A comprehensive review published in Biomarker Research proposes that tumor-draining lymph nodes (TDLNs) represent an underutilized source of biomarkers for predicting responses to immunotherapy in breast cancer. The study, authored by Zheng Li, Shengbin Pei, Zheng Qu, and colleagues, argues that current biomarkers derived from primary tumor tissue or peripheral blood fail to capture the adaptive immune response at its site of initiation. By examining the intersection of TDLN immunobiology, high-resolution multi-omics, and artificial intelligence, the review outlines a roadmap for developing next-generation biomarkers that could complement existing metrics such as PD-L1 immunohistochemistry and tumor mutational burden.
The authors emphasize that TDLNs are not merely passive conduits for metastatic dissemination but dynamic immunological hubs where anti-tumor T cell priming and immune evasion converge. Anatomically, these nodes are the first lymphoid organs to receive tumor-derived antigens, exosomes, and soluble mediators via afferent lymphatic vessels. Immunologically, they serve as primary sites for dendritic cell-mediated antigen cross-presentation to naive CD8+ T cells, clonal expansion of tumor-specific cytotoxic T lymphocytes, and the orchestration of humoral immunity through B cell activation and germinal center formation. The functional state of the TDLN, balanced between activation and suppression, may therefore influence the efficacy of systemic immunotherapy, a hypothesis supported by preclinical and observational studies but not yet prospectively validated in immunotherapy-treated breast cancer cohorts.
Current biomarkers for immune checkpoint inhibitor therapy, including PD-L1, tumor mutational burden, and microsatellite instability, share a fundamental conceptual limitation: they do not interrogate the initiation node of adaptive anti-tumor immunity. PD-L1 immunohistochemistry suffers from substantial inter-assay variability across antibody clones and spatial and temporal heterogeneity within and between lesions. Tumor mutational burden, while approved as a tumor-agnostic biomarker, demonstrates limited predictive utility in breast cancer due to the relatively modest mutational burden of breast tumors and the inability of TMB to resolve the immunogenicity or clonal architecture of the neoantigen landscape. Gene expression signatures such as the T-cell inflamed gene expression profile have shown promise in exploratory analyses but have not been prospectively validated as definitive predictive biomarkers across breast cancer subtypes.
The review details the complex cellular and molecular architecture of the breast cancer TDLN, highlighting T cell exhaustion, B cell-mediated immunity, and stromal crosstalk. In tumor-bearing hosts, the priming process is frequently suboptimal, with conventional type 1 dendritic cells often displaying an immature or tolerogenic phenotype characterized by reduced expression of co-stimulatory molecules and increased expression of inhibitory ligands such as PD-L1. This results in the generation of functionally impaired, anergic CD8+ T cells, a phenomenon of peripheral tolerance co-opted by tumors for immune evasion. The balance between progenitor exhausted T cells and terminally exhausted T cells within TDLNs is hypothesized to be an important factor influencing the magnitude and durability of the response to immune checkpoint inhibitors, based primarily on pan-cancer analyses and preclinical models.
High-throughput multi-omics platforms, including single-cell RNA sequencing, spatial transcriptomics, and single-cell T cell receptor and B cell receptor sequencing, have matured to enable comprehensive molecular profiling of clinical tissue specimens at high resolution. These technologies permit the deconvolution of cellular heterogeneity that is obscured in bulk transcriptomic analyses. For instance, single-cell transcriptomic analyses have revealed heterogeneity within the fibroblastic reticular cell population, with distinct subtypes occupying specific anatomical niches and expressing unique repertoires of chemokines, cytokines, and extracellular matrix components that collectively govern the spatial organization of immune cells. Spatial transcriptomics bridges the gap between single-cell resolution and tissue context by mapping gene expression within intact tissue architecture, enabling the identification of spatially proximal cell type pairs and inference of ligand-receptor interactions within defined spatial neighborhoods.
Artificial intelligence and deep learning methodologies, including graph neural networks, vision transformers, and large-scale biological foundation models, have demonstrated remarkable capacity for integrating heterogeneous, high-dimensional data modalities into clinically actionable predictive models. However, the application of these tools to TDLN specimens for immunotherapy biomarker discovery remains in its early stages. A major methodological gap exists: while AI frameworks are increasingly mature for primary tumors, their application to TDLNs remains unvalidated and limited by a severe scarcity of annotated multi-omics datasets. Tumor-trained models are optimized to recognize malignant epithelium and its immune contexture, whereas TDLNs are architecturally organized secondary lymphoid organs whose biologically meaningful features, such as follicles and germinal centers, are largely absent from tumor-trained representations. This domain shift poses a significant challenge for the direct transfer of primary-tumor AI approaches to TDLN datasets.
The authors propose a four-type biomarker taxonomy to organize candidate TDLN-derived features: cell-compositional, functional-state, spatial-architectural, and omics-signature. Cell-compositional biomarkers involve immune cell type proportions and ratios, such as the CD8+ T cell to regulatory T cell ratio, which has been associated with clinical outcomes in several solid tumor types. Functional-state biomarkers include molecular readouts of activation, differentiation, or exhaustion, such as transcriptomic signatures of T cell exhaustion versus T cell stemness. Spatial-architectural biomarkers capture spatially resolved features of tissue organization, such as the density and maturation status of germinal centers. Omics-signature biomarkers are high-dimensional signatures derived from multi-omics data through AI-driven analysis. These categories are not mutually exclusive but represent progressively richer levels of biological information, from local cellular abundance to global molecular architecture.
From a clinical perspective, TDLNs possess a practical advantage that distinguishes them from other potential biomarker sources: sentinel lymph node biopsy is a standard-of-care staging procedure in early breast cancer, generating routinely collected tissue specimens that are amenable to advanced molecular profiling. The progressive de-escalation of axillary surgery over the past three decades, driven by trials establishing the oncological safety of sentinel lymph node biopsy over axillary lymph node dissection, has made sentinel lymph node tissue widely available as a research resource while reducing surgical morbidity. Archival formalin-fixed, paraffin-embedded sentinel lymph node specimens in pathology repositories represent a largely untapped resource for biomarker discovery. However, the availability of sentinel lymph node tissue varies by disease stage, molecular subtype, and neoadjuvant treatment pathway, and recent trials support omitting sentinel lymph node biopsy entirely in selected low-risk, clinically node-negative patients.
The review addresses pivotal translational considerations, including the immunological implications of axillary surgery in the immune checkpoint inhibitor era. Preclinical studies in murine models have demonstrated that complete TDLN resection impairs the efficacy of PD-1/PD-L1 blockade, suggesting that TDLNs are functional participants in the therapeutic mechanism of immune checkpoint inhibitors. However, substantial considerations temper a direct clinical translation of these findings. Preclinical models involve complete TDLN resection, whereas sentinel lymph node biopsy removes only one to three nodes from a larger axillary basin. The choice between sentinel lymph node biopsy and axillary lymph node dissection is strongly confounded by baseline nodal tumor burden, treatment response, and established surgical indications. No dedicated clinical evidence from immune checkpoint inhibitor-treated breast cancer cohorts currently exists to support a causal relationship between axillary surgery extent and immunotherapy efficacy, and this hypothesis is not intended to influence current surgical decision-making.
The authors emphasize that no TDLN-derived biomarker has yet undergone prospective validation as a predictive biomarker for immunotherapy in breast cancer. Retrospective analyses of trial specimens must account for the timing of tissue acquisition relative to systemic therapy, as systemic therapy remodels the nodal immune landscape. The review charts immediate research priorities, including establishing TDLN reference atlases and developing clinically feasible specimen collection pathways for biomarker validation. A pragmatic TDLN biomarker development strategy requires distinct specimen pathways for baseline prediction, pharmacodynamic assessment, post-surgical prognosis, and longitudinal monitoring. The review provides a conceptual framework and translational roadmap to assess whether TDLN-derived features can complement existing tumor-centric metrics in guiding immunotherapy decisions, highlighting the need for rigorous prospective studies, methodological standardization, and multidisciplinary collaboration.
Subject of Research: Tumor-draining lymph nodes as candidate sources for immunotherapy biomarkers in breast cancer
Article Title: Tumor-draining lymph nodes as candidate sources of immunotherapy biomarkers in breast cancer: a multi-omics and artificial intelligence roadmap
Article References: Li, Z., Pei, S., Qu, Z., Zhang, L., Lu, J., Xian, K., Wang, J., & Fang, Y. (2026). Tumor-draining lymph nodes as candidate sources of immunotherapy biomarkers in breast cancer: a multi-omics and artificial intelligence roadmap. Biomarker Research, 14(1), Article 168. https://doi.org/10.1186/s40364-026-01047-2
Image Credits: AI Generated
DOI: 10.1186/s40364-026-01047-2
Keywords: breast cancer, immunotherapy, biomarkers, tumor-draining lymph nodes, multi-omics, artificial intelligence, single-cell RNA sequencing, spatial transcriptomics, immune checkpoint inhibitors, T cell exhaustion, sentinel lymph node biopsy, axillary surgery, B cell immunity, dendritic cells, clinical translation
News Source: Nathaniel Bowman. (November 18, 2026). Lymph Nodes May Reveal Immune Therapy Biomarkers. Scienmag.



