Relationship of Helicobacter pylori Infection with Nonalcoholic Fatty Liver Disease: A Meta-Analysis

Background and Aims Helicobacter pylori (H. pylori) and nonalcoholic fatty liver disease (NAFLD) have become increasingly recognized, both of which affect human health globally. The association of H. pylori infection with NAFLD remains unclear. Methods PubMed, EMBASE, and Cochrane Library databases were searched. Only a random-effects model was used. Odds ratios (ORs) and risk ratios (RRs) with 95% confidence intervals (CIs) were calculated for the combined estimates of raw data. Adjusted ORs (aORs) and hazard ratios (aHRs) with 95% CIs were calculated for the combined estimates of data adjusted for confounders. Results Thirty-four studies with 218573 participants were included. Based on unadjusted data from 26 cross-sectional studies and 3 case-control studies, H. pylori infection was significantly associated with the presence of NAFLD (OR = 1.26, 95% CI = 1.17–1.36, P < 0.001). Based on adjusted data from 15 cross-sectional studies and 1 case-control study, H. pylori infection was significantly associated with the presence of NAFLD (aOR = 1.25, 95% CI = 1.08–1.44, P < 0.001). Compared with control subjects without NAFLD, patients with moderate (OR = 1.67, 95% CI = 1.17–2.39, P = 0.005) and severe (OR = 1.71, 95% CI = 1.30–2.24, P < 0.001) NAFLD, but not those with mild NAFLD (OR = 1.14, 95% CI = 0.9–1.45, P = 0.286), had significantly higher proportions of H. pylori infection. The association of H. pylori infection with the occurrence of NAFLD was statistically significant based on adjusted data from 3 cohort studies (aHR = 1.18, 95% CI = 1.05–1.34, P = 0.007), but not based on unadjusted data from 3 cohort studies (RR = 1.41, 95% CI = 0.80–2.48, P = 0.237). Conclusion H. pylori infection is associated with NAFLD, especially moderate and severe NAFLD. The impact of H. pylori eradication on the prevention of NAFLD should be further explored.


Introduction
Helicobacter pylori (H. pylori) infects about half of the world's population, especially people living in developing countries and poor socioeconomic countries [1][2][3][4]. Considering such a high infection rate, it is recognized as a major public health problem worldwide [2]. H. pylori is a main pathogenic factor for chronic gastritis, peptic ulcers, gastric cancer, and gastric mucosa-associated lymphoid tissue (MALT) lymphoma [5]. H. pylori infection may also disturb a series of biological processes and determine or infuence the development and severity of various extragastric diseases [6], such as insulin resistance, metabolic syndrome, diabetes, nonalcoholic fatty liver disease (NAFLD), vitamin B 12 defciency, cardiovascular, neurological, dermatological, and ophthalmic diseases [7]. NAFLD is considered a major hepatic manifestation of metabolic syndrome [8] and includes a full spectrum of fatty liver disease from simple hepatic steatosis or nonalcoholic fatty liver (NAFL) to nonalcoholic steatohepatitis (NASH) and cirrhosis [9]. NAFLD has become the most common type of chronic liver disease, with a global prevalence of approximately 25% [10]. Recently, a new nomenclature of metabolic dysfunction-associated fatty liver disease (MAFLD) has been proposed to replace NAFLD with updated diagnostic criteria and recognized by worldwide experts [11,12]. Regardless, there is now growing evidence that the development of NAFLD is associated with gut microbiota imbalance [13]. Some studies suggested an association between H. pylori infection and NAFLD, and the presence of H. pylori or Helicobacter species has been observed in liver specimens from patients with various liver diseases [14][15][16]. However, others indicated no correlation between them [17][18][19]. Considering the importance of understanding potential risk factors for NAFLD on its management, we have conducted an updated meta-analysis of studies published to date to explore the association between H. pylori infection with NAFLD.

Registration.
Tis study was registered on the PROS-PERO with a registration number CRD42021247307.

Literature Search.
Te relevant publications were searched via PubMed, Cochrane library, and EMBASE databases. Te search terms were as follows: ("HP" or "H. pylori" or "Helicobacter pylori" or "Helicobacter infection" or "Helicobacter") and ("Nonalcoholic fatty liver disease" or "Fatty liver" or "Nonalcoholic fatty liver" or "Nonalcoholic steatohepatitis" or "NAFLD" or "NASH" or "NAFL"). Tere was no language restriction. Te last search was conducted on July 14, 2022.

Eligibility Criteria.
Inclusion criteria were as follows: (1) eligible studies should include patients who were diagnosed with NAFLD and detected the H. pylori infection, (2) eligible studies should clearly report the diagnostic methods of H. pylori infection and NAFLD, (3) eligible studies should provide the number of patients with positive and negative H. pylori infection in NAFLD patients and control subjects without NAFLD or report the odds ratios (ORs) or hazard ratios (HRs) with 95% confdence intervals (CIs) to evaluate the association between H. pylori infection and NAFLD, and (4) age >18 years old. If multiple publications were available for the same study, only the publication with the most complete data would be included.

Data Extraction.
Te following data were extracted from each study: frst author, publication year, study country, study design, publication form (abstract or full text), number of positive and negative H. pylori infection in the NAFLD patients and control subjects without NAFLD, and diagnostic methods of H. pylori infection and NAFLD. Adjusted ORs (aORs) and adjusted HRs (aHRs) with 95% CIs with confounders adjusted were extracted from the studies where multivariate regression analyses were performed to evaluate the association of H. pylori infection with NAFLD. If studies had multiple adjustment models, only the models that refected the greatest degree of adjustment for confounders and its corresponding aOR and aHR would be further considered in our meta-analysis.

Study Quality
Assessment. Te quality of cohort and case-control studies was assessed by the Newcastle-Ottawa Scale (NOS), a widely used tool for assessing the quality of observational/nonrandomized studies. It has three major domains: (1) selection, (2) comparability, and (3) exposure/ outcome. Te maximum score is 9. A score of 0-3, 4-6, and 7-9 represents low, moderate, and high quality, respectively. Te quality of cross-sectional studies was evaluated by the Agency for Healthcare Research and Quality (AHRQ) with an 11-item checklist. An item would be scored "0," if its answer was "NO" or "UNCLEAR"; and an item would be scored "1," if its answer was "YES." Te maximum score is 11. A score of 0-3, 4-7, and 8-11 represents low, moderate, and high quality, respectively.

Statistical Analyses.
All statistical analyses were performed using the Stata software version 12.0 (Stata Corp, College Station, USA) and Review Manager software version 5.4 (Cochrane collaboration, the Nordic Cochrane Centre, Copenhagen, Denmark). Only a random-efects model was employed. ORs and RRs with 95% CIs were calculated for the combined estimation of raw data, and aORs and aHRs with 95% CIs were calculated for the combined estimates of data adjusted for confounders. Te I 2 statistics and Cochran Q test were used to evaluate the heterogeneity, and P < 0.1 and/or I 2 > 50% were considered to indicate statistically signifcant heterogeneity. Subgroup and meta-regression analyses were performed to explore the sources of heterogeneity among the studies with and without adjustment for confounders. Tey were grouped according to the study design, region, study quality, diagnostic methods of H. pylori infection and NAFLD, sample size, adjustment for confounders, and publication form. Te interaction between subgroups was tested. Leave-one-out sensitivity analyses were assessed by sequentially omitting one study each time. Publication bias was evaluated by Egger test. P < 0.1 was considered as a statistically signifcant publication bias. In addition, the proportion of H. pylori infection was compared according to the severity of NAFLD (i.e., mild, moderate, and severe).

Study Quality.
Among the cohort and case-control studies, 6 and 1 were of high and moderate quality, respectively (Supplementary Table 1). Among the crosssectional studies, 16 and 11 were of high and moderate quality, respectively (Supplementary Table 2).

H. pylori Infection and Presence of NAFLD.
Based on the unadjusted data from 26 cross-sectional studies and 3 casecontrol studies, the meta-analysis showed that H. pylori infection was signifcantly associated with the presence of NAFLD (OR � 1.26, 95% CI � 1.17-1.36, and P < 0.001) ( Figure 2). Heterogeneity was statistically signifcant (I 2 � 88.7% nd P < 0.001). Such a statistically signifcant association between them disappeared in the subgroup analyses of studies using the rapid urease test and fecal antigen test to detect H. pylori infection, but remained in others. Te interaction between subgroups was statistically signifcant in the subgroup analyses according to the study design (P < 0.001) and diagnostic methods of NAFLD (P < 0.001), but not in others. Subgroup analyses did not identify any source of heterogeneity (Table 2). Metaregression analyses showed that the study design (P < 0.001) and diagnostic methods of NAFLD (P < 0.001) might be the sources of heterogeneity (Supplementary Table 3). Sensitivity analyses did not identify any source of heterogeneity (Supplementary Table 4). Egger test did not show any signifcant publication bias (P � 0.294).
Based on the adjusted data from 15 cross-sectional studies and 1 case-control study, the meta-analysis showed that H. pylori infection was signifcantly associated with the presence of NAFLD (aOR � 1.25, 95% CI � 1.08-1.44, and P < 0.001). Heterogeneity was statistically signifcant (I 2 � 90% and P < 0.001) (Figure 3). Such a statistically signifcant association between them disappeared in the subgroup analyses of non-Asian studies, those using the rapid urease test to detect H. pylori infection, those using surrogate markers for diagnosis of NAFLD, those with a sample size of >5000, and those published as abstracts but remained in others. Te interaction between subgroups was statistically signifcant in the subgroup analyses according to the study design (P < 0.001), study quality (P < 0.001), diagnostic methods of H. pylori (P � 0.01), and diagnostic methods of NAFLD (P < 0.001), but not in others. Subgroup analyses did not identify any source of heterogeneity (Table 3). Meta-regression analyses showed that the study design (P � 0.015) and diagnostic methods of NAFLD (P � 0.023) might be the sources of heterogeneity (Supplementary Table 5). Sensitivity analyses did not identify any source of heterogeneity (Supplementary  Table 6). Egger test did not show any signifcant publication bias (P � 0.591).

H. pylori Infection and Severity of NAFLD.
Te association between H. pylori infection and severity of NAFLD was explored in 4 studies (Table 4).
Te meta-analysis showed no statistically signifcant diference in the proportion of H. pylori infection between patients with mild NAFLD and those without NAFLD (OR � 1.14, 95% CI � 0.9-1.45, and P � 0.286). Heterogeneity was statistically signifcant (I 2 � 95.1% and P < 0.001) (Supplementary Figure 1). Because only a small number of studies was included, subgroup analyses were not performed to explore the sources of heterogeneity.
Te meta-analysis showed that the proportion of H. pylori infection was signifcantly higher in patients with moderate NAFLD than those without NAFLD (OR � 1.67, 95% CI � 1.17-2.39, and P � 0.005). Heterogeneity was statistically signifcant (I 2 � 92.7% and P < 0.001) (Supplementary Figure 2). Because only a small number of studies was included, subgroup analyses were not performed to explore the sources of heterogeneity.

H. pylori Infection and Occurrence of NAFLD.
Based on the unadjusted data from 3 cohort studies, the meta-analysis showed that H. pylori infection was not signifcantly associated with the occurrence of NAFLD (RR � 1.41, 95% CI � 0.80-2.48, and P � 0.237). Heterogeneity was statistically signifcant (I 2 � 98.3% and P < 0.001) (Supplementary Figure 4). Because only a small number of studies was included, subgroup analyses were not performed to explore the sources of heterogeneity. Based on the adjusted data from 3 cohort studies, the meta-analysis showed that H. pylori infection was associated with the occurrence of NAFLD (aHR � 1.18, 95% CI � 1.05-1.34, and P � 0.007). Heterogeneity was statistically signifcant (I 2 � 70.8% and P � 0.032) (Supplementary Figure 5). Because only a small number of studies was included, subgroup analyses were not performed to explore the sources of heterogeneity.

Discussion
Based on the data from cross-sectional studies and casecontrol studies, H. pylori infection was associated with the presence of NAFLD, especially moderate and severe NAFLD. Based on the data from cohort studies, H. pylori infection increased the risk of NAFLD occurrence after adjustment for confounders. Notably, seven previous metaanalyses [51][52][53][54][55][56][57] also concluded a signifcant association between H. pylori infection with NAFLD. Our current metaanalysis has several advantages compared to previous ones [51][52][53][54][55][56][57]. First, there was a more comprehensive collection of eligible studies by expanding the search strategy and updating the fnal search date. Tus, the number of studies included was larger in the current meta-analysis than in the previous ones. Second, some of the previous meta-analyses did not strictly follow the prespecifed inclusion and exclusion criteria to collect all relevant studies. For example, in both meta-analyses by Zhou et al. [51] and Heydari et al. [56], a cross-sectional study by Sumida et al. [40] would have been included based on their eligibility criteria, but neither of them included this study. Tird, some of the previous meta-analyses only calculated ORs to evaluate their association. By comparison, the current meta-analysis further pooled HRs to evaluate their cause-efect association. Fourth, the interaction between subgroups was tested to infer whether the impact of H. pylori infection on NAFLD was signifcantly infuenced by some confounding factors, which have not been performed in previous meta-analyses yet. Last, the association of H. pylori infection with the severity of NAFLD was evaluated in the current meta-       Canadian Journal of Gastroenterology and Hepatology analysis, which has not been performed in previous metaanalyses yet. Metabolic syndrome, including overweight/obesity, type 2 diabetes mellitus (T2DM), and metabolic disorders, is an important pathogenic factor of NAFLD/MAFLD [58,59]. It is also closely associated with H. pylori infection [60]. Notably, insulin resistance (IR) is a key factor in the development of metabolic syndrome [61]. Tus, the pathophysiological interrelationship between H. pylori infection and MAFLD/NAFLD may be explained by IR [62][63][64] (Supplementary Figure 6). First, H. pylori infection can stimulate the release of proinfammatory cytokines, such as tumor necrosis factor-α (TNF-α) [65][66][67]. TNF-α induces serine/threonine-mediated phosphorylation of insulin receptor substrate 1 (IRS-1), which attenuates IRS-1-mediated insulin signaling, leading to the occurrence of IR [68]. Second, H. pylori infection causes a decrease in adiponectin levels [48,69]. Adiponectin can reduce gluconeogenesis and  Canadian Journal of Gastroenterology and Hepatology lipogenesis in the liver, and therefore has the efect of an insulin sensitizer to inhibit intrahepatic lipid accumulation [70]. Tus, decreased adiponectin levels would result in increased intrahepatic fat content and IR [71][72][73]. Tird, there is an interaction of reciprocal inhibition between adiponectin and TNF-α in terms of their production and action, thereby enhancing IR [74]. Fourth, H. pylori infection leads to elevated fetuin-A levels [75]. Fetuin-A can stimulate adipocytes and macrophages to produce proinfammatory cytokines and then induce IR [76,77]. Besides, their association may be attributed to altered gut microbiota [13]. Chronic H. pylori infection causes significant changes in the gut microbiota composition [78]. Gut microbiota can release endotoxin composed of the outer wall of Gram-negative bacteria, which can introduce into the liver directly through the portal vein. Endotoxin can stimulate infammatory response via Toll-like receptor 4 (TLR4), thereby exacerbating hepatic infammation [79]. Indeed, some studies have shown that lipopolysaccharide, a surrogate marker of endotoxin, is elevated in patients with NAFLD [80,81].
Our previous study found a higher rate of H. pylori infection in young military personnel than in civilians [82]. Tis phenomenon is probably explained by the fact that increased mental stress caused by high-intensity military training suppresses the body's humoral and cellular immunity, thereby increasing the risk of H. pylori infection. On the other hand, high occupational and personal stress are independent predictors of NAFLD development [83]. Terefore, the association of H. pylori infection with the presence of NAFLD may be because both of them have a concomitant predisposing factor (i.e., stress).
Another previous meta-analysis by our group also showed a signifcant association between H. pylori infection and irritable bowel syndrome (IBS) [84]. It should be noted that multiple etiological factors, including obesity, gut microbiota, dietary factors, and immune-mediated causes [85], overlap between IBS and NAFLD. Tus, such factors should not be neglected to explain our current fndings about the association of H. pylori infection with NAFLD.
Current consensus recommends that dietary modifcation, exercise, and weight loss as the major treatment option for NAFLD to reduce liver fat and improve IR [86,87]. Besides, considering that silymarin has antioxidant, antiinfammatory, immunomodulatory, antifbrotic, and hepatoprotective activities and stimulates protein synthesis and liver tissue regeneration [88], silymarin may be used for the treatment of NAFLD. Notably, it seems that silymarin can also inhibit H. pylori activity [89][90][91][92]. Terefore, it may be hypothesized that silymarin can be helpful for the treatment of NAFLD and H. pylori infection.
Considering an association of H. pylori infection with NAFLD, it appears that H. pylori eradication is benefcial in preventing NAFLD. However, this is still controversial. A randomized controlled study by Maharshi et al. showed that successful eradication of H. pylori in patients with NAFLD resulted in signifcant improvement in IR [93]. However, a study by Jamali et al. found that H. pylori eradication may not infuence liver fat content, liver function tests, lipid profle, IR, and anthropometric measurements in patients with dyspeptic NAFLD [94]. Terefore, whether H. pylori eradication infuences the occurrence or progression of NAFLD needs to be confrmed by more studies in the future.
Consensus and practice guidelines recommend bismuth quadruple therapy, which consists of a proton pump inhibitor, a bismuth, and two antibiotics, for a duration of 10-14 days as the frst-line treatment of H. pylori. Commonly used antibiotics include amoxicillin, clarithromycin, and metronidazole [95][96][97][98]. However, considering an increased burden of multidrug resistant Gram-negative infections [99], some studies have suggested that a combination of tetracycline and tinidazole should achieve a higher rate of H. pylori eradication [100][101][102].
Our meta-analysis has some limitations. First, a majority of these included studies provided only cross-sectional data, which can only establish a possible association between H. pylori infection and NAFLD, but not any cause-efect association. Second, only some of these included studies adjusted the confounders in multivariate regression analyses, and the confounders adjusted were inconsistent among them. Tird, only a minority of these included studies evaluated the prevalence of H. pylori infection according to the severity of NAFLD. Finally, none of these included studies have evaluated the association of H. pylori infection with MAFLD.

Conclusion
Tere seems to be an association between H. pylori infection and NAFLD, but this association was weak. More prospective cohort studies are needed in the future to demonstrate the impact of H. pylori infection and its eradication on MAFLD, and experimental studies should also be necessary to elucidate their potential mechanisms. Undoubtedly, these studies may provide promising approaches for the management of MAFLD.

Data Availability
Data sharing is not applicable to this article as no new data were created in this study.

Conflicts of Interest
Te authors declare that they have no conficts of interest regarding the publication of this paper.

Authors' Contributions
Guangqin Xu reviewed and searched the literature, extracted and collated the data, discussed the fndings, and drafted the manuscript. Hongyu Li discussed the fndings and gave critical comments. Shaoze Ma searched the literature, extracted the data, and gave critical comments. Liyan Dong discussed the fndings and gave critical comments. Nahum Mendez-Sanchez discussed the fndings and gave critical comments. Xingshun Qi conceived the work, reviewed the literature, interpreted the fndings, and revised the manuscript. All authors have made an intellectual contribution to the manuscript and approved the submission.

Acknowledgments
Tis work was supported by Prof. Hongyu Li who received a grant from the National Key Research and Development Program of China (2019YFC1315901).

Supplementary Materials
Te paper includes supplementary tables 1-6 as supplementary materials. Teir descriptions are as follows: Supplementary  Table 4: results of leave-one-out sensitivity analysis in studies unadjusted for confounders. Supplementary Table 5: results of meta-regression analyses regarding the association of H. pylori infection with NAFLD in studies adjusted for confounders. Supplementary Table 6: results of leave-one-out sensitivity analysis in studies adjusted for confounders. Supplementary