Evaluation of histology as a Helicobacter pylori detection method and analysis of associated problems

Helicobacter pylori is regarded as a common cause of gastritis and peptic ulcer disease. The most commonly used H. pylori detection method in Sri Lanka is histology. However, the detection rate of H. pylori in routine histology practice is low. Therefore, we conducted the following study to evaluate the diagnostic efficacy of histology and to analyze the possible problems associated with H. pylori detection. Multiple endoscopic gastric biopsies were obtained from a sample of 205 patients detected to have endoscopic gastric erythema or ulcers. Biopsies were formalin fixed and paraffin embedded and stained with haematoxylin and eosin, toluidine blue and immunohistochemistry. Serum was collected for screening of anti H. pylori antibodies using an immunochromatography based kit method. Diagnostic efficacy of histology was evaluated against immunohistochemistry as the gold standard. Inter observer variation among four pathologists was assessed using the intraclass correlation coefficient. Haematoxylin and eosin showed a sensitivity of 100% and specificity of 99% and toluidine blue had 100% sensitivity and 98.5% specificity. Average measures of intra class correlation coefficient for H&E was 0.428 (95% CI 0.228 – 0.588) and for toluidine blue stain 0.320 (95% CI 0.085 – 0.513). The sero prevalence of anti – H. pylori antibodies was 4.9%. In conclusion, sensitivity, specificity and negative predictive values of histology in detecting H. pylori are shown to be high. Main limitations were, low positive predictive value and unsatisfactory interobserver agreement. Sampling errors and exposure to antibiotics appeared to be an unlikely cause of the low detection rate with histology.


Introduction
Helicobacter pylori is a spiral shaped bacterium which resides beneath the mucous layer of the gastric mucosa often adherent to the surface epithelium. (1,2) H. pylori has been reported to have a strong aetiological relationship with chronic gastritis, peptic ulcer disease, gastric carcinoma and lymphoma (1 -3) A high prevalence of H. pylori infection is observed among the developing countries and East Asian countries. (4) The H. pylori prevalence in the South Asian region, such as India and Bangladesh, has also been reported to be relatively high. (4.5) No single test has proven to be ideal for H. pylori detection. Commonly available test methods, which require endoscopic biopsies, include histology, rapid urease tests such as CLO test, culture and polymerase chain reaction (PCR). Serology, urea breath test and stool antigen test are minimally invasive test methods. Of these, urea breath test has been recognised as the best method. (6,7) Histology, haematoxylin and eosin stain combined with a special stain for H. pylori (Giemsa or toluidine blue), is the most commonly used detection method for H. pylori in Sri Lanka. Often, this is the only method available in the government hospitals. However, apart from one study, the general experience of pathologists in Sri Lanka is that H. pylori detection rate by histology is low. (8,9) The possible factors that can affect the H. pylori detection with histology include expertise of the pathologist and the density and distribution of H. pylori in the gastric mucosa.
Exposure to antibiotics, even for other reasons can reduce the bacterial density giving rise to false negative results. Furthermore, long term use of proton pump inhibitors are known to promote migration of the organisms to the body region giving rise to false negative results with biopsies taken only from the antrum. (10) Coccoid form of H. pylori has been described as an adaptive morphological transformation of the organism in a less favourable environment. (11,12) In histology, the pathologist depends on the spiral shape morphology and their distribution pattern for identification of H. pylori. Therefore, a high prevalence of coccoid forms could lead to a false negative result on histology.
Accordingly, we conducted the following study to evaluate the diagnostic efficacy of histology in detecting H. pylori infection, taking immunohistochemistry as the gold standard.
The other contributory factors that can give rise to false negative results such as sampling errors, prior exposure to antibiotics, prevalence of coccoid forms and inter -observer variability among pathologists were also evaluated. pylori antibodies. According to the manufacturer, the method has a high sensitivity (95.5%) and relatively low specificity (89.6%). Therefore, the test was used to screen for the presence of anti H. pylori antibodies to assess the exposure rate to the H. pylori in the study population.
The diagnostic efficacy of histology was evaluated in terms of sensitivity, specificity and positive and negative predictive values using immunohistochemistry as the gold standard.
Histology results of an investigator who had undergone a special training in gastrointestinal pathology were used for this purpose.

H&E and toluidine blue stained sections
were assessed for the presence or absence of H. pylori organisms by three independent pathologists and a trainee pathologist to assess the degree of inter-observer variation. The interobserver variation of the histological results were analysed using the intra-class correlation coefficient. The confidence interval was set at 95% for all statistical methods used.

Results
There were 5 cases positive for H. pylori with H&E stain, 6 with toluidine blue stain and 3 with immunohistochemistry. Six cases were positive on histology (H&E and toluidine blue combined). All the cases detected with immunohistochemistry were spiral shaped organisms and there were no coccoid forms.

Evaluation of haematoxylin and eosin stain
Results of H&E stain against immunohistochemistry are given in Table 1 10(2): [5][6][7][8][9][10][11] The main shortcoming of histology is unsatisfactory interobserver agreement, which was more prominent in the interpretation of the toluidine blue stain ( Table 3) Inclusion of additional biopsies from the stomach did not increase the H. pylori detection rate indicating that sampling errors are unlikely to be responsible for the low detection rate.
The low sero-prevalence (4.9%) of anti H. pylori antibodies indicated that the exposure rate of this study population to the bacterium was low. The test method we used is a sensitive method which collectively detects all types of anti H. pylori antibodies including IgG which indicate current as well as past infection. Therefore, the low detection rate by histology cannot be attributed to a low H. pylori density due to prior exposure to antibiotics.

Conclusions
The sensitivity, specificity and negative predictive value of haematoxylin and eosin and toluidine blue stain in detecting H. pylori were shown to be high. Main limitations were, the low positive predictive value and unsatisfactory interobserver agreement. Sampling errors and exposure to antibiotics appeared to be an unlikely cause for the low detection rate by histology.