MUSCLE DIFFERENTIATION AND CLINICOPATHOLOGICAL FEATURES OF GASTROINTESTINAL STROMAL TUMORS

We analyzed 46 gastrointestinal stromal tumors (GISTs) using a panel of antibodies to determine the frequency of smooth muscle differentiation and the relationship of immunophenotype to histopathologic features and clinical behavior. Thirty-six GISTs were classified as benign or malignant based exclusively on clinical behavior; a 2-year minimum follow-up was required for benign lesions. GISTs were immunopositive in the following categories: vimentin 45 of 46, desmin nine of 45, muscle-specific actin (MSA) 36 of 46, alpha-smooth muscle actin (SMA) 34 of 46, chicken gizzard actin-7 zero of 38, cytokeratin two of 46, S100 protein six of 46, glial fibrillary acidic protein (GFAP) zero of 46, synaptophysin zero of 46, and chromogranin one of 46. At least one muscle marker was positive in 39 of 46 tumors. Five GISTs were MSA positive/SMA negative, and three were MSA negative/SMA positive. All desmin-positive cases reacted with MSA or SMA. Eight GISTs were positive for vimentin, MSA, SMA, and desmin, whereas seven were vimentin positive only. Compared with the latter, the former tended to be smaller, less often necrotic, and clinically benign (p < 0.05 for each). All vimentin-positive only GISTs were malignant. Immunohistochemical features did not correlate with tumor site, cellularity, nuclear pleomorphism, or mitotic rate. Benign GISTs were less cellular than were malignant GISTs (p < 0.05), but they did not differ statistically in degree of nuclear pleomorphism, necrosis, mitotic rate, or size. We conclude that (a) 85% of GISTs react with at least one muscle antibody; (b) immunohistochemical features are unrelated to anatomic site; (c) SMA is, in effect, as sensitive as MSA, whereas desmin is less sensitive; and (d) simultaneous vimentin, MSA, SMA, and desmin positivity correlates with a benign outcome.