Proton Therapy for Uveal Melanoma on a Pencil Beam Scanning Gantry

Abstract

Purpose

We present our experience treating ocular tumors in a standard pencil beam scanning (PBS) gantry room without apertures, which could broaden access to proton therapy for patients with ocular cancer globally. Besides, this study explores the dosimetric benefits of beam-specific apertures.

Methods and Materials

We retrospectively evaluated 11 consecutive patients with uveal melanoma treated in a clinic gantry room. The dose deviations between the planned and received by the patient were investigated by assessing the forward calculation of the treatment plan on the synthetic computed tomography of cone beam computed tomography. Each plan was forward calculated with a beam-specific brass aperture (BSA) using a Monte Carlo algorithm to explore dosimetric improvements. We compared the plan quality to the delivered plan (DP) using target coverage (D95%) and mean/maximum doses to the adjacent organs.

Results

A close agreement between the planned and delivered dose was achieved, with D95% deviations within 3.6% for all treatments, maintaining dose constraints for critical organs. Similar target coverage was reached, with D95% at 101% ± 1.0% (DP) and 101% ± 3.2% (BSA). BSA was effective (P < .05) in reducing the mean [D_Mean (DP, BSA)Gy] and maximum [D_Max (DP, BSA)Gy] dose to organs: retina D_Mean (37.7, 29.5), cornea D_Mean (10.7, 2.4), conjunctiva D_Mean (13.6, 4.1), lacrimal gland D_Mean (25.5, 14.1), optic nerve D_Mean (19.6, 13.1), lens D_Max (22.4, 8.5), cornea D_Max (24.4, 10.2), eyebrow D_Max (15.3, 6.8). BSA lowered the mean dose to surrounding organs and significantly decreased the maximum dose to nonabutting organs (lens, cornea, eyebrow), but had little impact on the maximum dose to the abutting organs (retina, optic nerve).

Conclusions

We demonstrate the successful implementation of ocular proton treatment with a standard PBS gantry beamline without apertures. The beam-specific apertures effectively reduced doses to the organs adjacent to the target in the PBS proton treatment while maintaining similar target coverage. This approach offers an opportunity to expand access to ocular proton therapy widely.

Introduction

Uveal melanoma (UM) poses a threat to life, the eye, and vision.¹ The Collaborative Ocular Melanoma Study² demonstrated that compared to surgical removal of the eye (enucleation), radiation therapy (brachytherapy) yields an equivalent rate of survival, with more patients retaining their eyes and vision for medium-sized tumors (2.5-10 mm in height and within 16 mm in basal diameter). It has been observed that following plaque brachytherapy, tumors in a posterior location are more likely to recur than those in an anterior location.³ This may be due to the tumor’s proximity to the optic disc and the interference of the optic nerve in the placement of the plaque. In this situation, proton beam therapy is often the preferred radiation therapy strategy.4, 5, 6

Most centers using proton therapy (PT) for UM have a dedicated “eyeline” treatment room to deliver radiation through a single anterior beam with a small aperture.7, 8, 9 Image guidance is conducted with bony and tantalum marker alignments.¹⁰ An idealized eye model is used for treatment planning and predicting the dose-volume histogram based on patient-specific parameters.¹¹ Ultrasound and fundus photographs are often used for tumor delineation in treatment planning.^12,13 Although effective, this approach is limited in availability and the program is expanding slowly worldwide.^14,15

For this reason, investigators have explored a pencil beam scanning (PBS) proton solution for the treatment of UM. This technique has been developed rapidly in recent years, and new proton centers are typically equipped with PBS and advanced imaging techniques.¹⁶ Along with highly conformal dose distribution, an advanced 3D treatment planning system (TPS), and high-quality online volumetric images, PBS can potentially expand the use of PT for ocular treatments.

This study presents our initial experience with multiple-field PT for UM using a regular gantry and general TPS, potentially expanding the availability of PT for patients with UM. We also quantitively explored the dosimetric benefit of apertures.

Materials and Methods

Patient selection for proton beam therapy

A total of 11 consecutive patients with UM received PBS PT using the ProBeam (Varian Medical Systems) proton gantry beamline since 2020. They were retrospectively evaluated for this study, which the institutional review board approved.

Pretreatment evaluation and preparation

All patients were evaluated by an ophthalmic oncologist and radiation oncologist for diagnostic confirmation, tumor localization, and staging. During a subsequent procedure under anesthesia, the ophthalmic oncologist placed 2 to 3 tantalum fiducial markers on the scleral surface. These markers, lens/optical nerves structures, and the gantry-based cone beam computed tomography (CBCT)/orthogonal kV images were used for image guided patient and tumor alignment during daily treatment. A qualified medical physicist advised the marker placement based on potential PT beam angles. Generally, markers were placed at the distal and lateral edges of the beams, minimizes the dosimetric perturbations from these markers in dose optimization and calculation.¹⁷

Simulation

A custom-molded head mask (Qfix) was created to immobilize the head. A custom-designed gaze fixation device was constructed from Qfix thermoplastic material and then attached to the mask, typically contralateral to the eye being treated. Figure 1 presents the 11 patients’ computed tomography (CT) scans and an example of a customized gazing device. The eye gaze direction was selected based on the target location and the organs-at-risk (OARs) of interest, typically, the gazing device is placed above the contralateral eye (nondisease) to attract the disease eye and pull the target more superficially toward the temporal scalp surface (minimizing the exposure of healthy tissue to the proton beam path). The gazing device was attached to the disease side for patient 1 due to the loss of vision for the contralateral eye. CT scans (120 kVp, 1.0 mm in-plane resolution and 1.0 mm slice thickness) were performed in the treatment position with immobilization and gazing devices in place; eyelid retractors were not employed because no field was planned from the anterior direction. The effectiveness and reproducibility of the gaze fixation were verified with mini-scans (reduced slices) of the eye, with a short rest between scans. The CT slice thickness of 1.0 mm was used for scans, and verified for the accuracy of the dose calculation against a smaller slice thickness. The positions of the lens, markers, and optical nerves are assessed between CT scans to ensure the effectiveness of the gazing device and the intergazing target variations.

Figure 1.

The target volume was delineated and illustrated on an axial section of the simulation computed tomography (CT) for 11 patients in this study. The last figure in the third row shows the custom-molded head mask (a) with an eye gaze fixation device (b) attached to the mask and a marker (c) for the patient to gaze at. The images in the last row show the kV and cone beam computed tomography (CBCT) for patient 8 acquired for alignment, and the alignment contours from treatment planning.

The tantalum markers and the gazing device were contoured as reference structures for image guided radiation therapy purposes. Treatment beam angles were determined to avoid directly passing through the gazing device and tantalum markers to minimize the associated uncertainties. In addition to the gross target volume (GTV) delineated on CT using fundus mapping provided by the ophthalmologist, normal OARs, including the retina, optic nerve, cornea, lens, lacrimal gland, conjunctiva, sinonasal mucosa, and eyebrow, were contoured for plan optimization and evaluation. Because these tumors primarily arise in peripapillary or foveal regions, sparing structures such as the macula/fovea and optic disc is particularly challenging. As a result, those structures are not routinely delineated during treatment planning. Moreover, the ciliary body typically resides in the far lateral penumbra region, rendering it less critical than in conventional eyeline treatment and thereby not routinely contoured.

Treatment planning

Eclipse V15.6 TPS (Varian Medical Systems) was commissioned and validated for small-field dose optimizations and calculations for a specific treatment room. The gantry room used for ocular treatment was comprehensively commissioned. Machine parameters consistency such as spot position, spot size, as well as the output were measured and confirmed for every 10 MeV (70-180 MeV) at every 15° gantry angle. The analytical dose calculation algorithm was validated against the Monte Carlo algorithm demonstrating < 2.5% dose variations for small fields (1-2.5 cm in dose full width at half maximum). Additionally, the impact of CT slice thickness (1.0 mm) and dose grid size (1.0 mm) is evaluated to be used in the final clinical protocol. The plan optimization and dose calculation were performed using the analytical proton convolution superposition algorithm. Clinical plans were created, with 3 to 4 proton fields selected for dose optimization using either single-field optimization (SFO) or multifield optimization (MFO). All patients are prescribed to 50 Cobalt Gray Equivalent in 5 consecutive days. Constraints were considered in dose optimization for critical OARs: retina (<55 Gy), optic nerve (<55 Gy), cornea (<30 Gy), lens (<30 Gy), lacrimal gland (<55 Gy), conjunctiva (<55 Gy), and eyebrow (<20 Gy). Robust optimization considering customary perturbations with a combination of setup errors (3 mm) and range uncertainties (3.5%) was employed for planning^11,18 where 2 mm spot spacing and 1.5 sigma layer spacing were used. SFO typically achieves a balanced dose contribution from each field, and 8 out of 11 patients were treated with SFO plans. MFO was used for complex cases when the tumor had a concave shape (cases 1, 2, and 6) or critical OAR dose constraints were not achievable with SFO. Depending on the beam angle and target depth for a specific field, a 2 cm range shifter, mounted on the gantry at the exit of the beamline, with a minimal air gap (around 10 cm), was selectively used for dose optimization.¹⁹ The beams often avoided passing through tantalum markers and eyebrows. Most treatments were planned with 3 fields (8 out of 11 cases, and the rest had 4 fields). Common beam geometry includes lateral, lateral-posterior oblique, and superior oblique with a couch kick. For the SFO plans, the field weights were optimized to achieve a balance of robustness and minimal dose for critical organs. The dose calculation considered the heterogeneous eye tissue and functional structures with a calculation grid of 1.0 mm. Plans were forward calculated on the patient’s CT using the Monte Carlo algorithm, AcurosPT V16.1 in Eclipse, as the second dose and MU verification. Radiation oncology, ophthalmology, medical dosimetry, and physics were involved in the institutional peer review of the treatment plan and included in the author list.

Robustness evaluation was applied to each case. The target coverage goal was 95% of the GTV receiving ≥95% of the prescribed dose under the perturbations of 3 mm setup errors and 3.5% range uncertainties. All the planning goals for OARs were planned to be met when range uncertainties were considered. The output of each field was measured with a micro-ion chamber and 2D dose distribution with a scintillator detector at 2 different depths for plan quality assurance. Two-dimensional gamma analysis was applied with a passing rate of at least 95% with criteria of 2 mm/3% and 10% dose threshold. All the treatment plans went through institutional peer review before the start of treatment.

Treatment and imaging guidance

Patient positioning, eye alignment, and gaze fixation reproducibility were verified with daily orthogonal kV radiography (70 kVp, 0.0283 cm resolution at isocenter) and CBCT (100 kVp, 0.0565 cm in-plane resolution and 0.2 cm slice thickness) for each treatment. The last row of Fig. 1 illustrates an example for patient 8, showing kV and CBCT acquired for alignment alongside the alignment contours created during treatment planning. Bony alignment was applied to kV images first. The gazing device and marker (on the gazing device) were verified against their contours on the planning CT. The gazing device on CBCT not aligning well with that on the CT indicated a deviation in the position of the gazing device that required positional adjustment (of the device, not the patient). Once bone anatomy and gazing device were confirmed to be well aligned, the position of the tantalum markers and eye substructures (optic nerve and lens) was verified on CBCT. Three structures were assessed in the following order: the tantalum fiducial markers, the ipsilateral optic nerve, and the lens. The former structure is typically the easiest to visualize, and the latter may not be visible in patients who have undergone cataract extraction. The displacement of the eye’s position in CBCT typically indicates the patient is not looking at the gazing device because they were instructed during the simulation CT. Additional coaching and verification were repeated for optimal alignment.

The reproducibility of the patient’s positioning and anatomy was confirmed by daily CBCT before each treatment. The beam-on time is around 24 $\pm$10 seconds for each field. The typical treatment time is 40 minutes for the first fraction and 30 minutes for the following fractions. Between the treatment fields, kV images were retaken at the angle of the treatment field to ensure the consistency of the target positioning through the bone and markers alignment. When necessary, additional CBCTs were taken for realignment and verification purposes. A structure defined by the 20% isodose line was used as a volume of interest to guide the therapist during the imaging alignment. The volume helps the therapist to identify the beam path and confirm the tissue variations along each beam path, which eventually translates to range uncertainties and dose errors. It also serves as a reference in CBCT for beam path consistency in daily treatment. A physicist verified the patient’s setup before beam-on and was present during the treatments. The physician on the treatment day reviewed the anatomic alignment, provided clinical decisions before the treatment, and oversaw the treatment.

Delivered plan evaluation

Synthetic CT images based on daily CBCT images were generated using the image-correction-based method in MIM Software V7.2.11 (MIM Software Inc).²⁰ This method will address the shading artifact, fine-tune voxel values of CBCT to match the Hounsfield unit of planning CT via the deformation algorithm. Target volumes were transferred based on the image registration in treatment between planning CT and CBCT through marker-marker registration. Treatment plans were forward calculated on daily synthetic CTs using the proton convolution superposition algorithm in Eclipse, and the accumulated doses from daily treatments were generated to assess the dosimetric consistency between the planned dose and the dose received by the patient. ²⁰ The differences in dosimetric metrics were reported for target coverage and mean/maximum dose to critical organs.

Plan regeneration with beam-specific aperture

New plans were generated by forwardly calculating the treatment plans using the AcurosPT Monte Carlo algorithm after adding a 2 cm thickness beam-specific brass aperture (BSA) to each treatment field. The beam angles remained identical to the delivered plans (DPs). The aperture was created in the beam’s eye view of the field-specific target with 3 mm margins and placed close to the patient with a 4 cm or less air gap without collision with the patient or couch.²¹ The dosimetric metrics were compared and reported, including target coverage and doses to OARs.

Statistics

Multivariable regression analyses were performed on the dosimetric metrics of the 2 plan strategies with respect to the rule of thumb that more than 10 events are required per predictor. All P values are 2-sided, and the statistical significance was acceptable at the condition of P < .05.

Results

Figure 1 presents the tumor contours and locations for all 11 patients in this study. Most of the patients had a peripapillary tumor involving the optic disc. Table 1 summarizes the patients' characteristics and radiation treatment details for the 11 patients in this study, including the patients’ diseased eye, gender, age, prior radiation therapy, GTV (volume and diameter), the volume ratio between target and globe, the number of markers, the use of SFO or MFO, the number of treatment fields, field arrangement, using range shifter or not, and the location of the gaze device.

Table 1.

Patients’ characteristics and radiation treatment details

			CTV/GTV
Patient (diseased eye)	Gender/age	Eye with prior RT	Volume (cc)	Diameter (cm)	Vtarget/Vglobe	No. of markers	SFO/MFO	No. of fields	Field arrangement (Gantry angle-Couch angle)	Range shifter	Gazing
1 (left)	M/61	N	1.09	1.6	0.122	3	MFO	3	G325-C0, G115-C0, G135-C0	None	Left*
2 (right)	F/50	Yes, right	0.71	1.5	0.084	2	MFO	4	G30-C0, G270-C0, G270-C90, G240-C0	None	Left
3 (right)	M/57	Yes, right	0.77	1.5	0.067	3	SFO	4	G280-C0, G270-C30, G260-C0, G240-C0	2 cm	Left
4 (left)	M/57	N	0.30	1.0	0.037	2	SFO	3	G90-C0, G125-C0, G90-C335	2 cm	Right
5 (right)	M/59	N	0.34	0.8	0.033	2	SFO	3	G300-C0, G270-C0, G240-C0	3 cm	Left
6 (right)	M/74	N	2.98	2.5	0.303	2	MFO	4	G270-C0, G250-C0, G230-C0, G230-C40	None	Left
7 (left)	M/62	N	0.30	0.8	0.030	2	SFO	3	G120-C0, G90-C330, G85-C0	2 cm	Right
8 (left)	M/46	N	0.24	0.5	0.025	2	SFO	3	G120-C0, G90-C0, G90-C330	2 cm	Right
9 (right)	F/82	N	0.42	1.0	0.048	2	SFO	3	G270-C30, G265-C0, G240-C0	2 cm	Left
10 (left)	M/68	N	1.47	2.0	0.143	2	SFO	3	G125-C0, G100-C330, G250-C90	2 cm (G125, G100)	Right
11 (right)	F/60	N	0.38	1.0	0.040	2	SFO	3	G260-C0, G250-C0, G235-C0	2 cm (G260, G250)	Left

Abbreviations: CTV; GTV = gross target volume; MFO, multifield optimization; RT, radiation therapy; SFO, single-field optimization.

^⁎Patient 1’s gazing device was attached to the ipsilateral side due to poor vision in the contralateral eye.

To assess the planned versus delivered dose, a dose evaluation and analysis were performed on all 11 patients (Fig. 2). The deviations in D95% for the target coverage are all within 3.6%. In the worst case (patient 4), the target coverage in GTV D95% is reduced from 102% in the planned dose to 98.7% in the delivered dose. Additionally, the dose delivered on OARs for each case shows no violations of the dose constraint for any OAR. These results indicate that our system offers robust treatment with consistent quality.

Figure 2.

The differences in dosimetry metrics between the planned and the delivered doses. The differences were calculated as (delivered – planned dose)/planned dose in percentage. The planned dose was obtained from the simulation computed tomography (CT). The delivered dose was obtained from the synthetic CT.

Proton plans were regenerated for each patient using a BSA to assess the potential dosimetric improvement. A representative patient example (patient 8) is shown in Fig. 3. The first row depicts the 2D field-specific dose comparison between the delivered and aperture plans. The plan dose distribution comparison between the delivered and aperture plans is shown in the second row, where 2 arrows indicate 2 coplanar fields. The aperture plan resulted in a sharper dose distribution with the average lateral penumbra reduced from 8.6 mm to 4.1 mm and the average distal penumbra reduced from 8.7 mm to 7.8 mm, reducing the irradiated volume for nearby OARs. This improvement is more noticeable for the distant organs including the cornea, lens, conjunctiva, lacrimal gland, and eyebrow.

Figure 3.

Comparison of dose distribution for a single lateral field (first row) and all 3 fields (second row) among the delivered plan (left) and aperture plan (right, 2 cm air gap) for the eighth patient. Yellow arrows: 2 beam angles for the coplanar fields; the third beam is a noncoplanar field and is not shown here.

The dosimetric parameters of target coverage and dose to OARs for the delivered and aperture plans are compared and reported in Table 2 and Fig. 4. The conformity index (CI) proposed by Radiation Therapy Oncology Group was calculated as CI_RTOG = V95%/TV, where V95% is the 95% isodose volume, and TV is the volume of the target with 2 mm isometric expansion.²² Results show that the aperture method can reduce doses to the organs adjacent to the target in the PBS proton ocular treatment while achieving similar target coverage, with D95% at 101% ± 1.0% (DP) and 101% ± 3.2% (BSA). Better conformity of CI_RTOG = 1.08 ± 0.36 was achieved using BSA versus the CI_RTOG = 1.22 ± 0.25 for DP. BSA was effective in reducing the mean [D_Mean (DP, BSA) Gy] and maximum [D_Max (DP, BSA) Gy] dose to organs: retina D_Mean (37.7, 29.5), cornea D_Mean (10.7, 2.4)/D_Max (24.4, 10.2), conjunctiva D_Mean (13.6, 4.1), lacrimal gland D_Mean (25.5, 14.1), optic nerve D_Mean (19.6, 13.1), lens D_Max (22.4, 8.5), eyebrow D_Max (15.3, 6.8) (P < .05 for all). BSA lowered the mean dose to the surrounding organs and significantly decreased the maximum dose to the nonabutting organs (lens, cornea, and eyebrow), but had little impact on the maximum doses to the abutting or overlapped organs (retina and optic nerve).

Table 2.

Comparison of dosimetric metrics for target coverage and dose to OARs among delivered plan and planning method using beam-specific apertures

		Clinical goal	Delivered plan	Aperture plan
GTV	D95% (%)	$\ge$100	101 ± 1.0	101 ± 3.2
	CIRTOG	1	1.22 ± 0.25	1.08 ± 0.36
Retina	DMax (Gy) DMean (Gy)	<55	54.5 ± 0.7 37.7 ± 5.1	54.5 ± 0.6 29.5 ± 8.9
Lens	DMax (Gy)	<30	22.4 ± 9.7	8.5 ± 13.9
Cornea	DMax (Gy) DMean (Gy)	<30	24.4 ± 9.5 10.7 ± 5.8	10.2 ± 14.6 2.4 ± 4.2
Conjunctiva	DMax (Gy) DMean (Gy)	<55	43.2 ± 8.3 13.6 ± 4.9	40.2 ± 13.0 4.1 ± 4.3
Lacrimal gland	DMax (Gy) DMean (Gy)	<55	41.8 ± 9.0 25.5 ± 13.1	37.8 ± 14.0 14.1 ± 15.9
Optic nerve	DMax (Gy) DMean (Gy)	<55	52.7 ± 1.1 19.6 ± 4.2	53.1 ± 0.6 13.1 ± 4.7
Eyebrow	DMax (Gy)	<20	15.3 ± 8.5	6.8 ± 9.5

Abbreviations: OAR = organs-at-risk; GTV = gross target volume.

Figure 4.

Box plot of the (a) mean and (b) maximum organ dose comparison for delivered plans and plans with beam-specific apertures. Each box consists of 11 patients' doses of different organs-at-risk (OARs). The lowest, highest, 25th percentile, 50th percentile, 75th percentile, and average doses are presented in the plot. +: Average dose. The dosimetry of different plans is compared, and differences with P < .05 are tagged with stars (*).

Figure 4a shows the mean dosimetry of OARs on the DP and aperture planning method. The mean doses to the retina, cornea, conjunctiva, lacrimal gland, and optic nerve showed significant decreases using apertures compared to the DP.

Figure 4b shows the maximum dosimetry of OARs from the DPs and aperture planning method. Maximum doses to the lens, cornea, and eyebrow showed significant decreases from the apertured plans compared to the DPs. No statistical difference was found in the maximum doses to the retina, conjunctiva, lacrimal gland, and optic nerve.

In addition, the cases with OARs’ dose violations in the treatment plan all met the constraints after introducing apertures. For patient 2, the maximum dose to the cornea of the DP was 3239 cGy, which was 239 cGy higher than the dose constraint of 3000 cGy. The dose was reduced to 829 cGy using the beam-specific aperture. For patient 10, the maximum dose to the eyebrow of the DP was 2500 cGy, which was 500 cGy higher than the dose constraint of 2000 cGy. With the aperture planning method, the maximum dose on the eyebrow was reduced to 1305 cGy.

Discussion

According to the Particle Therapy Co-Operative Group, PT has expanded rapidly in recent years, and more than 110 proton centers are now in operation in 20 countries. However, access to a proton beam for ocular treatment is limited due to the high cost of a dedicated proton eyeline treatment room and a relatively low frequency of ocular malignancies treated with radiation therapy. A single en-face beam using a single-scattering approach and Bragg peak modulation is applied through a dedicated eyeline in many existing eyeline-based proton treatment facilities.²³ At centers using dedicated eyelines, EYELINE, OCTOPUS, or similar specificatlly designed treatment planning systems are used for most of the treatment planning.¹¹ In contrast, more recently developed proton centers are equipped with PBS delivery techniques with a full or partial gantry. Recently, pencil-beam-based techniques have been reported, such as uniform scanning with apertures²⁴ and CT-based 3D treatment planning with multiple beam directions.²⁵ PBS-based single en-face field ocular treatment demonstrated similar dosimetric properties to the traditional eyeline.²¹ Our work is the first study to systematically report the multiple-field PBS proton treatment of ocular disease using a general TPS and regular full-gantry room. In addition, potential dosimetric improvements are explored when the aperture option is available.

A PBS treatment can be planned by either SFO or MFO. Many studies have shown the pros and cons of each planning technique.^26,27 For small ocular targets, the dose homogeneity and modulation are spatially limited. Therefore, the difference in plan robustness is less pronounced between SFO and MFO. Because of the small size target, caution should be given to balancing the dose contributions from each field for MFO plans to reduce the dose inhomogeneity. Especially for reirradiation cases, MFO offers more flexibility in optimizing the dose for nearby organs.²⁶

One of the advantages of using a gantry room for ocular treatment is its flexibility for beam angle selection. Compared to fixed-beam-based treatment, where the preservation of the eye and vision was based on the optimization of aperture shape, gaze angle, and compensators or wedges,¹¹ a PBS gantry beamline can achieve OAR sparing by beam angle selection and inverse dose optimization. Initial efforts were placed to explore the optimal beam angles for a regular ocular target. For example, anterior and vertex beams were investigated. With electing to defer using an eyelid retractor for patient convenience and the absence of an eye-tracking system, the range uncertainties caused by eyelid position were of concern in the anterior field. The vertex field must pass through the brain to reach the target, thus unnecessarily irradiating some brain tissue, which is unfavorable. Our current ocular planning protocol recommends 3-4 lateral or lateral oblique beams on the target side, offering potential dose conformity to the target.²⁸ For melanoma of the iris or ciliary body, which is predominantly managed with plaque brachytherapy,⁴ PT offers a viable alternative.²¹ In many cases, the tumor and adjacent critical organs are closely apposed or overlapping, making a steep dose gradient indispensable to spare the normal organs. Consequently, adding apertures to the treatment beam becomes essential to achieve the optimal dose conformity and maximal protection to nearby normal structures. The tantalum markers are inserted based on a standard planning protocol to ensure the markers do not cause dose perturbations. Therefore, the markers are often placed on the medial aspect of the globe, superior or inferior to the target, and beyond the distal edge of the treatment beams.

Small-field dosimetry is a common challenge in radiation therapy. Several groups have explored the limitations and options of dosimeters for proton small-field dosimetry.^29,30 Unlike passive scattering and uniform scanning delivery techniques, beam modifiers such as apertures are not widely used for routine PBS clinics, and not all commercial vendors support the aperture. It is worth pointing out that spot size variations are more critical for ocular treatment than routine clinical cases due to the nature of high proximity of many critical organs to target. The change in spot size results in penumbra dose variation, which eventually cause dose variation to surround organs.³¹ Therefore, one of our gantry rooms is commissioned specifically to treat ocular patients or patients with very small targets (1-2.5 cm in dose full width at half maximum). The spot size variations for each energy at every 15° gantry angle are verified to be within 10%. All the patient-specific quality assurance (QA) for ocular patients is performed in the treatment room.

Given the small target and hypofractionated delivery, the reproducible eye position is crucial for successful ocular radiation therapy. Eye immobilization is achieved through a gaze fixation device (Fig. 1) attached to the patient’s mask. The position of the gazing device must be verified by kV images against its contour and fine-tuned as needed with imaging guidance to match its position at simulation. However, this simple gazing device does not have a function to monitor the patient’s eye position during the proton beam delivery. Thus, a large setup uncertainty was used in the plan robust optimization. Researchers are actively exploring camera-based solutions,^32,33 and an in-room camera system is reported to monitor the patient’s gaze throughout the treatment.^21,25,34, 35, 36 A gazing device with a tracking and monitoring camera is under development in our institution.

Besides the potential improvement of eye gazing, monitoring, and tracking by introducing a digital camera, reducing and sharpening the penumbra of the proton beam are worth further improving ocular treatment delivery. Spot size is critical in ocular treatment, especially for low-energy proton beams with a range shifter. A larger air gap between the patient and the range shifter is typically adopted to avoid a collision with the bulky nozzle. A range shifter with minimal air gaps was proven to maintain the spot size integrity and the capability of high-dose modulation.³⁷ An aperture further sharpens the dose fall-off of boundary spots.^38,39 Especially for ocular tumors, many OARs are within or abutting the globe, and their volumes are very small. Sharper penumbra provides sharper dose fall-off outside the target volume and improves the dosimetric parameters of these surrounding OARs. Our study indicates that clinical implementation of brass aperture helps to improve the dose conformity and sparing surrounding OARs for ocular treatment, although extra effort is required for the aperture plan QA and aperture milling support.

Conclusions

The ocular proton treatment delivered with a regular PBS gantry beamline was successfully implemented, which provides more data to support the improvement on the availability of proton treatment for ocular patients. We also demonstrate that beam-specific apertures may significantly reduce doses to the organs adjacent to the target while achieving similar target coverage. Additional efforts focus on developing an eye-tracking and monitoring system to maintain the imaging-confirmed eye position during the proton beam delivery and ongoing.

Disclosures

Haibo Lin reports receiving a research grant from Varian Medical System. Charles B. Simone, II reports receiving honoraria from the Varian Medical System. David H. Abramson reports receiving Cancer Center Support Grant P30 CA008748 from National Cancer Institute (NCI). Jasmine H. Francis reports receiving Cancer Center Support Grant P30 CA008748 from NCI. Christopher A. Barker reports receiving Cancer Center Support Grant P30 CA008748 from NCI; reports receiving investigator-initiated trial support from Regeneron, EMD Serono, Amgen, Elekta, Melanoma and Skin Cancer Trial Limited, Merck, Alpha Tau Medical and subcontract of investigator-initiated trial from University of California San Francisco; reports receiving subcontract of NCI SBIR grant from Physical Sciences Incorporated; reports receiving payment for participation in expert peer exchange from American Journal of Managed Care; reports receiving travel support from National Comprehensive Cancer Network, University of Washington and National Cancer Institute; reports receiving scientific advisory fees from Regeneron; reports uncompensated relationship with Castle Biosciences; and compensated as part of salaried employment of vice-chair for clinical research, Department of Radiation Oncology, Memorial Sloan Kettering Cancer Center.