Helicobacter pylori infection and increased diabetes prevalence were the risks of colorectal adenoma: a systematic review and meta-analysis

Background Helicobacter pylori (H. Pylori) infection and hyperglycemia may be associated with an increased risk of colorectal neoplasm. However these two factors affect colorectal neoplasm remain controversial. We aimed to carry out a meta-analysis to evaluate the study population diabetes prevalence rate and H. pylori infection rate with colorectal adenoma risk. We conducted a systemic research through English databases for medical reports. We also recorded the diabetes prevalence and H. pylori infection prevalence in each study. We classified these studies into 4 subgroups as their background population diabetes prevalence < between 6 to 8%(Group between 8 to 10 %(Group 3) and more than 10%(Group The random effects model had used to calculate pooled prevalence estimates with 95% confidence interval CI.

(including Australia, China, Indonesia, Japan, Taiwan, Korea and Vietnam) has a DM prevalence of approximately 8.1% [15]. However, as socioeconomic growth and industrialization are rapidly occurring in this area, the increasing prevalence of diabetes was also noted in these countries [16]. DM is also considered a risk factor for colon adenoma and carcinoma. Several studies have shown that subjects with DM had an OR of 1.45 for colon adenoma and a relative risk of 1.38 for colorectal adenocarcinoma [17,18].
Our previous study demonstrated that combined hyperglycemia and H. pylori infection was involved in colon adenoma formation and had a synergistic effect [19]. The risk of colorectal adenoma might decrease after H. pylori is successfully eradicated [20]. This means that hyperglycemia and H. pylori infection might interact and affect colorectal adenoma formation.
Therefore, we carried out a systematic review and meta-analysis of published studies to evaluate the association between H. pylori infection and colorectal adenoma formation.
Furthermore, we surveyed the DM prevalence in each published study as a subgroup analysis and tried to determine the relationship between H. pylori infection and DM prevalence in the risk of colon adenoma formation. We tried to use different DM prevalence rates to classify subgroups of these studies to answer Plummer's question and discuss the relationship between the DM prevalence rate and H. pylori infection involvement in colorectal adenoma formation.

Data sources and searches
PRISMA statement guidelines were followed for conducting and reporting meta-analysis data. The PICOS scheme was followed for reporting inclusion criteria. A systematic search was conducted using PubMed/MEDLINE, EMBASE, and the Cochrane Library for medical reports published until the end of October 2017 without language or date restrictions. The following search terms were used: ''Helicobacter pylori'' AND (''colorectal'' OR ''colonic'' OR "colon'' OR ''large intestine'') AND (''neoplasms'' OR "polyp'' OR ''adenoma'' OR ''cancer''). Abstracts of articles from the literature search were individually evaluated independently for possible inclusion by the three authors (Chen MJ, Wu MS and Yang HW).
For all databases, the last search was run on 22 October 2017.

Study selection
The eligible studies enrolled in the meta-analysis satisfied the following criteria: (i) investigated only the gastrin level or the recurrence of colorectal neoplasia, or (iii) included a pediatric population.

Data extraction
Three reviewers (Chen CC, Liu CJ, Shih SC) independently extracted data by using a predefined form, and disagreements were resolved by discussion and consensus.
Information was collected in each selected study as possible, concerning first author, publication year, research type, publication type, the number of subjects, the country of origin, the prevalence of DM rate of origin, the matching variables, the location of neoplasia, the H. pylori detection method, the sample characteristics (age and sex), the reported OR for colorectal adenoma with 95% CIs and the covariates adjusted for in the analysis.

Statistical analysis
Three reviewers (Lin YC, Yu LY, Kuo YC) independently assessed the quality of all eligible studies using the ROBINS I scale. The ORs were collected for analysis. Heterogeneity between studies was assessed using both the χ2 test with a p-value < 0.10 and inconsistency index (I2) with a cut-off of 50%. Pooled effects with 95% CIs were derived using a fixed effect model unless significant heterogeneity was present, in which case a random effects model was applied. When a study provided different OR estimates based on hospital-based controls and population-based controls, the latter estimates were selected for the combined analysis. We also calculated and presented a summary of the adjusted ORs for the studies that provided adjusted results. Publication bias was evaluated with funnel plot and Egger's test.
We examined the roles of several potential sources of heterogeneity using restricted maximum likelihood-based random-effects meta-regression analysis and subgroup analyses according to geographic location, prevalence of diabetes (DM) rate, prevalence of H. pylori infection rate, study design, sample size for the cases, subsite of the neoplasia, H. pylori detection method and sex. The estimated DM prevalence rate of each study was referenced from the International Diabetes Federation (IDF) Atlas and local country government data or study results (by Wang HY, Huang CT). We used the mean year of study duration as the DM prevalence rate. According to the diabetes prevalence rate of the study population, we classified the studies to 4 groups: group 1: DM prevalence < 6%; group 2: DM prevalence from 6 ~ 8%, group 3: DM prevalence from 8 ~10% and

Results
After initial screening, the full text of 281 potentially eligible articles was retrieved for detailed assessment, and 27 studies were eligible for analysis [3,6,8,9,19,. All of the eligible studies presented OR for colorectal adenoma related to H. pylori infection. [3,6,8,9,19, Nineteen of them are cross-sectional studies [3, 6, 8, 19, 21-24, 25- Most studies were carried out in Asia (7 in Korea, 3 in Taiwan A flow chart describing the process of study selection is shown in Figure 1. For the metaanalysis, data were extracted from twenty-seven studies with a total of 68,410 patients, and the pooled OR for colorectal adenoma related to H. pylori infection was 1.51 (95% CI 1.39-1.63), although there was significant heterogeneity (p < 0.001) ( Table 1 and Figure   2). Because of geographic and time differences in DM prevalence rate [43-52], subgroup analysis for the studies was performed by classifying them into 4 groups: group 1: DM prevalence < 6%; group 2: DM prevalence from 6 ~ 8%, group 3: DM prevalence from 8~10% and group 4: DM prevalence > 10% (Table 1 and Figure 2).
In the subgroup analysis, the background population DM prevalence < 6% studies had three studies with a total of 886 patients. Meta-analysis of these studies revealed that the H. pylori-infected patients have an increased risk of having colorectal adenoma with a pooled OR of 1.24 (95% CI 0.86-1.78) but no significant difference. There was no significant heterogeneity in group 1 studies (p = 0.279). Since the study background population DM prevalence increased more than 6%, the H. pylori-infected patients have significant increased risk of having colorectal adenoma. This upturn was more significant when the background population DM prevalence was approximately 6% to 8%. After metaanalysis of these studies, the H. pylori-infected people had a higher risk of colorectal adenoma than noninfected people. The pooled OR was 2.16 (95% CI 1.61-2.91), and a total of 11655 patients were included. In these studies, Lin's study noted the DM prevalence in his study participants, and males had an 8.1% prevalence of DM, and females had a 6.1% prevalence of DM. Due to Lin's study population, 6.94% of patients had DM and were thus classified into group 2 for subgroup analysis. There was significant heterogeneity in the group 2 studies (p < 0.01) In groups 3 and 4, similar results were observed. Compared to that of group 2 studies, the pooled OR of groups 3 and 4 was mildly decreased but still had a significantly elevated risk of colorectal adenoma in H. pylori-infected patients. For the H. pylori-infected patients, the risk of colorectal adenoma in group 3 was 1.40 (95% CI 1.24-1.57), and the group 4 OR was 1.52 (95% CI 1.46-1.57). The group 3 studies included 28,431 patients, and the heterogeneity across studies was marginal (I 2 = 58.6%, p = 0.013). There were 27,438 participants included in group 4 studies, and there was no significant heterogeneity (I 2 = 0.0%, p = 0.704). We also checked other variables, including gender, detection methods of H. pylori infection and study designs, and there was no evidence of statistical relevance to OR for adenoma with H. pylori infection.

Test of Heterogeneity and Publication Bias
Heterogeneity was assessed by chi-square and I-square for the included studies. The value of χ2 was < 0.01, and the I 2 was 61.9%. Due to the heterogeneity noted in our analysis, a random effects model was used. The shape of the funnel plots for studies on the association between H. pylori infection and the risk of colorectal adenoma appeared asymmetrical ( Figure 3A), which indicated that studies with positive correlation are reported more often. The p-value for Egger's linear regression method (p < 0.01) suggested that there was statistical evidence of publication bias. (Figure 3  H. pylori infection is the main cause of chronic gastritis and peptic ulcer disease and is the main carcinogen for gastric malignancy disease [56]. In the extragastric system, H. pylori infection has also been reported to be related to inflammatory bowel disease, colorectal neoplasms, and cardiovascular disease; to contribute to insulin resistance, associated metabolic syndrome, diabetes, autoimmune disease, and kidney disease; and to be associated with neurodegenerative disease, respiratory disease and hematologic disease [57,58]. In these studies, most scientists addressed the topic of H. pylori infection related to colorectal neoplasm formation [3,6,8,9,19,. Now that it is known that H. pylori is an infectious disease, it can be cured with a course of antibiotics. In recent studies, H. pylori infection rate decreased due to public health improvement in developed countries and some developing countries. In many countries, the incidence of H. pylori infection has been decreasing in association with improved standards of living and levels of hygiene. However, as the prevalence rate of H. pylori gastrin might act as mitogen. Third, H. pylori was found to be associated with metabolic diseases that associated with CRC risk. [59] It also hints to a link between H. pylori infection and colorectal adenoma formation, which may be due to the third factor in this connection.
Unlike H. pylori infection disease, the prevalence of DM in the global world has persisted.
The age-standardized DM prevalence increased from 4.3% in 1980 to 9.0% in 2014 in men and from 5.0% to 7.9% in women [60]. The rise in prevalence might be due to population growth and aging, as the number of adults with diabetes has increased nearly 4-fold over the past 35 years. The prevalence and number of adults with diabetes both increased and doubled in men and increased by 60% in women worldwide, shifting from an excess prevalence in women in 1980 to a higher male prevalence in 2014 [61]. Persistent high blood sugar concentrations lead to damage to the blood vessels and peripheral nerves.
This situation might result in an increased risk of cardiovascular diseases, such as heart attack and stroke, kidney disease, diabetic retinopathy and foot amputations [62]. DM is also considered an increased risk factor for colon adenoma and carcinoma [63]. These DMrelated complications lead to higher costs for the health care system [64] as well as lower quality of life and reduced life expectancy [65].
A past study showed that DM prevalence was low in much of Asia and sub-Saharan Africa in the 1980s and 1990s [57,66]. However, several recent reports have demonstrated that the DM prevalence has increased in China, India, Turkey and Saudi Arabia [67][68][69][70]. Some high-income English-speaking countries, such as the USA and the UK, [71,72] also reported increased DM prevalence. On the other hand, DM prevalence did not increase in Western Europe, and some reports from Sweden, Germany and Switzerland showed similar results [73,74]. In accordance with a previous statement and based on geographical distribution, the prevalence of H. pylori infection and DM seems to be inversely related. In our study, when the background population DM prevalence was below 6%, H. pylori infection did not significantly increase the risk of colorectal adenoma formation (Group 1). This association became significant when the study population DM prevalence was over 6% (Group 2-4). The odds ratio of H. pylori infection-related colorectal formation was 2.16 (95% CI 1.61-2.91); it was the highest when the population DM prevalence was from 6~8% (Group 2). However, when the DM prevalence rate was elevated to 8~10% or more than 10% (Group 3, Group 4), the OR was mildly decreased to 1.40 (95% CI 1.24-1.57) and 1.52 (95% CI 1. 46-1.57). H. pylori infection still significantly increased colorectal adenoma but was not distinct from Group 2 (DM prevalence between 6 and 8%). However, when DM prevalence increased, H. pylori infection might increase the risk of colorectal adenoma formation.
The cause of this condition may be related to the study period and location. Most studies in group 2 were carried out from 1996 to 2000, and group 3 studies were carried out from 2000 to 2010. The study's location may also affect this result. Most of the Group 3 studies were from Korea, and Group 4 studies were from the USA. The Group 2 studies were more heterogeneous in location, including Japan, Taiwan, Turkey, Israel and Germany. The H.
pylori infection rate of the study population was also collected and is shown in Table 1.
There was no significant difference in the ORs of H. pylori infection-associated colorectal adenoma between the lower infection rate area (USA or Germany, 35.3% to 35.6%) and the middle infection rate area (Japan, Korea, China, Taiwan; approximately 51.7% to 55.8%). Further evaluation for the other reason that Group 2 studies revealed higher ORs in H. pylori infection-associated colorectal adenoma is necessary. In addition, our study might partially answer Plummer's query "How to explain that some areas had a high prevalence of H. pylori infection but low CRC risk?" [10]  Our previous study demonstrated the interaction of hyperglycemia and H. pylori infection in colon adenoma formation. We found that the OR for adenoma was 1.437 (95% CI 1.197-1.726) if H. pylori was present or 1.629 (95% CI 1.239-2.14) if HbA 1c ≥ 6.5. When combining these two factors, the OR was elevated to 4.712 (95% CI 3.189-6.963), suggesting that these two factors may have a synergistic effect in colorectal adenoma [19]. The likely reason for the synergistic effect may involve several processes. These animal models and individualized study results support this study finding that states that an elevated DM prevalence rate enhances the risk of colorectal adenoma in H.
pylori-infected populations. Our study still had several limitations. First, in our subgroup analysis, we found that the Group 2 studies had more heterogeneity. The reason for this condition may be related to population studies with differences in study location, population, aging, and gender. Despite this heterogeneity, H. pylori infection also increased the risk of colorectal adenoma. Second, the asymmetry of the results of Egger's test and funnel plots suggested the possibility of publication bias. However, because most of our included studies [3,6,8,9,19, had statistically significant results, Egger's test and funnel plots would show asymmetry. Third, the population DM prevalence rate might not completely represent our included studies participant's diabetes condition. Only Lin et al. [24] and Hu et al. [19] included the DM prevalence rate in their studies. To the best of our knowledge, we tried to estimate the diabetes rate of each study as accurately as possible.

Conclusion
In conclusion, our study demonstrated that population DM prevalence affects the risk of colorectal adenoma with H. pylori infection. Since diabetes prevalence was over 6% in the background study population, H. pylori infection became a more significant factor in inducing colorectal adenoma formation. Given the increasing prevalence of diabetes in the world, H. pylori eradication and hyperglycemia control might have an impact on the prevention of colorectal neoplasm formation.