Journal of Pediatric Surgery 51 (2016) 16651669

Contents lists available at ScienceDirect

Journal of Pediatric Surgery

journal homepage: www.elsevier.com/locate/jpedsurg

The optimal timing of surgical resection in high-risk neuroblastoma

Yesenia Rojas a, Sergio Jaramillo b, Karen Lyons c, Nadia Mahmood c, Meng-Fen Wu d, Hao Liu d,
Sanjeev A. Vasudevan a, R. Paul Guillerman c, Chrystal U. Louis e, Heidi V. Russell e,
Jed G. Nuchtern a,1, Eugene S. Kim a,f,,1
a
Division of Pediatric Surgery, Michael E. DeBakey Department of Surgery, Texas Children's Hospital, Baylor College of Medicine, Houston, TX
b
Department of Radiation Oncology, Baylor College of Medicine, Houston, TX
c
Department of Radiology, Texas Children's Hospital, Baylor College of Medicine, Houston, TX
d
Division of Biostatistics, Dan L. Duncan Cancer Center, Baylor College of Medicine, Houston, TX
e
Section of Hematology-Oncology, Department of Pediatrics, Texas Children's Hospital, Baylor College of Medicine, Houston, TX
f
Division of Pediatric Surgery, Department of Surgery, Children's Hospital Los Angeles, Keck School of Medicine, University of Southern California, Los Angeles, CA

a r t i c l e i n f o a b s t r a c t

Article history: Background: While most high-risk neuroblastoma (HRNB) patients are enrolled in cooperative group or
Received 17 January 2016 institutional protocols, variability exists within these protocols as to when surgical resection of the primary
Received in revised form 20 May 2016 tumor should be performed after neoadjuvant induction chemotherapy. We sought to determine if the number
Accepted 30 May 2016 of chemotherapy cycles prior to surgery affects surgical or survival outcomes in HRNB patients.
Methods: We performed a retrospective review of all HRNB patients b 18 years of age from 2000 to 2010, at
Key words:
Texas Children's Hospital. Patients were stratied based on the number of neoadjuvant induction chemotherapy
High-risk neuroblastoma
Induction chemotherapy
cycles prior to surgical resection. Pre and post- chemotherapy tumor size, MYCN status, iodine-131-
Surgical resection metaiodobenzylguanidine (MIBG) score at diagnosis, extent of surgical resection, estimated surgical blood loss,
post-operative outcomes, and event free (EFS) and overall survival (OS) were evaluated. Data were analyzed
using Wilcoxon rank-sum test, KruskalWallis test, Fisher's exact test, KaplanMeier analyses, and Cox regres-
sion analyses. P-value b 0.05 was considered signicant.
Results: Data from 50 patients with HRNB were analyzed. Patients were stratied by the number of cycles of
chemotherapy received prior to surgery. Six patients received 2 cycles of chemotherapy (12%), 20 patients re-
ceived 3 cycles (40%), 13 patients received 4 cycles (26%), and 11 patients received 5 cycles (22%) prior to surgical
resection of the primary tumor. The 5-year OS was 33%, 45%, 83% and 36% in patients who received 2, 3, 4 and
5 cycles of chemotherapy prior to surgery, respectively (p = 0.07). Multivariate analysis revealed that patients
who received 4 cycles of chemotherapy had a signicantly lower mortality (HR: 0.11, 95% CI: 0.010.87, p =
0.04) compared to those with 2 cycles of chemotherapy. Among the different cohorts, there were no differences
with respect to MYCN status, MIBG score at diagnosis, incidence of bone marrow metastasis, extent of surgical
resection, estimated blood loss, incidence of post-operative complications, or length of stay.
Conclusion: HRNB patients who receive 4 cycles of chemotherapy prior to surgical resection have a superior OS
than patients who receive 2. Based on the superior survival of patients who received 4 cycles of chemotherapy
prior to surgery, further studies are warranted to elucidate these differences.
2016 Elsevier Inc. All rights reserved.

Neuroblastoma is the most common extracranial solid tumor in chil-
dren, arising from neural crest cells that differentiate to the sympathetic
ganglia and adrenal medulla [1]. It accounts for 7% of all malignancies in
children and 1015% of cancer-related deaths in childhood [1,2]. NB is a
heterogeneous disease, and its clinical course can range from spontane-
ous regression to very aggressive, metastatic and progressive disease.
Patients with high-risk neuroblastoma (HRNB) are at an increased risk
Corresponding author at: Children's Hospital Los Angeles, 4650 Sunset Blvd, MS# 100,
Los Angeles, CA 90027. Tel.: +1 323 361 8332; fax: +1 323 361 3534.
E-mail addresses: eugeneskim@chla.usc.edu, ek49@yahoo.com (E.S. Kim).
1
Denotes co-senior authors.
for rapid progression and relapse [2]. Despite multimodal therapy, 50%
of patients with newly diagnosed HRNB, and less than 10% of patients
with relapsed HRNB, will survive [1].
Current therapy for HRNB consists of some combination of intensive
multi-agent induction chemotherapy, surgery, radiation, myeloablative
consolidation therapy with stem cell rescue and transplantation, 13-cis
retinoic acid and immunotherapy [14]. The role of surgical resection of
the primary tumor in HRNB is controversial with conicting studies in
the literature regarding the benet of aggressive surgery in this group
[5,6]. Currently, the Children's Oncology Group (COG) recommends ag-
gressive surgical resection of the primary tumor with the goal of gross
total resection [2,7]. This current recommendation is supported by the

http://dx.doi.org/10.1016/j.jpedsurg.2016.05.021
0022-3468/ 2016 Elsevier Inc. All rights reserved.
1666 Y. Rojas et al. / Journal of Pediatric Surgery 51 (2016) 16651669
recent results of two studies from the COG and the International Society
of Pediatric Oncology Europe Neuroblastoma (SIPOEN), which demon-
strated benet and improved survival with gross total resection [7,8].
However, unlike the treatment of other pediatric solid tumors, such as
rhabdomyosarcoma and Ewing's sarcoma, which clearly dene when
surgery and radiation should occur, the optimal timing for surgical re-
section in HRNB has not been well dened, and thus the decision is
left to the discretion of individual oncologists and surgeons. At different
institutions, some patients may undergo surgery after the last cycle of
induction chemotherapy, while others advocate for an earlier surgical
intervention, based on when the tumor appears resectable on imaging
[9]. The purpose of this study was to determine if the number of chemo-
therapy cycles prior to surgical resection affects surgical or survival out-
comes in HRNB patients. Our hypothesis is that patients treated with
less neoadjuvant chemotherapy would have larger pre-resection tu-
mors, more difcult resections and thus higher post-operative compli-
cation rates and worse overall survival.
1. Methods
1.1. Patients and study design
This is a retrospective study of all patients 18 years of age diag-
nosed with HRNB at our institution from 2000 to 2010. Patients were in-
cluded if they were 18 years of age, had a diagnosis of HRNB based on
the Children's Oncology Group (COG) risk stratication and underwent
surgical resection of their primary tumor.
Our study focused on HRNB patients who presented with
unresectable primary tumors at diagnosis. Patients (n = 10) with re-
sectable tumors who did not receive neoadjuvant chemotherapy prior
to surgery were excluded. The upfront resection patients presented
with smaller tumors (mean 5.0 cm greatest diameter) at the time of di-
agnosis and were resected prior to receiving any chemotherapy. In addi-
tion, patients with no primary tumor (n = 2) and patients who
underwent surgical resection of the primary tumor at another facility
(n = 9) were excluded from this analysis. Moreover, a patient with a si-
multaneous diagnosis of Ewing's sarcoma and HRNB, a patient who had
no chemotherapy treatment because of pre-existing cardiomyopathy,
and a patient with incomplete medical records were excluded.
1.2. Data collection
All clinical data were obtained from electronic and paper medical re-
cords. Collected variables included patient demographics, age at the
time of diagnosis, stage based on the International Neuroblastoma Stag-
ing System (INSS) [10], International Neuroblastoma Risk Group (INRG)
stage [11,12], MYCN status, Shimada histology, bone marrow status
at diagnosis, MIBG score [13] at the time of diagnosis, induction
chemotherapy protocol, number of chemotherapy cycles prior to surgi-
cal resection of primary tumor, tumor volume reduction, extent of re-
section, estimated blood loss (EBL), intraoperative and post-operative
complications, length of hospital stay, and status at the end of follow-
up. Minor post-operative complications included post-operative fever,
Horner's syndrome, hypovolemia, and ileus. Major post-operative com-
plications included chylous ascites or chylothorax, post-operative hem-
orrhage (dened as loss of N30% blood volume), bacteremia, adrenal
insufciency and death.
1.3. Treatment protocols and surgery
All patients in this study were treated with a multidisciplinary
approach that consisted of high-dose chemotherapy, surgery, radiother-
apy, autologous stem cell transplantation, and 13-cis retinoic acid.
Two patients from this cohort were treated on the COG ANBL0032
study evaluating chimeric 14.18 immunotherapy with 13-cis retinoic
acid. The time period of this study spanned 2 different induction
chemotherapy regimens used at our institution: PEPI (an institutional
review board approved [IRB] Phase II institutional research protocol)
and COG ANBL00P1. The treatment protocol offered to each patient
was dependent on which protocol was open or the institutional
standard at the time of patient presentation. The PEPI induction
chemotherapy protocol used a combination of cisplatin, etoposide, cy-
clophosphamide, and doxorubicin. ANBL00P1 consisted of vincristine,
ifosfamide, and carboplatin, in addition to those used in the PEPI
protocol. Patients were placed on an individualized protocol with dose
reduction of either PEPI or ANBL00P1 chemotherapy agents if they dem-
onstrated renal or organ dysfunction and/or drug toxicity. All induction
protocols in this study were a total of 5 cycles. Surgical resection of the
primary tumor was performed by one of two experienced pediatric sur-
geons at our institution. The goal of surgery for all cases was complete
resection, dened as removal of N90% of gross total tumor volume.
The extent of tumor resection was determined by the surgeon's opera-
tive report. The decision to operate was made by multi-disciplinary con-
sensus based on imaging and efcacy of chemotherapy, such that
patients who did not appear to have signicant tumor reduction with
additional chemotherapy cycles underwent surgical resection. Surgical
resection after cycle 25 was allowed for patients enrolled on PEPI,
after cycle 4 or 5 if patients were enrolled on ANLB00P1 and at the dis-
cretion of the multidisciplinary team for all patients not enrolled on a
clinical trial.
1.4. Metaiodobenzylguanidine (MIBG) scoring system
The scoring method for MIBG scans performed at the time of diagno-
sis was based on criteria outlined by Messina et al. [13]. The patient's
skeleton was divided into 9 segments to assess bone lesions and a
10th segment to assess soft tissue involvement [13]. Extension and in-
tensity scores were assigned to each segment to quantify the extent
and degree of MIBG uptake. Extension score was graded as follows: 0,
no sites per segment; 1, one site per segment; 2, more than one site
per segment; and 3, diffuse involvement (N50% of the segment). Inten-
sity score was graded as follows: 0, no uptake; 1, doubtful uptake; 2, ob-
vious uptake; and 3, strong uptake. The maximum extension and
intensity scores are 30 and 30, respectively. All MIBG scans were
reviewed and scored by a pediatric radiologist (N.M.) blinded to the
patient's outcome.
1.5. Tumor volumes
To determine the extent of tumor reduction with chemotherapy,
computed tomography (CT) scans at the time of diagnosis and prior to
surgery were reviewed by a pediatric radiologist (K.L.) blinded to the
number of chemotherapy cycles of each patient at the time of the
scan. Tumor volume (cm3) was calculated using the following formula:
maximal orthogonal tumor length width height 0.52 (/6). Tumor
volume reduction was calculated as the difference between tumor vol-
ume at the time of diagnosis and prior to surgery.
1.6. Statistical analyses
Patients were stratied by the number of chemotherapy cycles re-
ceived prior to surgical resection. As the HRNB patients represented
unresectable tumors at the time of diagnosis, and since none of the pa-
tients underwent one cycle of chemotherapy prior to surgical resection,
the 2 cycle group serves as our minimum chemotherapy control group,
and cohorts of patients who received additional cycles of chemotherapy
were compared to this group. Data were summarized and compared be-
tween groups using non-parametric Wilcoxon rank-sum test or
KruskalWallis test for continuous variables and Fisher's exact test for
categorical variables. KaplanMeier analyses were used to estimate
overall survival (OS) and event free survival (EFS). Cox regression anal-
yses were performed to determine the association between the number
 Y. Rojas et al. / Journal of Pediatric Surgery 51 (2016) 16651669 1667

of chemotherapy cycles prior to surgical resection and survival
outcomes after adjusting for other covariates. A P-value b 0.05 was con-
sidered signicant. Analyses were conducted using SAS for Windows
(SAS Institute, Cary, North Carolina).
1.7. Ethics and policy
This study was approved by the IRB at Baylor College of Medicine (H-
30850).
2. Results
2.1. Patient demographics and clinical presentation
Fifty patients who met inclusion criteria were treated at our institu-
tion between 2000 and 2010. The median age of diagnosis was 2 years
(range 5 months9 years); 54% (n = 27) were male and 46% (n =
23) were female. The majority of the patients (84%) were INSS stage 4
and the remaining patients were stage 3 (16%). All of the patients had
unresectable tumors at the time of diagnosis with the majority of pa-
tients presenting with an adrenal primary tumor (n = 40, 80%), follow-
ed by abdominal non-adrenal primary tumor (n = 7, 14%), and thoracic
primary tumor (n = 3, 6%). At presentation, 33 (66%) had metastatic
disease with bone marrow inltration. Bone marrow metastasis was
negative in 15 (30%) patients and unknown in 2 (4%). Using uorescent
in situ hybridization (FISH), MYCN amplication was present in 32
(64%) patients, unknown in 2 cases (4%), and non-amplied in 16 (32%).
Surgical resection of the primary tumor was performed in 6 patients
after cycle 2 (12%), 20 after cycle 3 (40%), 13 after cycle 4 (26%), and 11
after cycle 5 (22%). Tumor characteristics and treatment regimens
across the different chemotherapy cycle groups are summarized in
Table 1. The various cohorts were largely equivalent and did not differ
signicantly with respect to MYCN status and MIBG score at the time
of diagnosis, and interestingly, did not differ with regards to extent of
tumor resection, post-operative complications, and length of stay. As
anticipated, we observed greater tumor volume reduction with
additional cycles of chemotherapy. Response to chemotherapy was
the greatest after 2 cycles, which was the minimum number of chemo-
therapy cycles patients received in our study. Tumor volume reduction
after 2 cycles of chemotherapy was 77% from the time of diagnosis. After
2 cycles, tumor reduction was increased by 23% more with each
succeeding treatment cycle, however this was not statistically signi-
cant (p = 0.27).
There was a signicant difference in the type of induction chemo-
therapy protocol across the groups with more patients who had surgery
after cycles 2 and 3 being treated with the institutional PEPI protocol
and more patients who underwent surgical resection after cycles 4
and 5 on the COG ANBL00P1 protocol (p = 0.01). Two patients treated
on individualized protocols were in the 5-cycle group, as compared to
none in the other groups. Patients who underwent surgical resection
after 2 cycles of chemotherapy had the greatest median blood loss
when compared to the other groups.
2.2. Survival analyses
The 5-year OS for our cohort, all cycles, of high-risk neuroblastoma
patients at our institution was 40%. When stratied by number of cycles
of chemotherapy, the 5-year OS for patients who received 2, 3, 4, and
5 cycles of chemotherapy prior to surgical resection was 33%, 45%,
83%, and 36%, respectively (p = 0.07) (Fig. 1).
By univariate analysis (Table 2), patients treated with 4 cycles of
chemotherapy prior to surgical resection had a signicantly superior
OS compared to patients who received 2 cycles (OS HR: 0.19, 95% CI:
0.040.85, p = 0.03). The EFS was also superior but not signicant
(EFS HR: 0.38, 95% CI: 0.101.41, p = 0.15). Patients who had greater
tumor volume reduction by chemotherapy had both superior OS (HR:
0.98, 95% CI: 0.971.0, p = 0.01) and EFS (HR: 0.98, 95% CI: 0.970.99,
p = 0.001). Pre-operative tumor volume was found to be signicantly
associated with OS (HR: 1.56, 95% CI: 1.062.30, p = 0.02) and EFS
(HR: 1.81, 95% CI: 1.252.61, p = 0.002) with larger tumor volume
portending a worse outcome. The negative prognostic indicator of
MYCN amplication has been well documented in neuroblastoma [1],

Table 1
Tumor characteristics and treatment regimens across different chemotherapy cycle groups, N = 50.

# Chemotherapy cycles prior to surgery P-value

2 3 4 5
n = 6 (%) n = 20 (%) n = 13 (%) n = 11 (%)

MYCN status
Negative 3(50) 5(25) 5(41.7) 3(30) 0.63
Positive 3(50) 15(75) 7(58.3) 7(70)

MIBG score at diagnosis

Extension 17.5 (325) 12.5 (126) 13 (018) 4 (212) 0.21
Intensity 19.5 (525) 16 (324) 14.5 (022) 9 (517) 0.46

Protocol type
ANBL00P1 1 (16.7) 17 (85) 10 (76.9) 6 (54.5) 0.01
PEPI 5 (83.3) 3 (15) 3 (23.1) 3 (27.3)
Individual 0 (0) 0 (0) 0 (0) 2 (18.2)
Tumor volume% reduction from diagnosis 77.3 (22.182.2) 79.7 (20.696.6) 83.2 (48.298.4) 85.6 (097.5) 0.27

Extent of resection
N90% 6 (100) 18 (90) 11 (84.6) 7 (100) 0.73
5090% 0 (0) 2 (10) 2 (15.4) 0 (0)
b50% 0 (0) 0 (0) 0 (0) 0 (0)
EBL (cc) 500 (301500) 185 (102000) 200 (30825) 350 (200650) 0.53

Post-op complication
Minor 0 (0) 2 (33.3) 4 (80) 2 (50) 0.12
Major 4 (100) 4 (66.7) 1 (20) 2 (50)

Length of stay
(days) 11.5 (623) 6 (416) 7 (419) 7 (213) 0.16

MIBG - metaiodobenzylguanidine, EBL - estimated blood loss.

KruskalWallis test for continuous variables and Fisher's exact for categorical variables

Negative values indicate tumor increase
1668 Y. Rojas et al. / Journal of Pediatric Surgery 51 (2016) 16651669

1 younger patients having a higher risk of death. A trend towards im-

Overall survival probability
0.8
0.6
0.4
Chemo Cycle
2
0.2 3
4
5
0
0 1 2 3
Year
Fig. 1. 5-year overall survival (OS) of high-risk NB patients based on the number of chemo-
therapy cycles received prior to surgical resection.
and patients with MYCN amplication in our study had a trend towards
higher mortality compared to patients with no MYCN amplication
(HR: 2.32, 95% CI:0.866.28, p = 0.10). Moreover, similar to recent re-
ports of improved survival in patients who undergo complete gross
total tumor resection, patients in our study who underwent incomplete
tumor resection (5090%) had a trend towards increased risk of death
compared to patients who underwent complete gross total tumor resec-
tion (N 90%), though not reaching statistical signicance (p = 0.58). In
addition, we did not observe a statistically signicant difference in sur-
vival with regards to age, post-operative complications, length of hospi-
tal stay, or based on the chemotherapy protocol received.
By multivariate analysis, patients who received 4 cycles of chemo-
therapy had a signicantly improved OS compared to patients who re-
ceived 2 cycles after adjusting for other variables (OS HR: 0.11, 95% CI:
0.010.87, p = 0.04). The EFS was also superior but not signicant
(EFS HR: 0.5, 95% CI: 0.112.34, p = 0.38)(Table 3). Patients with
MYCN amplied tumors had a signicantly worse OS (HR: 3.6, 95% CI:
1.2410.46, p = 0.02) and a trend towards worse EFS (HR: 2.53, 95%
CI: 0.956.76, p = 0.06) when compared to those with non-MYCN
amplied tumors. Age was found to be related to OS (p = 0.06) with
Table 2
Univariate analyses for overall survival (OS) and event free survival (EFS).
Overall survival (OS) Event free survival (EFS)
HR 95% CI P HR
P= 0.07
proved OS was observed for patients with greater chemotherapy-
induced tumor volume reduction (p = 0.09). Preoperative tumor size
was not signicantly associated with OS but was signicantly associated
with EFS (HR: 1.79, 95% CI: 1.023.12, p = 0.04). Greater extent of sur-
gical resection was not found to be associated with OS and EFS by mul-
tivariate analysis.
3. Discussion
4 5
The majority of patients with HRNB present with metastatic disease
and/or large, inoperable tumors, and thus the goal of induction chemo-
therapy is to decrease the size of the tumor to facilitate denitive, gross
total resection. According to the Goldie-Coldman hypothesis,
cytoreductive therapy may decrease the likelihood of developing resis-
tant clones, and hence increase the chance of curability [14]. With this
goal in mind, previous centers have shown that surgical resection of
the tumor is feasible after the 2nd cycle of induction chemotherapy;
however, the majority of patients with HRNB undergo surgery after
the 5th cycle [9].
Denitive surgery of large tumors in the early cycles of induction
chemotherapy has been shown to be feasible and yield gross total resec-
tion as the greatest tumor reduction occurs after the 2nd or 3rd cycle of
chemotherapy but it remains unknown if this earlier resection results in
improved survival or increased operative complexity [15]. Based on our
cohort of patients, we found no signicant differences in the extent of
tumor resection across the 4 chemotherapy groups. From our data, in-
creased tumor reduction with additional cycles of chemotherapy did
not necessarily translate into an increased likelihood of gross total re-
section. Additionally, patients who received only 2 or 3 cycles of chemo-
therapy were just as likely to have N 90% tumor resection as those
treated with 4 or 5 cycles. However, patients who underwent resection
after 2 cycles of chemotherapy were more likely to have increased blood
loss during surgery, which is likely related to the larger tumor size.
Moreover, the length of post-operative hospital stay did not signicant-
ly differ among the groups.
To address our hypothesis, patients who underwent early surgical
resection following 2 cycles of chemotherapy did not result in superior
survival but rather had the worst OS when compared to patients who
received 3, 4, and 5 cycles. Interestingly, patients with the best OS
(83%) were those who underwent surgery after 4 cycles of chemother-
apy. One may postulate that the patients resected after 4 cycles either
95% CI P had biologically different disease and/or a different disease burden at

Chemotherapy cycle
2 1 1
Table 3
3 0.51 0.161.64 0.25 0.77 0.252.36 0.64
Multivariate analyses for overall survival (OS) and event free survival (EFS).
4 0.19 0.040.85 0.03 0.38 0.101.41 0.15
5 0.85 0.032.88 0.79 0.80 0.242.66 0.71 Overall survival (OS) Event free survival
(EFS)
MYCN status
Negative 1 1 HR 95% CI P HR 95% CI P
Positive 2.32 0.866.28 0.09 2.23 0.905.55 0.08
Chemotherapy cycle
Protocol type 2 1 1
ANBL00P1 0.59 0.251.41 0.23 0.90 0.411.98 0.80 3 0.45 0.141.48 0.19 0.80 0.262.50 0.71
PEPI 1 1 4 0.11 0.010.87 0.04 0.50 0.112.34 0.38
Individual 0.57 0.074.55 0.59 0.52 0.074.09 0.53 5 1.72 0.456.56 0.43 0.87 0.233.29 0.84
Tumor reduction 0.98 0.971.00 0.01 0.98 0.970.99 0.001
Pre-op tumor volume (cm3) 1.56 1.062.30 0.02 1.81 1.252.61 0.001 MYCN status
Negative 1 1
Extent of resection Positive 3.60 1.2410.46 0.02 2.53 0.956.76 0.06
N90% 1 1 Tumor reduction % 0.98 0.961.00 0.09 0.99 0.981.01 0.42
5090% 1.51 0.356.52 0.58 1.03 0.244.35 0.97 Pre-op tumor volume (cm3) 1.43 0.802.57 0.23 1.79 1.023.12 0.04

Post-op complications Extent of resection

Minor 1 1 N90% 1 1
Major 1.62 0.475.57 0.44 1.92 0.645.80 0.25 5090% 1.65 0.2311.85 0.62 0.52 0.093.02 0.46
Length of stay 1.05 0.961.15 0.28 1.05 0.971.13 0.20 Age, month 0.98 0.951.00 0.06 0.99 0.971.01 0.31
Cycle 2 group is reference group as we hypothesized that patients treated with less Cycle 2 group is reference group as we hypothesized that patients treated with less
chemotherapy would have worse overall survival. chemotherapy would have worse overall survival.
Log scale Log scale
 Y. Rojas et al. / Journal of Pediatric Surgery 51 (2016) 16651669
the time of diagnosis based on the differences in survival. But, as shown
in Table 1, we found no signicant differences identied with respect to
MYCN status and extent of disease at diagnosis, as determined by MIBG
score, to explain the differences observed in survival. Furthermore, by
multivariate analysis, MYCN status and the number of presurgical che-
motherapy cycles were the only independent prognostic factors of over-
all survival, with MYCN non-amplied patients and 4-cycle patients
demonstrating a decreased risk of death.
Our study has a number of limitations, which include the retrospec-
tive nature of the data, the small number of patients in each chemother-
apy cycle group, and the different chemotherapy protocols used during
the study period. Although our institution is a high-volume center in the
treatment of neuroblastoma, the low incidence of the disease makes it
challenging to acquire signicant numbers from a single institution. Fu-
ture multi-centered, cooperative group studies are needed to better
study this question with larger numbers of patients. In addition, our
nding of superior survival in patients who received 4 cycles of chemo-
therapy prior to surgery is based on the assumption that all patients in
this study are inherently similar at the time of diagnosis and in their re-
sponse to chemotherapy. However, these patients may have subtle un-
accounted differences, which may account for the differences in survival
but were not detected by statistical analyses based on the small cohorts.
Another potential confounding factor is the variability in the treatment
protocol, as not all patients studied received the same chemotherapy
regimen. Patients resected after cycles 4 and 5 were more likely to be
treated according to a more established induction regimen. Therefore,
it is possible that the chemotherapy cycle cohort may be a surrogate
for the chemotherapy protocol.
Another variable that may introduce selection bias is the timing of
surgical resection in our study. At our institution, all decisions on
when to proceed with resection were determined by consensus agree-
ment by a multi-disciplinary team of oncologists and surgeons during
a weekly team meeting. This decision is based on a number of factors
which include resectability based on involvement of critical structures
and blood vessels, chemotherapy toxicity, and decreased tumor reduc-
tion with additional cycles of chemotherapy. Therefore, it is difcult to
ascertain why the patients who underwent surgical resection after
4 cycles of chemotherapy had markedly improved survival compared
to patients who received either more or less cycles prior to surgery. Al-
though patients in the 5-cycle group were similar to the other groups
with respect to MYCN status and extent of disease, the reasons for de-
layed resection after 5 cycles may introduce a component of selection
bias in and of itself. Without prospective documentation of the reasons
why primary resection versus additional chemotherapy was pursued, it
is presumed that patients in the latter group had tumors involving vital
structures that made resection more difcult and thus had surgery post-
poned until induction chemotherapy was complete [16]. However, by
the same token, one might also expect this type of selection bias to
have a positive impact compared to the survival of the 2- and 3-cycle
groups, as these patients would have more resectable tumors, but this
was not observed.
In addition, despite the fact that the 5-cycle chemotherapy patients
had the greatest tumor reduction among all of the groups, the overall
survival was poor at 36%. This may be explained by genetic mutations
that can occur with additional cycles of chemotherapy. Genomic analy-
ses have recently demonstrated that the genetics of sporadic neuroblas-
toma is far more complex, as other recurrent genetic abnormalities have
been sparse [17]. Recent studies analyzing genetic changes in relapse tu-
mors, however, have shown clonal evolution with new mutations follow-
ing chemotherapy when compared to those at diagnosis [18]. Further
genetic studies may elucidate the molecular basis of our ndings.
Another explanation for poor outcomes in this group is that patients
who received 5 cycles of chemotherapy were more likely to experience
treatment toxicity compared to patients who received less chemother-
apy. In fact, all patients who received individualized protocols because
of treatment toxicity were in the 5-cycle group. Moreover, at our
1669
institution, which favored earlier resection during this study period, a
patient who received 5 cycles of chemotherapy prior to surgery would
imply a particularly invasive tumor that had a poor response or a patient
with a number of complications and infections.
With regards to the improved survival seen in the 4-cycle patients,
one hypothesis could be the potential positive impact of post-
operative clean-up chemotherapy. Patients resected after 4 cycles
may have received the greatest possible tumor reduction benet from
chemotherapy before resection and then an additional benet from
the last clean-up cycle as the minimal residual post-surgical disease
burden would have been very small.
Induction chemotherapy in HRNB decreases tumor burden and facil-
itates surgical resection; however, there appears to be a tipping point in
which more chemotherapy may result in poorer survival. Despite the
limitations of our study, the observed superior survival of patients
who underwent surgical resection after 4 cycles of induction chemo-
therapy suggests timing of surgery may play a crucial role in the long-
term survival of HRNB patients. As is currently practiced in other pedi-
atric solid tumors such as rhabdomyosarcoma, Ewing's sarcoma, and
Wilms tumor, future COG protocols for HRNB should consider standard-
ized timing for surgical resection during induction chemotherapy based
on the results of prospective trials.
References
[1] Park JR, Bagatell R, Hogarty M, et al. Children's Oncology Group's 2013 blueprint for
research: Neuroblastoma. Pediatr Blood Cancer 2012;60:98593.
[2] Davidhoff AM. Neuroblastoma. Semin Pediatr Surg 2012;21:214.
[3] Matthay KK, Villablanca JG, Seeger RC, et al. Treatment of high-risk neuroblastoma
with intensive chemotherapy, radiotherapy, autologous bone marrow transplanta-
tion, and 13-cis-retinoic acid. N Eng Med 1999;341:116573.
[4] Yu AL, Gilman AL, Ozkaynak MF, et al. Anti-GD2 antibody with GM-CSF, interleukin-
2, and isotretinoin for neuroblastoma. N Eng Med 2010;363:132434.
[5] Simon T, Haberle B, Hero B, et al. Role of surgery in the treatment of patients with
stage 4 neuroblastoma age 18 months or older at diagnosis. J Clin Oncol 2013;31:
7528.
[6] LaQuaglia MP, Kushner BH, Heller G, et al. Stage 4 neuroblastoma diagnosed at more
than 1 year of age: Gross total resection and clinical outcome. J Pediatr Surg 1994;
29:11625.
[7] DVon AllmenAMDavidoffWLondonInuence of extent of resection on survival in
high risk neuroblastoma patients: A report from the COG A3973 study, Presented
at Advances in Neuroblastoma Research Conference. Cologne, Germany. Available
from: http://www.anr2014.com/leadmin/ANR_Kongress/Programm/ANR_2014_
Information_Book_2014-05-08.pdf; 2014.
[8] Holmes K, Sarnacki S, Poetschger U, et al. Inuence of Surgical Excision on Survival of
Patients with High Risk Neuroblastoma: Report from Study 1 of Siop Europe
(Siopen). Presented at Advances in Neuroblastoma Research Conference, Cologne,
Germany. Available from: http://www.anr2014.com/leadmin/ANR_Kongress/
Programm/ANR_2014_Information_Book_2014-05-08.pdf; 2014.
[9] Davidhoff AM, Corey BL, Hoffer FA, et al. Radiographic assessment of resectability of
locoregional disease in children with high-risk neuroblastoma during neoadjuvant
chemotherapy. Pediatr Blood Cancer 2005;44:15862.
[10] Brodeur GM, Seeger RC, Barrett A, et al. International criteria for diagnosis, staging
and response to treatment in patients with neuroblastoma. J Clin Oncol 1988;6:
187481.
[11] Monclair T, Brodeur GM, Ambros PF, et al. The International Neuroblastoma
Risk Group (INRG) staging system: an INRG Task Force report. J Clin Oncol 2009;
27:298303.
[12] Cohn SL, Pearson ADJ, London WB, et al. The International Neuroblastoma Risk
Group (INRG) classication system: An INRG Task Force report. J Clin Oncol 2009;
27:28997.
[13] Messina JA, Cheng S, Franc BL, et al. Evaluation of semiquantitative scoring system
for metaiodobenzylguanidine (mIBG) scans in patients with relapsed neuroblasto-
ma. Pediatr Blood Cancer 2006;47:86574.
[14] Goldie JH, Coldman AJ. Genetic origin of drug resistance in neoplasms: Implications
for systemic therapy. Cancer Res 1984;44:364353.
[15] Medary I, Aronso D, Cheung NV, et al. Kinetics of primary tumor regression with che-
motherapy: Implications for the timing of surgery. Ann Surg Oncol 1996;3(6):
5215.
[16] Cecchetto G, Mosseri V, De Bernardi B, et al. Surgical risk factors in primary sur-
gery for localized neuroblastoma: The LNESG1 study of the European Interna-
tional Society of Pediatric Oncology Neuroblastoma Group. J Clin Oncol 2005;
23:84839.
[17] Bosse KR, Maris JM. Advances in the translational genomics of neuroblastoma: From
improving risk stratication and revealing novel biology to identifying actionable
genomic alterations. Cancer 2016;133.
[18] Eleveld TF, Oldridge DA, Bernard V, et al. Relapsed neuroblastomas show frequent
RAS-MAPK pathway mutations. Nat Genet 2015;47(8):86471.