Gleason score

Gleason 6 Gleason ¼ 7 Gleason 8

Markersa Loci ROH LOH ROH LOH ROH LOH Pb

Gleason score and LOH/AI in epithelium D1S1612 1.36.23 42 5 14 7 2 5 0.001 D1S1597 1.36.21 33 15 7 15 2 2 0.014 D4S2397 4.15.2 36 9 7 12 2 3 0.002 D5S1470 5.13.3 31 22 4 17 2 5 0.004 D5S1462 5.15 28 20 4 16 4 2 0.008 D8S1477 8.12 30 17 6 15 2 7 0.005 D13S317 13.31.1 23 16 4 16 4 3 0.012 D15S659 15.21.1 32 23 5 18 1 6 0.002 D16S2624 16.22.3 38 11 3 15 6 3 <0.001 D22S277 22.12.3 38 18 6 17 3 5 0.002 Gleason score and LOH/AI in stroma D2S1400 2.25.1 41 12 10 12 6 2 0.028 D4S2397 4.15.2 31 13 5 14 1 4 0.001 D4S1647 4.23 31 22 5 16 4 4 0.023 D9S1825 9.33.3 41 10 11 11 6 4 0.028 D19S591 19.13.3 45 8 15 4 3 6 0.007

Abbreviation: ROH, retention of heterozygosity. aThree markers, which are associated with Gleason score, were excluded because they were missing in more than half of the samples. b Multiple testing adjustment is based on FPRP0.05<0.5. epithelium and metastatic cancer groups (Fig. 1B). Inter- volume, epithelial prostate cancer samples were also suc- estingly, the primary cancer samples were equally clustered cessfully separated into normal, primary cancer, and met- with normal epithelium or with metastatic disease. astatic disease in epithelial analysis (Supplementary Fig. S2A). In contrast, our stromal cell cultures did not cluster into tumor-derived stroma and normal stroma groups in Expression profiling of genes within LOH/AI hot/cold stromal analysis for tumor volume (Supplementary Fig. spot regions using our experimentally generated S2B). stromal microarray data We conducted a hierarchical clustering of stromal cells, Independent validation of candidate gene by supervised by 28 genes of LOH/AI hot/cold spot 250 kb immunohistochemical analysis regions from stroma only (Fig. 2). This accurately separated To further validate our genetic data, we selected our normal stromal cells and cancer-derived stromal cells. strongest candidate gene, STIM2, lying in D4S2397, which is the most significant stroma-associated LOH marker cor- Expression profiling of genes within LOH/AI regions related with Gleason score (Supplementary Table S2). We associated with clinicopathologic features examined STIM2 expression in 49 independent primary We further focused on genes residing in proximity to our prostate cancer samples that had both Gleason grade III LOH/AI regions associated with Gleason score and tumor and Gleason grade IV and/or V lesions by immunohisto- volume and analyzed their expression profiles using pub- chemical analysis (Fig. 3). Stromal protein expression of licly available microarray data (see Methods) for epithelial STIM2 (y-axis) showed significant downregulation in the analysis as well as our experimentally generated expression stroma of GIV/V lesions compared with that of GIII lesions data for stromal analysis. With only one exception, P3, in (P < 0.001; Fig. 3A). neoplastic epithelial analysis, samples were clearly separated into normal, primary cancer, and metastatic disease by 31 genes within the LOH/AI ( 250 kb) regions in neo- Discussion plastic epithelium associated with Gleason score (Supple- Recent molecular studies have provided a myriad of mentary Fig. S1A). Our stromal cell cultures derived from information about genetic changes in prostate cancer and normal tissues and from prostate cancer were accurately accumulating data over the last 2 decades have shown the clustered supervised by the 18 genes belonging to our importance of the tumor stroma in prostate carcinogenesis, stroma-only LOH/AI ( 250 kb) regions associated with with the latter based on mainly functional and cell biologic Gleason score (Supplementary Fig. S1B). As for tumor studies. Here, we have completed a genome-wide LOH/AI

www.aacrjournals.org Clin Cancer Res; 18(6) March 15, 2012 1583

Ashida et al.

A B N2 N1 N3 N4 NX1 NX2 P2 P5 PX1 PX2 P1 P4 P3 W1 W2 W3 W4 WX1 WX2 N2 N1 N3 N4 NX1 NX2 P3 P4 P1 P2 PX1 PX2 P5 W1 W2 WX1 WX2 W3 W4 C7 BANP C6 BANP FCRL1 BANP EMP1 ACVR1B EMP1 SSH2 EMP1 CA5A ZNF554 ANKRD28 CARD6 ERRFI1 CD1D KRT75 RPL37 AADAC CSNK2A1 SSH2 CSNK2A1 C12orf44 CSNK2A1 SSH2 ARSB HTR7 GOLGA1 KRT83 ZNF555 KRT86 DGKG KLF5 Figure 1. Dendrogram and heatmap CALCOCO2 ANKRD28 CALCOCO2 ANKRD28 of hierarchical clustering analysis CSNK2A1 SLC7A5 CSNK2A1 TMEM181 SRXN1 BLMH using 19 prostatic samples from TBC1D20 MCM6 HOXB13 SSH2 HOXB13 KRT80 public microarray database. A, PRKAA1 MCM6 PRKAA1 JPH3 supervised by 87 genes within PRKAA1 TULP4 TBC1D20 TULP4 C17orf92 TULP4 LOH/AI hot/cold spots 250 kb in MCTP2 KLHDC4 MCTP2 DDT epithelium and stroma. In the SFRS5 SUCNR1 SFRS5 LCT C20orf54 GRASP sample axis, samples were clearly HSPA5 JPH3 SERPINI1 APOL2 separated into 3 classes. In the DMGDH SYNJ2 BHMT2 SLC6A4 BHMT2 ZRANB3 gene axis, the genes were KIRREL CABIN1 SMOC1 CABIN1 FCRL1 KRT81 clustered in different branches SMOC1 NR4A1 ZNF555 EFCAB5 according to similarities in relative KNG1 KRT82 NRSN2 SUSD2 PRKAB1 SUSD2 expression. B, supervised by 61 RABEPK UTS2 ARSB APOL1 genes within LOH/AI hot/cold FCRL2 APOL3 FCRL2 GGT5 ARSB BTD spots 250 kb in only epithelium. HRG APOL2 CD5L GSTT1 Note that supervision only by the HOMER1 HTR7 CCDC60 ACVR1B ARSB ACVR1B genes within epithelial hot/cold PXDN SYNJ2 PXDN APOL4 CCDC64 APOL4 spots resulted in the primary HRG SYNJ2 GRIN2B GTF2H5 tumors not being able to be FCRL2 SERAC1 RABEPK SYNJ2 ﬁ GRIN2B SYNJ2 classi ed correctly and were AES KRT85 GNA11 MYH9 instead sorted equally as normal GNA11 FGF12 GNA11 FGF12 ZNF554 FGF12 prostate or metastatic disease. SOX12 EFCAB5 FETUB SLC6A4 This is in contrast to (A) where SOX1 HTR7 ARPC5L JPH3 ARPC5L HTR7 genomic information from both ARPC5L SLC6A4 ARPC5L ACVR1B epithelium and stroma was DNAJB11 GSTT2 HOXB5 BTD C9orf126 NR4A1 integrated with expression array CIT SPECC1L HOXB7 APOL4 HOXB7 KRT7 data. N, benign prostate; P, primary HOXB6 PARK7 HOXB2 HACL1 prostate cancer; W, metastatic HOXB3 UBXN4 ZBBX UBXN4 HOXB7 UBXN4 prostate cancer samples; NX, PX, TLE6 C3orf59 HOXB9 RBM9 or WX, respective mixtures of PRKAB1 DARS HOXB4 DARS CCDC64 CCDC55 benign, primary, or metastatic TRIB3 RBM9 BRUNOL5 PER3 prostate cancer samples (i.e., NX1 HOXB5 RBM9 SOX12 PER3 TTLL6 ZRANB3 indicates the mixture of N1-4 and ZNF77 KLF5 ZNF77 R3HDM1 BHMT TNFRSF9 NX2 is the duplicate of NX1). Red, BRUNOL5 UTS2 GAPVD1 KRT6B the transcript level is above the HOXB8 PER3 HOMER1 TNFRSF9 KIAA0247 median for that gene across all HOMER1 SOX1 samples; green, below the median SERPINI2 WDR49 WDR49 for that gene across all samples; SKAP1 SNTG2 black, unchanged expression; TLE6 GAPVD1 PDCD10 gray, no detectable expression. GAPVD1 GAPVD1 TBCCD1 TLE2 TLE2 GOLGA1 PPP6C PPP6C ZNF57 AHSG CRYGS FCRL2 GNA15 AHSG KNG1 NCLN HOXB3 TCF15 ZNF556 HOXB8 S1PR4 EMP1 HOXB1 KIRREL SLC10A1 KIRREL C20orf54 TPO C13orf28 ZNF555 HOXB9 ZNF555 RBCK1 RBCK1 TRIB3

1584 Clin Cancer Res; 18(6) March 15, 2012 Clinical Cancer Research

Integrated Genome-Wide Analyses for Prostate Cancer Stroma

can initiate apoptosis or autophagic cell death, as well as belong to a multicomponent innate immunity complex– building platform (35). Thus, loss of function of the APOL1- 4 region will lead to loss of apoptosis and autophagic cell death regulation as well as potential decreased innate immunity locally. Our observations here, that is, loss of APOL1-4 function leading to loss of autophagic cell death and consequent loss of essential nutrients and energy to tumor epithelial cells in a paracrine fashion are consistent with the autophagic tumor stroma model of cancer metab- olism underscored by the work of Witkiewicz and collea- gues (36). Furthermore, the relevance to innate immunity lends credence to early hypotheses of metabiomic relation- ships in initiating various neoplasias (35, 37), including

F-PZ-64 F-PZ-79 F-PZ-82 F-PZ-102 F-PZ-105 F-CA-31 F-CA-93 F-CA-52 F-CA-39 F-CA-67 that of the prostate. Of note, the activin/TGF-b signaling BCKDK subnetwork has already been acknowledged to be impor- C16orf58 POL3S tant in prostate carcinogenesis (38). We have shown that MYST1 STK4 D12S297 harbors the human activin receptor, ACVR1B, ROCK2 STX4 which has been described as having somatic mutations and TGFB1I1 ZNF668 deletions in at least 4 other types of cancers, and which is ARMC5 ACVR1B SYS1 expressed in prostate epithelium (39). encodes an VKORC1 b SDC4 activin receptor that signals via the TGF- pathway as well. FMN2 Second, we found 15 specific loci of LOH/AI in the KCNS1 TOMM34 epithelium or stroma associated with Gleason score (Table SLPI ZNF202 2). We suspect that these regions and genes are relevant to DBNDD2 TGFB1I1 invasion and progression and focused especially on stroma. TNS1 FUS Two loci (4p15.2 and 19p13.3) harbored significant LOH/ GREB1 SEMG1 AI in the stroma, correlated with Gleason score. One of the ZNF646 GREM2 genes located in the 4p15.2, which is the most significant ROCK2 region, is STIM2 (stromal interaction molecule 2). Our PYCARD PYCARD immunohistochemical analysis of 49 independent prostate E2F6 YWHAB cancer samples has given us an in-principle validation of YWHAB VKORC1 our genetic and functional genomic studies. STIM1 was originally proposed as a candidate tumor suppressor and regulates intracellular calcium (40, 41). STIM2 is also a Figure 2. Dendrogram and heatmap of hierarchical clustering analysis transmembrane protein that, in limited studies, has been supervised by 28 genes within stromal LOH/AI hot/cold spots 250 kb across 10 stromal cell cultures derived from normal tissues (F-PZ) or shown to mediate cellular proliferation (42). In view of our tumors (F-CA). Note that cell cultures were accurately clustered into each observations, further studies on the role of STIM2 in dif- group according to origin. VKORC1 and ROCK2 that appear twice and ferent cancers, especially in different compartments, are show different profiles represent probe sets that recognize different required. The 19p13.3 region contains AES which is known splice isoforms. Colors, sample names, and the other conditions are the to be associated with Wnt signaling pathway and mitogen— same as those of Fig. 1. activated protein kinase (MAPK)/extracellular signal— regulated kinase 1 (ERK1), which is an upstream kinase for scan and integrated these data with global expression array NF-kB activation (43, 44). The cluster of ZNFs including data for prostate cancer epithelium and stroma and corre- ZNF77 also at the 19p13.3 region was previously associated lated these with presenting clinicopathologic features and with different cancers, and ZNF77 status has been used as a validated our observations via functional genomics prognostic marker in early-stage renal cell carcinomas and approach and immunohistochemistry of a newly identified stage I breast cancers to predict recurrence (45, 46). stroma-relevant gene associated with Gleason score. We subsequently integrated global transcriptome data First, genome-wide LOH/AI scan revealed LOH/AI hot/ from public archives and those we experimentally derived cold spots that were important in each compartment alone with our LOH/AI hot/cold spots. Importantly, these pro- and in both compartments. Among the 225 genes in prox- vided functional validation of our genomic data. The gene imity to the 43 LOH/AI hot/cold spot regions germane to expression profile using neoplastic epithelium from public epithelium and/or stroma, several candidate tumor sup- sources with only genes from our epithelium-related LOH pressor genes, such as HOXB13 (epithelium and stroma), hot/cold spots, which is not the entire epithelium gene the APOL1-4 cluster and ACVR1B (epithelium), and expression profile, could not accurately cluster the samples. TP53TG5 (stroma), are notable. The APOL family of genes Similarly, we also attempted the cluster analysis supervised (APOL1-6) encodes programmed cell death proteins that by the genes from only stromal hot/cold spots, and as

www.aacrjournals.org Clin Cancer Res; 18(6) March 15, 2012 1585

Ashida et al.

that the genes playing roles in tumor stroma are almost ABG3 certainly relevant to invasion, progression, and metastasis, 5 as noted above. By integrating genomic and global expression data for the 4 epithelium and stroma of prostate cancer, we were able to 3 detect signiﬁcant regions and candidate genes important in the tumor-microenvironment cross-talk that leads to carci- 2 G4 nogenesis and progression. Importantly, we have conducted functional genomic validation of the genes in the 1 regions associated with Gleason score as well as tumor

Stromal expression of STIM2 Stromal expression volume, in a compartment-specific manner, as well as 0 G3 G4/5 validated a stromal gene associated with Gleason score in Gleason grade an independent series of tumors using a different technique, namely, immunohistochemistry for STIM2, as proof of principle. Parenthetically, we also noted that the epithelial Figure 3. Immunohistochemical analysis reveals decreased STIM2 expression of STIM2 (D4S2397 is also a significant epithe- expression in the stroma of Gleason grade 4 and/or 5 (G4/5) regions compared with that of Gleason 3 regions. A, the evaluation in lium-associated LOH marker correlated with Gleason score immunohistochemical staining of prostate cancer stroma with anti- as well) was downregulated in G4/5 over G3. However, we STIM2 antibody, indicating that STIM2 gene expressions were did not formally analyze the epithelial expression in this significantly downregulated in G4/5 in comparison with GIII (Wilcoxon study because the effect size did not appear as marked as for < sign-rank test, P 0.001). The x-axis shows Gleason grade and y-axis stroma. These data provide fundamental insights into com- indicates stromal expression of STIM2 (see Methods). B, representative staining lesions of GIII and G4/5 for the same case are shown (400). partment-specific roles in prostate carcinogenesis or progression and may also reveal the solid tumor microenvironment as a novel compartment for important biomarkers expected, it resulted in failure to cluster the samples (data of prognosis as well as targeted therapy. not shown). In contrast, these samples could be accurately fl clustered when taking into account the genes from the Disclosure of Potential Con icts of Interest No potential conflicts of interests were disclosed. stromal hot/cold spots, indicating that genomic alterations with consequent expressional alterations in the stroma are Acknowledgments necessary and play an important role in prostate carcino- The authors thank the members of the laboratory of C. Eng for technical genesis and progression. Equally important is a notable assistance and helpful discussion over the years. negative. The genes within the LOH hot/cold spot regions identified as important in correlating with tumor volume Grant Support This work was funded, in part, by the Department of Defense US Army were germane only in the epithelium. This was validated Prostate Cancer Research Program (C. Eng). C. Eng holds the Sondra J. when expression data supervised by the genes within the and Stephen R. Hardis Endowed Chair of Cancer Genomic Medicine at the LOH hot/cold spot regions relevant to tumor volume could Cleveland Clinic, and the ACS Clinical Research Program, funded, in part, by the F.M. Kirby Foundation. not accurately classify normal stroma from tumor stroma The costs of publication of this article were defrayed in part by the (Supplementary Fig. S2B). In other words, at least in pros- payment of page charges. This article must therefore be hereby marked advertisement tate cancer, tumor volume does not appear to be influenced in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. by stromal genomic or expressional alterations of genes located within tumor volume–associated stromal LOH Received October 1, 2011; revised January 11, 2012; accepted January 12, regions in this study. This makes teleology logical given 2012; published OnlineFirst January 24, 2012.

References 1. Bian Y, Knobloch TJ, Sadim M, Kaklamani V, Raji A, Yang GY, et al. 5. Joshua AM, Vukovic B, Braude I, Hussein S, Zielenska M, Srigley J, Somatic acquisition of TGFBR16A by epithelial and stromal cells et al. Telomere attrition in isolated high-grade prostatic intraepithelial during head and neck and colon cancer development. Hum Mol Genet neoplasia and surrounding stroma is predictive of prostate cancer. 2007;16:3128–35. Neoplasia 2007;9:81–9. 2. Fukino K, Shen L, Patocs A, Mutter GL, Eng C. Genomic instability 6. Kim HS, Lim HS, Lee SH, Lee JW, Nam SW, Park WS, et al. Mito- within tumor stroma and clinicopathological characteristics of spo- chondrial microsatellite instability of colorectal cancer stroma. Int J radic primary invasive breast carcinoma. JAMA 2007;297:2103–11. Cancer 2006;119:2607–11. 3. Hu N, Wang C, Hu Y, Yang HH, Kong LH, Lu N, et al. Genome-wide loss 7. Kurose K, Hoshaw-Woodard S, Adeyinka A, Lemeshow S, Watson PH, of heterozygosity and copy number alteration in esophageal squa- Eng C. Genetic model of multi-step breast carcinogenesis involving mous cell carcinoma using the Affymetrix GeneChip Mapping 10 K the epithelium and stroma: clues to tumour-microenvironment inter- array. BMC Genomics 2006;7:299. actions. Hum Mol Genet 2001;10:1907–13. 4. Ishiguro K, Yoshida T, Yagishita H, Numata Y, Okayasu T. Epithelial 8. Kurose K, Gilley K, Matsumoto S, Watson PH, Zhou XP, Eng C. and stromal genetic instability contributes to genesis of colorectal Frequent somatic mutations in PTEN and TP53 are mutually exclusive adenomas. Gut 2006;55:695–702. in the stroma of breast carcinomas. Nat Genet 2002;32:355–7.

1586 Clin Cancer Res; 18(6) March 15, 2012 Clinical Cancer Research

Integrated Genome-Wide Analyses for Prostate Cancer Stroma

9. Macintosh CA, Stower M, Reid N, Maitland NJ. Precise microdissec- evidence for multiple tumor suppressors and identifies novel candidate tion of human prostate cancers reveals genotypic heterogeneity. regions. Am J Hum Genet 2003;73:748–67. Cancer Res 1998;58:23–8. 27. Tarin D, Thompson EW, Newgreen DF. The fallacy of epithelial mes- 10. Man Y, Mannion C, Kuhls E, Moinfar F, Bratthauer GL, Albores- enchymal transition in neoplasia. Cancer Res 2005;65:5996–6000; Saavedra J, et al. Allelic losses at 3p and 11p are detected in both discussion 6000–1. epithelial and stromal components of cervical small-cell neuroendo- 28. Faraway JJ. Extending the linear model with R: generalized linear, crine carcinoma. Appl Immunohistochem Mol Morphol 2001;9:340–5. mixed effects and nonparametric regression models. Boca Raton, FL: 11. Matsumoto N, Yoshida T, Yamashita K, Numata Y, Okayasu I. Possible Chapman & Hall/CRC; 2006. alternative carcinogenesis pathway featuring microsatellite instability 29. McCullagh P, Nelder JA. Generalized linear models. 2nd ed. London, in colorectal cancer stroma. Br J Cancer 2003;89:707–12. UK; New York: Chapman and Hall; 1989. 12. Moinfar F, Man YG, Arnould L, Bratthauer GL, Ratschek M, Tavassoli 30. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N. FA. Concurrent and independent genetic alterations in the stromal and Assessing the probability that a positive report is false: an approach for epithelial cells of mammary carcinoma: implications for tumorigenesis. molecular epidemiology studies. J Natl Cancer Inst 2004;96:434–42. Cancer Res 2000;60:2562–6. 31. Varambally S, Yu J, Laxman B, Rhodes DR, Mehra R, Tomlins SA, et al. 13. Paterson RF, Ulbright TM, MacLennan GT, Zhang S, Pan CX, Sweeney Integrative genomic and proteomic analysis of prostate cancer reveals CJ, et al. Molecular genetic alterations in the laser-capture-microdis- signatures of metastatic progression. Cancer Cell 2005;8:393–406. sected stroma adjacent to bladder carcinoma. Cancer 2003;98: 32. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP. 1830–6. Summaries of Affymetrix GeneChip probe level data. Nucleic Acids 14. Patocs A, Zhang L, Xu Y, Weber F, Caldes T, Mutter GL, et al. Breast- Res 2003;31:e15. cancer stromal cells with TP53 mutations and nodal metastases. N 33. DeMarzo AM, Nelson WG, Isaacs WB, Epstein JI. Pathological and Engl J Med 2007;357:2543–51. molecular aspects of prostate cancer. Lancet 2003;361:955–64. 15. Shiraishi H, Mikami T, Yoshida T, Tanabe S, Kobayashi N, Watanabe 34. Peehl DM, Sellers RG. Cultured stromal cells: an in vitro model of M, et al. Early genetic instability of both epithelial and stromal cells in prostatic mesenchymal biology. Prostate 2000;45:115–23. esophageal squamous cell carcinomas, contrasted with Barrett's 35. Smith EE, Malik HS. The apolipoprotein L family of programmed cell adenocarcinomas. J Gastroenterol 2006;41:1186–96. death and immunity genes rapidly evolved in primates at discrete sites 16. Tuhkanen H, Anttila M, Kosma VM, Yla-Herttuala S, Heinonen S, of host-pathogen interactions. Genome Res 2009;19:850–8. Kuronen A, et al. Genetic alterations in the peritumoral stromal cells 36. Witkiewicz AK, Kline J, Queenan M, Brody JR, Tsirigos A, Bilal E, et al. of malignant and borderline epithelial ovarian tumors as indicated Molecular profiling of a lethal tumor microenvironment, as defined by by allelic imbalance on chromosome 3p. Int J Cancer 2004; stromal caveolin-1 status in breast cancers. Cell Cycle 2011;10: 109:247–52. 1794–809. 17. Weber F, Fukino K, Sawada T, Williams N, Sweet K, Brena RM, et al. 37. Eisenberg-Lerner A, Kimchi A. The paradox of autophagy and its Variability in organ-specific EGFR mutational spectra in tumour epi- implication in cancer etiology and therapy. Apoptosis 2009;14:376–91. thelium and stroma may be the biological basis for differential 38. Dowling CR, Risbridger GP. The role of inhibins and activins in prostate responses to tyrosine kinase inhibitors. Br J Cancer 2005;92:1922–6. cancer pathogenesis. Endocr Relat Cancer 2000;7:243–56. 18. Weber F, Xu Y, Zhang L, Patocs A, Shen L, Platzer P, et al. Microen- 39. Su GH, Bansal R, Murphy KM, Montgomery E, Yeo CJ, Hruban RH, vironmental genomic alterations and clinicopathological behavior in et al. ACVR1B (ALK4, activin receptor type 1B) gene mutations in head and neck squamous cell carcinoma. JAMA 2007;297:187–95. pancreatic carcinoma. Proc Natl Acad Sci U S A 2001;98:3254–7. 19. Wernert N, Locherbach C, Wellmann A, Behrens P, Hugel A. Presence 40. Parker NJ, Begley CG, Smith PJ, Fox RM. Molecular cloning of a novel of genetic alterations in microdissected stroma of human colon and human gene (D11S4896E) at chromosomal region 11p15.5. Genomics breast cancers. Anticancer Res 2001;21:2259–64. 1996;37:253–6. 20. Chung LW, Baseman A, Assikis V, Zhau HE. Molecular insights into 41. Sabbioni S, Barbanti-Brodano G, Croce CM, Negrini M. GOK: a gene at prostate cancer progression: the missing link of tumor microenviron- 11p15 involved in rhabdomyosarcoma and rhabdoid tumor development. J Urol 2005;173:10–20. ment. Cancer Res 1997;57:4493–7. 21. Sung SY, Hsieh CL, Law A, Zhau HE, Pathak S, Multani AS, et al. 42. El Boustany C, Katsogiannou M, Delcourt P, Dewailly E, Prevarskaya Coevolution of prostate cancer and bone stroma in three-dimensional N, Borowiec AS, et al. Differential roles of STIM1, STIM2 and Orai1 in coculture: implications for cancer growth and metastasis. Cancer Res the control of cell proliferation and SOCE amplitude in HEK293 cells. 2008;68:9996–10003. Cell Calcium 2010;47:350–9. 22. Tuxhorn JA, Ayala GE, Rowley DR. Reactive stroma in prostate cancer 43. Chen Y, Guo Y, Ge X, Itoh H, Watanabe A, Fujiwara T, et al. Elevated progression. J Urol 2001;166:2472–83. expression and potential roles of human Sp5, a member of Sp tran- 23. McCarthy RP, Zhang S, Bostwick DG, Qian J, Eble JN, Wang M, et al. scription factor family, in human cancers. Biochem Biophys Res Molecular genetic evidence for different clonal origins of epithelial and Commun 2006;340:758–66. stromal components of phyllodes tumor of the prostate. Am J Pathol 44. Tetsuka T, Uranishi H, Imai H, Ono T, Sonta S, Takahashi N, et al. 2004;165:1395–400. Inhibition of nuclear factor-kappaB-mediated transcription by associ- 24. Fukino K, Shen L, Matsumoto S, Morrison CD, Mutter GL, Eng C. ation with the amino-terminal enhancer of split, a Groucho-related Combined total genome loss of heterozygosity scan of breast cancer protein lacking WD40 repeats. J Biol Chem 2000;275:4383–90. stroma and epithelium reveals multiplicity of stromal targets. Cancer 45. Nikolic-Vukosavljevic D, Markicevic M, Grujic-Adanja G, Petrovic A, Res 2004;64:7231–6. Kanjer K, Neskovic-Konstantinovic Z. Cathepsin D-related disease- 25. Weber F, Shen L, Fukino K, Patocs A, Mutter GL, Caldes T, et al. Total- free interval in pT1 primary breast carcinomas: a pilot study. Clin Exp genome analysis of BRCA1/2-related invasive carcinomas of the Metastasis 2005;22:363–8. breast identifies tumor stroma as potential landscaper for neoplastic 46. Salama ME, Guru K, Stricker H, Peterson E, Peabody J, Menon M, et al. initiation. Am J Hum Genet 2006;78:961–72. pT1 substaging in renal cell carcinoma: validation of the 2002 TNM 26. Miller BJ, Wang D, Krahe R, Wright FA. Pooled analysis of loss of staging modification of malignant renal epithelial tumors. J Urol heterozygosity in breast cancer: a genome scan provides comparative 2005;173:1492–5.

www.aacrjournals.org Clin Cancer Res; 18(6) March 15, 2012 1587

Integrated Analysis Reveals Critical Genomic Regions in Prostate Tumor Microenvironment Associated with Clinicopathologic Phenotypes

Shingo Ashida, Mohammed S. Orloff, Gurkan Bebek, et al.

Clin Cancer Res 2012;18:1578-1587. Published OnlineFirst January 24, 2012.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-11-2535

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2012/01/24/1078-0432.CCR-11-2535.DC1

Cited articles This article cites 44 articles, 11 of which you can access for free at: http://clincancerres.aacrjournals.org/content/18/6/1578.full#ref-list-1

Citing articles This article has been cited by 4 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/18/6/1578.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/18/6/1578. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.