Gastroscopic biopsy: reliability of histological diagnosis with special reference to the single biopsy.

For three years (1970-72) endoscopic gastric biopsies from 883 lesions in 862 patients were histologically examined, every single biopsy (4088) separately. In 38% of the cases, namely 142 malignant and 194 benign lesions, the diagnoses could be verified. If the criterion for a positive examination was that one of the biopsies showed unambigous carcinoma, the sensitivity of the examination was 76%, the specificity 99%, the PVpos 98%, and the PVneg 85% for the first examination, using the terms of Vecchio. If the biopsy diagnoses carcinoma with some uncertaincy and precancerous biopsies were considered positive too, and if all examinations prior to verification were included, the corresponding figures were 88%, 98%, 97%, and 91%. The most common number of biopsies used in an examination for diagnosing carcinoma was 4 (range 1-11). Fewer biopsies were used if the carcinomas were localized to the whole stomach, were protruded or had a largest diameter of more than 7 cm. The characteristic 'protruded' was the only tumour criterion which significantly was found most common among positive examinations. Ulcer-like carcinomas were common among the false negative. On the basis of pathoanatomy no special number of biopsies to be taken per examination could be recommended.