1. Bio-2 The Geant4-DNA project
Takashi Sasaki – KEK, Japan
Sébastien Incerti – IN2P3/CENBG, France
on behalf of the Bio-2 team
TYL-FKPPL Joint Workshop, Seoul, June 4-6, 2013

2. Outline Context Proposed workplan for 2013 Collaboration matters
Development of a simulation platform dedicated to the modelling biological effects of ionising radiation down to the cellular & sub-cellular scales
the Geant4-DNA project
Proposed workplan for 2013
Validation of theoretical cross section models in biomolecules through dedicated measurements
Development of realistic cellular geometries
Efforts toward Geant4-DNA computation speedup
Application in targeted radiotherapy
Collaboration matters

3. Proton & hadrontherapy
Modelling biological effects of ionising radiation remains a major scientific challenge
Chronic exposure
Fukushima
Diagnosis
Space missions
ISS
Space exploration
Proton & hadrontherapy
« A major challenge lies in providing a sound mechanistic understanding of low-dose radiation carcinogenesis » L. Mullenders et al. Assessing cancer risks of low-dose radiation Nature Reviews Cancer (2009)
Mars
NCC

4. The Monte Carlo approach
Can « reproduce » with accuracy the stochastic nature of particle-matter interactions
Many Monte Carlo codes are already available today in radiobiology for the simulation of track structures at the molecular scale in biological medium
E.g. PARTRAC, TRIOL, PHITS, KURBUC, NOREC…
Include physics & physico-chemistry processes, detailed geometrical descriptions of biological targets down to the DNA size, DNA and chromosome damage simulation and even repair mechanisms (PARTRAC)…
Usually designed for very specific applications
Not always easily accessible
Is it possible to access the source code ?
Are they adapted to recent OSs ?
Are they extendable by the user ?
« To expand accessibility and avoid ‘reinventing the wheel’, track structure codes should be made available to all users via the internet from a central data bank» H. Nikjoo, IJRB 73, 355 (1998)

5. ATLAS, CMS, LHCb, ALICE @ CERN
BaBar, ILC…
Brachytherapy
PET Scan (GATE)
Hadrontherapy
DICOM dosimetry
Medical linac
Earth magnetosphere
ISS
GAIA
GLAST/FERMI (NASA)
Physics-Biology

6. Geant4 ?
Can we try to extend Geant4 to model biological effects of radiation ?
Strong limitations prevent its usage for the modelling of biological effects of ionising radiation at the sub-cellular & DNA scale
Condensed-history approach
No step-by-step transport on small distances, a key requirement for micro/nano-dosimetry
Low-energy limit applicability of EM physics models is limited
~250 eV for Livermore Low Energy EM models
~100 eV for Penelope Low Energy EM models
No description of target molecular properties
Liquid water, DNA nucleotides
Only physical particle-matter interactions
Physical interactions are NOT the dominant processes for DNA damage at low LET...

7. See Int. J. Model. Simul. Sci. Comput. 1 (2010) 157–178
The Geant4-DNA project
Initiated in 2001 by Dr Petteri Nieminen at the European Space Agency/ESTEC
Main objective: extension of the general purpose Geant4 Monte Carlo toolkit for the simulation of interactions of radiation with biological systems at the cellular and DNA level in order to predict early DNA damages in the context of manned space exploration missions (« bottom-up » approach)
Providing an open source access to the scientific community that can be easily upgraded & improved
First prototypes of physics models were added to Geant4 in 2007
It is now a full independent sub-category of the electromagnetic category of Geant4
$G4INSTALL/source/processes/electromagnetic/dna
Usable from external codes such as GATE since 2012
Currently an on-going interdisciplinary activity of the Geant4 collaboration « low energy electromagnetic physics » working group
Coordinated by CNRS/IN2P3 since 2008
Supported by several funding agencies : ESA (AO6041, AO 7146), ANR, INSERM...

8. How can Geant4-DNA model radiation biology ?
FJPPL
Physics stage
step-by-step modelling of physical interactions of incoming & secondary ionising radiation with biological medium (liquid water)
Excited water molecules
Ionised water molecules
Solvated electrons
Physico-chemistry/chemistry stage
Radical species production
Diffusion
Mutual interactions
FJPPL
Geometry stage
DNA strands, chromatin fibres, chromosomes, whole cell nucleus, cells… for the prediction of damages resulting from direct and indirect hits
Biology stage DIRECT DNA damages
Biology stage INDIRECT DNA damages low LET)
t=0
t=10-15s
t=10-6s

9. a) Physics stage: improving Physics

10. Physics models available in Geant4 9.6+P01 (Spring 2013)
Geant4-DNA physics models are applicable to liquid water, the main component of biological matter
They can reach the very low energy domain down to electron thermalization
Compatible with molecular description of interactions (5 excitation & ionisation levels of the water molecule)
Sub-excitation electrons (below ~9 eV ) can undergo vibrational excitation, attachment and elastic scattering
Purely discrete
Simulate all elementary interactions on an event-by-event basis
No condensed history approximation
Models can be purely analytical and/or use interpolated data tables
For eg. computation of integral cross sections
They use the same software design as all electromagnetic models available in Geant4 (« standard » and « low energy » EM models and processes)
Allows the combination of models and processes

11. Overview of physics models for liquid water
Electrons
Elastic scattering
Screened Rutherford and Brenner-Zaider below 200 eV
Partial wave framework model, 3 contributions to the interaction potential
Ionisation
5 levels for H2O
Dielectric formalism & FBA using Heller optical data up to 1 MeV, and low energy corrections
Excitation
Dielectric formalism & FBA using Heller optical data and semi-empirical low energy corrections
Vibrational excitation
Michaud et al. xs measurements in amorphous ice
Factor 2 to account for phase effect
Dissociative attachment
Melton et al. xs measurements
Protons & H
Excitation
Miller & Green speed scaling of e- excitation at low energies and Born and Bethe theories above 500 keV
Ionisation
Rudd semi-empirical approach by Dingfelder et al. and Born and Bethe theories & dielectric formalism above 500 keV (relativistic + Fermi density)
Charge change
Analytical parametrizations by Dingfelder et al.
He0, He+, He2+
Excitation and ionisation
Speed and effective charge scaling from protons by Dingfelder et al.,
Semi-empirical models from Dingfelder et al.
C, N, O, Fe
Speed scaling and global effective charge by Booth and Grant
Photons
from EM « standard » and « low energy »
-Electron ionisation: low energy phase-specific (liquid) exchange term proposed by Emfietzoglou et al.
-Electron excitation: semi empirical low energy corrections by Emfietzoglou et al.
-Electron vib. excitation: Measurements detailed by Michaud et al. (2003) were considered taking into account 9 excitation phonon modes; 2 translational, 2 vibrational, bending, stretching, asym-metrical stretching, stretching-vibrational combination and an overtone of the stretching modes. In the condensed water phase cross-section are expected to be higher than for ice, a common procedure among Monte-Carlo users is that cross-sections on ice films are multiplied by a factor of 2 to account for differences between solid and liquid phases, e.g. Meesungnoen et al. (2002). This factor was deduced somehow empirically by analyzing the expected results after comparing the difference between the gas phase data e.g. Banna et al. (1986) and the solid phase data of Michaud and Sanche (1987), Michaud et al. (2003), and also because it leads to a reasonable range and energy loss rate for subexcitation electrons.
-Electron attachment:
Melton’s experiments were carried on vapour water studying the energetic electron dissociation yields.Three species can result from such an interaction; O-, H- and OH-. The electron dissociation is important as it contributes to the energy loss processes and it is also important for the physico-chemical phase simulation that follows the physical stage. Summing up the molecules production cross-sections, from Melton’s experiment, one can obtain the total inelastic dissociative attachment cross-section as a function of the electron’s energy. Although there are some differences between data from different measurements,the threshold energies and the peak positions remain the same; this gives an idea about the reliability of the obtained results
Fermi density effect stands for medium polarization effects
-For ions: for ions heavier than alphas, it is complicated to calculate cross sections for every single charge state of the particle. Thus a global effective charge, including the charge change effect as well as the charge screening for the incident ion, is used. The effective charge expression of Booth and Grant is applied
-Born and Bethe have shown that the stopping power of a medium is proportional to the mass M, and to the square of atomic number, Z2, of the atoms in the medium and divided by E incoming energy of particle
See Med. Phys. 37 (2010) and Appl. Radiat. Isot. 69 (2011)

12. Electron process cross sections in liquid water
Vib. excitations
Ionization = excitation
Dissociative attachment
Electron process cross sections cover energy range up to 1 MeV down to either
7.4 eV for the partial wave elastic scattering model (default)
or 9 eV for the Screened Rutherford elastic scattering model

13. Extension to DNA material
We have implemented physics processes and models for the modeling of proton and neutral hydrogen atom interactions with
liquid water
the four DNA nucleobases : Adenine (A), Thymine (T), Guanine (G) and Cytosine (C)
within a Classical Trajectory Monte Carlo approach and taking into account specific energetic criteria
Mean energy transfers (potential and kinematical) during interactions were computed from quantum mechanics and are tabulated
First time that a general purpose Monte Carlo toolkit is equipped with such functionnalities
Dedicated Physics constructor: G4EmDNACTMCPhysics
Expected to be released in Geant4 X

14. Examples of integral cross sections
Liquid water
Adenine
Protons : single capture, single ionization, transfer ionization (new process), double ionization (new process)
Neutral hydrogen : single ionization, stripping
Adenine cross sections are of about one order of magnitude greater than their homologous in liquid water for all the ionizing processes (double ionization shows a 2 orders of magnitude ratio)

15. Is DNA equivalent to water ?
Nucl. Instrum. Methods B, in press (2013)
Is DNA equivalent to water ?
A commonly used approximation
Not necessarily true when studying elementary energy deposition events in nanometer size biological targets ...
Chromatine
DNA segment (10 bp)
Nucleosome
Incident protons

16. Need for experimental validation of theoretical cross section models
Experimental measurement of cross section in biomolecules remain very scarce today
Pr A. Itoh, Kyoto University, joined this proposal in order to propose an experimental workplan dedicated to the measurement of cross sections in biomolecules
Ion induced collisions on DNA & RNA components
Ionisation and electronic capture processes
Measurement of multi-differential and total cross sections
Will allow the validation Geant4-DNA theoretical models

17. Eg.: ionisation of Uracil
Nucl. Instrum. Methods B, in press (2013)
Eg.: ionisation of Uracil
Comparison between experimental and theoretical doubly differential cross sections (CDW-EIS and CB1: solid and dashed line, respectively) for 1 MeV-protons colliding with uracil at different electron emission angles.
First 1st Born approximation with correct boundary conditions (CB1 model) or the continuum distorted wave-eikonal initial state approach (CDW-EIS model)

18. Experimental setup
A beam of 1 MeV H+ is extracted from a Van de Graff accelerator at Kyoto University
Beam was well collimated to about 1x3 mm2 in size and was introduced into a double mu-metal shielded collision chamber.
A typical beam current of about 50 nA measured by a Faraday cup was used in this work.
An effusive molecular beam target of uracil was produced by heating crystalline uracil powder (99 % purity) in a stainless steel oven at a temperature of 473 K.
The energy and angular distributions of secondary electrons ejected from uracil molecules were measured by a 45° parallel plate electrostatic spectrometer.
The measurements were carried out for electron energies from 1 eV to 1 keV and emission angles from 15° to 165°.
Experimental errors on the obtained single and doubly differential cross sections are 8-13%.

19. b) Geometry stage: improving geometries

20. Towards DNA and cell geometries
Dose Point Kernel simulations and the CENBG« nanobeam » simulations (see Geant4 « nanobeam » advanced example) have shown that the Geant4/Geant4-DNA transport is accurate at nanometer scale
Being included in Geant4, Geant4-DNA can use Geant4 geometry modelling capabilities
We have built a cell nucleus (15 μm diameter) containing 6×109 base pairs of DNA in the B-DNA conformation
Containing randomly oriented and non-overlapping fragments of chromatine fibers
Based on voxellized nucleus phantom (see « microbeam » advanced example)

21. Frequency of energy deposition in DNA backbone
It is possible to simulate the frequency of energy deposition events in the sugar-phosphate volumes (backbone) of the DNA segments contained in the whole nucleus
This allows for the quantification of DNA direct strand breaks

22. Estimating DNA damages
Possibility to investigate direct single and double strand breaks
Results depend mainly on
Energy threshold value for induction of a single strand break
Fixed threshold (8.22 eV)
Linear probability (à la PARTRAC)
Geometrical volumes of sugar-phosphate groups (backbone)
Reasonnable agreement with experimental data
Difficult to clearly distinguish direct from non-direct effects
Nucl. Instrum. Methods B, in press (2013)

23. Including chromosomal territories
Voxellized appoach
Voxels cotain a chromatine fiber element of 6 nucleosomes, each including 2 x 100 base pair B-DNA loops
The B-DNA sequence follows the ratio 6:4 (A-T VS G-C)
46 chromosomes are built from a random walk approach
Each chromosome has a selectable overall shape
Sphere, cylinder, box
5.4x104 voxels
46 chromosomes
KEK/CENBG

24. Toward skin tissue
The geometrical model was extended to a skin-like tissue
Top view of three layers of skin-like tissue.
One layer consists 100 voxels with 100 x 100 x 10 micrometer µm3 volume each.
Each voxel contains one nucleus shown as a green sphere that includes the 46 choromosomes.
Deliver a Geant4-DNA example
KEK/CENBG

25. c) Software performance

26. Improving the computation speed of Geant4-DNA
Hot spots requiring heavy CPU usage will be identified through profiling
Improvement of existing models is proposed, focusing first on two major processes
Simulation of electron elastic scattering – dominant process at low energy
Electron & proton ionization speed-up (Born)
Main energy loss mecanism
Speed improvement on-going through switching to cumulated single differential cross sections

27. d) Application: targeted radiotherapy

28. I-125 located at cell nucleus I-125 located at cell membrane
Radionuclide therapy with 125I labeled antibodies to HER
Use Geant4-DNA physical processes: ionization, excitation, vibration excitation, electron attachment in water
Distribution of energy deposit
Electron trajectory (50keV)
100μm
100μm
to simulate energy deposition (or dose) in cell nucleus or membrane by Auger electrons (I-125)
Simple cell phantom
Energy deposit (dose)
/ decay of I-125
I-125 located at cell nucleus
I-125 located at cell membrane
Energy deposit
in nucleus (dose)
9.3keV (3.7mGy)
0.4keV (0.18mGy)
in cytoplasm (dose)
4.4keV (0.26mGy)
6.6keV (0.38mGy)
in membrane (dose)
0.001keV (0.09mGy)
0.15keV (10mGy)
human epidermal growth factor receptor, or HER1
Diameter of cell nucleus: 10μm
Size of cell: 20 μm
Thickness membrane: 5 nm
The simulation results will be useful for the optimization of injected dose and configuration of RI

29. TRI with Geant4-DNA accuracy More accurate cellular geometries
Cellular phantoms (eg. Rad Prot. Dos. 133 (2009) 2-11 )
Include new cross section models for biological targets
Theory (including quantum chemistry : GAUSSIAN)
Experiments at Kyoto U.
Simulate oxydative radical generation, interaction and transportation in cells
Other radioemitters
At-211 alpha emitter
accuracy

30. Bio-2 collaboration matters

31. Participants France Japan C. Champion CENBG K. Amako KEK S. Incerti
S. Hasegawa
NIRS
(M. Karamitros)
Y. Hirano
A. Itoh
Kyoto U.
A. Kimura
Ashikaga IT
K. Murakami
C. Omachi
Nagoya city hosp.
T. Sasaki

32. Budget request for 2013 Request from France Request from Japan
2 french visitors to KEK & Kyoto U.
Participation to experimental program
(possibly participate to a Geant4 Japanese tutorial)
Request from Japan
3 japanese visitors to CENBG for a week
Note: we do not befenit from any other funding for Japan-France collaboration around Geant4-DNA & Geant4

33. Geant4-DNA from the Internet
A unique web site for Geant4-DNA:
Fully included in Geant4

34. Thank you very much