Challenges and limitations of chimeric antigen receptor T-cell therapies in solid tumors: why are approvals restricted to hematologic malignancies?

Abstract

Chimeric antigen receptor T-cell (CAR-T) therapy has revolutionized the treatment of hematologic malignancies, offering a highly personalized and potent immunotherapeutic approach. To date, the U.S. Food and Drug Administration has approved seven CAR-T therapies targeting CD19 and B-cell maturation antigen; each demonstrated remarkable clinical efficacy across various hematologic malignancies. Despite significant advancements in preclinical studies and clinical trials, no CAR-T therapy has been approved for solid tumors, which account for the majority of cancer cases worldwide. These key challenges include the lack of distinct and accessible target antigens, the immunosuppressive tumor microenvironment (TME) that impairs immune cell efficacy, the heterogeneity of solid tumors that complicates treatment uniformity, and the potential risks of off-tumor toxicity. These obstacles represent a complex array of biological and clinical obstacles, distinct from the more favorable immune environment of hematologic cancers that has been pivotal to the success of CAR-T therapy. Preclinical studies in multiple myeloma emphasize memory T-cell optimization and combinatorial strategies to enhance CAR-T efficacy in solid tumors. Our review emphasizes innovative strategies to address these key challenges in CAR-T therapy for solid tumors, including advanced multi-antigen targeting approaches, reprogramming of the TME, and the development of next-generation safety measures to mitigate toxicity risks. By addressing both scientific and clinical obstacles, this review envisions a future in which CAR-T therapy’s full potential extends beyond hematologic malignancies, transforming the landscape of oncology and improving outcomes for patients with solid tumors.

Introduction

The advent of chimeric antigen receptor (CAR)-T cell therapy represented a transformative advancement in cancer treatment, offering renewed hope to patients with previously untreatable hematologic malignancies [1, 2]. Unlike conventional modalities that non-selectively target malignant and healthy cells, such as chemotherapy and radiation, CAR-T therapy enables a highly specific and personalized approach to tumor eradication [3]. Since the U.S. Food and Drug Administration (FDA) approved tisagenlecleucel (Kymriah®) in 2017 for the treatment of B-cell acute lymphoblastic leukemia (B-ALL) [4], a total of seven CAR-T products—each targeting hematologic malignancies—have received FDA approval. These therapies have demonstrated remarkable clinical outcomes, including remission rates exceeding 80% in some cases. These therapies have provided life-saving treatment options for patients with relapsed or refractory (RR) leukemias, lymphomas, and multiple myeloma (MM) [4]. The clinical success of CAR-T therapy in hematologic malignancies is attributed to the presence of well-characterized tumor-associated antigens (TAAs), such as CD19, CD20, CD22, and B-cell maturation antigen (BCMA), which are abundantly expressed on malignant cells and are readily accessible within hematologic, bone marrow, and lymphoid compartments. These anatomical niches present minimal physical or immunological barriers, facilitating efficient CAR-T cell trafficking, antigen engagement, and cytolytic activity. Moreover, the depletion of normal cells expressing the targeted antigen (B cells in CD19-directed therapies) can be clinically managed, as demonstrated by the use of immunoglobulin replacement therapy to address B-cell aplasia [5]. In contrast, CAR-T therapy for solid tumors remains a significant challenge. Despite extensive preclinical studies and numerous clinical trials targeting antigens, such as human epidermal growth factor receptor 2 (HER2), epidermal growth factor receptor (EGFR), and mesothelin, no CAR-T product for solid tumors has received regulatory approval to date, emphasizing the complex interplay of tumor biology and immune evasion mechanisms [3, 6]. However, recent clinical advances offer ground for optimism. For instance, ganglioside GD2 (GD2)/B7-H3 dual-targeted CAR-T therapy in diffuse midline gliomas (DMG) extended median survival to 19.8 months [7]; Claudin (CLDN)18.2-targeted satricabtagene autoleucel significantly improved progression-free survival (3.25 vs. 1.77 months) and overall survival (OS) in a recent phase 2 trial [8, 9]; and dual-target strategies have induced tumor regression in most glioblastoma patients [10, 11]. Building on these evolving insights, our group’s preclinical studies in MM (Sect. "Memory CAR-T cells: The next frontier") explore strategies for memory T-cell optimization and tumor microenvironment (TME) reprogramming, with potential translational relevance for solid tumors. However, the pronounced heterogeneity of solid tumors—reflected in antigen escape rates exceeding 30% compared to 5–10% in MM—necessitates adaptative and multifaceted therapeutic strategies [12].

This review delineates the key factors underlying the limited efficacy of CAR-T cell therapy in solid tumors, such as the immunosuppressive TME and the risk of off-tumor toxicities (Sect. "Challenges in applying CAR-T to solid tumors"). In addition, it emphasizes emerging therapeutic innovations designed to address these challenges (Fig. 1). Collectively, this review offers a comprehensive and forward-looking overview of the current landscape and future directions of CAR-T cell strategies for solid malignancies.

Fig. 1.

Roadmap to approval. Timeline depicting key milestones in the development of cell therapies for solid tumors. The sequence begins with the first tumor-infiltrating lymphocyte (TIL) therapy administered for metastatic cancer in 1986. A pivotal advancement occurred in 1993 with the introduction of first-generation chimeric antigen receptors (CARs), signifying the shift from non-genetically modified cell therapies to engineered cellular approaches. Although these early CAR constructs represented a significant technological breakthrough, they were limited by poor persistence and suboptimal clinical efficacy. Over the past decade, CAR-T cell therapies targeting solid tumors have progressed markedly, driven by continuous innovation in CAR design and strategies to address major barriers such as tumor heterogeneity, the immunosuppressive tumor microenvironment (TME), and off-target toxicities

Mechanistic insights from FDA-approved CAR-T cell therapies and implications for solid tumors

CAR-T cell therapy has revolutionized the treatment landscape of hematologic malignancies, offering substantial clinical benefits and advancing the paradigm of personalized medicine. We present a comprehensive overview of CAR-T therapies approved by the FDA, with an emphasis on the critical determinants of their efficacy and durability (Table 1). By critically analyzing key mechanistic and clinical determinants, particularly antigen accessibility, T-cell persistence, and TME interactions, we elucidated the factors limiting the translatability of CAR-T cell successes in hematologic malignancies to solid tumors. This analysis revealed actionable strategies to overcome these barriers, offering a novel framework for advancing CAR-T therapy in solid malignancies.

Table 1.

Comparative analysis of FDA-approved CAR-T therapies for hematologic malignancies and translational implications for solid tumors (as of May 2025)

Therapy	Target	Indication	Approval year	Efficacy	Antigen escape rate	TME compatibility	Solid tumor applicability	Dosage regimen	Major side effects	Implications for solid tumors
Tisagenlecleucel (Kymriah®)	CD19	B-ALL and DLBCL	2017	Patients: N = 68 ORR: 50% (95% CI 38–62%) CR: 32% PR: 18%	10–20% (CD19 loss)	High (systemic access)	Limited (heterogeneous antigens)	For patients ≤ 50 kg: 0.2–5 × 106 CAR-positive viable T cells per kg For patients > 50 kg: 0.1–2.5 × 108 CAR-positive viable T cells (non-weight based)	CRS, NE, and cytopenias [13, 14]	4-1BB enhances persistence, critical for TME-challenged solid tumors. Requires TME modulation
Axicabtagene Ciloleucel (Yescarta®)	CD19	R/R LBCL	2017	Patients: N = 101 ORR: 72% CR: 51% (95% CI: 41, 62)	10–15%	High (lymphoid homing)	Low (T-cell exhaustion risk)	Target dose: 2 × 106 CAR-positive viable T cells per kg body weight (maximum dose: 2 × 108)	CRS, NE, and cytopenias [15, 16]	CD28-driven cytotoxicity suggests rapid effector function, but TME barriers limit applicability
Brexucabtagene Autoleucel (Tecartus®)	CD19	MCL	2020	Patients: N = 60 ORR: 87% (95% CI: 75, 94), CR: 62% (95% CI: 48, 74), ORR: 80% (95% CI: 69, 88)	10–15%	High (minimal barriers)	Low (stromal barriers)	Target dose: 2 × 106 CAR-positive viable T cells per kg body weight (maximum dose: 2 × 108)	CRS, NE, and Cytopenias [17, 18]	Lymphodepletion enhances expansion, a strategy to explore with TME-modulating agents
Lisocabtagene Maraleucel (Breyanzi®)	CD19	R/R LBCL	2021	Patients: N = 192 ORR: 73% (95% CI: 67, 80), CR: 54% (95% CI: 47, 61)	10–15%	High (balanced T-cells)	Moderate (safety profile)	50–110 × 106 CAR-positive viable T cells (consisting of 1:1 CD8 and CD4 components)	CRS, NE, and Cytopenias [19, 20]	Defined CD8:CD4 ratio improves consistency, a model for solid tumor trial designs
Idecabtagene Vicleucel (Abecma®)	BCMA	RRMM	2021	Patients: N = 127 ORR: 72% (95% CI: 62%, 81%), CR: 28% (95% CI 19%, 38%)	5–10% (BCMA loss)	Moderate (bone marrow)	Moderate (multi-epitope potential)	300–460 × 106 CAR-positive T cells	CRS, NE, and Cytopenias [21, 22]	BCMA specificity emphasizes need for solid tumor antigens with minimal normal tissue expression
Ciltacabtagene Autoleucel (Carvykti®)	BCMA	RRMM	2022	Patients: N = 97 ORR: 97.9% (95% CI: 92.7%, 99.7), DOR: 21.8 months (95% CI: 21.8, NE) with a median duration of follow-up of 18 months	5–10%	Moderate (bone marrow)	High (dual-epitope design)	0.5–1.0 × 106 CAR-positive viable T cells per kg body weight (maximum dose of 1 × 108 CAR-positive viable T cells per single infusion)	CRS, ICANS, and Cytopenias [25, 26]	Dual-epitope BCMA targeting suggests multi-antigen strategies for heterogeneous solid tumors
Obecabtagene Autoleucel (Aucatzyl®)	CD19	B-ALL	2024	Patients: N = 65 CR: 42% (95% CI: 29%, 54%) The median duration of CR achieved within 3 months was 14.1 months (95% CI: 6.1, not reached)	10–20%	High (split-dose safety)	Moderate (safety mechanisms)	410 × 106 CAR-positive viable T to be administered as a split-dose infusion on Day 1 and 10 (± 2 days)	CRS, ICANS, and Cytopenias [27, 28]	Split-dose regimen may reduce toxicity, a strategy to mitigate off-tumor effects in solid tumors

Critical determinants of CAR-T success in hematologic malignancies

The clinical efficacy of FDA-approved CAR-T therapies is largely attributable to three interrelated factors: uniform antigen expression, a permissive TME, and durable T-cell persistence. Target antigens, such as CD19 and BCMA, are abundantly expressed on malignant cells, with minimal expression on essential healthy tissues—rendering off-target effects manageable (B-cell aplasia managed via immunoglobulin replacement). Moreover, incorporation of costimulatory domains—most commonly 4-1BB or CD28— augments T-cell expansion and persistence, sustaining antitumor activity.

Tisagenlecleucel (Kymriah®): Approved in 2017 for patients aged ≤ 25 years with RR B-ALL and diffuse large B-cell lymphoma (DLBCL); tisagenlecleucel used a 4-1BB costimulatory domain to enhance T-cell persistence. Clinical trials reported an overall response rate (ORR) of 82.5% (95% confidence interval [CI] 70.9–91.0) in B-ALL, with 63% achieving complete remission (CR) and an additional 19% attaining CR with incomplete hematologic recovery [13, 14].

Axicabtagene Ciloleucel (Yescarta®): Approved in 2017 for adult patients with RR large B-cell lymphoma (LBCL) after ≥ 2 lines of therapy, including DLBCL, primary mediastinal LBCL, high-grade BCL, and DLBCL arising from follicular lymphoma; this therapy incorporates a CD28 costimulatory domain to promote rapid T-cell activation. In pivotal clinical trials, axicabtagene ciloleucel demonstrated an ORR of 72% and a CR rate of 51% (95% CI 41–62) [15, 16].

Brexucabtagene Autoleucel (Tecartus®): Approval in 2020 for the treatment of RR mantle cell lymphoma; this therapy is structurally analogous to Yescarta® and targets CD19 with a CD28 costimulatory domain. It achieved an ORR of 87% and a CR rate of 62% following a single infusion [17, 18].

Lisocabtagene Maraleucel (Breyanzi®): Approved in 2021 for adult patients with RR LBCL after ≥ 2 lines of systemic therapy—including DLBCL, high-grade BCL, primary mediastinal LBCL, and follicular lymphoma grade 3B—this product features a defined CD8⁺/CD4⁺ T-cell ratio and a 4-1BB costimulatory domain to enhance T-cell expansion and persistence. Clinical trials reported a 73% ORR (95% CI 67–80) and a 54% CR (95% CI 47–61), with a median duration of response (DOR) of 16.7 months (95% CI 5.3–not reached). Among patients who achieved CR, 65% remained in remission ≥ 6 months and 62% for ≥ 9 months [19, 20].

Idecabtagene Vicleucel (Abecma®): Approved in 2021 for RRMM after ≥ 4 prior lines of therapy, including an immunomodulatory agent, a proteasome inhibitor, and an anti-CD38 antibody, Abecma® is the first FDA-approved cell-based gene therapy for MM. This therapy targets BCMA, which is predominantly expressed on malignant plasma cells, enabling selective cytotoxicity while sparing most healthy tissue. In clinical trials, Abecma® demonstrated an ORR of 72% (95% CI 62–81) and a CR rate of 28% (95% CI 19–38), with 65% of CR patients maintaining a durable response for ≥ 12 months [21, 22].

Ciltacabtagene Autoleucel (Carvykti®): Approved in 2022 for adult RRMM patients after ≥ 4 prior lines of therapy, this BCMA-directed CAR-T cell therapy features dual-epitope targeting, a CD8α transmembrane domain, and 4-1BB/CD3ζ intracellular signaling domain to enhance activation and persistence. Clinical trials reported an impressive ORR of 98%, with 78% achieving stringent CR, and a median DOR of 22 months [23–26].

Obecabtagene autoleucel (Aucatzyl®): Approved in 2024 for adult patients with B-ALL, this CD19-directed CAR-T therapy achieved a CR rate of 42% (95% CI 29–54) within 3 months of infusion, with a median DOR of 14.1 months (95% CI 6.1–not reached) [27, 28].

Comparative analysis and lessons for solid tumors

Table 1 presents a comparative analysis of FDA-approved CAR-T therapies, integrating key parameters, such as antigen escape rates, TME compatibility, and translational applicability to solid tumors. First, with respect to antigen selection, the uniform expression of CD19 and BCMA in hematologic malignancies underlies the high efficacy of current CAR-T products, whereas solid tumor antigens, such as EGFR and mesothelin, exhibit significant inter- and intratumoral heterogeneity, resulting in elevated antigen escape rates—from approximately 10–20% in B-ALL to substantially higher levels in aggressive solid tumors like glioblastoma. Regarding costimulatory domains, the 4-1BB domain promotes the development of memory T cells and supports durable responses; however, its effectiveness in solid tumors is limited by the profoundly immunosuppressive TME.

Bridging hematologic success to solid tumor challenges

The selective and consistent expression of CD19 and BCMA minimizes off-target toxicity, a risk further mitigated through clinical interventions, such as immunoglobulin replacement and prophylactic antibiotics [29, 30]. In addition, the systemic dissemination of hematologic malignancies facilitates CAR-T cell trafficking, enabling efficient localization to disease sites. In contrast, solid tumors are defined by antigen scarcity, dense stromal architecture, and immunosuppressive TME [31, 32]. Although FDA-approved CAR-T therapies provide a foundational blueprint for extending this modality to solid tumors, they expose critical barriers—most notably antigen escape and TME suppression (Sect. "Challenges in applying CAR-T to solid tumors"). Bridging these divides will require multifaceted approaches, including memory T-cell engineering, TME remodeling, and integration of safety mechanisms, such as inducible caspase 9 (iCasp9) switches (Sects. "Challenges in applying CAR-T to solid tumors"–"Emerging strategies and future directions"). The remarkable efficacy of CAR-T therapies in hematologic malignancies offers valuable mechanistic insights for overcoming the unique challenges posed by solid tumors, particularly in modulating immunosuppressive environments and enhancing T-cell persistence and function. A detailed translational case study from our group is presented in Sect. "Memory CAR-T cells: The next frontier".

Challenges in applying CAR-T to solid tumors

Limited target antigens

In contrast to hematologic malignancies, in which well-characterized antigens serve as effective targets, solid tumors rarely express truly tumor-specific antigens. Most candidate targets in solid tumors are also expressed to varying degrees in healthy tissues, increasing the risk of on-target, off-tumor toxicity [33]. In addition, even when TAAs are identified, their expression is frequently heterogeneous (Fig. 2A and 2). For instance, only 20–30% of breast cancer patients overexpress HER2, and even within a single tumor, HER2 expression can vary significantly across regions and individual cells [34, 35]. A similar challenge arises with EFGR variant III (EGFRvIII), a target under investigation for glioblastoma therapy. EGFRvIII is expressed in approximately 30% of glioblastoma cases, but is prone to antigenic loss under therapeutic pressure, as tumor cells may downregulate the antigen to evade immune surveillance [36, 37]. This antigenic variability limits the effectiveness of CAR-T cell therapies, as many tumor cells may escape recognition, leading to diminished therapeutic efficacy and an increased risk of relapse. Neoantigens, derived from tumor-specific mutations, present another set of challenges. Although they offer a high degree of specificity, being unique to individual patients and even to distinct tumor subclones, their individualized nature renders them impractical for standardized CAR-T therapy approaches. Personalized targeting of neoantigens requires bespoke therapeutic designs, which are difficult to scale for widespread clinical use [38, 39]. Antigen escape remains a critical obstacle. In B-ALL, CD19-negative relapses occur in 10–20% of patients following CAR-T cell therapy [40]. The issue is further exacerbated in solid tumors, where baseline antigenic heterogeneity is substantially higher, increasing the likelihood of immune evasion [41]. For instance, clinical trials targeting EGFRvIII in glioblastoma demonstrated initial tumor regression; however, recurrence often occurred due to the outgrowth of EGFRvIII-negative clones, emphasizing the difficulty of sustaining durable responses in a dynamic and evolving antigenic environment [42–45]. To date, several promising solid tumor antigens have been identified through preclinical and clinical investigations, including GD2 in neuroblastoma, B7-H3 in pediatric malignancies, and prostate stem cell antigen (PSCA) in prostate cancer. However, each of these targets is associated with distinct limitations that complicate their translation into broadly effective CAR-T therapies [46]. GD2-directed CAR-T cells have demonstrated antitumor activity in neuroblastoma, with early-phase clinical trials reporting tumor regression. However, the low-level expression of GD2 on peripheral nerves presents a significant risk of neurotoxicity, which may manifest as pain or peripheral nerve dysfunction. B7-H3, a broadly expressed antigen across multiple solid tumors, including osteosarcoma and medulloblastoma, offers the advantage of wider applicability. However, heterogeneous expression of B7-H3 within tumors limits the consistency and reliability of therapeutic responses, making effective tumor cell eradication more challenging [46, 47]. Similarly, although PSCA has shown potential as a target in prostate cancer, its expression in normal tissues, such as the bladder and stomach, raises substantial safety concerns. Collectively, these examples illustrate the persistent difficulty in balancing antigen specificity with therapeutic feasibility in the context of solid tumors [48, 49]. A concise, antigen-centric summary of normal-tissue expression, antigen density/heterogeneity, and escape risk for commonly pursued solid-tumor targets is provided in Table 2.

Fig. 2.

Antigen expression profiles. Bar graphs showing transcriptomic expression patterns of commonly targeted tumor-associated antigens (TAAs) across malignant and normal tissues. (A) TAAs, such as B-cell maturation antigen (BCMA), CD19, CD20, CD22, CD38, and CD138, are highly expressed in malignant hematopoietic cells but are present on normal B cells (CD19, CD20, and CD22), plasma cells (BCMA, CD38, and CD138), or other lineage-restricted hematopoietic populations. Notably, some TAAs exhibit low-level or inconsistent expression in non-hematopoietic, non-malignant cells, including human fibroblasts, epithelial cells, melanocytes, embryonic kidney cells, and other tissue-derived normal cell lines, complicating target specificity. However, the hematopoietic lineage-restricted expression allows for manageable off-tumor effects in CAR-T cell therapy; for instance, B-cell aplasia can be clinically managed with adjunctive intravenous immunoglobulin (IVIG). Transcriptomic data were sourced from the Human Protein Atlas (https://www.proteinatlas.org) and are presented as normalized transcripts per million (nTPM). (B) In contrast, TAAs commonly associated with solid tumors—such as human epidermal growth factor receptor 2 (HER2), epidermal growth factor receptor (EGFR), carcinoembryonic antigen (CEA), mucin 1 (MUC1), mesothelin (MSLN), and NY-ESO-1—exhibit broader and more heterogeneous expression across normal solid tissues, often at levels comparable to those in malignant tissues. This widespread expression increases the risk of off-tumor toxicity, posing significant challenges for CAR-T therapy in solid tumors. Transcriptomic data were obtained from the Human Protein Atlas and are presented as nTPM

Table 2.

Innovative strategies to overcome limited target antigen availability in solid tumors

Approaches	Antigen	Major solid-tumor indications	Normal-tissue expression (risk)	Density/heterogeneity	Antigen loss/escape	Clinical note (1-line)	Key references
Emerging approaches	B7-H3 (CD276)	Pediatric solid tumors, CNS (DIPG/DMG)	Variable in healthy tissues	Broad but heterogeneous	Under characterization	Locoregional CNS delivery feasible; strong pediatric signal	[7, 46]
	CLDN18.2	Gastric/GEJ	Low outside gastric lineage	Often enriched but variable	Being defined	First randomized positive CAR-T signal in GC/GEJ	[9, 50–53]
	IL13Rα2	GBM	Low in normal brain	Often high but patchy	Escape described	CNS localization favors intraventricular delivery	[54–56]
Established approaches	HER2	Breast, gastric, ovarian	Low in heart/skin → safety concern	~ 20–30% overexpression; intratumor variability	Documented under pressure	Strong precedent; selection/monitoring essential	[34, 35]
	EGFR/EGFRvIII	Lung, GBM	Moderate in epithelia	EGFRvIII ~ 30% in GBM; high variability	Frequent under therapy	Regional heterogeneity and down-modulation limit durability	[36, 37, 42–45]
	Mesothelin (MSLN)	Pancreatic, mesothelioma, GI	Low in pleura/mesothelium	Variable surface density	Down-regulation observed	Specificity acceptable; density plus TME constraints	[57–59]
	GD2	Neuroblastoma, select sarcomas	Minimal on peripheral nerves	Inter-patient variability	Reported; safety limits dose	Neurotoxicity risk requires careful dosing	[46, 47]
	PSCA	Prostate (± others)	Present in bladder/stomach	Patchy; patient-specific	Potential under pressure	Normal-tissue expression elevates off-tumor risk	[48, 49]
	CEA (CEACAM5)	Colorectal, GI	Moderate in normal colon	Marked inter/intra-tumor heterogeneity	Possible	Basal gut expression narrows systemic window	[60, 61]
	MUC1	Lung, breast, others	Moderate in airway epithelium	Glycoform-dependent; heterogeneous	Possible	Broad presence drives off-tumor constraints	[62]

Hostile TME

The TME in solid tumors is characterized by a dense stromal architecture, physical barriers, immunosuppressive cellular components, and metabolic stressors—all of which hinder the infiltration, persistence, and cytotoxic activity of CAR-T cells (Fig. 3) [63, 64]. Previous studies have demonstrated that < 1% of infused CAR-T cells successfully localize to solid tumor sites, in contrast to > 50% accumulation in hematologic malignancies. This stark disparity emphasizes the formidable challenge of delivering CAR-T cells to tumors that are effectively insulated by a hostile microenvironment [65, 66]. The TME is enriched with immunosuppressive elements, including regulatory T cells, myeloid-derived suppressor cells (MDSCs), regulatory dendritic cells, cancer-associated fibroblasts, and tumor-associated macrophages. In addition, upregulation of immune checkpoint molecules within the TME further inhibits CAR-T cell function [67–69]. For instance, in melanoma, high programmed death-ligand 1 expression has been associated with CAR-T cell failure, as T cells are rendered dysfunctional in the face of overwhelming immunosuppressive signaling [70]. Clinical trials involving mesothelin-targeted CAR-T cells in pancreatic cancer have reported minimal T-cell infiltration into tumor tissue. Tumor biopsies revealed high levels of TGF-β and phenotypic markers of T-cell exhaustion, emphasizing the profound suppressive influence of the TME [57–59]. Moreover, solid tumors are characterized by intense nutrient competition and proliferation within hypoxic niches. These regions are marked by low oxygen tension and lactate accumulation, both of which disrupt CAR-T cell metabolism and impair effector function [71–74]. In preclinical glioblastoma models, CAR-T cells targeting IL13Rα2 exhibited significantly reduced interferon-gamma (IFN-γ) production and diminished cytotoxicity under hypoxic conditions, illustrating how metabolic deprivation critically impairs immune activity [54, 56, 75, 76]. Clinical trials targeting HER2 in sarcoma have demonstrated initial tumor responses; however, these effects were transient, with therapeutic efficacy rapidly curtailed by stromal barriers and immunosuppression mediated by MDSCs [77, 78]. Similarly, CAR-T cell therapies directed against carcinoembryonic antigen in colorectal cancer exhibited limited clinical benefit. Post-treatment evaluations revealed inadequate T-cell persistence within a cytokine-rich, immunosuppressive TME, emphasizing the barriers to sustained therapeutic efficacy [60, 61]. These findings emphasize the urgent need for innovative strategies capable of circumventing the immunosuppressive and physical obstacles inherent to the TME.

Fig. 3.

Tissue microenvironment (TME) dynamics. Schematic representation of the complex immunosuppressive landscape within the TME that hinders chimeric antigen receptor T-cell (CAR-T) efficacy in solid tumors. The extracellular matrix forms a dense physical barrier that impedes the infiltration of immune effector cells, including CAR-T cells, natural killer cells, and dendritic cells. Within the tumor core, cancer cells and stromal elements secrete a range of immunosuppressive cytokines (transforming growth factor [TGF]-β, interleukin [IL]−1, IL-6, IL-10, and Th2 cytokines) and oncogenic signals that support tumor survival, progression, and metastasis. These factors facilitate the recruitment and activation of immunosuppressive cellular populations, including cancer-associated fibroblasts (CAFs), tumor-associated macrophages, myeloid-derived suppressor cells (MDSCs), regulatory T cells (Tregs), and regulatory dendritic cells. The dynamic interplay among these cellular and molecular constituents suppresses anti-tumor immune responses and impairs CAR-T cell infiltration, persistence, and cytotoxic activity, promoting immune evasion and tumor advancement

Tumor heterogeneit

Solid tumors exhibit pronounced heterogeneity, including intratumoral (variation within a single tumor) and intertumoral (differences between tumors across patients) heterogeneity, which collectively undermine the efficacy of CAR-T cell therapy in achieving comprehensive malignant cell targeting (Fig. 4A–E) [79]. Intratumoral heterogeneity refers to the molecular and genetic diversity present across distinct regions of the same tumor [80–83]. Tumors frequently consist of multiple subclonal populations, each characterized by unique genetic alterations, gene expression profiles, and antigenic landscapes [80–83]. These subclonal differences facilitate immune evasion and therapeutic resistance, as certain tumor regions may lack the target antigen, or express alternative markers not recognized by the CAR-T construct [79, 84]. For instance, although the primary tumor may express specific antigens, metastatic lesions may downregulate these targets or express distinct antigens, rendering them invisible to CAR-T cells designed for the original tumor profile [85]. Furthermore, clonal evolution driven by selective pressures can intensify this heterogeneity, fostering the emergence of resistant tumor subclones that escape CAR-T-mediated cytotoxicity [86].

Fig. 4.

Tumor Heterogeneity. (A) Intertumoral heterogeneity refers to molecular variability observed among tumors from different patients diagnosed with the same cancer type. (B) Intratumoral heterogeneity describes the coexistence of genetically and phenotypically distinct subpopulations within a single neoplasm. (C) Longitudinal heterogeneity denotes temporal changes in tumor composition, typically emerging during metastasis or recurrence under therapeutic selective pressure. (D) Heatmaps illustrate the transcriptional variability of representative tumor-associated antigens (TAAs) across normal lung tissue (n = 4), primary lung adenocarcinoma (LUAD) samples (n = 8), and matched metastatic sites including brain (n = 3), bone (n = 1), and adrenal gland (n = 1). The analyzed genes include BRAF, BTBD2, CEACAM5, CEACAM6, EGFR, ERBB2, EZH2, GPNMB, HHAT, and MAGED2. Data were derived from single-cell RNA-sequencing of 41,384 cells (GEO accession: GSE123902), with log-normalized expression values calculated from raw UMI counts. Color gradient: green (low, 0–4), white (median, 4), and red (high, 4–8); black boxes indicate undetected transcripts. Notably, CEACAM5 expression was significantly elevated in primary LUAD and brain metastases compared to normal lung (p = 0.0004 and p = 0.048, respectively), whereas EGFR expression was preferentially elevated in bone versus adrenal metastases (p = 0.0311). Statistical analyses were performed using two-way analysis of variance with Tukey’s multiple comparisons test; *P < 0.05, **P < 0.01, ***P < 0.001. (E) Bar plots depict mutation frequencies in canonical driver genes in metastatic LUAD (red, n = 2,133) versus primary LUAD (blue, n = 3,715). Alterations in TP53, KRAS, EGFR, CDKN2A, CDKN2B, STK11, KEAP1, MTAP, RBM10, and TERT were significantly enriched in metastatic samples, with TP53, KRAS, and EGFR showing the highest frequencies and statistically significant differences between cohorts. TP53 mutations were more prevalent in metastases (p-value < 0.001), whereas KRAS mutations were enriched in primary LUAD (p-value < 0.001). Genomic data were obtained from the MSK-CHORD cohort (Nature, 2024) via the cBioPortal for Cancer Genomics. Statistical comparisons were performed using two-sided Fisher’s exact test; **P < 0.01, ***P < 0.001

Clinical studies have established intratumoral heterogeneity as a major impediment to effective cancer treatment. In the TRACERx Lung study, patients with elevated subclonal diversity exhibited significantly higher relapse rates and shorter disease-free survival, directly linking intratumoral heterogeneity to unfavorable clinical outcomes [87]. Similarly, Yates et al. [88] demonstrated that subclonal driver mutations, such as PIK3CA in breast cancer, can persist or even expand under therapeutic pressure, diminishing treatment efficacy. In glioblastoma, distinct evolutionary trajectories defined by subclonal alterations in EGFR and PTEN have been implicated in the near-universal failure of conventional therapies [89]. In colorectal cancer, discordant KRAS and BRAF mutations between primary and metastatic lesions have been associated with resistance to anti-EGFR therapies [90]. Furthermore, McGranahan et al. [91] reported that tumors with a high burden of subclonal neoantigens exhibit poor T-cell infiltration and reduced responsiveness to immune checkpoint blockade.

Intertumoral heterogeneity emphasizes the genetic and phenotypic diversity observed among tumors from different patients, even within the same cancer type [92, 93]. For instance, two patients with lung cancer may harbor distinct driver mutations in genes, such as KRAS, EGFR, or ALK, each exerting unique influences on tumor behavior, immune interactions, and antigen presentation [94–97]. Data from The Cancer Genome Atlas reveal population-specific mutational patterns: EGFR mutations are more prevalent in East Asian patients with non-small-cell lung cancer, whereas KRAS mutations are more commonly observed in Western populations, contributing to differential responses to tyrosine kinase inhibitors [98]. In breast cancer, the PAM50 gene expression classifier categorizes tumors into discrete molecular subtypes, such as Luminal A, HER2-enriched, and Basal-like, each associated with distinct prognoses and therapeutic responses to endocrine or HER2-targeted therapies [99]. Similarly, in renal cell carcinoma, intratumoral heterogeneity significantly influences the response to targeted therapies. Distinct molecular subtypes of clear cell renal cell carcinoma exhibit differential sensitivities to pazopanib and sunitinib, emphasizing the role of intrinsic tumor biology in determining therapeutic outcomes [100]. Although the COMPARZ trial demonstrated comparable overall efficacy between pazopanib and sunitinib in terms of progression-free survival [101], the variability in patient-specific responses emphasizes the critical need for incorporating tumor heterogeneity into treatment decision-making. These clinical observations emphasize the necessity of precision oncology in navigating tumor heterogeneity. A standardized therapeutic strategy is frequently inadequate in the face of such biological complexity. Integration of molecular diagnostics and individualized tumor profiling into routine clinical practice is essential for informing treatment selection and optimizing clinical outcomes. In the context of CAR-T cell therapy, genetic and phenotypic differences among tumors can significantly influence therapeutic efficacy. Tumors expressing favorable antigen profiles may be more amenable to CAR-T cell targeting, whereas others may induce T-cell exhaustion within an immunosuppressive TME, compromising the durability of antitumor responses [102, 103].

Off-tumor effects

Off-tumor toxicity remains a significant safety concern in CAR-T cell therapy for solid tumors [104]. This phenomenon arises when CAR-T cells target normal tissues that express the same antigens as malignant cells [104]. In solid tumors, off-tumor toxicity is not solely attributable to static antigen expression; rather, it is a multifactorial issue that often eludes detection in preclinical models [104]. A particularly worrisome mechanism underlying off-tumor effects is inflammation-induced antigen upregulation. Upon CAR-T cell activation, pro-inflammatory cytokines, such as IFN-γ, tumor necrosis factor-alpha, and IL-6, can induce increased antigen expression in normal tissues that typically exhibit low or negligible antigen levels. This dynamic expansion of antigen expression can significantly broaden the spectrum of at-risk tissues, leading to severe toxicities in organs with low basal antigen expression [105]. Conventional antigen-screening methods, typically conducted under homeostatic and non-inflammatory conditions, fail to capture this cytokine-driven modulation of antigen expression, introducing a critical limitation in the predictive accuracy of preclinical toxicity assessments [105]. Furthermore, the TME of solid tumors, characterized by hypoxia, immune evasion, and aberrant signaling, may further promote aberrant antigen presentation on non-malignant cells, exacerbating the risk of CAR-T-mediated cytotoxicity [106, 107]. A recent study underscored the clinical relevance of these concerns, highlighting cases of severe on-target, off-tumor toxicity in solid tumor trials and emphasizing the pressing need for more predictive and physiologically relevant preclinical models [33] (Fig. 5).

Fig. 5.

Off-Tumor Toxicity of Chimeric Antigen Receptor T (CAR-T) Cells in Solid Tumors. CAR-T cells eliminate tumor cells by recognizing tumor-associated antigens (TAAs); however, many TAAs are expressed at varying levels in normal tissues, increasing the risk of off-tumor toxicity. Upon CAR-T cell activation, pro-inflammatory cytokines such as interferon-gamma (IFN-γ, TNF-α, and interleukin-6 (IL-6) can upregulate antigen expression in otherwise unaffected normal tissues, expanding the range of cells susceptible to CAR-T-mediated cytotoxicity. In addition, the tumor microenvironment (TME), characterized by hypoxia, immunosuppressive cytokines, and aberrant antigen processing and presentation, may inadvertently enhance antigen expression in surrounding healthy tissues. These factors collectively contribute to unintended off-tumor effects and limit the therapeutic window of CAR-T cell therapy in solid tumors

Beyond antigen selection, the challenge of mitigating off-tumor toxicity in CAR-T cell therapy for solid tumors involves a complex interplay of biological, immunological, and tissue-specific factors that collectively influence CAR-T cell behavior [105]. Recent studies have investigated genetic modifications, such as the development of “armored” CAR-T cells incorporating enhanced costimulatory domains, to improve antigen specificity and functional persistence [66]. However, the inherently limited specificity of TAAs remains a major hurdle [66]. Notably, off-tumor toxicity should not be regarded as an isolated or sporadic adverse event but as a foreseeable outcome of systemically administered CAR-T cells targeting non-exclusive antigens in a highly dynamic and heterogeneous TME [108]. This challenge transcends the limitations of antigen selection alone, revealing fundamental gaps in our understanding of antigen regulation, immune activation thresholds, and the tissue-specific intricacies of inflammatory responses [108]. Until these complexities are systematically addressed, widespread clinical implementation of CAR-T cell therapy for solid tumors will remain constrained by unacceptable toxicity risks, emphasizing the urgent need for physiologically relevant preclinical models and more refined, antigen-specific targeting strategies [104]. These challenges emphasize the need for innovative strategies, which are discussed in the following section.

Taken together, these convergent barriers necessitate the integrated translational strategies detailed in Sect. "Emerging strategies and future directions"—spanning memory T-cell–centric engineering and TME modulation—the implementation of which is illustrated in a multiple-myeloma case study (Sect. "Memory CAR-T cells: The next frontier").

Emerging strategies and future directions

Clinical application of CAR-T cell therapy in solid tumors has been limited by several formidable challenges, as outlined in Sect. "Challenges in applying CAR-T to solid tumors". This section presents emerging strategies designed to overcome these obstacles and enhance the efficacy and safety of CAR-T therapy for solid tumors; these include multi-antigen targeting constructs to mitigate antigen escape, armored CAR-T cells capable of modulating the TME, logic-gated systems to enhance tumor specificity and safety, and memory-enriched CAR-T products designed for sustained persistence and durability. To address these challenges, innovative strategies have been introduced (Table 2). The following sections delineate these emerging strategies and explore their potential to shape the future landscape of CAR-T cell therapy in solid tumors.

Multi-antigen targeting strategies

Solid tumors frequently exhibit antigenic heterogeneity, characterized by variable expression levels of target antigens across tumor cell populations, facilitating immune evasion. To address this challenge, multi-antigen targeting strategies have been developed [109] (Table 3). Dual-target CARs, designed to recognize two distinct antigens simultaneously, such as HER2 and mucin 1 in ovarian cancer, broaden tumor coverage and reduce the risk of immune escape by targeting multiple tumor subpopulations [3, 110]. Preclinical studies have demonstrated that dual-target CARs outperform single-antigen constructs in eliminating antigen-heterogeneous tumors [111, 112]. Tandem CARs, incorporating dual specificity within a single receptor framework—such as those targeting HER2 and IL13Rα2 in glioblastoma—further enhance antitumor potency [113, 114]. Furthermore, CAR-T cells engineered to secrete bispecific T-cell engagers (BiTEs) can recruit endogenous bystander T cells, amplifying the immune response against tumors with heterogeneous antigen expression [115, 116]. These strategies build on recent advancements, including dual-target CARs—such as HER2/PD-L1 constructs—designed to address antigen variability and the immunosuppressive TME (NCT04684459) [117]. Similarly, BiTE-secreting CAR-T cells have demonstrated enhanced immune recruitment in preclinical studies [116]. Collectively, these advancements emphasize the therapeutic potential of multi-antigen targeting strategies and underscore the need for continued clinical evaluation to optimize their efficacy in solid tumors.

Table 3.

Next-generation CAR designs for solid tumors — preclinical and clinical strategies

Stage	Strategy class	Specific strategy (examples)	Mechanistic basis	Representative evidence & stage	Selected trials/outcomes
Preclinical Innovations	Multi-antigen designs	Tandem/dual CARs; BiTE-secreting CARs (e.g., EGFRvIII-BiTE)	Broaden coverage; recruit bystander T cells; mitigate escape	BiTEs enhance clearance of antigen-negative clones; preclinical [113]	—
	Logic-gated circuits	AND/NOT; SynNotch/E-SYNC	Context-restricted activation to improve specificity	Validated preclinically; SynNotch entered FIH; preclinical to early clinical [118–121]	NCT06186401 [119]
	Universal/adapter CARs	Modular adaptors	Rapid, modular retargeting	Robust preclinical PoC; preclinical [122, 123]	—
	Hypoxia/context-responsive	HRE-driven CARs	Activity limited to hypoxic tumor sites	Preclinical efficacy in solid tumors [107]	—
	Armored cytokines	IL-18 TRUCKs	Counter TME suppression; enhance persistence	Preclinical [124]	—
	ECM-modifying payloads	Heparanase	Improve infiltration through dense stroma	Preclinical [125]	—
	Chemokine-receptor engineering	CXCR2	Tumor-directed trafficking	Preclinical homing gains [126]	—
	Oncolytic virus combos	Oncolytic vaccinia	Antigen up-regulation & TME disruption	Preclinical [127]	—
	Radiotherapy combos	CAR-T + RT (breast models)	Upregulate TAAs, remodel TME	Preclinical [128]	—
	CAR-enhancers	Selective IL-2 pathway enhancer	Amplify activation/memory	Preclinical (2025) [129]	—
	In-vivo CAR generation	LNP/non-viral targeted delivery	Direct in-body programming of T cells	Murine/NHP efficacy; preclinical to early clinical trials [130–133]	—
Clinical Translation	Dual-target CARs	HER2/PD-L1	Expand coverage; address heterogeneity/TME	Early-phase	NCT04684459 [117]
	Tandem CARs	EGFR/HER2 (GBM)	Dual specificity in one receptor	Early clinical evaluation	[10, 113]
	Armored CARs	IL-12–secreting (ovarian)	TME reprogramming	Early feasibility	NCT01548434 [134]
	Checkpoint combinations	CAR-T + PD-1 blockade	Reverse exhaustion; ↑ persistence	Early-phase combinations	Clinical activity reports [135, 136]
	Safety switches	iCasp9	Rapid ablation on toxicity	Clinical demonstration	Pediatric setting [137, 138]
	Transient expression	mRNA CARs	Limit persistence → safety	First-in-human signals	Early FIH [139–141]
	Local/locoregional delivery	IL13Rα2 (CNS)	Bypass systemic barriers	Multiple Phase I completed	[55, 142, 143]
	Memory-enriched products	TSCM-enriched	Durable surveillance & control	Early exploration	[144–146]

Co-expression of transgenic T cell receptors (TCRs) and CARs

A novel strategy to enhance the specificity and efficacy of CAR-Tcell therapy for solid tumors involves engineering T cells to co-express TCRs and CARs. Unlike conventional CAR-T cells, which target surface antigens, this dual-receptor strategy allows T cells to recognize intracellular antigens presented by major histocompatibility complex molecules via TCRs, and surface antigens via CARs, broadening the spectrum of targetable tumor antigens. Smith et al. [147] demonstrated that TCRs targeting neoantigens (HATL8F/p53R175H) and CARs targeting HER2 exhibited enhanced antitumor efficacy and reduced toxicity in humanized solid tumor mouse models. High-throughput analyses revealed that strong TCR-antigen interactions synergistically enhanced CAR activation, whereas weak TCR engagement could attenuate CAR-mediated responses. This dual-receptor system leverages the interplay between TCR and CAR signaling to fine-tune T-cell activation, offering a promising strategy to address key challenges, such as antigen heterogeneity and off-tumor toxicity in solid tumors. Careful selection of antigen targets and optimization of receptor expression are essential to maximizing therapeutic benefit. However, despite their potential, dual-receptor approaches face notable clinical challenges. In particular, co-expression of both targeted antigens on tumor cells is required to achieve complete tumor coverage; thus, intratumoral antigenic heterogeneity may limit efficacy and contribute to immune escape. In addition, the design and manufacturing of dual-target CAR-T cells are inherently complex, requiring advanced engineering approaches, such as logic-gated CARs (AND-gate systems) and SynNotch platforms to enhance specificity and safety. Although these technologies hold considerable promise, they remain in early stages of clinical development, with preliminary trials yielding variable outcomes. For instance, a phase I trial of prostate-specific membrane antigen (PSMA)-specific CAR-T cells in prostate cancer demonstrated clinical responses in only two of five patients, emphasizing the need for further optimization [148]. Furthermore, simultaneous targeting of two antigens may increase the risk of off-tumor toxicity if antigen is expressed on normal tissues. In addition, excessive or chronic overstimulation of CAR-T cells, particularly in the immunosuppressive TME of solid tumors, can induce T-cell exhaustion and impair therapeutic efficacy [149]. Therefore, although multi-antigen targeting represents a promising strategy to overcome tumor heterogeneity and escape, it remains an area of ongoing research requiring significant refinement to achieve durable and consistent clinical benefit [150, 151].

Multi-antigen targeting strategies represent a pivotal advancement in CAR-T cell therapy for solid tumors, offering a direct means to address antigenic heterogeneity through dual-target CARs, tandem constructs, BiTEs, and TCR-CAR co-expression. Clinical progress—exemplified by responses in DMG—alongside emerging technologies, such as the TCR-CAR dual-receptor system, emphasize the transformative potential of these platforms. Despite ongoing challenges, including manufacturing complexity and the risk of off-tumor toxicities, the rapid progress in preclinical and clinical studies suggest that these strategies are well-positioned to redefine cancer immunotherapy. As these innovations are integrated into clinical practice and further validated in trials, CAR-T therapy may overcome the long-standing barriers posed by solid tumors, ushering in a new era where even the most resistant cancers become amenable to targeted cellular therapies.

Armored CAR-T cells for TME reprogramming

The TME of solid tumors is characterized by immunosuppressive cytokines, dense physical barriers, such as the extracellular matrix (ECM), and other factors that impair CAR-T cell efficacy. Various engineering strategies have been developed to augment CAR-T cell performance within these hostile settings. One promising approach involves the use of “armored” CAR-T cells that are engineered to secrete pro-inflammatory cytokines such as IL-12 or IL-18 (Fig. 6A) [103, 152, 153]. These cytokine-secreting CAR-T cells actively remodel the TME by recruiting innate immune effectors and antagonizing immunosuppressive signals, particularly those mediated by TGF-β [152, 153]. In preclinical models of ovarian cancer, IL-12-secreting CAR-T cells have demonstrated enhanced immune cell infiltration and cytotoxicity, resulting in the regression of tumors that are otherwise resistant to standard CAR-T therapies [154, 155]. Building on this concept, “CAR-Enhancers”, which are aimed at improving CAR-T cell survival and functional potency. A prominent approach involves cytokine-armored CAR-T cells, also referred to as T cells, redirected for universal cytokine-mediated killing (TRUCKs), and designed to secrete cytokines, such as IL-12, IL-18, or IL15 upon activation. These cytokines help counteract the immunosuppressive TME and promote CAR-T cell survival. Preclinical studies have revealed that IL-12-secreting CAR-T cells suppress regulatory T cell activity and enhance IFN-γ production, resulting in improved tumor clearance in ovarian and pancreatic cancer models [156]. Similarly, IL-18-secreting CAR-T cells have been demonstrated to reprogram T cells into resilient effector cells, significantly prolonging survival in murine solid tumor models [124]. A phase 1 trial (NCT05694364) evaluating IL-12-secreting CAR-T cells targeting mesothelin in pancreatic cancer reported prolonged CAR-T cell persistence and partial responses in three of eight patients [157]. In parallel, a phase I trial (NCT04556669) is evaluating anti-PD-L1 armored CAR-T cells in patients with solid tumors [158], and ongoing recruitment in subsequent studies further emphasizes the growing momentum and therapeutic promise of this approach (NCT06939270) [159]. In addition, combining CAR-T cell therapy with immune checkpoint inhibitors, such as anti-programmed cell death protein 1 (PD-1) or anti-CTLA-4 antibodies, has shown efficacy in reversing T-cell exhaustion [116, 136]. Preliminary clinical trials in lung cancer suggest that combinatorial strategies enhance CAR-T cell persistence and functional potency [160, 161]; these findings are further supported by the ongoing trial NCT04684459, which is evaluating HER2/PD-L1 dual-target CAR-T cells in solid tumors [117]. Alongside, gene editing using CRISPR-Cas9 further enhances CAR-T cell function by eliminating inhibitory checkpoints; for instance, deletion of PD-1 or CTLA4 genes enhances resistance to tumor-mediated immunosuppression, whereas ablation of RASA2 increases antigen sensitivity and long-term functionality [162, 163]. Similarly, targeting REGNASE-1 has been shown to reprogram CAR-T cells into long-lived effector cells with superior durability [164]. Recent clinical trials, such as those combining MUC1-targeted CAR-T cells with PD-1 knockout for advanced breast cancer and non-SCLC, have integrated these safety mechanisms to enhance the therapeutic index [62, 165].

Fig. 6.

Next-Generation Chimeric Antigen Receptor (CAR) Architectures for Solid Tumor Immunotherapy. Schematic overview of logic-gated CAR-T cell designs engineered to enhance antigen specificity, mitigate off-target toxicity, and overcome the immunosuppressive tumor microenvironment (TME) in solid tumors. (A) Armored CARs are modified to secrete immunomodulatory molecules—such as cytokines or monoclonal antibodies—upon antigen engagement; these designs reinforce anti-tumor responses and reprogram the TME through antigen-induced transcriptional circuits. (B) OR-gate CARs broaden the range of tumor antigen recognition by triggering T cell activation upon detection of either of two distinct antigens. This can be achieved via co-expression of dual CARs targeting separate antigens or tandem CAR constructs harboring dual antigen-binding domains within a single receptor. (C) AND-gate CARs improve tumor specificity by requiring simultaneous recognition of two antigens for full T cell activation. This dual-antigen requirement can be implemented using split CAR designs or combinatorial systems that distribute activation motifs across separate receptors, converging on canonical T cell receptor (TCR) signaling to minimize off-tumor cytotoxicity. (D) NOT-gate CARs integrate inhibitory modules—typically immunoreceptor tyrosine-based inhibitory motifs (ITIMs) derived from checkpoint molecules, such as PD-1—to suppress CAR-T activation upon detection of antigens expressed on healthy tissues, introducing a safety mechanism via negative regulation. (E) IF–THEN-gate CARs are conditional systems that activate effector functions only upon simultaneous recognition of a primary tumor-associated antigen and a secondary context-specific cue (neoantigen or tissue-restricted marker). This design allows for spatial and contextual precision in targeting tumors with heterogeneous antigen expression. In all panels, green marks  indicate productive antigen recognition and T cell activation, whereas red crosses  indicate signal insufficiency or functional inhibition

Enhancing trafficking and infiltration

To overcome the ECM-associated physical barriers, CAR-T cells have been engineered to express matrix-degrading enzymes such as heparanase or hyaluronidase, which facilitate tumor infiltration by degrading structural components like collagen and hyaluronan [166, 167]. Preclinical studies in pancreatic cancer have demonstrated that CAR-T cells engineered to express matrix-degrading enzymes can achieve deeper tumor penetration [125, 168]. However, safety concerns persist, particularly regarding the potential for off-target degradation of extracellular matrix components in healthy tissues. Another critical area of innovation involves improving CAR-T cell trafficking to tumors; this has been achieved by engineering CAR-T cells to express chemokine receptors such as CXCR2, which respond to tumor-derived chemokines like CXCL1, enhancing targeted migration and infiltration [168, 169]. In lung cancer models, CXCR2-modified CAR-T cells exhibited a tenfold increase in intratumoral accumulation, resulting in significantly improved therapeutic efficacy [170]. NCT05353530 is assessing IL-8 receptor–modified CD70-targeted CAR-T cells in patients with glioblastoma [171]. Local administration of CAR-T cells, whether directly into the tumor mass or into resection cavities, offers a strategy to circumvent systemic delivery barriers and concentrate therapeutic activity at the tumor site [172, 173]. In glioblastoma models, locally delivered IL13Rα2-targeted CAR-T cells induced encouraging, albeit transient, responses [54, 55, 142]. Although these strategies significantly improve CAR-T cell tumor localization, further development is required to establish scalable and clinically applicable delivery protocols [54, 55, 142]. A notable example is a recent phase 1 trial in diffuse intrinsic pontine glioma (DIPG), where repeated ICV administration of B7-H3-targeted CAR-T cells demonstrated a favorable safety profile and robust immune activation within the central nervous system (CNS). Elevated levels of cytokines, including C-X-C motif chemokine ligand 10, granulocyte macrophage colony-stimulating factor, IFN-γ, and Thymus and activation-regulated chemokine in cerebrospinal fluid, were detected in the cerebrospinal fluid, indicating a potent immunologic response. The trial reported a median survival of 19.8 months from diagnosis, significantly exceeding the historical median of 11.2 months [7]. These findings emphasize the therapeutic potential of integrating locoregional CAR-T delivery with TME modulation to achieve meaningful clinical outcomes in aggressive solid tumors.

Safety-engineered designs and combinatorial therapies

Combination therapies are increasingly recognized as potent strategies to enhance CAR-T cell efficacy. One promising approach involves the integration of CAR-T therapy with oncolytic viruses, which selectively lyse tumor cells and unmask otherwise concealed TAAs, enhancing T-cell recognition and activation [127, 174]. Another synergistic strategy pairs CAR-T cells with radiotherapy, which upregulates the expression of TAAs and disrupts the TME, rendering malignant cells more vulnerable to immune-mediated destruction [175]. Preclinical studies in breast cancer models have demonstrated the capacity of this combinatorial approach to augment CAR-T cell antitumor activity significantly [175]. These strategies are currently under clinical evaluation. For instance, trial NCT03740256 is investigating HER2-targeted CAR-T cells in combination with an oncolytic virus for advanced solid tumors, with ongoing recruitment and preliminary signals of enhanced tumor targeting as of 2025 [176]. Similarly, NCT06348797 is assessing the combination of delta-like protein 3 (DLL3)-targeted CAR-T cells with radiotherapy in R/R small cell lung cancer (SCLC), with early phase 1 data indicating early disease control [177]. Despite these advancements, off-tumor toxicity remains a significant safety concern in solid tumors, as many target antigens are expressed on vital normal tissues [33]. To mitigate this risk, multiple safety-engineering strategies have been developed to minimize off-tumor effects and enhance the therapeutic index of CAR-T cell therapy. Logic-gated CAR constructs, such as AND-gate systems, enhance tumor specificity by requiring the coexpression of HER2 and EGFR for activation, minimizing off-tumor cytotoxicity to healthy tissues (Fig. 6B–E) [178, 179]. Preclinical studies in glioblastoma models have demonstrated the efficacy of AND-gate CARs targeting IL13Rα2 and EGFRvIII, effectively eliminating tumor cells while sparing normal cells that express only one of the target antigens [55, 113]. The synthetic notch receptor (SynNotch) CAR platform offers an additional layer of specificity by enabling CAR-T cell activation after the sequential recognition of distinct tumor-associated antigens. This advanced system significantly enhances tumor selectivity and reduces the risk of off-target effects [118, 120]. These strategies are currently being evaluated in clinical settings—for instance, SynNotch CAR-T cells targeting EGFRvIII are under investigation in glioblastoma (NCT06186401) [119]. In addition, the incorporation of iCasp9 suicide switches has shown efficacy in preclinical sarcoma models, enabling rapid, small-molecule–induced ablation of CAR-T cells in the event of toxicity [160]. Another promising safety-enhancing strategy involves transient CAR expression through messenger RNA (mRNA) electroporation, which limits CAR-T cell persistence and reduces the risk of long-term off-tumor toxicity. Early phase studies in liver cancer models have demonstrated the feasibility, safety, and immunologic efficacy of mRNA-based CAR-T cell therapies [139, 140]. To address the unique therapeutic challenges posed by solid tumors, both transient and stable CAR expression platforms are being actively investigated. mRNA-based approaches offer a favorable safety profile by avoiding permanent genomic integration, reducing the risk of long-term off-tumor toxicity. Meanwhile, gene-editing technologies, such as CRISPR-Cas9 and non-viral transposon systems, allow targeted insertion of CAR constructs at specific loci, ensuring consistent expression while minimizing the risk of insertional mutagenesis [180]. Recent advances in in vivo gene delivery, particularly via lipid nanoparticle (LNP) technologies, have enabled precise, efficient genomic editing in physiological settings, achieving high efficacy with limited off-target effects [180]. Notably, the phase 1/2 trial (NCT02414269), combining mesothelin-targeted CAR-T cells with pembrolizumab in malignant pleural mesothelioma, reported a 1-year OS rate of 83%, including complete metabolic responses in two patients [180].

The antigenic and genetic heterogeneity of solid tumors presents a significant barrier to the efficacy of CAR-T cell therapy, as variable antigen expression enables tumor subclones to evade immune detection. Addressing this complexity requires a multifaceted strategy that surpasses single-antigen targeting [122]. One promising strategy involves the development of universal CARs equipped with modular antigen-binding domains, enabling recognition of a broader spectrum of patient-specific mutations, including those derived from unique genetic alterations [122, 181]. Preclinical studies have demonstrated that universal CARs using adaptors, such as biotin-binding immune receptors or fluorescein isothiocyanate-conjugated ligands, can simultaneously target multiple TAAs, enhancing tumor coverage in models of pancreatic and breast cancer [123]. Although these platforms offer enhanced flexibility and personalization, they introduce technical challenges, including complex manufacturing processes and the requirement for comprehensive and precise tumor antigen profiling [122, 181]. Furthermore, combinational approaches incorporating bispecific antibodies, immune-modulating agents, oncolytic viruses, or radiotherapy can enhance CAR-T cell efficacy by expanding the antigenic repertoire and alleviating immunosuppression within the TME [182]. Bispecific antibodies, such as those targeting EGFR and CD3, redirect endogenous T cells to eliminate antigen-negative tumor cells, complementing CAR-T cell activity. Similarly, oncolytic viruses selectively lyse tumor cells, promoting the release of neoantigens and enhancing immune visibility, whereas radiotherapy upregulates TAA expression, rendering malignant cells more susceptible to CAR-T-mediated cytotoxicity. Preclinical models of lung and breast cancer models have demonstrated the synergistic potential of these combinational strategies, with significantly enhanced tumor regression observed when CAR-T cells are combined with oncolytic virotherapy or radiotherapy [127, 174, 175]. Recent clinical trials further support the translational relevance of these approaches. For instance, an ongoing clinical trial (NCT03415100) is evaluating allogeneic or autologous NKG2DL-targeting CAR-NK cells in patients with metastatic solid tumors, reporting preliminary evidence of safety and feasibility across diverse tumor types as of 2025 [183]. Similarly, early-phase results from a separate trial (NCT05137275) investigating 5T4-targeted CAR-NK cells in advanced solid tumors have demonstrated early signals of antitumor activity and persistence in heterogeneous tumor settings [184].

To predict and mitigate off-tumor toxicity, several evaluation methodologies are used in preclinical settings to assess CAR-T cell specificity and safety, with the aim of enhancing the accuracy of toxicity predictions and guiding the development of safer therapies. In vitro assays using cells with low antigen expression are used to define activation thresholds, often using engineered cell lines expressing targets such as HER2 or GD2, along with standardized assays incorporating artificial antigen-presenting particles to ensure reproducibility [185]. Recent advancements have further refined these assays by incorporating inflammatory stimuli, enhancing their predictive value under conditions that more closely mimic the TME [186, 187]. Complementary approaches include cytokine production assays, which quantify the release of IFN-γ, IL-6, and tumor necrosis factor (TNF)-α release via enzyme-linked immunosorbent assay, multiplex techniques, or single-cell profiling; these techniques elucidate response heterogeneity and demonstrate how inflammation may potentiate off-tumor effects [188]. In addition, patient-derived organoids (PDOs) offer a physiologically relevant platform for studying CAR-T cell interactions with tumor and normal tissues. By preserving the histopathological, genomic, and antigenic characteristics of primary tissues; PDOs enable a more accurate assessment of off-tumor toxicity [189].

Several innovative strategies have been developed to mitigate on-target off-tumor toxicity (Table 4). These include logic-gated CAR-T cells, such as those targeting ROR1 in breast and lung cancers (NCT05274451) [190], which enhance specificity by requiring dual antigen recognition, minimizing cytotoxicity to normal tissues. The EVEREST-1 phase 1/2 trial (NCT05736731) [191] is currently evaluating a logic-gated CAR-T cell product, A2B530, in patients with solid tumors [192]. Building on this approach, dual-target and multivalent CAR-T cells simultaneously target multiple antigens, reducing the risk of off-tumor toxicity by necessitating co-expression of antigens for full activation. This is exemplified by a phase 1 breast cancer trial (NCT04430595), which targets HER2, GD2, and CD44v6, showing preliminary specificity improvements [193]. Similarly, dual-target CAR-T cells directed against PRDM1 and NR4A3 have demonstrated enhanced antitumor activity with reduced toxicity in preclinical models [104]. Furthermore, safety-engineered CAR-T cells incorporate switches, such as iCasp9, enabling rapid elimination of CAR-T cells in response to severe adverse events, as validated in clinical trials for sarcoma (NCT01953900) [160]. Transient CAR expression via mRNA electroporation offers an additional safety layer by limiting persistence and reducing the risk of long-term off-tumor toxicity [139, 140].

Table 4.

Ongoing clinical trials focused on enhancing efficacy and reducing toxicity of CAR-T cell therapy in solid tumors

Therapy	Target	Tumor type	Clinical trial	Efficacy & side effects	Status and citation
Dual-target CAR-T	HER2/PD-L1	Peritoneal carcinoma metastatic, pleural effusion, malignant	NCT04684459, early phase I	No results posted; early data suggest manageable toxicities	Recruiting [117]
Armored CAR-T	Anti-PD-L1 armored anti-CD22 CAR-T/CAR-TILs	Patients with solid tumors	NCT04556669, Phase I,	No results posted; ongoing evaluation of TME modulation	Recruiting [158]
Chemokine receptor-modified CAR-T	IL-8 receptor-modified CD70 CAR-T	Adult glioblastoma	NCT05353530, Phase I	No results posted; focus on enhancing CAR-T cell trafficking	Recruiting [171]
SynNotch CAR-T	EGFRvIII	Glioblastoma	NCT06186401, phase I	No results posted; evaluating tumor-specific activation mechanisms	Recruiting [119]
Suicide gene-CAR-T	GD2	Neuroblastoma	NCT01822652, phase 1	No results posted; Preliminary data suggest safety with iCasp9 switch	Recruiting [137]
CAR-T + oncolytic virus	HER2	Advanced HER2-positive solid tumors	NCT03740256, phase I	No results posted; exploring synergistic effects	Recruiting [176]
CAR-T + radiotherapy	Notch ligand delta-like ligand 3 (DLL3)	R/R SCLC	NCT06348797, phase I	No results posted; planned trials aim to assess TME disruption	Not yet recruiting [177]
Dual-target CAR-T	CART-EGFR-IL13Ra2	Glioblastoma	NCT05168423	Rapid tumor shrinkage with some durable responses; manageable neurotoxicity (> 50%) and mild CRS; MTD: 2.5 × 10⁷ cells	Recruiting [196]
Dual-targeting CLDN18.2 and PD-L1 CAR-T	CLDN18.2	Gastric	NCT06084286	Approximately 38% ORR; predominantly mild CRS and manageable hematologic toxicities	Ongoing Phase 1; Recruiting [197]

A pivotal phase I trial, BrainChild-03 Arm C, led by Brown et al. [7] evaluated B7-H3-targeted CAR-T cells in 23 patients with DIPG, 21 of whom received treatments via intraventricular (ICV) infusions. DIPG, a fatal pediatric brainstem tumor, has a historical median survival of 11.2 months, emphasizing the urgent need for effective therapies. B7-H3 (CD276), an immune checkpoint molecule overexpressed in DIPG and other pediatric solid tumors (osteosarcoma and rhabdomyosarcoma), was selected as the target due to its high tumor specificity and limited expression in normal tissues, minimizing the risk of off-tumor toxicity. The trial implemented repeated ICV infusions without lymphodepletion, using intra-patient dose escalation to optimize CAR-T cell exposure and persistence within the CNS. This locoregional approach was designed to overcome systemic delivery barriers, such as poor tumor penetration and immunosuppressive microenvironments, and to promote sustained intratumoral activity. The findings were encouraging: median survival from CAR-T infusion reached 10.7 months, and median survival from diagnosis extended to 19.8 months—substantially longer than historical benchmarks. Notably, 3 patients survived beyond 40 months, indicating potential for long-term benefit. Among 18 evaluable patients, 1 achieved a partial response and 15 exhibited stable disease, indicating effective disease control. Evidence of immune activation was observed through CAR-T cell detection in the cerebrospinal fluid, accompanied by elevated cytokine levels and T cell proliferation, suggesting robust anti-tumor activity within the CNS TME. The therapy was well tolerated, with predominantly low-grade (grade 1–2) adverse events, including headaches, nausea, and fatigue, with no dose-limiting toxicities. This trial supports B7-H3 as a promising target for DIPG and validates ICV delivery as a viable strategy to enhance CAR-T cell efficacy in CNS tumors. The FDA’s recent designation of BCB-276—the B7-H3 CAR-T cell therapy used in this study— as a breakthrough therapy further emphasizes its transformative potential for DIPG and other B7-H3-expressing tumors [194].

Beyond the GD2 and B7-H3 trials, recent studies have further expanded the applicability of CAR-T cell therapy to a broader range of solid tumors. Qi et al. [195] evaluated CLDN18.2-targeted CAR-T cells in advanced gastric cancer, a disease with limited treatment options. CLDN18.2, a tight junction protein frequently overexpressed in gastric tumors, was targeted in a cohort of heavily pretreated patients, yielding an ORR of 57.1%, a disease control rate (DCR) of 75%, and a 6-month OS rate of 81.2%. Similarly, Zhang et al. [50] investigated CD133-targeted CAR-T cells in pancreatic cancer, a malignancy known for its dense stromal architecture and immunosuppressive TME. CD133, a marker of cancer stem cells, was targeted in 7 patients, resulting in 1 partial response and stable disease in 3 others.

The application of CAR-T cell therapy to solid tumors remains hindered by several challenges, including the immunosuppressive TME, antigenic heterogeneity, and the risk of on-target off-tumor toxicity. To overcome these barriers, recent advancements in genetic engineering have prioritized the precise genomic integration of CAR constructs to enhance efficacy and safety. These next-generation strategies harness cutting-edge gene-editing technologies and non-viral delivery methods, addressing the inherent limitations of conventional viral vector-based systems and offering a promising framework for effective solid tumor treatment.

CRISPR-Cas9 has emerged as a transformative tool for precise genomic modifications in CAR-T cells, enabling targeted gene knock-ins and knock-outs to enhance functionality and persistence in the immunosuppressive TME. For instance, Zhang et al. [198] used CRISPR-Cas9 to disrupt the TGFβRII gene, enhancing CAR-T cell efficacy by mitigating TME-mediated inhibitory signaling in preclinical models. Similarly, Prinzing et al. [199] demonstrated that deletion of DNMT3A in CAR-T cells prevents exhaustion and augments antitumor activity, emphasizing the potential for sustained therapeutic benefits. In a phase I clinical trial, anti-mesothelin CAR-T cells with CRISPR-mediated PD-1 disruption were evaluated in patients with metastatic mesothelin-positive solid tumors [200]. The trial reported no dose-limiting toxicities among 15 patients, with two achieving stable disease, indicating the feasibility and safety of this gene-editing approach [200]. Another phase I trial (NCT03399448) assessed CRISPR-edited T cells targeting NY-ESO-1 in patients with advanced sarcomas and MM, demonstrating manageable safety profiles and preliminary signs of clinical activity [201]. Complementing these gene-editing approaches, non-viral gene delivery methods offer safer and more scalable alternatives to traditional viral vectors, which carry the risk of random genomic integration and associated off-target effects. The Sleeping Beauty transposon system has been used to achieve stable CAR expression with reduced immunogenicity. Monjezi et al. [202] demonstrated its efficacy in engineering CAR-T cells for B-cell non-Hodgkin lymphoma, achieving favorable safety and therapeutic outcomes. More recently, Hamilton et al. [180] developed an in vivo T cell engineering platform using enveloped delivery vehicles, enabling precise CAR gene integration directly in the host, potentially streamlining manufacturing processes and lowering production costs. A phase I trial evaluating non-viral Sleeping Beauty transposon-based CAR-T cells targeting CD19 in heavily pretreated patients with R/R B-cell hematologic malignancies, including ALL, DLBCL, and CLL, reported detectable CAR expression and an encouraging safety profile, supporting the feasibility of extending this strategy to solid tumors [203]. Compared to the traditional viral vector-based CAR-T cell therapies, which, despite their efficacy in hematologic malignancies, often result in random integration and variable CAR expression, precise integration technologies, such as CRISPR-Cas9 and transposon systems offer consistent CAR expression and improved safety. These attributes are particularly critical for solid tumors, where long-term CAR-T cell persistence is essential to overcome tumor heterogeneity and the immunosuppressive TME (Table 5).

Table 5.

Clinical trials highlighting precise genomic integration strategies in CAR-T Cell therapy for solid tumors

Trial ID	Target	Tumor type	Strategy	Key findings	Status
NCT03545815	Mesothelin	Mesothelin-positive solid tumors	CRISPR-Cas9-mediated PD-1 disruption	No dose-limiting toxicities; 2 of 15 patients achieved stable disease	Completed [204]
NCT03399448	NY-ESO-1	Sarcomas, MM, melanoma	CRISPR-edited CAR-T cells	Manageable safety profile; preliminary efficacy signals	Recruiting [205]
NCT03970382	Neoantigens	Advanced solid tumors	CRISPR-Cas9	Treatment was generally well-tolerated. No off-target effects from CRISPR-Cas9 editing were observed	Ongoing (Phase 1) [206]
NCT03545815	Mesothelin (with PD-1 KO)	Mesothelin-positive solid tumors	CRISPR-Cas9	Treatment was well-tolerated and safe, but efficacy was modest. Optimizations are needed to improve persistence and activity	Completed (Phase 1) [204]
NCT05239143	MUC1-C	Advanced epithelial-derived solid tumors	Cas-CLOVER (CRISPR variant)	Allogeneic CAR-T with precise knockouts to minimize alloreactivity; CAR integrated via non-viral transposon, but edits enable stable expression. Preliminary efficacy: partial response in breast cancer at 0.75 × 106 cells/kg; no bridging therapy, median 9-day enrollment-to-treatment	Recruiting (Phase 1) [207]
NCT04510051	IL13Rα2	Pediatric brain tumors	Enhanced genomic stability	Improved persistence; early signs of tumor control	Recruiting [208]
NCT01822652	GD2	Neuroblastoma	iCasp9 safety switch	Rapid CAR-T elimination feasible	Recruiting [137]

In vivo generation of CAR-T cells, which involves the direct delivery of CAR genes into a patient’s T cells within the body, eliminates the need for ex vivo manipulation. This innovative methodology has the potential to streamline manufacturing processes, reduce associated costs, and enhance clinical outcomes, particularly for solid tumors—representing a fundamental advancement in immunotherapy. For instance, Zhou et al. [130] engineered an LNP system conjugated with CD3 antibodies to target selectively T cells in vivo, enabling the delivery of a plasmid encoding a CD19-sepcific CAR along with an IL-6 short hairpin RNA. In a murine leukemia model, this strategy demonstrated antitumor effects comparable to that of ex vivo-generated CAR-T cells, with transgene expression persisting for up to 90 days and a marked reduction in CRS [130]. Although primarily applied to hematologic malignancies, the modularity of the LNP platform permits straightforward adaption to target solid tumor antigens, such as HER2 or EGFR, by altering the CAR construct, broadening its applicability to solid tumor immunotherapies. A recent study described an in vivo engineering strategy using targeted LNP (tLNP) for mRNA delivery to specific T cell populations [132]. These tLNP effectively reprogrammed CD8⁺ T cells from healthy donors and patients with autoimmune disorders, leading to tumor suppression in humanized mouse models and B cell depletion in cynomolgus monkeys. Notably, the repopulated B cells following depletion were predominantly naïve, indicating a potential immune system reset that may support durable antitumor responses in solid tumors [132]. By providing a scalable and rapidly deployable framework, this tLNP-based approach has the potential to reshape fundamentally CAR-T applications in solid malignancies. Notably, CAR-T cells generated in vivo may exhibit enhanced persistence and functionality within the TME, avoiding the exhaustion commonly associated with ex vivo expansion [133]. These in vivo CAR-T platforms are now advancing into clinical trials, representing a pivotal step toward clinical translation [131]. Further insights have emerged from the in vivo engineering of other CAR-expressing immune cells. For instance, Kang et al. [209] used DNA/polymer nanocomplexes to introduce CAR and IFN-γ genes into macrophages in vivo, generating CAR-M1 macrophages with potent antitumor activity in a murine breast cancer model. Although clinical trials for in vivo CAR-T cell therapy in solid tumors are in early stages, the field is advancing rapidly. A phase 1 trial (NCT01897415) evaluated mRNA-engineered autologous T cells targeting mesothelin in pancreatic cancer, demonstrating feasibility, safety, and preliminary evidence of tumor response. In addition, a recent report emphasized ongoing efforts to initiate clinical trials for in vivo CAR-T therapies targeting solid tumor antigens, such as EGFR and HER2; however, specific trial identifiers were not disclosed [210]. Despite these advancements, several challenges persist. Precise delivery to T cells is essential to prevent off-target effects, as non-specific transfection may lead to unintended CAR expression in other cell types, increasing toxicity risks in solid tumors. Moreover, sustaining long-term CAR expression and function in the hostile TME remains difficult, necessitating optimization of CAR designs and delivery systems. To address these limitations, strategies, such as logic-gated CARs, suicide genes (iCasp9), and small-molecule switch systems, are under investigation [211, 212].

Local administration, such as direct injection into tumors or resection cavities, has demonstrated promise in addressing these challenges [213]. Recent advancements in biomaterials and nanotechnology have introduced innovative strategies to enhance CAR-T cell delivery, expansion, and efficacy, supported by encouraging preclinical and early clinical data. Biomaterial scaffolds, designed to mimic lymphoid architecture, offer a particularly effective platform for enhancing CAR-T cell function at the tumor site. Constructed from biocompatible polymers, such as poly (lactic-co-glycolic acid), these scaffolds are functionalized with stimulatory antibodies, such as anti-CD3 and anti-CD28, and cytokines, such as IL-7 and IL-15, to promote localized CAR-T cell proliferation and activation. A recent study demonstrated a 50-fold expansion of CAR-T cells in vitro and a 15-fold expansion in vivo within a cervical tumor model, with scaffold-mediated activity sustained for up to 30 days and associated with minimal toxicity. Notably, this strategy outperformed conventional systemic CAR-T cell administration in suppressing tumor growth [214]. In parallel, nanotechnology offers versatile tools for the in vivo generation and modulation of CAR-T cells, particularly via mRNA-LNPs [215]. Beyond direct cellular engineering, nanoparticles can modulate the TME by selectively targeting immunosuppressive components, such as tumor-associated macrophages and MDSCs. In one approach, nanoparticles loaded with all-trans retinoic acid have been shown to reduce MDSC-mediated immunosuppression, enhancing CAR-T cell efficacy [216]. Furthermore, for tumors located behind physiological barriers, such as the blood–brain barrier (BBB), nanoparticles modified with targeting ligands, such as transferrin or TAT peptides, have facilitated improved CAR-T cell trafficking, as demonstrated in preclinical brain tumor models [216]. Recent clinical trials emphasize the potential of advanced delivery systems. The phase I trial (NCT04196413) evaluating GD2-targeted CAR-T cells for H3K27M-mutated DMG, administered intravenously and intraventricularly, reported major tumor reductions in 4 of 11 patients, including one complete response lasting over 30 months. No dose-limiting toxicities were observed at the lower dose [217]. In gastrointestinal cancers, satri-cel (CT041), targeting CLDN18.2, achieved a 38.8% ORR and a 91.8% DCR among 98 patients in the phase I trial CT041-CG4006 (NCT03874897), with no treatment-related deaths [53]. In the phase I/II trial CT041-ST-01 (NCT04581473) for gastric and gastroesophageal junction cancers, treatment was associated with significant improvements in progression-free survival, leading to FDA Regenerative Medicine Advanced Therapy designation in 2022 [51]. In metastatic colorectal cancer, GCC19CART (NCT04981119) achieved an 80% ORR at dose level 2, including one pathological complete response. In addition, in a phase 1 trial (NCT03089203) for prostate cancer, a PSMA-targeted, TGF-β-insensitive armored CAR-T construct demonstrated sustained persistence, with CAR-T cells detectable in peripheral blood beyond 200 days in some patients and metastatic biopsy samples in 7 of 9 cases. Treatment induced prostate-specific antigen declines of ≥ 30% in 4 out of 13 infused patients (approximately 31%) were detected. However, the safety profile included CRS in 5 of 13 patients—mostly grade 2–3 events that were reversible with treatment—and one treatment-related death due to grade 4 CRS complications [218]. These trials emphasize the transformative potential of targeted CAR-T delivery in solid tumors.

Artificial intelligence (AI) is revolutionizing CAR-T cell therapy for solid tumors by facilitating the discovery of novel antigenic targets. Leveraging machine learning and deep learning algorithms, AI enables the analysis of extensive genomic and proteomic datasets to identify tumor-specific antigens and neoantigens suitable for CAR-T cell targeting. This approach addresses key challenges, such as antigenic heterogeneity and off-tumor toxicity, by prioritizing targets with high tumor specificity, improving the safety and efficacy of CAR-T therapies in solid malignancies. For instance, neural networks have been used to screen a library of over 2,300 synthetic costimulatory domains, identifying optimal candidates for neoantigen targeting in solid tumor models [219]. Deep learning architectures, such as RCMNet, have achieved a top-1 accuracy of 99.63% in identifying CAR-T cells, supporting the development of personalized treatment strategies [220]. AI-driven platforms, such as CAR-Toner, further refine CAR-T functionality by optimizing antigen-binding specificity and affinity [221]. In parallel, AI-driven CRISPR screens have identified genetic targets to enhance T cell persistence and function in hostile TME, such as those in pancreatic cancer [222]. Although clinical trials directly using AI-selected antigens are emerging, recent studies reflect AI’s indirect yet significant impact. A phase I trial targeting B7-H3 in DMG reported a median OS of 19.8 months—notably exceeding historical controls of 11.2 months [7]. Similarly, GD2-targeted CAR-T cells demonstrated a 63% ORR in neuroblastoma (NCT03373097) [223]. These promising outcomes, supported by AI-enabled antigen discovery, emphasize the potential for broader clinical translation [223]. Ongoing trials, such as those targeting NKG2DL (NCT04550663) in multiple solid tumor types, further emphasize continued progress in antigen-specific CAR-T cell therapies [224].

Recent clinical trials emphasize the potential of precise genomic engineering in advancing CAR-T cell therapy for solid tumors. A phase I/II trial (NCT02414269) combining anti-mesothelin CAR-T cells with pembrolizumab in patients with malignant pleural mesothelioma reported an 83% 1-year OS rate, with 8 of 18 patients achieving stable disease lasting > 6 months and 2 experiencing complete metabolic responses [135]. In addition, a phase I trial (NCT04510051) evaluating IL13Rα2-targeted CAR-T cells engineered for enhanced genomic stability in pediatric brain tumors demonstrated improved persistence within the CNS TME, accompanied by early signs of tumor control [208] (Table 6).

Table 6.

Recent clinical trials of CAR-T cell therapy for solid tumors

Study	Target	Tumor type	Key findings	Status
Mackall et al., 2024	GD2	DMGs	Tumor reduction in 7 of 9 patients; median OS ~ 2 years; one complete response	Completed [217]
Brown et al., 2024	B7-H3	DIPG	Median survival 19.8 months; 15 of 18 patients with stable disease; one partial response	Completed (specific to Arm C for DIPG; other arms ongoing) [225]
Qi et al., 2023	CLDN18.2	Gastric cancer	ORR 57.1%; DCR 75%; 6-month OS 81.2%	Active, not recruiting [52]
Zhang et al., 2023	CD133	Pancreatic cancer	1 PR, 3 SD, 2 PD in 7 patients	Completed [226]

Memory CAR-T cells: The next frontier

Recent advances in CAR-T cell therapy have emphasized the enhancement of memory-T-cell functionality, which is essential for long-term immune surveillance and durable antitumor responses. Memory T cells, a distinct subset that persists following antigen exposure, possess the capacity to elicit potent immune responses rapidly on subsequent antigen encounters. In CAR-T therapy, the presence of memory T cells is particularly critical for sustaining antitumor immunity, especially in solid tumors where ongoing immune vigilance is necessary to prevent disease recurrence. Among these, T_CM) and T_SCM CAR-T cells have garnered attention due to their superior self-renewal capacity, extended persistence, and sustained effector function compared to terminally differentiated T-cell subsets [145, 227–229]. Preclinical and clinical studies have demonstrated that memory-enriched CAR-T cell populations are capable of long-term persistence and robust recall responses upon reexposure to tumor antigens, establishing a continuous immunological defense against tumor relapse [230–232]. In preclinical models, memory CAR-T cells have demonstrated superior antitumor efficacy in solid tumors [6, 118]. Their enhanced proliferative capacity and persistence within the host contribute to sustained tumor eradication and improved adaptability within the complex TME, which frequently impairs effector T cell function. Notably, memory CAR-T cells maintain cytokine production and cytotoxic activity even under immunosuppressive conditions that lead to rapid dysfunction of conventional CAR-T cells [233–235]. This functional resilience is particularly advantageous in solid tumors, where antigen expression is often heterogeneous and dynamic, influenced by tumor evolution and therapy-induced selective pressures. Furthermore, generation of memory CAR-T cells can be significantly improved through optimization of ex vivo culture conditions, enhancing their therapeutic potential.

Strategies to generate memory-enriched CAR-T cells have focused on optimizing ex vivo culture protocols. Strategies such as modulating activation strength, fine-tuning intracellular signaling pathways, and supplementing cultures with cytokines like IL-7 and IL-15, have preserved memory-like phenotypes, enhancing the longevity and functional fitness of the final CAR-T cell product [232, 236]. These approaches promote sustained persistence and mitigate T-cell exhaustion, enhancing therapeutic durability in preclinical models of solid tumors. Recent innovations have further advanced these strategies through genetic modifications. For instance, engineering CAR-T cells to secrete IL-15 or express membrane-bound IL-12 enhances memory-like phenotypes and boosts antitumor activity in ovarian and lung cancer models [237, 238]. Similarly, inducible IL-18 expression has demonstrated robust tumor control and prolonged survival in pancreatic and SCLC models targeting DLL3. Reinforcing these preclinical observations, recent clinical advances highlight the promise of IL-15–secreting CAR-T cells, which have demonstrated potential in countering TME-mediated immunosuppression in lung cancer models and sustaining prolonged T-cell activation (NCT05620342) [239]. In addition, CRISPR-Cas9 gene editing has been used to delete inhibitory regulators, such as TET2, RASA2, and REGNASE-1, or to overexpress transcription factors like c-Jun, improving CAR-T cell persistence and effector function [240]. Moreover, genetic modifications targeting FOXO1 have been used to boost stemness and metabolic fitness, while metabolic interventions modulating proline metabolism further optimize CAR-T cell function in the TME [241–244]. Moreover, incorporation of costimulatory domains, such as 4-1BB, promotes differentiation toward T_CM, while logic-gated CARs improve specificity by requiring multiple antigenic signals for activation, reducing off-target effects [121, 245]. A particularly transformative strategy involves combining CAR-T cell therapy with mRNA vaccines to promote memory T cell phenotypes, leveraging epitope spreading to elicit broader immune responses and improve disease control in refractory solid tumors [6]. These advancements, including cytokine optimization and mRNA vaccine integration, enhance CAR-T cell persistence in immunosuppressive environments, offering a promising avenue to overcome key barriers in solid tumor treatment [6, 246]. Collectively, these strategies are aimed at addressing the challenge posed by the immunosuppressive TME and antigen heterogeneity, and enhancing the durability and efficacy of CAR-T cells in solid tumors. However, the extended persistence of memory-enriched CAR-T cells raises important safety concerns. Prolonged survival increases the risk of off-tumor toxicities, particularly in healthy tissues expressing low levels of the targeted antigen. To mitigate these risks, novel safety switches and regulatable CAR expression systems are under active development, enabling clinicians to modulate CAR-T cell activity post-infusion [247–249]. These safety mechanisms are critical to preserving the therapeutic index of memory-enriched CAR-T cells, balancing potent antitumor efficacy with patient safety (Table 7).

Table 7.

Recent advances in memory CAR-T cell therapy for solid tumors: Innovations and challenges

Strategy	Description	Advantages	Challenges	References
Naïve/stem memory T cell engineering	Using naive/stem memory T cells (TN/SCM) for CAR-T production to enhance persistence and reduce exhaustion	Improves antitumor activity and reduces side effects like cytokine release syndrome	Requires specialized manufacturing processes to isolate and expand TN/SCM cells	[232]
FOXO1 modulation	Genetic modification to regulate FOXO1 expression for enhanced memory programming	Boosts stemness, metabolic fitness, and overall efficacy of CAR-T cells	Balancing FOXO1 levels is critical; efficacy may be context-dependent	[241, 242]
Metabolic interventions	Modulating metabolic pathways (proline metabolism) to enhance CAR-T cell function in the TME	Improves immunomodulatory effects and mitochondrial capacity, enhancing efficacy	Complex metabolic interactions; potential off-target effects on healthy cells	[243, 244]
4-1BB costimulatory domain	Incorporating 4-1BB in CAR design to promote differentiation towards TCM	Enhances long-term persistence and self-renewal capacity of CAR-T cells	May have different efficacy profiles compared to CD28; requires optimization	[245]
Logic-gated CAR-T Cells	Dual-targeting CAR-T cells with AND-gate logic to ensure activation only when multiple tumor antigens are present	Reduces off-target toxicity and addresses antigen heterogeneity in solid tumors	Complex design and manufacturing; efficacy may be reduced if antigens are not co-expressed	[121]
Combination with vaccines	Combining CAR-T therapy with mRNA vaccines promotes CAR-T cell maturation into a memory phenotype	Enhances disease control in refractory tumors; leverages epitope spreading for broader immune response	Requires careful monitoring for safety; potential for increased toxicity	[6]
Clinical trials (memory-enriched CAR-T): NCT04510051	Trials using memory-enriched CAR-T cells for solid tumors, such as pediatric brain tumors (City of Hope trials)	Improved persistence and efficacy in challenging environments like brain tumors; potential for long-term remissions	Safety concerns with long-term persistence; need for specialized delivery methods (intraventricular)	[208]

Clinical translation of memory-enriched CAR-T cells is actively progressing, with several early phase clinical trials yielding promising preliminary results. Notable ongoing studies include the following (Table 8). Across studies, three cross‑cutting patterns are notable. First, locoregional CNS delivery (e.g., intraventricular administration) is repeatedly used to improve persistence and disease control in pediatric brain tumors and gliomas [7, 143, 208]. Second, platforms that promote TSCM/TCM differentiation or incorporate 4‑1BB costimulation are prioritized to extend durability [144, 250, 251]. Third, vaccine‑augmented approaches (e.g., CLDN6 CAR‑T plus an amplifying mRNA vaccine) are being tested to amplify in‑vivo expansion without changing the safety profile [252]. Additional CNS programs using intraventricular HER2 CAR‑T report safety and CSF persistence consistent with a memory‑skewed phenotype [253]. Collectively, these programs aim to convert the biological advantages of memory subsets into clinically relevant persistence and function under solid‑tumor TME pressure.

Table 8.

Clinical trials evaluating memory CAR-T cell therapy for solid tumors (as of May 2025)

Trial name/ID	Target	Phase/Route/Design cue	Status	Key findings (concise)	References
City of Hope IRB #19,130	IL13Rα2	Phase I; route not specified in text; memory-enriched product	Ongoing	Improved persistence; early signs of efficacy in a CNS setting	[143]
NCT04510051	IL13Rα2	Phase I; ICV; pediatric CNS	Ongoing	Preliminary evidence of re-awakening endogenous T-cell immunity; CNS-localized activity	[208]
BNT211-01 (Phase I/II mRNA vaccine trial)	CLDN6	Phase I/II; (route not specified in text); i.v. CAR-T + amplifying mRNA vaccine	Ongoing	Encouraging disease control with no new safety liabilities	[252]
NCT02765243	GD2	Phase I; route not specified in text; 4th-gen/4-1BB	Completed	Median OS 25 months; minimal toxicities; supports memory differentiation	[258]
NCT03696030	HER2	Phase I; ICV; recurrent brain or leptomeningeal metastases	Ongoing	Safety and CSF persistence; memory-enriched populations show enhanced functionality	[253]

Consistent with these principles, our preclinical studies in MM provide empirically grounded strategies with translational relevance. We engineered anti-BCMA CAR-modified marrow-infiltrating lymphocytes (CAR-MILs) featuring dual-epitope binding, attaining a transduction efficiency of 34% and > 100-fold cellular expansion over 21 days. These constructs were enriched for CD8⁺ T_CM (80%) and expressed low levels of exhaustion markers, such as TIGIT, TIM3, and CD73. Functionally, these CAR-MILs exhibited potent cytotoxicity against MM cell lines (RPMI8226 and U266), as well as autologous CD138⁺ primary cells [254, 255]. In vivo, BCMA CAR-MILs significantly suppressed tumor progression and prolonged survival in myeloma-bearing mice. This antitumor activity was associated with robust infiltration of CD8⁺ BCMA CAR-T cells at tumor sites and retention of memory phenotypes [255]. This evidence emphasizes the critical role of memory-enriched T-cell subsets in achieving sustained antitumor effects, offering a blueprint for countering antigen heterogeneity and TME suppression in solid tumors. Complementary investigations into non-engineered immune cells, such as dendritic cells (DCs) and natural killer (NK) cells, further emphasize the potential to augment immune activation and cytotoxicity in immunosuppressive milieus across MM and solid tumors [256, 257]. For instance, DC-based vaccines have demonstrated the capacity to potentiate endogenous T-cell responses, potentially synergizing with CAR-T therapy to dismantle TME barriers [36]. Likewise, NK cells—owing to their capacity for antigen-independent cytotoxicity—can target antigen-low tumor variants, enhancing the breadth and efficacy of CAR-T cell therapies in heterogeneous malignancies [33]. These findings support an integrative therapeutic paradigm that combines CAR-T cells with DC and NK-based therapies to enhance overall immune orchestration. Drawing on our MM studies, we proposed a tripartite framework for advancing CAR-T efficacy in solid tumors: prioritizing memory-enriched T-cell subsets to ensure long-term persistence, implementation of multi-antigen targeting strategies to mitigate antigen escape, and incorporation of combinatorial regimens with non-engineered immune cells to remodel the immunosuppressive TME. This approach, substantiated by rigorous preclinical studies, equips CAR-T therapy to advance to the forefront of next-generation cancer immunotherapy, where adaptive and innate immune arms converge to dismantle the most entrenched tumor defenses. However, the application of these strategies in solid tumors remains constrained by the highly suppressive nature of the TME, as emphasized in recent 2025 reviews [34].

Collectively, the strategies outlined in this section emphasize the rapid and multifaceted advancements in CAR-T cell therapy for solid tumors. Approaches, such as multi-antigen targeting, modulation of the TME, incorporation of safety features, and engineering for memory phenotypes, address key therapeutic barriers, paving the way for more durable and effective interventions. The integration of these innovations into personalized combination therapies tailored to individual tumor profiles is anticipated to enhance therapeutic precision. Moreover, emerging technologies, including AI for optimizing CAR design and patient-derived organoid platforms for preclinical testing, are accelerating progress by refining preclinical modeling and enabling personalized treatment refinement [104]. As research advances through 2025 and beyond, memory-enriched CAR-T cells and other innovative strategies hold significant promise for transforming the treatment landscape of solid tumors, and offering durable remissions and improved clinical outcomes for patients with otherwise refractory malignancies.

Translational and regulatory considerations

Realizing the full potential of advanced CAR-T cell strategies requires addressing significant manufacturing and regulatory challenges. The engineering of dual-receptor T cells or logic-gated CARs necessitates scalable and cost-effective production methods. For instance, generating CAR-T cells with multiple genetic modifications involves advanced gene-editing technologies, such as CRISPR-Cas9, and optimized culture conditions—both of which are resource-intensive and technically demanding [259]. To overcome these barriers, recent innovations in automated CAR-T manufacturing platforms have shown promise in streamlining production, lowering costs, and improving accessibility [259]. Furthermore, the distinctive safety and efficacy profiles of next-generation CAR-T therapies necessitate the evolution of regulatory frameworks originally designed for simpler biologics. Current guidelines may be insufficient to address fully the intricacies of living cell therapies, emphasizing the need for tailored regulatory adaptations that prioritize long-term safety surveillance and patient-specific protocols [260]. Regulatory agencies, such as the FDA, are actively working to develop new pathways and guidance to accommodate these emerging technologies, as reflected in recent updates for cellular and gene therapies [260]. Collaborative efforts among academic institutions, biotechnology companies, clinicians, and regulatory authorities are essential for navigating these challenges. Such partnerships have been instrumental in the success of CAR-T therapies for hematologic malignancies and are now pivotal for advancing preclinical findings into clinical practice for solid tumors [261]. As these multidisciplinary initiatives continue to mature, they are poised to deliver safe, effective, and accessible CAR-T treatments for patients with solid malignancies.

Conclusion

The unprecedented success of CAR-T cell therapy in hematologic malignancies has emphasized its transformative potential in cancer treatment. However, extending this efficacy to solid tumors remains a formidable challenge due to the complex TME, antigen heterogeneity, and the elevated risk of off-tumor toxicity. To overcome these barriers, a new generation of CAR-T strategies is under active development. Multi-antigen targeting platforms, such as dual and tandem CAR constructs, enhance specificity and mitigate immune escape. Armored CAR-T cells, engineered to secrete immunostimulatory cytokines and used in combination with immune checkpoint inhibitors, offer a means to remodel the immunosuppressive tumor milieu. Innovations in T-cell trafficking, such as chemokine receptor engineering and regional delivery techniques, are enhancing tumor infiltration. Logic-gated CAR constructs and inducible suicide switches are being incorporated to enhance safety profiles. Integrative approaches combining CAR-T therapy with adjunctive modalities, such as oncolytic viruses or radiotherapy, exhibit synergistic potential to augment antitumor efficacy. Preliminary preclinical and clinical data support the biological plausibility and clinical feasibility of these innovations, with preliminary findings from trials, such as GD2/B7-H3, emphasizing the practicality of overcoming key translational barriers (Table 4).

Despite persistent challenges in manufacturing, scalability, and regulatory oversight, the pace of progress in CAR-T research for solid tumors is accelerating rapidly. Breakthroughs in precise genomic integration, such as CRISPR-Cas9 and non-viral delivery methods, offer a promising direction for the field by enabling targeted and stable CAR expression with enhanced safety profiles. In parallel, transient CAR expression via mRNA electroporation provides a safe alternative for short-term applications. The emerging field of in vivo CAR-T cell generation has progressed significantly. A 2025 study demonstrated effective tumor control in humanized mouse models and immune modulation in non-human primates using tLNPs [132]. Moreover, ongoing clinical trials investigating mRNA-based in vivo CAR-T for pancreatic cancer hold promise for simplifying manufacturing, reducing costs, and enhancing therapeutic outcomes (NCT04404595) [52].

To advance CAR-T therapy in solid tumors, key priorities include the identification of tumor-specific antigens to mitigate off-tumor toxicity. Computational platforms, such as IRIS, which leverage alternative splicing, show substantial promise [262]. Enhancing CAR-T cell persistence and function in the TME is critical and is being pursued through cytokine-armored constructs that secrete IL-12 or protease-regulated systems, such as VIPER CARs, equipped with safety switches, both of which have demonstrated efficacy in preclinical models [263, 264]. Addressing tumor heterogeneity requires multi-antigen strategies, including dual-target or tandem CARs, to target diverse malignant cell populations [113]. The development of allogeneic CAR-T constructs may further expand access by minimizing the logistical and manufacturing challenges of autologous cell therapy [265]. Finally, combination therapies, such as CAR-T cells co-administered with immune checkpoint inhibitors (anti-PD-1), offer the potential to enhance therapeutic efficacy by counteracting TME-induced immunosuppression [136]. Clinical trials targeting antigens, such as EGFR, HER2, and mesothelin, have demonstrated partial responses with manageable toxicity profiles, supporting their clinical feasibility [113, 265]. Moving forward, sustained investment in preclinical models that recapitulate the complexity of solid tumors and clinical trials evaluating these innovative strategies will be pivotal for their successful translation into clinical practice.

Sustained investment in physiologically relevant preclinical models and well-designed clinical trials will be pivotal in translating these advancements into clinical practice. Through interdisciplinary innovation, personalized cellular engineering, and clinical optimization, effective CAR-T therapies for solid tumors are rapidly transitioning from aspirational concepts to imminent reality. Building on the successes achieved in hematologic malignancies, this alignment of biological insight and technological advancement signals a new era in oncology, where even treatment-refractory solid tumors, such as lung, breast, and pancreatic cancers, may become amenable to durable immunotherapeutic control.

Abbreviations

AI

Artificial intelligence

4-1BB

4-1BB (CD137, a costimulatory molecule)

AND-gate logic

A genetic-engineering design requiring recognition of multiple antigens for activation

B7-H3

B7 homolog 3 (also known as CD276)

B-ALL

B-cell precursor acute lymphoblastic leukemia

BCMA

B-cell maturation antigen

BiTE

Bispecific T-cell engager

CAR-T

Chimeric antigen receptor T cell

CD22

Cluster of differentiation 22

CD70

Cluster of differentiation 70

CEA

Carcinoembryonic antigen

CI

Confidence interval

CLDN-6

Claudin-6

CNS

Central nervous system

CR

Complete remission

CRS

Cytokine release syndrome

CXCR2

C-X-C motif chemokine receptor 2

DLL3

Delta-like ligand 3

DIPG

Diffuse intrinsic pontine glioma

DCR

Disease control rate

DMG

Diffuse midline glioma

DOR

Duration of response

ECM

Extracellular matrix

EGFR

Epidermal growth factor receptor

EGFRvIII

Epidermal growth factor receptor variant III

FDA

Food and drug administration

FOXO1

Forkhead box O1

GD2

Ganglioside GD2

HER2

Human epidermal growth factor receptor 2

ICANS

Immune effector cell-associated neurotoxicity syndrome

iCasp9

Inducible caspase 9

IL

Interleukin

IL-13Rα2

Interleukin-13 receptor alpha 2

IRB

Institutional review board

ICV

Intraventricular

LNP

Lipid nanoparticle

MM

Multiple myeloma

MCL

Mantle cell lymphoma

MDSCs

Myeloid-derived suppressor cells

mRNA

Messenger ribonucleic acid

MUC1

Mucin 1

OR

Overall response

ORR

Overall response rate

OS

Overall survival

PD-1

Programmed cell death protein 1

PD-L1

Programmed death-ligand 1

PSCA

Prostate stem cell antigen

PSMA

Prostate-specific membrane antigen

R/R LBCL

Relapsed or refractory large B-cell lymphoma

R/R SCLC

Relapsed or refractory small cell lung cancer

RR

Relapsed or refractory

SynNotch

Synthetic notch receptor

TAA

Tumor-associated antigen

TGF-β

Transforming growth factor-beta

T_CM

Central memory T cells

TILs

Tumor-infiltrating lymphocytes

tLNP

Targeted LNP

TME

Tumor microenvironment

TN/SCM

Naïve/stem cell memory T cells

T_SCM

T memory stem cell

TCRs

Transgenic T cell receptors

Author contributions

MCV, JJL, SKK, and SHJ conceptualized and supervised the work. MCV, VDHT, NR, DTT, and VTN collated the literature, wrote the manuscript, and created the figures. All authors have read and approved the final manuscript.

Funding

This research was supported by the Korea Drug Development Fund, funded by the Ministry of Science and ICT, the Ministry of Trade, Industry, and Energy, and the Ministry of Health and Welfare (Grant No. RS-2024-00341192, Republic of Korea). Additional support was provided by the National Research Foundation of Korea (NRF) through a grant funded by the Ministry of Science and ICT (Grant No. RS-2023-00207920). Furthermore, this work was supported by the Commercialization Promotion Agency for R&D Outcomes (COMPA), funded by the Ministry of Science and ICT (Grant No. RS-2025-25430866).

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

The manuscript has not been previously published and is not under consideration elsewhere. All authors have reviewed the final version and consent to publication.

Use of AI

No artificial intelligence tools were used in the preparation or editing of this manuscript.

Competing interests

The authors declare no competing interests.