Neuroblastoma: Issues in transplantation

Neuroblastoma is the most common extra-cranial solid malignancy of childhood, and has a broad spectrum of clinical presentations and behavior. While low- and intermediate-risk neuroblastoma are mostly curable(1, 2), high-risk neuroblastoma has proven refractory to conventional treatment modalities(3–5). Despite the unsatisfactory responses to conventional therapies, some improvements in outcome have been achieved through the escalation of therapeutic intensity(6). Although even the most intense conventional therapy results in long-term event-free survival of well less than 40%, improvements in event-free survival (EFS) can be achieved through the addition of consolidation therapy with high-dose therapies that exceed marrow tolerance.

Initial studies

Investigators in the late 1980s and early 1990s began exploring the hypothesis that increased treatment intensity beyond marrow tolerance would improve survival in patients with high-risk neuroblastoma. Multiple early single-arm or retrospective studies suggested that autologous transplant might indeed improve the EFS of these patients, although none of the studies were randomized and may have been influenced by selection bias(7–12). The largest retrospective analysis was performed through the EBMT in 1997. See Table 1 for a summary of the various NBL transplant studies mentioned below. 1070 transplants for high-risk neuroblastoma were analyzed, and 2-year survival among the group of patients who had reached a SCT procedure was 49%. Most relapses occurred within the first 18 months following transplant, and there were no survivors amongst the group of the 48 patients who relapsed and underwent a second SCT. Notably, late relapses were found as long as 7 years from transplant(13).

Table 1.

Results from selected studies of autologous SCT in high-risk neuroblastoma^a

Group	N	Study type	EFS from	EFS	Myeloablative regimen(s)
EBMT(13)	1070	Retrospective	Transplant	49% (2 year)	Various
			Transplant	33% (5 year)
CCG 3891(4)	539	Phase III	Estimated from diagnosis	38% (3.7 year)	CEM/TBI
Grupp et al.(3)	97	Phase II	Diagnosis	55% (3 year)	No. 1 CECtx
					No. 2 melphalan/TBI
Kletzel et al.(17)	25	Phase II	Diagnosis	57% (3 year)	No. 1 CE
					No. 2 CE
					No. 3 TCtx
Villablanca et al.b	73	Phase II	Transplant	49% (3 year)	CEM
			Transplant	47% (5 year)
Kreissman et al.(25)	489	Phase III	Diagnosis	49% (2 year)	CEM
	(489 randomized, 398 transplanted)		Diagnosis	~40% (3 year)
			Transplant	46% (3 year)
Ladenstein et al(20)	1577	Phase III	Transplant	33% (3 year)	CEM
	(598 randomized)		Transplant	49% (3 year)	Bu/Mel

Abbreviations: Bu=busulfan; C=carboplatin; CCG=Children’s Cancer Group; Ctx=cyclophosphamide; E=etoposide; EBMT=European Group for Blood and Marrow Transplantation; EFS=event-free survival; M=Mel=melphalan; T=thiotepa; TBI=total body irradiation.

^aStudy populations differed significantly in these six studies. The EBMT analysis included allogeneic transplants and transplants after relapse. Villablanca, et al. included only stage 4>1 year post-SCT in the group presented.

^bVillablanca et al., unpublished data.

Table extensively adapted from Fish et al., Bone Marrow Transplantation (2008) 41, 159–165(78)

Randomized trials

The promise suggested by these early studies propelled prospective evaluation of autologous transplant for high-risk neuroblastoma. The largest of the randomized, prospective studies was the Children’s Cancer Group phase III 3891 trial. In 3891, patients were randomized to a consolidation regimen consisting of autologous bone marrow transplant versus continuation chemotherapy. Following consolidation, patients were then randomized to biologic therapy with 13-cis retinoic acid versus no further therapy(4). The study found that those treated with HDC and transplant had a significantly better event-free survival than those treated with chemotherapy alone. It was also noted that treatment with 13-cis-retinoic acid further improved the outcome among patients without progressive disease. With an estimated 38% EFS 3.7 years from diagnosis in the best group, this study helped establish autologous transplant followed by 6 months of oral cis-RA therapy as the new standard of care for these patients. Other studies of autologous SCT in high-risk neuroblastoma have since built on the results of 3891. The conditioning regimens used in these studies have varied widely, with the greatest difference being that some studies have used total body irradiation (TBI) in the conditioning regimen and others have not. There have been no randomized trials of the use of TBI during conditioning, and while it may improve outcomes, it also results in significant late effects in this young (median age 3) patient population. Overall, these studies have led to the current core standard for neuroblastoma treatment: 5–6 cycles of induction chemotherapy, surgery, radiotherapy (at a minimum to the tumor bed) and SCT followed by oral cis-retinoic acid. To this, we may now add GD2-targeted immunotherapy (including the anti-GD2 antibody 14.18, IL-2, GM-CSF and cis-retinoic acid) based on recent data from the COG ANBL0032 study, which showed a superior outcome in patients who received this immunotherapy-based treatment(14).

Tandem transplantation

Given the evidence that dose-intensity correlates with outcome, and that HDC with autologous stem cell rescue renders a statistically significant improvement in survival, it was logical to examine sequential courses of HDC with stem cell rescue, otherwise known as tandem transplant. Tandem transplantation allows for even greater dose intensity in consolidation, with the potential to introduce different active agents at each transplant. A very early attempt to employ this technique was complicated by unacceptable TRM (15). Several groups have retested the tandem transplant approach with more promising results(3, 16, 17)_ENREF_16. The largest of these studies was conducted over 6 years at 4 cooperating institutions(3, 16). The study was designed using early collection of PBSC and two myeloablative regimens containing distinct agents: (1) carboplatin, etoposide and cyclophosphamide, followed by (2) melphalan and TBI. TRM in this study was 6%, and included two patients who died of EBV lymphoproliferative disease (EBV-LPD)(18). Longer followup of this treatment approach in a large phase II cohort has demonstrated a 3-year EFS from diagnosis of consecutively enrolled patients of 55% (19). A second multiple cycle SCT study, performed using 3 sequential SCT procedures, found comparable results in terms of 3 year EFS (57%), although there appeared to be instability of the curve out to 4–5 years(17). Based on these promising results, the current open phase III COG trial, ANBL0532, is testing single versus tandem transplant as consolidation therapy for high-risk neuroblastoma.

Bu/Mel vs. CEM

The combination of carboplatin, etoposide and melphalan is currently the effective standard of care for NBL SCT in the US, having been used in COG A3973 as well as the single transplant (standard) arm of COG ANBL0532. Recently the SIOPEN group has been testing the combination of busulfan and melphalan for autologous SCT, comparing this conditioning regimen to CEM. Although these data have not yet been published, they were presented in a plenary session by Dr. Ruth Ladenstein at ASCO this year(20). Although only about a third of enrolled patients actually underwent the randomization, the study reports a statistically significant difference in outcome between the Bu/Mel and CEM arms, with 3 year EFS in the Bu/Mel being 48% compared to 33% with CEM. Importantly, although rates of treatment-related mortality were similar in the two arms, Bu/Mel was far better tolerated, with lower rates of all post-SCT toxicities with the single exception of VOD/SOSS. The authors conclude that this result establishes Bu/Mel as the new standard of care for NBL transplant. At the same session Dr. Julie Park then provided a discussion of the study, contrasting it with recent US data. Although the groups of high-risk NBL undergoing SCT in COG and SIOPEN studies are not exactly congruent (largely due to criteria for required response), it appears that 3 year EFS from transplant in patients getting CEM who did not subsequently get immunotherapy is 46%, comparable to the SIOPEN Bu/Mel result.

Of course, this raises the key question of whether Bu/Mel would or will perform better in the US context, after the induction regimen most commonly used here (a modification of the Sloan- Kettering N6 regimen). Perhaps there was a negative interaction between the SIOPEN Rapid COJEC induction, although there is no obvious hypothesis about why this should be the case. This would potentially account for the lower results observed in the CEM arm. At the same time, even similar performance between Bu/Mel and CEM after N6 induction would be a win, given the reported lower acute toxicities of the Bu/Mel conditioning regimen. Another consideration is the issue of toxicities associated with local radiotherapy. Although a standard dose of 2160 cGy and an abdominal field is well-tolerated after Bu/Mel, there is unquestionably an interaction between Bu/Mel and some uses of radiotherapy. Whole-lung irradiation (almost never used in NBL) is absolutely contraindicated after Bu/Mel because of a very high risk of pulmonary fibrosis. An unknown is how a limited field in a thoracic NBL patient would be tolerated – it appears that such patients may have been taken off the SIOPEN study to get a CEM transplant, although the published report will further clarify this. Also unknown is whether the toxicity of local radiotherapy will be augmented in patients who receive involved-field boosts up to 3600 cGy after Bu/Mel, or even if such boosts, which are currently being tested in COG ANBL0532, are advisable. The approach being taken by the COG is to move with caution. Bu/Mel will be piloted within COG for high-risk NBL patients after a US-style induction, which will give us an initial sense for potential toxicities and interactions. At the same time, the next US phase III study is being designed with the assumption that Bu/Mel may indeed be used as the conditioning regimen, once these issues are being worked out.

Tumor cell purging

In addition to increasing dose intensity, graft manipulation has been used to attempt to improve survival following autologous SCT in neuroblastoma. The most researched manipulation in the context of neuroblastoma has been purging of malignant cells prior to the infusion of the HSC product. Early research suggested that clonogenic tumor cells can be infused with a HSC graft, and that these cells can result in relapse of the malignancy(21). This led to trials addressing the question of whether purging stem cell products of NBL cells could further improve post-transplant overall and disease-free survival. The most widely used technique in neuroblastoma has been anti-tumor monoclonal antibodies followed by a magnetic depletion step(22, 23), although CD34 selection has also been tested(24). Although the evidence suggests that purging of bone marrow may be important, PBSC are less likely to contain tumor cells than bone marrow, and no study to date has shown that purging itself improves outcome. This was borne out in COG A3973, which was a phase III, randomized comparison of purged versus unpurged PBSC given in the context of autologous SCT for high-risk neuroblastoma. Data from this trial presented at ASCO IN 2008(25) showed no advantage for patients receiving a purged PBSC. The 2-year EFS was 51% in the unpurged group, and 47% in the purged group (P=0.47). The overall estimated 3-year EFS was 40%. The lesson from these data may be that the key to success is purging tumor in the patient, rather than the PBSC product.

Impact of Biomarkers on Neuroblastoma Therapy

The International Neuroblastoma Risk Group (INRG) staging system’s risk stratification criteria for neuroblastoma are based on clinical and biologic factors, including stage, age at diagnosis, MYCN gene amplification status, DNA index, copy number aberrations at chromosome 11q, and histology(26). This approach has been most successful in identifying those with low and intermediate risk who have greater than 90% overall survival after receiving only moderate intensity chemotherapy(27, 28). Although treatment for children identified with high-risk neuroblastoma has improved over the past 20 years, only 45% of these patients become long-term, disease-free survivors (4, 14, 19, 29–33). The majority of failures occur within two years of diagnosis, and survival remains poor among patients with disease progression(32, 34, 35). Improvement in survival for this high-risk group necessitates development of effective new therapies that target biologically risk-stratified subgroups of patients and development of effective biomarker assays that provide prognostic information at the time of diagnosis and during the course of therapy. This approach has been successfully implemented in patients with acute lymphocytic leukemia, where risk based assignment using clinical and genetic markers at diagnosis are further refinement based on minimal residual disease (MRD) evaluation of blood and bone marrow early in the course of therapy.

Neuroblastoma Prognostic Markers at Diagnosis

There already is precedent for defining risk for all stages of neuroblastoma (local, regional, metastatic) based upon molecular biologic aberrations in primary tumors. The use of MYCN gene amplification to predict outcome is a paradigm for this(36). Chromosomal losses such as 1p and 11q also provide additional prognostic information(37–39), and additional studies have demonstrated importance of chromosome 17(40, 41). Several genomic and gene expression profiling studies of neuroblastoma have also reported associations with patient outcomes(39, 42–46). These studies provide insight into the clinical heterogeneity of neuroblastoma and have been most successful in predicting poor outcome for patients classified as low or intermediate risk as defined by INRG classification. A study focused primarily on metastatic tumors lacking MYCN gene amplification was the first to demonstrate a prognostic signature for patients with the same stage of disease (stage 4) based on RNA profiling(47). A 55-gene signature predicted the likelihood of progression-free survival (PFS) with < 20% error rate for patients over 12 or 18 months of age at diagnosis. Surprisingly in that study, the microarray signature showed that high expression of the immunoglobulin kappa gene was predictive of poor outcome. Gene Ontology (GO) and gene set enrichment analysis also identified antigen-binding genes and CCR5 and CXCR4 macrophage signaling pathway as being associated with the microarray-based high-risk group. Inclusion of inflammation-related markers into a clinically-applicable 14-gene assay using TaqMan Low Density Array (TLDA) has demonstrated the clinical significance of expression of inflammation-related genes in this high-risk group of patients. (unpublished data) This clinical assay has now been validated in two independent cohort of samples providing a high and ultra-high risk subgroup of patients with metastatic MYCN non-amplified tumors. These data demonstrate that expression profiling of primary tumors at diagnosis can provide clinically relevant prognostic information for high-risk neuroblastoma patients.

Neuroblastoma Prognostic Markers during Therapy

Advances in molecular technologies have also fueled research into discovery of gene-based assays for detections of rare tumor cells that serve as surrogate marker for tumor burden during therapy. Evaluations are performed using specimens from bone marrow, the most common site of disease recurrence(14, 32). The low sensitivity of morphologic assessment initially prompted development of immunocytology, which uses four different monoclonal antibodies to identify tumor cells. Bone marrow and blood were found to be positive for tumor cells in 81% and 58% of stage 4 patients respectively at diagnosis by immunocytology, and quantifying tumor cells throughout induction therapy provided independent prognostic information(48). Subsequently, RT-PCR assays have been reported to detect infrequent tumor cells and predict outcome(49–53). RNA expression of neuroblastoma specific genes including TH, GD2, and PHOX2b have been frequently used in quantifying neuroblastoma cells in the bone marrow. The selection of genes and design of primers for such markers requires careful evaluation to minimize false positives in normal bone marrow samples. The heterogeneity in biology of neuroblastoma is also evident in these studies, as multi-gene detection assays have been shown to have superior sensitivity and specificity for detection of MRD than any one single gene. In our own ongoing research, we have used data from microarray analyses of over 150 high-risk neuroblastomas, normal neuronal tissues, and bone marrow samples to develop a 5-gene RT-PCR detection assay that includes TH, PHOX2b, CHGA, DCX, DDC with high sensitivity and specificity. In general, the limit of sensitivity of MRD tests is as follows: morphology, one tumor cell among 100 normal cells (1%); immunocytology, one tumor cell among 100,000–200,000 normal cells; RT-PCR, one neuroblastoma cell among 500,000–1,000,000 bone marrow cells. Using RT-PCR MRD assessment, 10% of patients with metastatic disease have no detectable tumor cells in bone marrow at diagnosis as compared to 20% using immunocytology (48, 54). Seeger et al. reported 29% of patients had identifiable tumor cells at the end of 12 weeks of therapy using immunoctology(48), while Stutterheim et al. showed that 71% have detectable disease using RT-PCR 2–4 months after start of therapy. The latter group have also shown that 11 out of 38 (29%) of patients had early clearance of tumor cells at 3 months after diagnosis and was associated with favorable outcome (5-y-OS 62 ± 15.0% versus 19 ± 8%; p = 0.009). Analysis of peripheral blood stem cells from 238 patients, which were obtained 8–12 weeks after diagnosis, demonstrated that 49.6% had detectible tumor cells using the 5-gene RT-PCR MRD signature. (unpublished data) Detection of tumor cells was associated with a significantly worse EFS (35.0 ± 4.6% vs. 50.4 ± 4.6%; P=0.014) at 3 years after diagnosis than non-detection. Analysis of bone marrow obtained at the end of induction was reported to be MRD positive in 12 of the 29 samples (41%), while only 5 of these samples were also positive by immunocytology(54). All patients with MRD positive BM died of progressive disease. Further studies should provide additional clinical information about the robustness and validity of these assays during induction period.

Next generation sequencing (NGS) technologies have introduced new possibilities for monitoring tumor response to therapy. Tumor specific translocation sites can now be identified using NGS technologies which allow design of patient-specific PCR primers for quantification of circulating tumor DNA. However, it is unclear if these translocation events are preserved during the clonal evolution of tumors after multiple rounds of chemotherapy. Finally, ongoing analyses combining strategies that monitor response to therapy such as RT-PCR MRD or MIBG imaging strategies(55, 56) should allow identification of subgroups of rapid-responders among patients who are classified as high-risk or ultra-high risk using RNA signatures of pre-therapy samples. These strategies would facilitate appropriate selection of patients for AHSCT.

Radiolabeled metaiodobenzylguanidine (mIBG) and autotransplant

Neuroblastoma is a malignancy of sympathetic nervous system origin, one that has the ability to concentrate, store and secrete catecholamine metabolites. In more than 90% of cases, increased levels of urinary catecholamines, including dopamine, homovanillic acid (HVA) and/or vanillymandelic acid (VMA) are present at diagnosis.

Metaiodobenzylguanidine (mIBG) is a norepinephrine analogue initially developed at the University of Michigan in the 1970’s for adrenal imaging (57) (Figures 1 and 2). Similar to norepinephrine, mIBG has the ability to concentrate in neural crest tissues and neuroblastoma via cell surface norepinephrine transport (NET) channels. The ability to radiolabel mIBG with various iodine isotopes, including I-131 or I-123, led to its subsequent use in neuroblastoma imaging in the early 1980’s (58–62). Both 131I-mIBG and 123I-mIBG are currently licensed for scintigraphic imaging of neuroblastoma, with 131I-mIBG currently in use as a therapeutic modality. The potential impact of 131I-mIBG in neuroblastoma therapy is enhanced by its ability to concentrate within tumors with diverse presentations and biology. MIBG uptake is well described in both primary tumor sites and metastases, including soft tissue, bone marrow and bony lesions. MIBG concentrates in tumors with favorable or unfavorable histologic patterns, amplified or non-amplified MYCN oncogene expression, and low stage as well as advanced stage disease (60, 62–64). These features make 131I-mIBG an attractive therapy option for patients with high risk neuroblastoma.

Figure 1.

Chemical structure of ¹³¹I-metaiodobenzylguanidine and norepinephrine.

Figure 2.

¹²³I-mIBG scintigraphy. Extensive mIBG avid disease in cranium, orbits, sino-facial region, bilateral humeri, and sternum.

131I-mIBG therapy: General overview

Over the past two decades, 131I-mIBG therapy has focused on palliation for patients with refractory disease, specifically targeting those patients who fail induction or develop progressive disease post-transplant. The benefits of 131I-mIBG in this disease setting have now been established, and current protocols are investigating its role with adjuvant chemotherapy, biologic response modifiers or in combination with transplant conditioning regimen. A dose dependent response has been noted, with dose escalation from 8 to 21 mCi/kg associated with increasing levels of tumor response (65). When given as single agent therapy, response rates have ranged from 10–55% for patients with end-stage or refractory disease (66). In a large case series of 164 patients, responses were seen in 45% of patients with disease isolated to the bone and/or bone marrow, and 50% for patients with isolated soft tissue involvement (65). Improved response rates (RR) were also seen in patients > 12 years in age (55% RR) and those having received < 3 prior regimen (53% RR). Reported toxicities have been mild, including infusion related nausea and vomiting, myelosuppression, and the potential for hypothyroidism (65). The primary toxicity remains hematologic, with stem cell rescue required in patients receiving > 15 mCi/kg of 131I-mIBG (66). For current North American trials, 131I-mIBG therapy still requires an adequate number of stem cells to be collected and cryopreserved in advance of therapy. Late complications of 131I-mIBG therapy have been rare, with hypothyroidism (10%), adrenal insufficiency (<5%), and secondary malignancies (<5%) all reported (66).

131I-mIBG therapy: Role in the autologous transplant setting

As noted above, the ability to combine 131I-mIBG with chemotherapy, radiosensitizers, or biologic response modifiers has grown dramatically in the past decade. A pilot study in the 1990’s at the University of Michigan, incorporating 131I-mIBG directly into the transplant conditioning regimen for 12 patients who had failed induction therapy, demonstrated feasibility and efficacy (67). All 12 patients received 12 mCi/kg of 131I-mIBG given on day -21 pre-transplant, followed by carboplatin, etoposide, and melphalan (CEM) administered days -7 to -4 pre-transplant. Responses were seen in 5 of the 8 patients with metastatic disease and 3 of 4 patients with localized disease. A subsequent phase I trial performed by the New Approaches to Neuroblastoma Therapy (NANT consortium) identified 12 mCi/kg as the MTD dose for 131I-mIBG, when given in combination with a CEM conditioning regimen for patients with refractory disease (35). Six of 22 patients (27%) had a complete or partial response to therapy, with estimated 3-year event free survival 0.31 + 0.10 and overall survival 0.58 + 0.10 (Figure 3). The use of radiolabeled mIBG in the conditioning regimen did not appear to affect hematologic recovery post-transplant. Engraftment was brisk in both studies, with a median time to neutrophil recovery (> 500/uL) 10 days and platelet recovery 26–28 days for the two studies. Primary toxicities of the regimen included grade 3–4 oral mucositis, febrile neutropenia and hepatic toxicity, with hepatic veno-occlusive disease (VOD) seen in 6 of the 22 patients treated in the NANT study. The high incidence of VOD in this study stands in contrast to single agent 131I-mIBG therapy studies, in which hepatic toxicity is uncommon (< 5%), and in contrast to the Michigan study, in which no VOD was noted in the 12 patients treated. Given the above phase I results, a number of phase I/II studies have now been completed, both within North America and Europe, the majority focusing on patients with mIBG avid, non-responsive or progressive disease (68, 69). The combination of irinotecan, vincristine plus 131I-mIBG appears to be an attractive option to the 131I-mIBG-CEM transplant regimen, with a phase I study recently completed by the NANT consortium (70). Though myelo-ablative at the defined maximal tolerated dose (MTD) of 18 mCi/kg 131I-mIBG, the regimen is associated with significantly less epithelial injury than 131I-mIBG-CEM, with minimal oral mucosal and hepatic toxicity reported. Based upon the safety profile and potential efficacy of this regimen, the combination of irinotecan, vincristine, and 131I-mIBG with stem cell rescue is being considered for a randomized, phase III trial within the Children’s Oncology Group (COG) for the treatment of patients with high risk neuroblastoma in first response.

Figure 3.

Phase I study of 131I-mIBG with CEM chemotherapy and autologous stem cell rescue for patients with refractory neuroblastoma. Event free and overall survival. Matthay KK et al. JCO 2006;24: 500–506.

Prognostic impact of pre-transplant mIBG scans

Over the past 15 years, a semi-quantitative mIBG scoring system (Curie scoring system) has been developed to estimate the extent and severity of mIBG avid disease (71–73). Such assessments, when performed during or following induction therapy, have been predictive of subsequent tumor response and progression free survival (72–76). The Curie scoring system is based upon the extent of mIBG uptake in multiple anatomical regions, with numeric scores assigned to the degree of involvement in each region. The scores from each individual region are subsequently summated to determine a composite score (71, 72). Both composite and “relative” scores (obtained by dividing the absolute score at each time point by the corresponding pre-treatment score) have been extremely reliable in predicting response (72, 76). In particular, the presence of a > 75% decline in Curie score following 4 cycles of induction therapy has been associated with improved EFS and OS. The Children’s Oncology Group is currently conducting a mIBG scan review of nearly 300 patients treated on a homogenous high risk neuroblastoma protocol, COG A3973 (77). Curie scores are being determined from mIBG scans obtained at the time of diagnosis, post-induction therapy (pre-transplant), post-transplant and post-biotherapy. Initial analysis indicates that pre-transplant Curie scores will have a significant impact in determining subsequent post-transplant EFS (77). Ultimately, the Curie scoring method may prove to be a valuable tool in the decision-making process for transplant, both in terms of determining the optimal timing of transplant and intensity of the transplant regimen. Patients with high Curie scores pre-transplant could even be considered for alternative transplant regimen, including options that incorporate mIBG therapy directly into the transplant process.