Introduction

Duodenal ulcer (DU) is one type of peptic ulcer in duodenum, which may result from infection of Helicobacter pylori (H. pylori) bacteria, overuse of alcohol and medications (aspirin and non-steroidal anti-inflammatory drugs) [1, 2]. Among these etiological factors, H. pylori infection was the most common one [3], which could lead to increased gastric acid, degradation of mucus barrier and eventual ulceration. However, it was reported that >20 % of patients with H. pylori infection would develop DU [4].

The reason why the H. pylori-induced pathologies are different is unclear. The vacuolating cytotoxin A (vacA), one of major virulence factors, has been reported to be involved. The vacA protein was encoded by the polymorphic H. pylori vacA gene. It is a secreted cytotoxin being capable to form vacuoles in gastric epithelial cells [5]. The vacA gene is polymorphic, distributed in three principal regions: the signal (s), intermediate (i) and middle (m) regions. Each type can be further divided into two main subtypes, numbered 1 and 2 [6, 7]. The two subtypes of the s-region (s1 and s2) and m-region (m1 and m2) were mostly studied. Different s or m genotypes of H. pylori varied in the vacuolating activities, which might contribute to different clinical outcomes [6]. It was reported that vacA s1 genotype produced a large amount of cytotoxin, whereas s2 secreted few or no cytotoxin at all. In addition, both s1m1 and s1m2 subtypes were able to produce high or moderate levels of vacA, whereas s2m2 subtype was not [6]. Until now, numerous studies have reported the association between DU and vacA gene s- and m- region genotypes [8–50], but the conclusions were inconsistent. In addition, the relatively small sample size in each single published study may limit the credibility of the conclusions. Therefore, we designed this meta-analysis to evaluate the association between vacA s-/m-region genotypes and DU risk.

Materials and methods

Literature search

The meta-analysis followed the preferred reporting items for systematic reviews and meta-analyses (PRISMA) criteria [51]. We collected all published studies on humans up to Aug. 15, 2013 by systematically searching Embase (1966 to Aug. 2013), PubMed (up to Aug. 2013) and ISI Web of Science (2003 to Aug. 2013) with search strategy: “duodenal ulcer” AND “vacA OR vacuolating cytotoxin A” AND “variation OR variant OR mutation OR polymorphism OR genotype”. The language was limited to English and Chinese. In addition, references from identified articles, reviews and the abstracts presented at related scientific societies meetings were also checking.

Inclusion criteria

The inclusion criteria were as follows: (1) studies of the association between vacA s- or m-region genotypes and DU; (2) studies being published; (3) studies with case–control design; (4) studies with sufficient data for estimating odds ratio (OR) and 95 % confidence interval (CI). Two investigators (Zhang BB and Yang B) screened the title, abstract and full text of each study to determine inclusion independently. If they could not reach a consensus, a third author (Li Y) was consulted.

Data extraction

The following data from the included studies were carefully extracted by the same two authors (Zhang BB and Yang B) independently: name of first author; publication year; country; total numbers, gender ratios and age of cases and controls; frequency of vacA s- or m-regions genotypes in cases and controls. Patients infected with strains of multiple vacA genotypes or undetected vacA genotypes were excluded from the meta-analysis.

Quality score assessment

The quality assessments of all the included studies were performed by the same two authors (Zhang BB and Yang B) independently using the Newcastle–Ottawa Scale (NOS) [52]. Disagreement was settled as described above. The NOS ranges from 0 (worst) to 9 points (best). Studies ≥7 points were assessed to be of high quality.

Statistical analysis

All the statistical analyses were performed using Stata 11.0. Pooled ORs with 95 % CIs were used to calculate the effects of vacA s- and m- region genotypes on the risk of DU respectively. Between-study heterogeneity was assessed by the I2 statistic and Q-test [53, 54]. When there was significant between-study heterogeneity (P < 0.10 and I2 > 50 %), the random effects model was used [55]. Otherwise, the fixed effects model was used [56]. Kappa statistic was used to evaluate the strength of agreement between reviewers regarding study selection. Subgroup analyses were performed according to ethnicity (Asia, Europe and Latin America). Galbraith plot was created to assess the source of heterogeneity graphically. Both Egger’s test and Begg’s funnel plot were applied to analyze publication bias [57].

Results

Study characteristics

The study selection process is detailed in Fig. 1. There were 483 potentially relevant articles after the searching. Based on the titles and abstracts, we included 61 studies for full-text assessment with a Kappa value of 0.87. After full-text assessment, a total of 43 studies were included in qualitative synthesis. As a same population was studied in two studies [32, 33], we included the more comprehensive one [32]. Hence, 42 articles [8–32, 34–50] were finally included in the meta-analysis with a Kappa value of 0.85. The main characteristics of the studies included are shown in Table 1. The NOS results displayed that the average score was 5.69 (range 5–8). Studied countries included Cuba (one study), Bahrain (one study), Italy (one study), USA (one study), Poland (one study), France (one study), Netherlands (one study), Lithuania (one study), Thailand (one study), Iran (two studies), Korea (two studies), Japan (two studies), UK (two studies), Pakistan (two studies), India (three studies), Germany (four studies), Brazil (five studies), Turkey (five studies) and China (six studies). Six studies were from Latin America, 12 from Europe and 24 from Asia. One of the 42 articles was written in Chinese and the others in English.

Fig. 1
figure 1

Flow diagram of the study selection process

Full size image
Table 1 Characteristics of included studies
Full size table

Meta-analysis

There were 34 studies reporting the association between vacA s-region genotype and DU. The prevalence of vacA s1 was 90.9 % (1251/1376) in DU patients and 77.8 % (1190/1529) in controls. The combined OR and 95 %CI showed that vacA s1 increased the risk of DU significantly (OR = 2.96, 95 %CI = 2.34–3.75, Fig. 2). The prevalence of vacA s1 was 92.4 % (611/661) in DU patients and 87.1 % (676/776) in controls in Asian countries; 87.7 % (121/138) in DU patients and 63.3 % (88/139) in controls in European countries; 89.9 % (519/577) in DU patients and 69.4 % (426/614) in controls in Latin America countries. The combined ORs and 95 %CIs showed that vacA s1 increased the risk of DU in Asian countries (OR = 1.92, 95 %CI = 1.30–2.83), European countries (OR = 3.58, 95 %CI = 2.13–6.03) and Latin American countries (OR = 4.20, 95 %CI = 2.21–7.98).

Fig. 2
figure 2

Forest plot of the association between s1 and DU. The white diamonds indicate the summary pooled ORs and 95 % CIs

Full size image

There were 26 studies reporting the association between vacA m-region genotype and DU. The prevalence of vacA m1 is 48.8 % (503/1031) in DU patients and 41.7 % (458/1099) in controls. The combined OR and 95 %CI showed that vacA m1 increased the risk of DU significantly (OR = 1.46, 95 %CI = 1.05–2.04, Fig. 3). The prevalence of vacA m1 was 48.5 % (279/575) in DU patients and 41.9 % (247/590) in controls in Asian countries; 44.5 % (151/339) in DU patients and 40.3 % (162/402) in controls in European countries; 62.4 % (73/117) in DU patients and 47.6 % (49/103) in controls in Latin American countries. The combined ORs and 95 %CIs showed that vacA m1 increased the risk of DU in Latin American countries (OR = 2.98, 95 %CI = 1.59–5.56), whereas no significant associations were shown in Asian countries (OR = 1.27, 95 %CI = 0.75–2.16) and European countries (OR = 1.30, 95 %CI = 0.95–1.77).

Fig. 3
figure 3

Forest plot of the association between m1 and DU. The white diamonds indicate the summary pooled ORs and 95 % CIs

Full size image

There were 20 studies reporting the association between vacA s1m1 genotype and DU. The prevalence of vacA s1m1 is 63.3 % (451/712) in DU patients and 59.1 % (622/1053) in controls. The combined OR and 95 %CI showed that vacA s1m1 increased the risk of DU significantly (OR = 1.89, 95 %CI = 1.47–2.42, Fig. 4). The prevalence of vacA s1m1 was 72.2 % (322/446) in DU patients and 65.2 % (531/814) in controls in Asian countries; 34.4 % (42/122) in DU patients and 26.4 % (32/121) in controls in European countries; 60.4 % (87/144) in DU patients and 50.0 % (59/118) in controls in Latin America countries. The combined ORs and 95 %CIs showed that vacA s1m1 increased the risk of DU in Asian countries (OR = 2.04, 95 %CI = 1.12–3.73) and Latin American countries (OR = 2.05, 95 %CI = 1.20–3.48), whereas no significant association was shown in European countries (OR = 1.42, 95 %CI = 0.80–2.53).

Fig. 4
figure 4

Forest plot of the association between s1m1 and DU. The white diamonds indicate the summary pooled ORs and 95 % CIs

Full size image

There were 20 studies reporting the association between vacA s1m2 genotype and DU. The prevalence of vacA s1m2 was 28.9 % (206/712) in DU patients and 24.9 % (262/1053) in controls. The combined OR and 95 %CI showed no association between vacA s1m2 and DU (OR = 0.96, 95 %CI = 0.74–1.26, Fig. 5). The prevalence of vacA s1m2 was 25.3 % (113/446) in DU patients and 24.0 % (195/814) in controls in Asian countries; 49.2 % (60/122) in DU patients and 43.0 % (52/121) in controls in European countries; 22.9 % (33/144) in DU patients and 12.7 % (15/118) in controls in Latin American countries. The combined ORs and 95 %CIs showed no association between vacA s1m2 and DU in Asian countries (OR = 0.82, 95 %CI = 0.58–1.15), European countries (OR = 1.21, 95 %CI = 0.70–2.08) and Latin America countries (OR = 1.33, 95 %CI = 0.65–2.71).

Fig. 5
figure 5

Forest plot of the association between s1m2 and DU. The white diamonds indicate the summary pooled ORs and 95 % CIs

Full size image

There were 20 studies reporting the association between vacA s2m1 genotype and DU. The prevalence of vacA s2m1 was 25.0 % (178/712) in DU patients and 22.2 % (234/1053) in controls. The combined OR and 95 % CI showed no association between vacA s2m1 and DU (OR = 1.64, 95 % CI = 0.81–3.30, Fig. 6). The prevalence of vacA s2m1 was 24.7 % (110/446) in DU patients and 23.6 % (192/814) in controls in Asian countries; 9.0 % (11/122) in DU patients and 8.3 % (10/121) in controls in European countries; 39.6 % (57/144) in DU patients and 27.1 % (32/118) in controls in Latin America countries. The combined ORs and 95 % CIs showed that vacA s2m1 increased the risk of DU in Latin American countries (OR = 2.30, 95 % CI = 1.17–4.50), whereas no associations were shown in Asian countries (OR = 1.27, 95 % CI = 0.37–4.39) and European countries (OR = 1.54, 95 % CI = 0.59–4.03).

Fig. 6
figure 6

Forest plot of the association between s2m1 and DU. The white diamonds indicate the summary pooled ORs and 95 % CIs

Full size image

Heterogeneity analysis

Significant heterogeneity existed in m1 genotype (I2 = 60.0 %) and s2m1 genotype (I2 = 65.7 %). To explore the source of heterogeneity graphically, we created Galbraith plots. Two studies [23, 43] were identified as the main source of heterogeneity for m1 genotype. Four studies [23, 37, 44, 50] were identified as the main contributors to heterogeneity for s2m1 genotype (Fig. 7).

Fig. 7
figure 7

Galbraith plots. The dots outlier indicate the study of Ashour et al. [23] and Zhang et al. [43]for m1; Ashour et al. [23], Bolek et al. [37], Bindayna et al. [44] and Khan et al. [50] for s2m1

Full size image

Publication bias

No significant evidence of asymmetry was revealed by the funnel plots visually (Fig. 8). In addition, no statistical evidence of publication bias was found using Egger’s regression: P = 0.96 for s1, P = 0.70 for m1, P = 0.59 for s1m1, P = 0.28 for s1m2 and P = 0.54 for s2m1, respectively.

Fig. 8
figure 8

Funnel plots. SE: standard error; OR: odds ratio

Full size image

Discussion

VacA is one of the most commonly studied virulence markers of H. pylori. To date, numerous studies have evaluated the association between vacA genotype and DU, but the conclusions remained inconsistent. In addition, the credibility of results from single case–control study is questionable due to their relatively small sample size. Meta-analysis is of benefit to increase the sample size generating more precise conclusions, which has been widely used in genetic association studies [58, 59]. To our knowledge, the present study is the first meta-analysis assessing the association between vacA s-/m- region genotype and DU. There were 1,876 patients and 2,704 controls in the present study. Results of our study showed that s1 genotype was associated with increased DU risk in overall studied population, and also Asian, European or Latin American population; m1 genotype increased the risk of developing DU in overall studied population and Latin American population; s1m1 genotype increased the risk of DU in overall studied population, Asian population and Latin American population; s2m1 genotype increased the risk of DU in Latin American population.

The s region encodes part of the cytotoxin’s signal peptide and N-terminus, while the m region encodes part of the 55 kDa C-terminal subunit [6]. There were two types of s region: s1 and s2. The s2 genotype was reported to block the vacuolating activity since it encodes a shorter extension of the N-terminal peptide on the mature protein. On the contrary, the s1 genotype was reported to increase cytotoxin activity and thereby lead to gastric inflammation and duodenal ulceration [60]. Similar to s region, m region also has two subtypes: m1 and m2. Type m1 strains demonstrated more toxin activity than m2 strains [6, 61]. In this meta-analysis, we identified 34 articles focusing on the vacA s region genotype and 26 articles focusing on the vacA m region genotype. Among them, ten studies reported that the s1 genotype could increase the risk of DU; five studies reported that the m1 genotype was associated with increased DU risk; one study provided evidence that m1 was the protective factor for DU; the others revealed no significant difference between DU and control. Our study showed that vacA s1 increased the risk of DU by 2.96-folds and vacA m1 increased the risk of DU by 1.46-folds. Regarding the combination of s- and m- region, previous reports showed that the s1m1 genotype was closely tied to a large amount of toxin with high vacuolating activity in gastric epithelial cells, whereas the s1m2 genotype was associated with moderate amounts of toxin and s2m2 was associated with very little or no toxin [6, 60]. In this meta-analysis, we identified 20 articles reporting data on the combination of s- and m- genotypes. The combined results showed that only the vacA s1m1 could increase the risk of DU by 1.89-fold in the overall population. Taken the above results into consideration, it suggests that s1 and m1 are indeed the risk factors during the development of DU. Patients with s1 and m1 genotype would increase DU risk by 1.46–2.96-folds compared those without the same genotype.

To explore a more precise relationship between vacA s-/m- region genotypes and DU, we performed subgroup analyses by ethnicity. Our results demonstrated that vacA s1 increased the risk of DU by 1.92–4.20-fold in all the three subgroups. VacA m1 increased the risk of DU only in Latin American population. In Latin American population, the s1m1 and s2m1 increased the risk of DU by 2.05 and 2.30-folds respectively. In Asian population, only s1m1 showed risk effect on DU. The above data suggest that there is region difference in the vacA m genotype distribution. In addition, gene-environment interaction may also influence the effect of vacA m1 on the development of DU.

Some limitations of this meta-analysis should be noted. Firstly, we could not obtain the original data, which may limit the further evaluation of potential interactions among gene–gene and gene–environment. Secondly, only studies published in English or Chinese were included, which may lead to some inevitable bias, as eligible studies unpublished or reported in other languages would be missed.

Nonetheless, our meta-analysis with robust data and unbiased results demonstrated convincingly that VacA genotypes of H. pylori was well correlated with the risk of developing DU, and the correlation extent was various in different genotypes and also affected by region factors. Genotype testing of vacA s- and m- regions will be useful in screening susceptible individuals for DU development.