ESTRO 2025 - Abstract Book

S3992

Radiobiology - Tumour radiobiology

ESTRO 2025

Purpose/Objective: Hypoxia (<2% O 2 ) is a well-known adverse prognostic factor in cancer 1 which strongly links to radioresistance 2 . One of the biological mechanisms is through hypoxia-induction of ECM remodelling 3,4 . However, this process is poorly characterised. Here, we aimed to comprehensively analyse the hypoxic extracellular matrix (ECM) in cancer to identify novel radiosensitivity biomarkers. Material/Methods: A meta-analysis (n=21,920 patients; 11 cancer types) using validated cancer-specific hypoxia signatures and random-effect modelling identified hypoxia-associated changes (RNA level). Multi-omics (RNAseq, ChIPSeq, proteomics) confirmed in vitro (T24, J82, UMUC3, RT4) the predicted hypoxia ECM changes. Spatial transcriptomic (n=66 ROIs) validated changes in tumour and stromal (hypoxia identified using CA9 + staining). Cross-comparison identified a gene signature that was tested for prognostic and predictive significance in bladder cancer (n=1,102). The stratification approach was validated in multiple cohorts using Kaplan-Meier, Cox multivariate and neural network models. Mechanistic validation in vitro involved immunofluorescence, cell adhesion, and cell migration assays using irradiated (2-8 Gy) and non-irradiated cells (T24, J82, UMUC3, RT4) seeded onto hypoxic and normoxic ECM. Results: Meta-analysis identified 223 significant genes (fold change >1 or <-1, FDR<0.05) at pan-cancer level, highlighting ECM remodelling as the most prevalent pathway. We validated the meta-analysis predicted hypoxic-ECM in vitro , creating a 55-gene ECM atlas predominantly regulated by HIF1/HIF2 (75% of genes) (Figure 1). The atlas linked hypoxia to ECM organisation and collagen degradation, reproducing the ECM alterations predicted by the meta-analysis. Further validation using spatial transcriptomics revealed a 5-gene signature, which also highlighted different ECM remodelling processes in tumour and stromal areas. The 5-gene signature was predictive for radiotherapy benefit in bladder cancer (p<0.001, AUC=0.73, n=1,102; Figure 2), and prognostic for overall survival (sarcoma, p<0.05, n=259; lung, [p<0.01, n=904], glioblastoma [p<0.01, n=1,118]) and disease-free survival (glioblastoma, p<0.05, n=339; pancreas, p<0.01, n=138). Patients benefiting from radiotherapy had “high” hypoxia-ECM scores and better metastasis-free survival (p<0.0001). In vitro assays confirmed hypoxic ECMs decreased cell migration and increased adhesion, effects enhanced by irradiation. Furthermore, hypoxia decreased collagen fibres, while irradiation decreased cell/ECM interactions, providing mechanistic context.