CD5 is a surface marker of T-cell malignancies and is expressed in approximately 80%-95% of T-ALL or T lymphoblastic lymphoma (TLL) (56, 57). Typically, CD5 is also expressed on mature peripheral blood T cells, thymocytes, and some B-cell lymphocytes in healthy tissues, which lead to fratricide of CAR-T cells (58, 59). In 2015, Mamonkin et al. reported that CD5 CAR-T cells only exhibit partial fratricide, following which these could be expanded in vitro. The expanded CAR-T cells also maintained killing efficacy in T-ALL/TLL tumor cell lines (including Jurkat, CCRF-CEM, MOLT4, Hut78, and SupT1) and primary T-ALL cells in vitro and Jurkat and CCRF-CEM cell lines in vivo. Further investigation found that the incomplete fratricide of CD5 CAR-T cells was due to the internalization of surface CD5 molecules after ligand binding which, in turn, downregulated CD5 expression on the normal T-cell surface (60). To prevent fratricide completely, in 2017, Raikar et al. genetically knocked out CD5 expression on the surface of T cells with CRISPR-Cas9 genome editing, and CD5-edited effector T cells overcame the challenge of self-activation and fratricide, which demonstrating the feasibility for CD5 CAR-T therapy in T-cell malignancies (61). To prolong the persistence of CAR-T cells in vivo, in 2018, Mamonkin et al. designed a doxycycline (Dox) controlled Tet-Off system to inhibit CAR expression to prevent fratricide of 4-1BB CD5 CAR instead of CD28 CD5 CAR. In this study, CD5 expression in CAR-T cells occurred after Dox withdrawal, leading to improved and prolonged antitumor ability against CD5+ T-ALL cell lines Jurkat and CCRF-CEM in vitro, as well as the Jurkat mouse model in vivo ( 62). To explore the feasibility of CD5 CAR-T therapy as a bridge to HSCT in the clinic, in 2019, Hill et al. treated 9 heavily treated patients with autologous CD5 CAR-T cells, and 3/9 patients achieved CR, 4/9 patients obtained an objective response, and 2/9 patients relapsed at 6 weeks and 7 months post-infusion, all side-effects were manageable. These results proved that CD5 CAR-T cells could allow ineligible patients to proceed to HSCT with safety and clinical response for R/R CD5+ malignancies (63). Except for the fratricide, T-cell aplasia is also an important challenge for CD5 CAR. In 2020, Wada et al. used alemtuzumab, which targets CD52 as an inducible safety switch, to remove CAR-T cells from systemic blood circulation without affecting the anti-tumor efficacy in the mouse model, thus avoiding T-cell aplasia after therapy (64). T-cell malignancies with CNS infiltration always have poor outcomes and limited treatment options, IL-15 could strengthen the anti-tumor response. In 2021, Feng et al. modified a CD5-IL-15/IL15sushi CAR which secretes an IL-15/IL-15 complex, to explore its clinical effectiveness against one refractory T-cell lymphoma patient with CNS infiltration. This patient obtained a rapid ablation of the CNS lymphoblast and lymphoma and was accompanied by brief and transient T-cell aplasia (65). In another study, Dai et al. manufactured a new biepitopic CAR with fully human heavy-chain variables FHV_H3 and FHV_H1, which could bind different epitopes of CD5. As such, CD5KO FHV_H3/V_H1 CAR-T cells showed prolonged and sustained efficacy against CD5+ T-ALL cell lines, such as Jurkat, CCRF-CEM, MOLT4, SupT1 in vitro, and CCRF-CEM in vivo, with moderate cytokine production, proved that this new CD5 CAR deserved more exploration (66). Due to recurrence occurring in some patients after CAR-T therapy which targets single antigen CD5 or CD7 (63, 67), in 2022, Dai et al. designed a bispecific CAR that targeted CD5 and CD7 with a fully human variable heavy chain (FHV_H). The results showed that fratricide-resistant FHV_H-derived CD5/CD7 bispecific CAR-T cells showed potent antitumor activity to Jurkat, CCRF-CEM, MOLT-4, and SUP-T1 cell lines in vitro and CCRF-CEM T-ALL mouse model in vivo, it provided the possibility to the populations with antigen heterogeneous (68). Pan et al. explored the safety and efficacy of donor-derived CD5 CAR-T cells in CD7-negative relapsed patients after CD7 CAR therapy. The data showed that all five patients achieved CR at 1 month and no dose-limiting toxicities occurred. Donor-derived CD5 CAR showed promising clinical safety and response in R/R T-ALL, but the evaluation of durable remission and functional immune system reconstitution needs longer follow-ups (69). Possible approaches to overcome the challenges such as fratricide and T-cell aplasia include internalization of surface CD5, CRISPR-Cas9, or Tet-Off system technologies that could be used to deal with complete fratricide without affecting the efficacy of CAR-T cells, while alemtuzumab could be used as a safety switch to deplete CAR-T cells after therapy to prevent T-cell aplasia. Taken together, CD5 is a promising target for CD5+ hematological malignancies, there are 4 clinical trials about CD5 CARs that are being recruited, as shown in Table 1 .

CD3 is a pan-T-cell surface antigen, expressed predominantly on all mature T-cells (70), however, the development of CAR-T targeting CD3 remains limited due to the complete fratricide of CAR-T cells. In 2018, Rasaiyaah et al. tried to disrupt endogenous TCRαβ/CD3 on T cells by adopting transcription activator-like effector nuclease (TALEN) to construct a CAR targeting CD3ϵ and CAR-T cells exhibited specific toxicity against the T-ALL cell line Jurkat in vitro and in vivo ( 71). However, since T-ALL and TLL cells derived from patients typically express cytoplasmic CD3 (cCD3) rather than membrane CD3 (mCD3) (72), the therapeutic effect of CAR targeting CD3 is limited in its clinical application. No clinical trials are currently documented in the literature.

CD7 is a 40-kD Ig superfamily member expressed on normal T and NK cells (73–75) and is expressed in over 95% of ALL and 30% of AML, as well as some lymphomas (56, 76–79). CAR-T cells targeting CD7 showed complete fratricide and could not be expanded, thereby resulting in limited studies of CAR targeting CD7 until 2017. Gomes-Silva et al. first knocked out CD7 expression on normal T cells by CRISPR–Cas9 pre-T-transfection so that CD7-knocked out CAR-T cells could be expanded, and these cells showed specific cytotoxicity to both CD7+ cell lines Jurkat, CCRF-CEM, MOLT-4, Hut78 and SupT1, also primary blasts, these provided a new insight for CD7 CAR (76). Png et al. reported a study that blocked CD7 in the ER/Golgi by using a protein expression blocker (PEBL) system, in which CD7 could not be expressed on the surface of T cells. This method alleviated fratricide without affecting the proliferation of CAR-T cells, while CD7 CAR-T cells showed strong antileukemia activity to CD7+ cell lines including Jurkat, CCRF-CEM, Loucy, MOLT4, and KG1a (80). Due to CD7 expression on both normal and malignant T cells, fratricide and malignant contamination occurred during CAR-T cell preparation. To solve both issues, allogeneic T cells may be another choice to prepare CAR-T cells. Then in 2018, Cooper et al. knocked out CD7 and the T-cell receptor alpha chain (TRAC) of T cells by CRISPR–Cas9, and CD7 CAR-T cells showed antileukemia efficacy against T-ALL cell lines MOLT-3, MOLT-4, HSB-2, and CCRF-CEM without graft-versus-host-disease (GVHD) (81). In 2020, Zhang et al. explored the antitumor activity and toxicity of CD7 CAR in clinical trials for the first time. The author reported that 2 of 3 enrolled R/R ALL/TLL patients obtained complete remission (CR) with minimal residual disease (MRD) negative at day 28 post-infusion. At the same time, cytokine release syndrome (CRS) occurred which was controllable. Nevertheless, this study showed the significant potential value for the clinical application of CD7 CAR (82). To address CRS after CAR-T cell therapy, in 2021, Li et al. treated 2 T-ALL patients with “off-the-shelf” allogeneic CD7 CAR-T cells combined with ruxolitinib. Strong CAR-T-cell expansion and rapid tumor cell clearance were detected after CAR-T cell infusion, and both patients achieved CR with MRD negative. While, both patients developed grade 3 CRS, but it was manageable with co-treatment of ruxolitinib (83). Yang et al. conducted a phase I clinical trial of CD7 CAR-T cell therapy for 14 R/R T-ALL. The data showed that 13/14 (92.9%) patients achieved MRD negative CR at day 28 post-infusion and all patients experienced mild CRS (grade<2) (84). For T-ALL patients, there may be not enough normal T cells for CAR-T cells preparation, Pan et al. used donor-derived CD7 CAR-T cells to explore the efficacy and safety in R/R T-ALL, 18/20 patients achieved CR with 7 patients proceeding to HSCT and with a manageable safety profile (67). Furthermore, CAR-T/NK targeting CD7 will clear CD7-positive normal T/NK cells in vivo, HSCT is usually required after CAR-T therapy; otherwise, the treatment causes severe immunodeficiency and potentially life-threatening infection by pathogenic microorganisms. Kim et al. knocked out CD7 in hematopoietic stem cells (HSCs) and found that these CD7-KO HSCs differentiated into CD7-negative T-cells and NK cells that exhibited effector functions after transplantation into mice. This study suggests that such an approach could resolve T-cell immunodeficiency caused by CD7-CAR therapy, altogether providing a potential new approach for the development of CD7 CAR (85). In 2022, Dai et al. reported a case that haploidentical CD7 CAR-T cells induced remission in an 11-year-old TP53 mutated R/R ETP-ALL/LBL patient, and grade 3 CRS and macrophage activation syndrome were observed but manageable (86). Due to the GVHD occurring in some patients after donor-derived CD7 CAR-T cell therapy, Zhao et al. treated five R/R T-ALL/LBL with autologous CAR-T cell therapy. The results showed that 4/5 patients achieved CR at day 30 post-infusion without neurotoxicity, GVHD, or infection (87). Due to 3-10% CD7- T cells exiting in peripheral blood, Lu et al. obtained naturally selected 7CAR (NS7CAR) T cells, which are a subtype of CD7 negative T-cells that survived from fratricide. The NS7CAR T cells comprised a higher proportion of CAR+ cells and CD8+ central memory T cells while maintaining similar therapeutic activity to cell line CCRF-CEM in vitro and in vivo when compared with CD7 knocked-out 7CAR T cells. Most importantly, a study of 14 R/R T-ALL and 6 T-LBL patients who received NS7CAR T therapy found that 19/20 patients achieved MRD CR in bone marrow at day 28, all with manageable side effects (88). In another study, Freiwan et al. sorted CD7- T cells from PBMC before transducing them with CD7 CAR, the CD7- CAR-T cells contained more CD4+ memory phenotype and have a robust antitumor activity to CD7+ cell lines CCRF-CEM and MOLT3 in vitro and eliminated CCRF-CEM cells in the mouse model, as well as bypass fratricide (89). Li et al. combined donor-derived CD7 CAR-T therapy with allogeneic HSCT for a 3-year-old hepatitis B-positive T-ALL patient. The patient had CR at seven months post-infusion and the copy number of hepatitis B virus continuously decreased during treatment (90). Hu et al. resisted fratricide, GVHD, and allogeneic rejection in healthy donor-derived CD7 CAR (RD13-01) by genetic modifications. Twelve patients were recruited in the phase I clinical trial, and the data showed that 81.8% of patients achieved objective responses and 63.6% of patients received CR with no dose-limiting toxicity, GVHD, neurotoxicity, or severe CRS (91). CD7 CAR showed satisfied efficacy and safety in various clinical trials, gene editing technologies such as CRISPR-Cas9, TALEN, or PEBL, and recently reported natural selection could be applied to overcome fratricide, while HSCT could be applied in combination with CAR to address potential T-cell aplasia, and CRS could be controlled by ruxolitinib. Taken together, CD7 CAR is a promising therapeutic target for R/R T-cell malignancies. Currently, there are 23 clinical trials of CD7 CAR recruiting to explore its therapeutic effects in T-cell malignancies, as shown in Table 2 .

In addition to CARs targeting T-cell pan-antigens, some antigens are only expressed on a subset of T cells, which can avoid complete fratricide during CAR-T preparation and T-cell immunodeficiency during therapy.

CD4 is expressed in most TLLs and some T-ALLs and its expression is restricted to the hematopoietic compartment but it is not expressed in HSCs. As the target of CAR, it can potentially reduce the off-target side effects for non-hematological tissues. But the persistence of CD4 CAR T cells after removal of tumor cells can lead to aplasia of CD4 positive T cells and cause HIV/AIDS-like syndrome. In 2016, Ma et al. utilized CAMPATH (alemtuzumab) as a natural safety switch to deplete CD4 CAR T cells post-therapy in mice. They found that this approach was effective to kill Jurkat cells in vitro and in vivo with minimal toxic side effects from loss of CD4+ T-cells (92). CD4 is expressed on a subset of normal T cells, which could prevent complete fratricide and be used to kill CD4 high-expressed malignant T cells. Three CD4 CARs are currently being recruited to explore clinical efficacy in T-ALL, as shown in Table 3 .

The expression of CD1a is largely restricted to developing cortical thymocytes, and neither CD34+ progenitor cells nor T-cells are expressed during ontogeny. In contrast, however, CD1a is only expressed in cortical T-ALL (coT-ALL) and its expression persists in relapsed patients. Diego et al. found that CD1a CAR-T cells are fratricide-resistant and showed good efficacy against CD1a+ Jurkat and MOLT4 cell lines and primary coT-ALL cells in vitro and Jurkat T-ALL mouse models in vivo ( 93). Although CD1a CAR could be used to treat coT-ALL, some limitations exist to CD1a CAR in clinical use, such as that only a minority of cases of T-ALL express CD1a. Furthermore, CD1a is associated with relatively a favorable prognosis, and CD1a+ patients rarely relapsed or are refractory to treatment.

Others, such as CCR4 (C–C chemokine receptor type 4) and TRBC (T-cell receptor beta constant), are also expressed in some T-ALL and can be potential targets for T-ALL immunotherapy (94, 95).

One potential target for AML is CD123, an IL-3 receptor alpha chain that acts as a high-affinity receptor for stem cell factor (SCF) and is expressed at low levels in early hematopoietic cells, such as hematopoietic stem/progenitor cells (HSPCs) (96). CD123 is expressed in approximately 97% of AML patients and is overexpressed in 45% of patients, and at low levels by HSPCs, monocytes, a subset of dendritic cells, and endothelial cells (97–99). As early as 2013, Tettamanti et al. designed CD123 CAR-infected cytokine-induced killer cells (CIKs). These CIKs showed efficacy against CD123+ THP1 cell line and primary AML blasts, with minimal effects on healthy monocytes and endothelial cells with low expression of CD123, which proves the feasibility of CD123 as a therapeutic target (100). In 2014, Mardiros et al. demonstrated that CD123 CAR-T cells were effective in eliminating CD123+ LCL, KG-1a cell lines, and primary AML blasts without eliminating granulocyte/macrophage and erythroid colony formation in vitro. Notably, patients-derived CAR-T cells could eliminate autologous AML blasts in vitro ( 101). Gill et al. found that CD123 CAR was efficient in clearing human primary leukemia cells, meanwhile inducing myeloablation in xenograft mouse models, which suggests that CD123 CAR-based myeloablation may be used to bridge HSCT (102). Due to CD123 and CD33, both are expressed on AML cells and normal HSPCs, Pizzitola et al. modified CKI cells with the CD123 and CD33 CAR to compare their efficacy and safety. Both these CAR-T cells efficiently eliminated primary human AML KG-1a cells in mice, but only CD123 CAR showed limited killing efficacy on normal HSPCs compared to CD33 CAR (103). In 2015, Luio et al. first treated one relapsed AML-M2 (FLT3/ITD+), male patient with apoptosis-inducible CD123 CAR, and the patient subsequently achieved partial remission within 20 days. Although CRS occurred on day 4, it was effectively controlled with a single dose of Tocilizumab (104). To address myeloablation after CD123 CAR-T therapy in AML, Tasian et al. compared three CAR-T-cell clearance strategies: (1) transiently active anti-CD123 messenger RNA electroporated CAR T cells (RNA-CART123); (2) T-cell clearance with alemtuzumab after CD123 CAR-T-cell therapy; and (3) T cell ablation with rituximab to CD20-coexpressing CART123(CART123-CD20) after therapy. The author found that CD123 CAR-T cells showed strong antitumor activity to MOLM14 cell line and primary AML blasts in mouse xenograft models, and all these three approaches could efficiently deplete CAR-T cells without affecting antileukemic effects. Notably, the ablation of CAR-T cells allowed subsequent HSCT rescue in normal hematopoiesis xenograft models (105). Thus, strategies for posttreatment CAR-T cell clearance may be effective to alleviate detrimental side effects, such as myeloablation. Further to this area of investigation, Arcangeli et al. designed a rational mutation in the anti-CD123 CAR antigen binding domain to reduce the binding affinity of CAR, which could minimize the toxicity and side effects against normal tissues with low expression of CD123 without affecting the antileukemic activity to tumor cells; altogether demonstrating the manageable safety of CD123 CAR in AML (106). The same year (2017), Budde et al. reported in an abstract form that a Phase I dose escalation clinical trial, which evaluated the efficacy and safety of CD123 CAR in 6 patients with refractory AML after treatment with autologous HSCT (alloHSCT). In this study, 4 of 6 patients achieved CR, and 2 patients reported reduced blast counts but did not achieve remission after a single or second infusion of CAR-T cells. Notably, all side effects were reversible and controllable (107). Meanwhile, to ablate CAR-T cells after therapy in patients, Cummins et al. evaluated the safety of CD123 CAR in R/R AML patients with mRNA electroporate technology. CART123 cells did not expand successfully in vivo, and all 5 patients treated with mRNA CART123 eventually developed clinical progression (108). To improve the efficacy and persistence of CAR-T cells in vivo, in 2018, Mu et al. developed CD123 CAR-expressing IL-15. The results showed that genetically engineered CD123 CAR improved the anti-AML activity against CD123+ cell lines in vitro and eliminated primary AML cells in vivo ( 109). Loff et al. redirected CD123 CAR-T cells using a switch-controllable universal CAR T platform (UniCAR) based on two major elements: a non-reactive inducible CAR and a soluble targeting module (TM) enabling UniCAR-T reactivity in an antigen-specific manner. UniCAR T 123 exhibited potent cytotoxic activity against patient derived CD123+ leukemia cells in vitro and in vivo. Notably, in this study, the activation, cytolytic activity, and cytokine release profiles for UniCAR T123 were all tightly controlled. Compared with traditional CD123 CAR-T cells, UNICAR T 123 cells can additionally distinguish malignant leukemia cells with high CD123 expression from healthy tissues with low CD123 expression, features that further improve the safety of CD123 CAR (110). A phase I clinical trial of UNICAR-CD123-CAR is currently in progress (111). In 2019, Qin et al. developed a simple and highly selective D-domain, which was derived from de novo-designed α-helical bundle-α3D, to target CD123 for a unique CD123 CAR. The work revealed that CD123 CAR composed of D-domain mediated efficiently mediated T-cell activation and cytotoxicity to MOLM14, IM9, KG-1a, and NALM6 cell lines, and induced complete and durable remission in two AML xenograft mouse models. This work supports the development of multifunctional CARs through such an approach (112). Yao et al. used donor-derived CD123 CAR T cells as a conditioning regimen for haploidentical HSCT (haploid-HSCT) in a patient with FUS-ERG+ AML and found that CD123 CAR-T cells reduced chemotherapy-resistance blasts without affecting donor chimerism and myeloid implantation (113). To enhance the anti-tumor function of CD123 CAR-T cells, in 2020, You et al. combined decitabine treatment with CD123 CAR for AML. The results showed that decitabine enhances the anti-leukemia efficacy of CD123 CAR-T cells to THP1 cell line in vitro and in vivo, alongside CD123 CAR-T cells differentiate into naive and memory phenotypes (114). In 2021, UniCAR-T-123 with the targeting module TM123 was used to treat 3 R/R AML patients. 1 patient showed partial remission and 2 patients showed CR and adverse events were generally mild (115). This clinical trial is still ongoing. To minimize the side effect on normal hematopoietic progenitor cells and prolong the persistence of CD123 CAR-T cells, Khawanky et al. found that demethylating therapy could increase the CD123 expression on leukemia cells and increase CTLA4- CD123 CAR-T cell proportion and showed superior cytotoxicity against AML cells, accompanied by higher TNFα production in leukemia-bearing mice (116). Due to CD123 could distinguish HSC from leukemia stem cells(LSCs), to eliminate LSCs and preserve normal HSC, in 2022, TALEN gene-editing technology was used to produce a TCRαβ negative allogeneic CD123 CAR (UCART123), which preferentially eliminates primary AML than normal cells with modest toxicity in vitro ( 117). While CAR targeting CD123 has shown efficacy and safety in preclinical and some clinical trials, but myeloablation occur after CAR-T therapy. Therefore, some strategies such as transiently active messenger RNA for CAR, alemtuzumab or rituximab treatment, and UniCAR may be utilized to ablate CAR-T cells in vitro and in vivo. The depletion of CAR-T cells after therapy requires further investigation in clinical studies. Nevertheless, the safety and efficacy profiles of CAR therapy targeting CD123 are reflected in the finding that 7 clinical trials are recruiting to explore its feasibility to treat AML, as shown in Table 4 .

CD33 is a transmembrane receptor of the sialic-acid-binding immunoglobulin-like lectin (SIGLEC) family and is expressed on myeloid cells ranging from progenitors to well-differentiated cells, including neutral granulocytes, monocytes, and tissue-resident macrophages (118). However, in pathological states, CD33 is expressed in approximately 80-90% of AML patients and may also be expressed on LSCs (119–122). In 2010, to improve the effector functions of CIK cells, Marin et al. transduced anti-CD33-ζ and anti-CD33-CD28-OX40-ζ CARs into CIK cells and found that the CD33 CAR enhanced the anti-leukemic functions of CIK cells against HL60 and KG-1a cell lines and primary AML blasts (123). In 2012, to decrease tumor escape, Dutour et al. transduced human Epstein Barr virus (EBV)-specific cytotoxic T cells with CD33 CAR, and the CAR-T cells displayed EBV and HLA-unrestricted bispecificity in vitro and anti-AML tumor activity in CD33+ human AML-bearing mice without irreversibly disrupting the formation of CD34(+) hematopoietic progenitor clones (124). In 2014, Pizzitola compared the cytotoxicity of CD123 and CD33 CAR on AML, and find that no difference in anti-leukemic activity, yet CD33 CAR appeared to have stronger cytotoxicity on normal HSPCs compared with CD123 CAR (103). In 2015, Wang et al. reported an autologous CD33 CAR in one R/R AML patient, with no uncontrollable clinical toxicities, but with subsequent disease progression at 9 weeks post-T-cells infusion (125). Kenderian and colleagues used the scFv of gemtuzumab ozogamicin (clone My96) to develop a CD33 CAR (CART33). CART33 exhibited significant cytotoxicity against the MOLM14 cell line in vitro and eradication of leukemia in AML xenograft, yet a reduction of myeloid progenitors in xenograft models was also observed. Thus, the author prepared transient CART33 cells by expressing modified mRNA, which exhibited potent but self-limited activity against AML (126). To minimize the myelosuppression after CAR-T therapy, in 2016, Minagawa et al. designed Caspase9-CAR CD33T cells inducibly selected by ΔCD19, which could specifically lyse CD33+ MV4-11 tumor cells and primary leukemic blasts in vitro, following which CAR-T cells were largely eliminated by suicide gene activation (127). In 2018, Kim et al. proved that CD33 KO HSPC showed normal engraftment and differentiation in the mouse model, and autologous rhesus macaques CD33 KO HSPC showing normal myeloid function in vivo. Most importantly, CD33 CAR-T cells showed efficient elimination of leukemia, while CD33-deficient cells were spared without myelotoxicity, as observed in human xenograft models (128). Li et al. demonstrated that 4-1BB as a costimulator could endow CAR-T cells with increased central memory and prolonged survival for maximum efficacy, as compared to CD28, and both CD28 and 4-1BB (129). Due to PI3K pathway being activated in CD33 CAR-T cells, to increase the persistence of CAR-T cells in vivo, Zhang et al. used PI3K inhibitors to modulate the differentiation of CD33 CAR-T cells. The data showed that it maintained CAR-T cells at a lower differentiated state without affecting CAR-T-cell expansion (130). Schneider et al. constructed a humanized CAR33VH CAR, comprised of a human heavy-chain variable fragment, that exhibited antitumor activity to CD33 high expressed MOLM14 and HL60 cell lines and eliminated tumors in a MOLT-14 AML mouse model (131). In 2021, Tambaro and colleagues conducted a Phase I clinical trial to evaluate the efficacy and safety of CD33 CAR-T in R/R AML patients. It was reported that three patients received one low dose of CAR-T cells, and biologic activity was observed by associated symptoms and increased cytokine levels, however, an anti-leukemic response was not documented, and no dose-limiting toxicities were observed (132). To explore the impact of costimulatory domains on the CD33 CAR-T cells, Qin et al. developed six CD33-targeted CARs with one of three scFv of clinically tested CD33 antibodies, paired with CD28 or 4-1BB costimulatory domains. These six CARs exhibited cytotoxicity against CD33+ AML cell lines in vitro and in vivo and showed strong anti-leukemia activity against patient-derived xenograft (PDX) derived from pediatric AML patients. Furthermore, CD28-based CD33 CAR-T cells exhibited superior anti-leukemia compared with 4-1BB, and the safety and efficacy in patients were evaluated in a phase I clinical trial (133). In 2022, Liu et al. proved that the third generation CD33 CAR-T showed stronger vitality, proliferation ability, and stronger cytotoxicity than the second generation CAR. Notably, the third generation CD33 CAR-T preferentially killed leukemia cells while sparing CD33-deficient HSPCs (134). CD33 CAR showed efficient anti-AML activity in vitro, different costimulators, different generation CAR structures, and PI3K inhibitors may affect the anti-tumor activity, proliferation, and persistence of CD33 CAR-T cells. The efficacy of CD33 CAR in patients did not show promising results in some previous studies, more efforts are needed to improve the anti-tumor response of CD33 CAR. 6 clinical trials are recruited to explore the efficacy and safety of CD33 CAR in AML patients, as shown in Table 5 .

C-type lectin-like molecule-1 (CLL-1) is reportedly expressed in more than 80% of AML blasts, and LSCs. Its expression is restricted to the myeloid lineage and absent in normal CD34+CD38- HSCs. Significantly, CLL1 is present on a small subset of chemotherapy-resistant LSCs (135–137), suggesting that it may be a potential target for therapeutic intervention. In 2017, Tashiro et al. first designed a CLL1 CAR with specific killing efficacy on CLL1+ HL60 and THP1 cell lines and primary AML blasts in vitro, as well as exhibited anti-leukemia activity in an HL60-AML xenograft mouse model. Notably, CLL1 CAR-T cells eliminated mature normal myeloid cells yet selectively spared healthy HSCs, to allow immune recovery after therapy (138). Laborda and colleagues designed an scFv-based CLL1 CAR that showed cytotoxic activity against HL60 and MOLM14 AML cell lines and patient-derived AML blasts in vitro, as well as clearance against tumor cells in mouse xenografts, while without damaging healthy HSCs. Compared with CAR designs utilizing CD8 as a hinge domain, IgG4 induced higher cytotoxicity in cell lines (139). In 2018, Wang et al. constructed a new CLL1 CAR using scFv from C57BL/6 mouse-derived CLL antibody, which has good anti-leukemia activity to U937 cell line and primary AML blasts in vitro and eliminated human AML in xenograft mouse models without targeting normal HSCs (140). Liu et al. first explored the safety and efficacy of CLL1 CAR in patient. A compound CAR targeting CLL1 and CD33 was constructed and allennimumab was used to clear CAR-T cells after tumor eradication. A 6-year-old patient with a complex karyotype of FLT3-ITD mutation received two split doses of CAR-T cells and achieved CR on day 19, followed by HSCT (141). In 2019, Atilla and colleagues explored the impact of different combinations of spacers, transmembrane, and intracellular signaling domains to CLL-1 CAR. By comparing their proliferation, functional persistence, and antitumor activity in vitro and in vivo, the data showed that CD28z CAR with a short hinge region or with a CD8 intracellular domain is better than others for CLL1 CAR (142). In 2020, Zhang et al. reported the case of one 10-year-old patient with secondary AML treated with CLL CAR-T cells. This patient finally achieved morphological, immunophenotypic, and molecular CR over 10 months, which provides a new treatment option for secondary AML (143). Atilla et al. demonstrated that CLL-1 CAR T cells with additional transgenic IL15 supplementation, and combined with a TNFα blocker antibody as well as activation of caspase-9 control switcher increased expansion, persistence, and anti-leukemia of CLL1 CAR-T cells in PDX and HL60 xenograft mouse models while avoided excessive cytokine production (144). PD-1 expression was increased after the activation of CAR-T cells and caused T cells exhaustion, to evaluate the efficacy of combination of PD-1 silencing and CLL-1 CAR-T therapy, in 2021, Lin et al. designed a PD-1-silenced CLL-1 CAR and reported that PD-1 silencing enhanced the cytotoxicity of CLL-1 CAR (145). Similar to this, Zhang et al. designed a CLL-1 CAR based on the apoptosis-inducing gene FKBP-Caspase9 and evaluated the efficacy and safety in four R/R AML patients. The author reported that three patients achieved CR with MRD negativity, while the fourth survived for 5 months, with manageable side effects (146). CLL1 CAR showed promising anti-tumor efficiency in pre-clinical experiments and anti-AML response in AML patients and selectively spared normal HSCs. There are 8 clinical trials for CLL-1 CAR currently recruiting, as shown in Table 6 .

CD70 is a tumor necrosis factor (TNF) receptor ligand and is proven to be absent in normal tissues and hematopoietic cells, but highly expressed on most AML blasts and AML stem/progenitor cells (147, 148). In 2017, Riether et al. reported that CD70/CD27 signaling promotes blast stemness, and blocking CD70/CD27 by mAb could prolong survival in murine AML xenografts, representing that CD70/CD27 is a promising therapeutic strategy for AML (149). CD19 CAR-T cell therapy has presented revolutionary progression in CD19+ hematological malignancies, but some patients reccurence due to the exiting of CD19- tumor cells (150). Therefore, it is necessary to develop alternative antigens to avoid antigen escape. In 2021, Deng et al. designed L/H and H/L svFv-based CD70 CAR, truncated CD27-based CD70 CAR and anti-CD19 CAR as controls. The results showed that anti-CD70 (H/L) effectively killed CD19+ and CD19- Raji cells in vitro and in NSG xenograft mouse models, altogether providing a new therapeutic option for patients who have CD19- recurrence (151). Sauer et al. compared a panel of scFv-based CD70 CARs with the same scFv and different size and flexibility of the extracellular spacer, different transmembrane, and different costimulatory domains to CD27-based CD70 CAR. The results showed that the ligand CD27-based CD70 CAR presents superior proliferation and antitumor activity against AML cell lines Molm-13, THP-1, and IMS-M2 in vitro and Molm-13 AML mouse xenografts and primary AML in vivo (152). In 2022, Leick et al. used azacitidine to increase antigen density of CD7 in tumor cells and designed a CD8 hinger and transmembrane-modified CD27-based CD70 CAR to mitigate cleavage of the extracellular portion of CD27, altogether could enhance avidity and expansion of CD70 CAR-T cells and lead to more potent activity in vivo (153). The safety and efficacy of CD70 CAR in patients need more exploration in the clinical trial. There are only one clinical trial about CD70 CAR-T in AML currently recruiting as shown in Table 7 .

In addition, some antigens such as CD38, FLT3 et al., can be potential targets for AML CAR-T therapy. The recruiting clinical trials are shown in Table 8 .