Proposal of a new diagnostic algorithm for Helicobacter pylori in histopathology: Hematoxylin-eosin and alkaline blue...

To determine the diagnostic capacity and interobserver agreement of a new diagnostic algorithm for Helicobacter pylori in histopathology compared with the current gold standard.

A cross-sectional study analyzed 92 paraffin blocks from gastric biopsies obtained from the Taxa Oncology Laboratory in Lima, Peru. Each block underwent three new sections, and Warthin-Starry staining, immunohistochemistry, and alkaline blue stain were applied. The results obtained by two experts and those from the conventional hematoxylin and eosin method were then grouped into two diagnostic algorithms and subsequently compared.

As main results, the interobserver agreements of the new algorithm and the conventional method were K=0.90 (strong agreement) and K=0.85 (strong agreement), respectively. With adjusted Kappa, it was better for the new algorithm (K=0.91, almost perfect agreement) in comparison with the conventional one (K=0.87, strong agreement). Also, all diagnostic capacity statistics for the new algorithm compared with the conventional one (reference) were very high (sensitivity=93.5% and predictive positive value=90.3%), highlighting the specificity and predictive negative value (95.2% and 96.7%, respectively).

This proposed new diagnostic algorithm presented optimal statistical reliability, in addition to being faster, less toxic, and easier to implement than the conventional one, which suggests it has great potential for application in pathology laboratories, especially for those who cannot afford reference methods, such as immunohistochemistry or Warthin-Starry.