Dosimetric impact of tracheostomy devices in head and neck cancer patients

Abstract Introduction The tracheostomy site and adjacent skin is at risk for recurrence in head/neck squamous cell cancer patients. The tracheostomy tube is an in situ device located directly over the tracheostomy site and may have clinical implications on the radiation dose delivered to the peristomal region. This study aimed to investigate this effect by comparing the prescribed treatment planning dose with the actual dose in vivo to the peristomal clinical target region. A retrospective, dosimetric study was performed with approval of the institutional research ethics board. Methods Fifteen patients who had received high‐dose radiotherapy to the tracheostomy region with in vivo dose measurements were included. The radiation dose at the skin surface underneath the tracheostomy device was measured using an optically stimulated luminescent dosimeter (OSLD) and was compared with the prescribed dose from the radiation planning system. The effect of the tracheostomy flange and/or soft tissue equivalent bolus on the peristomal dose was calculated. Results and discussion Patients with tracheostomy equipment in situ were found to have a 3.7% difference between their prescribed and actual dose. With a tissue equivalent bolus there was a 2.0% difference between predicted and actual. The mean prescribed single fraction dose (mean = 191.8 cGy, SD = 40.18) and OSLD measured dose (mean = 194.02 cGy, SD = 44.3) were found to have no significant difference. However, with the flange excluded from the planning simulation (density = air) target skin dose deviated from predicted by an average of 55.3% (range = 12.4–72.9, SD = 22.5) and volume coverage was not achieved. Conclusion In summary, the tracheostomy flange acts like bolus with a twofold increase in the skin surface dose. Changes in the peristomal apparatus from simulation to treatment needs to be considered to ensure that the simulated dose and coverage is achieved.


| INTRODUCTION
Radiotherapy is the primary treatment for many head and neck cancer patients and plays an important role in the postoperative setting for patients with locally advanced disease. When locally advanced tumors cause dyspnea, orthopnea, and stridor, patients may undergo an emergency tracheostomy procedure to protect the airway. In addition, tracheostomy is required following total laryngectomy and other radical surgeries to help manage secretions. This is clinically important as it can affect the ability to effectively deliver radiation therapy to this region. Rates of peristomal recurrence have been described between 1 and 11% and are associated with significant morbidity and mortality. [1][2][3] Accurate dose delivery to the peristomal region is a key factor in reducing peristomal recurrence.4,5 It is generally recommended that clinical target volumes (CTVs) include the stoma site and adjacent skin as these areas are at risk for locoregional recurrence in patients who had preoperative or intraoperative tracheostomy. 6,7 Although modern treatment planning systems (TPS) are reliably accurate for regions located beyond the depth of maximum dose, there remains an element of dosimetric uncertainty in the surface and build-up regions. 8 Linear accelerators emit significant levels of electron contamination (EC) that are difficult to model in a TPS that computes dose based upon kernel superposition methods. A common way to compensate for the EC problem is to empirically model the EC effect and superimpose it on the kernel superposition dose calculation. Although this empirical fit technique does improve the modeling of the surface/build-up region considerably, accurate dosimetry is still a challenge for complex beam arrangements such as those seen in intensity-modulated radiation therapy (IMRT). 9 One indication of the challenges faced when modeling the build-up region is demonstrated in AAPM's Task Group 53 report on commissioning and quality assurance of TPS. 10 In this report an example recommendation for build-up region dose accuracy is stated as 20% for square, rectangular, and asymmetric fields. Modern TPS can typically achieve accuracies of better than ±10% for the surface region (0-0.5 cm depth), and ±5% in the build-up region (0.5 cm to depth at maximum dose) for simple square fields. [11][12][13] The issues around dose uncertainty in superficial regions is of particular relevance for head and neck cancer, where the planning target volume (PTV) often encroaches upon the patient surface. 14 In the present IMRT application, it is desired to confirm that the dose received by peristomal tissue lying beneath the plastic/silicone components of a tracheostomy flange is at the desired level. Since this peristomal region is in the surface/build-up region, there is an inherent uncertainty to the dose planned by the TPS. Chung et al 15 reported a phantom study that simulated head and neck IMRT treatments for shallow (0.5 cm depth) and deep (6 cm depth) targets.
Using Pinnacle 3 as the TPS and radio chromic film as the dosimeter, there was a 5.6% and 6.5% agreement for surface dose for the shallow and deep targets, respectively.
Due to the uncertainties in TPS predictions for dose in the surface/build-up regions, in vivo dosimetry is occasionally required. This technique allows for direct measurements of the dose to ensure that the patient is exposed by the appropriate amount for the region of interest. Traditionally, TLDs, diodes, or metal oxide semiconductor field effect transistors (MOSFETs) have been used for in vivo dosimetry. Recently, dosimeters based upon optically stimulated luminescence (OSL) have been proven to be useful and increasingly popular. 16,17 The objective of taking direct OSL measurements of peri-stomatic tissue is to confirm that the bolus effect of the tracheostomy equipment in the peristomal area is adequately modeled by the Pinnacle 3 TPS.
To our knowledge the dosimetric effect of the actual tracheostomy tube and flange in situ has not been previously described.
Therefore, a dosimetric study was performed to evaluate the impact of the tracheostomy hardware on the measured dose delivered to patients and the predicted dose calculated by the TPS.

| MATERIALS AND METHODS
A retrospective, dosimetric study to assess the impact of tracheostomy hardware was performed with approval of the institutional research ethics board. All head and neck cancer patients were identified from a retrospective database and included in the study if they met several criteria. These criteria included patients who: had tracheostomy, received radiotherapy and had a physical OSL dosimeter (OSLD) measurement of the dose at the stoma site between 2013 and 2017. The dosimeter location was known to be a predetermined region associated with the highest prescribed dose from the planning distribution. This is an institutional policy that is followed for all patients.

2.A | Radiation planning and treatment
Head and neck contouring was completed by the attending radiation oncologist based on the institutional standard agreed upon for contouring of organs at risk and target volumes. 18,19 The tracheostomy site and surrounding skin were considered to be a region at risk of microscopic disease and a CTV was contoured with a PTV margin of 5 mm. The prescribed doses were determined based on institutional practice and provincial guidelines. 19 If macroscopic disease was present, the prescribed dose was 70 Gy. However, the range of pre-  site PTV received the prescribed dose, for example, V56 Gy > 95%.
The radiation treatment plans for all patients were copied to a research database for review and dosimetric analysis.

2.B | Optically stimulated luminescence (OSL) measurements in vivo
For in vivo dosimetry, a commercial OSL system was used consisting of the InLight microStar reader (Landauer, Glenwood, IL) with Landauer nanoDot dosimeters. These devices were prescreened by the manufacturer for accuracy. This system can be used to measure dose at or near the skin surface. 20,21 The OSL sensitive material is aluminum oxide with carbon impurities (Al 2 O 3 :C) encapsulated in 0.2 mm thick, 5 mm diameter discs. This sensitive material is enveloped by a lighttight plastic casing that measures 10 × 10 × 2 mm. In the first year of using OSLDs at our institution, monthly quality control tests determined that the OSL system is accurate within ±3% for therapeutic doses (approximately 10-300 cGy/fraction) These results are in line with the manufacturer's specifications and prior data. 20,21 Quality control tests were then performed on an ongoing basis to ensure that the OSLD measurements stayed within this accuracy range.

2.C | Quality assurance/verification
For any high dose head and neck radiotherapy plan, the treating radiation oncologist may request OSLD measurement for verification of the delivered dose relative to the planned dose. The institutional policy was that the measurement would not result in any treatment changes unless a discrepancy of >5% was detected and felt to be clinically significant.
An OSLD measurement was performed for one of the fractions during the treatment course for each individual patient involved in this study.
Once the patient was set up on the treatment couch, an OSLD was placed on the skin directly adjacent to the stoma. This was a predetermined region associated with the highest prescribed dose from the plan- After irradiation, the OSLDs were read out by the microStar reader after at least 10 min had elapsed. 22 Each OSLD was read out three times, and the results were averaged and then reported. These results were compared to the OSLD dose as predicted by the TPS in the patient treatment plan. A contour was then drawn to approximate the OSLD in the location that the dosimeter was placed during the treatment fraction as seen in Fig. 1. The mean dose to this OSLD was reported and compared to the OSLD reading.

2.D | Measurement of tracheostomy material density
The density of the tracheostomy hardware was calculated using CT images. This result was then compared to physical measurements of the device's density. This was done in order to ensure that the tra-

2.F | Statistical methods
The patient, tumor, and treatment characteristics (n = 15) were analyzed and summarized using descriptive statistics. The difference between the prescribed dose from the radiation plan and the measured OSLD dose was calculated as an absolute value (cGy) and as a percentage difference to normalize for the variation in the absolute prescribed doses. The mean of the differences was calculated along with the standard deviation of the differences. For each patient, a paired t-test was carried out to assess the mean difference between the OSLD measurement and the radiation dose predicted by the planning system at the peristomal region.

| RESULTS
This single institution study identified 15 patients with biopsy proven head and neck cancer. These patients had tracheostomy prior to radiation and at least one measurement of the dose received using an OSLD. A complete set of patient, tumor, and characteristic information can be seen in Table 1. Density of the tracheostomy flange was measured directly to be 1.189 ± 0.2 g/cm 3 which was comparable to the predicted value from CT planning datasets, measured as 1.168 ± 0.2 g/cm 3   For patients with bolus, the target dose and coverage also decreased when the bolus and applicator were excluded from the planning CT (density set to air equivalent). This was seen to be an average of 62.7% (SD 3.50) for patients with a bolus and an applicator and 46.8% (SD = 1.22%) for patients with a bolus but no applicator. An important consideration for these findings is the accuracy of the OSLD measurements, and the differences between the measured and planned dosages. There are a couple of factors which can cause deviation between these values. The first major cause is related to the location of the dosimetry measurements. The dosimeters were placed on the skin directly adjacent to the stoma, within a region of high-dose gradient. In this region there is the largest potential for deviation between the OSLD measurement and predicted dose value. Since there is a large gradient in this area, minor changes in positional accuracy will have large effects on the OSLD's measurement accuracy as compared to the predicted value. Second, the calculated TPS density for the applicator flap could potentially cause a deviation from the actual radiation dose delivered. However, the calculated density from the CT images of the applicator flap correlated with the measured physical density. This indicates that the applicator is constructed out of a polymer material with a low average atomic number; as such, the electron density used for the dose computation is accurate. Despite these sources of potential error, the difference T A B L E 1 Patient, tumor, and treatment characteristics.

Age
Range 40-90, Median 62 between the planned and measured doses were on average <4%.
This is within acceptable limits as described by evidence-based treatment guidelines. 19 Accurate dosimetry is important to ensure proper treatment delivery and can help to limit peristomal recurrence, particularly for head/neck squamous cell cancer patients.
As seen with the tracheostomy devices in this study, medical devices can have a significant effect on radiation treatment planning. The dosage effects of medical devices must be carefully accounted for to ensure target coverage and to avoid excessive toxicity. 24,25 An optimal radiotherapy plan needs to be able to effectively deliver the radiation dose to the targeted treatment area while minimizing dose delivery to adjacent structures. To achieve this goal the impact of any internal or external medical devices must be measured and accounted for in the radiation plan. The impact of dental implants and amalgam, intravenous ports, and breast and hip prosthesis has previously been described. 26 F I G . 4. Mean difference between the measured, planned, and predicted dose plans for patients receiving treatment with/without bolus. Dose plan refers to the original dose calculated by the pinnacle plan. Originally there is minimal deviation between the planned dosage and the measured dosages from the optically stimulated luminescent dosimeter with or without a bolus. When the trach/bolus were set to air equivalent (ρ = air) a new predicted dose was calculated by the pinnacle software, this shows the effect that a bolus or tracheostomy equipment has on the planned pinnacle dosage. For the 13 patients that did not have a bolus, setting the Trach density to air equivalent changed the calculated plan by 55%. For the three patients with a Bolus, and the two patients with a bolus and no Trach, setting the Bolus/ Trach density to air equivalent changed the plan by 62% and 46%, respectively.