Author(s):
Esperança-Martins, Miguel ; Vasques, Hugo ; Ravasqueira, Manuel Sokolov ; Lemos, Maria Manuel ; Fonseca, Filipa ; Coutinho, Diogo ; López, Jorge Antonio ; Huang, Richard S.P. ; Dias, Sérgio ; Gallego-Paez, Lina ; Costa, Luís ; Abecasis, Nuno ; Gonçalves, Emanuel ; Fernandes, Isabel
Date: 2025
Persistent ID: http://hdl.handle.net/10362/188436
Origin: Repositório Institucional da UNL
Subject(s): CINSARC; consensus clustering; DNA-seq; precision treatment; prognosis; RNA-seq; SARCULATOR; soft tissue sarcomas; therapeutic targets; unsupervised machine learning; Oncology; Cancer Research; SDG 3 - Good Health and Well-being
Description
Funding Information: This research was sponsored by F.Hoffmann-LaRoche AG and by Foundation Medicine Inc. under the RNA LDT Research Programme. Funding Information: We would like to thank F.Hoffmann-LaRoche AG and Foundation Medicine Inc. for providing the FoundationOneCDx and FoundationOneRNA assays, and for all the technical support, specifically during the transfer of the sequencing data via the safe platform. We would like to thank Sarah Yacoub (Foundation Medicine Inc) for all the precious help in the logistical operationalization of the transfer of all of the different batches of samples between iMM (Lisbon, Lisbon, Portugal) and the Foundation Medicine Headquarters (Cambridge, Massachusetts, United States of America) and for her valuable assistance on sending the results of both DNA-seq and RNA-seq results via a safe platform. We would also like to thank Rachel Beth Keller-Evans (Foundation Medicine Inc) for her crucial support in the analysis of the data that has resulted from the employment of the FoundationOneRNA assay, namely, for her help in the analysis of the detected fusions. We would also like to show our deepest gratitude to the IPOLFG tumor biobank staff, that were responsible for the original retrieval and organization of the FFPE samples that were used in this study, the IPOLFG pathology department staff, and to the iMM Comparative Pathology unit team (that have sectioned the blocks), the iMM Translational Oncobiology Lab staff (that have helped in the preparation and shipment of the different batches of samples), and the iMM Technology Transfer Office staff. Emanuel Gon\u00E7alves work is supported by FCT (Funda\u00E7\u00E3o para a Ci\u00EAncia e Tecnologia), under projects UIDB/50021/2020 (DOI:10.54499/UIDB/50021/2020), SARC-RON-AI ( https://doi.org/10.54499/2024.07252.IACDC , through RE-C05-i08.M04), and SYNTHESIS (LISBOA2030-FEDER-00868200). We would like to thank Tiago Barroso for the design of the graphical abstract. We would finally like to than Brian Van Tine and Alliny C S Bastos for their exquisite input and feedback. Publisher Copyright: © 2025 by the authors.
Background: Soft tissue sarcomas (STSs) histopathological classification system and the clinical and molecular-based tools that are currently employed to estimate its prognosis have several limitations, impacting prognostication and treatment. Clinically driven molecular profiling studies may cover these gaps and offer alternative tools with superior prognostication capability and enhanced precision and personalized treatment approaches identification ability. Materials and Methods/Results: We performed DNA sequencing (DNA-seq) and RNA sequencing (RNA-seq) to portray the molecular profile of 102 samples of high-grade STS, comprising the three most common STS histotypes. The analysis of RNA-seq data using unsupervised machine learning models revealed previously unknown molecular patterns, identifying four transcriptomic subtypes/clusters (TCs). This TC-based classification has a clear prognostic value (in terms of overall survival (OS) and disease-free survival (DFS)), a finding that was externally validated using independent patient cohorts. The prognostic value of this TC-based classification outperforms the prognostic accuracy of clinical-based (SARCULATOR nomograms) and molecular-based (CINSARC) prognostication tools, being one of the first molecular-based classifications capable of predicting OS in STS. The analysis of DNA-seq data from the same cohort revealed numerous and, in some cases, never documented molecular targets for precision treatment across different transcriptomic subtypes. The functional and predictive value of each genomic variant was analyzed using the Molecular Tumor Board Portal. Conclusions: This newly identified TC-based classification offers a superior prognostic value when compared with current gold-standard clinical and molecular-based prognostication tools, and identifies novel molecular targets for precision treatment, representing a cutting-edge tool for predicting prognosis and guiding treatment across different stages of STS.
