After 1906, when the first human-to-human corneal graft was performed, there was a loss of interest in keratoprosthesis development and surgery. In 1920, Verhoeff reported on a single case of insertion of a quartz button into a patient's cornea; however, it had to be removed shortly afterward.⁷ Similarly, Filatov,⁸ in 1935, implanted a full penetrating glass device into a patient's opacified cornea and covered it with a double conjunctival flap after surgery. He left the flap in place, and after 6 months it had thinned sufficiently to give the patient an ambulatory vision of 1/200. More details on the history of keratoprosthesis surgery around this period have been reported previously in reviews by Day,⁹ Cardona,¹⁰ and Lund.¹¹

During World War II, Wunsche,¹² Stone,¹³ and others noticed that polymethylmethacrylate (PMMA) splinters embedded in the cornea of pilots were well tolerated. This led to their experiments showing that PMMA discs could be retained in the cornea of rabbits. Soon human applications followed, and many ophthalmologists attempted to refine their procedure using these new inert plastics, primarily PMMA. However, once again, many of these cases were associated with severe complications, and the procedure lost favor. Some surgeons did persevere in their techniques and refined them over the years. However, the combined experiences of their keratoprosthesis surgeries probably did not amount to more than four or five thousand cases during the past half century, which is quite small compared with the number of penetrating keratoplasties carried out on a worldwide basis.

The material most frequently used for the central optical portion of the keratoprosthesis has been medical-grade PMMA. PMMA is completely transparent and biologically inert, and there is now extensive experience with its use in intraocular lenses. However, many different types of material have been used for the skirts or plates used to secure the optical stem in the cornea, such as perforated grids of PMMA, Teflon, nylon, and ceramics. More recently, attention has been directed to porous designs using materials such as Proplast,¹⁴ polytetrafluorethylene,¹⁵ hydrogels,^16,17 collagen,¹⁸ and various copolymers.^19,20 These materials have been manufactured and developed with the idea that colonization with natural tissue elements would subsequently hold and anchor the device more securely in place, thus decreasing the likelihood of future extrusion. In the 1960s, Strampelli²¹ introduced a novel idea of inserting the PMMA stem into a slice of dental bone from the patient's jaw, with the hope that the bone would subsequently heal better into the patient's cornea. This is, however, a formidable multistage procedure. DeVoe and Cardona²² made an important observation that in an extremely dry eye, if the prosthesis is allowed to protrude through the skin rather than between the lids, safety is enhanced and the extrusion rate is subsequently reduced. Considerable modifications to various keratoprosthesis designs have occurred over the years, but whether an ideal keratoprosthesis can ever be designed that will integrate into the human cornea without the risk of extrusion or necrosis has yet to be determined.

My practice is to use an all-PMMA device like that shown in Figure 1 (Dohlman-Doane Keratoprosthesis types I and II). The standard type I device is used in all but extremely dry eyes, where a “through-the-lid” approach with a type II device is preferred. In the latter case, the anterior nub is designed to protrude through an opening in the lid skin. The dimensions described for the two devices allow satisfactory visual fields. Thus, the type I permits a peripheral field of approximately 60 degrees; type II, with a longer stem, can give a visual field of approximately 40 degrees. The manufacture of these devices has previously been described.²³ The Strampelli and Cardona devices are the other most frequently used models.

Of the patients in whom keratoprosthesis could be considered, the patients with the poorest outlook are those with Stevens-Johnson syndrome. These patients have binocular disease and are sometimes monocular, because they frequently have undergone numerous unsuccessful surgical procedures in the past. A further difficulty with this group is that they are young and require the keratoprosthesis to remain complication-free for many years. As already mentioned, they often have ongoing ocular inflammation, which increases the postoperative complication rate. Accordingly, at this time I advise against keratoprosthesis surgery in young patients with Stevens-Johnson syndrome; I would prefer waiting until techniques have improved.

Patients with ocular cicatricial pemphigoid tend to be older, and this is in their favor. They also appear to have a more favorable outcome long term. However, they are prone to postoperative skin retraction and glaucoma.

Chemical burn patients can have good results after keratoprosthesis surgery. A simultaneous glaucoma shunt (Ahmed valve) is nearly always placed in these inflammatory conditions: the incidence of postoperative long-term glaucoma without such a shunt is extremely high.

Perhaps the most controversial category of patients are those with previous graft failures after noninflammatory edema, previous dystrophies, or trauma. The frequency of postoperative uveitis and uncontrolled glaucoma in these patients is quite low, and vision is usually restored fairly rapidly, in most cases more rapidly than a successful regraft would permit. Many of these are elderly patients with graft failure after bullous keratopathy. In certain instances, these patients may have a better chance of seeing well during their remaining days with a keratoprosthesis rather than with another corneal graft.²⁴

In the early postoperative period, it is desirable that the conjunctival flap does not retract, because attempts at repair are usually futile. Extensive surgery, such as is done in keratoprosthesis, is associated with considerable postoperative inflammation. Accordingly, we have found that small doses of corticosteroids are needed, especially in eyes with a previous history of severe inflammation (e.g., Stevens-Johnson syndrome, pemphigoid, chemical burns). Routinely, 10 mg dexamethasone is given by peribulbar injection through the lower lid at the completion of surgery. After 2 to 3 weeks, 20 to 40 mg triamcinolone may be administered by peribulbar injection if inflammation persists. The total dose of corticosteroids may be difficult to judge: too much may jeopardize wound healing, leading to necrosis, fistulas, or leak of tears. Once the flap has been opened, it is easier to monitor the anterior chamber and vitreous reaction. If postoperative uveitis is still present, peribulbar steroids may have to be continued. Occasionally, in patients with Stevens-Johnson syndrome or ocular cicatricial pemphigoid, systemic immunosuppression in the form of oral steroid treatment or other immunosuppressive agents is indicated.

Intravenous antibiotics are administered at the time of surgery. After surgery, patients are given topical antibiotic drops and a 14-day course of oral antibiotics. We usually prescribe topical antibiotic drops for a prolonged period. These are rotated every 2 to 3 months to prevent the development of resistant organisms. Patients with Stevens-Johnson syndrome who are particularly at risk for infection need to be monitored closely and should probably take topical antibiotics for life. After approximately 2 months, when the conjunctival flap or lid skin is opened, the patient is given topical anticollagenase medication. We use 1% medroxyprogesterone, which reduces collagenase synthesis, and a 1% tetracycline solution, which is a potent direct enzyme inhibitor. Although the efficacy of these agents in reducing corneal necrosis has not been established in humans, their beneficial effects have been well documented in experimental animals.^26,27 Unfortunately, these suspensions are not commercially available and have to be made up in a hospital pharmacy.

Patients are seen frequently in the early postoperative period and are advised to report any symptoms of pain, lid edema, blurred vision, or discharge from the eye. The goal should be to achieve a noninflamed eye with intact tissue around the device (Figs. 3 and 4).

The major complication is necrosis of tissue around the keratoprosthesis (Fig. 5). Unless treated, this can lead to subsequent tissue breakdown (melt), aqueous leak, infection, or retinal detachment. Certain measures reduce the risk of such calamities. Thus, the type I keratoprosthesis in the wet and adequately blinking eye is usually covered with a total conjunctival flap. This is left in place for 2 months before opening, which reduces the incidence of subsequent necrosis. The use of collagenase enzyme-suppressing medication also lowers the incidence of this complication. However, in some cases necrosis still occurs. It typically begins around the edge of the cornea adjacent to the optical stem. If identified early, various repair mechanisms can be attempted to halt the process. A thin slice of corneal tissue with a central 3-mm trephined hole and a radial cut can be slipped under the anterior plate and sutured in place.²⁵ In many instances, this reinforces the base of the keratoprosthesis and allows further stabilization of the device. In some cases, this new donor cornea material becomes vascularized early; this is a favorable sign because it generally is associated with long-term retention of the added tissue. Once repaired, the patients need to be followed up closely. In some of these patients, more melting will occur, and they may require a further procedure. However, with the passage of time, the need for subsequent repair decreases markedly.

Skin retraction away from the plastic stem may occur with the type II keratoprosthesis. If allowed to proceed, it can again lead to subsequent leakage, infection, and extrusion. In the early stages, retraction can be repaired by mobilizing the surrounding skin and suturing it to keep the skin snug against the optical stem. Repeated skin revisions may be necessary in the initial postoperative period (see Fig. 5), but, once again, with the passage of time the situation tends to stabilize, with fibrosis occurring around the base of the keratoprosthesis stem.

Postoperative uveitis occurs frequently after keratoprosthesis surgery. It can be associated with subsequent secondary problems, such as the formation of retroprosthetic membranes and vitreous changes. This usually abates after a month or two. Long-term uveitis is rare but can result in persistent vitreous haze and membrane formation, retinal pathology, and even detachments. As mentioned previously, the use of corticosteroids by the peribulbar route, in particular in the early postoperative period, has reduced the incidence of uveitis significantly and has also reduced the incidence of secondary retroprosthesis membranes.

With the more liberal use of corticosteroids in the postoperative period, retroprosthetic membranes are less common. When they do occur, it is usually possible to open the membranes using the YAG laser. The back plate of the keratoprosthesis has a laser ridge that will keep the membrane away from the stem. In a few patients, a dense retroprosthesis membrane that is not amenable to opening with the YAG laser may require surgical opening with a Zeigler knife or a vitrectomy cutting instrument.

Glaucoma is a common complication after keratoprosthesis surgery and is one of the more common direct causes of eventual blindness. Its pathogenesis is probably multifactorial. It may be associated with prolonged uveitis, giving rise to trabecular meshwork blockage by inflammatory cells or the development of peripheral anterior synechiae. One of the major difficulties with these patients is monitoring intraocular pressure after surgery. Obviously, tonometers are useless in this setting, and digital palpation of the globe is the only method available to obtain a rough estimate of intraocular pressure. Disc appearance can be assessed using the 90 diopter lens or direct ophthalmoscopy, and disc photographs can be taken at repeated intervals. Visual fields should be monitored on a frequent basis and can also be used to assess glaucoma damage. Treatment is limited because the topical absorption of antiglaucoma medications is impossible in the type II (through-the-lid) patients. In the type I setting, topical medication has some penetration, as can be demonstrated by the action of topically applied phenylephrine on the pupil. Systemic carbonic anhydrase inhibitors are the mainstay of treatment. Bearing these difficulties in mind, in both the postoperative monitoring and treatment of glaucoma, most of our patients now receive an Ahmed valve glaucoma shunt, which is implanted together with the prosthesis.²⁹

Retinal detachment is fortunately not a common complication. It may be rhegmatogenous in nature and is probably associated with peripheral retinal breaks. In some patients, tractional retinal detachment occurs, particularly after chemical burns. Presumably this is secondary to the development of inflammatory vitreoretinal membranes stimulated by prolonged postoperative uveitis. Unfortunately, once a retinal detachment has occurred, treatment is rarely successful.

Infection is uncommon in the early postoperative period, but endophthalmitis can occur at any stage after surgery. Early detection and treatment of tissue melts can help reduce this complication. Nonetheless, patients sometimes present with a fulminant endophthalmitis with no obvious source of leakage around the keratoprosthesis. The outcome is usually devastating. It is our impression that patients with Stevens-Johnson syndrome are particularly prone to infection; this may be due to the fact that their tissue damage is so profound from chronic inflammation. Accordingly, we treat these patients with prophylactic antibiotics, both during and for prolonged periods after surgery.

The need for frequent follow-up and for repair of threatening complications should not be minimized. After type II surgery, skin revision is often indicated. After type I surgery, insertion of repair tissue behind the plate may be necessary many times. Also, YAG laser opening of membranes has frequently been necessary in both types of surgery. Thus, in 36 of 60 eyes, some surgical or laser repair had to be done, in many cases repeatedly. In 6 cases, the keratoprosthesis operation had to be repeated. After the first year or several years, the situation tends to stabilize and the need for repair becomes much less frequent.

The various causes of corneal blindness differ markedly in prognosis (Table 1). This difference is related to the degree and chronicity of previous inflammation in the eye. Thus, patients who have had repeated graft failures after edema, dystrophy, and so forth, with little inflammation, have had relatively little trouble after keratoprosthesis. At the other end of the spectrum, Stevens-Johnson syndrome carries the worst prognosis. Of seven such cases followed for at least 3 years, only one patient now sees well. It appears that the chronic mucous membrane inflammation makes the tissue around the prosthesis vulnerable to aseptic necrosis and melt, which can then lead to sudden and devastating endophthalmitis years after the initial surgery. Ocular pemphigoid and severe chemical burns occupy a middle ground in terms of outcome.

TABLE 1. Prognostic Categories in Keratoprosthesis

  Noninflammatory conditions such as graft failure and dystrophies
  Graft failure after inflammatory disease such as herpes simplex and zoster keratitis
  Ocular cicatricial pemphigoid
  Chemical burns
  Stevens---Johnson syndrome

The most serious complications seen have been seven cases of endophthalmitis and four cases of retinal detachment. These were mostly in patients with Stevens-Johnson syndrome. To put these complications into context, most of the eyes in the whole patient group had very severe disease in the past that frequently involved ocular tissues beyond the cornea itself

As mentioned earlier, glaucoma is often already present in cases requiring keratoprosthesis, and it is described as a frequent postoperative cause of long-term failure and blindness (Fig. 6). Glaucoma valve shunts, implanted at the same time as the keratoprosthesis or at a later date, have substantial prophylactic value. Of shunts implanted in 40 cases, there have been few complications, and in only 5 of these cases (4 of them with preexisting end-stage glaucoma) was further deterioration of the fields observed.²⁹ Glaucoma shunt implantation constitutes a major advance in keratoprosthesis surgery.

On the positive side, if complications have been minimal or absent and if the posterior segment has been normal, the vision obtained has often been spectacular. Thus, if the Stevens-Johnson syndrome cases are excluded, two thirds of the patients have ended up with vision between 20/200 and 20/20. Even including Stevens-Johnson syndrome patients, the percentage of substantial visual improvement is more than 50%. Length of follow-up is clearly a factor in evaluating success: it is always uncertain for a given patient how long successful vision can be retained.

Not all cases that end up with a disastrous complication such as blindness from glaucoma, endophthalmitis, or retinal detachment can be dismissed as total failures. In the case of a young patient, bilaterally blind from Stevens-Johnson syndrome, who receives a keratoprosthesis and excellent vision but after a year loses the eye from infection, the exercise must be deemed a cruel failure. However, an elderly person with pemphigoid who gets 5 years of good vision from the surgery before losing it again is, in the end, usually still grateful for what was achieved.

1. Pellier de Quengsy G: Precis au Cours d'Operations sur la Chirurgie des Yeux. Paris: Ditot, 1789

2. Nussbaum N: Cornea Artificialis, ein Substitut fur die Transplantatio Cornea. Deutsche Klinik 34:367, 1853

3. Heusser J: Die Einheilung einer Cornea artificialis. Oesterr Ztschr Pract Med 26:424, 1860

4. Dimmer F: Zwei Falle von Celluoidplatten der Hornhaut, Kim. Monatsbl Augenh 29:104, 1981

5. Salzer F: Uber den kunstlichen Hornhautersatz. Wiesbaden, 1898

6. von Hippel A: Uber die operative Behandlung totaler stationarer Hornhaut-Trubungen. Graefes Arch Clin Exp Ophthalm 23:79, 1887

7. Verhoeff FH: Cited in Cardona H: Keratoprosthesis. Am J Ophthalmol 54:284, 1962

8. Filatov VP: Alloplastik bei vollstandig “hoffnungslosen” Leukomen. Sov Viest Opht 9:400, 1936

9. Day R: Artificial corneal implants. Trans Am Ophthalmol Soc 55:455, 1957

10. Cardona H: Keratoprosthesis. Am J Ophthalmol 54:284, 1962

11. Lund O-E: Die Keratoprosthese. Berichte Deutsche Ophthalmologische Gesellschaft, p 173. Munich: Bergmann, 1977

12. Wunsche G: Versuche zur totalen Keratoplastie und zur Cornea arteficialis. Arztliche Forschung 1:345, 1947

13. Stone W Jr, Herbert E: Experimental study of plastic material and replacement for the cornea: Preliminary report. Am J Ophthalmol 36:168, 1953

14. Barber JC, Feaster F, Priour D: The acceptance of a vitreous carbon alloplastic material, Proplast, in the rabbit eye. Invest Ophthalmol Vis Sci 19:182, 1980

15. Legais J-M, Rossi C, Renard G et al: A new flurorocarbon for keratoprosthesis. Cornea 11:538, 1992

16. Trinkaus-Randall V, Capecchi J, Newton A et al: Development of a biopolymeric keratoprosthesis material. Invest Ophthalmol Vis Sci 29:393, 1988

17. Crawford GJ, Constable IJ, Chirila TV et al: Tissue interaction with hydrogel sponges implanted in the rabbit cornea. Cornea 12:348, 1993

18. DeVore D, Kelman CD, Stark W et al: Evaluation of collagen-based corneal grafts in the rabbit model. J Refract Corneal Surg 9:208, 1993

19. Kain HL: The silicon-carbon keratoprosthesis. Design and implant procedure [abstract]. Invest Ophthalmol Vis Sci 33:992, 1992

20. Kirkham SM, Dangel ME: Keratoprosthesis: Improved biocompatibility through design and surface modification. Ophthalmic Surg 22:455, 1991

21. Strampelli B: Osteo-odonto-keratoprosthesis. Ann Ottalmol Clin Ocul 89:1039, 1963

22. Cardona H, DeVoe AG: Prosthokeratoplasty. Trans Am Acad Ophthalmol Otolaryngol 83:271, 1977

23. Doane MG, Dohlman CH, Bearse G: Fabrication of a keratoprosthesis. Cornea 15:179, 1996

24. Donn A: Aphakic bullous keratopathy treated with prosthokeratoplasty. Arch Ophthalmol 94:270, 1976

25. Dohlman CH, Waller SG, Netland PA: Keratoprosthesis surgery. In Lindquist TD, Lindstrom RL (eds): Ophthalmic Surgery Update 4. Chicago: Mosby Year-Book, 1997

26. Newsome DA, Gross J: Prevention by medroxyprogesterone of perforation in the alkali-burned rabbit cornea: Inhibition of collagenolytic activity. Invest Ophthalmol Vis Sci 16:21, 1977

27. Seedor JA, Perry HD, McNamara TF et al: Systemic treatment of alkali-induced corneal ulceration in rabbits. Arch Ophthalmol 105:268, 1987

28. Dohlman CH, Doane MG: Some factors influencing outcome after keratoprosthesis surgery. Cornea 13:214, 1994

29. Netland PA, Terada H, Dohlman CH: Glaucoma associated with keratoprosthesis. Ophthalmology 105;751, 1998

30. Hicks CR, Chirila TV, Dalton PD et al: Development of an artificial corneal button for penetrating keratoplasty; design, biocompatibility and results in animals [abstract]. Invest Ophthalmol Vis Sci (Suppl) 37:2524, 1996

31. Hicks CR, Chirila TV, Clayton AB et al: Clinical results of implantation of the Chirila keratoprosthesis in rabbits. Br J Ophthalmol 82:18, 1998

32. Legais et al: Keratoprosthesis with biocolonizable microporous fluorocarbon haptic. Arch Ophthalmol 113:757, 1995