Prospective comparison of H&E, Giemsa, and Genta stains for the diagnosis of Helicobacter pylori

Background: H. pylori is more easily visualized with special stains than with H&E, but this adds time and expense to the diagnostic workup. We sought to determine if the diagnostic accuracy was improved with special stains. Methods: One hundred-one patients had two ''jumbo'' biopsies taken from the gastric antrum and two from the body for examination with H&E, Genta, and Giemsa stains. Four separate biopsy specimens were also taken from the antrum and the body for culture and for three types of rapid urease test, and C-13- urea breath tests were also performed. Mixed, coded biopsies were assessed for H. pylori, and density was scored from 0 to 4. A case was considered positive for H. pylori if culture was positive, two rapid urease tests and a urea breath test were positive, or two different stains were positive. Biopsy specimens were excluded from analysis if the slides were missing or there was inadequate tissue for review, or if the specimen showed a lack of staining. Results: Fifty-two (13%) of 404 specimens were excluded because of a poor Genta stain. Sensitivities were comparable for the three stains (H&E, 92%; Giemsa, 88%; Genta, 91%), while H&E specificity (89%) was significantly lower than that of the special stains (98%). Sensitivity for all three stains was significantly lower at low (grade 0 to 1) H. pylori density than at high (grade 2 to 4) density (H&E, 70% vs 98%; Giemsa, 64% vs 96%; Genta, 66% vs 97%), and 20 of 22 false positives were grade 1. Conclusions: The sensitivities of H&E and special stains are comparable at around 90%, but the specificity of H&E is significantly lower. The Giemsa stain appears to be the preferred stain for H. pylori diagnosis on the basis of its goad sensitivity, excellent specificity, and lack of technical difficulty in preparation. However, H&E provides excellent accuracy when more than minimal (grade 1) H. pylori density is present.