Radiation Protection Dosimetry (2011), Vol. 143, No. 2 –4, pp. 523 –527 doi:10.
1093/rpd/ncq466
Advance Access publication 15 December 2010
FROM RADIATION-INDUCED CHROMOSOME DAMAGE TO
CELL DEATH: MODELLING BASIC MECHANISMS AND
APPLICATIONS TO BORON NEUTRON CAPTURE THERAPY
F. Ballarini 1,*, S. Bortolussi 1, A.M. Clerici 2, C. Ferrari 2, N. Protti 1 and S. Altieri 1
1
Department of Nuclear and Theoretical Physics, University of Pavia, INFN-Pavia, via Bassi 6, I-27100
Pavia, Italy
2
Department of Surgery, Experimental Surgery Lab, University of Pavia, I-27100 Pavia, Italy
Cell death is a crucial endpoint in radiation-induced biological damage: on one side, cell death is a reference endpoint to
Downloaded from http://rpd.oxfordjournals.org/ at CERN Library on April 11, 2013
characterise the action of radiation in biological targets; on the other side, any cancer therapy aims to kill tumour cells.
Starting from Lea’s target theory, many models have been proposed to interpret radiation-induced cell killing; after briefly
discussing some of these models, in this paper, a mechanistic approach based on an experimentally observed link between
chromosome aberrations and cell death was presented. More specifically, a model and a Monte Carlo code originally devel-
oped for chromosome aberrations were extended to simulate radiation-induced cell death applying an experimentally observed
one-to-one relationship between the average number of ‘lethal aberrations’ (dicentrics, rings and deletions) per cell and – ln S,
S being the fraction of surviving cells. Although such observation was related to X rays, in the present work, the approach was
also applied to protons and alpha particles. A good agreement between simulation outcomes and literature data provided a
model validation for different radiation types. The same approach was then successfully applied to simulate the survival of
cells enriched with boron and irradiated with thermal neutrons at the Triga Mark II reactor in Pavia, to mimic a typical treat-
ment for boron neutron capture therapy.
INTRODUCTION acting via a multiple-track mechanism. The par-
ameter p is the assumed proportionality constant
Lea’s ‘target theory’ is one of the earliest interpretive
between cell death and the yield of DSBs, and h and
models for radiation-induced cell killing(1).
k are the probability constants for chemical bond
According to its ‘multitarget-single-hit’ extension,
rupture per bond and per unit dose (via single- or
the cell contains n critical targets, each target has
multiple-track mechanism, respectively). Finally, x
the same probability q of being hit by radiation, and
; n0f0 and f ; En1n2f1f2f0, where n1 and n2 are the
one-hit is sufficient to inactivate a target but not the
number of critical bonds on strands 1 and 2, n0 is
cell. The surviving probability after a dose D is then
the number of sites that can sustain a DSB, f1 and f2
are the fractions of unrestored bonds in strands 1
SðDÞ ¼ 1 ð1 eqD Þn ð1Þ and 2, f0 is the fraction of unrepaired DSBs, and E
the probability of a rupture occurring from two
The fact that the slope at zero dose is zero is a single-strand breaks (SSBs) associated in time and
major limitation for this model. This led to alterna- space. According to the usual Poisson-type cell
tive approaches including the ‘molecular model’(2), killing, cell survival is then given by S¼e – Q.
also called ‘linear-quadratic’ (LQ) model. The LQ Assuming that k and h are quite small, one gets the
approach assumes that the integrity of the DNA familiar LQ relationship:
double helix is essential for clonogenic survival and
that a double-strand break (DSB) is a critical S ¼ expðaD bD2 Þ ð3Þ
damage. Ionising radiation can cause DSBs and
eventually cell death. A DSB is caused either when
both DNA strands are broken by a single radiation where a ¼( f0, n0, h, D) and b ¼( f0, E, n1, n2, f1, f2,
track or when each strand is broken independently. k2, (12D)). Although the fundamental assumptions
The average number of lethal DSBs per cell follow- are not widely accepted, the LQ model is widely
ing a dose D is then used in radiobiology since in general it fits mamma-
lian cell survival data pretty well.
In 1972, Kellerer and Rossi (3) proposed the
Q ¼ p½xð1 ehDD Þ þ wð1 ekð1DÞD Þ2 ð2Þ Theory of Dual Radiation Action (TDRA), which is
based on the following assumptions: (1) ionising
Here, D is the fraction of dose acting via a single- radiation induces cellular ‘sub-lesions’, which are
track mechanism and 12D the fraction of dose proportional to the dose; (2) the interaction between
# The Author 2010. Published by Oxford University Press. All rights reserved. For Permissions, please email: [email protected]
F. BALLARINI ET AL.
two sub-lesions can produce a ‘lesion’, which has a presented, starting from an experimental study indi-
certain probability to lead to cell death. Such inter- cating a one-to-one relationship between – ln S and
action, which is possible only within a ‘sensitive the yield of ‘lethal aberrations’ (Giemsa-stained
site’, has 100 % probability for distances smaller dicentrics, rings and deletions) induced in AG1522
than the sensitive site dimensions (which according fibroblasts exposed to X rays(5). A mechanistic
to TDRA are of the order of the micrometre), model and a Monte Carlo code originally developed
whereas it is zero for larger distances. These assump- for radiation-induced chromosome aberrations were
tions lead to the following (LQ) expression for the extended to model cell death. The approach was
survival probability: applied not only to X rays but also to protons, alpha
particles and a mixed field of protons and alphas
S ¼ exp½kðzD þ D2 Þ ð4Þ that mimics the exposure scenario for boron neutron
capture therapy (BNCT), where cells loaded with
10
B are exposed to thermal neutrons.
Here, k reflects a biological property of the system,
since E(z)¼kz 2, where E is the average number of
lesions and z the specific energy, i.e. the energy A MODEL AND CODE FOR CHROMOSOME
Downloaded from http://rpd.oxfordjournals.org/ at CERN Library on April 11, 2013
imparted
Ð per event
Ð and per unit mass. Furthermore, ABERRATION INDUCTION
z¼ z21f (z1)dz1/ z1f (z1)dz1, where f (z1) is the distri-
More than 10 y ago(6), a mechanistic model and a
bution of single-event specific energies. This theory
Monte Carlo code for radiation-induced chromo-
has been criticised on the basis of ‘event-by-event’
some aberrations were initiated, relying on the
track-structure simulations, according to which the
following basic assumptions: (1) chromosome aber-
interaction between distinct events occurs at the
rations arise from DNA lesions that are clustered at
nanometre—and not the micrometre—level.
the nanometre level (cluster lesions, CLs), each CL
However, an alternative interpretation may be the
giving rise to two independent chromosome free-
following: on the basis of the relationship between
ends; (2) only pairs of free ends initially created
some chromosome aberrations and cell death (see
within a threshold distance d can join and thus
below), sub-lesions can be thought as (complex)
produce exchange-type aberrations. These assump-
DSBs, whereas lesions can be thought as lethal
tions reflect the idea that among the many initially
chromosome aberrations. This way a sensitive site of
induced breaks of the double helix, only those that
the order of the micrometre becomes consistent with
are severe enough and close enough are likely to be
the data, since chromosome aberrations are pro-
involved in the formation of chromosome aberra-
duced by pairwise interaction of DSBs, and such
tions. The current version of the code can deal
interaction occurs between chromosome domains,
either with spherical cell nuclei or with cylindrical
which have linear dimensions of the order of the
nuclei. Interphase chromosome territories are mod-
micrometre.
elled as irregular intra-nuclear regions consisting of
A more recent approach, used at GSI for carbon-
the union of small adjacent cubic boxes. The volume
therapy treatment plan calculations, is the ‘Local
of each territory is proportional to the chromosome
Effect Model’, based on the assumption that the
DNA content. The main code input is the yield of
damage in a small sub-volume (nm) of the cell
radiation-induced CLs (i.e. average number of CLs
nucleus is solely determined by the energy depo-
per Gy and per cell), which represent those initial
sition in that sub-volume, independent of the radi-
DNA lesions that can ‘evolve’ into chromosome
ation type. According to this assumption, the
aberrations. The yield of CLs primarily depends on
average number of lethal events per cell for heavy
radiation quality, but it can also be modulated by
ions can be written as
the repair ability of the considered cell line.
ð Although in previous works on lymphocytes
Nion ¼ ln SX ðdÞVn dVn ð5Þ exposed to protons or alpha particles(7) the CLs
yields have been taken from ‘event-by-event’ radi-
ation track-structure simulations in which a CL had
where SX is the surviving fraction for X rays, d the been defined as ‘at least two SSBs on each DNA
local dose and Vn the cell nucleus volume. The strand within 30 base-pairs’(8), more generally such
current version of the model, which recently has yield may be considered as a semi-free parameter
been refined taking into account free-radical diffu- that takes into account not only the radiation
sion, DNA strand-break clustering and an extension quality but also the specific cell type.
of the inner part of the particle track(4), led to good For sparsely ionising radiation, the various CLs
agreement with survival data for various cell lines. are randomly distributed in the cell nucleus, whereas
In this paper, a modelling approach to cell death for light ions, the lesions are randomly distributed
based on the mechanistic relationship between some along segments representing the particle track. The
chromosome aberrations and cell death was CLs induced by heavy ions like carbon and iron (of
524
MODELLING CHROMOSOME DAMAGE AND CELL DEATH
interest for hadron therapy and space research, study concerns AG1522 fibroblasts exposed to X
respectively) are located partly along a segment rays (and subject to delayed plating to allow for
representing the core of the primary track and partly potentially lethal damage repair), in the present
with a ‘radial shift’ with respect to the core, to repro- work, the approach was also applied to protons,
duce the effects of delta rays. For a heavy-ion track, alpha particles and a mixed field of protons and
the probability of inducing a lesion at distance r alphas, the latter to mimic a BNCT scenario where
from the core is assumed to be proportional to r 22. thermal neutrons impinging on 10B-loaded cells
The subsequent simulation steps consist of: (1) give rise to protons and alpha particles following
identification of the chromosome(s) and chromo- neutron capture by 14N and 10B, respectively. The
some arm(s) hit by each CL; (2) pairwise rejoining first step of the work, reported elsewhere(14), con-
between chromosome free-ends, assuming 100 % sisted of reproducing the experimental outcomes on
rejoining probability if the (initial) distance between X-irradiated AG1522 cells(5). A very good agree-
the two free ends is ,d and no rejoining if the dis- ment between model predictions and experimental
tance is .d; (3) aberration scoring (Giemsa or data confirmed the important role of lethal aberra-
FISH); (4) repetition for a statistically significant tions for radiation-induced cell death. In this work,
number of irradiated cells; and (5) repetition for a very good agreement was found with other
Downloaded from http://rpd.oxfordjournals.org/ at CERN Library on April 11, 2013
different dose values to obtain a dose –response photon data, concerning V79 cells exposed to X
curve for the main aberration types (dicentrics, rays(15) or gamma rays(16). As a second step, the
translocations, rings, deletions and more than 40 approach was applied to a parallel field of monoe-
different complex exchanges), directly comparable nergetic protons or alpha particles. A very good
with experimental data. Specific background (i.e. agreement was found with literature data on V79
prior to irradiation) yields for different aberrations cells irradiated with 3.2 MeV alpha particles
can be included, and the scoring of chromosome (average LET: 120 keV mm21)(16). The agreement
fragments smaller than a threshold value can be was less satisfactory for V79 cells exposed to 0.64
‘switched off ’, since these fragments can hardly be MeV protons (35 keV mm21), probably due to
detected experimentally when chromatin is con- the fact that at this energy, the irradiation is not
densed. Up to now, the model has been validated for under track segment conditions, i.e. the LET is not
the induction of the main types of chromosome constant (17). Higher energies will be tested as a
aberrations in lymphocytes exposed to X rays or future development. Finally, the approach was
gamma rays(9), protons and alpha particles(7), extended to model the survival of DHDK12TRb
carbon ions and iron ions(10, 11). The agreement (DHD) cells enriched with 10B and exposed to
between model predictions and literature experimen- thermal neutrons, which in a boron-rich tissue
tal data supports the model assumptions on the deposit their energy mainly through the capture
mechanisms governing chromosome aberration reactions 10B(n,a)7Li and 14N(n,p)14C. The energies
induction, including a fundamental role for DNA of the involved particles are 1.47 MeV for alphas
damage clustering at the nanometre scale and a (average LET: 190 keV mm21), 0.84 MeV for 7Li
step-like distance dependence at the micrometre (160 keV mm21) and 0.59 MeV for protons (38
scale for the rejoining probability between two keV mm21). The effects of this treatment were
chromosome free-ends. Furthermore, the model has directly modelled by a mixed field of alpha par-
been applied to predict the induction of chronic ticles and protons. 7Li was treated like an alpha,
myeloid leukaemia following acute exposure to since with these LET values, it is unlikely that the
gamma rays(12) and the induction of chromosome biological effectiveness is significantly different. The
aberrations in astronauts exposed to space irradiation section of the code was purposely modi-
radiation(13). fied to simulate a mixed field of protons and
alphas, with each particle starting from a random
position inside the cell with a random direction, to
EXTENSION OF THE CODE TO CELL DEATH
mimic a uniform intra-cellular distribution of 10B
AND APPLICATION TO BNCT
and 14N. The results described above are reported
The model/code described in the previous section in Figure 1, which, from top to bottom, shows the
was extended to simulate radiation-induced cell simulated cell survival curves following exposure to
death, applying the experimentally observed(5) one- either photons, or 0.64 MeV protons, or a mixed
to-one relationship between the yield of ‘lethal field of 1.47 MeV alpha particles and 0.59 MeV
aberrations’ (i.e. Giemsa-stained dicentrics, rings protons (86 and 14 % in dose, to simulate thermal
and deletions) and –ln S, S being the fraction of neutron irradiation of boron-enriched cells), or 3.2
surviving cells. The fraction of surviving cells after MeV alpha particles. The lines are model predic-
a dose D was calculated as S(D)¼e – LA(D), where tions, the points are literature data on V79 cells
LA(D) is the (simulated) average number of lethal exposed to X rays(15), gamma rays(16), 0.64 MeV
aberrations per cell. Although the experimental protons(17) or 3.2 MeV alpha particles(16), and
525
F. BALLARINI ET AL.
Downloaded from http://rpd.oxfordjournals.org/ at CERN Library on April 11, 2013
Figure 1. Cell survival for photons, 0.64 MeV protons, a mixed field of 1.47 MeV alphas and 0.59 MeV protons, or 3.2
MeV alphas. The lines are model predictions and the points are experimental data(15 – 18). The cell nuclei were modelled as
cylinders with height 4 mm and radius 6 mm, and the following values were used as CL yields: 2.5 for photons, 3.6 for
0.64 MeV protons, 9.2 for 3.2 MeV alpha particles, 9.4 for 1.47 MeV alpha particles and 3.6 for 0.59 MeV protons.
DHD cells enriched with 10B and exposed to Yoshiya Furusawa, Maria Antonella Tabocchini and
thermal neutrons at the Triga Mark II research Ben Phoenix for useful discussion and data sharing.
reactor of the University of Pavia(18)
FUNDING
CONCLUSIONS
This work was partially supported by INFN
A model and a Monte Carlo code for chromosome ( project ‘WIDEST1’).
aberrations were extended to simulate cell death by
different radiation types, basing on a link between
aberrations and cell death observed for AG1522 cells REFERENCES
exposed to X rays. A very good agreement was 1. Lea, D. E. Actions of radiations on living cells, second
found with data on X rays, gamma rays and alpha edn. Cambridge University Press (1955).
particles, as well as with data on DHD cells loaded 2. Chadwick, K. and Leenhouts, H. A molecular theory
with 10B and exposed to thermal neutrons. This of cell survival. Phys. Med. Biol. 18, 78– 87 (1973).
suggests that the link between lethal aberrations and 3. Kellerer, A. M. and Rossi, H. H. A generalized formu-
lation of dual radiation action. Radiat. Res. 75,
cell death may hold not only for AG1522 cells
471– 488 (1978).
exposed to X rays, but also for other cell lines and 4. Elsasser, T., Kramer, M. and Scholz, M. Accuracy
other radiation types, including protons and alphas. of the local effect model for the prediction of
Preliminary calculations with carbon ions are biologic effects of carbon ion beams in vitro and
encouraging and will represent a major future devel- in vivo. Int. J. Radiat. Oncol. Biol. Phys. 71, 866– 72
opment of this work. (2008).
5. Cornforth, M. and Bedford, J. A quantitative compari-
son of potentially lethal damage repair and the rejoining
ACKNOWLEDGEMENTS of interphase chromosome breaks in low passage normal
human fibroblasts. Radiat. Res. 111, 385–405 (1987).
The authors are grateful to Michael Cornforth, 6. Ballarini, F., Merzagora, M., Monforti, F., Durante,
Marco Durante, Michael Scholz, Thomas Friedrich, M., Gialanella, G., Grossi, G.F., Pugliese, M. and
526
MODELLING CHROMOSOME DAMAGE AND CELL DEATH
Ottolenghi, A. Chromosome aberrations induced by 13. Ballarini, F. and Ottolenghi, A. A model of chromo-
light ions: Monte Carlo simulations based on a mechan- some aberration induction: applications to space
istic model. Int. J. Radiat. Biol. 75, 35–46 (1999). research. Radiat. Res. 164(4), 567– 570 (2005).
7. Ballarini, F. and Ottolenghi, A. Chromosome aberra- 14. Ballarini, F. et al. Nuclear physics meets medicine and
tions as biomarkers of radiation exposure: modelling biology: boron neutron capture therapy. In: Proceedings
basic mechanisms. Adv. Space Res. 31(6), 1557–1568 of the 12th International Conference on Nuclear
(2003). Reaction Mechanisms, Varenna, Italy, June 15–19,
8. Ottolenghi, A., Merzagora, M., Tallone, L., Durante, 2009, Cerutti, F. and Ferrari, A. eds. Vol. 2, pp. 561–
M., Paretzke, H.G. and Wilson, W. The quality of DNA 571, CERN, Geneva, 2010.
double-strand breaks: a Monte Carlo simulation of the 15. Furusawa, Y., Fukutsu, K., Aoki, M., Itsukaichi, H.,
end-structure of strand breaks produced by protons and Eguchi-Kasai, K., Ohara, H., Yatagai, F., Kanai, T.
alpha particles. Radiat. Environ. Biophys. 34, 239– 244 and Ando, K. Inactivation of aerobic and hypoxic cells
(1995). from three different cell lines by accelerated 3He-, 12C
9. Ballarini, F., Biaggi, M. and Ottolenghi, A. Nuclear and 20Ne-ion beams. Radiat. Res. 154, 485– 496 (2000).
architecture and radiation-induced chromosome aberra- 16. Phoenix, B., Green, S., Hill, M., Jones, B., Mill, A.
tions: models and simulations. Radiat. Prot. Dosim. 99, and Stevens, D. Do the various radiations present in
175 –182 (2002). BNCT act synergistically? Cell survival experiments in
Downloaded from http://rpd.oxfordjournals.org/ at CERN Library on April 11, 2013
10. Ballarini, F., Alloni, D., Facoetti, A. and Ottolenghi, A. mixed alpha-particle and gamma-ray fields. Appl.
Heavy-ion effects: from track structure to DNA and Radiat. Isot. 67, S318–S320 (2009).
chromosome damage. New J. Phys. 10, 075008, http:// 17. Belli, M. et al. RBE-LET relationships for cell inacti-
www.njp.org (2008). vation and mutation induced by low energy protons in
11. Ballarini, F. Chromosome damage by ionizing V79 cells: further results at the LNL facility. Int. J.
radiation: a review. Il Nuovo Cimento B 124(4), Radiat. Biol. 74, 501–509 (1998).
443 –458 (2009). 18. Ferrari, C. et al. Boron neutron capture therapy of liver
12. Ballarini, F. and Ottolenghi, A. A model of and lung coloncarcinoma metastases: an in vitro Survival
chromosome aberration induction and CML incidence at Study. In: Proceedings of the 13th ICNCT, Florence, 2–
low doses. Radiat. Environ. Biophys. 43, 165– 171 7 November 2008. Zonta, A., Altieri, S., Roveda, L.
(2004). and Barth, R., eds (Roma: ENEA) pp. 331– 336 (2008).
527
