Paper
SECONDARY NEUTRON DOSES TO PEDIATRIC PATIENTS DURING
INTRACRANIAL PROTON THERAPY: MONTE CARLO SIMULATION
OF THE NEUTRON ENERGY SPECTRUM AND ITS ORGAN DOSES
Shinnosuke Matsumoto,*† Yusuke Koba,† Ryosuke Kohno,‡ Choonsik Lee,§
Wesley E. Bolch,** and Michiaki Kai*
Abstract—Proton therapy has the physical advantage of a Bragg
peak that can provide a better dose distribution than conventional
x-ray therapy. However, radiation exposure of normal tissues
cannot be ignored because it is likely to increase the risk of sec-
ondary cancer. Evaluating secondary neutrons generated by
the interaction of the proton beam with the treatment beam-line
structure is necessary; thus, performing the optimization of ra-
diation protection in proton therapy is required. In this research,
the organ dose and energy spectrum were calculated from sec-
ondary neutrons using Monte Carlo simulations. The Monte
Carlo code known as the Particle and Heavy Ion Transport
code System (PHITS) was used to simulate the transport pro-
ton and its interaction with the treatment beam-line structure
that modeled the double scattering body of the treatment nozzle
at the National Cancer Center Hospital East. The doses of the
organs in a hybrid computational phantom simulating a 5‐y-old
boy were calculated. In general, secondary neutron doses were
found to decrease with increasing distance to the treatment field.
Secondary neutron energy spectra were characterized by inci-
dent neutrons with three energy peaks: 110−7, 1, and 100 MeV.
A block collimator and a patient collimator contributed signifi-
cantly to organ doses. In particular, the secondary neutrons from
the patient collimator were 30 times higher than those from the
first scatter. These results suggested that proactive protection will
be required in the design of the treatment beam-line structures
and that organ doses from secondary neutrons may be able to
be reduced.
Health Phys. 110(4):380–386; 2016
*Graduate school, Oita University of Nursing and Health Sciences.
Oita city, Oita 870-1201, Japan; †Medical Exposure Research Project,
National Institute of Radiological Sciences. Chiba city, Chiba 263-8555,
Japan; ‡Division of Particle Therapy, National Cancer Center Hospital
East. Kashiwa city, Chiba 277-8577, Japan; §Division of Cancer Epi-
demiology and Genetics, National Cancer Institute, National Institute of
Health, Rockville, MD 20850, USA; **Department of Radiology, Univer-
sity of Florida, Gainesville, FL 32611, USA.
For correspondence contact: Shinnosuke Matsumoto, National Insti-
tute of Radiological Sciences at 4‐9‐1 Anagawa, Inage-ku, Chiba-shi,
Chiba 263‐8555, Japan, or email at shimatsu@nirs.go.jp.
(Manuscript accepted 29 October 2015)
0017-9078/16/0
Copyright © 2016 Health Physics Society
DOI: 10.1097/HP.0000000000000461
380
Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.
Key Words: dose, organ; Monte Carlo; pediatrics; radiation
therapy
INTRODUCTION
THE 5-Y survival rate of pediatric cancers has reached around
80% due to the recent improvement of multimodality ther-
apy. Radiation therapy plays an important role in multi-
modality therapy and is useful for the treatment of many
types of pediatric cancer. Sparing normal tissue is important
in pediatric radiation therapy because cancer risk in children
is higher than in adults, and children’s smaller statures
put their nontargeted organs in closer proximity to the
therapeutic fields.
Particle therapy has the physical advantage of a Bragg
peak, providing small lateral dispersion in biological tissues.
Therefore, particle therapy can provide a better dose dis-
tribution than conventional x-ray therapy (Bonnett, 1993;
Sisterson 1995; Olsen et al. 2007). In particular, consider-
able clinical research on proton therapy has indicated the
possibility of limiting the long-term side effects in pediatric
intracranial tumors (St. Clair et al. 2004; Kirsch and Tarbell
2004). Proton therapy will become more important and
common in the future.
However, secondary neutrons are generated by the
nuclear reaction of the primary proton beam with structures
in the proton beam line (Yan et al. 2002). Neutron exposure
of normal tissues cannot be ignored because neutrons have
a high biological effect in the form of an increased risk of
secondary cancer.
It is necessary to evaluate the secondary neutrons
generated by interaction of the treatment beam-line struc-
ture with the proton beam and, subsequently, to compare
these with other proton therapy facilities. This would pro-
vide basic information for optimizing the dose to patients
during proton therapy.
www.health-physics.com
 Doses during intracranial proton therapy c S. MATSUMOTO ET AL. 381

Fig. 1. Schematic view of the beam-line port at NCCHE. Simplified nozzle modeled in PHITS.

The purpose of this paper was to analyze and compare
organ doses from secondary neutrons received by pediatric
patients treated at the National Cancer Center Hospital East
(NCCHE) for intracranial tumors using a Monte Carlo sim-
ulation code, namely the Particle and Heavy Ion Transport
code System Version 2.7.6 (PHITS). This study used the
hybrid computational phantoms that were developed under
collaboration between the University of Florida and the
U.S. National Cancer Institute (Lee et al. 2007, 2010). The
absorbed dose and equivalent dose to each organ from
secondary neutrons were calculated for the nontargeted
regions located outside a clinical target volume. The con-
tribution rate of an organ dose to a nontargeted region from
each proton beam-line structure was also calculated. Finally,
the neutron-equivalent dose in NCCHE was compared with
other facilities in published studies. These results can pro-
vide useful information for optimizing shield materials and
the thicknesses of beam-line-components.
MATERIALS AND METHODS
Beam-line at NCCHE
The arrangement of the beam-line in port 2 of the
NCCHE is shown in Fig. 1. The double-scattering method,
which is a type of passive irradiation method, was used to
produce a uniform lateral dose distribution (Takada 1994).
There were three beam-limiting devices just ahead of pa-
tients: a pre-ring collimator (RC), a block collimator (BC),
and a patient-specific collimator (PC). The RC, comprising
50‐mm-thick brass with a 150‐mm diameter, was located
just before the ridge filter to prevent the leakage of primary
and secondary particles and to prevent the activation of the
ridge filter caused by the primary and secondary charged
particles. The BC comprised two pairs of one-dimensional
collimators of 50‐mm-thick brass; the PC comprised 5‐cm-
thick brass and was used as a final collimator. Three colli-
mators could be moved together along the beam axis to
make the lateral penumbra smaller by approaching the
patient. To obtain a uniform depth-dose distribution within
the spread-out Bragg peak (SOBP), a ridge filter (RF) com-
prised of aluminum was used. A range-shifter compris-
ing 10 Polymethyl methacrylate plates of various thicknesses
and a bolus made of polyethylene was used to compare
the distal end of the SOBP to that of the target volume.
The beam-line system was equipped with three identical
parallel-plate ionization chambers that acted as both pri-
mary and secondary beam monitors of the proton beam-
line (Fig. 2).
Monte Carlo simulation code
The general-purpose PHITS 2.7.6 (Sato et al. 2013)
was used in this study for the following reasons: PHITS
provides good agreement with the measurements of the
energy spectra and angular distributions of neutrons pro-
duced by reactions of materials with comparable high-Z
and proton beams. PHITS can transport neutrons from

Fig. 2. Schematic overview of the modeled treatment apparatus and computational phantom. The view is oriented for the left lateral field.
www.health-physics.com

Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.
382 Health Physics
thermal energy up to 200 GeV. The evaluated nuclear data
libraries were used for neutron transport and collision pro-
cesses in the low-energy region in the same way as MCNP,
whereas Bertini and/or JAM (Nara et al. 1999) models were
used for an intracascade in the high-energy region. For
protons and other hadrons, JAM was used from 1 MeV to
200 GeV, and only the ionization process was considered
below 1 MeV. In the ionization process of charged parti-
cles and nuclei, the SPAR code (Armstrong et al. 1973)
was used for the average stopping power dE/dx, the first
order of the Moliere model for the angle straggling, and
the Gaussian, Landau, and Vavilov theories for the energy
straggling around the average energy loss according to
the charge density and velocity.
Computational phantom
Monte Carlo calculations of secondary neutron doses
to organs required the use of realistic computational phantoms
representative of the treated patients, namely 5‐ to 10‐y-old
children, who comprise the age peak incidence for intra-
cranial tumors. Therefore, a 5‐y-old phantom (UFH5M)
was adopted out of the series of 12 hybrid phantoms (Lee
et al. 2007, 2010) that were developed at the University of
Florida (UF) and the U.S. National Cancer Institute (NCI).
These phantoms had the same anatomical and physiological
characteristics of reference individuals reported by ICRP Pub-
lication 89 (ICRP 2002). The UFH5M was converted for the
Monte Carlo code to PHITS through voxel format fitting.
Treatment plan of proton therapy
This study focused on child patients suffering from an
ependymoma of the cerebellum treated with proton therapy.
The standard ependymoma treatment plan delivers a total
Relative Biological Effectiveness (RBE)-weighted dose
of 49 Gy, where RBE of a proton is 1.1 to the tumor. The
dose calculation was performed under the conditions of
a 189 MeV proton beam, 9 cm SOBP, and opposing portal
irradiation. All these elements were modeled on the basis
of manufacturers and engineering drawings. The opening
sizes of each collimator were 150 mm diameter (RC), 200 
200 mm2 (BC), and 30  30 mm2 (PC). The bolus had a cu-
boid shape (height: 3 cm, length: 10 cm, width: 10 cm)
with a 2‐cm-deep concave of 2 cm square.
Monte Carlo calculations
Neutron-absorbed doses to organs were calculated in
the UHF5M computational phantom. This study collected
the deposited energy of the secondary charged particles
generated by stray neutrons to estimate the absorbed doses.
Therefore, the neutron-absorbed doses were based on aver-
age energy depositions in each organ. Internal and external
neutrons were not distinguished from one another. Previous
studies have clearly demonstrated that external neutrons
www.health-physics.com
Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.
April 2016, Volume 110, Number 4
dominated in most secondary neutron dose situations
(Zacharatou et al. 2008).
The neutron-equivalent dose (HT) was also calculated
for each organ by multiplying the neutron-absorbed dose
by the radiation-weighting factor (WR). The WR was deter-
mined according to the incident-neutron spectra for the
phantom using a continuous function in neutron energy
for the calculation of WR for neutrons (ICRP 2007).
Nontarget regions were defined as other organs and
tissues besides brain and bone. PHITS could identify the
location in the beam-line structure where secondary neu-
trons were generated by proton beams. Using this function,
the authors calculated how high the contribution rate was
to nontarget regions from secondary neutrons.
In order to identify how to reduce the secondary neu-
tron doses, neutron energy spectra generated from each
beam-line structure were calculated for neutrons incident
upon the computational phantom.
Comparison with other medical facilities
The neutron-equivalent doses in NCCE were compared
with the neutron-equivalent doses in 12 organs published by
Sayah et al. (2014) and Zacharatou et al. (2008).
Sayah et al. (2014) calculated neutron equivalent
doses for a 5‐y-old female patient treated for intracranial
tumors with five fields using a 178‐MeV proton beam. The
authors used the same phantom series (UFH05F) and a dif-
ferent Monte Carlo simulation code, MCNPX (Hendricks
2006). The authors calculated the secondary neutron doses
at the Curie Institut-Orsay (Orsay, France) proton therapy
center (ICPO).
Zacharatou et al. (2008) calculated the dose to an 8‐y-old
female patient treated for intracranial tumors using pro-
tons with 169.2–180.1 MeV. The authors used a different
phantom series [University of Florida B-series voxel phan-
toms (Lee 2006)] and a different Monte Carlo simulation
code, Geant4. The authors calculated the doses at the F. H.
Burr Proton Therapy Center at the Massachusetts General
Hospital (MGH).
RESULTS
Secondary neutron-absorbed dose
The neutron-absorbed doses calculated for the organs
of the phantom are presented in Fig. 3. The graph also
shows which organ absorbed relatively larger doses dur-
ing proton therapy. According to this result, the mean
neutron absorbed dose reached a maximum of 0.579 mGy
Gy−1 (RBE) for the organs considered in this study.
Secondary neutron-equivalent dose
The neutron-weighting factors, WR, were calculated
for the phantom using the neutron spectrum simulated with
PHITS. Using these energy spectra, the energy-dependent
WR value in the phantom calculated according to ICRP
 Doses during intracranial proton therapy c S. MATSUMOTO ET AL. 383

Fig. 3. Secondary neutron absorbed doses [mGy Gy−1(RBE)] calculated for various organs. The doses are normalized per the therapeutic dose.

103 (2007) was 11.34. The neutron equivalent doses calcu- than that of the first scatter. The ring collimator had relatively
lated for the organs of the phantom are presented in Table 1. little effect on the secondary neutron organ doses.

Contribution rate to non-target regions Secondary neutron energy spectra

The contribution rates to nontarget regions are shown
in Fig. 4. According to these results, the patient collimator
made a significant contribution to the secondary neutron or-
gan doses. The contribution of the patient collimator to the
secondary neutron-organ doses was about 30 times higher
Table 1. Neutron-equivalent doses per therapeutic dose (mSv/Gy).
The results are obtained from Monte Carlo simulations of a cerebellum
tumor using the voxel 313 computational 5-year-old boy phantom.
Organs and tissues Neutron equivalent doses (mSv/Gy)
The secondary neutron energy spectra calculated for
the whole body of the phantom are shown in Fig. 5. The
two peaks observable in the spectra corresponded to evapo-
ration and nuclear cascade neutrons. There existed little dif-
ference between the shapes of all energy spectra and the
energy spectra from each component. The relative errors
in the energy spectra of secondary neutrons caused by the

Pharynx, nose 6.57

Thyroid 6.44
Salivary glands 6.26
Esophagus 4.63
Lungs 4.10
Larynx 3.96
Heart 3.14
Gallbladder 2.95
Liver 2.60
Breast 2.44
Stomach 1.90
Spleen 1.84
Kidneys 1.61
Pancreas 1.38
Colon 1.39
Testes 1.28
Small intestine 1.22
Prostate 0.91
Urinary bladder 0.86 Fig. 4. Contribution rate of each structure to the secondary neutron
organ dose.
www.health-physics.com

Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.
384 Health Physics
Fig. 5. Secondary neutron energy spectra calculated for the whole
body of the phantom. The spectra are normalized to the proton thera-
peutic dose.
Monte Carlo simulation were about 25%, 20%, 20%, 10%,
8%, and 5% in first scatter, second scatter, RC, fine-
degrader, BC and PC, respectively.
Comparison with other medical facilities
The results of the calculation were compared with the
neutron-equivalent doses published by Sayah et al. (2014)
and Zacharatou et al. (2008). Fig. 6 compares the neutron-
equivalent doses at three medical facilities averaged over
all treatment fields for 12 organs. The secondary neutron
equivalent doses to the thyroid region showed a difference
of about 1.4 to 3.5 times among the facilities.
Fig. 6. Comparison of secondary equivalent doses in intracranial tumors at three facilities: the NCCHE (Black), MGH (Diagonal strokes), and
ICPO (White).
www.health-physics.com
Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.
April 2016, Volume 110, Number 4
DISCUSSION
The results for the neutron-absorbed dose to the non-
target region showed that the closer to the target, the higher
are the doses received, and that the closer to a patient, the
higher is the contribution to organ doses. In particular, the
head and neck regions received high secondary-neutron
doses because these regions were close to the patient and
block collimators in the intracranial tumor. These results
could explain why the patient and block collimators made
a significant contribution to secondary neutron-organ doses.
A high proportion of the neutrons were of 1 MeVenergy,
which had a large radiation-weighting factor. From these
results, practical dose reduction could be accomplished by
shielding the 1 MeV neutrons arising from the patient
and block collimators. All of the energy spectra had sim-
ilar shapes. Thus, these results could indicate a reduction of
about 50% of the secondary neutrons generated from the
block and patient collimators. The doses to the head-neck
region could be reduced by about 44%.
From these comparisons, the neutron equivalent doses
in NCCHE were higher than those at other facilities. Yonai
et al. (2008) reported that the secondary neutron ambient
dose equivalent at NCCHE was higher than in other facil-
ities. This report showed that the neutron ambient dose
equivalent in NCCHE was about 1.2 to 1.8 times higher
at a 50 cm distance from the isocenter than in other facili-
ties. The high neutron dose in NCCHE was likely to be
mainly because of the widespread angle of the primary pro-
ton beam at the beam-emitting window, as the patient colli-
mator played a role in the shaping of the therapeutic field.
To verify the angle dependence of the secondary neutron
doses, the neutron absorbed doses were calculated by alter-
ing the spread angles of the primary proton beam to values
 Doses during intracranial proton therapy c S. MATSUMOTO ET AL.
Fig. 7. Angular dependence of the secondary neutron dose. The
head–neck region includes the eyes, salivary glands, ET regions, oral
cavity, and thyroid. The chest region includes the esophagus, lungs,
breast, thymus, and heart. The upper abdomen region includes the
liver, gallbladder, stomach, pancreas, spleen, and adrenal glands.
The lower abdomen region includes the kidneys, urinary bladder,
small intestine, prostate, colon, and testes.
of 0.0°, 0.5°, 1.0°, 1.5°, and 1.8° from the default setting.
The results are shown in Fig. 7.
The secondary neutron dose decreased by 17%, 56%,
and 77% in the head-neck region when the proton emittance
angle was 1.5°, 1.0°, and 0.5°, respectively. The change in
dose was calculated compared with the dose at 1.8° as the
default setting. It was found that the proton emission angle
contributed largely to secondary neutron doses. It is likely
that the large angle of the primary proton beam had an in-
tense effect on increasing the secondary neutron dose in
NCCE. Hence, secondary neutron dose reduction could
be achieved by means of a shield against the secondary
neutron from the patient collimator and also by control-
ling the spread angle of the primary proton beams. The
statistical error by the Monte Carlo simulation was less
than 10% in most calculations. The systematic error would
be negligibly small because the proton beam-line was accu-
rately modeled on the basis of original drawings.
CONCLUSION
This paper analyzed the secondary neutron doses to
pediatric patients during intracranial proton therapy using
a Monte Carlo method. The results were compared with
published data. This study indicated that the patient colli-
mator contributed to the secondary neutron doses and
also that the proton emission angle had an intense effect.
Thus, this study suggested that a significant reduction of
www.health-physics.com
Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.
385
the secondary neutron doses could be achieved by optimiz-
ing the proton beam-line structure.
REFERENCES
Armstrong TW, Chandler KC. SPAR, a FORTRAN program for
computing stopping powers and ranges for muons, charged
pions, protons and heavy ions. Oak Ridge, TN: Oak Ridge Na-
tional Laboratory; ORNL‐4869; 1973.
Bonnett DE. Current developments in proton therapy: a review.
Phys Med Biol 38:1371–1392; 1993. DOI 10.1088/0031‐9155/
38/10/001.
Hall EJ, Phil D. Intensity-modulated radiation therapy, protons, and
the risk of second cancers cancer rates. Int J Radiat Oncol Biol
Phys 65:1–7; 2006.
Hendricks JS, McKinney GW, Durkee JW, Finch JP, Fensin ML,
James MR, Johns RC, Pelowitz DB, Waters LS. MCNPX Re-
port LA-UR-06-7991. Los Alamos, NM: Los Alamos National
Laboratory; 2006.
International Commission on Radiological Protection. 2007 rec-
ommendations of the International Commission on Radiologi-
cal Protection. Oxford: Pergamon; ICRP Publication 103; Ann
of the ICRP. 37(2–4); 2007.
International Commission on Radiological Protection. Basic ana-
tomical and physiological data for use in radiological protection.
Oxford: Pergamon; ICRP Publication 89; Ann of the ICRP, 32
(3–4); 2002.
Kirsch DG, Tarbell NJ. Conformal radiation therapy for childhood
CNS tumors. Oncologist 9:442–450; 2004. DOI 10.1634/
theoncologist.9‐4‐442.
Lee C, Lodwick D, Hasenauer D, Williams JL, Lee C, Bolch WE.
Hybrid computational phantoms of the male and female
newborn patient: NURBS-based whole-body models. Phys
Med Biol 52:3309–3333; 2007. DOI 10.1088/0031‐9155/
52/12/001.
Lee C, Lodwick D, Hurtado J, Pafundi D, Williams JL, Bolch WE.
The UF family of reference hybrid phantoms for computational
radiation dosimetry. Phys Med Biol 55:339–363; 2010. DOI
10.1088/0031‐9155/55/2/002.
Lee C, Williams JL, Bolch WE. Whole-body voxel phantoms of
paediatric patients—UF Series B. Phys. Med. Biol. 51:4649–
4661; 2006.
Nara Y, Otuka N, Ohnishi A, Niita K, Chiba S. Relativistic
nuclear collisions at 10A GeV energies from p+ Be to Au+
Au with the hadronic cascade model. Phys Rev C 61:
024901; 1999.
Olsen DR, Bruland ØS, Frykholm G, Norderhaug IN. Pro-
ton therapy—a systematic review of clinical effectiveness.
Radiother Oncol 83:123–132; 2007. DOI 10.1016/j.radonc.
2007.03.001.
Sato T, Niita K, Matsuda N, Hashimoto S, Iwamoto Y, Noda S,
Ogawa T, Iwase H, Nakashima H, Fukahori T, Okumura K,
Kai T, Chiba S, Furuta T, Silver L. Particle and heavy ion trans-
port code system. J Nucl Sci Technol 50:913-923; 2013. DOI
10.1080/00223131.2013.814553.
Sayah R, Farah J, Donadille L. Secondary neutron doses received
by paediatric patients during intracranial proton therapy treat-
ments. J Radiol Prot 34:279–296; 2014. DOI 10.1088/0952‐
4746/34/2/279.
Sisterson JM. Proton radiation therapy: a summary of the world
wide experience. Nucl Instr Meth Phys Res 99:827–829;
1995. DOI 10.1016/0168-583X(94)00598‐2.
St. Clair WH, Adams JA, Bues M, Fullerton BC, La Shell S,
Kooy HM, Loefler JS, Tarbell NJ. Advantage of protons
386 Health Physics April 2016, Volume 110, Number 4

compared to conventional x-ray or IMRT in the treatment of
a pediatric patient with medulloblastoma. Int J Radiat
Oncol Biol Phys 58:727–734; 2004. DOI 10.1016/S0360‐
3016(03)01574‐8.
Takada Y. Dual-ring double scattering method for proton beam
spreading. Jpn J Appl Phys 33:353–359; 1994. DOI 10.1143/
JJAP.33.353.
Yan X, Titt U, Koehler A, Newhauser W. Measurement of neutron
dose equivalent to proton therapy patients outside of the proton
radiation field. Nucl Instr Meth Phys Res 476:429–434; 2002.
DOI 10.1016/S0168‐9002(01)01483‐8.
Yonai S, Matsufuji N, Kanai T, Matsui Y, Matsushita K, Yamashita
H, Numano M, Sakae T, Terunuma T, Hishio T, Kohno R,
Akagi T. Measurement of neutron ambient dose equivalent
in passive carbon-ion and proton radiotherapies. Med Phys
35:4782; 2008. DOI 10.1118/1.2989019.
Zacharatou JC, Lee C, Bolch WE, Xu XG, Paganetti H. Assessment
of organ-specific neutron equivalent doses in proton therapy
using computational whole-body age-dependent voxel phan-
toms. Phys Med Biol 53:693–717; 2008. DOI 10.1088/0031‐
9155/53/3/012.
■■

www.health-physics.com

Copyright © 2016 Health Physics Society. Unauthorized reproduction of this article is prohibited.