Background: The aim of this study was to ascertain whether routine endoscopic biopsy specimens are of sufficient quality and representative enough to enable reliable classification of histologic type, grade of differentiation, and frequency of signet ring cells.
Methods: As part of a population-based etiologic study of gastric cancer in Sweden, the histologic slides of 687 incident cases were independently reviewed by one pathologist. In 305 cases both biopsy and surgical specimens were available for assessment of sensitivity and specificity of diagnoses on the basis of biopsy specimens, with the classification based on surgical specimens as the gold standard.
Results: The sensitivity and specificity of the diagnoses on the basis of biopsy specimens was 86% and 82%, respectively, for detecting the intestinal type, and 87% and 90%, respectively, for the diffuse type of gastric cancer. The percentage agreement between biopsy and surgical specimens with regard to grade of differentiation was 71%.
Conclusions: This study shows that misclassification of Laurén type due to random intraobserver variation needs to be taken into account in epidemiologic studies. The degree of misclassification appears to be of similar magnitude regardless of the source material.