Commentary on Multimodal profiling identifies CD4 + CXCR5 + PD-1 – Tfh cells as prognostic and predictive biomarkers in diffuse large B-cell lymphoma

As a highly heterogeneous malignant lymphoma, the mechanisms underlying relapsed and refractory diffuse large B-cell lymphoma (R/R DLBCL) and the prediction of treatment response have long been challenging issues in clinical research. Through technologies such as single-cell sequencing and multiplex immunohistochemistry (mIHC), this multimodal study by Yu et al[1] has clearly identified CD4+CXCR5+PD-1– follicular helper T (Tfh) cells (here after referred to as “CD4+ Tfh1 cells”, specifically referring to the CD4+CXCR5+PD-1– Tfh subset) as prognostic biomarkers for DLBCL patients and predictive biomarkers for the response to the R-CHOP regimen, providing a new perspective for the precise diagnosis and treatment of this disease. The core advantage of this study lies in the adoption of a “multimodal profiling” strategy, which constructs an evidence chain from three levels (cellular, molecular, and clinical) to ensure the reliability of the results. The study included 9 samples from newly diagnosed DLBCL patients (8 used for single-cell sequencing), validated the findings using single-cell data of 72,351 cells, and conducted in vivo experiments with mouse models. This formed a longitudinal research dimension covering DLBCL patients from onset to relapse. It integrated single-cell RNA sequencing (scRNA-seq) to analyze cell composition and gene expression, DNA sequencing to analyze copy number variations (CNVs), and mIHC to observe the spatial distribution of cells. Additionally, it innovatively introduced a conditional generative adversarial network (GAN) model to predict mIHC results from 4′,6-diamidino-2-phenylindole (DAPI)-stained images, addressing the practical challenges of obtaining mIHC images from clinical samples. By conducting in-depth analysis of the cell composition and functions in the DLBCL tumor microenvironment (TME), the study revealed differences between newly diagnosed and relapsed patients, and clarified the key role of CD4+ Tfh1 cells. It is crucial to distinguish this cell subset from the classic PD-1+ Tfh cells extensively studied in previous research: classic PD-1+ Tfh cells are mainly localized within germinal centers, characterized by a high level of expression of PD-1, inducible co-stimulator (ICOS), and interleukin 21 (IL-21), and their core function is to promote B cell differentiation and antibody production. In contrast, CD4+ Tfh1 cells identified in this study exhibit unique gene expression profiles (e.g., related to the tumor necrosis factor alpha/nuclear factor-kappa B [TNF-α/NF-κB] signaling pathway and nuclear factor of activated T cells 1 [NFATC1] regulation), and exert anti-tumor effects through synergistic interactions with CD8+ T cells, representing a functionally distinct subpopulation that enriches the understanding of Tfh cell heterogeneity in DLBCL. In relapsed patients, the proportion of malignant B cells (B_mal) increased, while the proportion of immune cells such as CD4+ T cells and CD8+ T cells decreased. Moreover, regulatory T cells (Tregs) were enriched. In contrast, newly diagnosed patients had a higher abundance of cells with anti-tumor functions, such as CD4+ Tfh1 cells and CD8+ effector memory T cells. The proportion of CD4+ Tfh1 cells was significantly higher in patients sensitive to R-CHOP regimen treatment. Additionally, CD8+ T cells were more densely distributed within a 100 μm range around these cells, and the shorter the distance between the two cell types, the longer the patient’s survival time. Further validation using animal experiments confirmed that a high level of CD4+ Tfh1 cells was directly associated with a better prognosis. These findings collectively support the role of these cells as dual biomarkers for disease prognosis and treatment response prediction, providing new evidence for the functional significance of this previously unreported Tfh subpopulation in DLBCL. In the context of existing DLBCL research, recent studies using single-cell sequencing to explore the TME have provided valuable comparative frameworks. For example, Dai et al[2] integrated single-cell transcriptomics, spatial transcriptomics, bulk RNA-seq, and immunohistochemistry, and dissected the interaction mechanisms, immunosuppressive effects and independent prognostic value of glycolysis-high B cells and IFN+ tumor-associated macrophages (TAMs) in the TME of DLBCL, but did not report the CD4+CXCR5+PD-1– Tfh subpopulation described here. This indicates that Yu et al’s study fills a gap in the identification of Tfh cell subsets in DLBCL. Compared with other studies focusing on Treg infiltration or CD8+ Tfh cell exhaustion, this research highlights the unique prognostic and predictive value of CD4+ Tfh1 cells, supplementing the existing cognition of DLBCL immune biomarkers. Notably, while some studies have reported the correlation between total Tfh cell infiltration and DLBCL prognosis, Yu et al’s work further refines this association by defining a specific PD-1– subpopulation, providing more precise targets for clinical application. This study fills the research gap in some recurrence mechanisms and treatment prediction of DLBCL, and also holds important practical value in terms of clinical translation. Currently, the efficacy evaluation of DLBCL relies on imaging results (such as positron emission tomography-computed tomography [PET-CT], magnetic resonance imaging [MRI], and CT) and lacks early predictive indicators. Compared with existing predictive tools for R-CHOP response, including the International Prognostic Index (IPI), molecular subtypes (e.g., germinal center B-cell-like vs. activated B- cell-like), and biomarkers such as MYC/BCL2 double expression, CD4+ Tfh1 cells offer unique advantages. IPI primarily reflects clinical risk factors but fails to capture the dynamic immune microenvironment; molecular subtypes focus on tumor cell intrinsic characteristics, while CD4+ Tfh1 cells integrate both tumor and immune system interactions. Preliminary data suggest that combining CD4+ Tfh1 cell proportion with IPI may improve prediction accuracy, especially for patients with intermediate IPI scores who are currently difficult to stratify. This supports the potential of CD4+ Tfh1 cells as a complementary predictive indicator to optimize clinical decision-making. The GAN model in this study can generate mIHC-like results from conventional DAPI-stained images, with a cell recognition accuracy rate of 0.86. This provides a low-cost and easily promotable technical approach for primary hospitals, effectively mitigating the challenges of high sample requirements and limited accessibility associated with traditional mIHC detection. It is important to note that while the model addresses key practical barriers, it does not fully solve all limitations of mIHC (e.g., it cannot replicate the full spectrum of protein expression detection or resolve ultra-fine spatial interactions with the same precision as direct staining). Moreover, the mechanisms identified in this study, such as the activation of the TNF-α/NF-κB signaling pathway and the abnormal expression of NFATC1, provide new directions for DLBCL targeted therapy. For example, inhibiting NFATC1 or regulating the TNF-α pathway can reverse the immunosuppressive microenvironment and enhance the anti-tumor synergistic effect of CD4+ Tfh1 cells and CD8+ T cells. To further elaborate on the mechanistic basis: scRNA-seq data from the study show that Tfh1 cells highly express TNF-α, which may activate NF-κB signaling in CD8+ T cells to enhance their cytotoxicity; CD4+ conversely, abnormal NFATC1 overexpression in malignant B cells may suppress CD4+ Tfh1 cell function through paracrine signaling. These interactions suggest a bidirectional regulatory network between CD4+ Tfh1 cells and tumor cells, laying the foundation for future mechanistic investigations. Despite the significant achievements of this study, there are still areas that need further improvement. The current study included a small number of de novo samples (9 cases) and lacked long-term follow-up data (such as 3–5-year progression-free survival and overall survival). Multicenter, large-sample cohort studies are needed to verify the clinical practical value of CD4+ Tfh1 cells as biomarkers. The study preliminarily revealed that CD4+ Tfh1 cells exert their effects through synergy with CD8+ T cells, but the specific molecular interaction mechanisms between the two (such as cytokine regulation and co-stimulatory molecule expression) remain unclear. For instance, do CD4+ Tfh1 cells secrete IL-2 or IFN-γ to enhance CD8+ T cell activity? Are there direct cell–cell interactions mediated by molecules like CD40L-CD40? Answering these questions requires in vitro co-culture experiments and protein interaction assays. In addition, further in vitro functional experiments are required to verify how NFATC1 regulates the immune escape of malignant B cells. The GAN model is still in the laboratory validation stage, and multicenter clinical trials are needed to evaluate the consistency of its detection results across different pathological platforms and sample types. Similar AI-driven studies have been reported in DLBC, such as models applied in the review on the performance and clinical utility of AI for digital pathology and the classification of DLBCL.[3] Yu et al’s GAN model is distinctive in its focus on immune cell subset quantification rather than tumor classification, but direct comparisons of technical performance (e.g., reproducibility, cross-cohort generalization) with these existing tools would help clarify its relative advantages. At the same time, targeted intervention strategies based on CD4+ Tfh1 cells (such as Tfh cell adoptive therapy and combined application of TNF-α inhibitors) also need to undergo clinical trials to explore their safety and efficacy. In summary, this study focuses on CD4+ Tfh1 cells and established a research framework of “cellular mechanism–biomarker–clinical translation” for DLBCL. By clarifying the unique characteristics of CD4+ Tfh1 cells relative to classic Tfh cells, comparing their predictive value with existing indicators, and contextualizing the findings within the broader landscape of DLBCL TME research, the study not only provides new insights into disease recurrence mechanisms but also is likely to promote the advancement of DLBCL diagnosis and treatment towards “precision and individualilzation”. With further validation and mechanistic exploration, CD4+ Tfh1 cells have the potential to become a cornerstone of DLBCL prognostic stratification and treatment guidance, thus possessing considerable academic value and clinical significance. Conflicts of interest None.