CD47-SIRPα blocking-based immunotherapy: Current and prospective therapeutic strategies

Abstract

Background: The CD47-signal regulatory protein alpha (SIRPα) 'don't eat me' signalling axis is perhaps the most prominent innate immune checkpoint to date. However, from initial clinical trials, it is evident that monotherapy with CD47-SIRPα blocking has a limited therapeutic effect at the maximum tolerated dose. Furthermore, treatment is associated with severe side effects, most notably anaemia, that are attributable to the ubiquitous expression of CD47. Nevertheless, promising clinical responses have been reported upon combination with the tumour-targeting antibody rituximab or azacytidine, although toxicity issues still hamper clinical application.

Main body: Here, we discuss the current state of CD47-SIRPα blocking therapy with a focus on limitations of current strategies, such as depletion of red blood cells. Subsequently, we focus on innovations designed to overcome these limitations. These include novel antibody formats designed to selectively target CD47 on tumour cells as well as tumour-targeted bispecific antibodies with improved selectivity. In addition, the rationale and outcome of combinatorial approaches to improve the therapeutic effect of CD47 blockade are discussed. Such combinations include those with tumour-targeted opsonizing antibodies, systemic therapy, epigenetic drugs, other immunomodulatory T-cell-targeted therapeutics or dual immunomodulatory CD47 bispecific antibodies.

Conclusion: With these advances in the design of CD47-SIRPα-targeting therapeutic strategies and increasing insight into the mechanism of action of this innate checkpoint, including the role of adaptive immunity, further advances in the clinical application of this checkpoint can be anticipated.

Keywords: CD47; bispecific antibody; immunotherapy; patient selection; tumour selective.

FIGURE 1

Toxicity observed with CD47‐signal regulatory protein alpha (SIRPα) blocking antibodies and strategies to improve them. CD47 is expressed on healthy and tumour cells, and therefore, targeting CD47 will also result in the loss of healthy cells with CD47 expression, such as red blood cells (RBCs) and thrombocytes. The thrombocythemia and anaemia that are the result of these ‘off‐target’ effects result in a low maximum tolerated dose in clinical trials, thus limiting the effects on the tumour. To overcome this, three different strategies are discussed in this review. (1) Novel body formats are designed to target only CD47 expressed on cancer cells. These antibodies are designed to bind to clustered CD47 only. Another method to prevent binding to RBCs is by designing an antibody that binds to the epitope of CD47 that is closely located to an N‐glycosylated on RBCs and therefore functions as a ‘shield’ for RBCs. (2) Instead of targeting CD47 with CD47 monoclonal antibody (mAb) or recombinant human SIRPα (rhSIRPα), it is also possible to target SIRPα on phagocytes, thereby circumventing the RBCs and thrombocytes. (3) Bispecific antibodies are designed to target CD47 only to tumour cells with a second arm that binds only to tumour‐selective targets

FIGURE 2

Combinatory strategies that improve the therapeutic effect of CD47‐signal regulatory protein alpha (SIRPα) blocking. Currently, five different strategies to improve CD47 blocking therapy are being evaluated. (1) In clinical trials with several different types of cancer parents, Fc receptor (FcR) crosslinking antibodies that directly target tumour antigens are combined with CD47 blocking. The combination of CD47 blocking with (5) stimulation of adaptive costimulatory signals or (2) blockade of adaptive ‘don't eat me’ signals is also a promising combinatory strategy, as increasing evidence states that the adaptive immune system has a pivotal role in the effect of CD47 blocking therapy. (3) The combination of anthracycline, epigenetic drugs (demethylating agents) and proteasome inhibitors also improved the therapeutic effect of CD47 blockade. Most likely, this improvement is at least partly caused by upregulation of ‘eat me’ signals on tumour cells triggering phagocytosis by immune cells. Finally, accumulating evidence points to CD47‐SIRPα blocking antibodies triggering not only phagocytosis but also autophagy of tumour cells and health issues. (4). Combination with autophagy blockers seems to improve the phagocytic index in non‐small cell lung cancer (NSCLC) and glioblastoma in a preclinical setting

FIGURE 3

Different bifunctional CD47‐targeting antibodies (bispecific antibodies, bsAbs) and fusion proteins compromising signal regulatory protein alpha (SIRPα). (A) A normal monoclonal antibody (mAb) demonstrating the different regions used in bifunctional proteins. (B) Kappa/lambda bsAb compromised of a kappa and lambda light chain and are fused by a common heavy chain with an active IgG1 domain. (C) Di‐single‐chain variable fragment (di‐ScFv) is compromised of the variable regions of RTX and CD47 mAb fused with a linker. (D) The dual‐variable‐fragment domain is compromised of a full antibody fused to the variable region of a CD47‐ or SIRPα‐directed antibody. (E) A full RTX antibody is fused to a CD47‐directed nanobody (a single‐domain antibody fragment derived from a naturally occurring heavy‐chain IgG antibody). (F) A bispecific trap antibody that is comprised of a human epithelial growth factor receptor 2 (HER2)‐directed full antibody and the variable domain of the extracellular domain (ED) of SIRPα. (G) A fusion protein comprising the full CD123 antibody fused to the ED of SIRPα. (H) A fusion protein compromised of the V_H V_L of an epidermal growth factor receptor (EGFR) antibody fused to the ED of SIRPα using the knobs‐into‐holes technique. (I) A fusion protein compromised of V_H V_L of a CD47 mAb fused with a programmed cell death ligand 1 (PD‐L1) mAb that consists of an Fc domain with two V_L. (J) A homotrimeric fusion protein compromised of the ED of 4‐1BBL and three EDs of SIRPα. (K) An Fc‐linked fusion protein compromised of the ED of SIRPα that is linked through an inactive IgG4 Fc domain with the ligand of CD40. DVD‐Ig, dual‐variable domain immunoglobulin; MSLN, mesothelin; V_H, variable domain of the heavy chain; V_L, variable domain of the light chain