Helicobacter pylori‐related risk predictors of gastric cancer: The latest models, challenges, and future prospects

ABSTRACT Helicobacter pylori is known as an important determinant of preneoplastic lesions or gastric cancer (GC) risk. The bacterial genotypes may determine the clinical outcomes. However, the evidence for these associations has varied between and within continents, and the actual effect of each gene and corresponding allelic variants are still debatable. In recent years, two new models have been proposed to predict the risk of GC; the phylogeographic origin of H. pylori strains and a disrupted co‐evolution between H. pylori and its human host, which potentially explain the geographic differences in the risk of H. pylori‐related cancer. However, these models and earlier ones based on putative virulence factors of the bacterium may not fully justify differences in the incidence of GC, reflecting that new theories should be developed and examined. Notably, the new findings also support the role of ancestry‐specific germline alteration in contributing to the ethnic/population differences in cancer risk. Moreover the high and low incidence areas of GC have shown differences in transmission ecology, largely affecting the composition of H. pylori populations. As a new hypothesis, it is proposed that any high‐risk population may have its own specific risk loci (or variants) as well as new H. pylori strains with national/maybe regional gene pools that should be considered. The latter is seen in the Americas where the rapid evolution of distinct H. pylori subpopulations has been occurred. It is therefore proposed that the deep sequencing of both H. pylori and its human host is simultaneously performed in GC patients and age‐sex‐matched controls from high‐risk areas. The expression and functional activities of the identified new determinants of GC must then be assessed and matched with human and pathogen ancestry, because some of risk loci are ancestry‐specific. In addition, potential study‐level covariates and moderator variables (eg physical conditions, life styles, gastric microbiome, etc) linked to causal relationships, and their impact, should be recognized and controlled.

(hspSAfrica, hspWAfrica, and hspCAfrica), hpAfrica2, hpNEAfrica, hpEurope, hpSahul, hpAsia2, and hpEastAsia (hspEAsia, hspAmerind, and hspMaori) Table 1. 15,17,21 These populations are the result of a bacterial genetic separation that led to a reduction in the recombination between strains in some populations. 13 In Africa, most strains are hpAfrica1, hpAfrica2 or hpNEAfrica and in South Asia, the HP strains are mostly hpAsia2 with a low rate of GC,while, in those areas with hpEastAsia, especially hspEAsia, the incidence of GC is so high Figure 1. Thus, East Asian strains are more related to the incidence of GC compared to other isolates designated by phylogeographic origins. 22 European strains are highly related to the increased premalignant histologic lesions (mean 3.9, 95% CI,3.6-4.2 for the hpEurope strains vs mean 2.9, 95% CI,2.5-3.3 for the hpAfrica1 strains, P = .001) and the DNA damage of human gastric epithelial cells (mean 46.0%, 95% CI, 40.1%-51.9% for the hpEurope strains vs mean 22.7%, 95% CI, 13.4%-31.9% for the hpAfrica1 strains, P = .001). 5 In Colombia, it has been shown that the population residing in the Andean highlands of Colombia who had just European ancestry were 25-times more exposed to GC risk than the people living in the coastal region who had European and African phylogeographic origin. 5 In Iran, the ethnic-geographic diversity is so high and almost 69% of Iranian people are infected with HP. 23 In addition, it has been shown that the frequency of GC is strongly impressed by the ethnic/geographic origin. 16 By studying the sequence of house-keeping genes of 68 strains belonging to 9 global HP populations and 14 Iranian strains, Latifi-Navid et al. showed that Iranian strains are in the same group as hpEurope population. HpEurope consisted of mixing the two populations of Ancestral Europe1 (AE1) and Ancestral Europe2 (AE2), the rates of which vary based on geographic situation. The Iranian strains have relatively equal contributions from the two ancestral sources, thus, it does not seem that there had been any sources from Iran before the expansion of populations throughout Western Eurasia. 16 Although all the Iranian HP strains showed hpEurope ancestry, there were considerable differences in age-standardized GC incidence rate within the country. 24,25 | 4811 BAKHTI eT Al.

| Mechanisms of carcinogenicity of HP risk factors
It seems that the prevalence of GC depends on the pathogenic factors of the bacterial strains, environmental co-factors, and host susceptibility. Several studies have shown the considerable differences in the prevalence of the vacA alleles among patients with different clinical outcomes. [26][27][28][29] Therefore, the particular genotypes of HP may be related to the severity of gastrointestinal diseases. 9,30 Various HP pathogenic factors including vacA, cagA, babA, hopQ, oipA, and homA/B are the predictors of gastric atrophy, intestinal metaplasia, and severe clinical consequences. 30,[31][32][33][34] The cagA (cytotoxin-associated gene A) and vacA (vacuolating cytotoxin A) genes are the two main determinants of HP-associated disease risk that are mainly involved in the chronic gastritis and damage of epithelial cells leading to GC. 26,35 These HP CagA and VacA oncogenic proteins induce GC with a multi-step process. 36,37 Injecting CagA into the gastric cells leads to the upregulation of oncogenes and silencing of tumor suppressor genes. It increases the level of reactive nitrogen and oxygen species, causing the DNA damage and break-down of the cells as well as changing the microRNA expression profile. VacA results in the EGFR (epidermal growth factor receptor) expression and promotes the GC formation. 38

| Role of the vacA genotypes in the incidence of GC
One of the first pathogenic factors discovered in HP is the VacA protein that plays an important role in the bacterium establishment in the gastric mucosa and the incidence of HPrelated diseases. 39 Although all HP strains carry the vacA gene, there are significant differences in the vacuole-creating activities between strains. These differences are related to the vacA gene structure in the signal (s), intermediate (i), and middle (m) regions. These regions include the s1 (s1a, s1b, and s1c), s2, i1, i2, m1, and m2 subtypes. 40 Region i plays an important role in the vacuole-creating activity and has a strong association with GC. 9,41 This relationship is independent of and bigger than the associations of the s-or m-types of vacA and cagA status with GC. 42 Recently, one deletion of 81 bp has been found between the m and i regions and called the d region (d1 or d2). The vacA d1 (no deletion) genotype showed a strong association with the neutrophil infiltration and gastric mucosal atrophy in both the corpus and the antrum. 8 The highest frequencies of the pathogenic vacA-s1/-i1 and cagA + genotypes are found in Eastern Asia (ASR-Both Sexes-= 24.2) and South-Eastern Asia (ASR-Both Sexes-= 6), although these two regions are very different in the incidence rate of GC. The vacA d1 genotype is also the most abundant bacterial genotype in Eastern Asia and has not been studied in South-Eastern Asia Table 2. However, the highest frequencies of the pathogenic vacA alleles (except m1) and the cagA + genotype are found in areas with the high incidence of GC (ASRs-Both Sexes-≥ 20), although there is often no association between these genotypes and the risk of GC Table 3. Heterogeneity in the vacA gene might be an important risk factor in the creation of different gastrointestinal diseases, especially GC, in HP-infected patients. 9,26,[43][44][45] HP strains with the vacA s1, i1, m1, or d1 genotypes from Western countries were strongly associated with the risk of GC (adjusted ORs: 3.17, 8.57, 10.65, and 8.04, respectively). In Spanish patients, infection with the HP strains representing the vacA s1m1/ cagA + genotype has strongly increased the progression of gastric precancerous lesions compared with the strains carrying the vacA s2m2/cagA − genotype (OR = 4.80). 6 Moreover, it is proposed that the d region genotype (d1 vs d2) is an important risk locus for GC in Western strains. 8 In contrast, in the HP strains from Eastern countries, there was no significant association between the vacA polymorphisms, clinical consequences, and histopathologic patterns. 8 A new polymorphism region at the 3 ʹ-end of the vacA gene has been recently identified and named as the vacA c (c1: with deletion of 15 bp) region. The vacA c1-type was significantly associated with the risk of GC. This association was independent of and larger than the associations of the i1-, m1-, and d1-types of the vacA alleles or cagA status with GC. 26 We also reported that the vacA c1 genotype had an intense association with the risk of cardia gastric adenocarcinoma (CGA), non-CGA, diffuse-type gastric adenocarcinoma (DGA), and intestinal-type gastric adenocarcinoma (IGA), whether the controls were non-tumor (ie those with non-atrophic gastritis (NAG) and PUs,the OR was 14.11, 9.59, 11.91, and 16.93, respectively) or those with NAG (the OR was 10.71, 8.11, 9.56, and 11.22, respectively). 46 Little is known about all the constellations of five polymorphic sites of H. pylori vacA and cagA status. Recently, we found novel five-and six-genotype combinations (vacA s1m1i1d1c1, s1m2i1d2c1, s1m2i2d2c1, and s1m2i2d2c-1cagA) associated with the risk of GC. The ORs were 2.433, 11.524, 4.200, and 6.263, respectively. These associations were largely dependent on the presence of c1-type of vacA. Therefore, analysis of all the combined genotypes of the vacA alleles and the cagA status may play an important role in determining clinical outcomes associated with H. pylori. 47 The detailed information regarding the associations of HP vacA gene polymorphisms with the risk of GC in low-, moderate-, and high-incidence areas are indicated in Table 4. In most areas, the association of some HP vacA alleles with the risk of GC has been demonstrated. However, most studies in  East Asia with the high incidence of GC did not show such a significant association, due mainly to the high prevalence of the pathogenic vacA alleles in this area.

| Role of the cagA genotype in the incidence of GC
cagA is another pathogenic gene of HP 48 and cagA + strains are more pathogenic than cagA − strains. 49 In this regard, epidemiological studies have disclosed that the attendance of the cagA is associated with the development of gastric precancerous lesions, GC, 50,51 increased mucosal inflammation, 52 and cell proliferation, 53 however, other reports have not shown such a relation. 54,55 It is reported that more than 90%-100% of HP strains from East Asian countries are cagA + . 56,57 However, this high prevalence of the cagA + genotype is independent of its association with clinical consequences. In contrast, just 50%-70% of HP strains from Western countries have the cagA + genotype. 43,56,58,59 Given the reports from Western countries, it is likely that the cagA + genotype may be a useful biomarker for prediction of clinical outcomes,however, in Eastern countries, this issue requires more investigations. It seems that the determination of the number of phosphorylation sites of CagA is more important than the detection of the cagA genotype alone. 59,60 A large epidemiological study of 2145 patients in Venezuela (in Latin America,a region with the high incidence of GC and HP infection) also showed a strong association between the cagA-positive genotype and the severity of gastric precancerous lesions. The odds ratios were increased with the severity of the disease. The OR was 2.00 for chronic gastritis, 7.35 for intestinal metaplasia type II and 16.7 for dysplasia. 61 Moreover, a large cross-sectional study in China (Linqu County) showed that individuals seropositive to Omp, CagA, VacA, and HP0305 significantly increased the risk of precancerous lesions (ORs = 5.37, 3.23, 3.75, and 3.85, respectively). The risk estimate was increased more than fourfold by seropositivity to 4, 5, or more than 6 specific HP antigens (Omp, HP0305, HyuA, HpaA, CagA, and VacA) relative to seropositivity to 3 or less than 3 antigens. However, the risk was increased to 7.43 when the individuals were seropositive to both Omp and HP0305. This indicate the fact that in East Asia these serological biomarkers are more important than CagA seropositivity alone for predicting the risk of advanced gastric lesions. 62 The detailed information regarding the association of the HP cagA + genotype with the risk of GC in low-, moderate-, and high-incidence areas are indicated in Table 4.

| Wide diversity at the EPIYA and CM motifs of CagA and the risk of GC
CagA protein from various HP strains has a wide diversity at the carboxyl terminus that includes repetitive phosphorylation sites (EPIYA motif). 59 The reports have shown that these repetitive EPIYA sites play a major role in the development of gastrointestinal disorders. 59,63 The diversity of EPIYA motif can influence the CagA structure and its interaction inside the gastro-epithelial cells. 64 Based on the sequences surrounding the EPIYA motif, EPIYA-A, EPIYA-B, EPIYA-C, or EPIYA-D segments have been specified. 65,66 The number and combination of various motifs, which vary according to geographic locations, determine clinical consequences. 11,43,67,68 There have been associations between the different motifs of the CagA protein, the origin country of each sequence type (ST), and the prevalence of gastrointestinal disorders. 11 The EPIYA-ABD type of CagA (East Asiantype) is more virulent compared to the EPIYA-ABC type (Western-type) and has a stronger association with the GC risk. 56,69 In Northeastern Thailand, it has been reported that most strains represent Western-type CagA and are significantly associated with PUD. 70 However, they had no relation with non-ulcer dyspepsia and GC, consistent with previous studies. 51,70 Furthermore, the recent study conducted in East China showed that EPIYA-ABD had no relation with gastric diseases (chronic gastritis and gastric or duodenal ulcerations), and only polymorphism at amino-acids 878 and 879, flanking the EPIYA-A motif, had a significant association with GC. 54 It seems that the number of motifs/segments of EPIYA, especially EPIYA-C is associated with the risk of atrophy, 71 GC, 11,43,72 and intestinal metaplasia. 43 The CagA multimerization motif (CM)-located within the EPIYA-C motif and downstream of the EPIYA-D motif-is responsible for the dimerization of CagA 73 and categorized as Eastern and Western based on CM sequences. 74 The type and number of CM motifs may influence the multimerization of CagA. 73 The CagA protein with two Western CM motifs (vs. that with the Western and Eastern CM motifs) causes cell elongation 74,75 and a greater affinity for SHP-2. 74 The Western CagA proteins with an Eastern CM motif may reduce the multimerization and CagA pathogenicity. 75 Unlike CagA EPIYA motifs, the association of CM motifs with gastrointestinal disorders has not been well studied. However, the presence of one or two Western CM motifs in the absence of an Eastern CM motif was linked with the risk of GC. 76

| Ancestral origin or pathogenic HP factors: Which is more important in GC?
Sablet et al showed that the ancestry of HP is associated with the incidence of GC risk. The Colombian people infected by the hpEurope strains showed more severe tissue damage compared to people infected by hpAfrica1 strains. 5 We have shown that although Iranian HP strains represent European ancestry with almost equal contributions from two ancestral sources (AE1 and AE2), a difference of 2-to 10-times were observed in the GC incidence between Southern and Northwestern-Northern regions of Iran. 16,77 This difference was not consistent with the new prediction model of GC based on bacterial ancestry, thus it has still remained obscure. However, it was proposed that the vacA d1/-i1 genotypes have an excellent potential for differentiating HP strains from the low-and high-incidence regions of GC in Iran. 77 A strong association was found between the HP vacA i1-/ d1-genotypes and the risk of GC in East Azerbaijan region (North-West Iran), where the GC risk is high in males. 78,79 Yamaoka et al have also previously studied the relationship between HP pathogenic factors such as cagPAI, oipA, babA, iceA, and vacA from Colombian people and their clinical consequences. Although cagPAI and oipA were associated with each other, just the oipA gene was inclined toward association with GC. 33 Therefore, they proposed that the oipA in cagA-positive strains might improve the geographic origin diversity which is specified by multi-locus sequence typing (MLST) analysis, challenging the Sablet's model. 80 As a result, the difference in tissue damage might be due to the various status of the oipA gene; therefore, it is required to study the relationship between the pathogenic factors and phylogenic tree by MLST. Also, it was stated that a selection bias was possible in Sablet's model as they merely considered 64 participants who were suffered from cagA + /vacA s1m1 strains and did not focus on those suffering from cagA − genotype. In fact, all cagA − strains were related to hpEurope, and a very low amount of tissue damage was observed in patients infected by these strains. It was stated that the hpEurope strains may manifest low clinical pathogenesis in the absence of cagA. Therefore, this factor could better predict GC risk in comparison with phylogeographic origin. 80 Shiota et al proposed another case from Okinawa in Japan, which challenged the Sablet's model. Many Okinawa strains have been categorized into three hpEastAsia clusters. A main difference between clinical outcomes and phylogeographic origin has been presented,however, the clinical results were dependent on the cagA (cagA − , Western-type, and East Asian-type) status and the vacA m1 or m2 genotypes; GC was prevalent in many clusters which mainly had the cagA of East Asian-type. The MLST analysis revealed that when only the cagA of Western-type or cagA − strains were applied, there was not any difference in the GC prevalence, indicating that cagA better predicts GC in comparison with phylogeographic origin. 80 Nevertheless, Okinawa in Southwestern Japan may show a lower GC incidence than expected from other areas of the country. Thus, the data cannot be generalized to Japan as a whole. In addition, Shiota et al conducted a study in Andean region of Colombia showing that the MLSTbased phylogeographic origin is not enough to specify the GC risk in the hpEurope strains. Hence, the phylogenetic analysis of HP virulence factors is essential for diagnosing clinical findings. 81 However, the phylogenetic tree achieved by conducting MLST did not correlate with the cagA or vacA types, though the vacA s2m2 or cagA − strains were clustered relatively, proposing that vacA and cagA genotypes cannot be assumed as indicating factors for phylogenetic origin based on MLST. They found that the cagA status was associated with a phylogenetic cluster. 80,82 Another study was carried out in Bangladesh which is a country in South Asia with a lower incidence of GC and a higher prevalence of HP infection (60.2%); it was shown that HP strains from Bengali population had two main hpAsia2 and hpEurope populations. The study demonstrated that the hpEurope population was associated with a higher GC risk in comparison to the hpAsia2 population. However, the hpAsia2 strains were strongly associated with an inflammation in the antrum compared to the hpEurope populations. This may be as a result of a higher proportion of less-pathogenic genotypes of the vacA and cagA genes in the hpEurope strains associated with a low histologic damage. Nevertheless, independent of the HP population types, strains with the vacA s1m1i1d1c1 genotype significantly caused greater inflammatory responses and atrophy in the antrum, unlike strains with the vacA s2m2i2d2c2, s2m2i2d2c2, or s1m2i1d1c2 genotypes. Bangladesh strains were categorized into two different HP populations with various genotypes. The low incidence of GC in Bangladesh is possibly because of a great number of less-virulent genotypes, which probably better predict the GC risk in comparison with the ancestral origin of HP strains. Thus, the findings may justify the "Asian enigma" in Bangladesh, with the higher HP infection prevalence and the lower GC risk. 14

| Interaction and co-evolution between HP and host ancestries can be the main determining risk factor for GC
It has been shown that in Western countries, cagA positivity is associated with an increased risk of GC; however, in East Asian countries, although CagA represents EPIYA-D segment and is biologically stronger, its presence does not indicate an elevated risk of GC. 6,8,9,35,55 In this regard, no significant relationship was found between the vacA genotypes, clinical consequences, and histopathologic variations in East Asian countries. 8,83 Clustering analyses have isolated six clinical Colombian strains from the regions with the higher and the lower prevalence of GC on the basis of geographic origin,the European strains showed the higher expression of the BabB, CagA, and VacA pathogenic factors and were correlated with the increased histopathologic lesions. European strains remarkably increased the expression of interleukin 8 in AGS cells compared to the African strains. The African strains considerably promoted apoptosis,while just one strain with the European origin considerably promoted apoptosis. These data proposed that the difference in the expression of virulence factors is resulted by the host-pathogen interactions and might have a remarkable role in carcinogenesis. 10 Increased expression of BabB, CagA, and VacA and decreased apoptosis in the European strains might indicate an increased damage to DNA and suggest GC risk in individuals infected with the European strains compared with those infected with the African strains. 5 Thus, it is assumed that the bacterial pathogenic factors which are determined by the phylogeographic origin of the HP strains might increase the risk of GC. 10 Bhutan, a South Asian country, has a high incidence of GC and HP infection in comparison to the neighborhood countries, like India. According to phylogeographic analyses, Indian strains are of the hpAsia2 and/ or hpEurope types and have Western-type CagA. However, most Bhutan strains are from the East Asian population (though some strains are a mixture of Amerindian and East Asian population types or relate to the European type) and mostly (90%) have East Asian-type CagA (AB'BD and AB'BBD), all strains have vacA s1 type. The difference in the incidence of GC between Bhutan and India is the result of the differences in the phylogeographic origin and virulence factors of HP. HP strains of Bhutan with East Asian-type CagA showed significantly increased inflammatory responses compared to those with Western-type CagA. This justifies the higher GC risk in Bhutan in comparison to India. 84 Kodaman et al studied two Colombian populations from different geographic regions with the same prevalence of HP infection, but an extremely different incidence of GC (150/100 000 vs 6/100 000). They showed that the cagA gene alone cannot be a reason for the difference in the risk of gastric precancerous lesions between these two populations and the interaction between bacterial and host ancestries can be the reason for this difference. The human host and HP genetics are both effective in pathogenesis; however, the impact of human ancestry on histopathology is dependent on the HP ancestry, and vice versa. In this model, African HP ancestry was comparatively benign in people of African ancestry; however, it was detrimental in people with substantial Amerindian ancestry, reflecting the fact that disease risk is modulated by co-evolution. Thus, the disrupted co-evolution between HP and human host could explain why the GC incidence is high in the population of Andes Mountains. When the genomic diversity of the human host and pathogen are analyzed together, it will be more predictive of histopathology compared to the time when just one of them is analyzed. Accordingly, it seems that the co-evolutionary relationships are the main determining risk factors for gastric diseases. 7 Recently, Thorell et al have investigated HP evolution during the recent colonization of Americas. In the United States, strains showing African and European ancestry have remained genetically distinct. In contrast, in Colombia and Nicaragua, bottlenecks and widespread genetic exchange among isolates have led to the rapid emergence of new HP subpopulations. Three outer membrane proteins showing atypical levels of Asian ancestry were found in American strains. In contrast, the alleles that were almost exclusively fixed in South American isolates were found. Thus, the ethnic composition of hosts may play an important role in the colonization of incoming strains. Moreover the high and low incidence areas of GC may show differences in transmission ecology, largely affecting the composition of HP populations. 85 Differences in the host risk genotypes, like pro-inflammatory cytokine polymorphisms, in alliance with the infection of HP strains were associated with the risk of GC in many populations. 86,87 Therefore, genetic variances of both HP strains and the host may demonstrate the difference in GC risk in mountainous areas. 88 In areas with greater GC incidence, HP infection is observed with strains having virulence factors linked to increased inflammatory response, atrophic gastritis, and an elevated GC risk. Hence, HP-host cell interactions via the induction of genetic instability of the host genome (double-stranded breaks, mismatch repair, methylation, etc) F I G U R E 2 A, Interaction and co-evolution between H. pylori and human host in gastric cancer (GC) susceptibility. B, Host responses and risk for GC progression and aberrant miRNA expression result in GC. 89 Moreover, IL-1β significantly amplifies inflammatory response during HP infections 90,91 and IL-1β-31 CT and TT may increase possibility of the infection by HP. 92 The IL-1β-511T, IL-1RN*2/*2, IL-10 (haplotype ATA/ATA), and TNF-α-308A genotypes were significantly associated with an increased risk of non-cardia GC with HP infection. 93,94 Furthermore, the cagA(+)/vacA s1(+)/IL-1β-511T genotype increased the risk of severe gastric anomalies. 95 These findings, in turn, emphasize that the cytokine gene polymorphisms should be considered in GC host-pathogen interaction models Figure 2.

| FUTURE DIRECTIONS
In recent years, it has been proposed two new models to predict GC; the phylogeographic origin of HP strains and a disrupted co-evolution between HP and its human host. As discussed above, several studies have challenged the first new model and proposed the virulence factors as the best predictors of GC. 33,77,80 In addition, the strains from populations of East Asian countries such as Korea and Japan, with an extremely high incidence of GC, belonged to the hspEAsia group. 15,17 In these populations, both the host and bacterial ancestries have remained relatively pure, 15,17,96,97 reflecting the fact that the disrupted co-evolution is unlikely to occur, challenging the Kodaman's model. Therefore, these models and earlier ones based on putative pathogenic factors of the HP may not fully justify differences in the incidence of GC.
In recent years, the genetic studies of complex phenotypes are shifting to include the lower frequencies of rare variants, which is dependent on 'common disease-rare variant' hypothesis and can only be allowed by an advanced sequencing technology, namely next-generation sequencing (NGS). This postulates that the multiple rare variants which represent the larger effect sizes are the important determinants of disease heritability. In contrast, genome-wide association studies (GWASs) are primarily examining the 'common disease-common variant' hypothesis, which is based on the principle of linkage disequilibrium. Therefore, exome sequencing combined with genotyping arrays is probably the best approach over the next decade to explore cancer-risk loci. [98][99][100] Notably, the new findings also support the role of ancestry-specific germline alteration in contributing to the ethnic/population differences in cancer risk. 101 Moreover, the high and low incidence areas of GC have shown differences in transmission ecology, largely affecting the composition of HP populations. 85 As a new hypothesis, it is proposed that any high-risk population may have its own specific risk loci (or variants) as well as new HP strains with national/maybe regional gene pools that should be considered. The latter is seen in the Americas where the rapid evolution of distinct HP subpopulations has been occurred. 85 It is therefore proposed that the deep sequencing of both HP and its human host is simultaneously performed in GC patients and agesex-matched controls from high-risk areas. The expression and functional activities of the identified new determinants of GC must then be assessed and matched with human and pathogen ancestry because some risk loci are ancestry-specific. 98,101 In addition, potential study-level covariates and moderator variables (eg physical conditions, life styles, gastric microbiome, etc) linked to causal relationships, and their impact, should be recognized and controlled.

ACKNOWLEDGMENTS
This study was supported by the National Institute for Medical Research Development (NIMAD) Grant No. 958117. The supporter had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. There was no additional external funding received for this study.