Background

Helicobacter pylori (H. pylori) is a S-shaped microorganism that colonize in the surface of gastric mucosa of half the world’s population, maybe even more [1]. Long last colonization with this bacterium leads to a chronic progressive gastric inflammation associated with severe gastrointestinal effects [2]. Nowadays, eradication of H. pylori is the main therapeutic strategy in management of patients who suffering from different complications including peptic ulcer disease (PUD), gastric cancer (GC), mucosa associated-lymphoid tissue (MALT) lymphoma, and atrophic gastritis [3]. According to the Kyoto Global Consensus Conference, eradication of H. pylori infection among the asymptomatic subjects seems an necessity [4]. Nevertheless, the rate of the treatment for H. pylori infection is declining annually; the emergence of clarithromycin-resistant strains has been declared a global threat by the World Health Organization (WHO) [5, 6].

The cure rate of H. pylori infection could be affected by both microbial (high bacterial load, point mutations, biofilm formation, efflux pumps, and virulence factors), and non-microbial (cytochrome P450 2C19 polymorphism, multidrug resistance transporter-1, pro-inflammatory cytokines polymorphism, smoking, life style, duration of treatment, high gastric acidity, poor patient compliance) factors; all of these factors play a role in the severity of the infection [3, 7, 8]. Vacuolating cytotoxin A (vacA) and cytotoxin associated gene A (cagA) are considered as the main virulence factors of H. pylori [9]. The toxin encoded by the vacA gene causes apoptosis, T-cell activation, and persistent infection (through inhibition of immune system), which these changes are lead to severe gastrointestinal outcomes [10]. Full-length sequence analysis of the vacA gene showed that this gene has a mosaic structure and is encoded by different subfamilies s1, m1 and m2 alleles, with its own biological activities [11]. The vacA s1/m1 genotype possess the highest toxicity property for host cells, while the vacA s2/m2 genotype biologically is inactive [12, 13]. CagA is encoded by cagA gene; this toxin is highly immunogenic, and upon entering the host cell, it activates kinases through EPIYA motifs in its C-terminal, which in turn disrupt signaling pathways [14].

Studies have shown that this protein induces IL-8 expression, which contributes to the formation of cytokine storms and eventually susceptibility to PUD as well as GC [15]. Both CagA and VacA antigens significantly affect the colonization and pathogenesis of this bacterium, and play a determining role in cure rate of disease [16, 17]. Although chromosomal mutations are considered to be the main mechanism of antibiotic resistance, but, the location of these single nucleotide polymorphisms (SNPs) is not the same in all populations, and therefore, understanding the mechanisms of antibiotic resistance of H. pylori is essential for the introduction of rational antibiotic combinations [18]. In recent studies, the eradication results associated with CagA and VacA status are highly inconsistent [19,20,21,22]. Interestingly, in meta-analysis by Wang et al. (collecting the data from 26 papers), it was represented that the eradication rate of infection in patients infected with vacA s1/cagA positive strains was more conducive compared to less virulent strains [8].

In this study, we performed a comprehensive literature search to demonstrate the relationship between cagA or vacA status and antibiotic resistance in H. pylori.

Methods

Eligibility of relevant studies

Using international databases such as the Clarivate analytics, PubMed, Scopus, EMBASE, DOAJ, and Google Scholar, related articles to the effect of cagA and vacA on the antibiotic resistance of H. pylori were reviewed, regardless of publication and language restrictions until April 2022. In this regard, we used keywords based on MeSH terms such as “Genotype”, “Antibiotic resistance”, “Helicobacter pylori”, “H. pylori”, “VacA”, “CagA”, and “Antimicrobial resistance”. The bibliography of articles was reviewed manually to retrieve missing related studies.

Inclusion and exclusion criteria

Our inclusion criteria were the following: (1) studies on the association between cagA/vacA status and antibiotic resistance; (2) studies on human subjects; (3) studies based on standard methodology (CLSI); (4) studies without repetitive samples. On the other hand, studies such as case reports, reviews, congress abstracts, duplicates, studies on non cagA/vacA genes, in vitro studies, as well as studies without clear results were excluded from this study.

Data extraction

Eligibility of studies was evaluated by the two authors separately, and conflicting of interest was resolved by discussion. The main items were including: first author, country, year of publication, number of H. pylori isolates, number of cagA + isolates, number of vacA s1m1 + isolates, antimicrobial susceptibility tests, and frequency of each genotype (cagA and vacA s1m1) resistant to clarithromycin, metronidazole, amoxicillin, tetracycline, and levofloxacin (Table 1) [23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63].

Table 1 Characteristics of included studies
Full size table

According to the literature, vacA s1m1 is the most virulent genotype of H. pylori, nevertheless, in the present meta-analysis, we evaluated the frequency of other vacA genotypes in all eligible studies. The distribution of antibiotic resistance of three genotypes vacA s1m2, vacA s2m1, and vacA s2m2 was assessed and their results are shown in Table 2.

Table 2 Distribution of antibiotic resistance in vacA genotypes
Full size table

Quality assessment

The Newcastle–Ottawa scale (NOS) was used to assess the quality of the included studies. The quality of studies was evaluated based on the items such as selection, comparability, and outcome, so that NOS scores in the range of 1–3, 4–6, and 7–9 were considered low, medium, and high respectively. The quality appraisal process was performed separately by the two authors, and the disagreement was resolved through discussion.

Statistical analysis

Retrieved studies was analyzed using Comprehensive Meta-Analysis (CMA) software version 2.2 (Biostat, Englewood, NJ, USA). Frequency of cagA- and vacA-positive strains was measured based on the event rate with 95% confidence interval (95%CI). Finally, the association between the genotypes of these virulence factors and resistance to clarithromycin, metronidazole, amoxicillin, tetracycline, and levofloxacin was calculated using the odds ratio (OR) and corresponding 95% CI. For measuring heterogeneity, we used from two parameters Cochran’s Q statistic and I2 statistic. The fixed-effects model was used when there was no significant heterogeneity (p value ≥ 0.10 and I2 ≤ 50%) between the studies [64]; a random-effect model based on the Dersimonian and Laird method was used if significant heterogeneity was identified [65]. Eventually, publication bias was assessed by Egger’s p value test, Begg’s p value test, and asymmetry of funnel plot [66]. We also used the “trim-fill” method to prove the correction effect on publication bias according to Duval and Tweedie [67, 68]. We performed subgroup analysis based on several items such as ethnicity, study sample size, diagnostic test, and developing/developed status of country. Moreover, the leave-one-out method as sensitivity analyses were performed to estimate the effect of each included study on overall effect [69]. A random effects meta-regression analysis was performed to assess the potential sources of heterogeneity to explore factors that may be associated with between-study variations in H. pylori antibiotic resistance.

Results

Characteristics of the included studies

A systematic literature search was conducted based on PRISMA guideline. In the first stage, 509 articles were selected as potential documents. According to the inclusion criteria 471 articles were deleted and finally 38 eligible articles were entered in the present research (Fig. 1). Of all eligible studies, 38 articles had evaluated the relationship of cagA and antibiotic resistance, while 23 articles had assessed the effect of vacA genotypes on antibiotic resistance. The NOS results showed that the quality of eligible studies was ranged between 6 and 8. All studies in had been performed in regions such as Asia, Europe, and Latin America during 2001–2020. Standard methods for detecting antibiotic resistance included agar dilution, modified disk-diffusion agar, E-test, PCR-RFLP, GenoType HelicoDR kit. In the present study, 5156 of clinical positive samples were evaluated, and consequently the frequency of infection with cagA and vacA s1m1 was computed 64.6% (95% CI: 58.4–70.4) and 41.9% (95% CI: 34.3–50.0), respectively.

Fig. 1
figure 1

The flowchart of included studies

Full size image

The vacA status and antibiotic resistance

Overall, 23 articles had appraised the vacA genotypes status and resistance to clarithromycin, metronidazole, amoxicillin, tetracycline, and levofloxacin. Interestingly, we found that the vacA s1m1 significantly reduced the risk of resistance to metronidazole (OR: 0.41; 95% CI: 0.20–0.86) (Fig. 2). After exclusion 4 studies, the sensitivity analysis was similar (OR: 0.34; 95% CI: 0.29–0.40) without significant heterogeneity rate. Moreover, the results were not significant for other antibiotics (Table 3). Due to the presence of a significant asymmetry in funnel plots, we performed trim and fill method to exclude potential publication bias. Adjusted OR according to the trim-and-fill method was lower than the original estimates but results were similar to the original findings (OR: 0.25; 95% CI: 0.11–0.57); however, a significant difference was not noted between before and after filling the potential missing studies (Fig. 3). Thus, trim and fill method did not change conclusion, indicating that our results were statistically robust regarding potential association between vacA s1m1 and resistance to metronidazole.

Fig. 2
figure 2

The forest plot associated with vacA s1m1 and resistance to metronidazole

Full size image
Table 3 Odds ratio (OR) with 95% CI for vacA s1m1 genotype and antibiotic resistance in H. pylori
Full size table
Fig. 3
figure 3

The funnel plot adjusted using a trim and fill method for evaluation of possible link between vacA s1m1 and resistance to metronidazole

Full size image

The details of overall estimates related to vacA s1m1 based on the sample size of the study, diagnostic test, and developing/developed status of country are given in the Table 4.

Table 4 The vacA s1m1-positive status and metronidazole resistance
Full size table

In subgroup analysis, the results showed that in an Asian population vacA s1m1 significantly increases the resistance of H. pylori to metronidazole (OR: 0.37; 95% CI: 0.15–0.90), while in Western countries, vacA s1m1 increases resistance to amoxicillin and levofloxacin. (OR: 16.58; 95% CI: 1.77–154.58, and OR: 6.25; 95% CI: 1.63–23.84, respectively). We showed that vacA s2m2 decreases resistance to all five antibiotics (clarithromycin, metronidazole, amoxicillin, tetracycline and levofloxacin). On the other hand, vacA s1m2 decreases resistance to clarithromycin and metronidazole, while vacA s2m1 only decreases resistance to clarithromycin. Details on the relationship between non-vacA s1m1 genotypes and antibiotic resistance are summarized in Table 5.

Table 5 Odds ratio (OR) with 95% CI for Non-vacA s1m1 genotypes and antibiotic resistance in H. pylori
Full size table

A meta-regression was performed to examine the sources of heterogeneity according to the publication year or NOS score; the results of meta-regression showed that H. pylori antibiotic resistance was significantly influenced by publication year (Slope intercept: -0.18; 95% CI: -0.24 to -0.12; SE: 0.029; p value: 0.01) or NOS score scale (Slope intercept: -7.30; 95% CI: -8.98 to -5.63; SE: 0.85; p value: 0.01). In subgroup analysis, we found no association between the high virulent strains containing cagA-vacA s1m1 and antibiotic resistance (Fig. 4). In general, it seems that the degree of antibiotic resistance in strains with high pathogenicity is not different from the strains with low virulence. Due to heterogeneity and publication bias, we need further studies with larger sample sizes.

Fig. 4
figure 4

The forest plot associated with cagA-vacA s1m1 and resistance to clarithromycin and metronidazole

Full size image

The cagA status and antibiotic resistance

Association between cagA status and resistance to clarithromycin, metronidazole, amoxicillin, tetracycline, and levofloxacin had been measured in 40 articles. Based on the current results, it seems that cagA significantly increases metronidazole resistance (OR: 2.69; 95% CI: 1.24–5.83; p value: 0.01), especially in Western countries (Fig. 5). By discovering the potential sources of heterogeneity, we excluded 3 studies. Sensitivity analysis showed a similar OR: 2.67 (95% CI: 1.20–5.94; p value: 0.01). The details of overall estimates related to cagA based on the sample size of the study, diagnostic test, and developing/developed status of country are addressed in the Table 6. However, the results of Egger’s regression test and asymmetry of funnel plot showed evidence of publication bias in overall estimates. Thus, we have performed the trim and fill method to adjust for publication bias. The pooled OR did not show the correlation between cagA status and antibiotic resistance (OR: 0.29; 95% CI: 0.13–0.64; p value: 0.001). Hence, after imputed missing studies by the trim and fill method, the adjusted estimate significantly dropped from OR: 2.69 (95% CI: 1.24–5.83) to OR: 0.29 (95% CI: 0.13–0.64) that revealed there is no relationship between cagA status and resistance to metronidazole. The population sample size was low in some included studies that may cause to this significant difference between adjusted OR and original estimates. More extensive research is needed to confirm the present findings.

Fig. 5
figure 5

The forest plot associated with cagA and resistance to metronidazole

Full size image
Table 6 The cagA-positive status and metronidazole resistance
Full size table

In addition, our findings showed a non-significant association between cagA status and resistance to clarithromycin, amoxicillin, tetracycline, and levofloxacin. The results of cagA status and resistance to these antibiotics are listed in Table 7. Sensitivity analysis also confirmed the stability of the overall estimates after excluding studies that may cause significant heterogeneity.

Table 7 Odds ratio (OR) with 95% CI for cagA genotype and antibiotic resistance in H. pylori
Full size table

A meta-regression was performed to examine the sources of heterogeneity according to the publication year or NOS score; the results of meta-regression showed that publication year (Slope intercept: − 0.150; 95% CI: − 0.20 to − 0.10; SE: 0.025; p value: 0.01) or NOS score scale (Slope intercept: − 5.26; 95% CI: − 6.82 to − 3.69; SE: 0.79; p value: 0.01) was disrupted the association between cagA status and H. pylori antibiotic resistance. In the subgroup analysis, our results showed that cagA increases resistance to metronidazole, amoxicillin, and levofloxacin only in the Western population (OR: 1.59; 95% CI: ‎0.78–3.21, OR: ‎19.68‎; 95% CI: 2.74–‎‎141.18, and OR: ‎11.33; 95% CI: ‎1.39–91.85‎, respectively), nonetheless, the results associated with the Asian countries were not significant (Table 8). After the trim and fill method, the adjusted OR was slightly lower than original estimates (but not significant difference) which indicates the reliability of the overall estimates.

Table 8 Results of subgroup analysis for both Asian and Europe/America (West) populations
Full size table

Publication bias

The results of Egger’s and Begg’s tests, as well as funnel plot asymmetry showed a significant publication bias; however, when the trim-and-fill method was performed to correct the results, the adjusted OR for vacA genotypes was decreased but no significant difference was observed compared to original estimates (Fig. 6). However, the adjusted OR for cagA status and resistance to metronidazole was dropped significantly that represents there is no association between cagA status and antibiotic resistance.

Fig. 6
figure 6

The trimmed and filled funnel plot represent the correlation between the standardized OR and its standard error with pseudo 95% confidence limits regarding possible association between cagA status and resistance to metronidazole

Full size image

Discussion

The cagA and vacA genes are the most well-known virulence factors of H. pylori, and previous studies have demonstrated that infection with cagA-vacA s1m1 positive strains can increase the risk of severe gastrointestinal disorders [70, 71]. Wang et al. understood that infection with strains carrying both cagA and vacA products could increase the chance of eradicating H. pylori infection, however, the reported heterogeneity was significant [8]. Infection with cagA-positive strains can be led to gastric mucosal inflammation, which in turn increases the diffusion of antibiotic (following an increase in blood flow, disruption of mucosal barrier, and inhibition of IL-1β-induced gastric acid secretion) and ultimately high cure rate [72, 73]. Interestingly, vacA s1-positive strains reduce the risk of treatment failure due to induce sever gastric inflammation and lower expression of somatostatin [74, 75].

To the best of our knowledge, this is the first meta-analysis study that investigated the potential association between H. pylori virulence factor and antibiotic resistance. Based on this analysis, a considerable association exists between the status of vacA-cagA genes and resistance of H. pylori to commonly used antibiotic agents. The results of the present study indicated that cagA-positive strains can significantly increase resistance to metronidazole (OR: 2.69; 95% CI: 1.24–5.83; p value: 0.01). Although, s1m1 genotype of vacA significantly reduces resistance to metronidazole, vacA s1m2 reduces resistance to both clarithromycin and metronidazole. Moreover, vacA s2m1 decreased resistance to clarithromycin, as well as vacA s2m2 decreased resistance to metronidazole, clarithromycin, amoxicillin, tetracycline, and levofloxacin. We showed that cagA-positive strains in particular in Western countries increase the risk of resistance to metronidazole, amoxicillin, and ciprofloxacin.

In their study, Chisholm et al. asserted that resistance against metronidazole was not merely due to mutation in the rdxA gene, but was influenced by a variety of mechanisms [76]. In a study by Kim et al., they showed that resistance to metronidazole could occur even in the lack of rdxA expression or truncated RdxA [77]. Correlation between cagA pathogenicity islands (PIA) and resistance to metronidazole first was investigated by Alfizah et al.; they found that strains containing an intact cagPAI region were sensitive to metronidazole, while strains possessing partially deleted cagPAI regions were resistant to metronidazole [42]. Variations in the 3’ terminal of cagA lead to the differentiation of new subclones with unique genetic characteristics, and due to this fact, Rengifo et al. in their study demonstrated that genetic changes in this region cause the formation of antibiotic-resistant subclones [43, 78]. Recent studies show that in patients treated with antibiotics, new subclones of cagA are formed due to recombination and quorum sensing, which differ in some features and this phenomenon is effective in antibiotic resistance [79, 80]. We showed that gastric colonization with cagA-positive strains, especially in Western countries, can potentially increase the risk of resistance to common antibiotics. In a study conducted by Yue et al., they realized that the prevalence of resistance to metronidazole in strains with Western-type cagA 3’ variable region was significantly higher than East Asian-type strains [81, 82]. Today, evidence suggests that CagA protein is involved in processes such as integron acquisition, biofilm formation, and efflux pump function [83,84,85]. In general, cagA-positive strains, especially in the Western population, seem to be considered as diagnostic biomarkers in the phenomenon of antibiotic resistance. Recently, Ayibatari et al. revealed that patients carrying Western-type cagA had higher rates of gastritis than East Asian-type cagA [86].

Our results showed that vacA s2m2 genotype was associated with a significant decrease in resistance to antibiotics. Strains containing vacA s2m2 genotype are not able to produce VacA cytotoxic antigen [87]. Krzyżek et al. observed that the change to coccoid form in vacA s1m1 strains was significantly higher than vacA s2m2 strains [88]. Studies show that vacA s2m2 strains have higher nutritional requirements and are also less compatible with antibiotics, so they are more sensitive to antibiotics [89,90,91]. Though, our results suggested that there is no meaningful association between cagA/vacA s1m1 double positive H. pylori infection and antibiotic resistance. The biofilm formation capacity of vacA s1m1 genotype is higher than other genotypes, which in turn is an effective strategy in antibiotic resistance [92, 93]. Our results (as several cross-sectional studies) showed that the s1m1 and s1m2 genotypes reduce the risk of resistance to metronidazole and clarithromycin [59, 94,95,96]. Strains containing s1 or m1 are strong immunogens to stimulate the immune system and gastritis, so antibiotic delivery in the stomach lumen increases due to increased blood flow [39]. Nevertheless, the effect of other virulence factors may be ignored, for example Brennan et al. showed that the incidence of infection with s1m1/s1m2 strains was higher in treatment-naïve patients than in those previously treated [91].

Overall, our statistical analysis showed that metronidazole resistance was significantly high in cagA-positive H. pylori strains. As well as, less virulent vacA s2m2 genotype was sensitive to all antibiotics. Our study had several limitation including: (1) small ample size; (2) study only on adult population; (3) high heterogeneity among the included studies; (4) imbalanced geographical distribution; (5) inaccessibility to raw data to assess bacterial density and other factors in cag PAI; (6) publication bias. However, we performed meta-regression and sensitivity analyses to diminish the effects of heterogeneity on the reliability of the pooled estimates. Meta-regression and sensitivity analyses assisted us exclude the impact of some positive data on the overall estimates. Moreover, we used random-effects models to establish associations among the moderate variables with high heterogeneity. Therefore, it is appropriate to present evidence, but the findings should be interpreted with more caution. In the current meta-analysis, publication bias considerably changed the association between cagA status and resistance ‎to metronidazole according to the trim-and-fill method. Meanwhile, adjusted OR for vacA genotype and antibiotic resistance after implementation of the trim and fill producer revealed that results were slightly lower without significant difference with overall estimates.

Conclusions

In the current meta-analysis, our findings showed that infection with cagA-positive strains of H. pylori significantly increases the risk of metronidazole resistance in Western countries. In addition, vacA s1m1 increases resistance to amoxicillin and levofloxacin in Western countries. According to our findings, the vacA s1m1 significantly increases resistance to metronidazole, while the vacA s1m2 decreases resistance to clarithromycin and metronidazole. Additionally, antibiotic resistance to clarithromycin, metronidazole, amoxicillin, tetracycline, and levofloxacin in less virulent H. pylori strains (carrying vacAs2m2 genotype) is significantly lower than others. We also performed the trim and fill method to exclude the potential bias from the overall estimates. Although, the adjusted OR was slightly lower than original estimates but this difference was not significant.