Adapting VMAT plans optimized for an HD120 MLC for delivery with a Millennium MLC

Abstract Linac downtime invariably impacts delivery of patients' scheduled treatments. Transferring a patient's treatment to an available linac is a common practice. Transferring a Volumetric Modulated Arc Therapy (VMAT) plan from a linac equipped with a standard‐definition MLC to one equipped with a higher definition MLC is practical and routine in clinics with multiple MLC‐equipped linacs. However, the reverse transfer presents a challenge because the high‐definition MLC aperture shapes must be adapted for delivery with the lower definition device. We have developed an efficient method to adapt VMAT plans originally designed for a high‐definition MLC to a standard‐definition MLC. We present the dosimetric results of our adaptation method for head‐and‐neck, brain, lung, and prostate VMAT plans. The delivery of the adapted plans was verified using standard phantom measurements.


| INTRODUCTION
Ideally, a patient's entire treatment course is delivered in sequential daily fractions (weekends excepted) on the linear accelerator (linac) for which the radiation plan was designed; however, linac malfunction occasionally causes downtime, and may require the transfer of patient plans between linacs with differing multileaf collimator (MLC) designs.
In our department, one linac is equipped with a Varian high-definition MLC (HD120), while multiple linacs have standard-definition Millennium MLCs (Millennium 120). All of these linacs are commissioned to deliver VMAT treatments. In addition to differences in leaf widths, the MLCs differ in material composition and geometric properties (leaf thickness, tongue-and-groove design, and leaf-end curvature 1,2 ), which would lead to dosimetric differences between VMAT plans.
Nevertheless, the single-fraction transfer of a treatment may be desirable to maintain the prescribed fractionation schedule. Fractionation plays a sensitive and demonstrable role in patient outcomes for headand-neck treatments, 3,4 and likely for other treatment sites.
A previous study 5 investigated transferring VMAT patients between linacs using a Pinnacle-based treatment planning system (TPS). Transferring a VMAT plan was not possible without reoptimization. With our method, a high-definition VMAT plan (from a high-definition linac, HDL) could be adapted to a standard-definition linac (SDL) by creating a new plan (the "adapted" plan) using the DICOM plan file. The resulting adapted plan is analyzed within the TPS (Varian Eclipse 13.6) and exported to our record-and-verify system for treatment delivery. Our goal with the plan adaptation system was to study a simple method that would allow for transferring of the VMAT patient from an HDL to an SDL.

| ME TH ODS AND MATERIALS
A MATLAB routine (MATLAB, R2007a, The MathWorks Inc., Natick, MA, USA) was written to copy VMAT delivery data (leaf aperture shapes, field weights and control point MU indices) contained in the planning DICOM file and place it into a prepared template. The template is a copy of the original plan which will act as a container for the transfer linac information.

2.A | MLCs and DICOM file creation
The Eclipse treatment planning system, version 13.6 (Varian Inc., Palo Alto, CA, USA), was used for this study. A copy of the original HDL plan was created in Eclipse and the linac was changed from the HDL to the SDL in the plan properties. This process automatically removes the dynamic MLC positions defining the control point apertures stored in the plan. A new MLC object was then added to each field (VMAT arc) in the SDL plan, and the number of control points was set by the user to match the original number found in the HDL fields. The TPS template method provides the beam modeling data required to calculate dose correctly for the SDL.

2.A.2 | Behavior of MLC leaf velocities under averaging
At the time that the original VMAT plan was created by the TPS, the optimization step included applying an MLC leaf-speed constraint to the leaves that defined potential aperture sets. Let an individual leaf's maximum travel speed between adjacent control points be C (leaf-speed constraint is the same for both linacs). A simple calculation shows that this constraint will also be satisfied by any leaf in the adapted plan (i.e., after the averaging process). Let x 1 and x 2 be the positions of two adjacent HDL leaves that are averaged to provide, x, the position of an SDL leaf: The average velocity of this leaf is then, Since each HDL leaf obeys the maximum-speed constraint, and, by the triangle inequality, Therefore, which was the original SDL leaf maximum-speed constraint.

3.A | Changes to the dose to structures
The dose to the patient structures in the adapted SDL plans was calculated in the TPS using the Eclipse AAA algorithm and compared via DVH analysis to the corresponding structure doses in the original 1. An illustration comparing the relative leaf widths and locations of the leaves between the HD120 MLC and the Millennium MLC 120.
HDL plan. The percent difference between HDL and SDL doses (maximum and mean) to individual planning structures were calcu- where D SDL is the mean or maximum dose to a structure in the SDL plan and D HDL is the corresponding dose in the original HDL plan.
The homogeneity index, HI, was calculated for target structures

3.B | Summation plan
The purpose of the adaptation method is to develop a deliverable plan for one fraction (Fig. 4) of a treatment course in order to maintain the patient's treatment schedule during linac downtime. The patient will most likely receive the remainder of their fractions of treatment on the original HDL machine. In a representative case a regular course of treatment of 1.8 Gy 9 25 fractions followed by several boost courses had one fraction from the 25-fraction course adapted and replaced in the summation plan (Fig. 5). The detail of the target structure's shoulder region shows a slight increase in the target structure's DVH in the adapted plan.

3.C | Clinical timeline
The intention of creating an adapted plan for an SDL is to allow a patient to continue treatment on the same day he/she was originally scheduled when the originally planned linac is down. The adapted plans need to be processed in the TPS and evaluated using DVH analysis and verified using phantom measurements in an efficient manner. We studied the time required to complete the adaptation tasks. We found that preparing an adapted fraction can be performed within an hour from end-to-end, that is, from the time that physics staff is notified of the desire to proceed with a treatment on a different linac to the evaluation of the The timeline in Fig. 6 outlines the general components involved in VMAT plan adaptation for a single patient with approximate times noted. There could be gains in speed due to tasks that are completed in parallel, that is, multiple physicists may work in tandem to process multiple patient plans.

4.A | Dose changes
Changes in the maximum, mean, and minimum dose to structures between HDL and SDL plans were observed ( Table 1). The mean doses to the structures both increased and decreased over a range T A B L E 1 Mean percent differences in the target and organs-at-risk for four sets of treatment sites.  The effects for one patient treatment field (Fig. 4)  After discussing the results of our adaptation method, our clinical group has established a cutoff of AE10% in changes to the mean or max dose-to-structure. Structures that go outside this threshold will trigger additional review of the adapted DVHs. No review of the adapted treatment is required for plans where the 10% threshold is not violated by any structures. Figure 7 shows the mean percent difference for maximum and mean structure doses were AE10% for all structures analyzed.

4.B | Clinical cases
Prostate treatments were the best performers under adaptation.
The mean dose to the rectum in prostate cases increased by

4.C | Clinical application and timing
Our clinic has a large proportion of VMAT case on our HDL. In a 1-2-week period, 51.7% of the cases treated on the HDL were VMAT. During a period of HDL downtime, there could be more than 10 patients who are potential linac transfer candidates. Let one assume that replanning of the patient treatments for the SDL will take at least 2 h each. With two members of clinic staff working solely on the replanning tasks, verification planning deliveries would commence after at least 10 work hours. Considering our 10-patient example, treatments would commence after an estimated 12.00 h (assuming two physicists performing verification planning and delivery at a cost of 25 min per patient plan). Our method allows the preparation of 10 adapted plans within 2 h (with two staff members) leading to a total of 5.42 h of time between the beginning of HDL-to-SDL adaptation until the completion of verification plan delivery.

| CONCLUSION
We have developed an efficient method to adapt VMAT plans from HDLs to SDLs using a leaf-position averaging process, and have shown the changes to patient dose distributions for a variety of treatment plans. Dose to structures in adapted plans were impacted as follows: the mean doses increased or decreased on the order of 1% while maximum doses increased on the order 10%. In our study of 33 cases, the DVH information indicates that clinically acceptable adapted plans can be produced for a single-fraction treatment. However, since the change in mean or maximum doses to particular structures reaches >10% for some adapted plans, each plan should be evaluated individually before proceeding with treatment.

ACKNOWLEDG MENTS
The authors gratefully acknowledge Jacqueline Zielan, CMD, for her assistance in the linac transfer trials.

CONFLI CT OF INTEREST
The authors declare no conflict of interest.