Expression of HOXB2, a Retinoic Acid Signalingtarget in Pancreatic Cancer and Pancreatic Intraepithelial Neoplasia Davendra Segara,1Andrew V

Cancer Prevention

Expression of HOXB2, a Retinoic Acid SignalingTarget in Pancreatic Cancer and Pancreatic Intraepithelial Neoplasia Davendra Segara,1Andrew V. Biankin,1, 2 James G. Kench,1, 3 Catherine C. Langusch,1Amanda C. Dawson,1 David A. Skalicky,1David C. Gotley,4 Maxwell J. Coleman,2 Robert L. Sutherland,1and Susan M. Henshall1

Abstract Purpose: Despite significant progress in understanding the molecular pathology of pancreatic cancer and its precursor lesion: pancreatic intraepithelial neoplasia (PanIN), there remain no molecules with proven clinical utility as prognostic or therapeutic markers. Here, we used oligonucleotide microarrays to interrogate mRNA expression of pancreatic cancer tissue and normal pancreas to identify novel molecular pathways dysregulated in the development and progression of pancreatic cancer. Experimental Design: RNA was hybridized toAffymetrix Genechip HG-U133 oligonucleotide microarrays. A relational database integrating data from publicly available resources was created toidentify candidate genes potentially relevant topancreatic cancer. The protein expression of one candidate, homeobox B2 (HOXB2), in PanIN and pancreatic cancer was assessed using immunohistochemistry. Results: We identified aberrant expression of several components of the retinoic acid (RA) signaling pathway (RARa,MUC4,Id-1,MMP9,uPAR,HB-EGF,HOXB6,andHOXB2),manyof which are known to be aberrantly expressed in pancreatic cancer and PanIN. HOXB2, a downstream target of RA, was up-regulated 6.7-fold in pancreatic cancer compared with normal pancreas. Immunohistochemistry revealed ectopic expression of HOXB2 in15% of early PanINlesions and 48 of 128 (38%) pancreatic cancer specimens. Expression of HOXB2 was associated with nonresectable tumors and was an independent predictor of poor survival in resected tumors. Conclusions: We identified aberrant expression of RA signaling components in pancreatic cancer, including HOXB2, which was expressed in a proportion of PanIN lesions. Ectopic expression of HOXB2 was associated with a poor prognosis for all patients with pancreatic cancer and was an independent predictor of survival in patients who underwent resection.

Pancreatic cancer is the fifth leading cause of cancer death that some of these variables lack accuracy. In addition, in Western societies with a 5-year survival rate of <10% (1). preoperative assessment of some variables such as lymph node Pancreatic cancer presents at an advanced stage; thus, only metastases is difficult. Whereas in other cancers assessment of 10% to 20% of patients are suitable for surgical treatment at aberrations in gene expression that cosegregate with therapeu- the time of presentation (1). Clinical management of these tic response and outcome are being adopted routinely to patients is complicated by inconsistencies in the influence of increase predictive power (e.g., ER and HER-2/neu in breast conventional clinicopathologic variables on outcome suggesting cancer), there remain no molecular markers of clinical utility in pancreatic cancer. This highlights the need for the identification of novel regulatory pathways important in Authors’ Affiliations: 1Cancer Research Program, Garvan Institute of Medical pancreatic cancer that may also have diagnostic, therapeutic Research and 2Division of Surgery, St. Vincent’s Hospital, Darlinghurst, Sydney, and prognostic utility. New South Wales, Australia; 3Institute of Clinical Pathology and Medical Research, There is now compelling histopathologic and molecular 4 Westmead Hospital, Westmead, New South Wales, Australia; and University of evidence to support the evolution of pancreatic cancer through Queensland, Department of Surgery, Princess Alexandra Hospital, Woolloongabba, Queensland, Australia a series of noninvasive duct lesions called pancreatic intra- Received 9/5/04; revised 2/1/05; accepted 2/10/05. epithelial neoplasia (PanIN; refs. 2, 3). Early duct lesions Grant support: Royal Australasian College of Surgeons, National Health and designated PanIN-1A and PanIN-1B show minimal cytologic Medical Research Council of Australia, St. Vincent’s Clinic Foundation Sydney, and architectural atypia and are associated with activating K-ras Cancer Council New South Wales, R.T. Hall Trust, and Prostate Cancer Foundation mutations (4), shortened telomeres (5), and overexpress of Australia (S.M. Henshall). WAF1/CIP1 The costs of publication of this article were defrayed in part by the payment of page p21 (6). PanIN-2 lesions exhibit mild to moderate charges. This article must therefore be hereby marked advertisement in accordance cytologic and architectural atypia and are associated with with 18 U.S.C. Section 1734 solely to indicate this fact. loss of p16INK4A expression (7) and cyclin D1 overexpression Note: D. Segara and A. Biankin contributed equally to this work. (6). PanIN-3 exhibits significant cytologic and architectural Requests for reprints: Robert L. Sutherland, Cancer Research Program Garvan atypia, manifests p53 mutations (8), and loss of DPC4/ Institute of Medical Research 384 Victoria Street, Darlinghurst. New South Wales 2010, Australia. Phone: 61-2-9295-8322; Fax: 61-2-9295-8321; E-mail: Smad4 expression (6). These molecular aberrations increase [email protected]. in frequency with advancing PanIN lesions through to invasive F 2005 American Association for Cancer Research. cancer.

During vertebrate development, retinoic acid (RA) signaling clinicopathologic data. Multiple samples of pancreatic tissue of f500 is important for the correct patterning of embryonic structures mg were excised intraoperatively from 12 patients, undergoing (9). Endodermal expression of pdx-1 (a homeobox-containing pancreatic resection for pancreatic cancer, immediately snap frozen in j transcription factor essential for pancreatic development) is liquid nitrogen and stored at À80 C, before RNA extraction. Total RNA was isolated from 12 pancreatic cancer specimens and six macroscopi- induced by RA (10) and marks a pluripotent population of cells cally and microscopically normal appearing pancreas from the same that give rise to all cell types in the pancreas. RA signaling patients (matched). Biotinylated cRNA for Affymetrix Genechip hybrid- regulates pancreas exocrine lineage selection, and treatment ization was prepared through a single round of reverse transcription with RA analogues can effect a shift from an acinar to a ductal with Superscript II (Life Technologies, Rockville, MD) followed by se- phenotype through epithelial-mesenchymal interactions (11). cond strand synthesis to create double stranded cDNA. After purification Such a shift from an exocrine to a predominantly ductal the cDNA was transcribed and labeled using a T7 polymerase (Enzo phenotype is characteristic of mouse models of pancreatic Technologies, New York, NY) and purified (27). Hybridization cocktails cancer development. In addition, pancreatic stellate cells, which were prepared as per the Affymetrix protocol (Affymetrix, Santa Clara, are essential for the development of fibrosis associated with CA) and quality assured on Affymetrix Test3 arrays, before hybridization chronic pancreatitis and pancreatic cancer, store retinoids in fat to HG-U133A and B oligonucleotide microarrays. Data analysis. A relational database was constructed using File- droplets, and in turn can have their function altered with RA Maker Pro (FileMaker, Inc., San Francisco, CA) to facilitate multiple analogue treatment in vitro (12). The retinoid signal is queries of gene expression data generated from the above experiments transduced by two families of nuclear transcription factors: and public domain data available electronically from the Internet. The RA receptors (RAR) and retinoid X receptors, that are members database incorporated (a) transcript profiles of pancreatic cancer and of the nuclear receptor superfamily, which in the presence of normal pancreas from the experiments done in this study (absolute ligand heterodimerize to activate the transcription of target values); (b) mathematical algorithms programmed within the database genes through RA response elements (13). Although few RA to generate fold change comparisons between the average expression response elements have been identified, one of the mechanisms across all samples of pancreatic cancer to the average in normal pancreas; by which retinoids exert their effects is thought to be through (c) linear statistical analyses generated using the Affymetrix Data Mining regulation of HOX gene expression (9, 14). Tool Software (MAS 5.0), which included t test and Mann-Whitney U Homeobox genes are transcription factors with established test data for comparisons between normal pancreas and pancreatic cancer and (d) interactive molecular pathway maps were generated using roles in development and cell function. The homeobox is a GenMAPP software (Gladstone Institutes UCSF, San Francisco, CA, highly conserved 183-bp DNA sequence coding for a 61-amino- http://www.GenMAPP.org/default.html), designed to incorporate tran- acid domain, the homeodomain (15). This region binds DNA script profile data into maps of known pathways including those elements, primarily those that contain a TAAT core motif involved in carcinogenesis and development. Data files using Swissprot (16). Accordingly, homeodomain containing proteins act as identification numbers were uploaded into the program, and various both activators and repressors of transcription. Human class 1 pathway maps available as part of the package were used to model homeobox genes called HOX genes consist of 39 genes numerous pathways. An existing RA signaling GenMAPP was modified arranged in four clusters HOXA, HOXB, HOXC, and HOXD to include all molecules thought to be regulated by RA signaling and is localized on chromosomes 7, 17, 12, and 2, respectively (17). presented in Fig. 1. Statistical data were generated using the t test and Mammalian development requires a complex interaction of Mann-Whitney U tests to compare the average expression across samples of pancreatic cancer to the average expression of all samples of normal HOX gene networks, with HOX gene expression commencing pancreas for the GenMAPP that is presented. during gastrulation and collectively controlling the identity of Patient cohort. We identified a cohort of 128 patients with a diag- various regions along the body axis from the hindbrain to the nosis of pancreatic adenocarcinoma that underwent pancreatic resection tail (18, 19). Aberrant expression of HOX genes has been or biopsy between January 1972 and November 2001 with available implicated in the development of solid tumors including renal archived tissue. This cohort represents a subset of a previously described carcinoma (20), colon cancer (21), ovarian carcinoma (22), group of 348 patients (28). Archival formalin-fixed, paraffin-embedded and breast carcinoma (23, 24). Given the emerging importance tissue from all 128 pancreata that were resected or biopsied were used to of developmental pathways in pancreatic cancer such as Notch construct seven pancreatic cancer tissue arrays, which contained up to (25) and sonic hedgehog (26), the role of RA signaling in early 55 Â 1.6 mm cores per slide. Conventional sections of 26 cases of normal pancreas development and evidence of RA signaling and pancreas from areas distal to the pancreatic cancer were used to assess homeobox gene network dysregulation in carcinogenesis, we gene expression in benign ductal epithelial cells and PanIN lesions. For this cohort, the average age at diagnosis was 63.8 years present data suggesting that aberrant RA signaling may be (median, 66.5; range, 34-86; Table 1). Of the 128 patients, 76 were important in pancreatic cancer. Based on these data, we from pancreatic resections, 46 intraoperative incision biopsies, and assessed HOXB2, a RA-responsive gene, and show that ectopic 6 postmortem specimens. Median follow-up for the cohort was expression of HOXB2 occurs in a significant proportion of 7.6 months (range, 0-117 months). Eight patients were alive at the pancreatic cancer, is detectable in a proportion of PanIN, and is census date (September 21, 2002). Median disease-specific survival associated with a poor prognosis, supporting a potential role of was 7.25 months. For the resected group of 76 patients, 39 (51%) had HOXB2 in the biological behavior of some pancreatic cancer. lymph node metastasis (Table 1). The mean tumor size was 31 mm. Resection margins were microscopically free of tumor in 40 patients (53%). Poorly differentiated tumors occurred in 25 patients (33%). Materials and Methods Median follow-up was 11.0 months with a median disease-specific survival of 10.1 months, 1-year survival of 48.6%, and 5-year survival RNA preparation and transcript profiling. Ethical approval was of 11%. The 30-day mortality for resection was 2 (3%). obtained from five teaching hospitals (The Princess Alexandra Hospital, Immunohistochemistry. Pancreatic tissue microarrays were cut at Brisbane, Australia, Westmead Hospital, Concord Hospital, Royal Prince 4 Am, deparaffinized, and rehydrated before unmasking in target Alfred Hospital, and St. Vincent’s Hospital Campus in Sydney, Australia) retrieval solution (EDTA and citrate, DAKO Co., Carpinteria, CA) in a for the acquisition of fresh and archival tissue and recording of microwave for 30 minutes. Using a DAKO autostainer, endogenous

Clin Cancer Res 2005;11(9) May 1, 2005 3588 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 9, 2021. © 2005 American Association for Cancer Research. HOXB2 and Pancreatic Cancer

Fig. 1. A customized GenMAPP of RA signaling components with statistically significant relative expression levels in pancreatic cancer (PC) compared with normal pancreas. Relative expression levels are represented as average fold change with those with statistically significant up-regulation marked red, those with statistically significant down-regulation marked blue (a < 0.05 on t test and/or Mann-Whitney U test) and nochange ( NC)markedgreenfor(A)RARs,(B) cellular RA-binding proteins (CRABP), (C) downstream targets of RA signaling previously described to be aberrantly expressed in pancreatic cancer and PanIN, (D) downstream targets of RA signaling aberrantly expressed in this study, (E) krox20-mediated regulation of HOXB2 and HOXB1expression in normal hindbrain development.Ps presented inTable 2. Abbreviations not mentioned in text: RBP4, retinoid binding protein 4; CRALBP, cellular retinaldehyde binding protein; RALDH, retinaldehyde dehydrogenase; hRADH, retinol dehydrogenase homologue. peroxidase activity was quenched in 3% hydrogen peroxide in methanol intensity of nuclear staining on a scale of 0 to 3 (0, no staining; 1, followed by avidin/biotin and serum-free protein blocks (DAKO). slight/weak heterogenous nuclear staining; 2, strong homogenous Sections were incubated for 30 minutes with 1:200 anti-HOXB2 (P-20) nuclear staining; and 3, intense homogenous nuclear staining). The goat polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA). A criteria to achieve a positive score were HOX B2 nuclear intensity of >1 streptavidin-biotin peroxidase detection system was used according in >20% of nuclei. to the manufacturer’s instructions (LSAB label + link kit; DAKO) with Statistical analysis. Kaplan-Meier and the Cox proportional hazards 3,3V-diaminobenzidine as a substrate. Counterstaining was done with model were used for univariate and multivariate analysis using Statview Mayer’s hematoxylin. HOXB2-positive breast cancer was used as a 5.0 Software (Abacus Systems, Berkeley, CA). P < 0.05 was accepted as positive control (24), whilst ovary was used as a negative control. statistically significant. Those factors that were prognostic on univariate Antibody specificity was confirmed using blocking peptide sc-17165 P analysis were assessed in a multivariable model to identify factors that (Santa Cruz Biotechnology), which abrogated nuclear staining for were independently prognostic and those that were the result of immunohistochemistry and eliminated a specific band on Western confounding. This analysis was done sequentially on all patients who blotting using pancreatic cancer cell lines. In addition, mRNA expres- had available tissue (n = 128) and on a subgroup of patients who sion in cell lines using reverse transcription-PCR correlated with protein underwent operative resection (n = 76). expression on Western blotting. Immunohistochemical scoring. Up to four separate samples of pancreas were examined per patient. Staining was assessed by two Results blinded independent observers (D.S. and J.G.K.). Standardization of scoring was achieved by comparison of scores between observers and by Transcript profiling data analysis. Whereas previous tran- conferencing, where any discrepancies were resolved by consensus. script profiling studies have been limited to identifying single Scores were given as the percentage of nuclei staining positive within genes aberrantly expressed in pancreatic cancer (29, 30), we the representative area of the tissue microarray core and the absolute employed a strategy that used GenMAPP software to identify

Ta bl e 1. Clinicopathologic and outcome data for all patients in the cohort

Whole cohort Median Resected Median Variable no. (%) survival (mo) P (log-rank) cohort no. (%) survival (mo) P (log-rank)

Sex 128 76 Male 72 (56) 45 (59) Female 56 (44) 31(41) Age (y) 128 76 Mean 63.8 61 Median 66.5 65 Range 34-86 34-83 Treatment 128 Resection 76 (59) 11 Operative biopsy 46 (36) 3.9 <0.0001 Nooperative intervention 6 (5) Outcome 128 76 Follow-up (mo) 0-117 0.2-117 Median 7.6 11 30-d mortality 2(3) Death from pancreatic cancer 114(89) 63 (83) Death from other cause 2 (2) 2 (3) Alive 8 (6) 8 (11) Losttofollow-up 4(3) 3(4) Stage 127 I27(21) II 13 (10) 13.7 III 70 (55) IV 17 (13) 6.4 <0.0001 Differentiation 127 Well 11 (9) 7 (9) Moderate 68 (53) 8.9 44 (58) 12.2 Poor 48 (38) 5 0.0152 25 (33) 8.6 0.0582 Tumor size (mm) V20 15 (20) 17.1 >20 61 (80) 9.7 0.0375 Margins Clear 40 (53) 14.5 Involved 36 (47) 8.5 0.0014 Lymph node status Positive 39 (51) 9.2 Negative 35 (46) 13.8 0.0235 HOXB2 expression 128 76 Positive 48 (38) 5 16 (21) 6.75 Negative 80 (72) 9.9 <0.0001 60 (79) 14.0 <0.0001 molecular pathways in which a significant proportion of genes cancer and PanIN from other studies, were also up-regulated: showed aberrant expression. Using this approach, we confirmed S100 calcium binding protein P (S100P; ref. 31; 152-fold), aberrations in molecular pathways known to be important in MUC4 mucin (ref. 32; 24.6-fold), matrix metalloproteinase 9 pancreatic cancer (transforming growth factor–h signaling, cell (MMP9; ref. 33; 2.0-fold), Id-1 (ref. 34; 2.3-fold), urokinase cycle regulation, and apoptosis; data not shown). In addition, plasminogen activator receptor (uPAR; ref. 35; 13.5-fold), and we identified aberrant expression of a significant number of heparin-binding epidermal growth factor-like growth factor components of RA signaling (Table 2; Fig. 1). RAR-a and RAR-g (HB-EGF; ref. 36; 2.5-fold; Table 2). Other genes, yet to be were up-regulated 2.9- and 2.2-fold, respectively, in pancreatic characterized in pancreatic cancer but thought to be regulated by cancer compared with normal pancreas. Expression of a RA were also aberrantly expressed, including a RA-induced G- substantial number of known RA-responsive genes was also protein–coupled receptor (26.3-fold) and RAR responders altered in pancreatic cancer, consistent with dysregulated RA RARRES 1 (16.5-fold) and RARRES 3 (3.8-fold; Table 2). signaling activity, primarily demonstrating up-regulation of Studies of hindbrain development have provided the greatest genes downstream of RAR-a. A substantial number of genes insights into the mechanism of RA signaling. RA-dependent regulated by RA and known to be highly expressed in pancreatic lineage restriction in rhombomeres 3 and 5 is marked by

Clin Cancer Res 2005;11(9) May 1, 2005 3590 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 9, 2021. © 2005 American Association for Cancer Research. HOXB2 and Pancreatic Cancer krox20 and HOXB2 expression. RA, through an as yet HOXB2 expression in pancreatic cancer and pancreatic intra- unknown mechanism that may involve CEBPh (37), results epithelial neoplasia. Representative examples of HOXB2 in increased krox20 expression, which in turn increases immunostaining are shown in Fig. 2. Nuclear expression was HOXB2 expression by directly binding promoter elements of identified in 48 of 128 cancers (38%). When HOXB2 HOXB2 (38). krox20 also suppresses HOXB1 expression. The expression was present within the tumor, >80% of the nuclei expression profile in the present study is consistent with stained positively. HOXB2 expression was detected in the activity of this pathway of HOXB2 regulation (Fig. 1). In histologically normal pancreatic ducts of 2 of 26 (8%) patients, addition, the variant promyelocytic leukemia fusion protein in 1 of 24 (4%) PanIN-1A lesions, 3 of 20 (15%) PanIN-1B, 3 of PLZF-RARA also regulates HOXB2 expression through a 10 (30%) PanIN-2, and 1 of 4 (25%) PanIN-3 lesions, showing similar mechanism and is thought to be important in that HOXB2 expression occurs in PanIN and may play a role in promyelocytic leukemia development and resistance to RA the evolution of PanIN. therapy (39). For these reasons, HOXB2,apreviously HOXB2 expression in the whole cohort was associated with a uncharacterized gene in pancreatic cancer, which showed a poor outcome (median survival, 5 versus 9.9 months; log-rank 6.7-fold increase (P < 0.001) compared with normal pancreas, P < 0.0001; Fig. 3A). In addition, operative resection (P < was selected for further study. 0.0001), low-stage (P < 0.0001), and non–poorly differentiated

Ta bl e 2 . Components of RA signaling with differential expression between pancreatic cancer and normal pancreas on Affymetrix U133 microarrays

Probe set Unigene cluster Gene name Fold change P 204351_at Hs.2962 S10 0 (calcium-binding protein P) 152 0.001 203108_at Hs.194691 GPCR (RA-induced 3) 26.3 0.007 217109_at Hs.198267 MUC4 24.6 0.001 206392_s_at Hs.82547 RAR responder1 (RARRES1)16.50.004 205366_s_at Hs.98428 HOXB6 14.4 0.009 211924_s_at Hs.179657 UPAR 13.5 0.002 202859_x_at Hs.624 Interleukin 8 (IL8 ) 12.5 0.001 205453_at Hs.2733 HOXB2 6.7 0.001 219799_s_at Hs.179608 Retinol dehydrogenase homologue (hRADH)4.70.016 203596_s_at Hs.27610 RA- and IFN-inducible protein (IFT5) 4.2 0.010 204070_at Hs.17466 RAR responder 3 (RARRES 3) 3.8 0.005 228601_at Hs.93574 HOXD3 3.4 0.024 205249_at Hs.1359 Krox20 (EGR2) 3.2 0.008 231936_at Hs.40408 HOXC9 3.2 0.009 213844_at Hs.37034 HOXA5 3.0 0.035 203749_s_at Hs.250505 RAR-a 2.9 0.007 201042_at Hs.512708 TGM2 2.9 0.001 202510_s_at Hs.101382 TNFAIP2 2.9 0.004 204420_at Hs.283565 FOSL1 (FOS-like antigen-1) 2.8 0.005 205601_s_at Hs.22554 HOXB5 2.7 0.003 206858_s_at Hs.820 HOXC6 2.6 0.016 202575_at Hs183650 Cellular RA-binding protein 2 (CRABP2) 2.6 0.018 38037_at Hs.799 HB-EGF 2.5 0.009 2214782_at Novel gene similar to retinaldehyde-binding 2.4 0.036 protein (sRABP) 208937_s_at Hs.75424 Id-1 (inhibitor of DNA binding1) 2.3 0.039 201505_at Hs.82124 Laminin b1 2.2 0.026 204118_s_at Hs.1497 RAR-c 2.2 0.049 212501_at Hs.99029 CEBP b (CCAATenhancer-binding protein h)2.10.005 203936_s_at Hs.151738 MMP9 2.0 0.008 221701_s_at Hs.24553 STRA6 1. 9 0.0 3 0 202449_s_at Hs.20084 Retinoid X receptor a (RXR-a)0.40.005 231906_at Hs.301963 HOXD8 0.6 0.010 202882_x_at Hs.106346 RA-repressible protein (RARG-1) 0.6 0.045 207914_x_at Hs.336963 Even-skipped homeobox1 (EVX1) 0.5 0.016 208224_at Hs.99992 HOXB1 0.4 0.010 205883_at Hs.37096 Zinc finger protein145 (ZNF145, PLZF) 0.3 0.002 203423_at Hs.101850 Cellular RA-binding protein1 (CRABP1) 0.2 0.001 209496_at Hs.37682 RAR responder 2 (RARRES 2)0.10.001

Fig. 2. Photomicrographs of HOXB2 nuclear staining (magnification, Â200). A, negative HOXB2 expression in normal pancreatic duct. B, positive nuclear staining in PanIN-1Blesion. C, positive nuclear staining in PanIN-3 lesion. D,positive nuclear staining in pancreatic cancer. E, negative nuclear staining in PanIN-1A lesion. F, negative nuclear staining in pancreatic cancer. G, HOXB2 staining in pancreatic cancer without blocking peptide. H, serial section of (G) stained with anti-HOXB2 antibody after incubation with blocking peptide sc-17165 P.

Clin Cancer Res 2005;11(9) May 1, 2005 3592 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 9, 2021. © 2005 American Association for Cancer Research. HOXB2 and Pancreatic Cancer tumors (P = 0.0152) were associated with significantly 14.0 versus 3.7 months; log-rank P < 0.0001; Fig. 3C). improved survival using Kaplan-Meier analysis. However, Survival for patients with tumors that were HOXB2 negative multivariate analysis identified resection and stage as the only and who underwent resection was significantly longer than independent prognostic factors when modeled together with survival in all other groups (14 versus 4.3 months; log-rank degree of differentiation and HOXB2 status (Table 3A). Whereas P < 0.0001; Fig. 3D). Hence, in this cohort, lack of HOXB2 HOXB2 expression was identified in 32 of 52 (62%) unresected expression cosegregated with operative resectability, with only tumors, it was present in only 16 of 76 (21%) resected those who were HOXB2 negative having a survival advantage pancreatic cancers. Hence, HOXB2 expression was associated from operative resection. with nonresectable tumors (m2; P < 0.0001) and consequently Survival analysis of patients that underwent operative was not an independent prognostic factor. Operative resection resection identified decreased survival associated with HOXB2 did not benefit those patients whose tumors expressed nuclear expression (median survival, 6.75 versus 14.0 months; HOXB2 (log-rank P = 0.37; Fig. 3B) but was beneficial to log-rank P < 0.0001; Fig. 3E). Kaplan-Meier analyses identified those patients who did not express HOXB2 (median survival, clear margin status (P = 0.0014), tumor size of V20 mm

Fig. 3. Kaplan-Meier survival curves for (A) HOXB2 nuclear expression in the whole cohort. Effect of resection on prognosis in the following subgroups: (B)HOXB2 positive, (C)HOXB2negative,(D)all patients stratified for HOXB2 status and resection. Kaplan-Meier survival curves for 76 patients whounderwent surgical resection: (E) HOXB2 nuclear expression, (F) margin status, (G) tumor size, and (H) lymph node involvement.

Ta bl e 3 . Multivariate analysis for clinicopathologic variables and HOXB2 expression in the whole cohort and resected pancreatic cancer

Variable Hazards ratio (95% confidence interval) P A.Whole cohort (n = 127) Stage III/IV versus I/II 2.30 (1.44-3.69) 0.005 Resection 0.43 (0.26-0.72) 0.0013 HOXB2 expression 1.56 (0.94-2.57) 0.085 B. Resected (n = 74) HOXB2 expression 2.90 (1.51-5.57) 0.0014 Margin involvement 1.89(1.02-3.48) 0.0428 Lymph node involvement 1.30 (0.71-2.40) 0.3981 C. Resected (n = 76) HOXB2 expression 2.82 (1.48-5.40) 0.0017 Margin involvement 2.04 (1.17-3.53) 0.0115 Tu m or s i z e , >20 mm 1.48 (0.75-2.90) 0.2567 D. Resected (n = 74) HOXB2 expression 2.69 (1.39-5.20) 0.0032 Margin involvement 1.75 (0.94-3.25) 0.0777 Lymph node involvement 1.34(0.73-2.46) 0.3525 Tu m or s i z e , >20 mm 1.49 (0.76-2.94) 0.2474

(P = 0.0375), and no lymph node involvement (P = 0.0235) as pancreatic fate (41). RARs also regulate exocrine pancreatic being associated with a survival advantage (Fig. 3F-H). Degree development at later stages (42), by modulating lineage of differentiation was not associated with a survival advantage selection favoring ductal rather than acinar differentiation (P = 0.0582). HOXB2 expression and involved surgical margins primarily through RAR-a (10, 11). Reactivation of develop- were independent prognostic factors when modeled against all mental pathways, specifically those that regulate exocrine cell combinations of HOXB2 expression, involved surgical margin, lineage, have been implicated previously in the early develop- lymph node involvement, and tumor size in the subgroup of ment of pancreatic cancer (25) and other pathways that patients who underwent surgical resection (Table 3B-D). determine duct cell versus acinar cell differentiation involving RA signaling, may also be important. In addition, forced Discussion expression of cellular retinol binding protein (CRABP1), a mediator of RA signaling, in transgenic mice results in the Expression profiling identified differential expression of a development of poorly differentiated pancreatic cancer (43), significant number of RA signaling pathway components and further supporting a role of aberrant RA signaling in pancreatic downstream responders in pancreatic cancer compared with cancer evolution. The transcript profile data presented here normal pancreas. These included some genes that are known to suggests the RA signaling pathway has a role in pancreatic be associated with pancreatic cancer: MUC4, MMP9, Id-1, cancer, specifically, a number of RA-responsive genes known to uPAR, HB-EGF, and S100P, as well as novel candidates. be important in pancreatic cancer and PanIN development Although there is substantial evidence implicating aberrant were aberrantly expressed in this study: MUC4 mucin is retinoid signaling in carcinogenesis (e.g., acute promyelocytic overexpressed in a significant proportion of pancreatic cancer leukemia; ref. 40), the mechanisms by which retinoid target (44) and PanIN (32) and can be induced through RAR-a genes exert these effects remains to be elucidated. Here we activation (45); similarly, MMP9 is expressed in pancreatic present evidence, implicating RA signaling in pancreatic cancer cancer (33) and is up-regulated by RA treatment (12) as is and show that a RA-responsive homeodomain transcription uPAR, HB-EGF, and p21WAF1/CIP1 (46). Id-1, which antago- factor, HOXB2, is ectopically expressed in a significant nizes basic helix loop helix proteins, inhibits differentiation proportion of pancreatic cancer, with a profound association and can enhance cell proliferation is overexpressed in PanIN with tumor progression. HOXB2, which is not normally lesions (34) and is also RA responsive (47). In the present expressed in the pancreas at any stage during development or study, HOXB2 expression was also detected in PanIN lesions. adult life, was expressed in 38% of pancreatic cancers and RA seems to exert its effect on exocrine lineage selection seemed to occur during the development of a proportion of towards a ductal phenotype during development through PanIN. Ectopic HOXB2 expression was associated with non- laminin-h1 (11), which was up-regulated in the present study. resectable tumors and was an independent prognostic factor in As was the putative tumor suppressor gene RARRES 3 (48). resected tumors when modeled with known clinicopathologic Stra6, a gene whose function is yet to be determined responds prognostic factors. In addition, only those patients that were to RA, is up-regulated in colon cancer (49) and was up- HOXB2 negative obtained a survival advantage with operative regulated in this study as was transglutaminase 2 (TGM2) and resection. tumor necrosis factor a–induced protein 2 (TNFAIP2), both Numerous lines of evidence from separate studies suggest also RA-responsive genes (47). There is clear evidence the importance of individual RA signaling components in supporting RA regulation of HOXB2 expression from studies pancreas development and pancreatic cancer evolution. RA of hindbrain patterning and branchial arch development (38); regulates early instructive signals from lateral plate mesoderm however, the mechanism by which RA signaling imparts its that is essential for specification of endoderm towards a effects on the HOX network and cellular function is poorly

Clin Cancer Res 2005;11(9) May 1, 2005 3594 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 9, 2021. © 2005 American Association for Cancer Research. HOXB2 and Pancreatic Cancer

understood. There were, however, some inconsistencies in the efficacy as a marker of prognosis in pancreatic cancer, as ectopic data where downstream targets of RA were down-regulated expression of a protein is a more reliable indicator than loss of such as HOXD8. Presumably, other mechanisms can also expression. Although pancreatic resection offers the best chance regulate the expression levels of these transcripts other than of cure in patients with pancreatic cancer, it is a procedure which through RA signaling alone. Validation of these data with carries significant morbidity and mortality. The development of manipulation of RA signaling is required to further investigate a reliable preoperative assessment of HOXB2 status would be an a putative functional role in pancreatic cancer; however, the important addition to a physician’s limited diagnostic arma- strength of data from the literature and evidence presented mentarium in this disease and may be used, together with here makes a strong case for an important role in pancreatic current clinicopathologic variables of disease outcome, to cancer. determine a patient’s suitability for operative resection. Multivariate analysis identified HOXB2 expression as an In conclusion, gene expression profiling of pancreatic cancer independent predictor of survival in the subgroup of patients has suggested that RA signaling is a potentially important that underwent pancreatic resection. Although HOXB2 expres- regulatory pathway in pancreatic cancer evolution. Ectopic sion was not identified as an independent predictor of survival expression of HOXB2 in pancreatic cancer is a frequent in the whole cohort due to its association with resection, lack of occurrence, an event which manifests in the development of HOXB2 expression combined with surgical resection conferred a PanIN in a proportion of cases, and is possibly a consequence of significant survival advantage. Because all known prognostic aberrant RA signaling. Current prognostic factors for pancreatic indicators in pancreatic cancer, such as tumor size, resection cancer remain poorly defined, depend upon examination of the margins, and lymph node status can only be determined post resected pancreas, and cannot be accurately determined preop- resection, HOXB2 expression has potential utility as a prognos- eratively. Assessment of HOXB2 expression may provide an tic indicator in pancreatic cancer, especially because it seems to alternative method for determining the suitability for resection have a profound independent influence on survival, with the and the prognosis of patients with pancreatic cancer. Further advantage that it can be assessed using biopsy techniques study to determine the effects of ectopic HOXB2 expression and without resection. We have previously identified that loss of other components of the HOX transcriptional network, its DPC4/Smad4 expression is associated with poor outcome in relationship to RA signaling, and clinical utility in pancreatic pancreatic cancer (28). However, HOXB2 may have more adenocarcinoma is required.

References 1. YeoCJ,CameronJL,SohnTA,etal.Sixhundred selection through epithelial-mesenchymal interac- 24. Cantile M, PettinatoG, ProcinoA, FelicielloI, fifty consecutive pancreaticoduodenectomies in the tions. Gastroenterology 2002;123:1331 ^ 40. CindoloL, CilloC. In vivo expression of the whole 1990s: pathology, complications, and outcomes. 12. Jaster R, Hilgendorf I, Fitzner B, et al. Regulation of HOX gene network in human breast cancer. Eur J Ann Surg 1997;226:248 ^ 57; discussion 257 ^ 60. pancreatic stellate cell function in vitro: biological and Cancer 2003;39:257 ^ 64. 2. Hruban RH, Adsay NV, Albores-Saavedra J, et al. molecular effects of all-trans retinoic acid. Biochem 25. Miyamoto Y, Maitra A, Ghosh B, et al. Notch medi- Pancreatic intraepithelial neoplasia: a new nomencla- Pharmacol 2003;66:633^41. ates TGF a-induced changes in epithelial differentia- ture and classification system for pancreatic duct 13. Mangelsdorf DJ, Evans RM. The RXR heterodimers tion during pancreatic tumorigenesis. Cancer Cell lesions. AmJ Surg Pathol 2001;25:579 ^ 86. and orphan receptors. Cell 1995;83:841 ^ 50. 2003;3:565^76. 3. Hruban RH, Takaori K, Klimstra DS, et al. An illus- 14. OosterveenT, Niederreither K, Dolle P, Chambon P, 26.Thayer SP, di MaglianoMP, Heiser PW, et al. Hedge- trated consensus on the classification of pancreatic Meijlink F, DeschampsJ. Retinoids regulate the anteri- hog is an early and late mediator of pancreatic cancer intraepithelial neoplasia (PanIN) and intraductal or expression boundaries of 5 VHoxb genes in posterior tumorigenesis. Nature 2003;425:851 ^ 6. papillary mucinous neoplasms (IPMNs). Am J Surg hindbrain. EMBO J 2003;22:262 ^ 9. 27. Baugh LR, Hill AA, Brown EL, Hunter CP. Pathol 2004;28:977 ^ 87. 15. Duboule D, Morata G. Colinearity and functional Quantitative analysis of mRNA amplification by 4. Moskaluk CA, Hruban RH, Kern SE. p16 and K-ras hierarchy among genes of the homeotic complexes. in vitro transcription. Nucleic Acids Res 2001; gene mutations in the intraductal precursors of human Trends Genet 1994;10:358^64. 29:e29. pancreatic adenocarcinoma. Cancer Res 1997;57: 16. Abate-Shen C. Deregulated homeobox gene ex- 28. Biankin AV, Morey AL, Lee C-S, et al. DPC4/ 2140 ^ 3. pression in cancer: cause or consequence? Nat Rev Smad4 expression and outcome in pancreatic 5. van Heek NT, Meeker AK, Kern SE, et al. Telomere Cancer 2002;2:777 ^ 85. ductal adenocarcinoma. J Clin Oncol 2002;20: shortening is nearly universal in pancreatic intraepithe- 17. ApiouF, FlagielloD, CilloC, MalfoyB, PouponMF, 4531 ^ 42. lial neoplasia. Am J Pathol 2002;161:1541 ^ 7. Dutrillaux B. Fine mapping of human HOX gene clus- 29. Iacobuzio-Donahue CA, Ashfaq R, Maitra A, et al. 6. Biankin AV, Kench JG, Morey AL, et al. Overexpres- ters. Cytogenet Cell Genet 1996;73:114^5. Highly expressed genes in pancreatic ductal adeno- sion of p21WAF1/CIP1is an early event in the develop- 18. Graham A, Papalopulu N, Krumlauf R. The murine carcinoma: a comprehensive characterization and ment of pancreatic intraepithelial neoplasia. Cancer and Drosophila homeobox gene complexes have comparison of the transcript profiles obtained from Res 2001;61:8830 ^ 7. common features of organization and expression. Cell three major technologies. Cancer Res 2003;63: 7. Wilentz RE, Geradts J, Maynard R, et al. Inactivation 1989;57:367^78. 8614^ 22. of the p16 (INK4A) tumor-suppressor gene in pan- 19. CilloC, Faiella A, Cantile M, Boncinelli E. Homeobox 30. Logsdon CD, Simeone DM, Binkley C, et al. Mole- creatic duct lesions: loss of intranuclear expression. genes and cancer. Exp Cell Res 1999;248:1 ^ 9. cular profiling of pancreatic adenocarcinoma and Cancer Res 1998;58:4740 ^ 4. 20. CilloC, Barba P, Freschi G, Bucciarelli G, Magli MC, chronic pancreatitis identifies multiple genes differen- 8. DiGiuseppe JA, Hruban RH, Goodman SN, et al. Boncinelli E. HOX gene expression in normal and neo- tially regulated in pancreatic cancer. Cancer Res Overexpression of p53 protein in adenocarcinoma of plastic human kidney. IntJCancer1992;51:892 ^ 7. 2003;63:2649 ^ 57. the pancreas. AmJ Clin Pathol 1994;101:684 ^8. 21. De Vita G, Barba P, OdartchenkoN, et al. Expres- 31. Crnogorac-JurcevicT, Missiaglia E, Blaveri E, et al. 9. Oosterveen T, van Vliet P, Deschamps J, Meijlink F. sion of homeobox-containing genes in primary and Molecular alterations in pancreatic carcinoma: expres- The direct context of a hox retinoic acid response ele- metastatic colorectal cancer. Eur J Cancer 1993;29A: sion profiling shows that dysregulated expression of ment is crucial for its activity. J Biol Chem 2003;278: 887 ^ 93. S100 genes is highly prevalent. J Pathol 2003;201: 24103 ^ 7. 22. Naora H, YangYQ, Montz FJ, Seidman JD, Kurman 63 ^74. 10. TulachanSS,DoiR,KawaguchiY,etal.All-trans RJ, Roden RB. A serologically identified tumor antigen 32. Swartz MJ, Batra SK, Varshney GC, et al. MUC4 retinoic acid induces differentiation of ducts and encoded by a homeobox gene promotes growth of expression increases progressively in pancreatic endocrine cells by mesenchymal/epithelial inter- ovarian epithelial cells. Proc Natl Acad Sci U S A intraepithelial neoplasia. Am J Clin Pathol 2002;117: actions in embryonic pancreas. Diabetes 2003;52: 2001;98:4060^5. 791 ^ 6. 76 ^ 84. 23. Lewis MT.Homeobox genes in mammary gland de- 33. Gong YL, Xu GM, Huang WD, Chen LB. Expression 11. Kobayashi H, Spilde TL, Bhatia AM, et al. Retinoid velopment and neoplasia. Breast Cancer Res 2000;2: of matrix metalloproteinases and the tissue inhibitors signaling controls mouse pancreatic exocrine lineage 15 8 ^ 6 9. of metalloproteinases and their local invasiveness

and metastasis in Chinese human pancreatic cancer. Krumlauf R, Zelent A. Regulation of Hoxb2 by 45. Choudhury A, Singh RK, Moniaux N, El-Metwally J Surg Oncol 2000;73:95 ^ 9. APL-associated PLZF protein. Oncogene 2003;22: TH, Aubert JP, Batra SK. Retinoic acid-dependent 34. Maruyama H, Kleeff J, Wildi S, et al. Id-1 and Id-2 3685^97. transforming growth factor-h 2-mediated induction are overexpressed in pancreatic cancer and in dys- 40. Melnick A, Licht JD. Deconstructing a disease: of MUC4 mucin expression in human pancreatic tu- plastic lesions in chronic pancreatitis. Am J Pathol RARa, its fusion partners, and their role in the patho- mor cells follows retinoic acid receptor-a signaling 1999;155:815^ 22. genesis of acute promyelocytic leukemia. Blood1999; pathway. J Biol Chem 2000;275:33929 ^ 36. 35. CanteroD, Friess H, Deflorin J, et al. Enhanced ex- 93:3167 ^ 215. 46. Liu T-X, Zhang J-W, TaoJ, et al. Gene expression pression of urokinase plasminogen activator and its re- 41. Kumar M, Jordan N, Melton DA, Grapin-Botton A. networks underlying retinoic acid-induced differentia- ceptor in pancreatic carcinoma. Br J Cancer 1997; Signals from lateral plate mesoderm instruct endo- tion of acute promyelocytic leukemia cells. Blood 75:388 ^ 95. derm towards a pancreatic fate. Dev Biol 2003;259: 2000;96:1496^504. 36. ItoY, Higashiyama S, Takeda T, YamamotoY, 109 ^22. 47. Ma Y, Koza-Taylor PH, DiMattia DA, et al. Micro- Wakasa KI, Matsuura N. Expression of heparin- 42. Kadison A, Kim J, Maldonado T, et al. Retinoid array analysis uncovers retinoid targets in human binding epidermal growth factor-like growth factor signaling directs secondary lineage selection in bronchial epithelial cells. Oncogene 2003;22: in pancreatic adenocarcinoma. Int J Gastrointest pancreatic organogenesis. J Pediatr Surg 2001;36: 4924 ^32. Cancer 2001;29:47 ^52. 115 0 ^ 6 . 48. DiSepioD, Ghosn C, Eckert RL, et al. Identification 37. Duprez E, Wagner K, Koch H, Tenen DG. C/EBPb: a 43. Giguere V. Retinoic acid receptors and cellular reti- and characterization of a retinoid-induced class II major PML-RARA responsive gene in retinoic acid- noid binding proteins: complex interplay in retinoid tumor suppressor/growth regulatory gene. Proc Natl induced differentiation of APL cells. EMBO J 2003; signaling. Endocr Rev 1994;15:61 ^ 79. Acad Sci U S A1998;95:14811^5. 22:5806 ^ 16. 44. Balague C, Gambus G, CarratoC, et al. Altered ex- 49. SzetoW, Jiang W,Tice DA, et al. Overexpression of 38. Gavalas A. ArRAnging the hindbrain.Trends Neuro- pression of MUC2, MUC4, and MUC5 mucin genes in the retinoic acid-responsive gene Stra6 in human sci 2002;25:62 ^ 4. pancreas tissues and cancer cell lines. Gastroentero- cancers and its synergistic induction by Wnt-1 and 39. Ivins S, Pemberton K, Guidez F, Howell L, logy1994;106:1054^61. retinoic acid. Cancer Res 2001;61:4197 ^ 205.

Clin Cancer Res 2005;11(9) May 1, 2005 3596 www.aacrjournals.org Downloaded from clincancerres.aacrjournals.org on October 9, 2021. © 2005 American Association for Cancer Research. Expression of HOXB2, a Retinoic Acid Signaling Target in Pancreatic Cancer and Pancreatic Intraepithelial Neoplasia

Davendra Segara, Andrew V. Biankin, James G. Kench, et al.

Clin Cancer Res 2005;11:3587-3596.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/9/3587

Cited articles This article cites 47 articles, 16 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/9/3587.full#ref-list-1

Citing articles This article has been cited by 19 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/9/3587.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 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/11/9/3587. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.