Background—The salivary diagnosis of Helicobacter pylori infection offers attractive possibilities for the epidemiological study of infection in children. Salivary enzyme linked immunosorbent assay (ELISA) is less reliable then serum ELISA, owing to variable transudation of immunoglobulin. In addition, children are more difficult to study because of lower specific serum antibody concentrations to H pylori. The performance of salivary western blotting in comparison with serum western blotting and serum ELISA was investigated in school children.
Subjects and methods—Paired serum and saliva specimens were obtained from 665 school children aged 9–11 in 10 British towns. All saliva and serum specimens were first analysed by ELISA; subsequently, western blotting of both specimens was performed on 31 and 34 specimens, respectively, to establish the criteria for positivity for western blotting. The remaining 121 specimens were then tested blindly and saliva was compared with the serum.
Results—The sensitivity and specificity of salivary ELISA in the 665 specimens was 32 of 50 (64%) and 530 of 691 (87%), respectively, when compared with serum ELISA. The western blotting validation was performed on 28 subjects with positive serum and positive salivary ELISA, 28 saliva positives with negative serum, 16 saliva negatives with positive serum, and 50 doubly negative subjects. Compared with serum western blots, the sensitivity and specificity of salivary western blots was 38 of 47 (81%) and 68 of 75 (91%), respectively. Using serum ELISA as the gold standard, the sensitivity and specificity were 32 of 44 (73%) and 72 of 78 (92%), respectively, the specificity being significantly higher than salivary ELISA (p < 0.001).
Conclusion—Salivary western blotting for IgG is useful in the diagnosis of H pylori infection and is superior to ELISA. It also permits the identification of pathogenic strains.
- Helicobacter pylori
- western blotting
- enzyme linked immunosorbent assay
- salivary specimens
Statistics from Altmetric.com
Helicobacter pylori is a chronic bacterial infection of the human stomach that is associated with peptic ulcer and gastric cancer, as well as with possible extra-intestinal manifestations. The diagnosis of H pylori can be made on biopsies obtained at endoscopy, but these are unsuitable for epidemiological studies, which have relied on serology and the urea breath test. Salivary antibody testing is a potentially useful alternative for epidemiological studies,1 especially in children, because it is non-invasive and no sample preparation is required.
The specific antibody response to H pylori can be detected using a variety of methods, but ELISA and immunoblotting (western blotting) for specific IgG are the most reliable.2,3 ELISA is more convenient than western blotting, but specific anti-H pylori IgG values tend to be lower in children than adults, and most assays have reported a slightly lower sensitivity when compared with adults.4
Although IgA is the predominant immunoglobulin present within saliva, poor sensitivity and specificity are obtained in comparison with IgG when assayed by ELISA.5 IgG gains access to saliva by transcapillary leakage, but values are approximately 1/1000 of those in serum. The low concentrations, together with variable transcapillary leakage, are the main reasons why salivary IgG determination is less reliable than serum determination. Immunoblotting is a sensitive and qualitative method for determining the presence of antibodies, and hence would be expected to overcome some of these difficulties. It also has the advantage of being able to detect antibodies to CagA and VacA, proteins that are markers of H pylori virulence.
Therefore, we investigated whether salivary western blotting was superior to salivary ELISA for the detection of IgG antibodies to H pylori compared with serum ELISA and serum western blotting. We also determined whether salivary western blotting had the ability to detect antibodies to VacA and CagA, as determined by serum western blotting.
Subjects and methods
Our report is based on the 8–11 year old examination of the “ten towns” children study. The general design of our study has been described elsewhere.6 Paired saliva and serum samples were available for 665 children from the last five towns screened; three with particularly low adult cardiovascular mortality (Leatherhead, Bath, and Tunbridge Wells) and two with particularly high cardiovascular mortality (Rhonda, Rochdale). ELISA antibody determinations were performed on both serum and saliva with the Launch Premier ELISA kitTM (Launch Premier, Longfield, Kent, UK). This has been shown previously to be 100% sensitive and 96% specific for serum in adults.7 Serum was tested according to the manufacturer's instructions. Optimum performance for salivary ELISA was obtained when undiluted saliva was incubated with IgG conjugate (A2290; Sigma, Poole, Dorset, UK) at a concentration of 1/1500.
The specimens selected for the validation study had an over representation of false positive saliva by ELISA to allow a better evaluation of specificity. All the seropositive specimens from the original 665 paired samples tested on ELISA with remaining saliva were included. In total, 122 paired saliva and serum samples were studied by western blotting. They comprised 28 subjects who were ELISA seropositive and salivary IgG positive, 16 subjects who were seropositive and saliva IgG negative, 28 subjects who were seronegative but saliva IgG positive (OD > 0.41), and 50 subjects who were seronegative and saliva IgG negative. A further seven samples were unreadable owing to gross contamination or insufficient saliva.
Criteria for a positive serum western blot were first established on 18 positive and 16 negative sera measured by ELISA, and for a positive salivary western blot on saliva specimens from 13 seropositive and 18 seronegative subjects by serum ELISA. When the criteria had been established the remainder of the samples were tested blindly.
Western blotting was performed as follows: 100 μg/ml of a crude lysate of a CagA VacA positive strain of H pylori was run on a 10–15% gradient gel and then blotted on to nitrocellulose for one hour at 20 V, 300 mAmps. The H pylori were grown on selective plates at 37°C in a controlled atmosphere (5% oxygen, 10% carbon dioxide) for five days and then washed and stored in phosphate buffered saline (PBS) at −70°C. The nitrocellulose was blocked overnight at 4°C in blocking solution (15 g Marvel milk in 200 ml PBS). The nitrocellulose was incubated with 5 μl serum and 1 ml saliva and control serum samples for one hour and at the dilutions stated below in TTBS (300 ml PBS/Tween, 9 g Marvel milk). The blots were washed in PBS/Tween (one litre PBS/4 ml Tween 20) three times for 10 minutes. The IgG conjugate (HRP conjugated goat antihuman IgG; catalogue number A2290; Sigma) was diluted at concentrations stated below in TTBS and incubated again for one hour. Washing was then performed four times for 15 minutes in PBS/Tween and the signal developed using the Amersham ECL (Amersham, Little Chalfont, Buckinghamshire, UK) detection system.
Conjugate, serum, and saliva concentrations were optimised using paired samples. The optimum dilution of saliva was 1/4 to minimise sample use and 1/500 for the IgG conjugate. The optimal concentrations for serum were a sample dilution of 1/5000 and IgG conjugate concentration of 1/6000.
Some saliva specimens contained an excess of debris on visual inspection and were centrifuged for five minutes at 5000 ×g before testing.
The sensitivity of saliva ELISA and saliva western blot test were compared by focusing on “true” positives as defined by serum ELISA and serum western blot, respectively. The sensitivity of the two tests was then compared using McNemar's test for paired samples with continuity correction. The specificity of the two saliva tests was similarly compared by focusing on the “true” negatives defined by the gold standard.
We calculated κ statistics for comparing the agreement for saliva and serum western blot using EPI INFO version 6.
PERFORMANCE OF SALIVARY ELISA VERSUS SERUM ELISA
Of the 665 sera and saliva samples tested by ELISA, 50 (8%) were seropositive (fig 1). The sensitivity and specificity of salivary ELISA against serum ELISA as gold standard was 32 of 50 (64%) and 530 of 619 (87%), respectively, at a cut off point of 0.41 OD (optical density) units for saliva and 0.1 OD units for serum (fig 2), as suggested by the manufacturer. The cut off point of 0.41 maximised specificity without compromising sensitivity too far (fig 3).
ESTABLISHING CRITERIA FOR POSITIVE AND NEGATIVE SALIVARY WESTERN BLOT
The criteria for positivity on serum/saliva western blotting were either the recognition of the CagA (120 kDa) band alone, which has previously been shown to be specific for the presence of H pylori infection,8 or a minimum of six bands from any other region of the blot, because several bands between 30 and 60 kDa were non-specific, as has been reported previously.8–10 Using these criteria, 17 of 18 (94%) positive sera were western blot seropositive and 16 of 16 (100%) negative sera were western blot seronegative (fig 4)
The same criteria were applied to the saliva western blot samples and 11 of 13 (84%) of the saliva specimens from seropositive subjects were western blot positive, and 18 of 18 (100%) of the saliva specimens from seronegative subjects were western blot negative (fig 5).
PERFORMANCE OF SALIVARY WESTERN BLOT COMPARED WITH SERUM WESTERN BLOT
A total of 16 samples had to be centrifuged before western blotting and of these seven still yielded uninterpretable blots and were excluded from the analysis. Using the serum western blot as the gold standard, the sensitivity and specificity of the salivary western blot was 38 of 47 (81%) and 68 of 75 (91%), respectively (table 1). The sensitivity and specificity of saliva ELISA on these same specimens was 29 of 47 (62%) and 48 of 75 (64%), respectively. Saliva western blotting was significantly more sensitive (p = 0.039) and significantly more specific (p = 0.00005) than saliva ELISA.
The presence of bands to the cagA and vacA proteins on the salivary and serum western blots were compared, with good overall agreement between the two methods (κ = 0.75, p < 0.000001 for CagA; κ = 0.66, p < 0.000001 for VacA; table 2).
PERFORMANCE OF SALIVARY WESTERN BLOTTING AGAINST SERUM ELISA
The sensitivity and specificity of saliva western blot compared with serum ELISA was 32 of 44 (73%) and 65 of 78 (83%), respectively (table 3). The sensitivity of salivary western blotting was not significantly greater than salivary ELISA (p = 0.39), whereas specificity was significantly greater (p = 0.0071).
COMPARISON OF SERUM WESTERN BLOTTING AGAINST SERUM ELISA
There was good overall agreement between serum western blot and serum ELISA (κ = 0.84, p < 0.000001; table 4). Of the six ELISA negative, but western blot positive, specimens three had ELISA optical densities between 0.050 and 0.099 OD units—that is, at the higher end of the range of negative sera.
Our study demonstrates that western blotting has a significantly better specificity than ELISA for the detection of H pylori antibodies in the saliva of children and is comparable with serum ELISA. Therefore, salivary western blotting might be useful in the study of children, where high specificity is of particular importance, owing to the low prevalence of infection. In comparison with serum western blotting, salivary western blotting was more sensitive than salivary ELISA. It was also superior when both were compared with serum ELISA, although this failed to reach significance.
Our study might have underestimated the specificity of salivary western blotting significantly, because a stratified sample was used, in which saliva positive serum negative specimens on ELISA were over represented. Although western blotting of saliva was very specific, it had a slightly lower sensitivity than serum ELISA, using serum western blot as the gold standard. However, specificity is more important than sensitivity in studying low prevalence populations, such as children in the developed world.
The qualitative nature of the western blot does appear to overcome the problems of variable transudation of serum immunoglobulin into saliva, which might affect ELISA optical density. Western blotting has the additional advantage of providing more information from the saliva samples, such as seropositivity to CagA or VacA, which can be used to type the organisms with which the patient is infected. There was good agreement between bands detected on serum and saliva western blotting, supporting the use of saliva for this purpose.
Saliva has some disadvantages compared with serum. In a number of specimens there was excessive debris, which required centrifugation. Unfortunately, there were still seven samples that yielded uninterpretable blots and these were not included in the results. Another potential disadvantage is the sample volume required, although only 1 ml was used, which is not excessive.
Other problems that might impair the sensitivity of salivary western blotting are the freeze thawing of samples and the circumstances of sample collection. In our study, the saliva was three years old and had been subjected to a previous freeze–thaw cycle. The means of collection might also be important: theoretically, unstimulated saliva is better than stimulated saliva.
Saliva has several advantages over other methods of determining H pylori status. It is non-invasive, unlike serum testing, and it is much cheaper and more convenient than C13 urea breath testing. New tests to detect H pylori antigen in faeces hold promise,11 but faecal specimens are more difficult to obtain, particularly in children, and harder to work with in the laboratory. Preliminary reports of evaluation in children have given promising results.12
Although western blotting is a reliable method and overcomes the inconvenience of a capture assay, such as ELISA, it is time consuming, but could and is being automated. However, because of the ease of collection and non-invasiveness, saliva remains attractive for the diagnosis of H pylori infection, and salivary western blotting is therefore a potentially useful method for epidemiology and paediatric diagnosis.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.