Chapter 4 – Biophysiology and Clinical Considerations in Radiotherapy
George E. Laramore
The use of ionizing radiation in medicine dates back almost to the very date of its discovery. In 1895, Wilhelm Roentgen discovered x-rays, and 3 years later, Pierre and Marie Curie announced that they had isolated radium from pitchblende. The first documented radiation biology experiment was performed inadvertently at about this time when Antoine Becquerel developed a “burn” on his chest from carrying a vial of radium salt in his vest pocket. It soon became apparent that this newly discovered entity—radiation—had the ability to affect profound biologic change. The public embraced this new agent, and it was touted as a cure for almost every ailment known to humans. The results of these early clinical trials are not well documented, but it is probably safe to assume that most were not very successful. However, the first “cure” of a malignant neoplasm achieved with ionizing radiation was reported in 1899.
During the early 1900s, most clinical radiotherapy was done by surgeons who used it as another form of cautery. Radiation was used in large doses to produce a “tissue slough” and the adverse side effects associated with its early use still color the attitudes many physicians have toward radiotherapy. Used properly, ionizing radiation produces selective modifications of cells through subtle changes introduced into deoxyribonucleic acid (DNA) and other cellular elements. Special training is required to understand these effects and how to best use them in clinical settings. From this need, radiation oncology has emerged as a separate medical specialty.
The capabilities of the radiation oncologist have increased in keeping with advancing technology. Initially, only low-energy x-rays were available, and these were capable of treating only superficial tumors, without causing severe side effects to the intervening healthy tissues. High-energy linear accelerators were then developed for research purposes and soon were used to produce “megavoltage” x-rays for medical use in a few large centers, although the “megavoltage” era in radiotherapy really began with the use of ?-ray beams from 60 Co sources. Now compact linear accelerators are used routinely in radiotherapy departments. Similarly, research into nuclear physics made it possible to produce many artificial radioisotopes that have had application in medicine; the field is no longer restricted to 226 Ra as it was in the past. Also, specialized, high-dose rate brachytherapy devices have been developed, which reduce the duration of the implant and simplify the radiation protection problem. Investigations in new areas such as particle beam radiotherapy, radiation protecting agents, hypoxic cell sensitizers, chemotherapy–radiotherapy combination treatments, and hyperthermia are taking place today and have the potential for changing the field of radiotherapy as much in the future as it has been changed in the past.
The purpose of this chapter is to provide the clinician with an overview of the basic principles of physics and biophysiology that underlie modern radiotherapy. Limitations of space necessitate the presentation of the overall picture only, rather than a detailed chronologic account of the development
of the field. Topics will be covered in a manner that assumes no previous expertise on the part of the reader. The references cited will be representative and illustrative in nature rather than comprehensive.
BASIC OVERVIEW OF PHYSICS
Conventional types of radiation
Radiotherapy is performed most commonly using high-energy photons or “quanta” of electromagnetic radiation. The electromagnetic spectrum is a continuum with radiowaves 10 to 1000 m in length lying at one end and energetic cosmic rays 10-12 cm in length lying at the other end. The ?-rays produced from a 60 Co source are about 1.3 million electron volts (MeV) in energy, which corresponds to a wavelength of 10-10 cm. Energies of 3 to 5 electron volts (eV) are needed to break chemical bonds, and this typically requires photons shorter than 10-4 cm. Microwaves used for heating purposes are less energetic than this and act by exciting bending and rotational modes in molecules (e.g., H2 O).
High-energy photons used in radiotherapy initially interact in matter (i.e., tissue) to produce high-energy electrons by one of three principal processes: photoelectric effect, Compton scattering, or pair production. In the photoelectric effect, a photon excites a tightly bound, inner-shell electron and is completely annihilated. This process scales like Z3 /E3 per gram of material, where Z is the “effective” nuclear charge of the material, and E is the photon energy. This process is most important for photon energies in the range of 10 to 50 kiloelectron volts (keV), which is the range typically used in diagnostic radiology. The higher effective “Z” of bone relative to soft tissue causes it to show up well on diagnostic films.
The Compton effect is most important in the 500 keV to 10 MeV range of photon energies used in therapy. It scales like Z0 per gram of material and decreases in a complex way with increasing energy. Physically, a photon can be thought of as transferring a part of its energy to a loosely bound outer electron and emerging at a lower energy and longer wavelength. Within this energy range, all tissues absorb photons at about the same rate on a gram-for-gram basis. This is important for therapeutic purposes, such as when managing soft-tissue tumors adjacent to bone. On films exposed with megavoltage x-rays, the distinction between bone and soft tissue is lost.
Pair production refers to a high-energy photon being annihilated in the strong electromagnetic field of an atomic nucleus and producing an electron–positron pair. The threshold energy for this process is 1.02 MeV. It scales like Z per gram of material and increases with increasing photon energy. For a 10 MeV photon, this accounts for about 28% of the total absorption cross-section in tissue. Other processes also can take place at higher photon energies.
Once one of these primary processes has occurred, a high-energy electron is produced, which creates secondary ionization events as it travels through tissue. Typically, about 34 eV of energy is lost for each ion pair that is produced. The resulting ionization clusters are relatively isolated on a scale of typical cellular distances. Most of the events involve water molecules in the cell cytoplasm, and their reaction products initiate complex sequences of chemical reactions that generally involve free radicals. The biologic properties of different megavoltage photon beams are equivalent per unit of energy deposited.
Radiation doses are specified in terms of the energy deposited in a unit quantity of material. In the past, the conventional dose unit was the rad, which was equivalent to 100 ergs being deposited per gram of material. More recently, an international commission has agreed that radiation doses should be specified in terms of gray (Gy), which corresponds to 1 joule being deposited per kilogram of material. The older literature will have radiation doses specified in terms of rad, whereas the newer literature will have the doses specified in terms of Gy. Doses in this chapter will be specified in terms of the latter unit. Numerically, doses in rad can be converted to equivalent doses in Gy by dividing by 100 (i.e., 100 rad = 1 Gy).
Typical depth–dose curves for photon beams used in the therapy of head and neck cancers are shown in the upper panel of Figure 4–1 . The plots are for the dose along the central axis for a 10 cm × 10 cm field size. The energy of the beam is specified by the energy to which the incident electron beam is accelerated before impacting the target and actually producing the x-rays. The x-ray beam is a continuum with the maximum energy equal to that of the electron beam. To express that a range of x-ray energies is produced, the term MV is used rather than MeV. Appropriate filtering elements also are used to “harden” and “shape” the beam, but for most practical purposes, at a given source-axis distance (SAD), the beams from given energy linear accelerators are essentially equivalent. The three curves have the same general shape but vary somewhat in specific details. Note that they do not start out at their maximum value, but rather there is a build-up region, which occurs because the initial, high-energy electrons produced by the photon beam are directed primarily in the forward direction. The number of these electrons increases with depth until a distance equal to the average electronic path length is reached. The deposited dose is low at the surface and then increases to a maximum, after which it decreases with depth because of attenuation of the radiation field. The distance of the dose maximum from the surface is referred to as Dmax. It varies from 1.2 cm for the 4 MV (80 cm SAD) beam, to 1.3 cm for the 6 MV (100 cm-SAD) beam, and to 3 cm for the 15 MV (100 cm SAD) beam. The skin and subcutaneous tissues are spared within this build-up region, enabling the delivery of
Figure 4-1 Typical depth–dose curves for megavoltage photon and electron beams commonly used in the therapy of head and neck cancers. The upper panel shows curves for 10 cm × 10 cm fields for a 4-MV (80-cm SAD) linear accelerator (dashed line), a 6-MV (100-cm SAD) linear accelerator (solid line), and a 15-MV (100-cm SAD) linear accelerator (dotted line). The lower panel shows depth–dose curves for 10 cm × 10 cm fields for 6-MeV (dashed line), 12-MeV (solid line), and 20-MeV (dotted line) electron energies.
a higher dose of radiation to a deeper tumor. Higher energy photon beams can be used with even greater values of Dmax, but these have increased usefulness for the more deeply seated tumors of the thorax, abdomen, or pelvis.
Alternatively, the high-energy electron beam produced by the linear accelerator can be used directly in patient treatments. Typical depth–dose curves for various electron energies are shown in the lower panel of Figure 4–1 . Note that these beams typically penetrate a given distance and then fall off rapidly. There is a slight amount of skin sparing for the 6 MeV beam but not for the others. These beams are useful for treating skin cancers, tumors of the buccal mucosa, or even superficial tumors of the oral cavity, provided that appropriate applicator cones are used. Optimal treatment of a given lesion may require some combination of electron and photon beams, and this in turn requires the services of a comprehensive radiation treatment facility. Megavoltage electron beams have the same biologic properties as megavoltage photon beams for an equivalent dose of absorbed radiation.
Figure 4-2 The upper panel shows a typical depth–dose curve for a neutron beam used in therapy. It is for a 10 cm × 10 cm field size and was generated from a 50-MeV p ? Be reaction at 150 cm SAD. The lower panel shows the pure Bragg curve (solid line) for a neon ion beam of energy 425 MeV/amu and the resulting curve (dotted line) when a 4-cm spiral ridge filter (SRF) is used to broaden the beam for therapy. The data in the lower panel are from the BEVALAC facility at the Donner Laboratories.
In the strictest sense, the electron beams used in conventional radiotherapy facilities are a type of “particle” radiation, but this section will be devoted to the heavier charged particles (e.g., protons, a-particles, heavy ions, p-mesons, and fast neutrons) used experimentally at a small number of radiotherapy centers throughout the world. These particles are of special interest because of their different radiobiologic properties or their better depth–dose characteristics, which allow for higher tumor doses without causing a commensurate increase in the dose to the surrounding healthy tissues.
The particle for which there has been the greatest amount of clinical work to date is the fast neutron. A depth–dose curve for a beam from the cyclotron facility at the University of Washington is shown in the upper panel of Figure 4–2 . Note that this is similar in general appearance to the photon beam curves in Figure 4–1 . Fast neutrons are of clinical interest because of their radiobiologic properties, which occur because of the much greater amount of energy they deposit when they go through tissue. Neutrons are neutral particles
and interact with the atomic nuclei, producing “heavy” charged particles such as protons, a-particles, or nuclear fragments that in turn create a dense chain of ionization events as they go through tissue. The distribution of these secondary particles depends on the energy spectrum of the neutron beam, and hence the biologic properties of the beam strongly depend on its energy spectrum. Neutrons used in therapy generally are produced by accelerating charged particles, such as protons or deuterons, and impacting them on a beryllium target. To a first approximation, the beam can be specified by indicating the charged particle that is accelerated, the energy of the particle when it impacts the target, and the distance between the target and the treatment axis (SAD). The curve in Figure 4–2 , for example, is for a 10 cm × 10 cm field for a beam produced by accelerating a stream of protons to 50 MeV and impacting them on a beryllium target of a thickness that absorbs about 50% of the beam energy. It has approximately the same penetration characteristics as the photon beam from a 6 MV linear accelerator. Most often, cyclotrons are used to accelerate the charged particle beams, but special linear accelerators can be used.
Neutrons also are produced using deuterium–tritium (DT) generating tubes that yield a quasimonoenergetic beam of 14 or 15 MeV neutrons. Although the cost of systems using the DT reaction is lower than cyclotron-based systems, their lower neutron output makes them less suitable for therapy. Although once popular, such DT systems now are used for clinical purposes only in a few centers in Europe. Neutrons in the energy range most commonly used in therapy deposit most of their energy via a “knock-on” reaction, whereby a hydrogen nucleus is impacted, producing a recoil proton. This process is more efficient in tissues that contain a greater quantity of hydrogen, such as adiopose or nerve tissue, and is less efficient in bone. Compared with muscle, the absorption can vary by ± 10%. Typically, the recoil fragments produced by therapy neutron beams deposit 50 to 100 times more energy than the electrons created by megavoltage photon beams. The energy deposited by a radiation beam is characterized by its linear energy transfer (LET) spectrum. The primary high-energy electrons produced by megavoltage photons have LETs in the range of 0.2 to 2 keV per micron traversed, whereas the recoil protons produced by fast neutrons have LETs in the range of 20 to 100 keV per micron. It is this difference in LETs that results in the special radiobiologic properties discussed in the next section.
There also is considerable interest in using the charged particle beams directly for therapeutic purposes, which generally requires beams of much higher energy than those used to produce neutrons. The lighter particles, such as protons and a-particles, are of interest because of their extremely favorable depth–dose characteristics. The radiobiologic properties of these beams are similar to those of conventional photon or electron beams. In the United States, proton beam radiotherapy is carried out at the Massachusetts General Hospital using a Harvard University cyclotron and at a dedicated clinical facility at Loma Linda, California. Heavy charged particles combine the favorable depth–dose properties of the proton and a-particle beams with the favorable biologic properties of the neutron beams. Energies are on the order of several hundred MeV per nucleon rather than the few MeV per nucleon for the recoil fragments produced by neutrons. These highly energetic particles do not deposit much energy in tissue until they reach the end of their path, where they are moving slowly. Hence, they do not produce much radiation damage in the intervening tissues.
The lower panel of Figure 4–2 shows a “pure” Bragg peak for a neon beam (solid line) and for its spread form (dotted line). These data are from the BEVALAC facility at the Donner Laboratories in Berkeley, California. Note the high ratio of the energy deposited at the peak compared with that deposited at shallower depths for the unspread beam. The Bragg peak itself is narrow, and so it must either be “scanned” across a tumor while its penetration depth is being varied, or it must be spread out by passing it through appropriate filters. The dotted curve shows the result after the beam is passed through a 4 cm spiral ridge filter (SRF). Note that this lowers the peak-to-plateau ratio of energy deposition, and at the same time, it broadens the trailing edge of the peak. Clearly, both are undesirable for therapeutic purposes, although the dose of radiation deposited along the initial portion of the path is still lower than that deposited across the spread peak, which represents an advantage over the other types of radiation discussed thus far in this chapter. The broadening of the trailing edge of the peak occurs because of fragmentation of the neon nuclei in the filter, and this does not occur with protons or a-particles. Thus, the spread peaks for the latter two particles have somewhat better localization than the curve shown here. A more sophisticated approach is to use true three-dimensional scanning, which changes the particle energy as the beam is swept across the target. Currently, heavy ion radiotherapy is available only at the Heavy Ion Medical Accelerator (HIMAC) facility in Chiba, Japan.
Another type of charged particle that has been used in radiotherapy is the p-meson. The p-meson is a subatomic particle produced by accelerating protons to energies of 400 to 800 MeV and then impacting them into an appropriate target. Magnetic fields are then used to focus the resulting p-mesons into a beam that can be used for therapy. The p-meson is much lighter than the other charged particles discussed in this section, being only 273 times the mass of the electron (the proton, for example, is 1836 times the mass of the electron). Like the other charged particles, it does not lose much energy until it is near the end of its path, resulting in a “Bragg-peak” type of energy deposition curve. When it stops, an atomic nucleus “captures” it and then explodes into massive charged fragments that deposit considerable energy in a very localized region. Neutrons also are produced in this process, and they deposit their energy throughout a somewhat greater volume. The biologic properties of a p-meson
beam are complex because of the large number of processes involved, but in a crude sense, they can be thought of as behaving like a mixture of low-LET and high-LET radiation.
FUNDAMENTALS OF RADIOBIOLOGY
Cell killing by radiation
Within the cell, there are certain key “targets” that must be affected by the radiation before the cell is killed. The nuclear DNA is probably the most critical target, but other elements, such as the nuclear membrane and mitochondria, also may be important. When any form of radiation interacts with the cell material, there is some probability that one or more of the key target areas will be directly affected. This is the “direct” mechanism of action. Conversely, the radiation interaction may be with some other element such as a molecule in the cell’s cytoplasm, and the loss of this molecule may not be critical to the cell’s continued function. The reaction products may be capable of damaging the critical targets, provided that they can diffuse to them and interact before being converted to nontoxic elements by other chemical interactions (for the OH radical produced by the interaction of radiation with H2 O in the cell, the diffusion distance is about 2 nm). This is the “indirect” mechanism of action. All forms of radiation interact by both mechanisms, but because of the smaller amount of energy deposited by low-LET radiation, it primarily interacts through the “indirect” mechanism. High-LET radiation kills a significant fraction of cells via the “direct” mechanism. Comparing the biologic effects of low- and high-LET radiation provides a way of studying the results of these two processes.
Perhaps the simplest biologic experiment imaginable is simply to irradiate a colony of cells with different amounts of a given type of radiation and see how many are alive and able to reproduce afterward. This is done by plating the cells out on a new growth medium and counting the resulting colonies. This assays for a reproductive viability that is the quantity of paramount importance in tumor control. The radiation is given in a single dose, and the cells are plated out immediately.
A plot of the surviving fraction of cells as a function of the radiation dose is shown in Figure 4–3 (Figure Not Available) . By convention, the surviving cell fraction is plotted on a logarithmic scale, and the radiation dose is plotted on a linear scale. This curve is representative of most mammalian cells. Consider the solid curve, which represents the survival data. Note that there are two distinct regions to the curve. There is an initial region for low radiation doses, where the slope of the curve is shallow. In this region, small incremental changes in the amount of radiation are not very effective at increasing the number of cells that are killed. This is called the shoulder region, and its width is characterized by the parameter Dq. It is the distance along the dose axis at a surviving fraction of unity between the abscissa and the point where the extrapolated linear portion of the curve is intersected. It is a measure of the ability of the cells to repair small amounts of radiation damage.
At higher doses of radiation, the curve becomes a straight line on a semilog plot. Its slope is characterized by Do, which is the incremental dose change required to reduce the surviving cell fraction to 1/e of its value. The steeper the slope in this region, the smaller is the value of Do and the more radiosensitive is the cell line. When extrapolated back to a zero radiation dose, it intersects the abscissa at a value N. A curve of this type can be modeled using the equation
where S is the surviving fraction, D is the radiation dose, and N and Do are as indicated in the figure. In target theory, N can be thought of as the number of distinct targets in the cell that should receive one radiation “hit” before the cell is inactivated. Other parameters also can be introduced into the analysis by requiring more than one radiation “hit” to
Figure 4-3 (Figure Not Available) A representative cell survival curve (solid line) for mammalian cells exposed to single doses of radiation. The surviving cell fraction is plotted on a logarithmic scale, and the radiation dose is plotted on a linear scale. Dq characterizes the width of the shoulder region, which in target theory also may be characterized by the extrapolation number, N. Do characterizes the slope of the “straight line” portion of the curve. The dotted line shows the extrapolation of the linear portion of the curve back to the abscissa. The dashed line shows the regeneration of the cell survival curve if, after giving a certain amount of radiation, 6 to 8 hours are allowed to pass before additional radiation is given. (Redrawn from Hall EJ: Radiobiology for the radiologist, Philadelphia, 1994, JB Lippincott.)
inactivate a given target, but such refinements are beyond the scope of this overview. Radiobiologic data also can be analyzed using a linear–quadratic model of the form
where a and ß are simply parameters used to fit the curve over some restricted dose range. Large ß:a ratios correspond to curves with large shoulder regions. There is one final point to note from Figure 4–3 (Figure Not Available) . If 5 Gy are given, resulting in a 10% cell survival, and then 6 to 8 hours pass before giving additional radiation, the shoulder region of the survival curve is regenerated as shown by the dashed curve. During the waiting period, the cells have recovered most of their original ability to recover from small doses of radiation. This is called sublethal damage repair.
The basic features of the cell survival curves can be qualitatively understood in terms of DNA repair processes as outlined in Figure 4–4 . The complementary strands of the helix are represented by the parallel straight lines, and the base pairings between the strands are represented by the open circles and dots that link the lines. In the upper panel, a photon schematically interacts with one strand of the DNA, which could either be via the direct or the indirect mechanism, with the particular nature of the damage event being irrelevant to the present discussion. What is important is that only one strand of the DNA is affected. Most cells contain repair enzymes that can excise the damaged portion and then, using the information on the complementary strand, can resynthesize the damaged portion. This is what is taking place in the shoulder region of the cell survival curve. If small amounts of radiation are given, there is a likelihood that many cells will experience only one damage event that can be repaired in this manner, although when larger amounts of radiation are given, a situation as shown in the lower panel occurs. Now many of the cells experience multiple damage events, and there is increased probability that some cells will have damage to both strands of the DNA. When the cell attempts to repair the radiation damage, a portion of both strands is excised, and a portion of the genetic information is lost. If this information loss occurs in a “silent” region of the DNA, the cell continues to live. If the information loss occurs in a key area of the genome, then the cell ultimately dies. This is the situation that occurs in the straight portion of the cell survival curve.
Relative biologic effectiveness and oxygen enhancement ratio
High-LET radiation deposits so much energy as it goes through the cell that radiation damage events are clustered closely in space and time, which means that if one strand of the DNA is damaged, there is a high probability that the other strand also will be damaged. Thus, the situation as shown in the lower panel of Figure 4–4 occurs, with an increasing portion of the radiation damage being irreparable. As the LET of the radiation is increased, expect to see the shoulder of the cell survival curve decrease (i.e., Dq ? 0) and the slope of the straight portion of the curve become steeper (i.e., Do ? 0). This effect is shown in Figure 4–5 , which shows survival curves for human kidney cells exposed to 250 kVp x-rays, 15 MeV neutrons from a DT generator, and 4 MeV a-particles. The LET of the radiation increases as indicated, and the curves change as expected.
Because the shapes of the cell survival curves shown in
Figure 4-4 Schematic illustration of the interaction of radiation with cellular deoxyribonucleic acid (DNA). In the upper portion, the radiation interacts with one strand of the DNA, and using the appropriate repair enzymes, the cell can excise the damaged portion and resynthesize the affected region using the genetic information on the complementary strand. In the lower portion, the radiation interacts with both strands of the DNA. When the cell attempts to repair the radiation damage, genetic information is lost.
Figure 4-5 Survival curves for cultured human cells exposed to radiation having different linear energy transfers (LETs). The triangles indicate data for 250 kVp x-rays, the open circles indicate data for 15 MeV neutrons from a DT reaction, and the closed circles indicate data for 4 MeV a-particles. Note that with increasing LET, the shoulder on the curve decreases, and the slope of the straight portion increases. (Redrawn from Hall EJ: Radiobiology for the radiologist, Philadelphia, 1994, JB Lippincott; original data from Broerse JJ, Bardensen GW, van Kersen GR: Int J Radiat Biol 13:559, 1967.)
Figure 4–5 differ according to the type of radiation used, it is difficult to define biologically equivalent doses for therapeutic purposes. Consider the neutron and the x-ray curves, for example. If one chooses as an endpoint the amount of radiation required to kill 99% of the cells, this requires about 9.3 Gy of x-rays but only about 4.2 Gy of neutrons. Hence on a physical dose basis, the neutrons are more effective, and a relative biologic effectiveness (RBE) of 9.3/4.2 = 2.2 can be defined. If one chooses as an endpoint the amount of radiation required to kill 50% of the cells, then the respective doses are 2.8 Gy of x-rays and 1.1 Gy of neutrons for an RBE of 2.5. This situation illustrates a general phenomenon: because of the increased shoulder on the cell survival curves for low-LET radiation, the RBE for neutrons and other high-LET radiation increases with lower dose increments. The change is greatest for cell lines that have the largest shoulders on the low-LET curves (e.g., gut, nerve tissue) and is smallest for cell lines having small shoulders (e.g., bone marrow, germ cells). In the early days of neutron radiotherapy, workers did not appreciate the dependence of the RBE on dose size and tissue type, which led to a high incidence of treatment-related complications. These effects now are being considered, and the incidence of complications is much lower.
Previously in this chapter, it was noted that low-LET radiation primarily killed cells through the “indirect” mechanism, which involved the radiation interacting with molecules in the cell cytoplasm. The sequence of chemical reactions that can take place is complex, but at some point, a free radical generally is involved. A free radical is a chemical species that contains an unpaired electron and is highly reactive. Oxygen acts to stabilize the free radicals, thus allowing them to diffuse to the DNA or other target regions where they react chemically to produce damage. An obvious question is how great an oxygen concentration is required. Experiments have been performed on many species of bacteria, yeasts, and mammalian cells; the overall conclusions are summarized in Figure 4–6 , which shows the relative radiosensitivity as a function of the oxygen concentration in Torr (1 Torr = 1 mm Hg). Note that the radiosensitivity does not change much until the oxygen concentration decreases below about 20 Torr, and then it decreases fairly rapidly. At essentially 0 Torr, the cells are 2.5 to 3.0 times less radiosensitive than they are on the flat portion of the curve. Healthy tissues of the body are at oxygen concentrations between that of arterial and venous blood—between 40 to 100 Torr—and so are on the radiosensitive portion of the curve. However, large tumors tend to outgrow their blood supply and develop regions of necrosis surrounded by cells in a very hypoxic state. These tumor cells lie on the radioresistant portion of the curve, and this is thought to be one reason why large tumors are not as well controlled by radiotherapy as small ones.
One way of avoiding this problem is to use a mode of radiotherapy that is not as dependent on the presence of oxygen for cell killing. One possibility is to use high-LET radiation, for which the “direct” mechanism of cell killing is more important. Figure 4–7 shows cell survival curves for human kidney cells irradiated in well-oxygenated (open circles) and hypoxic (closed circles) conditions. If a 90% cell kill is chosen as the endpoint, then for 250 kVp x-rays,it takes 2.5 times as much radiation to kill hypoxic cells as it does when they are well oxygenated. The oxygen enhancement ratio (OER) is 2.5. As the LET of the radiation increases—going to 15 MeV neutrons from a DT reaction and then to 4 MeV a-particles, and finally to 2.5 MeV a-particles—the OER decreases to 1. This shows the effect of the increasing importance of the “direct” mechanism as the LET of the radiation increases. In general, the OER decreases with increasing LET until a value of 1 is reached, for a LET of about 150 keV/micron.
Cell cycle effects
Cycling mammalian cells proliferate by undergoing mitotic divisions. To define terms, take mitosis or M phase as a starting point. After this comes a “resting” phase, G1 , before the cell starts undergoing DNA synthesis. After DNA synthesis (S), there is another “resting” phase, G2 , before the cell again enters mitosis. Although it is well recognized that many chemotherapeutic agents act at specific points along the cell cycle, it is not commonly appreciated that cells vary in their degree of radiosensitivity according to their position in the cell cycle. Synchronously dividing cell populations are needed in experiments that measure this effect.
Figure 4-6 Plot of relative radiosensitivity of cells as a function of the oxygen concentration in Torr. Well-oxygenated cells are 2.5 to 3.0 times more sensitive than their hypoxic counterparts. Oxygen concentrations for room air and 100% O2 at 1 atmosphere of pressure are indicated by the arrows. This curve is schematic and is not meant to represent any particular cell line.
One way of producing such a cell population is to exploit the fact that at the time of mitosis, many cells growing in monolayers attached to the surface of culture containers will take on a spherical shape and become loosely attached to the vessel wall. If the container is subjected to a gentle shaking motion, these cells will become detached and float to the surface of the growth medium where they can be collected. These cells can then be inoculated into a fresh growth medium, wherein they will grow in synchrony through several cell cycles. Radiobiologic experiments can be performed on these cells at different times after “shake-off,” and they can be caught at different points along the cycle.
The result of radiosensitivity measurements for typical mammalian cells is shown in Figure 4–8 . Relative radioresistance is shown along the abscissa as a function of position along the cell cycle. The position of the cells along the cycle is shown at the top of the figure. The cells are radiosensitive early in the M phase but become more resistant toward the end of this phase. They are resistant in the early G1 phase but then become more sensitive in the late G1 and early S phases. They then become sensitive again in the late G2 and M phases. Cell lines vary in the time they require to go through the cycle, but this is mostly caused by different lengths of the G1 phase. The exact mechanisms underlying this change in radiosensitivity are not clear, but it is interesting to note that at the beginning of mitosis, the DNA in the chromosomes aggregates into a discrete state, whereas in the late S phase, the DNA content of the cell has doubled. These points in the cycle correspond, respectively, to the points of maximum and minimum radiosensitivity. Other variations in radiosensitivity may correlate with different amounts of sulfhydryl compounds in the cell. Sulfhydryl compounds act as free radical scavengers and so act to protect the cell from the “indirect” effects of radiation.
Figure 4–9 shows specific cell survival curves for Chinese hamster ovary cells at different points along the cell cycle.   The open symbols are for cells exposed to ?-rays from a 60 Co source, and the closed symbols are for cells exposed to a fast neutron beam. Note that for each form of radiation there is the same type of variation along the cell cycle, but the degree of variation is about a factor of 4 less for the neutron beam. OERs are about the same for different points along the cycle, so this represents an effect apart from this.
Many tumor systems contain an appreciable fraction of cells in a noncycling or Go phase. Radiation damage to cells in this phase cannot be monitored until the cells are recruited back into the cycle and until it can be seen whether they produce viable progeny. Noncycling cells can be produced in the laboratory by allowing them to grow in a medium until some key nutrient is exhausted. Cell proliferation then stops, and if the cells are kept in this suboptimal medium, the number of cells remains constant. Such cells are said to be in the plateau phase of growth and are mostly in the Go phase. These cells can be irradiated and then can either be immediately inoculated into fresh growth medium or can be incubated for a period in the suboptimal medium before the inoculation takes place. Once they are placed in the fresh growth medium, they return to their normal cycling mode, although the cell survival curve varies depending on whether they have been incubated for a time before being placed in the fresh medium.
This effect is shown in Figure 4–10 . The circular data points indicate cells treated with 60 Co radiation, and for a
Figure 4-7 Cell survival curves for human kidney cells irradiated during hypoxic and well-oxygenated conditions for radiation beams having different LET values. The open circles represent the well-oxygenated cells, and the closed circles represent the hypoxic cells. A, 250 kVp x-rays; B, 15 MeV neutrons from a DT generator; C, 4 MeV a-particles; and D, 2.5 MeV a-particles. Values of the oxygen enhancement ratio (OER) are indicated in the respective panels. The OER decreases as the LET increases. (Redrawn from Hall EJ: Radiobiology for the radiologist, Philadelphia, 1994, JB Lippincott; original data from Broerse JJ, Bardensen GW, van Kersen GR: Int J Radiat Biol 13:559, 1967.)
Figure 4-8 Schematic illustration of the variation in the radiosensitivity of mammalian cells with their position along the cell cycle. The abscissa shows relative radioresistance as a function of time after “shake-off.” The relative position along the cell cycle is indicated along the top of the curve. The curve is schematic and not meant to represent any particular cell line.
Figure 4-9 Cell survival curves for synchronously dividing Chinese hamster ovary cells at different points along the cell cycle. The open symbols indicate cells irradiated with 60 Co ?-rays, and the closed symbols indicate cells irradiated with a 50 MeV D?Be neutron beam from the TAMVEC facility. The circles represent cells in late S and early G2 ; the squares represent cells in late G1 ; and the triangles represent cells in mitosis. (From Gass RL and others: Radiat Res 76:283, 1977 and 1978.)
Figure 4-10 Potentially lethal damage repair for Chinese hamster ovary cells irradiated in the plateau phase. The circular data points correspond to cells irradiated with 60 Co photons, and the square data points correspond to cells irradiated with a 50 MeV D?Be neutron beam from the TAMVEC facility. The open symbols indicate cells plated out immediately, and the closed symbols represent cells plated out after an 8-hour delay. Surviving cell fraction is plotted along the abscissa as a function of the radiation dose. (From Gass RL and others: Radiat Res 76:283, 1977 and 1978.)
given dose of radiation, there are more surviving cells after an 8-hour delay than if the cells immediately started cycling. This effect is called potentially lethal damage repair because the effect of the radiation damage depends on what happens to the cell after the irradiation. The dose is only “potentially” but not necessarily lethal to the cell because the cell can repair itself before reentering the mitotic cycle where it is expressed. The square data points are for cells irradiated with 50 MeV D?Be neutrons. For high-LET radiation, potentially lethal damage cannot be repaired (or can be repaired only to a limited extent), a fact that may be important in certain clinical settings.
THERAPEUTIC WINDOW CONCEPT
Dose–response curves for tumor control and normal tissue damage are sigmoidal in shape. Whether radiation can safely control a given tumor depends on the relative positions of these two curves. Dose–response curves for a “radiosensitive” tumor are shown in Figure 4–11 . Here, giving a therapeutic dose of radiation results in a 95% probability of tumor control and only a 5% probability of normal tissue complication. There is a large gap between the two curves—that is, there is a wide “therapeutic window.” This should be contrasted with the situation shown in Figure 4–12 for a “radio-resistant” tumor. In this situation, a dose of radiation that would result in a 95% probability of tumor control would result in an unacceptably high probability of normal tissue damage. Giving doses that are within the limits of normal tissue tolerance would yield only a low likelihood of tumor control, and the separation between the two curves is narrow. Clearly, the concept of a therapeutic window depends on the radiobiologic properties of the tumor and the healthy tissue in the irradiated volume.
In general, local control of tumors can be improved by better dose localization, which means moving higher on the tumor-response curve without moving higher on the normal
Figure 4-11 Dose–response curves for tumor control (solid line) and for healthy tissue damage (dashed line) for a “radiosensitive” tumor. This corresponds to a wide “therapeutic window,” in that doses that yield a high probability of tumor control have a low probability of causing healthy tissue damage.
Figure 4-12 Dose–response curves for tumor control (solid line) and for healthy tissue damage (dashed line) for a “radioresistant” tumor. This corresponds to a narrow “therapeutic window,” in that doses that yield a high probability of tumor control have a high probability of causing healthy tissue damage.
tissue complication curve, or by exploiting some intrinsic difference in the properties of the tumor and normal tissues, which effectively widens the gap between the two curves. Three-dimensional treatment planning and delivery, brachytherapy, intraoperative radiotherapy, and the use of charged particle radiation are examples of the former approach; the use of high-LET radiation, altered fractionation schedules, radiosensitization agents, and radioprotective agents are examples of the latter.
The intent of clinical radiotherapy is to sterilize tumors and at the same time to avoid untoward damage to the healthy tissues in the treatment volume. To accomplish this goal, fractionated schemes of delivering radiotherapy have evolved over time. The tumor and the healthy tissue consist of heterogeneous populations in regard to the position of the cells in the cycle. In addition, the tumor may have an appreciable fraction of its cells in a hypoxic state. Figure 4–13 shows what happens when such a mixture of cells is irradiated with equal-dose fractions of magnitude D. The
Figure 4-13 Illustration of the effects of fractionated radiotherapy on a heterogeneous cell population. Surviving cell fraction is plotted along the abscissa as a function of the radiation dose. The dose is given in increments, D¯, with the time interval between successive doses being long enough to allow for sublethal damage repair. The initial dose increment kills a greater fraction of well-oxygenated cells than it does their hypoxic counterparts. It also preferentially kills those cells in the radiosensitive phases of the cell cycle. The solid curve indicates when the remaining cells reoxygenate and redistribute along the cell cycle before the next radiation dose is given. The dashed curve indicates when there is no reoxygenation or redistribution, and successive radiation doses are delivered to a more radioresistant cell population. The figure is schematic and is not meant to represent any particular cell line.
first dose increment preferentially kills the cells that are well oxygenated and are in radiosensitive portions of the cell cycle. If several hours pass before delivering the next dose increment, during this period, there is repair of sublethal damage. With the killing of a substantial number of cells, there is less competition for the available oxygen, hence some of the formerly hypoxic cells can reoxygenate. Also, some of the cells can proceed along the cell cycle and thus be in a more radiosensitive phase when the next dose of radiation is delivered. Assuming that both effects occur, the result is the solid curve shown in Figure 4–13 . If there is no reoxygenation or redistribution throughout the cell cycle, then the result is the dotted curve, which shows less cell kill because the remaining cells are in a radioresistant state. These are not the only effects: there is continued cell division and regrowth during the time interval between radiation fractions. These tumor repopulation kinetics have not been considered in Figure 4–13 . To maximize the cell kill, it is important that the size of the dose fractions be greater than Dq —the width of the shoulder region of the single fraction cell survival curve.
These effects are known as the four Rs of radiotherapy: (1) repair (of sublethal damage), (2) redistribution (across the cell cycle), (3) repopulation, and (4) reoxygenation. Fractionated radiotherapy has evolved to exploit the differences in these effects between tumors and healthy tissues. With few exceptions, radiotherapy works not because tumors are intrinsically more radiosensitive than normal tissue (i.e., a smaller value of Do , but because normal tissues are better at repair and repopulation.
Time–dose considerations are important in estimating the effect of a given total radiation dose. If the dose were given in a single fraction, then the healthy tissues would experience more cell killing than if it were given in a fractionated manner. This difference occurs because single fractions allow no opportunity for sublethal damage repair. In general, smaller total radiation doses given over shorter total treatment times produce the same normal tissue effects as larger total radiation doses given over longer time intervals. The classic measurements that illustrate this point are the isoeffect measurements on skin that were made by Strandquist. He showed that the isoeffect lines for various degrees of skin damage and for curing skin cancer were straight when plotted on a log-log scale of total dose versus time. Moreover, the lines appeared to have the same slope (i.e., were parallel). The required dose to produce a given effect was proportional to time to the 0.33 power. Ellis extended this concept to clinical radiotherapy by allocating a portion of the exponent 0.33 to the overall treatment time, T, and a portion to the number of fractions, N. He defined the nominal standard dose (NSD) by
where Dt is the total radiation dose. The exponents in this expression are for skin and no doubt vary for other tissues.
The linear–quadratic model discussed previously in this chapter provides another way of comparing the biologic effectiveness of different radiation schedules. Assuming that there are “n” separated doses of radiation of magnitude, “D,” the cumulative biologic effect of the treatments can be given by
where Dt is the total dose of radiation. Dividing through by a the following is obtained
where E/a is the biologically effective dose. For purposes of comparing radiation schedules, a/ß = 3 can be used for late-responding tissues, and a/ß = 10 can be used for early-responding tissues (i.e., acute effects). It is also possible to modify this expression to crudely account for tumor proliferation during the radiation course.
Altered fractionation schedules
The highly fractionated radiotherapy schemes used today are the result of many years of clinical experience, but radiobiologic considerations may provide guidance for their future improvement. For example, acute radiation side effects such as mucositis and pharyngeal edema are caused by changes in tissues that are composed of rapidly proliferating cells. Late effects, such as subcutaneous fibrosis, vascular damage, radiation necrosis, and spinal cord injury, are caused by changes in tissues composed of more slowly proliferating cells. Radiobiologic measurements indicate that for low-LET radiation, the tissues experiencing late effects are characterized by cell survival curves having large shoulders. It is the late effects that ultimately limit the total dose that can be delivered in the treatment of head and neck cancer. Hence, a logical approach would be to give smaller radiation treatment fractions so as not to exceed the shoulder on the “late effects” tissue curves and then give a higher total dose, which, it is hoped, would result in greater tumor control. This would effectively widen the therapeutic window. Note that the assumption is implicitly made that the tumor will behave like the rapidly proliferating healthy tissues and thus will not have a large shoulder on its cell survival curve. To avoid too great a prolongation of the overall treatment time and hence allowing tumor repopulation kinetics to dominate, multiple daily fractions should be given. A sufficient time interval (generally =6 hours) should elapse between the multiple daily treatments to allow for adequate repair of sublethal and potentially lethal damage in the healthy tissues.
Hyperfractionation refers to giving multiple daily doses of radiation of such a size that the overall treatment time is about the same as for conventionally fractionated course of once-a-day radiotherapy. Several randomized clinical trials recently have been completed using the hyperfractionation approach. The European Organization for Radiation Therapy in Cancer (EORTC) reported on a trial comparing a “standard” radiation schedule of 2 Gy-fractions, once-a-day treatment to 70 Gy versus a hyperfractionation schedule of 1.15 Gy-fractions given twice daily to 80.5 Gy. A total of 356 patients with oropharyngeal lesions were studied. At the 5-year endpoint, the local control rates were 59% versus 40% (P = 0.02) in favor of the hyperfractionation arm. There was a suggestion of improved survival for the hyperfractionation arm, but this did not achieve statistical significance (P = 0.08). There was no increase in complications on the hyperfractionation arm, which agrees with the basic radiobiologic concepts discussed in a preceding section of this chapter. In the United States, comparative, hyperfractionation studies have been conducted that indicate the potential for improved local control for more advanced head and neck tumors. The Radiation Therapy Oncology Group (RTOG) has conducted a dose-searching study to determine the maximum dose that could safely be given for patients with head and neck cancers. Patients were randomized to receive either 67.2, 72, 76.8, or 81.6 Gy at 1.2 Gy given twice daily. A preliminary analysis based on 479 patients suggested an improvement in local tumor control at 2 years with increasing radiation doses for the lowest three dose arms: 25% versus 37% versus 42% (P = 0.08). No survival differences were noted, and the incidence of major late complications was the same at all three dose levels. Data analysis is still pending for the 81.6 Gy arm. A phase III clinical trial comparing hyperfractionation versus conventional fractionation for head and neck cancers is currently underway.
Accelerated fractionation refers to giving multiple daily doses of such a size that the overall treatment time is shortened relative to that of conventional radiotherapy. This may have a potential advantage for overcoming repopulation effects in rapidly proliferating tumors. Wang has used such a schema in the treatment of advanced head and neck tumors.  He uses 1.6 Gy fractions twice daily, which is too high a total daily dose for patients to tolerate without a planned treatment interruption to allow for repopulation and recovery of the mucosa. No randomized trials have been conducted using this schema, but a comparison with historical controls indicates a possible benefit.
One of the more extreme accelerated schedules is the continuous hyperfractionated accelerated radiotherapy treatment (CHART) regimen. This regimen consists of giving three daily radiation treatments of 1.5 Gy each to a total dose of 54 Gy without giving any weekend breaks. As might be expected, acute radiation reactions have been severe, but of more concern was the fact that there were two incidences of cervical myelitis. A comparative analysis with similar patient groups seems to show an improvement in local control, but as yet there have been no randomized studies with this regimen.
Another version of accelerated radiation that attempts to
limit the healthy tissue acute reactions is the concomitant “boost” regimen proposed by Ang and others, which delivers the accelerated portion of the radiation only during the last phase of treatment. In this approach, the volume of tissue receiving the twice-daily treatments is limited to the primary target volume, and no breaks in treatment are given. There is a further theoretic advantage in that the accelerated portion of the radiation is given at a time when the proliferation rate has been increased for the tumor and the healthy tissues. The RTOG is currently carrying out a randomized trial using this approach as one arm of the study.
The altered fractionation approaches discussed previously have their rationale in the basic radiobiology of tumor and healthy tissue response. They all incorporate at least a 6-hour interval between sequential radiation treatments to allow for repair of sublethal and potentially lethal damage in the irradiated healthy tissues. Other types of “hybrid” schemes have been reported in the context of phase I trials involving small patient numbers. Although conceptually attractive, nonstandard radiation schemes have inherent toxicities and should be used with caution in nonprotocol settings. Late morbidity and efficacy data are still accumulating.
Many radioactive isotopes are used in modern radiotherapy practice. Although radium needles are still used as implants in certain head and neck tumors, the trend now is toward afterloading techniques using 192 Ir sources. These sources produce a lower-energy ?-ray, thus simplifying the radiation protection requirements associated with routine patient care. These sources are left in place for a specified time and then are removed. Alternatively, permanent implants using 198 Au and 125 I can be used. These implants deliver their total radiation dose over the effective lifetime of the radioactive material.
One obvious advantage to using implants for a portion of the planned radiotherapy is better dose localization, which results in less radiation damage to the healthy tissue surrounding the tumor. Another advantage is the relatively prolonged time over which the radiation is delivered. External beam radiation is given at the rate of 1.5 to 2.0 Gy per minute. A typical 192 Ir implant delivers its dose at the rate of 0.4 to 0.8 Gy per hour. This can be thought of as “continuous” fractionation, and it allows for healthy tissue repair and reoxygenation of the tumor throughout the time course of the implant. A typical 125 I implant delivers its dose at an even slower rate. Often high total doses in the range of 100 to 200 Gy are given, but one half of the total dose is given during the first 60-day half-life, one fourth of the total dose is given during the next 60-day half-life, and so on. The actual radiobiology of such extremely low dose rates is somewhat uncertain.
More recently, high-dose rate remote afterloading devices have been developed. These devices push a single, high-activity 192 Ir source through a set of interstitial catheters,and a computer program controls the source dwell time at various points throughout the implant. Typically, about 3.0 to 3.5 Gy is given to a distance of about 1 cm from the periphery of the catheters each treatment, and two daily treatments are given about 6 hours apart. Each treatment takes about 15 to 30 minutes, depending on the strength of the radioactive source and the complexity of the implant. There are approximate guidelines to determine how a radiation dose delivered in this manner corresponds to the more familiar doses delivered via low-dose rate implants,  but long-term late effects data are still being accrued. Because these treatments are given in a shielded area in the radiation oncology department, no radioactive material is left in the catheters when patients return to their room, and the radiation protection problem is greatly reduced.
During the past several decades, there has been increasing interest in Japan and in the United States in radiotherapy directly administered to the exposed tumor bed at the time of surgery. Intraoperative radiotherapy (IORT) is given as a single, large fraction using either orthovoltage x-rays or megavoltage electrons. In this approach, it is often possible to move critical structures outside the radiation fields, and the surgeon can aid in identifying the areas at highest risk for residual tumor. A few institutions have dedicated equipment in operating rooms, but the majority of facilities offering IORT transport patients from the operating room to a sterilized unit in the radiation oncology center where the radiation actually is delivered.
Because the biologic effectiveness of a single large dose of radiation is much greater than if the same amount of radiation was given in multiple increments, the total dose given intraoperatively should be reduced compared with that given in a course of fractionated radiotherapy. Most of the IORT experience is for tumors of the abdomen and pelvis, but some general guidelines can be given regarding the tolerance of certain classes of normal structures of importance in the head and neck region. Major blood vessels tolerate single doses in the range of 20 to 25 Gy, whereas damage to peripheral nerves has been noted at doses higher than 20 Gy. Tumor hypoxia may be a greater problem when the radiation dose is given in a single increment because there is no time for reoxygenation to take place. High electron affinic radiation sensitizers, such as misonidazole or SR-2508, may play a role in future IORT study protocols. Similarly, tumor redistribution kinetics do not have time to operate during IORT, and thus tumor cells in radioresistant parts of the cell cycle may be preferentially spared with this technique.
IORT probably can best be used in patients in whom there is a limited number of well-defined sites at high risk for microscopic residual disease. Possible indications are (1) tumor fixation to the carotid artery or deep structures of the neck, (2) “close” margins because of the necessity to preserve vital structures, or (3) tumor extending to bony
structures such as the base of the skull, spinal column, sternum, or clavicle.
High-linear energy transfer radiation
Most clinical data on the use of high-LET radiation in the management of head and neck tumors are for fast neutrons. This will be the topic of this section.
Squamous cell carcinomas
The usefulness of fast neutron radiotherapy in the treatment of squamous cell carcinomas of the head and neck is a subject of considerable controversy. The first reported work dates back to the 1940s when Stone and others conducted a series of clinical studies using an early cyclotron at Berkeley. A total of 249 patients were treated, and about half of these patients had head and neck tumors. Although many dramatic tumor responses were reported, the late complication rate was unacceptably high. Interest in fast neutron radiotherapy waned until the late 1950s when a better understanding of fast neutron radiobiology indicated that most of Stone’s patients had inadvertently received extremely high doses of radiation. Investigation of fast neutron radiotherapy then began at Hammersmith Hospital, and an early report noted dramatic tumor response again, but this time with a more acceptable complication rate. Unfortunately, other trials in Europe and the United States failed to confirm this benefit.    They showed no improvement in either local control at the primary site or in survival rates with neutron radiation, although they seemed to show improved local control for clinically positive neck nodes—45% versus 26%, P = 0.004.  This fact can be qualitatively understood in terms of the basic radiation biology of these tumors. Battermann and others measured the response rates of pulmonary metastases from various tumor histologies using fast neutrons and conventional photon irradiation. They found that the RBE for squamous cell tumors was about the same as for the normal tissue side effects (RBE—3.0 to 3.8), so a therapeutic gain would not necessarily be expected if some other factor such as tumor hypoxia was not a problem and if OER effects would come into play. Guichard and others have demonstrated in animal models that metastatic lymph nodes often have a greater fraction of hypoxic cells than primary tumors of equal size. Measurements of oxygen partial pressure in humans show that hypoxic regions within cervical lymph node metastases constitute approximately 20% of their volume. Hence, it may be that tumor hypoxia in enlarged cervical lymph nodes, and not at the primary tumor site, accounts for the clinical observations reported thus far. In an attempt to resolve this matter, the RTOG undertook yet another randomized trial to study squamous cell tumors of the head and neck. The sophisticated treatment techniques now possible with modern neutron radiotherapy facilities were used, but unfortunately no particular benefit was noted for fast neutron radiotherapy for those with squamous cell tumors.
Tumors that recur after initial radiotherapeutic or after surgical treatment represent another situation wherein high-LET radiotherapy may offer some benefits over conventional radiotherapy. Such recurrences may derive from clones of cells exhibiting a resistance to conventional photon irradiation. Further, the initial treatment may have compromised the vascularity, and the recurrent tumors may have a greater degree of hypoxia than tumors treated de novo. Two nonrandomized clinical trials support this hypothesis. Fermi Laboratories reported an 85% initial response rate, a 45% complete response rate, and an ultimate local control rate of 35% in 20 patients irradiated with neutrons for squamous cell carcinoma recurrent in regions that had received previous photon irradiation. A report from Hammersmith on nine similar patients showed an 89% complete remission rate and a 56% local control rate at 1 year. The rate of major treatment complications was about 25%.
Salivary gland malignancies
Based on the radiobiologic data of Battermann and others, salivary gland tumors exhibit high RBEs for neutron irradiation. They found an RBE of 8 for fractionated neutron radiation of acinic cell carcinoma metastatic to lung, which would indicate a large therapeutic gain factor in using neutrons to treat this tumor system. Phase II clinical trials and a randomized phase III study support this conclusion.
The randomized trial and the historical series are summarized in Table 4–1 .  The data in this table are for patients treated for gross disease—either de novo or for tumor recurrent after surgery. Patients with microscopic residual disease after a surgical resection are not included. Although the number of patients in the randomized trial is small, the difference in the local control rates at 2 years is statistically significant (P = 0.005). The rates of complete tumor clearance in the cervical lymph nodes were six of seven (86%) for the neutron group and one of four (25%) for the photon group. There was an association between improved local control and survival rate at 2 years—62% for the neutron group versus 25% for the photon group (P = 0.1). Given the dramatic differences between the two groups of patients and historical control data that closely paralleled the trial results, it was
TABLE 4-1 — Local control rates for salivary gland tumors treated definitively with radiotherapy
The appropriate references are in the papers by Laramore and Griffin and others. (Laramore GE: Int J Radiat Oncol Biol Phys 13:1421, 1987 and Griffin TW and others: Int J Radiat Oncol Biol Phys 15:1085, 1988; by permission Pergamon Press.)
thought to be unethical to continue the trial further. Ten-year data on this study recently have been published that continue to show improved local and regional control in the neutron group (56% versus 17%, P = 0.009) but no difference in survival rate. The lack of correlation between improved local and regional control and survival rate was a result of distant metastases, which became of greater importance on the neutron arm because of a reduction in deaths caused by local disease. Neutron facilities now consider fast neutron radiation the treatment of choice for patients with either inoperable lesions or with gross residual disease after surgery. Salivary gland tumors constitute a diverse spectrum of histologies, and the fact that the number of patients in the randomized trial is small can certainly be criticized in this respect. Analysis of the historical series seems to indicate that all histologies of salivary gland tumors respond equally well to fast neutron treatments. There also was no apparent difference between major and minor salivary gland tumors. Given the rarity of these tumors and the current opinions of the radiotherapy community, it is unlikely that the randomized trial will be repeated, although data from larger patient series with longer follow-up times will continue to be of interest.
Charged particle radiotherapy
The use of “heavy” charged particles in radiotherapy allows the delivery of high radiation doses to tumors without causing much damage to the healthy intervening tissue. In terms of the curves in Figure 4–12 , this enables work to be done at comparatively low doses on the healthy tissue side effects curve and at high doses on the tumor-response curve. The trailing edge of the Bragg peak for protons and a-particles decreases very rapidly because there are no fragmentation effects. With such beams, it is possible to deliver very high doses to the target volume with millimeter precision. In certain patients, such as those with juxtaspinal cord tumors, some head and neck sarcomas, and cordomas of the clivus, these beams often are the only way of delivering curative doses of radiation without causing life-threatening complications. Local control rates using this approach are excellent.   These beams also are used in the treatment of ocular melanomas, wherein they allow eradication of the tumor and preservation of vision at the same time. A study is currently underway comparing this approach for ocular melanoma with 60 Co plaque therapy.
Hyperthermia and radiotherapy
Hyperthermia refers to the use of elevated temperatures in an attempt to control tumors. In killing cells with heat alone, the temperature to which the tissue is increased and the exposure time at that temperature are the critical factors. There are at least three basic mechanisms that have been proposed in heat-induced cell death: (1) altered membrane permeability, (2) microtubule breakdown, and (3) enhancement of antigen expression or antigen–antibody complexation.
A marked synergy has been shown between hyperthermia and ionizing radiation. Tissue culture experiments show that the cytotoxic effects of these two modalities are additive in the G1 phase of the cell cycle but are synergistic in late S phase. This may be a result of inhibition of DNA repair by heat shock proteins or by alterations of cellular membrane structures important in the repair process. Hyperthermia also seems to inhibit repair of potentially lethal damage in Go phase cells. A low pH renders cells more sensitive to heat, and in tumors, a low pH generally is associated with hypoxic cells. Hence, hyperthermia could potentially help to eradicate the fraction of cells most resistant to conventional photon irradiation.
The most significant impediment to a thorough study of hyperthermia is the inability to deliver and monitor thermal dosages in clinical trials. Methods of delivery include radiofrequency heating, use of microwaves, and ultrasound. In most patients, the resulting temperature profiles are highly inhomogenous, making it difficult to address fundamental issues such as the optimal sequencing of the two modalities. Although the relatively superficial tumors of the head and neck are easier to heat than more deeply seated tumors located elsewhere in the body, interest in the approach is waning because of lack of any documented clinical benefit.
Radiosensitizers and radioprotectors
Radiosensitizers are chemical agents that potentiate the effects of radiation. They should, ideally, be nontoxic in themselves. The basic idea is to increase the effect of the radiation on tumor cells but not on healthy tissue and thus “separate” the two dose–response curves. Hence, these agents should exploit some key differences between the two tissues. The halogenated pyrimidines such as 5-bromodeoxyuridine (BUdR) are preferentially incorporated into the DNA of rapidly proliferating cells in place of thymidine. After their incorporation, the cells are able to repair radiation damage to a lesser degree. The application of these agents for head and neck cancer may be limited because the oral mucosa is a rapidly cycling tissue and also is sensitized. High electron-affinic hypoxic cell sensitizers, such as misonidazole and SR-2508, preferentially sensitize hypoxic cells, which should be more common in tumors than in healthy tissue. Many studies using misonidazole have been done; the results are mixed. A review by Dische showed that misonidazole was beneficial in only five of 33 clinical trials involving various tumor sites.
More recently, several randomized trials using misonidazole have been done. As noted in the preceding section on altered fractionation, the EORTC conducted a trial combining misonidazole with an altered fractionation regimen and found no improvement in either local control or survival rate compared with a course of standard fractionation radiotherapy. A randomized trial was conducted in Denmark to
evaluate the effect of adding misonidazole to two different split-course radiotherapy regimens. A total of 626 patients was entered into the study. There was no difference in overall local control rates with the addition of misonidazole (37% versus 34%), but a subset analysis showed a benefit for the patients with pharyngeal lesions. The preirradiation hemoglobin level also was found to be of prognostic importance. The RTOG performed a trial of 298 patients, evaluating the addition of misonidazole to a “standard” course of radiotherapy.  There were no significant differences in either local control or survival rate, and subset analysis failed to reproduce the results of the Danish group with respect to either pharyngeal primaries or pretreatment hemoglobin levels.
A problem with the use of misonidazole as a hypoxic cell radiosensitizer relates to peripheral neuropathy, which is its principal toxicity. This limits the amount of radiosensitizer that can be used, and it may be that insufficient amounts have been used in the clinical trials reported to date. New agents, such as SR-2508 and Ro-03-8799, are more efficient radiosensitizers than misonidazole, and clinical trials using these agents may be more adequate tests of the radiosensitization concept.
Another approach to widening the therapeutic window is to shift the normal tissue-response curve to the right without changing the position of the tumor-response curve via the use of agents that selectively “protect” the healthy tissues in the radiation field. The radioprotective agent studied most extensively thus far is a thiophosphate derivative of cysteine known as WR-2721. This compound probably protects cells by neutralizing intracellular free radicals before they can interact with the key target areas. Clinical work shows that it protects the bone marrow during hemibody irradiation. It is known that WR-2721 preferentially concentrates in the salivary glands, and thus it may be advantageous in reducing the xerostomia that often is a result of the radiotherapeutic treatment of head and neck cancer. New and more effective agents are being developed.
CHEMOTHERAPY AND RADIOTHERAPY COMBINATIONS
There are two basic intents to the addition of chemotherapy to the treatment regimen for those with head and neck cancer: (1) there is potentially a synergistic effect with radiotherapy by the chemotherapy altering the radiobiologic parameters “a,” “ß,” and the effective tumor doubling time, and (2) the chemotherapy may be effective at eradicating micrometastases, thus reducing the incidence of distant metastases. By far, the most data have been accumulated on the sequential addition of chemotherapy to the regimen. To date, there has been no consistent, overall improvement in local and regional control or survival rate, although there have been several large, randomized studies that have shown a reduction in the incidence of distant metastases even though the basic intent of these studies was different. The Intergroup Study 0034 investigated the effect of adding sequential chemotherapy after surgery and before radiotherapy for patients with operable tumors. The Head and Neck Contracts study compared three arms—one with standard therapy consisting of surgery and postoperative radiotherapy, one with induction chemotherapy before standard therapy, and one arm with induction chemotherapy followed by standard therapy followed by maintenance chemotherapy.  A Southwest Oncology Group study investigated the use of induction chemotherapy before surgery, and the Veterans Administration laryngeal study investigated using the response to induction chemotherapy as a predictor of radioresponsiveness. The Padua, Italy, study compared the effect of four cycles of neoadjuvant chemotherapy plus radiation with radiation alone for patients with inoperable tumors. The common finding in all these studies was a reduction in the overall incidence of distant metastases for the patients on the chemotherapy arm (in the case of the Head and Neck Contracts study, it was only for the group on the maintenance chemotherapy arm). Because distant failure is not the main cause of death for patients with squamous cell tumors of the head and neck, there was, in general, no improvement in overall survival rate. The Padua, Italy, study was the only one of the five that also showed an improvement in local and regional control and survival rate. The data are consistent with some modest efficacy of current chemotherapeutic agents for this class of tumors.
Currently, there is more interest in using chemotherapy concomitantly with radiotherapy, which has greater potential for giving a synergistic rather than an additive effect but also is associated with increased acute toxicity. Large scale, randomized clinical trials using this approach are only now being done. An early success of this approach is Intergroup Study (IG0099) for locally advanced nasopharyngeal cancer that has been stopped early because an interim analysis showed a statistically significant advantage to the experimental arm. In the experimental arm, patients were given concomitant chemotherapy consisting of cisplatinum at 100 mg/m2 every 3 weeks along with radiotherapy followed by four cycles of consolidation chemotherapy with cisplatinum and 5-fluorouracil. In the control arm, patients were treated with standard fractionated radiotherapy. At the time of closure, median progression-free survival rate was 52 months on the experimental arm versus 13 months (P < 0.0001)on the control arm and respective absolute survival rates were “median not yet reached” versus 30 months (P = 0.0007). Nasopharyngeal cancer is unique among head and neck cancers in many respects, and the extension of this approach to other head and neck sites must await the results of other clinical trials.