A Retrospective Analysis of the Impact of Head and Neck Immobilization Devices on

Target Volumes and Organs at Risk with Various Treatment Planning System Algorithms

Nick Hopkins, BS, RT(T); Savannah Coleman, BS, RT(T); Jacquelyn Palermino, BS, RT(T);

Julie Puzzonia, BS, RT(T); Ashley Hunzeker, MS, CMD; Nishele Lenards PhD, CMD, RT(R)

(T), FAAMD; Alyssa Olson, MS, RT(T), CMD

ABSTRACT

Immobilization devices are essential in radiation therapy aiding in setup reproducibility and

limiting patient movement. One treatment site that is highly dependent on the use of

immobilization for radiotherapy includes head and neck (HN) cancers. There is limited literature

on the effects on immobilization in dose calculations for volumetric modulated arc therapy

(VMAT) planning. A retrospective study was performed to determine the dosimetric impact of

HN immobilization devices on the planning target volume (PTV) and organs at risk (OAR)

coverage using the Eclipse treatment planning system (TPS) with Anisotropic Analytical

Algorithm (AAA) and Acuros XB (AXB), along with Pinnacle TPS using Collapsed Cone

Convolution (CCC). This study retrospectively collected data from the treatment plans of 21 HN

patients. Comparison plans were created from the original treatment plan to either include or

exclude immobilization devices into the dose calculations depending on the TPS. For Pinnacle

systems, a contour was created for the HN immobilization, assigned a density equivalence of air,

and was excluded from dose calculations. For Eclipse planning, immobilization devices were

contoured as part of the body contour for consideration in dose calculations. All other planning

parameters remained the same and beams were re-calculated to reflect the effects of

immobilization. Evaluations of the treatment plans showed that PTV coverage decreased when

immobilization was factored into dose calculations. There was a notable difference when

assessing the percentage of the PTV that received 100% of the prescription dose (V100), with a

mean change in coverage from 91.4% to 74%. In addition, a reduction of dose to OAR tested

significant (p<0.05) between plans for the cord, brainstem, mandible, and left parotid. The right

parotid did not show significance (p=.09) when evaluated in this study. The skin dose to a

volume of 1 cc (D1.0 cc) and 0.03 cc (D0.03 cc) was assessed showing a mean percent increase of 5.62%

and 3.15% respectively. The plans for all 3 treatment planning algorithms displayed consistent

trends of decreased dose to the PTV and OAR, and increased dose to the skin with the inclusion

of immobilization devices in dose calculations. This study supports previous research regarding

the impact of immobilization devices on decreasing PTV coverage and increased skin dose due

to bolus effect. The results were displayed using 3 different treatment planning algorithms

suggesting that additional care should be taken in planning to consider immobilization devices

during dose calculation.

Keywords: Immobilization Devices, Radiotherapy, Attenuation, Head and Neck, VMAT

Introduction

Photon beam attenuation occurs any time a material with sufficient density intersects the

path of a radiation beam. There is no industry standard currently established accounting for

immobilization devices in contouring or in dose calculations.1 When delivering radiation therapy

treatment plans, immobilization devices in the path of the radiation beam could alter dose

distributions and decrease target coverage. Many institutions have implemented couch models to

incorporate photon beam attenuation when treating with posterior beam angles however,

immobilization devices often remain unaccounted for in dose calculations.

Immobilization devices are essential in radiation therapy aiding in setup reproducibility

and limiting intrafractional patient movement. One treatment site that is highly dependent on the

use of immobilization for radiotherapy includes HN cancers. Common devices used for

treatment in this anatomic region include head holders, thermoplastic masks, and table top

extensions/overlays. Previous studies have shown that immobilization devices attenuate a portion

of the beam and decrease skin sparing.2-6

While most current literature compares the effect of immobilization devices on intensity

modulated radiation therapy (IMRT) and three-dimensional (3D) based planning, Olson et al7

showed a statistically significant impact on PTV coverage due to HN immobilization devices

using VMAT alone. However, limitations of the study included using a single calculation

algorithm in the TPS without evaluation of dose to OAR. Thus, the goal of this retrospective

study was to determine the dosimetric impact of HN immobilization devices on PTV and OAR

coverage using the Eclipse TPS with AAA and AXB, along with the Pinnacle TPS using CCC.

Methods and Materials

Patient Selection

All patients selected for this study were diagnosed with cancer of the HN region. The

patients were prescribed definitive radiation doses between 60-70 Gy to the primary tumor. The

patients were selected from 3 different clinical institutions consisting of 8 patients with tonsillar

cancer, 5 patients with base of tongue (BOT) cancer, 3 patients with oropharyngeal cancer, 1

patient with laryngeal cancer, 1 patient with cutaneous squamous cell metastasis, 1 patient with

floor of mouth cancer, 1 patient with right parotid cancer, and 1 patient with an unknown

primary of squamous cell cancer. Each clinical site contributed 7 patients to this study, allowing

for an equal distribution of patients planned with each treatment planning algorithm. All

retrospective cases chosen were planned using a VMAT technique; plans generated using a non-

VMAT treatment technique were excluded from this study. In addition, patients with primary

gross tumor volume (GTV) involvement of the skin surface were excluded. Therefore, treatments

requiring the need for bolus were omitted from this study.

All patients in this study underwent a simulation procedure in which a CT scan was

obtained for treatment planning. The patients were simulated in the supine head-first treatment

position with the scanning field of view (FOV) to include the entirety of the immobilization

devices in the target area. The patients simulated at clinical site 1 included the use of the

following immobilization devices during simulation: CIVCO Type-S standard perforated

thermoplastic mask, CIVCO Type-S tabletop overlay board, QFix head holder, and CIVCO

AccuForm cushion. The patients simulated at clinical site 2 utilized the QFix Curve Board, S-

Frame perforated thermoplastic mask, and for some patients a CIVCO AccuForm cushion

(Figure 1). Finally, patients simulated at clinical site 3 were immobilized using CIVCO Type-S

head-only perforated thermoplastic mask, CIVCO Type-S tabletop overlay board, QFix

Silverman head holder, and for some patient's a CIVCO AccuForm cushion.

Contouring

The CT scan obtained from simulation was used for the contouring of normal structures

and target volume delineation. Treatment volumes such as the GTV, clinical target volume

(CTV), and PTV were delineated in the TPS by the physician. For the purpose of this study, the

PTV was specific to the volume encompassing the primary tumor bed.

The OAR evaluated in this study included the spinal cord, brainstem, mandible, skin, and

right and left parotid glands. The skin was defined as a 3 mm interior rind that was created from

the external surface contour. All normal tissue contours were delineated by the medical

dosimetrist and approved by the radiation oncologist. Any OAR that were missing from the

original data set were contoured for the completeness of data analysis. An exception to this was

when the primary tumor invaded the OAR, in which case, the dose to that structure was

excluded.

To assess the effect of immobilization devices in the planning process, specific contours

were created depending on the TPS utilized. The Eclipse TPS ignores any CT data that is outside

of the body contour with the exception of contours assigned a “support” type. Each support

contour also requires a single manually assigned density value for dose calculations. For the

purposes of this study in Eclipse, the body contour was extended to include all immobilization

devices which allowed the TPS to calculate dose to all CT densities within the body contour

(Figure 2). In the Pinnacle TPS, all CT densities beyond a specified tissue-air threshold are

automatically assessed and used in dose calculation. Clinically, this threshold is often set too

high (0.6 g/cm3) to recognize the lesser densities of immobilization devices. For this study,

assessments were performed with the threshold set to 0.0 g/cm3 to incorporate all densities within

the treatment scan.. A contour separate from the patient was created to encompass all

immobilization devices within the treatment area (Figure 3). This contour was overridden to the

density equivalence of air to be comparable to the Eclipse TPS plans without the extended body

contour.

Treatment Planning

All patients in this study were treated with a curative intent, using treatment doses

ranging from 60-70 Gy. There were 12 patients treated to a definitive dose of 70 Gy, 3 treated to

a dose of 69.96 Gy, 2 treated to a dose of 68 Gy, 2 treated to a dose of 66 Gy, and 3 treated to a

dose of 60 Gy. The patients were treated using a standard fractionation schedule with daily

treatments of 5 days per week.

The TPS used to carry out the dosimetric calculations included Varian Eclipse and

Philips Pinnacle. The treatment plans were generated using a VMAT technique consisting of 2-3

arcs with the smallest arc spanning 197° and the largest arc spanning 358°. All plans created

used a photon energy of 6 MV, with delivery carried out on the Elekta Agility at clinical site 1,

Varian Novalis or Varian TrueBeam at clinical site 2, and Varian Edge at clinical site 3.

Throughout the initial planning process, plans were optimized to achieve target coverage

goals while considering normal tissue constraint recommendations as proposed by published

sources such as Quantitative Analyses of Normal Tissue Effects in the Clinic (QUANTEC) and

Radiation Therapy Oncology Group (RTOG) protocols.8 The immobilization devices were then

introduced into the dosimetric calculation process to reflect either the addition or subtraction of

immobilization devices for the corresponding TPS. Throughout this process, plan normalizations

were kept consistent and beams were re-computed to assess any changes in target coverage and

dose to OAR.

Plan Comparisons

The process of plan evaluation was dependent on the TPS used for each patient. With the

Eclipse TPS, the PTV coverage and dose to OAR were evaluated first from the initial treatment

plan without the inclusion of immobilization devices. The plan was then copied, and the body

contour was extended to include all HN treatment devices. The beams were then re-computed

using fixed monitor units (MU) to ensure consistent normalization and dose to targets and OAR

were analyzed in the comparison plan.

In the Pinnacle TPS, the initial plan represented the inclusion of the immobilization

devices because the planning system detects all densities within the scan above the tissue-air

threshold. The treatment devices were then contoured, and the CT density of this contour was

assigned to 0 g/cm3, air equivalence, to simulate the exclusion of the devices. The beams were

then re-computed maintaining the same normalization and MU to achieve the comparison plan.

During plan evaluation, the spinal cord, brainstem, mandible, and right and left parotid

glands were assessed using the mean dose (Dmean) and the D0.03 cc with and without the inclusion

of immobilization devices. For the skin, D1 cc and the D0.03 cc were reported with and without

inclusion of immobilization devices. With regard to target coverage, the percent volume of the

PTV that received 95% of the prescription dose (V95), and the V100 were analyzed with and

without the inclusion of immobilization devices. To further evaluate the changes between plans,

the percent difference of each OAR and PTV metric was computed. This information was then

used to derive mean values and mean percent differences from the data collected. These

additional calculations allowed for data classification based on metric or TPS used.

Results

Target Coverage

The data obtained from this study showed a decrease in PTV coverage when

immobilization devices were accounted for in dose calculations. When evaluating the PTV V95,

the mean value of all cases decreased from 99.3% to 98.2% after the plans were recomputed to

account for the attenuation of the immobilization devices. When analyzing the mean value of the

PTV V100, the effects of immobilization were even more apparent with a change in coverage

from 91.4% to 74.0% (Figure 4). Furthermore, the mean of the percent difference was calculated

for the PTV V100, which showed a significant change of 45.0% between plans.

Organs at Risk

When evaluating the specific OAR in both plans, the data collected showed a decrease in

dose when immobilization was included in planning. The mean of the percent difference ranged

from 0.28% to 2.50% between plans (Figure 5). To ensure that the results were not due to

random error, a repeated measures T-Test was performed. This test found that all OAR presented

a significant change between plans (p <.05), except for the right parotid gland (p=.09).

Unlike the other OAR, the skin showed an increase in dose when including

immobilization devices into the dose calculation and was thus analyzed further (Figure 6). The

mean of the percent difference for the D1.0 cc was a 5.62% increase in dose and the D0.03 cc showed

a 3.15% increase in dose between plans. A repeated measures T-Test also confirmed that these

findings were significant (p<.05) between plans.

Planning Algorithms

This study gathered data using 3 different treatment planning algorithms including

Pinnacle’s CCC, and Eclipse’s AAA and AXB. The results showed a consistent trend across all

planning algorithms with an increase in skin dose when including immobilization devices in the

dose calculations. For example, the percent difference of D1.0 cc of the skin showed increases in

plans ranging from 0.88%-8.28% using CCC, 7.97%-12.71% using AAA, and 2.72%-7.56%

using AXB. In addition, all planning algorithms showed a trend of decreased PTV coverage and

dose to OAR when evaluating the mean of the percent difference between plans with and without

immobilization (Table 1).

Discussion

The results of this study were consistent with the findings of Olson et al7 suggesting that

the attenuating effects of immobilization have an impact on PTV target coverage. In addition,

these findings were confirmed using a larger sample size and were consistent when tested across

3 different treatment planning algorithms. This suggests that there is a deficit in treatment

planning practices across multiple systems in accounting for immobilization devices in dose

calculations.

This study also showed a reduction in dose to OAR when accounting for immobilization

devices in dose calculations. The inaccuracies of dose reporting to OAR could impact treatment

planning in numerous ways. For example, in HN planning, trade-offs often take place as the

medical dosimetrist tries to maximize PTV coverage and minimize dose to OAR. The reduced

dose to OAR from the inclusion of immobilization devices could indicate that there was more

room to optimize coverage to the PTV and enhance treatment outcomes.

Furthermore, this study also considered skin dose which is a topic of interest with regards

to VMAT treatment techniques. Previous studies suggest up to an 8% higher skin dose with a

marginally higher significance of Grade ≥ 2 acute skin toxicities when using a VMAT treatment

technique vs. IMRT in HN patients.9 Other studies report a bolus effect from thermoplastic

masks in HN cancer with an increase in the average skin dose by 18%.5 Although the previous

studies differ by reporting measurements of skin dose using thermoluminescent dosimeters, the

results are aligned with the findings of this study indicating an increase in skin dose due to HN

immobilization in VMAT planning. Therefore, it is suggested that care be taken in VMAT

planning with considerations of limiting skin toxicity.

Conclusion

The presence of immobilization devices in dose calculations influences PTV coverage,

OAR, and skin dose. The decrease in PTV coverage and dose to OAR can be attributed to the

increased attenuation of the photon beam when considering immobilization. For PTV coverage,

the greatest effects were seen when assessing the V100, which may be attributed to differences in

dose homogeneity within the PTV. For OAR, although the mean of the percent difference did not

suggest a large change, all changes tested statistically significant with the exception of the right

parotid.

This study also supports the reported increase in skin dose due to the inclusion of

immobilization devices in dose calculations. Multiple studies have shown the potential for a

bolus effect with the use of immobilization devices related to 3D conformal and IMRT

planning.3,5-6 This study continues to support the literature and, in addition, confirms the need for

considerations of skin toxicity in VMAT planning.

Finally, this study observed the effects of immobilization across 3 different treatment

planning algorithms which has not been addressed in current literature. All planning algorithms

used in this study supported previous conclusions recommending the standardization of

immobilization device inclusion in treatment planning. Further research may be completed to

evaluate the specific differences in dose calculations using one data set across multiple

algorithms with and without immobilization devices.

References

1. Olch AJ, Gerig L, Li H, Mihaylov I, Morgan A. Dosimetric effects caused by couch tops and

immobilization devices: Report of AAPM Task Group 176. Med Phys. 2014;41(6):61501–

61530. http://dx.doi.org/10.1118/1.4876299

2. Munjal RK, Negi PS, Babu AG, et al. Impact of 6MV beam attenuation by carbon fiber

couch and immobilization devices in IMRT planning and dose delivery. J Med Phys.

2006;31(2):67–71. http://dx.doi.org/10.4103/0971-6203.26690

3. Seppälä JKH, Kulmala JAJ. Increased beam attenuation and surface dose by different couch

inserts of treatment tables used in megavoltage radiotherapy. J Appl Clin Med Phys.

2011;12(4):15-23. http://dx.doi.org/10.1120/jacmp.v12i4.3554

4. Dieterich S, Ford E, Pavord D, Zeng J. Immobilization techniques in radiotherapy. In:

Dieterich S, Ford E, Pavord D, Zeng J, ed. Practical Radiation Oncology Physics.

Philadelphia, PA: Elsevier, Inc; 2016:87-94.

5. Lee N, Chuang C, Quivey JM, et al. Skin toxicity due to intensity-modulated radiotherapy for

head-and-neck carcinoma. Int J Radiat Oncol Biol Phys. 2002;53(3):630-637.

http://dx.doi.org/10.1016/S0360-3016(02)02756-6

6. Pashkovskaya OA, Bedny IV, Anikeeva OY, Polovnikov ES. The evaluation of skin toxicity

during brain tumor irradiation dose calculation. Int J Biomed. 2013;3(4):283-286.

https://elibrary.ru/item.asp?id=20922807. Accessed April 24, 2018.

7. Olson A, Phillips K, Eng T, et al. Assessing dose variance from immobilization devices in

VMAT head and neck treatment planning: A retrospective case study analysis. Med Dosim.

2018;43(1):39-45. http://dx.doi.org/10.1016/j.meddos.2017.08.001

8. Marks LB, Yorke ED, Jackson A, et al. Use of Normal Tissue Complication Probability

Models in the Clinic. Int J Radiat Oncol Biol Phys. 2010;76(3):S10-S19.

http://dx.doi.org/10.1016/j.ijrobp.2009.07.1754

9. Bredfeldt J, Sapir E, Masi K, Schipper M, Eisbruch A, Matuszak M. TH-EF-BRD-11:

Clinical Skin Toxicity Comparison and Phantom Dose Measurements for Head and Neck

Patients Treated with IMRT vs. VMAT. Med Phys. 2015;42(6):3742-3742.

http://dx.doi.org/10.1118/1.4926298

Figures

Figure 1. Patient positioning using a QFix Curve Board, thermoplastic mask and AccuForm headrest.

Figure 2. Contour representation of body (green) including HN immobilization devices in Eclipse TPS.

Figure 3. Contour (purple) used for immobilization devices separate from body contour in Pinnacle TPS.

PTV V95 PTV V1000.0%

20.0%

40.0%

60.0%

80.0%

100.0%

120.0%

99.3%91.4%

98.9%

74.0%

Mean PTV Coverage

Mean value PreMean value Post

Dose-Volume Objective

Perc

ent o

f PT

V v

olum

e (%

)

Figure 4. The mean PTV coverage for plans without immobilization devices (pre) vs. plans with immobilization devices (post).

Parotid

L D0.0

3cc

Parotid

L Mean

Parotid

R D

0.03c

c

Parotid

R M

ean

Spinal

Canal

D0.03c

c

Spinal

Canal

Mean

Brains

tem D

0.03c

c

Brains

tem M

ean

Mandib

le D0.0

3cc

Mandib

le M

ean0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

3.00% 0.00E+001.00E-022.00E-023.00E-024.00E-025.00E-026.00E-027.00E-028.00E-029.00E-021.00E-01

Organs at Risk

Mean Percent Difference Significance Value (p)

Organ at Risk

Mea

n Pe

rcen

t Diff

eren

ce (%

)

P-va

lue

Figure 5. The mean percent difference and significance value between plans with and without immobilization devices for organs at risk.

CC

CA

AA

AxB

Ave

rage

-12.00% -10.00% -8.00% -6.00% -4.00% -2.00% 0.00%

-3.17%

-9.65%

-4.03%

-5.62%

-1.17%

-6.24%

-2.03%

-3.15%

Skin D0.03 cc Skin D1.0 cc

Mean Percent Difference

Cal

cula

tion

Alg

orig

htm

Figure 6. Increase in skin dose for patients using the treatment planning algorithms CCC, AAA, and AXB.

Tables

CCC Mean %diff AAA Mean %diff AxB Mean %diffParotid L

D0.03 cc 2.84% 1.44% 1.95%Dmean 1.88% 1.12% 0.15%

Parotid RD0.03 cc 1.55% 1.36% 1.91%Dmean 0.40% 0.11% 0.33%

Spinal CanalD0.03 cc 1.94% 1.48% 1.97%Dmean 1.59% 1.37% 2.00%

BrainstemD0.03 cc 2.05% 1.79% 1.40%Dmean 1.20% 0.90% 0.37%

MandibleD0.03 cc 2.06% 1.31% 2.90%Dmean 2.47% 2.25% 2.78%

PTVV95 0.57% 0.00% 0.52%V100 14.69% 16.32% 104.10%

Table 1. Mean percent difference reported for PTV and OAR based on TPS algorithm used.