Are Helicobacter pylori highly cytotoxic genotypes and cardia gastric adenocarcinoma linked? Lessons from Iran

Abstract

BACKGROUND:

Although the most extensive studies revealed the role of H. pylori VacA and CagA toxins in the development of gastric adenocarcinoma, the magnitude of this association and the correlations of vacA mosaicism and cagA status with cardia gastric adenocarcinoma (CGA) still remain controversial.

OBJECTIVE:

We aimed to examine the linkage of H. pylori highly cytotoxic genotypes to CGA in Iranian populations as a model.

METHODS:

A total of 601 Iranian patients were enrolled. Biopsies were cultured, genotyped, and anatomically and histologically classified.

RESULTS:

The vacA c1 genotype, but not cagA status, showed a strong association with the risk of both CGA and non-cardia adenocarcinoma (NCGA), whether the controls were non-tumors, as those with either non-atrophic gastritis or peptic ulcerations, (the OR (95%CI) was 14.11 (4.91–40.52) and 9.59 (4.06–22.65), respectively) or those with NAG (the OR (95%CI) was 10.71 (3.49–32.82) and 8.11 (3.26–20.16), respectively). The vacA c1/cagA ${}^{\mathrm{+}}$ genotype was significantly associated with an increased risk of NCGA, whether the controls were non-tumors or those with NAG; the adjusted risk was 4.706 (1.41–15.67) and 4.85 (1.42–16.51), respectively.

CONCLUSIONS:

The H. pylori vacA c1 genotype, but not cagA status, might be the first important bacterial biomarker for predicting the cardia adenocarcinoma risk in male patients aged $⩾$ 55 in Iran.

1. Introduction

Although the incidence of stomach cancer is continually decreasing [1]; however, it remains as the fifth common cancer worldwide (constitutes 6.8% of the global cancer in the world) and the third cancer-related mortality (8.8%) [2]. Gastric adenocarcinoma (GA) accounts for approximately 90% of all stomach cancers that can be classified based on the anatomic site and histologic type of the tumor. Anatomically, there are two subtypes of GA, cardia (proximal: the upper part of the stomach adjoining the esophagus) gastric adenocarcinoma (CGA) and non-cardia (distal: the middle and distal parts of the stomach) gastric adenocarcinoma (NCGA). According to the Laurén classification, GA consists of two main histologic variants, intestinal and diffuse [3].

Several reports indicate that the incidence of CGA has steadily increased in the last several decades, especially in the Western countries [4, 5, 6]. This is in contrast to the decline in the incidence of NCGA worldwide [4, 7, 8]. NCGA is seen in higher rates in high-risk geographic regions, such as Eastern Europe, East Asia, Central, and South America [9], whereas the incidence of CGA is much higher in low risk populations [10], such as western populations [7, 8]. Intestinal-type gastric adenocarcinoma (IGA) is more common in high-risk areas, such as East Asia, East Europe, and South and Central America [11], whereas diffuse-type gastric adenocarcinoma (DGA) is uniformly distributed all over the world [12]. This difference in the incidence of GA rates based on the anatomic sub-site and histologic features may be due to differences in the prevalence of Helicobacter pylori (H. pylori) infection and differences in the genetic diversity of H. pylori strains. Many clinical and epidemiologic studies have shown that the colonization of the human stomach with H. pylori is as an important risk factor for NCGA development [13, 14] such that 77% of the cases of NCGA are caused by this bacterial infection [13]. In contrast, CGA has negative [14], positive [15] or no association with H. pylori infection [13, 16, 17].

Interestingly, in a meta-analysis study, the presence of H. pylori was related to an increased CGA risk in high-risk settings, but an inverse association existed in low-risk settings [18]. For example, several studies of Asian countries have shown a high positive correlation between H. pylori infection and CGA, whereas some studies of Western countries have found no association or an inverse association [13, 19]. It has been reported that H. pylori colonization is linked to both types, IGA and DGA [13, 20, 21], but in some studies have not found such a relationship between DGA and H. pylori infection; and more often associated with genetic abnormalities. Heterogeneity among H. pylori virulence genes may be an important factor in creation of the different types of GA. H. pylori heterogeneity has been strongly investigated the past two decades to identify possible virulence factors associated with the GA.

Polymorphic variations in H. pylori vacA gene result in differences in the levels of the cytotoxicity and pathogenicity. Variations in the signal (s) region (encodes a part of the N-terminal region of the 88-kDa mature protein), the middle (m) region (encodes a part of 55-kDa C-terminal (p ${}^{55}$ ) domain), the intermediate (i) region (situated between s and m regions, which has an important role in vacuolating activities), and the deletion (d) region (located between m and i regions) of vacA play a major role in increasing the risk of gastric diseases [22, 23, 24, 25, 26]. Recently, we have identified the fifth polymorphic site located at the 3’-end region of the H. pylori vacA gene and termed c-region. Two allelic variants of this region were denoted c1 (with 15 bp deletion) and c2 (no deletion). The c1-type association was independent of, and larger than, the associations of the m-, i-, and d-type of vacA or cagA status with GC [22]. cagA, which encodes an oncogenic CagA, is considered a virulence marker [27]. cagA-positive strains have strong association with GA risk [22, 28]. Several studies have shown that, compared to cagA ${}^{\mathrm{-}}$ strains, infection with strains carrying cagA gene is linked to a higher risk of NCGC [28, 29, 30, 31, 32, 33]. Other studies revealed a significant negative correlation between the presence of cagA-positive strains and the risk of CGC [14, 17, 34]. Moreover, some studies showed no correlation between the presence of cagA-positive strains and the risk of CGC or their protective effect on the incidence of CGC [32, 34, 35]. There was a significant relationship between the attendance of vacA and the risk of NCGA, but not CGA [17, 36]. It has been reported that the risk variation in the incidence of histologic sub-type of GA may also reflect differences between cagA ${}^{\mathrm{+}}$ and cagA ${}^{\mathrm{-}}$ H. pylori strains [37, 38, 39]. CagA and VacA were also associated with an increased risk of both IGA and DGA [36]. Although the most extensive studies revealed the role of H. pylori VacA and CagA toxins in the development of GC [17, 22, 23, 24, 40, 41, 42, 43], the magnitude of this association and the correlations of vacA genotypes and cagA status with the anatomic origin and histologic features of the tumor still remain controversial. Iran is a country with the higher rate of the outbreak of H. pylori infection (69%) [44]; and with a high incidence of gastric adenocarcinoma, ranked fourth in Asia [45, 46]. The highest rates of CGC have occurred in Central Asia countries, such as Iran [47]. Moreover, the role of bacterial genotypes in the incidence of GC has been much stressed in Iran [22, 24, 42, 48]. Therefore, we aimed to examine the linkage of H. pylori highly cytotoxic genotypes to cardia adenocarcinoma in Iranian populations as a model.

Table 1

Oligonucleotide primers used for PCR

Genes	Primers	Sequences (5’ $\to$ 3’)	Size of PCR products (bp)	Optimized annealing
				temperature ( ${}^{\circ }$ C)
16 S rDNA	HP1	GCA ATC AGC GTC AGT AAT GTT C	519	56
	HP2	GCT AAG AGA TCA GCC TAT GTC C
vacA
c1/-c2	c1-F	ATC ATY SGT TAT GRH AAT GTT TCT	c1: 600–700	55
	R-nd	TTA TGC TCT AAA CTG GCT A
	c2-F	ATT ATA ATT TAG TAG GAG TGC AAG G	c2: 600–700	55
	R-nd	TTA TGC TCT AAA CTG GCT A
cagA	CAG1	ACC CTAGTC GGT AAT GGG TTA	591–856	50
	CAG2	GTA ATT GTC TAG TTT CGC
cag PAI empty site	Cp1	ACTTTCACGCCCTTTCCCTCC	593	51
	Cp2	TTGCATGCGTTATTATTTCAC

2. Materials and methods

2.1 Subjects

A total of 601 patients from different regions of Iran participated in this study. They were referred to endoscopy units in hospitals and physicians’ offices across the country. Gastric biopsy samples were obtained from patients with non-atrophic gastritis (NAG), peptic ulcerations (PUs) and GA. The study was approved by the research Ethics Committee of GCRC. All patients signed written informed consent.

2.2 Endoscopy and biopsy sampling and cultivation

Gastric biopsies were taken from the antrum and/or the corpus, of which one was used for rapid urease test and another for cultivation. Tumor biopsies were also taken and examined histologically. The tumors originated from the lower one-third of the esophagus, above the $Z$ -line, were considered as esophageal adenocarcinoma, but not CGA, and excluded from the analysis. When determining the anatomic origin of a tumor was impossible by endoscopist, it was categorized as unspecified. The biopsies were then cultured on selective Brucella agar plates under micro-aerobic conditions. Brucella agar plates (Merck, Germany) contained 10% blood, vancomycin (10 mg/mL; Zakaria, Iran), trimethoprim (5 mg/mL; MP Biomedicals, France), and amphotericin B (4 mg/mL; Bristol-Myers Squibb, USA). The bacterial colonies were identified based on Gram’s staining, typical cell morphology, and positive reactions to catalase, oxidase, and urease, as well as PCR amplification of H. pylori-specific 16S rDNA.

2.3 Histologic examination

One of the biopsy specimens was fixed in 10% formalin and embedded in paraffin, and then the tissue sections for histopathologic examinations were prepared. The Lauren’s classification was performed based on morphology and mucin staining, and the histo-morphologic architecture of the tumors was expressed as intestinal-type or diffuse-type gastric adenocarcinomas. Histopathologic assessments were also applied regarding the updated Sydney classification system [49].

2.4 DNA extraction, primers and PCR conditions and genotyping

Total DNAs were extracted from H. pylori strains with the Genomic DNA purification kit (Fermentas, UK) based on the manufacturer’s protocol. PCR amplification was performed in 30 $\mu$ L reaction volume, containing 3 $\mu$ L of 10X PCR buffer (CinnaGen, Iran), 1 mM/L MgCl2, 2U of Taq DNA polymerase (CinnaGen Co., Iran), 200 $\mu$ M/L dNTP (CinnaGen Co., Iran), 0.5 $\mu$ M of each primer, and 25 ng of bacterial DNA. The utilized primers for PCR are summarized in Table 1. The PCR conditions were at 96 ${}^{\circ }$ C for 180 s; then 35 cycles of 96 ${}^{\circ }$ C for 40 s (denaturation), an optimized annealing temperature for each allele (Summarized in Table 1) for 40 s, and 72 ${}^{\circ }$ C for 40 s (extension), and a final incubation cycle at 72 ${}^{\circ }$ C for 7 min (final extension). Finally, PCR products were separated by gel electrophoresis, stained with ethidium bromide, and visualized using a UV transilluminator. The band sizes corresponding to each gene and allele are listed in Table 1. Escherichia DH5 $\alpha$ strain and deionized water were used as negative controls.

Table 2

Characteristics of patients enrolled in this study

Characteristics		No. of patients (%)
Age groups
	$⩾$ 55	123/282 (43.6)
	$<$ 55	157/282 (55.7)
	No data	2/282 (0.7)
Sex groups
	Males	178/282 (63.1)
	Females	104/282 (36.9)
Types of gastroduodenal diseases
Control		194/282 (68.8)
	Non-atrophic gastritis	136/194 (70.1)
	Peptic ulcer	58/194 (29.9)
Case		88/282 (31.2)
	Cardia gastric adenocarcinoma	38/88 (43.2)
	Non-cardia gastric adenocarcinoma	46/88 (52.3)
	Unspecified	4/88 (4.5)
	Intestinal-type adenocarcinoma	57/88 (64.8)
	Diffuse-type adenocarcinoma	25/88 (28.4)
	Mucin producing-type adenocarcinoma	2/88 (2.3)
	Signet ring-type adenocarcinoma	2/88 (2.3)
	Adenocarcinoma, poorly differentiated	1/88 (1.1)
	Adenocarcinoma, moderate differentiation	1/88 (1.1)
Total		282/282 (100)

For confirmatory purposes, the amplified fragments of each gene/allele from thirteen strains were purified and sequenced with both forward and reverse primers using a BigDye technology on an ABI3700XL DNA sequencer (Applied Biosystems). The BLAST program (http://www.ncbi.nlm.nih.gov) also was utilized to compare the nucleotide sequences with those in GenBank.

2.5 Statistical analysis

The Fisher’s exact and Chi-square tests were used to evaluate significant differences between the frequencies of each gene/allele and genotype combination and anatomic site and histologic variants of the tumor. To determine the effect of each genotype on the risk of CGA, NCGA, and the different histologic types of GA, we performed simple logistic regression analysis by the Enter method and multiple logistic regression analysis by the Forward Stepwise LR (Likelihood Ratio) method with adjustment for a threshold age of $⩾$ 55 years and sex. All the $P$ -values were two-sided, and $P$ -values $<$ 0.05 were indicated as statistically significant. The SPSS version 19 was used for data analysis. The proportions of associations might be incorrectly classified as significant due to multiple testing. We therefore estimated the false discovery rate (FDR) and its analog the $Q$ -value among the associations tested by using the $Q$ -value package in $R$ version 3.1.1.

3. Results

3.1 Classification of patients

The present study is an Iranian population-based case-control study. Of the 111 case patients, 56 had NCGA, 41 CGA, and 6 both the types of CGA and NCGA, and 8 with a pathologic diagnosis of “no-tumor”, “MALT lymphoma” or “invasive squamous cell-type carcinoma” were excluded. Control subjects included 490 with NAG and PUs. Overall, 390/593 patients (65.8%) were positive for H. pylori infection by rapid urease test. Due to some severe contaminations in the stomach of patients, only 282 H. pylori strains were successfully obtained from the biopsy cultures of gastric antrum and body mucosa, and were genotyped. They included 88 cases (38 with CGA, 46 with NCGA, and 4 with both the types of CGA and NCGA) and 194 controls (136 with NAG and 58 with PUs). The characteristics of patients enrolled in this study are summarized in Table 2.

3.2 Prevalence of H. pylori vacA c-region genotypes and cagA gene

In general, the vacA c-region genotypes, and the status of cagA gene were determined in 98.2% (277) and 98.6% (278) of strains, respectively. Neither vacA c-regions nor cagA gene could be amplified in five (1.8%) and four (1.4%) strains, respectively. The frequency of vacA c1 was 99/282 (35.1%), c2 163/282 (57.8%), c1c2 15/282 (5.3%) (6 controls, 7 those with CGA, and 2 those with NCGA), cagA ${}^{\mathrm{+}}$ 178/282 (63.1%), and cagA ${}^{\mathrm{-}}$ 100/282 (35.5%) (Table 3). The samples which had mixed infection with multiple strains in the same patient and showed the vacA c1c2 genotypes were excluded from the final analysis.

Table 3

Frequencies (%) of H. pylori vacA c-region and cagA statues

Genotypes		Frequency (N)	Percent (%)
vacA c region		277/282	98.2
	c1	99/282	35.1
	c2	163/282	57.8
	c1 and c2	15/282	5.3
	Undetectable	5/282	1.8
cagA status		278/282	98.6
	cagA ${}^{\mathrm{+}}$	178/282	63.1
	cagA ${}^{\mathrm{-}}$	100/282	35.5
	Undetectable	4/282	1.4

3.3 Association between age and sex and the anatomic origin and histologic type of the tumors

In GC group, more than 80% of patients were male and had 55 yr. of age and older. Statistical analysis showed also a significant correlation between sex or age and NCGA and CGA as well as IGA and DGA, whether the controls were non-tumors or those with NAG ( $P<$ 0.05; Table 4). Non-tumors involved the patients with either NAG or PUs, and those with benign tumors were not welcome. The association of age $⩾$ 55 with DGA was much larger than its association with IGA, whether the controls were non-tumors (ORs $=$ 29.8 vs. 8.37, $P=$ 0) or those with NAG (ORs $=$ 31.9 vs. 8.97, $P=$ 0; Table 4).

3.4 Associations of H. pylori vacA c-region genotypes and cagA status with the risk of NCGA and CGA as well as IGA and DGA

Table 6 describes the risk of GA with respect to H. pylori genotypes by simple logistic regression analysis, where the controls were non-tumors. The frequency of the c1-type of vacA was higher in patients with CGA (73.3%) and NCGA (68.2%) than in non-tumors (23.4%). The analysis confirmed the associations of this genotype with an increased risk of both CGA and the NCGA; the OR (95% CI) was 9.01 (3.74–21.70) and 7.02 (3.41–14.44), respectively. There was no significant difference in the prevalence of the strains carrying cagA gene between the patients with either CGA or NCGA and control groups (non-tumors). The vacA c1/cagA ${}^{\mathrm{+}}$ genotype was significantly associated with an increased risk of NCGA; the OR (95% CI) was 5.27 (1.79–15.50). As illustrated in Table 6, the results of simple logistic regression analysis for the risk of GA, where the controls were those with NAG, demonstrated that the vacA c1 and vacA c1/cagA ${}^{\mathrm{+}}$ genotypeswere remarkably associated with an increased risk of both CGA and the NCGA; the OR (95% CI) was 8.50 (3.45–20.96) and 6.62 (3.13–14.02), respectively, for c1 and 3.57 (1.20–10.55) and 6.07 (2.02–18.22), respectively, for vacA c1/cagA ${}^{\mathrm{+}}$ . No significant correlation was found between cagA statues and the risk of CGA or NCGA.

Following further analysis was conducted based on the histologic features of the tumor expressed as IGA and DGA. The frequency of the vacA c1 genotype in patients with IGA (70.6%) and DGA (77.3%) was significantly higher than in those with non-tumors (23.4%) (Table 6). It was significantly associated with the risk of IGA and DGA by simple logistic regression analysis; the OR (95% CI) was 7.87 (3.93–15.72), and 11.14 (3.88–31.98), respectively. The cagA ${}^{\mathrm{+}}$ genotype was not independently associated with the risk of IGA and DGA (Table 6). The presence of both the vacA c1 and cagA ${}^{\mathrm{+}}$ genotypes further increased the risk of DGA (OR $=$ 13.97; 95% CI, 1.70–114.77), but not IGA (OR $=$ 4.07, 95% CI, 1.63–10.13). The results of simple logistic regression analysis for the risk of GA, where the controls were those with NAG, demonstrated that the vacA c1 and vacA c1/cagA ${}^{\mathrm{+}}$ genotypes were associated with an increased risk of IGA and DGA (Table 6); the OR (95% CI) was 7.42 (3.60–15.28) and 10.15 (3.59–30.7), respectively for vacA c1 and 4.68 (1.83–11.95) and 16.07 (1.93–133.51), respectively for vacA c1/cagA ${}^{\mathrm{+}}$ . There was not a significant association between the cagA statues and the risk of both IGA and DGA.

Finally, the multiple logistic regression analysis revealed that the vacA c1 genotype was independently and significantly associated with the age- and sex-adjusted risk for CGA, NCGA, IGA, and DGA, whether the controls were non-tumors (the OR (95% CI) was 14.11 (4.91–40.52), 9.59 (4.06–22.65), 11.91 (4.99–28.45), and 16.93 (4.97–57.68), respectively) or those with NAG (the OR (95% CI) was 10.71 (3.49–32.82), 8.11 (3.26–20.16), 9.56 (3.86–23.63), and 11.22 (3.077–40.94), respectively) (Table 7). The multiple logistic regression analysis also showed that the vacA c1/cagA ${}^{\mathrm{+}}$ genotype was significantly correlated with the age- and sex-adjusted risk for NCGA, IGA, and DGA, where the controls were non-tumors (the OR (95% CI) was 4.706 (1.41–15.67), 3.92 (1.32–11.64), and 12.37 (1.29–118.33), respectively) and with the risk for the NCGA and the IGA, where the controls were those with NAG (the OR (95% CI) was 4.85 (1.42–16.51), and 3.93 (1.28–12.03), respectively) (Table 7).

Table 4

Association between age and sex and the anatomic origin and histologic type of the tumors

Genotypes		Cardia gastric adenocarcinoma					Non-cardia gastric adenocarcinoma					Intestinal type adenocarcinoma					Diffuse type adenocarcinoma
	Control ${}^{\text{a}}$ No.(%)	Case No.(%)	P value	OR ${}^{\text{b}}$	95% CI ${}^{\text{c}}$	Total	Case No.(%)	P value	OR	95% CI	Total	Case No.(%)	P value	OR	95% CI	Total	Case No.(%)	P value	OR	95% CI	Total
Gastric adenocarcinomas vs. non-tumors
Sex
Male	111(56.9)	31(81.6)	0.006	3.35	1.40–7.98	142(60.9)	35(76.1)	0.019	2.40	1.15–5.018	146(60.6)	42(75.0)	0.016	2.27	1.16–4.42	153(61.0)	21(84.0)	0.015	3.97	1.31–12.00	132(60.0)
Female	84(43.1)	7(18.4)	1 (ref)	1 (ref)	1 (ref)	91(39.1)	11(23.9)	1 (ref)	1 (ref)	1 (ref)	95(39.4)	14(25.0)	1 (ref)	1 (ref)	1 (ref)	98(39.0)	4(16.0)	1 (ref)	1 (ref)	1 (ref)	88(40.0)
Age, y
$⩾$ 55	54(27.8)	32(84.2)	0.00	13.82	5.47–34.93	86(37.1)	34(75.6)	0.00	8.01	3.79–16.94	88(36.8)	42(76.4)	0.00	8.37	4.17–16.81	96(38.6)	23(92.0)	0.00	29.81	6.79–130.79	77(35.2)
$<$ 55	140(72.2)	6(15.8)	1 (ref)	1 (ref)	1 (ref)	146(62.9)	11(24.4)	1 (ref)	1 (ref)	1 (ref)	151(63.2)	13(23.6)	1 (ref)	1 (ref)	1 (ref)	153(61.4)	2(8.0)	1 (ref)	1 (ref)	1 (ref)	142(64.8)
Gastric adenocarcinomas vs. non-atrophic gastritis
Sex
Male	64(46.7)	31(81.6)	0.00	5.05	2.08–12.25	95(54.3)	35(76.1)	0.001	3.62	1.70–7.73	99(54.1)	42(75.0)	0.00	3.42	1.71–6.83	106(54.9)	21(84.0)	0.002	5.98	1.95–18.36	85 (52.5)
Female	73(53.3)	7(18.4)	1 (ref)	1 (ref)	1 (ref)	80(45.7)	11(23.9)	1 (ref)	1 (ref)	1 (ref)	84(45.9)	14(25.0)	1 (ref)	1 (ref)	1 (ref)	87(45.1)	4(16.0)	1 (ref)	1 (ref)	1 (ref)	77(47.5)
Age, y
$⩾$ 55	36(26.5)	32(84.2)	0.00	14.81	5.72–38.36	68(39.1)	34(75.6)	0.00	8.58	3.93–18.71	70(38.7)	42(76.4)	0.00	8.97	4.32–18.61	78(40.8)	23(92.0)	0.00	31.944	7.16–142.34	59(36.6)
$<$ 55	100(73.5)	6(15.8)	1 (ref)	1 (ref)	1 (ref)	106(60.9)	11(24.4)	1 (ref)	1 (ref)	1 (ref)	111(61.3)	13(23.6)	1 (ref)	1 (ref)	1 (ref)	113(59.2)	2(8.0)	1 (ref)	1 (ref)	1 (ref)	102(63.4)

a: Non atrophic gastritis (NAG); b: Odds ratio; c: Confidence interval.

Table 5

Risk estimates for CGA, NCGA, IGA, and DGA in relation to H. pylori vacA c-region genotypes and cagA status in a simple logistic regression analysis, where the controls were non-tumors

Genotypes		Cardia gastric adenocarcinoma					Non-cardia gastric adenocarcinoma					Intestinal-type adenocarcinoma					Diffuse-type adenocarcinoma
	Control ${}^{\text{a}}$ No.(%)	Case No.(%)	P-value	OR ${}^{\text{b}}$	95% CI ${}^{\text{c}}$	Q-value ${}^{\text{d}}$	Case No.(%)	P-value	OR	95% CI	Q-value	Case No.(%)	P-value	OR	95% CI	Q-value	Case No.(%)	P-value	OR	95% CI	Q-value
vacA c-region
c1	43(23.4)	22(73.3)	9.21e-07 ${}^{\text{𝐞}}$	9.01	3.74–21.70	2.76e-06	30(68.2)	1.13e-07	7.02	3.41–14.44	3.39e-07	36(70.6)	5.22e-09	7.87	3.93–15.72	1.56e-08	17(77.3)	7.32e-06	11.14	3.88–31.98	2.19e-05
c2	141(76.6)	8(26.7)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	14(31.8)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	15(29.4)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	5(22.7)	1 (ref)	1 (ref)	1 (ref)	1 (ref)
cagA status
cagA ${}^{\mathrm{+}}$	127(65.5)	20(55.6)	2.57e-01	0.659	0.32–1.35	2.57e-01	28(63.6)	8.18e-01	0.923	0.46–1.82	8.18e-01	33(61.1)	5.54e-01	0.82	0.44–1.54	5.54e-01	16(64.0)	8.84e-01	0.93	0.33–2.23	8.84e-01
cagA ${}^{\mathrm{-}}$	67(34.5)	16(44.4)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	16(36.4)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	21(38.9)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	9(36.0)	1 (ref)	1 (ref)	1 (ref)	1 (ref)
Genotype combination
vacA c1/  cagA ${}^{\mathrm{+}}$	38(39.2)	12(66.7)	3.64e-02	3.105	1.07–8.97	5.46e-02	17(77.3)	2.46e-03	5.27	1.79–15.50	3.70e-03	21(72.4)	2.49e-03	4.07	1.63–10.13	3.74e-03	9(90.0)	1.41e-02	13.97	1.70–114.77	2.11e-02
vacA c2/  cagA ${}^{\mathrm{-}}$	59(60.8)	6(33.3)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	5(22.7)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	8(27.6)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	1(10.0)	1 (ref)	1 (ref)	1 (ref)	1 (ref)

Table 6

Risk estimates for CGA, NCGA, IGA, and DGA in relation to H. pylori vacA c-region genotypes and cagA status in a simple logistic regression analysis, where the controls were those with non-atrophic gastritis

Genotypes		Cardia gastric adenocarcinoma					Non-cardia gastric adenocarcinoma					Intestinal type adenocarcinoma					Diffuse type adenocarcinoma
	Control ${}^{\text{a}}$ No.(%)	Case No.(%)	P-value	OR ${}^{\text{b}}$	95% CI ${}^{\text{c}}$	Q-value ${}^{\text{d}}$	Case No.(%)	P-value	OR	95% CI	Q-value	Case No.(%)	P-value	OR	95% CI	Q-value	Case No.(%)	P-value	OR	95% CI	Q-value
vacA c-region
c1	32(24.4)	22(73.3)	3.28e-06 ${}^{\text{𝐞}}$	8.50	3.45–20.96	9.85e-06	30(68.2)	7.48e-07	6.62	3.13–14.02	2.24e-07	36(70.6)	5.31e-08	7.42	3.60–15.28	1.59e-07	17(77.3)	1.74e-05	10.15	3.59–30.78	5.24e-05
c2	99(75.6)	8(26.7)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	14(31.8)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	15(29.4)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	5(22.7)	1 (ref)	1 (ref)	1 (ref)	1 (ref)
cagA status
cagA ${}^{\mathrm{+}}$	80(58.8)	20(55.6)	7.23e-01	0.875	0.41–1.83	7.23e-01	28(63.6)	5.71e-01	1.22	0.60–2.47	5.71e-01	33(61.1)	7.72e-01	1.10	0.57–2.09	7.72e-01	16(64.0)	8.18e-01	1.24	0.51–3.01	8.18e-01
cagA ${}^{\mathrm{-}}$	56(41.2)	16(44.4)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	16(36.4)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	21(38.9)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	9(36.0)	1 (ref)	1 (ref)	1 (ref)	1 (ref)
Genotype combination
vacA c1/  cagA ${}^{\mathrm{+}}$	28(35.9)	12(66.7)	2.13e-02	3.57	1.20–10.55	3.19e-02	17(77.3)	1.30e-03	6.07	2.02–18.22	1.95e-03	21(72.4)	1.22e-03	4.68	1.83–11.95	1.83e-03	9(90.0)	2.46e-03	16.07	1.93–133.51	3.70e-03
vacA c2/  cagA ${}^{\mathrm{-}}$	50(64.1)	6(33.3)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	5(22.7)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	8(27.6)	1 (ref)	1 (ref)	1 (ref)	1 (ref)	1(10.0)	1 (ref)	1 (ref)	1 (ref)	1 (ref)

Table 7

Age- and sex-adjusted risk for CGA, NCGA, IGA, and DGA in relation to H. pylori vacA c-region genotypes and cagA status in a multiple logistic regression analysis

	Cardia gastric adenocarcinoma				Non-cardia gastric adenocarcinoma				Intestinal-type adenocarcinoma				Diffuse-type adenocarcinoma
Genotypes	P-value	OR ${}^{\text{a}}$	95% CI ${}^{\text{b}}$	Q-value ${}^{\text{c}}$	P-value	OR	95% CI	Q-value	P-value	OR	95% CI	Q-value	P-value	OR	95% CI	Q-value
Gastric adenocarcinoma vs. non-tumors
vacA c1 vs.  vacA c2	8.73e-07 ${}^{\text{𝐝}}$	14.11	4.91–40.52	2.62e-06	2.52e-07	9.59	4.06–22.65	7.56e-07	2.41e-08	11.91	4.99–28.45	7.24e08	6.04e-06	16.93	4.97–57.68	1.81e-05
cagA ${}^{\mathrm{+}}$ vs.  cagA ${}^{\mathrm{-}}$	3.63e-02	0.39	0.16–0.94	5.44e-02	1.81e-01	–	–	1.81e-01	8.40e-02	–	–	8.40e-02	1.86e-01	–	–	1.86e-01
vacA c1/cagA ${}^{\mathrm{+}}$ vs.  vacA c2/cagA ${}^{\mathrm{-}}$	9.52e-02	–	–	9.52e-02	1.16e-02	4.706	1.41–15.67	1.74e-02	1.37e-02	3.92	1.32–11.64	2.06e-02	2.89e-02	12.37	1.29–118.33	4.34e-02
Gastric adenocarcinoma vs. non-atrophic gastritis
vacA c1 vs.  vacA c2	3.30e-05	10.71	3.49–32.82	9.90e-05	6.61e-06	8.11	3.26–20.16	1.98e-05	1.00e-06	9.56	3.86–23.63	3.01e-06	2.50e-04	11.22	3.077–40.94	7.50e-04
cagA ${}^{\mathrm{+}}$ vs.  cagA ${}^{\mathrm{-}}$	1.32e-01	–	–	1.32e-01	5.24e-01	–	–	5.24e-01	2.98e-01	–	–	2.98e-01	3.70e-01	–	–	3.70e-01
vacA c1/cagA ${}^{\mathrm{+}}$ vs.  vacA c2/cagA ${}^{\mathrm{-}}$	1.30e-01	–	–	1.32e-01	1.15e-02	4.85	1.42–16.51	1.72e-02	1.61e-02	3.93	1.28–12.03	2.41e-02	3.61e-02	11.57	1.172–114.26	5.41e-02

a: Odds ratio; b: Confidence interval; c False discovery rate-adjusted P-value; d: Boldface data indicate statistically significant results.

4. Discussion

In the present study, 65.8% of patients were positive for H. pylori infection. The frequency of the c1-type of vacA was higher in patients with CGA (73.3%) and NCGA (68.2%) than in either non-tumors (23.4%) or those with NAG (24.4%). Eventually, after being adjusted for confounding factors, the multiple logistic regression analysis revealed that the c1 genotype had a strong correlation with an enhanced risk of CGA and NCGA, whether the controls were non-tumors (ORs 14.11 and 9.59, respectively) or those with NAG (ORs 10.71 and 8.11, respectively). These findings provide the first evidence for the determinant role of the H. pylori vacA c1 genotype in the causing of both cardia and non-cardia adenocarcinomas in male patients aged $⩾$ 55 in Iran. Recently, we have proposed that the H. pylori vacA i1 (OR 37.52) and d1 (OR 7.17) genotypes might be important determinants of non-cardia cancer risk in Ardabil, a very high-risk area in Northwestern Iran. The m1, not independently, but in combination might further define GC risk [42]. Furthermore, Shakeri et al. conducted a study on 272 cases of GA and 524 controls in Northeastern Iran, to assess the associations of seropositivity to H. pylori antigens with GA. They showed that VacA was related to an increased risk of NCGA (OR 2.8), but not CGA [17]. In one study, da Costa et al. observed that the frequencies of vacA s1, vacA m1, vacA s1m1, vacA s2m2, and cagA were 85.5%, 70.1%, 66.6%, 11.1%, and 64.9%, respectively in patients with GC. The distribution of these genotypes was similar in tumors from both regions cardia and non-cardia; so no relationship was found [50]. In the current study, the vacA c1 genotype also was linked to an elevated risk of both IGA and DGA, whether the controls were non-tumors (adjusted ORs 11.91 and 16.93, respectively) or those with NAG (adjusted ORs 9.56 and 11.22, respectively). Similar results were obtained in Ardabil, where the i1- and d1-types of vacA were linked to increased risks of intestinal- (OR 14.04) and diffuse-type (OR 7.71) adenocarcinomas, respectively [42] In a recent study, Figura et al. examined 226 H. pylori colonies from 15 patients with IGC and 13 patients with DGC; and tested the associations of polymorphisms of pathogenic cagA and vacA with histopathologic variables. They reported that 80.95% of vacA s1/m2-carrying strains were significantly isolated from DGC cases ( $P<$ 0.001) [51]. These findings reflect the fact that some H. pylori genotypes might also be important for the development of the histologic types of GC.

Our study showed no significant statistical correlation between the cagA ${}^{\mathrm{+}}$ genotype and the risk of CGA, whether the controls were non-tumors (OR 0.659) and or those with NAG (OR 0.875). These results correspond with previous studies [32, 36, 52, 53]. For example, Bornschein et al. study on 152 patients with GC (73 with CGC and 79 with NCGC) showed that the prevalence of CagA status was similar in proximal and distal GC (77.2 vs. 84.6%) as well as in intestinal- and diffuse-type GC (82.6 vs. 79.2%) by serologic assessment [53]. Another study from Sweden also showed no association between CagA antibodies and cardia cancer (OR 1.0) [16] However, other studies demonstrated a significant inverse relationship between cagA ${}^{\mathrm{+}}$ strains and the development of CGA, which reduces the risk for this type of tumor [14, 34]. In contrast, in Shakeri et al. study, CagA seropositivity was associated with an increased risk of both CGA (OR 1.9) and NCGA (OR 3.4) [17].

In the present study, no significant difference was found in the prevalence of strains carrying cagA gene between patients with NCGA (63.6%) and either non-tumors (65.5%) or those with NAG (58.8%). In some studies, infection with cagA ${}^{\mathrm{+}}$ compared with cagA ${}^{\mathrm{-}}$ strains was related to an increased risk of non-cardia cancer [28, 29, 30, 31]. For example, in a case-control study on Swedish community, there was a significant correlation between the presence of CagA (OR 9.2) and VacA (OR 3.5) and the risk of NCGA, but not CGA [36]. In a study on 41 cardia and 339 non-cardia cancer cases, and 380 controls, a positive association was not found for cagA ${}^{\mathrm{+}}$ strains with cardia cancer, but cagA ${}^{\mathrm{+}}$ strains was related to an increased risk of non-cardia cancer (OR 1.60) [32]. Wang et al., in a case-control study (257 cases with non-cardiac cancer and 514 controls) in Xi’an, China showed that the cagA ${}^{\mathrm{+}}$ strains of H. pylori had a strong correlation with non-cardiac cancer [33]. In contrast, in a study from New Jersey and western Washington, infection with cagA ${}^{\mathrm{+}}$ strains did not show a significant association with non-cardia cancer (OR 1.4), but an inverse association with the risk of esophageal and gastric cardia adenocarcinoma was found (OR 0.4) [34].

A simple logistic regression analysis in our study also showed that the cagA ${}^{\mathrm{+}}$ genotype was not associated with a statistically significant increased risk for IGA, and DGA. Our results were in compliance with previous studies from India andLatin America countries, so that a study on Indian patients indicated that 34% of the control gastric tissues were cagA ${}^{\mathrm{+}}$ , however only 26.2% of the gastric cancer groups (29.5% in DGA and 19.0% in IGA) were harboring the cagA gene. Statistical analysis showed no association between this genotype and either histologic variety of GA [54]. Cardenas-Mondragon et al. in a case-control study of patients with gastric disease in Latin America observed the frequency of cagA ${}^{\mathrm{+}}$ decreased in intestinal-type 28/50 (56%) but it was increased in the diffuse-type of GC 50/64 (78.1%) compared with NAG 157/225 (69.8%). However, these differences were not statistically meaningful [55]. These were not in agreement with the results of a recent case-control study in Sweden, where CagA and VacA were also associated with a heightened risk of both intestinal (ORs 6.0 and 3.7, respectively) and diffuse (ORs 20.6 and 3.9, respectively) histologic subtypes [36]. Furthermore, in Figura et al. study, a total of 214 strains were cagA-positive that were isolated from both IGC and DGC patients; and 12 strains were cagA-negative, all of which were isolated from DGC patients ( $P<$ 0.001) [51]. In our study, the presence of the cagA ${}^{\mathrm{+}}$ genotype in combination with the vacA c1 genotype was associated with NCGA, IGA, and DGA, while the cagA ${}^{\mathrm{+}}$ genotype alone was not associated with these diseases. These findings might reveal a significant concordance between the vacA c1 and cagA ${}^{\mathrm{+}}$ genotypes; however, it still remains obscure.

Our study has some strengths. We enrolled biopsy-proven gastric adenocarcinomas and examined directly the genetics of the active H. pylori strains isolated from gastric biopsies of all case and control individuals. Furthermore, we adopted two control groups; non-tumors and those with NAG. This, in turn, increased the reliability of association analyses because, in addition to NAG, PUs also was included in the first group. Most of the above studies, however, tested VacA or CagA seropositivity among H. pylori-infected subjects, and some of them did not even include information on prior attempts at H. pylori eradication in individuals. Our study has also some limitations. We did not find a significant association between the cagA ${}^{\mathrm{+}}$ genotype alone and the risk of NCGA and the different histologic types of GA. It also showed no effect on cardia adenocarcinoma risk, even in combination with the c1-type of vacA. One reason might be that we did not assess the diversity of the carboxyl-terminal region of CagA and its EPIYA motifs. It has been shown that the strains carrying CagA with a large number of EPIYA-C motifs are more likely linked to gastric precancerous lesions and GC [56, 57]. As previously shown, all the Iranian H. pylori strains are Western-type, representing type-C phosphoryaltion motif [58, 59]. Therefore, determining the number of EPIYA-C motifs and recruiting a larger sample size might be important for predicting the risk of cardia and non-cardia as well as intestinal- and diffuse-type adenocarcinomas in relation to the cagA ${}^{\mathrm{+}}$ genotype.

In sum, the H. pylori vacA c1 genotype, but not cagA status, might be the first important bacterial biomarker for predicting the cardia adenocarcinoma risk in male patients aged $⩾$ 55 in Iran. c1, whether independently or in combination with cagA, might also predict the risk of non-cardia and intestinal- or diffuse-type gastric adenocarcinomas. Further studies from other parts of the world are needed to assess the strength of these findings.