Radiomic Profiling of Tumor Heterogeneity Predicts Treatment Response in High-Risk Hodgkin Lymphoma: Results from the Phase III AHOD0831 Trial
Cancers · 20.11.2026 · 18 (22) · 3715
DOI: 10.3390/cancers18223715 ↗Kurz gefasst
Fachartikel in der Zeitschrift Cancers, online erschienen am 20.11.2026. Verfasst von Xuemei Ji und 6 weiteren Autor:innen. Die Arbeit nennt 51 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Background/Objectives: To determine whether baseline 18F-FDG PET radiomics improves prediction of treatment response in pediatric and adolescent patients with high-risk Hodgkin lymphoma relative to conventional PET-derived biomarkers.
Methods
Baseline FDG-PET/CT data from 115 patients enrolled in the multicenter AHOD0831 trial were analyzed. Total metabolic tumor volume (TMTV) and total lesion glycolysis (TLG) were calculated from aggregate volume of all lesions. Fifty-nine second-order textural radiomic features were extracted using three fixed SUV discretization bin widths (0.1, 0.25, and 0.5 SUV). Features were selected using univariable testing, correlation-based redundancy reduction, and repeated bootstrapped subsequent least absolute shrinkage and selection operator (LASSO) regression. Logistic-regression models based on selected radiomic features, radiomic features plus TMTV, and conventional PET biomarkers were evaluated using repeated five-fold cross-validation, receiver-operating-characteristic analysis, and decision-curve analysis.
Results
Of 115 patients, 54 (47.0%) achieved complete response, while 61 (53.0%) had partial response or stable disease. A core group of heterogeneity-related radiomic features—including gray-level co-occurrence matrix-based informational measures of correlation, gray-level dependence matrix-based gray-level non-uniformity, and neighboring gray tone difference matrix-based coarseness and strength—was retained across discretization schemes. Repeated LASSO resampling identified a limited subset of stable predictors. Across all datasets, radiomics-only models and hybrid models combining radiomic features with TMTV consistently outperformed conventional models based on TMTV and TMTV plus TLG. Exploratory decision-curve analysis suggested greater estimated net benefit for the best-performing radiomics model than for TMTV alone within a potentially clinically relevant threshold-probability range.
Conclusions
Baseline FDG-PET radiomics captures intratumoral metabolic heterogeneity and may improve treatment-response prediction in high-risk Hodgkin lymphoma beyond conventional PET-derived biomarkers. These findings warrant independent external validation to determine whether baseline radiomics can improve risk stratification and inform future response-adapted treatment strategies.
Zitieren
APA
Ji, X., Anderson, P., Huang, J., Kuker, R., Alderuccio, J. P., Dogan, N., & Yang, F. (2026). Radiomic Profiling of Tumor Heterogeneity Predicts Treatment Response in High-Risk Hodgkin Lymphoma: Results from the Phase III AHOD0831 Trial. Cancers, 18(22), 3715. https://doi.org/10.3390/cancers18223715
Vancouver
Ji X, Anderson P, Huang J, Kuker R, Alderuccio JP, Dogan N, et al. Radiomic Profiling of Tumor Heterogeneity Predicts Treatment Response in High-Risk Hodgkin Lymphoma: Results from the Phase III AHOD0831 Trial. Cancers. 2026;18(22):3715. doi:10.3390/cancers18223715
BibTeX
@article{Ji2026,
author = {Ji, Xuemei and Anderson, Parker and Huang, Jianfeng and Kuker, Russ and Alderuccio, Juan Pablo and Dogan, Nesrin and Yang, Fei},
title = {Radiomic Profiling of Tumor Heterogeneity Predicts Treatment Response in High-Risk Hodgkin Lymphoma: Results from the Phase III AHOD0831 Trial},
journal = {Cancers},
year = {2026},
volume = {18},
number = {22},
pages = {3715},
doi = {10.3390/cancers18223715}
}
EUVIMED