Giant Cell Tumors With HMGA2::NCOR2 Fusion : Clinicopathologic, Molecular, and Epigenetic Study of a Distinct Entity

Abstract

Giant cell tumors (GCTs) with high mobility group AT-Hook 2 ( HMGA2 )::nuclear receptor corepressor 2 ( NCOR2 ) fusion are rare mesenchymal tumors of controversial nosology, which have been anecdotally reported to respond to CSFR1 inhibitors. Here, we performed a comprehensive study of 6 GCTs with HMGA2::NCOR2 fusion and explored their relationship with other giant cell-rich neoplasms. Tumors occurred in 4 females and 2 males ranging in age from 17 to 32 years old (median 24). Three lesions originated in subcutaneous soft tissue and 3 in bone. Tumor size ranged from 20 to 33 mm (median 27 mm). The lesions had a nodular/multinodular architecture and were composed of sheets of mononuclear "histiocytoid" cells with uniform nuclei intermingled with multinucleated giant cells. Mitotic activity was low and nuclear atypia and metaplastic bone were absent. Variable findings included necrosis, cystic degeneration, lymphocytic infiltrate (sometimes forming nodules), and xanthogranulomatous inflammation. On immunohistochemistry, all cases focally expressed pan-keratin and were negative with SATB2 and H3.3G34W. Whole RNA-sequencing was performed in all cases of GCT with HMGA2::NCOR2 fusion and a subset of giant cell-rich tumors (tenosynovial-GCT, n = 19 and "wild-type" GCT of soft tissue, n = 9). Hierarchical clustering of RNA-sequencing data showed that GCT with HMGA2::NCOR2 fusion formed a single cluster, independent of the other 2 entities. Methylome profiling showed similar results, but the distinction from "wild-type" GCT of soft tissue was less flagrant. Gene expression analysis showed similar levels of expression of the CSF1/CSFR1 axis between GCT with HMGA2::NCOR2 fusion and tenosynovial-GCT, supporting their potential sensitivity to CSFR1 inhibitors. Clinical follow-up was available for 5 patients (range: 10 to 64 mo; median 32 mo). Three patients (60%) experienced local recurrences, whereas none had distant metastases or died of disease. Overall, our study confirms and expands previous knowledge on GCT with HMGA2::NCOR2 fusion and supports its inclusion as an independent entity.