Density of Helicobacter pylori Infection In Vivo as Assessed by Quantitative Culture and Histology

Transcription

1 552 Density of Helicobacter pylori Infection In Vivo as Assessed by Quantitative Culture and Histology John C. Atherton,* Kyi T. Tham, Richard M. Peek, Jr., Timothy L. Cover, and Martin J. Blaser Divisions ofinfectious Diseases and of Gastroenterology, Department of Medicine, and Department ofpathology, Vanderbilt University School ofmedicine, and Medical Service and Department ofpathology, Department of Veterans Affairs Medical Center, Nashville, Tennessee Helicobacter pylori density was assessed by quantitative culture and histologic examination of gastric biopsy specimens from 29 H. pylori-infected dyspeptic patients. Density was correlated with caga and vaca genotypes (assessed by polymerase chain reaction and colony hybridization), gastric inflammation and epithelial injury (assessed histologically), and peptic ulceration. Quantitative culture was more reproducible than histology, and antral density was more reproducible than corpus density. Mean antral density of caga +lvaea sl strains was 4-fold higher than that of caga -{vaea s2 strains (1.9 X 10 6 vs. 4.5 X los cfu/g, P =.02). Antral density was associated with mucosal neutrophilic and lymphocytic infiltration (P <.01) and with epithelial injury (P <.05). Mean antral bacterial density was 5-fold higher in duodenal ulcer patients than in others (P =.005). In conclusion, H. pylori density in vivo is easily quantified and is associated with bacterial virulence determinants, gastric inflammation, and duodenal ulceration, suggesting a central role in pathogenesis. Helicobacter pylori infection causes chronic gastritis and is a major risk factor for duodenal and gastric ulceration, gastric adenocarcinoma, and primary gastric lymphoma [1, 2]. Why some infected persons develop these sequelae and others do not is unknown, but one possible explanation is that some H. pylori strains are more pathogenic than others. Research has focused on two main groups of bacterial virulence detenninants: possession ofthe cagiregion (for which caga is a marker [3]) and differences in vacuolating cytotoxin (VacA) activity [4,5] and genotype [6]. Possession of caga [7-9], possession ofthe vaca sl genotype [6] (in particular its sla subtype [10]), and in vitro expression of vacuolating cytotoxin activity [4, 5] are all characteristics found more commonly in H. pylori isolates from patients with duodenal ulcers than from patients without ulceration. One hypothesis to explain why these strains are associated with ulcers is that they colonize the stomach more densely than do other strains. Increased gastric bacterial density may lead to increased levels of inflammation and epi- Received 8 January 1996; revised 7 May Presented in part: Seventh European Workshop on Gastroduodenal Pathology and Helicobacter pylori and Satellite Symposium of the Tenth World Congress of Gastroenterology, Houston, September 1994 (abstract 147; Am J Gastroenterol 1994; 85: 1322). Informed consent was obtained from all patients taking part in this study. The study followed the guidelines of the Department of Health and Human Services, and ethical approval was granted by the Vanderbilt University and Nashville Department of Veterans Affairs Institutional Review Boards. Grant support: NIH (DK and DK-45293); Medical Research Service, Department of Veterans Affairs. Reprints or correspondence: Dr. Martin J. Blaser, Division of Infectious Diseases, Dept. of Medicine A3310 MCN, Vanderbilt University School of Medicine, st Ave. S, Nashville, TN * Present affiliation: Division of Gastroenterology, Department ofmedicine, University Hospital, Queen's Medical Centre, Nottingham, UK. The Journal of Infectious Diseases 1996: 174: by The University of Chicago. All rights reserved /96/ $ thelial injury. This may be a marker for duodenal damage or may itself cause increased acid output [11, 12] and hence predispose to duodenal ulceration. The relationship between H. pylori virulence determinants and colonization density has not been studied. Previous studies have, with one exception [13], positively associated bacterial numbers with mucosal neutrophil and lymphocyte density [14, 15] and with duodenal ulceration [16]. However, apart from one study [17], assessment of bacterial density has been based on histology and has used semiquantitative and subjective scoring systems. To confirm and extend these observations, we first aimed to standardize the techniques of quantitative H. pylori culture and histology. We then sought to assess the relationship between H. pylori density in vivo and H. pylori virulence markers, gastric inflammation, gastric epithelial injury, and duodenal ulceration. Subjects and Methods Subjects. Twenty-nine consecutively recruited H. pyloriinfected dyspeptic patients were enrolled prospectively into the study. These subjects represent a subgroup ofpatients recruited to a larger study of inflammatory mediators and bacterial virulence determinants in H. pylori gastritis described elsewhere [6, 18]. Endoscopic procedure andulcer diagnosis. At upper gastrointestinal endoscopy, active peptic ulceration was defined as a circumscribed break in the mucosa, with apparent depth, measuring at least 1 ern in any dimension, and active erosion was defined as a circumscribed break in the mucosa not fulfilling these criteria. Peptic ulcer disease was defined as active peptic ulceration or an unequivocal diagnosis of peptic ulceration made at a previous endoscopy or upper gastrointestinal series. Gastric biopsy specimens weighing 3-10 mg were obtained, using Olympus FB-25K biopsy forceps (6 mm diameter), from the greater curve of the gastric antrum, 2-5 em from the pylorus, and from the greater

2 JID 1996; 174 (September) Density of H. pylori Infection In Vivo 553 curve of the gastric corpus, 6-10 em proximal to the termination of the gastric rugae. From each site, 2 gastric biopsy specimens were obtained for quantitative culture and 2 for histologic evaluation. Ofthe 29 patients, only 27 had all ofthese biopsies obtained; 2 did not have biopsies for quantitative culture obtained from the gastric corpus. Quantitative culture of H. pylori. Gastric biopsy specimens were transferred immediately to individual preweighed sterile microcentrifuge tubes containing 50 j.ll ofsaline at 4DC and processed 1 h later. Each biopsy was homogenized, using a sterile groundglass tissue minihomogenizer, then diluted to 250 j.ll in sterile saline. Serial 10-fold dilutions in saline were inoculated to trypticase soy agar with 5% sheep blood (BBL, Cockeysville, MD) and incubated under microaerobic conditions (Campypak plus; BBL) at 37 DC for 4 days. Viable counts were recorded, and a sweep of multiple colonies was transferred to a new plate for subsequent genotypic analysis. H. pylori was identified by positive urease, catalase, and oxidase tests and by typical appearance on Gram's stain. Histologic methods. Adjacent 4-j.lm paraffin sections were stained with hematoxylin-eosin and with a modified Giemsa stain, then examined by a single experienced histopathologist unaware ofthe patient's clinical diagnosis or the H. pylori strain characteristics. Bacterial density was quantified in the Giemsa sections by counting Helicobacter-like organisms on the mucosal surface and in the foveolae and dividing by the number of high-power fields (hpf) presented by the mucosal surface. Areas ofintestinal metaplasia were not colonized and thus were excluded from this analysis. The average number ofhpfcounted for each biopsy is 6.35, which is equivalent to 2.86 mm. Other histologic features were graded using the hematoxylin-eosin sections, as described [18]. Features scored were neutrophilic infiltration, lymphocytic infiltration and infiltration ofplasma cells, epithelial degeneration (irregular luminal cell borders, nuclear pyknosis, and cytoplasmic eosinophilia), mucus depletion, and epithelial erosion. These features were graded for 2 biopsies from each site (antrum or corpus), and the mean score was used. Assessment of bacterial virulence determinants. Assessment ofvirulence determinants has been fully described elsewhere [6]. In brief, caga status was determined by colony hybridization, vaca signal sequence type (sla, sib, or s2) by polymerase chain reaction, vaca midregion type (m 1 or m2) by polymerase chain reaction and colony hybridization, and cytotoxin production in vitro (Tox status) by a HeLa cell vacuolation assay done at least twice on each specimen. Data analysis. Of the 29 patients, 2 did not have corpus biopsies taken for quantitative culture, and 2 others did not have H. pylori identified histologically in antral or corpus biopsies. Thus, 25 patients were used for standardization of quantitative culture and histology in repeatability studies. Comparisons of H. pylori density with genotypes, inflammation, and clinical status are based on all 29 patients for antral quantitative culture, 27 patients for corpus quantitative culture, and 27 patients for quantitative histology in antrum and corpus. For statistical analysis, all data were normalized by log transformation. Repeatability of quantitative culture and histology were assessed by calculating their coefficients of repeatability, which are the 95% confidence intervals (95% CI) for the difference between results for 2 biopsies [19]. Continuous variables were correlated using Pearson's product moment correlation coefficient (r). Student's t test was used for comparison ofmean bacterial density values between groups. Analysis of trend across multiple ordered but nonparametric groups used the method of Cuzick [20, 21]. Results Twenty-nine H. pylori-infected patients were recruited to the study; the median age was 58 years (range, 32-80), all were men, and 28 were white and 1 was African-American. Of these, 17 (59%) had duodenal ulcer disease (12 active and 5 with no current ulcer or duodenal erosions), 15 (52%) were current smokers, 22 (76%) were regularly taking histamine; receptor antagonists, and 13 (45%) were regularly taking aspirin or nonsteroidal antiinflammatory drugs (NSAIDs). No significant association was found between any measure of bacterial density and age, smoking status, or histamine- receptor antagonist, aspirin, or NSAID use (data not shown). Comparison of quantitative culture and histology for assessing H. pylori density. On the basis of the 25 patients with complete biopsy sets, H. pylori density as assessed by quantitative culture was 2-fold higher in the antrum (mean, 1.7 X 10 6 [95% CI, 9.2 X X 10 6 ] cfu/g of biopsy wet weight) than in the corpus (7.9 X 105 [4.6 X X 10 6 ] cfu/g; P =.02). However, when assessed by histology, H. pylori densities in antrum (37 [22-64] bacteria/hpf) and corpus (31 [18-53]) were not significantly different. The coefficient of repeatability of H. pylori density in the antrum was ±3.5 fold for culture and ±5.9-fold for histology and in the corpus was ±8.5-fold for culture and ± 10.6-fold for histology. In the antrum there was a significant positive correlation between culture and histology values (r =.44, P =.027), but this was not seen in the corpus (r =.26, P =.20), where repeatability was poorer for both measures. Significant correlation was found between bacterial density in the antrum and corpus (r =.54, P =.006) as assessed by culture but not by histology (r =.20, P =.32). Relationship between H. pylori density in vivo and known bacterial virulence factors. Of the 29 H. pylori isolates obtained, 21 were caga ". The vaca genotypes, expressed as signal sequence type/midregion type, were slim1 for 10 isolates, s 11 m2 for 11, and s2/m2 for 8; of the 21 vaca sl strains, 14 were s1a and 7 were sib. In this study, all 21 vaca s 1 strains were caga + and all 8 vaca s2 strains were caga ". Twelve strains were Tox" in the HeLa cell vacuolation assay. caga+/vaca sl strains were associated with significantly more dense antral H. pylori colonization than caga - IvacA s2 strains, as assessed by both culture (mean, 1.9 X 10 6 [95% CI, 7.7 X X 10 6 ] vs. 4.5 X 105 [1.8 X X 10 6 ] cfu/g; P =.02; figure 1) and histology (44 [24-69] vs. 9.1 [1.7-50] bacteria/hpf; P =.01). A difference in corpus bacterial density between these groups of strains was observed by histology (39 [18-82] vs. 7.4 [3.9-14] bacteria/hpf; P =.02) but not by culture (6.6 X 105 [3.4 X X 10 6 ] vs. 6.3 X 10 5 [3.3 X X 10 6 ] cfu/g; P =.9). Among the caga+/vaca 81 strains, there was no significant difference in bacterial density between vaca

3 554 Atherton et al. JID 1996; 174 (September) ~ :..9::J ~u. :t:~ ~ ro.- V)... c: c: Q) « I I- caga- vacas2 (n=8) caga+ vacas1 (n=21) p=0.02 H. pylori genotype Ị - Figure 1. Relationship between antral H. pylori density, as assessed by quantitative culture, and vaca signal sequence and caga genotype of H pylori isolate. Bar, mean for each group. Difference between means was significant (P =.02, Student's t test). sla and s lb genotypes. No significant differences were found in bacterial density in the antrum or corpus between patients infected with Tox " and Tox" strains or between those infected with vaca midregion type ml and m2 strains. Relationship of H. pylori density to histologic gastric inflammation and epithelial damage. In the gastric antrum, H. pylori density, as measured by quantitative culture, was associated with both degree of neutrophilic infiltration (P =.009; figure 2A) and degree of lymphocytic infiltration (P =.007; figure 2B). These results were confirmed by quantitative histology (P =.006 for neutrophilic infiltration, P =.02 for lymphocytic infiltration; data not shown). Antral bacterial density by quantitative culture also was associated with both epithelial degeneration (P =.03; figure 1C) and mucus depletion (P =.05; figure ID). However, the associations of histologically assessed antral bacterial density with these measures of epithelial injury did not reach statistical significance (P =.1 for epithelial degeneration, P =.3 for mucus depletion). No significant association was found between bacterial density and histologic erosions, but only 8 of the 29 patients had erosions in one or more of the antral biopsy specimens. In the gastric corpus, no associations between bacterial density and any ofthe histologic parameters were statistically significant (neutrophilic infiltration, P =.11; lymphocytic infiltration, P =.23; epithelial degeneration, P =.11; mucus depletion, P =.12), although the trend was always in the same direction as for the antrum. Relationship between H pylori density in vivo and duodenal ulcer disease. The 17 patients with duodenal ulcer disease had a significantly higher mean antral density than the 12 without, as demonstrated by quantitative culture (2.5 X 10 6 [95% -Mean CI, 1.3 X X 10 6 ] vs. 4.9 X 105 [2.2 X X 10 6 ] cfu/g; P =.005; figure 3) and histology (58 [38~88] vs. 9.1 [4.1-20] bacteria/hpf; P =.001). In the gastric corpus, the density of infection was not significantly different between patients with and without duodenal ulcer disease, when measured by either quantitative culture (9.1 X 105 [4.4 X X 10 6 ] vs. 4.9 X 105 [2.1 X X 10 6 ] cfu/g; P =.28) or histology (30 [16-53] vs. 16 [7.3-36] bacteria/hpf; P.25). Discussion In this paper we describe two quantitative methods for measuring H. pylori density in vivo: quantitative culture and histology. Ofthese, culture is the more reproducible; that is, it gives results that agree more closely for 2 biopsy specimens from the same gastric region ofthe same patient. However, in practical terms, H pylori density had essentially the same associations whether measured by culture or histology; thus, the method used in further studies will be best determined by the type of expertise locally available. More substantial than differences between techniques was the reduced reproducibility of both methods in the gastric corpus compared to the antrum. H pylori infection may be patchy [22, 23], and our study implies that there is greater variation in bacterial density in the corpus than in the antrum. H pylori density is associated with caga and vaca genotype, in that caga+/vaca sl strains infect more densely than cag.a"] vaca s2 strains. These two genotypic characteristics are closely associated [6]; thus, which one is more important or whether both are markers for another genotypic determinant cannot be determined from this study. That vaca midregion type and Tox phenotype were not significantly associated with bacterial density was expected, as both ofthese features are less closely associated with duodenal ulcer disease than is vaca signal sequence type [6]. However, within the caga/vaca sl group, a difference between colonization density of s1a and sib strains might have been predicted, as s1a strains are associated with more inflammation and with increased prevalence of duodenal ulcer disease [10]. That such a difference was not found implies that these associations are independent of bacterial density, which is to be expected, as vaca signal sequence influences cytotoxin activity and thus would be expected to have a direct effect on inflammation. In agreement with most other studies [13, 14, 15, 24], we found that bacterial density was associated with inflammation in terms of both neutrophil and lymphocyte infiltration of the mucosa. In agreement with one other study [24], it was also found to be associated with epithelial injury in terms ofepithelial degeneration and mucus depletion. This is consistent with the hypothesis that H. pylori genotype determines bacterial density, which in tum determines level of inflammation and epithelial damage. There is no evidence from this study that caga +/vaca s1 strains cause more inflammation per bacterium than caga - /vaca s2 strains: The ratio of inflammatory grade

4 to bacterial number is actually higher for caga-lvaca s2 strains, although interpretation of this ratio is uncertain since inflammatory grade is scored nonparametrically. Even ifinflammation per bacterium is the same, it remains possible that caga +IvacA s I bacteria initially cause more inflammation but that this is subsequently down-regulated by the host so that a new steady state is reached [25]. An important finding confirmed in this study was the strong association between antral bacterial density and duodenal ulcer disease. Several possible mechanisms could explain this. Antral H. pylori density could be a marker for duodenal H pylori density, inflammation, and epithelial damage or it could be a marker for virulence factors important in the pathogenesis of duodenal ulceration, for example, products of caga or linked genes or of vaca. Alternatively, more dense infections, associated with more antral inflammation, may cause lower somatostatin expression [12], leading to higher levels of gastrin and of acid production [II], which may predispose to duodenal ulceration. Corpus H. pylori density was not found to be significantly associated with histologic measures ofinflammation or damage or with duodenal ulceration. However, in each case, a trend toward such an association was evident, although the size of the effect was invariably smaller for the corpus than for the antrum. The reduced size of the effects in the corpus together with reduced reproducibility of H. pylori density measurement in this region could explain why the effects were not formally statistically significant. That they were present in each case and always mimicked the antral associa-

5 556 Atherton et al. JID 1996; 174 (September) p= 't:: ~ -S2:=> ~LL 1- :t 8 -~ 10 I I 6 CO ~.- (/J... c:: c:: OJ « DU nodu (n=17) (n=12) Duodenal ulcer disease Figure 3. Relationship between antral H pylori density, as assessed by quantitative culture, and duodenal ulcer disease, defined as documented past or present duodenal ulceration (DU). Bar, mean for each group. Difference between means was significant (P =.005, Student's t test). tion implies that they may truly exist and be demonstrable in a larger study. In this study, we have described methodology for in vivo quantitation of H. pylori density suitable for use in further observational studies in humans and also potentially suitable for mechanistic studies in animal models. This study confirms that our methodology is sound, provides a framework for building hypotheses, and yields a data set with which theories of H pylori pathogenesis must be consistent. Acknowledgment We thank Wendy Bundy-Young for help with figures. References 1. Peterson WL. Helicobacter pylori and peptic ulcer disease. N Engl J Med 1991; 324: Blaser MJ, Parsonnet J. Parasitism by the "slow" bacterium Helicobacter pylori leads to altered gastric homeostasis and neoplasia. J Clin Invest 1994; 94: Berg DE, Blaser MJ, Labigne A, Rappuoli R, Taylor DE, Trust TJ. Proposal for development of a uniform genetic nomenclature for Helicobacter pylori [abstract]. Gut 1995;37(suppl 1):A Figura N, Guglielmetti P, Rossolini A, et al. Cytotoxin production by Campylobacter pylori strains isolated from patients with peptic ulcers and from patients with chronic gastritis only. J Clin Microbiol 1989; 27: Tee W, Lambert JR, Dwyer B. Cytotoxin production by Helicobacter pylori from patients with upper gastrointestinal tract diseases. J Clin Microbiol 1995;33: Atherton rc, Cao P, Peek RM, Tummuru MKR, Blaser Ml, Cover TL. Mosaicism in vacuolating cytotoxin alleles of Helicobacter pylori. Association of specific vaca types with cytotoxin production and peptic ulceration. J BioI Chern 1995;270: Crabtree le, Taylor JD, Wyatt JI, et al. Mucosal IgA recognition of Helicobacter pylori 120 kda protein, peptic ulceration, and gastric pathology. Lancet 1991;338: Covacci A, Censini S, Bugnoli M, et al. Molecular characterization ofthe 120 kda immunodominant antigen of Helicobacter pylori associated with cytotoxicity and duodenal ulcer. Proc Nat! Acad Sci USA 1993; 90: Cover TL, Glupczynski Y, Lage LP, et al. Serologic detection of infection with cag.a" Helicobacter pylori strains. 1 Clin Microbiol 1995;33: Atherton rc, Tham KT, Peek RM, Blaser Ml, Cover TL. Specific Helicobacter pylori vaca genotypes are associated with presence of duodenal and gastric ulceration, and degree of gastric inflammation [abstract]. Gut 1995;37(suppl 1):A3. II. el-omar EM, Penman ID, Ardill JE, Chittajallu RS, Howie C, McColl KE. Helicobacter pylori infection and abnormalities of acid secretion in patients with duodenal ulcer disease. Gastroenterology 1995; 109: Kaneko H, Nakada K, Mitsuma T, et al. Helicobacter pylori infection induces a decrease in immunoreactive somatostatin concentrations of human stomach. Dig Dis Sci 1992; 37: Langdale-Brown B, Haqqani MT. Acridine orange fluorescence, Campylobacter pylori, and chronic gastritis. Scand 1 Gastroenterol 1990;25: Bayerdorffer E, Lehn N, Hatz R, et al. Difference in expression ofhelicobacter pylori gastritis in antrum and body. Gastroenterology 1992; 102: Satoh K, Kimura K, Yoshida Y, Kasano T, Kihira K, Taniguchi Y. A topographical relationship between Helicobacter pylori and gastritis: quantitative assessment of Helicobacter pylori in the gastric mucosa. Am J Gastroenterol 1991; 86: Alam K, Schubert TT, Bologna SD, Ma CM. Increased density of Helicobacter pylori on antral biopsy is associated with severity of acute and chronic inflammation and likelihood of duodenal ulceration. Am 1 Gastroenterol 1992;87: Khulusi S, Mendall MA, Patel P, Levy l, Badve S, Northfield TC. Helicobacter pylori infection density and gastric inflammation in duodenal ulcer and non-ulcer subjects. Gut 1995;37: Peek RM, Miller QQ, Tham KT, et al. Heightened cytokine expression and inflammatory response in vivo to caga + Helicobacter pylori strains. Lab Invest 1995; 73: Bland JM, Altman DQ. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986; 1:307-IO. 20. Altman DG. Practical statistics for medical research. London: Chapman and Hall, 1991: Cuzick J. A Wilcoxon-like test for trend. Stat Med 1985;4: Bayerdorffer E, Oertel H, Lehn N, et al. Topographic association between active gastritis and Campylobacter pylori colonization. J Clin Pathol 1989; 42: Maaroos HI, Kekki M, Villako K, Sipponen P, Tamm A, Saeniemi L. The occurrence and extent of Helicobacter pylori colonization and antral and body gastritis profiles in an Estonian population sample. Scand 1 Gastroenterol 1990;25: Chan WY, Hui PK, Leung KM, Thomas TM. Modes of Helicobacter colonization and gastric epithelial damage. Histopathology 1992; 21: Kirschner DE, Blaser MJ. The dynamics of Helicobacter pylori infection in the human stomach. J Theor BioI 1995; 176: