T Cell Exhaustion in Cancer Immunotherapy: Heterogeneity, Mechanisms, and Therapeutic Opportunities.

Introduction

The rapid advancement of cancer immunology has spurred the emergence of multiple breakthrough therapeutic strategies, with immune checkpoint inhibitors (ICIs) and adoptive cell therapy (ACT) being representative examples. These strategies have significantly enhanced the survival prognosis of patients with various advanced cancers. Therapies targeting immune checkpoints, such as programmed death protein 1 (PD‐1), programmed death‐ligand 1 (PD‐L1), and cytotoxic T lymphocyte‐associated antigen 4 (CTLA‐4), have exhibited durable clinical responses in a wide range of solid tumors and hematological malignancies.

In some cases, patients have even achieved long‐term remission [1, 2]. Nonetheless, a substantial proportion of patients do not respond to treatment or ultimately develop resistance. This clinical challenge is, to a large extent, attributed to the dysfunctional state of T cells within the tumor microenvironment (TME), especially the critical state of T cell exhaustion [3, 4].

T cell exhaustion is a cellular state that occurs under the backdrop of persistent antigenic stimulation, such as during chronic infections or tumor progression. In this state, T cells gradually lose their effector functions, experience a marked decline in proliferation ability, and are accompanied by the sustained high expression of multiple inhibitory receptors, including PD‐1, lymphocyte‐activation gene 3 (LAG‐3), and T cell immunoglobulin and mucin‐domain containing‐3 (TIM‐3). This state not only impairs the tumor clearance ability mediated by T cells but also obstructs the formation of immune memory, thus significantly affecting the long‐term efficacy of immunotherapy [5].

Significantly, exhaustion does not signify cellular “ineptitude”. Instead, it represents a precisely regulated adaptive state. Underlying this state are unique transcriptomic, epigenetic, and metabolic reprogramming processes.

These processes can maintain a certain level of pathogen or tumor control while simultaneously limiting immunopathology [6].

In recent years, the remarkable progress in single‐cell multi‐omics technologies and epigenetic analysis methods has significantly enhanced our comprehension of the biological nature of T cell exhaustion. Studies have clearly shown that exhausted T (TEX) cells exhibit a high degree of heterogeneity. They consist of distinct subpopulations, including precursor exhausted T (TPEX) cells, intermediate exhausted T (TEX‐int) cells, and terminally differentiated exhausted T (TEX‐term) cells, which form a well – defined and continuous differentiation pathway.

Among these, TPEX characterized by the expression of T cell factor 1 (TCF‐1), possess self‐renewal capacity and are sensitive to immune checkpoint blockade (ICB) therapy. In contrast, terminally exhausted cells display irreversible functional impairment and epigenetic “entrapment” [6, 7]. Crucial transcription factors such as thymocyte selection‐associated high mobility group box protein (TOX), TCF‐1, and nuclear receptor subfamily 4, group A (NR4A) have been identified as the central regulators of the exhaustion program.

Through hierarchical epigenetic remodeling and metabolic – epigenetic cross – regulation, they jointly determine the fate of T cell exhaustion [8, 9, 10, 11]. Moreover, metabolic anomalies, such as impaired glycolysis/oxidative phosphorylation (OXPHOS) and mitochondrial dysfunction, along with immunosuppressive cytokines in the TME, such as Interleukin (IL)‐10 and transforming growth factor‐β (TGF‐β), have also been recognized as key determinants in maintaining and stabilizing the exhausted state [12, 13, 14, 15, 16]. Recent research has also disclosed that persistent antigenic signals, by driving chronic AKT activation, initiate a distinctive proteotoxic stress response (PSR).

This response is characterized by an increase in global protein synthesis, a specific upregulation of the endoplasmic reticulum stress chaperone proteins BiP/gp96, and the accumulation of protein aggregates. These events collectively disrupt proteostasis, directly leading to the functional exhaustion of T cells [17]. Together, these mechanisms form a multi‐tiered regulatory network that drives the progressive loss of T cell function, culminating in an irreversible exhausted state under continuous antigen exposure.

Building upon these insights into the underlying mechanisms, novel therapeutic strategies are advancing at a rapid pace, transcending the traditional single ICB paradigm and moving towards a paradigm of combinatorial interventions and precise regulation. Specifically, the concurrent application of histone deacetylase inhibitors (HDACi) or DNA methyltransferase inhibitors (DNMTi) in conjunction with ICB holds the potential to reverse the epigenetic silencing of T cells, thereby rekindling their effector functions. This approach aims to modulate the epigenetic landscape of T cells, unlocking gene expression programs that were previously repressed and restoring their ability to mount an effective anti – tumor immune response [11, 18, 19].

Furthermore, targeting immune metabolic pathways or obstructing inhibitory signals, including adenosine, indoleamine 2,3‐dioxygenase (IDO), and TGF‐β, has demonstrated promising synergistic antitumor capabilities in both pre‐clinical investigations and clinical trials [5, 16, 20, 21, 22]. In the realm of cell therapy, gene editing technologies such as CRISPR‐Cas9 can effectively knock out molecules associated with T cell exhaustion or enhance the expression of functional promoters. These interventions have the potential to significantly improve the persistence, stem cell‐like characteristics, and anti‐tumor efficacy of chimeric antigen receptor T cell (CAR‐T), T cell receptor‐engineered T (TCR‐T), and tumor‐infiltrating lymphocytes (TILs) [4, 23, 24].

Furthermore, oncolytic viruses (OVs), bispecific antibodies (BsAbs), antibody drug conjugates (ADCs), and costimulatory agonists offer innovative strategies for reversing immune exhaustion and overcoming immunosuppression through various mechanisms [23, 25, 26].

This article systematically reviews the latest progress in the field of T cell exhaustion. Starting from its definition and core characteristics, the key molecular mechanisms regulating the exhaustion state, the heterogeneity of cell subsets, and new biomarkers are discussed. On this basis, we will focus on innovative therapeutic strategies to target T cell exhaustion, including the optimized application of existing ICIs, the development of combination therapy strategies, and the improvement of the next generation of cell therapy technology, in order to provide in‐depth theoretical insights and feasible translational directions for overcoming current immunotherapy resistance.

The Biological Basis of T Cell Exhaustion

Discovery and Definition of T Cell Exhaustion

T cell exhaustion refers to a critical cellular state characterized by the progressive loss of effector function in T cells due to persistent antigen exposure during chronic infections and within the tumor microenvironment. The conceptual foundation of this phenomenon originated from studies of a mouse model of chronic lymphocytic choriomeningitis virus (LCMV) infection conducted in the 1990s [27]. It was observed that, in contrast to the functional effector and memory T cells generated during acute infection, virus‐specific CD8+ T cells in chronically infected mice exhibited a gradual decline in functionality.

These cells demonstrated significantly impaired proliferative capacity, reduced secretion of inflammatory cytokines, and diminished cytotoxic activity, ultimately resulting in a failure to clear the virus and increased susceptibility to apoptosis. This seminal study not only provided the first direct experimental evidence of T cell exhaustion but also established a classical animal model that remains widely used in immunological research today [28].

Since then, T cell dysfunction phenotypes, closely resembling those observed in the LCMV model, have been identified in patients with chronic viral infections, including human immunodeficiency virus (HIV), hepatitis C virus (HCV), and hepatitis B virus (HBV), as well as in TILs, thereby confirming the broad applicability and cross‐disease relevance of the TEX cell state [29, 30, 31, 32, 33, 34]. Notably, these TEX cells consistently exhibit elevated expression of multiple inhibitory receptors, with PD‐1 recognized as one of their most characteristic molecular markers [30, 35, 36]. These findings represent a pivotal transition in T cell exhaustion research—from experimental models to human disease contexts—and have established critical molecular targets for the development of ICB therapeutic interventions.

An important turning point in the field emerged from the systematic analysis of LCMV models conducted by the Wherry team. For the first time, they established a clear conceptual framework for T cell exhaustion and elucidated its core biological characteristics [37]. TEX cells are defined as a distinct differentiation state characterized by progressive loss of effector function, co‐expression of multiple inhibitory receptors, a unique transcriptomic and epigenetic signature, and an inability to develop into functional memory cells under persistent antigenic and inflammatory stimulation [36, 38, 39, 40].

The dysfunction of these cells follows a hierarchical pattern: diminished proliferative capacity and loss of IL‐2 production occur early, followed by reduced secretion of tumor necrosis factor‐alpha (TNF‐α) and interferon‐γ (IFN‐γ) [41, 42]. Mechanistically, TEX cells not only exhibit high expression of inhibitory receptors such as PD‐1, TIM‐3, LAG‐3, and CTLA‐4 [36, 43, 44], but also display a gene expression profile distinct from that of effector and memory T cells, as well as substantial alterations in chromatin accessibility. These epigenetic modifications are considered a fundamental mechanism underlying the irreversibility of the exhausted state [5].

This study not only formally established T cell exhaustion as a unique T cell differentiation lineage, but also provided a critical theoretical foundation for the development of ICB therapies.

The clinical success of ICIs, such as those targeting PD‐1 and CTLA‐4, has established the reversal of T cell exhaustion as a viable therapeutic strategy, catalyzing a transformative shift in cancer treatment. For their pioneering contributions to the discovery of immune checkpoint regulation, James P. Allison and Tasuku Honjo were awarded the 2018 Nobel Prize in Physiology or Medicine [45, 46].

In recent years, extensive research has further demonstrated that the gene expression profile of TEX cells is stabilized by durable epigenetic modifications, enabling their dysfunctional state to persist for prolonged periods even in the absence of antigenic stimulation. This mechanism elucidates the limitations of current ICB monotherapies in fully restoring the functionality of TEX‐term cells [47], while also identifying molecular‐level regulators—acting as both “on‐switches” and “state stabilizers”—of the exhaustion program. These insights provide a critical foundation for the development of next‐generation immunotherapeutic approaches that extend beyond conventional ICB strategies (Figure 1).

Multifactorial Mechanisms Underlying T Cell Exhaustion

The development of T cell exhaustion is a complex and dynamic process mediated by multiple signals within the TME. This phenomenon arises from multilayered and interconnected biological mechanisms, including persistent antigen stimulation, inhibitory receptor signaling, transcriptional and epigenetic reprogramming, metabolic dysregulation, and an immunosuppressive milieu. Collectively, these factors establish a positive feedback loop that progressively drives T cells from functional impairment to terminal exhaustion (Figure 2).

Persistent Antigen and Inhibitory Receptor Signaling: Initiation and Reinforcement of Exhaustion

Continuous antigenic stimulation serves as the initiating trigger for the T cell exhaustion program. In contrast to the transient stimulation observed during acute infections, persistent antigen exposure in the context of tumors leads to prolonged, low‐intensity activation of T cell receptor (TCR) signaling [48]. This abnormal TCR signaling not only directly leads to the loss of T cell effector function but also induces the expression of inhibitory receptors by activating key transcription factors such as nuclear factor of activated T cells (NFAT) and TOX, and triggers global transcriptional and epigenetic reprogramming.

See Section 2.2.2 for details. These changes persist even after antigen removal, establishing a durable “exhaustion signature” that is resistant to reversal and impedes the complete restoration of T‐cell function [49]. In addition, persistent TCR signaling acts synergistically with metabolic stresses in the tumor microenvironment—such as hypoxia and glucose deprivation—to rapidly induce mitochondrial dysfunction, impaired glycolysis, and reactive oxygen species (ROS) accumulation, thereby further promoting stabilization of the TEX cell phenotype [14].

Metabolic disturbances not only restrict the energy supply to T cells but also impair the activities of histone‐modifying enzymes and DNA methyltransferases, contributing to the epigenetic stabilization of exhaustion‐associated genes [12]. For further details, refer to Section 2.2.3.

Inhibitory receptor networks play a central role in establishing and stabilizing T cell exhaustion. In normal immune responses, molecules like PD‐1, CTLA‐4, TIM‐3, LAG‐3, and T cell immunoreceptor with Ig and ITIM domains (TIGIT) are transiently expressed to maintain immune homeostasis [50, 51, 52, 53, 54, 55, 56]. However, under persistent antigen exposure in the TME, these receptors become stably and highly expressed.

They form a synergistic inhibitory network that suppresses T cell function through multiple mechanisms. At the juxtamembrane level, they directly interfere with TCR signaling, disrupt the formation of immunological synapses, and attenuate the activation of key downstream kinases such as AKT and ERK [57, 58]. Moreover, the intracellular domains of these receptors contain immune receptor tyrosine‐based inhibitory motifs (ITIMs) or immune receptor tyrosine‐based switching motifs (ITSMs), which recruit phosphatases such as src homology region 2 domain‐containing phosphatase (SHP)‐1 and SHP‐2.

These phosphatases subsequently dephosphorylate critical signaling molecules in the TCR and CD28 pathways, thereby dampening activation signals at their origin [59, 60, 61]. Additionally, this inhibitory network can exert direct effects in the nucleus by modulating gene expression programs, thereby promoting the establishment and stabilization of exhaustion‐associated transcriptional profiles [62, 63]. Sustained overactivation of this network collectively results in the loss of T cell cytotoxicity, impaired cytokine production, and complete abrogation of proliferative capacity [64, 65].

Transcription and Epigenetic Reprogramming: Solidification of the Exhausted State

T cell exhaustion represents a progressive state driven by an aberrant transcriptional program and maintained through sustained epigenetic remodeling [66]. In the absence of costimulatory signals such as CD28, persistent antigenic stimulation leads to the activation of the transcription factor NFAT in a “chaperonless” conformation [67]. Without its canonical partner activator protein 1 (AP‐1), NFAT preferentially initiates a transcriptional program associated with T cell exhaustion and upregulates key transcription factors, including TOX/TOX2 and NR4A [68, 69].

These factors not only execute the exhaustion program but also function as “molecular architects” by orchestrating epigenetic changes. For instance, TOX modulates chromatin architecture by directing the recruitment of chromatin‐modifying complexes and sustaining the accessibility of gene loci encoding inhibitory receptors such as PD‐1, thereby reinforcing their sustained expression [70, 71].

At the epigenetic level, TEX cells exhibit a chromatin landscape that is distinctly different from that of effector and memory T cells. Central to this distinction is the establishment of an “epigenetic scar” which serves as the core molecular mechanism underlying the irreversible nature of T cell exhaustion [72]. This remodeling process originates with widespread alterations in chromatin accessibility.

Assay for transposase‐accessible chromatin using sequencing (ATAC‐seq) has revealed that open chromatin regions are highly enriched at gene loci associated with exhaustion markers such as PD‐1 and TIM‐3, whereas genes related to effector and memory functions display reduced chromatin accessibility [73, 74]. This differential accessibility pattern lays the foundation for subsequent stabilization through deeper epigenetic modifications. Concurrently, DNMT3A‐mediated de novo DNA methylation emerges as an early initiating event in the exhaustion program.

It targets promoters of key effector and memory‐associated genes, including Tcf7 and Ifng, leading to their transcriptional silencing and stable inheritance across cell divisions [19, 75, 76]. The progressive accumulation and consolidation of these methylation events ultimately give rise to a durable and largely irreversible epigenetic methylation signature. This provides a mechanistic explanation for the limited efficacy of ICIs in fully restoring the functional capacity of terminally TEX cells [11].

Subsequently, DNA methylation and histone modifications establish a tightly coordinated, self‐reinforcing synergistic network. Notably, repressive histone marks such as H3K27me3 can actively recruit the DNA methyltransferase DNMT3A to specific genomic loci, thereby facilitating the establishment of stable DNA methylation at these sites and enabling deep silencing of effector and memory‐associated genes [77, 78]. Conversely, DNA methylation can recruit methyl‐CpG‐binding domain proteins (MBDs), which in turn facilitate the recruitment of histone deacetylases (HDACs) and other repressive complexes, further reinforcing local transcriptional repression and forming a self‐sustaining feedback loop [79].

Importantly, this highly cooperative silencing network also harbors intrinsic counteracting mechanisms, such as the histone demethylase KDM6B, which may partially preserve the plasticity of TPEX cells [80, 81]. However, in TEX‐term cells, the aforementioned cooperative silencing mechanism becomes overwhelmingly dominant. This imbalance, combined with a state of “transcriptional arrest” imposed by bivalent chromatin domains marked by both H3K4me3 and H3K27me3, contributes to locking cells in a terminal state of severe functional impairment [80].

Critically, chromatin remodeling complexes such as mSWI/SNF act as ultimate executors of epigenetic information by regulating the three‐dimensional architecture of the genome and stabilizing covalent modifications within specific spatial conformations. This directly leads to the widespread phenomenon of “active mark–gene expression uncoupling” in TEX‐term cells. Specifically, epigenomic analyses including ChIP‐seq and CUT&RUN have revealed that although many genomic regions retain active enhancer marks such as H3K27ac, the physical constraints imposed by closed chromatin conformation and the absence of essential transcription factors prevent effective activation of the associated gene expression programs [82, 83].

More importantly, a profound crosstalk exists between metabolic perturbations and epigenetic reprogramming within the tumor microenvironment, further exacerbating the stabilization of the TEX cell phenotype. Metabolites, serving as essential substrates or regulators of epigenetic modifications, directly influence chromatin states [84]. For instance, recent studies have demonstrated that histone lactoylation, including H3K18la and H3K9la, is highly enriched in effector T cells and functions as a critical link between metabolism and epigenetics, coinciding with active expression of cytotoxic and effector function‐related genes [85, 86, 87].

In contrast, in TEX‐term cells, levels of lactoylation are markedly reduced due to metabolic dysregulation, while histone acetylation becomes the predominant regulatory modification—indicating that the metabolic state of T cells directly shapes their epigenetic landscape [88, 89]. Furthermore, it has been established that targeted metabolic or epigenetic interventions can partially remodel the epigenetic architecture of TEX and restore functional capacity. For example, L‐2‐hydroxyglutaric acid (L‐2‐HG), an immunomodulatory metabolite, enhances TEX cells functionality and antitumor immunity by modulating epigenetic marks such as H3K27me3 [90].

These findings indicate that the irreversibility of T cell exhaustion arises not only from persistent epigenetic locking but also from dysregulation of the metabolism‐epigenetics axis.

Therefore, the irreversibility of T cell exhaustion stems from persistent antigen stimulation, which initiates a self‐reinforcing, multilevel epigenetic network encompassing DNA methylation, histone modifications, and chromatin remodeling. This network stabilizes aberrant transcriptional programs into durable “epigenetic scars” with metabolic dysregulation serving as a key enabler in establishing this profound “molecular lock”.

Metabolic Disorders and Immunosuppressive Microenvironment: External Constraints on Functional Execution

Metabolic dysregulation represents a central extrinsic factor contributing to T cell exhaustion within the TME, with profound interconnections at the molecular level. Persistent antigen exposure, coupled with chronic hypoxia and nutrient deprivation in the TME, induces mitochondrial dysfunction in T cells, characterized by impaired mitochondrial biogenesis, loss of membrane potential, and reduced ATP production [14, 91]. Notably, this mitochondrial impairment is not merely a consequence of T cell exhaustion but also a key driver of its progression.

Accumulation of ROS suppresses PGC‐1α‐mediated mitochondrial biogenesis, leading to diminished mitochondrial quantity and function [92, 93]. This mitochondrial insufficiency not only compromises cellular energy metabolism but also promotes aberrant activation of transcription factors such as NFAT via ROS signaling, thereby upregulating expression of exhaustion‐associated genes, including Blimp‐1 and hypoxia‐inducible factor 1 alpha (HIF‐1α), and facilitating the differentiation of T cells toward a terminally exhausted phenotype [94]. Furthermore, defective mitophagy results in the accumulation of dysfunctional mitochondria, exacerbating both metabolic disturbances and functional decline in T cells [95, 96].

In recent years, sphingolipid metabolism—recognized as a critical component of metabolic reprogramming within the TME—has been established as a key regulator of T cell function and response to immunotherapy. Emerging evidence indicates that CERS4 mediates the suppression of CD8+ Tim‐3+ TEX cells via RhoB signaling, thereby promoting immune modulation and serving as a predictive biomarker for favorable responses to anti‐PD‐1 therapy [97]. Beyond sphingolipid metabolism, other metabolic pathways also significantly contribute to T cell exhaustion.

Tumor cells reprogram methionine metabolism to generate elevated levels of S‐adenosylmethionine (SAM) and 5’‐methylthioadenosine (MTA), which induce chromatin condensation and epigenetic remodeling in T cells, thereby altering the expression profiles of exhaustion‐related genes [20, 98]. Notably, supplementation with L‐2‐hydroxyglutarate (L‐2‐HG) has been shown to partially reverse the exhausted phenotype [90]. Moreover, methionine deficiency reduces H3K79me2 levels, suppresses AMPK expression, upregulates PD‐1, and promotes exhaustion in CD4+ T cells [99].

Concurrently, tumor cells and immunosuppressive populations deplete essential amino acids such as tryptophan and methionine, impairing T cell proliferation and effector functions. Kynurenine (Kyn), a tryptophan metabolite, activates the aryl hydrocarbon receptor (AhR) and directly drives T cell exhaustion [99, 100]. Additionally, TEX cells exhibit diminished glycolytic capacity and restricted glucose uptake and utilization.

Despite sustained activation of mTOR signaling, downstream metabolic pathways are inhibited by signals from immune checkpoint receptors such as PD‐1, leading to impaired OXPHOS, insufficient energy production, and compromised cytotoxic and proliferative capacities [14, 101, 102]. Notably, monocarboxylate transporter 11 (MCT11), a lactate transporter, is highly expressed in TEX‐term cells, facilitating increased lactate uptake and thereby exacerbating metabolic stress and dysfunction [103, 104]. Under conditions of hyperlactatemia, both glycolysis and mitochondrial metabolism in CD8+ T cells are impaired, further promoting the stabilization of the exhausted phenotype [105].

However, recent studies have revealed that lactate exerts a dual role in the regulation of T cell exhaustion. In addition to its traditionally recognized immunosuppressive effects, lactate contributes to the preservation of CD8+ T cell stemness through metabolic‐epigenetic crosstalk. Specifically, on one hand, lactate‐driven histone lactylation—such as H3K18la and H3K9la—can regulate the transcription of key genes involved in CD8+ T cell differentiation and function [86].

On the other hand, lactate inhibits HDAC activity, increases H3K27 acetylation at the Tcf7 super‐enhancer region, and enhances TCF‐1 expression, thereby supporting the stem‐like properties and self‐renewal capacity of CD8+ T cells [106]. Thus, metabolic dysregulation cooperates through multiple interconnected mechanisms—including mitochondrial dysfunction, epigenetic remodeling, impaired glycolysis, lactate accumulation, and amino acid depletion—to drive the initiation, maintenance, and progression of T cell exhaustion.

The immunosuppressive cell populations within the TME play a pivotal role in the induction and maintenance of T cell exhaustion [107]. Regulatory T cell (Treg) directly suppresses effector T cell function through the secretion of immunosuppressive cytokines such as IL‐10 and TGF‐β, as well as through competitive consumption of IL‐2 [108, 109]. Tumor‐associated macrophages (TAMs) and myeloid‐derived suppressor cells (MDSCs) collaboratively establish a robust immunosuppressive network by upregulating inhibitory ligands, including PD‐L1, and producing immunosuppressive mediators such as arginase‐1 (ARG1) and IDO [110].

Furthermore, dendritic cells (DCs) in the TME frequently exhibit a tolerogenic phenotype characterized by reduced expression of costimulatory molecules and elevated expression of PD‐L1 and IDO, thereby failing to effectively prime full T cell activation and instead promoting T cell exhaustion [111, 112, 113].

Emerging Drivers: The Regulatory Role of the Microbiome

The gut microbiome and its metabolites are emerging as systemic regulators of T cell exhaustion [114, 115]. Preclinical evidence demonstrates that chronic restraint stress can disrupt oral microbial homeostasis, thereby promoting premature CD8+ T cell exhaustion and accelerating the progression of head and neck squamous cell carcinoma [116]. Moreover, gut‐derived metabolites, including short‐chain fatty acids (SCFAs), secondary bile acids, and tryptophan derivatives (e.g., indole‐3‐lactic acid), can modulate immune responses.

These metabolites signal through various receptors, such as G protein‐coupled receptors (GPCRs), nuclear hormone receptors (NHRs), and AhR, to directly influence T cell differentiation, metabolism, and function [114, 117, 118]. Collectively, these findings highlight mechanisms underlying the “gut–tumor axis” and the broader microbiota–immune axis in the systemic regulation of anti‐tumor immunity and T cell fate, offering novel insights into the systemic determinants of T cell exhaustion.

The process of T cell exhaustion is highly complex, involving multiple regulatory layers, including persistent antigen stimulation, transcriptional control, epigenetic reprogramming, and metabolic dysregulation. Current studies indicate that under sustained antigen exposure, TCR signaling initiates the activation of key transcription factors such as NFAT and TOX in T cells. These factors not only directly regulate the expression of inhibitory receptors but also drive extensive epigenetic modifications—including DNA methylation, histone modifications, and chromatin remodeling—leading to the establishment of stable “epigenetic scars” that render the exhausted phenotype largely irreversible [11, 119].

Concurrently, this epigenetic locking interacts with metabolic impairments such as defective glycolysis, mitochondrial dysfunction, and the immunosuppressive tumor microenvironment, collectively amplifying T cell exhaustion and ultimately resulting in a permanent loss of effector function [12, 14]. These mechanisms are not isolated; rather, they engage in extensive crosstalk and jointly govern the initiation, maintenance, and therapeutic responsiveness of T cell exhaustion. Therefore, therapeutic strategies targeting T cell exhaustion should extend beyond isolated interventions—such as PD‐1 blockade—and instead focus on modulating early critical nodes within the regulatory network (e.g., preventing the formation of epigenetic scars) or leveraging synergistic approaches, such as combining epigenetic modulators with ICB, to achieve more durable and effective immune restoration.

Heterogeneity in T Cell Exhaustion

T cell exhaustion is not a uniform condition but exhibits substantial spatiotemporal heterogeneity, regulated by multiple factors within the TME, including the intensity and duration of antigen exposure as well as epigenetic programming, all of which significantly influence tumor progression and response to immunotherapy (Figure 3, Table 1).

Temporal Dynamics of Heterogeneity

From the perspective of temporal dynamics, T cell exhaustion follows a progressive differentiation trajectory initiated by chronic antigen stimulation [152]. During acute immune responses, Naïve T cells are activated through TCR engagement and costimulatory signals like CD28. They differentiate into effector T cells with potent cytotoxic activity; some subsequently develop into memory T cells [153, 154, 155].

In contrast, under persistent antigen exposure, Naïve T cells gradually diverge from the memory differentiation pathway and enter an exhaustion trajectory, characterized by the emergence of precursor TPEX that maintain self‐renewal capacity but exhibit gradually diminished effector functions [156, 157]. This transition is driven by sustained TCR signaling, which induces key transcription factors such as TOX and TCF‐1 [121]. TOX promotes chromatin remodeling and exhaustion‐associated gene programs, while TCF‐1 preserves the “precursor” state and functional plasticity [8, 158].

With the continuous stimulation of antigens, TPEX cells gradually downregulate the expression of TCF‐1 and differentiate into transitional TEX‐int cell subsets [159]. These subsets exhibit cytokine secretion and enhanced cytotoxic activity. However, they have irreversibly diverged from the memory/stem lineage [160].

Further differentiation leads to the TEX‐term cell stage, characterized by dominant TOX activity and accompanied by significant epigenetic alterations such as DNA methylation and histone modifications [123]. This process stabilizes the state of depletion even after antigen clearance. TEX‐term cells display severe functional impairment, including loss of proliferative capacity, reduced cytokine production (e.g., IFN‐γ and TNF‐α), increased secretion of inhibitory cytokines (e.g., IL‐10), and high expression of multiple inhibitory receptors (e.g., PD‐1, CTLA‐4, TIM‐3, LAG‐3), rendering them refractory to ICB therapy [161, 162].

Notably, the proportion and functional status of the TPEX subpopulation directly influence ICB efficacy and tumor immune control, suggesting that targeting the regulation of differentiation trajectories, maintaining the TPEX state, or blocking the differentiation to TEX‐term cell is a cutting‐edge direction for enhancing the effectiveness of immunotherapy [163, 164, 165]. Recent single‐cell multi‐omics investigations increasingly confirm that T cell exhaustion is not a simple linear process but involves a spectrum of rare and transitional subsets with distinct molecular and epigenetic features. For example, multi‐omics profiling in colorectal cancer identified TNFRSF18 and CXCL13 as dynamic markers of exhausted CD8+ T cells, with TNFRSF18‐high cells representing an early, stem‐like exhausted state and CXCL13‐high cells marking more terminally exhausted populations.

The transition from TNFRSF18 to CXCL13 expression parallels disease progression and is accompanied by changes in ribosomal stemness, highlighting a previously unappreciated link between tumor cell stemness and T cell exhaustion [166]. It should be noted that current research on temporal dynamics primarily focuses on CD8+ T cells, while temporal heterogeneity in other T cell subsets (e.g., CD4+ T cells) remains underexplored. For a detailed classification of CD8+ TEX cell subsets, refer to Section 3.4.1.

Heterogeneity in Spatial Distribution

In the tumor microenvironment, the spatial distribution of T cell exhaustion exhibits marked heterogeneity. Numerous single‐cell and spatial transcriptomic studies have demonstrated that TEX‐term cells are predominantly localized within the tumor core—a region characterized by profound immunosuppression, nutrient deprivation, and hypoxia—conditions that collectively impair T cell functionality [128, 167, 168]. Similarly, omics analyses in extramedullary myeloma (EMD) have further revealed that TIM3+ PD‐1+ T cells closely colocalize with myeloma cells in the tumor core, where enhanced activity of the oxidative phosphorylation pathway contributes to the exacerbation of T cell dysfunction [169].

In contrast, TPEX cells are preferentially enriched at the tumor periphery or invasive margin. These cells possess stem‐like properties, including self‐renewal capacity and high expression of TCF‐1, and are associated with robust immune responsiveness and greater sensitivity to ICB therapy [125, 135]. Moreover, specific subsets such as CD160+ CD8+ T cells demonstrate anti‐exhaustion and anti‐tumor activities within the tumor‐normal interface, highlighting the critical influence of the local microenvironment on T cell functional states [170, 171].

Notably, integrated spatial transcriptomic and single‐cell sequencing analyses have identified a rare transitional subset of NK‐like exhausted T cells at the tumor edge or invasion front. These cells frequently express natural killer (NK)‐associated receptors (e.g., members of the KLR family), co‐express partial effector functions and exhaustion markers, and may serve as a key cellular substrate underlying responses to immunotherapy [122, 160].

Recent studies have demonstrated that tertiary lymphoid structures (TLS), as ectopic immune activation sites within tumors, play a critical role in shaping spatial heterogeneity and sustaining T cell functionality [172, 173]. Emerging evidence from spatial omics and multiplex immunohistochemical analyses indicates that mature TLS (mTLS), functioning as newly formed lymphoid‐like organs in the tumor microenvironment, serve as niches for TCF1+ stem‐like CD8+ T cells and pre‐exhausted CD4+ T cells. These structures facilitate a multicellular cooperative network centered on B cells and DCs, thereby promoting sustained T cell activation and differentiation, supporting their proliferative capacity, and enhancing antitumor immunity [126, 174].

The presence of mTLS is strongly associated with improved survival prognosis and enhanced response to ICB. Spatial transcriptomic and imaging mass cytometry analyses have revealed that CD39+ PD‐1+ CD8+ stem‐like T cells are highly enriched within TLS and their surrounding regions. The density of TPEX in TLS‐positive tumors is significantly higher than in TLS‐negative counterparts, indicating that TLS serves as a reservoir for TPEX and facilitates their migration into the tumor parenchyma, thereby augmenting anti‐tumor immunity and ICB efficacy [126].

Mechanistically, mTLS promotes the sustained recruitment and functional maintenance of tumor‐specific T cells by enriching and activating TPEX, CD4+ T follicular helper (Tfh) cells, and B cells. This process suppresses the accumulation of TEX‐term cells, delays T cell dysfunction, and enhances responsiveness to ICB. In contrast, in the absence or dysfunction of TLS, the proportion of TPEX cells declines while the proportion of TEX‐term cells increases, contributing to immunotherapy resistance and an elevated risk of tumor immune escape [126, 174, 175].

Notably, the maturity and spatial distribution of TLS critically influence their immunomodulatory function. mTLS located within the tumor core are enriched with memory B cells, plasma cells, and CD4+ T cells, and promote B cell activation, antibody production, and sustained T cell activation. These features collectively enhance intratumoral anti‐tumor immunity and correlate with improved patient survival [174]. In contrast, immature TLS (imTLS) or those at the tumor periphery are enriched in endothelial cells, fibroblasts, and regulatory T cells, upregulate expression of epithelial‐mesenchymal transition (EMT)‐related genes, and may promote immunosuppression and tumor progression [127, 176].

Thus, TLS not only provides a spatial niche for the differentiation and activation of T cell subsets but also, through their maturity and anatomical localization, determines whether the immune microenvironment exhibits an anti‐tumor or pro‐tumorigenic phenotype.

Specificity Across Cancer Types

Although T cell exhaustion is commonly observed across various malignant tumors, the specific phenotype, regulatory mechanisms, and therapeutic responses exhibit considerable heterogeneity among different cancer types. This variability arises from the distinct tissue microenvironments, persistent antigen exposure, and metabolic characteristics inherent to each tumor type, collectively shaping a disease‐specific T cell exhaustion landscape.

Lung Cancer

The regulatory network underlying T cell exhaustion in lung cancer exhibits distinct molecular features and spatial distribution patterns. Mechanistic investigations have demonstrated that Rab37, a small GTPase, mediates the intracellular trafficking and membrane localization of PD‐1 in a lung cancer‐specific manner. Impaired Rab37 function results in intracellular accumulation of PD‐1 and disrupts its proper translocation to the cell membrane, thereby sustaining exhaustion signals and establishing the Rab37/PD‐1/TIM‐3 signaling axis as a potential prognostic biomarker in lung cancer [177].

In addition, recent single‐cell multi‐omics analyses have revealed that casein kinase II (CK2) upregulates TBX21 expression through HDAC8‐mediated epigenetic reprogramming, thereby reversing CD8+ T cell exhaustion and enhancing responsiveness to anti‐PD‐1 therapy in non‐small cell lung cancer (NSCLC). These findings indicate that CK2B is a potential therapeutic target for restoring the function of exhausted effector CD8+ T cells [129]. Notably, cross‐tissue single‐cell sequencing has systematically uncovered the spatial heterogeneity of the exhausted T cell population in NSCLC.

The study demonstrated that TPEX cell subsets are predominantly enriched in tumor‐draining lymph nodes, whereas terminally exhausted subsets are localized within the tumor core, delineating distinct differentiation trajectories and migratory patterns [178]. At the level of the immune microenvironment, M2 macrophages in small cell lung cancer (SCLC) contribute to CD8+ T cell inactivation through the recruitment of FOXP3+ Tregs and the upregulation of PD‐L1 and other immunosuppressive factors. This fosters an “immune‐excluded” microenvironment, where increased immune cell infiltration is accompanied by functional impairment [179].

Collectively, these findings elucidate the multi‐scale regulation of T cell exhaustion in lung cancer and provide a rational foundation for developing combination immunotherapies targeting distinct phases of T cell dysfunction and the associated tissue microenvironment.

Breast Cancer

The heterogeneous mechanisms underlying CD8+ TEX cell in breast cancer are primarily driven by molecular subtype‐specific factors, particularly in ERα‐negative subtypes such as triple‐negative breast cancer (TNBC) [180, 181]. In ERα‐negative tumors, twist family BHLH transcription factor 1 (TWIST1) binds to the PD‐L1 promoter and recruits the TIP60 acetyltransferase complex via BRD8, leading to increased PD‐L1 expression and enhanced CD8+ TEX cells [182]. Furthermore, extracellular vesicles secreted by breast cancer cells deliver the TGF‐β type II receptor (TβRII) to CD8+ T cells, thereby activating the SMAD3‐TCF1 signaling axis and promoting T cell exhaustion [16].

This immunosuppressive pathway is especially pronounced in TNBC, where the deubiquitinating enzyme USP8 stabilizes TβRII, thereby enhancing sustained TGF‐β signaling and immune evasion. Concurrently, exosome‐mediated transfer of immune cell adhesion molecule 1 (ICAM1) contributes to bone metastasis in TNBC through a related mechanism [183]. In addition, single‐cell transcriptomics and spatial proteomics have revealed that PD‐1high CXCL13+ TEX cells in breast cancer are spatially associated with an inflammatory immune microenvironment and imTLS, exhibiting significant functional and spatial heterogeneity across subsets [131, 132].

These findings elucidate subtype‐specific immune escape mechanisms in breast cancer and identify potential therapeutic targets for combination immunotherapies targeting distinct molecular pathways.

Glioblastoma

Glioblastoma (GBM) is characterized by a distinct immunosuppressive tumor microenvironment, in which T cell exhaustion is particularly pronounced among solid tumors [184]. A cross‐cancer comparative study revealed that GBM‐infiltrating T cells display an atypical exhaustion phenotype compared to those in melanoma, breast cancer, and lung cancer [185]. Specifically, clonally expanded infiltrating T cell populations often exhibit low PD‐1 and TIM‐3 expression and tend to differentiate into a GZMK+ effector‐like phenotype with limited cytotoxic capacity [186].

Single‐cell lineage tracing and multi‐omics analysis further revealed that myeloid cells—including TAMs and microglia—play a pivotal role in promoting T cell exhaustion through the secretion of immunosuppressive cytokines such as IL‐10 and TGF‐β, as well as the expression of inhibitory ligands including PD‐L1 and B7‐H3 [187]. Furthermore, anatomical barriers such as the blood‐brain barrier and bone marrow sequestration restrict T cell infiltration, while metabolic reprogramming‐induced nutrient competition and hypoxia collectively contribute to a complex, multilayered immunosuppressive milieu in GBM [188, 189]. Notably, a significant sex‐based difference in T cell exhaustion has been observed in GBM.

Male CD8+ T cells demonstrate a greater propensity for exhaustion, evidenced by a higher proportion of progenitor‐exhausted subsets, whereas female T cells preserve stronger effector functionality. This sexual dimorphism is driven by the X‐chromosome escape gene Kdm6a, highlighting unique features of the GBM immune landscape [190]. Furthermore, another study demonstrated that the knockdown of tumor necrosis factor receptor type II (TNFR2) led to the emergence of a T cell population expressing TIM3, yet characterized by reduced TOX expression and altered functional properties, encompassing both precursor and terminally exhausted subsets.

In a glioma mouse model, inhibition of TNFR2 signaling significantly prolonged the overall survival of tumor‐bearing mice [191]. These findings may provide new insights into the understanding of T cell exhaustion within intracranial tumor microenvironments. Clinical evidence further indicates that GBM patients experiencing disease progression following anti‐angiogenic therapy exhibit a markedly exacerbated state of T cell exhaustion, characterized by increased Treg infiltration, upregulated PD‐1 expression, and functional impairment—highlighting the intricate dynamic interplay between therapeutic response and immune modulation [192].

Colorectal Cancer

The heterogeneity of T cell exhaustion in colorectal cancer (CRC) exhibits distinct organ‐specific characteristics, primarily shaped by the unique immune microenvironment, molecular subtypes, and metabolic reprogramming within the gut. A recent landmark study has identified thyroid‐stimulating hormone (TSH)/thyroid‐stimulating hormone receptor (TSHR) signaling as a key intrinsic pathway that promotes CD8+ T cell exhaustion in CRC. Specifically, cancer cells deliver the TSHR to CD8+ T cells via the exosomal pathway, thereby inhibiting their effector differentiation—a mechanism that plays a pivotal role in establishing the CRC‐specific exhausted microenvironment [135].

In contrast, by comparing ileal and colonic samples from CRC patients with those from healthy donors, a group of researchers recently identified an ileum‐enriched population of CD160+ CD8+ T cells that exhibit resistance to terminal exhaustion and possess robust clonal expansion capacity. In both microsatellite instability‐high (MSI‐H) and anti‐PD‐1‐resistant CRC models, CD160+ CD8+ T cells enhance tumor‐infiltrating progenitor‐exhausted T cells, improve anti‐PD‐1 efficacy, and overcome therapeutic resistance, thereby significantly augmenting antitumor immunity [134]. In microsatellite stable (MSS) tumors, CD8+ TEX cells co‐express PD‐1 and TIGIT, display reduced cytokine secretion, and exhibit limited cytotoxic function, consistent with features of terminal exhaustion [193].

Furthermore, gut‐specific dysregulation of ammonia metabolism exacerbates T cell exhaustion through metabolic reprogramming, which may partially account for the widespread resistance to ICB observed in CRC and provides a rationale for therapeutic strategies targeting ammonia clearance [170].

Hematologic Malignancies

Hematologic malignancies arise within the hematopoietic system, and their systemic dissemination confers unique dynamic features to T cell exhaustion. In B‐cell acute lymphoblastic leukemia (B‐ALL), CD4+ T cells—rather than CD8+ T cells—are identified as the primary population undergoing exhaustion. In a clinical study of children with B‐ALL, researchers observed that CD4+ T cells in the bone marrow exhibited elevated expression of exhaustion markers, including TIM‐3 and PD‐1, along with impaired effector function and a diminished anti‐leukemic response—features strongly associated with a significantly increased risk of disease recurrence.

Patients with a high proportion of TIM‐3+ CD4+ T cells in the bone marrow at diagnosis had a 7.1‐fold increased risk of relapse, indicating that TIM‐3‐mediated dysfunction of CD4+ T cells plays a critical role in immune escape and disease recurrence in B‐ALL [137]. This exhaustion may be driven by leukemia‐induced TIM‐3 signaling, particularly through CD200/TIM‐3 interactions, rather than the classical PD‐1/PD‐L1 pathway [137]. On the other hand, Sean I.

Tracy et al. revealed, through single‐cell sequencing of bone marrow T cells from patients with B‐ALL, that tumor‐TEX encompasses multiple differentiation states and functional lineages. The study identified two distinct CD4+ T cell subsets, designated as Thctx (T helper/cytotoxic) and exhausted Thctx (Thctx‐ex). These cells co‐expressed cytotoxic genes such as GZMB and NKG7, as well as helper cytokine genes including IFNG and TNF, indicating dual functional capabilities encompassing both cytotoxic and helper activities.

Furthermore, the study demonstrated that the combination of the BCR‐ABL tyrosine kinase inhibitor nilotinib and the anti‐PD‐L1 monoclonal antibody significantly reversed this exhausted phenotype, thereby promoting clonal expansion of leukemia‐specific CD4+ T cells and markedly improving long‐term survival in a murine model of leukemia [138]. Compared with B‐ALL, other hematological malignancies, such as multiple myeloma (MM), exhibit a more complex immune microenvironment. Studies have shown that T cells in the bone marrow and peripheral blood of MM patients display elevated expression of exhaustion markers, including PD‐1, TIM‐3, and TOX.

However, the classical terminal exhaustion phenotype is not prominently observed, and exhausted T cell clones are typically small in size and non‐persistent. Functional impairment in these cells is primarily due to the immunosuppressive microenvironment and defects in antigen presentation, rather than classical antigen‐driven exhaustion [194, 195]. This provides a mechanistic explanation for the limited efficacy of PD‐1/PD‐L1 inhibitors in MM.

Similarly, in diffuse large B‐cell lymphoma (DLBCL), recent evidence indicates that although certain CD8+ T cells exhibit features of terminal exhaustion (e.g., IFN‐γ+ TPEX), they retain cytotoxic activity and cytokine secretion capacity—distinct from the functionally impaired TEX cells observed in chronic infections [196]. In contrast, CD8+ T cells may express exhaustion markers such as PD‐1 without complete loss of function in some contexts, which may account for the modest response of DLBCL to single‐agent ICIs.

Hepatocellular Carcinoma

The exhaustion phenotype in hepatocellular carcinoma (HCC) is driven not only by persistent tumor antigen stimulation but also by the liver's innate immune regulatory mechanisms and the presence of viral antigens, resulting in a more complex and sustained state of T cell dysfunction [171]. Transcriptomic analyses have demonstrated that TIGIT, an inhibitory receptor, is significantly more upregulated in HCC‐specific exhausted CD8+ T cells compared to PD‐1, suggesting that TIGIT may serve as a more reliable biomarker for identifying this subset of dysfunctional T cells [197]. Furthermore, the unique immune‐tolerant microenvironment characteristic of HCC influences the distinct differentiation trajectory of TEX cells.

This microenvironment is enriched with Tregs, MDSCs, and immunosuppressive cytokines such as TGF‐β and IL‐10, which collectively foster a highly tolerant milieu and promote the accelerated progression of CD8+ T cells toward terminal exhaustion—a state typically more profound and durable [198]. Notably, approximately 80% of HCC cases are associated with chronic HBV or HCV infection [199]. In such contexts, virus‐specific and tumor‐specific T cells coexist within the same microenvironment and are subjected to prolonged antigenic stimulation.

This dual pressure from viral and tumor antigens leads to earlier and more robust establishment of epigenetic programs linked to T cell exhaustion, thereby limiting the responsiveness of these cells to ICB therapies, including PD‐1 monotherapy [200, 201, 202]. Additionally, the liver‐specific metabolic environment—particularly the elevated levels of ROS and lactic acid resulting from mitochondrial dysfunction—exacerbates T cell exhaustion through metabolic stress, further impairing anti‐tumor immunity [203].

Cervical Cancer

Cervical cancer (CESC) exhibits a distinct pattern of T cell exhaustion, primarily attributable to its well‐established viral etiology. The persistent expression of human papillomavirus (HPV)‐derived E6 and E7 oncoproteins serves as a shared tumor‐associated antigen, driving sustained and intense activation of virus‐specific T cell clones. This chronic stimulation promotes early and stable induction of T cell exhaustion program via NFAT‐TOX‐mediated epigenetic remodeling [139, 204, 205, 206].

Beyond this canonical viral‐driven mechanism, CESC also demonstrates unique metabolic dysregulation in the tumor microenvironment. Specifically, tumor cells overexpress SERINC2, leading to competitive depletion of serine, which results in metabolic deprivation of T cells. This nutrient limitation directly impairs T cell proliferation and effector functions, thereby accelerating their functional exhaustion [140].

In addition, spatial omics and immunohistochemistry confirmed that tumor‐specific B cells and SPP1+ macrophages interact closely with T cell subsets to collectively shape the immunosuppressive tumor microenvironment [207, 208]. At the immune checkpoint level, in addition to upregulation of classical inhibitory receptors such as PD‐1 and TIM‐3, CESC exploits the non‐classical checkpoint B7‐H3, which aberrantly engages the costimulatory receptor 4‐1BB to deliver inhibitory signals—a mechanism that constitutes a distinctive pathway of immune evasion [209, 210]. Recent studies have reported that MiR‐379‐5p, as an epigenetic tumor suppressor, may serve as a novel biomarker for cancer immunotherapy by enhancing the function of CD8+ T effector cells and inhibiting T cell exhaustion [211].

Collectively, these molecular and immunological features underscore the potential responsiveness of CESC to emerging therapeutic strategies, including HPV antigen‐targeted vaccines, metabolic modulation, and B7‐H3‐directed interventions.

Heterogeneity of T Cell Exhaustion Subsets

T cell exhaustion manifests across diverse T cell lineages, including CD8+, CD4+, γδ T cells, and mucosa‐associated invariant T (MAIT) cells. However, the underlying regulatory mechanisms exhibit both conserved and subset‐specific features. Common drivers include persistent antigen stimulation, sustained expression of inhibitory receptors (e.g., PD‐1, TIM‐3), epigenetic reprogramming mediated by transcription factors such as TOX and NR4A, and metabolic dysregulation.

In contrast, the functional specialization and differentiation pathways of each subset also confer unique exhaustion phenotypes and therapeutic vulnerabilities.

CD8+ T Cell Exhaustion Subsets

The heterogeneity of CD8+ T cell exhaustion subsets has been elucidated through single‐cell omics and functional studies, revealing major populations including TPEX, TEX‐int, and TEX‐term cells. These subsets exhibit multi‐level complexity in terms of spatial distribution, molecular signatures, functional properties, and clinical relevance [122, 142]. The TPEX subset is characterized by high expression of TCF‐1 (Tcf7), Ly108 (SLAMF6), and CD69, and is predominantly localized in lymphoid tissues, tumor‐associated TLS, and the tumor stroma.

TPEX cells possess self‐renewal capacity and robust proliferative potential, with low expression of inhibitory receptors. They represent a key target for ICB therapy and play a pivotal role in mediating anti‐tumor immune responses and determining ICB efficacy [122, 141]. The TEX‐int subset is defined by high T‐bet expression, CX3CR1 positivity, and partial retention of effector functions, such as the secretion of IFN‐γ and TNF‐α, as well as cytotoxic activity.

Functionally and phenotypically intermediate between TPEX and TEX‐term cells, TEX‐int cells can respond to ICB and contribute to anti‐tumor immunity, although they have irreversibly lost stem‐like properties [74]. Recent single‐cell omics analyses have further identified rare transitional subpopulations within the TEX cell compartment, including NK‐like, ISG+, and other novel intermediate states, which express distinct molecular markers such as those from the KLR family, CXCL13, and TNFRSF18. These findings have significantly expanded the heterogeneity landscape of T cell exhaustion [122, 166, 212, 213].

However, the functional roles and clinical implications of these emerging subsets remain unclear. Their inclusion in the canonical classification framework requires further validation. The TEX‐term cell subset is characterized by high expression of transcription factors such as TOX, EOMES, and NR4A, along with upregulation of surface molecules including CD101, TIM‐3, and LAG‐3.

It is predominantly localized within the tumor parenchyma and exhibits severe functional impairment, epigenetic fixation, and poor responsiveness to ICB. As such, it represents a major contributor to immunotherapy resistance and tumor immune escape [142, 143]. Recent pan‐cancer single‐cell transcriptomic studies have established a unified clustering framework for CD8+ TEX cells across species and tumor types, identifying six distinct subsets (C1–C6) with divergent functional and molecular profiles.

Notably, the C1 subset is significantly expanded following ICB treatment, suggesting its potential as a predictive biomarker for therapeutic response [143]. In summary, the spatial distribution and functional states of CD8+ TEX cell subsets exhibit considerable plasticity and dynamic variation across different tumor types and tissue microenvironments [143, 213, 214].

CD4+ T Cell Exhaustion Subsets

CD4+ T cells are core coordinators of immune responses, and their exhaustion affects the integrity of the tumor immune network [146]. Recent studies have demonstrated that CD4+ T cells not only share a high degree of phenotypic and differentiation similarity with CD8+ T cells, but also possess distinct subset characteristics and regulatory mechanisms [146, 215]. Tfh cells were divided into non‐cytotoxic and cytotoxic groups.

The former had high expression of TCF1, low BLIMP1 and LAG‐3, self‐renewal ability, and sensitivity to PD‐1 monoclonal antibody, similar to CD8+ TEX cell subsets. The latter group has low TCF1, high BLIMP1 and LAG‐3, loss of self‐renewal, and is resistant to PD‐1 monoclonal antibody, but may be sensitive to PD‐1/LAG‐3 dual blockade, corresponding to the terminal exhaustion state [145, 146, 216]. This parallel differentiation pattern suggests a partially conserved regulatory mechanism underlying T cell exhaustion in both CD4+ and CD8+ T cell lineages.

Additionally, Tregs exhibiting exhaustion‐associated phenotypes (e.g., upregulation of inhibitory receptors such as PD‐1, TIM‐3, and altered suppressive function) are increasingly recognized as a distinct functional state [146, 147, 217]. Traditionally, Tregs have been regarded as key mediators of immunosuppression [218]. However, in the contexts of chronic inflammation and the tumor microenvironment, a subset of Tregs upregulates multiple inhibitory receptors, such as PD‐1, TIGIT, and TIM‐3, acquiring enhanced immunosuppressive capacity, while potentially compromising their lineage stability [219, 220, 221].

In this “super‐suppressive” state, Tregs not only exert more potent inhibition on effector T cell function but also actively contribute to shaping an immunosuppressive microenvironment through the secretion of cytokines, including IL‐10 and TGF‐β [15, 222, 223]. However, it is important to note that there exists a critical distinction between the “super inhibitory” activity of Tregs exhibiting exhaustion‐associated phenotypes and their therapeutic capacity in maintaining systemic immune homeostasis. Lamarche et al. demonstrated that although exhausted CAR‐Tregs exhibited enhanced suppressive function in vitro, they completely lost their protective efficacy against graft‐versus‐host disease (GVHD) in vivo [224].

This discrepancy highlights the functional complexity of Tregs and underscores the necessity of relying on in vivo models for accurate evaluation of cellular therapies. This loss of in vivo functionality may be attributed to metabolic rigidity and epigenetic stabilization associated with T cell exhaustion, which could impair adaptability in complex physiological environments.

It is worth emphasizing that although both CD8+ and CD4+ T cell exhaustion are characterized by functional impairment and elevated expression of inhibitory receptors driven by persistent antigen stimulation and an immunosuppressive microenvironment, there exist significant differences as well as overlapping features between these subsets in terms of molecular mechanisms, phenotypic characteristics, and functional outcomes. The exhaustion trajectory of CD8+ T cells is relatively well defined, involving a stepwise lineage differentiation from TPEX to TEX‐int and ultimately to TEX‐term cells [225]. In contrast, CD4+ T cell exhaustion follows a more heterogeneous differentiation pathway, encompassing both stem‐like to terminal differentiation within Tfh‐like subsets and the acquisition of a “hypersuppressive” state in Tregs [144, 216].

Functionally, CD8+ T cell exhaustion primarily results in diminished cytotoxic capacity against tumors, whereas CD4+ T cell exhaustion indirectly impairs CD8+ T cell activity, dendritic cell maturation, and immune microenvironment remodeling through dysfunctions in helper, regulatory, and cytotoxic functions [144, 215]. Moreover, exhausted CD4+ T cells promote the formation of tumor‐associated tertiary lymphoid structures by secreting chemokines such as CXCL13, thereby facilitating the recruitment and activation of CXCR5+ CD8+ TPEX cells [226, 227]. Both subsets play synergistic roles in tumor progression and immune escape, underscoring the importance of developing coordinated targeting strategies in future cancer immunotherapies.

Unconventional T Cell Exhaustion

In addition to the well‐documented depletion of CD8+ and CD4+ T cells, the functional exhaustion of γδ T cells has increasingly drawn attention in cancer research in recent years [228]. γδ T cells represent a distinct subset of T lymphocytes that are independent of MHC restriction and antigen‐presenting cells, and they play a critical role in immune surveillance and host defense through the secretion of diverse cytokines [229]. However, within the tumor microenvironment, γδ T cells frequently exhibit functional exhaustion. For instance, a recent single‐cell RNA sequencing (scRNA‐seq) study in a murine model of colon cancer identified tumor‐infiltrating CD8α− PD‐1+ γδ T cells with downregulated expression of cytotoxicity‐associated genes and upregulated expression of IL‐17‐related and tumor‐promoting genes [230].

More importantly, a mechanistic investigation in cervical cancer demonstrated that BTN3A1 expressed on tumor cell surfaces activates the TCR signaling pathway, leading to increased expression of the exhaustion‐associated transcription factors NR4A2 and NR4A3, thereby driving Vδ2 T cells into a state of functional exhaustion [148]. This finding provides a defined molecular mechanism underlying Vδ2 T cell exhaustion. Furthermore, terminally differentiated Vδ2+ γδ T cells display an exhausted phenotype that correlates with lymph node metastasis, and elevated PD‐1 expression on these cells in the peripheral blood of breast cancer patients underscores their potential as a biomarker for disease progression [231].

PD‐1 blockade has been shown to significantly restore the cytotoxic function of tumor‐infiltrating γδ T cells and enhance the antitumor efficacy of γδ T cell‐based adoptive immunotherapy in a murine model of prostate cancer [149]. Clinical studies have further demonstrated that the functional capacity of Vδ2+ γδ T cells is restored in a subset of patients with solid tumors following PD‐1/PD‐L1 immune checkpoint blockade, indicating that γδ T cell dysfunction is reversible and may represent a viable target for immunotherapeutic intervention. Similarly, tumor‐infiltrating Vγ9− γδ T cells are frequently observed in high numbers in both solid tumors and lymphomas.

Their abundance correlates with the extent of tissue residency and exhaustion and exhibits significant similarity to CD8+ T cells in terms of differentiation trajectory, gene expression profiles, and responsiveness to immune checkpoint therapy [232]. Collectively, these findings suggest that activated γδ T cells in the tumor microenvironment display exhaustion features analogous to those of CD8+ T cells, characterized by concurrent and sustained upregulation of immune checkpoint molecules—although this phenotype extends beyond checkpoint expression alone.

The exhaustion of MAIT cells has garnered significant attention in recent years [233]. MAIT cells play a critical role in mucosal immunity through the recognition of microbial metabolites presented by MR1 molecules. Evidence indicates that tumor‐infiltrating MAIT cells exhibit high expression of exhaustion markers, including PD‐1, Tim‐3, and CD39, consistent with a terminally exhausted phenotype.

Although these cells retain some proliferative capacity, their functional versatility is markedly impaired, particularly in terms of the production of key anti‐tumor cytokines such as IFN‐γ and TNF [150]. At the same time, recent studies have demonstrated that in cancer‐associated chronic conditions such as HCC, abnormal accumulation of polyunsaturated fatty acids (PUFAs) in MAIT cells induces lipid peroxidation, leading to impaired mitochondrial respiration and glycolysis and establishing a state of “metabolic exhaustion.” Sustained lipid peroxidation ultimately triggers ferroptosis, resulting in functional impairment and cell death of MAIT cells [151]. Encouragingly, in vitro studies have demonstrated that PD‐1 blockade can partially restore MAIT cell function, suggesting that this subset may represent a promising target for ICB therapies [150].

In summary, T cell exhaustion across different subsets shares a common molecular framework centered on the TOX/NR4A‐driven transcriptional program, epigenetic fixation, and upregulation of PD‐1, while also exhibiting significant heterogeneity due to differences in antigen recognition patterns, functional specialization, and microenvironmental cues. For CD8+ T cells, the primary therapeutic target is PD‐1 blockade to restore cytotoxic function. For CD4+ T cells, intervention should focus on correcting the imbalance between helper and suppressor activities.

For γδ T cells, strategies must address their MHC‐unrestricted activation pathways. Future therapeutic approaches should integrate broad‐spectrum checkpoint inhibitors with subset‐specific targeting modalities to achieve precise immune reconstitution.

Biomarkers of T Cell Exhaustion

The establishment and maintenance of T cell exhaustion represent a dynamic process that is tightly regulated by multilayered molecular networks. These molecules not only serve as “identity markers” for identifying exhausted states but also function as key effectors directly involved in driving T cell dysfunction. This section provides a systematic overview of the critical biomarkers defining T cell exhaustion, organized across three dimensions: surface inhibitory receptors, core transcription factors, and metabolism‐associated molecules, along with an in‐depth analysis of their intrinsic regulatory mechanisms.

Surface Inhibitory Receptors

Under conditions of chronic antigen stimulation, T cells co‐express multiple inhibitory receptors on their cell surface, thereby establishing a synergistic immune checkpoint network that collectively suppresses T cell functionality. A comprehensive understanding of both the distinct and overlapping regulatory mechanisms of these receptors is fundamental to the development of effective combination immunotherapies (Figure 4).

PD‐1

PD‐1, also known as CD279, is widely recognized as the most characteristic surface marker of T cell exhaustion [234]. It functions as a critical immunosuppressive molecule and is predominantly expressed on activated T cells, B cells, and macrophages, playing a significant role in immune regulation [235, 236]. As a key receptor on the surface of T cells, PD‐1 effectively suppresses T cell activation and dampens the immune response upon interaction with its ligands, thereby facilitating the transition of T cells into an exhausted state [237, 238].

To date, two ligands for PD‐1 have been identified: PD‐L1 (B7‐H1, CD274) and PD‐L2 (B7‐DC, CD273) [239]. The interaction between PD‐1 and its ligands is recognized as a pivotal immune checkpoint mechanism that negatively regulates T cell proliferation. Upon ligand binding, PD‐1 recruits intracellular SHP‐1 and SHP‐2, leading to the inhibition of key signaling molecules within the TCR signaling pathway [240].

Further mechanistic investigations have demonstrated that PD‐1 signaling mediates immunosuppression through multiple distinct pathways. First, it inhibits cell cycle progression by suppressing cyclin‐dependent kinase (CDK) activity [241, 242]. Second, it attenuates TCR signaling microcluster formation via downregulating CK2.

Additionally, in the TME, sustained PD‐1 activation promotes epigenetic remodeling—such as DNA methylation and histone modifications—which stabilizes the TEX cell phenotype [243, 245]. Moreover, PD‐1 signaling is not only prevalent in T cells during chronic viral infections but is also markedly upregulated in tumor‐infiltrating T cells, contributing to T cell exhaustion and impaired anti‐tumor immunity [246]. For instance, in approximately half of NSCLC patients treated with PD‐1 inhibitors, an increase in proliferating Ki‐67+CD8+ T cells was observed in peripheral blood, with these cells predominantly residing in the PD‐1‐expressing subset.

Additional evidence indicates that the CD28/B7 costimulatory pathway plays a critical role in modulating this response [243]. Notably, Ma et al. reported that co‐expression of PD‐1 and TIM‐3 at high levels is strongly associated with functional impairment of CD8+ T cells and reduced survival in HCC patients, highlighting the importance of further evaluating these receptors for potential therapeutic targeting [247].

The sustained high expression of PD‐1 on TEX cells has driven the development of ICB therapies targeting the PD‐1/PD‐L1 axis [248]. In 2014, the U.S. Food and Drug Administration (FDA) approved two PD‐1‐blocking monoclonal antibodies, pembrolizumab and nivolumab, for the treatment of advanced melanoma, squamous NSCLC, and renal cell carcinoma [249, 250].

In liver cancer research, GLIS family zinc finger 1 (GLIS1) has been identified as a transcriptional regulator of the SGK1‐STAT3‐PD‐1 signaling pathway, which maintains elevated PD‐1 expression on CD8+ T cells and promotes T cell exhaustion. Downregulation of GLIS1 expression has been shown to synergistically enhance CD8+ T cell infiltration and reduce their exhausted phenotype in combination with anti‐PD‐1 therapy, offering a novel and promising strategy for liver cancer immunotherapy [251]. Furthermore, PD‐1/PD‐L1 blockade has demonstrated the ability to augment the antitumor efficacy of peptide vaccines by promoting the induction of tumor‐specific CTLs in patients with HCC [252].

Despite these encouraging clinical advances, PD‐1 blockade still faces significant challenges. Multiple adaptive resistance mechanisms within the tumor microenvironment—including upregulation of alternative immune checkpoints, loss of costimulatory signals, and metabolic reprogramming—collectively contribute to heterogeneous treatment responses. These findings underscore the need for future research to comprehensively elucidate the regulatory mechanisms of the PD‐1 signaling network using multi‐omics approaches and to develop biomarker‐guided precision therapeutic strategies to overcome current limitations in immunotherapy [253].

CTLA‐4

CTLA‐4 functions as a dominant suppressor at the immune synapse by competing with CD28 for binding to CD80 and CD86 [254]. CTLA‐4 exhibits significantly higher affinity for CD80/CD86 than CD28 and delivers a potent inhibitory signal through its cytoplasmic domain. Additionally, through the process of trans‐endocytosis, CTLA‐4 can directly remove CD80/CD86 from the surface of antigen‐presenting cells, thereby attenuating the costimulatory signals required for T cell activation [58, 255, 256, 257].

Notably, CTLA‐4 has been shown to induce the expression of IDO in antigen‐presenting cells, leading to the catabolism of tryptophan into kynurenine—a metabolite with immunosuppressive properties—thus indirectly contributing to T cell functional exhaustion [258]. Elevated CTLA‐4 expression has been consistently associated with T cell dysfunction across various tumor types. In patients with HCC and chronic hepatitis B, both peripheral and tumor‐infiltrating CD8+ T cells exhibit high levels of CTLA‐4 expression, which strongly correlates with impaired effector function [259].

Furthermore, in the ovarian cancer microenvironment, CTLA‐4 promotes the development of a T cell senescence phenotype by antagonizing CD28‐mediated signaling pathways [260].

Given the central role of CTLA‐4 in immune regulation, its targeted therapy has emerged as a promising strategy in tumor immunotherapy [261]. In clinical trials involving patients with metastatic melanoma, the CTLA‐4 inhibitor ipilimumab demonstrated significantly improved progression‐free survival compared to PD‐1 checkpoint inhibitor monotherapy, along with a lower incidence of adverse events [262]. In models of chronic HCV infection, CTLA‐4 and PD‐1 have been shown to induce T cell exhaustion in a functionally redundant manner, and dual blockade of these pathways can synergistically restore T cell function in a CD4‐independent but CD28‐dependent manner, highlighting their therapeutic potential [44].

However, in highly immunosuppressive tumor microenvironments such as those observed in pancreatic cancer, CTLA‐4 blockade enhances infiltration of CD4+ T cells but exerts limited effects on CD8+ T cell recruitment, indicating that the efficacy of this intervention is strongly influenced by tumor type‐specific factors [263]. These findings underscore the importance of further investigating cell‐type‐specific mechanisms underlying CTLA‐4‐mediated immune regulation. Future research should focus on developing precise patient stratification strategies based on the characteristics of the tumor immune microenvironment (TIME) as well as optimizing combination regimens involving CTLA‐4‐targeted therapies and other immunomodulatory approaches to achieve more effective antitumor immune responses.

TIGIT

As an immune checkpoint molecule following CTLA‐4 and PD‐1, TIGIT has garnered significant attention, with its structure and functional mechanisms becoming increasingly well characterized [264]. The molecular architecture of TIGIT consists of an extracellular immunoglobulin variable (IgV) domain, a transmembrane domain, and an intracellular domain. The latter contains two highly conserved signaling motifs: an ITIM and an immunoreceptor tyrosine‐based tail (ITT)‐like motif [265].

Upon ligand binding, TIGIT undergoes phosphorylation and subsequently recruits adaptor proteins such as Grb2 and β‐arrestin 2, ultimately facilitating the recruitment of the SHIP1 phosphatase. This cascade leads to the negative regulation of the PI3K‐MAPK and NF‐κB signaling pathways, thereby directly suppressing TCR signaling and diminishing the activation capacity of CD8+ T cells. Additionally, TIGIT synergistically enhances Treg immunosuppressive function [266, 267].

TIGIT expression is closely associated with T cell exhaustion across various malignancies [268, 269]. It can be transiently upregulated following TCR activation and is stably expressed in NK cells, Tregs, Tfh cells, and CD8+ TEX cells [120]. Notably, TIGIT is specifically and highly expressed in a well‐defined subset of exhausted CD8+ T cells—namely, TOXhi TCF‐1hi cells—in both human and murine models, and is widely regarded as a key marker of T cell exhaustion [197, 246].

Elevated TIGIT expression has also been significantly correlated with T cell exhaustion and poor clinical outcomes in patients with CRC, further underscoring its role as a central mediator of negative immune regulation [270].

The interaction of TIGIT with its multiple ligands plays a pivotal role in mediating immunosuppression. Known TIGIT ligands include CD155 (PVR), CD112, CD113, and Nectin‐4, among which CD155 is recognized as the primary high‐affinity ligand for TIGIT in both humans and mice [271, 272, 273]. Targeted inhibition of the TIGIT–ligand interaction, particularly with CD155, has emerged as a promising strategy to restore antitumor immune responses.

For instance, in cervical cancer, Liu et al. demonstrated that elevated expression of CD155 and TIGIT within the TME is strongly associated with impaired immune cell function, and that blockade of the TIGIT/CD155 signaling axis can reverse CD8+ T cell exhaustion and enhance anti‐tumor immunity [139]. Similarly, Wu et al. reported in head and neck squamous cell carcinoma (HNSCC) that anti‐TIGIT therapy significantly ameliorates the immunosuppressive microenvironment by reducing ARG1 expression in MDSCs and suppressing TGF‐β1 secretion from Tregs [274]. Currently, several TIGIT‐blocking monoclonal antibodies are under clinical development.

Notably, Ociperlimab (BGB‐A1217), a humanized IgG1 monoclonal antibody with intact Fc functionality, effectively inhibits the binding of TIGIT to CD155 and CD112. Preclinical studies have indicated that combining Ociperlimab with PD‐1/PD‐L1 inhibitors yields synergistic anti‐tumor effects [275]. Furthermore, in an autologous stem cell transplantation model of multiple myeloma, dual blockade of TIGIT and PD‐1 markedly prolonged tumor control, potentially through modulation of the T‐bet/Eomes expression ratio to enhance T cell cytotoxicity and suppress IL‐10‐mediated immune microenvironment remodeling [276].

With ongoing advances in understanding the TIGIT signaling pathway and its immunoregulatory functions, TIGIT‐targeted therapies—particularly in combination with other immunotherapeutic approaches—are anticipated to offer novel therapeutic opportunities for cancer patients.

LAG‐3

LAG‐3 (CD223) is an inhibitory receptor that has garnered significant attention in the field of cancer immunotherapy in recent years. Marked upregulation of LAG‐3 on TEX cells in chronic viral infections and the TME suggests that LAG‐3 blockade may restore antitumor T cell function [40, 277]. Structurally, LAG‐3 is highly homologous to CD4 but binds MHC class II molecule (MHC‐II) with much higher affinity [278].

Ei Wakamatsu et al. demonstrated that LAG‐3 can reduce surface expression of MHC‐II on antigen‐presenting cells (APCs) through trans‐endocytosis, thereby impairing antigen presentation efficiency and suppressing CD4+ T cell activation. Anti‐LAG‐3 therapy can directly inhibit this process, alleviate the intrinsic inhibitory signals mediated by LAG‐3, prevent downregulation of MHC‐II, and promote the functional recovery of TEX cells, offering a promising strategy for cancer immunotherapy [279]. In addition to MHC‐II, fibrinogen‐like protein 1 (FGL‐1) has been identified as another key functional ligand of LAG‐3.

FGL‐1, primarily secreted by hepatocytes, promotes HCC cell proliferation [280, 281]. The binding of elevated FGL1 to LAG‐3 within the TME drives the exhaustion of CD8+ T resident memory (TRM) cells within the tumor, thus facilitating tumor immune escape [282]. Given its critical role in tumor immune regulation, targeting the LAG‐3/FGL‐1 signaling axis represents a promising avenue for the development of novel immunotherapeutic approaches [283].

To date, a diverse array of anti‐LAG‐3 monoclonal antibodies has entered clinical development [284]. Relatlimab (BMS‐986016), the first fully human IgG4‐κ monoclonal antibody targeting LAG‐3, exhibits high‐affinity binding to LAG‐3 and effectively blocks its interaction with MHC‐II. It has demonstrated favorable tolerability and promising clinical activity in phase I‐III trials involving patients with solid tumors and hematologic malignancies [285].

Additional investigational agents, including Ieramilimab (LAG525), TSR‐033, and Favezelimab (MK‐4280), are currently under evaluation in clinical studies for advanced solid tumors [286, 287, 288]. Notably, LAG‐3 is often co‐expressed with PD‐1, together driving T cell exhaustion [289]. Ngiow et al. reported that, in a melanoma model, dual blockade of LAG‐3 and PD‐1 can recalibrate the expression balance of CD94/NKG2 receptors on exhausted CD8+ T cells, thereby restoring their antitumor immune function [290].

Compared to monotherapy, combined inhibition of the LAG‐3 and PD‐1 pathways demonstrates superior efficacy in enhancing antitumor immunity [291]. Nevertheless, key questions remain regarding the distinct signaling pathways of PD‐1 and LAG‐3. Although the broad repertoire of LAG‐3 ligands suggests multifaceted regulatory roles, the precise mechanisms and functional outcomes of its interactions with individual ligands remain incompletely elucidated.

Further research is warranted to refine targeted therapeutic approaches [292].

TIM‐3

TIM‐3 was initially identified as a marker and regulator of T cell exhaustion in the context of HIV‐1 infection. Subsequent studies have consistently demonstrated that TIM‐3 plays a critical role in mediating negative immune regulation across various chronic viral infections, including HBV, HCV, and Friend virus, as well as within the tumor microenvironment [293, 294]. Elevated expression of TIM‐3 on TILs is strongly associated with T cell exhaustion, thereby facilitating tumor immune evasion [295].

Mechanistically, the intracellular domain of TIM‐3 contains conserved tyrosine residues that can be phosphorylated by Src family tyrosine kinases, such as Lck and Fyn. This phosphorylation enables the recruitment of multiple SH2 domain‐containing signaling proteins, including PI3K, PLC‐γ1, and Ras‐GAP, leading to the activation of NFAT/AP‐1 and NF‐κB signaling pathways [296]. While this signaling cascade may transiently enhance T cell responses, sustained activation promotes terminal T cell differentiation into an exhausted phenotype [297].

Clinical evidence underscores the pivotal role of TIM‐3 in T cell dysfunction. For instance, in patients with HBV‐related HCC and prostate cancer, increased TIM‐3 expression on tumor antigen‐specific CD8+ T cells correlates positively with disease progression, suggesting its potential utility as a prognostic biomarker [298, 299]. Furthermore, emerging research has highlighted the involvement of TIM‐3 in Treg function.

Activated human TIM‐3+ FoxP3+ Tregs isolated from tumor tissues or expanded in vitro exhibit enhanced immunosuppressive activity compared to their TIM‐3− FoxP3+ counterparts and contribute to the exacerbation of effector T cell exhaustion, underscoring the broad inhibitory influence of TIM‐3 across diverse immune cell subsets [300].

The mechanism of TIM‐3 is highly complex, as it exerts diverse inhibitory effects on various immune cells through interactions with multiple ligands. For instance, the binding of TIM‐3 to galectin‐9 (Gal‐9) induces apoptosis in Th1 cells and CTLs [301]. Furthermore, its interaction with Gal‐9 or carcinoembryonic antigen‐related cell adhesion colecule 1 (CEACAM‐1) enhances the suppressive function of Tregs and impairs the effector functions of CD8+ T and Th1 cells, thereby contributing to T cell exhaustion within the TME and correlating closely with poor patient prognosis [302, 303].

Consequently, blockade of the TIM‐3 signaling pathway may alleviate its inhibitory effects on multiple immune cell populations, restore anti‐tumor immune responses, and facilitate the development of immunological memory [304].

Currently, a range of immunotherapeutic agents targeting TIM‐3 have advanced into phase I/II clinical trials. Notably, sabatolimab (MBG453), a high‐affinity humanized IgG4 monoclonal antibody, effectively inhibits the interaction of TIM‐3 with Gal‐9 and phosphatidylserine (PtdSer) [305, 306]. This agent demonstrates a distinctive dual “immune‐bone marrow” mechanism of action in acute myeloid leukemia (AML) and myelodysplastic syndromes (MDS), enabling both restoration of immune‐mediated clearance of leukemia stem cells and direct suppression of tumor cell proliferation in TIM‐3‐expressing malignancies [307].

Similarly, the anti‐TIM‐3 antibody M6903 has been shown to concurrently disrupt the binding of Gal‐9, PtdSer, and CEACAM‐1 to TIM‐3, thereby enhancing T cell functionality [308]. Beyond monoclonal antibodies, small molecule inhibitors such as ML‐T7 bind to the FG‐CC loop of TIM‐3, interfering with its interactions with PtdSer and CEACAM‐1; preclinical studies indicate that ML‐T7 significantly improves survival and enhances the antitumor activity of CD8+ CTLs [309]. Importantly, TIM‐3 is frequently co‐expressed with PD‐1 on TEX‐term cells, suggesting that combined blockade may yield synergistic antitumor effects.

A phase I/II clinical trial (NCT02608268) evaluating sabatolimab as monotherapy or in combination with the anti‐PD‐1 antibody spartalizumab in patients with advanced solid tumors demonstrated limited efficacy with monotherapy, but the combination regimen exhibited preliminary antitumor activity and was well tolerated. In conclusion, TIM‐3 represents a promising immune checkpoint target with significant potential to reverse T cell exhaustion and augment antitumor immunity. Targeted therapies against TIM‐3, particularly when combined with PD‐1 inhibition, may offer a novel therapeutic strategy for patients with diverse cancer types.

Transcription Factors

The fate of TEX cells is governed by a core hierarchical network of transcription factors that establishes and maintains a dysfunctional state through the remodeling of the epigenetic landscape (Figure 5).

TOX

As an intranuclear DNA‐binding protein, TOX plays a pivotal role in the epigenetic regulation of T cell exhaustion [52]. Under conditions of persistent antigen stimulation, the calcineurin‐NFAT signaling pathway becomes aberrantly activated, leading to sustained expression of both TOX and its homolog TOX2 [310]. Notably, in contrast to conventional T cell activation—which relies on the NFAT‐AP‐1 transcriptional complex—NFAT can independently initiate the TOX transcriptional program in the absence of AP‐1 within the tumor microenvironment [10].

Furthermore, TOX stabilizes the epigenetic and transcriptional landscape of TEX cells by preserving the expression of key lineage‐defining factors such as the progenitor marker TCF‐1 and the survival regulator Eomes, while also establishing a calcineurin‐independent autoregulatory feedback loop [8]. Importantly, a robust positive feedback mechanism exists between TOX and co‐inhibitory receptors, which exacerbates T cell dysfunction. On one hand, TOX directly upregulates the transcription of inhibitory receptors, including PD‐1 and LAG‐3 [10].

On the other hand, in models of HCC, TOX may enhance PD‐1 protein stability by interfering with its lysosomal degradation pathway [311]. Intriguingly, different immune checkpoint molecules exert distinct regulatory effects on TOX expression. Depletion of LAG‐3 results in a significantly greater reduction in TOX levels compared to PD‐1 depletion, indicating a dominant role for LAG‐3 in sustaining TOX expression [312].

This differential regulatory pattern has been corroborated in models of chronic viral infection, where LAG‐3 primarily supports TOX expression, whereas PD‐1 predominantly modulates the proliferation of TEX cells [24]. Moreover, TOX directly suppresses the expression of effector cytokines such as IFN‐γ, TNF, and IL‐21, thereby further impairing the anti‐tumor functionality of CD8+ T cells [312].

Targeting TOX has emerged as a promising strategy to reverse T cell dysfunction and enhance antitumor immune responses. Experimental evidence demonstrates that knockout of TOX in established TEX cells significantly reduces apoptosis resistance, downregulates the expression of inhibitory receptors such as PD‐1 and TIM‐3, and promotes a shift toward an effector‐like phenotype [70]. These findings provide a robust theoretical foundation for the development of TOX‐targeted inhibitors.

Furthermore, several studies indicate that modulating upstream signaling pathways of TOX may hold therapeutic potential; for instance, overexpression of basic leucine zipper ATF‐like transcription factor (BATF) or blockade of vascular endothelial growth factor (VEGF)/VEGFR2 signaling can suppress TOX expression and restore T cell functionality [313, 314]. Notably, therapeutic approaches targeting TOX face significant challenges. While TOX ablation can drive reprogramming of TEX cells into effector‐like states, its intrinsically disordered protein structure poses substantial obstacles to the design of direct small‐molecule inhibitors [315].

More critically, TOX plays a protective role in preventing T cell overactivation and activation‐induced cell death, highlighting its dual biological functions, which must be carefully considered in therapeutic contexts [316]. Future research should prioritize targeting TOX upstream regulatory networks or exploring synergistic combinations with existing immunotherapeutic modalities.

TCF‐1

TCF‐1 (encoded by Tcf7), a key effector molecule of the Wnt/β‐catenin signaling pathway, plays an essential role in maintaining TPEX [9, 317]. Unlike conventional effector or memory T cells, TCF‐1+ TPEX cells retain self‐renewal capacity and proliferative potential in chronically antigen‐stimulated environments, while avoiding terminal differentiation through epigenetic reprogramming [318, 319, 320]. Specifically, TCF‐1 regulates TPEX cell fate decisions via multi‐level mechanisms.

At the epigenetic level, TCF‐1 significantly influences global chromatin accessibility and the transcriptome in Tex by upregulating a set of transcriptional regulators associated with critical effector functions—such as FOXO1, ZEB2, ID3, and Eomes—and by suppressing the expression of Blimp‐1 and Bim [321]. In contrast, TCF‐1 inactivation is partially mediated by DNA methylation, as demonstrated in a rapid in vitro CD8+ T cell exhaustion system [76]. Genetic deletion of Tcf7 leads to a marked reduction in the frequency of TPEX cells and an increased viral load in a chronic LCMV infection model, underscoring its pivotal role in sustaining antiviral immune responses [322].

Collectively, these findings highlight TCF‐1 as a promising therapeutic target for enhancing both antiviral and antitumor immunity.

Clinical transformation studies have demonstrated that the expression level of TCF‐1 in tumor‐infiltrating CD8+ T cells is closely associated with the therapeutic efficacy of PD‐1 blockade. In melanoma patients, pretreatment intratumoral enrichment of TCF‐1+ CD8+ T cells has been significantly correlated with improved treatment response and prolonged survival, whereas the accumulation of CD39+ Tim‐3+ TEX‐term cells is indicative of treatment resistance [323]. These findings establish TCF‐1 as a promising biomarker for predicting immunotherapy outcomes.

Moreover, TCF‐1+ TPEX cells have emerged as a novel therapeutic target for enhancing the durability of antitumor immunity. In vitro studies indicate that the expansion and differentiation of stem cell‐like TCF‐1+ memory T cells can be promoted through the combined application of antioxidants, TCR stimulation, and pro‐inflammatory cytokines, offering a potential strategy to augment antitumor immune responses [317]. Notably, memory T cells in human lymph nodes exhibit strong phenotypic and functional similarities to TCF‐1+ CD8+ T cells observed in mouse models following checkpoint blockade, including high TCF‐1 expression, robust proliferative capacity, and self‐renewal potential [323].

This suggests their critical role as a therapeutic target in chronic infections and cancer. TCF‐1 plays a pivotal role in modulating T cell exhaustion during chronic infections and malignancies, as well as in sustaining immune responses under PD‐1 blockade. Consequently, targeting TCF‐1—either directly or through its upstream regulators or downstream effectors—alone or in combination with other therapeutic approaches, may enhance and prolong antitumor and anti‐infective immunity.

Nevertheless, the precise regulatory mechanisms governing TCF‐1 expression in CD8+ T cells, along with its dynamic interactions with other transcription factors, require further investigation to identify more effective therapeutic interventions.

NFAT

The NFAT family, serving as central effectors in the calcium signaling pathway, exhibits a distinctive duality in regulating T cell function [324, 325]. Under physiological conditions, TCR‐mediated calcium influx activates calcineurin, leading to NFAT dephosphorylation and subsequent nuclear translocation. Once in the nucleus, NFAT forms a transcriptional complex with AP‐1 to cooperatively drive the expression of effector cytokines such as IL‐2 and IFN‐γ [326].

However, under persistent antigen stimulation within the tumor microenvironment, the absence of costimulatory signals results in inadequate AP‐1 expression [69]. This condition promotes NFAT to bind to regulatory regions of exhaustion‐associated genes in an atypical monomeric form. This “cofactor‐deficient” mode of NFAT activation selectively enhances the transcriptional upregulation of immune checkpoint molecules, including PD‐1, TIM‐3, and LAG‐3, while simultaneously suppressing the expression of memory‐related genes [69].

Mechanistic investigations have revealed that NFAT family members exhibit both functional redundancy and distinct roles in the T cell exhaustion program. Combined deletion of NFAT1 and NFAT2 markedly reduces the expression levels of inhibitory receptors, whereas additional knockdown of NFAT4 nearly abolishes residual expression, indicating a cumulative contribution of individual NFAT members to the exhaustion network [69]. Furthermore, NFAT can cooperate with transcription factors such as IRF4 and BATF to reinforce the exhaustion‐associated epigenetic program by establishing specific patterns of chromatin accessibility [327].

Notably, the role of NFAT5, a member of the NFAT family, has garnered increasing attention in recent years with respect to T cell exhaustion. Unlike other NFAT family members, NFAT5 operates independently of calcineurin and lacks an AP‐1 binding site. Instead, NFAT5 is primarily activated by TCR signaling and metabolic stresses such as hyperosmolality within the tumor microenvironment.

Importantly, NFAT5 exerts minimal influence on T cell exhaustion associated with chronic infections, and its pro‐exhaustion effects are largely restricted to the tumor context. This indicates that NFAT5 plays a context‐specific regulatory role across different pathological conditions [328].

Based on these findings, the therapeutic strategy targeting NFAT is evolving from conventional calcineurin inhibition toward more precise and selective interventions. Emerging research directions include the development of small‐molecule compounds designed to specifically disrupt the binding of unchaperoned NFAT to target genes, as well as the restoration of normal NFAT/AP‐1 transcription complex function through epigenetic modulation. These approaches aim to selectively reverse T cell exhaustion while preserving essential immune functions, thereby offering novel insights for enhancing immunotherapy efficacy [329].

NR4A

The NR4A orphan nuclear receptor family, comprising NR4A1, NR4A2, and NR4A3, is markedly upregulated in TEX‐term cells. Its expression is directly induced by persistent calcium‐NFAT signaling and further amplified in the absence of AP‐1 activity [330]. Similar to TOX, NR4A contributes to the establishment and maintenance of the T cell exhaustion phenotype through epigenetic remodeling [11, 331].

Specifically, NR4A mediates the formation of repressive chromatin states by recruiting histone‐modifying complexes to the promoter regions of effector genes, while simultaneously increasing chromatin accessibility at exhaustion‐associated gene loci. Notably, NR4A and the HMG‐box transcription factors TOX/TOX2 engage in a positive regulatory feedback loop, mutually enhancing each other's expression to establish a stable transcriptional network. This self‐reinforcing mechanism is especially pronounced in TEX‐term cells and may contribute to the limited efficacy of conventional ICIs in reversing this state.

In preclinical CAR‐T cell models, genetic ablation of NR4A not only substantially reduces the expression of inhibitory receptors such as PD‐1 and TIM‐3 but also enhances cytokine production and in vivo persistence, underscoring its pivotal role in driving T cell dysfunction [10].

Given the pivotal role of NR4A in T cell exhaustion, therapeutic strategies targeting the NR4A family hold considerable promise. Current research efforts are focused on developing small‐molecule inhibitors to suppress NR4A transcriptional activity and employing epigenetic approaches to disrupt its interactions with downstream effector molecules. However, given NR4A's essential role in T cell homeostasis, these physiological functions must be carefully considered [332].

Future studies should aim to elucidate strategies that achieve a precise balance between reversing T cell exhaustion and preserving immune homeostasis.

Other Key Transcription Factors

In addition to TOX, TCF‐1, NFAT, and NR4A, recent studies have revealed that a variety of key transcription factors also play critical roles in the initiation and maintenance of T cell exhaustion. Given the high complexity of their mechanisms and functions, and due to space limitations, this review provides only a brief overview.

Interferon regulatory factor 4 (IRF4) and BATF are key transcription factors involved in TCR signaling. Their expression is significantly elevated in chronic infectious and neoplastic conditions. IRF4 promotes the expression of the inhibitory receptor PD‐1 and suppresses TCF‐1, thereby contributing to metabolic dysregulation and functional exhaustion of TEX cells.

Reduction of IRF4 expression has been shown to partially restore T cell function and facilitate the development of memory‐like T cells, highlighting its central role in the transcriptional program of T cell exhaustion [100, 327]. BATF acts in concert with IRF4 to modulate the phenotype and functional state of TEX cells. Upregulation of BATF is closely associated with the downregulation of AP‐1 family transcription factors.

Notably, BATF overexpression can reverse the TEX cell phenotype and enhance antitumor immunity, whereas BATF depletion favors the generation of memory T cells, indicating a bidirectional regulatory mechanism between T cell exhaustion and effector function [314].

As a transcriptional repressor, Blimp‐1 (PRDM1) promotes the upregulation of inhibitory receptors and suppresses memory T cell differentiation during chronic infection, with its expression level playing a critical role in regulating the balance between effector function and exhaustion in TEX cells [333]. Deficiency of Blimp‐1 reduces the expression of inhibitory receptors such as PD‐1 and can partially restore T cell effector function; however, it may also compromise the persistence of antiviral and antitumor immune responses [334].

As classical regulators of T cell differentiation, EOMES and T‐bet also play critical roles in the differentiation of TEX cell subsets. The TEX cell subset characterized by high T‐bet expression exhibits enhanced proliferative capacity and responsiveness, whereas elevated EOMES expression is associated with a terminally exhausted phenotype. The dynamic balance between these two transcription factors determines the differentiation trajectory and functional state of TEX cells [122, 335].

In addition, members of the IKZF family—specifically Ikaros (IKZF1), Helios (IKZF2), and Aiolos (IKZF3)—have recently been identified as key regulators of TEX cell differentiation and functional maintenance. Dual knockdown of IKZF1 and IKZF3 significantly enhances antitumor and antiviral effector functions in TEX cells, highlighting their critical role in modulating the expansion and functional capacity of the TEX cell subpopulation [336]. Concurrently, the NRF2 signaling pathway influences TEX cells metabolism and differentiation by regulating the oxidative stress response and downstream PTGIR receptor expression, revealing a novel mechanism underlying the metabolic‐transcriptional regulatory axis in TEX cells development [337, 338].

The transcriptional regulatory network governing TEX cells is highly complex, with the coordinated interplay—including both synergistic and antagonistic interactions—of multiple transcription factors collectively determining cell fate and functional states. A comprehensive understanding of this regulatory machinery provides a crucial theoretical foundation for optimizing immunotherapeutic strategies.

Metabolic Markers

The functional state of T cells is closely linked to their metabolic health, with specific metabolic enzymes and their products now serving as key indicators for defining the exhaustion state.

CD39/CD73

The purinergic signaling pathway, comprising CD39 (ENTPD1) and its downstream effector CD73, plays a pivotal role in the metabolic regulation of T cell exhaustion. This pathway is highly expressed on terminally differentiated exhausted CD8+ T cells and mediates the sequential hydrolysis of extracellular ATP and ADP to AMP, which is subsequently converted by CD73 into adenosine—a potent immunosuppressive molecule. This process establishes an adenosine‐rich immunosuppressive network within the tumor microenvironment [339, 340].

Functionally, CD39hi CD8+ T cells exhibit a characteristic exhausted phenotype, marked by significantly reduced secretion of effector cytokines such as IFN‐γ and TNF‐α, along with co‐upregulation of inhibitory receptors PD‐1 and TIM‐3 [340]. This metabolic‐immune crosstalk not only directly compromises T cell antitumor activity but also leads to sustained activation of the cAMP‐PKA signaling cascade via the adenosine A2a receptor (A2aR) axis, thereby suppressing TCR signaling and impairing cell cycle progression [341]. Notably, recent studies indicate that adenosine‐mediated immunosuppression can also involve additional pathways, such as inhibition of the mTORC1 pathway and modulation of MAP kinase signaling, further dampening T cell metabolic fitness and effector function [342, 343].

Moreover, CD39/CD73+ T cells often co‐express immune checkpoints like PD‐1, and adenosine signaling may synergize with these pathways to enhance immunosuppression [344, 345]. Hypoxia within the TME enhances CD39 expression through an HIF‐1α‐dependent mechanism while concurrently inducing mitochondrial metabolic dysfunction, thus creating a self‐reinforcing cycle of immunometabolic suppression [346]. Preclinical evidence indicates that targeting the CD39/CD73‐adenosine axis can effectively reverse T cell exhaustion and synergize with ICB therapies [347].

In particular, when combined with chemotherapeutic agents such as oxaliplatin, this strategy demonstrates robust potential to remodel TIME, offering a promising metabolic intervention to overcome resistance to cancer immunotherapy [348].

IDO1

Indoleamine 2,3‐dioxygenase 1 (IDO1), a key rate‐limiting enzyme in the tryptophan‐kynurenine metabolic pathway, plays a critical role in establishing an immunosuppressive microenvironment [349]. It promotes T cell exhaustion through two synergistic mechanisms: first, by depleting tryptophan—an essential amino acid—within the microenvironment, IDO1 induces amino acid starvation stress, activates the general control nonderepressible 2 (GCN2) kinase signaling pathway, and consequently leads to T cell proliferation arrest and functional impairment [350]. Second, its catalytic product, kynurenine, directly modulates exhaustion‐associated transcriptional programs by activating the AhR signaling axis, thereby promoting Treg differentiation and recruitment of MDSCs [351, 352].

Mechanistically, IDO1‐mediated metabolic reprogramming is closely intertwined with epigenetic regulation. AhR not only upregulates the expression of inhibitory receptors such as PD‐1 and TIM‐3 but also alters chromatin accessibility in T cells via histone modifications, thus stabilizing the exhausted phenotype [353, 354]. Notably, studies in glioblastoma models have shown that while IDO1 knockdown transiently restores T cell function, it simultaneously triggers compensatory upregulation of IDO1 in other immune cells, underscoring the complexity of this regulatory network [355].

Based on current mechanistic insights, the combination of IDO1 inhibitors with radiotherapy and ICIs holds significant therapeutic potential [22]. Clinical studies in renal cell carcinoma have demonstrated synergistic antitumor effects with the combination of pembrolizumab and epacadostat [356]. However, the variability observed across clinical trials indicates that the efficacy of IDO1 inhibitors may be influenced by multiple factors, including tumor type, microbiome composition, and host metabolic status.

Therefore, future efforts should focus on developing more precise biomarker‐guided strategies for patient stratification [357, 358, 359, 360].

Other Metabolic Markers

In the study of T cell exhaustion, beyond the classical CD39/CD73 and IDO1 pathways, accumulating experimental evidence in recent years has identified additional key metabolic markers and elucidated their underlying mechanisms. Dysregulation of lactate metabolism represents a critical characteristic of T cell exhaustion within the tumor microenvironment. TEX‐term cells specifically exhibit high expression of MCT11, which facilitates lactate uptake, resulting in intracellular acidification and metabolic suppression, thereby exacerbating T cell exhaustion.

Conditional deletion of MCT11 or its blockade via antibodies has been shown to significantly reduce lactate influx, restore T cell effector function, and suppress tumor growth in preclinical tumor models, highlighting MCT11 as a specific metabolic marker and a promising therapeutic target for TEX cells [103]. Furthermore, dysregulated cholesterol metabolism has been firmly linked to T cell exhaustion. Cholesterol accumulation in tumor tissues enhances the expression of immune checkpoint molecules—including PD‐1, 2B4, TIM‐3, and LAG‐3—on CD8+ T cells, promoting an exhausted phenotype.

Mechanistically, cholesterol induces transcription of immune checkpoint genes by augmenting endoplasmic reticulum stress and activating the XBP1 signaling pathway. Reducing cholesterol levels in T cells or inhibiting XBP1 signaling effectively restores their antitumor functionality [361].

Reprogramming of methionine metabolism represents another critical mechanism underlying tumor‐induced T cell exhaustion. Tumor cells directly promote T cell exhaustion by enhancing the methionine cycle and accumulating high levels of SAM and MTA. CRISPR‐Cas9‐mediated knockout of MAT2A, the enzyme responsible for SAM synthesis, significantly attenuates T cell exhaustion and delays tumor progression, indicating that molecules involved in methionine metabolism serve as novel metabolic markers of T cell exhaustion [20].

Moreover, levels of L‐2‐hydroxyglutarate (L‐2‐HG), an immunomodulatory metabolite, are markedly reduced in TEX cells. Exogenous supplementation of L‐2‐HG has been shown to improve mitochondrial function, reverse epigenetic repression, promote memory T cell differentiation, and enhance antitumor immunity, highlighting its regulatory role in T cell exhaustion and its potential as a metabolic intervention target [90].

Mitochondrial dysfunction and oxidative stress represent additional key metabolic features of T cell exhaustion. TEX cells exhibit loss of mitochondrial membrane potential, impaired respiratory chain activity, and accumulation of ROS, leading to insufficient energy production and cellular dysfunction [95, 362]. Downregulation of regulators of mitochondrial biogenesis, such as PGC‐1α, is closely associated with the exhausted phenotype, and supplementation with NAD+ precursors or enhancement of mitochondrial function has been shown to partially restore T cell functionality [101].

Furthermore, dysregulated arginine metabolism leads to upregulation of ARG1, resulting in arginine depletion within T cells, inhibition of the mTOR‐T‐bet signaling axis, and promotion of the exhausted phenotype. Modulation of arginine metabolism can improve the metabolic fitness of T cells and enhance their antitumor efficacy [363].

Notably, the functional roles of certain metabolic markers remain controversial. For instance, fatty acid metabolism exhibits a dual role in T cell exhaustion: it can support T cell survival by providing energy through β‐oxidation, yet excessive accumulation may lead to lipid peroxidation and cellular dysfunction. Specifically, molecules such as fatty acid binding protein 5 (FABP5) promote fatty acid uptake and oxidation, enabling T cells to adapt to the glucose‐deprived tumor microenvironment.

However, sustained or excessive fatty acid uptake and oxidation can result in lipid peroxidation, mitochondrial damage, and impaired T cell function [364]. Moreover, the interplay between metabolic reprogramming and epigenetic regulation or signaling pathways has not been fully elucidated. Future studies should aim to comprehensively characterize the dynamic changes and underlying molecular mechanisms of metabolic markers during the induction, maintenance, and reversal of T cell exhaustion, thereby informing more precise immunometabolic intervention strategies.

Immunotherapy Strategies Targeting T Cell Exhaustion

The advent of cancer immunotherapy, particularly ICB and ACT, has profoundly transformed the treatment landscape for numerous advanced malignancies by restoring the antitumor activity of CTLs [365]. Nevertheless, clinical efficacy varies significantly across patient populations, and long‐term response rates remain suboptimal. T cell exhaustion is widely recognized as a central biological barrier contributing to this therapeutic challenge [366].

Within the TME, persistent antigen exposure, the synergistic effects of multiple immunosuppressive signals, and metabolic stress collectively drive T cells toward progressive functional impairment and epigenetic reprogramming, ultimately leading to the differentiation of TEX‐term cells, which have severely diminished effector capacity [161]. This dysfunctional state not only compromises the cytolytic potential of T cells but also confers resistance to both ICB and ACT, thereby representing a key mechanism underlying immunotherapy failure and disease relapse [3, 367].

Given the pivotal role of T cell exhaustion in promoting immune evasion, the development of therapeutic approaches capable of precisely modulating or reversing this state has become a critical priority in overcoming current limitations in cancer immunotherapy. The research paradigm is shifting from monotherapy‐based ICB toward mechanism‐driven, multidimensional combinatorial strategies [3, 368, 369]. As previously outlined, these emerging strategies aim to enhance the efficacy of existing ICB and ACT regimens by targeting co‐inhibitory receptor networks on T cell exhaustion (e.g., PD‐1, LAG‐3, TIGIT), reprogramming intrinsic transcriptional and epigenetic regulatory circuits (e.g., TOX and TCF‐1), or remodeling the metabolic environment of the TME.

This section provides a systematic and critical evaluation of the most recent therapeutic approaches directed at T cell exhaustion, with emphasis on their mechanisms of action, potential for synergistic integration, and translational prospects, aiming to establish a conceptual framework and strategic guidance for the design of next‐generation personalized combination immunotherapies (Figure 6 and Table 2).

ICB Therapy

ICB therapy inhibits immune checkpoint molecules and their ligands, restoring T cell function and suppressing tumors [370]. However, as a monotherapy, ICB exhibits limited efficacy, with objective response rates below 30% across multiple tumor types [371]. To better reverse T cell exhaustion, researchers are actively exploring combinatorial strategies that integrate ICB with other therapeutic modalities.

Combination with Immunomodulatory Factors

The initiation and persistence of T cell exhaustion are closely associated with a broad spectrum of immunoregulatory factors, including immunosuppressive cytokines (e.g., TGF‐β and IL‐10), immune checkpoint receptors (e.g., PD‐1, CTLA‐4, LAG‐3, and TIM‐3), and metabolic regulatory enzymes (e.g., IDO and ARG1). These factors contribute to T cell dysfunction by inhibiting T cell activation, amplifying suppressive signaling pathways, and disrupting metabolic homeostasis within the tumor microenvironment.

Cytokines play a central role in regulating T cell differentiation, functional maintenance, and immune microenvironment remodeling. The combination of cytokine‐targeted approaches with ICB has emerged as a key strategy for overcoming T cell exhaustion [372]. For example, IFN‐I drives CD8+ TILs toward an exhausted phenotype, a process mediated by chronic IFN‐I signaling that disrupts lipid metabolism and redox balance, exacerbates lipid peroxidation, and promotes T cell dysfunction.

Clinical evidence indicates that elevated IFN‐I levels are strongly associated with reduced ICB efficacy [373]. Furthermore, Poly(ADP‐ribose) polymerase 14 (PARP14), a critical mediator of IFN‐γ‐driven immune escape, not only facilitates terminal exhaustion of CTLs but also regulates the polarization of TAMs. In preclinical models, inhibition of PARP14 in combination with PD‐1 blockade significantly enhances antitumor immune responses and ameliorates the immunosuppressive tumor microenvironment [374].

Interleukin‐mediated regulation of T cell exhaustion is closely linked to immunotherapy outcomes. IL‐2 promotes the differentiation of TCF‐1+ stem‐like CD8+ T cells into effector phenotypes, thereby enhancing the antitumor efficacy of PD‐1 blockade [375]. Joonbeom Bae et al. further demonstrated that engineered mesenchymal stem cells (MSCs) delivering an IL‐2 mutant protein can effectively reverse CD8+ T cell exhaustion, highlighting the therapeutic potential of IL‐2‐based interventions [376].

IL‐6 contributes to CD8+ T cell dysfunction by upregulating exhaustion markers such as PD‐1. Dual blockade of IL‐6 and PD‐1 has been shown to significantly enhance the cytotoxic function of TILs and suppress tumor growth in vivo [377]. Thus, distinct cytokines exert diverse and context‐dependent effects on T cell exhaustion, and targeted modulation of specific cytokine signaling pathways enables multi‐level intervention in this process, thereby improving the efficacy of ICB.

Lipopolysaccharide (LPS) also exhibits a dual role in modulating T cell function. On one hand, LPS promotes T cell infiltration in pancreatic ductal adenocarcinoma (PDAC); on the other hand, it upregulates PD‐L1 expression via the TLR4/MyD88/AKT/NF‐κB signaling pathway, leading to impaired T cell function or induction of exhaustion. Based on these mechanistic insights, combining LPS with PD‐L1 blockade represents a promising integrative therapeutic strategy for PDAC [378].

Moreover, direct targeting of inhibitory molecules associated with T cell exhaustion offers novel opportunities to improve durable responses to immunotherapy. P‐selectin glycoprotein ligand‐1 (PSGL‐1) is a key inhibitory receptor that sustains the exhausted state of T cells and limits the functional recovery of TPEX cells. Rienk Offringa et al. demonstrated that targeting PSGL‐1 not only restores effector function in TEX cells but also acts synergistically with ICB, significantly enhancing T cell reactivation and providing a promising avenue for cancer immunotherapy [379].

Similarly, B7‐H3 impairs ICB efficacy by promoting CD8+ T cell exhaustion and reshaping the immunosuppressive TME across multiple cancer types, suggesting that co‐targeting B7‐H3 may represent a viable strategy to overcome limitations imposed by T cell exhaustion in current immunotherapies [380].

Targeting Tumor‐Related Signaling Pathway Regulation

Targeting specific signaling pathways to modulate T cell activation, proliferation, and differentiation has emerged as a pivotal strategy for enhancing the efficacy of ICB. In mismatch repair‐deficient (dMMR) tumor models, the CDK4/6 inhibitor abemaciclib significantly prolongs survival, increases the CD4+ to CD8+ T cell ratio among TILs, reduces expression levels of T cell exhaustion markers, and diminishes Treg populations. These findings reveal novel therapeutic opportunities for patients resistant to ICB or ineligible for monotherapy [381].

In tumors with low neoantigen burden and poor responsiveness to ICB, transient inhibition of SHP‐1 lowers the TCR activation threshold, markedly enhances the effector function and tumor infiltration of low‐affinity, endogenous autoantigen‐specific CD8+ T cells, and effectively reverses the exhausted phenotype [382].

Concurrently, inhibition of the DNA damage response pathway has been shown to augment T cell‐mediated immunity. In hepatocellular carcinoma models, the ATR inhibitor AZD6738 remodels the immunosuppressive TME via activation of the cGAS/STING pathway, reduces the frequency of Tregs and TEX cells, and synergizes with PD‐L1 blockade to enhance the durability of antitumor immune memory [383]. In metastatic renal cell carcinoma, 5α‐reductase inhibitors improve responses to ICB by reducing CD8+ T cell exhaustion and PD‐1 expression through modulation of androgen receptor signaling, indicating that androgen signaling represents a promising target for combination immunotherapy [384].

Furthermore, semaphorin 4A (Sema4A) enhances ICB efficacy by promoting the proliferation and cytotoxic activity of tumor‐specific CD8+ T cells while suppressing terminal exhaustion, highlighting its potential as both a predictive biomarker and a therapeutic target in combinatorial immunotherapy strategies [385]. Notably, IGF‐Trap, an inhibitor of type I insulin‐like growth factor receptor (IGF‐IR) signaling, effectively alleviates the local immunosuppressive microenvironment in the liver, reduces recruitment of MDSCs, and reverses T cell exhaustion, with its therapeutic efficacy further enhanced when combined with PD‐1 inhibitors [386].

Although targeting tumor‐associated signaling pathways has demonstrated promise in improving T cell function, mitigating exhaustion, and enhancing the efficacy of ICB across multiple preclinical models, significant challenges remain. First, signaling dependencies vary considerably across tumor types, limiting the broad applicability of single‐target interventions. Second, these approaches may induce immune‐related adverse events or disrupt physiological functions in non‐tumor tissues; for example, inhibition of SHP‐1 or IGF‐IR can perturb normal immune homeostasis or endocrine system activity.

Therefore, future research should prioritize multi‐target combination therapies, dose optimization, and biomarker‐guided patient stratification to balance therapeutic efficacy with safety and to ensure clinical feasibility across diverse tumor types.

Metabolic and Epigenetic Interventions

Metabolic reprogramming is a key mechanism underlying tumor immune escape. In recent years, targeting tumor metabolic pathways has emerged as a promising strategy to reverse T cell exhaustion and enhance the efficacy of ICB. In tumors exhibiting a preference for aerobic glycolysis, cancer cells efficiently compete for glucose and secrete lactate, thereby exacerbating immunosuppression and promoting the exhausted phenotype of CTLs.

Citrate serves as a critical metabolic regulator. Preclinical studies have demonstrated that high‐dose sodium citrate (SCT) can restore glucose availability and inhibit lactate accumulation in the TME by suppressing both tumor glycolysis and oxidative metabolism, consequently enhancing T cell functionality and driving the polarization of TAMs toward a pro‐inflammatory phenotype. The combination of ICIs with SCT therapy represents a promising approach to alleviate the immunosuppressive microenvironment at the metabolic level and improve immunotherapy response rates in solid tumors [387].

Moreover, mitochondrial function plays a crucial role in maintaining CD8+ T cell effector capacity and reversing exhaustion. The application of optogenetic technology, termed “OptoMito‐On”, enables remote enhancement of mitochondrial ATP production and improvement of mitochondrial membrane potential in CD8+ T cells, leading to increased migratory capacity and cytotoxic activity, effective reversal of the exhausted state, and augmented antitumor immune responses [388]. Thus, metabolic intervention constitutes a novel synergistic strategy to ameliorate the immunosuppressive TME and enhance ICB efficacy through modulation of glycolysis, oxidative metabolism, and mitochondrial function, demonstrating significant potential for combination therapy in solid tumors.

Recent studies have shown that epigenetic regulation is an effective strategy for improving the efficacy of ICB. The histone demethylase lysine‐specific demethylase 1 (LSD1) establishes a distinct epigenetic program in progenitor‐exhausted CD8+ T cells that antagonizes TCF‐1‐mediated stemness maintenance and drives their terminal differentiation. Genetic ablation or pharmacological inhibition of LSD1 significantly enhances the persistence of progenitor‐exhausted CD8+ T cells, thereby sustaining a pool of terminally differentiated effector cells with enhanced tumoricidal capacity and enabling more durable and robust responses to anti‐PD‐1 therapy [389].

However, although LSD1 inhibition promotes CD8+ T cell tumor infiltration, its cytotoxic potential is limited by the induction of TGF‐β expression, which compromises antitumor immunity. Recent evidence indicates that combining LSD1 inhibition with TGF‐β blockade and PD‐1 antibody treatment significantly enhances CD8+ T cell infiltration and cytotoxic function, reverses the exhausted phenotype, and achieves tumor eradication and long‐term immune protection in poorly immunogenic tumor models. These findings suggest that this triple‐combination regimen holds substantial promise for overcoming T cell exhaustion and improving the therapeutic efficacy of ICB [390].

Targeted Delivery

Targeted delivery strategies enable the precise localization of ICIs and related therapeutic agents to the TME, thereby enhancing T cell activity, mitigating T cell exhaustion, and improving the efficacy of combination immunotherapy. A variety of delivery platforms are currently under development, including viral vectors, engineered extracellular vesicles (EVs), and targeted systems for small molecules and siRNA, offering innovative approaches to overcome tumor‐induced immunosuppression.

The integration of viral vectors with ICB has demonstrated significant synergistic effects. For example, combining anti‐LAG‐3 therapy with cowpea mosaic virus‐based immunotherapy (CPMV IT) reduces T cell exhaustion and enhances local immune activation, providing new insights into improving cancer immunotherapy outcomes [391]. Similarly, a tumor‐selective oncolytic herpesvirus vector (YST‐OVH) enables site‐specific delivery of PD‐1 inhibitors and synergizes with other checkpoint blockades [392].

Beyond viruses, engineered EVs offer another promising platform. Zhang et al. designed EVs to overexpress PD‐1 protein and co‐administered them with CTLs, achieving dual functional benefits: not only blocking PD‐1/PD‐L1‐mediated immune evasion but also inducing tumor cell apoptosis via factor‐related apoptosis ligand (FasL) and granzyme B (GzmB). This strategy leverages the intrinsic trafficking capacity of EVs to deliver immunomodulatory agents directly to the TME, simultaneously activating effector T cells while ensuring high target specificity and enhanced immune stimulation, thus representing a compelling approach for reversing T cell exhaustion.

Targeted siRNA delivery has demonstrated unique advantages in treating tumors at anatomically challenging sites. In glioblastoma, for instance, EV‐loaded siRNA can specifically target PD‐L1 on tumor cell surfaces, cross the blood‐brain barrier, and achieve complete blockade of the PD‐1/PD‐L1 axis, thereby enhancing T cell cytotoxicity and reversing the exhausted phenotype [393]. This approach addresses key limitations of conventional drug delivery systems and expands the applicability of immunotherapy to tumors residing in immunologically hostile or physically inaccessible microenvironments.

Dual ICIs Strategy

T cells exhibit distinct functional states and differentiation trajectories across different stages of exhaustion. Therefore, the rational design of ICI combinations tailored to specific phases of exhaustion enables precise therapeutic intervention and optimization of clinical outcomes. In the early stage of T cell hyporesponsiveness, dual blockade of PD‐1 and CTLA‐4 effectively prevents the upregulation of multiple inhibitory receptors, limits the progression of T cell dysfunction, and preserves effector potential [394].

During the critical transition from effector to exhausted state, short‐term ICI intervention can downregulate inhibitory receptor expression and promote functional recovery of effector Th1 and CD8+ T cells. For example, in a peritoneal dissemination model of gastric cancer, while anti‐PD‐1 and anti‐CTLA‐4 therapy enhances CD8+ T cell tumor infiltration, resistance is frequently associated with aberrant activation of the JAK‐STAT pathway and accumulation of immunosuppressive cell populations. The addition of Janus kinase inhibitors restores CD8+ T cell function, remodels the TME, and significantly improves therapeutic efficacy [395].

Moreover, studies in melanoma patients have demonstrated that anti‐CTLA‐4 promotes the expansion of progenitor T cells, and when combined with anti‐PD‐1, drives clonal expansion of CD8+ T cells, thereby enhancing clinical response rates [396]. Interventions during this phase aim to selectively reactivate partially dysfunctional but not their TEX‐term cells, thereby enhancing downstream antitumor immunity.

In the terminal exhaustion phase, T cell function is severely compromised and exhibits limited responsiveness to monotherapy with ICIs. A neoadjuvant study in patients with head and neck squamous cell carcinoma showed that the combination of anti‐PD‐1 and anti‐LAG‐3 antibodies was more effective than CTLA‐4 blockade in reprogramming the transcriptional profile of CD8+ TILs, promoting their conversion from an exhausted phenotype toward effector or tissue‐resident memory T cell states. This dual regimen also enhanced TCR diversity and facilitated clonal expansion [397].

Combining ICIs with Other Therapies

A variety of therapeutic modalities, including chemotherapy, radiotherapy, nano‐immunomodulators, and cancer vaccines, offer promising combinatorial approaches for treating drug‐resistant and poorly immunogenic tumors by improving the TME, enhancing antigen presentation, and activating T cell effector functions.

The combination of radiotherapy and ICB demonstrates significant immunostimulatory effects. Key mechanisms include the induction of damage‐associated molecular patterns (DAMPs) and chemokine release, promotion of DC and CTL infiltration, reversal of T cell exhaustion, conversion of “cold” to “hot” tumor phenotypes, and enhancement of ICB responsiveness [398]. A randomized phase I trial showed that in patients with immunologically “cold” tumors and high tumor aneuploidy, concurrent administration of radiotherapy and ICB resulted in significantly higher local and distant response rates and prolonged overall survival compared to sequential therapy, an effect strongly associated with enhanced antigen presentation [399, 400].

Similarly, Philip J. Saylor et al. demonstrated that radium‐223 radionuclide therapy effectively alleviates T cell exhaustion in metastatic prostate cancer by promoting intratumoral CD8+ T cell proliferation and infiltration, with a marked survival benefit when combined with anti‐PD‐1 therapy [401]. Collectively, radiotherapy and targeted radionuclide therapies represent robust strategies to overcome T cell dysfunction through multi‐level activation of antitumor immunity.

Accumulating evidence indicates that conventional chemotherapy exerts not only direct cytotoxic effects but also enhances the efficacy of ICB by remodeling TIME. Recent single‐cell transcriptomic and trajectory analyses using CITE‐seq revealed that neoadjuvant low‐dose metronomic chemotherapy (MCT) induces dynamic reprogramming of myeloid cell subsets in TNBC, thereby establishing an immunologically favorable microenvironment conducive to improved immunotherapy outcomes [402]. In colorectal cancer models, FOLFOX chemotherapy promotes the transition of CD8+ TILs to a high‐responsiveness phenotype characterized by reduced expression of PD‐1 and TIM‐3, thereby increasing their sensitivity to ICB.

The combination of FOLFOX with anti‐PD‐1 therapy significantly suppresses tumor growth [403]. Furthermore, the paclitaxel‐loaded nano‐immunomodulator Nano‐PI, when administered in conjunction with anti‐PD‐1, enhances T cell activity, effectively inhibits breast cancer metastasis, and achieves durable responses [404]. Together, these findings highlight chemotherapy as a viable strategy for synergizing with ICB through immune microenvironment modulation and functional reprogramming of T cells.

Vaccines and costimulatory agonists can further enhance T cell function and antitumor responses. For example, the λ phage vaccine targeting the tumor‐associated antigen ASPH, when combined with PD‐1 blockade, synergistically improves TIME in TNBC and hepatocellular carcinoma by activating CD4+ Th1 cells and CD8+ CTLs, promoting the formation of tertiary lymphoid structures, and reducing Treg infiltration. This combination significantly inhibits tumor progression and prolongs survival [405].

Intratumoral dual CD40‐TLR4 stimulation reverses irreversible T cell exhaustion resistant to PD‐1 blockade by eliminating exhausted PD‐1hi T cells and inducing systemic antitumor CD8+ T cell responses, thereby restoring sensitivity to immunotherapy [406]. The novel bifunctional agonist STAR0602 targets specific Vβ TCR subpopulations and cis‐activates IL‐2 signaling, significantly expanding CD4+ and CD8+ T cell populations exhibiting atypical effector memory phenotypes. This overcomes the constraints of T cell exhaustion and demonstrates durable antitumor effects across multiple PD‐1–refractory solid tumor models [407].

Vaccines and costimulatory agonists offer novel pathways for immune reactivation in drug‐resistant tumors through direct activation or expansion of T cell populations.

Adoptive Cell Therapy

ACT enhances antitumor immune responses by expanding TILs or peripheral blood‐derived immune cells ex vivo and reinfusing them into patients [408]. Similar to ICB therapy, the efficacy of ACT relies on CD8+ T cell‐mediated antitumor immunity within the TME. Consequently, T cell exhaustion not only impairs anticancer immune responses but also constitutes a direct mechanism of resistance to ACT.

To overcome this limitation, various strategies have been developed, among which genetic engineering has emerged as a pivotal approach. Genetic engineering can partially restore T cell proliferation and effector function by modulating exhaustion‐associated transcription factors, immune checkpoints, or epigenetic programs, thereby offering a promising strategy to enhance the potency of ACT [409] (Table 3). It should be noted that such interventions may induce adverse effects, including aberrant T cell proliferation.

For instance, biallelic deletion of ten‐eleven translocation 2 (TET2) in combination with forced expression of BATF3 can lead to antigen‐independent clonal expansion of CAR‐T cells, posing potential safety risks [410]. In light of these challenges, this section presents a systematic review of therapeutic strategies aimed at reversing T cell exhaustion, with a focus on recent advances in three representative ACT modalities—CAR‐T, TCR‐T, and TILs therapy—providing a theoretical foundation and practical guidance for optimizing cellular immunotherapy (Figure 7).

CAR‐T Therapy

Among various ACT, CAR‐T therapy has demonstrated considerable potential due to its ability to specifically recognize tumor antigens and its MHC‐independent mechanism of action. By genetically engineering T cells to express chimeric antigen receptors (CARs), CAR‐T cells can specifically target and eliminate malignant cells, thereby mediating potent antitumor activity [437, 438]. However, approximately 75% of patients experience disease recurrence following treatment, a phenomenon closely associated with tumor antigen escape mechanisms as well as the limited persistence and functional durability of CAR‐T cells in vivo [240, 439].

Clinical evidence has clearly established that T cell exhaustion is a key factor limiting the efficacy of CAR‐T therapy. In a study of 24 patients with large B‐cell lymphoma (LBCL) treated with axicabtagene ciloleucel (Yescarta), the proportion of memory CD8+ T cells within the infused CAR‐T product was significantly higher in patients achieving sustained complete response (CR) compared to those with partial response (PR) or progressive disease (PD). Conversely, the frequency of exhausted CD8+ T cells showed an inverse correlation [440].

Another clinical study involving 32 patients with LBCL demonstrated that memory‐like CD8+ T cell populations were significantly associated with response to tisagenlecleucel (Kymriah), but not to axicabtagene ciloleucel [441]. These findings indicate that the differentiation status of T cells in CAR‐T products—particularly the balance between exhausted and memory T cell subsets—is a critical determinant of clinical outcomes.

Accumulating evidence suggests that the antitumor activity of CAR‐T cells is regulated by multiple molecular mechanisms, including transcriptional regulatory networks, chronic antigen stimulation, and antigen‐independent signaling pathways. Moreover, the degree of T cell exhaustion is inversely correlated with the proportion of memory T cells [3]. A comprehensive understanding of these regulatory mechanisms will provide a robust theoretical foundation for the rational design of next‐generation CAR‐T products and for developing effective strategies to mitigate T cell exhaustion.

CAR‐T Cell Type and Differentiation Status

The type and differentiation status of T cells are critical determinants of the efficacy of CAR‐T therapy. Accumulating evidence indicates that although Naïve T cells and T stem cell memory (TSCM)‐enriched CAR‐T products exhibit relatively limited cytotoxic activity in vitro, they demonstrate enhanced antitumor effects, greater proliferative capacity, and improved persistence in vivo, along with reduced T cell exhaustion and a more pronounced memory phenotype. Compared to unselected bulk T cell products, TN/SCM‐derived CAR‐T cells are associated with a lower risk of cytokine release syndrome (CRS), exhibit attenuated T cell activation, and display a higher therapeutic safety index [442].

Furthermore, co‐expression of IL‐15 with GD2‐targeting CAR effectively preserves the central memory and stem cell‐like phenotypes and reduces PD‐1 expression, thereby delaying functional exhaustion [443, 444].

The superior performance of less differentiated T cell subsets in CAR‐T therapy has been consistently demonstrated. CD19‐directed CAR‐T cells enriched for TSCM—the most primitive T cell subset with self‐renewal capacity—exhibit enhanced proliferative activity in vitro, as well as prolonged persistence and superior tumor control in vivo [445, 446, 447]. Cong He et al. reported that NKG2D‐IL7 CAR‐T cells promote CD8+ T cell proliferation and upregulate B‐cell lymphoma 2 (Bcl‐2) expression, thereby reducing apoptosis and mitigating functional exhaustion—a mechanism closely linked to the enrichment of poorly differentiated T cell phenotypes [417].

Since CAR‐T cell exhaustion predominantly occurs in terminally differentiated effector T cells, enriching early‐differentiated T cell populations—particularly naïve and TN/SCM or TSCM subsets—during the initial stages of CAR‐T product manufacturing can help prevent premature functional exhaustion and enhance long‐term antitumor efficacy. In summary, optimizing the composition and differentiation state of CAR‐T cells not only reduces T cell exhaustion but also establishes a robust theoretical and technical foundation for the development of durable, effective, and safe adoptive cell therapies.

Regulation of CAR‐T Cell Metabolism

In the heterogeneous TME, T cell function is impaired by metabolic constraints, which significantly limit the differentiation, proliferation, and antitumor persistence of CAR‐T cells, ultimately compromising therapeutic efficacy [448]. In recent years, multidimensional intervention strategies targeting metabolic reprogramming have shown considerable promise in mitigating T cell exhaustion and enhancing CAR‐T cell functionality. Restoring mitochondrial function represents a central mechanism.

Exogenous supplementation of mitochondria has been demonstrated to substantially improve energy metabolism and enhance functional capacity in T cells. This approach enhances antitumor activity and represents a novel synergistic strategy for T cell‐based immunotherapy [449]. Similarly, upregulation of Complement component 1 Q subcomponent‐binding protein (C1QBP) expression in T cells helps preserve mitochondrial integrity, suppress ROS accumulation, and activate anti‐apoptotic signaling pathways, thereby improving the antitumor efficacy of CAR‐T therapy [450].

Moreover, hypoxia‐responsive CAR‐T cells engineered with hypoxia response elements can selectively induce CAR expression within tumor‐specific hypoxic niches, effectively alleviating T cell exhaustion and enhancing both the safety and efficacy of solid tumor therapy [451].

Regulation of glucose metabolism has also been established as a critical determinant in reversing T cell exhaustion. Forced expression of the glucose transporter1 (GLUT1) confers metabolic advantages to CAR‐T cells in glucose‐deprived TME, promotes the development of a stem cell‐like memory phenotype, and reduces cellular exhaustion [452]. Stable overexpression of GLUT1 not only enhances glycolytic flux and antioxidant capacity but also increases secretion of proinflammatory cytokines and drives Th17 differentiation, thereby augmenting cytotoxicity and long‐term antitumor persistence [453].

During in vitro expansion, the addition of IL‐15 reduces glycolytic activity by inhibiting the mTOR‐GLUT4 signaling pathway, improves T cell metabolic fitness, mitigates exhaustion, and supports the survival and effector function of CAR‐T cells [454].

Cholesterol deficiency in the TME also impairs CAR‐T cell proliferation and induces autophagy‐mediated apoptosis. Targeted modulation of liver X receptor β (LXRβ) can restore cholesterol homeostasis, enhance membrane integrity, and improve antitumor efficacy [455]. Concurrently, sustained overactivation of calcium signaling leads to excessive store‐operated calcium entry (SOCE), a key driver of CAR‐T cell exhaustion.

The SOCE inhibitor BTP‐2 significantly reduces T cell exhaustion and terminal differentiation while enhancing in vivo persistence and antitumor activity, an effect mediated by suppression of the SOCE–calcineurin–NFAT and glycolytic pathways [456]. Furthermore, arginine deprivation in the TME limits CAR‐T cell proliferation and undermines clinical outcomes. Genetic engineering to express functional argininosuccinate synthase (ASS) or ornithine transcarbamylase (OTC) enables metabolic adaptation, restores proliferative capacity, and attenuates the exhausted phenotype [457].

Epigenetic Modification of CAR‐T Cells

Epigenetic regulation plays a critical role in the development of CAR‐T cell dysfunction and exhaustion, with DNA methylation representing a central mechanism. Studies have demonstrated that DNMT3A‐mediated DNA methylation restricts the functional plasticity of CAR‐T cells and promotes exhaustion, whereas genetic ablation of DNMT3A not only enhances CAR‐T cell functionality but also increases IL‐10 expression, revealing the molecular basis of DNA methylation‐driven exhaustion. These findings provide a novel rationale for improving the long‐term antitumor efficacy of CAR‐T cells through epigenetic modulation [75].

Further evidence shows that the DNA methyltransferase inhibitor decitabine sustains high expression levels of memory‐associated genes and reduces expression of exhaustion‐related genes, thereby delaying functional decline in CAR‐T cells [458]. Additionally, the use of piggyBac transposons with low CpG content has been shown to reduce TLR9‐mediated inflammatory responses, suppress DNA methylation‐induced T cell exhaustion, significantly enhance antitumor activity, promote expansion of TILs, and decrease pro‐inflammatory cytokine release, ultimately supporting sustained CAR‐T cell function [459]. Collectively, these results underscore that targeting DNA methylation is a promising strategy to mitigate CAR‐T cell exhaustion.

Histone modification represents another effective approach to enhancing CAR‐T cell functionality. High‐throughput studies have demonstrated that class I HDACi, such as M344 and chidamide, significantly improve memory maintenance and resistance to exhaustion by reducing HDAC1 expression, enhancing H3K27ac activity, and activating key genes in the Wnt/β‐catenin signaling pathway, thereby augmenting antitumor efficacy [460]. Maik Luu et al. further confirmed that in vitro treatment with valerate and butyrate inhibits class I histone deacetylase activity and promotes the secretion of effector molecules—including CD25, IFN‐γ, and TNF‐α—resulting in markedly enhanced antitumor activity of antigen‐specific CTLs and ROR1‐targeted CAR‐T cells in syngeneic mouse models of melanoma and pancreatic cancer [461].

Furthermore, Zhang et al. found that odd‐skipped related 2 (Osr2) recruits HDAC3 to reshape the epigenetic landscape by integrating TCR signaling with the Piezo1/calcium/CREB axis‐mediated mechanical stress response, selectively inducing terminal exhaustion and suppressing cytotoxic gene expression. Notably, Osr2 knockout markedly alleviates exhaustion in tumor‐specific T cells and CAR‐T cells and enhances their antitumor function [462].

Beyond DNA methylation and histone modification, chromatin remodeling also contributes to the regulation of CAR‐T cell exhaustion. Elena Battistello et al. conducted a systematic analysis of the genome‐wide occupancy of the mSWI/SNF chromatin remodeling complex during acute and chronic T cell stimulation, revealing that this complex modulates T cell activation and phenotypic differentiation by regulating chromatin accessibility. Pharmacological or genetic inhibition of mSWI/SNF function enhances T cell persistence, alleviates CAR‐T cell exhaustion, and improves both expansion and antitumor efficacy in vivo, thereby providing a novel theoretical foundation for combining small‐molecule therapeutics with immunotherapy [82].

Combining CAR‐T With Other Therapies

In the clinical application of CAR‐T therapy, monotherapy is frequently limited by TIME and T cell exhaustion, leading to suboptimal therapeutic outcomes in certain patient populations. Consequently, combining CAR‐T therapy with complementary treatment modalities to enhance antitumor activity and long‐term persistence has become a major focus of current research.

Among various combination strategies, small molecule inhibitors and OVs have demonstrated significant synergistic effects. Studies have shown that MEK1/2 inhibitors can reduce CAR‐T cell exhaustion and terminal differentiation by suppressing the MAPK‐c‐Fos‐JunB signaling axis, thereby substantially improving CAR‐T cell efficacy in both hematologic malignancies and solid tumors [25]. Dasatinib, a tyrosine kinase inhibitor targeting abelson tyrosine kinase (ABL), can transiently inhibit CAR‐T cell signaling, maintain cells in an early differentiation state, and prevent activation‐induced cell death [463].

Furthermore, OVs enhance antitumor immunity by releasing immunostimulatory signals, promoting DC maturation, and facilitating the recruitment of CAR‐T cells and other immune effector cells to tumor sites [464, 465].

Traditional therapies such as chemotherapy and radiotherapy also play crucial roles in combination regimens. Preclinical studies indicate that administering chemotherapy 3–6 days prior to CAR‐T cell infusion reduces tumor burden, improves the TME, and promotes CAR‐T cell infiltration, expansion, and long‐term persistence [466]. Radiotherapy, meanwhile, enhances CAR‐T cell antitumor responses by inducing tumor antigen release, promoting recruitment of cytotoxic T cells, and alleviating T cell exhaustion.

The combination of radiotherapy with CAR‐T therapy not only enhances therapeutic efficacy but also reduces the required radiation dose and improves CAR‐T cell migration to tumor sites [467].

Clinical evidence further supports the benefits of combination approaches. A study showed that PD‐1 blockade using pembrolizumab reversed post‐infusion T cell exhaustion following CD19‐directed CAR‐T therapy and significantly enhanced CAR‐T cell activation and proliferation. In a cohort of 12 patients with relapsed or refractory B‐cell lymphoma, the combination of pembrolizumab and CAR‐T therapy achieved an overall response rate of 25%, including one complete response, two partial responses, and one stable disease, resulting in a clinical benefit rate of 33% [468].

These results indicate that ICIs can effectively mitigate CAR‐T cell exhaustion and improve clinical outcomes.

Notably, the “drug‐induced transient rest” strategy proposed by Evan W. Weber et al. represents a novel approach to functional recovery of CAR‐T cells. By transiently downregulating CAR protein expression or using the multikinase inhibitor dasatinib, this strategy promotes acquisition of a memory‐like phenotype, induces global transcriptional and epigenetic reprogramming, and effectively reverses T cell exhaustion while restoring antitumor activity [469].

These findings challenge the traditional notion that T cell exhaustion is a fixed and irreversible state, and provide a robust theoretical foundation and promising clinical avenue for optimizing CAR‐T therapy through rational combination strategies.

TIL Therapy

TILs are a heterogeneous population of lymphocytes that naturally infiltrate the TME and actively participate in antitumor immune responses. A substantial proportion of these cells are T cells capable of recognizing tumor‐specific antigens and neoantigens. ACT therapy, which involves the ex vivo expansion and reinfusion of autologous TILs into cancer patients, has demonstrated considerable clinical potential in recent years [470].

Although TIL‐based ACT leverages patient‐derived T cells to target tumor‐specific antigens with high specificity, its therapeutic efficacy is frequently limited by T cell exhaustion. Upon prolonged exposure to an immunosuppressive TME in vivo, TILs progressively lose their proliferative capacity and effector function. They also exhibit upregulated expression of inhibitory receptors such as PD‐1 and TIM‐3.

Therefore, optimizing TIL expansion protocols, combining ICIs, and modulating metabolic or epigenetic states have emerged as key strategies to reverse TIL exhaustion and enhance the potency of ACT therapy. These approaches align closely with exhaustion‐targeted interventions in CAR‐T and TCR‐T therapies, highlighting that multidimensional regulation of T cell function constitutes a central paradigm for achieving durable and effective cancer immunotherapy.

Preoperative and Postoperative Strategies for Modulating the Tumor Microenvironment

In TIL therapy, T cell exhaustion represents a key limitation to antitumor efficacy. To enhance the survival, functionality, and in vivo persistence of TILs, researchers routinely combine TIL infusion with preoperative and postoperative modulation of the TME to optimize antitumor immune responses. Preoperative lymphodepletion—typically achieved through chemotherapy (e.g., fludarabine or cyclophosphamide) or total body irradiation (TBI)—has significantly improved objective response rates in patients with metastatic melanoma [471].

The therapeutic benefit of lymphodepletion is partially attributed to increased availability of homeostatic cytokines such as IL‐7 and IL‐15, which are produced by non‐lymphoid tissues and play critical roles in T cell proliferation and memory formation. Transient exhaustion of endogenous lymphocytes and NK cells results in elevated serum levels of IL‐7 and IL‐15, a phenomenon known as the “cytokine sink” effect that has been strongly associated with long‐term survival and favorable clinical outcomes in TIL‐treated patients [470]. Thus, preoperative optimization of the TME establishes a supportive milieu for adoptively transferred TILs by enhancing their activation, functional capacity, and antitumor activity through improved cytokine availability.

Postoperative interventions primarily involve vaccination within prime‐boost regimens. Vaccines can activate tumor‐specific T cells and act synergistically with TIL infusion to amplify antitumor immunity [472, 473]. Additionally, vaccines may be administered as a booster after ACT therapy to promote reactivation of infused TILs and facilitate their infiltration into tumor sites [474].

By providing sustained antigenic stimulation, vaccine‐based strategies combined with TIL therapy may delay the onset of T cell exhaustion and promote differentiation toward memory phenotypes, thereby enhancing the therapeutic efficacy of TIL therapy in solid tumors.

Optimization of In Vitro Expansion Protocols

In TIL therapy, ex vivo expansion is a critical step for generating sufficient numbers of functional T cells for adoptive transfer. The clinical response to TIL‐based adoptive cell transfer depends on the presence and persistence of tumor‐specific CD8+ T cell clones [475]. However, prolonged in vitro culture often induces T cell exhaustion, compromising TIL survival and antitumor efficacy upon reinfusion.

To address this limitation, researchers have developed optimized in vitro expansion strategies aimed at preserving T cell functionality, stemness, and antitumor potential.

Culturing TILs in a “synthetic immune niche” (SIN) composed of immobilized CCL21 and ICAM‐1 significantly enhances proliferative capacity and cytotoxicity, reduces expression of the exhaustion marker TIM‐3, and increases secretion of IFN‐γ and granzyme B, thereby effectively suppressing tumor growth [476]. Additionally, co‐stimulation with IL‐2, IL‐15, and IL‐21 during TIL expansion from glioma patients generates a population enriched in central memory phenotype cells, characterized by reduced expression of exhaustion markers and enhanced cytokine production and cytotoxic activity against autologous tumors and shared antigens [477]. Combining the SIN platform with multi‐cytokine stimulation provides an efficient and reproducible method for in vitro TIL expansion, enabling simultaneous optimization of cellular phenotype and functional potency.

The phenotypic composition of expanded TILs directly influences therapeutic efficacy, making the selective enrichment and maintenance of stem cell‐like and effector T cell subsets during in vitro expansion critically important. Studies have demonstrated that CD27− TILs exhibiting an effector phenotype exhibit superior therapeutic efficacy compared to CD27+ TILs with a memory phenotype [478]. Further analysis of TIL infusion products has revealed that a small subset of stem cell‐like, CD39−CD69− tumor‐reactive T cells is strongly associated with durable clinical responses [479].

Activation of the ProS1‐MerTK signaling pathway promotes central memory differentiation, enhances metabolic adaptability and proliferative capacity in CD4+ T cells, drives type 1 immune responses, and concurrently reinforces stemness while mitigating exhaustion during TIL expansion, thereby improving the efficacy of ACT [480]. Moreover, reprogramming exhausted TILs into induced pluripotent stem cells (iPSCs), followed by redifferentiation into heterogeneous T cell populations enriched in naïve and memory phenotypes, represents a promising strategy to overcome the limitations of conventional expansion protocols—particularly the scarcity of functional T cells—and significantly enhances antitumor activity [481].

Combining TILs With Other Therapies

The combination therapy strategy has demonstrated significant potential in optimizing TIL therapy. The core of this approach lies in delaying or reversing T cell exhaustion and enhancing the durability and antitumor functionality of TILs through the synergistic action of various immunomodulatory interventions. For instance, cytokine modulation via membrane‐bound IL‐15 reduces the proportion of terminally differentiated T cells, downregulates expression of inhibitory receptors such as LAG‐3, TIGIT, and TIM‐3, promotes the generation of stem‐like CD8+ T cells, and significantly enhances TIL persistence and antitumor efficacy in the treatment of NSCLC [482].

Combined epigenetic interventions provide a novel and optimized approach for enhancing TIL therapy. Studies have demonstrated that EZH2 inhibitors can block H3K27me3‐mediated gene silencing, improve the differentiation status and functional capacity of TILs, and enhance synergy with ICIs [483, 484, 485, 486]. Bromodomain and extra‐terminal domain (BET) inhibitors increase the proportion of stem cell‐like and central memory T cells, reduce expression of Th17 differentiation‐associated factors, and improve the durability and antitumor efficacy of TILs [487].

Furthermore, inhibition of DNA methyltransferases or histone deacetylases can also enhance T cell function to a certain extent [488, 489], offering a viable strategy to maintain the long‐term activity of TILs during ex vivo expansion and post‐infusion persistence.

The combination of ICIs with TIL therapy has also demonstrated significant advantages. In a TNBC mouse model, TIL therapy combined with PD‐1 inhibitor treatment significantly increases the secretion of IFN‐γ and TNF‐α, enhances tumor infiltration by activated T cells, and downregulates inhibitory markers such as PD‐1, thereby effectively alleviating T cell exhaustion and achieving tumor clearance [490]. Radiotherapy induces immunogenic cell death, promotes the release of tumor antigens and danger signals, and creates an immunological window for TIL activation, thereby enhancing the local antitumor effect [470].

Furthermore, spermidine‐induced restoration of autophagy reduces the expression of inhibitory receptors, increases TCR diversity, and significantly enhances the proliferative capacity and cytotoxic activity of TILs, offering a novel metabolic regulatory strategy to reverse T cell exhaustion [491].

TCR‐T Therapy

TCR‐T therapy involves introducing the TCR gene encoding a specific tumor antigen into the patient's T cells, enabling them to accurately recognize and eliminate antigen‐expressing tumor cells. Unlike CAR‐T cells, which target surface antigens in an MHC‐independent manner, TCR‐T cells recognize peptide antigens derived from intracellular protein processing through MHC molecule presentation, thereby expanding their applicability to a broad range of endogenous tumor antigens. Existing studies have demonstrated that TCR‐T cells hold significant potential for enhancing antitumor immunity, reversing T cell exhaustion, and serving as a component of combination immunotherapy.

However, their clinical efficacy remains limited by MHC restriction, tumor immune escape mechanisms, and immunosuppression within the TME. Therefore, strategies such as optimizing TCR affinity, alleviating T cell exhaustion, and combining TCR‐T therapy with ICIs or cytokine‐based interventions have become key approaches to improving its antitumor efficacy.

Studies have shown that a multi‐engineered T cell vaccine (Multi‐Tvac) can activate TCR‐T cells and alleviate their exhaustion by promoting DC maturation and enhancing tumor antigen presentation in secondary lymphoid organs. In a solid tumor model, the combined use of Multi‐Tvac and TCR‐T cell therapy prevented immune escape due to antigen loss and achieved complete tumor regression, highlighting the critical role of optimized antigen presentation in overcoming TCR‐T cell exhaustion [492].

In recent years, combination immunotherapy has been recognized as a key strategy to alleviate T cell exhaustion and enhance the efficacy of cellular immunotherapy [365]. For example, EBV‐positive tumors recruit mononuclear‐macrophages via CCL5 to secrete CSF1 and IL‐10, thereby inducing CD163+ M2 macrophage polarization. These M2 macrophages secrete high levels of MMP9, leading to TIL dysfunction and resistance to TCR‐T therapy.

Inhibition of MMP9 not only restores T cell function but also significantly suppresses the growth of EBV‐positive solid tumors when combined with TCR‐T treatment [493]. Additionally, Akt1 and Akt2 isoforms promote terminal differentiation of CD8+ T cells, thereby restricting the formation of central memory T cells (TCM). Selective inhibition of these two isoforms delays T cell exhaustion and preserves naïve and TCM populations, thus enhancing the proliferation, survival, and effector function of CD8+ T cells [494].

Moreover, studies have shown that linoleic acid (LA) promotes the formation of endoplasmic reticulum‐mitochondrial contact sites (MERCs), enhances calcium signaling and mitochondrial energy metabolism, improves the metabolic fitness of CD8+ T cells, prevents their exhaustion, promotes a memory‐like effector phenotype, and significantly enhances antitumor function [495]. Andrew Nguyen et al. found that MS‐275, a histone deacetylase inhibitor, reverses tumor burden‐induced immunosuppression, reduces exhaustion‐associated features, enhances IL‐2–STAT5 signaling pathway activity, and promotes the differentiation of CD8+ T cells into terminally differentiated effector cells, thereby improving durable regression in adoptive T cell therapy [496]. In a combinatorial TCR and CAR strategy, GPC3‐HLA‐A2 transgenic mice were used to screen efficient TCRs, which were then engineered into TCR‐T cells.

Combined administration of ACT and CAR‐T cells enhances the proliferation, effector function, and resistance to exhaustion of TCR‐T cells, thereby significantly improving the antitumor effect in HCC [497].

TCR‐T cell therapy has demonstrated unique targeting advantages and broad clinical potential in cancer treatment. Optimizing antigen presentation, metabolic regulation, signaling pathway modulation, and combining TCR‐T therapy with ICIs, cytokine‐based interventions, or other cell engineering approaches hold promise for significantly alleviating T cell exhaustion, enhancing long‐term antitumor immunity, and overcoming immune escape. In the future, multi‐dimensional combination strategies will enable more precise and effective TCR‐T–based immunotherapies for solid tumors.

OVs

OVs are self‐replicating viruses that selectively target and infect tumor cells, replicate extensively, and induce tumor cell lysis. Recent studies have demonstrated that, in addition to their direct tumoricidal activity, OVs can modulate the host immune response via their distinct structural features and mechanisms of action. This immunomodulatory capacity helps alleviate T cell exhaustion and enhances antitumor immunity.

Engineered OVs as a Platform for Delivering Immunomodulatory Factors

In recent years, a range of OVs have been developed for the localized or systemic delivery of immunomodulatory agents aimed at improving the TIME and mitigating T cell exhaustion. The tumor‐selective herpes simplex virus (HSV) vector, engineered by Fei Ju et al. to encode a human PD‐1 single‐chain variable fragment (hPD‐1scFv), enables targeted release of a PD‐1 inhibitor within the tumor microenvironment, thereby significantly enhancing CD8+ T cell infiltration and activation, reducing T cell exhaustion markers, and promoting antitumor immune responses [392]. Additionally, conditionally replicating adenoviruses (CRAds) genetically modified to express immune‐stimulatory transgenes mCD40L or m4‐1BBL have demonstrated efficacy in a murine melanoma model, where they delayed tumor progression, extended survival, and enhanced immune cell infiltration: m4‐1BBL primarily promotes CD8+ T cell recruitment, whereas mCD40L facilitates CD4+ T cell infiltration, and their combined action leads to a synergistic enhancement of anti‐tumor immunity [498].

The engineered adenovirus TILT‐322, which expresses an aMUC1×aCD3 T cell engager and IL‐2, has been shown to activate γδ T cells and various immune cell populations, elevate levels of granzyme B, perforin, and IFN‐γ, and reduce the proportion of exhausted CD8+ T cells in ovarian cancer ascites. These effects contribute to enhanced T cell functionality and improved tumor control [499]. Similarly, TILT‐517 expressing IL‐7 enhances T cell and NK cell activity and promotes remodeling of the TME in the renal cell carcinoma (RCC) model [500].

VSV‐IFNβ exerts direct oncolytic effects, activates innate immunity, and enhances CD8+ T cell responses through sustained IFNβ expression, while simultaneously reducing the frequency of Tregs. When combined with ICIs, VSV‐IFNβ helps preserve the functionality of tumor‐specific T cells and prevents their exhaustion [501]. Pexa‐Vec can convert the TME from an immunologically “cold” to a “hot” state by inducing interferon signaling and pro‐inflammatory cytokine production, thereby promoting CTL infiltration and activation [502].

In addition, mesenchymal stem cell‐mediated delivery of oncolytic adenovirus (OAd‐MSCs) has been shown to increase the proportions of TAMs, natural killer cells, and TILs, while reducing the frequency of PD‐1+ TILs, thereby partially reversing T cell exhaustion [503]. The oncolytic adenovirus Ad5/3Δ24‐GM‐CSF was demonstrated to significantly decrease TIM‐3+ CD8+ TILs and improve the intratumoral immune microenvironment [504]. Therefore, engineered OVs not only mediate antitumor effects through direct oncolysis but also contribute to the reversal of T cell exhaustion and enhancement of antitumor immune responses via immunomodulation.

Combination Immunotherapy Involving OVs

Notably, OVs possess the dual capabilities of directly lysing tumor cells and modulating the immune microenvironment. When combined with immunotherapy, OVs can mitigate T cell exhaustion and augment anti‐tumor immune responses. In recent years, accumulating evidence has demonstrated the potential of OVs in reversing T cell dysfunction.

For instance, an oncolytic virus expressing OX40L (OV‐mOX40L) administered in PDAC models has been shown to enhance CD4+ T cell activation, reduce Treg infiltration, and alleviate CTL exhaustion. Therapeutic efficacy is further enhanced when this approach is combined with anti‐IL‐6 and anti‐PD‐1 therapies, offering a promising multimodal strategy for PDAC [505]. Additionally, NKG2A blockade in combination with the KISIMA‐VSV‐GP‐TAg vaccine significantly reduced CD8+ T cell exhaustion within the tumor microenvironment, improved local T cell functionality, prolonged survival in murine models, and promoted the development of effector memory CD8+ T cells [506].

Furthermore, the combination of oncolytic adenovirus expressing interleukin‐7 (oAD‐IL7) with B7‐H3‐targeted CAR‐T cells has been shown to significantly enhance T cell proliferation and survival, as well as improve the persistence of tumor‐infiltrating CAR‐T cells [507]. Additionally, OAd‐MSCs combined with granulocyte colony‐stimulating factor (G‐CSF) have been shown to increase TIL counts, reduce T cell exhaustion, and suppress tumor growth in murine osteosarcoma models [508].

The development of multifunctional and multi‐target OVs therapy strategies offers novel insights into mitigating T cell exhaustion. Recombinant vaccinia virus encoding an EpCAM‐targeting bispecific T cell engager (VV‐EpCAM BiTE) is capable of infecting and lysing tumor cells, thereby releasing the BiTE molecule, which facilitates the formation of a bridge between tumor cells and T cells, activates naïve T cells, and enhances cytokine production. Intratumoral administration of this construct has been shown to significantly increase immune cell infiltration, reduce CD8+ T cell exhaustion, and augment anti‐tumor immune responses across various solid tumors [509].

Additionally, the virus‐like vesicle (VLV) vector CARG‐2020 enables co‐delivery of single‐chain IL‐12, PD‐L1‐targeted shRNA, and a dominant‐negative mutant of IL‐17RA, leading to sustained immune memory and systemic anti‐tumor effects through the activation of Th1 immunity and suppression of genes associated with T cell exhaustion and pro‐tumorigenic inflammation [510].

Cancer Vaccines

Cancer vaccines aim to reverse T cell exhaustion by reactivating antitumor T‐cell responses through the delivery of tumor‐specific antigens. However, within the profoundly immunosuppressive tumor microenvironment, antigen stimulation alone often yields limited efficacy and may even aggravate T cell exhaustion. Consequently, recent strategies have focused on combining cancer vaccines with immunomodulatory approaches to enhance their effectiveness.

These include blocking inhibitory immune checkpoints, remodeling the tumor microenvironment, and promoting the generation of memory T cells. For instance, integrating vaccines with small‐molecule inhibitors—such as those targeting the JAK‐STAT pathway or bromodomain proteins—has been shown to augment T cell activation, mitigate exhaustion, and facilitate tumor infiltration [511, 512]. Additionally, when combined with adoptive cell therapies, such as CAR‐T or TILtransfer, vaccines enhance effector T cell expansion, persistence, and antigen‐specific immune responses [513, 514].

Collectively, these integrated strategies demonstrate the potential of cancer vaccines to reverse T cell exhaustion and enhance the effectiveness of immunotherapy. To this end, we summarize the relevant studies and underlying mechanisms (Table 4).

It is noteworthy that the vaccine delivery strategy constitutes a critical factor in enhancing therapeutic efficacy and counteracting T cell exhaustion. Evidence from preclinical studies indicates that intravenous administration demonstrates superior efficacy compared to subcutaneous injection in generating ICIs‐responsive TPEX cells in MC38 tumor‐bearing mice, leading to improved tumor control [529]. Furthermore, targeted delivery of vaccines to lymph nodes via amphiphilic vectors has been shown to optimize DC function within lymphoid tissues and potentiate antitumor immune responses [530, 531].

Additionally, the selection of adjuvants can modulate the expression of costimulatory molecules and cytokines on DCs, thereby influencing the differentiation and functional fate of CD8+ T cells [532].

Costimulatory Agonists

Depleted T cells often lose critical costimulatory signals (e.g., OX40, 4‐1BB, GITR, CD40), compromising their functions and reducing the effectiveness of ICIs. Agonists targeting these receptors act as “functional restoration switches” delivering survival and effector signals to TEX cells and restoring or enhancing their antitumor capabilities. Costimulatory agonists, when used alone or in combination with ICIs and other immunotherapies, have demonstrated potential in delaying or reversing T cell exhaustion (Table 5).

Although costimulatory agonists may pose clinical risks such as cytokine release syndrome, autoimmune responses, and thrombosis [546, 547], they show significant potential as a “functional recovery switch” in combination immunotherapy. Nevertheless, their safety profile, optimal dosage, and administration regimen require further optimization and validation.

Bispecific Antibody

BsAbs are next‐generation immunotherapies that can simultaneously target two distinct antigens, making them highly promising anticancer agents. By bridging T cells to tumor‐associated antigens or modulating two immune signaling pathways, BsAbs can activate effector T cells, alleviate immunosuppression, reprogram the function of TEX cells, and restore their proliferation and cytotoxicity. Notably, BsAbs have shown substantial clinical potential in hematological malignancies and some solid tumors, offering a novel strategy to reverse T cell exhaustion and enhance cancer immunotherapy [548].

One of the key contributors to T cell exhaustion is the persistent presence of inhibitory immune signals, particularly through sustained activation of immune checkpoint receptors such as CTLA‐4 and PD‐1. BsAbs demonstrate distinct therapeutic potential in this context by simultaneously targeting tumor‐associated antigens and inhibitory receptors, or by engaging two immune regulatory molecules concurrently, thereby directly suppressing exhaustion‐related signaling pathways and restoring T‐cell functionality. For instance, MGD019, a bispecific antibody that co‐blocks PD‐1 and CTLA‐4, enables localized modulation of Tregs within the TME while preserving systemic anti‐CTLA‐4 activity, thus enhancing both safety and antitumor efficacy [549].

Similarly, in patients with B‐cell non‐Hodgkin lymphoma (B‐NHL), the combination of BsAbs targeting CD39 or CD73 has been shown to augment T‐cell responses, particularly among individuals exhibiting a high frequency of PD‐1+ TIM‐3+ T cells [345].

TEX cells exhibit not only diminished effector function but also a loss of costimulatory signaling, making durable immune reconstitution difficult to achieve through the mere blockade of inhibitory pathways. By incorporating costimulatory signals, BsAbs can promote the reactivation of TEX cells and sustain their functional capacity, thereby enhancing tumor control. In multiple myeloma, BCMA‐targeted immunotherapies have significantly improved patient outcomes; however, the recurrence rate remains high.

Next‐generation T cell–redirecting BsAbs, such as those targeting GPRC5D or FcRH5, have demonstrated clinical promise in patients who have relapsed after prior treatments. Trispecific antibodies (TsAbs), which integrate costimulatory signals while simultaneously engaging two tumor antigens, offer a strategic advancement by mitigating both T cell exhaustion and antigen escape [550], thus providing a novel approach to overcoming the limitations of conventional ICB.

In addition, the immunosuppressive characteristics of the TME, such as upregulation of PD‐1/PD‐L1 during treatment, can limit the efficacy of BsAbs in solid tumors [551]. Combining ICB with BsAbs, including bispecific T‐cell engagers (BiTEs), has been shown to significantly restore and enhance T‐cell function. For instance, the CD33/CD3 bispecific antibody AMG 330 induces cytotoxic activity against AML cells, accompanied by concurrent upregulation of PD‐L1 expression.

The blockade of PD‐1/PD‐L1 further enhances AMG 330‐mediated cell lysis, T‐cell proliferation, and interferon‐γ secretion [552]. In studies involving CEA‐targeting Bsabs, upregulation of PD‐1 and impaired effector T‐cell function have been observed; combining such agents with anti‐PD‐1/PD‐L1 therapy can augment tumor cell killing [553]. Although HER2‐TDB enhances the antitumor activity of anti‐HER2 therapies, it may also induce PD‐L1‐mediated resistance.

The combination of HER2‐TDB with anti‐PD‐L1 treatment has been demonstrated to significantly improve antitumor efficacy and prolong response duration [554].

BsAbs have demonstrated considerable potential in reversing T cell exhaustion and enhancing antitumor immune responses by concurrently blocking inhibitory signals and activating costimulatory pathways, particularly in hematological malignancies and certain solid tumors. Nevertheless, challenges remain in the context of solid tumors, including immunosuppressive tumor microenvironments, suboptimal target selection, and immune‐related toxicities. Future research should focus on refining BsAb design, improving tumor‐specific delivery strategies, combining BsAbs with ICB or other immunotherapies to mitigate adverse effects, and utilizing biomarkers for patient stratification to enable personalized, precision‐based therapeutic approaches.

A deeper understanding of the mechanisms underlying the response of TEX cell subsets to various BsAb formats will further contribute to enhanced therapeutic efficacy and sustained anti‐tumor immunity.

Antibody–Drug Conjugates

ADCs are composed of a monoclonal antibody, a chemical linker, and a highly potent cytotoxic agent, representing a targeted therapeutic strategy in oncology. By selectively delivering cytotoxic payloads to tumor cells, ADCs achieve potent antitumor efficacy while minimizing systemic toxicity. In comparison with conventional chemotherapy, ADCs offer not only direct tumor cell killing but also potential immunomodulatory effects.

Specifically, ADC‐induced tumor cell lysis through ICD leads to the release of tumor‐associated antigens and DAMPs, such as ATP and HMGB1. These molecules facilitate DC maturation and promote antigen‐specific T cell activation. Collectively, these immune‐mediated responses help overcome tumor immune tolerance, remodel the immunosuppressive microenvironment, and contribute to reversing T cell exhaustion.

Recent studies have demonstrated that a novel humanized AXL‐targeting antibody‐drug conjugate (AXL02‐MMAE) effectively suppresses tumor migration and EMT by selectively targeting tumor cells exhibiting high AXL expression as well as tumor‐associated M2 macrophages. This agent also attenuates the immunosuppressive TME and restores CTL function, leading to significant reversal of T cell exhaustion [555]. Similarly, ADCs incorporating pyrrolobenzodiazepine or tubulysin payloads have been shown to activate the immune system and enhance CD8+ T cell function through induction of ICD.

These ADCs exhibit synergistic antitumor activity when combined with PD‐1/PD‐L1 checkpoint inhibitors or costimulatory agonists targeting OX40 and GITR pathways [556].

The cGAS‐STING pathway is activated upon recognition of foreign DNA released by necrotic tumor cells, leading to the production of type I interferons and subsequent activation of the innate immune system, thereby promoting T cell infiltration into the tumor microenvironment [557, 558, 559, 560]. Building on this mechanism, STING‐ADC enhances antitumor immunity while maintaining systemic tolerance through targeted delivery of STING agonists, and demonstrates synergistic efficacy when combined with PD‐L1 inhibitors, highlighting its potential for clinical translation [561]. For instance, XMT‐2056 utilizes HER2‐targeted delivery to administer a STING agonist, thereby inducing tumor regression via activation of innate immunity, reversal of immunosuppressive conditions, reduction of T cell exhaustion, and enhancement of T cell effector function [562].

Additionally, trastuzumab deruxtecan (T‐DXd) can induce immunogenic cell death by releasing cytotoxic payloads, activate myeloid cells, and augment the antitumor activity of CD8+ T cells. However, it may concurrently upregulate tumor expression of CD47 and contribute to T cell exhaustion. Combination with CD47 blockade can counteract this immunosuppressive effect, significantly improve the therapeutic efficacy of antibody‐drug conjugates, enhance CD8+ T cell memory responses, and reduce the risk of tumor recurrence [563].

ADCs not only exert direct tumoricidal effects but also reverse T cell exhaustion through the induction of immunogenic cell death, enhancement of antigen presentation, and remodeling of the tumor microenvironment. However, their immunoregulatory efficacy is limited by tumor heterogeneity, immunosuppressive features of the TME, and potential systemic toxicity. Future strategies to enhance immune reconstitution and anti‐tumor activity may include optimization of ADC design, combination with costimulatory signals or ICIs, augmentation of cGAS/STING pathway activation, and integration of biomarker‐driven personalized therapeutic approaches.

Challenges and Future Directions

T cell exhaustion resulting from persistent tumor antigen stimulation has emerged as one of the most significant biological barriers in cancer immunotherapy. Although ICB has demonstrated remarkable efficacy in reinvigorating TEX cells, critical challenges persist, shaping key priorities for future research.

T cell exhaustion is a dynamic differentiation process with multiple distinct subsets, rather than a uniform cellular state. At the two extremes of this lineage are TPEX cells, which possess self‐renewal and proliferative capacity, and TEX‐term cell, which exhibit limited functionality. However, the origins and differentiation pathways of these subsets remain a subject of debate [122, 564].

Some studies propose that TEX cells arise from conventional effector T cells, which progressively lose effector function under persistent antigen stimulation through upregulation of inhibitory receptors and gradual impairment of cytokine production, following a model of “continuous phenotypic evolution” [565]. In contrast, accumulating evidence indicates that T cells may commit to a distinct differentiation trajectory early during activation in the setting of chronic antigen exposure. Key transcription factors such as TOX are rapidly induced and drive extensive epigenetic reprogramming, establishing an exhaustion‐specific chromatin landscape that actively directs T cell fate toward exhaustion rather than representing mere functional decline [316].

Heterogeneity in the TME—such as differences in tumor type, antigen burden, and metabolic conditions—futher complicates differentiation pathways and increases therapeutic uncertainty [566, 567]. However, traditional analytical approaches are limited in their ability to fully capture these complex divergence trajectories and spatial distributions. The integration of single‐cell multi‐omics technologies—such as scRNA‐seq, ATAC‐seq, and single‐cell T cell receptor sequencing (scTCR‐seq)—with spatial profiling methods, including spatial transcriptomics, has overcome this technical limitation.

This integrated approach enables the simultaneous analysis of transcriptional, epigenetic, and spatial localization data at single‐cell resolution, thereby allowing for a precise characterization of the differentiation trajectories, spatial organization, and interactions of TEX cells within the tumor microenvironment [123, 568, 569, 570, 571].

A major challenge is the epigenetic basis of T cell exhaustion. Upon entering the exhausted state, T cells undergo extensive epigenetic remodeling, characterized by alterations in chromatin accessibility, DNA methylation, and histone modifications—collectively forming an “epigenetic scar” [24]. This epigenetic stabilization renders the exhausted phenotype persistent even after antigen removal, which explains why ICB therapies, such as PD‐1 inhibition, often fail to fully reverse TEX‐term cells.

Instead, therapeutic efficacy primarily depends on the expansion of residual TCF1+ progenitor‐like cells [572, 573]. Thus, safely and effectively reprogramming the epigenetic landscape of TEX cells remains a critical hurdle in immunotherapy. Although direct ablation of exhaustion‐associated genes such as TOX and Tim‐3 can transiently restore T cell function, it frequently results in rapid T cell exhaustion or severe immunopathological consequences, underscoring the essential biological role of the exhaustion program in preserving T cell longevity and systemic homeostasis [297].

Novel approaches—including low‐dose epigenetic modulators (e.g., DNMT3A or HDAC inhibitors) combined with ICB, as well as CRISPR‐Cas9‐mediated targeted epigenome editing—are currently under investigation in clinical settings [24, 574, 575]. Moreover, T cell exhaustion is not limited to transcriptional and epigenetic dysregulation but also involves significant disruptions at the translational level [576]. Recent evidence indicates that tumor‐infiltrating T cells adopt an aberrant “hypertranslation” state during exhaustion.

In particular, the RNA‐binding protein LARP4 enhances the translational efficiency of nuclear‐encoded OXPHOS mRNAs, leading to mitochondrial translational imbalance and accumulation of ROS, thereby accelerating T cell dysfunction [577]. Targeting this hyperactive translation state may offer a promising strategy for reversing T cell exhaustion; however, research in this area remains in its early stages.

In addressing the complex challenge of T cell exhaustion, a fundamental paradigm shift in therapeutic strategy is required—from attempting to reverse terminally exhausted states toward early intervention and fate regulation. Rather than pursuing the impractical goal of fully reversing terminal T cell exhaustion, greater emphasis should be placed on preserving the population of TPEX cells with stem‐like properties, delaying their differentiation into terminal states, and integrating strategies for remodeling the TME to enhance T cell functionality and persistence. For instance, inhibition of suppressive signals such as TGF‐β and IL‐10, or provision of supportive cytokines including CD28 costimulatory signals and IL‐21, has been shown to effectively delay the exhaustion process [578, 579, 580, 581].

Notably, T cell exhaustion should not be viewed as an isolated dysfunction of CD8+ T cells, but rather as a component of a complex, multicellular immune network. The impairment of CD4+ T cells results in the loss of critical helper signals, thereby accelerating the functional decline of CD8+ T cells [41, 582]. Furthermore, dysregulation of Tregs during chronic inflammation may transiently reduce immunosuppressive activity, yet over time leads to disruption of immune homeostasis and increased susceptibility to autoimmune pathology [583].

Consequently, any effective therapeutic approach must account for these intricate intercellular interactions within the immune system.

Innovative technologies are advancing the field at an unprecedented pace. Emerging technologies such as CRISPR‐Cas9 gene editing enable precise modulation of exhaustion‐related pathways (e.g., NR4A, TOX), offering new avenues for durable T cell rejuvenation. Recent studies have demonstrated that deletion of the three NR4A family genes (NR4A1, NR4A2, and NR4A3) in CAR‐T cells significantly enhances their durable cytotoxic capacity against solid tumors and improves mitochondrial metabolic adaptability, with consistent efficacy observed across multiple donor‐derived T cell populations [584].

Similarly, knockdown of TOX downregulates the expression of multiple immune checkpoint molecules, delays the progression of T cell exhaustion, and has been associated with improved prediction of immunotherapy response [335]. Synthetic biology enables T cells to sense and self‐regulate exhaustion signals through the construction of artificial genetic circuits. Recent evidence shows that T cells can be reprogrammed into an atypical exhausted state via orthogonal cytokine engineering—such as expression of IL‐2 variants, IL‐33, or PD‐1 decoy receptors—which effectively suppresses TOX expression and enhances T cell persistence and antitumor efficacy in solid tumor models [160, 585].

Furthermore, logic‐gated systems based on AND, OR, and NOT Boolean circuits have been employed to precisely control T cell responses to tumor microenvironment signals, thereby reducing off‐target toxicity while preserving effector function [586, 587]. In addition, advances in nanotechnology and delivery systems have enabled the development of efficient and precise therapeutic platforms; for instance, lipid nanoparticle (LNP)‐based mRNA delivery systems allow localized and controlled expression of the cytokine IL‐12, minimizing systemic toxicity while enhancing T cell function [368]. However, it must be emphasized that the efficient and specific delivery of therapeutic agents to target cells within the tumor microenvironment, while minimizing off‐target effects, remains a major barrier to clinical translation.

Future strategies are likely to rely on the development of intelligent, stimuli‐responsive delivery systems—such as nanomaterial‐based drug carriers sensitive to the unique pH, enzymatic activity, or redox conditions of the TME—or the engineering of viral and non‐viral vectors designed to recognize specific T cell surface antigens, enabling controlled in vivo drug release [588, 589]. Interestingly, biomimetic physical barriers can transiently block immunosuppressive interactions between T cells and tumor cells, creating a “protective microenvironment” that promotes T cell accumulation and functional preservation, with restored cytotoxic activity upon timely release [590]. A deeper understanding of the mechanisms underlying T cell exhaustion offers promising avenues for therapies that precisely modulate this state, enabling the immune system to mount robust and sustained antitumor responses—ultimately leading to durable, safe, and effective cancer immunotherapy.

Conclusion

T cell exhaustion represents a central mechanism underlying tumor immune escape. Because of its multi‐layered and dynamic nature, single therapeutic interventions are unlikely to yield significant breakthroughs. Future research should build upon a comprehensive understanding of the heterogeneity of TEX cells, epigenetic regulation, and metabolism‐translation coupling networks to develop multi‐targeted, temporally controlled combination strategies.

Integrating interdisciplinary approaches from immunology, epigenetics, synthetic biology, and nanomaterials science may enable precise reprogramming of T cell states. This holds the potential to stabilize stem‐like T cell subpopulations, inhibit terminal differentiation, remodel the TIME, and ultimately enable durable and safe anti‐tumor immunity. Achieving this goal requires both innovative scientific insights and rigorous translational and clinical validation.

Conflicts of Interest

The authors declare no conflicts of interest.