Dosimetric effect of body contour changes for prostate and head and neck volumetric modulated arc therapy plans

Abstract Body contour changes are commonly seen in prostate and head and neck (H&N) patients undergoing volumetric modulated arc therapy (VMAT) treatments, which may cause a discrepancy between the planned dose and the delivered dose. Dosimetrists, radiation oncologists or medical physicists sometimes are required to visually assess the dosimetric impact of body contour changes and make a judgment call on whether further re‐assessment of the plan is needed. However, an intuitive judgment cannot always be made in a timely manner due to the complexity of VMAT plans as well as the complicated forms of body contour changes. This study evaluated the dosimetric effect of body contour changes for prostate and H&N patients to help with clinical decision‐making. By analyzing the one‐dimensional spatial dose profiles from the original body and the body with different body contour deformations, rules of thumb for dose percentage change and isodose line shift due to body contour changes were ascertained. Moreover, based on dose distribution comparison using three‐dimensional gamma analysis, the response of the clinical prostate and H&N VMAT plans to body contour changes was assessed. Within center specific dose deviation tolerances, prostate patients who had less than 2 cm single side body contour change or less than 1 cm uniform body contour change were unlikely to need plan re‐assessment; H&N VMAT plans with less than 1 cm uniform body contour change or less than 1 cm shoulder superior–inferior positional change were also unlikely to trigger further evaluation. Dose percentage change and isodose line shift were considered independently from the problem of volume changes in this study, but clinically, both aspects must be considered.

In a conventional planning process, the patient's treatment plan is created based on the anatomy present in the planning computed tomography image set (p-CT), which is typically taken 1 to 2 weeks before the start of radiation therapy. However, during the course of radiation therapy, the patient's anatomy may change in ways that cannot be corrected by image guided radiation therapy (IGRT). Examples for prostate patients include weight change, buttock flex, and abdominal position change. 3,4 Weight change, tumor shrinkage, and shoulder position variations are commonly seen in head and neck (H&N) patients. [5][6][7][8][9] As a result, the patient's body contour on the treatment day can deviate from the p-CT and this can be clearly visualized on the volumetric images (e.g., cone-beam CTs) taken on the treatment day.
Essentially, body contour changes can cause changes in beam path length, entry angle, and degree of phantom scatter in a given field. When combined with the high conformality and steep dose gradients generated with VMAT, body contour changes can potentially lead to significant differences between planned and delivered dose. In most cancer centers, it is common for changes in body contour to trigger re-assessment of the plan. 3,[10][11][12] Dosimetrists, radiation oncologists or medical physicists need to make a judgment call on further examining the dosimetric impact of body contour changes based on the cone-beam CT (CBCT) taken before treatment as well as the decision to reposition and treat the patient.
In the era of 3D conformal radiation therapy, the effect of body contour changes was commonly estimated by the tissue phantom ratio (TPR) for isocentric setups. For intensity modulated radiation therapy (IMRT), the effect could be approximated by the weighted TPRs for all the fields, which could be done on the fly. However, for VMAT plans, where the dose rate, gantry speed, and multileaf collimators' movements are changing during the 360°delivery around the patient, it is not intuitive to assess the dosimetric impact of body contour changes. One can perform full dose calculations on the CBCT image sets. This time consuming method can be problematic because of the lack of an accurate electron density conversion curve for CBCT systems and problems associated with image registration between the CBCT and p-CT. On the other hand, the number of patients who can be re-assessed and re-planned is constrained by the limitation of resources in a clinic. Thus, it is important to have some efficient ways to evaluate the impact of body contour changes on the spatial dose distribution as well as knowing how sensitive the plans are to body contour changes.
In this study, we provide rules of thumb for dose percentage change and isodose line shift due to body contour changes for prostate and H&N VMAT plans. Our analysis is based on one-dimensional dose profile comparison and 3D gamma analysis.

2.A | Patient population
Twelve early-to-intermediate stage prostate cancer patients (six prescribed with 78 Gy in 39 fractions and six prescribed with 60 Gy in 20 fractions) and ten loco-regionally advanced oropharyngeal cancer patients [70 Gy in 33 fractions to the primary gross tumor and high risk nodal regions, namely, high risk clinical target volume (CTV_H) and 59.4 Gy in 33 fractions to low-risk nodal region, namely, low risk clinical target volume (CTV_L)] were randomly selected. All of them were treated with VMAT and retrospectively analyzed.

2.B | VMAT treatment planning
For prostate plans, the CTV was defined as the prostate with or without proximal seminal vesicles; the planning target volume (PTV) was a 10 mm expansion of the CTV, 7 mm posteriorly. For the oropharyngeal plans, the high dose PTV and low dose PTV were created by a 3 mm uniform expansion of the corresponding CTVs but the PTVs were cropped 3 mm from skin.
All the clinical plans were made by the dosimetrists in the treatment planning system using the progressive resolution optimizer (Eclipse version 11.0, Varian Medical System, Palo Alto, CA) as per PROFIT trial protocol (60 Gy/20fractions), 13 departmental prostate protocol (78 Gy/39fractions), 14 and departmental H&N protocol 14 .
Each clinical plan had two to three full arcs with the energy of 6 MV and the anisotropic analytical algorithm with a dose grid resolution of 2.5 mm was used for dose calculation.

2.C | Body contour deformation
To create CT image sets with deformed body contours and to assess the theoretical impact of body contour changes, body contours in the p-CT were deformed using the margin tool in Eclipse Contouring for body contour shrinkage and expansion. For body contour expansion, the air gap between the original body and the new contour was assigned HU = 0 for simplicity. Although in reality, this part of the tissue may be mostly adipose tissue (mass density~0.9 g/cm 3 ) with typical HU in the range of −190 to −30, 15 it is unlikely to greatly affect the dose deposition. In Eclipse, dose is calculated within the body contour structure and any material outside of the body contour is treated as vacuum (with the exception of designated support structures).
The body contour change may also lead to the shrinkage or expansion (shift) of clinically relevant isodose lines (lines connecting the voxels of equal dose), which may be a concern. For example, after body contour change, the 95% isodose line may not fully cover the PTV, leading to increased risk of loco-regional recurrence, or, a larger portion of the OARs may be covered by the high isodose lines, which may not be acceptable. Therefore, isodose line shift, the

2.D.1 | 3D gamma index
The gamma index is clinically used for quantitative evaluation of the treatment planning system calculated dose distribution and the measured dose distribution using the acceptability criteria. 17     One trend in Fig. 4 is that high dose isodose lines (e.g., 95% isodose line) and relative low dose isodose lines (e.g., 60% isodose line)

3.B | Rules of thumb for H&N plans
In Fig. 5, for uniform body contour change, the medians of ΔD(%/ cm) had little variability and the variations of the medians were within 1%. Thus, it was reasonable to take the average of the medians as the rule of thumb (Table 1). This value (4%) was close to the reference value from single beam because the facial geometry was close to a cylinder and the body contour change was happening uniformly. However, the results from patients' shoulder position change had larger variations (as high as 6.5% or as low as 1.5%). This may be due to the fact that the original shoulder position on the p-CT varied tremendously between patients (the slice examined was five slices inferior to the most superior shoulder contour on the p-CT) and there may be shoulder contour variability.
In Fig. 6  One limitation of the rules of thumb for H&N plans is that these rules cannot assess the dosimetric effect in the buildup region (<1.5 cm from surface), where charged particle equilibrium is absent and the dose fall-off is very steep. In this region, the dose percentage and isodose line shift are highly dependent on the depth.

3.C | 3D gamma index
The results for the 3D gamma passing rate are shown in Table 2.

| DISCUSSION
In this study, we find that prostate patients who have body contour changes less than 2 cm at a single side or less than 1 cm uniformly are unlikely to need further assessment. This corresponds to roughly For H&N patients, we find that a uniform body contour change less than 1 cm in facial area is unlikely to warrant further assessment due to dose change. However, the anatomical changes may cause the OARs (e.g., parotid glands) to enter high dose regions, which may not be acceptable even with less than 1 cm body contour change. There is evidence showing that weight loss is correlated with body contour changes. 16,25 Weight loss for H&N patients is well-known 5-8 and body contour change due to weight loss has been quantified in some studies. Yang et al. 21 reported that the transverse diameter at the odontoid process level decreased on average by 4.6 mm from the first fraction to the 16th fraction and 7.9 mm from the first fraction to the 25th fraction for their 23 patients. Ahn et al. 25 showed that the average skin separation at the isocenter decreased by 3. In this study, we evaluated the dosimetric effect of body contour changes and found the rules of thumb for dose percentage change and isodose line shift ( (3%-4%) as those dosimetric parameters (e.g., CTV D99%, rectum and bladder D30%) mentioned above.
In this study, the rules of thumbs were derived from "controlled" cases. However, in real clinical situations, the patients' body contour change may not be as regular. For cases like those, we can apply the idea of mean surface distance to approximate the equivalent uniform body contour change and then use the rules of thumb. Figure 7 shows In the future, daily online re-planning for prostate and H&N patients in a timely fashion may be achievable. However, currently, as a result of the limited resources in busy cancer centers, it is unlikely that daily online or offline re-planning will be applied to all patients on a routine basis. The decision on flagging the plans for further assessment and potentially re-planning is mostly based on the anatomical changes seen on the pre-treatment images (body contour change as the one that can be easily visualized). Thus, it is still essential to establish ballpark dosimetric consequences that result from body contour changes.
There are some limitations of this study. Firstly, 1D dose profile on a single slice was used for establishing the rules of thumb and a limited number of isodose lines were examined. As a result, the rules of thumb may not work well for low isodose lines (e.g., <50%). Overall, the rules of thumb tend to underesti- can be useful to help the radiation oncologist/physicist/RT staff to make a quick decision on treat or not treat due to body contour changes while the patient is on bed, or to indicate the priority of the full dosimetric calculation, and as a "sanity check" when reviewing such calculations.
F I G . 7. Dose distributions on the planning computed tomography (p-CT) (a) and the synthetic CT (b, planning CT deformed to the conebeam CT on the 28th fraction). The body contour for the 28th fraction is marked on the planning CT (the orange contour) and there is significant body contour change on the left and posterior region. regions. In addition, guidelines were given for patients who underwent body contour changes but were unlikely to require plan reassessment. However, the judgment is dependent on center-specific tolerances for dose deviations.

CONF LICT OF I NTEREST
The authors declare that they have no conflict of interest.