T cell activation mainly requires two signals: the first signal is generated by the binding of MHC-presented antigens to the T cell receptor (TCR), while the second signal involves co-stimulatory and co-inhibitory signaling pathway. However, T cells are not continuously activated, in order to prevent the immune system from over-activating and damaging its own tissues. This regulation is achieved through a series of signaling molecules on the surface of activated T cells, which act as “brakes” on T cell activation [1, 23]. These negative regulatory signaling molecules are known as immune checkpoints, which play a crucial role in maintaining immune homeostasis.

Typical immune checkpoint molecules include PD-1, PD-L1, and CTLA-4. PD-1 is expressed on activated T cells and in non-lymphoid tissues, and thus plays a major role in maintaining peripheral tolerance [24, 25]. Those molecules binding with their ligands inhibit TCR-mediated T cell proliferation and cytokine secretion [26]. PD-L1 is a ligand for PD-1 expressed by tumor cells, stromal cells, and immune cells such as lymphocytes and myeloid cells. The PD-1/PD-L1 pathway regulates immune tolerance within the tumor microenvironment (TME). CTLA-4 is induced upon T cell activation and primarily modulates T cell responses mainly in the early stages within lymphoid organs. For competing with co-stimulatory molecule CD28 for binding to B7 proteins, CTLA-4 prevents excessive immune responses, thereby protecting normal tissues and avoiding autoimmunity [26].

However, many tumor cells also “exhaust” T cells by overexpressing checkpoint molecules that bind to T cell co-inhibitory receptors, thereby limiting the body’s anti-tumor immune response and promoting immune escape. Based on this mechanism, ICIs are developed to block the binding of immune checkpoints to their ligands, relieving the immune suppression of T cells and reactivating the immune cells to exert anti-tumor effects, thus enhancing the body’s anti-tumor immune response [1]. Typical ICIs include monoclonal antibodies against PD-1, PD-L1, and CTLA-4, which have been approved for clinical use in treating various solid tumors such as melanoma [3, 27, 28], NSCLC [2, 29–31], RCC [5, 13, 30, 32], and TNBC [4, 33–35].

It is important to point out that it was previously believed that the immune response induced by ICIs mainly occurs within the tumor, particularly in the TME [36, 37]. However, a growing number of papers suggests that ICI-induced immune responses occur not only in the TME, often referred to as the “battlefield”, but also highlight the essential role of various cell types in TDLNs, which serve as the “training camp” for immune cells [38–41].

According to the cancer-immunity cycle, tumor antigens, which include tumor-associated antigens (TAAs) and tumor-specific antigens (TSAs), released from dying tumor cell death following immunotherapy are captured by APCs. These APCs then present immunogenic tumor antigens to naive T cells (T_naive) in the lymph nodes and make decisions about tumor tolerance or immune activation [42]. Activated effector T cells (T_eff) leave the lymph node and infiltrate the tumor, specifically recognizing and killing cancer cells, thereby releasing additional tumor antigens. These TAAs/TSAs released during tumor cell death, are subsequently captured by APCs in the TME, perpetuating the immune cycle. Figure 1 illustrates a brief process of the cancer-immune cycle.

The efficacy of ICIs hinges on the mobilization of a systemic immune cycle, a coordinated series of immune responses involving both the tumor TME and peripheral immune organs especially in the TDLNs, facilitating the activation, proliferation, and infiltration of T_eff that sustain tumor cell destruction. Therefore, TDLNs have the potential to serve as valuable “windows of observation” for predicting ICIs efficacy.

The lymph node, a highly organized immune organ, is situated at the confluence of the lymphatic system’s blood vessels and can be anatomically subdivided into three primary regions: the B cell zones, T cell zones, and medulla zones [47]. Each zone plays a distinct role in the node’s immune functions.

Removal of lymph nodes, especially TDLNs, has been found to negatively impact ICIs therapeutic efficacy in mouse models. Chamoto et al. [48] demonstrated that deleting of CD8⁺ T cells or ablating of TDLNs in PD-L1 mAb-treated mouse model of colon cancer resulted in impaired tumor growth inhibition by the PD-L1 mAb, suggesting TDLNs play a critical role in the initiation of CD8⁺ T cells and the generation of CTLs during PD-1 blockade therapy. Similarly, Fransen et al. [39] highlighted the significant involvement of TDLNs in the therapeutic efficacy of PD-1 and PD-L1 inhibitors in the MC38 colon cancer mouse model. Immune activation induced by ICIs was primarily observed in TDLNs rather than non-draining lymph nodes, characterized by the local accumulation of CD8⁺ T cells. These studies underscore the importance of TDLNs as sites where APCs interact with T cells to initiate tumor-specific CTL amplification. Given that many ICIs target this interaction between tumor and immune cells, the microenvironment within the TDLNs may significantly influence their efficacy.

T cell immunity is an important part of the body’s immune defense against cancer. Upon antigenic stimulation, the T_naive differentiates into T_mem and T_eff. T_mem serve as a functional reservoir for T_eff, retaining the potential to re-differentiate into T_eff when needed. However, in anti-tumor processes, prolonged antigen exposure promotes terminal differentiation or “exhaustion” of T cells, which hinders the formation of T_mem [49]. These exhausted T cells exhibit sustained upregulation of inhibitory receptors (including PD-1, LAG-3, TIM-3, CTLA-4, etc.), which suppress the TCR signaling pathway through SHP-1/2 phosphatases. This leads to reduced secretion of IFN-γ and tumor necrosis factor-α (TNF-α), ultimately resulting in the loss of cytotoxic T cell tumor-killing capacity and consequently diminishing the efficacy of ICIs.

Notably, there exists a unique subset of CD8⁺ T cells that possess stem-like characteristics, including self-renewal and multilineage differentiation capabilities, encompassing both stem-like exhausted T cells and T_mem. These cells play a pivotal role in antitumor immunity, particularly in the context of ICIs therapy. They are among the cellular populations capable of effectively responding to treatment and mediating sustained antitumor immune responses [50]. Typically, these stem-like CD8⁺ T cells express high levels of the transcription factor T cell factor 1 (TCF1)/TCF7 while maintaining low levels of effector differentiation markers [51]. With higher proliferative potential and stem cell-like developmental plasticity, these stem-like CD8⁺ T cells not only survive for a longer period of time under conditions of sustained antigenic stimulation, but also regulate immune responses by differentiating into effector or depleted T cells [51, 52].

Stem-like CD8⁺ T cells have been found to differentiate into two major classes: CD8⁺ precursor exhausted T cells (T_pex), which maintain the immune function of specific CD8⁺ T cells in a chronic antigen-stimulated environment and are considered to be important target cells for PD-1/PD-L1 blockade therapy [53]; and tumor-specific T_mem, which are usually generated in TDLNs and act as key responders of anti-tumor immunity in ICIs therapy [54].

As mentioned previously, two signals are required for T cell activation in the body: a primary antigen-inducing signal from the TCR and a secondary signal from a co-stimulatory receptor [47]. CD28 is precisely a key T cell co-stimulatory molecule and member of the immunoglobulin superfamily, which is mainly expressed on the surface of T cells, especially on initial T cells and T_mem. It provides the co-stimulatory signals required for T cell activation and proliferation by binding to B7-1 (CD80) and B7-2 (CD86) on APCs, thereby lowering the activation threshold of the TCR complex and promoting T cell survival and effector functions [47, 58]. As a critical co-stimulatory molecule for T cell activation and survival, CD28 is expressed on all T_naive in neonates. However, as T cells undergo prolonged antigen stimulation and differentiation, particularly during chronic viral infections and within the TME, they may lose CD28 expression [59]. This phenotypic change is associated with T cell exhaustion, characterized by high levels of inhibitory receptors such as PD-1 and impaired functionality, with the absence of CD28 exacerbating this process [60]. Furthermore, the co-stimulatory signaling function of CD28 can be competitively inhibited by ligands such as CTLA-4. Due to its structural homology with CD28 and higher binding affinity, CTLA-4 can outcompete CD28 for ligands, thereby suppressing T_eff responses [47]. Consequently, CD28 expression on T cells is not only a key biomarker for assessing T cell exhaustion but also a homologous target of the CTLA-4 immune checkpoint, and its presence is associated with ICIs treatment response.

Kamphorst et al. [61] demonstrated in both mouse models and samples from patients with advanced NSCLC that PD-1-targeted therapy enhances T cell responses, but this rescue requires the involvement of the CD28/B7 co-stimulatory pathway. In mouse models of colorectal cancer and melanoma, they found that the CD28/B7 co-stimulatory pathway is essential for effective PD-1 therapy during tumor burden and chronic viral infections. Conditional gene deletion studies showed that the proliferation of CD8⁺ T cells after PD-1 blockade depends on the presence of CD28, and B7 co-stimulation is also necessary for effective PD-1 therapy. Additionally, their group analyzed blood samples from advanced NSCLC patients and observed that the proliferating CD8⁺ T cells following PD-1 therapy were predominantly CD28-positive. Specifically, in approximately half of the patients, an increase in Ki-67-expressing CD8⁺ T cells was observed after PD-1 treatment, primarily within the PD-1⁺ population, and these PD-1⁺ CD8⁺ T cells were mainly CD28⁺. This suggests that PD-1 therapy induces the selective proliferation of CD8⁺ T cells, highlighting the potential of CD28 as a biomarker for predicting CD8⁺ T cell responses in cancer patients.

Further supporting the importance of CD8⁺ T cells, studies by Duraiswamy et al. [62] found that tumor-specific CD8⁺ T cells accumulate in ovarian cancer niches and are closely associated with intratumoral myeloid APCs. These APCs support tumor-infiltrating T cells and provide CD28 co-stimulation during PD-1 blockade therapy, enabling sustained antitumor immune responses and facilitating the efficacy of PD-1 blockade.

Collectively, these findings underscore the necessity of CD28 co-stimulation for rescuing CD8⁺ T cells [38], highlight the critical role of the CD28/B7 pathway in PD-1 therapy for cancer patients, and suggest that increased expression of CD28 on PD-1⁺ CD8⁺ T cells may indicate a better response to ICIs.

CD4⁺ T cells exhibiting an activated T_H1 phenotype serve as biomarkers that predict response to ICIs therapy. A study showed that the combination of anti-PD-L1 and anti-CTLA-L1 in patients with recurrent or metastatic HNSCC, compared to anti-PD-L1 therapy alone, is more likely to affect tumor progression [66]. The combined therapy effectively activates and expands T_H1-type CD4⁺ cells, which correlated with a favorable response to the combination immunotherapy. The article highlights that pre-existing T_H1 immunity predicts expansion following anti-PD-L1⁺ anti-CTLA-4 treatment. Researchers compared changes in T cells within TDLNs and tumor tissues of patients treated with the combined therapy, and found that TDLNs consisted mainly of initial/central memory CD4⁺ T cells, whereas tumors were dominated by T_eff. Analysis of TCR-seq data from tumors and TDLNs revealed that under dual treatment with anti-PD-L1 and anti-CTLA-4, initial/central memory CD4⁺ T cells at the site of the TDLNs were recruited, and differentiated and expanded under anti-CTLA-4 treatment, thereby transforming into T_eff to achieve tumor cell killing. This indicates that TDLNs are important sites for immune antigen depots and anti-tumor T cell training. Thus, CD4⁺ T cell activation and recruitment from TDLNs are hallmarks of early response to anti-PD-L1 plus anti-CTLA-4 in HNSCC.

In addition, researchers conducted a detailed analysis of patient tumor tissue sections. Spatial single-cell proteomic results showed that expanded CD4⁺ T cells and CD8⁺ T cells exhibited more pronounced co-localized in the tumor tissues of patients receiving combination therapy. Moreover, the expansion of these T cell populations was closely associated with the co-localization of DCs and plasma cells. This suggests that a similar spatial analysis of the lymph nodes could identify immune cells that contribute to the phenotypic changes in how such CD4 T cells become T_H1-type, such as IL-12 production by DCs and macrophages, and IFN-γ production by natural killer (NK) cells and CD4⁺ T_H1 cells themselves, which is essential for CD4⁺ T_H1 cell differentiation. Therefore, it can be assumed that activated CD4⁺ T_H1 cells are not only a biomarker of ICI treatment response, but also a key predictor of the efficacy of combination immunotherapy.

Recently Nature also reported the existence of a stem-like CD4⁺ T cells within TDLNs that are capable of differentiating into CD4⁺ T_H1 cells and promoting effector differentiation of CD8⁺ T cells by secreting IFN-γ, thereby controlling tumor growth and improving response to ICI therapy [67]. This is the first identification of a population of PD1⁺ TCF1⁺ CD4 T cells that reside predominantly in TDLNs and exhibit the ability to self-renew and differentiate into classical CD4⁺ T_eff, and are therefore considered stemness. In the context of T_reg cell depletion, these stem-like CD4⁺ T cells can differentiate into CD4⁺ T_H1 cells and are augmented by the production of IFN-γ to drive the differentiation of stem-like CD8⁺ T cells into effector cytotoxic T cells. Notably, patients with RCC with a strong CD4⁺ T_H1 response have been associated with a higher frequency of cytotoxic effector CD8⁺ T cells after ICI treatment and an improved progression-free survival. Thus, differentiation of PD1⁺ stem-like CD4⁺ T cell cells into CD4⁺ T_H1 cells is not only critical for orchestrating effective antitumor immunity, but also predicts patients who are more likely to respond well to ICIs therapy.

T_reg are a heterogeneous population of CD4⁺ T lymphocytes with immunosuppressive functions, which play a key role in maintaining immune tolerance, preventing autoimmune diseases, and modulating immune responses. T_reg exert immunosuppressive functions through multiple mechanisms, such as secreting immunosuppressive cytokines [e.g., IL-10, transforming growth factor-β (TGF-β)], depleting key growth factors such as IL-2, and generating immunosuppressive cytokines through cell contact, as well as inhibiting the activation and proliferation of T_eff via cell contact. In tumor immunity, T_reg are regarded as key participants in immune escape within the TMEs and TDLNs, providing support for tumor immune privilege by suppressing anti-tumor effector immune responses. This makes them potential negative predictive markers in ICIs therapy.

TDLNs serve as the primary site for the induction and activation of de novo tumor-specific T_reg, which maintain the immune tolerance within the TDLNs by inhibiting the function of T_eff. This phenomenon mirrors the immunosuppression observed in the TME, allowing sustained tumor growth despite high levels of circulating tumor-specific T cells [68]. Huang et al. [69] found that T_reg in the TDLNs secreted TGF-β, which significantly inhibited the cytotoxic activity of tumor-specific CD8⁺ T cells, thereby promoting tumor growth in mouse tumor models. The presence of T_reg also affects the differentiation of CD4⁺ T cells. Alonso et al. [66] reported in a study of lung adenocarcinoma mouse model that CD4⁺ T cell differentiation in TDLNs favors the generation of T_reg over effector CD4⁺ T cells. In 2024, Cardenas et al. [67] published an article in Nature that showed that the differentiation of stem-like CD4⁺ T cells in TDLNs was a major factor in the growth of tumor cells, with T_reg in TDLNs being more prone to producing TGF-β compared to effector CD4⁺ T cells.

In addition to inhibiting CD8⁺ T cell function or affecting tumor antigen-specific CD4⁺ T cell differentiation, T_reg also impact the overall function of the cancer-immunity cycle. Through surface expression of CD39 and CD73, T_reg catalyze the degradation of ATP into adenosine, which subsequently downregulates the expression of adhesion molecules (such as ICAM-1 and VCAM-1) on tumor vascular endothelial cells via A2A adenosine receptor signaling, thereby reducing immune cell infiltration into tumors [70]. Additionally, peripheral blood T_reg can suppress systemic antitumor immunity through both adenosine-dependent and independent mechanisms, and their abundance is negatively correlated with the efficacy of ICIs [71].

Besides, T_reg can shape the immune-tolerant environment of lymph nodes by modulating PD-L1 expression and thus affect ICIs efficacy. A mouse model of lymph node metastasis melanoma constructed by Reticker-Flynn et al. [72] revealed a critical role for lymph nodes in tumor immune escape and distant metastasis. The lymph nodes formed an immunosuppressive microenvironment by upregulating the expression of MHC-I and PD-L1, which not only facilitated tumor cells’ evasion from NK cell mediated killing, but also suppressed T cell activity. This suppression generated tumor-specific immune tolerance, creating favorable conditions for distant tumor colonization and growth. The “Metastatic Tolerance” model proposed in the article emphasizes that lymph node metastasis not merely serves as a transit point for tumor cells to spread to distant sites, but rather induces tumor-specific immune tolerance, making distant tissues more receptive to tumor cell colonization. In this process, T_reg play a key role. Besides secreting immunosuppressive cytokines such as TGF-β, T_reg are closely associated with PD-L1 expression. Despite the possibility of a false negative, PD-L1 expression remains an important predictor of response to ICIs therapy, it can be hypothesized that T_reg within lymph nodes and the immune tolerance microenvironment they induce play a crucial role in predicting and modulating the response to ICIs therapy.

DCs are able to acquire tumor antigens from the TME and activate CD8⁺ T cells and CD4⁺ T cells and maintain antitumor immune responses by cross-presentation via MHC-I molecules or direct antigen presentation via MHC-II molecules. As the most important dedicated APCs, functionally active DCs of different subtypes play an important role in the immunotherapeutic response, which is mainly related to the crosstalk or communication between DCs and antitumor T cells [93]. Preclinical data suggest that the antitumor effect of anti-PD-1 antibodies requires the interaction between DCs (especially cDC1) and T cells for “authorization”. Following PD-1 inhibition, activated CD8⁺ T cells secrete IFN-γ, which activates DCs, particularly cDC1, to produce IL-12. This cDC-secreted IL-12 further stimulates cytotoxic CD8⁺ T cells and enhances the recovery of T cell dysfunction mediated by anti-PD-(L)1 antibodies [92].

Studies have shown that the functional status of DCs is closely related to the efficacy of ICIs. For example, in human HNSCC patients treated with anti-PD-L1 showed a closer spatial localization of T_pex and DCs in uiLNs [57], which further validates the importance of DCs and antitumor T cells interactions in the treatment of ICIs; and in melanoma, activation of the β-catenin signaling pathway inhibits CD103⁺ DCs and T cells from entering the tumor, but by injecting bone marrow-derived DCs into the tumor site, T cell recruitment can be promoted, making the tumor more sensitive to PD-L1 and CTLA-4 monotherapy [94, 95]. Meanwhile, activation of TLR4 on CD103⁺ DCs enhances CD8⁺ T cell infiltration and significantly improves the efficacy of ICIs [95, 96]. The CCR7⁺ DC subpopulation also plays an important role in antitumor immunity, with mature DCs signaling via the chemokine CCR7 migrating from the TME to TDLNs, where they initially activate and expand tumor-specific T cells. This CCR7⁺ DC subpopulation exhibits high levels of PD-L1 expression and plays a key role in interaction with CD8⁺ T cells.

Macrophages play multiple roles in the regulation of anti-tumor immune responses, with subpopulations that promote anti-tumor immunity and also exert immunosuppressive functions after being affected by the TME. Macrophages can be categorized into subcapsular sinus (SCS) macrophages and medullary sinus macrophages based on their anatomical location in the lymph node [97, 98]. Clinical studies have shown that the density of CD169⁺ SCS macrophages in TDLNs is significantly associated with a favorable patient prognosis. These macrophages function as APCs, activating T cell responses in lymph nodes and limiting the spread of tumor cells by capturing tumor-derived antigens or metastatic tumor cells [99]. In addition, CD169⁺ SCS macrophages are able to capture subcutaneously injected apoptotic tumor cells and activate CD8⁺ T cells through antigen presentation, thereby enhancing antitumor immunity.

In addition to acting as APCs, macrophages can also participate in the cytotoxicity and clearance of target cells through activating Fcγ receptors (FcγRI, FcγRIIA, FcγRIIC, and FcγRIIIA) on their surfaces, which are particularly prominent in monoclonal antibody therapy [95]. Studies have shown that ipilimumab, an antibody against CTLA-4, is able to reduce the number of T_reg through FcγR-mediated effects, thereby enhancing the activity of anti-tumor T cells [1, 100]. In addition, melanoma patients who responded to ipilimumab had a higher proportion of FcγRIIIA⁺ CD16⁺ monocytes, suggesting that activated Fcγ receptors may serve as a potential predictor of response to therapy.

On the other hand, macrophages may also limit antitumor immune responses in some cases, and tumor-associated macrophages are more likely to antagonize immune checkpoint therapy than tumor-infiltrating DCs. For example, the inhibitory Fcγ receptor (FcγRIIB) regulates myeloid cell maturation by cross-linking with agonistic CD40 mAb in a mouse tumor model, thereby indirectly affecting T cell activation [95]. In some tumor settings, macrophage-secreted IL-10 inhibits IL-12 expression by DC, thereby impairing CD8⁺ T cell-mediated antitumor immunity.

Although the direct correlation between macrophages and ICIs therapy in TDLNs has yet to be thoroughly investigated, available studies suggest that macrophages in TDLNs have the potential to serve as predictive markers of response to ICIs therapy. This dual role of macrophages underscores the need to precisely modulate macrophage function to maximize its role in promoting anti-tumor immune responses when developing new immunotherapeutic strategies. Figure 3 summarizes the immune cell characteristics within TDLNs as predictive features for the response to ICIs.

The methods to study lymph nodes include surgical staining to observe tumor metastasis, hematoxylin and eosin (H&E) staining of pathological sections, immunohistochemistry, single-cell sequencing, imaging scanning, metabolomics study, and spatial multi-omics study with tissue morphology preservation. These methods enable us to identify the characteristics of lymph nodes from different perspectives and provide multi-dimensional information for the prediction of ICIs efficacy. Despite these advancements, the approach also has certain limitations. For instance, there are notable discrepancies in consistency between various detection methods and clinical presentations, and conducting a comprehensive analysis of these differences poses a significant challenge. Additionally, single-cell sequencing is not only relatively costly but also involves complex sample processing procedures. Spatial multi-omics analysis similarly faces challenges in data integration and interpretation. In this context, comprehensive analyses of human TDLNs biological samples based on histological or clinical parameters using a variety of methodologies will be of paramount importance.

Retrospective clinical studies of patients with melanoma, thyroid cancer, and breast cancer have found that prophylactic lymph node removal does not improve survival, and the research featured in Xue et al. [107] has shown that surgical removal of TDLNs before tumor implantation and/or anti-PD-1 therapy could reduce treatment efficacy. This supports a central role for TDLNs in the induction of CTL-mediated antitumor cell killing after PD-1 blockade. Therefore, in practice, we prefer to evaluate the status of single lymph nodes to predict the efficacy of ICIs, rather than perform extensive lymph node dissection to reduce the risk of trauma and complications for patients. This selective lymph node evaluation may provide a basis for individualized prediction of the efficacy of ICIs.

In order to predict the efficacy of ICIs, it is important to determine the best timing of lymph node detection, which may include before ICIs treatment, during a specific course of treatment after ICIs treatment, or when sentinel lymph node surgery is performed after immunotherapy. The selection of these timing is of great significance for evaluating the efficacy of ICIs and guiding subsequent treatment strategies. If TDLNs remain in situ prior to ICIs treatment, pre-treatment biopsies or systemic immune population sampling of TDLNs may become important predictive biomarkers. Post-treatment evaluation of resected TDLN may help prognosis or determine whether further adjuvant therapy is needed. Ongoing clinical trials of neoadjuvant ICIs may help define the most effective timing of therapy.

There are two main methods for lymph node detection: non-invasive and invasive. Surgical resection of the whole lymph node or needle biopsy is widely used in clinical practice. Utilizing non-invasive detection techniques for sampling TDLNs together with more clinical information may help us delineate the relationship between TDLNs immune environment and ICIs efficacy. Evaluation of TDLNs in patients with NSCLC by fine-needle aspiration can determine the composition of immune cells [108], which may help predict the response to ICIs.

TDLNs of different tumor types exhibit distinct immune characteristics, which implies that predictive markers effective in one tumor type may not be applicable to others. This increases the complexity of research and application, necessitating more in-depth and specific studies for different tumor types.