Case 1: A 49-year-old man with neovascular glaucoma (NVG) in the right eye (RE), who had undergone phacoemulsification, posterior chamber lens (PCL) implantation, and AGV implantation, was referred to us. His intraocular pressure (IOP) (RE) was 40 mmHg despite medical therapy including dorzolamide 2% / timolol maleate 0.5% fixed combination (Cosopt®, Merck & Co., Inc., Blue Bell, PA, USA), latanoprost 0.005% (Xalatan®, Pfizer, Inc., New York, NY, USA), and brimonidine purite 0.15% (Alphagan P® 0.15%; Allergan, Inc., Irvine, CA, USA). The AGV tube as well as its body was found exposed with complete melting of the patch graft and conjunctiva. The body of the valve was located close to the limbus (4 mm posterior to the limbus): in addition, this diabetic patient suffered from severe ocular pain (Fig. 1). Surgery was planned to remove the preexisting AGV and insert a new AGV in another quadrant because we believed that the preexisting AGV was non-functioning and pain-evoking. In addition, we believed that insertion of a new AGV into the same place might predispose the patient to a recurrence of exposure.

Under peribulbar anesthesia, a conjunctival peritomy was performed alongside the area of transconjunctival tube erosion, and the tube of the AGV was removed from the anterior chamber. To seal the space previously occupied by the tube, the fistula was debrided with a No. 11 Bard-Parker blade and a Weckcel® sponge. A piece of processed pericardium (Tutoplast®), measured 1 mm in width, was then plugged into the fistula. The pericardium was trimmed flush to the external opening and secured with two interrupted 10-0 nylon sutures (Fig. 2). Next, the plate of the AGV was explanted from the underlying sclera. The conjunctiva was drawn back in place and sutured. For implantation of a new AGV, a fornix-based incision was made through the conjunctiva and the Tennon's capsule at the superonasal quadrant. A pocket was created between the superior rectus and the medial rectus muscles by blunt dissection of the Tennon's capsule from the episclera. A new valve body was inserted into the pocket with its leading edges 8 mm posterior to the limbus. The tube of the valve was trimmed to permit a 2 mm insertion into the anterior chamber. A paracentesis was performed, and the anterior chamber was entered at the limbus with a sharp 23 G needle, parallel to the iris. The drainage tube was inserted through the needle track. The tube was secured to the sclera with two 10-0 Nylon sutures, and covered with pericardium (Tutoplast®). The conjunctiva and the Tennon's capsule were closed with 10-0 Nylon sutures.

The IOP ranged from 8 to 20 mmHg during the postoperative one month. At postoperative 14 months, IOP was 18 mm Hg and no fistula-related complication such as wound dehiscence or pericardium melting was noted.

Case 2: A 61-year-old man with diabetes mellitus, hypertension and NVG (LE) had undergone AGV implantation (LE) 19 months prior to its explantation. At 4 months postoperative, exposure of the AGV tube was noted. Except for mild hyperemia around the exposed tube, no significant inflammatory sign was noted in the anterior chamber or around the tube. IOP stayed under 20 mm Hg. However, the cataract of his left eye worsened enough to preclude a view of the fundus. Nineteen months after the AGV implantation, phacoemulsification, posterior chamber lens (PCL) implantation, AGV explantation and trabeculectomy with an intraoperative application of Mitomycin-C (0.2 mg/ml for 2 minutes) were performed concurrently on his left eye. The AGV was explanted by the method described previously. The IOP ranged from 9 to 23 mmHg during the early postoperative 3 months. At the latest follow-up at postoperative 23 months, the IOP was 13 mm Hg and no fistula-related complication was found (Fig. 3).

Case 3: A 56-year-old diabetic man who had undergone pars plana vitrectomy, phacoemulsification, PCL implantation and AGV implantation (LE) complained of ocular pain at postoperative 2 years. Hypopyon and exposure of AGV tube were noted. The patient was immediately treated with a topical fluoroquinolone (ciprofloxacin) hourly for presumed endophthalmitis. The next day the AGV was explanted using the aforementioned method. With topical antibiotics, the anterior chamber reaction resolved completely. No cultures yielded any organisms. However, at the postoperative day one, the IOP went up to the 40s. Diode laser transscleral cyclophotocoagulation was performed to control IOP. At postoperative 12 months, and the IOP was in the late 10s without any complications (Fig. 4).

Case 4: A 17-year-old man, who had undergone trabeculotomy, trabeculectomy and AGV implantation (LE) for congenital glaucoma, underwent a second AGV implantation (LE). Two months postoperatively, the tube of the second AGV was noted exposed. Explantation of the second AGV was performed using the pericardial plug in the aforementioned manner. The next day after the surgery, the IOP rose to 28 mmHg. Two hypotensive medications (Cosopt® and Alphagan-P®) were added postoperatively. During the postoperative 26 months, the anterior segment remained stable with IOP under control (Table 1).

Transconjunctival tube erosion is an infrequent but well-known complication of GDIs surgery. It is estimated that 2~7% of patients undergoing GDIs procedure develop melting of the overlying scleral or pericardial patch with erosion of the tube through the conjunctiva.8-13 Possible causes of conjunctival erosion include mechanical abrasion of the conjunctiva by the lid, excessive conjunctival tension over the tube, tube malposition, or lack of a smooth and tapered surface between the patch graft and the host along with poor ocular lubrication.10 Patch grafts like sclera, fascia lata, or dura mater may thin or disappear after implantation, secondary to ocular inflammation or a host-graft reaction.10 Pericardial patch grafts, which are supposed to be cell-free and without any antigenic stimuli, have also been reported to melt.4,10

Management of the exposed tube includes conjunctival autograft, scleral patch graft, amniotic membrane patch graft, and removal of the tube.3-6 If the tube exposure is accompanied by the infectious inflammatory signs which do not respond to antibiotics, the tube and the valve plate should be removed to help stop the propagation of the infectious process and prevent the subsequent development of endophthalmitis. Although the explantation of a GDIs is a simple procedure, it leaves an opening in the cornea and sclera that may not self-seal. If the GDIs has been implanted for a long period of time, it can be difficult to close the fistula which has been occupied by the tube.7 Mere direct sutures may close the defect. However, this would require a great amount of tissue tension and thus induce a significant amount of astigmatism which may be bothersome to an eye with good visual potential. Even if the fistula is sealed tight initially, there is still a chance that the fistula may recur later and lead to aqueous leakage if the suture is not strong enough to withstand the tissue tension. Therefore, tissue replacement may be an effective method to seal off the fistula, because it induces much less astigmatism postoperatively. Since the pericardium is highly biocompatible and provides an immediate and permanent seal, we used a pericardial plug to replace the corneoscleral fistula when explanting the AGV in all four of our cases.

Since Sibayan et al.7 first description of this surgical technique, there has been no published report of utilization of this method and its postoperative results. In Sibayan et al. study, they presented only the postoperative fourth week outcome of a single case. Our present study demonstrates the postoperative results of four cases that were followed up for a longer period. No fistula-related complication such as wound inflammation, leakage, or pericardium melting has been noted in any of our cases till the latest postoperative follow-up ranging from 12 to 26 months. Although the number of cases in our study is fairly small and the follow-up period is still short to verify the efficacy and stability of the method, the preliminary data of our cases may give support to the use of pericardium plug in such aforementioned conditions.

In conclusion, the pericardium plug seems to be an effective method in the repair of corneoscleral fistulas resulting from explantation of glaucoma valve implants. However, a prospective study of a larger number of cases with a long-term follow-up will be needed to ensure the stability of this method.