International Journal of

Radiation Oncology
biology

physics

www.redjournal.org

Editorial

The Power of Radiation BiophysicsdLets Use It

J. Donald Chapman, PhD,* and Colin J. Gillespie, PhDy
*CRM Consulting Services, Penticton, BC, Canada; and yBFP Inc, Taylor McCaffrey LLP, Winnipeg, MB, Canada
Received Mar 20, 2012, and in revised form Apr 11, 2012. Accepted for publication Apr 12, 2012
written in a form that yields a straight line such that the 2
parameters are the intercept at 0 dose and the slope, respectively.

Two recent editorials in this Journal about the linear quadratic

(LQ) model (1, 2) have debated its use on essentially empirical
grounds. Missing are its roots in biophysics and what they state
about how it can (and cannot) be used. In the 1970s, 2 roads
led to analysis of in vitro cell killing by ionizing radiation in
terms of the underlying mechanisms. Kellerer and Rossi (3)
explored the statistics of energy deposition in small volumesdthe microdosimetric approach. Chadwick and Leenhouts
(4) related cell killing to lesions in 2-stranded DNAdthe
biophysical approach. Both approaches postulate that radiation
can kill cells by 2 distinct processes, a 1-hit mechanism and
another that requires 2 independent events. We began to use the
LQ equation in 1973 for the pragmatic reason that it fits the
survival data better than the other models we tested (5). It soon
emerged that, if key factors were controlled, this equation offers
insights into the mechanisms involved in cell death and survival
during and after irradiation. Here, we reviewed 4 forms of this
equation that describe cell survival in specific circumstances and
how they might help those working to improve radiotherapeutic
protocols.
The basic LQ equation states that the cell surviving fraction (S)
is the product of 2 Poisson escape probabilities. The mean
numbers of events for the underlying mechanisms are proportional
to the first and second powers of dose, D, respectively.



SZexpaD  exp b0 D2
LQ 1

The basic LQ equation applies to cell populations of sufficiently homogeneous radiosensitivity. Thus, much of our
research investigated the cell-survival and cell-killing mechanisms with synchronized cells in mitosis (tetraploid) or in G1
phase (diploid) of the cell cycle. Stationary (G0) phase cells
(diploid) were also used in experiments in which it was not
feasible to synchronize the cells. In this model, the a/b ratio
defines the dose at which the different mechanisms contribute
equally to cell killing.
However, most radiobiology research uses asynchronous
populations of normal or tumor cell lines whose survival can also
be characterized with an LQ equation. Although LQ Eq. 1 can be
fitted to such data, the resultant a and b parameters do not
precisely express the underlying mechanisms. Rather, they are
associations of values for subpopulations in which the range of
a values affects the fitted b values and vice versa (5). This is
reflected by a different form of the equation:

The sublesions associated with b0-inactivation (likely singlestrand breaks [6]) repair rapidly at mammalian body temperatures. Thus, to obtain precise values of a and b0, cells are
irradiated at dose rates >10 Gy/min or at 2 C-6 C, such that the
first sublesion of a 2-hit event will not be removed before a second
has been induced. The sources available to most radiobiology
research have dose rates of 1-2 Gy/min at positions convenient for
cell irradiation; thus, low temperatures should be used to obtain
the maximal and true value of b0. This basic LQ equation can be

where nx is the fraction of cell population x, with parameters ax

and b0x, and a and b0 are effective values for the mixed population. The mechanistic ambiguity of the parameter values from
mixtures of cells of different radiosensitivity should be clear. This
form of the equation will be useful for asynchronous cell populations, for mixtures of oxygenated and hypoxic cells, for
mixtures of quiescent and proliferating cells, for cells in planning
tumor volume voxels exposed to particles of the variable linear
energy transfer, and so forth. The variation in a for various human

Reprints requests to: J. Donald Chapman, PhD, CRM Consulting

Services, Penticton, BC V2A 8N1 Canada. Tel: (250) 493-7722; Fax: (250)
493-7722; E-mail: vino@vip.net
Int J Radiation Oncol Biol Phys, Vol. 84, No. 2, pp. 309e311, 2012
0360-3016/$ - see front matter 2012 Elsevier Inc. All rights reserved.
doi:10.1016/j.ijrobp.2012.04.020

lnS=DZa b0 D


SZexpaD  exp  b0 D2
X 



Z nx expax D  exp b0x D2

LQ 1a

LQ 2

Conflict of interest: none.

310

International Journal of Radiation Oncology  Biology  Physics

Chapman and Gillespie

tumor cell lines correlates well with the different radioresponses

q
of the tumors from which they were derived; in contrast, b0
varies only slightly among cell types (7, 8).
Cell killing by radiation was also characterized at intermediate
dose rates and different temperatures. These studies identified
a first-order repair process that accounts for the reduced cell
inactivation observed (9). This effect can be expressed in another
form of the basic equation:



SZexpaD  exp  b0 D2 f1  1  expm=mg=m
LQ 3
where a and b0 are unchanged, m is the ratio kD/R, k is the
measured rate constant for repair of sublesions at the specific
temperature of interest, and R is dose rate. This form of the
equation should be used when low dose rates or prolonged
exposures are used. It tends toward a straight line at a high dose on
a semi-log plot. Our studies yielded a value of k Z 0.03/min at
37 C for Chinese hamster fibroblasts (9); however, human tumor
cell lines might have different repair rates. Nevertheless, this form
of the equation quantitatively expresses the reduced cell killing
due to repair of sublesions during exposure of more than a few
minutes, which should come as no surprise. Additional research
might indicate that an additional and much slower first-order
repair process, such as is observed for single-strand break repair
(6), might result in reduced cell killing at low dose rates.
Those working to introduce radiobiology parameters into
treatment planning protocols need another form of the basic equation. Because most radiation therapy is delivered in daily fractions,
5 d/wk, there is enough time between dose fractions for complete
repair of the sublethal damage associated with the 2-hit mechanism
and for clonogen reassortment. Thus, fractionated doses will
produce cell killing, as described by the following equation:




SZexp  n ad b0 d 2
LQ 4
where n fractions of dose d deliver a total dose D Z nd. On a plot
of log(S) vs D, these survival curves also take on a linear
appearance over several decades of cell kill (10).
If one knows the total number and the intrinsic radiosensitivity
of clonogens in a group of similar tumors, the tumor control
probability models can predict local tumor control as a function
of the total dose. The radiation inactivation parameters for
pathologically similar tumors in different patients or for
clonogens in individual tumors are known to vary (11). Consequently, guesstimates are made for specific tumor-treatment cases
and controversy can ensue. A case in point, the analysis of
prostate cancer brachytherapy yielded excellent results when
a very low value for a (0.036 Gy1) and 15  2 clonogens were
assumed for tumors with Gleason scores of mainly 6 and 7 (12).
Several pathologists and urologists consider this clonogen number
to be low for tumors of this volume. Equally good fits to these
clinical data were obtained using average values of a and b for
prostate cancer cell lines reported in published studies, with
clonogen numbers of 1-10 million and the recently obtained
knowledge of severe hypoxia in some human prostate cancers
(10). That such widely disparate values can produce equally good
fits to clinical data underlines the need for better tumor biology
information. Additional studies could fine tune the more important parameters such as a and b, with some estimate of errors,
quiescent vs. proliferating compartments, oxygenated vs hypoxic
cells, sublethal damage repair rates, and clonogen number.

With realistic parameters, radiation oncology could go beyond

empiricism to tumor control probability modeling to predict
confidently the relative merits of fractionation schemes and to
better understand the 4 or 5 Rs of radiobiology that apply to
tumor radiation therapy (13, 14).
As for normal tissue complication probability modeling, even
less is known of the underlying mechanisms of normal tissue
damage that account for deleterious effects of radiation. If the
damage results from killing stem cells in the tissue, for example,
quantitative LQ radiobiology could come into play. However,
when the late effects of radiation involve interactions between
stem cell, stromal, and vascular elements in the tissues, it is
unclear how the LQ analyses might apply.
Most current radiobiologic research does not attempt to meet
the conditions for LQ analysis and thus does not readily relate to
underlying physical mechanisms. Our research has suggested that
single-hit inactivation by X-rays results from electron track ends
depositing energy in compacted chromatin (with some indication
of a lipid/DNA composition), producing nonreparable lesions
(15). However, 2-hit inactivation results from simple DNA
lesions (mostly reparable strand breaks) produced by simple
ionizations (w60 eV), those that radiation chemists term spurs.
Hydroxyl radicals generated in cellular water mediate a large
proportion of the killing of cells by both mechanisms (9). How
these different lesions can produce the aberrations observed in
the mitotic chromosomes of irradiated cells has been elegantly
described (16).
Using the appropriate LQ equation, this biophysical understanding of tumor cell killing by ionizing radiation can be
extremely useful in the design of improved treatment protocols.
For example, hypofractionated radiation therapy causes less
stress to the patient (shorter overall treatment times) and has
lower cost (fewer patient setups) and thus should be studied
vigorously. Current treatments with 1.8-2.0 Gy/fraction were
optimized empirically using inferior imaging and radiation
delivery technologies. Because the 2 cell-killing mechanisms
exhibit entirely different properties (8), LQ analysis can be used
to usefully guide adjustments to current protocols in light of the
expected biologic consequences. The previous radiation
biophysic studies of tumor cell killing should be embraced and
the appropriate LQ equations exploited for the design of
improved treatment strategies.

References
1. Glatstein E. The omega on alpha and beta. Int J Radiat Oncol Biol Phys
2011;81:319-320.
2. Brenner DJ, Sachs RK, Peters LJ, et al. We forget at our peril the
lessons built into the a/b model. Int J Radiat Oncol Biol Phys 2012;82:
1312-1314.
3. Kellerer AM, Rossi HH. The theory of dual radiation action. Curr
Topics Radiat Res Q 1972;8:85-158.
4. Chadwick KH, Leenhouts HP. A molecular theory of cell survival. Phys
Med Biol 1973;18:78-87.
5. Gillespie CJ, Chapman JD, Reuvers AP, et al. The inactivation of
Chinese hamster cells by X-rays: synchronized and asynchronous cell
populations. Radiat Res 1975;65:353-364.
6. Dugle DL, Gillespie CJ, Chapman JD. DNA strand breaks, repair and
survival in X-irradiated mammalian cells. PNAS 1976;73:809-812.
7. Deacon J, Peckham MJ, Steel GG. The radioresponsiveness of human
tumors and the initial slope of the cell survival curve. Radiother Oncol
1984;2:317-323.

Volume 84  Number 2  2012

8. Chapman JD. Single-hit mechanism of tumor cell killing by radiation.
Int J Radiat Biol 2003;79:71-81.
9. Chapman JD, Gillespie CJ. Radiation induced events and their time
scale in mammalian cells. Adv Radiat Biol 1981;9:143-198.
10. Nahum AE, Movsas B, Horwitz EM, et al. Incorporating clinical
measurements of hypoxia into tumor local control modeling of
prostate cancer: implications for the a/b ratio. Int J Radiat Oncol Biol
Phys 2003;57:391-401.
11. Nahum AE, Sanchez-Nieto B. Tumor control probability modeling:
basic principles and applications in treatment planning. Phys Med
2001;17:369-380.

Beyond empiricism

311

12. Brenner DJ, Hall EJ. Fractionation and protraction for radiotherapy of
prostate carcinoma. Int J Radiat Oncol Biol Phys 1999;43:1095-1101.
13. Withers HR. The four Rs of radiotherapy. Adv Radiat Biol 1975;5:
241-271.
14. Steel GG, Peacock JH. Why are some human tumors more radiosensitive than others? Radiother Oncol 1989;15:63-72.
15. Chapman JD. Biophysical models of mammalian cell inactivation by
radiation. In: Meyn RE, Withers HR, editors. Radiation Biology in
Cancer Research. New York: Raven Press; 1980. p. 21-32.
16. Chadwick KH, Leenhouts HP. The Molecular Theory of Radiation
Biology. New York: Springer-Verlag; 1981.