Review of histological classifications of gastrointestinal epithelial neoplasia: differences in diagnosis of early carcinomas between Japanese and Western pathologists

Gastrointestinal lesions considered to be high grade adenoma/dysplasia by W estern pathologists using the conventional Western classification are often diagnosed as carcinoma by Japanese pathologists using the Japanese group c lassification. To overcome these differences, the Padova classification, th e Vienna classification, and a revision of the Vienna classification have r ecently been proposed. The clinical usefulness of these five classification s needs to be reviewed for early gastric, esophageal, and colorectal neopla sias. In 1998, 31 pathologists from 12 countries individually diagnosed the same 35 gastric, 21 esophageal, and 20 colorectal specimens. Their histolo gical diagnoses can be classified conventionally and according to the newly proposed terminology, and from these data, the extent of agreement between pathologists with Western and Japanese viewpoints can be calculated, using kappa statistics. With the conventional Western, Japanese, Padova, Vienna, and revised classifications, the agreement scores were 37%, 37%, 71%, 71%, and 80%, respectively, for gastric lesions; 14%, 14%, 57%, 62%, and 67% fo r esophageal lesions; and 45%, 50%, 65%, 65%, and 70% for colorectal lesion s. The kappa values were lower than 0.3 with the conventional Western and J apanese classifications, but higher than 0.5 for gastric lesions, higher th an 0.3 for esophageal lesions, and higher than 0.4 for colorectal lesions w ith the newly proposed classifications. When the literature regarding treat ment indications for early neoplastic lesions is reviewed, it becomes appar ent that the categories of the revised classification would fit best with c urrent clinical treatment considerations. This classification would be part icularly useful for endoscopically resected specimens, to determine whether additional surgery with lymph node dissection is required. In conclusion, the use of the newly proposed terminology can, in large part, resolve the i ntercountry differences in the diagnosis of adenoma/dysplasia and early car cinoma. However, the newly proposed classifications should be used with cau tion for biopsy specimens, as sampling error may result in an underestimati on of the neoplastic grade or depth of invasion. For the choice between end oscopic and surgical treatment, assessment of the depth of invasion by endo scopic inspection and ultrasound or radiography is essential.