Abstract
Background:
Several U.S. subgroups have increased risk of gastric cancer and gastric intestinal metaplasia (GIM) and may benefit from targeted screening. We evaluated demographic and clinical risk factors for GIM and examined the interaction between race/ethnicity and birthplace on GIM risk.
Methods:
We identified patients who had undergone esophagogastroduodenoscopy with gastric biopsy from 3/2006-11/2016 using the pathology database at a safety-net hospital in Houston, Texas. Cases had GIM on ≥1 gastric biopsy histopathology, while controls lacked GIM on any biopsy. We estimated odds ratios (OR) and 95% confidence intervals (CI) for associations with GIM risk using logistic regression and developed a risk prediction model of GIM risk. We additionally examined for associations using a composite variable combining race/ethnicity and birthplace.
Results:
Among 267 cases with GIM and 1842 controls, older age (vs. <40 years: 40-60 years adjOR 2.02; 95%CI 1.17-3.29; >60 years adjOR 4.58; 95%CI 2.61-8.03), black race (vs. non-Hispanic white [NHW]: adjOR 2.17; 95%CI 1.31-3.62), Asian race (adjOR–2.83; 95%CI 1.27-6.29 and current smoking status (adjOR 2.04; 95%CI 1.39-3.00) were independently associated with increased GIM risk. While non-U.S. born Hispanics had higher risk of GIM (vs. NHW: adjOR 2.10; 95%CI 1.28-3.45), we found no elevated risk for U.S. born Hispanics (adjOR 1.13; 95%CI 0.57-2.23). The risk prediction model had area under the receiver operating characteristic (AUROC) of 0.673 (95%CI 0.636-0.710) for discriminating GIM.
Conclusions:
We found Hispanics born outside the U.S. were at increased risk of GIM while Hispanics born in the U.S. were not, independent of H. pylori infection. Birthplace may be more informative than race/ethnicity when determining GIM risk among U.S. populations.
Keywords: gastric intestinal metaplasia, gastric cancer, race/ethnicity, birthplace, risk factors, epidemiology
INTRODUCTION
Gastric cancer is the third leading cause of cancer-related mortality worldwide1, 2. In the United States, non-cardia gastric cancer has higher incidence, hospitalization and mortality rates in minorities compared to whites3. Gastric intestinal metaplasia (GIM), occurring mostly in the setting of Helicobacter pylori (H. pylori) infection, is a known precursor lesion for non-cardia gastric cancer4. Therefore, understanding the determinants of GIM is important for explaining the demographic variations in gastric cancer and improving the detection and prevention of gastric cancer.
Birthplace may be an additional risk factor for gastric cancer5–7. The rates of gastric cancer are highest among first generation immigrants and decrease with subsequent acculturated generations8–10, although the risk does not drop to the risk of the host country. Non-white race has been identified as a risk factor for GIM in Western populations11–14. However, few studies have taken into account place of birth and immigration status while examining the role of race/ethnicity on GIM risk in the U.S. There has been a disproportionate increase in non-H. pylori gastric cancer over time in certain sub-groups of the U.S. population15. This suggests other risk factors apart from H. pylori may be implicated in GIM and gastric cancer pathogenesis.
It is important and efficient to study risk factors for GIM in U.S. populations enriched with these high-risk individuals who would have the greatest potential benefit from gastric cancer screening. Therefore, we aimed to examine demographic and clinical risk factors for GIM in a high-risk U.S. population enriched with racial/ethnic minorities and immigrants. In addition, we specifically examined the association that race/ethnicity and birthplace has on the risk of GIM to determine the interplay between immigration status and race/ethnicity on GIM risk.
METHODS
Study Population and Design
We performed a retrospective cross-sectional study of all patients who underwent esophagogastroduodenoscopy (EGD) for any elective indication and had at least one gastric biopsy obtained from March 3, 2006 to November 30, 2016 at Ben Taub Hospital, a safety net hospital for Harris County providing care for a multiethnic population in Houston, Texas. The data sources included the Ben Taub Hospital pathology database, which is a clinical database of systematically recorded histopathology findings, and the EPIC electronic medical record (EPIC Hyperspace; Verona, WI). This study was approved by the Institutional Review Board for Human Subjects Research for Baylor College of Medicine.
Definition of Cases and Controls
GIM cases were defined using reports in the hospital pathology database. All patients with gastric intestinal metaplasia on ≥1 gastric biopsy were identified. Two independent reviewers reviewed the electronic medical record to confirm presence of GIM on gastric biopsy histopathology report. GIM cases were defined by consensus agreement between both reviewers. Patients with discordant GIM definition between the two reviewers or patients where neither reviewer found GIM were excluded from the study. An independent pathologist conducted a secondary review of the histopathology slides in a subset of 180 GIM cases, which was used in a sensitivity analysis. Controls were identified using the pathology database as patients who underwent EGD with at least one gastric biopsy for any indication and did not have evidence of GIM on any gastric biopsy report. GIM was defined as the presence of goblet cells on hematoxylin and eosin (H&E) stain. Histopathologic determination of H. pylori infection was determined by presence of the bacteria on H&E stain or in instances where the bacteria was difficult to find or limited in number, immunohistochemical stain was used at the discretion of the pathologist.
Data collection
We conducted automated abstraction of the EPIC electronic medical record for demographic characteristics including age at EGD, sex, race/ethnicity (non-Hispanic white [NHW], black, Hispanic/Latino, Asian, other), marital status (married, single, divorced/widowed, unknown), citizenship status (American citizen-born in USA, American citizen-naturalized, legal alien/legal sponsor/restricted visa, undocumented alien, unknown), and body mass index (BMI). We also conducted reviews and manual abstraction of lifestyle and clinical variables that were less reliable on automated abstraction, including smoking status (former, current, or never user), alcohol use (former, current, or never user; use defined as >1 drink per week), and presence of H. pylori prior to and at the time of EGD. Current H. pylori infection was defined as presence of H. pylori on gastric biopsy histopathology at the time of EGD. Ever H. pylori infection was defined as either current or previous H. pylori infection on gastric biopsy, positive stool antigen, serum antibody (i.e., IgG, IgA, IgM), or urease breathe test, receipt of H. pylori treatment, or H. pylori positivity or treatment at any outside hospital or clinic.
Statistical Analysis
We compared demographic, lifestyle, and clinical risk factors between GIM cases and controls using Chi-squared tests for categorical variables and Student’s t-tests for continuous variables. We conducted logistic regression models estimating odds ratios (OR) and associated 95% confidence intervals (CI) to examine independent associations between demographic, lifestyle, and clinical risk factors and GIM risk. We constructed a multivariable model by including risk factors with p<0.10 from univariate analysis (i.e., age, sex, race/ethnicity, smoking status) as well as conceptually important risk factors (i.e., birthplace, current H. pylori) irrespective of statistical significance. Birthplace was categorized as “Born in U.S.” for American citizen-born in USA and “Not born in U.S.” for American citizen-naturalized, legal alien, legal sponsor, restricted visa, and undocumented alien. We additionally examined a multivariable model replacing current H. pylori infection with ever H. pylori infection. We dropped all cases and controls with other/unknown race/ethnicity, unknown birthplace, unknown smoking status, or unknown H. pylori status from the logistic regression model. In a sensitivity analysis, we included those with these missing variables and included a “dummy” missing category for the variable.
In a sensitivity analysis, we compared the subset of GIM cases confirmed on a secondary histopathology review to the same controls to determine independent associations between demographic, lifestyle, and clinical risk factors with GIM risk. We assessed whether the association between race/ethnicity and risk of GIM was modified by birthplace by performing likelihood ratio tests of nested models with and without the race/ethnicity-birthplace interaction term.
To additionally examine the influence of birthplace and race/ethnicity on GIM risk, we created a single variable capturing both birthplace and race/ethnicity (NHW, black U.S. born, black non-U.S. born, Hispanic U.S. born, Hispanic non-U.S. born, Asian U.S. born, Asian non-U.S. born). In a secondary analysis, we examined for independent associations with GIM risk using the composite variable for birthplace and race/ethnicity.
We developed a risk prediction model of the risk of GIM based on the multivariate model. We included variables associated with GIM on the univariate analysis (p<0.10) or of clinical significance. Predictive discrimination, or ability of the model to discriminate between cases and controls, was reported using the area under the receiver operating characteristic curve (AUROC; [also known as the c-statistic]) and its 95% CI. Calibration of the model, which compares the predicted probabilities with the observed risk, was calculated using the Hosmer-Lemeshow goodness-of-fit statistic16 where a p-value >0.05 indicates sufficient calibration of the model.
All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC), and a 2-tailed p-value of < 0.05 was considered statistically significant.
RESULTS
Our cohort consisted of 267 cases with GIM and 1842 controls (Table 1). GIM cases were older than controls (57.7 vs 51.8 years; p<0.001) and more likely to be male (47.6% vs 38.6%; p=0.006). GIM cases were more likely to be black (27.7% vs 23.7%) and Asian (5.2% vs 2.4%) and less likely to be NHW (8.6% vs. 14.6%) or Hispanic (48.3% vs. 53.5%) (p=0.003). GIM cases were more likely to be current smokers (25.5% vs 18.7%; p=0.025) as compared to controls. There were no significant differences in the proportions of patients between cases and controls according to marital status, birthplace, alcohol use, drug use, BMI, or medication usage. Of 267 cases, 93 (34.6%) had a test for H. pylori in addition to gastric histopathology (stool antigen, serum antibody, urea breath test), while 619 out of 1842 (33.6%) controls had an additional test. There were no differences in proportions of patients with current (p=0.506) or previous H. pylori infection (p=0.315) between cases and controls.
TABLE 1:
Demographic and clinical characteristics of cases with gastric intestinal metaplasia (GIM) and controls
| GIM Cases (n=267) |
Controls (n=1842) |
P-value | |
|---|---|---|---|
| Age | <0.001 | ||
| <40 | 19 (7.12) | 288 (15.64) | |
| 40-60 | 136 (50.94) | 1113 (60.42) | |
| >60 | 112 (41.95) | 441 (23.94) | |
| Sex | 0.006 | ||
| Male | 127 (47.57) | 710 (38.55) | |
| Female | 140 (52.43) | 1132 (61.45) | |
| Race/Ethnicity | 0.003 | ||
| Non-Hispanic White | 23 (8.61) | 268 (14.55) | |
| Hispanic | 129 (48.31) | 985 (53.47) | |
| Black | 74 (27.72) | 437 (23.72) | |
| Asian | 14 (5.24) | 45 (2.44) | |
| Other/Unknown | 27 (10.11) | 107 (5.81) | |
| Marital Status | 0.474 | ||
| Married | 119 (44.57) | 746 (40.50) | |
| Divorced/Widowed | 61 (22.85) | 433 (23.51) | |
| Single | 86 (32.21) | 657 (35.67) | |
| Unknown | 1 (0.37) | 6 (0.33) | |
| Birthplace | 0.148 | ||
| Born in U.S. | 100 (37.45) | 785 (42.62) | |
| Not Born in U.S. | 164(61.42) | 1038(56.35) | |
| Unknown | 3 (1.12) | 19 (1.03) | |
| Smoking status | 0.025 | ||
| Never | 131 (49.06) | 1099 (59.66) | |
| Current smoker | 68 (25.47) | 344 (18.68) | |
| Former smoker | 66 (24.72) | 391 (21.23) | |
| Unknown/missing | 2 (0.75) | 8 (0.43) | |
| Alcohol use | 0.559 | ||
| Never | 185 (69.29) | 1272 (69.06) | |
| Current | 58 (21.72) | 438 (23.78) | |
| Former | 22 (8.24) | 126 (6.84) | |
| Unknown/missing | 2 (0.75) | 6 (0.33) | |
| BMI (kg/m2) | 0.990 | ||
| <25 | 47 (17.60) | 317 (17.21) | |
| 25-29.9 | 154 (57.68) | 1052 (57.11) | |
| ≥30 | 66 (24.72) | 473 (25.68) | |
| H. pylori ever | 0.315 | ||
| Negative | 119 (44.57) | 900 (48.86) | |
| Positive | 146 (54.68) | 936 (50.81) | |
| Unknown | 2 (0.75) | 6 (0.33) | |
| H. pylori current | 0.506 | ||
| Negative | 199 (74.53) | 1398 (75.90) | |
| Positive | 66 (24.72) | 438 (23.78) | |
| Unknown | 2 (0.75) | 6 (0.33) | |
| PPI use | 0.744 | ||
| No | 52 (19.48) | 298 (16.18) | |
| Yes | 183 (68.54) | 1321 (71.72) | |
| Unknown | 32 (11.99) | 223 (12.11) | |
| H2RA use | 0.997 | ||
| No | 156 (58.43) | 1092 (59.28) | |
| Yes | 79 (29.59) | 527 (28.61) | |
| Unknown | 32 (11.99) | 223 (12.11) | |
| Aspirin use | 0.829 | ||
| No | 192 (71.91) | 1369 (74.32) | |
| Yes | 43 (16.10) | 250 (13.57) | |
| Unknown | 32 (11.99) | 223 (12.11) | |
| NSAID use | 0.534 | ||
| No | 107 (40.07) | 640 (34.74) | |
| Yes | 128 (47.94) | 979 (53.15) | |
| Unknown | 32 (11.99) | 223 (12.11) |
BMI: body mass index; PPI: proton pump inhibitor; NSAID: nonsteroidal anti-inflammatory drug
Demographic Risk Factors
In the univariate logistic regression analyses, older age (vs. <40 years: 40-60 years OR 2.13; 95% CI 1.23-3.70; >60 years OR 4.65; 95% CI 2.65-8.19), male sex (OR 1.35; 95% CI 1.02-1.77), Black race (vs. NHW; OR 2.05; 95% CI 1.24-3.38), and Asian race (OR 3.76; 95% CI 1.79-7.89) were associated with GIM (Table 2). In the multivariate model, older age (vs. <40 years: 40-60 years adjOR 2.02; 95% CI 1.17-3.49; >60 years adjOR 4.58; 95% CI 2.61-8.03), Black race (vs. NHW: adjOR 2.17; 95% CI 1.31-3.62), Asian race (adjOR 2.83; 95% CI 1.27-6.29), and non-U.S. birthplace (adjOR 1.75; 95% CI 1.16-2.66) were significantly associated with increased risk for GIM while male sex (adjOR 1.19; 95% CI 0.88-1.60) and Hispanic race was not (adjOR 1.39; 95% CI 0.82-2.36). In a sensitivity analysis including those with unknown/missing race/ethnicity and birthplace, the effect sizes for race/ethnicity and birthplace were similar to the primary analysis.
TABLE 2:
Associations of demographic and clinical risk factors with risk of gastric intestinal metaplasia
| Univariate OR (95%CI) | Multivariate OR (95%CI) Current H. pylori |
Multivariate OR (95%CI) Ever H. pylori |
|
|---|---|---|---|
| Age (ref: <40 years) | |||
| 40-60 | 2.13 (1.23-3.70) | 2.02 (1.17-3.49) | 2.02 (1.17-3.49) |
| >60 | 4.65 (2.65-8.19) | 4.56 (2.60-8.01) | 4.58 (2.61-8.03) |
| Sex (ref: Female) | |||
| Male | 1.35 (1.02-1.77) | 1.19 (0.88-1.60) | 1.19 (0.88-1.60) |
| Race/Ethnicity (ref: NHW) | |||
| Hispanic | 1.57 (0.98-2.52) | 1.42 (0.84-2.41) | 1.39 (0.82-2.36) |
| Black | 2.05 (1.24-3.38) | 2.23 (1.34-3.70) | 2.17 (1.31-3.62) |
| Asian | 3.76 (1.79-7.89) | 2.85 (1.28-6.36) | 2.83 (1.27-6.29) |
| Birthplace (ref: Born in U.S.) | |||
| Not Born in U.S. | 1.21 (0.92-1.59) | 1.80 (1.19-2.72) | 1.75 (1.16-2.66) |
| Smoking status (ref: Never) | |||
| Current Smoker | 1.68 (1.21-2.34) | 2.05 (1.39-3.01) | 2.04 (1.39-3.00) |
| Former Smoker | 1.39 (0.99-1.95) | 1.33 (0.92-1.90) | 1.33 (0.92-1.91) |
| H. pylori current (ref: Negative) | |||
| Positive | 1.02 (0.73-1.41) | ||
| H. pylori ever (ref: Negative) | |||
| Positive | 1.20 (0.91-1.58) | 1.14 (0.86-1.51) |
CI: confidence interval; NHW: Non-Hispanic White; OR: odds ratio.
When black, Hispanic and Asian race/ethnicity was further categorized based on birthplace, black U.S. born (adjOR 1.98; 95% CI 1.19-3.30), Hispanic non-U.S. born (adjOR 2.10; 95% CI 1.28-3.45) and Asian non-U.S. born (adjOR 3.96; 95% CI 1.82-8.62) were associated with increased GIM risk compared to NHWs. For black non-U.S. born and Hispanic U.S. born, the risk for GIM was no different to NHWs. We had too few Asians patients born in the U.S. to draw a meaningful conclusion (Table 3). We found no evidence for statistical interaction between race/ethnicity and birthplace on GIM risk (Wald chi-square=5.97, p=0.202).
TABLE 3.
Association of combined race/ethnicity and birthplace with risk of gastric intestinal metaplasia
| Multivariate OR (95% CI) | |
|---|---|
| Age (ref: <40 years) | |
| 40-60 | 1.97 (1.14-3.40) |
| >60 | 4.49 (2.56-7.87) |
| Sex (ref: Female) | |
| Male | 1.21 (0.90-1.62) |
| Race/Ethnicity with Birthplace (ref: NHW) | |
| Black U.S. born | 1.98 (1.19-3.30) |
| Black non-U.S. born | 2.26 (0.90-5.66) |
| Hispanic U.S. born | 1.13 (0.57-2.23) |
| Hispanic non-U.S. born | 2.10 (1.28-3.45) |
| Asian U.S. born | 9.56 (0.86-105.69) |
| Asian non-U.S. born | 3.96 (1.82-8.62) |
| Smoking status (ref: Never) | |
| Current Smoker | 1.92 (1.31-2.82) |
| Former Smoker | 1.27 (0.89-1.82) |
| H. pylori ever (ref: Negative) | |
| Positive | 1.17 (0.88-1.55) |
CI: confidence interval; NHW: Non-Hispanic White; OR: odds ratio
Clinical Risk Factors
On univariate analysis, smoking status (current smoker OR 1.68; 95% CI 1.21-2.34) was positively associated with GIM while alcohol use and current H. pylori infection were not (Table 2). Smoking status remained associated with GIM on multivariate analysis (vs. never smoker; current smoker adjOR 2.04; 95% CI 1.39-3.00), while ever having H. pylori infection was not (adjOR 1.14; 95% CI 0.86-1.51). In the multivariate model including current H. pylori infection, the magnitude and direction of associations were similar for the demographic and clinical risk factors.
Sensitivity analysis
In a sensitivity analysis, we compared a subset of 180 GIM cases who underwent a secondary review and confirmation of the histopathology by an independent pathologist to 1842 controls (Supplementary Table 1). We found that the magnitude of association with age (ref <40 years: 40-60 years adjOR 1.77; 95% CI 0.94-3.33; >60 years adjOR 4.31; 95% CI 2.26-8.23), race/ethnicity (vs. NHW; Black adjOR 1.81; 95% CI 1.03-3.20; Asian adjOR 2.55; 95% CI 1.06-6.15), non-U.S. birthplace (adjOR 1.89, 95% CI 1.16-3.09), and smoking status (vs. never smoker: current smoker adjOR 2.02, 95% CI 1.29-3.18) and risk of GIM were similar to our primary analysis involving all 267 cases with GIM (Supplementary Table 2).
Risk Prediction Model Performance
The final risk prediction model, which included age, sex, race/ethnicity, birthplace, smoking status, and H. pylori infection, had an AUROC of 0.673 (95% CI 0.636-0.710; Figure 1) for discriminating GIM cases from controls. The model was well-calibrated (Hosmer-Lemeshow test p=0.829).
FIGURE 1:
Receiver operating characteristic (ROC) curve of GIM risk prediction model, including age, sex, race/ethnicity, birthplace, smoking status, and H. pylori infection
DISCUSSION
We found that birthplace outside the U.S. was associated with GIM, even after adjusting for race/ethnicity, in addition to older age (>40 years) and current smoking status. Black and Asian race/ethnicities were independently associated with GIM risk in addition to non-U.S. born Hispanic race/ethnicity while U.S. born Hispanic race/ethnicity was not. In our multiethnic cohort of patients undergoing EGD for various indications, over half of patients had evidence for H. pylori infection (current or prior), and we observed no association of H. pylori infection with risk of GIM. We developed a risk prediction model that included these risk factors, which had AUROC 0.673 for discriminating GIM risk.
The racial/ethnic disparities in prevalence of GIM reported in our study are in line with previous studies in other U.S. populations. In one U.S. veteran cohort, the prevalence of GIM was 50% in Hispanics and Blacks compared to 13% in NHWs undergoing EGD for symptomatic indications12. In another study where both asymptomatic and symptomatic patients were recruited to undergo systematic gastric mapping biopsies, the prevalence of GIM was also higher in Hispanics (29.5%) and Blacks (25.5%) compared to NHWs (13.7%)15. Studies with large populations of race/ethnic minorities have found non-white race/ethnicity to be independently associated with higher risk of GIM (Hispanic OR 1.63-6.7412, 15, 17, 18, Black OR 1.8715, Asian OR 2.0-7.3917–19). Our study found similar associations of Black and Asian race/ethnicity and GIM, and further found the elevated risk associated with Hispanic race/ethnicity to be restricted to Hispanics who were non-U.S. born.
Examining the association of birthplace and immigration with GIM risk is a novel aspect of our study. We found that non-U.S. birthplace was associated with risk of GIM independent of race/ethnicity. Furthermore, among Hispanics, immigrant status or non-U.S. birthplace confers increased risk of GIM compared to U.S. born Hispanics. The association with birthplace likely reflects additional familial and environmental factors associated with GIM that race/ethnicity inadequately capture. Few U.S. studies have addressed the risk of GIM based on birthplace and immigration status. This is of particular importance as gastric cancer mortality rates are higher in foreign-born individuals compared to individuals born in the U.S.20 Among Hispanic and Asian immigrants, the rates of gastric cancer are not as high as in the country of origin but still elevated compared to rates in the host country (i.e., U.S.)8, 21. One study found 1.5-2 times higher gastric cancer rates in Asia-born Japanese-Americans compared to U.S.-born Japanese-Americans, which still had 3-times elevated risk compared to U.S.-born whites10.
The association of race/ethnicity and gastric cancer is likely through immigration and importation of H. pylori as well as other familial and environmental factors from high-risk countries. Therefore, immigration and birthplace may be a more direct risk factor to estimate gastric cancer risk than race/ethnicity alone, especially among Hispanic and Asian populations in the U.S. A large systematic review and meta-analysis of 38 studies conducted in the U.S. found the gastric cancer risk among first-generation immigrants from high incidence areas (e.g. East Asia, Mesoamerica, Eastern Europe, Andean, Central Asia, Middle East) to low incidence areas (e.g. United States, Canada, Australia, Western Europe, Brazil, Israel) was up to 1.2-5.1-times higher compared to the destination population incidence8. They also found that the risk was lower in subsequent generations birthed from these immigrants, but the risk was never diminished to the same as that of the destination population. This meta-analysis included 5 U.S. studies that found a 1.5-5.8-fold increased risk of gastric cancer in the immigrant population compared to the general U.S. population. These results suggest that environmental risk factors and acculturation may play a role in gastric cancer risk following immigration.
We did not find a significant association with H. pylori infection, both current or ever, and risk of GIM in our study. With progression of gastric inflammation and atrophy in the setting of H. pylori, the resulting achlorhydric environment becomes increasing inhospitable to H. pylori making it more difficult to detect on gastric biopsy22, and the number of patients who had documentable evidence of non-biopsy H. pylori testing was small (33.8%). Thus, GIM may be the “point of no return” where H. pylori eradication has minimal benefit to gastric cancer risk23. This finding may also reflect a selection bias in the study. Furthermore, because of the high prevalence of H. pylori infection in our cohort (51.3%) in both cases and controls, our study population may not effectively be able to distinguish the role of H. pylori in GIM risk. Special stains and immunohistochemistry staining for H. pylori was performed at the discretion of the pathologist based on histopathology findings.
To our knowledge, there is no published risk model incorporating demographic risk factors to predict risk of GIM in U.S. populations. Moreover, the U.S. does not have a general population-based gastric cancer screening strategy based on unfavorable cost effectiveness consideration, although certain subpopulations remain at high enough risk for a favorable cost effectiveness argument. Other high-risk countries (i.e., Korea, Japan) have universal gastric cancer screening recommendations using only age cut-off of ≥40 years24. Such a screening strategy would not be cost effective in the U.S. given the overall low prevalence of gastric cancer. Biomarkers (e.g., pepsinogens) have been incorporated into risk models to estimate risk of atrophic gastritis and GIM in Asia. However, pepsinogen tests are not widely available or used in the U.S. as the sensitivity varies greatly based on cut-off values and population prevalence of GIM25. Using a high-risk population in a U.S. safety net hospital, we developed a risk model for GIM detection to be used prior to endoscopy using readily available, measurable demographic and clinical risk factors (i.e., age, sex, smoking status, H. pylori, race/ethnicity, and birthplace). Such a risk model may inform clinical decisions about which high-risk patients to refer for endoscopic screening of GIM.
The prevalence of GIM may have been underestimated in our study as our cohort only included patients who underwent EGD with biopsies for either symptomatic indication or endoscopic findings. Those with normal endoscopic findings were less likely to be biopsied, and therefore even the control group in the study was more likely to have abnormalities, such as suspected H. pylori infection. While we do not have information on the exact number of gastric biopsies obtained in this study, a common practice at our hospital is to perform 5 gastric biopsies including the lesser and greater curvatures of the antrum and corpus. However, most of the gastric biopsies were placed into one jar so location of GIM could not be determined. A limitation is the lack of GIM location in our study as a previous U.S. study found stronger magnitude of risk factors in extensive GIM (antrum and corpus) as compared to antrum-only GIM11. Furthermore, we did not include the indication for the EGD or the presence of high-risk symptoms (e.g., anemia, dyspepsia) as risk factors for GIM in the model. All included cases and controls likely had a clinical reason to warrant referral to gastroenterology clinic and/or an approval of an EGD request by a gastroenterologist. Furthermore, high-risk symptoms (i.e., iron deficiency anemia, epigastric pain, abnormal imaging, gastric ulcer, weight loss) have been shown to poorly correlate with the presence of GIM19. Thus, presence of high-risk symptoms was likely non-differential between cases and controls and likely not a significant predictor of GIM. Furthermore, patients with gastric cancers were excluded for our analysis.
Despite the potential limitations related to selection bias and limited generalizability, GIM cases were determined by consensus independent review of the histopathology reports by two reviewers, thus minimizing misclassification of cases. We further validated our findings in a sensitivity analysis of a subset of 180 cases that underwent independent secondary histopathology review and found similar magnitude of associations among the risk factors. We attempted a secondary review of all GIM cases to reduce misclassification of cases and controls but only a subset of 180 GIM cases were available for review; these 180 cases were selected nearly randomly. We abstracted variables using two methods of traditional manual medical record review along with information technology automated abstraction of the medical record, thus efficiently and accurately collecting demographic and clinical data.
In this study, we found that Black and Asian race/ethnicity and birthplace outside the U.S., particularly in Hispanics and Asians, as well as older age and current smoking status to be independent risk factors for GIM in a multiethnic U.S. population. Although race/ethnicity is a strong risk factor for GIM in the U.S., birthplace and immigration status may be more important risk factors for GIM. Identification of these demographic and clinical risk factors would aid in risk stratification to better identify high-risk populations who would benefit most from gastric cancer screening programs.
Supplementary Material
STUDY HIGHLIGHTS.
What is known:
Gastric cancer and gastric intestinal metaplasia (GIM) disproportionately affect several subgroups in the U.S.
Understanding demographic and clinical risk factors for GIM among U.S. populations may identify high risk groups with greatest potential benefit from targeted screening.
What is new here:
Older age, Black race, Asian race, smoking, and non-U.S. birthplace were independently associated with GIM risk in a multiethnic U.S. population.
Among Hispanics, non-U.S. born Hispanics had elevated GIM risk, while the risk in U.S. born Hispanics was not different from non-Hispanic whites.
Grant support:
This work was supported in part by National Institutes of Health grant P30 DK056338 (Study Design and Clinical Research Core), which supports the Texas Medical Center Digestive Diseases Center and in part by the Caroline Wiess Law Fund for Research in Molecular Medicine. This research was supported in part with resources at the VA HSR&D Center for Innovations in Quality, Effectiveness and Safety (#CIN 13-413), at the Michael E. DeBakey VA Medical Center, Houston, TX. The opinions expressed reflect those of the authors and not necessarily those of the Department of Veterans Affairs, the U.S. government or Baylor College of Medicine.
Footnotes
Conflicts of interest: The authors report no competing interests for this publication.
REFERENCES
- 1.Bray F, Ferlay J, Soerjomataram I, et al. Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2018;68:394–424. [DOI] [PubMed] [Google Scholar]
- 2.Thrift AP, El-Serag HB. Burden of Gastric Cancer. Clin Gastroenterol Hepatol 2020;18:534–542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Schlansky B, Sonnenberg A. Epidemiology of noncardia gastric adenocarcinoma in the United States. Am J Gastroenterol 2011;106:1978–85. [DOI] [PubMed] [Google Scholar]
- 4.Correa P, Haenszel W, Cuello C, et al. A model for gastric cancer epidemiology. Lancet 1975;2:58–60. [DOI] [PubMed] [Google Scholar]
- 5.SEER*Explorer: An interactive website for SEER cancer statistics [Internet]. Beta Version: Surveillance Research Program, National Cancer Institute. [Google Scholar]
- 6.Lui FH, Tuan B, Swenson SL, et al. Ethnic disparities in gastric cancer incidence and survival in the USA: an updated analysis of 1992-2009 SEER data. Dig Dis Sci 2014;59:3027–34. [DOI] [PubMed] [Google Scholar]
- 7.Miller BA, Chu KC, Hankey BF, et al. Cancer incidence and mortality patterns among specific Asian and Pacific Islander populations in the U.S. Cancer Causes Control 2008;19:227–56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Pabla BS, Shah SC, Corral JE, et al. Increased Incidence and Mortality of Gastric Cancer in Immigrant Populations from High to Low Regions of Incidence: A Systematic Review and Meta-Analysis. Clin Gastroenterol Hepatol 2020;18:347–359.e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Ziadeh C, Ziogas A, Anton-Culver H. Cancer risk in different generations of Middle Eastern immigrants to California, 1988-2013. Int J Cancer 2017;141:2260–2269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Kamineni A, Williams MA, Schwartz SM, et al. The incidence of gastric carcinoma in Asian migrants to the United States and their descendants. Cancer Causes Control 1999;10:77–83. [DOI] [PubMed] [Google Scholar]
- 11.Tan MC, Mallepally N, Liu Y, et al. Demographic and Lifestyle Risk Factors for Gastric Intestinal Metaplasia Among US Veterans. Am J Gastroenterol 2020;115:381–387. [DOI] [PubMed] [Google Scholar]
- 12.Fennerty MB, Emerson JC, Sampliner RE, et al. Gastric intestinal metaplasia in ethnic groups in the southwestern United States. Cancer Epidemiol Biomarkers Prev 1992;1:293–6. [PubMed] [Google Scholar]
- 13.IARC Helicobacter pylori Working Group. Helicobacter pylori eradication as a strategy for preventing gastric cancer (IARC Working Group Reports, No. 8). Lyon, 2014. [Google Scholar]
- 14.Asaka M, Kato M, Kudo M, et al. Atrophic changes of gastric mucosa are caused by Helicobacter pylori infection rather than aging: studies in asymptomatic Japanese adults. Helicobacter 1996;1:52–6. [DOI] [PubMed] [Google Scholar]
- 15.Nguyen TH, Tan MC, Liu Y, et al. Prevalence of Gastric Intestinal Metaplasia in a Multiethnic US Veterans Population. Clin Gastroenterol Hepatol 2020;14:S1542-3565(20)30325-6. Online ahead of print. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Steyerberg EW, Vickers AJ, Cook NR, et al. Assessing the performance of prediction models: a framework for traditional and novel measures. Epidemiology (Cambridge, Mass.) 2010;21:128–138. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Genta RM, Turner KO, Sonnenberg A. Demographic and socioeconomic influences on Helicobacter pylori gastritis and its pre-neoplastic lesions amongst US residents. Aliment Pharmacol Ther 2017;46:322–330. [DOI] [PubMed] [Google Scholar]
- 18.Choi CE, Sonnenberg A, Turner K, et al. High Prevalence of Gastric Preneoplastic Lesions in East Asians and Hispanics in the USA. Dig Dis Sci 2015;60:2070–6. [DOI] [PubMed] [Google Scholar]
- 19.Da B, Jani N, Gupta N, et al. High-risk symptoms do not predict gastric cancer precursors. Helicobacter 2019;24:e12548. [DOI] [PubMed] [Google Scholar]
- 20.Hallowell BD, Endeshaw M, Senkomago V, et al. Gastric cancer mortality rates among US and foreign-born persons: United States 2005–2014. Gastric Cancer 2019;22:1081–1085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Kim GH, Bang SJ, Ende AR, et al. Is screening and surveillance for early detection of gastric cancer needed in Korean Americans? Korean J Intern Med 2015;30:747–58. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Genta RM, Gürer IE, Graham DY, et al. Adherence of Helicobacter pylori to areas of incomplete intestinal metaplasia in the gastric mucosa. Gastroenterology 1996;111:1206–11. [DOI] [PubMed] [Google Scholar]
- 23.Chen HN, Wang Z, Li X, et al. Helicobacter pylori eradication cannot reduce the risk of gastric cancer in patients with intestinal metaplasia and dysplasia: evidence from a meta-analysis. Gastric Cancer 2016;19:166–75. [DOI] [PubMed] [Google Scholar]
- 24.Kim GH, Liang PS, Bang SJ, et al. Screening and surveillance for gastric cancer in the United States: Is it needed? Gastrointest Endosc 2016;84:18–28. [DOI] [PubMed] [Google Scholar]
- 25.Huang YK, Yu JC, Kang WM, et al. Significance of Serum Pepsinogens as a Biomarker for Gastric Cancer and Atrophic Gastritis Screening: A Systematic Review and Meta-Analysis. PLoS One 2015;10:e0142080. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.