AUTHOR=Almeida Jani Sofia , Sousa Luana Madalena , Couceiro Patrícia , Alves Vera , Rodrigues Joana , Fonseca Ruben , Freitas-Tavares Paulo , Santos-Rosa Manuel , Casanova José Manuel , Rodrigues-Santos Paulo TITLE=AI-assisted peripheral immune profiling reveals unconventional lymphocyte signatures associated with prognosis in soft tissue sarcoma patients JOURNAL=Frontiers in Immunology VOLUME=Volume 16 - 2025 YEAR=2025 URL=https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2025.1677408 DOI=10.3389/fimmu.2025.1677408 ISSN=1664-3224 ABSTRACT=Immunotherapy has reshaped the treatment of several cancers, yet patient responses remain highly variable, partly due to differences in immune competence. In soft tissue sarcomas (STS), the immune landscape is poorly characterized, limiting the development of prognostic markers and immune-based therapeutic strategies. This study aimed to comprehensively profile circulating and tumor-infiltrating cytotoxic lymphocyte populations in STS. Peripheral blood from patients and healthy donors was analyzed by multiparametric flow cytometry combined with AI-assisted unsupervised clustering, enabling the identification of both conventional and unconventional subsets. In a pilot cohort, tumor-infiltrating lymphocytes were evaluated using the same approach to explore systemic–local immune compartmentalization. STS patients displayed systemic immune imbalance with increased CD8+ T cells and reduced NK cells and CD161+ CD8+ T cells, consistent with overall immunosuppression. Several unconventional populations showed prognostic associations: elevated CD8+ γδ T cells and CD4+ NKT-like cells correlated with poorer survival, whereas CD8+ NKT-like cells were enriched in immune-competent patients and linked to better outcomes, suggesting potential protective functions. Pilot tumor analyses identified γδ NKT-like cells that were nearly absent from circulation, suggesting their selective enrichment within the tumor microenvironment. Together, these findings highlight the contribution of rarely profiled cytotoxic lymphocytes to systemic immune fitness and disease outcome in STS. Importantly, despite clinical and histological heterogeneity, patients showed consistent immune alterations, suggesting shared immunological features across STS subtypes. While limited by small tumor sample size and lack of functional assays, this study provides proof-of-concept that immune-based profiling can uncover novel prognostic markers and candidate populations of therapeutic relevance. Future work in larger, longitudinal cohorts, coupled with functional characterization, will be essential to validate these subsets and to define their role in STS immune surveillance and responsiveness to immunotherapy.