(P 382) NTRK1/2/3 FUSIONS DETECTION USING RNA-SEQ IN MESENCHYMAL TUMORS REVEALS ULTRA-RARE AND HISTO-SPECIFIC TARGETABLE ALTERATIONS: EXPERIENCE OF AN EXTENSIVE ROUTINE RESEARCH PROGRAM.

Objective: Since the histo-agnostic approvals of NTRK1/2/3 tyrosine kinase inhibitors (NTRKi), a broad research effort for NTRK1, NTRK2 or NTRK3 fusion transcripts by RNA-seq has been developed. In sarcomas, 5 typically NTRK-fused histotypes have been identified (infantile fibrosarcomas (IFS), lipofibromatosis-like neural tumors(LLNT), ALK/ROS1 wild type inflammatory mesenchymal tumors(IMT), NTRK-rearranged spindle cell neoplasms (NRSCN), and KIT/PDGFRA wild type gastro-instestinal tumors (wtGIST)). Yet, little has been described regarding the prevalence and relevance of NTRK translocations identified in mesenchymal tumors. The objective of this study is to provide an extensive landscape of FFPE samples whole-RNA-seq (WERS) in mesenchymal tumors and to propose a several steps process to ensure relevance of identified NTRK fusions.

Methods: This study was performed as a retrospective translational research program on mesenchymal tumor of pediatric and adult patients from French cancer centers diagnosed between 2016 and March 2023. All diagnosis were histologically and molecularly reviewed by expert pathologists. WERS was performed with FFPE material using TruSeq RNA Access Library Prep Kit (Illumina®, San Diego, USA). Bioinformatics analyses (described in PMID: 35080790) included 4 to 6 different algorithms for fusion detection. A multidisciplinary molecular board reviewed the WERS results in term of technical, biological and clinical interpretations.

Results: WERS was performed in a cohort of 3015 cases composed of suspicions of both malignant and benign mesenchymal tumors which most important subtypes are summarized (Table1). At least 1 fusion of NTRK1, NTRK2 or NTRK3 was reported in n=111 (3.68%) cases bioinformatically. Fourty-three cases were untargetable fusions by TKI due to a loss of kinase domain in the fusion or to a frameshift in the NTRK1/2/3 fused, so remaining 2.25% (68/3015) of the cohort were biologically actionable. Another 11 samples had fusions with criteria of technical false positive results due association of to a very low number of split/span reads associated to inconsistant results in fusion calling algorithm and inconsitant partner. Hence, 1.92% (57/3015) of all tumors had biologically and technically reliable NTRK fusions accounting for 51% (57/111 ) of the baseline possibly NTRK-fused tumors bioinformatically identified.
This cohort (see Figure 1) was composed of a large majority of tumors were pathologist had no clear differential diagnosis and could be classified typical NTRK-fused histological subtypes (11 IFS, 14 LLNT, 9 IMT , 19 NRSCN, 1 GIST). Only 3 sarcomas with a prior clear differential diagnosis by pathologist had a possibly targetable NTRK fusions (1 dedifferenciated liposarcoma with MDM2-amplification, 1 intimal sarcoma of the pulmonary arteria with at leat MDM2-overepression, 1 leiomyosarcoma. All had unsual fusions and , no NTRK inibitor has been administered to these patients with the clinical experience available.

Conclusion: WERS is sensitive to detect NTRK fusions in FFPE mesenchymal tumors although the bioinformatically detected fusions must be filtered for biological or technical irrelevance in order to further identify targetable alterations. The prevalence of possibly targetable NTRK1/2/3 fusions were mainly "histo-specific" to the 5 subtypes of “simple genomic sarcomas”. In these cases, the presence of the fusion could be histologically suspected and/or no clear sarcoma subtype was reach prior to the NTRK identification. Besides, the fusion is extremely rare and clinical activity of NTRKi remains unreported in MDM2-amplified or “complex genomic sarcomas”. Our study may help to refine the NTRK fusion screening strategy in mesenchymal tumors and discuss biases of NTRK fusion detection.