CONTEXT: Most gastric gastrointestinal stromal tumors (GISTs) express CD117/c-kit, as do a subset of metastatic melanomas, leading to a diagnostic dilemma in some cases. OBJECTIVE: To further differentiate GISTs from melanoma, we investigated expression of melanoma markers in GISTs using a well-characterized set of gastric lesions on tissue microarrays. DESIGN: Tissue microarrays from paraffin-embedded tissue cores from 38 patients were stained with S100 protein, HMB-45, and Melan-A antibodies. All cases had been previously stained with CD117/c-kit and CD34 antibodies. All were reactive with CD117/c-kit, and 88.2% expressed CD34. RESULTS: S100 protein was focally expressed in 2 (5.3%) of 38 GISTs; these lesions lacked HMB-45 and Melan-A labeling. No tumor labeled with HMB-45, but 4 (10.6%) of 38 cases labeled with Melan-A antibodies. The Melan-A-reactive cases were all S100 negative and CD34 positive. The S100-reactive cases were spindle cell neoplasms, whereas the Melan-A-reactive cases were epithelioid neoplasms (4/9; 44%). An additional 15 standard sections of separate cases of epithelioid GISTs were then labeled with Melan-A, and 5 (33%) of 15 showed at least focal labeling. CONCLUSIONS: Melan-A staining can be encountered in a subset of epithelioid GISTs, a finding that can suggest a differential diagnosis of melanoma. In this series, the Melan-A-reactive cases lacked S100 protein and expressed CD34, both of which would be unlikely in melanoma. As such, a panel approach is best in differentiating epithelioid GISTs from melanoma.
CONTEXT: Most gastric gastrointestinal stromal tumors (GISTs) express CD117/c-kit, as do a subset of metastatic melanomas, leading to a diagnostic dilemma in some cases. OBJECTIVE: To further differentiate GISTs from melanoma, we investigated expression of melanoma markers in GISTs using a well-characterized set of gastric lesions on tissue microarrays. DESIGN: Tissue microarrays from paraffin-embedded tissue cores from 38 patients were stained with S100 protein, HMB-45, and Melan-A antibodies. All cases had been previously stained with CD117/c-kit and CD34 antibodies. All were reactive with CD117/c-kit, and 88.2% expressed CD34. RESULTS: S100 protein was focally expressed in 2 (5.3%) of 38 GISTs; these lesions lacked HMB-45 and Melan-A labeling. No tumor labeled with HMB-45, but 4 (10.6%) of 38 cases labeled with Melan-A antibodies. The Melan-A-reactive cases were all S100 negative and CD34 positive. The S100-reactive cases were spindle cell neoplasms, whereas the Melan-A-reactive cases were epithelioid neoplasms (4/9; 44%). An additional 15 standard sections of separate cases of epithelioid GISTs were then labeled with Melan-A, and 5 (33%) of 15 showed at least focal labeling. CONCLUSIONS: Melan-A staining can be encountered in a subset of epithelioid GISTs, a finding that can suggest a differential diagnosis of melanoma. In this series, the Melan-A-reactive cases lacked S100 protein and expressed CD34, both of which would be unlikely in melanoma. As such, a panel approach is best in differentiating epithelioid GISTs from melanoma.