Neuroblastoma: the association of anatomical tumour site, molecular biology and patient outcomes

Numerous factors have been identified as carrying prognostic value in neuroblastoma (NB) and therefore incorporated in risk stratification of disease. Here, we investigate the association of anatomical site of NB with molecular biology and clinical outcomes.


Introduction
Neuroblastoma (NB) is the most common extracranial solid tumour in infants and children accounting for 8-10% of all malignant childhood tumours. 1 The most common tumour sites include abdomen (adrenal gland 48% and extra-adrenal retroperitoneum 25%) and thorax (16%) but NBs can also rarely arise from the pelvis (3%) and the neck (3%). 2 Numerous factors have been identified as carrying prognostic value in NB, many of which have been incorporated into various classification or staging systems. [2][3][4] The currently accepted International Neuroblastoma Risk Group staging system characterizes pre-treatment tumours according to the absence or presence of multiple image-defined risk factors and genetic characteristics of the tumour further stratified patients into low-/high-risk groups. 2 Amplification of MYCN oncoprotein is strongly associated with rapid disease progression and poor outcome in patients independent of age and tumour staging. 4,5 Other chromosomal aberrations associated with either whole DNA copy number alterations or incomplete segmental alterations have also been shown to predict NB behaviour. In particular, aggressive tumour behaviour and poor outcome are associated with deletions at the chromosomal region 1p36.3 6 or 11q23, 7 and with unbalanced gain of the long arm of chromosome 17 (17q21 to 17qter). 8 Tumour cell DNA content in NB falls into two main categories: near-diploidy or hyperdiploidy (e.g. triploidy)in patients younger than 18 months with metastatic disease, a near-diploid DNA content is a predictor of poor outcome. 9,10 More recently, mutations in several genes such as ALK, TERT, ATRX and PTPRD have been implicated in the outcome of NB patients 11 and authors have thus also suggested incorporating these novel biomarkers into existing risk prognostication system. 12 Several studies have reported that neuroblastic tumours originating from different anatomical sites follow diverse clinical outcomes. [13][14][15][16][17][18] Nevertheless, it is unclear here whether the tumour site alone carries prognostic significance or whether any survival benefit is due to the biological and molecular characteristics of the tumour cells. There is limited evidence available which has directly compared characteristics and clinical outcomes of abdominal and extra-abdominal NBs. 19,20 Therefore, in this current study, we aim to further investigate the association between anatomical site of NB and their clinical, biological and molecular characteristics with resultant clinical outcomes.

Methods
We undertook a retrospective analysis of all children diagnosed with NB between 1985 and 2013 identified from our institution's oncology database. One hundred and seventeen patients were identified, and tumour location was confirmed with computed tomography/magnetic resonance imaging. Data on molecular biology were obtained as testing became available. Chi-squared, Fisher's exact test and Kaplan-Meier log-rank tests were used for statistical/survival analysis. A significance level of P ≤ 0.05 (two-tailed) was set. Analyses were performed using JMP Pro, version 13.1.0 for Windows (SAS Institute Inc., Cary, NC, USA). Study was approved by the Department of Oncology and Pathology, Alder Hey Children's Hospital, Liverpool, UK. Ethical approval was not needed as this article does not contain any studies with human participants or animals performed by any of the authors.

Statistical modelling
We performed Cox proportional hazards regression analysis to fully investigate the associations between tumour site, age at diagnosis, MYCN status and tumour stage with 5-year OS. In our study series, only MYCN (P = 0.014) and age at diagnosis (P = 0.001) were identified to be independent prognostic factors (Table 4).

Discussion
The current study has shown statistically significant relationships between NB tumour site, their genetic characteristics and clinical outcome(s). Historical studies have previously suggested that thoracic neuroblastic tumours may be associated with better overall prognosis. 14,17,21 These observations have now been further reinforced by more contemporaneous work which has compared NB tumour site(s) with prognostic factors such as histology, MYCN status and biochemical markers. 15,19,22 Our findings have herein demonstrated that TNB has significantly better outcome(s) than ANB (Table 1) and that these patients are more likely to present with only locoregional disease (INSS stage II-III) compared to ANB lesions ( Table 2). TNB tumours likewise tended to exhibit favourable molecular biology profile(s), namely: negative MYCN amplification, negative segmental chromosomal alterations and DNA index >1 (Table 3).
Previously held consensus has identified TNB as a distinct disease subset that presents at earlier age. 17,23 We have found in this study no difference(s) in presenting age in our population which is also in keeping with findings from recent works. 15,19 The underlying mechanism(s) as to why TNB tumours have better survival outcome than ANB and why the thoracic location of the lesion itself confers independent prognostic value is subject to much debate. Multivariate analyses from a number of multicentre retrospective studies have shown conflicting findings.
Data from the Pediatric Oncology Group published by Morris et al. 17 showed that the thoracic location of tumour confers survival advantage(s) independent of DNA index, MYCN status and serum lactate dehydrogenase levels. This finding is also supported by a report from the International Neuroblastoma Risk Group 24 that demonstrated thoracic tumours had a lower hazard ratio compared to non-thoracic tumours after adjusting for patient age, MYCN status and stage of disease.
However, data from the German Cooperative Study Group NB90 15 showed that only tumour stage, MYCN status and serum lactate dehydrogenase were independent prognostic factors and not the location of tumour. In our current study, we herein report only MYCN status and age at primary diagnosis as independent prognostic factors and not the tumour location itself. This implies that the OS advantage of TNB over ANB is due to the inherent characteristics of TNB tumours rather than the anatomical location alone. These results also confirm our previous observation from a smaller cohort of patients. 25 Our findings are limited by sample size and availability of genetic analysis, nevertheless they corroborate well with other larger studies. 11 There is growing evidence in NB that genetic and molecular differences exist resulting in the fascinating and enigmatic behaviour of this tumour. Cooper et al. have shown that NB cells can 'arrest' at various levels of adrenal medullary cell differentiation and that a process of differentiation/dedifferentiation maybe responsible for the biological 'switch' from malignant to benign tumour phenotype in some cases of NB. 26 This intriguing hypothesis may also usefully be supported by in vivo laboratory work from our science group in the chick embryo NB model, which has demonstrated evidence of cell differentiation, reduced cell division and undetectable MYCN expression in MYCN-amplified NB cells implanted in the avian system that then migrate into the sympathetic ganglia. In non-neural locations, the implanted MYCN NB cells in the chick continued to rapidly proliferate aggressively and overexpress MYCN. 27 NB is therefore a 'molecular defined disease' greatly influenced by the genetic properties of the tumour cell. 11,28-32 NB tumours at   anatomical sites likely derive from a very distinct embryological milieu associated with unique genetic profiling and survival outcome(s). Future research should therefore be vigorously directed to encompass complete genetic and molecular biology profiling of neuroblastic tumours.

Conclusion
TNB tumours have better overall outcome(s) than primary adrenal neoplasms. This may be due to varied factors reported here including non-metastatic disease at presentation, non-amplification of the MYCN oncogene and favourable molecular biology.