Promising therapeutic targets in neuroblastoma

Increasing tumor radiation dose

Attempts to further improve response rate included increasing the dose, using a different iodine isotope, or a higher specific activity form of the molecule (Table 1A). Due to radiation safety limitations, the increase in the dose was accomplished by administering a rapid sequence double infusion given two weeks apart and supported by HCT. In this fashion, a single patient could receive as much as 42 mCi/kg over a two-week period in a recent phase I study; however, the response rate in that limited number of patients did not appear to be different than the standard 18 mCi/kg dose (25). Other investigators have reported benefit from repeated MIBG infusions given six to 12 weeks apart, with continued improved response in some of the patients with each successive infusion (26, 27). One small phase I trial used ¹²⁵I-MIBG, rather than the usual ¹³¹I-MIBG, based on the hypothesis that the different isotope would be more effective for microscopic tumors or single cells, since the Auger electrons travel only a few nanometers, whereas the beta particles of ¹³¹I are more effective in tumors > 1 millimeter. Myelosuppression was significant, despite a lower whole body radiation dose (28). Recently, a the New approaches to Neuroblastoma Therapy (NANT) phase I trial of “no-carrier added” ¹³¹I-MIBG was completed, based on pre-clinical data that non-radioactive “carrier” MIBG molecules in the standard preparation of MIBG (specific activity 1.2 MBq/μg) inhibit uptake of ¹³¹I-MIBG, resulting in less tumor radiation and increased risk of cardiovascular toxicity (29). In the no-carrier added preparation of MIBG, virtually every molecule is radioactive (specific activity of 165 MBq/μg). The NANT trial showed a similar toxicity and response profile to the standard preparation, but had the advantage of being able to be infused over 30 min, instead of 90 to 120 minutes (30).

Combining MIBG with radiosensitizers or chemotherapy

Another approach to improving efficacy has been using combination therapy, by adding chemotherapy or other radiosensitizers to the MIBG (Table 1A). Exploratory studies showed the feasibility of giving 7–15 mCi/kg of ¹³¹I-MIBG followed10–14 days later by myeloablative doses of carboplatin, etoposide and melphalan (31–34). A Phase I study of 24 patients with refractory neuroblastoma established the MTD at 12 mCi/kg of ¹³¹I-MIBG, with carboplatin 1,500 mg/m², etoposide 1,200 mg/m², and melphalan 210 mg/m². The dose-limiting toxicities were mucositis, vascular leak, veno-occlusive disease, and sepsis, but only one toxic death (35). A subsequent NANT phase II study of 50 patients reported the main toxicity as veno-occlusive disease, and 7/50 responses in this highly refractory group of patients (36).

Other studies focused on using the ¹³¹I-MIBG combined with chemotherapy or radiosensitizers in a non-myeloablative regimen in refractory patients. Italian investigators treated 16 patients with relapsed or refractory neuroblastoma with 200 mCi ¹³¹I-MIBG combined with cisplatin and cyclophosphamide with or without etoposide and vincristine, and obtained 12 partial responses(37). Two studies have combined MIBG with a camptothecin, with tolerable toxicity and measurable responses, including the European study of topotecan with double infusion of MIBG(38), and the NANT Phase I study of irinotecan and vincristine with MIBG (39). Recently, preclinical data showed additive growth inhibition of the combination of a histone deacetylase inhibitor, vorinostat, with radiation in a metastatic murine neuroblastoma model (40). In addition, vorinostat increased uptake of MIBG in the neuroblastoma tumors, via increased expression of hNET(41). These findings led to a NANT phase I study of vorinostat combined with ¹³¹I-MIBG, which is currently continuing enrolling at Dose Level 5 with no dose-limiting toxicity to date (MIBG, 18 mCi/kg; vorinostat, 230 mg/m²). Other radiosensitizers are under study in pre-clinical testing for neuroblastoma, which may prove useful in the future (42).

A few recent studies have tested the incorporation of MIBG therapy into induction for newly diagnosed patients (Table 1). Forty-one patients at an Amsterdam center treated with two cycles of ¹³¹I-MIBG prior to addition of chemotherapy had a response rate to the MIBG of 66% (43). German studies evaluated the addition of MIBG therapy at the end of induction and prior to myeloablative therapy for patients with residual MIBG positive disease, with a response rate of 46%, but no improvement in overall survival (44). A Children's Oncology Group pilot trial (ANBL09P1) will test the addition of MIBG with vincristine and irinotecan prior to myeloablative therapy for all high-risk patients, regardless of residual disease, and if feasibility and tolerability is demonstrated, a randomized trial will be undertaken.

Future perspectives for MIBG

¹³¹I-MIBG therapy is a promising strategy which now requires randomized testing in newly diagnosed high risk patients. The practicality of this targeted radiopharmaceutical is rapidly increasing, with nine North American and at least seven European pediatric centers regularly administering MIBG therapy, and more centers in development. Due to the whole body radiation accompanying the tumor dose, further testing of compounds that will increase tumor uptake and sensitivity without increasing normal organ toxicity is essential, as is the investigation of different radioisotopes, such as the alpha emitter, astatine-211 (45). Better pre-therapy tumor dosimetry using SPECT-CT with tracer doses of ¹³¹I-MIBG (46), or PET-CT with ¹²⁴I-MIBG (47) may also help to personalize the dose.

First generation anti-GD2 mAbs

14G2a is an IgG 2a murine anti-GD2 monoclonal antibody. In two phase I trials, the dose of mAb14G2a was escalated from 25 to 500mg/m²/course (Table 1B). Toxicities mainly consisted of reversible pain, tachycardia, fever, changes in blood pressure, hyponatremia and urticaria, and were more severe in adult patients. Pain is thought to occur due to binding of antibody to peripheral nerve fibers expressing GD2 (50). These two phase I trials included 19 neuroblastoma, 3 osteosarcoma and 11 melanoma patients. Therapeutic activity was observed, with 1 complete response and 2 partial responses in neuroblastoma patients, and 6 minor responses including 3 in neuroblastoma patients (51, 52). Pharmacokinetic studies revealed a beta t ½ of 18.3 ± 11.8 h (53). As antibody-dependent cellular cytotoxicity (ADCC) is the key antitumor mechanism of therapeutic antibodies, and IL-2 was shown to augment lymphocyte-mediated ADCC in vitro (54) and anti-tumor activity of 14G2a in vivo (55), a phase I trial of14G2a in combination with IL2 was conducted in 31neuroblastoma patients and 2 osteosarcoma, with1 complete response in osteosarcoma and 1 partial response in neuroblastoma patients (56).

3F8 is a murine IgG3 anti-GD2 antibody against GD2 developed in the 1980's, with similar side effects and indications of anti-neuroblastoma activity as 14G2A (57). A phase II study of 3F8 showed that 13/34 patients with Stage 4 neuroblastoma in first or subsequent response remained progression free for 40–130 months (57). A follow-up phase II report of 3F8 + GM-CSF in 136 patients showed an overall 5-year EFS of 38% for patients without prior relapse, and a better outcome for patients with the FCGR2A (R/R) genotype polymorphism, which favors the binding of the IgG3 antibody. (58)

MAb Ch14.18 consists of the variable regions of murine IgG3 anti-GD2 mAb 14.18 and the constant regions of human IgG1-κ (59). Phase I clinical trials confirmed activity in relapsed neuroblastoma and a similar toxicity profile as mAb14G2a (60, 61). As expected, the half-life of ch14.18 was longer than 14G2a, with a beta t 1/2 of 66.6_+27.4h. (60, 62). Since GM-CSF not only raises the number of leukocytes but also enhances their anti-GD2 mediated ADCC (63), a pilot study of ch14.18 + GM-CSF showed responses in patients with recurrent/refractory neuroblastoma (64). This led to a phase II national study which confirmed the efficacy of ch14.18 + GM-CSF, with 2CR, 2 PR, 1 MR and 2 SD in 32 patients with recurrent/refractory neuroblastoma (65). Since most responses were in bone marrow or bone, it was hypothesized that this approach would be most effective in the setting of minimal residual disease (MRD). Subsequently, the feasibility of giving ch14.18 in combination with GM-CSF, IL2 and isotretinoin during the early post-transplant period was demonstrated in 2 pilot-phase I studies (66, 67).

These clinical trials led to the pivotal randomized COG phase III study, to determine if immunotherapy with ch14.18 combined with GM-CSF and IL2 on the backbone of isotretinoin would improve survival compared to isotretinoin alone for children with high-risk neuroblastoma in first response after myeloablative therapy and stem cell rescue. Eligible patients were randomized after stem cell transplantation to 6 cycles of isotretinoin (standard) or isotretinoin with 5 intercalated cycles of ch14.18 combined with GM-CSF or IL2 in alternating cycles (immunotherapy). Analysis of 226 eligible patients showed that EFS was significantly higher for 113 patients randomized to immunotherapy, with 2-yr estimated EFS from randomization of 66%±5% vs. 46%±5% (p=0.0115) for the 113 randomized to istotretinoin alone. Overall survival (OS) was also significantly higher for immunotherapy group (86%±4% vs 75%±5% at 2 yrs, p=0.0223) (5). This major advance is the first effective immunotherapy for high-risk neuroblastoma and also the first successful immunotherapy to target a non-protein antigen.

Although ch14.18 in combination with IL2 and GM-CSF has been shown to improve the outcome of high-risk neuroblastoma, the treatment is associated with significant toxicities, especially in cycles containing IL2. Furthermore, the exact contribution of cytokines in vivo remains unclear. To address these issues, SIOPEN is conducting a study that randomizes patients with high-risk neuroblastoma to ch14.18 alone or in combination with subcutaneous infusion IL2 to ameliorate toxicities associated with intravenous administration of IL2.

Second generation GD2-targeted immunotherapy and future perspectives

With the successful demonstration that ch14.18 + cytokines significantly improved outcome of patients with high-risk neuroblastoma, a COG phase III trial is ongoing to collect comprehensive toxicity data for regulatory approval of ch14.18. In addition, immunophenotypes such as FCGR3A and FCGR2A polymorphism (58, 68) or Killer-Immunoglobulin-like Receptors (KIR) -ligand mismatch (69) that may affect ADCC activity and thereby clinical response to ch14.18 is under investigation. Meanwhile, second-generation anti-GD2 antibodies and an anti-idiotype antibody vaccine have been developed and are currently in early phase clinical trials.

Hu14.18-IL2 is a fusion protein of humanized anti-GD2 antibody (hu14.18) and IL-2. The maximum tolerated dose in a phase I trial was 12 mg/m²/d, approximately 50% that of ch14.18. Clinical toxicities were similar to those reported with IL-2 and anti-GD2 mAbs, and antitumor activity was noted in three of 27 neuroblastoma patients, although there were no measurable complete or partial responses (70). A phase II study of this immunocytokine showed 5 CR in 23 patients with neuroblastoma evaluable only by MIBG and/or bone marrow histology, but no responses for patients with measurable disease (71). In this study, patients with KIR -ligand mismatch seemed to be associated with better clinical response to immunotherapy with anti-GD2. (69) A larger Phase II study adding GM-CSF is ongoing in the COG.

Hu14.18K332A is a humanized ch14.18 with a mutation to alanine at lysine 322 that limits its ability to fix complement and thereby reduces the pain associated with ch14.18, while retaining ADCC capabilities. Preclinical studies in rats confirmed that hu14.18K322 elicited significantly less allodynia than ch14.18 (72). Preliminary findings of a phase I clinical trial of hu14.18K322 showed the MTD to be 70 mg/m²/d × 4 with reduced neuropathic pain (73).

mAb1A7 is an anti-idiotype antibody directed against a murine anti-GD2, 14G2a, and in effect, mimics the GD2 antigen. Yu et al conducted a clinical trial of mAb 1A7 as a surrogate GD2 vaccine in 31 patients with high risk neuroblastoma who achieved first or subsequent responses (74). There were no systemic toxicities seen with subcutaneous injections given periodically over two years, but only local reactions, transient fever in four patients and serum sickness in one. All patients generated anti-mAb1A7 and their immune sera displayed CDC and ADCC activities. Sixteen of 21 patients who enrolled during first remission had no evidence of disease progression at a median of 6 years, while only 1 of 10 patients in second remission remains progression free. These findings indicated that mAb1A7 vaccine has little toxicity, is effective in inducing biologically active anti-GD2, and may be useful in controlling MRD (75) (unpublished data from Yu et al).

In light of the observed low toxicity profile of mAb1A7, and hu14.18K332A, it will be desirable to replace ch14.18 with one of these products, or possibly to substitute the Hu14.18-IL2 for the treatment with the ch14.18 with exogenous IL2. Thus, it will be important to ascertain whether mAb1A7 vaccine or the mutant anti-GD2 mAb will have similar therapeutic efficacy as ch14.18 in future clinical trials. Given the documented synergism between chemotherapy and anti-cancer mAbs such as rituximab (76) and trastuzumab (77), it will be timely to assess the efficacy of anti-GD2 mAbs with chemotherapy for treatment of neuroblastoma. Recently, genetic engineering of human T lymphocytes to express GD2-directed chimeric antigen receptors (CAR) has been produced by grafting the specificity of 14G2a onto a T-cell receptor (78). These CAR-T cells have been used to mediate tumor regression in patients with neuroblastoma and offer another testable strategy for GD2-directed immunotherapy (79).

Crizotinib and other ALK inhibitors

The first drug to be FDA approved for the treatment of ALK-rearranged cancers was crizotinib, an orally bioavailable ATP-competitive 2,4-pyrimidinediamine derivative (PF2341066; Pfizer Inc.)(84). Crizotinib binds to the inactive conformation of ALK and has shown striking efficacy against ALK-rearranged tumors such as non-small-cell lung cancer (NSCLC) and inflammatory myofibroblastic tumor. In early phase clinical testing, the overall response rate was 57% in 82 patients with EML4-ALK positive NSCLC (10) (80, 81). Currently, crizotinib is being tested in an ongoing phase I/II trial for children with neuroblastoma and other solid tumors bearing ALK mutations and rearrangements (NCT00939770, ClinicalTrials.gov). Preclinical testing has demonstrated the sensitivity of neuroblastoma cell lines with ALK amplification and the R1275Q mutation to crizotinib, marked by complete and sustained regression of xenografts (85), By contrast, the F1174L mutation, which possesses greater transforming capacity than other ALK mutations (83), and is associated with a poor response to standard therapy (83), has shown only minimal response in cell lines and none at all in neuroblastoma xenografts bearing this mutation (85). ALK^F1174L also mediates acquired resistance to crizotinib in adults with ALK translocation-positive cancers (86). The structural basis of this resistance is not yet understood, but could reflect the observation that this mutation is not in direct contact with the crizotinib-binding site within the ATP-binding pocket (86) or that it has a greater affinity for ATP than crizotinib (85). In view of the lack of sensitivity of the ALK^F1174L to crizotinib, the pediatric phase I study has been amended to increase the recommended phase II dose based on preclinical data that suggests that the resistance can be overcome in a dose-dependent manner (85).

Aggressive efforts to develop new ALK inhibitors are under way. An example is CH5424802 (Chugai Pharmaceuticals Ltd) (87), an orally available, benzo[b]carbazole derivative that potently inhibited the growth of neuroblastoma cells expressing amplified ALK. This agent also displayed in vitro inhibitory activity against neuroblastoma cells expressing ALK^F1174L, which was shown to be comparable to that achieved in cells with wild-type ALK. Whether CH5424802 will demonstrate efficacy for in vivo models of neuroblastoma with mutated ALK must be determined. Additionally, LDK378 (Novartis Pharmaceuticals), an orally active inhibitor of the ALK kinase, is currently being tested in Phase 1 trials in adult patients (NCT01283516, ClinicalTrials.gov), and may hold promise for the treatment of neuroblastoma patients with mutated ALK.

As seen with other kinase inhibitors, eventual resistance to crizotinib and other ALK inhibitors will likely develop, either through mutations within the ALK kinase domain or by upregulation of an alternative signaling pathway (86, 88). This phenomenon has already been described in adult patients, and will undoubtedly emerge in children who initially respond to crizotinib. Preclinical models of resistance are needed that would enable prediction of resistance mechanisms and design drug combinations to delay or circumvent resistance. Effective targeting of resistance mutations would be aided by resolution of the crystal structure of ALK in its active confirmation, and of the F1174L mutant in particular.

Future Prospects for clinical application of ALK targeting

Although the number of patients whose tumors have ALK aberrations is approximately half that associated with MYCN oncogene amplification, there are several reasons why inhibition of ALK would be a feasible therapeutic option in neuroblastoma. Firstly, ALK is expressed on the surface of most neuroblastoma tumor cells (89) and is restricted to the brain following development. Thus ALK, like, GD2, would be an ideal tumor-associated antigen for targeting with immune-based therapies (Figure 3). Anti-ALK antibodies may be beneficial not only to patients with aberrant ALK in their tumors, alone or in combination with small molecule inhibitors (90), but also to the larger number of patients whose tumors express wild-type ALK.

Secondly, a proportion of neuroblastoma cells overexpressing phosphorylated ALK that is neither mutated nor amplified respond to ALK depletion by undergoing apoptosis. Such tumor cells are also responsive to ALK inhibition by small molecules (91) and may be even be amenable to RNA interference (RNAi)-based therapeutic strategies (92). Thus, ALK is apparently activated in such cells by alternative mechanisms, suggesting that ALK inhibition may afford a useful therapeutic strategy for a larger subset of patients than previously projected.

Thirdly, ALK inhibition may provide an effective targeting strategy against MYCN-amplified tumors, which comprise nearly half of all high-risk neuroblastomas. ALK F1174L co-segregates with MYCN amplification in patients, and this combination is associated with a particularly poor prognosis, as demonstrated by the fatal outcome of 9 of 10 children with ALK F1174L/MYCN amplified tumors(83). ALK F1174L also has been shown to potentiate the oncogenic activity of MYCN in vivo, although by itself, ALK F1174L appears to be insufficient to cause the malignant transformation of neural crest derived m cells (93, 94). This latter observation may in part explain the occurrence of ALK mutations in low stage neuroblastomas. Both wild-type and F1174L ALK upregulate MYCN expression in vitro (95), and in vivo (R. George, unpublished). Moreover these two genes signal through a common signaling pathway, PI3K/AKT/mTOR, and targeting ALK with short hairpin RNAs (shRNA) has been shown to inhibit cell growth in tumor lines with concomitant MYCN amplification (6). Together, these observations suggest that inhibition of ALK, whether wild-type or mutant, could be a means to induce the death of MYCN-amplified tumor cells, either alone or when combined with emerging strategies that target MYC, such as BET bromodomain inhibition (96).

Neuroblastoma has been at the forefront of pediatric cancers in terms of stratification of patients based on molecular aberrations, the prototypical example of which is MYCN amplification. More recently, the discovery of ALK mutations has bolstered the prospects of effective molecularly targeted therapy for high-risk tumors, although considerable work needs to be done before ALK inhibition can occupy the same high-priority ground as hNET-MIBG and anti-GD2 antibody therapy.