1. RADIATION RESEARCH 164, 212–220 (2005)
0033-7587/05 $15.00
2005 by Radiation Research Society.
All rights of reproduction in any form reserved.
Epithermal Neutron Beams for Clinical Studies of Boron Neutron
Capture Therapy: A Dosimetric Comparison of Seven Beams
P. J. Binns,a,1 K. J. Rileya and O. K. Harlingb
a
Nuclear Reactor Laboratory and b Department of Nuclear Science and Engineering, Massachusetts Institute of Technology, 138 Albany Street,
Cambridge, Massachusetts 02139
cade using newly designed beams of epithermal neutrons
Binns, P. J., Riley, K. J. and Harling, O. K. Epithermal for irradiating deep-seated lesions. The use of these higher-
Neutron Beams for Clinical Studies of Boron Neutron Cap-
ture Therapy: A Dosimetric Comparison of Seven Beams. Ra-
energy epithermal neutrons attempts to ensure that satisfac-
diat. Res. 164, 212–220 (2005). tory dose distributions of thermal neutrons are achieved at
depth without requiring surgical intervention to reflect the
A comparison of seven epithermal neutron beams used in
clinical studies of boron neutron capture therapy (BNCT) in scalp or remove part of the skull. Thus far the studies have
Sweden (Studsvik), Finland (Espoo), Czech Republic (ReZ), been restricted to dose-finding Phase I/II trials that seek first
The Netherlands (Petten) and the U.S. (Brookhaven and Cam- to establish normal tissue tolerance with a view thereafter
bridge) was performed to facilitate sharing of preclinical and to demonstrate some therapeutic effect (1–3). As the dif-
clinical results. The physical performance of each beam was ferent programs conduct the extramural research necessary
measured using a common dosimetry method under condi- to develop a successful paradigm and accrue individual
tions pertinent to brain irradiations. Neutron fluence and ab- clinical experience, the need will arise to compare and pos-
sorbed dose measurements were performed with activation
sibly share results among the different participants. This
foils and paired ionization chambers on the central axis both
in air and in an ellipsoidal water phantom. The overall quality imperative will reduce duplication of effort while hopefully
of each beam was assessed by figures of merit determined expediting optimization and determination of the possible
from the total weighted dose profiles that assumed the pres- efficacy for the modality.
ence of boron in tissue. The in-air specific beam contamination The description of the physical characteristics of an ep-
from both fast neutrons and rays ranged between 8 and 65 ithermal neutron beam is a prerequisite for any exchange
10 11 cGy(w) cm2 for the different beams and the epither- of either experimental or clinical results between different
mal neutron flux intensities available at the patient position facilities. This, however, is complicated by the differences
differed by more than a factor of 20 from 0.2–4.3 109 n
cm 2 s 1. Percentage depth dose profiles measured in-phantom
in beam designs and an absence of a standardized meth-
for the individual photon, thermal and fast-neutron dose com- odology of appraisal. Fission reactors are currently the
ponents differed only subtly in shape between facilities. As- source of choice for clinical use that employ beam lines of
suming uptake characteristics consistent with the currently various lengths and configurations to direct neutrons to the
used boronated phenylalanine, all the epithermal beams ex- patient’s position (4). These beam lines contain different
hibit a useful penetration of 8 cm or greater that is sufficient filter moderator assemblies that moderate and shape the fis-
to irradiate a lesion seated at the brain midline. The perfor- sion energy spectrum of the source beam to the desired
mance of the existing facilities will benefit from the introduc-
epithermal range. Both photon shielding and collimation
tion of advanced compounds through improved beam pene-
trability. This could increase by as much as 2 cm for the pur- are used to remove unwanted background contaminants and
est of beams, although the beam intensities generally need to to reduce beam spreading. No single, ideal neutron energy
be increased to between 2–5 109 n cm 2 s 1 to maintain exists, and the facilities, because of practical considerations,
manageable irradiation times. These data provide the first all produce a range of epithermal neutron energies. The
consistent measurement of beam performance at the different neutron energy spectrum, field size and degree of colli-
centers and will enable a preliminary normalization of the mation all affect beam performance, and these differ be-
calculated patient dosimetry. 2005 by Radiation Research Society
tween facilities. In addition, the beams are invariably con-
taminated with photons and fast neutrons that are detri-
INTRODUCTION mental to the depth–dose characteristics, unnecessarily in-
creasing the dose to healthy tissue. The minimization of
Clinical research into boron neutron capture therapy
these contaminants is achieved with differing degrees of
(BNCT) for brain cancer has advanced during the last de-
success at the different facilities.
1
Address for correspondence: Nuclear Reactor Laboratory, 138 Albany An adequate description of the physical characteristics of
Street, Cambridge, MA 02139; e-mail: binns@mit.edu. an epithermal beam for experimental as well as therapeutic
212

2. A COMPARISON OF SEVEN EPITHERMAL BEAMS FOR BNCT 213
purposes requires an in-phantom assessment of the three that are based upon outputs of beam monitors such as fission counters to
principal radiation constituents contributing to the dose, indicate the relative radiation fluence during the experiment.
Measurements were performed in each beam on the central axis both
namely the thermal neutron flux, which is used to determine in-air and in an ellipsoidal phantom of total volume of 2.50 liters that
the dose from neutron capture in boron and other nuclei as was filled with distilled water (13). The geometry of the phantom is based
well as the fast-neutron and the photon absorbed dose com- upon the Snyder brain model and has outer dimensions of 13.6 (X), 19.6
ponents. These physical characteristics vary depending (Y) and 16.6 cm (Z) for the axes of the three major ellipses. The phantom
upon the geometry and material of the phantom in which was placed in the collimated fields with the X (13.6 cm) axis parallel to
the beam and aligned with the central axis for each measurement using
they are measured. Since there is no standard protocol for the patient positioning lasers. The phantom was positioned in the usual
mixed-field dosimetry in epithermal therapy beams, differ- treatment position, that was, for most facilities, without an air gap be-
ent methods using active and passive dosimeters are cur- tween the apex of the ellipsoid and the plane that forms the end of the
rently used by the various research groups (5). The methods collimator. At the highly collimated HFR, however, it was placed 30 cm
all have comparatively large uncertainties by general radio- downstream from the collimator exit. In-air measurements were per-
formed with the detectors similarly positioned in the center of the field
therapy standards because of the complexities of interpret- and either held in a retort stand or affixed across the collimator aperture
ing dosimeter responses to the individual radiation com- with Mylar tape.
ponents. These together with differences in the assumed Thermal and epithermal neutron fluxes were determined using the cad-
tissue compositions for which absorbed dose is specified, mium difference technique with 0.05-mm-thick gold foils and 0.51-mm-
the use of different elemental kerma coefficients, and def- thick cadmium covers (6). The foils were irradiated in runs between 0.5
and 16.1 h long for which the incident epithermal neutron fluence was
initions of thermal neutron flux all preclude a simple eval- in the range of 3–19 1012 n cm 2. Foils were counted within 24 h of
uation of the data routinely reported between the various the end of bombardment at the host institute using an HPGe detector with
groups. an energy and efficiency calibration traceable to a European Standards
To afford a direct comparison of the dosimetric proper- Laboratory. Irradiated foils from Studsvik, Espoo and Petten were also
ties of the various BNCT research facilities currently avail- counted at MIT as a further test of consistency, using an HPGe detector
calibrated with a mixed radionuclide source traceable to the National
able in Europe and the U.S., the four different European Institute of Standards and Technology (NIST). Fast-neutron and photon
facilities presently conducting clinical trials were appraised absorbed dose rates were measured using the dual-chamber technique
in quick succession using a common experimental method with separate A-150 and graphite walled ionization chambers that have
with a single set of uncertainties. Beams in Sweden (Studs- identical active volumes of 0.1 ml (8). Flow meters and gas mixtures
vik), Finland (Espoo), Czech Republic (ReZ) and The were provided by each of the host laboratories, and chamber responses
were corrected to standard environmental conditions using calibrated ther-
Netherlands (Petten) were characterized under reference mometers and barometers.
conditions pertinent to clinical irradiations using the dosim- The absorbed dose rates arising from thermal neutron capture in boron
etry procedures developed and adopted by the Massachu- and nitrogen were calculated from measurements of the 2200 m s 1 neu-
setts Institute of Technology (MIT). These studies comple- tron flux using bare and cadmium-covered gold activation foils (7, 14).
ment those performed previously in the U.S. at MIT, Cam- The determined neutron flux was multiplied by the weight fraction of
boron or nitrogen for the tissue or material of interest and kerma coef-
bridge (M67) (6), Brookhaven (BMRR) (7) and more re- ficients of 8.66 10 6 and 7.88 10 10 cGy cm2 for boron and nitrogen,
cently the new MIT fission converter beam (FCB) (8). The respectively. The energy-dependent responses of the two ionization cham-
measured depth–dose data were used to determine various bers were converted to absorbed dose using values of the various physical
figures of merit assessing the performance of each beam parameters described recently (8). Neutron energy spectra from the Eu-
under uniform experimental conditions and provide, for the ropean beams were compared with those previously assumed for the M67,
BMRR and FCB facilities to confirm the appropriateness of the different
first time, a consistent set of characterizations of all facili- values with depth in phantom assigned to kt for the sensitivity of the A-
ties accruing clinical results in the Western world. This 150 plastic walled chamber to neutrons. Fast-neutron absorbed doses are
study is intended to illustrate the suitability of the beams reported for A-150 plastic that is tissue (muscle) equivalent (15).
used to date for BNCT and emphasize the possible require- To compare the performance of each beam under more realistic con-
ments of those that might be needed in future, more clini- ditions pertinent to the irradiation of brain tumors where boron would be
present in tissue, the boron dose was added to provide the total effective
cally based settings. It is also envisaged that these mea- dose profiles from the measured data. Two uniform sets of parameters
surements could be extended to include other centers in the were applied that are representative of either boronated phenylalanine
world as and when they commence clinical trials using ep- (BPA) or a more advanced boron delivery compound such as the por-
ithermal neutrons. phyrins currently under development. Boron concentrations of 18 and 65
g g 1 were assumed in normal brain and tumor tissue, respectively,
which are characteristic of the uptakes observed using BPA in clinical
MATERIALS AND METHODS trials at MIT (16). Additional profiles determined for a more selective
porphyrin compound assumed a boron concentration of 65 g g 1 in
The epithermal neutron beams at the facilities of the R2–0 (Studsvik) tumor to match the loading achievable with BPA, but with only 0.65 g
(9), FiR-1 (Espoo) (10), LVR-15 (ReZ) (11) and HFR (Petten) (12) re- g 1 in brain (17). Brain tissue was considered to be composed of 10.7%
actors were characterized with the standard dosimetry apparatus and pro- hydrogen, 14.5% carbon, 2.2% nitrogen and 71.2% oxygen by mass (15).
cedures used in clinical trials at MIT. This is a direct extension of pre- Skin doses were estimated by combining the activation foil measurements
vious characterizations for the M67 (6), BMRR (7) and FCB facilities on the surface of the phantom with extrapolations of the photon and fast-
(8). Irradiation conditions and beam-line configurations for the study were neutron absorbed dose rates measured in phantom. The composition of
similar to those routinely used during therapy at the different facilities. skin was represented as soft muscle tissue containing 10.2% hydrogen,
Beam intensity was assessed with the respective dose control systems 12.3% carbon, 3.5% nitrogen and 72.9% oxygen (15). Boron concentra-

3. 214 BINNS, RILEY AND HARLING
TABLE 1
In-Air Beam Parameters Measured in Seven Different Epithermal Beams used for Clinical NCT
MIT FCBa Studsvikb FiR-1 BMRRc ReZ HFR MIT M67c
Reactor power (MW) 5.0 0.50 0.25 3.0 9.0 45 5.0
Aperture diameter (cm) 12 14 10 14 12 12 12 15
epi (10 n cm s 1) 4.29 1.43 1.18 1.11 0.60 0.33 0.20
9 2
Dfn (cGy min 1) 3.6 7.1 2.3 1.7 6.0 2.4 1.6
D (cGy min 1) 9.4 10.8 0.6 1.0 3.9 0.8 1.7
Dfn/ epi (10 11 cGy cm2) 1.4 8.3 3.3 2.6 16.9 12.1 13.0
D / epi (10 11 cGy cm2) 3.6 12.6 0.9 1.5 10.8 3.8 14.3
Dfn/ epi (10 11 cGy(w) cm2) 4.5 26.5 10.4 8.3 54.1 38.7 41.6
DTotal/ epi (10 11 cGy(w) cm2) 8.1 39.2 11.5 9.8 64.9 42.5 55.9
Notes. Overall uncertainties (1 ) of 15, 5 and 16% are associated respectively with the epithermal flux, photon and fast-neutron absorbed dose
determinations. The data from the Brookhaven (BMRR) and two MIT (M67 and FCB) clinical facilities were obtained previously (6–8). The total
specific doses were determined assuming an RBE of 3.2 for neutrons and 1.0 for photons.
a
Fission converter operating at 83 kW.
b
Beam line configured with 9-mm-thick 6Li filter installed.
c
Decommissioned.
tions of 27 and 0.65 g g 1 were assumed in skin for BPA (18) and tron dose components. The individual and total specific
porphyrin, respectively. Individual absorbed dose components were com- background doses were determined by weighting the pho-
bined to yield estimates of the total weighted dose to normal brain and
tumor tissue by applying RBE values of 1.0 and 3.2 for photons and ton and neutron contributions with RBEs of 1.0 and 3.2,
neutrons (thermal and fast), respectively. Differences in the microdistri- respectively, and then summing the photon and fast-neutron
bution of boron delivered by BPA in both tissue and tumor were also components. Epithermal neutron flux ranges from 0.2–4.3
accommodated using cRBE factors of 1.3 for normal brain, 3.8 for tumor 109 n cm 2 s 1 with the attendant beam purity varying
(19), and 2.5 for skin (18). In the absence of boron microdistribution data significantly between the different facilities. A practical
for an advanced compound, a conservatively high estimate of 3.8 was
applied as the cRBE to the absorbed dose produced from thermal neutron limit of approximately 2.8 10 11 cGy(w) cm2 was pro-
capture by boron in normal brain, skin and tumor. The measurements and posed (4) as a goal for negligible beam contamination. The
calculations include no boron in the water-filled phantom that would oth- additional dose burden this represents is equal to only about
erwise be accounted for in treatment planning. 10% of the unavoidable peak effective absorbed dose pro-
The total weighted dose profiles obtained from the in-phantom mea- duced from neutron capture reactions with elemental hy-
surements were then used to determine several figures of merit, namely
the advantage parameters (20), to assess the different beams under uni- drogen and nitrogen during a brain irradiation with epither-
form irradiation conditions. The advantage depth (AD) is defined as the mal neutrons alone. The three facilities, FCB, FiR-1 and
depth at which the total weighted dose to tumor equals the maximum BMRR, possessing the lowest beam contamination of ap-
weighted dose received by normal tissue during an irradiation and is a proximately 10 10 11 RBE cGy cm2 exceed the proposed
measure of the maximum depth at which therapeutic benefit is obtained. diminimus level but are considerably less than the 39–65
The advantage ratio (AR) is the ratio of the integral tumor dose to the
normal tissue dose averaged from the surface where the beam is incident 10 11 cGy(w) cm2 exhibited by the others.
to the advantage depth. The advantage depth–dose rate (ADDR) is also Figure 1 shows the thermal neutron flux as a function of
determined and specifies the therapeutic dose rate at the AD. The ADDR depth in phantom on the central axis for the different
gives the total dose rate achievable to treat tumor at the maximum useful beams. The estimated uncertainties for all these measure-
depth of the beam and is also the maximum absorbed dose rate to normal ments, which are omitted from the figure for clarity, range
tissue. Last, the therapeutic ratio, defined as the quotient of the biologi-
cally weighted dose to tumor and the maximum weighted dose to brain, between 4.4% at shallow depth to 6.0% at 10 cm depth.
was determined from the measured dose profiles as a function of depth Curves are smoothed spline fits to the experimental data to
in the water phantom. help guide the eye and are shown only for the four Euro-
pean facilities. The data are normalized, for ease of com-
parison, to each measured thermal-neutron flux maximum
RESULTS
given in Table 2. The shapes of the profiles are all generally
Free in-air measurements of the epithermal neutron in- similar except for those of Studsvik and Petten, which peak
tensity and purity provide a useful first-order indicator of at slightly greater depths. Although the experimental un-
beam performance. The epithermal neutron flux and spe- certainties make separating the different curves difficult, it
cific fast-neutron and photon absorbed doses measured in- is apparent that the HFR has the lowest relative thermal
air for the specified aperture size at each of the seven beams flux on the central axis at the beam entrance point on the
are shown in Table 1 scaled to the nominal reactor oper- phantom. The FiR-1 curve peaks at the shallowest depth
ating power used during therapy. Fast-neutron absorbed and is then attenuated with depth at the same rate as the
doses are those measured in A-150 plastic. The measure- FCB, M67 and BMRR beams. This is more rapid than for
ments have estimated uncertainties (1 ) of 15% for epi- either of the Studsvik, Petten or ReZ beams. At Studsvik,
thermal flux, 5% for the photon and 16% for the fast-neu- the deeper penetrating profile is attained with 6Li filtration,

4. A COMPARISON OF SEVEN EPITHERMAL BEAMS FOR BNCT 215
FIG. 1. Thermal neutron (2200 m s 1) flux obtained from the analysis FIG. 2. Photon absorbed dose measured with a graphite walled ioni-
of gold foil activation at seven different clinical NCT facilities in Europe zation chamber are shown for seven different NCT facilities in Europe
and the U.S. Measurements were performed in an ellipsoidal water phan- and the U.S. Measurements were performed in an ellipsoidal water phan-
tom under conditions pertinent to clinical irradiations. The results for each tom under conditions pertinent to clinical irradiations. The results for each
beam are normalized to the measured maximum given in Table 2 and beam are normalized to the measured maximum given in Table 2 and
have an estimated uncertainty of 4.4% at shallow depths increasing to have an estimated uncertainty of 4.4%.
6.0% at 10 cm.
attenuated at depths greater than 3 cm at a similar rate
which preferentially decreases the number of lower-energy except for the HFR and M67 beams. The possibility of
neutrons in the epithermal range, while at the HFR it is significant differences between the in-phantom photon en-
likely related to high beam collimation. ergy spectra of these two beams compared to the others is
The photon and fast-neutron dose components associated not considered a likely explanation. This is because the
with the beams are shown in Figs. 2 and 3 as depth–dose majority of the photon dose is produced from the same
curves normalized to the respective measured maxima stat- interactions of thermal neutrons with the medium irrespec-
ed in Table 2. All photon dose maxima were observed at tive of the facility. Poorer shielding and the presence of
depth in-phantom, but those for fast neutrons were gener- leakage radiation striking the phantom from the beam line
ally obtained in-air. The estimated uncertainty for the pho- as well as a beam size approaching the phantom size (21)
ton dose component is 4.4%. Beams with higher specific may explain the deeper penetration of the photons for the
photon doses have proportionally greater photon doses in M67 beam. However, the different photon attenuation pro-
the first 3 cm from the surface of the phantom where dose file for the HFR may be related to the high collimation of
buildup is prominent. The photon curves peak at approxi- the epithermal neutron beam that ensures a more radially
mately the same depth in all the different beams and are uniform irradiation of the phantom volume relative to less
TABLE 2
Measured Maxima for Thermal Neutron Flux as well as Photon and Fast-Neutron
Absorbed Dose Rates at the Seven Different NCT Facilities in Europe and the U.S.
In-phantom maxima MIT FCBa Studsvikb FiR-1 BMRRc ReZ HFR MIT M67c
2200 (10 n cm s 1) 5.70 2.68 1.95 1.48 1.18 0.72 0.20
9 2
Photon dose rate (cGy min 1) 31.0 18.1 9.8 6.3 9.5 5.0 2.6
Fast-neutron dose rate (cGy min 1) 3.2 4.3 2.3 1.6 6.7 1.6 1.1
Note. These values are used to normalize the flux depth profile and depth–dose curves illustrated in Figs. 1–3.
a
Fission converter operating at 83 kW.
b
Beam line configured with 9-mm-thick 6Li filter installed.
c
Decommissioned.

5. 216 BINNS, RILEY AND HARLING
certainties depend upon the relative proportion of the fast-
neutron component to the total dose. Only two curves are
shown that represent the shallowest (Studsvik, FCB and
HFR) and the deepest (M67) profiles for the normalized
data sets, with points from the other facilities generally ly-
ing between. The normalized fast-neutron dose profiles all
appear to be generally similar in shape within the large
stated uncertainties, and the doses are attenuated rapidly in
the first few centimeters irrespective of the beam.
The overall quality of the different beams is compared
using advantage parameters (figures of merit) presented in
Table 3. These were calculated from the weighted depth–
dose profiles assuming a boron uptake characteristic of
BPA and an advanced compound. Uncertainties associated
with the different advantage parameters depend upon the
relative error between the RBE weighted dose profiles for
normal tissue and tumor in each beam. The effect on the
weighted dose profiles of separately varying each absorbed
dose component by its respective uncertainty was studied
in a sensitivity analysis to give conservatively estimated
uncertainties of 1, 6 and 3% for the AD, AR and ADDR,
respectively. Skin doses for either type of compound did
FIG. 3. Fast-neutron absorbed dose measured with an A-150 plastic not exceed the peak dose to brain for any of the beams and
walled ionization chamber are presented for seven different NCT facilities
in Europe and the U.S. Measurements were performed in an ellipsoidal consequently had no influence on the derived advantage
water phantom under conditions pertinent to clinical irradiations. The parameters. Advantage depths (useful penetration) of 8.0
results for each beam are normalized to the measured maximum given in cm or greater are realized with all the beams studied if the
Table 2 and have an estimated uncertainty ranging between 29 and 300% uptake characteristics of BPA are assumed. The AD must
depending upon the depth in-phantom. Higher uncertainties are associated
with greater depth in-phantom. The measured in-air fast-neutron dose
be larger than the maximum of tumor depth for effective
rates from Table 1 are plotted at a depth of 0 cm with data points for therapy, and values 8.0 cm or greater are needed to irradiate
each beam normalized to the maximum measured either in-air (Studsvik, a target as deep as the midline of the average-sized head
HFR, FCB, M67 and BMRR) or in-phantom (FiR-1 and ReZ). with bilateral field placements. The ADs of the facilities
that are still operational vary by only 1.1 cm for the dif-
ferent field sizes assessed. When considering BPA, a best
forward-directed beams for field sizes approaching the di- value of 9.7 cm is realized by both Studsvik and the HFR,
mensions of the phantom (21). Normalized fast-neutron which possess the most penetrating thermal flux depth pro-
depth–dose data are depicted in Fig. 3 with the in-air values files. Use of an advanced compound increases the AD by
also plotted at a depth of 0 cm. The uncertainties for these between 11 and 22% depending upon the beam.
measurements are large, varying in the different beams be- The advantage ratio (AR) accounts for the detrimental
tween 29% near the surface to 300% at depth. These un- effects of beam contamination in achieving the required
TABLE 3
Advantage Parameters or Figures of Merit Derived from the Dosimetric Measurements for the Nominal
Operating Conditions of the Different Epithermal Beams Using a Uniform Set of Boron Uptake Parameters
Appropriate for BPA and an Advanced Delivery Compound in Parentheses
MIT FCBa Studsvikb FiR-1 BMRRc ReZ HFR MIT M67c
AD (cm) 9.3 (11.3) 9.7 (11.2) 9.0 (10.5) 9.3 (10.6) 8.6 (9.5) 9.7 (11.0) 8.0 (9.0)
AR 6.0 (11.8) 5.6 (10.1) 5.8 (10.9) 6.0 (11.9) 4.2 (6.2) 5.4 (9.3) 3.5 (4.7)
ADDR (cGy(w) min 1) 125 (60) 67 (41) 45 (24) 33 (16) 46 (35) 19 (12) 9 (7)
Time to reach 12.5 Gy(w) (min) 10 (21) 19 (31) 28 (52) 38 (77) 27 (36) 66 (104) 139 (177)
Notes. Boron concentrations of 18 and 65 g g 1 are assumed in normal brain and tumor tissue, respectively, for BPA and 0.65 and 65 g g 1 for
an advanced compound. The applied RBEs are 1.0 for photons and 3.2 for neutrons. The cRBEs are 1.3 in brain and 3.8 in tumor for BPA and 3.8
in both tissue and tumor for an advanced compound. The time required to reach a normal tissue tolerance dose of 12.5 Gy(w) for these irradiation
conditions is also given.
a
Fission converter operating at 83 kW.
b
Beam line configured with 9-mm-thick 6Li filter installed.
c
Decommissioned.

6. A COMPARISON OF SEVEN EPITHERMAL BEAMS FOR BNCT 217
FIG. 5. Variations in therapeutic ratio as a function of depth deter-
FIG. 4. Variations in therapeutic ratio as a function of depth deter-
mined for the current clinical epithermal neutron facilities using param-
mined for the different clinical epithermal neutron facilities using param-
eters for an advanced compound as the boron delivery agent. Boron con-
eters for BPA as the boron delivery agent. Boron concentrations of 18
centrations of 0.65 and 65 g g 1 are assumed in normal brain and tumor
and 65 g g 1 are assumed in normal brain and tumor tissue, respectively.
tissue, respectively. The applied RBEs are 1.0 for photons and 3.2 for
The applied RBEs are 1.0 for photons and 3.2 for neutrons with a cRBE
neutrons with a cRBE of 3.8 in tumor. The dashed horizontal line rep-
of 1.3 in brain and 3.8 in tumor. The dashed horizontal line represents a
resents a TR of one.
TR of one.
penetration and dose delivery to tumor. The advantage ratio 125 cGy(w) min 1 or between 7 and 60 cGy(w) min 1 with
with BPA varies between 5.4 and 6.0 at most of the facil- the advanced compound. Irrespective of the class of com-
ities, with beams that possess lower contamination at the pound considered, however, the range in ADDR between
upper end of this range. Lower values of 4.2 and 3.5 are different beams is large, varying by more than a factor of
associated with facilities having the highest beam contam- five for the existing facilities. The time to reach a tolerance
ination. The FiR-1 beam, for instance, with the lowest com- dose to normal tissue of 12.5 Gy(w) at the specified ADDR
bined in-air contamination of the four European facilities, is also given and ranges from 10–139 min and 21–177 min
has a high AR of 5.8 even though the thermal-neutron flux with BPA and an advanced compound, respectively.
profile appears to be the shallowest of those measured. This Estimates of expected clinical performance for the dif-
is in contrast to the facility at ReZ, which offers a superior ferent beams are compared as a function of depth in phan-
distribution of thermal neutrons at depth but which is ac-
tom using the therapeutic ratio (TR). These curves are
companied by much higher beam contamination and con-
shown for brain irradiations in the different beams with the
sequently a poorer AR of only 4.2. The FCB and BMRR
two classes of compound in Figs. 4 and 5. The peak TR
show an AR of 6.0, which is the best in this comparison.
values are observed at depths of 2–3 cm and are large,
These two examples illustrate the benefit of providing ad-
equate depth profiles of thermal flux in combination with ranging between 3.3 and 6.3 with BPA and favoring most
a low specific fast-neutron contamination. The AR increas- those facilities with lower specific dose contamination. Us-
es markedly if an advanced compound is considered and ing an advanced compound significantly improves the peak
almost doubles for beams with the lowest specific dose TR values for all the existing facilities and extends the use-
rates for the RBEs considered. If a cRBE lower than 3.8 ful therapeutic penetration of these beams by up to 2 cm.
was found appropriate for normal tissues (as with BPA), At a depth of 8 cm (the brain midline), the TR for the FCB
the dose to normal tissue would decrease and the AD and is highest for the advanced compound, with a value of 3.1,
AR values would increase still further. which is higher than for Studsvik (2.9), HFR (2.8), FiR1
The advantage depth–dose rate (ADDR), which is the (2.8) and ReZ (1.5). Again these results for overall beam
therapeutic dose rate at the AD, is also given in Table 3. If performance exemplify the advantage of high purity linked
parameters for BPA are used, the ADDRs range from 9 to to adequate thermal flux penetration.

7. 218 BINNS, RILEY AND HARLING
TABLE 4
An Illustration of the Influence of Aperture Size and 6Li Filtration on the Advantage
Parameters for the FCB Using Boron Uptake Parameters Appropriate for BPA and an
Advanced Compound in Parentheses
ADDRa Time to reach
Aperture diameter (cm) AD (cm) AR (cGy(w) min 1) 12.5 Gy(w) (min)
16 9.7 (12.0) 5.7 (10.8) 173 (88) 7.4 (14.2)
12 9.3 (11.3) 6.0 (11.8) 125 (60) 9.6 (20.8)
12 (with 8-mm 6Li filter) 9.8 (11.7) 5.7 (10.7) 66 (32) 18.9 (39.1)
Notes. Boron concentrations of 18 and 65 g g 1 are assumed in normal brain and tumor tissue, respectively, for
BPA and 0.65 and 65 g g 1 for an advanced compound. The applied RBEs are 1.0 for photons and 3.2 for neutrons.
The cRBEs are 1.3 in brain and 3.8 in tumor for BPA and 3.8 in both tissue and tumor for an advanced compound.
The time required to reach a normal tissue tolerance dose of 12.5 Gy(w) for these irradiation conditions is also
given. Results for the 12- and 16-cm fields are from measurements (8) except those with the 6Li filter, which were
calculated using the experimentally validated beam model of the FCB.
a
Fission converter operating at 83 kW.
DISCUSSION figures of merit for an epithermal neutron beam can be
quite pronounced, as illustrated for the FCB in Table 4.
The measured epithermal neutron intensities vary con-
Increasing circular field size from 12 to 16 cm in diameter
siderably among the different facilities, spanning over an
increases the AD by 0.4 cm to 9.7 cm, decreases the AR
order of magnitude between the least and most intense. Re-
from 6.0 to 5.7, and significantly improves the ADDR from
sults are for the typical operating powers of the reactors
1.25 to 1.73 Gy(w) min 1 if BPA is considered as the boron
that produce the neutron sources. Higher power operation
delivery compound. These changes in AD and AR are sim-
is possible at some of the facilities, such as Studsvik, where
ilar in magnitude to those that would be expected by in-
the reactor can run at 1 MW rather than the current 0.5
MW, and at MIT, where increases of up to 250% could be corporating a final 8-mm-thick 6Li filter in the beam line
realized (22). Beam purity is most easily assessed by the for the 12-cm-diameter field. These results were obtained
contamination evident from in-air measurements and using the experimentally validated beam model of the FCB
should be minimized or reduced to levels that have an in- (8). The inclusion of the 6Li filter, however, reduces beam
significant effect on in-phantom beam performance. Total intensity considerably, by almost half, whereas increasing
specific dose rate can be a good predictor of overall in- field size has the opposite effect. Field size may also ex-
phantom beam performance since all the beams with low plain in part the better penetration observed at the HFR.
specific dose rates of approximately 10 10 11 cGy(w) Although a field with a 12-cm-diameter aperture was stud-
cm (i.e. FCB, FiR-1 and BMRR) exhibit good figures of
2 ied, the phantom was positioned 30 cm from the collimator
merit. Having a higher than optimum specific dose rate aperture at the HFR where anything less than ideal colli-
does not, however, preclude good in-phantom performance, mation will produce a divergent beam. Previously a full
as illustrated by the Studsvik and HFR beams. The gener- width at half maximum of 13.2 cm has been reported for
ally good overall performance of these two beams can be the beam (24) at this location. This effectively larger field
understood qualitatively in terms of the known dosimetric impinging upon the phantom should contribute to pushing
behavior of epithermal neutron beams (21, 23). The Studs- the thermal flux deeper, as observed.
vik beam uses a similar filter-moderator arrangement of Tef- The measured flux and dose–depth profiles are generally
lon and aluminum as the FCB to produce a useful spectrum similar for all the beams, although small discernible differ-
of epithermal neutrons but incorporates an additional 9- ences are evident that influence the figures of merit derived
mm-thick 6Li filter that removes lower-energy neutrons for each beam. The thermal flux profile on central axis is
through absorption (9). This filter hardens the final beam the most important of the quantities measured since this is
incident upon the patient, enhancing penetration of the ep- used to directly determine the 10B dose component in tissue.
ithermal neutron flux and the resultant thermal neutrons This dose contribution is significant with a compound like
(23). Combined with this is the influence of field size, BPA. However, thermal neutron flux should not be consid-
which for increasing dimensions up to those of the target ered as the sole criterion for judging beam performance.
(ellipsoidal water phantom) improves the depth profile for The Studsvik and HFR beams have better penetrating ther-
thermal neutron flux and the advantage depth. At Studsvik, mal neutron profiles (and ADs) than the FCB, FiR-1 or
the field aperture has an area of 140 cm2 compared to 113 BMRR but inferior advantage ratios. The more penetrating
cm2 for that studied at the FCB. Predictions for a colli- thermal neutron component of the ReZ beam does not pro-
mated, forward-directed beam show that the AD and duce superior figures of merit compared to the FCB, FiR-
ADDR increase as the AR decreases with increasing field 1 and BMRR beams, because the better penetration at ReZ
size (21). The effect of changing field size on in-phantom is insufficient to offset the deleterious effects of the high

8. A COMPARISON OF SEVEN EPITHERMAL BEAMS FOR BNCT 219
specific beam contamination (65 10 11 cGy(w) cm2). The apparent if full advantage is to be taken of future improve-
majority of the specific beam contamination at ReZ is from ments in compounds. This is illustrated by plotting the ther-
fast neutrons that have not been filtered from the beam and apeutic ratio as a function of depth for the different beams
that presumably contribute to a harder neutron energy spec- with the two classes of compound. The more the normal
trum and a better depth penetration of thermal neutrons. tissue dose is reduced, whether from beam contamination
The ReZ beam does offer superior performance compared or the delivery compound selectivity, the higher the TR
to the M67 facility that had a comparable specific dose of peak and the greater the improvement in performance of a
56 10 11 cGy(w) cm2 but a less penetrating thermal com- particular beam for BNCT at each depth. The comparative-
ponent. ly high uptake of boron in normal tissue from the BPA
Of practical significance to BNCT is the ADDR, which compound currently employed produces a large normal tis-
determines how quickly a clinical irradiation field can be sue dose that obscures the importance of beam purity, al-
completed. The ADDRs for the conditions studied depend though the detrimental effect of high specific doses (56–65
mostly on the intensity of the epithermal neutron source 10 11 cGy(w) cm2) is evident in the limited TR and AD
available. A large variation between 9 and 125 cGy(w) attained in the ReZ and the M67 beams. Should future com-
min 1 is apparent amongst the various facilities if BPA is pounds with better selectivity result in lower boron con-
considered. The FCB, Studsvik, FiR-1 and ReZ facilities centration in normal tissue, beam contamination will be
can deliver a tolerance dose to normal tissue of 12.5 Gy(w) more important in limiting the physical performance of a
in less than the 30 min suggested as reasonable for clinical beam for BNCT. This is illustrated in Fig. 5 by the mark-
studies (4). The use of an advanced compound with a high- edly improved TRs with an advanced compound. The rel-
er tumor to normal tissue selectivity, although increasing ative improvement in the TR for the FCB and FiR-1 beams,
the tumor dose relative to normal tissue, significantly which have the lowest measured specific dose, is large and
lengthens irradiation times compared to those for BPA with exceeds that for any of the other beams at all depths.
the same boron concentration in tumor. This is because the Apart from providing a consistent measurement of beam
boron in normal tissue would be much less than with BPA, performance, it is also envisaged that these data will be
but the clinician would probably still want to irradiate until used in a preliminary comparison of the calculated patient
normal tissue tolerance is approached. Nonetheless, the dosimetry from the participating clinical centers. The re-
FCB, Studsvik and ReZ beams would be able to complete sults will be used in the near future to normalize comple-
irradiations in about 30 min or less. In practice, however, mentary treatment planning calculations from each center
it might also be feasible to increase the ADDR when using that are currently being prepared for the identical irradiation
an advanced compound by increasing the overall boron conditions. These will use CT images of the ellipsoidal wa-
loadings in both tumor and normal tissue above those spec- ter phantom2 together with the RBE as well as cRBE values
ified, although this will eventually be limited by drug tox- and actual boron concentrations appropriate for each infu-
icity. Routine therapy would require high patient through- sion schedule. This separate study will be an integral test
put as well as multiple beams possibly for each irradiation relating the absorbed doses specified by treatment planning
fraction, and ADDRs of 1 Gy(w) min 1 or greater would calculations with those actually measured at all the different
be desirable. Some of the current facilities already have the centers using a common dosimetry method to facilitate a
intensities required or the capacity to attain them. direct comparison of clinical dosimetry protocols.
The improvements that a more selective capture agent This study illustrates that the dosimetric performance of
could bring to the figures of merit depend upon the beam. an epithermal neutron beam for BNCT can be properly as-
AD and AR increase for all the facilities currently pursuing sessed only by using figures of merit that account for all
clinical trials. The advantage depth improves by between radiation components in the mixed field rather than by com-
11 and 22% for the different facilities, favoring those beams parison of any single dose component. The adequacies of
with lower specific doses. A significant improvement is ob- the different epithermal neutron beams investigated are
served for the AR, increasing between 80–97% for the demonstrated by advantage depths of 8.0 cm or greater
cleaner beams. An advanced compound would offer better when uptake parameters consistent with BPA as a boron
dose distributions for tumor relative to normal tissue and delivery agent are assumed. The advent of more advanced
could provide more flexibility in treatment planning. In ad- compounds with a better tumor to normal tissue selectivity
dition, the useful penetration of a beam would be extended will, however, place further demands on the existing and
by up to 2 cm in some cases. This could be important if future facilities if the full potential of NCT is to be realized.
sites other than the brain are to be considered as future The challenge in epithermal neutron beam design is to
candidates for NCT. The greater tissue selectivity promised achieve a TR greater than unity that extends to the greatest
by improved compounds will allow a higher tumor dose possible depth while maintaining adequate intensity for
during NCT and will effectively shift the AD deeper, de- clinical studies and eventually routine therapy. It is clear
crease the ADDR, and increase the irradiation time. The that by hardening the epithermal energy spectrum from fis-
need for an epithermal neutron source that has the com-
bined attributes of high purity and high intensity becomes 2
W. S. Kiger III, private communication, 2004.

9. 220 BINNS, RILEY AND HARLING
sion sources through 6Li filtration (Studsvik), high beam Metamorphosis of a 35 year-old TRIGA reactor into a modern BNCT
facility. In Frontiers in Neutron Capture Therapy, Vol. I (M. F. Haw-
purity (FCB, FiR-1, BMRR), and high collimation (HFR), thorne, K. Shelly and R. J. Wiersema, Eds.), pp. 267–275. Kluwer
a favorable effect on the maximum useful beam penetration Academic Publishers/Plenum Press, New York, 2001.
is achieved. However, all these approaches to enhanced 11. M. Marek, L. Viererbl, S. Flibor, J. Burian and J. Rejchrt, Validation
beam performance are costly in intensity and emphasize the of the epithermal neutron beam at LVR-15. In Proceedings of the
Ninth International Symposium on Neutron Capture Therapy, pp. 41–
importance of producing neutron sources of the highest 42. Osaka, Japan, 2000. [Abstract]
possible strength. 12. R. L. Moss, F. Stecher-Rasmussen, K. Ravensberg, G. Constantine
and P. Watkins, Design, construction and installation of an epithermal
neutron beam for BNCT at the high flux reactor Petten. In Progress
ACKNOWLEDGMENTS in Neutron Capture Therapy for Cancer (B. J. Allen Ed.), pp. 63–
66. Plenum Press, New York, 1992.
The authors thank the staff at the different facilities for their cooper-
13. O. K. Harling, K. A. Roberts, D. J. Moulin and R. D. Rogus, Head
ation and assistance throughout the course of this study. In particular we
phantoms for neutron capture therapy. Med. Phys. 22, 579–583
are grateful to Drs. Jacek Capala, Kurt Skold and Per Munck af Rosensc-
¨ (1995).
hold (Studsvik), Iiro Auterinen and Tom Seren (FiR-1), Milan Marek and
¨
14. ASTM, Standard Method for Determining Thermal Neutron Reaction
Jiri Burian (ReZ) as well as Finn Stecher-Rasmussen, Ray Moss and Jim
and Fluence Rates by Radioactivation Techniques. Standard E262-
Morrissey (HFR) for scheduling and making these measurements possible 97, American Society for Testing and Materials, West Conshohocken,
at their respective facilities. The U.S. DOE provided partial support for PA, 1998.
this work under contract numbers DE-FG02-97ER62489 and DE-FG02-
15. ICRU, Tissue Substitutes in Radiation Dosimetry and Measurement.
96ER62193. Report 44, International Commission on Radiation Units and Mea-
surements, Bethesda, MD, 1998.
Received: March 15, 2004; accepted: March 10, 2005 16. W. S. Kiger, III, M. R. Palmer, K. J. Riley, R. G. Zamenhof and P.
M. Busse, A pharmacokinetic model for the concentration of 10B in
blood after boronphenylalanine-fructose administration in humans.
REFERENCES Radiat. Res. 155, 611–618 (2001).
1. P. M. Busse, O. K. Harling, M. R. Palmer, W. S. Kiger, III, J. Kaplan, 17. M. Miura, G. M. Morris, P. L. Micca, D. T. Lombardo, K. M. Youngs,
I. Kaplan, C. F. Chuang, J. T. Goorley, K. J. Riley and R. G. Za- J. A. Kalef-Ezra, D. A. Hoch, D. N. Slatkin, R. Ma and J. A. Coderre,
menhof, A critical examination of the results from the Harvard-MIT Boron neutron capture therapy of a murine mammary carcinoma us-
NCT program Phase I clinical trial of boron neutron capture therapy ing a lipophilic carboranyltetraporphyrin. Radiat. Res. 155, 603–610
for intracranial disease. J. Neuro-Oncol. 62, 111–121 (2003). (2001).
2. H. Joensuu, L. Kankaanranta, T. Seppala, I. Auterinen, M. Kallio, M.
¨¨ 18. J. A. Coderre and G. M. Morris, The radiation biology of boron
Kulvik, J. Laakso, J. Vahatalo, M. Kortesniemi and S. Savolainen,
¨ ¨ neutron capture therapy. Radiat. Res. 151, 1–18 (1999).
Boron neutron capture therapy of brain tumors: Clinical trials at the 19. J. A. Coderre, M. S. Makar, P. L. Micca, M. M. Nawrocky, H. B.
Finnish facility using boronophenylalanine. J. Neuro-Oncol. 62, 123– Liu, D. D. Joel, D. N. Slatkin and H. I. Amols, Derivations of relative
134 (2003). biological effectiveness for the high-LET radiations produced during
3. W. Sauerwein and A. Zurlo, The EORTC Boron Neutron Capture boron neutron capture irradiations of the 9L rat gliosarcoma in vitro
Therapy (BNCT) Group: Achievements and future projects. Eur. J. and in vivo. Int. J. Radiat. Oncol. Biol. Phys. 27, 1121–1129 (1993).
Cancer 38, S31–S34 (2002). 20. S. D. Clement, J. R. Choi, R. G. Zamenhof, J. C. Yanch and O. K.
4. O. K. Harling and K. J. Riley, Fission reactor neutron sources for Harling, Monte Carlo methods of neutron beam design for neutron
neutron capture therapy—a critical review. J. Neuro-Oncol. 62, 7– capture therapy at the MIT research reactor (MITR-II). In Neutron
17 (2003). Beam Design, Development, and Performance for Neutron Capture
Therapy (O. K. Harling, J. A. Bernard and R. G. Zamenhof, Eds.),
5. IAEA, Current Status of Neutron Capture Therapy. IAEA, Vienna, pp. 51–69. Plenum Press, New York, 1990.
2001.
21. J. C. Yanch and O. K. Harling, Dosimetric effects of beam size and
6. R. D. Rogus, O. K. Harling and J. C. Yanch, Mixed field dosimetry collimation of epithermal neutrons for boron neutron capture therapy.
of neutron beams for boron neutron capture therapy at the MITR-II Radiat. Res. 135, 2488–2493 (1993).
research reactor. Med. Phys. 21, 1611–1625 (1994).
22. O. K. Harling, K. J. Riley, T. H. Newton, B. A. Wilson, J. A. Bernard,
7. K. J. Riley, P. J. Binns, D. D. Greenberg and O. K. Harling, A phys- L-W. Hu, E. J. Fonteneau, P. T. Menadier, S. J. Ali and P. M. Busse,
ical dosimetry intercomparison for BNCT. Med. Phys. 29, 898–904 The fission converter-based epithermal neutron irradiation facility at
(2002). the Massachusetts Institute of Technology reactor. Nucl. Sci. Eng.
8. K. J. Riley, P. J. Binns and O. K. Harling, Performance characteristics 140, 223–240 (2002).
of the MIT fission converter based epithermal neutron beam. Phys. 23. S. Sakamoto, W. S. Kiger, III and O. K. Harling, Sensitivity studies
Med. Biol. 48, 943–958 (2003). of beam directionality, beam size, and neutron spectrum for a fission
9. J. A. Capala, B. H. Stenstam, K. Skold, P. M. af Rosenschold, V.
¨ ¨ converter-based epithermal neutron beam for boron neutron capture
Giusti, C. Persson, E. Wallin, A. Brun, L. Franzen and R. Hendriks- therapy. Med. Phys. 26, 1979–1988 (1999).
son, Boron neutron capture therapy for glioblastoma multiforme: 24. C. P. J. Raaijmakers, M. W. Konijnenberg and B. J. Mijnheer, Clinical
Clinical studies in Sweden. J. Neuro-Oncol. 62, 135–144 (2003). dosimetry of an epithermal neutron beam for neutron capture therapy:
10. I. Auterinen, P. Hiismaki, P. Kotiluoto, R. J. Rosenberg, S. Salmen-
¨ dose distributions under reference conditions. Int. J. Radiat. Oncol.
hara, T. Seppala, T. Seren, V. Tanner, C. Aschan and P. Valimaki,
¨¨ ¨ ¨ Biol. Phys. 37, 941–951 (1997).