Background

In the United States, it is estimated that 14,440 men and 3,550 women will be diagnosed with and 15,210 people will die of cancer of the esophagus in 2013 [1]. The incidence rate for esophageal cancer significantly increased from 1975 through 2009 in Surveillance, Epidemiology and End Results (SEER 9) Registries [2]. Esophageal adenocarcinoma arises from glandular cells present at the junction of the esophagus and stomach [3]. There are two major histological types of the cancer: squamous cell carcinoma and adenocarcinoma. While squamous cell carcinoma incidence rates have decreased, rates for adenocarcinoma have increased over 6-fold in the United States since the 1970s [2]. Most of the increasing incidence and death rates are attributed to the rising burden of esophageal adenocarcinoma experienced among white males [4].

Because the esophagus is located near other vital organs, lesions are most treatable when detected as the premalignant condition of Barrett’s esophagus [5] or as early-stage cancer. In SEER Registries, overall 5-year esophageal cancer survival increased from 5 to 20% from the 1970s through 2004. Furthermore, from 2001 to 2007, 5-year esophageal cancer survival was 38% among localized-stage cases and 20% among regional-stage cases, but only 3% among distant-stage cases [2]. Although there is currently no standard or routine screening test for esophageal cancer, detection of Barrett’s esophagus and early-stage esophageal cancers are still possible through medical procedures including endoscopy, biopsy, brush and balloon cytology, chromoendoscopy, and fluorescence spectroscopy [6]. Further experimental studies are needed to evaluate the effectiveness of these screening procedures for esophageal cancers.

Low socioeconomic status may be a risk factor for esophageal adenocarcinoma [7,8]. Specific socioeconomic factors that are significantly associated with increased EAC risk include education [8], wealth [8], occupational status [7], and marital status [7,8]. Other established risk factors for esophageal adenocarcinoma include older age [5,9,10], male gender [9,10], white race [11], obesity [9], and smoking [5,9,10,12,13].

Social support may encourage lifestyle and screening behaviors that reduce the risk of esophageal adenocarcinoma [7]. Positive social support can confer health benefits in several ways [14], including improving awareness of and access to available health care resources and reducing behavioral risk factors that predispose to esophageal adenocarcinoma. In SEER Registries, we examined associations between esophageal adenocarcinoma stage, case characteristics, and area-level attributes associated with this diagnosis including median household income and percent foreign-born population.

Methods

The NCI IRB and the IRB of the contributing SEER registries have granted a waiver from HIPAA requirements because the SEER data used in this analysis have been de-identified, protecting against disclosure of personal information. The study was restricted to incident cases from 16 Surveillance, Epidemiology and End Results Registries that were diagnosed between the years 2000 and 2007. The Registries included eight states (New Mexico, Hawaii, Utah, Iowa, Connecticut, Kentucky, Louisiana, New Jersey) and eight areas (Atlanta, Rural Georgia, Greater California, San Francisco-Oakland, San Jose/Monterey, Los Angeles, Seattle, Detroit), which together represent approximately 20 percent of the United States [15]. Census tract- and county-level census data regarding the percent foreign-born population were obtained from the Census Bureau [16].

Results

Discussion

In this study, no association was seen between reported birth inside the United States versus outside the country and distant-stage esophageal cancer. While the majority of cases were born in the United States, at both the census tract and county level, cases that lived in areas with a high percentage of non-U.S. born residents had significantly higher odds of distant-stage esophageal adenocarcinoma. This finding has potential implications for cancer control because screening to detect early-stage cancer may improve prognosis [6]. The findings also demonstrate that area-level and individual attributes can operate independently, with the potential for misclassification if area-level measures are used as proxies for personal attributes.

At the individual level, the risk of developing esophageal adenocarcinoma is associated with limited social support and low socioeconomic status [7,8]. We were, however, unable to identify studies that related esophageal adenocarcinoma stage at diagnosis to area-level percent foreign-born population. Studies of other cancer sites may be informative in this regard. In SEER registries, Hispanics living in neighborhoods with higher density Hispanic populations were more likely to be diagnosed with late-stage breast, cervical, and colorectal cancers [18]. Foreign-born Hispanic women in California residing in high-enclave neighborhoods were also more likely than U.S.- born Hispanic women who reside in assimilated neighborhoods to be diagnosed with distant-stage breast cancer [19]. Low income, poor health insurance coverage, and language barriers are also often experienced within less assimilated Hispanic communities [20]. Another interesting finding at the individual level was the lower risk of distant-stage EAC for men compared to women. This association was further strengthened in the adjusted model. Men have a higher risk of developing EAC and Barrett’s esophagus, therefore, they may be identified earlier through screening. Since screening practices for EAC have changed over time with no clear guidelines for a standard screening procedure, we further stratified by year of diagnosis, however, no clear time trend was evident that could be explained by screening.

Keegan and colleagues hypothesized that the surplus of distant-stage cancer in enclave communities was a function of limited access to screening [19]. Low income [21], poor health insurance coverage [22], and language barriers [18] are often encountered in incompletely assimilated communities [20]. Thus, low social position in areas with sizeable non-native populations may contribute to the association in the present report between high area-level percent foreign-born population and distant-stage esophageal adenocarcinoma. Better access to health care [22] and higher SES [18] may contribute to the finding that white cases in our study were less likely to be diagnosed with distant-stage esophageal adenocarcinoma than non-white cases.

Strengths of the study included the large sample size and linkage to area-level attributes. There was, however, limited availability of individual-level socioeconomic data, and nearly 30% of cases were missing data regarding place of birth. A large proportion of cases with missing place of birth had localized/regional stage of disease, were younger than 70 years of age, and were diagnosed in more recent years. We suspect that place of birth is most complete among cases with worse outcomes because death certificates are a common source of this information. Nonetheless, stratified analyses in registries with the most complete data on nativity supported the null finding. A limited amount of demographic data is typically collected in cancer registries. To overcome this limitation of cancer surveillance, area-level data are sometimes used as proxies for individual data. This practice is based on the assumption that people residing in a common area often share similar socioeconomic attributes. As our findings suggest, caution is warranted against this practice, because area-level measures can operate differently from individual-level measures. First, attributes of residents can differ from prevailing attributes. In addition, while area-level socioeconomic status affects social support and quality of care within a community [20], it occurs at the ecologic level.

In summary, while residing in an area with a high percent of non-U.S. born individuals was associated with distant-stage esophageal adenocarcinoma, no such association was evident among cases with individual level data on place of birth. The findings suggest a need for caution regarding the use of area-level measures as proxies for individual level attributes. The findings of this study could also help target resources toward specific neighborhoods and individuals at risk for late-stage esophageal adenocarcinoma.

Acknowledgments

We would like to thank Steve Scoppa and Todd Gibson for database assistance, Benmei Liu for small area estimates, Zaria Tatalovich for consultation on geocoding, Li Zhu, Deborah Rohm Young, and Mandi Yu for critical review, and Linda Anderson for scientific editing.