Dosimetric and Radiobiological Impact of Patient Setup Errors in Intensity-modulated Radiotherapy for Esophageal Cancer

Abstract

Purpose

To evaluate the impact of patient setup errors on the dosimetry and radiobiological models of intensity-modulated radiotherapy (IMRT) for esophageal cancer.

Methods and materials

This retrospective study with 56 patients in thermoplastic mask (TM) and vacuum bag (VB) groups utilized real setup-error (RSE) data from cone-beam CT scans to generate simulated setup-error (SSE) data following a normal distribution. The SSE data were applied to simulate all treatment fractions per patient by shifting the plan isocenter and recalculating the dose. A simulated plan sum (SPS) was created by accumulating all simulated fraction plans. Comparisons of target dose, improved homogeneity index (iHI), conformity index (CI), tumor control probability (TCP) and normal tissue complication probability (NTCP) were conducted between SPSs and original treatment plans (OTPs). Correlations between RSE and TCP/NTCP were analyzed.

Results

Compared to OTPs in the TM group, the planning target volume (PTV) of SPSs showed reductions in D_95%, D_98%, iHI, CI and TCP by 1.2%, 2.2%, 2.3%, 7.3% and 1.2%, while D_2% increased by 0.3%; D_2% of clinical target volume (CTV) increased by 0.2% (P < .05). In the VB group, D_95%, D_98%, iHI, CI and TCP of PTV decreased by 2.5%, 4.5%, 4.2%, 15.6% and 2.0%, with D_2% increasing by 0.5%; D_2% of CTV increased by 0.5% while D_98% decreased by 0.2% (P < .05). The dose of organs at risk (OARs) changed slightly in both groups. The mean and standard deviation of absolute RSE negatively correlated with the TCP of PTV, while the mean RSE positively correlated with the NTCP of lung and spinal cord.

Conclusions

Setup errors may reduce dose homogeneity and conformity, potentially reducing TCP of PTV and increasing NTCP, especially when mean RSE shifts the isocenter towards OARs. VB immobilization may result in relatively larger impacts of setup errors, but this needs future validation.

Keywords

esophageal cancer setup error intensity modulated radiotherapy (IMRT)radiation dosimetry radiobiology

Introduction

Esophageal cancer is the seventh most common cancer and the sixth-leading cause of cancer death worldwide.¹ Treatment approaches for esophageal cancer encompass surgery, chemotherapy and radiotherapy.² Intensity-modulated radiation therapy (IMRT) stands out as an advanced technique offering improved target coverage, dose homogeneity, conformity, and reduced toxicity to normal tissues compared to traditional 3D-conformal radiation therapy (3D-CRT).³ IMRT achieves this through its ability to generate steep dose gradients between the target volume and surrounding normal tissue.^4,5 However, due to the heightened level of target conformity and steep dose gradients, accurate patient setup is crucial. Significant errors during setup may lead to inadequate radiation dose to the tumor or excessive dose to critical organs, potentially resulting in uncontrolled tumor growth or severe organ complications.

Image-guided radiotherapy (IGRT) can achieve online correction of setup errors by obtaining images before or during treatment, thereby reducing the above differences.^6,7 Although IGRT plays an important role in the implementation of treatment plans, in many regions of China, due to the lack of linear accelerator mounted on-board imager (OBI) and professionals (radiotherapy physicians, physicists and technicians), image correction of setup errors is not performed for each patient in each treatment. In addition, studies^8–10 have reported that frequent use of cone-beam computed tomography (CBCT) scans can increase the risk of secondary cancer by increasing the additional dose to patients. The lack of setup-error data for all treatment fractions makes it challenging to assess the impact of setup errors on dosimetry and radiobiology. Therefore, this study proposes a new method to simulate the setup errors to solve this problem.

Moreover, radiobiological models, including tumor control probability (TCP) and normal tissue complication probability (NTCP) models, are commonly employed to evaluate the radiobiological effects of radiation therapy.^11–15 Previous studies have investigated the impact of patient setup errors on radiotherapy outcomes using radiobiological models.^16–18 These studies demonstrate the value of TCP and NTCP models as useful tools for quantifying the biological consequences of setup errors by translating dosimetric effect into biological responses using mathematical formulas. These models provide insights into how setup errors impact tumor control and normal tissue toxicity. Several well-established radiobiological models are frequently used in this context. The equivalent uniform dose (EUD)-based TCP/NTCP model developed by Niemierko¹⁹ is popular due to its straightforward formula and effectiveness. Additionally, the Webb-Nahum TCP model (WN model) and the Lyman-Kutcher-Burman (LKB) model are widely employed. The WN model combines the linear quadratic (LQ) model for cell survival with Poisson statistics, assuming a normal distribution of tumor radiosensitivity. The LKB model is particularly favored for estimating the risk of complications in normal tissues after radiation therapy and has been clinically validated for predicting radiation-induced complications.²⁰ Additionally, the second cancer complication probability (SCCP) model is often utilized for assessing long-term clinical outcomes, such as in Zhang's study¹⁷ where it was employed to evaluate the long-term risk of developing contralateral breast cancer. Although these radiobiological models were widely applied in the investigation of impact of setup errors, there has been limited attention specifically devoted to the radiobiological implications of setup errors in esophageal cancer radiotherapy.

This study utilized CBCT scans to acquire setup error data, which were then incorporated into simulated treatment plans to investigate their effects on dosimetry and radiobiology, and to analyze the correlation between setup errors and TCP/NTCP. The objective is to provide reference for clinical optimization of treatment.

Methods and Materials

Patients’ General Data

This retrospective cohort study encompassed 56 patients diagnosed with middle and upper esophageal cancer undergoing radical radiotherapy. The patients were diagnosed with squamous cell carcinoma through pathological biopsy and had a Karnofsky Performance Status (KPS) score of 70 points or higher, indicating the ability to cooperate with treatment actively. All patient details were de-identified. Based on the immobilization device employed, patients were divided into two groups: the thermoplastic mask (TM) group and the vacuum bag (VB) group. The general information of patients was summarized in Table 1. The differences in clinical stage, tumor location, gender distribution and planning target volume (PTV) with or without supraclavicular lymph node metastasis (LNM) between the two groups were tested using the chi-square test (with continuity correction for clinical stage comparison) while the age, body mass index (BMI), loss of weight during radiotherapy, KPS score and PTV volume were tested using t-test. Consequently, no significant differences were observed and the two groups were deemed comparable for the purposes of this study. The reporting of this study conforms to STROBE guidelines.²¹

Table 1.

General Information of the Patients.

Patient characteristic	items	TM group	VB group	Test value	P value
Tumor location	Upper segment	19(67.9%)	13(46.4%)	χ²= 2.625	.105
	Middle segment	9(32.1%)	15(53.6%)	χ²= 2.625	.105
Gender	Male	19(67.9%)	22(78.6%)	χ²= 0.820	.365
	Female	9(32.1%)	6(21.4%)	χ²= 0.820	.365
Age	Averaged age	64.0 ± 7.3	67.9 ± 8.2	t = −1.868	.067
BMI	Averaged BMI	20.1 ± 3.0	21.3 ± 3.4	t = −1.313	.195
Weight	Loss of weight during RT (kg)	2.2 ± 1.0	2.2 ± 1.7	t = −0.088	.931
Performance status	KPS score	88.6 ± 7.6	87.1 ± 8.1	t = 0.682	.498
Clinical stage	T₁₋₂N₀₋₁M₀, T₃N₀M₀	4(14.3%)	5(17.9%)	χ²= 0.000	1.000
	T₁₋₂N₂₋₃M₀, T₃N₁₋₃M₀, T₄N₀₋₃M₀	24(85.7%)	23(82.1%)	χ²= 0.000	1.000
PTV	Averaged volume(cm³)	409.8 ± 117.8	478.9 ± 193.9	t = −1.611	.114
	With supraclavicular LNM	23(82.1%)	19(67.9)	χ²= 1.524	.217
	Without supraclavicular LNM	5(17.9%)	9(32.1%)	χ²= 1.524	.217

PTV, planning target volume; TM, thermoplastic mask; VB, vacuum bag; BMI, body mass index; RT, radiotherapy; KPS, Karnofsky Performance Status; LNM, lymph node metastasis.

Body Position Fixation and Original Treatment Planning

Patients in the TM group were positioned in a supine posture, and immobilized utilizing head-neck-shoulder thermoplastic masks (Guangzhou Klarity Medical and Equipment Co, Ltd, China). Patients in the VB group were positioned in a supine position with immobilizing vacuum bags (Medtec Medical, Inc, Buffalo Grove, IL, USA). A spiral CT scanner (Philips Brilliance Big Bore Oncology Configuration, Cleveland, OH) was utilized to perform scans for the patients and the acquired images were then imported into the Eclipse v15.5 treatment planning system. The target volume was delineated by the radiotherapy physicians. The gross target volume (GTV) included the primary esophageal tumor (GTV_primary) and suspicious para-esophageal lymph nodes (GTV_ln). The clinical target volume (CTV) included areas that might be clinically invaded by tumor cells, including 2.0-4.0 cm longitudinal extension of GTV_primary, 0.5-1.0 cm lateral extension of GTV_primary, 0.5-1.0 cm uniform extension of GTV_ln, mediastinal lymph drainage area and bilateral supraclavicular lymph drainage area. Each plan incorporated margins of 0.5 cm for the PTV in the left-right (LR), cranio-caudal (CC) and anterior-posterior (AP) directions based on the CTV. The treatment plans were designed using five fields, with gantry angles set at 210°, 300°, 0°, 60°, and 150°, based on a TrueBeam linear accelerator (Varian Medical Systems, Palo, Alto, CA, USA). A prescribed dose of 50 Gy (administered in 2 Gy/fraction over 25 fractions), was specified. These plans underwent optimization to generate clinically acceptable plans, referred to as the original treatment plans (OTPs). Normalization of the OTPs was performed to ensure 95% of the PTV volume covered by the prescribed dose.

KV-CBCT Scanning and Setup-Error Recording

Prior to the first radiotherapy session, all patients underwent KV-CBCT validation using the OBI of the TrueBeam, subsequently, weekly CBCT scan was performed, and thus, five to six scans were performed per patient throughout the treatment course. The CBCT scans were operated by two different groups of radiotherapy technicians. The scanning parameters were 120 kV voltage and 80 mA current. According to the image gray scale, automatic registration was used firstly and manual registration was used to achieve a substantial overlap between the target volumes in the CBCT images and the planning CT images. Displacement between these image sets in the LR, CC and AP directions was measured. A total of 289 sets of CBCT images were acquired, and real setup error (RSE) data in the LR, CC and AP directions were recorded.

Generation of Simulated Plan

By the central limit theorem, the sum of multiple distributions of any shape will approach a Gaussian distribution. According to the views of van Herk et al,²² setup errors are caused by several smaller errors and therefore are likely to approach a normal distribution. In this study, the mean and standard deviation (SD) of the RSE were calculated for each patient. Randomly simulated setup error (SSE) data, following a normal distribution with the mean and SD of the RSE, were generated for 25 fractions per patient. Finally, the reliability of the SSE data needed to be verified. The specific generation process of SSE data is shown in Figure 1. In Eclipse, duplicates of the OTPs were created, and the SSE for each of the 25 fractions was incorporated into these duplicates by adjusting the isocenter position in three spatial directions and subsequently recalculating the dose distribution. This process resulted in 25 simulated plans, with the gantry angle, collimator angle, monitor unit (MU), and other conditions kept unchanged. Finally, the 25 simulated plans were accumulated to generate a simulated plan sum (SPS).

Figure 1.

Flow chart of simulated setup error (SSE) data generation.

Evaluation of Treatment Plan

Dose volume histogram (DVH) was employed to assess the dose delivered to the target volume and organs at risk (OARs) in each plan. To better quantify the variation between OTPs and SPSs, we refined the homogeneity index (HI) recommended by ICRU report 83.²³ The novel HI was denoted as the improved homogeneity index (iHI). We transformed the HI values ranging from positive infinity to 0 into a normal scale from 0 to 1 using the following formula: iHI = 1/(HI + 1). This transformation allows for easy quantification of differences in the comparison between treatment plans. The following parameters were used to evaluate the PTV and CTV: D_2%, D_50%, D_95%, D_98%, iHI and conformity index (CI). iHI was calculated using the formula D_50% / (D_2% − D_98%+ D_50%), while CI was calculated using the formula (TV_RI / TV) × (TV_RI / V_RI).²⁴ Here, D_2%, D_50% and D_98% represent the dose received by 2% (approximately maximum dose), 50% (median dose) and 98% (approximately minimum dose) of the target volume, respectively. TV_RI is the volume of the target covered by the reference isodose line. TV represents the target volume. V_RI represents the volume covered by the reference isodose line.

D_mean represents the average dose, and D_max represents the maximum dose. V_xGy represents the percentage volume receiving a dose ≥ x Gy, and D_x% represents the dose received by x% volume. The evaluation indices of OARs were as follows: D_mean, V_5Gy, V_10Gy, V_20Gy, V_30Gy for lung tissue; D_max, D_2% for spinal cord and spinal cord-planning organ at risk volume (SC-PRV); D_mean, V_30Gy for heart; and V_5Gy, V_10Gy, V_20Gy, V_30Gy for normal tissue (NT, defined as the body volume minus the PTV).

The TCP was calculated using the WN model,^25,26 where the parameter values used were ρ = 10⁷ cm⁻¹, α₀= 0.4 Gy⁻¹ and σ_α = 0.08 Gy⁻¹.²⁷

The LKB model,²⁸ which comprises three equations, was utilized to calculate the NTCP. For the lung, the parameter values used were n = 1, m = 0.45, TD50 = 31.4 Gy,²⁹ while for the spinal cord, the parameter values used were n = 0.05, m = 0.175, TD50 = 66.5 Gy,³⁰ and for the heart, the parameter values used were n = 0.64, m = 0.13, TD50 = 50.6 Gy.³¹

Statistical Analysis

Statistical analyses were performed utilizing Statistical Package for the Social Sciences (SPSS), version 19.0 (SPSS Inc., Chicago, IL, USA). In this study, the dosimetric and radiobiological parameters of the target volumes and OARs were compared between the OTPs and SPSs, and the data were presented as mean ± SD or median (quartile range). Prior to comparison, Shapiro–Wilk test was employed to test the data for normality. If the data obeyed normal distribution, paired-sample t-test was used. If the data were not normally distributed, the Wilcoxon rank-sum test was chosen. Furthermore, Mann-Whitney U test was utilized to compare the setup errors in the LR, CC and AP directions between the TM and VB groups. Pearson correlation analysis or Spearman correlation analysis was used to analyze the correlation between the RSE and the variation of TCP or NTCP. Statistical difference was considered significant when two-sided P-value was less than .05.

Results

Setup Errors

CBCT scans were performed on 28 patients in the TM group and 28 patients in the VB group prior to their treatment, yielding 145 and 144 sets of RSE data, respectively. In the TM group, the mean absolute RSE values (Mean ± SD) in the LR, CC, and AP directions were 2.1 ± 1.9 mm, 2.0 ± 1.5 mm and 1.1 ± 0.9 mm, respectively, while in the VB group, the corresponding values were 1.7 ± 1.5 mm, 3.1 ± 2.6 mm and 2.6 ± 2.0 mm. Statistical analysis revealed that the setup errors in the CC and AP directions were significantly more pronounced in the VB group compared to the TM group (P < .05). Table 2 provides a comparison of the setup errors, presenting the mean absolute RSE values in the LR, CC, and AP directions for both the TM and VB groups.

Table 2.

Comparison of the Setup Errors Between the TM and VB Groups.

Direction	TM group (mm)	VB group (mm)	Z value	P value
LR	2.1 ± 1.9	1.7 ± 1.5	−1.760	.078
CC	2.0 ± 1.5	3.1 ± 2.6	−3.640	.000
AP	1.1 ± 0.9	2.6 ± 2.0	−6.904	.000

LR, the left-right direction; CC, the cranio-caudal direction; AP, the anterior-posterior direction; TM, thermoplastic mask; VB, vacuum bag.

Based on the RSE data, a total of 700 sets of SSE data were generated for the TM and VB groups, respectively. Figures 2 and 3 present the statistical histograms that illustrate the distribution of RSE and SSE in the LR, CC and AP directions for 28 patients in each group, respectively. These histograms reveal a pattern of normal distribution for both RSE and SSE data, and the distributions were similar across the groups. Upon comparative analysis, the mean value, SD, systematic error (treatment preparation variations, was calculated from the SD of all average setup errors for each patient) and random error (treatment execution variations, was calculated from the root mean square of all SD values for each patient) of the RSE and SSE data were either equal or extremely close (with differences not exceeding 0.2 mm). Furthermore, t-test confirmed that there was no statistically significant difference between the mean values of RSE and SSE (P > .05), indicating a strong consistency between these two datasets. In essence, the SSE data closely mirrors the RSE data, validating their compatibility and reliability. Table 3 provides a comprehensive comparison between the RSE and SSE data.

Figure 2.

Histograms of the real setup errors (RSE) and simulated setup errors (SSE) in the left-right (LR), cranio-caudal (CC) and anterior-posterior (AP) directions for 28 patients in the thermoplastic mask (TM) group.

Figure 3.

Table 3.

Comparison Between the RSE and SSE Data.

		RSE (mm)				SSE (mm)
		M	SD	∑	σ	M	SD	∑	σ	t value	P value
The TM group	LR	0.4	1.5	1.5	2.6	0.4	1.6	1.6	2.5	−0.060	.952
	CC	0.8	1.9	1.9	1.6	0.8	1.9	1.9	1.5	0.028	.978
	AP	−0.5	1.1	1.1	1.1	−0.4	1.1	1.1	1.1	−0.012	.990
The VB group	LR	−0.4	1.5	1.5	1.9	−0.4	1.4	1.4	1.9	−0.028	.978
	CC	0.3	3.3	3.3	2.7	0.2	3.1	3.1	2.7	0.103	.918
	AP	−0.9	2.6	2.6	2.1	−0.8	2.6	2.6	2.1	−0.036	.971

TM, thermoplastic mask; VB, vacuum bag; M, the mean; SD, the standard deviation; ∑, the systematic error; σ, the random error; LR, the left-right direction; CC, the cranio-caudal direction; AP, the anterior-posterior direction; t value, the result of t-test comparison between RSE mean and SSE mean, P > .05, there was no statistical difference between RSE mean and SSE mean.

Dosimetric Effect of Setup Errors on Target Volume

As shown in Table 4, notable differences are observed between the OTPs and SPSs for the PTV in the TM and VB groups. Specifically, in the TM group, the SPSs exhibited a reduction of 1.2% and 2.2% in the D_95% and D_98%, respectively, as compared to the OTPs, while the D_2% increased by 0.3% (P < .05). Furthermore, the iHI and CI of the SPSs exhibited inferior by 2.3% and 7.1%, respectively (P < .05). In the VB group, the D_95% and D_98% of the SPSs decreased by 2.5% and 4.5%, the iHI and CI decreased by 4.2% and 15.6%, respectively, while the D_2% increased slightly by 0.5% (P < .05). Regarding the CTV, the D_2% of the SPSs in the TM group increased by 0.2% compared to the OTPs (P < .05), while the D_2% of the SPSs in the VB group increased by 0.5%, and the D_98% decreased slightly by 0.2% (P < .05). Figures 4 and 5 provide visual comparisons of the DVHs for PTV and CTV, respectively, for a representative case considering the OTP and SPS in the TM and VB groups, respectively. Figure 6 provides the isodose profiles of the OTP and SPS for the representative case.

Figure 4.

Dose-volume histograms (DVHs) of the planning target volume (PTV) and clinical target volume (CTV) between the original treatment plan (OTP) and simulated plan sum (SPS) for a representative case in the thermoplastic mask (TM) group.

Figure 5.

Figure 6.

Isodose curves of the original treatment plan (OTP) and simulated plan sum (SPS) for a representative case.

Table 4.

Dose Comparison Between the OTPs and SPSs for the Target Volume in the TM and VB Groups.

Target	Parameter	TM group					VB group
Target	Parameter	OTP	SPS	Diff(%)	t / Z	P	OTP	SPS	Diff(%)	t / Z	P
PTV	D_2%(Gy)	51.9 (0.7)	52.2 (1.0)	0.3 (0.9)	−3.880	.001	51.4 (0.6)	51.7 (1.0)	0.5 (0.8)	−4.035	<.001
	D_50%(Gy)	51.1 (0.3)	51.0 (0.4)	−0.1 (0.3)	1.065	.296	50.7 (0.2)	50.8 (0.4)	0.0 (0.6)	1.231	.229
	D_95%(Gy)	50.0 (0.0)	49.4 (0.9)	−1.2 (1.6)	−4.623	<.001	50.0 (0.0)	48.8 (1.1)	−2.5 (2.0)	−4.623	<.001
	D_98%(Gy)	49.6 (0.2)	48.5 (1.3)	−2.2 (2.6)	−4.623	<.001	49.7 (0.1)	47.5 (2.2)	−4.5 (4.0)	−4.623	<.001
	iHI	1.0 (0.0)	0.9 (0.0)	−2.3 (2.0)	−4.571	<.001	1.0 (0.0)	0.9 (0.1)	−4.2 (3.0)	−4.636	<.001
	CI	0.9 (0.1)	0.8 (0.1)	−7.3 (6.9)	−4.629	<.001	0.8 (0.0)	0.7 (0.1)	−15.6 (11.6)	−4.624	<.001
CTV	D_2%(Gy)	51.9 (0.7)	52.3 (1.1)	0.2 (0.9)	−3.336	.001	51.5 (0.6)	51.8 (1.1)	0.5 (0.8)	−3.803	<.001
	D_50%(Gy)	51.2 (0.4)	51.2 (0.4)	0.0 (0.3)	−0.219	.828	50.8 (0.3)	50.9 (0.5)	0.1 (0.6)	0.464	.647
	D_95%(Gy)	50.5 (0.3)	50.5 (0.5)	0.0 (0.3)	0.453	.654	50.3 (0.1)	50.3 (0.4)	0.0 (0.8)	−1.867	.062
	D_98%(Gy)	50.3 (0.3)	50.2 (0.4)	0.0 (0.5)	−1.401	.161	50.2 (0.1)	50.1 (0.5)	−0.2 (1.0)	−3.040	.002

OTP, original treatment plan; SPS, simulated plan sum; PTV, planning target volume; CTV, clinical target volume; Diff(%) = (SPS − OTP) / OTP × 100%.

Dosimetric Effects of Setup Errors on OARs

Table 5 respectively illustrates the comparative dosimetric results for the OARs between the OTPs and SPSs in the TM and VB groups. For the lung, the D_mean, V_5Gy, V_10Gy and V_30Gy of the SPSs in the TM group increased by 1.2%, 0.6%, 0.5% and 0.1%, respectively, compared to the OTPs (P < .05). In the VB group, only the V_10Gy of the SPSs increased by 0.5% compared to the OTPs (P < .05), with no statistical differences observed in other parameters. For the SC-PRV, the D_2% and D_max of the SPSs in the VB group decreased by 2.1% and 3.9% compared to the OTPs (P < .05), and there was no statistical difference in the TM group. Regarding the NT, the TM group SPSs showed 0.3% and 0.2% increments in V_5Gy and V_10Gy, respectively, compared to the OTPs (P < .05), while the VB group SPSs exhibited similar trends with 0.3% and 0.1% rises in these metrics (P < .05). However, no statistical differences were observed in the doses to the heart and spinal cord between OTPs and SPSs in both groups. Figures 7 and 8 present the DVH comparisons of the OARs for a representative case, considering the OTP and SPS in the TM and the VB groups, respectively.

Figure 7.

Dose-volume histograms (DVHs) of the lung, spinal cord, spinal cord planning organ at risk volume (SC-PRV), heart and normal tissue (NT) between the original treatment plan (OTP) and simulated plan sum (SPS) for a representative case in the thermoplastic mask (TM) group.

Figure 8.

Table 5.

Dose Comparison Between the OTPs and SPSs for the OARs in the TM and VB Groups.

OAR	Parameter	TM group					VB group
OAR	Parameter	OTP	SPS	Diff(%)	t / Z	P	OTP	SPS	Diff(%)	t / Z	P
Lung	D_mean(Gy)	11.4 (6.4)	11.4 (6.2)	1.2 (3.2)	−2.276	.003	13.7 (2.3)	13.8 (2.7)	0.2 (3.5)	−1.086	.287
	V_5Gy (%)	51.7 ± 19.9	52.3 ± 19.8	0.6 ± 1.0	−3.536	.001	68.8 ± 14.4	69.3 ± 14.5	0.5 ± 1.6	−1.616	.118
	V_10Gy(%)	39.6 ± 14.6	40.2 ± 14.5	0.5 ± 0.8	−3.710	.001	51.2 ± 9.2	51.7 ± 9.5	0.5 ± 1.2	−2.117	.044
	V_20Gy(%)	19.8 ± 6.8	20.0 ± 6.7	0.2 ± 0.5	−1.799	.083	24.1 ± 6.1	24.2 ± 6.0	0.1 ± 0.8	−0.490	.628
	V_30Gy(%)	6.6 ± 2.7	6.7 ± 2.7	0.1 ± 0.2	−3.828	.001	9.1 ± 3.8	9.2 ± 3.9	0.1 ± 0.5	−1.011	.321
Spinal cord	D_2%(Gy)	38.9 (5.5)	39.9 (5.7)	0.1 (3.2)	0.469	.643	35.2 ± 1.9	35.1 ± 2.1	−0.4 ± 3.0	0.699	.491
Spinal cord	D_max(Gy)	42.1 (4.7)	42.4 (4.3)	0.6 (4.3)	0.174	.863	38.6 ± 2.1	38.1 ± 2.4	−1.1 ± 4.5	1.330	.195
SC- PRV	D_2%(Gy)	43.5 (5.0)	42.8 (5.9)	−0.9 (4.1)	1.612	.119	39.0 ± 2.1	38.2 ± 2.4	−2.1 ± 4.1	2.727	.011
SC- PRV	D_max(Gy)	47.8 (1.7)	47.4 (2.8)	−0.4 (4.2)	−1.913	.056	45.7 ± 2.0	43.9 ± 2.3	−3.9 ± 4.3	4.887	<.001
Heart	D_mean(Gy)	6.8 (20.8)	6.6 (20.0)	3.5 (8.7)	−1.735	.094	12.4 ± 9.1	12.5 ± 9.2	1.3 ± 11.3	−0.676	.505
Heart	V_30Gy(%)	6.4 (22.0)	6.5 (23.0)	0.0 (1.1)	−1.591	.123	8.3 (17.3)	7.7 (16.5)	0.0 (1.6)	−0.569	.569
NT	D_mean(Gy)	6.7 ± 1.6	6.7 ± 1.6	0.4 ± 1.0	−1.665	.108	6.2 ± 1.5	6.2 ± 1.5	−0.1 ± 1.2	0.304	.764
	V_5Gy (%)	34.1 ± 8.4	34.4 ± 8.4	0.3 ± 0.4	−3.035	.005	30.9 ± 7.7	31.2 ± 7.8	0.3 ± 0.3	−3.165	.002
	V_10Gy(%)	22.7 ± 5.6	22.9 ± 5.6	0.2 ± 0.3	−3.536	.001	20.9 ± 5.2	21.0 ± 5.2	0.1 ± 0.3	−2.595	.015
	V_20Gy(%)	12.3 ± 3.4	12.3 ± 3.4	−0.1 ± 0.2	−1.600	.110	11.4 ± 3.1	11.2 ± 3.1	−0.1 ± 0.3	2.834	.009
	V_30Gy(%)	4.9 ± 1.8	4.9 ± 1.8	0.0 ± 0.1	−1.513	.130	4.9 ± 2.0	4.8 ± 1.9	−0.1 ± 0.1	−2.436	.015

OAR, organ at risk; OTP, original treatment plan; SPS, simulated plan sum; SC-PRV, spinal cord planning organ at risk volume; NT, normal tissue; Diff(%) = (SPS − OTP) / OTP × 100%, used for D_mean, D_2% and D_max; Diff(%) = SPS − OTP, used for V_5Gy, V_10Gy, V_20Gy and V_30Gy.

Influence of Setup Errors on TCP and NTCP

Table 6 shows that setup errors resulted in a significant reduction in the TCP of PTV in the TM and the VB groups, respectively, when utilizing the WN model to calculate. Specifically, in both the TM and VB groups, the TCP of PTV decreased by 1.2% and 2.0%, respectively, in the SPSs compared to the OTPs (P < .05). Furthermore, in the SPSs, the NTCP of the lung in the TM group increased slightly by 0.1% compared to the OTPs (P < .05), while the NTCP of the lung in the VB group did not differ between the OTPs and SPSs (P > .05). There was a statistical difference in the NTCP of the SC-PRV between OTPs and SPSs in the VB group, but the difference was small enough to be negligible. However, the setup errors had no effect on the TCP of CTV and NTCP of the spinal cord and heart in the TM and VB groups (P > .05).

Table 6.

Comparison of TCP/NTCP Between the OTPs and SPSs in the TM and VB Groups.

Project	Parameter	TM group					VB group
Project	Parameter	OTP	SPS	Diff(%)	t / Z	P	OTP	SPS	Diff(%)	t / Z	P
CTV	TCP_WN	36.6 ± 3.1	36.7 ± 3.4	0.1 ± 0.8	−0.582	.565	33.8 ± 4.2	33.6 ± 4.3	−0.2 ± 1.4	−0.091	.927
PTV	TCP_WN	30.2 (3.2)	28.2 (4.6)	−1.2 (2.9)	−4.600	<.001	27.0 (5.0)	24.9 (8.4)	−2.0 (3.7)	−4.509	<.001
Lung	NTCP	7.3 ± 2.9	7.4 ± 2.9	0.1 ± 0.2	−2.743	.011	10.2 ± 2.6	10.3 ± 2.7	0.1 ± 0.4	−1.205	.239
Spinal cord	NTCP	0.2 ± 0.1	0.2 ± 0.1	0.0 ± 0.0	−1.166	.254	0.0 ± 0.0	0.0 ± 0.0	0.0 ± 0.0	0.487	.630
SC-PRV	NTCP	0.4 ± 0.2	0.4 ± 0.3	0.0 ± 0.1	0.361	.721	0.1 ± 0.1	0.1 ± 0.1	0.0 ± 0.1	2.904	.007
Heart	NTCP	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	−0.284	.776	0.0 (0.0)	0.0 (0.0)	0.0 (0.0)	−0.944	.345

OTP, original plan; SPS, simulated plan sum; CTV, clinical target volume; PTV, planning target volume; SC-PRV, spinal cord planning organ at risk volume; TCP_WN, represents the tumor control probability (TCP) calculated using Webb-Nahum model; NTCP, normal tissue complication probability; Diff(%) = SPS − OTP.

Correlation Between the RSE and the Variation of TCP / NTCP

Table 7 demonstrates a negative correlation (P < .05) between the mean absolute value of RSE in the CC and AP directions and the TCP of the PTV. Specifically, it highlights that an increase in the mean absolute RSE values in these directions leads to a corresponding decrease in the TCP of PTV. Furthermore, the study reveals that the SD of RSE in the LR, CC and AP directions also exerts a negative impact (P < .05) on the TCP of PTV, indicating that an increase in the SD of RSE is associated with a decrease in the TCP of PTV.

Table 7.

Correlations Between the RSE and the Variations of TCP/NTCP (r / P value).

RSE (mm)	PTV_TCP variation (%)	CTV_TCP variation (%)	Lung_NTCP variation (%)	Spinal cord_NTCP variation (%)
LR_mean	0.062/0.648	−0.122/0.370	−0.156/0.252	−0.113/0.407
LR_SD	−0.390/0.003	−0.165/0.224	0.163/0.230	0.105/0.440
CC_mean	−0.580/0.000	−0.117/0.392	0.760/0.000	−0.141/0.302
CC_SD	−0.499/0.000	−0.198/0.144	0.152/0.264	−0.173/0.201
AP_mean	−0.384/0.003	0.233/0.090	0.583/0.000	0.677/0.000
AP_SD	−0.438/0.001	−0.149/0.272	0.075/0.582	−0.241/0.074

RSE, real setup errors; LR_mean, the mean (for lung and spinal cord) or mean absolute value (for PTV and CTV) of RSE in the left-right direction; LR_SD, the standard deviation of RSE in the left-right direction; CC_mean, the mean (for lung and spinal cord) or mean absolute value (for PTV and CTV) of RSE in the cranio-caudal direction; CC_SD, the standard deviation of RSE in the cranio-caudal direction; AP_mean, the mean (for lung and spinal cord) or mean absolute value (for PTV and CTV) of RSE in the anterior-posterior direction; AP_SD, the standard deviation of RSE in the anterior-posterior direction; PTV_TCP variation, the variation of tumor control probability in the planning target volume; CTV_TCP variation, the variation of tumor control probability in the clinical target volume; Lung_NTCP variation, the variation of normal tissue complication probability in the lung; Spinal cord_NTCP variation, the variation of normal tissue complication probability in the spinal cord.

Table 7 also reveals a positive correlation (P < .05) between the mean RSE in the CC and AP directions and the NTCP of the lung. Specifically, it indicates that a positive mean RSE, reflecting a setup error biased towards the head and face, leads to an increase in the NTCP of the lung. Conversely, a negative mean RSE, indicative of a setup error biased towards the feet and back, results in a decrease in the NTCP of the lung. Moreover, the mean RSE in the AP direction displays a similar positive correlation (P < .05) with the NTCP of the spinal cord, where a positive mean RSE (bias towards the face) augments the NTCP of the spinal cord, while a negative mean RSE (bias towards the back) diminishes it.

Figure 9(a) and (b) illustrate scatter plots depicting the correlations between the mean absolute value of RSE in the CC and AP directions and the TCP variation of the PTV, respectively. Figure 9(c) through (e) present scatter plots showing the correlations between the SD of RSE in the LR, CC and AP directions and the TCP variation of the PTV. Figure 9(f) displays a scatter plot exhibiting the correlation between the mean RSE in the CC direction and the NTCP variation of the lung. Figure 9(g) and (h), respectively, present scatter plots that reveal the correlations between the mean RSE in the AP direction and the NTCP variation of the lung and the spinal cord.

Figure 9.

Scatter plots showing correlations between the mean absolute value of real setup errors (RSE) in the cranio-caudal (CC) direction and the tumor control probability (TCP) variation of the planning target volume (PTV) (a), the mean absolute value of RSE in the anterior-posterior (AP) direction and the TCP variation of the PTV (b), the standard deviation (SD) of RSE in the left-right (LR) direction and the TCP variation of the PTV(c), the SD of RSE in the CC direction and the TCP variation of the PTV (d), the SD of RSE in the AP direction and the TCP variation of the PTV (e), the mean RSE in the CC direction and the normal tissue complication probability (NTCP) variation of the lung (f), the mean RSE in the AP direction and the NTCP variation of the lung (g), the mean RSE in the AP direction and the NTCP variation of the spinal cord (h).

Discussion

In this study, a new setup error simulation method was proposed, and based on this method, the dose distribution, TCP and NTCP changes caused by setup errors were simulated. The results show that the setup errors may significantly decrease the D_95%, D_98%, iHI, CI and TCP of the PTV, increase the high dose of the PTV and CTV, and slightly decrease the low dose of the CTV in the VB group. However, the dosimetric impact on the OARs is minimal, and the TCP of CTV, NTCP of spinal cord and heart remain unaffected.

Our findings align with the research conducted by Tamihardja et al,³² who conducted a CBCT-based dose accumulation analysis for the dominant intraprostatic lesion in primary radiotherapy of prostate cancer. They observed an increase in the D_2% of the target and a decrease or no significant change in the D_98%, indicating a decline in dose homogeneity. Similarly, they reported only minor dose variation for OARs. A comparable trend was also noted in the study on postmastectomy radiotherapy by Zhang et al.¹⁷ Additionally, the study on Nasopharyngeal carcinoma by Ding et al³³ revealed deteriorated dose homogeneity, decreased TCP and significant dose variations in several OARs due to setup uncertainties. The observed differences in OAR dose variation compared to our study may be attributed to the closer proximity of the OARs to the PTV in their research.

In radiation therapy simulation planning, several methods are utilized to determine the values of setup errors. Currently, these typically include direct numerical input, CBCT-based estimation, and estimation through surface guidance systems. (1) Direct numerical input is straightforward and simple, but it may not accurately reflect the actual variability and randomness of setup errors, potentially leading to biased impact assessments. (2) CBCT-based estimation^33,34 provides accurate data, but it prolongs scanning time and may increase the patient's additional radiation exposure. (3) Surface guidance systems offer real-time monitoring and quick management of setup errors,^35,36 but they are more appropriate for tumors near the body surface. In our study, we opted to use CBCT to obtain the setup error value due to the deep location of esophageal cancer. Typically, after determining the values of setup errors, CT or CBCT images may be used for dose recalculation for simulation of setup errors. CBCT could capture shape changes in internal organs for different treatment fractions, but it has limitations such as a low signal-to-noise ratio, unclear images, inaccurate electron density, a limited field of view (FOV). If the target area is too extensive, the CBCT scanning range may not encompass the entire target, complicating the evaluation of the complete target dose distribution. Therefore, CT images were used for dose recalculation in this study.

Previous studies have employed a methodology that simulates fixed-distance errors to assess their dosimetric consequences. For instance, Prabhakar et al³⁷ demonstrated that a 3 mm shift in the isocenter position along the right and left lateral followed by anterior directions led to the parotid gland exceeding its tolerated dose threshold. Similarly, Komori et al³⁸ found that the dose deviations in the tumor and mucosa could be contained within ±5% when setup errors were confined to ±3.0 mm in all directions. These results defined the allowable limits for setup errors under extreme conditions. In contrast, the simulation method employed in the present study is tailored to evaluate the dose impact under normal treatments, offering a more realistic assessment of the dosimetric consequences under normal operating conditions. This novel method may be used for other researchers in future study. This simulation method can be applied to daily work, and it is recommended that this simulation method be integrated into the treatment planning system in the future. By inputting the parameters of the setup errors, the impact of the setup errors on dosimetry and biology can be evaluated, which is conductive to the timely adjustment and optimization of the treatment plan.

The choice of immobilization has a direct impact on the amount of setup error. Song et al³⁹ identified the immobilization device as an independent factor influencing setup errors in radiotherapy. Our analysis of setup errors, as presented in Table 2, the notable disparity in setup errors between the TM and VB groups, particularly in the CC and AP directions, may be attributed to several factors. Firstly, the VB's immobilization in these directions offers less rigidity and restraint compared to the TM, allowing for greater patient movement. Secondly, the marking line used with the VB is applied directly to the patient's skin, necessitating bodily movement during positioning, thereby complicating the process. Therefore, the VB group exhibited a higher sensitivity to setup errors, such as the D_95%, D_98%, iHI and CI of PTV. The variations of VB group was 2.5%, 4.5%, 4.2% and 15.6%, respectively, while the variations of TM group was 1.2%, 2.2%, 2.3% and 7.3%. The assessment of setup error is crucial in determining the appropriate setup margin, which is used to reduce sensitivity to setup error.³³ If the setup error is overestimated, the setup margin may be unnecessarily wide, leading to increased target volume and potentially affecting more normal tissue. Conversely, underestimating setup error may result in a narrow setup margin, which could compromise target coverage and increase the risk of uncontrolled tumor growth, as the TCP is more likely to depend on the volume that received the minimum dose.¹⁹ By accurately assessing setup error, we can optimize the setup margin to ensure adequate target coverage while minimizing the irradiation of normal tissue. An appropriate setup margin⁴⁰ is essential for maintaining the intended dosimetric distribution. A well-defined setup margin allows for better alignment of the treatment field with the target volume, reducing the risk of underdosage to the target and overdosage to surrounding normal tissues.⁴¹ This, in turn, can lead to improved treatment outcomes and reduced toxicity.

In this study, we refined the HI as defined by ICRU report 83.²³ The traditional formula for HI, (D_2% − D_98%) / D_50%, aimed to quantify the homogeneity of the target volume, with a value nearing zero indicating better homogeneity. When calculating the relative change in the original HI, sometimes it may lead to unexplained values for variation. For instance, if the HI value in OTP is 0.04 and the corresponding HI value in SPS is 0.12, then the calculated relative change in HI would be (0.12-0.04) / 0.04 = 200%. This suggests that the HI for SPS is 200% worse than that for OTP, which is a misleading interpretation. If we adopt the newly improved iHI for calculation, the iHI for OTP would be 0.962 and 0.893 for SPS. In this condition, the change in iHI would be (0.893-0.962) / 0.962 = −7%, which provides a more interpretable result and indicates that the iHI in SPS is 7% lower than that in OTP. Overall, the iHI scales the original HI to a 0-1 range, making the quantification of dose uniformity comparison between plans more interpretable.

The findings of this study show that setup errors led to inhomogeneous dose distribution in the target volume, and then led to the increase of cold spots and high dose, as illustrated in Figure 6, which depicts the isodose profiles of the target volume in the SPS. Crucially, the emergence of cold spots bears a significant correlation with the reduction in TCP, signifying that a decrease in the approximately minimum dose delivered to the target volume may directly compromise TCP,¹⁹ thereby diminishing the local tumor control efficacy. In this study, it was observed that the VB group exhibited a heightened risk of cold spots compared to the TM group. On the other hand, the dose uniformity of the OTP in the planning system was optimized to an exemplary level. However, in the SPS, the introduction of setup errors resulted in a displacement of the radiation field, worsening the dose uniformity. Consequently, an increase in high dose was observed. The proliferation of these high dose posed a heightened risk to the esophagus, potentially exposing it to elevated levels of irradiation, thereby increasing the risk of esophageal toxicity and even potentially serious consequences such as esophageal perforation or esophageal fistula.⁴²

Based on the findings regarding the implications of setup errors on OAR doses, it was observed that the impact of setup errors on OAR doses was minimal. As shown in Table 5, there was no effect on dose in the spinal cord and heart, except for a slight change in the lung and SC-PRV. This is because the effect of setup errors on the OAR doses is related not only to the size of error but also to the direction of error. Specifically, if a setup error causes the field's isocenter to shift towards the side of the OAR, it will result in an increased dose to that organ. Conversely, if a setup error shifts the isocenter away from the OAR, it will lead to a reduced dose to that organ. Therefore, the minor variations in doses to the OARs may be attributed to the varying directions of setup errors throughout the treatment course.

Furthermore, our study revealed that the SDs and mean absolute values of RSE in various directions, were associated with the TCP of PTV. Increasing SD or mean absolute value of RSE led to a gradual decrease in the TCP of PTV. In terms of NTCP of OARs, it was observed that the mean RSE in the CC direction impacted the NTCP of the lung positively. A positive mean RSE in the CC direction resulted in a bias of the field isocenter towards the patient's feet, leading to increased lung volume irradiation, lung dose and NTCP. Additionally, the mean RSE in the AP direction positively influenced the NTCP of the spinal cord and lung. A positive mean RSE in the AP direction caused a bias of the field isocenter towards the patient's back, resulting in increased irradiation volume of the spinal cord and lung, as well as their respective NTCP values. Therefore, in addition to be careful and meticulous to avoid large setup errors, special attention should be paid to the mean setup errors in the CC and AP directions. When a positive mean value occurs, it should be alert to the possibility of increasing the risk of normal tissue complications, especially radiation pneumonitis and radiation myelitis. Radiation pneumonitis is a well-known complication of thoracic radiotherapy, typically manifesting acutely and effectively treated with corticosteroids, but the progression to radiation-induced pulmonary fibrosis is difficult to predict.^43,44 On the other hand, radiation myelitis is a severe complication of radiation therapy that can result in paraesthesia, paralysis, and, ultimately, death due to secondary infections.⁴⁵ Radiation myelitis, although rare, may occur, particularly for patients with target volume close to the spinal cord. Currently, there are no effective treatments for radiation-induced pulmonary fibrosis and radiation myelitis, except for prevention. Therefore, when monitoring setup errors, more attention should be paid to errors in some specific directions to minimize the risk of toxicity in normal organs.

In this study, we utilized the WN model and LKB model to calculate the TCP and NTCP, respectively. The WN model calculated the TCP based on the DVH for the PTV,²⁷ and it had been widely used in various studies across different types of cancer, such as esophageal cancer,⁴⁶ lung cancer⁴⁷ and nasopharyngeal carcinoma.⁴⁸ On the other hand, the LKB model is frequently used for evaluating and comparing treatment plans,^49–53 selecting prescription doses,⁵⁴ and for the biological optimization of radiotherapy plans.^55,56 In 1991, Emami et al⁵⁷ analyzed and summarized acceptable dose values of 28 OARs for uniform irradiation, based on literature data and personal experience. Subsequently, Burman et al³⁰ fitted these acceptable dose value data into the NTCP model proposed by Lyman.⁵⁸ The Lyman model was originally developed for uniform irradiation, but normal tissues were typically exposed to non-uniform irradiation. To address this, Kutcher and Burman⁵⁹ used an efficient volume conversion method to convert non-uniform doses into equivalent uniform doses, which could then be used with the Lyman model to calculate NTCP. This integrated NTCP model is commonly referred to as the LKB model. Tsougos’ study demonstrated that LKB model exhibits sufficient predictive capacity when appropriate model parameters are selected.²⁰

To the best of our knowledge, there are few reports on the impact of simulated setup errors on TCP and NTCP for IMRT of esophageal cancer. Our study fills a knowledge gap in this area. However, our study has five limitations. Firstly, we only monitored inter-fraction errors, and did not include an analysis of potential intra-fraction errors. Further investigation is needed to assess the impact of intra-fraction errors. Secondly, our study did not account for potential changes in tumor shape, volume and anatomical structure during treatment, as we solely focused on the impact of setup errors. Thirdly, rotational error data was unavailable due to the absence of a six-dimensional couch on our Varian TrueBeam linac. Fourthly, it should be noted that the setup error simulation in this study was solely based on weekly CBCT scans, and the use of more frequent CBCT scans would have provided greater precision. Lastly, the sample size of our study is relatively small for the comparisons of dosimetric and radiobiological parameters between the TM and VB groups, which may lead to underpowered results for detecting small effect sizes, so the findings have to be considered descriptive and require further validation in larger studies.

Conclusions

This study presents a novel approach to simulate the effect of setup errors on dosimetry and radiobiology. The results show that setup errors may affect the dose distribution of the PTV, CTV and OARs, reducing dose homogeneity and conformity, and resulting in a reduction in the TCP of PTV as well as increasing the cold spots and high dose within the CTV. The impact of setup errors on OAR doses and NTCP is minimal. Meanwhile, relatively large mean setup errors in certain specific directions may significantly increase the toxicity of OARs. Therefore, it is particularly important to monitor setup errors during IMRT for esophageal cancer, and more attention should be paid to the mean value of the setup errors in certain specific directions, especially when they shift the isocenter towards the OARs. The impact of setup errors should be evaluated to allow for timely adjustments to the treatment plan. Additionally, VB immobilization may result in relatively larger impacts of setup errors compared to TM, but this needs future validation.

Footnotes

Data Availability Statement

The original contributions presented in the study are included in the article material. Further inquiries can be directed to the corresponding author.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work is sponsored by Medical Scientific Research Foundation of Guangdong Province (A2022218) and Special Fund Project for Science and Technology Innovation Strategy (Vertical Collaborative Management Direction) of Guangdong Province [Shan Fu Ke (2018) No. 157].

Ethics Statement

This retrospective study was approved by the Medical Ethics Committee of Cancer Hospital of Shantou University Medical College (approval number: 2023053). Due to the retrospective nature of this study, the ethics committee of the hospital waived the informed consent of the patients and confirmed compliance with the Declaration of Helsinki and the confidentiality of the patient data.

ORCID iD

Jia-Yang Lu

References

Sung

Ferlay

Siegel

, et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin. 2021;71(3):209-249. doi: https://doi.org/10.3322/caac.21660.

Vovk

Duda

Vovk

. The effect of a multimodal approach on the results of treatment in surgery: Integration of chemotherapy, surgery, and radiotherapy. Georgian Med News. 2024;(347):41-46.

Krishnapriya

Sivanandan

Roshni

, et al. Dosimetric comparison of 3DCRT and IMRT in radical chemoradiotherapy of squamous cell carcinoma esophagus. J Cancer Res Ther. 2023;19(7):1844-1851. doi: https://doi.org/10.4103/jcrt.jcrt_1664_21.

Mazloomi

Abedi

Shanei

, et al. Investigating the number of radiation fields in intensity-modulated radiotherapy plans of optic nerve sheath meningioma patients using dose gradient index. Biomed Phys Eng Express. 2022;8(5). doi: https://doi.org/10.1088/2057-1976/ac6059.

Esmailzadeh

Abedi

Khorasani

. Investigating variation of dose gradient index by different grid sizes in intensity-modulated radiation therapy of optic nerve sheath meningioma. Appl Radiat Isot. 2023;193:110657. doi: https://doi.org/10.1016/j.apradiso.2023.110657.

Roos

CTG

Faiz

Visser

, et al. A comprehensive motion analysis – consequences for high precision image-guided radiotherapy of esophageal cancer patients. Acta Oncol. 2021;60(3):277-284. doi: https://doi.org/10.1080/0284186X.2020.1843707.

Kuo

Fang

Lin

, et al. Effectiveness of image-guided radiotherapy for locally advanced esophageal squamous cell carcinoma patients treated with definitive concurrent chemoradiotherapy. Thorac Cancer. 2020;11(1):113-119. doi: https://doi.org/10.1111/1759-7714.13244.

Kan

Leung

Wong

, et al. Radiation dose from cone beam computed tomography for image-guided radiation therapy. Int J Radiat Oncol Biol Phys. 2008;70(1):272-279. doi: https://doi.org/10.1016/j.ijrobp.

Yuasa

Shiinoki

Onizuka

, et al. Estimation of effective imaging dose and excess absolute risk of secondary cancer incidence for four-dimensional cone-beam computed tomography acquisition. J Appl Clin Med Phys. 2019;20(11):57-68. doi: https://doi.org/10.1002/acm2.12741.

10.

Zhou

Bai

Zhang

, et al. Imaging dose, cancer risk and cost analysis in image-guided radiotherapy of cancers. Sci Rep. 2018;8(1):10076. doi: https://doi.org/10.1038/s41598-018-28431-9.

11.

Dashnamoorthy

Jeyasingh

Rajamanickam

. Validation of esophageal cancer treatment methods from 3D-CRT, IMRT, and rapid arc plans using custom python software to compare radiobiological plans to normal tissue integral dosage. Rep Pract Oncol Radiother. 2023;28(1):54-65. doi: https://doi.org/10.5603/RPOR.a2023.0002.

12.

Mazonakis

Tzanis

Lyraraki

, et al. Automatic radiobiological comparison of radiation therapy plans: An application to gastric cancer. Cancers (Basel). 2022;14(24):6098. doi: https://doi.org/10.3390/cancers14246098.

13.

Tai

Phat

Ngoc Anh

, et al. Dosimetric and radiobiological comparison between conventional and hypofractionated breast treatment plans using the Halcyon system. Front Oncol. 2023;13:1259416. doi: https://doi.org/10.3389/fonc.2023.1259416.

14.

Chen

Huang

Wang

, et al. Dosimetric comparison and biological evaluation of fixed-jaw intensity-modulated radiation therapy for T-shaped esophageal cancer. Radiat Oncol. 2021;16(1):158. doi: https://doi.org/10.1186/s13014-021-01882-7.

15.

Balasubramanian

Shobana

Anabalagan

, et al. Dosimetric and radiobiological comparison of IMRT, VMAT, and helical tomotherapy planning techniques in hippocampal sparing whole brain radiotherapy with simultaneous integrated boost for multiple brain metastases. Radiat Environ Biophys. 2024;63(1):47-57. doi: https://doi.org/10.1007/s00411-023-01052-1.

16.

Nakano

Tanabe

Sasamoto

, et al. Radiobiological evaluation considering setup error on single-isocenter irradiation in stereotactic radiosurgery. J Appl Clin Med Phys. 2021;22(7):266-275. doi: https://doi.org/10.1002/acm2.13322.

17.

Zhang

Peng

, et al. Evaluation of hybrid VMAT advantages and robustness considering setup errors using surface guided dose accumulation for internal lymph mammary nodes irradiation of postmastectomy radiotherapy. Front Oncol. 2022;12:907181. doi: https://doi.org/10.3389/fonc.2022.907181.

18.

Hamming-Vrieze

Depauw

Craft

, et al. Impact of setup and range uncertainties on TCP and NTCP following VMAT or IMPT of oropharyngeal cancer patients. Phys Med Biol. 2019;64(9):095001. doi: https://doi.org/10.1088/1361-6560/ab1459.

19.

Gay

Niemierko

. A free program for calculating EUD-based NTCP and TCP in external beam radiotherapy. Phys Med. 2007;23(3-4):115-125. doi: https://doi.org/10.1016/j.ejmp.2007.07.001.

20.

Tsougos

Mavroidis

Theodorou

, et al. Clinical validation of the LKB model and parameter sets for predicting radiation-induced pneumonitis from breast cancer radiotherapy. Phys Med Biol. 2006;51(3):L1-L9. doi: https://doi.org/10.1088/0031-9155/51/3/L01.

21.

von Elm

Altman

Egger

, et al. The strengthening the reporting of observational studies in epidemiology (STROBE) statement: Guidelines for reporting observational studies. Ann Intern Med. 2007;147(8):573-577.

22.

van Herk

Remeijer

Rasch

Lebesque

. The probability of correct target dosage: Dose-population histograms for deriving treatment margins in radiotherapy. Int J Radiat Oncol Biol Phys. 2000;47(4):1121-1135. doi: https://doi.org/10.1016/s0360-3016(00)00518-6.

23.

Hodapp

. Der ICRU-report 83: Verordnung, Dokumentation und Kommunikation der fluenzmodulierten Photonenstrahlentherapie (IMRT) [The ICRU report 83: Prescribing, recording and reporting photon-beam intensity-modulated radiation therapy (IMRT)]. Strahlenther Onkol. 2012;188(1):97-99. doi: https://doi.org/10.1007/s00066-011-0015-x.

24.

Feuvret

Noël

Mazeron

Bey

. Conformity index: A review. Int J Radiat Oncol Biol Phys. 2006;64(2):333-342. doi: https://doi.org/10.1016/j.ijrobp.2005.09.028.

25.

Lee

Wynne

Webb

, et al. A comparison of proton and megavoltage X-ray treatment planning for prostate cancer. Radiother Oncol. 1994;33(3):239-253. doi: https://doi.org/10.1016/0167-8140(94)90359-x.

26.

Isacsson

Montelius

Jung

, et al. Comparative treatment planning between proton and X-ray therapy in locally advanced rectal cancer. Radiother Oncol. 1996;41(3):263-272. doi: https://doi.org/10.1016/s0167-8140(96)01851-8.

27.

Bedford

Viviers

Guzel

, et al. A quantitative treatment planning study evaluating the potential of dose escalation in conformal radiotherapy of the oesophagus. Radiother Oncol. 2000;57(2):183-193. doi: https://doi.org/10.1016/s0167-8140(00)00258-9.

28.

Chang

Gehrke

Prabhakar

, et al. RADBIOMOD: A simple program for utilising biological modelling in radiotherapy plan evaluation. Phys Med. 2016;32(1):248-254. doi: https://doi.org/10.1016/j.ejmp.2015.10.091.

29.

Marks

Bentzen

Deasy

, et al. Radiation dose-volume effects in the lung. Int J Radiat Oncol Biol Phys. 2010; 76(3 Suppl):S70-S76. doi: https://doi.org/10.1016/j.ijrobp.2009.06.091.

30.

Burman

Kutcher

Emami

Goitein

. Fitting of normal tissue tolerance data to an analytic function. Int J Radiat Oncol Biol Phys. 1991;21(1):123-135. doi: https://doi.org/10.1016/0360-3016(91)90172-z.

31.

Gagliardi

Constine

Moiseenko

, et al. Radiation dose-volume effects in the heart. Int J Radiat Oncol Biol Phys. 2010; 76(3 Suppl):S77-S85. doi: https://doi.org/10.1016/j.ijrobp.2009.04.093.

32.

Tamihardja

Cirsi

Kessler

, et al. Cone beam CT-based dose accumulation and analysis of delivered dose to the dominant intraprostatic lesion in primary radiotherapy of prostate cancer. Radiat Oncol. 2021;16(1):205. doi: https://doi.org/10.1186/s13014-021-01933-z.

33.

Ding

Xiang

Zeng

, et al. Evaluation of plan robustness on the dosimetry of volumetric arc radiotherapy (VMAT) with set-up uncertainty in nasopharyngeal carcinoma (NPC) radiotherapy. Radiat Oncol. 2022;17(1):1. doi: https://doi.org/10.1186/s13014-021-01970-8.

34.

Xiao

Han

Wei

, et al. Setup errors analysis in iterative kV CBCT: A clinical study of cervical cancer treated with volumetric modulated arc therapy. J Appl Clin Med Phys. 2024;9:e14480. doi: https://doi.org/10.1002/acm2.14480.

35.

Laaksomaa

Aula

Sarudis

, et al. Surface-guided radiotherapy systems in locoregional deep inspiration breath hold radiotherapy for breast cancer – a multicenter study on the setup accuracy. Rep Pract Oncol Radiother. 2024;29(2):176-186. doi: https://doi.org/10.5603/rpor.99673.

36.

Malone

Ryan

Nicholson

, et al. Intrafraction motion in surface-guided breast radiation therapy and its implications on a single planning target volume margin strategy. Pract Radiat Oncol. 2024;12:S1879-8500(24)00196-6. doi: https://doi.org/10.1016/j.prro.2024.06.017.

37.

Prabhakar

Laviraj

Haresh

, et al. Impact of patient setup error in the treatment of head and neck cancer with intensity modulated radiation therapy. Phys Med. 2010;26(1):26-33. doi: https://doi.org/10.1016/j.ejmp.2009.05.001.

38.

Komori

Hirose

Sato

, et al. Dosimetric effect of set-up error in accelerator-based boron neutron capture therapy for head and neck cancer. J Radiat Res. 2022;63(4):684-695. doi: https://doi.org/10.1093/jrr/rrac017.

39.

Song

Peng

Chen

Luo

. Compare of interfractional setup reproducibility between vacuum-lock bag and thermoplastic mask in radiotherapy for breast cancer. Technol Cancer Res Treat. 2021;20:1533033821104303. doi: https://doi.org/10.1177/15330338211043037.

40.

Luo

Yin

Zhang

, et al.

Does the protocol-required uniform margin around the CTV adequately account for setup inaccuracies in whole breast irradiation?

Radiat Oncol. 2021;16(1):143. doi: https://doi.org/10.1186/s13014-021-01863-w.

41.

Pruthi

Nagpal

Pandey

Effectiveness of 6D couch with daily cone beam computed tomography in reducing PTV margins for glioblastoma multiforme. J Neurosci Rural Pract. 2023; 14(1):78-83. doi: https://doi.org/10.25259/JNRP_2_2022. Epub 2022 Oct 22.

42.

Cheung

Huang

, et al. Improving target coverage and organ-at-risk sparing in intensity-modulated radiotherapy for cervical oesophageal cancer using a simple optimisation method. PLoS One. 2015;10(3):e0121679. doi: https://doi.org/10.1371/journal.pone.0121679.

43.

Bledsoe

Nath

Decker

. Radiation pneumonitis. Clin Chest Med. 2017;38(2):201-208. doi: https://doi.org/10.1016/j.ccm.2016.12.004.

44.

Hanania

Mainwaring

Ghebre

, et al. Radiation-Induced lung injury: Assessment and management. Chest. 2019;156(1):150-162. doi: https://doi.org/10.1016/j.chest.2019.03.033.

45.

Ong

Wong

Soliman

, et al. Radiation myelopathy following stereotactic body radiation therapy for spine metastases. J Neurooncol. 2022;159(1):23-31. doi: https://doi.org/10.1007/s11060-022-04037-0.

46.

Warren

Partridge

Carrington

, et al. Radiobiological determination of dose escalation and normal tissue toxicity in definitive chemoradiation therapy for esophageal cancer. Int J Radiat Oncol Biol Phys. 2014;90(2):423-429. doi: https://doi.org/10.1016/j.ijrobp.2014.06.028.

47.

Huang

Lin

, et al. Radiobiological modeling analysis of the optimal fraction scheme in patients with peripheral non-small cell lung cancer undergoing stereotactic body radiotherapy. Sci Rep. 2015;5:18010. doi: https://doi.org/10.1038/srep18010.

48.

Hou

Jiang

, et al. Benefits of intraoral stents for sparing normal tissue in radiotherapy of nasopharyngeal carcinoma: A radiobiological model-based quantitative analysis. Transl Cancer Res. 2021;10(10):4281-4289. doi: https://doi.org/10.21037/tcr-21-1324.

49.

Chiou

Huang

, et al. The potential of integrating stereotactic ablative radiotherapy techniques with hyperfractionation for lung cancer. Thorac Cancer. 2024;15(22):1679-1687. doi: https://doi.org/10.1111/1759-7714.15335.

50.

Dashnamoorthy

Jeyasingh

Rajamanickam

, et al. Assessment of the dosimetric Index from IMRT and rapid arc plan for oropharyngeal cancer with simultaneous integrated boost (SIB) technique in combination with EUD-based NTCP and TCP radiobiological models. Asian Pac J Cancer Prev. 2024;25(5):1515-1528. doi: https://doi.org/10.31557/APJCP.2024.25.5.1515.

51.

Raturi

Motegi

Zenda

, et al. Comparison of a hybrid IMRT/VMAT technique with non-coplanar VMAT and non-coplanar IMRT for unresectable olfactory neuroblastoma using the RayStation treatment planning system-EUD, NTCP and planning study. J Radiat Res. 2021;62(3):540-548. doi: https://doi.org/10.1093/jrr/rrab010.

52.

Ming

Mao

, et al. Intensity-modulated proton and carbon-ion radiotherapy using a fixed-beam system for locally advanced lung cancer: dosimetric comparison with x-ray radiotherapy and normal tissue complication probability (NTCP) evaluation. Phys Med Biol. 2024;69(1). doi: https://doi.org/10.1088/1361-6560/ad13d1.

53.

Balasubramanian

Shobana

. A dosimetric and radiobiological comparison of intensity modulated radiotherapy, volumetric modulated arc therapy and helical tomotherapy planning techniques in synchronous bilateral breast cancer. Asian Pac J Cancer Prev. 2022;23(12):4233-4241. doi: https://doi.org/10.31557/APJCP.2022.23.12.4233.

54.

Thaper

Singh

Kamal

, et al. Impact of dose heterogeneity in target on TCP and NTCP for various radiobiological models in liver SBRT: Different isodose prescription strategy. Biomed Phys Eng Express. 2021;7(1):015020. doi: https://doi.org/10.1088/2057-1976/abd3f0.

55.

Momeni

Broomand

Roozmand

, et al. Estimating the dose-response relationship for ocular pain after radiotherapy of head and neck cancers and skull base tumors based on the LKB radiobiological model. J Biomed Phys Eng. 2023;13(5):411-420. doi: https://doi.org/10.31661/jbpe.v0i0.2210-1554.

56.

Feng

Tao

Zhu

, et al. An integrated strategy of biological and physical constraints in biological optimization for cervical carcinoma. Radiat Oncol. 2017;12(1):64. doi: https://doi.org/10.1186/s13014-017-0784-1.

57.

Emami

Lyman

Brown

, et al. Tolerance of normal tissue to therapeutic irradiation. Int J Radiat Oncol Biol Phys. 1991;21(1):109-122. doi: https://doi.org/10.1016/0360-3016(91)90171-y.

58.

Lyman

. Complication probability as assessed from dose-volume histograms. Radiat Res Suppl. 1985;8:S13-S19.

59.

Kutcher

Burman

. Calculation of complication probability factors for non-uniform normal tissue irradiation: The effective volume method. Int J Radiat Oncol Biol Phys. 1989;16(6):1623-1630. doi: https://doi.org/10.1016/0360-3016(89)90972-3.