Radiotherapy with Heavy Charged Particles

G. Kraft, GSI Biophysik, Planckstraße 1, 64291 Darmstadt, Germany



Only nine years after the advantages of an inversed dose profile has been recognised by R. Wilson, the first proton therapy was put into operation at Berkeley in 1954. To this day, 45 years later, 20,000 patients have been treated successfully with heavy-charged particles. The majority of them with protons and only a few hundred with heavier ions like neon at Berkeley and carbon at Chiba.

In consequence, proton therapy starts to become the treatment modality of choice when an improved physical selectivity and a greater tumor dose is the basic condition for tumor control. In the next years another four proton therapy facilities are scheduled for the US and another three for Japan. In Europe the spot scanning technique, developed at Villingen in Switzerland, opens a new field of target-conform treatment.

In the case of the heavier ions the clinical trials at Berkeley were terminated because of the shut down of the Bevelac accelerator. Presently, the therapy at Himac at Chiab in Japan is the only heavy-ion facility in operation where a significant number of patients are treated. The heavy-ion therapy at GSI in Germany is an experimental facility, where an extreme tumor-conform treatment combined with a radiobiology-based treatment planning will explore the advantage of an improved biological selectivity which should be of clinical relevance for slow growing, radioresistant tumors..


In the 100 years’ history of radiation therapy two ways for better tumor control have been successfully applied (1). The first was the use of higher photon energies in order to improve the dose localization and to shape the irradiated volume according to the tumor contours: Therefore, some decades ago the low voltage x-ray machines were replaced by the megavolt therapy using Co-gamma radiation, that is now replaced by the Röntgen Bremsstrahlung from high energy electron linear accelerators. These high energy photons combined with inverse treatment planning from multiple ports produce extremely well defined dose contours even for deep-seated tumors. However, due to the physical nature of an exponential dose decay with increasing penetration depth the integral dose to the healthy tissue mostly exceeds the target dose and causes severe complications.

The second way to improve radiotherapy was the change of the radiobiological interaction mechanisms. Neutron beams are densely ionizing radiation because of the high linear energy transfer (LET) and are able to kill radioresistant cells with high efficiency. Therefore, the local tumor control by neutron irradiation is drastically improved, especially for radioresistant tumors. Unfortunately, neutron beams show a similar dose depth curve like Co-gamma radiation, thus producing a high amount of biologically very effective damage in the healthy tissue around the tumor. In consequence, neutron therapy has been terminated in the most facilities in spite of the good tumor control because of its severe side effects.

For further development of radiotherapy, beams of heavy-charged particles like carbon ions yield a much better dose distribution than any other modality and have an increased RBE which can be restricted to the target volume (2).


In contrast to photons, heavy-charged particles like protons or heavier ions like carbon exhibit an inversed dose profile: With increasing penetration depth the energy deposition i.e. the dose increases to a sharp maximum, the Bragg peak (Fig. 1). Beyond the Bragg peak the dose has a sharp fall off. The position of this Bragg maximum in depth depends on the primary energy of the beam and can be shifted by varying the particle energy.

It was R. Wilson who first realized the medical potential of the inversed dose profiles of protons and carbon ions for therapy in 1946 (3). But it took twelve years before the first patient was treated with a proton beam at LBL in 1958. In these first exposures the beam was mainly focussed on small lesions like the pituitary, a small gland in the center in the head. But also other locations were tried. In this very first study, particle doses had to be established that were adequate to control the tumor but would not cause elevated side effects to the healthy tissue.

At this early time particle radiobiology was also at its beginning and most knowledge for therapy had to be taken from patient response and could not be predicted by radiobiological measurements. Therefore, patient treatment had to start very exploratory. In addition, there was the misfortunate experience of the neutron therapy that had just been terminated for the first time.

However, particle therapy developed to be a great success: Presently twenty thousand patients have been treated with heavy charged particles mainly protons showing very good results. In some indications like the melanoma of the eye or special tumors of the skull base and prostate cancer where the precise targeting is important, proton therapy became the treatment of choice. A pioneering role for proton treatment played the Harvard University where approximately one third of the world total has been treated. The good results of these trials yielded the construction of the Loma Linda facility, a proton therapy that treats presently the largest number of particle patients.

In addition at Harvard, a new proton facility is under construction and three proton units have been ordered by the Tecnet Corporation a private company operating a large number of US hospitals. Apart from the US, Japan and the Soviet Union have been pioneering proton therapy, too. While Russia has difficulties, new facilities for particle treatment are planed and under construction in Japan (4). In Europe, France has two proton facilities for the treatment of eye melanomas, at Nice and Paris that will reach larger penetration depth in near future. In Switzerland at PSI a new technique of conform beam delivery, the spot scanning has been developed and incorporated in a gantry system allowing beam delivery from all angles.

From these developments it is evident that proton therapy will play an important role in future radiotherapy. Precision radiotherapy and the initial attempts of proton therapy using old and disposed nuclear accelerators have successfully initiated a large innovation in medicine.


Because of the great success of proton therapy new approaches are presently studied in order to further increase the efficiency of radiotherapy. There are two ways for significant improvement: One is the use of active scanning systems in order to reach the maximal possible conformation of the irradiated volume with the planed target volume. The second is to use heavy ions to improve the dose distribution and to produce an increased biological efficiency in the tumor only.

In the particle therapies running up to now the shaping of the target volume is performed in straight analogy to photon therapy using passive absorber systems like degrader foils, collimators and boli. A much better way of beam shaping can be achieved by deflecting the charged particles using magnets. This is partially realized in the spot scanning of PSI where the beam is deflected in one dimension and the patient is moved in the other. A fully 3 dimensional conformity can be reached using the rasterscan system described below.

Heavy ions like carbon exhibit the most precise physical dose distribution, because lateral and range scattering decreases quadratically with atomic number and has reached for realistic treatment depth values still below one millimeter. For these penetration depth a low dose tail is found beyond the Bragg maximum. This dose tail originates from lighter products of nuclear reactions. Concerning the precision of dose distribution this process is not wanted but for carbon it is not significant. However a part of these nuclear reactions are producing positron emitters and can be used to trace the beam inside the patient by Positron Emission Tomography (PET). This is a very important and new feature in radiotherapy for the tracing of the beam inside the patient without any additional dose application.

Finally heavy ion beams like carbon change their relative biological efficiency along their path. At the entrance, the energy of the carbon ion is high and the ionisation density is low producing a biological efficiency close to sparsely ionizing radiation, like photons where the uppermost part of the initial damage can be repaired. At the end of the carbon range, in the last two centimeters the local ionisation density increases causing biological damage that is difficult to repair. In consequence radioresistant tumors that have a high repair capacity against photon irradiation become sensitive for heavy ion treatment.

A large facility for heavy ion treatment started its operation three years ago at Chiba, Japan. Although Chiba is up to now not equipped with the active scanning system allowing for extreme precision and online PET, the results are very good and triggered further heavy ion facilities in Japan that are presently in planing or under construction (4). A facility where the whole beauty of the advantages of heavy-ion therapy are realized exists only at GSI in Darmstadt.


At GSI, a novel technique of beam delivery has been developed: the intensity-controlled rasterscan system (5). In the rasterscan technique the target volume is dissected in slices of equal particle range and each slice is painted with a dose using a small pencil beam having a diameter of a few millimeters only. For this procedure each slice is covered by lines of picture points, pixels, and the beam is sweeped from pixel to pixel using two pairs of fast deflection magnets. For each pixel the number of particles has been calculated to achieve lateron the desired biological effect. The requested particle distribution is normally not homogeneous over the treated field because planning corrections have to be made for density inhomogeneities as well as for the effects of previous dose depositions when more distal layers will be treated and in addition for variations in the RBE. With the rasterscan technique these variations can be followed to a large extend in the same way as a TV is able to produce images of various contours and intensities (Fig. 2). But in contrast to the TV imaging the rasterscan produces a 3-dimensional ‘volume-picture’ by changing the beam energy and accordingly the penetration depth from slice to slice. In the technical realization of this concept the energy variation is produced by the accelerator. The complete particle range between 2 cm and 30 cm corresponding to 80MeV/u to 430MeV/u particle energy is dissected into 255 steps but a subset of 30-60 energies is usually needed to fill the volume (6). A complete treatment of a larger tumor in the skull basis using 60 energy slices took 12 minutes altogether and can be reduced in time furtheron. These short treatment times are made possible by the high speed of the scan which is in an order of 10m/sec. The overall treatment time is rather determined by the speed of the safety system than by the maximum possible speed of the scanner.


The basic issue of the safety system is to protect the patient against any possible failure. In such a complex machinery like a heavy-ion accelerator many components have to be controlled for a correct beam delivery (7). For the safety system only components that could cause an irradiation at a false position or a false intensity of the beam are important. Malfunction that could cut off the beam (like vacuum problems) are not risky but may appear as warning signals. The inhibitive signals of the safety system are mainly created by the beam diagnostics at the treatment area. There, the beam is monitored in a non-destructive way in a veto counter in front of the raster system and its localization and intensity is measured just before it enters the patient. For the latter purpose a pair of ionisation chambers and multiwire proportional counters are installed that compare of the beam status with the precalculated values every 100m /sec. Intensity deviations of a single pixel up to 5% yield warnings, larger deviations cause interrupts. Spatial deviation of the beam position of more than 30% of the halfwidth of the beam cause interrupts, too. From the experience with many phantom irradiations and the first patient treatments it is obvious that these extremely strict conditions can be fulfilled. Only very few beam interrupts occurred during the daily course of treatments. This demonstrates that the synchrotron beam achieves a very high stability.

The fully active beam delivery has also the important advantage that no individual beam shaping elements have to be manufactured for each individual patient in time taking and expensive procedures. In order to change the contours of the target volume only the computer control has to be changed. This makes the system not only more flexible but also much cheaper than the passive beam-shaping methods used in conventional and heavy-particle therapy up to now.


Due to nuclear reactions, a small fraction of the stable carbon ions is converted to 11C and 10C. Both isotopes are radioactive and decay with a half life of 20 min and 19 sec, respectively under the emission of a positron (8). The positron annihilates with a target electron and emits two 511 keV gamma rays coincidentally under 180o. A fraction of these coincident events can be detected by two gamma cameras on opposite sites of the patient and their origin i.e. the region of the stopping carbon ions can be reconstructed immediately after each treatment session. Using this technique of positron emission tomography (PET) it is possible to verify with a precision of two millimeters whether the target volume was irradiated correctly. The use of PET for the localization of stopping ions as developed by the FZ Rossendorf represents a completely new technique of on vivo monitoring. It is only possible for heavy ions like carbon but not for proton beams. PET monitoring guarantees also for complicated tumor locations the correctness of the application of the beam in each fraction. In case of uncertainties a small dose suffices to verify the target volume by PET prior to the main irradiation. This will make the heavy-ion irradiation an ideal tool for radiosurgery. A small dose can be given first to verify the irradiated volume before the main treatment is delivered.


In addition to the physical selectivity, superior to any other kind of radiation including protons, heavier ions like carbon supply a greater biological selectivity because of their increased LET and consequently an increased RBE at the end of their range. DNA experiments measuring the fraction of irreparable double strandbreaks after heavy ion exposure revealed that at the end of the range of the carbon ions the rate of repair drops from 80% to 20% or even less in experiments using cultured cells (9). This finding has also been confirmed in inactivation measurements (Fig. 3). However, the relative biological efficiency RBE is no parameter that can be measured in vitro experiments and then transferred as a fixed number to the exposed tissues in patient treatment. As increased RBE is caused by reduced repair, the repair capacity of the different tissues determines the RBE. Generally it can be said, that slowly growing tumors - usually very radioresistant to photon irradiation - show the largest effect in RBE when exposed to carbon beams while for radiosensitive and fast proliferating tumors the gain in biological efficiency is smaller. But even in this case the extreme tumor-conform dose delivery always remains a strong argument for carbon ions.

RBE and treatment planning

Because of the different radiosensitivities of the tissues involved in patient treatment it is neither possible to determine experimentally the RBE distribution over the treatment fields for each individual patient nor for typical treatment scenarios. Therefore, it was fundamental for the heavy-ion therapy to develop a theory of RBE that allows to calculate the RBE i.e. the most likely tissue reactions for a given dose. The local effect model (LEM) explains the RBE on the basis of the X-ray sensitivity of the irradiated tissue and the radial dose distributions within the particle tracks and their dependence on energy and atomic number of the particles (10).

In many experiments, where in vitro systems as well as animals had been used, the accuracy of the LEM was tested with great success for single fraction exposure and for fractionated exposure up to many fractions of the skin or nerve tissue.

In treatment planning the results of LEM are combined with the dose optimization using a physical beam model that includes beam fragmentation, energy loss angular scattering, etc. (Fig. 4). This treatment planning system is an adaptation of the Voxelplan system of the DKFZ Heidelberg to the modalities of heavy particles (11). Apart from the biology, the major problem of heavy particles is the density inhomogeneity caused by the traversal of the beam through very different types of tissue like skin, fat, bone, muscles and air. In order to correct the particle ranges regarding these inhomogeneities the gray values given as Hounsfield numbers of the CT scan are transformed to density values that are taken into account in the planning procedure.

The physical optimization of the treatment field takes only a few seconds on the computer while the biological optimization that takes into account all the different tissue sensitivities as well as the dose levels take much longer. However, for the first time the treatment planning in radiotherapy is based on biology and not on the physical dose optimization that is a surrogate only of the biological effect to be achieved.

First patient treatment at GSI

The first two patients using the elaborate scan and control systems of GSI have been treated successfully in December 1997. These patients suffered from large tumors at the base of skull and were exposed to a carbon boost during their treatment with conforming photon therapy at Heidelberg. The target volume was assessed using CT scans and treatment planning performed optimizing the biological effect. Each patient was treated in each session with two nearly opposing fields and a total target dose of 3.5 Gye was given in each fraction. Beam delivery was monitored online using the wire chambers combined with the ionization chambers and with the PET system. This carbon boost treatment was given within five daily fractions. Although the safety limits for this treatment were very narrow, there was no major interrupt. In Figure 5, the expected PET signal is compared with the actual measured signal. These data are compared to the measured PET signal and confirm the extreme precision of carbon therapy.


The logical extensions of the leading trend of external radiotherapy in the last century is the use of heavy ions like carbon because of the ultimate dose distribution that can be monitored by PET and the increased RBE in the tumor. First patient treatments at GSI demonstrated the feasibility of an extreme target-conform and biologically optimized carbon treatment in full beauty. The target-conform treatment was performed without major interrupts and PET measurement during irradiation as well as MRI later confirmed the precision as well as the great effectiveness of carbon ions. Larger clinical trials with carbon beams have to confirm these initial findings.


The heavy-ion therapy at GSI is a joint project of the radiobiological clinic and the German cancer research center and the FZ Rossendorf with GSI Darmstadt. I have to thank many colleagues from these institutions for valuable help and discussion. Especially, I would like to extend these thanks to people from the GSI biophysics group.

Note: parts of this manuscript were also used in other proceedings.


Figure Captions


Comparison of the depth dose distribution of photon of various energies with a carbon beam.


This famous portrait of Albert Einstein was recorded using a 430MeV/u carbon beam. It demonstrates the flexibility of the rasterscan system that is able to produce homogeneous low dose and high dose regions as well as rapid transitions. (Courtesy to U. Weber, GSI)


Comparison of absorbed dose, biological effect and RBE. Top: the Bragg curve for a 270/Mev/u for two particle fluences compared with measured cell survival. Bottom panel: the corresponding RBE values were calculated in comparison to the sparsely ionizing radiation. (Courtesy to W. K.-Weyrather)


Treatment plan for a tumor in the base of a skull shown as iso-doses that yield a homogeneous biological iso-effect in the target volume (11).


PET verification of the treated volume. Top: expected PET distribution according to the dose distribution of Fig. 4. Bottom: Measurement of PET distribution. During the exposure the annihilation of the 10,11C positrons are measured and the area of stopping carbon ions can be reconstructed that is not identical with the dose distribution but indicates the accuracy of the irradiation (8).




A. Wambersie, The future of high-let radiation in cancer therapy. Justification of the heavy-ion therapy programs, Proceedings of Eulima Workshop on the Potential Value of light ion beam therapy, Nice, November 1988


G. Kraft, The radiobiological and physical basis for radiotherapy with protons and heavier ions, Strahlentherapie und Onkologie 166, (Nr. 1), pp. 10-13, Urban & Vogel, 1990


R.R. Wilson: Radiobiological use of fast protons, Radiology 47, 487, 1946


G. Kraft: heavy-ion therapy at GSI in J. Sisterson: Particles, No 21, 1998


Th. Haberer, W. Becher, D. Schardt, G. Kraft

Magnetic scanning system for heavy-ion therapy

Nucl. Instr. and Meth. in Phys. Res. A 330, 296-305, (1993)


H. Eickhoff, B. Franczack, U. Krause, C. Riedel, R. Steiner,

Accelerator developments and tests for the GSI therapy project

Scientific Report GSI Darmstadt, 1996


E. Badura, H. Brand, H.G. Essel, H. Hardel, J. Hoffmann, N. Kurz, P. Liebold, W. Ott, K. Poppensieker, G. Ribitzki, M. Richter, D. Schall

A control and safety system for the therapy project at GSI

Scientific Report GSI Darmstadt, 1996


W. Enghardt, W.D. Fromm, H. Geissel, H. Keller, G. Kraft, A. Magel, P. Manfraß, G. Münzenberg, F. Nickel, J. Pawelke, D. Schardt, C. Scheidenberger, M. Sobiella, The spatial distribution of positron-emitting nuclei generated by relativistic light-ion beams in organic matter, Phys. Med. Biol., 37, 2127-2131 (1992)


J. Heilmann, G. Taucher-Scholz, T. Haberer, M. Scholz and G. Kraft

Measurement of intracellular DNA double-strand break induction and rejoining along the track of carbon and neon particle in water

Int. J. Radiat Oncology Biol. Phys. Vol. 34, No. 3, 599-608, 1996


M. Scholz, G. Kraft, Calculation of heavy-ion inactivation probabilities based on track structure, X-Ray sensitivity and target size, Rad. Prot. Dosi. 52, 29 (1994)


M. Krämer and O. Jäkel, Treatment planning for heavy-ion irradiation I/II in "Heavy-Charged Particles in Biology and Medicine", GSI-Report 09/97, 1997



Titel: Radiotherapy with Heavy Charged Particles

Author: G. Kraft

Affiliation: GSI Biophysik, Planckstraße 1, 64291 Darmstadt, Germany

Headings: SUMMARY






In vivo beam localization by PET

Radiobiological advantages of heavy ions

RBE and treatment planning

First patient treatment at GSI



Figure Captions