Gastroesophageal Glomus Tumors: Clinicopathologic and Molecular Genetic Analysis of 26 Cases With a Proposal for Malignancy Criteria. (15th October 2022)

• Record Type: Journal Article
• Authors: Papke, David J.; Sholl, Lynette M.; Doyle, Leona A.; Fletcher, Christopher D.M.; Hornick, Jason L.
• Is Part Of: American journal of surgical pathology. Volume 46:Number 10(2022)
• Page Start: 1436
• Page End: 1446
• Publication Date: 2022-10-15
• Subjects: glomus tumor -- gastric -- gastroesophageal -- mesenchymal tumor -- malignant -- malignancy criteria -- NOTCH2 -- CARMN::NOTCH2; Pathology, Surgical -- Periodicals; 617.0705
• DOI: 10.1097/PAS.0000000000001925
• Languages: English
• ISSNs: 0147-5185

Abstract

Abstract : Although criteria for malignancy have been established for glomus tumors of soft tissue, there are no accepted criteria for gastroesophageal glomus tumors, the behavior of which is considered to be unpredictable. Recently, both benign and aggressive gastroesophageal glomus tumors have been shown to harbor CARMN :: NOTCH2 fusions, but, as yet, there are no described genetic features that predict clinical behavior. Here, we evaluated 26 gastroesophageal glomus tumors to investigate histologic and genetic features that might predict malignant behavior. Seventeen patients (65%) were male. The median age at presentation was 54.5 years (range: 16 to 81 y). Primary sites were stomach (25 tumors) and distal esophagus (1). The median tumor size was 4.05 cm (range: 0.8 to 19.5 cm). Tumors were composed of lobules of rounded cells with sharp borders, palely eosinophilic to clear cytoplasm, and round nuclei. All tumors involved the muscularis propria, and 12 also involved the serosal surface. Mitoses ranged from <1 to 53/10 HPF (median: 5/10 HPF). Sixteen tumors, including all 15 with mitoses ≥2/10 HPF, showed atypia (3 mild, 10 moderate, 3 severe), defined as spindle cell morphology, nuclear irregularity, nuclear size variability, enlarged nuclei, or coarse chromatin. Considering these histologic features and clinical behavior, tumors were classified as malignant (15 tumors) if they measured ≥5 cm or showed both atypia and mitoses ≥2/10 HPF, or benign (11 tumors) if these criteria were not met. Follow-up was available for 19 patients (73%; range: 1 to 15 y; median: 5.8 y), including 7 with benign tumors and 12 with malignant tumors. Two patients with malignant tumors had metastases at presentation, and 7 developed metastases subsequently. Follow-up was available for 8 of 9 patients with metastatic disease. Two were alive with disease at most recent follow-up. One underwent resection of a liver metastasis, with no subsequent metastases in 3 years of follow-up. Five patients died of metastatic disease. By immunohistochemistry, smooth muscle actin was diffusely positive in all evaluated tumors, and caldesmon and synaptophysin were positive in 94% and 73%, respectively. Sequencing identified NOTCH2 alterations in 4 benign tumors (80%) and 8 malignant tumors (80%), including CARMN :: NOTCH2 fusions in 2 benign and 5 malignant tumors. All 5 sequenced benign tumors lacked complex copy number alterations (CNAs), whereas all 10 sequenced malignant tumors showed complex CNAs, including recurrent loss of 9p21.3 (4/10, variably including CDKN2A / B and MTAP ) and ATRX inactivation (4/10). Complex CNAs were identified in all sequenced tumors that were ≥5 cm, exhibited both cytologic atypia and ≥2 mitoses/10 HPF, involved the serosa or metastasized. We propose that gastroesophageal glomus tumors ≥5 cm or with both atypia and mitoses ≥2/10 HPF should be considered malignant. Copy number analysis might be helpful in borderline cases.