If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Patient specific estimation of interplay effects improves ahead of treatment.
•
Investigated a novel software allowing virtual phantom motion during beam delivery.
•
Developed in-house software to generate user defined breathing traces.
•
Physically moved the 3D dose measuring phantom by programmed couch for validation.
•
Investigated interplay for 14 Lung VMAT SABR plans and various breathing traces.
Abstract
Purpose
In lung SABR, interplay between target motion and dynamically changing beam parameters can affect the target coverage. To identify the potential need for motion-management techniques, a comprehensive methodology for pre-treatment estimation of interplay effects has been implemented.
Methods
In conjunction with an alpha-version of VeriSoft and OCTAVIUS 4D (PTW-Freiburg, Germany), a method is presented to calculate a virtual, motion-simulated 3D dose distribution based on measurement data acquired in a stationary phantom and a subsequent correction with time-dependent target-motion patterns. In-house software has been developed to create user-defined motion patterns based on either simplistic or real patient-breathing patterns including the definition of the exact beam starting phase. The approach was validated by programmed couch and phantom motion during beam delivery.
Five different breathing traces with extremely altered beam-on phases (0 % and 50 % respiratory phase) and a superior-inferior motion altitude of 25 mm were used to probe the influence of interplay effects for 14 lung SABR plans. Gamma analysis (2 %/2mm) was used for quantification.
Results
Validation measurements resulted in >98 % pass rates. Regarding the interplay effect evaluation, gamma pass rates of <92 % were observed for sinusoidal breathing patterns with <25 number of breaths per delivery time (NBs) and realistic patterns with <18 NBs.
Conclusion
The potential influence of interplay effects on the target coverage is highly dependent on the patient’s breathing behaviour. The presented moving-platform-free approach can be used for verification of ITV-based treatment plans to identify whether the clinical goals are achievable without explicit use of a respiratory management technique.
Radiotherapy techniques such as intensity modulated radiotherapy (IMRT) and volumetric modulated arc therapy (VMAT) offer a better dose conformity compared to conventional radiotherapy. Multileaf Collimators (MLCs) are the key to altering the beam fluence and delivery of continuous modulated dose to the planned volume. In VMAT, the MLC aperture shape, the dose rate and the gantry rotation speed vary dynamically, improving the precision of external beam therapy.
Relative motion between the treatment apparatus and the target volume makes the VMAT technique susceptible to challenges with potential dosimetric impacts. A clinically applied respiratory motion-management strategy is the internal target volume (ITV) approach. The ITV is derived by means of 4-dimensional computed tomography (4DCT) scans. For SABR, the ITV consists of the full range of target position using the maximum image projection (MIP) and/or a sum of the gross target volumes (GTVs) for each individual breathing phase. The planning target volume (PTV) encloses the ITV with an additional margin (typically 3–5 mm) and is designed to account for other treatment uncertainties (e.g., setup uncertainties, etc.) [
]. A recent study has evaluated CTV to PTV margin for Liver stereotactic body radiotherapy using the ITV approach employing daily CBCT set-up correction [
Although the ITV/PTV encloses the peak-to-peak breathing amplitude of the target volume, the interplay between dynamic MLC and target motion (interplay effects) may lead to under- and over dosage within the PTV or organs at risk (OAR). This has especially been reported for hypofractionated VMAT treatments with high doses delivered in only a few breathing cycles [
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
]. As hypofractionated concepts are increasingly being used in techniques such as stereotactic ablative radiotherapy (SABR), investigating the interplay between dynamic MLCs and target motion on a treatment plan specific basis is significant.
There is contrasting evidence on the dosimetric impact of the interplay effect. The methodology used to simulate motion during measurements in some reports had limitations in the choice of motion trajectories and beam-on phase [
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
], however, the devices normally require a physical motion system that are heavy (e.g., 60 kg), complex, and expensive. Additionally, the attenuation of such motion platforms often has to be accurately modelled in the treatment planning system (TPS), which is time consuming and can add uncertainties.
It has been shown that enabling the beam-on in two different breathing phases results in different mutual movement between the target and the MLC and consequently different doses to the moving target [
]. For a real patient treatment, this makes the extent of interplay effects extremely challenging to predict as the initial breathing phase of the treatment delivery is typically unknown unless gating techniques are implemented [
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
The interplay effect is reduced with double arcs compared with a single arc as the longer time averages out some of the interplay effect dependent dose errors [
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
Investigation of 4D dose in volumetric modulated arc therapy-based stereotactic body radiation therapy: does fractional dose or number of arcs matter?.
]. This different outcome may be affected by the fact that deformable image registration was used for accumulating doses to the target region and although the accuracy of the image registration algorithm was confirmed, a voxel-by-voxel matching within the target volume was not guaranteed [
Investigation of 4D dose in volumetric modulated arc therapy-based stereotactic body radiation therapy: does fractional dose or number of arcs matter?.
]. A number of investigators found that the interplay effect is less pronounced with longer treatment times, larger number of fractions, multiple fields and arcs [
]. The extended beam-on-time and consequently the increase of NBs during irradiation leads to a decrease of dose variation in the target. This is particularly important for single-fraction FFF lung SABR [
Siva S, Bressel M, Mai T, Le H, Vinod S, de Silva H, et al. Stereotactic Ablative Fractionated Radiotherapy Versus Radiosurgery for Oligometastatic Neoplasia to the Lung (SAFRON) II Study Investigators. Single-Fraction vs Multifraction Stereotactic Ablative Body Radiotherapy for Pulmonary Oligometastases (SAFRON II): The Trans Tasman Radiation Oncology Group 13.01 Phase 2 Randomized Clinical Trial. JAMA Oncol. 2021 Oct 1;7(10):1476-85. 10.1001/jamaoncol.2021.2939.
The absence of broad consistency between the results from different studies implies that altered combinations of breathing patterns and machine-related parameters may result in dissimilar impact levels [
]. Availability of easy to use tools and techniques for interplay investigations on measured 3D dose distributions across clinics could provide confidence and better consensus. Some commercial methods are currently available that could be used to assess the interplay effect on delivered target doses with and without moving platforms. The treatment fields don't pass through the Delta4 HexaMotionTM (ScandiDos, Uppsala, Sweden) platform which applies 6D breathing motion to a 3D dose measuring phantom [
]. Hence it does not require complex modelling of motion platform for TPS dose calculations. The MotionSim feature on the ArcCHECKTM (Sun Nuclear, Melbourne, FL) measurement system uses target motion trajectory applied to the measured dose on a static phantom to create a dose distribution as if the voxels were virtually moving during measurement [
]. To our knowledge this is the only commercial system available for 3D dose verification allowing virtual simulation of target motion but it involves the TPS plan and dose distribution to calculate the measured dose distribution. Currently, there are no reports on the effect of beam-on phase on measured 3D dose distributions other than the studies employing these two phantoms.
The aim of this study was to investigate a comprehensive and user-friendly methodology for patient specific 3D treatment plan verification (i) incorporating virtual intra-fractional motion (ii) with control over the motion trajectory and beam-on phase and (iii) independent of the TPS dose calculation.
2. Materials and methods
2.1 General approach
The high sensitivity of VMAT treatments to respiratory motions requires high confidence in the accuracy of the planning and delivery process. The methodology described in the subsequent sections can be summarized as follows. The patient plans were delivered to a stationary OCTAVIUS 4D phantom (PTW-Freiburg, Germany) (OCT4D). Breathing patterns were used to generate time-dependent target motion coordinates. A novel 3D dose reconstruction algorithm in VeriSoft (PTW-Freiburg, Germany) calculates the motion-simulated 3D phantom dose distributions with different beam starting phases set to either end-inhalation (0 % breathing phase) or end-exhalation (50 % breathing phase). 3D Gamma analysis (2 %2mm criterion) was used to compare the out-of-phase phantom dose distributions. In the clinical scenarios with fractionated treatment and with multiple arcs per plan, the starting phases of the fields will most likely be spread out. These comparisons between completely out of phase plan delivery for a single fraction allowed estimation of the influence of the interplay effect for an extreme scenario.
2.2 Plan characteristics
For the present study, fourteen VMAT-SABR plans of patients with primary lung targets were selected. The patients underwent RapidArc® treatment using two 6 MV FFF arcs on a Varian TrueBeam™ at a dose rate of 1400 MU/min using fractionation scheme of either 8 × 7.5 Gy, 3 × 18 Gy, or 5 × 11 Gy.
The plan parameters are summarized in Table 1. Beam-on-time is the time interval from the start of the beam at the beginning of first arc to the moment that the beam is switched off at the end of second arc, including the time required for changing the collimator angle between the two partial arcs. Depending on the dose rate and the total number of MU the beam-on-time varied between 89 s and 225 s.
Table 1Treatment plan parameters of the fourteen studied patients (PTV stands for planning target volume).
All plans were re-calculated on a cylinder representing the OCT4D in the Eclipse TPS (Varian Medical Systems, Palo Alto, USA). The first five cases were used to validate the presented approach (see section 2.5). Afterwards, all 14 lung SABR plans were delivered in order to evaluate the dose inhomogeneity across the target in the presence of intrafraction motions by means of the validated approach.
2.3 Dose measurement system
All dose measurements were performed with the treatment plan verification system OCT4D in conjunction with the patient plan verification software VeriSoft V6.2 (PTW-Freiburg, Germany), used to collect the measured data [
Allgaier B, Schüle E, Würfel J. Dose reconstruction in the OCTAVIUS 4D phantom and in the patient without using dose information from the TPS. PTW-Freiburg, White Paper, 2013.
The OCT4D comprises a 2D array detector positioned in the centre of a motorized, cylindrical polystyrene phantom with 32 cm diameter, 34.3 cm length, and an angular range of ±360° [
]. Using an inclinometer the phantom rotation is synchronized with the gantry keeping the 2D array detector always perpendicular to the beam during the treatment plan delivery. The 2D dose is measured time and angle dependent.
In this study the OCTAVIUS Detector 1000SRS (PTW-Freiburg, Germany) array detector was positioned inside the cylindrical phantom. The array detector is comprised of 977 liquid-filled ionization chambers spread across an active area of 11 × 11 cm2. The detector spacing varies throughout the sensitive area from 2.5 mm in the central region of 5.5 × 5.5 cm2 and 5.0 mm in the surrounding area. Dose rates ranging from 0.2 to 36.0 Gy/min can be measured with a resolution of 0.1 mGy/min. Liquid-filled ionisation chambers present almost water equivalent spectral response and high signal to noise ratio compared to air-filled chambers. The measurement interval was 200 ms and acquisitions were performed without resetting the detector system between the arcs for each plan (dose integration time was 900 s). Influences due to dose rate and dose-per-pulse were to be expected [
Characterisation of a two-dimensional liquid-filled ion chamber detector array using flattened and unflattened beams for small fields, small MUs and high dose-rates.
]. In order to minimize these saturation effects, the cross-calibration procedure was performed at field sizes and dose rates close to the treatment conditions of the patient plans.
2.4 Dose reconstruction and evaluation principles
An alpha version of VeriSoft was utilized for data processing and evaluation. It includes an enhanced dose reconstruction algorithm that allows the application of arbitrary motion patterns to the measurement data acquired by a stationary OCT4D and the calculation of a 3D phantom dose distribution as if the phantom had moved during the measurement. To receive a 3D dose distribution in such a virtually moving phantom the algorithm processes the time- and angle-dependent measurement data using in-water measured percentage depth dose curves (source to surface distance (SSD) 85 cm, field sizes ranging from 4x4 cm2 to 26x26 cm2) and a time-dependent phantom motion information (shift/translation file in XML format). The motion information is indispensable as it accounts for the spatial translation of the dose information and corrects for the changing SSDs and the geometry the phantom would undergo during movement in all spatial directions. In the present study only the motion simulation along the longitudinal direction was validated and used to investigate interplay effects. All phantom dose distributions were reconstructed with a voxel resolution of 2.5 mm.
The dose distribution comparisons were evaluated using gamma analysis which is a standard procedure for planned dose verification [
]. 3D gamma analysis (2 %/2mm criterion) was performed with global normalisation to the maximum dose of the reference dose distribution. Following the recommendations of TG-218 for 3D matrix comparison [
], all analyses were performed for a criterion of 2 % dose difference and 2 mm relative distance-to-agreement with dose suppression thresholds of 10 % and 50 %.
2.5 In-house developed software
In‐house software was developed using MATLAB (MATLAB, Natick, Massachusetts: The MathWorks Inc.) programming platform to generate the translation file with explicit structure, compatible with the format required by the VeriSoft alpha version. A raw OCT4D measurement file and a breathing motion trace are needed as the input. For each interval in the measurement file, an entry in the shift/translation file is used to define the respective translation and off-axis values [
]. The target motion trace can either be derived from a real patient target motion pattern or a sinusoidal curve – simulating the patient breathing patterns – with different periods and amplitudes. More important for the interplay effect studies, the software gives the possibility to define specific points within the motion trace as the beam starting phase. To synchronize the beam starting point of the measured data with the desired respiratory phase a dose threshold trigger was implemented.
In this study we assumed that the dominant target displacement is in the superior inferior (SI) direction [
]. Therefore, the translation files carry exclusively longitudinal motion patterns derived from target motion patterns.
2.6 Approach validation
The novel algorithm embedded in the VeriSoft alpha version together with the translation file from the in-house developed software allows any motion pattern to be applied to the measured dose distributions acquired by a stationary OCT4D and reconstructs the dose as if the phantom was moving with a predefined beam starting phase. To validate this for motion along the longitudinal direction, the Varian TrueBeam Developer Mode (DM) was used, which allows continuous movement of the couch, following selected motion patterns during the beam delivery. The couch oscillation as well as the beam starting time are specified through XML scripts readable by the Varian DM.
The accuracy of the Varian DM couch motion was verified by acquiring kV images of a ball bearing (BB) block placed on the couch (initially positioned at the isocentre). Longitudinal motion patterns were created in DM for 7 lung SABR plans and were delivered simultaneously with continuous on-board kV imaging (acquisition rate 15 frames per second). Five plans were delivered with 4 s period and the other two with 20 s period. Images were analysed with an in-house developed MATLAB software to identify the BB and calculate the coordinates of its centre. The software also read the acquisition time stamp of each image from the DICOM header. The position coordinates of the BB’s centre were plotted against time and fitted to the planned motion curve and compared to the actual amplitude and period of the DM’s moving couch. The kV image analysis showed that the oscillation period and amplitude of the DM couch motion is consistent with the programmed motion trace within ±0.1 s and ±0.7 mm, respectively.
For the validation measurements, five patient plans (case 1–5, Table 1) were delivered to a stationary OCT4D system. The same plans have been delivered again with the treatment couch used to drive the OCT4D system, longitudinally, following a sinusoidal breathing trace (25 mm amplitude) using DM. These measurements were performed for two different periods of 4 s and 20 s with beam starting at 0 % respiratory phase (end-inhalation) and 50 % respiratory phase (end-exhalation). All measurements were performed three times in order to check their reproducibility. Fig. 1 presents a schematic view of the measurement setup. Using the novel dose reconstruction approach the measurements with the stationary OCT4D where reconstructed, receiving a motion-simulated phantom dose distribution. The measurements acquired in the physically moving OCT4D where reconstructed classically because they already included the motion influence. Same period, amplitude and beam starting phase were used for both the virtually and physically moving phantoms in order to quantify the difference between the dose matrices. A 3D Gamma analysis of the dose distributions from the motion-simulated and physically moved OCT4D measurements was performed.
Fig. 1Schematic view of the experimental setup for validation measurements. The OCTAVIUS 4D phantom is positioned on the treatment couch. The couch and phantom movement is controlled via the Varian TrueBeam Developer Mode according to pre-defined oscillation curves.
For the interplay effect evaluation using our validated approach, all VMAT patient plans in Table 1 have been delivered on a stationary OCT4D system (i.e. not moving in the SI direction). The MATLAB tool was used to create longitudinal motion patterns derived from simplistic sinusoidal breathing patterns with 0 % and 50 % beam starting phases. For each patient, the breathing periods were chosen such to have the NBs per delivery time ranging from 5 to 50.
3D gamma analysis (2 %/2mm criterion) between dose distributions measured with the beam starting at different respiratory phases can give an estimation of the interplay effect. To investigate the extreme scenario, gamma pass rates have been calculated for the motion-simulated doses measured with beam starting at 0 % respiratory phase compared to the motion-simulated doses measured with the beam starting at 50 %. Global normalisation to the maximum measured dose with beam starting at 0 % respiration phase was applied.
Additionally, the comparison was also made for more realistic breathing patterns which were available from QUASAR™ breathing simulator system (Modus Medical Devices, Ontario, Canada). Four traces representing irregular, typical, typical-fast and typical-slow respiration were selected to generate translation files with target motion in the SI direction using the in-house developed MATLAB software. For comparison purposes, the amplitudes have been rescaled to 25 mm. The resulting longitudinal target motion traces are shown in Fig. 2 and the associated breathing periods are listed in Table 2. Virtually moved 3D dose distributions were calculated for each motion scenario and subsequently the gamma passing rate between the dose distributions with the beam starting at 0 % and 50 % respiratory phases were calculated.
Fig. 2Longitudinal target motion traces derived from the QUASARTM breathing simulator system with irregular, typical, typical-fast, and typical-slow frequencies.
A comprehensive approach for investigating the impact of interplay between target motion and dynamically changing beam parameters has been introduced and validated by the Varian TrueBeam developer mode, as explained in section 2.6. The results from the validation measurements for the five evaluated VMAT patient plans are depicted in Fig. 3. The error bars represent the standard deviation calculated from the three measurement sets. For all cases the 2 %/2 mm gamma criteria are satisfied in more than 98 % of the total points for both dose suppression thresholds of 10 % and 50 %. This indicates agreement between the DM motion versus our approach and, thereby, confirming the feasibility of the methodology based on the stationary measurement for interplay effect evaluation.
Fig. 3Gamma passing rate (2 %/2mm, 10 % and 50 % dose threshold) of the comparison between motion-simulated phantom dose distributions and phantom dose distributions affected by physical phantom movement. Two periods (4 s and 20 s) and two beam starting phases (0 % and 50 %) upon a sinusoidal motion trace with 25 mm amplitude have been considered.
For the interplay effect investigation 14 patient plans have been measured with the stationary OCT4D phantom. Initially the dose was reconstructed under simplistic sinusoidal breathing conditions. The comparison between the dose distributions with beam starting phases at end-inhalation and end-exhalation shows a clear dependency of the gamma passing rate with respect to the NBs per beam extent. As can be seen in Fig. 4, for each of the 14 cases, the gamma passing rate decreases as the number of breathing cycles reduces. This trend is more pronounced at the matrix comparisons using a dose suppression threshold of 10 % compared to 50 % threshold. More than 92 % of points pass the 2 %/2mm criteria for more than 25 NBs using both the 10 % and 50 % thresholds.
Fig. 4Gamma passing rate with 2 %/2 mm tolerance versus number of breaths per delivery time for 14 plans. Left: Dose suppression threshold of 10 % and Right: Dose suppression threshold of 50 %. Gamma analysis is based on the comparison between motion-simulated phantom dose distributions with beam starting at end-inhalation (0 %) and end-exhalation (50 %).
To investigate the effect of realistic abnormal and normal breathing variability on the dose distribution across the target, a comparison of measurements affected by irregular, typical, typical-fast, and typical-slow respiratory patterns was carried out. The number of cycles per beam delivery time for all 14 plans are calculated and presented in Fig. 5, graphically. Fig. 6 shows the gamma evaluation between the 3D dose distributions of the motion-simulated measurements with beam starting phase at 0 % versus 50 %. In all treatment plans the gamma passing rate for the typical slow breathing condition is lower compared to the other breathing patterns. The difference between gamma results of the irregular, typical and typical-fast breathing patterns are small (passing rate difference <5 %). The average gamma passing rate for dose suppression threshold of 10 % over all the 14 treatment plans is 79.8 % (±11.7 % SD), 98.6 % (±1.3 % SD), 98.7 % (±1.0 % SD), and 99.1 % (±0.7 % SD) for typical-slow, irregular, typical, and typical-fast breathing traces, respectively.
Fig. 5Number of breathing cycles per delivery time (NB) for all 14 investigated treatment plans applying typical-slow, irregular, typical, and typical-fast breathing patterns. Boxes illustrate median and interquartile range of the NB and whiskers represent the spread of the evaluated data.
Fig. 6Gamma analysis (2 %/2mm) of the motion-simulated dose distributions using realistic breathing patterns with beam starting at phase 0 % compared to beam starting at phase 50 %. Left: Dose suppression threshold of 10 % and Right: Dose suppression threshold of 50 %. Boxes illustrate median and interquartile range of the gamma pass rates for all 14 plans. Whiskers represent the spread of the data.
For VMAT plan no. 1, Fig. 7 shows the comparison between phase 0 % and phase 50 % using two different breathing patterns: typical-fast and typical-slow. The results indicate that failed points are more dominant in case of the typical-slow breathing pattern. Furthermore, in both cases, the failing points are located especially in peripheral regions. The interplay investigation results are for comparison between single plan deliveries where the breathing motions is completely out of phase. This is an extreme scenario to probe interplay effects. Clinically comparisons will be between TPS dose calculation for a static phantom and motion simulated measured dose distribution calculated for chosen starting phases for individual fields. Spreading of beam starting phases of individual fields, fractionated treatment and the irregular breathing motion in real patients will have an averaging effect on the dose distribution [
Siva S, Bressel M, Mai T, Le H, Vinod S, de Silva H, et al. Stereotactic Ablative Fractionated Radiotherapy Versus Radiosurgery for Oligometastatic Neoplasia to the Lung (SAFRON) II Study Investigators. Single-Fraction vs Multifraction Stereotactic Ablative Body Radiotherapy for Pulmonary Oligometastases (SAFRON II): The Trans Tasman Radiation Oncology Group 13.01 Phase 2 Randomized Clinical Trial. JAMA Oncol. 2021 Oct 1;7(10):1476-85. 10.1001/jamaoncol.2021.2939.
Fig. 7Example failed-points in comparison between motion-simulated 3D dose distributions (end-inhalation vs end-exhalation) for plan no. 1 using typical-fast breathing pattern (a) and typical-slow breathing pattern (b). The red colour represents points failing the 2 %/2mm gamma criteria due to overdosage while the blue colour represents points failing due to underdosage. (c) and (d) are profile comparisons of the dose distributions corresponding to (a) and (b).
In the present study, a comprehensive scheme for individual pre-treatment estimation of the interplay effect – based on patient breathing pattern, beam starting phase, and dose distribution measured in a static phantom – was introduced and validated. Subsequently, the presented methodology was used to assess the interplay effect for 14 lung SABR treatment plans. In this approach, the treatment plan is delivered to the static OCT4D phantom; a simplistic or realistic breathing pattern is used to generate translation coordinates; 3D phantom dose distributions were calculated with the beam starting phase set to end-inhalation (0 %) and end-exhalation (50 %); and finally a gamma analysis is applied to compare motion-simulated dose distribution measured with beam starting at 0 % against 50 % in order to estimate the highest potential breathing motion induced interplay effect. With the present methodology, investigation of up to 6D target motion (3D translation, 3D rotation) is possible. We have validated the approach in 1D for simplicity and consistency with other studies [
] assuming that the dominant target displacement was in the superior inferior direction. Extreme beam starting phases of 0 % and 50 % with periods of 4 s and 20 s were probed. The unrealistic 20 s motion period was used to highlight the extent of interplay and its effect on target dose coverage.
Many studies have investigated the intra‐fractional motion induced interplay effect and it has been shown that there are numbers of patient- and machine-related parameters that influence the impact of interplay effects [
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
]. These parameters include, but are not limited to, prescribed dose, dose per fraction, number of fractions, dose rate, target size and location, target motion pattern (period, amplitude, degree of freedom, phase shift, level of irregularity), beam starting phase (target location at the start of dose delivery), plan complexity, MLC motion speed, size and shape of MLC openings, and number/length of arcs [
]. Due to the wide range of impacting parameters, in each study only a part of those have been considered. Comparison among different studies depends on the similarities between the examined patients/plans parameters. Even for similar parameters, potential abnormalities in the target motion patterns makes the comparisons hard [
Validation results specify a good agreement between the DM motion and the VeriSoft-based method (see Fig. 3). In the DM motion plans, secondary jaws were kept static to avoid any potential beam delays that could happen with jaw tracking technique. Including the jaw tracking technique to fully reproduce a SABR treatment scenario requires further investigation.
By following our validated approach, for 14 lung SABR plans with FFF beams, it was shown that the maximum potential interplay effect is significant when the breathing period is large and the ratio between the beam-on time and the breathing period is small (see Fig. 4, Fig. 5). This is consistent with other studies, demonstrating increased dose discrepancies within the target for longer period times [
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
]. Fig. 4 indicates how the patient’s number of breaths alters the interplay between moving target and the dynamic MLCs. All the 14 patients follow a similar trend and the dose deviations due to interplay effects will, to a greater extent, average out if there are more respiratory cycles while the plan is delivered. In other words, the higher the number of respiration cycles (per total beam time) the higher the gamma passing. Using sinusoidal breathing patterns it was revealed that for over 25 NBs the interplay effect is not significant. Real patient motion is, however, not as regular as this.
To obtain more realistic breathing motions trajectories, irregular, typical, typical-fast, and typical-slow motion patterns were created. The number of cycles per beam delivery time are presented graphically in Fig. 5, which helps to envisage the severity level of consequent interplay effect. Due to the longer period and fewer NBs, the typical-slow respiratory pattern is expected to most likely result in a higher dose discrepancy within the target. Fig. 6 shows for all patients, minimum gamma passing rate corresponds to the typical-slow breathing trace – as expected. The irregular pattern does not show much difference with respect to the typical ones. This is because the chosen irregular pattern features a short mean period and as a result high number of cycles per beam time (see Table 2 and Fig. 5). Based on the results from the realistic target motions, one can state that the maximum interplay effect is negligible for NBs >18, which is consistent with the findings from Leste et al [
]. The percent of failed gamma voxels are more for the 10 % threshold compared to the 50 % threshold, which indicates the effects of interplay to occur especially in peripheral regions. This can also be inferred from Fig. 4 where the gamma pass rates are generally lower for the 10 % threshold compared to the 50 % threshold. Furthermore, the gamma pass rate was plotted against target volume and no significant correlations were found, which is consistent with what has been reported by Edvardsson et al [
It is important to note that the NBs threshold, which is discussed by different authors, are only applicable for that specific evaluated conditions (dose rate, MLC motion, dose per fraction, the number and length of arcs, etc.) and more data may be required to determine a patient’s suitability for lung VMAT-SABR [
]. Even though the investigated dose measurement system can measure dose distributions simulating 3D motion, only motion in the longitudinal direction has been validated and used to analyse interplay effects. The novel approach presented here allows investigation of the influence of target motion patterns with varying NBs for the same patient plan, which can potentially be used to define motion management strategies. The proposed methodology as well as the validation technique may open new possibilities to define a standardized procedure for pre-treatment verification of critical high-dose hypofractionated treatments (such as lung SABR plans) without the need for bulky motion platforms. Further investigations are ongoing in this respect.
In clinical practice, the target motion trace could be obtained from 4DCT image sets. Target volumes could be outlined on each individual phase image set and the trajectory of the centroids of these volumes used as a representation of the target motion. Coupled with time-resolved breathing patterns obtained for example by the RPM System (Varian, Palo Alto), a time-dependent target translation file could be generated.
The methodology presented could also be improved for clinical use by integrating the translation file generation algorithm into the measured dose reconstruction software with the actual patient respiratory pattern and the target motion pattern as the input. The motion simulated measured dose distributions would be compared against the convolved TPS calculated dose distribution in the phantom. Along with gamma analysis, phantom measured dose volume parameters of a simulated moving GTV could provide a better insight into the interplay effects. Using the methodology discussed here, a study is ongoing to evaluate dose perturbation within the moving GTV for lung SABR plans.
4.1 Conclusion
Our study confirms that the potential influence of respiration-induced interplay effects on the lung target treatments are highly dependent on the patient, plan, and machine specific parameters. The actual dose delivered to the moving target will depend on the nature of the motion as well as the individual breathing conditions, even for optimized margins encompassing the entire target excursion. The large patient-to-patient variation in the respiratory cycles and the varying dynamic MLC sequences for different plans prevent a clear and universally valid conclusion. Therefore, patient specific pre-treatment verification including an evaluation of the dosimetric impact of relative motion between the MLC leaves and target is a highly valuable tool using frameworks such as that presented.
Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
Authors would like to thank Bernd Allgaier (PTW-Freiburg, Germany) for the support during this study and the Friends of the Cancer Centre organization (Northern Ireland Cancer Centre, Belfast) for funding Dr Mohammad Varasteh.
References
Zeng L.
Wang X.
Zhou J.
Gong P.
Wang X.
Wu X.
et al.
Analysis of the amplitude changes and baseline shifts of respiratory motion using intra-fractional CBCT in liver stereotactic body radiation therapy.
Dosimetric impact of the interplay effect during stereotactic lung radiation therapy delivery using flattening filter-free beams and volumetric modulated arc therapy.
Investigation of 4D dose in volumetric modulated arc therapy-based stereotactic body radiation therapy: does fractional dose or number of arcs matter?.
Siva S, Bressel M, Mai T, Le H, Vinod S, de Silva H, et al. Stereotactic Ablative Fractionated Radiotherapy Versus Radiosurgery for Oligometastatic Neoplasia to the Lung (SAFRON) II Study Investigators. Single-Fraction vs Multifraction Stereotactic Ablative Body Radiotherapy for Pulmonary Oligometastases (SAFRON II): The Trans Tasman Radiation Oncology Group 13.01 Phase 2 Randomized Clinical Trial. JAMA Oncol. 2021 Oct 1;7(10):1476-85. 10.1001/jamaoncol.2021.2939.
Allgaier B, Schüle E, Würfel J. Dose reconstruction in the OCTAVIUS 4D phantom and in the patient without using dose information from the TPS. PTW-Freiburg, White Paper, 2013.
Characterisation of a two-dimensional liquid-filled ion chamber detector array using flattened and unflattened beams for small fields, small MUs and high dose-rates.
02096-8&id=gr1.jpg){kind=link}
02096-8&id=gr2.jpg){kind=link}
02096-8&id=gr3.jpg){kind=link}
02096-8&id=gr4.jpg){kind=link}
02096-8&id=gr5.jpg){kind=link}
02096-8&id=gr6.jpg){kind=link}
02096-8&id=gr7.jpg){kind=link}