The clinical significance of MAGEA3 expression in pancreatic cancer

Joseph Kim1,2, Howard A. Reber

4, Oscar J. Hines

4, Kevork K. Kazanjian

4, Andy Tran

1, Xing Ye

3, Farin F. Amersi

1,2,

Steve R. Martinez1,2, Sarah M. Dry5, Anton J. Bilchik2 and Dave S.B. Hoon1*

1Gastrointestinal Research Section, Department of Molecular Oncology, John Wayne Cancer Institute, Saint John’s Health Center,Santa Monica, CA,USA2Division of Surgical Oncology, John Wayne Cancer Institute, Saint John’s Health Center, Santa Monica, CA, USA3Division of Biostatistics, John Wayne Cancer Institute, Saint John’s Health Center, Santa Monica, CA, USA4Section of Gastrointestinal Surgery, Department of Surgery, David Geffen School of Medicine, University of California,Los Angeles, CA, USA5Department of Pathology, David Geffen School of Medicine, University of California, Los Angeles, CA, USA

The MAGEA gene family that encodes cancer testis antigens isdifferentially expressed in many cancers. Though MAGEA3 ex-pression has been detected in gastrointestinal malignancies, itsrole in pancreatic ductal adenocarcinoma (PDAC) has not beenwell established. We assessed 57 patients who underwent intent-to-cure surgery for PDAC. Total RNA from paraffin-embeddedpancreatic tumors was extracted and assessed for MAGEA3 geneexpression by an optimized probe-based quantitative real-timeRT-PCR (qRT) assay. MAGEA3 gene expression was detected byqRT in 25 (44%) patients. For the entire cohort, detection ofMAGEA3 expression was associated with significantly decreasedoverall survival (median, 16 vs 33 months; log-rank, p 5 0.032).When clinicopathologic factors, including age, gender, stage,tumor extent, lymph node metastasis, tumor grade, perineuralinvasion and lymphovascular invasion were assessed by univariateanalysis, MAGEA3 gene expression and tumor grade were signifi-cant prognostic factors for poor survival (HR 2.1, 95% CI: 1.0–4.4, p 5 0.041; and HR 3.7, 95% CI: 1.8–7.6, p 5 0.0004, respec-tively). Immunohistochemistry (IHC) was performed and con-firmed MAGEA3 protein in PDAC specimens. In conclusion,MAGEA3 is differentially expressed in patients with PDAC; itsexpression correlates with significantly worse survival. Molecularassessment for MAGEA3 should be considered to improve prog-nostic evaluation and to identify eligible patients for potentialimmune-based therapy.' 2005 Wiley-Liss, Inc.

Key words: cancer testis antigen; quantitative real-time RT-PCR;MAGEA3; pancreatic cancer

Pancreatic ductal adenocarcinoma (PDAC) has been a leadingcause of cancer-related mortality, in part, due to ineffective treat-ment options. Although efforts to characterize the molecular fea-tures of PDAC have identified dysregulated signaling and tran-scriptional pathways (e.g., KRAS and DPC4/SMAD4 mutations,respectively)1,2 that promote aggressive properties in pancreaticcancer, therapies to abrogate or block such pathways have beendisappointingly ineffective. This may reflect nonspecific targetingof regulated, benign cells that share identical signaling pathwayswith cancer cells. Such may be a limitation for recently developedtyrosine kinase inhibitors.3 One potential method to bypass theseconstraints is to identify pancreatic tumor antigens that are notexpressed on benign cells.

The MAGE family of cancer testis antigens (MAGEA, MAGEBand MAGEC) are unique tumor markers that are expressed nor-mally only in the placenta and male germ cells.4–6 The remainingmembers of the human MAGE family, MAGED and necdin, areubiquitously expressed.7 The MAGE antigens arise from tran-scription of genes that belong to a family of more than 20 closelyrelated genes located on chromosome X.8–15 Although originallyidentified in melanoma, the MAGE gene is commonly expressedin various tumors of epithelial origin, including breast, lung andcolorectal carcinomas.16–19 The MAGEA subfamily, which iscomprised of 12 genes, has had one or more of its antigensdetected in various gastrointestinal malignancies, including pan-creatic cancer.20–26

We have previously assessed MAGEA3 gene expression in mela-noma and other cancers and have found it to be a specific molecularbiomarker for cancer cells.27–30 However, DNA sequences ofMAGEA3 and MAGEA6 are almost identical and, therefore, RT-PCR assays may not discriminate between the 2 MAGEA var-iants.31 We hypothesized that, as an antigen unique primarily tocancer cells and expressed in gastrointestinal malignancies,MAGEA3 may be expressed in pancreatic cancer. A recent reportexamined MAGEA3 expression in pancreatic cancer, albeit by gel-based RT-PCR methods.26 Here, we assessed PDAC specimensfrom patients who underwent surgical resection and sought to deter-mine patterns of MAGEA3 gene expression by quantitative real-time RT-PCR (qRT) and whether differential patterns of expressioncorrelated with clinical outcomes.

Material and methods

Patients and resources

A multi-institutional cohort of patients (n 5 57) was evaluatedfor this study. Patients were accrued from John Wayne CancerInstitute (JWCI), UCLA School of Medicine and the CooperativeHuman Tissue Network (CHTN). All patients underwent intent-to-cure surgery for PDAC between the years 1996 and 2004.Benign and matched normal pancreas specimens (n 5 15) werealso obtained from all 3 sites. During this time interval PDACpatients were offered various adjuvant chemotherapy regimensconsisting of 5-fluorouracil, leucovorin, mitomycin C, dipyrida-mole or gemcitabine. Adjuvant radiation therapy was offered topatients at the discretion of their treating physicians. A HumanSubjects Institutional Review Board (IRB) approval was obtainedfor the purposes of this study at the participating institutions.Informed consent waivers were obtained for all patients in thisstudy to allow collection of retrospective clinical data and to con-duct an analysis of archived paraffin-embedded specimens.

RNA isolation

Eight established pancreatic cancer cell lines (MIA PaCa-2,PANC-1, Capan-1, BxPC-3, COLO357, CFPAC-1, AsPC1 and Hs766T) were obtained from the American Type Culture Collection(American Type Culture Collection, ATCC, Manassas, VA) and

Grant sponsors: Harold J. McAlister Foundation; Martin H. WeilResearch Laboratories, John Wayne Cancer Institute; Hirshberg Founda-tion for Pancreatic Cancer Research, UCLA School of Medicine, LosAngeles, CA, USA.*Correspondence to: Dr. D.S.B. Hoon, Department of Molecular

Oncology, John Wayne Cancer Institute, 2200 Santa Monica Blvd., SantaMonica, CA, 90404, USA. Fax:11-310-449-5282, 11-310-449-5261.E-mail: hoon@jwci.orgReceived 22 July 2005; Accepted after revision 27 September 2005DOI 10.1002/ijc.21656Published online 5 December 2005 in Wiley InterScience (www.

interscience.wiley.com).

Int. J. Cancer: 118, 2269–2275 (2006)' 2005 Wiley-Liss, Inc.

Publication of the International Union Against Cancer

maintained as recommended. The immortalized normal humanpancreas ductal epithelial (HPDE) cell line32,33 was kindly pro-vided by Dr. Tsao (University of Toronto, Toronto, Ontario, Can-ada) and was used as a negative cell line control for MAGE-A3gene expression. All cells were incubated at 37�C with 5% CO2.Total RNA from cell lines was extracted, isolated and purifiedusing Tri-Reagent (Molecular Research Center, Cincinnati, OH)as previously described.28,29

Paraffin-embedded archival tissue (PEAT) blocks of PDACspecimens were reviewed by a surgical pathologist to confirm thediagnosis of PDAC and ensure the presence of tumor in the tissueblocks. Additional paraffin blocks of tumor from the same patientwere unavailable due to procedural restrictions of tumor procure-ment. Therefore, potential tumor heterogeneity was not assessed.Also, tumor blocks were not microdissected because MAGEA3 isnot expressed in normal tissues other than placenta or male germcells.4–6

The paraffin blocks were sectioned (20 lm) under RNAse-freeconditions and placed in sterile microcentrifuge tubes (Eppendorf,Westbury, NY). After deparaffinization with xylene and washingswith 100% ethanol, the specimens were treated with a proteinaseK digestion buffer for 3 hr. Total RNA was extracted, isolated andpurified using a modification of the RNAWiz (Ambion, Austin,TX) phenol–chloroform extraction method as previously des-cribed.34 RNA was quantified and assessed for purity by ultravio-let spectrophotometry and RIBOGreen detection assay as previ-ously described (Molecular Probes, Eugene, OR).30

Primers and RT-PCR

Primer and probe sequences were designed as previously des-cribed.35 Specific primers were designed to amplify sequentialexon–exon regions to avoid potential amplification of contami-nating genomic DNA and to produce amplicon sizes less than150 bp to optimally amplify PEAT RNA. The primers and FRETprobe sequences used were: MAGEA3: 50-AGGAGAAGATCT-GCCAGTGG-30 (forward); 50-AGTGCTGACTCCTCTGCTCA-30(reverse); and 50-FAM-AGCTCCTGCCCACACTCCCGCCTG-T-BHQ-1-30 (FRET probe). Glyceraldehyde-3-phosphate dehydro-genase (GAPDH): 50-GGGTGTGAACCATGAGAAGT-30 (forward);50-GACTGTGGTCATGAGTCCT-30 (reverse); and 50-FAM-CAG-CAATGCC TCCTGCACCACCAA-BHQ-1-30 (FRET probe). PCRproducts were assessed by gel electrophoresis to confirm ampliconsizes.

Reverse transcription of total RNA was performed using Molo-ney Murine Leukemia Virus RT (Promega, Madison, WI). Toensure more robust cDNA production, Oligo dT (Gene Link, Haw-thorne, NY) and random hexamers (Roche, Indianapolis, IN) wereadded to the PEAT reaction mixtures as previously described.30

The qRT assay was performed with the iCycler iQ RealTime PCRDetection System (Bio-Rad Laboratories, Hercules, CA) using250 ng of total RNA for each reaction.

Each PCR reaction was performed with 1 lM of each primer,200 lM each deoxynucleotide triphosphate, 4.0 mM MgCl2, 103AmpliTaq Buffer and 1 U of AmpliTaq Gold Polymerase (AppliedBiosystems, Branchburg, NJ) and was subjected to 45 cycles at95�C for 60 sec, 58�C for 60 sec and 72�C for 60 sec forMAGEA3; 45 cycles at 95�C for 60 sec, 55�C for 60 sec and 72�Cfor 60 sec for housekeeping gene, GAPDH. Each sample wasassayed in triplicate by qRT and cDNA derived from cancer celllines and volunteer lymphocytes served as positive and negativecontrols, respectively, for the MAGEA3 qRT assay. The expres-sion of the housekeeping gene GAPDH was assessed in all pancre-atic cancer specimens to verify mRNA integrity. Only specimenswith adequate RNA (positive MAGEA3 expression or adequateGAPDH gene expression, i.e., copy numbers �1,000) were in-cluded in the study.

Immunohistochemistry

Immunohistochemistry (IHC) was performed to confirm thetranslation of MAGEA3 mRNA to protein in PDAC specimens.Tumor blocks were sectioned (5 lm), dried overnight at 37�C andthen deparaffinized with xylene. The sections were treated with anantigen retrieval solution (Target Retrieval, DakoCytomation,Carpinteria, CA) at 95�C for 15 min, cooled to room temperatureand then treated with dilute hydrogen peroxide to block endoge-nous peroxidase as previously described.35 Nonspecific antibodybinding was diminished with 5% milk. The sections were in-cubated overnight at 4�C with a polyclonal rabbit anti-humanMAGEA3 antibody (Abgent Inc., San Diego, CA) at a dilution of1:100. We note that this anti-MAGEA3 antibody may cross-react

FIGURE 1 – MAGEA3 gene expression in 8 pancreatic cancer celllines and one normal pancreas ductal epithelial cell line. MAGEA3expression is plotted as a ratio to the expression of the housekeepinggene GAPDH.

TABLE I – PANCREATIC CANCER DEMOGRAPHICS

Clinicopathologic factors N

Total patients 57Men 30Women 27

Age (yr)<50 350–70 27>70 27

UICC/TNM stagespI 18pII 39

T-stagepT1 9pT2 41pT3 7

Lymph node metastasispN0 21pN1 36

Pathologic gradeWell 9Moderate 25Poor 23

Tumor size (cm)<2 82–5 44>5 5

Perineural invasionNone 18Present 39

Lymphovascular invasionNone 41Present 16

2270 KIM ET AL.

with MAGEA6 because of the protein sequence homologybetween the 2 variants. The next day, sections were labeled with asecondary Link-Streptavidin HRP solution (Dako), developedwith diaminobenzidine (DAB), counterstained with hematoxylinand examined at 2003 and 4003 magnifications.

Statistical analysis

Patient characteristics andMAGEA3 expression were summarizedusing mean, median and frequency. Clinicopathologic factors ofMAGEA3 negative and positive patients were compared by Stu-dent’s t-test and Fisher’s Exact test. Overall survival curves withrespect to MAGEA3 expression were compared using Kaplan–Meier’s method. The log-rank test was used to compare the equalityof the 2 curves. Univariate analysis of prognostic factors includingage, gender, stage, tumor extent, lymph node metastasis, grade, size,perineural invasion and lymphovascular invasion was assessed. Amultivariate analysis using the Cox proportional hazard regressionmodel was also performed to evaluate the prognostic significance ofMAGEA3 expression when clinical prognostic factors were adjusted.A stepwise method was chosen for covariate selection. The relation-ship of MAGEA3 expression with tumor grade was determined byChi square analysis. All analyses were performed using SAS (SAS /STAT User’s Guide, version 8; SAS Institute Inc, Cary, NC) andtests were 2-sided with significance level of p < 0.05.

Results

MAGEA3 gene expression in control specimens

In 8 of 8 established pancreas cancer cell lines (MIA PaCa-2,PANC-1, Capan-1, BxPC-3, COLO357, CFPAC-1, AsPC1 and Hs766T), MAGEA3 gene expression was detected by qRT. All but 2of the lines (PANC-1 and BxPC-3) were derived from metastaticpancreatic cancers and only one (PANC-1) was derived from theexocrine ducts. MAGEA3 expression was undetectable by qRT in

the normal control line HDPE (Fig. 1). BxPC3, COLO357 and Hs766T had the highest levels of MAGEA3 gene expression relativeto GAPDH expression. Additionally, mRNA from the peripherallymphocytes of healthy volunteers (n 5 5) were negative forMAGEA3 expression by qRT. Matched benign pancreas speci-mens (n 5 10) and pancreas tissue with chronic pancreatitis (n 55) were obtained and analyzed. In 14 of 15 (94%) specimens,MAGEA3 expression was not detected. One pancreas specimenfrom a patient with chronic pancreatitis was positive for MAGEA3gene expression.

MAGE-A3 expression in patients with pancreatic cancer

Fifty-seven patients were assessed for MAGEA3 gene expres-sion by qRT; the patient demographics are listed in Table I.Because of our use of paraffin-embedded samples, we assessed forMAGEA3 expression and stratified specimens into a group with(1) expression of any level and a group with (2) expression. Weutilized zero copy number as the cut-off between (1/2) expres-sion. This cut-off was determined by generating standard PCRcurves using serially diluted cDNA standard templates forMAGEA3 and using the threshold cycle (Ct) of templates withknown numbers of copies. We utilized the optimized thresholdcycle number that would amplify at least 1 mRNA copy numberof MAGEA3. At 45 cycles, our MAGEA3 positive controls (pan-creas cancer cell lines) were PCR (1) and our negative controls(normal pancreas ductal epithelial cell line and volunteer lympho-cytes from healthy volunteers) were PCR (2). Accordingly,MAGEA3 gene expression was recorded as a binary result (1/2)for all PDAC specimens.

MAGEA3 gene expression was detected in the PDAC specimensof 25 (44%) patients. All 57 patients were then stratified into2 groups based on MAGEA3 expression. Clinicopathologic fac-tors of MAGEA3 negative and positive patients were compared(Table II). All patients in the cohort had negative surgical marginsby H&E. The groups were similar for all other factors except forPDAC tumor grade, with a higher number of differentiated tumorsin the MAGEA3 negative group. Kaplan–Meier curves were con-structed to determine whether MAGEA3 gene expression corre-lated with survival in these groups. A significant difference inoverall survival was present between patients with MAGEA3 posi-tive vs negative gene expression (median 16 vs 33 months, respec-tively; log-rank, p 5 0.032) (Fig. 2). During a median follow-uptime of 15 months, 35 (61%) patients had succumbed to disease.

To determine the prognostic significance of MAGEA3 geneexpression, clinicopathologic data, including age, gender, stage,tumor extent, lymph node metastasis, tumor grade and size, peri-

TABLE II – COMPARISON OF CLINICOPATHOLOGIC FACTORS BETWEENMAGEA3 NEGATIVE AND POSITIVE PATIENTS

Clinicopathologic factorsMAGEA3

p-value1Negative(n 5 32)

Positive(n5 25)

Age (yr) 0.58Mean 6 SD 676 12 696 9Range 42–90 54–84

Gender 1.0Female 15 12Male 17 13

UICC/TNM stage 0.15pI 13 5pII 19 20

Primary tumor 0.61pT1 5 4pT2 22 19pT3 5 2

Lymph node metastasis 0.27pN0 14 7pN1 18 18

Tumor size (cm) 0.430–2 7 4>2–5 23 18>5 2 3

Pathologic grade 0.028Well 7 2Moderate 16 9Poor 9 14

Perineural invasion 0.78Absent 11 7Present 21 18

Lymphovascular invasion 0.14Absent 26 15Present 6 10

1Comparison for age was performed by Student’s t-test, all othercomparisons were assessed by Fisher’s Exact test.

FIGURE 2 – Kaplan–Meier curves showing the difference in sur-vival between patients with MAGEA3 negative and positive pancreaticcancers.

2271MAGEA3 EXPRESSION IN PANCREATIC CANCER

neural invasion and lymphovascular invasion were compared byunivariate analysis for the 2 groups (Table III). By univariate anal-ysis,MAGEA3 and tumor grade were significant prognostic factorsfor poor survival (HR 2.1, 95% CI: 1.0–4.4, p 5 0.041; and HR3.7, 95% CI: 1.8–7.6, p 5 0.0004, respectively). On multivariateanalysis, only poorly-differentiated histologic tumor grade re-mained a significant prognostic factor for poor survival (HR 6.6,95% CI: 1.8–23.6, p 5 0.0007). Lymph node metastasis, a prog-nostic factor for poor survival in large cohort studies was not sig-nificant here.36,37

Immunohistochemistry

IHC was performed on representative PEAT sections ofMAGEA3 gene expression positive and negative PDAC specimens(n 5 14). IHC of specimens having absent MAGEA3 gene expres-sion by qRT demonstrated no detectable immunostaining. In ourstudy, matched histologically benign pancreas and colon served asnegative control specimens. Distinct patterns of membranous andcytoplasmic immunostaining for MAGEA3 protein were observedin undifferentiated cancer cells in the PDAC specimens (Fig. 3a),whereas little or no staining was present in representative desmo-plastic tissues (Fig. 3b) or well-differentiated tissues (Fig. 3c).

Discussion

As antigens with expression patterns unique to many tumors,cancer testis antigens from the MAGEA family have garneredattention as potential targets for vaccine-based immunotherapy ofcancer.9,10 The lack of gene expression of the encoding genes inhealthy tissues theoretically ensures strict tumor-specific targetedimmune responses after patient vaccination. Recently, clinical tri-als have been initiated with MAGEA-derived immunogens.38–41

Therefore, the determination of patients eligible for immunizationwith a defined MAGEA antigen mandates analysis of tumors forexpression of the MAGEA gene along with the appropriate HLAspecificity. Whether the appropriate criteria have been met can bereadily tested by HLA typing and by RT-PCR on RNA extractedfrom tumor samples.

Expression of the MAGE family of cancer testis antigens hasbeen detected in several epithelial cancers of varying embryologicorigin.17–20 The role of these MAGE antigens during embryogene-sis, cellular growth and differentiation is still not entirely clearsince their initial discovery, and their function in epithelial cancersis largely unknown.4 In these cancers, clinical correlations havebeen identified as a result of differential expression patterns. Somereports have noted that expression of MAGE antigens was associ-ated with worse pathological tumor stage or poor clinical out-comes.42,43 In contrast, Hansel et al.44 discovered that absence ofMAGEA1 expression indicated a worsened prognosis. In our cur-rent report, we detected MAGEA3 gene expression in a large per-centage of our patient cohort (44%). Patients whose tumors ex-pressed MAGEA3 had an �50% decrease in overall survival com-pared to tumors with absent expression.

The considerable difference in survival relative to MAGEA3gene expression was significant by univariate analysis. When clin-icopathologic factors were adjusted, MAGEA3 expression was nolonger significant (Table III). Because of the large percentage ofdeaths (n 5 20 of 23; 87%) in patients with poorly differentiatedtumors, tumor grade overshadowed MAGEA3’s role as a prognos-tic factor for survival in this patient cohort. Exclusion of tumorgrade from the multivariate analysis identifies MAGEA3 as a sig-nificant prognostic factor for poor survival (HR 2.1, 95% CI 1.0–4.4; p 5 0.04). Furthermore, there is a significant association ofMAGEA3 gene expression with tumor grade. Only 24% (n 5 11

TABLE III – UNIVARIATE AND MULTIVARIATE ANALYSIS

Clinicopathologic Factors #Death/nUnivariate analysis Multivariate analysis

Hazard ratio(95% CI)

p-value Hazard ratio(95% CI)

p-value

Age NS NS<50 2/3 1.050–70 17/27 1.1 (0.3–4.7)>70 16/27 1.3 (0.3–5.5)

Gender NS NSFemale 16/27 1.0Male 19/30 1.1 (0.6–2.1)

Stage NS NSp1 10/18 1.0p2 25/39 1.5 (0.7–3.1)

Tumor extent NS NSpT1 4/9 1.0pT2 27/41 1.5 (0.5–4.4)pT3 4/7 1.1 (0.3–4.4)

Lymph node disease NS NSpN0 11/21 1.0pN1 24/36 1.6 (0.8–3.2)

Tumor size (cm) NS NS<2 5/8 1.02–5 25/44 0.8 (0.3–2.2)>5 5/5 2.4 (0.8–8.5)

GradeWell 4/9 1.0Moderate 11/25 2.2 (0.6–8.0)Poor 20/23 3.7 (1.8–7.6) 0.0004 6.6 (1.8–23.6) 0.0007

Perineural invasion NS NSAbsent 10/18 1.0Present 25/39 1.6 (0.8–3.5)

Lymphovascular invasion NS NSAbsent 25/41 1.0Present 10/16 1.3 (0.6–2.8)

MAGEA3 expression NSNo 20/32 1.0Yes 15/25 2.1 (1.0–4.4) 0.041

2272 KIM ET AL.

of 45) of well- and moderately differentiated tumors demonstratedMAGEA3 gene expression, whereas 61% (n 5 14 of 23) of poorlydifferentiated tumors were MAGEA3 positive (v2, p 5 0.035). Theimplication is that MAGEA3 protein contributes to poor survivalthrough its expression in primarily poorly differentiated pancreatictumors.

Recent studies illustrate that epigenetic mechanisms may regu-late the differential patterns of MAGE gene expression.25 Thismechanism may account for the varying levels of MAGEA3 geneexpression by qRT in our pancreatic cancer cell lines (Fig. 1). Ademonstration of the epigenetic silencing of MAGEA3 wasreported by Sigalotti et al.,45 who utilized 5-aza-20-deoxycytidineto reverse methylation silencing of the MAGE genes. This findingis informative since the MAGEA3 gene on the X-chromosome hasmethylated CpG islands in normal somatic tissues.46,47 However,additional unknown epigenetic events may regulate MAGEA3expression and, therefore, require further studies. It is possible thatvariable silencing of MAGEA3 gene expression by epigeneticmechanisms may decrease levels of gene expression to suchdegrees, that it cannot be detected with semi-quantitative methods.This explanation may account for the discrepancies in resultsbetween the report by Kubuschok et al.26 and our study. Theirgroup, using gel-based RT-PCR, noted MAGEA3 gene expressionin only 2 of 10 pancreatic cancer cell lines and in none of theirpancreatic cancer biopsies.

Current therapeutic options, which include chemotherapy,monoclonal antibodies and radiation therapy remain limited inefficacy, although new agents and regimens show promise.48

MAGEA3, however, has been shown to be immunogenic inhumans, eliciting both cellular and antibody responses.49–52 Vari-ous peptide vaccines are under investigation and have shown toelicit specific immune responses, and more importantly clinicalresponses.41,52–54 In this report, we evaluated patients with limiteddisease, who were candidates for curative resection. Therefore,our patient cohort may not be representative of the 80–90% ofpatients with pancreatic cancer, who typically present with unre-sectable disease.55 It is unknown, whether MAGEA3 gene expres-sion is differentially expressed in the primary and metastaticlesions. We are currently accruing patients to evaluate whetherMAGEA3 gene expression has prognostic value in patients withadvanced disease.

Pancreatic cancer is an appropriate target for MAGE immune-based treatment strategies, specifically for those patients withtumors that express MAGEA3. In a recent trial, 25 tumor-bearingHLA-A1 melanoma patients were immunized with a MAGEA3peptide presented by HLA-A1, where objective regression ofmetastases was observed in 7 patients.38 Three of these regressionswere complete. No adverse side effects are expected in germlinecells because HLA molecules are not present at the surface ofthese cells. In conclusion, the clinical characteristics of MAGEA3,its absence in normal tissues and association with worse outcomesmake it an attractive target for immune-based treatment strategiesin patients with pancreatic cancer.

We, the authors of this study, certify that we have not enteredinto any agreement that interferes with our access to the data ofthis research study, our ability to analyze the data independently,our preparation of the manuscript and our publication of thismanuscript.

We, the authors of this study, have no financial interests to declare.

Acknowledgements

We would like to thank Dr. K. Washington at the CooperativeHuman Tissue Network, Vanderbilt University School of Medi-cine (Nashville, TN, USA) for their assistance in obtaining pan-creas specimens and we would like to thank the Tissue Procure-ment Core Laboratory of the Department of Pathology at UCLAfor their expertise and support in this project.

FIGURE 3 – (a) Representative MAGEA3 protein expression inundifferentiated pancreatic cancer cells by immunohistochemistry fol-lowed by staining with hematoxylin at 3400 magnification. Therewas absence of immunostaining in (b) desmoplastic tissues next tomalignant cells and absence of immunostaining in (c) well-differenti-ated tissues at3200 magnification.

2273MAGEA3 EXPRESSION IN PANCREATIC CANCER

References

1. Almoguera C, Shibata D, Forrester K, Martin J, Arnheim N, PeruchoM. Most human carcinomas of the exocrine pancreas contain mutantc-K-ras genes. Cell 1988;53:549–54.

2. Iacobuzio-Donahue CA, Song J, Parmiagiani G, Yeo CJ, Hruban RH,Kern SE. Missense mutations of MADH4: characterization of themutational hot spot and functional consequences in human tumors.Clin Cancer Res 2004;10:1597–1604.

3. Xiong HQ, Abbruzzese JL. Epidermal growth factor receptor-targetedtherapy for pancreatic cancer. Semin Oncol 2002;29(Suppl 14):31–7.

4. van der Bruggen P, Traversari C, Chomez P, Lurquin C, De Plaen E,Van den Eynde B, Knuth A, Boon T. A gene encoding an antigen rec-ognized by cytolytic T lymphocytes on a human melanoma. Science1991;254:1643–7.

5. Gaugler B, Van den Eynde B, van der Bruggen P, Romero P, GaforioJJ, De Plaen E, Lethe B, Brasseur F, Boon T. Human gene MAGE-3codes for an antigen recognized on a melanoma by autologous cyto-lytic T lymphocytes. J Exp Med 1994;179:921–30.

6. Chi DD, Merchant RE, Rand R, Conrad AJ, Garrison D, Turner R,Morton DL, Hoon DS. Molecular detection of tumor-associated anti-gens shared by human cutaneous melanomas and gliomas. Am J Pathol1997;150:2143–52.

7. Kirkin AF, Dzhandzhugazyan KN, Zeuthen J. Cancer/testis antigens:structural and immunobiological properties. Cancer Invest 2002;20:222–36.

8. Oaks MK, Hanson JP, Jr, O’Malley DP. Molecular cytogenetic map-ping of the human melanoma antigen (MAGE) gene family to chro-mosome region Xq27-qter: implications for MAGE immunotherapy.Cancer Res 1994;54:1627–9.

9. De Plaen E, Arden K, Traversari C, Gaforio JJ, Szikora JP, De SmetC, Brasseur F, van der Bruggen P, Lethe B, Lurquin C, et al. Struc-ture, chromosomal localization and expression of twelve genes of theMAGE family. Immunogenetics 1994;40:360–9.

10. Lucas S, De Smet C, Arden KC, Viars CS, Lethe B, Lurquin C, BoonT. Identification of a new MAGE gene with tumor-specific expressionby representational difference analysis. Cancer Res 1998;58:743–52.

11. Lucas S, Brasseur F, Boon T. A new MAGE gene with ubiquitousexpression does not code for known MAGE antigens recognized by Tcells. Cancer Res 1999;59:4100–3.

12. Lucas S, De Plaen E, Boon T. MAGE-B5, MAGE-B6, MAGE-C2 andMAGE-C3: four new members of the MAGE family with tumor spe-cific expression. Int J Cancer 2000;87:55–60.

13. Lurquin C, De Smet C, Brasseur F, Muscatelli F, Martelange V, DePlaen E, Brasseur R, Monaco AP, Boon T. Two members of thehuman MAGEB gene family located in Xp. 21.3 are expressed intumors of various histological origins. Genomics 1997;46:397–408.

14. Muscatelli F, Walker AP, De Plaen E, Stafford AN, Monaco AP. Iso-lation and characterization of a new MAGE gene family in the Xp21.3region. Proc Natl Acad Sci USA 1995;92:4987–91.

15. Pold M, Zhou J, Chen GL, Hall JM, Vescio RA, Berenson JR. Identi-fication of a new, unorthodox member of the MAGE gene family.Genomics 1999;59:161–7.

16. Rogner UC, Wilke K, Steck E, Korn B, Poustka A. The melanomaantigen gene (MAGE) family is clustered in the chromosomal bandXq28. Genomics 1995;29:725–31.

17. Fujie T, Mori M, Ueo H, Sugimachi K, Akiyoshi T. Expression ofMAGE and BAGE genes in Japanese breast cancers. Ann Oncol1997;8:369–72.

18. Otte M, Zafrakas M, Riethdorf L, Pichlmeier U, Loning T, Janicke F,Pantel K. MAGE-A gene expression pattern in primary breast cancer.Cancer Res 2001;61:6682–7.

19. Scanlan MJ, Gure AO, Jungbluth AA, Old LJ, Chen YT. Cancer/testisantigens: an expanding family of targets for cancer immunotherapy.Immunol Rev 2002;188:22–32.

20. Chen YT, Scanlan MJ, Sahin U, Tureci O, Gure AO, Tsang S, Wil-liamson B, Stockert E, Pfreundschuh M, Old LJ. A testicular antigenaberrantly expressed in human cancers detected by autologous anti-body screening. Proc Natl Acad Sci USA 1997;94:1914–8.

21. Lin J, Lin L, Thomas DG, Greenson JK, Giordano TJ, Robinson GS,Barve RA, Weishaar FA, Taylor JM, Orringer MB, Beer DG. Mela-noma-associated antigens in esophageal adenocarcinoma: identifica-tion of novel MAGE-A10 splice variants. Clin Cancer Res 2004;10:5708–16.

22. Sienel W, Varwerk C, Linder A, Kaiser D, Teschner M, Delire M,Stamatis G, Passlick B. Melanoma associated antigen (MAGE)-A3expression in stages I and II non-small cell lung cancer: results of amulti-center study. Eur J Cardiothorac Surg 2004;25:131–4.

23. Sugita M, Geraci M, Gao B, Powell RL, Hirsch FR, Johnson G,Lapadat R, Gabrielson E, Bremnes R, Bunn PA, Franklin WA. Com-bined use of oligonucleotide and tissue microarrays identifies cancer/testis antigens as biomarkers in lung carcinoma. Cancer Res 2002;62:3971–9.

24. Utsunomiya T, Inoue H, Tanaka F, Yamaguchi H, Ohta M, OkamotoM, Mimori K, Mori M. Expression of cancer-testis antigen (CTA)genes in intrahepatic cholangiocarcinoma. Ann Surg Oncol 2004;11:934–40.

25. Bert T, Lubomierski N, Gangsauge S, Munch K, Printz H, PrasnikarN, Robbel C, Simon B. Expression spectrum and methylation-depend-ent regulation of melanoma antigen-encoding gene family membersin pancreatic cancer cells. Pancreatology 2002;2:146–54.

26. Kubuschok B, Xie X, Jesnowski R, Preuss KD, Romeike BF, Neu-mann F, Regitz E, Pistorius G, Schilling M, Scheunemann P, IzbickiJR, Lohr JM, et al. Expression of cancer testis antigens in pancreaticcarcinoma cell lines, pancreatic adenocarcinoma and chronic pancrea-titis. Int J Cancer 2004;109:568–75.

27. Wascher RA, Bostick PJ, Huynh KT, Turner R, Qi K, Giuliano AE,Hoon DS. Detection of MAGE-A3 in breast cancer patients’ sentinellymph nodes. Br J Cancer 2001;85:1340–6.

28. Bostick PJ, Morton DL, Turner RR, Huynh KT, Wang HJ, ElashoffR, Essner R, Hoon DS. Prognostic significance of occult metastasesdetected by sentinel lymphadenectomy and reverse transcriptase-poly-merase chain reaction in early-stage melanoma patients. J Clin Oncol1999;17:3238–44.

29. Takeuchi H, Morton DL, Kuo C, Turner RR, Elashoff D, Elashoff R,Taback B, Fujimoto A, Hoon DS. Prognostic significance of molecu-lar upstaging of paraffin-embedded sentinel lymph nodes in mela-noma patients. J Clin Oncol 2004;22:2671–80.

30. Kuo CT, Hoon DS, Takeuchi H, Turner R, Wang HJ, Morton DL,Taback B. Prediction of disease outcome in melanoma patients bymolecular analysis of paraffin-embedded sentinel lymph nodes. J ClinOncol 2003;21:3566–72.

31. Kufer P, Zippelius A, Lutterbuse R, Mecklenburg I, Enzmann T,Montag A, Weckermann D, Passlick B, Prang N, Reichardt P, DugasM, Kollermann MW, et al. Heterogeneous expression of MAGE-Agenes in occult disseminated tumor cells: a novel multimarker reversetranscription-polymerase chain reaction for diagnosis of micrometa-static disease. Cancer Res 2002;62:251–61.

32. Ouyang H, Mou Lj, Luk C, Liu N, Karaskova J, Squire J, Tsao MS.Immortal human pancreatic duct epithelial cell lines with near normalgenotype and phenotype. Am J Pathol 2000;157:1623–31.

33. Furukawa T, Duguid WP, Rosenberg L, Viallet J, Galloway DA, TsaoMS. Long-term culture and immortalization of epithelial cells fromnormal adult human pancreatic ducts transfected by the E6E7 gene ofhuman papilloma virus 16. Am J Pathol 1996;148:1763–70.

34. Takeuchi H, Fujimoto A, Tanaka M, Yamano T, Hsueh E, Hoon DS.CCL21 chemokine regulates chemokine receptor CCR7 bearingmalignant melanoma cells. Clin Cancer Res 2004;10:2351–8.

35. Kim J, Takeuchi H, Lam ST, Turner RR, Wang HJ, Kuo C, Foshag L,Bilchik AJ, Hoon DS. CXCR4 expression in patients with colorectalcancer promotes liver metastasis. J Clin Oncol 2005;23:2744–53.

36. Yeo CJ, Cameron JL, Sohn TA, Lillemoe KD, Pitt HA, Talamini MA,Hruban RH, Ord SE, Sauter PK, Coleman J, Zahurak ML, GrochowLB, et al. Six hundred fifty consecutive pancreatico-duodenectomiesin the 1990s: pathology, complications, and outcomes. Ann Surg1997;226:248–57.

37. Neoptolemos JP, Stocken DD, Dunn JA, Almond J, Beger HG, Peder-zoli P, Bassi C, Dervenis C, Fernandez-Cruz L, Lacaine F, Buckels J,Deakin M, et al. Influence of resection margins on survival forpatients with pancreatic cancer treated by adjuvant chemoradiationand/or chemotherapy in the ESPAC-1 randomized controlled trial.Ann Surg 2001;234:758–68.

38. Escudier B, Dorval T, Chaput N, Andre F, Caby MP, Novault S, Fla-ment C, Leboulaire C, Borg C, Amigorena S, Boccaccio C, BonnerotC, et al. Vaccination of metastatic melanoma patients with autologousdendritic cell (DC) derived-exosomes: results of the first phase I clini-cal trial. J Transl Med 2005;3:10.

39. Toungouz M, Libin M, Bulte F, Faid L, Lehmann F, Duriau D,Laporte M, Gangji D, Bruyns C, Lambermont M, Goldman M, VeluT. Transient expansion of peptide-specific lymphocytes producingIFN-g after vaccination with dendritic cells pulsed with MAGE pepti-des in patients with mage-A1/A3-positive tumors. J Leukoc Biol2001;69:937–43.

40. Nishiyama T, Tachibana M, Horiguchi Y, Nakamura K, Ikeda Y,Takesako K, Murai M. Immunotherapy of bladder cancer using autol-ogous dendritic cells pulsed with human lymphocyte antigen-A24-specific MAGE-3 peptide. Clin Cancer Res 2001;7:23–31.

41. Kruit WH, van Ojik HH, Brichard VG, Escudier B, Dorval T, DrenoB, Patel P, van Baren N, Avril MF, Piperno S, Khammari A, Stas M,et al. Phase 1/2 study of subcutaneous and intradermal immunizationwith a recombinant MAGE-3 protein in patients with detectable meta-static melanoma. Int J Cancer 2005;117:596–604.

42. Mou DC, Cai SL, Peng JR, Wang Y, Chen HS, Pang XW, Leng XS,Chen WF. Evaluation of MAGE-1 and MAGE-3 as tumour-specific

2274 KIM ET AL.

markers to detect blood dissemination of hepatocellular carcinomacells. Br J Cancer 2002;86:110–16.

43. Hasegawa H, Mori M, Haraguchi M, Ueo H, Sugimachi K, AkiyoshiT. Expression spectrum of melanoma antigen-encoding gene familymembers in colorectal carcinoma. Arch Pathol Lab Med 1998;122:551–4.

44. Hansel DE, House MG, Ashfaq R, Rahman A, Yeo CJ, Maitra A.MAGE1 is expressed by a subset of pancreatic endocrine neoplasmsand associated lymph node and liver metastases. Int J GastrointestCancer 2003;33:141–7.

45. Sigalotti L, Fratta E, Coral S, Tanzarella S, Danielli R, Colizzi F, Fon-satti E, Traversari C, Altomonte M, Maio M. Intratumor heterogeneityof cancer/testis antigens expression in human cutaneous melanoma ismethylation-regulated and functionally reverted by 5-aza-20-deoxycy-tidine. Cancer Res 2004;64:9167–71.

46. De Smet C, De Backer O, Faraoni I, Lurquin C, Brasseur F, Boon T.The activation of human gene MAGE-1 in tumour cells is correlatedwith genome-wide demethylation. Proc Natl Acad Sci USA 1996;93:7149–53.

47. De Smet C, Lurquin C, Lethe B, Martelange V, Boon T. DNA methyl-ation is the primary silencing mechanism for a set of germ line- andtumour-specific genes with a CpG-rich promoter. Mol Cell Biol1999;19:7327–35.

48. Jafari M, Abbruzzese JL. Pancreatic cancer: future outlook, promisingtrials, newer systemic agents, and strategies from the GastrointestinalIntergroup Pancreatic Cancer Task Force. Surg Oncol Clin N Am2004;13:751–60.

49. Yu JS, Liu G, Ying H, Yong WH, Black KL, Wheeler CJ. Vaccina-tion with tumor lysate-pulsed dendritic cells elicits antigen-specific,

cytotoxic T-cells in patients with malignant glioma. Cancer Res 2004;64:4973–9.

50. Vantomme V, Dantinne C, Amrani N, Permanne P, Gheysen D, BruckC, Stoter G, Britten CM, Keilholz U, Lamers CH, Marchand M, DelireM, et al. Immunologic analysis of a phase I/II study of vaccination withMAGE-3 protein combined with the AS02B adjuvant in patients withMAGE-3-positive tumors. J Immunother 2004;27:124–35.

51. Schaft N, Dorrie J, Thumann P, Beck VE, Muller I, Schultz ES, Kamp-gen E, Dieckmann D, Schuler G. Generation of an optimized polyvalentmonocyte-derived dendritic cell vaccine by transfecting defined RNAsafter rather than before maturation. J Immunol 2005;174:3087–97.

52. Zhang Y, Renkvist N, Sun Z, Schuler-Thurner B, Glaichenhaus N,Schuler G, Boon T, van der Bruggen P, Colau D. A polyclonal anti-vaccine CD4 T cell response detected with HLA-DP4 multimers in amelanoma patient vaccinated with MAGE-3.DP4-peptide-pulsed den-dritic cells. Eur J Immunol 2005;35:1066–75.

53. Zhang Y, Sun Z, Nicolay H, Meyer RG, Renkvist N, Stroobant V,Corthals J, Carrasco J, Eggermont AM, Marchand M, Thielemans K,Wolfel T, et al. Monitoring of anti-vaccine CD4 T cell frequencies inmelanoma patients vaccinated with a MAGE-3 protein. J Immunol2005;174:2404–11.

54. Lonchay C, van der Bruggen P, Connerotte T, Hanagiri T, Coulie P,Colau D, Lucas S, Van Pel A, Thielemans K, van Baren N, Boon T.Correlation between tumor regression and T cell responses in mela-noma patients vaccinated with a MAGE antigen. Proc Natl Acad SciUSA 2004;101(Suppl 2):14631–8.

55. Jemal A, Murray T, Ward E, Samuels A, Tiwari RC, Ghafoor A,Feuer EJ, Thun MJ. Cancer statistics, 2005. CA Cancer J Clin 2005;55:10–30.

2275MAGEA3 EXPRESSION IN PANCREATIC CANCER