0360-3016/90 $3.00 + .I3 Copyright 0 1990 Pelgamon Press plc
Ini J Radmmn Oncology Bml. Phys, Vol. 18, pp. 1521-1527 Printed in the U.S.A. All rights reserved.
??Technical Innovations and Notes
PHYSICAL ASPECTS OF EXTERNAL TREATMENT OF MALIGNANT M. SOUBRA, M.Sc.,*
P. B. DUNSCOMBE,
PH.D.,*
BEAM RADIOTHERAPY FOR THE PLEURAL MESOTHELIOMA D. I. HODSON, M.D.+ AND G. WONG, M.D.+
Manitoba Cancer Treatment and Research Foundation, 100 Olivia Street, Winnipeg Manitoba, R3E OV9 Canada The optimization of radiotherapy for the treatment of malignant mesothelioma highlights many of the currently outstanding problems in clinical radiation physics. The experimental investigation of an intuitively attractive irradiation technique with combined photon and electron beams using a specially constructed phantom has established that, due to the penetration in low density material of both primary electrons and those secondary to photon irradiation, the normal lung tissue is not spared to any significant degree by such a technique. Furthermore, great care needs to be exercised in the treatment planning calculations for this approach if absolute dosimetry errors as large as 50% are to be avoided. Malignant mesothelioma,
Radiotherapy, Treatment planning, Combined photon, Electron beams.
Although pleural mesothelioma is an uncommon malignancy, the increasing incidence and the difficulties associated with therapy represent a therapeutic challenge. The insidious progression of symptoms and the diffuse pathogenesis generally preclude a definitive surgical extirpation and, although prospective comparisons are scarce, it appears that combined modality therapy with surgery and radiotherapy is superior to either modality alone ( 12). The role of chemotherapy remains speculative in spite of extensive studies using a variety of investigational and conventional regimen (1). The bulk of the therapeutic experience confirms the clinical impression that this is a relatively radioresistant malignancy requiring doses for control that are well above accepted pulmonary tolerance levels (12). If radiation therapy is to be used optimally in this disease, then the difficult dosimetric problems associated with the treatment of a “shell” (the pleural surface) need to be addressed. Hilaris and coworkers (5) have suggested a novel and intuitively attractive irradiation technique whereby the pleural surfaces lying predominantly parallel to the radiation beam axis are treated with photons and the anterior and posterior pleural surfaces by electron beams. The intention of such an approach is to limit the radiation dose to the lung which is encapsulated by the target. In less than half the patients treated by Hilaris et al. (5) it
was desirable or possible to augment the dose delivered by external beam radiotherapy with brachytherapy techniques. In this communication we consider some of the physical aspects of implementing mixed external beam radiotherapy of the pleural surface. Such a treatment approach involves many of the currently outstanding problems in clinical radiation physics. These are: abutting photon and electron beams with different penumbras; the dose distribution resulting from photon irradiation in the neighborhood of gross electron density inhomogeneities and the behavior of electron beams in the presence of inhomogeneities. In the absence of appropriate computational tools for tackling these problems our approach has been experimental. A phantom which simulates the essential features of the treated hemithorax albeit with simplified geometry has been constructed. Irradiation of this phantom with photon and electron beams under appropriate clinical conditions has served to elucidate some of the dosimetric aspects of the treatment planning of pleural mesothelioma. We have investigated the effect of photon energy by using beams of both 4 and 25 MV. In addition, the combined photon-electron approach has been compared with the clinically simpler technique of irradiation with photons alone. Since completing the experiments and analysis reported in this communication more details of the original Hilaris technique have been published (7). It is apparent that di-
Reprint requests to: M. Soubra, MSc., Department of Medical Physics, Ottawa Regional Cancer Centre, 190 Melrose Ave., Ottawa, KlY 4K7 Canada. Acknowledgements-The skill of Jules Legal and his colleagues who constructed the phantom used in this study is gratefully
acknowledged. Dr. G. Froese is thanked cussions. * Department of Medical Physics. + Department of Radiation Oncology. Accepted for publication 20 December
INTRODUCTION
1521
for many useful dis-
1989.
1522
1. J. Radiation Oncology 0 Biology 0 Physics
June 1990. Volume 18, Number 6
100% from one electron field
5 cm
100% from two photon fields
Fig. 2. The irradiation geometry. The hatched region at the interface between acrylic (soft tissue) and cedar or balsa (lung) represents the target area. Arrows indicate the normalization points for both photon and electron fields. X
4 or 25 MV
Fig. 1. Photon and electron portals used to irradiate the phantom. For the combined treatment technique the electron field was blocked during photon irradiation.
rect comparison between the present work and that recently reported (7) is precluded by the different electron energies and normalizations in the two studies. The irradiation conditions used here were determined by the design of the phantom and by common clinical practice in this institution for the normalization of electron fields. METHODS
AND MATERIALS
Irradiation technique The recent clinical treatments of pleural mesothelioma in this institution which followed the combined modality technique outlined by Hilaris et al. (5) were performed on a linear accelerator.* This machine is capable of delivering a 25 MV photon beam and electron beams of energies between 7 and 32 MeV. With this approach combined photon-electron treatments required the use of only one machine and could be performed efficiently and with the least inconvenience to the patient. Figure 1 illustrates the fields used during a combined photon-electron treatment. For the photon irradiation of lateral, superior and inferior surfaces a lung block was positioned to shield the central area shown in Figure 1. The electron treatment of the anterior and posterior lung surfaces was implemented by irradiating the previously shielded region with an electron beam of appropriate energy. For the phantom measurements the energy was selected to be 16 MeV as this was the lowest available energy which delivered a dose of at least 95% of maximum at the depth of
* Sagittaire
Therac 40.
the interface (in the absence of inhomogeneity). For ease of set up, the lung shield used for the photon irradiation and the lead alloy cut-out used during the electron component of the treatment were placed in the same shadow tray holder. In addition both fields were on the same center and their edges, defined by the 50% dose point, were coincident at the phantom surface. This approach ignores the marked differences between photon and electron beam penumbras and could be expected to lead to dose nonuniformity in the join up region. This aspect of the technique is commented upon later in this communication. It is recognized that dosimetry problems exist with small fields at high photon energies due to the loss of lateral electronic equilibrium. Although it was not considered practicable, for clinical treatments, to use different machines for the electron and photon components, the phantom study described in this report includes photon irradiation at 4 MV.+ The significance of this reduction in photon energy from 25 to 4 mV for the dosimetry of combined photon-electron treatment is discussed below. In view of the difficulties expected in the combined photon-electron approach, irradiations of the phantom were also performed with photon beams alone. These were carried out in the absence of the lung shielding block at both 4 and 25 MV for the entire field, as shown in Figure 1. Dose specification and normalization for the irradiation of a target volume in the form of a shell is rarely required in clinical radiotherapy. Figure 2 illustrates how such problems were handled in this study and how the clinical situation was modeled. The hemi-thorax being treated was simulated by a phantom, described in detail below, of rectangular cross section. The target volume was considered to be a shell extending 5 mm on either side of the lung-soft-tissue interface and is indicated by the hatched area in Figure 2. The symmetry of the simple geometrical
’ SHM Therapi
4.
Radiotherapy for pleural mesothelioma 0 M. SOUBRA
model used necessitated the irradiation of only one lateral surface to yield the required information. Photon beam dose distributions, whether or not they were used in conjunction with electrons, were normalized to 50% from each of the anterior and posterior beams at the lateral lung-soft tissue interface at the simulated mid-separation. This normalization point, shown in Figure 2, lays on the axis of rotation of the gantry. The dose to the anterior and posterior regions of the target was normalized at the center of the field to 100% from one electron beam at the depth of the interface. Such a normalization implies that only one electron beam contributes significantly to the dose to these points. The phantom The phantom used in this study was constructed of acrylic to simulate soft tissues and wood (either balsa or cedar) to simulate lung. Figure 2 shows the geometry of the situation being simulated although in practice the phantom had only two acrylic sides with the posterior side absent. Such a design was easier to construct and handle and, as shown below, the lack of backscatter resulting from the missing side did not significantly distort the dose distribution. The low density material simulating lung tissue was 20 X 20 cm in the plane parallel to the central axis of the beam and the average electron densities relative to water or acrylic, cedar and balsa were determined from the CT method of Battista and Bronskill(2) to be 1.14,0.42, and 0.11, respectively. The two electron densities for the simulated lung material bracket the values found in both normal and diseased lungs (11). The phantom was constructed in two identical halves of thickness 10 cm such that a film could be located in the mid plane of the phantom (the plane shown in Fig. 2) parallel to the beam axis. A clamping device ensured that air pockets on either side of the film were excluded. Experimental technique Wrapped radiotherapy verification film+ was used in this study. In mounting the film care was taken to ensure that the edge on the source side was flush with the phantom surface. The films were exposed in a plane containing the radiation beam central axis with an exposure time selected to produce a maximum optical density of about I. 1 on the film. Optical density measurements were performed with a densitometer,§ using a 1 mm diameter aperture with the conversion from optical density to absorbed dose being based upon experimentally determined calibration curves. As pointed out by Williamson et al. (13) the dose versus density calibration curve at 4 MV depends upon the depth of interest in the phantom. For this energy of irradiation calibration curves were measured in an acrylic phantom at the depths at which data were to be generated. At 25 MV agreement between the central axis depth dose measured with film (based on a single * Kodak
X-Omat
V.
1523
et al.
calibration curve) and that measured with an ionization chamber was better than 2.5%. At this higher photon energy it was not considered necessary to apply a depth correction to the optical density measurements. Calibration of the film for use during electron irradiation was also carried out in an acrylic phantom. As there is no evidence that the mass stopping power of lung is significantly different from that of soft tissue and, further, that stopping powers for muscle and acrylic are close over the range of interest in this study (6) electron dose values reported here may be interpreted as the dose to lung tissue embedded in the simulated lung material at the position of the film. During the photon beam irradiation for the mixed beam study a lung shield was placed on an acrylic shadow tray which was positioned at 69.5 cm from the radiation source. The block was 8 cm thick and made of lead alloy with its faces tapered to follow beam divergence. For the electron beam irradiations a lead alloy cut-out of 1.2 cm thickness was placed at the same position with respect to the source and phantom as the lung block for the photon fields. This cut-out defined the shape of the electron field shown in Figure 1. In all the measurements the phantom was placed in the radiation beam with its surface at 95 cm from the radiation source. With this arrangement the plane in which the photon beams were normalized (defined at 10 cm depth from the phantom surface, Fig. 2) contained the axis of gantry rotation. The open field dimensions were kept at 13.6 X 22.8 cm with the field center being the same for all irradiations. Beam profiles from single photon or electron fields were read at 3.7 cm, 4.7 cm, 10 cm, and 15.4 cm depths from the phantom surface. As illustrated in Figure 2 the beam profile data from single electron fields were normalized to 100% at the proximal acrylic-wood interface on the central axis. The photonbeam profiles were normalized to 50% at a point 10 cm deep and on the lateral acrylic-wood interface. The profiles representing the combined distribution of parallel opposed fields were generated by summing two single beam profiles at 10 cm to yield the profile at 10 cm or for the combined profile at 4.7 cm by summing profiles at 4.7 cm and 15.4 cm depth. As was pointed out above, the phantom differed from the geometry being simulated (Fig. 2) in that the distal simulated tissue layer was absent. To ensure that the lack of backscatter in the phantom study did not invalidate the results, limited measurements were performed in both photon and electron beams with acrylic placed immediately distal to the phantom. Profiles at 15.4 cm depth with and without backscatter differed by less than 1.5% in both photon and electron beams. RESULTS
AND DISCUSSION
Electron jields Figure 3 shows depth dose curves, normalized to 100% at the depth of maximum, along the central axis of a 16 5 Sargent Welsh.
1524
1. J. Radiation Oncology 0 Biology 0 Physics
MeV electron beam in an homogeneous acrylic phantom and when balsa or cedar was present to represent the lung. The arrow in Figure 3 indicates the position ofthe acrylicbalsa interface at 4.2 cm from the surface. Two important effects are observed in the presence of the lower density materials: firstly, the increased penetration of the electron beam with decreasing material density and secondly, the existence of a dose reduction in the region of the woodacrylic interface due to the changes in electron backscatter conditions. It can be immediately seen from Figure 3 that the exit dose from the simulated lung can be significant. At the exit from a balsa ‘lung’ (15.9 cm from the phantom surface) the dose is in excess of 20% of its maximum value. Under such conditions the treatment planning process described by Figure 2 in which the anterior and posterior surfaces are considered to be irradiated by single electron fields is clearly invalid. Note, however, that the more usual application of electrons in the thorax, for the treatment of the chest wall, will not lead to such severe exit dose problems. In the case under consideration here, the intention is to deliver a high percentage of the maximum dose to the pleural surface and to achieve this end the energy of the electrons leaving the target volume and entering the lung must necessarily be high. Profiles at three depths in the phantom with a balsa ‘lung’ irradiated with 16 MeV electrons are shown in Figure 4. Lateral scatter back into the ‘lung’ from the acrylicbalsa interface is particularly clear at 10 cm depth. This leads to an asymmetry in the dose profile with a higher dose in the balsa adjacent to the acrylic than at the symmetrically positioned point on the other side of the beam axis. Similar profiles were measured for the higher density cedar ‘lung’. In this case the dose was observed to drop
Depth (cm)
Fig. 3. Central axis depth dose curves for a 16 MeV electron beam in composite and homogeneous phantoms. Normalization is to the depth of maximum. The arrow in this and subsequent diagrams indicates the location of the interface.
June 1990, Volume 18, Number 6
Distance from unshielded field centre (cm)
Fig. 4. Dose profiles resulting from 16 MeV electron beam irradiation of the phantom with a balsa ‘lung’ (at 3.7, 4.7, and 10 cm depths from the phantom surface).
off more rapidly with depth as expected and, in addition, the profiles exhibited greater flatness in the central region of the beam. This latter effect is due to the shorter electron path length in cedar than balsa and hence a closer approach to lateral electronic equilibrium in the center of the field.
Shielded photon fields Figures 5 and 6 show beam profiles from a single field at various depths in the balsa phantom and for 25 and 4 MV, respectively. These profiles indicate the existence of dose perturbations in the neighborhood of the lateral interface region. These perturbations are particularly pronounced for 25 MV photon irradiation and are noted to increase with depth. Of greater significance for the combined photon-electron treatment of pleural mesothelioma is the relatively
-5
0
5
Distance from unshielded field centre (cm)
Fig. 5. Dose profiles resulting from irradiation of the phantom with a balsa ‘lung’ by a shielded 25 MV photon beam (at 3.7, 4.7, and 10 cm depths from the phantom surface).
1525
Radiotherapy for pleural mesothelioma 0 M. SOUBRAef a/
/
xe /
I
/‘-‘, \ I \
I /‘-.
!
!,
4
‘k,._.-
/i_ -5
, 0
5
Distance from unshielded field centre
t -5
(cm)
Fig. 6. Dose profiles resulting from irradiation of the phantom with a balsa ‘lung’ by a shielded 4 MV photon beam (at 3.1,4.7, and 10 cm depths from the phantom surface).
large dose under the lung block. At 25 MV and with a balsa ‘lung’ the profile at 10 cm depth illustrates that under the center of the block the dose exceeds 25% of its maximum value at that depth. Replacement of the balsa by cedar at 25 MV yielded profiles (not shown) which were similar to those of Figure 6. The dose under the block remained significant but was less pronounced than that in balsa at 25 MV. The primary transmission of the block used for lung shielding was calculated to be 4% and hence the cause of the high dose in the shielded lung must be sought elsewhere. From previous studies (9, 14) it is apparent that the cause of this effect is the scattering of high energy electrons into the low density material. Under the worst conditions studied here (25 MV and a balsa ‘lung’) the dose penumbra is diffused so greatly by this effect that it spreads out under the entire lung block.
t
I
I
-5
0
5
Distance from unshielded field centre km)
Fig. 7. Combined 25 MV photon- 16 MeV electron dose profiles with balsa ‘lung’ (at 4.7 and 10 cm depths from the phantom surface).
Distance
I
I
0
5
from unshielded
field centre
km)
Fig. 8. Combined 25 MV photon- 16 MeV electron dose profiles with cedar ‘lung’ (at 4.7 and 10 cm depths from the phantom surface).
With either a lower energy or a more dense lung the effect of electron scattering under the block is reduced. However, at 10 cm depth it remains at 14% (25 MV photons and a cedar ‘lung’) and 18% (4 MV photons and balsa ‘lung’) of the maximum dose at that depth. It is well recognized that commercially available dose computation algorithms do not explicitly account for photon induced secondary electron transport (4, 8). Thus, conventional treatment planning approaches ignore this potentially significant effect.
Combined electron and photon Jields Figures 7-9 show combined dose profiles at 4.7 and 10 cm depth under three conditions. It can be seen that under all the conditions considered there is a very significant absolute dose error when the treatment planning tech-
t -5 Distance
I
I
0
5
from field centre
km)
Fig. 9. Combined 4 MV photon-16 MeV electron dose profiles with balsa lung (at 4.7 and 10 cm depths from the phantom surface).
1526
I. J. Radiation Oncology 0 Biology 0 Physics
nique illustrated in Figure 2 is used. Due principally to the large extent of the penumbra from the penetrating electron beam the dose in the lateral acrylic sheet exceeds that intended by between 40 and 60%. It is also remarkable that the lung dose not only exceeds that expected from a simplistic planning approach (4) but also exceeds by a very large amount that prescribed to the target. The exceptionally high lung dose is due to the high penetrating ability of the opposed electron beams in low density material and to the contribution from side scattered electrons set in motion during photon irradiation. Reducing the incident electron energy from 16 MeV to 13 MeV accompanied by a corresponding reduction in acrylic thickness (to 2.5 cm) has been shown to produce very similar results to those presented. As was noted earlier no special attempt was made to optimize the dose distribution across the photon-electron join-up region. The penalty for adopting the simple to implement approach is apparent from Figures 7. 8 and 9 and is not unexpected. Dose inhomogeneities up to 423% were observed under the present conditions. If this technique of combining photon and electron fields were to meet its objective of treating the pleural membrane without significantly irradiating the encapsulated lung, efforts to improve the join-up region would be justified. However, in view of the other major dosimetry difficulties uncovered during this study penumbra modification to achieve a more uniform distribution across the join-up region is not warranted. Unblocked photon jields
Figure 10 shows two dose profiles for parallel opposed 25 MV photon irradiation of the phantom with a balsa ‘lung’. Normalization to 100% from the combined fields was at the lateral acrylic-balsa interface at 10 cm depth. Measurements (not shown) were also made with a cedar
t -5
Distance
I
0 from unshielded
5 field centre (cm)
Fig. 10. Dose profiles resulting from opposed 25 MV photon fields with balsa ‘lung’ (at 4.7 and 10 cm depths from the phantom surface).
June 1990, Volume 18. Number 6
‘lung’ at 25 MV and at 4 MV photon energy with both cedar and balsa ‘lungs’. From these measurements it was observed that the mid-line dose (10 cm depth) increased with respect to that at 4.7 cm as both the ‘lung’ density was reduced (i.e. cedar to balsa) and the irradiating energy was reduced (i.e. 25 MV to 4 MV). In comparing Figure 10 to Figures 7,8 and 9 it is noted that the dose homogeneity is greater with irradiation by photons alone although a dose gradient does exist in the region of the lateral interface. ‘Lung’ dose is higher than peak ‘soft tissue’ dose for photons alone in contrast with the combined photon-electron treatment. A final comparative observation is that treatment planning of photons alone is far simpler than that of the combined treatment. Good agreement has been reported by other workers when studying computational algorithms for large photon fields irradiating inhomogeneous tissue (8).
CONCLUSIONS The treatment of patients with malignant pleural mesothelioma using a combined photon-electron technique (5) has been shown, under our experimental conditions, to deliver a high dose to the lung relative to the pleural surface. Although a strict comparison with a recently reported assessment of such a technique (7) is not possible due to differences in the irradiation techniques used, our results appear to indicate a higher relative lung dose than has been previously suggested (7). In addition, our experiments have highlighted the dosimetric difficulties of using photon and electron beams in regions containing low density lung. Failure to acknowledge the significant penetration in the lung of both primary electrons and those secondary to photon irradiation can lead to a dose to the anterior and posterior pleural surfaces 50% greater than that prescribed on the basis of the assumptions implicit in Figure 2. The clinically expedient approach to field matching used here has been shown, predictably, to lead to significant dose inhomogeneity in the join-up region with excessive dose delivered to those regions of the target volume lying parallel to the beam axis. However, in view of the failure of this technique to meet its primary objective of sparing normal lung tissue it is not clear that effort directed to improving either the absolute dosimetry or the field matching, in this case, is warranted. Photon irradiation alone will not, of course, spare the lung. However, the dose is reasonably uniform and calculable using commercially available algorithms. The techniques we report on here for the irradiation of a target volume encapsulating lower density normal tissue have been shown to have serious shortcomings. An external beam technique which offers the possibility of producing more desirable dose distributions is that of Conformation Therapy (3, 10). This newer approach is certainly worthy of investigation for the treatment of malignant mesothelioma.
Radiotherapy for pleural mesothelioma 0 M. SOUBRA et al.
1527
REFERENCES 1. Aisner, J.; Wiemik, P. H. Chemotherapy in the treatment of malignant mesothelioma. Sem. Oncol. 8:335-343; 198 1. 2. Battista, J. J.; Bronskill, M. J. Compton-scatter imaging of transverse sections: an overall appraisal and evaluation for radiotherapy planning. Phys. Med. Biol. 26:8 l-99; 198 1. 3. Brace, J. A.; Davy, T. J.; Skeggs, B. M.; Williams, H. S. Conformation therapy at the Royal Free Hospital. A progress report on the tracking cobalt project. Br. J. Radiol. 54: 10681074; 1981. 4. El-Khatib, E.; Battista, J. J. Accuracy of lung dose calculations for large-field irradiation with 6-MV x-rays. Med. Phys. l:lll-116; 1985. 5. Hilaris, B. S.; Dattatreyudu, N.; Kwong, E.; Kutcher, G. J.; Martini, N. Pleurectomy and intraoperative brachytherapy and postoperative radiation in the treatment of malignant pleural mesothelioma. Int. J. Radiat. Oncol. Biol. Phys. 10: 325-331; 1984. 6. ICRU Report 35. Radiation dosimetry: electron beams with energies between 1 and 50 MeV. Bethesda, Maryland; ICRU: 1984. I. Kutcher, G. J.; Kestler, C.; Greenblatt, D.; Brenner, H.; Hilaris, B. S.; Nori, D. Technique for external beam treatment for mesothelioma. Int. J. Radiat. Oncol. Biol. Phys. 13:1747-1752; 1987.
8. Mackie, T. R.; El-Khatib, E.; Battista, J.; Scrimger, J. Lung dose corrections for 6- and 15-MV x-rays. Med. Phys. 12: 327-332; 1985. 9. Mackie, T. R.; Scrimger, J. W.; Battista, J. J. A convolution method of calculating dose for 15-MV x-rays. Med. Phys. 12:188-196; 1985. 10. Takahashi, K.; Purdy, J.; Liu, Y. Y. Work in progress: treatment planning system for conformation radiotherapy. Radiology 147:567-573; 1983. 11. Van Dyk, J.; Keane, T. J.: Rider, M. B. Lung density as measured by computerized tomography: implications for radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 8: 1363-l 37 1; 1982. 12. Wanebo, H. J.; Martini, N.; Melamed, M. R.; Hilaris, B. S.; Beattie, E. J. Pleural mesothelioma. Cancer 38:248 l2488; 1976. 13. Williamson, J. F.; Khan, F. M.; Sharma, S. C. Film dosimetry of mega-voltage photon beams: a practical method of isodensity-to-isodose curve conversion. Med. Phys. 8:9498; 1981. 14. Young, M. E. J.; Kornelsen, R. 0. Dose corrections for lowdensity tissue inhomogeneities and air channels for IO-MV x-rays. Med. Phys. 10:450-455; 1983.
Ini J Radmmn Oncology Bml. Phys, Vol. 18, pp. 1521-1527 Printed in the U.S.A. All rights reserved.
??Technical Innovations and Notes
PHYSICAL ASPECTS OF EXTERNAL TREATMENT OF MALIGNANT M. SOUBRA, M.Sc.,*
P. B. DUNSCOMBE,
PH.D.,*
BEAM RADIOTHERAPY FOR THE PLEURAL MESOTHELIOMA D. I. HODSON, M.D.+ AND G. WONG, M.D.+
Manitoba Cancer Treatment and Research Foundation, 100 Olivia Street, Winnipeg Manitoba, R3E OV9 Canada The optimization of radiotherapy for the treatment of malignant mesothelioma highlights many of the currently outstanding problems in clinical radiation physics. The experimental investigation of an intuitively attractive irradiation technique with combined photon and electron beams using a specially constructed phantom has established that, due to the penetration in low density material of both primary electrons and those secondary to photon irradiation, the normal lung tissue is not spared to any significant degree by such a technique. Furthermore, great care needs to be exercised in the treatment planning calculations for this approach if absolute dosimetry errors as large as 50% are to be avoided. Malignant mesothelioma,
Radiotherapy, Treatment planning, Combined photon, Electron beams.
Although pleural mesothelioma is an uncommon malignancy, the increasing incidence and the difficulties associated with therapy represent a therapeutic challenge. The insidious progression of symptoms and the diffuse pathogenesis generally preclude a definitive surgical extirpation and, although prospective comparisons are scarce, it appears that combined modality therapy with surgery and radiotherapy is superior to either modality alone ( 12). The role of chemotherapy remains speculative in spite of extensive studies using a variety of investigational and conventional regimen (1). The bulk of the therapeutic experience confirms the clinical impression that this is a relatively radioresistant malignancy requiring doses for control that are well above accepted pulmonary tolerance levels (12). If radiation therapy is to be used optimally in this disease, then the difficult dosimetric problems associated with the treatment of a “shell” (the pleural surface) need to be addressed. Hilaris and coworkers (5) have suggested a novel and intuitively attractive irradiation technique whereby the pleural surfaces lying predominantly parallel to the radiation beam axis are treated with photons and the anterior and posterior pleural surfaces by electron beams. The intention of such an approach is to limit the radiation dose to the lung which is encapsulated by the target. In less than half the patients treated by Hilaris et al. (5) it
was desirable or possible to augment the dose delivered by external beam radiotherapy with brachytherapy techniques. In this communication we consider some of the physical aspects of implementing mixed external beam radiotherapy of the pleural surface. Such a treatment approach involves many of the currently outstanding problems in clinical radiation physics. These are: abutting photon and electron beams with different penumbras; the dose distribution resulting from photon irradiation in the neighborhood of gross electron density inhomogeneities and the behavior of electron beams in the presence of inhomogeneities. In the absence of appropriate computational tools for tackling these problems our approach has been experimental. A phantom which simulates the essential features of the treated hemithorax albeit with simplified geometry has been constructed. Irradiation of this phantom with photon and electron beams under appropriate clinical conditions has served to elucidate some of the dosimetric aspects of the treatment planning of pleural mesothelioma. We have investigated the effect of photon energy by using beams of both 4 and 25 MV. In addition, the combined photon-electron approach has been compared with the clinically simpler technique of irradiation with photons alone. Since completing the experiments and analysis reported in this communication more details of the original Hilaris technique have been published (7). It is apparent that di-
Reprint requests to: M. Soubra, MSc., Department of Medical Physics, Ottawa Regional Cancer Centre, 190 Melrose Ave., Ottawa, KlY 4K7 Canada. Acknowledgements-The skill of Jules Legal and his colleagues who constructed the phantom used in this study is gratefully
acknowledged. Dr. G. Froese is thanked cussions. * Department of Medical Physics. + Department of Radiation Oncology. Accepted for publication 20 December
INTRODUCTION
1521
for many useful dis-
1989.
1522
1. J. Radiation Oncology 0 Biology 0 Physics
June 1990. Volume 18, Number 6
100% from one electron field
5 cm
100% from two photon fields
Fig. 2. The irradiation geometry. The hatched region at the interface between acrylic (soft tissue) and cedar or balsa (lung) represents the target area. Arrows indicate the normalization points for both photon and electron fields. X
4 or 25 MV
Fig. 1. Photon and electron portals used to irradiate the phantom. For the combined treatment technique the electron field was blocked during photon irradiation.
rect comparison between the present work and that recently reported (7) is precluded by the different electron energies and normalizations in the two studies. The irradiation conditions used here were determined by the design of the phantom and by common clinical practice in this institution for the normalization of electron fields. METHODS
AND MATERIALS
Irradiation technique The recent clinical treatments of pleural mesothelioma in this institution which followed the combined modality technique outlined by Hilaris et al. (5) were performed on a linear accelerator.* This machine is capable of delivering a 25 MV photon beam and electron beams of energies between 7 and 32 MeV. With this approach combined photon-electron treatments required the use of only one machine and could be performed efficiently and with the least inconvenience to the patient. Figure 1 illustrates the fields used during a combined photon-electron treatment. For the photon irradiation of lateral, superior and inferior surfaces a lung block was positioned to shield the central area shown in Figure 1. The electron treatment of the anterior and posterior lung surfaces was implemented by irradiating the previously shielded region with an electron beam of appropriate energy. For the phantom measurements the energy was selected to be 16 MeV as this was the lowest available energy which delivered a dose of at least 95% of maximum at the depth of
* Sagittaire
Therac 40.
the interface (in the absence of inhomogeneity). For ease of set up, the lung shield used for the photon irradiation and the lead alloy cut-out used during the electron component of the treatment were placed in the same shadow tray holder. In addition both fields were on the same center and their edges, defined by the 50% dose point, were coincident at the phantom surface. This approach ignores the marked differences between photon and electron beam penumbras and could be expected to lead to dose nonuniformity in the join up region. This aspect of the technique is commented upon later in this communication. It is recognized that dosimetry problems exist with small fields at high photon energies due to the loss of lateral electronic equilibrium. Although it was not considered practicable, for clinical treatments, to use different machines for the electron and photon components, the phantom study described in this report includes photon irradiation at 4 MV.+ The significance of this reduction in photon energy from 25 to 4 mV for the dosimetry of combined photon-electron treatment is discussed below. In view of the difficulties expected in the combined photon-electron approach, irradiations of the phantom were also performed with photon beams alone. These were carried out in the absence of the lung shielding block at both 4 and 25 MV for the entire field, as shown in Figure 1. Dose specification and normalization for the irradiation of a target volume in the form of a shell is rarely required in clinical radiotherapy. Figure 2 illustrates how such problems were handled in this study and how the clinical situation was modeled. The hemi-thorax being treated was simulated by a phantom, described in detail below, of rectangular cross section. The target volume was considered to be a shell extending 5 mm on either side of the lung-soft-tissue interface and is indicated by the hatched area in Figure 2. The symmetry of the simple geometrical
’ SHM Therapi
4.
Radiotherapy for pleural mesothelioma 0 M. SOUBRA
model used necessitated the irradiation of only one lateral surface to yield the required information. Photon beam dose distributions, whether or not they were used in conjunction with electrons, were normalized to 50% from each of the anterior and posterior beams at the lateral lung-soft tissue interface at the simulated mid-separation. This normalization point, shown in Figure 2, lays on the axis of rotation of the gantry. The dose to the anterior and posterior regions of the target was normalized at the center of the field to 100% from one electron beam at the depth of the interface. Such a normalization implies that only one electron beam contributes significantly to the dose to these points. The phantom The phantom used in this study was constructed of acrylic to simulate soft tissues and wood (either balsa or cedar) to simulate lung. Figure 2 shows the geometry of the situation being simulated although in practice the phantom had only two acrylic sides with the posterior side absent. Such a design was easier to construct and handle and, as shown below, the lack of backscatter resulting from the missing side did not significantly distort the dose distribution. The low density material simulating lung tissue was 20 X 20 cm in the plane parallel to the central axis of the beam and the average electron densities relative to water or acrylic, cedar and balsa were determined from the CT method of Battista and Bronskill(2) to be 1.14,0.42, and 0.11, respectively. The two electron densities for the simulated lung material bracket the values found in both normal and diseased lungs (11). The phantom was constructed in two identical halves of thickness 10 cm such that a film could be located in the mid plane of the phantom (the plane shown in Fig. 2) parallel to the beam axis. A clamping device ensured that air pockets on either side of the film were excluded. Experimental technique Wrapped radiotherapy verification film+ was used in this study. In mounting the film care was taken to ensure that the edge on the source side was flush with the phantom surface. The films were exposed in a plane containing the radiation beam central axis with an exposure time selected to produce a maximum optical density of about I. 1 on the film. Optical density measurements were performed with a densitometer,§ using a 1 mm diameter aperture with the conversion from optical density to absorbed dose being based upon experimentally determined calibration curves. As pointed out by Williamson et al. (13) the dose versus density calibration curve at 4 MV depends upon the depth of interest in the phantom. For this energy of irradiation calibration curves were measured in an acrylic phantom at the depths at which data were to be generated. At 25 MV agreement between the central axis depth dose measured with film (based on a single * Kodak
X-Omat
V.
1523
et al.
calibration curve) and that measured with an ionization chamber was better than 2.5%. At this higher photon energy it was not considered necessary to apply a depth correction to the optical density measurements. Calibration of the film for use during electron irradiation was also carried out in an acrylic phantom. As there is no evidence that the mass stopping power of lung is significantly different from that of soft tissue and, further, that stopping powers for muscle and acrylic are close over the range of interest in this study (6) electron dose values reported here may be interpreted as the dose to lung tissue embedded in the simulated lung material at the position of the film. During the photon beam irradiation for the mixed beam study a lung shield was placed on an acrylic shadow tray which was positioned at 69.5 cm from the radiation source. The block was 8 cm thick and made of lead alloy with its faces tapered to follow beam divergence. For the electron beam irradiations a lead alloy cut-out of 1.2 cm thickness was placed at the same position with respect to the source and phantom as the lung block for the photon fields. This cut-out defined the shape of the electron field shown in Figure 1. In all the measurements the phantom was placed in the radiation beam with its surface at 95 cm from the radiation source. With this arrangement the plane in which the photon beams were normalized (defined at 10 cm depth from the phantom surface, Fig. 2) contained the axis of gantry rotation. The open field dimensions were kept at 13.6 X 22.8 cm with the field center being the same for all irradiations. Beam profiles from single photon or electron fields were read at 3.7 cm, 4.7 cm, 10 cm, and 15.4 cm depths from the phantom surface. As illustrated in Figure 2 the beam profile data from single electron fields were normalized to 100% at the proximal acrylic-wood interface on the central axis. The photonbeam profiles were normalized to 50% at a point 10 cm deep and on the lateral acrylic-wood interface. The profiles representing the combined distribution of parallel opposed fields were generated by summing two single beam profiles at 10 cm to yield the profile at 10 cm or for the combined profile at 4.7 cm by summing profiles at 4.7 cm and 15.4 cm depth. As was pointed out above, the phantom differed from the geometry being simulated (Fig. 2) in that the distal simulated tissue layer was absent. To ensure that the lack of backscatter in the phantom study did not invalidate the results, limited measurements were performed in both photon and electron beams with acrylic placed immediately distal to the phantom. Profiles at 15.4 cm depth with and without backscatter differed by less than 1.5% in both photon and electron beams. RESULTS
AND DISCUSSION
Electron jields Figure 3 shows depth dose curves, normalized to 100% at the depth of maximum, along the central axis of a 16 5 Sargent Welsh.
1524
1. J. Radiation Oncology 0 Biology 0 Physics
MeV electron beam in an homogeneous acrylic phantom and when balsa or cedar was present to represent the lung. The arrow in Figure 3 indicates the position ofthe acrylicbalsa interface at 4.2 cm from the surface. Two important effects are observed in the presence of the lower density materials: firstly, the increased penetration of the electron beam with decreasing material density and secondly, the existence of a dose reduction in the region of the woodacrylic interface due to the changes in electron backscatter conditions. It can be immediately seen from Figure 3 that the exit dose from the simulated lung can be significant. At the exit from a balsa ‘lung’ (15.9 cm from the phantom surface) the dose is in excess of 20% of its maximum value. Under such conditions the treatment planning process described by Figure 2 in which the anterior and posterior surfaces are considered to be irradiated by single electron fields is clearly invalid. Note, however, that the more usual application of electrons in the thorax, for the treatment of the chest wall, will not lead to such severe exit dose problems. In the case under consideration here, the intention is to deliver a high percentage of the maximum dose to the pleural surface and to achieve this end the energy of the electrons leaving the target volume and entering the lung must necessarily be high. Profiles at three depths in the phantom with a balsa ‘lung’ irradiated with 16 MeV electrons are shown in Figure 4. Lateral scatter back into the ‘lung’ from the acrylicbalsa interface is particularly clear at 10 cm depth. This leads to an asymmetry in the dose profile with a higher dose in the balsa adjacent to the acrylic than at the symmetrically positioned point on the other side of the beam axis. Similar profiles were measured for the higher density cedar ‘lung’. In this case the dose was observed to drop
Depth (cm)
Fig. 3. Central axis depth dose curves for a 16 MeV electron beam in composite and homogeneous phantoms. Normalization is to the depth of maximum. The arrow in this and subsequent diagrams indicates the location of the interface.
June 1990, Volume 18, Number 6
Distance from unshielded field centre (cm)
Fig. 4. Dose profiles resulting from 16 MeV electron beam irradiation of the phantom with a balsa ‘lung’ (at 3.7, 4.7, and 10 cm depths from the phantom surface).
off more rapidly with depth as expected and, in addition, the profiles exhibited greater flatness in the central region of the beam. This latter effect is due to the shorter electron path length in cedar than balsa and hence a closer approach to lateral electronic equilibrium in the center of the field.
Shielded photon fields Figures 5 and 6 show beam profiles from a single field at various depths in the balsa phantom and for 25 and 4 MV, respectively. These profiles indicate the existence of dose perturbations in the neighborhood of the lateral interface region. These perturbations are particularly pronounced for 25 MV photon irradiation and are noted to increase with depth. Of greater significance for the combined photon-electron treatment of pleural mesothelioma is the relatively
-5
0
5
Distance from unshielded field centre (cm)
Fig. 5. Dose profiles resulting from irradiation of the phantom with a balsa ‘lung’ by a shielded 25 MV photon beam (at 3.7, 4.7, and 10 cm depths from the phantom surface).
1525
Radiotherapy for pleural mesothelioma 0 M. SOUBRAef a/
/
xe /
I
/‘-‘, \ I \
I /‘-.
!
!,
4
‘k,._.-
/i_ -5
, 0
5
Distance from unshielded field centre
t -5
(cm)
Fig. 6. Dose profiles resulting from irradiation of the phantom with a balsa ‘lung’ by a shielded 4 MV photon beam (at 3.1,4.7, and 10 cm depths from the phantom surface).
large dose under the lung block. At 25 MV and with a balsa ‘lung’ the profile at 10 cm depth illustrates that under the center of the block the dose exceeds 25% of its maximum value at that depth. Replacement of the balsa by cedar at 25 MV yielded profiles (not shown) which were similar to those of Figure 6. The dose under the block remained significant but was less pronounced than that in balsa at 25 MV. The primary transmission of the block used for lung shielding was calculated to be 4% and hence the cause of the high dose in the shielded lung must be sought elsewhere. From previous studies (9, 14) it is apparent that the cause of this effect is the scattering of high energy electrons into the low density material. Under the worst conditions studied here (25 MV and a balsa ‘lung’) the dose penumbra is diffused so greatly by this effect that it spreads out under the entire lung block.
t
I
I
-5
0
5
Distance from unshielded field centre km)
Fig. 7. Combined 25 MV photon- 16 MeV electron dose profiles with balsa ‘lung’ (at 4.7 and 10 cm depths from the phantom surface).
Distance
I
I
0
5
from unshielded
field centre
km)
Fig. 8. Combined 25 MV photon- 16 MeV electron dose profiles with cedar ‘lung’ (at 4.7 and 10 cm depths from the phantom surface).
With either a lower energy or a more dense lung the effect of electron scattering under the block is reduced. However, at 10 cm depth it remains at 14% (25 MV photons and a cedar ‘lung’) and 18% (4 MV photons and balsa ‘lung’) of the maximum dose at that depth. It is well recognized that commercially available dose computation algorithms do not explicitly account for photon induced secondary electron transport (4, 8). Thus, conventional treatment planning approaches ignore this potentially significant effect.
Combined electron and photon Jields Figures 7-9 show combined dose profiles at 4.7 and 10 cm depth under three conditions. It can be seen that under all the conditions considered there is a very significant absolute dose error when the treatment planning tech-
t -5 Distance
I
I
0
5
from field centre
km)
Fig. 9. Combined 4 MV photon-16 MeV electron dose profiles with balsa lung (at 4.7 and 10 cm depths from the phantom surface).
1526
I. J. Radiation Oncology 0 Biology 0 Physics
nique illustrated in Figure 2 is used. Due principally to the large extent of the penumbra from the penetrating electron beam the dose in the lateral acrylic sheet exceeds that intended by between 40 and 60%. It is also remarkable that the lung dose not only exceeds that expected from a simplistic planning approach (4) but also exceeds by a very large amount that prescribed to the target. The exceptionally high lung dose is due to the high penetrating ability of the opposed electron beams in low density material and to the contribution from side scattered electrons set in motion during photon irradiation. Reducing the incident electron energy from 16 MeV to 13 MeV accompanied by a corresponding reduction in acrylic thickness (to 2.5 cm) has been shown to produce very similar results to those presented. As was noted earlier no special attempt was made to optimize the dose distribution across the photon-electron join-up region. The penalty for adopting the simple to implement approach is apparent from Figures 7. 8 and 9 and is not unexpected. Dose inhomogeneities up to 423% were observed under the present conditions. If this technique of combining photon and electron fields were to meet its objective of treating the pleural membrane without significantly irradiating the encapsulated lung, efforts to improve the join-up region would be justified. However, in view of the other major dosimetry difficulties uncovered during this study penumbra modification to achieve a more uniform distribution across the join-up region is not warranted. Unblocked photon jields
Figure 10 shows two dose profiles for parallel opposed 25 MV photon irradiation of the phantom with a balsa ‘lung’. Normalization to 100% from the combined fields was at the lateral acrylic-balsa interface at 10 cm depth. Measurements (not shown) were also made with a cedar
t -5
Distance
I
0 from unshielded
5 field centre (cm)
Fig. 10. Dose profiles resulting from opposed 25 MV photon fields with balsa ‘lung’ (at 4.7 and 10 cm depths from the phantom surface).
June 1990, Volume 18. Number 6
‘lung’ at 25 MV and at 4 MV photon energy with both cedar and balsa ‘lungs’. From these measurements it was observed that the mid-line dose (10 cm depth) increased with respect to that at 4.7 cm as both the ‘lung’ density was reduced (i.e. cedar to balsa) and the irradiating energy was reduced (i.e. 25 MV to 4 MV). In comparing Figure 10 to Figures 7,8 and 9 it is noted that the dose homogeneity is greater with irradiation by photons alone although a dose gradient does exist in the region of the lateral interface. ‘Lung’ dose is higher than peak ‘soft tissue’ dose for photons alone in contrast with the combined photon-electron treatment. A final comparative observation is that treatment planning of photons alone is far simpler than that of the combined treatment. Good agreement has been reported by other workers when studying computational algorithms for large photon fields irradiating inhomogeneous tissue (8).
CONCLUSIONS The treatment of patients with malignant pleural mesothelioma using a combined photon-electron technique (5) has been shown, under our experimental conditions, to deliver a high dose to the lung relative to the pleural surface. Although a strict comparison with a recently reported assessment of such a technique (7) is not possible due to differences in the irradiation techniques used, our results appear to indicate a higher relative lung dose than has been previously suggested (7). In addition, our experiments have highlighted the dosimetric difficulties of using photon and electron beams in regions containing low density lung. Failure to acknowledge the significant penetration in the lung of both primary electrons and those secondary to photon irradiation can lead to a dose to the anterior and posterior pleural surfaces 50% greater than that prescribed on the basis of the assumptions implicit in Figure 2. The clinically expedient approach to field matching used here has been shown, predictably, to lead to significant dose inhomogeneity in the join-up region with excessive dose delivered to those regions of the target volume lying parallel to the beam axis. However, in view of the failure of this technique to meet its primary objective of sparing normal lung tissue it is not clear that effort directed to improving either the absolute dosimetry or the field matching, in this case, is warranted. Photon irradiation alone will not, of course, spare the lung. However, the dose is reasonably uniform and calculable using commercially available algorithms. The techniques we report on here for the irradiation of a target volume encapsulating lower density normal tissue have been shown to have serious shortcomings. An external beam technique which offers the possibility of producing more desirable dose distributions is that of Conformation Therapy (3, 10). This newer approach is certainly worthy of investigation for the treatment of malignant mesothelioma.
Radiotherapy for pleural mesothelioma 0 M. SOUBRA et al.
1527
REFERENCES 1. Aisner, J.; Wiemik, P. H. Chemotherapy in the treatment of malignant mesothelioma. Sem. Oncol. 8:335-343; 198 1. 2. Battista, J. J.; Bronskill, M. J. Compton-scatter imaging of transverse sections: an overall appraisal and evaluation for radiotherapy planning. Phys. Med. Biol. 26:8 l-99; 198 1. 3. Brace, J. A.; Davy, T. J.; Skeggs, B. M.; Williams, H. S. Conformation therapy at the Royal Free Hospital. A progress report on the tracking cobalt project. Br. J. Radiol. 54: 10681074; 1981. 4. El-Khatib, E.; Battista, J. J. Accuracy of lung dose calculations for large-field irradiation with 6-MV x-rays. Med. Phys. l:lll-116; 1985. 5. Hilaris, B. S.; Dattatreyudu, N.; Kwong, E.; Kutcher, G. J.; Martini, N. Pleurectomy and intraoperative brachytherapy and postoperative radiation in the treatment of malignant pleural mesothelioma. Int. J. Radiat. Oncol. Biol. Phys. 10: 325-331; 1984. 6. ICRU Report 35. Radiation dosimetry: electron beams with energies between 1 and 50 MeV. Bethesda, Maryland; ICRU: 1984. I. Kutcher, G. J.; Kestler, C.; Greenblatt, D.; Brenner, H.; Hilaris, B. S.; Nori, D. Technique for external beam treatment for mesothelioma. Int. J. Radiat. Oncol. Biol. Phys. 13:1747-1752; 1987.
8. Mackie, T. R.; El-Khatib, E.; Battista, J.; Scrimger, J. Lung dose corrections for 6- and 15-MV x-rays. Med. Phys. 12: 327-332; 1985. 9. Mackie, T. R.; Scrimger, J. W.; Battista, J. J. A convolution method of calculating dose for 15-MV x-rays. Med. Phys. 12:188-196; 1985. 10. Takahashi, K.; Purdy, J.; Liu, Y. Y. Work in progress: treatment planning system for conformation radiotherapy. Radiology 147:567-573; 1983. 11. Van Dyk, J.; Keane, T. J.: Rider, M. B. Lung density as measured by computerized tomography: implications for radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 8: 1363-l 37 1; 1982. 12. Wanebo, H. J.; Martini, N.; Melamed, M. R.; Hilaris, B. S.; Beattie, E. J. Pleural mesothelioma. Cancer 38:248 l2488; 1976. 13. Williamson, J. F.; Khan, F. M.; Sharma, S. C. Film dosimetry of mega-voltage photon beams: a practical method of isodensity-to-isodose curve conversion. Med. Phys. 8:9498; 1981. 14. Young, M. E. J.; Kornelsen, R. 0. Dose corrections for lowdensity tissue inhomogeneities and air channels for IO-MV x-rays. Med. Phys. 10:450-455; 1983.
