Landscape of T-cell bispecific antibodies in cancer therapy: therapeutic strategies, challenges and future prospection

Abstract

T-cell bispecific antibodies (TCBs), a specialized subclass of bispecific antibodies (BsAbs), are engineered to simultaneously engage T cells and tumor cells by binding two distinct antigens—or two epitopes on a single antigen—thereby redirecting cytotoxic T lymphocytes to eliminate malignant cells. These agents enhance anti-tumor immunity through dual mechanisms: direct activation of T cell receptor (TCR)-mediated signaling and modulation of T cell function via immune checkpoint (ICP) pathways. Clinically, TCBs have demonstrated transformative efficacy, particularly in hematologic cancers, heralding a new era in tumor immunotherapy. Despite their therapeutic promise, widespread clinical adoption of TCBs is impeded by significant challenges—including severe immune-related toxicities such as cytokine release syndrome (CRS) and immune effector cell-associated neurotoxicity syndrome (ICANS), as well as resistance conferred by the immunosuppressive tumor microenvironment (TME). This review provides a comprehensive analysis of TCBs, encompassing their molecular design, mechanisms of action, and clinical performance across oncology indications. We systematically compare current TCB platforms with respect to therapeutic strategy, structural format selection, safety profiles, and clinical outcomes. Additionally, we critically evaluate the major translational hurdles facing TCB-based therapies and explore emerging solutions—including the development of novel format antibodies, rational combination regimens with immune checkpoint inhibitors or other immunomodulators, the application of artificial intelligence (AI) in de novo antibody design, and antibody drug conjugates (ADCs). In summary, this review not only highlights the current landscape and limitations of TCBs but also outlines actionable strategies to overcome existing barriers—serving as a valuable resource for researchers, clinicians, and biopharmaceutical developers striving to advance the next generation of T-cell redirecting therapies into clinical practice.

Introduction

T cells play a central role in adaptive immunity and serve as potent effectors in anti-tumor therapy. Upon antigen presentation by antigen-presenting cells (APCs), T cells become activated through engagement of the T cell receptor (TCR) and initiation of downstream signaling cascades [1–3]. Activated CD8⁺ cytotoxic T lymphocytes form immunological synapses with cognate target cells—such as virus-infected or malignant cells—and deliver cytotoxic granules containing perforin and granzymes, triggering target-cell apoptosis [4–6]. Meanwhile, activated CD4⁺ helper T cells orchestrate broader immune responses by secreting key cytokines, including interleukin-2 (IL-2), interferon-γ (IFN-γ), and tumor necrosis factor-α (TNF-α) [7, 8]. Given these attributes, T cells have been extensively investigated and utilized as an ideal tool in the field of antitumor immunotherapy [9].

The activity of T cells is finely tuned by a network of membrane-bound receptors that either promote or suppress their function. Co-stimulatory molecules such as CD28 enhance TCR signaling via the PI3K–Akt pathway, thereby promoting T cell activation and survival [10]. Conversely, inhibitory receptors—collectively termed immune checkpoint proteins (ICPs)—serve as critical negative regulators of T cell activity [10–12]. Tumor cells frequently exploit these pathways by upregulating ICP ligands, thereby dampening T cell effector functions and enabling immune evasion, which significantly limits the efficacy of conventional immunotherapies. Targeting these regulatory axes represents a promising strategy to restore anti-tumor immunity.

Bispecific antibodies (BsAbs) are genetically engineered immunoglobulins designed to simultaneously bind two distinct antigens—or two epitopes on a single antigen—thereby enabling precise redirection of immune effector functions [13, 14]. Evidence suggests that BsAb can achieve superior antitumor activity compared with anti-TAA antibody monotherapy, particularly in hematologic malignancies, due to their ability to actively redirect T cells to tumor cells. For example, in a clinical trial involving patients with relapsed or refractory lymphoma, epcoritamab—a CD3×CD20 TCB—demonstrated substantially enhanced objective response rate (ORR, 95.2% vs. 31%) and complete response rate (CR, 76.2% vs. 9%) relative to the CD20-targeting monoclonal antibody rituximab, thereby effectively curtailing disease progression in the treated cohort [15, 16]. To date, over 100 BsAbs have entered clinical trials, and 13 have received regulatory approval, predominantly for oncologic indications [17, 18]. Among these, T-Cell bispecific antibodies (TCBs) represent a rapidly evolving subclass that redirects T cells to tumor cells, bypassing the need for MHC-restricted antigen recognition and leveraging endogenous T cell cytotoxicity through diverse mechanistic strategies [19, 20].

Despite these successes, broad application is hindered by severe toxicities, TME suppression, solid tumor heterogeneity, and high manufacturing costs. This review categorizes TCBs according to their therapeutic mechanisms and highlights clinically validated agents that have successfully completed pivotal trials. We aim to provide researchers and clinicians with a comprehensive, mechanistically grounded perspective on this transformative class of immunotherapeutics, underscoring both their current achievements and future potential.

Mechanisms and target selection of TCBs

TCBs target a broad spectrum of antigens with considerable functional heterogeneity, resulting in distinct antitumor mechanisms. Based on the specific antigens engaged and their corresponding modes of action, TCBs can be classified into three major categories: those targeting tumor-associated antigens (TAAs) and CD3, those targeting TAAs and ICPs, and those targeting dual ICPs. Each class employs unique strategies to enhance or reprogram T cell–mediated antitumor effects, offering tailored approaches to inhibit tumor growth (Fig. 1).

Fig. 1.

Mechanism of T-cell bispecific antibodies in tumor therapy. A TCBs redirect T cells to tumor cells by co-engaging CD3 and tumor-associated antigens (TAAs), and harness effector molecules such as perforin and granzymes released upon T-cell activation to mediate tumor cell killing. B TCBs target immune checkpoint proteins (ICPs) and TAAs, thereby sustaining T-cell activity and enabling tumor cell recognition to counteract immune evasion, ultimately inhibiting tumor progression and metastasis. C TCBs enhance T-cell-mediated antitumor efficacy by targeting distinct ICPs. This encompasses simultaneously engaging multiple immunosuppressive ICPs to circumvent compensatory inhibitory mechanisms on T cells, and co-targeting immunosuppressive ICPs together with T-cell costimulatory molecules to further potentiate T-cell activation and cytokine secretion, thereby enabling precise modulation of the antitumor immune response. (Created in https://BioRender.com)

TCBs targeting TAA and CD3

TCBs in this type typically detect tumor cell by targeting specific TAAs and engages T cells via CD3. CD3 is the representative marker of T cells and consists of 4 subunits. After antigen presentation reaction between antigen–major histocompatibility complex (MHC) and TCR complex, CD3 initiates downstream pathways for T cell activity [21–23]. By bridging TAAs and CD3, these TCBs activate T cells and strengthen their interaction with tumor cells, thereby promoting T cell–mediated cytotoxicity to targeted tumor cells [24, 25]. This class of TCBs predominates the existing BsAbs. By targeting distinct TAAs, these TCBs are applicable to the treatment of a wide range of tumor types. Currently, numerous anti-CD3 TCBs have been approved to market (Table 1).

Table 1.

Clinical trials of CD3xTAA TCBs

Name	Target	Structure	Tumor type	mDOR	ORR	mOS	AE (over grade3)	Phase
Teclistamab	CD3/BCMA	DuoBody	Multiple Myeloma	24	63%	22.2	88.5%	Approved
Elranatamab	CD3/BCMA	Knob and hole IgG	Multiple Myeloma	Not reached	61%	24.6	70.7%	Approved
Blinatumomab	CD3/CD19	ScFv tandem link	B-cell precursor acute lymphoblastic leukemia	6.7	68%	9.8	67%	Approved
Epcoritamab	CD3/CD20	IgG like	B-cell lymphoma	20.8	59%	18.5	> 10%	Approved
Glofitamab	CD3/CD20	IgG like	B-cell lymphoma	16.8	51.6%	25.5	57.8%	Approved
Mosunetuzumab	CD3/CD20	IgG like	Follicular lymphoma	35.9	77.8%	Not reached	51.1%	Approved
Talquetamab	CD3/GPRC5D	IgG like	Multiple Myeloma	17.5	74%	34.0	> 30%	Approved
Tarlatamab	CD3/DLL3	ScFv-Fc	Non-small-cell lung cancer	9.7	23.4%	15.2	30.8%	Approved
Linvoseltamab	CD3/BCMA	Knob and hole IgG	Multiple Myeloma	29.4	71%	31.4	73.5%	Apply for a listing
Odronextamab	CD3/CD20	Knob and hole IgG	Follicular lymphoma, B-cell lymphoma	26.0	80.5%	54.2	86.7%	Apply for a listing
CM355	CD3/CD20	IgG like	Non-hodgkin lymphoma	Not reached	100%	Not reached	22.2%	Phase I/II
GB261	CD3/CD20	IgG like	Non-hodgkin lymphoma	Not reached	73%	Not reached	25.5%	Phase III
MBS303	CD3/CD20	IgG-ScFv	Non-hodgkin lymphoma	Not reached	71%	Not reached	NR	Phase I/II
SMET12	CD3/EGFR	Knob and hole IgG	Metastatic breast cancer, gastric carcinoma, lung cancer, colon cancer	Not reached	83.3%	Not reached	NR	Phase I/II
ABBV-383	CD3/BCMA	IgG like	Multiple Myeloma	Not reached	66%	Not reached	79%	Phase III
Alnuctamab	CD3/BCMA	IgG-Fab	Multiple Myeloma	33.6	54%	Not reached	80%	Phase III
M701	CD3/EpCAM	Knob and hole IgG like	Malignant ascites	Not reported	61.5%	3.7	4%	Phase III
AMG701	CD3/BCMA	IgG like	Multiple Myeloma	3.8	83%	Not reached	7%	Phase I
EX103	CD3/CD20	Knob and hole IgG like	Non-hodgkin lymphoma	Not reached	67.7%	Not reached	1.4%	Phase I
Acapatamab	CD3/PSMA	ScFv-Fc	Prostate cancer	Not reached	7.4%	Not reached	16.1%	Phase I
JNJ-79,635,322	CD3/BCMA /GPRC5D	IgG-ScFv	Multiple Myeloma	Not reached	100%	Not reached	3.3%	Phase III
MBS314	CD3/BCMA /GPRC5D	IgG-ScFv	Multiple Myeloma	Not reached	63.6%	Not reached	NR	Phase I/II

BCMA

B-cell maturation antigen (BCMA) is a member of the TNF receptor superfamily, is predominantly expressed on B cells and plays a critical role in regulating B-cell maturation, proliferation, and survival [26, 27]. Previous studies have shown that BCMA is widely expressed across multiple myeloma cell lines and exhibits high expression levels in malignant plasma cells [28, 29].

Teclistamab is generated by the DuoBody platform and targets CD3 and BCMA. Preclinical studies demonstrate that Teclistamab effectively eliminates BCMA-positive multiple myeloma (MM) cell lines and promotes robust T-cell activation even in vitro and in vivo [30]. Clinically, teclistamab still presented impressive therapeutic effect. In the phase 2 trial of Teclistamab (NCT04557098), 65% of the MM patients achieved an objective response rate (ORR) and 40% of them achieved a complete response (CR). During the trial, 80% of patients occurred adverse event (AE) over grade 3 and the most common syndromes was CRS (70%) [31]. With extended treatment duration, Teclistamab demonstrated durable efficacy. Among patients receiving long-term Teclistamab therapy, the median duration of response (mDOR) was prolonged to 24.0 months. Furthermore, the median progression-free survival (mPFS) and overall survival (mOS) improved to 11.4 months and 22.2 months, respectively [32].

CD19

As a member of the immunoglobulin superfamily, CD19 is expressed on the surface of B cells and serves as a classical marker of early B-cell lineage development. Within the B-cell receptor (BCR)-mediated signaling pathway, CD19 functions as a co-receptor that lowers the threshold for B-cell activation by amplifying both signal strength and sensitivity, thereby promoting B-cell proliferation, differentiation, and antibody secretion. Current evidence indicates that CD19 expression is significantly upregulated in a variety of B-cell-derived hematological malignancies, making it an ideal TAA for TCB therapy [33, 34].

Blinatumomab is a T-cell-bispecific antibody composed of tandemly linked single-chain variable fragments (scFvs) that recognize CD19 and CD3. In early preclinical investigations, blinatumomab exhibited highly potent cytotoxicity against CD19-positive B-cell acute lymphoblastic leukemia (B-ALL) cell lines both in vitro and in vivo [35, 36]. Clinically, blinatumomab demonstrated robust antitumor activity in a phase II trial (NCT01209286) enrolling patients with B-ALL: 69% of patients achieved CR or complete remission with partial hematologic recovery (CRh), and the mOS reached 9.8 months. During treatment, 67% of patients experienced grade ≥ 3 adverse events [37].

CD20

CD20, also known as membrane-spanning 4-domains subfamily A member 1 (MS4A1), is a four-transmembrane protein localized on the surface of B cells. Previous studies have shown that CD20 deficiency leads to abnormalities in IgG secretion by B cells. Although interactions between CD20 and various B cell surface proteins, including CD40 and MHC class II molecules, have been observed, the underlying mechanism of action remains largely unclear. CD20 is widely expressed across multiple subsets of the B cell lineage, making it an ideal therapeutic target for various B-cell malignancies [38].

Epcoritamab, a TCB targeting both CD3 and CD20 developed using the DuoBody platform, was approved by the FDA in 2023 for the treatment of several types of lymphoma. Preclinically, this antibody demonstrated potent cytotoxicity against a range of patient-derived lymphoma cells [39]. In a clinical trial (NCT03625037) for relapsed/refractory large B-cell lymphoma (LBCL), 59% of patients achieved an ORR following treatment. The mOS and mDOR were also favorable, at 18.5 and 20.8 months, respectively [40].

GPRC5D

As an emerging therapeutic target, G protein–coupled receptor class C group 5 member D (GPRC5D) is highly expressed in multiple myeloma (MM) cells. Although the function and underlying mechanisms of GPRC5D remain largely unclear, several GPRC5D-targeted therapies have already entered clinical trials [41, 42].

Talquetamab is the first FDA-approved TCB targeting both CD3 and GPRC5D, indicated for the treatment of relapsed/refractory multiple myeloma (RRMM). In preclinical validation studies, talquetamab demonstrated potent cytotoxicity against GPRC5D-positive MM cells [43], this property that was maintained in clinical settings. In Phase I/II clinical trials involving patients with RRMM (NCT03399799, NCT04634552), treatment with talquetamab resulted in an objective response rate (ORR) of 74%. No treatment discontinuations due to adverse events were reported. Hematologic toxicities were the most common grade ≥ 3 adverse events, which occurred only during the initial treatment cycles and were reversible [44, 45].

DLL3

Delta-like ligand 3 (DLL3) serves as a ligand for the NOTCH receptor and functions as an inhibitor in NOTCH pathway [46, 47]. Previous studies have shown that DLL3 is intrinsically involved in embryonic development [48, 49]. In small cell lung cancer (SCLC) cells, DLL3 is aberrantly localized on the cell surface and exhibits markedly higher expression in normal tissues [50, 51].

Tarlatamab targets CD3 and DLL3 and is comprised of two tandem ScFvs and Fc region. Preclinical studies demonstrated effectively cytotoxicity of tarlatamab against DLL3 positive cell lines [52]. In the phase I trial of tarlatamab (NCT03319940), 23.4% of patients received tarlatamab achieved ORR [53].

PSMA

PSMA (prostate-specific membrane antigen) is a type II integral membrane glycoprotein that exists specifically on the surface of prostate cells in either monomeric or dimeric form. PSMA participates in cellular metabolism by hydrolyzing folate and glutamate, thereby regulating cell proliferation. Compared with normal tissues, PSMA expression is significantly elevated in various prostate tumors and contributes to tumor angiogenesis, making it an ideal target for prostate cancer therapy [54].

Acapatamab targets to PSMA and CD3, and the antigen recognition module is consists of tandem scFvs, with its C-terminus linked to an Fc region to enhance stability. In preclinical studies, Acapatamab demonstrated potent cytotoxicity against PSMA-positive prostate cancer cells by activating T cells [55]. In a Phase I clinical study (NCT03792841) involving patients with metastatic castration-resistant prostate cancer (mCRPC), 30.4% of patients achieved a confirmed prostate-specific antigen (PSA) response (PSA50), and 7.4% achieved a radiographic partial response. During treatment, 16.1% of patients experienced grade ≥ 3 cytokine release syndrome [56].

Although this class of TCBs has demonstrated excellent antitumor efficacy in clinical trials, the emergence of non-responsive patients during treatment, post-treatment disease relapse, and drug resistance have consistently posed challenges to their therapeutic effectiveness. In exploring the underlying causes, the phenomenon of T-cell exhaustion has gradually garnered increasing attention [57].

During chronic infection and tumor progression, T cells enter a state of exhaustion upon persistent stimulation by corresponding antigens [58, 59]. Compared with normal activated T cells, exhausted T cells exhibit metabolic reprogramming, characterized by mitochondrial damage and accumulation of reactive oxygen species, which collectively lead to a sharp decline in their metabolic capacity [60–62]. Concomitantly, the immune effector functions of exhausted T cells progressively deteriorate during the immune response, as evidenced by reduced secretion of cytokines such as IL-2 and TNF-α, decreased expression levels of cytotoxic effector molecules including perforin and granzymes, slowed proliferation rates, and upregulated expression of inhibitory receptors on T cells [63, 64].

Accumulating evidence suggests that T-cell exhaustion is a critical factor influencing the therapeutic efficacy of TCBs. In multiple clinical trials involving different TCBs, patients who presented positive responses to the antibodies demonstrated relatively high levels of T-cell activity, whereas non-responsive patients displayed prominent features of T-cell exhaustion [65, 66]. Furthermore, acquired T-cell exhaustion resulting from TCB treatment during therapy has been strongly associated with eventual treatment failure [57]. These findings indicate that maintaining T-cell activity by preventing exhaustion holds promise for overcoming the current bottlenecks in TCB therapy. Therefore, in the design of TCBs and the selection of treatment strategies, alongside activating T-cell responses, preserving the antitumor activity of T cells represents another highly promising approach.

TCB targeting TAA and ICPs

This type of TCBs target TAA and ICPs [67]. By preventing ICP–receptor interactions, these TCBs sustain T‑cell activity and reduce the possibility of tumor immune escape [68, 69]. Currently, among these TCBs, those targeting both PD-1 and TAAs are relatively well-developed, whereas the majority of those targeting other ICPs remain in the early research stage. The Table 2 lists the TCBs under clinical trials (Table 2).

Table 2.

Clinical trial data of part of ICP/TAA TCBs

Name	Target	Structure	Tumor type	mDOR	ORR	mOS	AE (over grade3)	Phase
Ivonescimab	PD-1/ VEGF	IgG-ScFv	NSCLC	Not reached	39.8%	Not reached	22.2%	Phase III
IBI322	PD-L1/ CD47	Knob and hole IgG	Solid tumor	Not reached	47.8%	Not reached	22.4%	Phase II
PM8002	PD-L1/ VEGF	IgG-nanobody	NSCLC	11.7	85.4%	Not reached	86%	Phase II
Bintrafusp alfa	PD-L1/ ⋅TGF-β	IgG fusion protein	Solid tumor	11.7	28.3%	13.7	72.6%	Phase III
SHR-1701	PD-L1/ TGF-β	IgG fusion protein	Solid tumor	10.2	56.5%	16.8	18.5%	Phase III
IBI315	PD-1/ HER2	IgG like	Solid tumor	Not reached	20.0%	Not reached	41.4%	Phase I
RC148	PD-1/ VEGF	IgG like	Solid tumor	Not reached	61.9%	Not reached	40.9%	Phase I
HX009	PD-1/ CD47	IgG fusion protein	Solid tumor	Not reached	17.6%	Not reached	15.7%	Phase I

Ivonescimab targets PD‑1 and vascular endothelial growth factor (VEGF). VEGF induces intratumoral angiogenesis [70, 71], and serves as a crucial biomarker of disease progression and bleak prognosis in cancer patients [72]. In preclinical studies, Ivonescimab effectively promoted T‑cell activation and well performed in inhibition of tumor cell proliferation [73]. In the phase Ib trial (NCT04900363), patients with non‑small cell lung cancer (NSCLC) received ivonescimab. 39.8% of patients achieved ORR. 22.2% of patients occurred grade ≥ 3 adverse events, one patient discontinued treatment, and three deaths were attributed to TRAE [74].

As research into the mechanisms regulating T‑cell activity continues to advance, in addition to the initially identified PD‑1 and CTLA‑4, an increasing number of emerging ICPs, such as LAG‑3, TIM‑3, and TIGIT, have been reported [75]. During antitumor immune responses, these ICPs suppress the antitumor activity of T cells through multiple mechanisms, including blocking TCR signaling pathways, competitively binding to costimulatory receptors on T cells, and inducing T‑cell apoptosis, thereby facilitating immune evasion [76, 77].

Based on the differences in immunosuppressive mechanisms, the mechanisms of action of antibodies targeting corresponding ICPs also vary considerably. Antibodies targeting CTLA-4 block the interaction between CTLA-4 and B7 in secondary lymphoid organs such as lymph nodes, thereby ensuring the proper initiation of T cell activation [75]; In contrast, antibodies that bind to PD-1 primarily exert their effects within the tumor microenvironment by blocking the inhibitory action of tumor cells on T cell activity [78]. Furthermore, anti-CTLA-4 antibodies also influence the immune landscape within the tumor tissue by participating in the regulation of regulatory T (Treg) cell proliferation and survival [79, 80]. Therefore, in the design of TCBs, the differences and underlying mechanisms of ICPs represent critical considerations for final target selection.

Results from a series of clinical trials indicate that the therapeutic outcomes of antibodies targeting distinct ICPs also differ considerably. Specifically, antibodies targeting CTLA‑4 are associated with a relatively lower ORR and a narrower scope of indications compared with antibodies targeting PD‑1, yet they yield a longer OS. In terms of drug safety, the adverse event profiles of CTLA‑4 antibodies differ from those of PD‑1 antibodies, with CTLA‑4 antibodies exhibiting generally higher toxicity and a greater therapeutic risk [81].

Although the mechanisms of action of various ICPs are different, monotherapy with a single ICP blockade consistently encounters the issue of drug resistance during treatment. In some patients, T cell activity remains suppressed throughout therapy, while others experience relapse after an initial response [82]. As research has progressed, the phenomenon of compensatory upregulation (also known as acquired resistance) of ICPs has garnered increasing attention [83]. For instance, in a mouse model of metastatic ovarian cancer, monotherapy with antibodies targeting PD-1, LAG-3, or CTLA-4 individually led to upregulation of other ICPs pathways, and all treated mice showed tumor relapse [84]. In another study, PD-1 and LAG-3 were found to act synergistically during T cell exhaustion; blocking either ICP alone failed to reverse the exhausted state of T cells [85, 86]. This functional redundancy exhibited by ICPs in suppressing T cell activity severely compromises the antitumor efficacy of this class of TCBs, thereby giving rise to therapeutic strategies involving blockade of multiple ICPs.

TCBs targeting two ICPs

This type of TCBs target two distinct ICPs simultaneously to avoid the diminished therapeutic efficacy and unfavorable outcomes caused by redundant pathways of ICPs. Accordingly, these TCBs co‑block distinct ICPs to more fully relieve immunosuppression and enhance T‑cell–mediated antitumor activity [68, 87, 88]. The following table lists the TCBs under clinical trials (Table 3).

Table 3.

Clinical trials of ICPs TCBs

Name	target	Structure	Tumor type	mDOR	ORR	mOS	AE (over grade3)	Phase
Cadollimab	PD-1/ CTLA4	IgG-ScFv	Solid tumor	12.7	32.3%	Not reached	28%	Phase III
Volrustomig	PD-1/ CTLA4	Knob and hole IgG	Solid tumor	17.0	48.4%	Not reached	62.5%	Phase III
Lorigerlimab	PD-1/ CTLA4	DART	mCRPC and solid tumor	4.1	25.7%	Not reached	32.3%	Phase II
Erfonrilimab	PD-L1/ CTLA4	SdAb with Fc region	TNBC	16.6	12.5%	Not reported	14%	Phase I/II
Vudalimab	PD-L1/ CTLA4	IgG	Solid tumor	Not reported	14.1%	Not reported	Not reported	Phase I/II
Rilvegostomig	PD-1/ TIGIT	Knob and hole IgG	NSCLC	10.3	61.8%	Not reported	2.1%	Phase III
IBI318	PD-1× PD-L1	IgG like	NSCLC, NPC	3.56	45.5%	Not reached	9.7%	Phase I
SI-B003	PD-1/ CTLA4	Tetravalent	Solid tumor	4.2	14.3%	Not reported	3%	Phase I
EMB-02	PD-1/ LAG3	FIT-Ig®	Solid tumor	Not reached	6.4%	Not reached	12.8%	Phase I
Tobemstomig	PD-1/ LAG3	IgG like	melanoma	Not reached	47.5%	Not reached	2.5%	Phase II
HLX301	PD-1/ TIGIT	IgG like	Solid tumor	Not reached	12.5%	Not reached	11.1%	Phase I
Acasunlimab	PD-L1/ 4-1BB	DuoBody	Solid tumor	2	31%	Not reached	8.7%	Phase II
IBI363	PD-1/ IL-2	Cytokine fusion protein	melanoma	14.0	26.4%	Not reached	29.7%	Phase I/II
FS222	PD-L1/ 4-1BB	IgG like	melanoma	Not reached	60%	Not reached	> 10%	Phase I
MCLA145	PD-L1/ 4-1BB	IgG like	Solid tumor	Not reached	9.6%	Not reached	33%	Phase I
Ragistomig	PD-L1/ 4-1BB	IgG-ScFv	Solid tumor	Not reached	15.3%	Not reached	18.4%	Phase I

Although their immunosuppressive mechanisms differ, various ICPs are intricately interconnected across multiple dimensions and levels, including expression regulation, signaling pathway crosstalk, transcriptional co-regulation, and functional complementarity, with their effector functions being similarly intertwined. Consequently, in antitumor therapies that combine different antibodies or TCBs to block distinct ICPs, the choice of specific ICP combinations requires careful consideration of the underlying therapeutic logic and strategy rationality. To date, combination strategies involving PD-1 and other ICPs are relatively mature, with multiple combinatorial regimens having been validated through clinical trials. In contrast, combination strategies targeting other ICPs remain largely in the preclinical or early-stage clinical research phases, necessitating further in-depth exploration [89].

PD-1 / CTLA-4

Cadonilimab is an IgG1-type tetravalent TCB targeting CTLA‑4 and PD‑1. The preclinical data have reported that cadonilimab presented high affinity for PD‑1 and CTLA‑4, and the incubation with peripheral blood mononuclear cells (PBMCs) effectively promoted secretion of IFN‑γ and IL‑2 [88]. In the phase Ib/II clinical trials of cadonilimab (NCT03852251), patients with advanced solid tumors were enrolled in the trials. among evaluable patients, 32.3% of them achieved ORR. During the trial, 28% of the patients occured grade ≥ 3 AEs [90].

Compared with single‑agent ICP blockade, concomitant targeting of PD‑1 and CTLA‑4 further promotes T cell proliferation and activation of cytotoxic T lymphocytes (CTLs), thereby achieving enhanced therapeutic efficacy [91, 92]. For instance, in a phase III trial of nivolumab (anti‑PD‑1) in combination with ipilimumab (anti‑CTLA‑4), patients with PD‑L1‑negative metastatic melanoma who received the combination therapy demonstrated improved survival rates compared with those receiving monotherapy [93]. Additionally, among previously untreated patients with advanced melanoma, those treated with the combination regimen achieved a superior ORR and progression‑free survival (PFS) compared with patients receiving ipilimumab alone [94]. To date, the FDA has approved concurrent PD‑1 and CTLA‑4 blockade for the treatment of various solid tumors, further underscoring the high reliability of this combinatorial strategy [95–97].

PD-1/ LAG-3

Previous studies have revealed that PD‑1 and LAG‑3 synergistically promote T cell exhaustion by co‑regulating the activity of the transcription factor TOX, thereby exerting complementary effects on T cell suppression through distinct yet interconnected signaling pathways [98, 99]. Preclinical data indicate that simultaneous blockade of PD‑1 and LAG‑3 significantly enhances CD8 + T cell proliferation and cytokine secretion, while effectively mitigating the resistance issues associated with single‑agent blockade therapies [89, 100, 101]. In the clinical setting, the combination therapy of relatlimab (anti‑LAG‑3) and nivolumab (anti‑PD‑1) was approved by the FDA in 2022 for the treatment of metastatic melanoma [102]. Currently, multiple clinical trials are evaluating combination strategies targeting both PD‑1 and LAG‑3. Notably, in a phase III clinical trial for colorectal cancer (NCT02720068), the dual blockade strategy of favezelimab (anti‑LAG‑3) and pembrolizumab (anti‑PD‑1) demonstrated favorable antitumor efficacy in patients with PD‑L1‑positive tumors [103, 104].

PD-1/ TIGIT

Previous studies have demonstrated a synergistic interaction between PD‑1 and TIGIT in suppressing T cell activity. Both molecules regulate the costimulatory signaling pathway mediated by CD226 by inhibiting the phosphorylation level of CD226 [105]. Furthermore, dual blockade of TIGIT and PD‑L1 effectively promotes the expansion of tumor‑specific cytotoxic T lymphocytes (CTLs) and the formation of immunological memory, and is superior to anti‑PD‑1 monotherapy in enhancing tumor‑infiltrating lymphocyte (TIL) proliferation and cytokine secretion [106, 107]. Clinically, the combination strategy of TIGIT and PD‑L1 blockade has also achieved breakthroughs. In a phase II clinical trial (NCT03563716) involving patients with metastatic non‑small cell lung cancer (mNSCLC), the combination of tiragolumab (anti‑TIGIT) and atezolizumab (anti‑PD‑L1) exhibbited superior antitumor efficacy compared with atezolizumab alone in patients with high PD‑L1 expression. Patients receiving the combination therapy demonstrated significantly improved ORR, PFS, and OS [108].

PD-1 / TIM-3

Accumulating evidence indicates that PD-1 and TIM-3 act synergistically in suppressing T cell activity by forming a complex with TIM-3 and its ligand Galectin-9, thereby preventing the apoptosis of exhausted T cells [109]. Furthermore, in clinical samples from patients with renal cell carcinoma (RCC), the proportion of PD-1⁺TIM-3⁺ T cells within the tumor tissue were positively correlated with aggressive tumor phenotypes and tumor volume, and was associated with a higher risk of recurrence and shorter overall survival (OS) [110]. In preclinical studies, concurrent blockade of PD-1 and TIM-3 demonstrated significantly stronger antitumor efficacy than single-agent blockade [111]. Clinical investigations of combination therapy targeting PD-1 and TIM-3 are also steadily advancing. In multiple clinical trials (NCT02608268, NCT02817633, and NCT03680508), dual blockade of PD-1 and TIM-3 using antibodies has shown antitumor activity in patients with solid tumors [112–114].

PD-1 / 4-1BB

During the adaptive immune response, T cell activity is maintained in a finely tuned steady state through the precise regulation of multiple distinct mechanisms. In addition to the inhibitory effects of ICPs, a series of costimulatory molecules—including CD28 and 4-1BB(CD137)—further promote T cell activation [115–117]. Within the tumor microenvironment, dysfunction of both ICPs and costimulatory molecules severely disrupts the normal regulation of T cell activity. Simultaneously targeting ICPs and costimulatory molecules holds promise for overcoming this challenge.

Acasunlimab (GEN1046) targets PD‑L1 and 4‑1BB and was developed on the DUOBODY platform [118, 119]. Preclinical result demonstrated that GEN1046 effectively promotes T‑cell activity to inhibit the proliferation of PD‑L1 positive tumor cell [120], and in a phase II clinical trial (NCT05117242), patients with metastatic non‑small cell lung cancer (mNSCLC) treated with Acasunlimab achieved an unconfirmed ORR of 31%. During the treatment, 8.7% of patients experienced adverse events over grade 3 [121].

Different structures of TCBs

As artificially engineered, non‑natural antibodies, TCBs exhibit a high degree of structural diversity. The modular design philosophy employed during their development further expands the potential for structural modification, enabling TCBs to be tailored to the therapeutic needs of various diseases. Based on their structural characteristics, TCBs can be classified into immunoglobulin and non‑immunoglobulin formats [20]. These distinct structural classes differ significantly in terms of engineering and optimization strategies during design, as well as in their eventual drug characterization and mechanisms of action.

IgG-based formats

Immunoglobulin format TCBs retain the antibody scaffold and achieve multi‑antigen/epitope targeting via variable‑region substitution or scFv appendages on light/heavy‑chain termini. In knob‑into‑hole designs, Fc engineering achieves heavy‑chain heterodimerization and reduces the possibility of non‑specific homodimerization interactions [20, 122, 123]. Figure 2A illustrates several representative formats of IgG-based TCBs.

Fig. 2.

Representative formats of T-cell bispecific antibodies. A Example of IgG-based format TCBs, including symmetric structures(scFv light chain linked TCBs, scFv heavy chain linked TCBs, variable region substitute TCBs, Fabs-in-tandem TCBs and variable region-in-tandem TCBs) and asymmetric structures (Bi-specific scFv tandem linked TCBs, knob and hole type TCBs and asymmetric IgG like TCBs) (B) Example of Non-IgG-based format TCBs, including variable region direct linked structures (scFv tandem linked TCBs, nanobody tandem linked TCBs, Diabody, nanobody and scFv tandem linked TCBs and tri-specific TCBs) and Fc region or receptor-ligand refused structures (Fc region fused TCBs, receptor-ligand linked TCBs and scFv Fc fused TCBs). (Created in https://BioRender.com)

Human immunoglobulins comprise five classes: IgA, IgD, IgE, IgG, and IgM. Among these, IgG is the predominant class found in serum [124]. IgG is further divided into four subclasses: IgG1, IgG2, IgG3, and IgG4, with IgG1 being the most abundant and IgG4 the least (60% vs. 5%) [124, 125]. Although these subclasses share high sequence homology, they exhibit numerous structural and functional differences. The hinge region of IgG3 is significantly longer than that of other IgG subclasses, rendering it less stable and conferring a relatively shorter half-life [126]. The affinity of the Fc domain for Fc receptors varies considerably among IgG subclasses; IgG1 and IgG3 bind Fc receptors with high affinity, thereby eliciting stronger effector functions (ADCC, CDC, and ADCP). In contrast, IgG2 and IgG4 exhibit relatively weaker Fc receptor affinity, resulting in correspondingly weaker Fc-mediated effector functions [125, 127]. Unique among the subclasses, IgG4 displays the phenomenon of Fab-arm exchange (FAE), in which reduction and reassortment of disulfide bonds in the hinge region led to the exchange of Fab domains between different IgG4 molecules, generating naturally occurring monovalent half-antibodies or bispecific antibodies [124, 128]. The subclasses of some IgG-based TCBs are presented in Table 4.

Table 4.

Subclasses of IgG-based TCBs

Name	Target	Strategies in treatment	subclass of IgG
Acasunlimab	PD-L1/4-1BB	Block ICP and activate T cells	IgG4
MEDI7526	PD-1/CD40	Target tumor cells and activate T cells	IgG4
KN052	PD-L1/OX40	Block ICP and activate T cells	IgG4
HX009	PD-1/CD47	Target tumor cells and block ICP	IgG4
Tebotelimab	PD-1/LAG-3	Block ICPs	IgG4
Teclistamab	CD3 ⋅/BCMA	Target tumor cells and activate T cells	IgG4
Cadonilimab	PD-L1/CTLA-4	Block ICPs	IgG1
ATOR-1144	CTLA-4/GITR	Activate T cells and eliminate Treg cells	IgG1
HX-044	CTLA-4 ⋅/CD47	Eliminate Treg cells	IgG1
Lomvastomig	PD-1/TIM-3	Block ICPs	IgG1

In the design of immunoglobulin format TCBs, selecting the appropriate IgG subclass as the antibody backbone based on functional differences and therapeutic goals holds promise for improving the likelihood of subsequent clinical success. Given its relatively short half-life, IgG3 is unsuitable for engineering into TCBs [129, 130]. In a study investigating the binding characteristics of different IgG subclass-based TCBs targeting CD19 and CD3, although all subclass-based TCBs were able to bind their target antigens, the IgG2-based TCB failed to bind both target antigens simultaneously. Subsequent experiments further revealed that this IgG2-based TCB could not induce the aggregation of T cells with target tumor cells. However, a chimeric IgG2-based TCB, in which the F(ab)₂ region of IgG1 or IgG4 was grafted onto the Fc domain of IgG2, restored dual-targeting functionality [131]. Currently, the vast majority of immunoglobulin-format TCBs that are either approved or in clinical trials utilize IgG1 or IgG4 as the antibody backbone. During the engineering of IgG4-based TCBs, the introduction of the S228P mutation in the core hinge region and the R409K mutation in the CH3 domain effectively prevents Fab arm exchange, thereby maintaining structural stability of the antibody [128, 132].

Given that the antitumor mechanisms elicited by TCBs vary depending on the specific combination of targets, the selection of the appropriate IgG subclass for IgG‑format TCBs is more complex than that for conventional antibodies. The primary considerations in this process are the localization of the TCB targets and the synergistic interactions between them. For example, when designing TCBs targeting a tumor‑associated antigen (TAA) and CD3, IgG4 should be selected as the antibody backbone to prevent Fc‑mediated ADCC/CDC effects from eliminating effector T cells. Conversely, when designing TCBs that target immune checkpoint ligands on tumor cells or immunosuppressive Treg cells, choosing IgG1 as the backbone can effectively enhance tumor cell clearance or ameliorate the immunosuppressive TME [133, 134].

Non-IgG-based formats

In contrast, TCBs with non-IgG formats adopt diverse structures. The most common design is two distinct variable regions that are tandemly linked, such as scFv or nanobody fragments, via specific short peptide chains as a linker. Beyond this tandem linear arrangement, alternative structural formats are also employed [20]: for example, diabodies use linkers to connect the variable region sequences of two different heavy and light chains, forming a dimeric structure through non-covalent interactions [135]. Additionally, ligands and their receptors (e.g., IL-15 and IL-15R) can be conjugated to antibody termini, enabling heterodimer formation via ligand–receptor interaction [136]. Figure 2B illustrates several representative formats of non-IgG-based TCBs.

The substantial conformational differences among non-IgG-format TCBs represent a critical factor influencing target binding. Previous studies have identified that the structural flexibility of non-IgG-format TCBs is an important determinant of antibody potency. In a comprehensive comparison of four distinct TCB structures targeting CD3×HER2, lower flexibility of the TCB upon antigen binding resulted in closer proximity between the two target antigens. Moreover, TCBs with lower flexibility formed more stable immunological synapses, thereby exhibiting enhanced cytotoxicity against target tumor cells [137].

In the context of antitumor therapy, non-IgG-format TCBs with distinct structures exhibit different therapeutic outcomes. Tarlatamab and HPN328 both target CD3 and DLL3 but possess entirely distinct spatial conformations. The antigen-recognition module of Tarlatamab consists of tandem single-chain variable fragments (scFvs) that bind CD3 and DLL3, with an Fc dimer appended at its C-terminus [138]. In contrast, the antigen-recognition module of HPN328 comprises an anti-CD3 scFv and an anti-DLL3 nanobody, positioned at the N-terminus and C-terminus, respectively, of an anti-human serum albumin (HSA) nanobody [139]. Antibody binding models indicate that Tarlatamab forms a relatively extended immunological synapse, whereas HPN328 forms a more compact one [140]. Preclinical data demonstrate that, compared with Tarlatamab, HPN328 effectively activates T cells and promotes cytotoxicity against tumor cells at lower concentrations [140]. In clinical trials involving patients with SCLC, HPN328 has shown superior therapeutic efficacy to Tarlatamab, with a higher ORR among patients receiving HPN328 (50% vs. 40%) [141, 142].

Comparison of two formats

There are several differences between the two TCB formats. Due to their structural similarity to conventional antibodies, IgG-format TCBs exhibit greater stability and a relatively longer half-life [143, 144]. However, their molecular weight exceeds that of standard antibodies, which limits their tissue penetration [20]. In contrast, lower molecular weight IgG TCBs are capable of infiltrating deep regions of target tissue. Nevertheless, their pronounced structural divergence from conventional antibodies results in reduced stability and a shorter half-life. To address this, conjugation with the Fc region or HSA is commonly employed to extend IgG TCBs’ half-life [143–146].

Similar to conventional antibodies, immunoglobulin-format TCBs contain eukaryotic-specific modifications in constant regions like glycosylation. Consequently, their production requires mammalian cell expression systems, and the high costs increase overall manufacturing expenses. Furthermore, the relatively large molecular size of immunoglobulin-format TCBs further complicates expression. In contrast, non-immunoglobulin-format TCBs generally lack constant regions, allowing expression in prokaryotic bacteria or yeast. These expression systems are more efficient than mammalian cells and significantly reduce production costs. Moreover, their compact molecular size minimizes manufacturing complexity, facilitating downstream development and application [147].

Challenges of TCB therapy

Despite its promising anti-tumor efficacy, TCB therapy still faces several unresolved challenges that warrant in‑depth investigation, including: [1] off‑target toxicities (e.g., CRS and neurotoxicity); [2] TME-mediated suppression due to inadequate T‑cell infiltration.

Potential toxicity

Off-target toxicity

The TAAs targeted by TCBs do not exist specifically on tumor cells; rather, they are also present on the surface of normal cells, albeit at lower levels. Consequently, TCBs can bind the corresponding antigens on normal cells and thereby damage normal tissues [148–151]. The principal manifestations of off-target toxicity include CRS and neurotoxicity.

Cytokine release syndrome (CRS)

In TCB‑induced antitumor immunotherapy, activated T cells potentiate the immune response by secreting inflammatory cytokines—like IFN‑γ, IL‑6, and TNF‑α—to eliminate tumor cells. During this process, an excessive immune response can lead to massive secretion of diverse inflammatory cytokines, giving rise to a cytokine storm and a clinical called CRS [152]. The principal symptoms of CRS include fever, hypotension, dyspnea, and multiple‑organ dysfunction or failure; severe cases of CRS can be life‑threatening. In clinical trials of TCBs, CRS is the primary adverse reaction, which significantly impacts safety and overall druggability [153].

TCBs with different CD3 affinities elicit distinct cytokine‑release profiles and exhibit differing pharmacokinetics during T‑cell engagement. In vitro, moderating CD3 affinity reduces cytokine secretion without significantly compromising antitumor activity [154, 155]. AMG-340, a TCB targeting prostate-specific membrane antigen (PSMA) and CD3, which has relatively weak CD3 binding, demonstrated robust antitumor activity in preclinical studies but underperformed clinically, with an ORR of 0% and a low incidence of CRS. Accordingly, TCB design must balance efficacy against CRS risk by carefully tuning the affinity for the target antigen [156, 157].

Neurotoxicity

With the expansion of TCBs in clinical application, their neurotoxic adverse effects have drawn increasing scrutiny. Early reports labeled several reactions as immune‑related adverse events, but subsequent studies have clarified that many are neurotoxic. Compared with more common toxicities, immunotherapy‑related neurotoxicity is frequently subtle and difficult to characterize [158, 159]. Clinically, TCB‑associated neurotoxicity can be partitioned by anatomical site into (i) central nervous system (CNS) manifestations—fatigue, headache, dizziness, memory impairment, cognitive dysfunction, mood instability—and (ii) peripheral neuropathies—burning pain, paresthesia, hyperalgesia, numbness, and muscle weakness. The mechanism of neurotoxicity may include cytokine‑driven blood–brain barrier disruption, directed of T‑cell trafficking into the CNS, and T‑cell‑mediated autoimmunity [160, 161].

The current mitigation strategies for neurotoxicity include dose/regimen optimization and combination with anti‑inflammatory agents [162, 163]. Unlike conventional antitumor drugs, TCBs often achieve optimal efficacy at doses below the maximum tolerated dose (MTD); setting the MTD as the recommended dose therefore carries substantial risk. Safety can be improved by step‑up dosing (SUD) to define a target dose, optimizing dosing frequency and treatment duration, and altering the administration route [164–166]. For example, in a phase II trial of mosunetuzumab (NCT02500407), subcutaneous administration was associated with a lower CRS incidence than intravenous dosing (20% vs. 39%) [167]. Corticosteroids, as non‑specific anti‑inflammatory agents, limit leukocyte trafficking and down‑regulate cytokines, chemokines, and adhesion molecules, thereby dampening systemic immune responses; they are widely used for CRS and immune effector cell-associated neurotoxicity syndrome (ICANS). Prophylactic steroids can curb CRS progression during immunotherapy [168–170]. Although cumulative steroid exposure is negatively associated with progression-free survival (PFS) and overall survival (OS) in retrospective analyses, short‑term use does not impede immunotherapy efficacy. Dexamethasone, with high blood–brain barrier penetration, can mitigate neurotoxicity, including via intrathecal administration [171]. IL‑6 is a key driver of inflammatory activation, so the blockade of this pathway is effective: tocilizumab (IL‑6R antagonist) reduces CRS incidence, and siltuximab (high‑affinity IL‑6 binding) sequesters circulating IL‑6, attenuating CRS and limiting CNS entry; multiple trials report remission of CRS and neurotoxicity with siltuximab [172–174]. In addition to anti‑inflammatory approaches, deeper TCB optimization is needed to further reduce off‑target toxicity.

Hindrance of tumor microenvironment

The internal environment of tumor tissue differs markedly from that of normal tissue. The types and functions of stromal cells, immune cells surrounding tumor cells, the vascular system, the extracellular matrix, and multiple factor molecules influence tumor progression to various degrees. Based on the distinct activity of the immune response, the TME can be categorized into hot tumors, cold tumors, and immune-privileged tumors [175, 176].

In hot tumor tissues, cytokine and chemokine levels are elevated, promoting robust infiltration of immune cells. In the marginal region of immune-privileged tumors, immune cell infiltration does not markedly differ from that of hot tumors; however, a physical barrier formed by the surrounding stroma severely impedes penetration, resulting in a scarcity of immune cells in the core. In cold tumor tissues, immunosuppression is considerably higher, and immune cell infiltration is correspondingly weaker [177].

In cold tumor tissues, regulatory immune cells—such as regulatory T cells (Tregs) and tumor-associated macrophages (TAMs)—secrete cytokines that promote immunosuppression, thereby interfering with the cytotoxicity of effector immune cells against tumor cells. In addition, the low expression of chemokines in cold tumors further hinders the infiltration of effector immune cells. These characteristics significantly diminish the therapeutic efficacy of TCBs against cold tumors. At present, the utilization of oncolytic viruses represents the feasible approach for enhancing the therapeutic effect of TCB in this context [178].

Oncolytic viruses are engineered to selectively recognize tumor cells and induce tumor cell death upon infection. Using gene-editing techniques, DNA sequences encoding the target protein can be inserted into the viral genome. Upon infection, this integrated construct is delivered into tumor cells, enabling expression of the effector protein within tumor tissues to achieve therapeutic benefit. Owing to their small size, oncolytic viruses possess high capability for intratumoral infiltration [179]. Beyond effector expression, infection of tumor cells by oncolytic viruses can also enhance adaptive immune responses, thereby modulating the TME. At present, most oncolytic virus drugs expressing TCB remain in the preclinical stage for efficacy verification. However, the oncolytic virus BS006 developed by Binhui Biotechnology was granted conditional approval for clinical trials on August 15, 2025; this virus expresses a PD-L1×CD3 TCB and has entered Phase I clinical testing (NCT05938296) [180].

Compared with hematologic malignancies, solid tumors display greater cellular and microenvironmental complexity. During tumor progression, the genetic features of malignant cells evolve continuously with proliferation, generating pronounced intratumoral heterogeneity; tumor cells within the lesion tissue differentiate into distinct subpopulations, and their uneven distribution across the tissue establishes spatial heterogeneity; moreover, cellular phenotypes shift across developmental stages of the tumor, giving rise to temporal heterogeneity. This heterogeneity seriously interferes with TCB‑mediated targeted immunotherapy, rendering the identification of effective solutions an urgent priority [181, 182].

In addition to combination therapy with antitumor agents, increasing the abundance of TCB‑targeted antigens through diverse strategies is expected to help address treatment challenges in heterogeneous solid tumors [183]. Due to the signaling pathways governing key physiological processes in tumor cells are multilayered and composed of numerous proteins, the pathway cascade is complicated; therefore, the genes encoding these pathway components tend to be relatively conserved across subpopulations. This feature suggests a practical direction for overcoming heterogeneity‑driven therapeutic resistance [184, 185]. Accordingly, selecting TCB targets that represent core drivers of tumor growth may reduce susceptibility to heterogeneity‑related interference during treatment. JANX008, an EGFR×CD3 TCB, has shown strong antitumor effects in preclinical studies and has entered a Phase I clinical trial (NCT05783622) [186].

Future direction of TCB therapy

The future development of TCBs is poised to revolutionize oncology through four key areas: novel TCB formats engineering to overcome immune evasion, synergistic combination therapies, AI‑optimized design for precision targeting, and the progression of antibody-drug conjugates (ADCs).

Novel TCB formats

Multi-specific antibody

To further enhance the anti-tumor therapeutic efficacy of TCBs, one strategy is to incorporate additional antigen-recognition modules specific to another TAA, thereby extending the bispecific format into a multi-specific antibody configuration. For example, tri-specific T cell engagers (TriTEs) can simultaneously engage three targets, enabling activation of T cells through multiple mechanisms while precisely targeting tumor cells. The Comprehensive studies demonstrate that TriTCE exhibits superior anti-tumor efficacy compared with the original TCBs, and these findings validate the feasibility of boosting the antitumor activity of T cell-engaging antibodies through additional co-stimulatory signaling, highlighting the broad therapeutic potential of this approach [187]; IBI3003 is a TriTE developed by Sinopharm Biotech that targets the GPRC5D, BCMA, and the T-cell marker CD3. This antibody employs Roche’s CrossFab bispecific technology and is being investigated for the treatment of multiple myeloma. IBI3003 exhibited excellent anti-tumor effect in Phase I/II clinical trial (NCT06083207), 100% of patients achieved OR. Regarding safety, although all patients experienced adverse events (AEs), only one case of dose-limiting toxicity (DLT) was observed [188, 189].

Antibody-cytokine fusions

With advances in antibody engineering, the capacity to modify natural antibodies has steadily increased, overcoming inherent structural limitations and enabling the development of antibody-cytokine fusion proteins. Unlike TCB molecules, which recognize two distinct antigens, these fusion proteins target a single antigen while leveraging their cytokine component to modulate immune‑cell activity [190].

IL‑2 plays a critical role in promoting T‑cell activation and supporting T‑cell survival. It has been widely used in the treatment of tumors, viral infections, immunodeficiency disorders, and autoimmune diseases. Common adverse effects of IL‑2 therapy include fever and vomiting; more severe toxicities may encompass disturbances in water and electrolyte balance, organ dysfunction, and capillary leak syndrome. Accordingly, minimizing these adverse effects is essential to improve the safety profile of IL‑2‑based therapy [191–193]. ANV419 is a fusion protein that comprises IL‑2 fused to an anti‑IL‑2 antibody, with the IL‑2 amino‑acid sequence inserted between the CDR1 and CDR2 regions of the antibody light chain. This configuration introduces steric hindrance that blocks interaction between IL‑2 and the α subunit of the IL‑2 receptor, thereby biasing binding toward the β/γ heterodimeric IL‑2 receptors rather than the α/β/γ trimeric complex. Compared with native IL‑2, preclinical studies have shown that ANV419 has an extended half‑life and selectively promotes activation and proliferation of CD8⁺ T cells and natural killer (NK) cells while markedly minimizing activation of Tregs [194]. In the Phase I clinical trial (NCT04855929) of ANV419, 64% of evaluable patients with advanced solid tumors achieved disease stabilization, and one patient exhibited a PR [195]. IBI363 is a cytokine fusion protein–based TCB that targets PD‑1 and IL‑2R. In preclinical studies, IBI363 demonstrated robust antitumor activity and effective promotion of T‑cell infiltration into tumor tissues [196]. In the Phase I clinical trial of IBI363 (NCT05460767), 67.7% of patients exhibited symptom relief [197].

Combination therapy

The conventional treatments for cancer such as chemotherapy, radiotherapy, and small‑molecule agents lack precise tumor specificity, leading to substantial toxicity in normal tissues. To mitigate these adverse effects and reduce drug dosage—thereby alleviating systemic burden—combination strategies integrating antibodies with other modalities have emerged. In such regimens, antibodies confer highly specific targeting of tumor cells, while radiotherapy, chemotherapy, and small‑molecule agents penetrate tumor cells to damage DNA or disrupt defined cellular pathways, ultimately inhibiting tumor‑cell proliferation. Compared with individual antibody therapy or traditional monotherapies, the concurrent utilization of combination therapy in tumor treatment yields superior therapeutic outcomes [198–201].

Beyond their application in combination with classical antibodies, traditional treatment modalities also hold significant promise when paired with TCBs. Chemotherapy can effectively modulate the TME of solid tumors, thereby facilitating TCB‑induced immune‑cell infiltration. Consequently, the therapeutic outcome of the combined regimen is not merely a simple superposition of the effects of individual therapies but rather a process of mutual enhancement. Ba‑PL is a TCB targeting the immune checkpoints PD‑L1 and lymphocyte activation gene 3 (LAG3). Despite its ability to effectively inhibit rapid in vivo tumor‑cell proliferation, it remains unable to prevent further expansion of tumor tissues. When combined with the chemotherapeutic agent doxorubicin, tumor volume in mice was effectively controlled, and survival was significantly increased [202]. Across multiple studies evaluating combinations of various TCBs with chemotherapy or small‑molecule inhibitors, these regimens have demonstrated notable anti‑tumor activity, underscoring their value for clinical treatment strategies [203–205].

Bispecific antibody-drug conjugates

In the course of cancer therapy development, although cytotoxic drugs that kill tumor cells exert potent antitumor effects, their cytotoxicity toward normal tissues during treatment leads to severe adverse effects [206, 207]. Consequently, in efforts to enhance tissue specificity, antibody-drug conjugates (ADCs) have emerged, which couple chemotherapeutic agents or small-molecule inhibitors to tumor-targeting antibodies through various strategies. ADCs not only trigger antibody-mediated antitumor immune responses but also precisely deliver the conjugated cytotoxic payloads to tumor tissues [208, 209]. Since their inception, ADCs have held substantial promise. Through three generations of development, ADCs have gradually transitioned from a conceptual proposal to clinical reality [210]. To date, 15 ADCs have received FDA approval for the treatment of various tumor types [211, 212].

Tumor heterogeneity represents a major constraint limiting the efficacy of various antitumor therapies. The downregulation of target antigen expression, which leads to acquired drug resistance, along with alterations in the TME, is a primary factors undermining the effectiveness of ADCs [213, 214]. In the development of next‑generation ADCs, the bispecific antibody‑drug conjugate (BsADC) strategy, which involves conjugating therapeutic payloads to BsAbs, holds substantial promise [215]. Compared with conventional ADCs, BsADCs expand the repertoire of recognized tumor cells by targeting multiple distinct tumor‑specific antigens, thereby reducing the impact of tumor heterogeneity [216, 217]. Furthermore, the multiple binding specificities of BsAbs can further decrease off‑target toxicity, enabling precise delivery of the payload [218]. During the design of BsADCs, selecting different target combinations can further extend the functional capabilities of the drug, thereby accommodating various therapeutic contexts [215]. BL‑B01D1 is the first BsADC to have been submitted for marketing approval. It consists of Lunkang Yilongtumab (an antibody targeting EGFR and HER3) conjugated to the topoisomerase I inhibitor Ed‑04. In a Phase Ib clinical trial involving patients with metastatic esophageal squamous cell carcinoma (NCT05262491), 39.6% of patients achieved ORR. Regarding drug safety, 63.3% of patients experienced TRAEs of grade 3 or higher [219]. Collectively, these findings demonstrate that BL‑B01D1 exhibits favorable therapeutic efficacy against metastatic esophageal squamous cell carcinoma, with an overall manageable safety profile.

In addition to BsAbs targeting distinct TAAs, the use of TCBs as the antibody component of ADCs also holds considerable development potential. Unlike conventional BsADCs, TCB-based ADCs are expected to achieve enhanced therapeutic outcomes through the combination of the precise cytotoxicity of ADCs and immunotherapy. Although the current development of BsADCs remains primarily focused on dual-TAA targeting, the development of TCB-based ADCs is gradually gaining momentum. IBI3014 is an ADC targeting TROP2 and PD-L1, conjugated with the DNA topoisomerase I inhibitor NT1. In a series of preclinical studies, IBI3014 demonstrated favorable antitumor activity and has since entered Phase I/II clinical trials (NCT06974812) [220, 221].

Artificial-intelligence-driven TCB engineering

With advances in computer science, the application of artificial intelligence (AI) systems in antibody engineering has steadily improved. In contrast to traditional machine learning, AI utilizes deep learning and does not require models built around predefined parameters or handcrafted features; instead, they learn hierarchical representations directly from large volumes of raw data [222, 223]. AlphaFold achieves high‑accuracy structure inference and shows strong potential for breakthroughs in protein structure prediction and analysis [222, 224].

TCB discovery faces long‑standing challenges. The hyper-diverse amino‑acid sequences within variable regions generate substantial structural heterogeneity and hampers analysis and prediction [225–227]. IgGM is a new generative foundational model that can substantially improve the efficiency of engineered antibody design. Within a unified framework, the model employs a pre‑trained protein sequence model (PPSM) to extract antibody sequence features, constructs an antigen–antibody binding model via a feature encoder, and ultimately predicts antibody sequence and structure. Leveraging the antigen–antibody interaction model, IgGM can generate antibodies based on antigen information; it also performs diverse antibody design tasks, including maturation, and humanization modification. In addition to conventional antibodies, IgGM supports nanobody prediction. This model demonstrates functional diversity and efficiency and holds strong application potential [228].

Beyond structure prediction, applying AI to modify and optimize existing antibodies can further enhance the therapeutic performance. Tezspire (tezepelumab) is a marketed asthma treatment drug that targets TSLP. After changing 20% of the CDR regions by the optimizing of the AI protein generation model Chroma, GB-0895 was created. Compared to the original antibody, the affinity of GB-0895 to TSLP increased by 20 times, and the dosing frequency was extended from once a month to once every six months. Currently, GB-0895 has entered the Phase III clinical trial [229].

Discussion

The influence of epitopes on the therapeutic effect of antibodies

Differences in epitope specificity constitute a critical determinant of the therapeutic efficacy of antibodies directed against the same target. Antibodies engaging distinct epitopes on a given antigen may operate through divergent mechanisms. Both pertuzumab and trastuzumab target human epidermal growth factor receptor 2 (HER2), yet they recognize discrete extracellular epitopes located within subdomain II and subdomain IV, respectively. Upon HER2 binding, trastuzumab exerts its antitumor activity primarily by inhibiting mitogenic signaling in HER2-overexpressing tumor cells, whereas pertuzumab blocks ligand-dependent downstream signaling by preventing HER2 from forming heterodimers with other HER family members [230]. Notably, trastuzumab has demonstrated superior clinical performance relative to pertuzumab, as reflected by an ORR of 11.6% versus 3.4% [231, 232]. Beyond therapeutic efficacy, epitope variation also substantially influences the safety profiles of therapeutic antibodies. In an analysis of ten currently marketed TCBs targeting CD3, which were categorized according to amino acid sequence homology, six distinct complementarity-determining region (CDR) specificities for CD3 were identified. Among these agents, mosunetuzumab—despite exhibiting comparable CD3-binding affinity to other TCBs with disparate epitope footprints—elicited a markedly lower incidence of AEs during treatment [233]. Collectively, these findings underscore that, in addition to careful selection of the target antigen, comprehensive evaluation of distinct epitope specificities represents an essential facet of optimizing the ultimate therapeutic performance of TCBs.

Differences in ICPs TCBs compared to combination of two distinct ICP antibodies

ICP TCBs and combinations of ICP inhibitors employ analogous target engagement strategies, TCBs are characterized by considerably more intricate mechanisms of action [234, 235]. In preclinical studies, volrustomig—a TCB directed against PD-1 and CTLA-4—exhibited preferential binding to tumor-infiltrating T cells with high PD-1 expression relative to T cells expressing low levels of PD-1, thereby endowing volrustomig with selective tumor tissue tropism. Beyond checkpoint blockade, volrustomig further promotes PD-1 internalization and subsequent degradation, thus profoundly attenuating PD-1-mediated immunosuppressive signaling. Spatially coordinated dual targeting of PD-1+/CTLA-4 + T cells markedly amplify T cell-driven antitumor activity. Compared with the corresponding monoclonal antibody combination strategy, volrustomig elicited elevated secretion of immune response-associated cytokines and conferred more effective suppression of tumor growth [235].

Although combination regimens of distinct ICIs can further augment the antitumor efficacy of single-agent checkpoint blockade, they are frequently accompanied by exacerbated treatment-related toxicity [236]. In phase 3 trial (NCT02477826), the incidence of grade 3 or 4 treatment-related adverse events in patients treated with nivolumab plus ipilimumab was markedly higher than nivolumab alone (27.0% vs. 19.4%) [237]. In contrast to the combination of ICIs, ICP TCBs not only demonstrate enhanced therapeutic efficacy but also afford an improved safety profile. In a randomized, phase 1b/2 umbrella trial (NCT05116202) evaluating patients with stage III melanoma, tobemstomig—a PD-1×LAG-3 bispecific antibody—exhibited a pathological response rate (pRR) comparable to that of the nivolumab plus ipilimumab combination (80% vs. 77.3%), yet the incidence of grade ≥ 3 adverse events was significantly reduced (2.5% vs. 22.7%). Moreover, no treatment discontinuations attributable to treatment-related adverse events were recorded in the tobemstomig group, in stark contrast to the 13.6% discontinuation rate observed with combination therapy [238]. The superior therapeutic index conferred by ICPs-TCBs—delivering potent antitumor activity while substantially mitigating severe toxicity—further underscores the translational promise of this approach.

The evaluation and analysis of future directions in TCB therapy

Although the attractive potential presented in the prospections of TCB therapy, however, the drawbacks that exist in the development of these emerging fields cannot be ignored either.

As emerging therapeutic modalities, multi-specific antibodies and cytokine fusion antibodies represent an appealing strategy for further augmenting antitumor efficacy, and current evidence indicates considerable promise for this therapeutic approach. Nevertheless, several challenges encountered during exploratory investigations merit particular attention [239–241]. Compared with conventional antibodies and BsAbs, these antibody-based agents deviate even more substantially from the canonical antibody architecture. In addition to compromised structural stability, their physicochemical properties and pharmacokinetic profiles are markedly more complex than those of traditional antibodies, thereby elevating both manufacturing costs and the uncertainty associated with clinical trials [242]. At present, the vast majority of such novel antibodies remain in early-stage research or at the initial phases of clinical development, underscoring the need for more extensive and in-depth investigation.

The strategy of combination therapy has substantially accelerated progress in antitumor treatment, as substantiated by numerous clinical trials demonstrating its feasibility. Nonetheless, several issues encountered during therapeutic administration warrant careful consideration, including overlapping drug toxicities and complications arising from the concomitant use of multiple agents. Accordingly, in the process of clinical translation, personalized selection of drug combinations tailored to distinct biomarker profiles, rational optimization of administration regimens, and comprehensive evaluation of therapeutic risks hold considerable promise for overcoming existing treatment challenges and unlocking the full therapeutic potential of combination regimens [243, 244].

Although ADCs exhibit remarkable therapeutic efficacy in clinical practice, this class of agents still faces numerous challenges during treatment. As the linker that bridges the antibody and the payload, it plays a critical role in maintaining ADC stability and enabling precise drug release [245, 246]. During drug delivery, the instability of the linker in plasma and the uncertainty of drug release timing increase the clinical risk associated with treatment [247]. Current strategies to address this issue primarily fall into two categories. The first involves further engineering of the antibody to anchor the payload conjugation site, thereby preventing premature drug exposure [248, 249]. The second approach focuses on developing linkers that are specifically responsive to the tumor microenvironment, thereby enhancing the tissue selectivity of drug release [250, 251]. Furthermore, current understanding of the subsequent intracellular trafficking and processing mechanisms of the conjugate following ADC internalization remains remarkably limited—an issue that is of particular significance for optimizing the pharmacokinetic properties of ADCs [252, 253].

The formidable computational and predictive capabilities of AI have instigated profound technological transformations and advancements in the field of antibody development; however, both the technology itself and its subsequent applications continue to face considerable challenges. Compared with the vast datasets required for model training, the current corpus of experimental antibody data remains limited in both scale and diversity, which consequently predisposes trained models to overspecialization and undermines their generalizability for predicting unknown antibody structures [254]. Furthermore, although AI models—through deep learning and LLMs—can generate attribution-based predictions of antibody architecture, the underlying biological rationale for these predictions often remains elusive, thereby posing impediments to regulatory evaluation and approval [255]. Additionally, AI-designed antibodies frequently exhibit suboptimal performance during clinical translation and validation, with notable discrepancies persisting between simulated outcomes and empirical observations [256]. Despite the formidable nature of these challenges, the integration of AI and antibody engineering represents a highly promising emerging technology, warranting sustained and in-depth investigation.

Conclusion

TCBs have revolutionized cancer immunotherapy, particularly in hematological malignancies; approved agents such as teclistamab and blinatumomab have demonstrated excellent therapeutic effects across multiple indications in clinical trials.

However, in solid tumors, TCBs have not consistently prevented disease progression or deterioration. Off‑target effects and treatment‑emergent neurotoxicity challenge drug safety, and the immunosuppressive TME markedly interferes with antitumor activity.

To overcome these barriers, future TCB development will integrate novel-format antibody engineering; combine TCBs with other antitumor modalities (e.g., CAR‑T therapies or novel small‑molecule inhibitors); pursue TCB-based ADC engineering and leverage AI‑optimized design platforms to enhance efficacy. By addressing these challenges, TCBs may become cornerstone therapies for personalized immuno‑oncology, with global research pipelines and market projections indicating potential to redefine cancer treatment paradigms over the coming decade.

Abbreviations

ADCC

Antibody‑dependent cell‑mediated cytotoxicity

ADCP

Antibody‑dependent cellular phagocytosis

AE

Adverse event

AI

Artificial intelligence

APC

Antigen‑presenting cell

BCMA

B-cell maturation antigen

TCB

T-cell bispecific antibodies

BsAb

Bispecific antibodies

CD4

Cluster of differentiation 4

CDC

Complement‑dependent cytotoxicity

CDR

Complementarity-determining regions

CNS

Central nervous system

CR

Complete response

CRS

Cytokine release syndrome

CTLA-4

Cytotoxic T-lymphocyte-associated protein 4

DLL3

Delta-like ligand 3

DLT

Dose-limiting toxicity

EGFR

Epidermal growth factor receptor

FAE

Fab arm exchange

GPRC5D

G protein-coupled receptor class C group 5 member D

HER2

Human epidermal growth factor receptor 2

HSA

Human serum albumin

ICANS

Immune effector cell-associated neurotoxicity syndrome

ICP

Immune checkpoint

IFN-γ

Interferon gamma

IL-2

Interleukin 2

IL-2R

Interleukin 2 receptor

LAG3

Lymphocyte activation gene 3

mDOR

median duration of response

MHC

Major histocompatibility complex

MM

Multiple myeloma

MTD

Maximum tolerated dose

ORR

Objective response rate

OS

Overall survival

PBMC

Peripheral blood mononuclear cell

PD-1

Programmed cell death protein 1

PFS

Progression-free survival

PPSM

Pre‑trained protein sequence model

PR

Partial response

PSMA

Prostate-specific membrane antigen

SCLC

Small cell lung cancer

SMITE

Simultaneous multiple interaction T‑cell engager

SUD

Step‑up dosing

TAA

Tumor-associated antigen

TAM

Tumor-associated macrophages

TCE

T‑cell engager

TCR

T cell receptor

TIGIT

T cell immunoreceptor with Ig and ITIM domains

TIM-3

T cell immunoglobulin and mucin domain-containing protein 3

TME

Tumor microenvironment

TNF-α

Tumor necrosis factor alpha

Treg

Regulatory T cells

TsAb

Trispecific antibodies

TSLP

Thymic stromal lymphopoietin

VEGF

Vascular endothelial growth factor

Authors’ contributions

Q.Z., and G.J conceived and designed the review. Q.Z., Y.D. and G.J drafted the manuscript text and prepared the table figures. H.P., Y.Y. and X.L. assisted in the literature search. Z.L gave the revise suggestion and Y.L. participated in the revision of the manuscript. Y.H Contributed to Resources. All authors have read and approved the article.

Funding

This work was supported by the Science and Technology Development Fund of Macau (FDCT/0150/2025/AFJ, FDCT/0010/2023/AKP, FDCT/0009/2023/RIC and 0065/2025/ITP1), the University of Macau (MYRG-GRG2023-00158-FHS-UMDF, MYRG-GRG2024-00172-FHS).

Data availability

Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.