AUTHOR=Zhou Xiaofang , Su Xiaoli , Yu Lan , Wang Feng , Yu Shujie , Yu Feifei , Lin Xiaoye , Song Yang , Cao Dairong , Wang Xingfu , Xing Zhen 

TITLE=Evaluation of programmed cell death ligand-1 expression in primary central nervous system lymphoma using whole-tumor histogram analysis of multiparametric MRI: implications for immunotherapy selection

JOURNAL=Frontiers in Immunology

VOLUME=Volume 16 - 2025

YEAR=2025

URL=https://www.frontiersin.org/journals/immunology/articles/10.3389/fimmu.2025.1676273

DOI=10.3389/fimmu.2025.1676273

ISSN=1664-3224

ABSTRACT=ObjectiveTo assess the diagnostic performance of whole-tumor histogram analysis of multiparametric MRI in predicting programmed cell death ligand-1 (PD-L1) expression in primary central nervous system lymphoma (PCNSL).MethodsA total of 130 patients with PCNSL (61 males, aged 21–80 years) were included in the study. Histogram features derived from T2-weighted imaging (T2WI), T1-weighted imaging (T1WI), fluid-attenuated inversion recovery (FLAIR), contrast-enhanced T1-weighted imaging (T1WI+C), and apparent diffusion coefficient (ADC) were compared between the low and high PD-L1 expression groups using the Mann-Whitney U test. Receiver operating characteristic (ROC) curves and logistic regression analysis were applied to assess the diagnostic performance of both individual and combined models in predicting PD-L1 expression levels in PCNSL.ResultsEighteen histogram features extracted from multiparametric MRI exhibited significant differences between high and low PD-L1 expression in PCNSL (all P < 0.05). The predictive performance of single-sequence models was relatively modest, with areas under the curve (AUC) ranging from 0.637 to 0.705, and no significant differences were observed between these models (all P > 0.05). The combined model demonstrated the highest diagnostic performance (AUC = 0.809), significantly outperforming the single-sequence models (all P < 0.05).ConclusionsWhole-tumor histogram analysis of multiparametric MRI shows potential as a non-invasive method for evaluating PD-L1 expression in PCNSL, which may assist in the identification of immunotherapy-eligible patients.