[Gastrointestinal stromal tumors: clinical, pathologic features and effectiveness of new diagnostic criteria].

Background/aims: The effective and reproducible diagnostic parameters for differentiating benign from malignant gastrointestinal stromal tumors (GISTs) are still not clear. In this study, GISTs were diagnosed and classified by immunohistochemistry and their clinical and pathologic features were investigated. GISTs were re-evaluated by Amin's and NIH's criteria, and prognostic relevance of these two criteria were compared.

Methods: Fifty cases of gastrointestinal mesenchymal tumor diagnosed from May 1990 to February 2000, were evaluated by immunohistochemical staining for CD117, CD34, smooth muscle actin, and S-100 protein. GISTs were diagnosed according to Amin's and NIH's criteria. The relationship between the prognosis and diagnosis based on Amin's or NIH's classification were analyzed.

Results: Thirty cases of gastrointestinal mesenchymal tumors were diagnosed as GISTs. The stomach (40%) and small bowel (40%) were the most common origin for GISTs. Immunophenotypically, null, myoid, neural, combined type were 70.0%, 10.0%, 16.7% and 3.3%, respectively. Seven cases showed metastasis and one case showed recurrence. According to Amin's criteria, 5 benign, 8 borderline and 17 malignant tumors were diagnosed. The NIH's criteria showed 2 very low risk, 6 low risk, 7 intermediate risk, and 15 high risk tumors. Metastasis or recurrence of GISTs had no significant relationship with malignancy according to Amin's criteria (p=0.4069) but had significant correlation with high risk tumor based on NIH's criteria. (p=0.0352).

Conclusions: GISTs showing local invasion, distant metastasis or recurrence were related with high risk tumors based on NIH's criteria. NIH's criteria might be better reliable scheme than Amin's for predicting the prognosis of GISTs.