Introduction
Nasal natural killer/T-cell lymphoma (NNKTCL), which is characterized by regional and ethnic differences, is a rare entity in western countries but more common in Eastern Asia [1, 2]. Radiotherapy (RT) is highly recommended to treat NNKTCL due to good local tumor control and long-term survival [3-5]. However, the local failure rate remains >40% for patients with stage I–II NNKTCL after RT [6-8]. Retrospective literature studies have demonstrated that insufficient target volume dosages are one of the main causes of RT failure in NNKTCL patients [7, 8].
As a tissue-equivalent material, a bolus is always placed on the skin surface to increase the dose to the skin, especially when photons are used to treat superficial tumors and lymph nodes [9, 10]. Phua et al. [11] proved that the intensity-modulated radiotherapy (IMRT) plan with a bolus resulted in superior target coverage compared to the plan without a bolus for nasopharyngeal carcinoma patients with enlarged lymph nodes. Since the tumor target area is close to the nasal surface, a bolus needs to be applied in RT for patients with NNKTCL. Although the bolus and the skin of the nose fit closely according to the RT plan designed using a treatment planning system (TPS), an air gap exists inevitably between them in actual treatment delivery. Therefore, utilizing a bolus in RT for treating NNKTCL results in dose deviations to the target and organs at risk (OARs) between the treatment plan and the treatment applied.
The application of a bolus in IMRT and volumetric-modulated arc therapy (VMAT) for NNKTCL leads to dose differences to the target and OARs between the treatment plan and the actual dose delivered, which may result in treatment failure or tumor recurrence. Therefore, we undertook this study to describe the impact of a bolus on target quality, OARs sparing, and tumor control probability (TCP) in IMRT and VMAT for patients with NNKTCL.
Materials and Methods
Patients and Treatment Plans
Ten patients with stage I–II NNKTCL were randomly selected at our institution, 2 patients were in stage II, and 8 patients were in stage I. The gross tumor volume comprised the primary tumor and regional lymph nodes, which were identified with magnetic resonance imaging or computer tomography (CT) and endoscopic and physical examinations. There were no overlaps between the gross tumor volume and the OARs in our patients. The clinical target volume contained the gross tumor volume and adjacent tissue at risk for contiguous spread. The clinical target volume was expanded symmetrically by 3 mm in all dimensions to obtain the planning target volume (PTV).
The prescribed PTV dose was 50 Gy divided into 25 fractions over 5 weeks. All plans met the clinical planning criteria for target dose and coverage (95% receiving at least 100% of the prescribed PTV) with a maximum hot spot to the PTV of 107% of the prescription dose. The critical structures, such as optic chiasm, optic nerves, eyes, lenses, spinal cord, brain stem; and parotid glands were contoured. The planning objectives for the OARs were adopted from previous studies [1, 2] and defined as follows: the maximum dose (Dmax) to the optic chiasm, optic nerves, eyes; or brain stem were limited to <50 Gy, respectively; the dose limit to lenses was slightly modified and set to 20 Gy; the dose limit to the spinal cord was set to 45 Gy; the mean dose (Dmean) to the parotid gland was limited to <26 Gy.
Using the patients’ original CT images for planning, a clinical 6-MV, 8-field IMRT plan and a 6-MV, 2-arc VMAT plan were created for each patient in the Eclipse 13.6 TPS. All IMRT plans contained 8 fields with the gantry angles 255/290/325/0/0/35/70/105°. The collimator angle for all IMRT plans was set to 0°, and the jaw position in the 2 fields with gantry angles of 0° should be adjusted as described in the literature in order to protect the eyes and lenses [12]. All IMRT plans were optimized with the dynamic sliding-window IMRT delivery technique and a fixed dose rate of 300 monitor units (MUs)/min. All VMAT plans included 2 arcs (clockwise: from 255° to 105°; counterclockwise: from 105° to 255°), with the collimator rotation of 30° and 330°, and the couch rotation set to 0°. All the VMAT plans were optimized with a maximum dose rate of 600 MUs/min.
The non-coplanar fields and the beams passing through the couch were not considered in either IMRT or VMAT plans. These plans were designed, and the dose was calculated adding a 0.5-cm bolus, which was close to the skin of the nose. The bolus limitations were described previously [1, 2]. These plans are referred to as the clinical plans.
Each clinical plan was then copied and pasted, and the bolus was edited to simulate the actual bolus position in the actual dose delivery and then linked to each field in all plans. The bolus was edited as follows:
Selecting the CT layer with the highest apex of the nose and drawing a vertical line down the nearby apex of the nose under the guidance of 0.5 cm grid lines in TPS, a circle was firstly drawn with the diameter of the distance between 2 intersection points of a vertical line and the structures of body, and with the center of the intermediate point of 2 intersection points. And then the circle was drawn with a circular sketch tool slice by slice at all slices to obtain a cylindrical structure in TPS. Finally, a structure of edited body was generated by the intersection volume of the structure of the cylinder and the original body, and a new bolus named edited bolus was inserted and linked to all fields of the copied plans.
Finally, the dose distributions of the pasted plans were recalculated using the fields and MU of the clinical plans. These newly generated plans were referred to as the edited bolus plans. Therefore, each patient had a total of 2 IMRT and 2 VMAT plans.
Dose-Volume Histograms
All dose-volume histogram (DVH) data obtained from the 40 plans were analyzed. Representative DVH for 2 types of plans for IMRT and VMAT are shown in the Figure 1. The maximum and minimum doses, Dmean, the conformal index (CI), and the homogeneity index (HI) of PTV were considered to evaluate the quality of target volume coverage. D2% and D98% (PTV doses of 2 and 98%, respectively) were defined as the maximum and minimum dose of the PTV. Utilizing the Paddick conformity index, CI was defined as CI = TV2PTV/(TV × PIV), where TVPTV represented the PTV volume receiving 95% of the prescribed dose, TV was the PTV volume, and PIV represented the body volume covered by 95% of the prescribed dose. A CI value approaching 1 signified fine conformity of PTV. HI expressed the uniformity of target volume coverage, which was defined as the dose delivered to 5% of the PTV (D5%) minus the dose delivered to 95% of the PTV (D95%), and then divided by Dmean of the PTV. A higher value of HI signified poor homogeneous irradiation of the target volume.
Fig. 1.
Representative dose-volume histograms for 2 types of plans for intensity-modulated radiotherapy (IMRT) and volumetric-modulated arc therapy (VMAT).
To evaluate the radiation dosage to OARs (optic chiasm, left and right optic nerve, left and right eye, left and right lens, spinal cord, brain stem, and left and right parotid gland), Dmax and Dmean of OARs were calculated and compared.
TCP Analysis
In the Niemierko model, TCP is given by equation (1) [13].
where TCD50 represents the dose controlling 50% of the tumor cells when the tumor is homogeneously irradiated, r50 is a parameter describing the slope of the dose-response curve, and for tumors the equivalent uniform dose (EUD) represents the biologically EUD for tumors, which leads to the same probability of local control as the actual inhomogeneous dose distribution. The formula for calculating EUD utilizing the differential DVH data is given as equation (2).
where a represents a unitless number, which is specific to the tumor of interest and describes the dose-volume effect, and vi represents the i’th partial volume receiving a dose of Di in Gy. The vi and Di data pairs are obtained from the differential DVH from a given RT plan.
Additionally, the value for TCD50, r50, and a were needed in this study. Due to lack of relevant biological data for stage I–II NNKTCL, a value of 14.65 Gy for TCD50, with a corresponding value of 0.41 for r50, was taken from the data for stage I–II Hodgkin’s lymphoma in a study by Okunieff et al. [14]. The value of −13 for a was taken from a study evaluating head-and-neck cancer cases with this model [13].
Statistical Analysis
SPSS software (SPSS, Chicago, IL, USA) was used for statistical analyses. The paired t test was performed to determine significant differences for each parameter examined. A value of p < 0.05 was considered statistically significant.
Results
PTV Coverage
The results of dosimetric comparison for PTV using DVH data are listed in Table 1. In this study, dosimetric parameters of D98%, D2%, Dmean, CI, and HI were compared to evaluate the quality of target coverage. Compared with clinical plans, D98%, D2%, and Dmean of edited bolus plans for IMRT decreased 0.48, 0.09, and 0.14 Gy (p = 0.002, 0.015, and 0.000), respectively, but the HI of edited bolus plans increased 0.003 (p = 0.024); D98%, D2%, and Dmean of edited bolus plans for VMAT decreased 0.48, 0.09, and 0.14 Gy (p = 0.000, 0.000, and 0.000), respectively, but the HI of edited bolus plans increased 0.002 (p = 0.010).
Organs at Risk
The radiation doses to OARs of IMRT and VMAT are listed in Tables 2 and 3, respectively. In comparison to clinical plans, the edited bolus plans with IMRT enlarged Dmean to left eye (p = 0.011) and right eye (p = 0.027) and Dmean to the left lens (p = 0.002) and right lens (p = 0.012), but it reduced Dmean to the brain stem and left parotid, and Dmax to the right parotid (p = 0.006, 0.015, and 0.031); the edited bolus plans with VMAT reduced Dmax to the spinal cord and brain stem (p = 0.040, 0.020) and Dmean to the brain stem and left parotid (p = 0.000, 0.020).
Tumor Control Probability
The results of TCP in IMRT and VMAT plans are listed in Table 4. Compared with clinical plans, the TCP of edited bolus plans for IMRT decreased 0.09% (p = 0.000); similarly, the TCP of edited bolus plans for VMAT decreased 0.09% (p = 0.000).
Discussion
Research involving the clinical impact of a bolus on IMRT and VMAT for stage I–II NNKTCL is rare. Hence, this study focusing on the impact of a bolus on target coverage, OARs sparing, and TCP for treating stage I–II NNKTCL is significant. The purpose of the study is to estimate the clinical effect caused by a bolus and to provide guidance in clinical practice for stage I–II NNKTCL patients.
The tissue-equivalent bolus was used to enhance the surface dose in RT, which was confirmed by a comparative dosimetric study [11]. However, there were uncertainties regarding bolus preparation and utilization, especially when the skin is irregularly shaped [15]. In TPS, the virtual bolus form contacted with the body contour perfectly, even in the irregular area of the nose, ear, and scalp; nevertheless, an air gap was inevitably formed between the commercially available flat bolus and the irregular skin area in actual dose delivery. Due to the unpredictable depth, the air gap variously affects the maximum and surface dose, resulting in differences between planned and delivered doses [15-17]. A report demonstrated that a 3D printed customized bolus can provide good contact with the irregular surface and sufficient surface dose enhancement [15]. In our study on IMRT and VMAT dose delivery, the bolus produced minor decreases or increases in D98%, D2%, Dmean,and HI, and altered the delivery dose to some OARs. Our results suggested that the bolus can slightly alter RT dose exposure to the patient in practice, and the clinical effects of this change need further study and analysis.
The edited bolus enlarged or reduced the radiation dose to some OARs, but the dose change was not very large in this study. However, the dose changes still pose a significant risk to the patient. Firstly, dose increases may possibly reach or exceed the threshold of radiotoxicity. Radiation toxicity to the optic nerves and chiasm was markedly increased at doses >60 Gy at 1.8 Gy per fraction, and severe xerostomia was usually avoided if at least 1 parotid gland received a mean dose <20 Gy, or both parotid glands received a mean dose <25 Gy [18, 19]. In addition, the edited bolus might change the radiation dose to some sensitive organs, such as the thyroid gland, which was inevitably exposed to out-of-field radiation in this study. Previous studies stressed that the thyroid gland has a low dose threshold (0.05 Gy) for radiation-induced cancer in children and young adults [12, 20]. Though the OARs radiation dose was not high enough to cause serious side effects in our study, the possible dose increases caused by a deformed bolus may reach the dose threshold of radiotoxicity or result in radiation-induced head-and-neck cancer at high RT prescription doses.
Planning IMRT or VMAT with the given physical constraints in the cost function, TPS employs an optimization algorithm to target the desired aims at the scheduled doses. However, the design and evaluation of RT plans should be directly based on biological indices rather than physical dose parameters. Various radiobiological models have been developed to predict the clinical results, such as the Poisson model, the Niemierko model, and the Marsden model [21-23]. These models apply the radiobiological parameters and the DVH data to predict TCP and the normal tissue complication probability, and further indicate the clinical outcomes. Unfortunately, application of the models is markedly restricted by the lack of reliable biological data. In this study, due to the lack of NNKTCL data, the radiobiological parameters used for calculating TCP were from stage I–II Hodgkin lymphomas. The Niemierko model was applied to calculate the TCP for IMRT and VMAT plans, and the differences in TCP between paired plans were used to indicate the quality of the plans. The results showed that a deformed bolus in IMRT and VMAT reduced the TCP <0.1% in actual treatment; however, in a previous study, TCP was reduced by as much as 10.5% by the impact of the couch top and rails on IMRT and VMAT [24]. The dose change caused by a deformed bolus was less than that by the couch with movable rails, which was possibly reflected by the TCP reduction of <1% in this study. Yet, the exact percent reduction in TCP depended on the radiobiological model and associated parameters, which could impair the TCP calculation.
The deformed bolus caused little dose change to the target, some OARs, and TCP in clinical practice, and the clinical impact of a bolus could be ignored in both IMRT and VMAT for stage I–II NNKTCL.
Acknowledgment
This work was generously supported by the National Natural Science Foundation of China (grant Nos. 11805025), the Foundation and Frontier Research of the Chongqing Science and Technology Commission (CSTC) (No. cstc2015jcyjBX0055 and cstc2018jcyjAX0404], the CSTC Performance Motivation and Guidance Special Project (cstc2017jxjl130017), the Performance Incentive Guide Special Project of Chongqing Scientific Research Institute [No. cstc2019jxjl130031], and the Medical Scientific Research Projects of the Joint of Chongqing Science and Health Commission (No. 2018MSXM041).
Statement of Ethics
This study was approved by the Ethics Committee of the Chongqing University Cancer Hospital, and informed consent was obtained from each patient enrolled.
Disclosure Statement
The authors have no conflicts of interest to declare.

Get Permission