Minimizing dose variation from the interplay effect in stereotactic radiation therapy using volumetric modulated arc therapy for lung cancer

Abstract It is important to improve the magnitude of dose variation that is caused by the interplay effect. The aim of this study was to investigate the impact of the number of breaths (NBs) to the dose variation for VMAT‐SBRT to lung cancer. Data on respiratory motion and multileaf collimator (MLC) sequence were collected from the cases of 30 patients who underwent radiotherapy with VMAT‐SBRT for lung cancer. The NBs in the total irradiation time with VMAT and the maximum craniocaudal amplitude of the target were calculated. The MLC sequence complexity was evaluated using the modulation complexity score for VMAT (MCSv). Static and dynamic measurements were performed using a cylindrical respiratory motion phantom and a micro ionization chamber. The 1 standard deviation which were obtained from 10 dynamic measurements for each patient were defined as dose variation caused by the interplay effect. The dose distributions were also verified with radiochromic film to detect undesired hot and cold dose spot. Dose measurements were also performed with different NBs in the same plan for 16 patients in 30 patients. The correlations between dose variations and parameters assessed for each treatment plan including NBs, MCSv, the MCSv/amplitude quotient (TMMCSv), and the MCSv/amplitude quotient × NBs product (IVS) were evaluated. Dose variation was decreased with increasing NBs, and NBs of >40 times maintained the dose variation within 3% in 15 cases. The correlation between dose variation and IVS which were considered NBs was shown stronger (R 2 = 0.43, P < 0.05) than TMMCSv (R 2 = 0.32, P < 0.05). The NBs is an important factor to reduce the dose variation. The patient who breathes >40 times during irradiation of two partial arcs VMAT (i.e., NBs = 16 breaths per minute) may be suitable for VMAT‐SBRT for lung cancer.


| INTRODUCTION
Stereotactic body radiation therapy (SBRT) using volumetric modulated arc therapy (VMAT) has been widely investigated in recent years. 1,2 Verbakel et al. 1 showed that SBRT using VMAT allows delivery of hypofractionated doses over much less time than conventional SBRT using 10 static noncoplanar fields, with the additional advantage of the plans being more conformal compared with those in conventional SBRT in peripheral stage I lung cancer.
VMAT-SBRT is, however, susceptible to dose variation from the interplay effect between the multileaf collimator (MLC) sequence and tumor motion. 3 The dose variation may not be negligible in VMAT-SBRT, which is generally completed with a few fractions. Jiang et al. 4 showed that the maximum dose variation due to the interplay effect can be up to 30% for one intensity modulated radiation therapy (IMRT) field over one fraction, and 18% for all five IMRT fields over one fraction. Court et al. 5 showed that dose error could be >5% for the area of 40% in the target when target motion was 2 cm for VMAT plans (2 Gy/1 fraction). Tyler et al. 6 found that VMAT-SBRT deliveries showed an increased interplay effect with maximum deviation of AE4.8% in dose received at least 1% (D1%) of the gross tumor volume (GTV). It is important to improve the magnitude of dose variation that is caused by the interplay effect. The number of breaths (NBs) during irradiation was focused on an improving factor for the dose variation in this study. The aim of this study was to investigate the impact of NBs to the dose variation for VMAT-SBRT to lung cancer.

2.A | Patient selection
The data of 30 consecutive patients who underwent treatment with VMAT-SBRT for lung cancer between July 2011 and July 2015 at our institution were selected. Of these 30 patients, 17 patients were primary lung cancer and 13 patients were metastatic lung cancer.
The tumor location, amplitude of respiratory motion in the craniocaudal direction, and respiratory cycle are summarized in Table 1.

2.B | Treatment planning
Four-dimensional computed tomography (4DCT; GE Medical Systems, Waukesha WI, USA) was employed with breathing phases identified by an infrared marker and camera system (Real-time Position Management System (RPM); Varian Medical Systems, Palo Alto CA, USA). For each patient, 10 three-dimensional computed tomography (3DCT) images, corresponding to equally spaced phases of a respiratory cycle, were reconstructed from 4DCT images and imported into a treatment planning system (TPS; Eclipse ver. 10, Varian Medical Systems, Palo Alto CA, USA). The amplitude of each tumor trajectory was assessed by measuring the peak-to-peak tumor position from the phases of the breathing cycle with the TPS.
The GTV was contoured on lung window level CT images over all breathing phases by an oncologist. The GTVs were then merged to generate the internal target volume (ITV) on the average CT image which was reconstructed from the 4DCT. A planning target volume (PTV) was created by adding an isotropic margin of 5 mm around the ITV. In all plans, the prescription dose was 70 Gy delivered in 10 fractions, 7 with at least 95% of the PTV being covered by the prescribed dose. The dose was calculated based on the average CT image using the anisotropic analytical algorithm with inhomogeneity correction. A dose calculation grid size of 2.5 mm was used.
Plans were derived using a NovalisTx linear accelerator (Varian Medical Systems, Palo Alto CA, USA) equipped with a high-definition (HD120) MLC (2.5 mm leaf width in the central region). All VMAT plans used 6 MV and were delivered in two partial arcs (0°-180°, clockwise and counterclockwise) to avoid the contralateral lung. Collimator angles of 30°and 330°were used for each arc to reduce the cumulative effects of interleaf transmission and the tongueand-groove effect. The created treatment plans were actually used to treat the patients. The measured doses were compared with the mean dose recalculated with TPS on each phased CT image of the phantom. These phased CT images were reconstructed from 4DCT image which was acquired with the phantom moved by the amplitude and respiratory waveform. Dose calculation was performed with inhomogeneity correction and the grid size was 2.5 mm. The standard deviation calculated from the results of 10 dynamic measurements was defined as dose variation due to interplay effect.

2.C.2 | Dose distribution
The dose distributions were also verified with radiochromic film (Gafchromic EBT3 ISP, Wayne, NJ, USA) to detect undesired hot and cold dose spot. The film was placed in the coronal plane through the isocenter of the cylindrical insert. While the cylindrical insert moved according to the respiratory waveform and amplitude of each patient, measurements were performed twice with different starting points of the respiratory cycle. EBT3 films were scanned at least 24 hr after irradiation with an Epson ES-10000G flatbed scanner (Seiko Epson Corp., Nagano Japan). Images were acquired in transmission mode and landscape orientation. RBG images were collected at a depth of 16 bits per color channel with a spatial resolution of 300 dpi and were saved in .tiff format. 8 The two dose distributions were aligned by localizing the films using installed room laser before the irradiation. And then, they were compared with dose profile and gamma passing rate using RIT 113 version 5.4 (Radiological Imaging Technology, Colorado Springs CO, USA) as first measured dose T A B L E 1 Descriptive statistics of patient characteristics and plan parameters.

2.D | Plan parameters
To assess the relationship between dose variation and factors related to the interplay effect, quantitative analysis of parameters related to respiratory movement of the tumor and the complexity of the MLC sequence was performed. The complexity of the MLC sequence was evaluated using the modulation complexity score for VMAT (MCSv) introduced by Masi et al. 10 The implies a more modulated/complex plan, and it was assumed that more modulated plans can produce more interplay. A smaller amplitude may cause a smaller interplay, therefore the value of the MCSv divided by the amplitude was smaller for a more dose variation.
where A is amplitude. TMMCSv is assumed that the value is closer to 0 according to the smaller dose variation. The other index (IVS: interplay effect variable score) is calculated as the product of TMMCSv and NBs. It was assumed that undesired hot and cold dose spot can be improved by a larger NBs.
Likewise, IVS is assumed that the value is the closer to 0 according to the smaller dose variation.
The parameters obtained from all plans were analyzed with descriptive statistics. The correlations between the dose variation and the described NBs, MCSv, TMMCSv, and IVS were evaluated using the coefficient of determination (R 2 ). Statistical analyses were performed using IBM SPSS Statistics software, version 22 (IBM Corp., Armonk NY, USA).

3.A | Case-specific QA
The absolute doses measured by static measurement were maintained within 3% of recalculated TPS data on a static CT image of the phantom in all plans, with an average of À1.2 AE 0.63%. Figure 1 shows the dose errors for each patient using the ionization chamber.

| DISCUSSION
The NBs during irradiation affected dose variation from the following results: (a) Dose variation was reduced to less than 3% by increasing the NBs to approximately 40 or more (i.e., number of breaths per minute ≥16 times) during irradiation except for patient number 10 (Fig. 3). There was an increased likelihood of larger dose variations when the NBs was <40 times during irradiation. (b) TMMCSv and IVS showed a correlation with the dose variation Relative dose (%) Craniocaudal direction (cm) cycles obtained from the 4DCT, but for the large motion and increased cycle (60 s), a significant interplay effect was observed, with D99% ranging from À16% to 17%. Therefore, extended beam-  Ehrbar et al. 20 concluded that the interplay effect was not correlated to the modulation factor. In the present study, the correlation between MCSv alone and the dose variation was not found. The

| CONCLUSIONS
The NBs is an important factor to reduce the dose variation caused by the interplay effect with VMAT-SBRT for lung cancer. The patient who breathes >40 times during irradiation of two partial arcs VMAT (i.e., NBs = 16 breaths per minute) may be suitable for VMAT-SBRT for lung cancer.

CONF LICT OF I NTEREST
The authors report no conflicts of interest.