Immunotherapeutic Strategies for Neuroblastoma: Present, Past and Future

Neuroblastoma is the most common extracranial pediatric solid tumor with a heterogeneous clinical course, ranging from spontaneous regression to metastatic disease and death, irrespective of intensive chemotherapeutic regimen. On the basis of several parameters, children affected by neuroblastoma are stratified into low, intermediate and high risk. At present, more than 50% of high-risk patients with metastatic spread display an overall poor long-term outcome also complicated by devastating long-term morbidities. Thus, novel and more effective therapies are desperately needed to improve lifespan of high-risk patients. In this regard, adoptive cell therapy holds great promise and several clinical trials are ongoing, demonstrating safety and tolerability, with no toxicities. Starting from the immunological and clinical features of neuroblastoma, we here discuss the immunotherapeutic approaches currently adopted for high-risk patients and different innovative therapeutic strategies currently under investigation. The latter are based on the infusion of natural killer (NK) cells, as support of consolidation therapy in addition to standard treatments, or chimeric antigen receptor (CAR) T cells directed against neuroblastoma associated antigens (e.g., disialoganglioside GD2). Finally, future perspectives of adoptive cell therapies represented by γδ T lymphocyes and CAR NK cells are envisaged.


General Features of Neuroblastoma
Neuroblastoma (NB) is the most widespread extracranial solid tumor in childhood [1,2] and the most common malignancy in the first year of life. The median age at diagnosis is 18 months and 90% of cases are diagnosed before the patient is 10 years of age [3]. It arises from the sympathetic nervous system, mainly from the adrenal medulla [4] and, in most cases, affects the adrenal glands or sympathetic ganglia [5]. At clinical presentation, NB is quite heterogeneous, ranging from asymptomatic tumors to diffuse metastases with systemic manifestations, thus reflecting on differences in outcome of patients that may evolve into spontaneous regression as well as into unfavorable progression, metastasis and death, irrespective of the intensive therapies adopted [6]. Metastases are present at diagnosis in about 50% of patients and mainly involve bone marrow (BM), bone and regional lymph nodes, while involvement of the central nervous system and lungs is rare, being present in less than 5% of metastatic patients at diagnosis [7,8]. In addition, extensive liver involvement may be observed in infants and causes liver disease, renal and lung dysfunction as a consequence of abdominal distention.
Diagnosis is based on a combination of laboratory tests, radiographic imaging and pathology, but many additional biological factors may be helpful to predict the clinical behavior in NB, including histologic and cytogenetic features, as well as molecular changes, in particular the amplification of the MYCN oncogene [9,10].
Remarkable efforts have been done by the International Neuroblastoma Risk Group (INRG) with the help of international groups, i.e., the Children's Oncology Group and the International Society of Paediatric Oncology European Neuroblastoma, that created a cooperative task force in order to identify homogeneous risk groups before any treatment [11]. The extent of disease was determined by the presence or absence of image defined risk factors and/or metastatic disease at the time of diagnosis, defining disease stages as local (L1 and L2) or metastatic (M and MS). Furthermore, risk stratifications were defined including not only the stage, but different aspects of tumor biology [12] (Table 1). The INRG collected data from over 8000 patients thanks to cooperation with different groups in North America, Europe and Japan and, when available, explored 35 different potential risk factors including prognostic factors such as age at diagnosis, pathology and genomic characterization (e.g., MYCN amplification and 11q status, cell ploidy and segmental chromosomal abnormalities), comparing these features to event-free and overall survival. Such efforts were of particular relevance since the precise risk stratification of patients were needed to guide therapy, improve the outcome for high-risk patients by intensification or changing treatment, and modify appropriately the chemotherapy for lower risk patients, with the aim of minimizing toxicity and late effects.
Thus, the INRG classified patients as low, intermediate or high risk: for the low and intermediate risk patients high overall survival greater than 90% has been achieved, while minimizing therapy [13][14][15]. By contrast, the high-risk patients show overall poor longterm outcome also complicated by devastating long-term morbidities, indicating that this group is specifically associated with chemo-resistance. The overall survival of high risk patients has improved over the past 20 years, from 29% for patients diagnosed from 1990 to 1994 to 50% for patients diagnosed from 2005 to 2010 [16,17]. Such results were presumably due to the intensification of therapy, myeloablation and immunotherapy, but prognosis of these patients still remains unsatisfactory. Nonetheless, patients with refractory or relapsed NB can rarely be cured and for this reason novel efficacious therapies are urgently needed.

Conventional Therapies for High Risk Patients
High-risk patients require intensive and complex therapies that include (i) the induction phase with multiple cycles of chemotherapy before surgery, (ii) a consolidation phase which may include myeloablation and autologous hematopoietic stem cell transplantation, local radiation and anti-disialoganglioside GD2 antibodies (Ab) and (iii) a maintenance phase with immunotherapy and/or differentiation agents [2].
The most widely used conventional cytotoxic chemotherapies are topotecan with either cyclophosphamide or temozolomide [18] or irinotecan and temozolomide [19][20][21] that may offer partial or even complete response with improvement in symptoms and quality of life, especially for low or intermediate risk patients.
At the end or soon after the end of induction chemotherapy, a surgical resection of the tumor mass, when possible, is applied in order to eliminate the remaining primary tumor.
Concerning the consolidation phase, it has been reported that myeloablation may significantly improve the outcome [22][23][24]. Although the autologous hematopoietic stem cell transplantation is commonly used, only marginal effects on event-free survival have been obtained and for this reason the optimal conditioning regimen is still under investigation. In this regard, long-term cures have been achieved by induction and stem-cell transplantation followed by anti-GD2 Ab therapy [25]. Alternatively, radiation therapy can be used locally.
The maintenance phase is generally composed of a combination of anti-GD2 Ab (that will be discussed in the following paragraph) and isotretinoin, known for its ability to induce differentiation and death in tumor cells, finally improving event-free survival in a randomized trial [23]. A phase III clinical trial is still active to test the side effects and efficacy of treating patients with NB (NCT01041638).
Another therapy is represented by the use of metaiodobenzylguanidine (MIBG), based on the finding that 90% of NB tumors express the norepinephrine transporter and therefore take up the sympathomimetic MIBG [26]. Clinical trials conducted in relapsed or refractory high-risk NB patients, using a high dose of 131 I-MIBG as monotherapy or in combination with other agents, demonstrated a 30-40% response rate [27][28][29].
To complicate this picture, the tumor microenvironment is hypoxic thus contributing to a metabolic challenge for tumor cells themselves [37][38][39] and for infiltrating immune cells, leading to immune-suppression. The latter finding is supported by infiltration of tumor-associated macrophages (TAM) that are effective in paralyzing T cell responses, inducing T cell apoptosis through Fas-Fas ligand interactions and activating myeloid derived suppressor cells and regulatory T cells that, in turn, suppress active immune response [40][41][42]. Furthermore, NB cells express constitutively high levels of gangliosides which further contribute to the immune suppressive microenvironment. Of these, GD2 is one of the most studied surface antigens in NB also used for target therapies [43,44] 4. Immunotherapeutic Approaches for Neuroblastoma 4.1. Antibodies Targeting NB in Clinical Settings: the GD2 Disialoganglioside Prototype and Other Tumor-Associated Antigens GD2 is an oncofetal antigen expressed during fetal development. It remains expressed after birth in neurons, peripheral nerves and skin melanocytes and it is consistently found in NB and osteosarcoma [43]. In these tumors, specific Ab targeting GD2 have been developed and used in clinical settings with encouraging results [25,45,46].
The effectiveness of GD2 Ab is related to different mechanisms as (i) direct induction of cell death, (ii) Fcγ receptor (FcγR)-mediated antibody-dependent cell-mediated cytotoxicity (ADCC) by immune cells such as NK, macrophages and neutrophils and (iii) complement-dependent cytotoxicity (CDC) [47][48][49][50]. Of note, these mechanisms of cytotoxicity may be highly potentiated by the use of immune stimulating cytokines or adoptive cell therapy. FDA approved two different GD2 Abs for clinical trials that are dinutuximab (Ch14.18, chimeric murine/human antibody to GD2) and naxitamab (hu3F8, a humanized murine antibody to GD2) [45,46,51], currently used in clinical trials in recruiting phase or still active.
Only one of these clinical trials reported data regarding progression-free survival (PFS) and overall survival (OS) comparing patients treated with irinotecan and temozolomide in combination with dinutuximab or temsirolimus [52]. Improved PFS and OS (76.5% and 88.2%, respectively) were observed in patients receiving dinutuximab compared to those treated with temsirolimus (24.7% and 64.7%, respectively) Figure 1 summarizes the antibodies used for NB immunotherapy with corresponding antigens and mechanisms of action. These antibodies specific for NB-associated antigens may induce direct cell death, complement-dependent cytotoxicity (CDC) and antibody-dependent cell-mediated cytotoxicity (ADCC). In addition, cytokines such as IL-2 and GM-CSF may increase the anti-tumor activity exerted by these antibodies.  These Abs showed similar clinical impact and toxicities mainly manifested as neuropathic pain, fever and allergic reactions. However some differences have been reported probably due to their different half life [30,53].
Cytokines are the most factors used in combination with a GD2 antibody, especially IL-2 and granulocyte-macrophage colony stimulating factor (GM-CSF). IL-2 induces activation T and NK cells [54] and it was approved by FDA for treating adult tumors. In children with high-risk NB several phase I and II trials tested IL-2 as monotherapy, but no significant effective results have been reported [55][56][57]. Recently, Ladestein et al. demonstrated only marginal effects of dinutuximab beta in combination with IL-2. In fact, the 3-year event-free survival was 56% (95% CI 49-63) with dinutuximab beta (83 patients had an event) and 60% (53-66) with dinutuximab beta and subcutaneous IL-2 (80 patients had an event; p = 0.76) [25,48]. By contrast, IL-2 administration with alternating cycles of GM-CSF in combination with dinutuximab resulted in higher rates of event-free (66% versus 46%) and overall survival after 2 years, compared to standard therapy alone (86% versus 75%) [25,48]. GM-CSF is a myeloid growth factor that stimulates differentiation of progenitor cells into granulocytes and monocytes, and boosts immune responses [58]. A human recombinant GM-CSF, named LEUKINE ® (sargramostim), has been approved by FDA. Several clinical trials evaluating the role of GM-CSF combined with anti-GD2 in highrisk NB were conducted over the last decade (Table 2). Although this immunotherapy has shown substantial anti-neuroblastoma activity (NCT01757626) [51] even against minimal residual disease (NCT02100930), patients should be closely monitored for the onset of posterior reversible encephalopathy syndrome (NCT01183897) [26] (Table 2). Table 2. Clinical trials using granulocyte-macrophage colony stimulating factor (GM-CSF) and anti-GD2 in the last ten years.  Abbreviations: 3F8, a murine immunoglobulin 3 monoclonal antibody specific for disialoganglioside; Hu3F8, Humanized murine IgG3 anti-GD2 antibody m3F8; GM-CSF, subcutaneous granulocyte-macrophage-colony-stimulating factor.

Other NB Associated Antigens as Targets for Antibody-Mediated Immunotherapy
Another target recently considered for development of therapeutic approaches against high risk patients is represented by anaplastic lymphoma kinase (ALK) that was found to be mutated in 9% of NB [59,60]. In 2009 the first clinical trial in pediatric patients with refractory solid tumors was conducted by the Children's Oncology Group using an ALK inhibitor (i.e., crizotinib) providing discouraging results [61]. Afterwards, better results were obtained using other ALK inhibitor compounds, alone or in combination with chemotherapy, in solid tumors such as lung cancer, but some problems emerged due to secondary mutation and amplification of ALK and off-target mechanisms including activation of 'bypass' signaling pathways [62][63][64]. Recently, Sano R. et al. developed an Ab drug conjugate directly targeting ALK receptor, the CDX-0125-TEI, that exhibited cytotoxicity against both wild-type and mutated ALK in patient-derived xenografts [65] ( Figure 1).
Finally, an additional interesting target for NB therapy is B7-H3 (CD276), a type I transmembrane glycoprotein molecule homogeneously expressed in both primary and metastatic NB (Figure 1) [62]. Loo D. et al. developed the anti B7-H3 Ab MGA27 that demonstrated to mediate potent cytotoxicity against a broad range of tumor cell types in xenograft models, while avoiding toxicities [66]. This antibody, later named as enoblituzumab, is currently in phase I trials for diverse solid tumors including refractory tumors and pediatric cancers.

Adoptive Cell Therapy Based on NK Cells
NK cells represent a subset of lymphocytes belonging to innate immune system, which originate in the BM and exert anti-tumor and anti-viral activities [67][68][69]. NK cell functions are regulated by the integration of multiple signals derived from activating and inhibitory receptors which bind specific ligands expressed on tumors and viral-infected cells. NK cells may be expanded ex vivo and stimulated by different cytokines, including IL-15, IL-12 and IL-18 [67][68][69]. Since NK cells do not cause graft versus host disease (GvHD), they represent an ideal source for allogeneic "off-the-shelf" cellular therapy. The development of therapeutic protocols for NB patients, based on the infusion of NK cells, arose from the success obtained in different clinical trials using monoclonal antibodies targeting NB. In these trials, the observed complete remission or stabilization of the disease was related to an increase of NK cell frequency and activity, including NK cell-mediated ADCC [70][71][72][73][74]. Of note, NK cells are known for their ability to target and lyse tumor cells lacking HLA-class I molecules, which represent the ligands of killer inhibitory receptors [75]. For this reason metastatic NB cells, which show low to absent HLA-class I expression [76], paralleled by the presence of different ligands for NK cell activating receptors [77], represent an ideal target for NK cell mediated lysis.
Different preclinical studies on NB have been described using NK cells. Castriconi et al. [78] analyzed the therapeutic effect of infusion of NK cells activated in vitro with IL-2 in a xenograft model of metastatic NB. SCID/NOD mice, inoculated intravenously with the HTLA-230 NB cell line, showed significant prolonged survival when injected with multiple NK cell doses soon after the inoculation of tumor cells. Such therapeutic effect was further increased by treatment with IL-2 and IL-15 that stimulated NK cells through up-regulation of the activating markers CD69 and CD25 as well as of NKp44 and DNAM-1 receptors [78]. Similarly, Liu and coworkers [79] tested NK cells, either from normal donors or NB patients, in SCID/NOD mice injected with CHLA-255-Fluc cells. At variance with the previous studies, NK cells were expanded in vitro, using a feeder composed of irradiated clinical-grade K562 cells expressing membrane-bound IL-21 (K562 mbIL-21), in the presence of IL-2. The resulting NK cells displayed high expression of DNAM-1, NKG2D, NKp46, CD56 and CD16, and efficiently lysed NB cell lines in vitro, through mechanisms such as (i) granzyme A and B release and (ii) induction of ADCC in the presence of anti-GD2 Ab. Preclinical studies highlighted a limited therapeutic effect of expanded NK cells alone, but a significant inhibition of tumor growth and a prolonged survival of mice when combined with the anti-GD2 ch14.18 Ab. These data suggested that the main therapeutic effect exerted by NK cells against NB in vivo was related to induction of ADCC rather than direct NK cell lysis [79]. The importance of this mechanism was confirmed in the study by Barry et al. that used primary NB xenografted in murine kidney, then analyzed for metastatic disease in the liver and BM. In this model, the administration of NK cells, expanded in vitro using K562 and IL-21, in combination with dinutuximab, not only reduced metastasis in the liver and BM, but also prolonged survival [80].
Starting from the concept that TGFβ is an immunosuppressive cytokine released by NB cells able to down-regulate activating receptors on NK cells and to inhibit the release of granzyme and perforin, Tran and coworkers performed preclinical studies using the TGFβR1 inhibitor galunisertib. They reported that mice, injected in the kidney with NB cell lines and treated with galunisertib plus expanded NK cells and dinutuximab, showed prolonged survival associated with reduction of tumor growth, as compared to mice treated with NK cells and dinutuximab alone. Thus, they paved the way to clinical application of this combined therapy for high-risk NB patients [81].
Clinical trials using NK cell infusions for NB patients are still ongoing (Table 3). A pilot trial was performed by Federico et al. [82] on NB patients that had undergone six different courses of induction chemotherapy, including cyclophosphamide and topotecan, or irinotecan and temozolomide, or ifosfamide, carboplatin and etoposide. Each group was treated with the anti-GD2 Ab hu14.18K322A, GM-CSF and IL2, in the presence or absence of haploidentical parental NK cell infusion. Such GMP-graded NK cells were isolated from leukapheresis by CliniMACS and infused at a median dose of 15.5 × 10 6 /Kg. NK cell infusion was safe and well tolerated, although toxicities related to chemotherapy, anti-GD2 or GM-CSF infusion were observed. The overall response rate of patients was 61.5%, with a complete response/very good partial response rate of 38.5%. More importantly, progression of the disease was never observed. However, the contribution of NK cells to therapeutic success in this limited study was not assessed [82]. A phase II clinical trial was performed in NB patients subjected to multiple cycles of induction chemotherapy with different drugs (including cyclophosphamide, topotecan, cisplatin, etoposide, doxorubicin and vincristine) and infused with the anti-GD2 Ab hu14.18K322A with GM-CSF and IL-2 [83]. Patients were treated, as consolidation therapy, with busulfan/melphalan plus infusion of autologous hematopoietic stem cells (collected in the induction therapy phase). In addition, some patients were treated with an experimental immunotherapy, composed of the hu14.18K322A Ab and GM-CSF in the presence or absence of GMP-graded haploidentical NK cells, which were purified using CliniMACS system and not expanded ex vivo. No difference in toxicities was observed between patients infused with or without NK cells. NK cells were still detected until 18 days post-infusion demonstrating an NK cell engraftment [83]. A longitudinal analysis of this clinical trial revealed that patients receiving haploidentical NK cell infusion had an increased NK cell count at day 7, and a decreased NK cell count at day 21 post-infusion [84]. However, donor NK cells infused at higher doses were still detected after 21 days. Such cells were able to respond to ex vivo stimulation with IL-15 and IL-2 and to lyse K562 cell line in vitro, thus suggesting that NK retained their cytotoxic function. Finally, patients with stable disease displayed a higher number of alloreactive NK cells than those with residual disease [84].
Modak and coworkers [85] performed a phase I clinical study on NB patients receiving an induction therapy with vincristine, cyclophosphamide and topotecan, and then different doses of haploidentical parental NK cells and the anti-GD2 murine Ab 3F8. In this study, NK cells were isolated by CliniMACS and then cultured overnight with IL-2 before being infused. Toxicities were almost related to the infusion of 3F8 Ab, whereas NK cell administration was safe and well tolerated. Moreover, they demonstrated that patients with detectable NK cells in the periphery at day 14 display a complete remission and, in general, patients who received a higher dose of NK cell infusion showed an improved progression-free survival compared to those receiving lower NK cell doses (p = 0.018) [85].   Of note, Heinze and coworkers [86] recently described different methods to expand NK cells in vitro for clinical use analyzing viability, purity and composition of NK cell preparations starting from normal donors' buffy coat preparations. Either CD56 + enriched cells and CD3 + /CD19 + depleted cells were cultured in media supplemented with different combinations of cytokines (i.e., IL-2+IL-15 or IL-15+IL-21). The authors concluded that the highest purity and expansion rate can be obtained starting from CD3 + /CD19 + depleted cells using IL-15 and IL-21. These cells showed also increased cytotoxicity and degranulation, when cultured in the presence of NB cell lines, thus suggesting that these NK cell preparations could be useful to test in clinical trials [86].
Collectively, these studies suggested that NK cell infusion, as consolidation therapy, in NB patients is feasible and well tolerated. Therapeutic effects of NK cells against NB hold promise and have to be carefully evaluated in phase III randomized clinical trials. At present, no phase III/IV clinical trials using NK cell infusions are ongoing, whereas 20 phase I/II trials are registered (clinicaltrials.gov), and some of them completed (Table 3).

Another innate cytotoxic cell population, with intermediate features between NK and T cells, is represented by NKT cells which express an invariant
TCRα chain, Vα24-Jα18, and are able to recognize self-and microbial-derived glycolipids presented by the monomorphic HLA class I-like molecule CD1d [87]. The contribution of NKT cells to antitumor responses has been demonstrated in different tumor models [88][89][90], as underlined by the finding that these cells are decreased in number and/or functionality in cancer patients [91][92][93]. NKT-based clinical trials are ongoing for several malignancies, and are based on the injection of alpha-Galactosylceramide (α-GalCer) or α-GalCer-pulsed APC to expand NKT cells in vivo. Alternatively, autologous NKT cells have been expanded ex vivo using IL-2 and anti-CD3 mAbs [94]. Recent evidence suggested that NKT cells may have anti-tumor activities in NB patients. Liu et al. [95] reported that NKT cell migration towards a co-culture of NB cells and monocytes was increased in hypoxic conditions, a typical feature of NB microenvironments. Indeed, hypoxia down-regulated the secretion of the tumor-associated chemokine CCL2 (which attracts NKT cells) by NB cells, whereas the chemokine CCL20 was increased upon co-culture of NB cells with monocytes. Accordingly, neutralizing monoclonal Abs against CCL20 abrogated NKT cell migration in hypoxic conditions [95]. The authors demonstrated that humanized NOD/SCID Il2rg −/− (NSG) mice inoculated with NB cells in the renal capsule showed a high infiltration of human TAM in the tumor microenvironment, with a high percentage of M2 macrophages. Afterwards, they inoculated expanded NKT cells that highly infiltrated the tumors, but this feature was abrogated by the treatment of mice with anti-CCL2 and anti-CCL20 mAbs. Furthermore, expanded NKT cells, inoculated in preclinical model of metastatic NB, preferentially migrated in the hypoxic areas of metastasis in liver and BM, were inhibited by hypoxia and by contact with tumor cells [95]. To improve NKT cell functions, they transduced NKT cells with a cDNA encoding human IL-15 (NKT/IL-15), demonstrating that the latter cells were able to proliferate in hypoxic conditions in vitro, whereas parental NKT cells were inhibited in the same settings. Finally, they setup a metastatic model of NB in NSG and humanized NSG mice, showing that tumors developed more rapidly in the latter mice, due to the infiltration of human hematopoietic cells in the tumor. Such tumor-promoting effect was totally abrogated by inoculating NKT/IL-15 (but not NKT) and restored in the presence of CD1d blocking mAbs. These findings suggest that adoptive therapy with expanded NKT cells in combination with IL-15 may represent a therapeutic strategy for patients with metastatic NB [95]. So far, no active clinical trials based on NKT cells as therapy are present for NB patients.

CAR T Cells for Therapy of High Risk NB Patients
The use of T cells genetically modified to express a chimeric antigen receptor (CAR) is a new promising approach of adoptive cell therapy in cancer, combining the antigen specificity of a monoclonal Ab with the effector function and long-term persistence of T cells [96][97][98]. CAR provided success in treating B cell acute lymphoblastic leukemias, with limited clinical benefit in solid tumors, although more than 100 clinical studies have been developed. Indeed, the success of CAR T cell therapy in solid tumors relies on the ability of CAR T cells enter in the tumor site, overcome the immunosuppressive tumor microenvironment, and persist for a long period.
CAR is composed of different domains, including (i) the single-chain variable fragment (scFv) of tumor antigen-specific Ab, (ii) a hinge region, (iii) the trans-membrane domain of CD8α or CD28 molecules and (iv) the intracellular signaling region. Thus, T cells endowed with CAR showed an improved T-cell antigen recognition, T-cell activation and tumor cell lysis. CARs are classified as first, second and third generation on the basis of the presence of one, two or more T cell co-stimulatory molecules [99,100].
First generation CAR included activating/signaling domains of CD3ζ or FcγRIII molecules, whereas second generation CAR included co-stimulatory domains of other molecules, including CD28, ICOS, 4-1BB, OX-40 and CD27). The third generation CAR are endowed with co-stimulatory domains in tandem (i.e., CD28 in combination with 4-1BB), which may increase T cell expansion and anti-tumor functions [101].
FDA approved two different products for CAR T cell therapy in 2017, named Kymriah ® (Novartis) and Yescarta ® (Kite/Gilead), for the treatment of hematological malignancies [102].
The most common side effects of CAR αβ T cells are cytokine release syndrome (CRS) and GvHD [103]. Of note, to prevent unexpected T-cell-related toxicity such as CRS, an inducible caspase-9 (iC9) gene was introduced in frame with the TCR, thus allowing the prompt elimination of genetically modified T cells by the use of chemical induction of dimerization (CID) drug, AP1903 [104,105].
Two clinical studies have been conducted with first generation CAR T cells in patients with NB [106,107]: in the first GD2 CAR T cells were well tolerated, but only one of six patients had a partial response, in the second five of 11 patients with active disease showed tumor responses and three of them had complete responses. Nonetheless, the response rates in patients with NB remain significantly lower than those observed in patients with acute lymphoblastic leukemia treated with CD19-CAR constructed with the same approach [108]. More recently, Heczey A et al. implemented a third-generation GD2 CAR in which the specific scFv was derived from the murine 14.G2a mAb coupled with the ξ-chain endodomain and two costimulatory endodomains in tandem (CD28 and OX40) [109]. CAR T cells, administered in combination with lymphodepletion to patients with NB, were well tolerated and promoted some objective clinical responses [107,109,110]. Afterwards, Quintarelli et al. [111] reported that insertion of 4-1BB costimulatory domain, in the place of OX40 in a third generation CAR, significantly improved the anti-tumor efficacy of GD2 CAR. The choice of 4-1BB signaling was particularly effective in terms of T cell activation and persistence, control of tumor cell growth and T cell exhaustion [111].
Finally, Chen and coworkers incorporated IL-15 and iC9 within the GD2 CAR construct and demonstrated that transduced T cells enriched in central memory/stem cell-like cells, expressed PD-1 at low level, promoted superior antitumor activity, expansion and survival in vitro and in vivo [112].
Several clinical trials are in progress (Table 4), including those that evaluate CAR T cells that target CD171 [113] in patients with NB and ganglioneuroblastoma (NCT02311621) or B7H3 (NCT04483778) in relapsed/refractory pediatric solid tumors. Many others continue to explore GD2-CAR T cells for NB and include NCT02761915, NCT03373097, NCT03721068, NCT03635632, NCT01822652 and NCT01953900. In addition, several studies have addressed strategies to implement the efficacy of CAR T cell therapies in NB patients [114].   (1) To assess the safety and tolerability of cellular immunotherapy utilizing ex-vivo expanded autologous T cells genetically modified to express B7H3-specific CAR (2) To assess the safety and tolerability of cellular immunotherapy utilizing ex vivo expanded autologous T cells genetically modified to express a bispecific B7H3xCD19 CAR.
To determine the maximum tolerated dose of B7H3-specific CAR. (4) To determine the maximum tolerated dose of bispecific B7H3xCD19 CAR. (5) To assess the dose limiting toxicities and describe the full toxicity profile for each study arm type, frequency, severity and duration of adverse events will be tabulated and summarized. (6) To assess the feasibility of manufacturing B7H3 specific CARs from patient-derived lymphocytes.
To assess the feasibility of manufacturing B7H3xCD19 bispecific CARs from patient-derived lymphocytes.

Future Prospects: CAR NK Cells and γδ T Cells
γδ T cells and NK cells may represent an ideal source for adoptive cell therapy due to some shared features. Both cell types recognize and kill tumor cells irrespective of the expression of a single tumor-associated antigen, thus avoiding tumor immune escape related to single antigen loss [67][68][69]94,114,115]. In addition, γδ T and NK cells act through MHC-independent recognition of target cells, thus reducing the risk of alloreactivity and GvHD [116] and are able to provide immediate release of effector cytokines in different tissues. Of note γδ T cells show a natural tissue tropism which provide a migratory advantage over αβ T cells, leading to a superior ability to infiltrate in hypoxic tumors [117,118]. Finally, γδ T cells can interact with APC and other immune cells, and also act as APCs by priming the antigens for αβ T cells, thus orchestrating the anti-tumor immune responses [119,120].
Taken together, these features render NK and γδ T cells an optimal source for immunotherapy, and different genetic engineering strategies have been developed to enhance and redirect their cytotoxicity toward tumor antigens as well as to improve their in vivo

Future Prospects: CAR NK Cells and γδ T Cells
γδ T cells and NK cells may represent an ideal source for adoptive cell therapy due to some shared features. Both cell types recognize and kill tumor cells irrespective of the expression of a single tumor-associated antigen, thus avoiding tumor immune escape related to single antigen loss [67][68][69]94,114,115]. In addition, γδ T and NK cells act through MHC-independent recognition of target cells, thus reducing the risk of alloreactivity and GvHD [116] and are able to provide immediate release of effector cytokines in different tissues. Of note γδ T cells show a natural tissue tropism which provide a migratory advantage over αβ T cells, leading to a superior ability to infiltrate in hypoxic tumors [117,118]. Finally, γδ T cells can interact with APC and other immune cells, and also act as APCs by priming the antigens for αβ T cells, thus orchestrating the anti-tumor immune responses [119,120].
Taken together, these features render NK and γδ T cells an optimal source for immunotherapy, and different genetic engineering strategies have been developed to enhance and redirect their cytotoxicity toward tumor antigens as well as to improve their in vivo or ex vivo expansion that is crucial for their broad clinical application [121].
In this regard, although γδ T cells may be difficult to expand efficiently ex vivo, sufficient numbers for adoptive transfer immunotherapy especially for pediatric patients are easily obtained. For example, the Vγ9Vδ2 population may be efficiently expanded using zoledronic acid and IL-2 both ex vivo and in vivo starting from peripheral blood mononuclear cells [99]. γδ T cell infusion proved to be safe and showed promising results in recent clinical trials. Indeed, ongoing studies are addressing the increase of γδ T cell survival and potency for clinical purposes [122].
Although CAR γδ T cells have shown preclinical efficacy, their optimization and the choice of the best co-stimulatory domains are still open [123]. γδ T cells express a wide panel of co-stimulatory molecules, including CD28, CD27 and 4-1BB that enhance γδ T cell survival and proliferation. Thus, the interaction between signals derived from γδ TCR and these molecules should be studied in detail to design more suitable CAR constructs [124].
The clinical efficacy of CAR T cells in solid tumors may be inhibited by immune suppressive conditions in the tumor microenvironment [123]. In this regard, γδ T cells may be resistant to immune-suppression due to the expression of peculiar receptors. In addition, the addition of treatments targeting the stroma and checkpoint inhibitors may improve the treatment efficacy in solid tumors.
Polito et al. [125] recently described an efficient method to produce allogeneic thirdparty and ready-to-use γδ CAR T cells. They expanded memory Vδ1 subpopulation using engineered APCs expressing CD86, 4-1BBL, CD40L and the CMV-antigen-pp65, obtaining polyclonal γδ T cells expressing activation and memory markers, with potent anti-tumor activity in vitro and in vivo with no sign of alloreactivity. Finally, the authors engineered these cells with a third generation anti-GD2 CAR (GD2.CD28.4-1BBζ), further enhancing anti-NB cytotoxicity.
NK cells represent another ideal source for allogeneic "off-the-shelf" cellular therapy, since such cells (i) do not cause GvHD, (ii) lack in vivo clonal expansion and (iii) display a limited persistence. These features reduce the risk of CRS, a life-threatening toxicity observed in several CAR αβ T cell trials [127]. However, a limitation of this therapeutic approach may be related to a different persistence between CAR T and CAR NK cells according to disease, the tumor burden, lymphodepletion settings and other factors (i.e., the potency of the administered T cells and CAR construct). Thus, some strategies have been developed to overcome this issue, including the co-transfection of stimulatory cytokines (e.g., IL-15) and multiple administration of CAR cells at different time points.
Esser [141] and Romanski [142] used amphotropic retroviral vectors to generate CAR NK cells. The construct was composed of scFv of ch14.18 anti-GD2 chimeric Ab or of anti-CD19 Ab, respectively, combined with IgVH signal peptide, myc tag, the hinge region of CD8α and trans-membrane/intracellular domains of CD3ζ chain. Such CAR NK cells recognized and eliminated GD2 expressing cells which were resistant to parental NK-92 cells. In addition, the authors observed increased killing of NK-sensitive tumor cell lines using retargeted NK-92 cells. Such lysis was strictly dependent on specific recognition of the target antigen.
The potency of this therapeutic approach was further implemented by Kailayangiri et al. [143] by the integration of co-stimulatory molecules (i.e., trans-membrane domain of CD28 and signaling domains of 4-1BB and CD3ζ) and by extended cytokine support during NK cell expansion in vitro. However, the in vivo response was low, due to the immunosuppressive activities of HLA-G molecule.

Conclusions
In the last 20 years several biological parameters that impact on the course of NB and on patients' survival have been identified. These are of particular relevance for the best risk stratification, thus identifying high risk patients who need more aggressive treatments to improve their clinical outcome. In this line, a plethora of therapeutic strategies have been setup and applied in preclinical and clinical settings for patients with refractory or relapsed tumors. Nonetheless, these patients can rarely be cured, and for this reason innovative and more effective protocols are urgently needed.
Anti-GD2 Abs represent the first immunotherapy applied in NB which greatly increase overall and progression-free survival of children, especially when used in combination with cytokines, such as IL-2 and GM-CSF.
Recent advances have been achieved using NK cells as adoptive cell therapy in support of consolidation phase, and infusion of CAR T cells. In this regard, many phase I/II clinical trials have been completed, some are still ongoing, demonstrating that adoptive cell therapy is safe, well tolerated and without relevant toxicities. However, efficacy of both adoptive cellular therapeutic strategies has to be confirmed in future phase III/IV studies. Of note, NK cells will be used as adjuvant therapy after at least one cycle of standard therapy plus anti-GD2 mAbs whereas CAR T cells will be infused in patients with relapse or refractory to conventional treatments.
Future perspectives of adoptive cell therapies for high-risk NB patients are represented by the use of γδ T cells and by the generation of novel CAR cells obtained by the transfection of NK and γδ T cells. Undoubtedly, the peculiar advantage of these cells is the generation of "off the shelf" therapies avoiding a strict selection of the donors irrespective of the recipients, with no risk of GvHD. To date, few phase I/II on NB patients are ongoing, however promising results have been obtained in other malignancies, demonstrating the safety of these therapeutic approaches.

Conflicts of Interest:
The authors declare no conflict of interest.