Aqueous Shunts and Keratoplasty

Introduction

Aqueous shunts (glaucoma drainage devices, drains, implants, tubes) have a long history of application in surgical control of intraocular pressure (IOP) in complicated glaucomas. Three devices, the Ahmed implant ( Fig. 119-1 ), the Baerveldt implant (see Fig. 112-1 ), and the Molteno implant ( Fig. 119-2 ), all available in more than one model, have been most frequently reported in recent publications as useful for controlling IOP in eyes undergoing penetrating keratoplasty (PK). The incidence of reported glaucoma following corneal transplantation is significant, and varies from 6.5–45%, depending on multiple pre- and postoperative factors.

Indications

The principal indication for installing an aqueous shunt combined with a penetrating (or partial) keratoplasty is known or anticipated glaucoma unlikely to be controllable with topical or systemic medications in an eye with scarring of the conjunctiva. Often, IOP control may be marginal or clearly unacceptable preoperatively in spite of maximal tolerated therapy. Laser trabeculoplasty has often either been done previously or is not possible due to corneal opacity. Most eyes undergoing PK have had prior cataract surgery or prior inflammatory disease such as herpes keratitis, and often come to keratoplasty with at least partially closed angles. Many also have had complications of cataract surgery, perhaps including placement of an anterior chamber intraocular lens (IOL) with resulting peripheral anterior synechiae formation around the haptics and variable additional clock hours of angle closure. In the case of repeat PK, the angle is often partially or completely closed by peripheral anterior synechiae sometimes extending onto clear cornea with segmental or complete obliteration of the anterior chamber. Hence, the development of glaucoma following corneal transplantation is multifactorial, but most often the result of chronic synechial angle closure from inflammatory disease or surgically induced, prolonged corticosteroid use, or decompensated pre-existing glaucoma.

Contact lens dependence, judged to preclude a standard filtering bleb; epithelial ingrowth not amenable to surgical excision; congenital glaucomas not responsive to standard pediatric glaucoma surgery (goniotomy, trabeculotomy); surface disease (pemphigoid, Stevens–Johnson syndrome); trauma precluding standard surgery; chemical burns; iridocorneal endothelial syndromes (ICE) failing standard surgery; and uveitis-related glaucomas are also indications for considering an aqueous shunt. Since aqueous shunts require substantial periocular space they may not be applicable to some congenital anomalies associated with contracted orbits or small eyes.

The suspicion of poor endothelial function in an existing graft based on clinically appreciable corneal edema at relatively low IOP bodes poorly for graft survival after an aqueous shunt is installed. However, if IOP is high and presumed to exceed endothelial tolerance, shunt installation may sufficiently lower IOP so as to allow clearing of corneal edema and prolong graft life. When faced with an apparently failing graft at moderately high IOP, installing a shunt may forestall or preclude the need for repeat grafting. In any case, a subsequent repeat PK can be done with favorable outcome if IOP has been well controlled by the tube. Cyclodestruction, via diode transscleral or endoscopic technique, can compliment prior tube installation if IOP is unacceptable prior to repeat PK or be applied simultaneously with repeat PK in eyes with poorly functioning shunts.

The choice of device, especially whether a ‘large’ or ‘small’ surface area explant is to be used, is probably best made by estimating aqueous production as normal or reduced and the expected longevity of the patient. In a young otherwise healthy patient with a long life expectancy, logic favors choosing a large surface area device. In a systemically fragile, elderly patient, especially with end-stage diabetes and presumed low aqueous production, a smaller surface area device is probably preferable and safer, since aqueous filtration has been related to explant surface area. Overfiltration after aqueous shunt implantation is rare but may necessitate closure of the tube or cutting down the explant size. If IOP control is not adequate after placement of a smaller device, a second device can usually be placed in an adjacent quadrant. The choice between ‘valved’ or ‘nonvalved’ devices can perhaps be best made based on the urgency of IOP control. If an immediate postoperative drop in IOP is considered essential, the Ahmed shunt may be the best choice as it functions without delay. A nonvalved Baerveldt or Molteno requires temporary ligation of the drainage tube to allow encapsulation, typically delaying shunt function for several weeks. Fenestration of the tube anterior to the ligature may allow immediate function but is relatively unpredictable. If the eye in question has a relatively healthy optic nerve, the delay in function associated with ligatured nonvalved devices may be acceptable. All shunts can be associated with immediate undesirable hypotony, and additionally the single-plate Ahmed's long-term efficacy is probably less than that of the larger Molteno or Baerveldt devices with larger explants. The nonvalved Molteno and Baerveldt devices should always be temporarily ligatured to allow encapsulation of the explant before tube opening and avoidance of profound temporary hypotony and its complications. Several methods of temporary ligature of nonvalved tubes have been described and can be applied to pars plana or anterior chamber tube installations with equal effectiveness.

Corneal grafts, whether full thickness or Descemet's stripping endothelial keratoplasty (DSEK), have been shown to have a higher failure rate and complication rate in eyes with pre-existing glaucoma. The aqueous shunts may accelerate corneal graft failure when there is tube–endothelium touch, or perhaps even when the tube tip rests near but does not actually hit the endothelium. Clinically localized corneal edema or focal fibrous metaplasia of endothelium may result when direct tube–endothelium touch occurs. Fibrous metaplasia can be limited and may not result in corneal edema, focal or diffuse. Possibly even normal blinking or intermittent eye rubbing may sufficiently indent the cornea or change a tube's position so as to push the endothelium into contact with a closely approximated tube tip. Topographically related corneal edema (wedge or localized) over the tube is logically interpreted as tube-caused. Even if actual physical contact between tube and endothelium does not occur, it has been theorized that there may be sufficient irritation from two-way flow in nonvalved shunts to explain localized endothelial damage. Diffuse endothelial failure (edema) after shunt installation or other forms of filtering surgery might more logically be explained via host defense immune reactions due to exposure to nonepithelialized bleb capsules (deficient blood–eye barrier). One-way valved shunts, one can argue, would seem less likely to facilitate such blood–eye interaction. Regardless, after PK, graft rejection is probably mediated via limbal–corneal vascularization and probably ultimately responsible for most graft failures, whether or not a shunt has been installed.

Regardless of the choice of device, major considerations when placing a shunt in any previously operated or complex case should include conjunctival status and the feasibility of wound closure. The most desirable site for insertion of a large single-plate explant such as a Baerveldt 350 mm ² or Molteno3 is superotemporally, although both can be safely placed with care superonasally, or alternatively in either inferior quadrant. Double-plate devices (Molteno or Ahmed) would generally be placed with the primary plate and tube in the superotemporal quadrant with the interconnecting tube over or under the superior rectus muscle. Right and left labels on the double-plate Molteno device place the primary plate nasally and date to times when extracapsular cataract surgery and large limbal wounds were routine. Nasal placement of the primary plate and tube left more room temporally for subsequent cataract extraction. Evolution of cataract surgery has largely obviated concern about which quadrant to use for the initial device. Smaller explants (single-plate Ahmed [pediatric polypropylene or silicone; or adult polypropylene or silicone], Baerveldt 250 mm ² , Molteno single-plate pediatric or Molteno adult-size single-plates) can all be placed relatively easily in any quadrant, but the superotemporal quadrant is preferable to minimize the risk of strabismus or globe displacement after bleb formation. Preexisting conjunctival scarring, especially in the perilimbal areas, may preclude easy installation of the connecting tube into the anterior chamber. The tubes on all devices are sufficiently long to permit routing around scarred perilimbal conjunctiva for installation via either the pars plana or anterior chamber in an adjacent superior quadrant if necessary. Good mobility of the perilimbal conjunctiva is probably the best indicator that installation of a shunt is feasible in that quadrant.