UMEÅ UNIVERSITY MEDICAL DISSERTATIONSNew series No 315 - ISSN 0346-6612
From the Department of Oncology, University of Umeå, Umeå, Sweden
BIOLOGICAL EFFECTS OF HIGH ENERGY RADIATIONAND
ULTRA HIGH DOSE RATES
Björn Zackrisson
University of Umeå Umeå 1991
UMEÅ UNIVERSITY MEDICAL DISSERTATIONSNew series No 315 - ISSN 0346-6612
From the Department of Oncology, University of Umeå, Umeå, Sweden
BIOLOGICAL EFFECTS OF HIGH ENERGY RADIATIONAND
ULTRA HIGH DOSE RATES
AKADEMISK AVHANDLINGsom med vederbörligt tillstånd av Rektorsämbetet vid
Umeå universitet för avläggande av medicine doktorsexamen kommer att offentligt försvaras i Onkologiska klinikens föreläsnings
sal, 244, 2 tr, torsdagen den 21 november 1991, kl 09.00
av
Björn Zackrisson
Umeå 1991
ABSTRACT
BIOLOGICAL EFFECTS OF HIGH ENERGY RADIATION AND ULTRA HIGH DOSE RATES.Björn Zackrisson, Department of Oncology, University of Umeå, S-901 85 Umeå, Sweden.
Recently a powerful electron accelerator, 50 MeV race-track microtron, has been taken into clinical use. This gives the opportunity to treat patients with higher x-ray and electron energies than before. Furthermore, treatments can be performed were the entire fractional dose can be delivered in parts of a second.
The relative biological effectiveness (RBE) of high energy photons (up to 50 MV) was studied in vitro and in vivo. Oxygen enhancement ratio (OER) of 50 MV photons and RBE of 50 MeV electrons were investigated in vitro. Single-fraction experiments, in vitro, using V-79 Chinese hamster fibroblasts showed an RBE for 50 MV x-rays of approximately 1.1 at surviving fraction 0.01, with reference to the response to 4 MV x- rays. No significant difference in OER could be demonstrated. Fractionation experiments were carried out to establish the RBE at the clinically relevant dose level, 2 Gy. The RBE calculated for the 2 Gy/fraction experiments was 1.17. The RBEs for 20 MV x-rays and 50 MeV electrons were equal to one. In order to investigate the validity of these results, the jejunal crypt microcolony assay in mice was used to determine the RBE of 50 MV x-rays. The RBE for 50 MV x-rays in this case was estimated to be 1.06 at crypt surviving fraction 0.1. Photonuclear processes are proposed as one possible explanation to the higher RBE for 50 MV x-rays.
Several studies of biological response to ionizing radiation of high absorbed dose rates have been performed, often with conflicting results. With the aim of investigating whether a difference in effect between irradiation at high dose rates and at conventional dose rates could be verified, pulsed 50 MeV electrons from a clinical accelerator were used for experiments with ultra high dose rates (mean dose rate: 3.8 x 10^ Gy/s) in comparison to conventional (mean dose rate: 9.6 x 10" ̂ Gy/s). V-79 cells were irradiated in vitro under both oxic and anoxic conditions. No significant difference in relative biological effectiveness (RBE) or oxygen enhancement ratio (OER) was observed for ultra high dose rates compared to conventional dose rates.
A central issue in clinical radiobiological research is the prediction of responses to different radiation qualities. The choice of cell survival and dose response model greatly influences the results. In this context the relationship between theory and model is emphasized. Generally, the interpretations of experimental data are dependent on the model. Cell survival models are systematized with respect to their relations to radiobiological theories of cell kill. The growing knowledge of biological, physical, and chemical mechanisms is reflected in the formulation of new models. This study shows that recent modelling has been more oriented towards the stochastic fluctuations connected to radiation energy deposition. This implies that the traditional cell survival models ought to be complemented by models of stochastic energy deposition processes at the intracellular level.
Key words: Radiobiology, 50 MV x-rays, 50 MeV electrons, in vitro, in vivo, RBE, OER, ultra high dose-rate, radiobiological models, model selection, models of stochastic processes.
UMEÅ UNIVERSITY MEDICAL DISSERTATIONSNew series No 315 - ISSN 0346-6612
From the Department of Oncology, University of Umeå, Umeå, Sweden
BIOLOGICAL EFFECTS OF HIGH ENERGY RADIATIONAND
ULTRA HIGH DOSE RATES
Björn Zackrisson
University of Umeå Umeå 1991
Copyright 1991 (C) Björn Zackrisson
ISBN 91-7174-614-5
Printed in Sweden by the Printing Office of Umeå University
Umeå 1991
"The great tragedy of Science - the slaying of a beautiful hypothesis by an ugly fact". T.H. Huxley 1883
ABSTRACT
BIOLOGICAL EFFECTS OF HIGH ENERGY RADIATION AND ULTRA HIGH DOSE RATES.Björn Zackrisson, Department of Oncology, University of Umeå, S-901 85 Umeå, Sweden.
Recently a powerful electron accelerator, 50 MeV race-track microtron, has been taken into clinical use. This gives the opportunity to treat patients with higher x-ray and electron energies than before. Furthermore, treatments can be performed were the entire fractional dose can be delivered in parts of a second.
The relative biological effectiveness (RBE) of high energy photons (up to 50 MV) was studied in vitro and in vivo. Oxygen enhancement ratio (OER) of 50 MV photons and RBE of 50 MeV electrons were investigated in vitro. Single-fraction experiments, in vitro, using V-79 Chinese hamster fibroblasts showed an RBE for 50 MV x-rays of approximately 1.1 at surviving fraction 0.01, with reference to the response to 4 MV x- rays. No significant difference in OER could be demonstrated. Fractionation experiments were carried out to establish the RBE at the clinically relevant dose level, 2 Gy. The RBE calculated for the 2 Gy/fraction experiments was 1.17. The RBEs for 20 MV x-rays and 50 MeV electrons were equal to one. In order to investigate the validity of these results, the jejunal crypt microcolony assay in mice was used to determine the RBE of 50 MV x-rays. The RBE for 50 MV x-rays in this case was estimated to be 1.06 at crypt surviving fraction 0.1. Photonuclear processes are proposed as one possible explanation to the higher RBE for 50 MV x-rays.
Several studies of biological response to ionizing radiation of high absorbed dose rates have been performed, often with conflicting results. With the aim of investigating whether a difference in effect between irradiation at high dose rates and at conventional dose rates could be verified, pulsed 50 MeV electrons from a clinical accelerator were used for experiments with ultra high dose rates (mean dose rate: 3.8 x 10^ Gy/s) in comparison to conventional (mean dose rate: 9.6 x 10‘2 Gy/s). V-79 cells were irradiated in vitro under both oxic and anoxic conditions. No significant difference in relative biological effectiveness (RBE) or oxygen enhancement ratio (OER) was observed for ultra high dose rates compared to conventional dose rates.
A central issue in clinical radiobiological research is the prediction of responses to different radiation qualities. The choice of cell survival and dose response model greatly influences the results. In this context the relationship between theory and model is emphasized. Generally, the interpretations of experimental data are dependent on the model. Cell survival models are systematized with respect to their relations to radiobiological theories of cell kill. The growing knowledge of biological, physical, and chemical mechanisms is reflected in the formulation of new models. This study shows that recent modelling has been more oriented towards the stochastic fluctuations connected to radiation energy deposition. This implies that the traditional cell survival models ought to be complemented by models of stochastic energy deposition processes at the intracellular level.
Key words: Radiobiology, 50 MV x-rays, 50 MeV electrons, in vitro, in vivo, RBE, OER, ultra high dose-rate, radiobiological models, model selection, models of stochastic processes.
6
7
CONTENTS
ORIGINAL PAPERS 9INTRODUCTION 11
Aspects on radiation energy 12Ultra high dose rates 14Radiobiological cell survival models 15
AIMS OF THE STUDY 16MATERIALS AND METHODS 17
Cell cultures 17Cell cloning assay 17Animals 17Jejunal crypt assay 17Radiation qualities 18Irradiation techniques 18Dosimetry 19Experimental design and statistical methods 22
Single-fraction experiments 22Three-fractions experiments 22Jejunal crypt assay 24
RESULTS AND COMMENTS 25RBE of high energy x-rays and electrons 25RBE and OER of pulsed high dose rate electrons 29Radiobiological cell survival models 31
CONCLUSIONS 33ACKNOWLEDGEMENTS 34REFERENCES 35
8
9
ORIGINAL PAPERS
This thesis is based on the following papers, which will be referred to by their Roman numerals.
I Relative biological effectiveness and oxygen enhancement ratio of 50 MV x-rays. Zackrisson B, Johansson B, Östbergh P. Acta Oncol 1989; 28: 529-35.
II Relative biological effectiveness of high energy photons (up to 50 MV) and electrons (50 MeV). Zackrisson B, Johansson B, Östbergh P. Radiat Res. In press 1991.
III Relative biological effectiveness of 50-MV X rays on jejunal crypt survival, in vivo. Zackrisson B, Karlsson M. Submitted.
IV Biological response, in vitro, to pulsed high dose rate electrons from a clinical accelerator. Zackrisson B, Nyström H, Östbergh P. Acta Oncol. In press 1991.
V Radiobiological cell survival models - a methodological study. Zackrisson B. Submitted.
10
11
INTRODUCTION
In radiation therapy, the development of new equipment and techniques has resulted in an improvement of clinical results. Since megavoltage x-rays became available in clinical radiotherapy, it has been possible to treat tumours with a high radiation dose without exceeding the limits for normal tissue tolerance. Today, more tumours can be controlled by radiotherapy than ever before because of an increased knowledge in radiotherapy, radiobiology, radiation physics and related research areas. Technical achievements have made it possible to use some of the new knowledge in clinical practice. Still, there are a substantial number of tumours that cannot be controlled by radiation therapy or other treatment modalities. Different approaches are needed to improve treatment results.
In radiotherapy, a major obstacle is that the delivery of higher doses, which can indeed eradicate the tumour, cause damage to normal tissues in the target volume and result in side effects that are not acceptable. A small increase in dose significantly increases the probability of tumour control but the observed side effects increase proportionally. The dose-effect relationships are often steep at these levels so that a relatively small difference in dose can have large effects in terms of tumour control and side effects (Brahme 1984, Nias 1990). The type and extent of side effects also depends on which tissues are included in the target volume.
Consequently, one objective is to find a way to deliver a higher radiation dose to tumours but not to normal tissues. Multiple field techniques or arc therapy provides a means for "focusing" the radiation on the tumour in a way that reduces the dose to tissues outside the target volume below the tolerance level. To achieve as low a dose as possible to tissues outside deep sited target volumes, the highest available x-ray energies are used, usually with accelerating potentials of 10-30 MV. In other tumour locations multiple or single fields of x-rays of lower energies or high energy electrons (or combinations) give dose distributions that are suitable. Other types of radiation such as protons, ir'-mesons, a-particles, and heavy ions give dose distributions that may be of great value in radiotherapy but are not available for routine clinical use. Presently, electron accelerators with energies up to 50 MeV are being produced for radiation therapy to provide electrons with energies up to 50 MeV and the corresponding x-ray qualities.
Another important issue is to change the effectiveness of the radiation dose by increasing the sensitivity of tumour cells relative to that of normal cells. This can, at least theoretically, be done by means of, for example, chemical substances that either act
12
as sensitizers of tumour cells or protectors of normal cells, or by fractionation of the radiation dose in a way that increases the therapeutic ratio between normal tissues and tumour. It is also well established that hypoxic cells are less sensitive to radiation than well oxygenized cells (Holthusen 1921, Read 1952). The dose to anoxic cells has to be approximately three times as large as to oxic cells in order to give equal proportions of surviving cells. This relationship is known as the oxygen enhancement ratio (OER) (ICRU 1979). In normal tissues virtually all cells are well oxygenized, whereas in tumours, hypoxic cells regularly have been demonstrated (Thomlinson and Gray 1955). This implies that in a radiation therapy course, these cells are relatively resistant to radiation, and may survive, causing a treatment failure. In fact, there are clinical data that support this theory (Henk 1986). Different approaches to overcome the problem with hypoxic cells have been used. One approach is to specifically sensitize hypoxic cells by; different chemical compounds acting as a substitute for oxygen (Urtasun et al. 1975), oxygen treatment of the patient (Churchill-Davidson et al. 1955), increasing the blood flow to the tumour or combinations of these methods (Rojas 1991, Rojas et al. 1991). Also drugs that have a toxic effect exclusively on hypoxic cells will belong to this category (Adams and Stratford 1986). An alternative way to deal with hypoxia is to use different ways of delivering the radiation dose. It is known that the OER decreases at smaller doses per fraction (Littbrand 1970). This is the rationale for fractionation of the radiation into a larger number of fractions with smaller doses per fraction during the same overall treatment time used in conventional treatment (hyperfractionated radiotherapy). Also, clinical data have shown improved treatment results with hyperfractionation (Edsmyr et al. 1985). Delivering the radiation dose continuously over a prolonged period of time as, for example, in many instances in interstitial or intracavitary treatments with radioactive isotopes, may also decrease the OER (Hall and Cavanagh 1967, Ling et al. 1985). The use of densely ionizing radiation e.g. neutron radiation is yet another way to decrease the OER (Barendsen et al. 1966).
Aspects on radiation energy
The quality of x-ray beams is often described by the accelerating potential in units of megavolt (MV). In a x-ray beam of 10 MV, there is a spectrum of energies where the maximum photon energy is 10 MeV. The mean energy of the photons is often about 1/3 of the maximum energy. Regarding electron radiation, the spectrum of energies is often very narrow and the maximum and mean energies are generally close. However, the energy given by the manufacturer, often refers to the maximum energy. As for all particle radiations the electron energy is given in its proper unit, a multiple of eV. Thus,
13
the mean energy of electrons generated by a 10 MV x-ray beam is considerably lower than that of a 10 MeV electron beam.
The relative biological effectiveness (RBE) describes the effectiveness of a radiation quality, compared to a reference radiation, in producing the same response (ICRU 1979). The RBE of sparsely ionizing radiation has been the objective of several studies (Sinclair 1962, Hettinger et al. 1965, Wambersie et al. 1966, Kim et al. 1968, Dutreix et al. 1969, Robinson and Ervin 1969, Huang et al. 1974, Williams and Hendry 1978, Amols et al. 1986, Bistrovic et al. 1986). It is well known that x-rays with accelerating potentials up to about 300 keV have a higher RBE than megavoltage x-rays and GOO) gamma-rays. This is explained by the fact that kilovoltage x-rays, when interacting with matter, have a larger proportion of more densely ionizing radiation of low energy. Densely ionizing radiation is known to be more effective in cell killing per unit dose than sparsely ionizing. The term linear energy transfer (LET) describes ionization density. Photon and electron radiations are usually referred to as low-LET radiation while e.g. neutron radiation is a high-LET quality.
The photoelectric absorption dominates the interaction processes for x-rays up to ~0.1 MV. Compton scattering dominates for x-rays from about 0.1 MV. Above 15-20 MV, production of electron pairs becomes a significant ( - 10%) photon interaction process (Hubbel 1969). In addition to this, photonuclear processes occur to a small extent for high photon energies. The probability for photonuclear processes is largest for photon energies between 20 and 25 MeV (NCRP 1984). The given examples are approximately valid for water, however, the elemental composition of the irradiated material will determine at which energy, and in which proportion, each interaction will take place.
Biological experiments have been carried out to study the RBE of megavoltage x-rays of different accelerating potentials (Sinclair 1962, Robinson and Ervin 1969, Amols et al. 1986). Most commonly, x-ray qualities up to 30 MV have been studied but results from up to 42 MV are available (Bistrovic et al. 1986). No significant differences in RBE are reported from these studies.
The interactions of electrons of different energies with matter roughly consist of ionization, excitation and loss of energy by setting delta rays in motion (ICRU 1984). The probability for an electron to interact directly with a nucleus is about two orders lower than for corresponding x-ray qualities (SCRAD 1966, Laughlin et al. 1979). Several studies of RBE for different electron energies compared to megavoltage x-rays have shown an RBE of unity (Sinclair 1962, Hettinger et al. 1965, Wambersie et al.
14
1966, Kim et al. 1968, Dutreix et al. 1969, Robinson and Ervin 1969, Huang et al. 1974, Williams and Hendry 1978).
Ultra high dose rates
The mean absorbed dose rates used in clinical external beam therapy are often in the range of 1.7 x 10"2 to 5.0 x 10"2 Gy/s (1-3 Gy/min). With the introduction of scanning electron beams, it has become possible to treat patients with ultra high dose rates (i.e. the entire radiation dose is delivered in parts of a second, which usually will mean > > 2 Gy/s (Hall 1972)).
The interest in ultra high dose rates increased dramatically when Dewey and Boag (1959, 1960) showed that bacteria irradiated with ultra high dose rates under hypoxic conditions, at high doses, exhibited the characteristics of being anoxic. This phenomenon was due to oxygen consumption in the cells by the radiation. Investigations of this phenomenon in mammalian cells failed to reveal a similar behaviour of the mammalian cell survival curves (Todd et al. 1968, Nias et al. 1969 & 1970, Berry and Stedeford 1972, Epp et al. 1972, Michaels et al. 1978).
The attention of the modified OER was due to the hypothetical possibility of making all cells in a radiation target volume act anoxically, thus eliminating the difference in response of hypoxic tumour cells compared to normal cells.
An inability of cells to repair sublethal damage after irradiation with ultra high dose rates have been reported (Nias et al. 1973). In opposition to this, Schultz et al. (1978) and Gerweck et al. (1979) found evidence for retained repair capacity.
Purdie et al. (1978, 1980) found that human kidney cells showed an increased sensitivity to high dose rates under both oxic and anoxic conditions. In addition, there was a reduction of OER. This was the case for both single pulses of electrons with dose rates of 2.5 x 107 . 2.5 x 10 ̂Gy/s and gamma irradiation with 0.7 Gy/s. Furthermore, Ling et al. (1984) found a decreased OER when comparing pulsed x-rays at mean dose rate 1.7 x 10'1 to 1.7 x 10'2 Gy/s. The dose rates in each 2 microsecond pulse were 300 and 150 Gy/s, respectively. In summary, despite many investigations in this research area, seemingly diverging results have been reported.
Recently a powerful electron accelerator, 50 MeV race-track microtron, has been taken into clinical use. This provides the opportunity to treat patients with higher x-ray and
15
electron energies than before. Furthermore, treatments can be performed where the entire fractional radiation dose can be delivered in parts of a second (i.e. an extremely high dose rate compared to conventional radiotherapy).
Radiobiological cell survival models
Radiobiological research usually precedes the start of clinical trials or the use of new equipment. This is important since new hypotheses relevant for clinical research may be formulated. In order to predict the effects, advantages, or hazards of a new treatment, an important role is also played by experimental radiobiology. Different theories in radiobiology, as in many other research fields, have resulted in different mathematical expressions to predict dose-response relationships. These models are often used to analyze experimental results. The ability to predict certain effects from these models depends on the validity of the theories related to the models.
At low doses of radiation, the dose response relationships are difficult to establish experimentally since the response is relatively small in comparison to the experimental variations. Subsequently, the studied phenomenon or the experimental system used make the inference from low doses (i.e. < 2 Gy) complicated. Therefore, the use of models for cell survival are used in radiobiology to interpret experimental results. This means that the results from levels of response, where the response to a small change in dose is relatively large, can be used for inference from the low dose levels. However, the conclusions drawn under such circumstances will be critically dependent on the validity of the theory behind the model. A number of cell survival models and related theories have been proposed. This prediction aspect has been paid special attention by some investigators (Littbrand and Révész 1969, Goodhead 1983, Palcic et al. 1983), and important deviations from the predicted- results have been shown. The dominating cell survival models have been the single-hit multitarget model (Lea 1955), and the linear- quadratic (LQ) model (Kellerer and Rossi 1972, Chadwick and Leenhouts 1973, 1981). The growing knowledge of biological, physical and chemical mechanisms of cell killing by radiation is reflected in the formulation of new models (Kappos and Pohlit 1972, Tobias et al. 1980, Goodhead 1985, Curtis 1986). Recently, the modelling process has been more aware of stochastic fluctuations of the radiation effects (Albright and Tobias 1985, Curtis 1988, Albright 1989, Sachs et al. 1990). An alternative approach for studying the stochastic processes in radiation effects is to empirically study the cell kill at fixed doses.
16
AIMS OF THE STUDY
The development, in recent years, of new technical equipment for radiation therapy, includes the use of powerful electron accelerators. Before these accelerators are taken into clinical practice the biological effects of the radiation beams produced by this type of equipment need to be investigated. In the present thesis, the effects of high energy and ultra high dose rates were investigated according to the following specific aims:
- To determine the RBE and OER of high energy x-rays (50 MV) in relation to well known radiation qualities (I).
- To find plausible explanations to a difference in RBE (II).
- To investigate whether the difference in RBE can be detected at clinically relevant dose levels (i.e. ~2 Gy/fraction) (II).
- To evaluate the validity of the results in vitro, in an in vivo experimental system (III).
- To investigate potential clinical benefits from using ultra high dose rates (IV).
- To study radiobiological survival models from a methodological point of view (V).
- To find a useful experimental method of determining RBE with limited model dependence (II).
17
MATERIALS AND METHODS
Cell cultures (I,II,IV)
Fibroblasts from lung tissue of Chinese hamster (V-79-379-A) were used for the experiments. Monolayers of exponentially growing cells were cultured on glass plates (diameter 16 mm, thickness 0.1 mm) (I,IV) or plastic tissue culture discs (Lux Scientific, diameter 11 mm, thickness 0.1 mm) (H) on the bottom of Petri dishes in Eagle’s MEM supplemented with NaHCC^, L-glutamine, penicillin, streptomycin, amphotericin B and fetal calf serum. The cells were incubated in an CC^-incubator at 37°C and a relative humidity of 90% in an atmosphere containing 5% CC>2 in air. The plates were transferred to the irradiation chamber, which is described in more detail below.
Cell cloning assay (I,II,IV)
The cells were trypsinized not more than 15 min after the irradiation, and then plated for cloning assay. Seven days after irradiation, clones were stained in situ and those comprising more than 50 cells were counted as survivors.
Animals (HI)
Male, pathogen free, CBA/J mice aged 8 weeks (range 7-9), were used. All mice were obtained in the same shipping in order to minimize inter-individual variation.
Jejunal crypt assay Hill
After total body irradiation the mice were kept under standard conditions for 72 hours, at which time they were sacrificed. The duodenum was identified and the following 10 cm of the jejunum was used for the study of microcolonies.
Neutral formalin was used for fixation of jejunum. Following fixation, each specimen was divided into 6-10 pieces, which were then bound together in a bundle with surgical tape (Potten and Hendry 1985). In this way the pieces from one individual mouse could be handled easily during the embedding procedure, and the specimen could also be oriented in such a way as to ensure true cross-sections were obtained. The specimens were embedded in paraffin.
18
Histological sections (thickness approximately 5 mm) were prepared from the embedded specimens. When several sections were made from each specimen, they were taken at least 200 /xm apart. The sections were stained with haematoxylin and eosin. Surviving microcolonies were counted in a light microscope. In order to be counted as a survivor, the crypt should contain 10 or more cells with prominent nuclei and basophilic cytoplasm. Cells of Paneth type were excluded (Withers and Elkind 1970, Potten and Hendry 1985). Since the assessment of crypt viability is partly subjective, the slides were coded in order to avoid bias.
Radiation qualities (I,II,m,IV)
20 and 50 MV x-rays and 50 MeV electrons were produced by a race-track microtron (MM50, Scanditronix, Uppsala, Sweden). The 4 MV x-rays used for reference (I,II) were produced by a linear accelerator (Clinac 4-800, Varian, USA). No flattening filters are used in the beams of the race-track microtron since they are generated by scanning 20 or 50 MeV electrons on a target. Therefore, the dose distribution and energy spectrum are constant over the radiation field. The race-track microtron was set to give a homogeneous dose distribution to the cells during the scanning cycle, and to give a mean dose rate, pulse dose rate, pulse repetition frequency, and pulse duration similar to the 4 MeV linear accelerator (I,II,HI). The radiation qualities were specified according to the NACP-protocol (NACP 1980) and the IAEA-protocol (IAEA 1987) (I-IV).
In order to get a ultra high dose rate (IV), the 50 MeV electron beam was extracted from the race-track microtron by switching off the first gantry bending magnet. Thus, the beam did not pass the built-in system of scanning magnets, foils, collimation and dose- monitoring devices. Also, the pulse duration was set at its maximum (6 ms). T o increase the homogeneity of the dose distribution at the position of the cells, a thin gold scattering foil was positioned at the exit window.
Irradiation techniques (I,II,III,IV)
A specially designed irradiation chamber was used in the in vitro studies (I,II,IV). The glass (I,IV) or plastic (H) plate on which the cells had been grown was attached to a holder in the chamber. The chamber was then sealed and filled with cell growth medium which thus surrounded the cell culture plate. The chamber was attached to a gas flow system that allowed the medium to equilibrate with a desired mixture of gases. For the experiments performed under anoxic conditions (I,IV), a mixture of 95% purified nitrogen and 5% purified carbon dioxide was used (in total < 2 ppm oxygen as measured
19
with a Hagan oxygen meter). In the oxic experiments, a mixture of 95% air and 5% carbon dioxide was used. When the chamber was attached to the gas-flow system the gas went through the cell growth medium in order to increase the gas exchange. Therefore, only serum-free medium was used since the serum protein would produce foam when gas passed through the medium. The irradiation chamber and gas flow system were manufactured from metal and glass since all plastic materials contain oxygen and may compromise the anoxic environment. This is also the reason why cells were grown on glass plates in some investigations (I,TV). The chamber was then put into a water phantom. The water was heated to a temperature that kept the cell growth medium in the chamber at a temperature of 37°C. The cells were irradiated after gas and temperature equilibration (15 min). After irradiation, cells were removed and plated for the cloning assay described above.
The mice were given total-body irradiation (HI). They were irradiated four at a time in a perforated cylindrical Lucite cage. The cage was divided into six compartments by Lucite walls. During the irradiations one mouse was kept in each of four compartments. The mice were unanaesthetised during the experiments. All irradiations were performed on the same day. The different doses and energies were delivered in random order.
In all cases (I,II,HI,TV) Lucite of appropriate thickness was positioned between the source and the cells (I,n,IV) or the abdominal cavity (HI) in order to ensure that they were positioned at the maximum dose for the radiation quality tested.
Dosimetry (I,II,IH,IV)
The absorbed doses were calculated using the NACP-protocol (NACP 1980) (I) or the LAEA-protocol (IAEA 1987) (n,HI,TV). In the conventional dose rate experiments, the mean absorbed dose in the cells was calculated from the measured mean dose to a ferrous sulphate solution (Fricke and Hart 1966). The ferrous sulphate solution was irradiated in cylindrical Lucite capsules in the same position as the culture plates. Corrections were made for any dose gradient over the dosimeter. Measurements for calibration of the dose monitor were made in the same way. Additional measurements were performed with air ionization chambers to confirm the results of the FeSC^- dosimetry.
An air ionization chamber was also used for control measurements during the cell irradiation experiments with conventional dose rates (H, TV). This ionization chamber was positioned behind the cell irradiation chamber in a fixed geometry.
20
In paper (I), the additional dose from backscatter from the glass, the cells were attached to was considered. Since the amount of backscatter could be assumed to be dependent on the radiation quality, paired experiments were performed with cells on glass plates being irradiated in parallel to cells on plastic plates. The ratios of survival from each experiment were used as response (Fig. 1). The dose modifying factor obtained in the analysis of these experiments was then used for correcting the absorbed dose calculated in the conventional way. The analysis showed that the dose modifying factor was significantly different from unity for 4 MV, but not for 50 MV x-rays.
3 . 0 -
2 . 0 -
. 5 -
10 121198
Absorbed dose / Gy
Figure 1. The ratio of surviving fraction as a function of dose for cells cultured on glass plates and plastic plates irradiated with 4 MV x-rays. The steepness of the curve represents the dose modifying factor (1.085 with 95% confidence interval ± 0.011) due to backscatter from the glass. This includes an assumption of approximate log-linearity of the survival curve in the dose range investigated. Each symbol represents the mean value of 9 individual experiments. The error bars represent the 95% confidence intervals.
In the high dose rate experiments, a coil (pulse transformer) was used for monitoring the beam current (IV). The beam passes through the coil and a current is therefore induced which is proportional to the dose rate. The total dose could be determined, by integrating the pulse transformer signals from the radiation pulses delivered. The calibration of the transformer was done by irradiation of ferrous sulphate solution. The
21
calibration of the transformer was done by irradiation of ferrous sulphate solution. The results of ferrous sulphate dosimetry is linear beyond the doses per pulse («1.6 Gy) used in this study (ICRU 1982).
The target volume in the animal experiment (ID) is the abdominal cavity of the mouse. A cylindrical (Sinclair 1969) perspex phantom with dimensions according to those measured on the mice was manufactured for the dosimetry. A calibrated air ionization chamber was inserted in the phantom at the position of the abdominal cavity. The dosimetry was done according to the LAEA-protocol (IAEA 1987) and verified at the time of the animal experiments. A comparison between this dosimetry and Fricke dosimetry was made for the two beam qualities. The radiation dose was delivered as a single field from below. In order to achieve a homogeneous dose in the abdominal cavity, specially designed build up and back scattering material was used. The thickness was individually optimized to the different beam qualities so that the dose variation within the target volume should vary less than 1 percent from the estimated average dose value (Fig. 2).
100 50 MV
0 1 2 3 4 5 wh Buildup material..Lucite I Abdominal wall I I Abdominal cavity
100 20 MV
0 1 2 3 4 5Depth / cm
Figure 2. Schematic description of the irradiation geometry. Depth dose curve, build-up material, mouse anatomy and back scattering material. The air space surrounding the mice in the experimental situation (not shown in the figure) does not significantly affect the dose distribution.
22
Experimental design and statistical methods (I-IV)
Single-fraction experiments (I,II,IV)
The survival data were analyzed using the LQ-model (Chadwick and Leenhouts 1973). In spite of its weak theory connection, it is widely used in radiobiological research for evaluating data. This model is of the form
InS = - aD - ßE>2 (1)
where S is the surviving fraction, a and ß are constants and D is the absorbed dose. The doses in the experiments were chosen partly because of the relatively small experimental errors in this dose domain. The model was given an intercept by the unirradiated controls. The parameters were estimated by the ordinary least squares method. The model was then adjusted to pass through S = 1 when D = 0. The model parameters were tested simultaneously by application of a dummy variable. The statistical F-test was applied for tests of significance. To calculate the confidence intervals a normal distribution of the experimental errors was assumed. The confidence intervals of RBE were transformed from the confidence interval of the model parameters. For each experiment a number of different dose levels and an unirradiated control were used. Each experiment was repeated at least 7 times. For practical reasons only one experiment could be performed each day.
Three-fractions experiments (II)
By using multiple fractions of radiation the possibility of detecting small differences in response is increased. In this case, 3 fractions of 2.0 Gy were delivered to the cells. The interval between each fraction was 4 h to allow for repair of sublethal damage. In each experiment, one tissue culture plate was irradiated with 50 MV one with 20 MV, and one plate remained unirradiated as a control. All three plates were drawn from the same tissue culture vessel. The cells were treated in parallel throughout the experiment. Due to the experimental design, a situation with paired observations in the statistical analysis is required. The experimental series consists of 14 pairs.
In this experiment, a partial synchronization of cells in the cell cycle might occur. However, it is shown at small doses per fraction (1.5 - 3 Gy) the effect of synchronization is very small (McLarty 1976). Furthermore, since these experiments compare two radiation qualities which are only moderately different, the possible
23
influence of redistribution is therefore assumed to affect the cells similarly in both arms and thus be reduced due to the experimental design.
In fractionated-dose experiments, in vitro, where there is complete repair between each fraction, the dose-response relationship will be approximately log-linear, provided the number of fractions is small (< 5), and the fraction sizes are equal (McNally and de Ronde 1976). In this case, the assumption of equal effect per fraction means that making observations after one or two fractions will give information of limited practical value. The surviving fractions are described by the equations
' lnS20MV = kl x D1 (2)and
-lnS50MV = k2 x D2 (3)
where S20MV an<̂ S50MV rePresents surviving fraction at 20 and 50 MV x-rays respectively, and k2 are the regression coefficients. and D2 are the total absorbed doses. The intercepts are 0 since S = 1 when D = 0. By definition RBE = 'D^/T>2 wken
-lnS20MV = "̂ n^50MV an<̂ consequently
RBE = 0 ^ 2 = k2/k 1 (4)
The statistical t-test for paired observations was applied for test of significance.
Experiments were also carried out with the purpose of verifying that complete repair of sublethal damage was completed after 4 hours. V-79 cells were exposed to two fractions of 60(2o radiation and incubated at 37°C for various time intervals between the two doses. The cells were treated in exactly thè same way as described for the three fraction experiments, except for being irradiated only twice. For practical reasons the experiment could not be designed to allow analysis as paired observations. The results are shown in Fig. 3. In the case where 2 Gy x 2 were delivered, a small increase in surviving fraction is indicated after 0.5 h. However, no statistically significant differences between the surviving fractions are present with intervals between fractions up to 8 h. In order to investigate whether any inhibition of repair was present, a second series of experiments was performed where 4 Gy x 2 were delivered to the cells. The results are shown in Fig. 3. In this case a statistically significant increase of the surviving fraction is present, as an expression of repair of sublethal damage. The repair process is virtually complete after 2 h. Obviously, four hours seems to allow for complete repair of sublethal damage in this system. Furthermore, the experiments with 2 Gy x 2 indicate that any effects from
24
redistribution of cells within the cell cycle will be of minor importance, in agreement with the findings of McLarty (1976). The inability to demonstrate evidence of repair after a fraction of 2 Gy is most likely due to the increase in surviving fraction being small compared to the experimental errors.
0 0.8 I— 0.7 ̂ 0.6
Î o ,g 0.4
>> 0.3 DCz>U)
0.2
0.1 -
H
0 0.5 1 3 4 5 ......... 6... 7 £TIME B E T W E E N FRACTIONS ( h o u r s )
Figure 3. Surviving fraction as a function of time intervals between fractions for cells irradiated with 2 Gy x 2 (□) and 4 Gy x 2 (■). Symbols represent the mean values of 9 individual experiments. Error bars represent the 95% confidence intervals.
Jejunal crypt assay (ID)
Five different dose levels were used in order to acquire a survival curve for each of the two x-ray energies. Since small differences in survival between the two qualities were expected, special attention was given to reducing inter-individual variations. All animals were irradiated within 3 hours, in order to avoid the influence of diurnal variations in radiosensitivity. At all dose levels, 8 mice were irradiated at each quality. 8 unirradiated mice were used as a control in order to assess the surviving fraction.
25
In the dose domain investigated (9.5 - 15.5 Gy) the data were analyzed by a double log- linear model which is an approximation often advocated (Gilbert 1974, Moore et al. 1983, Potten and Hendry 1985, Thames and Hendry 1987). The model has the form
ln(-lnP) = ln N -D /D 0 (5)
P denotes the probability of the crypt being sterilized (i.e. 1 - crypt surviving fraction). The expression -In P is considered to be proportional to the number of surviving clonogenic cells per crypt. D is the absorbed dose, D q is the absorbed dose that is required to decrease the number of clonogens per crypt to 0.37 of its previous value (i.e. it describes the steepness of the curve), and N is the intercept, proportional to the initial number of clonogens per crypt. The parameters were estimated by the ordinary least squares method. The statistical t-test was applied for the tests of significance. To calculate the confidence intervals of the parameters a normal distribution of the experimental errors was assumed. The confidence intervals of the RBE were transformed from the confidence intervals of the model parameters.
RESULTS AND COMMENTS
RBE of high energy x-rays and electrons (1-ffi)
The RBE for 50 MV x-rays was found to be significantly greater than unity, when using 4 MV x-rays as reference (I). The RBE at surviving fraction (S) 0.1 was estimated to 1.12 with 95% confidence interval (Cl) ±0.08 and 1.10 with 95% Cl ± 0.07 at S 0.01, for oxic cells (I). The corresponding values for the anoxic cases were 1.11 ±0.09 at S 0.1 and 1.10 ±0.06 at S 0.01. There is a tendency toward an increase in RBE at lower doses. This increase is not statistically significant. The results suggest that a phenomenon related to the radiation energy could be responsible for the higher RBE of 50 MV x-rays compared to 4 MV x-rays. The difference between the LET-distributions for 4 and 50 MV are moderate, if just the electron fluences in the irradiated phantom are considered. However, the photonuclear reactions ( y,p; y ,n; y ,a; y,2n; y,np) fluences make a small contribution to the absorbed dose. In tissue such reactions are mainly due to photon interactions with oxygen, carbon and nitrogen. The threshold energies for the (y ,n) reaction with these elements are between 10.6-18.7 MeV. The giant resonance peaks for the interactions occur at photon energies between 20-25 MeV (Bülow and Forkman 1974, NCRP 1984). It should therefore be expected that such reactions are of minor importance for x-ray qualities below 20 MV (i.e. having a mean photon energy much
26
lower than 20 MeV). For electrons, the dose contribution due to nuclear reactions is very small (SCRAD 1966, Laughlin et al. 1979). In order to investigate whether photonuclear reactions offer a plausible explanation to the higher RBE of 50 MV x-rays, a new set of experiments were set up in which 50 and 20 MV x-rays and 50 MeV electrons were used (H). 4 MV x-rays were chosen as reference quality. The results confirm those of (I) for 50 MV x-rays. However, the results from 20 MV x-rays and 50 MeV electrons showed that the RBE for those qualities did not differ significantly from unity (II). The estimates of RBE are given in Table 1.
Table IEstimates of RBE for single fraction experiments, in vitro, with 4 MV x-rays as reference (II). (The figures in brackets represent the 95 % Cl).
Quality RBE at surviving RBE at survivingfraction 0.1 fraction 0.01
50 MV x-rays 1.14 (± 0.07) 1.12 (± 0.05)20 MV x-rays 0.99 (± 0.07) 1.00 (± 0.05)50 MeV electrons 1.03 (± 0.08) 1.02 (± 0.07)
These results support the hypothesis that a high-LET component is responsible for the higher RBE of 50 MV x-rays. In this experiment there also is a tendency for an increasing RBE at lower doses, which is consistent with a high-LET component.
The OER was 2.83 with 95 % Cl ± 0.02 for 4 MV x-rays at surviving fraction 0.01. The corresponding result for 50 MV x-rays was 2.81 ± 0.03. This difference is not statistically significant. A decreased OER is expected if high-LET is involved (Hall 1978). The expected decrease in OER is, however, hard to predict quantitatively. There are experimental data from some studies on OER for mixed low- and high-LET beams (Railton et al. 1974). The fitted data of Railton et al. for different compositions of the mixed gamma-ray and neutron beam can provide a crude estimate of the expected decrease in OER in this case. Analyzing their data this way, predicts an approximate decrease in OER from 2.72 to 2.66 for 1 % neutrons and to 2.42 for 5 % neutron contribution. The predicted decrease in OER for 1 % neutron contribution is small and of the same order as the estimate in the present study.
27
Several investigators have made both actual measurements, and theoretical calculations based on available cross sections for photonuclear processes in different atoms, of this high-LET component, i.e. the total dose arising from photonuclear reactions, of high energy photon radiation (Horsley et al. 1953, Frost and Michel 1964, Allen and Chaudhri 1982, 1988). Allen and Chaudhri (1982) calculated the dose contribution from neutrons, protons and a-particles produced by photonuclear processes in tissue for 24 MV x-rays. The dose contribution of all these particles together was calculated to be 0.1% of the photon dose. In another later study (Allen and Chaudri 1988), the photoneutron yield from tissue-equivalent material irradiated with 28 MV x-rays from a betatron was measured. The authors found that the measured results were in good agreement with the calculated results. The total dose contribution from all photonuclear processes for 30 MV x-rays was calculated to be 0.22% of the photon dose. A fourfold increase of neutron dose at 50 MV x-rays compared to 30 MV has been calculated (Laughlin et al. 1979). This estimate of the size of the high-LET component means that a dose contribution of «0.9% would be expected for the 50 MV x-rays in this investigation. However, deviations from this value might follow from different energy distributions of the x-rays from different accelerators.
Most authors conclude that in radiotherapy, no significant biological effect should be expected from photonuclear processes. However, for tissues irradiated with 24 MV x- rays, protons, neutrons and a-particles contribute 69%, 24% and 7%, respectively of the additional dose (Biilow and Forkman 1974, Allen and Chaudhri 1982). Recoil nuclei also contribute to the absorbed dose (Horsley et al. 1953). With present knowledge, it is not possible to predict a common RBE for this mixture of particles. Consequently, biological experiments are needed to investigate the effect of photonuclear processes.
It is also known that a synergistic effect' occurs when cells are irradiated sequentially with a small dose of a high-LET radiation and a dose of x-rays (Durand and Olive 1976, Hornsey et al. 1977, Ngo et al. 1977, Ngo et al. 1981, Bird et al. 1983, McNally et al. 1984, McNally et al. 1988). This synergism is reported to be even more pronounced when the two radiation qualities are used simultaneously (Higgins et al. 1983).
For the race-track microtron the neutron contamination of the beam is very small (Gudowska 1985) as flattening filters are not used. However, the dose contribution from in-phantom photonuclear reactions might be of biological significance. In this experimental situation (1,11), photonuclear reactions will be produced mainly in the Lucite, in water or in the cells. The main elements for production of photonuclear processes are oxygen and carbon. In soft tissues nitrogen may also contribute. The
28
photonuclear cross-sections for these elements show only small differences (NCRP 1984). Therefore, the situation in this experiment should simulate the situation in soft tissues reasonably well (II). An aluminium chamber was used partly because of the similarities between the cross sections for photonuclear reactions and in the relevant elements (O, C, N). However, slightly more photonuclear reactions are to be expected in aluminium, and the high-LET contribution might therefore be slightly larger than in tissue. This is only true for the neutron component and not for the other components since the range of these particles is too short to reach the cells in this system (II). From the cross-section data (NCRP 1984) the increase in dose from photonuclear reactions in aluminium compared to soft tissue will not exceed 20%.
In radiotherapy, the RBE at clinically relevant dose levels (i.e. doses in the order of 2 Gy/fraction) is of importance. For that reason, fractionated experiments were performed with 2 Gy given three times to the cells with 4 hours intervals (II). The RBE for 50 MV x-rays was estimated to 1.17 with 95% Cl ±0.11. The results obtained with this method (II) will be independent from any cell survival model and give an estimate of the RBE at 2 Gy. The RBE obtained with this method does have a higher value (although, not statistically significant) than with high single doses, which is consistent with the hypothesis of a high-LET component.
Since a higher RBE will have consequences for radiation therapy, it is important to investigate the validity of the in vitro experiments by using a different biological system. The jejunal crypt survival, in vivo, was studied in mice for this purpose (III). The parameters of the models were significantly different, both with respect to D q and N (p< 0.001 in both cases). The estimated RBE shows a higher RBE for 50 MV x-rays compared to 20 MV. At a crypt surviving fraction of 0.5, the RBE is estimated to be 1.02 with 95% Cl ±0.14. For higher doses the RBE was significantly greater than unity. RBE was 1.06 (±0.02) at a crypt surviving fraction of 0.1 and the corresponding result was 1.10 (±0.04) at a crypt surviving fraction of 0.01. The estimates of RBE in this experiment have a tendency to increase as the dose increases, ranging from 1.02-1.10 for the surviving fraction levels 0.5-0.01. However, when the confidence intervals are taken into account, they do not differ significantly. In most cases, the RBE show a variation with absorbed dose, often pronounced only at low doses. In this experiment, the absorbed doses are considerably higher than in most cell survival studies. For this reason it might be expected that the variation of RBE as a function of dose would be virtually zero, and the variation observed in this case might only reflect experimental variation and possible shortcomings of the survival model used. Since N is proportional to the initial number of clonogens per crypt, according to the assumptions of the model, there
29
is no obvious reason for N to differ between the two treatments. Consequently, one possibility of analyzing the data would be to force the two dose-response curves to the same intercept. However, in this case there would be an obvious risk for introducing an error in the analysis of the data.
The values of RBE are similar in both the in vitro, and in vivo studies. The point estimates of the RBE in the in vivo study seem to be somewhat lower compared to in vitro studies. The differences could be explained by stochastic variations. No conclusions of the differences can be drawn from this material. However, there are some possible explanations for the different RBE values that may be worthy of further discussion. Firstly, a high-LET component could be responsible for the RBE. Different biological systems may give different RBE values (Kellerer and Rossi 1972). Secondly, the experimental irradiation system (i.e. the aluminium irradiation chambers) used for the in vitro studies might have introduced a source of error, as mentioned earlier.
In conclusion, it is known that in a target with a composition comparable to human tissue, high energy x-rays will produce a small high-LET contribution to the absorbed dose. Also, low doses of high-LET radiation can have a fairly high RBE. Furthermore, the synergism between high- and low-LET radiation will give an additional biological effect. It seems reasonable that these effects will explain the RBE of 50 MV x-rays. In clinical radiotherapy an RBE higher than unity should be taken into consideration since even quite small deviations in dose-effect might, together with uncertainties in the radiotherapy procedure, influence the outcome of the treatment (ICRU 1976). Both in vitro and in vivo experiments show an increased RBE when the radiation effect of 50 MV x-rays is compared to 4 or 20 MV. The results strongly indicate that the higher RBE should be considered in the clinical use of this high energy. A reduction of the total dose of 5-10 % would be reasonable. However, it is not possible to deduce a fixed "correction factor" from this material since the RBE may be different from one tissue to another. Furthermore, a higher radiation dose-effect than conventionally used might be acceptable if a decrease in energy imparted follows from careful shielding and advantages from the depth dose distribution for the high photon energy.
RBE and PER of pulsed high dose rate electrons (IV)
The results of Purdie et al. (1978, 1980) and Ling et al. (1984) indicate that ultra high dose rates might give a different biological response, mainly due to a reduced OER. Such effects could be of major interest in clinical radiotherapy since hypoxic cells can be demonstrated in human tumours.
30
However, in the present investigation there was no statistically significant difference between the response to high dose rates (mean dose rate: 3.8 x 102 Gy/s; pulse dose rate 2.7 x 10 ̂ Gy/s) and conventional (mean dose rate: 9.6 x 10'2 Gy/s; pulse dose rate 1.2 x 102 Gy/s), in either the oxic or anoxic cases. Subsequently, there was no significant difference in RBE or OER. The estimates of RBE at survival level 0.01 were 1.03 with 95% Cl ±0.08 in the oxic state, and 1.02 (±0.07) for the anoxic state. OER at the same level of survival were 2.78 (±0.24) for the high dose rate cases and 2.74 (±0.14) for the conventional dose rate. The conventional dose rate in this investigation is higher than in common clinical practice. The purpose was to minimize the influence of recovery from sublethal damage during irradiation.
Although technical considerations prevented the use of single pulses, each pulse used contained a dose of approximately 1.6 Gy which is close to the usual dose per fraction given in clinical radiotherapy. Of course, the negative results of this study do not preclude the existence of dose rate effects at even higher dose rates and doses per pulse. If considerably larger doses per radiation pulse are needed in order to obtain a different biological response, it is unlikely that this effect will be useful in clinical radiotherapy.
Some physico-chemical mechanisms for effects of ultra high dose rates:
- Oxygen is consumed by radiation (Dewey and Boag 1959, 1960, Epp et al. 1968). This is not relevant in radiotherapy since doses in the kGy range would be needed in order to obtain enough hypoxia to be detected in biological materials (Kiefer and Ebert 1970).
- At extremely high dose rates, a greater number of free radicals will be present simultaneously in the irradiated volume. These free radicals may recombine in a way that decreases the radiation effect. This will require pulse doses of a magnitude that is not relevant in radiation therapy (Tallentire and Barber 1978).
These mechanisms are clearly beyond clinical application. Furthermore, biological mechanisms are not likely to be differently affected by ultra high dose rates than by conventional doses since the known repair processes in the cells will act during minutes and hours, and not during the irradiation lasting for parts of a second. Some investigators, however, have reported results that might seem to be in conflict with this (Purdie et al. 1978, Ling et al. 1984). They observed a significant variation in cell survival depending on dose rate. One explanation of this conflict is that the reference dose rate used by Purdie et al. and Ling et al (3.5 x IO'2 Gy/s and 1.6 x IO'2 Gy/s
31
respectively) requires exposure times which are too long to eradicate the effect of repair of sublethal damage during the irradiation. Also, Ling et al. (1985) show a variation of OER with dose rate in the dose rate range 2 x 10"2 - 7.6 x 10'2 Gy/s which could offer an explanation of their results.
In conclusion, the results of this study are in agreement with several earlier investigations (Todd et al. 1968, Nias et al. 1969, Berry and Stedeford 1972, Epp et al. 1972, Michaels 1978). However, small differences in biological response could be overlooked due to experimental variations. In this investigation, a difference in OER or RBE in the order of 10% would have been required to find a statistically significant effect of high dose rates. However, a decrease in OER of about 10% or less would be too small to be of practical value in fractionated radiotherapy (Denekamp and Joiner 1982). From a clinical point of view, there seems to be few arguments for further investigations on effects of ultra high dose rates.
Radiobiological cell survival models (V)
The study has established that in radiobiology, a number of models have been utilized, and empirical studies have verified many effects of them. A given set of experimental data has also shown a good fit to different models (Palcic et al. 1983). As a consequence, questions concerning the methodology for model evaluation and clinical implementation arise. The key question can be formulated; How does one select or develop an effective quantitative model?
The issue is of great clinical relevance and concerns the prediction aspect from the high dose domain ( >4 Gy) to the low (< 2 Gy). In the high dose domain, several models have a good fit to observations, but the predictions of the low dose response may diverge significantly. Portier and Hoel (1983) tried a low dose extrapolation from carcinogenesis data and verified the expectation. Their argument to choose a multistage model was that it was the only one which has any basis in the biological theory of carcinogenesis. The authors also conclude: "The results of this research emphasizes the need for incorporation of the biological theory in model selection". The conclusion is easy to accept, but on the other hand, many theories can be associated with a specific model and many models can be related to a single theory. Simplicity (e.g. in context of the number of explaining mechanisms) is another argument for model choice, which further complicates the issue. Otherwise, it is not possible to provide an unambiguous answer to the question raised. Clinical experience, evaluation of experimental results, and new research are natural elements which interact in a process to come closer to the truth. In
32
this view development of new quantitative models can be characterized as a permanent search process closely related to the development of new theories.
The question can also be answered in connection with the research process. The interaction between theories (explanations), models, and observations can be regarded as the driving force in the research process. That view has been demonstrated through this study. Many models have been tried, some of them have been long-lived, and dominated (e.g. the LQ approach) the growth of knowledge. Obviously, the propensity of model changes is not of the same order as the propensity of theory changes. That implies, in the long run, a risk of conservatism which makes the interpretations of experimental studies inefficient and perhaps also invalid. One of the major reasons appears to be the inability to establish an effective methodology for model evaluation to sort out deficient models and with which users can be guided to the model best suited for their purposes.
The experimental research in radiobiology is dependent on the measuring capability, and thus directed to a high dose domain. Palcic et al. (1983) demand experimental data from all the dose domains. Without measurements in the low dose domain, a model is not valid. They go very far and conclude that statements of low dose effects are not meaningful. This describes the ideal situation of model use, but never possible to satisfy in research. Prediction is always a part of the research process to create new hypotheses and hence new knowledge, at least from a popperian point of view (Popper 1959). In this context, models and theories are considered as preliminary. Model approaches will always be replaced. On the contrary, the research and the clinical experience in radiobiology have shown that such predictions are not only meaningful, but also valuable instruments in the development of new knowledge.
In conclusion: What is expected in future is a closer relationship between theories of cellular mechanisms and the model choice and development. An application of stochastic models and the theory of stochastic processes seem to be very fruitful and will extend the ability to improve new theories. This development will provide a new direction for radiobiological cell survival model formulation in direction to an infinitesimal approach, even if the phenomenological one will still be a useful simplification of very complicated systems. The approach, however, demands a strict problem and theory formulation to be effective.
33
CONCLUSIONS
- The RBE of 50 MV x-rays is significantly greater than unity compared to 4 MV x- rays. No statistically significant difference in OER could be detected.
- Photonuclear processes at high photon energies seem to offer a plausible explanation for the higher RBE of 50 MV x-rays.
- The higher RBE of 50 MV x-rays is present at clinically relevant doses i.e. 2 Gy/fraction.
- Similar results for 50 MV x-rays, are obtained from jejunal crypt assay, in vivo as from in vitro experiments.
- Ultra high dose rates do not have a different biological effect with respect to RBE or OER, when compared to a conventional dose rate.
- A closer relationship between description of cellular mechanisms and the model development is required for more effective analyses of experimental results.
- RBE at doses around 2 Gy can be determined, in vitro, with a reasonable sensitivity by paired experiments with fractionated radiation.
34
ACKNOWLEDGEMENTS
I wish to express my sincere gratitude to:
Professor Bo Littbrand, Head of the Department of Oncology, my tutor, for introducing me to the field of radiobiology, for constructive criticism, and valuable advice during the investigation.
Professor Hans Svensson, Head of the Department of Radiation Physics and Head of Dosimetry Section IAEA, for taking so much interest in this study, and for generously sharing his vast knowledge.
Bengt Johansson, Peter Östbergh, Håkan Nyström, Mikael Karlsson, for stimulating collaboration, and for co-authorship.
Margaretha Bäckström, Kjell Brännström, Cenneth Forsmark, Tor Granqvist, Rolf Hahlin, Annika Holmberg, Yvonne Jonsson, Stellan Sandström, for expert technical assistance in the experimental work.
Karin Gladh, Anna Wernblom, for excellent typing, editing and helping in administrative problems.
Rut Jonsson, Parviz Mothlag, Allan Sandström, Helena Tano, for excellent stand-in performances of technical assistance.
Christina, Hanna and Emil, for being there.
This study was supported by grants from the Swedish Cancer Society and Lion’s Research Foundation, University of Umeå.
35
REFERENCES
Adams GE, Stratford IJ. Hypoxia-mediated nitroheterocyclic drugs in the radio- and chemotherapy of cancer. Int J Radiat Biol 1986; 12: 71-6.
Albright NW, Tobias CA. Extension of the time-independent repair-misrepair model of cell survival to high-LET and multicomponent radiation. In: LeCam L, Olshen RA, eds. Proceedings of the Berkely conference in honor of Jerzy Neyman and Jack Kiefer. Belmont: Wadsworth, 1985; 397-424.
Albright N. A Markov formulation of the repair-misrepair model of cell survival. Radiat Res 1989; 118: 1-20.
Allen PD, Chaudhri MA. The dose contribution due to photonuclear reactions during radiotherapy. Med Phys 1982; 9: 904-6.
Allen PD, Chaudhri MA. Photoneutron production in tissue during high energy bremsstrahlung radiotherapy. Phys Med Biol 1988; 9: 1017-36.
Amols HI, Lageaux B, Cagna D. Radiobiological effectiveness (RBE) of megavoltage x- ray and electron beams in radiotherapy. Radiat Res 1986; 105: 58-67.
Barendsen GW, Koot CJ, van Kersen GR, Bewley DK, Field SB, Parnell CJ. The effect of oxygen on impairment of the proliferative capacity of human cells in culture by ionizing radiations of different LET. Int J Radiat Biol 1966; 10: 317-27.
Berry RJ, Stedeford JBH. Reproductive Survival of mammalian cells after irradiation at ultra-high dose- rates: further observations and their importance for radiotherapy. Br J Radiol 1972; 45: 171-7.
Bird RP, Zaider M, Rossi HH, Hall EJ, Marino SA, Rohrig N. The sequential irradiation of mammalian cells with x-rays and charged particles of high LET. Radiat Res 1983; 93: 444-52.
Bistrovic M, Biscan M, Viculin T. RBE of 20 kV and 70kV x-rays determined for survival ov V-79 cells. Radiother Oncol 1986; 7: 175-80.
36
Brahme A. Dosimetrie precision requirements in radiation therapy. Acta Oncol 1984; 23: 379-91.
Biilow B, Forkman B. Photonunclear cross-sections. In: Handbook on nuclear activation cross-sections. IAEA Technical Report Series No. 156. Vienna: IAEA, 1974: 475-558.
Chadwick KH, Leenhouts HP. A molecular theory of cell survival. Phys Med Biol 1973; 18: 78-87.
Chadwick KH, Leenhouts HP. The molecular theory of radiation biology. Heidelberg: Springer-Verlag, 1981.
Churchill-Davidson I, Sanger C, Thomlinson RH. High-pressure oxygen and radiotherapy. Lancet 1955; 1095-9.
Curtis SB. Lethal and potentially lethal lesions induced by radiation - a unified repair model. Radiat Res 1986; 106: 252-70.
Curtis SB. The lethal and potentially lethal model - a review and recent development. In: Kiefer J, ed. Quantitative mathematical models in radiation biology. Berlin Heidelberg: Springer-Verlag, 1988; 137-46.
Denekamp J, Joiner MC. The potential benefit from a perfect radiosensitizer and its dependence on reoxygenation. Br J Radiol 1982; 55: 657-63.
Dewey DL, Boag JW. Modification of the oxygen effect when bacteria are given large pulses of radiation. Nature 1959; 183: 1450-1.
Dewey DL, Boag JW. Inaktivierung von Bakterien mittels einzelner Elektronenimpulse. Z Naturforsch 1960; 15B(6): 372-4.
Durand RE, Olive PL. Irradiation of multi-cell spheroids with fast neutrons versus x- rays: A qualitative difference in sub-lethal damage repair capacity or kinetics. Int J Radiat Biol 1976; 30: 589-92.
Dutreix A, Dutreix J, Wambersie A. Experiments on relative biological effectiveness of 20 MeV electrons. Ann NY Acad Sci 1969; 161: 272-81.
37
Edsmyr F, Andersson L, Esposti PL, Littbrand B, Nilsson B. Irradiation with multiple small fractions per day in urinary bladder cancer. Radiother Oncol 1985; 4: 197-203.
Epp ER, Weiss H, Santomasso A. The oxygen effect in bacterial cells irradiated with high-intensity pulsed electrons. Radiat Res 1968; 34: 320-5.
Epp ER, Weiss H, Djordjevic B, Santomasso A. The radiosensitivity of cultured mammalian cells exposed to single high intensity pulses of electrons in various concentrations of oxygen. Radiat Res 1972; 52: 324-32.
Fricke H, Hart EJ. Chemical dosimetry. In: Attix FH, Roesch WC, eds. Radiation dosimetry. Instrumentation. Second edition. New York: Academic Press, 1966;2 : 167- 239.
Frost D, Michel L. Über die zusätzliche Dosiskomponente durch Neutronen bei der Therapie mit schnellen Elektronen sowie mit ultraharten Röntgenstrahlen. Strahlenterapie 1964; 124: 321-349.
Gerweck LF, Epp ER, Michaels HB, Ling CC, Peterson EC. Repair of sublethal damage in mammalian cells irradiated at ultrahigh dose rates. Radiat Res 1979; 77: 156- 69.
Gilbert CW. A double minus log transformation of mortality probabilities. Int J Radiat Biol 1974; 25: 633-4.
Goodhead DT. Cellular effects of ultrasoft x-radiation. Implications for mechanisms of radiation cell killing. In: Steel GG, Adams GE, Peckham MJ, eds. The biological basis of radiotherapy. Amsterdam: Elsevier Science Publishers, BV, 1983: 81-92.
Goodhead DT. Saturable repair models of radiation action in mammalian cells. Radiat Res (Suppl 8) 1985; 104: S-58-S-67.
Gudowska I. Neutron fluence and dose equivalent measurements around medical accelerators using foil activation and a 235u fission chamber. Internal report RI 1985 1985-02, Dept, of Radiation physics, Karolinska Institute, Box 60211, S-104 01 Stockholm, Sweden, 1985.
38
Hall EJ, Cavanagh J. The oxygen effect for acute and protracted radiation exposures measured with seedlings of Vicia faba. Br J Radiol 1967; 40: 128-33.
Hall EJ. Radiation dose rate: a factor of importance in radiobiology and radiotherapy. BrJ Radiol 1972; 45:81-97.
Hall EJ. LET and RBE. In: Radiobiology for the radiologist. Second edition. New York: Harper & Row, 1978: 93-110.
Henk JM. Late results of a trial of hyperbaric oxygen in head & neck cancer: a rationale for hypoxic cell sensitizers. Int J Radiat Oncol Biol Phys 1986; 12: 1339-41.
Hettinger G, Bergman S, Österberg S. The relative biological efficiency (RBE) of 30- MeV electrons on haploid yeast. Biophysik 1965; 2: 276-81.
Higgins PD, DeLuca Jr PM, Pearson DW, Gould MN. V79 survival following simultaneous or sequential irradiation by 15 MeV neutrons and 60co photons. Radiat Res 1983; 95: 45-56.
Holthusen H. Beitrage zur biologie der Strahlenwirkung. Pflüger’s Archiv für die Gesamte Physiologie 1921; 187: 1-24.
Hornsey S, Andreozzi U, Warren PR. Sublethal damage in cells of the mouse gut after mixed treatment with x-rays and fast neutrons. Br J Radiol 1977; 50: 513-7.
Horsley RJ, Johns HE, Haslam RNH. Energy absorption in human tissue by nuclear processes with high-energy x-rays. Nucleonics 1953; 11: 28-31.
Huang DP, Mauldon GF, Shen CM, Ho JHC. Relative biological effectiveness (RBE) of 30 MeV electrons at varying depths in tissue. Br J Radiol 1974; 47: 795-9.
Hubbel JH. Photon cross-sections, attenuation coefficients and energy absorption coefficients from 10 keV to 100 GeV. In: NSRDS-NBS 29. Washington DC: National Bureau of Standards, 1969: 64.
International Atomic Energy Agency (IAEA): Absorbed dose determination in photon and electron beams. An international code of practice. IAEA Technical Report Series No. 277. Vienna, 1987.
39
International Commission on Radiation Units and Measurements (ICRU). Report No. 24. Determination of absorbed dose in a patient irradiated by beams of x or gamma rays in radiotherapy procedures. Washington, DC: ICRU, 1976.
International Commission on Radiation Units and Measurements (ICRU). Report No. 30. Quantitative concepts and Dosimetry in Radiobiology, Washington, DC: ICRU, 1979.
International Commission on Radiation Units and Measurements (ICRU). Report No34. The dosimetry of pulsed radiation. Bethesda: ICRU, 1982.
International Commission on Radiation Units and Measurements (ICRU). Report No.35. Radiation dosimetry: Electron beams with energies between 1 and 50 MeV. Bethesda: ICRU, 1984.
Kappos A, Pohlit WA. Cybernetic model for radiation reactions in living cells. Int J Radiat Biol 1972; 22: 51-6.
Kellerer AM, Rossi HH. The theory of dual radiation action. Curr Top Radiat Res Q 1972; 8: 85-158.
Kiefer J, Ebert M. Some theoretical considerations concerning ultra high dose rate experiments. In: Proceedings of the second symposium on microdosimetry. Stresa, Italy. 1969, edited by H.G. Ebert, Luxembourg. Commission of the European Communities, (EUR 4452 d-f-e.), Luxembourg, 1970.
Kim JH, Pinkerton A, Laughlin JS. Studies on the biological effectiveness of high-energy electron beams at different depths in tissue-absorbing material. Radiat Res 1968; 33: 419-25.
Laughlin JS, Reid A, Zeitz L, Ding J. Unwanted neutron contribution to megavoltage x- ray and electron therapy. In: Special Publication No. 554. Washington DC: National Bureau of Standards. 1979: 1-14.
Lea DE. The target theory. In: Actions of radiations on living cells. Second edition, reprinted 1962. Cambridge: Cambridge University Press, 1955; 69-99.
40
Ling CC, Spiro IJ, Stickler R. Dose rate effect between 1 and 10 Gy/min in mammalian cell culture. Br J Radiol 1984; 57: 723-8.
Ling CC, Spiro IJ, Mitchell J, Stickler R. The variation of OER with dose rate. Int J Radiat One Biol Phys 1985; 11: 1367-73.
Littbrand B, Révész L. The effect of oxygen on cellular survival and recovery after radiation. Br J Radiol 1969; 42: 914-24.
Littbrand B. Survival characteristics of mammalian cell lines after single or multiple exposures to Roentgen radiation under oxic or anoxic conditions. Acta Oncol 1970; 9: 257-81.
McLarty JW. Mathematical modeling methods in radiobiology. Ph.D. Thesis, University of Texas at Houston Graduate School of Biomedical Sciences 1976. Referred in: Thames HD, Hendry JH. Radiobiological guide for the radiobiologist. In: Fractionation in radiotherapy. London: Taylor & Francis. 1987: 170-244.
McNally NJ, de Ronde J. Effect of repeated small doses of radiation on recovery from sub-lethal damage by Chinese hamster cells irradiated in the plateau phase of growth. Int J Radiat Biol 1976; 29: 221-34.
McNally NJ, de Ronde J, Hinchliffe M. The effect of sequential irradiation with x-rays and fast neutrons on the survival of V79 Chinese hamster cells. Int J Radiat Biol 1984; 45: 301-10.
McNally NJ, de Ronde J, Folkard M. Interaction between x-ray and a-particle damage in V79 cells. Int J Radiat Biol 1988; 53: 917-20.
Michaels HB, Epp ER, Ling CC, Peterson EC. Oxygen sensitization of CHO cells at ultrahigh dose rates: Prelude to oxygen diffusion studies. Radiat Res 1978; 76: 510-21.
Moore JV, Hendry JH, Hunter RD. Dose-incidence curves for tumour control and normal tissue injury, in relation to the response of clonogenic cells. Radiother Oncol 1983; 1: 143-57.
41
National Council on Radiation Protection and Measurements (NCRP). Neutron production and transport. In: Report 79, Neutron contamination from medical accelerators. Bethesda: NCRP, 1984: 5-60.
Ngo FQH, Han A, Elkind MM. On repair of sublethal damge in V79 Chinese hamster cells resulting from irradiation with fast neutrons or fast neutrons combined with x-rays. Int J Radiat Biol 1977; 32: 507-11.
Ngo FQH, Blakely EA, Tobias CA. Sequential exposures of mammalian cells to low- and high-LET radiations. Radiat Res 1981; 87: 59-78.
Nias AHW, Swallow AJ, Keene JP, Hodgson BW. Effects of pulses of radiation on the survival of mammalian cells. Br J Radiol 1969; 42: 553.
Nias AHW, Swallow AJ, Keene JP, Hodgson BW. Survival of HeLa cells from 10 nanosecond pulses of electrons. Int J Radiat Biol 1970; 17: 595-8.
Nias AHW, Swallow AJ, Keene JP, Hodgson BW. Absence of a fractionation effect in irradiated HeLa cells. Int J Radiat Biol 1973; 23: 559-69.
Nias AHW. Fractionated radiotherapy. In: An introduction to radiobiology. Chichester: John Wiley & Sons, 1990: 256-76.
The Nordic Association of Clinical Physics (NACP). Procedures in external radiation therapy dosimetry with electron and photon beams with maximum energies between 1 and 50 MeV. Acta Radiol Oncol 1980; 19: 55-79.
Palcic B, Brosing JW, Lam GYK, Skarsgaard LD. The requirement of survival measurements at low doses. In: Proceedings of the Neyman-Kiefer conference in honours of Jerry Neyman and Jack Keifer. New York: Wadsworth Inc. 1983: 57-8.
Popper KR. Logic of scientific discovery. London: Huitchinson, 1959.
Portier C, Hoel D. Low-dose-rate extrapolation using the multistage model. Biometrics 1983; 39: 897-906.
42
Potten CS, Hendry JH. The micro-colony assay in mouse small intestine. In: Potten CS, Hendry JH. eds. Cell clones: Manual of mammalian cell techniques. Edinburgh: Churchill and Livingstone, 1985: 50-60.
Purdie JW, Inhaber ER, Klassen NV. Increased sensitivity of anoxic mammalian cells irradiated with single pulses at very high dose rates. In: Sixth Symposium on Microdosimetry, volume 2, edited by J. Booz and H.G. Ebert. London: Harwood Academic Publishers Ltd. 1978: 1023-32.
Purdie JW, Inhaber ER, Klassen NV. Increased sensitivity of mammalian cells irradiated at high dose rates under oxic conditions. Int J Radiat Biol 1980; 37: 331-5.
Railton R, Porter D, Lawson RC, Hannan WJ. The oxygen enhancement ratio and relative biological effectiveness for combined irradiations of Chinese hamster cells b y . neutrons and gamma-rays. Int J Radiat Biol 1974; 25: 121-7.
Read J. The effect of ionizing radiation on broad bean root. X. The dependence of the x-ray sensitivity on dissolved oxygen. Br J Radiol 1952; 25: 89-99.
Robinson JE, Ervin TL. A study of relative biological effectiveness for 35 MeV electrons with mammalian cells (100%-10% depth-dose levels). Ann NY Acad Sci 1969; 161: 301-9.
Rojas A. Radiosensitization with normobaric oxygen and carbogen. Radiother Oncol 1991; 20(suppl 1): 65-70.
Rojas A, Joiner MC, Collier JM, Kjellén E, Johns H. Carbogen plus nicotinamide in experimental radiotherapy. In: ARCON - A workshop held in the Fowler Scott Library, Gray laboratory, Mount Vernon Hospital, Northwood, UK, 2-3 June 1991.
Sachs RK, Hlatky L, Hahnfeldt P, Chen P. Incorporating dose-rate effects in Markov radiation cell survival. Radiat Res 1990; 124: 216-26.
Schultz RJ, Nath R, Testa JR. The effects of ultra-high dose rates on survival and sublethal repair in Chinese-hamster cells. Int J Radiat Biol 1978; 33: 81-8.
43
Sinclair WK. The relative biological effectiveness of 22 MeVp x-rays, cobalt-60 gamma rays, and 200 kvcp x-rays. VII. Summary of studies for five criteria of effect. Radiat Res 1962; 16: 394-8.
Sinclair WK. Radiobiological dosimetry. In: Attix FH, Tochlin E, eds. Radiation dosimetry. New York: Academic Press, 1969: 617-76.
The Sub-Committee on Radiation Dosimetry (SCRAD). Protocol for the dosimetry of high energy electrons. Phys Med Biol 1966; 11: 505-20.
Tallentire A, Barber DJW. The interrelationship between pulse length and dose-rate within the pulse in the enhancement of anoxic damage in wet Bacillus megaterium spores irradiated with pulses of electrons. In: Sixth Symposium on Microdosimetry, volume 2, edited by J. Booz and H.G. Ebert, London. Harwood Academic Publishers Ltd, London, 1978. 1061-9.
Thames HD, Hendiy JH. Radiobiological guide for radiotherapists. In: Fractionation in radiotherapy. London - New York - Philadelphia: Taylor and Francis, 1987: 170-244.
Thomlinson RH, Gray LH. The histological structure of some human lung cancers and the possible implications for radiotherapy. Br J Cancer 1955; 9: 539-49.
Tobias CA, Blakely EA, Ngo FQH, Yang TCH. The repair-misrepair model and cell survival. In: Meyn R, Withers HR, eds. Radiation Biology in Canncer Research. New York: Raven Press, 1980; 195-230.
Todd P, Winchell HS, Feola JM, Jonès GE. Pulsed high-intensity roentgen rays. Inactivation of human cells cultured in vitro and limitations on usefulness in radiotherapy. Acta Radiol 1968; 7: 22-6.
Urtasun RC, Chapman JD, Band P, Rabin HR, Fryer CG, Sturmwind J. Phase I study of high dose metronidazole: A specific in vivo and in vitro radiosensitizer of hypoxic cells. Radiology 1975; 117: 129-33.
Wambersie A, Dutreix A, Dutreix J, Lellouch J, Moustacchi E, Tubiana M. Efficacité biologique relative d’un faisceau d’électrons de 20 MeV en fonction de la profondeur. Int J Radiat Biol 1966; 10: 261-75.
44
Williams PC, Hendry JH. The RBE of megavoltage photon and electron beams. Br J Radiol 1978; 51: 220.
Withers HR, Elkind MM. Microcolony survival assay for cells of mouse intestinal mucosa exposed to radiation. Int J Radiat Biol 17; 1970: 261-7.
