Background

Helicobacter pylori (H. pylori) is a bacterial pathogen associated with the gastrointestinal (GI) tract of over 50% of the world’s population [1]. H. pylori, is a Gram-negative spiral-shaped bacterium that colonizes the stomach, was graded as a Group I carcinogen in 1994 by the International Agency for Research on Cancer [2]. With its flagella, H. pylori is capable of moving and can survive on stomach acids, leading to colonization of GI tract cells and irritation and inflammation [3]. Epidemiologic and clinical data have demonstrated the role of H. pylori in up to 75% of non-cardia gastric malignancies and up to 98% of gastric cardia malignancies [4]. There is a strong correlation between gastric cancer (GC) and H. pylori infection [5].

Gastric cancer (GC) is the fifth most common cancer in the world and has the third highest mortality rates, for both sexes [6]. In 2020, actually 1.09 million new GC cases and 0.77 million deaths from GC was estimated all over the world [7]. The overall yearly incidence rates globally are 15.6 to 18.1 and 6.7 to 7.8 per 100,000 individuals in men and women, respectively [8]. According to anatomical subsites, GC can be classified into two categories: cardia GC and non-cardia GC [9]. Cardia and non-cardia GC are treated as two different diseases due to different epidemiological characteristics and distinct pathogeneses. Non-cardia GC is more common than cardia GC. Non-cardia GC accounted for up to 82% of all GC cases around the world in 2018 [10].

The high incidence of H. pylori infection is not always associated with high prevalence of GC. This enigma of H. pylori infection and GC, defined by a very high incidence of infection but a low rate of GC, was first described by Holcombe in 1992 as the "African Enigma" [11]. Hence, the African enigma represents a modification of the inflammatory response triggered by the infection, leading to the absence of any neoplastic manifestations [11]. Other countries including China, Colombia, India, Costa Rica, and Malaysia have described similar enigmas [11]. Several previous studies have suggested that an increased risk of GC is associated with lifestyle behaviors, such as cigarette smoking, intensive alcohol consumption, high salt intake, consumption of processed meat, and low intake of fruits [12]. In addition, host’s genetics has been associated with GC. Mutation in CDH1 gene that encodes E-cadherin protein for cell–cell adhesion has been associated with more than 80% increased risk of GC, and patients with reduced expression of the E-cadherin protein have a poor prognosis [13].

The majority of infections are asymptomatic, therefore a screening and treatment program cannot be justified except for high-risk patients [14]. However, the inflammatory response to an infection in a host and the virulence of the infection vary between individuals. Additionally, environmental exposures may also contribute to the increase in the risk of GCs [15]. Infection prevalence shows large geographical variations. In general, the prevalence of infection is higher in developing countries than developed countries such as Europe and North America [16]. Despite the global prevalence of GC in people with H. pylori infection was reported by Pormohammad et al. [1], a complete up-to-date research on the prevalence of GC in people with H. pylori infection has not been done yet. In the previous study, only studies conducted until 2016 were evaluated. However, in this review, statistics until 2021 were considered. Also, there were several differences between 2 studies in terms of the data bases, time period of search, eligibility criteria, and keywords. Hence, this study aimed to update the GC estimate in H. pylori positive patients after reviewing existing evidence and reassessing the global burden of GC caused by H. pylori in different regions.

Methods

Search strategy

PubMed, Web of Science, and Embase were searched from1 January 2011 to 20 April 2021 to retrieve all relevant studies in the world. MeSH keywords and search strategy were as below: 'Stomach Neoplasm' [tiab], OR 'Cancer of Stomach' [tiab], OR 'Gastric Cancer' [tiab], OR 'Cancer of Gastric' [tiab], OR '' Stomach Cancer '[tiab], OR 'Neoplasm of Stomach' [tiab] AND 'Helicobacter pylori' [tiab], OR 'Campylobacter pylori' [tiab], OR 'Campylobacter pylori subsp. pylori' [tiab] OR, 'Campylobacter pyloridis' [tiab], OR 'Helicobacter nemestrinae' [tiab] AND 'Prevalence' [tiab], OR 'Frequency' [tiab].

Eligibility criteria

We set our inclusion and exclusion criteria based on PECOTS criteria (population, exposure, comparison, outcome, time and study design) (Table 1). For that, all cross-sectional, prospective and retrospective case-series studies which reported the prevalence of GC in H. pylori patients were included. However, case reports and case series with less than five patients (as study population) and also clinical trial studies were excluded. Also, studies without reported prevalence data, definite sample sizes, and clear correct estimates of the prevalence, as well as case–control studies and abstracts presented in scientific meetings with no sufficient data were excluded from this study.

Table 1 PECOTS criteria of the study
Full size table

Study selection

There were 17,438 results from the initial search. Two authors (SK and RP) separately assessed these papers' eligibility, and any discrepancies were settled by consensus. The following step involved excluding 5380 duplicate articles. Also, after reviewing the titles and abstracts of the remaining publications, 11,058 papers were omitted. Of the remaining 1053 articles, 904 ineligible articles were omitted during the review of the entire texts. Eventually, 149 articles that qualified for inclusion were examined.

Quality assessment

Newcastle Ottawa scale (NOS) was used to measure the quality of studies (Table 2). This scale is used to measure the quality of observational studies including cohort, cross-sectional and case series studies. The validity and reliability of this tool have been proven in various studies [17, 18].

Table 2 Quality assessment of studies by Newcastle Ottawa Scale (NOS) checklist
Full size table

Data extraction

Two authors independently performed the study selection and validity assessment and resolved any disagreements by consulting a third researcher. First author, country, enrollment time, published time, type of study, number of Hp+ patients, mean age in Hp+ patients, detection method of Hp, number of patients with cancer, sort (name) of cancer, diagnosis method of GC, and prevalence (95% CI) were extracted from articles.

Statistical analysis

All statistical tests in this study were performed with Stata 14.0. As previous researches [91, 92] the sample size, the number of patients with H. pylori, number of cancer cases in patient with H. pylori, and prevalence of GC in H. pylori positive patients were extracted. We applied Cochran's Q test to determine the heterogeneity. We also quantified it with the I2 index. Based on the Higgins classification approach, I2 values above 0.7 were determined as high heterogeneity. We used random effects model to estimate pooled values where that heterogeneity was high. Also we used the subgroup analysis and meta-regression analysis to find out the heterogeneity sources. Metaprop package were used to calculate the pooled prevalence with 95% confidence interval. Random-effects model was applied to estimate the pooled prevalence. This package applies double arcsine transformations to stabilize the variance in the meta-analyses. The effects of publication time, continents, age mean, sample size and study design on the studies heterogeneity were analyzed by univariate and multiple meta-regression analysis. Publication bias evaluated by “metabias” command. In case of any publication bias, we adjusted the prevalence rate with “metatrim” command applying trim-and-fill approach. Statistical significance was considered 0.05.

Result

A total of 149 studies with 352,872 total sample size were included in our study. Selection process flow chart is available in Fig. 1, and Table 3 shows the studies’ characteristics such as first author, country, published time and type of study. Several primary studies reported overall number of gastric cancer and do not present more detail about cancer. But some primary studies presented more detail about cancer such as anatomical location of it. Many studies mentioned they used histopathology method to detection of cancer. The highest studies number belonged to Asia continent (114 studies) area and Africa continent (6 studies) was the lowest one. All the included studies were published during 1 January 2011 to 20 April 2021. The minimum and maximum age range of the subjects was for Haddadi et al. [93] article with the age ranges (mean age = 26 years old) and Shibukawa et al. [94] study with the mean age = 73 years old, respectively. Sixty-nine (46.31%) of studies were cross sectional, sixty-four (42.95%) of studies were case series and sixteen (10.73%) of studies were cohort.

Fig. 1
figure 1

Flow diagram of study selection

Full size image
Table 3 Characteristics of studies included in the meta-analysis
Full size table

Pooled prevalence of GC in H. pylori positive patients

Figure 2 shows the forest plot of prevalence of GC in H. pylori positive patients. Minimum and maximum prevalence were in Doorakkers et al. [107] study (Prevalence: 0.07%; 95% CI: 0.06–0.09) from the Sweden and Tanaka et al. [161] (Prevalence: 90.90%:95% CI: 83.61–95.14) from Japan, respectively. Due to high heterogeneity and different study design, results don’t merge and presented based on different subgroups

Fig. 2
figure 2

Forest plot of prevalence of gastric cancer in Helicobacter pylori positive patients

Full size image

Pooled prevalence of gastric cancer in H. pylori positive patients based on different subgroups

Pooled prevalence of GC in H. pylori positive patients based on study design and continents are listed in Fig. 3 and Table 4. Based on design, the highest and lowest prevalence was observed in prospective case series (pooled prevalence: 23.13%; 95% CI: 20.41 − 25.85; I2: 97.70%) and retrospective cohort (pooled prevalence: 1.17%; 95% CI: 0.55 − 1.78; I 2: 0.10%), respectively. Also based on continents, the highest and lowest prevalence was observed in America (pooled prevalence: 18.06%; 95% CI: 16.48 − 19.63; I2: 98.84%) and Africa (pooled prevalence: 9.52%; 95% CI: 5.92 − 13.12; I2: 88.39%) continents, respectively.

Fig. 3
figure 3

Pooled prevalence with 95% confidence interval [CI] and heterogeneity indexes of gastric cancer in Helicobacter pylori positive patients based on type of the design and continents places. The diamond mark illustrates the pooled prevalence and the length of the diamond indicates the 95% CI

Full size image
Table 4 Result of meta-analysis, publication bias and fill-trim method for prevalence estimate and corresponding 95% confidence interval of gastric cancer in H.pylori positive patients
Full size table

Heterogeneity and meta‐regression

Heterogeneity results are available in Table 4. Cochran's Q test showed the included studies had high heterogeneity (p < 0.001). The I2 index for total prevalence was up to 98%. The result of univariate meta‐regression analysis (Table 5) showed the age (Coefficient: 0.59; p: 0.009), sample size (Coefficient: − 0.1; p: 0.003) and study design (based WHO regional office) (Coefficient: 3.72; p: 0.015) possess significant effect on the studies heterogeneity (Fig. 4A and B) and have eligible to include to multiple model. The result of multiple meta‐regression analysis showed the just age (Coefficient: 0.66; p: 0.003) have a significant effect on the studies heterogeneity. The R2-adj for multiple model was 13.63% and this mean the age, Sample size and study design explained the about 14% of total heterogeneity of prevalence.

Table 5 The univariate and multiple meta-regression analysis on the determinant heterogeneity in effect of iron therapy on depression
Full size table
Fig. 4
figure 4

Association between Pooled prevalence of gastric cancer in Helicobacter pylori positive patients with age (A) and publication year (B) by means of meta-regression. The size of circles indicates the precision of each study. There is a positive significant association with respect to the pooled prevalence with age

Full size image

Publication bias

The results of Egger’s test showed significant publication bias in our meta-analysis which provided in Table 4. For adjustment of pooled prevalence, fill and trim method was used that result was showed in Table 4. Based on this result, publication-bias-adjusted pooled prevalence estimation for cross sectional was 7.89% (95% CI: 6.78—9.01) which was different with pooled prevalence estimation based on meta-analysis 19.46% (95% CI: 18.34 to 20.57). Result of fill and trim method for other subgroups was showed in Table 4.

Discussion

Infection with H. pylori causes chronic inflammation and significantly increases the risk of developing duodenal and gastric ulcer disease and GC. H. pylori primarily infect the epithelial cells in the stomach and can survive in humans for decades by inhibiting the immune system responsiveness, results inducing chronic inflammatory responses. Because of endotoxin elaboration and other inflammatory exudates, the colonization of the gastric mucosa by H. pylori has been observed with gastric atrophy [173]. Researchers have recently reported molecular aspects that highlight the importance of certain apoptotic genes and proteins including C-Myc, P53, Bcl2, and Rb-suppressor systems in H. pylori pathogenesis. H. pylori infection has also been shown to be related to nitric oxide (NOSi genotype) [70]. Induction of apoptosis in gastric mucosa by H. pylori involves upregulation of Bax and Bcl-2 [70].

With H. pylori involvement in the gastric intestinal pH alteration, dysplasia has been observed in patients with H. pylori infection [174]. Previous studies have been shown that individuals who had been infected with H. pylori were six times more likely to develop GC compared with healthy people [175]. In this study, using random-effects model approach, pooled prevalence of GC in H. pylori positive patients was 8.97% (95% CI: 8.62–9.33) [N = 149; I2 = 98.68%]. Therefore, from every 1000 H. pylori positive patients, 8.62 to 9.33 individuals get GC. The frequency of H. pylori in people less than 50 years old was reported as 41.9%.

The study by Vohlonen et al. showed risk ratio (RR) of stomach cancer in people with H. pylori infection was 5.8 (95%CI: 2.7–15.3) compared to people with healthy stomachs, and 9.1 (95%, CI: 2.9–30.0) in men with atrophic gastritis [86]. The present observation also demonstrated that an H. pylori infection alone (non-atrophic H. pylori gastritis) is by itself a clear risk condition for GC as was suggested by the IARC/WHO statement in 1994 [176]. In study conducted before 1998, by approximately 800 GC cases, the analysis yielded a risk ratios of 2.5 (95% CI: 1.9–3.4) for GC in H. pylori-seropositive people [177]. Another study including 233 GCs and 910 controls, yielded a risk ratios of 6.5 (95%CI: 3.3–12.6) for non-cardia GC in subjects infected with a cytotoxic (CagA) H. pylori strain [178]. In another study, the risk ratios of GC was 3.1 (95%CI: 1.97–4.95) between H. pylori infected and non-infected persons [179]. The risk ratios, based on case–control study designs, varied between 1.6 and 7.9 in three published papers from two extensive prospective nutritional intervention trials of over 29,000 males at age of 50–69 years in Linxian, China and Finland [180,181,182].

Our estimate of the prevalence of GC due to H. pylori infection in cross sectional studies was 19.46% (95% CI: 18.34—20.57) [N = 69; I2 = 98.59%], Therefore, from every 1000 H. pylori positive patients, 183 to 206 individuals get GC.

The simple infection markedly increases the cancer risk when compared to a healthy stomach. The risk varies between the populations with the highest and lowest by 15 to 20 times. East Asia (China and Japan), South America, Eastern Europe, and Central America are the high-risk regions. North and East Africa, North America, Southern Asia, New Zealand, and Australia are the low-risk regions [183].

Our study noted the lowest prevalence of GC in H. pylori positive patients from the Sweden (Prevalence: 0.07%; 95% CI: 0.06–0.09) [107] and the highest from the Japan (Prevalence: 90.90%:95% CI: 83.61–95.14) [161].

This difference may be due to the following reasons: dietary habits, socio-economic status and racial disparities. Suerbaum et al. [184] have mentioned that populations with lower socioeconomic status were more likely to be infected with H. pylori. Data based on National Health and Nutrition Examination Surveys of the United States have also shown that racial disparities played a certain role in the prevalence of H. pylori. The prevalence of H. pylori in African Americans was higher than whites [185]. The findings of the studies showed that Blacks and Hispanics consistently have higher H. pylori prevalence, serologic markers, and histologic signs than whites. Generally, the prevalence of CagA in adult people with H. pylori positivity ranged from 71%-90% in blacks, 64%-74% in Hispanics, and 36% to 77% in whites. Studies that amplified the VacA m allelic region for genomic characterization discovered that Blacks and Hispanics were more likely than whites to carry the virulent VacA-m1 genotype [186]. It has been hypothesized that racial discrepancies associated with H. pylori are contributed to GC incidence and mortality.

The evidence that is currently available implies that practitioners should be aware that the prevalence of H. pylori varies depending on race [187]. Perhaps it would be better if we personalized GC prevention and improved clinical management for all patients.

The results of subgroup analysis, based on our design, the highest and lowest prevalence was observed in prospective case series (pooled prevalence: 23.13%; 95% CI: 20.41 − 25.85; I2: 97.70%) and retrospective cohort (pooled prevalence: 1.17%; 95% CI: 0.55 − 1.78; I 2: 0.10%). The highest and lowest prevalence of GC in H. pylori patients was observed in America (pooled prevalence: 18.06%; 95% CI: 16.48 − 19.63; I2: 98.84%) and Africa (pooled prevalence: 9.52%; 95% CI: 5.92 − 13.12; I2: 88.39%) continents, respectively.

Steady declines in GC incidence rates have been observed worldwide in the last few decades [183]. The general declining incidence of GC may be explained by higher standards of hygiene, improved food conservation, a high intake of fresh fruits and vegetables, and by H. pylori eradication [188]. Current treatment for H. pylori infection includes antisecretory agents or bismuth citrate plus two or more antimicrobials. Clarithromycin and metronidazole are the most commonly used antibiotics to treat H. pylori infection. Increasing resistance of H. pylori to metronidazole and clarithromycin has made current therapies with these antibiotics less successful [68]. Bismuth triple therapy is not very effective in the presence of a high prevalence of metronidazole resistance, unless higher doses of metronidazole are prescribed to increase the cure rate of therapy. Resistance to the major anti-H pylori antibiotics, the final duration of therapy, and the prescribed antibiotic dose are all factors that affect the efficacy of therapy. Host genetic polymorphisms may also influence the efficacy of therapy [189].

The results of our study indicated a significant heterogeneity (p < 0.001) in the prevalence of H. pylori in GC across different geographical regions. The result of univariate meta‐regression analysis showed the age, sample size and study design possess significant effect on the studies heterogeneity and have eligible to include to multiple model. The results of multiple meta‐regression analysis showed the just age have a significant effect on the studies heterogeneity. The R2-adj for multiple model was 13.63% and this mean the age, sample size and study design explained the about 14% of total heterogeneity of prevalence. This was in accordance with a recent study that assessed the prevalence of H. pylori in gastrointestinal disease cases [97]. Study performed by Spineli et al. [98] revealed that subgroup analysis may not be powerful enough to test for relationships between variables when fewer studies are involved. However, type of sample was significantly associated with H. pylori prevalence [184]. Although subgroup analysis and meta-regression were performed to minimize the heterogeneity across the included studies, significant heterogeneity still could be observed in subgroup analysis. Moreover, some important factors like drinking and dietary habit could not be extracted from the included studies, which might have potential influence on the heterogeneity.

Therefore, these results should be considered with caution and more studies are needed to further confirm these results in the future.

In general, limitations of meta-analyses are that the validity is dependent on the quality of the included studies, on heterogeneity between studies, and on possible publication bias; but we tried to deal of them by statistical manner. Indeed we dealt to heterogeneity by using random effects model, subgroup and meta-regression analysis. Also we tried to deal publication bias by use the fill and trim method to estimate the publication-bias-adjusted-pooled.

Conclusions

In our study by evaluate the 149 studies and 352,872 sample size illustrated that prevalence of GC in patient with H. pylori was considerable. But the rate was varied based on different subgroups so that the rate was highest among in America continent but was lowest in Africa continent. Also, using meta-regression and assessment the effect of several variables, indicated that age, sample size and study design explained the about 14% of total heterogeneity. It is advised to launch appropriate control guidelines for high-risk region. The risk of different factors should also be taken into account when developing GC decrease strategies, even though H. pylori eradication may be a promising method for preventing the disease.