Association of dietary patterns and endoscopic gastric mucosal atrophy in an adult Chinese population

Atrophy gastritis harbor a high risk for the development of dysplasia and gastric cancer. The study investigated the relationships of specific dietary patterns and endoscopic gastric mucosal atrophy. In this cross-sectional study, we enrolled 574 consecutive outpatients who were diagnosed as chronic gastritis according to endoscopic examination. Dietary intakes of study individuals was assessed using the semi-quantitative food group frequency questionnaire. Logistic regression analyses were used to evaluate the relationship between dietary patterns and endoscopic gastric mucosal atrophy adjusted for potential confounders. A total of 574 participants were included, 286 with endoscopic gastric mucosal atrophy. Three dietary patterns were identified by factor analysis. “Alcohol and fish” (tertile 1 vs. tertile 3: adjusted odds ratio = 1.85, 95% confidence interval: 1.06–3.22) and “coarse cereals” (tertile 1 vs. tertile 3: adjusted odds ratio = 2.05, 95% confidence interval: 1.24–3.39) were associated with an increased risk for endoscopic gastric mucosal atrophy but a “traditional” pattern was not. Dietary pattern was not associated with gastric mucosal atrophy in women or in participants with H. pylori infection. A high adherence to both “Alcohol and Fish” and “Coarse cereals” dietary patterns seem to be associated with higher odds of endoscopic gastric mucosal atrophy in men and in patients without H. pylori infection. Further prospective cohort studies needed to confirm these findings.

Gastric mucosal atrophy is a common, age-related multifactorial disease associated with Helicobacter pylori infection and diet [1][2][3] . It is estimated that as many as 25% of people worldwide are at risk of loss or atrophy of normal gastric mucosal glands 4 . The prevalence of chronic atrophy gastritis varies regionally, from 25.8% in China 5 to 40.7% in Korea 6 . Gastric mucosal atrophy is a precursor lesion in the development of gastric carcinoma 7,8 . The overall incidence of gastric cancer has been declining in Asian countries including China 9 , but most people infected by H. pylori remain asymptomatic for life despite evidence of chronic gastritis 10 . H. pylori infection alone might not be sufficient to cause gastric cancer even though the World Health Organization considers it to be a class-I carcinogen. Diet may also play a key role in gastric carcinogenesis. For example, some studies have found that a high-salt diet was independently associated with gastric carcinoma and that Western diets rich in salt, meat, animal fat, starchy food and alcohol increase the risk of precancerous lesions or gastric cancer [11][12][13][14][15][16][17] . Diets rich in fruits and vegetables appear to reduce gastric cancer risk 18,19 . Completion of the sequence of events that occur in the gastric mucosa before the manifestation of gastric cancer may take decades 8 . Most studies of the role of diet have focused on gastric cancer, but increased awareness of the association of dietary factors with development of precancerous lesions would help in the formulation of targeted prevention.
Rather than focusing on individual nutrients or foods, dietary pattern analysis might be the most feasible method to assess the relationship between diet and disease and to help nutritionists to provide dietary recommendations in public health practice 20 . The characterization of diet patterns by factor analysis has been proven to predict the risk of coronary heart disease 21 , colorectal cancer 22 , and gallbladder disease 23 . To our knowledge, the relationship of dietary factors and endoscopic gastric mucosal atrophy has not been well established. The study aim was to identify the role of dietary patterns and risk factors associated with gastritis and of potential help in the prevention of gastric carcinoma.

Methods
Study population. This cross-sectional study investigated the relationship of diet and endoscopic gastric mucosal atrophy in a Han Chinese population in Binzhou city Shandong Province, China. The participants were recruited between April and October 2018 at the Outpatient Digestive Endoscopy Center at the Binzhou Medical University Affiliated Hospital, which is the only tertiary hospital in the region. Patients ≥18 years of age with a endoscopic diagnosis of chronic gastritis were eligible for inclusion. Patients with a history of gastrointestinal surgery, malignancy, or severe systemic of neurological disease, a history of gastritis or H. pylori infection, pregnancy or at lactation, or unable to communicate or walk normally were excluded. A total of 574 consecutive outpatients who met the above eligibility were included in the analyses (Fig. 1). The study was approved by the ethics committee of Binzhou Medical University Hospital and conducted in line with the ethical guidelines of Declaration of Helsinki. Informed consent was obtained from all participants.
Assessment of dietary intake. Dietary intake during the previous year was assessed with a semiquantitative food group frequency questionnaire (FGFQ) administered by medical staff and study investigators during face-to-face interviews. The questionnaire included items on the frequency ("three times a day", "twice a day", "once a day", "two to three times a week", "once a week", "once every two weeks", "once a month", or "never") and portion size in grams of 23 food group items. The daily intake in grams was calculated by multiplying the portion size of each food group item by the daily consumption frequency (3, 2, 1, 0.357, 0.143, 0.071, 0.033, or 0). The patients were asked to recall consumption of food groups rather than nutrients during the previous year because of the short 20-minute interval between administration of the spray oral anesthetic and the endoscopy procedure. Commonly consumed food groups and recipes were chosen by a nutritionist (Supplementary Table 1  and Stata version 15.1 (Stata Corporation, College Station, TX, USA). All p-values were two-sided. Dietary patterns were analyzed by principle component analysis (PCA) with varimax rotation. The number of patterns was determined by the cumulative contribution rate of variance, eigenvalue >1.45, and the interpretability of the results. Food groups with a rotated factor loading of >|0.4| were considered the primary components of each pattern. Dietary pattern scores were calculated by summing the products of the standardized intakes of each food group multiplied by regression coefficients and were reported as tertiles. The values of continuous variables were reported as means and standard deviations. Categorical variables were reported as numbers and percentages. Between-group differences in participants with or without endoscopic gastric mucosal atrophy were compared by Student's t-tests. Chi-square tests were used to compare percentage differences of categorical variables in participants with or without endoscopic gastric mucosal atrophy. Differences in patient characteristics across dietary pattern scores were evaluated by chi-square tests for categorical variables and the analysis of variance for continuous variables. The association of adherence to each dietary pattern and the odds of endoscopic gastric mucosal atrophy was tested by binary logistic regression. Model 1 was adjusted for age and sex. Model 2 included BMI, education (<6, 6-12, or >12 years), residence (urban or rural), marital status (single, divorced, widowed, or married), occupation (supervisory, company employee, laborer, self-employed, student, farmer, or retried) income (<1000, 1000-3000, >3000 CNY/month/individual), smoking (yes or no), drinking (yes or no). Model 3 was further adjusted for H. pylori infection (yes or no), history of diabetes (yes or no), history of hypertension (yes or no), history of anticoagulant use (yes or no), and total energy intake. Assuming that total energy intake was missing in randomly distributed participants, missing data were compensated by multivariate imputation of chained equations 26,27 . Twenty imputed datasets were created to reduce sampling variability and the estimates were combined based on Rubin's rules 28 .

Results
The characteristics of the 574 participants are shown in Table 1. The point prevalence of endoscopic gastric mucosal atrophy was 49.8%. Participants with or without endoscopic gastric mucosal atrophy showed significant differences in terms of marital status (p = 0.044), occupation (p = 0.001), H. pylori infection status (p < 0.001), history of hypertension (p = 0.001), history of taking anticoagulants (p = 0.015). Compared with those without endoscopic gastric mucosal atrophy, participants with gastric mucosal atrophy showed older age (p < 0.001), lower education (p < 0.001), and higher income (p = 0.012). extraction of dietary patterns. The PCA analysis yielded three independent diet patterns that explained 24.5% of the variation in food consumption ( Table 2). The first was "alcohol and fish, " which included high consumption of wine, beer, fresh water and sea fish. The second was "traditional, " with high consumption of vegetables, wheat products and red meat, and the third was "coarse cereals, " with high consumption of whole grain cereals, legumes, and poultry. The participant characteristics and dietary pattern tertiles are shown in Table 3. Participants with an "alcohol and fish" pattern were more likely to be men, smokers, drinkers, supervisory or company employees, urban residents with high incomes and education levels, and histories of hypertension or anticoagulant use. Those adhering to a "traditional" pattern were more likely to be men, younger, smokers, drinkers, have high incomes, low BMI, and low education level. Those adhering to the "coarse cereals" pattern were more likely to be men, younger, supervisory employees, and urban residents with high education and income levels.
After controlling for potential confounders, participants who adhered to the "alcohol and fish" pattern were more likely to be drinkers, non-farmers, without a history of hypertension. Those who adhered to the "traditional" pattern were more likely to be men, younger than 50 years of age, and drinkers. Those who adhered to the "coarse cereals" pattern were more likely to be smokers, have an H. pylori infection, and have a high education level (Supplementary Table 3). Table 4 shows the effect of adherence to the dietary patterns on the risk of endoscopic gastric mucosal atrophy. Model 1 found that the "alcohol and fish" and "coarse cereals" patterns were both positively associated with endoscopic gastric mucosal atrophy but that the "traditional" pattern was not. After controlling for potential confounders, adherence to the "alcohol and fish" dietary pattern was related to an increased prevalence of endoscopic gastric mucosal atrophy, but the association was significant only for the highest tertile (tertile 1 vs. tertile 3 adjusted OR = 1.84, 95% CI: 1.06-3.20, p-value for trend = 0.040). Adherence to the "coarse cereals" pattern also increased the risk of endoscopic gastric mucosal atrophy (tertile 1 vs. tertile 3: adjusted OR = 2.04, 95% CI: 1.24-3.38, p-value for trend = 0.005). Adherence to the "traditional" pattern was not found to influence the risk of endoscopic gastric mucosal atrophy. ( www.nature.com/scientificreports www.nature.com/scientificreports/ The association of dietary pattern and endoscopic gastric mucosal atrophy after stratifying the participants by selected variables is shown in Supplementary Table 4. After controlling for potential confounders, The "coarse cereals" pattern was associated with endoscopic gastric mucosal atrophy in men (tertile 1 vs. tertile 3: adjusted OR = 3.05, 95% CI: 1.45-6.38, p-value for trend = 0.003). The "alcohol and fish" pattern was associated with H. pylori infection status (adjusted OR = 2.12 (95% CI: 1.08-4.15, p-value for trend = 0.031). The "coarse cereals" pattern was associated with endoscopic gastric mucosal atrophy in patients without H. pylori infection (adjusted OR = 2.35 (95% CI: 1.28-4.32, p-value for trend = 0.005). The results of sensitivity analysis with or without adjusting for total energy intake were similar (data not shown).

Discussion
After adjusting for potential confounders, participants with a dietary pattern including high alcohol and fish intake had increased odds of endoscopic gastric mucosal atrophy. The "coarse cereals" pattern also increased the risk of endoscopic gastric mucosal atrophy. Vegetable, wheat, and red meat consumption were not associated with the presence of endoscopic gastric mucosal atrophy. Analysis after stratification revealed that the "alcohol and fish" pattern was associated with an increased risk for endoscopic gastric mucosal atrophy in men but not in women, which might reflect the inclusion of only five female drinkers. The "alcohol and fish" and "coarse cereals" patterns were associated with endoscopic gastric mucosal atrophy in patients without H. pylori infection. Although the most common cause of gastritis is H. pylori infection, H. pylori-negative gastritis was common and more likely in past alcohol drinkers than in patients with H. pylori infection 29 .
To the best of our knowledge, this is the first study to investigate the relationship of dietary patterns and chronic atrophic gastritis. Most previous studies investigated the association of dietary patterns and gastric cancer. A meta-analysis of 23 studies reported that diets rich in vegetables, fruit, fish, low-fat milk, and whole grains may decrease the risk of stomach cancer. Western-style diets rich in meats, refined grains, sweets, high-fat dairy products, high-fat gravies, and alcohol may increase the risk of stomach cancer 30 . Another recent meta-analysis of cohort studies found an inverse relationship between the adherence the Mediterranean Diet and gastric cancer risk 31 .
The results of previous studies of the association of diet and chronic atrophic gastritis were limited to consumption of single foods or alcohol consumption. You et al. 32 revealed that alcohol was a risk factor (OR = 3.2), but Gao et al. 33 found that life-time moderate beer consumption (≤51.38 g of alcohol/day) was inversely related to chronic atrophic gastritis (OR = 0.73). The majority of endoscopic and histological data do not support a relationship of alcohol consumption with chronic atrophic gastritis [34][35][36] . These cross-sectional studies did not find an independent association of alcohol drinking and chronic atrophic gastritis, but small sample sizes make it difficult to confirm the presence or absence of a relationship. A recent retrospective cohort study of 10,185 subjects did find that alcohol consumption was an independent risk factor of mucosal atrophy (adjusted HR = 1.001, 95% CI: 1.001-1.002) 37 . Both acute and chronic alcohol consumption can alter gastric acid secretion and induce acute gastric mucosal injury by mediating the release of inflammatory factors, granulocyte activation, protease secretion,  www.nature.com/scientificreports www.nature.com/scientificreports/ production of reactive oxygen species, vasoconstriction, and increased vascular permeability 38,39 . A meta-analysis of cross-sectional studies found that moderate alcohol intake had a dose-response association with a reduction in H. pylori infection, but after adjusting for H. pylori infection status, alcohol drinking was significantly associated with gastritis in this study 40 .
There is little evidence of a direct relationship of fish consumption and chronic atrophic gastritis. A study by Tanigawa et al. 41 , reported that fish consumption promoted H. pylori-induced gastritis in a Mongolian gerbil model. More data are available on the effect of fish consumption on H. pylori infection. Tongtawee et al. 42 reported that H. pylori infections were more frequent in those who ate pickled fish than in those who did not. Ikezaki et al. 43 reported that increased fish consumption was negatively associated with the success of oral treatment to eradication of H. pylori infection. The incorporation in cell membranes of highly unsaturated n-3 fatty acids from fish could increase the risk of H. pylori gastritis because of disruption of the clustering of lipid rafts and alteration of immune responses 44,45 . Yoshinori et al. 46 found that the serum concentration of docosahexaenoic acid, an n-3 polyunsaturated fatty acid, increased the risk of chronic atrophic gastritis (OR = 2.2).
A high intake of legumes and cereal grains other than wheat and rice, has long been recognized as a poor man's dietary pattern 47 . Coarse cereals are recommended for their anti-inflammatory, anti-atherogenic, and antioxidant activities and for reducing the risk of type 2 diabetes, cardiovascular diseases, and cancer, but the effect   www.nature.com/scientificreports www.nature.com/scientificreports/ of adherence to a "coarse cereals" diet on chronic gastritis has not been addressed [47][48][49] . The unfavorable effects of coarse cereals on chronic gastritis may be related to its relatively limited their nutritive value. For example some study participants routinely ate millet porridge for breakfast. The pericarp of millet is unpalatable and may contain antinutrients such as phytates and tannins 47,50 . Another possible explanation is that cereals other than yellow corn 51 , corn do not contain vitamin B 12 , vitamin C, or vitamin A, which could lead to gastritis-induced vitamin deficiency 52 and gastric mucosal atrophy 53 . Tsigane et al. 3 reported that frequent intake of soybean products was associated with an increased risk of atrophic gastritis.
This study has several strengths. First, to our knowledge, it is the first study to evaluate the association between dietary patterns and endoscopic gastric mucosal atrophy among adults in the city of Binzhou, Shandong Province, China. It provides evidence into the potential role of dietary modification in the prevention of gastric mucosal atrophy. Second, face-to-face interview ensured that the data we collected are accurate. Furthermore, we have adjusted for potential known confounders in our analyses. Nevertheless, several limitations also need to be considered. The analysis was controlled for many potential confounders, but family history of gastric cancer, physical activity, and mental health status were not considered and could have affected the results. Secondly, some participants may have had pre-existing gastric mucosal atrophy. Consequently, changing their usual dietary intake might have influenced our results. Thirdly, the cross-sectional design did not allow determining the causal relationship of dietary patterns and endoscopic gastric mucosal atrophy. Fourthly, recall bias and misclassification of study participants was not avoidable because the FGFQ had not been previously validated. Finally, because the subjects were recruited in Binzhou, China, the study findings might not be generalizable.
In conclusion, adherence to "alcohol and fish" or "coarse cereals" dietary patterns was associated with increased odds of endoscopic gastric mucosal atrophy. The results support confirmation in large prospective cohort studies.