Expression of the Caudal-Type Homeodomain Transcription Factors CDX 1/2 and Outcome in Carcinomas of the Ampulla of Vater
http://www.100md.com
《临床肿瘤学》
the Departments of Pathology, Surgery, Oncology, and Oncology Biostatistics, Sidney Kimmel Comprehensive Cancer Center, The Johns Hopkins University School of Medicine, Baltimore, MD
ABSTRACT
PURPOSE: Adenocarcinomas of the ampulla of Vater demonstrate a characteristic histology but vary significantly in outcome. As a consequence, prognostic factors for these cancers are poorly defined. The caudal-type homeodomain transcription factors 1 (CDX1) and 2 (CDX2) regulate axial development and intestinal differentiation. We assessed the expression of these putative intestinal epithelial-specific transcription factors and their influence on patient outcome.
PATIENTS AND METHODS: Fifty-three resected carcinomas of the ampulla of Vater, 31 pancreatic ductal adenocarcinomas, and 15 extrahepatic biliary carcinomas were analyzed for CDX1 and CDX2 expression using immunohistochemistry.
RESULTS: Forty percent of carcinomas of the ampulla of Vater but less than 5% of pancreatic and biliary adenocarcinomas expressed CDX. Expression of CDX was associated with a better prognosis (P = .0009). Individually, both CDX1 (P = .02) and CDX2 (P = .02) expression were associated with a survival advantage on univariate analysis. Advanced T stage (P = .02), lymph node metastases (P = .004), and vascular space invasion (P = .0009) were associated with a poor prognosis. Multivariate analysis revealed vascular space invasion (P = .01) and CDX expression (P = .01) to be independent prognostic factors.
CONCLUSION: Expression of CDX was an independent marker of outcome in patients with resected adenocarcinoma of the ampulla of Vater. Expression of CDX may distinguish good prognosis intestinal-like tumors, which potentially arise within intestinal epithelium, from poorer prognosis pancreatobiliary tumors, which arise in adjacent pancreatic and/or biliary ductal epithelium.
INTRODUCTION
The ampulla of Vater is located at the confluence of the main biliary and pancreatic ducts at the duodenum. Consequently, it is composed of the following three distinct epithelial elements: duodenal mucosa, pancreatic ductal epithelium, and biliary ductal epithelium. Survival rates of patients with nonampullary carcinomas arising in the periampullary region differ in outcome based on epithelial origin, with operatively resected duodenal carcinomas demonstrating superior 5-year survival rates (33% to 59%) compared with biliary (27% to 47%) and pancreatic (5% to 15%) primaries. Patients with ampullary carcinomas have an intermediate 5-year survival rate of 35% to 50%.1–5 Despite the fact that the ampulla is composed of multiple epithelial elements, ampullary carcinomas share similar histologic features, which limit the ability to identify the specific epithelium of origin. Despite being similar in histologic appearance, these tumors demonstrate a wide spectrum of biologic behavior, and outcomes for patients vary considerably. In addition, the influence of pathologic factors, such as tumor size and lymph node (LN) involvement, differ substantially between studies,1,3–8 suggesting an inherent variability in the biology of these tumors. Because ampullary carcinomas are likely to be composed of neoplasms arising from one of the three types of epithelium that converge in this region, we hypothesized that markers of intestinal epithelium as compared with pancreatic or biliary epithelium may be valuable in determining prognosis of ampullary carcinomas.
The caudal-type homeodomain transcription factors 1 (CDX1) and 2 (CDX2) regulate axial development and intestinal differentiation, both during embryonic life and postnatally.9–11 In the adult, expression of CDX1 and CDX2 is restricted to intestinal and colonic epithelial cells, where these proteins have been implicated in regulation of gut epithelial differentiation.12,13 Specifically, CDX1 seems to be expressed within the intestinal crypts, whereas CDX2 is localized to the more differentiated cells of the mouse intestinal villus.14,15 Mice with a homozygous deletion of CDX1 demonstrate axial skeletal abnormalities but normal intestinal development.9 Homozygous deletion of CDX2 in mice leads to early embryonic death; however, heterozygotes develop multiple intestinal hamartomas early in life.10 Inactivation of CDX2 in chimeric mice results in the development of ectopic gastric epithelium in pericecal intestine and inability of affected cells to differentiate to an intestinal epithelial cell fate.16 Loss of CDX1 and/or CDX2 has been well documented in colon cancer and colon cancer cell lines13,17–19 and has been associated with a poor prognosis in colon cancer20 and gastric cancer.21 These associations may reflect the role of CDX in promoting a more differentiated cellular phenotype.22 Ectopic expression of CDX1 and CDX2 has been described in various gastrointestinal carcinomas, including gastric and esophageal cancer.19,23,24 In gastric epithelium, CDX expression is associated with the expression of the intestinal mucin MUC2 and the presence of intestinal-type metaplasia.23 Ectopic expression of CDX2 in the gastric epithelium of transgenic mice results in intestinal metaplasia, a precursor lesion of gastric cancer.25
Within the normal pancreas, CDX2 is present within non-beta cells of the islets of Langerhans, where it primarily regulates the production of glucagon26,27 but is only detected in a small subset of pancreatic ductal and acinar cells.24 Expression of CDX1 in the human pancreas has not been described.
Using tissue microarrays, we examined CDX1 and CDX2 expression in normal ampullary and periampullary tissue, as well as in a large group of resected ampullary, distal biliary, and pancreatic carcinomas. Expression of CDX1 and CDX2 was evident in a subset of ampullary carcinomas, and distinguished good prognosis intestinal-like tumors, which may arise within intestinal epithelium, from poorer prognosis pancreatobiliary tumors, which possibly arise in immediately adjacent pancreatic and/or biliary ductal epithelium.
PATIENTS AND METHODS
Patient Specimens and Microarray Construction
Permission for this study was obtained from the Johns Hopkins Joint Committee for Clinical Investigation. Paraffin-embedded material from a series of pancreaticoduodenectomy specimens (Whipple procedure) containing pancreatic ductal adenocarcinoma (n = 31), extrahepatic biliary tract carcinomas (n = 15), or ampullary cancers (n = 53) resected at the Johns Hopkins Hospital (Baltimore, MD) between 1996 and 2001 were used for this study. Seven tissue microarrays, including three ampullary cancer, two biliary carcinoma, and two pancreatic ductal adenocarcinoma arrays, were generated from these lesions.28 Each specimen was represented by up to four 1.4-mm cores on the tissue microarrays to obtain adequate representation of neoplastic cells.
Immunohistochemistry
Three- to four-micrometer sections from paraffin-embedded tissue (microarrays) were used for immunohistochemistry. Slides were deparaffinized in fresh xylene and rehydrated through sequential graded ethanols. Antigen retrieval was performed by citrate buffer incubation (10 mmol/L, pH 6) using a microwavable pressure cooker for 20 minutes. Slides were cooled for 20 minutes, incubated for 5 minutes with 3% hydrogen peroxide, washed in phosphate-buffered saline-0.1% Triton X-100 (pH 7.6), blocked for 20 minutes in 20% normal donkey serum, and incubated in appropriate antibody dilution for CDX1 (1:200, rabbit polyclonal; gift from D. Silberg, University of Pennsylvania School of Medicine, Philadelphia, PA) or CDX2 (1:400, mouse monoclonal; Biogenex, San Ramon, CA) overnight at 4°C. The next day, slides were washed in phosphate-buffered saline-0.1% Triton X-100 and incubated for 30 minutes in a 1:200 dilution of biotinylated antimouse or antirabbit secondary antibody. The ABC Elite kit (Vectastain, Burlingame, CA) with 3,3'-diaminobenzidine development was used to visualize antibody binding, and slides were subsequently counterstained with hematoxylin.
Immunolabeling Analysis and Scoring
Slides were examined by three independent, blinded observers (D.E.H., A.M., and A.V.B.) using an Olympus BH-2 microscope (Olympus Optical, Tokyo, Japan). The labeling index criterion was set at less than 25% v 25% positively labeled nuclei, using all hematoxylin and eosin-stained epithelial cell nuclei as a denominator, as based on previously published data for the gastrointestinal tract.24 Only nuclear labeling for CDX1 and CDX2 was included; occasional cytoplasmic reactivity was sporadically seen and was considered artifactual. Duodenal and colonic epithelia were present on each microarray slide and were used as a positive immunolabeling control.
Statistical Analysis and Patient Outcomes
Statistical evaluation was performed using 2 analysis of proportions, Kaplan-Meier survival for univariate survival analysis, and the Cox proportional hazards model for multivariate survival analysis using Statview 5.0 Software (Abacus Systems, Berkeley, CA). P < .05 was accepted as statistically significant. Those factors that were prognostic on univariate analysis were then assessed in a multivariate model to identify those factors that were independently prognostic and those that were the result of confounding. The variables assessed were patient age at diagnosis, tumor size, T stage, surgical margin involvement by tumor, presence of LN metastases, presence of vascular space invasion, degree of tumor differentiation, CDX expression (either CDX1, CDX2, or both), and CDX1 expression and CDX2 expression alone. All analyses are presented so that the worse risk is compared with the best risk, with the best risk being the first category and the worse risk being the second category in each case.
RESULTS
CDX1 and CDX2 Expression in Normal Ampulla of Vater, Pancreas, Duodenum, and Colon
CDX1 immunolabeling of the adult human pancreas did not reveal significant nuclear labeling of normal pancreatic ductal epithelium, islet cells, or acinar cells (Figs 1A and 1B). Normal duodenal and colonic epithelia, which were present on the same slide, demonstrated robust nuclear labeling in both basal and apical epithelial cells (Fig 1C). Similarly, CDX2 did not appear to label significant numbers of nuclei of pancreatic islets, acini, or ducts but demonstrated robust labeling throughout duodenal epithelium (Figs 1D to 1F). The epithelium of the ampulla of Vater was positive for CDX1 and CDX2 on its duodenal aspect, where the duodenal mucosa formed its wall, whereas the epithelium lining the ampulla and the main biliary and pancreatic duct as they entered the ampulla did not express either CDX1 or CDX2 (Fig 1F).
Ampullary Cancers Demonstrate Differential CDX Expression
We examined 53 ampullary carcinomas, 31 pancreatic ductal adenocarcinomas, and 15 extrahepatic biliary tract carcinomas for CDX1 and CDX2 expression. To characterize individual tumors based on the frequency of CDX1 and CDX2 expression, we used a nuclear labeling cutoff of 25% cells, as previously described for CDX2.24
Analysis of CDX1 and CDX2 immunolabeling revealed nuclear staining of either CDX1 or CDX2 expression in 21 (40%) of 53 ampullary carcinomas. In contrast, expression of CDX1 or CDX2 was present in only one (7%) of 15 extrahepatic biliary tract carcinomas and one (3%) of 31 pancreatic ductal adenocarcinomas (Figs 1G to 1J). Ampullary cancers demonstrated 25% nuclear labeling in 16 (30%) of 53 cancers for CDX1 and 14 (27%) of 53 cancers for CDX2. Nine (17%) of 53 cancers expressed both CDX1 and CDX2. Pancreatic ductal adenocarcinoma demonstrated 25% nuclear labeling in only one (3%) of 31 cancers for CDX1 and in zero (0%) of 31 cancers for CDX2. Extrahepatic biliary tract carcinomas demonstrated labeling of 25% in only one (7%) of 15 cancers for CDX1 and one (7%) of 15 cancers for CDX2.
CDX Expression Is Associated With Improved Patient Outcomes in Ampullary Carcinoma
The cohort consisted of patients with a median age of 68 years (range, 47 to 88 years); 57% of patients were male. The median cumulative survival for patients with resected ampullary carcinoma was 29 months, with a 1-year survival of 80% and a 5-year survival of 40%. There was one perioperative death, and three patients were lost to follow-up. Factors influencing survival on univariate analysis were LN metastases (median survival > 60 months for LN-negative tumors v 18.6 months for LN-positive tumors; P = .004), tumor stage (45 months for T1-2 v 18 months for T3-4; P = .02), and vascular space invasion (> 60 months for tumors with no evidence of vascular space invasion v 14 months for tumors with vascular space invasion; P = .0009; Table 1, Fig 2). Neither tumor size nor degree of differentiation influenced survival (P = .67; well and moderate v poor). Patient age at diagnosis influenced survival when modeled as a continuous variable (P = .007; hazard ratio [HR], 1.07; 95% CI, 1.02 to 1.12) where the HR reflects the increased risk of death per year of age. Therefore, a 10-year difference in age is equal to a HR of 1.0710 (= 1.97). Only one resected carcinoma involved surgical margins, precluding analysis of this variable (P = .95).
Expression of CDX (either CDX1 or CDX2, or both) was associated with a longer median survival (> 60 months for CDX-positive tumors v 20 months for CDX-negative tumors; P = .0009). Patients with CDX-positive tumors demonstrated a 1-year survival of 96% and a 5-year survival of 73%; patients with CDX-negative tumors demonstrated a 1-year survival of 73% and a 5-year survival of only 20%. Assessment of CDX1 and CDX2 expression individually demonstrated that expression of CDX1 within the ampullary carcinoma was associated with better median survival (> 60 months for CDX1-positive tumors v 20 months for CDX1-negative tumors; P = .02). Patients with CDX1-positive tumors demonstrated an actuarial 1-year survival of 100% and a 5-year survival of 68%; patients with CDX1-negative tumors had a 1-year survival of 79% and a 5-year survival of 30%. Expression of CDX2 was also associated with a significant survival advantage. Patients with CDX2-positive tumors had a median survival of more than 45 months v 20 months for patients who had CDX2-negative tumors (P = .02). The 1- and 5-year actuarial survival for patients with CDX2-positive tumors was 91% and 76%, respectively, versus 79% and 26%, respectively, in CDX2-negative patients. Although those patients whose tumors did not express CDX1 or CDX2 had a worse prognosis (P = .01) than patients who expressed either marker or both, numbers are too small to draw conclusions concerning differences in survival between patients who expressed either CDX1 or CDX2 alone and patients who expressed both markers. Expression of both CDX1 and CDX2 in a tumor was not associated with a survival advantage compared with the rest of patients (patients with tumors that did not express CDX at all or only expressed one of the CDX subtypes; P = .15). Results are listed in Table 1, and Kaplan-Meier survival curves are shown in Figure 2.
On multivariate analysis using the Cox proportional hazards model for all variables significant on univariate analysis, CDX expression (defined as CDX1 and/or CDX2 expression) and age at diagnosis were independent prognostic factors (Table 2A). Standard step-down variable selection procedures with removal of T stage and, subsequently, LN metastases resolved into the final model (Table 2B), which eliminated partial redundancies among these variables and identified CDX expression (P = .01; HR, 4.5; 95% CI, 1.51 to 13.4), vascular space invasion (P = .01; HR, 2.78; 95% CI, 1.23 to 6.25), and age at diagnosis (P = .01; HR, 1.07; 95% CI, 1.02 to 1.12) as independent prognostic factors. CDX1 and CDX2 expression were also modeled individually but were not presented together because of colinearity of the two variables (Table 2C). Step-down variable selection for models including CDX2 expression demonstrated that CDX2 expression (P = .04; HR, 3.68; 95% CI, 1.06 to 12.7), LN metastases (P = .04; HR, 2.63; 95% CI, 1.03 to 6.67), vascular space invasion (P = .02; HR, 2.78; 95% CI, 1.18 to 6.67), and age at diagnosis (P = .01; HR, 1.07; 95% CI, 1.02 to 1.12), but not T stage, were independent prognostic factors (Table 2D). CDX1 expression was associated with lower T stage (P = .02), lack of LN metastases (P = .01) on 2 analysis of proportions, and a trend against vascular space invasion (P = .06) and, therefore, was not an independent predictor of survival (Table 2E), with the final model incorporating CDX1 resolving to include LN metastases, age, and vascular space invasion (Table 2F).
DISCUSSION
CDX expression was detected in normal duodenal epithelium of the duodenal papilla and the periduodenal region of the ampulla of Vater but not in other regions of the ampulla or in the biliary or pancreatic ducts. Expression of CDX (CDX1 and/or CDX2) and CDX2 expression alone were independent predictors of survival after resection of ampullary carcinomas. Expression of CDX1, although prognostic on univariate analysis, was not an independent predictor of outcome, reflecting associations with lower T stage, lack of LN metastases, and a trend against vascular space invasion. Advanced T stage, LN metastases, vascular space invasion, and advancing age were associated with a poor prognosis on univariate analysis. Vascular space invasion and age but not T stage or LN status were independent predictors of survival on multivariate analysis when modeled with CDX expression.
Both CDX1 and CDX2 have been implicated in the maintenance of intestinal and colonic epithelial differentiation,13,14,19,25,29 and loss of expression is associated with increased tumorigenicity of colon cancer13 and colon cancer-derived cell lines.20 CDX2 may exert this effect through regulation of liver intestine-cadherin expression, which is involved in cell adhesion.30 CDX2 expression is associated with improved outcome in gastric cancer21 and is associated with intestinal metaplasia of gastric epithelium,23,30,31 which is a precursor lesion of gastric cancer. In the current study, CDX1 expression in ampullary carcinomas was associated with lower T stage and lack of LN metastases, with a trend against vascular space invasion. CDX2 seems to be an independent prognostic factor, suggesting that mechanisms apart from tumor stage and LN metastases are influencing patient outcome in patients with CDX2-positive tumors. CDX2 labeled a different subset of cancers compared with CDX1, and although numbers are too small to examine those tumors that expressed CDX2 and not CDX1 in multivariate analysis, the small degree of codependence of CDX2 expression with T stage and LN metastases in some of the multivariate models is likely to be the result of CDX1 expression in some CDX2-positive cancers.
It has been suggested that CDX expression, in particular CDX2, is a specific marker of intestinal differentiation. Intestinal metaplasia of the stomach and distal esophagus (Barrett's esophagus) is also characterized by expression of CDX.14,19,23,25,32 Because the ampulla is composed of epithelium from the duodenal mucosa and the biliary and pancreatic ducts, it is tempting to speculate that CDX labeling in tumors of the ampulla of Vater indicates an intestinal origin. However, other studies have reported an incidence of CDX expression in both biliary tract and pancreatic cancers,24 suggesting that this analogy is not absolute. It is also likely that some duodenal cancers do not express CDX because loss of CDX2 expression is a feature of colonic carcinogenesis17 and may also occur in duodenal carcinoma development.
In summary, expression of CDX influenced outcome in carcinomas of the ampulla of Vater. CDX2 expression was an independent marker of outcome, but CDX1 was associated with more favorable pathologic criteria and was not an independent prognostic factor. Expression of CDX is evident in a subset of ampullary carcinomas and may distinguish good prognosis intestinal-like tumors, which may arise within intestinal epithelium of the ampulla, from poor prognosis pancreatobiliary tumors, which may possibly arise in immediately adjacent pancreatic or biliary ductal epithelium. A mechanism by which CDX influences tumor behavior has yet to be elucidated.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank Debra Silberg, MD, PhD, University of Pennsylvania School of Medicine, Philadelphia, PA, for her gift of CDX1 antibodies.
NOTES
Supported by National Institutes of Health grant Nos. DK-56211 (S.D.L.) and T32-DK 077130 (Johns Hopkins Training Grant in Gastrointestinal Surgical Research), the family of Margaret Lee, the Ken Warren Fellowship of the International Hepato-Pancreato-Biliary Association (A.V.B.), the Neil Hamilton Fairley postdoctoral fellowship from the National Health and Medical Research Council of Australia (A.V.B.), and the Paul K. Neumann Professorship in Pancreatic Cancer at Johns Hopkins University (S.D.L.).
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Yeo CJ, Sohn TA, Cameron JL, et al: Periampullary adenocarcinoma: Analysis of 5-year survivors. Ann Surg 227:821-831, 1998
Naef M, Buhlmann M, Metzger D, et al: Periampullary carcinomas: A special entity of duodenal tumors. Swiss Surg 5:11-13, 1999
Sellner FJ, Riegler FM, Machacek E: Implications of histological grade of tumour for the prognosis of radically resected periampullary adenocarcinoma. Eur J Surg 165:865-870, 1999
Bouvet M, Gamagami RA, Gilpin EA, et al: Factors influencing survival after resection for periampullary neoplasms. Am J Surg 180:13-17, 2000
van Geenen RC, van Gulik TM, Offerhaus GJ, et al: Survival after pancreaticoduodenectomy for periampullary adenocarcinoma: An update. Eur J Surg Oncol 27:549-557, 2001
Willett CG, Warshaw AL, Convery K, et al: Patterns of failure after pancreaticoduodenectomy for ampullary carcinoma. Surg Gynecol Obstet 176:33-38, 1993
Matory YL, Gaynor J, Brennan M: Carcinoma of the ampulla of Vater. Surg Gynecol Obstet 177:366-370, 1993
Talamini MA, Moesinger RC, Pitt HA, et al: Adenocarcinoma of the ampulla of Vater: A 28-year experience. Ann Surg 225:590-599, 1997
Subramanian V, Meyer BI, Gruss P: Disruption of the murine homeobox gene Cdx1 affects axial skeletal identities by altering the mesodermal expression domains of Hox genes. Cell 83:641-653, 1995
Tamai Y, Nakajima R, Ishikawa T, et al: Colonic hamartoma development by anomalous duplication in Cdx2 knockout mice. Cancer Res 59:2965-2970, 1999
Beck F, Tata F, Chawengsaksophak K: Homeobox genes and gut development. Bioessays 22:431-441, 2000
Bonner CA, Loftus SK, Wasmuth JJ: Isolation, characterization, and precise physical localization of human CDX1, a caudal-type homeobox gene. Genomics 28:206-211, 1995
Mallo GV, Rechreche H, Frigerio JM, et al: Molecular cloning, sequencing and expression of the mRNA encoding human Cdx1 and Cdx2 homeobox: Down-regulation of Cdx1 and Cdx2 mRNA expression during colorectal carcinogenesis. Int J Cancer 74:35-44, 1997
Silberg DG, Swain GP, Suh ER, et al: Cdx1 and cdx2 expression during intestinal development. Gastroenterology 119:961-971, 2000
James R, Erler T, Kazenwadel J: Structure of the murine homeobox gene cdx-2: Expression in embryonic and adult intestinal epithelium. J Biol Chem 269:15229-15237, 1994
Beck F, Chawengsaksophak K, Luckett J, et al: A study of regional gut endoderm potency by analysis of Cdx2 null mutant chimaeric mice. Dev Biol 255:399-406, 2003
Ee HC, Erler T, Bhathal PS, et al: Cdx-2 homeodomain protein expression in human and rat colorectal adenoma and carcinoma. Am J Pathol 147:586-592, 1995
Vider BZ, Zimber A, Chastre E, et al: Deregulated expression of homeobox-containing genes, HOXB6, B8, C8, C9, and Cdx-1, in human colon cancer cell lines. Biochem Biophys Res Commun 272:513-518, 2000
Silberg DG, Furth EE, Taylor JK, et al: CDX1 protein expression in normal, metaplastic, and neoplastic human alimentary tract epithelium. Gastroenterology 113:478-486, 1997
Mallo GV, Soubeyran P, Lissitzky JC, et al: Expression of the Cdx1 and Cdx2 homeotic genes leads to reduced malignancy in colon cancer-derived cells. J Biol Chem 273:14030-14036, 1998
Mizoshita T, Tsukamoto T, Nakanishi H, et al: Expression of Cdx2 and the phenotype of advanced gastric cancers: Relationship with prognosis. J Cancer Res Clin Oncol 129:727-734, 2003
Suh E, Traber PG: An intestine-specific homeobox gene regulates proliferation and differentiation. Mol Cell Biol 16:619-625, 1996
Almeida R, Silva E, Santos-Silva F, et al: Expression of intestine-specific transcription factors, CDX1 and CDX2, in intestinal metaplasia and gastric carcinomas. J Pathol 199:36-40, 2003
Werling RW, Yaziji H, Bacchi CE, et al: CDX2, a highly sensitive and specific marker of adenocarcinomas of intestinal origin: An immunohistochemical survey of 476 primary and metastatic carcinomas. Am J Surg Pathol 27:303-310, 2003
Silberg DG, Sullivan J, Kang E, et al: Cdx2 ectopic expression induces gastric intestinal metaplasia in transgenic mice. Gastroenterology 122:689-696, 2002
Jin T, Drucker DJ: Activation of proglucagon gene transcription through a novel promoter element by the caudal-related homeodomain protein cdx-2/3. Mol Cell Biol 16:19-28, 1996
Laser B, Meda P, Constant I, et al: The caudal-related homeodomain protein Cdx-2/3 regulates glucagon gene expression in islet cells. J Biol Chem 271:28984-28994, 1996
van Heek NT, Meeker AK, Kern SE, et al: Telomere shortening is nearly universal in pancreatic intraepithelial neoplasia. Am J Pathol 161:1541-1547, 2002
Mizoshita T, Inada K, Tsukamoto T, et al: Expression of Cdx1 and Cdx2 mRNAs and relevance of this expression to differentiation in human gastrointestinal mucosa-with special emphasis on participation in intestinal metaplasia of the human stomach. Gastric Cancer 4:185-191, 2001
Hinoi T, Lucas PC, Kuick R, et al: CDX2 regulates liver intestine-cadherin expression in normal and malignant colon epithelium and intestinal metaplasia. Gastroenterology 123:1565-1577, 2002
Mizoshita T, Inada KI, Tsukamoto T, et al: Expression of the intestine-specific transcription factors, Cdx1 and Cdx2, correlates shift to an intestinal phenotype in gastric cancer cells. J Cancer Res Clin Oncol 130:29-36, 2004
Phillips KW, Goldsworthy SM, Bennett LM, et al: Brca1 is expressed independently of hormonal stimulation in the mouse ovary. Lab Invest 76:419-425, 1997(Donna E. Hansel, Anirban )
ABSTRACT
PURPOSE: Adenocarcinomas of the ampulla of Vater demonstrate a characteristic histology but vary significantly in outcome. As a consequence, prognostic factors for these cancers are poorly defined. The caudal-type homeodomain transcription factors 1 (CDX1) and 2 (CDX2) regulate axial development and intestinal differentiation. We assessed the expression of these putative intestinal epithelial-specific transcription factors and their influence on patient outcome.
PATIENTS AND METHODS: Fifty-three resected carcinomas of the ampulla of Vater, 31 pancreatic ductal adenocarcinomas, and 15 extrahepatic biliary carcinomas were analyzed for CDX1 and CDX2 expression using immunohistochemistry.
RESULTS: Forty percent of carcinomas of the ampulla of Vater but less than 5% of pancreatic and biliary adenocarcinomas expressed CDX. Expression of CDX was associated with a better prognosis (P = .0009). Individually, both CDX1 (P = .02) and CDX2 (P = .02) expression were associated with a survival advantage on univariate analysis. Advanced T stage (P = .02), lymph node metastases (P = .004), and vascular space invasion (P = .0009) were associated with a poor prognosis. Multivariate analysis revealed vascular space invasion (P = .01) and CDX expression (P = .01) to be independent prognostic factors.
CONCLUSION: Expression of CDX was an independent marker of outcome in patients with resected adenocarcinoma of the ampulla of Vater. Expression of CDX may distinguish good prognosis intestinal-like tumors, which potentially arise within intestinal epithelium, from poorer prognosis pancreatobiliary tumors, which arise in adjacent pancreatic and/or biliary ductal epithelium.
INTRODUCTION
The ampulla of Vater is located at the confluence of the main biliary and pancreatic ducts at the duodenum. Consequently, it is composed of the following three distinct epithelial elements: duodenal mucosa, pancreatic ductal epithelium, and biliary ductal epithelium. Survival rates of patients with nonampullary carcinomas arising in the periampullary region differ in outcome based on epithelial origin, with operatively resected duodenal carcinomas demonstrating superior 5-year survival rates (33% to 59%) compared with biliary (27% to 47%) and pancreatic (5% to 15%) primaries. Patients with ampullary carcinomas have an intermediate 5-year survival rate of 35% to 50%.1–5 Despite the fact that the ampulla is composed of multiple epithelial elements, ampullary carcinomas share similar histologic features, which limit the ability to identify the specific epithelium of origin. Despite being similar in histologic appearance, these tumors demonstrate a wide spectrum of biologic behavior, and outcomes for patients vary considerably. In addition, the influence of pathologic factors, such as tumor size and lymph node (LN) involvement, differ substantially between studies,1,3–8 suggesting an inherent variability in the biology of these tumors. Because ampullary carcinomas are likely to be composed of neoplasms arising from one of the three types of epithelium that converge in this region, we hypothesized that markers of intestinal epithelium as compared with pancreatic or biliary epithelium may be valuable in determining prognosis of ampullary carcinomas.
The caudal-type homeodomain transcription factors 1 (CDX1) and 2 (CDX2) regulate axial development and intestinal differentiation, both during embryonic life and postnatally.9–11 In the adult, expression of CDX1 and CDX2 is restricted to intestinal and colonic epithelial cells, where these proteins have been implicated in regulation of gut epithelial differentiation.12,13 Specifically, CDX1 seems to be expressed within the intestinal crypts, whereas CDX2 is localized to the more differentiated cells of the mouse intestinal villus.14,15 Mice with a homozygous deletion of CDX1 demonstrate axial skeletal abnormalities but normal intestinal development.9 Homozygous deletion of CDX2 in mice leads to early embryonic death; however, heterozygotes develop multiple intestinal hamartomas early in life.10 Inactivation of CDX2 in chimeric mice results in the development of ectopic gastric epithelium in pericecal intestine and inability of affected cells to differentiate to an intestinal epithelial cell fate.16 Loss of CDX1 and/or CDX2 has been well documented in colon cancer and colon cancer cell lines13,17–19 and has been associated with a poor prognosis in colon cancer20 and gastric cancer.21 These associations may reflect the role of CDX in promoting a more differentiated cellular phenotype.22 Ectopic expression of CDX1 and CDX2 has been described in various gastrointestinal carcinomas, including gastric and esophageal cancer.19,23,24 In gastric epithelium, CDX expression is associated with the expression of the intestinal mucin MUC2 and the presence of intestinal-type metaplasia.23 Ectopic expression of CDX2 in the gastric epithelium of transgenic mice results in intestinal metaplasia, a precursor lesion of gastric cancer.25
Within the normal pancreas, CDX2 is present within non-beta cells of the islets of Langerhans, where it primarily regulates the production of glucagon26,27 but is only detected in a small subset of pancreatic ductal and acinar cells.24 Expression of CDX1 in the human pancreas has not been described.
Using tissue microarrays, we examined CDX1 and CDX2 expression in normal ampullary and periampullary tissue, as well as in a large group of resected ampullary, distal biliary, and pancreatic carcinomas. Expression of CDX1 and CDX2 was evident in a subset of ampullary carcinomas, and distinguished good prognosis intestinal-like tumors, which may arise within intestinal epithelium, from poorer prognosis pancreatobiliary tumors, which possibly arise in immediately adjacent pancreatic and/or biliary ductal epithelium.
PATIENTS AND METHODS
Patient Specimens and Microarray Construction
Permission for this study was obtained from the Johns Hopkins Joint Committee for Clinical Investigation. Paraffin-embedded material from a series of pancreaticoduodenectomy specimens (Whipple procedure) containing pancreatic ductal adenocarcinoma (n = 31), extrahepatic biliary tract carcinomas (n = 15), or ampullary cancers (n = 53) resected at the Johns Hopkins Hospital (Baltimore, MD) between 1996 and 2001 were used for this study. Seven tissue microarrays, including three ampullary cancer, two biliary carcinoma, and two pancreatic ductal adenocarcinoma arrays, were generated from these lesions.28 Each specimen was represented by up to four 1.4-mm cores on the tissue microarrays to obtain adequate representation of neoplastic cells.
Immunohistochemistry
Three- to four-micrometer sections from paraffin-embedded tissue (microarrays) were used for immunohistochemistry. Slides were deparaffinized in fresh xylene and rehydrated through sequential graded ethanols. Antigen retrieval was performed by citrate buffer incubation (10 mmol/L, pH 6) using a microwavable pressure cooker for 20 minutes. Slides were cooled for 20 minutes, incubated for 5 minutes with 3% hydrogen peroxide, washed in phosphate-buffered saline-0.1% Triton X-100 (pH 7.6), blocked for 20 minutes in 20% normal donkey serum, and incubated in appropriate antibody dilution for CDX1 (1:200, rabbit polyclonal; gift from D. Silberg, University of Pennsylvania School of Medicine, Philadelphia, PA) or CDX2 (1:400, mouse monoclonal; Biogenex, San Ramon, CA) overnight at 4°C. The next day, slides were washed in phosphate-buffered saline-0.1% Triton X-100 and incubated for 30 minutes in a 1:200 dilution of biotinylated antimouse or antirabbit secondary antibody. The ABC Elite kit (Vectastain, Burlingame, CA) with 3,3'-diaminobenzidine development was used to visualize antibody binding, and slides were subsequently counterstained with hematoxylin.
Immunolabeling Analysis and Scoring
Slides were examined by three independent, blinded observers (D.E.H., A.M., and A.V.B.) using an Olympus BH-2 microscope (Olympus Optical, Tokyo, Japan). The labeling index criterion was set at less than 25% v 25% positively labeled nuclei, using all hematoxylin and eosin-stained epithelial cell nuclei as a denominator, as based on previously published data for the gastrointestinal tract.24 Only nuclear labeling for CDX1 and CDX2 was included; occasional cytoplasmic reactivity was sporadically seen and was considered artifactual. Duodenal and colonic epithelia were present on each microarray slide and were used as a positive immunolabeling control.
Statistical Analysis and Patient Outcomes
Statistical evaluation was performed using 2 analysis of proportions, Kaplan-Meier survival for univariate survival analysis, and the Cox proportional hazards model for multivariate survival analysis using Statview 5.0 Software (Abacus Systems, Berkeley, CA). P < .05 was accepted as statistically significant. Those factors that were prognostic on univariate analysis were then assessed in a multivariate model to identify those factors that were independently prognostic and those that were the result of confounding. The variables assessed were patient age at diagnosis, tumor size, T stage, surgical margin involvement by tumor, presence of LN metastases, presence of vascular space invasion, degree of tumor differentiation, CDX expression (either CDX1, CDX2, or both), and CDX1 expression and CDX2 expression alone. All analyses are presented so that the worse risk is compared with the best risk, with the best risk being the first category and the worse risk being the second category in each case.
RESULTS
CDX1 and CDX2 Expression in Normal Ampulla of Vater, Pancreas, Duodenum, and Colon
CDX1 immunolabeling of the adult human pancreas did not reveal significant nuclear labeling of normal pancreatic ductal epithelium, islet cells, or acinar cells (Figs 1A and 1B). Normal duodenal and colonic epithelia, which were present on the same slide, demonstrated robust nuclear labeling in both basal and apical epithelial cells (Fig 1C). Similarly, CDX2 did not appear to label significant numbers of nuclei of pancreatic islets, acini, or ducts but demonstrated robust labeling throughout duodenal epithelium (Figs 1D to 1F). The epithelium of the ampulla of Vater was positive for CDX1 and CDX2 on its duodenal aspect, where the duodenal mucosa formed its wall, whereas the epithelium lining the ampulla and the main biliary and pancreatic duct as they entered the ampulla did not express either CDX1 or CDX2 (Fig 1F).
Ampullary Cancers Demonstrate Differential CDX Expression
We examined 53 ampullary carcinomas, 31 pancreatic ductal adenocarcinomas, and 15 extrahepatic biliary tract carcinomas for CDX1 and CDX2 expression. To characterize individual tumors based on the frequency of CDX1 and CDX2 expression, we used a nuclear labeling cutoff of 25% cells, as previously described for CDX2.24
Analysis of CDX1 and CDX2 immunolabeling revealed nuclear staining of either CDX1 or CDX2 expression in 21 (40%) of 53 ampullary carcinomas. In contrast, expression of CDX1 or CDX2 was present in only one (7%) of 15 extrahepatic biliary tract carcinomas and one (3%) of 31 pancreatic ductal adenocarcinomas (Figs 1G to 1J). Ampullary cancers demonstrated 25% nuclear labeling in 16 (30%) of 53 cancers for CDX1 and 14 (27%) of 53 cancers for CDX2. Nine (17%) of 53 cancers expressed both CDX1 and CDX2. Pancreatic ductal adenocarcinoma demonstrated 25% nuclear labeling in only one (3%) of 31 cancers for CDX1 and in zero (0%) of 31 cancers for CDX2. Extrahepatic biliary tract carcinomas demonstrated labeling of 25% in only one (7%) of 15 cancers for CDX1 and one (7%) of 15 cancers for CDX2.
CDX Expression Is Associated With Improved Patient Outcomes in Ampullary Carcinoma
The cohort consisted of patients with a median age of 68 years (range, 47 to 88 years); 57% of patients were male. The median cumulative survival for patients with resected ampullary carcinoma was 29 months, with a 1-year survival of 80% and a 5-year survival of 40%. There was one perioperative death, and three patients were lost to follow-up. Factors influencing survival on univariate analysis were LN metastases (median survival > 60 months for LN-negative tumors v 18.6 months for LN-positive tumors; P = .004), tumor stage (45 months for T1-2 v 18 months for T3-4; P = .02), and vascular space invasion (> 60 months for tumors with no evidence of vascular space invasion v 14 months for tumors with vascular space invasion; P = .0009; Table 1, Fig 2). Neither tumor size nor degree of differentiation influenced survival (P = .67; well and moderate v poor). Patient age at diagnosis influenced survival when modeled as a continuous variable (P = .007; hazard ratio [HR], 1.07; 95% CI, 1.02 to 1.12) where the HR reflects the increased risk of death per year of age. Therefore, a 10-year difference in age is equal to a HR of 1.0710 (= 1.97). Only one resected carcinoma involved surgical margins, precluding analysis of this variable (P = .95).
Expression of CDX (either CDX1 or CDX2, or both) was associated with a longer median survival (> 60 months for CDX-positive tumors v 20 months for CDX-negative tumors; P = .0009). Patients with CDX-positive tumors demonstrated a 1-year survival of 96% and a 5-year survival of 73%; patients with CDX-negative tumors demonstrated a 1-year survival of 73% and a 5-year survival of only 20%. Assessment of CDX1 and CDX2 expression individually demonstrated that expression of CDX1 within the ampullary carcinoma was associated with better median survival (> 60 months for CDX1-positive tumors v 20 months for CDX1-negative tumors; P = .02). Patients with CDX1-positive tumors demonstrated an actuarial 1-year survival of 100% and a 5-year survival of 68%; patients with CDX1-negative tumors had a 1-year survival of 79% and a 5-year survival of 30%. Expression of CDX2 was also associated with a significant survival advantage. Patients with CDX2-positive tumors had a median survival of more than 45 months v 20 months for patients who had CDX2-negative tumors (P = .02). The 1- and 5-year actuarial survival for patients with CDX2-positive tumors was 91% and 76%, respectively, versus 79% and 26%, respectively, in CDX2-negative patients. Although those patients whose tumors did not express CDX1 or CDX2 had a worse prognosis (P = .01) than patients who expressed either marker or both, numbers are too small to draw conclusions concerning differences in survival between patients who expressed either CDX1 or CDX2 alone and patients who expressed both markers. Expression of both CDX1 and CDX2 in a tumor was not associated with a survival advantage compared with the rest of patients (patients with tumors that did not express CDX at all or only expressed one of the CDX subtypes; P = .15). Results are listed in Table 1, and Kaplan-Meier survival curves are shown in Figure 2.
On multivariate analysis using the Cox proportional hazards model for all variables significant on univariate analysis, CDX expression (defined as CDX1 and/or CDX2 expression) and age at diagnosis were independent prognostic factors (Table 2A). Standard step-down variable selection procedures with removal of T stage and, subsequently, LN metastases resolved into the final model (Table 2B), which eliminated partial redundancies among these variables and identified CDX expression (P = .01; HR, 4.5; 95% CI, 1.51 to 13.4), vascular space invasion (P = .01; HR, 2.78; 95% CI, 1.23 to 6.25), and age at diagnosis (P = .01; HR, 1.07; 95% CI, 1.02 to 1.12) as independent prognostic factors. CDX1 and CDX2 expression were also modeled individually but were not presented together because of colinearity of the two variables (Table 2C). Step-down variable selection for models including CDX2 expression demonstrated that CDX2 expression (P = .04; HR, 3.68; 95% CI, 1.06 to 12.7), LN metastases (P = .04; HR, 2.63; 95% CI, 1.03 to 6.67), vascular space invasion (P = .02; HR, 2.78; 95% CI, 1.18 to 6.67), and age at diagnosis (P = .01; HR, 1.07; 95% CI, 1.02 to 1.12), but not T stage, were independent prognostic factors (Table 2D). CDX1 expression was associated with lower T stage (P = .02), lack of LN metastases (P = .01) on 2 analysis of proportions, and a trend against vascular space invasion (P = .06) and, therefore, was not an independent predictor of survival (Table 2E), with the final model incorporating CDX1 resolving to include LN metastases, age, and vascular space invasion (Table 2F).
DISCUSSION
CDX expression was detected in normal duodenal epithelium of the duodenal papilla and the periduodenal region of the ampulla of Vater but not in other regions of the ampulla or in the biliary or pancreatic ducts. Expression of CDX (CDX1 and/or CDX2) and CDX2 expression alone were independent predictors of survival after resection of ampullary carcinomas. Expression of CDX1, although prognostic on univariate analysis, was not an independent predictor of outcome, reflecting associations with lower T stage, lack of LN metastases, and a trend against vascular space invasion. Advanced T stage, LN metastases, vascular space invasion, and advancing age were associated with a poor prognosis on univariate analysis. Vascular space invasion and age but not T stage or LN status were independent predictors of survival on multivariate analysis when modeled with CDX expression.
Both CDX1 and CDX2 have been implicated in the maintenance of intestinal and colonic epithelial differentiation,13,14,19,25,29 and loss of expression is associated with increased tumorigenicity of colon cancer13 and colon cancer-derived cell lines.20 CDX2 may exert this effect through regulation of liver intestine-cadherin expression, which is involved in cell adhesion.30 CDX2 expression is associated with improved outcome in gastric cancer21 and is associated with intestinal metaplasia of gastric epithelium,23,30,31 which is a precursor lesion of gastric cancer. In the current study, CDX1 expression in ampullary carcinomas was associated with lower T stage and lack of LN metastases, with a trend against vascular space invasion. CDX2 seems to be an independent prognostic factor, suggesting that mechanisms apart from tumor stage and LN metastases are influencing patient outcome in patients with CDX2-positive tumors. CDX2 labeled a different subset of cancers compared with CDX1, and although numbers are too small to examine those tumors that expressed CDX2 and not CDX1 in multivariate analysis, the small degree of codependence of CDX2 expression with T stage and LN metastases in some of the multivariate models is likely to be the result of CDX1 expression in some CDX2-positive cancers.
It has been suggested that CDX expression, in particular CDX2, is a specific marker of intestinal differentiation. Intestinal metaplasia of the stomach and distal esophagus (Barrett's esophagus) is also characterized by expression of CDX.14,19,23,25,32 Because the ampulla is composed of epithelium from the duodenal mucosa and the biliary and pancreatic ducts, it is tempting to speculate that CDX labeling in tumors of the ampulla of Vater indicates an intestinal origin. However, other studies have reported an incidence of CDX expression in both biliary tract and pancreatic cancers,24 suggesting that this analogy is not absolute. It is also likely that some duodenal cancers do not express CDX because loss of CDX2 expression is a feature of colonic carcinogenesis17 and may also occur in duodenal carcinoma development.
In summary, expression of CDX influenced outcome in carcinomas of the ampulla of Vater. CDX2 expression was an independent marker of outcome, but CDX1 was associated with more favorable pathologic criteria and was not an independent prognostic factor. Expression of CDX is evident in a subset of ampullary carcinomas and may distinguish good prognosis intestinal-like tumors, which may arise within intestinal epithelium of the ampulla, from poor prognosis pancreatobiliary tumors, which may possibly arise in immediately adjacent pancreatic or biliary ductal epithelium. A mechanism by which CDX influences tumor behavior has yet to be elucidated.
Authors' Disclosures of Potential Conflicts of Interest
The authors indicated no potential conflicts of interest.
Acknowledgment
We thank Debra Silberg, MD, PhD, University of Pennsylvania School of Medicine, Philadelphia, PA, for her gift of CDX1 antibodies.
NOTES
Supported by National Institutes of Health grant Nos. DK-56211 (S.D.L.) and T32-DK 077130 (Johns Hopkins Training Grant in Gastrointestinal Surgical Research), the family of Margaret Lee, the Ken Warren Fellowship of the International Hepato-Pancreato-Biliary Association (A.V.B.), the Neil Hamilton Fairley postdoctoral fellowship from the National Health and Medical Research Council of Australia (A.V.B.), and the Paul K. Neumann Professorship in Pancreatic Cancer at Johns Hopkins University (S.D.L.).
Authors' disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
Yeo CJ, Sohn TA, Cameron JL, et al: Periampullary adenocarcinoma: Analysis of 5-year survivors. Ann Surg 227:821-831, 1998
Naef M, Buhlmann M, Metzger D, et al: Periampullary carcinomas: A special entity of duodenal tumors. Swiss Surg 5:11-13, 1999
Sellner FJ, Riegler FM, Machacek E: Implications of histological grade of tumour for the prognosis of radically resected periampullary adenocarcinoma. Eur J Surg 165:865-870, 1999
Bouvet M, Gamagami RA, Gilpin EA, et al: Factors influencing survival after resection for periampullary neoplasms. Am J Surg 180:13-17, 2000
van Geenen RC, van Gulik TM, Offerhaus GJ, et al: Survival after pancreaticoduodenectomy for periampullary adenocarcinoma: An update. Eur J Surg Oncol 27:549-557, 2001
Willett CG, Warshaw AL, Convery K, et al: Patterns of failure after pancreaticoduodenectomy for ampullary carcinoma. Surg Gynecol Obstet 176:33-38, 1993
Matory YL, Gaynor J, Brennan M: Carcinoma of the ampulla of Vater. Surg Gynecol Obstet 177:366-370, 1993
Talamini MA, Moesinger RC, Pitt HA, et al: Adenocarcinoma of the ampulla of Vater: A 28-year experience. Ann Surg 225:590-599, 1997
Subramanian V, Meyer BI, Gruss P: Disruption of the murine homeobox gene Cdx1 affects axial skeletal identities by altering the mesodermal expression domains of Hox genes. Cell 83:641-653, 1995
Tamai Y, Nakajima R, Ishikawa T, et al: Colonic hamartoma development by anomalous duplication in Cdx2 knockout mice. Cancer Res 59:2965-2970, 1999
Beck F, Tata F, Chawengsaksophak K: Homeobox genes and gut development. Bioessays 22:431-441, 2000
Bonner CA, Loftus SK, Wasmuth JJ: Isolation, characterization, and precise physical localization of human CDX1, a caudal-type homeobox gene. Genomics 28:206-211, 1995
Mallo GV, Rechreche H, Frigerio JM, et al: Molecular cloning, sequencing and expression of the mRNA encoding human Cdx1 and Cdx2 homeobox: Down-regulation of Cdx1 and Cdx2 mRNA expression during colorectal carcinogenesis. Int J Cancer 74:35-44, 1997
Silberg DG, Swain GP, Suh ER, et al: Cdx1 and cdx2 expression during intestinal development. Gastroenterology 119:961-971, 2000
James R, Erler T, Kazenwadel J: Structure of the murine homeobox gene cdx-2: Expression in embryonic and adult intestinal epithelium. J Biol Chem 269:15229-15237, 1994
Beck F, Chawengsaksophak K, Luckett J, et al: A study of regional gut endoderm potency by analysis of Cdx2 null mutant chimaeric mice. Dev Biol 255:399-406, 2003
Ee HC, Erler T, Bhathal PS, et al: Cdx-2 homeodomain protein expression in human and rat colorectal adenoma and carcinoma. Am J Pathol 147:586-592, 1995
Vider BZ, Zimber A, Chastre E, et al: Deregulated expression of homeobox-containing genes, HOXB6, B8, C8, C9, and Cdx-1, in human colon cancer cell lines. Biochem Biophys Res Commun 272:513-518, 2000
Silberg DG, Furth EE, Taylor JK, et al: CDX1 protein expression in normal, metaplastic, and neoplastic human alimentary tract epithelium. Gastroenterology 113:478-486, 1997
Mallo GV, Soubeyran P, Lissitzky JC, et al: Expression of the Cdx1 and Cdx2 homeotic genes leads to reduced malignancy in colon cancer-derived cells. J Biol Chem 273:14030-14036, 1998
Mizoshita T, Tsukamoto T, Nakanishi H, et al: Expression of Cdx2 and the phenotype of advanced gastric cancers: Relationship with prognosis. J Cancer Res Clin Oncol 129:727-734, 2003
Suh E, Traber PG: An intestine-specific homeobox gene regulates proliferation and differentiation. Mol Cell Biol 16:619-625, 1996
Almeida R, Silva E, Santos-Silva F, et al: Expression of intestine-specific transcription factors, CDX1 and CDX2, in intestinal metaplasia and gastric carcinomas. J Pathol 199:36-40, 2003
Werling RW, Yaziji H, Bacchi CE, et al: CDX2, a highly sensitive and specific marker of adenocarcinomas of intestinal origin: An immunohistochemical survey of 476 primary and metastatic carcinomas. Am J Surg Pathol 27:303-310, 2003
Silberg DG, Sullivan J, Kang E, et al: Cdx2 ectopic expression induces gastric intestinal metaplasia in transgenic mice. Gastroenterology 122:689-696, 2002
Jin T, Drucker DJ: Activation of proglucagon gene transcription through a novel promoter element by the caudal-related homeodomain protein cdx-2/3. Mol Cell Biol 16:19-28, 1996
Laser B, Meda P, Constant I, et al: The caudal-related homeodomain protein Cdx-2/3 regulates glucagon gene expression in islet cells. J Biol Chem 271:28984-28994, 1996
van Heek NT, Meeker AK, Kern SE, et al: Telomere shortening is nearly universal in pancreatic intraepithelial neoplasia. Am J Pathol 161:1541-1547, 2002
Mizoshita T, Inada K, Tsukamoto T, et al: Expression of Cdx1 and Cdx2 mRNAs and relevance of this expression to differentiation in human gastrointestinal mucosa-with special emphasis on participation in intestinal metaplasia of the human stomach. Gastric Cancer 4:185-191, 2001
Hinoi T, Lucas PC, Kuick R, et al: CDX2 regulates liver intestine-cadherin expression in normal and malignant colon epithelium and intestinal metaplasia. Gastroenterology 123:1565-1577, 2002
Mizoshita T, Inada KI, Tsukamoto T, et al: Expression of the intestine-specific transcription factors, Cdx1 and Cdx2, correlates shift to an intestinal phenotype in gastric cancer cells. J Cancer Res Clin Oncol 130:29-36, 2004
Phillips KW, Goldsworthy SM, Bennett LM, et al: Brca1 is expressed independently of hormonal stimulation in the mouse ovary. Lab Invest 76:419-425, 1997(Donna E. Hansel, Anirban )