As previously described[15], between January and July, 2000, 800 consecutive patients referred to the endoscopy service at the Calderón Guardia Hospital in San José for dyspeptic symptoms were interviewed. This hospital is a tertiary hospital that, however, also provides gastroenterology service at primary and secondary levels. Patients were excluded if they were under 18 years of age, had not resided in Costa Rica for the previous two years, had received H pylori eradication therapy, had taken antibiotics during the 3 mo preceding endoscopy, had taken bismuth compounds at the time of endoscopy, had a history of gastric surgery, suffered from diseases associated with coagulation, or presented digestive bleeding. A total of 501 patients fulfilled the inclusion criteria and signed an informed consent form before 25 mL of fasting blood was obtained. The project had been previously approved by the ethic committees of the University of Costa Rica and the Hospital. A questionnaire with information concerning sociodemographic factors, family history, general health status, consumption of salt, coffee, alcohol, fruit, vegetables, and smoking habits was filled out[15]. Blood pressure and height and weight were recorded to calculate the body mass index.

Seven biopsies were obtained: Two biopsies from the middle part of the antrum were taken from the major and minor curvatures and two from the gastric body at the anterior and posterior walls of the central body for histopathological diagnosis. One biopsy from the antrum and one from the body were taken for culture and one from the antrum for rapid urease test[15].

Sections of formaline-fixed and paraffin-embedded biopsies, stained with hematoxylin and eosin, were used for the histopathological diagnosis according to the Sydney Classification[16]. All slides were independently read by two specialists in gastric pathology (JA Ramírez in San José, Costa Rica, and A De Mascarel in Bordeaux, France). Cases with discrepant diagnoses were reviewed by both pathologists until a consensus was reached. A patient was considered to have: (1) atrophic body gastritis (ABG), if atrophy or metaplasia was present in any of the biopsies from the body; (2) atrophic antral gastritis (AAG), when atrophy was found in any of the biopsies from the antrum and not in the body; (3) non-atrophic gastritis (NAG), in the case of gastritis without atrophy, and (4) Normal mucosa, when no pathology was found in the antrum or the body and no granulocytic activity or lymphoid follicles were observed. In 22 of the patients, there was no histopathological diagnosis due to insufficient tissue in the biopsies[15].

Infection with H pylori was determined by histology, culture and serology. For histological detection, the slides were stained with toluidine blue. One biopsy from the antrum and one from the body were cultured for H pylori. Biopsies from the antrum and corpus were ground in brucella broth with an electric homogenizer. The suspensions were plated on two in-house media: A Wilkins Chalgren agar containing 10% human blood and antibiotics and a Columbia blood agar without antibiotics incubated for 10 d in a microaerobic atmosphere at 37°C. Standard identification was performed. Serum antibodies to H pylori were measured by an in-house enzyme linked immunosorbent assay (ELISA) developed in our laboratory and based on a modification of a previously described ELISA[17]. The antigen preparation and determination of cutoff points was previously reported[15]. A patient was considered positive for H pylori when positive by any of the three methods: culture or histology or serology and negative when none of the methods were positive.

The cagA status was determined by polymerase chain reaction (PCR) on isolated strains using primers cagA₁-A₂ and, if negative, a second set of primers, cagA₃-A_4, were used[18]. Serum antibodies to CagA were measured by ELISA as described by Blaser et al, 1995[19]. A patient was considered positive for H pylori CagA⁺ infection when positive by PCR and/or serology.

The concentrations of PGI and PGII in sera were measured by an enzyme immunosorbent assay (EIA) (Eiken Chemical Company, Tokyo, Japan) according to the manufacturers’ recommendations. The validation of pepsinogen levels for this population was described previously[15]. In order to detect as many patients as possible with ABG, a cut-off point that favored sensitivity was selected. The optimal cut-off point for this population had been set at a PGI/PGII ratio of 3.4, which gave a sensitivity of 91.2% and a specificity of 38.5%[15]. The predictive value for a positive sample was 11.2% and for a negative 98.1%. Pepsinogen levels could not be measured in 51 patients, because insufficient amounts of blood were obtained or because the patients did not accept being bled.

IL-1B polymorphism analysis was performed by PCR-restriction fragment length polymorphism (PCR-RFLP). 100 ng of genomic DNA were amplified with the primers and PCR conditions previously reported[20]. The PCR products for IL-1B + 3954 were digested with the restriction enzyme TaqI and the allele designation for this polymorphism was the same as previously reported[21].

Genomic DNA (100 ng) was amplified using the same primers and PCR conditions as previously reported for IL-1RN[20]. For statistical analysis purposes and because of the low frequency of alleles 3, 4 and 5, this polymorphism was treated as biallelic by dividing alleles into short and long categories, in which the short allele has two repeats (allele 2) and long allele has more than two repeats (allele L). For 110 (22%) patients genotyping was not done because of insufficient leukocyte samples or problems in the isolation or analysis of the DNA.

χ² statistics were used for comparing genotype frequencies among the groups studied and to assess Hardy-Weinberg equilibrium for each of the loci studied. Allelic frequencies were assessed using Estimating Haplotype Frequencies (EH) (available at ftp://linkage.rockefeller.edu/software/eh/). Politomic logistic regression was used to compare genotypic frequencies among the groups studied here using STATA/SE 8.0 (STATA Corporation, College Station, TX).

A logistic regression model was used to calculate the odd ratios (ORs) for (dependent variable): (1) ABG compared to dyspepsia without atrophic gastritis; (2) AAG compared to dyspepsia without atrophic gastritis; (3) Gastric ulcer compared to all of the others; (4) Duodenal ulcer compared to the rest; (5) Levels of PGI < 60 μg/L compared to levels of PGI > 60 μg/L; and 6) Values of PGI/PGII < 3.4 compared to values of PGI/PGII > 3.4. The systematic independent variables were H pylori CagA infection, IL-1B + 3954 allele T carriers, IL-1RN allele 2 carriers, and those that had been associated with the dependent variables in a previous study of the same population: Age, sex, overweight, frequency of alcohol and fruit consumption and the PGI/PGII ratio for atrophic gastritis and age; gender and cigarette smoking for peptic ulcers[15]. The logistic regression analyses were performed including either, only H pylori infected individuals or, all of the participants. The association was determined as an OR at a confidence interval of 95%. Statistical significance was set at P < 0.05. The software STATA/SE 8.0 (STATA Corporation, College Station, TX) was used for the statistical analysis.

PGI y PGII concentrations could be measured in 450 patients. Serum concentrations of PGI as well as the PGI/PGII ratio were lower in patients with ABG as compared to other dyspeptic patients (Table 1). Dyspeptic patients with a ratio below 3.4 were at increased risk of having ABG compared to those with a ratio higher than 3.4 (Table 2).

The frequency of H pylori infection was 86%, out of which 67.8% were CagA⁺, however, the prevalence varied according to the pathology. Patients with a normal histopathological diagnosis, although few, showed a lower incidence of infection with H pylori and H pylori CagA⁺ (15.0% and 5.3%, respectively) as compared to the rest, NAG (88.9% and 55.5%), AAG (93.3% and 81.6%) or ABG (82.9% and 58.8%). The prevalence of infection with H pylori and H pylori CagA⁺ for gastric ulcer was 100% and 52.2%, respectively, and for duodenal ulcer 98% and 85%, respectively. Among patients with PGI/PGII < 3.4, the prevalence of infection with H pylori was 87.7% and with H pylori CagA⁺ 61.3% while that of patients with ratios > 3.4 were at 82.1% and 50.0%. H pylori CagA+ was associated with a PGI/PGII ratio < 3.4 and PGI < 60 μg/L (Table 3).

Genotyping of the cytokine polymorphisms was successfully performed in 371 patients (74%). Genotypic distributions of IL-1RN and IL-1B + 3954 polymorphisms were not significantly different among the studied groups (Table 4). There was Hardy-Weinberg equilibrium for the IL-1RN locus in the groups used here as controls (dyspepsia without atrophic gastritis and all of the groups together excluding ABG). However, in the case of the IL-1B + 3954 polymorphism, there was no equilibrium for any of the control groups (data not shown). The frequency of the allele 2 for IL-1RN polymorphism was not significantly different in normal patients, compared to those with some type of gastritis (Table 4).

An increased risk of ABG was observed among patients older than 50 years, those consuming alcohol and those with a PGI/PGII ratio < 3.4. Patients infected with CagA⁺ strains of H pylori and those with lower consumption of fruit were at higher risk of developing AAG, (Table 2). The presence of pro-inflammatory alleles, IL-1B + 3954T and IL-RN*2, did not confer an enhanced risk of any type of atrophy (Table 2).

Duodenal ulcer was more frequent than gastric ulcer (52/24) in both genders. However, it was twice as common in males and the average age of patients with duodenal ulcer (average 46; 95% IC, 42.5-50.2) was lower than that of those with gastric ulcer (average 58; 95% CI, 52.2-63.6). Males, smokers and patients infected with H pylori CagA⁺ were at increased risk of duodenal ulcer (Table 5).

Costa Rica has a high incidence of and mortality from gastric cancer. Infection with H pylori and proinflammatory interleukin (IL) gene polymorphisms have been associated with precancerous gastric lesions and gastric adenocarcinoma. Atrophic gastritis may lead to the development of cancer. This study addresses the association of atrophic antral and body gastritis with infection with H pylori-CagA+ strains and IL-1 gene polymorphisms in a dyspeptic population in an area at high risk of gastric cancer in Costa Rica.

In a dyspeptic population in Costa Rica, H pylori CagA⁺ is not associated with atrophic body gastritis, but it is a risk factor for atrophic antral gastritis (AAG). The pro-inflammatory cytokine polymorphisms IL-1B + 3945 and IL-1RN are not associated with the atrophic lesions.

Studies of various populations have indicated an association between gastric cancer and H pylori-CagA as well as proinflammatory gene variants. The etiology of gastric cancer varies among populations. The present study shows that H pylori-CagA is related to active AAG, but not to more advanced atrophy of the gastric body in a high-risk population. It is speculated that the bacteria may have disappeared at this stage. Furthermore, the study demonstrates no relation between two proinflammatory IL-1 gene polymorphisms and atrophy of the stomach in dyspeptic patients.

In developing countries with high incidence of H pylori infection and dyspepsia it is not feasible to screen large populations with endoscopy. The information generated here may be used to create a battery of markers to detect risk factors for gastric cancer from a blood sample. Screening and intervention may then be concentrated on people with a high-risk profile.

The report is interesting. It could be useful in clinical settings. It indicated that in a dyspeptic population in Costa Rica, H pylori CagA⁺ is a risk factor for AAG.