Background: It is important to distinguish gastrointestinal stromal tumors (GISTs) from other gastrointestinal mesenchymal tumors (GIMTs), because of the malignant potential of GISTs and the availability of molecular targeted therapy. GISTs represent the most common subgroup of GIMTs, and rarely occur in the colon and rectum. The first objective of our retrospective study was to reclassify colorectal mesenchymal tumors, from files collected over 20 years, to determine if, based on immunohistologic features, the lesions were truly GISTs. The second objective was to identify the relationship between clinicopathologic features and prognostic factors of GISTs in the colon and rectum.
Methods: We evaluated all cases of colorectal mesenchymal tumor identified from the database of the Department of Surgical Pathology at Taichung Veterans General Hospital for the period 1983-2001. For 25 patients, clinical data, and information about tumor characteristics, surgical procedures, and survival outcomes, were obtained and analyzed. Histopathologic evaluations, and appropriate immunohistochemical markers, were used to distinguish between various GIMT subtypes. The relationship between KIT expression and clinicopathologic features was investigated.
Results: The following variables were significantly associated with different CD117 results: symptomatic presentation, location, gross features, tumor size, mitotic count, cellularity, and type of surgery. Only 18 tumors were identified as GISTs. For these, the following variables were significantly associated (by univariate analysis) with increased lethality: tumor size (p = 0.049); mitotic count (p = 0.019); nuclear atypia (p = 0.019); and tumor necrosis (p = 0.045). However, only mitotic activity showed a significant difference in the survival analysis (p = 0.0304; log-rank test).
Conclusion: Two clinicopathologically different categories were identified from our colorectal mesenchymal tumors: intramural GISTs and polypoid submucosal leiomyomas. Our study suggests that GIST is a better categorization than smooth muscle tumor because of the malignant potential. Prognosis is strictly related to the number of mitoses. However, tumor size, nuclear atypia and tumor necrosis are probably also significant predictive factors of lethality. Future studies with DNA analysis and larger patient numbers are essential to evaluate the prognostic significance of our findings.