The authors investigated whether stage at diagnosis, cancer treatments, and survival of Asian and Pacific Islander (API) gastric cancer patients in the United States vary by birthplace.
The authors investigated whether stage at diagnosis, cancer treatments, and survival of Asian and Pacific Islander (API) gastric cancer patients in the United States vary by birthplace.
The authors studied 6454 API and 10,099 non-Hispanic white (NHW) patients diagnosed with gastric cancer from the Surveillance, Epidemiology, and End Results program between 1992 and 2005. In descriptive analyses, stage, receipt of adequate lymph node examination (ALNE), and surgery were compared among US-born APIs, foreign-born (FB) APIs, and NHWs. Multivariate polytomous logistic and proportional hazards regression models were used to assess differences in cancer stage and survival, respectively, adjusted for clinical and demographic factors.
As a group, APIs were more likely than NHWs to present with earlier-stage diagnoses and receive surgery and ALNE (P < .001). However, FB (adjusted odds ratios [aOR], 0.79; 95% confidence interval [CI], 0.73-0.86) but not US-born APIs (aOR, 1.05; 95% CI, 0.92-1.20) were significantly more likely to present at earlier stages than NHWs. Compared with NHW patients, FB and US-born APIs were more likely to receive surgery (adjusted risk ratio [aRR], 1.06; 95% CI, 1.03-1.09 and aRR, 1.09; 95% CI, 1.03-1.14, respectively) and ALNE (aRR, 1.29; 95% CI, 1.19-1.41 and aRR, 1.14; 95% CI, 1.00-1.32, respectively). In fully adjusted models, FB (adjusted relative hazard ratios [aHR], 0.86; 95% CI, 0.82-0.90) but not US-born APIs (aHR, 0.96; 95% CI, 0.89-1.04) had more favorable survival than NHWs.
The earlier-stage diagnosis, more complete surgical treatment, and improved survival of Asians and Pacific Islanders with gastric cancer may result from less aggressive tumors or more prompt recognition and thorough evaluation of early symptoms. Further study of these factors could improve outcomes for all patients with gastric cancer. Cancer 2009. © 2009 American Cancer Society.
Despite the declining incidence of gastric cancer in the United States in recent decades, this disease remains a major concern. Current estimates suggest that approximately 21,500 new people were diagnosed and 11,000 died from gastric cancer in 2008.1 The incidence of gastric cancer varies by race and ethnicity, disproportionately affecting Asian and Pacific Islanders (APIs).2, 3 Compared with their counterparts residing in Asia and the Pacific Islands, API gastric cancer patients residing in the West experience reduced overall survival rates and stage-stratified survival.4-8
Currently, 13 million APIs reside in the United States. This number is projected to triple by 2050, making APIs the fastest growing minority group in the United States. APIs also comprise the second largest immigrant group, representing 25% of all immigrants.9 Given the sizable numbers of US-born and foreign-born APIs, further investigation of their health outcomes is warranted.
Although early migration studies found that mortality because of gastric cancer was higher in immigrant Asians than in whites,10, 11 more recent evidence indicates that Asian race is associated with superior overall survival.12-16 However, previous analyses have not considered whether stage at diagnosis and survival differ by birthplace. Some studies suggest that foreign birth is associated with worse survival for APIs with breast and colorectal cancer.17, 18 Yet it remains unknown whether birthplace correlates with cancer outcomes for API patients with gastric cancer.
Therefore, we investigated the associations between birthplace and stage at diagnosis, initial treatment, and overall mortality among APIs with gastric cancer, and compared their experiences to non-Hispanic white (NHW) patients. We further explored differences in survival among the larger API ethnic subgroups, including Japanese, Chinese, Korean, Filipino, Vietnamese, Pacific Islander, and Hawaiian adults.
We used publicly available data from 6 US cancer registries participating in the Surveillance, Epidemiology, and End Results (SEER) program19 from 1992 to 2005. We used data from registries where at least 5% of gastric cancer patients were APIs (Hawaii, Los Angeles, San Francisco/Oakland, San Jose/Monterey, Seattle/Puget Sound, and Atlanta). Of all APIs with gastric cancer recorded by the SEER program during these years, 97% of APIs patients resided in these areas. SEER captures all incident cancers (98% case ascertainment) as well as information on patient demographics at diagnosis, cancer characteristics, and initial course of treatment, defined as within 4 months of diagnosis from 1973 to 1998 and within 12 months of diagnosis after 1998. Ascertainment of surgery and radiation therapy by SEER is generally complete.20, 21 However, chemotherapy ascertainment is known to be incomplete and is not publicly released in SEER data. Vital status is tracked annually, and death certificates are used to capture underlying cause of death.
Because we used publicly available, de-identified data, our study was deemed exempt from review by the institutional review board.
NHW and API patients from the 6 registries were eligible for our study if they were ≥18 years old when diagnosed with a pathologically confirmed first primary invasive gastric adenocarcinoma between 1992 and 2005 (n = 23,185). Adenocarcinoma was defined using International Classification of Disease for Oncology, 3rd edition codes 8140-8145, 8260-63, 8310, 8323, 8480, 8481, 8490, 8510, 8510, 8560, and 8570-8576.13 We excluded 84 cases where the diagnosis was determined from death certificates or autopsy. Our final study sample consisted of 16,553 patients, including 10,099 (61%) NHW and 6454 (39%) API patients.
Race/ethnicity and birthplace were primary factors of interest. SEER collects information about race/ethnicity largely from chart review. We classified race/ethnicity into 2 categories, NHW and API, which included Chinese, Japanese, Filipino, Hawaiian, Korean, Asian Indian/Pakistani, Vietnamese, Laotian, Hmong, Kampuchean, Thai, Micronesian, Chamorran, Guamanian, Polynesian, Samoan, Tahitian, Tongan, Melanesian, Fiji Islander, New Guinean, and other Asian and Pacific Islander.
We classified birthplace into US-born, foreign-born, and unknown. SEER obtains information about birthplace from multiple sources, including medical records, state motor vehicle records, and death certificates. The sensitivity and positive predictive value of birthplace data among APIs has been reported as >90%.22, 23 Overall, 19% of API patients were missing birthplace data (ranging from 14% of Japanese to 30% of Chinese patients). Therefore, we included API patients with unknown birthplace as a separate group rather than impute birthplace information or exclude them from our analyses. We classified patients as NHW (n = 10,099), US-born API (n = 1944), foreign-born API (n = 3253), or API of unknown birthplace (n = 1257).
We did not examine birthplace among NHW patients, because it is not well documented in medical records for white patients, and US Census data indicate that only 3.9% of NHW individuals in the United States are foreign-born.9
Finally, recognizing that APIs are a heterogeneous group and that aggregate data may mask important disparities, we examined 7 API ethnic groups with sufficient samples: Japanese (n = 2026), Chinese (n = 1460), Korean (n = 1187), Filipino (n = 641), Vietnamese (n = 391), Hawaiian (n = 290), and Pacific Islander (n = 140). We further explored the Japanese, Chinese, Korean, and Filipino groups, comparing US-born and foreign-born patients.
Our 2 primary outcomes were stage at diagnosis and overall mortality after diagnosis. Because stage using the American Joint Committee on Cancer (AJCC) classification system for patients with gastric cancer is not available in SEER data until 2004, we used the SEER historical staging system (local, regional, and distant) in our primary analyses. In addition, we derived AJCC stage from individual data elements when available. For this subset (44%), we performed sensitivity analyses to assess the impact of derived AJCC stage on survival. We assessed deaths from all causes and censored observations of patients alive on December 31, 2005 (n = 3478).
We examined the association of birthplace with surgical treatment, adequate lymph node examination (ALNE), and radiation therapy. ALNE was defined as examination of at least 15 lymph nodes as suggested by the AJCC.24 Receipt of surgery was dichotomized for analyses investigating the likelihood of surgery, but was categorized in other models as none, partial gastrectomy, total/near gastrectomy, or gastrectomy en bloc. Patients with unknown surgery status were excluded. In analyses examining ALNE, only patients who received surgery (n = 8575) and had known lymph node assessment were included (n = 8031). Receipt of radiation was dichotomized for all analyses.
We performed bivariate analyses to evaluate the relationship between race/ethnicity (according to birthplace) and stage at diagnosis, receipt of ALNE, and surgery. To examine differences in stage, we used multivariate ordinal polytomous logistic regression25 and adjusted for covariates previously shown to be associated with stage, including sex, age and marital status at diagnosis, geographic location (SEER tumor registry), and year of diagnosis. Patients with unknown stage (n = 1861) were excluded from this analysis. We estimated adjusted odds ratios and 95% confidence intervals (CIs) for later stage (regional or distant) at diagnosis. We used multivariate logistic regression to examine differences in initial treatment (surgery, ALNE) by race/ethnicity and birthplace after adjusting for characteristics at diagnosis, geographic location, and tumor grade and location. Because treatment was relatively common (>10%), we derived adjusted risk ratios (aRRs) and 95% CIs using a log binomial regression model.26 Because previous studies have demonstrated geographic variations in cancer care, we further explored the association between race/ethnicity and SEER region for each outcome.
Finally, we computed Kaplan-Meier estimates of overall survival and fit Cox proportional hazards regression models to examine differences in overall survival by race/ethnicity and birthplace after adjustment for the factors listed above, stage, and treatment. We present adjusted relative hazard ratios (aHRs) and 95% CIs; aHR <1.00 indicates longer survival among API patients relative to NHW patients. Because several API patients had unknown birthplace, we performed sensitivity analyses by alternatively categorizing API patients of unknown birthplace as US-born APIs then as foreign-born APIs. All statistical analyses were conducted using SAS version 9.1 (SAS Institute, Cary, NC).
The characteristics of patients with gastric cancer are presented in Table 1. APIs were primarily concentrated in Hawaii and California. Compared with NHW patients, APIs were more likely to be women, to be married, to be diagnosed at younger ages, to have tumors located in the lower third of their stomach, and to present at earlier stages of disease and with lower grade (P < .001). Compared with all other groups, US-born APIs were significantly less likely to have signet-ring cell histology.
|NHW (n=10,099), %||APIs|
|US-Born (n=1944), %||Foreign-Born (n=3253), %||Unknown Birthplace (n=1257), %|
|Age at diagnosis, mean±SD*||71±13||72±12||66±15||69±14|
|Location of tumor*|
|Cardia and fundus, upper third||42||16||12||13|
|Body, middle third||7||12||9||10|
|Antrum and Pylorus, lower third||19||30||37||38|
|Stomach, lesser, greater curve, NOS||24||29||32||31|
|Stage at diagnosis*|
Among API patients, US-born APIs were less likely to be women, to be married, and to be diagnosed at younger ages (P < .05) than foreign-born APIs. US-born APIs were more likely to present with tumors located in the proximal stomach and with lower-grade tumors than foreign-born APIs (P < .05) (Table 1).
In bivariate analyses, API patients overall and by birthplace were less likely to present with later-stage disease than NHW patients (Table 2). These differences remained significant after adjustment for the API groups, except for US-born APIs.
|Race/Ethnicity||Unadjusted Odds Ratio* (95% CI)||Adjusted Odds Ratio*† (95% CI)|
|All API||0.77 (0.72-0.82)||0.78 (0.72-0.84)|
|API by birthplace|
|US-born||0.83 (0.76-0.91)||1.05 (0.92-1.20)|
|Foreign-born||0.82 (0.76-0.89)||0.79 (0.73-0.86)|
|Unknown||0.55 (0.49-0.62)||0.58 (0.51-0.65)|
Table 3 presents relative rates of surgery and ALNE by race/ethnicity and birthplace, adjusted for demographic factors, geographic location, and tumor characteristics. Regardless of birthplace, APIs were more likely than NHW patients to receive surgery and undergo ALNE, although the difference between US-born APIs and NHWs was not statistically significant.
|Race/Ethnicity||Receipt of Surgery, Adjusted* RR (95% CI)||Receipt of Adequate Lymph Node Examination,† Adjusted‡ RR (95% CI)|
|All API||1.08 (1.05-1.11)||1.24 (1.15-1.33)|
|API by birthplace|
|US-born||1.09 (1.03-1.14)||1.14 (1.00-1.32)|
|Foreign-born||1.06 (1.03-1.09)||1.29 (1.19-1.41)|
|Unknown||1.13 (1.09-1.17)||1.21 (1.08-1.37)|
Because the proportion of APIs varied substantially across the SEER registries, and previous studies have shown geographic variations in cancer care,13, 27 we further explored regional variations in stage at diagnosis, receipt of surgery, and ALNE (Table 4). We found that in Hawaii, where APIs comprised 88.6% of the sample, patients were more likely to be diagnosed at earlier stages and had the highest rates of surgery (60.8%) and ALNE (56.0%). After adjustment, we found no differences in stage at diagnosis across the regions.
|Registry||No.||Proportion of APIs in Registry,* %||Diagnosed at Local Stage,* %||Receipt of Surgery,* %||Receipt of Adequate Lymph Node Examination,* %|
We found regional variations in receipt of surgery (P < .001) across the 6 registries. In contrast to Hawaii, Atlanta and Seattle/Puget Sound, areas with the lowest proportion of APIs (12.8% and 14.4%, respectively) also had the lowest surgical rates (49.8% and 49.5%, respectively). In each registry, APIs were more likely to receive surgery than NHWs. However, we found no statistically significant interaction between race/ethnicity and SEER area. After adjustment, only patients in Los Angeles were more likely to receive surgery than those in Seattle/Puget Sound (aRR, 1.05; 95% CI, 1.01-1.09). With the exception of Hawaii, there were small differences in receipt of ALNE across the other registries (range, 27.6%-30.2%). After adjustment, only patients in Hawaii (aRR, 1.66; 95% CI, 1.49-1.86) were more likely to receive ALNE than those in Seattle/Puget Sound.
APIs experienced more favorable overall and stage-specific survival compared with NHWs (Fig. 1). The median survival was 14 months for API patients and 8 months for NHW patients (P < .0001). Among API patients, median survival was 14 months for foreign-born and 10 months for US-born patients. Table 5 presents unadjusted and adjusted hazard ratios of overall mortality comparing APIs with NHW patients. After adjustment for demographic and tumor characteristics (Table 5, second column), APIs had significantly lower mortality than NHW, even after considering birthplace. Further adjustment for differences in treatment (Table 5, third column) had a modest impact on the observed difference in mortality between NHW patients and API groups, except for US-born APIs, where the survival difference compared with NHWs was no longer statistically significant.
|Unadjusted HR (95% CI)||Adjusted* HR (95% CI)||Treatment Adjusted† HR (95% CI)|
|All API||0.70 (0.68-0.73)||0.80 (0.76-0.83)||0.84 (0.80-0.87)|
|API by birthplace|
|US-born||0.86 (0.81-0.90)||0.88 (0.82-0.95)||0.96 (0.89-1.04)|
|Foreign-born||0.70 (0.66-0.73)||0.83 (0.79-0.87)||0.86 (0.82-0.90)|
|Unknown||0.52 (0.48-0.56)||0.64 (0.59-0.70)||0.67 (0.62-0.73)|
In unadjusted analyses, patients in Hawaii and Los Angeles experienced longer survival compared with those in Seattle/Puget Sound. However, these differences were no longer significant after adjustment (data not shown).
We also determined the hazard ratios of overall mortality comparing APIs and NHW patients, substituting derived AJCC stages for the SEER historical staging system in our adjusted models. Although AJCC stage could be derived for only 44% of gastric cancer patients in this study, our findings did not change; foreign-born (aHR, 0.86; 95% CI, 0.79-0.94) but not US-born API patients (aHR, 1.01; 95% CI, 0.89-1.14) experienced significantly longer survival compared with NHW patients.
API patients with unknown birthplace had better survival than the US-born and foreign-born APIs. Thus, in sensitivity analyses that included APIs with unknown birthplace with US-born APIs, the adjusted hazard ratio (including treatment) was 0.81 (95% CI, 0.76-0.86). When APIs with unknown birthplace were included with the foreign-born APIs, this adjusted hazard ratio was 0.80 (95% CI, 0.71-0.85).
For API ethnic groups with sufficient samples, we explored ethnic variations in overall mortality compared with NHW patients (Fig. 2) and found that overall mortality was lower for each ethnic group. After adjustment, differences for Japanese, Chinese, Korean, and Vietnamese patients remained statistically significant. Regardless of birthplace, API patients did not experience significantly worse survival outcomes compared with NHW patients (Fig. 3). Within API subgroups, adjusted overall survival was more favorable for foreign-born Korean patients than their US-born counterparts, compared with NHW patients. Overall survival for Chinese patients was more favorable than NHW patients regardless of birthplace.
We examined stage at presentation, initial treatment, and survival of API gastric cancer patients categorized by birthplace and ethnicity, distinguishing this study from previously published analyses. Compared with NHWs, API gastric cancer patients residing in the United States were more likely to present at earlier stages of disease, receive surgery, and have an adequate number of lymph nodes examined. API gastric cancer patients as a group, foreign-born patients in particular, and subgroups of Japanese, Chinese, Korean, and Vietnamese patients had significantly longer adjusted survival than NHW patients.
This finding of improved survival among API gastric cancer patients as an aggregate group is consistent with other studies,12-16, 28 although it appears that earlier stage at diagnosis and longer survival for foreign-born APIs is limited to gastric cancer patients. Theuer et al16 also found a survival advantage specific for Asian (mostly foreign-born) patients with gastric cancer in California, but not for Asian patients with colorectal cancer. Previous studies indicate that foreign-born API patients with other malignancies are more likely to present with advanced-stage disease and experience worse survival than their US-born counterparts.29, 30 For some cancer diagnoses, this may be because foreign-born APIs are less likely to undergo cancer screening,31, 32 perhaps due inadequate access to care, limited English proficiency, and other cultural factors.33-35 Because no formal population screening programs exist in the United States for early detection of gastric cancer, disparities in stage at presentation cannot be attributed to differences in screening. However, because minorities and immigrants typically have worse access to healthcare, it is counterintuitive that API patients with gastric cancer present at earlier stages, particularly for those who are foreign-born. Because gastric cancer is a greater public health concern in Asian countries, APIs may be more aware of symptoms associated with gastric cancer, especially those who have migrated to the United States. Alternatively, physicians may have a lower threshold for recommending endoscopic evaluation of API patients with upper gastrointestinal symptoms because of their greater risk of gastric cancer.
Several studies highlight the impact of appropriate surgery and ALNE on increased survival for patients with gastric cancer. Surgical intervention may extend survival even if performed for palliative purposes.36, 37 Furthermore, examination of at least 15 lymph nodes correlates with longer survival, presumably because of more accurate staging.13, 27, 38, 39 We were surprised to find that API patients were more likely than NHW patients to receive surgery and have the recommended number of lymph nodes assessed. Although surgery and ALNE had only a modest impact on aggregate API survival in our study, differences in initial treatments explained some of the observed survival advantage among US-born APIs compared with NHWs. Findings from our unadjusted analyses raise the possibility that areas with a very high concentration of APIs, such as Hawaii and Los Angeles, may also have a greater cultural awareness such that gastric cancer patients are diagnosed earlier and have higher rates of surgery and ALNE. Future research should examine the characteristics of providers and the hospitals where gastric cancer patients received treatment. API patients, particularly the foreign-born, may seek treatment from clinicians familiar with APIs and who may suspect gastric cancer more readily when API patients present with common symptoms and also provide more aggressive treatment.
Differences in tumor biology according to ethnicity may explain some of the improved survival among API patients. Gastric tumors among API patients may be less aggressive than in other racial groups.7, 40 We observed significantly longer survival among foreign-born APIs, although they were as likely as NHWs to have the more aggressive signet-ring cell histology. However, APIs were also more likely to have tumors located in the distal stomach, a characteristic associated with less aggressive disease and more favorable surgical outcomes and survival.41 This finding is consistent with other studies that have reported a higher prevalence of distal gastric tumors and increased survival for APIs compared with NHWs.12, 14, 16, 42, 43
Furthermore, differences in prevalence of distal gastric tumors may be related in part to Helicobacter pylori infection, a well-known risk factor that also correlates with distal gastric adenocarcinoma and not with more aggressive proximal tumors.44 The incidence of H. pylori infection is higher in Asian countries than in the United States. It is possible that foreign-born APIs migrated to the United States with H. pylori infection, which affected their gastric tumor characteristics and survival.
Recent studies demonstrate differences in E-cadherin expression,45 and microsatellite instability,46 which may reflect less aggressive gastric tumors in Japanese patients compared with American patients of European descent or British patients. Further investigation of biological differences in gastric tumors by ethnicity is necessary to determine whether biologic characteristics indicate less aggressive disease and therefore superior survival for APIs.
Although not available in SEER, environmental factors, such as smoking, alcohol intake, diet, and body mass index,6, 47 may also influence survival in patients with gastric cancer. While the prevalence of these risk factors varies substantially by Asian ethnicity; in the United States they are generally less prevalent in APIs than NHWs3, 47-50 and may contribute to survival differences.40, 51
Our study has important limitations. We were unable to assess chemotherapy, which may have a positive impact on survival.52-54 However, a recent study of gastric cancer patients in the United States found that differences in adjuvant therapy use did not explain the survival advantage experienced by APIs.12
AJCC staging was unavailable for most of our patients. Therefore, we used the SEER historic staging system. Although this system employs broader stage classifications, the categories are standardized and consistent over time. Furthermore, sensitivity analyses of the subset of cases with AJCC stage did not alter our findings.
Although SEER mandates the collection of birthplace, this variable is often coded as unknown. Rather than excluding patients with unknown birthplace (19% of API cases), we performed sensitivity analyses. Grouping APIs with unknown birthplace as either US-born or foreign-born APIs did not alter our findings.
Among API ethnic groups, there is substantial variation in socioeconomic status (SES) and length of time since immigrating to the United States, which may influence ethnic-specific health outcomes. However, these factors are not captured by SEER. Further research should assess the relationship between SES and gastric cancer outcomes. There are also important differences in patterns of immigration among API ethnic groups, with the Japanese having 1 of the earliest immigration histories, whereas groups such as Chinese have come in waves. In our study, API patients were generally from more affluent ethnic groups with a longer history of US immigration, including Japanese and Filipinos,3 although Vietnamese patients, generally considered a less affluent group with a more recent immigration history,3 also demonstrated more favorable survival compared with NHW patients.
In summary, we found that API patients with gastric cancer, particularly those who are foreign-born, present with earlier stages of disease, are more likely to receive surgery and ALNE, and experience significantly longer survival than NHW patients. Although specific to gastric cancer, these results suggest that patients are not always disadvantaged by foreign birth. This conclusion warrants further study to determine whether biological differences or other factors not examined in this study influence survival differences. These results may help guide future interventions for all gastric cancer patients. Furthermore, our study suggests that the survival of NHW gastric cancer patients may be improved with more extensive surgical resection and thorough lymph node assessments.
Supported by a National Cancer Institute Training Grant, Program in Cancer Outcomes Training R25 (CA92203-04 for S.A.D.B. while a postdoctoral fellow from July 2006 to June 2008 at the Institute of Technology Assessment at Massachusetts General Hospital, Boston, Massachusetts).