The first clinical results regarding CAR-T therapy for AML are those targeting the Lewis Y (LeY) carbohydrate antigen, which is associated with the Lewis blood group system [9]. LeY is a universal antigen that is expressed in various tumors, including AML and solid tumors. In this study, infused LeY CAR-T cells were detected in peripheral blood for up to three months, with CAR-T cells infiltrating tumor sites (skin lesions), and no Grade 3 or higher adverse events were observed. However, although modest clinical responses were observed in 3 of 4 patients including cytogenetic remission, reduction in peripheral blood blasts, all three patients eventually relapsed with leukemic blast expressing LeY at comparable levels as compared to those before the treatment. As a result, the clinical development of LeY CAR-T cells for AML was discontinued. When targeting universal antigens, CAR-T cells are less likely to cause severe hematologic toxicity, as described above. However, this approach may limit the effectiveness of CAR-T cells as these antigens are not essential for AML cell survival and proliferation.

CD33 (Siglec-3) is a sialic acid receptor that is extensively found on myeloid and progenitor cells in the bone marrow and is expressed in 96% of AML cases. CD33 have been a representative target antigen for AML, and indeed an antibody–drug conjugate, gemtuzumab ozogamicin, is already approved.

In 2015, Chinese group reported the first adult patient who was treated with CD33 CAR-T cells [10] (Table 1). The patient received single dose of CD33 CAR-T cells and then developed CRS with worsening pancytopenia. Despite of a transient decrease in bone marrow blasts, the patient did not achieve CR, and the patient’s condition rapidly deteriorated. CAR-T cells were detected in the patient’s peripheral blood and bone marrow for upto 2 months after administration, and the blasts remained CD33 positive. Several factors have been proposed as explanations for the lack of efficacy, including heterogeneous CD33 expression on AML cells and the presence of CD33-positive subpopulations in approximately 15% of administered CAR-T cells, potentially leading to T cell exhaustion and fratricide (Fig. 1). Another possible explanation is that the single-chain variable fragment (scFv) recognizes only one of the two CD33 splicing variants [10].

In 2021, the MD Anderson Cancer Center group published the results of a study involving ten adult patients in the first cohort [11]. Two patients were unable to undergo leukapheresis due to disease progression, and eight patients received poor apheresis material with high blast ratios. Even among the four patients whose CAR-T cells were successfully manufactured; one did not proceed to administration due to disease progression. Eventually, three patients received CAR-T cells, but none of them achieved disease control and they died prematurely [11] (Table 1). The researchers then attempted to improve the efficacy of CD33 CAR-T cells by incorporating membrane-bound IL-15 into the CAR construct (PRGN-3006), which demonstrated significant antitumor effects in a preclinical model. Twenty-four patients were enrolled and treated with PRGN-3006 either without (cohort 1) or with (cohort 2) lymphodepleting (LD) chemotherapy. Three of 10 patients in cohort 1 and 7 of 14 patients in cohort 2 developed CRS, with 1 patient experiencing Grade 3 CRS in cohort 2. However, objective responses such as CR with incomplete hematologic recovery (CRi), CR with partial hematologic recovery (CRh), and partial remission (PR) were observed in only 3 patients in cohort 2 [12], indicating that the incorporation of membrane IL-15 is insufficient to induce potent efficacy of CD33 CAR-T cells.

Meanwhile, interim results of a Phase I/II multicenter trial of CD33 CAR-T cells were reported in 2023 [13]. Of the 24 enrolled subjects, CD33 CAR-T products were successfully manufactured in 23 subjects, and 4 dose levels (DL1–4, 0.3–10 × 106/kg) of CAR-T cells were infused after LD chemotherapy. CRS was observed in 13 (68%) patients with 4 patients experiencing Grade ≥ 3. Two patients in the highest cell dose (DL4) cohort achieved CR. These two patients also achieved MRD-negative CR with myeloid aplasia. Again, although manufacturing of CD33 CAR-T cells was feasible, clinical efficacy remained limited (Table 1).

CD33 has been a representative target antigen for many research groups over the years for AML-targeted CAR-T therapy. Initially, the main concern was on on-target/off-tumor toxicity, such as myelosuppression and veno-occlusive disease or sinusoidal obstruction syndrome. However, evidence from recent clinical studies indicates that manufacturing failure and poor efficacy are more serious concerns. It should be noted that CD33 is expressed in ex vivo expanded CAR-T cells during the manufacturing process, which raises concerns about its suitability as a target antigen for CAR-T cells.

NKG2D is one of the natural killer (NK) cell-activating receptor that binds to its ligands (NKG2DL), such as MICA, MICB, and ULBP1-6. NKG2DL is activated by stressors, such as infection, heat, and tumorigenesis. While normal tissue cells generally do not express NKG2DL [14], their overexpression has been reported in various cancer types including AML and solid tumors.

Celyad Oncology, a Belgium-based company, developed CAR-T cells targeting NKG2DL (CYAD-01) with a structure that includes the NKG2D extracellular region and the CD3ζ intracellular signaling domain. Although this CAR lacks costimulatory molecules, such as CD28 or 4-1BB, it can form a complex with DAP10 that contains the intracellular signaling domain of the introduced NKG2D, thus allowing the transmission of NK cell-like activation signals to T cells (Fig. 2). In their first human dose-escalation trial conducted in the United States, 12 adult patients (7 with AML and 5 with multiple myeloma) received a single infusion of CYAD-01 without LD chemotherapy [15]. However, the efficacy was limited, with only one patient with AML showing PR (Table 1). Consequently, a Phase I clinical trial for AML, myelodysplastic syndrome (MDS), and multiple myeloma was conducted. Three infusions of CYAD-01 were given without LD chemotherapy in three escalating dose cohorts (3 × 108, 1 × 109, and 3 × 109) every two weeks (THINK trial) [16]. Seven of the 25 patients had manufacturing failure, and 2 were excluded during screening. Finally, CYAD-01 was administered to 16 patients. Seven patients had Grade 3 or 4 toxicities (44%), and five patients (31%) had CRS of Grade ≥ 3. Although one patient in cohort 3 experienced dose-limiting toxicity (DLT), there were no deaths as a result of adverse events, and the maximum tolerated dose was not reached. Three (25%) of the 12 evaluable patients with AML or MDS achieved hematological CR [16, 17]. Among the responders, two AML patients underwent HSCT and maintained CR for 5 and 61 months, respectively (Table 1). Although this trial demonstrated tolerability, the efficacy of NKG2D ligand-targeted CAR-T cells remained unsatisfactory despite multiple infusions, most likely due to a lack of LD chemotherapy and NKG2DL overexpression on the CAR-T cell surface during manufacturing. As a result, Celyad Oncology launched additional trials, such as the DEPLETHINK trial, which combines LD chemotherapy, and the CYCLE-1 trial, which uses CYAD-02 CAR-T cells with shRNA to suppress MICA and MICB expression in CAR-T cells to prevent fratricide in ex vivo manufacturing [14].

Molecular basis of CAR-T therapy for AML. First-generation CAR lacks costimulatory molecules, whereas second-generation CAR harbors costimulatory molecules, such as CD28, 41BB, etc. Switchable CAR-T can recognize TM123 molecules that function as an adaptor and bind to CD123+ AML cells. Although clinically available NKG2D CAR lacks costimulatory molecules, they can form a complex with DAP10 that works as a signaling domain to stimulate T cells

More recently, Sauter et al. published the results of a Phase 1 study of off-the-shelf allogeneic CAR-NK cells targeting NKG2DL for AML. Six patients with r/r AML were enrolled in the study and received 3 doses of allogeneic CAR-NK cells (day 0, 7, and 14). They used a novel LD chemotherapy regimen that included fludarabine and high-dose cytarabine (2 g/m2). Surprisingly, four out of six patients achieved CR/CRi, with three patients having no detectable MRD by flow cytometry. Furthermore, no CRS or ICANS were found in this study. The PK study showed the persistence of CAR-NK cells for upto 3 weeks [18] (Table 1). Specific lymphodepleting chemotherapy as well as donor-derived NK cells may be associated with the favorable outcomes in this study.

CD123 is an interleukin-3 receptor alpha chain that is expressed in 78%–89% of AML cases and also prevalent in AML leukemic stem cells [19]. CD123 has been identified as a promising target for immunotherapy, such as CAR-T cells, bispecific antibodies, and antibody–drug conjugates. There have been several studies and clinical trials that explore CD123 as a CAR-T cell therapy target; however, controlling on-target/off-tumor toxicity is still a major challenge as CD123 is widely expressed in normal cells, including multipotent progenitors, myeloid progenitors, lymphoid progenitors, and immature blood cells.

The first clinical application of CD123 CAR-T cells was reported in 2015 in China [20]. An adult patient with AML received single infusion of fourth-generation CD123 CAR-T cells with three costimulatory molecules (CD28, 4-1BB, and CD27) as a signaling domain and an inducible caspase-9 (iC9) safety switch. However, despite the severe CRS observed, the therapeutic effect was limited and only resulted in a modest reduction in peripheral blasts (Table 1).

A US-based company, Mustang Bio, reported preliminary results of CD123 CAR-T cell therapy (either autologous or allogeneic) in six adult patients [21]. Two of the six patients achieved morphologic CR after CAR-T infusion. Another patient who was in CR at the time of CAR-T cell administration remained in CR. Notably, no severe GVHD, cytopenia, or DLT was observed [21] (Table 1).

Yao et al. [22] published a case report of relapsed patients with AML 3 months post-HSCT who were treated with donor-derived CD123 CAR-T cells with 41BB as a conditioning regimen for 2nd haploidentical HSCT. The patient was given a low-intensity conditioning protocol, followed by 9 × 107 CART123 cells. The patient developed Grade 4 CRS within 24 h of the infusion, which was treated with tolicizumab and methylprednisolone. Although the authors described the reduced percentages of bone marrow blasts (40.8% > 10.3%) and observed 2 weeks after CART123 infusion, the effects of prior conditioning cannot be ruled out. While the patient engrafted and achieved CRi after the 2nd HSCT, he succumbed to Grade 4 GVHD on day 56 [22].

St. Jude Children’s Research Hospital researchers conducted a Phase I trial of CD123 CAR-T cells in patients upto 21 years old as a bridging therapy for HSCT [23]. They presented preliminary findings from five patients in the first and second-dose cohorts. One patient in the second-dose cohort achieved CR, while the other one had a reduction in blast cells. A patient who achieved CR with extramedullary disease received a second CAR-T cell infusion after the CAR-T cells disappeared, resulting in another short-term CR [23].

A French biotechnology company, Cellectis, has initiated a phase I clinical trial for their allogeneic CD123 CAR-T cells (UCART123), which use genome editing (TALEN technology) to knock out the T cell receptor alpha chain and CD52 to prevent GVHD and rejection [24]. Patients with blastic plasmacytoid dendritic cell neoplasm (BPDCN) and adults with AML were enrolled in the initial trials. However, two patients experienced severe Grade 4 or higher adverse events, causing both trials to be halted. The first patient with BPDCN died 9 days after infusion due to CRS and capillary leak syndrome. The second patient with AML developed Grade 3 CRS and Grade 4 capillary leak syndrome. These findings indicated that the low CD123 expression in vascular endothelial cells may play a role in the development of capillary leak syndrome [24]. Subsequently, the trial was restarted with significant protocol changes, including lowering the age limit to 65 years, excluding BPDCN, lowering the cell dose, and reducing LD chemotherapy. Up until July 2022, 16 patients received UCART123v1.2 infusions in combination with either fludarabine and cyclophosphamide (FC arm) or FC plus alemtuzumab (FCA arm) before LD chemotherapy. The first eight patients in the FC arm received four dose levels (DL1, DL2, DL2i, and DL3) of UCART123v1.2 following LD chemotherapy. Because three patients experienced DLT (two with Grade 4 CRS, and one with Grade 3 ICANS) at higher dose levels (DL2i, 1.5 × 106/kg and DL3, 3.03 × 106/kg) in the FC arm, the next eight patients in the FCA arm received a fixed cell dose of 6.25 × 105/kg (DL1). Despite this, another patient died as a result of CRS in the FCA arm. Overall, 15 of the 16 patients developed CRS, and four had DLTs with Grade 3 or higher CRS (two Grade 4 and one Grade 5) [24]. Regarding efficacy, four of the 16 patients demonstrated the best overall responses, including two with stable disease, one with morphologic remission, and one with MRD-negative CR. One patient with CR had MRD negativity for 8 months. These findings suggest that UCART123v1.2 may be effective for AML, but they also raise concerns about severe CRS.

To reduce on-target/off-tumor toxicity against CD123, several groups incorporated safety switches (iCaspase9, CD20, etc.) within CAR constructs that enable CAR-T cells to be eliminated in case of adverse effects, as previously reviewed [25]. Meanwhile, by employing two unique products, AvenCell Europe developed a new technology of adapter-type CAR-T cells called the “UniCAR-T system” (Fig. 2). The target module (TM) is a soluble adapter with a short half-life that contains a scFv specific to the target antigen (CD123) as well as a certain peptide sequence. Then, they used Uni CAR-T-123 cells, which can specifically bind to TM by recognizing certain peptide sequences [26]. As these CAR-T cells can only bind to TM and not to any human-surface antigens, this system can control CAR-T function by systemically administering TM for the duration of the treatment. In 2021, this group published preliminary results of three cases who were treated with the UniCAR-T system from their first-in-human study [27]. TM123 (0.5–1.0 mg/day) was administered continuously from days 0 to 24, with 100–250 × 106 UniCAR-T 123-specific CAR-T cells administered on day 1. PR was achieved in the first case, whereas CRi was observed in the second and third cases with administration of higher doses. Notably, the second case maintained CRi for 100 days after UniCAR-T administration. All three cases experienced Grade 4 neutropenia during TM123 administration; however, the neutropenia improved quickly after TM123 was discontinued. Grade 1 CRS was observed in the second and third cases, with no cases of ICANS or DLT observed in any of the three cases [27]. This group recently updated the results in 2023 [28] for 19 patients who received UniCAR-T-123 treatment. Treatment was generally well tolerated with mild CRS in 12 of 19 patients, except for 3 cases of Grade 3 CRS and 1 case of Grade 2 ICANS. Eight of 15 (53%) r/r patients with AML achieved ORR, and three of four (75%) MRD-positive patients achieved ORR [29]. Therefore, UniCAR-T seem to have the potential to directly control on-target/off-tumor toxicity while also ensuring efficacy and safety.

Overall, the findings of multiple clinical trials have demonstrated certain clinical benefits of CD123 CAR-T cells. However, concerns have also been expressed about severe adverse events, particularly for severe CRS. Thus, more efforts are required to improve efficacy while maintaining a balance with safety in CD123 CAR-T.

CLL-1 (CLEC12A) is one of the inhibitory C-type lectin receptors that is expressed in 85–92% of AML cases. It is expressed in normal bone marrow cells, such as granulocytes, monocytes, macrophages, and dendritic cells; however, its expression is low in hematopoietic stem and progenitor cells as well as in nonhematopoietic cells [30].

Clinical development of CLL-1 CAR-T cells has been led by China. The first and second cases were reported from Chinese group showing the successful treatment of dual-target CAR-T cells targeting CLL-1 and CD33 for AML in a phase I clinical trial. The first adult patient achieved MRD-negative CR after CAR-T infusion with pancytopenia and Grade 1 CRS. The patient underwent consolidative HLA-matched HSCT and remained relapse-free during the observation period [31]. The second pediatric patient achieved MRD-negative CR 19 days after the treatment. The patient also experienced severe pancytopenia and Grade 3 ICANS. The patient underwent consolidative HSCT after achieving CR but later succumbed to severe infection [32].

Although no further updates have been reported from this group, several promising results have been reported regarding the clinical efficacy of CLL1-targeted CAR-T cells by other Chinese groups.

Researchers from Guangzhou Women and Children’s Medical Center infused CLL-1 CAR-T cells into four pediatric patients, and three of the four patients achieved MRD-negative CR (Table 1) [33]. These patients experienced Grade 2 or lower CRS, which resolved quickly. One patient maintained a CR without undergoing consolidative HSCT and is still alive 9 months post CAR-T therapy. The other two patients remained in remission for a relatively long time but eventually died 23 and 10 months after treatment. They later conducted another clinical trial including eight pediatric patients treated with CLL-1 CAR-T cells [34]. CRS up to Grade 2 occurred in all 8 patients without ICANS or DLT. All patients had Grade 4 neutropenia and thrombocytopenia. Four of the seven patients who achieved PR or CR/CRi underwent HSCT following treatment on days 28–82, and four of them maintained MRD-negative CR during follow-up [34].

Jin et al. [35] described the treatment of ten adult patients with CLL-1 CAR-T cells. All 10 patients developed CRS, with 6 developing Grade 3 CRS, but no ICANS was observed. All patients had severe pancytopenia including Grade 3 or higher neutropenia in 9 patients. One patient underwent HSCT for rescue, and two others died from severe infections caused by prolonged neutropenia after achieving CRi. Among seven patients who achieved CR/CRi including four patients attaining MRD-negative status, six patients underwent HSCT and maintained MRD-positive CR for a median observation time of 173 days (range 15–488) [35]. They recently updated the results [36] of 30 patients who received CAR-T cell infusions ranging from 0.5 (DL1) to 2.0 (DL4) × 106/kg, followed by LD chemotherapy. All patients developed CRS, with 12 experiencing Grade ≥ 3 and one experiencing Grade 4 neurotoxicity. All patients had hematological toxicity, with neutropenia (Grade 3–4) in 29 (97%) and thrombocytopenia in 30 (100%). Twenty-two patients (73%) achieved CR/CRi, with 12 (40%) achieving MRD-negative CR/CRi. Although it is difficult to assess the long-term efficacy of CAR-T cells because almost all patients were bridged to HSCT 0.7 (0.3–1.13) months after CAR-T infusions, their efficacy is promising with a median PFS of 300 (191–409) days and a median OS of 348 (234–462) days.

Ma et al. [37] infused PD1-silenced anti-CLL-1 CAR-T cells into two r/r patients with AML to treat relapse after HSCT and multiline salvage therapies including anti-CD38 CAR-T cell infusion. The patients experienced Grade 1–2 CRS but no neurotoxicity. Both patients achieved CRi with molecular remission on day 28 after CAR-T infusion.

Pei et al. [38] reported about seven pediatric patients who were treated with CLL1 CAR-T cells with either 41BB or CD28/CD27 costimulatory molecules. All patients experienced Grade 1–2 CRS, and one patient developed Grade 2 ICANS. The overall response rate was 75% and 66.7% in the CD28/CD27 and 41BB groups, respectively.

Overall, the reported clinical results demonstrated the potent efficacy and safety of CLL1 CAR-T cells against AML, indicating that further research is warranted in this regard. However, as all clinical results were reported from China, where CAR-T therapy regulations and clinical study design appear to be different from those in other countries, the findings should be interpreted with caution.

While CD7-targeting CAR-T cells were originally developed for the treatment of T-ALL, some AML cells have been found to express CD7. Therefore, several groups are conducting clinical trials of CD7 CAR-T cells for the treatment of AML. Recently, Zhang et al. [39] reported promising results for naturally selected CD7 CAR-T cells for AML in a Phase 1 clinical trial. Ten patients were enrolled in the study and infused with 0.5–1.0 × 106/kg CD7 CAR-T cells. All ten patients developed CRS, with two patients developing Grade 3 CRS. None of these patients developed neurotoxicity. Seven of ten patients achieved CR, with six of them achieving MRD-negative CR. Notably, all nonresponding patients had CD7-negative relapse. Therefore, CD7 CAR-T cells should have potent antitumor effects against AML while antigen-loss may limit their efficacy.