Optimizing next-generation CAR-macrophages against solid tumors: challenges and potential strategies

Abstract

Chimeric antigen receptor macrophage (CAR-M) therapy has emerged as a highly promising novel platform in solid tumor immunotherapy. Leveraging its inherent tumor-homing capacity, potent phagocytic function, and potential to remodel the tumor microenvironment (TME), CAR-M offers a new strategic approach to address the limitations faced by CAR-T therapy in solid tumors, such as poor infiltration and immunosuppression. Despite these mechanistic advantages, clinical outcomes with first-generation CAR-M constructs have been modest, largely due to their limited in vivo persistence and effector activity. In this review, we summarize the core challenges limiting the efficacy and clinical application of CAR-M, and provide an in-depth discussion of engineering strategies aimed at enhancing its anti-tumor activity through optimization of the CAR molecular structure. Beyond CAR-M engineering, we also outline recent advances in combining CAR-M with other therapeutic modalities and discussing their underlying synergistic mechanisms. Looking forward, we highlight next-generation CAR-M platforms, such as in vivo edited CAR-M and CAR-monocytes, which aim to simplify manufacturing, reduce costs, and enable more precise immune modulation. Although challenges remain in manufacturing, durability of response, and safety, continuous technological innovation and rational combination strategies are accelerating the translation of CAR-M therapy from proof-of-concept toward clinical application, holding promise for opening new avenues in solid tumor treatment.

Introduction

Over the past decade, chimeric antigen receptor (CAR)-based immune cell therapy platforms have seen revolutionary advancements [1–3]. The efficacy of CAR-T therapy has improved markedly, while manufacturing strategies are evolving from autologous and allogeneic sources toward in vivo editing [4, 5]. Its clinical applications have also broadened from hematologic malignancies to include autoimmune diseases [6, 7]. However, consistent and robust clinical responses remain elusive in the highly challenging field of solid tumors, with only some clinical trials reporting encouraging results [8, 9]. The CAR is a synthetic transmembrane receptor whose basic structure includes an extracellular antigen-binding domain, a hinge region, a transmembrane domain, and an intracellular activation domain [10]. This modular design enables the rational engineering of CARs to directly mediate the antigen-dependent activation of immune effector cells [11, 12] (Fig. 1A). In these contexts, several alternative CAR-engineered immune cell candidates have emerged, such as CAR-NKT, CAR-NK cells, CAR-macrophages (CAR-M), and CAR-neutrophils [13–16] (Fig. 1B). Among these, macrophages have attracted considerable interest owing to their distinct tumor-homing capacity, potent phagocytic mechanisms, and their role as a functional bridge between innate and adaptive immunity [17].

Fig. 1.

CAR immune cell therapy: structural foundation, cellular platforms, and the evolution of CAR-M. (A) Overview of the basic structure and function of the CAR molecule. (B) Overview of CAR technology platforms: Major cell types and potential indications. The CAR immune cell therapy platform primarily comprises T-cells, natural killer T-cells(NKT), natural killer (NK) cells, regulatory T-cells (Tregs), dendritic cells (DCs), macrophages, and neutrophils. Its potential indications extend to a range of conditions driven by dysregulated pathological cells, such as cancers, autoimmune diseases, fibrotic disorders and infectious diseases. These conditions are often characterized by the presence of aberrant cells such as cancer cells, activated fibroblasts, and immune-dysregulated B cells. (C) Intergenerational evolution of CAR structure in CAR-M anti-tumor therapy. The intracellular domain design of CAR-M has evolved significantly, from first-generation constructs relying on CD3ζ or FcεRIγ for basic activation, to second-generation versions incorporating costimulatory domains such as TIR, and further to third-generation designs that integrate cytokines (e.g., IL-15, IL-21). This figure presents the generational progression of CAR-M structures and their functional enhancement pathways, highlighting the pivotal role of synthetic biology in the design of next-generation CAR-M

Advancing cell-based therapies for solid tumors requires overcoming a series of unique challenges, such as limited immune cell infiltration and persistence, an immunosuppressive tumor microenvironment (TME), tumor heterogeneity, and a lack of tumor-specific antigens [18]. These factors constrain the infiltration and survival of adaptive immune cells and drive them toward exhaustion [19]. In contrast, macrophages, a major immunosuppressive cell population in the TME of many solid tumors, can freely enter tumor sites due to their innate tumor-homing capability [20]. Moreover, their functional plasticity allows for therapeutic reprogramming [21]. Strategies such as inhibiting the CSF-1/CSF-1R signaling pathway [22], targeting the CCL2/CCR2 chemokine axis [23], and blocking phagocytosis checkpoints can shift tumor-associated macrophages (TAMs) from an immunosuppressive to an anti-tumor phenotype [24, 25]. Nevertheless, although such reprogramming strategies have shown promising results in pre-clinical studies, no related drug has yet gained regulatory approval to date, primarily due to limited efficacy or toxicity observed in clinical trials [26].

CAR-M therapy represents a novel immunotherapeutic strategy utilizing engineered macrophages [27]. In CAR-M therapy, macrophages target tumor cells via CARs and directly activate phagocytic activity. Notably, two mechanisms come into play during this process. One is trogocytosis, a relatively low-intensity process in which the macrophage “nibbles” on tumor cells to acquire and present antigens, resulting in limited cytotoxic effects. The other is phagocytosis, where the CAR-M extends pseudopodia to completely engulf the target, forming a phagosome that subsequently fuses with lysosomes. There, enzymes and an acidic environment thoroughly degrade the target [28, 29]. Subsequently, they remodel the TME by secreting cytokines and activating adaptive immunity, thereby achieving the suppression and eradication of solid tumors [29, 30] (Fig. 2). Currently, CAR-M has demonstrated superior anti-tumor activity in pre-clinical studies across various solid tumors, including refractory cancers such as ovarian cancer [31], glioma [32], renal carcinoma [33], liver cancer [34], and pancreatic cancer [35]. Concurrently, exploration into its application for non-oncological diseases is also advancing [36].

Fig. 2.

The anti-tumor mechanism of CAR-M. 1. CAR-M cells are recruited to tumor sites through homing mechanisms. 2.Upon arrival, they specifically recognize and bind to tumor cells expressing the target antigen. The hinge region provides spatial flexibility to the single-chain variable fragment (scFv), enabling it to effectively access and engage target antigens at varying heights on the tumor cell membrane. Meanwhile, the transmembrane domain anchors the CAR structure stably within the cell membrane. This binding triggers a cascade of membrane and cytoskeletal rearrangements, leading to the formation of a highly organized immune synapse centered on the CAR-antigen complex, between the CAR-M cell and the tumor cell. Synapse formation promotes CAR clustering, phosphorylation of immunoreceptor tyrosine-based activation motifs in the intracellular domains, and initiation of downstream activation signaling. 3. Subsequently, CAR-M cells release inflammatory cytokines, which remodel the TME by inducing tumor cell apoptosis and counteracting immunosuppressive cells. 4. CAR-M attacks tumor cells through phagocytosis and trogocytosis. 5. In parallel, antigen presentation and cytokine secretion activate T-cells, establishing an adaptive anti-tumor immune response. 6. Furthermore, epitope spreading may extend cytotoxic activity to tumor cells not originally targeted by the CAR. Together, these mechanisms form a potent pro-inflammatory and tumoricidal network within the tumor, enabling effective suppression and clearance of the lesion

Despite the progress made, CAR-M therapy still faces challenges in achieving significant therapeutic efficacy against solid tumors, with first-generation CAR-M treatments demonstrating only limited anti-tumor activity [37]. To enhance the functionality and activity of CAR-M cells, novel intracellular domain activating CAR-M cells with strengthened intracellular signaling domains have been developed, and advances have been made in identifying key genes essential for sustaining CAR-M activity, alongside the application of new synthetic biology technologies. Strategies involving combination therapies with existing treatments, such as monoclonal antibodies, oncolytic viruses, and other immunotherapies, are under evaluation. Furthermore, innovative therapeutic approaches, including CAR-monocytes and in vivo editing, have also been proposed. These strategies aim not only to improve the efficacy of CAR-M therapies but also to mitigate associated toxicities, thereby broadening their therapeutic applicability and ensuring treatment safety.

In this review, we comprehensively summarize the engineering strategies and methods for enhancing the activity and functionality of CAR-M, outline future directions for CAR-M development, and thoroughly discuss the determinants that will advance CAR-M technological progress and clinical applications.

Expanding anti-tumor spectrum and advancing clinical research

Since 2020, when Professor Saar Gill’s team first confirmed in an ovarian cancer model that CAR-M could serve as a potential adoptive immunotherapy based on the CAR platform [38], researchers have begun to explore the feasibility of CAR-M therapy across various solid tumor models. For example, Chen et al. [32] investigated in situ engineered CAR-M for preventing recurrent brain tumors; Dong et al. [39] provided the first evidence of the therapeutic potential of CAR-M derived from patient peritoneal macrophages in gastric cancer; Zheng et al. [35] examined the role of c-MET-targeted CAR-M in pancreatic cancer; and Jing et al. [33] established the efficacy of CAR-M therapy in renal carcinoma. In clinical trials, CAR-M therapy has primarily enrolled patients with advanced breast or gastroesophageal cancer (NCT04660929), advanced peritoneal metastatic cancer (NCT06224738), and refractory ovarian cancer (ChiCTR2400080078). Beyond solid tumors, CAR-M therapy has also been validated in hematologic malignancies. Jiang et al. [40] developed CD26-targeted CAR-M and confirmed its efficacy in a mouse model of chronic myeloid leukemia. These advances indicate a continuously expanding anti-tumor profile for CAR-M, laying a foundation for its future application in a broader range of cancer types (Fig. 3).

Fig. 3.

Anti-tumor spectrum of CAR-M at clinical and pre-clinical stages. Since the concept of engineering macrophages with CAR was first proposed in 2018, the field has advanced rapidly. The therapeutic spectrum of CAR-M has continued to expand, and both pre-clinical studies and clinical research in this area have progressed steadily

Recent clinical research data on CAR-M therapy provide critical insights (Table 1). In the first-in-human Phase I trial reported by Reiss et al. [37], 14 previously treated adult patients with metastatic HER2‑over-expressing solid tumors received intravenous infusion of CT-0508, an adoptive anti-HER2 CAR-M cell product, at doses up to 5.0 × 10⁹ CAR-M cells, administered as either a single or multiple infusions, without prior lymphodepleting chemotherapy. Among the 9 patients with HER2 3 + tumors, 4 achieved stable disease at the 8-week follow-up, whereas all other patients (5 with HER2 3 + and 5 with HER2 2+) experienced disease progression. No dose-limiting toxicities, severe cytokine release syndrome (CRS), or immune effector cell-associated neurotoxicity syndrome (ICANS) were observed. Some patients developed grade 1–2 CRS, which resolved within 72 h. Importantly, the anti-tumor mechanism of CAR-M is supported by clinical data, which demonstrate not only lesion reduction but also, through serial biopsy analysis, that CT-0508 can migrate to and remodel the TME. This remodeling was associated with expansion of CD8⁺ T-cells, suggesting the therapy enhances anti-tumor immunity via TME modulation, a key insight for optimizing future treatment strategies.

Table 1.

Clinical research and advances of CAR-M in anti-tumor therapy

Current status	Disease	Study start	CAR-M cell	Primary objectives/outcome	Lymphodepletion	Dose schedule	Progress achieved	Registration ID
Phase 1 Active, not recruiting	Solid tumors that overexpress HER2 metastasis	2021-02-02	HER2 targeted CAR-M (CT-0508)	• Evaluate the efficacy and safety of CT-0508 in patients with solid tumors.	No	A total of 5.0 × 10⁹ CT-0508 cells were administered via intravenous infusion (divided into single-dose and multiple-dose groups)	• Among the 14 patients who received two different treatment interventions, the SD was 28.6%	NCT04660929
Phase 1 Phase 2 Suspended	CD5 + Relapsed/Refractory T Cell Lymphoma	2021-12-15	CD5 targeted CAR-monocytes (MT-101)	• Test the safety, tolerability, and efficacy of the investigational agent MT-101 in patients with T cell Lymphoma.	Yes	Intravenous infusion MT-101	• In evaluable patients with relapsed/refractory PTCL, partial and complete responses were observed.	NCT05138458
Not Applicable Recruiting	Advanced Solid Tumors	2023-04-12	MSLN targeted CAR-M (SY001)	• Evaluate the safety, tolerability and initial effectiveness of SY001. λ Evaluate the pharmacokinetic characteristics, cytokines and the correlation between the efficacy of SY001.	No	Intravenously infused with 1.0 × 10⁹ CAR-M cells	• Two patients with ovarian cancer treated with SY001 showed stable disease and no serious adverse events.	NCT06562647
Phase 1 Recruiting	Relapsed/refractory OC with high HER 2 expression	2024-01-19	HER2 targeted CAR-M (RR-M01)	• Evaluate the safety and efficacy of RR-M01 in recurrent and refractory OC patients with HER2 overexpression.	No	Administered via intraperitoneal injection of 1.0 × 10⁹ CAR-M cells (repeated and sustained dosing)	• Among the 7 patients, RR-M01 had good tolerability and a DCR of 71.4%.	ChiCTR2400080078
Phase 1 Active, not recruiting	Solid tumors that overexpress HER2 metastasis	2024-01-08	HER2 targeted CAR-monocytes (CT-0525)	• Evaluate the safety, tolerability, and manufacturing feasibility of CT-0525 in HER2 positive solid tumor patients.	No	Not report	• Not report	NCT06254807
Early Phase 1 Not yet recruiting	Advanced peritoneal metastatic gastric cancer	2024-03-01	HER2 targeted CAR-M (MAC-001)	• Evaluate the safety and efficacy of MAC-001 in advanced HER2 + gastric cancer.	No	A single intraperitoneal infusion of CAR-M cells at a dose of 3.0 × 10⁸ cells, with subsequent escalating dose	• The first subject medication was completed in 2023.	NCT06224738
Phase 1 Recruiting	Advanced or metastatic GPC3-expressing cancers	2024-07-01	GPC3 targeted CAR-monocytes (MT-303)	• Evaluate the safety, tolerability and define the RP2D of MT-303 alone and in combination with Atezo/Bev in participants with advanced HCC expressing GPC3.	No	Intravenous injection of MT-303 (dose escalation)	• Not report	NCT06478693

Data were obtained from https://clinicaltrials.gov/ and https://www.chictr.org.cn/

CAR, chimeric antigen receptor; DCR, disease control rate; LNPs, lipid nanoparticles; OC, ovarian cancer; PTCL, peripheral T-cell lymphoma; SD, stable disease

At the 2025 European Society for Medical Oncology (ESMO) Congress, RocRock Biotech presented the latest clinical trial data for its CAR-M therapy product RR-M01. In a study involving seven patients with advanced HER2-positive solid tumors, RR-M01 monotherapy demonstrated promising efficacy signals, with a disease control rate of 71.4%. For patients previously treated with trastuzumab deruxtecan, RR-M01 continued to show significant therapeutic effects and exhibited potential efficacy in patients with HER2 2 + expression levels [41]. Additionally, in a separate study targeting mesothelin, a CAR-M candidate (SY001) exhibited a favorable safety profile in patients with ovarian cancer. Two treated patients with advanced disease experienced only transient cytokine release and neutropenia, and remained stable without disease progression during the 28-day follow-up period [42]. Although large-scale clinical validation and high-level evidence are still lacking, these clinical cases support the development of CAR-M as an effective immune cell therapy.

Biological challenges in CAR-M therapy

Current clinical data on CAR-M therapy have largely clarified its anti-tumor mechanisms. The combined effects of its tumor infiltration capacity, targeted phagocytosis, and activation of adaptive immunity collectively demonstrate therapeutic potential in solid tumors [37]. Nonetheless, there are still numerous challenges to overcome in advancing CAR-M as a breakthrough therapy for solid tumors. The core issues lie in the interactions between CAR-M and tumor cells along with their microenvironment, specifically how to enhance the efficiency and capacity of CAR-M in attacking tumor cells and how to foster a microenvironment conducive to the survival and activity of CAR-M [43].

Solid tumors possess distinct features that limit CAR-M efficacy (Fig. 4): (1) Although macrophages possess natural tumor tropism, the physical barriers formed by dense extracellular matrix and abnormal vascular structures hinder the infiltration of CAR-M cells and the recruitment of other immune cells [44]. While CAR-M cells themselves can secrete matrix metalloproteinases (MMPs) to degrade the extracellular matrix, this process is often blocked by the high expression of tissue inhibitor of metalloproteinase in the TME [45]. This further limits the secondary anti-tumor immune mechanisms triggered after phagocytosis by CAR-M cells.

Fig. 4.

Major biological limitations of CAR-M in anti-tumor therapy. The application of CAR-M in solid tumor therapy is hampered by tumor heterogeneity and the immunosuppressive microenvironment. Furthermore, the unique architecture of solid tumors significantly restricts a crucial anti-tumor mechanism of CAR-M: facilitating adaptive immunity. Additionally, safety concerns, including off-target distribution and off-target toxicity, remain critical challenges to be addressed

The immunosuppressive TME remains a major biological hurdle limiting the success of cancer immunotherapies. The inherent phenotypic plasticity of macrophages, while advantageous for engineering purposes, renders them highly susceptible to reprogramming into a pro-tumoral state by dominant suppressive signals within the TME, such as TAMs and inhibitory cytokines. This functional repolarization can reinforce the immunosuppressive niche, thereby fostering cancer progression and metastasis [46]. Specifically, the expression of phagocytosis checkpoints on engineered CAR-Ms significantly curbs their baseline phagocytic activity, which in turn impairs downstream antigen presentation and the secretion of pro-inflammatory cytokines [47]. Furthermore, metabolic stress within the TME imposes additional constraints; for instance, lactate accumulation and amino acid deprivation can suppress macrophage mTOR signaling and inflammatory function, leading to a state of metabolic quiescence or exhaustion [48].

Heterogeneous tumor cell populations with immune-escape mechanisms further compromise CAR-M therapeutic potential [49, 50]. For example, point mutations or deletions in the gene encoding the target antigen, or the production of variants lacking critical epitopes through alternative splicing, render the CAR incapable of recognition. In certain tumor types, under therapeutic pressure, tumor cells alter their differentiation state, shifting from a lineage that expresses the target antigen to one that does not [51, 52]. Additionally, the outgrowth of antigen-negative clones can lead to relapse [32].

Furthermore, safety concerns warrant attention, particularly off‑target toxicity and the pro‑tumor risk associated with immunosuppressive polarization [46]. As the scarcity of ideal tumor-specific antigens limits therapeutic targeting, CAR-M carries the risk of attacking normal tissues that express the target antigen. This risk is particularly pronounced in organs with high macrophage abundance, such as the lungs and liver, where such off-tumor recognition can lead to severe inflammation and tissue damage [53]. Although the underlying mechanisms of such off-tumor toxicity have been extensively studied in CAR-T therapy, the risk profile in macrophages is likely to be more complex, given that macrophages are potent secretors of inflammatory cytokines [54].

Overcoming functional challenges

Based on the concept of modular design, CAR-M cells can be dissembled into two components: the CAR molecule and the macrophage. Rational design targeting the modular CAR molecule and direct intervention on the macrophage itself may both lead to potentially beneficial changes. We will explore strategies to enhance CAR-M function at two levels: (1) optimizing the CAR molecule itself, including its structural configuration and the incorporation of additional functional payloads; (2) genetically engineering the macrophages themselves, for instance through metabolic reprogramming or the introduction of synthetic receptors.

Unlike T-cells, which require stringent activation procedures, macrophages do not rely on specific co-stimulatory signals. Therefore, the inclusion of co-stimulatory domains in CAR-M is optional, and their precise functions and mechanisms of action have yet to be definitively studied. Researchers often use CD3ζ or FcR alone to activate macrophage functions, rather than employing the fixed combination of CD28-CD3ζ or 4-1BB-CD3ζ commonly used in CAR-T studies [55, 56]. The design of CAR structures in CAR-M has evolved through several generations, with the core goals of enhancing functionality and optimizing safety, particularly in endowing CAR-M with sustained anti-tumor activity and overcoming suppressive microenvironments (Fig. 1C).

The intracellular domain is the core of CAR molecular structural modification, directly determining the fate and behavioral programming of host cells upon activation [57]. Rational design and structural modification of intracellular activation domains are primary approaches to optimizing the safety, efficacy, and applicability of CAR immune cell therapies [58, 59]. CAR-M have now progressed through three iterations, largely driven by innovations in their intracellular domains. Building on advances in related platforms such as CAR-T and CAR-NK cells, a fourth generation of CAR-M therapy is now within reach. This evolution underscores the growing role of synthetic biology in CAR-M development, where traditional genetic engineering approaches fall short of meeting the need for smarter, controllable, and multifunctional systems. Synthetic biology offers modular genetic circuits, logic-gated CARs, inducible expression systems, and engineered synthetic receptors, thereby enabling the design of next-generation CAR-M cells capable of sensing and responding to complex signals within the TME [60–62].

The intracellular domains of first-generation CAR-M are typically derived from canonical signaling molecules associated with antibody-dependent cellular phagocytosis (ADCP), such as FcεRI-γ and its homologue CD3ζ. Studies have confirmed that these domains can activate the phagocytic function of CAR-M. For instance, Klichinsky et al. [38] constructed CAR-M using CD3ζ as the intracellular domain, demonstrating for the first time in animal models that CAR-mediated targeting by macrophages leads to tumor cell phagocytosis and a reduction in tumor burden. Multiple studies have also validated the activating role of FcεRI-γ as an intracellular domain in CAR-M, showing anti-tumor effects in various solid tumors including gastric, ovarian, and breast cancers [39, 63]. In current clinical research, products such as CT-0508 and SY001 belong to this first-generation CAR-M. Although some therapeutic effects have been observed in clinical trials, their overall anti-tumor efficacy remains unsatisfactory [37, 42].

For the macrophage itself, genetic engineering techniques are commonly employed to reprogram its metabolism or phenotype, thereby enhancing its survival, polarization, antigen presentation, and other functions within the TME [64]. The modular design concept endows CAR-M therapy with remarkable flexibility. By freely combining CAR molecules with distinct functions and macrophages pre-conditioned through various methods, researchers can design and construct customized CAR-M cells tailored to specific disease types and therapeutic goals to achieve optimal efficacy (Table 2).

Table 2.

Overview of the intracellular domain of CAR-M

Intracellular domain/payload	Pathway engaged	Phenotype effect	Key risks/Deficiencies	Best-fit tumor context
CD3ζ	Phosphorylated CD3ζ-ITAM recruits SYK, thereby initiating a cascade of downstream signaling events	Does not directly affect the CAR-M phenotype	There is a risk of being reprogrammed by the tumor microenvironment into a pro-tumor phenotype; its generated activity is limited, necessitating combination with other therapies	As an immunomodulatory agent, it provides a pre-environment for subsequent immunotherapy
CD147	Promoting the secretion of MMPs, possibly involving the FAK signaling pathway	Does not directly affect the CAR-M phenotype	No direct tumor-killing activity; MMPs degrade the basement membrane, which may release tumor cells and promote metastasis	“Hard” tumors rich in dense ECM
FcεR1γ	SYK and the PI3K/AKT pathway	Antigen-dependent M1 polarization	Insufficient maintenance of the M1 phenotype; limited duration of effect	Clearance of peritoneal or local residual lesions
IFN-γR1/IFN-γR2/TLR4	NF-κB and the JAK-STAT1 pathway are activated	Antigen-dependent M1 polarization	Risk of CAR-M overactivation leading to CRS; sustained IFN-γ signaling may lead to T cell exhaustion	“Cold” tumors that need to break immune suppression
Megf10-CD3ζ	Activating signaling pathways such as NF-κB, ERK, and STAT1 while suppressing the STAT3 pathway.	Favors M1 polarization and resists M2 polarization	Constitutive M1 activation-associated metabolic burden; non-native signaling combinations require further validation	Immunosuppressive tumors with high and uniform expression of the target antigen
TIR-CD3ζ	Activating the TLR4-NF-κB signaling pathway	Sustained antigen-dependent M1 polarization and remodeling of the TME	Risk of excessive activation of CAR-M-mediated cytokine storm; metabolic exhaustion of macrophages	Solid tumors with abundant TAM infiltration
IFN-γ	IFN-γ-JAK-STAT1 pathway	IFN-γ drives M1 polarization, remodeling the TME into a pro-inflammatory state	Sustained expression of IFN-γ may trigger autoimmune responses or CRS	Solid tumors with abundant TAM infiltration
IL-15	InvolvingJAK-STAT5, MAPK/ERK, and mTOR signaling pathways	Does not directly participate in the remodeling of the CAR-M polarization state	Limited penetration and killing capacity for larger metastatic lesions; chronic activation leads to exhaustion of CAR-M and other immune cells	Tumors with NK cell infiltration but function suppressed by the TME
IL-21	IL-21-JAK-STAT, PI3K-AKT and ERK/MAPK signaling pathways	Maintaining the pro-inflammatory phenotype	Massive activation of T cells can trigger CRS; IL-21 may also activate Treg cells or induce T cell exhaustion	“Cold” tumors with T cell infiltration but functional inhibition
CD47 inhibitor	Blocking the CD47-SIRPα axis	Does not directly affect the CAR-M phenotype	Hematologic toxicity; non-specific inflammation and tissue damage caused by macrophage activation in normal tissues	Solid tumors with high antigen heterogeneity and high CD47 expression
STING agonist	activation of the cGAS-STING pathway	Promoting M1 polarization	Overactivation of STING may lead to suppression of mRNA translation by type I interferons, decreased CAR expression, or cytotoxicity	Solid tumors with abundant TAM infiltration

CAR-M: Chimeric antigen receptor macrophage; cGAS: cyclic GMP-AMP synthase; CRS: Cytokine release syndrome; ECM: Extracellular Matrix; ERK: Extracellular regulated protein kinases; IFN-β: Interferon beta; IFN-γ: Interferon gamma; ITAM: Immunoreceptor tyrosine-based activation motifs; MAPK: Mitogen-activated protein kinase; MMPs: Matrix metalloproteinases; NF-κB: Nuclear factor kappa-B; TAM: Tumor-associated macrophage; TIR: Toll-like receptor 4 intracellular toll/IL-1R; TLR4: Toll like receptor 4; TME: Tumor microenvironment; STING: Stimulator of interferon genes; SYK: Spleen tyrosine kinase

Sustaining the pro-inflammatory phenotype of CAR-M

Macrophage polarization is a highly dynamic and reversible process. Under physiological conditions, the M1-like pro-inflammatory phenotype mediates pathogen clearance and anti-tumor immunity, whereas the M2-like anti-inflammatory phenotype facilitates tissue repair, inflammation resolution, and immune regulation [65]. This plasticity is essential for macrophage-mediated tissue homeostasis but presents a dual challenge in CAR-M therapy. For effective solid tumor treatment, CAR-M products should maintain a sustained M1-like phenotype to ensure robust tumor cell killing and efficient priming of adaptive immune responses. However, the TME, enriched with M2-polarizing cytokines such as IL-4, IL-10, and TGF-β, can redirect infiltrating macrophages toward a pro-tumorigenic state [66]. This not only compromises the direct effector functions of CAR-M but also risks their conversion to a tumor-promoting phenotype. Therefore, a central challenge for CAR-M therapy in solid tumors is to preserve their pro-inflammatory activation within this suppressive milieu [30]. Sustaining M1-like effector functions, including potent phagocytosis of tumor cells, secretion of pro-inflammatory cytokines, and effective antigen cross-presentation, is critical for ensuring durable anti-tumor efficacy.

Development of second-generation pro-inflammatory CAR-M

Regarding the design of CAR molecules, the first-generation CAR-M designs, which incorporated phagocytosis-associated signaling domains within their intracellular regions, failed to fully exploit the pro-inflammatory potential of macrophages. Consequently, the development of second-generation CAR-M has focused on enhancing their overall activity and pro-inflammatory function to better counteract the immunosuppressive TME. Lei and colleagues developed a CAR-M therapy based on toll-like receptor 4 intracellular toll/IL-1R (TIR) domain activation [30]. TIR is a key transduction structure in the TLR4 signaling pathway, capable of promoting the expression of pro-inflammatory cytokines, chemokines, and tumor necrosis factors [67]. The tandem intracellular domain of CD3ζ-TIR demonstrated robust pro-inflammatory activation and maintained the pro-inflammatory state of CAR-M in various solid tumor models, showing superior activity compared to the first-generation CAR-M. Similarly, Duan et al. [68] proposed the idea of designing pro-inflammatory CAR-M by leveraging the TLR4 inflammatory signaling pathway and IFN-γ signaling. They constructed CAR-M with TLR4 and IFN-γR as intracellular domains, which, upon antigen stimulation, could ameliorate the TME by secreting inflammatory cytokines.

In CAR-T therapy, the incorporation of costimulatory domains (such as CD28 and 4-1BB) provides a complete signal for CAR-T cell activation, addressing the core bottleneck of insufficient in vivo expansion and poor persistence of CAR-T cells and thereby promoting the clinical translation of this therapeutic approach [69, 70]. However, whether the presence of costimulatory domains plays a decisive role in CAR-M therapy remains unclear. Currently, most CAR-M cells in pre-clinical and clinical studies are equipped only with activation domains, such as CD3ζ, which are already sufficient to deliver adequate activation signals. Recently, Ibrahim et al. [71] provided new insights into CAR-M activity based on costimulatory domains. They found that adding the CD86 costimulatory domain to the CAR structure, while not significantly enhancing the direct tumor-killing capacity of CAR-M cells, led to increased CD86 expression on the surface of CAR-M cells, enhanced M1 polarization and activation of cytotoxic T lymphocytes, along with elevated secretion of inflammatory cytokines such as IFN-γ and TNF-α. This study clarified the potential role of costimulatory domains in CAR-M therapy, demonstrating their ability to influence the polarization state of CAR-M cells, enhance adaptive immunity, and remodel the TME.

Metabolic reprogramming mediates inflammatory phenotype remodeling

In addition to modifying the CAR molecular structure, interfering with the expression of key genes in macrophages, such as metabolism-related genes and regulatory molecules of critical signaling pathways, serves as an effective approach to enhancing CAR-M activity and maintaining the inflammatory phenotype [72](Fig. 5A).

Fig. 5.

Other enhancement strategies for CAR-M. (A) Mechanisms of gene editing to enhance CAR-M related signaling pathways, such as knocking out ACOD1 and inhibiting furin. (B) CAR-M with different extracellular domain designs to overcome immune escape, including tandem, bicistronic, and sequential CAR-M

The functional polarization of macrophages is tightly coupled with the reprogramming of their core metabolic pathways [73]. Metabolic status is not only a consequence of energy supply but also a key driver that determines and sustains their functional phenotypes [72]. Existing research indicates that pro-inflammatory macrophages primarily rely on glycolysis, characterized by impaired tricarboxylic acid cycle activity and expression of inducible nitric oxide synthase, whereas alternatively activated macrophages predominantly depend on mitochondrial oxidative phosphorylation [74]. This provides an opportunity to intervene in macrophage phenotype remodeling driven by metabolic reprogramming. Wang et al. [75] employed CRISPR-Cas9 screening to identify the key metabolic regulatory signaling pathway ACOD1/KEAP1/NRF2 involved in pro-inflammatory activation of macrophages. By knocking out ACOD1, the level of the immunometabolite itaconate was reduced, allowing KEAP1 to prevent NRF2 from entering the nucleus and activating anti-inflammatory programs. This resulted in sustained pro-inflammatory activation of CAR-M, enhanced phagocytosis, and increased cytotoxicity against cancer cells.

Beyond the well-established aconitate decarboxylase 1 (ACOD1), other gene editing targets that demonstrate transformative potential in engineered macrophages may achieve similar effects in CAR-M therapy. For example, research by Wang et al. [76] revealed that Parkin regulates macrophage immune phenotypes via a non-canonical AMPK pathway, linking mitochondrial stress to dysregulated antigen presentation. Knockout of the Parkin gene can activate both innate and adaptive immunity, thereby remodeling the immunosuppressive TME. Additionally, Park2-knockout mice exhibited increased sensitivity to PD-L1 monoclonal antibody treatment, suggesting its potential for combination therapy with immune checkpoint inhibitors. For adoptive CAR-M therapy, metabolism-related genes such as ACOD1 and Park2 represent suitable intervention targets, as they can reprogram macrophage metabolism through different mechanisms, effectively “locking” them into a more potent anti-cancer state. Given the complexity and redundancy of metabolic networks, more effective strategies in the future may involve combined editing of multiple complementary metabolic nodes [77, 78].

Additionally, modulating key signaling pathways represents a promising strategy to maintain CAR-M cells in an inflammatory state. For example, proprotein convertases (PCs) are serine proteases that cleave inactive protein precursors into their bioactive forms within cells. In macrophages, PCs regulate the maturation of critical cytokines and chemokines, such as TNF-α, IL-1β, and TGF-β, through their proteolytic activity [79]. Among them, furin modulates T-cell adaptive immune responses and contributes to macrophage phenotypic differentiation. Studies have shown that furin deletion promotes inflammatory polarization in macrophages. Building on this, Ziane-Chaouche et al. [80] used siRNA to knock down furin expression in CAR-M cells, demonstrating that furin silencing promotes inflammatory activation and significantly enhances the tumor-killing capacity of CAR-M cells. The latest study by Wang et al. [81] revealed a post-transcriptional regulatory mechanism in TAMs mediated by RNA modification. The tRNA m¹A methyltransferase TRMT61A is highly expressed in pro-inflammatory macrophages. Loss of TRMT61A disrupts STING protein translation and diminishes IFN-β secretion, thereby promoting tumor growth. Conversely, over-expression of TRMT61A significantly augments the anti-tumor efficacy of CAR-Ms through activation of the STING-IFN-β pathway.

In summary, these gene-editing approaches offer new perspectives for CAR-M design by reprogramming endogenous gene expression to control macrophage phenotype and improve function. Furthermore, combining gene editing with optimized CAR signaling components, such as integrated co-stimulatory domains, could synergistically steer an ideal CAR-M phenotype through both intrinsic metabolic and extrinsic signaling pathways. This dual approach may help sustain optimal CAR-M activity within the complex TME, ultimately advancing therapeutic outcomes for solid tumors.

Persistence and Functional Maintenance of CAR-M in the Immunosuppressive TME

Development of third-generation cytokine-enhanced CAR-M

To more fully harness the functional capabilities of CAR-M cells, enhance their communication with other immune cells, and mobilize the immune population within the TME, researchers exploring third-generation CAR-M therapies have incorporated specific cytokines or secretory proteins into the CAR construct. For instance, Shah et al. [82] integrated mIL-15 into their CAR design, which not only conferred anti-apoptotic protection to CAR-M cells but also activated NK cells and T-cells. Their findings demonstrated that IL-15-expressing CAR-M cells could persistently eliminate tumor cells both in vitro and in vivo, bolstering CAR-M-mediated intervention in the immunosuppressive TME and triggering a potent adaptive immune response. In a separate study, Chen et al. [83] developed a CAR molecule co-expressing IL-21. Compared with first-generation CAR-M, the inclusion of IL-21 substantially remodeled the solid TME by sustaining an inflammatory phenotype, through enhanced macrophage phagocytic activity and suppression of M2 polarization, while simultaneously activating T-cells and NK cells. This approach significantly improved CAR-M-mediated regulation of the immunosuppressive TME and reduced tumor burden in mice models.

Enhancement strategies based on synthetic receptors

With the rapid advancement of synthetic biology, beyond modifications to CAR molecular structures and the co-expression of secreted proteins (such as cytokines and antibodies) [84], the co-expression of other synthetic membrane proteins has emerged as a significant frontier in CAR‑M research. This strategy enables the enhancement of engineered cell function and survival through the expression of specific surface receptors. Furthermore, the design of chimeric receptors can induce selective macrophage activation by triggering specific intracellular components.

Typically, the co-expression strategy of cytokine or chemokine receptors has been extensively explored in CAR immune cell therapy platforms and has achieved phased progress [85]. For example, Lin et al. [86] conducted a Phase I clinical trial for pediatric high-grade central nervous system (CNS) tumors, introducing a constitutively active IL-7 receptor to enhance the anti-tumor activity and persistence of CAR-T-cells in CNS tumors. The results showed that co-expression of the IL-7 receptor could persistently activate the STAT5 pathway without exogenous IL-7. Compared with CAR-T monotherapy, patients in the IL-7 receptor co-expression group exhibited longer-lasting neurological improvement and significantly prolonged progression-free survival. Beyond directly co-expressing specific cytokine receptors (such as IL-7R, IL-15R, etc.) to receive positive signals and enhance immune cell survival, another approach to improve the function and persistence of CAR-based immune cells in the suppressive TME involves using dominant-negative receptors to interfere with or block immunosuppressive signals [87]. For example, to counteract the inhibitory effects of TGF‑β on CAR-T-cells, Li et al. [88] developed a CAR-T therapy co-expressing dnTGFβRII, which renders CAR-T-cells insensitive to the negative regulatory signals of TGF‑β, thereby improving their survival and function in solid tumors.

Beyond natural receptors, synthetic receptor engineering, enabled by advances in synthetic biology, has opened new directions for CAR-based immunotherapy. For example, Zheng et al. [89] developed an inverse cytokine receptor by fusing the extracellular domain of TGF-β receptor II with the intracellular domain of IL-15 receptor α, termed TB15 ICR. Upon TGF-β binding, this synthetic receptor converts the normally inhibitory signal into an IL-15–mediated activating signal, enhancing CAR-T-cell proliferation and function. Similarly, in CAR-M research, Jing et al. [33] designed an IL-2R-TLR4 chimeric signaling receptor (CSR) by combining the extracellular segment of the IL-2 receptor with the intracellular signaling domain of TLR4. Stimulation with exogenous IL-2 activates CSR-mediated pro-inflammatory pathways, including NF-κB downstream of TLR4, thereby stabilizing the inflammatory phenotype and improving the anti-tumor function of CAR-M cells. Furthermore, chimeric receptors can be engineered to target negative regulators such as IL-4, IL-10, and TNF-β, while incorporating intracellular domains from inflammatory signaling molecules like TLR4 or IFN-γ. When expressed in engineered macrophages, these receptors help sustain an activated inflammatory state, potentially reversing the immunosuppressive TME under continuous ligand exposure. This signal-reprogramming concept also extends to non-oncological applications. In inflammatory diseases, for instance, a TNF-α-targeted CAR containing the intracellular domain of IL-4Rα has been shown to redirect macrophages from pro-inflammatory TNF signaling toward immunosuppressive and tissue-repair pathways upon trimeric TNF stimulation [90]. These findings demonstrate that CAR-mediated signal switching can effectively reprogram macrophage function, yielding significant anti-inflammatory effects in both acute and chronic inflammatory contexts.

Evidently, such chimeric signaling receptors can be triggered not only by endogenous immunosuppressive factors but also by exogenously infused cytokine signals, enabling precise and dynamic control over immune cell functions [91]. These synthetic receptors not only alter the source of signal input but, more importantly, redefine the cellular output behaviors. They can be engineered to perform complex functions beyond the natural capabilities of immune cells, significantly enhancing the adaptability, persistence, and cytotoxic activity of therapies such as CAR-T and CAR-M in hostile TMEs [92, 93]. This innovation paves the way for the next generation of smarter and more potent cellular immunotherapies.

It should be noted that excessively enhancing the activity and persistence of CAR-M may increase the risk of off-target toxicity or CRS [94]. This necessitates stringent dose control or the incorporation of logic-gated safety switches. Furthermore, introducing potent artificial signaling pathways, such as TLR4/NF-κB, could disrupt the macrophage’s intrinsic signaling networks, potentially impairing normal functions including metabolism, differentiation, and memory formation. From a technical standpoint, the stable and efficient co-expression of multiple exogenous genes in CAR-M places greater demands on vector systems, such as lentiviral vectors, lipid nanoparticle (LNP)-mRNA, or transposon systems [95]. Future advancements will depend on more sophisticated design strategies (e.g., logic gates, inducible switches), stricter control over targeting (e.g., using TME-specific promoters), and more rigorous pre-clinical safety evaluations [96]. These measures are essential to ensure therapeutic efficacy within a safe and controllable framework.

Overcoming tumor antigen heterogeneity and immune escape

The antigen expression of solid tumors exhibits significant spatiotemporal heterogeneity, which is one of the core challenges in CAR-M therapy. Intratumoral heterogeneity is characterized by varying levels of target antigen expression among different cells within the same tumor tissue, including a population of cells that are negative for the target antigen [97]. When CAR-M cells eliminate tumor cells expressing the target antigen, these antigen-negative or low-expressing cells gain a selective advantage, leading to tumor recurrence. Furthermore, intertumoral heterogeneity means that different metastatic lesions in the same patient may display distinct antigen expression profiles, making it difficult for a single-target CAR-M therapy to achieve comprehensive coverage [98]. Adding to the complexity, tumor cells evolve multiple immune evasion mechanisms to actively circumvent recognition and clearance by macrophages, such as expressing phagocytosis checkpoint signaling molecules to block macrophage phagocytosis [99].

While intracellular domain optimization is crucial, extracellular elements and costimulatory domains also play pivotal roles in CAR immune cell functionality [100]. The antigen-binding domain serves as the extracellular core region of the CAR molecule. In current CAR-based immune cell therapies, the antigen types under development primarily include glycan antigens and protein antigens [101]. For CAR-M, currently validated antigens are predominantly focused on protein antigens, such as surface receptors like HER2 and EGFR [31, 68], as well as transmembrane proteins like CD19 and c-Met [35]. Additionally, glycan antigens such as GPC3 and GD2 are also under development [30]. Owing to the shared structural framework of CAR-based immunotherapy platforms, the target antigen spectrum for CAR-M therapy is expected to broaden in the future [102].

Typically, researchers use the single-chain variable fragment (scFv) of corresponding antibodies to achieve targeted binding [103]. Unfortunately, truly tumor-specific antigens that are exclusively expressed on the surface of tumor cells and absent on normal cells are rare, which creates a opportunity for tumor immune escape [104]. To address this limitation, two extracellular design strategies have emerged as promising solutions (Fig. 5B). The first is the tandem CAR design, which incorporates two distinct scFvs within a single CAR molecule [105]. This architecture broadens target recognition to heterogeneous cell populations expressing different antigens, thereby mitigating immune escape due to downregulation of a single target. A potential drawback, however, is that steric hindrance or conflicting affinities between the two scFvs may reduce overall binding efficiency [106]. Another related strategy is the bicistronic CAR approach, wherein two independent CAR molecules are co-expressed on the immune cell surface [107, 108]. Since the CARs are expressed separately, interference between them is minimized. This design also offers flexibility, allowing the use of different promoters and intracellular signaling domains to orchestrate complex activation profiles. Nevertheless, expressing two full-length CAR genes places greater demands on vector design and requires precise regulatory control to ensure balanced and uniform functionality [109]. Additionally, both approaches demand advanced engineering techniques and necessitate greater investment and pre-clinical research to establish their safety and efficacy.

Beyond engineering strategies, adjusting therapeutic approaches by adopting sequential therapy also represents a potential rescue strategy for immune escape [110]. Sequential therapy refers to the administration of a second or multiple reinfusions of CAR immune cells, targeting either the same or different antigens, after a patient shows no response or experiences relapse following initial CAR immune cell treatment [111]. This approach aims to address the limitations of single-cell therapies, such as tumor immune escape and insufficient cytotoxicity [112, 113]. Although not yet validated in CAR-M, sequential therapy in CAR‑T has been confirmed to improve immune escape and patient prognosis [114, 115]. It is foreseeable that this strategy also holds potential application value in the future development of CAR‑M.

Macrophage-specific CAR and next-generation smart CARs

It is noteworthy that current CAR design strategies are primarily guided by CAR-T cell engineering, and the design of macrophage-specific CAR molecules remains insufficient. T-cell activation relies on the phosphorylation of immunoreceptor tyrosine-based activation motifs (ITAMs) within the CD3ζ chain, as well as co-stimulatory signals such as CD28 or 4-1BB, to drive proliferation, survival, and cytotoxic functions [116]. However, the effector functions of macrophages, such as phagocytosis, pro-inflammatory cytokine release, and antigen presentation, are regulated by a distinct set of signaling pathways. Therefore, to fully harness the therapeutic potential of CAR-Ms, it is essential to move beyond the traditional T-cell signaling paradigm and integrate the latest technological advancements to design next-generation smart CAR immune cells optimally suited for macrophages.

Development of macrophage-specific CARs

The pioneering study by Morrissey and colleagues revealed that the CD3ζ domain of CAR-T can also induce phagocytosis in macrophages, with efficacy comparable to that of Megf10 (a phagocytosis-associated signaling molecule) [29]. In-depth mechanistic studies demonstrated that, in T-cells, the ITAMs within CD3ζ bind to the tandem SH2 domains of the kinase ZAP70 to transmit activation signals. In macrophages, although ZAP70 is absent, the highly expressed spleen tyrosine kinase (SYK), which also contains a tSH2 domain, can bind to phosphorylated CD3ζ and trigger downstream phagocytic signals. This suggests that ITAM signaling domains may possess cross-cell-type universality, provided they can couple with downstream effector molecules.

Macrophages express various natural receptors responsible for recognizing and eliminating targets, and their signaling mechanisms could offer new insights for CAR design. While the first-generation CAR-M utilized the FcεRI-γ chain, other FcR, such as CD32a, contain intrinsic ITAM motifs and are coupled with specific downstream signaling molecules, effectively mediating phagocytosis and ADCP [117]. Future macrophage-specific CARs could integrate these distinct ITAM domains, potentially in combination, to finely regulate phagocytic efficacy and inflammatory responses.

For instance, Zhen et al. [55] designed an intracellular domain that fused megf10 with CD3ζ. In this construct, phagocytosis is driven by SYK activation mediated by Megf10, while CD3ζ synergistically amplifies ERK signaling alongside NF-κB and STAT1 pathways. This combined action enhances pro-inflammatory signaling while actively suppressing STAT3-mediated immunosuppressive signals, ultimately achieving stable M1 polarization of CAR-M and remodeling the TME. Similarly, research by Lei et al. [30] has provided successful examples of combining TIR domains with CD3ζ. Notably, the assembly of phagocytic signaling domains may yield different outcomes depending on the disease context. For example, CAR-M constructs incorporating MERTK did not induce specific phagocytosis of tumor cells, although studies have found that this intracellular domain can mediate endocytic activity against viral particles [118]. These findings underscore the importance of engineering intracellular signaling domains tailored to the functional requirements of specific disease types.

Intelligent CAR-M therapy platform

Furthermore, for the next generation of immune cell therapies, there is a gradual shift from “living drugs” to integrated “drug factories,” with modular CAR molecular designs providing a window into this vision [119]. These payloads can intervention of immunosuppressive cells or factors, ablation of the tumor stroma, recruitment of anti-tumor cells, and direct enhancement of CAR immune cell activity and function [120]. In CAR-M, the co-expression of membrane surface molecules or secreted proteins, such as cytokines or chemokines and their receptors, immunomodulatory antibodies, bispecific adaptor molecules, and other therapeutic proteins, accelerates this trend. For example, Chen et al. [121] incorporated an endogenously secreted CD47 blocker into the CAR molecule, which synergizes targeted phagocytosis with CD47 blockade, enhancing the clearance of tumor cells with differential antigen expression. Han et al. [122] proposed a strategy of co-expressing stimulator of interferon genes (STING) agonists to enhance the pro-inflammatory activation of CAR-M through the cyclic GMP-AMP synthase (cGAS)-STING signaling cascade, thereby remodeling the TME and eliminating tumor cells. In summary, the ultimate goal of these designs is to evolve CAR-M cells from relatively simple cytotoxic tools into a synthetic biology platform with programmable functions and controllable behavior. Despite significant challenges, this represents an essential pathway to overcoming the bottlenecks in solid tumor therapy and achieving safer, more durable therapeutic efficacy.

Design Principles of CAR-M

In summary, distinct from CAR-T cells, which are designed for sustained proliferation and cytotoxicity, the design of CAR-M must be centered around the unique biological properties of macrophages, such as phagocytosis, polarization, antigen presentation, and transient survival. The future of CAR-M design will place greater emphasis on coordinating the entire CAR molecular system with macrophage functions. It is not merely about engineering macrophages but rather about designing and constructing a living cellular drug system capable of sensing TME signals, integrating multiple instructions, executing coordinated functions, and incorporating built-in safety redundancies.

Target selection must balance specificity and density

Target selection for CAR-M therapy affects not only efficacy but also the risk of on-target/off-tumor toxicity. Tumor-specific antigens, such as EGFRvIII which is absent in normal tissues [123], should be prioritized due to their favorable safety profile. Tumor-associated antigens require cautious evaluation, necessitating a comprehensive assessment of their expression patterns in normal tissues and associated toxicity risks. Furthermore, clinical data from CT-0508 highlight the influence of antigen expression levels on therapeutic outcomes, responses were observed exclusively in patients with HER2 3 + expression [37]. Therefore, antigen density must be considered to avoid therapeutic failure due to insufficient density or exacerbated toxicity resulting from excessive density.

The Choice of Signaling Domain Determines Functional Status

The intracellular signaling domain of CAR-M is central to its effector functions and polarization state. Research indicates that different signaling domains confer distinct functional properties on CAR-M. Accordingly, the selection or combination of signaling domains should align with the intended therapeutic objective, such as promoting phagocytosis, remodeling the extracellular matrix, or activating immune responses. While single signaling domains often yield limited effects, the use of dual tandem signaling domains and the incorporation of cytokine-derived signals (e.g., CD3ζ-Megf10, CD3ζ-TIR) represent a prevailing strategy to enhance CAR-M functionality [124].

Regulation of the polarization state is key to balancing efficacy and safety

The high plasticity of macrophages presents both advantages and risks. CAR design must consider how to enable CAR-M to maintain an anti-tumor phenotype within the TME and resist immunosuppression. An ideal CAR design should allow activation and maintenance of the M1 phenotype specifically at tumor sites, while remaining quiescent or minimally active in non-tumor regions.

Functional modularization and synergy is the future trend

CAR-M activation mediated solely by a single intracellular domain is insufficient to eliminate heterogeneous solid tumors. Consequently, CAR-M functions are evolving toward modularization and synergy. Modular design enables the integration of multiple functional modules within the same cell to achieve synergistic effects, such as the incorporation of cytokines, nanobodies, or small-molecule drugs [83]. At the same time, attention must be paid to the coordination of these functional modules to avoid exhaustion or metabolic disorders caused by over-activation.

Safety by design should be considered upfront

The design of CARs should not focus solely on functionality; safety considerations must be integrated at an early stage. This includes assessing the immunogenicity of the vector, the rationality of target selection, and the selective elimination of overactivated CAR-M. Particularly, controllable in vivo clearance of CAR-M is essential. Compared to CAR-T, macrophages, as pro-inflammatory cytokine secretors, require a more timely and efficient clearance mechanism. This can be achieved by introducing a suicide switch, such as the truncated EGFR (tEGFR) system [82]. Additionally, the inducible caspase 9 (iCasp9) system, which has demonstrated good efficacy in CAR-T therapy [125], is currently the most reliable direct apoptosis-inducing system and also holds potential for application in CAR-M.

Combined therapy strategy

Current clinical trial data indicate that CAR-M monotherapy remains unlikely to displace conventional strategies for solid tumors. Due to intratumoral and intertumoral heterogeneity, CAR‑M cells targeting a single antigen are expected to exhibit limited efficacy [126, 127]. Moreover, the immunosuppressive TME and immune‑escape mechanisms further constrain the effectiveness of standalone therapies [128]. Thus, focusing solely on CAR design optimization or functional enhancement of CAR‑M cells is insufficient. Instead, CAR‑M therapy holds promise when integrated into combination regimens, where it may yield greater clinical benefit (Fig. 6). With continued innovation and accelerated clinical translation, multifaceted combinations of CAR‑M with other treatment modalities could offer more precise and effective therapeutic pathways for solid tumors.

Fig. 6.

CAR‑M‑based combination strategies. Combining CAR-M with other therapeutic strategies may enhance anti-tumor efficacy through synergistic or complementary mechanisms. Current combination approaches primarily include ICIs, phagocytosis checkpoint inhibitors, chemotherapy, oncolytic viruses, and other CAR immune cell therapies

Immune checkpoint inhibitors

In recent years, Immune checkpoint inhibitors (ICIs) have emerged as a cornerstone of anti-tumor immunotherapy. Although both CAR immune cell therapy and ICIs fall under the umbrella of immunotherapy, their mechanisms of action against tumors are fundamentally different. Immune checkpoint inhibitors work by activating the anti-tumor activity of tumor-infiltrating T lymphocytes, primarily through alleviating T-cell exhaustion and disrupting immunosuppressive mechanisms [129]. However, a major challenge lies in the limited infiltration of T-cells and other immune cells into solid tumors, which often hinders immune checkpoint inhibitors from achieving the desired therapeutic outcomes [130]. In clinical oncology, tumors characterized by a paucity of immune cell infiltration, or where immune cells are unable to penetrate the tumor core, are termed “cold” tumors [131]. Such malignancies typically show limited responsiveness to immunotherapies that depend on endogenous immune cells, such as ICIs and therapeutic vaccines [132]. In contrast, CAR-M reprograms “cold” into “hot” tumors by establishing a robust immune-activating network that involves phagocytosis, secretion of inflammatory cytokines, and enhanced antigen presentation. This transformation creates a window of opportunity for immune checkpoint inhibitors to intervene effectively [133]. Therefore, combining these two approaches can theoretically overcome the limitations of monotherapies and yield synergistic effects.

In CAR-M therapy, its combination with immune checkpoint inhibitors has demonstrated remarkable efficacy in pre-clinical models. In animal models of various solid tumors, the combined application of CAR-M and PD-1/PD-L1 antibodies has shown enhanced anti-tumor activity. The underlying mechanism involves the recruitment and infiltration of T-cells and granulocytes by CAR-M, as well as the activation of T-cells by PD-1/PD-L1 antibodies [34, 71, 75, 134]. Furthermore, Gu et al. [135] reported a novel synergistic mechanism of PD-L1 antibodies in CAR-M therapy. Their study found that upon activation of CAR-M, not only was the expression of MHC-related genes significantly up-regulated, but PD-L1 expression also increased. Knocking out PD-L1 in CAR-M via siRNA enhanced the enrichment of the Tpex subset, which represents the primary responders to PD-1 immune checkpoint inhibitors [136]. These results suggest that immune checkpoint inhibitors not only rescue endogenous T-cells suppressed by tumor cells but also counteract T-cell inhibition by CAR-M itself, while potentially modulating CAR-M function, collectively establishing an integrated and synergistic immune circuit.

Phagocytosis checkpoint inhibitors

Beyond adaptive immune checkpoints, recent studies have revealed that innate immune checkpoints expressed on antigen-presenting cells play a critical role in immune escape [99, 137]. Among these, phagocytosis checkpoints refer to molecular signaling pathways in which molecules expressed by human cells bind to corresponding receptors on the surface of macrophages (or other myeloid cells) and actively transmit inhibitory “don’t eat me” signals [24]. Physiologically, these checkpoints aid myeloid cells in distinguishing healthy tissues, thereby protecting normal cells from phagocytosis. However, tumors frequently exploit this mechanism by over-expressing phagocytosis checkpoint molecules to evade phagocytic clearance and suppress innate immunity [138]. Typical phagocytosis checkpoints include molecules such as CD47/SIRPα, CD24/Siglec-10, MHC-I/LILRB1, β2M/LILRB2, and SLAMF/SFRs [139, 140].

Phagocytosis checkpoints are now recognized as a significant constraint on the anti-tumor efficacy of CAR-M therapies. Therapeutic strategies targeting these axes, such as CD47-SIRPα and CD24-Siglec-10, to potentiate phagocytosis have garnered substantial experimental support [141, 142]. Although CAR expression enables macrophages to recognize and phagocytose tumor cells directly, phagocytosis checkpoint signaling remains a potent negative regulator. For example, in a patient-derived glioblastoma humanized mouse model, Chen et al. [32] administered a combination therapy of CAR-M and an anti-CD47 monoclonal antibody. Compared with monotherapy, the combination treatment increased the frequency of positive immune response cells, induced potent tumor-killing immunity, and suppressed postoperative glioblastoma recurrence. Similarly, research by Zhou et al. [143] reached comparable conclusions: monotherapy with an anti-CD47 antibody partially reduced tumor burden in mice, while the combination therapy group with CAR-M showed approximately a 4.2-fold reduction in bioluminescence intensity compared to the anti-CD47 monotherapy group, along with significantly prolonged survival. These data indicate that the combination of CAR-M and anti-CD47 monoclonal antibody achieves remarkable anti-tumor efficacy. However, considering the progress in drug development targeting CD47, the potential impact of systemically administered antibodies such as anti-CD47 on normal cells expressing these molecules, particularly hematopoietic cells, still requires attention and further optimization [144].

In addition to combination treatment, advances in genetic engineering and synthetic biology now allow the integration of components such as shRNA and nanobodies into CAR constructs, enabling the generation of enhanced, all-in-one CAR-M cells. An additional advantage is that these modifications occur exclusively within CAR-M cells, enabling localized action at solid tumor sites with potentially lower systemic toxicity. For example, Zhang et al. [145] co-expressed shRNA targeting SIRPα with CAR molecules, disrupting the reception end of the CD47-SIRPα signaling axis and endowing CAR-M cells with enhanced phagocytic activity, along with significant cytotoxic effects against target antigen-positive tumor cells. Another interesting design involves expressing ITAM-deficient Siglec-G mRNA on the surface of CAR-M cells, enabling competitive binding with CD24 against normal Siglec-G to reduce interference from this phagocytosis checkpoint on CAR-M activity [146]. The results were evident, showing a significant enhancement in the phagocytic function of liver macrophages.

Tumor types with high TAM infiltration are likely to derive greater benefit from this combination therapeutic strategy. Beyond simply enhancing the phagocytic activity of CAR-M cells, the concurrent blockade of phagocytosis checkpoints serves to reprogram TAMs. This reprogramming not only dismantles the physical and functional barriers TAMs impose on CAR-M therapy but also converts this predominantly immunosuppressive population into potent anti-tumor effector cells. Through this dual mechanism, relieving immunosuppression while simultaneously boosting cytotoxic activity, a durable adaptive immune response can be established [147, 148]. Nevertheless, further investigation is required to determine whether phagocytosis checkpoint blockade can consistently achieve TAM reprogramming by CAR-M within the complex TME, and whether the safety risks associated with potentially overactivated macrophages.

Chemotherapy

The combination of chemotherapy and immunotherapy, particularly immune checkpoint inhibitors, has been established as a standard treatment for various solid tumors. The mechanisms of action include alleviating tumor-induced immunosuppression, enhancing the anti-tumor activity of cytotoxic immune cells, and promoting immune cell infiltration into tumor sites [149, 150]. Based on these mechanisms, combining chemotherapy with CAR-M therapy is theoretically well-justified.

Recent years have witnessed promising outcomes in several proof-of-concept studies. In gastric cancer models, the combination of oxaliplatin, a first-line chemotherapeutic agent, with CAR-M therapy has demonstrated efficacy [39]. This combined approach enhanced tumor cell apoptosis and promoted the pro-inflammatory polarization of CAR-M cells, characterized by elevated secretion of IL-6 and IFN-γ, which synergistically strengthened CAR-M-mediated tumor killing. Beyond exerting phagocytic functions, modifying other biological properties of macrophages has also become an important direction in combination therapies. For example, Wang et al. [151] developed CAR-M cells targeting cancer-associated fibroblasts (CAFs), capable of disrupting the fibrotic barrier in solid tumors. In pancreatic cancer, the dense fibrotic stroma formed by CAFs often limits the penetration and distribution of intravenous chemotherapy, thereby restricting its efficacy. Their study showed that CAR-M-mediated clearance of CAFs and reduction of collagen deposition markedly disrupted the fibrotic barrier, leading to enhanced local accumulation of gemcitabine within tumors and improved anti-tumor responses without significant toxicity. These studies indicate that the combined application of CAR-M and chemotherapy is based on deep functional complementarity: chemotherapy not only directly kills tumor cells and alters the TME, but also enhances the immune-activating functions of CAR-M. Meanwhile, CAR-M can be designed as “wall breaker” capable of overcoming both physical and immune barriers of tumors, thereby improving the delivery and efficacy of chemotherapeutic drugs.

Cancer vaccine

Beyond the combination approaches discussed above, the integration of vaccine strategies with CAR-M therapy represents an emerging frontier with significant synergistic potential. Vaccines, particularly those based on mRNA or circRNA platforms, can elicit durable humoral and cellular immunity against tumor-associated antigens. When combined with CAR-M, these vaccines may amplify anti-tumor responses through multiple complementary mechanisms [152, 153]. Wang et al. [154] developed a synergistic approach combining in vivo generation of CAR cells with circRNA-based vaccines. In this system, immunocyte-tropic lipid nanoparticles (LNPs) deliver circRNAs encoding CARs to generate “panCAR” cells in vivo, including CAR-T, CAR-NK, and CAR-M. When combined with circRNA vaccines encoding the corresponding HER2 antigens, this approach demonstrated several synergistic effects relevant to CAR-M therapy. The combination of in vivo CAR generation and vaccination exhibited significantly improved anti-tumor efficacy across multiple mouse models compared to either approach alone.

Oncolytic viruses are a class of viruses capable of selectively infecting and lysing tumor cells, with their core characteristic lying in the dual capacity for direct oncolytic activity and secondary immune activation [155]. Furthermore, through genetic engineering, oncolytic viruses can serve as efficient and targeted gene delivery vectors, enabling sustained expression of therapeutic proteins (such as cytokines, antibodies, and immunomodulatory factors) within the TME [156]. These features make oncolytic viruses ideal partners for combination with CAR immune cell therapies, offering a synergistic strategy to overcome key obstacles in solid tumor treatment, including the immunosuppressive TME, inadequate immune cell infiltration, and antigen heterogeneity [157].

For example, Qi et al. [158] designed and synthesized an oncolytic adenovirus carrying a CD47 antibody (oAd-CD47). This virus not only kills tumor cells through direct oncolysis, releasing tumor-associated antigens to activate adaptive immunity, but also produces CD47 antibodies in situ to block the CD47-SIRPα axis, thereby reprogramming macrophage phagocytic function and enhancing innate anti-tumor immunity. In various solid tumor models, the combination of CAR-M and oAd-CD47 significantly improved the tumor immune microenvironment, reduced the proportion of immunosuppressive cells such as MDSCs and Tregs, and reversed the exhaustion state of CD8⁺ T-cells. More importantly, the combination therapy stimulated a robust neoantigen-specific T-cell response and induced systemic, specific adaptive immune memory in mice. These findings indicate that the combination of oncolytic viruses and CAR-M represents a mechanistically complementary and logically rigorous strategy for enhanced efficacy. The introduction of oncolytic viruses effectively alleviates immunosuppression within the TME, providing CAR-M with additional immunological benefits. While this combination offers a promising paradigm for advancing solid tumor immunotherapy, clinical translation remains challenging and requires systematic evaluation of safety, dosing schedules, long-term efficacy, and potential resistance in more complex models.

CAR-T therapy

In addition to classical chemotherapy, emerging ICIs, and oncolytic viruses, immune cell therapy represents a significant frontier in immunotherapy, particularly the CAR immune cell therapy platform. As an advanced therapy within this platform that has achieved clinical translation at the forefront, CAR-T therapy has been established as a salvage treatment for various hematologic malignancies [159]. Recently, Yescarta and Breyanzi have successively received FDA approval for second-line treatment of diffuse large B-cell lymphoma [160, 161], further demonstrating the tremendous potential of this therapy. Given the antigen-presenting and cytokine-secreting functions of macrophages, which enhance T-cell mobilization, the combined application of CAR-M and CAR-T holds potential mechanistic advantages.

As early as 2022, research using in vitro models proposed the concept of combining CAR-M and CAR-T therapies [162]. The study indicated that inflammatory factors secreted by CAR-T cells could induce macrophages to polarize toward the M1 phenotype, enhancing their cytotoxicity. Additionally, it promoted the adaptability and activation of CAR-T cells by upregulating the expression of co-stimulatory ligands (such as CD86 and CD80) on the surface of CAR-M cells, ultimately leading to significantly enhanced combined cytotoxicity. Clearly, while this positive immune feedback mechanism can generate synergistic effects and amplify therapeutic outcomes, it is also accompanied by notable risks of adverse reactions, such as CRS potentially triggered by excessive activation of CAR-T-cells. To date, no in vivo results of combined CAR-T and CAR-M therapy based on adoptive cell transfer have been reported [163].

The advent of in vivo editing platforms in recent years has created novel opportunities for such combined strategies. It is now feasible to simultaneously generate a variety of engineered immune cells, including CAR-T, CAR-NK, and CAR-M cells, via in vivo delivery systems [164]. For instance, Wang et al. [154] developed a circular RNA carrying a CAR-encoding sequence that, when delivered locally to tumors using LNPs, can edit multiple immune cell types in vivo to form a “pan-CAR cell” population. This approach not only reduces the proportion of immunosuppressive cells and shifts the TME toward a pro-inflammatory, immunologically active state, but also concurrently engages both adaptive and innate immunity, thereby inducing a more comprehensive and durable anti-tumor immune response.

Although in vivo editing platforms offer novel technological pathways for the combination therapy of CAR-engineered immune cells, their clinical translation still faces multiple challenges. On the one hand, a key focus for future research lies in precisely regulating the proportion, activity, and persistence of different CAR-engineered immune cells in vivo to maximize synergistic effects while controlling toxic side effects [165]. On the other hand, the interaction network among immune cells within the context of combination therapy is complex, necessitating in-depth analysis of its dynamic regulatory mechanisms through systems immunology and single-cell technologies [166]. In particular, the risk of adverse reactions arising from the over-activation of multiple CAR immune cells must be effectively controlled. Nevertheless, it is undeniable that in vivo-generated combination therapies involving multiple CAR-engineered immune cells hold promise for transcending the current limitations of tumor immunotherapy. They may provide new breakthroughs for the treatment of solid tumors and drug-resistant malignancies, with potential applications extending to other refractory diseases.

Principles of CAR-M-based combination therapy

The expanding landscape of CAR-M combination therapy strategies raises a critical translational question: how to select the optimal combination regimen for a given patient and tumor type. Clearly, this must be precisely matched with the biological characteristics of the tumor and the functional properties of CAR-M themselves.

The unique immune landscape and stromal architecture of different solid tumors form the basis for selecting specific combination strategies. For immunologically “cold” tumors, such as glioblastoma and pancreatic cancer, the primary obstacle is the lack of T-cell infiltration. In such cases, the immunomodulatory capacity of CAR-M can be leveraged to convert “cold” tumors into “hot” tumors [122]. Priority should be given to combining CAR-M with agents that sustain their function, such as phagocytosis checkpoint inhibitors like anti-CD47, or those that directly recruit T-cells, such as oncolytic viruses. Without such an initial immune recruitment step, relying solely on the baseline activity of CAR-M may result in limited efficacy. For “hot” tumors with pre-existing immune infiltration, such as melanoma, renal cancer, and non-small cell lung cancer, combining CAR-M with immune checkpoint inhibitors is a highly rational approach [134]. CAR-M can enhance antigen presentation and create a pro-inflammatory microenvironment that amplifies T-cell responses. Pre-clinical data support this synergy, the research by Pierini et al. [134] demonstrating that CAR-M can sensitize HER2 + solid tumors to PD-1 blockade. Furthermore, in tumors characterized by a dense stromal barrier, CAR-M can be engineered to target cancer-associated fibroblasts. By disrupting the fibrotic stroma, these CAR-M can enhance the penetration and efficacy of subsequent chemotherapies, such as gemcitabine [151].

Additionally, the timing and route of administration in combination therapies are equally critical. The optimal treatment sequence depends on the underlying mechanism of synergy. For instance, in the combination of CAR-M with oncolytic viruses, pretreatment with oncolytic viruses can remodel the TME and enhance CAR-M infiltration and function. Conversely, for CAR-M combined with immune checkpoint inhibitors, simultaneous administration or sequential dosing timed to coincide with the peak of CAR-M-mediated antigen presentation may maximize T-cell function and promote long-term immunological memory. While systemic administration offers convenience, local delivery may provide advantages for specific combination regimens. For example, in ovarian cancer or peritoneal carcinomatosis, intraperitoneal injection of CAR-M combined with intraperitoneal chemotherapy can achieve high local drug concentrations while minimizing systemic toxicity [167].

Systematic considerations for Safety

Safety risks associated with targets

On-target/off-tumor toxicity, often triggered by CAR molecules targeting antigens with low-level expression on healthy cells, is a well-documented clinical challenge in CAR immune cell therapies [54]. For CAR-M, the situation may be even more complex. In addition to direct phagocytosis, enhanced CAR-M designs typically possess potent M1 polarization capabilities, leading to the release of large amounts of pro-inflammatory cytokines upon activation. This can trigger localized inflammatory storms, potentially causing broader damage to surrounding normal tissues. Furthermore, infused macrophages actively infiltrate organs such as the liver, lung, and spleen. If cells in these organs express the target antigen, CAR-M may accumulate extensively, further increasing the risk of toxicity [83].

Strategies targeting specific antigens have been proposed, such as the design of dual-target “AND” logic-gated CAR-M. This design requires the CAR molecule to simultaneously recognize two antigens to fully activate macrophage functions, thereby significantly enhancing specificity. Even if one antigen is expressed in normal tissues, as long as the other is absent, complete activation does not occur [168]. However, the design of this CAR molecule is complex, and it is necessary to ensure that the two antigens are co-expressed on the tumor and the construction of the CAR molecule is difficult.

With the growing understanding of the importance of antigen affinity in CAR immune cell therapy, designing CARs with low affinity has emerged as a potential option. Such CAR molecules are sensitive only to tumor cells with high antigen density and do not respond to normal cells with low antigen density [169]. However, this approach imposes strict thresholds for antigen expression and may not be suitable for tumors with high antigen heterogeneity.

Additionally, rational design of administration protocols can reduce off-target distribution of CAR-M cells. For specific tumor types, approaches such as intratumoral injection, intraperitoneal injection, hepatic artery infusion, or intracerebroventricular injection can deliver CAR-M directly to the lesion site [39, 170, 171]. This reduces non-specific accumulation of CAR-M in organs such as the liver and lungs, as well as secondary systemic toxicity, but may not be applicable to widely metastatic or inaccessible tumors.

CAR-M intrinsic risks

The unique functional attributes of CAR-M introduce inherent safety risks for therapeutic development. As described above, their phenotypic plasticity, while providing a rationale for engineering pro-inflammatory phenotypes, also presents a dual challenge: the risk of hyperactivation versus the potential for inadequate activation, which could lead to re-polarization towards a M2-like state. Therefore, achieving a calibrated inflammatory response is critical. Incorporating safety switch mechanisms, such as iCasp9 or tEGFR, which enable rapid elimination of CAR-M via a small molecule or antibody in the event of intolerable toxicity, represents a prudent strategy [172].

Furthermore, advances in synthetic biology have led to the application of non-native signaling domain combinations, such as TIR-CD3ζ and CD3ζ-Megf10 [30, 55]. While these constructs may demonstrate short-term efficacy, their long-term stability and functional integrity under the sustained selective pressure of the complex TME warrant further investigation. Additionally, the high signaling demand from such potent, non-physiological domains could lead to signaling molecule exhaustion, potentially compromising other critical CAR-M functions.

Vector-related Risks

The method by which the CAR gene is introduced into cells directly influences genomic stability and the risk of insertional mutagenesis. Current approaches, such as lentiviral vectors, adenoviral vectors, and CRISPR-Cas9 technology, each carry specific safety concerns. Lentiviral vectors integrate the CAR gene randomly into the host genome, which may lead to insertional activation of nearby proto-oncogenes or disruption of tumor suppressor genes, thereby posing a risk of tumorigenesis [173]. Adenoviral vectors are inherently immunogenic and, while they enhance the inflammatory activation of CAR-M macrophages, may also exacerbate CRS [38]. CRISPR-Cas9 can introduce off-target DNA double-strand breaks, resulting in unintended mutations. Additionally, the bacterially derived Cas9 protein may provoke an immune response in the host [174].

The transition from randomly integrating viral vectors to targeted integration using gene-editing technologies represents a new direction in vector optimization. For instance, Zhen et al. [55] utilized the HITI technique to precisely insert the CAR gene into a safe harbor locus (e.g., hH11), thereby avoiding the tumorigenic risks associated with random integration. Optimizing the immunogenicity of vectors will contribute to the stability of CAR-M cells in vivo. Additionally, prior to clinical translation, comprehensive off-target analysis and insertion site characterization, using high-precision sequencing technologies such as rhAmpSeq, argeted locus amplification, and whole-genome sequencing, must be performed on CAR-M clones to ensure the absence of deleterious mutations.

Key challenges and advances in the clinical translation of CAR-M therapy

Beyond the biological challenges directly linked to efficacy and safety, the translation of CAR-M therapies from bench to bedside also faces significant hurdles in manufacturing and scalable production. These issues, concerning product accessibility, cost, and quality control, are critical to enabling broad clinical adoption [175]. Compared to the relatively well-established CAR-T manufacturing system, the unique cellular biological characteristics of CAR-M present distinct obstacles for scalable and standardized manufacturing, while simultaneously driving targeted process innovations [176] (Table 3).

Table 3.

Major challenges and optimization strategies in CAR-M manufacturing

Dimension	Challenge	Traditional method	Emerging strategies
Cell source	Patient-derived autologous cells are often limited in number and exhibit suboptimal quality	Isolate autologous monocytes from the patient and induce their differentiation in vitro	Alternative cell sources, such as iPSCs, allogeneic cells, and immortalized cell lines; in vivo editing
CAR gene editing	Low transfection/expression efficiency	Adenovirus/lentivirus transduction	Enhanced viral platform; non-viral delivery platform; transposon system
Phenotypic control ability	Carries the risk of polarizing toward an immunosuppressive phenotype.	Cytokine stimulation to control phenotypic polarization (GM-CSF/IFN-γ)	Synthetic gene circuit control
Scalable expansion	Non-proliferative in vitro, difficult to scale up.	Multiple rounds of cytokine stimulation, high cost, and limited production capacity	In vivo differentiation of CAR-monocytes; Large-scale cultivation in bioreactors
Accessibility	Technically complex and costly	Individualized treatment	The development of universal CAR-M therapies and in vivo editing technologies

CAR, chimeric antigen receptor; iPSC: induced pluripotent stem cell

Preparation of CAR-M

The first challenge facing CAR-M therapy manufacturing is low exogenous CAR protein expression efficiency. Due to their innate immune defense mechanisms, macrophages recognize exogenous nucleic acids or transfection complexes as foreign, rapidly internalize them, and subsequently degrade them via the lysosomal pathway [177, 178]. This prevents the genetic material from effectively entering the cytoplasm or nucleus for expression. Concurrently, the activation of pattern recognition receptors by these exogenous substances triggers downstream immune signaling pathways, which can induce a cascade of intracellular immune responses, further interfering with the expression and function of the foreign genetic material [179]. Currently, enhanced CAR gene delivery platforms have been developed, including novel adenovirus vectors based on AD5F35 [37, 180], enhanced lentiviral vectors incorporating vpx [181], and non-viral platforms [62]. Promoter engineering and the optimization of regulatory elements also offer novel strategies for achieving more stable and precise CAR protein expression in macrophages [182].

Secondly, limited sources of clinical-grade cells restrict the dose-dependent efficacy of CAR-M therapy. Since CAR-M cells lack proliferative capacity, the adoptive transfer requires a substantial cell quantity. The current approach, mobilizing, collecting, and genetically engineering a patient’s autologous monocytes, can only yield a limited number of CAR-M cells [37]. Alternative sources, such as induced pluripotent stem cells (iPSCs), immortalized cell lines, or allogeneic macrophages, have not yet met clinical-grade application standards [183–186]. Among these, macrophages directionally induced from iPSCs through multi-step differentiation protocols are considered promising for addressing issues such as the significant individual variability and limited quantity of peripheral blood-derived cells. A key point is that the transduction efficiency of CAR engineering in iPSCs is significantly higher than in macrophages, and CAR-M derived from CAR-iPSC differentiation can achieve high-purity CAR molecule expression (exceeding 80%), providing a new solution for the standardized production of CAR-M [30, 187]. However, challenges remain, such as whether iPSC-derived macrophages fully possess functional maturity equivalent to that of macrophages derived from in vivo sources or primary monocytes, the potential accumulation of gene mutations during reprogramming and long-term in vitro culture, the increased risk of malignant transformation, and the tumorigenic risk posed by residual undifferentiated iPSCs. Additionally, the use of the immortalized THP-1 monocyte cell line for CAR-M preparation offers the advantage of large-scale expansion, but its potential tumorigenic risk currently limits its application to pre-clinical validation, and it has not yet advanced to related human studies [31]. Furthermore, both patient-derived monocytes and these alternative sources require directed differentiation using specific cytokine cocktails, and their final phenotype and function are highly dependent on culture conditions [188, 189]. Therefore, establishing a stable, reproducible differentiation and activation protocol to ensure that each batch of CAR-M products possesses potent tumor-killing capability and a consistent immunologically activated phenotype is central to process development [190, 191].

Finally, high costs and lengthy production timelines hinder the accessibility of CAR-M therapy. The entire manufacturing process, from cell collection, differentiation, genetic engineering, expansion, to quality control, is time-consuming, creating a conflict with patients’ urgent therapeutic needs. Additionally, relying on expensive cytokines and transfection reagents imposes a substantial financial burden on patients [192]. Optimizing the workflow, implementing automated closed-system production, and reducing material costs are practical issues that must be resolved to make this therapy available to a broader patient population in the future [193–195].

Logistics factor

In addition to the production of CAR-M, the successful clinical translation of CAR-M therapy critically depends on logistical and supply chain factors that determine product accessibility, reliability, and cost-effectiveness. Unlike traditional small-molecule drugs, cell therapies require the coordination of a complex network involving cell collection, engineering modification, quality control, storage, transportation, and final administration [196]. For CAR-M, the unique biological characteristics of macrophages, including sensitivity to cryopreservation, limited post-thaw viability, and functional plasticity, further complicate these logistical considerations.

The ability to cryopreserve CAR-M products is essential for achieving an “off-the-shelf” supply, repeated dosing, and centralized manufacturing models. However, macrophages present unique challenges for cryopreservation due to their large size, complex membrane structure, and phagocytic activity [197]. Currently, research on iPSC-derived CAR-M progenitors has established important benchmark parameters. Shah et al. [82] demonstrated that CD45⁺ myeloid progenitors (CAR-iMacP) can be cryopreserved in CS10 cryopreservation medium for at least two months, maintaining post-thaw viability of 80–85% compared to fresh control cells. Importantly, cryopreserved progenitors retained their ability to differentiate into mature CAR-M and maintained phagocytic function and antigen presentation capacity. This progenitor cell banking strategy offers a practical solution: cryopreserving intermediate-stage progenitor cells. These cells can be thawed and differentiated on-demand at treatment centers, rather than cryopreserving terminally differentiated macrophages, which may be more susceptible to freeze-thaw injury.

However, several issues require further consideration. These include variability in post-thaw viability across different differentiation batches and cryopreservation conditions, necessitating stringent lot release testing. Additionally, long-term stability data for CAR-M progenitors after cryopreservation are lacking, and the impact of prolonged storage on genomic integrity and CAR expression remains to be determined. Furthermore, ensuring that cryopreserved progenitors from different batches can differentiate into phenotypically homogeneous CAR-M products presents a significant quality control challenge.

Next-generation CAR-M cell technologies

Following the limited efficacy observed in clinical trials of CAR-M adoptive therapy, besides enhancement strategies for CAR-M and combination therapy approaches, novel platforms such as CAR-monocytes and in vivo engineered CAR-M have been proposed (Fig. 7). CAR-monocytes represent a significant evolution of the CAR-M adoptive therapy approach. Rather than using terminally differentiated macrophages, this strategy engineers their precursor monocytes [198]. The rationale centers on the innate ability of monocytes to actively migrate and infiltrate tumor and inflammatory sites. Upon infusion, CAR-monocytes are more efficiently recruited to tumor tissue, where they differentiate locally into functional, tumor-killing CAR-M under microenvironmental cues. Furthermore, this strategy may also streamline the ex vivo manufacturing process and leverage the relatively abundant pool of monocyte precursors in vivo to provide a potential sustained cellular source for therapeutic effects.

Fig. 7.

Overview of the CAR-M therapy platform. Currently, the mainstream platform for CAR-M therapy remains adoptive therapy, with CAR-monocytes emerging as a novel therapeutic modality. Compared with traditional CAR-M therapies, they offer simpler manufacturing processes and lower costs. However, their unpredictable in vivo differentiation requires exogenous guidance and intervention. In vivo editing represents a paradigm-shifting approach, utilizing engineered nanocarriers to deliver CAR genes into the body. This enables in vivo or in situ reprogramming of CAR-M cells to eliminate pathological cells. This innovative strategy directly edits tumor-associated macrophages, streamlining production and treatment workflows, reducing therapeutic costs, improving patient accessibility, and holding exceptional promise for clinical application

Concurrently, a more disruptive in vivo CAR-M engineering platform is emerging. This method entirely bypasses complex ex vivo cell manufacturing by delivering CAR-encoding genetic instructions directly to a patient’s monocytes or macrophages via targeted vectors, generating CAR-M in situ [191]. This platform could convert costly individualized cell therapies into more convenient, repeatably administrable genetic medicines, a promising direction for improving accessibility and reducing costs. Together, CAR-monocyte and in vivo engineering platforms represent a new generation of strategies that, alongside conventional enhancement and combination approaches, form a diversified arsenal against solid tumors (Table 4).

Table 4.

Comparison of different types of CAR-M therapy platforms

Dimension	CAR-M adoptive therapy	CAR-monocytes	In vivo CAR-M
Cell source	Isolation of autologous monocyte cells and in vitro differentiation	Isolation of autologous monocyte cells	In vivo editing of autologous macrophages
Preparation time	1 weeks to complete product preparation and re-infusion	1–2 days to complete product preparation and re-infusion	Preparation in advance and immediate infusion
Relative cost	High	Moderate	Low
Clinical toxicity manifestations	CRS; hematotoxicity; potential long-term side effects	CRS; hematotoxicity; potential long-term side effects	CRS; hematotoxicity; potential long-term side effects; carrier related toxicity
Technology maturity	Moderate, relevant clinical trials have been completed	Low, still in clinical studies	Low, still in pre-clinical studies
Characteristics	Classical treatment mode, high technical maturity	An optimized platform for adoptive therapy, with relatively simple manufacturing and short production cycles	Emerging technology platform with high attention; avoiding difficulties in in vitro manufacturing
Inadequacy	The manufacturing process is complex, time-consuming, and has limited product output capacity; limited persistence in the body	The differentiation efficiency in vivo is unknown; delay in effectiveness and activation; limited persistence in the body	The delivery technology is not mature; the in vivo process is not controllable

CAR, chimeric antigen receptor; CRS, cytokine release syndrome

The emergence of in vivo CAR immune cell editing platforms

Beyond adoptive therapy, the emergence of novel in vivo editing technologies has advanced the development and accessibility of CAR immunotherapy [199]. This includes expanding the patient population that can benefit, alleviating financial burdens on patients, and enabling faster access to CAR-based treatments. Programming in situ CAR immune cells to attack pathological targets by delivering gene-editing vectors locally at tumor sites or within the circulatory system represents a highly efficient approach. It also offers significant advantages in terms of manufacturing cost and time. Moreover, this in vivo editing strategy avoids several major limitations associated with adoptive therapy, such as the heterogeneity risks during ex vivo cell expansion, the impact of logistics and transportation on the viability of cell therapy products, and the mismatch between prolonged ex vivo manufacturing processes and the urgent treatment needs of critically ill patients [200].

Currently, multiple companies have entered the field of in vivo CAR immune cell therapy, focusing on advancing the clinical translation of this technological platform [201]. The primary technical approaches involve using LNPs to deliver CAR genes or employing targeted viruses to achieve precise targeting and editing of immune cells in vivo [202]. Several in vivo CAR-T-cell therapies have now advanced to clinical trials, demonstrating breakthrough progress and establishing novel therapeutic avenues for both cancer and autoimmune diseases [5]. Most recently, Professor Zhu Chen’s team conducted the world’s first human clinical study of an LNP-mRNA-based in vivo CAR-T cell therapy [203]. The results showed that five treated patients with refractory systemic lupus erythematosus generated functionally active CD19 CAR-T cells in situ. These cells efficiently cleared pathogenic B cells, significantly reduced autoantibody levels, improved disease activity, and demonstrated a favorable safety profile. In the field of anti-tumor therapy, a nanobody-targeted and immune-shielded lentiviral vector, ESO-T01, successfully edited CAR-T-cells in four adult patients with multiple myeloma, exhibiting promising anti-tumor efficacy [204]. After two months of follow-up, two patients achieved complete response, while the other two showed partial response. Notably, neither of these two different in vivo CAR-T generation approaches led to severe adverse reactions. Typical symptoms included low-grade CRS and hematological toxicity, with all adverse events being manageable and under stable control.

The application of viral vectors for the in vivo editing of macrophages is constrained by several inherent biological properties of these cells. These include intrinsic resistance to transduction by viral vectors, a terminally differentiated state that limits the efficiency of viral genome integration or transgene expression, degradation of vectors and editing tools by the robust lysosomal system, and nonspecific uptake by tissue-resident macrophages [205]. Furthermore, viral vector-mediated genomic integration carries risks of insertional mutagenesis, oncogenesis, and immunogenicity. Consequently, researchers are increasingly favoring non-viral approaches for in vivo macrophage editing [206]. For instance, LNP-mediated delivery offers a transient editing profile with no risk of genomic integration and substantially lower immunogenicity than viral vectors. This transient nature allows for precise control over the editing window through repeated administration or dose adjustment [207].

Progress and challenges of in vivo CAR-M editing

In CAR-M therapy, the feasibility of in vivo editing has been validated across multiple solid tumor models, primarily through non‑viral methods for in vivo delivery of CAR gene vectors. By editing TAMs, a key immunosuppressive cell type within the TME, this approach reverses their pro‑tumor niche and transforms them into anti‑tumor‑active CAR‑M cells, achieving a kill two birds with one stone effect. Clearly, the safety and efficacy of this in vivo manufacturing approach are constrained by the efficiency of gene delivery and the precision of editing techniques [208].

In in vivo CAR‑M editing, the main delivery target is the typical transporter mannose receptor, which is widely expressed on the surface of TAMs [209]. For instance, Kang et al. [210] targeted the mannose receptor using mannose‑conjugated polyethyleneimine (MPEI) and achieved sustained CAR molecule expression in macrophages via a transposon system. Similarly, Liu et al. [211] synthesized mannosylated chitosan oligosaccharide‑arginine‑based liponanocarriers and successfully edited CAR‑M cells in mice. Jing et al. [33] constructed an ionizable lipid with a citric acid core structure and employed DSPE‑PEG‑mannose complexes to mediate targeting of the mannose receptor, which efficiently delivered cyclic RNA into macrophages for expression.

Beyond active delivery through macrophage surface receptors, passive targeting strategies have also been explored for in vivo CAR-M programming. These approaches leverage the natural tendency of nanomedicine delivery systems to be absorbed by the mononuclear phagocyte system. For instance, Zhou et al. [212] designed an LNP‑mRNA complex that capitalizes on the innate, nonspecific phagocytic activity of macrophages toward nanoparticles. The complex enters cells via the lysosomal pathway, escapes from lysosomes within macrophages, and enables CAR molecule expression, leading to potent anti-tumor effects. It should be noted, however, that the mouse model used in this study lacked mature T and B cells while retaining NK cells and macrophages, which limits insight into the transduction efficiency of this complex in other immune cell types.

Small extracellular vesicles (sEVs) are nanoscale, membrane-bound vesicles that naturally transport bioactive macromolecules to recipient cells, facilitating intercellular communication and functional modulation [213]. They possess favorable characteristics such as biocompatibility, engineerability, low immunogenicity, and stability [214]. Utilizing sEVs for mRNA delivery represents a viable strategy for in situ genome editing. In an innovative approach, Xiao et al. [215] developed engineered sEVs targeted via an anti-CD206 single-chain variable fragment and loaded with mRNA encoding a CAR. These sEVs were designed to generate CAR‑M cells directly within the lung metastatic microenvironment. Interestingly, for lung cancer, the researchers innovatively adopted nebulized inhalation for administration. Unlike intravenous injection, which primarily enriches the vesicles in the liver and spleen, inhalation delivery enables specific and prolonged accumulation of the small extracellular vesicles in lung tissues, allowing penetration into tumor sites. This significantly reduces carrier loss in non-target organs. This inhalation-based in situ CAR-M cell generation strategy using small extracellular vesicles demonstrated remarkable therapeutic efficacy and the potential to induce long-term immune memory in a lung metastasis model.

A method similar to small extracellular vesicles, which utilizes enucleated mesenchymal stem cells to deliver plasmids for in vivo CAR-M engineering, has recently been validated. By removing the nucleus of mesenchymal stem cells via physical or chemical approaches, cytoplasts retaining the cell membrane and partial cytoplasmic components are obtained. This structure is no longer a complete living cell and thus cannot divide, proliferate, or form tumors, but it preserves the membrane proteins and tumor-homing capability of mesenchymal stem cells, essentially functioning as an engineered extracellular vesicle. Plasmids are loaded into the enucleated mesenchymal stem cells through techniques such as electroporation, co-incubation, or membrane fusion, constructing a targeted delivery vehicle carrying gene-editing tools [216]. Zhou et al. [143] employed enucleated mesenchymal stem cells as carriers, leveraging their tumor-homing properties to achieve intratumoral delivery. Subsequently, due to the absence of a nucleus, the enucleated mesenchymal stem cells undergo endogenous apoptosis, triggering macrophage-specific endocytosis to precisely deliver plasmids encoding CAR into glioma-associated microglia or macrophages, thereby enabling in vivo generation of CAR-M cells.

Current advancements in in vivo CAR-M cell engineering remain critically dependent on breakthroughs in gene delivery technology. Existing limitations are primarily constrained by the efficiency and specificity of delivery, the immunogenicity of exogenous materials such as viral vectors and gene carriers, the uncertain anti-tumor efficacy of variably generated CAR-M cells in vivo, and overarching safety concerns [217]. It is anticipated that these challenges will be systematically addressed through the rapid development of novel delivery platforms and more profound investigation into CAR-M biology [202]. Furthermore, combining this innovative approach with complementary therapeutic strategies presents a highly promising direction. For instance, the co-delivery of mRNA encoding immune checkpoint or phagocytosis checkpoint inhibitors via CAR gene delivery vectors could enable the integrated generation of next-generation, multifunctional CAR-M cells.

In summary, in vivo editing represents a critical advancement in CAR‑M therapy development, providing key support for its broader application. Its core advantages include a simplified treatment process, reduced overall costs, and a distinct mechanistic approach: by directly reprogramming TAMs, it alters the composition of the immunosuppressive network within the TME. This strategy avoids a major uncontrolled variable in adoptive CAR‑M therapy, the interference of pro‑inflammatory adoptive macrophages by pro‑tumor TAMs. When confronting the localized immunosuppressive network of solid tumors, repurposing its central cellular component, the TAM, offers a more pertinent therapeutic strategy than attempting to override the entire network [218, 219].

CAR-monocyte therapy

In addition to exploring combination therapies and novel in vivo gene editing strategies, the engineering of CAR-monocytes is emerging as a promising avenue for advancing CAR-M therapeutics [198]. This approach primarily addresses a key limitation in adoptive cell therapy: the limited availability and plasticity of autologous monocytes and macrophages obtained from patient peripheral blood. As macrophage precursors, monocytes are relatively abundant in circulation and can be more readily collected in sufficient quantities via leukapheresis [220]. Early-phase clinical trials of CAR-M further indicate that differentiated macrophages possess a reduced tumor infiltration capacity compared to monocytes. Moreover, ex vivo-generated CAR-M cells often suffice for only a single infusion, complicating the sustained delivery of engineered macrophages. In contrast, CAR-monocytes exhibit enhanced persistence in vivo and superior tumor-homing ability. They also offer the potential for rapid manufacturing and repeat administration. Following infusion, these engineered cells migrate actively to tumor sites, where they differentiate into functional CAR-M cells within the TME, thereby contributing to phagocytosis, antigen presentation, and immunomodulation.

Yang and colleagues engineered a novel CAR-monocyte therapy based on the THP-1 cell line [198]. In mouse models, these CAR-monocyte enhanced T-cell infiltration and the secretion of pro-inflammatory factors, thereby linking innate and adaptive anti-tumor immune responses. The THP-1-based CAR-M approach holds promise for scalable production, combination regimens, and application in solid tumors, positioning it as a promising avenue for next-generation cancer immunotherapy. Nevertheless, its clinical translation faces several challenges, including those related to safety, persistence, and the optimization of combination strategies. In vivo engineering of CAR-monocytes is also progressing. Although data remain scarce, emerging pre-clinical evidence suggests that reprogramming myeloid cells in mice by employing LNPs to deliver CAR-encoding mRNA can suppress tumor growth and potentially reduce metastasis [221]. In clinical research, MT-302 has emerged as the first in vivo CAR-myeloid cell therapy for solid tumors, utilizing LNP-delivered mRNA to program CAR-myeloid cells in vivo. To date, 27 patients with advanced solid tumors across seven dose cohorts have been administered MT-302. Repeated dosing was well tolerated, with minimal reactogenicity. Biopsies of tumor tissue revealed infiltration of CAR-positive myeloid cells, recruitment of CD8⁺ T-cells, and release of pro-inflammatory mediators such as IFN-γ, indicating that in vivo CAR engineering of myeloid cells can remodel the immunosuppressive TME. However, the overall therapeutic efficacy of MT-302 remained limited, with only one patient achieving a partial response [222]. This suggests that, for most patients with advanced solid tumors, immune remodeling driven solely by CAR-myeloid cells is insufficient to induce significant tumor regression. Further development may therefore focus on combination therapies with immune checkpoint inhibitors or chemotherapeutic agents.

In summary, the investigation of CAR-monocytes represents a novel and distinct direction within CAR-M therapy. Nevertheless, the field remains nascent, with limited pre-clinical and clinical evidence to robustly support its efficacy and safety profile. Consequently, its successful development into a viable therapeutic candidate for solid tumors will depend on achieving an optimal balance among potency, safety, and controllability.

Conclusion and outlook

To advance the innovation and clinical translation of CAR-M therapy, it is imperative to integrate novel technological developments and innovative strategies to address existing challenges and improve therapeutic outcomes. The rapid progress of gene editing and synthetic biology has established a solid foundation for CAR-based immune cell therapy platforms. Advancements range from the structural optimization of CAR molecules and the design of synthetic gene circuits to the site-specific integration of CAR genes and precise knockout or knock-in of functional genes. In addition, combination approaches further embed CAR-M within a broader immunotherapeutic network, leveraging synergistic anti-tumor mechanisms to maximize treatment efficacy. Integrating these technological advancements to drive CAR-M iteration and accelerate clinical application represents the primary focus of ongoing development.

Additionally, emerging in vivo editing platforms present new opportunities to expand patient access to CAR immune cell therapies, expedite product delivery, and simplify treatment protocols. Concurrently, the increasing technical complexity, along with safety, ethical, and regulatory challenges, must be rigorously addressed through targeted strategies. Further exploration of in vivo editing platforms holds the potential to disrupt the current treatment paradigm, transitioning CAR-M from a customized cell therapy into an off-the-shelf, reusable genetic drug, a novel technological path advancing in parallel. In the long term, the ultimate goal is to create intelligent synthetic biological systems that transcend the biological limitations of natural macrophages, which depends on deeper fundamental insights into macrophage biology and advancements in multi-gene circuit technology. Although this remains theoretical at present, some studies, as mentioned above, are already providing valuable technical references.

In summary, CAR-M therapy is currently at a critical juncture of accelerating translation from proof-of-concept to clinical solutions. Continuous technological innovation, combined with the exploration and application of combination therapies, provides robust support for enhancing efficacy, advancing CAR-M technology toward clinical use, and expanding the beneficiary population. The existing research foundation and clinical cases offer confidence in realizing CAR-M therapy for solid tumors, with the potential to bring new hope to patients.

Abbreviations

ACOD1

Aconitate decarboxylase 1

ADCP

Antibody-dependent cellular phagocytosis

CAF

Cancer-associated fibroblast

CAR

Chimeric antigen receptor

CAR-M

Chimeric antigen receptor macrophage

CAR-N

Chimeric antigen receptor neutrophil

CAR-NK

Chimeric antigen receptor natural killer cell

CAR-T

Chimeric antigen receptor T-cell

CD3ζ

CD3 zeta chain

cGAS

cyclic GMP-AMP synthase

c-MET

Cellular mesenchymal epithelial transition factor

CNS

Central nervous system

CRS

Cytokine release syndrome

CSR

Chimeric signaling receptor

DC cell

Dendritic cell

FcεRIγ

High-affinity IgE receptor gamma chain

ICANS

Immune effector cell-associated neurotoxicity syndrome

iCasp9

inducible caspase 9

ICI

Immune checkpoint inhibitor

IFN-β

Interferon beta

IFN-γ

Interferon gamma

iPSC

induced pluripotent stem cell

ITAM

Immunoreceptor tyrosine-based activation motifs

LNP

Lipid nanoparticle

MMPs

Matrix metalloproteinases

NF-κB

Nuclear factor kappa-B

NK cell

Natural killer cell

PCs

Proprotein convertases

PD-1

Programmed cell death 1

PD-L1

Programmed cell death ligand 1

scFv

Single-chain variable fragment

SFRs

Signaling lymphocytic activation molecule family receptors

sEVs

Small extracellular vesicles

STING

Stimulator of interferon genes

SYK

Spleen tyrosine kinase

TAM

Tumor-associated macrophage

tEGFR

truncated EGFR

TGF-β

Transforming growth factor beta

TIR

Toll-like receptor 4 intracellular toll/IL-1R

TLR4

Toll like receptor 4

TME

Tumor microenvironment

TNF-α

Tumor necrosis factor alpha

TNF-β

Tumor necrosis factor beta

Treg

Regulatory T-cells

Author contributions

Yizhao Chen: Writing - original draft, Visualization.Lucheng Zhou: Writing - original draft.Xinlei Chen: Writing - original draft.Shuai Wang: Supervision, Writing - review & editing.Weiwei Chen: Supervision, Writing - review & editing.Zixuan Li: Supervision, Writing - review & editing.Ji Qiu: Supervision, Writing - review & editing.Ruilin Li: Supervision, Writing - review & editing.Jiajie Tu: Conceptualization, Supervision, Writing - review & editing.Ning Lin: Conceptualization, Supervision, Writing - review & editing.

Funding

This study was supported by the National Natural Science Foundation of China (82504836), and Scientific and Technological Project of Bengbu Medical University under the “Healthcare Alliance” in 2024 (2024byzd361).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.