Intratumoral Metabolic Heterogeneity Predicts Treatment Response in High-Risk Hodgkin Lymphoma: Evidence from the AHOD0831 Trial

Background: Conventional imaging biomarkers for high-risk Hodgkin lymphoma—such as metabolic tumor volume (MTV) and maximum standardized uptake value (SUVmax)—do not account for intratumoral metabolic heterogeneity (IMH). This study evaluates the predictive value of baseline IMH quantified by Image Biomarker Standardization Initiative (IBSI)-compliant radiomics relative to these conventional metrics in pediatric patients with high-risk Hodgkin lymphoma.

Methods: The current study analyzed 110 pretreatment FDG-PET/CT scans from high-risk Hodgkin lymphoma patients enrolled in AHOD0831. Radiomics texture features (GLCM: 24; GLRLM/GLSZM: 16 each; GLDM: 14; NGTDM: 5) were extracted from predefined metabolic tumor volumes following IBSI guidelines. Complete response (CR) vs. non-complete response (nCR) served as the primary endpoint. Binary logistic regression models (BLRMs) were built using features most strongly associated with the endpoint, with performance evaluated by receiver operating characteristic (ROC) analysis against benchmarks using MTV and SUVmax.

Results: BLRM using the top radiomics feature (IBSI: glcm_LMC1) achieved an AUC of 0.68, outperforming MTV alone (AUC 0.62; DeLong p < 0.001) and SUVmax alone (AUC 0.50; p < 0.001). Incorporating the second-best feature (IBSI: ngtdm_Coarseness) further improved AUC to 0.70, surpassing the combined MTV and SUVmax model (AUC 0.62; p < 0.001).

Conclusion: Radiomics-derived IMH captures tumor metabolic heterogeneity overlooked by conventional metrics, demonstrating superior predictive performance for treatment response. These findings position IMH as a promising biomarker to guide risk-adaptive therapies in high-risk Hodgkin lymphoma.