Relationship between Helicobacter pylori infection and obesity in Chinese adults: A systematic review with meta-analysis

Background Obesity is highly prevalent worldwide. More and more studies have been conducted on the relationship between H. pylori infection and obesity or overweight. But the relationship between them is controversial in the literatures and there is no comprehensive evidence for the correlation. Aim To evaluate the prevalence of H. pylori infection in Chinese adult subjects who received routine physical examinations and the relationship between H. pylori and obesity. Methods Literatures on H. pylori infection and obesity in Chinese population were searched in online databases. Relevant data were extracted independently by two researchers and meta-analysis was performed by using Review manager 5.3 software. Results 22 articles were selected with a total sample size of 178033. The pooled prevalence of H. pylori was 42% (95%CI: 37% to 47%) and mean difference of BMI between subjects with and without H. pylori infection was 0.94 (95%CI: -0.04 to 1.91). 9 eligible studies with 27111 subjects were used to calculated pooled OR value because they contained obesity groups. The OR value showed that H. pylori-positive subjects tended to be obese at a risk of 1.20 (95% CI: 1.13 to 1.28). Conclusion In China, obesity has association with H. pylori infection. H. pylori infection may be one of the risk factors for obesity.

Introduction the sample sizes of both the H. pylori-positive group and the H. pylori-negative group, as well as overweight or obesity in each group, or the mean and standard deviation of BMI in each group can be obtained. Studies were excluded if: (1) they were articles such as conferences and reviews; (2) they were about pathological analysis or animal experiments; (3) the samples or contents of them were repeated or very similar to others.

Data abstraction
Two investigators reviewed all literature independently and retrieved studies according to inclusion or exclusion criteria. Any disagreement will be determined by discussion participated by a third investigator. Data extraction was carried out from literatures meeting inclusion criteria, mainly including the following contents: authors, publish year, survey year, survey region and methods to diagnose H. pylori, demographic of subjects, prevalence of H. pylori, number of cases with overweight or obesity or the mean and standard deviation of BMI in population with and without H. pylori infection. The eligibility of relevant studies was evaluated using the cross-sectional study quality evaluation criteria recommended by the Agency for Healthcare Research and Quality (AHRQ) [18]. There are 11 items in the evaluation criteria, including the selection of subjects, quality control and data processing, and the answers are "yes", "no" and "unclear". An item would be scored "0" if it was answered "no" or "unclear"; if it was answered "yes", then the item scored "1". Article quality was assessed as follows: low quality = 0-3; moderate quality = 4-7; high quality = 8-11. The process was independently conducted by two researchers at the same time. In case of disagreement, the dispute shall be discussed and decided by a third investigator.

Statistical analysis
The Review Manager 5.3 software was adopted for meta-analysis of prevalence, mean difference and the odds ratio. The results of continuous data were represented by weighted mean difference (WMD) and 95%CI, and the results of classified data were represented by OR and 95%CI. The heterogeneity between studies was determined by I 2 test. If I 2 > 50%, the random effect model was adopted; if I 2 < 50%, the fixed effect model was adopted. To ensure the stability of meta-analysis results, sensitivity analysis was performed, removing one at a time to compare whether there were significant differences in effect values before and after removal. And the publication bias assessment was conducted on the included literature. If the plot was symmetric, it was considered that there was no publication bias.

Results
The literature selection process is shown in Fig 1. Of 435 potentially relevant articles (2 articles were added through related citations), 289 relevant articles left after removing duplicates, 185 articles were then excluded due to their titles and abstracts not meeting inclusion criteria. Subsequently, 43 articles were excluded due to the following reasons: not observational studies (n = 6), subjects included did not came from China (n = 6), lacking data about H. pylori and BMI (n = 35), subjects not from Check-up crowd (n = 24), subjects selected by designated BMI (n = 4), data or subjects duplicated (n = 7). Finally, twenty-two observational studies (all crosssectional studies) met all inclusion criteria. The characteristics and quality assessment of the included cross-sectional studies are presented in Table 2. Totally 178033 subjects involved in Relationship between Helicobacter pylori infection and obesity in Chinese adults this review and 70761 of them infected with H. pylori. Age of studied subjects ranged from 17 years to 91 years or elder age. Test methods of H. pylori infection used in these studies were urea breath test (n = 15), serum diagnosis method through H. pylori-specific IgG antibody (n = 5), rapid urease test (n = 5), biopsy or histology method (n = 1). Stool antigen test was not used in any study.

The prevalence of H. pylori in different subgroups
All included studies were used in the analyses to assess the prevalence of H. pylori. The prevalence of H. pylori among all subjects was 42% (95%CI: 37% to 47%), as shown in Table 3. Subgroup analysis according to test methods and regions was also performed on 22 studies that provided prevalence on adult subjects. The results showed that the pooled detection rate of urea breath test method was higher than that of other methods (serology, biopsy, rapid urease test), and the detection rate of H. pylori in first-tier cities (Beijing, Shanghai) was higher than that in other cities. Stratified analysis was also carried out for subset of those 22 studies which contained stratified information of gender, age and body type. The prevalence of H. pylori in elderly population was higher than it in the middle-aged population and was lowest in the young population. Obese population had highest H. pylori infection rate of 51% (95CI: 42% to 59%) in three body shape groups. After grouping analysis, there was still high heterogeneity, and the data were processed by random effect model. Due to the limited information provided in included literatures, more subgroup analysis were failed to be conducted.

Estimated differences in BMI between subjects with and without H. pylori infection
As shown in Table 4, main anthropometric and biochemical characteristics per H. pylori group at baseline were provided in parts of included studies. For original data which presented as median, interquartile, sample mean and standard deviation were estimated from Luo [41], Wan [42]. Mean differences of the physiological and biochemical indexes were estimated, concluded in Table 5. The average HDL-C of the Hp-positive group was lower than that of the H. pylori negative group, but the BMI, age, SBP, DBP, TG, TC and LDL-C of the H. pylori positive group were all higher than that of the negative group. There was still great heterogeneity in the study corresponding to each index.

Meta-analysis of the impact of H. pylori on obesity
Nine references were included for this meta-analysis, because only in these studies, we could get case number of patients with H. pylori in both obesity and normal weight groups. There was no statistical heterogeneity being found (I 2 = 0.0%), so we selected a fixed-effect model for this analysis (Fig 2). The population size, demographic data, and the H. pylori infection rate for this subset can be queried by their name in    diastolic blood pressure, triglycerides, total cholesterol and serum LDL cholesterol of the H. pylori-positive group were all enhanced, and the average HDL cholesterol decreased, which testify to a worse energy metabolism. The precise mechanism underlying these findings is not well established. There is a number of potential factors which may be involved. (1) Ghrelin and leptin, gastrointestinal hormones, are both involved in metabolic control and energy balance. Ghrelin is produced in the stomach and can stimulate food intake. Leptin has an opposite effect. It was mainly synthesized in adipose tissue [45] and also produced by P cells of the gastric epithelium [46]. Studies reported lower serum leptin levels [47] and lower serum Ghrelin levels in H. pylori-positive patients [47,48,49]. Leptin can inhibit eating, and its reduction may be involved in excessive eating and obesity. While the decrease of plasma Ghrelin concentration represented a physiological adaptation to the positive energy balance associated with obesity [50]. (2) Insulin resistance is an important risk factor in lots of common metabolic disorders [51,52]. H. pylori infection was found to have a potential role in promoting insulin resistance as observed in a Japanese study [53]. So, people with H. pylori infection may be more likely to get obesity. (3) Obesity may interact with H. pylori infection. Recently, growing evidence has implicated the intestinal immune system as an important contributor to metabolic disease including obesity [54]. It has been reported that the ability of monocytes to convert into macrophages was decreased in morbid obesity patients [55], which indicate that immune environment of obese people is more powerful for H. pylori survival. On the other hand, pre-adipocytes could develop phagocytic activity toward microorganisms as macrophage-like cells until they stop proliferating and differentiate into adipocytes [56], which suggests that H. pylori infection may stimulate the growth of adipose tissue to participate in the immune process.
Our result supports a positive relationship between H. pylori infection and obesity. But, studies included in this review are all the cross-sectional studies so this study cannot establish a causal relationship between them. In addition, this study has several limitations. First, studies included are unable to distinguish the H. pylori genotypes. H. pylori type I strains which expressing cagAand vacA are more virulent than type II strains and may bring more effect on metabolism. Studies have shown that the variation of the 3 0 Region of the cagA Gene in H. pylori is closely related to the pathological changes and clinical outcomes caused by the infection of the strain [57]. But how these gene differences affect obesity remains unclear. Second, subjects in our study were people who took health examination from urban area, so we may underestimate the prevalence of H. pylori in China. More researches on H. pylori infection and obesity in rural areas are needed. Last, obesity is a chronic disease that is also affected by heredity and lifestyle. There are existing more than one major gene influencing BMI in Chinese sample [58]. In Chinese population, Genetic variation in the FTO gene is strongly associated with obesity and BMI, and its effect size on BMI is comparable with that in the European population [59]. NOC Gene, which is one of circadian clock genes, is also associated with obesity and BMI [60]. Besides, the AC3 genetic polymorphisms are associated with obesity in adults but not in children [61]. So, further studies are needed to control the confounders and to verify or strengthen the association. It would be important to note that the evaluation of the H. pylori infection rate in several subgroups and the calculation of pooled values were not determined by all studies but some subsets. Because only these subsets contains the data needed for the calculation. Notice that there was significant heterogeneity between studies when we compared the mean difference of biochemical characteristics, but hardly any heterogeneity when evaluated the risk of obesity for H. pylori-positive subjects, we thought this was due to the intrinsic properties of the object being analyzed. For the former's heterogeneities, we considered difference in survey regions, H. pylori test methods, socioeconomic status of subjects, and other factors as sources of them.
H. pylori, a pathogenic bacteria of gastrointestinal diseases, has relationship with many extra-intestinal diseases. However, H. pylori is not on the adverse effects of all diseases. It presents a protective factor in the onset of certain diseases, which may inspire new diagnosis and therapeutic methods for obesity and other diseases.