Serology and13C-urea breath test have been widely used as noninvasive tests to detect Helicobacter pylori infection. However, easier collection of samples and lower costs are desirable for diagnosis of the individual patient or for use in epidemiologic studies. Our aim was to study the diagnostic accuracy of a recently developed urine-based enzyme-linked immunosorbent assay (ELISA) kit for the detection of H pylori-specific immunoglobulin G (IgG) antibodies in children.
Specimens of serum and randomly voided urine were collected from 816 children (0–15 years old) and were analyzed using 2 serum-based ELISA kits and a urine-based ELISA kit, respectively. Based on results of serology, the sensitivity, specificity, and accuracy of the urine-based ELISA kit were evaluated. With regard to false-positive and false-negative results, urinary IgG concentrations and IgG/creatinine levels were studied.
Both serum-based ELISAs were positive in 41 children and were negative in 666, who were enrolled in this study. The remaining 109 children were excluded because of disagreement between the results of the 2 serum-based ELISAs, including indeterminate values. Overall sensitivity, specificity, and accuracy of urine-based ELISA test compared with serology were 85.4%, 95.5%, and 94.9%, respectively. On positivity rates, the urine-based ELISA was closely coincident with the serum-based ELISA in each age group. There was no correlation between antibody levels detected by urine-based ELISA and each serum-based ELISA. Urinary IgG concentrations and IgG/creatinine levels were significantly higher in false-positives and were lower in false-negatives than in true-positives plus true-negatives for serology. Most of those with false-positive results had trace to moderate proteinuria.
The urine-based ELISA is an alternative to serum-based ELISA for diagnosis of H pylori infection in children and should be suitable for large-scale epidemiologic studies concerning the organism. In children with proteinuria, results of the test should be interpreted with caution. It is possible that the urine-based ELISA method would be applicable to diagnosis of other infectious diseases.
In the June issue of Pediatrics, Kato et al. reported on the diagnostic accuracy of a urine-based enzyme-linked immunosorbent assay (ELISA) kit for the detection of Helicobacter pylori specific immunoglobulin G antibodies in children. This study raises important methodological problems.
As no single test has yet proved accurate enough to be used as the gold standard, the validation of a new test (i.e., the urinary test) should need the concordance of at least two tests among biopsy-based (culture, histology, and urease quick test) and noninvasive (urea breath test, stool test, and antibodies) tests for classifying subjects as H pylori positive or negative. This allows to minimize the possibility of false-negative and false-positive results. Nevertheless, authors in their study used the concordant results of two serum commercial ELISAs, whose sensitivity and specificity is not given, as the gold standard. Furthermore, the number of children seropositive for H pylori is too small (n=41) relatively to that of the seronegative ones (n=666) in order to obtain a reliable diagnostic performance of the urinary test. Indeed, eventhough 46% of urine-positive children (30 out of 65) are false- positive, the low proportion of seropositive children (5.7%) overcome this important finding leading to 95% specificity. On the other hand, the absolute small number of seropositive children negatively affect the 95% confidence interval (70-94% are the correct figures instead of 83-88%, as reported at the bottom line of Table 1) of the sensitivity of the urinary test. Finally, data on the reproducibility (i.e., intra and inter-assay coefficient of variation) of this new test is not given.
Collectively, the above concerns negatively affect the validity of the study.