Choice of optimum megavoltage for accelerators for photon beam treatment

Inl. J. Radiation Oncology Biol. Phys.. Vol. 12, pp. 1551-1557 Printed in the U.S.A. All rights reserved.

Copy-i@

0

0360-3016/86 1986 Pqamon

$3.00 + .OO Journals Ltd.

??Original Contdbution

CHOICE OF OPTIMUM MEGAVOLTAGE FOR ACCELERATORS FOR PHOTON BEAM TREATMENT JOHNS.LAUGHLIN,PH.D.,RADHEMOHAN,PH.D.ANDGERALDJ.KUTCHER,PH.D. Department of Medical Physics, Memorial Sloan-Kettering Cancer Center, 1275 York Ave., New York, NY 10021 Over three decades ago, the development of megavoltage accelerators revolutionized radiation oncology and provided the therapist with photons and electrons of any desired energy. The initial advantages cited for high energy photon therapy, listed below, have proved valid and accelerators have almost totally replaced orthovoltage units. Initially, it appeared that most of these cited advantages should continue to improve with increasing energy, and there has been an impetus for the production of ever higher megavoltage accelerators. Some of these advantages are reviewed in this paper. Also., recent investigations have indicated increasing diffuseness of the photon beam boundary with increasing energy lbecause of lateral transport of electrons. The impact on treatment planning as a function of energy of the increase in volume dose due to the diffuseness of beam boundaries, “build-down” and “rebuild-up” effects in tissues at cavity and inhomogeneity interfaces, bone absorption, and photoneutron production are discussed. Consideration of the behavior of these parameters indicates that optimum photon energies have been achieved and that the impetus for higher megavoltages is unwarranted for most treatment. For many therapeutic applications, there are major advantages of 4 MV to 8 MV photon beams relative to 6oCo gamma rays. For large lesions in the abdomen or pelvis there is an advantage to energies above those provided by 15 MV units. The various considerations above are discussed and summarized as a function of lesion site and megavoltage. Megavoltage radiotherapy, Optimum radiotherapy energy, Treatment plan optimization. INTRODUCIION

1.

Low skin dose and maximum dose below the surface;

2.

Slow fall-off of depth dose because of high penetration of the photons; 3. Less lateral photon scatter relative to orthovoltage photons with increasing photon energy because of the decrease in the mean photon scattering angle; 4. Reduced absorption in bone due to predominance of the Compton process.

This

report examines diflFerent factors that apply to the choice of a preferred photon energy range for accelerators employed in radiation oncology. Although optimum X ray energies for diagnostic purposes were rapidly achieved in the years following R.oentgen’s discovery, radiation therapists had to wait until after World War II before optimum treatment energies with both photons and electrons were achieved. The first instrument to provide megavoltage energy photons was the betatron with which patient treatment was initiated in 1948.22 This was followed within 5 years by t.he linear accelerator,5,” which is now the predominantly used machine for energetic photon treatment. Following a suggestion by W. V. Mayneord, teletherapy units for the use of ‘%o were developed and clinical application was initiated in Canada and in the United States in 1955.13 The physical advantages cited initially for the high energy photon beams produced by these megavoltage accelerators, relative to the orthovoltage radiation, included the following dose distribution properties:

Since the first three of these properties appeared to improve even further with increasing energy, it has been a widely-held belief that radiation treatment machines of increasingly higher photon energies are preferable, and some manufacturers have so responded. Relative to orthovoltage X rays, the advantages of megavoltage X rays have been so dramatic as to revolutionize radiation treatment and to help its establishment as a discipline distinct from radiology. Protocol studies were unnecessary to establish the advantage of megavoltage treatment relative to orthovoltage and few have been carried out. In 1970, KaplanI compared survival experience statistics in several lesions treated with orthovoltage or megavoltage X rays University for supplying us with pelvic dose distributions, and of Mary Martel, Ph.D., Columbia University, for helping on some of the computations. The interest in and critique of this analysis by Zvi Fuks, M.D., is appreciated by the authors. Accepted for publication 24 March 1986.

Presented at the 3rd Ann4 Meeting of ESTRO, September 12, 1984, Jerusalem, Israel. Reprint requests to: John S. Laughlin, Ph.D. Acknowledgments-The authors wish to acknowledge the assistance of Karen Doppke, MS., Massachusetts General Hospital and of Nagalingam Suntharalingam, Ph.D., Jefferson Medical 1551

1552

I. J. Radiation Oncology 0 Biology 0 Physics

and demonstrated decisively the superiority of megavoltage X rays. An instance of a protocol study inside the megavoltage range is the random trial by All? comparing the external application of either 6oCo gamma rays or of 22 MV betatron X rays in advanced cancer of the uterine cervix which, despite some questions, indicated a higher survival in betatron treated cases. More recently, however, the following problems related to very high energy photons have become evident. Contrary to the original perception, it has been demonstrated that the beam boundary becomes more diffuse with increasing energy; “build-down” and “rebuild-up” at interfaces may be quite significant at tissue interfaces: their clinical impact needs further investigation; the decreased bone absorption feature lessens with increasing photon energy; and photoneutrons become an increasingly serious problem as the accelerator energy increases. Review of some cited advantages It is the purpose of this paper to examine the possibility of an optimum photon beam energy range and, accordingly, to review the behavior of some of these cited advantages and disadvantages in their application to radiation therapy as a function of energy. Sharpness of lateral fall oflat the photon beam boundary. The third advantage listed above was based on the decrease in mean scattering angle of photons with increasing energy. However, it has been possible to examine this phenomenon using the Differential Pencil Beam (DPB) dose computation model of Mohan et a1.19This model has been developed with the aid of Monte Carlo methods and incorporates transport of electrons in the dose deposition phenomena. We have employed this method to examine the sharpness of the lateral fall off at the beam boundary diffuseness as a function of energy. Although the decrease in mean photon scattering angle is real, the increasing range of the secondary electrons has the consequence that the sharpness of the beam at its lateral boundaries decreases monotonically with energy over the energy range considered. The shape of the beam boundary occurs as a result of a number of factors including the size of the source, scatter from the collimation system, the shape of the beam flattener, and scattering and transport of electrons and photons in the medium. As the energy increases, photon scatter becomes more and more forward directed. This fact apparently led to the assumption in the past that, if all other factors remained the same, the boundary of the higher energy beam should be increasingly sharp with increasing energy. However, if the transport of electrons knocked out by photons in the medium is also considered, one arrives at a different conclusion. Figure 1 shows portions of beam profiles near the lateral boundary region at a depth of 10 cm. It shows the increase in boundary diffuseness for the bremsstrahlung beam as the accelerator megavoltage is increased from 4 MV to 15 MV, and to 24 MV. These profiles were calculated

September 1986, Volume 12, Number 9

a

9

10

11

12

X (cm) Fig. 1. Computed lateral dose profiles produced by parallel photon beams from accelerators with the indicated peak megavoltages.

using the DPB model. The energy spectra required for these calculations were generated using the ElectronGamma-Shower (EGS) Monte Carlo code.’ In these calculations, a point source of radiation at an infinite distance from the phantom was assumed thereby eliminating the effects of geometry of the source, the beam flattener, and the collimating system. The plots shown represent the consequences of only the physical phenomena, that is, transport of photons and electrons in the medium as a function of photon energy. As the energy of the incident photon increases, the boundary of the beam becomes more diffuse. This is a direct consequence of the lateral transport of electrons ejected by the primary photon, although the scattered photons for high energy incident photons are forward peaked in their angular distribution. Higher dose outside the beam boundary for the 4 MV beam is the result of photon scattering. For realistic beams, the beam flattener causes an excess of energy to be delivered near the beam boundaries resulting in boundaries that are sharper than the theoretical predictions. By manipulating the shape of the flattening filter, sharpness of the beam boundary defined by the collimator can be adjusted up to a point. Sharpening the boundary beyond this point would result in unacceptably high horns. For beams with blocks, however, there is no mechanism built into the treatment machine that delivers the required excess of energy to sharpen the boundaries defined by the blocks. Thus, depending upon the location of the blocks, the boundary of a high energy beam may be observed to be significantly more diffuse than that for a lower energy beam. The data presented in Figure 1 would be realistic for a half beam block. One could, in principle, increase the primary intensity, even for irregular shaped blocked fields, by employing custom-made modifiers. Biggs and Shipley4 have described the effective application of this method to decrease lateral boundary sharpness in individual cases where the foils can be shaped empirically relative to the contours of the field. It is not always convenient or practical to prepare such filters for all of the

Optimum treatment megavoltage 0 J.S. LAUGHLIN etal.

different field sizes and fshapes that may be desired. Although such filters can be effective for certain sites, their general use on a unit would complicate routine quality assurance and calibration procedures. Thus, for a large number of situations, an advantage, previously considered to exist for high energy photon beams, decreases with increasing energy. Consequences of boundary unsharpness are illustrated by the following examp:le. For the 4 MV* data shown, the distance between the beam boundary (taken as 50%) and the point where the dose is 90% of the central axis is 0.3 cm. For the 24 MVP data, the corresponding distance is 0.65 cm. Thus, the area of a beam is a smaller fraction of the width for a high energy beam than for a low energy one. As a result, a larger field around the tumor is required for high energy photons. The consequence for treatment planning is that, if the distribution of dose in three dimensions is employed with attention to the boundary dose unsharpness, the influence of penetrability of high energy photons on the merit of the treatment plans is n.ot as marked as previously believed. Dependence of treatment plan merit on energy. It is difficult to characterize complex treatment plans by quantitative “merit” values and give adequate consideration to the reduction of dose to critical organs and other pertinent factors. The “integral dose” concept introduced originally by Mayneord” has been employed for this purpose for many years. An early instance of this use of integral dose was by Garrison et al. lo They compared the energy absorbed by the body for the same tumor dose for a variety of treatments employing either 400 kV X rays or 23 MV X rays, and the superiority of the higher energy photons was demonstrated quantitatively for five different sites. It is understood that the ratio of integral doses is a relatively simplistic energy ratio and is not a true measure of the merit of a treatment plan; rather, a quantity such as the probability of uncomplicated control is a more meaningful representation. However, the ratio of integral doses is a straightforward way of comparing the physical characteristics of different machine energies and is adequate to indicate the trend of the suitability of different dose distributions as a function of energy. We have computed ratios of integral
* Clinac-4.

1553

6 MV, 10 MV, 15 MV, 18 MV, 25 MV, and 45 MV have been developed. To demonstrate qualitatively the relative importance of beam penumbra and depth dose, three sites were chosen: pituitary, esophagus, and pelvis, each with a progressively larger mean diameter. Only conventional treatment plans, that is, those which would be used in a variety of radiation therapy centers, were used in the analysis. Inhomogeneities such as the lung were ignored since this simplified the calculations and does not affect the conclusions of this comparison. For each technique the beam energy and field size were modified but the beam directions were kept the same, thus permitting only the variables depth dose and penumbra width to be analyzed. Illustrative examples of some of these plans appear in Figures 2(a), (b), and (c) for pituitary (3 field; 6 MV, 18 MV); esophagus (3 field; 6 MV, 10 MV); and for a small whole pelvis (four field; 6 MV, 18 MV). A rigid requirement was that the 95% dose contour (for a normalization of 100% at the isocenter) include the target volume to within 1 mm. The resulting uniformity of the target volume dose was analyzed and indicated that the largest standard deviation in dose was 1.5% except for the Clinac4 which, due to its “horns”, had a maximum standard deviation of 2.3% for large fields. Plots of the ratio of the whole body absorbed energy to that for the target volume are shown in Figures 3(a) through (d). The variation in the integral dose ratio indicates the competing effects of increased diffuseness of beam boundary definition and that of increased photon penetration, both as a function of increasing photon beam energy. In the case of the pituitary (Fig. 2) the best (lowest) ratio is obtained at about 6 MV, with some deterioration above 6 MV. Although, in the case of the esophagus the competition of these two factors results in apparent optima at 6 MV and 15 MV; in both of these cases there is less than 5% variation in the figure of merit between 4 and 18 MV, which is because of the small differences in target volume coverage. Thus, based upon this figure of merit, there is really no gain, for small regions, in going to energies above about 6 MV. However, it should be realized that the distribution of integral dose is different at lower and higher energies. At low energies, the lower isodose levels surround a larger volume, whereas at higher energies the isodose levels between 95 and 70% or so surround a larger volume. That is, for the same integral dose, lower energy accelerators produce tighter dose distributions around the target volume. Consequently, the dose to critical organs near the target will be lower at the expense of somewhat higher dose in the surface regions near the beam entry port. Thus, although the integral dose ratio varies little between 4 and 18 MV, clinical effects could be quite different. In the case of the small pelvis with parallel-opposed fields, substantial improvement occurs above 6oCo gamma-ray energy with only a slight improvement above

t Clinac-2500.

I. J.

RadiationOncology 0 Biology 0 Physics

/

\ Esophagus

Small

Pelws

4- Ftcld

Fig. 2. (a) Three-field treatment of pituitary with wedges using 6 MV photons from a Clinacd and 18 MV photons from a Therac-20. (b) Three-field treatment of the esophagus using 6 MV photons from a Clinac-6and 10 MV photons from a Therac20. No lung corrections. (c) Four-field plan for a small pelvis (AP separation of 20 cm and lateral separation of 3 1 cm) using 6 MV photons from a Clinac-6 and 18 MV photons from a Therac-20.

September 1986, Volume 12, Number 9

does the higher energy radiation lead to a distinct improvement in the dose distributions. Dose variation near inhomogeneities and air cavities. Features of the reduced skin dose and the build-up of electron equilibrium in superficial layers of tissue have been studied. The build-up of electron dose recurs within the patient when inhomogeneities and air cavities are present. This effect is important even at 6oCo energies and has been the subject of analysis by Epp et al,’ Young and Kornelsen, 24 Doppke,6 Epp et al. ’ Nilsson et aL2’and others. Build-up for internal air ca&es has been measured using small ionization chambers and thin TLD wafers as a function of the size of the air cavity, and represents one limiting parameter in the use of high energy photons for radiation treatment. As demonstrated in measurements by Doppke,6 this effect increases in significance with increasing energy. This effect is shown also in Figure 4. This figure contrasts the dose distribution produced by a 15 MV photon beam in a phantom containing an air gap as determined by measurement and by predictions of the EGS Monte Carlo program and a traditional TPR-SPR model. Although the lack of build-up is mitigated with more than one field in some body sites, particularly in the upper respiratory air passages, it does represent a limitation that becomes more significant with increasing energy. “Build-down” and “rebuild-up” effects give rise to cold spots, which are not predicted by conventional dose computation models, but may be quite significant from the clinical point of view. The magnitude of the dose variation and the size of the region of d&equilibrium increases with increasing energy and is most important for small fields and near the boundaries of beams and blocks.

4

;" 2.6 L

f

dose and beam penumbra are more balanced and the slight improvement above 6oCo is gradual. The improvement is more marked with the large pelvis where beam penetration is more important (Fig. 3d). These results indicate that the sharper beam penumbra of lower energy machines (4-6 MV) leads to a tighter distribution of dose around the target volume for small to moderate target volumes and patient dimensions. Only where large target volumes in large patients are considered

Small Pd"lS _ _

12

24 EnerqyiMV)

4

8

12

16

20

Energy(WI

TvoFy 0

20

~ FaurF,e,d i

$&

10 MV. With four fields the competing effects of depth

16

8 12 EnerqyIMVI

36

f ,

~~l~~'"',

*, 48

0

12

24 EnergyMY)

36

40

3. (a) The ratio of whole body to tumor integral dose vs. energy for the 3-field treatment of the pituitary. (b) The ratio of whole body to tumor integral dose vs. energy for the 3-field

Fig.

treatment of the esophagus. (c) Ratio of whole body to tumor integral dose vs. energy for a small pelvis using parallel opposed and four-field techniques. (d) Ratio of whole body to tumor integral dose vs. energy for the large pelvis (AP separation 28.5 cm, lateral separation 44.5 cm) using 4 fields. The points at 25 MV are for a Clinac-35 and at 45 MV for a Brown-Boveri betatron.

Optimum treatment megavoltage 0 J. S. LAUGHLIN et al.

) MV Photbns

Measurements

Gap I

20

Depth (cm)

I : 30

Fig. 4. Dose build down and build up in a polystyrene phantom as measured on the proximal and distal surface of an air cavity. The computed dose by the EGS Monte Carlo program and the traditional TPR/SPR model are also shown. Courtesy of Mohan and Chui.” Lack of increased absorption in bone. It was discovered clinically that the use of orthovoltage X rays in fields containing bone could result. in bone necrosis. The increased absorption in bone at lower photon energies, largely because to the photoelectric effect, had been a major limitation in their use for treatment. The predominance of Compton absorption in the megavoltage range is a definite advantage with respect to irradiation of healthy bone. The dependence of this effect on the photon beam energy can be estimated by integration of bremsstrahlung spectrum over the variation of the mass energy absorption coefficent as a function of energy:

s I(E)

&YE) -

ItEJ s

-tissue

Sbone(E)dE

P /.&Y(E)

dE

P

where I (E) is the intensity (energy X number of photons) spectrum of the accelerator, p/p’s are the mass absorption coefficients and S (E) is the average stopping power ratio of electrons ejected by monoenergetic photons of energy E. Results of this computation using accelerator spectra I(E) generated with the EGS Monte Carlo code, are shown in Table 1. This table represents absorption of dose in tissue imbedded in bone, relative to dose in tissue. It is apparent that this advantage has been fully realized in the lower megavoltages and that there is a gradual increase in bone absorption with increasing accelerator voltage above 6 MV. Neutron exposure. Production of neutrons in nuclear photo-disintegration reactions was recognized as a problem as early as 1948 in the therapeutic application of the betatron.15 Methods were developed then to determine the energy spectrum of tlhe neutrons surrounding the accelerator as well as their contribution of dose to the patient

1555

and to staff. Since then measurements of the neutrons surrounding different medical accelerator installations have been reported extensively in the literature.2*3*17Neutrons are produced by this mode of disintegration reaction in the target as well as in the differential filter and collimator of the accelerator. The fundamental reactions are giant nuclear resonances that have definite thresholds. As the energy of the photon beam is increased, more of the photons are above the pertinent thresholds and the probability of these interactions increases rapidly in the 10 MV to 20 MV accelerator peak voltage range, reaching maximum saturation level by approximately 25 MV. The neutron exposure doubles approximately above 18 MV. Data on neutron production by accelerators were summarized at a conference at the National Bureau of Standards in 1980.i2 Although it has been considered that patient exposure is not greatly affected by neutrons in and near the beam,20,23staff exposure is a major consideration. Accordingly, with photon beams of 18 MV and above, accelerator manufacturers have had the problem of incorporating neutron absorbing material in the head, and the shielding design of the installation has been more complicated to protect staff from the more biologically effective neutrons. In fact, at the low dose rates of the leakage neutrons in the console area, the appropriate quality factor may be greater than the value of 10, which is assumed. This shielding problem can be managed but it represents still another problem which is exacerbated with increasing accelerator energy. DISCUSSION

In summary, the computations and measurements discussed above briefly have indicated that, despite the decrease in photon scattering angle with increasing energy, the increased lateral electron range causes increasing unsharpness of dose distributions at the lateral boundaries of the photon beam. This realization affects the consideration of optimum photon energy. Formulation of treatment plans with rigourous criteria for inclusion of the target volume within the 95% dose contour has been illustrated above. Computation of a physical figure of merit based on the energy absorbed in the tumor and totally in the patient contrasts the influence of the increasing penetration of the beam with increasing unsharpness of the Table 1. Dose in tissue imbedded in bone relative to dose in tissue as a function of accelerator voltage Machine

Nominal energy

Superficial 60~0 Clinac-4 Clinac-6 Clinac- 18 Clinac-20 Clinac-2500

100 KeV 1.25 MeV 4MV 6MV 10 MV 15 MV 24 MV

Dbone/Dtissue 2.71 1.03 1.04 1.05 1.06 1.07 1.10

I. J. Radiation Oncology 0 Biology 0 Physics

1556

Table 2. Importance

of beam characteristics

vs site

Beam characteristic

Site

Depth dose

Brain Head and neck Breast Thorax Lymphoma Pancreas Whole pelvis

+ + + +

Build-up

Beam penumbra

Bone dose

Neutrons

+ + + +

+ + + + +

+ + +

+

+

+

+

+

Cone down Pediatrics

+ +

+

+

beam boundaries, both as a function of increasing photon energy. The choice of an optimum photon energy depends upon a combination of the previously discussed beam characteristics, depth dose, build-up, beam penumbra, bone dose, and neutron levels. In defining an optimum, each of these characteristics should be weighted differently, depending upon the treatment site. The importance of each characteristic as a function of site is shown in a binary way (f when important) in Table 2. It is clear from Table 2 that depth dose is important (although to varying degrees) in essentially all sites located below the neck. Buildup or build-down are important, in breast, head and neck, thorax, and lymphomas since the aim is to keep the skin dose (depth of 0.5 mm) at a minimum, and the target dose (l-5 mm) at a maximum. Also, each of these sites has natural inhomogeneities (air cavities, lung) adjacent to target structures. The beam penumbra is very important for most sites with the exception of the abdomen and whole pelvic fields. The bone doses are important for many sites but particularly for head and neck (mandible), breast (ribs), lymphoma (bone marrow), and pelvis (femoral heads). The neutron levels for patients are especially important for sensitive tissues such as breast and in younger patients with potentially long term survivals. As discussed above, at higher megavoltages neutron levels increase and protection of staff requires increased shielding.

Table 3. Optimum

September 1986. Volume 12. Number 9

With this information optimum energy ranges may be assessed as shown in Table 3. For most sites, megavoltages below 10 MV will probably lead to the best therapeutic ratios whereas higher energies are preferred mainly for abdominal and pelvic cases. For these latter sites the previous calculations demonstrate that lo- 15 MV is adequate for all but the largest patients when four field techniques are used. Although 6oCo is included there is a physical advantage of sharper penumbra in the use of 4-6 MV accelerators. Which energy machine or machines should be purchased by a department also depends upon other issues which include the distribution of treatment sites, the number of treatment units in the facility, and economic factors. For single unit facilities 6 MV is close to the optimum machine energy in that it offers the best compromise between depth dose and beam penumbra. It is important to realize that if a higher energy machine is chosen ( lo- 15 MV) that the build-up characteristics are unsuitable for many sites such as breast, head and neck, and lymphoma. The use of bolus or beam degraders cannot change the build-up to mimic that of a 4-6 MV beam. Thus, more skin reactions or reduced target doses will be obtained. For two unit facilities one machine at 4-6 MV and the other at lo- 18 MV seems appropriate. The choice of optimum units for larger departments is too facility dependent for specific recommendation. However, the use of machines with energies in excess of 18 MV to treat large patients should be considered carefully. For abdominal and pelvic treatments there is only a small gain in the tumor to normal tissue integral dose ratio for small to moderate sized patients when multifield techniques are used. For large patients there is a definite gain in the integral dose ratio with higher energies. The clinical manifestation of this should be in reduced small bowel and other tissue toxicities for the same prescribed target dose. However, these factors must be weighed against a rapid increase in neutron production, larger penumbra widths, and increased bone absorption. In facilities where a high energy machine could be dedicated mainly to large pelvic and abdominal radiation its purchase may be warranted. However, if it is necessary

energy vs site Optimum

Site Brain Head and neck Breast Lung Lymphoma Pancreas Whole pelvis Pelvic cone down Pediatrics

4MV

@+zo

Energy lo-15

6MV

MV

a18 MV

>

< <

> <

> <

9
<

4 >

> > >

Optimum treatment megavoltage 0 J. S. LAUGHLIN ef al.

to treat many other sites with this high energy beam, then an overall reduction in the quality of the dose distributions can be expected (and with it the therapeutic ratio) unless great care is taken in patient selection. Another consid-

1557

eration for the future is that the implementation of multileaf collimation and dynamic therapy, which is equivalent to the use of many fields, may further reduce the need for higher energies.

REFERENCES 1. Allt, W.E.: Supervoltage radiation treatment in advanced cancer of the uterine cervix. Can. Med. Assoc. J. 100: 792797, 1969. 2. Almond, P.R.: Neutron leakage from current machines. In Proceedings of a Confirence on Neutrons from Electron Medical Accelerators, NBS Special Publication 554, Heaton,

H.T. and Jacobson, R. (Eds.). U.S. Gov’t. Printing Office. 1979, pp. 129-138. 3. Axton, E.J., Bardell, A.G.: Neutron production from electron accelerators used fo:r medical purposes. Phys. Med. Biol. 17: 293-298,

1972.

4. B&s, P.J., Shipley, W.V.: A beam width improving device for a 25 MV x ray beam. Int. J. Radiat. Oncol. Biol. Phys.

11: 123, 1985. 5. Day, M.J., Farmer, F.T.: The 4 MeV linear accelerator at Newcastle Upon Tyne. Br. J. Radiol. 31: 669-682, 1958. 6. Doppke, K.P.: Interface effects with 10 MV and 25 MV xrays (Abstr.). Scientific .Exhibit at the 1984 AAPM Annual Meeting, Chicago. 7. Epp, E.R., Lot&reed, M.N., McKay, J.W.: Ionization buildup in upper respiratory air passages during teletherapy with cobalt-60 radiation. Br. J. Radiol. 31: 361-367, 1958. 8. Epp, E.R., Boyer, A.L., Doppke, K.P.: Underdosing of lesions resulting from lack of electronic equilibrium in upper respiratory air cavities Iirradiated by 10 MV x-ray beams. Int. J. Radiat. Oncol. Biol. Phys. 2: 6 13-6 19, 1977. 9. Ford, R.L., Nelson, W.R.: The EGS code. Computer programs for the Monte Carlo simulation of electromagnetic cascade showers. SLAC-210. Stanford University, CA, 1978. 10. Garrison, H., Anderson, J., Laughlin, J.S., Harvey, R.A.: Comparison of dose distributions in patients treated with x-ray beams of widely different energies. Radiology 58: 36 l368, 1952. 11. Haimson, J., Karzmark, C.J.: A new design 6 MeV linear accelerator system for supervoltage therapy. Brit. J. Radiol. 36: 650-659,

1963. 12. Heaton, H.T.: In Proceedings of a Conference on Neutrons from Electron Medical Accelerators, NBS Special Publication

554, Heaton, H.T. and Jacobson, R. (Eds.). U.S. Gov’t. Printing Office, 1979. 13. Johns, H.E.: The physicist in cancer treatment and detection. Int. J. Radiat. Oncol. Biol. Phys. 7: 801-808, 198 1. 14. Kaplan, H.: Radiotherapeutic advances in the treatment of neoplastic disease. Israel J. Med. Sci. 13: 808-8 14, 1977. 15. Laughlin, J.S.: Considerations in the use of a 23 MeV-medical betatron. Nucleonics 8: 5- 16, 195 1. 16. Mayneord, W.V.: Energy absorption. Brit. J. Radiol. 13: 235-247,

1940.

17. McCall, R.C., Swanson, W.P.: Neutron sources and their characteristics. In Proceedings of a Conference on Neutrons from Electron Medical Accelerators, NBS Special Publication 554, Heaton, H.T. and Jacobson, R. (Eds.). U.S. Gov’t. Printing Office. 1979, pp. 75-86. 18. Mohan, R., Chui, C.S.: Validity of the concept of separating primary and scatter dose. Med. Phys. 12: 726-730, 1985. 19. Mohan, R., Chui, C., Lidofsky, L.: Differential pencil beam dose computation model for photons. Med. Phys. 13: 6473, 1986. 20. Nath, R., Epp, E.P., Laughlin, J.S., Swanson, W.P., Bond, V.P.: Neutrons from high-energy x-my medical accelerators: an estimate of risk to the radiotherapy patient. Med. Phys. 11: 231-241, 1984. 2 1. Nilsson, B., Schnell, P.O.: Build-up effects at air cavities measured with thin thermoluminescent dosimeters. Acta Radiol. Ther. Phys. Biol. 15: 427-432, 1976. 22. Quastler, H., Adams, G.D., Almy, G.M., Dancoff, S.M.,

Hanson, A.O., Kerst, D.W., Koch, H.W., Lanzl, L.H., Laughlin, J.S., Riesen, D.E., Robinson, C.S., Austin, V.T., Kerley, T.G., Lanzl, E.F., McClure, G.Y., Thompson, E.A., Skaggs, L.S.: Techniques for application of the betatron to medical therapy. Am. J. Roentgenol. 61: 591-625, 1949. 23. Rawlinson, J.A., Johns, H.E.: Communication to the Editor. Med. Phys. 4: 456-457, 1977. 24. Young, E.J., Kornelsen, R.O.: Dose corrections

for lowdensity tissue inhomogeneities and air channels for IO-MV x-ray. Med. Phys. 10: 450-455, 1983.