Gastric cancer incidence has markedly decreased in some countries, such as United State 1 but it remains high in others such as Japan 2 and Iran 34. It is the first leading cause of cancer-related deaths in men, and the second one among women in Iran 3. According to the identification of many predisposing factors in recent years, incidence of gastric cancer has declined in most countries, especially in developed countries 25. Meanwhile, many studies show that adenocarcinomas arising from gastric cardia have increased, especially in areas with low incidence of gastric cancer 678910111213141516171819202122. By contrast, some of the studies have not shown any increase in cardia carcinoma incidence 2324. Two important population-based studies indicate that the cardia carcinoma incidence in different countries has been inconsistent 2526. There is also a belief that this rising incidence has been confined to the areas with low-risk for gastric carcinoma 2324, although there are some contradictory studies 192027. It should be remembered that in spite of a large number of studies carried out in low-risk countries 2, very rare ones have been performed in high-risk countries.

Although Iran is one of the countries with high incidence of gastric cancers, there is no information about possible gastric carcinoma location change in Iranian patients. Therefore, this study intends to evaluate the gastric carcinoma location trend during the last 36 years.

As a preliminary step, all cases of gastrectomy specimens due to carcinoma during 1969–2004 in Cancer Institute of Tehran Medical University, the oldest main referral center for cancerous patients in Iran, were collected. All new cases with diagnosis of adenocarcinoma were also included in the course of this study. All the related pathology reports and hematoxylin and eosin sections were re-evaluated and reviewed by two pathologists. In cases with inappropriate slides new sections were cut and stained. Cases with damaged paraffin blocks were excluded from the study.

Anatomic site of each tumor was first determined by referring to the relevant macroscopic description of pathology report and then correlated with the latest guidelines for gastric cancer classification by Japanese Research Society for Gastric Cancer, in which the stomach is anatomically delineated into the upper, middle and lower thirds by dividing the lesser and greater curvature in two equidistant points and joining these points 28. Tumors located predominantly in the gastro-esophageal junction and cardia were determined to be in the upper third of the stomach, those located in the pylorus were considered to be in the lower third and those located in the mid-body were determined to be in the middle third of the stomach. If tumor was located across adjacent regions, the region containing the greater proportion of the tumor was considered to be the tumor's main location. Cardia was defined as region between 1 cm proximal and 2 cm distal to gastro-esophageal junction 29. In regard of rather low number of cases per year, each five years categorized in one group and then the trend of location during this 35 years was evaluated.

Clinical data, including sex and age, were gathered from pathologic reports. Demographic and clinical data were analyzed via Pearson's X², ANOVA, and Somer's d. A value of P < 0.05 was considered to be statistically significant.

All 1310 cases of gastrectomy for adenocarcinoma during 1969 to 2004 in Tehran Cancer Institute were collected, from which 78.5%were male. The mean age was 56.6 ± 21.2. In gastric sublocation analysis, 89 cases were excluded because of unavailability of clinical data and 12 cases for multicentricity. Among the rest, 298 (22.7%) were in the upper third, 325 (24.8%) in the middle third and 586 (44.7%) in the lower third. According to lauren classification 54.9% were intestinal type and the remaining was diffuse type.

There was no sexual difference between locations (Pearson chi-square, X² = 1.47, P = 0.479). Analysis of variance showed a significant difference in age at diagnosis between adenocarcinomas appeared in three part of stomach (P < 0.001) and in two genders (P < 0.006), but no significant correlation between location of tumors and gender (P > 0.05).

There was significant difference between locations and the degree of invasion (Pearson chi-square, X² = 26.5, P = 0.000) which means serosal invasion was higher in adenocarcinoma of upper and middle third (Figure 1).

Prevalence of adenocarcinoma of upper and middle third has increased in recent 36 years. By contrast, that of lower third has decreased (Somer's d = -0.252; P <0.001)(Figure 2). The trend of location was seen in both sexes; and it was approximately equal (Somer's d: -0.250 in males and -0.263 in female; both P < 0.001)(figure 3). This increasing trend did not vary by age (under 50 versus upper 50) but was more pronounced among the younger (Somer's d: -0.239 in youngers and -0.227 in elders; both P < 0.001) (figure 4).

Adenocarcinoma of the upper and middle third of stomach shows an increasing trend during the last 30 years in Iran. Gastric cancers are the most common malignant neoplasm among Iranians 34. They are the first leading causes of cancer-related deaths in Iranian males and the second in Iranian females 3 and, despite progress in diagnosis and treatment they have still poor prognosis (five year survival less than 20%)1. This study is the first one that focusing on changing of location of gastric adenocarcinoma in Iran. This study had some limitations; the most important one is that our study was not a population-based study, because cancer registry system in Iran is not complete yet and designing a population-based study is impossible now. However we believe that our study has minimal selection bias because 1) The selected center for this study is the most important referral center in Iran, 2) The cases were collected for a period of 36 years from all over the country. The other limitation was the incomplete macroscopic description of reports, particularly among early cases, a fact which leads to uncertainty in defining location sub-classification of tumors.

The advantage of this study is that all the related pathology reports and hematoxylin and eosin sections were reviewed by two pathologists. Consequently the error of misclassification that is propounded for increasing incidence of cardia adenocarcinoma in some studies 631 was minimized.

Our study showed significant change in distribution of gastric adenocarcinoma as prevalence of carcinoma of upper and lower third reached from 15% in the first 10 years to 36% in the last 10 years of study period. One of the reasons that has usually been brought up for the increasing incidence of upper third gastric carcinoma is the decreasing incidence of lower third carcinoma due to of H. pylori eradication treatments leading to the relative increase of upper third carcinoma 323334353637. But this study showed not only decrease of prevalence of lower third adenocarcinomas but also independently increase of prevalence of upper and middle third adenocarcinoma (Spearman's rho: upper: 0.750 (p ~ 0.05); middle: 0.750 (p ~ 0.05); lower: -0.964 (P < 0.01))(Figure 5). Misclassification is another reason that has been brought up in some studies 731, but as will be shown later, this was limited in this study.

Disclosing of separate risk factors for cancers of upper third of stomach made this increasing incidence explanation available. Although many evidences show that treatment of H. pylori infection, especially cagA⁺ strains can decrease incidence of distal gastric adenocarcinomas, several recent studies have noted that this treating might increase the risk of peptic esophagitis and adenocarcinoma of esophagus and cardia 32373839. However some studies show that adenocarcinomas of distal esophagus are inversely associated with H. pylori but gastric cardia cancers have an unclear association with it 21404142434445. Other risk factors that former studies have propounded are amount of dietary antioxidant 33, fruit and vegetable consumption 46, cigarette smoking 4754 and obesity 535556. However in some reports gastric cardia cancers are only modestly associated with obesity 42 and are known to have no association with antioxidant intake 41. In addition to the above risk factors Epstein – Barr virus (EBV) has been recently found to be associated with cardia tumors and it increases the risk of gastric cardia tumors five times 57. Genetic factors are also important 5859 and it has been shown that familial gastric cancers are frequently located in the cardia and are usually more aggressive than sporadic gastric cancers 6061.

Urban development and industrialized living in recent decades in Iran have increased the probability of confrontation with much of the above risk factors, and this can be the cause of increasing prevalence of adenocarcinomas of upper third of stomach in this study. Decreasing prevalence of carcinoma of lower third of stomach can be due to early treatment of H. pylori infection in recent years and elimination of noxious diet. Cancers of middle third and lower third of stomach have the same risk factors; therefore we may normally expect a decreasing incidence of them. However our findings show an increasing prevalence of middle third gastric adenocarcinoma; the same was seen in another study in Japan 27. We have no explanation for this and it needs more confirmatory studies.

In this study the trend of location was seen in both sexes, and it was approximately equal in both sexes. But Devesa's study in united state 6 and Botterweek's study in Slovakia, England, Wales and Scotland show the trends was more pronounced in females25. In other studies it was seen exclusively in males 12162762. Less confrontation with risk factors such as smoking and other genetic and hormonal differences are propounded for explanation; however regarding of contradictory results this gender difference is largely unexplained and we could not find it 26.

With regard to small number of cases in our study an accurate evaluation of age effect in distribution of gastric cancers was impossible. However in a general analysis increasing prevalence of carcinomas of upper and middle third was a little more pronounced among the younger. This finding is against with other studies 627.

Generally increasing prevalence of upper third adenocarcinomas in this study is similar to increasing incidence seen in the United States 6216364, England 1011, Denmark 12, the Netherlands 13, Sweden 14, New Zealand 15, Norway 16, Slovakia 17, Austria 18 and France 65. It was not seen in all areas of the world. In two important population-based studies – Corley's 26 and Botterweek's 25 study – the results were not consistent in different areas and this can be due to different risk factors. There are also some contradictory results in the same areas. For example in Japan in one 30 year period study no change in gastric cancer location was seen 24, but in two other studies increasing incidences of gastric cardia cancer were seen 1920.

Several studies have suggested that increasing incidence of proximal gastric cancers is seen in areas with low-risk for gastric cancer 2324, but our findings in Iran which is one of the regions with high frequency did not show this change, similar to some other studies in Japan192027.

In conclusion, our findings show an increasing prevalence of proximal gastric cancers in our center during the last four decades. These findings express the changing in life style and so the risk factors; however we should consider other factors such as misclassification, and probability of selection bias in this study in regard of a few numbers of cases. Therefore we recommend a confirmatory population-based study. There are also hypothesis about histological trends in gastric adenocarcinoma, so in regard of lacking such a study in Iran our next step will be analysis of histological trend in gastric adenocarcinoma.

EBV: Epstein – Barr virus