Boron concentration measurement in biological tissues by charged particle spectrometry

ORIGINAL PAPER

Boron concentration measurement in biological tissues by chargedparticle spectrometry

S. Bortolussi • S. Altieri

Received: 1 February 2013 / Accepted: 22 June 2013 / Published online: 9 July 2013

� Springer-Verlag Berlin Heidelberg 2013

Abstract Measurement of boron concentration in bio-

logical tissues is a fundamental aspect of boron neutron

capture therapy, because the outcome of the therapy

depends on the distribution of boron at a cellular level,

besides on its overall concentration. This work describes a

measurement technique based on the spectroscopy of the

charged particles emitted in the reaction 10B(n,a)7Li

induced by thermal neutrons, allowing for a quantitative

determination of the boron concentration in the different

components that may be simultaneously present in a tissue

sample, such as healthy cells, tumor cells and necrotic

cells. Thin sections of tissue containing 10B are cut at low

temperatures and irradiated under vacuum in a thermal

neutron field. The charged particles arising from the sample

during the irradiation are collected by a thin silicon

detector, and their spectrum is used to determine boron

concentration through relatively easy calculations. The

advantages and disadvantages of this technique are here

described, and validation of the method using tissue stan-

dards with known boron concentrations is presented.

Keywords BNCT � a-Spectrometry � Boron

concentration � Thermal neutrons

Introduction

Boron neutron capture therapy (BNCT) is an experimental

radiotherapy whose most interesting characteristic is that it

is selective at a cellular level (Barth et al. 2005). Boronated

drugs able to load the tumor with a higher concentration of10B compared to normal tissues are administered to the

patient; following this, the tumor target is irradiated with

thermal neutrons. Low-energy neutron capture in 10B

occurs with a cross section of 3,837 barns at 0.025 eV,

which is the most probable energy in a Maxwellian flux at a

temperature of 293.61 K, and the charged particles pro-

duced in the reaction 10B (n,a)7Li deposit locally in a dose

proportional to the boron concentration present. As the

high-LET a-particles and 7Li ions have ranges comparable

with a cell diameter, the energy deposition is local and does

not affect much the surrounding cells. Boron neutron

capture therapy selectivity depends on the preferential

boron uptake in the tumor rather than on the characteristics

of the irradiation beam, the latter being important for

conventional photon therapy and for proton and carbon-ion

therapy. Depending on the absolute values of the boron

concentration in normal and tumor cells, and on their ratio,

it is possible to assess an irradiation plan that allows

delivery of a lethal dose to the tumor and a dose below the

tolerance limits to the normal (healthy) tissues. This aspect

could be exploited in case of tumors that cannot be oper-

ated or treated by other radiotherapies, such as metastatic

spreads and highly infiltrative malignancies. At the TRIGA

Mark II reactor of the University of Pavia, research is

ongoing to apply BNCT to liver metastases from colon

carcinoma (Zonta et al. 2009), to lung disseminated tumors

(Bortolussi et al. 2011; Protti et al. 2009) and to limb

osteosarcoma (Ferrari et al. 2009). For all these studies,

conducted both in vitro and in vivo using animal models, it

S. Bortolussi (&) � S. Altieri

Department of Physics, University of Pavia, Via Bassi 6, 27100

Pavia, Italy

e-mail: [email protected]

S. Bortolussi � S. Altieri

National Institute for Nuclear Physics, INFN, Section of Pavia,

via Bassi 6, 27100 Pavia, Italy

123

Radiat Environ Biophys (2013) 52:493–503

DOI 10.1007/s00411-013-0480-y

was firstly necessary to prove the available boronated

drugs, mainly boronophenylalanine (BPA), to concentrate

preferentially on the tumor.

Different methods to measure boron concentration are

reported in the literature and are presently employed in

BNCT: inductively coupled plasma atomic emission

spectrometry (ICP-AES) (Probst 1999 and Wittig et al.

2008), inductively coupled plasma mass spectrometry

(ICP-MS) (Probst 1999), secondary ion mass spectrometry

(SIMS) (Chandra 2003) and prompt gamma neutron acti-

vation analysis (PGNAA) (Riley and Harling 1998; Ver-

bakel et al. 2003). In some clinical trials, blood samples

taken at different times after administration of the boro-

nated drug are taken from the patient and measured; boron

concentration in tumor and normal tissues is then inferred

based on the previous pharmacokinetic studies. For

example, a couple of hours after administration of BPA,

normal tissues contain a similar boron concentration as

blood, while the tumor concentration is a factor 3.5 higher

(Coderre et al. 1998). In the case of skin, the uptake of the

normal tissue is usually higher than the concentration

measured in blood (Fukuda et al. 1999; Menendez et al.

2009). Alternatively, the tumor-to-healthy tissues concen-

tration ratio may also be obtained by positron emission

tomography (PET) imaging after administration of BPA

labeled with 18F, a method that only gives a rough esti-

mation of the boron concentration in tissues (Imahori et al.

1998). The knowledge of boron concentration in the irra-

diated tissues is thus limited, and the dose delivered to the

patients could vary considerably due to the biological

variability of boron uptake, even with the same boron

administration protocols.

In Pavia, the first application of BNCT was a chal-

lenging protocol of liver autotransplantation, where the

liver of the patient, after BPA administration, was ex-

planted, irradiated in the thermal column of the TRIGA

reactor for about 10 min and then reimplanted in the

patient (Zonta et al. 2009). For this treatment, a boron

measurement technique based on charged particle

spectroscopy of tissue samples was developed (Chiaravi-

glio et al. 1989): this technique has been refined, and it is

extensively described in this paper. Biopsies from the

healthy liver and from some metastases were taken during

the explantation surgery, sectioned and measured at the

reactor. During the surgery, a full analysis of the samples

was conducted before the liver irradiation, and thus, the

assessment of the most effective irradiation plan was pos-

sible while the organ was prepared to be carried to and

irradiated at the reactor. Other applications, such as irra-

diation of the thorax with epithermal neutron beams to treat

disseminated lung metastases, could also take advantage

from this boron measurement technique, especially during

the preclinical in vitro and in vivo research. Then, in a

clinical trial, it could be employed on samples obtained in

explorative examinations where the boron carrier is

administered to the patient before taking a biopsy.

The measurement method is described in the following

paragraphs, together with its validation both by means of

tissue standards at different known boron concentrations

and by means of Monte Carlo calculations.

Materials and methods

Sample preparation

Biopsies taken from treated animals or from patients are

divided into smaller samples of about 1 cm3 and frozen in

liquid nitrogen. Then, 70-lm-thick sections are produced

with a Leica cryostat at a temperature of -20 �C and

deposited on 100-lm-thick Mylar disks (Fig. 1).

Measurement setup

The setup allows irradiation of up to ten samples simulta-

neously without the need to shut down the reactor. It

consists of a rotating holder made of Teflon with 12

housings for the Mylar supports, ten of which contain the

Fig. 1 Preparation of the tissue sections: a the sectioning at the cryostat. b The tissue sections deposited on Mylar disks on a sample holder

494 Radiat Environ Biophys (2013) 52:493–503

123

tissue samples, one the calibration standard described

below and one a Mylar disk only as a control. A cap is

placed on the top of the housings to protect the samples,

with holes (diameter 0.4 cm) that act as collimators, in

order to define the measurement area. The holder is then

screwed to the body of a chamber where a thin silicon

detector is also fixed. The chamber is connected to a pump

that keeps the setup under vacuum during the irradiation

(about 1 Pa). The whole chamber is positioned at the end

of the thermal column of the TRIGA reactor, where the

thermal neutron flux is approximately 2 9 109 cm-2 s-1.

The holder is rotated by remote control in a way that one

section a time is irradiated in front of the detector during

the established time (usually 10 min per sample). The

geometry of the setup is shown in Fig. 2. The signals

produced in the detector by the charged particles, after a

convenient shaping, are sent to a Ortec MCB module

connected to an Ortec Maestro software.

The background measured by this setup consists of three

components: (a) an exponential component due to the cbackground present in the irradiation position (Agosteo

et al. 2003), (b) protons from nitrogen in the residual air

present in the vacuum chamber and (c) particles from

neutron capture in 10B present in the detector as a dopant.

The exponential component can be reduced using detectors

of small volume; for this reason, a thin detector was chosen

(Ortec, 18–25 lm thick, area 450 mm2).

Many attempts were made to reduce the background in

the detector in the absence of a tissue sample. The best

results were obtained by positioning the detector in a pure

graphite ring and placing a disk of pure silicon behind the

detector (Fig. 2) that protects it from charged particles

arising from the neutron capture in impurities present in the

materials of the setup.

Energy calibration

The energy calibration is accomplished using a boron

standard sample purchased at National Institute of Standard

and Technology (NIST), consisting of a squared support of

pure silicon in which a known quantity of 10B was

implanted. The capture reaction occurs in the following

branches:

n ? 10B ? 11B ? 7Li ? a ? 2.79 MeV (6.1 %)

where the emitted a-particle has an energy of 1.78 MeV;

and

n ? 10B ? 11B ? (7Li)* ? a ? 2.31 MeV ? 7Li ? c?0.478 MeV (93.9 %) where the emitted a-particle has

an energy of 1.47 MeV.

The standard is an almost superficial implantation (peak

depth 0.188 lm from the surface). The spectra obtained by

irradiating such a sample with thermal neutrons are char-

acterized by Gaussian peaks corresponding to a-particles

and Li ions of the two branches of the neutron capture

reaction, including a small energy absorption in the stan-

dard sample governed by the implantation depth of 10B in

the sample. The detector resolution allows separation of the

two peaks due to a-particles but not the peaks due to Li

ions (Fig. 3). The a-energy of 1,470 keV (which is actually

1,422 keV due to energy absorption in the standard sam-

ple) is used as a reference for the energy calibration of the

obtained spectra. The energy resolution of the detector at

this energy is 2.5 %.

Fig. 2 Sketch of the setup for charged particle spectrometry (not

drawn to scale)

Fig. 3 Energy spectrum obtained by irradiating the standard sample

consisting of a silicon wafer with an almost superficial implantation

of 10B atoms. For details see text

Radiat Environ Biophys (2013) 52:493–503 495

123

Spectra from tissue samples

The main elements in biological tissue are H, C, O and N,

with traces of Na, Cl, P, S and K, with the addition of B.

Because the neutron flux at the measurement position is

mainly thermal and the epithermal and fast contamination

is only about 2 % (Protti et al. 2012), the spectrum mea-

sured with the silicon detectors will be essentially produced

by charged particles emitted in reactions showing a posi-

tive Q-value. These reactions are reported in Table 1.

In addition to these reactions, there is also the thermal

neutron capture in hydrogen, 1H(n, c)2H, with a cross

section of 0.322 b that produces 2.2 MeV gamma rays. Due

to the concentration of each element in tissue and to the

reaction cross section, the most relevant reactions are14N(n,p)14C and 1H(n,c)2H. Note, however, that the gamma

radiation from hydrogen cannot be detected by the system

that is optimized for the charged particle detection; it only

contributes to a low-energy background. In contrast, the

range of 588 keV protons from the nitrogen reaction is of

the order of 10.8 lm in tissue and 7.7 lm in silicon,

resulting in a further peak in the energy spectrum obtained

from the irradiation of samples, in addition to the peaks

from the 10B(n,a)7Li reactions discussed above. The range

of the charged reaction products is shorter than 11 lm in

tissue; this means that even with tissue samples of only a

few tenths of micrometer, the charged particles loose part

of their energy in the samples and, thus, deposit different

residual energies in the detector depending on their point of

origin and their flight directions. For this reason, the

spectra of any particle type measured by the silicon

detector are not Gaussian but show a typical shape (Fig. 4).

Calculation of the boron concentration

The basic idea is to select a part of the energy spectrum

obtained by the irradiation of a tissue sample and to

establish a correlation between the total events in this

energy range and the concentration in tissue. In order to

obtain this correlation, it is necessary to calculate the tissue

volume where the events collected in the selected energy

range were generated (Fig. 5). The fact that the spectra

present an absorbed profile implies that the contribution of

different particles overlaps in most parts of the energy

range; thus, it is not possible to separate the different

components. However, it is possible to select a range in the

histogram where the collected events are only a-particles.

The interval chosen is DE between 1,100 and 1,350 keV. It

is important to note that the limiter present in front of the

sample (Fig. 2) is not meant to collimate the reaction

products but to define a certain area in the sample. The

particles emitted from that area that reach the detector and

that show a residual energy in the energy range specified

above are then used to infer the boron concentration. The

error associated with the presence of particles flying in

other directions than the parallel one will be discussed

below.

The integral of the histogram between these energy

values corresponds to the number of particles that left the

tissue sample with a residual energy between 1,100 and

1,350 keV. Using the relation between the residual energy

Table 1 Principal reactions

with positive Q-values that

contribute to charged particle

emission in biological tissue,

during thermal neutron

irradiation

Reaction Thermal microscopic

cross section (barn)

Q-value

(keV)

Isotopic

abundance

(%)

Elemental percentage by

weight of lung tissue (ICRU 46)

14N(n,p)14C 1.8 630 99.634 3.117O(n,a)14C 0.24 1,800 0.038 74.932S(n,p)33P 0.002 530 95.02 0.332S(n,a)29Si 0.007 1,500 95.02 0.333S(n,a)30Si 0.2 3,500 0.75 0.335Cl(n,p)35S 0.4 620 75.77 0.340K(n,p)40Ar 4.0 930 0.012 0.210B(n,a)7Li 3,837 2,790 19.9 \0.5 ppm

Fig. 4 Example of charged particle spectra of tissue samples; dotted

line background without tissue sample, dashed line tissue sample

without boron, solid line tissue sample with boron


123

of a-particles and the distance covered in tissue, the depths

x and x ? Dx in the tissue corresponding to the residual

energies 1,350 and 1,100 keV can be calculated, allowing

for the determination of the volume of the sample in which

these particles were produced. The cross section of neutron

capture in boron and the geometrical efficiency of the

apparatus are then used to obtain the boron concentration in

that sample volume. Assuming that boron is uniformly

distributed in the analyzed sample, the concentration in the

calculated subvolume is equal to the concentration in the

whole tissue section. If the sample is irradiated with a uniform

thermal neutron field, the a-particles are isotropically emitted

in all directions and from different depths in the tissue. In the

following calculations, only the particles moving parallel to

the x-axis toward the negative values will be considered, as

illustrated in Fig. 5. If E0 is the initial energy of the particle

and x the tissue depth where it was produced, the residual

energy Eres at the position where the particle leaves the tissue

and that is deposited in the detector is:

EresðxÞ ¼ E0 � Edissðx! 0Þ ¼ E0 �Z0

x

dE

dxdx ð1Þ

Similarly, if the particle comes from a depth x ? Dx, its

residual energy is:

Eresðxþ DxÞ ¼ E0 � Edissðxþ Dx! 0Þ ¼ E0 �Z0

xþDx

dE

dxdx

ð2Þ

The interval of the residual energy of all the particles that

come from the depth interval Dx is thus:

Eresðxþ DxÞ � EresðxÞ ¼ DEres ¼Z0

x

dE

dxdx�

Zx

xþDx

dE

dxdx

ð3Þ

The number N of a-particle events collected in the interval

between Eres (x ? Dx) and Eres(x) depends on the number

of reactions that took place in the depth interval Dx.

Defining K as the number of events per time interval and

per energy interval:

K ¼ N

DE � Dtð4Þ

The following relations hold:

K � DE

g¼ N

g � Dt¼ R � U � DV ¼ n � r � U � S � Dx ð5Þ

where g is the measurement efficiency (geometrical and

detector efficiency); R is the macroscopic cross section of

the boron neutron capture reaction; r is the microscopic

cross section of the boron neutron capture reaction; U is

the thermal neutron flux in n/(cm2 s); n is the number of

nuclei of 10B per unit volume in the sample; S is the

surface of the sample that the detector sees through the

collimator; and DV is the volume of the sample

considered.

From Eq. (5), the nuclear boron density in the sam-

ple is derived provided it is possible to measure K

(Eq. 6):

n ¼ K

g � r � / � SDE

Dxð6Þ

Finally, the boron concentration in the sample is given by:

mB

mt

¼ n � DVAw

NA

� 1

mt

¼ K

g � r � / � S �DE

DðqxÞ �AW

NA

where DV is the volume of the sample emitting a-particles

arriving in DE; mB is the 10B mass in DV ; mt is the tissue

mass in DV; Aw is the atomic weight of boron; and NA is

the Avogadro number.

The quantity to be calculated is the concentration in

fresh tissue. However, the present technique uses thin

sections put under vacuum that rapidly dry after the sec-

tioning. Thus, the concentration in fresh tissue could only

be directly obtained if it would be possible to keep the

characteristics of the tissues during the measurements

unchanged. Let the quantities with subscript ‘‘f’’ relate to a

Fig. 5 Measurement principle. Left a-particles have a maximum

range in tissue of R0, and those that are produced between x and

x ? Dx and fly in a direction perpendicular to the detector surface

reach the detector with an energy between E and E ? DE. The

detector collects these events, and the integral of the curve between

E and E ? DE is N (right)


123

fresh sample, while those with subscript ‘‘d’’ relate to a dry

sample. Then Eq. (6) becomes:

nf ;d ¼Kf;d

g � r � / � Sf;d

DEf;d

Dxf;dð7Þ

Using the quantities measured in the dry samples, it is

possible to obtain information on the fresh tissue, as

described in the following considerations.

Besides (Eq. 7), also the following Eq. (8) holds:

nf;d ¼mBf;d

Vf;d� NA

AW

ð8Þ

Under the assumption that no boron loss occurs during the

drying process, mBf = mBd, Eq. (9) follows.

nf

nd

¼ mBf

mBd

� Vd

Vf

¼ Vd

Vf

¼ Sd � Dxd

Sf � Dxf

) nf ¼ nd

Sd � Dxd

Sf � Dxf

ð9Þ

Using Eq. (6) for the dry tissue and multiplying and

dividing by qd (density of dry tissue) leads to Eq. (10).

nf ¼Kd

g � r � / � Sd

� DEd

DðqdxdÞ� mtd

Vf

ð10Þ

The boron concentration in the fresh tissue (concF) is then

given by:

concF ¼mBf

mtf

¼ nf � Vf

Aw

NA

� 1

mtf

which is connected to the quantities measured in the dry

samples through Eq. (10):

concF ¼Kd

g � r � / � Sd

� DEd

DðqdxdÞ� AW

NA

� mtd

mtf

ð11Þ

Hence, all the quantities needed for the determination of

the boron concentration in fresh tissue can be measured in

the dry samples, including two factors, (a) Kd which is the

net integral of the count rate in the selected DEd measured

in the dry sample and (b) g � r � / � Sd which is a factor

depending on the thermal neutron flux, the reaction cross

section and the counting efficiency of the apparatus.

The second factor can be obtained from a measurement

of the standard NIST (see Fig. 4), for which the superficial

density of 10B is certified (N = (1.018 ± 0.035) 9 1015

at./cm2). The rate R of events recorded under the a-peaks

is:

R ¼ g � R/V ¼ g � r � / � ntot

Scoll

Stot

) g � r � / � Scoll

¼ R � Stot

ntot

This relationship takes into account that only a part of

the standard surface is seen by the detector (Scoll),

because of the presence of the collimator. This method

implies that it is not necessary to measure the neutron

flux and the efficiency of the system. Moreover, the

reaction rate is measured using the same reaction that is

also employed for the measurement of boron concentra-

tion. The surface of the section of dry tissue where the

reactions take place is defined by the collimator. It is

important to note that the surface of the dry section is the

same as that of the fresh one because of the strong

adhesion of the tissue to the Mylar support. This fact has

been verified by microscope observation. If the section

has a surface smaller than the collimator aperture, the

area of the sample (Scoll) is measured using a stereomi-

croscope connected to a camera employing image anal-

ysis software (Image Pro-Plus).

The next factor in Eq. (11), DEd

DðqdxdÞ, i.e., the energy lost

per path length unit by a-particles in dry tissue of the

thickness DðqdxdÞ corresponding to the selected energy

interval DEd, was experimentally measured for the tissue of

interest as described in (Stella et al. 2009) using thin tissue

sections and an a-source of 241Am. Otherwise, it can be

obtained by the Stopping and Range of Ions in Matter

(SRIM) (Ziegler et al. 1985).

Finally, mtd

mtfin Eq. (11), which is the ratio of the mass

of the dry tissue sample and the corresponding mass of

the fresh one before water loss, can be measured for each

kind of tissue analyzed as follows: When preparing the

samples for the irradiation, some sections are cut,

deposited on aluminum foils and immediately weighted

with a digital scale connected to a personal computer. The

weight of the sections is recorded every second, and the

diagram of the weight change is plotted as a function of

time. When the weight is stable, the ratio of the dry to

fresh mass is determined [see Fig. 8 in (Gadan et al.

2012)].

Boron concentration in nonuniform samples

The described strategy holds for homogeneous samples

such as healthy tissues, where boron is usually uniformly

distributed. When in the same section different histological

types of tissues are simultaneously present, however, such

as viable tumor tissue, normal tissue, fibrotic or necrotic

tissue, this technique alone cannot separate the boron

concentration in the various components. In fact, from the

residual energy spectrum, it is not possible to determine

whether the a-particles came from a tumor or a healthy

area. Rather, the described measurement for a mixed

sample provides an average boron distribution depending

on the percentage of the various tissues present in the

irradiated section. Thus, it is necessary to couple charged

particle spectroscopy with a histological preparation and an

image of the boron distribution in the section obtained by

neutron autoradiography (Altieri et al. 2006, 2008). To this


123

end, three contiguous cryostatic sections are prepared: the

first one is used for standard hematoxylin–eosin staining,

the second is deposited on Mylar for the spectroscopy,

and the third is deposited on a solid-state nuclear track

detector (CR-39). The histological preparation with

standard hematoxylin–eosin evidences the morphological

characteristics of the analyzed section such as the per-

centage of viable tumor cells and the possible presence of

necrosis or fibrosis. The images of boron distribution are

obtained after the irradiation of CR-39 detectors and a

proper etching with NaOH solution, where darker areas

correspond to higher boron concentrations than lighter

areas. Comparing the boron distribution images with the

histological images, it is possible to verify whether boron

uptake is higher in the tumor and to delimit its area. In this

case, the area of the tumor with respect to the whole

section area is measured using software for image analysis

(Fig. 6).

A spectroscopic measurement of a healthy sample is

necessary, in order to properly weight the contribution to

the spectrometry of the tumor in case of mixed samples. If

only tumor and normal parenchyma are present in the

sample, being b = VT/Vtot the tumor-to-total volume ratio

for the sample inside the collimator, the concentration in

the mixed sample is:

concM ¼ b � concT þ ð1� bÞ � concH ð12Þ

where concM, concT and concH are the average

concentrations obtained in the mixed sample, the tumor

and healthy tissue, respectively. Thus, the concentration in

the tumor is obtained by:

concT ¼concH

b� concM

concH

� ð1� bÞ� �

ð13Þ

A previous measurement on a healthy sample obtained

from the same animal gives concH; then it is possible to

determine boron concentration in the viable tumor using

Eq. (12).

Results and discussion

Precision of the measurement method

To investigate the validity of the present method and its

uncertainties, Monte Carlo calculations were performed

using the Transport of Ions in Matter (TRIM) tool of the

simulation environment the Stopping and Range of Ions in

Matter (SRIM). For this, a slab of tissue with a thickness of

1 mg/cm2 (which is the range of protons with the energy of

588 keV) was simulated, where the charged particles were

generated as a source. A layer of gold (thickness 40 lg/

cm2) was also simulated to take into account the energy

loss inside the electrode at the entrance of the detector.

These layers were put at the same distance as in the

experimental setup. Between the sample and the detector, a

collimator was inserted, in contact with the tissue. The

starting coordinates of protons, a and 7Li ions, their ener-

gies and their flight directions were randomly extracted

using an external Monte Carlo program. The characteristics

of these particles whose directions hit the detector area

were stored in a file written in a standard input format to be

read by SRIM as an input. For each transported particle, the

output file reported the initial characteristics, the residual

energy and the direction cosines after the transport in the

tissue. In this way, the information concerning the particles

after the transport was kept in correlation with the starting

points inside the tissue.

Firstly, as the total spectrum in the detector includes the

spectra due to protons, a-particles and lithium ions, these

different components were separately simulated and sum-

med over the whole energy range. A boron concentration of

13 ppm was assumed, in order to compare the simulated

total spectrum with the experimental one obtained from a

tissue sample that gave 13 ppm as a result of the mea-

surement. The results of the simulations were broadened

using a Gaussian function, in order to reproduce the reso-

lution of the detector (Fig. 7). Figure 7 demonstrates a

Fig. 6 Procedure to separate

the boron concentration in

tumor and in healthy tissues by

measuring a mixed sample. The

areas of the different kinds of

tissues are measured by image

analysis, and the boron

concentration in healthy tissue

is obtained by a previous

measurement of a healthy

uniform tissue. The neutron

autoradiography of a contiguous

section is analyzed in order to

prove that in the tumor areas,

the boron concentration is

higher


123

good agreement between the calculated and the measured

spectra.

The TRIM simulation tool was also used to investigate

the accuracy of the method to determine the thickness

Dx where the particles with residual energy in the chosen

DE are emitted. Determination of this thickness Dx could

be impaired by different effects, in particular by the

direction of the particles crossing the tissue and by the

contribution to DE due to a-particles of the two considered

energies (1,470 and 1,780 keV). As for the latter, it is clear

that the Dx values corresponding to the two energy com-

ponents that have residual energy in this DE are different,

as shown in Fig. 8.

Thus, in the energy range considered, there are events

coming from a deeper region, and the correct procedure

consists in subtracting these events from the integral of the

spectrum and correcting the number of reactions for the

branching ratio 0.94 (see section ‘‘Energy calibration’’ in

‘‘Materials and methods’’). Considering all the particles as if

they were of low energy without correcting for 0.94 overes-

timates the result by 0.5 %.

Another important issue concerns the particles that travel in

tissue following paths other than the orthogonal ones, thus

depositing more energy and contributing to other ranges in the

residual energy histogram. This fact leads to an error in

assigning the proper Dx to the chosen DE. The results of the

simulation performed with SRIM were therefore used to check

the energy distribution of the particles produced in the volume

considered. The Dx distribution shown in the upper part of

Fig. 8 was used, and the distribution of a-particles produced

there was superimposed to the total spectrum broadened with

the Gaussian function that describes the resolution of the

detector. It turned out that due to the directions of flight of the

particles travelling in tissue, some of the particles starting from

the chosen Dx do not fall into the chosen interval of residual

energy. The difference between the integral of the a-spectrum

Fig. 7 Comparison between simulated and experimental spectra for a

tissue sample with 13.3 ppm of 10B. The vertical dotted lines show

the part of spectrum used for boron concentration evaluation

Fig. 8 Histogram of the

distance from the sample

surface where a-particles with

residual energy between 1,100

and 1,350 keV are produced:

top—a with an initial energy of

1,470 keV; middle—a with an

initial energy of 1,780 keV;

bottom—total


123

between the energy limits indicated and the total number of

particles coming from Dx is 4 % (Fig. 9).

To summarize, the boron concentration present in a

homogeneous sample can be measured with the technique

described here with a precision that depends mainly on the

accuracy with which the following factors can be deter-

mined: (a) the number of counts in the energy interval DE,

(b) the areas of the healthy tissue and the tumor in the

sample, (c) the thickness (Dqx) of the tissue within the

sample that contributes to the signal in the energy range

DE selected for the analysis and (d) the dry-to-fresh mass

ratio of the tissue sample.

As for (a), the measurement time is set such that the

collected number of events is sufficient to get an error

less than 1 %. As for (b), the areas of the healthy tissue

and of the tumor are measured several times, and the

results thus represent average values. Typically, the

error associated with these average values is of the

order of 0.5 %. As for (c), as described in (Stella et al.

2009), the error associated with the determination of

Dx is around 17 %. Finally, regarding (d) in order to

determine the dry-to-fresh mass ratio, in the experi-

mental routine at least 10 sections are measured with a

digital scale, the values are averaged, and the corre-

sponding standard deviation is usually less than 5 %.

Combination of these independent sources of errors

results in an overall error of about 18 %.

Experimental validation

The method was also tested by experimental measurements

of tissue standards with known boron concentration. The

samples were prepared as described in (Gadan et al. 2012),

mixing a suspension of hepatic cells with a BPA solution at

different known concentrations. The suspension was then

frozen and sectioned as a normal biopsy, and the sections

were deposited on Mylar and irradiated in the described

setup. Five boron concentrations were tested, and for each

at least three samples were measured. As can be seen in

Table 2, the expected and measured values agreed within

6–11 %, with a standard deviation less than 3 % except in

the case of the control samples, where the concentration is

low and, accordingly, the technique is less precise.

Conclusions

The presented technique to measure boron concentrations

in biological tissues was developed to quantify the boron

content in the different components of tissues that must be

irradiated for BNCT. Most of the existing methods

described in the ‘‘Introduction’’ measure boron concentra-

tions in macroscopic samples and provide average con-

centration values that strongly depend on the percentage of

viable tumor cells present in the analyzed sample. In

contrast, the technique described here offers the possibility

to measure macroscopic samples (with a surface of the

order of tenth of mm2) and to separate the boron concen-

tration values in tumor cells from those in normal (healthy)

cells. Moreover, it allows taking into account the presence

of necrosis, characterized by cell death and thus by the

absence of boron. Of course, the concept of a-spectrometry

coupled to neutron autoradiography and histology repre-

sents a complex procedure, requiring time and resources.

Furthermore, it can be performed only when it is possible

to obtain biopsies from patients, after boron administration.

Nevertheless, when possible, it offers an insight regarding

boron biodistribution that cannot be obtained by any other

technique, thus allowing a more precise dosimetry and a

better understanding of the irradiation therapy outcome.

The error associated with the proposed procedure is about

20 %, with a major contribution due to the limited preci-

sion of the thickness Dx associated with the selected energy

range DE. This, in turn, depends on the uncertainties of the

Fig. 9 Solid line energy distribution of a-particles broadened by a

Gaussian function; dashed line energy distribution of a-particles

coming from the Dx shown in the upper part of Fig. 8 (0.06–0.2 mg/

cm2)

Table 2 Results of boron concentration measurements by a-spectrometry of standard tissue samples with known boron content

True 10B concentration (ppm) 0.0 (Control) 12.0 ± 1.2 23.1 ± 2.3 33.4 ± 3.4 51.5 ± 5.2

Measured 10B concentration (ppm) 0.5 ± 0.1 13.0 ± 2.0 25.0 ± 5.0 38.0 ± 8.0 54.0 ± 10.0

SD (5 samples for each concentration) (ppm) 0.2 0.2 0.8 0.8 1.5


123

residual energy as a function of the distance travelled by a-

particles below 2 MeV in tissues below a thickness of

0.7 mg/cm2 (Stella et al. 2009). In particular, in the mea-

surement of the residual energy of a-particles in tissue, the

major source of error was associated with the determina-

tion of the thickness of the tissue sections prepared for the

experiment. The relation that was obtained by Stella et al.

is:

DðqxÞ mg

cm2

h i¼ ð4:7� 0:1Þ � 10�4Eres½keV� þ ð1:03

� 0:03Þ

From this equation, the energy range between 1,100

and 1,350 keV corresponds to a tissue thickness of

Dqx = 0.12 ± 0.02 mg/cm2 with a relative error of

about 17 %. The study presented here demonstrates that

the other contributions to the overall error associated

with the measurement of boron concentration are all

below 5 %. Thus, the precision of the results could be

significantly improved if the experimental error of the

residual energy as a function of the distance travelled in

tissue could be reduced. To this end, a new experiment

has been planned to measure the described curve with

lower uncertainties.

The described boron measurement procedure may be

time-consuming, but it offers a deep insight into the

behavior of the boronated carrier in the tissues. This could

be exploited also to characterize new boron carriers, as one

of the most active branches of BNCT research is the

development of new vectors able to concentrate boron in

tumor cells, ensuring higher concentration ratios than those

achieved by the use of BPA.

More generally, the availability of a precise knowledge

of tumor-to-normal boron concentration ratios and of the

concentration in each tissue component is a prerequisite to

ensure an effective dose delivery, in order to exploit the

selective boron uptake. An accurate dosimetry is a very

complicated task in BNCT because of the mixed radiation

field (a, Li, p and c) and the boron biodistribution in the

target. Moreover, the analysis of the radiation effects both

for tumor and for normal tissues can be fully understood

only if they are strictly correlated to the delivered radiation

dose. It is our opinion that a more precise insight into the

boron distribution in biological tissues is needed to

improve BNCT and to allow its application becoming a

routine treatment in the next future. When biopsy is pos-

sible, boron concentration measurement by charged parti-

cle spectrometry coupled with imaging by neutron

autoradiography offers a relatively easy and precise

method to fulfill this requirement.

Acknowledgments The authors would like to thank Mr Piero

Bruschi for his invaluable technical contribution.

References

Agosteo S, Birattari C, D’Angelo G, Dal Corso F, Foglio Para A,

Lippi I, Pola A, Zotto P (2003) Neutron spectrometry with a

recoil radiator-silicon detector device. Nucl Instrum Methods A

515:589–604

Altieri S, Bortolussi S, Bruschi P, Fossati F, Vittor K, Nano R,

Facoetti A, Chiari P, Bakeine J, Clerici A, Ferrari C, Salvucci O

(2006) Boron absorption imaging in rat lung colon adenocarci-

noma metastases. J Phys Conf Ser 41:484–495

Altieri S, Bortolussi S, Bruschi P, Chiari P, Fossati F, Stella S, Prati

U, Roveda L, Zonta A, Zonta C, Ferrari C, Clerici A, Nano R,

Pinelli T (2008) Neutron autoradiography imaging of selective

boron uptake in human metastatic tumours. Appl Radiat Isot

66(12):1850–1855

Barth RF, Coderre JA, Graca H, Vicente M, Blue TE (2005) Boron

Neutron Capture Therapy of cancer: current status and future

prospects. Clin Cancer Res 11:3987–4002

Bortolussi S, Bakeine JG, Ballarini F, Bruschi P, Gadan MA, Protti N,

Stella S, Clerici A, Ferrari C, Cansolino L, Zonta C, Zonta A,

Nano R, Altieri S (2011) Boron uptake measurements in a rat

model for Boron Neutron Capture Therapy of lung tumours.

Appl Radiat Isot 69:394–398

Chandra S (2003) SIMS ion microscopy as a novel, practical tool for

subcellular chemical imaging in cancer research. Appl Surf Sci

203–204:679–683

Chiaraviglio D, De Grazia F, Zonta A, Altieri S, Braghieri A, Fossati

F, Pedroni P, Pinelli T, Perotti Specchairello A, Perlini G, Rief H

(1989) Evaluation of selective boron absorption in liver tumours.

Strahlenther Onkol 165(2/3):170–172

Coderre JA, Chanana AD, Joel DD, Elowitz EH, Micca PL,

Nawrocky MM, Chadha M, Gebbers JO, Shady M, Peress NS,

Slatkin DN (1998) Biodistribution of boronophenylalanine in

patients with glioblastoma multiforme: boron concentration

correlates with tumor cellularity. Radiat Res 149:163–170

Ferrari C, Zonta C, Cansolino L, Clerici AM, Gaspari A, Altieri S,

Bortolussi S, Stella S, Bruschi P, Dionigi P, Zonta A (2009)

Selective uptake of p-boronophenylalanine by osteosarcoma

cells for boron neutron capture therapy. Appl Radiat Isot

67:341–344

Fukuda H, Honda C, Wadabayashi N, Kobayashi T, Yoshino K,

Hiratsuka J, Takahashi J, Akaizawa T, Abe Y, Ichihashi M,

Mishima Y (1999) Pharmacokinetics of 10B-p-boronophenyl-

alanine in tumors, skin and blood of melanoma patients: a study

of boron neutron capture therapy for malignant melanoma.

Melanoma Res 9:75–83

Gadan MA, Bortolussi S, Postuma I, Ballarini F, Bruschi P, Protti N,

Santoro D, Stella S, Cansolino L, Clerici A, Ferrari C, Zonta A,

Zonta C, Altieri S (2012) Set-up and calibration of a method to

measure 10B concentration in biological samples by neutron

autoradiography. Nucl Instrum Methods B 274:51–56

Imahori Y, Ueda S, Ohmori Y, Sakae K, Kusuki T, Kobayashi T,

Takagaki M, Ono K, Ido T, Fujii R (1998) Positron emission

tomography-based boron neutron capture therapy using boron-

ophenylalanine for high-grade gliomas: part I. Clin Cancer Res

4:1825–1832

Menendez P, Roth BM, Pereira MD, Casal MR, Gonzalez SJ, Feld

DB, Santa Cruz GA, Kessler J, Longhino J, Blaumann H,

Jimenez Rebagliati R, Calzetta Larrieu OA, Fernandez C, Nievas

SI, Liberman SJ (2009) BNCT for skin melanoma in extremities:

updated Argentine clinical results. Appl Radiat Isot 67(7–8):

50–53

Probst TU (1999) Methods for boron analysis in boron neutron

capture therapy (BNCT). Rev Fresenius J Anal Chem 364:

391–403


123

Protti N, Bortolussi S, Stella S, Gadan MA, De Bari A, Ballarini F,

Bruschi P, Ferrari C, Clerici AM, Zonta C, Bakeine JG, Dionigi

P, Zonta A, Altieri S (2009) Calculations of dose distributions in

the lungs of a rat model irradiated in the thermal column of the

TRIGA reactor in Pavia. Appl Radiat Isot 67:210–213

Protti N, Bortolussi S, Prata M, Bruschi P, Altieri S, Nigg D (2012)

Neutron spectrometry for the University of Pavia TRIGATM

thermal neutron source facility. Trans Am Nucl Soc 107:

1269–1272

Riley KJ, Harling O (1998) An improved prompt gamma neutron

activation analysis facility using a focused diffracted neutron

beam. Nucl Instrum Methods B 143(3):414–421

Stella S, Bortolussi S, Bruschi P, Portella C, Altieri S (2009)

Measurement of a particle energy loss in biological tissue below

2 MeV. Nucl Instrum Methods B 267:2938–2943

Verbakel WFAR, Sauerwein W, Hideghety K, Stecher-Rasmussen F

(2003) Boron concentrations in brain during boron neutron

capture therapy: in vivo measurements from the phase I trial

EORT 11961 using a gamma-ray telescope. Int J Radiat Oncol

Biol Phys 55(3):743–756

Wittig A, Michel J, Moss RL, Stecher-Rasmussen F, Arlinghaus HF,

Bendel P, Mauri PL, Altieri S, Hilger R, Salvadori PA,

Menichetti L, Zamenhof R, Sauerwein WA (2008) Boron

analysis and boron imaging in biological materials for Boron

Neutron Capture Therapy (BNCT). Crit Rev Oncol Hematol

68:66–90

Ziegler JF, Biersack JP, Littmark U (1985) The stopping and range of

ions in solids. Pergamon Press, Oxford. Code available on

http://www.srim.org

Zonta A, Pinelli T, Prati U, Roveda L, Ferrari C, Clerici AM, Zonta

C, Mazzini G, Dionigi P, Altieri S, Bortolussi S, Bruschi P,

Fossati F (2009) Extra-corporeal liver BNCT for the treatment of

diffuse metastases: what was learned and what is still to be

learned. Appl Radiat Isot 67:67–75


123

http://www.srim.org

Boron concentration measurement in biological tissues by charged particle spectrometry

Documents