Helicobacter pylori infection is a common infection that affects human beings. Its incidence is estimated to be 0.5% per year in developed countries and 3 to 10% per year in developing countries. Also, its incidence throughout the world is higher in children than in adults. H. pylori are capable of causing a number of gastrointestinal diseases including: acute and chronic gastritis, atrophic gastritis, gastric and duodenal ulcer, intestinal metaplasia, lymphoma and even gastric adenocarcinoma. High prevalence of the infection and its curable pattern and also the importance of problems caused by this infection make its diagnosis very important. Moreover, many complications of the chronic H. pylori infection could be prevented by ideal treatment (1-5). There are several ways to diagnose H. pylori and they are divided into two main groups: invasive methods including culture, histological evaluation and rapid urease test and non-invasive methods including urea breath test, serology and evaluation of stool antigens (6-11).

Since Iran is a developing country with a high prevalence of H. pylori infection, accurate and economic diagnostic methods are needed. Diagnosis of this infection in our center is now based on the combination of histological evaluation and rapid urease test on samples obtained from gastroscopy however these methods are invasive and difficult in children. As there is not a cut-off point for antibodies in children and the serological method is a simple, economic and non-invasive method, we aimed to evaluate the efficacy of this method for diagnosis of H. pylori infection in children and if possible define a cut-off point for related antibodies in order to use them in diagnosis of this infection.

During a two-year period, 104 children younger than 18 years old with clinical suspicion of H. pylori infection were entered in the study. Suspicion to H. pylori infection was defined as having dyspepsia, pain or discomfort in the upper abdomen, fullness, early satiety, anorexia, eructation, nausea and vomiting. All cases underwent upper gastrointestinal (GI) endoscopic evaluation and tissue samples were taken from the antrum. All samples were evaluated for H. pylori histologically using hematoxylin and eosin and the modified Giemsa method. Accompanied with endoscopic evaluation rapid urease test was also done. During the 24 hours after the test, if peripheral color changed to pink, the test results were interpreted as positive or presence of H. pylori was assumed.

We considered these children infected with H. pylori if they had both positive histology and rapid urease test as the gold standard. Overall, 77 patients (74%) had positive rapid urease test and histology for H. pylori and were considered as the H. pylori infected group. Of the 27 reminders 16 patients (59.3%) had positive histology but negative rapid urease test and 11 patients (40.7%) had positive rapid urease test but negative histology and all of them were excluded from the study. Controls were also younger than 18 years, without any gastrointestinal (GI) symptoms and clinical suspicion of H. pylori infection. Controls and their first-degree families were also free of any GI symptom during the study period and their entire life, before the study. Neither cases nor controls received antibiotics during the last week of the study.

It was supposed that all controls were free of H. pylori and they did not undergo endoscopic evaluation due to ethical issues. However, blood samples were obtained from all subjects (cases and controls) and checked for three antibodies against H. pylori. The IBL Hamburg kits for each antibody were obtained from Germany. Using the enzyme-linked immunosorbent assay (ELISA) (sandwich principle), titers of IgA, IgM and IgG antibodies were detected. Therefore, 77 cases were evaluated by the rapid urease test, histology and measuring IgM, IgG, and IgA antibody titers against H. pylori. However, controls were only checked for IgM, IgG, and IgA antibody titers against H. pylori. All data were entered into Microsoft Excel 2003 and were analysed by the ROC-analysis method and finally three cut-off points for IgM, IgG, and IgA were obtained.

There were 42 (45.4%) boys and 35 (54.6%) girls in the case group. According to age, 6 patients (7.7%) were between 1 and 5, 34 patients (44.1%) between 6 and 10, 26 patients (33.7%) between 11 and 16, and 11 patients (14.2%) were between 16 and 18 years. The most common clinical symptoms were abdominal pain (96%), anorexia (82%), nausea (57%), heartburn (31%) and vomiting (23%). Sixty-two patients (80.5%) had used medicines such as antacid, cimetidine, ranitidine and omeprazole. Four patients (5.2%) had undergone an endoscopic examination in the past and one of them had a history of GI bleeding. Due to GI problems, 27.2% of the parents of cases underwent endoscopic examinations and 2.4% reported a history of GI bleeding; also, 52% reported a history of use of gastric acid suppressor medications.

Using ROC-analysis, three cut-off points for IgM, IgA and IgG were obtained. The cut-off point for IgA was 3.3 U/ML, for IgM 6.4 U/ML and for IgG 9.9 U/ML. To calculate sensitivity, specificity and accuracy, positive and negative predictive values for each of these antibody titers above these points were supposed positive and those below these points as negative. With 3.3 U/ML as the cut-off point for IgA, sensitivity, specificity, accuracy, positive and negative predictive value for this antibody was 64, 58, 59.3, 36.7 and 81.1%, respectively. IgM had 76% sensitivity, 36.1% specificity, 74.2% accuracy, 31.5% positive predictive value, and 76.9% negative predictive value with 6.4 U/ML as the cut-off point. Furthermore, 58.6% sensitivity, 61.3% specificity, 60.6% accuracy, 36.9% positive predictive value and 79.3% negative predictive value was obtained for IgG with the cut-off point equal to 9.9 U/ML.

In this study histological evaluation and rapid urease test were considered as the gold standard and results of serological tests in H. pylori infected group were compared with the control group. IgA, IgM and IgG have a high negative predictive value that could be used (scale) for a screening and diagnostic test. For all three antibodies, the negative predictive value was about 80%. These findings show that negative results have a higher value compared to positive results. When the test is negative with 80% probability, the person is not infected by H. pylori, while when it is positive, IgG has the highest (60.6%) and IgM the lowest (42.2%) accuracy. IgA accuracy is about 59.3% and the most specific antibody is IgG and the most sensitive is IgM.

Since Iran is a developing country with high prevalence of H. pylori infection, it is ideal to evaluate all controls by endoscopy to see how many of them were infected by H. pylori and not suppose that all are free of H. pylori. Therefore, the number of false positive and negatives would decrease and sensitivity, specificity, accuracy and positive and negative values of antibodies would increase. However, because of ethical issues this was impossible. Predictive value of a test is dependent on sensitivity, specificity and prevalence. In this study we could not estimate the exact prevalence of H. pylori with the small sample size. Thus, a study with a large sample size should be performed in Shiraz city to estimate the prevalence and predictive values.

A great deal of studies has been conducted in different countries. In an adults study, serum IgG with cut-off point of 15.2 U/mL had 94.1% sensitivity and 97.9% specificity for diagnosis of H. pylori infection (12). In another adult study, IgG antibody titers higher than 10 U/mL were considered positive for H. pylori infection; cut-off point of 3 U/mL had 100% sensitivity and 99% specificity for eradication of H. pylori (13). A study in Spain reported sensitivity, specificity, positive and negative predictive values for IgG as 81, 97, 89, and 93% and for IgA as 90, 76, 36, and 96%, respectively, in children. However, IgM was not evaluated in this study (2). Another pediatric study in Italy showed sensitivity, specificity, positive and negative predictive values for IgG to be 86, 80, 72 and 90%, respectively (14).

In Brazil it was shown that sensitivity, specificity, positive and negative predictive values and accuracy for IgG was 64, 83.7, 82, 66.6 and 73.1% and for IgA this was 72, 65.9, 72, 67.4 and 69.8%, respectively (15). Different values for sensitivity, specificity, positive and negative predictive values have been reported by various studies. This could be due to differences in prevalence of H. pylori between populations because of the variety in socioeconomic status, health status, crowding and other factors. Differences in prevalence have an obvious impact on statistical indexes.

A serological assay for H. pylori infection may be useful in epidemiological survey of prevalence, transmission mode and spontaneous clearance of the H. pylori infection, while it also may be helpful in the development of preventive measures for H. pylori infection. In Asia, the strains of H. pylori are different from those that are prevalent in other continents. Thus, the specificity, sensitivity and positive and negative predictive values of one kind of serology kit may differ in various geographic or ethnic populations.

We concluded that these antibody tests have a relatively high negative predictive value and a low positive predictive value. Thus, serological tests are not useful on their own, for diagnosis of H. pylori infection because the sensitivity and specificity of these tests for detection of antibodies against H. pylori in children vary widely and they have not been recommended for clinical practice in pediatric patients. More researches with a larger sample size are needed in the future to confirm the findings of this study.