Introduction
The remarkable therapeutic efficacy of chimeric antigen receptor (CAR) T-cell therapy, together with the substantial logistical and clinical challenges it presents, underscores the need to broaden the spectrum of cellular immunotherapies [1]. Recent developments have expanded adoptive cell therapy beyond CAR-T cells to include regulatory T cells (Tregs), natural killer (NK) cells, macrophages (M cells), γδ T cells, and other innate or innate-like lymphocyte subsets [2]. These emerging platforms are now being clinically evaluated, with meaningful implications for improving patient survival and moving closer to curative outcomes in both hematologic malignancies and solid tumors [3].
In particular, leveraging non-T-cell sources may help overcome several intrinsic limitations of autologous CAR-T products, such as manufacturing delays, high cost, patient-specific variability, and life-threatening toxicities, including cytokine release syndrome (CRS) and neurotoxicity [2, 4].
Therefore, exploring alternative cell sources and engineering strategies represents a promising avenue for overcoming these barriers and expanding the therapeutic horizon of cell-based immunotherapy [5]. Among these emerging modalities, CAR-engineered NK cells (CAR-NK) have rapidly advanced toward clinical translation as an “off-the-shelf” platform owing to their intrinsic safety, minimal risk of graft-versus-host disease (GvHD), and capacity to target malignant cells through both CAR-dependent and CAR-independent mechanisms [6].
At the same time, engineered αβ T-cell receptors (TCR-T), γδ T-cell platforms, macrophage-derived CAR therapies (CAR-M) and natural killer T (CAR-NKT) offer complementary advantages by enabling the targeting of intracellular antigens, recognizing stress-induced ligands, and remodeling the tumor microenvironment, features particularly relevant for solid tumors where CAR-T therapies have struggled [7].
Collectively, these next-generation immune effector cells reflect a paradigm shift from single-platform T-cell approaches toward a diversified cellular immunotherapy landscape that integrates innate, adaptive, and myeloid-based strategies [8]. As engineering technologies, such as genome editing, synthetic biology circuits, and iPSC-derived universal cell sources continue to advance, the field is rapidly moving toward scalable, customizable, and more predictable cellular therapies [9, 10].
This review examines current advances in five major platforms, including CAR-NK cells, TCR-engineered αβ T cells, γδ T cells, CAR macrophages, and CAR-NKT cells, with an emphasis on biological foundations, engineering innovations, clinical development, and remaining challenges. Together, these modalities illuminate a rapidly expanding frontier in cancer immunotherapy beyond CAR-T cells.
CAR-NK cells
NK cell biology
NK cells are cytotoxic lymphocytes of the innate immune system that play a central role in host defense by eliminating infected, foreign, stressed, or pathologically altered cells [11]. Beyond their direct cytolytic function, NK cells serve as critical regulators of antitumor immunity through dynamic cell–cell interactions and the secretion of immunomodulatory cytokines and chemokines, thereby shaping both innate and adaptive immune responses [12]. In contrast to T cells, NK cells do not express rearranged, antigen-specific receptors such as TCRs somatically [13].
Instead, NK-cell recognition is governed by the integration of signals derived from adhesion molecules, activating receptors, and inhibitory receptors, which collectively determine the target cell’s susceptibility.
NK cells compensate for the absence of antigen-specific receptors through two principal strategies. First, they express high levels of adhesion molecules such as neural cell adhesion molecule (NCAM/CD56), which facilitate stable immunological synapse formation with target and accessory cells [14]. Second, NK cells rely on a diverse repertoire of germline-encoded activating and inhibitory receptors that recognize stress-induced ligands, antibody-coated targets, and self–non-self cues, enabling discrimination between healthy and transformed tissues [15, 16].
Target cell elimination by NK cells is executed through the delivery of a cytotoxic “death package.” This involves the polarized release of lytic granules containing perforin and granzymes or the engagement of death receptor pathways, including Fas ligand (FasL)/Fas interactions, culminating in apoptotic cell death [17–20]. In parallel, NK cells secrete cytokines and chemokines such as interferon-γ (IFN-γ) and RANTES (CCL5), which recruit and activate dendritic cells (DCs), T cells, and additional innate immune effectors, thereby amplifying antitumor immune cascades [21, 22].
Historically, the equilibrium phase of cancer immunoediting was primarily attributed to adaptive immune responses, particularly T-cell–derived cytokines such as IFN-γ, IL-12, and IL-23 [23]. However, accumulating evidence indicates that NK cells also contribute substantially to this phase by exerting cytostatic effects on disseminated tumor cells (DTCs). Through sustained IFN-γ signaling, NK cells can induce cell-cycle arrest and suppress tumor cell proliferation, thereby maintaining long-term tumor dormancy rather than immediate elimination [24].
These findings underscore the dual capacity of NK cells to either eradicate malignant cells or enforce clinically favorable states of immune-mediated tumor control.
Importantly, NK-cell cytotoxicity is inherently modular, encompassing both CAR-dependent and CAR-independent mechanisms, as schematically illustrated in Fig. 1. Even in the absence of CAR engagement, NK cells retain intrinsic effector functions, including antibody-dependent cellular cytotoxicity (ADCC) mediated by FcγRIIIa (CD16) [25], activation through stress-sensing receptors such as NKG2D in response to MICA/B expression [26], and apoptosis induction via death receptor pathways (Fas/FasL and TRAIL/TRAIL-R) [27].
CAR engineering therefore augments, rather than replaces, this multifaceted cytotoxic repertoire, enabling CAR-NK cells to integrate synthetic antigen recognition with native NK effector programs.
CAR design and genetic engineering in NK cells
The generation of CAR-NK cells requires the rational design of chimeric antigen receptors that are compatible with NK-cell signaling architecture and functional biology. Canonical CAR constructs comprise an extracellular antigen-recognition domain, typically a single-chain variable fragment (scFv), linked via a hinge and transmembrane domain to one or more intracellular signaling and co-stimulatory modules [28]. The antigen-binding domain is most commonly derived from murine monoclonal antibodies [29, 30], although increasing attention is being directed toward ligand-based or receptor-based designs that exploit natural NK–tumor interactions to reduce immunogenicity and enhance physiological signal transduction [31].
CAR designs are conventionally classified into generations based on their intracellular signaling domains. First-generation CARs contain a single activation domain, most often CD3ζ, whereas second- and third-generation CARs incorporate one or more co-stimulatory motifs such as CD28 or 4-1BB to enhance expansion, persistence, and effector function [32–34]. However, direct translation of T-cell–optimized CAR architectures into NK cells has revealed limitations, including tonic signaling and suboptimal functional fitness [35].
To address these challenges, CAR-NK constructs are increasingly tailored to exploit native NK signaling pathways. Intracellular domains such as CD3ζ and DNAX-activating protein of 12 kDa (DAP12) are frequently combined with NK-specific co-stimulatory receptors, including 2B4 (CD244) and DNAM-1 (CD226), to potentiate cytotoxic granule release and cytokine secretion [13, 36]. This “NK-centric” design philosophy aims to avoid excessive tonic signaling observed with some T-cell–optimized CAR backbones and to maintain a more physiological activation threshold in NK cells [37, 38].
Functionally, these design choices directly map onto the killing mechanisms depicted in Fig. 1. DAP12-based CAR signaling preferentially activates SYK/ZAP70-dependent pathways, enhancing perforin–granzyme polarization and serial killing capacity [39, 40], while 2B4 co-stimulation supports IFN-γ production and sustained activation without exhausting the cell [41].
Consequently, CAR-NK cells maintain both CAR-triggered cytotoxicity and CAR-independent effector functions, including ADCC and death receptor–mediated apoptosis.
Efficient genetic modification of NK cells remains a technical challenge. Gene transfer is typically achieved using retroviral or lentiviral vectors [42, 43], although non-viral approaches such as electroporation of plasmid DNA or mRNA are increasingly explored to enable transient CAR expression or reduce genomic integration risks [44–46]. Following gene transfer, CAR-NK products undergo rigorous functional validation, including assessment of CAR expression, antigen-specific cytotoxicity, cytokine secretion (e.g., IFN-γ, TNF-α), and specificity against antigen-negative controls [31, 47].
Standardization of these assays is essential for cross-trial comparability and correlation of in vitro potency with clinical outcomes. Taken together, these receptor-level and gene-transfer strategies illustrate that CAR-NK development is not limited to antigen targeting alone, but instead relies on integrated engineering of signaling architecture, delivery platform, and functional output.
Advanced engineering strategies in CAR-NK cells
Beyond CAR design itself, next-generation CAR-NK development increasingly relies on multilayered engineering strategies aimed at improving persistence, resistance to exhaustion, and manufacturability. These strategies include cytokine armoring, targeted genome editing, and the development of standardized allogeneic platforms such as induced pluripotent stem cell (iPSC)-derived NK products. Framing CAR-NK therapy through these engineering layers more accurately reflects the translational direction of the field, in which efficacy is shaped not only by antigen recognition, but also by the fitness, durability, and scalability of the engineered cell product.
Cytokines such as IL-15 and IL-21 are integral to NK-cell biology and are widely used to support ex vivo expansion and in vivo persistence of CAR-NK cells. IL-15 is a central regulator of NK-cell development, survival, proliferation, and effector differentiation and has been widely shown to enhance CAR-NK antitumor activity [48]. Incorporation of IL-15 or IL-15/IL-15Rα complexes into CAR-NK constructs activates PI3K/AKT and JAK/STAT signaling pathways, upregulates genes associated with survival and function (e.g., FGFR2, FLT3), and reduces inhibitory checkpoint expression, including PD-1 and LAG-3, thereby conferring resistance to functional exhaustion [49].
In this context, IL-15 should be viewed not merely as a supportive cytokine, but as a cytokine-armoring strategy that functionally reprograms CAR-NK cells toward enhanced persistence and sustained antitumor activity.
IL-21 exerts overlapping yet distinct effects on NK cells. Constitutive IL-21 expression in NKG2D-CAR-NK-92 cells enhances cytotoxicity, IFN-γ production, and tumor control in xenograft models through PI3K/AKT pathway activation [50]. These effects are associated with improved proliferation, reduced apoptosis, and mitigation of functional exhaustion, supporting the use of IL-21 as a complementary cytokine to enhance CAR-NK efficacy, particularly in solid tumors.
Together, IL-15- and IL-21-based approaches exemplify how cytokine engineering can be used to reinforce CAR-NK persistence and function under hostile tumor microenvironmental conditions.
Mechanistically, CAR activation through CD3ζ has been shown to engage the PKA–CREB–CREM axis, inducing cyclic AMP response element modulator (CREM), a transcriptional repressor that suppresses effector genes such as PRF1 and GZMB [51]. Genetic deletion of CREM, particularly in the context of IL-15 stimulation, restores full cytotoxic competence and unleashes the antitumor potential of CAR-NK cells [51]. This finding highlights an additional engineering layer in CAR-NK therapy, namely genome-guided functional optimization through disruption of intrinsic negative regulators.
Genome editing technologies have substantially expanded the engineering landscape of CAR-NK therapies by enabling precise modification of endogenous genes that regulate NK-cell activation, persistence, and immune recognition [52]. In particular, CRISPR–Cas9–based strategies are increasingly used to disrupt inhibitory signaling pathways and enhance NK-cell functional fitness [53]. Among these targets, cytokine-inducible SH2-containing protein (CISH) has emerged as a key intracellular checkpoint regulating IL-15 signaling, and its genetic deletion has been shown to enhance NK-cell proliferation, metabolic fitness, and antitumor activity in preclinical models [54].
Consistent with these findings, additional studies show that targeting CISH lowers the activation threshold of NK cells and amplifies signaling through natural cytotoxicity receptors, thereby strengthening antitumor responses [55].
Beyond intrinsic cytotoxicity, cytokine-supported CAR-NK cells exert indirect immunomodulatory effects, as highlighted in Fig. 1. Elevated secretion of IFN-γ and chemokines such as RANTES promotes recruitment and activation of dendritic cells and endogenous T cells, thereby enhancing antigen presentation and amplifying adaptive immune responses [56, 57].
This immune orchestration may extend antitumor activity beyond the persistence window of the infused CAR-NK product.
Another major engineering advance in this field is the emergence of iPSC-derived CAR-NK platforms, which provide a renewable, standardized, and clonally defined cellular source that is highly suitable for multiplex genetic engineering [58]. In contrast to donor-derived primary NK cells, these platforms allow the controlled incorporation of CAR constructs together with additional functional modules such as high-affinity non-cleavable CD16 (hnCD16), IL-15/IL-15 receptor fusion support, and selected gene-editing strategies [59]. This is particularly important because limited expansion, donor variability, and poor in vivo persistence have remained major barriers in conventional NK-cell therapy [59].
In this context, products such as FT596 and FT576 illustrate how CAR-NK development is moving beyond simple receptor transfer toward a more integrated engineering strategy [60].
At the same time, these platforms also reflect a broader shift toward multi-functional cellular design. For example, FT576 combines a BCMA-directed CAR with hnCD16, IL-15RF, and CD38 knockout, thereby aiming to improve persistence, expand combinatorial targeting, and reduce fratricide during antibody-based therapy [60]. However, despite these clear engineering advantages, the field should still be viewed with some caution, as much of the current rationale remains based on preclinical data and early translational experience [61].
Thus, while iPSC-derived CAR-NK cells represent a highly promising off-the-shelf platform, their long-term clinical value will ultimately depend on whether this added engineering complexity translates into durable and consistent patient benefit.
Collectively, these advances illustrate how CAR-NK development is increasingly driven by integrated engineering strategies that combine receptor design, cytokine programming, genome editing, and scalable manufacturing platforms.
Clinical translation of CAR-NK therapies
CAR-NK cells represent a promising alternative to CAR-T therapies, particularly in light of the logistical and safety limitations associated with autologous CAR-T manufacturing. While CAR-T therapy has revolutionized the treatment of B-cell malignancies, its application is constrained by patient-specific production, high cost, and severe toxicities, including CRS and neurotoxicity [62–64]. In contrast, CAR-NK cells can be administered without HLA matching, enabling the development of allogeneic, off-the-shelf cellular products [65].
NK cells for therapeutic use can be derived from diverse sources, including peripheral blood, umbilical cord blood, hematopoietic stem cells, iPSCs, and NK cell lines such as NK-92 [66, 67]. This source flexibility is particularly advantageous for heavily pretreated patients with lymphopenia or dysfunctional T-cell compartments [61]. Early clinical studies highlighted challenges in expanding and genetically modifying primary NK cells, whereas NK-92 cells demonstrated higher transduction efficiency but were limited in vivo persistence due to irradiation requirements [68, 69].
Among these sources, iPSC-derived CAR-NK platforms have attracted particular translational interest because they combine standardized manufacturing with extensive genetic programmability. iPSC-based production enables large-scale expansion, with one study reporting approximately 3,000-fold growth and projected generation of more than 200 doses containing > 1 × 10⁹ cells each, thereby addressing major limitations related to donor dependence and batch variability [70]. CNTY-101 exemplifies this strategy as an allogeneic anti-CD19 CAR-iNK product incorporating allo-evasion edits, IL-15 support, and an EGFR safety switch; preclinical studies demonstrated durable serial killing, improved persistence of high IL-15–expressing clones, and preserved safety-switch functionality [71]. In the ongoing first-in-human ELiPSE-1 trial, CNTY-101 has shown a manageable early safety profile, with no reported dose-limiting toxicities, GvHD, or ICANS and preliminary evidence of antitumor activity, while repeat dosing without renewed lymphodepletion appeared feasible in selected patients [72].
A clinical case study further suggested that repeated CNTY-101 infusions may promote adaptive immune activation within the tumor microenvironment alongside tumor shrinkage [73]. Nevertheless, these findings remain preliminary, and larger datasets with longer follow-up will be required to define the durability, reproducibility, and true clinical advantage of iPSC-derived CAR-NK products [71–73].
Despite these challenges, early-phase clinical trials have demonstrated encouraging safety and preliminary efficacy. For example, in a phase I study (NCT05020678), NKX019, an allogeneic CD19-directed CAR-NK product, exhibited a favorable safety profile with no high-grade CRS, neurotoxicity, or GvHD, alongside evidence of antitumor activity across non-Hodgkin lymphoma subtypes [74]. Hematologic toxicities were primarily attributable to lymphodepleting chemotherapy, and pharmacokinetic analyses revealed dose-dependent expansion [74].
Fig. 1Schematic overview of the antitumor mechanisms mediated by chimeric antigen receptor–engineered natural killer (CAR-NK) cells. CAR-NK cells eliminate tumor cells through direct and indirect mechanisms.
Direct cytotoxicity includes CAR-dependent tumor recognition, leading to perforin (PFN) and granzyme B (GzmB) release and subsequent tumor cell lysis. CAR-NK cells also retain intrinsic CAR-independent cytotoxic functions, including antibody-dependent cellular cytotoxicity (ADCC) via CD16 engagement, NKG2D-mediated recognition of stress-induced ligands (MICA/B), and death receptor–mediated apoptosis through Fas/FasL and TRAIL/TRAIL-R interactions. In addition, CAR-NK cells exert indirect antitumor effects by secreting cytokines and chemokines such as IFN-γ and RANTES, which promote the recruitment and activation of dendritic cells and T cells, thereby enhancing antitumor immune responses.
This figure was independently designed using Adobe Illustrator 2020
Collectively, the growing body of clinical evidence summarized in Table 1 highlights the favorable safety profile of CAR-NK therapies, with consistently low rates of severe CRS and neurotoxicity across platforms and tumor types.
Table 1Summary of Clinical Trials of CAR-NK Cell TherapiesNK Source / PlatformCAR TargetTrial Phase / DesignKey Findings / OutcomesCytokine Support / ConditioningNCTHaploidentical NKCD19Phase ISafe infusion in relapsed/refractory B-lineage ALL; MTD, persistence, phenotype, and preliminary efficacy evaluatedMembrane-bound IL-15 & 41BBL (via K562 feeder cells)NCT00995137Haploidentical NK (PB-derived)CD19Phase IFeasible and safe; transient CAR expression; reduced tumor burden; transient B-cell lymphopenia as expectedIL-2 post-infusion; K562-mb15-41BBL feeder during ex vivo expansionNCT01974479Autologous/Allogeneic NK (PB-derived)NKG2D ligandsPhase ISafe and feasible; transient CAR expression via mRNA electroporation; monitored adverse events and anti-tumor response in metastatic solid tumorsIL-2 post-infusion in some patientsNCT03415100Allogeneic NK (TABP EIC)PSMAEarly Phase IEvaluating safety, tolerability, preliminary efficacy in metastatic castration-resistant prostate cancer; monitored AE, PSA response, PFS, time to progressionLymphodepletion: Cyclophosphamide + Fludarabine prior to NK infusionNCT03692663Allogeneic NKCD19 / CD22Early Phase ISafety and preliminary efficacy in relapsed/refractory B cell lymphoma; total dose 50–600k/kg administered; monitored adverse eventsNot specifiedNCT03824964Allogeneic NK (QN-019a)CD19Phase I, Monotherapy & Combo (Safety, tolerability, preliminary efficacy in relapsed/refractory B-ALL (monotherapy) or B-cell lymphoma (with Rituximab); monitored DLT, TEAEs, ORR, DOR, PFS, OS, PKLymphodepletion: Cyclophosphamide + Fludarabine + VP-16; combination with RituximabNCT05379647Allogeneic NK (NKX101)NKG2D ligandsPhase I, AML / MDSSafety, tolerability, response in relapsed/refractory AML or high-risk MDS; dose-finding (Part 1), dose expansion (Part 2); monitored TEAEs, response rate, persistence, immune responseLymphodepletion: Fludarabine/Cyclophosphamide ± Decitabine or Fludarabine/Ara-CNCT04623944Allogeneic NKCD33 / CLL1Early Phase IDose escalation; safety, tolerability, PK, MTD, RP2D, efficacy in relapsed/refractory AML; monitored DLT, OS, ORR, MRD, DOR, cytokine releaseTwo doses per cycle (Day 1 & 3), 2–3 × 10^9 cells per dose; 28-day cycleNCT05215015Umbilical/Cord Blood NKCD19Phase I/IISafe dose-escalation; evaluated overall response; persistence in relapsed/refractory B-cell malignanciesIL-15 (transduced in CAR construct)NCT03056339Umbilical/Cord BloodBCMAEarly Phase ISafety and feasibility in relapsed/refractory multiple myeloma; Infused BCMA CAR-NK; monitored ORR, DLT, PFS, DORLymphodepletion: Fludarabine, Cyclophosphamide prior to infusionNCT05008536Cord Blood-derived NK (IL-15 transduced)TROP2Phase I, Dose Escalation + ExpansionStudy ongoing; primary goals: safety, tolerability, MTD, RP2D; ctDNA clearance at 3 months; monitored PFS; persistence of infused NK cells; exploratory immune profiling and QoL assessmentsLymphodepletion: Fludarabine + Cyclophosphamide; combination with Cetuximab; Rimiducid (AP1903)NCT06358430Cord Blood-derived NKClaudin18.2Phase IStudy ongoing; MTD and DLT assessment in advanced gastric & pancreatic cancer; secondary: ORR, DCR, PFS, OS, DOR; CAR-NK amplification and persistence monitoredDose groups: 2 × 10^6/kg, 4 × 10^6/kg, 8 × 10^6/kg on days 0,7,14; 3 infusions total; 28-day observation; 2-year follow-upNCT06464965NK-92HER2Phase ISafety and tolerability of intracranial NK-92/5.28.z; preliminary anti-tumor activity in HER2 + glioblastoma; combined with anti-PD-1 (Ezabenlimab)Not specified; in vivo support through intracranial administrationNCT03383978NK-92 / t-haNK (irradiated)PD-L1Phase II, Open-labelEvaluating safety and efficacy of PD-L1 CAR-NK cells combined with Pembrolizumab and N-803 in recurrent/metastatic gastric/GEJ and HNSCC; monitored CR + PR, PFS, DOR, safety; weekly infusion first 6 weeks, then biweeklyN-803 subcutaneous every 4 weeks; Pembrolizumab IV every 6 weeks; irradiated PD-L1 CAR-NK 2 × 10^9 IV weekly → biweeklyNCT04847466iPSC-Derived NK (FT596)CD19Phase I, Monotherapy & ComboDose-finding; monotherapy and combination with Rituximab/Obinutuzumab in relapsed/refractory B-cell lymphoma and CLL; monitored DLTs, AEs, ORR, DOR, PFS, OS, pharmacokinetics, chimerismLymphodepletion: Cyclophosphamide, Fludarabine, Bendamustine prior to infusionNCT04245722iPSC-Derived NK (FT576)BCMAPhase I, Monotherapy & ComboDose-finding in relapsed/refractory multiple myeloma; monotherapy or combination with Daratumumab; single and multiple doses; monitored DLT, RP2D, TEAEs, ORR, DOR, PFS, OS, PK; up to 15-year follow-upiPSC-derived NK with IL-15/IL-15RF, hnCD16; lymphodepletion: Cyclophosphamide + Fludarabine + Bendamustine; Daratumumab in combinationNCT05182073iPSC-Derived NKCD19Phase I, Dose EscalationSafety, pharmacokinetics, preliminary efficacy in relapsed/refractory CD19 + B-cell malignancies; single or three infusions over 3 weeks; monitored MTD, RP2R, ORR, CRR, DOR, PFS, OS, Tmax, Cmax, AUC, TEAEsLymphodepletion chemotherapy prior to infusion; supplemental IL-2 subcutaneous injectionNCT05336409
Despite the favorable safety profile of CAR-NK cells, their clinical efficacy still requires careful interpretation. Most published CAR-NK trials are early-phase studies with small and often heterogeneous patient cohorts, so many of the reported responses should be considered preliminary rather than clearly sustainable (NCT05336409-NCT05182073-NCT04245722). In other words, tumor shrinkage or early objective response is encouraging, but it does not yet indicate that CAR-NK cells can produce the same depth or duration of remission as has been achieved with approved CAR-T products in similar settings of B-cell malignancies or multiple myeloma [59].
This is important because a safer, more reproducible or long-lasting antitumor effect will not be widely accepted as a treatment (NCT05215015). Compared with CAR-T cells, CAR-NK cells may have lower risks of severe CRS, ICANS, and GvHD, and are more suitable for ready-to-use development [69]. At the same time, they still face important efficacy challenges, including poorer in vivo proliferation, shorter survival, donor- or source-dependent variability, and poorly defined pharmacokinetic and pharmacodynamic markers of response [75].
In solid tumors, these problems become even more difficult because CAR-NK cells must also cope with poor trafficking, stromal shedding, antigen heterogeneity, metabolic stress, and suppressive cytokines in TME [46]. Therefore, the limited clinical acceptance of CAR-NK therapy is not only due to safety concerns, but also because sustained efficacy, product stability, and a clear benchmark against approved CAR-T therapies have not yet been fully established. Future studies should demonstrate not only that CAR-NK cells are tolerable, but also that they can expand, persist, and induce durable responses in a predictable manner.
Challenges and future directions
Despite promising advances, several limitations continue to constrain the clinical impact of CAR-NK therapies and can be broadly categorized into manufacturing challenges, TME–mediated suppression, and the need for advanced genetic optimization [76]. Manufacturing hurdles include limited in vivo persistence of irradiated NK cell lines and the technical complexity of expanding primary NK cells from diverse sources, including peripheral blood, cord blood, and pluripotent stem cells [66, 77]. CAR architecture also influences stability and tonic signaling; for example, 4-1BB/CD3ζ combinations can induce deleterious tonic signaling in NK-92 and CD34⁺ HSC-derived NK cells, whereas CD28 incorporation has been shown to enhance maturation and CAR stability in select contexts [78–83].
Solid tumors impose additional barriers, including physical exclusion, metabolic stress, and immunosuppressive cytokines such as TGF-β and IL-10 [84]. Localized cytokine delivery systems represent an emerging strategy to overcome these obstacles. For instance, a combination of HER2-CAR-NK cells with an oncolytic adenoviral vector encoding IL-12, IL-15, and IFN-α (CAdTrio) improved tumor infiltration and persistence while minimizing systemic cytokine toxicity [85].
Localized cytokine delivery systems represent an emerging strategy to overcome these obstacles. For instance, a combination of HER2-CAR-NK cells with an oncolytic adenoviral vector encoding IL-12, IL-15, and IFN-α (CAdTrio) improved tumor infiltration and persistence while minimizing systemic cytokine toxicity [86]. These findings support a next generation of CAR-NK therapies that integrate optimized CAR design, cytokine armoring, and multiplex gene editing to simultaneously address trafficking, persistence, and resistance to immunosuppression.
TCR-engineered αβ T Cells
Biological distinctions and therapeutic implications
Although CAR T-cell therapy has demonstrated transformative clinical efficacy in B-cell malignancies, its success in solid tumors remains limited by restricted antigen availability, intratumoral heterogeneity, and the inability of CARs to recognize intracellular proteins [87]. TCR-engineered αβ T cells provide a complementary therapeutic modality by harnessing the native biology of the T-cell receptor (TCR) to target intracellular tumor antigens presented in the context of human leukocyte antigen (HLA) molecules [88].
Because more than 99% of tumor-specific neoepitopes originate from intracellular proteins, TCR-based therapies dramatically expand the antigenic landscape accessible to adoptive cellular immunotherapy [89–91]. This capability enables targeting of both shared oncogenic drivers and private, patient-specific neoantigens that arise from somatic mutations, thereby addressing key limitations of surface antigen–restricted CAR approaches [92].
Engagement of peptide–HLA complexes by engineered TCRs initiates canonical T-cell activation through the CD3ε/γ/δ/ζ signaling complex, leading to phosphorylation of ZAP70, activation of NFAT and AP-1 transcriptional programs, metabolic rewiring, clonal expansion, cytokine secretion, and perforin–granzyme–mediated cytotoxicity [88, 93, 94]. These processes closely mirror physiological T-cell responses and underpin the potent effector function of TCR-engineered products.
As illustrated in Fig. 2, tumor elimination by TCR-engineered αβ T cells encompasses both direct and indirect mechanisms. In addition to granule-mediated cytotoxicity [95], TCR-T cells activate death receptor pathways, including Fas/FasL and TRAIL/TRAIL-R interactions [96], inducing apoptosis even under conditions of suboptimal immunological synapse formation.
This layered cytotoxic architecture enables TCR-T cells to maintain antitumor activity in heterogeneous and dynamically evolving tumor microenvironments.
Collectively, these features position TCR-engineered αβ T cells as a versatile platform with superior antigen breadth and enhanced suitability for solid tumors compared with CAR-based therapies. The fundamental biological and functional distinctions between CAR-T and TCR-engineered αβ T cells are summarized in Table 2.
Table 2Key Biological and Functional Differences Between CAR-T and TCR-Engineered αβ T CellsFeatureCAR‑T cellsTCR‑engineered αβ T cellsReferencesAntigen recognition scopeRestricted to surface antigens; HLA-independentRecognize intracellular and surface-derived peptides presented by HLA; can target neoantigens[97, 98]Receptor architectureSynthetic receptor composed of an scFv, hinge, transmembrane region, and CD3ζ ± co-stimulatory domainsNatural or minimally modified αβ TCR that partners with endogenous CD3 (ε, γ, δ, ζ) complex[99]Signaling & immunological synapseSignaling driven mainly via CD3ζ; CAR synapses form non-physiological architecturesEngage natural TCR–pMHC synapse, with full complement of CD3 signaling subunits; IS formation, signal transduction, and downstream pathways more closely mirror physiological T cell responses[100]PolyaminesHighly effective when target antigen is a stable surface protein (e.g., CD19) but limited in solid tumors due to antigen heterogeneitySuperior potential in solid tumors due to ability to target intracellular neoantigens and diverse mutated proteins[99]
Engineering strategies to enhance persistence and function
TCR-engineered αβ T cells can be optimized through multiple complementary levels of engineering, including genetic modifications that reshape intracellular signaling pathways, protein-level receptor engineering that improves antigen recognition and signaling fidelity, and chemical or cytokine-based interventions that support expansion and persistence following adoptive transfer. Framing TCR-T therapy through these layered engineering strategies highlights how therapeutic efficacy depends not only on antigen specificity but also on the durability and functional fitness of the engineered T cells within the tumor microenvironment.
Genetic engineering strategies
TCR-engineered T cells are frequently modified through genetic engineering approaches designed to overcome inhibitory signaling pathways within the tumor microenvironment. One prominent example is the expression of dominant-negative TGF-β receptor II (DNRII), which prevents downstream SMAD signaling and protects T cells from TGF-β-mediated suppression. In patients with relapsed Epstein–Barr virus–positive Hodgkin lymphoma, DNRII-modified LMP-specific T cells exhibited robust in vivo expansion and long-term persistence, with detectable cells observed for up to four years after infusion [101–103].
Consistent with the engineering-enhanced mechanisms depicted in Fig. 2, DNRII-based cytokine resistance not only preserves TCR-T effector function but also remodels the suppressive tumor milieu, amplifying both direct cytotoxicity and secondary immune activation.
Additional genetic strategies aim to enhance survival and functional persistence. For example, genetic incorporation of IL-15 signaling modules or membrane-tethered IL-15 constructs can promote the expansion and metabolic fitness of antigen-specific T cells, supporting sustained cytotoxic responses following infusion [104]. Together, these approaches illustrate how genetic engineering can reprogram TCR-T cells to resist immunosuppressive cues and maintain durable antitumor activity.
Consistent with the engineering-enhanced mechanisms depicted in Fig. 2, DNRII-based cytokine resistance not only preserves TCR-T effector function but also remodels the suppressive tumor milieu, amplifying both direct cytotoxicity and secondary immune activation.
Another emerging strategy involves genome editing of immune checkpoint pathways. For instance, CRISPR/Cas9-mediated knockout of PD-1 in cytotoxic T lymphocytes has been explored using lentiviral delivery systems that co-express Cas9 and guide RNAs, enabling stable introduction of the editing machinery into activated primary T cells during ex vivo manufacturing. This approach illustrates how checkpoint disruption can be implemented not only by selecting relevant inhibitory targets but also by integrating gene-editing platforms capable of modifying therapeutic T cells in clinically relevant production workflows [105].
Genome editing can also be used to improve the safety and compatibility of engineered T cells by disrupting endogenous receptor components. For example, CRISPR/Cas9-mediated knockout of TRAC and CD52 in primary human T cells has been investigated using electroporation-based plasmid delivery following T-cell isolation and activation, demonstrating the feasibility of multiplex genome editing to reduce endogenous TCR expression and improve compatibility with adoptive immunotherapy platforms. Methodologically, these studies highlight the diversity of gene-editing strategies currently explored in T-cell engineering, including viral delivery systems, plasmid-based editing, and emerging ribonucleoprotein (RNP) platforms that offer improved editing efficiency and reduced genomic integration risks during therapeutic T-cell manufacturing [106–108].
Chemical and cytokine-based engineering
In addition to genetic, chemical or cytokine-based strategies are frequently used to support the expansion and persistence of TCR-engineered T cells after infusion. Interleukin-15 plays a central role in maintaining memory-phenotype T cells and promoting metabolic fitness. In adoptive cellular therapy settings, exogenous IL-15 supplementation or cytokine-supported expansion protocols can selectively enrich tumor-reactive T-cell clones, increasing polyfunctionality and cytotoxic activity against tumor targets [109].
Ex vivo studies using peripheral blood mononuclear cells (PBMCs) and post-treatment tumor-infiltrating lymphocytes (TILs) demonstrated that low-dose IL-15 selectively expanded tetramer-positive NY-ESO-1–specific T cells, increased the frequency of dual IFN-γ⁺TNF-α⁺ polyfunctional populations, and doubled cytotoxic activity in three-dimensional tumor killing assays compared with IL-15–deficient conditions [109]. These effects were reproducible across manufacturing centers and, in some patients, correlated with dynamic cytokine fluctuations following lymphodepletion.
Endogenous IL-7 and IL-15 levels transiently peak early after conditioning regimens but rapidly decline, potentially limiting the persistence of transferred TCR-T cells. This observation has motivated the development of post-infusion IL-15 supplementation strategies as well as genetically engineered TCR-T products that secrete IL-15, express membrane-tethered IL-15, or trans-present IL-15 via IL-15Rα [109].
In line with Fig. 2, IL-15 armoring reinforces persistence and polyfunctionality of TCR-engineered αβ T cells, sustaining effector differentiation and enabling repeated tumor engagement under metabolically and immunologically hostile conditions.
Protein-level engineering
A central safety concern in TCR-engineered T-cell therapy is mispairing between endogenous TCR chains and transgenic TCR α and β chains. Such hybrid receptors may exhibit unpredictable specificity, potentially leading to autoreactivity or off-target tissue damage [110–112]. As the potency of TCR-T therapies increases, particularly in combination with checkpoint blockade, these risks become increasingly clinically relevant [113–115].
Several protein-level engineering strategies have therefore been developed to improve receptor pairing fidelity and signaling architecture in TCR-engineered T cells. Domain-swapping approaches, in which constant regions of engineered TCR chains are modified to favor correct pairing, effectively prevent association with endogenous chains while preserving antigen specificity [116]. In parallel, competition for limiting CD3 subunits between endogenous and transgenic TCRs can reduce surface expression and signaling strength.
Co-expression of the complete CD3 complex alongside the therapeutic TCR improves receptor assembly, signaling fidelity, and cytotoxic function [117, 118].
Additional protein-level modifications include affinity tuning of TCR variable domains and structural optimization of constant regions to enhance peptide–HLA recognition while minimizing cross-reactivity with unrelated self-antigens [119–121]. Alternative approaches seek to expand the scope of TCR targeting through antibody-derived or TCR-like receptors directed against surface antigens. However, these platforms face challenges including suboptimal affinity, limited immunogenicity, and increased risk of cross-reactivity [122, 123].
Prioritizing neoantigen-specific TCRs, which exhibit high intrinsic specificity and reduced off-target potential, may provide a safer and more effective path forward [123].
HLA restriction and safety engineering in TCR-based therapies
Although TCR-engineered αβ T cells expand the therapeutic antigen space by recognizing intracellular peptides, this advantage is inseparable from HLA restriction. In practice, a given therapeutic TCR is usually useful only for patients who express the required HLA allele, which limits broad clinical applicability and makes patient selection more complex [88]. Tumors may also escape TCR-based recognition through HLA downregulation, loss of heterozygosity, or defects in antigen processing and presentation [124].
Safety is another major issue, because increasing TCR affinity does not always mean better therapy; it may also increase recognition of unrelated self-peptides and create unexpected off-target toxicity [125]. For this reason, current TCR-T development increasingly depends on layered safety engineering rather than potency alone [126]. These approaches include endogenous TCR disruption to reduce mispairing, affinity tuning to preserve a safe therapeutic window, suicide or depletion switches to control severe toxicity, combinatorial antigen-recognition systems to reduce single-antigen errors, and computational or immunopeptidomic screening to predict cross-reactive peptide–HLA targets before clinical use [127, 128].
The main barriers and safety-oriented engineering strategies for TCR-based therapies are summarized in Table 3.
Table 3Major barriers and safety-engineering strategies in TCR-engineered αβ T-cell therapiesBarrier / safety issueWhy it matters clinicallyMain engineering or screening strategiesReferencesHLA restrictionA therapeutic TCR recognizes antigen only in the context of a specific HLA allele, so only patients carrying that allele are eligible. This reduces broad applicability compared with HLA-independent CAR platforms.HLA-stratified patient selection; development of TCR libraries across common HLA alleles; personalized neoantigen-TCR pipelines [88]HLA loss or antigen-presentation defectsTumors may escape by reducing HLA expression, losing the relevant HLA allele, or disrupting antigen processing/presentation machinery.Combination with therapies that restore antigen presentation; targeting multiple peptide–HLA complexes; integration with immune-checkpoint or cytokine-based approaches [127]TCR mispairingTransgenic TCR chains may pair with endogenous TCR chains, generating hybrid receptors with unpredictable specificity and possible autoreactivity.Endogenous TCR knockout, especially TRAC/TRBC disruption; murinized or modified constant regions; domain swapping; optimized TCR chain pairing [128]Cross-reactivity with self-peptidesEngineered TCRs, especially affinity-enhanced TCRs, may recognize unrelated self-peptides with similar peptide–HLA features, leading to off-target tissue toxicity.Affinity tuning; alanine/glycine scanning; peptide–HLA similarity screening; immunopeptidomics; structural modeling [125]Excessive TCR affinityVery high affinity can increase potency but may narrow the safety window and allow recognition of low-abundance self-peptides.Moderate-affinity optimization rather than maximal affinity; functional testing against normal tissue panels; dose-escalation strategies [129]On-target/off-tumor toxicityThe target peptide may also be expressed at low levels in normal tissues, especially when tumor-associated rather than tumor-specific antigens are used.Preferential selection of tumor-specific neoantigens; combinatorial antigen recognition; logic-gated activation systems; tissue-expression screening [130]Uncontrolled toxicity after infusionIf severe toxicity occurs, the infused cells may need to be rapidly eliminated or suppressed.Safety switches such as inducible caspase-9, EGFRt, or RQR8; drug-inducible depletion systems; transient receptor expression [131]Incomplete off-target predictionPreclinical assays may miss rare but clinically relevant cross-reactive peptide–HLA interactions.Computational off-target prediction; machine-learning peptide–HLA binding models; large-scale peptide libraries; mass-spectrometry-based immunopeptidomics [132]
Future perspectives
Advances in neoantigen discovery and cancer vaccine development are creating new opportunities to synergize with TCR-engineered therapies. A notable example is the DNA-based neoantigen vaccine GNOS-PV02, administered in combination with plasmid IL-12 and pembrolizumab in a patient with hepatocellular carcinoma [75, 133]. Treatment induced a 44% reduction in tumor burden and profoundly reshaped the TCR repertoire [133].
High-throughput TCRβ sequencing revealed the emergence of 67,893 new peripheral blood clones and 5,126 intratumoral clones, the majority of which were absent prior to vaccination [133]. Newly expanded clones were predominantly activated CD8⁺ T cells, including high-frequency CD8⁺CD69⁺ populations. T cells engineered with these vaccine-induced TCRs selectively recognized GNOS-PV02–encoded neoantigens, confirming functional specificity [133].
These findings support a therapeutic paradigm in which personalized vaccines prime and diversify the neoantigen-specific T-cell repertoire, while TCR-engineered αβ T cells amplify and deliver these precise specificities into tumor sites. Integration of neoantigen pipelines, high-throughput TCR cloning, and non-viral gene editing may ultimately enable scalable, individualized TCR-T products with enhanced safety and durability.
Fig. 2Schematic illustration of the antitumor mechanisms employed by TCR-engineered αβ T cells. TCR-αβ T cells mediate tumor elimination primarily through direct recognition of intracellular tumor antigens presented on human leukocyte antigen (HLA) molecules, triggering T cell receptor (TCR) signaling and the release of cytotoxic granules containing perforin (PFN) and granzyme B (GzmB), ultimately leading to tumor cell lysis.
Additional direct killing occurs via death receptor–mediated apoptosis, involving Fas/FasL and TRAIL/TRAIL-R interactions. To enhance therapeutic efficacy, TCR-T cells can be further optimized through engineering-enhanced strategies, including IL-15 armoring to promote T cell persistence and polyfunctionality, and expression of a dominant-negative TGF-β receptor II (DNRII) to overcome immunosuppressive TGF-β signaling within the tumor microenvironment. Beyond direct cytotoxicity, TCR-engineered T cells exert indirect antitumor effects through cytokine secretion, particularly IFN-γ, which supports antigen presentation and amplifies adaptive immune responses against tumor cells.
This figure was independently designed using Adobe Illustrator 2020
Overall, TCR-engineered T cells complement and extend the therapeutic reach of CAR-based therapies, offering a pathway to target the vast landscape of intracellular and mutation-derived tumor antigens inaccessible to CARs.
γδ T cells
Gamma-delta (γδ) T cells constitute a distinct lymphocyte lineage that functionally bridges innate and adaptive immunity, enabling rapid, MHC-independent immune surveillance against malignant and infected cells [134]. In contrast to conventional αβ T cells, which rely on peptide–MHC complexes for activation, γδ T cells recognize a spectrum of stress-induced ligands and metabolic alterations directly, allowing immediate responses in tissues undergoing oncogenic transformation or cellular distress [135]. This unique antigen-recognition paradigm underpins their role as early sentinels in tumor immunity.
Functionally specialized γδ T-cell subsets further enhance this versatility. Tissue-resident Vδ1 cells, enriched in epithelial compartments, display pronounced NK-like cytotoxicity and are particularly effective against solid tumors [136]. Conversely, circulating Vγ9Vδ2 T cells detect phosphoantigens (pAgs) or aminobisphosphonate-induced metabolites through butyrophilin-mediated sensing mechanisms, generating robust antitumor responses upon activation [137].
Together, these features confer γδ T cells with an innate–adaptive hybrid identity that is highly compatible with allogeneic, off-the-shelf immunotherapy and associated with a minimal risk of GvHD.
γδ T cells in the tumor microenvironment
Within the TME, γδ T cells infiltrate malignant tissues and recognize a broad array of stress-associated ligands, including MICA/MICB, butyrophilin (BTN) family members, and endothelial stress markers [138]. As illustrated in Fig. 3, tumor cell recognition initiates multiple, partially overlapping cytotoxic pathways that collectively mediate efficient tumor elimination.
Activated γδ T cells induce direct tumor cell lysis through perforin–granzyme B release, NKG2D-dependent cytotoxic granule exocytosis, and death receptor–mediated apoptosis involving Fas–FasL and TRAIL–TRAIL-R interactions [139].
Beyond direct cytotoxicity, γδ T cells exert profound immunomodulatory effects within the TME. Through the secretion of IFN-γ and TNF-α, they promote dendritic-cell maturation, enhance macrophage activation, and facilitate the recruitment and priming of downstream αβ T-cell responses, thereby amplifying antitumor immunity at the tissue level (Fig. 3, indirect mechanisms).
Clinically, elevated γδ T-cell infiltration has been associated with improved survival outcomes in several malignancies, including triple-negative breast cancer [140–142].
However, γδ T-cell plasticity represents a double-edged sword. Under specific cytokine milieus, γδ T cells can differentiate into regulatory phenotypes, such as IL-4–induced Vδ1 regulatory cells, which suppress antitumor immunity through IL-10 secretion and inhibitory cross-talk with effector lymphocytes [143, 144]. Consequently, the balance between cytotoxic and regulatory γδ subsets emerges as a critical determinant of therapeutic efficacy.
Additionally, γδ T cells display specialized tissue-residency programs characterized by differential expression of CXCR6, CCR5, and CD103, influencing their trafficking, retention, and spatial organization within tumors. Strategic modulation of these residency circuits may represent a key lever for enhancing γδ T-cell persistence and function in solid tumors [145–147].
Engineering γδ T cells for cancer therapy
Recent advances in viral and non-viral gene-transfer technologies have reinvigorated efforts to engineer γδ T cells for adoptive immunotherapy [148]. Similar to other immune cell platforms discussed in this review, γδ T-cell–based therapies can be optimized through multiple complementary engineering layers, including genetic modifications that introduce synthetic receptors or modify endogenous pathways, protein-level receptor engineering that improves antigen recognition and signaling fidelity, and cytokine or pharmacologic conditioning strategies that enhance proliferation and persistence after infusion.
Genetic engineering strategies
Engineering approaches encompass retroviral and lentiviral transduction, plasmid or mRNA electroporation, Sleeping Beauty transposon systems, and CRISPR-mediated targeted gene integration, collectively enabling the scalable production of clinical-grade γδ CAR-T products [149–151]. These gene-transfer technologies provide the foundational platform for introducing synthetic antigen receptors, modifying endogenous signaling pathways, or generating multiplex-edited γδ T cells suitable for allogeneic cellular therapies.
Adapting αβ CAR-T design principles to γδ T cells represents a particularly promising strategy. CAR constructs introduced into γδ T cells can combine antigen-specific CAR signaling with the endogenous stress-recognition capacity of the γδ TCR, thereby enabling dual recognition of tumor antigens and cellular distress signals [152]. Emerging evidence indicates that γδ CAR-T cells exhibit enhanced tumor specificity and favorable safety profiles compared with unmodified γδ T cells, with potent activity across both hematologic and solid malignancies [153].
Cytokine and pharmacologic engineering
Cytokine support remains a critical determinant of γδ T-cell performance. IL-2 and IL-15 supplementation markedly enhances γδ T-cell activation, proliferation, metabolic fitness, and inflammatory cytokine production. In three-dimensional lymphoma spheroid models, IL-2 and IL-15 significantly improved the antitumor activity of CD20 CAR γδ T cells, reinforcing the importance of cytokine conditioning strategies [154].
Beyond simple supplementation, cytokine-driven expansion protocols represent a form of chemical or pharmacologic engineering that can shape the differentiation state, persistence, and functional potency of therapeutic γδ T cells during manufacturing and after infusion [155–158].
Protein-level engineering
Protein-level engineering strategies are increasingly explored to optimize receptor signaling and antigen recognition in γδ T cells. Additional receptor-targeted stimulation strategies further refine γδ T-cell activation. Monoclonal anti-CD3 antibodies provide potent co-stimulatory signals that amplify γδ TCR activation, particularly when combined with phosphoantigens or cytokines [159].
Moreover, butyrophilin family receptors, notably BTN3A (CD277) and BTN2A, serve as central regulators of Vγ9Vδ2 T-cell activation. Therapeutic anti-BTN3A antibodies have elucidated key activation mechanisms and represent promising adjuncts for γδ T-cell–based therapies [160].
Beyond direct receptor optimization, recent evidence suggests that CD16 should be viewed not merely as an ADCC-associated surface receptor, but as a functionally informative biomarker for selecting higher-potency Vγ9Vδ2 T-cell donors. In a preclinical ovarian cancer study, CD16^high^ donor-derived Vδ2 T cells consistently showed superior cytotoxicity, stronger antibody-dependent cellular cytotoxicity, and increased effector-molecule production relative to CD16^low^ counterparts, whereas forced CD16 overexpression alone did not reproduce the same phenotype [161]. Importantly, when this biomarker-guided donor selection strategy was combined with CAR and IL-15 engineering, the resulting Vδ2 products demonstrated improved persistence and tumor control in preclinical models, highlighting how donor stratification can be integrated with multi-layered γδ T-cell engineering to enhance translational potential [161].
Fig. 3Overview of the antitumor effector mechanisms mediated by γδ T cells. γδ T cells exert potent antitumor activity through a combination of direct and indirect mechanisms that are largely independent of classical MHC-restricted antigen presentation. Direct cytotoxicity is initiated by MHC-independent recognition of phosphoantigens (pAgs) on tumor cells via the γδ T cell receptor (γδ TCR), facilitated by butyrophilin family members BTN2A1 and BTN3A1, leading to the release of perforin (PFN) and granzyme B (GzmB) and subsequent tumor cell lysis.
In addition, γδ T cells retain innate-like cytotoxic pathways, including antibody-dependent cellular cytotoxicity (ADCC) mediated through CD16, NKG2D-dependent recognition of stress-induced ligands such as MICA/B, and death receptor–mediated apoptosis via Fas/FasL and TRAIL/TRAIL-R interactions. Beyond direct tumor killing, γδ T cells contribute to indirect antitumor immunity by secreting pro-inflammatory cytokines and chemokines, including IFN-γ and TNF-α, which promote the recruitment and activation of dendritic cells and macrophages, thereby amplifying innate and adaptive immune responses within the tumor microenvironment. This figure was independently designed using Adobe Illustrator 2020
Clinical translation of γδ T-cell–based immunotherapies
Accumulating preclinical and early clinical data underscore the translational promise of γδ T-cell–based immunotherapies. Compared with αβ T cells, engineered γδ T cells offer several intrinsic advantages, including minimal GvHD risk, natural tumor tropism, resilience within immunosuppressive TMEs, and dual innate–adaptive recognition capabilities [162–165].
Recent studies demonstrate that γδ CAR-T cells derived from either Vδ1 or Vγ9Vδ2 subsets exhibit superior tumor infiltration, reduced exhaustion phenotypes، and sustained cytotoxic activity relative to αβ CAR-T counterparts [145, 166]. Incorporation of NK-like co-stimulatory domains, such as 2B4 or DAP10, further augments effector function by promoting granule polarization, accelerated cytotoxic kinetics, and sustained activity within suppressive TMEs [167].
Clinically, γδ T-cell therapies have displayed favorable characteristics, including durable in vivo expansion, low exhaustion propensity, potent tumor control in xenograft models, and an absence of xenogeneic GvHD even at high cell doses [168, 169]. Collectively, these findings position γδ T cells as a leading foundation for allogeneic, off-the-shelf cellular immunotherapy platforms. Key representative preclinical and clinical studies are summarized in Table 4.
Table 4Representative Preclinical and Clinical Studies of γδ T-Cell–Based ImmunotherapiesStudy / YearTherapyIndication/ModelsKey FindingsReferencesAlnaggar et al.,2019Allogenic Vγ9Vδ2 T cellsStage IV Cholangiocarcinoma (recurrent mediastinal LN metastasis)Elimination of metastatic lymph nodes; AFP and CA19-9 remained low.No infusion-related complications; stable liver and blood biomarkers.↑ Naïve & functional CD4+/CD8 + T cells; ↓ exhausted T cells; ↑ NK cell maturation.[162]Fisher et al., 2017GD2-DAP10 “co-stimulation-only” CAR Vδ2 γδ T cellsStimulation-only” CAR Vδ2 γδ T cellsNeuroblastoma / GD2⁺ tumors (in vitro)Cytotoxicity only when both γδTCR + CAR engaged; robust killing of GD2⁺ neuroblastoma & Ewing sarcoma lines; no killing of GD2⁻ cells.Designed to avoid on-target/off-tumor toxicity by requiring dual-signal activation; no in-vivo toxicity data due to GD2 expression limits.Tumor toxicity by requiring dual-signal activation; no in-vivo toxicity data due to GD2 expression limitsDual-signal requirement prevents unwanted IFN-γ release; ↑ TNF-α via DAP10; less exhausted phenotype vs. CD3ζ CARs; limited secondary expansion[167]Nishimoto et al., 2022CD20 CAR⁺ Vδ1 γδ T cells (allogeneic, off-the-shelf)B-cell lymphomas (Raji, JVM-2; xenografts)Potent in vitro killing; IFN-γ production; significant in vivo tumor control; capable of killing rituximab-resistant cells.No xenogeneic GvHD or toxicity in NSG mice; GMP batches required IL-2 support for durability.Naïve-like, non-exhausted phenotype; high proliferative capacity (> 20,000-fold expansion); dual innate/adaptive mechanisms; strong cytotoxic gene signature (GZMB, PRF1, IFNG, NKG2D pathways).[168]Davies et al., 2024Dual-specificity engineered Vγ9Vδ2 TCR (G115 + A12)αvβ6⁺ tumors; PAg-sensitive tumors (K562, integrin models)Dual reactivity: kills αvβ6⁺ targets, PAg⁺ targets, or both; enhanced killing at low PAg levels.Preclinical only; no toxicity signals reported.Synthetic CDR3 insertion gives two antigen specificities; preserves PAg reactivity; CDR3 length tunes sensitivity; stable surface expression when optimally sized.[140]Wang et al., 2022IL-15–expanded γδ T cells (PAM + IL-2 ± IL-15)Neuroblastoma (patient-derived PBMCs; SH-SY5Y cell line)IL-15 significantly enhanced in-vitro expansion and survival of γδ T cells via anti-apoptotic signaling. IL-15 maintained Mcl-1 expression and activated JAK/STAT5 and MAPK/ERK pathways, reducing apoptosis after cytokine withdrawal.
Upregulation of activation markers CD56, CD69, HLA-DR and enhanced cytotoxicity against SH-SY5Y neuroblastoma cells. Supports feasibility of IL-15-preconditioned autologous γδT-cell adoptive therapy for NB.[157]Sawaisorn ea al., 2024Ex vivo expanded Vγ9Vδ2 T cells (Zoledronate + IL-2 ± IL-15/IL-18)Cholangiocarcinoma cell lines (HuCCT1, TFK-1)Zoledronate-activated Vγ9Vδ2 T cells showed strong cytotoxicity against cholangiocarcinoma cells; zoledronate pretreatment of tumor cells enhanced killing via IPP accumulation. Addition of IL-15 or IL-18 improved expansion, persistence, and reduced pro-apoptotic gene expression (caspase-3/8/9).
High CD107a expression indicated perforin-dependent cytotoxic mechanism. Supports feasibility of Vγ9Vδ2 adoptive therapy for cholangiocarcinoma.[156]Xu et al., 2018Anti-CD19 α/β TCR-T cells (γδTCR-based TCR-fusion receptor)CD19⁺ B-cell lymphoma & ALL (Raji, NALM-6, PDX)Comparable to 2nd-gen CAR-T; robust tumor rejection; effective re-challenge protection.Significantly reduced cytokine release compared to CAR-T; lower PD-1/exhaustion; potential for reduced CRS/neurotoxicity.More naïve/stem-cell–like phenotype; less exhaustion; lower inflammatory cytokines; maintains potency without fused co-stimulation.[169]
Future perspectives
γδ T cells are emerging as a uniquely versatile and scalable platform for next-generation cancer immunotherapy [170]. Their lack of MHC restriction, intrinsic safety profile, and multipronged tumor-recognition mechanisms render them particularly attractive for solid tumors, an arena where αβ CAR-T therapies have faced persistent challenges [171].
Promising future directions include (i) Integrating γδ TCR and CAR Signaling; Hybrid constructs combining antigen-specific CAR modules with preserved γδTCR signaling may enable multi-layered recognition of tumor stress, metabolic dysregulation, and specific antigens, reducing antigen escape [172, 173]. (ii) Establishment of allogeneic γδ T-cell banks; Industrial-scale banks of γδ CAR-T products may represent one of the most scalable cell therapy platforms, given the inherently low risk of alloreactivity [174, 175].
Macrophage therapies
Macrophages have recently emerged as compelling candidates for cancer immunotherapy owing to their intrinsic immunomodulatory capacity, robust tumor-infiltrating ability, and professional phagocytic function [176]. As central regulators of tissue homeostasis and inflammation, macrophages exhibit remarkable phenotypic plasticity shaped by microenvironmental cues. Classically activated M1-like macrophages, induced by lipopolysaccharide (LPS) and IFN-γ, exert pro-inflammatory and antitumor functions, whereas alternatively activated M2-like macrophages, driven by IL-4 or IL-10, support tissue repair but frequently promote tumor progression and immune evasion [177].
This duality places macrophages at the nexus of tumor progression and anti-tumor immunity.
Rationale for macrophage-based therapies
The clinical success of adaptive immune–based strategies, including CAR-T cells and immune checkpoint inhibitors (e.g., CTLA-4 and PD-1/PD-L1 blockade), has catalyzed renewed interest in harnessing myeloid cells as active therapeutic agents [178]. Macrophages are essential mediators of antibody-dependent cellular phagocytosis (ADCP), a mechanism exploited by multiple therapeutic antibodies. For example, daratumumab (anti-CD38) mediates tumor clearance in part through macrophage engagement, while anti-CD20 and anti-KIT antibodies similarly rely on macrophage-driven elimination of malignant cells [179].
However, tumor-associated macrophages (TAMs) are typically skewed toward an immunosuppressive M2-like phenotype, characterized by inhibitory Fc receptor expression, impaired antigen presentation, and limited tumor specificity. These limitations have driven the development of CAR-Ms, designed to confer antigen specificity, enhance phagocytic capacity, and resist suppressive signals within the TME [180].
Engineering strategies in CAR-macrophages
CAR-M technologies are rapidly evolving as macrophage-based immunotherapies transition from conceptual frameworks to translational platforms. Similar to other immune cell therapies discussed in this review, macrophage-based cellular therapies can be optimized through multiple engineering layers, including genetic introduction of synthetic receptors, cytokine or metabolic programming that modulates macrophage polarization, and protein-level receptor design that enhances phagocytic signaling and tumor recognition.
Genetic engineering of CAR-macrophages
CAR-Ms are generated by introducing chimeric antigen receptors into monocyte-derived macrophages, induced pluripotent stem cell–derived macrophages (iMACs), or related differentiation platforms [181]. Unlike CAR-T cells, CAR-Ms do not rely on T-cell receptor signaling machinery but instead activate macrophage-specific pathways governing phagocytosis, cytokine secretion, antigen presentation, and TME remodeling [182].
Cytokine and metabolic programming
Cytokine signaling plays a central role in shaping macrophage functional polarization and therapeutic performance [183]. Cytokine-driven programming strategies are therefore increasingly used to reinforce M1-like inflammatory states and enhance macrophage-mediated antitumor responses [184].
Emerging research integrates metabolic engineering to further enhance CAR-M function. The ACOD1–KEAP1–NRF2 axis, a central regulator of redox balance and macrophage polarization, has recently been implicated in tuning CAR-M activity. A metabolic CRISPR screen identified KEAP1 as a key driver of macrophage inflammatory signaling, whereas ACOD1 deletion, by suppressing itaconate production and dampening NRF2 activation, augmented pro-inflammatory cytokine release, ROS generation, and M1-like polarization.
In MSLN-CAR-iMACs, ACOD1 deficiency significantly strengthened anti-tumor responses, positioning ACOD1 as a promising metabolic engineering target and highlighting metabolic reprogramming as a strategy to potentiate CAR-M–based immunotherapies [185].
Protein-level engineering of phagocytic receptors
Protein-level engineering of CAR constructs is another important determinant of macrophage therapeutic function [186]. Unlike CAR-T platforms that primarily rely on CD3ζ signaling, CAR-M designs often incorporate intracellular signaling domains derived from phagocytic receptors such as FcRγ or Megf10 to promote actin remodeling and efficient engulfment of tumor cells [187].
As depicted in Fig. 4, CAR-Ms exert antitumor activity through multiple, non-mutually exclusive mechanisms. (i), Enhanced Phagocytosis; CAR engagement increases actin polymerization and phagocytic cup formation, promoting tumor-cell engulfment [188].
(ii), Pro-inflammatory Repolarization; Several CAR-M platforms acquire an M1-like phenotype, secreting IL-12, TNF-α, and nitric oxide [189]. This M1-skewing effect transforms TAM-rich immunosuppressive tumors into inflamed, T-cell–permissive microenvironments [189]. (iii) Antigen Presentation & T-cell Recruitment; CAR-Ms upregulate MHC-I, MHC-II, co-stimulatory molecules (CD80, CD86), which leads to enhanced T-cell priming and infiltration [190, 191].
Limitations of current preclinical models
Despite promising preclinical data, most CAR-M studies rely on murine macrophage systems, including syngeneic BALB/c tumor models such as 4T1-cHER2 and CT26-hHER2. While these platforms enable mechanistic interrogation, they incompletely recapitulate human macrophage biology, particularly with respect to trafficking dynamics, cytokine responsiveness, cellular lifespan, and immunogenicity of human CAR constructs in murine hosts [192].
Nonetheless, these models have provided important proof-of-concept evidence, demonstrating that CAR-M therapy can synergize with PD-1 blockade to enhance antitumor efficacy. These findings support the central premise that CAR-M–mediated TME remodeling can potentiate adaptive immune responses and improve therapeutic outcomes [192].
Safety considerations and clinical monitoring
Safety monitoring and systematic adverse-event reporting are fundamental to the clinical development of CAR-M therapies, given their capacity to induce serious immune-mediated toxicities, including CRS. Current risk-management strategies are largely informed by established CAR-T cell protocols but have been adapted to reflect the distinct biological behavior and functional properties of macrophages. These frameworks prioritize proactive surveillance, early clinical recognition, and timely intervention to limit the severity of CRS, which remains a central safety concern across all CAR-based therapeutic platforms [193].
In parallel, optimization of CAR-M design is essential to overcome the challenge of tumor antigen heterogeneity. Tumor cells often display variable antigen expression both within individual tumors and across different tumor types, which can compromise therapeutic efficacy when CAR constructs target a single, selectively expressed antigen. Such heterogeneity may allow antigen-negative or low-expression tumor subpopulations to evade immune recognition.
Consequently, the development of CAR-M strategies capable of multi-antigen targeting or adaptable antigen recognition is critical for improving tumor coverage and enhancing overall treatment effectiveness [194, 195].
Future perspectives
CAR-M therapies are rapidly progressing from concept to clinical development. Their unique biology offers several advantages over T-cell–based therapies, particularly in solid tumors. Future research directions include (i) iPSC-Derived Macrophage Platforms; Stable, scalable iPSC-derived macrophages could enable universal, off-the-shelf CAR-M products with consistent phenotype and superior engineering flexibility [196, 197].
(ii) Combinatorial Cellular Therapies; Because macrophages potentiate antigen presentation and promote broad immune activation: CAR-M + CAR-T, CAR-M + γδ T cells [198, 199]. CAR-Ms may function as the “first wave” that inflames tumors and opens stromal niches for subsequent cytotoxic lymphocyte therapies.
Fig. 4Tumor-Killing Mechanisms of Chimeric Antigen Receptor–Engineered Macrophages (CAR-M). his schematic illustrates the principal anti-tumor mechanisms mediated by chimeric antigen receptor–engineered macrophages (CAR-M), encompassing direct, engineering-enhanced, and indirect pathways. Direct mechanisms include CAR-mediated recognition of tumor-associated antigens (TAAs), leading to antibody-independent phagocytosis of tumor cells and degradation of the extracellular matrix (ECM) via the secretion of matrix metalloproteinases (e.g., MMP-2 and MMP-9).
CAR-M cells also promote a pro-inflammatory tumor microenvironment through the release of cytokines such as IFN-γ, IL-12, and TNF-α, resulting in the repolarization of tumor-associated macrophages from an M2 to an M1 phenotype and subsequent induction of tumor cell apoptosis. Engineering-enhanced mechanisms involve antibody-dependent cellular phagocytosis (ADCP) mediated by Fc receptor engagement (e.g., CD16). Indirectly, CAR-M cells function as antigen-presenting cells, activating T cells via MHC-I and MHC-II pathways and thereby amplifying adaptive anti-tumor immune responses.
This figure was independently designed using Adobe Illustrator 2020
CAR-NKT cells
Biological characteristics
Natural killer T (NKT) cells represent a specialized lymphocyte population that bridges innate and adaptive immunity through rapid cytokine production and cytotoxic responses upon activation [200]. These cells share phenotypic and functional characteristics with both conventional T lymphocytes and natural killer (NK) cells. Similar to T cells, NKT cells express a TCR and surface markers such as CD4 or CD8, while also displaying NK-associated molecules including NK1.1 (CD161), CD16, and CD56.
However, unlike conventional T cells that possess a highly diverse TCR repertoire, NKT cells exhibit a restricted TCR composition characterized by a limited TCRα and TCRβ chain repertoire [201]. defining feature of NKT cells is their antigen-recognition mechanism: rather than recognizing peptide antigens presented by classical MHC molecules, NKT cells recognize glycolipid antigens presented by the non-polymorphic MHC class I–like molecule CD1d on antigen-presenting cells (APCs) [202]. This recognition system enables NKT cells to exert MHC-independent cytotoxicity and supports their potential application in allogeneic “off-the-shelf” cellular therapies with reduced risk of graft-versus-host disease [203].
NKT cells can be activated through several complementary mechanisms, including TCR-mediated recognition of CD1d-presented glycolipid antigens, engagement of NK-like receptors such as KIRs and other natural killer receptors, and cytokine-driven stimulation mediated by IL-12, IL-18, and IFN-γ [28]. One of the most potent activators of invariant NKT cells is α-galactosylceramide (α-GalCer), a synthetic glycolipid ligand originally isolated from the marine sponge Agelas mauritianus. When presented by CD1d molecules on APCs, α-GalCer strongly activates NKT cells, inducing rapid secretion of diverse cytokines and promoting activation of downstream immune effectors such as NK cells and conventional T cells within the tumor microenvironment [204].
Through this rapid cytokine release and cytotoxic activity, NKT cells function as “hybrid effectors” that coordinate both innate and adaptive immune responses [205, 206]. These intrinsic properties make NKT cells particularly attractive for cellular engineering approaches. When modified with CARs, CAR-NKT cells combine antigen-specific tumor targeting with endogenous TCR- and NK receptor–mediated cytotoxicity, enabling simultaneous elimination of tumor cells and modulation of immunosuppressive myeloid populations within the tumor microenvironment [207].
Furthermore, CAR-NKT platforms demonstrate reduced risk of graft-versus-host disease, compatibility with allogeneic manufacturing strategies, and flexible CAR design, where modifications in co-stimulatory domains can influence cytokine secretion profiles and functional responses. Consequently, CAR-NKT cells are increasingly recognized as a promising cellular immunotherapy platform with substantial potential for future cancer treatment [208, 209].
Engineering strategies for CAR-NKT cells
Engineering strategies for CAR-NKT cells increasingly aim to enhance antitumor efficacy, persistence, and resistance to immunosuppressive tumor microenvironments through integrated molecular design. One major direction involves the development of armored CAR constructs capable of secreting immunostimulatory cytokines such as IL-12, IL-15, or IL-7, thereby promoting immune activation and recruiting additional effector populations within the tumor microenvironment [210]. In parallel, efforts to counteract inhibitory signaling pathways have led to checkpoint-focused engineering strategies, including CRISPR-mediated disruption of PD-1 or the incorporation of chimeric PD-1/CD28 receptors that convert inhibitory signals into costimulatory activation [210].
Advances in gene-engineering technologies have further expanded the design landscape of CAR-NKT platforms by enabling sophisticated genomic modifications and cytokine programming. For example, hematopoietic stem cell–derived CAR-NKT cells have been generated using lentiviral systems that co-express invariant NKT TCRs, CAR constructs, and cytokines such as IL-15 to support enhanced persistence and sustained antitumor activity [211].
Beyond genomic modifications, optimization of CAR receptor architecture represents another key engineering dimension. Experimental comparisons of different signaling modules indicate that the selection of intracellular costimulatory domains critically shapes CAR-NKT cell function, with 4-1BB–containing constructs generally supporting improved metabolic fitness, persistence, and sustained antitumor activity relative to alternative signaling configurations [212]. In parallel, innovations in antigen-recognition domains are being explored to improve targeting specificity and structural stability.
For instance, nanobody-based CAR designs targeting tumor antigens such as TM4SF1 have demonstrated improved stability, tumor penetration, and reduced immunogenicity compared with traditional scFv-based CAR constructs [213]. Collectively, these advances illustrate how coordinated improvements in genetic engineering, receptor architecture, and immune-modulatory programming are shaping next-generation CAR-NKT cell therapies for both hematologic and solid malignancies.
Clinical progress
Recent preclinical investigations indicate that CAR-NKT cell therapy is emerging as a potentially valuable strategy for targeting solid tumors, although most evidence is still derived from early translational models. For example, one study reported the development of Allo15 MCAR-NKT cells generated from cord-blood CD34⁺ hematopoietic stem and progenitor cells through a feeder-free differentiation system, which enabled the production of highly pure and reproducible CAR-NKT populations suitable for scalable manufacturing [214]. Nevertheless, the translational significance of these findings remains uncertain because the experiments were largely performed in immunodeficient mouse models, which do not fully reproduce the complexity of human immune responses and therefore limit the predictive value of these results for clinical settings [214].
Parallel work in ovarian cancer models further supports the potential therapeutic relevance of CAR-NKT platforms. In this context, one study demonstrated that allogeneic stem-cell–derived CAR-NKT cells were capable of simultaneously targeting tumor cells and modulating the immunosuppressive tumor microenvironment, thereby producing stronger antitumor responses compared with conventional CAR-T strategies [215]. However, despite these encouraging findings, the study emphasized that clinical evidence is still limited, and the safety and efficacy advantages observed in experimental models must be validated in human trials before firm therapeutic conclusions can be drawn [215].
Further mechanistic insights have been provided by research exploring the manufacturing scalability and engineering flexibility of CAR-NKT cells. One investigation described a feeder-free differentiation platform capable of producing CAR-NKT cells targeting multiple tumor antigens while incorporating additional genetic modules such as cytokine-enhancement systems or safety switches, demonstrating the adaptability of this platform for diverse oncologic applications [216]. Importantly, unlike conventional allogeneic CAR-T therapies that require extensive gene editing to eliminate endogenous TCR or HLA expression, CAR-NKT cells inherently exhibit reduced alloreactivity, which simplifies manufacturing and may reduce the risk of graft-versus-host disease [216].
Complementary comparative analyses between CAR-NKT and CAR-T therapies revealed notable biological differences in vivo: CAR-NKT cells showed more localized tumor infiltration, improved metabolic stability, and a balanced mixture of effector and memory-like states over time, whereas CAR-T cells tended to display broader tissue distribution and stronger exhaustion signatures that could potentially contribute to systemic toxicity and reduced persistence [217]. Nevertheless, these studies also identified several unresolved challenges, including variability between stem-cell–derived and peripheral-blood–derived CAR-NKT products and the need for more comprehensive clinical testing across different tumor types [217].
In vivo activation and recruitment strategies for CAR-NKT cells
A newer and quite interesting direction in CAR-NKT therapy is to not only engineer the cells outside the body, but also to help them, after injection, once they are inside the patient [218]. This is particularly important in solid tumors, because even if CAR-NKT cells reach the tumor site, they may not remain activated for a long time. The local environment can be poor in stimulatory signals and rich in suppressive cytokines, metabolic stress, and other inhibitory factors.
In this regard, the recently reported iNKT cell-targeted microparticle recruitment and activation system, known as iMRAS, is an important example. This system has been described as an in vivo “charging station” for CAR-directed stationary NKT cells [218]. It is designed to recruit, activate, and expand CAR-iNKT cells locally in the tumor microenvironment by providing chemotaxis and activation signals [219].
In preclinical models of lymphoma and melanoma, this strategy improved CAR-iNKT cell persistence and tumor control. The value of this approach is that it takes CAR-NKT therapy beyond a purely ex vivo generation strategy [220]. Rather than assuming that the injected cells must have all the required functions before injection, systems like iMRAS attempt to support and reactivate them directly at the tumor site.
Although this approach is still in its early stages and requires further validation before clinical use, it represents a promising future direction for solid tumor immunotherapy, particularly if it can improve local immune activation, reduce repeated ex vivo manipulations, and limit systemic toxicity [218].
Emerging paradigms: Allogeneic platforms and in vivo CAR engineering
Recent advances in molecular and cellular engineering, together with improved understanding of tumor biology, have driven the development of targeted immunotherapies, including monoclonal and bispecific antibodies, antibody–drug conjugates, checkpoint inhibitors, and adoptive cell therapies [221, 222]. Within this framework, off-the-shelf CAR-T and CAR-NK platforms have been further optimized through genetic and cellular modifications, resulting in improved functionality and broader clinical applicability [223, 224]. A key advantage of these allogeneic approaches is the ability to produce multiple doses from a single manufacturing batch, which reduces cost, improves accessibility, and simplifies treatment logistics without the need for bridging therapies [225, 226].
In contrast to autologous CAR-T therapies, which depend on patient-derived leukapheresis material, allogeneic CAR-T products are generated from donor cells and offer a more scalable and practical alternative [227, 228]. Currently, peripheral blood mononuclear cells (PBMCs) from healthy donors are the most commonly used source. However, other sources such as umbilical cord blood (UCB) and iPSCs are increasingly being explored to enable standardized and renewable manufacturing [229].
Despite these advantages, allogeneic approaches face important challenges, including GVHD, host immune rejection, and the need for lymphodepleting conditioning to support persistence [230]. To address these issues, both gene-editing strategies and non–gene-editing approaches are being developed to improve immune compatibility and enhance in vivo performance [223].
Early clinical evidence supports the feasibility of these platforms. For example, a phase I study of an allogeneic CAR-modified EBV-specific T-cell (EBV-VST) product demonstrated a favorable safety profile, with no severe CRS, neurotoxicity, or dose-limiting toxicities, and enabled repeated dosing in heavily pretreated patients with relapsed/refractory B-cell malignancies [231–233]. However, clinical outcomes appear to be influenced by factors such as the intensity and duration of lymphodepleting regimens, highlighting the need for further optimization of treatment protocols [231, 234, 235].
Overall, these findings emphasize both the potential and the current limitations of allogeneic CAR therapies, supporting continued efforts to improve their safety, persistence, and clinical efficacy.
Concurrently, in vivo CAR engineering has emerged as a promising strategy to bypass ex vivo manufacturing by directly programming immune cells within the patient using viral or LNP-based delivery systems [236, 237]. This approach reduces production complexity and cost while maintaining CAR-mediated cytotoxic function [238]. Although early studies, including first-in-human trials, demonstrate feasibility and preliminary efficacy, safety concerns such as CRS, infection risk, and limited long-term data remain significant, alongside challenges in targeting specificity and durability [239].
Ongoing efforts to incorporate targeted delivery, transient expression systems, and safety switches aim to address these limitations, positioning in vivo CAR engineering as a potentially transformative next-generation immunotherapy platform [240].
Comparative translational synthesis across engineered immune-cell platforms
Comparisons between platforms suggest that these engineered immune cell therapies should not be viewed as competing versions of the same strategy, but rather as distinct biological tools suited to different clinical problems. In hematological malignancies, where tumor antigens such as CD19 or BCMA are more readily available and malignant cells are more exposed to circulating immune effectors, platforms such as CAR-NK cells may be of particular interest due to their ready feasibility, favorable safety profile, and ability to combine CAR-dependent killing with native NK cell functions [151]. However, in solid tumors, the therapeutic challenge is more complex and involves poor delivery, stromal exclusion, antigen heterogeneity, hypoxia, suppressive cytokines, and myeloid-dominated immune resistance [241].
In this setting, TCR-engineered αβ T cells may be valuable because they can target intracellular antigens and neoantigens, while γδ T cells provide MHC-independent stress recognition, CAR macrophages can reconstitute myeloid-rich, poorly inflamed tumor niches, and CAR-NKT cells may provide a hybrid immune bridging function [242]. However, each platform has its own limitations: CAR-NK cells often lack stable persistence, TCR-T cells are limited by HLA risks and cross-reactivity, γδ T cells exhibit subset diversity and expansion, CAR macrophages may be reprogrammed by the tumor microenvironment, and CAR-NKT cells remain clinically less mature [199]. Across these platforms, several common engineering principles are emerging, including locally controlled tumor activation, cytokine armoring, checkpoint or genome editing, metabolic reprogramming, improved transduction, and scalable allogeneic or iPSC-derived generation [199].
These strengths, weaknesses, and relative best engineering strategies are summarized in Table 5.
Table 5Comparative strengths, limitations, and best-fit clinical contexts of next-generation engineered immune-cell platformsPlatformBest-fit clinical contextMain strengthsMain limitationsMost relevant engineering strategiesCAR-NK cellsHematologic malignancies; selected solid tumors in combination settingsOff-the-shelf feasibility, low GvHD risk, low severe CRS/ICANS tendency, CAR-dependent and CAR-independent killing, ADCC through CD16Limited persistence, uncertain response durability, weaker expansion than CAR-T in some settings, trafficking barriers in solid tumorsIL-15/IL-21 armoring, CISH or checkpoint editing, hnCD16 engineering, iPSC-derived NK platforms, chemokine-receptor engineeringTCR-engineered αβ T cellsSolid tumors with intracellular antigens or defined neoantigensAccess to intracellular antigen space, physiological TCR signaling, neoantigen targetingHLA restriction, HLA loss, TCR mispairing, off-target cross-reactivity, individualized antigen discoveryAffinity tuning, endogenous TCR knockout, safety switches, computational off-target screening, neoantigen-guided TCR selectionγδ T cellsSolid tumors and allogeneic/off-the-shelf settings; some hematologic malignanciesMHC-independent recognition, stress-ligand sensing, low GvHD risk, innate–adaptive immune bridgingSubset heterogeneity, expansion variability, possible regulatory polarization, limited clinical standardizationCAR-γδ design, BTN/CD16-based selection, IL-15/IL-2 conditioning, subset-specific expansion, dual-recognition strategiesCAR-macrophagesSolid tumors with myeloid-rich, stromal, or poorly inflamed TMETumor infiltration, phagocytosis, antigen presentation, ECM remodeling, M1-like repolarization, T-cell recruitmentPolarization instability, short persistence, inflammatory toxicity, manufacturing complexityPhagocytic CAR domains, M1-stabilizing circuits, metabolic editing, cytokine programming, iPSC-derived macrophage platformsCAR-NKT cellsSolid tumors and allogeneic settings; still early translational stageCD1d-related biology, reduced alloreactivity, rapid cytokine release, NK/T hybrid function, immune bridgingLess mature clinical evidence, manufacturing variability, uncertain durability and comparative advantageIL-15 armoring, optimized co-stimulation, stem-cell-derived CAR-NKT production, safety switches, in vivo activation/recruitment systems
Discussion and conclusion
The rapid evolution of engineered immune-cell platforms beyond CAR-T therapy reflects a fundamental shift in cancer immunotherapy toward more diversified, adaptable, and biologically integrated strategies. Across the five major modalities reviewed, CAR-NK cells, TCR-engineered αβ T cells, γδ T cells, CAR-macrophages, and CAR-NKT cells, several converging principles emerge, including the need to overcome structural and immunosuppressive barriers within solid tumors, enhance cellular persistence and trafficking, and engineer resistance to inhibitory signals such as TGF-β, metabolic stress, and checkpoint pathways. Recent mechanistic advances further emphasize that fine-tuning CAR signaling dynamics, rather than simply increasing activation strength, represents a critical determinant of therapeutic durability [243].
Each platform contributes distinct and complementary biological advantages, and these should increasingly be viewed as components of an integrated, multi-effector immunotherapeutic ecosystem rather than isolated modalities.
Despite these advances, key translational challenges persist. Manufacturing variability and limited persistence remain major bottlenecks across platforms. At the molecular level, emerging strategies now focus on controlling CAR activity through protein-level regulation and signaling calibration.
For instance, CCT fusion–based CAR designs enable tunable receptor endocytosis and recycling, reducing trogocytosis, antigen loss, and fratricide while improving persistence and memory formation in vivo [243]. In parallel, novel intracellular engineering approaches, such as incorporation of pre-TCR α (pTα) domains, have been shown to enhance CAR-T expansion, persistence, and long-term tumor control through regulation of translational programs (e.g., YBX1 signaling), introducing mRNA translation as an additional layer of therapeutic control [244].
Beyond receptor design, the tumor microenvironment remains a dominant barrier. Recent CRISPR-based strategies have addressed this by coupling transgene expression to endogenous, tumor-responsive promoters (e.g., NR4A2, RGS16), enabling spatially restricted cytokine release and reducing systemic toxicity [245, 246]. This approach not only improves safety but also enhances antitumor immunity through mechanisms such as epitope spreading and recruitment of endogenous T-cell responses, thereby partially overcoming tumor heterogeneity and poor immune infiltration [245].
Importantly, these findings underscore a broader shift from constitutive “armored” CAR designs toward context-dependent, dynamically regulated systems.
A major conceptual advance highlighted in this review is the transition toward multi-layered and programmable CAR engineering, integrating receptor design, gene regulation, and microenvironment-responsive elements. In parallel, iPSC-derived platforms and allogeneic systems are enabling scalable, off-the-shelf manufacturing, although challenges related to immune rejection and persistence remain. Notably, the integration of precise gene editing with context-specific transgene expression and tunable CAR signaling suggests that next-generation therapies will rely on coordinated control across multiple biological layers rather than single-modality optimization.
Ultimately, the future of cancer immunotherapy is likely to be defined by systems-level integration of complementary immune platforms together with advanced engineering strategies that regulate CAR function at genetic, transcriptional, and protein levels. For example, macrophage-mediated remodeling of the tumor microenvironment, NK-cell–driven rapid cytotoxicity, and TCR- or CAR-T–mediated antigen-specific killing may be combined with dynamic CAR control strategies (e.g., CCT fusion), translational reprogramming (pTα/YBX1 axis), and tumor-restricted payload delivery (CRISPR knock-in systems) to overcome resistance mechanisms. As clinical translation accelerates, these innovations collectively support a transition toward precision-engineered, context-adaptive immunotherapies, with the potential to achieve more durable and broadly effective responses across both hematologic and solid malignancies.