Histological score for cells with irregular nuclear contours for the diagnosis of reflux esophagitis in children.

Histological criteria for the diagnosis of reflux esophagitis include basal zone hyperplasia, stromal papillae elongation, and inflammatory infiltrate. However, endoscopic esophageal biopsy specimens may include little or no lamina propria. Intraepithelial T lymphocytes, seen in hematoxylin and eosin-stained sections as cells with irregular nuclear contours (CINC), may have a higher density in children with esophagitis. We evaluated the diagnostic accuracy of a numerical score built up by grading the "classical" parameters and its correlation with CINC density in grasp biopsy specimens obtained from children undergoing esophagogastroduodenoscopy with and without esophagitis. We analyzed esophageal biopsy specimens from 349 children (median age, 5 years) subdivided in 4 groups according to the previous routine histology report: group 1, 144 children with esophagitis; group 2, 65 controls; group 3, 51 children with dubious esophagitis; and group 4, 75 children with esophagitis on endoscopy but a normal histology report. A numerical value was assigned to each parameter; the sum of these values represented the histological score. We also evaluated intraepithelial CINC density (ie, number of CINC per high-power field). We separately analyzed histological sections with and without lamina propria. For both total score and for CINC density, we calculated a cutoff using a receiver operating characteristic curve. Cutoffs of 6 for score and of 4 for CINC density provided the best sensitivity and specificity. Sensitivity of the histological score was better in biopsy specimens containing lamina propria (94%) than in those without lamina propria (4%). Sensitivity of CINC density was satisfactory in both specimens with (78%) and without (75%) lamina propria. Specificity was satisfactory for both parameters. In conclusion, when lamina propria was present in sections of endoscopic esophageal biopsy specimens, histological score provided a better diagnostic accuracy for the diagnosis of esophagitis. However, when no lamina propria was present, as was the case in 67% of our children, CINC density had better sensitivity. In addition, this latter parameter showed esophageal mucosa damage in 34% of previously dubious cases or cases with esophagitis at endoscopy but a previous routine histology report of normal mucosa.