lnt. .I. Radiation Oncology Biol. Phys., Vol. 37, No. 2, pp. 445453, 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. Ail rights reserved 0360.3016/97 $17.00 + .oO
PI1 SO360-3016(96)00459-2
ELSEVIER
l
Physics Contribution CONSISTENCY
OF THREE-DIMENSIONAL PLANNING TARGET ACROSS PHYSICIANS AND INSTITUTIONS
VOLUMES
CASE H. KETTING, M.D.,* MARY AUSTIN-SEYMOUR, M.D.,+ IRA KALET, PH.D.,? JONATHAN UNGER, M.S.,* SHARON HUMMEL, R.T.T.+ AND JON JACKY, PH.D.+ *Department of Radiation Medicine, Loma Linda University Medical Center, Loma Linda, CA; ‘Department of Radiation Oncology, University of Washington, Seattle, WA; and ‘RSA, Inc., Burlington, MA Purpose: Three-dimensional treatment planning depends upon exact and consistent delineation of target volumes. This study tested whether different physicians from different institutions vary significantly in their creation of planning target volumes (PTVs). Methods and Materials: Eight physicians from three different institutions created partial planning target volumes for nine clinical test cases. Their target volumes were evaluated qualitatively and quantitatively. Quantitative results were tested for significant differences. Results: Qualitative analysis showed the physicians to vary in (a) the margin placed around the clinical target volume, (b) the margin used near criticai structures, and (c) handling of concavities in the clinical target volume. Quantitative analysis showed these variations to result in statistically significant differences in the measured volume of the physicians’ planning target volumes. Conclusions: Individual physicians and institutions differ significantly in their creation of planning target volumes, suggesting individual and institutional differences in the working definition for the PTV. Implications of this fact are discussed, along with areas where standardization can he improved. 0 1997 Elsevier Science Inc. Planning target volume, ICRU-50, Three-dimensional treatment planning.
The objective of radiation therapy is to deliver a lethal dose of radiation to disease with minimal dose to surrounding structures. When drawing portals for conventional radiation therapy, however, radiation oncologists always include a margin around the areas of known or presumed disease. This margin represents a tacit recognition that several factors degrade the spatial accuracy of radiation delivery over a course of treatment. Treatment setups, for example, are never exactly reproduced from day to day; consequently, location of disease relative to the treatment beam isocenter will vary. In addition, patients must breathe during treatments, and they may fidget or shift as well. These movements also affect relative position of disease, and thus, accuracy of treatment delivery. In recent years, as three-dimensional imaging modalities have become more widely available, the concept of “area of disease” has been refined to “disease volume.” This has led to formal definitions of treatment volumes and nomenclature. In 1978, the International Commission on Radiation Units (ICRU) published ICRU Report 29,
which provided some of these definitions (5). The International Commission on Radiation Units Report 50, published in 1993, advanced and improved this work (6). Under ICRU 50, macroscopic disease apparent on clinical exam or imaging is designated the Gross Tumor Volume (GTV). The Clinical Target Volume (CTV) incorporates this volume and adds to it volumes in which gross disease is not present, but in which the disease is expected to be microscopically present based upon natural history and failure patterns. For example, the CTV for epidermoid cancers of the nasopharynx includes the draining lymph nodes of the head and neck, even if they are not clinically abnormal. Over a course of radiotherapy, several CTVs may be defined and treated. This is most frequently done when using a shrinking field technique in which the final CTV is identical to the GTV. The ICRU 50 report goes on to address the issue of margin by defining the Planning Target Volume (PTV). This volume is an enlargement of the CTV, which accounts for motion of the target volume during and between fractions of radiation. As described above, such motion is attributed to daily variation in treatment setup and to phys-
Reprint requests to: Mary Austin-Seymour, M.D., Department of Radiation Oncology, University of Washington, Box 356043, Seattle, WA 98 195-6043.
Grant Number LM04174-08 from the National Library of Medicine. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the National
Acknowledgements-This
Cancer Institute or the National Library of Medicine. Accepted for publication 16 September 1996.
INTRODUCTION
publication
was supported
by Grant
NumberNOl-CM97566 from the National CancerInstitute and 445
446
I. J. Radiation Oncology 0 Biology
l
Physics
Volume 37, Number 2, 1997
Table 1. Clinical characteristics of the test cases presented to the physicians Site
Histology
Lung 1 Lung 2 Lung 3 Lung 4 Lung 9 NP 1 NP la NP 3 NP 4
adeno adeno adeno adeno squamous squamous squamous squamous
R L L L R
Subsite
Stage
Immobilization
hilum hilum upper lobe upper lobe lower lobe
T4N2 T4N2 T2NO T3NO T2N2 T4NO T4NO T2NO T4N2
none none none none none mask none mask mask
Notes
fixed to chest wall
NP = nasopharynx, adeno = adenocarcinoma. iological causes such as respiration and other patient movements. Notably, it does not include margin added to account for radiation beam penumbra. A considerable number of studies have been published quantifying the various factors contributing to PTV margin at different anatomic sites, particularly head and neck (3, 4, 14, 16, 17), breast (26, 28), thorax (15, 16, 20, 21, 22, 27), and pelvis (16). However, studies of how consistently physicians utilize this information in drawing PTVs have been largely lacking from the literature. A study of brain tumor treatment by Leunens ef al. has shown considerable variability between different physicians’ definition of planning target volume as evidenced by field margins drawn on simulation films (11). However, in this study, the margin used for creating the planning target volume was specified by protocol, while physicians were required to independently determine the tumor volume and map this volume from the computed tomography (CT) studies to the simulation film. Thus, the great majority of the variability seen in PTV outlines was due to these latter factors. Furthermore, the study addressed only the two-dimensional case of conventional radiation treatment planning. With the advent of three-dimensionally planned conformal radiation treatments, precise and consistent definition of PTVs becomes even more important. In conforma1 plans, the high dose volumes are smaller owing to tighter tolerances on patient setups and improved shaping of the dose volume to the clinical target volume. Consequently, if PTV margins are too narrow, the risk of underdosing rises even more rapidly than it would for conventional treatment plans. Conversely, conformal treatment plans frequently specify higher doses than conventional plans, thus magnifying the risk of adverse late sequelae when overly generous PTVs are used. In an effort to evaluate current practice in defining PTVs, we isolated the task of generating three-dimensional planning target volumes for several clinical cases, presented this task to several different physicians, and tested agreement between the PTVs they drew. Our objectives were to determine if physicians differed significantly from each other and to define sources of variation in practice.
METHODS
AND
MATERIALS
Data collection The experiment consisted of presenting several clinicians with clinical cases for which they were asked to generate PTVs. These were collected and compared for consistency. Physicians from three academic institutions active in three-dimensional treatment planning (University of Washington, Washington University, and University of North Carolina at Chapel Hill) participated in the study. The eight physicians selected represented 81 years of clinical practice (range: 1 to 18 years). Each physician was experienced in three-dimensional treatment planning, although not all routinely drew PTVs in their clinical practice. Four of the physicians drew PTVs on all nine test cases, while the remaining four produced PTVs on only two cases. Each physician was given identical instruction, including a written definition of the PTV, regarding the required task. The physicians did not see or discuss the PTVs of other physicians before generating their own. The nine test cases consisted of five lung adenocarcinomas and four nasopharyngeal squamous carcinomas. In addition to clinical data summarized in Table 1, the physicians were provided with CT scans containing outlines of the clinical target volume and critical (radiation sensitive) normal structures. Because the physicians were being asked to generate PTVs for the final, or boost, phase of treatment in which the CTV was identical to the gross tumor volume, CTVs were not drawn separately. For use by the physicians, the GTV was mapped from the patients original CT to a clinically normal CT. If the original scans showing disease had been used, it is highly likely that each physician would have based her PTV on her own assessment of the extent of gross disease rather than on the provided GTV outlines. The resulting failure to isolate the task of generating PTVs from that of determining extent of disease would have nullified the study’s conclusions. The CT scans and contours for each case were distributed to the participating physicians in electronic form. These were then translated into each institution’s native treatment planning software format. In this way, each phy-
447
Consistency of planning target volumes 0 C. H. KETTING et al.
sician could use familiar software for drawing the PTV contours. To decreaseworkload and increase compliance, the physicians were asked to draw PTV contours on only three randomly designatedslices of each CT. This resulted in generation of 132 contours (four physicians x nine patients X three slices per patient + four physicians X two patients X three slices per patient). The contour data was returned to University of Washington for analysis, still in electronic form.
Data analysis Qualitative analysis was achieved by simply viewing the various physicians’ PTV contours and categorizing the differences among them. Several methods were considered for quantitative analysis. From a computational standpoint, the easiestwas to calculate the areasenclosed by the PTV contours and derive volume from the product of area and CT slice thickness. Volumes were then compared among physicians. PTVs of equivalent volume might actually differ in shape, but those of differing volume could not be equivalent. The qualitative analysis revealed, in part, that physicians tended to generate PTVs by simply enlarging the tumor contour by some radial margin. This finding suggested an additional method for quantitative analysis. Referring to Fig. 1, it is evident that the area between the tumor and PTV contours may be estimated by the product of the physician-selected margin and the mean circumference of the two contours. The calculation is analogousto determining rectangular area from the product of length and width. By reversing the calculation, therefore, a physician’s selected margin may be estimated from the areas and circumferences of the tumor contour and the physician’s PTV contour. That is, the difference between the area of the PTV contour and the GTV contour, divided by the mean of the two contours’ circumferences, gives an estimate of the average radial margin between the contours. This “radial margin estimate” was computed for each physician’s PTV contours. Analysis of variance (ANOVA) was employed to test for significance of the differences found between the physicians. This statistical technique quantifies variability in the data set and partitions it between the factors that are thought to influence the data values. For example, in this experiment, a planning target volume might be influenced by which physician drew it and which clinical casehe was working on. In addition, the volumes would vary with the size of the underlying gross tumor. Using ANOVA, a statistical test is applied to the variability attributable to each factor. This test determineswhether the factor’s variability significantly exceeds what would be expected from the inherent random variability of the data set. If so, that factor is deemed to have an effect on the data. In this case, if the variability attributable to the use of different physicians is significant, the physician’s PTVs cannot be said to agree with each other.
mean circumference Fig. 1. Cross-sectionof clinical target volume (CTV) andplanning target volume (PTV). The radial margin estimate,an approximate average of the distance between the two contours, can be derived by dividing the area between the two contours by their mean circumference.
RESULTS Qualitative
analysis
Qualitative analysis showed the physicians to differ in three main areas.First, physicians varied in the amount of margin they placed around the gross tumor volume (Figs. 2 and 3) Second, their shaping of the margin near critical structures varied significantly. For example, in Fig. 3, some physicians drew the PTV contour closely abutting the spinal cord, while others gave a margin on the cord as well as on the gross tumor. Finally, physicians varied in their handling of concavities in the gross tumor volume contour. Some would use a corresponding concavity in their PTV, while others drew strictly convex PTVs. A significant oversight common to all the physicians was also noted. All physicians failed to adequately account for margin in the dimension orthogonal to the plane of the CT. This problem was particularly acute in the lung cancer cases, where any reasonable estimation of respiratory movement would result in portions of the grosstumor volume falling outside most physicians’ PTVs during a portion of the respiratory cycle (Fig. 4).
Quantitative
analysis
Quantitative results are listed in Tables 2 and 3. Table 2 lists the measuredvolume included in the PTV by each physician for each case. These volumes reflect only the three slices outlined; the values would be much larger for a complete set of PTV contours. Figure 5 plots the data. The increasing volumes across the graph reflect the increasing sizes of the underlying gross tumor volumes.
448
I. J. Radiation Oncology 0 Biology 0 Physics
Volume 37, Number 2, 1997
Fig. 2. One computed tomography slice from a lung cancer case showing contours surrounding the red clinical target volume.
Table 3 lists the mean of the three radial margin estimates computed for each case from each physician’s set of three contours. These data, plotted in Fig. 6, clearly show the larger margins appropriately used for casesof lung cancer. Nevertheless, there is still considerable variability among the different physicians’ radial margins.
Statistical analysis Statistical analysis was restricted to the four physicians who addressed all nine test cases. Analysis of variance found both physician and clinical case to be highly significant factors affecting PTV volume and radial margin (p < 0.005). The effect due to caseis, of course, expected, as the clinical scenario and underlying gross tumor volume differed among cases. However, the effect due to physician implies that the physicians differed significantly from each other in their PTVs. Some subsetanalyseswere investigated. First, multiple paired comparisons between the four physicians, using Bonferroni’s method, were carried out. This test is necessarily lesspowerful than ANOVA; however, significant differences were detected in the pairs (MD 2, MD 1) and (MD 3, MD 1). A second analysis was carried out to determine if the effect of clinical case on radial margin could be entirely explained by the larger margins used for the lung cancers
the physicians’
planning
target volume
relative to the nasopharyngeal cancers. An analysis nesting clinical case as a factor within site (lung vs. nasopharynx) found both factors to be significant 0, < 0.005). This suggeststhat physicians based their choice of radial margin on clinical factors beyond just site of disease. Finally, the hypothesis was raised that differences between physicians’ PTVs were attributable to differences in efficacy of immobilization techniques at each institution. There was insufficient data to test this hypothesis statistically. However, referring to Fig. 7, it is evident that the lung PTVs of the four physicians from University of Washington differed significantly, as did those of the two physicians from University of North Carolina. Some discrepancy is also evident between nasopharynx PTVs drawn by physicians at the sameinstitution.
DISCUSSION In summary, the physicians’ PTVs differed significantly, as determined by volume measurementsand radial margin estimates. Qualitatively, this was seen to be due to use of different margins around the gross tumor and to different handling of nearby critical structures and tumor concavities. Furthermore, all physicians failed to adequately consider margin in the direction orthogonal to the CT image plane.
Consistency of planning target volumes 0 C. H. KETTINC er al
Fig. 3. One computed tomography slice from a nasopharynx volume contours surrounding the red clinical target volume.
Incorporation of an adequatePTV in three-dimensional treatment planning is important to tumor control, treatment complications, and the comparison of treatment plans. With respect to tumor control, the concept of marginal relapse is well established in patient series (1, 7, 9, 10, 13, 18, 29). Kim et al. related marginal relapse to inadequate PTV in conventionally treated cervical cancer casesby showing a 30% decreasein local control for Stage IB and II patients whose margins were judged inadequate (8). In articles discussing conformal three-dimensional treatment planning, Rudat et aZ. (22) and Urie et al. (25) suggest that the rapid falloff of dose outside the target volume achieved by this technique will only exacerbate the problem of marginal relapse where inadequate PTVs are used. Patient movement and setup errors displace the tumor volume relative to the dose volume. Therefore, the only way to assurefull dose to the tumor volume is to deliver full dose to the PTV, which is designed to account for this displacement. Urie et al. (25) examined how well the dose from a selection of three-dimensional treatment plans covered the PTV. In reviewing plans submitted by several participating institutions, they noted that, despite
cancer case showing
the physicians’
449
planning
target
agreements to treat the PTV (termed mobile target volume in the paper), some institutions’ treatment plans targeted only the gross or clinical disease volume. As a consequence, several plans underdosed from 12 to 50% of the PTV, while the best plans kept this value under 6% for all cases. Rudat et al. (22) studied the effects of positioning errors in 3D conformal treatments. They planned therapy for several prostate and esophagealcancer cases,specifying 80% isodose to the PTV. The ideal dose distributions resulting from their plans were then modified to reflect the effects of patient motion, which they had quantified in a separateanalysis. Interestingly, this modification had minimal effect on the dose-volume histograms (DVH) for their esophageal cancer cases.The plans for these cases were not highly conformal and were, thus, less sensitive to motion. In contrast, the DVHs for the prostate cancers, whose dose volumes conformed more closely to the PTV, were visibly affected. These findings were also seenin the calculated tumor control probabilities (TCP) for the different sites.TCPs for the esophagealcancer casesdropped an average of 1.9% (range: 0 to 6.3%), vs. 5.3% (range: 0.6 to 11.5%) for prostate cancer.
I. J. RadiationOncology0 Biology0 Physics
450
Lateral margin adequate, vertical margin too narrow
Fig. 4. A diseasevolume is shownin cross-sectionorthogonal to the planeof the imageslices.Seenon edge,thedisease volume contoursbecomehorizontallines. For eachslice,the widestextent of the planningtarget volume (PTV) contouris designated by two smallcircles.The PTV contoursshownaredrawn using a two-dimensionalperspectiveonly. Consequently,the lateral(in plane) marginsare adequate,while the vertical marginsare in severalplacesinsufficient to completelyencompass the disease when it is displacedvertically.
Several factors should be emphasizedin interpreting the findings of the latter study. First of all, with ideal planning of conformal therapy using accurate PTVs, no effect from patient motion should have been seen. As noted previously, in the ideal case, the PTV will encompassall expected movements and will receive full dose, thus ensuring that the diseasereceives full dose over the course of therapy. The changes in this study resulting from the inclusion of patient motion presumably stem from the PTV definition used, which was not clear from the article, and coverage of the PTV by the 80%, rather than lOO%, isodoseline. Secondly, only patient movement was addressed by this study. Movement of tumor relative to the patient, an occurrence well recognized for both esophageal(23) and prostate (19, 24) cancers, was not included. Its inclusion would likely have accentuated the findings.
Volume37,Number2, 1997 Finally, becausethe assumptionsunderlying TCP calculations remain inexact, the TCP findings of this study should be regarded as qualitative rather than quantitative. The study suggests that tumor control will decrease if treatment plans do not sufficiently addressthe issue of patient movement. The magnitude of this effect may average from 2 to 5% for esophageal and prostate tumors planned similarly to those in this study, but may be greater or less. The effect will be greater, of course, for some individual patients, and would certainly be greater if the PTV is ignored entirely in the planning process. An accurate and consistent definition of PTV is also important from the standpoint of treatment complications. Normal tissue complication probability (NTCP) is dependent upon the volume of normal tissue treated; a small increment in PTV margin can significantly affect that volume. For example, assuming spherical disease,a change in margin from 0.5 to 1.0 cm. (Table 3) around a 2.0 cm. tumor results in a 47 cc. or 73% increase in PTV. For a 4.0 cm tumor, the figures are 142 cc and 37%. These values do not include the increase in irradiated volume resulting from the enlarged entering and exiting treatment beams. Finally, becausePTVs represent the ideal dose region, calculation of dose-volume histograms will be affected by inconsistency in definition of PTV. Comparison of DVHs originating with different institutions, physicians, or protocols will be inadequate as a means of comparing treatment plan quality if the PTVs are defined differently. It is apparent from our study that, despite the significant advance represented by the ICRU 50 definitions, substantial differences in PTV delineation exist between individual physicians and institutions. At a minimum, the varied handling of the different clinical and anatomic situations presented in this study would lead to difficulties in evaluating and comparing treatment plans. At worst, patient outcomes could be compromised. Our findings suggest three main areas in which standardization would lead to more consistent and reproducible practice. First, there appeared to be fundamental disagreement on the margin required to create an adequatePTV for each
Table 2. Aggregatevolume (cc) of the partial planningtarget volumesdrawnby the participatingphysicians CTV
MD 1
Lung 1 Lung 2 Lung 3 Lung 4 Lung 9 NP 1 NP la
72.5 22.6 32.1
142.6 70.3 93.7
31.2 13.1 13.3 13.3
107.6
NP3 NP4
4.5 17.2
58.0 23.0 40.8
10.0 29.7
MD2
MD3
MD4
MD5
MD6
180.5 90.4 92.8
117.3
41.8
182.8
173.6
111.7 112.1 120.4
100.5 119.4
110.4
89.3 30.1 41.2 15.9 36.2
81.6 41.6 29.5 19.1 41.9
61.2 35.6 41.9 17.6 38.5
95.9
Volume of the underlying clinical target volume (CTV) is shownfor reference. NP = nasopharynx.
MD7
MD8
113.7
87.0
71.6
39.5
34.1
34.8
Consistency of planning target volumes 0 C. H. KETTING et al. Table 3. Mean (cm) of the three radial margin estimates derived from the physicians’ MD2
MD 1 Lung 1 Lung 2 Lung 3 Lung 4 Lung 9 NP 1 NP la NP 3 NP4
1.21 1.21 1.43 1.38 1.38 0.45 1.11 0.46 0.53
MD3
1.77 2.00 1.75 1.60 2.04 0.74 1.13 0.85 0.77
MD4
1.63 1.81 1.46 1.60 1.87 1.13 0.71 1.01 0.94
MD5
1.74 1.62 1.40 1.41 1.45 0.87 1.17 0.93 0.83
4.51
planning
target volume contours
MD6
MD7
MD8
1.84
1.78
.1.28
0.99
0.94
0.87
0.71
0.70
NP = nasopharynx.
clinical situation. Standards for three-dimensional PTV margin, similar to those existing for conventional therapy, may best be established and disseminated in the radiation oncology community by treatment protocols. This process has begun with the cooperative trial on conformal treatment of prostate cancer (30). As work continues, it is likely to become apparent that margins should be anisotropic, e.g., they may vary according to orthogonal dimension (anteroposterior, superior-inferior, or left-right), and they may also vary according to subsite within the CTV (12, 21). The method used to deal with concavities in the underlying clinical or gross diseasevolumes also contributed to differences between physicians. Ideally, concavities should be preserved in the PTV. Practically speaking, however, becauseconcave dosedistributions are relatively rare in conformal teletherapy plans, this may not be an
urgent consideration for many institutions. Those utilizing three-dimensional planning for brachytherapy or charged particle beams may produce concave dose distributions with relative ease. Consequently, these institutions will find PTVs that preserve concavities indispensableto representing, via dose-volume histograms and DVH-based metrics, the true quality of their treatment plans. Handling of nearby critical structures, the final area contributing to discrepancies posesa similar problem. To accurately represent the ideal plan, the PTV should not be modified to account for nearby critical structures such as spinal cord, rectum, or small bowel. This approach will most accurately represent, via DVHs calculated over the PTV, the compromises in ideal dose volume (and, thus, potential for cure) that are necessitatedby nearby critical
Physicians’ Mean Radial Margins Volumes of Physician PWs 200
T
180 + I
160 140
40
1 *MD4
20
0c---me--NP3
NPl
/ NP4
NPla
Lung9
Case
Lung2
I Lung3
Lung4
Lung1
Identifier
Fig. 5. Volumes of the partial planning target volumes (PTVs) drawn by each physician on each case. Differences between physicians’ PTVs are evident.
0.0 -I NP3
NPl
1
I I NP4 NPla Lung9 Lung2 Lung3 Lung4 Lung1 Case Identifier
Fig. 6. Mean of the radial margin estimates for each physician and each case. Larger margins were, in general, used for lung cancer cases.
I. J. Radiation Oncology 0 Biology 0 Physics
452
PlVs
50 definitions,
by Institution
80 --
20
t
0 I Univ Wash
Univ Univ Wash Wash
Univ Wash
Institutional
Volume 37, Number 2, 1997
I Wash UNC UNC U Chapel Chapel Hill Hill Source of PTV Wash U
Fig. 7. Volume of the planning target volumes (PTVs) for the two test cases as addressed by all eight physicians. Data are grouped by institution, showing disparities within as well as between institutions.
structures. In our study, most physicians excluded critical structures from the PTV, frequently including a margin around the structure. While not in conflict with the ICRU
this practice
engenders
an approach,
dis-
tinct from that above, in which a series of shrinking PTVs are generated for use as critical structures reach tolerance dose. Using either approach, organ movements and setup errors pertaining to the critical structure, similar to those applied to the CTV in generating the PTV, must ultimately be considered. Finally, our study found that physicians fail to consider out-of-plane movement when drawing PTVs. Instead, it appears that physicians generate PTVs using two-dimensional reasoning similar to that used when drawing radiation portals around gross disease on simulation films. Although this practice was common to all physicians and, thus, did not contribute to discrepancies between the PTVs, a correct geometric definition of PTVs would require true three dimensional reasoning. This final issue raises the question of how to consistently and efficiently generate PTVs in the clinical environment. Accounting for out-of-plane movement requires simultaneous consideration of the disease contours superior and inferior to the drawing plane currently being addressed. This is not a simple task for a human operator to carry out and would almost certainly further degrade efficiency in a process that is already time and labor intensive (2). However, it seems likely that, with the increasing use of computers, the task of producing PTVs may be made automatic or semiautomatic, thus enhancing consistency, repeatability, and efficiency. For this to become the case, however, physicians will need to reach consensus on a more exact definition of PTVs in clinical practice.
REFERENCES 1. Austin, J. P.; Urie, M. M.; Cardenosa, G.; Munzemider, J. E. Probable causes of recurrence in patients with chordoma and chondrosarcoma of the base of skull and cervical spine. Int. J. Radiat. Oncol. Biol. Phys. 25:439444; 1993. 2. Dowsett, R. J.; Galvin, J. M.; Cheng, E.; Smith, R.; Epperson, R.; Harris, R.; Henze, G.; Needham, M.; Payne, R.; Peterson, M. A.; et al. Contouring structures for 3-dimensional treatment planning. Int. J. Radiat. Oncol. Biol. Phys. 22:1083-1088; 1992. 3. Dunscombe, P. B.; Fox, K.; Loose, S.; Leszczynski, K. The investigation and recitification of field placement errors in the delivery of complex head and neck fields. Int. J. Radiat. Oncol. Biol. Phys. 26:155-161; 1993. 4. Huizenga, H.; Levendag, P. C.; De Porre, P. M. Z. R.; Visser, A. G. Accuracy in radiation field alignment in head and neck cancer: A prospective study. Radiother. Oncol. 11:181-187; 1988. 5. ICRU Report 29. Dose specification for reporting external beam therapy with photons and electrons. International Commission on Radiation Units and Measurements, Washington, DC; 1978. 6. ICRU Report 50. Dose specification for reporting external beam therapy with photons and electrons. International Commission on Radiation Units and Measurements, Washington, DC; 1978. (ICRU Report issued September 1993). 7. Keus, R. B.; Pontvert, D.; Brunin, F.; Jaulerry, C.; Bataini, J. P. Results of irradiation in squamous cell carcinoma of
8.
9.
10.
11.
12.
the soft palate and uvula. Radiother. Oncol. 11:3 11-3 17; 1988. Kim, R. Y.; McGinnis, L. S.; Spencer, S. A.; Meredith, R. F.; Jennelle, R. L.; Salter, M. M. Conventional four-field pelvic radiotherapy technique without computed tomography-treatment planning in cancer of the cervix: Potential geographic miss and its impact on pelvic control. Int. J. Radiat. Oncol. Biol. Phys. 31:109-l 12; 1995. Kuten, A.; Ben-Shahar, M.; Epelbaum, R.; Haim, N.; Cohen, Y.; Robinson, E. Results of radiotherapy in stage I to II extranodal nonHodgkin’s lymphoma of the head and neck. Strahlenther. Onkol. 165:578-583; 1989. Leibel, S. A.; Wara, W. M.; Hill, D. R.; Bovill, E. G., Jr.; de-Lorimier, A. A.; Beckstead, J. H.; Phillips, T. L. Desmoid tumors: Local control and patterns of relapse following radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 9: 11671171; 1983. Leunens, G.; Menten, J.; Weltens, C.; Verstraete, J.; vander-Schueren, E. Quality assessment of medical decision making in radiation oncology: Variability in target volume delineation for brain tumors. Radiother. Oncol. 29: 169-175; 1993. LoSasso, T.; Chui, C. S.; Kutcher, G. J.; Leibel, S. A.; Fuks, Z.; Ling, C. C. The use of a multi-leaf collimator for conformal radiotherapy of carcinomas of the prostate and nasopharynx. Int. J. Radiat. Oncol. Biol. Phys. 25:161-170; 1993.
Consistency of planning target volumes 0 C. H. KETTING et al. 13. Lusinchi, A.; Wibault, P.; Marandas, P.; Kunkler, I.; Eschwege, F. Exclusive radiation therapy: The treatment of early tonsillar tumors. Int. J. Radiat. Oncol. Biol. Phys. 17:273-277; 1989. 14. Marks, J. E.; Haus, A. G.; Sutton, H. G.; Griem, M. L. Localization error in the radiotherapy of Hodgkin’s and malignant lymphoma with extended mantel fields. Cancer 34:8390; 1974. 1.5. Marks, J. E.; Haus, A. G.; Sutton, H. G.; Griem, M. L. The value of frequent treatment verification films in reducing localization error in the irradiation of complex fields. Cancer 37:2755-2761; 1976. 16. Michalski, J. M.; Wong, J. W.; Gerber, R. L.; Yan, D.; Cheng, A.; Graham, M. V.; Renna, M. A.; Sawyer, P. J.; Perez, C. A. The use of on-line image verification to estimate the variation in radiation therapy dose delivery. Int. J. Radiat. Oncol. Biol. Phys. 27:707-716; 1993. 17. Mitine, C.; Leunens, G.; Verstraete, J.; Blanckaert, N.; VanDam, J.; Dutreix, A.; van-der-Schueren, E. Is it necessary to repeat quality control procedures for head and neck patients?. Radiother. Oncol. 21:201-210; 1991. 18. Munzenrider, J. E.; Verhey, L. J.; Gragoudas, E. S.; Seddon, J. M.; Urie, M.; Gentry, R.; Bimbaum, S.; Ruotolo, D. M.; Crowell, C.; McManus, P.; et al. Conservative treatment of uveal melanoma: Local recurrence after proton beam therapy. Int. J. Radiat. Oncol. Biol. Phys. 17:493498; 1989. 19. Pickett, B.; Roach, M., III; Verhey, L.; Horine, P.; Malfatti, C; Akazawa, C.; Dea, D.; Varad, B.; Rathbun, C.; Phillips, T. L. The value of nonuniform margins for six-field conforma1 irradiation of localized prostate cancer. Int. J. Radiat. Oncol. Biol. Phys. 32:211-218; 1995. 20. Reinstein, L. E.; Pai, S.; Meek, A. G. Assessment of geometric treatment accuracy using time-lapse display of electronic portal images. Int. J. Radiat. Oncol. Biol. Phys. 22: 1139-l 146; 1992. 21. Ross, C. S.; Hussey, D. H.; Pennington, E.; Stanford, W.; Doombos, J. F. Analysis of movement of intrathoracic neoplasms using ultrafast computerized tomography. Int. J. Radiat. Oncol. Biol. Phys. 18:671-677; 1990.
453
22. Rudat, V.; Flentje, M.; Oetzel, D.; Menke, M.; Schlegel, W.; Wannenmacher, M. Influence of the positioning error on 3D conformal dose distributions during fractionated radiotherapy. Radiother. Oncol. 3356-63; 1994. 23. Smoron, G. L.; O’Brien, C. A.; Sullivan, C. A. Tumor localization and treatment technique for cancer of the esophagus. Radiology 111:735-736; 1974. 24. Ten-Haken, R. K.; Forman, J. D.; Heimburger, D. K.; Gerhardsson, A.; McShan, D. L.; Perez-Tamayo, C.; Schoeppel, S. L.; Lichter, A. S. Treatment planning issues related to prostate movement in response to differential filling of the rectum and bladder. Int. J. Radiat. Oncol. Biol. Phys. 20:1317-1324; 1001. 2.5. Urie, M. M.; Goitein, M.; Doppke, K.; Kutcher, J. G.; LoSasso, T.; Mohan, R.; Munzenrider, J. E.; Sontag, M.; Wong. J. W. The role of uncertainty analysis in treatment planning. Int. J. Radiat. Oncol. Biol. Phys. 21:91-107; 1991. 26. van-Tienhoven, G.; Lanson, J. H.; Crabeels, D.; Heukelom, S.; Mijnheer, B. J. Accuracy in tangential breast treatment setup: A portal imaging study. Radiother. Oncol. 22:317322; 1991. 27. Weltens, C.; Leunens, G.; Dutreix, A; Cosset, J. M.; Eschwege, F.; van-der-Schueren, E. Accuracy in mantle field irradiations: Irradiated volume and daily dose. Radiother. Oncol. 29: 18-26; 1993. 28. Westbrook, C.; Gildersleve, J.; Yarnold, J. Quality assurance in daily treatment procedure: Patient movement during tangential fields treatment. Radiother. Oncol. 22:229-303; 1991. 29. Willich, N.; Bayer, M.; Krimmel, K.; Lengsfeld, M.; Rohloff, R.; Wendt, T. A ventral mantle technic with dorsolateral mediastinal saturation in the radiotherapy of Hodgkin’s disease. Strahlenther. Onkol. 164:393-401; 1988. 30. 3D-CRT Oncology Group. 3D/OG 94-06: A Phase I/II dose escalation study using three-dimensional conformal radiation therapy for adenocarcinoma of the prostate. Philadelphia, PA: American College of Radiology Operations Center; June 16, 1995.