Measurement variability and diagnostic sensitivity of gastric mucosal inflammatory cell morphometry

To assess the variability and sensitivity of morphometric measures of gastric mucosal lymphocyte and plasma cells to determine a systematic procedure for evaluating the density of these mononuclear inflammatory cells (MNC).

Gastric biopsies of antrum (n = 3), incisura angularis (n = 2) and corpus (n = 3) from two controls and three patients with Helicobacter pylori-related gastritis (antral, diffuse or multifocal gastritis) were considered. In each biopsy, three fields from each of three sections were selected. In each field, stromal area was obtained by subtracting gland area (GA) from total area, and MNC were counted. Results were expressed as MNC/total mm2 and MNC/stromal mm2. Correlations with GA, coefficients of variation (CV), discriminant power analysis and analysis of variance were performed.

Correlations always existed between GA and MNC/total mm2 and rarely between GA and MNC/stromal mm2. CV of MNC/stromal mm2 were lower (18%) than those of MNC/total mm2 (30%). High sensitivity (95.4%) and specificity (95.8%) were found for MNC/stromal mm2 but not for MNC/total mm2. Differences in MNC/stromal mm2 existed in all subjects (P < .0001). Highly significant differences in MNC/stromal mm2 were also found between normal and inflammatory states, gastric sites and sections.

In contrast to MNC/total mm2, MNC/stromal mm2 is an unbiased estimate of MNC density. The following sampling procedure is proposed: two biopsies from each gastric site, two sections from each biopsy and two microscopic fields from each section.