Helicobacter pylori Infection Is Associated with an Increased Risk of Hyperemesis Gravidarum: A Meta-Analysis

Background. Several studies have shown a possible involvement of Helicobacter pylori (H. pylori) infection in individuals with hyperemesis gravidarum (HG), but the relationship remains controversial. This meta-analysis was performed to validate and strengthen the association between HG and H. pylori infection. Methods. PubMed, Embase, and Web of Science databases up to March 20, 2014, were searched to select studies on the prevalence of H. pylori infection between pregnant women with HG and the normal pregnant control subjects. Results. Of the HG cases, 1289 (69.6%) were H. pylori-positive; however, 1045 (46.2%) were H. pylori-positive in control group. Compared to the non-HG normal pregnant controls, infection rate of H. pylori was significantly higher in pregnant women with HG (OR = 3.34, 95% CI: 2.32–4.81, P < 0.001). Subgroup analysis indicated that H. pylori infection was a risk factor of HG in Asia, Africa, and Oceania, especially in Africa (OR = 12.38, 95% CI: 7.12–21.54, P < 0.001). Conclusions. H. pylori should be considered one of the risk factors of HG, especially in the developing countries. H. pylori eradication could be considered to relieve the symptoms of HG in some intractable cases.


Introduction
Hyperemesis gravidarum (HG), the most severe form of pregnancy-associated nausea and vomiting, is leading to weight loss, nutritional deficiencies, and metabolic disturbance, for example, dehydration, acidosis from starvation, hypokalemia, and transient hepatic dysfunction, often requiring hospitalization and medical treatment to avoid lifethreatening complications [1]. HG complicates 0.3-2% of all pregnancies and is a multifactorial disease; however, little is known about the etiology of HG. But a variety of mechanisms may play a role in this disease, such as endocrine factors like human chorionic gonadotropin (HCG), estradiol, and progesterone and immunologic factors, as well as personal factors in which increased body weight has been proposed as possible underlying cause [2]. Recently, several studies have put emphasis on the correlation between Helicobacter pylori (H. pylori) infection and the risk of HG [3][4][5][6].
H. pylori, as a gram-negative flagellated spiral bacterium, colonizes stomach and creates the basis of pathogenesis of gastric pathologies, including chronic gastritis, duodenal and gastric ulcers, gastric adenocarcinoma, and mucosaassociated lymphoid tissue lymphoma [7][8][9]. H. pylori infected the stomach of 50% of the world population, and it is more prevalent in developing countries [10]. The prevalence of H. pylori infection in pregnant women varies according to socioeconomic conditions, geographic area, and even the method used to test H. pylori infection.
Several studies have shown a possible involvement of H. pylori infection in individuals with HG; however, other studies did not share with them. Although three previous meta-analyses have shown the positive association between H. pylori and HG, the role of H. pylori infection in the pathogenesis of HG has not reached a consensus. Moreover, the previous meta-analyses did not use the comprehensive search method, unable to include the overall studies, and did not perform a detailed analysis on subgroup to explore the potential factors in HG. On the other hand, the role of some factors such as different populations, geographic areas, ethnicity, and low socioeconomic status is unclear in HG. Therefore, our meta-analysis was undertaken to strengthen the hypothesis that H. pylori is a risk factor of HG and to describe the underlying factors in HG.

Data Sources.
We systematically identified studies in PubMed, Embase, and Web of Science (inception through March 20, 2014) databases by two independent investigators (LL and LLL) for all relevant literatures between the risk of HG and the infection of H. pylori, by using the MeSH terms, "Helicobacter pylori" OR "H. pylori" OR "Helicobacter infection" AND "hyperemesis gravidarum. " These were searched both as thesaurus terms and as text words. Even "nausea and vomiting" as text word was searched too. Finally, a Google Scholar search was done. Moreover, the references at the end of selected papers were also examined for additional relevant studies.

Inclusion Criteria.
In our meta-analysis, the included articles had to meet the following criteria: (1) they must be case-control studies or prospective studies or cross-sectional studies; (2) studies must refer to the association of H. pylori infection with HG risk and must provide raw data on H. pylori infection in both HG and control groups; (3) the participants must have had a clinical diagnosis of hyperemesis gravidarum as follows: pernicious vomiting, weight loss, and at least one positive ketonuria; (4) the confirmation of H. pylori infection was detected by enzyme-linked immunosorbent assay (ELISA), stool antigen test, 13 C-urea breath test ( 13 C-UBT), mucosal biopsy, and polymerase chain reaction (PCR). At least one positive result was considered as confirmation of infection.

Exclusion Criteria.
We excluded studies as follows: (1) reports without control groups; (2) reviews and duplicated publications; (3) studies in which the source of H. pylori infection in cases and control subjects and other essential information were not offered; hyperthyroidism, multiple gestation trophoblastic neoplasia, gastrointestinal and hepatic disorders, urinary tract or other infections, psychosocial or any other maternal disorders, and any treatment with antacids or antibiotics within the previous 7 days were excluded as well.

Data Extraction.
According to the inclusion criteria, data was carefully extracted independently by two reviewers (LL and LLL). For each study analyzed, we collected data including first author, year of publication, country of the population studied, study design, sources of pregnant women (primipara or multipara), gestational age, diagnosis of hyperemesis gravidarum, methods of H. pylori detection, total number of persons in cases and controls, and the numbers of H. pyloripositive and H. pylori-negative patients in the HG group and the control group of each study which were recorded, respectively. For conflicting evaluations, an agreement was reached by consensus and agreement with another reviewer (XYZ), referring back to the original articles.

Statistical Analysis.
We used pooled odds ratio (OR) with its corresponding 95% confidence interval (CI) to estimate the strength of the association between HG and H. pylori infection. Post hoc subgroup analyses were also performed to explain the heterogeneity in results. Subgroups were explored as follows: detection of H. pylori infection (serum H. pylori IgG/IgM/IgA antibody by ELISA, stool antigen test, mucosal biopsy from endoscopy, or H. pylori genome by PCR), publication period (1996-2000, 2001-2005, 2006-2010, or 2011-2014), and region (Asia, North America, Europe, Africa, or Oceania).
The heterogeneity of the studies included in this metaanalysis was assessed using the statistic test and the 2 statistic test. The random-effects model was selected when value < 0.1 or 2 > 50%; otherwise, the fixed-effects model was selected. Possible publication bias was evaluated by visual inspection of funnel plots and application of Begg's and Egger's test [11][12][13][14][15]. values of less than 0.05 from Egger's test were considered statistically significant.
All statistical analyses were done with STATA statistical software package version 12.0 (2000; STATA Corp., College Station, TX, USA); < 0.05 was identified as statistically significant. Figure 1, after rigorous searching, we identified 104 citations. Of these, fifty-six irrelevant papers were excluded after screening the titles. In the remaining 48 articles, 16 studies included five studies without control group, two did not provide sufficient data, and 9 reviews or meta-analyses were discarded. Thus, a total of thirty-two studies included twenty-nine case-control studies and three cross-sectional studies published between 1998 and 2014 fulfilled our inclusion criteria and were included in the meta-analysis [3][4][5][6].

Characteristics of Included Studies.
With respect to the H. pylori detection methods, serum H. pylori IgG/IgM/IgA antibody was detected by ELISA in twenty-nine, one, and one studies, respectively. However, H. pylori stool antigen (HpSA) was used in seven articles, H. pylori genome by PCR (Hp PCR DNA) tested in two studies, and biopsy and histological examination from endoscopy in one literature.  traits), three from Europe (Norway and Greece were grouped in Europe according to similarities in geographic position and racial traits), three from Africa, and one from Oceania. Apparently, all studies used hospital-based controls. All of the studies followed a standard definition of HG excluding subjects with differential diagnosis such as infections, gastrointestinal and endocrine diseases, or psychiatric illness.

3.3.
Overall. Thirty-two articles published between 1998 and 2014 met the inclusion criteria and their characteristics were shown in Tables 1 and 2  were H. pylori-positive in control group. Since overall 2 was 81.5%, we used the random-effects model for our analysis. As shown in Figure 2, the overall OR was 3.34 (95% CI: 2.32-4.81) and the overall effect value was 6.51 ( < 0.001), which indicated that there was a powerful association between H. pylori infection and risk of HG.

Subgroup Analysis.
To evaluate the underlying confounding factors that may have impacted the overall results, we further conducted subgroup analyses based on detection of H. pylori infection, publication period, and region, respectively. Since 2 were 0 in subgroups of Hp PCR DNA test and Africa, we selected the fixed-effects model. In the remaining subgroups, 2 were >50%, so the random-effects models were used. As shown in Table 3

Bias Diagnostics.
Begg's test was created for assessment of possible publication bias ( Figure 3). The values for Egger's tests were = 0.067 ( > 0.05), indicating the absence of heterogeneity and implying that the results of the present meta-analysis were relatively stable and that the publication bias might exert little influence on the overall results.

Main Findings.
This updated meta-analysis suggests that exposure to H. pylori is associated with an increased risk of HG. The studies included in this meta-analysis, containing 1851 HG patients of which 1289 cases were confirmed with H. pylori infection, implied that the rate of H. pylori infection was much greater in HG patients (1289/1851) than that in non-HG patients (1045/2262) after adjusting for confounding variables ( < 0.001).
Our meta-analysis contained 32 articles including 29 case-control studies and 3 cross-sectional articles, which contained comprehensive articles and added new primary studies. We enrolled much more patients with HG (1851) and controls (2262) than those in published studies to further confirm this relationship. We separated subgroups in detection of H. pylori infection, publication period, and region to comprehensively evaluate the underlying confounding factors that may have impacted the overall results. As reported before, three previous articles reported similar meta-analysis results [44][45][46]. Golberg et al. [44] performed a meta-analysis of 14 studies and did subgroup analysis only  Overall (I 2 = 81.5%, P = 0.000) Note: weights are from random-effects analysis 10      and a specificity of 55% in the diagnosis of HG as compared to controls without HG. Our study paid attention not only to the detection method of H. pylori IgG, but also to the methods of Hp PCR DNA and HpSA. We also described other detection methods used in pregnant women. With respect to detection methods used in pregnant women, Hp PCR DNA test seemed much more efficient than ELISA test of serum H. pylori IgG antibody, and the latter was more reliable than HpSA test. Of note, Cevrioglu et al. [30] collected both serum and feces samples from pregnant women with HG to investigate specific antibodies for H. pylori (immunoglobulin-IgG, IgA) and HpSA, and only HpSA test suggested significant association between H. pylori infection and HG while serologic assessment failed to reflect the association (40% versus 12.4%, < 0.001). By using PCR with specimen of saliva, Güngören et al. [42] found a positive relationship between the symptoms of HG and H. pylori positivity, while test of H. pylori IgG/IgM antibody failed to detect this association between the symptoms of HG and H. pylori positivity. From biopsies of the gastric antrum and corpus, Bagis et al. [23] found that, compared to controls, HG patients were detected with higher H. pylori density, degree of inflammation, and H. pylori activation, implying that H. pylori density might be related to HG since the bacterium density of controls was lower. These results suggested that the degree of gastric complaints might be related to density of H. pylori.
In addition, two studies from New Zealand indicate that incidence of hyperemesis gravidarum differs according to women's ethnic origin [47,48]. Just as shown in our subgroup analysis of region, studies in North America and Europe suggested no association between H. pylori infection and risk of HG, while in Asia, Africa, and Oceania analysis indicated H. pylori infection was a risk factor of HG, especially in Africa. Sandven et al. [39] also found this association between H. pylori infection and HG was much stronger in Africans as compared to non-Africans. This might be explained by ethnicity and low socioeconomic status. H. pylori-infected Africans possibly carry an aggressive variant of the bacterium and the host immune mechanisms might be a key to different responses to H. pylori in different populations and geographic areas. As well documented, the prevalence rate of H. pylori infection is much higher in developing countries than that in developed countries [49][50][51]. As Eshraghian reviewed, the overall prevalence of H. pylori infection in Iran and other Eastern Mediterranean Regional Office countries such as Egypt and Afghanistan, irrespective of time and age group, ranged from 30.6% to 82% and ranged from 22% to 87.6%, respectively. However the H. pylori prevalence in North Africa was 76% [52]. The prevalence is high in developing countries, while pregnant women with HG in these countries have higher rate of H. pylori infection. For example, it is 50%-70% in Turkey [53], more than 80% in Egypt [54]. The above studies all suggested that H. pylori infection was a risk factor of HG. Strategies to improve sanitary facilities, educational status, and socioeconomic status should be implemented to minimize H. pylori infection and come into the result of decline prevalence of HG.
Our meta-analysis suggested that H. pylori infection was a risk factor of HG. Frigo et al. suggested that the H. pylori may contribute to its persistence beyond the normal time course [16]. It was once reported that HG was an oxidative stress condition induced by increasing reactive oxygen species (ROS) activity and decreasing antioxidant status [55]. Meanwhile, H. pylori colonizes gastric mucosa and generates ROS as well as downregulating levels of plasma antioxidants such as ascorbic acid [7], similar to Güney et al. [37] who found that, compared to the control, level of serum malondialdehyde (MDA) was significantly higher and activities of antioxidant enzymes such as superoxide dismutase (SOD), catalase (CT), and glutathione peroxidase (GSH-Px) were significantly lower in the HG group ( < 0.01). Therefore, they hypothesized that increased ROS activity or decreased antioxidant potential, possibly induced by H. pylori, might have a pathogenic role in HG. To date, however, the knowledge of how H. pylori causes HG is still very limited, and we assumed the following. Firstly, hormonal mechanisms, in the early phase of pregnancy, as a result of the elevated steroid and human chorionic gonadotropin (HCG) levels, accumulation of fluid, and a displacement of intracellular and extracellular volume occur which in turn leads to a shift in pH in the gastrointestinal tract during pregnancy [40,56]. Secondly, emotional factors, the moods of pregnant women change frequently due to the changes of endocrine hormones that might increase women's susceptibility to infection caused by altered cell-mediated immunity that causes changes of various classes of antibodies during different gestational periods [57,58]. Thirdly, H. pylori infection might be one potential reason for HG. Dysmotility of gastrointestinal tract and prolonged gastric emptying and intestinal transit time induced by pregnancy might favor infectionof H. pylori [56,58]. On the other side, host inflammation response to varies of virulence of H. pylori strains also different from each other. The virulence of the organism might be another factor creating a possible link between H. pylori and the precipitation of HG. As we all know, cytotoxin-associated gene A product (CagA) and vacuolating cytotoxin A (VacA) are used as markers for genomic diversity of H. pylori. In Western countries VacA, rather than CagA, was associated with more severe diseases, while in East Asian countries it is the opposite [59][60][61]. Similarly, as shown in On the other hand, eradication of H. pylori could relieve the symptoms of HG. Strachan et al. reported a case of eradication in a 38-year-old woman in her third pregnancy, who orally omeprazole 20 mg bid, metronidazole 400 mg bid, and amoxicillin 500 mg tid for 7 days in her 30 weeks of pregnancy. This led to prompt resolution of her vomiting and improvement of her reflux symptoms [62]. El Younis et al. also reported two cases in which first-trimester patients with severe HG required intravenous fluid replacement.
Rapid improvement of the HG was observed with complete resolution of all symptomatology after using of erythromycin therapy orally, which possibly suggests a new therapeutic modality for similar patients [63]. On the other hand, omeprazole is not licensed for use in pregnancy and ought to be used with caution although experience from case-control and observational studies has not revealed any increase in congenital malformations or pregnancy complications [64]. Large-scale studies were not performed. However, the above case reports could verify that the eradication of H. pylori might relieve the symptoms of HG.

4.2.
Interpretation. This meta-analysis has some limitations to be acknowledged. Firstly, our meta-analysis only focused on papers published in the English language and might miss some eligible studies that were unpublished in other languages. Secondly, the articles identified were limited to those openly published up to March 2014, and it is possible that some related published or unpublished studies that might meet the inclusion criteria were missed. Finally, despite using a precise literature searching strategy to identify eligible studies, it is possible that a few studies meeting the inclusion criteria were not included, resulting in any inevitable bias, though the funnel plots and Egger's tests failed to show any significant publication bias.

Conclusion
In conclusion, our meta-analysis suggested that there was a strong association between H. pylori infection and HG, allowing us to conclude that H. pylori should, therefore, be considered as one of the risk factors of HG. Screening for H. pylori should be added to the investigations for HG, especially in the developing countries. Appropriate therapeutic regimens for eradication of H. pylori could be considered to relieve the symptoms of HG in some intractable cases. and Guoxin Zhang contributed materials/analysis tools; Lin Li and Guoxin Zhang contributed to the writing of the manuscript; Guoxin Zhang proofread the paper.