Abutting Orthogonal Electron and Photon Beams in the Head and Neck Region Using Asymmetrical

Transcription

1 ORIGINAL ARTICLE Abutting Orthogonal Electron and Photon Beams in the Head and Neck Region Using Asymmetrical Photon Beam Edges Dag R. Olsen, Lina Backaert and Jan F. Evensen From the Departments of Medical Physics (L. Backaert, D.R. Olsen) and Medical Oncology and Radiotherapy (J.F. Evensen), The Norwegian Radium Hospital, Oslo, and Department of Health Science, The College of Oslo, Oslo (Dag R. Olsen), Norway Correspondence to: D. R. Olsen, Dept. of Medical Physics, The Norwegian Radium Hospital, Box 20, Montebello, N-0310 Oslo, Norway Acta Oncologica Vol. 36, No. 6, pp , 1997 Irradiation of target volumes within the ethmoid sinus and nasal cavity region has, at our institution, been performed using one anterior electron beam, abutted to two opposing lateral photon beams, in order to exclude the orbit from the irradiated volume. However, this technique give rise to larger dose hetereogeneities within the target volume, as well as undesired hot spots in the chiasma region. To reduce the inhomogenious distribution, we have introduced asymmetrical photon beam edges to broaden the penumbra of the two opposing photon beams resembling the dose gradient of the perpendicular electron beam. The achieved reduction in dose heterogeneity, measured from the differential dose-volume-histogram, is in the order of 30-40% when applying the asymmetrical abutting technique compared with the conventional two opposing photon beam technique. As demonstrated by calculating the tumour control probability (TCP), the increase in dose homogeneity within the target volume, may be of clinical significance. Received 2 August 1996 Accepted 3 July 1997 In some external beam radiotherapy applications, a combination of electron and photon beams may be used to optimally cover the target volume. This is exemplified by tumours in the nasal cavity and the ethmoid sinus in which the target volumes may involve both the posterior orbital regions as well as the nasal cavity and the ethmoid sinus (1, 2). When there is no involvement of the orbit itself, it will usually be desirable to exclude the orbit from the treatment volume. To our knowledge, there are no field arrangements suitable for this kind of target volumes which has reached general acceptance. At our institution the treatment technique for this kind of target volume has been two lateral, opposing photon beams covering the posterior regions, and a frontal electron beam covering the involvement of the nasal cavity as well as the ethmoid sinus (3) (Fig. 1). The abutment of the distal fall-off of the electron beam to the edge of the photon beams normally result in poor dose homogeneity within the field matching zone. Maximum dose uniformity within the field matching zone of the treatment volume is important to accomplish for at least two reasons: to avoid reduced tumour control due to under-dosage, and to avoid blindness due to radiation induced damage of the optical pathway. In matching of parallel beams, dose homogeneity is primarily determined by the penumbra and the divergence of the beams applied (4-6). When orthogonal electron and photon beams are abutted, both the penumbra of the photon field, the depth dose distribution of the electron field, and the distance between electron beam entrance and photon field edge are of great influence for the dose homogeneity of the field matching area. Optimal matching conditions is assumed to be obtained when 1) the mean dose gradient of the penumbra region of the two opposing photon fields is equal to the magnitude of the dose gradient of the electron beam, and 2) the 50% dose level of the two opposing photon beams is positioned at the depth of the 50% dose level of the electron beam. The electron and photon beam dose gradients can be matched by decreasing the dose gradient of the penumbra region of the two opposing photon beams. In the present study, we have accomplished this by applying asymmetrical field edges of the photon beams. The same penumbra broadening could also be achieved by blocking one side of a symmetrical field. The present study quantifies the dose inhomogeneity reduction by using displaced field edges of the photon fields when matched to an orthogonal electron beam. The 0 Scandinavian University Press ISSN X Acta Oncologica

2 586 D. R. Olsen et al. Acta Oncologica 36 (1997) clinical benefit is assessed by calculation of tumour control probability (TCP) (7). Dose distribution calculations are performed for a number of patients with lesions in the nasal cavity and ethmoid sinus. The calculations is validated by performing both dose distribution measurements and calculations in the Alderson phantom. MATERIAL AND METHODS Patient material For 8 previously treated patients with cancer in the nasal cavity, maxillary and the ethmoid sinuses, retrospective dose planning was performed applying orthogonal electron and photon beams. Field matching were performed both with and without asymmetrical field edges of the photon beams. For all the patients, information on computed tomography (CT) of the region of interest was present. Dose distributions were calculated and presented as differential dose-volume-histograms (DDVH). The DDVH were also used as input for calculation of TCP (7). External beam arrangement and jield matching Two lateral, opposing 6 MV (Clinac 2100 C/D, Varian, Palo Alto) photon beams of size 70 x 85 mm were applied for lateral irradiation of the Alderson phantom. The central axis of the beams were positioned at the level of intersection of slab 2 and 3 of the phantom. The angle between the two photon beams was somewhat less than 180", yielding a non-divergent anterior field edge and constant dose gradients in the lateral direction. A fixed distance of 7 mm between the anterior field edges of the two opposing photon beams was accomplished by use of asymmetrical fields. The photon beams were abutted to a orthogonal anterior 12 MeV electron beam of 35 x 40 mm. Matching of the orthogonal beams were performed at the 50% depth of the electron beam, R,, and at the 50% dose level of the penumbra region of the two opposing photon fields. The distance between electron beam entrance and penumbra region of the photon beams was adjusted to fulfil this matching criterion, using a wax bolus covering the nose (Tenax Wax, White Manufacturing Ltd.). The bolus was so dimensioned that the thickness of the bolus added to the distance between the surface and the edge of the two opposing photon beams became equal to R,, of the applied electron beam everywhere at the same horizontal level (Fig. 1). The same field arrangement was applied for the retrospective dose calculations of the selected patients. The size of the two opposing photon beam fields were ranging in size from 65 to 143 mm x 80 to 176 mm, and electron beam field from 35 to 50 mm x 35 to 80 mm. For the purpose of the present study, they were all calculated with an electron beam energy of 12 MeV and a anterior field edge asymmetry of 7 mm. Photon beam energy applied was either 5 or 6 MV. (a) beam entrance (elechons) Bolus beam entrance Fig. I. CT based treatment plan showing the mid-orbital plane of a selected patient (a). Irradiation is performed using two opposing photon beams with asymmetrical field edges and an anterior electron beam. A wax bolus covering the nose was applied to accomplish a distance between the beam entrance and the edge of the two opposing photon beams equal to R, of the electron beam. The positions of 12 dose measurements points in the Alderson phantom (b), where both TLD measurements and dose calculations were performed.

3 Acta Oncologica 36 (1997) Matching of orthogonal electron and opposing photon beams 587 Dose gradient calculations The dose gradient of the electron beam at the depth of 50% dose level, (dd/dz)50, is calculated from the equation: h E 10 where G,, is the normalized dose gradient taken at depth of 50% dose level, as recommended by Brahme & Svensson (8) to describe dose reduction beyond the therapeutic depth of electron beams. The parameter G is obtained by standard parametrization of measured depth dose curves using the RFA-7 field analyser (Scanditronix). Re is the depth of intersection of the tangent at the 50Y0 dose level and the 0% dose level, and D, is the maximum dose level. Dose gradient calculations were performed for four different fields ranging in size from 30 x 30 to 60 x 60 mm. The mean dose gradient in the penumbra region of two opposing photon beams was calculated as the dose reduction from 80% to 2OYn dose level divided by the distance between the two dose levels, dso-zcum,. Calculation of the mean photon dose gradient was performed for a number of different field edge distances in the range of 0 to 14 mm. All calculations were performed applying a cubic 150 x 150 x 150 mm phantom, using the dose planning system TMS (Helax, Sweeden). Dose distribution measurements and irradiation A 6 mm polystyrene slab was manufactured and introduced in between slab 2 and 3 of the Alderson phantom. Holes for positioning of 1 2 thermo-luminescence-dosimeter (TLD) rods were drilled in the polystyrene slab (Fig. lb) at the dose distribution matching zone. TLD-100, LiF detectors, from Harshaw were applied for dose measurements. Absorbed dose in each of the 12 selected points were based on three individual measurements. The rods were read not later than 24 h after irradiation. No correction for fading was therefore applied. After preheating at 100 C for 10 min, read-out of the TLD rods were performed at approximately 300"C, where integrated light output was converted to dose by an energy specific conversion factor. Read-out was performed on the Harshaw 2000D, TL detector. Absorbed dose from the electron beam and the two photon beams were measured at separate sessions, since the TL sensitivity differed for the two beam qualities applied. Photon beam irradiation of the Alderson phantom was performed so that the dose at the intersection of all three beam axis were 2 Gy. For the electron beam irradiation, the dose at D, was 2.00 Gy. Dose distribution calculation and evaluation CT-based 3D dose distribution calculations of the Alderson phantom and the 8 selected patients were performed on the TMS treatment planning system (Helax, Sweden). Y K a,.- m a, u) B 2-1 L a, n 6- c.- a, '0 m 4 - L m a, u) B 2-0 ' I I I I I I I I E 8 Field border distance (mm) 0 1 I I I I I I I Electron beam energy (MeV) Fig. 2. Mean dose gradient of the penumbra region of the two opposing photon beams measured as a function of the distance between the field edges (a), and the dose gradient of the electron beam depth dose distribution calculate according to eq. [I] as a function of beam energy, for field size 60 x 60 mm (filled circles), 50 x 50 mm (open triangles), 40 x 40 mm (filled rectangles), and 30 x 30 mm (open squares), (b). Electron beam dose calculations were performed using a pencil-beam algorithm accomplishing correction for patient/phantom inhomogeneities (9-12). The photon beam algorithm applied for dose calculations also accomplished correction for inhomogeneities. The dose distribution was normalised (1000/0) at the point of intersection of all the three beam axis. A minimum of 5 CT images were applied with an interspacing of Evaluating the dose homogeneity was based on DDVH taken at three different volumes of interest, VOI, (Fig. lb), all within the dose distribution matching zone: 40 x 15 x 10 mm, 25 x 15 x 10 mm, and 15 x 15 x 10 mm (Fig. Ib). For the largest VOI, also TCP was calculated using the formula suggested by Webb & Nahum (7), but with the modification that only a fixed cc-value is considered: b

4 588 D. R. Olsen et al. Acta Oncologica 36 (1997) g h 2.0 alv) 0 '0 1.5 e! I Calculated dose (Gy) Fig. 3. Calculated vs. measured total dose (both electrons and photons) in the Alderson phantom of the 12 selected positions of the field matching zone indicated in Fig. Ib, using symmetrical beam edges (open circles), and asymmetrical edges (filled circles). TCP = l { exp[ - p *VT exp( - a *di)]} where p is the tumour cell density assumed to be lo7 Vi is the volume of one bin in the DDVH, and di is the dose of this bin. A number of 64 bins are applied for the DDVH used in the TCP calculations. 0.7 a, 0.4 E : PI RESULTS 0.4 n 0.3 -/ I Dose gradient measurements The dose gradients, (dd/dz)go, of the electron beams were found to be in the range of 2-4.2% per mm for a beam energy of 16 MeV, dependent of the field size, and is increasing by decreasing energy to % per mm for a 8 MeV beam (Fig. 2b). At 12 MeV the dose gradient is approximately 4.5% per mm for a 40 x 40 mm field and close to 5.5% per mm for a 30 x 30 mm field (Fig. 2b). Most of the electron beam fields applied in this treatment technique are of the size 30 x 30 to 40 x 40 mm. A field edge distance corresponding to the same mean dose gradient of the penumbra region of the two opposing photon beams were found to be in the range of 6 to 8 mm (Fig. 2a). To fulfil the requirement of the first of the two field matching criteria, a field edge distance of 7 mm and an electron beam energy of 12 MeV were chosen for the retrospective dose planning of the Alderson phantom and the selected patients. Dose calculation and measurements in the Alderson phantom Verification of the performed dose calculations is accomplished by plotting the calculated dose vs. the measured dose for a number of selected points in the target volume (Fig. 3) of the Alderson phantom. As can be seen from the figure, the calculations performed on the dose planning 0 0.o \ C n II 1 I I I I I I I Relative dose (%) Fig. 4. DDVH of a selected patient irradiated by the conventional technique with symmetrical field edges (a,c,e), and the novel asymmetrical field edge technique (b,d,f) for the three different VOIs described (a and b smallest VOI, c and d: medium VOI, and e and f: largest VOI)

5 Acta Oncologica 36 (1997) Mutching of' orthogonal electron and opposing photon beams asymetrical symetrical a" 0 to 1.0 a patients 7 8 Fig. 5. Dose distributic het 'ogeneity evaluat~ I by the ratio between the dose of the 90% and 10% percentiles of the DDVH of the largest VOI for the 8 individual patients.included. system tend to overestimate the dose in the low dose region (< 2 Gy), whereas an underestimation is seen at higher dose levels (>2 Gy). The range of the measured and the calculated doses is similar for both techniques. Dose distribution homogeneity and TCP The general quality of the dose distributions obtained by abutting orthogonal electron beam and photon beams with symmetrical and asymmetrical field edges, is shown for one of the patients in Fig. 4. For the smallest VOI, a characteristic double-peak DDVH is seen when applying symmetrical field edges, with an expectation value of about 75% and 120% of the specification dose of the low and high dose peak respectively (Fig. 4a). If instead asymmetrical field edges of the two opposing photon fields are applied, the dose distribution is significantly narrowed, and the characteristic double-peak characteristic no longer to be seen (Fig. 4b). When analyzing larger VOIs (Fig. 4 c, d, e, 0, the double-peak characteristic is less pronounced also for the asymmetrical technique, since more and more bins with dose levels close to the normalization dose is included. The same degree of distribution broadening is, however still present for the non-asymmetrical field edge technique, and also the same distribution narrowing is obtained when changing to the asymmetrical irradiation technique. The narrowing'of the DDVH by introducing the asymmetrical photon field edge when abutting the electron and photon beams, can be evaluated from Fig. 5, where the ratio between doses at the 90% and 10Y0 percentile is shown for the largest VOI. The dose distribution narrowing, measured by the 90% to 10% percentile ratio is in the range of 30% patients Fig. 6. Calculated TCP ratio for all the 8 individual patients included in the study, using eq. [2], considering only the largest VOI, and assuming an a value of 0.35 Gy-', a target volume dose of 50 Gy, and lo7 tumour cells per m3. TCPJTCP, is the ratio between the TCP accomplished by the asymmetrical technique and the TCP obtained using the conventional symmetrical field edge technique. to 43.5% for these 8 patients. The mean ratio is decreasing from 1.67 to 1.32 when introducing asymmetrical field edges of the two opposing photon beams, a reduction of more than 30% in dose inhomogeneity. An even larger reduction L a (W') Fig. 7. Mean TCP ratio for the 8 patients calculated, using the same conditions as in Fig. 6, but now as function of a.

6 590 D. R. Olsen et al. Acta Oncologica 36 (1997) is seen if the same analysis is applied to the two smaller VOIS. The possible clinical effect of the reduction in dose inhomogeneity is assessed by calculation of TCP, based only on the DDVH for the largest VOI. As can be seen from Fig. 6, most of the patients would have had an increased probability of tumour control, if the asymmetrical irradiation technique had been applied. The mean gain in TCP is 18.5%, ranging from less than 1% to over 63%, for an target volume dose of 50 Gy, assuming an u value of 0.35 Gyp'. The mean gain in TCP is, however, dependent on the U-value selected. As illustrated in Fig. 7, an even larger TCP gain can be accomplished for lower u values. On the other hand, for higher a-values the TCP ratio is approaching unity. DISCUSSION Selected electron beam energy and photon beam edge asymmetry were chosen to utilize an increased dose distribution homogeneity in a limited number of patients. As can be seen from Fig. 2, there are a number of combinations of electron beam energy and field edge asymmetry, fulfilling the criterion of optimal field matching of orthogonal electron and photon beams. For individual optimization, different sets of beam energy and field edge asymmetry may then be appropriate. Choosing a higher electron beam energy, the dose gradient, (dd/dz),,, will decrease and the distance to the photon field edges, R,,, increase. Higher electron beam energies therefore entail a more posterior positioning of the photon fields as well as larger photon field edge asymmetry. This may be acceptable and even desirable for more posterior located target volumes without involvement of regions close to the posterior part of the orbit. One must bear in mind that large beam edge asymmetry entails a less smooth dose profile of the penumbra region of the photon fields, resulting in a changing dose gradient in the transversal direction of the photon beams. The mean dose gradient may then no longer be a good parameter for optimal field matching of orthogonal beams. This characteristics is increasingly pronounced when the field edge asymmetry approaches twice the initial photon beam penumbra. A better way to accomplish the desired penumbra broadening might be the use of a graduate wedge technique (14). Also the addition of small doses from small laterally incident photon beams may be an alternative (13). Even better will probably be the use of dynamically shrinking photon fields. The steepness, and therefore also (dd/dz),,, of the descending part of the depth dose curve is larger for lower electron beam energies than for beams of high energy. The use of low energy beams allows smaller asymmetrical field edge distances, and may be advantageous to anterior located target volumes with involvement close to the orbit. For a number of patients, electron beams of energies lower than 10 MeV, with R,,s of mm, will entail an anterior positioning of the photon beams including parts of the orbit. The photon beams will then give rise to unacceptable dose levels at the lens and retina. Electron beam energies lower than 10 MeV should therefore be of limited use in this treatment technique. Also, when applying low electron energy beams, the increasing dose gradient requires high precision manufacturing of the wax bolus. The application of the combined irradiation technique using orthogonal photon and electron beams, including the suggested modification in abutting these fields, requires accurate dose distribution calculations. These calculations are also required to be performed in 3D. Of crucial interest is the dose distribution calculation of the anterior electron beam since it most often includes air cavities and other inhomogeneities. As can be seen from Fig. 3, significant differences between measured and calculated dose is found. Therefore, to refine the abutment technique when applying orthogonal electron and opposing photon beams, better models for calculation of 3D dose distribution from electron beams in the presence of inhomogeneities are highly desirable. Despite the uncertainties in electron beam calculations, the broadening of the penumbra by applying asymmetrical photon beam edges when abutting orthogonal electron and photon beams, will reduce the dose distribution heterogeneity compared with that of the conventional abutting technique. The TD,,, of the chiasma is approximately 50 Gy, whereas the complication probability is increased to 50% for 65 Gy, using blindness as endpoint (15). Hot regions in the anterior part of the photon fields may, for a number of patients, occur, or partly occur, at the position of chiasma (Fig. la). Reduction of the dose level in this region may be of great importance. If the required target dose to the 8 selected patients were 50 Gy, most of the patients have received chiasma doses in the range of Gy (data not shown), using symmetrical matching technique. From a calculated risk of radiation induced blindness of 5%, the dose inhomogeneity may have raised the complication probability to a magnitude of 50%, assuming that the hot region encompasses the chiasma. A significant reduction in dose level is accomplished by applying the asymmetrical field edge matching technique. The dose to the hot region will, for the 8 selected patients, be in the order of Gy, a reduction by approximately 8 Gy. For a number of the patients this reduction may be sufficient to avoid a high complication probability of radiation induced blindness of the chiasma. For others, the chiasma dose level will still be close to TD,,,,. For anterior located target volumes, low and medium electron energies is applicable. By choosing low and medium energies, the position of the hot-spots will be moved anterior of the chiasma. Medium and low energy electron beams may be preferred when hot regions occur at the position of the chiasma. Unlike the approach by Webb & Nahum (7), TCP is calculated by using one single value of a for the 8 individ-

7 Acta Oncologica 36 (1997) Matching of orthogonal electron and opposing photon beams 591 ual patients (Fig. 6). Since our calculations of tumour control is based on a limited number of patients, characterized by relatively large differences in 3D dose distribution (Fig. 5), the quantitative gain in tumour control in a larger patient population, using the modified abutting technique, cannot be reached from this material even if a radiation sensitivity distribution were included. For larger populations of patients, a radiation sensitivity distribution, however, would be appropriate when performing retrospective calculations. For our limited number of patients we have found it more relevant to include an analysis of the consequences of differences in the radiation sensitivity parameter CI (Fig. 7). As indicated in Fig. 7, the gain in TCP by applying asymmetrical beam edges is approaching unity for high sensitivity tumour cells, assuming a mean target volume dose of 50 Gy. This is due to the fact that doses in the order of 50 Gy, for large values of a, and a tumour cell density of lo7 cells cmp3, result in a very high degree of tumour control probability and that a reduction in cold spots within the target volume results in minor TCP gain only. However, at lower mean target volume doses, corresponding to lower initial TCP, a significant gain in TCP when applying the modified abutting technique even for more sensitive tumour cells (higher values of a), may be obtained. REFERENCES 1. Ho JHC. Treatment of nasopharyngeal carcinoma. In: Treatment of cancer Chapman and Hall, 1982; Lee AWM, Law SCK, Foo W, et al. Nasopharyngeal carcinoma: local control by megavoltage irradiation. Br J Radiol 1993; 66: Ang KK, Kanders JHAM, Peters L. Radiotherapy for head and neck cancers-indications and techniques. Lea and Febiger, 1994; Harms WB, Purdy JA. Abutment of high energy electron fields. Int J Radiat Oncol Biol Phys 1991; 20: Hopfan S, Reid A, Simpson L, Ager PJ. Clinical complications arising from overlapping of adjacent radiation fieldsphysical and technical considerations. Int J Radiat Oncol Biol Phys 1977; 2: Kalend AM, Zwicker RD, Wu A, Sternick ES. A beam-edge modifier for abutting electron fields. Med Phys 1985; 12: Webb S, Nahum AE. A model for calculating tumour control probability in radiotherapy including effects of inhomogeneous distribution of dose and clonogenic cell density. Phys Med Biol 1993; 38: Brahme A, Svensson H. Specification of electron beam quality from the central-axis depth dose distribution. Med Phys 1976: 3: Lax I. Accuracy in clinical electron beam dose planning using pencil beam algorithms. Radiother Oncol 1987; 10: Lax I, Brahme A, Andreo P. Electron beam dose planning using Gaussian beams. Improved radial dose profiles. Acta Radiol 1983; (Suppl. 364): Lax I. Inhomogeneity corrections in electron-beam dose planning. Limitations with semi-infinite slab approximation. Phys Med Biol 1986; 31: Lax I, Brahme A. Electron beam dose planning using Gaussian beams. Energy and spatial scaling with inhomogeneities. Acta Radiol Oncol 1985; 24: Woudstra E. Improvement of depth dose distribution by addition of small doses from laterally incident beams. Radiother Oncol 1988; 13: Tobler M, Prows J, Leavitt DD. Graduate block technique for the treatment of paranasal sinus tumors. Medical Dosimetry 1991; 16: Emami B, Lyman J, Brown A, et al. Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys 1991; 21: