A Performance Comparison of Commonly Used Assays to Detect RET Fusions

Abstract

Abstract Purpose: Selpercatinib and pralsetinib induce deep and durable responses in patients with advanced RET fusion–positive lung and thyroid cancer. RET fusion testing strategies with rapid and reliable results are critical given recent FDA approval. Here, we assess various clinical assays in a large pan-cancer cohort. Experimental Design: Tumors underwent DNA-based next-generation sequencing (NGS) with reflex to RNA-based NGS if no mitogenic driver or if a RET structural variant of unknown significance (SVUS) were present. Canonical DNA-level RET fusions and RNA-confirmed RET fusions were considered true fusions. Break-apart FISH and IHC performance were assessed in subgroups. Results: A total of 171 of 41,869 patients with DNA NGS harbored RET structural variants, including 139 canonical fusions and 32 SVUS. Twelve of 32 (37.5%) SVUS were transcribed into RNA-level fusions, resulting in 151 oncogenic RET fusions. The most common RET fusion–positive tumor types were lung (65.6%) and thyroid (23.2%). The most common partners were KIF5B (45%), CCDC6 (29.1%), and NCOA4 (13.3%). DNA NGS showed 100% (46/46) sensitivity and 99.6% (4,459/4,479) specificity. FISH showed 91.7% (44/48) sensitivity, with lower sensitivity for NCOA4-RET (66.7%, 8/12). A total of 87.5% (7/8) of RET SVUS negative for RNA-level fusions demonstrated rearrangement by FISH. The sensitivity of IHC varied by fusion partner: KIF5B sensitivity was highest (100%, 31/31), followed by CCDC6 (88.9%, 16/18) and NCOA4 (50%, 6/12). Specificity of RET IHC was 82% (73/89). Conclusions: Although DNA sequencing has high sensitivity and specificity, RNA sequencing of RET SVUS is necessary. Both FISH and IHC demonstrated lower sensitivity for NCOA4-RET fusions.