Improving treatment efficiency via photon optimizer (PO) MLC algorithm for synchronous single‐isocenter/multiple‐lesions VMAT lung SBRT

Abstract Purpose Elderly patients with multiple primary or oligometastases (<5 lesions) lesions with associated co‐morbidities may not retain their treatment position for the traditional long SBRT treatment time with individual isocenters for each lesion. Treating multiple lesions synchronously using a single‐isocenter volumetric arc therapy (VMAT) plan would be more efficient with the use of the most recently adopted photon optimizer (PO) MLC algorithm and improve the patient comfort. Herein, we quantified the clinical performance of PO versus its predecessor progressive resolution optimizer (PRO) algorithm for single‐isocenter/multiple‐lesions VMAT lung SBRT. Materials and methods Fourteen patients with metastatic non‐small‐cell lung cancer lesions (two to five, both uni‐ and bilateral lungs) received a highly conformal single‐isocenter co/non‐coplanar VMAT (2–6 arcs) SBRT treatment plan. Patients were treated with a 6X‐FFF beam and Acuros algorithm with a single‐isocenter placed between/among the lesions, using PO for MLC optimization. Average isocenter to tumor distance was 5.5 ± 1.9 cm. Mean combined PTV derived from 4D‐CT scans was 38.7 ± 22.7 cc. Doses were 54 Gy/50 Gy in 3/5 fractions prescribed to 70%–80% isodose line so that at least 95% of the PTV receives 100% of prescribed dose. Plans were re‐optimized using PRO algorithm. Plans were compared via ROTG‐0915 protocol criteria for target conformity, heterogeneity and gradient indices, and dose to organs‐at‐risk (OAR). Additionally, total number of monitor units (MU), modulation factor (MF) and beam‐on time were compared. Results All plans met SBRT protocol requirements for target coverage and OAR doses. Comparison of target coverage and dose to the OAR showed no statistical significance between the two plans. PO had 1042 ± 753 (P < 0.001) less MU than PRO resulting in a beam‐on time of about 0.75 ± 0.5 min (P < 0.001) less, on average. For similar dose distribution, a significant reduction of beam delivery complexity was observed with PO (average MF = 3.7 ± 0.7) vs PRO MLC algorithm (average MF = 4.4 ± 1.3) (P < 0.001). Conclusions PO MLC algorithm improved treatment efficiency without compromising plan quality when compared to PRO algorithm for single‐isocenter/multi‐lesions VMAT lung SBRT. Shorter beam‐on time can potentially reduce intrafraction motion errors and improve patient compliance. PO MLC algorithm is recommended for future clinical lung SBRT plan optimization.

errors and improve patient compliance. PO MLC algorithm is recommended for future clinical lung SBRT plan optimization.

K E Y W O R D S
lung SBRT, MLC, photon optimizer, single-isocenter/multi-lesions, VMAT

| INTRODUCTION
With recent technological advances, SBRT treatment to solitary primary or metastatic lung lesions for medically inoperable non-smallcell lung cancer (NSCLC) patients is safe, effective and has a high cure rate comparable to surgery. [1][2][3][4] SBRT can be beneficial for elderly patients. 5 However, elderly patients who developed multiple primary or oligometastases (<5 lesions) lung lesions with associated co-morbidities may not retain their treatment position for traditional long SBRT treatment times with an individual isocenter placed for each lesion. Treating multiple lung lesions synchronously with a single-isocenter plan, either using intensity-modulated radiation therapy (IMRT) or volumetric arc therapy (VMAT), has been studied by a few researchers. [6][7][8][9] Furthermore, utilizing flattening filter free (FFF) beam 10 for single-isocenter multiple-lesion VMAT lung stereotactic body radiation therapy (SBRT) treatment was fast and efficient, improved the patient comfort and is gaining popularity in clinical practice. [11][12][13] Recently, Varian Eclipse treatment planning system (TPS, Varian Medical Systems, Palo Alto CA, Version 13.6) has implemented a new multi leaf collimators (MLC) optimization algorithm, called photon optimizer (PO). Photon optimizer was created to be more efficient for IMRT/VMAT optimization over its predecessor, progressive resolution optimizer (PRO). 14 The main difference between PO and PRO algorithms is that PO uses a new structure model. For PO, the structures, dose-volume histogram calculations and dose sampling are defined spatially using a single matrix over the image instead of a point-cloud model defining structures that was used in the PRO algorithm. In this configuration, PO algorithm under-samples voxels at the periphery of the target. However, PO configuration uses multiresolution dose calculation approach to increase the dose calculation accuracy. Fixed voxel resolutions of 1.25, 2.5 or 5 mm can be used during multiresolution optimization. For a single-lesion treatment, a few investigators have reported dosimetric differences between PO and PRO optimization for IMRT/VMAT plans. [15][16][17][18] For instance, the advantages and limitations of PO algorithm compared to its predecessor PRO for IMRT plans were evaluated by Binny et al. 16 Eleven plans including prostate, brain, and head and neck treatments were optimized using both PO and PRO algorithms. For similar target coverage and dose to critical structures, they reported that PO algorithm gave higher MLC variability and more monitor units. However, Liu et al. 18 compared PO with PRO algorithms for VMAT planning of lung SBRT and brain stereotactic treatments.
Their retrospective study included 20 lung SBRT patients (10 received 54 Gy in 3 fractions and 10 received 50 Gy in 5 fractions) and 10 brain stereotactic patients received 25 Gy in five fractions.
For identical target coverage, PO algorithm provided comparable plan quality to PRO, with less MLC complexity, thus improving the treatment delivery and contradicting Binny et al. 16 Although dosimetric differences with PO algorithm for a singlelesion treatment with SBRT have been studied previously by Liu et al. 18 , the dosimetric impact and treatment delivery complexity of this algorithm with a FFF-beam in the treatment of multiple lesions simultaneously using a single-isocenter VMAT lung SBRT plan has not yet been reported. When using a single-isocenter for VMAT lung SBRT, the MLCs must travel a longer distance to provide adequate coverage to each lesion simultaneously. Moreover, due to under sampling of the voxels at the periphery of each tumor by the PO algorithm, this distance could cause higher nontarget normal tissue dose to the organs-at-risk (OAR) adjacent to the tumor. This prompted us to quantify the effect of PO MLC algorithm for our clinical implementation of single-isocenter/multi-lesions VMAT lung SBRT approach. Dose to radiosensitive nontarget OAR is a major concern in VMAT lung SBRT treatment, 19,20  A single-isocenter was placed approximately between/among the tumors in each patient. Average isocenter to tumors distance was 5.6 ± 1.9 cm. Highly conformal, clinically optimal VMAT treatment plans were generated on the free-breathing CT scan using 2-6 co/ non-coplanar full/partial arcs (5°-10°, couch kicks were used for noncoplanar partial arcs) for the Truebeam linear accelerator ( Advanced Acuros-based dose calculation and dose to medium was used. A dose of 54 or 50 Gy in 3 and 5 fractions was prescribed to 70%-80% isodose line such that at least 95% of the each PTV received the prescription dose. In addition to optimization ring structures, the generalized normal tissue objective (NTO) parameters were used to control the gradients for each target. Planning objectives for the OAR were per RTOG 0915 guidelines. 3 The main tumor characteristics of the patients included in this study is shown in       Table 2). However, the total number of MU, MF, and BOT show statistically significant differences between the two plans (see Table 2). PO algorithm pro- An example of the corresponding MLC control points for PO and PRO algorithms of a representative patient is shown in Fig. 2 Fig. 3. For the given single-isocenter/multilesions lung SBRT plan, the total number of MU was reduced significantly while using PO algorithm for VMAT plan optimization, suggesting that the PO plan had smaller MF (P < 0.001). Because of this, the average beam-on time for PO plans was 0.75 min less (maximum up to 2.0 min) than PRO plan due to less total MU.

2.D | Plan analysis
The MF for PO vs PRO algorithms and the MF as a function of the isocenter to tumors distance is shown in Fig. 4.   Table 2). Most importantly, the beam-on time was improved by The isodose distribution is shown for photon optimizer (PO) (left) and progressive resolution optimizer (PRO) (right) for an example case patient who was treated for bilateral lung lesions, synchronously using 2-full co-planner arcs. This patient received a synchronous SBRT treatment to a total dose of 50 Gy to each lesion in 5 fractions. The single-isocenter location is shown by the cross-hair. Tumors were located in bilateral lungs. Isocenter to tumor distance was an average of 5. A few investigators have reported the dosimetric differences in PO algorithm for IMRT/VMAT planning in a digital phantom, 15 conventional prostate, head and neck, and brain treatments, 16

CONF LICT OF I NTEREST
The authors declare no conflict of interest.