Radiogenomics of neuroblastomas: Relationships between imaging phenotypes, tumor genomic profile and survival
Radiogenomics of neuroblastomas: Relationships between imaging phenotypes, tumor genomic profile and survival
Editor: Michael Baudis
HerveÂ J. Brisse 0
Gudrun Schleiermacher 1
VeÂ ronique Mosseri
Eve Lapouble 1
Paul FreÂ neaux
Jean Michon 1
0 Imaging Department, Institut Curie, Paris, France, 2 Paris Sciences et Lettres Research University , Paris , France , 3 Department of Pediatric Surgery, Necker-Enfants-Malades University Hospital, Assistance Publique-H oÃpitaux de Paris, Paris, France, 4 Paris-Descartes Sorbonne-Paris-Cit eÂ University , Paris , France
1 Department of Pediatric Oncology, Institut Curie, Paris, France, 6 Unit eÂ de GeÂneÂ tique Somatique, Institut Curie, Paris, France, 7 Department of Biostatistics, Institut Curie, Paris, France, 8 Department of Pediatric Surgery and Urology, Robert DebreÂ University Hospital, Assistance Publique-HoÃ pitaux de Paris, Paris, France, 9 Paris-Diderot Sorbonne-Paris-CiteÂ University , Paris , France , 10 Institut Curie, INSERM, U830, Equipe LabelliseÂe Ligue Contre le Cancer, Paris, France, 11 Department of Pathology, Robert DebreÂ University Hospital, Assistance Publique-H oÃpitaux de Paris, Paris, France, 12 Department of Biopathology, Institut Curie, Paris, France, 13 Department of Pathology, Necker-Enfants-Malades University Hospital, Assistance Publique-HoÃ pitaux de Paris , Paris , France
Funding: The funders had no role in study design,
data collection and analysis, decision to publish, or
preparation of the manuscript. This study was
supported by the following grants: Grant-4654
(Site de Recherche InteÂgreÂ sur le Cancer, Institut
National du Cancer, Direction GeÂneÂrale de l'Offre de
This study investigated relationships between neuroblastomas (NBs) imaging phenotypes, tumor genomic profile and patient outcome.
Patients and methods
This IRB-approved retrospective observational study included 133 NB patients (73 M, 60 F;
median age 15 months, range 0±151) treated in a single institution between 1998 and 2012.
A consensus review of imaging (CT-scan, MRI) categorized tumors according to both the
primarily involved compartment (i.e., neck, chest, abdomen or pelvis) and the sympathetic
anatomical structure the tumors rose from (i.e., cervical, paravertebral or periarterial chains,
or adrenal gland). Tumor shape, volume and image-defined surgical risk factors (IDRFs) at
diagnosis were recorded. Genomic profiles were assessed using array-based comparative
genomic hybridization and divided into three groups: ªnumerical-only chromosome
alterationsº (NCA), ªsegmental chromosome alterationsº (SCA) and ªMYCN amplificationº
(MNA). Statistical analyses included Kruskal±Wallis, Chi2 and Fisher's exact tests and the
Kaplan-Meier method with log-rank tests and Cox model for univariate and multivariate survival analyses.
Soins); http://www.e-cancer.fr/, http://social-sante.
AOM 02014 and PHRC IC2007-09 (Direction
GeÂneÂrale de l'Offre de Soins); http://social-sante.
Institut National de la SanteÂ et de la Recherche
MeÂdicale, http://www.inserm.fr/; Ligue Nationale
contre le Cancer (Equipe labelliseÂe), https://www.
ligue-cancer.net/; « Carte d'IdentiteÂ des Tumeurs »
program (Ligue Nationale Contre le Cancer),
https://www.ligue-cancer.net/; CEST of Institut
Curie, https://curie.fr/. The authors also thank the
following associations for their continuous
support: The Annenberg foundation, https://www.
annenberg.org/ The APAESIC (Association des
Parents et des Amis des Enfants SoigneÂs à l'Institut
Curie), http://www.apaesic.org/; The « Association
Hubert Gouin Enfance et Cancer », http://www.
enfance-et-cancer.org/; « Les Bagouz à Manon »,
http://www.lesbagouzamanon.org/; « Enfants et
SanteÂ », http://www.enfants-cancers-sante.fr/; «
Les Amis de Claire», https://curie.fr/page/
Competing interests: The authors have declared
that no competing interests exist.
A significant association between the sympathetic structure origin of tumors and genomic
profiles was demonstrated. NBs arising from cervical sympathetic chains were all NCA.
Paravertebral NBs were NCA or SCA in 75% and 25%, respectively and none were MNA.
Periarterial NBs were NCA, SCA or MNA in 33%, 56% and 11%, respectively. Adrenal NBs
were NCA, SCA or MNA in 16%, 36% and 48%, respectively. Among MNA NBs, 92%
originated from the adrenal gland. The sympathetic anatomical classification was significantly
better correlated to overall survival than the compartmental classification (P < .0003). The
tumor volume of MNA NBs was significantly higher than NCA or SCA NBs (P < .0001).
Patients with initial volume less than 160 mL had significantly better overall survival (P <
.009). A ªsingle massº pattern was significantly more frequent in NCA NBs (P = .0003). The
number of IDRFs was significantly higher in MNA NBs (P < .0001).
Imaging phenotypes of neuroblastomas, including tumor origin along the sympathetic system, correlate with tumor genomic profile and patient outcome.
Neuroblastomas (NBs) are the most common extracranial solid tumors in children. NBs derive
from the sympathetic nervous system originating from neural crest cells. Therefore, these
tumors may theoretically arise from any migratory pathway [
]. NBs mostly arise from the
abdomen (adrenal gland 48%, extra-adrenal retroperitoneum 25%), less frequently from the
chest (16%) and rarely from the pelvis (3%) or the neck (3%). NBs are associated with
remarkable biological heterogeneity and outcome. Some tumors undergo spontaneous
regression, some are cured by surgery alone or after chemo-reduction, while other exhibit extremely
Among prognostic factors previously identified, many are associated with each other and
define pretreatment risk groups[
]. Major prognostic factors are: the age at diagnosis (better
prognosis if younger than 18 months), the tumor stage (according to the International
Neuroblastoma Staging System (INSS)[
] or the International Neuroblastoma Risk Grouping Staging
]), the pathology based on the International Neuroblastoma Pathology
Classification (INPC)[6±8], various biological factors [
], and somatic genetic abnormalities,
especially the amplification of the MYCN oncogene, which occurs in 20 to 25% of NBs[
Whole-genome DNA copy number analysis with array-based comparative genomic
hybridization (aCGH) provided further critical prognostic information, especially in patients without
MYCN amplification [
]. Tumors that present exclusively whole-chromosome copy number
variations are associated with excellent survival, but tumors with any type of segmental
chromosome alterations exhibit a high risk of relapse[11±13], which increases with the number of
The anatomical location of the primary tumor, which can be assessed by imaging methods,
was also described as a prognostic factor [15±28]. However, the primary site was always
reported according to the anatomical compartment (e.g., neck, chest, abdomen or pelvis)
although NBs may arise from distinct sympathetic structures within a single compartment. A
sympathetic anatomical classification might be more relevant in terms of prognosis than a
simple compartmental anatomical classification.
2 / 17
Imaging techniques (CT-scan, MRI) recommended at diagnosis for disease staging[
allow identifying the tumor origin precisely, its volume, shape and its local extension and
might therefore represent a non-invasive method to get relevant prognostic information. To
the best of our knowledge, no study specifically investigated the relationships between imaging
patterns of NBs and survival or other prognostic factors, especially the genomic profile.
The present study therefore investigated relationships between the anatomical origin of
NBs along the sympathetic system, their imaging pattern, their genomic profile and patient
outcome. As a secondary objective, this study assessed the accuracy of imaging for the
identification of the primary tumor site compared to surgical and pathological findings.
Patients and methods
The Institut Curie Institutional Review Board approved this study. Written informed consent
was obtained from parents or guardians for inclusion in the clinical trials. Analysis of tumor
samples was performed according to the relevant national law on the protection of participants
in biomedical research. This retrospective observational study was conducted according to
The following inclusion criteria were used: (1) patient age 18 years or younger at diagnosis, (2)
patient referred to our institution before 2013, (3) cytologically or histologically proven NB,
(4) frozen tumor material obtained at the time of diagnosis that enabled DNA extraction and
molecular analysis, (5) availability of DICOM (Digital Imaging and Communications in
Medicine) imaging data (i.e., CT scan or MRI) at diagnosis and during follow-up, (6) operative
report availability in cases of surgery, and (7) availability of pathological reports of material
Our institutional database identified 203 eligible patients. DICOM data were missing for 56
patients, aCGH data for 11 patients and follow-up data for 3 patients. As a result, 133 patients
were included for analysis.
Clinical, surgical and pathological data
Clinical, surgical and pathological data was extracted from medical charts including surgical
and pathological reports. INPC classification (i.e., ªfavorableº or ªunfavorableº subtypes)
derived from the pathological description at diagnosis, except when cellular material was
obtained using only fine-needle aspirates.
All patients were treated using previous or ongoing protocols or trials of the International
Society of Paediatric Oncology European Neuroblastoma (SIOPEN) (NCT 01704716, 00025
428, 00025597, 00025649, 00025623, 00025610; details available in S1 File): 18 patients with
localized disease had surgical removal of their tumor only, 86 patients were operated after
neoadjuvant chemotherapy, and 27 patients were treated with chemotherapy without further
surgery (either no local treatment or radiation therapy in case of high-risk tumors, i.e.
INSSstage 4 or INSS-stage 2±4 MNA tumors). Two patients had only clinical and radiological
follow-up without any treatment.
Imaging data analyses
Imaging data was extracted from the Picture Archiving and Communicating System (PACS)
of our institution (V 11.3, Carestream Health, Vaughan, Canada) and centrally reviewed in
consensus by one senior pediatric radiologist (HJB, 20 years' experience), two senior (SS, PPC,
3 / 17
20 and 15 years' experience) and one fellow (TB, 3 years' experience) pediatric surgeons who
were blinded to any other data. Data included 207 CT scans and 62 MRI (24 patients had both
examinations). Due to the retrospective nature of the study, images used for assessing tumor
location and size were either enhanced-CT images, T1-weighted images (WI), T2-WI or
contrast-enhanced T1-WI. Each tumor was classified as originating from one of the four
anatomical compartments (i.e., neck, chest, abdomen or pelvis) and from one of the following
sympathetic structure groups (according to the international Terminologia Anatomica): (1)
cervical sympathetic chains (i.e., superior, middle and inferior cervical and cervicothoracic
ganglia), (2) paravertebral sympathetic chains (i.e., thoracic, lumbar and sacral ganglia), (3)
periarterial sympathetic pathways (i.e., thoracic aortic, abdominal aortic and celiac plexus,
aorticorenal ganglia, superior and inferior mesenteric, superior hypogastric and iliac plexus), and
(4) adrenal glands (Fig 1). NBs originating from lumbar ganglia and iliac plexus were classified
as originating from the abdomen, and tumors originating from the superior hypogastric plexus
(i.e., in the division angle of the aorta at L5 level) were classified in the pelvis. Tumor origin
according to sympathetic anatomy was first assessed on imaging at diagnosis only and then
compared to imaging after chemo-reduction and surgical and pathological data. The primary
site was finally allocated by consensus on the basis of all available data.
Tumor volume at diagnosis was calculated from measurements in three perpendicular
dimensions based on an elliptical estimate (volume = length x width x thickness x ᴨ/6). When
preoperative chemo-reduction was used, preoperative tumor volumes and tumor volume
decrease were calculated (Tumor volume decrease (%) = 100 x [1ÐResidual Volume / Initial
Volume]). Tumor shape was classified as ªsingle massº or ªmultiple confluent massesº.
ImageFig 1. Radiogenomics classification of neuroblastomas according to anatomical origin.
Neuroblastomas may be classified based on the anatomical compartment (i.e., neck, chest, abdomen or pelvis) or
according to the sympathetic structure the tumors arise from, i.e., (1) the cervical sympathetic chains (i.e.,
including the superior, middle and inferior cervical and the cervicothoracic ganglia (g.)); (2) the paravertebral
(PV) sympathetic chains (i.e., including all thoracic, lumbar and sacral ganglia); (3) the periarterial (PA)
sympathetic pathways (i.e., including the thoracic aortic, abdominal aortic and celiac plexus (pl.)), the
aorticorenal ganglia, and the superior and inferior mesenteric, superior hypogastric and iliac plexus); and (4)
the adrenal glands. For each compartment or sympathetic group, the pie charts show the distribution of the
genomic profile of the tumors, i.e., numerical-only chromosome alterations (NCA), segmental chromosome
alterations (SCA) or MYCN amplification (MNA).
4 / 17
defined surgical risk factors (IDRFs) were assessed according to the published list and
The following genetic analyses were included: (1) the MYCN status assessed using fluorescence
in situ hybridization (FISH) and an MYCN probe (Zymed Laboratories, San Francisco, CA,
USA) on frozen sections for tumor fragments or cytogenetic preparations of fine-needle
aspirates according to recommendations of the INRG Biology Committee[
], and (2) aCGH
performed on BAC/PAC or NimbleGen arrays, as previously described[
]. DNA was
extracted from tumor specimens that were obtained at diagnosis and exhibited a tumor
cellularity higher than 50%. The results were analyzed on the VAMP site using the GLAD algorithm
and submitted to visual inspection[
]. Three genomic types were defined:
ªnumericalonly chromosomal alterationsº (NCA) type, which only included numerical changes in whole
chromosomes without any detectable structural rearrangement; ªsegmental chromosomal
alterationsº (SCA) type, which was characterized by any partial chromosome imbalances,
excluding MYCN amplification, with or without associated numerical aberrations; and
ªMYCN-amplifiedº type (MNA), which exhibited MYCN amplification, with or without
segmental or numerical aberrations[
Comparisons between genomic profiles and continuous variables were performed using the
non-parametric Kruskal±Wallis test. Comparisons between genomic profiles and discrete
variables were performed using the Chi2 test or Fisher's exact test, if necessary. Event-free survival
(EFS) was defined as the time from diagnosis to first event (local or metastatic failure, or
death). Patients with no events were censored at the time of last follow-up. Overall survival
(OS) was defined as the time from diagnosis to death from any cause or last follow-up. Survival
curves were analyzed using the Kaplan-Meier method and results were compared using the
log-rank test. The 5-year rates were expressed together with their standard error (SE). For each
variable, relative risks were estimated using a univariate Cox model and expressed with their
95% confidence interval. Relationships between anatomical origin and imaging pattern of
neuroblastomas and survival were assessed by a multivariate analysis using a Cox model with a
forward procedure. Multivariate analysis was performed among variables demonstrating
significance by univariate analysis. Hence, the relationship between survival and the two
anatomical classifications (compartmental and sympathetic) and tumor volume was assessed on OS
only. P-values less than .05 were considered statistically significant. The anonymized data set
necessary to replicate our study findings are at the European Genome-phenome Archive
(study accession number: EGAS00001002651).
Study cohort characteristics and relationships between patient age,
INPC, stage and genomic profile
A total of 133 patients (73 males, 60 females) treated between 1998 and 2012 were included.
Median age at diagnosis was 15 months (range, 0±151) (Table 1). Fifty-nine percent of the
children were younger than 18 months at diagnosis. Fifty-one percent of the children had
metastatic disease (63 children stage 4, 5 children stage 4S). Univariate analyses demonstrated that
patient age, INPC and stage significantly correlated with the genomic profile, i.e., a NCA
5 / 17
(1) NCA: numerical-only chromosome alterations; SCA: segmental chromosome alterations; MNA: MYCN-amplification.
(2) Available data for only 74 of the 133 patients.
profile being observed more frequently in younger children, in NBs with favorable INPC and
in lower stage disease (Table 1).
Relationships between tumor origin, genomic profile and outcome
Both anatomical classifications, i.e., the compartmental one and the sympathetic one,
significantly correlated with the genomic profile (Table 2, Fig 1).
According to the compartmental anatomical classification, all tumors arising from the neck
or pelvis were NCA NBs. In the chest, 62% were NCA, 33% SCA and only 5% MNA.
Abdominal tumors were more widely distributed with 21% NCA, 42% SCA and 37% MNA.
Eightyfive percent of SCA and 97% of MNA NBs were observed in the abdomen.
With regards to their sympathetic structure origin, all tumors arising from the cervical
sympathetic chains were NCA. Among paravertebral tumors, 75% were NCA, 25% SCA and none
MNA. Among periarterial tumors, 56% were SCA, 33% NCA and 11% were MNA. The
distribution of genomic profiles was wider among adrenal tumors: 16% were NCA, 36% SCA and
48% MNA. Ninety-two percent of MNA NBs originated from the adrenal gland.
Within a single anatomic compartment, the sympathetic chain or structure from which
NBs originated allowed differentiation of tumors with distinct biological features. Ninety
percent (19/21) of chest tumors arose from the paravertebral chains (Table 2, Fig 2A and 2B) and
these tumors were mostly NCA (68%) or SCA (32%), whereas the two tumors that originated
from the mediastinal periarterial sympathetic pathways were SCA and MNA, respectively (Fig
2D and 2E). Within the abdomen, 73% (69/95) of NBs were of adrenal origin (Fig 3A), 23%
(22/95) were periarterial (27% NCA, 64% SCA and 9% NMA) (Fig 3B) and 4% (4/95) were
lumbar paravertebral (3 NCA and 1 SCA) (Fig 3C). Among pelvic tumors, 62.5% (5/8) were
presacral paravertebral NBs and 37.5% (3/8) periarterial (superior hypogastric plexus) and all
were NCA NBs.
The anatomical origin of NBs also significantly related to outcome (Table 3, Fig 4). The
compartmental classification revealed that abdominal NBs exhibited a significantly lower
6 / 17
5-year OS and a trend of lower 5-year EFS than extra-abdominal primaries. Survival was not
significantly different between neck, chest and pelvic tumors. The sympathetic classification
demonstrated that adrenal NBs had a significantly lower 5-year OS (and a trend of lower
5-year EFS) compared to extra-adrenal primary tumors. EFS was not significantly different
among the four sympathetic origin groups (P = .113). No significant EFS difference was
observed between cervical and paravertebral NBs (P > .87) on one hand, neither between
adrenal and periarterial NBs (P > .90) on the other hand. However, pooled NBs originating
from adrenal gland or periarterial sympathetic chains had significantly lower EFS that pooled
NBs originating from cervical or paravertebral chains (P < .0015). Multivariate analysis
demonstrated that the sympathetic anatomic classification was significantly more relevant than the
compartmental one for the prediction of OS (P < .0003).
Contribution of initial imaging to the classification of tumor origin according to sympathetic system anatomy
Initial imaging was judged relevant to allocate the sympathetic origin of the primary in 77%
(102/133) of cases. Review of imaging after chemotherapy (Fig 2C) modified the interpretation
of tumor origin in 14% of cases. Surgical and pathological reports provided additional relevant
anatomical details that were not depicted by imaging at diagnosis in 11% and 12% of cases,
respectively. The mean tumor volume of patients for whom post-chemotherapy imaging,
surgical or pathological reports provided additional information was significantly higher than
that of patients with anatomical location that was correctly judged on initial imaging (362 mL
7 / 17
Fig 2. Imaging phenotypes of chest neuroblastomas. (a, b, c) Newborn with L2-stage left posterior
mediastinal NCA neuroblastoma. MRI at diagnosis (a, b: sagittal and axial T2-weighted sequences). The
primary tumor is a unique well-delineated mass (*) with focal contact with the thoracic aorta (arrowhead) and
intra-spinal extension (arrow). Follow-up MRI 3 months later (c) after neoadjuvant chemotherapy (2 courses
of cyclophosphamide-vincristine and 2 courses of etoposide-carboplatin) shows the tumor residue precisely
located at the costo-vertebral junction, i.e., a paravertebral sympathetic chain location. (d, e) 9-year-old girl
8 / 17
with M-stage mediastinal SCA neuroblastoma. Enhanced CT scan at diagnosis (axial and coronal views).
The primary tumor (*) is ill-defined and diffusely infiltrates the posterior mediastinum, pleura and chest wall
(arrows), crosses the midline and encases the thoracic aorta (arrowhead). The presumed origins are the
mediastinal sympathetic fibers surrounding the descending aorta.
versus 223 mL, P = .0087). The anatomical compartment in the former group was mainly the
abdomen (25/31), and the sympathetic origin was primarily the adrenal gland (14/31) or
periarterial pathways (10/31).
Relationships between tumor imaging pattern (IDRFs, shape, volume, tumor volume decrease), genomic profiles and outcome
The occurrence of any IDRF was not significantly related to the genomic profile. However, the
median number of IDRFs among the 122 IDRF-positive tumors was significantly higher in
MNA NBs (Table 4). The 5-year OS and EFS rates of IDRF-negative patients were higher than
IDRF-positive patients but these differences were not significant (Table 3).
The ªsingle massº pattern was significantly more frequent than the ªmultiple confluent
massesº pattern in NCA (84%) compared to SCA and MNA NBs (55% and 44%, respectively).
Mean tumor largest diameter and volume at diagnosis were 8.5 cm and 255 mL,
respectively. A tumor volume of 160 mL (median volume) or less at diagnosis was significantly
associated with better OS, but not significantly better EFS. Initial tumor volume of MNA NBs was
significantly higher than NCA and SCA NBs. The tumor volume decrease after neoadjuvant
chemotherapy was significantly higher in MNA tumors but was not significantly associated
NBs in a single anatomical compartment may derive from distinct sympathetic structures
which are associated with distinct genomic profiles. Therefore, the precise anatomical origin
of the primary tumor is of special interest. We described statistically significant relationships
between the sympathetic origin of tumors and their genomic profile and outcome. Our data
also confirmed that initial imaging efficiently depicted the sympathetic origin of the tumor
compared to surgical and pathological findings, except for very large abdominal masses for
which post-chemotherapy imaging was more accurate, i.e., when the tumor shrinks on its
original sympathetic structure.
It has long been suggested that tumor behavior may differ based on the primary tumor
location and adrenal NBs are known to be associated with poorer prognosis[
]. Other studies
enhanced this concept by comparing the primary tumor site with patient survival rates or
other prognostic factors, such as tumor stage, histology and serum markers[
]. Chest NBs
are associated with better outcome among the extra-abdominal sites[
multivariate analysis in a large retrospective study[
] did not identify the chest location as an
independent prognostic factor. Cervical and pelvic NBs are also associated with better prognosis,
although these results were based on smaller series[
15, 18, 19, 22, 24, 25
]. The INRG recently
confirmed that adrenal tumors were more likely than non-adrenal tumors to have MNA, and
thoracic tumors were less likely than non-thoracic tumors to have MNA. Our
wholegenome DNA copy number analysis allowed the identification of significant relationships
between cervical and pelvic locations and NCA profile, and between chest location and
nonMNA NBs. It also confirmed the strong association between abdominal location and MNA
profile and poorer outcome.
9 / 17
Fig 3. Imaging phenotypes of abdominal neuroblastomas. (a) 18-month-old girl with M-stage right
adrenal MNA neuroblastoma. Enhanced CT scan at diagnosis. The primary tumor (*) is centered on the
right adrenal gland between the right kidney and the inferior vena cava (arrow) and extends medially in
contact with the aorta (arrowhead). (b) 12-month-old girl with M-stage retroperitoneal periarterial SCA
neuroblastoma. Enhanced CT scan at diagnosis. The primary tumor (*) is centered in the median
retroperitoneum around the aorta (arrowhead) and behind the inferior vena cava (arrow). (c) Newborn with
L2-stage lumbar dumbbell paravertebral NCA neuroblastoma. Axial T2-weighted MRI at diagnosis. The
primary tumor (*) is centered on the right paravertebral chain and invades the psoas and spinal muscles and
fills the spinal canal, compressing the spinal cord (dotted arrow). The tumor is totally separated from the
inferior vena cava (arrow) and the aorta (arrowhead).
10 / 17
(1) EFS: event-free survival; OS: overall survival; SE: standard error. The mean follow-up of the cohort was 83 months (range, 1±175 mo). The 5-year EFS
and OS of the entire cohort were 58.6% (+/- 4.3%) and 74.6% (+/- 3.8%), respectively.
(2) RR: relative risk of events; CI: confidence interval.
(3) RR: relative risk of death.
(4) Relative risk has been estimated in abdominal tumours compared to cervical, thoracic or pelvic tumours, as no death occurred in patients with pelvic
tumours (RR could not be estimated by the Cox model for this subgroup of patients). For homogeneity of results, the same cluster has been done for
estimating EFS' hazard ratio. However, Logrank tests compare survival and EFS of the 4 anatomical compartments.
(5) (1 missing data).
By using a sympathetic anatomical classification for the first time instead of the classical
compartmental one, our study provided a better differentiation of outcome. This result is
explained by the occurrence of distinct genomic profiles within each compartment. The
recognition of a paravertebral sympathetic origin is notable because these tumors are not associated
with MNA type and mostly associated to the favorable NCA profile (75% in this series).
Imaging identifies paravertebral tumors as arising in the chest (ªcosto-vertebralº NB), abdomen
(ªlumbarº tumors) or pelvis (ªpresacralº tumors), possibly associated with intra-spinal
extension (ªdumbbellº tumors). Imaging also identifies periarterial tumors, which are observed in
various compartments, primarily the abdomen around the aorta or its branches, and
occasionally in the pelvis (superior hypogastric plexus). In the chest, the use of the sympathetic
classification allows differentiation between periarterial mediastinal and paravertebral tumors.
Although more widely distributed, genomic profiles of periarterial NBs were less favorable
than those of paravertebral NBs, including 56% SCA and 11% MNA types. Finally, the adrenal
gland was the origin of most MNA NBs (92% in this series).
Together our data supports the hypothesis that genomic profiles and the aggressiveness of
NBs may be associated with distinct neural crest cell-derived elements. During embryogenesis,
neural crest cells emerge early in development and a defined region gives rise to precursor cells
that differentiate into the adrenal medulla and sympathetic ganglia[
]. The exact mechanisms
that lead to tumorigenesis are not fully determined. According to our data, it is remarkable that
the most distally (i.e., cervical and presacral) and dorsally (i.e., paravertebral) migrating cells are
mostly associated with the favorable NCA genomic profile, whereas adrenal and periarterial
tumors are more associated with the less favorable SCA and MNA profiles.
11 / 17
Fig 4. Kaplan-Meier survival analysis according to anatomical classifications of primary tumors. Event-free
survival (EFS) and overall survival (OS) according to the anatomical compartment (a, b) and sympathetic system
origin (c, d) of the primary tumor.
Our study also demonstrated links between tumor volume, shape, tumor volume decrease
and genomic profile and outcome. A small tumor volume was significantly associated with
non-MNA NBs and better survival. A single mass was more frequently associated with
favorable NCA NBs than multiple confluent masses. The link between IDRFs and outcome remains
controversial in the literature[
]. We did not identify a significant link with genomic
profile or outcome among localized tumors, but this result may be related to insufficient statistical
power. However, a high number of IDRFs was significantly related to the MNA type. Tumor
volume reduction during the early phase of induction chemotherapy in high-risk NBs was
reported as associated with a better outcome. The tumor volume decrease in this study,
which included any risk-group NBs, was not associated with survival, and MNA NBs were
associated with a higher tumor volume decrease compared to SCA and NCA tumors.
We acknowledge that our study includes limitations. As the primary sympathetic structure
is usually distorted by the tumor, the exact origin of the tumor may be difficult to assess. The
12 / 17
(1) NCA: numerical-only chromosome alterations; SCA: segmental chromosome alterations; MNA: MYCN-amplification.
(2) IDRF number among the 122 IDRF-positive NBs.
reference anatomical location used in this study was based on imaging data obtained before
and after neoadjuvant chemotherapy, as well as on the macroscopic and microscopic
descriptions from the surgeons and pathologists. For adrenal tumors, the residual gland invaded by
tumor cells and surrounding the tumor was helpful in defining the tumor origin. For other
locations, we were only able to define the most probable sympathetic chain involved. Other
limitations of our work are the relatively small size of our cohort and the non-inclusion of
highly differentiated NBs, such as intermixed ganglioneuroblastomas and ganglioneuromas,
for which aCGH profiles are not contributive. Actually, the quality of aCGH interpretation is
directly correlated to tumor cells content and the assessment of the exact balance between
immature neuroblastic cells, ganglion cells, Schwann cells and stroma is challenging in mature
tumors. Very few samples reach the minimum threshold of tumor cells content required (i.e.,
50%). Actually, normal cells and subclonal populations tend to lessen the dynamic of genomic
profiles which can be totally flat. When aCGH fits all quality controls but shows a flat profile,
the result is tagged as ªnot contributiveº. However, prognostic information is less useful for
those tumors that share a comparable good prognosis[
]. Functional imaging was not
addressed because of the retrospective nature of the study. Early results using
diffusionweighted MRI suggested that neuroblastoma and ganglioneuroma / ganglioneuroblastoma
might be differentiated using this method[
]. Among metastatic NBs, 123-I-MIBG scan was
recently used to differentiate MNA from non-MNA NBs[
]. Functional imaging may actually
provide additional diagnostic and prognostic information in the future.
Imaging phenotypes of neuroblastomas correlate with tumor genomic profile and patient
outcome. If confirmed in a larger study cohort, the combination of anatomical data (sympathetic
structure origin) and morphological pattern (volume, shape, number of IDRFs) may represent
relevant prognostic criteria. In addition, our data potentially suggest distinct tumor genesis
pathways according to the neural crest cells origin that could contribute to a better
understanding of the observed genomic profiles.
S1 File. Supporting information on treatment regimens.
13 / 17
The authors thank the following physicians, pediatric oncologists, radiologists and surgeons of
the SocieÂteÂ FrancËaise de lutte Contre les Cancers et les leuceÂmies de l'Enfant et de l'adolescent
(SFCE) for their contribution: Daniel Orbach, FrancËois Doz, HeÂlène Pacquement, Franck
Bourdeaut, Isabelle Aerts, CeÂcile Cellier, Sylvia Neuenschwander, Yves Aigrain, and Sabine Irtan.
The authors also thank the "Tumorothèque Necker-Enfants-Malades" team, which partially
contributed to the management of frozen tumor specimens.
Conceptualization: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, VeÂronique
Mosseri, Pascale Philippe-Chomette, Isabelle Janoueix-Lerosey, Gaelle Pierron, Michel
Peuchmaur, Jean Michon, Olivier Delattre, Sabine Sarnacki.
Data curation: Thomas Blanc, VeÂronique Mosseri, Gaelle Pierron, Eve Lapouble, Nathalie
Formal analysis: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, VeÂronique Mosseri,
Michel Peuchmaur, Jean Michon, Sabine Sarnacki.
Funding acquisition: Gudrun Schleiermacher, VeÂronique Mosseri, Jean Michon, Olivier
Investigation: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, VeÂronique Mosseri,
Pascale Philippe-Chomette, Gaelle Pierron, Michel Peuchmaur, Paul FreÂneaux, Louise
Galmiche, Nathalie Algret, Matthieu Peycelon, Jean Michon, Sabine Sarnacki.
Methodology: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, VeÂronique Mosseri,
Isabelle Janoueix-Lerosey, Gaelle Pierron, Michel Peuchmaur, Nathalie Algret, Jean
Michon, Olivier Delattre, Sabine Sarnacki.
Project administration: HerveÂ J. Brisse.
Resources: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, Pascale
Philippe-Chomette, Isabelle Janoueix-Lerosey, Gaelle Pierron, Michel Peuchmaur, Paul FreÂneaux, Louise
Galmiche, Matthieu Peycelon, Jean Michon, Sabine Sarnacki.
Supervision: HerveÂ J. Brisse, Isabelle Janoueix-Lerosey, Michel Peuchmaur, Jean Michon,
Olivier Delattre, Sabine Sarnacki.
Validation: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, VeÂronique Mosseri,
Pascale Philippe-Chomette, Isabelle Janoueix-Lerosey, Gaelle Pierron, Eve Lapouble, Michel
Peuchmaur, Paul FreÂneaux, Louise Galmiche, Nathalie Algret, Matthieu Peycelon, Jean
Michon, Olivier Delattre, Sabine Sarnacki.
Visualization: HerveÂ J. Brisse, Thomas Blanc, VeÂronique Mosseri.
Writing ± original draft: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher, VeÂronique
Mosseri, Isabelle Janoueix-Lerosey, Gaelle Pierron, Michel Peuchmaur, Sabine Sarnacki.
Writing ± review & editing: HerveÂ J. Brisse, Thomas Blanc, Gudrun Schleiermacher,
VeÂronique Mosseri, Gaelle Pierron, Michel Peuchmaur, Sabine Sarnacki.
14 / 17
15 / 17
16 / 17
1. Turkel SB , Itabashi HH . The natural history of neuroblastic cells in the fetal adrenal gland . Am J Pathol . 1974 ; 76 ( 2 ): 225 ± 44 . Epub 1974/08/01. PMID: 4843383; PubMed Central PMCID : PMCPMC1910868 .
2. Takahashi Y , Sipp D , Enomoto H . Tissue interactions in neural crest cell development and disease . Science . 2013 ; 341 ( 6148 ): 860 ±3. https://doi.org/10.1126/science.1230717 PMID: 23970693 .
3. Cohn SL , Pearson AD , London WB , Monclair T , Ambros PF , Brodeur GM , et al. The International Neuroblastoma Risk Group (INRG) classification system: an INRG Task Force report . J Clin Oncol . 2009 ; 27 ( 2 ): 289 ± 97 . https://doi.org/10.1200/JCO. 2008 . 16 .6785 PMID: 19047291; PubMed Central PMCID : PMCPMC2650388 .
4. Brodeur GM , Pritchard J , Berthold F , Carlsen NL , Castel V , Castelberry RP , et al. Revisions of the international criteria for neuroblastoma diagnosis, staging, and response to treatment . J Clin Oncol . 1993 ; 11 ( 8 ): 1466 ± 77 . https://doi.org/10.1200/JCO. 1993 . 11 .8.1466 PMID: 8336186 .
5. Monclair T , Brodeur GM , Ambros PF , Brisse HJ , Cecchetto G , Holmes K , et al. The International Neuroblastoma Risk Group (INRG) staging system: an INRG Task Force report . J Clin Oncol . 2009 ; 27 ( 2 ): 298 ± 303 . https://doi.org/10.1200/JCO. 2008 . 16 .6876 PMID: 19047290; PubMed Central PMCID : PMCPMC2650389 .
6. Shimada H , Umehara S , Monobe Y , Hachitanda Y , Nakagawa A , Goto S , et al. International neuroblastoma pathology classification for prognostic evaluation of patients with peripheral neuroblastic tumors: a report from the Children's Cancer Group . Cancer . 2001 ; 92 ( 9 ): 2451 ± 61 . PMID: 11745303 .
7. Shimada H , Ambros IM , Dehner LP , Hata J , Joshi VV , Roald B . Terminology and morphologic criteria of neuroblastic tumors: recommendations by the International Neuroblastoma Pathology Committee . Cancer . 1999 ; 86 ( 2 ): 349 ± 63 . PMID: 10421272 .
8. Peuchmaur M , d'Amore ES , Joshi VV , Hata J , Roald B , Dehner LP , et al. Revision of the International Neuroblastoma Pathology Classification: confirmation of favorable and unfavorable prognostic subsets in ganglioneuroblastoma, nodular . Cancer . 2003 ; 98 ( 10 ): 2274 ± 81 . https://doi.org/10.1002/cncr.11773 PMID: 14601099
9. Seeger RC , Brodeur GM , Sather H , Dalton A , Siegel SE , Wong KY , et al. Association of multiple copies of the N-myc oncogene with rapid progression of neuroblastomas . N Engl J Med . 1985 ; 313 ( 18 ): 1111 ± 6. https://doi.org/10.1056/NEJM198510313131802 PMID: 4047115 .
10. Schleiermacher G , Michon J , Huon I , d'Enghien CD , Klijanienko J , Brisse H , et al. Chromosomal CGH identifies patients with a higher risk of relapse in neuroblastoma without MYCN amplification . British journal of cancer . 2007 ; 97 ( 2 ): 238 ± 46 . https://doi.org/10.1038/sj.bjc.6603820 PMID: 17579628; PubMed Central PMCID : PMCPMC2360301 .
11. Janoueix-Lerosey I , Schleiermacher G , Michels E , Mosseri V , Ribeiro A , Lequin D , et al. Overall genomic pattern is a predictor of outcome in neuroblastoma . J Clin Oncol . 2009 ; 27 ( 7 ): 1026 ± 33 . https://doi. org/10.1200/JCO. 2008 . 16 .0630 PMID: 19171713 .
12. Schleiermacher G , Mosseri V , London WB , Maris JM , Brodeur GM , Attiyeh E , et al. Segmental chromosomal alterations have prognostic impact in neuroblastoma: a report from the INRG project . British journal of cancer . 2012 ; 107 ( 8 ): 1418 ± 22 . https://doi.org/10.1038/bjc. 2012 .375 PMID: 22976801; PubMed Central PMCID : PMCPMC3494425 .
13. Schleiermacher G , Michon J , Ribeiro A , Pierron G , Mosseri V , Rubie H , et al. Segmental chromosomal alterations lead to a higher risk of relapse in infants with MYCN-non-amplified localised unresectable/ disseminated neuroblastoma (a SIOPEN collaborative study) . British journal of cancer . 2011 ; 105 ( 12 ): 1940 ± 8 . https://doi.org/10.1038/bjc. 2011 .472 PMID: 22146831; PubMed Central PMCID : PMCPMC3251887 .
14. Schleiermacher G , Janoueix-Lerosey I , Ribeiro A , Klijanienko J , Couturier J , Pierron G , et al. Accumulation of segmental alterations determines progression in neuroblastoma . J Clin Oncol . 2010 ; 28 ( 19 ): 3122 ± 30 . https://doi.org/10.1200/JCO. 2009 . 26 .7955 PMID: 20516441 .
15. Abramson SJ , Berdon WE , Ruzal-Shapiro C , Stolar C , Garvin J . Cervical neuroblastoma in eleven infantsÐa tumor with favorable prognosis. Clinical and radiologic (US, CT, MRI) findings . Pediatr Radiol . 1993 ; 23 ( 4 ): 253 ± 7 . PMID: 8414748 .
16. Adams GA , Shochat SJ , Smith EI , Shuster JJ , Joshi VV , Altshuler G , et al. Thoracic neuroblastoma: a Pediatric Oncology Group study . J Pediatr Surg . 1993 ; 28 ( 3 ): 372 ± 7; discussion 7±8 . PMID: 8468649 .
17. Cotterill SJ , Pearson AD , Pritchard J , Foot AB , Roald B , Kohler JA , et al. Clinical prognostic factors in 1277 patients with neuroblastoma: results of The European Neuroblastoma Study Group 'Survey' 1982±1992 . Eur J Cancer . 2000 ; 36 ( 7 ): 901 ± 8 . PMID: 10785596
18. Cruccetti A , Kiely EM , Spitz L , Drake DP , Pritchard J , Pierro A . Pelvic neuroblastoma: low mortality and high morbidity . J Pediatr Surg . 2000 ; 35 ( 5 ): 724 ± 8 . Epub 2000/05/17. https://doi.org/10.1053/jpsu. 2000 . 6076 PMID: 10813335 .
19. Haase GM , O'Leary MC , Stram DO , Lukens JN , Seeger RC , Shimada H , et al. Pelvic neuroblastomaÐ implications for a new favorable subgroup: a Children's Cancer Group experience . Ann Surg Oncol . 1995 ; 2 ( 6 ): 516 ± 23 . PMID: 8591082
20. Haberle B , Hero B , Berthold F , von Schweinitz D. Characteristics and outcome of thoracic neuroblastoma . Eur J Pediatr Surg . 2002 ; 12 ( 3 ): 145 ± 50 . https://doi.org/10.1055/s-2002-32721 PMID: 12101494 .
21. Horiuchi A , Muraji T , Tsugawa C , Nishijima E , Satho S , Takamizawa S , et al. Thoracic neuroblastoma: outcome of incomplete resection . Pediatric surgery international. 2004 ; 20 ( 9 ): 714 ±8. https://doi.org/10. 1007/s00383-003-1049-7 PMID: 15278374 .
22. Leclair MD , Hartmann O , Heloury Y , Fourcade L , Laprie A , Mechinaud F , et al. Localized pelvic neuroblastoma: excellent survival and low morbidity with tailored therapyÐthe 10-year experience of the French Society of Pediatric Oncology . J Clin Oncol . 2004 ; 22 ( 9 ): 1689 ± 95 . https://doi.org/10.1200/JCO. 2004 . 04 .069 PMID: 15117991 .
23. Morris JA , Shcochat SJ , Smith EI , Look AT , Brodeur GM , Cantor AB , et al. Biological variables in thoracic neuroblastoma: a Pediatric Oncology Group study . J Pediatr Surg . 1995 ; 30 ( 2 ): 296 ±302; discussion - 3 . PMID: 7738754 .
24. Moukheiber AK , Nicollas R , Roman S , Coze C , Triglia JM . Primary pediatric neuroblastic tumors of the neck . Int J Pediatr Otorhinolaryngol . 2001 ; 60 ( 2 ): 155 ± 61 . PMID: 11518594 .
25. Qureshi SS , Kembhavi S , Ramadwar M , Chinnaswamy G , Laskar S , Talole S , et al. Outcome and morbidity of surgical resection of primary cervical and cervicothoracic neuroblastoma in children: a comparative analysis . Pediatric surgery international . 2014 ; 30 ( 3 ): 267 ± 73 . https://doi.org/10.1007/s00383-013- 3431-4 PMID: 24264427 .
26. Suita S , Tajiri T , Sera Y , Takamatsu H , Mizote H , Ohgami H , et al. The characteristics of mediastinal neuroblastoma . Eur J Pediatr Surg . 2000 ; 10 ( 6 ): 353 ± 9 . Epub 2001/02/24. https://doi.org/10.1055/s2008-1072390 PMID: 11215774.
27. Sung KW , Yoo KH , Koo HH , Kim JY , Cho EJ , Seo YL , et al. Neuroblastoma originating from extraabdominal sites: association with favorable clinical and biological features . J Korean Med Sci . 2009 ; 24 ( 3 ): 461 ± 7 . Epub 2009/06/23. https://doi.org/10.3346/jkms. 2009 . 24 .3.461 PMID: 19543510; PubMed Central PMCID : PMCPMC2698193 .
28. Vo KT , Matthay KK , Neuhaus J , London WB , Hero B , Ambros PF , et al. Clinical, biologic, and prognostic differences on the basis of primary tumor site in neuroblastoma: a report from the international neuroblastoma risk group project . J Clin Oncol . 2014 ; 32 ( 28 ): 3169 ± 76 . https://doi.org/10.1200/JCO. 2014 . 56 . 1621 PMID: 25154816; PubMed Central PMCID : PMCPMC4171360 .
29. Brisse HJ , McCarville MB , Granata C , Krug KB , Wootton-Gorges SL , Kanegawa K , et al. Guidelines for imaging and staging of neuroblastic tumors: consensus report from the International Neuroblastoma Risk Group Project . Radiology . 2011 ; 261 ( 1 ): 243 ± 57 . https://doi.org/10.1148/radiol.11101352 PMID: 21586679 .
30. Ambros PF , Ambros IM , Brodeur GM , Haber M , Khan J , Nakagawara A , et al. International consensus for neuroblastoma molecular diagnostics: report from the International Neuroblastoma Risk Group (INRG) Biology Committee . British journal of cancer . 2009 ; 100 ( 9 ): 1471 ± 82 . https://doi.org/10.1038/sj. bjc.6605014 PMID: 19401703; PubMed Central PMCID : PMCPMC2694415 .
31. Hupe P , Stransky N , Thiery JP , Radvanyi F , Barillot E. Analysis of array CGH data: from signal ratio to gain and loss of DNA regions . Bioinformatics . 2004 ; 20 ( 18 ): 3413 ± 22 . https://doi.org/10.1093/ bioinformatics/bth418 PMID: 15381628 .
32. La Rosa P , Viara E , Hupe P , Pierron G , Liva S , Neuvial P , et al. VAMP: visualization and analysis of array-CGH, transcriptome and other molecular profiles . Bioinformatics . 2006 ; 22 ( 17 ): 2066 ±73. https:// doi.org/10.1093/bioinformatics/btl359 PMID: 16820431 .
33. Coldman AJ , Fryer CJ , Elwood JM , Sonley MJ . Neuroblastoma: influence of age at diagnosis, stage, tumor site, and sex on prognosis . Cancer . 1980 ; 46 ( 8 ): 1896 ± 901 . Epub 1980/10/15. PMID: 7427894 .
34. Monclair T , Mosseri V , Cecchetto G , De Bernardi B , Michon J , Holmes K. Influence of image-defined risk factors on the outcome of patients with localised neuroblastoma. A report from the LNESG1 study of the European International Society of Paediatric Oncology Neuroblastoma Group . Pediatric blood & cancer . 2015 ; 62 ( 9 ): 1536 ± 42 . https://doi.org/10.1002/pbc.25460 PMID: 25663103 .
35. Simon T , Hero B , Benz-Bohm G , von Schweinitz D , Berthold F. Review of image defined risk factors in localized neuroblastoma patients: Results of the GPOH NB97 trial . Pediatric blood & cancer . 2008 ; 50 ( 5 ): 965 ±9. https://doi.org/10.1002/pbc.21343 PMID: 17914735 .
36. Yoo SY , Kim JS , Sung KW , Jeon TY , Choi JY , Moon SH , et al. The degree of tumor volume reduction during the early phase of induction chemotherapy is an independent prognostic factor in patients with high-risk neuroblastoma . Cancer . 2013 ; 119 ( 3 ): 656 ± 64 . Epub 2012/09/07. https://doi.org/10.1002/cncr. 27775 PMID: 22952047 .
37. De Bernardi B , Gambini C , Haupt R , Granata C , Rizzo A , Conte M , et al. Retrospective study of childhood ganglioneuroma . J Clin Oncol . 2008 ; 26 ( 10 ): 1710 ±6. https://doi.org/10.1200/JCO. 2006 . 08 .8799 PMID: 18375900 .
38. Gahr N , Darge K , Hahn G , Kreher BW , von Buiren M , Uhl M . Diffusion-weighted MRI for differentiation of neuroblastoma and ganglioneuroblastoma/ganglioneuroma . Eur J Radiol . 2011 ; 79 ( 3 ): 443 ± 6 . Epub 2010/05/14. https://doi.org/10.1016/j.ejrad. 2010 . 04 .005 PMID: 20462716 .
39. Bleeker G , van Eck-Smit BL , Zwinderman KH , Versteeg R , van Noesel MM , Kam BL , et al. MIBG scans in patients with stage 4 neuroblastoma reveal two metastatic patterns, one is associated with MYCN amplification and in MYCN-amplified tumours correlates with a better prognosis . European journal of nuclear medicine and molecular imaging . 2015 ; 42 ( 2 ): 222 ± 30 . Epub 2014/10/01. https://doi.org/10. 1007/s00259-014-2909-1 PMID: 25267348; PubMed Central PMCID : PMCPMC4315489 .