Endocrine Abstracts

Point mutations constitute the majority of thyroid cancer mutations. Depending on screening strategy and patient populations, however, rearrangements also serve as oncogenic drivers for roughly one-in-five tumors. This underscores the importance of a comprehensive molecular approach that extends beyond assessment of single nucleotide variants (SNV). Receptor tyrosine kinases (RTK) feature prevalently within the described gene fusions in thyroid entities. Such fusions impart a constitutively active phenotype that deregulates multiple signaling cascades. RTKs are highly indiscriminate, both in their fusion partners and the underlying genomic mechanism of rearrangement, which can fall below the detection limit of conventional diagnostic approaches. As a consequence, targeted multiplex amplicon sequencing has emerged as a preferred method for characterizing thyroid neoplasia. However, such approaches require a priori knowledge of fusion partners and/or breakpoints. Therefore, it was hypothesized that gene partner and/or breakpoint agnostic approaches might impart improved clinical sensitivity. To assess this hypothesis in a primary sample across four marketed thyroid sequencing panels (ThyroSPEC, Oncomine Focus Assay (OFA), Illumina TruSight Fusion Panel (IFP), and Archer FusionPlex CLT), we performed an in silico analysis on n = 20 unpublished cases of fusion-driven thyroid carcinomas identified using retrospective FusionPlex sequencing, between 2022-2023. To supplement this data, we identified 16 publications reporting at least five thyroid cancer fusion-driven cases, however, 13/16 (81%) were excluded due to lack of transcript or genomic reference IDs and 1/16 (6%) was excluded due to utilization of a targeted multiplex amplicon approach, for a final total of n = 168. Sensitivity analysis was performed through stratification via the five most common RTK fusion partners/families: RET, NTRK1-3, BRAF, ALK, and MET. Proportions tests were used to evaluate significance, with multiple hypothesis correction. We observed that FusionPlex could identify 100% of the 168 described fusions, compared to a detection frequency of 57%, 60%, and 65% for ThyroSpec, IFP, and OFA, respectively (P =1.11e-20). Subset analysis revealed similar significance for BRAF (P =7.36e-19), NTRK1-3 (P =3.69e-7), and MET (P =0.023), however RET and ALK did not reach significance. Collectively, these findings demonstrate the critical importance of panel selection for identifying fusion-driven thyroid cancers and that a gene partner and/or breakpoint-agnostic approach to fusion discovery has improved clinical sensitivity. Additional work will need to quantify the diagnostic utility of agnostic sequencing approaches against both broad scale and hot-spot panels as they relate to metrics of cost-effectiveness and theranostic consequence.