Clinical implementation of respiratory‐gated spot‐scanning proton therapy: An efficiency analysis of active motion management

Abstract Purpose The aim of this work is to describe the clinical implementation of respiratory‐gated spot‐scanning proton therapy (SSPT) for the treatment of thoracic and abdominal moving targets. The experience of our institution is summarized, from initial acceptance and commissioning tests to the development of standard clinical operating procedures for simulation, motion assessment, motion mitigation, treatment planning, and gated SSPT treatment delivery. Materials and methods A custom respiratory gating interface incorporating the Real‐Time Position Management System (RPM, Varian Medical Systems, Inc., Palo Alto, CA, USA) was developed in‐house for our synchrotron‐based delivery system. To assess gating performance, a motion phantom and radiochromic films were used to compare gated vs nongated delivery. Site‐specific treatment planning protocols and conservative motion cutoffs were developed, allowing for free‐breathing (FB), breath‐holding (BH), or phase‐gating (Ph‐G). Room usage efficiency of BH and Ph‐G treatments was retrospectively evaluated using beam delivery data retrieved from our record and verify system and DICOM files from patient‐specific quality assurance (QA) procedures. Results More than 70 patients were treated using active motion management between the launch of our motion mitigation program in October 2015 and the end date of data collection of this study in January 2018. During acceptance procedures, we found that overall system latency is clinically‐suitable for Ph‐G. Regarding room usage efficiency, the average number of energy layers delivered per minute was <10 for Ph‐G, 10‐15 for BH and ≥15 for FB, making Ph‐G the slowest treatment modality. When comparing to continuous delivery measured during pretreatment QA procedures, the median values of BH treatment time were extended from 6.6 to 9.3 min (+48%). Ph‐G treatments were extended from 7.3 to 13.0 min (+82%). Conclusions Active motion management has been crucial to the overall success of our SSPT program. Nevertheless, our conservative approach has come with an efficiency cost that is more noticeable in Ph‐G treatments and should be considered in decision‐making.


| INTRODUCTION
A four-dimensional computed tomography (4D-CT) study obtained at time of treatment simulation, used to approximate motion throughout treatment, is the current gold standard for motion assessment for spot-scanning proton therapy (SSPT). However, using the 4D-CT to define a geometric uncertainty margin for targets and organ at risks (OARs) is typically not adequate for SSPT given that (a) protons are extremely sensitive to heterogeneities in their path 1 (i.e., their water-equivalent depth, WET) 2 and (b) tumor motion in the context of an energetically-and spatially sequential SSPT delivery can result in dosimetric patterns of constructive and destructive interference so called "motion interplay". [3][4][5] WET variation has the potential to produce a significantly different dose distribution for any defined sub-portion of the respiratory cycle. Without motion management, movement of both the target and upstream normal tissues along the beam path (such as the diaphragm) can lead to unacceptable differences in planned vs delivered proton range. 6 The severity of the effect is beam specific. Motion interplay in SSPT can occur both as a consequence of target motion (perpendicular to beam direction) and also as a consequence of WET changes along the beam path associated with motionthat is, volumetric interplay. 7 The overall magnitude of the associated dose perturbation depends on the target motion dynamics, spot delivery/ timing parameters, and their mutual degree of synchronization. A well-demonstrated mitigation technique for the interplay effect is "repainting," also referred to as rescanning. Depending on the capabilities of the delivery system, the associated spot rescanning pattern can be delivered in a volumetric mode (i.e., delivering the whole field N times in succession with a spot weight reduction per volume rescan of 1/N) or each layer can be fully or partially rescanned in succession, the latter having been developed with numerous variants such as iso-layered rescanning, 8,9 scaled rescanning, 8 and the recently developed evenly spread spot-adapted rescanning. 10 An alternative basic/simple mitigation strategy for the interplay effect that has been shown to be effective is the use of larger pristine proton spots, 3 for example, through use of a range shifter if available. Other related motion management strategies that can mitigate interplay include: aligning the preferential directionality of spot scanning with the principle (cranial-caudal) motion direction; the use of fractionation 11 ; and reduction of target motion through actively gated beam delivery or some form of mechanical restrictions such as compression. 12,13 Lastly, not all SSPT optimization approaches are equivalent in terms of motion robustness. Two adopted frameworks are single-field optimization (SFO) and multi-field optimization (MFO). Per-field dose distributions derived from Multi-field Optimization (MFO) tend to be more heterogeneous than SFO fields and are, therefore, more sensitive to anatomical variation, inclusive of motion-related variation.
For these reasons, target motion in the context of SSPT requires a mitigation strategy. At our institution, target motion is managed with a potential combination of techniques, chosen on the basis of target motion amplitude and treatment site. We present a summary of our experience combining motion-mitigation and SSPT. The PTCOG Thoracic and Lymphoma Subcommittee published initial guidelines 14 on implementing active motion management, the launch of our program, however, precedes the release date of that report, in addition, this manuscript presents an efficiency study on patient data.

2.A.2. | Gating interface
Working in collaboration with Mayo Clinic's Division of Engineering, a custom "Respiratory Interface" was designed and fabricated to provide compatibility between the RPM (Varian Medical Systems, Inc.) and the Hitachi Synchrotron (Fig. 1).
As part of the implementation of our motion mitigation program, a separate "rescanning" machine based on the iso-layered repainting (also known as a "Max-MU" threshold approach) was commissioned in Eclipse. Max-MU repainting was achieved by simply setting the maximal MU/spot in the planning system to a value smaller than the actual deliverable maximum MU, in our case MU MAX = 0.005, (only 2.5× our minimum deliverable MU of 0.002).

Acceptance testing
Our institution and the manufacturer agreed on a 200 and 0.5 ms proton beam delivery latency specification for beam-on and beamoff respectively. The acceptance testing performed involved generation of TTL gating logic signals with a vendor-provided device, with simultaneous monitoring of these logic signals in relation to synchrotron delivery signals (e.g., gantry room bending magnet) during delivery of treatment plans sent via ARIA in clinical/DICOM mode (with a tag in the DICOM metadata indicating that the given plan requires an external gating signal).

Commissioning
For bench testing purposes, a plan was designed to deliver a simpli-  F I G . 1. Respiratory gating interface system diagram. The interface provides health logic (i.e., ready) signals to the real-time position management system (RPM) software via the Varian "Gating Switch Box," and in turn, our interface forwards gating (beam-on/beam-off) logic signals it receives from the RPM (via the same Gating Switch Box) to the synchrotron delivery/control system. channels) on a high-resolution flatbed scanner and dose-converted using FilmQA Pro [Ashland]).

2.B. | Active motion management program
The following sections describe our clinical decision-making process for patients undergoing proton treatment simulation with abdominal or thoracic moving tumors.

2.B.1 | Simulation procedure
Currently, all motion-managed cases are simulated head-first supine, with 1.5 mm CT slices. A 4D-CT is required, and the data set is used for two purposes: as a tool for motion evaluation and to create a 4D-average representation for treatment planning. Prior to 4D-CT acquisition, the respiratory trace derived from the RPM (used to generate the 4D-CT binning) is evaluated for reasonable regularity (regarding both cycle-to-cycle amplitude and period) and for having one discernable inspiratory "peak" per respiratory cycle. If an irregular trace is observed or if the motion analysis performed on the reconstructed phases yields displacements larger than 10 mm, a breathhold scan may be additionally performed.
Our institution's protocol for BH simulation is determining the BH gate level based on patient comfort (typically moderate-deep inspiration -mDIBH; with an RPM amplitude gate width of 5 mm); then subsequently acquiring three BH scans to verify that BHs are performed consistently by the patient. The choice of which scan to use is oftentimes based on visual review of the fused BH CTs so as to pick the scan that best approximates the "mid-position" anatomy.
Sometimes we may objectively throw out one or more of the multiple BH scans from consideration because of poor compliance observed with respect to the target BH gate during the scan. (This is necessary because the CT scan is not truly "gated" like the proton therapy machine.) Deep-inspiration BH is not used due to reproducibility concerns. 18

2.B.2 | Motion management guidelines and treatment planning
Our clinic's motion-management generic decision-making scheme is shown in Fig. 2  Breast/Chest-wall lesions are typically treated with two enface oblique fields. A 4.5 cm range shifter is used to allow superficial coverage. Similar to conventional x-ray therapy, breath-hold may be utilized to geometrically displace the heart away from the targeted chest wall volume. Generally, chest wall motion observed is <5 mm (and is often principally along the en-face beam direction); hence, when breath holding is not used for heart sparing, free breathing treatments are preferred.

Deviations from standard practice
For patients with a highly irregular breathing trace and noncompliant for BH, coached shallow breathing may be considered as alternative.
In shallow breathing, the patient is given feedback through the goggles to maintain a small-amplitude breathing trace. A phased-based 4D-CT for residual motion assessment and planning can be acquired and reconstructed with the RPM used in an amplitude mode.
Another source of deviation from our motion management SOP's is the presence of metal hardware in the potential beam path. Geometric beam avoidance of these objects would be a logical preferred strategy, likely allowing for SFO planning. However, in some cases this compromise may excessively hamper the ability to spare critical structures. In these unique scenarios, MFO, combined with more elaborate STVs and robust optimization including inter-field variability likely provides the best compromise.

2.B.3 | Image-guided radiation therapy (IGRT) process
Our standard IGRT methodology is summarized in Fig. 3  As a consequence, for BH cases we may opt to adjust the breathhold gate level to allow both bony and soft-tissue alignment to be consistent with simulation. After isocenter localization at couch angle 270°, verification x-ray imaging may also be performed at the actual treatment couch angles. Optical surface imaging is increasingly utilized in our clinic, primarily as a tool for reduction in frequency of x-ray imaging: e.g., monitoring accuracy of couch rotation, global patient motion, and BH variability.     repainted a large number of times (e.g., >30) and the average F I G . 3. General decision-making scheme for IGRT using our standard stereoscopic kV imaging system. X-Ray imaging always starts at our defined "setup" couch angle (270°); verification x-ray imaging is typically performed at the actual treated couch angles. Issues with localization encountered during treatment may require us to return to couch angle 270°.

3.A | Commissioning test
number of repaints is higher for distal layers (due to higher-MU spots being utilized).

3.B | Efficiency analysis
The average delivery time per field for different sites is quantified in Fig. 7 using energy layers delivered per minute as a proxy for delivery efficiency. The plot compares treatments delivered using BH, Ph-G, and FB + repainting. Due to relatively small sample sizes for these treatment sites and motion management techniques, no substratifications were included in terms of, for example, plan complexity or treatment volume. The general trend observed is that, compared with FB treatments (+repainting), Ph-G treatments take longer to deliver than BH treatments. In the case of lung, the majority of patients included in the study were treated with Free-Breathing (N = 13), and only a few of them included active motion management (N = 3). The inverted trend between BH and Ph-G for this particular site can likely be attributed to patient-specific characteristics such as, target size or patient breathing performance at the time of treatment. Which is to say, given the small numbers we were likely comparing dissimilar plans on average. FB+repainting. Ph-G cases suffer from the highest time penalty due primarily to two factors, namely (a) the addition of repainting to gated delivery: many of our Ph-G plans also are combined with repainting when residual motion within the gate is deemed unacceptable; (b) the effective/realizable duty cycle is often less than the planned/ideal duty cycle, due to variable patient breathing as well as synchrotron control system behavior in the context of unpredictable beam-on gating signals (since beam cannot be held infinitely due to space-charge-related instabilities). In our experience, the effective duty cycle of the system is significantly improved by adjusting RPM's "predictive filter." 19 As shown in Fig. 7, in some cases BH treatments can achieve similar delivery speed performance as FB + repainting; this is explained by the short beam-on time per rescan produced by iso-layered repainting. 20 Other factors which extend treatment times F I G . 6. Number of repaints per spot normalized to the total number of unique spot positions. Energy layers were broken down in two groupsdistal (1/3) and proximal (2/3). The histograms were created using DICOM RTPlan files of treatments were "Max-MU" based rescanning was used.   23 We plan to deploy this device clinically at our facility very soon, recognizing its potential relevance in terms of treatment efficiency, motion management, and general plan robustness. Replanning frequency per treatment site is the subject of an ongoing internal investigation, which is outside the scope of this report. A preliminary and high-level review of these data suggest that replanning is required for approximately 25% of cases involving disease sites relevant for motion management. However, since each patient, each treatment site and each specific motion management strategy pose unique challenges, more in-depth study and granular reporting is needed.

| CONCLUSION S
This work presents a review of the current processes at our institution in the context of motion management. After developing and Ph-G Delivery F I G . 8. Comparison of treatment time and continuous delivery mode (QA) using boxplots as previously described (Fig. 7). The sites in the phase-gated bin for various sites include: liver, esophagus, lung, pancreas, and bile-duct. The results are presented in two formats: (a) contrasting absolute time of each delivery mode and (b) ratios between treatment time and continuous delivery.

CONF LICT OF I NTEREST
The authors have no conflicts of interest to disclose. No external funding was received.