3.1 Monoclonal antibodies

Monoclonal antibodies, as one of the most basic and established targeted therapies, offer numerous advantages such as high purity, strong specificity, and good reproducibility, and are widely applied in fields such as oncology, haematological disorders, immunology, and infectious diseases.^{53, 54} They also form the foundation for many novel therapies, such as ADCs and CAR-T.

Cusatuzumab (ARGX-110) is the first CD70-targeted monoclonal antibody developed by Argenx that is extensively studied and applied. It is a defucosylated IgG1 monoclonal antibody that restores systemic immunosurveillance for tumours through inactivating CD70.⁵⁵ In addition, cusatuzumab directly kills CD70-expressing tumour cells through enhanced antibody-dependent cell-mediated phagocytosis (ADCP), complement-dependent cytotoxicity (CDC), and antibody-dependent cell-mediated cytotoxicity (ADCC).⁵⁵ A phase I trial adopting a dose escalation scheme investigated cusatuzumab among 26 individuals suffering from late-stage solid or hematologic malignancies, focusing on safety and early efficacy. All the participants displayed satisfactory tolerance to the drug, revealing neither dose-limiting toxicities nor maximally tolerated dose. Stable disease (SD) was observed in 53.8% of participants, suggesting acceptable tolerability and initial antitumour potential.⁵⁶ A phase I/II investigation (NCT01813539) recruited individuals suffering from refractory or recurrent cutaneous T-cell lymphoma (CTCL). Twenty-seven CD70-positive patients were enrolled, comprising 9, 14, and 4 participants, respectively suffering from Sézary syndrome (SS), mycosis fungoides (MF), and additional subtypes of the disease.⁵⁷ Patients received 1 or 5 mg/kg cusatuzumab once in 21 days, and the regimen was generally tolerated, with adverse events predominantly mild to moderate.⁵⁷ A total of 1, 5, and 9 cases of complete response (CR), partial response (PR), and SD were, respectively, documented, yielding an overall response rate of 23%.⁵⁷ Notably, in patients with SS, the PR rate reached 50%, indicating a better efficacy compared to other subtypes, and this improvement was positively correlated with dose, increasing from 33% to 60% when cusatuzumab dose was elevated from 1 to 5 mg/kg.⁵⁷ However, in a phase Ib trial involving 11 NPC cases, the efficacy of cusatuzumab was limited, with SD observed in only 2 patients, while the rest showed progressive disease (PD).⁵⁸ Although both CTCL and NPC express CD70, the observed efficacy of cusatuzumab differs significantly. In CTCL, the therapeutic effect is likely mediated by direct tumour cytotoxicity via ADCC, leading to tumour cell apoptosis.⁵⁷ In NPC, investigators have associated the limited efficacy with several possible reasons, including the small number of enrolled cases, the absence of CD70-guided stratification and prior treatment-related immunosuppression, which may have compromised the immune-modulating efficacy of the antibody.⁵⁸ Notably, the relative contributions of effector function (ADCC/CDC/ADCP) and the inhibition of CD70's interaction with CD27 (especially Tregs modulation) to the antitumour activity of cusatuzumab remain insufficiently defined and may vary across tumour types depending on immune contexture and tumour biology.^{55, 56, 58}

In addition, Li et al.⁵⁹ designed and synthesized another high-affinity humanized antibody, IMM40H. Compared with cusatuzumab, IMM40H has a higher binding affinity for CD70 and stronger Fc-mediated effector functions. In xenograft mouse models of multiple myeloma, Burkitt lymphoma, and RCC, .3 mg/kg IMM40H exhibited superior antitumour activity than 1.0 mg/kg cusatuzumab, suggesting enhanced therapeutic potential.⁵⁹ Its clinical evaluation has entered Phase I (NCT05549557).

3.2 Antibody–drug conjugates

Since the use of cytotoxic drugs in cancer treatment, researchers have been seeking ways to improve efficacy while reducing overall toxicity to patients. ADCs, as an innovative therapeutic approach, have been developed based on monoclonal antibodies. ADCs covalently combine monoclonal antibodies and cytotoxic drugs (payloads) with linkers to, respectively, utilize their high specificity and high toxicity toward cancer cells.⁶⁰ Due to the presence of the antibody, these drugs can locate CD70-positive tumour cells, forming an ADC–antigen complex. This complex enters the cell through endocytosis and undergoes intralysosomal dissociation to release the payload.⁶¹ These cytotoxic drugs precisely kill tumour cells by inducing DNA damage, disrupting microtubules, and other mechanisms, while minimizing damage to normal tissues.^{62, 63}

Several CD70-targeted ADCs are currently in clinical or preclinical development. Nakae et al.⁶⁴ developed a CD70-ADC loaded with monomethyl auristatin F (MMAF) for treating uterine leiomyosarcoma via cytotoxic mechanisms. Shiomi et al.⁶⁵ conjugated vorsetuzumab with MMAF to form a CD70-ADC for the treatment of platinum-resistant ovarian cancer. Although several ADCs showed promising results in early studies, some have been discontinued due to suboptimal efficacy or safety concerns.⁶⁶ SGN-75, developed by Seattle Genetics (now Seagen), is an ADC formed by conjugating a humanized IgG1 monoclonal antibody with MMAF via a maleimidocaproyl (mc) linker.⁶⁷ Although SGN-75 showed significant antitumour activity in pancreatic and ovarian cancer models, the first-in-human NCT01015911 study recruiting CD70-positive metastatic RCC and non-Hodgkin lymphoma (NHL) sufferers revealed that its clinical efficacy did not meet expectations. Among the 47 patients treated every 3 weeks, only 1 achieved CR and 2 achieved PR.⁶⁸ Due to limited clinical efficacy and the emergence of dose-limiting toxicities, Seagen ultimately discontinued the development of SGN-75. The cytotoxic payload MMAF has been associated with ocular toxicity and thrombocytopenia, both of which were identified as key adverse events during the trial and contributed to the decision to terminate further development.⁶⁶ MDX-1203, another CD70-targeted ADC comprising a fully human antibody and a duocarmycin-derived cytotoxic payload, entered a phase I study (NCT00944905) involving 26 patients diagnosed with ccRCC or B-cell NHL. Although 69% of patients achieved SD, no PR or CR were observed. Moreover, high-dose administration (15 mg/kg) was associated with notable toxicities, including pleural and pericardial effusions, causing permanent therapy stoppage in 50% participants receiving this dose.⁶⁹

However, the development of CD70-targeted ADCs has not stalled, and the next-generation CD70-ADCs, such as SGN-CD70A and AMG 172, have entered the research phase, aiming to achieve a better balance between efficacy and toxicity. SGN-CD70A, also developed by Seagen, is a CD70-targeted ADC with a pyrrolobenzodiazepine (PBD) dimer as its payload, a particularly effective DNA crosslinker. Compared with MMAF, the PBD dimer has a more direct mechanism of action and can exhibit higher toxicity at lower doses. Preclinical studies demonstrated that SGN-CD70A suppressed proliferation and triggered apoptosis in CD70-positive CTCL cells, with multiple high-dose administrations significantly prolonging the survival of lymphoma mice.⁷⁰ However, data from a phase I trial (NCT02216890) revealed concerns regarding both the therapeutic effectiveness and safety profile. In 20 CD70-positive NHL patients, SGN-CD70A showed some monotherapy activity, including 1 CR and 3 PR. However, up to 75% of patients developed frequent and severe thrombocytopenia.⁷¹ Similarly, in RCC, high-dose treatment with SGN-CD70A was also associated with severe toxicities, including anaemia and thrombocytopenia. Despite achieving a disease control rate of 78%, only one patient achieved PR, with the median progression-free survival being as short as 3.5 months.⁷² These results restrict the clinical application potential of SGN-CD70A. Similar to SGN-CD70A, AMG 172, comprising a monoclonal antibody against human IgG1 linked by a noncleavable linker to maytansinoid as the payload, was evaluated by the NCT01497821 clinical study involving advanced ccRCC sufferers.⁷³ Among 37 evaluable cases, only 5.4% achieved PR while 35.1% experienced PD, reflecting limited clinical activity despite manageable toxicity.⁷³

Overall, it is challenging to develop ADCs targeting CD70 and apply them clinically.⁶⁶ Their limited success to date is primarily due to insufficient antitumour activity in clinical settings and unacceptable toxicity profiles associated with current payloads and linker designs. The inability of preclinical models to accurately mimic human physiological complexity, particularly in toxicity prediction, further hinders translational success. Many ADCs that appear safe in animal studies may exhibit unanticipated adverse effects in clinical trials. Additional factors such as unstable linkers and nonspecific binding further narrow the therapeutic window. Therefore, future efforts should focus on optimizing ADC design, incorporating cleavable and tumour-specific linkers, selecting better-tolerated payloads, and exploring combination strategies or next-generation platforms like bispecific ADCs or conditionally activated ADCs to improve therapeutic precision and reduce systemic toxicity.^{61, 74, 75}

3.3 CAR-T cell therapy

As a type of adoptive immunotherapy, CAR-T involves genetically modifying T cells to enable the expression of tumour antigen-recognizing chimeric antigen receptors (CARs), thereby precisely targeting and eliminating tumour cells⁷⁶ (Figure 3A). Identifying suitable new targets, optimizing existing CAR-T designs, and developing novel CAR-T therapies remain major research directions.⁷⁷ The safety and efficacy of several CAR-T strategies targeting CD70 have been assessed preclinically and within patient cohorts.^{78, 79} We summarize the common strategies for constructing CD70-targeted CARs, including nanobody-based CAR-T, dual-target CAR-T, and allogeneic CAR-T, which will be discussed in detail later (Figure 3B).

3.4 CAR-NK cell therapy

NK cells possess a unique target recognition mechanism that differs from T cells, allowing them to exert innate cytotoxicity against tumour cells. When engineered with CARs, CAR-NK cells gain enhanced specificity and retain their natural antitumour properties. Since NK cells lack TCR expression, CAR-NK cells have a minimal risk of inducing GvHD and exhibit a low incidence of serious toxicities such as CRS or neurotoxicity.¹¹⁵ More importantly, they can be sourced from allogeneic donors, overcoming the limitations of autologous cell therapy, expanding cell sources, and significantly reducing treatment costs.¹¹⁶

CD70-targeted CAR-NK cells remain in early-stage research, yet current findings show promising application potential. The incorporation of cytokines, including IL-15, has been reported to evidently enhance their persistence and antitumour function¹¹⁷ (Figure 4). Eynde et al.¹¹⁸ designed CD70-CAR-IL-15 NK cells to target CD70-positive colorectal and pancreatic ductal adenocarcinoma cells as well as cancer-associated fibroblasts (CAFs). That research exhibited that IL-15 enhanced the cytotoxicity of CAR-NK cells through promoting CAR expression and elevating proinflammatory cytokine secretion.¹¹⁸ Additionally, IL-15 can form an autocrine positive feedback loop that further activates and enhances the antitumour capacity of NK cells.¹¹⁸ Another study demonstrated in CD19-negative BCL that repeated dosing strategies, by continuously increasing IL-15 levels in vivo, further enhanced the persistence and proliferation of CD70-CAR NK cells, thereby promoting their antitumour effects.³⁷ Wang et al.¹¹⁹ developed human induced pluripotent stem cell (iPSC)-derived 70CAR-iNK cells that incorporated IL-15 receptor α/IL-15 fusion protein (IL15RF), high-affinity noncleavable CD16, and CD70 gene knockout, significantly enhancing NK cell cytotoxicity and persistence (Figure 4). These cells also reduced allogeneic T-cell immune responses, avoiding immune rejection issues commonly seen in traditional cell therapies.¹¹⁹

In a preclinical study, Acharya et al.¹²⁰ investigated the impact of different co-stimulatory molecules on CAR-NK cell activity. They generated CD70-targeted CAR constructs by fusing the extracellular domain of CD27 with cytoplasmic domains from multiple co-stimulatory molecules. Despite the lack of endogenous CD28 expression in NK cells, CAR27-28ζ NK cells featuring CD28 co-stimulation exhibited superior long-term cytotoxicity and multifunctionality¹²⁰ (Figure 4). Corresponding phase I/II clinical trials are currently underway to assess the safety and efficacy of CAR27-28ζ NK cells in patients with CD70-positive hematologic and solid malignancies (NCT05092451, NCT05703854).

Despite encouraging progress, the clinical translation of CAR-NK therapies remains limited by several challenges, including poor gene transfer efficiency, limited in vivo persistence, and suboptimal trafficking to solid tumours.^{121, 122} Strategies such as IL-15 armouring, inhibitory receptor knockout, and metabolic reprogramming have shown promise, but improvements in manufacturing scalability and functional durability are still needed to enable broader clinical application.^{115, 123}

3.5 Combination therapy strategies

Although each of these therapies has made significant progress in its respective fields, the limitations of monotherapy remain prominent. Heterogeneity of tumour cells, development of drug resistance, immune escape, and loss or alteration of therapeutic targets often limit the durability and efficacy of therapy.¹²⁴ In addition, the systemic toxicity of certain therapies cannot be ignored, which further narrows the patient's treatment options. In this context, combination therapy strategies have become the optimal solution for antitumour therapy.^{125, 126}

Current combination therapy strategies targeting CD70 typically integrate traditional therapies with emerging modalities such as monoclonal antibodies or CAR-T/CAR-NK cell therapies. These combinations have been investigated in preclinical models and have shown potential to augment antitumour responses through multiple mechanisms, including upregulating CD70 antigen expression, enhancing the cytotoxicity of CD70-targeted immune cells, and remodelling the immunosuppressive tumour microenvironment to overcome resistance.¹²⁷

In AML, research has found that hypomethylating agents (HMAs) can upregulate CD70 expression in LSCs by reducing CD70 promoter methylation and enhance stem cell self-renewal and survival by activating the CD70/CD27 signalling pathway.¹²⁸ Preclinical studies have further shown that decitabine-induced CD70 upregulation sensitizes AML LSCs to cusatuzumab, with the combination leading to a marked reduction of functional LSCs in vitro and in vivo. Quantitative analysis using the Chou–Talalay method confirmed a synergistic interaction between the two agents.¹²⁹ Based on this mechanism, several clinical trials have evaluated the combination of cusatuzumab with azacitidine. A phase I/II clinical trial (NCT03030612) demonstrated promising activity, with 8 of 12 AML patients achieving CR, 2 complete remissions with incomplete blood count recovery, and 2 PR.¹²⁹ Although the trial was not designed as a comparative study, the observed response rate was described as favourable in relation to historical data on HMA monotherapy.¹²⁹ A subsequent phase I/II trial refined the dosing schedule of cusatuzumab (1, 3, 10, and 20 mg/kg) in combination with azacitidine, establishing 10 mg/kg as the recommended phase II dose.¹³⁰ Of the 38 participants enrolled 19 exhibited objective responses, including 14 with CR, supporting the therapeutic potential of this combinatorial approach.¹³⁰ In the CULMINATE randomized phase II clinical trial (NCT04023526), researchers compared the efficacy of two doses of cusatuzumab (10 vs. 20 mg/kg) combined with azacitidine. The results demonstrated that the CR rate (27%) in the 20 mg/kg group was significantly higher than in the 10 mg/kg group (12%), with superior overall efficacy.¹³¹ Besides, in another study involving six Japanese AML patients (NCT04241549), the combination therapy was found to be safe, but its efficacy was relatively limited, with no CR observed.¹³²

Moreover, Cheng et al.¹⁰⁸ found that the epigenetic modulators decitabine and chidamide were able to upregulate CD70 expression in AML cells, which in turn enhanced the targeted killing effect of nanobody-based nb70CAR-T cells. However, the precise regulatory mechanism of epigenetic modulators in promoting CAR-T therapy and their in vivo efficacy still needs further validation. Similarly, in chronic myeloid leukaemia, imatinib, a tyrosine kinase inhibitor (TKI), was found to upregulate CD70 expression on LSCs via epigenetic mechanisms, and combining anti-CD70 antibodies with TKIs effectively eliminated resistant LSCs by disrupting compensatory CD70/CD27-mediated Wnt signalling.¹³³ In solid tumours, a preclinical study has shown that low-dose chemotherapy agents such as docetaxel and cisplatin can further upregulate the expression of CD70 in NSCLC. Sequential treatment combining chemotherapy and cusatuzumab can enhance the antitumour activity of NK and T cells, thereby inhibiting tumour growth.¹³⁴

In addition to upregulating CD70 expression, multiple preclinical combination strategies have been investigated to enhance the efficacy of CAR-T/CAR-NK cells. These strategies act through different mechanisms, including boosting CAR cell persistence and activity, altering tumour cell susceptibility, and modulating the immunosuppressive microenvironment. Leick et al.¹³⁶ reported that azacitidine upregulated CD70 expression in AML cells and facilitated the expansion of CD8H&TM CD70 CAR-T cells modified with CD8 hinge and transmembrane domains, thereby enhancing antitumour efficacy.¹³⁵ Activation of the IL-21/IL-21R signalling pathway has been implicated in enhancing the sensitivity of LSCs to CD70 CAR-T cell therapy and cytarabine by modulating their differentiation and metabolic state, thereby contributing to enhanced therapeutic efficacy in combination settings.¹³⁷ Proteasome inhibitors such as bortezomib and carfilzomib have been shown to enhance the cytotoxic activity of CAR-NK cells against AML by downregulating HLA-I molecules expression and upregulating stress-associated proteins.¹³⁸ In RCC, Ji et al.¹³⁹ reported that PARP inhibitors enhanced the efficacy of CD70-targeted CAR-T therapy via activation of the cGAS-STING signalling cascade in tumour cells, resulting in elevated CXCL10 and CCL5 expression and promoting CAR-T cell infiltration and cytotoxicity. Additionally, in glioblastoma, oncolytic herpes simplex virus-1 (oHSV-1) was shown to synergize with CD70 CAR-T cells by promoting IFN-γ release, suppressing TGF-β1 expression, and modulating Tregs infiltration, thereby remodelling the immunosuppressive microenvironment and enhancing treatment responses.¹⁴⁰

CD70-targeted therapies have also been investigated in combination with ICIs. CD70 has been found to correlate with immune checkpoint gene expression across various cancers.^{25, 26} In pleural mesothelioma, co-expression of CD70 and PD-L1 was associated with poor prognosis and increased infiltration of FOXP3⁺ Tregs and CD163⁺ macrophages, supporting their cooperative role in immune evasion.¹⁴¹ In thyroid cancer, Ning et al.¹⁴² identified a mechanism of PD-1 resistance in which M2 macrophage-derived extracellular vesicles delivered miR-21-5p to suppress METTL3 expression, resulting in the demethylation and stabilization of CD70 mRNA. This upregulation of CD70 promoted Tregs accumulation and T cell exhaustion, contributing to an immunosuppressive tumour microenvironment. Blockade of CD70 with cusatuzumab reversed these effects and partially restored sensitivity to PD-1 inhibition.¹⁴² In NPC, a combination of cusatuzumab and the anti-PD-1 antibody nivolumab demonstrated superior tumour control compared with either monotherapy in patient-derived organoids and humanized mouse models.²⁹

These combination strategies, involving CD70-targeted therapies with chemotherapy, targeted agents, ICIs and other adjunctive approaches, have shown promise in remodelling the immunosuppressive tumour microenvironment and enhancing tumour-specific immune responses. While these strategies show therapeutic potential, most remain at the experimental stage, and clinical data remain limited. Furthermore, whether these combinations offer superior clinical efficacy over monotherapy has yet to be confirmed, and synergistic effects have not been systematically evaluated in most studies. These limitations underscore the need for mechanistic investigations and well-designed clinical trials to determine their true therapeutic value.