Biological basis of ion beam therapy

As already widely discussed, ion therapy shows some interesting physical features respect to conventional radiotherapy and also to more recent proton therapy. However, what makes ion beams really attractive is their enhanced bio-logical effect. In some sense, heavier ions combine more favourable dose distri-butions of the protons and benefits from the high local ionization and, in particu-lar, carbon ions represent the best compromise between local control of the tu-mour and negative side effects.

Changes in the biological effectiveness are the result of complex interplay between physical parameters, such as ionization density and biological parame-ters. In order to understand the relation between these different aspects, it is im-portant to discuss at least from a qualitative point of view the modality of energy deposition at a microscopic scale. The deposited dose in a volume is not the only variable to take into account, because it gives indication of the total energy re-leased. For the biological response the track of the particle represents the key in-formation. For an incident charged particle, the ionization occurs along the tra-jectory of the particle and most of the energy loss is transferred to the liberated electrons, which form a sort of “electron cloud” around the trajectory of the pri-mary ion, i.e. the ion track. Finally, the action of these electrons determines the biological response together with the primary ionization. It is the higher elec-tron-density, and consequently the ionization-density, that yields a greater bio-logical effectiveness.

The formation of a particle track can be regarded more in detail as a two-steps process: first the emission of the electrons by the ion impact and second the transport of these electrons through the material around the particle track, causing secondary ionization and energy deposition. The calculation of track structure is not a trivial task and it has been subject of many publications in which different approaches have been examined. Some groups have developed semi-empirical or analytical models, starting by simplified assumptions [46][47], others have exploited Monte Carlo simulations where each basic interaction is

1.2 Rationale of ion beam therapy

31 treated individually [48-50]. Although the various models vastly differ in their basic assumptions and their input data, the obtained radial dose distributions are very similar.

The diameter of a track depends on the range of the electrons and, conse-quently, on the velocity of the ion. At higher energy the track is wide and the en-ergy loss is low, therefore the ionization events are well separated. With decreas-ing energy, the track narrows and the energy loss becomes larger. Consequently, the produced damage has a higher local density resulting in a diminished repara-bility of the lesion and, therefore, an increased biological effect.

Figure 1.11. Left: schematic view of an undamaged part of DNA (A), two separate sgle strand breaks (B), a double strand break (C), and a “clustered lesion” (D); the (*) in-dicate a base damage. Right: the structure of a proton and a carbon track in nanometre resolution with a schematic representation of a DNA molecule; the higher density of secondary electrons, produced by carbon ions, creates a large amount of clustered DNA damage.

The main target of the radiation attack is the DNA (deoxyribonucleic acid) inside cells nuclei. DNA is a very complex system and its integrity is essential for cells survival. Therefore DNA is highly protected by an extremely elaborate repair system so that DNA violations like single strand breaks (SSB) or double strand breaks (DSB) are rapidly restored. But when DNA is exposed to very high local doses, where local refers to the scale of a few nanometres, the DNA lesions become concentrated or clustered and repair system fails to correct the damage, as seen in Figure 1.11 (left) [45].

In general, the energy released by the primary ions is distributed over the volume of the track with a steep gradient of local dose over many orders of

32

magnitude, from milli-Grays at the maximum track radius- to mega-Grays in the center [44].

From the previous considerations is evident the difference between densely ionizing radiations, like carbon ions or more in general heavier particles, and sparsely ionizing radiations, like X-rays. The latter, in fact, produces damage in a stochastic manner, where the dose is distributed randomly throughout the cell, giving rise to more reparable lesions. For carbon ions, higher local ionization densities are reached also respect to those obtained with protons. Indeed, for the latter the energy loss is small and the individual ionization events are far from each other, giving rise to more easily reparable DNA damages (Figure 1.11, right).

From the quantitative point of view, the energy deposition along the track of the particles in the tissue is represented by the Linear Energy Transfer (LET), measured in keV/µm and defined as the ratio between the energy dE deposited by a charged particle in a track element and its length dx, considering only single collisions characterized by energy deposition within a specific value ∆:

∆

It is sometimes called also “restricted energy loss” or “restricted LET”. If no limitation in the amount of energy released in any single collision is considered, it is called “unrestricted energy loss” and it is indicated wit L_∞. Hence, the pre-vious classification of radiations is strictly dependent on LET values: high LET radiations produce more microscopic damages and, thus, they are more biologi-cally effectiveness respect to low LET radiations. Thus, in a high LET track the damage is produced in high density and consequently as “clustered lesions” that are, to a large amount, irreparable. Moreover, by considering a specific kind of particle, LET is sensibly variable with the penetration depth.

Even if there is not a sharp limit between high and low LET values, for many cell systems the biological effectiveness starts being important if LET be-comes greater than about 20 keV/µm. LET values of light ions are summarized in table 2 for the range corresponding to 200 MeV protons (262mm of water)

1.2 Rationale of ion beam therapy

33 [45]. There it is showed that the LET of carbon ions is larger than 20 keV/µm in the last 40mm of their range in water, while in the initial part of an approxi-mately 20 cm range in matter (the so called “entrance channel”) LET is smaller that 15 keV/µm. Helium shows high LET values only in the last millimetre. For protons, the range of elevated effectiveness is restricted to a few micrometers at the end of the range, which is too small to have a significant clinical impact. For ions heavier than carbon the residual range of elevated LET starts too early and extends to the normal tissues located before the tumour.

After the work done at Berkeley with neon and helium ions, in the beginning of the 1990s, carbon ions were chosen as optimal for the therapy of deep-seated tumours as the increased biological effectiveness, owing to the variation of the LET along the track, could be restricted mainly to the target volume [51].

LET (keV µm^-1) at various residual ranges in water (mm) Charged

Table 1.2. LET values for various charged particles at different residual ranges. The en-ergies of column 2 correspond to a range of 26.2 cm in water [45].

Anyway, LET alone is not the only parameter to be considered for an evaluation of the biological damage. As mentioned before, also parameters strictly related to the biological side have to be taken into account, such as the DNA repair capacity of the cell. The Relative Biological Effectiveness (RBE) combines together both physical and biological aspects and it is the final pa-rameter to be considered in the optimization of the dose to the patient. For ions,

34

RBE shows complex dependences on absorbed dose, particle energy, atomic number of the radiation, cell line and survival level and it is defined starting from the biological response for sparsely ionizing radiation, such as X-rays. In this case the biological response, which is quantitatively described by the sur-vival S, is a non-linear function of dose and for doses up to a few Gray cell inac-tivation can be approximated with good accuracy by a linear-quadratic expres-sion in dose D given by:

( ²)

0 D

e D

S

S = ^{α +β} (1.12)

where α and β are specific coefficients characterizing the radiation response.

Figure 1.12. Survival curves for CHO-K1 cells irradiated with X-rays (1) nand carbon ions of different energies: (2) 266.4 AMeV LET = 13.7 keV/µm, (3) 11.0 AMeV LET = 153 keV/µm, (4) 2.4 AMeV LET = 482.7 keV/µm [25].

A plot of survival in a semi-logarithmic scale leads to the characteristic shouldered survival curve Figure 1.12, with the size of the shoulder being a measure of the repair capacity of the cell tissue. For particle radiation of increas-ing LET, the β term becomes smaller and, finally, the radiation response is given by a pure linear dose relationship. The figure clearly shows that the same level

1.2 Rationale of ion beam therapy

35 of survival is reached by a lower dose in case of carbons, and, in this last case, for lower energies. This fact experimentally demonstrates the different level of damage produced by sparsely and density ionizing radiation, respectively, and its relation with LET. The different action of radiation is quantitatively described in terms of the RBE, which is defined as the ratio of the sparsely ionizing radia-tion (mostly 220 keV X-rays) and the dose of particles radiaradia-tion producing the same biological effect [52]:

particle ray X

D

RBE= D ⁻ (1.13)

Thus, by definition RBE for photons is 1. Because RBE refers to the linear quadratic X-ray dose effect curve, it strongly depends on effect level: it is high for low doses and it decreases with increasing dose. Thus the effect level has to be always given as an index with the RBE level. The ratio of the α-terms as the initial slope of the dose effect curve defines the maximum value RBEα. The value at the 10% survival level RBE10 is often used in literature for comparisons.

In radiotherapy, the topic of the correct RBE has been always crucial for par-ticles therapy and RBE values have to be used that are as close as possible to re-ality. As mentioned at the beginning of the chapter, the first neutron trials failed because too little was known about RBE, especially, about the behaviour of RBE when the dose is not applied in a single exposure but in many small fractions.

For the first period of heavy ion therapy at the Bevalac a large effort was in-vested for RBE determination in in-vitro experiments. The RBE determination at NIRS mainly uses in-vivo data from corresponding neutron trials and from ani-mal experiments using Carbon beams while the GSI therapy developed a theory for the calculation of RBE that is based on X-ray data taken from the same tu-mour or healthy tissue that is under particle exposure. For proton therapy a clini-cal RBE of 1.1 is widespread used.

Figure 1.12 also shows that different energies of the carbon ions yield differ-ent dose effect curves. Qualitatively, this can be easily explained: for high ener-gies the track is wide and the LET low, thus the ionisation events occur far

36

enough to make repair possible, yielding shouldered curves similar to sparsely ionising radiation. When energy decreases the diameter of the track shrinks and the LET increases. This leads to a higher ionisation density where the ionisation events occur closely together with a high possibility of interaction, diminishing the influence of repair and yielding a significantly increased RBE. At very high LET values at the end of the particle range (for carbon ions this is above 200 keV/µm) the local dose density becomes higher than necessary for a lethal dam-age and RBE decreases again [25]. In other words in these conditions there is an over production of local damage, which gives rise to the so called over-kill ef-fect, resulting also in an effective saturation, while the dominator of equation (1.13) continues to increase linearly. This effect is shown in Figure 1.13 con-verted into a depth distribution in water.

Figure 1.13. RBEα for carbon irradiated CHO-K1 cells. Data from W. K. Weyrather [53].

According to what has been previously discussed it could be possible to summarize that ions heavier than carbon could achieve the same biological ef-fects or even enhanced. That is in part true but high RBE characterizing a spe-cific radiation is not enough to say that it will be suitable for therapy because, as already stated before, what is crucial is the compromise between sterilization of

1.2 Rationale of ion beam therapy

37 the tumour and damage to the surrounding healthy tissues. Hence, the impact of the enhanced RBE on tumour killing is higher when the RBE maximum overlaps sufficiently with the Bragg maximum, thus getting together both effects of dose and high RBE but, at the same time, minimizing the biological effects before the peak. For carbon this condition is well satisfied, in fact the strongly elevated RBE region is restricted to the end of the particle range, where RBE has values ranging from 2 to 5, while in the entrance channel it is about 1, which means that reparable DNA damage predominates. In Figure 1.13 that behaviour is already clearly showed but it is demonstrated unambiguously if comparisons with other heavier ions are shown.

Figure 1.14. RBE for C, Ne, Si and Ar ions as function of the penetration depth. For C ions RBE is relatively low at the entrance and becomes higher closeness the peak.

Indeed, for ions like Neon, Silicon and Argon the irreparable damage be-comes more important in entrance channel, too (Figure 1.14). This is one of the reasons why carbon ions have been considered as the most suitable for had-rontherapy also from the biological point of view and they were chosen in the beginning of 1990 as optimal for treating deep-seated tumours.

However, because RBE depends also on the possibility to repair the damage produced in the DNA, the repair capacity of the irradiated tissues becomes rele-vant. In this contest, a major problem in radiotherapy is the oxygen effect, which

38

means that irradiation in the presence of oxygen causes higher biological dam-age than in the absence of oxygen. The ratio of the doses leading to the same ef-fect in aerobic and in hypoxic cells or tissue is called Oxygen Enhancement Ra-tio (OER). The effect is related to the indirect acRa-tion of the radiaRa-tion; indeed the reaction is mainly caused by radiation induced free radicals, which cause bio-logical damage after a chain of events. In the absence of oxygen most of these ionized target molecules can repair themselves and recover. If oxygen is present, it reacts with the free radicals and in this way “fixes” the radiation lesion. Thus, the often hypoxic tumours may be up to three times more radio-resistant than the well blood-supplied, and therefore aerobic, normal tissues [54]. As the damage from high LET particles does not differ for aerobic or anaerobic cells, the OER is smaller for heavy stopping particles (carbon and heavier). In this case, RBE is higher for the poor oxygenated tissues [55] and, consequently, high LET radia-tion is more effective. Hence, ion beams are better suited to slow growing and good repairing tumours, which can be radio-resistant to photons or protons.

In order to compare the RBE dependence over a large number of experi-ments with different LET radiation, RBE is frequently given as function of the linear energy transfer. Figure 1.15 summarizes in a qualitative way the RBE-LET relationship for cell inactivation, obtained from a “historical” series of ra-diobiological data [56]. Even if it should be considered that the RBE depends on a number of parameters other than LET, the general trend is an increase of RBE with LET, up to a maximum in the region 100-200 keV/µm, while the RBE de-crease observed for greater RBE values is related to the previously discussed over-kill effect. This figure adds some more hints about the rationale for choos-ing the most suitable particles to be used in hadrontherapy. Protons show radio-biological properties close to those of photons, so that their main advantage re-lies on the superior dose distribution compared to low-let radiations.

Moreover the figure shows that among light ions heavier than protons, car-bon ions have the most suitable RBE-LET combination for the energy range of interest in hadrontherapy. Indeed, LET values in a SOBP needed to treat deep-seated tumours typically cover the range 40-90 keV/µm. This range is still be-fore the maximum in the RBE-LET relationship, therebe-fore bebe-fore the drop due to

1.2 Rationale of ion beam therapy

39 the over-kill effect. As a consequence, carbon ions give the best combination for RBE at the tumour and RBE at the entrance [57].

Figure 1.15 Summary of LET (here intended as L∞) dependence of RBE at 10% survival level. The band includes the results for nine cellular systems exposed to carbon, neon and argon ions at Bevalac (adapted from Blakely et al. [56]). The LET ranges available with protons, carbon and neon ions at the energies of interest in radiobiology and radio-therapy are also shown.

Even if what discussed above gives a reasonable and general description of RBE-LET relation, in reality dependence of RBE on LET is different for ions, showing a separate maximum for each atomic number, shifting from 25 keV/µm for protons [58] to higher LET values for heavier ions. In particular maximum measured for carbon ion beams was found at 200 keV/µm [59]. For He ions the RBE maximum is higher than for Carbon and so on. Moreover, the location of the RBE maximum is shifted to greater LET values with increasing atomic num-ber. The finding of separate RBE maxima at different LET values for different atomic numbers indicates that the general energy deposition or LET of a particle does not determine the biological response alone. The experimental finding can be explained with the assumption that the local distribution of ionization density

40

inside the particle track is as important as the total energy, as already pointed out (Figure 1.11). The LET value gives the energy released in a track element but the radial distribution of the dose depends on the projectile energy. Conse-quently, the combination of the two parameters, LET and energy, finally deter-mine the RBE and its position in the LET spectrum.

These considerations imply that in order to take into account biological ef-fects, beyond the physical dose, the spatial distribution of LET, energy and flu-ence of all the particles has to be known. This is mandatory in case of mixed ra-diation fields, as for carbon ion treatments, where LET and energies of primary particles as well as of secondaries have to be considered for an accurate treat-ment planning. Such a kind of calculation is a very complicated task, which sim-ple analytical algorithms cannot easily carry out. Monte Carlo simulations are useful for these calculations, because they are able to consider the nuclear

These considerations imply that in order to take into account biological ef-fects, beyond the physical dose, the spatial distribution of LET, energy and flu-ence of all the particles has to be known. This is mandatory in case of mixed ra-diation fields, as for carbon ion treatments, where LET and energies of primary particles as well as of secondaries have to be considered for an accurate treat-ment planning. Such a kind of calculation is a very complicated task, which sim-ple analytical algorithms cannot easily carry out. Monte Carlo simulations are useful for these calculations, because they are able to consider the nuclear