The purpose of this study was to identify the extent of optic nerve motion using MRI and to further evaluate the dosimetric impact on perioptic lesions that have undergone radiation treatment. Optic nerve motion was found to be in the opposite direction of globe motion and followed a nearly conical shape. The displacement of the PEs of the optic nerves, where they adjoin the optic canal, was minimal (within 1 mm) in all subjects, yet the displacement of the AEs of the nerves, where they connect to the globes, could be over 10 mm. A simple rigid motion model was applied to three clinical scenarios to evaluate the dosimetric effects of optic nerve movement based on the finding of this study. In these three cases, non-negligible dosimetric changes were observed.
A few studies have investigated optic nerve motion previously. Clarke et al.  demonstrated that by having patients look at different sides the optic nerves could move up to 6 mm as compared to the neutral position, based on CBCT images of four patients. In a study investigating optimal MRI sequence design for optic nerve disease by Moodley et al. , it was found that the mean total distance that the optic nerves travel during eye movement was 11.8 mm; however, none of these prior studies systematically evaluated the movement patterns for the optic nerves in multiple directions, nor did they further assess the dosimetric impact of these movements.
Although the absolute risks of RION remain relatively low in most patients, the risk for toxicity increases exponentially as therapeutic dose increases. This is especially true in patients with perioptic lesions such as meningiomas, pituitary adenomas etc.,, where hypofractionated SRS can be the definitive treatment. Due to sharp dose falloff required near critical structures with SRS, even small uncertainties in position could have a sizeable effect on the radiation dose delivered. The risk of RION can be further increased in patients who have undergone prior radiation treatments to the same area, where avoiding excessive dose to the optic nerve is crucial. For SRS, most clinical sites use either no margin or 1 mm margins for the optic nerve contours/PRV. For conventional external beam treatment, a 3 to 5 mm margin is generally used, pending image guidance and treatment technique. The results obtained from this study demonstrated that the displacement of the optic nerves as patients change the direction of their gaze can exceed these defined margins, and this may lead to insufficient dosing to the target volume if the tumor is of the optic nerve (i.e., optic sheath meningiom), or excessive dosage to the organs at risk (i.e., optic nerve itself) if the PTV is adjacent to the nerve. It should be noted that the three clinical scenarios were evaluated assuming patients kept staring at sideways throughout the treatment. For conventionally fractionated external beam radiation therapy, treatment is divided into multiple fractions over a longer course of. The eyeball movement will most likely be averaged out in all directions. This differs with SRS, where optic nerve motion could result in a higher impact, especially for patients treated with high-dose-rate beams (e.g. 1400MU/min for 6FFF beam). Additionally, a majority of patients treated in radiation oncology are simulated with CT images, which only take a few to tens of seconds. If the patient is simulated with optic neves at non-neutral positions the dosimetric estimation regarding the nerves itself could be inaccurate since the treatment plan was purely generated based on that image set. A possible solution to this could be that patients may need to be counseled to look straight ahead (or in a pre-determined direction based on target location) or to utilize eye-tracking if deemed necessary [9, 10]. Furthermore, during treatment, it may be benifical to guide the patient to look towards the desired optimal side/direction, in order to minimize the radiation dose to the optic nerve. Nevertheless, data on how far and frequently the patients move their eyes during the entire course of radiation treatment and how stable the optic nerve can stay at neutral position with/without couching, although beyond the scope of the current study, warrants larger-scale clinical investigations.
This study has a few limitations. First, to shorten the imaging acquisition time for improved patient comfort, two-dimensional imaging sequences with relatively thick slices (3mm) were used. Even though the in-plane resolution is relatively high (0.7 mm), the thicker slices could still lead to uncertainty in the measured range of movement. Given the imaging parameters, this uncertainty should not exceed 1–2 mm, in order to avoid affecting the conclusion of this study. In addition, a rigid movement model was assumed throughout the course of the study. The optic nerve, being an organ consisting of soft tissue, deforms with motion. Based on imaging results, the anterior portion of the optic nerves, which are close to the posterior walls of the globes, followed the rotation of the globes and bent towards their posterior aspects. This slightly reduces the actual movement of the optic nerves following the motion of the globes. The rigid movement model used in this study serves as a conservative estimate for the range of motion. It also shows that optic nerves maintain linear shapes following movement. This is consistent with the findings that displacement of the middle points of the optic nerves is always close to half of the displacement values measured for the anterior portions. Therefore, a rigid movement model, which may overestimate the movement slightly, should still provide a very reasonable and relatively accurate estimation of the range of motion. Finally, the clinical impact of the dosimetric changes found based on the motion-inclusive model is difficult to assess and a prospective study would better eludicate the overall impact on reducing the incidence of RION or improving tumor control.