Dosimetric Evaluation of Body Contour Changes to Target Volumes and Organs at Risk for Cervix and Head and Neck Radiotherapy Plans

Purpose: To investigate how much dose discrepancy would be caused by the anatomy changes during the radiotherapy (RT) course. Methods: Ten cervical cancer and ten nasopharyngeal carcinoma (NPC) CT datasets from RT patients were enrolled. The body contour from different directions changed to simulate the weight loss or gain for cervical cancer patients, who had been treated with external-beam RT using intensity-modulated radiation therapy (IMRT) or volumetric modulated arc therapy (VMAT). Moreover, the body contour from facial and shoulder superior-inferior positional change had been also assessed for NPC patients using IMRT or VMAT. The new CT (n-CT) was generated by the body contour changes with different directions based on original CT datasets. The dosimetric parameters to target volumes and organs at risk (OARs) were evaluated in Eclipse based on n-CT. Results: The target volumes and OARs were influenced by the body contour changes. Body contour expansion resulted in coverage loss, whereas body contour shrinkage increased the dose to the OARs. These findings were generally con-sistent for both IMRT and VMAT plans. Over a course of research, the dose to 95% of the target volumes for cervical cancer decreased by up to 2.83% per cm for IMRT and 2.87% per cm for VMAT (P < 0.05). And the influence on H&N plans was that the dose to 95% of the target volumes (low risk regions) decreased by up to 4.45% per cm. Conclusions: The RT staff could determine whether resimulation and replaning or not according to which body contour directions were changed.


Introduction
A radiotherapy (RT) treatment starts with the acquisition of computed tomography (CT) scan, which is used to plan an individualized treatment for the patient. Highly precise RT techniques enable delivered dose in accordance with planned dose to the clinical target volumes (CTVs) and organs and risk (OARs), based on the premise that the anatomy is unchanged since the planning stage [1]. However, it takes about 1 to 2 weeks from initial CT scan to the start of RT, and a course of intensity-modulated radiation therapy (IMRT) or volumetric modulated arc therapy (VMAT) for cervical cancer or nasopharyngeal carcinoma (NPC) is typically 5 to 7 weeks [2]. During this long time, the weight change, tumor shrinkage, and shoulder position variations possibly happen, so the delivered dose to CTVs and OARs are not the same as the planned dose owing to the body contour changes.
The abdomen of cervical cancer patients usually has a lot of adipose tissue deformed with weight fluctuation. The dosimetric effects of changes in body contour, e.g., due to weight change, have been studied for head and neck [3] and prostate cancer patients [4]. Lee et al. [5] reported the weight change during RT on the development of toxicity in patients with locally advanced cervical cancer treated with IMRT. Stauch et al. [6] and Sun et al. [7] reported dosimetric effects of the body contour changes for prostate and H&N cancer.
Chow et al. [8] reported the dosimetric estimation on variations of patient size in prostate VMAT therapy. Astrid et al. [9] studied the dosimetric effects of changes in body contour for pancreatic cancer. The pieces of literature [3]- [9] showed that involuntary weight changes have a dosimetric effect on RT plans for abdominal neoplasms. Several studies show that H&N patients' anatomy changes during the course of the treatment, and that this results in dosimetric changes from the original plans [1] [3] [7]. As a result, the patient's body contour on the treatment day could differ from the CT scan compared with cone-beam CT (CBCT) images taken on the treatment day [10]. Figure 1 illustrated a cervical cancer patient's pre-treatment and post-treatment image which is from CBCT registration, the anterior body contour shrank 1.55 cm in this example. So body contour changes induced perturbations in the dose distribution, although generally only for large changes. There were no studies found about the dosimetric evaluation of target volumes and OARs from different body contour directions for tumor sites in the abdomen and pelvis, including the cervix.
The purpose of this study is to quantify for RT of cervical cancer and NPC patients the impacts on target dose coverage and OARs sparing dose parameters as a result of changes in body contour in IMRT or VMAT plans to ensure treatment efficacy, and provide a prejudgment whether further re-assessment of the plan is needed for RT staff.

Patient Selection
This study included ten cervical cancer patients and ten NPC patients. A total of 20 patients were randomly selected and retrospectively analyzed according to the prescription dose based on the TNM staging. The basic information of 20 selected patients was shown in Table 1. RT planning images of the 20 patients were used for this dosimetric evaluation. Their mean age of cervical cancer patients and NPC were 46.8 ± 9.5 and 58.3 ± 12.4 years old, respectively. The patients were positioned supine and immobilized with thermoplastic fixation. For each of cervix and H&N patient, a CT image with a slice thickness of 3 mm was obtained by a CT simulator (Brilliance-16, Philips Medical Systems Inc., Cleveland, OH, USA).

RT Plans
All scans were exported to Eclipse for target volumes and OARs delineation and Z. Wu et al.  treatment planning. The plan was delivered with 6-MV photon beams from a linear accelerator (Trilogy and Millennium 120 MLC, Varian Medical System, Palo Alto, CA, USA). Ten cervical cancer patients had been treated with external-beam radiation therapy using IMRT (n = 5) and VMAT (n = 5). Of the 5 patients who underwent IMRT, 1 was planned to 46 Gy at 2.0 Gy/fraction, 4 was planned to 50 Gy at 2.0 Gy/fraction. Of the 5 patients who underwent VMAT, 2 were planned to 46 Gy, 3 was planned to 50 Gy at 2.0 Gy/fraction. The IMRT plans consisted of an 8-beam arrangement, with gantry angles of 0˚, 45˚, 90˚, 135˚, 179˚, 225˚, 270˚, and 315˚. The VMAT plans used two 360˚ arcs.
The original plan (o-plan) in o-CT was copied to n-CT to recomputed (not reoptimized) to evaluate the effect of weight change on the dose distribution over numerous regions of interest. The recomputed plan was named a new plan (n-plan). The dosimetric parameters of PTV and OARs for body contour changes were statistically compared for IMRT vs VMAT plans, using a 2-sided Wilcoxon rank-sum test in SPSS (IBM Armonk, NY, USA). A p-value of 0.05 or less was considered statistically significant.
The dose to D 95% (the dose that covers 95% of the volumes) variation of the target was defined as ΔD 95% which could be expressed as follows: where, D 95%,n-plan and D 95%,o-plan are the D 95% of the n-plan and o-plan, respectively.
The maximal dose variation of OARs was defined as ΔD max which could be expressed as follows: where, D max,n-plan and D max,o-plan is the D max of the n-plan and o-plan, respectively.  2cm in the anterior direction, similarly for other directions. The ΔD 95% of target volumes increased near linearly as body contour shrank. The ΔD 95% variations were fitted linearly. The mean slopes of the lines were −1.98% ± 0.1%, −1.21% ± 0.4%, −2.83% ± 0.5% and −1.3% ± 0.2% per cm for IMRT in A, P, U and L direction expanding, respectively. And the mean slopes of the lines were −2.16% ± 0.1%, −1.25% ± 0.6%, −2.87% ± 0.3% and −1.18% ± 0.3% per cm for VMAT.

Target Volumes
The difference between IMRT and VMAT was significant (p < 0.05) in A, U, L directions. For both IMRT and VMAT plans, the anterior region ΔD 95% showed larger variations than do in the posterior region, this was mainly because the anterior body contour changed length is longer than the posterior body contour change. In A and U direction, the ΔD 95% for VMAT was overall larger than that for IMRT, which is mainly because in VMAT plans the dose is delivered by two full arcs with multileaf collimator modulation and the IMRT plans the dose is delivered in an 8-beam arrangement. Whereas in L direction, the ΔD 95% shows slightly larger than that for VMAT, probably because in IMRT plans, the gantry angle of 45˚, 90˚ and 135˚ had a greater influence than VMAT plans during the The results of ΔD 95% variation as body contour change in superior(S), inferior (I), facial (F) direction for ten H&N plans are shown in Figure 4. The ΔD 95% of PTV1 decreased near linearly as body contour expanded in F direction, and the body contour change of S, I directions almost did not influence PTV1. The mean slopes of the PTV1 ΔD 95% lines were −3.88% ± 0.18% per cm for IMRT and −4.13% ± 0.25% per cm for VMAT in F direction expanding (P < 0.05). The ΔD 95% of PTV2 has a similar influence to PTV1 whereas the PTV2 of p5, p8 covered to supraclavicular nodes, so the PTV2 of p5 and p8 was influenced by S and I directions. The mean slopes of the PTV2 ΔD 95% lines were −3.41% ± 0.31% per cm for IMRT and −4.08% ± 0.22% per cm for VMAT in F direction expanding.
The ΔD 95% for VMAT was overall larger than that for IMRT (P < 0.05), which is also mainly because in VMAT plans the dose is delivered by two full arcs with multileaf collimator modulation and the IMRT plans the dose is delivered in a 9-beam arrangement. Whether the body contour change affects the PTV3 or not depends on the position of PTV3. As shown in Figure 4 PTV3, the ΔD 95% loss was seen in the C7-T2 region in p1 and p2, and the ΔD 95% loss was seen in the p4, p6: without PTV2; p2, p3, p5, p9: without PTV4.

OARs Sparing
In addition to decreasing the dose to the target volumes, body contour change also has the potential to increase the dose to OARs. The change in dose to the bladder, rectum, small bowel for cervical patients and brain stem and the spinal cord for NPC patients in each direction change was evaluated in Eclipse for both IMRT and VMAT plans.
For the ten selected cervical patients, the rectum, bladder and small bowel were investigated from the dose-volume histogram. Figures 5(a)   For the ten selected NPC patients, the D max change to the brain stem and spinal cord was displayed in Figure 8. The ΔD max of the brain stem of all of the examined dosimetric cases was up to 195 cGy for IMRT and 210 cGy for VMAT in F-1 direction, and the ΔD max of the spinal cord increased by up to 209 cGy for IMRT and 628 cGy for VMAT in S-2 direction.

Discussion
Accurate dose delivery to target volumes and organs at risk (OARs) is essential to ensure the radiotherapeutic effect and minimize the risk of normal tissue toxicity, whereas weight fluctuation frequently occurs during the whole radiotherapy course, which may cause the body contour changes to effect on treatment accuracy. Therefore, we evaluated the impacts of body contour changes to target volumes and OARs in the radiotherapy plans of cervical cancer and head and neck tumors. A course of RT is typically several weeks. During the long period from the CT scans to RT finish, the weight change for abdominal and H&N patients was reported in previous literature [13] [14] [15] [16]. There is evidence showing that weight change is correlated with external contour changes [17].
Booth et al. [18] reported that 68% of the 198 analyzed CBCT images from 19 prostate patients were in the range of 0 -1 cm, 28% 1 -2 cm, and 4% > 2 cm with deviations occurring mostly in the postero-lateral direction. Chow et al. [13] studied body contour shrunk by 0.5, 1, 1.5, 2 cm in anterior, left and right direction for five patients' IMRT and VMAT prostate plans. Sun et al. [7] reported that prostate patients who have body contour change less than 2 cm at a single side or less than 1 cm uniformly are unlikely to need further assessment.
For H&N patients, Chen et al. [14] reported 25 NPC patients shrank the external contours with different margins (2, 3, and 5 mm). Our study illustrated the de- shoulder based on the previous study by Sun et al. [7] and Neubauer et al. [19]. Neubauer et al. [19] examined ten patients and 243 CTs, and found that 2% of shoulder shifts were greater than 1 cm. Noble et al. [1] measured lateral neck diameter which is 175 mm on the first day and 162 mm on the final treatment day. In our study, we chose shoulder changed 1 cm and 2 cm in S-I directions and 0.5 cm and 1 cm in the facial area. Chen et al. [14] shrank external contour with different margins (2, 3, and 5 mm) and reported that the D 95% of PTV1 was increased by 1.9% to 2.9%, which was similar to our result 1.95% ± 0.58%/5mm.
The PTV3/4 in Figure 4 shows relative irregular changes for D 95% compared with PTV1/2. The PTV3/4 extended outside of the shrunk body contour, so the volumes within the new contour were affected by the build-up effect. Zhao et al. [20] found an increase in the maximum dose to the spinal cord and brainstem volumes of 560 cGy and 250 cGy, respectively, by comparing repeat CT imaging to the dose distribution on the original planning CT. Wang et al. [21] reported the NPC repeated CT scan after 18 fractions, the mean volume of the left and right parotid decreased 6.19 mL and 6.44 mL, and the center of C2 vertebral body slices contracted with the mean contraction of 8.2 mm, 9.4 mm, and 7.6 mm while the maximum dose to the brain stem and spinal cord increased by 0.08 to 6.51 Gy and 0.05 to 7.8 Gy. Our study showed that the ΔD max of the brain stem of all of the examined dosimetric cases was up to 195 cGy for IMRT and 210 cGy for VMAT in F-1 direction, and the ΔD max of the spinal cord increased by up to 209 cGy for IMRT and 628 cGy for VMAT in S-2 direction. The dosimetrists need to take the dosimetric changes into account during the RT plan design.
The factors which affect the dose to target volumes and OARs were anatomy and setup error, this paper studied dose discrepancy which resulted from the anatomy change. The limitation of this study was that the location, geometry, and size of the tumor, OARs may change during the RT period. Moreover, the patients' body contour change may not be as regular in real clinical status. These situations were not discussed in this study. The body contour changed methods in this work have been widely reported in previous literature [4] [7] [8] [13] [14]. It is a simplified way but good choice to predict such complicated dosimetric problems. Therefore, the RT staff could make a preliminary judgment of dosimetric parameters induced by body contour changes for cervical cancer and NPC patients based on the findings throughout this work.

Conclusion
The dosimetric evaluation of body contour changes to PTV and OARs for cervical cancer and NPC plans was studied. The body contour shrinkage or expan-