A novel and clinically useful dynamic conformal arc (DCA)‐based VMAT planning technique for lung SBRT

Abstract Purpose Volumetric modulated arc therapy (VMAT) is gaining popularity for stereotactic treatment of lung lesions for medically inoperable patients. Due to multiple beamlets in delivery of highly modulated VMAT plans, there are dose delivery uncertainties associated with small‐field dosimetry error and interplay effects with small lesions. We describe and compare a clinically useful dynamic conformal arc (DCA)‐based VMAT (d‐VMAT) technique for lung SBRT using flattening filter free (FFF) beams to minimize these effects. Materials and Methods Ten solitary early‐stage I‐II non‐small‐cell lung cancer (NSCLC) patients were treated with a single dose of 30 Gy using 3–6 non‐coplanar VMAT arcs (clinical VMAT) with 6X‐FFF beams in our clinic. These clinically treated plans were re‐optimized using a novel d‐VMAT planning technique. For comparison, d‐VMAT plans were recalculated using DCA with user‐controlled field aperture shape before VMAT optimization. Identical beam geometry, dose calculation algorithm, grid size, and planning objectives were used. The clinical VMAT and d‐VMAT plans were compared via RTOG‐0915 protocol compliances for conformity, gradient indices, and dose to organs at risk (OAR). Additionally, treatment delivery efficiency and accuracy were recorded. Results All plans met RTOG‐0915 requirements. Comparing with clinical VMAT, d‐VMAT plans gave similar target coverage with better target conformity, tighter radiosurgical dose distribution with lower gradient indices, and dose to OAR. Lower total number of monitor units and small beam modulation factor reduced beam‐on time by 1.75 min (P < 0.001), on average (maximum up to 2.52 min). Beam delivery accuracy was improved by 2%, on average (P < 0.05) and maximum up to 6% in some cases for d‐VMAT plans. Conclusion This simple d‐VMAT technique provided excellent plan quality, reduced intermediate dose‐spillage, and dose to OAR while providing faster treatment delivery by significantly reducing beam‐on time. This novel treatment planning approach will improve patient compliance along with potentially reducing intrafraction motion error. Moreover, with less MLC modulation through the target, d‐VMAT could potentially minimize small‐field dosimetry errors and MLC interplay effects. If available, d‐VMAT planning approach is recommended for future clinical lung SBRT plan optimization.

available, d-VMAT planning approach is recommended for future clinical lung SBRT plan optimization.

K E Y W O R D S
DCA-based VMAT, FFF-beam, lung SBRT, single dose

| INTRODUCTION
The use of stereotactic body radiation therapy (SBRT) has become a standard curative treatment for medically inoperable early-staged non-small-cell lung cancer (NSCLC) patients providing a high cure rate and minimal treatment-related toxicity. [1][2][3][4][5] For the selected peripherally located NSCLC patients, single dose of lung SBRT has become a curative-intent treatment option as shown by the randomized clinical trials. [6][7][8][9][10][11][12][13] Most recently, clinical use of flattening filter free (FFF) beams has been of interest in delivering lung SBRT treatments due to dosimetric advantages compared to a flattened beam. 14-18 FFF beams can significantly reduce beam-on time due to their higher dose rates, resulting in better patient compliance, potentially reducing dose delivery uncertainty due to less intrafraction motion error and reduction in out-of-field dose with less head scatter and electron contamination. [14][15][16] A single-large dose of 30 Gy in one fraction lung SBRT treatment is an extreme form of hypofractionation dosing schemata used in our clinic for extracranial lesions where the dose calculation accuracy could potentially suffer by tumor size, tumor location, and the presence of tissue heterogeneity in the lung. Utilizing volumetric modulated arc therapy (VMAT) with FFF-beams 17,18 resulted in better tumor dose coverage and faster treatment delivery of complex lung SBRT treatments compared to historically used plans with 8-15 noncoplanar fixed fields or several coplanar DCA fields with flattened beams. [19][20][21][22] Similar results were observed when compared to linacbased intensity modulated radiation therapy (IMRT), VMAT plans, helical TomoTherapy, or optimized robotic CyberKnife plans (showing significant increases in SBRT treatment times). [23][24][25][26] However, for a single dose of lung SBRT treatments, highly modulated IMRT/ VMAT plans are susceptible to delivery uncertainties due to smallfield dosimetry error 27 and interplay effects 28  users can control the field aperture shape and create a 3D plan using dynamic conformal arc (DCA) therapy before VMAT optimization.
Although a few investigators have studied the clinical use of PO-MLC algorithm for VMAT lung SBRT plan optimization, 30,31 the dosimetric impact and treatment delivery complexity of this planning approach with a FFF beam in the treatment of single high dose (30 Gy in 1 fraction) using non-coplanar VMAT lung SBRT plan has not yet been reported.
As part of SBRT commissioning of Eclipse TPS (Version 15.6), it is important to stress the importance of investigating new planning features to provide the highest quality and most accurate plan. Dose to radiosensitive non-target OAR is a major concern in VMAT lung SBRT treatments, 32,33 specifically while delivering a single-large fraction dose as described here. Herein, we have retrospectively evaluated 10 consecutive early-stage NSCLC patient's plans who underwent a single dose of VMAT lung SBRT treatment in our clinic.
For comparison, the clinical VMAT plans were re-optimized using a DCA-based VMAT (d-VMAT) planning approach with identical beam geometry, dose calculation algorithm, grid size, planning objectives, and parameters. The d-VMAT plans utilized DCA-based dose with the highest strength of field aperture-shape control priority before VMAT optimization; therefore, less beam modulation through the target is expected. The original clinical VMAT plans and re-optimized d-VMAT plans were compared via lung SBRT protocol compliance criteria for the target conformity, intermediate dose-spillage, and dose to OAR per RTOG requirements. 6 Furthermore, treatment delivery efficiency and accuracy were reported.

2.A | Patient characteristics
After obtaining an institutional review board (IRB) approval from our institution, 10 consecutive Stage I-II NSCLC patients with peripherally located tumors who underwent a single dose of lung SBRT treatments (30 Gy) were included in this study.

2.B | Imaging and target definition
All patients were immobilized using Body Pro-Lok TM platform (CIVCO system, Orange City, IA) in the supine position with their arms above their head using an armrest. The free-breathing planning 3D-CT scan was acquired on a GE Lightspeed 16 slice CT scanner (General Electric Medical Systems, Waukesha, WI) with 512 × 512 pixels at 2.5 mm slice thickness in the axial helical mode.
Following the 3D-CT scan, these patients underwent a respiration- Medical Systems, Palo Alto, CA) and co-registered for target delineation. An internal target volume (ITV) was created using the 4D-MIP co-registered with planning 3D-CT images. Planning target volume (PTV) was generated by adding a 5 mm isotropic margin around the ITV per RTOG-0915 recommendation. 6 The relevant critical structures included bilateral lungs excluding the ITV (healthy lung), spinal cord, ribs, heart, trachea/bronchus, esophagus, and skin.

2.C | Clinical VMAT plans and treatment delivery
For the 10 consecutive patients, clinically optimal VMAT-SBRT plans were generated in Eclipse TPS using 3-6 (mean, 4) partial non-coplanar arcs (with ± 5-10°couch kicks) for a Truebeam Linac (Varian Palo Alto, CA) consisting of standard millennium 120 MLC and 6 MV-FFF (1400 MU/min) beam. The isocenter position was set to the geometric center of the PTV. These partial non-coplanar arcs had an arc length of approximately 200-220°. Collimator angles (between 30 and 135°) were manually optimized to reduce the MLC tongue-and-groove dose leakage throughout the arc rotation on a per-patient basis. Additionally, the jaw-tracking option was used during plan optimization to further minimize out-of-field dose leakage.
The prescription dose was 30 Gy in 1 fraction to the PTV while covering at least 95% of the PTV with prescription dose and ensuring that all hot spots (between 120 and 130%) fall within the ITV. All clinical treatment plans were calculated with the advanced AcurosXB (Varian Eclipse TPS, Version 15.6) dose calculation algorithm 34-37 on the planning 3D-CT images with heterogeneity corrections with 1.25 × 1.25 × 1.25 mm 3 CGS and the PO-MLC algorithm. In these clinical plans, low priority of MLC aperture shape controller was used. The dose to medium reporting mode was applied, and the planning objectives followed the RTOG-0915 requirements (Arm 1). 6 Before delivering each VMAT-SBRT plan, a daily quality assurance (QA) check on kilovoltage to megavoltage imaging isocenter coincidence was performed, including IsoCal measurement for the precise and accurate target localization. IsoCal localization accuracy for our Truebeam Linac was < 0.5 mm. All the QA procedures, including patient-specific QA, were in compliance for SBRT treatment delivery. 5 On the treatment day, patient set up prior to singledose lung SBRT was performed using SBRT/IGRT protocol; 5,6 by coregistering the pretreatment conebeam CT with the planning CT scan. Image registration was performed, automatically, based on a bony landmark region of interest, followed by manual refining per-   • Conformity index, CI: ratio of prescription isodose volume to the PTV. CI less than 1.2 is desirable; CI = 1.2-1.5 is acceptable with minor deviations.

2.D | Re-optimized d-VMAT plans
• Gradient index, GI: ratio of 50% prescription isodose volume to the PTV. GI has to be smaller than 3-6, depending on the PTV.
• Maximum dose at any point 2 cm away from the PTV margin in any direction, D 2cm : D 2cm has to be smaller than 50-70%, depending on the PTV size.
• Percentage of normal lung receiving dose equal to 20 Gy or more, V 20 : Per protocol, V 20 should be less than 10%, V 20 less than 15% is acceptable with minor deviations.
• Heterogeneity index, HI: HI = Dmax/prescribed dose was used to evaluate the dose heterogeneity within the PTV.
• Gradient distance, GD: GD is the average distance from 100% prescribed dose to 50% prescribed dose, which indicates how sharp the dose falls off. The GD is used to evaluate dose sparing to normal lung volume. The smaller the value of GD, the faster the dose fall-off around the target.
• Total number of monitor units (MU).
• Modulation factor, MF: ratio of total number of MU to the prescription dose in cGy.
• Beam-on time, BOT: BOT was recorded during VMAT QA phantom measurement at the machine for both plans.
Furthermore, all clinical VMAT and d-VMAT plans were evaluated for the relative volume of normal lung receiving 10 Gy, dose to the spinal cord (maximum and 0.35 cc), heart (maximum and 15 cc), and esophagus (maximum and 5 cc). Since these tumors were peripherally located, the doses to ribs (maximum and 1 cc) and skin (maximum and 10 cc) were also documented. The mean and standard deviation for each dose metric were compared using two-tailed students t-tests (using an upper bound P value of < 0.05, being sta-

3.B | OAR Sparing
The dosimetric differences (mean and standard deviation) between clinical VMAT and d-VMAT plans for the OAR (spinal cord, heart, esophagus, trachea/bronchus, ribs, skin, and normal lung) are listed in

3.C | Treatment delivery efficiency and accuracy
The improvement of treatment delivery efficiency and accuracy is directly associated with Eclipse's new feature of adjustable aperture T A B L E 2 Evaluation of target coverage for all 10 lung SBRT patients for both plans.  Fig. 3). In addition to the uncertainty of modeling small-field dosimetry, there is a potential concern that the interplay effects between the very high dynamic MLC mod-

| DISCUSSION
A novel and clinically useful lung SBRT planning approach via DCAbased dose followed by VMAT optimization is presented per RTOG-0915 compliance for rapid delivery of a single dose of 30 Gy to lung lesions. The new d-VMAT-SBRT plans were highly conformal and achieved similar target coverage (see Table 2) compared to clinical VMAT plans. For all patients, the d-VMAT plans provided similar or better OAR (spinal cord, heart, esophagus, trachea/bronchus, ribs, and skin, see Table 3) sparing and were well below protocol dose requirements. The d-VMAT plans required less total number of MU to deliver the same total prescribed dose due to less beam modulation across the target. Therefore, the beam-on time was reduced sig-  was improved significantly (see Table 4) with measurements analyzed at 2%/2 mm gamma passing criteria. believe that delivering 30 c/n-coplanar fields to treat lung SBRT patients would be clinically impractical for current Linac/clinic workflows. In contrast, utilizing our d-VMAT approach with 6MV-FFF beam can deliver quicker (within a few minutes) and effective curative single-dose SBRT treatments for selected early-staged NSCLC patients.
Potential concerns of changing respiratory motion patterns between the planning CT simulation and the time of treatment have been studied in the past by many investigators. [39][40][41][42] It has been reported in the literature that there were only small changes (within ± 3 mm) due to intrafractional and interfractional motion in lung SBRT treatments. In addition, the mean patient set up time from tumor localization to the end of treatment cone beam CT scan was about 40 min. 41 It was recommended that a symmetrical 5 mm PTV margin around the ITV was adequate to address these potential set up errors. Furthermore, the interplay effect between the MLC modulation and gantry rotation as a function of tumor motion could introduce dose blurring on highly modulated VMAT plans, which can be of another concern for a single high dose of lung SBRT treatments. 28 Table 4). The d-VMAT planning can be easily adopted to any other disease sites (including 3-5 fractions lung SBRT) such as stereotactic treatment of brain or any abdominal/ pelvis lesions such as liver, pancreas, or adrenal glands SBRT. Due to decreased total number of MU/treatment and smaller beam-on time with d-VMAT planning approach, deep inspiration breath-hold lung SBRT treatments may be of value in future investigations. Moreover, the potential use of MLC shape controller strength in d-VMAT planning approach for highly irregular large targets that overlapped with adjacent OAR will be further investigated.

CONFLI CT OF INTEREST
None.