Targeting CD47 for cancer immunotherapy

Cluster of differentiation 47 (CD47) is a trans-membrane protein ubiquitously expressed on human cells but overexpressed on many types of tumor cells. It is a cell surface glycoprotein molecule, belonging to the immunoglobulin superfamily, binding to various proteins including integrin, thrombospondin-1 and signal regulatory protein α (SIRPα). CD47 is an important tumor antigen for the development and progression of various cancers. The interaction of CD47 with SIRPα triggers a "don't eat me" signal to the macrophages, inhibiting phagocytosis. Thus, overexpression of CD47 enables tumor cells to evade immune surveillance via the blockade of phagocytic mechanisms. CD47 blockade alone is not sufficient to trigger macrophage anti-tumor activity. Macrophages also need an “eat-me” (pro-phagocytic) signal. In recent years, accumulating data suggest that the CD47-SIRPα axis is a key immune checkpoint in different cancers including hematological malignancies, similar to that of the PD-1/PD-L1 checkpoint for solid tumors. CD47-SIRPα blockade has emerged as a next-generation immune checkpoint disruption strategy in various malignancies after PD-1/PD-L1. Many CD47 monoclonal antibodies (mAbs) not only block CD47 from engaging SIRPα, but also engage the activating Fc gamma receptor (FcγR) on macrophages. Together they deliver a potent phagocytic signal to macrophages.

In addition to the CD47-SIRPα signaling pathway, significance of CD47 expression has been investigated in different malignancies. Association between CD47 expression, clinical characteristics and prognosis in patients with different malignancies, including advanced non-small cell lung cancer [1], gastric cancer [2], colorectal adenocarcinoma [3] and pancreatic neuroendocrine tumor [4] has been explored. For example, in patients with high-grade lung neuroendocrine tumors, the high expression of CD47 was strongly associated with a worse progression-free survival, especially in patients with a Ki-67 > 40% [1]. CD47 expression also has correlation with adverse clinicopathologic features and an unfavorable prognosis in colorectal adenocarcinoma [3]. Similarly, CD47 has been found playing important roles in hematological malignancies, including in non-Hodgkin lymphomas (NHL), lymphoblastic lymphoma/acute lymphoblastic leukemia (LBL/ALL), acute myeloid leukemia (AML) and multiple myeloma (MM) [5, 6]. CD47 has become a potential therapeutic target and is being studied in various preclinical studies and clinical trials to prove its safety and effectiveness in the treatment of hematological neoplasms [7], and continues to grow rapidly around the world [7‐10]. Many therapeutic products targeting CD47 are being developed, including anti-CD47 mAbs, bi-specific antibodies (BsAbs) to CD47 and other molecules, as well as SIRPα-related fusion proteins. Some of the therapeutic products have demonstrated promising results in clinical research. This review will help readers to comprehensively understand the advances of clinical development of CD47-targeted therapeutic products including mAbs, SIRPα-Fc fusion proteins and BsAbs. The trend of technological innovation development is also discussed.

Prior to becoming the current new focus in the field, the prospect of CD47 just a few years ago was quite dim. In 2017, a phase I clinical trial of CD47 mAb Ti-061 in Europe (EudraCT number, 2016-004372-22) was terminated [11]. Then, in 2018, CD47 mAb CC-90002 failed in the phase I clinical trial (NCT02641002). The severe hemolytic reaction caused by hemagglutination (HA) induced by CD47 monoclonal antibody was the main problem of those clinical failures. The repetitive setbacks made the prospect of developing CD47 for cancer treatment very pessimistic. In 2019, combination of CD47 mAb magrolimab and azacitidine in the treatment of acute myeloid leukemia (AML)/myelodysplastic syndrome (MDS) showed excellent and sustainable efficacy with manageable hematological toxicity when the initial dose of 1 mg/kg was given one week before the treatment dose (10–30 mg/kg) [12]. With that positive result, since then, CD47-targeted drug development gained renewed interests, was resurrected and thrusted into a new era.

CD47-related clinical research and trial information was retrieved through PubMed database and the online open resources for clinical trial registration platforms from the US national clinical trials registry (NCT) system www.​clinicaltrials.​gov and China drug trials registry (CDT) system www.​chinadrugtrials.​org.​cn. Frontiers of CD47 clinical research are summarized, and its clinical trials-related information including organizations, identity numbers, phases, participating centers, conditions, status, enrollment, targets, spectrum of MOA, start date is collected. Clinical trials are summarized based on the territory of pharmaceutical companies: international or China-initiated or—involved as presented in Tables 1 and 2.

In the past few years, the clinical research on CD47 mAbs has made rapid progress in the USA. As of August 28, 2021, 46 clinical trials on CD47-targeted therapy were registered on clinicaltrials.gov platform. Different types of cancer patients were studied in these clinical trials, including 29 trials in solid tumor (ST) including gastric cancer (GC), head and neck squamous cell carcinoma (HNSC), and leiomyosarcoma (LMS), 14 trials in hematological malignancies including non-Hodgkin's lymphoma (NHL), AML, MDS, MM and chronic myeloid leukemia (CML), as well as 3 trials in both solid tumors and hematological malignancies. Among these clinical trials, different anti-CD47 mAbs were investigated with various strategies [47]. The types of mAbs and its MOAs were quite diverse (Table 1, 2).

In recent years, much progress has also been made in anti-CD47 therapeutic development in China. According to the information from the national pharmaceutical products administration (NMPA) website, at least 17 CD47-targeted drugs have been approved by NMPA for clinical research, accounting for more than half of the CD47-targeted agents globally. Currently, there are 15 registered clinical trials using 8 different therapeutic products in the CDT system; 9 of them are in patients with solid tumors and 6 are in patients with AML/MDS. The eight therapeutic agents from China include 6 mAbs, 2 BsAbs and 1 fusion protein, and 7 of them are registered in both NCT and CDT registry platforms, one drug is registered in CDT platform only. Here we summarize and review the advances on these potential therapeutic products from China.

Immune checkpoint inhibitor (ICI) therapy, such as PD-1/PD-L1, is considered as a revolutionary anti-tumor immunotherapy strategy. However, certain tumors do not respond to the ICI therapy, owing to the lack of T lymphocytes pre-infiltration and co-existing of intricate immune-negative microenvironment including the macrophage-suppressed “Don’t eat me” CD47 signal overexpression. Therefore, studies to boost both the innate and adaptive immune reactions by cutting off PD-1/PD-L1 and CD47-SIRPa signal pathways have been tried. The preclinical experimental results showed that the nanocomposites exhibited a stronger anti-tumor immune effect, promoted infiltration of T lymphocyte into the tumor site and strengthened phagocytosis of macrophages. More importantly, it can significantly reverse pro-tumor M2-like tumor-associated macrophages (TAMs) to anti-tumor M1-like TAMs [79]. This provided a promising strategy to develop high-efficiency and low-toxicity immunotherapy based on nanotechnology.

BsAbs against dual checkpoint blocking of CD47 and PD-L1 have been developed to provide new insights into how bone marrow cells and T cells are uniquely regulated by dual congenital and adaptive checkpoint mAbs. Results demonstrated their potential in clinical development (NCT04881045) to improve patient prognosis through concurrent PD-L1 and CD47-targeted therapy [80]. Other BsAbs have also been developed by targeting different immune check point pathways (Table 1). Multiple clinical trials targeting dual molecules such as CD47/PD-1(HX009), CD47/41BB (DSP107), CD47/PD-L1 (IBI322, PF-07257876), CD47/CD20 (IMM0306), CD47/CD19 (TG-1801) have been started.

Based on the high expression of cytotoxic T lymphocyte-associated protein 4 (CTLA-4) on regulatory T (Treg) cells and abundant Fc receptor-expressing active phagocytes inside the TME, preclinical study had been done by combining anti-CTLA-4 mAbs with the SIRPα, targeting Treg cells and selectively blocking CD47 on intratumoral Treg cells, respectively. The results demonstrated that simultaneously modulating both "eat me" and “do not eat me” signals induced Treg cell depletion inside the TME and could be an effective strategy for treating solid tumors [81].

CD47 is a target expressed in many cancers. Due to ubiquitous CD47 expression on senescent RBCs, however, the therapeutic utility of CD47-SIRPα blockade mAbs is largely compromised due to significant RBCs toxicities and fast target-mediated clearance as a result of extensive expression of CD47 on normal cells. A priming strategy has been developed to avoid the RBCs depletion toxicities. Patients are first treated at a low dose to remove the aging RBCs, thereby inducing compensatory hematopoiesis [10].

To overcome these limitations and further improve therapeutic efficacy, new generation CD47 antibodies that efficiently target tumor cells while exerting a minimal adverse effect on RBCs to avoid severe anemia were discovered [51]. A few ideally designed BsAbs have been developed, such as IBI322 for CD47/PD-L1, IMM0306 for CD47/D20 and IMM2902 for CD47/Her2, which can preferentially bind to CD47 on tumor cells with the help of higher affinity of the bi-specific molecules for another tumor target, effectively inhibiting CD47-SIRPα signal and triggering strong tumor cell phagocytosis in vitro, but only with minimal impact on CD47 single positive cells such as human RBCs. However, as discussed previously, multiple factors have to be considered such as target selection for signaling pathway, different subclasses with different effector functions such as ADCC and ADCP, antigen cellular distribution and the related biological activity [82].

The on-target/off-tumor effects, especially on platelets, can also cause potential problems with thrombocytopenia. However, this toxicity of anti-CD47 antibody on erythrocytes and platelets seems to be Fc dependent [51]. These findings suggest that optimizing the Fc structure of anti-CD47 mAs may ameliorate these adverse effects. Some of the new generation CD47 mAbs had been designed with this optimization. In addition, considering that old erythrocytes may be more susceptible to phagocytosis, the age of patients should be considered in the future CD47/SIRPα-

targeted therapy studies [83, 84].

The so-called antigen sink effect may also add challenges to the anti-CD47 therapy. The widespread expression of CD47 means that a drug may need a large initial dose and/or frequent administration to effectively block CD47. Compared with CD47, SIRPα has more limited histological distribution, which may have a better blocking effect and less therapeutic toxicities in targeted therapy [85]. Due to the polymorphisms of SIRP family members (SIRPβ and SIRPγ), cross-reactivity may occur between SIRP members. However, the consequences of targeting these different isoforms of receptors are not yet clear [86].

The anti-CD47 mAbs therapy may also face challenges in the combination therapies, including tumor cell-specific antibodies and T-cell checkpoint inhibitors. Usually a given anti-CD47 mAb is human IgG4 that has minimal activity of FcγR engagement and cannot stimulate the so-called Eat me signal on its own and thus would need another IgG1 antibody to fulfill the task. This will limit the use of CD47 mAb in combination with another IgG4 antibody (such as PD-1 antibody). Conversely, CD47-specific IgG1 antibody or recombinant protein would have the advantages of “One stone, two birds” if the hematological toxicity could be resolved. As mentioned earlier, IMM01 is such a molecule that can simultaneously block the “Don’t eat me” signal and stimulate the “Eat me” signal and thus would have little limitation when combining with other tumor therapeutics.

From the laboratory to the bedside, different types of CD47-related drugs including mono-specific and BsAbs as well as fusion proteins have been developed into variable clinical research stages. Although IgG4 is the most popular isotype, however, other isotypes including IgG2 and IgG1 have been used with the consideration of multiple factors as it was discussed previously [82]. Promising results have been achieved from the clinical trials. New technologies for the development of CD47 mAbs have been explored in different therapeutic agents. In addition to CD47 blockades on target cells and SIRPα blockades on immune effector cells, the third category of CD47-SIRPα signaling pathway inhibitor is the glutaminyl-peptide cyclotransferase-like protein (QPCTL) which is required for CD47 maturation [87]. QPCTL could be a novel target to augment antibody therapy of cancer. Meanwhile, novel drug carriers such as modified biomimetic nanoparticles have been explored in cancer therapy [88], which may be a novel pathway worthy to deliver the therapeutic CD47 mAbs.

It is important to emphasize that because TME is so intricate that blocking single signaling pathways on a subset of immune cells only have limited or mild impact. For further investigation and better tumor treatment efficacy, multi-target combination therapies are particularly important. By precisely and comprehensively reprogramming TME, combination therapies containing CD47 and other immune checkpoint inhibitors may exert better effects than monotherapy. Considering the nature of SIRPα-expressing macrophages and dendritic cells as professional antigen-presenting cells [88, 89], concurrently activating both innate immune cells via CD47-SIRPα blockade and adaptive immune cells via immune checkpoint inhibitors in conjunction with other tumor-targeted therapies so to achieve long-lasting anti-tumor activity would be the future of drug development for tumor immunotherapy. Most recently, a study demonstrated that SIRPα-αCD123 antibodies combined local CD47 blockade with specific AML LSCs targeting in a single molecule, minimized the risk of targeting healthy cells and efficiently eliminated AML LSCs. This indicates that SIRPα-αCD123 fusion antibodies targeting CD123 in conjunction with CD47 blockade could enhance the clearance of AML-initiating cells, which may be used as promising therapeutic interventions for AML [90].