You are on page 1of 6

Int. J. Cancer (Radiat. Oncol. Invest): 96 (Suppl.

) 125–130 (2001) Publication of the International Union Against Cancer

© 2002 Wiley-Liss, Inc.

Doses Near the Surface during Total-body

Irradiation with 15 MV X-Rays
A. Kassaee, * Y. Xiao, Ph.D.,2 P. Bloch, Ph.D.,1 J. Goldwein, M.D.,1
D.I. Rosenthal,1 and B.E. Bjärngard, Ph.D.1
Department of Radiation Oncology, Hospital of the University of Pennsylvania,
Philadelphia, Pennsylvania
Department of Radiation Oncology, Thomas Jefferson University, Philadelphia,

SUMMARY An extended source–surface distance (SSD) is often employed in total-body

irradiation (TBI) to obtain fields sufficiently large to encompass the total body. It is
clinically desirable to administer a uniform dose to all tissues, including the surface,
especially to counteract the skin-sparing of high-energy X-rays. The electrons generated
from the air contribute more to the dose near the surface at extended distances than at
regular SSDs. However, if further increase of the dose at shallow depths is desired a
spoiler can be employed. Monte Carlo simulations were used to study these effects for the
electron component of the beam and the parameters needed for the pencil-beam dose
calculation of this component. Measurements were performed to confirm these studies.
The energy spectrum of electrons generated in air is independent of SSD. All the electrons
in the beam, however, originate from the spoiler when one is used. The spectrum of these
electrons is nearly independent of the SSD. The pencil-beam dose kernels were calculated
using the derived electron spectra. To calculate doses in the buildup region with and
without a spoiler, we use one pencil-beam kernel for the electrons generated in air and one
for the electrons generated from the spoiler. Measured depth-dose data agreed with the
calculation results. © 2002 Wiley-Liss, Inc.

Key words: surface dose; extended SSD; TBI; pencil-beam kernel

INTRODUCTION small children. This dose depends on the source–

surface distance (SSD), field size, and the photon-
Total-body irradiation (TBI) is used for the purpose of beam energy [2]. It can be increased with a beam
achieving total malignant cell kill and for immuno- spoiler [3– 6], the effects of which are determined by
suppression. The technique is being used with in- the atomic number, the thickness of the spoiler, field
creasing frequency as a component of preparatory size, and the spoiler-to-surface distance (STSD)
regimens for bone-marrow reconstitution of patients [6 – 8]. The electrons generated in the spoiler have a
with refractory malignancies, including leukemia, wide angular distribution and modify the dose at
non-Hodgkin’s lymphoma, and neuroblastoma [1]. In shallow depths both inside and outside the photon
order to attain uniform dose throughout the whole beam [9,10]. Previous studies of spoilers have been
body, high-energy photon beams are often employed. limited to SSDs of approximately 100 cm [6 – 8,11],
Sometimes it is necessary to increase the dose to the which we here extend to the long SSDs used for TBI.
superficial tissues to kill the circulating cells close to For short SSDs, 80 –100 cm, the electrons gen-
the skin surface, especially for thin patients and for erated in air contribute little to the dose near the

*Correspondence to: Alireza Kassaee, Ph.D., Department of Radiation Oncology, 3400 Spruce Street, 2 Donner,
Philadelphia, PA 19104. Phone: (215) 662-6204; Fax: (215) 349-5978; E-mail:
Received 7 May 2001; Revised 21 September 2001; Accepted 2 October 2001
DOI 10.1002/ijc.10349
Published online 14 March 2002 in Wiley InterScience (
126 Kassaee et al.: Dose Near Surface for TBI

The electron fluence and spectrum in water were

evaluated on the central axis of the beam for the three
geometries shown in Figure 1. The variation of the
electron energy fluence with distance from the central
ray was determined with air and vacuum between the
lead plate and the water and, in the case of air, with
and without the spoiler. The sampling volumes were
1 cm high circular annuli with 5-cm radial increments.
The Monte Carlo simulations employed approxi-
mately 8 ⫻ 106 histories, with photon and electron
cutoffs of 0.01 and 0.1 MeV, respectively.
Fig. 1. Geometrical configuration for Monte Carlo calcu- With the electron spectrum derived, a pencil-
lations. The spoiler consists of a 1-cm thick water-equiva-
lent plate. Source–surface distance (SSD) ⫽ 100, 300, 500 beam energy-deposition kernel for the electron com-
cm, and spoiler-to-surface distance (STSD) ⫽ 1, 30, 100 ponent in the photon beam was calculated. The kernel
cm. was incorporated in a previously described dose-cal-
culation algorithm [11]. This kernel was used to cal-
culate the depth dose in the superficial region due to
electrons generated in air or in spoiler for similar
surface compared to those originating in the acceler- geometries.
ator head [12–14]. At longer SSDs, the relative con-
tribution from electrons generated in air increases Measurements
with increasing irradiated air volume and ultimately The dose was measured at shallow depths with a
dominates as the head-generated electrons are lost parallel-plate ionization chamber (Capintec Model
[14]. PS-033) having 0.5 cm3 air volume and 0.5 mg/cm2
In this study, our objective was to evaluate the window thickness. The collimator setting (field size
electron component at long SSDs generated in air and 100 cm from the source) was 40 ⫻ 40 cm2 A solid-
from a spoiler. We focused on the following ques- water phantom (30 ⫻ 30 cm2 entrance area, 15 cm
tions: 1) Is a spoiler necessary or are the “contami- depth) was surrounded with acrylic plates for a total
nating” electrons produced in the head and in air volume of 100 ⫻ 50 ⫻ 30 cm2 Depth dose was
sufficient? 2) How can one easily determine a kernel measured for a 15-MV photon beam for SSDs of 300
for the incident electrons to be used for dose calcula- and 500 cm, with and without a beam spoiler. The
tions? STSD was 10, 20, or 30 cm. The ion-chamber re-
sponse was corrected for the secondary electrons scat-
MATERIALS AND METHODS tered from the side-wall of the chamber [17] as de-
Monte Carlo Studies scribed by Gerbi and Khan [18]. This correction was
approximately 2% at the surface and 0.3% at a 7-mm
A 15-MV X-ray beam, produced by a Varian 2100
C/D linear accelerator, was selected for this study.
The photon spectrum of this beam had been deter- RESULTS
mined previously by Monte Carlo simulations using
the manufacturer’s design of the target and flattening Monte Carlo Simulations
filter [11]. From this spectrum, the spectrum and the Electrons Generated in Air
fluence of the contaminating electrons were calculated The calculated electron spectra 500 cm from the
using the cylindrical option Cyltran of the Integrated source in vacuum and air are shown in Figure 2. The
Tiger Series (ITS3/PC) of coupled electron-photon sources of these “contaminating” electrons are both
Monte Carlo transport code [15,16]. The geometry the lead plate (simulating the flattening filter) and the
shown in Figure 1 was employed, with the following intervening air. The figure demonstrates that electrons
parameters: 1) the circular field has a radius of 50 cm generated in air shift the spectrum towards lower
at 500 cm from the source; 2) a 0.5-cm thick lead energies. Additional calculations indicated that the
plate positioned 10 cm from the X-ray target is used electron spectrum does not change with distance for
to simulate a source of “contaminating electrons” in SSD greater than 300 cm. The electron fluence times
the accelerator head; 3) the SSD is 100 cm, 300 cm, the square of the distance from the lead plate is shown
or 500 cm, and this volume is either air or vacuum; 4) in Figure 3 for intervening air and vacuum. In the
a spoiler consisting of 1 cm of water-equivalent ma- latter case, the electron fluence obeys the inverse
terial is positioned at 1, 30, or 100 cm (STSDs) from square law, with the lead plate being a well-defined
the water phantom. source of electrons. However, in air the electron
Kassaee et al.: Dose Near Surface for TBI 127

Fig. 2. Monte Carlo generated electron spectra in air and Fig. 4. Electron spectrum as a function of distance from
in vacuum 500 cm from the source. the 1-cm thick spoiler with spoiler-to-surface distances
(STSD) of 1, 30, 100 cm without the phantom.

Fig. 3. Electron fluence ⫻ L2 in air (circles) and in Fig. 5. Calculated depth dose, normalized to 5 mm depth,
vacuum (squares) as a function of distance L from the lead for the contaminating-electron component using the con-
plate used to simulate the treatment head. The error bars taminating-electron kernels at 1 (circles) and 30 cm
represent the standard deviation in the Monte Carlo simu- (squares) from the spoiler.
lations. The normalization point is at 1 cm from the lead
ergetic electrons corresponded to the electron spectra
at the phantom surface, with STSDs of 1 and 30 cm.
fluence exceeds that from the lead, as electrons are The kernel was incorporated in a superposition dose-
generated along the air path. calculation algorithm [11] and then was used to cal-
Variation of Electron Spectrum with Distances culate the depth dose for the contaminating electrons.
Monte Carlo results showed no significant change of Figure 5, in which the doses are normalized to 0.5 cm
the contaminating electron spectrum as a function of depth, demonstrates that the depth dose of the elec-
SSD with the spoiler in the beam. Thus, the electrons trons is nearly independent of distance from the
generated in the accelerator head and the air are spoiler for depths greater than 3 mm.
virtually completely absorbed in the spoiler, which Electron Fluence Profile
produces the entire electrons incident on the phantom. The in-air electron fluence profiles with or without a
The electron spectra at the surface of a phantom with spoiler, shown in Figure 6, were calculated with the
500 cm SSD and STSD of 1, 30, and 100 cm are Monte Carlo program, with air between the source
shown in Figure 4. and the phantom with an SSD of 500 cm. For calcu-
An energy-deposition kernel for the electrons lation with the spoiler, the STSD was 30 cm and the
from the spoiler at a fixed 500-cm SSD was generated electron fluence profiles are calculated in air at 1 and
by taking a weighted linear combination of kernels for 5 cm from the spoiler. Contaminating electrons from
monoenergetic electrons. The weights of the monoen- the accelerator head were neglected by removing the
128 Kassaee et al.: Dose Near Surface for TBI

Fig. 7. Measured depth doses for open beams (symbols,

Fig. 6. Variation of in-air electron fluence from central dashed curves) at source–surface distance (SSD) ⫽ 100 cm
ray at 5 cm from a spoiler located 400 cm away from the (triangles), 300 cm (squares), and 500 cm (diamonds). Mea-
source (solid line and symbols) and without the spoiler sured depth doses with a spoiler at spoiler-to-surface dis-
(solid line). tance (STSD) ⫽ 30 cm (solid curve), SSD ⫽ 300 cm
(squares), and 500 cm (not shown, overlaying the one for
SSD ⫽ 300 cm).

lead plate in the beam for the Monte Carlo simula-

tions. The profile 5 cm from the spoiler is shown in
Figure 6, together with the profile without the spoiler
at the same distance from the source. When the
spoiler is used, the profile of contaminating electrons
is flatter inside the beam, with a sharper fall-off at the
edges. These are consequences of the angular spread
of the electrons: when the spoiler is used the electrons
from it have little opportunity to diverge and the
characteristics of the flat, sharply collimated photon
beam are reflected by the electron fluence. Corre-
spondingly, the penumbra broadens with increasing
distance from the spoiler. For example, at 1 and 5 cm
from the spoiler, the penumbra region (90 –10%) is
14.3 cm and 16.3 cm, respectively. Fig. 8. Measured depth doses (upper curves) for 500 cm
The contaminating electrons have limited range source–surface distance for open beam (diamonds) and
and they influence the dose distribution only near the spoiled beam with spoiler-to-surface distance (STSD) ⫽ 30
surface. At 3-cm depth in water, the approximate cm (triangles) and 10 cm (squares), normalized to 6 cm
depth, and the difference in depth dose (lower curves)
depth of dose maximum, the profiles from photons between the open and spoiled beams with STSD ⫽ 30 cm
alone and with electrons generated in air and the (triangles) and 10 cm (squares).
spoiler were the same within ⫾1%. Hence, although
some contaminating electrons may reach the depth of
increase with decreasing spoiler-phantom distance,
dose maximum, they do not influence the dose profile
but these changes do not extend to depths greater than
at this or greater depths.
0.3 cm. Also shown in the figure is the difference in
Results of Measurement the depth doses between the open field and with the
The measured depth doses in the buildup region for an spoiler in place. At depths greater than 3 mm, the
irradiated area of 100 ⫻ 50 cm2 are shown in Figure differences with STSD are insignificant.
7 for 300 and 500 cm distance to the source with and
without the spoiler 30 cm above the phantom surface. DISCUSSION
The depth doses are normalized to a 6-cm depth. For In an X-ray beam without a spoiler, the contaminating
an open field, the dose at the surface and in the electrons are originated in the accelerator head (here,
buildup region increases at longer SSDs but is nearly the lead plate) and in the irradiated air. The electrons
independent of the SSD with a spoiler. Figure 8 shows from the accelerator head are degraded in energy by
that the measured depth doses in the buildup region the intervening air (Fig. 2). The number of electrons
Kassaee et al.: Dose Near Surface for TBI 129

As shown in Figure 7, 500 cm (0.65 g/cm2) of air

is not sufficient to achieve full electron equilibrium
for a 15-MV photon beam, although the fractional
surface dose increases significantly at the extended
SSDs. If further increase in surface dose is desired, a
spoiler can be employed. This absorbs the electrons
generated in the accelerator head or in air but pro-
duces new ones that, if the spoiler–skin distance is not
Fig. 9. Energy-deposition kernels for electrons generated too large (ⱕ30 cm), effectively provide electron equi-
from air (left) and electrons generated from the spoiler librium at the surface.
(right). The curves are for fractions of energy deposited. Can we extrapolate the findings to beams of
other energies? To answer this question the same
generated in air increases simultaneously (Fig. 3) and Monte Carlo analysis was performed for a 6-MV
at an SSD of 500 cm there is 3.4 times as many X-ray beam. Depth doses in the water phantom were
electrons generated in air as from the simulated ac- obtained for an air column and with a spoiler in the
celerator head for a 15-MV beam. In Figure 7, the beam with similar geometries used for the 15-MV
measured depth dose at the surface at 500 cm SSD is beam. The results were similar to those obtained for
nearly twice that at 100 cm. Since air-generated elec- the 15-MV beam with surface percent dose 80% and
trons dominate at extended SSD, the spectrum of the 95% without and with the spoiler, respectively, for an
electrons is nearly independent of the distance from SSD of 500 cm and an STSD of 30 cm. In general, the
the source beyond 300 cm from the source. Thus, a dose due to contaminating electrons emanating from
single pencil-beam kernel, shown in Figure 9, can be the spoiler depends on the clinical photon beam en-
used for calculating the dose due to electrons gener- ergy, STSD field size, and spoiler thickness.
ated in air regardless of the distance from the source
When a spoiler is inserted in the beam, most of CONCLUSION
the electrons originated in the accelerator head are In TBI treatment with high-energy photons without a
absorbed in or scattered away by the spoiler [4,7,8]. spoiler at typical distances of 300 –500 cm from the
Figure 7 shows that 1 cm water-equivalent material source, the air-generated electrons are a dominant
eliminates nearly all electrons produced upstream. source of contaminated electrons. The dose in the
Instead, the electrons generated by photon interac- buildup region at these extended distances are signif-
tions in the spoiler appear in the beam. The electron- icantly greater than the standard treatment distances
fluence characteristics downstream of the spoiler are of 80 –100 cm. At SDD of 300 cm and greater, the
approximately independent of the distance to the electron spectrum is nearly independent of distances
source, as shown for the electron spectrum (Fig. 4). and could be represented with a single pencil-beam
Therefore, this permits the use of a single pencil-beam kernel for dose calculations. Depending on the desired
kernel to calculate doses for these electrons. This dose in the superficial tissues, these electrons gener-
kernel is shown in Figure 9. ated in air may be sufficient enhancement to satisfy
In Figure 4, there is a larger fraction of low- the dose prescription requirements. However, if full
energy electrons present near the spoiler. These low- dose relative to dose at the depth of maximum is
energy electrons dissipate energy and are lost to the required, the use of a beam spoiler is warranted. With
beam at larger spoiler phantom distances. Electrons a beam spoiler, the dose in the buildup region is
from the spoiler have limited range and affect the dose increased, depending on the STSD, field size, beam
only close to the surface, down to the depth of max- energy, and spoiler thickness and material. Use of a
imum dose (Fig. 7). beam spoiler in the beam with a thickness of about
The broad, roughly Gaussian directional distri- half of the electron range effectively stops the elec-
bution of the electrons generated in air and the accel- trons generated in air. The spoiler becomes the new
erator head manifests itself at shallow depths as a source of contaminating electrons. At the typical
large dose-profile variation inside the beam and a STSDs of 10 –30 cm, the electrons generated within
broadened penumbra [1] (Fig. 6). When a spoiler is the spoilers traveling downstream are nearly indepen-
introduced these electrons are absorbed, resulting in dent of STSD and therefore a single pencil-beam
the flat profile and the sharper beam edge that just kernel for the contaminating electrons component
below the spoiler characterize the collimated X-ray may be used for dose calculations. This approxima-
beam. Further away from the spoiler, however, the tion results in errors of less than 1% for depths greater
penumbra width at shallow depths is larger due to the than 3 mm (Fig. 5). If the dose very close to the
wide angular spread of electrons from the spoiler. surface needs to be precisely known, the contaminat-
130 Kassaee et al.: Dose Near Surface for TBI

ing electron kernel needs to be calculated using finer 8. Fontenla D, Napoli J, Hunt M, Fass D, McCormick B,
energy bins down to 0.2 MV and finer depth incre- Kutcher G. Effects of beam modifiers and immobiliza-
ments less than 1 mm. tion devices on the dose in the build-up region. Int J
A spoiler increases the dose at the surface by a Radiat Oncol Biol Phys 1994;30:211–219.
significant amount, depending on the distance be- 9. Scrimger J, Kolitsi Z. Scattered radiation from beam
tween the spoiler and surface. The dose profile at the modifiers used with megavoltage therapy units. Radi-
ology 1979;130:233–236.
phantom surface with a spoiler is flatter inside the
10. Shiraz S, Kase K. Measurements of dose from second-
beam and sharper at the edges compared to that from
ary radiation outside a treatment field: effects of wedges
air-generated electrons. The profile at the surface at and blocks. Int J Radiat Oncol Biol Phys 1985;11:
extended SSD depends on SSD and spoiler–phantom 2171–2176.
distance, which needs to be measured for clinical 11. McKenna MG, Chen XG, Altschuler MD, Bloch P.
beams, especially when matching fields are used. Calculation of the dose in the build-up region for high
REFERENCES energy photon beam. Treatment planning when beam
spoilers are employed. Radiother Oncol 1995;34:63–
1. Galvin JM, D’Angio GJ, Walsh G. Use of tissue com- 68.
pensators to improve the dose uniformity for total body 12. Sjögren R, Karlsson M. Electron contamination in clin-
irradiation. Int J Radiat Oncol Biol Phys 1980;6:767–
ical high energy photon beams. Med Phys 1996;23:
2. Mackie TR, Scrimger JW. Contamination of a 15-MV
13. Yorke ED, Ling CC, Rustgi S. Air-generated electron
photon beam by electrons and scattered photons. Radi-
contamination of 4 and 10 MV photon beams: a com-
ology 1982;144:403– 409.
parison of theory and experiment. Phys Med Biol 1985;
3. Planskoy B, Tapper PD, Bedford AM, Davis FM. Phys-
ical aspects of total-body irradiation at the Middlesex
Hospital (UCL group of hospitals), London 1988-1993. 14. Petti PL, Goodman MS, Sisterson JM, Biggs PJ, Gab-
Phys Med Biol 1996;41:2327–2343. riel TA, Mohan R. Sources of electron contamination
4. Nilsson B. Electron contamination from different ma- for the Clinac-35 25-MV photon beam. Med Phys
terials in high energy photon beams. Phys Med Biol 1983;10:856 – 861.
1985;30:139 –151. 15. Mehlhorn TA, Young MF. UPEML—a machine-porta-
5. Chu JCH, Coia LR, Aziz D, Stafford PM. Dose to ble CDC update emulator. SAND 1984:84:1896.
superficial node for patients with head and neck cancer 16. Halbleib JA, Kensek RP, Mehlhorn TA. The integrated
treated with 6 MV and 60 Co photons. Radiother Oncol TIGER series of coupled electron/photon Monte Carlo
1991;21:257–260. transport codes. SAND 1984:84:573.
6. Kubo H, Russell MD, Wang CC. Use of 10 MV spoiled 17. Nilsson B, Montelius A. Fluence perturbation in photon
x-ray beam for treatment of head and neck tumors. Int beams under non-equilibrium conditions. Med Phys
J Radiat Oncol Biol Phys 1982;8:1795–1798. 1986;13:191–195.
7. Khan F, Moore V, Levitt S. Effects of various atomic 18. Gerbi BJ, Khan FM. Measurement of dose in the
number absorbers on skin dose for 10 MV X-rays. buildup region using fixed-separation plane-parallel
Radiology 1973;109:209 –212. ionization chambers. Med Phys 1990;17:17–26.