Three dimensional conformal radiation therapy in pediatric parameningeal rhabdomyosarcomas

Int. J. Radiation

Oncology

Biol.

Pergamon

Phys., Vol. 33, No. 5. pp. 985-991. 1995 Copyright 0 1995 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/95 $9.50 + .OO

0360-3016(95)00551-6

l

Clinical

Report

THREE

DIMENSIONAL CONFORMAL RADIATION THERAPY PARAMENINGEAL RHABDOMYOSARCOMAS

JEFF M. MICHALSKI,

M.D.,*

RANJAN

K. SUR, M.D.,

AND JAMES A. PURDY,

D.N.B.,+ PH.D.*

WILLIAM

IN PEDIATRIC B. HARMS,

B.S.,*

*Radiation Oncology Center, Mallinckrodt Institute of Radiology, WashingtonUniversity Medical Center, St. Louis, MO, ‘Departmentof Radiotherapy,Medwin Hospitals,Hyderabad,India Purpose: We evaluated the utility of three dimensional (3D) treatment planning in the management of children with parameningeal head and neck rhabdomyosarcomas. Methods and Materials: Five children with parameningeal rhabdomyosarcoma were referred for treatment at our radiation oncology center from May 1990 through January 1993. Each patient was evaluated, staged, and treated according to the Intergroup Rhabdomyosarcoma Study. Patients were immobilized and underwent a computed tomography scan with contrast in the treatment position. Tumor and normal tissues were identified with assistance from a diagnostic radiologist and defined in each slice. The patients were then planned and treated with the assistance of a 3D treatment planning system. A second plan was then devised by another physician without the benefit of the 3D volumetric display. The target volumes designed with the 3D system and the two-dimensional (2D) method were then compared. The doshnetric coverage to tumor, tumor plus margin, and normal tissues was also compared with the two methods of treatment planning. Results: The apparent size of the gross tumor volume was underestimated with the conventional 2D planning method relative to the 3D method. When margin was added around the gross tumor to account for microscopic extension of disease in the 2D method, the expected area of coverage improved relative to the 3D method. In each circumstance, the minimum dose that covered the gross tumor was substantially less with the 2D method than with the 3D method. The inadequate dosimetric coverage was especially pronounced when the necessary margin to account for subclinical disease was added. In each case, the 2D plans would have delivered substantial dose to adjacent normal tissues and organs, resulting in a higher incidence of significant complications. Conclusions: 3D conformal radiation therapy has a demonstrated advantage in the treatment of sarcomas of the head and neck. The improved doshnetric coverage of the tumor and its margin for subclinical extensions may result in improvement in local control of these tumors. In addition, lowering of radiation dose to adjacent critical structures may help lower the incidence of adverse late effects in children. Conformal,

Radiotherapy,

Pediatric

neoplasms.

INTRODUCTION Radiation therapy is a key element in the multimodality managementof pediatric rhabdomyosarcoma (11). These tumors have the propensity to infiltrate beyond their organ of origin and invade adjacent normal tissues. To ensure adequate coverage of the infiltrating tumor edge, a margin of 2 to 5 cm around the prechemotherapy gross tumor is typically encompassedby radiation therapy portals (17). Because of the risk of serious late morbidity related to the administration of irradiation in young patients, it may

be tempting to reduce this margin in an effort to minimize radiation dose to surrounding critical structures (8). Inadequate margins have resulted in an increased rate of local failure and death, as demonstrated by the Intergroup Rhabdomyosarcoma Study in the management of parameningeal tumors (16). The use of computed tomography (CT) for planning the shape of radiation fields directed to the primary tumor has made a significant impact in the local control of these tumors (5). The introduction of three dimensional (3D) radiotherapy planning software provides the potential to better

Reprint requeststo: Jeff M. Michalski, M.D., Mallinckrodt Institute of Radiology, Radiation Oncology Center, Box 8224, 4939 Children’sPlace, Suite 5500, St. Louis, MO 63110. Acknowledgements-Dr. Sur’s participation in this study has been made possibleby funding from the International Union Against Cancer(UICC), Geneva,throughits InternationalCan-

cer Technology Transferaward (#629/92).Dr. Sur alsowishes to acknowledgethe helpand supportof Dr. CarlosPerez,Director, RadiationOncology Center,Mallinckrodt Institute of Radiology, for obtainingthe awardandaccommodations at theInstitute. Accepted for publication 13 October 1994. 985

986

I. J. Radiation Oncology 0 Biology 0 Physics

Volume 33. Number 5, 1995

Table 1. Patient characteristics Patient

Age (years)

Site

Intracranial extension

Skull base erosion

IRS stage

IRS group

Radiation dose (Gy)

1 2 3 4 5

4 12 4 1 9

ITF ITF ME ITF ITF

Yes Yes No No No

Yes Yes Yes Yes Yes

T2bNO T2bNO T2aNO T2bNl T2bNl

IV III III III III

59.40 59.40 59.40 59.40 50.40

ITF = Infratemporal fossa. ME = Middle ear. IRS = Intergroup Rhabdomyosarcoma Study-IV.

define target volumes and their relationship to adjacent critical structures than CT based planning alone. This report summarizes our initial experience in the use of 3D radiation therapy planning in the management of pediatric head and neck sarcomas.

METHODS

AND

MATERIALS

Patient population From May 1990 through January 1993, five children with parameningeal head and neck rhabdomyosarcomas were referred for treatment at our radiation oncology center. Patient characteristics are presented in Table 1. Each patient was enrolled in the Intergroup Rhabdomyosarcoma Study (IRS-IV) and was evaluated, staged, and treated according to protocol (Intergroup Rhabdomyosarcoma Study-IV, for Stage II and III diseases, written communication, unpublished, 1991). This Intergroup Study randomized patients with gross disease (group III) to receive either single daily fraction (conventional) radiation therapy to 50.4 Gy (1.8 Gylfraction) or twice daily (hyperfractionated) irradiation to 59.4 Gy (1.2 Gy/ fraction). Each patient underwent diagnostic cross-sectional imaging of the head and neck with both a contrastenhanced treatment planning CT and magnetic resonance imaging (MRI) scanning. In the development of both plans for each patient, a review of these crosssectional studies with a diagnostic radiologist, as well as the clinical history and physical examination, served as the basis for target volume delineation for radiation therapy planning purposes. In each case, a 3D radiotherapy plan was created by the treating physician (J.M.M.) and was used to execute the actual treatments. A second radiation oncologist (RKS) created target volumes and developed a 2D radiotherapy plan using all the clinical information available (history, physical exam findings, and diagnostic imaging studies) but without the prior knowledge of the 3D plan. No constraints were made in creating an optimum 2D plan.

‘Siemens, DRH Scanner, Iselin, NJ.

30 planning system The 3D radiation therapy planning system in use in our facility has been previously described (14). Features on this system that facilitate the 3D planning process include the beam’s eye view and “room view” displays. These displays are complementary in design of beam apertures (customized portals) and choosing optimal beam projections. A diagnostic CT scanner’ has been interfaced with our 3D treatment planning system (13). This CT simulator allows scanning of patients in the treatment position on a modified flat couch with appropriate immobilization thermoplastic masks in place. Computed tomography scanning was done at intervals of at least 8 mm scan thickness with detail at the tumor region with a 4 mm scan thickness. Scans typically were taken from the head vertex to the lung apices after administration of intravenous contrast material. The CT scan data were then transferred from the CT scanner to our 3D radiation therapy planning system using computer network communication software and then converted to the format required by the planning system. Virtual simulation Target volumes and critical structures were contoured on the CT scans on a slice-by-slice basis. We implemented contouring software (IMEX) developed by the University of North Carolina computer science group that displays the CT data and allows the treatment planner or radiation oncologist to draw contours representing tumor, target, and normal tissues using a computer mouse (12). These contours are used to generate ring stack or shaded solid surface representations of each contoured structure in the 3D planning system. This provided the 3D anatomic display required to plan beam geometrys including isocenter placement, beam direction, and projection, and beam modifiers (custom blocks, wedges, compensating filters). These contours also provided the volumetric data required to compute dose volume histograms for plan analysis (4). Target volume delineation Target volumes were defined using the International Commission on Radiation Units and Measurements, NO.

Conformal radiotherapy in pediatric sarcoma 0 Table 2. Relative target volume projections

.I. M.

MICHAELSKI

(area) on lateral digital reconstructed

987

et al.

simulation

CTV

GTV

radiographs PTV

Pt. no.

2D (cm’)

3D (cm’)

3D/2D (ASo)

2D (cm’)

3D (cm’)

3D/2D (A%)

2D (cm’)

3D (cm?

3D/2D (A%)

1 2 3 4 5

44.7 4.4 29.7 44.1 11.9

58.9 5.6 33.4 57.8 15

+24 +21.5 +11 +23.7 +22.2

90.5 31.4 84.9 89.2 49.7

96.7 29.1 77.4 98.3 55.4

+6.4 -8.7 -9.7 +9.3 + 10.3

113.0 42.6 100.7 105.1 65.4

116.3 39.2 96.1 115.2 60.7

+2.8 -8.5 -4.8 +8.8 -7.8

GTV = gross tumor volume. CTV = clinical target volume. PTV = planning target volume. Pt. No. = patient identification number. A% = percentage change of target area.

(10). The gross tumor volume (GTV) includes any imaged abnormality thought to represent actual tumor. When findings on the treatment planning CT were diagnostic information from the pavague or equivocal, tient’s pretherapy MRI scan or the clinical physical examination was used to determine the extent of grossdisease. The IRS-IV study requires a margin of at least 2 cm

50 guidelines

around

the gross tumor

to account

for subclinical

micro-

scopic extension of diseaseoutside the imaged gross tumor. This volume corresponds with the ICRU clinical target volume (CTV) and was added on the 3D tumor volume by concentric expansion of the predefined tumor volume. When a contour came to the skin surface, the

Fig. 1. The beam’s eye view display of a right lateral treatment portal. The 3D gross tumor volume (central, solid) and planning target volumes (outer, transparent) are evident in the infratemporal region of patient 2. The portal aperture design with the 2D technique does not give an adequate margin around either the PTV or CTV and shields gross tumor in the inferiormost region.

*Claris CAD, Claris Corp., SantaClara, CA.

margin was less than 2 cm to keep the contour within the body. An additional margin of 3 to 7 mm was then added circumferentially around the CTV to account for uncertainty in daily setup or patient motion within the immobilization device. This is the planning target volume (PTV) that the prescription dose optimally should cover. For the 2D plans, a 2 cm margin was added around the perimeter of the GTV to define the CTV and another 5 mm around the CTV to define the PTV. In each circumstance in the definition of the CTV and PTV, a superior and inferior cap are created on CT slices above and below the visualized GTV. The IRS protocol sets dose limits to critical structures such as the optic nerve and chiasm (46.8 Gy), spinal cord (45 Gy), lacrimal glands (41.4 Gy), and lens (14.4 Gy). When possible, these tolerance levels were respected unless it was deemed clinically necessary to treat beyond these dose guidelines to ensure adequate tumor control and, ultimately, survival.

20 treatment planning The presumed location and shape of the GTV, CTV, and PTV were outlined on a lateral digital reconstructed radiograph (DRR) to compare the ability to define target volumes with each of the planning methods. The same DRRs were then printed with the 3D defined target volumes delineated on them. The area of each target volume projection was then calculated from the 2D and 3D exercises using a graphics program2 on a desktop computer.’ For 2D planning purposes, no restrictions on beam geometry, blocking, or modifiers were made. In all but one circumstance opposedlateral fields were chosenby the 2D planning physician because of the massive and complex nature of these patients’ tumors. In one case (patient #3) a coplanar wedge plan was created to treat a middle ear tumor. These 2D plans with customized shielding were then imported into the 3D radiation therapy planning sys-

‘Macintosh

IIsi, Apple Computer,

Inc., Cupertino,

CA.

988

I. J. Radiation

GROSS

TUMOR

7

^r 60 !2 50

.I. .i.

ii

.I, ‘? .i.. .i..

Oncology

.......... I 0 Biology

0 Physics

Volume

VOLUME

5, 1995

PLANNING

TARGET

VOLUME

.(.

i

~

4 5 P&NT IDEN3TIFICATION (4

GROSS

33, Number

TUMOR

1

2 PATIENT

IDENTIFICATION

(b)

VOLUME

PLANNING

TARGET

VOLUME

n 70 n ag

60

g

40

0

r 8 Q 0 I’

0

2-D PLAN 3-D PLAN

50

P

30

=9

20

E

‘O 0 1

PATIENT

IDENTIFICATION (cl

PATIENT

IDENTIFICATION

5

(d)

Fig. 2. (a) Minimum doseto the grosstumor volume. (b) Minimum doseto the planningtarget volume. (c)Mean doseto the grosstumor volume. (d) Mean doseto the planningtarget volume. tern for quantitative comparison of the two methods of treatment planning. Two sets of treatment plans were devised for each patient. In the first instance, the 3D treatment planning system features were used to the full extent to develop and optimize a “conformal” treatment plan to encompassthe 3D planning target volumes, yet respecting when feasible, the tolerance of surrounding critical structures. In the second instance, the plan was implemented using the 2D defined target volumes drawn on the DRR. Beam orientation, custom blocking, differential dosimetric weighting, and wedge selection were all done without the 3D display for these 2D plans. Plan evaluation Room view display of the various target volumes and critical structures with wire cage isodosecontours allowed

for a rapid qualitative evaluation of each set of plans (14). Quantitative interplan evaluation was bestperformed with the assistanceof cumulative and differential dose-volume histograms using our Graphical Plan Evaluation Tool (GPET) (3). Dose statistics, such minimum dose(smallest dose delivered to less than 5% of target volume), maximum dose (highest dose delivered to less than 5% of target volume), and mean dose were calculated with the dose volume histograms.

RESULTS Targei volume definition In each case, the apparent size of the GTV was underestimated, with the conventional 2D planning method relative to the 3D method. The 3D GTV projections were

Conformal

CNS STRUCTURE

radiotherapy

in pediatric

BRAINSTEM :

2

:

3

:

4

:

:

5

:

2

3

4

5

IDENTIFICATION

(a)

OPTIC APPARATUS

60 9 s=

50

i

40

0

:

30

s

20

I

10 0

MICHAELSKI

et al.

989

Target volume dosimetry Using the 2D and 3D defined beam shaping and field arrangements, plans were generated in the 3D planning system for dosimetric comparisons. The minimum and mean dose planned to the GTV and PTV are displayed in Fig. 2. In each circumstance, the minimum dose that covers the gross tumor is less when planned by the 2D method than in the 3D method. The lack of adequate coverage is especially pronounced when one takes into account the margin necessary to cover both extensions of subclinical disease and uncertainties in daily setup. Planning target volume is substantially underdosed with the 2D plan in each of these patients. The mean dose delivered to the GTV is not significantly different from that of the 3D plan in any of the five cases. Likewise, the mean dose to the PTV shows some differences between the 2D and 3D plan, but they are not as striking as the differences in the minimum dose. The 3D plans typically irradiated the target volumes in a more homogenous fashion; hence, there are little differences between the minimum, mean, and maximum doses delivered to the target volumes in the 3D treatment plans. On the other hand, the 2D treatment plans frequently had substantial heterogeneity in their dose distributions through the target volume as well as some of the surrounding normal structures. As a result, the mean doses in the 2D plan compare favorably to those in the 3D plan despite “hot and cold” spots. This emphasizes the need to review all information provided in the 3D treatment plan including the dose statistics, dose-volume histograms, and isodose distributions.

..... ... ...... ..... 1.II

PATIENT

IPSILATERAL NERVE

0 J. M.

either shielded (two cases)or had less than the minimum 2 cm margin requirement (Fig. 1).

DOSES

70

sarcoma

DOSES

rC

............... ........

.........

d PATIENT

.... ... .... ...

~

2

3

4

5

IDENTIFICATION

(b)

Fig. 3. Maximum dosedeliveredto normalstructures.Dark bars indicate the doseplannedwith the 2D treatmentplan, and the light barsindicatethe doseplannedwith the 3D treatmentplan. (a) Radiationdoseto CNS critical structures.(b) Radiationdose to the optic nervesand chiasm. consistently larger (11 to 24%) than the volumes estimated with manual measurementsfrom the CT or MRI data (Table 2). When margins were added around the GTV, the differences between the apparent 2D and 3D CTVs and PTVs were not as great, in part because the differences were normalized to larger target areas. In some instances, addition of a geometric margin around the 2D GTV resulted in a treatment area significantly larger than that estimated with the 3D system. However, a qualitative review of these target areas disclosed that tumor coverage was not always improved, despite a similar or larger area being treated. Areas of grosstumor were

Normal tissue dosimetry In each case, the 2D plans would have delivered substantial dose to adjacent normal tissues and organs. In nearly all cases, the dose would have resulted in a high risk of clinically significant complications, including optic chiasmal (bilateral) blindness, cataracts, xerostomia, brain or brain stem injury, or mandibular injury. The 3D plans performed extremely well with one patient receiving a high ipsilateral optic nerve dose, one receiving a high peak mandibular dose, and two others getting a moderate contralateral parotid dose. In each of these, the GTV was in close proximity to these normal structures. Figure 3 illustrates the differences in maximum dose delivery to normal neurologic structures by each method. In only one instance would the 3D plan have irradiated the ipsilateral optic nerve above tolerance. This child’s tumor was adjacent to this structure. The 2D plan would have exceeded normal tissue tolerances in as many as eight CNS structures in these five patients. DISCUSSION

The introduction of CT scanninginto the routine practice of radiation oncology has greatly improved our ability to

990

I. J. Radiation Oncology 0 Biology l Physics

encompasstarget volumes with external beam irradiation (7). It has been reported that modifications in beam shaping may be necessarywhen CT scanning is used to define of the treatment area when compared with plain film radiographsor physical assessment alone (1,2,6,9). In this study of pediatric head and neck rhabdomyosarcomapatients, we have shown that when cross-sectionalimaging is available to the treatment planner, tumor targeting may still be inadequate. In each instancea clinician’s perception of projected gross tumor volume on lateral radiograph differed significantly from that which was projected from the 3D reconstructions. Typically, the perceived tumor size was smaller with the 2D method than with the 3D method, and treatment marginswould have beencompromised.A similar improvement in treatment planning has been described using 2D techniquesin patients with paranasalsinuscarcinomas(18). Theseperceived differences may be exaggeratedin pediatric head and neck cancers becauseof the clinician’s bias to reduce morbidity to the many surroundingcritical structures. It is routine in the radiotherapeutic managementof soft tissue sarcomas,including rhabdomyosarcoma, to encompassthe GTV with generousmargins becauseof the infiltrative nature of these neoplasms. This margin corresponds to the ICRU CTV. Attempts at narrowing this margin for subclinical diseasehave, in somecases,proved disastrous with high relapse rates (16). It is possible that tight margins increase the likelihood of gross tumor missessecondary to poor perception of tumor location in two dimensions. When appropriate margins were added to our 2D volumes, the magnitude of the differences between the 2D and the 3D areas of coverage decreased.

Volume 33, Number 5, 1995

The adding of a margin to ensure adequate GTV and CTV coverage in 2D plans always came at the cost of increasing dose and potential for toxicity to adjacent critical structures. A few of the 3D plans occasionally exceeded the tolerance of an adjacent normal structure, but only when that structure was in extremely close proximity to the gross tumor. The expected risks and complications were substantially lower in 3D plans when compared with the 2D plans.

CONCLUSIONS This group of patients treated with 3D planning illustrates the potential usefulness of this new technology in the management of patients with head and neck infratemporal malignancies. This is especially applicable to the pediatric population in which maturing tissuesin the head and neck, such as the brain, skull, and facial musculature, may have a substantial risk for morbidity with high-dose irradiation (8, 15). Many radiosensitive structures such as the optic nerve, eyes, salivary glands, brain, and spinal cord are in close proximity to the infratemporal fossa, and 3D planning offers the ability to limit radiation doses to them. This is especially pertinent in the pediatric population where growing normal structures (i.e., the brain) may have lower thresholds for late effects. 3D conformal radiation therapy has a demonstrated advantage in the treatment planning of children with head and neck rhabdomyosarcomas.The likelihood of lowering the risk of late effects, while maintaining or enhancing the possibility of local control, has substantial merit.

REFERENCES 1. Bentzen, S.; Jessen,K. A.; Jorgensen,J.; Sell, A. Impact

of CT basedtreatment planning on radiation therapy of carcinomaof the bladder.Acta Radiol.Oncol. 23:199-203; 1984.

8.

2. Brizel, H. E.; Livingston, P. A.; Grayson, C. V. Radiotherapeutic applicationof pelvic computedtomography.J. Comput. Assist. Tomogr. 3:453-466; 1979. 3. Drzymala, R. E.; Holman, M.; Yan, D.; Harms, W. B.; Jain, N. L.; Kahn, M. G.; Emami, B.; Purdy, J. A. Integrated

9.

software tools for the evaluation of radiotherapy treatment plans (Abst). Int. J. Radiat. Oncol. Biol. Phys. 24(1):157;

10.

1992. 4. Drzymala, R. E.; Mohan, R.; Brewster,L.; Chu, J; Goitein, M; Harms, W.; Urie, M. Dose volume histograms. Int. J. Radiat. Oncol. Biol. Phys. 21:71-78; 1991. 5. Gasparini, M.; Lombardi, F.; Gianni, M. C.; Massimino, M.; Gandola, L.; Fossati-Bellani, F. Questionable role of CNS radioprophylaxis in the therapeutic management of childhood rhabdomyosarcoma with meningeal extenuation. J. Clin. Oncol. 8:1854-1857; 1990. 6. Glatstein, E.; Lichter, A.; Fraass, B. A.; Kelly, B. A.; Geijin, J. V. D. The imaging resolution and radiation oncology: The use of CT, ultrasound and NMR for localization, treatment planning and treatment delivery. Int. J. Radiat. Oncol. Biol. Phys. 11:299-314; 1985. 7. Goitein, M.; Wittenberg, J.; Mendiondo, M.; Doucette, J.; Friedberg, C.; Ferrucci, J.; Gunderson, L.; Linggood, R.; Shipley, W. U.; Fineberg, H. V. The value of CT scanning

11.

12.

13.

in radiation therapy treatment planning: A prospective study. Int. J. Radiat. Oncol. Biol. Phys. 5: 1787- 1798; 1979. Heyn, R.; Ragab, A.; Raney, B.; Ruymann, F.; Tefft, M.; Lawrence, W.; Soule, E.; Maurer, H. M. Late effects of therapy in orbital rhabdomyosarcoma in children. Cancer 57:1738-1743; 1986. Holday, P.; Hodson, N. J.; Husband, J.; Parker, R. P.; Mcdonald, J. S. Computed tomography applied to radiotherapy treatment planning technique and results. Radiology 133:477-482; 1979. ICRU, Report No. 50, Prescribing, recording, and reporting photon beam therapy. Washington, DC: International Commission on Radiation Units and Measurements; 1993. Maurer, H. M.; Gehan, E. A.; Beltangady, M.; Crist, W.; Dickman, P. S.; Donaldson, S. S.; Fryer, C.; Hammond, D.; Hays, D. M.; Herrmann, J.; Heyn, R.; Jones, P. M.; Lawrence, W.; Newton, W.; Ortega, J.; Ragab, A. H.; Raney, R. B.; Ruymann, F. B.; Soule, E.; Tefft, M.; Webber, B.; Wiener. E.; Wharam, M.; Vietti, T. J. The intergroup rhabdomyosarcoma study II. Cancer 7 1: 1904- 1922; 1993. Mills, P. H.; Fuchs, H.; Pizer, S. M. A tool for image display and contour management in a windowing environment. Newport Beach, CA: Proceedings of Medical Imaging III: Image Capture and Display; 1989:132- 142. Perez, C. A.; Graham, M. V.; Emami, B.; Gerber, R. L.; Matthews, J. W.; Harms, W. B.; Purdy, J. A. A fully integrated CT-simulator: Conceptual design and clinical applications (Abst). Int. J. Rad. Oncol. Biol. Phys. 27(1):140; 1993.

Conformal radiotherapy in pediatric sarcoma 14. Purdy, J. A.; Harms, W. B.; Matthews, J. W.; Drzymala, R. E.; Emami, B.; Simpson, J. R.; Manolis, J. Advances in 3-D radiation treatment planning systems: Room-view display with real time interactivity. Int. J. Rad. Oncol. Biol. Phys. 27(4):933-944, 1993. 15. Razek, A. A.; Perez, C. A.; Lee, F. A.; Ragab, A. B.; Askin, F.; Bietti, T. Combined treatment modalities of rhabdomyosarcoma in children. Cancer 39:2415-2421; 1977. 16. Tefft, M.; Femandez, C.; Donaldson, M.; Newton, W.; Moon, T. E. Incidence of meningeal involvement by rhab-

0 J.

M.

MICHAELSKI

et al.

991

domyosarcoma of the head and neck in children. Cancer 42~253-258; 1978. 17. Tepper, J. E. Role of radiation therapy in the management of patients with bone and soft tissue sarcomas. Semin. Oncol. 16:281-288; 1989. 18. Thornton, A. F.; McLaughlin, P. W.; Sandler, H. M.; Urba, S. G.; Wolf, G. T. Contributions of 3-dimensional volumetrics and treatment planning to irradiation of advanced paranasal sinus tumors (Abst). Int. J. Radiat. Oncol. Biol. Phys. 21(1):145; 1991.