Long-Term Outcome and Role of Biology within Risk-Adapted Treatment Strategies: The Austrian Neuroblastoma Trial A-NB94

Simple Summary Neuroblastoma, the most common extracranial malignancy of childhood, shows a highly variable course of disease ranging from spontaneous regression or maturation into a benign tumor to an aggressive and intractable cancer in up to 60% of patients. To adapt treatment intensity, risk staging at diagnosis is of utmost importance. The A-NB94 trial was the first in Austria to stratify therapy intensity according to tumor staging, patient’s age, and MYCN amplification status, the latter being a biologic marker turning otherwise low-risk tumors into high-risk disease. Recent publications showed a prognostic impact of various genomic features including segmental chromosomal aberrations (SCAs). We retrospectively investigated the relevance of SCAs within this risk-adapted treatment strategy. The A-NB94 approach resulted in an excellent long-term survival for the majority of patients with acceptable long-term morbidity. An age- and stage-dependent frequency of SCAs was confirmed and SCAs should always be considered in future treatment decision making processes. Abstract We evaluated long-term outcome and genomic profiles in the Austrian Neuroblastoma Trial A-NB94 which applied a risk-adapted strategy of treatment (RAST) using stage, age and MYCN amplification (MNA) status for stratification. RAST ranged from surgery only to intensity-adjusted chemotherapy, single or multiple courses of high-dose chemotherapy (HDT) followed by autologous stem cell rescue depending on response to induction chemotherapy, and irradiation to the primary tumor site. Segmental chromosomal alterations (SCAs) were investigated retrospectively using multi- and pan-genomic techniques. The A-NB94 trial enrolled 163 patients. Patients with localized disease had an excellent ten-year (10y) event free survival (EFS) and overall survival (OS) of 99 ± 1% and 93 ± 2% whilst it was 80 ± 13% and 90 ± 9% for infants with stage 4S and for infants with stage 4 non-MNA disease both 83 ± 15%. Stage 4 patients either >12 months or ≤12 months but with MNA had a 10y-EFS and OS of 45 ± 8% and 47 ± 8%, respectively. SCAs were present in increasing frequencies according to stage and age: in 29% of localized tumors but in 92% of stage 4 tumors (p < 0.001), and in 39% of patients ≤ 12 months but in 63% of patients > 12 months (p < 0.001). RAST successfully reduced chemotherapy exposure in low- and intermediate-risk patients with excellent long-term results while the outcome of high-risk disease met contemporary trials.


Introduction
Neuroblastoma is the most common extracranial malignancy of childhood and originates from the sympathetic nervous system. It shows a highly heterogeneous behavior ranging from spontaneous regression or maturation into a benign ganglioneuroma to an aggressive and intractable disease. Risk classification systems are using clinical and biological characteristics to predict survival and adapt treatment intensity [1,2].
At this study's initiation, recognized risk driving factors included stage defined by the International Neuroblastoma Staging System (INSS), age at diagnosis, and MYCN oncogene amplification (MNA) status [3]. MNA, a strong biologic marker associated with rapid tumor growth [4,5], transforms otherwise favorable risk profiles of infants [6,7] and children with localized resectable [8,9] or unresectable [10] disease into high-risk [11]. Metastatic disease in children older than 18 months constitutes per se an unfavorable risk group regardless of MYCN status [12]. Intratumoral heterogeneous MNA (hetMNA) refers to the coexistence of clustered or scattered single MNA cells and non-MYCN-amplified (non-MNA) tumor cells [13], a phenomenon that was largely unexplored at the initiation of A-NB94. A recent study highlights the importance of viewing it separately from the MNA profile and its unfavorable risk implication, however, prognostication and therapy allocation are still unsolved issues [14,15].
Here, we present long-term outcomes of the Austrian neuroblastoma trial A-NB94, initiated in 1994 to apply a risk-adapted strategy of treatment (RAST) based on age (≤/>12 months), INSS stage and MYCN Status (Table 1). Encouraging results of the Lyon-Marseille-Curie-Est cooperative group (LMCE2) [16] using tandem high-dose chemotherapy (HDT) followed by autologous stem cell rescue (ASCR) prompted the adoption of a similar approach for patients with incomplete response to induction therapy. In addition, we show a post hoc genomic analysis to investigate pattern and potential influence of biomarkers on long-term outcomes. Table 1. Overview of the A-NB94 risk-adapted strategy of treatment.

Influence of Stage
Patients presenting with localized (stage 1-3) neuroblastoma (n = 109) had a 10y-EFS and OS of 93 ± 2% and 99 ± 1% ( Figure 1A), respectively. Six relapses were reported and all affected non-MNA patients ≤ 12 months of age (n = 39). Four of these patients had loco-regional relapses: two were salvaged by six cycles cyclophosphamide/vincristine (CV), one by second surgery, and one was only observed as the parents declined further chemotherapy and the tumor ultimately regressed. Two infants had a relapse in infant age developing liver metastases, very much in line with a stage 4S pattern. They revealed no adverse genomic features at relapse, were closely observed, and ultimately showed spontaneous regression without further therapy. GNB was only found in localized non-MNA tumors and included the nodular (n = 4) or intermixed (n = 20) subtype. All patients with localized disease became long-term survivors, apart from one GNB patient with underlying neurofibromatosis type 1 dying later outside and unrelated to the A-NB94 trial, resulting in a 10y-EFS of 93 ± 3% for localized neuroblastoma versus 96 ± 4% for GNB (p = 0.584) with a 10y-OS of 100% versus 96 ± 4% (p = 0.062), respectively.  Infants with stage 4S neuroblastoma (n = 10) had a 10y-EFS and OS of 80 ± 13% and 90 ± 9% ( Figure 1A), respectively. The tumors of four patients regressed without intervention while four patients underwent surgery of the primary tumor. Chemotherapy was given to two infants with clinical symptoms: one received electively four cycles of vincristine monotherapy and surgery whereas the other one, suffering from congenital Infants with stage 4S neuroblastoma (n = 10) had a 10y-EFS and OS of 80 ± 13% and 90 ± 9% ( Figure 1A), respectively. The tumors of four patients regressed without intervention while four patients underwent surgery of the primary tumor. Chemotherapy was given to two infants with clinical symptoms: one received electively four cycles of vincristine monotherapy and surgery whereas the other one, suffering from congenital neuroblastoma, showed uncontrollable disease progression, and died of respiratory failure despite chemotherapy escalation.
Anti-GD2 monoclonal antibody ch14.18/SP0/2 became available in 1996 for compassionate use in 12 stage 4 patients. The landmark time identified the median time between HDT and initiation of immunotherapy as 87 days. When comparing the anti-GD2 monoclonal antibody pilot population (n = 12) to the pre-immunotherapy population accrued in the A-NB94 trial and considering only stage 4 patients without progressive disease at the landmark time point of 87 days after HDT, no difference in outcome was observed shown by a 10y-EFS of 67 ± 14% versus 64 ± 13% (p = 0.77) and a 10y-OS of 67 ± 14% versus 71 ± 12% (p = 0.907).

Occurrence and Influence of Biomarkers
MYCN status was evaluated for all patients prospectively during the risk stratification process. MNA was observed in 6% (7/109) of localized, in 43% (20/44) of stage 4, and in none of the stage 4S tumors (p < 0.001). We found four infants harboring hetMNA; one of them was only recognized in a later tumor sample and this patient was treated according to the original result as per MNA protocol. The other three patients were treated in the non-MNA arm.
1p loss was recorded prospectively during the active trial period and data were available for 122 patients. Quantity and quality of frozen tumor samples allowed for post hoc genomic analysis of 108 patients (multiplex ligation-dependent probe amplification (MLPA), n = 68; single nucleotide polymorphism (SNP) array, n = 32; interphase fluorescent in-situ hybridization (iFISH), n = 8) including SCAs for 1q gain/1qloss , 2p gain , 3p loss , 4p loss , 5p loss , 6q loss , 9p loss , 11q loss , 14q loss , 17p loss , 17q gain , 19q loss , and 22q loss . Only in two cases of GNB, genomic analysis was interpretable and revealed no SCAs (data included in mentioned numbers); in the other Schwann cell stroma-rich GNB tumors, the neoplastic clone was masked by the normal Schwann cells [17].
SCAs were observed in 92% of stage 4 but only in 29% of localized tumors (p < 0.001). Patients > 12 months showed tumor SCAs in 63% while SCAs were found in only 39% of patients ≤ 12 months (p < 0.001) ( Figure 2). of GNB, genomic analysis was interpretable and revealed no SCAs (data included in mentioned numbers); in the other Schwann cell stroma-rich GNB tumors, the neoplastic clone was masked by the normal Schwann cells [17].
SCAs were observed in 92% of stage 4 but only in 29% of localized tumors (p < 0.001). Patients > 12 months showed tumor SCAs in 63% while SCAs were found in only 39% of patients ≤ 12 months (p < 0.001) (Figure 2). This analysis found patients with tumors showing one or more SCAs (n = 56) with a 10y-EFS and OS of 64 ± 6% and 69 ± 6% while it was 87 ± 5% (p = 0.014) and 96 ± 3% (p < 0.001) for patients without SCAs (n = 52) ( Figure 1C). In univariate analysis, presence of 1p loss , 1q gain /1q loss , 3p loss , 11q loss , or 17q gain had significant predictive power for worse 10y-EFS and/or OS (Table 3). To investigate the potential added impact of segmental chromosomal alterations (SCAs), we performed a multivariate analysis (MVA) corrected by the risk-stratifying factors of age (≤/>12 months), INSS stage and MYCN amplification This analysis found patients with tumors showing one or more SCAs (n = 56) with a 10y-EFS and OS of 64 ± 6% and 69 ± 6% while it was 87 ± 5% (p = 0.014) and 96 ± 3% (p < 0.001) for patients without SCAs (n = 52) ( Figure 1C). In univariate analysis, presence of 1p loss , 1q gain /1q loss , 3p loss , 11q loss , or 17q gain had significant predictive power for worse 10y-EFS and/or OS (Table 3). To investigate the potential added impact of segmental chromosomal alterations (SCAs), we performed a multivariate analysis (MVA) corrected by the risk-stratifying factors of age (≤/>12 months), INSS stage and MYCN amplification status. Neither model A nor model B (Table 4) was able to identify an added risk for SCAs in the investigated trial population. ATRX del (n = 3) and TERT gain (n = 4) were found only in stage 4 patients > 18 months of age and were mutually exclusive. All but one patient with TERT gain died with progressing disease.

Role of Treatment Elements to Achieve Remission Induction
Low-to intermediate-dose chemotherapy during first-line treatment was given to 18% (20/109) of patients with localized disease, 20% (2/10) of patients with stage 4S, and 100% (6/6) of patients ≤ 12 months with stage 4 non-MNA tumors. The overall response rate to cytotoxic treatment was 93% (26/28) in this cohort with 50% (14/28) entering a complete clinical remission (CR). One patient with stage 4 non-MNA and one with stage 4S disease did not respond, and both died of disease progression despite chemotherapy escalation.
Stage 4 patients > 12 months (n = 32) and ≤12 months with MNA tumors (n = 6) received dose-intensive induction therapy with a metastatic response rate of 74% (28/38) and a CR rate of 16% (6/38) including the effects of surgery. In this group, three patients progressed during induction. 2/38 patients (5%) experienced an infection-related septic shock during induction therapy and died with multi-organ failure.

Surgery
Surgical resection of the primary tumor was performed in 93% (152/163) of patients. In case of adrenal primary, 5% (6/113) underwent unilateral total nephrectomy while other patients only had unilateral adrenalectomy. Two patients needed revision surgery due to rebleeding.

High-Dose Therapy (HDT)
Patients eligible for HDT (Table 1) received one (n = 18), or, in case of mIBG metastatic incomplete response, two (n = 11) or three (n = 2) courses of HDT. In addition, two patients with stage 3 MNA but post-surgical macroscopic tumor residues, and one patient 11 months of age with stage 4 non-MNA but post-induction unresectable disease received a single course of HDT. Seven patients with single HDT were in metastatic CR (mCR) prior HDT; 93% (13/14) of other patients responded to HDT with a mCR rate of 50% (7/14). Within the multiple HDT group, response rate was 77% (10/13), CR rate 54% (7/13).
In the first attempt of stem cell apheresis in 27 evaluable patients, median yield was 4.5 × 10 6 CD34 positive cells per kilogram body weight (range 0.45 to 30.9 × 10 6 ). Six patients needed a second (median yield 3.76 × 10 6 , range 1.26 to 8.5 × 10 6 ), and two patients a third apheresis (median 4.99 × 10 6 , range 4.97 to 5 × 10 6 ) in order to reach the attempted minimum of 3 × 10 6 cells per kilogram body weight for each of the planned HDT courses according to their respective response status.

Acute Toxicities and Long-Term Morbidity
Acute toxicities and intervention related morbidity was related to respective treatments (Table A2) and clearly showed a higher acute toxicity burden with increased treatment intensity.
Of surviving patients receiving low-and intermediate-dose chemotherapy (localized, stage 4S, or stage 4 patients ≤12 months with non-MNA disease) but no HDT (n = 24), only two patients had treatment-associated long-term disabilities manifesting as reduced renal glomerular filtration rate (GFR). Treatment-unrelated morbidity was persistent paraplegia of the lower limbs already present at diagnosis (n = 1) and subsequent craniopharyngioma (n = 1).

Discussion
The A-NB94 trial was the first in Austria to apply a risk-adapted treatment strategy for children with neuroblastoma considering INSS stage, age, and MYCN amplification status [3]. Risk stratification was implemented in 1994 and results need to be viewed from this perspective. Only 15 years later, the INRG established staging based on pre-surgical image-defined factors [18], and included additional biologic makers of histopathology, ploidy, and 11q status [2]. We retrospectively stratified A-NB94 patients according to the INRG classification system which resulted in an upstaging of patients with localized MNA tumors to high-risk while other patients remained within a risk group similar to the A- Compared with published data, the Austrian screening program detected a comparatively high number of localized tumors harboring unfavorable biologic features (MNA, 1p loss , diploidy) [19], suggesting early adverse clonal evolution. However, the A-NB94 trial was not designed for answering the question if screening helped increasing the detection of high-risk disease at an early stage [20,21]. Overall, our approach resulted in excellent survival for these patients as all became long-term survivors apart from one GNB patient dying from complications after surgery not related to neuroblastoma treatment.
Notably, the very favorable outcome of the localized subgroup also includes 25 stage 3 patients, marking a major improvement compared to the preceding A-NB87 trial which applied an intensive chemotherapy for children with Evan's Stage 3 disease (especially when >2 years of age, increased neuron-specific enolase, and/or ferritin (Table A3), resulting in a 28% (8/29) toxic death rate and a 5y-OS of only 50% [22]. However, detailed comparison is hampered by differing staging criteria and MYCN not being a stratifying marker in A-NB87. In A-NB94, 82% (89/109) of patients with localized neuroblastoma received little or no chemotherapy in first line treatment. These numbers include six patients with non-MNA stage 3 tumors defined by tumors crossing the midline which could otherwise be surgically removed upfront. While two relapses were treated successfully with chemotherapy, potential long-term side effects following intense therapy could be circumvented in other patients by using a surgery only approach.
While our data support the notion of improved outcome despite therapy reduction in patients with intermediate-risk neuroblastoma [23,24], contemporary trials would submit patients with stage 2 or 3 MNA [25] or diploid [26] tumors to receive HDT/ASCR. Furthermore, the presence of certain SCAs was reported to identify patients of higher risk for relapse, especially in unresectable tumors [27]. While SCAs, especially 1p loss and 11q loss , may lower EFS but not OS in children ≤18 months, they also diminish OS in older children [28]. In A-NB94, 15 patients with localized tumors matched these criteria (MNA, n = 7; diploidy, n = 5; >18 months and 1p loss , n = 1; or 11q loss , n = 2) and all survived without relapse after receiving low-/intermediate-dose chemotherapy apart from two already described patients that received HDT/ASCR. The decision for adding HDT in the latter two patients was based on the presence of post-surgical residual tumors. These small numbers might suggest that patients with residual tumors benefited from therapy escalation while others were adequately treated by conventional chemotherapy and irradiation. In this context, it may be assumed that the remarkable good outcome in A-NB94 of patients with localized tumors including especially stage 3 patients relates to the surprisingly low prevalence of SCAs observed in this subgroup.
Outcome of infantile stage 4 metastatic disease is related to MNA, diploidy/tetraploidy, 1p loss , 11q loss , or 17q gain [29]. While most stage 4S tumors are treated adequately by observation only [30,31], certain markers including 11q loss and diploidy might predict less favorable outcomes even in stage 4S disease [32]. However, disease progression of two A-NB94 stage 4S patients clearly could not be explained by these biologic factors, as neither MNA, diploidy or 11q loss were detected. Stage 4S poses threats independent of tumor biology as extensive hepatomegaly causes a variety of clinical complications.
Of 44 stage 4 patients, six infants without MNA received moderate chemotherapy without HDT or radiation according to RAST. While survival was good, one of three patients with tumors showing SCAs involving 11q loss and 17q gain progressed and died during induction, in line with reports that these features are often associated with higher risk for relapse [33].
Stage 4 patients > 12 months and stage 4 infants with MNA [34,35] underwent an induction protocol similar to the "N6" protocol previously published to achieve fast cytoreduction [36], following the notion that increased response rates precede higher survival rates [37]. Furthermore, A-NB94 successfully introduced tandem or triple HDT/ASCR to children with incomplete metastatic response to induction therapy. While benefits of multiple cycle HDT have been reported consistently by European groups [16,38], recent reports demonstrated clearly improved survival in contemporary trials [39,40].
In 1996, anti-GD2 monoclonal antibody ch14.18/SP0/2 became available for compassionate use in stage 4 patients. The identified landmark time identified between HDT and start of immunotherapy was 87 days; thus, only stage 4 patients without progressive disease were included in the pre-immunotherapy control population. Comparing these two cohorts did not result in an advantage for the small immunotherapy population. However, overall outcome data are quite in line with later publications on the use of ch14.18 monoclonal antibody in first-line maintenance treatment [41,42]. Considering the rather small population in both cohorts, we only may hypothesize that RAST, using repetitive HDT adapted to response in this high-risk population, was probably a major contributor to the observed favorable outcomes and that immunotherapy applied with a less dose intensive monotherapy schedule did not result in a measurable added benefit.
In addition, while the prognostic impact of hetMNA remains unclear [43,44], recent data suggests it has to be viewed in light of the overall genetic background in order to determine the importance of the MNA cell clone [45]. Localized tumors usually show a favorable background with reported improved outcomes compared to homogeneous MNA. In A-NB94, hetMNA tumors showed only additional 1p loss in two cases but no other SCA or other unfavorable biologic markers. One of these cases was only discovered in a later tumor sample as the original sample suggested homogeneous MNA; this patient with stage 4 disease received high-dose therapy, which may, together with the age of ≤12 months and lack of additional high-risk features, have resulted in over-treatment.
SCAs were present in higher frequency in stage 4 disease, especially in patients > 12 months, when compared to children with localized tumors of any age (Figure 2), supporting the notion of SCAs being accumulated during tumor development [44] and an unfavorable biology being associated with overall poor prognosis [45]. This study confirms SCAs are commonly involving chromosome arms 1p, 11q, and 17q [46], however, high-risk metastasized tumors often show a combination of various chromosomal defects [47].
The acute toxicity burden as well as long-term morbidity is clearly related to RAST, showing the therapy burden of the curative efforts mainly in the high-intensity treatment groups. Particularly after additional HDT/ASCR, which is in line with previous reports [48,49], although and in contrast to previous approaches [50], no total body irradiation (TBI) was used. Observed surgery-related morbidities were in line with previous reports [51]. Irreversible inner ear hearing loss, a known side-effect following cisplatin therapy [52], affected 29% (10/34) of long-term survivors receiving cisplatin. The un-derstanding of the mechanism of uptake and accumulation in the stria vascularis of the cochlea provides now an important target for preventing ototoxicity in future trials [53]. Overall, organ-related long-term side-affects were low as well as incidence of secondary malignancies.

Patients
Children with histologically confirmed, previously untreated neuroblastoma or GNB up to 20 years of age were eligible for enrolment on A-NB94. Staging followed INSS guidelines and treatment response was assessed by the 1993 International Neuroblastoma Response Criteria (INRC) [3]. Written informed consent for treatment and data procession was obtained in compliance with institutional review board rules and in accordance with the Declaration of Helsinki [54].

Treatment Concepts
RAST was adjusted for age (≤/>12 months), stage, and MYCN status (Table 1). Stage 1 or 2 non-MNA tumors were planned for surgery only; incompletely resected MNA tumors received chemotherapy adapted to stage and radiotherapy. Stage 4S infants were observed up to 3 months and then underwent surgery but received chemotherapy in case of life-threatening symptoms or progression. Stage 3 and 4 were treated with neoadjuvant intensity-adjusted chemotherapy with surgery planned after four cycles. MNA tumors were irradiated age-adapted with 24 Gray (≤12 months) or 30 Gray (>12 months).
Stem cell harvest, aiming at >3 × 10 6 CD34 positive cells per kilogram body weight per high-dose treatment (HDT), was first attempted after 4 chemotherapy cycles if cytomorphological and histological bone marrow (BM) remission was achieved, or was otherwise postponed following cycle 6 or after an in vivo purge following the first or second HDT, eventually.
Depending on metastatic response, patients were eligible for a single (complete response, CR) or repetitive HDT (partial/minor response, PR/MR). In case of CR after second HDT, the third HDT was omitted. In 1996, the anti-GD2 monoclonal antibody ch14.18/SP0/2 became accessible as compassionate use (provided by the University of Tübingen, Prof. Dr. Rupert Handgretinger) to optimize treatment of HR disease after HDT and radiotherapy. Treatment consisted of three anti-GD 2 ch14.18/SP0/2 cycles (20 mg/m 2 over 5 days as 8-h infusions). Supportive care and treatment of infections followed institutional guidelines.

Disease and Response Assessment
Disease evaluation was planned at diagnosis, after 2, 4, and 6 courses of chemotherapy, before and after each HDT, and before and after immunotherapy. Bone marrow examination included aspirates/trephines obtained from two sites and used, apart from cytomorphology, an automated image analysis system (MetaSystems GmbH, Altlußheim, Germany) to quantify GD 2 /CD56-positive cells [55,56]. Skeletal disease was defined by pathologic iodine-123 meta-iodobenzylguanidine (mIBG) or, in mIBG non-avid tumors, by tecnetium-99m-hydroxymethylenediphosphanate (Tc99m) scans [57]. Primary tumors were investigated by MRI/CT scan. Tumor marker evaluation included urinary catecholamine metabolites, lactate dehydrogenase (LDH), and neuron specific enolase (NSE). Response was assessed based on local institution reporting following INRC guidelines [3].

Toxicity Evaluation
Acute and long-term toxicities were evaluated using case report forms (CRF) documenting organ-specific toxicities according to the Common Terminology Criteria for Adverse Events (CTCAE) Version 5.0 above grade 2, secondary malignancies, and other adverse events (AE). Acute toxicity evaluation focused on patients with intense induction chemotherapy and HDT.

Biologic Studies
Frozen tumor specimens were evaluated using high-density SNP arrays (CytoScan HD Array) [47], MLPA, or iFISH for the detection of MNA or SCAs. Recording/interpretation of data were done according to international standards [58].

Statistical Analysis
Event-free survival (EFS) was calculated from the time of diagnosis until first occurrence of relapse, progressive disease, secondary malignancy, or death from any cause, or until last contact with patients. Overall survival (OS) was calculated from time of diagnosis to death from any cause. The median time between HDT and initiation of immunotherapy was 87 days, therefore only patients without progressive disease at this landmark time point were included in the pre-immunotherapy cohort. Categorical data were analyzed by chi-square test or, in case of an estimated case-count of ≤5 per field, Fisher's exact test. Data were analyzed with SAS 9.4 program. All statistical tests were two-sided.

Conclusions
The risk-adapted approach resulted in an excellent long-term survival for the majority of patients with acceptable long-term morbidity. An age-and stage-dependent frequency of SCAs was confirmed and should be considered in future treatment decision-making processes.  Data Availability Statement: Biomarker data as presented in this cohort study are available in anonymized form on request at the CCRI by addressing the corresponding author.

Conflicts of Interest:
The following authors declare conflict of interest: Georg Ebetsberger-Dachs-Consulting or advisory role: Novartis International AG, Amgen Inc.; Travel, accommodations, expenses: Shire PLC., Servier Laboratories. The other authors declare no conflict of interest.   Summary of acute treatment-related toxicities according to induction treatment split by chemotherapy intensity with low-and intermediate-dose (LD/ID) and high-dose (HD) induction schedules, the latter including stage 4 > 12 months and stage 4 ≤ 12 months but MYCN amplified tumors, the former all other treatment groups; surgical complications; and high-dose therapy (HDT). Chemotherapy elements and treatment regimens according to the Austrian Neuroblastoma Trial A-NB87. Staging was done according to Evans criteria: stage I (localized without macroscopic lymphatic node involvement), stage IIA (macroscopic lymphatic node involvement), IIB (macroscopic residual tumor with ipsilateral lymphatic node involvement), IIIA (<2 years of age, ferritin < 300 µg/mL, neuron-specific enolase < 100 ng/mL), IIIB (any positivity of the markers described for stage IIIA), stage 4 (distant metastasized disease).