BASELINE PET RADIOMICS OUTPERFORMS CLINICAL RISK SCORES IN PREDICTING PRIMARY MEDIASTINAL B‐CELL LYMPHOMA OUTCOME: INSIGHTS FROM THE IELSG37 STUDY

Introduction: A major challenge in treating primary mediastinal B-cell lymphoma (PMBCL) is the early identification of patients who are likely to fail first-line therapy. To address this issue, present analysis explored potential using PET radiomics, a new technique for extracting quantitative parameters from images, predict treatment outcomes PMBCL patients. Methods: The study examined baseline PET/CT scans 501 treated with rituximab and doxorubicin-based immunochemotherapy regimens, 342 them receiving consolidation radiotherapy. Functional indicating tumor burden (metabolic volume, MTV), glucose consumption (total lesion glycolysis, TLG), heterogeneity (area under curve cumulative SUV-volume histogram, AUC-CSH) were calculated each patient segmentation algorithm 25%-SUVmax threshold. In addition, 107 radiomics features (RF) extracted 495 PyRadiomics package. 74 RFs uncorrelated SUVmax MTV selected then machine-learning approach based on recursive-partitioning classification tree (CTree) method. Statistical was performed either Stata-17 or R software packages, as appropriate. Results: CTree identified single RF significant prognostic impact: Grey-Level-Run- Length-Matrix run variance (GLRLMrv), which marker metabolic (MH). Patients low MH (GLRLMrv <0.137) had 93% progression-free survival (PFS) rate at 5 years, while those high 65% PFS (log-rank test, p < 0.0001). Overall (OS) rates 96% 80%, respectively subsequent conducted, including GLRLMrv, well other dichotomized clinical variables metrics that association (p 0.05) univariable (TLG, MTV, SUVmax, SUVpeak, maximal diameter stage, sex, LDH, number extra-nodal localizations). selection generated model combination GLRLMrv TLG, discriminated different OS 0.0001 both). Concordance probability estimate showed higher predictive accuracy (highest Harrell’s C value) radiomic comparison main international indices, namely IPI, revised-IPI age-adjusted IPI. better discriminatory power also confirmed by its lowest Akaike’s information criterion. research funded part grants Swiss National Science Foundation (SNSF) – Project 32003B_146931, Krebsforschung Schweiz KFS-2852-08-2011 Cancer Research UK (C30423/A16247) enrolment supported Group Clinical (SAKK). This grant League KLS-5406-08-2021. Keywords: diagnostic biomarkers, PET-CT No conflicts interests pertinent abstract.