Host gastric Lewis expression determines the bacterial density of Helicobacter pylori in babA2 genopositive infection

Abstract

Background and aims: We tested if host gastric Lewis antigens and the babA2 genotype of Helicobacter pylori correlated with clinicohistological outcome.

Methods: We enrolled 188 dyspeptic patients (45 with duodenal ulcer, 45 with gastric ulcer, and 98 with chronic gastritis) with H pylori infection, proved by culture and gastric histology, reviewed by the updated Sydney system. Gastric expression of Lewis (Le) antigens Le^a, Le^b, Le^x, and Le^y was determined immunochemically to determine intensity (range 0–3). The corresponding 188 H pylori isolates were screened for babA2 genotype by polymerase chain reaction.

Results: All H pylori isolates had a positive babA2 genotype. We identified Le^a in 33.5%, Le^b in 72.9%, Le^x in 86.2%, and Le^y in 97.4% of biopsies from these 188 patients. Patients who expressed Le^b had a higher H pylori density than those who did not express Le^b (p<0.001). Among 139 patients who expressed Le^b, H pylori density increased with a higher Le^b intensity (p<0.05). Gastric atrophy decreased with Le^b intensity and thus resulted in lower H pylori density in the antrum (p<0.05). For the 49 patients without gastric Le^b expression, H pylori density was positively related with Le^x and Le^a expression (p<0.05).

Conclusions: Taiwanese H pylori isolates are 100% babA2 genopositive. Gastric Le^b as well as Le^x intensity may be major determinants of H pylori density. While lacking gastric Le^b expression, Le^x and Le^a were closely related to H pylori colonisation.

Helicobacter pylori is a well recognised gastric pathogen in humans.^1,2 The ability of H pylori to achieve persistent colonisation in the human stomach has become the focus of intense research.³ Several studies have proposed that the molecular mimicry of H pylori lipopolysaccharide antigens to human Lewis (Le) antigens may help H pylori to evade the immune response and enhance bacterial adherence to gastric epithelium.^3–6 As Le antigens are also found on the gastric epithelium in humans,⁷ Le antigen expression may mediate the attachment of H pylori to the gastric mucosa.⁸ Strong evidence was provided by Ilver et al who purified the blood group antigen binding adhesin (BabA) of H pylori and found that BabA selectively adheres to the Le^b antigen of the host.⁹ Their findings suggest that gastric Le^b antigens selectively interact with the products of the babA2 (blood group associated binding gene) allele of H pylori and thus may possibly facilitate a more dense colonisation in the stomach. However, contradictory data focused on the role of the babA2 genotype in terms of clinicohistological outcome without analysing the host status for Le^b expression in the stomach.^10–12 Therefore, we conducted this study to elucidate if the interaction of the babA2 genotype of H pylori and gastric Le^b antigen expression of the host are correlated with different clinical outcomes.

As gastric Le^b antigens cannot be found in all humans, some other pathways must exist to facilitate adherence of H pylori. In contrast with the rare expression of Le^b, Le^x and Le^y antigens are commonly expressed.^3–5,13 As the adhesion pedestal formation contained Le^x on both H pylori and gastric epithelium, these Lewis antigens may be required to establish or maintain infection.^3,13 Thus we tested if these Lewis antigens have a role in bacterial adherence, when the host has weak or no gastric Le^b expression, interacting with the BabA of H pylori.

MATERIALS AND METHODS

Patients and study design

A total of 188 dyspeptic patients (112 men and 76 women; mean age 44.8 years) gave informed consent and were consecutively enrolled after they were proved to have H pylori infection, defined as a positive culture. None had a previous history of anti-H pylori therapy. Each patient had undergone panendoscopy to obtain a gastric biopsy for culture and histology of H pylori infection. The endoscopic diagnosis of these 188 study patients included uncomplicated chronic active gastritis (n=98), duodenal ulcer (n=45), and gastric ulcer (n=45).

At gastric biopsy, five samples, including two from the antrum, two from the corpus, and one from the cardia, were obtained during endoscopy. Three gastric specimens, each one from antrum, corpus, and cardia, were stained with haematoxylin and eosin as well as with modified Giemsa stains. Apart from analysis of H pylori related gastric histology, these three gastric specimens were stained immunochemically for expression of Lewis antigens Le^a, Le^b, Le^x, and Le^y. The remaining two gastric specimens were used for H pylori culture.¹⁴ Genomic DNA of these H pylori isolates were then extracted by polymerase chain reaction (PCR) to detect the babA2 genotype. Extraction of DNA was performed using the same method as reported in our previous publication.¹⁵

PCR and primers for babA2 genotypes

Extracted DNA from each strain was subjected to PCR for amplification of the babA2 genes, applying one pair of primers (babo-F: CTT AAA TAT CTC CCT ATC CC, corresponding to bp 1 to 20 of AF033654; babo-R: CGA TTT GAT AGC CTA CGC TTA TG, corresponding to bp 369 to 391 of AF033654) designed by Ilver and colleagues⁹ or another self designed primer (bab7-F: CCA AAC GAA ACA AAA AGC GT, corresponding to bp 105 to 124 of AF033654; bab7-R: GCT TGT GTA AAA GCC GTC GT, corresponding to bp 357 to 375 of AF033654).

The PCR mixtures were performed in a volume of 50 μl containing 0.2 μM of each primer, 0.2 mM each of deoxynucleoside triphosphates, reaction buffer with MgCl₂, and 1 unit of DyNAzyme II DNA polymerase (Finnzymes OY, Espoo, Finland). Amplification was carried out over 30 cycles consisting of 94°C for one minute, 45°C for one minute, and 72°C for one minute in a thermal cycler (Perkins-Elmer Cooperation, Norwalk, Connecticut, USA). The two primers achieved a 391 bp product (by primers designed by Ilver et al) and a 271 bp product (using the self designed primers in this study), respectively. The sequences of these two PCR products were determined using an ABI PRISM 377 DNA Sequencer (Applied Biosystems, Foster, California, USA).

In addition, we randomly selected 30 babA2 genopositive strains (proven by the presence of the 271 bp PCR product) to test their BabA producing phenotype by western blotting using BabA specific antiserum. BabA specific antiserum was obtained from Drs Thomas Boren and Stefan Odenbreit.^8,9,16 Each selected H pylori extract was analysed on a sodium dodecyl sulphate-10% polyacrylamide gel. The blot was then subjected to a 1:500 dilution of the anti- BabA antibody and detected with goat antirabbit antibody conjugated to horseradish peroxidase (Chemicon International Inc., Temecula, California, USA).

Analysis of H pylori related histology

The same pathologist, unaware of the endoscopic and culture results, analysed the gastric histology. H pylori density for each specimen was scored according to Yang and colleagues¹⁷: score 0, no bacteria; score 1, one or two small clusters with less than 10 bacteria; score 2, less than half the superficial crypt area with less than 10 bacteria in each crypt; score 3, less than half the area but with more than 10 bacteria, or more than half the area with less than 10 bacteria in each crypt; score 4, >10 bacteria in forvelae with some free area; and score 5, >10 bacteria without a free area. Total H pylori density (HPD) was defined as the sum of the densities from the three biopsy samples, obtained from the antrum, corpus, and cardia. Thus the HPD score ranged from 1 to 15. The acute inflammatory score (range 0–3), chronic inflammation score (range 0–3), atrophic change (absent, 0; present, 1), and intestinal metaplasia (IM) (absent, 0; presence, 1) were graded using the updated Sydney system.¹⁸ The total acute (AIS) and chronic (CIS) inflammatory scores were also a sum of the three specimens (range 0–9).

Immunochemical staining for gastric Lewis expression

Immunostaining of biopsy specimens for Lewis antigens was performed using the standard avidin-biotin-peroxidase technique. Formalin fixed paraffin embedded tissue sections, including topographical specimens from the antrum, corpus, and cardia from each patient, were deparaffinised through xylene and hydrated with ethanol. Slides were washed with distilled water and then placed in 1× phosphate buffered saline (PBS) for five minutes. Incubation with 3% hydrogen peroxide for three minutes blocked the endogenous peroxidase activities of these sections. After incubation with 2% bovine serum albumin for two hours and washing with PBS, the primary monoclonal antibodies for detection of gastric Lewis antigens were used (anti-Lewis Le^a, Le^b, Le^x, and Le^y ; Signet Laboratories, Inc., Dedham, Massachusetts, USA). The reaction time for the primary monoclonal antibodies (anti-Le^a, Le^b, Le^x, and Le^y) was three hours at 25°C. These slides were again washed with PBS and incubated with the secondary antibody to achieve a 1:2000 dilution of antimouse IgG and IgM conjugated to horseradish peroxidase (Chemicon International Inc., Temecula, California, USA) for two hours at 25°C. These slides were finally washed with PBS, and the AEC kit (Sigma, St Louis, USA) was used as substrate to illustrate the stain. All slides were evaluated blindly by the same pathologist. For each gastric site, the intensity of Le^a, Le^b, Le^x, and Le^y was scored from 0 to 3 (0, no staining; 1, staining of either surface mucous cells or deep gastric glands; 2, staining of surface cells, intercryptal epithelium, and deep glands but expressed in ⩽50% of the analysed specimens; 3, diffuse staining of ⩾50% of the analysed specimens on surface cells, intercryptal epithelium, and deep glands). Examples of intensity 1 and intensity 2 gastric Le^b expression are shown in fig 1A ▶ and 1B ▶, respectively. Total gastric Lewis antigen intensity (TLI) for Le^a, Le^b, Le^x, and Le^y was the sum of three biopsy samples from the antrum, corpus and cardia (range 0–9).

Figure 1.

(A, B) Gastric immunohistochemical stains of Lewis antigen Le^b expression. (A) Positive staining over the surface epithelium only. (B) Diffuse staining over the intercryptal epithelium. (C, D) Gastric immunohistochemical stains of Le^x expression. (C) Positive staining over the surface epithelium only. (D) Diffuse staining over the deep glands.

Statistics

The Student’s t test and paired t test were used as appropriate for parametric differences. One way ANOVA with Bonferroni’s method was used for multiple testing of data. Pearson’s χ² test was used for non-parametric proportions. All significance tests were two tailed and a p value <0.05 was taken as significant.

RESULTS

Prevalence of babA2 genotypes of H pylori infection in Taiwan

Fifty per cent (94/188) of H pylori isolates had a positive babA2 genotype by PCR, applying the primers used by Ilver et al to obtain a band of 371 bp. However, the nucleotide sequence of this 391 bp PCR product from the Taiwanese isolates was confirmed as not being babA2 in origin but had >90% homology with the published sequence of adenine specific DNA methyltransferase in H pylori 26695. To detect the babA2 genotype for the domestic strains, we self designed a pair of primers and achieved a 271 bp PCR product whose nucleotide sequence was confirmed with >95% homology to the babA2 gene of CCUG 17875 (fig 2 ▶). Based on PCR using these primers to obtain a 271 bp band, the prevalence rate of the babA2 genotype was 100% in all 188 Taiwanese H pylori isolates. Western blotting also confirmed that the 30 randomly selected 271 bp genopositive strains had a uniformly positive phenotype.

Figure 2.

The nucleotide sequence of the 271 bp polymerase chain reaction product gained from the self designed babA2 primers. The nucleotide sequence of the 10 randomly selected domestic strains (hp250, hp258, hp222, hp116, hp657, hp130, hp238, hp82, hp 639, and hp76) was confirmed to be babA2 in origin with >90% homology to the published sequence of the babA2 gene (afbabA2) of CCUG 17875. The 10 nucleotides (ATG AAA AAA C), representative of the babA2 gene of Helicobacter pylori, are indicated by “=”.

Topographic gastric Lewis antigen expression in H pylori infected Taiwanese

Based on the presence of staining of any one of the three gastric specimens, we identified Le^a in 33.5%, Le^b in 72.9%, Le^x in 86.2%, and Le^y in 97.4% of gastric biopsies in these 188 patients. As shown in table 1 ▶, the topographic intensity of gastric Le^y expression was higher in the corpus than in the antrum or cardia (1.76 v 1.61 and 1.68; paired t test, p<0.05). The intensity of Le^b expression was also higher in the corpus and cardia than in the antrum (1.74 v 1.37, and 1.59 v 1.37; paired t test, p<0.05). In contrast, the topographic intensity of Le^a or Le^x was higher in the antrum than in the corpus and cardia (Le^a: 0.56 v 0.41 and 0.45, p<0.05; Le^x: 1.19 v 0.88 and 0.75, p<0.05).

Table 1.

Topographic distribution of the intensity of gastric Lewis antigen expression in 188 patients with Helicobacter pylori infection

Lewis antigen	Antrum	Body	Cardia	Significance*
Intensity (range 0–3)
Lea	0.56 (0.77)	0.41 (0.65)	0.45 (0.81)	A>B; A>C
Leb	1.37 (1.24)	1.74 (0.95)	1.59 (1.03)	B>A; C>A
Lex	1.19 (0.72)	0.88 (0.71)	0.75 (0.73)	A>B; A>C
Ley	1.61 (0.95)	1.76 (1.07)	1.68 (0.88)	B>A; C>A

Values are mean (SD).

*Significant difference by paired t test with two tailed analysis (p<0.05).

A, antrum; B, body; C, cardia.

Lewis antigen expression and clinicohistological features of H pylori infection

There was no difference in ulcer rate between patients with or without Lewis antigen expression in the stomach (table 2 ▶). Patients with gastric Le^b expression had significantly higher HPD and CIS than those without Le^b expressions (HPD: 9.29 v 6.35, p<0.001; CIS: 7.59 v 6.15, p<0.05). We also found that mean HPD of 12 Le^a+b− patients was significantly lower than that of either 88 Le^a−b+ patients or 51 Le^a+b+ patients (7.42 v 9.22 and 9.41; p<0.05 by one way ANOVA). In fig 3A ▶, TLI of Le^b was found to be positively correlated with HPD (one way ANOVA, p<0.05). In table 2 ▶, although the statistical significance was limited, HPD was evidently higher in those patients who expressed Le^a, Le^x, and Le^y in the stomach. Furthermore, patients who expressed Le^a and Le^x had a higher bacterial density in biopsies (p<0.05). HPD was even elevated when the total number of gastric Lewis expression (Lewis-N) of each study patient increased (fig 3B ▶). Multivariate logistic regression disclosed that the intensity of Le^b and Le^x expression, rather than Lewis-N, was an independent factor correlated with HPD in H pylori infected patients (table 3 ▶).

Table 2.

Lewis antigen expression and clinicohistological features of Helicobacter pylori infection

	Lea		Leb		Lex		Ley
Parameter (mean)	(+) (n=63)	(−) (n=125)	(+) (n=139)	(−) (n=49)	(+) (n=162)	(−) (n=26)	(+) (n=183)	(−) (n=5)
AIS (0–9)§	2.31	2.61	2.62	2.21	2.81	2.45	2.55	1.18
CIS (0–9)†	6.89	6.97	7.59	6.15	6.95	6.88	6.96	6.40
AT (%)	57.1	57.6	57.6	57.1	54.9	73	56.3	100
IM (%)‡§	28.6	29.6	28.1	32.7	26.7	57.7	25.1	80
Ulcer rate (%)	49	45.6	47.5	44.9	47.5	42.3	46.9	40
HPD (1–15)†	9.03	8.29	9.29	6.35	8.59	8.03	8.57	7.41
Antrum (1–5)†	2.89	2.69	2.87	2.41	2.73	2.71	2.77	2.14
Body (1–5)*†‡	3.46	3.08	3.52	2.31	3.31	2.69	3.21	2.85
Cardia (1–5)†	2.67	2.51	2.89	1.63	2.66	2.53	2.57	2.41

AIS, acute inflammatory score; CIS, chronic inflammatory score; AT, antral atrophy; IM, intestinal metaplasia; HPD, total density of H pylori.

Significant difference (p<0.05): *between Le^a (+) and Le^a (−) patients; †between Le^b (+) and Le^b (−) patients; ‡between Le^x (+) and Le^x (−) patients; §between Le^y (+) and Le^y (−) patients.

Figure 3.

(A) Total Helicobacter pylori density (HPD) was positively correlated with an increase in Le^b intensity (*p<0.05, significant difference analysed using one way ANOVA). (B) Total Lewis number (Lewis-N) of each patient was also positively correlated with higher HPD in the stomach (one way ANOVA, p<0.05).

Table 3.

Multivariate logistic regression for independent factors relevant to total Helicobacter pylori density of the stomach

Parameter	Coefficient	Standard error	p Value	95% confidence interval
Total 188 patients
Total Leb intensity	0.491	0.052	0.000*	(0.389~0.593)
Lewis-N	−0.327	0.294	0.267	(−0.906~0.253)
Total Lex intensity	0.230	0.075	0.002*	(0.083~0.378)
Total Ley intensity	0.092	0.056	0.116	(−0.022~0.199)
Total Lea intensity	0.123	0.076	0.221	(−0.057~0.244)
In 139 patients with Leb
Total Leb intensity	0.518	0.066	0.000*	(0.386~0.649)
Total Lex intensity	0.183	0.074	0.015*	(0.037~0.329)
Total Ley intensity	0.061	0.055	0.269	(−0.048~0.169)
Total Lea intensity	−0.044	0.073	0.546	(−0.187~0.099)
In 49 patients without Leb
Total Lex intensity	0.594	0.136	0.000*	(0.318~0.868)
Total Lea intensity	0.242	0.105	0.025*	(0.031~0.456)
Total Ley intensity	0.111	0.093	0.241	(−0.077~0.298)

Lewis-N, total number of four Lewis antigen expressions in the stomach of each patient.

*Significant difference.

As also shown in table 2 ▶, patients with no expression of Le^x or Le^y had higher rates of IM in the stomach (p<0.01). Among those patients who expressed Le^x, TLI of Le^x was significantly lower in the presence of IM (2.68 v 3.83; p< 0.01) but higher in the presence of antral atrophy (4.05 v 3.01; p< 0.01). There was no decrease in TLI of Le^a, Le^b, or Le^y, despite the presence of IM or antral atrophy.

Factors correlating with HPD in non-Le^b patients

Of the 49 H pylori infected patients without Le^b expression, HPD was higher in patients who expressed gastric Le^a and Le^x (Le^a: 7.41 v 6.05, p<0.05; Le^x: 6.81 v 4.25, p<0.005). In table 3 ▶, multivariate logistic regression confirmed that Le^a and Le^x antigens were both independently correlated with HPD in the 49 patients without Le^b expression. In contrast, for the 139 patients who expressed Le^b, only Le^x but not Le^a antigens were independently correlated with HPD (table 3 ▶).

Compensatory effect of Le^x to maintain HPD for weak Le^b intensity in antral atrophy

Among the 139 Le^b positive patients, the topographic distribution of H pylori density and the intensity of Le^a, Le^b, and Le^x were compared between patients with and without antral atrophy (table 4 ▶). There was no decrease in either Le^a or Le^x intensity over the antrum despite the presence of atrophy. In contrast, patients with antral atrophy had a lower Le^b intensity over the antrum (p<0.05) and thus had a significantly lower bacterial density (p<0.05).

Table 4.

Topographic distribution of Helicobacter pylori density and gastric Lewis antigen intensity in 139 Le^b positive patients with and without antral atrophy

Parameter (mean)	Antrum	Body	Cardia	Body+cardia	Total
Lea intensity
AT (n=80)	0.54	0.39	0.54	0.93	1.36
Non-AT (n=59)	0.59	0.51	0.49	1.03	1.64
Leb intensity
AT (n=80)	1.63*	2.43	2.23	4.66	6.05
Non-AT (n=59)	2.15	2.26	2.25	4.51	6.44
Lex intensity
AT (n=80)	1.21	1.31*	0.94*	2.24 *	2.49
Non-AT (n=59)	1.28	0.88	0.77	1.66	2.24
H pylori density
AT (n=80)	2.52*	3.65*	3.05*	6.71*	9.23
Non-AT (n=59)	3.35	3.25	2.67	5.98	9.38

Body+cardia, the summation of the data from the body and cardia; Total, the sum of antrum, body, and cardia,

*Significant difference between patients with (AT) and without (non-AT) antral atrophy, analysed by the two tailed Student’s t test (p<0.05).

Although both bacterial density and the intensity of Le^b over the antrum were lower, HPD was not decreased by the presence of antral atrophy (table 4 ▶). The paradoxical increase in bacterial density on the body and cardia were found to maintain HPD under the presence of antral atrophy (p<0.05). However, there was no significant increase in Le^b intensity elsewhere in the body or cardia (table 4 ▶). In contrast, a significant increase in the intensity of Le^x over the gastric body and cardia was found in those patients with antral atrophy compared with those without antral atrophy (p<0.05).

DISCUSSION

Identification of specific receptors for H pylori on the gastric mucosa may explain why the organism can only adhere to those cells in humans. Ilver et al disclosed that the babA2 gene of H pylori is a putative determinant allowing it to adhere to Le^b of the gastric epithelium and thus could promote bacterial invasion of the human stomach.⁹ Our prospective study enrolled 188 H pylori infected patients and is the first to analyse both bacterial babA2 genotype and gastric antigen expression (including Le^b), thus further elucidating the impact of any interactions between BabA and Le^b on the clinicohistological outcome after H pylori infection.

In the present study, after applying the primer of Ilver et al to obtain a 391 bp PCR product, we discovered it was non-babA2 in origin. By applying our self designed primers, a 271 bp PCR product was found and was confirmed to have >95% homology to the published sequence of babA2. The nucleotide sequence data confirmed that our self designed pair of primers were suitable for babA2 genotyping in Taiwan and all 188 isolates in this study were uniformly proven to have a babA2 positive genotype. The prevalence was higher than in previous reports (38–85%).^10–12,19 Moreover, such an extremely high prevalence of babA2 in Taiwan suggests this could be an ideal country in which to study whether babA2 is a good target for preventive vaccination if BabA interacts strongly with Le^b to impact on H pylori colonisation of patients.

The prevalence rates of the different Lewis antigens in our study were compatible with Kobayashi et al, who reported that Le^a had the lowest incidence and that gastric Le^x or Le^y may disappear when H pylori infection is induced by IM.⁷ Such a finding was indirectly supported by our data (table 1 ▶) which showed that patients without expression of Le^x or Le^y had higher rates of IM than those with Le^x and Le^y (p<0.05).

Patients with gastric Le^b expression had a higher bacterial density of H pylori than those without Le^b expression (p<0.05) (table 2 ▶). TLI of Le^b expression was also positively correlated with HPD (fig 3A ▶). Moreover, HPD was higher in Le^a−b+ weak and Le^a+b+ strong secretors than in Le^a−b− non-secretors (p<0.05). Accordingly, the intensity of Le^b was proved to be an independent factor in determining HPD (table 3 ▶). As all domestic strains were babA2 positive, our study from Taiwan may be the most rational in elucidating the fact that gastric Le^b really serves as an important receptor for H pylori adherence.

There were 49 patients with H pylori infection but no expression of Le^b in the stomach. Bacterial densities of the body remained higher in patients with positive expression of Le^a and Le^x (p<0.05) (table 2 ▶). An increasing trend for HPD was found in patients whose Lewis-N ranked high (p<0.05, by one way ANOVA) (fig 3B ▶). These data imply that there may be some additive effect of expression of other Lewis antigens, apart from Le^b, serving as adherent receptors for H pylori. This is compatible with the finding of Clyne and Drumm who confirmed that blocking with a monoclonal antibody for the Le^b antigen on the gastric epithelium could not totally abolish adherence of H pylori.²⁰ Thus we tested if other gastric Lewis types also enhanced bacterial adherence in the 49 patients without Le^b expression. Our study found that patients who expressed Le^x and Le^a had higher HPD than those who did not express Le^x and Le^a (Le^a: 7.41 v 6.05, p<0.05; Le^x: 6.81 v 4.25, p<0.005). By multiple logistic regression, Le^x and Le^a were further confirmed to be independent factors in enhancing colonisation of H pylori (table 3 ▶). These clinical data thus support the laboratory findings of Taylor et al which found adhesion pedestal formations stained with Le^x on both H pylori and gastric epithelium. Accordingly, our study confirmed that Lewis antigens other than Le^b can be used to establish or maintain H pylori infection in the stomach.^3,10,16

Expression of Le^b was stronger in the body, in contrast with the antral dominant distribution of Le^a and Le^x (table 1 ▶). Therefore, we tested whether Le^x and Le^a had additive effects when present with Le^b for enhancement of H pylori colonisation in the 139 patients with Le^b expression. Patients with antral atrophy had different topographic distributions of bacterial density but the total density of H pylori did not differ (table 4 ▶). The presence of antral atrophy decreased the intensity of Le^b, which was expected, as Le^b usually stained the superficial glands.^7,9,19 When the intensity of Le^b was lower, bacterial density here decreased. However, overall HPD was maintained by the paradoxical increased density over the body and corpus. As the intensity of Le^x over the body and cardia were higher in the presence of antral atrophy, increased bacterial densities here could be mediated by Le^x expression. These clinical data supported the finding that gastric Le^x antigen can enhance H pylori adherence.¹⁶ Moreover, Le^x may have compensatory or additive effects with Le^b to maintain bacterial loads during ongoing atrophy changes.

Among those patients who expressed Le^x, TLI of Le^x in the presence of IM was 2.68 versus 3.83 (p<0.01) but was higher in the presence of antral atrophy (4.05 v 3.01; p< 0.01). These data confirm that IM and antral atrophy may change Le^x expression and thus alter the H pylori colonisation pattern.

In summary, Taiwanese H pylori isolates are 100% babA2 positive. Gastric Le^b intensity as well as Le^x intensity appear to be major determinants of bacterial density of H pylori. When lacking gastric Le^b expression, Le^x and Le^a are closely related to H pylori colonisation. To overcome H pylori adherence, genomic targets such as babA2 (or others interacting with gastric Lewis antigens) may be promising.

Abbreviations

• Le, Lewis

• Lewis-N, total Lewis number

• BabA, blood group antigen binding adhesin

• babA2, blood group associated binding gene

• PCR, polymerase chain reaction

• TLI, total gastric Lewis antigen expression intensity

• HPD, Helicobacter pylori density

• IM, intestinal metaplasia

• CIS, chronic inflammatory score

• PBS, phosphate buffered saline