The tumor microenvironment (TME) is a dynamic, heterogeneous ecosystem in which malignant cells coexist with stromal elements, vasculature, and a spectrum of immune cell populations. It encompasses immune cell subsets with immunostimulatory and immunosuppressive activities (1, 2). In many solid tumors, the TME evolves into an immunosuppressive niche, characterized by nutrient depletion, hypoxia, suppressive cytokines, and the accumulation of immunosuppressive cells, which allow cancer to evade anti-tumor immune responses and consequently limit the efficacy of anticancer therapies (1). Myeloid-derived suppressor cells (MDSCs) are a heterogeneous population of immature myeloid-derived immune cells that play a prominent role in driving immune suppression and limiting the response to immunotherapies (3, 4). These heterogeneous, pathologically activated immature myeloid cells (monocytic and polymorphonuclear subsets) employ multiple mechanisms – arginine depletion, reactive oxygen and nitrogen species, immune-checkpoint ligand expression, and secretion of suppressive cytokines and lipid mediators – to inhibit T cell activation and support tumor progression. MDSC abundance and activity are correlated with poor prognosis and a reduced response to immune checkpoint blockade across several cancer types (5).

Metabolic reprogramming is a key axis by which the TME enforces MDSC function. Rather than merely passive bystanders, MDSCs rewire their metabolism to survive in a nutrient-depleted TME (6). The rewired metabolic adaptation also enables MDSCs to generate immunosuppressive mediators (e.g., PGE2, arginase 1 (ARG1), NOS), which together support MDSC persistence in hostile microenvironments and directly impair effector lymphocyte metabolism and function (6). Since MDSCs are at the intersection of cellular immunosuppression and metabolic dysregulation, they represent an attractive therapeutic target to broaden and deepen anti-tumor immunity (7). Preclinical and early clinical strategies that deplete MDSCs, block their recruitment, inhibit key metabolic pathways, or reprogram them toward differentiation have shown promise in restoring T-cell activity and improving responses to checkpoint inhibitors and other therapies (8). Nevertheless, clinical translation faces considerable challenges, including the heterogeneity of MDSCs, overlapping expression of metabolic targets across cell types, and potential systemic toxicities that limit clinical therapeutic outcomes. Therefore, rational combination approaches that concurrently target MDSCs and the complex immunosuppressive network in the TME, together with biomarker-guided patient selection to enable personalized therapies, are critical for clinical success (7).

In this review, we explore recent insights into the immunometabolic characteristics of MDSCs, focusing on newly identified features in metabolism and immune regulation. We also discuss therapeutic strategies targeting metabolic and immune pathways, including metabolic targeting of MDSCs, strategies to deplete MDSC populations, blocking their recruitment into the TME, and promoting their differentiation into mature non-immunosuppressive myeloid cells. We highlight novel approaches that combine immune checkpoint blockade and chimeric antigen receptor (CAR)-based therapies with MDSC-targeted approaches. We also discuss ongoing clinical trials that target MDSCs immune and metabolic regulation, with the aim of improving the efficacy and clinical outcomes of immunotherapies in hematological and solid malignancies. Finally, we address future perspectives and challenges associated with the immunometabolic targeting of MDSCs in cancer treatment.

MDSCs emerge upon persistent pathological immune activation of the myeloid cell compartment by prolonged exposure to myeloid growth factors and inflammatory signals arising from tumors, chronic infections, inflammation, or autoimmune diseases (9). In the tumor setting, MDSC ontogeny is shaped by complex interactions between tumor-derived factors and the bone marrow niche (Figure 1, left panel). These immature myeloid cells originate primarily from hematopoietic stem cells in the bone marrow but can also develop from myeloid progenitors in secondary lymphoid organs such as the spleen (10) (Figure 1, right panel). Tumor-secreted cytokines such as G-CSF, GM-CSF, and IL-6 promote the expansion of immature myeloid progenitors in the bone marrow while simultaneously blocking their differentiation into mature dendritic cells, macrophages, or granulocytes. These expanded MDSC populations are then recruited to the tumor site through chemokine gradients, where they undergo further activation and acquire enhanced immunosuppressive functions (10). According to the latest classification system, the human MDSC population comprises three distinct groups: monocytic MDSCs (M-MDSCs), polymorphonuclear MDSCs (PMN-MDSCs), and early stage MDSCs (e-MDSCs) (11). The morphological and phenotypic characteristics of M-MDSCs and PMN-MDSCs closely resemble those of monocytes and neutrophils, respectively (12). e-MDSCs are the least mature population of MDSCs that are phenotypically devoid of granulocytic and monocytic surface markers (11).

Several phenotypic markers have been established to distinguish between MDSC subtypes in humans and murine models (Table 1). Moreover, recent studies have advanced the identification of MDSCs by introducing novel surface and intracellular markers to distinguish MDSC subtypes from other myeloid cells across various cancers, thereby enhancing diagnostic and prognostic precision (Table 1). For instance, a combination of markers, including CD14, CD68, CD163, CD206, and S100A9, has been employed in immunofluorescent multiparametric assays to accurately delineate MDSCs, offering improved specificity over traditional markers (13). A CRISPR-Cas9 screen recently identified CD300ld, expressed specifically in neutrophils, as a crucial marker of tumoral PMN-MDSCs (14). Moreover, excluding CD123⁺ cells from e-MDSC populations helps to distinguish them from basophils, addressing concerns about marker specificity and improving diagnostic accuracy (15).

MDSCs are pivotal orchestrators of immunosuppression within the TME and deploy a repertoire of mediators and molecules to suppress key immune components, including T cells, natural killer (NK) cells, macrophages, and dendritic cells (DCs) (16). Recent studies have elucidated the molecular mechanisms underlying these interactions, highlighting their roles in tumor immune evasion and progression across cancers. By leveraging metabolic reprogramming, cytokine/chemokine signaling, and immune checkpoint molecules, MDSCs create a suppressive network that hampers antitumor immunity, making them critical targets for therapeutic interventions. MDSCs reprogram their metabolism to deplete essential nutrients and produce toxic by-products in a subset specific manner (6, 10, 17–19). PMN-MDSCs primarily suppress T cells by generating large amounts of reactive oxygen species (ROS) through NADPH oxidase 2 (NOX2) and peroxynitrite (ONOO-) (6, 10). These reactive intermediates induce nitration of T cell receptors and inhibit antigen-specific interactions, disrupting T cell responses (6, 10). In contrast, M-MDSCs rely more heavily on Arginase-1 (ARG1), indoleamine 2,3-dioxygenase (IDO), and inducible nitric oxide synthase (iNOS) to deplete L-arginine and nitric oxide (NO) production to impair receptor signaling, inhibit T cell proliferation, and induce regulatory T-cell expansion to repress T-cells (6, 10, 17–19). MDSCs also secrete immunosuppressive cytokines, such as IL-10, IL-1RA, and TGF-β, which inhibit T-cell activation, promote regulatory T-cell expansion, and polarize macrophages toward M2 phenotypes (6, 20). Chemokines such as CCL2 and CXCL8 recruit additional MDSCs and Tregs, thereby sustaining the suppressive milieu (21). MDSCs directly suppress T cells via programmed death-ligand 1 (PD-L1) expression and amino acid depletion, thereby reducing cytotoxicity and proliferation (22). MDSCs expressing PD-L1 engage PD-1 on T cells to induce anergy, whereas VISTA and TIM-3 drive anti-PD-1 resistance, and TIGIT/CD155 interactions inhibit NK cell IFN-γ across solid and hematological malignancies (6, 23, 24). MDSCs induce downregulation of the activating receptor NKG2D on both CD8+ T cells and NK cells via membrane-bound TGF-b and soluble NKG2D ligands (MICA/MICB, ULBP); and impair maturation and cross-presentation of dendritic cells via IL-10, NO, and oxidized lipid transfer (6, 25, 26). CD39 and CD73 ectonucleotidases in MDSCs generate adenosine, activating A2A/A2B receptors on T cells, NK cells, and macrophages to suppress their effector functions (27). This multifaceted crosstalk between MDSCs and other immune components promotes immune tolerance in the TME and facilitates tumor progression.

Immune checkpoint blockade (ICB) has been a promising avenue in immunotherapy in recent years; however, favorable treatment outcomes are often limited by the development of therapy resistance and a lack of durable response (157). Clinically, high intratumoral and circulating levels of MDSCs have been correlated with poor response to ICIs across multiple cancer types, including melanoma and non-small cell lung cancer (158). Anti-PD-1 therapy reduces MDSC infiltration in ICB-sensitive tumor models but fails in refractory tumors, highlighting the need for combinatorial strategies to sensitize resistant tumors and enhance ICB efficacy (159).

PD-L1-overexpressing NSCLC cells have been shown to enhance both the migratory capacity and immunosuppressive function of MDSCs by activating JAK2/STAT3 signaling, contributing to ICB resistance (160). Moreover, targeting the G-protein-coupled receptor GPR84 expressed on MDSCs with the selective GPR84 inhibitor GLPG1205 restores sensitivity to anti-PD-1 therapy in melanoma, lung cancer, and esophageal cancer models, highlighting MDSCs as a pivotal barrier to effective immunotherapy and an attractive adjunctive target to sensitize ICB-resistant tumors to ICIs (161). Treatment with artemisinin, an anti-malarial drug with immune-regulatory and anti-tumor effects, synergizes with anti-PD-L1 therapy in melanoma tumor models, inhibiting MDSC immunosuppression and accumulation and promoting T cell anti-tumor effects (162). Bezafibrate (BEZ), a peroxisome proliferator-activated receptor (PPAR) agonist with mitochondrial complex I/III inhibitory activity, displays potent synergism with anti-PD-1 therapy (163, 164, 166–168). More recently, an analog of BEZ (mito-BEZ) with increased mitochondrial localization significantly reduced both PMN-MDSCs and M-MDSCs, while increasing tumor infiltration of CD8⁺ T cells, highlighting BEZ as a potential combinatorial therapy with ICB to target MDSC-mediated immunosuppression on T cells (165). CD100ld, an emerging PMN-MDSC marker, facilitates their recruitment to tumors and T cell suppression (14). Blocking CD100ld in combination with anti-PD-1 treatment has a pronounced synergistic effect on tumors, suggesting a PMN-MDSC-specific approach for combinatorial immunotherapy (14).

Prokineticin 2 (BV8) is a key mediator in the recruitment of MDSCs to tumors (169, 170). Anti-BV8 treatment in preclinical models of lung, breast, and renal carcinomas reduces PMN-MDSCs by inhibiting their recruitment and promoting T cell activation, thereby sensitizing anti-PD-1-resistant tumors to anti-PD-1 therapy (170). Notably, MDSC-depleting anti-DR5 antibody MD5–1 synergizes with immune checkpoint inhibitors such as anti-PD-L1, depletes M-MDSCs with high DR5 expression via TRAIL-mediated apoptosis, and reduces tumor infiltration without affecting T cells or DCs, leading to significantly improved tumor suppression and durable immunological memory in syngeneic models of gastric and colon cancer (93). S-1, an oral 5-FU formulation, targets MDSCs by downregulating key immunosuppressive recruiters, thereby inhibiting MDSC accumulation in spleen/tumors without directly affecting MDSC survival/differentiation (171). Moreover, in AB1-HA mesothelioma and LLC lung models, S-1 suppresses tumor progression and MDSC infiltration, synergizes with anti-PD-1 to enhance CD8⁺ infiltration, and overcomes immunosuppression, supporting an avenue for combinatorial immunotherapy in thoracic cancers (171). Dual inhibition of CD39 (an adenosine-producing ectonucleotidase) and VISTA (a checkpoint ligand on MDSCs) reduces MDSC infiltration by 40–50% in syngeneic models, downregulating suppressive mediators, such as TGF-β and IL-10 (172). This strategy reverses CD8⁺ T-cell exhaustion, enhances IFN-γ and granzyme B levels, and synergizes with anti-PD-1 (172). By remodeling the TME to promote T-cell activity, the CD39/VISTA blockade offers a promising approach to overcome resistance to ICB and radiotherapy (172). Tyrosine kinase with Ig and EGF (epidermal growth factor) homology domain 2 (TIE-2), a receptor for the proangiogenic factor angiopoietin-2, is expressed on circulating m-MDSCs in melanoma patients, where TIE-2⁺ M-MDSCs overexpress immunosuppressive mediators and potently inhibit melanoma-specific T-cell responses (173). This TIE-2/angiopoietin-2 signaling axis represents a novel tumor immune escape mechanism, and combining TIE-2 inhibitors with ICIs shows promise in overcoming MDSC-mediated suppression in melanoma (173).

In the translational I-RENE trial, HLA-DR- PD-L1⁺ MDSCs were reduced in metastatic clear cell renal carcinoma (RCC) responders upon anti-PD1 nivolumab treatment; however, patients with progressive disease and non-responders had an enriched MDSC signature (174). These studies highlight the importance of targeting MDSCs in resistant tumors that fail to respond to ICB.

Recent preclinical studies have demonstrated that combining MDSC targeting with cancer vaccines and CAR-engineered immune cells can markedly enhance the antitumor efficacy in solid tumor models (175). Administration of an anti-Tumor mucin 1 (MUC1) vaccine in combination with tadalafil, a phosphodiesterase-5 (PDE5) inhibitor, significantly reduced both circulating and intratumoral MDSCs and Tregs while enhancing tumor-specific CD8 + T-cell responses in patients with head and neck squamous cell carcinoma (HNSCC) in a randomized clinical trial (169).

The efficacy of CAR-T cell therapy, especially third generation GD-2 specific CART cells, is hindered by high levels of peripheral PMN-MDSCs in neuroblastoma patients, illustrating the critical need to target MDSCs alongside adoptive cell therapy to maximize therapeutic outcomes (176). Preclinical evidence shows that pretreatment with the Bruton tyrosine kinase inhibitor ibrutinib can attenuate MDSC-driven immunosuppression of CART19 cell function in vitro (177). Moreover, when ibrutinib is combined with standard lymphodepleting chemotherapy, it further enhances the in vivo anti-tumor activity of CART19 cells by weakening the immunosuppressive TME (177).

A diverse set of small-molecule inhibitors provides additional combinatorial opportunities. For example, the PARP inhibitor olaparib disrupts the SDF-1/CXCR4 axis to impair MDSC recruitment, leading to superior tumor reduction when combined with EGFRvIII-targeted CAR-T cells (178). CAR-T cells engineered to co-express CXCR4 with CLDN18.2 block cancer-associated fibroblasts (CAF)-mediated recruitment of MDSCs, decrease infiltration of MDSCs, and increase CAR-T cell infiltration, ultimately enhancing the efficacy of CLDN18.2-positive PDAC models (179). The multi-targeted tyrosine kinase inhibitor lenvatinib lowers MDSC levels, impairs immunosuppression, boosts tumor-infiltrating T cell IFN-γ, and works synergistically with CAR-T therapy to reduce tumor burden (180). Anti-CD33 gemtuzumab ozogamicin depletes MDSCs and enhances the activity of anti-GD2-/mesothelin-/EGFRvIII-CAR T-cells (86). Activation of TLR3 by Poly I:C in conjunction with EGFRvIII-targeted CAR-T cells polarizes MDSCs toward tumoricidal M1 macrophages and produces durable synergistic anti-tumor responses (181).

Engineering CAR-T cells that engage MDSC-derived signals offers a dual-action approach that can robustly reprogram the TME (182). Dual-target CAR-T cells co-expressing mucin 1 (MUC1) and TRAIL-R2 (TR2.41BB) have improved efficacy by simultaneously targeting TR2 expressing MDSCs and optimally activating T cells via two co-stimulatory domains (CD28 and 41BB), supporting T cell persistence and expansion in the TME (183).

Fibroblast activation protein (FAP) is abundantly expressed by activated cancer-associated fibroblasts (CAFs) in pancreatic ductal adenocarcinoma, marking the desmoplastic stroma that impedes effective immunotherapy (184). In immunocompetent PDAC mouse models, sequential infusion of FAP-targeted CAR-T cells selectively eradicated FAP⁺ CAFs, leading to a pronounced drop in intratumoral CXCL12, which in turn disrupted MDSC recruitment and accumulation (185). This stromal remodeling creates a more permissive microenvironment for subsequent CLDN18.2-specific CAR-T cells, enhancing their infiltration, persistence, and antitumor activity at the tumor site (185).

In the CAR-NK landscape, NKG2D.ζ-engineered NK cells exploit high NKG2D ligand expression on MDSCs to selectively eliminate these suppressors, reprogram the cytokine milieu toward IFN-γ and TNF-α dominance, and rescue the efficacy of GD2-CAR-T cells in neuroblastoma xenografts (186).

Emerging CAR-macrophage (CAR-M) platforms, exemplified by anti-HER2 CAR-Ms, not only mediate the direct phagocytosis of tumor cells but also potentiate antigen presentation and pro-inflammatory cytokine release (187). Although clinical combinations with MDSC checkpoint inhibitors remain under exploration, theoretical models predict that blocking MDSC-mediated IL-10/TGF-β signaling will sustain CAR-M M1 polarization and reinforce durable antitumor immunity (187). CT-0508, a CAR-M platform, is currently undergoing phase 1 trials (NCT04660929) for HER2⁺ solid tumor malignancies (187).

Despite considerable progress in elucidating the immunosuppressive and tumor-promoting roles of MDSCs in the TME, several key challenges must be addressed in order to translate these insights into effective therapies. MDSCs exhibit profound subset-specific heterogeneity across both tumor types and individual patients, with variable phenotypic markers, immunosuppressive mechanisms, and metabolic reprogramming depending on the MDSC subsets, tissue type niche, complicating their identification and selective targeting. Hence, subset-specific analyses are crucial for refining therapeutic strategies aimed at MDSC depletion or functional inhibition. Global depletion of MDSCs risks unwanted immunotoxicity or compensatory expansion of therapy-resistant subsets. Conversely, targeting subset-specific markers (e.g., TIE-2 in M-MDSCs, CD100d in PMN-MDSCs) or metabolic dependencies (e.g., iNOS in M-MDSCs, NOX2 in PMN-MDSCs) could allow for precise modulation of immunosuppressive networks without compromising essential myeloid cell functions (Table 3). Further studies should therefore emphasize multi-parametric flow cytometry, single cell RNA sequencing, and metabolic profiling to clarify temporal evolution and functional plasticity of MDSC subsets within distinct microenvironments.

Within the evolving TME, MDSCs are capable of shifting their phenotypes in response to metabolic cues from the surrounding tumor or stromal cells (hypoxia, lactate) and resistance mechanisms following standard chemotherapies or immunotherapies, further hindering consistent phenotypic identification and selective targeting. Non-specific broad-range inhibitors targeting survival, recruitment, or differentiation of MDSCs risk collateral damage to essential myeloid populations (e.g., neutrophils and macrophages), potentially leading to neutropenia or impaired wound healing. To overcome these challenges, next-generation approaches are being developed to confine action to the tumor niche and to spare healthy myeloid compartments. For instance, antibody constructs that bridge MDSCs to cytotoxic T-cells only in the presence of tumor antigens (e.g., CD123 × CD3 BiTEs) promise minimal off-target myeloid depletion (195). Preclinical data suggest that CAR-T cells co-expressing receptors for MDSC‐derived signals (e.g., TR2.4-1BB) outperform unmodified CAR-T cells in solid models, but clinical translation is pending (183).

The lack of reliable blood or tissue markers to evaluate MDSC burden or functional state is a major barrier to patient selection and personalization of therapy. Although functional gene-expression signatures (such as S100A8/A9 and LOX‐1) are promising, robust prospective validation is lacking (196). Dynamic assays to monitor circulating MDSCs and their suppressive activity, along with the integration of advanced profiling technologies such as single-cell RNA sequencing, proteomics, spatial transcriptomics in patient biopsies, and imaging mass cytometry to spatially identify tumor-infiltrating MDSC signatures and immunosuppressive interactions within TME could allow precise stratification of patients and enable real-time adaptation of therapeutic regimens (197).

Efforts to enhance MDSC-targeted therapy specificity have focused on minimizing off-target effects and reducing systemic toxicities. Context-activated checkpoint inhibitors such as the novel pH-sensitive linker-based co-delivery of anti-PD-1 and Midkine-siRNA that unveil inhibitory domains only in the acidic TME potentially sharpen spatial precision to neutralize MDSCs (198). Protease-sensitive pro-antibodies whose Fc or Fab regions are masked until exposure to tumor-specific proteases could also restrict MDSC blockade to the malignant niche, reducing systemic toxicity (199). Concurrently, controlled delivery vehicles such as lipid or polymer-based nanoparticles and exosome-mimetic vesicles are being engineered to deliver metabolic modulators directly to intratumoral MDSCs, thereby enhancing the local potency while minimizing systemic toxicity (200).

Single-pathway blockade of MDSCs can trigger compensatory immunosuppressive circuits, necessitating multitarget regimens for durable responses (201). Combining MDSC inhibitors with immune checkpoint blockade and adoptive cellular therapies (such as CAR-T or CAR-M) is promising, but requires careful pharmacodynamic and scheduling optimization to avoid antagonism and maximize T cell priming (5). This intensified immune activation heightens the risk spectrum of adverse immune-related events, mandating routine immune monitoring and standardized mitigation protocols in clinical trials (202). Additionally, the complex crosstalk of MDSC subsets with Tregs, TAMs, and TANs reinforces the need for multi-targeted suppression of immunosuppressive and tumor-supporting networks within the TME.

The translation of advanced MDSC-targeting modalities will depend on embedding adaptive, biomarker-guided trial designs testing multiple MDSC-targeted approaches, subset-specific MDSC analyses and real-time monitoring (e.g., PET imaging of surface level and functional MDSC signatures) to stratify patients and personalize regimens. Integration of single-cell multi-omics profiling, artificial intelligence, deep learning platforms, and longitudinal monitoring will enable early identification of non-responders, predict prognosis in responders displaying MDSC signatures, and validate the safety and long-term efficacy of immune-metabolic targeting of MDSCs in solid and hematological malignancies.

MDSCs play a central role in shaping the immunosuppressive TME and represent a major barrier to the success of current immunotherapy. Advances over the past decade have greatly expanded our understanding of their ontogeny, metabolic adaptations, and functional interactions with other immune and stromal cell populations. These insights have catalyzed the development of multiple therapeutic strategies, including metabolic modulation, direct depletion, and inhibition of recruitment and signaling, and promotion of differentiation, many of which have shown strong preclinical efficacy and are advancing into clinical trials. However, significant challenges remain, including the heterogeneity and plasticity of MDSC populations, lack of standardized biomarkers, and need for precise, selective interventions that minimize collateral effects on other immune compartments. The rational integration of MDSC-targeting approaches with immune checkpoint blockade, cancer vaccines, and adoptive cell therapies holds particular promise for overcoming resistance and achieving durable tumor control. A combination of mechanistic insights from immunometabolism and spatial profiling, coupled with innovative therapeutic design and biomarker-guided clinical translation, will be essential to fully realize the potential of MDSC-targeted interventions. By dismantling these key suppressive networks, MDSC-focused strategies have the potential to broaden and deepen the responses to cancer immunotherapy and improve patient outcomes across a wide spectrum of malignancies.