Dose optimization strategies are available for I-125 plaques, which are not possible for the commercial Co-60 and Ru-106 applicators. The removable I-125 seeds permit modification of the activity distribution on the plaque surface, while the low energy of I-125 enables shielding and source anisotropy to be exploited. If these relatively simple optimization strategies were to be implemented clinically, they might result in fewer late, radiation-induced, vision-degrading complications. Clinical implementation, however, will not be a simple matter. Major differences exist between the environment of the computer model and current clinical practice. These differences include the precision and accuracy of tumor localization, measurement of ocular dimensions, and accuracy of surgical placement of the plaque.
Regarding surgical placement, orientation of the suture eyelets, and distances between the suture eyelets or lip of the shell and anatomic landmarks such as the anterior pole or limbus could be used to facilitate precise placement of the plaque. The major difficulty lies in immobilization of the plaque while anchoring the sutures. Our clinical experience, however, suggests that plaque placement can be achieved to within 1 mm of the preplanned position using this approach. The accuracy of the prescription for plaque placement depends somewhat on correct dimensions for the eye. Clearly, the more information concerning the specific anatomy that is available, the more accurately the model can reflect actual conditions of treatment. We have addressed this problem by allowing our model to conform to a transecting CT image. A problem which remains to be investigated, however, is whether the presence of the plaque significantly distorts the geometry of the eye compared to its imaged state. Transillumination, ultrasound, and magnetic resonance imaging techniques (27) are available to verify plaque placement.
The level of precision required for plaque placement will vary with tumor size, location, and the optimization strategy employed. Our results suggest that exploiting shielding by the lip is one of the more effective means of reducing dose to the macula. It is clear, however, that strategies which depend strongly on lip shielding will require greater precision of plaque placement and better documentation of ocular anatomy than strategies which are primarily based on distance, activity distribution, and source anisotropy.
Perhaps the most critical problem is precise documentation of the tumor dimensions and location within the eye. We have implemented a technique for direct digitization of posterior tumors from fundus camera photography which presumes the retina is a spherical surface. The photography itself, however, is subject to various sources of optical distortion. The eye is composed of three lenses; the aqueous lens which is shaped by the cornea, the crystalline lens, and the vitreous lens. Problems in any of the ocular lenses or in the fundus camera will result in optical distortion of the image of the retinal surface. For example, the index of refraction of the crystalline lens is not constant since the lens is more dense near the center. Furthermore, digitization of fundus photos provides no information about the three-dimensional structure of the tumor, such as the location and height of the apex. Presently, this information is obtained from ultrasonic examination. We have recently developed an interface to our CT and MR imaging systems which permits reconstruction of the three-dimensional ocular anatomy from sequential tomographic images. This will eventually provide an additional means of localizing tumors. It should be particularly valuable for anteriorly positioned tumors which are inaccessible to fundus photography.
Additional optimization strategies beyond the simple maximization of the T:M ratio presented in this report are (of course) possible and may be more appropriate in various circumstances. It may be desirable, for instance, to emphasize other sites such as the optic nerve, particularly for tumors located very close to the nerve. It may also be possible to develop new plaque designs which optimize shielding and orientation characteristics for individual seeds and for tumors in specific locations. The design process is assisted by our computer model which permits rapid modification of plaque design using interactive graphics.
The development of the computer model makes it possible to seriously consider optimization of I-125 plaque brachytherapy. Furthermore, it emphasizes the immediate necessity for improving our methods of tumor localization and documentation of patient-specific ocular anatomy. If the potential for dose reduction to critical structures outside the tumor volume (such as the macula and fovea) suggested by this study could be implemented clinically, it may lead to improved vision for patients with ocular tumors without compromising tumor control.
Address reprint requests to: Melvin A. Astrahan Ph. D., Department of Radiation Oncology, University of Southern California School of Medicine, Kenneth Norns Cancer Hospital, 1441 Eastlake Ave., Los Angeles, California, 90033, USA