When employing two adjacent radiation therapy beams, the nature of the dose distribution at their junction is a long-standing topic of study because of the risk of the dose becoming higher or lower than the expected dose (i.e., forming hot or cold spots). This is particularly pertinent when treating very large lesions using a linear accelerator with a limited maximum field size, as more than two fields are required to cover the whole lesion. The cranio–spinal irradiation technique introduced by Faiz M. Khan is a representative example of this, and other cases have been described wherein radiation therapy for the lower abdominal region requires dealing with large lesions [1-4].

Volumetric modulated arc therapy (VMAT) [5,6] is a widely used technique in modern radiation therapy. VMAT employs the benefits of two other techniques namely, intensity modulated radiation therapy (IMRT), which allows for complex dose distribution, and rotational therapy, which enables rapid dose delivery and minimizes damage to the surrounding normal tissues; at our institution, VMAT has been adopted and used as a standard technique for IMRT. In addition, when large lesions exceed the maximum field size of the linear accelerator, treatment is performed using two adjacent VMAT fields to maximize its benefits. In particular, the radiation treatment planning (RTP) system in our institute can simultaneously optimize two VMAT beams with different beam centers, so the planning stage does not need to consider hot/cold spots that may occur at the junction.

This study aimed to analyze the effect of patient setup errors on dose distribution when using two adjacent VMAT fields for treatment. When treating with two adjacent fields, the patient must be positioned exactly where each field specifies, particularly for VMAT techniques that deliver complex modulated dose distributions to the patient. However, a degree of patient setup error is inevitable, and, consequently, the relative setup error between the two VMAT fields crucially requires prior analysis of how such errors affect dose distribution. Computational simulations using RTP system were performed to exclude the possible experimental errors and accurately evaluate the effect of patient setup errors on dose delivery. Finally, the calculated dose distributions for each situation were compared and analyzed.

1. Plan conditions

The three scan ranges of computed tomography images used in different VMAT plans for setup error simulation in this study were >50 cm along the cranial–caudal direction, and the total volumes of the planning target volumes (PTVs) were 861–3,638 cm³. MONACO (v6.1.2; Elekta) RTP was used to generate the plan. To cover a large treatment area, fields A and B were used with different field centers while their field borders were in contact with each other (Fig. 1). The details of the three planning parameters are presented in Table 1.

Table 1 . Plan parameters for the three patients.

Patient	Target			MU		Calculation time (s)
	Range	Volume (cm3)		Field A	Field B
1	L-spine, femur	3,638		2,017	1,136	1,724
2	T-spine, S-spine	986		867	795	1,387
3	C-spine, L-spine	861		674	1,028	980

MU, monitor unit..

Figure 1. Description of field gap (FG) simulation between fields A and B. (a) Negative FG, (b) golden standard, and (c) positive FG.

2. Simulation of setup error

To computationally simulate setup errors, verification plans for the three patients were created. Furthermore, to implement the setup error when a patient was moved to field B position after being treated in field A, the position of field B was moved by x cm in the axis direction to create the corresponding verification plan with a field gap (FG). Finally, the dose was recalculated to obtain a dose distribution with the setup error. Six FGs, viz., −5, −3, −1, 1, 3, and 5 mm were used, and plans having FGs for the three patients were generated so the total number of plans with non-zero FGs was 18. The setup error conditions established are depicted in Fig. 1.

3. Analysis

To analyze the effect of the size of setup error on dose distribution, gamma evaluation was performed using seven gamma criteria of 3%/3, 3%/2, 2%/3, 2%/2, 2%/1, 1%/2, and 1%/1 mm. Homogeneity index (HI) was calculated as shown in Equation 1 using the dose D_(2%) that occupies <2%, dose D_(98%) that occupies >98%, and dose D_(50%) that occupies 50% of PTV volume.

HI has a value of 0 in the ideal case and it increases as the uniformity of the target dose coverage decreases, making it a representative factor to assess the homogeneity of dose distribution delivered to PTV. Both the gamma pass rate (GP) and HI were determined to comprehensively analyze the effects of changes in FG size on dose distribution and target coverage.

1. GP

Several dose distributions computationally simulated for FGs and gamma evaluation results are shown in Figs. 2 and 3, respectively. All GP using a gamma criterion of 3%/3 mm were >90% at an FG of less than or equal to ±3 mm. However, when the FG was ±5 mm, all GPs were <90%. Furthermore, GPs were <90% for most FG conditions when the GP criteria were stricter than 3%/3 mm (Table 2). GP of patient 2 was lower than that of other patients (Fig. 3) because the PTV occupancy ratio of field B over field A was the highest for patient 2.

Table 2 . Gamma pass rates and homogeneity indices.

Patient	FG (mm)	Gamma pass rate (%)							HI
		3/3*	3/2	2/3	2/2	2/1	1/2	1/1
1	−5	86.1	80.9	83.5	77.0	63.9	70.0	55.3	0.17451
	−3	91.6	86.5	89.2	82.6	68.6	75.0	59.4	0.13080
	−1	97.7	95.1	96.2	91.8	76.5	84.1	65.4	0.09601
	1	97.6	94.8	95.9	91.6	76.7	84.1	65.9	0.09524
	3	92.0	86.9	89.8	83.0	69.2	75.6	60.0	0.11476
	5	86.3	80.9	83.6	77.2	64.8	70.5	56.3	0.14643
2	−5	76.8	71.0	73.5	67.0	59.2	61.5	52.3	0.25927
	−3	90.4	81.3	87.0	76.7	65.7	69.7	56.6	0.15625
	−1	98.7	97.2	98.0	95.0	79.3	88.6	66.4	0.06975
	1	98.7	97.1	98.0	94.7	79.3	88.2	66.3	0.06311
	3	90.0	81.5	86.7	76.8	65.9	70.1	57.3	0.12905
	5	76.6	71.3	73.7	67.8	60.1	62.6	53.0	0.19547
3	−5	86.4	80.9	84.2	76.8	63.9	70.3	53.2	0.36541
	−3	93.2	88.4	91.4	84.9	70.6	78.0	58.4	0.28887
	−1	99.1	97.5	98.3	95.4	81.6	89.3	66.5	0.22989
	1	98.9	97.1	98.1	94.8	81.2	88.7	66.6	0.21641
	3	93.5	88.6	91.6	84.8	70.7	78.2	58.9	0.22386
	5	87.1	81.3	84.8	77.3	64.5	71.1	54.4	0.23933

FG, field gap; HI, homogeneity index..

*Gamma criteria are presented as percent dose/distance to agreement..

Figure 2. Comparison of calculated dose distribution with five examples for patient 1 having different field gap (FG) values in the dose range of 80%–110%. (a) –5.0 mm (FG), (b) –3.0 mm, (c) 0 mm, (d) 3.0 mm, and (e) 5.0 mm.

Figure 3. Gamma pass rates of all plans using the gamma criteria of (a) 3%/3 mm, (b) 2%/2 mm, and (c) 1%/1 mm with different setup errors.

2. HI

The highest HI values were 0.26639 and 0.19374 on an average for FGs of −5 and 5 mm, respectively. The lowest HI values were 0.13188 and 0.12492 on an average for FGs of −1 and 1 mm, respectively. Fig. 4 shows a comparison of HI across all plans. HI increased with increased absolute value of FG, which was higher when FG had negative values.

Figure 4. Homogeneity indices of all plans with different setup errors.

GP values were <90% with a gamma criterion of 3%/3 mm, which is one of the widely used GP criteria in our simulation study. Furthermore, when stricter GP criteria were applied, GP values dropped further, with values approaching 50% at 1%/1 mm. This suggests that reducing setup errors when performing VMAT while using two different fields is vital for precisely covering large PTVs.

The decrease in GP with increasing FG was independent of the sign of FG values (Fig. 4), which indicates the direction of setup errors. However, HI seemed to be more affected when FG had negative values (Fig. 4). This implies that PTV coverage is likely to be poorer when two neighboring VMAT fields overlap each other because of setup errors, which must be considered during patient setup in clinical settings.

Although PTV volumes of the three patients differed, this difference did not significantly affect GP results. In the treatment scenario assumed in this study, field A was delivered after accurate patient setup via image guidance and assuming that some setup errors may occur while moving the patient to field B. Therefore, GP value would be lower with a higher PTV ratio occupied by field B rather than the absolute PTV volume. In this study, patient 2 had the highest PTV percentage in field B among the three patients and the lowest GP values with FGs of ±3 and ±5 mm.

This study has several limitations. We considered only one specific target that spanned the pelvis and femur regions; therefore, further research may be needed on different regions. However, as the types of targets with size greater than 40×40 cm² are quite rare, the trends derived from this study are considered sufficiently meaningful. In addition, only setup errors in the longitudinal direction were considered to evaluate the effects between two adjacent fields along the longitudinal axis. The effects of setup errors in the anterior–posterior and lateral directions must be investigated in further study.

The effect of setup error sizes between neighboring fields on dose distribution and target coverage in treatments involving large pelvic areas requiring two VMAT fields was computationally simulated. Setup errors that exceed approximately 3 mm could severely aggravate both GP and target coverage. Therefore, caution must be exercised to navigate setup errors between fields within ≤3 mm for large-area VMAT treatments.

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea, funded by the Korean government (2022R1F1A1064176), and a 2023 research grant from Pusan National University Yangsan Hospital.

The authors have nothing to disclose.

All relevant data are within the paper and its Supporting Information files.

Conceptualization: Hojeong Lee, Hosang Jeon. Data curation: Hojeong Lee, Dong Woon Kim. Formal analysis: Dong Woon Kim, Dong Hyeon Kim. Funding acquisition: Ji Hyeon Joo, Dong Hyeon Kim. Investigation: Dong Woon Kim, Wontaek Kim. Methodology: Hojeong Lee, Jiho Nam. Project administration: Yongkan Ki, Hosang Jeon. Resources: Hojeong Lee, Wontaek Kim. Software: Hojeong Lee, Dahl Park. Supervision: Ji Hyeon Joo, Yongkan Ki. Validation: Hojeong Lee, Hosang Jeon. Visualization: Hojeong Lee, Dahl Park. Writing – original draft: Hojeong Lee, Hosang Jeon. Writing – review & editing: Hosang Jeon, Jiho Nam.