- Email: [email protected]
Determining the optimal block margin on the planning target volume for extracranial stereotactic radiotherapy
Int. J. Radiation Oncology Biol. Phys., Vol. 45, No. 2, pp. 515–520, 1999 Copyright © 1999 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/99/$–see front matter
PII S0360-3016(99)00203-5
PHYSICS CONTRIBUTION
DETERMINING THE OPTIMAL BLOCK MARGIN ON THE PLANNING TARGET VOLUME FOR EXTRACRANIAL STEREOTACTIC RADIOTHERAPY ROBERT M. CARDINALE, M.D., QIUWEN WU, PH.D., STANLEY H. BENEDICT, PH.D., BRIAN D. KAVANAGH, M.D., EDWARD BUMP, B.A., AND RADHE MOHAN, PH.D. Department of Radiation Oncology, Medical College of Virginia Hospitals, Virginia Commonwealth University, Richmond, VA Purpose: To determine the block margin that minimizes normal tissue irradiation outside of the planning target volume (PTV) for body stereotactic radiotherapy (Body-SRT) of lung and liver tumors. Methods and Materials: Representative patient cases of lung and liver tumors were chosen for analysis. A PTV was constructed for each case and plans were generated which employed an array of block margins ranging from ⴚ2.5 mm to 10 mm at isocenter. Plans were generated for cerrobend blocks and for a multileaf collimator. The prescription isodose coverage was renormalized for each case and dose–volume histograms (DVH) and normal tissue complication probabilities (NTCP) were determined for each plan. Results and Conclusion: For the cases studied, the optimal block margin was in the 0.0 mm range. The ranking of plans was identical for both dose–volume based and biological based criteria. The method of blocking had no significant effect on treatment plans. The use of narrow margins for Body-SRT results in normal tissue sparing and creates significant target dose inhomogeneity which may be beneficial for tumor control. © 1999 Elsevier Science Inc. Stereotactic radiosurgery, Body radiosurgery, 3-D treatment planning, Multileaf collimator.
delivery can be done with a variety of methods. At our institution, we employ multiple static noncoplanar beam arrangements shaped by cerrobend blocks or a micro-multileaf collimator system (MLC), which is commercially available (3). The purposes of the current study were to determine the optimal block margin for minimizing normal tissue irradiation outside of the planning target volume (PTV), and to compare cerrobend blocked with MLC fields on test cases in the lung and liver.
INTRODUCTION The application of stereotactic radiotherapy to locations outside of the cranium is now possible because of the availability of stereotactic body frames and three-dimensional treatment planning systems. The purpose of body stereotactic radiotherapy (Body-SRT) is to deliver high tumor doses while sparing surrounding normal tissues as much as possible. Body-SRT can be used as primary or boost therapy given in one or more treatment sessions. Body-SRT shares many features with traditional cranial radiosurgery. They both rely on three-dimensional treatment planning tools, use stereotactic coordinates for localization and set-up, require accurate immobilization techniques, and in general attempt to ablate target tissues within the prescription isodose volume while sparing surrounding nearby and distant normal tissues. Tumors in the body, however, are more difficult to immobilize and require an extra margin for motion and set-up error as compared to cranial radiosurgery. Initial promising results of Body-SRT have been reported by Blomgren et al. (1, 2) in treating patients with lung and liver tumors. Although some immobilization devices for Body-SRT treatment delivery are commercially available, most of the techniques are in the developmental stages. Physical dose
METHODS AND MATERIALS Target volume construction Two adult patients, one with a solitary liver metastases (liver case) and another with a lung tumor (lung case), who were eligible for an institutionally approved protocol of Body-SRT treatment were chosen for analysis. These patients represent typical cases for which we consider BodySRT treatment in regard to tumor size and anatomic location. The patients had previously undergone computed tomography (CT) simulation, and the entire chest and abdomen were scanned with 5-mm slices. The CT scans were downloaded into the ADAC Pinnacle-3 treatment planning system (4).
Supported in part by NCI Grant R01- CA74043. Reprint requests to: Robert M. Cardinale, M.D., Department of Radiation Oncology, MCV Hospitals, 401 College Street, Rich-
mond, VA 23298-0058. Accepted for publication 14 May 1999. 515
516
I. J. Radiation Oncology
●
Biology
●
Physics
The gross tumor volume (GTV), clinical tumor volume (CTV), and PTV were defined according to ICRU Report 50 (5). Specifically, the GTV was constructed by outlining the visualized macroscopic tumor as seen on consecutive contrast-enhanced CT scan slices. The CTV was made equal to the GTV because no margin was added for subclinical extensions of tumor. The PTV was defined as a geometric volume used for treatment planning and for specification of dose that takes into account target motion and set-up uncertainty. The PTV was constructed by a three-dimensional volume expansion, which added 5 mm to the GTV surface in all directions. In our clinical practice of Body-SRT we add 5–12 mm in the superior–inferior direction depending on a target motion analysis done at the time of CT simulation. The specific amount of GTV expansion used to derive the PTV in this study should have no significant effect on the results, since we are focused on the normal tissue dose beyond a given PTV volume. Lung and liver organ motion due to diaphragm movement has been demonstrated to be substantially decreased by the use of a stereotactic body frame employing positive abdominal pressure. Lax et al. (6) have demonstrated the reproducibility and target motion accuracy to be within 5– 8 mm for most cases. The PTVs for the lung and liver target were 23 and 11 cm3 respectively. Anatomic organs were also outlined on each CT slice including liver, ipsilateral lung, spinal cord, kidneys, intestine, and heart. Three-dimensional (3D) treatment plan generation Treatment planning used “beams eye view” blocking in which each beam aperture shape was constructed with continuous custom cut cerrobend (BLOCKS) or a MLC. Blocking conformed to the outline of the PTV with a specified margin at isocenter in the plane perpendicular to the central ray of the beam. The ADAC Pinnacle-3 treatment planning system was used to generate the blocked conformal fields and to calculate dose distributions using 6 MV photons. All blocks were placed at the standard tray distance. Multiple isocentric noncoplanar static fields were arranged to minimize normal tissue irradiation outside of the PTV in the treated organ and to avoid nearby critical anatomic structures. The number, direction, and weighting of the treatment beams were chosen by a trial-and-error iterative process based on dose volume information of the PTV and normal tissue structures. Seven and 10 nonopposing beams were used for the lung and liver cases respectively as specified in Table 1. The treatment planning system utilizes an adaptive convolution algorithm for calculating doses. The overall capabilities of this treatment calculation technique and its accuracy have been described in detail (7, 8). For each of the two targets, plans were generated with block margins of 10, 5, 2.5, 0, and ⫺2.5 mm from the PTV at isocenter. All plans were normalized such that the prescription isodose volume covered exactly 99% of the PTV in each case. This coverage was chosen because of the increased tissue volumes that must be irradiated to cover the corners of the PTV on each consecutive CT slice if 100%
Volume 45, Number 2, 1999
Table 1. Beam geometries for lung and liver cases Beam
Couch angle (degrees)
Gantry angle (degrees)
I. Lung case
1 2 3 4 5 6 7
0 0 0 45 45 315 315
350 60 130 50 310 50 310
II. Liver case
1 2 3 4 5 6 7 8 9 10
0 0 0 0 0 90 90 90 50 230
200 160 40 260 320 315 45 155 45 45
coverage is required. The plans were generated with both BLOCKS and MLC. The MLC used (BrainLAB, Inc.) is a detachable design accessory, with 52 tungsten leaves of varying widths (3). The central 14 leaves have a width of 3 mm at isocenter. For MLC plans, the chosen margin corresponded to the position of the middle portion of the leaf at isocenter such that an equal area of each leaf end was inside and outside of the margin specified.
Dose–volume based plan comparison Since the goal of Body-SRT treatment is to “ablate” tissues within the PTV, these tissues were not considered at risk for complications in our analysis. Therefore, dose inhomogeneity inside the PTV of any degree was considered acceptable and not considered a priority in plan design. For each plan, dose–volume histograms (DVH) were calculated for the PTV and non-PTV tissue volumes. A partial nonPTV tissue volume DVH was also obtained which extended 2 cm from the PTV surface in all directions. For the lung tumor, the only nontarget tissue receiving a significant dose (⬎20% of prescription dose) was the surrounding ipsilateral lung, and likewise for the liver tumor only the surrounding nontarget liver was at risk. For each plan the prescription isodose (20 Gy delivered in one fraction) covered exactly 99% of the PTV and this isodose surface was normalized to 100%. DVHs were obtained using trilinear interpolation of dose at each CT pixel in the volumes of interest. The DVH bin size used was 10 cGy and the dose calculation grid was 0.2 ⫻ 0.2 ⫻ 0.5 cm3. The following data were calculated for each plan: maximum dose divided by prescription dose (MDPD), the volume of the prescription isodose surface divided by the PTV volume (PITV), and the volume of non-PTV tissue contained in the 100, 90, 80, 50, and 25% of prescription isodose volumes.
Block margin determination for Body-SRT
●
R. M. CARDINALE et al.
517
Fig. 1. Central-axis treatment plans using 0-mm block margin. The 100, 90, 80, 50, and 25% of prescription isodoses are shown.
Normal tissue complication probabilities (NTCP) based plan comparison DVH data often have only an indirect relationship for predicting treatment outcomes. Biologically based models have been used with increasing frequency to rank radiotherapeutic techniques or individual treatment plans as an adjunct to physical dose volume based information (9). Since the clinical endpoints have not been realized, and the models are often based on crude estimates of biologic parameters, the absolute NTCP values have to be interpreted with caution when making clinical judgments. We have used the NTCP analysis as an adjunct to dose-based criteria solely as a ranking tool. NTCPs were computed for all BLOCK plans using the cumulative DVH data sets as described by Lyman (10). For the lung case, NTCPs were calculated for the entire lung volume outside the PTV using the following parameters as derived from Burman (11): D50 ⫽ 2450 cGy, slope factor ⫽ 0.18, and volume factor ⫽ 0.87. For the liver tumor case NTCPs of non-PTV liver were computed with the following parameters: D50 ⫽ 4000cGy, slope factor ⫽ 0.15, and volume factor ⫽ 0.32. Since Body-SRT is PTVablative therapy we chose not to include the volume of normal tissue between the GTV and PTV in our calculation.
Because parallel organ NTCP calculations have a large volume dependence and the volume of normal tissue inside the PTV is very small compared to the whole organ volume, omitting this volume from the calculation should have no significant impact on plan ranking. Kutcher has demonstrated that in general the ranking of plans is also not significantly affected by the specific choice of D50, slope factor, and volume factor parameters (12, 13). The Lyman model is based on standard fractionated radiotherapy. For purposes of ranking plans, doses were chosen so that the best plan for each target had a resulting NTCP value of 5%. These doses were 120 Gy in 2 Gy fractions for the lung case and 108 Gy in 2 Gy fractions for the liver case.
RESULTS Figure 1 shows the central-axis isodose distribution using a block margin of 0.0 mm for the lung and liver plans. Table 2 shows the MDPD, PITV, and non-PTV volumes encompassed in various percent of prescription isodose volumes for the selected block margins. These results are for both BLOCK and MLC plans. Figures 2–5 show DVHs for PTV and non-PTV tissues for 0-mm and 10-mm plans.
Table 2. Lung tumor MDPD, PITV, and cm3 of non-PTV lung tissue contained in the 100, 90, 80, 50, and 25% isodose volumes for different block margins Blocks
MLC
Margin (mm)
MDPD
PITV
100%
90%
80%
50%
25%
MDPD
PITV
100%
90%
80%
50%
25%
10 5 2.5 0.0 ⫺2.5
1.3 1.4 1.5 1.6 1.9
2.2 1.6 1.6 1.4 1.4
27 14 13 9 10
58 32 25 18 18
86 49 40 29 31
205 140 123 105 113
525 424 400 387 444
1.3 1.4 1.5 1.6 1.9
2.2 1.7 1.6 1.4 1.5
28 16 13 10 11
6 34 25 19 19
91 53 39 31 32
214 144 125 105 114
547 436 415 391 445
518
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 45, Number 2, 1999
Fig. 2. DVH of the lung case PTV for treatment plans with 0 and 10-mm block margins.
Fig. 4. DVH of the liver case PTV for treatment plans with 0 and 10-mm block margins.
Table 3 shows NTCP values for non-PTV normal organ tissue for the BLOCK plans.
Liver tumor case As can be seen in Table 3, the MDPD increases with decreasing block margin from 1.1 to 2.3. The 2.5 and 0.0 mm block margin plans are superior in terms of PITV and normal tissue irradiation volumes for both BLOCK and MLC plans. The ⫺2.5 mm plan includes a larger volume of tissue at low isodose levels (⬍25%). The MLC plans are similar to the BLOCK plans for all chosen margins. In summary, plans improve with decreasing block margin to the 0.0 to 2.5 mm range.
Lung tumor case As can be seen in Table 2, the MDPD increases with decreasing block margin from 1.3 to 1.9. The 0.0 mm block margin is the best plan in terms of PITV and normal tissue irradiation volumes at all chosen isodose levels (25–100%) for both BLOCK and MLC plans. The ⫺2.5-mm plan is the second best plan in terms of PITV and high isodose normal lung volumes, but this plan irradiates more lung in the 25% isodose volume as compared to the 0.0 mm plan. The MLC plans are nearly identical to the BLOCK plans for all chosen margins. In summary, the plans tend to improve with tighter block margins down to 0.0 mm and worsen slightly at ⫺2.5 mm.
Fig. 3. DVH of the partial non-PTV lung (extending 2 cm from PTV surface) for treatment plans with 0 and10-mm block margins.
NTCP comparison The complication probabilities are shown in Table 4 for each target/margin combination. The results parallel the dose–volume comparison above, and demonstrate improv-
Fig. 5. DVH of the partial non-PTV liver (extending 2 cm from PTV surface) for treatment plans with 0 and 10-mm block margins.
Block margin determination for Body-SRT
●
R. M. CARDINALE et al.
519
Table 3. Liver tumor PITV, MDPD, and cm3 of non-PTV liver tissue contained in the 100, 90, 80, 50, and 25% isodose volumes for different block margins Blocks
MLC
Margin (mm)
MDPD
PITV
100%
90%
80%
50%
25%
MDPD
PITV
100%
90%
80%
50%
25%
10 5 2.5 0.0 ⫺2.5
1.1 1.1 1.3 1.6 2.3
2.9 1.7 1.5 1.5 1.6
20 7 5 5 6
42 15 11 8 9
56 23 15 14 14
119 59 44 45 47
338 199 163 170 214
1.1 1.1 1.2 1.6 2.3
3.0 1.7 1.4 1.6 1.6
21 7 4 6 6
43 16 10 9 10
59 23 14 13 15
124 60 43 41 48
355 205 158 177 211
ing NTCP with decreasing block margins to 0.0 mm. The NTCP increases with ⫺2.5-mm margins, but is still superior to 5-mm margin plans. DISCUSSION The PTV for Body-SRT takes into account target motion and set-up error. Once the PTV is defined, the planner attempts to cover the PTV with the prescription isodose surface as is done with standard radiation therapy delivery. For standard radiation therapy delivery an additional block margin is added to the PTV to account for penumbra effects and to ensure that an “acceptable” target dose homogeneity requirement is satisfied. Typically, this margin is in the 5–15 mm range. However, for Body-SRT, the goal of treatment is to destroy all of the tumor/stromal cells within the PTV and the requirement for a high degree of target dose homogeneity is of less importance. For the two test cases chosen, we have systematically studied various block margins to determine which range would lead to a greater sparing of tissues outside of the PTV while maintaining the same peripheral target dose in each case. The optimal block margin on the PTV with the treatment technique used is 0.0 mm for the lung case and between 0.0 and 2.5 mm for the liver case. It is difficult to arrive at a general mathematical solution to determine block margin because many parameters can affect the result such as beam number/orientation, anatomic location of the target, target size, treatment method, etc. We know from cranial radiosurgery that DVHs may often look similar, but on careful analysis of high-dose regions of normal tissues, significant differences can arise which may be the most important determinant of complication risk (9). Lax et al. have described a method of delivery of BodySRT for malignancies of the abdomen (6). They perform
Table 4. Normal tissue complication probabilities Margin (mm)
Lung case NTCP (%)
Liver case NTCP (%)
10 5 2.5 0.0 ⫺2.5
46 13 9 5 6
62 15 5 5 10
treatment with a limited number of fixed beams with “beams eye view” blocking. They favor highly heterogeneous dose distributions because of the possibility that improved treatment results may be obtained if the central portions of the target receive a higher dose than the periphery. They studied the impact of a highly heterogeneous dose distribution on a spherical target for 21 MV photon beams. The dose distributions are created by using beams with apertures at isocenter that are smaller than the target. The prescription isodose is chosen based on the dose that covers the periphery of the PTV. They show, in a theoretical study, that the dose for beam radii of 0.8R and 0.9R deliver a higher central target dose than for beam radii of R, whereas the dose in the volume outside the target is only marginally increased. This magnitude of increased normal tissue dose is not quantified, however. They do not state the exact beam aperture radius that is actually used in clinical practice, but they state that typically, the central parts of the PTV usually receive a dose that is 50% higher than the prescribed dose at the periphery (MDPD ⫽ 1.5). This degree of inhomogeneity is very similar to our results. The approach taken by Lax assumes that the “marginally increased” dose outside of the target is outweighed by an increase central target dose for a given peripheral target dose. Our approach to choosing block margin focuses on the dose outside of the target and does not assume that an increase in target inhomogeneity will necessarily lead to improved tumor control and that this difference outweighs increased normal tissue irradiation. There are various arguments as to the potential benefit of target dose inhomogeneity in regard to tumor control. When we placed the block margin within the PTV, our resulting MDPDs were 1.9 and 2.3. This margin also increased normal tissue doses and NTCP. We have shown, for the test cases chosen, that there is no difference in treatment plans between MLC or BLOCK plans. BLOCK plans, as generated by the treatment planning computer, assume ideal construction and mounting which would be the case for an “ideal” multileaf collimator. BLOCK fabrication and mounting is labor-intensive and may result in some uncertainty in precision therapy as compared to MLC treatment delivery. Another potential advantage of MLC is that it can be used to deliver nonuniform dose distributions through beam intensity modulation.
520
I. J. Radiation Oncology
●
Biology
●
Physics
We have shown that intensity modulation results in improved dose delivery for selected intracranial targets as compared to non-IMRT treatment methods and we are exploring its use for Body-SRT (14). For thoracic and liver tumors treated with Body-SRT, we
Volume 45, Number 2, 1999
recommend that users analyze three treatment plans with margins of ⫺2.5 mm, 0 mm, and 2.5 mm and select the plan that minimizes non-PTV tissue irradiation. The resultant plans in each case will have marked target inhomogeneity, which may contribute to increased tumor control.
REFERENCES 1. Blomgren H, Lax I, Naslund I, Svanstrom R. Stereotactic high dose fraction radiation therapy of extracranial tumors using an accelerator. Acta Oncol 1995;6:861– 870. 2. Blomgren H, Lax I, Goranson H, et al. Radiosurgery for tumors in the body: Clinical experience using a new method. J Radiosurg 1998;1:63–74. 3. M3 micro-multileaf collimator users guide. Palo Alto, CA: BrainLAB USA, Inc. 1998. 4. Pinnacle-3 users guide, version 3.0d. Milpitas, CA: ADAC Laboratories; 1997. 5. International Commission on Radiation Units and Measurements. ICRU Report 50. Prescribing, recording, and reporting photon beam therapy. Bethesda, MD: ICRU. 1993. 6. Lax I, Blomgren H, Naslund I, Svanstrom R. Stereotactic radiotherapy of malignancies of the abdomen. Acta Oncol 1994;6:677– 683. 7. Mackie TR, Scrimger JW, Battista JJ. A convolution method of calculating dose for 15 MV x rays. Med Phys 1985;12:189 – 197. 8. Liu HH, Mackie TR, McCullough EC. A dual source photon beam model used in convolution/superposition dose calcula-
9. 10. 11. 12.
13.
14.
tions for clinical megavoltage x-ray beams. Med Phys 1997; 24:1960 –1974. Smith V, Verhey L, Serago CF. Comparison of radiosurgery treatment modalities based on complication probabilities. Int J Radiat Oncol Biol Phys 1998;40:507–513. Lyman JT. Complication probability as assessed from dose– volume histograms. Rad Res 1985;104:S13–S19. Burman C, Kutcher GJ, Emami B, et al. Fitting of normal tissue tolerance data to an analytical function. Int J Radiat Oncol Biol Phys 1991;21:123–135. Jackson A, Kutcher GJ, Yorke ED. Probability of radiation induced complications for normal tissues with parallel architecture subject to non-uniform irradiation. Med Phys 993;20: 613– 625. Kutcher GJ, Burman C. Calculation of complication probability factors for non-uniform normal tissue irradiation: The effective volume method. Int J Radiat Onc Biol Phys 1989; 16:1623–1630. Cardinale RM, Benedict SH, Wu Q. A comparison of three stereotactic radiotherapy techniques: Arcs vs. noncoplanar fixed fields vs. intensity modulation. Int J Radiat Oncol Biol Phys 1998;42:431– 436.
PII S0360-3016(99)00203-5
PHYSICS CONTRIBUTION
DETERMINING THE OPTIMAL BLOCK MARGIN ON THE PLANNING TARGET VOLUME FOR EXTRACRANIAL STEREOTACTIC RADIOTHERAPY ROBERT M. CARDINALE, M.D., QIUWEN WU, PH.D., STANLEY H. BENEDICT, PH.D., BRIAN D. KAVANAGH, M.D., EDWARD BUMP, B.A., AND RADHE MOHAN, PH.D. Department of Radiation Oncology, Medical College of Virginia Hospitals, Virginia Commonwealth University, Richmond, VA Purpose: To determine the block margin that minimizes normal tissue irradiation outside of the planning target volume (PTV) for body stereotactic radiotherapy (Body-SRT) of lung and liver tumors. Methods and Materials: Representative patient cases of lung and liver tumors were chosen for analysis. A PTV was constructed for each case and plans were generated which employed an array of block margins ranging from ⴚ2.5 mm to 10 mm at isocenter. Plans were generated for cerrobend blocks and for a multileaf collimator. The prescription isodose coverage was renormalized for each case and dose–volume histograms (DVH) and normal tissue complication probabilities (NTCP) were determined for each plan. Results and Conclusion: For the cases studied, the optimal block margin was in the 0.0 mm range. The ranking of plans was identical for both dose–volume based and biological based criteria. The method of blocking had no significant effect on treatment plans. The use of narrow margins for Body-SRT results in normal tissue sparing and creates significant target dose inhomogeneity which may be beneficial for tumor control. © 1999 Elsevier Science Inc. Stereotactic radiosurgery, Body radiosurgery, 3-D treatment planning, Multileaf collimator.
delivery can be done with a variety of methods. At our institution, we employ multiple static noncoplanar beam arrangements shaped by cerrobend blocks or a micro-multileaf collimator system (MLC), which is commercially available (3). The purposes of the current study were to determine the optimal block margin for minimizing normal tissue irradiation outside of the planning target volume (PTV), and to compare cerrobend blocked with MLC fields on test cases in the lung and liver.
INTRODUCTION The application of stereotactic radiotherapy to locations outside of the cranium is now possible because of the availability of stereotactic body frames and three-dimensional treatment planning systems. The purpose of body stereotactic radiotherapy (Body-SRT) is to deliver high tumor doses while sparing surrounding normal tissues as much as possible. Body-SRT can be used as primary or boost therapy given in one or more treatment sessions. Body-SRT shares many features with traditional cranial radiosurgery. They both rely on three-dimensional treatment planning tools, use stereotactic coordinates for localization and set-up, require accurate immobilization techniques, and in general attempt to ablate target tissues within the prescription isodose volume while sparing surrounding nearby and distant normal tissues. Tumors in the body, however, are more difficult to immobilize and require an extra margin for motion and set-up error as compared to cranial radiosurgery. Initial promising results of Body-SRT have been reported by Blomgren et al. (1, 2) in treating patients with lung and liver tumors. Although some immobilization devices for Body-SRT treatment delivery are commercially available, most of the techniques are in the developmental stages. Physical dose
METHODS AND MATERIALS Target volume construction Two adult patients, one with a solitary liver metastases (liver case) and another with a lung tumor (lung case), who were eligible for an institutionally approved protocol of Body-SRT treatment were chosen for analysis. These patients represent typical cases for which we consider BodySRT treatment in regard to tumor size and anatomic location. The patients had previously undergone computed tomography (CT) simulation, and the entire chest and abdomen were scanned with 5-mm slices. The CT scans were downloaded into the ADAC Pinnacle-3 treatment planning system (4).
Supported in part by NCI Grant R01- CA74043. Reprint requests to: Robert M. Cardinale, M.D., Department of Radiation Oncology, MCV Hospitals, 401 College Street, Rich-
mond, VA 23298-0058. Accepted for publication 14 May 1999. 515
516
I. J. Radiation Oncology
●
Biology
●
Physics
The gross tumor volume (GTV), clinical tumor volume (CTV), and PTV were defined according to ICRU Report 50 (5). Specifically, the GTV was constructed by outlining the visualized macroscopic tumor as seen on consecutive contrast-enhanced CT scan slices. The CTV was made equal to the GTV because no margin was added for subclinical extensions of tumor. The PTV was defined as a geometric volume used for treatment planning and for specification of dose that takes into account target motion and set-up uncertainty. The PTV was constructed by a three-dimensional volume expansion, which added 5 mm to the GTV surface in all directions. In our clinical practice of Body-SRT we add 5–12 mm in the superior–inferior direction depending on a target motion analysis done at the time of CT simulation. The specific amount of GTV expansion used to derive the PTV in this study should have no significant effect on the results, since we are focused on the normal tissue dose beyond a given PTV volume. Lung and liver organ motion due to diaphragm movement has been demonstrated to be substantially decreased by the use of a stereotactic body frame employing positive abdominal pressure. Lax et al. (6) have demonstrated the reproducibility and target motion accuracy to be within 5– 8 mm for most cases. The PTVs for the lung and liver target were 23 and 11 cm3 respectively. Anatomic organs were also outlined on each CT slice including liver, ipsilateral lung, spinal cord, kidneys, intestine, and heart. Three-dimensional (3D) treatment plan generation Treatment planning used “beams eye view” blocking in which each beam aperture shape was constructed with continuous custom cut cerrobend (BLOCKS) or a MLC. Blocking conformed to the outline of the PTV with a specified margin at isocenter in the plane perpendicular to the central ray of the beam. The ADAC Pinnacle-3 treatment planning system was used to generate the blocked conformal fields and to calculate dose distributions using 6 MV photons. All blocks were placed at the standard tray distance. Multiple isocentric noncoplanar static fields were arranged to minimize normal tissue irradiation outside of the PTV in the treated organ and to avoid nearby critical anatomic structures. The number, direction, and weighting of the treatment beams were chosen by a trial-and-error iterative process based on dose volume information of the PTV and normal tissue structures. Seven and 10 nonopposing beams were used for the lung and liver cases respectively as specified in Table 1. The treatment planning system utilizes an adaptive convolution algorithm for calculating doses. The overall capabilities of this treatment calculation technique and its accuracy have been described in detail (7, 8). For each of the two targets, plans were generated with block margins of 10, 5, 2.5, 0, and ⫺2.5 mm from the PTV at isocenter. All plans were normalized such that the prescription isodose volume covered exactly 99% of the PTV in each case. This coverage was chosen because of the increased tissue volumes that must be irradiated to cover the corners of the PTV on each consecutive CT slice if 100%
Volume 45, Number 2, 1999
Table 1. Beam geometries for lung and liver cases Beam
Couch angle (degrees)
Gantry angle (degrees)
I. Lung case
1 2 3 4 5 6 7
0 0 0 45 45 315 315
350 60 130 50 310 50 310
II. Liver case
1 2 3 4 5 6 7 8 9 10
0 0 0 0 0 90 90 90 50 230
200 160 40 260 320 315 45 155 45 45
coverage is required. The plans were generated with both BLOCKS and MLC. The MLC used (BrainLAB, Inc.) is a detachable design accessory, with 52 tungsten leaves of varying widths (3). The central 14 leaves have a width of 3 mm at isocenter. For MLC plans, the chosen margin corresponded to the position of the middle portion of the leaf at isocenter such that an equal area of each leaf end was inside and outside of the margin specified.
Dose–volume based plan comparison Since the goal of Body-SRT treatment is to “ablate” tissues within the PTV, these tissues were not considered at risk for complications in our analysis. Therefore, dose inhomogeneity inside the PTV of any degree was considered acceptable and not considered a priority in plan design. For each plan, dose–volume histograms (DVH) were calculated for the PTV and non-PTV tissue volumes. A partial nonPTV tissue volume DVH was also obtained which extended 2 cm from the PTV surface in all directions. For the lung tumor, the only nontarget tissue receiving a significant dose (⬎20% of prescription dose) was the surrounding ipsilateral lung, and likewise for the liver tumor only the surrounding nontarget liver was at risk. For each plan the prescription isodose (20 Gy delivered in one fraction) covered exactly 99% of the PTV and this isodose surface was normalized to 100%. DVHs were obtained using trilinear interpolation of dose at each CT pixel in the volumes of interest. The DVH bin size used was 10 cGy and the dose calculation grid was 0.2 ⫻ 0.2 ⫻ 0.5 cm3. The following data were calculated for each plan: maximum dose divided by prescription dose (MDPD), the volume of the prescription isodose surface divided by the PTV volume (PITV), and the volume of non-PTV tissue contained in the 100, 90, 80, 50, and 25% of prescription isodose volumes.
Block margin determination for Body-SRT
●
R. M. CARDINALE et al.
517
Fig. 1. Central-axis treatment plans using 0-mm block margin. The 100, 90, 80, 50, and 25% of prescription isodoses are shown.
Normal tissue complication probabilities (NTCP) based plan comparison DVH data often have only an indirect relationship for predicting treatment outcomes. Biologically based models have been used with increasing frequency to rank radiotherapeutic techniques or individual treatment plans as an adjunct to physical dose volume based information (9). Since the clinical endpoints have not been realized, and the models are often based on crude estimates of biologic parameters, the absolute NTCP values have to be interpreted with caution when making clinical judgments. We have used the NTCP analysis as an adjunct to dose-based criteria solely as a ranking tool. NTCPs were computed for all BLOCK plans using the cumulative DVH data sets as described by Lyman (10). For the lung case, NTCPs were calculated for the entire lung volume outside the PTV using the following parameters as derived from Burman (11): D50 ⫽ 2450 cGy, slope factor ⫽ 0.18, and volume factor ⫽ 0.87. For the liver tumor case NTCPs of non-PTV liver were computed with the following parameters: D50 ⫽ 4000cGy, slope factor ⫽ 0.15, and volume factor ⫽ 0.32. Since Body-SRT is PTVablative therapy we chose not to include the volume of normal tissue between the GTV and PTV in our calculation.
Because parallel organ NTCP calculations have a large volume dependence and the volume of normal tissue inside the PTV is very small compared to the whole organ volume, omitting this volume from the calculation should have no significant impact on plan ranking. Kutcher has demonstrated that in general the ranking of plans is also not significantly affected by the specific choice of D50, slope factor, and volume factor parameters (12, 13). The Lyman model is based on standard fractionated radiotherapy. For purposes of ranking plans, doses were chosen so that the best plan for each target had a resulting NTCP value of 5%. These doses were 120 Gy in 2 Gy fractions for the lung case and 108 Gy in 2 Gy fractions for the liver case.
RESULTS Figure 1 shows the central-axis isodose distribution using a block margin of 0.0 mm for the lung and liver plans. Table 2 shows the MDPD, PITV, and non-PTV volumes encompassed in various percent of prescription isodose volumes for the selected block margins. These results are for both BLOCK and MLC plans. Figures 2–5 show DVHs for PTV and non-PTV tissues for 0-mm and 10-mm plans.
Table 2. Lung tumor MDPD, PITV, and cm3 of non-PTV lung tissue contained in the 100, 90, 80, 50, and 25% isodose volumes for different block margins Blocks
MLC
Margin (mm)
MDPD
PITV
100%
90%
80%
50%
25%
MDPD
PITV
100%
90%
80%
50%
25%
10 5 2.5 0.0 ⫺2.5
1.3 1.4 1.5 1.6 1.9
2.2 1.6 1.6 1.4 1.4
27 14 13 9 10
58 32 25 18 18
86 49 40 29 31
205 140 123 105 113
525 424 400 387 444
1.3 1.4 1.5 1.6 1.9
2.2 1.7 1.6 1.4 1.5
28 16 13 10 11
6 34 25 19 19
91 53 39 31 32
214 144 125 105 114
547 436 415 391 445
518
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 45, Number 2, 1999
Fig. 2. DVH of the lung case PTV for treatment plans with 0 and 10-mm block margins.
Fig. 4. DVH of the liver case PTV for treatment plans with 0 and 10-mm block margins.
Table 3 shows NTCP values for non-PTV normal organ tissue for the BLOCK plans.
Liver tumor case As can be seen in Table 3, the MDPD increases with decreasing block margin from 1.1 to 2.3. The 2.5 and 0.0 mm block margin plans are superior in terms of PITV and normal tissue irradiation volumes for both BLOCK and MLC plans. The ⫺2.5 mm plan includes a larger volume of tissue at low isodose levels (⬍25%). The MLC plans are similar to the BLOCK plans for all chosen margins. In summary, plans improve with decreasing block margin to the 0.0 to 2.5 mm range.
Lung tumor case As can be seen in Table 2, the MDPD increases with decreasing block margin from 1.3 to 1.9. The 0.0 mm block margin is the best plan in terms of PITV and normal tissue irradiation volumes at all chosen isodose levels (25–100%) for both BLOCK and MLC plans. The ⫺2.5-mm plan is the second best plan in terms of PITV and high isodose normal lung volumes, but this plan irradiates more lung in the 25% isodose volume as compared to the 0.0 mm plan. The MLC plans are nearly identical to the BLOCK plans for all chosen margins. In summary, the plans tend to improve with tighter block margins down to 0.0 mm and worsen slightly at ⫺2.5 mm.
Fig. 3. DVH of the partial non-PTV lung (extending 2 cm from PTV surface) for treatment plans with 0 and10-mm block margins.
NTCP comparison The complication probabilities are shown in Table 4 for each target/margin combination. The results parallel the dose–volume comparison above, and demonstrate improv-
Fig. 5. DVH of the partial non-PTV liver (extending 2 cm from PTV surface) for treatment plans with 0 and 10-mm block margins.
Block margin determination for Body-SRT
●
R. M. CARDINALE et al.
519
Table 3. Liver tumor PITV, MDPD, and cm3 of non-PTV liver tissue contained in the 100, 90, 80, 50, and 25% isodose volumes for different block margins Blocks
MLC
Margin (mm)
MDPD
PITV
100%
90%
80%
50%
25%
MDPD
PITV
100%
90%
80%
50%
25%
10 5 2.5 0.0 ⫺2.5
1.1 1.1 1.3 1.6 2.3
2.9 1.7 1.5 1.5 1.6
20 7 5 5 6
42 15 11 8 9
56 23 15 14 14
119 59 44 45 47
338 199 163 170 214
1.1 1.1 1.2 1.6 2.3
3.0 1.7 1.4 1.6 1.6
21 7 4 6 6
43 16 10 9 10
59 23 14 13 15
124 60 43 41 48
355 205 158 177 211
ing NTCP with decreasing block margins to 0.0 mm. The NTCP increases with ⫺2.5-mm margins, but is still superior to 5-mm margin plans. DISCUSSION The PTV for Body-SRT takes into account target motion and set-up error. Once the PTV is defined, the planner attempts to cover the PTV with the prescription isodose surface as is done with standard radiation therapy delivery. For standard radiation therapy delivery an additional block margin is added to the PTV to account for penumbra effects and to ensure that an “acceptable” target dose homogeneity requirement is satisfied. Typically, this margin is in the 5–15 mm range. However, for Body-SRT, the goal of treatment is to destroy all of the tumor/stromal cells within the PTV and the requirement for a high degree of target dose homogeneity is of less importance. For the two test cases chosen, we have systematically studied various block margins to determine which range would lead to a greater sparing of tissues outside of the PTV while maintaining the same peripheral target dose in each case. The optimal block margin on the PTV with the treatment technique used is 0.0 mm for the lung case and between 0.0 and 2.5 mm for the liver case. It is difficult to arrive at a general mathematical solution to determine block margin because many parameters can affect the result such as beam number/orientation, anatomic location of the target, target size, treatment method, etc. We know from cranial radiosurgery that DVHs may often look similar, but on careful analysis of high-dose regions of normal tissues, significant differences can arise which may be the most important determinant of complication risk (9). Lax et al. have described a method of delivery of BodySRT for malignancies of the abdomen (6). They perform
Table 4. Normal tissue complication probabilities Margin (mm)
Lung case NTCP (%)
Liver case NTCP (%)
10 5 2.5 0.0 ⫺2.5
46 13 9 5 6
62 15 5 5 10
treatment with a limited number of fixed beams with “beams eye view” blocking. They favor highly heterogeneous dose distributions because of the possibility that improved treatment results may be obtained if the central portions of the target receive a higher dose than the periphery. They studied the impact of a highly heterogeneous dose distribution on a spherical target for 21 MV photon beams. The dose distributions are created by using beams with apertures at isocenter that are smaller than the target. The prescription isodose is chosen based on the dose that covers the periphery of the PTV. They show, in a theoretical study, that the dose for beam radii of 0.8R and 0.9R deliver a higher central target dose than for beam radii of R, whereas the dose in the volume outside the target is only marginally increased. This magnitude of increased normal tissue dose is not quantified, however. They do not state the exact beam aperture radius that is actually used in clinical practice, but they state that typically, the central parts of the PTV usually receive a dose that is 50% higher than the prescribed dose at the periphery (MDPD ⫽ 1.5). This degree of inhomogeneity is very similar to our results. The approach taken by Lax assumes that the “marginally increased” dose outside of the target is outweighed by an increase central target dose for a given peripheral target dose. Our approach to choosing block margin focuses on the dose outside of the target and does not assume that an increase in target inhomogeneity will necessarily lead to improved tumor control and that this difference outweighs increased normal tissue irradiation. There are various arguments as to the potential benefit of target dose inhomogeneity in regard to tumor control. When we placed the block margin within the PTV, our resulting MDPDs were 1.9 and 2.3. This margin also increased normal tissue doses and NTCP. We have shown, for the test cases chosen, that there is no difference in treatment plans between MLC or BLOCK plans. BLOCK plans, as generated by the treatment planning computer, assume ideal construction and mounting which would be the case for an “ideal” multileaf collimator. BLOCK fabrication and mounting is labor-intensive and may result in some uncertainty in precision therapy as compared to MLC treatment delivery. Another potential advantage of MLC is that it can be used to deliver nonuniform dose distributions through beam intensity modulation.
520
I. J. Radiation Oncology
●
Biology
●
Physics
We have shown that intensity modulation results in improved dose delivery for selected intracranial targets as compared to non-IMRT treatment methods and we are exploring its use for Body-SRT (14). For thoracic and liver tumors treated with Body-SRT, we
Volume 45, Number 2, 1999
recommend that users analyze three treatment plans with margins of ⫺2.5 mm, 0 mm, and 2.5 mm and select the plan that minimizes non-PTV tissue irradiation. The resultant plans in each case will have marked target inhomogeneity, which may contribute to increased tumor control.
REFERENCES 1. Blomgren H, Lax I, Naslund I, Svanstrom R. Stereotactic high dose fraction radiation therapy of extracranial tumors using an accelerator. Acta Oncol 1995;6:861– 870. 2. Blomgren H, Lax I, Goranson H, et al. Radiosurgery for tumors in the body: Clinical experience using a new method. J Radiosurg 1998;1:63–74. 3. M3 micro-multileaf collimator users guide. Palo Alto, CA: BrainLAB USA, Inc. 1998. 4. Pinnacle-3 users guide, version 3.0d. Milpitas, CA: ADAC Laboratories; 1997. 5. International Commission on Radiation Units and Measurements. ICRU Report 50. Prescribing, recording, and reporting photon beam therapy. Bethesda, MD: ICRU. 1993. 6. Lax I, Blomgren H, Naslund I, Svanstrom R. Stereotactic radiotherapy of malignancies of the abdomen. Acta Oncol 1994;6:677– 683. 7. Mackie TR, Scrimger JW, Battista JJ. A convolution method of calculating dose for 15 MV x rays. Med Phys 1985;12:189 – 197. 8. Liu HH, Mackie TR, McCullough EC. A dual source photon beam model used in convolution/superposition dose calcula-
9. 10. 11. 12.
13.
14.
tions for clinical megavoltage x-ray beams. Med Phys 1997; 24:1960 –1974. Smith V, Verhey L, Serago CF. Comparison of radiosurgery treatment modalities based on complication probabilities. Int J Radiat Oncol Biol Phys 1998;40:507–513. Lyman JT. Complication probability as assessed from dose– volume histograms. Rad Res 1985;104:S13–S19. Burman C, Kutcher GJ, Emami B, et al. Fitting of normal tissue tolerance data to an analytical function. Int J Radiat Oncol Biol Phys 1991;21:123–135. Jackson A, Kutcher GJ, Yorke ED. Probability of radiation induced complications for normal tissues with parallel architecture subject to non-uniform irradiation. Med Phys 993;20: 613– 625. Kutcher GJ, Burman C. Calculation of complication probability factors for non-uniform normal tissue irradiation: The effective volume method. Int J Radiat Onc Biol Phys 1989; 16:1623–1630. Cardinale RM, Benedict SH, Wu Q. A comparison of three stereotactic radiotherapy techniques: Arcs vs. noncoplanar fixed fields vs. intensity modulation. Int J Radiat Oncol Biol Phys 1998;42:431– 436.