Acute Myeloid Leukemia (AML) is the most common acute leukemia in adults, with around 20,000 newly diagnosed cases annually in the US. Chemotherapy and hematopoietic cell transplants lead to positive results in some AML patients, but some patients are not responsive to these treatments and could benefit from other therapeutic options. Because AML cells express surface CD33 in most patients, this cell surface marker is a potential target for treatment.
Gemtuzumab ozogamicin (Mylotarg) is a drug comprised of a cytotoxic compound conjugated to a CD33 antibody and was developed as an AML treatment. A new approach to targeting CD33-expressing leukemia cells using CD33-targeting antibodies was recently reported by Fred Hutch scientists. Instead of delivering a toxic drug to CD33-expressing AML, as Mylotarg does, recent studies have focused on recruiting T cells to the leukemic blasts. This might be a more effective approach without harmful side effects. Development of antibody fragments specific for both CD3 and CD33, the so-called bispecifics, have shown promise in this regard. However, these antibody fragments are rapidly cleared from circulation and require continuous delivery with infusion pumps for maximum effectiveness. Dr. Roland Walter and his research team in the Clinical Research Division at Fred Hutch collaborated with Covagen and developed a bispecific CD3/CD33 FynomAb named COVA4231. COVA4231 is an intact CD3 antibody with a CD33 Fynomer attached. A Fynomer is a small protein that can bind to other proteins, functioning in a similar manner as antibodies.
In a study published in the journal Leukemia, the researchers showed that COVA4231 binds specifically to T cells expressing CD3 and CD33+ tumor cells. This binding activated CD4+ and CD8+ T cells and induced secretion of effector cytokines such as IFN-gamma and IL-2 in CD8+ T cells. Interestingly, activation of T cells did not occur when the target cells lacked CD33 expression. Using either a CD33 Fynomer fused to a non-binding IgG1 antibody or a CD3 antibody without a Fynomer did not activate the T cells, suggesting that T cells and CD33+ cells must be brought into close proximity for this activation to occur.
Walter and his team then wanted to demonstrate that in addition to T cell activation, COVA4231 facilitated the T cell-mediated cytotoxicity. To do this, they treated human leukemia cell lines expressing CD33 and human T cells with COVA4231. The T cells lysed the leukemia cells. Next, they tested the ability of COVA4231 to kill freshly isolated AML blasts. Flow cytometry experiments revealed a significant increase in dead AML blasts when T cells and COVA4231 were incubated for 48 hours in 15 primary AML samples.
These early results were promising, but the authors questioned whether COVA4231 would work in mouse models. As FynomAbs use a full-sized IgG antibody as the base, the scientists expected COVA4231 to remain in circulation longer than an antibody fragment would. Indeed, COVA4231 administered to mice lingered in circulation with a half-life of 15.8 days. Walter compared this to bispecific antibodies researched in other studies, saying the “Best established in the clinic is the efficacy of the class of Bispecific T cell Engager (BiTE) molecules, with the CD19/CD3 BiTE blinatumomab being the most prominent example and the one that has recently received FDA approval. One important shortcoming of BiTE and similar fragment bispecific molecules is the very short half-life (in the magnitude of 2 hours) which requires patients to walk around with infusion pumps for several weeks. The Covagen construct addresses this limitation as it is a bispecific molecule that is built on a regular antibody format and has, therefore, an expected half-life in people in the order of 2 weeks – this will allow intermittent (i.e. every several days) dosing – a huge simplification of care.”
Testing the ability of COVA4231 to promote T cell-mediated killing in vivo required the use of mice to model tumor growth. The scientists implanted HL-60 leukemia and T cells expanded from healthy donors into mice that were treated with COVA4231 every third day. Tumor growth was reduced in mice treated with COVA4231 as compared to control mice treated with an anti-CD3 antibody .
Other disease-bispecific FynomAbs have been studied, including one designed to treat psoriasis. Walter commented on their potential: “A second cool thing with the Covagen molecule is that is makes a pretty compelling case that you can use non-antibody-based structures (in this case, Fynomers) to build custom molecules that recognize a target with high-affinity, similar to what an antibody can do.”
This study was funded in part through a sponsored research agreement with Covagen.
Roland Walter is a Fred Hutch/UW Cancer Consortium faculty member.
Klupsch K, Baeriswyl V, Scholz R, Dannenberg J, Santimaria R, Senn D, Kage E, Zumsteg A, Attinger-Toller I, von der Bey U, König-Friedrich S, Dupuy F, Lembke W, Albani C, Wendelspiess S, Dinkel L, Saro D, Hepler RW, Laszlo GS, Gudgeon CJ, Bertschinger J, Brack S, Walter RB. 2018. COVA4231, a potent CD3/CD33 bispecific FynomAb with IgG-like pharmacokinetics for the treatment of acute myeloid leukemia. Leukemia. doi: 10.1038/s41375-018-0249-z. [Epub ahead of print].