When Is There a Need for Keratoprostheses?
A keratoprosthesis (artificial cornea) needs to be considered when there is little or no prospect of success with conventional cadaveric corneal transplantation or limbal stem cell transplantation. This may be due to a hostile environment such as a dry keratinised ocular surface, lid deficiency, a highly vascularised cornea, or multiple previous corneal graft failure.

Classification of Keratoprostheses
There are a number of ways to classify keratoprostheses, based on the type of fixation (nut and bolt, intrastromal, epicorneal with tissue covering, supra-Descemetic) or the material of the haptic {PMMA e.g. Choyce and Boston (previously known as Dohlman Doane) devices, Dacron (Pintucci), titanium, ceramics, hydrogel (AlphaCor), silicone, expanded PTFE (Legeais), biological materials and analogue such as tooth and bone (Strampelli, Falcinelli), tibial bone.

To our knowledge, at present only the Boston keratoprosthesis, the Pintucci device, the Worst Singh device and the AlphaCor artificial cornea are commercially available, while the haptics of the osteo-odonto-keratoprosthesis (OOKP) and the tibia bone prosthesis are made usually from the patients' own tissue by the surgeon himself. This article focuses on the use of two distinctly different devices, the OOKP and AlphaCor artificial cornea.

The Osteo-Odonto-Keratoprosthesis
OOKP was first described by Strampelli in the early 1960s. Basically, a single rooted tooth root and surrounding jaw bone is fashioned into a plate measuring some 12-15 mm long and 3 mm thick, through which a PMMA optical cylinder is cemented into a hole previously drilled through this plate. The anterior surface of the plate is entirely bony, and the posterior surface mostly dentine. The osteo-odonto-acrylic plate is buried in a submuscular pocket to acquire soft tissue investment. At the same time as fashioning the plate or lamina, the ocular surface is cleared down to Bowman's membrane and exposed to the level of insertion of the recti muscles. The bare area is covered with buccal mucous membrane harvested from the inside of the cheek. Two to four months later, at stage 2 surgery, the lamina is retrieved and placed on the surface of the cornea, having trephined a corneal opening for the posterior part of the optical cylinder to protrude intraocularly. The iris, lens and anterior vitreous will also have been removed.

There was an initial international enthusiasm, but it was mostly forgotten after poor results were reported by some authors. The current technique is that of Falcinelli, who made stepwise improvement of the original technique from the 1970s to the 1990s. Falcinelli's excellent long term visual and retention results were independently surveyed and his modified technique began to spread after he took on three initial international students in the mid-1990s. The initial spread to Austria, Germany and England has propagated further to Egypt, Japan, Singapore, India and Hong Kong through the efforts of the OOKP teaching group.

The strength of the OOKP lies in its ability to be retained even in the dry eye situation. It is not entirely understood as to why the retention rate should be so good, but there are at least three factors. Firstly, there is gradation of materials from non living rigid PMMA to previously living rigid dentine, which in turn is interfaced with living porous jaw bone through living flexible alveolar ligament. Secondly, the jaw bone is interfaced with buccal mucous membrane which not only provides a blood supply to the bone, but also a biological seal against infection. Finally, the buccal mucous membrane is mechanically strong and biologically resilient, thus restraining (shrink wrapping) the OOKP complex even in a hostile ocular environment such as the dry eye.

Indications
The OOKP can be used to rehabilitate all cases of corneal blindness, but because of the gravity of the surgical programme, should be limited to end stage ocular surface diseases such as severe Stevens Johnson syndrome, ocular cicatricial pemphigoid, trachoma, other forms of cicatrising conjunctivitides, chemical injury, uncorrectable loss of eyelid, and absolute/severe dry eye states. It is not to be used in unilateral corneal blindness because of the severity of the surgical programme, the possibility of severe complications, the requirement for lifelong follow up, and because of the difference of image size the OOKP optics would cause, compared to a phakic or pseudophakic eye. In cases of bilateral corneal blindness, only one eye should be rehabilitated although attention should be given to the fellow eye for any glaucoma present. The fellow eye is kept as a spare eye.

Contraindications
The lower age limit is 18 or 19 years of age. Obviously, in eyes with no light perception, there is no hope for visual amelioration. Eyes with severe damage to the posterior segment, i.e. persistent retinal detachment, advanced glaucoma or damage to the optic nerve, should not be operated on either.

Patient Assessment
The patient and their family need to be assessed as a unit, to determine if they are committed to a severe surgical programme and lifelong follow up at the OOKP centre. The patient's psychological state and wish for regaining sight needs to be assessed. Prolonged blindness and onset of blindness can lead to depression and various states of dependence on spouse and family. Some patients are happily blind. Some can lead an independent life despite becoming blind, especially if the onset of blindness was at a young age. Some carers cherish the dependence of the blind relative they are caring for, and may subconsciously resent to their regaining independence. Others may push the patient towards surgery when the patient is not seeking surgery. All in all, it is important that it is the patient who wants to have surgery to restore sight, but both the patient and relatives must understand the severity of the surgery, the possibility of serious complications and thus the possibility of further surgery, and the requirement of lifelong follow up.

Ocular assessment consists of full history and examination, concentrating on ascertaining retinal and optic nerve integrity. Previous surgery (especially lid, cornea, lens) and ocular perforation should be noted. Attention should be paid to the possibility of glaucoma, which may be worsened following OOKP surgery due to unintended additional surgical alterations of the anterior chamber angle and the inefficacy of topically applied glaucoma drugs, which will not be absorbed into the anterior chamber through the buccal mucosa covering the haptic. The axial length needs to be measured to determine the optical power of the optical cylinder to be used. Typically, ocular examination includes assessment of relative afferent pupil defect, projection of light in quadrants (patients usually have PL, HM or at best CF vision by the time they come for OOKP surgery), slit-lamp examination, digital palpation of intraocular pressure (the corneae are usually too dry and scarred for applanation tonometry), B-scan to exclude and assess retinal detachment, peripheral anterior synechiae, lens status, and pre-phthisis presenting as a short eye, and A-scan for biometry. Very few cases require electrodiagnostic tests which also cannot be entirely relied upon.

Oral assessment is best done by an oral surgeon, who will also be working on harvesting the tooth together with root and surrounding jaw bone, and buccal mucous membrane. Oral hygiene is assessed. The presence of canine teeth as well as gum and bone recession is assessed, followed by imaging of relevant teeth and their juxtaposition to adjacent teeth with orthopantomography (a form of dental panoramic radiography), individual canine teeth x-rays, and even a spiral CT scan. A single-rooted tooth, usually a canine, with a straight long root of good girth, good bone quality as close to the base of the crown, and good separation from its two adjacent teeth is selected. Buccal mucous membrane is also assessed for area available, quality and whether previous harvesting has taken place and if so on which side. Sometimes, it is necessary to harvest lower labial mucous membrane for lining the upper and lower eyelids.

Results of OOKP
The performance of any keratoprosthesis can be measured in terms of a number of parameters including retention (long-term anatomical success), vision (visual acuity, visual field, quality of vision include glare, etc.), patient satisfaction including cosmesis, and sight-threatening complications such as glaucoma, retinal detachment, etc. The size of the visual field depends on the geometry of the optical cylinder and therefore on the size of the available tooth. But often it is much less than its theoretically calculated value. With a classical PMMA cylinder it varies between 30 and 50° and can be enlarged by a larger or conical posterior part of the optic. The visual acuity may be as good as 1.5 (20/12 = 6/4) depending only on retinal status.

Falcinelli et al. reported an anatomic success rate of 96.5% after a follow-up of approximately 5 years, 94.1% in 10 years and 88.8% in about 20 years with a maximum of follow-up of 27 years In patients with dry eyes, which represent 41.5% of all of Falcinelli et al.'s patients, an improvement of visual acuity at long-term follow-up was found in over 90%of cases.

In a smaller German series of 25 patients, one-third had excellent visual acuity of 0.9 or better, two-thirds reading vision (-0.5 or 6/12) and 80% ambulatory vision (>0.05 or 1/20). To compare the results of OOKP with other keratoprostheses, figure 4 shows the Kaplan-Meier anatomic survival analysis of K.H.'s series (Homburg) of different keratoprostheses (25 patients with an OOKP, 4 with a tibial-bone KPro, 5 with a Pintucci KPro, and 1 with Legeais keratoprosthesis). Figure underlines the superiority of OOKP (and the biologic support tibial bone) over other alloplastic haptics.

At the Sussex Eye Hospital in Brighton, we have operated on 35 cases since 1996 (21 male, 14 female, with a mean age of 50.6 years at stage 2). Presenting diagnoses were: 15 Stevens-Johnson syndrome, 5 chemical injury, 6 ocular cicatricial pemphigoid, 4 miscellaneous dry eye diseases, 1 trachoma, 1 linear IgA disease, 1 ectodermal dysplasia, 1 post-bomb blast, and 1 congenital trigeminal nerve hyoplasia. 34/35 proceeded through stage 1 and stage 2 (1 developed a total retinal detachment between stage 1 and stage 2). Mean time interval between stages 1 and 2 was 19.8 weeks (range 9-84). Preoperative visual acuity was: PL (15), HM (15), CF (5). Postoperative visual acuity at last follow-up was: 6/5 (4), 6/6 (2), 6/9 (5), 6/18 (3), 6/24 (1), 6/36 (2), <6/60 (2), HM (4), CF (1), PL (3), andNPL (6) (2 patients required subsequent evisceration). Mean follow-up was: 34.7 months (range 1-96). There were 3 allografts. 7 exhibited clinical resorption of OOKP lamina. 2 subsequently required repeat OOKP surgery. Postoperative complications (episodes, not eyes) were: 5 retro-prosthetic membranes, 4 retinal detachments (including the 2 secondary to laminar resorption and endophthalmitis: see below), 3 vitreous haemorrhages, 3 extrusions of lamina, 2 infections of buccal membrane graft, 2 endophthalmitis, 1 expulsive haemorrhage and 1 epiretinal membrane. OOKP surgery is usually carried out in two stages. In the first stage a monoradicular tooth is harvested to prepare an osteo-odonto-lamina.

The root and surrounding jaw bone is sliced sagittally, whilst the crown is grasped with extraction forceps, to expose pulp which is removed. A hole is drilled through dentine through which the anterior part of a PMMA optical cylinder is cemented in place (there are various designs and sources of OOKP optical cylinder, the Brighton school uses cylinders from Lamda Polytech Ltd, Brackley, Northants., UK, the cylinders implanted in Germany had been fabricated by Morcher GmbH, Stuttgart, Germany). The crown is removed prior to drying with filtered oxygen and cementing of the optical cylinder. The saw, flywheel and drill and burr tips are constantly irrigated with balanced salt solution to provide cooling. Where periosteum has been detached, it is glued back with fibrin glue. The KPro is then implanted into a submuscular pouch (often the lower eyelid of the fellow eye) for a period of 2-4 months. A tooth allograft can be considered in edentulous patients, but HLA-matching, screening for blood-borne infections, and long-term immunosuppression with cyclosporin will be required. An allograft may be more rapidly resorbed compared with an autograft. One of us (K.H.) prefers using tibial bone in edentulous patients, but it is known that a tibial bone haptic is resorbed three times as fast as the osteo-odontolamina.

A buccal mucous membrane graft of about 3 cm diameter is used to cover the OOKP lamina, as there are stem cells present, it has proliferating capability and is adapted to high bacterial load. It will be vascularised by the time of stage 2 surgery and will provide the blood supply to the bone part of the OOKP lamina. Once harvested the fat from the buccal mucous membrane graft is removed with curved scissors and the graft soaked in an antibiotic solution until required. The eye is prepared by isolating the recti with stay sutures, a 360° per-itomy performed and the conjunctiva and tenons separated from underlying sclera. Corneal epithelium and Bowman's membrane are removed. The buccal mucosa is then trimmed to obtain an oval piece of adequate size to fit snugly on the front of the eye. The mucous membrane graft is sutured onto the side of the insertion of the four recti muscles and to the sclera with interrupted 6-0 vicryl. If possible, the cut edge of the graft should also be sutured to the conjunctiva.

When Not to Do Ocular Surface Reconstruction and Tooth Harvesting Together If the eye is very dry or there is a risk of the mucous membrane graft not taking, it may be better to perform stage 1 surgery in two steps. The mucous membrane graft to the eye is done first, and it is only when the graft has been shown to be well established before the patient is readmitted for tooth harvesting and preparing an OOKP lamina. Otherwise if there is a significant delay in mucous membrane healing, or if further partial or full repeat mucosal grafting proves necessary, the lamina may be resorbed whilst buried in the lid for an excessively long time. Stage 2 surgery is carried out 2-4 months after stage 1 in order for soft tissue to become integrated into the bone pores of the lamina. The interval also allows the lamina to recover from thermal damage, and any infection introduced from the oral cavity can be treated whilst the lamina is in the submuscular rather than on the eye. If the lamina is implanted submuscularly for a longer period of time, there may be significant resorption of the lamina. The first step in stage 2 surgery is to retrieve the buried lamina for inspection. It is only if this is of adequate size that the surgeon proceeds to prepare the eye for receiving the device.

After the OOKP lamina is retrieved from its submuscular pocket, soft tissue is excised from the posterior surface and the pseudocapsule that has grown around the OOKP is opened in the corners to form four pentagonal-shaped flaps of fibrovascular tissue which are used to fix it to the recipient sclera. Excess tissue is also trimmed from the anterior surface of the implant. A template is made of the lamina in order to plan placement of a Flieringa ring, and preplaced sutures for securing the lamina. The lamina is temporarily returned to its submuscular pocket until the cornea is about to be trephined. Traction sutures are applied to the lids for access to the eye. A superior rectus stay suture is placed and a buccal graft flap is fashioned by making an arcuate incision from 3 to 9 o'clock under constant irrigation with BSS and adrenaline. The flap is reflected and the cornea exposed. The buccal mucous flap is then reflected and a Flieringa ring sutured in place with sutures left long at 3 and 9 o'clock for traction. The centre of the cornea is marked and the template placed on the cornea and cardinal sutures are preplaced. Intravenous mannitol has by then been administered to reduce the intraocular pressure before trephination.

The cornea is partially trephined, the size depending on the diameter of the posterior part of the optical cylinder. This is completed with scissors or a blade. The iris is then completely removed using forceps. If the patient is phakic the lens is removed either by ICCE or ECCE (Falcinelli advocates an ICCE as he fears a capsular rim may occlude the trabecular meshwork. The Brighton school favours ECCE to avoid traction on the vitreous base). A posterior capsulotomy is made (if an ECCE has been used) and an anterior vitrectomy performed, with adequate traction provided by the surgical assistant on the two Flieringa ring sutures. The lamina is then sutured to the cornea with the posterior part of the optical cylinder traversing the corneal opening. Sterile air is then injected to reinflate the eye and indirect fundoscopy performed to ascertain adequate centration, to take note of the appearance of the posterior pole of the eye, and any presence of blood in the vitreous. Further interrupted sutures are applied to secure the lamina onto the sclera. The Flieringa ring is then removed, the buccal mucous membrane is repositioned and sutured in place, with a hole cut through the membrane to allow the anterior part of the optical cylinder to protrude.

Complications - Diagnosis and Management Complications can be divided into operative and postoperative complications. Significant damage to adjacent teeth is uncommon. Very occasionally, the maxillary sinus space may be breached. Facial and jaw fractures are also possible. Overheating of the OOKP lamina can take place without adequate irrigation. Buttonholing the buccal mucous membrane can also take place and requires suture repair. Vitreous haemorrhage can take place, especially if the surgeon does not wait long enough for the iris root bleeding to stop following total iridodialysis. Expulsive choroidal haemorrhage is possible. Mannitol infusion before opening the eye, and hypotensive anaesthesia help to prevent this catastrophic complication. The optical cylinder may become tilted if suture tensioning of the OOKP lamina is not done correctly. Among the most common postoperative complications is ulceration or necrosis of the mucous membrane graft. This is less common with a vascularised cornea, otherwise tenons tissue should be mobilised to lie over the cornea and to help supply the central part of the buccal mucous membrane graft with blood.

When the mucous membrane should thin or ulcerate, it is important to exclude infection and treat with prophylactic topical and systemic antibiotic. The thinned or ulcerated area, if not healed, will necessitate a bipedicular flap from the peripheral part of the graft, followed by a new graft to the peripheral bare area thus created. The periphery often has a better blood supply. Occasionally it is necessary to repeat the buccal mucous membrane graft altogether. If this has to be done following stage 2, then it will be necessary to remove the device and close the corneal opening with a small corneal graft whilst the device is returned to a submuscular pocket usually under the lower lid of the contralateral eye. It is also possible to use lid skin as cover as a last resort, having removed muscle and fat from the skin first. Glaucoma may be pre-existent or secondary to OOKP surgery. This needs to be tackled, otherwise continued field loss and finally loss of central vision will ensue despite technical success and retention of the device. Assessment of glaucoma is difficult but it is possible to estimate intraocular pressure with digital palpation through the upper lid, with the patient looking down to avoid the rigid OOKP lamina.

Other parameters which can be assessed include optic disc appearance, visual field examination and electrodiagnostic tests. Treatment is limited to systemic carbonic anhydrase inhibitor and surgery (cyclode-struction or drainage tube) which has not yet been worked out fully. It is not known how the OOKP lamina is absorbed, but it is in an ectopic situation, and the fact that it no longer bears a load (it previously did with mastication) may be responsible for increased osteoclastic activity. Bacterial action around the area bound by the mucous membrane graft opening may also play a part. The strategy has been to insert an adequately sized lamina of a minimum thickness of 3 mm throughout to allow for loss. The diameter of the cylindrical opening through the lamina is dictated by the diameter of the anterior part of the optical cylinder. For a thin root, a narrower optical cylinder needs to be used. When there is inadequate bone and dentine, the lamina (or the optical cylinder) can extrude through the overlying buccal mucous membrane. The subsequent entry of microorganisms can lead to the permanent loss of the eye. Thus the laminar dimensions are assessed at each clinic visit, using a cotton bud, the stability of the optical cylinder ascertained, including measuring the refraction, since any change in refraction is most likely to have been caused by an axial shift or tilt of the optical cylinder.

Serial photography from the side will also document and demonstrate protrusion of the anterior optical cylinder using the buccal mucous membrane as a reference point. In the absence of a change of refraction, such a protrusion may be due to thinning of the mucous membrane or the OOKP lamina, or a combination of both. Even though imaging will be at the limits of resolution, spiral CT and electron beam tomography have been useful in confirming clinical suspicion, leading to fashioning of a new lamina ready for lamina exchange after 2-3 months, in anticipation of device failure. Retinal detachment is a distinct possibility, especially in a young person following disturbing vitreous. Patients are warned of symptoms of posterior vitreous detachment, and each clinic follow-up is accompanied by a B-scan of the fundal periphery which cannot be viewed through the optical cylinder. Repair is now usually done with the BIOM system, although the endoscopio vitrectomy approach is sometimes called for. Neither technique requires the removal of the OOKP device.

AlphaCor The Device An alternative to rigid KPro technology is the AlphaCor™ artificial cornea (Addition Technology Inc., Des Planies, 111., USA) which has been described previously and surgical techniques discussed]. It is a flexible poly(2-hydroxyethyl methacrylate) (PHEMA) core-and-skirt keratoprosthesis with a peripheral macroporous skirt region that biointegrates through tissue ingrowth surrounding an effectively non-porous transparent optic. It is implanted within a lamellar pocket of which the central 3.5 mm posterior to the optic is removed, with the optic becoming a full-thickness corneal replacement 3 months later at stage 2 of the procedure, when tissue anterior to the optic is removed. A typical postoperative appearance of the device in the eye before and after stage 2 opening is seen in figure. A Gunderson flap, as pictured in figure, was originally felt a necessary adjunct to AlphaCor implantation, but it is now rarely performed. The device is designed for use as a corneal replacement in an eye with a reasonable corneal tear film, but may be used in mild-moderate dry eye states (if appropriately managed, for example with artificial tears, lateral tarsorrha-phy, bandage lens and some surgeons believe, with restasis), and in eyes with limbal stem cell abnormality.

Although a specially modified earlier prototype was evaluated in animals for a severely dry eye situation, this has not been developed for human application to date. The hydrophilic nature of PHEMA requires it to be positioned in a wet environment such that it retains a tear film to provide a good quality refractive surface. An inadequate or inflammatory tear film could increase the risk of postoperative stromal melting around the device, and of contact-lens type deposition on the optic. AlphaCor's conservative design, which lacks protuberant parts so as to minimise mechanical stresses, does entail the optic being recessed in relation to surround corneal tissue, and this impacts adversely on optic wetting and refractive performance in some cases. Indications AlphaCor has gained widespread regulatory clearance for use in adults with corneal opacity unsuited to a corneal graft due to a high risk of failure. Nearly 90% patients to date had previous graft failures (1-13, mean 2.4). Glaucoma affected 56.0% cases preoperatively and 24.9% had a drainage tube in situ. A history of chemical injury was reported in 12.6% of the series, and aniridics represented 6.9%. Detailed data concerning patient profiles and outcomes are voluntarily compiled and updated in the manufacturer's anonymous database, and available to user surgeons. Outcomes 300 AlphaCors have been implanted to date, with a maximum of just over 7 years follow-up, mean over 1 year.

Protocol cases have a better chance of 1-year retention in situ than a donor graft in similar preoperative conditions, as has been discussed previously, and risk and protective factors for the device have been described elsewhere. Risk factors for graft survival, such as vascularisation, number previous failed grafts, and glaucoma tubes, do not appear to impact adversely on AlphaCor outcomes, A system for evaluating risk of graft or AlphaCor success preoperatively has been suggested but requires more data to validate. Postoperative best corrected visual acuity outcomes for AlphaCor range between Light Perception and 6/6, with patients demonstrating a mean gain of 2.5 lines visual acuity. The best corrected visual acuity achieved with AlphaCor matches that (paired t-test) of patients achieved from their previous donor tissue graft prior to its failure. Discussion The choice of keratoprosthesis in the severely dry eye is straightforward, as only one device the OOKP will work. With careful assessment, adequate technique, regular follow up and early recognition and management of complications, most patients can look forward to many years of sight with an OOKP.

The surgery is complex and requires extensive support including long general anaesthesia, as well as input from oral surgical, oculoplastic, glaucoma, vitreoretinal and radiological colleagues. OOKP surgery should be confined to specialist centres. Shared care with the referring ophthalmologist is only appropriate if the referring ophthalmologist is willing to be rapidly available and is capable of recognising and treating complications of OOKP surgery. In the appropriately managed marginally dry eye, an AlphaCor keratopros-thesis may be considered, although the device should be seen as an alternative to high-risk keratoplasty in multiple graft rejection and vascularised corneae. Despite encouraging results from AlphaCor in eyes that would be high risk for penetrating keratoplasty and falling incidences of complications as described previously as risk and protective factors have been learned and management improved, this is clearly a device still very much on the learning curve and there is room for continued development and improvement of techniques and postoperative management to optimise outcomes.

Current data support the concept that patient selection is critical for success and suggest that AlphaCor outcomes result in higher probabilities of success at 1 and 2 years than is achieved by high-risk cases undergoing conventional donor penetrating keratoplasty. However, ongoing data collection and analysis will be critical in determining definitively which patients should have AlphaCor and which should have a further graft or other procedure in order to achieve the best possible outcome. KPro surgery is a complex and growing field. Those interested are invited to join the KPro Study Group (www.kpro.org), to add their contribution to the ongoing research and device evaluation.