S2535
Radiobiology – Immuno-radiobiology
ESTRO 2026
0.67, and 0.72 at 1, 3, 12, and 18 months, respectively), with HR(TRIscore)=1.45 [1.16–1.81], p=0.0012. Kaplan– Meier analysis of the full model (Figure 1) showed clear separation between high- and low-risk groups derived from the Youden cut-off of the training Cox model.
Digital Poster Highlight 2927
Decoding systemic inflammation: the TRIscore as a unifying prognostic biomarker in radiotherapy & radio-immunotherapy Lisa BOUARROUDJ 1,2 , Nicolas HARDY 1 , Antonin LEVY 1,3 , Christophe MASSARD 4,5 , Yohann LORIOT 6,5 , Eric DEUTSCH 1,2 , Daphné MOREL 1,2 1 Department of Radiation Oncology, Gustave Roussy, Villejuif, France. 2 Inserm U1030, Université Paris- Saclay, Gustave Roussy, Villejuif, France. 3 Faculté de Médecine, Université Paris Saclay, Le Kremlin-Bicêtre, France. 4 Gustave Roussy, Inserm U1030 Université Paris-Saclay, Villejuif, France. 5 Drug Development Department (DITEP), Gustave Roussy, Villejuif, France. 6 Department of Medical Oncology, Gustave Roussy, Villejuif, France Purpose/Objective: Accurate prognostication in advanced cancer is critical for guiding treatment strategies and improving outcomes. However, reliable biomarkers remain limited. We developed and evaluated the Tumor- Related Inflammation Score (TRIscore), a composite index derived from routine blood counts, designed to capture the systemic inflammatory response associated with cancer progression. Its prognostic value and generalizability were assessed in patients treated with radiotherapy (RT) with and without immunotherapy. Material/Methods: The TRIscore was computed from baseline blood counts combining three systemic inflammation ratios: monocyte-to-lymphocyte ratio (MLR), neutrophil-to- lymphocyte ratio (NLR), and systemic inflammation index (SII), using the formula √ (MLR ² +NLR ² +SII ² ). Prognostic performance for overall survival (OS) was evaluated using Cox models. The full model included TRIscore, age, ECOG performance status, and gender as covariates; a simplified model with TRIscore alone was also tested.Discrimination was assessed by time-dependent area under the ROC curve (AUC(t)), and robustness by leave-one-out cross- validation (LOOCV). Training was performed on two radio-immunotherapy cohorts (SABR-PDL1, NCT02992912; ABIMMUNE, NCT03212469). External validation was conducted on three retrospective cohorts of deceased patients treated at Gustave
The TRIscore-only model provided comparable performance (HR=1.59 [1.29–1.96]), confirming that the score alone is a strong prognostic factor. LOOCV confirmed internal robustness (AUC(t) ≈ 0.66– 0.72).Kaplan–Meier analyses of the TRIscore-only model (Figure 2) showed consistent patient stratification across RT-alone (n=924), immunotherapy-alone (n=448), and immunotherapy- combination (n=308) cohorts, with higher TRIscore values systematically associated with shorter OS (AUC(1 month)=0.83, 0.72, and 0.70, respectively).
Conclusion: TRIscore, derived from routine blood parameters, emerges as a robust, self-sufficient, and cost-free biomarker for prognostication in patients receiving RT with or without immunotherapy. Beyond its clinical utility, TRIscore captures systemic inflammation as a key determinant of treatment efficacy and provides a framework for immune-preserving radiation strategies. Prospective and longitudinal validation will clarify its value for adaptive treatment guidance in combined radio-immunotherapy. Keywords: Prognostic score, Systemic inflammation
Roussy between 2020 and 2024: RT alone, immunotherapy alone, and immunotherapy
combinations. For Kaplan–Meier analyses, patients were stratified into high- and low-risk groups using the optimal Youden cut-off from the linear predictor of the training model. Results: In the training set (n=129), the full model achieved stable discrimination over time (AUC(t)=0.69, 0.71,
Made with FlippingBook - Share PDF online