Ninjaflex vs Superflab: A comparison of dosimetric properties, conformity to the skin surface, Planning Target Volume coverage and positional reproducibility for external beam radiotherapy

Abstract Background and purpose When planning and delivering radiotherapy, ideally bolus should be in direct contact with the skin surface. Varying air gaps between the skin surface and bolus material can result in discrepancies between the intended and delivered dose. This study assessed a three‐dimensional (3D) printed flexible bolus to determine whether it could improve conformity to the skin surface, reduce air gaps, and improve planning target volume coverage, compared to a commercial bolus material, Superflab. Materials and methods An anthropomorphic head phantom was CT scanned to generate photon and electron treatment plans using virtual bolus. Two 3D printing companies used the material Ninjaflex to print bolus for the head phantom, which we designated Ninjaflex1 and Ninjaflex2. The phantom was scanned a further 15 more times with the different bolus materials in situ allowing plan comparison of the virtual to physical bolus in terms of planning target volume coverage, dose at the prescription point, skin dose, and air gap volumes. Results Superflab produced a larger volume and a greater number of air gaps compared to both Ninjaflex1 and Ninjaflex2, with the largest air gap volume of 12.02 cm3. Our study revealed that Ninjaflex1 produced the least variation from the virtual bolus clinical goal values for all modalities, while Superflab displayed the largest variances in conformity, positional accuracy, and clinical goal values. For PTV coverage Superflab produced significant percentage differences for the VMAT and Electron3 plans when compared to the virtual bolus plans. Superflab also generated a significant difference in prescription point dose for the 3D conformal plan. Conclusion Compared to Superflab, both Ninjaflex materials improved conformity and reduced the variance between the virtual and physical bolus clinical goal values. Results illustrate that custom‐made Ninjaflex bolus could be useful clinically and may improve the accuracy of the delivered dose.


| INTRODUCTION
Bolus materials are conventionally used in radiotherapy practice to alter the delivered dose to the skin surface and compensate for irregular patient contours. Naturally or synthetically developed materials have been used such as wet gauze, wax, and vinyl gels among others. 1 Synthetic gel-type commercial bolus for example, Superflab (Civco, Orange City, IA, USA), is in common use owing to its tissue equivalency (quoted density of 1.02 g/cm 3 ) and being latex free. In clinical practice the positioning of bolus should be reproducible, and must maintain its shape and properties throughout the course of treatment. 2 Direct contact of bolus with the surface of the skin is ideal as this is perceived to be more efficient by increasing the dose to superficial tissues and by improving dose uniformity. If bolus has not been applied closely to the skin surface then variations in air gaps during treatment may lead to a discrepancy between intended and delivered dose. 3,4 Kong and Holloway 5 found that for electron beams the impact of air gaps was dependent on field size, beam energy, air gap size, and bolus thickness.
For a 3 cm diameter circular field, 6 MeV beam, 20 mm air gap, and 15 mm bolus, both the maximum dose and surface dose were reduced by approximately 60%, and the depth of the dose maximum shifted by 3.5 mm. They recommended that air gaps should be avoided to improve the accuracy of treatment delivery. Butson et al. 6 assessed the impact of air gaps for 6 MV beams using field sizes of 8 × 8 cm and 10 × 20 cm. They found that small air gaps (<10 mm) slightly decreased the surface and skin dose, but still allowed for at least 90% of the maximum dose being delivered to the skin regions.
Investigating alternative bolus materials for radiotherapy treatment may help to improve the accuracy of treatment delivery and patient outcomes. A promising tool that is enabling significant developments in radiotherapy is the three-dimensional (3D) printer. 3D printing provides scope for printing out the exact patient surface contour, incorporating any unique indentations, and thereby accounting for individual differences.
A recent study by Park et al. 7 investigated the use of patientspecific breast bolus using 3D printed polylactic acid (PLA) bolus compared to their currently used Super-Flex bolus. The 3D printed bolus was created from a computed tomography (CT) scan of a breast phantom. Treatment plans were generated to assess the effect of unwanted air gaps between the bolus and phantom's surface on the dose distribution. The results showed that 3D printed solid bolus reduced variation in the daily setup and helped to overcome the dose discrepancy resulting from unwanted air gaps, leading to a more accurate treatment. It was concluded that 3D printed bolus could replace the currently used commercial boluses. Robar et al. 8 analyzed the use of 3D printed PLA bolus for patients receiving chest wall radiotherapy compared with standard sheet bolus.
Cone beam CT scanning was used to quantify the accuracy of fit with regards to air gaps between each type of bolus and the skin surface. For the sheet bolus, approximately 30% of all fractions involved air gaps of more than 5 mm, compared to 13% for the 3D printed bolus. They concluded that the accuracy of fit was improved significantly with the 3D printed bolus.
A literature review by Pugh et al. 9 found that the improved conformity of 3D printed bolus could prove advantageous for volumetric modulated arc therapy (VMAT) and intensity-modulated radiotherapy (IMRT) techniques as the presence of air gaps, small field sizes, and large beam obliquity can result in a reduction of 10% in the dose at the skin surface.
Recent studies have assessed the dosimetric properties and use of rigid, solid 3D printed plastics for boluses. [9][10][11][12] The disadvantage of these materials is their lack of flexibility. Ninjaflex, however, is a lightweight, flexible material that is rigid enough to hold its shape following printing. 13 Robar et al. 11 assessed the use of Ninjaflex as a bolus compared to standard sheet and 3D printed PLA bolus using a chest wall phantom. They found that both types of 3D printed boluses improved spatial conformity to the chest wall and that this improvement appeared to create a more uniform surface dose.
The aim of our study was to evaluate the suitability of 3D printed Ninjaflex bolus for external beam radiotherapy compared with the department's standard bolus material, Superflab, by assessing dosimetric properties, conformity to the skin surface, planning target volume (PTV) coverage, and reproducibility of setup against a virtually created bolus produced within the treatment planning system (TPS).

| MATERIALS AND METHODS
The RANDO head phantom (The Phantom Laboratory, USA) was used for the evaluation as it was considered a difficult test surface to avoid air gaps between the bolus and skin surface owing to its contour. 6 Internally the head phantom has the skeletal bone structure of a human skull; nasal, oral, and trachea air cavities; and teeth, all other tissue has a density of 1.00 g/cm 3 . An initial CT scan, using a Canon Aquilion Large Bore Multi Slice CT scanner (Canon Medical Systems Ltd, UK), was acquired and the image files were sent to two different 3D printing companies to generate the 3D printed Ninjaflex bolus. Ninjaflex is a thermoplastic polyurethane (TPU) material that is lightweight, promoted for its flexibility and longevity but is rigid enough to hold its shape following printing, 13 suggesting that it could be an ideal material to create bespoke bolus. Company1 used a Ultimaker S5 to create Nijaflex1 while Company2 used a Lulzbot Taz 5 printer to create Ninjaflex2.
The generated Ninjaflex bolus had a depth of 5 mm, and was designed to cover the right-hand side of the head phantom including the nose, lips, chin, mandible, submaxillary triangle, extending posteriorly around the neck to the mastoid process and inferiorly to the thyroid cartilage ( Figure 1). Each was requested to be 100% infill to minimize any density differences throughout the material.
The attenuation properties of the Ninjaflex material compared to

| RESULTS
The density of the phantom bolus was measured using the TPS and was found to be 0.96 g/cm 3 for Ninjaflex1 and 1.02 g/cm 3 for Nin-jaflex2. The physical properties of each 5 and 10 mm square bolus sheet were measured to verify the density ( Table 1). The 10 mm 3D printed Ninjaflex1 sheet was found to have a lower density than expected, which was apparent by the observed physical weight and feel of the bolus. Air gaps between the bolus and skin surface were analyzed on each CT and found to have statistically significant difference (X 2 (2) = 11.601, P < 0.05), with a mean rank of 12

| DISCUSSION
This study aimed to assess the feasibility of improving radiotherapy treatment by using a 3D printable flexible material for bolus. This the Bolus Virt demonstrating that it too could be used as a more superior bolus material. Clinically, this would mean plans could be produced using Bolus Virt and using the higher software version (Raystation ver. 7 or above) the bolus data can be exported to the F I G . 4. Summarizes the distribution in data between the different boluses and clinical goals depicting the median, first and third quartile, whiskers that extend no more than 1.5 times the interquartile (IQ) range and outliers whose values are between 1.5 and 3 times the IQ range. The numbered outliers refer to the Kruskal-Wallis test ranking value.
3D printer, resulting in improved conformity and positioning of the

| CONCLUSION S
We have tested the feasibility of using 3D printed Ninjaflex bolus as a substitute and improvement to our current standard, Superflab.
We did this by assessing the dosimetric properties, conformity to the skin surface, PTV coverage, and reproducibility of setup against a virtually created bolus produced within the TPS. We found that Ninjaflex had excellent conformity and reduced the variance between the resultant virtual and physical bolus clinical goal values.
The clinical user experience revealed that Ninjaflex was easier to position and setup compared with the Superflab as the Ninjaflex was 'patient' specific and conformed exactly to the surface of the phantom. There may also be infection control advantages in certain situations using a bolus that is patient specific. There are, however, cost considerations for single-patient use bolus, although it may also be a preferred option for patients given the infection control benefits. Customized 3D printed Ninjaflex bolus could be used clinically and it may be that 3D printed bolus should be considered for certain types of treatment that appear to be more affected by the presence of air gaps, for example, electron therapy.

CONFLI CT OF INTEREST
Dr. Adamson reports other from NHS Tayside, during the conduct of the study; other from GSK, grants from Roche and Boehringer Ingelheim, other from Roche, outside the submitted work. Ms.
Robertson reports grants from NHS Tayside, during the conduct of the study; other from GSK outside the submitted work.