Long-term effects on vertebral body height growth of dose sculpting intensity modulated radiation therapy for children with neuroblastoma

PDF

Download

Open document

Flip pages

Summarize

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content
LONG-TERM EFFECTS ON VERTEBRAL BODY HEIGHT GROWTH OF
DOSE SCULPTING INTENSITY MODULATED RADIATION THERAPY FOR
CHILDREN WITH NEUROBLASTOMA

by

Chia-Ling Wu

A Thesis Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA

In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(APPLIED BIOSTATISTICS AND EPIDEMIOLOGY)

May 2017

Copyright 2017 Chia-Ling
i

ACKNOWLEDGEMENT
I would like to take this moment to express my gratitude to all the experts who have guided
and supported me to complete my master thesis.
I would like to thank my advisor, Dr. Richard Sposto. Dr. Sposto has always been patient
and kind to explain every question about my research and writing. His consistent instruction
always steered me in the right direction.
I would like to acknowledge Dr. Lydia Ng, Dr. Arthur Olch, Dr. Kenneth Wong and
Children's Hospital Los Angeles (CHLA) for kindly providing me this valued data and the
thoughts from clinical experts. Thank you for letting me have this chance to collaborate with
and learn from you.
Additionally, I would like to thank my thesis committee members, Dr. Donald Barkauskas
and Dr. Arthur Olch for taking the time out of their schedule to review my thesis and
providing me valuable comments.
Finally, I must express my gratitude to my family for unfailing support and continuous
encouragement throughout my years of study. This accomplishment would not have been
possible without them.

ii

TABLE OF CONTENTS
ACKNOWLEDGEMENT ..................................................................................................... i
TABLE OF CONTENTS ...................................................................................................... ii
LIST OF TABLES ............................................................................................................... iii
LIST OF FIGURES ............................................................................................................. iv
ABSTRACT ........................................................................................................................... v
INTRODUCTION ................................................................................................................. 1
Background of Neuroblastoma ........................................................................................... 1
Long-term Vertebral Growth Anomalies in Neuroblastoma .............................................. 1
Risk Factors of Vertebral Growth Defect ........................................................................... 2
Research Objectives ........................................................................................................... 3
METHODS ............................................................................................................................ 5
Study Design ...................................................................................................................... 5
Participants ......................................................................................................................... 6
Statistical Methods ............................................................................................................. 6
RESULTS .............................................................................................................................. 9
Characteristics of Participants ............................................................................................ 9
Vertebral Body Height Growth Rate in Absolute Difference .......................................... 10
Vertebral Body Height Growth Rate in Difference Ratio ................................................ 14
Vertebral Body Height Growth in 10 Years vs. Radiation Dose ..................................... 16
DISCUSSION/ CONCLUSION .......................................................................................... 18
REFERENCE ....................................................................................................................... 22
TABLES ............................................................................................................................... 23
FIGURES ............................................................................................................................. 30

iii

LIST OF TABLES
Table 1. Characteristics of Participants 23
Table 2. The Relationship of Predictors with Growth Rate in Absolute Difference 24
Table 3. The Relationship of Growth in Absolute Difference and Follow-up Years
Accounted for Interactions 25
Table 4. The Relationship of Growth in Absolute Difference and Follow-up Years
Accounted for All Potential Interactions 26
Table 5. The Relationship of Predictors with Growth Rate in Difference Ratio 27
Table 6. The Relationship of Growth in Difference Ratio and Follow-up Years Accounted
by Interaction 28
Table 7. The Relationship of Growth in Difference Ratio and Follow-up Years Accounted
for All Potential Interactions 29

iv

LIST OF FIGURES
Figure 1. Regression Plot of Height Growth in Absolute Difference on Follow-up Years,
Interacts by Age. 30
Figure 2. Regression Plot of Height Growth in Absolute Difference on Follow-up Years,
Interacts by Radiation Dosage. 31
Figure 3. Regression Plot of Height Growth in Absolute Difference on Follow-up Years,
Interacts by Gender. 32
Figure 4. Regression Plot of Height Growth in Absolute Difference on Follow-up Years,
Interacts by Spine. 33
Figure 5. Regression Plot of Height Growth in Absolute Difference on Follow-up Years,
Interacts by Age and Radiation Dose. 34
Figure 6. Regression Plot of Height Growth in Absolute Difference on Follow-up Years,
Interacts by Spine and Radiation Dose. 35
Figure 7. Regression Plot of Height Growth in Difference Ratio on Follow-up Years,
Interacts by Age. 36
Figure 8. Regression Plot of Height Growth in Difference Ratio on Follow-up Years,
Interacts by Radiation Dose. 37
Figure 9. Regression Plot of Height Growth in Difference Ratio on Follow-up Years,
Interacts by Gender. 38
Figure 10. Regression Plot of Height Growth in Difference Ratio on Follow-up Years,
Interacts by Spine. 39
Figure 11. Regression Plot of Height Growth in Difference Ratio on Follow-up Years,
Interacts by Age and Radiation Dose. 40
Figure 12. Regression Plot of Height Growth in Difference Ratio on Follow-up Years,
Interacts by Spine and Radiation Dose. 41
Figure 13. Scatter Plot of Height Growth in 10 Years Difference on Radiation Dose. 42
Figure 14. Regression Plot of 10-Year Height Growth in Absolute Difference on Radiation
Dose, Interacts by Spine. 43
Figure 15. Regression Plot of 10-Year Height Growth in Difference Ratio on Radiation
Dose, Interacts by Spine. 44

v

ABSTRACT

Background: Children with neuroblastoma who are treated with radiotherapy have
increased incidence of vertebral body growth deformities. Intensity-modulated radiation
therapy (IMRT) has been adopted to attempt to reduce the exposure of normal tissues to the
toxicities of radiation therapy, with the expectation of reducing growth inhibition without
sacrificing tumor coverage. This study aimed to estimate the influence of radiotherapy and
other clinical factors- age at treatment, gender, vertebral spine location- on vertebral body
height growth for patients who had childhood neuroblastoma treated with dose sculpting
IMRT.
Methods: A retrospective review was performed on children with neuroblastoma treated
with radiation therapy at Children’s Hospital Los Angeles (CHLA) from year 2000 - 2011,
with a minimum of 3 years of follow-up imaging. Vertebral body height was measured prior
to and several years after radiation in the thoracic and lumbar spinal section for each patient.
Delivered radiation dose was also recorded for these spinal sections. Vertebral body height
growth was defined in two ways: vertebral body height growth in absolute difference (cm)
and in difference ratio (%). We used generalized estimating equations (GEE) models to
evaluate the associations between vertebral height growth and follow-up years, and the
influence of age at treatment, gender, mean dose of radiation, and location of radiated spine
on growth rate.
Results: Data were available on 34 participants with 56% male and 44% female, mean age
at treatment of 5 years old, and average length of follow-up of 5.5 years. Of 133 vertebral
vi

bodies, 48 (36%), 44 (33%), and 41 (31%) were designated as target, spared and control
respectively. Target vertebral body radiation dose was 23.6 Gray (Gy) on average whereas
spared dose was 13.0 Gy. For absolute difference (cm), older age at treatment, higher
radiation mean dose, male sex and thoracic spine location were significantly associated with
lower growth rate. The vertebral body height grew 0.0536 cm per body per year for the
thoracic spine of male children aged two years old in the control group. An increase of one
year of age at time of irradiation significantly decreased vertebral body height growth by
0.0015 cm/ body/ year (p=0.04); an increase of one Gy radiation dose decreased growth by
0.0012 cm/ body/ year (p<0.0001); females grew faster than males by 0.0085 cm/ body/ year
(p=0.007); lumbar spine grew faster than thoracic spine by 0.0159 cm/ body/ year which was
statistically significant (p<0.0001) as well. In difference ratio (%), increasing age at
treatment, higher radiation dosage and male sex had significant negative effects on vertebral
height growth. An increase of one year of age at time of radiation decreased height growth
by 0.23% (p<0.0001); a one Gy increase in radiation dose decreased growth by 0.09%
(p<0.0001); males grew slower than females by 0.52% (p=0.04). The effect of radiation
mean dose on growth was different for different ages (p=0.0008), which indicates older
children were less affected by radiation dose.
Conclusion: Radiation dosage, treatment age and gender significantly influenced the
vertebral body height growth, both in absolute difference and for difference ratio. Lumbar
spine grew faster than thoracic spine in absolute difference. We also found lumbar spine
and younger age were more sensitive to radiation dose.
1

INTRODUCTION
Background of Neuroblastoma
Neuroblastoma is a common extracranial tumor of childhood. It is the third most common
pediatric cancer; however, it is the most common tumor in infants younger than 1 year old
(Irwin & Park, 2015; Whittle et al., 2017). The incidence of neuroblastoma in North America
and Europe is 10.5 per million children between 0 and 14 years of age, with a slight male
predominance (1.2:1.0) (Hartley et al., 2008; Ries, 1999; Stiller & Parkin, 1992). Each year,
around 650 new cases are diagnosed in the United States (Ries, 1999). Most patients with
neuroblastoma are diagnosed between 0 and 4 years of age, with a median age of 19 months
(Irwin & Park, 2015). There has been a noticeable improvement in clinical neuroblastoma
outcomes during the past several decades. The overall survival rates for patients with low
and intermediate risk neuroblastoma are greater than 90% whereas high-risk patients remain
at 40-50% (Irwin & Park, 2015; Whittle et al., 2017).
Long-term Vertebral Growth Anomalies in Neuroblastoma
At least half of neuroblastoma patients now reach adulthood but with time may develop
adverse chronic health conditions. Long-term survivors of neuroblastoma have an increased
cumulative incidence of musculoskeletal anomalies (Dorr, Kallfels, & Herrmann, 2013;
Paulino & Fowler, 2005; Probert & Parker, 1975; Yu et al., 2015). The reasons for
musculoskeletal anomalies are many, including direct effects of cancer, poor nutrition and
health condition, and lack of physical activity after disease onset (Halton et al., 1995).
However, several studies showed that radiotherapy may have additional direct or indirect
2

negative effects on musculoskeletal development (Dorr et al., 2013; Paulino & Fowler, 2005;
Probert & Parker, 1975; Yu et al., 2015). Around half (51%) of patients receiving
radiotherapy for pediatric cancers had recorded pathological findings in skeletal system and
soft tissues, including short stature, scoliosis, or kyphosis (Dorr et al., 2013). A study in 2005
indicated the overall 5, 10, and 15 years’ follow-up for scoliosis-free rates were 87.6%,
79.0%, and 76.0% respectively (Paulino & Fowler, 2005). A retrospective cohort study for
neuroblastoma patients treated with local radiotherapy and half with total body irradiation
(TBI) compared their height percentiles with national growth charts, at the time of diagnosis,
pre-radiotherapy and follow-up with magnetic resonance imaging (MRI) examination, and
showed the height percentiles of patients had a dramatic decrease at the time of follow-up
MRI examine (Yu et al., 2015). Based on these studies, we know radiotherapy impairs
musculoskeletal and height growth.
Intensity-modulated radiation therapy (IMRT) has been adopted in an attempt to reduce the
exposure of normal tissues to the toxicities of radiation therapy. IMRT conforms the
radiation dose more precisely to match the shape of the tumor by modulating, or controlling,
the intensity of the radiation beam in multiple small volumes. IMRT also allows higher
radiation doses to be focused to regions within the tumor while minimizing the dose to
surrounding normal critical structures. Therefore, IMRT is expected to reduce growth
inhibition without sacrificing tumor coverage.
Risk Factors of Vertebral Growth Defect
Though multiple studies revealed that radiation dose correlated with musculoskeletal
anomalies, not many studies specifically investigated the risk factors of short stature and
3

growth delay in children receiving radiotherapy. A study (Hartley et al., 2008) in the United
States of patients receiving craniospinal irradiation (CSI) and chemotherapy indicated that
age, gender, and dose of craniospinal irradiation had significant effects on the growth rate
of some vertebral body segments, especially in T4-T5, C6-T3, T4-T7 and L1-L5. Age and
female sex were negatively correlated with growth rate. Compared with average irradiation
dose (23.4 Gy), spinal sections receiving high dose (36-39.6 Gy) had significantly lower
growth rate. Moreover, lumbar spine, e.g. L1-L5, was more sensitive to radiation, compared
with cervical and thoracic spine (Hartley et al., 2008). However, another study presented
different results that indicated the total radiation dose and age at radiotherapy were not
significantly associated with low height percentile, but the number of irradiated vertebrae
and having TBI were associated (Yu et al., 2015). The inconsistency may be because the
height measurement used in the Hartley’s study was vertebral height, whereas the Yu’s study
used standing height.
Research Objectives
Though the long-term negative effect of radiation on height growth is well-know, there is
less discussion about the statistical relationship between radiation and vertebral body height.
This study is the first study to assess the effects on vertebral body height growth for
neuroblastoma patients treated with dose sculpting IMRT. The study aim is to estimate how
radiotherapy and other clinical factors, such as age at treatment, gender, vertebral spine
location, influence vertebral body height growth per follow-up year for patients who had
childhood neuroblastoma treated with dose sculpting IMRT.
4

The research question of this study is whether vertebral height growth per year of follow-up
is affected by radiation dose, age at treatment, gender and vertebral spine location for
patients who had childhood neuroblastoma treated with dose sculpting IMRT at Children’s
Hospital Los Angeles (CHLA) from years 2000 - 2011. This study hypothesis is that
radiation dose, age at treatment, gender and vertebral spine location all affect the children’s
vertebral body height in a linear relationship.

5

METHODS
Study Design
This study aimed to assess the effects of radiation dose, age at treatment, gender and spine
location on vertebral body height growth rate for patients who had childhood neuroblastoma
treated with dose sculpting IMRT. This study is an internal control study that compared
vertebral body height change prior to and several years after radiation in different spinal
segments that received different radiation doses (target, spared, control). Data on patients at
Children’s Hospital Los Angeles (CHLA) from years 2000 - 2011 were collected by Dr.
Lydia Ng in collaboration with Dr. Arthur Olch and Dr. Kenneth Wong and were kindly
provided for the analysis.
The vertebral height of thoracic spine and lumbar spine were measured separately, due to
reported differences in growth rates both before and during puberty, as well as a reported
difference in sensitivity to craniospinal irradiation (Dimeglio & Canavese, 2012; Hartley et
al., 2008). The heights of vertebral bodies with different mean radiation doses- target, spared,
and control- were measured separately as well. In each patient’s pre-treatment planning CT,
craniocaudal height was measured en-bloc for target vertebral bodies and for spared
vertebral bodies (Ng et al., 2015). An internal control group of out-of-field vertebral bodies
was measured, if available. Craniocaudal heights of target, spared, and control vertebral
bodies were similarly measured on follow-up imaging: on the CT scout if available, or on
coronal T2-weighted MRI series if not. Dose-volume data for target versus spared vertebral
bodies were extracted and correlated to vertebral body growth rates (Ng et al., 2015).
6

Participants
This study collected data on patients with high-risk neuroblastoma (Cohn et al., 2009) who
underwent dose sculpting IMRT at CHLA from January 2000 to December 2011. Patients
less than 15 years of age at treatment were enrolled in this study. Included children were
those with paravertebral primary tumors with a minimum of 3 years of evaluable post-
treatment thoracolumbar imaging. Patients who underwent re-irradiation of the spine prior
to follow-up imaging were excluded (Ng et al., 2015).
Statistical Methods
Each subject had measured vertebral body height prior to and following radiation with height
and dosage measurement separately for different spinal sections. Subject’s characteristics
including age at treatment and gender were also collected. Each subject had multiple spinal
sections with two timepoint measurements.
This study presumed that the main effect to influence vertebral body height growth was
follow-up years, with age at treatment, gender, mean dose of radiation, location of radiated
spine interacting with follow-up years to modify vertebral body height growth rate. Outcome
assessment of vertebral body height growth was defined in two ways: vertebral body height
growth in absolute difference (cm) and in difference ratio (%). Height growth in absolute
difference is equal to the height change in centimeters (cm) for each spinal section between
the time of treatment start and the time of follow-up. Height growth in difference ratio is
equal to vertebral body height at follow-up divided by vertebral body height at treatment
minus 1.
7

𝑉𝑒𝑟𝑡𝑒𝑏𝑟𝑎𝑙 𝐵𝑜𝑑𝑦 𝐻𝑒𝑖𝑔 ℎ𝑡 𝐺𝑟𝑜𝑤𝑡 ℎ 𝑖𝑛 𝐴𝑏𝑠𝑜𝑙𝑢𝑡𝑒 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 =
𝐻𝑒𝑖𝑔 ℎ𝑡 𝑎𝑡 𝐹𝑜𝑙𝑙𝑜𝑤 𝑢𝑝 (𝑐𝑚 ) − 𝐻𝑒𝑖𝑔 ℎ𝑡 𝑎𝑡 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑐𝑚 )
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑆𝑝𝑖𝑛𝑎𝑙 𝑆𝑒𝑐𝑡𝑖𝑜𝑛
𝑉𝑒𝑟𝑡𝑒𝑏𝑟𝑎𝑙 𝐵𝑜𝑑𝑦 𝐻𝑒𝑖𝑔 ℎ𝑡 𝐺𝑟𝑜𝑤𝑡 ℎ 𝑖𝑛 𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑅𝑎𝑡𝑖𝑜 =
𝐻𝑒𝑖𝑔 ℎ𝑡 𝑎𝑡 𝐹𝑜𝑙𝑙𝑜𝑤 𝑢𝑝 (𝑐𝑚 )
𝐻𝑒𝑖𝑔 ℎ𝑡 𝑎𝑡 𝑇𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 (𝑐𝑚 )
− 1
Considering the correlations between vertebral body height in each section and timepoint,
we used generalized estimating equations (GEE) models (Zeger & Liang, 1986) to evaluate
the associations between vertebral height growth and follow-up years, and the influence of
age at treatment, gender, mean dose of radiation and location of radiated spine. We assumed
a linear relationship between growth and time, as study subjects were mostly within 5 years
old during which sitting height growth is approximately linear (Dimeglio & Canavese, 2012).
This study assumed normal distribution for those participants’ height growth. This study
assumed zero-intercept should fit the data, since theoretically there is no height growth with
zero follow-up years. Hence, we estimated the models with intercept and without intercept
to see if the zero-intercept assumption is appropriate and we can eliminate the intercept in
the model. Below is our study formula:
𝑦 = 𝛼 + 𝛽 1
𝑋 1
+ 𝛽 2
𝑋 2
𝑋 1
+ 𝛽 3
𝑋 3
𝑋 1
+ 𝛽 4
𝑋 4
𝑋 1
+ 𝛽 5
𝑋 5
𝑋 1

y: Vertebral body height growth
X1: Follow-up years
X2: Age at treatment
X3: Radiation Dose
X4: Gender (0: Male, 1: Female)
X5: Location of spine (0: Thoracic, 1: Lumbar)
Descriptive statistics were analyzed for patients’ personal and clinical variables, including
age at treatment, age at follow-up, follow-up years, gender, radiation mean dose, and spinal
8

location. Multivariate GEE regression analysis was used to analyze possible influence
factors for the vertebral body height growth rate. In our model, p-values less than 0.05 were
considered to indicate statistical significance; parameter estimates (β) and its 95%
confidence interval (CI) were reported. At first, univariate GEE regression analysis was used
to see the raw relationship between growth rate with follow-up years, age at treatment,
gender, radiation mean dose and spinal location, respectively. Next, we evaluated the main
effect of follow-up years, adding the interaction of follow-up years and age at treatment,
gender, radiation dose and spinal location into the model separately and individually to see
the interaction effect. Finally, we evaluated all the interactions in multivariate model to
evaluate the significance of the interaction effects in a stepwise fashion. In the beginning,
follow-up years and all interactions with covariates were included in the model. Interactions
with p-value less than 0.05 were included in the final model and considered as significant
influencing factors. We examined the zero-intercept model with all significant influence
factors to see if the intercept met our zero-intercept assumption, e.g. the coefficient closed
to zero and p-value is not significant. Considering that some of the variables may moderate
differently across the levels of other variables, for instance, male and female may have
different growth rates by age, we assessed the three-way interactions that follow-up years
interacts with treatment age and radiation dose, treatment age and gender, as well as radiation
dose and spinal location.
All data were analyzed using Statistical Analysis System software version 9.4 (SAS institute,
Inc., Cary, North Carolina).

9

RESULTS
Characteristics of Participants
A total of 88 children with high-risk neuroblastoma underwent radiotherapy at CHLA from
year 2000 to 2011. Eligible patients were those with paravertebral primary tumors with a
minimum of 3 years of evaluable post-treatment thoracolumbar imaging, and we excluded
children who underwent re-irradiation of the spine prior to follow-up imaging. We had 34
evaluable cases. All of them had high-dose chemotherapy prior to radiation, almost all (33
of 34) had planned stem cell rescue. They were irradiated with a Varian 2100C linear
accelerator with 120 leaf multileaf collimator, using IMRT with 6MV photons (Ng et al.,
2015).
Table 1 shows characteristics of study participants. A total of 34 participants had
measurements on 133 spinal sections. Participants were 19 males (56%) and 15 females
(44%). Age at treatment ranged from 1.8 to 15.8 years old, mean age at treatment was 5
years with 3.0 standard deviation. Median age at treatment was 4.4 years, and 23 (68%)
children were under 5 years of age. On average, patients had 5.5 years of follow up, while
age at follow-up ranged from 5.5 to 20.9 years, with an average age of 10.6 years. With dose
sculpting IMRT, 32 children had vertebral bodies with spared dose, and 27 children had
vertebral bodies intentionally targeted to 18 Gy or more. At time of data collection 79% of
children were still alive. The 133 spinal sections evaluated were about equally divided
between thoracic spine and lumbar spine (48% vs 52%). The distribution of target, spared
and control vertebral body sections was 48 (36%), 44 (33%), and 41 (31%). The average
dose to target vertebral bodies was 23.6 Gy whereas the spared vertebral bodies received
10

13.0 Gy, but the dose ranges overlapped: 6.4 - 23.3 Gy for spared dose and 17.7 - 34.1 Gy
for target dose.
Vertebral Body Height Growth Rate in Absolute Difference
Vertebral body height growth in absolute difference is the average height change in
centimeters (cm) for each spinal section between treatment start and follow-up. We used the
GEE model to assess the associations between vertebral body height growth and potential
influence factors.
Table 2 shows the univariate relationships of height growth in absolute difference with
follow-up years, age at treatment, radiation mean dose, gender and location of spine. Follow-
up years and location of spine were strongly statistically significant associated with growth
rate in absolute difference (p<0.0001), on average, vertebral bodies grow 0.0526 cm per
body per year (95% CI: 0.0391 - 0.0662). Compared with thoracic spine, lumbar spine
sections grew faster in 0.0853 cm/ body/ year (95% CI: 0.0487 - 0.1219). Radiation mean
dose was also strongly significantly associated with growth rate (p<0.0001), in a negative
relationship. A one Gy increase of radiation mean dose decreased height growth by 0.0061
cm/ body/ year (95% CI: 0.0041 - 0.0082) for each spinal section. Age at treatment start and
gender were not significant associated with growth rate in this model. But it indicated that
the older age at radiation, and male sex tend to have less growth on average.
The study model presumed vertebral body height is linearly associated with follow-up years,
and that age at treatment, radiation mean dose, gender and location of spine interacts with
follow-up years to influence the association. We aimed to assess how the association was
11

changed by these covariate interactions, both univariately and multivariately. Hence, we
conducted the multivariate model with zero-intercept including follow-up years plus
significant interactions of the variables with follow-up years.
Table 3 presents the relationships of growth in absolute difference with follow-up years, and
included interactions with each covariate individually to assess its influence on growth rate.
Model 1 included the main effect follow-up years only, while model 2 to 5 were based on
model 1 and accounted for interaction of follow-up years with treatment age, radiation mean
dose, gender and location of spine, respectively. The interaction with treatment age (model
2) was not significantly associated with vertebral height growth (p=0.53). In model 3, the
interaction with radiation dosage was significant (p<0.0001), indicating that a one Gy
increase in dosage decreased growth rate by 0.0011 cm/ body/ year (95% CI: 0.0007 -
0.0015). Model 4 and 5 showed that interactions with gender and spine were significant
influencing factors as well (p=0.01, p=0.0008 respectively), with vertebral body height
growth rate in females significantly higher than in males by 0.0119 cm/ body/ year, and
lumbar spine growth rate higher than thoracic spine by 0.0131 cm/ body/ year. It is notable
that the coefficients of the intercept terms were all close to zero and not significant, which
confirmed our zero-intercept assumption that the vertebral body height growth should be
zero if the number of follow-up years is zero.
The effect of follow-up years on vertebral body height growth in absolute difference,
accounting for potential interactions in multivariate analysis is presented in Table 4. Model
1 included follow-up years as a main effect and all potential influencing factors. All factors
were significant in the model except intercept, which affirmed our zero-intercept assumption.
12

Hence, the final model without intercept indicated follow-up years and the interactions all
influenced vertebral body height growth. After adjusting for interactions, the beta coefficient
of follow-up years was 0.0566 (95% CI: 0.0496 - 0.0733) and strongly significant
(p<0.0001). Each vertebral body height grew 0.0536 cm for the thoracic spine of male
children aged two years old in the control group in each year. An increase of one year of
treatment age significantly decreased vertebral body height growth rate by 0.0015 cm/ body/
year (95% CI: 0.0029 - 0.0001, p=0.04); an increase of one Gy radiation dose statistically
significantly decreased growth rate by 0.0012 cm/ body/ year (95% CI: 0.0016 - 0.0009;
p<0.0001); females grew faster than males by in 0.0085 cm/ body/ year (95% CI: 0.0026,
0.0168; p=0.007); lumbar spine grew faster than thoracic spine by 0.0159 cm/ body/ year
(95% CI: 0.0092 - 0.0225) which was statistically significant (p<0.0001) as well. Generally,
the results from univariate and multivariate models were consistent -- all factors significantly
influenced vertebral body height growth rate with one exception: interaction with age at
treatment was significant (p=0.04) in multivariate analysis but not significant (p=0.53) when
considered alone.
We assessed three-way interactions of follow-up years with treatment age and gender,
follow-up years with treatment age and radiation dose, and follow-up years with spinal
location and radiation dose. Only the interaction of follow-up years with spinal location and
radiation dose was convincingly significant (p<0.0001). The p-value was 0.51 for the
interaction of follow-up years with age and gender; the p-value was 0.053 for the interaction
of follow-up years with age and radiation dose.
13

Figure 1 to 6 present the interaction effects between follow-up years and the influencing
factors in the GEE multivariate model. The main effect of follow-up years and all significant
interactions were included, with continuous predictors categorized. The points are observed
values. For all figures, the regression lines were evaluated at the mean value of each group
for the factors of interest, and were evaluated at the overall mean value for the remaining
factors in the model. For age at treatment (Figure 1), we used the mean age in the group aged
less than 4 and over 4, and overall mean value of radiation dose, gender and spinal location.
This showed height growth rate for children aged at treatment less than 4 was higher. For
radiation dosage, we used the mean dose of control, spared and target groups. Height growth
rate was highest in the control group, and was successively lower in spared and target groups
(Figure 2). Females grew faster than males (Figure 3) and lumbar spine grew faster than
thoracic spine (Figure 4). The p-values shown are those from the final model in Table 4.
Figure 5 and 6 added three-way interactions in the model. Figure 5 presents the interaction
of follow-up years with treatment age and radiation mean dose. Although the growth rate
decreased more among dose groups for the age less than 4 group, showing that the dose
effect may be greater in younger treatment age group, this interaction was not significant in
absolute difference scales. Figure 6 presents three-way interaction of follow-up years with
spinal location and radiation doses. This shows that height growth in lumbar spine was
higher than in thoracic spine when treated with control and spared dose, however it was
lower when treated with target dose, and the effect was significant (p<0.0001). This suggests
lumbar spine may have higher susceptibility to radiation dose in the absolute difference scale.

14

Vertebral Body Height Growth Rate in Difference Ratio
Vertebral body height growth in difference ratio is the height at follow-up divided by the
height at first radiation treatment minus 1. This measurement indicates the proportion change
in growth of each spinal section between treatment start and follow-up time.
The univariate relationships between vertebral body height growth in difference ratio and
follow-up years, age at treatment, radiation mean dose, gender and location in spine are
displayed in Table 5. Follow-up years, age and radiation mean dose were strongly
statistically associated with growth in difference ratio (p<0.0001, p=0.002, p<0.0001
respectively). The vertebral body height grew 3.29% per year (95% CI: 2.38% - 4.20%).
Age and radiation dosage negatively affected vertebral body growth. Each additional year
of age decreased growth by 1.02% (95% CI: 0.36% - 1.68%). Each additional one Gy of
radiation dose decreased growth by 0.46%. Females and lumbar spine had higher growth,
although these were not significant (p=0.50 for gender, p=0.15 for spine).
Table 6 demonstrates the relationships of growth rate in difference ratio with follow-up years,
and its interactions with other variables. Model 1 accounted for the main effect of follow-up
years only, model 2 to 5 added the influence factors: age at treatment, radiation mean dose,
gender and location of spine, respectively. The interaction with treatment age (model 2) and
the interaction with radiation dosage (model 3) both had significant negative effect
(p<0.0001), which showed that an increase of one year of age at start of radiation decreased
vertebral height growth rate by 0.16%, and an increase of one Gy radiation dose decreased
growth rate by 0.09% (95% CI: 0.06 - 0.11%). As in Table 5, gender and location of spine
did not significantly influence growth rate in difference ratio (p=0.051 for gender, p=0.55
15

for spine). The coefficients of the intercepts were all statistically close to zero which was
compliant with our zero-intercept assumption. The p-values of follow-up years remained
significant which indicated time is the strongest influence factor of height growth, with other
factors interacting with time to influence height growth.
The effect of follow-up years on vertebral body height growth in difference ratio accounting
for potential interactions in multivariate analysis is shown in Table 7. Model 1 included
follow-up years and all potential influencing factors. The interaction with location of spine
was not significant (p=0.09). Hence, we next excluded the interaction with spine location in
model 2, in addition to excluding the intercept in final model. After adjusting for interactions,
the vertebral body height grew 4.3% each year for male children aged two years old in the
control group. Increasing radiation age, dosage and male sex had significant negative effects
on vertebral height growth rate. Each additional one year of age at treatment decreased
growth rate by 0.23% (95% CI: 0.14% - 0.31%; p<0.0001); one Gy increase of radiation
dose decreased growth rate by 0.09% (95% CI: 0.07% - 0.12%; p<0.0001); males grew
slower than females by 0.52% (95% CI: 0.03% - 1.02%; p=0.04). The only difference
between this multivariate model with the previous univariate model was that gender changed
to significant (p=0.04) in multivariate one, thus was included in the final model.
Three-way interaction of follow-up years with treatment age and gender was not significant
(p=0.29). Three-way interaction of follow-up years with treatment age and radiation dose
was significant (p=0.0008) with estimates 0.01% (Figure 11), which indicated growth rate
of younger children decreased more with higher radiation dose. That means younger age was
more sensitive to radiation dose on the percent change scale. We also tested the three-way
16

interaction of follow-up years with spinal location and radiation doses. Height growth rate
in lumbar spine decreased more when dose increased, and it became lower than thoracic
spine when treated with target dose (Figure 12), and the effect was significant (p=0.0001).
That suggested lumbar spine has higher susceptibility to radiation dose in percent change
scale.
Figure 7 to 12 present the interaction effects between follow-up years and influence factors
in multivariate GEE model. The main effect of follow-up years and all significant
interactions in multivariate analysis are included, as the final model in Table 7. The points
are observed values. The regression lines were evaluated at the mean value of each group
for the factors of interest and overall mean value of the rest of factors in the model. In order
to present the effect of spinal location, the plot for spine (Figure 9) was based on the final
multivariate model adding the effect of spine location. Hence the p-value reflects the
interaction of follow-up years and spine with all other influence factors in the model. It
showed height growth rate of lumbar was slightly larger than thoracic spine but it was not
significant (p=0.10). Treatment age, dosage and gender significantly influenced height
growth rate in the percent change scale (Figure 6 - 8) with p-values <0.0001.
Vertebral Body Height Growth in 10 Years vs. Radiation Dose
Vertebral body height growth in 10 years was also assessed in absolute difference and ratio
scale. 10-year height growth in absolute difference equals the difference in height between
2 timepoints, divided by number of vertebral bodies and follow-up years (cm/ body/ year),
then multiplied by ten. 10-year height growth in ratio equals the height growth in proportion
divided by follow-up years then multiplied by ten. In Figure 13, a scatter plot for the 10-year
17

height growth vs radiation dose, we could see height growth was getting lower with higher
dosage, and almost obeyed a linear relationship. This affirmed our assumption that radiation
dose linearly decreases the vertebral body height growth. We further assessed whether the
relationship between 10-year height growth rate with radiation dose differs by spinal location.
Figure 14 shows that the 10-year height growth in difference scale decreased when dose
increased. Compared with thoracic spine, lumbar spine grew faster at low doses, but the
growth rate decreased more with increasing dose, hence lumbar spine began to grow more
slowly than thoracic spine at high doses. This condition applied in ratio scale as well (Figure
15). The intersection of two regression lines for thoracic and lumbar spinal sections
suggested the effects of dose were different by spine locations (p<0.0001 for difference scale,
p=0.0001 for ratio scale).

18

DISCUSSION/ CONCLUSION
This study evaluated the effect of radiation on vertebral body height growth, and how the
effect was influenced by treatment age, radiation dosage, gender and location in the spine.
In this sample, 34 neuroblastoma children were treated with dose sculpting IMRT at CHLA
from years 2000 - 2011, with a total of 133 spinal sections. The participants in this study
were 56% male and 44% female, which was consistent with the male predominance (Irwin
& Park, 2015). Median treatment age was 4.4 years, which is reasonable as most children
are diagnosed before age 4 years (Irwin & Park, 2015).
We found that radiation dosage, treatment age and gender significantly influenced the
vertebral height growth rate, whether evaluated using absolute difference or difference ratio.
However, the height growth rate in ratio was not significantly different based on spinal
location, which may be because the size of lumbar spine vertebral bodies are larger than
thoracic spine, therefore it was easier to have larger increase in absolute difference. Similarly,
age at treatment was close to non-significant in absolute difference, which may because the
spinal section of younger children was smaller, resulting in a more significant increase in
ratio than in difference.
Age at treatment was associated with lower growth in our study -- younger children had
higher growth, especially age less than 4 years. However, radiation effect is higher for
younger age when measurement was on the percent scale. Though the effects of age were
contrary between studies (Hartley et al., 2008; Probert & Parker, 1975; Yu et al., 2015), in
general lower age appeared to have higher susceptibility to the effects of radiation dose.
Therefore, minimizing radiation dosage is more important when treating younger children.
19

Radiation dosage has a strong negative effect on height growth in both difference and ratio
level. We can see the growth rate of target dose group was lower than spared and control
group, with dose range for target group 17.7 - 34.1 Gy, with 23.6 Gy in average. The result
was consistent with previous studies, which indicated dosage higher than 17.5 - 23.4 Gy was
associated with spinal growth abnormality (Hartley et al., 2008; Paulino & Fowler, 2005).
Females tend to grow faster than males from our study. In contrast, a previous study
indicated that females grow slower in lower thoracic and lumbar spine (T4 - T7 and L1 - L5)
but found no difference when considering single vertebral bodies (T4 and L3) (Hartley et al.,
2008). It is possibly because the previous study estimated growth in different segment groups,
while we estimated growth in height per vertebral body. In addition, the age distribution was
larger in the previous study which may also lead to different result, as we know puberty may
affect male/ female growth in different age. A spinal growth study also supported that female
height growth was higher than male during childhood (Dimeglio & Canavese, 2012).
Spinal location is an important factor to predict vertebral height growth in absolute
difference. Our finding that the lumbar spine grows faster than thoracic spine is consistent
with other studies (Hartley et al., 2008). That lumbar spine is more affected by radiation
dosage by an order of magnitude was also found in a previous study (Hartley et al., 2008).
This study estimated the associations based on a linear relationship. Since our study
participants were mostly distributed at age 2 to 5 years (68%), their sitting height growth
(Dimeglio & Canavese, 2012; Probert & Parker, 1975) supported our assumption - follow-
up years and age were linearly associated with vertebral height growth rate. We reviewed
the relationship between vertebral height growth and radiation dosage in Figure 13, which
20

helped us to confirm the linear relationship assumption. We examined the relationship with
treatment age and radiation dosage in quadratic term, though this was not significant and
therefore excluded.
This study has provided evidence that vertebral height growth effects are multifactorial, that
age at treatment, gender, radiation dosage and spinal location moderate with follow-up years
on height growth rate. This is the first study investigating the effect of factors for vertebral
height growth of neuroblastoma in children with dose sculpting IMRT. Though this study
only contained 34 subjects, the sample size is reasonable considering that the incidence rate
of neuroblastoma is rare and it is an internal control study with multiple measurements for a
subject. Since the participants were limited to high risk neuroblastoma group, the result may
not represent low to intermediate risk group, but those patients would not be treated by
radiation in general. In addition, the doses tend to be clustered in both the spared dose and
target dose groups, so that there were only few doses in the low ranges. Therefore, it is
difficult to see if there is a threshold dose of radiation below which there is no effect on
vertebral height growth. About three quarters of subjects had age at follow-up less than age
13 years, so they may not have experienced their pubertal spurt. Hence, there is a need for
further study to explore long-term effects after puberty. At the end, though this study did not
find evidence of a quadratic association of age and dose on height growth, further research
is encouraged to explore more possibilities of a non-linear relationship.
In conclusion, radiation dose, age at treatment and gender significantly affect vertebral
height growth. Higher dose, older treatment age and male sex lead to lower growth. Lumbar
spine has higher absolute height growth (cm). Younger age and lumbar spine have higher
21

susceptibility to the effect of radiation dose. These models support that dose sculpting IMRT
techniques are helpful to reduce growth deformities by sparing the volume of bone receiving
high dose, and hopefully can provide an estimate to assist caregivers in predicting the
vertebral height growth impairment.

22

REFERENCE
Cohn, S. L., Pearson, A. D., London, W. B., Monclair, T., Ambros, P. F., Brodeur, G. M.,
Matthay, K. K. (2009). The International Neuroblastoma Risk Group (INRG)
classification system: an INRG Task Force report. J Clin Oncol, 27(2), 289-297.
Dimeglio, A., & Canavese, F. (2012). The growing spine: how spinal deformities influence
normal spine and thoracic cage growth. Eur Spine J, 21(1), 64-70.
Dorr, W., Kallfels, S., & Herrmann, T. (2013). Late bone and soft tissue sequelae of
childhood radiotherapy. Relevance of treatment age and radiation dose in 146
children treated between 1970 and 1997. Strahlenther Onkol, 189(7), 529-534.
Halton, J. M., Atkinson, S. A., Fraher, L., Webber, C. E., Cockshott, W. P., Tam, C., & Barr,
R. D. (1995). Mineral homeostasis and bone mass at diagnosis in children with acute
lymphoblastic leukemia. J Pediatr, 126(4), 557-564.
Hartley, K. A., Li, C., Laningham, F. H., Krasin, M. J., Xiong, X., & Merchant, T. E. (2008).
Vertebral body growth after craniospinal irradiation. Int J Radiat Oncol Biol Phys,
70(5), 1343-1349.
Irwin, M. S., & Park, J. R. (2015). Neuroblastoma: paradigm for precision medicine. Pediatr
Clin North Am, 62(1), 225-256.
Ng, L., Wong, K., & Olch, A. (2015). Dose Sculpting Intensity Modulated Radiation
Therapy for Vertebral Body Sparing in Treatment of Neuroblastoma. Poster session
presented at the American Society for Radiation Oncology, San Antonio, TX.
Paulino, A. C., & Fowler, B. Z. (2005). Risk factors for scoliosis in children with
neuroblastoma. Int J Radiat Oncol Biol Phys, 61(3), 865-869.
Probert, J. C., & Parker, B. R. (1975). The effects of radiation therapy on bone growth.
Radiology, 114(1), 155-162.
Ries, L. S., MA. Gurney, JG. Linet, M. Tamra, T. Young, JL. Bunin, GR. (1999). Cancer
Incidence and Survival among Children and Adolescents: United States SEER
Program 1975-1995. National Cancer Institute, SEER Program: Bethesda (MD):
NIH Pub.
Stiller, C. A., & Parkin, D. M. (1992). International variations in the incidence of
neuroblastoma. Int J Cancer, 52(4), 538-543.
Whittle, S. B., Smith, V., Doherty, E., Zhao, S., McCarty, S., & Zage, P. E. (2017). Overview
and recent advances in the treatment of neuroblastoma. Expert Rev Anticancer Ther.
Yu, J. I., Lim, D. H., Jung, S. H., Sung, K. W., Yoo, S. Y., & Nam, H. (2015). The effects
of radiation therapy on height and spine MRI characteristics in children with
neuroblastoma. Radiother Oncol, 114(3), 384-388.
Zeger, S. L., & Liang, K. Y. (1986). Longitudinal data analysis for discrete and continuous
outcomes. Biometrics, 42(1), 121-130.

23

TABLES

Table 1. Characteristics of Participants
N (%) Mean SD Min Max
By Participant
Age at Treatment (Years) 34 (100) 5.0 3.0 1.8 15.8
Age at Follow-up (Years) 34 (100) 10.6 3.6 5.5 20.9
Follow-up Years 34 (100) 5.5 2.2 3.3 10.7
Gender 34 (100) - - - -

Male 19 (56) - - - -

Female 15 (44) - - - -
Alive at Present 34 (100) - - - -

Yes 27 (79) - - - -

No 5 (15) - - - -
Unknown 2 (6) - - - -
By Measurement Site

Radiation Mean Dose 133 (100) 18.5 6.7 6.4 34.1

Target 48 (36) 23.6 4.6 17.7 34.1

Spared 44 (33) 13.0 3.5 6.4 23.3

Control 41 (31) 0.0 0.0 0.0 0.0
Spine 133 (100) - - - -

Thoracic 69 (52) - - - -
Lumbar 64 (48) - - - -
A participant has multiple measurement sites.
The mean and standard deviation estimates for overall mean dose were excluded control group.

24

Table 2. The Relationship of Predictors with Growth Rate in Absolute Difference
β 95% CI p-value
Follow-up Years 0.0526 0.0391, 0.0662 <0.0001***
Age at Treatment (Years) -0.0062 -0.0178, 0.0055 0.30
Radiation Mean Dose -0.0061 -0.0082, -0.0041 <0.0001***
Gender

Male (ref) - -
Female 0.0330 -0.0673, 0.1333 0.52
Location of Spine

Thoracic (ref) - -
Lumbar 0.0853 0.0487, 0.1219 <0.0001***
* p <0.05, ** p<0.01, *** p<0.001.

25

Table 3. The Relationship of Growth in Absolute Difference and Follow-up Years Accounted for Interactions
Model 1 Model 2 Model 3

β 95% CI P-value β 95% CI P-value β 95% CI P-value
Intercept -0.0109 -0.1313,0.0399 0.80 -0.0430 -0.1279,0.0419 0.32 -0.0254 -0.1050,0.0543 0.53
Follow-up Years 0.0478 0.0391,0.0662 <0.0001*** 0.0548 0.0380,0.0716 <0.0001*** 0.0626 0.0517,0.0735 <0.0001***
FU_Yrs*Age - - - -0.0005 -0.0022,0.0012 0.53 - - -
FU_Yrs*Dose - - - - - - -0.0011 -0.0015,-0.0007 <0.0001***
FU_Yrs*Gender(F) - - - - - - - - -
FU_Yrs*Spine(L) - - - - - - - - -

Model 4 Model 5

β 95% CI P-value β 95% CI P-value

Intercept -0.0491 -0.1291,0.0308 0.23 -0.0523 -0.1367,0.0320 0.22

Follow-up Years 0.0482 0.0359,0.0606 <0.0001*** 0.0477 0.0344,0.0609 <0.0001***

FU_Yrs*Age - - - - - -

FU_Yrs*Dose - - - - - -

FU_Yrs*Gender(F) 0.0119 0.0034,0.0204 0.01* - - -

FU_Yrs*Spine(L) - - - 0.0131 0.0054,0.0208 0.0008***

* p <0.05, ** p<0.01, *** p<0.001.

Model 1: Included intercept and follow-up years.

Model 2-5: Included intercept, follow-up years and each interaction between follow-up years and other covariates.

Gender presented for female; Spine presented for lumbar spine.

26

Table 4. The Relationship of Growth in Absolute Difference and Follow-up Years Accounted for All Potential Interactions
Model 1 Final Model

β 95% CI P-value β 95% CI P-value

Intercept -0.0261 -0.1002,0.0481 0.49 - - -

Follow-up Years 0.0601 0.0496,0.0733 <0.0001*** 0.0566 0.0477,0.0654 <0.0001***

FU_Yrs*Age -0.0015 -0.0029,-0.0001 0.04* -0.0015 -0.0030,-0.0001 0.04*

FU_Yrs*Dose -0.0012 -0.0016,-0.0009 <0.0001*** -0.0012 -0.0016,-0.0009 <0.0001***

FU_Yrs*Gender(F) 0.0098 0.0030,0.0167 0.005** 0.0085 0.0026,0.0168 0.007**

FU_Yrs*Spine(L) 0.0159 0.0093,0.0225 <0.0001*** 0.0159 0.0092,0.0225 <0.0001***

* p <0.05, ** p<0.01, *** p<0.001.

Numbers were rounded up to four decimals for β and two decimals for p-value.

Model 1: Included intercept, follow-up years and the interactions between follow-up years and other covariates.

Final Model: Included follow-up years and the significant interactions between follow-up years and other covariates in model 1.

27

Table 5. The Relationship of Predictors with Growth Rate in Difference Ratio
β 95% CI p-value
Follow-up Years 0.0329 0.0238,0.0420 <0.0001***
Age at Treatment (Years) -0.0102 -0.0168,-0.0036 0.002**
Radiation Mean Dose -0.0046 -0.0061,-0.0031 <0.0001***
Gender

Male (ref) - -
Female 0.0223 -0.0426,0.0872 0.50
Location of Spine

Thoracic (ref) - -
Lumbar 0.0185 -0.0068,0.0438 0.15
* p <0.05, ** p<0.01, *** p<0.001.

28

Table 6. The Relationship of Growth in Difference Ratio and Follow-up Years Accounted by Interaction

Model 1 Model 2 Model 3

β 95% CI P-value β 95% CI P-value β 95% CI P-value
Intercept -0.0309 -0.0867,0.0249 0.28 -0.0226 -0.0765,0.0314 0.41 -0.0147 -0.0699,0.0406 0.60
Follow-up Years 0.0329 0.0238,0.0420 <0.0001*** 0.0396 0.0299,0.0494 <0.0001*** 0.0408 0.0313,0.0504 <0.0001***
FU_Yrs*Age - - - -0.0016 -0.0025,-0.0008 0.0001*** - - -
FU_Yrs*Dose - - - - - - -0.0009 -0.0011,-0.0006 <0.0001***
FU_Yrs*Gender(F) - - - - - - - - -
FU_Yrs*Spine(L) - - - - - - - - -

Model 4 Model 5

β 95% CI P-value β 95% CI P-value

Intercept -0.0327 -0.0863,0.0208 0.23 -0.0317 -0.087,0.0237 0.26

Follow-up Years 0.0304 0.0210,0.0399 <0.0001 0.0323 0.0227,0.0419 <0.0001***

FU_Yrs*Age - - - - - -

FU_Yrs*Dose - - - - - -

FU_Yrs*Gender(F) 0.0068 -0.0000,0.0135 0.05 - - -

FU_Yrs*Spine(L) - - - 0.0016 -0.0037,0.0070 0.55

* p <0.05, ** p<0.01, *** p<0.001.

Model 1: Included intercept and follow-up years.

Model 2-5: Included intercept, follow-up years and each interaction between follow-up years and other covariates.

Gender presented for female; Spine presented for Lumbar spine.

29

Table 7. The Relationship of Growth in Difference Ratio and Follow-up Years Accounted for All Potential Interactions
Model 1 Model 2 Final Model

β 95% CI P-value β 95% CI P-value β 95% CI P-value
Intercept -0.0053 -0.0531,0.0426 0.83 -0.0042 -0.0525,0.0442 0.86 - - -
Follow-up Years 0.0475 0.0378,0.0572 <0.0001*** 0.0486 0.0395,0.0576 <0.0001*** 0.0480 0.0410,0.0549 <0.0001***
FU_Yrs*Age -0.0023 -0.0032,-0.0015 <0.0001*** -0.0023 -0.0031,-0.0014 <0.0001*** -0.0023 -0.0031,-0.0014 <0.0001***
FU_Yrs*Dose -0.0009 -0.0012,-0.0007 <0.0001*** -0.0009 -0.0012,-0.0006 <0.0001*** -0.0009 -0.0012,-0.0007 <0.0001***
FU_Yrs*Gender(F) 0.0052 0.0003,0.0100 0.04* 0.0053 0.0004,0.0101 0.04* 0.0052 0.0003,0.0102 0.04*
FU_Yrs*Spine(L) 0.0038 -0.0007,0.0083 0.09 - - - - - -
* p <0.05, ** p<0.01, *** p<0.001.

Numbers were rounded up to four decimals for β and two decimals for p-value.

Model 1: Included intercept, follow-up years and the interactions between follow-up years and other covariates.
Model 2: Included intercept, follow-up years and the significant interactions between follow-up years and other covariates in model 1.
Final Model: Included follow-up years and the significant interactions between follow-up years and other covariates in model 1.

30

FIGURES

Figure 1. Regression Plot of Height Growth in Absolute Difference on Follow-up Years, Interacts by Age.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age,
radiation mean dose, gender and spine (refer to table 4). The regression line was evaluated at mean age of
each group, and overall mean value of radiation dose, gender and spine.
0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
0.6
0.8
Vertebral Body Heigth Growth in Absolute Difference (cm/body)
>=4 <4
p-value: 0.0401
31

Figure 2. Regression Plot of Height Growth in Absolute Difference on Follow-up Years, Interacts by Radiation Dosage.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation dose, gender
and spine (refer to table 4). The regression line was evaluated at mean radiation dose of each group, and overall mean value of
age, gender and spine.
0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
0.6
0.8
Vertebral Body Heigth Growth in Absolute Difference (cm/body)
Target Spared Control
p-value: <.0001
32

Figure 3. Regression Plot of Height Growth in Absolute Difference on Follow-up Years, Interacts by Gender.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation
dose, gender and spine (refer to table 4). The regression line was evaluated at overall mean value of age, radiation
dose and spine.
0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
0.6
0.8
Vertebral Body Heigth Growth in Absolute Difference (cm/body)
F M
p-value: 0.0074
33

Figure 4. Regression Plot of Height Growth in Absolute Difference on Follow-up Years, Interacts by Spine.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation
dose, gender and spine (refer to table 4). The regression line was evaluated at overall mean value of age, radiation
dose and gender.
0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
0.6
0.8
Vertebral Body Heigth Growth in Absolute Difference (cm/body)
L T
p-value: <.0001
34

Figure 5. Regression Plot of Height Growth in Absolute Difference on Follow-up Years, Interacts by Age and Radiation Dose.

The regression line estimated by GEE multivariate model of follow-up years, interactions with age, radiation dose, gender and spine,
and a three-way interaction with age and radiation dose. The regression line was evaluated at overall mean value of gender and spine.

Target Spared Control
3 way interaction p-value: 0.0525
Follow-up Years
0 2 4 6 8 10 0 2 4 6 8 10
0.0
0.2
0.4
0.6
0.8
Vertebral Body Heigth Growth in Difference (cm/body)
Age Group = >=4 Age Group = <4
35

Figure 6. Regression Plot of Height Growth in Absolute Difference on Follow-up Years, Interacts by Spine and Radiation Dose.

The regression line estimated by GEE multivariate model of follow-up years, interactions with age, radiation dose, gender and spine,
and a three-way interaction with age and radiation dose. The regression line was evaluated at overall mean value of gender and spine.

Target Spared Control
3 way interaction p-value: <.0001
Follow-up Years
0 2 4 6 8 10 0 2 4 6 8 10
0.0
0.2
0.4
0.6
0.8
Vertebral Body Heigth Growth in Difference (cm/body)
SPINE = L SPINE = T
36

Figure 7. Regression Plot of Height Growth in Difference Ratio on Follow-up Years, Interacts by Age.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age,
radiation dose, gender and spine (refer to table 4). The regression line was evaluated at plugged in mean age
of each group, and overall mean value of radiation dose, gender and spine.

0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
Vertebral Body Heigth Growth in Difference Ratio
>=4 <4
p-value: <.0001
37

Figure 8. Regression Plot of Height Growth in Difference Ratio on Follow-up Years, Interacts by Radiation Dose.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation dose,
gender and spine (refer to table 4). The regression line was evaluated at mean dose of each radiation group, and overall
mean value of age, gender and spine.

0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
Vertebral Body Heigth Growth in Difference Ratio
Target Spared Control
p-value: <.0001
38

Figure 9. Regression Plot of Height Growth in Difference Ratio on Follow-up Years, Interacts by Gender.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation
mean dose, gender and spine (refer to table 4). The regression line was evaluated at overall mean value of age,
radiation dose, and spine.

0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
Vertebral Body Heigth Growth in Difference Ratio
F M
p-value: 0.0383
39

Figure 10. Regression Plot of Height Growth in Difference Ratio on Follow-up Years, Interacts by Spine.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation
mean dose, gender and spine (refer to table 4). The regression line was evaluated at overall mean value of age,
radiation dose and gender.

0 2 4 6 8 10
Follow-up Years
0.0
0.2
0.4
Vertebral Body Heigth Growth in Difference Ratio
L T
p-value: 0.1006
40

Figure 11. Regression Plot of Height Growth in Difference Ratio on Follow-up Years, Interacts by Age and Radiation Dose.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation mean dose, gender
and spine (refer to table 4). The regression line was evaluated at overall mean value of spine and gender.

Target Spared Control
3 way interaction p-value: 0.0008
Follow-up Years
0 2 4 6 8 10 0 2 4 6 8 10
0.0
0.2
0.4
Vertebral Body Heigth Growth in Ratio
Age Group = >=4 Age Group = <4
41

Figure 12. Regression Plot of Height Growth in Difference Ratio on Follow-up Years, Interacts by Spine and Radiation Dose.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation mean dose, gender
and spine (refer to table 4). The regression line was evaluated at overall mean value of age and gender.

Target Spared Control
3 way interaction p-value: 0.0001
Follow-up Years
0 2 4 6 8 10 0 2 4 6 8 10
0.0
0.2
0.4
0.6
Vertebral Body Heigth Growth in Ratio
SPINE = L SPINE = T
42

Figure 13. Scatter Plot of Height Growth in 10 Years Difference on Radiation Dose.

The loess line drew by radiation mean dose that provided the line of best fit.

43

Figure 14. Regression Plot of 10-Year Height Growth in Absolute Difference on Radiation Dose, Interacts by Spine.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation mean dose,
gender, spine and three-way interaction of spine and radiation dose. The regression line was evaluated at overall mean value
of age and gender.

0 10 20 30
Radiation Mean Dose (Gy)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
Spinal Height Growth in Difference per 10 years(cm/body/10yrs)
L T
p-value: <.0001
44

Figure 15. Regression Plot of 10-Year Height Growth in Difference Ratio on Radiation Dose, Interacts by Spine.

The regression line estimated by GEE multivariate model of follow-up years and interactions with age, radiation mean
dose, gender, spine and three-way interaction of spine and radiation dose. The regression line was evaluated at overall
mean value of age and gender.

0 10 20 30
Radiation Mean Dose (Gy)
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Spinal Height Growth in Ratio per 10 years
L T
p-value: 0.0001

Abstract (if available)

Abstract Background: Children with neuroblastoma who are treated with radiotherapy have increased incidence of vertebral body growth deformities. Intensity-modulated radiation therapy (IMRT) has been adopted to attempt to reduce the exposure of normal tissues to the toxicities of radiation therapy, with the expectation of reducing growth inhibition without sacrificing tumor coverage. This study aimed to estimate the influence of radiotherapy and other clinical factors—age at treatment, gender, vertebral spine location—on vertebral body height growth for patients who had childhood neuroblastoma treated with dose sculpting IMRT. ❧ Methods: A retrospective review was performed on children with neuroblastoma treated with radiation therapy at Children’s Hospital Los Angeles (CHLA) from year 2000-2011, with a minimum of 3 years of follow-up imaging. Vertebral body height was measured prior to and several years after radiation in the thoracic and lumbar spinal section for each patient. Delivered radiation dose was also recorded for these spinal sections. Vertebral body height growth was defined in two ways: vertebral body height growth in absolute difference (cm) and in difference ratio (%). We used generalized estimating equations (GEE) models to evaluate the associations between vertebral height growth and follow-up years, and the influence of age at treatment, gender, mean dose of radiation, and location of radiated spine on growth rate. ❧ Results: Data were available on 34 participants with 56% male and 44% female, mean age at treatment of 5 years old, and average length of follow-up of 5.5 years. Of 133 vertebral bodies, 48 (36%), 44 (33%), and 41 (31%) were designated as target, spared and control respectively. Target vertebral body radiation dose was 23.6 Gray (Gy) on average whereas spared dose was 13.0 Gy. For absolute difference (cm), older age at treatment, higher radiation mean dose, male sex and thoracic spine location were significantly associated with lower growth rate. The vertebral body height grew 0.0536 cm per body per year for the thoracic spine of male children aged two years old in the control group. An increase of one year of age at time of irradiation significantly decreased vertebral body height growth by 0.0015 cm/ body/ year (p=0.04)

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Survival of children and adolescents with low-risk non-rhabdomyosarcoma soft tissue sarcomas (NRSTS) treated with surgery only: an analysis of 234 patients from the Children’s Oncology Group stud...

PDF

The evaluation of the long-term effectiveness of zero/low fluoroscopy workflow in ablation procedures for the treatment of paroxysmal and persistent atrial fibrillation

Asset Metadata

Creator Wu, Chia-Ling (author)

Core Title Long-term effects on vertebral body height growth of dose sculpting intensity modulated radiation therapy for children with neuroblastoma

School Keck School of Medicine

Degree Master of Science

Degree Program Applied Biostatistics and Epidemiology

Publication Date 04/14/2017

Defense Date 03/30/2017

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag children spinal growth,dose sculpting intensity modulated radiation therapy,IMRT,long-term effect,neuroblastoma,OAI-PMH Harvest,vertebral body height growth

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Sposto, Richard (committee chair), Barkauskas, Donald (committee member), Olch, Arthur (committee member)

Creator Email chialing.ally.wu@gmail.com,wuchiali@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-353452

Unique identifier UC11255907

Identifier etd-WuChiaLing-5183.pdf (filename),usctheses-c40-353452 (legacy record id)

Legacy Identifier etd-WuChiaLing-5183.pdf

Dmrecord 353452

Document Type Thesis

Rights Wu, Chia-Ling

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA