Radiomics signature for dynamic monitoring of tumor inflamed microenvironment and immunotherapy response prediction

• DOI: 10.1136/jitc-2024-009140
• Publication Date: 2025-01-12T04:45:13Z
• Authors: Kinga Bernatowicz; Ramon Amat; Olivia Prior; Joan Frigola; Marta Ligero; Francesco Grussu; Christina Zatse; Garazi Serna; Paolo Nuciforo; Rodrigo Toledo; Manel Escobar; Elena Garralda; Enriqueta Felip; Raquel Perez-Lopez
• Keywords: Signature (topology)

The efficacy of immune checkpoint inhibitors (ICIs) depends on the tumor microenvironment (TIME), with a preference for T cell-inflamed TIME. However, challenges in tissue-based assessments via biopsies have triggered exploration non-invasive alternatives, such as radiomics, to comprehensively evaluate TIME across diverse cancers. To address these challenges, we develop an ICI response signature by integrating radiomics gene-expression profiles. We conducted pan-cancer investigation into utility assessment, including 1360 tumors from 428 patients. Leveraging contrast-enhanced CT images, characterized through RNA gene expression analysis, using signature. Subsequently, CT-radiomic predicting inflamed (CT-TIME) was developed and externally validated. Machine learning employed select robust radiomic features predict study also integrated independent cohorts longitudinal baseline biopsies, comprehensive immunohistochemistry panel evaluation assess biological associations, spatiotemporal landscape clinical CT-TIME. CT-TIME signature, comprising four linked T-cell microenvironment, demonstrated performance AUCs (95% CI) 0.85 (0.73 0.96) (training) 0.78 (0.65 0.92) (external validation). scores exhibited positive correlations CD3, CD8, CD163 expression. Intrapatient analysis revealed considerable heterogeneity between tumors, which could not be assessed biopsies. Evaluation aggregated per-patient highlighted its promising dynamically assessing immunotherapy scenarios advanced cancer. Despite demonstrating progression disease at first follow-up, patients within status group, identified CT-TIME, significantly prolonged progression-free survival (PFS), some surpassing 5 months, suggesting potential phenomenon pseudoprogression. Cox models images statistically significant reduction risk PFS cohort (HR 0.62, 95% CI 0.44 0.88, p=0.007), Kaplan-Meier further confirmed substantial differences uninflamed (log-rank test p=0.009). holds promise impacting decision-making, patient stratification, treatment outcomes therapies.