- Email: [email protected]
Intensity-modulated stereotactic radiosurgery using dynamic micro-multileaf collimation
Int. J. Radiation Oncology Biol. Phys., Vol. 50, No. 3, pp. 751–758, 2001 Copyright © 2001 Elsevier Science Inc. Printed in the USA. All rights reserved 0360-3016/01/$–see front matter
PII S0360-3016(01)01487-0
CLINICAL INVESTIGATION
Brain
INTENSITY-MODULATED STEREOTACTIC RADIOSURGERY USING DYNAMIC MICRO-MULTILEAF COLLIMATION STANLEY H. BENEDICT, PH.D.,* ROBERT M. CARDINALE, M.D.,* QIUWEN WU, PH.D.,* ROBERT D. ZWICKER, PH.D.,* WILLIAM C. BROADDUS, M.D., PH.D.,† AND RADHE MOHAN, PH.D.* Departments of *Radiation Oncology and †Neurosurgery, Medical College of Virginia Hospitals of Virginia Commonwealth University, Richmond, VA Purpose: The implementation of dynamic leaf motion on a micro-multileaf collimator system provides the capability for intensity-modulated stereotactic radiosurgery (IMSRS), and the consequent potential for improved dose distributions for irregularly shaped tumor volumes adjacent to critical organs. This study explores the use of IMSRS to provide improved tumor coverage and normal tissue sparing for small cranial tumors relative to plans based on multiple fixed uniform-intensity beams or traditional circular collimator arc-based stereotactic techniques. Methods and Materials: Four patient cases involving small brain lesions are presented and analyzed. The cases were chosen to include a representative selection of target shapes, number of targets, and adjacent critical areas. Patient plans generated for these comparisons include standard arcs with multiple circular collimators, and fixed noncoplanar static fields with uniform-intensity beams and IMSRS. Parameters used for evaluation of the plans include the percentage of irradiated volume to tumor volume (PITV), normal tissue dose–volume histograms, and dose– homogeneity ratios. All IMSRS plans were computed using previously established IMRT techniques adapted for use with the BrainLAB M3 micro-multileaf collimator. The algorithms comprising the IMRT system for optimization of intensity distributions and conversion into leaf trajectories of the BrainLab M3 were developed at our institution. The ADAC Pinnacle3 radiation treatment-planning system was used for dose calculations and for input of contours for target volumes and normal critical structures. Results: For all cases, the IMSRS plans showed a high degree of conformity of the dose distribution with the target shape. The IMSRS plans provided either (1) a smaller volume of normal tissue irradiated to significant dose levels, generally taken as doses greater than 50% of the prescription, or (2) a lower dose to an important adjacent critical organ. The reduction in volume of normal tissue irradiated in the IMSRS plans ranged from 10% to 50% relative to the other arc and uniform fixed-field plans. Conclusion: The case studies presented for IMSRS demonstrate significant dosimetric improvements for small, irregularly shaped lesions of the brain when compared to treatments using multiple static fields or standard SRS arc techniques with circular collimators. For all cases, the IMSRS plan yielded a smaller volume of normal tissue irradiated, and/or a reduction in the volume of an adjacent critical organ (i.e., brainstem) irradiated to significant dose levels. © 2001 Elsevier Science Inc. Stereotactic radiosurgery, Intensity-modulated radiotherapy, Micro-multileaf collimation.
INTRODUCTION
tensity-modulated radiotherapy (IMRT) to produce dose distributions that are superior to conventional plans has been widely presented in the literature (6 –9). The majority of studies demonstrating improvement in dosimetry with IMRT have been for large field sizes using relatively large multileaf collimation systems, generally with leaves 1.0 cm in width. With the recent introduction of micro-multileaf collimator systems the advantages of IMRT may be further extended to the case of small intracranial targets (10). In a previous investigation, we demonstrated the dosimetric advantages of IMRT applied to intracranial targets which were relatively large (9.6 –36.7 cm3) and of various geometric
There have been numerous publications on the development of linac-based radiosurgery techniques, from the initial arcbased approach with circular collimators (1) to fixed-field arrangements and dynamic SRS (2, 3). Methods to evaluate radiosurgery techniques (4, 5) are also well established. The present work extends these studies by using clinical examples to confirm and quantify the advantages of a technique that combines the precision of stereotactic positioning with the dose-delivery capabilities of IMRT to treat small critically located targets. The potential of dynamic multileaf collimation and inReprint requests to: Dr. Stanley H. Benedict, Department of Radiation Oncology, Medical College of Virginia Hospitals of Virginia Commonwealth University, 401 College Street, Richmond, VA 23298-0058. E-mail: [email protected]
This work is supported by Grants CA74043 and R29 CA7643901A1 from the National Cancer Institute. Accepted for publication 24 January 2001. 751
752
I. J. Radiation Oncology
●
Biology
●
Physics
shapes, such as an ellipsoid and hemisphere (11). In this study, we present a selection of actual patient cases selected for radiosurgery with small intracranial structures (1.2–3.5 cm3) to demonstrate that IMRT using a set of fixed fields and delivered with dynamic micro-multileaf collimation leads to improved dose distributions when compared to multiple arcs or fixed uniform-intensity fields. The collimation system used for this investigation is the BrainLAB M3 micro-multileaf collimator (BrainLAB AG, Heimstetten, Germany), the characteristics of which have been previously described in the literature (12, 13). The primary innovative feature of the M3 is that it has leaves of variable width, which are narrower than those generally available as a standard feature on the current generation of accelerators. The M3 has 14 pairs of 0.3 cm, 6 pairs of 0.45 cm, and 6 pairs of 0.55 cm leaves. All of the IMRT computer algorithms used for optimization of intensity distributions and conversion into M3 leaf trajectories were developed at our institution (14). METHODS AND MATERIALS Overview For this study, 4 patient cases involving small brain lesions were selected, with a range of shapes, number of targets, and adjacent critical areas for investigation. All patients were immobilized, planned, and treated using established stereotactic techniques. Patient 1 presented with lung metastases to the brain with two foci 5 cm apart. Patient 2 had a recurrent ependymoma in the posterior fossa adjacent to brainstem. Patient 3 presented with sphenoid wing meningioma, and Patient 4 presented with an acoustic neuroma. For all 4 patients, stereotactic arc plans using circular cone collimators and plans based on noncoplanar static, uniform-intensity beams defined by the micro-multileaf collimator (micro-MLC) were generated and optimized to give a minimum percentage of irradiated-volume to tumor-volume ratio (PITV) and minimal normal tissue irradiation at high dose levels. The PITV and dose–volume histogram (DVH) to adjacent critical organs were then used to compare the plans with the corresponding IMSRS plans. Multiple IMSRS plans were generated for each patient using our in-house programs with varying beam configurations, the same configurations used in the fixed-field plans described below. For all of the plans, the dose homogeneity within the planning target volume (PTV) was assigned a low priority relative to target conformity and adjacent critical structure doses. The ADAC Pinnacle3 radiation treatment-planning system (ADAC Laboratories, Milpitas, CA) was used for all dose calculations and for input of the target and critical structure contours used by our in-house IMRT optimization program. The dose grid used for these calculations was 0.3 ⫻ 0.3 ⫻ 0.3 cm, a practical limitation imposed by the current computer hardware capabilities. For comparison purposes, all plans were prescribed to a dose of 10 Gy, with the requirement that 99% of the target volume receive the prescription dose or greater. For planning pur-
Volume 50, Number 3, 2001
poses, and in accordance with routine SRS procedures, the PTV was assigned the same volume as the gross tumor volume (GTV), with no further expansion. Fixed-field plan design For the fixed-field arrangement, an initial plan of 15 fields was set up for each plan. The design of the initial fixed-field arrangement was established to incorporate 3 fields each along 5 planes of the patient’s skull, which provides approximately 2 steradian solid-angle coverage. The beam arrangement utilized left and right laterals, left and right oblique fields, and vertex fields, designed for a minimum of beam opposition. Once the template of 15 fields was scripted to the patient, adjustments of ⫾ 10° were made in each of the couch/gantry orientations to minimize dose to adjacent critical areas. Beam weights were also optimized by trial and error to provide the prescribed target coverage while minimizing dose contributions to adjacent critical areas. Block margins for these beams were optimized to yield a minimum PITV, and were typically 0.0 – 0.3 cm for each beam. Block margin optimization for fixed fields has been previously investigated at our institution (15). The M3 collimator angle was also chosen such that the sum of the unblocked field areas external to the optimal block margin due to the leaf shapes, and the blocked areas internal to the margin is minimized, typically at a value of less than about 0.5 cm2. Arc-based plans with multiple circular collimators For evaluation of the linac-based stereotactic arc plans using multiple circular collimators, each isocenter was assigned 4 –5 arcs of approximately 100° per arc, in accordance with well-established techniques (1). A number of studies have been published that demonstrate the superiority of linac plans that use conformal fixed-field arrangements over those using arcs with multiple circular collimators (11, 16 –21). In the course of the present studies, this superiority was demonstrated for all patient cases investigated. For this reason, only the first patient case is presented with all three techniques, while the remaining cases compare only the fixed uniform-intensity fields with the IMRT fields. Intensity-modulated radiosurgery (IMSRS) methods The in-house IMRT optimization system was coupled to the ADAC Pinnacle3 three-dimensional (3D) dose calculation engine to generate the IMSRS treatment plans. All fields were divided into intensity matrices of size 0.1 cm ⫻ 0.3 cm to provide the highest resolution consistent with the geometry of the M3 collimator. The Pinnacle3 3D radiation treatment-planning system was used to input and display the target and critical structure contours and the beam parameters needed by the IMRT calculation routines, and also to generate the quantities used for the final evaluation of treatment plans. For the delivery of treatment, the leaf trajectories, in the form of leaf position data stored as a function of monitor units, obtained from the optimized beam intensity patterns, were transferred to the dynamic
Intensity-modulated stereotactic radiosurgery
multileaf collimator (DMLC) controller. This DMLC controller utilizes the driver software and was developed by the treatment machine manufacturer (Varian Medical Equipment, Palo Alto, CA) to move the leaves to their designated positions while the beam is on. After the fixed-field plan had been optimized, all of the beam geometry parameters were transferred to the IMRT program for intensity distribution optimization. No margin was assigned to the gross target volume for IMRT to minimize the volume of normal brain exposed to high doses. This requires accepting higher target dose inhomogeneity in the final plan. Once the IMRT technique was applied to the optimized 15 fixed-field plan, further iterations were applied to obtain a minimum score based on the penalties and parameters of the objective function of the program. The objective function for optimizing intensity distributions in these calculations was specified in terms of dosevolume limits. The criteria used in this study included the requirement that exactly 99% of the target volume should be enclosed in the prescription isodose surface. Penalties for dose inhomogeneity within the target were marginalized relative to those assigned to the index of conformity. In the absence of any strong data demonstrating the importance of dose homogeneity in these cases, hot spots within the target volume were assigned a low priority for our planning purposes. The dose to normal tissue was constrained to limit the volume allowed to receive a dose higher than a chosen dose limit without a high penalty. Such constraints are expressed mathematically as variance of dose (sum of squares of actual and desired dose values at all points of interest) from desired dose ranges, and only the values lying outside the dose ranges are penalized, with penalty weights specified by the user. Of course, the specified constraints are rarely met and the values ultimately achieved are compromises between competing constraints. The IMRT plans shown here are derived from many plans produced, and multiple plans were generated for different beam configurations, and for a range of values of the objective function parameters. Often trial-and-error is needed to adjust constraints and penalties to achieve the final plan. We present 4 illustrative patient case examples of the application of these methods in the next section. RESULTS Case 1: Brain metastases The first patient case is that of recurrent brain metastases with two tumor foci 5 cm apart: a medial lesion (1.12 cm3) and a lateral lesion (1.15 cm3). Three treatment plans were generated for comparison, including a standard circularcone arc plan, an MLC-blocked static field arrangement (15 fields) as described above, and a similar 15 field plan further optimized with IMRT. For the circular collimator arc-based case, two isocenters were selected with a combination of 4 arcs for one and 5 arcs for the second, each arc spanning a gantry rotation angle of 100 –130°. In an effort to improve conformation and minimize the normal tissue irradiated, the
●
S. H. BENEDICT et al.
753
prescription isodose lines were chosen at the 80% and 70% levels for the two lesions. It is important to note that the SRS arc plans required two separate isocenters for planning, while both the fixed static field and fixed-field IMRT plans utilized only a single isocenter located in the normal tissue between the two lesions. One of the important advantages of the M3 is that it allows simultaneous treatment to multiple lesions, which is considerably more efficient for treatment delivery, and potentially produces more conformal dose distributions. Figure 1 depicts the relative IMRT intensities for one of the 15 beams (right anterior oblique) used to treat the two metastatic lesions. The intensities have been plotted in beamlets of 0.1 ⫻ 0.3 cm, the resolution used in the calculations. It is important to note that both lesions are treated simultaneously with the micro-MLC system, which has a maximum field size of 9.8 ⫻ 9.8 cm. The nonuniform beamlet intensities are particularly useful in depositing additional dose along the boundary of the target volume to compensate for dose loss due to penumbra effects. A transverse CT slice comparison is presented in Fig. 2 for all three techniques. As described previously, the plans were normalized so as to deliver the prescription dose of 10 Gy/fraction to 99% of the target volume. With the large arc degrees available (approximately 100° per arcs), the circular collimator arc-based plan has a sharp dose gradient from the prescription dose to the 8-Gy dose. The IMRT provides a modest improvement for the higher dose conformity. This is quantified in Table 1, which shows the dose–volume data for the total brain, including the PITV, and the total volumes (including the targets) enclosed the 9.0 Gy, 8.0 Gy, and 5.0 Gy isodose surfaces. The SRS arc case is presented because the target volumes are relatively spherical and lend themselves to treatment planning with circular collimators and arcs. However, from these data it is shown that the fixedfield arrangements are more conformal than arcs and that the application of IMRT to the fixed-field arrangement provides an additional measure of conformity. The maximum to minimum dose ratios for the two tumors and three plans were as follows: for the stereotactic arcs: 2.0 (15.0/7.5 Gy) and 1.68 (14.1/8.4 Gy); for fixed static fields: 1.57 (13.7/8.7 Gy) and 1.46 (13.0/8.9 Gy); and for IMRT fields: 1.45 (12.9/8.9 Gy) and 1.35 (13.1/9.7 Gy). Not surprisingly, the SRS arc plan yields the highest dose inhomogeneity ratio, due in large part to the use of multiple isocenters, which necessitated prescribing to lower percent isodose lines. However, the fixed static fields and IMRT fields yield comparatively similar dose inhomogeneity, which is consistent with the low priority assigned to this index in the planning. Case 2: Recurrent ependymoma The second example is a recurrent ependymoma of the posterior fossa. From Fig. 3, one can see that the delineated planning target volume and the brainstem are positioned very closely. Multiple plans were generated for this case, but due to the irregularity of the target shape and proximity
754
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
Fig. 1. IMRT intensities plotted for one of the 15 beams (right anterior oblique) of Case 1, used to treat the two metastatic lesions simulataneously. The relative intensities are plotted in beamlets of 0.1 ⫻ 0.3 cm.
of the brainstem, the blocked fixed-field arrangement was demonstrated to be far superior to the traditional circularcollimator arc plan. This is consistent with the findings of previous investigators (11, 16 –21) for cases of this type. Figure 3 shows isodose curves from these two plans overlaid on a transverse CT slice through the lesion. It is evident that the fixed static field plan is less conformal than the IMRT plan. Moreover, the IMRT plan provides a modest additional degree of conformation along the border with the brainstem. The advantages realized from the improved
conformity of IMRT dose delivery are demonstrated in the reduced PITV shown in Table 2. The dose-volume (including target) data in Table 2 further confirms the sparing of normal tissue with the IMRT technique, showing a reduction of about 20% in the volume of normal tissue only treated at the 9 Gy level. The data also shows improved sparing of the brainstem with the IMRT technique. The maximum to minimum doses for the tumor in case two were as follows: fixed static fields: 2.33 (14.9/6.4 Gy), and fixed IMRT fields: 1.75 (13.3/7.6 Gy). The IMRT fields
Fig. 2. Case 1: Brain metastases. Transverse CT of dose distributions for the two metastatic lesions for three plan types: linac-based arcs with multiple circular collimators (left), fixed uniform-intensity fields (center), and intensity-modulated radiotherapy (IMRT) fields (right). All plans were normalized to deliver 10 Gy to 99% of the planning target volume (PTV). The lesions are the white contours, and the isodose lines surrounding the lesion show 13-, 10-, 8-, 7-, 6-, and 5-Gy levels.
Intensity-modulated stereotactic radiosurgery
Table 1. Case no. 1: Brain metastases; all plans prescribe 10 Gy to 99% of PTV; total volume of 2 lesions is 2.3 cm3 Plan type IMRT Fixed fields Arcs
PITV
Volume 9.0 Gy
Volume 8.0 Gy
Volume 5.0 Gy
2.12
6.68
8.77
20.70
2.23 2.82
6.84 8.82
9.08 10.42
21.46 22.63
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensitymodulated radiotherapy.
thus yield an improved dose homogeneity, despite the low penalty assigned to this parameter in the IMRT calculations. Case 3: Sphenoid wing meningioma The third example is a meningioma involving the cavernous sinus with extension into the suprasellar region. The patient received a partial resection and was referred for radiotherapy to include a stereotactic boost because significant residual disease remains within the cavernous sinus and sella turcica. The patient presented with postoperative cranial nerve deficits on the involved side (left), and radiation doses to the intact right optic apparatus were to be minimized. As in the previous cases, a plan was generated with 15 beams and optimized for both for fixed static fields and with the subsequent application of IMRT, and the results are presented below.
●
S. H. BENEDICT et al.
755
Table 2. Case no. 2: Reccurent ependymoma; all plans prescribe 10 Gy to 99% of PTV; volume of lesion 2.6 cm3 Plan type
Volume Volume Volume 9.0 8.0 5.0 Brainstem PITV Gy Gy Gy 9.0 Gy
IMRT 1.79 Fixed fields 2.12
Brainstem 5.0 Gy
6.14
6.76
15.55
0.89
2.42
6.84
8.44
17.39
0.95
2.41
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensitymodulated radiotherapy.
Figure 4 shows a transverse CT slice for comparison of the fixed-field and IMRT techniques for Case 3. As described previously the plans were normalized so as to deliver the prescription dose of 10 Gy/fraction to 99% of the target volume. The IMRT plan in this case provides a modest improvement in higher dose conformity. This is quantified in Table 3, which shows the dose-volume data for the total brain, including the PITV, and volume enclosed in the 9.0 Gy, 8.0 Gy, and 5.0 Gy isodose surfaces. From these data, it is shown that the optimization available with IMRT applied to the fixed-field arrangement provides an additional measure of tumor conformity and sparing of the adjacent brainstem. The maximum-to-minimum dose ratios for the tumor in Case 3 were as follows: for fixed static fields: 1.48 (13.0/8.8 Gy), and for IMRT fields: 1.41 (12.7/9.0). The static fields and IMRT fields in this case yield comparatively similar dose homogeneity.
Fig. 3. Case 2: Recurrent ependymoma. Transverse CT of dose distributions on a section through the planning target volume (PTV), comparing a plan with 15 fixed-gantry uniform-intensity fields (left) and the same fixed-field arrangement with intensity-modulation (right). All plans were normalized to deliver 10 Gy to 99% of the PTV. The lesion and brainstem are the dark contours, and the dose lines surrounding the lesion are 11, 10, 9, 8, and 5 Gy, respectively.
756
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
Fig. 4. Case 3: Sphenoid wing meningioma. Transverse CT of dose distributions on a section through the planning target volume (PTV), comparing a plan with 15 fixed-gantry uniform-intensity fields (left) and the same fixed-field arrangement with intensity-modulation (right). All plans were normalized to deliver 10 Gy to 99% of the PTV. The lesion and brainstem are the dark contours, and the dose lines surrounding the lesion are 11, 10, 9, 8, and 5 Gy, respectively.
Case 4: Acoustic neuroma The fourth case is that of a patient that presented with an enhancing lesion in the eighth cranial nerve complex on the left side, who underwent craniotomy resection with a residual mass remaining adjacent to the brainstem. With the radiographic evidence of progressive acoustic schwannoma, the patient was referred for stereotactic radiosurgery to treat the affected area. As in the previous cases, a plan was generated with 15 beams and optimized for both the uniform intensity and with the subsequent application of IMRT. Figure 5 shows a transverse CT slice for comparison of the fixed-field and IMRT techniques for Case 4. The plans were designed to deliver the prescription dose of 10 Gy per fraction to 99% of the target volume. In this case the IMRT plan did not yield an improved PITV or decrease the volume of normal tissue irradiated at higher dose levels. However, the IMRT constraints were successful in reducing the volume of adjacent
Table 3. Case no. 3: Sphenoid wing meningioma; all plans prescribe 10 Gy to 99% of PTV; volume of lesion 3.5 cm3 Plan type
Volume Volume Volume 9.0 8.0 5.0 Brainstem Brainstem PITV Gy Gy Gy 9.0 Gy 5.0 Gy
IMRT 2.86 FIXED FLDS 3.05
14.32
18.48
38.31
0.19
1.13
13.88
17.87
37.73
0.25
1.37
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensitymodulated radiotherapy.
brainstem treated to high dose levels, and this may be a critical consideration for the clinician in treatment plan selection. The data comparing the two plans are provided in Table 4. These data show that the static field arrangements are relatively conformal in this case, and that the application of IMRT does not improve conformity, but greatly reduces the volume of brainstem treated at high-dose levels. The maximum-to-minimum dose ratios for the tumor in Case 4 were as follows: for fixed static fields: 1.37 (12.5/9.1 Gy), and for IMRT fields: 1.56 (14.2/9.1 Gy). Thus, the reduction in dose to the brainstem is accomplished in the IMRT plan only by increasing the overall volume of normal tissue irradiated to high-dose levels, and by reducing dose homogeneity in the target volume. DISCUSSION Four patient case examples were presented to examine the potential of thin-leafed, dynamic MLC-based IMRT (IMSRS) for stereotactic irradiation of small lesions. Plan comparisons included the use of dose–volume data, volumes of normal adjacent tissue and critical structures irradiated, indices of conformity including the PITV, and dose homogeneity. For 3 of the 4 cases the IMSRS plan yielded a smaller volume of normal tissue irradiated in the significant range of doses, generally described as doses greater than 50% of the prescription. In general, the IMSRS plans showed a reduction in PITV, with values ranging from 5% to 18% lower than those seen in the fixed static field stereotactic plans. In the fourth patient case, although not
Intensity-modulated stereotactic radiosurgery
●
S. H. BENEDICT et al.
757
Fig. 5. Case 4: Acoustic neuroma. Transverse CT of dose distributions on a section through the planning target volume (PTV), comparing a plan with 15 fixed-gantry uniform-intensity fields (left) and the same fixed-field arrangement with intensity-modulation (right). All plans were normalized to deliver 10 Gy to 99% of the PTV. The lesion and brainstem are the white contours, and the dose lines surrounding the lesion are 11, 10, 9, 8, and 5 Gy, respectively.
demonstrated improved conformality in their treatments of small lesions, but the consequence is usually a dose inhomogeneity index on the order of 2.0, with the inhomogeneity due in large part to dose overlapping, producing substantial hot spots within the irradiated volume. All CT scans were acquired with a minimum slice thickness of 0.3 cm. We have used CT slice thickness on the order of 0.1 cm for clinical studies, and these provide additional information with regard to target contours. However, other obstacles stand in the way of improving the precision of dose calculations with our present IMSRS program, including the M3 minimum leaf width of 0.3 cm, minimum practical dose calculation grid of 0.3 ⫻ 0.3 ⫻ 0.3 cm, and beamlet intensity of 0.1 ⫻ 0.3 cm. However, we can assume that future improvements in the IMSRS may not only require reductions in the dose grid sizes and beamlet intensities, but also thinner CT acquisition slices for some highly irregular target contours. In the results presented here, all field sizes were small enough that they were fully covered by the 0.3-cm leaf width portion of the M3. For larger fields, we must consider variable leaf widths.
providing a lower PITV, the IMRT plan was able to reduce the volume of brainstem irradiated to approximately 50% of that seen with static fields only. The dose inhomogeneity ratio for two of the plans was less for the IMRT fields in two of the cases, and greater in the other two, which is an indication of the low priority assigned to this parameter. Improvements in dose homogeneity could serve as a point of focus for future investigations, even though there appears to be some controversy over the importance of homogeneity in treating small intracranial lesions. For the treatments presented here the prescriptions were all based on 99% coverage of the PTV, and therefore a substantial contribution to the dose inhomogeneity occurs in the comparatively small volume (1%) which incurs a reduced prescription dose. Clinical practice may vary on what volume to prescribe, and values of 90% and 95% target coverage are commonly reported, but further investigations may be needed to find an acceptable resolution to this question. It is important to note that 60Co gamma knife plans that may require a large number of isocenters, which would be impractical for linac-based treatments, have
Table 4. Case no. 4: Acoustic neuroma; all plans prescribe 10 Gy to 99% of PTV; volume of lesion 1.2 cm3 Plan type
PITV
Volume 9.0 Gy
Volume 8.0 Gy
Volume 5.0 Gy
Brainstem 9.0 Gy
Brainstem 5.0 Gy
IMRT FIXED FIELDS
2.51 2.17
4.24 3.64
5.40 4.83
11.09 10.51
0.05 0.16
0.54 1.01
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensity-modulated radiotherapy.
758
I. J. Radiation Oncology
●
Biology
●
Physics
Our approach for accommodating variable leaf widths will be described elsewhere. The traditional use of arcs with multiple circular collimators has provided an efficient method of adapting linear accelerators for a wide range of clinical stereotactic programs. The arc-circular collimator technique has also been previously demonstrated to be very effective and highly conformal in the treatment of relatively spherical lesions, and in providing a steep dose gradient from prescription dose to a clinically insignificant dose (22). However, the circular collimator technique has been demonstrated both in this study, and in a variety of others (11, 16 –21) to yield dose distributions less conformal than fixed-field plans for very irregularly shaped targets. The purpose of this study was to demonstrate the potential for further improvement in the fixed-field plans by the application of IMRT using dynamic micro-multileaf collimation. In fact, it would seem apparent that the main advantage of arc based SRS (steep dose gradients) and intensity modulation (conformal dose distributions) could be combined to yield even more highly optimized treatment plans using arc based IMRT, provided an efficient means of delivery can be devised. Although IMRT has been demonstrated to produce sig-
Volume 50, Number 3, 2001
nificant dosimetric improvements for larger tumors and modest improvements for small intracranial lesions, its efficacy is not guaranteed for small tumors due to the lateral transport of radiation (6). An investigation into the problems inherent to small leaf collimation, and possible gain in the limitations of leaf width, is underway. Other future investigations include automated beam configurations, automated beam weight optimization, and the use of conformal arcs with dynamic collimation. CONCLUSION IMSRS has been demonstrated to provide dosimetric improvements for small, highly irregularly shaped lesions of the brain when compared to complex, multi-isocenter linac-based stereotactic arc plans or to uniform-intensity fixed static field plans. In addition to improved dose conformation and minimization of dose outside of the PTV, the IMSRS provides an additional option for prioritizing dose minimization to adjacent critical areas. This paper demonstrates that IMSRS with a micro-MLC has potential for superior treatment planning relative to uniform-intensity fixed-field or arc-based methods with circular collimators.
REFERENCES 1. Lutz W, Winston KR, Maleki N. A system for stereotactic radiosurgery with a linear accelerator. Int J Radiat Oncol Biol Phys 1988;14:373–381. 2. Podgorsak EB, Olivier A, Pla M., et al.Dynamic stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1988;14:115–126. 3. Leavitt DD, Gibbs FA, Heilbrun MP, et al. Dynamic field shaping to optimize stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1991;21:1247–1255. 4. Schell MC, Smith V, Larson DA, et al.Evaluation of radiosurgery techniques with cumulative dose volume histograms in linac-based stereotactic external beam irradiation. Int J Radiat Oncol Biol Phys 1991;20:1325–1330. 5. Podgorsak EB, Pike GB, Olivier A, et al.Radiosurgery with high energy photon beams: A comparison among techniques. Int J Radiat Oncol Biol Phys 1989;16:857– 865. 6. Mohan R, Wu Q, Wang X, et al.Intensity modulation optimization, lateral transport of radiation and margins. Med Phys 1996;23:2011–2022. 7. Webb S. Configuration options for intensity-modulated radiation therapy using multiple static fields shaped by a multileaf collimator. Phys Med Biol 1998;43:241–260. 8. Wang XH, Mohan R, Jackson A, et al.Optimization of intensity-modulated 3D conformal treatment plans based on biological indices. Radiother Oncol 1995;37:140 –152. 9. Boyer AL, Geis P, Grant W, et al.Modulated beam conformal therapy for head and neck tumors. Int J Radiat Oncol Biol Phys 1997;39:227–236. 10. Woo SY, Grant WH, Bellezza D, et al.A comparison of intensity modulated conformal therapy with a conventional external beam stereotactic radiosurgery system for the treatment of single and multiple intracranial lesions. Int J Radiat Oncol Biol Phys 1996;35:593–597. 11. Cardinale RM, Benedict SH, Wu Q, et al.A comparison of three stereotactic radiotherapy techniques: Arcs vs. non-coplanar fixed fields vs. intensity modulation. Int J Radiat Oncol Biol Phys 1998;42:431– 436.
12. Xia P, Geis P, Xing L, et al.Physical characteristics of a miniature multileaf collimator. Med Phys 1999;26:65–70. 13. Cosgrove VP, Jahn U, Pfaender M, et al.Commissioning of a micro multi-leaf collimator and planning system for stereotactic radiosurgery. Radiother Onco 1999;50:325–336 14. Wu Q, Mohan R. Algorithms, and functionality of an IMRT optimization system. Med Phys 2000;27:701–711 15. Cardinale RM, Benedict SH, Bump EA, et al.Determining optimal block margin for conformal stereotactic irradiation of lung and hepatic tumors. Int J Radiat Oncol Biol Phys 1999; 45:515–520. 16. Kubo HD, Pappas CTE, Wilder RB. A comparison of arcbased and static mini-multileaf collimator-based radiosurgery plans. Radiother Oncol 1997;45:89 –93. 17. Bourland JD, McCollough KP. Static field conformal stereotactic radiosurgery: Physical techniques. Int J Radiat Oncol Biol Phys 1993;28:471– 479. 18. Hamilton RJ, Kuchnir FT, Sweeney P, et al.Comparison of static conformal field with multiple noncoplanar arc techniques for stereotactic radiosurgery or stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1995;33:1221–1228. 19. Shiu AS, Kooy HM, Ewton JR, et al.Comparison of miniature multileaf collimation (mmlc) with circular collimation for stereotactic treatment. Int J Radiat Oncol Biol Phys 1997;37: 679 – 688. 20. Marks LB, Sherouse GW, Das S, et al.Conformal radiation therapy with fixed shaped coplanar or noncoplanar radiation beam boquets: A possible alternative to radiosurgery. Int J Radiat Oncol Biol Phys 1995;33:1209 –1219. 21. Laing RW, Bentley RE, Nahum AE, et al.Stereotactic radiotherapy of irregular targets: A comparison between static conformal beams and non-coplanar arcs. Radiother Oncol 1993;28:241–246. 22. Kooy HM, Nedzi LA, Loeffler JS, et al. Treatment planning for stereotactic radiosurgery of intracranial leasions. Int J Radiat Oncol Biol Phys 1991;21:683– 693
PII S0360-3016(01)01487-0
CLINICAL INVESTIGATION
Brain
INTENSITY-MODULATED STEREOTACTIC RADIOSURGERY USING DYNAMIC MICRO-MULTILEAF COLLIMATION STANLEY H. BENEDICT, PH.D.,* ROBERT M. CARDINALE, M.D.,* QIUWEN WU, PH.D.,* ROBERT D. ZWICKER, PH.D.,* WILLIAM C. BROADDUS, M.D., PH.D.,† AND RADHE MOHAN, PH.D.* Departments of *Radiation Oncology and †Neurosurgery, Medical College of Virginia Hospitals of Virginia Commonwealth University, Richmond, VA Purpose: The implementation of dynamic leaf motion on a micro-multileaf collimator system provides the capability for intensity-modulated stereotactic radiosurgery (IMSRS), and the consequent potential for improved dose distributions for irregularly shaped tumor volumes adjacent to critical organs. This study explores the use of IMSRS to provide improved tumor coverage and normal tissue sparing for small cranial tumors relative to plans based on multiple fixed uniform-intensity beams or traditional circular collimator arc-based stereotactic techniques. Methods and Materials: Four patient cases involving small brain lesions are presented and analyzed. The cases were chosen to include a representative selection of target shapes, number of targets, and adjacent critical areas. Patient plans generated for these comparisons include standard arcs with multiple circular collimators, and fixed noncoplanar static fields with uniform-intensity beams and IMSRS. Parameters used for evaluation of the plans include the percentage of irradiated volume to tumor volume (PITV), normal tissue dose–volume histograms, and dose– homogeneity ratios. All IMSRS plans were computed using previously established IMRT techniques adapted for use with the BrainLAB M3 micro-multileaf collimator. The algorithms comprising the IMRT system for optimization of intensity distributions and conversion into leaf trajectories of the BrainLab M3 were developed at our institution. The ADAC Pinnacle3 radiation treatment-planning system was used for dose calculations and for input of contours for target volumes and normal critical structures. Results: For all cases, the IMSRS plans showed a high degree of conformity of the dose distribution with the target shape. The IMSRS plans provided either (1) a smaller volume of normal tissue irradiated to significant dose levels, generally taken as doses greater than 50% of the prescription, or (2) a lower dose to an important adjacent critical organ. The reduction in volume of normal tissue irradiated in the IMSRS plans ranged from 10% to 50% relative to the other arc and uniform fixed-field plans. Conclusion: The case studies presented for IMSRS demonstrate significant dosimetric improvements for small, irregularly shaped lesions of the brain when compared to treatments using multiple static fields or standard SRS arc techniques with circular collimators. For all cases, the IMSRS plan yielded a smaller volume of normal tissue irradiated, and/or a reduction in the volume of an adjacent critical organ (i.e., brainstem) irradiated to significant dose levels. © 2001 Elsevier Science Inc. Stereotactic radiosurgery, Intensity-modulated radiotherapy, Micro-multileaf collimation.
INTRODUCTION
tensity-modulated radiotherapy (IMRT) to produce dose distributions that are superior to conventional plans has been widely presented in the literature (6 –9). The majority of studies demonstrating improvement in dosimetry with IMRT have been for large field sizes using relatively large multileaf collimation systems, generally with leaves 1.0 cm in width. With the recent introduction of micro-multileaf collimator systems the advantages of IMRT may be further extended to the case of small intracranial targets (10). In a previous investigation, we demonstrated the dosimetric advantages of IMRT applied to intracranial targets which were relatively large (9.6 –36.7 cm3) and of various geometric
There have been numerous publications on the development of linac-based radiosurgery techniques, from the initial arcbased approach with circular collimators (1) to fixed-field arrangements and dynamic SRS (2, 3). Methods to evaluate radiosurgery techniques (4, 5) are also well established. The present work extends these studies by using clinical examples to confirm and quantify the advantages of a technique that combines the precision of stereotactic positioning with the dose-delivery capabilities of IMRT to treat small critically located targets. The potential of dynamic multileaf collimation and inReprint requests to: Dr. Stanley H. Benedict, Department of Radiation Oncology, Medical College of Virginia Hospitals of Virginia Commonwealth University, 401 College Street, Richmond, VA 23298-0058. E-mail: [email protected]
This work is supported by Grants CA74043 and R29 CA7643901A1 from the National Cancer Institute. Accepted for publication 24 January 2001. 751
752
I. J. Radiation Oncology
●
Biology
●
Physics
shapes, such as an ellipsoid and hemisphere (11). In this study, we present a selection of actual patient cases selected for radiosurgery with small intracranial structures (1.2–3.5 cm3) to demonstrate that IMRT using a set of fixed fields and delivered with dynamic micro-multileaf collimation leads to improved dose distributions when compared to multiple arcs or fixed uniform-intensity fields. The collimation system used for this investigation is the BrainLAB M3 micro-multileaf collimator (BrainLAB AG, Heimstetten, Germany), the characteristics of which have been previously described in the literature (12, 13). The primary innovative feature of the M3 is that it has leaves of variable width, which are narrower than those generally available as a standard feature on the current generation of accelerators. The M3 has 14 pairs of 0.3 cm, 6 pairs of 0.45 cm, and 6 pairs of 0.55 cm leaves. All of the IMRT computer algorithms used for optimization of intensity distributions and conversion into M3 leaf trajectories were developed at our institution (14). METHODS AND MATERIALS Overview For this study, 4 patient cases involving small brain lesions were selected, with a range of shapes, number of targets, and adjacent critical areas for investigation. All patients were immobilized, planned, and treated using established stereotactic techniques. Patient 1 presented with lung metastases to the brain with two foci 5 cm apart. Patient 2 had a recurrent ependymoma in the posterior fossa adjacent to brainstem. Patient 3 presented with sphenoid wing meningioma, and Patient 4 presented with an acoustic neuroma. For all 4 patients, stereotactic arc plans using circular cone collimators and plans based on noncoplanar static, uniform-intensity beams defined by the micro-multileaf collimator (micro-MLC) were generated and optimized to give a minimum percentage of irradiated-volume to tumor-volume ratio (PITV) and minimal normal tissue irradiation at high dose levels. The PITV and dose–volume histogram (DVH) to adjacent critical organs were then used to compare the plans with the corresponding IMSRS plans. Multiple IMSRS plans were generated for each patient using our in-house programs with varying beam configurations, the same configurations used in the fixed-field plans described below. For all of the plans, the dose homogeneity within the planning target volume (PTV) was assigned a low priority relative to target conformity and adjacent critical structure doses. The ADAC Pinnacle3 radiation treatment-planning system (ADAC Laboratories, Milpitas, CA) was used for all dose calculations and for input of the target and critical structure contours used by our in-house IMRT optimization program. The dose grid used for these calculations was 0.3 ⫻ 0.3 ⫻ 0.3 cm, a practical limitation imposed by the current computer hardware capabilities. For comparison purposes, all plans were prescribed to a dose of 10 Gy, with the requirement that 99% of the target volume receive the prescription dose or greater. For planning pur-
Volume 50, Number 3, 2001
poses, and in accordance with routine SRS procedures, the PTV was assigned the same volume as the gross tumor volume (GTV), with no further expansion. Fixed-field plan design For the fixed-field arrangement, an initial plan of 15 fields was set up for each plan. The design of the initial fixed-field arrangement was established to incorporate 3 fields each along 5 planes of the patient’s skull, which provides approximately 2 steradian solid-angle coverage. The beam arrangement utilized left and right laterals, left and right oblique fields, and vertex fields, designed for a minimum of beam opposition. Once the template of 15 fields was scripted to the patient, adjustments of ⫾ 10° were made in each of the couch/gantry orientations to minimize dose to adjacent critical areas. Beam weights were also optimized by trial and error to provide the prescribed target coverage while minimizing dose contributions to adjacent critical areas. Block margins for these beams were optimized to yield a minimum PITV, and were typically 0.0 – 0.3 cm for each beam. Block margin optimization for fixed fields has been previously investigated at our institution (15). The M3 collimator angle was also chosen such that the sum of the unblocked field areas external to the optimal block margin due to the leaf shapes, and the blocked areas internal to the margin is minimized, typically at a value of less than about 0.5 cm2. Arc-based plans with multiple circular collimators For evaluation of the linac-based stereotactic arc plans using multiple circular collimators, each isocenter was assigned 4 –5 arcs of approximately 100° per arc, in accordance with well-established techniques (1). A number of studies have been published that demonstrate the superiority of linac plans that use conformal fixed-field arrangements over those using arcs with multiple circular collimators (11, 16 –21). In the course of the present studies, this superiority was demonstrated for all patient cases investigated. For this reason, only the first patient case is presented with all three techniques, while the remaining cases compare only the fixed uniform-intensity fields with the IMRT fields. Intensity-modulated radiosurgery (IMSRS) methods The in-house IMRT optimization system was coupled to the ADAC Pinnacle3 three-dimensional (3D) dose calculation engine to generate the IMSRS treatment plans. All fields were divided into intensity matrices of size 0.1 cm ⫻ 0.3 cm to provide the highest resolution consistent with the geometry of the M3 collimator. The Pinnacle3 3D radiation treatment-planning system was used to input and display the target and critical structure contours and the beam parameters needed by the IMRT calculation routines, and also to generate the quantities used for the final evaluation of treatment plans. For the delivery of treatment, the leaf trajectories, in the form of leaf position data stored as a function of monitor units, obtained from the optimized beam intensity patterns, were transferred to the dynamic
Intensity-modulated stereotactic radiosurgery
multileaf collimator (DMLC) controller. This DMLC controller utilizes the driver software and was developed by the treatment machine manufacturer (Varian Medical Equipment, Palo Alto, CA) to move the leaves to their designated positions while the beam is on. After the fixed-field plan had been optimized, all of the beam geometry parameters were transferred to the IMRT program for intensity distribution optimization. No margin was assigned to the gross target volume for IMRT to minimize the volume of normal brain exposed to high doses. This requires accepting higher target dose inhomogeneity in the final plan. Once the IMRT technique was applied to the optimized 15 fixed-field plan, further iterations were applied to obtain a minimum score based on the penalties and parameters of the objective function of the program. The objective function for optimizing intensity distributions in these calculations was specified in terms of dosevolume limits. The criteria used in this study included the requirement that exactly 99% of the target volume should be enclosed in the prescription isodose surface. Penalties for dose inhomogeneity within the target were marginalized relative to those assigned to the index of conformity. In the absence of any strong data demonstrating the importance of dose homogeneity in these cases, hot spots within the target volume were assigned a low priority for our planning purposes. The dose to normal tissue was constrained to limit the volume allowed to receive a dose higher than a chosen dose limit without a high penalty. Such constraints are expressed mathematically as variance of dose (sum of squares of actual and desired dose values at all points of interest) from desired dose ranges, and only the values lying outside the dose ranges are penalized, with penalty weights specified by the user. Of course, the specified constraints are rarely met and the values ultimately achieved are compromises between competing constraints. The IMRT plans shown here are derived from many plans produced, and multiple plans were generated for different beam configurations, and for a range of values of the objective function parameters. Often trial-and-error is needed to adjust constraints and penalties to achieve the final plan. We present 4 illustrative patient case examples of the application of these methods in the next section. RESULTS Case 1: Brain metastases The first patient case is that of recurrent brain metastases with two tumor foci 5 cm apart: a medial lesion (1.12 cm3) and a lateral lesion (1.15 cm3). Three treatment plans were generated for comparison, including a standard circularcone arc plan, an MLC-blocked static field arrangement (15 fields) as described above, and a similar 15 field plan further optimized with IMRT. For the circular collimator arc-based case, two isocenters were selected with a combination of 4 arcs for one and 5 arcs for the second, each arc spanning a gantry rotation angle of 100 –130°. In an effort to improve conformation and minimize the normal tissue irradiated, the
●
S. H. BENEDICT et al.
753
prescription isodose lines were chosen at the 80% and 70% levels for the two lesions. It is important to note that the SRS arc plans required two separate isocenters for planning, while both the fixed static field and fixed-field IMRT plans utilized only a single isocenter located in the normal tissue between the two lesions. One of the important advantages of the M3 is that it allows simultaneous treatment to multiple lesions, which is considerably more efficient for treatment delivery, and potentially produces more conformal dose distributions. Figure 1 depicts the relative IMRT intensities for one of the 15 beams (right anterior oblique) used to treat the two metastatic lesions. The intensities have been plotted in beamlets of 0.1 ⫻ 0.3 cm, the resolution used in the calculations. It is important to note that both lesions are treated simultaneously with the micro-MLC system, which has a maximum field size of 9.8 ⫻ 9.8 cm. The nonuniform beamlet intensities are particularly useful in depositing additional dose along the boundary of the target volume to compensate for dose loss due to penumbra effects. A transverse CT slice comparison is presented in Fig. 2 for all three techniques. As described previously, the plans were normalized so as to deliver the prescription dose of 10 Gy/fraction to 99% of the target volume. With the large arc degrees available (approximately 100° per arcs), the circular collimator arc-based plan has a sharp dose gradient from the prescription dose to the 8-Gy dose. The IMRT provides a modest improvement for the higher dose conformity. This is quantified in Table 1, which shows the dose–volume data for the total brain, including the PITV, and the total volumes (including the targets) enclosed the 9.0 Gy, 8.0 Gy, and 5.0 Gy isodose surfaces. The SRS arc case is presented because the target volumes are relatively spherical and lend themselves to treatment planning with circular collimators and arcs. However, from these data it is shown that the fixedfield arrangements are more conformal than arcs and that the application of IMRT to the fixed-field arrangement provides an additional measure of conformity. The maximum to minimum dose ratios for the two tumors and three plans were as follows: for the stereotactic arcs: 2.0 (15.0/7.5 Gy) and 1.68 (14.1/8.4 Gy); for fixed static fields: 1.57 (13.7/8.7 Gy) and 1.46 (13.0/8.9 Gy); and for IMRT fields: 1.45 (12.9/8.9 Gy) and 1.35 (13.1/9.7 Gy). Not surprisingly, the SRS arc plan yields the highest dose inhomogeneity ratio, due in large part to the use of multiple isocenters, which necessitated prescribing to lower percent isodose lines. However, the fixed static fields and IMRT fields yield comparatively similar dose inhomogeneity, which is consistent with the low priority assigned to this index in the planning. Case 2: Recurrent ependymoma The second example is a recurrent ependymoma of the posterior fossa. From Fig. 3, one can see that the delineated planning target volume and the brainstem are positioned very closely. Multiple plans were generated for this case, but due to the irregularity of the target shape and proximity
754
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
Fig. 1. IMRT intensities plotted for one of the 15 beams (right anterior oblique) of Case 1, used to treat the two metastatic lesions simulataneously. The relative intensities are plotted in beamlets of 0.1 ⫻ 0.3 cm.
of the brainstem, the blocked fixed-field arrangement was demonstrated to be far superior to the traditional circularcollimator arc plan. This is consistent with the findings of previous investigators (11, 16 –21) for cases of this type. Figure 3 shows isodose curves from these two plans overlaid on a transverse CT slice through the lesion. It is evident that the fixed static field plan is less conformal than the IMRT plan. Moreover, the IMRT plan provides a modest additional degree of conformation along the border with the brainstem. The advantages realized from the improved
conformity of IMRT dose delivery are demonstrated in the reduced PITV shown in Table 2. The dose-volume (including target) data in Table 2 further confirms the sparing of normal tissue with the IMRT technique, showing a reduction of about 20% in the volume of normal tissue only treated at the 9 Gy level. The data also shows improved sparing of the brainstem with the IMRT technique. The maximum to minimum doses for the tumor in case two were as follows: fixed static fields: 2.33 (14.9/6.4 Gy), and fixed IMRT fields: 1.75 (13.3/7.6 Gy). The IMRT fields
Fig. 2. Case 1: Brain metastases. Transverse CT of dose distributions for the two metastatic lesions for three plan types: linac-based arcs with multiple circular collimators (left), fixed uniform-intensity fields (center), and intensity-modulated radiotherapy (IMRT) fields (right). All plans were normalized to deliver 10 Gy to 99% of the planning target volume (PTV). The lesions are the white contours, and the isodose lines surrounding the lesion show 13-, 10-, 8-, 7-, 6-, and 5-Gy levels.
Intensity-modulated stereotactic radiosurgery
Table 1. Case no. 1: Brain metastases; all plans prescribe 10 Gy to 99% of PTV; total volume of 2 lesions is 2.3 cm3 Plan type IMRT Fixed fields Arcs
PITV
Volume 9.0 Gy
Volume 8.0 Gy
Volume 5.0 Gy
2.12
6.68
8.77
20.70
2.23 2.82
6.84 8.82
9.08 10.42
21.46 22.63
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensitymodulated radiotherapy.
thus yield an improved dose homogeneity, despite the low penalty assigned to this parameter in the IMRT calculations. Case 3: Sphenoid wing meningioma The third example is a meningioma involving the cavernous sinus with extension into the suprasellar region. The patient received a partial resection and was referred for radiotherapy to include a stereotactic boost because significant residual disease remains within the cavernous sinus and sella turcica. The patient presented with postoperative cranial nerve deficits on the involved side (left), and radiation doses to the intact right optic apparatus were to be minimized. As in the previous cases, a plan was generated with 15 beams and optimized for both for fixed static fields and with the subsequent application of IMRT, and the results are presented below.
●
S. H. BENEDICT et al.
755
Table 2. Case no. 2: Reccurent ependymoma; all plans prescribe 10 Gy to 99% of PTV; volume of lesion 2.6 cm3 Plan type
Volume Volume Volume 9.0 8.0 5.0 Brainstem PITV Gy Gy Gy 9.0 Gy
IMRT 1.79 Fixed fields 2.12
Brainstem 5.0 Gy
6.14
6.76
15.55
0.89
2.42
6.84
8.44
17.39
0.95
2.41
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensitymodulated radiotherapy.
Figure 4 shows a transverse CT slice for comparison of the fixed-field and IMRT techniques for Case 3. As described previously the plans were normalized so as to deliver the prescription dose of 10 Gy/fraction to 99% of the target volume. The IMRT plan in this case provides a modest improvement in higher dose conformity. This is quantified in Table 3, which shows the dose-volume data for the total brain, including the PITV, and volume enclosed in the 9.0 Gy, 8.0 Gy, and 5.0 Gy isodose surfaces. From these data, it is shown that the optimization available with IMRT applied to the fixed-field arrangement provides an additional measure of tumor conformity and sparing of the adjacent brainstem. The maximum-to-minimum dose ratios for the tumor in Case 3 were as follows: for fixed static fields: 1.48 (13.0/8.8 Gy), and for IMRT fields: 1.41 (12.7/9.0). The static fields and IMRT fields in this case yield comparatively similar dose homogeneity.
Fig. 3. Case 2: Recurrent ependymoma. Transverse CT of dose distributions on a section through the planning target volume (PTV), comparing a plan with 15 fixed-gantry uniform-intensity fields (left) and the same fixed-field arrangement with intensity-modulation (right). All plans were normalized to deliver 10 Gy to 99% of the PTV. The lesion and brainstem are the dark contours, and the dose lines surrounding the lesion are 11, 10, 9, 8, and 5 Gy, respectively.
756
I. J. Radiation Oncology
●
Biology
●
Physics
Volume 50, Number 3, 2001
Fig. 4. Case 3: Sphenoid wing meningioma. Transverse CT of dose distributions on a section through the planning target volume (PTV), comparing a plan with 15 fixed-gantry uniform-intensity fields (left) and the same fixed-field arrangement with intensity-modulation (right). All plans were normalized to deliver 10 Gy to 99% of the PTV. The lesion and brainstem are the dark contours, and the dose lines surrounding the lesion are 11, 10, 9, 8, and 5 Gy, respectively.
Case 4: Acoustic neuroma The fourth case is that of a patient that presented with an enhancing lesion in the eighth cranial nerve complex on the left side, who underwent craniotomy resection with a residual mass remaining adjacent to the brainstem. With the radiographic evidence of progressive acoustic schwannoma, the patient was referred for stereotactic radiosurgery to treat the affected area. As in the previous cases, a plan was generated with 15 beams and optimized for both the uniform intensity and with the subsequent application of IMRT. Figure 5 shows a transverse CT slice for comparison of the fixed-field and IMRT techniques for Case 4. The plans were designed to deliver the prescription dose of 10 Gy per fraction to 99% of the target volume. In this case the IMRT plan did not yield an improved PITV or decrease the volume of normal tissue irradiated at higher dose levels. However, the IMRT constraints were successful in reducing the volume of adjacent
Table 3. Case no. 3: Sphenoid wing meningioma; all plans prescribe 10 Gy to 99% of PTV; volume of lesion 3.5 cm3 Plan type
Volume Volume Volume 9.0 8.0 5.0 Brainstem Brainstem PITV Gy Gy Gy 9.0 Gy 5.0 Gy
IMRT 2.86 FIXED FLDS 3.05
14.32
18.48
38.31
0.19
1.13
13.88
17.87
37.73
0.25
1.37
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensitymodulated radiotherapy.
brainstem treated to high dose levels, and this may be a critical consideration for the clinician in treatment plan selection. The data comparing the two plans are provided in Table 4. These data show that the static field arrangements are relatively conformal in this case, and that the application of IMRT does not improve conformity, but greatly reduces the volume of brainstem treated at high-dose levels. The maximum-to-minimum dose ratios for the tumor in Case 4 were as follows: for fixed static fields: 1.37 (12.5/9.1 Gy), and for IMRT fields: 1.56 (14.2/9.1 Gy). Thus, the reduction in dose to the brainstem is accomplished in the IMRT plan only by increasing the overall volume of normal tissue irradiated to high-dose levels, and by reducing dose homogeneity in the target volume. DISCUSSION Four patient case examples were presented to examine the potential of thin-leafed, dynamic MLC-based IMRT (IMSRS) for stereotactic irradiation of small lesions. Plan comparisons included the use of dose–volume data, volumes of normal adjacent tissue and critical structures irradiated, indices of conformity including the PITV, and dose homogeneity. For 3 of the 4 cases the IMSRS plan yielded a smaller volume of normal tissue irradiated in the significant range of doses, generally described as doses greater than 50% of the prescription. In general, the IMSRS plans showed a reduction in PITV, with values ranging from 5% to 18% lower than those seen in the fixed static field stereotactic plans. In the fourth patient case, although not
Intensity-modulated stereotactic radiosurgery
●
S. H. BENEDICT et al.
757
Fig. 5. Case 4: Acoustic neuroma. Transverse CT of dose distributions on a section through the planning target volume (PTV), comparing a plan with 15 fixed-gantry uniform-intensity fields (left) and the same fixed-field arrangement with intensity-modulation (right). All plans were normalized to deliver 10 Gy to 99% of the PTV. The lesion and brainstem are the white contours, and the dose lines surrounding the lesion are 11, 10, 9, 8, and 5 Gy, respectively.
demonstrated improved conformality in their treatments of small lesions, but the consequence is usually a dose inhomogeneity index on the order of 2.0, with the inhomogeneity due in large part to dose overlapping, producing substantial hot spots within the irradiated volume. All CT scans were acquired with a minimum slice thickness of 0.3 cm. We have used CT slice thickness on the order of 0.1 cm for clinical studies, and these provide additional information with regard to target contours. However, other obstacles stand in the way of improving the precision of dose calculations with our present IMSRS program, including the M3 minimum leaf width of 0.3 cm, minimum practical dose calculation grid of 0.3 ⫻ 0.3 ⫻ 0.3 cm, and beamlet intensity of 0.1 ⫻ 0.3 cm. However, we can assume that future improvements in the IMSRS may not only require reductions in the dose grid sizes and beamlet intensities, but also thinner CT acquisition slices for some highly irregular target contours. In the results presented here, all field sizes were small enough that they were fully covered by the 0.3-cm leaf width portion of the M3. For larger fields, we must consider variable leaf widths.
providing a lower PITV, the IMRT plan was able to reduce the volume of brainstem irradiated to approximately 50% of that seen with static fields only. The dose inhomogeneity ratio for two of the plans was less for the IMRT fields in two of the cases, and greater in the other two, which is an indication of the low priority assigned to this parameter. Improvements in dose homogeneity could serve as a point of focus for future investigations, even though there appears to be some controversy over the importance of homogeneity in treating small intracranial lesions. For the treatments presented here the prescriptions were all based on 99% coverage of the PTV, and therefore a substantial contribution to the dose inhomogeneity occurs in the comparatively small volume (1%) which incurs a reduced prescription dose. Clinical practice may vary on what volume to prescribe, and values of 90% and 95% target coverage are commonly reported, but further investigations may be needed to find an acceptable resolution to this question. It is important to note that 60Co gamma knife plans that may require a large number of isocenters, which would be impractical for linac-based treatments, have
Table 4. Case no. 4: Acoustic neuroma; all plans prescribe 10 Gy to 99% of PTV; volume of lesion 1.2 cm3 Plan type
PITV
Volume 9.0 Gy
Volume 8.0 Gy
Volume 5.0 Gy
Brainstem 9.0 Gy
Brainstem 5.0 Gy
IMRT FIXED FIELDS
2.51 2.17
4.24 3.64
5.40 4.83
11.09 10.51
0.05 0.16
0.54 1.01
Abbreviations: PTV ⫽ planning target volume; PITV ⫽ percentage of irradiated volume to tumor volume; IMRT ⫽ intensity-modulated radiotherapy.
758
I. J. Radiation Oncology
●
Biology
●
Physics
Our approach for accommodating variable leaf widths will be described elsewhere. The traditional use of arcs with multiple circular collimators has provided an efficient method of adapting linear accelerators for a wide range of clinical stereotactic programs. The arc-circular collimator technique has also been previously demonstrated to be very effective and highly conformal in the treatment of relatively spherical lesions, and in providing a steep dose gradient from prescription dose to a clinically insignificant dose (22). However, the circular collimator technique has been demonstrated both in this study, and in a variety of others (11, 16 –21) to yield dose distributions less conformal than fixed-field plans for very irregularly shaped targets. The purpose of this study was to demonstrate the potential for further improvement in the fixed-field plans by the application of IMRT using dynamic micro-multileaf collimation. In fact, it would seem apparent that the main advantage of arc based SRS (steep dose gradients) and intensity modulation (conformal dose distributions) could be combined to yield even more highly optimized treatment plans using arc based IMRT, provided an efficient means of delivery can be devised. Although IMRT has been demonstrated to produce sig-
Volume 50, Number 3, 2001
nificant dosimetric improvements for larger tumors and modest improvements for small intracranial lesions, its efficacy is not guaranteed for small tumors due to the lateral transport of radiation (6). An investigation into the problems inherent to small leaf collimation, and possible gain in the limitations of leaf width, is underway. Other future investigations include automated beam configurations, automated beam weight optimization, and the use of conformal arcs with dynamic collimation. CONCLUSION IMSRS has been demonstrated to provide dosimetric improvements for small, highly irregularly shaped lesions of the brain when compared to complex, multi-isocenter linac-based stereotactic arc plans or to uniform-intensity fixed static field plans. In addition to improved dose conformation and minimization of dose outside of the PTV, the IMSRS provides an additional option for prioritizing dose minimization to adjacent critical areas. This paper demonstrates that IMSRS with a micro-MLC has potential for superior treatment planning relative to uniform-intensity fixed-field or arc-based methods with circular collimators.
REFERENCES 1. Lutz W, Winston KR, Maleki N. A system for stereotactic radiosurgery with a linear accelerator. Int J Radiat Oncol Biol Phys 1988;14:373–381. 2. Podgorsak EB, Olivier A, Pla M., et al.Dynamic stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1988;14:115–126. 3. Leavitt DD, Gibbs FA, Heilbrun MP, et al. Dynamic field shaping to optimize stereotactic radiosurgery. Int J Radiat Oncol Biol Phys 1991;21:1247–1255. 4. Schell MC, Smith V, Larson DA, et al.Evaluation of radiosurgery techniques with cumulative dose volume histograms in linac-based stereotactic external beam irradiation. Int J Radiat Oncol Biol Phys 1991;20:1325–1330. 5. Podgorsak EB, Pike GB, Olivier A, et al.Radiosurgery with high energy photon beams: A comparison among techniques. Int J Radiat Oncol Biol Phys 1989;16:857– 865. 6. Mohan R, Wu Q, Wang X, et al.Intensity modulation optimization, lateral transport of radiation and margins. Med Phys 1996;23:2011–2022. 7. Webb S. Configuration options for intensity-modulated radiation therapy using multiple static fields shaped by a multileaf collimator. Phys Med Biol 1998;43:241–260. 8. Wang XH, Mohan R, Jackson A, et al.Optimization of intensity-modulated 3D conformal treatment plans based on biological indices. Radiother Oncol 1995;37:140 –152. 9. Boyer AL, Geis P, Grant W, et al.Modulated beam conformal therapy for head and neck tumors. Int J Radiat Oncol Biol Phys 1997;39:227–236. 10. Woo SY, Grant WH, Bellezza D, et al.A comparison of intensity modulated conformal therapy with a conventional external beam stereotactic radiosurgery system for the treatment of single and multiple intracranial lesions. Int J Radiat Oncol Biol Phys 1996;35:593–597. 11. Cardinale RM, Benedict SH, Wu Q, et al.A comparison of three stereotactic radiotherapy techniques: Arcs vs. non-coplanar fixed fields vs. intensity modulation. Int J Radiat Oncol Biol Phys 1998;42:431– 436.
12. Xia P, Geis P, Xing L, et al.Physical characteristics of a miniature multileaf collimator. Med Phys 1999;26:65–70. 13. Cosgrove VP, Jahn U, Pfaender M, et al.Commissioning of a micro multi-leaf collimator and planning system for stereotactic radiosurgery. Radiother Onco 1999;50:325–336 14. Wu Q, Mohan R. Algorithms, and functionality of an IMRT optimization system. Med Phys 2000;27:701–711 15. Cardinale RM, Benedict SH, Bump EA, et al.Determining optimal block margin for conformal stereotactic irradiation of lung and hepatic tumors. Int J Radiat Oncol Biol Phys 1999; 45:515–520. 16. Kubo HD, Pappas CTE, Wilder RB. A comparison of arcbased and static mini-multileaf collimator-based radiosurgery plans. Radiother Oncol 1997;45:89 –93. 17. Bourland JD, McCollough KP. Static field conformal stereotactic radiosurgery: Physical techniques. Int J Radiat Oncol Biol Phys 1993;28:471– 479. 18. Hamilton RJ, Kuchnir FT, Sweeney P, et al.Comparison of static conformal field with multiple noncoplanar arc techniques for stereotactic radiosurgery or stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 1995;33:1221–1228. 19. Shiu AS, Kooy HM, Ewton JR, et al.Comparison of miniature multileaf collimation (mmlc) with circular collimation for stereotactic treatment. Int J Radiat Oncol Biol Phys 1997;37: 679 – 688. 20. Marks LB, Sherouse GW, Das S, et al.Conformal radiation therapy with fixed shaped coplanar or noncoplanar radiation beam boquets: A possible alternative to radiosurgery. Int J Radiat Oncol Biol Phys 1995;33:1209 –1219. 21. Laing RW, Bentley RE, Nahum AE, et al.Stereotactic radiotherapy of irregular targets: A comparison between static conformal beams and non-coplanar arcs. Radiother Oncol 1993;28:241–246. 22. Kooy HM, Nedzi LA, Loeffler JS, et al. Treatment planning for stereotactic radiosurgery of intracranial leasions. Int J Radiat Oncol Biol Phys 1991;21:683– 693