Abstract
Introduction
In recent years, the role of thermoplastic masks and the development of image-guided radiotherapy have made invasive frame-based stereotactic radiosurgery (SRS) no longer the only choice for patients. Moreover, many studies have indicated that frameless, image-guided radiosurgery exhibits comparable overall accuracy to standard frame-based radiosurgery.1–4
During the treatment process, the duration of treatment primarily relies on the quantity of lesions and the intricacy of the treatment plan, ranging from 20 min to over 1 hour. 5,6 At different treatment times, the movement of the patient during treatment is unpredictable, and the patient can only be immobilized with a suitable head support and a thermoplastic mask when immobilized, but this does not eliminate intrafractional motion.7–9 In addition, treatment accuracy is also greatly impacted by the type of fixation mask and head supports used in SRS.10–13 The CyberKnife robotic radiosurgery system utilizes real-time image-guided positioning technology and a flexible robotic arm to correct for patient intrafractional motion.
Improving clinical treatment accuracy has always been our institution's goal. The aim of this study was to investigate whether variations in cranial angles and treatment accuracy during CyberKnife robotic radiosurgery necessitate adjustments to the margins of the planning target volume (PTV).
Material and Methods
Patients
This study has been approved by our institution's institutional review board (No. 202200891B0). Because we used standard treatment data in this noninvasive retrospective analysis, the institutional review board waived the requirement for informed permission from each patient. In this study, data from 66 patients receiving CyberKnife robotic radiosurgery for brain tumors at our clinic were collected (9022 data points); 47 patients underwent SRS and 19 underwent stereotactic radiotherapy. The patients’ age range was 20 to 87 years with a median of 62 years. Among the patients, one had a case of acoustic neuroma, 7 had arteriovenous fistula, 2 had arteriovenous malformation, 10 had carotid-cavernous fistula, 4 had meningioma, 33 had metastases, 1 had pituitary tumor, and 8 had internal acoustic canal tumor. Within a range of 3° to 18°, the mean cranial angle was 10.2°. Group A consisted of 38 patients with cranial angles ≤10°, whereas Group B included 28 patients with cranial angles >10°. The patients received doses ranging from 12 to 40 Gy in 1 to 5 fractions, with a mean dose of 18.9 Gy, prescribed to the 75 % to 93% isodose line. The treatment time per fraction was 41 to 86 min. Table 1 shows the characteristics of the patients in both groups.
Patient Characteristics (N = 66).
Abbreviations: SD, standard deviation; MLC, multileaf collimator; PTV, planning target volume; IAC, internal acoustic canal.
Immobilization and Simulation of Patients
To immobilize each patient, a U-frame thermoplastic mask with a thickness of 2.4 mm and a Silverman headrest (CIVCO Medical Solutions) equipped with MoldCare cushions (ALCARE Co.) were utilized (Figure 1).
The MoldCare cushioned Silverman head support.
The Silverman headrest was made of thin, transparent plastic, and the MoldCare cushion was a soft fabric bag packed with polystyrene beads wrapped in moisture-cured polyurethane resin. When water is sprayed onto the cushions, they become rigid. After 5 to 10 min, the cushion hardens to create a personalized and supportive structure. A GE Discovery RT-16 Slice CT Simulator (GE Healthcare) was used to perform CT simulations on all patients while they were supine, with a 0.625-mm slice thickness.
Cranial Angle Measurement
For all 66 patients, the cranial angle was measured at the Advantage Sim MD workstation using planning CT images. The angle between the glabellomeatal line (GML) and the vertical line at the center of the external auditory meatus is known as the cranial angle (Figure 2).
The cranial angle and vertical line are shown.
CyberKnife® M6 System and Imaging-Guided Tracking Method
The Cyberknife M6 system (Accuray Inc.) is a computer-controlled 6-axis robot that is equipped with a small 6 MV linear accelerator (1000 MU/min LINAC). The imaging system is comprised of 2 ceiling-mounted x-ray sources and matching floor-mounted image detectors.
Using kV x-ray imaging can provide target localization during treatment. During treatment, a tracking system captures 2 real-time images at set intervals (ranging from 15 to 150 s) using real-time imaging. The Multiplan Treatment Planning Software (version 5.1.3; Accuray Inc., Chesapeake Terrace) generates Digitally Reconstructed Radiograph (DRR) images, which are promptly registered with the live images.
When treating intracranial diseases, the 6D skull tracking system can follow the skull's bone structure directly. Target tracking and motion correction can be performed using the contrast and brightness discrepancies between DRR and live pictures (Figure 3).
A demonstration of monitoring skull movements using 6D skull tracking.
The position of the patient on the treatment couch is determined by the coordinates of 3 axes: X, Y, and Z. The X-axis represents the patient's up-down direction (superior [–] to inferior [+]) and their left-right direction (right [–] to left [+]). The Y-axis represents the patient's head-to-toe direction (foot lift [–] to head lift [+]). The Z-axis represents the patient's posterior-to-anterior direction (posterior [–] to anterior [+]).
Roll, pitch, and yaw are 3 concepts that can be used to define rotation. Roll is the left [–] to right [+] rotation around the X-axis. The rotation from foot lift [–] to head lift [+] around the Y-axis is known as pitch. The movement from clockwise [–] to counterclockwise [+] around the Z-axis is known as yaw.
Statistics and Data Analysis
During the study, we analyzed the movement of patients in 6 directions. Out of 66 individuals, 38 did not make any adjustments to the treatment couch. Patient data were collected prior to adjusting the couch to establish a consistent baseline for all the data. A total of 28 patients were included in this study. The data analysis process documented a duration ranging from approximately 0 to 50 min.
Data are collected every 10 min using 6 axes (X, Y, Z, roll, pitch, and yaw) to calculate the average value of all images at each time point. This helps to display the displacement as the total population trend and reduces the analytical bias caused by an extreme value captured by a single image at a given time point. To calculate the 3-dimensional (3D) vector error, the absolute deviations in the 3 translational axes for each patient at each time point are computed. The formula for calculating the 3D vector error is as follows:
To guarantee that 95% of the prescribed dose covers 90% of the treatment plan volume, the PTV margin (MPTV) was calculated according to van Herk et al
14
as follows:
Results
Table 2 shows the gradual changes in translational and rotational errors observed during five 10-min sessions. Significant differences in translational errors were found in the X-axis (P < .02) for Group A and in the Z-axis (P < .03) for Group B.
Progressive Changes of Translational and Rotational Errors Within 0 to 40 min in Different Head Supports Divided into Four 10-min Sessions.
Abbreviations: SD, standard deviation; X, superior–inferior direction; Y, right-left direction; Z, anterior–posterior direction; 3D, 3-dimensional.
P value calculated from the Friedman ANOVA test.
The unstandardized coefficient in the linear regression analysis.
P value calculated from the linear regression analysis.
Neither group had significant differences in the 3D vector and rotational errors. Upon comparison of the 2 groups, it was found that Group A had significantly lower errors in the Y-axis (P < .045) and roll axis (P < .005) compared to Group B. Table 3 presents the results of the systematic error (Σ), random error (σ), and estimated PTV margin (M) for both groups. In the X, Y, and Z axes, Group A had estimated margins of 0.56 mm, 0.46 mm, and 0.47 mm, respectively. In Group B, the estimated PTV margins were found to be 0.62 mm, 0.48 mm, and 0.46 mm in the X, Y, and Z axes, respectively. We found that Group A had a higher random error on the Y-axis than Group B, while Group B had a lower random error on the X-axis.
The Systematic Errors, Random Errors, and Estimated Margins in the Overall 0 to 50 min Time Sessions.
Abbreviation: PTV, planning target volume.
The cumulative error frequency is depicted in Figure 4 for incremental translation axis deviations of 0.1 mm. To ensure accuracy, all deviations are calculated using absolute values, regardless of direction. In order to cover 95% of the intrafractional motion during a 50-min duration, Group A required margins of 0.49 mm on the X-axis, 0.50 mm on the Y-axis, and 0.49 mm on the Z-axis. In addition to this, a margin of 0.69 mm on the 3D vector was necessary. On the other hand, Group B required a margin of 0.60 mm on the X-axis, 0.50 mm on the Y-axis, 0.48 mm on the Z-axis, and 0.79 mm on the 3D vector in order to account for 95% of the intrafractional motion within a 50-min period.
The cumulative frequency of translational (A and C) and rotational deviations in 2 groups (B and D).
In clinical practice, using a 1-mm margin in the evaluation, complete coverage rate can be achieved in all axes in Group A, while only 2.1% of the 3D vector in Group B will exceed 1 mm. The reporting of this study conforms to STROBE guidelines. 15
Discussion
This was a retrospective study of frameless radiosurgery using the CyberKnife system. The mean cranial angle in this study was 10.2°. We used the closest integer to median as the point of data partitioning, primarily to minimize the influence of potential outliers or extreme values. The dataset was divided into 2 groups based on their values relative to the median: one group with values above the median and another with values below it. The median cranial angle in this study was 10°. The selection of a 10-degree threshold for categorizing patients based on cranial angle was primarily guided by clinical observations and initial data analysis. Nevertheless, we acknowledge the potential existence of more optimal cutoff values. To ascertain the threshold of an optimal cutoff value, a comprehensive analysis encompassing a broader range of cranial angles is warranted. This involves evaluating intrafractional stability across various angles to pinpoint the juncture at which stability undergoes significant changes. Such an analysis would yield a more nuanced understanding of the relationship between cranial angle and stability, enabling the development of more tailored patient positioning strategies. Nevertheless, given the significance of the results at a 10-degree threshold, this finding can readily be implemented in clinical practice for ease of recall. Further exploration of this aspect employing a data-driven approach should be designed to establish a more optimal cutoff value for t cranial angles.
We analyzed a total of 66 patients, primarily comparing changes in different cranial angles and intrafractional skull motion. Initially, the duration for each treatment fraction was observed to range from 41 to 86 min. The selection of five 10-min intervals for data collection sessions (totaling 50 min) was driven by the imperative to strike a judicious balance between comprehensive data acquisition and practical considerations. The data were systematically gathered at 10-min intervals, facilitating a meticulous examination of patient motion throughout a significant segment of the treatment duration. This temporal framework was chosen to afford a comprehensive delineation of trends in patient movement, ensuring sufficient granularity without unduly extending the observation period beyond the bounds of a conventional treatment session.
Notable changes in translation errors were observed in Group A along with the X-axis (P < .02) and in Group B along with the Z-axis (P < .03) when comparing treatment time and displacement. In the comparison of the 2 groups, it was evident that Group A exhibited significant differences in the Y-axis (P < .045) and the roll axis (P < .005) compared to Group B.
The impact of cranial angle on intrafractional stability is intricate, involving a nuanced interplay of biomechanical and patient-specific factors. This angle significantly influences head positioning and support during treatment, thereby affecting the patient's capacity to maintain a consistent position. A steeper angle may induce heightened pressure in specific areas, potentially causing discomfort and unintended movement. Conversely, a more neutral angle can enhance comfort and stability. Our analysis indicates that patients with a cranial angle of ≤10° exhibit superior stability, possibly attributable to a more natural alignment, reducing strain on the neck and head and minimizing involuntary movement. These findings underscore the potential significance of optimizing cranial angle during positioning as a pivotal factor in enhancing treatment accuracy by mitigating intrafractional motion. Subsequent investigations should delve deeper into the biomechanical dimensions of cranial angle to provide a more comprehensive understanding. We calculated the coverage rate based on previously suggested margin values. Group A had a coverage rate of 97.69% along with the X-axis, 94.02% along with the Y-axis, and 95.81% along with the Z-axis, while Group B had coverage rates of 96.4%, 94.42%, and 93.54% along with the 3 axes, respectively. These results are in close approximation to the 95% coverage calculated using all the data.
Previous research has compared various head support methods in frameless radiosurgery. 12 Tryggestad et al evaluated 4 thermoplastic mask-based immobilization systems for intracranial radiotherapy using cone-beam CT. 16 Studies have shown that different frameless thermoplastic mask systems with image-guided technology require PTV margins of just 1 mm, as confirmed in our study. In clinical practice, a 1-mm margin resulted in complete coverage in all axes of Group A, while only 2.1% of the 3D vectors in Group B exceeded 1 mm. Ensuring robust fixation during frameless robotic radiosurgery is essential, as random patient displacement during treatment cannot always be predicted in advance and greatly impacts treatment accuracy.
Our noninvasive fixation device mainly consists of a headrest, fixation holder, and thermoplastic mask. In the past, patient comfort was the primary criterion for mask molding, followed by the evaluation of head and neck support. We have been actively exploring ways to enhance treatment quality by improving patient stability and accuracy during treatment. We can use the results of this study as a reference for assessing patient comfort and treatment stability, particularly before molding patient masks, with commonly used skull positioning landmarks such as the GML, orbitomeatal line (OML), infra OML, and others. Determining the optimal patient fixation method for each institution is also crucial.
There are several limitations to our study. First, it is important to note that this study is retrospective in nature, which may introduce potential selection bias in the enrolled patient population. Second, the patient fixation procedures and treatments may not have been consistently performed by the same therapist at the time of simulation. Recognizing potential variations in therapist techniques impacting data, we emphasize adherence to standardized procedures to mitigate variability. While acknowledging inherent patient and therapist differences, our analysis, designed to average these diversities across a larger sample, minimizes their impact. Third, this study primarily collected imaging data within 50 min of treatment time and did not include data collected after adjustments to the treatment couch. Although standard operating procedures were employed, the techniques for adjusting and correcting displacement may not have been consistent across all therapists. As a result, the findings and conclusions of this study may not fully represent patient displacement over extended treatment periods.
Based on our results, fixation factors appear to be critical for intrafractional stability in radiotherapy. A further retrospective study should be conducted to verify the requisite fixation parameters, in order to reduce the unstable conditions during intrafraction and enhance the precision of radiotherapy treatment. Besides, more stringent controls and enhanced therapist training should be considered to standardize procedures, reducing variability and enhancing research reliability. This investigation is aligned with the objective of augmenting the quality of radiotherapy, underscoring the paramount importance of systematically addressing fixation variables. This emphasis is placed with the overarching goal of enhancing not only the overall precision of radiotherapy but also ameliorating treatment outcomes.
Conclusion
Adjusting cranial angle to ≤10° during thermoplastic mask molding provided better or similar intrafractional stability compared to >10°.
Footnotes
Abbreviations
Acknowledgments
Some abstract contents, tables, and figures were previously presented as a digital poster at the 2023 European Society for Radiotherapy and Oncology (ESTRO) Annual Meeting in Vienna, Austria.
Author Contributions
CLK, YYH, and CCH were involved in the conception and design. CLK, YYH, SHT, YRC, and CCH were involved in the analysis and interpretation of the data. CLK, YYH, SHT, and YRC drafted the paper. YRC, YJH, and CCH revised it critically for intellectual content. All authors gave their final approval of the version to be published.
Authors’ Note
Ethics Approval: This study received approval from the Institutional Review Board (IRB) of Chang Gung Medical Foundation (No. 202200891B0). Ethics approval obtained, but patient consent not required. Ethics Statement: The Institutional Review Board of the Chang Gung Medical Foundation authorized this study (No. 202200891B0). Informed consent: The necessity for obtaining informed consent from individual patients was waived by the IRB. This waiver was granted because it is an IRB-approved retrospective study, all patient information was deidentified, and patient consent was not required.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
