10TH ANNIVERSARY ARTICLE
Molecular Biology and Genetics
of Human Neuroblastoma
Garrett M. Brodeur and Chin-to Fong
Neuroblastoma, a tumor of postganglionic sympathetic neurons, is perhaps the most
fascinating and enigmatic of childhood neoplasms from both a clinical and a
biological viewpoint. Over 500 new cases are diagnosed in the United States each
year, making this the most common solid tumor in children . However, improve-
ments in cancer treatment over the past 25 years have had little impact on the
long-term survival of children with neuroblastoma. Despite its resistance to conven-
tional modalities of therapy, there are clues that neuroblastoma might be particularly
susceptible to innovative· approaches to treatment. These clues include the good
. prognosis of infants, even with disseminated disease (e.g., Stage IV-S), the propensity
of the tumor to occasionally undergo spontaneous regression in patients, and its
ability to undergo spontaneous or induced differentiation to a benign ganglioneu-
roma. Thus, a better understanding of this disease at the biological level may suggest
new and potentially more effective approaches to treatment.
A great deal of progress has been made in the past few years in advancing our
knowledge of human neuroblastoma at the cellular and molecular levels [2-5]. The
genetic predisposition to this disease is becoming clarified, a specific oncogene
amplified in neuroblastoma cells has prognostic significance, and the deletion of the
short arm or chromosome 1 has been more precisely defined. These and other recent
genetic observations have contributed to our understanding of tumor predisposition,
tumorigenesis, genetic heterogeneity, tumor progression, and prognosis.
In this review, the clinical and biological significance of the following major
topics will be addressed: 1) the genetics of human neuroblastoma-including
hereditary predisposition to neuroblastoma and constitutional cytogenetic abnormal-
ities in patients with this disease; and 2) cytogenetic and molecular abnormalities in
neuroblastoma cells-including an overview of cytogenetic abnormalities, a review
of N-myc amplification and other evidence of oncogene activation, and a discussion
of chromosome 1 p deletion and loss of heterozygosity. Finally, conclusions and some
future prospects will be presented.
From the Department of Pediatrics, Washington University School of Medicine, St. Louis, Missouri
Address reprint requests to: Garrett M. Brodeur, M.D., Washington University School of
Medicine, 400 South Kingshighway Boulevard, St. Louis, MO 63110.
Received January 30, 1989; accepted June 1, 1989.
© 1989 Elsevier Science Publishing Co., Inc.
655 Avenue of the Americas, New York, NY 10010
Cancer Genet Cytogenet 41:153-174 (1989)
G. M. Brodeur and C. Fang
GENETICS OF HUMAN NEUROBLASTOMA
Hereditary Predisposition to Neuroblastoma
A subset of patients exhibit a predisposition to develop neuroblastoma, and this
predisposition follows an autosomal dominant pattern of inheritance. Indeed,
Knudson and Strong  estimated that as many as 22% of all neuroblastomas could
be the result of a germinal mutation. Regression analysis of their data indicated that
neuroblastoma fits the two-mutation hypothesis proposed by Knudson for the origin
of childhood cancer . According to this hypothesis, the nonhereditary form of
neuroblastoma would result from two postzygotic (somatic) mutations in a single
cell, causing malignant transformation of the cell, which then develops into a
single tumor. Hereditary tumors would arise in individuals in whom the first mu-
tation is acquired as a prezygotic (germinal) event, so it is present in all cells.
Only one additional mutation in any cell of the target tissue would be needed
to induce malignant transformation, so thE;Jse individuals have a higher incidence
of neuroblastoma with a peak incidence at an earlier age. In addition, they may
develop tumors at multiple primary sites, either simultaneously or sequentially.
If such persons survive, half of their offspring could expect to be carriers of the
germinal mutation, with an estimated 63% chance of developing a neuroblas-
The nature of the two mutations was not specified by Knudson, but Comings 
suggested that these mutations inactivated two alleles of a specific gene on homol-
ogous chromosomes. If the two-mutation hypothesis is correct for neuroblastoma,
it has several important implications. First, the baseline incidence of this tumor
may remain fairly stable, reflecting a constant rate of spontaneous mutation. How-
ever, an increase in exposure to environmental mutagens would cause an increase
in both hereditary and nonhereditary forms. Second, all patients with a family
history of neuroblastoma, those with multifocal primary tumors and a subset of
the remaining patients, would be carriers ·of a germinal mutation. Obligate and
suspected carriers could be screened regularly during early childhood when the
risk of neuroblastoma is highest. The task of identifying predisposed individuals
would be greatly simplified if the chrompsomal location of the predisposition
locus or loci were identified and if informativ:e probes were available for genetic
There have been a number of reports of familial neuroblastoma, as well a.s bilateral
or multifocal disease. These are consistent with the existence of a hereditary
predisposition to neuroblastoma in some patients (9-13]. The topic of familial
neuroblastoma was reviewed recently by Kushner et al. (14]. The median age at
diagnosis of unselected patients with neuroblastoma is 30 months, with about 25% of
cases diagnosed within the first year of life. In contrast, the median age of patients
with familial neuroblastoma is 9 months, and 60% are diagnosed by 1 year of age. At
least 20% of patients with familial neuroblastoma have bilateral adrenal tumors or
multifocal primary tumors. The concordance or discordance for neuroblastoma in
monozygotic siblings was reviewed recently . This study suggests that hereditary
factors may be predominant in neuroblastoma diagnosed in infants, whereas random
environmental mutations may be more important in older children. There have been
a few reports of neuroblastoma associated with ,the fetfl.l hydantoin and fetal alcohol
syndromes [ 16-18]. While these reports suggest that prenatal exposute to hydantoins
or ethanol may increase the risk of neuroblastoma, these associations have not been
confirmed with certainty.
Biology and Genetics of Neuroblastoma
Constitutional Cytogenetic Abnormalities in Patients with Neuroblastoma
Retinoblastoma and Wilms' tumor are two common childhood neoplasms that fit the
two-mutation model for hereditary neoplasms. Each is associated with charactP!istic
congenital anomalies and specific constitutional chromosome abnormalities [19, 20].
The latter may represent an exaggerated form of the germinal mutation required for
oncogenesis in hereditary cases. However, no one has as yet identified a constitu-
tional cytogenetic predisposition syndrome or associated congenital anomaly with
predisposition to the development of neuroblastoma .
Neuroblastomas were reported in three patients with trisOmy D (or trisomy 13)
prior to the time banding analysis became routine [22, 23], but this finding has not
been reported since.Jndeed, these cases were newborns who died in the first week of
life and two were found to have only microscopic (in situ?) neuroblastoma at
autopsy. The association of trisomy 13 and neuroblastoma should be interpreted with
caution as microscopic foci of adrenal neuroblasts in a newborn might represent
residual elements of normal development of the fetal adrenal rhedulla [24, 25], rather
than representing in situ neuroblastoma .-
Table 1 summarizes the cases of constitutional chromosome abnormalities de-
tected by banding that have been reported in individuals with neuroblastoma, but no
consistent pattern is evident [27-34]. In one study by Moorhead and Evans , two
out of 37 patients with neuroblastoma whose constitutional f:_aryotypes were anal-
yzed by banding had balanced translocations. The other repOrts were of individual
cases with constitutional chromosome abnormalities. The case: originally reported by
Pegelow et al.  is particularly interesting. This child had two constitutional
cytogenetic abnormalities that were subsequently determined! to be inv(11)(q21q23)
and del(21)(p11), one inherited from each parent . Bothpiarents apparently had
had children with neuroblastoma by other partners. N e v e r t h e ~ ~ s s , the child with both
constitutional cytogenetic abnormalities was p h e n o t y p i c a l l ~ normal and did not
have obvious multifocal disease. Indeed, Hecht et al.  have ~ u b s e q u e n t l y reported
that neuroblastoma predisposition was not linked to the abnofmal chromosomes 11
or 21 in this family. Finally, Rudolph et al.  reported the occurrence of a fragile
site on chromosome 1 (1p13.1) in lymphocy:tes from nine ont of 20 patients with
neuroblastoma. Although this report is interesting with r e g a r ~ to the association of
abnormalities of chromosome 1p in n e u r o b l a ~ t o m a cells (see ,below), the region in-
volved in these patients is quite proximal to the site usually' deleted in the tumor
cells and unlikely to be related.
Neuroblastoma has also been associated with neurofibromatosis and aganglionosis
of the colon, suggesting that it might be an expression of neuroyristopathy [6, 35-40].
Table 1 Constitutional chromosome abnormalities in patients with
Partial trisomy (2p) and monosomy (16p)
Partial trisomy (3q) and monosomy (Bp)
Partial trisomy (15q) and monosomy (13q)
One from each parent
Hereditary fragile site
G. M. Brodeur and C. Fang
However, a recent analysis of the simultaneous occurrence of neuroblastoma and
neurofibromatosis in the same patient suggests that the coincidence can be accounted
for by chance alone . A variety of other congenital anomalies and genetic
syndromes has been reported in association with neuroblastoma. These include
tuberous sclerosis, Beckwith-Wiedemann syndrome, congenital heart disease, Frie-
dreich ataxia, dermatomyositis, nesidioblastosis, cystic fibrosis, and asymmetric
crying facies [41-48]. However, only one or a few patients with each association have
been reported. Several studies have shown a general increase in the incidence of
congenital anomalies in patients with neuroblastoma, but no specific congenital
abnormality occurred with increased frequency [49-51].
Other malignant diseases have been observed in individuals with neuroblastoma
following treatment, such as pheochromocytoma, renal cell carcinoma, astrocytoma,
and acute leukemia [52-56]. Interestingly, in a case of acute lymphoblastic leukemia,
the malignant karyotype had a t(4;11)(q21;q23) , and another case with acute
monoblastic leukemia had a t(9;11)(p21;q23) . The karyotypes seen in leukemias
that occur as second malignant diseases in patients with neuroblastoma are identical
to those seen in de novo leukemias. However, second cancers are not frequent in
patients with neuroblastoma, and no particular second cancer has occurred with
sufficient frequency to indicate a specific relationship with neuroblastoma .
CYTOGENETIC AND MOLECULAR ABNORMALITIES
IN NEUROBLASTOMA CELLS
Overview of Cytogenetic Abnormalities
Consistent cytogenetic abnormalities have been identified in a variety of human
leukemias and solid tumors . The most consistent types of changes are deletions,
translocations, extrachromosomal double minute chromatin bodies (dmin), and
chromosomally integrated homogeneously staining regions (HSRs). The latter two
abnormalities are cytogenetic manifestations of gene amplification. In some cases,
these chromosomal changes have suggested the location of genes that are preferen-
tially activated or inactivated. In addition, they may suggest mechanisms for altered
gene expression . To date, examples of amplification and of a characteristic
chromosome deletion have been identified in human neuroblastomas.
A substantial number of neuroblastomas have either dmin, HSRs, or both in
subpopulations of cells. The presence of dmin in neuroblastomas was first described
in 1965 by Cox et al. , and HSRs were identified in 1976 by Biedler et al. [60, 61].
Indeed, Biedler suggested that these abnormalities might be cytogenetic manifesta-
tions of gene amplification. The dmin are found in about one third of primary tumors
and about half the neuroblastoma cell lines . The HSRs are uncommon in pri-
mary tumors, but they are found in about half the cell lines analyzed (Table 2). Either
dmin or HSRs (or both in subsets of cells) occur in about 90% of neuroblastoma cell
Table 2 Selected chromosome abnormalities in near-diploid neuroblastomas
Biology and Genetics of Neuroblastoma
lines. Evidence suggests that there is selection in vitro for cell lines derived from
tumors that have preexisting dmin or HSRs, and there is no evidence to date that
these abnormalities develop with time in culture, at least in neuroblastomas.
Deletion of the short arm of chromosome 1 was first described in 1975 by Brodeur
et al. as the most characteristic cytogenetic abnormality in human neuroblastomas
(63, 64]. This finding has been confirmed by additional studies of chromosome
abnormalities in neuroblastomas and tumor-derived cell lines (62, 65-90], indicating
that 70-80% of neuroblastomas karyotyped have this abnormality (Ta,ble 2). We have
reviewed the published karyotypes of 60 n e a r ~ d i p l o i d neuroblastomas and cell lines
(Fig. 1). In this analysis, 50 out of 60 (83%) had a deletion or rearrangement of the
!Neuroblastomas: Numerical Changes!
: li.,__.-..,...,. ___ . , . G _ a ~ i n • s . . - . , . . , _ , , . . . . . . . , J . . . , . . . . , __ . , _ ~ ~
· · · · · · · ~ · · · · · · ·
~ .1 .••• 1...... ·•···
12 3 4 56 7 8 910111213141516171819202122X Y
!Neuroblastomas: Structural Changes I
1 2 3 4 56 7 8 910111213141516171819202122 X Y
~ : r · · · · · · - · · · · · ~ · · · - · · · - ·
Figure 1 A statistical analysis of modal karyotypes from 60 near-diploid human neuroblasto-
mas and tumor-derived cell lines is shown. The top panel gives the number of cases, with gain
or loss of the chromosomes indicated on the abscissa. No more than eight cases showed gain or
loss of any chromosome. The bottom panel shows an analysis of structural abnormalities of the
short or long arms of the chromosomes listed on the abscissa. While 50 of the 60 cases had an
abnormality of the short arm of chromosome 1, no more than ten cases had abnormalities of
other chromosome arms.
G. M. Brodeur and C. Fang
short arm of chromosome 1. This was the only cytogenetic abnormality (numerical or
structural) that occurred with statistically increased frequency (p < 0.001). Other
types of cytogenetic abnormalities occurred in several cases, such as rearrangements
of chromosome 17 , but no chromosome rearrangement other than the chromo-
some 1p abnormalities occurred with a frequency that reached statistical signifi-
Flow cytometric analysis of the DNA content of human neuroblastomas from
infants was first reported by Look et al. . In this analysis, abnormally high DNA
content was associated with lower stages of tumor and a response to initial
chemotherapy, whereas those with a "normal" DNA content were likely to have
advanced stages of tumor (especially stage D) and a poor response to chemotherapy.
This latter group of tumors likely had subtle genetic abnormalities that could not be
detected by flow cytometry, such as dmin or chromosome 1 deletions. Nevertheless,
subsequent studies have confirmed the prognostic importance of flow cytometric
measurements of DNA content [92-94]. This modality provides a complementary
approach to the genetic analysis of human neuroblastomas and in predicing outcome
and even response to therapy.
Dmin, HSRs and N-myc Amplification
Cytogenetic analysis of human neuroblastomas has revealed extrachromosomal dmin
and chromosomally integrated HSRs in about a third of primary tumors and about
90% of established cell lines . Although it has been known for over 10 years that
these abnormalities were cytogenetic manifestations of gene amplification, the nature
of the amplified sequences in neuroblastomas was not known until recently. Initially,
evidence for amplification of genes associated with drug resistance was sought, but
none was found.
A study was undertaken to determine if an oncogene was amplified in these
tumors. Schwab et al.  found that an oncogene related to the viral oncogene
v-myc, but distinct from the initially identified human homolog c-myc, was ampli-
fied in eight of nine neuroblastoma cell lines tested. This finding has been confirmed
independently in other laboratories [96, 97]. The amplified N-myc sequence was
mapped to the HSRs in neuroblastoma cell lines, and the normal single-copy locus
was mapped to the distal short arm of chromosome 2 [96, 98]. Thus, current evidence
suggests that amplification of a large region from this chromosome, which includes
the N-myc locus, becomes amplified initially as extrachromosomal dmin . In a
small percentage of primary tumors and about half of established neuroblastoma cell
lines, the amplified DNA occurs as an HSR that is linearly integrated into a
chromosome. It is unclear why HSRs are a more common form of amplified DNA in
established cell lines than they are in primary tumors.
We collaborated with Seeger and colleagues to determine if N-myc amplification
occurrecl. in primary tumors from untreated patients. In this analysis of 63 primary
tumors, amplification ranging from 3- to 300-fold per haploid genome was found in
24 tumot.'s (38%) . All cases with N-myc amplification in this initial study came
from patients with advanced stages of disease (stages III and IV by the Evans staging
system).iNext, the progression-free survival of 89 patients was analyzed according to
the stage of disease.and N-myc copy number [101, 102]. N-myc amplification was
clearly ~ s s o c i a t e d with rapid tumor progression and a poor outcome, independent of
the s t a g ~ of the tumor. Interestingly, in this later study two out of 16 patients
classified as stage II had amplification. Although patients with stage II in general have
a good grognosis, both stage II patients with N-myc amplification had rapid tumor
p r o g r e s s ~ o n , whereas only one of the remaining 14 patients without amplification
Biology and Genetics of Neuroblastoma
These studies were extended to over 600 patients with neuroblastoma enrolled in
protocols of the Children's Cancer Study Group (CCSG) and the Pediatric Oncology
Group (POG) . Examples of N-myc amplification seen in some of the primary
tumors are shown in Figure 2. In general, the same pattern seen in our earlier studies
has been borne out by the larger study (Table 3). Moreover, we demonstrated
unequivocally that about 5-10% of patients with stages of tumor traditionally
associated with a good prognosis (CCSG stages I, II and I V ~ S ; POG stages A, B and
D-S) have tumors with N-myc amplification . Our data indicate that these
patients are destined to have rapid tumor progression and a poor outcome, similar to
their counterparts with advanced stages of disease. Over 30% of patients with more
advanced tumor stages had N-myc amplification, and they also had an expectedly
poor outcome. Our findings that N-myc amplification is associated with a poor
outcome regardless of the clinical stage of tumor is supported by preliminary studies
and case reports from elsewhere in the United States, as well as Europe and Japan
A study was undertaken of N-rnyc copy number in multiple simultaneous or
consecutive samples of neuroblastoma tissue from 60 individual patients . This
study attempted to determine if N-myc copy number was heterogeneous in different
tumor samples from a given patient or if it changed with time in vivo along with
progression of the tumor. Indeed, a consistent pattern of N-myc copy number (either
amplified or unamplified) was found in different tumor samples taken from an
individual patient, either at the same point in time or fit different times during
treatment (e.g., at the time of diagnosis, at second-look surgery, at relapse, with
progressive disease, or at autopsy). These results suggest that N-myc amplification is
an intrinsic biological property of a subset of neuroblastomas . Tumors that
develop N-myc amplification generally do so by the time of !diagnosis, and so far no
cases of neuroblastoma with a single copy of N-rnyc at the! time of diagnosis have
developed amplification subsequently.
About 30% of the children with neuroblastoma have N-rnyc amplification in their
tumors, and virtually all of these children have rapidly prbgressive disease and a
poor prognosis. However, there is not yet a biological explin1-ation of why half of the
remaining patients with single-copy tumors do not survive. 4- general correlation has
been shown between N-myc copy number and expression' [106, 110-'--115], and the
Figure 2 Southern hybridization of N-myc to DNA from nine different neuroblastomas
obtained from untreated patients. Note increased hybridization intensity in samples 2, 4, 6, 7,
and 8, demonstrating amplification, compared to the normal copy number seen in samples 1, 3,
5, and 9.
Table 3 N-myc copy number in 662 POG and CCSG
patients with neuroblastoma
G. M. Brodeur and C. Fang
Stage at diagnosis Frequency of N-myc amplification
Low stages (A,B;I,II)
Stages IV -S or D-S
Advanced stages (C,D;III,IV)
level of expression per gene may be "activated" relative to N-myc expression in
nonneural tumors and tissues . It was shown that 20-25% of tumors overexpress
a single copy of N-myc at the RNA or protein level , and there appears to be an
inverse correlation between N-myc amplification or mRNA expression and features
of histologic differentiation [104, 112, 113]. The relationship between N-myc
expression and disease progression or outcome is still controversial [110-115], but
activation of N-myc by mechanisms other than amplification may play an important
role. In addition, it is possible that activation of other oncogenes, deletion of
chromosome 1p, or other genetic lesions may contribute to the poor clinical outcome
in these patients (see below).
Other Evidence of Oncogene Activation
We sought evidence for amplification of other oncogenes, including c-myc, L-myc,
c-N-ras, c-H-ras, c-K-ras, c-erbB1, c-erbB2, c-sis, c-src, c-myb, c-fos, and c-ets in over
100 neuroblastomas, but none was found . Similarly, we investigated expression at
the RNA level of the above oncogenes in ten neuroblastomas and a subset of primary
tumors, but marked overexpression was found only for N-myc in the cell lines and
tumors with amplification of this gene .
There is preferential expression of N-myc and c-ets in neuroblastomas compared
to a related tumor called peripheral neuroepithelioma (or primitive neuroectodermal
tumor) . The latter tumor expresses c-myc and does not express N-myc or c-ets
appreciably. However, both tumors express c-myb. An activated member of the ras
oncogene family, called N-ras, was first identified as the transforming gene of a
human neuroblastoma cell line [117, 118), but activation by specific mutations has
not been reported for this or other ras genes in other human neuroblastomas. There is
a recent report that a c-H-ras is expressed in neuroblastomas that are more differenti-
ated compared to their undifferentiated counterparts . Finally, there is evidence
for a posttranslational activation of the c-src gene, resulting in increased tyrosine
kinase activity in some human neuroblastoma cell lines , but primary tumor
tissue has not yet been ap.alyzed. In summary, there is no consistent evidence impli-
cating activation of an oncogene other than N-myc with poor outcome seen in a
subset of the patients lacking N-myc amplification.
Deletion and Loss of Heterozygosity of Chromosome 1p
Partial 1 p monosomy is found in 70-80% of primary tumors and cell lines that have
been karyotyped [62-68]. The region most commonly deleted is between 1p32 and
1pter (Fig. 3), although there are two cases that reported a proximal breakpoint at
1p34 (Fig. 4). This deletion is thought to represent the loss of a gene, the putative
neuroblastoma (Nb) suppressor gene [5, 6]. Loss or inactivation of one or both copies
of this gene may be an important step in the development of neuroblastoma.
Abnormalities of chromosome 1 have been reported in a variety of other malignant
diseases [121, 122]. However, in almost all other cases, the consistent finding is
Biology and Genetics of Neuroblastoma
cases, the normal chromosomes 1 are shown on the right and the deleted or rearranged
chromosomes 1 on the left. The NCG has a deletion of one homolog beyond band 1p13. The
NWC is near-tetraploid and has four chromosomes 1; there are two identical copies of a deleted
chromosome 1 beyond band 1p22. The NMB2 has a deletion of a homolog of chromosome 1
beyond 1p32, but additional material of uncertain origin is translocated to the breakpoint. The
NJF also appears to have a deletion of one homolog of chromosome 1 beyond 1p32 with three
additional bands translocated to the breakpoint. In all cases, the deleted material from
chromosome 1 is apparently deleted from the karyotype (modified from reference 62).
Partial karyotypes showing the chromosome 1 pairs of four neuroblastomas. In all
trisomy for the long arm of chromosome 1, not monosomy for the short arm. Trisomy
1q may be manifested by an extra intact chromosome 1, an extra chromosome 1 with
a deleted short arm, or an extra (1q). Indeed, trisomy 1q may be the most prevalent
cytogenetic change in all of neoplasia, but consistent monosomy for 1p is largely
restricted to neuroblastomas or other tumors of neuroectodermal origin.
It is possible that this genetic lesion is analogous to the 13q14 chromosome
deletion in retinoblastomas or the deletion of 11p13 in Wilms' tumors [19, 20].
Although constitutional deletions of the respective chromosomes have been iden-
tified in these two tumors, no chromosome deletion syndrome has been identified as
yet that predisposes to the development of neuroblastoma [5, 21]. Because there have
not been any descriptions of chromosome 1p deletion syndromes, it is likely that
deletions large enough to be visible cytogenetically are incompatible with life, due to
the expression of embryonic lethal mutations. Nevertheless, there are a number of
reports of familial neuroblastoma, suggesting that a gene inherited in an autosomal
dominant manner confers genetic predisposition to the development of this tumor
[9-15], as is seen in retinoblastoma and in Wilms' tumor.
Partial monosomy of chromosome 1p is the most consistent cytogenetic abnor-
mality found in human neuroblastomas, but its overall frequency is unclear. This is,
in part, because most of the karyotypes reported to date have come from patients with
advanced stages of disease or from established neuroblastoma cell lines. In addition,
cytogenetic analysis of primary tumor tissue is not always successful and may be
difficult to interpret. Therefore, we used a panel of chromosome-1-specific DNA
probes that identify restriction fragment length polymorphisms (RFLPs) along the
short and long arms of chromosome 1 to assess chromosome deletion or somatic loss
of heterozygosity (LOH) [103, 123, 124]. By comparing the pattern seen in normal
DNA of heterozygous patients with that obtained in tumdr DNA from the same
patient, we could assess LOH, which is the molecular e q ~ i v a l e n t of a deletion.
Examples of constitutional RFLP and LOH in four neuroblastomas are shown in
Actually, reduction at a given locus from two bands in constitutional DNA to one
band in the tumor can be accounted for by several possible mechanisms: 1) loss of
one entire homolog; 2) loss of one homolog with dupliqation of the other; 3)
2 12 15 10 4 3 2
G. M. Brodeur and C. Fang
mosome 1 showing deletions of the short arm
of chromosome 1 seen in 60 near-diploid
neuroblastomas. The brackets at the right of
the chromosome show the region of chromo-
some lp deleted from the karyotype, and the
number above the bracket indicates the num-
ber of cases in which the deletion was seen.
All of the above deletions resulted in loss of
the chromosome material bracketed (partial
lp monosomy]. Two other cases not shown
had rearrangements of chromosome 1p in
which there was no obvious loss of chromo-
some 1 material.
Schematic representation of chro-
chromosome deletion; and 4) mitotic recombination involving a portion of the
chromosome (Fig. 6). Both mitotic recombination and deletion would result in
retention of heterozygosity (by RFLP analysis) for most of the chromosome, with LOH
for the distal short arm probes. The two could be distinguished by identifying a gross
chromosome deletion (cytogenetic analysis) or by the relative intensity of the bands
compared to normal DNA (molecular analysis). Loss and duplication as well as
monosomy for chromosome 1 would result in LOH for all polymorphic probes along
the entire chromosome. The two could be distinguished by cytogenetic or molecular
analysis, as described above.
We studied pairs of human neuroblastoma DNA and constitutional DNA from 47
individual patients . Of the 47 tumor samples, there were 45 primary neuroblas-
tomas and two tumor-derived cell lines. Thirteen of the 47 tumors (28%) showed
LOH at one or more loci. Twelve of the 13 tumors revealed a pattern of LOH
consistent with terminal deletions of chromosome 1p, whereas the remaining tumor
may have had an interstitial deletion. Thus, the common region of LOH in these 13
cases lies at the distal end of the short arm of chromosome 1 from 1p36.1 to 1p36.3.
Figure 5 Southern hybridization with a hypervariable probe (pYNZ2 or D1S57) to normal
and tumor DNA after digestion with the restriction enzyme Taql. The first panel (from the left)
in the figure shows a case in which polymorphism was not seen in the constitutional DNA (C),
so the case is uninformative with respect to allelic loss or LOH in the tumor DNA (T). The
second panel shows a case in which polymorphism was seen in the constitutional DNA, so it
was informative, and no LOH was seen in the tumor. On the otherhand, the last two panels
show cases in which the constitutional DNA was informative and' LOH was detected in the
tumor, as demonstrated by the absence of the lower band in both c ~ s e s .
= ~ - =
Figure 6 Schematic representation of two normal chromosomes 1 :with the normal alleles at
the putative neuroblastoma locus (shown with a plus sign). Also shown are the mechanisms
detectable by cytogenetic and molecular analysis whereby a cell with a single mutation at this
locus (NB) can become homozygous or hemizygous for the mutated allele. Current evidence
suggests that LOH by the last two mechanisms rarely, if ever, occurs.
G. M. Brodeur and C. Fong
We postulate that loss or inactivation of a gene (or genes) at this site is critical for the
development or progression of neuroblastoma, and so the putative neuroblastoma
locus most likely lies within the boundaries defined by these two probes. Mutation in
the critical region on one chromosome, followed by deletion of the same region on
the homologous chromosomes (as manifested by LOH), may be an important
mechanism in the tumorigenesis of human neuroblastoma.
Cytogenetic evidence would suggest that deletion of chromosome 1 p occurs in
-70% of primary neuroblastomas, whereas our molecular studies demonstrated LOH
for distal chromosome 1p in only -30%. However, there are several possible
explanations for why LOH was not seen in a higher proportion of tumors. It is
possible that 1) most successful cytogenetic analyses are derived from advanced-
stage tumors with near-diploid karyotypes in which the incidence of large chromo-
some 1p deletions may be more common; 2) mutational events at the critical region in
some neuroblastomas may be too small to be detectable by LOH analyses with
currently available probes; 3) LOH for chromosome 1p may play a role in malignant
transformation in only a subset of neuroblastomas (such as those with advanced
stages of disease); or 4) LOH for chromosome 1 p may be a secondary event that occurs
in some tumors during the course of clonal evolution. Although we found LOH on
chromosome lp in an early stage neuroblastoma, the data showing LOH for 1p
predominantly in advanced stages of disease (see below) would argue for the latter
two explanations . Additional probes for the critical region should allow us to
distinguish definitively among these possibilities.
In addition to neuroblastoma, an increasing number of other tumors have in
common a loss of tumor DNA from a particular chromosome that is specific for each
type of tumor (Table 4) [125-173]. These genetic lesions are thought to represent the
loss of suppressor genes, whose absence may play an important role in malignant
transformation. In retinoblastoma, deletion of a specific gene on chromosome 13
thought to be critical in tumorigenesis has been identified, and its RNA and protein
product are being characterized [156-168]. Our results identify a critical region on
chromosome 1 p that may contain a recessive gene. The mutation or deletion of this
gene may be important in the transformation or progression of human neuroblastoma.
In some melanomas, medullary thyroid carcinomas (MCTs) and pheochromocyto-
mas, deletion or somatic loss of heterozygosity in the tumor tissue has been demon-
strated at loci on distal chromosome 1p [124, 125, 175]. Since all four neoplasms are
embryologically derived from neural crest cells, this suggests that there may be a
common mechanism underlying the formation or progression of these embryologi-
cally related tumors.
The other genetic lesion consistently associated with neuroblastoma is amplifi-
cation of the protooncogene N-myc [95-109]. The presence of N-myc amplification
has been shown to correlate strongly with advanced clinical stage and poor prognosis
in human neuroblastomas [100-109]. Our recent studies  showed a very strong
correlation between N-myc amplification and chromosome 1p LOH (p < 0.001),
indicating that LOH was common in patients with amplification (eight out of nine
patients, or 89%). Both N-myc amplification [100-109] and deletion of chromosome
1p as detected by cytogenetic analysis [85-90] appear strongly correlated with a poor
clinical outcome and with each other, but it is not yet clear if they are independent
prognostic variables. Nevertheless, they appear to characterize a genetically distinct
subset of very aggressive neuroblastomas .
A urinary catecholamine screening program has been underway in Japan for over a
decade to identify infants with neuroblastoma before they present clinically. Not
only does this promise to improve the prognosis of the infants detected, but it is also
providing important insights into the biology and evolution of human neuroblasto-
mas. Hayashi et al. [89, 90] have found that the majority of patients identified have
lower stages of disease, and virtually all of the tumors are in the hyperdiploid or
Biology and Genetics of Neuroblastoma
Table 4 Chromosome deletion or LOH associated with certain tumors
Tumors with deletion or LOH
of the indicated chromosomes
Medullary carcinoma of the thyroid
Small cell lung cancer
Non-small cell lung cancer
Renal cell carcinoma
triploid range. Earlier flow cytometric analyses of the DNA content of neuroblasto-
mas, as well as cytogenetic studies of modal chromosome number, had associated
hyperdiploid and triploid karyotypes with lower stages of disease and favorable
outcomes [85-88, 91-94]. The screening study suggested two possibilities: either 1)
all neuroblastomas begin as tumors with a more favorable genotype and phenotype
and some evolve into more aggressive tumors with adverse genetic features; or 2)
there are two different subsets of neuroblastoma, and the more favorable group
presents earlier and therefore is the predominant group detected.
Indeed, patterns are emerging, based on cytogenetic, molecular, and flow cytome-
tric analyses, that allow neuroblastomas to be assigned to two genetically distinct
groups [5, 90]. The first comprises those with hyperdiploid or triploid modal
karyotypes (or compatible DNA content by flow cytometry). Deletion of chromosome
1p, dmin, HSRs, or molecular evidence of N-myc amplification are rarely seen. These
patients are more likely to be infants with low stages of disease, [Stages 1, 2, or 4-5
by the International Neuroblastoma Staging System (174)], and they generally have a
very favorable prognosis. The second group consists of tumors that are near-diploid
or tetraploid modal chromosome number or DNA content. These tumors are more
likely to have chromosome 1p deletion, DMs, HSRs, or N-myc amplification. The
patients also are more likely to be over 1 year of age and have advanced stages of
disease (Stages 3 or 4). It remains to be determined if tumors in the favorable group
ever evolve or "progress" into the unfavorable group, but current evidence would
suggest that they are genetically distinct.
G. M. Brodeur and C. Fong
A subset of patients with neuroblastoma exhibit a predisposition to develop this
tumor, and this predisposition appears to follow an autosomal dominant pattern of
inheritance. Indeed, as many as 22% of neuroblastomas may result from germinal
mutations. However, a constitutional chromosome abnormality or congenital anom-
aly syndrome has not yet been identified that predisposes to neuroblastoma. Current
evidence does not support an increased risk of second malignant neoplasms in
patients with neuroblastoma, apart from the potential oncogenesis associated with
chemotherapy and radiation therapy.
In neuroblastoma cells, at least two genetic events have been identified in the
course of tumor evolution. These are loss of a critical region on the short arm of
chromosome 1, and activation (usually by amplification) of the N-myc protoon-
cogene. Our studies suggest that the two genetic events may be related, and that LOH
for chromosome 1 p may precede the development of amplification. Indeed, these two
genetic lesions may characterize a genetically distinct and aggressive subset of
neuroblastomas in the context of a near-diploid or near-tetraploid modal karyotype
(or DNA content). They generally present as unresectable or metastatic tumors in
older patients and are associated with a very poor prognosis.
The molecular and cytogenetic analysis of human neuroblastomas promises to
make available a great deal of information that would be difficult to obtain otherwise.
First, the localization of the neuroblastoma predisposition locus would make it
possible to distinguish between hereditary and sporadic cases so that family
counseling could be provided. Prenatal diagnosis of affected individuals in informa-
tive families would also be possible. Second, genetic markers should provide more
objective classification of tumors that may appear similar histologically. Third,
genetic analysis by karyotype, flow cytometry, and determination of N-myc copy
number provides information that has prognostic significance and can more appro-
priately direct the choice of treatment. Finally, genetic analysis tumor-specific
rearrangements may permit the identification of specific genetic lesions on which
future therapeutic approaches may be focused.
This work was supported in part by National Institutes of Health grants ROl-CA-39771 (GMB)
and K04-CA-01027 (GMB), the National Cancer Center [CTF), The American Cancer Society
Institutional Grant IN-36-29-4 (CTF), the Children's United Research Effort, and the Fern
Waldman Memorial Fund for Research in Childhood Cancer.
1. Young JL Jr, Ries LG, Silverberg E, Horm JW, Miller RW (1986): Cancer incidence, survival
and mortality for children younger than 15 years. Cancer 58:598-602.
2. Evans AE (ed.) (1980): Advances in Neuroblastoma Research. Progr Cancer Res Ther
3. Evans AE, D'Angio GJ, Seeger RC (eds.) (1985): Advances in Neuroblastoma Research.
Progr Clin Bioi Res 175:1-605.
4. Evans AE, D'Angio GJ, Knudson AG Seeger RC (eds.) (1988): Advances in Neuroblastoma
Research 2. Progr Clin Bioi Res 175:1-748.
5. Brodeur GM, Fong CT ( 1989): Molecular biology and genetics of human neuroblastoma. In:
Pochedly C and Tebbi C: Neuroblastoma: Tumor Biology and Therapy, C Pochedly, C.
Tebbi, eds. CRC Press, New York (in press).
6. Knudson AG Jr, Strong LC (1972): Mutation and cancer: Neuroblastoma and pheochromo-
cytoma. Am J Hum Genet 24:514-522.
Biology and Genetics of Neuroblastoma Download full-text
7. Knudson AG (1971): Mutation and cancer: Statistical study of retinoblastoma. Proc Natl
Acad Sci USA 68:820-823.
8. Comings DE (1973): A general theory of carcinogenesis. Proc Natl Acad Sci USA
9. Chatten J, Voorhess ML (1967): Familial neuroblastoma: Report of a kindred with multiple
disorders, including neuroblastomas in four siblings. N Eng! J Med 227:1230-1236.
10. Wong KY, Hanenson IB, Lampkin BC (1971): Familial neuroblastoma. Am J Dis Child
11. Hardy PC, Nesbit ME Jr (1972): Familial neuroblastoma: Report of a kindred with a high
incidence of infantile tumors. J Pediatr 80:74-77.
12. Arenson EB Jr, Hutter JJ, Restuccia RD, Holton CP (1976): Neuroblastoma in a father and
son. JAMA 235:727-729.
13. Wagget J, Aherne G, Aherne W (1973): Familial neuroblastoma: Report of two sib pairs.
Arch Dis Child 48:63-66.
14. Kushner BH, Gilbert F, Helson L: Familial neuroblastoma (1986): Case reports, literature
review, and etiologic considerations. Cancer 57:1887-1893.
15. Kushner BH, Helson L (1985): Monozygotic siblings discordant for neuroblastoma:
Etiologic implications. J Pediatr 107:405-409.
16. Seeler RA, Israel JN, Royal JE, Kaye CI, Rao S, Abulaban M (1979): Ganglioneuroblastoma
and fetal hydantoin-alcohol syndromes. Pediatr 63:524-527.
17. Allen Rw, Ogden B, Bentley FL, Jung AL (1980): Fetal hydantoin syndrome, neuroblas-
toma, and hemorrhagic disease in a neonate. 1 Amer Med Assoc 244:1464-1465.
18. Kinney H, Faix R, Brazy J (1980): The fetal alcohol syndrome and neuroblastoma.
19. Vogel F (1979): Genetics of retinoblastoma. Hum Genet 52:1-54.
20. Matsunaga E (1981): Genetics of Wilms' tumor. Hum Genet 57:231-246.
21. Brodeur GM (1980): Genetics and cytogenetics of human neuroblastoma. In: Neuroblas-
toma: Clinical and Biological Manifestations, C Pochedly, ed. Elsevier, New York, pp.
22. Feingold M, Gheradi GJ, Simmons C (1971): Familial neuroblastoma and trisomy 13. AmJ
Dis Child 121:451.
23. Nevin NC, Dodge JA, Allen IV (1972): Two cases of trisomy D associated with adrenal
tumors. J Med Genet 9:119-123.
24. Turkel SB, Itabashi HH (1975): The natural history of neuroblastic cells in the fetal adrenal
gland. Am J Pathol 76:225-243.
25. Ikeda Y, Lister J, Bouton JM, Buyukpamukcu M (1981): Congenital neuroblastoma,
neuroblastoma in situ, and the normal fetal development of th!'l adrenal. J Pediatr Surg
26. Beckwith J, PerrinE (1963): In situ neuroblastomas: A contribution to the natural history
of neural crest tumors. Am J Pathol 43:1089-1104.
27. Pegelow CH, Ebbin AJ, Powars D, Towner JW (1975): Familial neuroblastoma. J Pediatr
28. Hecht F, Kaiser-McCaw B (1981): Chromosomes in familial neuroblastoma. J Pediatr
29. Hecht F, Hecht BK, Northrup JC, Trachtenberg N, Wood ST, Cohen JD (1982): Genetics of
familial neuroblastoma: Long-range studies. Cancer Genet Cytogenet 7:227-230.
30. Moorhead PS, Evans AE (1980): Chromosomal findings in patients with neuroblastoma.
Progr Cancer Res Ther 12:109-118.
31. Nagano H, Kano Y, Kobuchi S, Kajitani T (1980): A case of partial 2p trisomy with
neuroblastoma. Jpn J Human Genet 25:39-45.
32. Robinson MG, McCorquodale MM (1981): Trisomy 18 and neurogenic neoplasia. J Pediatr
33. Sanger WG, Howe J, Fordyce R, Purtilo DT (1984): Inherited partial trisomy #15
complicated by neuroblastoma. Cancer Genet Cytogenet 1 1 : 1 5 3 ~ 1 5 9 .
34. Rudolph B, Harbott J, Lampert F (1988): Fragile sites and neuroblastoma: Fragile site at