Fax: (415) 833-3364
Marked regional variation in adenocarcinomas of the esophagus and the gastric cardia in the United States
Article first published online: 31 OCT 2002
Copyright © 2002 American Cancer Society
Volume 95, Issue 10, pages 2096–2102, 15 November 2002
How to Cite
Kubo, A. and Corley, D. A. (2002), Marked regional variation in adenocarcinomas of the esophagus and the gastric cardia in the United States. Cancer, 95: 2096–2102. doi: 10.1002/cncr.10940
- Issue published online: 31 OCT 2002
- Article first published online: 31 OCT 2002
- Manuscript Accepted: 12 JUN 2002
- Manuscript Revised: 6 JUN 2002
- Manuscript Received: 17 APR 2002
- United States National Institutes of Health Mentored Clinical Scientist Development Award. Grant Number: K08 DK02697
- esophageal neoplasms;
- gastrointestinal neoplasms;
Adenocarcinomas of the esophagus and the gastric cardia recently have experienced rapidly increasing incidence rates. Although these sites frequently are combined, they may have different risk factors.
The authors compared regional incidence rates of esophageal adenocarcinoma, gastric cardia adenocarcinoma, and esophageal squamous cell carcinoma within the U.S. Surveillance, Epidemiology, and End Results (SEER) cancer registry for the years 1973–1998.
Regional incidence rates varied considerably. The Seattle-Puget Sound registry's recent average esophageal adenocarcinoma rates were over twice as high as those of the Utah registry (5.3 vs. 2.4 per 100,000 persons per year; P < 0.01); gastric cardia rates also differed (4.0 vs. 2.8 per 100,000 persons per year; P < 0.01). The incidence rate increase also varied markedly between regions. Since 1974, white male esophageal adenocarcinoma rates increased by 800% in Seattle compared with an increase of only 300% in Utah. In contrast, white male cardia adenocarcinoma rates increased by only 16% in Seattle (from 3.1 per 100,000 persons per year in 1974 to 3.6 per 100,000 persons per year in 1998) compared with 300% in Utah (from 0.7 to 2.2 per 100,000 persons per year). Both types of adenocarcinoma were more common in males and in the white population in all regions, but recent esophageal adenocarcinoma rates for black males in Connecticut were significantly higher than the U.S. black male average (3.1 vs. 0.8 per 100,000 persons per year; P < 0.01) and equaled the rates for the white population in some areas. Esophageal adenocarcinoma rates continued rising for white males through 1998, whereas cardia adenocarcinoma rates stabilized after 1988.
There are substantial regional, temporal, and ethnic differences between esophageal adenocarcinoma incidence rates and gastric cardia adenocarcinoma incidence rates within a single cancer registry system. Thus, these malignancies may differ in important ways and should not be combined routinely in research studies. Individual-level studies are needed to explain these substantial regional and ethnic differences. Cancer 2002;95:2096–102. © 2002 American Cancer Society.
Recent reports indicate that incidence rates for adenocarcinomas of the esophagus and gastric cardia (a.k.a. the gastroesophageal junction) have increased up to 300% in the last few decades, more rapidly than any other malignancy in the United States.1–3 In contrast, rates for esophageal squamous cell carcinoma and distal gastric carcinomas have been stable or have decreased.2, 3 The increasing frequency and high mortality rate for these malignancies provide a compelling argument for detailed evaluation of regional differences in incidence rates.4 These differences may provide clues to modifying disease risk for these malignancies, for which the most effective approaches are prevention and early detection.
Previous studies frequently have considered esophageal and gastric cardia adenocarcinomas together rather than as separate entities, precluding an examination of potentially distinct epidemiologic patterns between these two adjacent anatomic sites.5, 6 Esophageal and gastric cardia adenocarcinomas may have different risk factors, ethnic distributions, and temporal trends. In addition, international incidence rate comparisons may be influenced by differences in data quality, site classification methods, and case ascertainment between the registries.1, 7 Examination of incidence rate patterns within a single, high-quality cancer registry system would delineate more reliably the geographic and ethnic differences between the anatomic sites; such an evaluation has not been performed.
We used the U.S. Surveillance, Epidemiology, and End Results (SEER) database to evaluate geographic, ethnic, and temporal variability between esophageal and gastric cardia adenocarcinomas. We also contrasted these rates with the current incidence rates and temporal trends of esophageal squamous cell carcinoma. These comparisons may provide additional insights into potential etiologic factors for these malignancies.
MATERIALS AND METHODS
We evaluated regional differences in cancer rates using the National Cancer Institute SEER cancer registry data. The SEER database collects cancer case data from 11 population-based tumor registries throughout the United States, covering approximately 14% of the U.S. population.8 Its population is comparable to the general U.S. population with regard to measures of poverty and education; however, its population is somewhat more urban and has a higher proportion of foreign-born persons.8
Anatomic site and histology definitions used the International Classification of Disease for Oncology, second edition (ICD-O-2).9 The site and histology codes were: esophageal carcinoma, codes C15.0–C15.9; gastric cardia (defined as the cardioesophageal junction, the esophagogastric junction, and the gastroesophageal junction), code C16.0; squamous cell carcinoma, codes 8050–8082; and adenocarcinoma, codes 8140–8573.9
All analyses were performed using SEER Stat software (version 4.;1 National Cancer Institute, Bethesda, MD) and Stata software (release 7; Stata Corporation, College Station, TX). Annual incidence rates per 100,000 persons were age adjusted to the 1970 U.S. standard population and were stratified by region, gender, and ethnicity (white and black). To maximize the stability of the incidence estimates, regional comparisons use 3-year averages (1996–1998). To provide maximal contrasts, we compared regional registries with the highest and lowest incidence rates using the two-tailed Z statistic (α = 0.05); registries with unstable data (i.e., extremely wide confidence intervals) were excluded from these comparisons. Temporal trends between the registries were analyzed with linear regression. The overall percent change in rate used 2-year averages of the first years (1973–1974) and last years (1997–1998) of data.8 The Los Angeles and San Jose-Monterey registries were excluded from the trend analyses, because they lacked data prior to 1992. Temporal trend graphs used the cubic spline median-smoothing technique.10 All incidence rates are expressed as new cancer diagnoses per 100,000 persons per year.
Esophageal adenocarcinoma incidence rates had significant variability in their geographic, gender, and ethnic distributions (Tables 1, 2). For example, in 1996–1998, white males in the Seattle-Puget Sound registry had twice the incidence of white males in the Utah registry (5.3 vs. 2.4 per 100,000 persons per year; P < 0.001). Nationally, during this period, white males had an 8-fold higher incidence compared with white females (4.0 vs. 0.5 per 100,000 persons per year; P < 0.01) and a 5-fold higher incidence compared with black males (4.0 vs. 0.8 per 100,000 persons per year; P < 0.01). However, rates for black males in Connecticut were significantly higher than the U.S. black male average (3.1 vs. 0.8 per 100,000 persons per year; P < 0.01) and equaled the rates for the white population in some areas.
|SEER registry||Tumor type|
|Esophagus (adenocarcinoma)||Gastric cardia||Esophagus (squamous)|
|11 SEER registries||4.0||3.8–4.2||3.3||3.1–3.4||1.8||1.7–2.0|
|11 SEER registries||0.8||0.6–1.1||1.9||1.5–2.3||8.8||7.9–9.7|
|SEER registry||Tumor type|
|Esophagus (adenocarcinoma)||Gastric cardia||Esophagus (squamous)|
|11 SEER registries||0.5||0.5–0.6||0.7||0.6–0.7||0.9||0.8–1.0|
|11 SEER registries||0.2||0.1–0.4||0.6||0.5–0.9||3.4||2.9–3.9|
Females had moderate geographic variation despite relatively low incidence rates in all registries (Table 2). For 1996–1998, white females in the Seattle registry had twice the incidence of white females in the Atlanta registry (0.7 vs. 0.3 per 100,000 persons per year; P < 0.01). No regional differences were detected in black females; however, low case counts created unstable estimates.
There were substantial regional differences in the rates of incidence change over time (Fig. 1). Average white male incidence rates in Seattle increased over 800% between 1974–1975 and 1997–1998 (from 0.6 per 100,000 persons per year to 5.5 per 100,000 persons per year), with an average annual increase of 0.2 (95% confidence interval [95%CI], 0.18–0.22). In contrast, the incidence rates in Utah increased by only 300% during this time (from 0.7 per 100,000 persons per year to 2.8 per 100,000 persons per year; average increase, 0.08; 95%CI, 0.06–0.10), despite the baseline comparable incidence rates for these two regions in 1973–1974. In addition, the rate of increase in Utah did not begin to increase substantially until approximately 1985. Black males had a small but significant incidence increase during this time (0.03 per 100,000 persons per year; 95%CI, 0.02–0.04).
Gastric Cardia Adenocarcinoma
Gastric cardia adenocarcinoma incidence rates also had significant variability in their geographic, gender, and ethnic distributions, and these distributions sometimes differed substantially from the patterns seen for esophageal adenocarcinoma. Similar to esophageal adenocarcinoma, the incidence rates of cardia adenocarcinoma in 1996–1998 were higher in males compared with females and in the white population compared with the black population (white males vs. black males: 3.3 vs. 1.9 per 100,000 persons per year; P < 0.01). Compared with esophageal adenocarcinoma, however, no significant differences were observed between white females and black females. The average national rates were highest in white males (3.3 per 100,000 persons per year; 95%CI, 3.1–3.4) and lowest in black females (0.6 per 100,000 persons per year; 95%CI, 0.5–0.9).
Although there was significant geographic variation in the incidence of gastric cardia adenocarcinoma, it was substantially less compared with the incidence of esophageal adenocarcinoma. White males in the Seattle-Puget Sound registry had significantly higher rates compared with white males in the Utah registry (4.0 vs. 2.8 per 100,000 persons per year; P < 0.01) or the New Mexico registry (4.0 vs. 2.2 per 100,000 persons per year; P < 0.01). White females in the Seattle and Los Angeles registries had almost three-fold higher rates compared with white females in Utah (0.8 vs. 0.3 per 100,000 persons per year; P < 0.01). In black males and females, no significant geographic patterns were observed.
Nationally, the average male cardia adenocarcinoma rates increased significantly by approximately 90% between 1973 and 1998 (from 1.7 to 3.3 per 100,000 persons per year in white males; from 0.9 to 1.6 per 100,000 persons per year in black males). However, the cardia adenocarcinoma rates of change varied substantially by region and compared with the rates of change for esophageal adenocarcinoma (Fig. 2). White male incidence rates increased by < 20% in Seattle (from 3.1 per 100,000 persons per year in 1974 to 3.6 per 100,000 persons per year in 1998) compared with an increase of > 300% in Utah (from 0.7 to 2.2 per 100,000 persons per year). Although the rate of increase was significantly higher in Utah compared with the rate of increase in Seattle, the absolute incidence rate still was higher in Seattle compared with the rate in Utah for any given year. Compared with esophageal adenocarcinoma rates, average gastric cardia adenocarcinoma rates stabilized after the late 1980s.
Esophageal Squamous Cell Carcinoma
Esophageal squamous cell carcinoma incidence rates also had significant variability in their geographic, gender, and ethnic distributions. The black population had a 4-fold higher average incidence compared with the white population (black males vs. white males: 8.8 vs. 1.8 per 100,000 persons per year; P < 0.001; black females vs. white females: 3.4 vs. 0.9 per 100,000 persons per year; P < 0.001). Black females, who were the group with the lowest incidence for esophageal and gastric cardia adenocarcinomas, had higher rates of esophageal squamous cell carcinomas compared with white males (3.4 vs. 1.8 per 100,000 persons per year; P < 0.001). The Connecticut registry had the highest black male incidence rates, whereas the San Francisco-Oakland registry had the lowest rates (16.4 vs. 6.2 per 100,000 persons per year; P < 0.001). The Atlanta and San Francisco-Oakland registries were high incidence areas for black females and white females, respectively.
Compared with esophageal and gastric cardia adenocarcinomas, all regions in the United States reported declining incidence rates of squamous cell carcinoma (Fig. 3). The average national incidence rate for the black population decreased by 40% for males (from 13.9 to 8.3 per 100,000 persons per year) and by 25% for females (from 4.3 to 3.2 per 100,000 persons per year) since 1973. The incidence for white males decreased by 42% (from 3.3 to 1.9 per 100,000 persons per year) during the same period.
We found that esophageal and gastric adenocarcinoma incidence rates increased significantly in all SEER regions between 1974 and 1998; however, the regional, temporal, and ethnic incidence patterns frequently differed between these two disease sites. The incidence rates for esophageal adenocarcinoma are continuing to rise, whereas incidence rates for gastric cardia adenocarcinoma have been relatively stable since the late 1980s. Among the 11 regions investigated, the Seattle and Connecticut registries experienced some of the highest rates for all cancers studied, whereas Utah experienced some of the lowest rates for all cancers regardless of gender and ethnicity (Tables 1, 2). These regional patterns varied by gender and ethnicity.
The current regional differences in esophageal adenocarcinoma incidence rates are largely from different rates of increase over time between registries rather than from baseline regional differences in disease incidence; the opposite is true for adenocarcinomas of the gastric cardia. For example, although esophageal adenocarcinoma incidence rates were comparable between Seattle and Utah in the middle 1970s (0.6 vs. 0.7 per 100,000 persons per year, respectively), marked differences in the rate of increase (800% in Seattle vs. only 300% in Utah) created the current substantial incidence differences. In contrast, gastric cardia adenocarcinoma rates already differed substantially between Seattle and Utah in the middle 1970s (3.1 vs. 0.7 per 100,000 persons per year), and increases of only 20% in Seattle and 300% in Utah between 1973–1974 and 1998 actually decreased the absolute incidence difference between the registries.
The current results extend prior reports of rising incidence rates for esophageal and cardia adenocarcinomas, the presence of incidence variation by ethnicity and gender, and stable or declining rates for squamous cell carcinoma.11–16 We now have demonstrated that esophageal and cardia adenocarcinoma rates also vary widely between regions within a single country, and the two disease sites (i.e., esophagus and gastric cardia) have different incidence patterns. These patterns suggest that regional variations in environmental risk factors are influencing disease incidence and that the two disease sites may have different risk factor profiles. This variability in regional rates complements similar site specific variation recently reported between countries, suggesting that the described international differences are not due solely to differences in classification and reporting methods between countries.1, 7 Our results also confirm reports of incidence variation by ethnicity and gender2, 3; however, we have now documented that these differences sometimes are overcome by regional differences (e.g., the relatively high rates of esophageal adenocarcinoma in black males in Connecticut).
There are several possible explanations for the geographic variability in incidence rates. First, although the SEER registry employs a standardized set of diagnostic criteria and classification methods, implementation of these criteria may vary between regions. In particular, the classification of tumors near the gastroesophageal junction as either esophageal or gastric in origin may differ between registries or over time.2, 3, 17, 18 However, if differences in classification of location alone explained the observed regional variability, then we would expect the combined esophageal and cardia rates to be similar across regions. However, the composite rates (esophageal plus gastric cardia adenocarcinoma) demonstrate persistent substantial differences between high incidence rates and low incidence areas (e.g., average esophageal plus gastric cardia adenocarcinoma rates in Seattle vs. Utah: 9.3 vs. 5.2 per 100,000 persons per year; P < 0.01). Furthermore, prior studies suggest that the rate trends for esophageal and cardia adenocarcinomas are not accounted for solely by potential diagnostic shifts of gastroesophageal junction or gastric carcinomas, although these may explain a portion of the variability.2, 3, 17
Second, regional incidence differences may reflect variable regional distribution of environmental or lifestyle risk factors. Putative risk factors include smoking, gastroesophageal reflux, obesity, micronutrient intake, and gastric colonization with the strains of the bacteria Helicobacter pylori.19–27 For example, Utah, which is a relatively low incidence area for both of these malignancies, has a high percentage of the population (> 70%) who are members of the Church of Jesus Christ of Latter-Day Saints,28 a population with lower exposure levels to cigarettes and some other putative cancer risk factors.29 Reflecting this, Utah's smoking rate in 1999 was substantially lower (14%) than some higher incidence states, e.g., Washington (22%) and Connecticut (22%).30 However, the variability of esophageal and cardia adenocarcinoma risk factors corresponding to the described geographic, ethnic, and gender incidence distributions have not been evaluated fully.
The strengths of our study included its use of the SEER cancer registry, a comprehensive population-based cancer registry with a high case ascertainment rate (98%) and consistent quality assurance procedures.8 Analyses of a wide range of geographic regions within a single registry system make SEER data less susceptible to potential biases in site classification. Weaknesses of the study included the possibility that residual classification differences between registries or over time within SEER may have contributed to the variability found; however, these differences are unlikely to account completely for our reported findings, as noted above.2, 3, 17
In summary, we report considerable regional, temporal, and ethnic differences between the incidence rates of esophageal adenocarcinoma compared with cardia adenocarcinoma. In addition, the national incidence rates for esophageal adenocarcinoma continue rising, whereas cardia adenocarcinoma rates have stabilized. These findings suggest these two malignancies may differ in important ways and should not necessarily be combined routinely in clinical or epidemiologic studies. We also report that the previously noted substantial incidence differences between white and black populations in the United States are influenced by regional differences, with the Connecticut registry reporting black male esophageal adenocarcinoma incidence rates comparable to those found in white males. Further studies examining temporal trends of diagnostic criteria will help evaluate any influence of changes in classification. In addition, ecologic and individual level studies evaluating regional and ethnic differences of putative risk factors are needed. Plausible explanations for the increasing incidence of esophageal and cardia adenocarcinomas must be able to account for the described regional, gender, temporal, and ethnic differences in incidence and for the differences between these two malignancies.
- 8SEER cancer statistics review, 1973–1997. Bethesda: National Cancer Institute, 2000., , , et al.
- 9International classification of diseases for oncology. 2nd edition. Geneva: World Health Organization, 1990., , .
- 10Stata Corporation. Stata statistical software: release 7.0. College Station: Stata Corporation, 2001.
- 28Church of Jesus Christ of Latter-day Saints statistics [accessed at http://www.adherents.com/largecom/com_lds.html], 2001.
- 30Centers for Disease Control and Prevention. Investment in tobacco control state highlights—2001. Volume URL. Atlanta: U.S. Department of Health and Human Services, Center for Disease Control and Prevention, 2001.