ExpressionofCdx2andHepatocyteAntigeninGastricCarcinoma ... · histostaining reagent, with PV-6001, (Dako, Glostrop, Denmark) as the secondary antibody and were visualized with diaminobenzedine

Expression of Cdx2 and HepatocyteAntigen in Gastric Carcinoma:Correlationwith HistologicType and Implications for PrognosisZhaoqing Fan,1Jiyou Li,2 Bin Dong,2 and Xinfu Huang1

Abstract Purpose:This studywas designed to (a) analyze the correlationbetween the expressionofCdx2and Hep and the clinicopathologic features of patients with gastric carcinoma, and (b) determinethe value of combined analysis of Cdx2 andHep expression in distinguishinghistologic types andprognoses of gastric carcinomas.Experimental Design:The expression of Cdx2 and Hep were studied using immunohisto-chemistry of paraffin-embedded tumor specimens from 109 patients who underwent D2resection for gastric adenocarcinoma from 1995 to 1998.Results:Nuclear Cdx2 andHep expressionwas detected in 36.7% (40 of109) and 54.1% (59 of109) of gastric carcinoma cases, respectively. Expression of Cdx2 and Hep was significantlyhigher in intestinal-type carcinomas than in diffuse-type carcinomas (P = 0.027 and P = 0.037,respectively). There was a clear negative correlation between Cdx2 expression and lymph nodemetastasis (P = 0.029), as well as between Hep expression and depth of wall invasion(P = 0.011). The patients with Cdx2-positive or Hep-positive expression shows higher survivalrate than those with Cdx2-negative or Hep-negative expression (P = 0.0008 and P = 0.003,respectively). Multivariate analysis revealed that the expression of Cdx2 and Hep were indepen-dent prognostic indicators of gastric carcinoma. The combination of Cdx2 and Hep expressionwas significantly lower in diffuse-type carcinoma than in intestinal or mixed-type carcinoma.Multivariate analysis revealed that Cdx2 and Hep expression was an independent prognosticindicator of gastric carcinoma (P < 0.001).Conclusions:These data suggest that combined analysis of Cdx2 and Hep has significant valuein distinguishinghistologic types and in predicting the prognosis of gastric carcinoma.

Gastric carcinoma is one of the most common malignantdiseases and is the second most common cause of cancer-related death in China and in the world (1).

Gastric carcinoma is a markedly heterogeneous disease inhistologic feature and biological characters, especially in theadvanced stages. For pathologic diagnosis and classificationof gastric carcinoma, two classification systems, the Laurenand WHO systems, are commonly used. The number ofclinical documents showed that the biological behavior andprognosis could be significantly different among the patientswith the same stage, histologic type, or differentiation grade.Therefore, searching for the biomarkers to indicate the bio-logical characters, and predicting the outcome of patients

with gastric carcinoma, is the major focus of research ongastric carcinoma.

Gastric carcinoma can be divided into two main types bythe Lauren classification system (2): intestinal and diffuse-type carcinomas. Intestinal-type carcinoma is thought todevelop from well-characterized sequential stages that includechronic gastritis, atrophy, intestinal metaplasia, and dysplasia(3). Recently, it was observed that intestinal-type gastriccarcinoma developed from intestinal metaplasia in Cdx2-transgenic mice (4). This is the first direct evidence whichshows that intestinal metaplasia is the precancerous lesion ofgastric carcinoma.

The human Cdx2 genes are the homologues of the caudal-related homeobox gene in the Drosophila . In normal adults,

Cdx2 is expressed in mucosa from the duodenum to the distalcolon but not in gastric or esophageal mucosa (5). As trans-cription factors, Cdx2 proteins are important for regulating theproliferation and differentiation of intestinal cells and main-

taining the intestinal phenotype by binding the promoters ofseveral intestine-specific genes and activating transcriptionprocesses. Cdx2 was reported to be involved in tumorigenesisof the colon (6) and is known to be a tumor suppressor (7).

Studies have shown that Cdx2 expression is absent in normalgastric mucosa but is present in >85% of intestinal meta-plasias (8–10), suggesting that Cdx2 might be associated withintestinal metaplasia formation. At the stage of chronicgastritis without intestinal metaplasia, expression of Cdx2

Imaging, Diagnosis, Prognosis

Authors’ Affiliations:Departments of 1Surgery and 2Pathology, Peking UniversitySchool of Oncology and Beijing Cancer Hospital, Beijing, P.R. ChinaReceived 2/7/05; revised 5/10/05; accepted 6/7/05.Grant support: Key Technologies Research and Development Program2002BA711A06 and National Basic Research Priorities Program 973 Project1998051203 from the Ministry of Science and Technology of China, and grantH020920030390 from the Beijing Science andTechnology Commission.The costs of publication of this article were defrayed in part by the payment of pagecharges.This article must therefore be hereby marked advertisement in accordancewith18 U.S.C. Section1734 solely to indicate this fact.Requests for reprints: Jiyou Li, Peking University School of Oncology andBeijing Cancer Hospital, Haidian District, No. 52 Fucheng Road, Beijing 100036,P.R. China. Phone: 86-10-8812-2450; E-mail: [email protected].

F2005 American Association for Cancer Research.doi:10.1158/1078-0432.CCR-05-0278

www.aacrjournals.orgClin Cancer Res 2005;11(17) September1, 2005 6162

Cancer Research. on February 26, 2021. © 2005 American Association forclincancerres.aacrjournals.org Downloaded from

http://clincancerres.aacrjournals.org/

but not Cdx1, or other gene markers of intestinal metaplasiawere detected. These findings suggest that the expression ofCdx2 precedes that of Cdx1 and other intestine-specific genesin the progression of intestinal metaplasia from chronicgastritis. Therefore, Cdx2 might be the triggering factor forintestinal metaplasia (10). Although expression of Cdx2 hasbeen detected in some gastric carcinomas (8), few studiesreported the relationship between Cdx2 expression andprognosis of gastric carcinoma (11, 12).

The epitope of hepatocyte antigen (Hep) is believed to be acomponent of the membrane of hepatocellular mitochondria(13). Due to the high sensitivity and specificity of Hepexpression in normal and carcinomatous liver tissue, Hep isthought to be a useful marker for the diagnosis ofhepatocellular carcinoma and for differential diagnosis ofprimary and metastatic hepatic malignancies (14, 15).Recently, Hep expression was found to be high in theintestinal epithelium (16, 17) and was found in 100% ofintestinal metaplasia of gastric and esophageal mucosa (17).Fan et al. (18) reported that Hep expression was detected in47% of gastric carcinomas. The clinicopathologic features andprognosis of gastric carcinoma in relation to Hep expressionhave not been reported.

In the present study, we examined the expression of Cdx2and Hep in gastric carcinoma in order to analyze theircorrelation with histologic type, clinicopathologic features,and clinical outcomes of the patients in gastric carcinoma. Wealso investigated the efficacy of combined analysis of Cdx2 andHep expression to distinguish histologic types and predict theprognosis of gastric carcinoma.

Materials andMethods

Patients and tissue samples. A consecutive series of 109 patients withgastric carcinoma were studied. All patients were treated by D2 resectionin the Beijing Cancer Hospital between September 1995 and December1998. There were 75 males and 34 females with a mean age of 59 years(range, 29-82 years). None of the patients had received chemotherapyor radiation therapy preoperatively. The age and gender of patients,maximum tumor size, depth of wall invasion, status of lymph nodemetastasis, vascular invasion, and histologic grade were obtained fromhistopathology reports. Stage of gastric carcinoma was describedaccording to the 1997 tumor-node-metastasis (TNM) classification ofmalignant tumors by the International Union Against Carcinoma. Allpatients were given follow-ups until January 2004 with a minimum 5years of follow-up. In addition, 6 samples of normal stomach tissue, 17samples of normal small intestine tissue (duodenum and ileum), and 9samples of colon tissue were selected from the surgical pathology files atthe Beijing Cancer Hospital between 1996 and 1998. Nine cases ofintestinal metaplasia were obtained from gastroscopic biopsies files atBeijing Cancer Hospital in 2003.

Immunohistochemistry. The tissue was fixed in 10% neutralformalin and embedded in paraffin. One paraffin-embedded block oftumorous tissue was selected from each case and was cut into 4 Amsections. The sections were put in the oven at 60jC for 4 hours,deparaffinized in xylene, rehydrated in a graded ethanol series, andtreated with 3% hydrogen peroxide solution for 10 minutes. Antigenretrieval was done by microwaving tissue in 10 mmol/L citric acidbuffer for 10 minutes, then cooling at room temperature for 2 hours.The sections were incubated with an anti-Cdx2 monoclonal antibody(1:100, BioGenex, San Ramon, CA) and an anti-Hep monoclonalantibody (1:100, Zymed, San Francisco, CA) overnight at 4jC. Theprimary antibodies were detected using the Powervision two-step

Fig. 1. Immunohistochemical staining for Cdx2 in normal gastrointestinal mucosaand intestinal metaplasia (�200). A , Cdx2 staining is negative-in normal gastricmucosa; B, Cdx2 staining is observed in the nuclei of ileum epithelial cells; C, Cdx2staining is observed in the nuclei of colon epithelial cells;D, Cdx2 staining isobserved in intestinal metaplasia of the stomach.

Cdx2 and HepatocyteAntigen and Prognosis of Gastric Carcinoma

www.aacrjournals.org Clin Cancer Res 2005;11(17) September1, 20056163



histostaining reagent, with PV-6001, (Dako, Glostrop, Denmark) as thesecondary antibody and were visualized with diaminobenzedine. Tissuewas counterstained with hematoxylin, 1% hydrochloric acid of alcoholand eosin, dehydrated with graded ethanols, and mounted. Positiveand negative immunohistochemistry controls were routinely used.

Assessment of staining. Two experienced pathologists independent-ly examined nuclear Cdx2 and cytoplasmic Hep staining while blindto the clinicopathologic data and clinical outcomes of the patients. Atleast 10 high-power fields at 400� magnification were chosenrandomly and >1,000 carcinoma cells were counted for each section.Cases with >5% positive tumor cells in a section were regarded aspositive expression.

Statistical analyses. All data were analyzed using SPSS 10.0software. The association of Cdx2 and Hep expression with variousclinicopathologic features was analyzed using the m2 test. Cumulativesurvival was estimated by the Kaplan-Meier method and differences

Fig. 2. Immunohistochemical staining for Hep in normal gastrointestinal mucosaand intestinal metaplasia. A , Hep staining is negative-in normal gastric mucosa(�40); B, Hep staining is observed in the cytoplasm of ileum epithelial cells (�40);C, Hep staining is negative-in colonmucosa (�40);D, Hep staining is observed inintestinal metaplasia of the stomach (�100).

Fig. 3. Immunohistochemical staining for Cdx2 and Hep in gastric carcinoma(�200). A , Cdx2 staining is observed in the nuclei of carcinoma cells; B, granularpositive staining of Cdx2 is observed in the cytoplasm of carcinoma cells; C, Hepstaining is observed in the cytoplasm of carcinoma cells but is absent from adjacentnormal tissue.





between survival curves were analyzed by the log-rank test. Theinfluence of each variable on survival was analyzed by the multivariateanalysis of Cox proportional hazard model (backward, stepwise).Differences at P < 0.05 were considered to be statistically significant.

Results

Cdx2 and Hep expression in normal gastric, intestinal tissue,and colon tissue, and intestinal metaplasia. Cdx2 expressionwas detected in the nuclei of most of the normal smallintestinal mucosa (14 of 17, 82.4%), normal colon mucosa (8of 9, 88.9%), and intestinal metaplasia (7 of 9, 77.8%), but notin any of the six cases of normal gastric mucosa (0 of 6, 0%;Fig. 1). Hep expression was detected in the cytoplasm of one

case of normal gastric mucosa (1 of 6, 16.7%), most cases ofnormal small intestinal mucosa (15 of 17, 88.2%), one case ofnormal colon mucosa (1 of 9, 11.1%), and in all cases ofintestinal metaplasia (9 of 9, 100%; Fig. 2).

Cdx2 and Hep expression in gastric carcinoma. Cdx2 wasexpressed in the nuclei of gastric carcinoma cells in 36.7% (40 of109) of cases, whereas it was expressed granularly in the cyto-plasm of gastric carcinoma cells in 33.9% (37 of 109) of cases.Cdx2 was expressed in both nucleus and cytoplasm in six cases.Hep was expressed in the cytoplasm of gastric carcinoma cells in54.1% (59 of 109) of cases and exhibited focal positivity (Fig. 3).

Cdx2 and Hep expression in relation to clinicopathologicfeatures. The expression of Cdx2 in the nucleus was signifi-cantly higher in males than in females (P = 0.019), in stages I

Table1. Correlation between Cdx2 nuclear or cytoplasmic expression and clinicopathologic features

Factor Cases Cdx2 nuclear expression P Cdx2 cytoplasmic expression P

Cdx2� (n = 69) Cdx2+ (n = 40) Cdx2� (n = 72) Cdx2+ (n = 37)

GenderMale 75 42 33 0.019 51 24 >0.05Female 34 27 7 21 13

Age (y)MeanF SE 109 58.8F 1.6 57.3F 1.7 >0.05* 57.9F 1.4 58.9F 2.1 >0.05*Tumor size (cm)[5.0 59 39 20 >0.05 38 21 >0.05<5.0 50 30 20 34 16

Depth of wall invasionT1 10 6 4 >0.05 7 3 >0.05T2 23 11 12 15 8T3 59 39 20 40 19T4 17 13 4 10 7

LymphnodemetastasisN0 30 15 15 0.029 22 8 >0.05N1 46 28 18 33 13N2 22 15 7 12 10N3 11 11 0 5 6

Distant metastasisM0 94 58 36 >0.05 65 29 >0.05M1 15 11 4 7 8

Vascular invasionV (�) 53 31 22 >0.05 38 15 >0.05V (+) 56 38 18 34 22

DifferentiationWell 7 5 2 >0.05 3 4 >0.05Moderated 35 19 16 24 11Poorly 67 45 22 45 22

TNMstagesI + II 39 19 20 0.018 27 12 >0.05III + IV 70 50 20 45 25

Lauren classificationc

Diffuse 52 40 12 33 19Intestinal 30 16 14 0.027b 19 11 >0.05bMixed 21 10 11 0.015b 15 6 >0.05b

NOTE: m2 test.*t test.cSix cases of mucinous carcinomawere not included in Lauren classification.bCorrelationwith diffuse-type carcinoma.





and II than that in stages III and IV (P = 0.018), and in intestinal-type carcinoma (46.7%) than diffuse-type gastric carcinoma(23.1%, P = 0.027). There was a negative correlation betweennuclear Cdx2 expression and lymph node metastasis (P = 0.029).The expression of nuclear Cdx2 was not found to correlate withage, maximum tumor diameter, depth of wall invasion, distantmetastasis, vascular invasion, or histologic grade. No significantassociation was observed between cytoplasmic Cdx2 expressionand any of these clinicopathologic features (Table 1).

There was a negative correlation between Hep expressionand depth of wall invasion (P = 0.011). There wassignificantly less Hep expression in diffuse-type carcinoma

(46.2%) than in intestinal-type carcinoma (70.0%, P = 0.037)but expression did not correlate with other clinicopathologicfeatures (Table 2).

Prognostic implications of Cdx2 or Hep expression in gastriccarcinoma. Cdx2-nucleus-positive patients had significantlyhigher cumulative postoperative 5-year overall survivalrates (57.5%) than Cdx2-nucleus-negative patients (23.2%,P = 0.0009; Fig. 4A). However, Kaplan-Meier survival curvesshowed no statistically significant correlation between cytoplas-mic positive (21.6%) and negative cases (43.1%, P = 0.1601;Fig. 4B). Cases were divided using the Lauren classification andgrade of differentiation, and the survival rates of groups withdifferent Cdx2 nuclear expression status were analyzed. Inintestinal-type carcinoma, the 5-year survival rate of Cdx2-positive patients tended to be higher than that of Cdx2-negativepatients (60.0% versus 33.3%). However, statistical significancewas not reached (P = 0.083). In diffuse or mixed-typecarcinoma, the Cdx2-positive group (58.3%, 54.6%, respec-tively) showed a significantly higher 5-year survival rate than theCdx2-negative group (22.5%, P = 0.0279; 10.0%, P = 0.0162,respectively). In moderately or poorly differentiated carcinoma,the Cdx2-positive groups (56.3% and 59.1%, respectively) hadsignificantly higher 5-year survival rate than that of negativegroup (15.8%, P = 0.0141; 24.4%, P = 0.0039, respectively).There were too few cases of well-differentiated carcinoma to beanalyzed by statistical analysis (Table 3).

The Hep-positive group had significantly higher 5-yearsurvival rates (44.1%) than the Hep-negative group (26.0%,P = 0.003; Fig. 4C). Using a Cox regression analysis for the 109patients, positive Cdx2 nuclear expression (P = 0.001), Hepexpression (P = 0.029), TNM stage (P < 0.001), and vascularinvasion (P = 0.015) seemed to be independent prognosticindicators (Table 4).

Combined analysis of Cdx2 and Hep expression in distinguish-ing histologictype. No significant association was observedbetween Cdx2 and Hep expression by m2 test. When the twomarkers were combined to evaluate (Cdx2 and Hep expres-sion), positivity was defined as Cdx2 and/or Hep expression.Table 5 shows the correlation between the expression profilesof Cdx2 and Hep expression and Lauren classification. Cdx2and Hep expression positivity in diffuse gastric carcinoma

Table 2. Correlation between Hep expression andclinicopathologic features

Factor Cases Hep expression P

Hep� (n = 50) Hep+ (n = 59)

GenderMale 75 36 39 >0.05Female 34 14 20

Age (y)MeanF SE 109 57.7F1.9 58.6F 1.5 >0.05*Tumor size (cm)[5.0 59 32 27 >0.05<5.0 50 18 32

Depth of wall invasionT1 10 3 7 0.011T2 23 5 18T3 59 30 29T4 17 12 5

LymphnodemetastasisN0 30 10 20 >0.05N1 46 24 22N2 22 12 10N3 11 4 7

Distant metastasisM0 94 40 54 >0.05M1 15 10 5

Vascular invasionV (�) 53 24 29 >0.05V (+) 56 26 30

DifferentiationWell 7 1 6 >0.05Moderated 35 20 15Poorly 67 29 38

TNMstagesI + II 39 13 26 >0.05III + IV 70 37 33

Lauren classificationc

Diffuse 52 28 24Intestinal 30 9 21 0.037b

Mixed 21 11 10 >0.05b

NOTE: m2 test.*t test.cSix cases of mucinous carcinomawere not included in Lauren classification.bCorrelation with diffuse-type carcinoma.

Table 3. Postoperative survival and Cdx2 nuclearexpression in eachhistologic type and grade

Cases 5-yearsurvival rate (%)

P

Cdx2� Cdx2+

Lauren system*Intestinal 30 33.3 60.0 0.0853Diffuse 52 22.5 58.3 0.0279Mixed 21 10.0 54.6 0.0162

Histologic gradeWell differentiated 7 K K KModerately differentiated 35 15.8 56.3 0.0141Poorly differentiated 67 24.4 59.1 0.0039

*Six cases of mucinous carcinomawere not included in Lauren classification.





(50%) was significantly less than that in intestinal-type (93.3%,P < 0.001) or mixed-type carcinoma (76.2%, P = 0.04).

Survival analysis of combined evaluation of Cdx2 and Hep.On the basis of the expression profiles of Cdx2 and Hep, the 109patients were categorized as (a) Cdx2+/Hep+, (b) Cdx2+/Hep�,(c) Cdx2�/Hep+, and (d) Cdx2�/Hep�. We analyzed thepostoperative survival (log-rank test) and found significantdifferences between groups a and b (P = 0.0165) and betweengroups c and d (P = 0.0436). However, there was no significantdifferences between groups b and c (P = 0.7411), and thus theywere combined: (a) Cdx2+/Hep+ (n = 21), (b) Cdx2+/Hep� orCdx2�/Hep+ (n = 57), (c) Cdx2�/Hep� (n = 31). There weresignificant differences in survival rates among the three groups(P < 0.0001; Fig. 5). The survival rate of group a was significantlyhigher than that of group b (P = 0.0039), and the survival rate ofgroup b was significantly higher than that of group c (P = 0.0106)and the survival rate of group a was significantly higher than thatof group c (P < 0.0001). Using Cox regression analysis of the 109patients, Cdx2 and Hep expression (P < 0.001), TNM stage (P <0.001), and vascular invasion (P = 0.001) seemed to beindependent prognostic indicators (Table 6).

Discussion

In this study, we found that Cdx2 and Hep, especiallycombined analysis of these two markers are very sensitive andspecific for identifying intestinal-type carcinoma, as well asbeing significant for understanding carcinogenesis and biolog-

ical behavior, and predicting outcome of patients with gastriccarcinoma.

Extensive epidemiologic investigation revealed that intestinalmetaplasia increased the risk for development of gastriccarcinoma, especially intestinal-type carcinoma. Intestinalmetaplasia is an important precancerous lesion in the multi-steps of gastric carcinoma proposed by Correa (3). Severalstudies to elucidate the mechanism underlying Cdx2 expressionand its role have been done. The study by Silberg et al. (19)showed that gastric expression of Cdx2 alone was sufficient toinduce intestinal metaplasia in mice. A recent study in Cdx2-transgenic mouse by Mutoh et al. (4) provided direct evidencethat intestinal-type carcinoma could develop from intestinalmetaplasia. Therefore, intestinal metaplasia and intestinal-typecarcinoma may partly share similar expression of somemarkers, such as Cdx2 and Hep.

Cdx2 is one member of the caudal-related homeobox genefamily and is an intestine-specific transcription factor. Innormal adults, Cdx2 is expressed in mucosa from theduodenum to the distal colon but not in gastric oresophageal mucosa (5). As a transcription factor, Cdx2 hasa key role for regulating the proliferation and differentiationof intestinal cells and maintaining intestinal phenotypes(20–24). Thus, Cdx2 is thought to be a marker for intestinalepithelium. Others reported that expression of Cdx2 was alsoobserved in non–gastrointestinal cancers with intestinalphenotype, including pancreatic carcinomas and carcinomasof ampulla of Vater (25, 26). In our study, expression of

Table 4. Multivariate analysis of the prognostic factors in109 cases by Cox proportional hazardmodel

Variables B SE RR (95% CI) P

TNMstage <0.001II versus I 0.428 0.342 1.534 (0.785-2.999) 0.211III versus I 0.978 0.344 2.660 (1.355-5.224) 0.004IV versus I 1.894 0.389 6.646 (3.102-14.240) <0.001

Vascular invasion (� versus +) �0.642 0.265 0.526 (0.313-0.885) 0.015Cdx2 nuclear staining (� versus +) 0.753 0.229 2.124 (1.356-3.327) 0.001Hep (� versus +) 0.471 0.216 1.602 (1.050-2.446) 0.029

NOTE: B, coefficient; RR, relative risk; CI, confidence interval.

Table 5. Correlation between expression of Cdx2 and Hep combined (Cdx2 and Hep expression) and Laurenclassification

Cases Cdx2 and Hep expression Positivity (%) P

� (n = 33) + (n = 70)

Lauren classification*Diffuse 52 26 26 50Intestinal 30 2 28 93.3 <0.001cMixed 21 5 16 76.2 0.04c

NOTE: Cdx2 and Hep expression (+) defined as one of Cdx2 and Hep stained positively at least.*Six cases of mucinous carcinomawere not included in Lauren classification.cm2 test, comparedwith diffuse type carcinoma.





Cdx2 was detected in the majority of cases of intestinalmetaplasia and was higher in intestinal-type than diffuse-typegastric carcinomas. Marchetti et al. (27) reported that chronicacid exposure induced Cdx2 expression in murine esophagealepithelium cells. Ectopic expression of Cdx2 in the gastricmucosa could cause transdifferentiation of the epithelium toan intestinal type (19). Lorentz et al. (24) has shown thatCdx2 plays a key role in the extracellular matrix-mediatedintestinal cells differentiation. Liver intestine–cadherin is anintestine-specific cell adhesion molecule and expression inintestinal metaplasia and intestinal-type gastric carcinoma.Hinoi et al.’s study showed that liver intestine–cadherinexpression was strongly induced by Cdx2 and was suggestedto mediate Cdx2 function in intestinal cell morphogenesisand differentiation (28).

Hep is initially used as a specific marker for diagnosis ofhepatocellular carcinoma. Later studies found that Hep wasalso expressed in other types of cancer. Interestingly, Hep wasexpressed specifically in intestinal metaplasia (17). Our data areconsistent with this result. A study by Lee et al. (29) showedthat gastric carcinoma with Hep expression was significantlyassociated with early stage and well differentiation. The Hepantigen was thought to be a component of the membrane ofhepatocellular mitochondria (13). Hep-positive expression ingastric and intestinal mucosa can be explained by commonembryologic derivation of liver, stomach, and small intestine,which raises the possibility of expression of common antigensin cells from these organs. However, the mechanism of Hepexpression in intestinal metaplasia and gastric carcinomaremains to be investigated.

Taken together, this and other studies have providedevidence that Cdx2 and Hep were special and sensitive markersfor intestinal metaplasia and intestinal-type gastric carcinoma.Moreover, our results showed that combined analysis of Cdx2and Hep could offer a biological variable to identify intestinal-type carcinoma.

In the present study, the Cdx2-positive patients show betteroutcome than the Cdx2-negative patients. Multivariate analysisrevealed that Cdx2 represents an independent prognostic

Fig. 4. Kaplan-Meier survival curves for overall survival of109 patients with D2resected gastric adenocarcinoma. A , categorized by nuclear Cdx2 expression,survival was significantly better for patients with positive Cdx2 nuclear expressionthan those with negative expression (P = 0.0008); B, Categorized by cytoplasmicCdx2 expression, no significant difference canbeobservedbetween the two groupsof patients (P = 0.1064); C, categorized by Hep expression, survival wassignificantly better for patients with positive Hep expression than those withnegative expression (P = 0.003).

Fig. 5. Kaplan-Meier survival curves for three phenotypes of Cdx2 and Hepexpression in gastric carcinoma. Patients with Cdx2�/Hep� expression profile hadthe worst outcome for overall survival among the three phenotypes (P < 0.0001).





indicator. The results are consistent with reports from Seno andMizoshita (11, 12). Our data show a negative correlationbetween Cdx2 expression and tumor stage or lymph nodemetastatic status, which suggests that Cdx2 may play asuppressive role in the progression of gastric carcinoma.Previous studies suggested that Cdx2 is a tumor suppressor ofcolon carcinoma. Cdx2 can promote differentiation, inhibitproliferation, and increase sensitivity to apoptosis of intestinalepithelium cells and colon cancer–derived cells (21, 30). Cdx2is involved in the signaling pathways of some tumorsuppressors. For example, Cdx2 can transactivate the transcrip-tion of the cyclin-dependent kinase inhibitor p21/WAF1/CIP1via binding to the promoter of p21 (31), which induces tumorcell growth suppression. Cdx2 also contributes to APC’s tumorsuppressive effects (32). Cdx2 may have a similar function ingastric carcinoma. Seno et al. reported that Cdx2-positive casesshowed a lower Ki-67 index (11), which suggested that Cdx2-positive carcinoma might have low rate of cell proliferation.Recently, Cdx2 was reported to have a close relation to theintestinal-like tumor as well as being a prognostic factor ofpancreatic invasive ductal carcinomas and carcinomas ofampulla of Vater (25, 26). Consistent with our viewpoint,Cdx2 is regarded as an intestinal phenotypic marker thatimplicates better biological behavior and outcome.

As a marker of intestinal epithelium, Cdx2 expression ingastric carcinoma suggests that not only intestinal-type but thata part of diffuse-type carcinomas also have biological character-istics of intestinal-type carcinoma. The prognosis of Cdx2-positive cases is better than that of Cdx2-negative cases indiffuse-type gastric carcinoma. This indicates that although theformer did not have morphologic characteristics of intestinalepithelium, it was similar in biological behavior of intestinal-type carcinoma. Cdx2 expression was significantly correlatedwith prognosis in spite of the same histologic type ordifferentiation grade. These results suggest that expression ofCdx2 may reflect the biological behavior and prognosis ofgastric carcinoma more accurately than histologic classification.

Our data show that Hep-positive expression has a negativecorrelation with the depth of wall invasion, suggesting that Hepmay inhibit the progression of cancer. In this study, Hepexpression was associated with good prognosis and was anindependent prognostic indicator of gastric carcinoma, whichhas not yet been previously reported. However, the mechanismof Hep expression in normal or neoplastic tissues is, thus far,still unclear.

Given that Cdx2 and Hep are both intestinal markers andindependent prognostic factors, we studied the significance ofCdx2 and Hep coexpression of the prognosis for gastriccarcinoma. The patients could be categorized into three groupsaccording to Cdx2 and Hep coexpression profiles: (a) Cdx2+/Hep+, (b) Cdx2+/Hep� or Cdx2�/Hep+, (c) Cdx2�/Hep�.Significant differences of prognosis were found among thesegroups. Patients with both positive expression of Cdx2 and Hephave the best outcome among the three groups. In contrast,patients with both negative expression of Cdx2 and Hep havethe worst outcome among the three groups (relative risk, 3.087;95% confidence interval, 1.946-7.444). Moreover, the survivalrate of group b is significantly different from groups a and c,and indicates that combined evaluation of Cdx2 and Heppredicts the prognosis more accurately than using a singlemarker of each. Multivariate analysis revealed that the Cdx2and Hep coexpression profile was an independent prognosticindicator of gastric carcinoma. These results suggest that Cdx2and Hep coexpression may be a new and powerful indicator forpredicting the prognosis of gastric carcinoma. This may behelpful for the individualized therapy of patients with gastriccarcinoma.

In conclusion, our findings indicate that the prognosis ofgastric carcinoma patients with Cdx2- or Hep-positive expres-sion is significantly better than that with Cdx2- or Hep-negativeexpression. We revealed for the first time that combinedanalysis of Cdx2 and Hep may provide additional evidence ofbiological behavior to histologic type and might be useful topredict the prognosis for gastric carcinoma.

Table 6. Prognostic value of Cdx2 and Hep expression inmultivariate analysis by Cox proportional hazardmodel

Variables B SE RR (95% CI) P

TNMstage <0.001II versus I 0.332 0.340 1.393 (0.716-2.711) 0.329III versus I 0.901 0.343 2.463 (1.256-4.827) 0.009IV versus I 1.959 0.385 7.091 (3.333-15.089) <0.001

Vascular invasion (� versus +) �0.866 0.263 0.421 (0.251-0.704) 0.001Cdx2 and Hep expression <0.001b versus a 0.968 0.296 2.632 (1.473-4.703) 0.001c versus a 1.337 0.342 3.807 (1.946-7.444) <0.001

NOTE: B, coefficient; RR, relative risk; CI, confidence interval.

References1. Parkin DM, Pisani P, FerlayJ. Global cancer statistics.CACancerJClin1999;49:33^64.2. Lauren P. The two histological main types of gastriccarcinoma: diffuse and so-called intestinal type carci-noma. Acta Pathol Microbiol Scand1965;64:31^49.3. Correa P. Human gastric carcinogenesis: a multi-step and multifactorial processKFirst American

Cancer Society Award Lecture on Cancer Epide-miology and Prevention. Cancer Res 1992;52:6735^40.4. Mutoh H, Sakurai S, Satoh K, et al. Development ofgastric carcinoma from intestinal metaplasia in Cdx2-transgenic mice. Cancer Res 2004;64:7740^7.5. Silberg DG, Furth EE,Taylor JK, SchuckT, ChiouT,

Traber PG. CDX1protein expression in normal, meta-plastic, and neoplastic human alimentary tract epithe-lium. Gastroenterology1997;113:478^86.6.Mallo GV, Rechreche H, Frigerio JM, et al. Molecu-lar cloning, sequencing and expression of the mRNAencoding human Cdx1and Cdx2 homeobox. Down-regulation of Cdx1 and Cdx2 mRNA expression





during colorectal carcinogenesis. Int J Cancer 1997;74:35^44.7. Bonhomme C, Duluc I, Martin E, et al. The Cdx2homeobox gene has a tumour suppressor function inthe distal colon in addition to a homeotic role duringgut development. Gut 2003;52:1465^71.8. Bai YQ, Yamamoto H, Akiyama Y, et al. Ectopicexpressionof homeodomainproteinCDX2 in intestinalmetaplasia and carcinomas of the stomach. CancerLett 2002;176:47^55.9.Mizoshita T, Inada KI,Tsukanoto T, et al. Expressionof Cdx1 and Cdx2 mRNAs and relevance of differ-entiation in human gastrointestinal mucosaKwithspecial emphasis on participation in intestinal meta-plasia of the human stomach. Gastric Cancer 2001;4:185^91.10. Eda A, Osawa H, Yanaka I, et al. Expression ofhomeobox gene CDX2 precedes that of CDX1duringthe progression of intestinal metaplasia. J Gastroen-terol 2002;37:94^100.11. Seno H, Oshima M, Taniguchi MA, et al. CDX2expression in the stomach with intestinal metaplasiaand intestinal-type cancer: prognostic implications.IntJOncol 2002;21:769^74.12.MizoshitaT, TsukamotoT, Nakanishi H, et al. Expres-sion of Cdx2 and the phenotype of advanced gastriccancers: relationship with prognosis. J Cancer ResClin Oncol 2003;129:727^34.13.Wennerberg AE, Nalesnik MA, ColemanWB. Hepa-tocyte paraffin 1: a monoclonal antibody that reactswithhepatocytes and canbeused for differential diag-nosis of hepatic tumors. AmJPathol1993;143:1050^4.14. Zimmerman RL, Burke MA,Young NA, SolomidesCC, Bibbo M. Diagnostic value of hepatocyte paraffin1antibody to discriminate hepatocellular carcinomafrom metastatic carcinoma in fine-needle aspirationbiopsies of the liver. Cancer 2001;93:288^91.15. Chu PG, Ishizawa S,Wu E, et al. Hepatocyte anti-

gen as a marker of hepatocellular carcinoma: animmunohistochemical comparison to CEA, CD10,and AFP. Am J Surg Pathol 2002;26:978^88.16. Ramos-Vara JA, Miller MA. Immunohistochemicalcharacterizationof canine intestinal epithelial andmes-enchymal tumours with a monoclonal antibody tohepatocyte paraffin 1 (Hep Par 1). HistochemJ 2002;34:397^401.17. Chu PG, Jiang Z, Weiss LM. Hepatocyte antigenas a marker of intestinal metaplasia. AmJSurg Pathol2003;27:952^9.18. Fan Z, van de Rijn M, Montgomery K, Rouse RV.Hep Par 1antibody stain for the differential diagnosisof hepatocellular carcinoma: 676 tumors tested usingtissue microarrays and conventional tissue sections.Mod Pathol 2002;16:137^44.19. Silberg DG, Sullivan J, Kang E, et al. Cdx2 ec-topic expression induces gastric intestinal metapla-sia in transgenic mice. Gastroenterology 2002;122:689^96.20. Troelsen JT, Mitchelmore C, Spodsberg N, et al.Regulationof lactase-phlorizinhydrolase gene expres-sion by the caudal-related homeodomain proteinCdx-2. BiochemJ1997;322:833^8.21. Suh E,Traber PG. An intestine-specific homeoboxgene regulates proliferation and differentiation. MolCell Biol1996;16:619^25.22.Drummond F, SowdenJ,Morrison K, et al.The cau-dal-type homeobox protein Cdx-2 binds to the colonpromoter of the carbonic anhydrase1gene. EurJ Bio-chem1996;236:670^81.23. LambertM, Colnot SSE, L’Horset F, et al. cis-Actingelements and transcription factors involved in the in-testinal specific expression of the rat calbindin-D9Kgene: binding of the intestine-specific transcriptionfactor Cdx-2 to theTATA box. Eur J Biochem 1996;236:778^88.24. Lorentz O, Duluc I, Arcangelis AD, Simon-

Assmann P, Kedinger M, Freund JN. Key role of theCdx2 homeobox gene in extracellular matrix-mediat-ed intestinal cell differentiation. J Cell Biol 1997;139:1553^65.25. Matsumoto K, Mizoshita T, Tsukamoto T, et al.Cdx2 expression in pancreatic tumors: relationshipwith prognosis of invasive ductal carcinomas. OncolRep 2004;12:1239^43.26. Hansel DE, Maitra A, Lin JW, et al. Expression ofthe caudal-type homeodomain transcription factorsCDX 1/2 and outcome in carcinomas of the ampullaof Vater. J Clin Oncol 2005;23:1811^8.27. Marchetti M, Caliot E, Pringault E. Chronic acidexposure leads to activation of the cdx2 intestinalhomeobox gene in a long-term culture of mouseesophageal keratinocytes. J Cell Sci 2003;116:1429^36.28. Hinoi T, Lucas PC, Kuick R, Hansh S, Cho KR,Fearon ER. CDX2 regulates liver intestine-cadherinexpression in normal and malignant colon epitheliumand intestinal metaplasia. Gastroenterology 2002;123:1565^77.29. Lee HS, Kim WH, Kang GH. Hepatocyte expres-sions in hepatocellular carcinomas, gastrointestinalneoplasms, and non-neoplastic gastrointestinalmucosa: its role as a diagnostic marker. J KoreanMed Sci 2003;18:842^8.30. Mallo GV, Soubeyran P, LissitzkyJC, et al. Expres-sion of the Cdx1and Cdx2 homeotic genes leads toreduced malignancy in colon cancer-derived cells.JBiol Chem1998;273:14030^6.31. Bai YQ, Miyake S, Iwai T, et al. CDX2, a homeo-box transcription factor, upregulates transcription ofthe p21/WAF1/CIP1 gene. Oncogene 2003;22:7942^9.32. da Costa LT, HeTC,YuJ, et al. CDX2 is mutated in acolorectal cancer with normal APC/h-catenin signal-ing. Oncogene1999;18:5010^4.





2005;11:6162-6170. Clin Cancer Res Zhaoqing Fan, Jiyou Li, Bin Dong, et al. Implications for PrognosisCarcinoma: Correlation with Histologic Type and Expression of Cdx2 and Hepatocyte Antigen in Gastric

Updated version

http://clincancerres.aacrjournals.org/content/11/17/6162

Access the most recent version of this article at:

Cited articles

http://clincancerres.aacrjournals.org/content/11/17/6162.full#ref-list-1

This article cites 31 articles, 8 of which you can access for free at:

Citing articles

http://clincancerres.aacrjournals.org/content/11/17/6162.full#related-urls

This article has been cited by 1 HighWire-hosted articles. Access the articles at:

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

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. (CCC)Click on "Request Permissions" which will take you to the Copyright Clearance Center's

.http://clincancerres.aacrjournals.org/content/11/17/6162To request permission to re-use all or part of this article, use this link



http://clincancerres.aacrjournals.org/content/11/17/6162.full#ref-list-1

http://clincancerres.aacrjournals.org/content/11/17/6162.full#related-urls

http://clincancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



Documents

ExpressionofCdx2andHepatocyteAntigeninGastricCarcinoma ... · histostaining reagent, with PV-6001, (Dako, Glostrop, Denmark) as the secondary antibody and were visualized with diaminobenzedine