Early carcinoma of the gastric cardia in Japan

Is it different from that in the West?

Authors


Abstract

BACKGROUND

The incidence of adenocarcinoma of the gastric cardia has increased recently in the West. However, in Japan, most patients with gastric carcinoma have disease that is situated in the body and the distal stomach. The objectives of this study were to compare the clinicopathologic findings of patients with early gastric carcinoma (EGC) arising at the cardia and those with carcinoma in more distal parts of the stomach, then comparing the findings with those from patients with carcinoma of the gastric cardia in the West.

METHODS

Three thousand one hundred forty-four patients with EGC who underwent surgical resection between 1962 and 1997 at the National Cancer Center Hospital in Tokyo were studied. Seventy patients with EGC at the cardia were compared with those who had lesions in the middle and lower parts of the stomach. The body mass index (BMI), smoking, and drinking were evaluated using all patients with cardia EGC and 344 patients in a matched cohort in the latter group.

RESULTS

Seventy patients had an EGC located just at the cardia, whereas 2796 patients had lesions in the lower two-thirds of the stomach. The former lesions were different from those in the distal two-thirds of the stomach: More often, they were of an elevated type (34% vs. 14%, respectively, they were histologically well differentiated in 89% (vs. 59%), and there were more submucosal tumors (53% vs. 41%). The BMI, smoking, and drinking in the two groups were not different. The incidence of Barrett esophagus and gastroesophageal reflux disease (GERD) in patients with EGC were 2.9% (2 of 70 patients) and 5.7% (4 of 70 patients), respectively.

CONCLUSIONS

There were many significant differences in clinicopathologic characteristics between patients with carcinoma of the cardia and patients with carcinoma of the distal stomach in Japan. The incidence of early cardia carcinoma was very low in Japan, and obesity, smoking, drinking, Barrett esophagus, or GERD were not related to its occurrence, in contrast to reports in the West. Cancer 2000;89:2555–9. © 2000 American Cancer Society.

The incidence of adenocarcinoma of the gastric cardia has recently increased in the West, and it now accounts for 45% of early gastric carcinoma (EGC).1–3 This trend is in contrast with a decrease in the incidence of distal gastric carcinoma.4, 5

Carcinoma of the gastric cardia is believed to have a different etiology and different clinicopathologic features than distal gastric carcinoma.4, 6, 7 Many epidemiologic studies have demonstrated a close correlation between infection with Helicobacter pylori and distal gastric carcinoma,7, 8 whereas the risk of carcinoma of the gastric cardia remains unknown. Recent papers have reported that obesity was correlated with an increased risk of carcinoma of the gastric cardia in limited case–control studies in Western populations.9 Furthermore, it has been reported that specialized columnar epithelium, a form of intestinal metaplasia associated with Barrett esophagus, may be the origin of adenocarcinoma of the gastric cardia.10–12 In this study, we compared the clinicopathologic findings in patients with early carcinoma of the gastric cardia and patients with distal gastric carcinoma, and we examined the association between obesity and cardia carcinoma in a Japanese population.

MATERIALS AND METHODS

A total of 3144 patients who underwent surgery for EGC from 1962 to 1997 at the National Cancer Center Hospital (NCCH) were studied. Gastric carcinoma was classified according to the Japanese Classification of Gastric Carcinoma,13 and the stomach was divided anatomically divided into three portions: the upper one-third, the middle one-third, and the lower one-third. Adenocarcinoma of the cardia was defined as a lesion, the center of which was within 20 mm from the gastroesophageal junction,14, 15 and distal gastric carcinoma was defined as lesions located in the middle one-third and the lower one-third of the stomach.

The body mass index (BMI) was calculated by division of the weight in kilograms by the square of the height in meters. The BMI of 70 patients with carcinoma of the gastric cardia was compared with that of 334 patients with distal gastric carcinoma matched for age and gender (matched cohort).

Details of smoking and drinking habits in patients with EGC of the cardia were obtained using a self-administered questionnaire and were compared with age- and gender-matched cohorts. Barrett esophagus and GERD were evaluated by the review of endoscopic photomicrographs: This was possible because 40 photomicrographs are taken routinely during upper gastrointestinal endoscopies in our institution. All of the available endoscopic findings were analyzed by two independent endoscopists, and patients were classified according to the Los Angeles classification of esophagitis for definite Barrett esophagus and GERD.16

Statistical Analysis

Clinicopathologic findings were compared by using t tests, chi-square tests, Aspin–Welch tests, and Fisher exact tests, and the changing trends were evaluated by using Cochran–Mantel–Haenszel statistics. The SAS program (SAS Institute, Inc., Cary, NC) was used for all analyses. A level of P < 0.05 was considered significant.

RESULTS

Table 1 shows a summary indicating that, among 3144 patients with EGC who underwent surgical resection at NCCH, only 70 patients(51 males and 19 females) had carcinoma of the gastric cardia, and 2796 patients (1892 males and 904 females) had distal gastric carcinoma. Patients with cardia carcinoma (mean age, 65.2 years) were significantly older than those with distal carcinoma (mean age, 57.9 years).

Table 1. Comparison of Clinicopathologic Findings between Patients with Cardia and Distal Carcinoma
CharacteristicCardia carcinomaDistal carcinomaP value
  • a

    Student t test.

  • b

    Chi-square test.

  • c

    A Spin–Welch test.

  • d

    Fisher exact test.

No. of patients702796
Mean age in yrs (range)65.2 (35–86)57.9 (20–91)0.0001a
Male/female ratio2.72.10.359b
Mean tumor size in mm (range)24.8 (4–80)30.7 (1–150)0.006c
Macroscopic type (%)0.004d
 Superficial elevated3414
 Flat31
 Superficial depressed5473
 Mixed912
Depth (%)0.026b
 Intramucosal invasion4759
 Submucosal invasion5341
Histology (%)0.001b
 Differentiated8959
 Undifferentiated1141

When macroscopic configurations were classified into four categories according to the Japanese classification of gastric carcinoma, there were significant differences between cardia carcinoma and distal carcinoma. The frequency of the elevated type was much greater in patients with carcinoma of the gastric cardia (34%) compared with patients with distal gastric carcinoma (14%).

Histologically, the differentiated type of adenocarcinoma accounted for 89% of cardia tumors, whereas only 59% of distal tumors were of the differentiated type. The incidence of submucosal invasive carcinoma was significantly greater in patients with carcinoma of the gastric cardia compared with patients with distal gastric carcinoma.

Trends of EGC regarding the location in the stomach are shown in Table 2. The absolute value and the rates of cardia carcinoma to EGC have been increasing significantly, although there were no significant differences in trends of location classified according to the Japanese classification.

Table 2. Changing Trends in the Number of Patients with Early Gastric Carcinoma
 No. of patients (%) with carcinoma of the cardiaabNo. of patients (%) with gastric carcinomaTotal (%)
UbML
  • U: upper one-third of the stomach; M: middle one-third of the stomach; L: lower one-third of the stomach.

  • a

    The absolute value and the rates have been significantly increasing by Cochran–Mantel–Haenszel statistics (P = 0.001).

  • b

    The numbers of patients with cardia carcinoma are included in those of the upper one-third of the stomach.

1962–19691 (0.4)34 (12.1)171 (60.6)77 (27.3)282 (100)
1970–19792 (0.4)51 (9.6)295 (55.3)187 (35.1)533 (100)
1980–198926 (2.5)106 (10.3)564 (54.5)364 (35.2)1034 (100)
1990–199741 (3.2)157 (12.1)690 (53.3)448 (34.6)1295 (100)
Total70 (2.2)348 (11.1)1720 (54.7)1076 (34.2)3144 (100)

When 70 patients with carcinoma of the gastric cardia were compared with 344 patients with distal gastric carcinoma who were matched for age and gender, no difference existed in weight, height, or BMI. Furthermore, no significant differences in smoking or drinking habits were found among the two groups (Table 3).

Table 3. Height, Weight, Body Mass Index, Smoking History, and Drinking History in Patients with Cardia Carcinoma and Age-Matched and Gender-Matched Patients with Distal Carcinoma
VariableCardia carcinoma (n = 70)Distal carcinoma matched cohort (n = 344)P value
  • BMI: body mass index.

  • a

    Student t test.

  • b

    Chi-square test.

Height in cm (range)158.6 (134.0–175.1)160.1 (127.6–180.0)0.2114a
Weight in kg (range)57.3 (35.5–81.7)58.5 (32.0–98.0)0.3836a
BMI (range)22.8 (15.82–31.72)22.7 (15.31–34.34)0.9776a
Smoking (%)0.212b
 Daily14 (20.0)102 (29.7)
 Past or sometimes17 (24.3)84 (24.4)
 No smoking39 (55.7)158 (45.9)
Drinking (%)0.368b
 Daily6 (8.5)37 (10.8)
 Occasional15 (21.5)97 (28.2)
 No drinking49 (70.0)210 (61.0)

Endoscopically, no patients had long-segment Barrett esophagus (≥2 cm), although two patients (2.9%) with carcinoma of the gastric cardia had short-segment Barrett esophagus. GERD was noted in four patients (5.7%) and was classified as Grade A according to the Los Angeles classification.

DISCUSSION

The incidence of adenocarcinoma of the gastric cardia has increased gradually in the West, and it is now more common than distal gastric carcinoma.1, 2 In Japan, however, carcinoma of the gastric cardia remains very rare, and it accounts for only 2% of total EGC cases. Although cardia carcinoma has been reported previously to be increasing in Japan15 as well as in our series, this increase may be derived from advances in endoscopic techniques and equipment. The current study was designed to determine whether carcinoma of the gastric cardia is similar clinicopathologically to distal gastric carcinoma in Japan, which has the highest incidence of gastric carcinoma in the world.

There are several definitions of the term “gastric cardia.”14, 17–19 We conform to the consensus report of the International Gastric Cancer Congress, because their criteria are used commonly and are accepted widely.14 Our results showed that carcinoma of the gastric cardia obviously was different from distal gastric carcinoma in several clinicopathologic features, such as size, depth, and histologic and macroscopic type. Carcinoma of the gastric cardia tended to be smaller than distal gastric carcinoma; however, this was partly because we defined cardia carcinoma as lesions, the centers of which were located within 20 mm of the esophagogastric junction. Despite the smaller size of gastric cardia tumors, the rate of submucosal invasion was greater than that in distal gastric tumors, suggesting that cardia gastric tumors may be more malignant than distal gastric tumors. The proportion of differentiated histologic type was significantly greater in cardia gastric tumors compared with distal gastric tumors, suggesting that different etiologies contribute to its tumorigenesis. Macroscopically, it is important that the superficial elevated type (IIa type) was relatively common in cardia tumors, although the superficial depressed type (IIc type) accounted for more than half of the cases. We should pay attention to elevated lesions for the diagnosis of early carcinoma of the gastric cardia during an endoscopic examination.

A close correlation has been demonstrated between H. pylori infection and the incidence of distal gastric carcinoma.3, 7, 8 However, it seems unlikely that H. pylori infection is a significant factor in the rising incidence of carcinoma of the gastric cardia.20–22 Our previous studies also showed that the rate of H. pylori infection was greater in patients with distal gastric carcinoma compared with patients with carcinoma of the gastric cardia, although the difference was not statistically significant.23 In the current study, adequate information about H. pylori infection was not available serologically or histologically due to the processing methods used for resected specimens and the retrospective nature of the study.

Recently, it was demonstrated that the increased incidence of carcinoma of the gastric cardia was related to high BMI.9 We compared the BMI in patients with cardia carcinoma with that in age- and gender-matched patients with distal gastric carcinoma and found that the BMI was within normal range in the two groups. There seemed to be no association between obesity and carcinoma of the gastric cardia in our series, as reported in patients with distal disease.24

Previous authors reported that there was an association between smoking or drinking habits and tumors of the cardia,25, 26 although the association remains controversial. In the current study, neither smoking nor drinking showed any definite correlation with adenocarcinoma of the gastric cardia.

In the West, it has been reported that 35–42% of patients with adenocarcinoma in the region of gastric cardia or the esophagogastric junction had Barrett esophagus, and 25–37% had GERD,2, 12, 27–29 suggesting a close correlation between carcinoma of the gastric cardia and Barrett esophagus/GERD. Conversely, in our series, the incidence rates of Barrett esophagus and GERD were 2.9% and 5.7%, respectively. No significant differences were found compared with healthy controls.30–32 EGC of the cardia in Japan was not associated with Barrett esophagus or GERD, and it is unclear why the rates of Barrett esophagus and GERD were so remarkably different between adenocarcinoma of the gastric cardia in Japan and the West. Although EGC accounts for only 3.4% of all EGC patients currently at the NCCH, it seems that the incidence of carcinoma of the gastric cardia is increasing gradually (Table 2) with the rising prevalence of GERD.32

The current results suggest the following: 1) EGC of the cardia accounted for only 2% of EGC patients in a Japanese population compared with 45% of EGC patients in the West. 2) There were significant differences in clinicopathologic features between carcinoma of the gastric cardia and distal gastric carcinoma at an early stage. 3) Unlike in the west, BMI, smoking, and alcohol were not associated with the occurrence of carcinoma of the gastric cardia. 4) There was a much lower incidence of both GERD and Barrett esophagus in the Japanese cardia carcinoma patients; furthermore, the Barrett esophagus segments were <2 cm in greatest dimension, and the GERD was all of a low grade endoscopically.

Adenocarcinoma of the gastric cardia remains a rare disease in Japan and simply may occur by chance. However, it is increasing in incidence, and the reasons for this are not obvious. To determine whether this disease is a different from that in the West, we would have to continue to collect epidemiologic and pathologic data as this increase occurs.

Ancillary