Clinical history
A 34-year-old man received proton-beam radiation therapy in 2003 for a 17.0x14.0x10.5 mm melanoma involving the left choroid and ciliary body. He was treated with proton-beam therapy to a dose of 70 Cobalt Gray Equivalent, in five fractions, over ten days. In 2017, the patient was in a motor vehicle accident, which resulted in rupture of his left globe. He underwent enucleation of the left globe and was found to have recurrent melanoma. The patient healed well from surgery and a left eye prosthesis was fitted. However, over the next nine months the patient reported that the prosthesis became progressively displaced and increasingly painful to wear. A diagnostic CT revealed a heterogeneous lobular soft tissue mass in the anterior and inferior left orbit measuring 27.0x26.0x19.0 mm. He underwent salvage left orbital exenteration in March 2018. Surgical pathology confirmed multiple recurrent melanoma with a positive inferior-medial surgical margin. Restaging imaging revealed no evidence of metastatic disease and he was referred for consideration of re-irradiation.
Written informed consent was obtained from the patient for publication of this case report and accompanying images.
Applicator design
An applicator for HDR brachytherapy was designed in the AutoCAD Inventor Suite (Autodesk, San Rafael, CA) based on the latest diagnostic CT. The primary contours of interest were the patient’s surface and the PTV. The patient surface was used to generate an applicator with a flush fit against the left orbital cavity and a protruding horizontal surface 10.0 mm anteriorly from the supraorbital ridge. Support wings with a thickness of 5.0 mm were designed to extend superiorly and inferiorly by 15.0 mm and laterally by 60.0 mm. The wings were designed to be flush against the patient’s skin in order to provide a stable and reproducible fit.
The involved left orbital surfaces, including the residual mucosa and soft tissue abutting the mass found on pre-operative imaging and the sites with positive margins, were contoured as the clinical target volume (CTV; Volume~4.0 cm3) and radially expanded by 2.0 mm to generate a PTV. The PTV was used to devise channels that allowed for access and sufficient coverage of the target with the Ir-192 HDR source. The channels were constructed to fit an endobronchial HDR source guide tube with an outer diameter of 2.0 mm (Varian Medical Systems, Palo Alto, CA). While considering the size of the orbital cavity and the distance of the PTV to the cavity surface, it was found that sufficient target access would be provided when all channels were tilted 15° toward the patient’s right and 10° superiorly. For the most superior channel, patient anatomy did not allow for the 10° tilt. Under these constraints, the applicator was designed with 16 channels of varying length. The tip of the channels, corresponding to the location of the first possible dwell position, was chosen to be at 5.0 mm from the surface of the orbital cavity. Figure 1 shows the applicator geometry overlaid on patient anatomy. The material for printing the applicator, an acrylic photopolymer (Polymerized TangoPlus and Agilus30 Family, Stratasys Ltd., Eden Prairie, MN), was selected based on similarity with the biomechanical properties of human skin (Edwards and Marks 1995). Note that these materials are not approved per the International Standard ISO-10993-1 as a biocompatible material. The applicator was covered in a sterile wrap to prevent any contact with patient skin. Given the flexibility of this material, a solid sheath with a thickness of 2.0 mm was designed as an outer support-wall for each channel to prevent bending or accidental puncturing of the surface of the applicator. The assembly was created using a PolyJet 3D printer (Stratasys Ltd., Eden Prairie, MN) which allows for simultaneous printing of materials with varying physical properties. The primary applicator was designed as 80/20% mixing of polymers in the Agilus30/TangoPlus family; the sheath was 80/20% mixing of TangoPlus/Agilus30. Print time was ~20 hours, while cost was ~400$. Figure 2 presents the design of the applicator and a model of the channel sheath. Prior to use in the patient, the applicator was imaged with a helical CT scanner (120kVp, 1.0 mm3 isotropic voxels). The mean Hounsfield unit (HU) values were measured in a large region-of-interest in the applicator and found to be (mean ± std.dev.) 85±11 HU, rendering it tissue-equivalent for dose calculations.
Testing and validation
At the time of treatment simulation, the patient was immobilized supine with a custom thermoplastic mask and head holder. Serial axial CT images were obtained for treatment planning after placing the patient-specific applicator, covered in a plastic and sterile latex wrap, inside the left orbital cavity and securing it using self-adherent wrap. Treatment planning and dose calculation were performed in the Brachytherapy Planning module of Eclipse (Varian Medical Systems, Palo Alto, CA) based on the AAPM TG-43 (Rivard et al. 2004) formalism, using Ir-192 at a nominal source strength of 10 Ci. The planned prescription was 3400 cGy, to be delivered in 10 fractions, twice daily, over five consecutive days (Finger et al. 2014). Figure 3 shows the dose-volume histogram for the PTV, orbit bones, right eye, right lens, and brain.
An end-to-end quality assurance test was designed to determine feasibility for clinical use. Two calibrated pairs of optically stimulated luminesce dosimeters (OSLDs) were firmly placed on the surface of the applicator at two locations representing dose to PTV and another high-resolution CT was acquired. The clinical HDR plan was transferred to the CT containing the OSLDs in order to calculate the mean dose to the dosimeters. The applicator was then immersed in a water-filled container to mimic scattering conditions, and the clinical HDR plan was delivered. In the two sampled positions on the surface of the applicator, the mean difference in measured and calculated dose was 12% and 18%. Finally, all ten fractions of the clinical HDR plan were consecutively delivered, amounting to a dose of at least 300 Gy to the surface of the applicator, and the applicator was monitored for structural damage. No damage was found over the course of two weeks.
Treatment
A second CT simulation scan was obtained without the brachytherapy applicator for the purpose of generating an alternative stereotactic body radiotherapy (SBRT) plan. The orbital air density was assigned to water (i.e., 0 HU) in the treatment planning system to simulate a fluid filled cavity. A 4-arc volumetric modulated arc therapy plan with 6MV photons, utilizing superiorly oriented non-coplanar beams to avoid entry or exit into the contralateral eye, was created in Eclipse version 13.6 (Varian Medical Systems, Palo Alto, CA). The prescription dose was 2500 cGy, in 5 daily fractions.
Although most of the previously irradiated soft tissue was resected, the patient was consented for osteoradionecrosis and non-healing wounds. He was treated with SBRT instead of brachytherapy because the 3D printed material was not approved for biocompatibility and because filling the orbital cavity with sterile saline provided a reproducible bolus with fewer air gaps. To confirm consistency with treatment planning, daily cone-beam CT was performed after filling the orbit with sterile saline.