Abstract
The objective of this study was to examine the associations between dietary patterns and the risk of Barrett’s esophagus, a precursor to esophageal adenocarcinoma. The authors conducted a case-control study within the Kaiser Permanente Northern California population between 2002 and 2005. Patients with a new diagnosis of Barrett’s esophagus (n = 296 cases) were matched to persons with gastroesophageal reflux disease (n = 308) without Barrett’s esophagus and to population controls (n = 309). Dietary information was obtained from a validated, 110-item food frequency questionnaire. A principal component analysis was used to identify major dietary patterns. Two major dietary patterns were “Western” (high in fast food and meat) and “health-conscious” (high in fruits, vegetables, and nonfried fish). When cases and population controls were compared, strong adherence to the health-conscious dietary pattern was inversely associated with Barrett’s esophagus (odds ratio = 0.35, 95% confidence interval: 0.20, 0.64; fourth vs. first quartile comparison). In contrast, data suggested an adverse effect of the Western dietary pattern on the risk of Barrett’s esophagus, although no dose-effect relation was found. Results suggest strong associations between a diet rich in fruits and vegetables and the risk of Barrett’s esophagus.
Keywords: Barrett esophagus, diet, esophageal neoplasms, factor analysis, statistical, food habits
The incidence of esophageal adenocarcinoma has increased dramatically in the last three decades. Barrett’s esophagus is a metaplastic transformation of the esophageal squamous epithelium into specialized intestinal columnar epithelium; it is associated with gastroesophageal reflux disease (GERD) and is considered a strong risk factor for and a common precursor of esophageal adenocarcinoma (1–3). Patients with Barrett’s esophagus are at a 30–40-fold increased risk of developing esophageal adenocarcinoma (1, 2); thus, identification of risk factors for Barrett’s esophagus may pinpoint early events in the carcinogenic pathway for esophageal adenocarcinoma amenable to interventions to reduce cancer risk (4–8). Diet is a modifiable risk factor that may influence cancer risk through several mechanisms, but little is known about the associations of overall dietary patterns with the carcinogenic sequence for Barrett’s esophagus and esophageal adenocarcinoma.
Dietary factors such as antioxidants (9), fruits and vegetables (10–15), and fiber (10, 13, 16, 17) are reported to have inverse associations with the risk of esophageal adenocarcinoma, whereas fat, saturated fat, and cholesterol may increase cancer risk (13, 16). However, studies of esophageal adenocarcinoma alone are unable to identify where in the proposed carcinogenic sequence from normal mucosa to GERD to Barrett’s esophagus to esophageal adenocarcinoma these factors may influence development of the cancer. Diet may impact the risk of esophageal adenocarcinoma by 1) influencing the risk of GERD (a strong risk factor for both Barrett’s esophagus and esophageal adenocarcinoma), 2) modifying the risk of developing Barrett’s esophagus by influencing mucosal integrity or healing, or 3) altering the rate of neoplastic progression from Barrett’s esophagus to cancer by changing the propensity for genetic damage to the cells (3, 18). Examination of dietary effects on the risk of Barrett’s esophagus may enable us to look at the early events in the carcinogenic process and thus may have both clinical and public health implications.
Compared with conventional analyses of individual nutrients, a study of overall dietary patterns more effectively captures the complexities of dietary intake. By identifying the most common dietary patterns in a population, we can evaluate the unified effects of nutrients and food items often consumed in combination. In addition, dietary pattern analysis is potentially useful in making dietary recommendations because it may be easier for patients to understand and incorporate recommendations for dietary patterns than methods to increase dietary intake of a particular nutrient (19).
Although a recent study reported a nonsignificant inverse association between fruits and vegetables intake and the risk of Barrett’s esophagus (20), we know of no previous study that has investigated the effects of overall dietary patterns. We examined this association by using a case-control design that compared patients with a new diagnosis of Barrett’s esophagus with population controls and, separately, with patients with GERD to evaluate risk factors for Barrett’s esophagus among persons with GERD.
MATERIALS AND METHODS
Study population
We conducted a case-control study within the population of Kaiser Permanente Northern California, an integrated health services delivery organization. Kaiser Permanente Northern California serves approximately 3.3 million persons; its membership demographics closely approximate those of the underlying census population of Northern California (21, 22). Eligible subjects were all adult (aged 18–79 years) Kaiser Permanente Northern California members who were continuously enrolled for at least 2 years prior to their index period, met the case or control definitions outlined below, and were able to understand spoken and written English. The index date for cases was the date of Barrett’s esophagus diagnosis. The index date for controls was the midpoint of each 2–3-month selection interval for the cases. The questionnaires asked participants to report exposures in the year prior to their index date. The population and GERD comparison groups were frequency matched to the cases by gender, age at the index date, and geographic region (each subject’s home facility).
Case definition
Cases were eligible Kaiser Permanente Northern California members with a new Barrett’s esophagus diagnosis made between October 2002 and September 2005 according to International Classification of Diseases, Ninth Revision code 530.2 (which, at Kaiser Permanente Northern California, is uniquely coded as “Barrett’s esophagitis”) or College of American Pathologists code 73330 (“Barrett’s esophagus”). A single board-certified gastroenterologist (D. A. C.) then reviewed the endoscopy and pathology records of potentially eligible cases. Subjects were included if the endoscopist clearly described a visible length of columnar-type epithelium proximal to the gastroesophageal junction/gastric folds, this area was biopsied, and the biopsies showed specialized intestinal epithelium (23). The following patients were excluded: those with only gastric-type metaplasia of the esophagus according to all pathology evaluations; those with columnar metaplasia without features of intestinal metaplasia according to all pathology readings; those without a biopsy of esophageal origin; those whose biopsies showed only a mildly irregular squamocolumnar junction (i.e., an “irregular z-line”); and those with a prior Barrett’s esophagus diagnosis. Pathology slides underwent a separate manual review by a gastrointestinal pathologist (G. J. R).
Population controls
Population controls were randomly selected from among members without an electronic diagnosis of Barrett’s esophagus at the time that the Barrett’s esophagus cases were identified.
GERD comparison group
GERD comparison group members were randomly selected from among persons with the following characteristics prior to their index date: a GERD-related diagnosis code (International Classification of Diseases, Ninth Revision code 530.11 (reflux esophagitis) or 530.81 (gastroesophageal reflux)); a prescription for at least a 90-day supply of a histamine-2 receptor antagonist or a proton pump inhibitor (medications used for treating GERD symptoms) in the previous year (from electronic pharmacy records); no prior Barrett’s esophagus; and an esophagogastroduodenoscopy administered close to the index date that did not demonstrate esophageal columnar metaplasia of any type.
Exposure assessment
We assessed subjects’ dietary intake by using a validated food frequency questionnaire (Block 1998, full length, 110 food items) for intake of various food items during the year prior to diagnosis of Barrett’s esophagus (24–27). This questionnaire has good correlations with multiple-day dietary records in a wide variety of population subgroups and in assessing past dietary habit (10–15 years ago) (28), and it correlates with serum measures for lipids and antioxidants (29, 30). We excluded subjects if more than 20 food items were missing or if total caloric intake was extremely out of the range of the normal distribution curve (e.g., >6,000 kcal/day or <400 kcal/day).
Subjects included in the analysis
Interviewed in the original study were 320 Barrett’s esophagus cases, 316 GERD controls, and 317 population controls. Of these persons, 296 Barrett’s esophagus cases, 308 GERD controls, and 309 population controls were included in the analyses, after exclusion of subjects who did not complete the food frequency questionnaire (n = 28) or whose responses on the questionnaire were extreme, as described previously (>20 skipped items, n = 5; total caloric intake >6,000 kcal/day, n = 3; total caloric intake <400 kcal/day, n = 4.) The interviewed subjects represented 57 percent of all living, eligible subjects who were able to be contacted by phone. For the entire group of potentially eligible subjects identified, the interviewed subjects represented 43 percent of all those eligible. Reasons for nonparticipation for the entire group were as follows: declined to participate (33 percent), unable to contact by phone (18 percent), severe physical or mental disorders (5 percent) (these subjects primarily were excluded by their physician prior to contact), or deceased (1 percent).
Statistical analysis
We first assigned number of servings of each food item per week for each subject. We used the principal component analysis option in the Proc Factor command (SAS statistical software; SAS Institute, Inc., Cary, North Carolina) to identify and help interpret major dietary patterns. Principal component analysis is a data reduction technique that reduces the number of variables by creating a small number of patterns that account for much of the variance in the individual food items. In determining which patterns or factors to retain, we used a combination of 1) the Kaiser-Guttman rule (retaining components with an eigenvalue >1), 2) the Scree test, and 3) the “interpretability criteria,” which interprets the substantive meaning of the retained patterns and verifies whether the interpretation makes sense given current knowledge (31–33). Factor loadings, which indicate each food item’s contribution to a specific dietary pattern, were calculated by using orthogonal rotation, which provides patterns that are not correlated with each other to achieve a simpler structure (31).
Once major dietary patterns were identified, each subject was assigned a score for each one. The scores, also computed with the Proc Factor command, are weighted sums of intakes, with weights equal to factor loadings. Subjects with a high score for a dietary pattern, compared with subjects with lower scores, had a stronger tendency to follow that type of dietary pattern. The scores were categorized into quartiles by using the distribution of the population controls.
We assessed the association between dietary pattern scores and Barrett’s esophagus by using standard analytic techniques for evaluating case-control studies and unconditional logistic regression (34, 35). We evaluated the following additional variables as potential confounders: ethnicity (White vs. non-White), smoking (ever vs. never smoker), body mass index (weight (kg)/height (m)2), recent alcohol consumption (number of drinks/week), aspirin or nonsteroidal antiinflammatory drug use, a comorbidity index (the DxCg score, which creates a predictive comorbidity score based on demographic data, medical coding, and pharmacy utilization) (36, 37), and total daily caloric intake. Confounders were included in the final model if they altered the β coefficient by more than 10 percent. SAS statistical software was used for all analyses, and all tests of statistical significance are two sided. The data collection and analyses were approved by the institutional review board.
RESULTS
Baseline characteristics
The baseline characteristics of the study subjects are presented in table 1. The demographic characteristics were fairly evenly distributed among the three groups, although cases tended to include a slightly higher proportion of ever smokers compared with the other groups. The average intakes of energy and total fat were comparable among the three groups, although population controls tended to have somewhat higher intakes of fruits and vegetables. Among the cases, the length of the Barrett’s segment was <3 cm in 118 subjects (37 percent) and ≥3 cm in 151 subjects (47 percent); the length was not reported for 51 subjects (16 percent).
TABLE 1.
Characteristics of groups of subjects in the case-control population studied regarding an association of dietary patterns with risk of Barrett’s esophagus, Northern California, 2002–2005
| Cases (n = 296) |
GERD* controls (n = 308) |
Population controls (n = 309) |
||||
|---|---|---|---|---|---|---|
|
|
|
|
||||
| No. | % | No. | % | No. | % | |
| Age (years)† | 62.3 (10.7) | 61.7 (10.8) | 62.3 (10.3) | |||
| Race | ||||||
| White | 255 | 86 | 247 | 80 | 262 | 85 |
| Black | 5 | 2 | 20 | 7 | 16 | 5 |
| Hispanic | 24 | 8 | 20 | 6 | 12 | 4 |
| Asian | 4 | 1 | 8 | 3 | 9 | 3 |
| Others | 8 | 3 | 13 | 4 | 10 | 3 |
| Gender | ||||||
| Male | 218 | 74 | 211 | 68 | 208 | 67 |
| Female | 77 | 26 | 97 | 32 | 101 | 33 |
| Smoking status (ever smoker) | 197 | 67 | 182 | 59 | 173 | 56 |
| Alcohol consumption (no. of drinks/week) | ||||||
| None | 77 | 26 | 109 | 36 | 78 | 25 |
| Light (≤1) | 153 | 52 | 131 | 42 | 157 | 51 |
| Moderate (1–2) | 27 | 9 | 27 | 9 | 48 | 15 |
| Heavy (≥2) | 39 | 13 | 41 | 13 | 26 | 8 |
| Body mass index (kg/m2)† | 29.4 (5.8) | 28.9 (5.3) | 29.5 (5.8) | |||
GERD, gastroesophageal reflux disease.
Values are expressed as mean (standard deviation).
Dietary patterns
Eight factors had an eigenvalue >1; from the Scree plot and interpretability criteria, we identified from these eight factors two major interpretable dietary patterns. Factor-loading matrixes for these factors are listed in table 2; for simplicity, only those dietary scores greater than 0.35 or smaller than −0.35 are shown.
TABLE 2.
Factor-loading* matrix for the major factors (dietary patterns†) identified by using the food frequency questionnaire, Northern California case-control population, 2002–2005‡
| Food item | “Western” diet | “Health-conscious” diet |
|---|---|---|
| Beef | 59 | |
| French fries | 53 | |
| Hamburger | 52 | |
| Pizza | 49 | |
| Refried beans | 49 | |
| Mustard | 47 | |
| Tacos | 45 | |
| White potatoes | 44 | |
| Pork chop | 44 | |
| Pork spareribs | 44 | |
| Chili | 43 | |
| Hot dog | 43 | |
| Salty snacks | 43 | |
| Spaghetti | 40 | |
| Fried chicken | 38 | |
| Bacon | 36 | |
| Sausage | 36 | |
| Jelly | 35 | |
| Broccoli | 60 | |
| Carrots | 51 | |
| Other fruits | 50 | |
| Peach | 49 | |
| Nonfried fish | 48 | |
| Orange | 46 | |
| Other vegetables | 46 | |
| Green salad | 45 | |
| Apple | 43 | |
| Tomatoes | 43 | |
| Green beans | 42 | |
| Spinach | 42 | |
| Cantaloupe | 39 | |
| Strawberry | 39 | |
| Vegetable soup | 39 | |
| Salad dressing | 36 | |
| Vegetable stew | 36 | |
| Baked beans | 35 | |
| Tofu | 35 |
Factor loadings indicate each food item’s contribution to a specific dietary pattern.
Refer to the Results section of the text for a definition of these dietary patterns.
Values reflect factor loading multiplied by 100 and rounded to the nearest integer. For simplicity, only those food groups with factor loading values ≥35 are listed in the table.
The first factor was loaded heavily with meat products, fast food (e.g., French fries, pizza, hamburgers, and tacos), soft drinks, beer/liquor, and coffee and was low in tofu, cooked cereals, fruits, and water; we labeled this factor the “Western” dietary pattern (range of dietary score: −1.3 to 3.2). The second factor was loaded heavily with fruits and vegetables, nonfried fish, and tofu and was low in meat, salty snacks, fried foods, and soft drinks; we labeled this factor the “health-conscious” dietary pattern (range of dietary score: −1.5 to 5.9). The first factor explained 6.7 percent of the total variance of the data, and the second factor explained 6.0 percent of the total variance.
Western dietary pattern and Barrett’s esophagus
Barrett’s esophagus cases were more likely than the population controls to adhere to a Western diet rich in meats and fast food, although there was not a conclusive dose trend across the quartile categories (table 3). When Barrett’s esophagus cases were compared with population controls, those in the third quartile of this dietary pattern were more than twice as likely to have Barrett’s esophagus compared with subjects in the lowest quartile of adherence to this dietary pattern (odds ratio (OR) = 2.30, 95 percent confidence interval (CI): (1.26, 4.21); third vs. first quartiles) (table 3). However, the association between the fourth quartile and the first quartile was weaker and nonsignificant, suggesting that there is no dose-effect relation.
TABLE 3.
Adjusted risk estimates for each dietary pattern*: Barrett’s esophagus cases vs. population controls from the Northern California case-control study, 2002–2005
| Quartile | “Western” diet | “Health-conscious” diet | ||||
|---|---|---|---|---|---|---|
|
|
|
|||||
| Cases/controls (no.) |
Odds ratio† |
95% confidence interval |
Cases/controls (no.) |
Odds ratio† |
95% confidence interval |
|
| 1 | 57/78 | 1 | 106/78 | 1 | ||
| 2 | 81/77 | 1.78 | 1.08, 2.93 | 72/77 | 0.65 | 0.41, 1.03 |
| 3 | 85/76 | 2.30 | 1.26, 4.21 | 73/76 | 0.65 | 0.39, 1.06 |
| 4 | 73/78 | 1.39 | 0.66, 2.93 | 45/78 | 0.35 | 0.20, 0.64 |
| p for trend | 0.12 | 0.001 | ||||
Refer to the Results section of the text for a definition of these dietary patterns.
The model was adjusted for age, sex, race (White vs. non-White), location of the facility, and energy intake.
Regarding the comparison of cases with GERD controls, the associations were weaker and nonsignificant (table 4). This finding suggested that this dietary pattern is not a risk factor for Barrett’s esophagus among persons with GERD.
TABLE 4.
Adjusted risk estimates for each dietary pattern*: Barrett’s esophagus cases vs. GERD† controls from the Northern California case-control study, 2002–2005
| Quartile | “Western” diet | “Health-conscious” diet | ||||
|---|---|---|---|---|---|---|
|
|
|
|||||
| Cases/controls (no.) |
Odds ratio‡ |
95% confidence interval |
Cases/controls (no.) |
Odds ratio‡ |
95% confidence interval |
|
| 1 | 57/67 | 1 | 106/96 | 1 | ||
| 2 | 81/70 | 1.53 | 0.91, 2.58 | 72/63 | 1.01 | 0.64, 1.59 |
| 3 | 85/89 | 1.29 | 0.71, 2.36 | 73/80 | 0.81 | 0.51, 1.29 |
| 4 | 73/82 | 0.79 | 0.36, 1.74 | 45/69 | 0.56 | 0.32, 0.96 |
| p for trend | 0.38 | 0.04 | ||||
Refer to the Results section of the text for a definition of these dietary patterns.
GERD, gastroesophageal reflux disease.
The model was adjusted for age, sex, race (White vs. non-White), location of the facility, and energy intake.
Health-conscious diet and Barrett’s esophagus
Barrett’s esophagus cases were less likely than the population controls to adhere to a health-conscious diet high in fruits, vegetables, and nonfried fish. Persons with the strongest adherence to this dietary pattern had a 65 percent lower risk of Barrett’s esophagus compared with those in the lowest quartile (OR = 0.35, 95 percent CI: 0.20, 0.64; fourth vs. first quartiles), with a trend for decreasing risk across the quartiles (table 3).
Similar associations were observed regarding comparisons between Barrett’s esophagus cases and GERD controls (table 4). The health-conscious diet was inversely associated with risk of Barrett’s esophagus (OR = 0.56, 95 percent CI: 0.32, 0.96; fourth vs. first quartile), with a statistically significant test for trend across the quartiles (p = 0.04). These data suggest that, among patients with GERD, a diet high in fruits, vegetables, and nonfried fish is inversely associated with risk of Barrett’s esophagus.
Supplemental analyses
Analyses of confounding and interaction
We found no evidence of confounding by smoking, body mass index, recent alcohol consumption, aspirin or nonsteroidal antiinflammatory drug use, vitamin supplement use, or a comorbidity index. A fully adjusted model (containing all these factors plus age, gender, ethnicity, location of the facility, and energy) for a comparison of the fourth with the first quartile regarding the health-conscious diet (OR = 0.35, 95 percent CI: 0.20, 0.64; cases vs. population controls) was not substantially different from a model that contained only age, gender, ethnicity, location of the facility, and energy (OR = 0.39, 95 percent CI: 0.22, 0.74). Longterm (≥2 years) vitamin supplement intake did not confound the association.
We also evaluated whether the dietary pattern associations differed depending on the use of dietary supplements containing vitamin C, obesity, or GERD symptoms (i.e., the presence of interaction with these factors). None of these variables showed statistically significant interactions.
Analysis of GERD symptoms
GERD is associated with Barrett’s esophagus; thus, dietary patterns may be associated with Barrett’s esophagus through a dietary pattern causing GERD. If this were true, we would expect that including GERD in the model would decrease or abolish the association between each dietary pattern and Barrett’s esophagus. However, inclusion of a variable of at least weekly GERD symptoms did not substantially modify any of the associations shown, suggesting that these associations are relatively independent of GERD symptoms. For example, the risk estimate for the health-conscious diet in a model containing GERD (OR = 0.39, 95 percent CI: 0.21, 0.77; fourth vs. first quartile comparison; case vs. population control) was similar to that from a model without the GERD variable (OR = 0.35, 95 percent CI: 0.20, 0.64). Similar patterns were observed for the Western dietary pattern (data not shown).
Stratification by the length of Barrett’s esophagus
To evaluate whether the effects of diet differ by the lengths of Barrett’s esophagus segments, we stratified the results by short (<3 cm) and long (≥3 cm). When cases and population controls were compared, the association became stronger for a long segment of Barrett’s esophagus (≥3 cm) (OR = 0.18, 95 percent CI: 0.08, 0.40; fourth vs. first quartile) and weaker for shorter segments (OR = 0.53, 95 percent CI: 0.23, 1.22). There was no clear association for the Western dietary pattern (data not shown).
Evaluation for nonresponse bias
To evaluate nonresponse bias, we used electronic medical data available on many nonresponders. When electronic data were used, participants did not differ significantly from nonparticipants by gender, smoking status, or body mass index. Participants were less likely to be Asian or Hispanic and were slightly older (age 62 vs. 59 years). When telephone data were used, we found no differences by aspirin/nonsteroidal antiinflammatory drug use. Analyses confined to the population controls suggested that participants were more likely than nonparticipants to have an electronic diagnosis of GERD (25.9 percent vs. 13.6 percent, p < 0.01) and a slightly higher total comorbidity score (2.8 vs. 2.0, p < 0.01).
DISCUSSION
To our knowledge, this is the first community-based study to evaluate the effect of overall dietary patterns on the risk of Barrett’s esophagus. Individuals adhering to dietary patterns high in fruits, vegetables, and nonfried fish had a substantially lower risk of developing Barrett’s esophagus when cases were compared with either population controls or GERD controls. The inverse association was the strongest between this health-conscious diet and long-segment Barrett’s esophagus. In contrast, persons adhering strongly to a Western diet (high in fast food and meat) were at a moderately increased risk of Barrett’s esophagus, in comparisons of cases with population controls, but no association was observed when cases were compared with GERD controls. These findings suggest that dietary patterns may influence the risk of Barrett’s esophagus among both the general population and persons with GERD.
These results extend the prior finding of a possible inverse association between intakes of antioxidants, fruits, and vegetables and esophageal adenocarcinoma (9, 10, 15); however, dietary pattern analyses provide additional information beyond individual micronutrients or food item intakes alone. Particular nutrients may simply be surrogates for certain dietary habits. For instance, antioxidant intake, a factor commonly reported to have a beneficial effect against cancer, may be a proxy for a diet rich in fruits and vegetables, which contain many other chemicals and nutrients. A prior analysis by our group, for example, suggested that antioxidants taken from food sources were inversely associated with the risk of Barrett’s esophagus, but antioxidants taken as dietary supplements were not (data not shown), a finding also reported in a study of dietary antioxidants and the risk of esophageal adenocarcinoma (38). In addition, complex interactions among known and unknown compounds within and between different dietary patterns are also likely.
Second, although our analysis suggested that GERD symptoms did not influence the association between dietary patterns and the risk of Barrett’s esophagus (through diet simply causing GERD symptoms, and GERD being associated with Barrett’s esophagus), we cannot preclude the possibility that the association between diet and Barrett’s esophagus may be at least partially mediated by GERD. Persons with GERD may be similar to persons with Barrett’s esophagus in many ways, making it more difficult to discriminate individual risk factors between the groups. In addition, there is an imperfect correlation between GERD symptoms and GERD-induced reflux damage, and there is a lack of reflux symptoms in a portion of persons with documented Barrett’s esophagus or esophageal adenocarcinoma (39–43). Finally, the Barrett’s esophagus cases and the GERD controls both had symptoms that led to an endoscopic procedure. Persons who undergo endoscopy may differ from those who do not in many ways, including their dietary choices.
There are numerous strengths of this study. First, this is the first study to our knowledge to investigate the effect of overall dietary patterns on the risk of Barrett’s esophagus, a metaplastic precursor to esophageal adenocarcinoma. Ours complements prior studies of dietary components as risk factors for esophageal adenocarcinoma by determining that diet may be associated with early changes in the carcinogenic pathway (e.g., Barrett’s esophagus). Studies of cancer patients may be biased because of recall bias or behavioral/biologic changes such as those in weight or diet due to the diagnosis or treatments for the cancer itself. By considering only those patients with a new diagnosis of Barrett’s esophagus and by identifying all patients with a new diagnosis within this population, we were able to evaluate whether diet acts early in the carcinogenic sequence, with less opportunity for bias.
Second, study subjects were ascertained from a diverse, broad population base that closely approximates the region’s census demographics; thus, the results can likely be generalized to similar large populations. Third, the availability of two comparison groups enabled us to evaluate the effect of diet in the general population, as well as whether diet is a risk factor for Barrett’s esophagus among persons with GERD.
There are several potential limitations of this analysis. First, case-control studies cannot definitively establish cause and effect (35). Second, observational studies are subject to confounding by other unmeasured variables. Although analyses that evaluated several potential confounders provided little evidence of confounding, we cannot exclude the possibility that some measured or unmeasured factors might have influenced the results. Third, potential random measurement error in dietary data from a food frequency questionnaire is often an issue in nutritional epidemiology, although such measurement error would bias the results toward the null, strengthening the finding of the present study. Lastly, the presence of nonresponders may lead to bias; however, the electronic data suggested that nonresponders were, on average, somewhat healthier than responders, with lower comorbidity scores. This finding would suggest that the effect of nonresponse, if any, may bias the results toward the null (making the population controls more similar to the cases), thereby potentially increasing the strength of our findings.
In summary, using a principal component analysis technique in a community-based population, we found that two dietary patterns may influence the risk of Barrett’s esophagus. Diets rich in fruits, vegetables, and fish were inversely associated with Barrett’s esophagus, whereas Western dietary patterns may be adversely associated with the risk of Barrett’s esophagus. These analyses provide data not available from evaluations of individual nutrients or foods, and, combined with other data suggesting that antioxidant supplementation alone is not associated with the risk of Barrett’s esophagus, they suggest that overall dietary patterns may influence early events in the carcinogenic pathway for esophageal adenocarcinoma. Future studies are needed to evaluate the role of GERD in the etiologic pathway and whether dietary interventions influence the risk of Barrett’s esophagus or esophageal adenocarcinoma among high-risk persons (e.g., persons with GERD).
Acknowledgments
This study was funded by the US National Institutes of Health (grants RO1 DK63616 and K08 DK02697).
Abbreviations
- CI
confidence interval
- GERD
gastroesophageal reflux disease
- OR
odds ratio
Footnotes
Conflict of interest: none declared.
References
- 1.Cameron AJ, Lomboy CT, Pera M, et al. Adenocarcinoma of the esophagogastric junction and Barrett’s esophagus. Gastroenterology. 1995;109:1541–6. doi: 10.1016/0016-5085(95)90642-8. [DOI] [PubMed] [Google Scholar]
- 2.Pera M, Cameron AJ, Trastek VF, et al. Increasing incidence of adenocarcinoma of the esophagus and esophagogastric junction. Gastroenterology. 1993;104:510–13. doi: 10.1016/0016-5085(93)90420-h. [DOI] [PubMed] [Google Scholar]
- 3.Reid BJ, Barrett MT, Galipeau PC, et al. Barrett’s esophagus: ordering the events that lead to cancer. Eur J Cancer Prev. 1996;5(suppl 2):57–65. doi: 10.1097/00008469-199612002-00009. [DOI] [PubMed] [Google Scholar]
- 4.Blot WJ, Devesa SS, Fraumeni JF., Jr Continuing climb in rates of esophageal adenocarcinoma: an update. (Letter) JAMA. 1993;270:1320. [PubMed] [Google Scholar]
- 5.Blot WJ, Devesa SS, Kneller RW, et al. Rising incidence of adenocarcinoma of the esophagus and gastric cardia. JAMA. 1991;265:1287–9. [PubMed] [Google Scholar]
- 6.Blot WJ, McLaughlin JK. The changing epidemiology of esophageal cancer. Semin Oncol. 1999;26:2–8. [PubMed] [Google Scholar]
- 7.Kubo A, Corley DA. Marked multi-ethnic variation of esophageal and gastric cardia carcinomas within the United States. Am J Gastroenterol. 2004;99:582–8. doi: 10.1111/j.1572-0241.2004.04131.x. [DOI] [PubMed] [Google Scholar]
- 8.Kubo A, Corley DA. Marked regional variation in adenocarcinomas of the esophagus and the gastric cardia in the United States. Cancer. 2002;95:2096–102. doi: 10.1002/cncr.10940. [DOI] [PubMed] [Google Scholar]
- 9.Kubo A, Corley DA. Meta-analysis of antioxidant intake and the risk of esophageal and gastric cardia adenocarcinoma. Am J Gastroenterol. 2007;102:2323–30. doi: 10.1111/j.1572-0241.2007.01374.x. [DOI] [PubMed] [Google Scholar]
- 10.Brown LM, Swanson CA, Gridley G, et al. Adenocarcinoma of the esophagus: role of obesity and diet. J Natl Cancer Inst. 1995;87:104–9. doi: 10.1093/jnci/87.2.104. [DOI] [PubMed] [Google Scholar]
- 11.Cheng KK, Sharp L, McKinney PA, et al. A case-control study of oesophageal adenocarcinoma in women: a preventable disease. Br J Cancer. 2000;83:127–32. doi: 10.1054/bjoc.2000.1121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Gonzalez CA, Pera G, Agudo A, et al. Fruit and vegetable intake and the risk of stomach and oesophagus adenocarcinoma in the European Prospective Investigation into Cancer and Nutrition (EPIC-EURGAST) Int J Cancer. 2006;118:2559–66. doi: 10.1002/ijc.21678. [DOI] [PubMed] [Google Scholar]
- 13.Mayne ST, Navarro SA. Diet, obesity and reflux in the etiology of adenocarcinomas of the esophagus and gastric cardia in humans. J Nutr. 2002;132(suppl):3467S–70S. doi: 10.1093/jn/132.11.3467S. [DOI] [PubMed] [Google Scholar]
- 14.Terry P, Lagergren J, Hansen H, et al. Fruit and vegetable consumption in the prevention of oesophageal and cardia cancers. Eur J Cancer Prev. 2001;10:365–9. doi: 10.1097/00008469-200108000-00010. [DOI] [PubMed] [Google Scholar]
- 15.Tzonou A, Lipworth L, Garidou A, et al. Diet and risk of esophageal cancer by histologic type in a low-risk population. Int J Cancer. 1996;68:300–4. doi: 10.1002/(SICI)1097-0215(19961104)68:3<300::AID-IJC6>3.0.CO;2-5. [DOI] [PubMed] [Google Scholar]
- 16.Kabat GC, Ng SK, Wynder EL. Tobacco, alcohol intake, and diet in relation to adenocarcinoma of the esophagus and gastric cardia. Cancer Causes Control. 1993;4:123–32. doi: 10.1007/BF00053153. [DOI] [PubMed] [Google Scholar]
- 17.Zhang ZF, Kurtz RC, Yu GP, et al. Adenocarcinomas of the esophagus and gastric cardia: the role of diet. Nutr Cancer. 1997;27:298–309. doi: 10.1080/01635589709514541. [DOI] [PubMed] [Google Scholar]
- 18.Clarkson PM, Thompson HS. Antioxidants: what role do they play in physical activity and health? Am J Clin Nutr. 2000;72(suppl):637S–46S. doi: 10.1093/ajcn/72.2.637S. [DOI] [PubMed] [Google Scholar]
- 19.National Research Council CoDaH. Implications for reducing chronic disease risk. Washington, DC: National Academy Press; 1989. Diet and health. [PubMed] [Google Scholar]
- 20.Anderson LA, Watson RG, Murphy SJ, et al. Risk factors for Barrett’s oesophagus and oesophageal adenocarcinoma: results from the FINBAR study. World J Gastroenterol. 2007;13:1585–94. doi: 10.3748/wjg.v13.i10.1585. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Hiatt RA, Friedman GD. The frequency of kidney and urinary tract diseases in a defined population. Kidney Int. 1982;22:63–8. doi: 10.1038/ki.1982.133. [DOI] [PubMed] [Google Scholar]
- 22.Krieger N. Overcoming the absence of socioeconomic data in medical records: validation and application of a census-based methodology. Am J Public Health. 1992;82:703–10. doi: 10.2105/ajph.82.5.703. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Sharma P, McQuaid K, Dent J, et al. A critical review of the diagnosis and management of Barrett’s esophagus: the AGA Chicago Workshop. Gastroenterology. 2004;127:310–30. doi: 10.1053/j.gastro.2004.04.010. [DOI] [PubMed] [Google Scholar]
- 24.Block G, Hartman AM. Issues in reproducibility and validity of dietary studies. Am J Clin Nutr. 1989;50:1133–8. doi: 10.1093/ajcn/50.5.1133. discussion 1231–5. [DOI] [PubMed] [Google Scholar]
- 25.Block G, Hartman AM, Dresser CM, et al. A data-based approach to diet questionnaire design and testing. Am J Epidemiol. 1986;124:453–69. doi: 10.1093/oxfordjournals.aje.a114416. [DOI] [PubMed] [Google Scholar]
- 26.Block G, Thompson FE, Hartman AM, et al. Comparison of two dietary questionnaires validated against multiple dietary records collected during a 1-year period. J Am Diet Assoc. 1992;92:686–93. [PubMed] [Google Scholar]
- 27.Block G, Woods M, Potosky A, et al. Validation of a self-administered diet history questionnaire using multiple diet records. J Clin Epidemiol. 1990;43:1327–35. doi: 10.1016/0895-4356(90)90099-b. [DOI] [PubMed] [Google Scholar]
- 28.Sobell J, Block G, Koslowe P, et al. Validation of a retrospective questionnaire assessing diet 10–15 years ago. Am J Epidemiol. 1989;130:173–87. doi: 10.1093/oxfordjournals.aje.a115310. [DOI] [PubMed] [Google Scholar]
- 29.Block G, Norkus E, Hudes M, et al. Which plasma antioxidants are most related to fruit and vegetable consumption? Am J Epidemiol. 2001;154:1113–18. doi: 10.1093/aje/154.12.1113. [DOI] [PubMed] [Google Scholar]
- 30.Coates RJ, Eley JW, Block G, et al. An evaluation of a food frequency questionnaire for assessing dietary intake of specific carotenoids and vitamin E among low-income black women. Am J Epidemiol. 1991;134:658–71. doi: 10.1093/oxfordjournals.aje.a116138. [DOI] [PubMed] [Google Scholar]
- 31.Hatcher L. A step-by-step approach to using SAS for factor analysis and structural equation modeling. Cary, NC: SAS Publishing, SAS Institute, Inc; 1994. [Google Scholar]
- 32.SAS/STAT user’s guide, version 8. Cary, NC: SAS Publishing, SAS Institute, Inc; 1999. [Google Scholar]
- 33.Kleinbaum D, Kupper L, Muller K. Applied regression analysis and other multivariable methods. Boston, MA: PWS-KENT Publishing Company; 1988. Variable reduction and factor analysis. [Google Scholar]
- 34.Kleinbaum D, Kupper L, Morgenstern H. Epidemiologic research: principles and quantitative methods. New York, NY: Van Nostrand Reinhold; 1982. [Google Scholar]
- 35.Rothman KJ, Greenland S, editors. Modern epidemiology. 2. Philadelphia, PA: Lippincott-Raven; 1998. [Google Scholar]
- 36.Zhao Y, Ash A, Ellis R, et al. Predicting pharmacy costs and other medical costs using diagnoses and drug claims. Med Care. 2005;43:34–43. [PubMed] [Google Scholar]
- 37.Zhao Y, Ellis RP, Ash AS, et al. Measuring population health risks using inpatient diagnoses and outpatient pharmacy data. Health Serv Res. 2001;36:180–93. [PMC free article] [PubMed] [Google Scholar]
- 38.Mayne ST, Risch HA, Dubrow R, et al. Nutrient intake and risk of subtypes of esophageal and gastric cancer. Cancer Epidemiol Biomarkers Prev. 2001;10:1055–62. [PubMed] [Google Scholar]
- 39.El-Serag HB, Petersen NJ, Carter J, et al. Gastroesophageal reflux among different racial groups in the United States. Gastroenterology. 2004;126:1692–9. doi: 10.1053/j.gastro.2004.03.077. [DOI] [PubMed] [Google Scholar]
- 40.Farrow DC, Vaughan TL, Sweeney C, et al. Gastroesophageal reflux disease, use of H2 receptor antagonists, and risk of esophageal and gastric cancer. Cancer Causes Control. 2000;11:231–8. doi: 10.1023/a:1008913828105. [DOI] [PubMed] [Google Scholar]
- 41.Gerson LB, Shetler K, Triadafilopoulos G. Prevalence of Barrett’s esophagus in asymptomatic individuals. Gastroenterology. 2002;123:461–7. doi: 10.1053/gast.2002.34748. [DOI] [PubMed] [Google Scholar]
- 42.Lagergren J, Bergstrom R, Lindgren A, et al. Symptomatic gastroesophageal reflux as a risk factor for esophageal adenocarcinoma. N Engl J Med. 1999;340:825–31. doi: 10.1056/NEJM199903183401101. [DOI] [PubMed] [Google Scholar]
- 43.Rex DK, Cummings OW, Shaw M, et al. Screening for Barrett’s esophagus in colonoscopy patients with and without heartburn. Gastroenterology. 2003;125:1670–7. doi: 10.1053/j.gastro.2003.09.030. [DOI] [PubMed] [Google Scholar]