CD7 CAR-T: a bridge to transplant in AML

In this issue of Blood, Lu et al¹ report early results from a phase 1 study of autologous CD7 chimeric antigen receptor–modified T cells (CD7 CAR-T) in patients with acute myeloid leukemia (AML). The authors describe the safety and preliminary efficacy of CD7 CAR-T, which induced disease remission in most patients. However, the responses were durable only in those who received subsequent allogeneic hematopoietic stem cell transplant (alloHSCT).

Unlike lymphoblastic malignancies, AML remains resistant to CAR T-cell therapy due to a paucity of safe target antigens, high antigen heterogeneity, cancer immunosuppression, and other factors. Approaches targeting pan-myeloid markers such as CD123, CD33, or CLL-1 with CAR T cells have revealed early dose-limiting toxicities and have not yet yielded the response rates observed in lymphoid malignancies.² CD7 is a T- and natural killer-lineage marker expressed at targetable levels in ∼20% of AML cases, predominantly in less-differentiated subtypes. We and others have found activity of CD7 CAR-T in preclinical models of CD7⁺ AML,³ but the development of CD7 CAR-T and their clinical application has been primarily focused on T-cell acute lymphoblastic leukemia, where CD7 is often expressed on blasts at high and uniform levels.^4-8 

In the current report, the investigators manufactured CD7 CAR-T for 10 patients with CD7⁺ AML, 7 of whom had relapsed after a previous alloHSCT. One patient had extramedullary disease. Most CAR T-cell products were patient derived, except 1 which was manufactured from a previous stem cell donor. The authors used a second-generation, 4-1BB–costimulated CD7 CAR-T harboring 2 nanobody-derived antigen binders in tandem. Expression of the CD7 CAR in T cells induced self-targeting (fratricide) in culture, leading to a selection of fratricide-resistant CD7 CAR-T in which CD7 expression was absent or masked by the CAR. The resulting “naturally selected” CD7 CAR-T cells were expanded ex vivo and infused at 0.5 × 10⁶ or 1 × 10⁶ CAR T-cell per kilogram after lymphoablation with cyclophosphamide and fludarabine. At enrollment, all patients had detectable disease in the bone marrow, but the leukemia burden was reduced in most patients by lymphodepleting chemotherapy before CAR T-cell administration. T-cell infusions invoked grade 1 or 2 cytokine-release syndrome in 8 patients and grade 3 in the remaining 2 patients, both of whom had >25% blasts in the bone marrow before CAR T-cell infusion. No neurotoxicity was observed. CAR T cells expanded in the peripheral blood and peaked between 7 and 21 days after infusion, followed by a gradual decline. Complete remissions were observed in 7 of 10 patients, with no measurable residual disease in 6 of them. A partial response was observed in a patient with extramedullary disease. In 4 patients with initial complete response, disease had relapsed within 100 days after the infusion. The remaining 3 responders proceeded to a second alloHSCT, 1 of whom remained free of disease >400 days after, whereas the other 2 succumbed to nonrelapse complications. Furthermore, 2 patients with early disease relapse after CAR T-cell treatment received salvage HSCT and remained alive at the time of reporting.

Resistance to treatment and early relapses in 6 patients were associated with the loss of CD7 on malignant cells. Five of these patients had leukemia with double mutations in the CCAAT/enhancer binding protein alpha (CEBPA) gene, an AML genotype associated with aberrant CD7 expression.⁹ The limited number of patients makes it difficult to conclude whether CEBPA mutations underpinned antigen escape. Other factors, including low disease burden (<5% bone marrow blasts in 5 patients, including 1 without detectable disease before CAR T-cell treatment) or young age of patients with sustained remissions, may have contributed to clinical responses to CD7 CAR-T therapy. Treatment of 2 patients with a substantial (>25%) leukemia burden in the bone marrow resulted in severe cytokine release syndrome but still produced temporary remissions.

This study not only validates CD7 as a clinically targetable antigen in a subset of patients with AML but also illustrates high heterogeneity in its expression on malignant cells. Loss of CD7 observed in this study is more frequent than that observed in T-cell leukemias, likely reflecting differences in the mechanisms of CD7 gene regulation in lymphoid and myeloid malignancies. These results also highlight the role of consolidative alloHSCT to sustain remission after CAR T-cell therapy and echo a recent report where CD7 CAR-T manufactured from healthy donors were effective as a bridge to transplant in patients with refractory or relapsed AML.¹⁰ Although this therapy would benefit only a few patients with AML, CD7 targeting can help reduce global myeloablation mediated by CAR T-cells targeting myeloid lineage markers and thus potentially enable a curative stem cell transplant with fewer toxicities.

Conflict-of-interest disclosure: M.M. is a cofounder of March Biosciences with equity, serves on the scientific advisory board of March Biosciences and NKILT Therapeutics, and receives patent fees and royalties from Fate Therapeutics, Allogene Therapeutics, Beam Therapeutics, and March Biosciences. He receives research support from Fate Therapeutics and March Biosciences and honoraria from Amgen and Galapagos NV. An up-to-date declaration of interests can be found at https://www.bcm.edu/academic-centers/cell-and-gene-therapy/research/disclosure-of-outside-interests.