Clinicopathologic and protein expression differences between cardia carcinoma and noncardia carcinoma of the stomach

Authors

  • Min A Kim M.D.,

    1. Department of Pathology, Seoul National University College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Hye Seung Lee M.D, Ph.D.,

    1. Department of Pathology, Seoul National University College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Han-Kwang Yang M.D., Ph.D.,

    1. Department of Surgery, Seoul National University College of Medicine, Seoul, Korea
    2. Cancer Research Institute, Seoul National University College of Medicine, Seoul, Korea
    Search for more papers by this author
  • Woo Ho Kim M.D., Ph.D.

    Corresponding author
    1. Department of Pathology, Seoul National University College of Medicine, Seoul, Korea
    2. Cancer Research Institute, Seoul National University College of Medicine, Seoul, Korea
    • Department of Pathology, Seoul National University College of Medicine, 28 Yongon-dong, Chongno-gu, Seoul 110-799, Korea
    Search for more papers by this author
    • Fax: (011) 82 27655600


Abstract

BACKGROUND

Although the incidence of adenocarcinoma of the stomach has decreased over the past several decades, gastric cardia carcinoma has increased over the same period.

METHODS

The clinicopathologic characteristics and immunohistochemical staining results of 21 proteins were investigated in 165 patients with cardia carcinoma, including 74 patients with true cardia carcinoma and 91 patients with subcardia carcinoma, and the results were compared with the results from 564 patients with noncardia carcinoma.

RESULTS

In the clinicopathologic analysis, patients who had cardia carcinoma tended to have tumors with poorly differentiated histology according to the World Health Organization classification system (P = 0.012), diffuse type according to the Lauren classification system (P = 0.049), and advanced pathologic TNM stage (P < 0.001). On immunohistochemical staining, loss of the p16 (P = 0.038) and smad4 (P < 0.001) tumor suppressor genes was more frequent in cardia carcinoma than in noncardia carcinoma. Carcinoembryonic antigen and CD44 overexpression were more frequent in patients with cardia carcinoma (P < 0.05). Conversely, patients who had cardia carcinoma exhibited less frequent expression of MUC1 (P = 0.008) and MUC5AC (P = 0.006) compared with patients who had noncardia carcinoma. Epstein–Barr virus infection was more common in patients with cardia carcinoma (P < 0.001). In the survival analysis, the patients with cardia carcinoma had a poorer prognosis. In the multivariate analysis, tumor location in the cardia was confirmed as an independent, poor prognostic factor in patients with gastric carcinoma.

CONCLUSION

Cardia carcinoma and noncardia carcinoma differed in their clinicopathologic characteristics and in their alterations of gene expression, as evaluated by immunohistochemistry. The current results support the hypothesis that cardia carcinoma forms a specific category of gastric carcinoma that is distinct from noncardia carcinoma. Cancer 2005. © 2005 American Cancer Society.

The gastric cardia is described histologically as a narrow circular zone at the transition between the esophageal squamous mucosa and the oxyntic gastric mucosa. It cannot be identified at the macroscopic level. At the microscopic level, the cardia is characterized by a thin mucosa with clear glandular cells and without any acid-producing cells. It ranges in width from 1 mm to 5 mm, and the size increases with age.

The past 2 decades have seen a sharp, worldwide decline in both the incidence and mortality of gastric carcinoma. However, gastric carcinoma remains a major cause of death from malignant tumors worldwide.1–5 In particular, the incidence of adenocarcinoma arising around the gastroesophageal junction has shown steadily rising rates over the past decades in the Western world, whereas the incidence of gastric adenocarcinoma arising in the antral region has shown a tendency to decrease.6, 7 A 5–6-fold increase was reported between 1970 and 1990.8, 9

In a narrow sense, cardia carcinoma of the stomach applies to tumors located exclusively in the region of the cardia. However, it has not been established whether these tumors originate from the true cardia mucosa or from the neighboring upper fundic mucosa. Furthermore, in patients with carcinoma that involves both the stomach and the esophagus, it is very difficult to differentiate between gastric and esophageal carcinoma. Therefore, the accurate definition of cardia carcinoma has been controversial for a long time. In this background, in 1987, Siewert et al. proposed a classification for carcinoma of the cardia that now is accepted widely.10 This classification is based on the anatomic characteristics of adenocarcinoma located at the gastroesophageal junction and consists of three groups. If the epicenter of the tumor is located > 1 cm above the gastroesophageal junction, then it is classified as Type I carcinoma (adenocarcinoma of the distal esophagus). If the epicenter of the tumor is located within 1 cm proximal and 2 cm distal of the gastroesophageal junction, then it is classified as Type II carcinoma (true cardia carcinoma of the stomach). If the epicenter of the tumor is located between 2 cm and 5 cm distal of the gastroesophageal junction, then it is classified as Type III carcinoma (subcardia carcinoma of the stomach).

The importance of cardia carcinoma is not only its rising incidence but also the possibility that it is a possible specific subtype of gastric carcinoma. Recent studies suggest that cardia carcinoma has different characteristics from adenocarcinomas arising in the rest of the stomach. For example, MacDonald and MacDonald reported that patients with cardia carcinoma had less in common with a family history of gastric carcinoma and more in common with a past history of chronic heartburn or duodenal ulcer.11 Gray et al. reported the association of tobacco smoking and alcohol drinking with cardia carcinoma.12 Furthermore, a greater tendency to proceed to deep wall penetration, lymph node metastasis, and a poor prognosis were described as characteristics of cardia carcinoma.13–15 Therefore, cardia carcinoma is considered to form a clinically distinct subtype among gastric carcinomas. However, the detailed clinicopathologic and immunohistochemical differences between cardia carcinoma and noncardia carcinoma of the stomach still have not been well characterized. In this study, using the tissue array method, we investigated the clinicopathologic characteristics, including survival and the expression status of 21 known tumor-associated proteins and Epstein–Barr virus (EBV), in 165 patients with cardia carcinoma, including 74 patients with true cardia carcinoma and 91 patients with subcardia carcinoma, and we compared with them with the survival and expression status in 564 consecutive patients with noncardia carcinoma.

MATERIALS AND METHODS

Patients and Samples

The files from 1145 patients who underwent surgical resection for gastric carcinoma diagnosed at the Department of Pathology, Seoul National University Hospital over a period of 2 years (from January, 1995 to December, 1996) were examined to evaluate their tumor location. To divide tumors into true cardia, subcardia, and noncardia carcinomas according to the classification system of Siewert et al., first, the pathology reports and surgical records were reviewed.10 Then, tumors were classified into the possible groups of true cardia, subcardia, and noncardia carcinomas. Next, macroscopic photographs and glass slides were reviewed, including precise measuring to classify true cardia, subcardia, and noncardia carcinomas correctly. The histologic gastroesophageal junction was identified on the basis of the following histopathologic landmarks: the most distal site of esophageal glands in the submucosal layer and the most distal site of squamous epithelium, including residual squamous epithelium.16 Consequently, 165 patients (14.4%) with cardia carcinoma, including 74 patients (6.5%) with true cardia carcinoma and 91 patients (7.9%) with subcardia carcinoma, were identified from a total of 1145 consecutive patients with gastric carcinoma. To compare cardia carcinoma and noncardia carcinoma, we selected 564 consecutive patients with noncardia carcinoma who were treated over a period of 1 year (from January, 1995 to December, 1995). The age, gender, tumor location, lymphatic invasion status, and pathologic Tumor-Lymph Node-Metastases (pTNM) stage were evaluated by reviewing the medical charts and pathology records.17 Glass slides were reviewed to determine the histologic type (according to the World Health Organization [WHO] and Lauren classification systems). Ninety-three percent of the patients had undergone curative resection (R0 according to the International Union Against Cancer guidelines). The clinical outcome of the patients was followed from the date of surgery to either the date of death or December 1, 2000, resulting in a follow-up that ranged from 1 months to 72 months (mean, 47 months). Patients who were lost to follow-up and patients who died from any cause other than gastric carcinoma were regarded as censored data in the survival analysis.

Tissue Array Methods

Core tissue biopsies (2 mm in greatest dimension) were taken from individual paraffin embedded gastric tumors (donor blocks) and were arranged in new recipient paraffin blocks (tissue array block) using a trephine apparatus (Superbiochips Laboratories, Seoul, Korea). Each tissue array block contained up to 60 cores, with 14 array blocks containing a total of 729 biopsies. Because it has been shown previously that there is excellent agreement between the staining results obtained from different intratumoral areas of gastric carcinomas, a core was sampled from each tumor.18 An adequate sample was defined as tumor occupying > 10% of the core area. Each block contained an internal control, which consisted of nonneoplastic gastric mucosa from the body, antrum, and intestinal metaplasia. Sections with a thickness of 4 μm were cut from each tissue array block, deparaffinized, and dehydrated.

Immunohistochemistry

Immunostaining was performed on the 14 array slides that contained 564 noncardia carcinoma samples and 165 cardia carcinoma samples. Twenty-one antibodies among various commercially available antibodies were selected by following a test procedure that employed a human control slide for immunohistochemistry (Superbiochips Laboratories). Table 1 and Figure 1 show the antibodies used in this study.

Table 1. Antibodies Used for Immunohistochemical Study
AntibodyRetrieval methodDilutionSourceExpression pattern in nonneoplastic mucosaExpression pattern in carcinoma
  • FHIT: fragile histidine triad; KAI1: kangai 1; MGMT: O6-methylguanine DNA-methyltransferase; CEA: carcinoembryonic antigen; CK: cytokeratin.

  • a

    Positive cytoplasmic staining was seen in intestinal metaplasia and was negative in the gastric gland proper.

  • b

    MUC5AC was positive in superficial glands, and MUC6 was positive in deep glands of the gastric mucosa.

p16Autoclave1:50Pharmingen (San Diego, CA)NucleusLoss
Smad4Microwave1:50Santa Cruz (Santa Cruz, CA)NucleusLoss
FHITMicrowave1:250Zymed (South San Francisco, CA)NucleusLoss
KAI1Microwave1:200Santa CruzCytoplasmicLoss
MGMTMicrowave1:50Chemicon (Temecula, CA)NucleusLoss
p53Microwave1:100DAKO (Carpinteria, CA)NegativeNucleus
E-cadherinMicrowave1:200Transduction (Lexington, KY)MembranousLoss
RbMicrowave1:50PharmingenNucleusLoss
MUC1Microwave1:100Novocastra (Newcastle, UK)NegativeCytoplasmic
MUC2Microwave1:100NovocastraCytoplasmaCytoplasmic
MUC5ACMicrowave1:100NovocastraCytoplasmbCytoplasmic
MUC6Microwave1:100NovocastraCytoplasmbCytoplasmic
CK7Microwave1:50DAKOMembranousMembranous
CK19Microwave1:100DAKOMembranousMembranous
CK20Microwave1:50DAKOMembranousMembranous
CEAMicrowave1:50DAKONegativeCytoplasmic
C-erbB2Microwave1:75DAKONegativeMembranous
bcl-2Microwave1:100DAKO (Glostrup, Denmark)NegativeNucleus
CD44Microwave1:40NovocastraNegativeMembranous
β-cateninMicrowave1:200Transduction (Lexington, KY)MembranousNucleus
p63Microwave1:100Santa CruzNegativeNucleus
Figure 1.

Immunohistochemical analyses in cardia carcinomas. (A) A p16-negative tumor sample. (B) An smad4-negative tumor sample. (C) An MUC1-positive tumor sample. (D) An MUC5AC-positive tumor sample. (E) A carcinoembryonic antigen-positive tumor sample. (F) A CD44-positive tumor sample (original magnification × 200).

Formalin fixed, paraffin embedded, 4-μm-thick sections were dewaxed in xylene, rehydrated through graded alcohol, and placed in an endogenous peroxide block for 15 minutes. Sections were washed in water before antigen retrieval was done. Then, they were placed in a citrate buffer (10% citrate buffer stock in distilled water, pH 6.0) and, for staining (except p16), microwaved for 10 minutes. Nonreactive staining was blocked by 1% horse serum in Tris-buffered saline, pH 6.0, applied for 3 minutes. The primary antibody was applied, and the antibody binding was detected using an avidin-biotin-peroxidase complex (Universal Elite ABC Kit PK-6200; Vectastain; Vector Laboratories, Burlingame, CA) for 10 minutes and diaminobenzidine tetrahydrochloride solution (Kit HK153-5K; Biogenex, San Ramon, CA). For the statistical analysis of these large-scale data, the results of immunostaining were considered positive if ≥ 10% of the neoplastic cells were stained.

EBV-Encoded Small RNA in Situ Hybridization

An in situ hybridization detection kit (Novocastra, Newcastle, United Kingdom) was used to visualize the EBV-encoded small RNAs (EBERs) that were encoded by EBV. Four-micrometer-thick sections were dewaxed in xylene and rehydrated in serially graded ethanol (100% and 95%), digested with proteinase K (20 mg/mL) for 30 minutes at 37 °C, then hybridized with fluorescein-labeled oligonucleotide probes for 2 hours at 37 °C. After washing, the hybridization products were identified with alkaline phosphatase-conjugated rabbit antibodies to fluorescein isothiocyanate (affinity-isolated rabbit F[ab′]). The chromogenic development was processed in nitroblue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate, and the slides were counterstained with Mayer hematoxylin. A positive signal was recognized as an intense blue-black nuclear stain under a light microscope. The positive tissue controls consisted of gastric carcinoma samples that were proven to contain EBERs. The normal cells that were present in every tissue sample served as a negative control.

Statistical Analyses

The chi-square test or the Fisher exact test (two-tailed) was used to compare clinicopathologic characteristics, including the expression status of the tumor-associated proteins and EBV, between cardia carcinomas and noncardia carcinomas. The results were considered statistically significant at P values < 0.05. Survival curves were estimated using the Kaplan–Meier product-limit method, and the significance of the differences between the survival curves was determined using the log-rank test. Multivariate survival analysis was performed using the Cox proportional hazards model. The enter method was used to determine a final Cox model. All statistical analyses were conducted using the SPSS 11.0 statistical software program (SPSS, Chicago, IL).

RESULTS

Comparison of Clinicopathologic Features of Cardia and Noncardia Carcinomas

Table 2 displays the clinicopathologic differences between cardia and noncardia carcinomas. Cardia carcinoma was associated with the poorly differentiated type according to WHO classification (P = 0.012) and the diffuse type of the Lauren classification (P = 0.049). Cardia carcinomas tended to have a more advanced pTNM stage than noncardia carcinomas (P < 0.001). However, there were no differences in either age or gender between patients with cardia carcinoma and patients with noncardia carcinoma. When we compared the clinicopathologic features of patients with true cardia carcinoma and patients with subcardia carcinoma, no differences were found in distribution according to age, gender, WHO classification, Lauren classification, or pTNM stage.

Table 2. Comparison of Clinicopathologic Parameters between True Cardia Carcinoma and Subcardia Carcinoma and between Cardia Carcinoma and Noncardia Carcinoma of the Stomach
ParameterNo. of patients (%)P value
1. True cardia carcinoma (n = 74)2. Subcardia carcinoma (n = 91)Total (1 + 2) (n = 165)3. Noncardia carcinoma (n = 564)1 vs. 21 + 2 vs. 3
  1. SD: standard deviation; NS: not significant; WHO: World Health Organization; W/D: well differentiated tubular adenocarcinoma; M/D: moderately differentiated tubular adenocarcinoma; P/D: poorly differentiated tubular adenocarcinoma; SRC: signet ring cell carcinoma; pTNM stage: pathologic Tumor-Lymph Node-Metastasis stage.

Mean age ± SD (yrs)52.1 ± 13.255.0 ± 12.053.7 ± 13.254.9 ± 12.5NSNS
Gender    NSNS
 Male44 (59.5)56 (61.5)100 (60.6)381 (67.6)  
 Female30 (40.5)35 (38.5) 65 (39.4)183 (32.4)  
WHO classification    NS0.012
 W/D 1 (1.4) 4 (4.4)  5 (3.0) 55 (9.8)  
 M/D23 (31.1)22 (24.2) 45 (27.3)166 (29.4)  
 P/D38 (51.4)50 (54.9) 88 (53.5)224 (39.7)  
 Mucinous 4 (5.4) 2 (2.2)  6 (3.6) 35 (6.2)  
 SRC 8 (10.8)13 (14.3) 21 (12.7) 84 (14.9)  
Lauren classification    NS0.049
 Intestinal24 (32.4)25 (27.5) 49 (29.7)226 (40.1)  
 Diffuse45 (60.8)57 (62.6)102 (61.8)302 (53.5)  
 Mixed 5 (6.8) 9 (9.9) 14 (8.5) 36 (6.9)  
Lymphatic emboli    NSNS
 Present23 (31.1)29 (31.9) 52 (31.5)152 (27.0)  
 absent51 (68.9)62 (68.1)113 (68.5)412 (73.0)  
pTNM stage    NS< 0.001
 I18 (24.3)24 (26.4) 42 (25.5)263 (46.6)  
 II22 (29.7)23 (25.3) 45 (27.3)113 (20.0)  
 III19 (25.7)18 (19.8) 37 (22.4)116 (20.6)  
 IV15 (20.3)26 (24.8) 41 (24.8) 72 (12.8)  

Differences in the Protein Expression Profiles between Cardia and Noncardia Carcinomas

Table 3 shows that there was a distinct pattern of protein expression in the patients with cardia carcinoma, who showed more frequent loss of smad4 (P < 0.001). The loss of p16 also was frequent in cardia carcinomas with marginal significance (P = 0.038). The overexpression of p53 protein was less frequent in cardia carcinomas than in noncardia carcinomas, although the difference did not achieve statistical significance (P = 0.058). However, there were no significant correlations between patients with cardia and noncardia carcinomas in terms of the loss of other tumor suppressor proteins, such as rb, fragile histidine triad, E-cadherin, kangai 1, and O6-methylguanine DNA-methyltransferase (P > 0.05). In addition, no differences were found in the expression of cytokeratins between the cardia and noncardia carcinomas. With regard to the mucin phenotype, MUC1 and MUC5AC expression was less frequent in cardia carcinomas than in noncardia carcinomas (P = 0.008 and P = 0.006, respectively). However, no significant difference was found in the expression of MUC2 and MUC6 between cardia and noncardia carcinomas. More frequent overexpression of carcinoembryonic antigen (CEA) (P = 0.001) and CD44 (P = 0.003) was observed in cardia carcinomas compared with noncardia carcinomas. In addition, positive EBV status was associated strongly with cardia carcinoma (P < 0.001). When we compared the immunohistochemical results between true cardia carcinomas and subcardia carcinomas, there were no statistically significant differences in protein expression levels, with the exception of CK7 expression (Table 3).

Table 3. Differences in Proteins Expressions and Epstein–Barr Virus Status between Cardia and Noncardia Carcinomas of the Stomach
Protein expression1. True cardia carcinoma (%)2. Subcardia carcinoma (%)Total (1 + 2) (%)3. Noncardia carcinoma (%)P value
1 vs. 21 + 2 vs. 3
  1. NS: not significant; FHIT: fragile histidine triad; KAI1: kangai 1; MGMT: O6-methylguanine DNA-methyltransferase; CEA: carcinoembryonic antigen; CK: cytokeratin; EBV: Epstein–Barr virus.

Tumor suppressor proteins with frequent loss in cardia carcinomas
 p16 (loss)40.834.837.528.8NS0.038
 smad4 (loss)29.623.626.312.8NS< 0.001
Tumor suppressor proteins without significant correlation between cardia and noncardia carcinomas
 p53 (positivity)23.329.526.734.7NSNS
 rb (loss)4.34.54.43.4NSNS
 FHIT (loss)47.147.647.446.5NSNS
 E-cadherin (loss)45.848.247.141.0NSNS
 KAI1 (loss)23.313.818.116.2NSNS
 MGMT (loss)16.415.916.112.5NSNS
Cytokeratin expression and mucin phenotype
 CK7 (positivity)54.471.463.871.20.030NS
 CK19 (positivity)85.982.484.088.3NSNS
 CK20 (positivity)29.634.932.525.8NSNS
 MUC1 (positivity)9.718.214.424.3NS0.008
 MUC2 (positivity)18.325.022.029.0NSNS
 MUC5AC (positivity)30.637.534.446.7NS0.006
 MUC6 (positivity)11.014.813.018.7NSNS
Proteins with frequent overexpression in cardia carcinomas
 CEA (positivity)53.459.856.942.5NS0.001
 CD44 (positivity)22.532.227.817.1NS0.003
Proteins without correlation between cardia and non-cardia cancers
 c-erb B2 (positivity)15.117.416.412.2NSNS
 bcl-2 (positivity)12.39.010.59.2NSNS
 β-catenin (positivity)20.519.520.018.5NSNS
 p63 (positivity)2.71.11.91.9NSNS
EBV status in cardia and noncardia carcinomas
 EBV (positivity)16.911.213.85.3NS< 0.001

Survival Analysis in Patients with Cardia and Noncardia Carcinomas

During follow-up, 265 of 721 patients (36.8%) died (89 patients in cardia carcinoma group and 176 patients in the noncardia carcinoma group). The survival rate among the patients with cardia carcinoma, as determined by the log-rank test, was significantly lower compared with the survival rate among the patients with noncardia carcinoma (P = 0.001) (Fig. 2A). When the patients were divided by pTNM stage, the prognosis for patients who had cardia carcinoma also was worse compared with the prognosis for patients who had noncardia carcinoma of the stomach in Stage III (P = 0.017) (Fig. 2B). In the multivariate analysis, tumor location in the cardia also was a significant and independent prognostic indicator in gastric carcinomas, although pTNM stage remained the strongest predictive factor (Table 4). However, there were no significant differences in survival between patients with true cardia carcinoma and patients with subcardia carcinoma (data not shown).

Figure 2.

Survival curves using the Kaplan–Meier method. Survival curves showed that patients who had cardia carcinoma had an unfavorable prognosis compared with patients who had noncardia carcinoma. (A) Overall survival curve (P < 0.001). (B) Survival curve for patients with advanced pathologic TNM Stage III cardia and noncardia carcinoma (P = 0.017).

Table 4. Multivariate Analysis of Predictive Factors for Survival: Cox Proportional Hazards Model
Prognostic factorHazard ratio (95% CI)P valuea
  • 95%CI, confidence interval; pTNM stage: pathologic Tumor-Lymph Node-Metastasis stage; WHO: World Health Organization.

  • a

    P values were determined using the enter method.

Cardia vs. noncardia carcinomas0.653 (0.502–0.849)0.001
pTNM stage2.473 (2.156–2.836)< 0.001
Lymphatic emboli1.533 (1.175–1.998)0.002
Radical surgery1.207 (0.994–1.465)0.057
Lauren classification1.106 (0.866–1.413)0.419
WHO classification1.034 (0.897–1.193)0.643

DISCUSSION

Carcinogenesis of the stomach, like that of other organs, consists of a multistep process of accumulation of several genetic alterations, including the activation of oncogenes and the inactivation of tumor suppressor genes.19, 20 In addition, multiple environmental factors, including Helicobacter pylori infection and dietary factors, have been implicated in the initiation of gastric carcinogenesis.21, 22 Among the different kinds of gastric carcinoma, cardia carcinoma may form a specialized subtype that differs from carcinomas of other parts of the stomach according to recent studies of the clinicopathologic characteristics and alterations of carcinoma-associated genes.11–15, 23–26 The main objective of the current study was to demonstrate the characteristics of cardia carcinoma by investigating the detailed clinicopathologic features and alterations in protein expression associated with this type of gastric carcinoma.

In the current study, we found that cardia carcinoma was associated with a poorly differentiated histology and a more advanced pTNM stage. However, there were no significant differences in the age, gender, or lymphatic emboli of the patients with cardia and noncardia carcinomas. MacDonald and MacDonald reported that the median age of patients who were diagnosed with cardia carcinoma was similar to the median age of patients who were diagnosed with noncardia carcinoma.11 Some previous reports revealed that cardia carcinoma was associated with a worse prognosis compared with noncardia carcinoma. This finding was related to the more advance stage of cardia carcinomas at the time of presentation and involved such factors as high incidences of deep penetration and regional lymph node and hepatic metastases.14, 15 Those reports generally are consistent with the findings of the current study.

Second, we compared the various protein expression profiles between patients with cardia carcinoma and patients with noncardia carcinoma. The loss of p16 and smad4 protein expression and positive EBV status were more frequent in cardia carcinomas than in noncardia carcinomas. To date, many reports have revealed that EBV-positive gastric carcinomas have distinct features, including a unique histology, a predominantly proximal location, and a positive correlation with the loss of p16 and smad4 protein expression.27–29 Driessen et al. reported that the characteristics of cardia carcinoma, including clinical data and cytokeratin expression patterns, were closer to esophageal adenocarcinoma than to distal gastric adenocarcinoma.25, 26 Their reports support our findings that there are distinct profiles of protein and EBV expression in cardia carcinoma compared with noncardia carcinoma of the stomach.

We demonstrated that the survival rate among patients with cardia carcinoma was lower compared with the survival rate among patients with non-cardia carcinoma, irrespective of disease stage or tumor histology. It was reported previously that cardia carcinoma shows a greater tendency to proceed to deep wall penetration, is associated more often with lymph node metastasis, and implies a poorer prognosis compared with noncardia carcinoma of the stomach.13–15 In the current study, we used multivariate analysis to confirm that tumor location in the cardia is a significant and independent prognostic indicator in patients with gastric carcinoma.

We adopted the classification of Siewert et al., which is based on the anatomic characteristics of adenocarcinoma located at the gastroesophageal junction. This classification scheme contains three groups, including Type I disease (adenocarcinoma of the distal esophagus), Type II disease (true cardia carcinoma of the stomach), and Type III disease (subcardia carcinoma of the stomach). Ichikura et al. insisted that true cardia carcinoma should be categorized as a distinct entity and should be separated from subcardia carcinoma not only on the basis of its clinicopathologic features but also from a therapeutic point of view.30 In the current study, true cardia carcinoma showed similar clinicopathologic characteristics to those of subcardia carcinoma. The protein expression patterns obtained using immunohistochemistry were almost the same in the true cardia carcinomas and the subcardia carcinomas. In the survival analysis, no difference was found between true cardia carcinoma and subcardia carcinoma. In this clinicopathologic and immunohistochemical study, we believe that true cardia carcinoma and subcardia carcinoma are similar entities but that cardia carcinoma should be separated from carcinomas of the rest of the stomach.

In summary, the clinicopathologic findings revealed that cardia carcinoma was associated with a poorly differentiated histology and advanced pTNM stage. In the immunohistochemical staining experiments, cardia carcinoma showed different expression patterns for several proteins, such as p16, smad4, CEA, CD44, MUC1, and MUC5AC, compared with noncardia carcinoma. With regard to EBV status, as evaluated by EBER in situ hybridization, there was a correlation between cardia carcinoma and EBV infection. In the survival analysis, the patients with cardia carcinoma had a poorer prognosis; and, in the multivariate analysis, tumor location in the cardia was confirmed as a significant and independent, poor prognostic factor in patients with gastric carcinoma. In conclusion, in immunohisochemical evaluation, cardia carcinoma and noncardia carcinoma differed both in clinicopathologic characteristics and in gene expression alterations.

Ancillary