Investigating volumetric repainting to mitigate interplay effect on 4D robustly optimized lung cancer plans in pencil beam scanning proton therapy

Abstract Purpose The interplay effect between dynamic pencil proton beams and motion of the lung tumor presents a challenge in treating lung cancer patients in pencil beam scanning (PBS) proton therapy. The main purpose of the current study was to investigate the interplay effect on the volumetric repainting lung plans with beam delivery in alternating order (“down” and “up” directions), and explore the number of volumetric repaintings needed to achieve acceptable lung cancer PBS proton plan. Method The current retrospective study included ten lung cancer patients. The total dose prescription to the clinical target volume (CTV) was 70 Gy(RBE) with a fractional dose of 2 Gy(RBE). All treatment plans were robustly optimized on all ten phases in the 4DCT data set. The Monte Carlo algorithm was used for the 4D robust optimization, as well as for the final dose calculation. The interplay effect was evaluated for both the nominal (i.e., without repainting) as well as volumetric repainting plans. The interplay evaluation was carried out for each of the ten different phases as the starting phases. Several dosimetric metrics were included to evaluate the worst‐case scenario (WCS) and bandwidth based on the results obtained from treatment delivery starting in ten different breathing phases. Results The number of repaintings needed to meet the criteria 1 (CR1) of target coverage (D95% ≥ 98% and D99% ≥ 97%) ranged from 2 to 10. The number of repaintings needed to meet the CR1 of maximum dose (ΔD1% < 1.5%) ranged from 2 to 7. Similarly, the number of repaintings needed to meet CR1 of homogeneity index (ΔHI < 0.03) ranged from 3 to 10. For the target coverage region, the number of repaintings needed to meet CR1 of bandwidth (<100 cGy) ranged from 3 to 10, whereas for the high‐dose region, the number of repaintings needed to meet CR1 of bandwidth (<100 cGy) ranged from 1 to 7. Based on the overall plan evaluation criteria proposed in the current study, acceptable plans were achieved for nine patients, whereas one patient had acceptable plan with a minor deviation. Conclusion The number of repaintings required to mitigate the interplay effect in PBS lung cancer (tumor motion < 15 mm) was found to be highly patient dependent. For the volumetric repainting with an alternating order, a patient‐specific interplay evaluation strategy must be adopted. Determining the optimal number of repaintings based on the bandwidth and WCS approach could mitigate the interplay effect in PBS lung cancer treatment.

repaintings based on the bandwidth and WCS approach could mitigate the interplay effect in PBS lung cancer treatment. encounters low-and high-density interfaces in its path, such as in the case of lung cancer treatment, [1][2][3][4] as well as if the proton beam traverses a range shifter, which creates an air gap between the distal end of the range shifter and patient body. [5][6][7] The second challenge in treating lung cancer with the PBS proton beam is the interplay effect between dynamic pencil proton beams and motion of the lung tumor. [8][9][10][11][12][13][14][15][16][17][18][19] To mitigate the interplay effect in proton therapy, several strategies have been proposed. These strategies include breath-hold, abdominal compression, gating, and repainting. 8,12,20 Repainting (also referred to as rescanning) allows the energy layers of the proton beam to be delivered more than once to achieve statistical averaging of motion effects. 12,17,18 Volumetric repainting is delivered by repetitive scanning through the whole target volume, whereas in layered repainting, the energy layer is rescanned more than once before switching to the next energy layer. 12,17,18 A benefit of volumetric and layered repainting over gating and breath-hold is the lack of external equipment that could require patient cooperation. 19 Several studies 4,12,14,17 have investigated the potential use of volumetric repainting in lung cancer and compared the volumetric repainting against layered repainting, providing contradictory conclusions. For instance, Seco et al. 17 showed that the volumetric repainting produced better results than layered repainting, whereas Grassberger et al. 14 showed that layered repainting is superior or equal to volumetric repainting. Engwall et al. 12 found that offline breath-sample layered repainting is superior to simple layer repainting and volumetric repainting. In offline breath-sample layered repainting strategy, the layer rescans for each energy level are spread uniformly over the breathing cycle. 12 Recently, Wang et al. 4 reported that a total of four volumetric repaintings were found to be optimal on ProteusPLUS proton system (Ion Beam Applications, Louvain-la-Neuve, Belgium) when they examined it on a moving anthropomorphic lung phantom. Wang et al. 4 demonstrated the feasibility of delivering volumetric repainting plans in a clinical setting.
To take advantage of the volumetric repainting technique in mitigating interplay in lung cancer, the proton delivery system needs to have a faster layer switching mechanism. 18,19 However, the volumetric repainting technique is manufacturer specific, and volumetric repainting capability may vary among different proton machines from the same manufacturer. 19 The above-mentioned volumetric repainting studies 4,12,14,17 on the lung cancer were conducted with beam delivery sequence in "down" direction only such that the proton beam is delivered from the deepest layer (highest energy) to the most proximal layer (lowest energy), and then scans are repeated (i.e., from the highest energy to the lowest energy). Figure 1 At Miami Cancer Institute, ProteusPLUS proton therapy system with a PBS dedicated nozzle is employed. 21,22 Recently, in an effort to decrease the layer switching time, a magnetic field regulation feature has been implemented on the proton delivery system. 23,24 For magnetic field regulation mode, Hall probes are mounted inside specific groups of magnets in the beamline. This allows the reduction in beam stabilization delays and layer switching time in both "down" and "up" directions. 23,24 The "up" direction means the proton beam is delivered from the most proximal layer (lowest energy) to the distal layer (highest energy). Figure 1 The use of magnetic field regulation has decreased the layer switching time to~0.9 s in the "down" direction and~1.3 s in the "up" direction. 23,24 This provides the feasibility of delivering volumetric repainting using an alternating order with beam delivery sequences in "down" and "up" directions as shown in Fig. 1.
The availability of faster energy layer switching in PBS proton therapy has generated a renewed interest in utilizing volumetric repainting in a clinical environment. In the current study, the authors aim to investigate the interplay effect on the volumetric repainting lung plans that are generated using an alternating order ("down" and "up" directions), and explore the number of volumetric repaintings needed to achieve acceptable lung cancer PBS proton treatment plan. For a volumetric repainting plan with beam delivery sequence in "down" direction only, the beamline needs to be switched from the lowest energy to the highest energy of the given treatment field when scans are repeated in depth. Such a delivery technique with big energy steps in magnetic field regulation mode may lead to destabilization of the magnets. It has been reported that big energy steps (of the order of the full energy range) can cause the beam positioning displacements of 1 to 3 mm. 25 A faster layer switching time in both "up" and "down" directions in magnetic field regulation mode provides the choice in terms of delivering a volumetric repainting plan, that is, repainting in both "up" and "down" directions. To date, previous volumetric repainting studies 4,12,14,17 on the lung cancer utilized beam delivery in "down" direction only. In the current study, the authors investigated the volumetric repainting technique with an alternating order ("up" and "down" directions) with a focus on several key items that are relevant for its clinical implementation: (i) the interplay effect evaluation on 4D robustly optimized volumetric repainting plans with an alternating order, (ii) the worst-case scenario (WCS) evaluation based on ten different breathing phases from 4D computed tomography (4DCT) as the starting phases, and (iii) a method to determine the number of volumetric repaintings needed for an acceptable PBS lung cancer treatment plan.

2.A | Patient cohort
The current retrospective study includes ten lung cancer patients.
The selection of the patients was made based on the following criteria: I. 4DCT data set includes all ten phases.
II. Tumor motion is greater than 3 mm but less than 15 mm. 2.B | Contouring, registration, and treatment planning In this retrospective study, gross tumor volume (GTV) was contoured in all ten phases of the 4DCT. The CTV was then generated by an isotropic margin of 5 mm around the GTV in all ten phases. Deformable registration was performed between the average intensity projection CT and ten phases from the 4D CT data set using ANAtomically CONstrained Deformation Algorithm (ANACONDA) within RayStation TPS. 11 In the current study, the total dose prescription to the CTV was 70 Gy(RBE) with a fractional dose of 2 Gy(RBE). PBS plans were generated in RayStation TPS (v9B; RaySearch Laboratories, Stockholm, Sweden) using the beam model of an IBA ProteusPLUS PBS machine that has an in-air one sigma spot size of 3 mm (at the isocenter) for the highest energy of 226.5 MeV. 26,27 All treatment plans were robustly optimized (patient setup uncertainty = 5 mm; range uncertainty = 3.5%) on all ten phases in the 4D data set.
Specifically, 4D robust optimization 11,28 was performed with the goal of 99% of the CTV receiving at least 99% of the prescription dose.
The organs at risk (OARS) such as the heart, spinal cord, normal lung, Examples of beam delivery directions in nominal VR1 (no repainting) and volumetric repainting plans with an alternating order; VR2 = 2 repaintings, VR5 = 5 repaintings, VR8 = 8 repaintings; Note: beam delivery starts from the distal energy layer to the proximal energy layer, and then follows an alternating order. RANA AND ROSENFELD | 109 and esophagus were included in the 4D optimization. The robust objective was applied to the CTV only. All treatment plans were based on the single field uniform dose (SFUD) technique utilizing two to three proton fields. The layer spacing was set by default in RayStation using automatic with scale 1. The layer spacing is calculated based on the Bragg peak width between the proximal 80% and distal 80% of each layer. 29 The spot spacing was also set by default in using automatic with scale 1. The spot spacing varies as a function of depth. 29 The Monte Carlo algorithm was used for the 4D robust optimization (10,000 ions/spot), as well as for the final dose calculation (statistical uncertainty of 0.5%) with a grid size of 3 mm. Treatment plans were then normalized, such that 99% of the CTV received 6930 cGy(RBE). These plans are referred to as nominal plans (VR1) with beam delivery in "down" direction only.

2.C | Volumetric repainting
For each patient, the VR1 plan is used to generate volumetric repainting plans with an alternating order, as shown in Fig. 1. The scripting environment within RayStation was utilized to generate the volumetric repainting plans. Engwall et al. 11,12 has detailed the method to generate the volumetric repainting plans using a script in RayStation TPS.
A minimum monitor unit (MU) of 0.015 as the spot weight was applied for all volumetric repainting plans to ensure the deliverability of the spots on the machine. If the alternating order includes X paintings, the plan is denoted as a VRX plan. For instance, the plan with five paintings with an alternating order is denoted as a VR5 plan.

2.D | Interplay effect
The interplay effect study was also performed within the RayStation scripting environment. 11,12 The interplay effect was evaluated for both the nominal VR1 plan (i.e., without repainting) as well as volu-  30 The interplay evaluation was carried out for each of the ten different phases as the starting phases.

2.E | Worst-case scenario analysis
For each treatment plan (VR1 and VRX) of a given patient, the results were obtained for each phase including the starting phase.
The following metrics were used to evaluate the WCS values and DVH bandwidths from the results of treatment delivery starting in ten different phases.

2.F | Criteria for acceptable plan
Currently, there is no consensus on acceptable interplay effect evaluation criteria for lung proton therapy. The acceptance criteria used in the current study are provided in Table 1. If a given treatment plan with X number of repaintings met the criteria 1 (CR1) of all seven metrics, it was considered "acceptable." However, if X number of repaintings did not meet the CR1 of all seven metrics, the number of repaintings was increased until CR1 of each metric was fulfilled.
The maximum allowable repainting was set to 10. If a final plan met The location of the CTV (red contour) in ten lung patients in the current study. The CTV ranged from 22.10 cc to 181.03 cc, whereas the tumor motion ranged from 3.8 mm to 13.2 mm T A B L E 1 Metrics to evaluate the WCS and bandwidth based on the results obtained from treatment delivery starting in ten different breathing phases. Each phase is considered as one scenario. was considered "acceptable with a minor deviation."

DVH BW avg for high Dose
where, BW = bandwidth of all ten scenarios from the DVHs at dose to the X% of the CTV; X = 1%, 2%, 3%, 4%, and 5%. to meet the CR1 for D 99% but were able to satisfy the CR1 for D 95% . Figure 3(d) shows the WCS results for the CTV ΔD avg . The number of repaintings needed to meet the CR1 (ΔD avg < 2%) ranged from 2 to 10. One patient (#8) did not meet the CR1 of D 99% and ΔD avg . but met the CR2 of these metrics. (ΔD 1% < 1.5%) ranged from 2 to 7. Two out of ten patients needed more than five repaintings to achieve ΔD 1% < 1.5%.  Fig. 4(c)], the number of repaintings needed to meet CR1 (<100 cGy) ranged from 1 to 7.

3.D | Overall plan evaluation
Based on the criteria described in section 2.6, the final plan evaluation showed that acceptable plans were achieved for nine patients, whereas one patient had an acceptable plan with a minor deviation.     • For patient #8 (CTV = 181.03 cc; tumor motion = 13.2 mm), ten repaintings were needed to achieve acceptable plan with a minor deviation.

3.F | Treatment Delivery Time
The Since the current study was primarily focused on the interplay effect, the robustness of the lung plans was not investigated. However, treatment plans were generated with the objective of achieving clinically acceptable and deliverable robust plans. To achieve this, the current study was performed using a 4D robust optimization feature available in the RayStation TPS. As described in Engwall's papers, 11 patients, a patient-specific interplay evaluation strategy must be adopted. This will yield an optimal number of volumetric repaintings for an individual lung cancer patient.
The current study was focused on the volumetric repainting technique with an alternating order. We did not investigate other repainting strategies such as layer repainting and volumetric repainting technique in the "down" direction only. These are the limitations of our study. One of the goals of the current study was to demonstrate the feasibility of mitigating the interplay effect using a volumetric repainting technique with an alternating order rather than to make a comparison against the layer repainting. The use of layer repainting to mitigate the interplay effect cannot be ignored. In the next study, we will make the direct comparison between the volumetric repainting (with alternating order) and layer repainting by providing the dosimetric and radiobiological results.
On the IBA ProteusPLUS machine, if the current regulation feature is employed, the energy layer switching time in the "up" direction can take up to 6 s. 23 More technical details on the magnetic field and current regulation features and volumetric repainting technique on the IBA Pro-teusPLUS machine can be found in previous publications. 23,24 The interplay effect results presented herein are more relevant for an IBA ProteusPLUS PBS machine, which employs magnetic field regulation feature, but not for the current regulation feature. The readers must be aware of the fact that the proton beam delivery systems are machine-and manufacturer specific, and performance of repainting can vary among different PBS proton machines. 34 An independent machine-specific validation for the repainting techniques can provide more accurate estimations of the interplay effect.
We acknowledge that the experimental measurements for the interplay effect were not performed in the current study. In the near future, we aim to perform an experiment using a moving phantom simulating different magnitudes of motion and investigate the computed vs measured doses of repainting plans (layer and volumetric).
At present, to the best of our knowledge, an interplay evaluation module is not currently available in the clinical versions of the proton TPSs. The implementation of interplay evaluation within TPS would not only give us the confidence in using the volumetric repainting technique in the clinical environment but also provides a tool to the clinicians to select the optimal number of repaintings.
This can result in homogenous dose distributions and maintain target coverage leading to a better clinical outcome for PBS lung cancer patients.

| CONCLUSION
The interplay effect was evaluated on the 4D robustly optimized lung cancer plans (tumor motion < 15 mm) for the volumetric repainting technique with an alternating order. The number of repaintings required to mitigate the interplay effect was found to be patient dependent. Determining the optimal number of repaintings based on the bandwidth and WCS approach from DVHs of ten breathing phases could mitigate the interplay effect in PBS lung cancer treatment. It is recommended to perform patient-specific interplay evaluation for PBS lung cancer plans.

ACKNOWLEDG MENTS
The authors thank Erik Engwall, Ph.D., (RaySearch Laboratories, Stockholm, Sweden) for his guidance on the volumetric repainting planning and interplay evaluation.

CONFLI CT OF INTEREST
No conflict of interest.