3F8

The murine IgG3 antibody 3F8 is the first anti-GD2 antibody developed and studied for effectiveness in treating neuroblastoma. An early phase I study of anti-GD2 monotherapy showed anti-tumour responses, ranging from complete clinical remission to mixed tumour regressions, in 7 of the 17 patients with neuroblastoma and melanoma treated. Typical acute side effects of severe pain and urticaria were prevalent throughout this study, thought to be the result of 3F8 interaction with pain sensory fibers, but long-term effects were not observed.20 Despite demonstrating effectiveness against minimal residual disease, studies of 3F8 have failed to produce notable efficacy against bulky disease.21 22 Due to observations that human anti-mouse antibody (HAMA) responses following treatment correlated with reduced persistence of anti-GD2 antibodies in circulation,23 24 a humanized 3F8 was developed (hu3F8) to overcome potential reductions in ADCC and complement-dependent cytotoxicity (CDC).25

Naxitamab

Naxitamab (Hu3F8) is a humanized IgG1 antibody being tested clinically as both a monotherapy and in combinations with immunogenic cytokines (table 1). A phase I dose-escalation study (NCT01757626) of naxitamab, in combination with granulocyte-macrophage colony-stimulating factor (GM-CSF) to stimulate monocyte production and maturation, demonstrated modest and manageable toxicity with no maximum tolerated dose, in addition to significant anti-tumour responses and durable progression-free survival (PFS) achieved even in patients with refractory disease.26 Following striking results from a phase II multicenter trial (NCT03363373) showing naxitamab treatment induced a 61% complete response rate in high-risk patients with refractory or relapsed bone disease, naxitamab with GM-CSF received Food and Drug Administration approval for this indication.27 Interestingly, many trial participants were managed as outpatients due to easily manageable toxicities, indicating better tolerance to humanized anti-GD2 compared with the murine m3F8. Despite this, naxitamab treatment typically requires supervision and further intervention to manage severe pain and hypotension.

Dinutuximab

Several anti-GD2 antibodies have been generated and characterized over the past decade, with dinutuximab (Ch14.18/SP2/0) being the first to receive marketing approval in 2015 for the treatment of high-risk neuroblastoma.28 It is a chimeric IgG1 anti-GD2 monoclonal antibody derived from SP2/0 hybridoma cells and received approval after a phase III clinical trial (NCT00026312) demonstrated that patients receiving a combination of GM-CSF, IL-2, isotretinoin and dinutuximab exhibited significantly better PFS and overall survival (OS) compared with those treated with standard chemotherapy.29 However, significant toxicities were reported throughout this study, with a high frequency of participants exhibiting high-grade pain, capillary leak syndrome and serious infusion reactions.18 Toxicity is often so severe that side effects require careful management with other pharmaceuticals such as analgesics. Long-term follow-up revealed late-relapse in a subset of patients, although benefit to survival outcome across the cohort remained significant.29

Dinutuximab beta

Dinutuximab beta (Ch14.18/CHO) is a dinutuximab analog produced by the International Society of Paediatric Oncology European Neuroblastoma Group (SIOPEN) using Chinese hamster ovary (CHO) cells to optimize protein glycosylation patterns and reduce rapid drug clearance by endogenous antibodies which recognize non-human sialic acid epitopes in antibodies derived from SP2/0 cells.30 The re-cloned Ch14.18/CHO exhibited similar pharmacokinetics in both humans and mice, with a small-scale phase I trial from SIOPEN showing similar toxicity profiles and evidence of clinical neuroblastoma responses.30 Dinutuximab beta is approved as a second-line treatment for patients with high-risk neuroblastoma over 12 months of age who initially responded to induction chemotherapy, or in combination with IL-2 for those with refractory disease.31 Incorporation of IL-2 is supported by preclinical and clinical studies showing an enrichment of activated NK cells enhances anti-GD2-mediated ADCC.32 33 Interestingly, a phase III multicenter trial (NCT01704716) found that inclusion of subcutaneous IL-2 with dinutuximab beta did not improve the survival of patients with prior responses to standard induction therapy, but significantly increased the incidence of toxicities such as hypersensitivity and neurotoxicity.34 A follow-up study showed that decreasing the dose of IL-2 significantly reduced associated toxicities.35 Together they support the reservation of IL-2 as an adjuvant treatment for particularly hard-to-treat tumours. Although fine-tuning of immunomodulatory treatments to maximize anti-tumour responses while avoiding immune-related adverse events is highly challenging, optimization of treatment schedules may yet improve clinical responses to dinutuximab in neuroblastoma.

hu14.18K322A

Other anti-GD2 antibodies in development have followed a similar approach of humanization to counteract dose-limiting toxicities and HAMA responses. Indeed, phase I clinical testing of humanized 14.18 (IgG1 isotype, hu14.18K322A) demonstrated reduced pain and other toxicities associated with ch14.18 treatment, while still inducing an objective response in 18% of patients with refractory neuroblastoma.36 Efforts to augment anti-neuroblastoma responses progressed from subcutaneous administration of IL-2 to the design of a hu14.18-IL-2 fusion protein. A phase II trial of hu14.18-IL-2 monotherapy in patients with refractory neuroblastoma demonstrated complete responses (21.7%) in patients with minimal residual disease, although patients with bulky disease did not respond.37 The study suggested that this antibody when used along with induction chemotherapy improves early responses, reduces tumour volume and leads to an event-free survival of 3 years.38

Overall, multiple anti-GD2 antibodies have undergone molecular revision and refinement to reduce immune-related toxicities and severe neuropathic pain associated with CDC and HAMA responses. Despite the attractive tissue expression and restricted distribution of GD2 as a target, the ADCC-based mechanism of anti-GD2 induces often intolerable toxicity made worse by combination with activating cytokines such as IL-2. Alternative methods of enhancing anti-GD2 efficacy are required, since of the three clinically approved antibodies available, currently dinutuximab is the only one deemed effective for standard therapy.

Targeting PD-1

The fully human IgG4 antibody nivolumab was evaluated in two phase II trials designed to assess the toxicity and pharmacokinetics of anti-PD-1 monotherapy in patients with refractory neuroblastoma (NCT02304458, NCT02813135). The first study demonstrated that nivolumab is well-tolerated in children with similar pharmacokinetics to those seen in adults, however objective responses were not observed and benefit was limited to stable disease only.44 The second study showed that use of nivolumab monotherapy was ineffective,45 which is consistent with the results of preclinical studies of single-agent immune checkpoint inhibitors (ICIs).46 Another monoclonal antibody used clinically to target the PD-1 pathway in a range of adult cancers is the humanized IgG1 anti-PD-L1 antibody atezolizumab. A phase I/II paediatric study of atezolizumab in children with solid tumours such as neuroblastoma (NCT02541604) demonstrated safety but limited anti-tumour responses.47 Thus, PD-1/PD-L1 blockade in paediatric patients with neuroblastoma has proven ineffective in clinical trials to date, despite the significant clinical successes in adult cancers. This may be a result of trial design and patient recruitment since adults approved for treatment with nivolumab or atezolizumab undergo pathology assessment to determine tumour mutational burden (TMB) and PD-L1 positivity. A caveat of the trials discussed is that the patients with neuroblastoma recruited were known to exhibit tumours with low TMB and low PD-L1 expression, so the lack of sensitivity to ICB is unsurprising.48 Future trials with specific ICI-related eligibility criteria may reveal a subset of patients sensitive to PD-1/PD-L1 blockade; potentially those exhibiting mismatch repair deficiency and a sufficient antigen load to drive anti-tumour T-cell responses.

Preclinical studies have sought to overcome the challenge of low-sensitivity to anti-PD-1 in neuroblastoma by inducing target cell PD-L1 expression. One provocative study demonstrated that targeting GD2 in LA-N-1 neuroblastoma cells with ch14.18/CHO in co-culture with primary leukocytes induced strong PD-L1 expression in what is usually a PD-L1 negative model.49 Interestingly, LA-N-1 PD-L1 upregulation effectively inhibited ch14.18-mediated ADCC, an effect abrogated by nivolumab treatment. Indeed, the nivolumab-ch14.18 combination synergistically inhibited the growth of NXS2-HGW murine neuroblastoma tumours in vivo and enhanced survival relative to either agent alone.49 This strategy was tested clinically in an interesting case involving two heavily pretreated patients with refractory disease.50 Strikingly, one patient exhibited a complete response and regression of multiple positron emission tomography (PET)-positive lesions, whilst the other exhibited tumour shrinking and partial remission. Although this was a minor case study, the observations are notable for the responses achieved in patients previously unresponsive to dinutuximab and potentially indicate that anti-GD2-induced PD-L1 expression renders neuroblastoma susceptible to nivolumab treatment.50

Targeting CTLA-4

Ipilimumab is a first-in-class human monoclonal antibody targeting the immune checkpoint CTLA-4. A phase I clinical trial (NCT01445379) assessing pharmacokinetics and immunogenicity in a mixed cohort off refractory solid tumours showed that it is tolerated in children, although immune-related adverse events (irAEs) consistent with those seen in adults were observed. Despite the lack of tumour regressions following single-agent ipilimumab, patients who exhibited irAEs had an increased OS relative to those who did not, suggesting that disruption of immunological tolerance with ipilimumab was beneficial in terms of survival outlook.51

Targeting both PD-1 and CTLA-4

Poor clinical responses to single-agent ICB have motivated preclinical investigations into combination strategies. One such study showed that the murine syngeneic NB-SQ neuroblastoma model exhibits an immunosuppressive TME characterized by a high frequency of myeloid-derived suppressor cells and Tregs.52 Consistent with clinical studies described here, the treatment of NB-SQ tumours with single-agent anti-PD-1 or anti-CTLA-4 failed to control the growth of established tumours. Notably, mice treated with dual anti-PD-1/anti-CTLA-4 following a course of cyclophosphamide exhibited an enrichment of effector CD8+T cells and significantly increased survival compared with those treated with ICI monotherapy.52 An alternative preclinical study evaluated an induced neuroblastoma model with immunogenic features and high-level PD-L1 expression.53 The authors observed that the murine N2a cell line mimics features of human SK-N-SH and SH-SY5Y neuroblastoma cells and that each cell line exhibits interferon (IFN)-γ-inducible PD-L1 expression. Interestingly, subcloning of N2a produced an aggressive AgN2a model which lacked PD-L1 expression and sensitivity to IFN-γ, reflecting features of high-risk clinical neuroblastoma. Combining anti-PD-1 or anti-CTLA-4 with tumour cell vaccines for these neuroblastoma cell lines induced complete regression of established tumours in vivo. It should be noted that ipilimumab with nivolumab is approved for the treatment of unresectable melanoma, and given that one safety trial has established comparable safety profiles for ipilimumab in adults and children, this represents a potential avenue for future progress.

Targeting CD47 to promote phagocytosis

In contrast to GD2, the transmembrane integrin-associated protein CD47 is ubiquitously expressed throughout a range of different tissue types. Being frequently overexpressed in tumours, CD47 enrichment is strongly associated with worse prognosis in both haematological and solid cancers.63 Inhibition of CD47 promotes macrophage-mediated phagocytosis of target tumour cells which also express secondary pro-phagocytic signals, such as calreticulin.64 Magrolimab (Hu5F9-G) is a first-in-class humanized IgG4 anti-CD47 which exhibits ICI activity in a wide range of preclinical solid tumour models.63 65 A first-in-human phase I trial of magrolimab (NCT02216409) revealed mild-to-moderate toxicity and objective tumour responses were reported in two patients, one with ovarian and another with fallopian tube tumours.63 A preclinical study demonstrated potent Hu5F9-G4 activity against multiple patient-derived xenograft models of paediatric brain tumours, observing substantial macrophage recruitment in treated tumours.66 An important finding of this study was that Hu5F9-G4 did not induce toxicity in neural progenitor cells, or promote phagocytosis of differentiated progeny, indicating differentiation between normal and cancer cells. Several drugs targeting the CD47-SIRPα axis are being tested against multiple tumour types; these include Hu5F9, TTI-621, ALX-148, AO-176 and OSE-172.67 Of these, only Hu5F9 is currently being studied for neuroblastoma treatment in combination with dinutuximab in a first-in-human/first-in-child phase I trial (NCT04751383). Although no clinical data has been released, recent assessment of this strategy in preclinical xenograft and murine syngeneic neuroblastoma models demonstrated potent synergy and tumour clearance.68 Notably, the authors identified the likely synergistic mechanism as GD2 ligation-dependent upregulation of the pro-phagocytic signal calreticulin, in addition to blocking GD2 interactions with the inhibitory immunoreceptor Siglec-7. Indeed, these encouraging findings are supported by other preclinical studies which demonstrated CD47 upregulation following anti-GD2 treatment functions as an adaptive immune checkpoint which limits effective ADCC in neuroblastoma models.69 Thus, dual targeting of GD2 and CD47 represents a promising strategy for clinical translation in neuroblastoma.

Targeting the colony stimulating factor-1 signaling

Colony stimulating factor-1 (CSF-1)/CSF-1R interactions can also be targeted to modulate TAM migration and accumulation in tumour sites.57 CSF-1R monoclonal antibodies such as RG7155 starve CSF-1-dependent macrophages of a crucial growth factor and suppress the growth of multiple cancers.60 A preclinical study assessing the effect of the selective CSF-1R inhibitor BLZ945 against the transgenic spontaneous TH-MYCN model of high-risk neuroblastoma demonstrated significant growth-limiting effects after treating established tumours.70 Interestingly, intratumoural macrophage and monocyte densities were unaffected by BLZ945, but significant depletion of splenic macrophages was observed, potentially indicating myeloid sustenance by tumour-cell derived GM-CSF.71 Due to observations that intratumoural myeloid cells are a significant source of stromal PD-L1, and that CSF-1R inhibition increases T-cell PD-1 expression, the same study assessed the effect of ICB on BLZ945 activity. Treatment with ICIs alone had no effect on tumour growth, while combination with BLZ945 induced potent synergy and complete regression of small tumours and inhibited progressive growth of large tumours in a majority of mice.70 Since CSF-1R⁺ myeloid infiltration predicts poor neuroblastoma outlook and PD-L1 is strongly associated with CSF-1R expression in human neuroblastomas, this may represent a promising therapeutic avenue.72 However, preclinical evidence for CSF-1R inhibition reported to date is often conflicting and highly model and context-dependent. Some studies report dramatic remodeling of the myeloid compartment, but only moderate effects on tumour growth,73 while others report significant anti-tumour activity following intratumoural macrophage repolarization.74

Clinical trials aiming to validate the therapeutic benefit observed with preclinical CSF-1R-PD-1 dual inhibition using both small molecule (NCT02452424) and monoclonal antibody inhibitors (NCT02526017 and NCT02323191) are assessing responses in patients with neuroblastoma.

Cell-based NKT cell therapy

Cell-based therapies involving the adoptive transfer of chimeric-antigen receptor (CAR) T cells has revolutionized the treatment of haematological cancers, inducing striking and durable regressions of B-cell acute lymphoblastic leukaemia.82 83 Engineered CAR systems consist of antigen-binding domains fused to a T cell receptor (TCR)ϛς chain and additional co-stimulatory domains, such as those for CD28 or OX40.82 Multiple iterations of the technology have produced a potent method of targeting cancer cells for immune-cell mediated killing, independently of MHC function. However, the use of patient-derived autologous T cells is time consuming and susceptible to clinical manufacturing failures, making “off-the-shelf” allogeneic alternatives an attractive option.84 Vα24-invariant–natural killer T (iNKT) cells are an innate-like cell type which exhibit potent cytotoxicity and rapid cytokine secretion following the recognition of glycolipids presented by CD1d.85 Crucially, CD1d monomorphism overcomes human leukocyte antigen (HLA)-restricted graft-versus-host disease usually seen with allogeneic transfer of T cells, making iNKT cells a promising and readily available CAR-T system which may benefit wider use of the therapy.84 Preclinical studies have demonstrated the efficient trafficking of NKT cells to tumours such as neuroblastomas in response to tumour-derived chemokines.86 Furthermore, infiltration of iNKT cells into neuroblastomas evokes anti-tumour immunity by targeting and killing CD1d-expressing TAMs.87 Observations that intratumoural abundance of iNKT cells is a favourable prognostic indicator in neuroblastoma86 supported efforts to study the effect of adoptively transferred activated cells on disease progression.

A first-in-human phase I clinical trial (NCT03294954) administered chimeric antigen receptor NKT (CAR-NKT) cells to children with refractory neuroblastoma.88 After administration, the CAR-NKT persisted in vivo, trafficked to tumours and induced objective responses and metastasis regression in a subset of patients.88 Another phase I study (NCT00877110) tested a combination of the anti-GD2 antibody m3F8 and adoptive transfer of haploidentical NK cells for high-risk refractory neuroblastoma.89 The study demonstrated that combination treatment was well tolerated and induced anti-tumour responses in a subset of patients.

STING agonists in neuroblastoma treatment

Around 50% of high-risk neuroblastomas exhibit MYCN amplification100 which is associated with reduced IFN-I pathway activity and insensitivity to ICB, therefore activating the STING pathway may be beneficial in targeting MYCN amplified neuroblastoma tumours. A study using genetically engineered mouse models of neuroblastoma used the STING agonist DMXAA to evaluate the anti-tumour effect of dual STING activation and MYCN inhibition.101 While MYCN-amplified tumours are associated with reduced IFN-I pathway activity in primary metastatic neuroblastomas and established cellular models, depletion of N-Myc potently augmented IFN pathway activity in response to DMXAA treatment.101 Another study evaluated cytosolic CDN administration to overcome the issues with membrane impermeability; after cytosolic administration CDNs activated STING and IFN-I production, and induced expression of anti-tumour cytokines and chemokines, causing tumour cell apoptosis.102 STING signaling in neuroblastoma is poorly characterized and not yet studied in clinical trials. In addition, due to their inherent instability, the majority of CDN-based STING agonists currently under clinical development are normally administered directly into the tumour site.103 CDNs also face obstacles such as limited intracellular bioavailability and suboptimal pharmacokinetic properties. These factors pose challenges that hinder their effectiveness and limit their clinical applicability to tumours that are easily accessible. To address this caveat, recent research endeavors have focused on generating STING-activating compounds that are amenable to systemic delivery or STING-activating nanoparticles designed to efficiently deliver the endogenous STING ligand into the cytosol.102 Such an approach has been explored in preclinical models of neuroblastoma that led to inhibition of tumour growth and improved response to ICB.102 However, the utility of STING activating compounds against neuroblastoma remains to be explored in clinical trials.