Editorial for “Intratumoral Spatial Heterogeneity at Dynamic Contrast‐Enhanced <scp>MRI</scp> for Assessing Tertiary Lymphoid Structures in Hepatocellular Carcinoma”

Hepatocellular carcinoma (HCC) remains a global health challenge, characterized by complex biological behavior that critically influences prognosis and treatment outcomes. The advent of immunotherapy has reshaped the treatment paradigm; however, patient responses exhibit marked heterogeneity. Tertiary lymphoid structures (TLSs), ectopic lymphoid aggregates functioning as sites for lymphoid recruitment and immune activation, have emerged as robust biomarkers for antitumor immunity and favorable response to immune checkpoint blockade in HCC [1, 2]. Despite their clinical relevance, accurate preoperative assessment of TLS status remains challenging. Biopsy, the current standard, is invasive and prone to sampling bias due to the pronounced spatial heterogeneity of TLS distribution within the tumor. This underscores an urgent need for noninvasive imaging biomarkers capable of holistically capturing the tumor immune microenvironment. Intratumoral heterogeneity is a hallmark of cancer, reflecting underlying genetic, molecular, and microenvironmental diversity [3]. While conventional radiomics extracts high-throughput features from the entire tumor volume, it may obscure critical spatial variations. Habitat imaging, an emerging paradigm, addresses this by partitioning tumors into physiologically distinct subregions (habitats) based on voxel-wise imaging characteristics, offering a novel framework to spatially quantify heterogeneity [4, 5]. By delineating subregions with distinct hemodynamic patterns, dynamic contrast-enhanced MRI (DCE-MRI)–based habitat analysis potentially bridges between macroscopic imaging and microscopic pathology. In the recent issue of JMRI, the study by Dong et al. [6] presents a compelling investigation into the value of DCE-MRI–based habitat analysis for identifying TLS status in a cohort of 334 HCC patients. By applying a voxel-wise kinetic feature-based unsupervised clustering to quantify intratumoral spatial heterogeneity in HCC, the authors identified four distinct tumor habitats, each exhibiting unique hemodynamic profiles. A key strength of this study lies in its comprehensive integration of four distinct families of habitat-derived features, specifically basic habitat metrics, multiregional spatial interaction (MSI) features, ITH scores, and habitat map-derived radiomics, into a unified predictive framework, systematically exploring the synergistic value of multidimensional information in identifying TLS status. Furthermore, the study rigorously assessed the reproducibility of these novel features by performing inter-observer stability analysis, excluding unstable features to ensure model robustness. The results are encouraging. The proposed hybrid model, which fused clinical variables, conventional radiomics, and habitat-derived features, demonstrated superior discrimination in the training and independent test sets (AUC = 0.83 and 0.84, respectively). Crucially, the hybrid model significantly outperformed the clinical-radiomics model (test set AUC = 0.77; p = 0.002), suggesting that habitat-derived features provide incremental diagnostic value by capturing spatial information potentially missed by global radiomics analysis. These findings substantiate the hypothesis that imaging-defined habitats can serve as a noninvasive surrogate for the underlying immune architecture of the tumor. While the study represents a significant step forward, several open questions remain. First, as a single-center retrospective study, the generalizability of the defined habitats across different scanners, protocols, and diverse patient populations necessitates validation through prospective, multi-center cohorts. Second, although the study establishes a link between habitats and TLS status, the precise biological underpinnings of each habitat, such as their association with specific gene expression programs or cellular compositions, merit further investigation through correlative radiogenomic analyses. Finally, achieving standardization in habitat segmentation and analysis pipelines will be essential to facilitate broader clinical translation. In summary, the study highlights the potential of DCE-MRI habitat imaging as a noninvasive tool for identifying TLS in HCC. By effectively decoding intratumoral spatial heterogeneity, this approach paves the way for more precise patient stratification and personalized immunotherapy strategies. This work was supported by the Taishan Scholar Foundation of Shandong Province (tsqn202211378), the National Natural Science Foundation of China (82371933, 82572188), and the Natural Science Foundation of Shandong Province (ZR2021MH120, ZR2024YQ075).