Re-Planning for Compensator-Based IMRT with Original Compensators
Article Outline
Abstract
Compared with multileaf collimator (MLC)–based intensity-modulated radiotherapy (IMRT) for moving targets, compensator-based IMRT has advantages such as shorter beam-on time, fewer monitor units with potentially decreased secondary carcinogenesis risk, better optimization-to-deliverable dose conversion, and often better dose conformity. Some of the disadvantages include additional time for the compensators to be built and delivered, as well as extra cost. Patients undergoing treatment of abdominal cancers often experience weight loss. It would be necessary to account for this change in weight with a new plan and a second set of compensators. However, this would result in treatment delays and added costs. We have developed a method to re-plan the patient using the same set of compensators. Because the weight changes seen with the treatment of abdominal cancers are usually relatively small, a new 4D computed tomography (CT) acquired in the treatment position with markers on the original isocenter tattoos can be registered to the original planning scan. The contours of target volumes from the original scans are copied to the new scan after fusion. The original compensator set can be used together with a few field-in-field (FiF) beams defined by the MLC (or beams with cerrobend blocks for accelerators not equipped with a MLC). The weights of the beams with compensators are reduced so that the FiF or blocked beams can be optimized to mirror the original plan and dose distribution. Seven abdominal cancer cases are presented using this technique. The new plan on the new planning CT images usually has the same dosimetric quality as the original. The target coverage and dose uniformity are improved compared with the plan without FiF/block modification. Techniques combining additional FiF or blocked beams with the original compensators optimize the treatment plans when patients lose weight and save time and cost compared with generating plans with a new set of compensators.
Key Words: Compensator, IMRT, Field-in-field, Treatment planning
Introduction
Intensity modulation across the radiotherapy x-ray beam can be accomplished by either segmenting the field aperture with a multileaf collimator (MLC)1 or by introducing variable-thickness attenuators (compensators).2 Figure 1 illustrates how the compensator modulates the radiation beam intensity. The fluence maps for a set of beams are optimized to deliver the dose to the treatment target and spare the nearby critical structures. The uniform beam from the accelerator is differentially attenuated following the variation of the thickness of the compensator designed to reproduce this fluence map as closely as possible. Intensity-modulated radiotherapy (IMRT) can therefore be delivered using an isocentric set of beams modulated by the compensators.2
Fig. 1.
Illustration of intensity modulation by a compensator. The variation in attenuation, introduced by the variable compensator thickness, modulates the uniform beam from an accelerator.
Compared with MLC-based IMRT, compensator-based IMRT has the advantages of shorter treatment times,2, 3, 4 fewer monitor units (MUs)5 with less potential secondary carcinogenesis, better optimization-to-deliverable dose conversion,6 and usually better target conformity.7, 8 Prolonged fraction time could reduce treatment efficiency.9 Compensator-based IMRT in combination with a high-dose-rate linear accelerator would biologically enhance the treatment efficiency. With the compensator technique, IMRT treatments can be delivered with accelerators not equipped with a MLC. Disadvantages of compensator-based IMRT include limited intensity modulation range2 and additional time and cost for the compensators to be manufactured.
In the treatment of abdominal cancers, a change in patient weight before or during treatment is a frequent clinical problem. A re-plan could potentially improve dose distribution after weight change, avoiding the potential unwanted increase in dose to normal tissue.10 A new IMRT plan using the same compensators would save time without adding any extra cost. This paper introduces 2 methods of re-planning while retaining the original compensators: one uses MLC-defined field-in-field (FiF) technique and the other uses cerrobend blocks to simulate the use of accelerators without a MLC.
As a forward IMRT planning technique, FiF is often used to generate homogeneous dose in target volumes.11 Cerrobend blocks are often used in blocking normal structure volumes in 3D conformal treatment planning and delivery.12
Materials and Methods
Clinical cases
All cases included in this study were pancreatic cancer cases. The patients were treated to a dose of either 50 Gy prescribed to 95% of the planning target volume (PTV), with 2 Gy per fraction in 25 daily fractions, or 45 Gy to 95% of the PTV with 1.8 Gy per fraction in 25 fractions, followed by a 5.4 Gy boost to the gross disease. The patients treated to 50 Gy had borderline resectable disease so a dose painting strategy of 50 Gy to gross disease and 45 Gy to the clinical target volume was designed with volumes drawn on 4D CT scans to account for respiratory motion. In other cases, the patients had locally advanced disease and were treated on a protocol specifying that all target volumes receive 180 cGy per fraction. Compensator-based IMRT was chosen to optimize coverage to targets moving with respiration. All patients had weight loss during the treatment course. Re-planning was performed to confirm tumor coverage after weight loss and to reduce the possibility of overdosing the target.
Treatment planning
All treatment plans were generated using the XiO planning system (Version 4.34.02.1, CMS, Inc, St. Louis, MO) for an Oncor linear accelerator (Siemens Medical Solutions USA, Inc., Malvern, PA). A 5-beam IMRT plan was originally generated for each case. Based on the fluence maps of the intensity-modulated beams, the brass compensators were manufactured by a commercial vendor (.decimal, Inc., Sanford, FL).
New treatment planning CT scans were taken when weight loss was clinically noticeable, usually after the treatment source-to-surface distances (SSDs) were recorded to be out of tolerance. For the 7 cases studied in this paper, the average weight loss was 8.31 ± 4.39 kg (range 3.4–17.1 kg), or 11.67% ± 6.45% (range 3.6%–23.1%), with 1 standard deviation. The new CT images in the free breathing mode were registered to the original set. The original isocenter was determined before the PTV was contoured. After the registration of the new CT to the original one, the old PTV was copied to the new CT. The original isocenter was not automatically copied to the new CT. The new isocenter was often not at the same location as the original one in relation to the external markers on the isocenter tattoos. Necessary shifts were determined to place the isocenter at the same anatomical location after the patient was virtually triangulated on the original tattoos. Normal structures were drawn anew on the repeat CT. The original IMRT beams were copied to the new CT and aligned to the new anatomically correct isocenter.
Review of these plans demonstrated that the PTV coverage would be altered with increased hot spots if the number of monitor units (MUs) were not reduced for the new plan, because of decreased treatment depth. The number of MUs of each beam was reduced proportionally, so that the resulting PTV coverage was the same as in the original plans. With the reduced MU and same coverage to the PTV, the new plans often demonstrated higher dose inhomogeneity. The hot spots could be a few percentage points higher compared with the original plan.
MLC-defined FiF beams, often two of them, were introduced in the new plans to reduce the hot spots. The FiF beams included the compensators, so that the therapists did not have to enter the treatment room to remove the compensators to deliver the FiF beams. The selection of the beams to add the FiF beams is based on the locations of hot spots in the beam's eye view (BEV). The hot spots are easier to block with an MLC when they are at the field edge in the BEV. Because the MLC cannot block “island” areas, for the beams with the hot spots in the center of the aperture, FiF cannot improve the dose homogeneity. Figure 2A shows an example of an added FiF beam. The isodose line used to define the aperture was 111% of the prescribed dose. The high dose volume close to the treatment aperture edge was blocked, whereas those closer to the center of the field were not. To do otherwise would reduce the PTV coverage. The MU used for all the FiF beams were in the range of 10–12. The parent beam MUs were reduced by the number of the FiF MUs.
Fig. 2.
(A) An example of a FiF beam added to 1 of the compensator-based IMRT beams. The isodose volume to block in this example was 111% of the prescribed dose (PTV shown as red colorwash). The isodose volume close to the BEV edge was blocked by MLC, whereas the ones off the edge were not. (B) An example of a cerrobend block field added to 1 of the compensator-based IMRT beams. The blocked isodose level was 108% of the prescribed dose. All the hot spots were blocked regardless of the location in the BEV.
For the new plans that did not use a MLC, the blocks were drawn in the BEVs to reduce the hot spots. Two of the 5 beam angles were selected to add blocked beams to be consistent with the FiF plans. The compensators were removed from the blocked beams. Figure 2B shows an example of a blocked beam with the same gantry angle as one of the compensator-based IMRT beams. An attempt was made to block all the hot spots visible in the BEV, regardless of location relative to the field edge. MUs used in all the new beams were in the range of 4–6 when using the block technique. To maintain the same dose to the PTV from the pair of a parent and blocked beams, more MUs were deducted from the corresponding original beam to account for the compensator attenuation. The compensator was removed when the cerrobend block beam was delivered, whereas the compensator was in place when the MLC FiF beam was used. To obtain a similar dosimetric effect, more MUs are needed in the FiF technique to compensate for the attenuation of the compensator.
Because the primary concern with weight loss cases is the increased hot spots, the plan evaluation is first focused on the hot spot reduction. Dose homogeneity is also analyzed.
Results
Isocenter registration
Table 1 lists the shifts from the original tattoos needed to place the new isocenter at the same anatomical location as the previous locations.
Table 1. New isocenter shifts relative to the original isocenters⁎
| Case | AP/cm | Lateral/cm | SI/cm |
|---|---|---|---|
| 1 | 0.3 | −0.2 | 0.3 |
| 2 | 1.3 | −0.5 | 0.9 |
| 3 | 0.2 | −0.3 | 1.2 |
| 4 | 0.5 | −0.2 | 0.0 |
| 5 | −0.3 | 0.3 | 0.6 |
| 6 | 0.2 | 0.3 | 0.2 |
| 7 | 0.5 | 0.6 | 0.4 |
| Average | 0.4 ±0.5 | 0.0±0.4 | 0.5±0.4 |
⁎Abbreviations: AP, anterior-posterior; SI, superior-inferior. |
MLC FiF technique
With the FiF technique, the plan based on the new CT typically exhibits the dosimetric quality similar to the original one. This would not be the case without the added FiF fields. For example, in Case 1, the plan was modified on the new planning CT 2 weeks into the treatment due to weight loss. The FiF modification was applied to 2 of the 5 beams. Without the FiF fields, the high PTV dose on the new plan was 111% of the prescribed dose. With the FiF modification, the hot spot was reduced to 105%, which was only slightly higher than the original plan. Table 2 shows the reduction of high doses in all 7 cases. In 1 case (Case 3), the maximum dose was even lower than that in the original plan.
Table 2. Percentage high dose in PTV in reference to the prescribed dose
| Case | Weight Loss (kg/%) | Average SSD Increase (cm) | Original | New Without FiF | New With FiF |
|---|---|---|---|---|---|
| 1 | 9.5/14.4 | 1.0 | 103.2% | 111.4% | 105.4% |
| 2 | 7.5/7.8 | 1.0 | 103.8% | 111.0% | 107.0% |
| 3 | 5.8/11.6 | 1.1 | 112.2% | 116.7% | 111.6% |
| 4 | 8.9/14.5 | 0.6 | 104.9% | 109.3% | 106.3% |
| 5 | 6.0/6.7 | 1.0 | 102.8% | 105.6% | 104.8% |
| 6 | 3.4/3.6 | 0.3 | 103.8% | 104.9% | 104.2% |
| 7 | 17.1/23.1 | 1.2 | 111.5% | 121.9% | 112.2% |
Figure 3 shows a comparison of the isodose lines for Case 1 between the plans on the new planning CT with and without the FiF modification. Dose homogeneity improvement can be clearly seen. Figure 4 shows the DVH comparison for the same case.
Fig. 3.
A comparison of the isodose lines between the plans on the new planning CT with (A) and without (B) the FiF modification.
Fig. 4.
DVH comparison. (A) Original plan on the original CT; (B) new plans with and without FiF modification on the new planning CT.
The dose homogeneity improvement is not only reflected in the maximum relative dose reduction. For the same isodose level, the size of the hot spots is also reduced. For example, in Case 3, 8.0% of PTV receives 110% dose in the new plan without FiF, whereas 0.8% PTV receives the same relative dose in the new plan with FiF. Figure 5 shows the comparison of the percentage volume of PTV covered by the 105% isodose.
Fig. 5.
The hot spot size reduction of the new plans.
PTV coverage of the new plan could be improved with the FiF modification compared with the original plan on the original CT (Fig. 4). In Case 1, originally 97% of the PTV received the prescribed 50 Gy dose, whereas with the new plan 100% of the PTV was covered with the prescription isodose. This better coverage was the major reason that the maximum dose in the new plan was often slightly higher than that in the original, and they would be about the same if the PTV coverage was scaled down to the same value.
Cerrobend block technique
The same new isocenters used in the FiF plans were used in the plans with blocks.
The PTV dose-volume histogram (DVH) in all the cases was better for the block technique compared with the MLC FiF in terms of the dose homogeneity. For most of the normal structures, block technique also shows better DVH curves (Fig. 6). Unlike the MLC FiF, the block technique has little limitation on the hot spot location in the BEV. It can be used for either the hot spots at the aperture edge or “island” ones. Because of this advantage, the plans using block technique usually exhibit better dose homogeneity compared with the ones using FiF. Figure 7 demonstrates the hot spot reduction for both the FiF and block techniques. Noticeably high doses can be found if the original plans were used with the new CT data (second bar from left in each case). The hot spots are reduced by reducing the number of MUs only (third bar). The dose homogeneity is further improved after FiF or block modifications are made (fourth and fifth bars, respectively).
Fig. 6.
Comparison of new plans using FiF (dashed curves) and blocks (solid curves). Two additional beams were used in both plans.
Fig. 7.
Dose homogeneity improvement corresponding to the different replanning steps. High dose in all the cases is the result of using original plans on the new CT data without any change (light purple). Reducing the number of MUs reduces the hot spots (red). The high doses are further reduced by adding FiF (green) or blocked (dark purple) beams.
Because there is no hot spot location limitation when using the block technique, more beams can be used in the new plan. However more time and careful experimentation is required to adjust the beam weightings to optimize the dose homogeneity when more beams are added. The magnitude of dose homogeneity improvement is much smaller for the additional beams. Considering the prolonged planning and treatment delivery times with the limited dosimetric benefit, more than 2 blocked beams are not recommended.
MU adjustment only
We encountered 2 cases where an MU adjustment was made but no FiF or block beams were necessary in the new plans. In these cases, patient weight loss was relatively modest (less than 4%) compared with the ones that required FiF or block beam modifications. For all the cases in which FiF or block beam modification were found useful, the average SSD increase between the 5 beams was between 0.6 and 1.1 cm. When only the MUs needed to be adjusted, the average SSD increase was 0.1 and 0.3 cm.
Minimum, maximum, and mean doses and dose homogeneity
The new plans using either technique follow the trend of the original plans in minimum, maximum, and mean doses in the PTV. Figure 8 demonstrates close similarity in these parameters. Table 3 quantitatively shows the average differences of the 7 cases studied. The maximum difference between the new plans, including both FiF and block plans, and the original plans is <4% over all the minimum, maximum, and mean doses, with the exception of Case 7. This case exhibited the most significant weight loss. The relatively big drop in minimum doses in the new plans for Case 7 is because of the significant anatomical change.
Fig. 8.
Comparison of the minimum, maximum, and mean doses in PTV between the new and original plans for 7 cases. Cases 2, 3, 4, and 7 were treated with 45 Gy to 95% of the PTV, whereas the rest were treated with 50 Gy.
Table 3. Relative mean differences of minimum, maximum, and mean doses between the new and original plans
| Difference | Minimum Dose (%) | Maximum Dose (%) | Mean Dose (%) |
|---|---|---|---|
| FiF-original | −2.71±7.36 | 1.16±1.66 | 0.58±0.67 |
| Block-original | −1.51±5.76 | 0.66±1.72 | 0.50±0.64 |
Although dose homogeneity is worse in the new plans compared with the original ones, it is significantly improved with the FiF or block fields compared with the plans on the new CT but without the additional fields (Fig. 9). Although most of the original plans cover 95% of the PTV in the 95%–105% of the prescribed dose range, most of the new plans cover the 95% PTV in the 92%–108% range, which is clinically acceptable. Despite the significant weight loss and anatomical change in Case 7, the 95% PTV was covered within the dose range of 92%–108% with FiF added in the new plan.
Fig. 9.
Statistical dose homogeneity comparisons for the PTV between the original and new plans. The vertical error bars are the PTV coverage ranges of all 7 cases studied.
Normal structures
The normal structures contoured on the new CT are often quite different from the original. An example of stomach contour variation is shown in Fig. 10. Most likely, the differences are not because of weight change. For example, the volume and shape of the stomach depend on how much food and gas are present inside at the time of the CT scan.
Fig. 10.
The contour of the stomach from the repeat CT (orange) overlapped on the original CT with the original contour (purple).
In most cases, the DVH changes for the normal structures were more significant than for the PTV. Figure 11 shows the DVH differences for the kidneys, bowel, and stomach, which is the most significant variation encountered in all the cases. Because of the random nature of the daily variation of the normal structures, no certain pattern of the DVH variation can be derived from the studied cases. The DVH curves of the normal structures in the new plan could be better or worse than those in the original plans. With the original compensators, optimization in the new plan aiming at DVH improvement for normal structures is difficult. This optimization is not necessary unless adaptive plans are generated daily, which may not be practical for compensator-based IMRT in terms of planning and delivery time.
Fig. 11.
Example of the normal structure DVH variations.
Discussion
The reduction of hot spots using the FiF technique depends on their location. If the hot spots are at the edges of the beam aperture in 1 or more BEV, decreasing the dose in these regions can be easily accomplished. Further reduction cannot be achieved once the hot spots are no longer at the edges of any BEV aperture. For the plans with hot spots in the center of all the BEV openings, the FiF technique can do little to improve the dose homogeneity. Because of this, usually only 1 or 2 beams can be used to add FiF.
The reduction of the hot spots with the cerrobend blocks is easier to realize in treatment planning than using the FiF technique, because of the greater flexibility in the aperture shape afforded by the blocks. This also simplifies beam selection for modification with the block technique. The blocked beams can be added to all the IMRT beams, and can even be added as a new beam with its separate gantry angle. However, adding many new beams increases both planning and delivery time. One or 2 additional blocked beams appear to be a reasonable compromise in terms of both plan quality and delivery time.
The dose homogeneity improvement is achieved by adding the FiF fields or the blocked beams to the new plans. The maximum dose reduction is a result of the combination of MU reduction and the FiF/block modification of the new plans.
The treatment delivery time is longer for the new optimized IMRT treatments compared with the original ones because of the additional beams, especially if the block technique is used. The beam-on time for the new beams is short (10–12 MUs in each FiF beam and 4–6 MUs in each block beam).
Because the dosimetric difference between using the 2 techniques is not significant, when an accelerator equipped with a MLC is used, the FiF technique is preferable because of the shorter treatment time.
Conclusions
Failure to account for weight change during IMRT treatment of the abdomen could lead to suboptimal target coverage and unwanted normal tissue morbidity. To save time and cost, the original compensators can be used to generate a new IMRT plan when the patient's weight changes during treatment. Adding a few FiF segments can achieve a plan on the new CT, which is dosimetrically very similar to the original. For accelerators that are not equipped with MLC, cerrobend blocks can be used to improve dose homogeneity in the new IMRT plans using the same compensators.
References
- . Multileaf collimator leaf sequencing algorithm for intensity modulated beams with multiple static segments. Med. Phys. 1998;25:1424–1434
- Compensators: An alternative IMRT delivery technique. J. Appl. Clin. Med. Phys. 2004;5:15–36
- A multi-institutional retrospective study on clinical IMRT treatment delivery efficiency. Med. Phys. 2008;35:2746
- Quality assurance and dosimetric analysis of intensity modulated radiotherapy using compensators for head and neck cancers. Med. Phys. 2008;35:2763
- Dosimetric comparison of compensator and multi-leaf collimator (MLC) based intensity modulated radiosurgery (IMRS). Med. Phys. 2008;35:2826
- Comparison of compensator-IMRT and segmental MLC-IMRT techniques: A retrospective study on treatment time, monitor units, and dosimetry in clinical application. Med. Phys. 2007;34:2428
- . Investigation of the reliability, accuracy, and efficiency of gated IMRT delivery with a commercial linear accelerator. Med. Phys. 2007;34:2928–2938
- . Intensity modulation delivery techniques: “Step & shoot” MLC auto-sequence versus the use of a modulator. Med. Phys. 2000;27:948–959
- The effect of fraction time in intensity modulated radiotherapy: Theoretical and experimental evaluation of an optimisation problem. Radiother. Oncol. 2003;68:181–187
- Repeat CT imaging and replanning during the course of IMRT for head-and-neck cancer. Int. J. Radiat. Oncol. Biol. Phys. 2006;64:355–362
- Intensity modulation to improve dose uniformity with tangential breast radiotherapy: Initial clinical experience. Int. J. Radiat. Oncol. Biol. Phys. 2000;48:1559–1568
- Randomized trial of radiotherapy to the thorax in limited small-cell carcinoma of the lung treated with multiagent chemotherapy and elective brain irradiation: A preliminary report. J. Clin. Oncol. 1984;2:1200–1208
PII: S0958-3947(10)00013-0
doi:10.1016/j.meddos.2010.01.004
© 2011 American Association of Medical Dosimetrists. Published by Elsevier Inc. All rights reserved.
