The impact of the field width on VMAT plan quality and the assessment of half field method

Abstract Purpose The goal of this work is to investigate the field width dependence of the volumetric modulated arc therapy (VMAT) plan quality and to propose a half field method to irradiate large volumes effectively with VMAT. Materials and methods We compared four different VMAT methods; namely three full field (3ff), four full field (4ff), three half field (3hf), four half field (4hf). To evaluate the impact of the field width on VMAT plan quality, 12 different size PTVs were created in the virtual phantom and treatment plans generated for each PTV were compared. The effectiveness of our half field method was tested using computed tomography (CT) data of 10 nasopharyngeal carcinoma patients. Results In the virtual phantom study, organs at risk (OAR) mean dose, the maximum point dose, and Homogeneity Index (HI) were found to be field width dependent. Conformation Number (CN) was not significantly affected. In the clinical study, 4hf plans obtained statistically significant dose reduction at brainstem (P < 0.001), right parotid (P = 0.034), oral cavity (P < 0.001), larynx (P = 0.003), cochlea (P = 0.017), lips (P = 0.024), and Body‐PTV (P = 0.04) compared to 4ff plans. Conclusion Our results indicate that VMAT plan quality is dependent on the field width. Half field VMAT method, with the help of reduced field width, shows a clear advantage for the irradiation of large size targets compared to traditionally used full field VMAT plans.


| INTRODUCTION
A novel form of radiotherapy called intensity modulated arc therapy (IMAT) was proposed by Yu in 1995 1 and its clinically applicable implementation, called as volumetric modulated arc therapy (VMAT), was developed by Otto in 2008. 2 Since then, it has been received broad interest mostly due to its reduced monitor units (MU) and treatment delivery time compared to intensity modulated radiation therapy (IMRT). [3][4][5][6][7] Otto's VMAT technique relies on a continuous modification of the gantry speed, the dose rate and the multileaf collimator (MLC) position to modulate the delivered radiation. 2 Initially, a coarse sampling method was used to optimize dynamic gantry motion. However, Otto proposed a progressive sampling method to optimize gantry and MLC positions. According to this method, rotational delivery of radiation is modeled by as a series of static fields. At the beginning of the optimization, a small number of evenly distributed static field samples is initialized. After some iterations, an additional set of field samples is added to the optimization process.
Field samples are continuously added until the whole gantry span is successfully covered. 2 In theory, VMAT can deliver a fraction dose in a single rotation.
However, for complex shaped target volumes more than one rotation is required to achieve equivalent results compared to IMRT. 8 This is mainly caused by constraints using by the optimization engine. Efficiency constraints, like MLC leaf positions and MU weights, 2  Because of this limitation, Varian Eclipse TM Treatment Planning System (TPS), may offer a multi-isocentric treatment plan, which requires multiple setups. Multiple setups should be avoided whenever possible, since they may increase the treatment time and potential for set-up errors. 9 Additional to this, when a VMAT field has an x-jaw aperture more than 15 cm, then one leaf cannot cover the whole field by itself. In other words, it cannot reach the opposite end of the field. This sometimes causes undesirable irradiation of the healthy tissue and degrades the plan quality. One example can be seen from beam's eye view (BEV) of a nasopharyngeal carcinoma VMAT plan (see Fig. 1).
The physical limitations of the treatment machine and the efficiency constraints of VMAT restraint freedom required to get higher quality treatment plans. The purpose of this study is to investigate the impact of the field width (x-jaw aperture) on VMAT plan quality and to propose a mono-isocentric treatment method for irradiation of large targets with VMAT. This method is referred as half field VMAT.

2.A | Treatment planning strategies
We used a similar approach that has been investigated with flattening filter free beams for breast cancer to create half fields. 10 We define the following methods: three full field (3ff), four full field (4ff), three half field (3hf), four half field (4hf). The number in the names of the plans refers to the number of rotations used. In full field plans, jaw apertures are selected so that planned target volume is fully covered at all gantry angles. Half field plans are generated from full field plans by blocking half of the fields. To get half blocked fields, one x-jaw is closed at the center of the field. The same x-jaw was blocked whenever possible to avoid high and low dose regions at the junctions. The strategy underlines these four different treatment techniques is as follows: 3ff is considered to be a baseline treatment plan that is widely accepted in clinical applications. 4ff benefits from one more rotation that provides extra freedom to get more desirable dose distribution. 3hf uses two half blocked fields and one full field (Since half fields require complementary field to cover the whole target, the third field must be fully opened). Half blocked fields have reduced field widths; therefore, MLC leaves travel shorter distance and reach higher modulation capability. 4hf benefits from the advantages of four half blocked fields.
Plans were created on the Eclipse TM , version 13.7.14 (Eclipse, Varian Medical Systems, Palo Alto, CA, USA). All plans were calculated using the Anisotropic Analytical Algorithm (AAA) version 13.7.14. The dose calculation grid was set to 1.25 mm. All plans were generated by using 6 MV photon beams from a Varian Trilogy iX linear accelerator equipped with a 120 leaf Millennium MLC which has a leaf length of 15 cm at isocenter.

2.B | Virtual phantom case
A homogeneous water equivalent virtual phantom (had the width and height of 45 cm) was created within the Eclipse TPS. To evaluate the impact of the field width on VMAT plan quality, 12 spherical planning target volumes (PTVs) and corresponding spherical organs at risk (OARs) were contoured (see Fig. 2). Each PTV had a different diameter (7,10,13,16,19,22, 25, 28, 31, 34, 37, 40 cm) and located at the center of the virtual phantom. For each PTV, two OAR contours (will be regarded as a single parallel organ both in the optimization and the dose calculation process) were created at the anterior and posterior sides of the PTV and located so that their center is on the border of the PTV contour. Then each PTV contour F I G . 1. A BEV at gantry angle 92°. PTVs are shown as red contour. The selected MLC leaves (highlighted as green) reached their maximum traveling distance, thus cannot cover brain stem (blue contour), temporal lobe (yellow contour), and cochlea (green contour). The field width was 25 cm cropped from its corresponding OAR to crate convex shaped PTV structure. The diameter of each OAR contour set equal to the half of their corresponding PTV's diameter. Corresponding OARs for PTV 34, 37, and 40 cm did not completely fit into the virtual phantom.
Hence, exterior parts of them were cropped which reduced their volumes and gave the optimization an additional difficulty.
Four techniques (3ff, 4ff, 3hf, and 4hf) described above were applied to 12 PTVs. In total 48 VMAT treatment plans were generated. The prescription dose was 30 Gy in 10 fractions. All plans were normalized so that 95% of the PTV gets the prescribed dose. The dose volume constraints and priorities used in the optimizations were kept the same and are included in Table 1.

2.C | Clinical case
To test the effectiveness of half field method in a clinical case, a ret- Since 4ff plans expected to give more desirable dose distribution, the 4ff plan was first optimized. ALARA (as low as reasonably achievable) principle was used in the optimization process. The optimization objectives were adjusted until reaching a trade-off plan.
After a successful 4ff plan was obtained, other plans were optimized by using the same optimization criteria to ensure a fair comparison between different techniques. All plans were generated and optimized by the same physicist.

2.D | Data analysis and Comparison Criteria
All plans were quantitatively evaluated using dose-volume histogram In the virtual phantom case, we only considered HI, CN, organ mean dose, and the maximum point dose as comparison criteria. To visualize the quality of the plans we used a plan quality index (QI):

3.A | Virtual phantom case
The smallest PTV has a diameter of 7 cm and a volume of 93.7 cm 3 , whereas the largest PTV has a diameter of 40 cm and a volume of 30124.1 cm 3 . Fig. 3 presents OAR mean dose, the maximum point dose, HI and CN for 3ff, 4ff, 3hf, and 4hf plans. Fig. 4 provides QI values.
OAR mean dose increases quickly for all plans of PTV larger than 33 cm except 4hf method. Reduced OAR mean dose for all PTVs were achieved by 4hf method (see Fig. 3 Table 2. Half field plans used significantly more MU, while virtual phantom mean doses were almost the same in all methods.  Table 4 shows results of statistical comparisons on OAR doses.

3.B | Clinical case
High sparing of OARs near the targets was achieved by half field plans. There was a slight dose reduction in the maximum point dose at brain stem and cord with 3hf (0.3 Gy and 0.4 Gy median dose reduction at brain stem and cord, respectively) and 4hf (0.5 Gy and 0.4 Gy median dose reduction at brain stem and cord, respectively) plans. Although the maximum dose of cord was found to be lower for half field plans, the difference was not significant (P > 0.05).  QI values (see Fig. 4). 4ff method on average achieved better plan quality as compared with 3ff method. One extra rotation helped the optimization to get a more desirable plan. However, a similar degree of decrease in plan quality as the targets get larger was observed for 4ff plans too. OAR mean dose and the maximum point dose were highly affected from the field width. CN was the only comparison criteria that did not seem to be depended on the field width. For the smallest PTV, CN was found to be the lowest. This might be due to the challenge to reach high conformity in small convex shaped targets.
To irradiate large targets with VMAT, we have reported a monoisocentric half field VMAT method. Compared with traditionally used full field VMAT plans, 6,13-16 our approach offers a potential to provide greater flexibility in the dose delivery. This is mainly due to the additional dose modulation capability in the fields having smaller xjaw apertures. Our half field method uses half blocked fields to reduce the field width. We have tested our method in a clinical case. Comparing plans generated for nasopharyngeal carcinoma tumors, our method offers similar PTV coverage while reducing mean OAR doses. Using multiple rotations in VMAT has been reported to achieve superior target coverage compared with single arc VMAT. 8 The findings were consistent.
In overall four field plans generated better HI and lower target mean dose compared to three field plans. However, our half field plans were able to achieve even lower target mean dose. By using four half blocked fields, 4hf method was able to get the lowest HI. Comparing our half field method with other studies, HI was similar to that reported by Szu-Huai et al 15 for high dose PTV and better to that reported by Johnston et al 6 for all PTVs. We achieved superior CN values comparing with previously reported studies. 13,14 However, all of our plans obtained clinically acceptable CN and we did not observe any statistically significant difference. For OAR doses, half field method achieved significant dose reductions especially for OARs located closer to the targets. 3hf and 4hf plans were able to reach statistically significant dose reduction at the mean doses of right parotid, oral cavity, larynx, cochlea, and lips. Chiasm, optic nerves, eyes, and lenses did not differ substantially with different methods. This might be due to the larger distance of these OARs to the targets.
The main drawback of our half field method is the increased MU. It uses higher MU than a typical three field VMAT plan and less MU than a typical 7-9 field IMRT plan. 16 MU is a measurement of the amount of radiation produced by the linear accelerator. Using higher MU results in increased scatter, which increases the dose to healthy tissue and potentially increases the risk of secondary cancers. 17 However, irradiating large targets with VMAT requires the use of large fields. As it can be seen in Fig. 1, MLC leaves cannot cover large fields sufficiently. Because of the physical limitations of currently available treatment machines, using large fields in VMAT may cause undesirable direct irradiation of healthy tissue and OARs.
Our findings indicate that, even though half field plans used considerably more MU, they have shown statistically significant dose reduction at the mean dose of healthy tissue and OARs.
As it can be seen from the QI values of the virtual phantom study, traditional full field method cannot offer clinically acceptable plans for large targets while it gives sufficient results for small targets. However, large targets are a part of modern radiotherapy.
Bilateral breast irradiation, pelvic irradiation for gynecological malignancies, late stage prostate, scalp irradiation, and whole body irradiation require large fields. Our nasopharyngeal carcinoma study demonstrated that half field method has a potential to offer monoisocentric and effective treatment plans with VMAT technique.
This study only covers only Varian accelerators. Our half field method may not be applicable to other vendor's treatment machines.
Jaw-tracking was not available in our clinic at the time this study was conducted, and therefore it was excluded.

| CONCLUSION S
We have demonstrated the field width effect on VMAT plan quality and the capability of our half field method. It offers an improved dosimetric plan quality along with higher amount of MU usage. We also aimed to increase the usability of mono-isocentric VMAT technique for large sized targets.

CONF LICT OF I NTEREST
No conflicts of interest.