- Email: [email protected]
Rotational setup errors in pediatric stereotactic radiation therapy
Practical Radiation Oncology (2013) 3, 194–198
www.practicalradonc.org
Original Report
Rotational setup errors in pediatric stereotactic radiation therapy Cem Altunbas PhD a,⁎, Todd C. Hankinson MD b , Moyed Miften PhD a , Tiffany Tello BS a , Steven R. Plimpton MS a , Kelly Stuhr MS c , Arthur K. Liu MD, PhD a a
Department of Radiation Oncology, University of Colorado School of Medicine, Aurora, Colorado Department of Neurosurgery, Children's Hospital Colorado, Aurora, Colorado c Department of Radiation Oncology, University of Colorado Hospital, Aurora, Colorado b
Received 30 April 2012; revised 20 August 2012; accepted 20 August 2012
Abstract Purpose: Stereotactic radiation therapy (SRT) is an increasingly commonly used technique in children. The use of image guidance increases the ability to accurately position patients. With our robotic couch, rotational errors that can be corrected are limited to approximately 3 degrees. Given this limitation, we reviewed the rotational setup errors in our pediatric brain tumor population. Methods and Materials: We reviewed the rotational corrections for all pediatric (age ≤21 years old) patients treated at our facility from 2009 to 2011. We compared children b5 years old treated to children between 5 and 21 years old (≥5 years old). Also, we analyzed the effect of steroid use and trends in rotational errors over the treatment period in each age group. Results: The mean pitch, roll, and yaw rotational setup errors for younger children are −0.70 ± 2.60 degrees, −0.06 ± 1.89 degrees, and 0.69 ± 2.42 degrees, respectively; for children ≥5 years old, they are 0.46 ± 2.09 degrees, −0.06 ± 1.89 degrees, and 0.69 ± 2.42 degrees, respectively. The mean pitch corrections are larger for children b5 years old (P b .001) and the variance of the pitch, roll, and yaw corrections are all larger for children b5 years old (P b .001). The frequency of rotational errors above 3 degrees for pitch, roll, and yaw is 21.7%, 10.6%, and 20.9% for children b5 years old, and 15.6%, 2.1%, and 13.8% for children ≥5 years old. In both age groups, pitch and roll corrections were larger for children treated with steroids. Conclusions: Rotational errors in our pediatric population occur more frequently than previously reported, and are more common in younger children and in children treated with steroids. These rotational set up errors may not be fully correctable due to mechanical and safety limitations. We have altered our planning and treatment process to better account for rotational errors in children receiving SRT. © 2013 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.
Conflicts of interest: None. ⁎ Corresponding author. Department of Radiation Oncology, University of Colorado School of Medicine, Anschutz Cancer Pavilion, MS F706, 1665 Aurora Ct, PO Box 6510, Aurora, CO 80045. E-mail address: [email protected] (C. Altunbas). 1879-8500/$ – see front matter © 2013 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.prro.2012.08.003
Practical Radiation Oncology: July-September 2013
Introduction Management for pediatric brain tumors typically includes surgery, chemotherapy, and for many children, radiation therapy. Improvements in radiation therapy delivery include the development of stereotactic radiation therapy (SRT), which provides accurate immobilization and daily image guidance. 1-3 SRT allows a reduction in the expansion margin required for setup errors and as a result reduces the amount of irradiated normal tissue. Treatment of children with SRT increases the complexity of patient positioning. Young children typically require anesthesia to tolerate treatment. Patient setup for anesthesia is often different than setup for awake patients as airway positioning is critical for children receiving general anesthesia. Also, immobilization devices are not specifically designed for the smaller anatomy of children. The use of image guidance can reduce setup uncertainties. Translational errors can typically be fully corrected on the majority of treatment machines. However, rotational errors may not be fully correctable. Existing studies of rotational errors in children and adults have shown small rotational errors in the majority of cases. 4-8 In our facility, we utilize a Novalis Brainlab (BrainLAB AG, Feldkirchen, Germany) with orthogonal kilovolt imaging for daily image guidance. An Exactrac (BrainLAB) system aligns the patient, allowing for position correction in 6 degrees of freedom. While all translational errors can be corrected, the Exactrac robotic couch is limited in the amount of rotation (pitch, roll, and yaw) that can be achieved due to mechanical limitations and concerns regarding patient safety. Excessive rotation of the table could result in the patient falling off the table. This limit is approximately 3 degrees, but maximum achievable correction depends on the total rotational correction required and actual couch position parameters. Given this rotational correction limitation, we examined the amount and frequency of rotational errors in our pediatric population.
Methods and materials We reviewed the rotational corrections for pediatric patients (age ≤21 years old) treated at our facility from 2009 to 2011 (prior data were not available for analysis). For all patients treated on our Novalis Brainlab linac, patients are immobilized using Brainlab's thermoplastic mask, and positioned using the Novalis Body system and a robotic couch. If a patient is treated with anesthesia, a small aperture around the mouth and nose is cut in the mask to accommodate oxygen delivery and carbon dioxide monitoring. The patient's mask is attached to a cranial array, which is a mask holder with 6 infrared reflective markers. For children requiring anesthesia, the patient is placed on a 7-cm thick Styrofoam (Dow Chemical
Rotational error in pediatric SRT
195
Company) platform to optimize airway positioning. An infrared camera detects the couch position via reflective markers, and drives the couch to the treatment position. Then, orthogonal kilovolt images are taken and aligned with the bony anatomy from corresponding digitally reconstructed radiographs (DRRs) from the planning computed tomography (CT). After the verification of DRR-kV (kilovolt) image registration, position corrections are applied in both translational and rotational directions using the robotic couch. This localization procedure was used for every treatment fraction for all patients. The system records the translational and rotational corrections that were required for patient alignment, along with the actual rotational correction that was achieved. This system has a positioning accuracy of less than 1 mm. 1-3 We evaluated the rotational corrections calculated after DRR-kV image registration process. For further analyses, we compared the corrections for children b5 years old and ≥5 years old. In addition, we compared the mean of the rotational errors using a Student t test and the variance of the rotational errors using an F test. In each age group, we investigated the effect of steroid use and trends in rotational errors over the course of treatment.
Results Forty-three pediatric patients who were treated for intracranial lesions were available for review. The most common treatment site was the posterior fossa (n = 15). For these patients, 873 rotational corrections were available. A summary of the patients is given in Table 1. The mean and standard deviations of rotational corrections for each individual patient in the order of increasing age are shown in Fig 1. The mean rotational errors for the 2 age groups are shown in Table 2. The difference between the younger (b5 years old) and the older (≥5 years old) children in mean pitch errors was 1.16 degrees (P b .001).
Table 1
Summary of patients evaluated
Variable
b5 years old
≥5 years old
No. patients Median (range) age No. patients treated under anesthesia No. patients receiving steroids Total no. of image localization data analyzed Median (range) no. of localization data per patient
18 2.5 (0.88-4.87) 18
25 11.3 (5.01-21.9) 4
5
11
387
486
25 (2-41)
18 (3-36)
196
C. Altunbas et al
Practical Radiation Oncology: July-September 2013
Figure 1 Mean and standard deviation of angular shifts per patient. Patients were sorted as a function of age. Horizontal lines indicate the overall mean and standard deviation of mean patient shifts. The vertical line indicates the age threshold of 5 (ie, patients 1 to 18 are b5 years old, and patients 19 to 43 are ≥5 years old).
There were no statistically significant differences in mean roll or yaw errors. When comparing the variance of the 2 groups, younger children had a larger variance in rotational errors for all directions (P b .001). As there are limitations to the amount of rotational errors that can be corrected, we also determined the frequency of large errors. The cumulative shift histograms for the magnitude of pitch, roll, and yaw rotational errors are shown in Fig 2. The fractions of pitch, roll, and yaw errors above 3 degrees, which is approximately the amount of rotational error that is correctable by the robotic couch, are 21.7%, 10.6%, and 20.9% for children b5 years old, and 15.6%, 2.1%, and 13.8% for children ≥5 years old. Large pitch and yaw corrections are more common than roll corrections for all children. The younger children have more frequent large rotational errors than the older children. In each age group, we evaluated whether steroid treatments correlated with rotational errors. Sixteen patients received steroids during the treatment period and 5 of these patients were b5 years old. In both age groups, patients who were treated with steroids had larger pitch and roll corrections on the average (Tables 3 and 4). However, in older children the mean yaw rotational errors were larger in those who did not receive steroids. All children b5 years old and only 4 children ≥5 years old were treated under anesthesia; hence, we did not evaluate the effects of anesthesia in each age group. Finally, we investigated whether the length of treatment course has an impact on rotational shifts. A statistically significant
Table 2
Mean rotational setup errors in each age group
Rotation b5 years old Pitch Roll Yaw a b
≥5 years old
Difference Difference in mean a in variance b
−0.70 ± 2.60 0.46 ± 2.09 b.001 −0.06 ± 1.89 0.13 ± 1.3 b.08 0.69 ± 2.42 0.76 ± 1.84 b.683 Student t test P value. F test P value.
b.001 b.001 b.001
correlation between the variation in rotational shifts and the length of the treatment course was not identified.
Discussion Rotational errors in patient positioning have been reported in adult brain tumor patients. Using 300 daily pretreatment megavoltage CT scans from a tomotherapy unit, mean roll rotational errors were 0.8 degrees with only 5% of the errors greater than 3 degrees, 0.7% greater than 4 degrees, and no roll errors greater than 5 degrees. 7 The couch for tomotherapy units cannot correct for pitch or yaw errors, so those were not reported. In a similar study of rotational setup errors in pediatric brain tumor patients, 1016 cone-beam CT scans from 21 patients were analyzed. 8 Mean rotational errors for pitch, roll, and yaw were 0.37, 0.29, and 0.03 degrees, respectively. Eighteen percent of the fractions had rotational errors greater than 2 degrees, 5% greater than 3 degrees, and 0.9% greater than 4 degrees. They found no effect on rotational errors in children treated under general anesthesia. Other studies of rotational errors in adult patients with prostate, head and neck, and lung cancer also report small rotational errors that are typically less than 2 degrees. 4-7 In our pediatric patients, rotational errors occurred more frequently in patients b5 years old compared with children who were older. Errors greater than 3 degrees, which can result in uncorrected rotational error, occurred in over 20% of the fractions of younger children compared with 15% in the older children. Our higher incidence of large rotational errors compared with the other published pediatric study 8 may reflect differences in the immobilization devices used. To a lesser extent, the localization imaging modality (orthogonal kilovolt image pairs versus 3-dimensional) might have also played a role in rotational error differences between the 2 studies. The use of steroids correlated with larger errors in pitch and roll. Steroids are well known to cause an increase in
Practical Radiation Oncology: July-September 2013
Rotational error in pediatric SRT
197
Figure 2 Cumulative shift histograms (CSH) of rotational error. Patients b5 years old (y.o.) are shown with the solid line and ≥5 years old are shown with the dashed line.
weight. These changes in anatomy likely result in the increase in setup errors. However, we are not able to explain the decrease in yaw errors in older children who received steroids. Although this work did not specifically study the cause of the larger variance in rotational errors seen in younger patients, we hypothesize 2 potential issues that may lead to increased rotational setup errors for these children. First, we have observed that properly fitting an immobilization mask is more challenging for younger children. The smaller head size, less well-developed facial features, and anesthesia requirements can result in less well fitting masks. Also, we have found that the thermoplastic mask continues to shrink. We scanned 5 children 1 day after the initial CT simulation and have found in all cases there are changes in head positioning. Our clinical investigations indicated that mask shrinkage is a short-term effect, and it mainly occurs within the first 1 to 2 days after simulation. Based on these observations, we modified our CT simulation work flow for pediatric patients; instead of making the mask and performing the CT scan in the same session, we perform the planning CT scan 2 days after the mask was made. As a result of this change, we observed that the magnitude of rotational setup corrections was reduced during patient localization. Setup errors are accounted for in the planning target volume (PTV). Estimates of systematic and random translational errors can be incorporated into the PTV margin
in a straightforward manner. 9 However, the inclusion of rotational setup errors into the PTV will depend on the shape of the target and the location of the center of the rotational error. For example, if the target is perfectly spherical and the rotation center is in the center of the sphere, rotational errors should not result in additional setup error. However, for a more irregular shape and eccentrically placed isocenter, rotation errors can result in large position errors. An example is shown in Fig 3, where a 5-degree rotational error results in a shift of over 4 mm at the edge of the target, which exceeds our typical stereotactic PTV expansion of 3 mm. This error would be in addition to translational setup errors. In this example, for a prescription dose of 54 Gy, the minimum dose to the PTV would be reduced from 51.8 Gy to 43.2 Gy in the region of the cervical spine. A planning study on pediatric brain tumor patients showed that uncorrected errors greater than 4 degrees can result in changes in the generalized equivalent uniform dose of the PTV of 5% and for serial normal structures of 10%. 8 Even with the use of image guided radiation therapy, rotational errors of large magnitude are often of concern because in many situations those errors cannot be corrected, in contrast to translation errors. For example, our robotic couch cannot fully correct for rotational errors above approximately 3 degrees; tomotherapy can only correct for roll and other couches cannot correct for any rotational error. In the majority of adult studies, large rotational errors (greater than 3 degrees) are uncommon. 7
Table 3 Mean rotational setup errors in the younger group (b5 years old) stratified based on steroid treatments
Table 4 Mean rotational setup errors in the older age group (≥5 years old) stratified based on steroid treatments
Rotation Steroid treatment
Rotation Steroid treatment
Pitch Roll Yaw a b
No steroid treatment
Difference Difference in mean a in variance b
−1.49 ± 2.60 −0.4 ± 2.55 b.001 1.32 ± 1.83 −0.59 ± 1.63 b.001 0.56 ± 2.42 0.75 ± 1.84 .53 Student t test P value. F test P value.
.39 .07 b.02
Pitch Roll Yaw a b
No steroid treatment
Difference Difference in mean a in variance b
1.02 ± 2.12 −0.105 ± 1.91 b.001 0.26 ± 1.26 −0.01 ± 1.33 b.03 0.25 ± 1.77 1.28 ± 1.78 b.01 Student t test P value. F test P value.
b.06 .2 .46
198
C. Altunbas et al
Practical Radiation Oncology: July-September 2013
Figure 3 Visualization of pitch error on a sagittal computed tomographic slice of a patient b5 years old. On the left, the patient is in the correct position and planning target volume is displayed. On the right, the patient has 5 degrees of uncorrected pitch. Isocenter is marked with a cross. In this example, 5 degrees of pitch error leads to a 4.4-mm error at 5-cm distance from the isocenter.
In our pediatric patient population, we found that these errors occurred in a significant portion of treatments. As a result of our study, we have altered our process for treating young children. We perform a repeat CT simulation 1 to 2 days after the initial simulation to allow for additional mask shrinkage. We increase the PTV expansion when there is a large distance from the edge of the clinical target volume to the isocenter. We also have set a threshold at 3 degrees for uncorrected rotation errors that require corrective action. In those situations we remove the immobilization mask and reposition the patient. If the uncorrected rotational errors are still above the 3-degree threshold, physician evaluation is required. The decision to resimulate and replan the patient is case specific and depends on multiple parameters such as magnitude of shifts, PTV shape and location, and number of fractions remaining to be delivered.
Conclusions Rotational positioning errors in children (in particular those b5 years old) treated with SRT are larger and more frequent than in adults. These errors may not be fully corrected even with image guided radiation therapy and require modifications to the planning and treatment localization verification.
References 1. Yan H, Yin FF, Kim JH. A phantom study on the positioning accuracy of the Novalis Body system. Med Phys. 2003;30:3052-3060. 2. Lamba M, Breneman JC, Warnick RE. Evaluation of image-guided positioning for frameless intracranial radiosurgery. Int J Radiat Oncol Biol Phys. 2009;74:913-919. 3. Gevaert T, Verellen D, Tournel K, et al. Setup accuracy of the Novalis ExacTrac 6DOF system for frameless radiosurgery. Int J Radiat Oncol Biol Phys. 2012;82:1627-1635. 4. Aubry JF, Beaulieu L, Girouard LM, et al. Measurements of intrafraction motion and interfraction and intrafraction rotation of prostate by three-dimensional analysis of daily portal imaging with radiopaque markers. Int J Radiat Oncol Biol Phys. 2004;60:30-39. 5. Guckenberger M, Meyer J, Vordermark D, Baier K, Wilbert J, Flentje M. Magnitude and clinical relevance of translational and rotational patient setup errors: a cone-beam CT study. Int J Radiat Oncol Biol Phys. 2006;65:934-942. 6. Kaiser A, Schultheiss TE, Wong JY, et al. Pitch, roll, and yaw variations in patient positioning. Int J Radiat Oncol Biol Phys. 2006;66:949-955. 7. Schubert LK, Westerly DC, Tomé WA, et al. A comprehensive assessment by tumor site of patient setup using daily MVCT imaging from more than 3,800 helical tomotherapy treatments. Int J Radiat Oncol Biol Phys. 2009;73:1260-1269. 8. Beltran C, Pegram A, Merchant TE. Dosimetric consequences of rotational errors in radiation therapy of pediatric brain tumor patients. Radiother Oncol. 2012;102:206-209. 9. van Herk M, Remeijer P, Lebesque JV. Inclusion of geometric uncertainties in treatment plan evaluation. Int J Radiat Oncol Biol Phys. 2002;52:1407-1422.
www.practicalradonc.org
Original Report
Rotational setup errors in pediatric stereotactic radiation therapy Cem Altunbas PhD a,⁎, Todd C. Hankinson MD b , Moyed Miften PhD a , Tiffany Tello BS a , Steven R. Plimpton MS a , Kelly Stuhr MS c , Arthur K. Liu MD, PhD a a
Department of Radiation Oncology, University of Colorado School of Medicine, Aurora, Colorado Department of Neurosurgery, Children's Hospital Colorado, Aurora, Colorado c Department of Radiation Oncology, University of Colorado Hospital, Aurora, Colorado b
Received 30 April 2012; revised 20 August 2012; accepted 20 August 2012
Abstract Purpose: Stereotactic radiation therapy (SRT) is an increasingly commonly used technique in children. The use of image guidance increases the ability to accurately position patients. With our robotic couch, rotational errors that can be corrected are limited to approximately 3 degrees. Given this limitation, we reviewed the rotational setup errors in our pediatric brain tumor population. Methods and Materials: We reviewed the rotational corrections for all pediatric (age ≤21 years old) patients treated at our facility from 2009 to 2011. We compared children b5 years old treated to children between 5 and 21 years old (≥5 years old). Also, we analyzed the effect of steroid use and trends in rotational errors over the treatment period in each age group. Results: The mean pitch, roll, and yaw rotational setup errors for younger children are −0.70 ± 2.60 degrees, −0.06 ± 1.89 degrees, and 0.69 ± 2.42 degrees, respectively; for children ≥5 years old, they are 0.46 ± 2.09 degrees, −0.06 ± 1.89 degrees, and 0.69 ± 2.42 degrees, respectively. The mean pitch corrections are larger for children b5 years old (P b .001) and the variance of the pitch, roll, and yaw corrections are all larger for children b5 years old (P b .001). The frequency of rotational errors above 3 degrees for pitch, roll, and yaw is 21.7%, 10.6%, and 20.9% for children b5 years old, and 15.6%, 2.1%, and 13.8% for children ≥5 years old. In both age groups, pitch and roll corrections were larger for children treated with steroids. Conclusions: Rotational errors in our pediatric population occur more frequently than previously reported, and are more common in younger children and in children treated with steroids. These rotational set up errors may not be fully correctable due to mechanical and safety limitations. We have altered our planning and treatment process to better account for rotational errors in children receiving SRT. © 2013 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved.
Conflicts of interest: None. ⁎ Corresponding author. Department of Radiation Oncology, University of Colorado School of Medicine, Anschutz Cancer Pavilion, MS F706, 1665 Aurora Ct, PO Box 6510, Aurora, CO 80045. E-mail address: [email protected] (C. Altunbas). 1879-8500/$ – see front matter © 2013 American Society for Radiation Oncology. Published by Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.prro.2012.08.003
Practical Radiation Oncology: July-September 2013
Introduction Management for pediatric brain tumors typically includes surgery, chemotherapy, and for many children, radiation therapy. Improvements in radiation therapy delivery include the development of stereotactic radiation therapy (SRT), which provides accurate immobilization and daily image guidance. 1-3 SRT allows a reduction in the expansion margin required for setup errors and as a result reduces the amount of irradiated normal tissue. Treatment of children with SRT increases the complexity of patient positioning. Young children typically require anesthesia to tolerate treatment. Patient setup for anesthesia is often different than setup for awake patients as airway positioning is critical for children receiving general anesthesia. Also, immobilization devices are not specifically designed for the smaller anatomy of children. The use of image guidance can reduce setup uncertainties. Translational errors can typically be fully corrected on the majority of treatment machines. However, rotational errors may not be fully correctable. Existing studies of rotational errors in children and adults have shown small rotational errors in the majority of cases. 4-8 In our facility, we utilize a Novalis Brainlab (BrainLAB AG, Feldkirchen, Germany) with orthogonal kilovolt imaging for daily image guidance. An Exactrac (BrainLAB) system aligns the patient, allowing for position correction in 6 degrees of freedom. While all translational errors can be corrected, the Exactrac robotic couch is limited in the amount of rotation (pitch, roll, and yaw) that can be achieved due to mechanical limitations and concerns regarding patient safety. Excessive rotation of the table could result in the patient falling off the table. This limit is approximately 3 degrees, but maximum achievable correction depends on the total rotational correction required and actual couch position parameters. Given this rotational correction limitation, we examined the amount and frequency of rotational errors in our pediatric population.
Methods and materials We reviewed the rotational corrections for pediatric patients (age ≤21 years old) treated at our facility from 2009 to 2011 (prior data were not available for analysis). For all patients treated on our Novalis Brainlab linac, patients are immobilized using Brainlab's thermoplastic mask, and positioned using the Novalis Body system and a robotic couch. If a patient is treated with anesthesia, a small aperture around the mouth and nose is cut in the mask to accommodate oxygen delivery and carbon dioxide monitoring. The patient's mask is attached to a cranial array, which is a mask holder with 6 infrared reflective markers. For children requiring anesthesia, the patient is placed on a 7-cm thick Styrofoam (Dow Chemical
Rotational error in pediatric SRT
195
Company) platform to optimize airway positioning. An infrared camera detects the couch position via reflective markers, and drives the couch to the treatment position. Then, orthogonal kilovolt images are taken and aligned with the bony anatomy from corresponding digitally reconstructed radiographs (DRRs) from the planning computed tomography (CT). After the verification of DRR-kV (kilovolt) image registration, position corrections are applied in both translational and rotational directions using the robotic couch. This localization procedure was used for every treatment fraction for all patients. The system records the translational and rotational corrections that were required for patient alignment, along with the actual rotational correction that was achieved. This system has a positioning accuracy of less than 1 mm. 1-3 We evaluated the rotational corrections calculated after DRR-kV image registration process. For further analyses, we compared the corrections for children b5 years old and ≥5 years old. In addition, we compared the mean of the rotational errors using a Student t test and the variance of the rotational errors using an F test. In each age group, we investigated the effect of steroid use and trends in rotational errors over the course of treatment.
Results Forty-three pediatric patients who were treated for intracranial lesions were available for review. The most common treatment site was the posterior fossa (n = 15). For these patients, 873 rotational corrections were available. A summary of the patients is given in Table 1. The mean and standard deviations of rotational corrections for each individual patient in the order of increasing age are shown in Fig 1. The mean rotational errors for the 2 age groups are shown in Table 2. The difference between the younger (b5 years old) and the older (≥5 years old) children in mean pitch errors was 1.16 degrees (P b .001).
Table 1
Summary of patients evaluated
Variable
b5 years old
≥5 years old
No. patients Median (range) age No. patients treated under anesthesia No. patients receiving steroids Total no. of image localization data analyzed Median (range) no. of localization data per patient
18 2.5 (0.88-4.87) 18
25 11.3 (5.01-21.9) 4
5
11
387
486
25 (2-41)
18 (3-36)
196
C. Altunbas et al
Practical Radiation Oncology: July-September 2013
Figure 1 Mean and standard deviation of angular shifts per patient. Patients were sorted as a function of age. Horizontal lines indicate the overall mean and standard deviation of mean patient shifts. The vertical line indicates the age threshold of 5 (ie, patients 1 to 18 are b5 years old, and patients 19 to 43 are ≥5 years old).
There were no statistically significant differences in mean roll or yaw errors. When comparing the variance of the 2 groups, younger children had a larger variance in rotational errors for all directions (P b .001). As there are limitations to the amount of rotational errors that can be corrected, we also determined the frequency of large errors. The cumulative shift histograms for the magnitude of pitch, roll, and yaw rotational errors are shown in Fig 2. The fractions of pitch, roll, and yaw errors above 3 degrees, which is approximately the amount of rotational error that is correctable by the robotic couch, are 21.7%, 10.6%, and 20.9% for children b5 years old, and 15.6%, 2.1%, and 13.8% for children ≥5 years old. Large pitch and yaw corrections are more common than roll corrections for all children. The younger children have more frequent large rotational errors than the older children. In each age group, we evaluated whether steroid treatments correlated with rotational errors. Sixteen patients received steroids during the treatment period and 5 of these patients were b5 years old. In both age groups, patients who were treated with steroids had larger pitch and roll corrections on the average (Tables 3 and 4). However, in older children the mean yaw rotational errors were larger in those who did not receive steroids. All children b5 years old and only 4 children ≥5 years old were treated under anesthesia; hence, we did not evaluate the effects of anesthesia in each age group. Finally, we investigated whether the length of treatment course has an impact on rotational shifts. A statistically significant
Table 2
Mean rotational setup errors in each age group
Rotation b5 years old Pitch Roll Yaw a b
≥5 years old
Difference Difference in mean a in variance b
−0.70 ± 2.60 0.46 ± 2.09 b.001 −0.06 ± 1.89 0.13 ± 1.3 b.08 0.69 ± 2.42 0.76 ± 1.84 b.683 Student t test P value. F test P value.
b.001 b.001 b.001
correlation between the variation in rotational shifts and the length of the treatment course was not identified.
Discussion Rotational errors in patient positioning have been reported in adult brain tumor patients. Using 300 daily pretreatment megavoltage CT scans from a tomotherapy unit, mean roll rotational errors were 0.8 degrees with only 5% of the errors greater than 3 degrees, 0.7% greater than 4 degrees, and no roll errors greater than 5 degrees. 7 The couch for tomotherapy units cannot correct for pitch or yaw errors, so those were not reported. In a similar study of rotational setup errors in pediatric brain tumor patients, 1016 cone-beam CT scans from 21 patients were analyzed. 8 Mean rotational errors for pitch, roll, and yaw were 0.37, 0.29, and 0.03 degrees, respectively. Eighteen percent of the fractions had rotational errors greater than 2 degrees, 5% greater than 3 degrees, and 0.9% greater than 4 degrees. They found no effect on rotational errors in children treated under general anesthesia. Other studies of rotational errors in adult patients with prostate, head and neck, and lung cancer also report small rotational errors that are typically less than 2 degrees. 4-7 In our pediatric patients, rotational errors occurred more frequently in patients b5 years old compared with children who were older. Errors greater than 3 degrees, which can result in uncorrected rotational error, occurred in over 20% of the fractions of younger children compared with 15% in the older children. Our higher incidence of large rotational errors compared with the other published pediatric study 8 may reflect differences in the immobilization devices used. To a lesser extent, the localization imaging modality (orthogonal kilovolt image pairs versus 3-dimensional) might have also played a role in rotational error differences between the 2 studies. The use of steroids correlated with larger errors in pitch and roll. Steroids are well known to cause an increase in
Practical Radiation Oncology: July-September 2013
Rotational error in pediatric SRT
197
Figure 2 Cumulative shift histograms (CSH) of rotational error. Patients b5 years old (y.o.) are shown with the solid line and ≥5 years old are shown with the dashed line.
weight. These changes in anatomy likely result in the increase in setup errors. However, we are not able to explain the decrease in yaw errors in older children who received steroids. Although this work did not specifically study the cause of the larger variance in rotational errors seen in younger patients, we hypothesize 2 potential issues that may lead to increased rotational setup errors for these children. First, we have observed that properly fitting an immobilization mask is more challenging for younger children. The smaller head size, less well-developed facial features, and anesthesia requirements can result in less well fitting masks. Also, we have found that the thermoplastic mask continues to shrink. We scanned 5 children 1 day after the initial CT simulation and have found in all cases there are changes in head positioning. Our clinical investigations indicated that mask shrinkage is a short-term effect, and it mainly occurs within the first 1 to 2 days after simulation. Based on these observations, we modified our CT simulation work flow for pediatric patients; instead of making the mask and performing the CT scan in the same session, we perform the planning CT scan 2 days after the mask was made. As a result of this change, we observed that the magnitude of rotational setup corrections was reduced during patient localization. Setup errors are accounted for in the planning target volume (PTV). Estimates of systematic and random translational errors can be incorporated into the PTV margin
in a straightforward manner. 9 However, the inclusion of rotational setup errors into the PTV will depend on the shape of the target and the location of the center of the rotational error. For example, if the target is perfectly spherical and the rotation center is in the center of the sphere, rotational errors should not result in additional setup error. However, for a more irregular shape and eccentrically placed isocenter, rotation errors can result in large position errors. An example is shown in Fig 3, where a 5-degree rotational error results in a shift of over 4 mm at the edge of the target, which exceeds our typical stereotactic PTV expansion of 3 mm. This error would be in addition to translational setup errors. In this example, for a prescription dose of 54 Gy, the minimum dose to the PTV would be reduced from 51.8 Gy to 43.2 Gy in the region of the cervical spine. A planning study on pediatric brain tumor patients showed that uncorrected errors greater than 4 degrees can result in changes in the generalized equivalent uniform dose of the PTV of 5% and for serial normal structures of 10%. 8 Even with the use of image guided radiation therapy, rotational errors of large magnitude are often of concern because in many situations those errors cannot be corrected, in contrast to translation errors. For example, our robotic couch cannot fully correct for rotational errors above approximately 3 degrees; tomotherapy can only correct for roll and other couches cannot correct for any rotational error. In the majority of adult studies, large rotational errors (greater than 3 degrees) are uncommon. 7
Table 3 Mean rotational setup errors in the younger group (b5 years old) stratified based on steroid treatments
Table 4 Mean rotational setup errors in the older age group (≥5 years old) stratified based on steroid treatments
Rotation Steroid treatment
Rotation Steroid treatment
Pitch Roll Yaw a b
No steroid treatment
Difference Difference in mean a in variance b
−1.49 ± 2.60 −0.4 ± 2.55 b.001 1.32 ± 1.83 −0.59 ± 1.63 b.001 0.56 ± 2.42 0.75 ± 1.84 .53 Student t test P value. F test P value.
.39 .07 b.02
Pitch Roll Yaw a b
No steroid treatment
Difference Difference in mean a in variance b
1.02 ± 2.12 −0.105 ± 1.91 b.001 0.26 ± 1.26 −0.01 ± 1.33 b.03 0.25 ± 1.77 1.28 ± 1.78 b.01 Student t test P value. F test P value.
b.06 .2 .46
198
C. Altunbas et al
Practical Radiation Oncology: July-September 2013
Figure 3 Visualization of pitch error on a sagittal computed tomographic slice of a patient b5 years old. On the left, the patient is in the correct position and planning target volume is displayed. On the right, the patient has 5 degrees of uncorrected pitch. Isocenter is marked with a cross. In this example, 5 degrees of pitch error leads to a 4.4-mm error at 5-cm distance from the isocenter.
In our pediatric patient population, we found that these errors occurred in a significant portion of treatments. As a result of our study, we have altered our process for treating young children. We perform a repeat CT simulation 1 to 2 days after the initial simulation to allow for additional mask shrinkage. We increase the PTV expansion when there is a large distance from the edge of the clinical target volume to the isocenter. We also have set a threshold at 3 degrees for uncorrected rotation errors that require corrective action. In those situations we remove the immobilization mask and reposition the patient. If the uncorrected rotational errors are still above the 3-degree threshold, physician evaluation is required. The decision to resimulate and replan the patient is case specific and depends on multiple parameters such as magnitude of shifts, PTV shape and location, and number of fractions remaining to be delivered.
Conclusions Rotational positioning errors in children (in particular those b5 years old) treated with SRT are larger and more frequent than in adults. These errors may not be fully corrected even with image guided radiation therapy and require modifications to the planning and treatment localization verification.
References 1. Yan H, Yin FF, Kim JH. A phantom study on the positioning accuracy of the Novalis Body system. Med Phys. 2003;30:3052-3060. 2. Lamba M, Breneman JC, Warnick RE. Evaluation of image-guided positioning for frameless intracranial radiosurgery. Int J Radiat Oncol Biol Phys. 2009;74:913-919. 3. Gevaert T, Verellen D, Tournel K, et al. Setup accuracy of the Novalis ExacTrac 6DOF system for frameless radiosurgery. Int J Radiat Oncol Biol Phys. 2012;82:1627-1635. 4. Aubry JF, Beaulieu L, Girouard LM, et al. Measurements of intrafraction motion and interfraction and intrafraction rotation of prostate by three-dimensional analysis of daily portal imaging with radiopaque markers. Int J Radiat Oncol Biol Phys. 2004;60:30-39. 5. Guckenberger M, Meyer J, Vordermark D, Baier K, Wilbert J, Flentje M. Magnitude and clinical relevance of translational and rotational patient setup errors: a cone-beam CT study. Int J Radiat Oncol Biol Phys. 2006;65:934-942. 6. Kaiser A, Schultheiss TE, Wong JY, et al. Pitch, roll, and yaw variations in patient positioning. Int J Radiat Oncol Biol Phys. 2006;66:949-955. 7. Schubert LK, Westerly DC, Tomé WA, et al. A comprehensive assessment by tumor site of patient setup using daily MVCT imaging from more than 3,800 helical tomotherapy treatments. Int J Radiat Oncol Biol Phys. 2009;73:1260-1269. 8. Beltran C, Pegram A, Merchant TE. Dosimetric consequences of rotational errors in radiation therapy of pediatric brain tumor patients. Radiother Oncol. 2012;102:206-209. 9. van Herk M, Remeijer P, Lebesque JV. Inclusion of geometric uncertainties in treatment plan evaluation. Int J Radiat Oncol Biol Phys. 2002;52:1407-1422.