The IL-18 Antagonist IL-18–Binding Protein Is Produced in the Human Ovarian Cancer Microenvironment

Published OnlineFirst July 19, 2013; DOI: 10.1158/1078-0432.CCR-13-0568

Clinical Cancer Human Cancer Biology Research

Grazia Carbotti1, Gaia Barisione1, Anna Maria Orengo1, Antonella Brizzolara1, Irma Airoldi3, Marina Bagnoli4, Patrizia Pinciroli4, Delia Mezzanzanica4, Maria Grazia Centurioni2, Marina Fabbi1, and Silvano Ferrini1

Abstract Purpose: Interleukin (IL)-18 is an immune-enhancing cytokine, which induces IFN-g production, T-helper 1 responses, and antitumor effects. In turn, IFN-g stimulates IL-18–binding protein production, which blocks IL-18 activity. In view of the potential use of IL-18 in epithelial ovarian cancer (EOC) immunotherapy, here, we studied IL-18BP expression and its regulation by cytokines in EOC cells in vitro and in vivo. Experimental Design: Expression and production of IL-18BP in EOC cell lines, primary ovarian carcinomas, and the corresponding normal tissues, patients’ serum, and ascites were investigated by immunochemistry, ELISA, screening of gene expression profiles, and reverse-transcription PCR. Results: Analysis of gene expression profiles revealed that IL18BP mRNA is increased in EOC tumors compared with normal ovary cells. Release of IL-18BP was detectable in EOC sera and to a greater extent in the ascites, indicating production at the tumor site. Indeed, immunochemical analyses on cells isolated from the ascites and on tumor sections indicated that IL-18BP is expressed in both tumor cells þ and tumor-associated leukocytes, which displayed a CD3 CD20 NKp46 CD13 CD14low phenotype. EOC cell lines do not constitutively express IL-18BP. However, its release is inducible both by IFN-g stimulation in vitro and by xenotransplantation of EOC cells in immune-deficient mice, suggesting a role for the microenvironment. In vitro experiments and immunochemistry indicated that IL-27 is also involved in IL-18BP upregulation in EOC cell lines and primary cells through STAT1 activation. Together, these data indicate that IL-18BP, which is produced in EOC in response to microenvironmental factors, may inhibit endogenous or exogenous IL-18 activity. Clin Cancer Res; 19(17); 1–10. 2013 AACR.

Introduction and that T-cell–mediated immunity may have an impact on Epithelial ovarian cancer (EOC) represents 80% of all clinical course of EOC (3). However, similar to other ovarian malignancies, is frequently diagnosed at advanced tumors, EOC has the ability to escape from immune system stages, and has a poor 5-year survival rate (1). Evidence from control through several mechanisms (4, 5). several studies indicates that T lymphocytes capable of Several cytokines and chemokines were found to be ele- recognizing EOC cells are present at the tumor site (2, 3) vated in serum and in the ascites of patients with EOC and have been implicated in tumor development, angiogenesis (6, 7), progression (8), drug resistance (9), or immune Authors' Affiliations: Departments of 1Integrated Oncological Therapies suppression (10). Some of these cytokines have been con- 2 and Surgery, IRCCS AOU San Martino-IST Istituto Nazionale per la sidered as serologic biomarkers of EOC (11). Among them, Ricerca sul Cancro, and 3Laboratory of Oncology, IRCCS Istituto G. Gaslini, Genoa, Italy; and 4Department of Experimental Oncology and Molecular interleukin (IL)-18 was proposed as a potential biomarker by Medicine, Fondazione IRCCS Istituto Nazionale dei Tumori, Milan, Italy gene expression profiling (12) and, indeed, IL-18 levels were Note: Supplementary data for this article are available at Clinical Cancer elevated in serum and ascites of patients with EOC (12, 13). Research Online (http://clincancerres.aacrjournals.org/). IL-18 is a proinflammatory and immune-enhancing cytokine, which induces IFN-g production by T cells and natural M. Fabbi and S. Ferrini contributed equally to this work. killer (NK) cells, mediates T-helper 1 cell (TH1) polariza- G. Carbotti is enrolled in the Doctorate School of Genetics, University of tion, and is involved in the defense from pathogens (14– Genoa, Italy. 16). IL-18 is synthesized as an inactive precursor (pro-IL- Corresponding Authors: Silvano Ferrini, UOC Terapia Immunologica, 18), which is converted to a mature form (mat-IL-18) by IRCCS AOU San Martino-IST Istituto Nazionale per la Ricerca sul Cancro, Largo R. Benzi 10, 16132 Genova, Italy. Phone: 0039-010-555-8372; Fax: caspase-1 (17). IL-18 is released from cells as both pro- and 0039-010-555-8374; E-mail: [email protected]; and Marina Fabbi, mat-IL-18 forms, but only mat-IL-18 can bind to IL-18R. E-mail: [email protected] Moreover, in preclinical tumor models, mat-IL-18 exhibited doi: 10.1158/1078-0432.CCR-13-0568 antitumor properties through the induction of an immune 2013 American Association for Cancer Research. response, whereas pro-IL-18 had no activity (18, 19).

www.aacrjournals.org OF1

Carbotti et al.

10% FCS, and antibiotics (Lonza). NK-92 cells (ATCC) Translational Relevance were grown in medium containing 600 IU IL-2 (Novartis). Interleukin (IL)-18 is an immune-enhancing cyto- A vial of each cell line master stock was recently genotyped kine, which is being studied in clinical trials of immu- using the Cell ID System (Promega) and GeneMapper notherapy. IL-18BP is a natural antagonist of IL-18 software, version 4.0. and limits IL-18 biologic activity. Here, we show that Cells (50 103/well) were seeded in 24-well plates in IL-18BP is produced by both tumor cells and tumor- culture medium. The following day, culture medium was associated leukocytes. Expression of IL-18BP in ovarian replaced with medium with or without human recombi- cancer cells in vitro is induced by cytokines such as nant IFN-g (PeproTech), human recombinant IL-27 (R&D IFN-g and IL-27, which may play a role in the tumor System) or human recombinant IL-35 (Enzo Life Sciences). environment. The high local levels of IL-18BP at the For IL-27R blocking experiments, an anti-gp130 antibody tumor site may limit the induction of an efﬁcient immune was added [monoclonal antibody (mAb) 228, R&D Sys- response by either endogenous or therapeutic IL-18. tems)]. Treatment was carried out for 48 hours. Condi- tioned media were then collected, centrifuged, and used for IL-18BP detection.

EOC cell lines release pro-IL-18, but not mat-IL-18, due to RT-PCR analysis of IL18BP mRNA expression defective caspase-1 expression or activation, whereas nor- Cells were detached by trypsin and washed, and total mal ovarian epithelial cells secrete mat-IL-18 (20). Indeed, RNA was isolated by the NucleoSpin RNA II kit (Macherey- IL-18 present at high levels in EOC ascites is predominantly Nagel) and reverse-transcribed using the SuperScript II the inactive pro-IL-18 (13), although we cannot exclude Reverse Transcriptase (Invitrogen). Quantitative RT-PCR that low levels of mat-IL-18, eventually present, may con- analysis was conducted using the following primers: tribute to the immune response. POLR2A upper primer GACAATGCAGAGAAGCTGG, lower IL-18 binding protein (IL-18BP) is an inhibitor of IL-18 primer GCAGGAAGACATCATCATCC; GAPDH upper prim- activity, as it binds the mature form of this cytokine and er GAAGGTGAAGGTCGGAGT, lower primer CATGGGTG- blocks its interaction with IL-18R (21) and its isoform "a" is GAATCATATTGGAA; IL18BP upper primer GTGTCCAGC- the mostly expressed form (22). IL-18BP is produced by ATTGGAAGTGACC, and lower primer GGAGGTGCTCAA- monocytes and macrophages (23) and by prostatic (24) TGAAGGAACC. Amplification was carried out by the Mas- and colorectal tumor cells (25) in response to IFN-g stim- tercycler ep realplex instrument (Eppendorf International) ulation. IFN-g is the physiologic inducer of IL-18BP, which using the iQ SYBR Green Supermix system (Bio-Rad in turn inhibits IL-18 biologic activity in a negative feed- Laboratories). Relative quantification of mRNAs was calcu- back loop. IL-18BP accumulates in the serum in chro- lated by the DDCt method. For semiquantitative RT-PCR, nic renal failure, in which it may contribute to the defective 2 mL of cDNA was separately amplified with 0.25 U of immune response (26). Taq DNA Polymerase (Roche) in the presence of 1 mmol/L To further address the biologic role of endogenous of the following primers: IL18BP upper primer ACCATG- IL-18 in human EOC, we conducted an integrated ana- AGACACAACTGGACACCAG, lower primer TTAACCCTG- lysis of IL-18 and IL-18BP in patients with EOC. The CTGCTGTGGACTGCTG; housekeeping gene ACTB upper study of IL-18BP also seemed relevant, as IL-18–based primer GGCATCGTGATGGACTCCG, lower primer GCTG- immunotherapy is being evaluated in patients with EOC GAAGGTGGACAGCGA in an Eppendorf Mastercycler ep (refs. 27, 28; NCT00659178) and high levels of IL-18BP Gradient S. PCR products were analyzed on 1% agarose gel may interfere with this treatment. We found that, indeed, stained with ethidium bromide. IL-18BP levels are elevated in the serum and particularly in the ascites of patients with EOC and that IL-18BP Patients is expressed by EOC cells and by reactive leukocytes. Our Clinical samples were obtained upon written informed data also suggest that IL-18BP production may be the consent and previous approval by the Institutional outcome of cross-talk between tumor cells and the micro- Review Board from patients and tumor-free, age-matched environment, involving IFN-g and IL-27, a member of (median ¼ 60 years; range ¼ 43–75) women. All 55 patients the IL-12 cytokine family (29, 30). Together, these data showed evidence of disease and untreated or off treatment support an immune-regulatory role for IL-18BP in EOC, for at least 2 months (Table 1). Ascitic fluids were collected as it may limit the activity of endogenous or therapeutic during surgical procedures. Tumor histopathology, grade, IL-18. and stage were assigned according to the International Federation of Gynecology and Obstetrics (FIGO) criteria.

Materials and Methods EOC xenotransplant model Cells and cell treatments Female homozygous nonobese diabetic/severe com- The human EOC cell lines SKOV3 (ATTC), A2780 bined immunodeﬁcient (NOD/SCID) mice (The Jackson (ICLC), A2774 (IST, Genoa, Italy), and OVCAR5 (INT, Laboratory) were bred in-house. Nude mice were from Milan, Italy) were grown in RPMI-1640, with L-glutamine, Janvier. The experiments were carried out according to the

OF2 Clin Cancer Res; 19(17) September 1, 2013 Clinical Cancer Research

IL-18BP in Ovarian Cancer

Table 1. Distribution by tumor characteristics high-pH citrate buffer in a microwave oven. The sections were immunostained using rabbit anti-IL-18BP (clone for EOC patients with evidence of disease EP1088Y, Epitomics), anti-EBI3 (Novus Biologicals Eur- ope), or anti-IL27A (LifeSpan BioSciences) overnight at Cases at Cases at 4C. The antibody complex was revealed with diagnosis, n relapse, n the EnVisionþ System-Peroxidase (Dako) and 3-amino- Total 48 7 9-ethylcarbazole (AEC, Calbiochem). The sections were Age at diagnosis, y counterstained with modified Mayer hematoxylin and >55 35 5 mounted in PermaFluor (Thermo Scientific). The sections i 55 12 were observed with a Nikon Eclipse 80 light microscope NA 1 2 equipped with a color camera imaging head, using a 40 objective. Stagea I15 Public EOC datasets of gene expression II 7 We explored IL18BP gene expression in our dataset III 22 3 (Iorio, GSE19532) and in public EOC datasets pro- IV 3 3 cessed through the Affymetrix HG U133 Plus 2.0 arrays NA 1 1 (Tothill, GSE9891 and Anglesio GSE12172). Only type II Histotype tumors (31) were considered for all three datasets when Serous 30 4 exploring pattern of correlation among different relevant Endometrioid 10 1 genes. We considered only the platforms including Mucinous 3 probes 222868_s_at because this is the only probe that Others 3 1 comprises the IL-18BPa isoform. The Tothill dataset con- NA 2 1 sists of 18 low malignant potential (LMP), 10 type I and Grade 210 type II EOC; the Anglesio dataset consists of 30 LMP 113 and 58 type II EOC cases; and the Iorio dataset includes 17 2193EOC high-grade tumors, two preparations of normal ovarian surface epithelial cells (OSE), and six EOC cell 3153 lines. Raw data were downloaded from GEO and normal- Borderline ized through the RMA algorithm (with Expression Con- NA 1 1 sole software, Affymetrix) except for the Iorio dataset, for Abbreviation: NA, not available. which the processed matrix was downloaded from GEO. aStage at diagnosis. Statistical analysis The normal distribution of the data was verified before applying parametric tests by transforming data to loga- National Regulation on Animal Research Resources and rithms. The one-way ANOVA and appropriate multiple approved by the Institutional Review Board. Six-week-old comparison tests were used to compare expression NOD/SCID animals were injected intraperitoneally with levels between patients and control subjects. The paired 5 106 SKOV3 or A2774 cells. When ascites developed, Student t test was used when appropriate. Parametric animals were euthanized and blood and ascites collected. methods were used to examine the correlation among Nu/Nu mice were surgically implanted in the left ovary with gene or protein expression levels (Pearson correlation 2 106 SKOV3 cells. Tumor masses were excised and fixed coefficient). An a level of 0.05 was used for all statistical in 10% buffered formalin. tests. GraphPad Prism 5.0d software and R statistical language (http://www.R-project.org; version 12.2) were ELISA for IL-18, IL-18BP, and IFN-g used. Samples were tested with commercially available ELISA kits for human IL-18 (MBL), IL-18BPa (DuoSet R&D Western blot analysis Systems), and IFN-g (Quantikine, R&D Systems). Assays For the analysis of phosphorylated proteins, 107 EOC were conducted in duplicate and background values were cells were incubated for 10 minutes at 37Cwithor subtracted. without 20 ng/mL of rIL-27 in 0.5 mL of medium. Cells were lysed in 100 mL of buffer containing 1 mmol/L Immunochemistry sodium orthovanadate. Lysates were resolved under Immunochemistry detection of IL-18BP was conduct- reducing conditions by 10% SDS-PAGE and analyzed ed on sections of formaldehyde-fixed paraffin-embedded by Western blotting using rabbit anti-phospho-STAT1 cell pellets or tumors explanted from mice or on com- (pY701) anti-serum (Cell Signaling Technology) or mercially available tissue microarrays of patients with anti-b actin or tubulin mAb (Sigma) and chemilumines- EOC (Super Bio Chips). Antigen retrieval was done with cence detection.

www.aacrjournals.org Clin Cancer Res; 19(17) September 1, 2013 OF3

Carbotti et al.

Figure 1. Expression of IL18BP mRNA and secretion of IL-18BP protein in EOC. A, normalized IL18BP gene expression levels in EOC tumors, cell lines, and OSE. Expression levels were significantly higher in tumor samples (one-way ANOVA and Bonferroni post-test). B, IL18BP mRNA expression was significantly higher in high-grade tumors (type II) relative to low malignant potential tumors (LMP), in two independent datasets by two-sided unpaired Student t test (Anglesio dataset) and one-way ANOVA and Tukey post-test (Tothill dataset). Boxes and whiskers represent median and quartiles with minimum and maximum. C, IL-18BP ELISA levels were significantly higher in sera from EOC patients at diagnosis (n ¼ 48) or at relapse (n ¼ 7) than in age-matched female controls (n ¼ 13) by Kruskal–Wallis test. IL-18BP levels were similar in stage I/II and stage III/IV patients. D, IL-18BP levels were higher in the ascites than in sera simultaneously collected from the same patient (by two-sided paired Student t test). Mean and SD are indicated in C and D.

Results between the two groups was observed (P ¼ 0.43), suggest- Gene expression of IL18BP in EOC tumors ing that IL-18BP is altered at early stages of EOC and that IL18BP mRNA was first to be analyzed in our gene IL-18BP serum levels are independent from tumor burden expression dataset (Iorio) including primary EOC tumors, (Fig. 1C). In addition, when patients were stratified in tumor cell lines, and normal ovarian surface epithelial type I and in type II, according to a recent classification (OSE) cells. A significantly higher expression intensity of (31), no difference was observed between the two groups IL18BP was observed in tumor samples than in EOC cell (data not shown). Consistently, no differences were lines (P < 0.0001) and in OSE cells (P ¼ 0.01; Fig. 1A). In observed by stratifying patients in accordance with tumor addition, IL18BP expression was significantly higher in type grading (data not shown). Performance of serum IL-18BP II (high-grade) tumors (31) relative to low malignant as classifier, evaluated by Receiver Operating Characteris- potential tumors, in two independent datasets (Tothill and tic analysis, yielded an area under the curve of 0.734, Anglesio, P ¼ 0.0043 and P < 0.0001, respectively), suggest- suggestive of a poor performance (Supplementary Fig. S1). ing a possible relationship of high IL18BP expression with To gain information on the possible tumor origin of malignancy (Fig. 1B). the high serum IL-18BP levels, we tested 18 EOC serum and ascites pairs, collected at the same time. By paired analysis, IL-18BP levels were significantly higher in the IL-18BP levels are elevated in EOC sera and ascites ascites than in serum of the same patients (mean SD ¼ We then tested IL-18BP serum levels in 48 patients with 31.9 14.7, median 32.8 vs. mean SD ¼ 11.7 5.9, EOC (Table 1), all with evidence of disease and untreated, median 10.5 ng/mL, P < 0.0001; Fig. 1D), thus suggesting in 7 patients at relapse, and in 13 age-matched healthy that IL-18BP derives from the tumor site, where it could women by ELISA for IL-18BPa, the major isoform of IL- limit the IL-18–driven immune response. 18BP (22). IL-18BP serum levels were higher in both untreated (onset, mean SD ¼ 11.06 5.7; median ¼ Analyses of IL-18BP correlation with IL-18 and IFN-g 9.45 ng/mL, P ¼ 0.03) and in patients with relapsing EOC It is known that IL-18BP inhibits IL-18 as the result of a (mean SD ¼ 11.76 3.2; median ¼ 11.21 ng/mL, P ¼ negative feedback loop mediated by IL-18–induced IFN-g 0.01) than in healthy women (mean SD ¼ 7.4 3.5; (21, 23). In agreement with our previous data (13), IL-18 median ¼ 6.86 ng/mL; Fig. 1C). By stratifying patients in ELISA levels were elevated in EOC serum and ascites (Sup- stage I/II and stage III/IV, no significant difference plementary Fig. S2A and S2B).

OF4 Clin Cancer Res; 19(17) September 1, 2013 Clinical Cancer Research

IL-18BP in Ovarian Cancer

showed positivity for IL-18BP, although stronger expression was found in tumor-associated leukocytes (Fig. 3A). Staining appeared speciﬁc, as no reactivity was found on a contiguous section stained with secondary antiserum. Immunohistochemical analyses of tissue microarrays revealed that most EOC tumors expressed IL-18BP, irre- spective of the tumor histotype, although at variable intensity in different tumors. Both neoplastic cells and inﬁltrating leukocytes showed expression of IL-18BP (Fig. 3B). The IL-18BP–expressing leukocytes showed monocyte- or granulocyte-like morphologic features and had a nonlymphoid (CD3 CD20 NKP46 ) but myeloid þ (CD13 CD14 /low) surface phenotype (Supplemen- tary Fig. S3). These features may be consistent with

Figure 2. Analysis of the correlation among IL18, IL18BP,andIFNG mRNA levels in high-grade (typeII)tumors.A,asignificant correlation was found between IL18 and IL18BP in both datasets. B, correlations between IL18 and IFNG, and between IL18BP and IFNG were also found only in the Tothill dataset. Pearson correlation coefficients are shown (r). Lines represent the best fit linear regression analysis with the 95% confidence interval.

We therefore first evaluated whether there was any relationship between IL18 and IL18BP mRNA in two public access datasets of ovarian cancers and found a moderate, yet significant correlation (Anglesio dataset, P ¼ 0.0056; Tothill dataset, P < 0.0001; Fig. 2A). In addition, a correlation was found for IFNG and IL18BP and for IFNG and IL18 in the Tothill dataset (IFNG vs. IL18, P < 0.0001; IL18BP vs. IFNG, P < 0.0001), although only a minority of patients showed elevated levels of IFNG expression (Fig. 2B). However, when we investigated possible correlations at the protein level, we found no significant correlation between IL-18 and IL-18BP in the sera (r ¼ 0.12, P ¼ ns by Pearson correlation) and in the ascites (r ¼ 0.3, P ¼ ns) of patients with EOC (Supplementary Fig. S2A and S2B). Moreover, IFN-g levels were low or undetectable in ascites and sera from patients with EOC, in agreement with previous reports, and showed no correlation with IL-18 and IL-18BP (Supplementary Fig. S2C). Although microarray data suggest that the IL-18/IFN-g axis may be involved in the regulation of IL-18BP expression in the tumor tissue of a few patients, no evidence of this Figure 3. Immunochemical analysis of IL-18BP expression in EOC and in regulation was found in sera and ascites. normal tissue counterparts. A, IL-18BP staining of cells isolated from ascites. Both tumor cell nests and particularly tumor-associated leukocytes show staining for IL-18BP (top). Negative control of a Both EOC cells and reactive cells from the contiguous section stained only with secondary antibody is shown. Scale microenvironment express IL-18BP bar, 100 mm. Arrows indicate tumor cell nests. Arrowheads indicate To address the possible cellular origin of the elevated tumor-associated leukocytes. B, EOC tissue array staining. Different EOC histotypes show expression of IL-18BP in tumor cells and infiltrating IL-18BP in EOC, we conducted immunochemical analyses leukocytes. C, normal ovary and Fallopian tube show virtually no on cells isolated from the ascites and on tissue arrays. positivity for IL-18BP staining. Negative controls of contiguous sections Among cells present in EOC ascites, tumor cell nests are shown for each panel.

www.aacrjournals.org Clin Cancer Res; 19(17) September 1, 2013 OF5

Carbotti et al.

"myeloid-derived suppressor cells" (32). Together, these sera, it was found in the ascites of nude mice bearing data indicate that different cell populations express IL- orthotopic xenotransplants of the human A2774 and 18BP in the EOC microenvironment. In contrast, normal SKOV3 cell lines (Fig. 4C), further suggesting that EOC ovary and Fallopian tube tissue showed no expression cells can contribute to IL-18BP production in vivo. Indeed, of IL-18BP by immunohistochemistry (Fig. 3C). A2774 and SKOV3 cells grown in immune-deficient mice showed IL-18BP expression by immunohistochemistry, IFN-g and IL-27 upregulate IL-18BP in human EOC cell whereas IL-18BP was virtually undetectable in the same lines EOC cell lines in vitro (Supplementary Fig. S4). These Unlike EOC cells present in tumor specimen, four EOC findings suggest that factor(s) present in the microenviron- cell lines showed no constitutive expression of IL-18BP ment are responsible for production of IL-18BP in vivo. mRNA or protein (Fig. 4). However, culture in the presence Although IFN-g mediates IL-18BP expression in EOC cell of IFN-g increased IL-18BP protein secretion (Fig. 4A) and lines in vitro, it should not be involved in vivo, as it was IL18BP mRNA (Fig. 4B) expression in EOC cell lines. In virtually undetectable in human ascites from patients and, addition, although human IL-18BP was undetectable in moreover, mouse IFN-g is inactive on human cells. These considerations prompted us to examine whether other cytokines, known to be elevated in patients with EOC, such as IL-6, TNF-a, VEGF-A, EGF, IL-18, and IL-8, could mediate IL-18BP expression in EOC cell lines, but none proved active (data not shown). A recent report indicated that the heterodimeric cytokine IL-27, consisting of EBI3 and IL-27 chains, could mediate IL-18BP production by human keratinocytes in an IFN-g–independent manner (33). We then tested whether IL-27 or the related cytokine IL-35 could mediate IL-18BP expression in EOC cells. Indeed, IL-27 inducedIL-18BPsecretion(Fig.5A;S5AandB)infour EOC cell lines in a dose-dependent fashion, whereas IL-35 showed no activity (not shown). IL-27 also increased IL18BP mRNA expression (Supplementary Fig. S5C) Importantly, the IL-18BP–containing supernatant of IL-27–stimulated A2780 cells significantly inhibited IL- 18 bioactivity in a concentration-dependent manner, as detected through IFN-g release by the human NK cell line NK-92. Controls such as IL-27–containing medium or the supernatant from unstimulated EOC cells produced no inhibition (Fig. 5B). Because IL-18BP expression is activated through the STAT1 pathway, we also analyzed STAT1 activation by IL- 27 in EOC cells. Indeed, IL-27 activates STAT1 signaling in EOC cell lines (Fig. 5C and Supplementary Fig. S5D), as reported for other IL-27–sensitive cell types (33, 34). In addition, Western blot analysis showed that STAT1 was constitutively tyrosine-phosphorylated in cells isolated from ascites ex vivo and was further activated by in vitro treatment with IL-27 (Fig. 5C). Confocal microscopy con- firmed constitutive STAT1 phosphorylation, which increased with IL-27 stimulation, in both EPCAM-positive EOC and EPCAM-negative inflammatory cells (Supplementary Fig. S5E). Consistently, cells isolated from ascites showed spontaneous IL-18BP secretion in culture, which could be further enhanced by in vitro IL-27 stimulation (Fig. 5A). Figure 4. IL-18BP expression by EOC cell lines. A, IL-18BP secretion is Because IL-27 activity is mediated through a heterodi- detected by ELISA in culture supernatants following EOC cell treatment meric receptor consisting of gp-130 and WSX-1 molecules, with IFN-g (100 and 1,000 U/mL for 24 hours. , P < 0.05; , P < 0.01 by we asked whether antibodies neutralizing gp-130 could in- two-sided unpaired Student t test). B, IFN-g–induced IL18BP mRNA hibit the effect of IL-27. Anti-gp130 mAb significantly expression in EOC cell lines as detected by RT-PCR analysis. C, Human IL-18BP ascites levels were measured in NOD/SCID mice bearing inhibited IL-27–mediated IL-18BP production in two dif- orthotopic xenotransplants of the SKOV3 (n ¼ 5) and A2774 (n ¼ 2) EOC ferent EOC cell lines, further supporting the involvement of cell lines. the IL-27/IL-27R pathway in IL-18BP regulation (Fig. 5A).

OF6 Clin Cancer Res; 19(17) September 1, 2013 Clinical Cancer Research

IL-18BP in Ovarian Cancer

Figure 5. Involvement of IL-27 in IL-18BP expression in EOC cells. A, IL-18BP secretion is detected by ELISA in culture supernatants following A2780 and A2774 EOC cell stimulation with IL-27 (20 ng/mL for 48 hours, P < 1E-05 vs. untreated control). Anti-gp130 mAb (from 0.5 to 500 ng/mL) significantly inhibits IL-18BP secretion mediated by IL-27 stimulation (, P < 0.05; , P < 0.001; P < 1E-05). Black bars show constitutive and IL-27–induced IL-18BP in vitro secretion by cells isolated from patients' ascites A98, representative of three cases with similar results. B, the þ conditioned medium of IL-27–stimulated A2780 cells containing 2 ng/mL IL-18BP (BP cm) significantly inhibits IL-18–induced IFN-g release by the NK-92 cells in a concentration (50, 25, or 12%v/v)-dependent fashion (, P < 0.05; , P < 0.001). As controls, conditioned medium of unstimulated cells (ctrl cm) and IL-27–containing medium (IL-27þ cm) were used. C, Western blot analysis of tyrosine-phosphorylated STAT1 protein in A2780 cells and in cells isolated from three different ascites unstimulated or stimulated for 10 minutes with IL-27 (20 ng/mL); b-actinwasusedasloading control. D, tumor-associated leukocytes show staining for IL-27A, whereas tumor cell nests show no reactivity (top). Negative control of a contiguous section is shown. Scale bar, 100 mm. Arrows indicate examples of negative tumor cell nests. E, EOC tissue array staining. Different EOC histotypes show expression of IL-27A in infiltrating leukocytes (enlarged in the inset).

IL-27A and EBI3 are expressed in EOC tissues gested a possible paracrine loop of IL-18BP induction Further analyses of two microarray datasets of EOC indi- in vivo. This hypothesis was also suggested by the use of cated a correlation between the expression of EBI3 and an anti-IL-27A (p28) speciﬁc antibody in immunochem- IL18BP mRNA in EOC primary tumors (Supplementary istry, which revealed IL-27 expression predominantly in a Fig. S6A). In addition, although IL18BP mRNA expression fraction of tumor-associated leukocytes isolated from the showed no signiﬁcant correlation with outcome (not ascites and in tissue microarrays, whereas tumor cell nests shown), high levels of EBI3 expression correlated with a appeared negative (Fig. 5C and D). Also, EBI3 protein shorter relapse-free survival (Supplementary Fig. S6B). The showed a similar distribution both in ascites (Supplemen- correlation between IL18BP and EBI3 gene expression sug- tary Fig. S7A) and within tumor tissues (Supplementary

www.aacrjournals.org Clin Cancer Res; 19(17) September 1, 2013 OF7

Carbotti et al.

Fig. S7B). Together, our data are consistent with a para- Evidence exists to suggest that IL-18BP production is a crineactivationofIL-18BPexpressioninthetumor result of the interaction between EOC cells and the micro- microenvironment. environment. Human EOC cell lines do not produce IL-18BP in culture but, once grafted in immune-deficient mice, they display IL-18BP expression, suggesting a role Discussion for factor(s), which function across the species. This is not In this study, we show that IL-18BP levels are elevated in the case for IFN-g, in view of its species specificity. Because the serum of patients with EOC and are even higher in the other cytokines, which are elevated in the ascites of EOC ascites, reaching four-fold higher levels than those found in failed to induce IL-18BP expression in vitro, we focused on normal serum. This finding suggested a local production of IL-27, which was recently shown to stimulate IL-18BP IL-18BP in the microenvironment of EOC. Indeed, immu- expression in human keratinocytes (33). nochemical analyses showed that IL-18BP is expressed by IL-27 is a member of the IL-12 family that may have pro- neoplastic cells of different EOC histotypes and by tumor- or anti-inflammatory properties in different systems associated leukocytes with myeloid features. Therefore, the (29, 30). IL-27 is a heterodimeric cytokine, composed of high concentration of IL-18BP in the tumor environment p28 and EBV-induced gene 3 (EBI3), which upregulates of EOC may limit the effect of endogenous or exogenously IL-12R expression and is relevant for TH1 polarization administered IL-18. On the other hand, IL-18BP also binds (41, 42). However, the precise contribution of IL-27 to the anti-inflammatory cytokine IL-37 (35), which may immune response, inflammation, and cancer is still poorly suppress the host immune response (36), and this may understood. On one hand, IL-27 has proinflammatory result in a beneficial effect for the host. However, to our effects through the induction of CXCL10 in macrophages knowledge, no evidence for IL-37 expression in ovarian in inflammatory skin disorders (43). However, IL-27 cancer has been provided to date. may limit the proinflammatory and immune-enhancing Elevated IL-18BP levels were recently described also in activities of IL-18 in the skin through IL-18BP induction serum of patients with prostatic (24) and pancreatic cancer (33) and may dampen autoimmunity, as Il27 / mice (37). In the latter, the concomitantly elevated levels of free- were more susceptible to experimental autoimmune en- IL-18 in the serum suggested the existence of a biologic cephalomyelitis (44). paradox, in view of the immune-enhancing properties of Here, we show that IL-27 induces the expression of IL-18. Indeed, increased levels of both IL-18 and its natural IL-18BP mRNA and protein in human EOC cell lines in inhibitor IL-18BP were also found in patients with systemic culture and activates STAT1 signaling in these cells, whereas lupus erythematosus, in whom biologically active free IL-18 IL-35, another EBI3-containing cytokine (30), was inactive. was still higher than in controls and was a marker of disease Such activity was specifically induced through the IL-27R activity and a potential contributor to autoimmunity (38). complex as indicated by the significant inhibition of It was previously shown that high levels of IL-18 are IL-18BP induction upon treatment with a neutralizing present in serum and ascites of patients with EOC (12) antibody against the gp130 chain. A potential role of and that pro-IL-18 is largely predominant (13). In fact, IL-27 in vivo was suggested by the expression of IL-27A and although IL-18 ELISA preferentially recognized mat-IL-18 EBI3 found by immunochemistry in tumor-associated leu- in sera, this assay also detected pro-IL-18, albeit with a kocytes in both ascites and tumor tissues and by the cor- reduced sensitivity. Therefore, the presence of "free IL-18" relation between EBI3 and IL18BP mRNA expression in in EOC could be explained by the predominance of pro-IL- two different EOC datasets. Interestingly, high EBI3 expres- 18, which is unable to bind IL-18BP and is detected by IL-18 sion correlated with a shorter progression-free survival in ELISA. It is likely that a similar situation may occur in other type II tumors. The finding that IL18BP gene expression tumors in which alterations of IL-18 processing have been had no significant correlation with relapse-free survival of reported (39). patients with type II tumors may reflect the multiplicity of Because the presence of mat-IL-18 in the ascites of EOC components driving clinical outcome. A correlation with could not be formally excluded and mat-IL-18 is an inducer IL27A mRNA expression could not be found (data not of IL-18BP, via IFN-g production (25, 40), we explored shown), but this may relate to technical limitations, as only possible correlations between immune-reactive IL-18 and one probeset was present in the arrays. The possible role of IL-18BP or IFN-g levels. No significant correlation was IL-27 in vivo was reinforced by the detection of constitutive found in the ascites and serum of patients with EOC, and STAT1 activation in both neoplastic and reactive cells iso- IFN-g levels were very low to undetectable in the ascites. lated from the ascites, in the absence of measurable IFN-g However, a correlation between IL18 and IL18BP mRNA levels. Moreover, these ascites cells showed spontaneous levels was found in two independent datasets of EOC gene secretion of IL-18BP in culture, which could be further expression profiles. Moreover, IFNG mRNA showed a cor- enhanced by the addition of exogenous IL-27. Consistently, relation with IL18BP in one dataset, although IFNG mRNA STAT1 phosphorylation also was increased by IL-27. was elevated only in a minority of cases. These data suggest Our study may open new perspectives on understanding that the IL-18/IFN-g loop may be active in the tumor tissue the role of IL-27 in cancer, as its involvement in the microenvironment in some patients and that other factor(s) antitumor immune response is still poorly understood. In may participate in the induction of IL-18BP expression. some hematologic neoplasia, including multiple myeloma

OF8 Clin Cancer Res; 19(17) September 1, 2013 Clinical Cancer Research

IL-18BP in Ovarian Cancer

(45) and acute leukemias (46, 47), IL-27 displays direct and Disclosure of Potential Conflicts of Interest indirect antitumor effects. We thus hypothesize that in EOC No potential conflicts of interest were disclosed. IL-27 may be part of an immune-regulatory network, which limits the induction of TH1 responses and IFN-g production Authors' Contributions in the microenvironment by inhibiting IL-18 activity. In Conception and design: M. Fabbi, S. Ferrini Development of methodology: G. Carbotti, G. Barisione, A. Brizzolara support of this concept, studies in murine models highlight- Acquisition of data (provided animals, acquired and managed patients, ed a predominant role of IL-27 as an immune-regulatory provided facilities, etc.): G. Carbotti, A.M. Orengo, A. Brizzolara, M.G. Centurioni and anti-inflammatory agent that generates and maintains Analysis and interpretation of data (e.g., statistical analysis, biosta- T-regulatory cell functions (48) and induces IL-10 produc- tistics, computational analysis): I. Airoldi, M. Bagnoli, P. Pinciroli, tion by T lymphocytes (49). D. Mezzanzanica, M. Fabbi Writing, review, and/or revision of the manuscript: G. Carbotti, I. Airoldi, Although the role of IL-18 in tumor cell biology has M. Bagnoli, P. Pinciroli, D. Mezzanzanica, M. Fabbi, S. Ferrini been debated (50), preclinical studies indicated that IL-18 Administrative, technical, or material support (i.e., reporting or orga- displays antitumor activity through its ability to trigger nizing data, constructing databases): I. Airoldi Study supervision: M. Fabbi, S. Ferrini IFN-g production and to favor the induction of a TH1 response (18, 19). Therefore, recombinant IL-18 is under- going testing in clinical trials of cancer immunotherapy Grant Support This study was supported by AIRC (Associazione Italiana per la Ricerca sul (27, 28) and in particular a clinical phase I study of IL-18 Cancro, IG5509, IG13018 and IG13518), Ministry of Health Project and doxorubicin in advanced-stage EOC (NCT00659178) "Tumori Femminili," and Compagnia di San Paolo. The costs of publication of this article were defrayed in part by the has recently concluded patient recruitment. It is possible payment of page charges. This article must therefore be hereby marked that the high local levels of IL-18BP present in EOC may advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate limit the biologic effects of low levels of endogenous IL- this fact. 18 or of therapeutically administered IL-18, particularly Received February 27, 2013; revised June 21, 2013; accepted July 8, 2013; at the tumor site. published OnlineFirst July 26, 2013.

References 1. Martin LP, Schilder RJ. Management of recurrent ovarian carcinoma: 12. Le Page C, Ouellet V, Madore J, Hudson TJ, Tonin PN, Provencher DM, current status and future directions. Semin Oncol 2009;36:112–25. et al. From gene profiling to diagnostic markers: IL-18 and FGF-2 2. Ferrini S, Biassoni R, Moretta A, Bruzzone M, Nicolin A, Moretta L. complement CA125 as serum-based markers in epithelial ovarian Clonal analysis of T lymphocytes isolated from ovarian carcinoma cancer. Int J Cancer 2006;118:1750–8. ascitic fluid. Phenotypic and functional characterization of T-cell 13. Orengo AM, Fabbi M, Miglietta L, Andreani C, Bruzzone M, Puppo A, clones capable of lysing autologous carcinoma cells. Int J Cancer et al. Interleukin (IL)-18, a biomarker of human ovarian carcinoma, is 1985;36:337–43. predominantly released as biologically inactive precursor. Int J Cancer 3. Nelson BH. The impact of T-cell immunity on ovarian cancer out- 2011;129:1116–25. comes. Immunol Rev 2008;222:101–16. 14. Okamura H, Tsutsui H, Komatsu T, Yutsudo M, Hakura A, Tanimoto T, 4. Frumento G, Piazza T, Di Carlo E, Ferrini S. Targeting tumor-related et al. Cloning of a new cytokine that induces interferon-gamma. Nature immunosuppression for cancer immunotherapy. Metab Immune Dis- 1995;378:88–91. ord Drug Targets 2006;6:233–7. 15. Takeda K, Tsutsui H, Yoshimoto T, Adachi O, Yoshida N, Kishimoto T, 5. Preston CC, Goode EL, Hartmann LC, Kalli KR, Knutson KL. Immunity et al. Defective NK cell activity and Th1 response in IL-18-deficent and immune suppression in human ovarian cancer. Immunotherapy mice. Immunity 1998;8:383–90. 2011;3:539–56. 16. Dinarello CA. IL-18: a Th1-inducing, proinflammatory cytokine and 6. Nilsson MB, Langley RR, Fidler IJ. Interleukin-6, secreted by human new member of the IL-1 family. J Allergy Clin Immunol 1999;103: ovarian carcinoma cells, is a potent proangiogenic cytokine. Cancer 11–24. Res 2005;65:10794–800. 17. Ghayur T, Banerjee S, Hugunin M, Butler D, Herzog L, Carter A, 7. Leinster DA, Kulbe H, Everitt G, Thompson R, Perretti M, Gavins FN, et al. Caspase-1 processes IFN-gamma-inducing factor and reg- et al. The peritoneal tumour microenvironment of high-grade serous ulates LPS-induced IFN-gamma production. Nature 1997;386: ovarian cancer. J Pathol 2012;227:136–45. 619–23. 8. Berek JS, Chung C, Kaldi K, Watson JM, Knox RM, Martinez-Maza O. 18. Heuer JG, McClung CT, Hock RA. Neuroblastoma cells expressing Serum interleukin-6 levels correlate with disease status in patients with mature IL-18, but not pro-IL-18, induce a strong and immediate epithelial ovarian cancer. Am J Obstet Gynecol 1991;164:1038–43. antitumour immune response. J Immunother 1999;22:324–35. 9. Duan Z, Feller AJ, Penson RT, Chabner BA, Seiden MV. Discovery of 19. Nakamori M, Iwahashi M, Nakamura M, Ueda K, Zhang X, Yamaue H. differentially expressed genes associated with paclitaxel resistance Intensification of antitumor effect by T helper 1-dominant adoptive using cDNA array technology: analysis of interleukin (IL) 6, IL-8, and immunogene therapy for advanced orthotopic colon cancer. Clin monocyte chemotactic protein 1 in the paclitaxel-resistant phenotype. Cancer Res 2003;9:2357–65. Clin Cancer Res 1999;5:3445–53. 20. Wang ZY, Gaggero A, Rubartelli A, Rosso O, Miotti S, Mezzanzanica D, 10. Curiel TJ, Coukos G, Zou L, Alvarez X, Cheng P, Mottram P, et al. et al. Expression of interleukin-18 in human ovarian carcinoma and Specific recruitment of regulatory T cells in ovarian carcinoma fosters normal ovarian epithelium: evidence for defective processing in tumor immune privilege and predicts reduced survival. Nat Med 2004;10: cells. Int J Cancer 2002;98:873–8. 942–9. 21. Novick D, Kim SH, Fantuzzi G, Reznikov LL, Dinarello CA, Rubinstein 11. Lambeck AJ, Crijns AP, Leffers N, Sluiter WJ, ten Hoor KA, Braid M, M. Interleukin-18 binding protein: a novel modulator of the Th1 cyto- et al. Serum cytokine profiling as a diagnostic and prognostic tool in kine response. Immunity 1999;10:127–36. ovarian cancer: a potential role for interleukin 7. Clin Cancer Res 22. Kim SH, Eisenstein M, Reznikov L, Fantuzzi G, Novick D, Rubinstein M, 2007;13:2385–91. et al. Structural requirements of six naturally occurring isoforms of

www.aacrjournals.org Clin Cancer Res; 19(17) September 1, 2013 OF9

Carbotti et al.

the IL-18 binding protein to inhibit IL-18. Proc Natl Acad Sci U S A free IL-18 are correlated with poor survival. J Immunother 2009;32: 2000;97:1190–5. 920–31. 23. Veenstra KG, Jonak ZL, Trulli S, Gollob JA. IL-12 induces monocyte IL- 38. Migliorini P, Anzilotti C, Pratesi F, Quattroni P, Bargagna M, Dinar- 18 binding protein expression via IFN-gamma. J Immunol 2002;168: ello CA, et al. Serum and urinary levels of IL-18 and its inhibitor IL- 2282–7. 18BP in systemic lupus erythematosus. Eur Cytokine Netw 2010; 24. Fujita K, Ewing CM, Isaacs WB, Pavlovich CP. Immunomodulatory IL- 21:264–71. 18 binding protein is produced by prostate cancer cells and its levels in 39. Pages F, Berger A, Henglein B, Piqueras B, Danel C, Zinzindohoue F, urine and serum correlate with tumor status. Int J Cancer 2011;129: et al. Modulation of interleukin-18 expression in human colon carci- 424–32. noma: consequences for tumour immune surveillance. Int J Cancer 25. Paulukat J, Bosmann M, Nold M, Garkisch S, Kampfer€ H, Frank S, et al. 1999;84:326–30. Expression and release of IL-18 Binding Protein in Response to INF-g. 40. Fantuzzi G, Reed D, Qi M, Scully S, Dinarello CA, Senaldi G. Role of J Immunol 2001;167:7038–43. interferon regulatory factor-1 in the regulation of IL-18 production and 26. Dinarello CA, Novick D, Rubinstein M, Lonnemann G. Interleukin 18 activity. Eur J Immunol 2001;31:369–75. and interleukin 18 binding protein: possible role in immunosuppres- 41. Pflanz S, Timans JC, Cheung J, Rosales R, Kanzler H, Gilbert J, et al. sion of chronic renal failure. Blood Purif 2003;21:258–70. IL-27, a heterodimeric cytokine composed of EBI3 and p28 protein, 27. Robertson MJ, Mier JW, Logan T, Atkins M, Koon H, Koch KM, et al. induces proliferation of naive CD4(þ) T cells. Immunity 2002;16: Clinical and biological effects of recombinant human interleukin-18 779–90. administered by intravenous infusion to patients with advanced 42. Lucas S, Ghilardi N, Li J, de Sauvage FJ. IL-27 regulates IL-12 cancer. Clin Cancer Res 2006;12:4265–73. responsiveness of naive CD4þ T cells through Stat1-dependent and 28. Robertson MJ, Kirkwood JM, Logan TF, Koch KM, Kathman S, Kirby -independent mechanisms. Proc Natl Acad Sci U S A 2003;100: LC, et al. A dose-escalation study of recombinant human interleukin- 15047–52. 18 using two different schedules of administration in patients with 43. Shibata S, Tada Y, Kanda N, Nashiro K, Kamata M, Karakawa M, et al. cancer. Clin Cancer Res 2008;14:3462–9. Possible roles of IL-27 in the pathogenesis of psoriasis. J Invest 29. Trinchieri G, Pflanz S, Kastelein RA. The IL-12 family of heterodimeric Dermatol 2010;130:1034–9. cytokines: new players in the regulation of T cell responses. Immunity 44. Batten M, Li J, Yi S, Kljavin NM, Danilenko DM, Lucas S, et al. 2003;19:641–4. Interleukin 27 limits autoimmune encephalomyelitis by suppressing 30. Vignali DA, Kuchroo VK. IL-12 family cytokines: immunological play- the development of interleukin 17-producing T cells. Nat Immunol makers. Nat Immunol 2012;13:722–8. 2006;7:929–36. 31. Kurman RJ, Shih IeM. Molecular pathogenesis and extraovarian origin 45. Cocco C, Giuliani N, Di Carlo E, Ognio E, Storti P, Abeltino M, et al. of epithelial ovarian cancer-Shifting the paradigm. Hum Pathol 2011; Interleukin-27 acts as multifunctional antitumor agent in multiple 42:918–31. myeloma. Clin Cancer Res 2010;16:4188–97. 32. Ostrand-Rosenberg S, Sinha P. Myeloid-derived suppressor cells: 46. Zorzoli A, Di Carlo E, Cocco C, Ognio E, Ribatti D, Ferretti E, et al. linking inflammation and cancer. J Immunol 2009;182:4499–506. Interleukin-27 inhibits the growth of pediatric acute myeloid leu- 33. Wittmann M, Doble R, Bachmann M, Pfeilschifter J, Werfel T, Muhl€ H. kemia in NOD/SCID/Il2rg/ mice. Clin Cancer Res 2012;18: IL-27 regulates IL-18 binding protein in skin resident cells. PLoS ONE 1630–40. 2012;7:e38751. 47. Canale S, Cocco C, Frasson C, Seganfreddo E, Di Carlo E, Ognio 34. Bachmann M, Paulukat J, Pfeilschifter J, Muhl€ H. Molecular mechan- E, et al. Interleukin-27 inhibits pediatric B-acute lymphoblastic isms of IL-18BP regulation in DLD-1 cells: pivotal direct action of leukemia cell spreading in a preclinical model. Leukemia 2011; the STAT1/GAS axis on the promoter level. J Cell Mol Med 2009;13: 25:1815–24. 1987–94. 48. Wojno ED, Hosken N, Stumhofer JS, O'Hara AC, Mauldin E, Fang Q, 35. Bufler P, Azam T, Gamboni-Robertson F, Reznikov LL, Kumar S, et al. A role for IL-27 in limiting T regulatory cell populations. J Immunol Dinarello CA, et al. A complex of the IL-1 homologue IL-1F7b and 2011;187:266–73. IL-18-binding protein reduces IL-18 activity. Proc Natl Acad Sci U S A 49. Awasthi A, Carrier Y, Peron JP, Bettelli E, Kamanaka M, Flavell RA, et al. 2002;99:13723–8. A dominant function for interleukin 27 in generating interleukin 10- 36. Nold MF, Nold-Petry CA, Zepp JA, Palmer BE, Bufler P, Dinarello CA. producing anti-inflammatory T cells. Nat Immunol 2007;8:1380–9. IL-37 is a fundamental inhibitor of innate immunity. Nat Immunol 50. Vidal-Vanaclocha F, Mendoza L, Telleria N, Salado C, Valcarcel M, 2010;11:1014–22. Gallot N, et al. Clinical and experimental approaches to the patho- 37. Carbone A, Vizio B, Novarino A, Mauri FA, Geuna M, Robino C, et al. physiology of interleukin-18 in cancer progression. Cancer Metastasis IL-18 paradox in pancreatic carcinoma: elevated serum levels of Rev 2006;25:417–34.

OF10 Clin Cancer Res; 19(17) September 1, 2013 Clinical Cancer Research

The IL-18 Antagonist IL-18−Binding Protein Is Produced in the Human Ovarian Cancer Microenvironment

Grazia Carbotti, Gaia Barisione, Anna Maria Orengo, et al.

Clin Cancer Res Published OnlineFirst July 19, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-13-0568

Supplementary Access the most recent supplemental material at: Material http://clincancerres.aacrjournals.org/content/suppl/2013/07/23/1078-0432.CCR-13-0568.DC1

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 Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2013/08/12/1078-0432.CCR-13-0568. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.