INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Helicobacter pylori is a common human gastric pathogen causing chronic gastritis and duodenal ulcers (6, 12). There is strong evidence that H. pylori infection is also associated with gastric cancers and gastric lymphomas (6, 13). Both the prevalence of H. pylori infection and the incidence of gastric cancer are higher in Asia, South America, and the Caribbean than in Europe and the United States. The prevalence of H. pylori infection is also higher among blacks than among caucasians in the United States (10). Because H. pylori infection persists for life in the absence of treatment (6), its clinical sequelae continue to present a major public health burden in areas in which this bacterium is endemic.

One of the challenges in epidemiologic studies of H. pylori infection has been the population-specific performance of serologic assays, which has made it difficult to interpret existing data across populations. Variations in bacterial genotype, antigen selections for the immunoassays employed, and host immune responses may affect the performance of serologic assays and their suitability for particular populations.

In the present study of Jamaican patients, we evaluated the performance of two commercial enzyme-linked immunosorbent assays (ELISAs) for immunoglobulin G (IgG) antibody to H. pylori and a research ELISA which had been validated in epidemiologic investigations of populations from diverse geographic regions (5, 7, 26). We also describe here the relationship of H. pylori infection to clinical and pathological findings and the molecular genotypes of Jamaican H. pylori strains.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Study subjects. We evaluated 30 sequential adult patients who underwent diagnostic gastroduodenoscopy for various upper gastrointestinal symptoms at the Gastroenterology Service of the University Hospital of the West Indies, Kingston, Jamaica. Patients with the following backgrounds were excluded from this study: history of cardiac, neurological, or pulmonary diseases precluding safe procedure; immunodeficiency; and/or antibiotic therapy during the month before the procedure. Antisecretory medications were withheld from these patients for at least 2 weeks before endoscopy. No patients had previously received H. pylori eradication therapy. The study protocol was approved by the institutional review boards of the National Cancer Institute and the University Hospital of the West Indies. Written informed consent was obtained from all patients. Experienced nurses collected demographic and clinical data at initial study enrollment, before H. pylori infection status was determined. From each subject, a total of seven biopsy specimens were obtained during endoscopy; two samples (fundus and antrum) of each were submitted for histopathological analysis, rapid urease test, and primary culture. One additional biopsy specimen was obtained from the duodenum. Biopsy specimens for histopathological analysis were fixed in 10% buffered formaldehyde and embedded in paraffin for sectioning. Other biopsy specimens were placed into a sterile cryovial (Nalgene 1 ml) with Trypticase soy medium and 20% glycerol and were frozen until used. In addition, 10 ml of blood was drawn from each subject. Serum samples were stored at 70°C until subsequently used.

Laboratory methods. (i) Rapid urease test and histological analyses. The rapid urease test (CLO-test; Trimed Laboratories, Draper, Utah) was performed on fresh biopsy specimens in the endoscopy suite, following the manufacturer's specifications. The results were read by an experienced nurse and confirmed by a physician (M.G.L.). Biopsy specimens were assessed for the presence of inflammation, as well as for H. pylori (via hematoxylin-eosin and Warthin-Starry stains), by an experienced pathologist (B.H.), who was blinded to the rapid urease test, H. pylori culture, and serologic results. Six sections per biopsy sample were inspected using the high-power (40×) objective. When no organisms were identified at this magnification, the sections were further examined with an oil immersion objective (100×). Specimens were graded according to the updated Sydney classification system for gastritis (1). Scores of 1 in the "mononuclear cell component" indicate chronic gastritis, with the normal (score = 0) being the presence of two to five mononuclear cells per high-power field (40×) in the lamina propria.

(ii) Primary culture. After inoculation onto both selective (Skirrows; Sigma, St. Louis, Mo.) and nonselective (brain heart infusion agar with 5% sheep blood; Centers for Disease Control and Prevention, Atlanta, Ga.) solid-medium plates, incubations at 37°C for 5 to 10 days under microaerobic conditions (5% O₂, 10% CO₂, 85% N₂) were performed until small, translucent colonies consistent with the morphology of H. pylori were obtained. Biochemical analyses for catalase, oxidase, and urease were performed. A dark-field examination with a flagellum stain was used to confirm the presence of H. pylori at each site before freezing and storage of specimens (20).

(iii) Serology. Over 70 H. pylori strains of various genotypes were utilized for the development of the research ELISA (5, 7). Outer membrane proteins and whole-cell antigens from a single strain and pooled strains were employed in the initial assay development process (5, 7). In addition, in order to attain improved assay performance, both outer membrane proteins and whole-cell protein antigens of H. pylori isolates obtained from the Jamaican patients in the present study were purified and employed in our research ELISA. Bacteria were grown overnight in Brucella broth (Life Technologies, Gaithersburg, Md.) with 10% fetal bovine serum (Sigma), 5 µg of trimethoprim/ml, and 10 µg of vancomycin (Sigma)/ml. Antigen extraction and protein isolation were done by gentle freeze-thaw sonication (Misonix Sonicator XL-2015; Heat Systems Inc., Farmingdale, N.Y.) (7, 15). A standard protein assay (Pierce, Rockford, Ill.) was used to determine the accurate and reproducible quantity of solid-phase antigen for our microtiter research ELISA (17).

(iv) Molecular genotyping. Genomic DNA from H. pylori was isolated under standard conditions (24). Analyses of vacuolating cytotoxin gene vacA (s and m regions) and of cytotoxin-associated gene cagA were performed by PCR techniques, using one microliter of DNA from an H. pylori culture lysate (21). PCR products from the vacA s and m regions as well as from cagA were analyzed simultaneously by reverse hybridization, using a line probe assay (19). Allele-specific PCR assays were used for analysis of iceA1 and iceA2. Among iceA2 strains, different subtypes were distinguished on the basis of the size of the iceA2-specific amplimer, as described in detail elsewhere (4). IceA amplimers were examined by electrophoresis on a 2% agarose gel according to standard procedures (18).

Statistical analysis. For the purpose of this analysis, H. pylori positivity was determined by the detection of the organism by one or more biopsy-based methods. Multiple biopsy sections were examined by histopathologic methods in order to minimize false-negative results due to sampling errors. Samples giving indeterminate OD values in ELISAs were considered H. pylori negative. Concordance of the results of the research ELISA and those of two commercial assays and intraassay variability (reproducibility) were assessed by using Spearman's correlation coefficient. Concordance of the biopsy-based H. pylori positivity and seropositivity determined by the research ELISA was examined by the ² test or Fisher's exact test. Sensitivity of serologic assays was determined as the proportion of biopsy-positive patients who are also seropositive; specificity was measured as the percentage of biopsy-negative patients who were also serologically negative. P values were two sided, with statistical significance set at the 0.05 level.

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Thirty symptomatic gastroendoscopic patients (17 males and 13 females) were enrolled in the study. The mean age of these patients was 53 (range, 23 to 84). Twenty-five (84%) of the 30 patients were black. Overall, H. pylori was detectable by one or more biopsy-based methods in 18 (60%) of the 30 patients. Among these 30 patients, the H. pylori positivity values determined by culture, rapid urease test, and histologic examination were 50%, 50%, and 40%, respectively. Overall, 21 (70%) of the 30 patients were positive for H. pylori IgG antibodies by our research ELISA.

Correlation among the triplicate OD values of the research ELISA was greater than 0.95 (P = 0.0001), suggesting excellent assay reproducibility. The results obtained by the three serologic assays were concordant for >70% of the samples. The sensitivity and specificity of the research ELISA were determined to be 100% (18 of 18) and 75% (9 of 12), respectively, in an analysis that used biopsy-based H. pylori positivity as the "gold standard." The sensitivities and specificities of two commercial assays, calculated in the same manner, were identical (100% and 50%, respectively). Additionally, we employed proteins from H. pylori isolates obtained from Jamaican patients in our research ELISA and compared its performance with that of the original research ELISA and two commercial assays. The research ELISA using Jamaican H. pylori antigens did not demonstrate any increased accuracy compared with the others.

Twenty-eight of the 30 patients had biopsy specimens evaluable by histopathologic methods (Table 1). Specimens from two patients with gastric cancer were not evaluable by the Sydney scoring method. Histopathology consistent with a diagnosis of chronic gastritis was present in 19 (68%) of the 28 evaluable patients, of whom 17 had gastric atrophy and 7 exhibited metaplasia. One patient with metaplasia also had a gastric cancer. Of 21 patients who were H. pylori antibody positive by our research ELISA, three were H. pylori negative by all biopsy-based methods. Two of these three patients had gastric adenocarcinomas; the other had normal histology (Sydney score = 0/9).

H. pylori was successfully isolated and cultured from 15 patients, including 2 patients who did not have detectable H. pylori by either the rapid urease test or histologic examination of the biopsy specimen (JHP022 and JHP028). As shown in Table 2, the cagA⁺ iceA2 vacA slb-ml genotype was found in 7 of the 15 patients (JHP002, -008, -011, -013, -019, -025, and -028). Two subjects were concurrently infected with a vacA sla strain, a genotype common in North America, Europe, and Australia but rare in South America (23, 25), in addition to a predominant slb strain. The vacA s genotype appeared to be absent from three isolates. Twelve of 15 (80%) isolated strains were cagA positive. Only one strain exhibited an iceA1 genotype.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The epidemiology of H. pylori infection in the Caribbean islands remains an important agenda for public health investigation because of the high prevalence of this infection and its association with gastric cancer. Although a few studies have evaluated H. pylori infection in this population by utilizing commercially available serologic assays, the resultant data are difficult to interpret because of differences in the performance of serologic assays across different populations. To understand H. pylori transmission and disease pathogenesis in this part of the world, the development of a validated, sensitive and specific serologic assay is critical. A serologic assay has an important advantage over endoscopy-based methods for large population-based epidemiologic studies because it is noninvasive and easily employed.

In the present study, we evaluated the performance of our research ELISA in two independent laboratories, using data from patients with gastrointestinal symptoms, and found excellent reproducibility and minimal intralaboratory variation for our research ELISA results. While the sensitivities of all three serologic assays were perfect, the specificity of the research ELISA was higher than those of the two commercial assays, indicating that the use of our research ELISA would minimize false-negative results. The accuracy of our research ELISA in the biopsy-based detection of H. pylori has been previously validated in many asymptomatic and symptomatic populations around the world in the same manner, with sensitivities of 89 to 96% and specificities of 92 to 97% (5, 7, 26). The use of a Jamaican H. pylori antigen in our research ELISA in the present study did not increase the accuracy of the assay, further confirming an excellent performance of our research ELISA for various H. pylori strains across different populations.

In our series, 18 (60%) of the 30 patients were positive for H. pylori by one or more biopsy-based methods, whereas 21 (70%) of our patients were positive by serologic assay. The seroprevalence of 70% is similar to previously reported values for H. pylori infection in Jamaica and Barbados (2, 8, 9). Two of the three discordant results of H. pylori infection status determined by serologic and biopsy-based methods were seen in gastric cancer patients with chronic atrophic changes. In both patients, all biopsy-based methods were negative for H. pylori, while serology was positive. These observations are consistent with the hypothesis that H. pylori may be no longer detectable in tissue in the presence of chronic gastric atrophy. In previous studies, the presence of gastric atrophy, which is thought to precede gastric carcinoma, has been observed to result in a decrease of the H. pylori load and a subsequent decline in levels of IgG antibodies to H. pylori (11, 14). Colonization of H. pylori is also at times less dense and has a different distribution (i.e., antrum versus body predominant) in achlorohydric patients, including those treated with acid inhibitors, resulting in false-negative results of either the rapid urease test, histological examination, or primary culture (3). In addition, sampling errors during biopsy may result in false negatives.

Thus, in clinical settings in which a high prevalence of H. pylori-related pathology is expected, any one biopsy-based method may not be sufficient to reliably detect infection. In such instances, the use of additional confirmatory methods, such as breath tests and stool antigen detection tests, in identifying truly H. pylori-infected cases may be recommended. The breath test exploits the urease enzyme produced by H. pylori, with H. pylori infection status being determined by detection of ¹³C- or ¹⁴C-labeled CO₂ in the expired air subsequent to ingestion of ¹³C- or ¹⁴C-labeled urea. The stool antigen test detects H. pylori DNA in the stool by using H. pylori-specific antibodies in an enzyme immunoassay. Although these noninvasive tests are highly sensitive and specific, they are logistically difficult to employ in large population-based studies. In contrast, our serologic test was found to be easy to employ and detected all biopsy-based H. pylori-positive cases as well as cases with H. pylori-associated pathology which could not have been detected by any biopsy-based method. Furthermore, only one patient with positive serologic results in our series had no apparent histopathological abnormalities, suggesting that the probability of having a false-positive serologic result is low. Taken together, serologic tests appear to be a better tool than biopsy-based methods for detecting H. pylori infection in epidemiologic studies of populations in which this bacterium is endemic.

The molecular characteristics of H. pylori infection in Jamaica had not been well described prior to this study. Our investigation indicated that the predominant H. pylori strain in Jamaica, including the one isolated from a patient with gastric adenocarcinoma (JHP025), has the genotype cagA⁺ iceA2 vacA slb-ml. The predominance of the cagA⁺ vacA slb-ml genotype is consistent with the findings for other populations from Central and South America. Two subjects were concurrently infected with a vacA sla strain, a genotype rare in South America but common elsewhere, including North America, Europe, and Australia (23, 25). The lack of a vacA s genotype in three isolates in the present study is currently unexplained but may have been caused by the existence of additional vacA s molecular variants, ones not readily detectable by the PCR-line probe assay. Recent studies suggest that the genotype of H. pylori potentially correlates with the severity of gastroduodenal disease associated with this infection and that the genotype distribution differs by geographic region (6, 22). The association of the observed genotypes and their impact on disease manifestation remains to be described further in a larger patient series.

In sum, the present study demonstrates that serum antibodies to H. pylori are a useful marker for epidemiologic studies of this infection in Jamaica and that the performance of our research ELISA for this purpose is excellent. Because levels of antibodies against H. pylori may wane after bacterial eradication or under conditions resulting in gastric atrophy and hypo- or achlorohydria, analysis of serum samples collected prior to diagnosis to establish causal associations is desirable. Thus, the pathogenesis of H. pylori and the risk of its transmission in Jamaica and elsewhere need to be evaluated in a prospective study.