www.redjournal.

org

Physics Contribution

Deciphering Time-Dependent DNA Damage

Complexity, Repair, and Oxygen Tension: A
Mechanistic Model for FLASH-Dose-Rate
Radiation Therapy
Hans Liew, MSc,*,y,z,x,k Stewart Mein, PhD,*,y,z,x Ivana Dokic, PhD,*,y,z,x
Thomas Haberer, PhD,{ Jürgen Debus, MD, PhD,z,x,k,{,#
Amir Abdollahi, MD, PhD,*,y,z,x and Andrea Mairani, PhD#,**
*Clinical Cooperation Unit Translational Radiation Oncology, National Center for Tumor Diseases (NCT),
Heidelberg University Hospital (UKHD) and German Cancer Research Center (DKFZ), Heidelberg,
Germany; yDivision of Molecular and Translational Radiation Oncology, Department of Radiation
Oncology, Heidelberg Faculty of Medicine (MFHD) and Heidelberg University Hospital (UKHD),
Heidelberg Ion-Beam Therapy Center (HIT), Heidelberg, Germany; zGerman Cancer Consortium (DKTK)
Core-Center Heidelberg, German Cancer Research Center (DKFZ), Heidelberg, Germany; xClinical
Cooperation Unit Radiation Oncology, Heidelberg Institute of Radiation Oncology (HIRO), National
Center for Radiation Oncology (NCRO), Heidelberg University and German Cancer Research Center
(DKFZ), Heidelberg, Germany; kFaculty of Physics and Astronomy, Heidelberg University, Heidelberg,
Germany; {Heidelberg Ion-Beam Therapy Center (HIT), Department of Radiation Oncology, Heidelberg
University Hospital, Heidelberg, Germany;#National Center for Tumor diseases (NCT), Heidelberg,
Germany; and **National Center of Oncological Hadrontherapy (CNAO), Medical Physics, Pavia, Italy

Received Aug 13, 2020, and in revised form Dec 4, 2020. Accepted for publication Dec 28, 2020.

Purpose: Irradiation with ultrahigh dose rates (FLASH) has reemerged as a promising radiation therapy approach to effec-
tively lower potential damage burden on normal tissue without sacrificing tumor control. However, the large number of recent
FLASH studies have been conducted under vastly different experimental conditions and circumstances (ie, investigated

Corresponding author: Andrea Mairani, PhD; E-mail: Andrea. Healthcare GmbH, Merck KGaA, Solution Akademie GmbH, Ergomed
[email protected] PLC Surrey Research Park, Quintiles GmbH, Pharmaceutecal Research
This work was supported by the German Research Council (DFG- Associates GmbH, Boehringer Ingelheim Pharma GmbH Co, PTW-
KFO214), Deutsche Krebshilfe, Germany (Max-Eder 108876), and intra- Freiburg Dr Pychlau GmbH, and Nanobiotix AA, outside the submitted
mural funds from the National Center for Tumor Diseases work. AA reports grants and other support from Merck, EMD, Fibrogen,
(NCT3.0_2015.21/22 NCT-PRO and Biodose programs), as well as a PhD BMS, BioMedX, and Roche, outside the submitted work.
stipend from the Helmholtz International Graduate School for Cancer Data Sharing: Research data are stored in an institutional repository
Research in Heidelberg (to H.L.), The funders had no role in study design, and will be shared upon request to the corresponding author.
data collection and analysis, decision to publish, or preparation of the Supplementary material for this article can be found at https://doi.org/
manuscript. 10.1016/j.ijrobp.2020.12.048.
Disclosures: JD reports grants from CRIeThe Clinical Research AcknowledgmentsdWe thank Dr Kristoffer Petersson and Dr Gabriel
Institute GmbH, View Ray Inc, Accuray International Sarl, Accuray Adrian for sharing the raw cell survival data as a function of oxygen
Incorporated, RaySearch Laboratories AB, Vision RT Limited, Merck concentration.
Serono GmbH, Astellas Pharma GmbH, Astra Zeneca GmbH, Siemens

Int J Radiation Oncol Biol Phys, Vol. 110, No. 2, pp. 574e586, 2021
0360-3016/$ - see front matter Ó 2021 Elsevier Inc. All rights reserved.
https://doi.org/10.1016/j.ijrobp.2020.12.048
Volume 110  Number 2  2021 FLASH-dose-rate radiation therapy in the dynamic UNIVERSE 575

biological endpoint, radiation quality, and environmental oxygen level), with unverified biological mechanisms of action and
unexplored interplay effect of the main dependencies. To facilitate radiobiological investigation of FLASH phenomena and
assessment of clinical applicability, we present an extension of the mechanistic radiobiological model “UNified and VER-
Satile bio response Engine” (UNIVERSE).
Methods and Materials: The dynamic (time-dependent) extension of UNIVERSE was developed incorporating fundamental
temporal mechanisms necessary for dose-rate effect prediction, ie, DNA damage repair kinetics [DDRK], oxygen depletion
and reoxygenation during irradiation. Model performance in various experimental conditions is validated based on a large
panel of in vitro and in vivo data from the literature. The effect of dose, dose rate, oxygen tension, tissue-type, beam quality
and DDRK is analyzed.
Results: UNIVERSE adequately reproduces dose-, dose-ratee and oxygen tensionedependent influence on cell killing. For
the studied systems, results indicate that the extent of cell/tissue sparing effect, if present at all, strongly depends on DDRK
and beam quality used for reference conventional irradiation. A validated mechanistic framework for predicting clinically
relevant endpoints comparing conventional and FLASH high-dose-rate effect has been successfully established, relying on
time-dependent processing of radiation-induced damage classes taking variable oxygen tension into account.
Conclusions: Highlighted by UNIVERSE itself, the multidimensional nature of this relative sparing effect using high-dose-
rate radiation compared with conventional means underlines the importance of robust quantification of biophysical charac-
teristics and consistent, well-documented experimental conditions both in vitro and in vivo before clinical translation. To
further elucidate underlying mechanisms and appraise clinical viability, UNIVERSE can provide reliable prediction for bio-
physical investigations of radiation therapy using ultrahigh dose rate. Ó 2021 Elsevier Inc. All rights reserved.

Introduction radio-resistance (oxygen depletion effect) is commonly

considered to explain sparing in general.1,4 Based on this
hypothesis and/or other dose-rateedependent radiochem-
Irradiation with ultrahigh dose rates (>>10 Gy/s) has
ical mechanisms, such as radical recombination, models
reemerged as a promising therapeutic tool affording clinical
have recently been established to describe bioeffects at
conditions where elevated normal tissue radio-resistances
FLASH dose rate.14-16 Nonetheless, quantitative bench-
are observed. Although initial studies date back to the
1960s,1-3 interest was recently reignited by multiple pub- marks of these models are limited or lacking completely.
In essence, experimental results regarding the FLASH
lications reporting significant sparing of normal tissue
effect describe differing biological endpoints obtained under
while maintaining bioeffect in the tumor with application of
vastly different settings. To date, there is an absence of
high dose levels at high dose rates compared with con-
cohesive and extensively benchmarked approaches to model
ventional radiation therapy using low dose rate, known as
all relevant conditions of biological systems when exposed to
the FLASH effect.4,5
FLASH dose rates, crucial for facilitating assessment of the
Experimental data in the literature, however, entail
underlying mechanisms. To this end, the “UNIfied and
highly diverse biological endpoints ranging from in vitro
cell survival with human cells6 and zebrafish embryos4 to VERSatile bio response Engine” (UNIVERSE) is a mecha-
nistic model for predicting response of biological systems to
neurocognitive functionality of mice after whole-brain
ionizing radiation through consideration of several key
irradiation7-9 and tail necrosis in mice.10 Furthermore, in-
conditions.17-19 In this work, time-dependent processing of
stitutions have applied differing radiation sources such as
radiation-induced DNA damage,20,21 oxygen depletion, and
102 keV x-rays,8 10 MeV electrons,6,10 and 224 MeV
reoxygenation mechanisms14 are implemented toward ac-
protons.11 Sparing effects have been reported at doses be-
curate prediction of biological response over a broad range of
tween 10 and 50 to 75 Gy, and applying dose rates from w
conventional and ultrahigh (FLASH) dose rates. Moreover,
106 Gy/s down to w10 Gy/s.7-10 Moreover, conventional
radiation sources used a reference range from a Cs-137 UNIVERSE is benchmarked against in vitro and in vivo data
from the literature, obtained under various experimental
source with a dose rate of  0.03 Gy/s12 up to a 10
conditions. Physical, biological, and clinical implications of
MeV electron source with a dose rate of 0.23 Gy/s,6 and
the UNIVERSE to support consistent planning and conduc-
oxygenation readings of radiobiological setups are often
tion of much needed foundational experimental studies of
unreliable. In short, experimental conditions of reported
FLASH bioeffect are explored.
data on the FLASH effect are subject to great variability,
and it is therefore challenging to unravel the underlying
dependencies and define explicit clinical parameters. Methods and Materials
The mechanisms that induce differential sparing be-
tween normal tissue and tumors remain highly debated.13 UNIVERSE is a mechanistic model for biological response
However, the radiochemical depletion of oxygen via high to ionizing radiation based on interaction with biological
dose-rate delivery and resulting transient hypoxia causing substructure and cell functionality. The time dependent, or
576 Liew et al. International Journal of Radiation Oncology  Biology  Physics

“dynamic,” version of UNIVERSE is introduced in this In the static UNIVERSE, a change in the oxygen level
work, extended from its time-independent, or “static,” can be accounted for by solely reducing the DSB yield by a
O2
predecessor presented in previous publications,17-19 which hypoxia reduction factor (HRFDSB ), which resembles the
we will briefly recall as follows: classical oxygen enhancement ratio (OER), while the
In the case of sparsely ionizing radiation (eg, x-rays, lethality parameters are assumed to be invariant.17-19 The
gamma rays, electrons), the dose deposition throughout the reduced DSB yield, aO DSB ; can be expressed by:
2

cell nucleus is assumed to be homogeneous. The number of aDSB

aODSB
2
Z O2 ð4Þ
DNA double strand breaks (DSB), considered the most HRFDSB
impactful type of DNA lesion,22 can be calculated using a
cell-line independent DSB yield of aDSB Z 5  If both hypoxic and normoxic data are available,
O2
103 DSB⁄ ððMbp  GyÞ Þ.23,24 Assuming this yield to be HRFDSB can be determined by fitting the model to both
constant over the clinically applied dose range, the ex- datasets, while KiDSB and KcDSB are kept constant. If only
O2
pected total number of DSB (hNtDSB i) can be expressed as: normoxic or hypoxic datasets are available, HRFDSB for a
hNtDSB i Z aDSB  D  DNAc ð1Þ given oxygen concentration ½O2  is estimated using an
empirical parametrization:
m$K þ ½O2 
where DNAc is the DNA content of a cell (w6000 mega O2
HRFDSB Z : ð5Þ
K þ ½O2 
base pairs [Mbp]) and D the applied dose in units of Gy.
Low-energy photon radiation sources have been found to introduced in an earlier publication17 following previous
induce a larger amount of DSBs for the same physical dose works.36,37 Due to the variety of endpoints included in the
applied, compared with photon radiation sources with study and the distinct approaches to their analysis, an
higher energy.25 To account for this change in effectiveness, automized global fitting of the parameters m and K was not
the DSB yield aDSB may be modified by a relative bio- feasible in the scope of this work. Instead, the values
logical effectiveness factor (RBEDSB ). derived in a previous publication17 were adapted and the
In the mechanistic view of UNIVERSE, a certain type of values m Z 3.1 and K Z 0.27 were found to be suitable for
chromatin substructure, so-called giant loops of about 2 the investigated datasets. However, the dataset of Epp
Mbp of DNA,26-28 play a central role in classifying local et al38 was best described with the values m Z 3.4 and
distributions of DSBs. Multiple DSBs inside such giant K Z 0.41. The effect of the chosen values of m and K on
loops are repaired significantly slower29,30 and are associ- the DSB yield as function of ½O2  is shown in Figure E1.
ated with an increased lethality for the cell due to the high The paragraphs above described the time-independent
risk of chromatin loss.31 Classifying giant loops containing (static) version of UNIVERSE. From here forward, the
exactly 1 lesion as isolated DSB (iDSB) and 2 or more temporal extension of UNIVERSE will be described. In the
lesions as complex DSB (cDSB) can accurately predict resulting dynamic version of UNIVERSE, the total irradi-
populations of swiftly and slowly repairing lesions in ation time (Trad ) is divided into several sequential time-
rejoining studies.32,33 UNIVERSE shares this classification steps (Nt Z 100). The relationship between applied dose
of lesions with other models.34,35 The total number of giant _ total applied dose (D) and total irradiation time is
rate (D),
loops (Ngl ) having a DNA content of DNAgl is given by: given by:
DNAc D
Ngl Z ð2Þ Trad Z ð6Þ
DNAgl D_
Following a Monte Carlo approach, the actual total The damage pattern induced by the partial dose applied
number of DSB induced in the nucleus (NtDSB ) is sampled at a given time-step, Dpart Z NDt ; is computed analogously to
for > 104 iterations based on a Poisson distribution with the the static UNIVERSE described earlier. However, to ac-
expectation value from Equation (1). For each iteration, the count for possible oxygen depletion and reoxygenation
sampled amount of DSBs is then distributed randomly over occurring during irradiation, the oxygen level at the current
the giant loops contained in the nucleus. Thereafter, the time-step OðtÞ is determined using14:
   
number of giant loops with 1 DSB (NiDSB ) or 2 and more l l _
OðtÞ Z Oenv þ 1 eðgDþlÞt ð7Þ
DSB (NcDSB ) are counted. The lethality parameters KiDSB gD_ þ l gD_ þ l
and KcDSB describe the probabilities of one isolated DSB or
complex DSB lead to cell inactivation, respectively. The where t is the time passed since the start of the irradiation,
ultimate probability of a cell surviving the irradiation (S) is Oenv is the environmental oxygen level ([O2]) at tZ 0; g is
then given by18,31: the depletion rate constant, and l is the reoxygenation
S Z ð1  KiDSB Þ iDSB ,ð1  KcDSB Þ cDSB :
N N
ð3Þ constant. Values from Petersson et al14 for in vitro analysis,
gZ 0:053 Gy1 and lZ 1 s1 ; were adopted. The time-
The average value of S over all iterations is used to dependent oxygen concentration OðtÞ was used to calcu-
predict the surviving fraction of the cell population. The late the current hypoxia reduction factor HRFDSB O2
ðtÞ
cell-line dependent parameters KiDSB and KcDSB are derived following equation 5, effectively modifying the oxygen
by fitting the model to data of cell survival. dependent radiosensitivity of the cells in real-time. Using
Volume 110  Number 2  2021 FLASH-dose-rate radiation therapy in the dynamic UNIVERSE 577

equations 4 and 1, one can compute the reduced number of probability of each damage class to ultimately trigger
DSBs to be distributed at each time-step. The resulting radio-necrosis or death of the mouse, respectively.
trend of DSB yield as a function of irradiation time is
depicted in Figure E1.
To model the time-dependent repair process of induced Results
damage in the cell nucleus, each iDSB and cDSB is
attributed a random lifetime drawn from an exponential Survival over dose rate
1=2 1=2
distribution based on repair half-life times TiDSB and TcDSB ;
respectively. These half-life times can be obtained either To investigate and survey general effect of different con-
from literature (eg, biexponential fits to DSB repair ki- ditions and parameters considered in the model on pre-
netics32) or fitted to available data. If by the application of a dicted trend of the dose-rate effect, survival fraction was
partial dose at a given time-step, any number of DSB is computed over a range of dose rates (0.01 Gy/s to 104 Gy/s)
added to a giant loop that harbors exactly one DSB (iDSB), with several representative inputs for a given model
it is reclassified as a cDSB and a new lifetime is drawn parameter, while all other parameters are fixed to certain
1=2
based on TcDSB : At every time-step, any damage that has values. These fixed values were chosen such that trends of
exceeded its lifetime is removed from the damage pattern. interest are clearly visible (eg, enough dose to induce suf-
However, when such a repair event takes place, there is a ficient oxygen depletion) while remaining within the range
probability equal to the values of KiDSB for isolated DSB of experimentally relevant values. During analysis, lethality
and KcDSB for cDSB to trigger a “misrepair” event that sets parameters were set for demonstrative purposes to values
the survival probability of this cell (or rather Monte Carlo found for the DU145 cell-line obtained from Adrian et al6
iteration) to zero. This approach ensures the consistency of (Table 1).
the model when repair processes are considered over a
large period of time. Key concepts of the implementation Total dose
described here had been introduced and validated in works Figure 1A displays predictions for various total applied
by Herr et al.20,21 In our implementation, if by the end of 1=2 1=2
dose levels. TiDSB Z 30 minutes, TcDSB Z 5 hours, and [O2]
the irradiation time no lethal event was triggered due to the Z 2.5 % were chosen as fixed values. A general trend of
failure of a repair process, the survival probability of the increased cell killing up to dose rates of about 1 Gy/s was
given Monte Carlo iteration is calculated using Equation 3. observed before survival increases up to roughly hundreds
Again, the survival fraction of the population is determined of Gy/s after which survival plateaus for the given settings.
by the mean value of the survival probabilities determined However, this trend is progressively pronounced with
by each Monte Carlo iteration. increasing total delivered dose and not observable for the 2
A fully mechanistic prediction of in vivo effects post- lowest dose levels (2 and 8 Gy).
irradiation would pose a considerable jump in complexity
from the already challenging description of cell populations Environmental oxygen level
in vitro. However, to explore the principle possibilities of Figure 1B depicts the effect of different oxygen levels
UNIVERSE to provide estimates on higher level systems, ([O2]) on survival prediction with the following fixed set-
1=2 1=2
major simplifications and assumptions were made to pre- tings: total dose of 16 Gy, TiDSB Z 30 minutes and TcDSB Z
dict in vivo endpoints, without explicit mechanistic 5 hours. Results indicate that [O2] substantially influences
description of the transition from in vitro to in vivo. sparing effects at higher dose rates. Given the settings,
Bhouris et al have reported the effect of different dose significant survival increase (beginning at w1 Gy/s) was
rates of radiation on the growth delay of tumors with the observed only for intermediate [O2] at 7.5%, 2.5%, and 1%.
expression 1  VVradctrl
; where Vrad is the volume of the tumor For normoxia (20%) and severe hypoxia (0.1%) and anoxia
measured 15 days postirradiation, while Vctrl is the volume of (0.01%), no significant sparing at higher dose rates was
untreated tumors after the same period.39 In this work, we observed with the selected parameters. Moreover,
describe this value as the relative tumor volume suppression increasing [O2] increased the slope of the initial decrease of
(RTVS), and heuristically approximate the ratio VVrad ctrl
to be survival at lower dose rates.
given by the survival fraction of the cell population within the
irradiated tumor leading to the assumption RTVS Z 1  S: Repair half-life times
For approximation of ND50 and LD50, the dose at which In Figure 1, C and D, predictions are shown for different
50% of mouse tails show radionecrosis as reported by repair half-life times of isolated and complex DSB,
Hendry et al10 and mice that died within 4 days after whole- respectively, with the following fixed settings: total dose of
1=2 1=2
body irradiation as reported by Hornsey et al,40,41 respec- 16 Gy and [O2] of 2.5%. With variable TiDSB ; TcDSB was set
1=2 1=2
tively, we assumed both to be equal to the dose at 50% to 5 hours, whereas for variable TcDSB ; TiDSB was set to 30
1=2
survival fraction. minutes. Figure 1C makes evident that the effect of TiDSB is
1=2
For in vivo data, lethality parameters remain as free highest under w1 Gy/s. In this region, a shorter TiDSB leads
parameters for each endpoint but can be reinterpreted as the to increasing slopes in survival for decreasing dose rates. At
578 Liew et al. International Journal of Radiation Oncology  Biology  Physics

Table 1 Model parameters of UNIVERSE applied in this work

Endpoint K iDSB K cDSB T 1=2
iDSB [min] T 1=2
cDSB [min] m K
42,43
In vitro survival (CHO) 5.9E-3 0.19 80.22 300 3.1 0.27
In vitro survival (HeLa)38 6.7E-3 0.21 14 130 3.4 0.41
In vitro survival (DU145)6 5.9E-3 0.17 4 100 3.1 0.27
RTVS (U87 xenograft)39 0.9E-3 0.095 60 300 3.1 0.27
Mice tail radionecrosis10 0.17E-3 0.006 60 300 3.1 0.27
LD50 whole-body irradiation mice40,41 0.1E-3 0.065 6 300 3.1 0.27

Abbreviations: CHO Z Chinese hamster ovary.

A B

10-1
10-1

10-3
10-3
Survival

Survival

10-5

10-5
10-7
2 Gy 24 Gy pO2 0.01 % pO2 2.5%
8 Gy 32 Gy pO2 01 % pO2 7.5%
16 Gy 10-7 pO2 1 % pO2 20%
10-9
10-2 10-1 100 101 102 103 104 10-2 10-1 100 101 102 103 104
Dose Rate [Gy/s] Dose Rate [Gy/s]
C D

10-2 10-2
Survival

Survival

T1/2 iDSB 5 mins

T1/2 iDSB 15 mins
T1/2 iDSB 30 mins T1/2 cDSB 1 hour
T1/2 iDSB 45 mins T1/2 cDSB 2 hours
T1/2 iDSB 60 mins T1/2 cDSB 5 hours
10-3 10-3
10-2 10-1 100 101 102 103 104 10-2 10-1 100 101 102 103 104
Dose Rate [Gy/s] Dose Rate [Gy/s]

Fig. 1. Effect of different model parameters: dose (A), environmental oxygen level (B), repair halftimes for isolated double
strand breaks (iDSB) (C), and complex double strand breaks (cDSB) (D). (A) Initial decrease of the survival fraction is
followed by onset of a sparing effect. No such effect is observed for lower doses (2 and 8 Gy). (B) Onset of significant sparing
is only visible at intermediate oxygen levels (7.5%, 2.5%, and 1%). (C) Effect is larger with lower dose-rates, with shorter
half-lives increasing the survival. No effect is seen above w1 Gy/s for the applied values. (D) Half-lives of cDSB have no
effect on survival in the given dose-rate range for the chosen values.
Volume 110  Number 2  2021 FLASH-dose-rate radiation therapy in the dynamic UNIVERSE 579
1=2
dose rates higher than w10 Gy/s, the effect of TiDSB complex DSB were simulated for 3ns pulses over the
1=2
diminishes. In contrast, TcDSB does not seem to affect analyzed dose range. Based on these values, the lethality
survival for the dose rates and parameters comparable to parameters were fit to the measured survival data using the
those used in this study. curve_fit routine of the scipy library for Python. This makes
2 fitted free parameters (both lethality parameters), 2
Benchmark against in vitro data globally set parameters (HRFDSB O2
parametrization), and 3
parameters taken from literature (both repair half-life times
Numerical values of applied model parameters for the and the RBEDSB ) for the predictions of this dataset. The R2
following datasets are presented in Table 1. values (coefficient of determination) for the prediction of
the Co-60, electrons and x-ray survival data (Figure 2A)
Survival of Chinese hamster ovary cells under various were determined to be 0.98, 0.98, and 0.99, respectively.
radiation qualities and in split dose experiments While Co-60 and the 3ns electron pulses show near iden-
Figure 2A shows Chinese Hamster Ovary (CHO) cell sur- tical effect, an increased bioeffect for 280 kVp radiation is
vival after irradiation with a Co-60 source (0.01 Gy/s), clearly visible. The mean relative difference between
single 3 ns pulses of w450 keV electrons, and 280 kVp measured and predicted survival (excluding the timepoint at
x-rays (0.033 Gy/s) collected from Michaels et al42 with 0 minutes) for the split doses of electrons (Fig. 2B) and x-
corresponding model predictions. Figure 2, B and C, rays (Fig. 2C) were 3.5% and 10%, respectively.
display CHO cell survival after irradiation with 2 fractions
of the same pulsed electron source and x-ray source at Survival of HeLa cells after irradiation with single 3ns
various doses with different separation times43 against electron pulses under different oxygen levels
model prediction, which are normalized (at t Z 0) to Figure 3 presents survival of HeLa cells postirradiation
account for statistical deviation and highlight time evolu- with single 3ns pulses of w350 keV electrons at various
tion of survival. An RBEDSB of 1.2 was set for the 280 kVp [O2] taken from Epp et al38 and respective model pre-
x-rays.43 Using the repair half-life times of CHO cells dictions. Furthermore, to illustrate the effect of oxygen
found in the literature,20 the number of isolated and depletion (OD) and reoxygenation (RO) kinetics, an

A B Split Dose, 3ns e-Pulses

100 10-1

5.85 Gy + 5.85 Gy
-2
10 6.9 Gy + 6.9 Gy
Survival

10-1 7.2 Gy + 7.2 Gy

10-3

Model
Data
-4
10
0 20 40 60 80 100 120 140
Survival

10-2 Time [min]

C Split Dose, 280 kVp X-Rays
10-1
5.0 Gy + 5.0 Gy

10-2
Survival

10-3
5.5 Gy + 5.5 Gy
Model Co-60
Model 3 ns e-Pulses
10-3
Model 280 kVp
Data Co-60 0.01 Gy/s 6 Gy + 6 Gy
Data 3 ns e Pulses
Data Xrays 280 kVp 0.033 Gy/s
10-4 10-4
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 0 20 40 60 80 100 120 140
Dose [Gy] Time [min]

Fig. 2. Effect of radiation quality ad repair kinetics in split dose experiments. (A) Chinese hamster ovary (CHO) cells
survival after irradiation with a Co-60 source (0.01 Gy/s), single 3 ns pulses of w450 keV electrons42 and 280 kVp x-rays
(0.033 Gy/s),43 with respective simulations by UNIVERSE. RBEDSB of 1.2 was applied to the x-ray simulation. (B and C)
Survival of CHO cells after irradiation with 2 fractions of w450 keV electrons (3 ns pulses) and 280 kVp x-rays (0.033 Gy/s)
with different split doses and various times in between,43 with respective simulations by UNIVERSE. The simulations by
UNIVERSE were normalized to the data point at t Z 0:
580 Liew et al. International Journal of Radiation Oncology  Biology  Physics

additional prediction is shown with both mechanisms overestimated. In contrast, survival of high dose rate ap-
deactivated. The repair half-life times were taken from the pears slightly underestimated by UNIVERSE at interme-
literature44 and used to calculate the number of iDSB and diate [O2] levels. However, predictions lie within the large
cDSB under normoxic conditions within the analyzed dose error margins of the corresponding data.
range. Using these values, both lethality parameters were fit
to the normoxic survival data (Fig. 3A) applying the cur- Benchmark against in vivo data
ve_fit command of the scipy library for Python. This results
O2
in 4 free parameters (if we add the 2 HRFDSB parameters Relative tumor growth supression of U87 xenografts in
that deviate from the global parametrization of the other mice
datasets to the fitted lethality parameters) and 2 parameters Figure 5A shows measured RTVS of tumors based on U87
taken from literature (both repair half-life times) for the human glioblastoma cells engrafted subcutaneously in mice
predictions of this dataset. The resulting mean R2 value for postirradiation with 5 to 6 MeV electrons at conventional
the data in Figure 3 was determined to be 0.73. Although no dose rates (0.1 Gy/s) and at high dose rates (between 125
effect of the OD and RO mechanisms are visible for the Gy/s and single pulse of 1.8 ms) from Bourhis et al,39 with
normoxic and anoxic cases, a significant elevation in sur- respective UNIVERSE simulations. Findings (Fig. 1D)
vival was observed for these mechanisms under hypoxic suggest the negligible effect of the chosen repair half-life
conditions. The sparing effect becomes visible at w10 Gy times of complex damages on the overall effect, thus we
and subsequently increases for higher dose levels. set its value to 5 hours in cases in which no literature values
were available, simplifying the determination of the
Survival of DU145 cells under distinct dose rates and remaining parameters.20 Both lethality parameters and the
oxygen levels repair half-life time of the isolated lesions were determined
Figures 4, A and B, depict DU145 cell survival post- simultaneously based on the analysis of c2 values,
irradiation with low (0.23 Gy/s) and high (600 Gy/s) dose comparing the predictions against the low dose-rate data. A
rates of 10 MeV electron radiation gathered from Adrian [O2] of 1% was assumed based on literature values for
et al6 alongside UNIVERSE predictions under normoxia xenografts.46 Ultimately, we have 3 free parameters (both
(20% oxygen) and hypoxia (1.6% oxygen). The repair half- lethality parameters and the repair half-life time of the
life times for DU145 cells were taken from the literature45 isolated lesions), 2 globally set parameters (HRFDSB O2

and used to predict the number of iDBS and cDSB for the parametrization), and 2 parameters set according to litera-
high dose rate and normoxic situation. These values were ture (oxygen tension and repair half-life time of complex
used to fit both lethality parameters to the measured sur- lesions) for the predictions of this dataset. R2 for the model
vival data with the curve_fit command of the scipy library predictions of the RTVS was determined to be 0.99 and
for Python. Thus, for the predictions for this dataset, we 0.96 under low- and high-dose-rate conditions, respectively.
have 2 fitted free parameters (both lethality parameters), 2 A slight sparing effect for the higher dose rate is visible
O2
globally set parameters (HRFDSB parametrization), and 2 between w10 and w30 Gy.
parameters taken from literature (both repair half-life time).
The R2 value of the model prediction was found to be 0.99 ND50 of mouse tail radionecrosis: dose-rate dependence
for the low and high dose rate under normoxia (Fig. 4A), as ND50 (dose required to produce necrosis in half of the
well as under hypoxia (Fig. 4B). Predicted survival for the 2 cohort) of mouse tail radionecrosis over the dose rate of 10
dose rates are essentially indistinguishable up to w10 Gy. MeV electrons, as measured by Hendry et al10 and
At higher doses, UNIVERSE correctly reproduces higher respective UNIVERSE predictions are shown in Figure 5B.
survival for high dose rate under hypoxia as observed in the In one case, the system is oxygenated, whereas in the other,
data. On the contrary, in normoxia, a slight increase in the mouse tail was under an anoxic atmosphere and clam-
survival for the lower dose rate is predicted at higher ped to minimize blood flow. Following the arguments
applied doses. However, this effect lies within the error described earlier, the repair half-life time for the complex
margins of the data, which largely overlap for both dose DSB was set to 5 hours. Both lethality parameters and the
rates. In Figure 4C, DU145 survival measurement and repair half-life time of the isolated DSB were determined as
prediction are presented under different [O2] levels after described in the supplementary material (Fig. 3E). Exact
irradiation with 18 Gy using the same beams and parame- values of the oxygen status in both cases were unknown,
ters as described earlier. The bounds of the UNIVERSE but oxygen levels of 0.4% and 0.12% for the oxygenated
predictions correspond to the lowest and highest experi- and clamped situation, respectively, were found to appro-
mentally measured doses of each dose rate (low dose rate: priately describe trends within a reasonable range.10,47 For
18.0-18.2 Gy; high dose rate: 17.0-19.0 Gy [personal the predictions of this dataset, we used 4 free parameters
communication, Petersson, April 2020]). A sparing effect is (both lethality parameters, the repair half-life time of the
predicted at intermediate [O2] levels, whereas at upper and isolated lesions and the oxygen tension) and 2 parameters
lower boundaries, both dose rates converge. The data of the set based on the literature (repair half-life time of the
low dose-rate radiation are predicted well for hypoxic [O2] complex lesions and RBEDSB ). The mean relative differ-
levels, whereas the normoxic data point appears to be ences between the measured and predicted ND50 were
Volume 110  Number 2  2021 FLASH-dose-rate radiation therapy in the dynamic UNIVERSE 581

A B
101 101
Data
Model with OD/RO
100 Model no OD/RO 100

10-1 10-1
Survival

Survival
10-2 10-2

10-3 10-3
O2 = 21% O2 = 0.91%
10-4 10-4
0 10 20 30 40 0 10 20 30 40
C Dose [Gy] D 101 Dose [Gy]
101

100 100

10-1 10-1
Survival

Survival
10-2 10-2

10-3 10-3
O2 = 0.77% O2 = 0.59%
10-4 10-4
0 10 20 30 40 0 10 20 30 40
Dose [Gy] Dose [Gy]
E 101 F 101

100 100

10-1 10-1
Survival

Survival

10-2 10-2

10-3 10-3

O2 = 0.26% O2 = 0%
10-4 10-4
0 10 20 30 40 0 10 20 30 40
Dose [Gy] Dose [Gy]

Fig. 3. Effect of environmental oxygen status and the oxygen depletion/reoxygenation mechanism. (A-F) HeLa survival
after irradiation with single 3ns pulses of w350 keV electrons38 at various environmental oxygen levels and respective
simulations by UNIVERSE (solid line). To illustrate the effect of implementing the oxygen depletion/reoxygenation
mechanism, the dotted line shows the simulation results with both mechanisms deactivated. An effect is visible above w10
Gy for hypoxic cases (B-E), whereas normoxic (A) and anoxic (F) scenarios are evidently not affected by the oxygen
depletion mechanism.

determined to be 6.2% and 0.3% for the oxygenated and 1.2) measured by Hornsey et al40,41 over a range of dose
clamped situation, respectively. rates with respective UNIVERSE prediction. Following the
same reasoning as described above, the repair half-life time
LD50 (4 days) of whole-body irradiation of mice: dose- for the complex lesions was set to 5 hours. Again, both
rate dependence lethality parameters and the repair half-life time of the
LD50 (dose at which 50% of the subjects have died after a isolated lesions were determined based on the c2 value of
given period of time; here t Z 4 days) is shown in the predictions as KiDSB Z 10 4, KcDSB Z 0.065 and
1=2
Figure 5C for whole body irradiation of mice with w8MeV TiDSB Z 6 minutes, respectively. An oxygen tension of 3%
electron and 250kVp x-ray sources (as above: RBEDSB Z was found to describe data best, fitting the expected O2
582 Liew et al. International Journal of Radiation Oncology  Biology  Physics

A DU 145, Normoxia
B DU 145, Hypoxia (1.6 %)
C DU 145, variable O2
101 101 100
Model 0.23 Gy/s Model 0.23 Gy/s
Model 600 Gy/s Model 600 Gy/s
100 Data 0.23 Gy/s 100 Data 0.23 Gy/s
Data 600 Gy/s Data 600 Gy/s 10-1

10-1 10-1
10-2

Survival
Survival
Survival

-2 -2
10 10

10-3
10-3 10-3

Model 0.23 Gy/s

10-4 10-4 10-4
Model 600 Gy/s
Data 0.23 Gy/s
Data 600 Gy/s
10-5 10-5 10-5
0 0 10 15 20 25 30 0 0 10 15 20 25 30 10-3 10-2 10-1 100 101
Dose [Gy] Dose [Gy] O2 concentration [%]

Fig. 4. Sparing effects in hypoxic conditions. DU145 survival after irradiation with conventional (0.23 Gy/s) and high dose
rate (600 Gy/s) of 10 MeV electron radiation under normoxia (20% [O2]) (A), hypoxia (1.6% [O2]) (B) and different oxygen
levels after irradiation with 18 Gy (C),6 with corresponding predictions by UNIVERSE. Upper and lower bounds in (C)
represent range of measured doses.

A RTVS of U87 Xenografts in Mice B ND50 Mice Tail Necrosis C LD50 of Mice Whole-Body-Irradiation
90 20

100
18
80
16
80
70
14
ND50 [Gy]

LD50 [Gy]

60
RTVS [%]

60 12

40 10
50
8 Model electrons
20 Model 0.1 Gy/s Model Air Model x-rays
Model 125 Gy/s 40 Model Clamping + N2 Data electrons Hornsey & Alper 1966
6
Data 0.1 Gy/s Data Air Data electrons Hornsey & Bewley 1971
Data 125 Gy/s Data Clamping + N2 Data x-rays Hornsey & Alper 1966
0 30
0 5 10 15 20 25 30 35 10-1 100 101 102 10-3 10-2 10-1 100 101 102
Dose [Gy] Dose-rate[Gy/s] Dose-rate[Gy/s]

Fig. 5. Effect of dose rate on in vivo endpoints. (A) relative tumor volume suppression (RTVS) measured in U87 human
glioblastoma engrafted subcutaneously in mice, after irradiation with 5 to 6 MeV electrons at conventional dose-rates (red:
0.1 Gy/s) and ultrahigh dose-rates (black: between 125 Gy/s and 1 pulse of 1.8 ms),39 with respective simulations by
UNIVERSE. An oxygen level of 1% was assumed in the xenograft. (B) ND50 of mice tail necrosis with different dose rates
of 10 MeV electrons10 in an oxygenated (blue) and anoxic environment (green: N2 þ clamping of tail) with corresponding
descriptions by UNIVERSE. (C) LD50 after whole-body-irradiation of mice with either 250kV x-rays (purple) or w8 MeV
electrons (orange) over a range of dose-rates40,41 and respective predictions by UNIVERSE. For consistency in defining and
interpreting the investigated endpoint (with tLD50 Z 4 days), the data for LD50 (tLD50 Z 5 days) reported by Hornsey and
Bewley41 (square) has been normalized to the corresponding LD50 at the same dose-rate. (A color version of this figure is
available at https://doi.org/10.1016/j.ijrobp.2020.12.048.)

level range.46,47 Thus, we used 4 free parameters (both Discussion

lethality parameters, the repair half-life time of the isolated
lesions and the oxygen tension) and 2 parameters set based An extension of the mechanistic biomodeling framework
on literature (repair half-life time of the complex lesions UNIVERSE is introduced, implementing time-dependent
and RBEDSB ). The mean relative differences between the repair of DSB lesions and oxygen depletion as well as
measured and predicted LD50 for the electron source were reoxygenation mechanisms, expanding the capabilities of
determined to be 4.3%.
Volume 110  Number 2  2021 FLASH-dose-rate radiation therapy in the dynamic UNIVERSE 583

UNIVERSE to predict the dose-rate-dependent bioeffect potentially even the inverse effect may appear due to the
from lower conventional (clinical) settings up to ultrahigh- “classical” dose-rate effect). This effect is modified by the
rate delivery. The effect of different parameters on general assumed repair half-life of the 2 DSB classes, in which a
survival trends as a function of dose rate are presented and shorter half-life leads to increased survival at lower dose
discussed first, providing an overview of FLASH de- rates. For the dose-rate range, and thus irradiation times,
pendencies and context for discussion of UNIVERSE considered in this work, this is especially relevant for the
development and validation. shorter repair half-life times of the isolated DSB (w5-60
Evolution of survival with dose rate exhibited by most minutes), whereas the significantly longer repair half-life
curves in Figure 1 can be explained by 2 mechanisms: (1) times of the complex DSB (several hours) have little effect
an initial survival decrease at lower dose rates is driven by (Fig. 1, C and D). DSB repair half-life times could also
the reduced repair of DSB taking place during the short- explain in part the potential tissue-specificity of a sparing
ening irradiation times (“classical” dose-rate effect48) and effect discussed in the literature.51
(2) a subsequent survival increase can be attributed to the Taken together, generalized UNIVERSE predictions
onset of the oxygen depletion effect. As for the latter, to over the range of dose rates and dependencies with relevant
trigger this effect a sufficiently high dose rate is needed to parameters are consistent with existing experimental evi-
deplete oxygen quicker than it is replenished in the system. dence. However, to further demonstrate validity of UNI-
Furthermore, even without reoxygenation mechanisms, a VERSE, quantitative benchmarking was performed based
specific dose level (ie, dose-threshold) is necessary to on datasets from the literature, which were acquired under
deplete enough oxygen to induce a sparing effect.14 Sur- diverse combinations of radiation sources, cell-lines, and
vival plateaus because irradiation times become virtually [O2] levels. Furthermore, many of the discussed features
instantaneous, allowing no intraradiation repair and are visibly reflected in these datasets.
limiting the ability of elevated dose rates to deplete more In line with prior discussion, under normoxic conditions,
oxygen. The dose-rate range in which we find the survival CHO cell survival in Figure 2A does not exhibit a sparing
increasing (w5-100 Gy/s) and the doses needed to observe effect under HDR radiation (3ns electron pulses). Further-
substantial sparing (8-16 Gy) in Figure 1A fit well with more, this dataset highlights the importance of considering
observations made in in vivo experiments: significant potential differences in biological effectiveness between
sparing effects have been mostly reported at doses 10 Gy radiation sources (eg, increased effectivity of 280-kVp x-
and dose rates greater than w40 Gy/s.4,5 The observed rays). The predicted survival curves of the Co-60 and x-rays
dose-rate region of the survival increase is also in agree- for their respective dose rates would be virtually the same
ment with a dimensional analysis by Zhou et al,49 which in the given dose range, as illustrated in Figure 1A, if the
predicts the order of magnitude of the minimum dose rate RBEDSB of the x-ray source is not taken into account.
to observe a sparing effect in the range of 10 to 100 Gy/s. However, including the RBEDSB reported by the literature
However, even for less complex cases of in vitro effects, for the given x-ray energy leads to a convincing match
one cannot give a precise and simultaneously general pre- between data and prediction. If such information is
diction concerning dose and dose-rate thresholds and neglected, severe misinterpretations of dose-rate effects
whether a sparing effect in comparison to conventional may arise. Although repair half-life times taken from the
dose rates can be observed. Not only are thresholds highly literature were applied to both the survival curves (Fig. 2A)
dependent on assumed oxygen kinetics,14 but a sparing and the split dose experiments (Fig. 2, B and C), UNI-
effect relative to conventional dose rates is dependent on VERSE provides satisfactory description in both cases,
several parameters. adding validity to the implementation of the time-
One of the major parameters is the [O2] level (Fig. 1B). dependent repair processes. However, survival is slightly
The observation of no significant oxygen depletion effect at overestimated at higher split times for the 2 higher doses of
normoxic and anoxic environments in this work is further electron pulses and lowest dose of x-rays. For the other 3
supported by experimental data in the literature, reporting datasets, survival is somewhat underestimated for lower
no sparing effects in systems with known normoxic4,6,41,50 split times. One may achieve improved descriptions of the
and anoxic50 conditions. Furthermore, this effect is pre- split dose experiments by using a separate set of parame-
dicted by existing quantitative oxygen depletion effect ters,20 which is supported by known discrepancies observed
models.14,15 Absence of a sparing effect at normoxia can be between values obtained from dose-rate and split-dose ex-
explained by the large doses required to deplete enough periments, potentially caused by temperature fluctuations
oxygen to observe significant radioprotection, whereas at during split dose experiments.20,52
the lowest oxygen concentrations, the absolute amount of Modeling survival of HeLa cells (Fig. 3) primarily ex-
possible depletion is insufficient to observe any change in emplifies the capabilities of UNIVERSE in the lower
radiosensitivity.4,5,14 hypoxic range (<1% oxygen) (Fig, 2, B-E) and visualizes
Another possible influence is the selected dose-rate level the effect of the oxygen depletion and reoxygenation
O2
of the reference radiation. More specifically, one could mechanism on survival. The chosen HRFDSB implementa-
argue that the lower the reference dose rate, the less likely a tion for this dataset itself leads to an increased oxygen ef-
relative sparing at higher dose rates would be observed (and fect in comparison to the other datasets. However, without
584 Liew et al. International Journal of Radiation Oncology  Biology  Physics

the oxygen depletion implemented, even more extreme explanation for the differential effect reported in the liter-
parameters must be chosen, underlining the necessity for ature (FLASH effect), which has been the primary reason
such a sparing mechanism. As expected, our predictions for the renewed research interest. As one possibility, the
show the sparing effect to emerge at w10 Gy, and no lower oxygen levels in tumors compared with normal tissue
sparing effect is visible at normoxia (Fig. 3A) and anoxia has been widely discussed.5,14 It has been pointed out,
(Fig. 3F). The predicted (and expected) lack of sparing however, that tumors are indeed regularly more hypoxic but
effect for normoxia and anoxia can be seen also for not anoxic and are expected to show sparing within the
U87MG and HT144 cell data taken from Cygler et al50 oxygen depletion models as well.5,47 Furthermore, Pratx
(Fig. 2E). et al15 have argued that no sparing effect should be visible
Not only does UNIVERSE describe the survival data of in the well perfused normal tissue and hypothesize sparing
the DU145 cells accurately for both dose rates under to be “driven by normal tissue stem cells, which reside in
distinct oxygenation conditions (Fig. 4, A and B), it also hypoxic niches away from the vasculature.” As discussed
shows appropriate predictions over the range of oxygen earlier, our predictions of the extent of sparing do not only
levels greater than 1% using the same parameter set depend on the exact levels of oxygen present in both sys-
(Fig. 4C). The sole outlier of the normoxic LDR lower dose tems, but also on a variety of other parameters, preventing
rate might be caused by issues with detecting very low any generalized statements regarding differential sparing
survival levels as a large fraction of the raw data points based only on oxygenation. Other potential explanatory
included in the depicted mean values were zero (personal approaches in the literature include different processing of
communication, Petersson et al, April 2020). The large radical oxygen species, repair capabilities of DNA damage,
error associated with the data of the high dose rate might and immunologic response between tumors and normal
reflect the dosimetric challenges at such high dose rates. tissue.13 One must note that implementation of the dynamic
Again, a sparing effect appears at hypoxia (Fig. 4B) but not UNIVERSE is not strictly bound to the oxygen depletion
normoxia (Fig. 4A), and a differentiation occurrs at doses hypothesis. In fact, due to its mechanistic architecture, any
greater than w10 Gy. Figure 4C markedly illustrates the mechanism that in principle can be reduced to a functional
appearance of a sparing effect at intermediate oxygen dependency of DSB yield on the dose rate can be imple-
levels. mented to reproduce the same general sparing effect and
At the current stage of development, the in vivo pre- potential mechanisms for differential sparing could princi-
dictions of UNIVERSE are not based on an explicit pally be included.
mechanistic description of the transition between in vitro
and in vivo, but rather on phenomenological assumptions
and approximations. However, they provide a first projec- Conclusion
tion on the consequences of the in vitro model in higher-
level systems. For in vivo tumors, although our descrip- The time-dependent UNIVERSE presented in this study
tion of the RTVS must be considered an approximation, quantitatively predicts bioeffect of dose rates across a
with the oxygen level in a xenograft derived from literature diverse panel of biological conditions endpoints from the
(1% O2), UNIVERSE manages to ably describe RTVS lowest clinically relevant values up to those applied in
measured in U87 human glioblastoma xenografts in mice radiobiological studies using nanosecond pulses of radia-
(Fig. 5A). For in vivo normal tissue, description of ND50 tion. Benchmarks on in vitro and in vivo data were per-
for mouse tail radio-necrosis (Fig. 5B) reproduces principle formed, making UNIVERSE a prime candidate to aid
trends of the data considering the same oxygen depletion development, experimentation, and assessment of
and reoxygenation parameters as for in vitro scenarios. ultrahigh-dose-rate (FLASH) radiation as a whole, despite
Furthermore, UNIVERSE accurately predicts LD50 of mice the highly distinct design of existing studies. In the future,
after whole body irradiation over a large range of dose rates the model could serve as a platform to test and implement a
(Fig. 5C). more advanced mechanistic hypothesis in an iterative
The successful description of in vivo data of both tumors approach, whereby an initially crude hypothesis is pro-
and normal tissue for both low and high dose rate suggests gressively refined as more experimental data becomes
that the UNIVERSE framework can, in principle, be available. The multidimensional dependency highlighted
extended to predicting in vivo biological effects at FLASH here illustrates that any general formulation of a threshold
dose rates. But, as no mechanism on in vitro level was dose/dose rate to observe a relative sparing effect would be
implemented (or known), which leads to a general differ- unreasonable without explicitly accounting for the specific
entiation between normal tissue cell and tumor cell setup. Thus, to further elucidate underlying mechanisms
response, our model would not predict their distinct and appraise clinical viability, experiments must be con-
response to high- and low-dose-rate radiation in vivo under ducted under actively controlled settings and factors. Here,
identical conditions. Moreover, tumor tissue is not explic- one should focus on reducing the degrees of freedom in the
itly excluded from a sparing effect in UNIVERSE, as experimental designs (eg, by using the same radiation
illustrated in the case of the RTVS of U87 xenografts source, while only varying the applied dose rate). If
(Fig. 5A). This might hint toward a more physiological possible, the oxygenation status of the irradiated system
Volume 110  Number 2  2021 FLASH-dose-rate radiation therapy in the dynamic UNIVERSE 585

should be monitored for in vitro systems and at least esti- 19. Liew H, Mein S, Debus J, Dokic I, Mairani A. Modeling direct and
mated in in vivo endpoints. Most important, however, is the indirect Action on cell survival after photon irradiation under nor-
moxia and hypoxia. Int J Mol Sci 2020;21:3471.
comprehensive documentation of each of the discussed 20. Herr L, Friedrich T, Durante M, Scholz M. A model of photon cell
parameters and their careful consideration during the killing based on the spatio-temporal clustering of DNA damage in
interpretation of the results. Ultimately, UNIVERSE could higher order chromatin structures. PLOS One 2014;9:e83923.
serve to unify biophysical investigations of radiation ther- 21. Herr L. Modeling of time-dose-LET effects in the cellular response to
apy using ultrahigh dose rate. radiation [Dissertation]. Darmstadt, Germany: Technische Universität;
2015.
22. Jeggo PA, Löbrich M. DNA double-strand breaks: their cellular and
clinical impact? Oncogene 2007;26:7717-7719.
References 23. Frankenberg-Schwager M. Review of repair kinetics for DNA damage
induced in eukaryotic cells in vitro by ionizing radiation. Radiother
1. Town CD. Effect of high dose rates on survival of mammalian cells. Oncol 1989;14:307-320.
Nature 1967;215:847-848. 24. Prise KM, Ahnström G, Belli M, et al. A review of dsb induction data
2. Berry RJ, Hall EJ, Forster DW, Storr TH, Goodman MJ. Survival of for varying quality radiations. Int J Radiat Biol 1998;74:173-184.
mammalian cells exposed to x rays at ultra-high dose-rates. Br J 25. Liang Y, Fu Q, Wang X, et al. Relative biological effectiveness for
Radiol 1969;42:102-107. photons: implication of complex DNA double-strand breaks as critical
3. Nias AH, Swallow AJ, Keene JP, Hodgson BW. Effects of pulses of lesions. Phys Med Biol 2017;62:2153-2175.
radiation on the survival of mammalian cells. Br J Radiol 1969;42:553. 26. Yokota H, van den Engh G, Hearst JE, Sachs RK, Trask BJ. Evidence
4. Vozenin M-C, Hendry JH, Limoli CL. Biological benefits of ultra-high for the organization of chromatin in megabase pair-sized loops ar-
dose rate FLASH radiotherapy: sleeping beauty awoken. Clin Oncol ranged along a random walk path in the human G0/G1 interphase
(R Coll Radiol) 2019;31:407-415. nucleus. J Cell Biol 1995;130:1239-1249.
5. Wilson JD, Hammond EM, Higgins GS, Petersson K. Ultra-high dose 27. Sachs RK, van den Engh G, Trask B, Yokota H, Hearst JE. A random-
rate (FLASH) radiotherapy: silver bullet or fool’s gold? Front Oncol walk/giant-loop model for interphase chromosomes. Proc Natl Acad
2020;9:1563. Sci U S A 1995;92:2710-2714.
6. Adrian G, Konradsson E, Lempart M, Bäck S, Ceberg C, Petersson K. 28. Solovjeva L, Svetlova M, Stein G, et al. Conformation of replicated
The FLASH effect depends on oxygen concentration. Br J Radiol segments of chromosome fibres in human S-phase nucleus. Chromo-
2019;93:20190702. some Res 1998;6:595-602.
7. Montay-Gruel P, Petersson K, Jaccard M, et al. Irradiation in a flash: 29. Johnston PJ, Bryant PE. A component of DNA double-strand break
unique sparing of memory in mice after whole brain irradiation with repair is dependent on the spatial orientation of the lesions within the
dose rates above 100Gy/s. Radiother Oncol 2017;124:365-369. higher-order structures of chromatin. Int J Radiat Biol 1994;66:531-
8. Montay-Gruel P, Bouchet A, Jaccard M, et al. X-rays can trigger the 536.
FLASH effect: ultra-high dose-rate synchrotron light source prevents 30. Gauter B, Zlobinskaya O, Weber K-J. Rejoining of radiation-induced
normal brain injury after whole brain irradiation in mice. Radiother DNA double-strand breaks: pulsed-field electrophoresis analysis of
Oncol 2018;129:582-588. fragment size distributions after incubation for repair. Radiat Res
9. Montay-Gruel P, Acharya MM, Petersson K, et al. Long-term neuro- 2002;157:721-733.
cognitive benefits of FLASH radiotherapy driven by reduced reactive 31. Friedrich T, Scholz U, Elsässer T, Durante M, Scholz M. Calculation
oxygen species. Proc Natl Acad Sci U S A 2019;116:10943-10951. of the biological effects of ion beams based on the microscopic spatial
10. Hendry JH, Moore JV, Hodgson BW, Keene JP. The constant low damage distribution pattern. Int J Radiat Biol 2012;88:103-107.
oxygen concentration in all the target cells for mouse tail radio- 32. Tommasino F, Friedrich T, Scholz U, Taucher-Scholz G, Durante M,
necrosis. Radiat Res 1982;92:172-181. Scholz M. A DNA double-strand break kinetic rejoining model based
11. Beyreuther E, Brand M, Hans S, et al. Feasibility of proton FLASH on the local effect model. Radiat Res 2013;180:524-538.
effect tested by zebrafish embryo irradiation. Radiother Oncol 2019; 33. Tommasino F, Friedrich T, Jakob B, Meyer B, Durante M, Scholz M.
139:46-50. Induction and processing of the radiation-induced gamma-H2AX
12. Favaudon V, Caplier L, Monceau V, et al. Ultrahigh dose-rate FLASH signal and its link to the underlying pattern of DSB: a combined
irradiation increases the differential response between normal and experimental and modelling study. PLOS One 2015;10:e0129416.
tumor tissue in mice. Sci Transl Med 2014;6:245ra93. 34. Elsässer T, Weyrather WK, Friedrich T, et al. Quantification of the
13. Zhou G. Mechanisms underlying FLASH radiotherapy, a novel way to relative biological effectiveness for ion beam radiotherapy: direct
enlarge the differential responses to ionizing radiation between normal experimental comparison of proton and carbon ion beams and a novel
and tumor tissues. Radiat Med Protect 2020;1:35-40. approach for treatment planning. Int J Radiat Oncol Biol Phys 2010;
14. Petersson K, Adrian G, Butterworth K, McMahon SJ. A quantitative 78:1177-1183.
analysis of the role of oxygen tension in FLASH radiation therapy. Int 35. Friedrich T, Durante M, Scholz M. Modeling cell survival after photon
J Radiat Oncol Biol Phys 2020;107:539-547. irradiation based on double-strand break clustering in megabase pair
15. Pratx G, Kapp DS. A computational model of radiolytic oxygen chromatin loops. Radiat Res 2012;178:385-394.
depletion during FLASH irradiation and its effect on the oxygen 36. Carlson DJ, Stewart RD, Semenenko VA. Effects of oxygen on
enhancement ratio. Phys Med Biol 2019;64:185005. intrinsic radiation sensitivity: a test of the relationship between aerobic
16. Labarbe R, Hotoiu L, Barbier J, Favaudon V. A physicochemical and hypoxic linear-quadratic (LQ) model parameters. Med Phys 2006;
model of reaction kinetics supports peroxyl radical recombination as 33:3105-3115.
the main determinant of the FLASH effect. Radiother Oncol 2020; 37. Alper T, Howard-Flanders P. Role of oxygen in modifying the
153:303-310. radiosensitivity of E. coli B. Nature 1956;178:978-979.
17. Mairani A, Böhlen TT, Dokic I, Cabal G, Brons S, Haberer T. 38. Epp ER, Weiss H, Djordjevic B, Santomasso A. The radiosensitivity of
Modelling of cell killing due to sparsely ionizing radiation in nor- cultured mammalian cells exposed to single high intensity pulses of
moxic and hypoxic conditions and an extension to high LET radiation. electrons in various concentrations of oxygen. Radiat Res 1972;52:
Int. J. Radiat. Biol 2013;89:782-793. 324-332.
18. Liew H, Klein C, Zenke FT, Abdollahi A, Debus J, Dokic I, 39. Bourhis J, Montay-Gruel P, Gonçalves Jorge P, et al. Clinical trans-
Mairani A. Modeling the effect of hypoxia and DNA repair inhibition lation of FLASH radiotherapy: why and how? Radiother Oncol 2019;
on cell survival after photon irradiation. Int J Mol Sci 2019;20:6054. 139:11-17.
586 Liew et al. International Journal of Radiation Oncology  Biology  Physics

40. Hornsey S, Alper T. Unexpected dose-rate effect in the killing of mice phosphorescence during fractionated radiation therapy. Nat Comm
by radiation. Nature 1966;210:212-213. 2020;11:573.
41. Hornsey S, Bewley DK. Hypoxia in mouse intestine induced by 47. McKeown SR. Defining normoxia, physoxia and hypoxia in
electron irradiation at high dose-rates. Int J Radiat Biol Relat Stud tumoursdimplications for treatment response. Br J Radiol 2014;87:
Phys Chem Med 1971;19:479-483. 20130676.
42. Michaels HB, Epp ER, Ling CC, Peterson EC. Oxygen sensitization of 48. Steel GG, Down JD, Peacock JH, Stephens TC. Dose-rate effects and the
CHO cells at ultrahigh dose rates: prelude to oxygen diffusion studies. repair of radiation damage. Radiother Oncol 1986;5:321-331.
Radiat Res 1978;76:510-521. 49. Zhou S, Zheng D, Fan Q, et al. Minimum dose rate estimation for
43. Gerweck LE, Epp ER, Michaels HB, Clifton Ling C, Peterson EC. pulsed FLASH radiotherapy: a dimensional analysis. Med Phys 2020;
Repair of sublethal damage in mammalian cells irradiated at ultrahigh 47:3243-3249.
dose rates. Radiat Res 1979;77:156-169. 50. Cygler J, Klassen NV, Ross CK, Bichay TJ, Raaphorst GP. The sur-
44. Kampinga HH, Hiemstra YS, Konings AW, Dikomey E. Correlation vival of aerobic and anoxic human glioma and melanoma cells after
between slowly repairable double-strand breaks and thermal radio- irradiation at ultrahigh and clinical dose rates. Radiat Res 1994;140:
sensitization in the human HeLa S3 cell line. Int J Radiat Biol 1997; 79-84.
72:293-301. 51. Venkatesulu BP, Sharma A, Pollard-Larkin JM, et al. Ultra high dose
45. El-Awady RA, Dikomey E, Dahm-Daphi J. Radiosensitivity of rate (35 Gy/sec) radiation does not spare the normal tissue in cardiac
human tumour cells is correlated with the induction but not with and splenic models of lymphopenia and gastrointestinal syndrome. Sci
the repair of DNA double-strand breaks. Br J Cancer 2003;89:593- Rep 2019;9:17180.
601. 52. Stephens TC, Eady JJ, Peacock JH, Steel GG. Split-dose and low
46. Cao X, Rao Allu S, Jiang S, et al. Tissue pO2 distributions in dose-rate recovery in four experimental tumour systems. Int J Radiat
xenograft tumors dynamically imaged by Cherenkov-excited Biol Relat Stud Phys Chem Med 1987;52:157-170.