Immune checkpoint inhibitors help only a small fraction of soft-tissue sarcoma (STS) patients, and there's no reliable, non-invasive way to tell whose tumors are immunologically "hot." Using a prospective cohort of 85 high-grade STS patients from the NEOSARCOMICS trial, the team asked whether metabolic activity on routine 18F-FDG-PET/CT tracks the immune landscape. Tumors classified as metabolic-high showed enriched CD8 T-cell transcriptomic programs and significantly denser CD8+, CD14+, CD45+, CD68+, and c-MAF infiltration than metabolic-low tumors. Tertiary lymphoid structure status was independent of metabolism — pointing to two complementary axes for immunotherapy patient selection.
Published in Biomarker Research, this correspondence comes out of the NEOSARCOMICS precision-medicine trial (NCT02789384), with Amandine Crombé (Institut Bergonié; Pellegrin University Hospital) as first and corresponding author and Prof. Antoine Italiano (Institut Bergonié; INSERM U1312 BRIC, SARCOTARGET team) as senior author. Explicyte was in charge of the multiplex immunofluorescence/IHC immune-profiling panels that anchored the imaging–immune correlation at the protein level.
The team extracted maximum, mean, and peak SUVs, metabolic tumor volume, and total lesion glycolysis from pre-treatment 18F-FDG-PET/CT for all 85 patients, then ran cross-validated principal component analysis. PC1 — correlated with all PET metrics — split the cohort into three metabolic profiles: metabolic-low (60%), metabolic-intermediate (15.3%), and metabolic-high (24.7%).
RNA sequencing on 32 tumors identified 67 differentially expressed genes between metabolic-low and metabolic-high samples, with the CD8 T-cell pathway enriched in the metabolic-high group. In the 7 most extreme metabolic-low tumors, 391 genes were differentially expressed and 13 LM22 immune genesets were suppressed — including ICOS, CD27, interferon-γ, and the CXCL9-10-11/CXCR3 axis — alongside downregulated cell-cycle genes E2F1, CDKN2A, and CCNB1. No association was found with the CINSARC signature (P = 0.176).
Explicyte ran a six-marker multiplex immunofluorescence/IHC panel (c-MAF, CD8, CD14, CD20, CD45, CD68) on 31 patients. Densities of CD8+, CD14+, CD45+, CD68+, and c-MAF cells were significantly higher in metabolic-high than metabolic-low tumors, and correlated positively with metabolism along PC1 (P-value range 0.0247–0.0499). The contrast was stark in matched examples: a metabolic-high tumor (SUVmax 22.4) showed a CD8+ density of 539 mm⁻² versus 1 mm⁻² in a metabolic-low tumor (SUVmax 5.1).
TLS status, assessed in all 85 patients, showed no association with metabolic group, PC1, PC2, or any raw PET metric — indicating that tumor metabolism and TLS capture distinct, non-overlapping features of the STS immune microenvironment.
The findings reframe routine 18F-FDG-PET/CT as a potential non-invasive readout of the sarcoma immune microenvironment — and position metabolism and TLS as complementary, independent stratifiers.
For drug developers, metabolic imaging that patients already receive could enrich immunotherapy trials by flagging immunologically active sarcomas non-invasively. Because metabolism and TLS status move independently, combining PET metrics with TLS scoring may stratify STS patients better than either alone. The immune-suppressed profile of extreme metabolic-low tumors — with silenced IFN-γ and CXCL9/10/11 signaling — also marks a subset unlikely to respond to checkpoint blockade without combination strategies.
