In Aug. 2021, immuno-oncology company, Trillium Therapeutics, was acquired by Pfizer for $2.3 billion, adding Trillium’s two CD47-targeting biologics, TTI-621 and TTI-622, to Pfizer’s oncology pipeline. The acquisition followed Gilead’s $4.9B buyout of Forty Seven in early 2020, after clinical data was presented on Forty Seven’s lead, the anti-CD47 monoclonal antibody, magrolimab. Magrolimab was granted breakthrough therapy designation by the FDA for myelodysplastic syndrome (MDS) and is currently the most advanced anti-CD47 therapy in the clinic. Here is a target minireview to get readers caught up on the science behind this emerging therapeutic axis.

CD47 – A Checkpoint for Innate Immune Cell Activity

Cancer immunotherapy has been a hot area in clinical development for some time thanks to the success of anti-PD-1/L1 and anti-CTLA-4 checkpoint inhibitors. Checkpoints like PD-1 can be thought of as “don’t kill me” signals on cancer cells, preventing T-cells from clearing tumors (there are additional mechanisms, but the framework is useful). Blocking checkpoints like PD-1 allows T-cells to proceed with killing tumors, leading in some cases to durable tumor regression.

If T-cells can be activated by checkpoint inhibitors against cancers, are there similar checkpoints whose inhibition could allow innate immune cells like neutrophils or macrophages to be turned against tumors? In the late 1980s to early 1990s, scientists led by Eric Brown developed an antibody that inhibited peptide-enhanced phagocytosis (by neutrophils) and identified its antigen to be a 50 kDa transmembrane protein belonging to the immunoglobulin superfamily. This antigen with a role in phagocytosis turned out to be CD47.

Figure 1. The interaction between CD47 and Signal Regulatory Protein α (SIRPα) elicits the intracellular phosphorylation of SIRPα and subsequent downstream signalling events that tell phagocytic cells like macrophages not to engulf the cell it’s confronting (“don’t eat me”). Inhibiting this CD47/SIRPα signalling process can lead to the phagocytosis of tumor cells. Image generated with PDB structures: 2WNG, 4CMM, and 7MYZ.

Scientists led by Stanford’s Irving Weissman later discovered that directly blocking CD47 in tumor cells led to macrophage-mediated phagocytosis of such cells (e.g. acute myeloid leukemia stem cells, metastatic leiomyosarcoma cells, and liver cancer cells). The potential importance of CD47 as a “don’t eat me” checkpoint signal was not lost on Weissman’s team amidst the immunotherapy boom. In 2014, Weissman co-founded Forty Seven as a Stanford spinout to develop CD47 inhibitors, and the anti-CD47 mAb from Forty Seven, Hu5F9-G4 (magrolimab), soon became the first CD47 inhibitor to be tested in clinical trials

Overcoming On-Target Anemia – a Challenge for CD47 Drug Discovery

Because CD47 is expressed on both tumor cells and healthy blood cells, CD47-targeting agents have the potential to cause premature turnover of healthy cells, resulting in anemia and thrombocytopenia (reviewed here). This effect has been observed in a number of preclinical animal models and in humans. Several strategies have been employed to maximize the therapeutic index of CD47-targeting drugs or manage their effects clinically. Four clinical candidates are highlighted below to illustrate strategies to overcome this challenge – a more comprehensive list of molecules targeting this axis can be found here

Magrolimab (Hu5F9-G4)

The discovery of magrolimab started with the identification of anti-CD47 mAbs by immunizing mice with the extracellular IgV domain of human CD47 (Figure 1, above). After generating hybridomas with mouse B-cells, the 5F9 antibody was identified and humanized based on structural modeling with the closest human germline sequences. 

In anticipation of the cytotoxicity that could be induced by the antibody’s Fc domain, Hu5F9-G4 was engineered with an IgG4 scaffold, which produces less antibody-dependent cell-mediated and complement-dependent cytotoxicity (ADCC and CDC). Although ADCCCDC-inducing activities were insignificant in xenograft mice, cynomolgus monkeys still exhibited dose-dependent anemia upon treatment due to the action of Hu5F9-G4 on erythrocyte CD47. However, by administering an initial, low “priming” dose that would fill part of the antigen sink and stimulate red blood cell production, the researchers found that subjects acclimated to the anemia and maintained normal hemoglobin levels even after subsequent higher doses. 

Hu5F9-G4 has been tested in several clinical trials as a monotherapy or in combination with other therapeutics for patients with osteosarcoma, neuroblastoma, colorectal, ovarian, urothelial, or blood cancer. The first-in-human trial found that with an initial priming dose, patients tolerated the investigational drug. In another magrolimab trial in non-Hodgkin’s lymphoma patients given the priming dose-containing regime, 50% achieved an objective response and 36% reaching a complete response. No clinically significant safety events were observed. 

An ongoing trial evaluating magrolimab’s efficacy in MDS and acute myeloid leukemia (AML) patients recently showed that 63% of patients achieved an objective response and 42% reached complete remission. Among AML patients, some carried a genetic variant associated with treatment-resistance AML and poor prognosis, however, 45% of patients reached complete remission after treatment, impressive given how few options there are in AML. Based on the Ph. 1b results, the FDA granted magrolimab Breakthrough Designation in 2020.

Evorpacept (ALX148)

Another approach to mitigate anemia is by using a Fc-fusion protein that binds CD47 with high affinity without eliciting antibody-dependent cell mediated phagocytosis (ADCP) through the Fc domain. The high-affinity CD47 binding portion competes for CD47 binding sites while a mutant Fc portion limits phagocytosis of healthy cells. 

The native IgV domain of SIRPa binds to CD47 with micromolar affinity, so to find stronger binding mutants, researchers at Stanford and Dusseldorf University used yeast display to identify a variant that bound CD47 in a similar manner to native SIRPα, but with picomolar affinity. To generate ALX148, scientists at ALX Oncology fused the high-affinity IgV variant with an IgG1 Fc mutant unable to bind FcγR and C1q, thereby avoiding ADCP and complement cascade but still maintaining a longer drug half-life compared to the IgV domain alone. In preclinical studies, ALX148 demonstrated single-agent and combination efficacy without effects on non-human primate red blood cells.

Encouraging results were recently observed with ALX148 in a Phase 1b clinical trial: patients with HER2+ gastric or gastroesophageal junction cancer on a trastuzumab/chemo/ALX148 combination therapy had an objective response rate of 72% and 76% overall survival at 12 months, significantly better than historical control trials. While the industry has turned wary of historical controls in small trials (think IDO), the data is encouraging and ALX148 is currently being investigated in Phase 2/3 trials.


Researchers at Trillium Therapeutics took a slightly different approach. TTI-621 was developed by fusing a high affinity IgV domain mutant with the pro-phagocytotic Fc domain of IgG1. They found that in cynomolgus monkeys, TTI-621 induced dose-limiting anemia, but interestingly, did not interact strongly with erythrocytes from the blood of healthy human donors. TTI-621 showed significantly less binding to human erythrocytes compared to other CD47-targeting mAbs, comparable to negative control immunoglobulins, but targeted tumor cells with remarkable anti-tumor activity in AML xenograft mice. The scientists hypothesized that the preference of TTI-621 for tumor cell CD47 was a result of the fact that the CD47 antigens on healthy cells are tethered to the erythrocyte spectrin cytoskeleton, which made them less membrane mobile and less clustered on the surface of normal cells. 

In a Phase 1 trial evaluating TTI-621 alone or in combination with rituximab, a 13% overall response rate was observed in patients with various relapsed or refractory hematologic malignancies. Roughly a quarter of patients showed a response when TTI-621 was administered as a monotherapy for large B-cell lymphoma or T-cell non-Hodgkin’s lymphoma. TTI-621, in combination with doxorubicin, is currently being investigated in patients with leiomyosarcoma.

NI-1701/TG1801 – CD47xCD19 Bispecific Antibody

Bispecific antibodies are another potential approach to divert the drug away from the antigen sink, by increasing its targeting specificity through the binding of two tumor surface-expressing antigens. NovImmune researchers used phage-display to select for two different antigens and identified the anti-CD47 and anti-CD19 bispecific, NI-1701/TG1801. Their process is compatible with manufacture-scale production of a bispecific with different light chains and a common heavy chain. In preclinical studies, NI-1701 was shown to kill cells of various B-cell malignancies and limited tumor growth in xenograft mouse models. The mechanism by which NI-1701 inhibits growth is by preventing the clustering and migration of CD19 to B-cell receptors, which promote growth and differentiation. NI-1701 was found to specifically bind to B-cells compared to an anti-CD47 monospecific mAb that also bound to T cells, erythrocytes, and platelets. 


The CD47/SIRPα axis is a promising target pathway for cancer immunotherapy, and is likely to become increasingly competitive. If magrolimab is approved, we might expect another PD-1/PD-L1-like industry race across indications, biomarkers, and drug mechanisms. Additional anti-CD47 mAbs such as Innovent’s letaplimab (IBI188) are already in development for broader indications including solid tumors, and numerous new CD47/SIRPα-targeting molecules of different mechanisms are entering early clinical development. This is certainly an interesting scientific area to watch for signs of differentiation between large molecule drug modalities.

We hope this is useful. Explore for more drug discovery content.

Thanks to Jennifer Huen, PhD for assistance researching, illustrating, and drafting this article, and Rebecca Burnham, PhD for conducting research for this article.

Categories: Highlights