- Email: [email protected]
Dosimetric comparison of treatment planning systems in irradiation of breast with tangential fields
Int. J. Radiation
Biol.
Phys., Vol. 38, No. 4, pp. 835-842. 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. All rights reserved 0360.3016/97 $17.00 + .OO
PI1 SO360-3016( 97)00078-3
ELSEVIER
l
Oncology
Physics Contribution DOSIMETRIC COMPARISON OF TREATMENT PLANNING SYSTEMS IRRADIATION OF BREAST WITH TANGENTIAL FIELDS
IN
CHEE-WAI CHENG, PH.D.,? INDRA J. DAS, PH.D., F.I.P.S.M.,# WALTER TANG, M.S.,” SHA CHANG, PH.D., 8 JEN-SAN TSAI, PH.D., * CRISTER CEBERG, PH.D.,’ BARBARA DE GASPIE, RT(T),” RAJINDER SINGH, D.A.B.R.,* * DOUGLAS A. FEIN, M.D.# AND BARBARA FOWBLE, M.D.# +Department of Radiation Oncology, University of Arizona, Tucson, AZ, #Department of Radiation Oncology, Fox Chase Cancer Center, Philadelphia, PA, “Medical Radiological Physics Consortium, Minneapolis, MN, “Department of Radiation Oncology, University of North Carolina, Chapel Hill, NC, *Department of Radiation Oncology, New England Medical Center, Boston, MA, *Department of Radiation Physics, Lund University Hospital, Lund, Sweden, “Radiation Oncology, University of Massachusetts Medical Center, University of Massachusetts, Worchester, MA, * *Department of Radiation Oncology, Glendale Memorial Hospital, Glendale, CA Purpose: The objectives of this study ate: (1) to investigate the d&metric differences of the diiferent treatment planning systems (TPS) iu breast h-radiation with tangential fields, and (2) to study the effect of beam characteristics on dose distributions in tangential breast irradiation with 6 MV linear accelerators from different manufacturers. Methods and Materials: Nine commercial and two university-based TPS are evaluated in this study. The computed “ medium” and “large” based on their tomographic scan of three representative patients, labeled as “small”, respective chest wall separations in the central axis plane (CAX) were used. For each patient, the tangential fields were set up in each TPS. The CAX distribution was optimized separately with lung correction, for each TPS based on the same set of optimization conditions. The isodose distributions in two other off-axis planes, one 6 cm cephalic and the other 6 cm caudal to the CAX plane were also computed. To investigate the effect of beam characteristics on dose distributions, a three-dimensional TPS was used to calculate the isodose distributions for three different linear accelerators, the Varian Clinac 6/100, the Siemens MD2 and the Philips SL/7 for the three patients. In addition, dose distributions obtained with 6 MV X-rays from two different accelerators, the Varian Clinac 6/100 and the Varian 21OOC, were compared. Results: For all TPS, the dose distributions in all three planes agreed qualitatively to within ? 5% for the “small” and the “medium” patients. For the “large” patient, all TPS agreed to within ? 4% on the CAX plane. The isodose distributions in the caudal plane differed by ? 5 % among all TPS. In the cephalic plane in which the patient separation is much larger than that in the CAX plane, six TPS correctly calculated the dose distribution showing a cold spot in the center of the breast contour. The other five TPS showed that the center of the breast received adequate dose. Isodose distributions for 6 MV X-rays from three diierent accelerators diifered by about -i- 3% for the “small” patient and more than -C 5% for the “large” patient. For two different 6 MV machines of the same manufacturer, the isodose distribution agreed to within i- 2% for all three planes for the “large” patient. Conclusion: The differences observed among the various TPS in this study were within + 5% for both the “small” and the “medium” patients while doses at the hot spot exhibit a larger variation. The large discrepancy observed in the off-axis plane for the “large” patient is largely due to the inability of most TPS to incorporate the collimator angles in the dose calculation. Only six systems involved agreed to within -C 5% for all three patients in all calculation planes. The difference in dose distributions obtained with three accelerators from different manufacturers is probably due to the difference in beam profiles. On the other hand, the 6 MV X-rays from two different models of linear accelerators from the same manufacturer have similar beam characteristics and the dose distributions are within k 2% of each other throughout the breast volume. In general, multi-institutional breast treatment data can be compared within a 2 5% accuracy. 0 1997 Elsevier Science Inc. Treatment
planning
systems, Tangential
breast irradiation.
INTRODUCTION Currently, there are over a dozen treatment planning systerns (TPS ) used in the different radiation oncology clin-
its in the U.S. Some of them were developed by commercial vendors, some were developed as joint ventures between universities and commercial vendors, and others were developed solely in-house at some universities. The
Reprint requests to: Chee-Wai Cheng, Ph.D., Department of Radiation Oncology, University of Arizona, Tucson, AZ 857245081. Acknowledgement-We would like to thank Katherine Buchneit
at the Department of Radiation Oncology, Fox Chase Cancer Center, for her editorial help. We would also like to thank the technical help of Mr. Robert Costa of the ADAC, Milpitas, CA. Accepted for publication 3 1 January 1997. 835
836
I. J. Radiation Oncology 0 Biology 0 Physics
Volume 38, Number 4, 1997
Table 1. Treatmentplanningsystems,lung correctionalgorithmsandlinear accelerators
Dplan UNC plan Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Renderplan
Machine
Lung corr. algorithm
Horns(%)
Clinac 6/100 SiemensMD2 Mevatron 67 Varian 2100C SiemensMD2 Mevatron 67 Varian 6OOC/D Philips SLi75 Varian 2300C Varian 2 1OOC Clinac 6/100
Ratio-of-TMR (RTMR) Batho’s RTMR Effective path length (EPL) EPL EPL Batho’s EPL and 1-D convolution RTMR Convolution EPL
103 105 102 102 103 102 103 104 103 103 102
Collimatorangles Yes Yes No No No No Yes Yes No Yes Yes
Field size of 25 x 25 cm at d,,,,.
early commercial TPS were two-dimensional (2D) TPS, later ones are marketed as 2SD (called pseudo-3D in the present study) and the most recent development has led to three-dimensional (3D) TPS. University-based TPS, on the other hand, are in general 3D TPS. The capabilities of 2D, pseudo-3D and 3D treatment planning systems have been discussedpreviously (6) ; however, the dividing line between a 2D system and a pseudo-3D TPS becomes fuzzy with the advancement in computer technology, as some of the vendors upgraded their 2D TPS to allow multiple-slice display, beams-eye-view, wire frame, colorwash display, surface rendering, etc., while leaving the dose calculation algorithm unchanged. For simplicity, we consider any TPS which has the capabilities of displaying multiple computed tomographic (CT) slices, beam’s-eyeview and other three-dimensional (3-D) display options as either a pseudo-3D or a 3D TPS. All other TPS which can handle only the central axis slice and which do not offer the various 3D display options are considered 2D; thus the distinction between a pseudo-3D and a 3D system is in the degree of sophistication of the dose calculation algorithm employed. A true 3D algorithm must take into account the shape of the irradiated volume with 3D divergence effects and with an accurate model for scatter dose integration (with or without blocks and with or without inhomogeneity correction) For example, in one of the 3D planning systems used in this study, the use of collimator angles in a tangential field setup can be incorporated in the dose calculation (3 ) and agreement within t 2% between calculation and ion chamber measurements has been reported (4). The samemay not be true for pseudo-3D algorithms. There is no formal definition as to what constitutes a pseudo-3D algorithm. A most commonly employed dose algorithm in pseudo-3D systemsis to approximate the entire volume by a cylinder with the same cross-sectional view as the slice in question (basically a 2D approach). The “z’‘-dependence (cephalic-caudal) of the doseprofile; however, is included in the dose calculation. A much more sophisticated approach combining precalculated Monte Carlo energy deposition kernels and measured beam data in a water phantom is usedin one of the pseudo-
3D systems ( 1) . The scatter integration is performed over the volume defined by the treatment fields and the patient contours. It is not a true 3D system since the SSD is assumed to be the same on the CAX over the entire field. It is not known how accurate those pseudo-3D systems are compared to true 3D systems when dose distributions are calculated in 3D throughout a treatment volume, especially one which changesvastly in shapein a treatment field. The question we would like to answer in this study is: do all TPS produce dose distributions within 5% of each other throughout a given treatment volume? To answer this question, we have carried out a study to compare isodosedistributions obtained with different TPS throughout a breast volume irradiated with tangential fields. Unlike the irradiation of prostate, lung or abdomen which generally involves fairly regularly shaped external contours throughout the radiation fields, the irradiation of breast with tangential fields involves a complex geometrical setup. The shape of the breast, the presence of lung and the use of collimator angles are often the complicating factors in dose calculations. The sophistication and the accuracy of a dose algorithm is best tested with such a complex volume. In addition to the many different TPS used in treatment planning, there are also a variety of linear accelerators from different vendors used in the clinics. It is generally believed that all linear accelerators of the same energy have similar characteristics in terms of beam profiles, depth doses, etc.; however, the same may not be true if wedges are usedbecauseof the different design of wedges among the different vendors. It should be emphasizedthat the differences in beam characteristics, wedge profiles, etc. between any two machines may not be a problem clinically since each clinic measuresits own beam data for each machine. The differences may complicate situations such as dose comparison between two centers with different TPS and linear accelerators. The objectives of this study are: 1) to study the dosimetric differences of various TPS in breast irradiation with tangential fields, and 2) to study the effect of beam characteristics on dose distributions with 6 MV linear accelerators from different manufacturers. Similar studies
Dosimetric comparison of treatment planning systems in irradiation of breast 0 C.-W.
CHENG
ef al.
837
have been conducted in Europe to investigate the performance of computer systems some of which were European-based and are not available in the U.S. (2, 8).
MATERIALS
AND METHODS
In this study, dose distributions obtained from nine commercial and two university-based TPS are compared. The commercial systems are the Modulex’ and the Focus systems,’ the Target system,2the ROCS3 the Prowess4 the Theraplan5 the TMS6 the Pinnacle,’ and the Render Plan.’ The two university-based systemsare the Dplan developed at the Joint Center for Radiation Therapy in Boston and the 3D treatment planning system developed at the University of North Carolina in Chapel Hill (UNC Plan). A summary of the TPS and the type of linear acceleratorsused in the study is given in Table 1. It is assumedthat all of the systems have been evaluated basedon the criteria for commissioning and quality assurance(7). Three representative patients, labeled as “small,” “medium” and “large” basedon their respective chest wall separationsof 18,2 1, and 28 cm in the central axis plane (CAX) were usedin this study. These three patients were chosento cover the range of chest wall and breast sizes encountered in the clinic. For the treatment planning, CT studiesof these three patients were performed. The external and the lung contours were extracted automatically on each CT slice using Dplan and printed out on a laser printer. The accuracy of scaling the printouts had been tested and was found to be accurate to within a millimeter of the true dimension. The hard copy printouts were sentto the different institutions and digitized manually into the various TPS. In this study, the breast irradiation technique consisted of two opposing isocentric tangential fields. The posterior field edges of the tangential fields were made coincident to minimize the volume of lung inside the fields. For each patient, the tangential fields were set up in each planning system. To facilitate comparison of dose distributions obtained with the different TPS, treatment planning was carried out using 6 MV X-rays only. Since the beam characteristics for 6 MV X-rays may be different for different accelerators, the CAX distribution was optimized separately for each TPS based on the same set of optimization conditions: (a) the dose distributions must be symmetrically distributed, and the medial and lateral dose heterogeneity equalized, (b) the dose at the apex must be equal to or smaller than the medial and lateral dose heterogeneity, and (c) the center of the breast must receive 2 100% of the prescribed dose provided that condition (b) was not violated. The dose normalization point was chosen at the lung-chest wall interface, anterior to the interface and on the perpendicular bisector of the posterior field edges.
Wedge and open beam combination were used for each field to obtain the optimal dose distribution. Lung correction was also used in the dose calculation. The dosimetry of tangential breast irradiation with lung correction has been reported previously (3 ) . Once an optimal distribution on the CAX was obtained, isodose distributions in two off-axis planes, one 6 cm cephalic and the other 6 cm caudal to the CAX plane, were computed. In addition, doses were computed at selected points inside each contour as shown in Fig. 1. The interest points were chosen since they represent the prescription points used in the different clinics (5 >; furthermore, the locations of these points provide a measure of the uniformity of the distribution in the breast. These, together with the hot spots at the apex, medial and the lateral subcutaneous locations characterize the dose distribution within a contour. In this study, a hot spot is defined as a dose region 2 110% of the prescribed dose. The choice of the 110% isodosevalue is somewhat arbitrary and based on the clinical practice in that it is considered as the maximum dose acceptable for a daily fraction size of 180 cGy. For consistency, dose distributions generated in each planning system were sent to the Department of Radiation Oncology, University of Arizona, and analyzed by one of the authors (Cheng). To investigate the effect of the different beam characteristics on dose distributions, three 6 MV linear acceler-
‘Computerized
5TheratronicsInternationalLtd., Kanada,Ontario, Canada.
Medical
System, St. Louis, MO.
‘GeneralElectric Medical System,Milwaukee, WI. 3Radiation Oncology Computer 4Prowess System, Chico, CA.
Systems, Inc., San Diego, CA.
Fig. 1. Schematicdiagramof the centralaxis of a tangentialfield breastirradiation showingdosecalculation points and the hot spots.1-NSABP point at h/3 distance,2-center of breast(h/ 2), 3-isocenter, 4-apex hot spot,5-medial hot spotand6lateralhot spot.
6Ma.rketed by Siemens Medical ‘ADAC, Milpitas, CA.
Systems Inc., Concord, CA.
‘PrecisionTherapy International,North Miami, Beach,FL.
838
I. J. Radiation Oncology l Biology l Physics
Volume 38, Number 4, 1997 Table 2. Comparison of doses at selected points for the “small” patient (a) Cephalic slice, z = +6 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
100 -5 +l 0 0 0 - 1 0 -1 +2 +2
100 0 +l 0 0 0 0 +3 -1 +2 -2
Isocenter 106 -1 -2 -1 -2 -4 0 +1 -1 +2 +2
Apex
Medial
Lateral
108 -3 -5 -3 -5 -6 -2 +2 -3 +2 0
105 -3 -3 +4 -4 -2 -2 -1 -5 +5 +2
105 -3 0 0 -3 0 -3 -1 -5 +5 +2
Apex
Medial
Lateral
104 +1 -4 -2 -3 -2 -1 +2 -2 +4 +3
106 +2 -1 +4 -1 -1 0 +2 -1 +5 +5
106 +2 -1 +4 -2 -1 0 +2 -1 +5 +4
(b) Central axis slice, z = +0 cm
‘M,yjhm~’
kdh?lIt
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
104 -4 -1 +l -3 -1 -2 -1 -1 +2 +1
100 +1 0 +l 0 -1 -2 +1 -1 +2 0
Isocenter 102 +2 -2 -1 -2 0 -1 +3 -2 +4 +1
(c) Caudal slice, z = -6 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
Fig. 2. Isodose distributions in the central axis for the three patients, “small, ” “medium,” and “large,” to illustrate the variation of hot spots with patient size. I = isocenter and N = normalization point.
ators from three different manufacturers, the Clinac 61 100, 9 MD2 lo and the SL/75, ‘i were used in Dplan for the three patients. In addition, dose distributions using 6 MV X rays from a Clinac 6/100 and a 2100 C9 were also compared for the “large” patient. ‘Varian Associates Inc., Palo Alto, CA. “‘Siemens Medical Systems Inc., Concord, CA.
h/3
h/2
Isocenter
Apex
Medial
Lateral
108 -5 +l -3 +2 -3 -1 -3 -2 +2
108 -4 +l -2 +2 -1 -1 -1 - 1 +1
-
112 -4 -3 -4 -2 -2 1;
106 +2
106 +2
+3 0 +4 +3 +2 0 i-6 +6
+4 0 +4 +1 +2 0 +6 +4
-
+3 0
Dose at isocenter cannot be calculated in the caudal slice since the projection of the isocenter is outside of the contour. The normalization point is chosen at the lung-chest wall interface, on the perpendicular bisector of the posterior field edges. For ease of comparison, all dose values are expressed as “+%” (above) and “-%” (below) the corresponding doses from Dplan. Shaded entries in Tables 2 (a)-(c) represent deviations outside of the t 5% range.
RESULTS Comparison of the difberent treatment planning systems Figure 2 shows the representative distributions for the “small, ” “medium,” and “large” breasts using Dplan to
“Philips
Medical
System, Bridgeport,
CT.
Dosimetric
comparison
of treatment
planning
systems
illustrate the differences in the apex, the medial and the lateral hot spots as a function of patient size. Qualitatively, the dose distributions for the “small” patient in all three planes agreed to within + 5% in the central portion of the breast volume for all TPS. Table 2 (ac) compares the dosesat the selected points and hot spots for the “small” patient. In general, the doses at the different selected points (at h/3, h/2 and isocenter) agreed to within + 3% with Dplan for all TPS. There is a larger variation of doses at the hot spots among the different TPS. The Pinnacle TPS showed the largest deviation from the dose values calculated from Dplan and are consistently higher than Dplan. In the caudal plane (Table 2c), the dose at the isocenter cannot be calculated since the projection of the isocenter is outside of the contour at that level. The Modulex TPS failed to produce a dose distribution in that plane. As shown in Table 1, the differences in the in-plane beam profiles for a field size of 25 X 25 at d,,, are generally small ( 102.9 2 0.9%) for the different 6 MV linacs. The largest horn of 105% occurred for one of the MD2 machines involved in the study. The rest were in the range 102- 104%. For breast irradiation, the planning was performed with a field size in the range of (S- 12) cm X ( 1824) cm. The horns for these field sizes are much smaller than that of the 25 X 25 field size. A comparison of the percent depth doses for 6 MV X-rays for the different accelerators in the study is shown in Appendix I. The maximum deviation for the different depth doses is less than 5% which occurs at depths around 5 cm. At larger depths (d 2 8 cm), the difference is less than 3%. The effect of the differences in percent depth doses is reduced when two opposing fields are used; furthermore, since dosedistributions are optimized separately for each machine, the differences observed in distributions for the different TPS cannot be due to the differences in beam profiles and percent depth doses. On the other hand, the larger variation in the doses at the hot spots may be partially due to the fact that in this study, a hot spot is taken as a high dose region which has a finite volume within a slice. No attempt is made to ensure that hot spots are of the same size for the different TPS; rather, the choice of a hot spot is based on clinical practice. There is an inherent uncertainty in comparing the dose at a hot spot, especially in the region of high dose gradient. Based on our previous study (4)) we estimated that there is a ? 2% uncertainty in the dose at a hot spot. Taking into account the dose uncertainty in the hot spots, all TPS agree to within -+ 5% for the “small” patient. For the “medium” patient, similar observations were noted as shown in Table 3. Again, taken into account the t 2% uncertainty in the dosesat the hot spots,the agreement between
all TPS were within
+ 5% in the three calculation
planes. For the “large” patient, a more detailed comparison is necessary due to the significant change in the patient contour and the unusual setup parameters in the tangential fields.
In the CAX
plane,
all TPS
agreed
to within
in irradiation
of breast 0 C-W.
CHENG
839
et al.
Table 3. Comparison of doses at selected points for the “medium” patient (a) Cephalic slice, z = + 6cm h/3
h/2
Isocenter
Apex
Medial
Lateral
98 98 10.5 102 103 Dplan UNC -3 +2 +3 +2 +3 Modulex +l +2 0 +1 0 Prowess 0 0 0 +3 +2 +4 +1 ROCS +1 0 -3 +2 +1 Target 0 +1 +3 Theraplan -3 -2 +1 +2 +2 0 +2 +2 +1 TMS +3 +2 +2 +3 Focus +3 +5 Pinnacle +l 0 -2 +6 +5 -3 -3 0 +3 0 Render Dose at isocenter in the cephalic slice cannot be calculated since the projection of the isocenter is outside of the contour. (b) Central axis slice, z = +0 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
Isocenter
Apex
Medial
Lateral
100 0 +2 +2
102 -1 0 0 -1
104
105
107
107
-2 0
0 +2
-1
0
-3
0
-2
-1
+3
-3 -1
-2 -2
0 +3
0
-1
+1
-2 0 +3 +3 -2 -3 +3 +4
-3
-1
+5
+4
-3
-4 -3
+2
+5
-2 0 +3 +2 -2 -3 +3 +3 +5 +3
Lateral
+4 0 - 1
0 +1 +1
(c) Caudal slice, z = -6 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
Isocenter
Apex
Medial
108
109
110
112
110
110
-3
+1
+2 +2
+l +l -1 0 0
+5 -2
+3 +2 -2 -2 -2 +2 +2
0 +2 0 0
0 +2 0 0
-1
-1
0 0 +2 +5 +4
0 0 +2 +2
+2 +1 -2
+2 +2
+3
0 0 0
+3 0
+1
0
0
0
-1
-2
-3
+6 +3
+4 +2
+1
The normalization point is chosen at the lung-chest wall interface, on the perpendicular bisector of the posterior field edges. For ease of comparison, all dose values are expressed as “+%” (above) and “-%” (below) the corresponding doses from Dplan. Shaded entries in Tables 3 (a)-(c) represent deviations outside of the ? 5% range.
-+ 4% at the selected dose points (Table 4b). The deviation in the hot spots are again larger compared to the interest points. Both the UNC plan and the Pinnacle TPS showed consistently higher doses compared to Dplan. In the cephalic plane (Table 4a), Theraplan, the TMS, the Pinnacle, the Render plan, Dplan and the UNC plan all showed an underdosed region (<99% of the prescribed dose) in the center of the breast contour due to
840
I. J. Radiation Oncology 0 Biology 0 Physics Table 4. Comparison of doses at selected points for the “large” patient
(a) Cephalic slice, z = +6 cm
Dplan UNC Modulex
Prowess ROCS Target
h/3
h/2
Isocenter
Apex
Medial
Lateral
96 -1 +5
97 +2 +4
-
107 +3 -2 +4
108 +4 0 +1
+1
0
+4 +2
+5 +2
-
h/3
h/2
Isocenter
Apex
Medial
Lateral
101 +1 -1 -2 0 -1 -4 -2 +1 +1 -4
103 +2 -1 0 0 -1 -1 0 0 +2 -2
103 +4 +1 +l -1 +1 +1 +1 +1 +3 -1
106 +5 -2 2 -1 +3 +3 +1 +4 0
110 +5 -1 0 0 0 -1 +2 -1 +4 +5
110 +5 -1 0 +2 0 -1 +2 -1 +2 +5
Isocenter
Apex
Medial
Lateral
114 +3 -2 -3 -4 +2 -2
113 +2 -5 -3 -3 -4 -5
-3 -5 +4 -4
2; +1 -3
110 +5 +2 +5
0
+3 0 Theraplan +3 0 TMS +5 0 Focus 0 +4 Pinnacle 0 0 -1 +4 Render -2 -3 -2 0 Dose at isocenter is not defined since it is outside of the contour in the cephalic slice. (b) Central axis slice, z = +0 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
+2 +2 -2 +2 0 +2
(c) Caudal slice, z = -6 cm
Dplan UNC Modulex Prowess
ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
108 0 -5 -4 -3 2
110 +2 -5 -4 -4 -5 ” -4
110 +3 -4 -3 -4 I23
112 +‘F. -5 9:
-1 -5 +2 -3
-3 0 +2 -3
-4 0 +4 -3
+4 -2 +5 0
,:
The normalization point is chosen at the lung-chest wall interface, on the perpendicular bisector of the posterior field edges. For ease of comparison,all dose values are expressed as “+%” (above) and “-%” (below) the corresponding doses from Dplan. Shaded entries in Tables 4 (a)-(c) represent deviations outside of the -C 5% range.
the large separation compared to that at CAX, while the other five TPS showed that the center of the breast is adequately dosed. By removing the collimator angles in the tangential field setup in Dplan ( 17” and 19” for the medial and lateral tangential fields, respectively), the dose distributions in the three calculation planes obtained from Dplan were identical for all practical purpose with the other TPS which do not allow collimator
Volume 38, Number 4, 1997
angle in the setup (see Table 1). In the caudal plane (Table 4c), it is generally true that all TPS agreed to within + 5% at all points and hot spots, taking into account the 2% uncertainty in hot spot dose. An exception is the dose at the NSABP point for the Theraplan which is just outside the 2 5% limit. Comparison of dose distributions from different linear accelerators using the sameplanning system Isodose distributions from three 6 MV linear accelerators, namely, the Clinac 61100, the MD2 and the SL/75, were compared using Dplan, for the “small” and the “large” patients. The dose distributions from the SL/75 were generally “warmer” than those from Clinac 6/ 100 and the MD2 throughout the breast volume. The difference was about 3% for the “small” patient and more than 5% for the “large” patient in the cephalic plane. On the other hand, distributions between the Clinac 6/100 and the MD2 agreed to within 2 2% in all calculation planes. Similar comparisonswere also made between isodosedistributions of the “large” patient irradiated on a Clinac 6/ 100 and the 6 MV X-rays from a 2100 C. The dose distributions agreed to within 2% for the three calculation planes between the two machines. The dose distributions from the Clinac 6/ 100 were generally “warmer” throughout the breast volume compared to those treated on the 2 1OOCmachine. DISCUSSION In this study, we have compared isodose distributions in tangential breast irradiation for a number of TPS commonly used in the clinic. A similar study was reported by a European group (2) and found significant differences among the systems tested. For both the “small” and the ‘ ‘medium’ ’ patients, isodose distributions for all TPS agreed to within + 5% in three representative planes, taking into account the + 2% uncertainty in the dosesat the hot spots. Considering the differences in beam profiles for the different linear accelerators, and the vast differences in dose algorithms between true 3D and pseudo-3D systems, the + 5% difference is reasonable. For the “large” patient, while agreement among all TPS were within ? 5% on the CAX plane, larger differences in the two off-axis planes were observed. In the cepahlic plane, the Theraplan, the TMS system, the Pinnacle, the Render plan, Dplan and the UNC plan all correctly calculated the dosesat the center of the breast which were below the prescription dose, although the extent of underdosage is different, ranging from 94 to 99%. The other five TPS, however, showed that the sameregion is adequately dosed. Similar differences were noted for some patients with separation > 24 cm (5 ) . The discrepancy is largely due to the inability of most pseudo-3D systemsto incorporate the collimator angles in the dose calculation. By removing the collimator angle in the beam setup, dose distributions from Dplan agreed with the other five TPS to within 2%.
Dosimetric comparison of treatment planning systems in irradiation of breast 0 C.-W.
Of the four 3D TPS, there is a slightly better agreement between the Render plan and Dplan compared to Pinnacle and the UNC plan. On the other hand, the TMS system, even though not a true 3D system, agreed to within + 3% with Dplan for most of all interest points and hot spot doses for all three patients. Similar good agreement between Theraplan (a pseudo-3D system) and Dplan are observed except for the “large” patient in the caudal plane. It is interesting to note that for scatter dose calculation, the TMS system utilizes a hybrid model of Monte Carlo calculated convolution kernels and measured beam data ( 1) but without taking into account the true shape of the scattering volume; the Pinnacle uses convolution technique and Monte Carlo calculated kernels; the Render plan uses an irregular field algorithm taking into account the patient curvature and the presence of beam modifying devices; the Dplan employs a strip integration technique; while both the Theraplan and the UNC plan employ the differential scatter-air-ratio method. The difference in the dosesat the interest points among all TPS has a more pronounced repercussion compared to the differences in the hot spot doses. As pointed out earlier, the selected points in Fig. 1 represent dose prescription points used in most clinics. The difference in doses at these points may pose a problem in dose comparison and treatment evaluation among the different clinics participating in certain national or regional breast protocols. The effect of the selection of prescription point on dose comparison is the subject of a separate study (5). The difference in dose distributions obtained with the SL/75, the Clinac 6/ 100 and the MD2 is probably due to the difference in beam profiles, especially since the SL/ 75 has a thick 60” wedge built into the treatment head. The good agreement between Clinac 6/ 100 and MD2 in-
CHENG
et al.
841
dicates that the two machines have similar beam characteristics even though the wedges associatedwith the MD2 generally are thicker compared to those with the Clinac 6/100. The 6 MV X-rays from two different models of linear accelerators from the same manufacturer, namely the Clinac 6/ 100 and the Clinac 21OOC,have similar beam characteristics and the dose distributions are within ? 2% of each other throughout the breast volume. CONCLUSION In summary, the use of pseudo-3D TPS in tangential breast irradiation seemsto be adequate clinically for more than 90% of all breast patients. This is true even for those systems which employ a simple 2D approach in the dose calculation. For breast patients whose breast contours change dramatically in the tangential fields or require large collimator angles in the setup, a true 3D TPS is necessary to appreciate the dose variation throughout the breast variation; however, the use of a true 3D TPS would not improve the dose distribution throughout the breast volume. Only 3D tissue compensators could improve the distributions. The usefulness of a true 3D system in tangential breast irradiation then is to reveal the actual distributions in the breast volume, to help design the 3D compensators and to verify the dose distributions when the compensators are fabricated. This study suggeststhat radiation treatment outcome in a multi-institutional study can be compared with confidence. The choice of TPS or machine may not be a factor in treatment. Most TPS provide adequate and nearly accurate dose distributions within + 5% as long as the dose specification point is consistent among the participating institutions.
REFERENCES Ahnesjo, A. Saxner, M. Trepp, A. A pencil-beam model for photon-dose calculation. Med. Phys. 19:263-273; 1992. Bree, N. A. M., Battum, L. J., Huizenga, H., Mijnheer, B. J., Three-Dimensional dose distribution of tangential breast treatment: a national dosimetry intercomparison. Radiother. Oncol. 22:252-260; 1991. Chin, L. M., Cheng, C.-W., Siddon, R. L., Rice, R. K., Mijnheer, B. J., Harris, J. R., Three dimensional photon dose distributions with and without lung corrections for tangential breast intact treatments. Int. J. Radiat. Oncol. Biol. Phys. 17:1327-1335; 1989. Cheng, C.-W., Das, I. J., Stea, B. The effect of the number of computed tomographic slices on dose distributions and evaluation of treatment planning systems for radiation therapy of intact breast. Int. J. Radiat. Oncol. Biol. Phys. 30: 183- 195; 1994.
5. Das I. J., Cheng, C.-W., Fein, D. A., Fowble, B., Patterns of dose variability in radiation prescription of breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 32 (Suppl):256; 1995. 6. Fraass B. A., McShan, D. L. Development in radiation therapy treatment planning. Computers in Medical Physics, AAPM Monograph no. 17, Benedetto, A.R., Huang, H.K., Ragan, D.P.; eds. p. 303-316; 1990. 7. Van Dyk, J., Bamett, R. B., Cygler, J. E., Shragge, P. C. Commissioning and quality assurance of treatment planning computers. Int. J. Radiat. Oncol. Biol. Phys. 26:261-273; 1993. 8. Westermann, C. F., Mijnheer, B. J., Kleffens, H. J. Determination of the accuracy of different computer planning systerns for treatment with external photon beams. Radiother. Oncol. 9:339-347; 1984.
842
I. J. Radiation
Oncology
0 Biology
0 Physics
Volume
APPENDIX
38, Number
4, 1997
I
Comparison of percent depth doses of 6 MV X-rays for the different accelerators the study. The bold line is the average of all the depth doses.
T @ $ 8 5 E 8 tl n
_-
.._( ,
! i
I !
.._..._
._._..___. _.___..___.___. .__. __ .__. .__. .___._..
..,
1
t / / /
10
15 depth
cm
in
Biol.
Phys., Vol. 38, No. 4, pp. 835-842. 1997 Copyright 0 1997 Elsevier Science Inc. Printed in the USA. All rights reserved 0360.3016/97 $17.00 + .OO
PI1 SO360-3016( 97)00078-3
ELSEVIER
l
Oncology
Physics Contribution DOSIMETRIC COMPARISON OF TREATMENT PLANNING SYSTEMS IRRADIATION OF BREAST WITH TANGENTIAL FIELDS
IN
CHEE-WAI CHENG, PH.D.,? INDRA J. DAS, PH.D., F.I.P.S.M.,# WALTER TANG, M.S.,” SHA CHANG, PH.D., 8 JEN-SAN TSAI, PH.D., * CRISTER CEBERG, PH.D.,’ BARBARA DE GASPIE, RT(T),” RAJINDER SINGH, D.A.B.R.,* * DOUGLAS A. FEIN, M.D.# AND BARBARA FOWBLE, M.D.# +Department of Radiation Oncology, University of Arizona, Tucson, AZ, #Department of Radiation Oncology, Fox Chase Cancer Center, Philadelphia, PA, “Medical Radiological Physics Consortium, Minneapolis, MN, “Department of Radiation Oncology, University of North Carolina, Chapel Hill, NC, *Department of Radiation Oncology, New England Medical Center, Boston, MA, *Department of Radiation Physics, Lund University Hospital, Lund, Sweden, “Radiation Oncology, University of Massachusetts Medical Center, University of Massachusetts, Worchester, MA, * *Department of Radiation Oncology, Glendale Memorial Hospital, Glendale, CA Purpose: The objectives of this study ate: (1) to investigate the d&metric differences of the diiferent treatment planning systems (TPS) iu breast h-radiation with tangential fields, and (2) to study the effect of beam characteristics on dose distributions in tangential breast irradiation with 6 MV linear accelerators from different manufacturers. Methods and Materials: Nine commercial and two university-based TPS are evaluated in this study. The computed “ medium” and “large” based on their tomographic scan of three representative patients, labeled as “small”, respective chest wall separations in the central axis plane (CAX) were used. For each patient, the tangential fields were set up in each TPS. The CAX distribution was optimized separately with lung correction, for each TPS based on the same set of optimization conditions. The isodose distributions in two other off-axis planes, one 6 cm cephalic and the other 6 cm caudal to the CAX plane were also computed. To investigate the effect of beam characteristics on dose distributions, a three-dimensional TPS was used to calculate the isodose distributions for three different linear accelerators, the Varian Clinac 6/100, the Siemens MD2 and the Philips SL/7 for the three patients. In addition, dose distributions obtained with 6 MV X-rays from two different accelerators, the Varian Clinac 6/100 and the Varian 21OOC, were compared. Results: For all TPS, the dose distributions in all three planes agreed qualitatively to within ? 5% for the “small” and the “medium” patients. For the “large” patient, all TPS agreed to within ? 4% on the CAX plane. The isodose distributions in the caudal plane differed by ? 5 % among all TPS. In the cephalic plane in which the patient separation is much larger than that in the CAX plane, six TPS correctly calculated the dose distribution showing a cold spot in the center of the breast contour. The other five TPS showed that the center of the breast received adequate dose. Isodose distributions for 6 MV X-rays from three diierent accelerators diifered by about -i- 3% for the “small” patient and more than -C 5% for the “large” patient. For two different 6 MV machines of the same manufacturer, the isodose distribution agreed to within i- 2% for all three planes for the “large” patient. Conclusion: The differences observed among the various TPS in this study were within + 5% for both the “small” and the “medium” patients while doses at the hot spot exhibit a larger variation. The large discrepancy observed in the off-axis plane for the “large” patient is largely due to the inability of most TPS to incorporate the collimator angles in the dose calculation. Only six systems involved agreed to within -C 5% for all three patients in all calculation planes. The difference in dose distributions obtained with three accelerators from different manufacturers is probably due to the difference in beam profiles. On the other hand, the 6 MV X-rays from two different models of linear accelerators from the same manufacturer have similar beam characteristics and the dose distributions are within k 2% of each other throughout the breast volume. In general, multi-institutional breast treatment data can be compared within a 2 5% accuracy. 0 1997 Elsevier Science Inc. Treatment
planning
systems, Tangential
breast irradiation.
INTRODUCTION Currently, there are over a dozen treatment planning systerns (TPS ) used in the different radiation oncology clin-
its in the U.S. Some of them were developed by commercial vendors, some were developed as joint ventures between universities and commercial vendors, and others were developed solely in-house at some universities. The
Reprint requests to: Chee-Wai Cheng, Ph.D., Department of Radiation Oncology, University of Arizona, Tucson, AZ 857245081. Acknowledgement-We would like to thank Katherine Buchneit
at the Department of Radiation Oncology, Fox Chase Cancer Center, for her editorial help. We would also like to thank the technical help of Mr. Robert Costa of the ADAC, Milpitas, CA. Accepted for publication 3 1 January 1997. 835
836
I. J. Radiation Oncology 0 Biology 0 Physics
Volume 38, Number 4, 1997
Table 1. Treatmentplanningsystems,lung correctionalgorithmsandlinear accelerators
Dplan UNC plan Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Renderplan
Machine
Lung corr. algorithm
Horns(%)
Clinac 6/100 SiemensMD2 Mevatron 67 Varian 2100C SiemensMD2 Mevatron 67 Varian 6OOC/D Philips SLi75 Varian 2300C Varian 2 1OOC Clinac 6/100
Ratio-of-TMR (RTMR) Batho’s RTMR Effective path length (EPL) EPL EPL Batho’s EPL and 1-D convolution RTMR Convolution EPL
103 105 102 102 103 102 103 104 103 103 102
Collimatorangles Yes Yes No No No No Yes Yes No Yes Yes
Field size of 25 x 25 cm at d,,,,.
early commercial TPS were two-dimensional (2D) TPS, later ones are marketed as 2SD (called pseudo-3D in the present study) and the most recent development has led to three-dimensional (3D) TPS. University-based TPS, on the other hand, are in general 3D TPS. The capabilities of 2D, pseudo-3D and 3D treatment planning systems have been discussedpreviously (6) ; however, the dividing line between a 2D system and a pseudo-3D TPS becomes fuzzy with the advancement in computer technology, as some of the vendors upgraded their 2D TPS to allow multiple-slice display, beams-eye-view, wire frame, colorwash display, surface rendering, etc., while leaving the dose calculation algorithm unchanged. For simplicity, we consider any TPS which has the capabilities of displaying multiple computed tomographic (CT) slices, beam’s-eyeview and other three-dimensional (3-D) display options as either a pseudo-3D or a 3D TPS. All other TPS which can handle only the central axis slice and which do not offer the various 3D display options are considered 2D; thus the distinction between a pseudo-3D and a 3D system is in the degree of sophistication of the dose calculation algorithm employed. A true 3D algorithm must take into account the shape of the irradiated volume with 3D divergence effects and with an accurate model for scatter dose integration (with or without blocks and with or without inhomogeneity correction) For example, in one of the 3D planning systems used in this study, the use of collimator angles in a tangential field setup can be incorporated in the dose calculation (3 ) and agreement within t 2% between calculation and ion chamber measurements has been reported (4). The samemay not be true for pseudo-3D algorithms. There is no formal definition as to what constitutes a pseudo-3D algorithm. A most commonly employed dose algorithm in pseudo-3D systemsis to approximate the entire volume by a cylinder with the same cross-sectional view as the slice in question (basically a 2D approach). The “z’‘-dependence (cephalic-caudal) of the doseprofile; however, is included in the dose calculation. A much more sophisticated approach combining precalculated Monte Carlo energy deposition kernels and measured beam data in a water phantom is usedin one of the pseudo-
3D systems ( 1) . The scatter integration is performed over the volume defined by the treatment fields and the patient contours. It is not a true 3D system since the SSD is assumed to be the same on the CAX over the entire field. It is not known how accurate those pseudo-3D systems are compared to true 3D systems when dose distributions are calculated in 3D throughout a treatment volume, especially one which changesvastly in shapein a treatment field. The question we would like to answer in this study is: do all TPS produce dose distributions within 5% of each other throughout a given treatment volume? To answer this question, we have carried out a study to compare isodosedistributions obtained with different TPS throughout a breast volume irradiated with tangential fields. Unlike the irradiation of prostate, lung or abdomen which generally involves fairly regularly shaped external contours throughout the radiation fields, the irradiation of breast with tangential fields involves a complex geometrical setup. The shape of the breast, the presence of lung and the use of collimator angles are often the complicating factors in dose calculations. The sophistication and the accuracy of a dose algorithm is best tested with such a complex volume. In addition to the many different TPS used in treatment planning, there are also a variety of linear accelerators from different vendors used in the clinics. It is generally believed that all linear accelerators of the same energy have similar characteristics in terms of beam profiles, depth doses, etc.; however, the same may not be true if wedges are usedbecauseof the different design of wedges among the different vendors. It should be emphasizedthat the differences in beam characteristics, wedge profiles, etc. between any two machines may not be a problem clinically since each clinic measuresits own beam data for each machine. The differences may complicate situations such as dose comparison between two centers with different TPS and linear accelerators. The objectives of this study are: 1) to study the dosimetric differences of various TPS in breast irradiation with tangential fields, and 2) to study the effect of beam characteristics on dose distributions with 6 MV linear accelerators from different manufacturers. Similar studies
Dosimetric comparison of treatment planning systems in irradiation of breast 0 C.-W.
CHENG
ef al.
837
have been conducted in Europe to investigate the performance of computer systems some of which were European-based and are not available in the U.S. (2, 8).
MATERIALS
AND METHODS
In this study, dose distributions obtained from nine commercial and two university-based TPS are compared. The commercial systems are the Modulex’ and the Focus systems,’ the Target system,2the ROCS3 the Prowess4 the Theraplan5 the TMS6 the Pinnacle,’ and the Render Plan.’ The two university-based systemsare the Dplan developed at the Joint Center for Radiation Therapy in Boston and the 3D treatment planning system developed at the University of North Carolina in Chapel Hill (UNC Plan). A summary of the TPS and the type of linear acceleratorsused in the study is given in Table 1. It is assumedthat all of the systems have been evaluated basedon the criteria for commissioning and quality assurance(7). Three representative patients, labeled as “small,” “medium” and “large” basedon their respective chest wall separationsof 18,2 1, and 28 cm in the central axis plane (CAX) were usedin this study. These three patients were chosento cover the range of chest wall and breast sizes encountered in the clinic. For the treatment planning, CT studiesof these three patients were performed. The external and the lung contours were extracted automatically on each CT slice using Dplan and printed out on a laser printer. The accuracy of scaling the printouts had been tested and was found to be accurate to within a millimeter of the true dimension. The hard copy printouts were sentto the different institutions and digitized manually into the various TPS. In this study, the breast irradiation technique consisted of two opposing isocentric tangential fields. The posterior field edges of the tangential fields were made coincident to minimize the volume of lung inside the fields. For each patient, the tangential fields were set up in each planning system. To facilitate comparison of dose distributions obtained with the different TPS, treatment planning was carried out using 6 MV X-rays only. Since the beam characteristics for 6 MV X-rays may be different for different accelerators, the CAX distribution was optimized separately for each TPS based on the same set of optimization conditions: (a) the dose distributions must be symmetrically distributed, and the medial and lateral dose heterogeneity equalized, (b) the dose at the apex must be equal to or smaller than the medial and lateral dose heterogeneity, and (c) the center of the breast must receive 2 100% of the prescribed dose provided that condition (b) was not violated. The dose normalization point was chosen at the lung-chest wall interface, anterior to the interface and on the perpendicular bisector of the posterior field edges.
Wedge and open beam combination were used for each field to obtain the optimal dose distribution. Lung correction was also used in the dose calculation. The dosimetry of tangential breast irradiation with lung correction has been reported previously (3 ) . Once an optimal distribution on the CAX was obtained, isodose distributions in two off-axis planes, one 6 cm cephalic and the other 6 cm caudal to the CAX plane, were computed. In addition, doses were computed at selected points inside each contour as shown in Fig. 1. The interest points were chosen since they represent the prescription points used in the different clinics (5 >; furthermore, the locations of these points provide a measure of the uniformity of the distribution in the breast. These, together with the hot spots at the apex, medial and the lateral subcutaneous locations characterize the dose distribution within a contour. In this study, a hot spot is defined as a dose region 2 110% of the prescribed dose. The choice of the 110% isodosevalue is somewhat arbitrary and based on the clinical practice in that it is considered as the maximum dose acceptable for a daily fraction size of 180 cGy. For consistency, dose distributions generated in each planning system were sent to the Department of Radiation Oncology, University of Arizona, and analyzed by one of the authors (Cheng). To investigate the effect of the different beam characteristics on dose distributions, three 6 MV linear acceler-
‘Computerized
5TheratronicsInternationalLtd., Kanada,Ontario, Canada.
Medical
System, St. Louis, MO.
‘GeneralElectric Medical System,Milwaukee, WI. 3Radiation Oncology Computer 4Prowess System, Chico, CA.
Systems, Inc., San Diego, CA.
Fig. 1. Schematicdiagramof the centralaxis of a tangentialfield breastirradiation showingdosecalculation points and the hot spots.1-NSABP point at h/3 distance,2-center of breast(h/ 2), 3-isocenter, 4-apex hot spot,5-medial hot spotand6lateralhot spot.
6Ma.rketed by Siemens Medical ‘ADAC, Milpitas, CA.
Systems Inc., Concord, CA.
‘PrecisionTherapy International,North Miami, Beach,FL.
838
I. J. Radiation Oncology l Biology l Physics
Volume 38, Number 4, 1997 Table 2. Comparison of doses at selected points for the “small” patient (a) Cephalic slice, z = +6 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
100 -5 +l 0 0 0 - 1 0 -1 +2 +2
100 0 +l 0 0 0 0 +3 -1 +2 -2
Isocenter 106 -1 -2 -1 -2 -4 0 +1 -1 +2 +2
Apex
Medial
Lateral
108 -3 -5 -3 -5 -6 -2 +2 -3 +2 0
105 -3 -3 +4 -4 -2 -2 -1 -5 +5 +2
105 -3 0 0 -3 0 -3 -1 -5 +5 +2
Apex
Medial
Lateral
104 +1 -4 -2 -3 -2 -1 +2 -2 +4 +3
106 +2 -1 +4 -1 -1 0 +2 -1 +5 +5
106 +2 -1 +4 -2 -1 0 +2 -1 +5 +4
(b) Central axis slice, z = +0 cm
‘M,yjhm~’
kdh?lIt
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
104 -4 -1 +l -3 -1 -2 -1 -1 +2 +1
100 +1 0 +l 0 -1 -2 +1 -1 +2 0
Isocenter 102 +2 -2 -1 -2 0 -1 +3 -2 +4 +1
(c) Caudal slice, z = -6 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
Fig. 2. Isodose distributions in the central axis for the three patients, “small, ” “medium,” and “large,” to illustrate the variation of hot spots with patient size. I = isocenter and N = normalization point.
ators from three different manufacturers, the Clinac 61 100, 9 MD2 lo and the SL/75, ‘i were used in Dplan for the three patients. In addition, dose distributions using 6 MV X rays from a Clinac 6/100 and a 2100 C9 were also compared for the “large” patient. ‘Varian Associates Inc., Palo Alto, CA. “‘Siemens Medical Systems Inc., Concord, CA.
h/3
h/2
Isocenter
Apex
Medial
Lateral
108 -5 +l -3 +2 -3 -1 -3 -2 +2
108 -4 +l -2 +2 -1 -1 -1 - 1 +1
-
112 -4 -3 -4 -2 -2 1;
106 +2
106 +2
+3 0 +4 +3 +2 0 i-6 +6
+4 0 +4 +1 +2 0 +6 +4
-
+3 0
Dose at isocenter cannot be calculated in the caudal slice since the projection of the isocenter is outside of the contour. The normalization point is chosen at the lung-chest wall interface, on the perpendicular bisector of the posterior field edges. For ease of comparison, all dose values are expressed as “+%” (above) and “-%” (below) the corresponding doses from Dplan. Shaded entries in Tables 2 (a)-(c) represent deviations outside of the t 5% range.
RESULTS Comparison of the difberent treatment planning systems Figure 2 shows the representative distributions for the “small, ” “medium,” and “large” breasts using Dplan to
“Philips
Medical
System, Bridgeport,
CT.
Dosimetric
comparison
of treatment
planning
systems
illustrate the differences in the apex, the medial and the lateral hot spots as a function of patient size. Qualitatively, the dose distributions for the “small” patient in all three planes agreed to within + 5% in the central portion of the breast volume for all TPS. Table 2 (ac) compares the dosesat the selected points and hot spots for the “small” patient. In general, the doses at the different selected points (at h/3, h/2 and isocenter) agreed to within + 3% with Dplan for all TPS. There is a larger variation of doses at the hot spots among the different TPS. The Pinnacle TPS showed the largest deviation from the dose values calculated from Dplan and are consistently higher than Dplan. In the caudal plane (Table 2c), the dose at the isocenter cannot be calculated since the projection of the isocenter is outside of the contour at that level. The Modulex TPS failed to produce a dose distribution in that plane. As shown in Table 1, the differences in the in-plane beam profiles for a field size of 25 X 25 at d,,, are generally small ( 102.9 2 0.9%) for the different 6 MV linacs. The largest horn of 105% occurred for one of the MD2 machines involved in the study. The rest were in the range 102- 104%. For breast irradiation, the planning was performed with a field size in the range of (S- 12) cm X ( 1824) cm. The horns for these field sizes are much smaller than that of the 25 X 25 field size. A comparison of the percent depth doses for 6 MV X-rays for the different accelerators in the study is shown in Appendix I. The maximum deviation for the different depth doses is less than 5% which occurs at depths around 5 cm. At larger depths (d 2 8 cm), the difference is less than 3%. The effect of the differences in percent depth doses is reduced when two opposing fields are used; furthermore, since dosedistributions are optimized separately for each machine, the differences observed in distributions for the different TPS cannot be due to the differences in beam profiles and percent depth doses. On the other hand, the larger variation in the doses at the hot spots may be partially due to the fact that in this study, a hot spot is taken as a high dose region which has a finite volume within a slice. No attempt is made to ensure that hot spots are of the same size for the different TPS; rather, the choice of a hot spot is based on clinical practice. There is an inherent uncertainty in comparing the dose at a hot spot, especially in the region of high dose gradient. Based on our previous study (4)) we estimated that there is a ? 2% uncertainty in the dose at a hot spot. Taking into account the dose uncertainty in the hot spots, all TPS agree to within -+ 5% for the “small” patient. For the “medium” patient, similar observations were noted as shown in Table 3. Again, taken into account the t 2% uncertainty in the dosesat the hot spots,the agreement between
all TPS were within
+ 5% in the three calculation
planes. For the “large” patient, a more detailed comparison is necessary due to the significant change in the patient contour and the unusual setup parameters in the tangential fields.
In the CAX
plane,
all TPS
agreed
to within
in irradiation
of breast 0 C-W.
CHENG
839
et al.
Table 3. Comparison of doses at selected points for the “medium” patient (a) Cephalic slice, z = + 6cm h/3
h/2
Isocenter
Apex
Medial
Lateral
98 98 10.5 102 103 Dplan UNC -3 +2 +3 +2 +3 Modulex +l +2 0 +1 0 Prowess 0 0 0 +3 +2 +4 +1 ROCS +1 0 -3 +2 +1 Target 0 +1 +3 Theraplan -3 -2 +1 +2 +2 0 +2 +2 +1 TMS +3 +2 +2 +3 Focus +3 +5 Pinnacle +l 0 -2 +6 +5 -3 -3 0 +3 0 Render Dose at isocenter in the cephalic slice cannot be calculated since the projection of the isocenter is outside of the contour. (b) Central axis slice, z = +0 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
Isocenter
Apex
Medial
Lateral
100 0 +2 +2
102 -1 0 0 -1
104
105
107
107
-2 0
0 +2
-1
0
-3
0
-2
-1
+3
-3 -1
-2 -2
0 +3
0
-1
+1
-2 0 +3 +3 -2 -3 +3 +4
-3
-1
+5
+4
-3
-4 -3
+2
+5
-2 0 +3 +2 -2 -3 +3 +3 +5 +3
Lateral
+4 0 - 1
0 +1 +1
(c) Caudal slice, z = -6 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
Isocenter
Apex
Medial
108
109
110
112
110
110
-3
+1
+2 +2
+l +l -1 0 0
+5 -2
+3 +2 -2 -2 -2 +2 +2
0 +2 0 0
0 +2 0 0
-1
-1
0 0 +2 +5 +4
0 0 +2 +2
+2 +1 -2
+2 +2
+3
0 0 0
+3 0
+1
0
0
0
-1
-2
-3
+6 +3
+4 +2
+1
The normalization point is chosen at the lung-chest wall interface, on the perpendicular bisector of the posterior field edges. For ease of comparison, all dose values are expressed as “+%” (above) and “-%” (below) the corresponding doses from Dplan. Shaded entries in Tables 3 (a)-(c) represent deviations outside of the ? 5% range.
-+ 4% at the selected dose points (Table 4b). The deviation in the hot spots are again larger compared to the interest points. Both the UNC plan and the Pinnacle TPS showed consistently higher doses compared to Dplan. In the cephalic plane (Table 4a), Theraplan, the TMS, the Pinnacle, the Render plan, Dplan and the UNC plan all showed an underdosed region (<99% of the prescribed dose) in the center of the breast contour due to
840
I. J. Radiation Oncology 0 Biology 0 Physics Table 4. Comparison of doses at selected points for the “large” patient
(a) Cephalic slice, z = +6 cm
Dplan UNC Modulex
Prowess ROCS Target
h/3
h/2
Isocenter
Apex
Medial
Lateral
96 -1 +5
97 +2 +4
-
107 +3 -2 +4
108 +4 0 +1
+1
0
+4 +2
+5 +2
-
h/3
h/2
Isocenter
Apex
Medial
Lateral
101 +1 -1 -2 0 -1 -4 -2 +1 +1 -4
103 +2 -1 0 0 -1 -1 0 0 +2 -2
103 +4 +1 +l -1 +1 +1 +1 +1 +3 -1
106 +5 -2 2 -1 +3 +3 +1 +4 0
110 +5 -1 0 0 0 -1 +2 -1 +4 +5
110 +5 -1 0 +2 0 -1 +2 -1 +2 +5
Isocenter
Apex
Medial
Lateral
114 +3 -2 -3 -4 +2 -2
113 +2 -5 -3 -3 -4 -5
-3 -5 +4 -4
2; +1 -3
110 +5 +2 +5
0
+3 0 Theraplan +3 0 TMS +5 0 Focus 0 +4 Pinnacle 0 0 -1 +4 Render -2 -3 -2 0 Dose at isocenter is not defined since it is outside of the contour in the cephalic slice. (b) Central axis slice, z = +0 cm
Dplan UNC Modulex Prowess ROCS Target Theraplan TMS Focus Pinnacle Render
+2 +2 -2 +2 0 +2
(c) Caudal slice, z = -6 cm
Dplan UNC Modulex Prowess
ROCS Target Theraplan TMS Focus Pinnacle Render
h/3
h/2
108 0 -5 -4 -3 2
110 +2 -5 -4 -4 -5 ” -4
110 +3 -4 -3 -4 I23
112 +‘F. -5 9:
-1 -5 +2 -3
-3 0 +2 -3
-4 0 +4 -3
+4 -2 +5 0
,:
The normalization point is chosen at the lung-chest wall interface, on the perpendicular bisector of the posterior field edges. For ease of comparison,all dose values are expressed as “+%” (above) and “-%” (below) the corresponding doses from Dplan. Shaded entries in Tables 4 (a)-(c) represent deviations outside of the -C 5% range.
the large separation compared to that at CAX, while the other five TPS showed that the center of the breast is adequately dosed. By removing the collimator angles in the tangential field setup in Dplan ( 17” and 19” for the medial and lateral tangential fields, respectively), the dose distributions in the three calculation planes obtained from Dplan were identical for all practical purpose with the other TPS which do not allow collimator
Volume 38, Number 4, 1997
angle in the setup (see Table 1). In the caudal plane (Table 4c), it is generally true that all TPS agreed to within + 5% at all points and hot spots, taking into account the 2% uncertainty in hot spot dose. An exception is the dose at the NSABP point for the Theraplan which is just outside the 2 5% limit. Comparison of dose distributions from different linear accelerators using the sameplanning system Isodose distributions from three 6 MV linear accelerators, namely, the Clinac 61100, the MD2 and the SL/75, were compared using Dplan, for the “small” and the “large” patients. The dose distributions from the SL/75 were generally “warmer” than those from Clinac 6/ 100 and the MD2 throughout the breast volume. The difference was about 3% for the “small” patient and more than 5% for the “large” patient in the cephalic plane. On the other hand, distributions between the Clinac 6/100 and the MD2 agreed to within 2 2% in all calculation planes. Similar comparisonswere also made between isodosedistributions of the “large” patient irradiated on a Clinac 6/ 100 and the 6 MV X-rays from a 2100 C. The dose distributions agreed to within 2% for the three calculation planes between the two machines. The dose distributions from the Clinac 6/ 100 were generally “warmer” throughout the breast volume compared to those treated on the 2 1OOCmachine. DISCUSSION In this study, we have compared isodose distributions in tangential breast irradiation for a number of TPS commonly used in the clinic. A similar study was reported by a European group (2) and found significant differences among the systems tested. For both the “small” and the ‘ ‘medium’ ’ patients, isodose distributions for all TPS agreed to within + 5% in three representative planes, taking into account the + 2% uncertainty in the dosesat the hot spots. Considering the differences in beam profiles for the different linear accelerators, and the vast differences in dose algorithms between true 3D and pseudo-3D systems, the + 5% difference is reasonable. For the “large” patient, while agreement among all TPS were within ? 5% on the CAX plane, larger differences in the two off-axis planes were observed. In the cepahlic plane, the Theraplan, the TMS system, the Pinnacle, the Render plan, Dplan and the UNC plan all correctly calculated the dosesat the center of the breast which were below the prescription dose, although the extent of underdosage is different, ranging from 94 to 99%. The other five TPS, however, showed that the sameregion is adequately dosed. Similar differences were noted for some patients with separation > 24 cm (5 ) . The discrepancy is largely due to the inability of most pseudo-3D systemsto incorporate the collimator angles in the dose calculation. By removing the collimator angle in the beam setup, dose distributions from Dplan agreed with the other five TPS to within 2%.
Dosimetric comparison of treatment planning systems in irradiation of breast 0 C.-W.
Of the four 3D TPS, there is a slightly better agreement between the Render plan and Dplan compared to Pinnacle and the UNC plan. On the other hand, the TMS system, even though not a true 3D system, agreed to within + 3% with Dplan for most of all interest points and hot spot doses for all three patients. Similar good agreement between Theraplan (a pseudo-3D system) and Dplan are observed except for the “large” patient in the caudal plane. It is interesting to note that for scatter dose calculation, the TMS system utilizes a hybrid model of Monte Carlo calculated convolution kernels and measured beam data ( 1) but without taking into account the true shape of the scattering volume; the Pinnacle uses convolution technique and Monte Carlo calculated kernels; the Render plan uses an irregular field algorithm taking into account the patient curvature and the presence of beam modifying devices; the Dplan employs a strip integration technique; while both the Theraplan and the UNC plan employ the differential scatter-air-ratio method. The difference in the dosesat the interest points among all TPS has a more pronounced repercussion compared to the differences in the hot spot doses. As pointed out earlier, the selected points in Fig. 1 represent dose prescription points used in most clinics. The difference in doses at these points may pose a problem in dose comparison and treatment evaluation among the different clinics participating in certain national or regional breast protocols. The effect of the selection of prescription point on dose comparison is the subject of a separate study (5). The difference in dose distributions obtained with the SL/75, the Clinac 6/ 100 and the MD2 is probably due to the difference in beam profiles, especially since the SL/ 75 has a thick 60” wedge built into the treatment head. The good agreement between Clinac 6/ 100 and MD2 in-
CHENG
et al.
841
dicates that the two machines have similar beam characteristics even though the wedges associatedwith the MD2 generally are thicker compared to those with the Clinac 6/100. The 6 MV X-rays from two different models of linear accelerators from the same manufacturer, namely the Clinac 6/ 100 and the Clinac 21OOC,have similar beam characteristics and the dose distributions are within ? 2% of each other throughout the breast volume. CONCLUSION In summary, the use of pseudo-3D TPS in tangential breast irradiation seemsto be adequate clinically for more than 90% of all breast patients. This is true even for those systems which employ a simple 2D approach in the dose calculation. For breast patients whose breast contours change dramatically in the tangential fields or require large collimator angles in the setup, a true 3D TPS is necessary to appreciate the dose variation throughout the breast variation; however, the use of a true 3D TPS would not improve the dose distribution throughout the breast volume. Only 3D tissue compensators could improve the distributions. The usefulness of a true 3D system in tangential breast irradiation then is to reveal the actual distributions in the breast volume, to help design the 3D compensators and to verify the dose distributions when the compensators are fabricated. This study suggeststhat radiation treatment outcome in a multi-institutional study can be compared with confidence. The choice of TPS or machine may not be a factor in treatment. Most TPS provide adequate and nearly accurate dose distributions within + 5% as long as the dose specification point is consistent among the participating institutions.
REFERENCES Ahnesjo, A. Saxner, M. Trepp, A. A pencil-beam model for photon-dose calculation. Med. Phys. 19:263-273; 1992. Bree, N. A. M., Battum, L. J., Huizenga, H., Mijnheer, B. J., Three-Dimensional dose distribution of tangential breast treatment: a national dosimetry intercomparison. Radiother. Oncol. 22:252-260; 1991. Chin, L. M., Cheng, C.-W., Siddon, R. L., Rice, R. K., Mijnheer, B. J., Harris, J. R., Three dimensional photon dose distributions with and without lung corrections for tangential breast intact treatments. Int. J. Radiat. Oncol. Biol. Phys. 17:1327-1335; 1989. Cheng, C.-W., Das, I. J., Stea, B. The effect of the number of computed tomographic slices on dose distributions and evaluation of treatment planning systems for radiation therapy of intact breast. Int. J. Radiat. Oncol. Biol. Phys. 30: 183- 195; 1994.
5. Das I. J., Cheng, C.-W., Fein, D. A., Fowble, B., Patterns of dose variability in radiation prescription of breast cancer. Int. J. Radiat. Oncol. Biol. Phys. 32 (Suppl):256; 1995. 6. Fraass B. A., McShan, D. L. Development in radiation therapy treatment planning. Computers in Medical Physics, AAPM Monograph no. 17, Benedetto, A.R., Huang, H.K., Ragan, D.P.; eds. p. 303-316; 1990. 7. Van Dyk, J., Bamett, R. B., Cygler, J. E., Shragge, P. C. Commissioning and quality assurance of treatment planning computers. Int. J. Radiat. Oncol. Biol. Phys. 26:261-273; 1993. 8. Westermann, C. F., Mijnheer, B. J., Kleffens, H. J. Determination of the accuracy of different computer planning systerns for treatment with external photon beams. Radiother. Oncol. 9:339-347; 1984.
842
I. J. Radiation
Oncology
0 Biology
0 Physics
Volume
APPENDIX
38, Number
4, 1997
I
Comparison of percent depth doses of 6 MV X-rays for the different accelerators the study. The bold line is the average of all the depth doses.
T @ $ 8 5 E 8 tl n
_-
.._( ,
! i
I !
.._..._
._._..___. _.___..___.___. .__. __ .__. .__. .___._..
..,
1
t / / /
10
15 depth
cm
in