Limbal stem cell deficiency

The maintenance of a healthy functional corneal epithelium under both normal and stress conditions is provided by a particular and unique subpopulation of stem cells (SC) located in the limbal region.1,2 The existence of these limbal epithelial SC is supported by the following five well-established facts: (1) These cells do not express such corneal epithelial differentiation markers as keratin 3,1 keratin 12,3,4,5 and connexin 43.6,7 (2) Epithelial neoplasias are commonly located in the limbal area.8 (3) These cells have a superior proliferative capacity under both in vivo and in vitro conditions.2,9,10,11 (4) A deficiency in corneal healing is observed when the limbal area is partially12,13 or totally destroyed.14,15 (5) Destruction of the limbal region has catastrophic consequences for the central cornea wound healing and anatomy.13,14

When limbal epithelial SC or their supporting stromal environment—known as niche—is destroyed or becomes dysfunctional, a state known as limbal stem cell deficiency (LSCD) becomes manifest (also reviewed in Grueterich et al16). Clinically, LSCD carries the hallmark of conjunctivalization, that is, the corneal surface is covered by ingrowing conjunctival epithelium containing goblet cells. This conjunctivalization process is invariably associated with the destruction of the basement membrane, emergence of superficial neovascularization, chronic inflammation, scarring, and poor corneal epithelia.14,17,18,19 Consequently, patients with severe SC loss experience severe irritation, photophobia, decreased vision, and conventional corneal transplants do not fare well.

According to the underlying aetiology, corneal diseases associated with LSCD can be subdivided into two major categories.18 In the first category, limbal epithelial SC are destroyed by known or recognizable offenders such as a chemical or thermal burn, Stevens–Johnson syndrome/toxic epidermal necrolysis, multiple surgeries or cryotherapies or medications (iatrogenic), contact lens, severe microbial infection, radiation, and antimetabolites including 5-fluorouracil and mitomycin C.18,20,21,22 The second category is characterized by a gradual loss of the SC population without known or identifiable precipitating factors. In this category, the limbal stromal niche is presumably affected and progressively deteriorates by a variety of aetiologies that include aniridia18,23 and coloboma,24 neoplasia, multiple hormonal deficiencies, peripheral ulcerative corneal diseases, neurotrophic keratopathy, and idiopathic limbal deficiency.18,25 Table 1 summarizes ocular diseases with limbal deficiencies according to the underlying hereditary pattern and presumed alteration of stromal niche.

Table 1 Corneal diseases manifesting LSCD
Full size table

A comprehensive diagnosis and classification of LSCD is crucial for planning a successful strategic management and may shed light on the prognosis. Among all the signs of LSCD, only conjunctivalization is specific for the diagnosis. Other signs such as superficial vascularization, chronic inflammation, irregular epithelium, recurrent erosion, and persistent epithelial defect with/without ulceration can be also observed in many other corneal diseases without LSCD. To confirm conjunctivalization, ophthalmologists may look for clinical signs such as the loss of limbal palisades of Vogt under slit-lamp examination,23 perform late fluorescein staining,26 which reflects poor epithelial barrier function,27 and impression cytology, which detects conjunctival epithelial and goblet cells that have invaded the corneal surface.18 Among these three diagnostic clues, impression cytology is the most definitive one to clearly demonstrate and confirm the diagnosis of LSCD. Figure 1 illustrates four clinical cases with clinical history and appearance suspicious of LSCD, while their diagnosis was confirmed in two, but ruled out in the other two.

Figure 1
figure 1

Different clinical settings where corneal vascularization is observed, and yet not all cases have the diagnosis of LSCD with conjunctivalization detected by impression cytology.18 (a) A 37-year-old man suffered a chemical burn showing superficial vascularization in the entire corneal circumference most pronounced in the nasal region. Two previously performed corneal transplants failed. Impression cytology showed complete conjunctivalization of the entire limbus except for the temporal aspect (see inset, goblet cells observed in the central cornea). (b) A 68-year-old woman with diffuse corneal vascularization and idiopathic LSCD diagnosed by impression cytology showing corneal conjunctivalization (see inset, goblet cells observed in the central cornea). (c) An 18-year-old man suffered a chemical injury in the right eye presenting with sectorial vascularization in the inferior cornea, mimicking LSCD. Nevertheless, impression cytology disclosed a normal limbus without conjunctivalization (inset shows a demarcation between conjunctiva with goblet cells and normal cornea where goblet cells are not observed), excluding the diagnosis of LSCD. (d) A 42-year-old man with a failed corneal graft and superficial vascularization and dryness was thought to have LSCD, which was excluded as impression cytology showed squamous metaplasia (see inset, big picnotic cells desquamated easily from the corneal surface).

Full size image

Casual examination without cytology to confirm the diagnosis of LSCD could result in both unnecessary transplantation procedures and systemic immunosuppression regimens (for classification of various procedures see Tseng et al,28 Holland et al29,30). Conversely, a correct diagnosis of LSCD will not only avoid detrimental corneal epithelial debridements but also futile lamellar or penetrating keratoplasties.

According to the severity of involvement, LSCD can be further classified as partial or total. In partial LSCD, quadrants of the limbus are spared with some corneal areas preserving a true corneal epithelial phenotype. However, total LSCD is characterized by a complete loss of limbal SC population accompanied by conjunctivalization encompassing the entire corneal surface.

Transplantation of limbal epithelial stem cells

Once an accurate diagnosis of total LSCD is obtained, the transplantation of limbal SC is the ultimate solution to improve patient's symptoms and vision. According to the unilateral or bilateral compromise, the extent of LSCD involvement, and the patient's acceptance and expectations of the proposed treatment, a number of limbal SC transplantation procedures may be considered. They are autologous conjunctival limbal autograft, autologous ex vivo limbal expansion, allogeneic conjunctival limbal allograft (CLAV) from living related donors, allogeneic ex vivo limbal expansion, and allogeneic keratolimbal allograft (KLAL) from cadavers. Nomenclature of these procedures have been proposed by Holland and Schwartz30 for CLAU, living related conjunctival allograft (Ir-CLAL), and KLAL. See Appendix A.

In patients with unilateral total or partial LSCD, transplantation of tissue from the contralateral healthy limbus can be performed by removing a lamellar graft encompassing a small portion of conjunctiva, the limbus and the most peripheral portion of the peripheral cornea from one-third to one-half of the limbal circumference (CLAU). There exists the concern that removing two large pieces, usually spanning at least two to three clock-hours each (6–9 mm in arc length each), will affect the healthy and normal donor eye.31,32,33,34 Autologous ex vivo expanded limbal epithelial SC on amniotic membrane (AM) can also be an alternative in patients with total unilateral LSCD.35,36 In situations of total LSCD compromising both eyes or in unilateral involvement, when the patient decides not to use the fellow eye as a donor due to the reported complications including localized haze,33 pseudopterygium,37,38 filamentary keratitis,39 microperforation during surgery40 and ‘corneal depression’,41 transplantation of allogeneic (allografts) limbal SC is the only remaining alternative for a successful reconstruction.

In the latter circumstance, CLAL from living related donors42,43 or KLAL29,44,45,46 or allogeneic ex vivo expansion of SC from a cadaveric donor47,48,49 are the only available choices for SC replenishment. They all carry the disadvantage of systemic immunosuppression, which is needed for a prolonged, if not indefinite, period of time. A KLAL is a 360° lamellar ring graft encompassing a minimal portion of scleral tissue, the entire limbus and the most peripheral portion of the cornea. When compared to Ir-CLAL, KLAL offers the advantages of providing more SC supply because the entire eye limbal circumference is removed for reconstruction, and eliminating the concern of removing healthy limbal tissue from a normal eye. Although both KLAL and Ir-CLAL require a prolonged use of systemic immunosuppression, the latter might offer a theoretical advantage of having lesser chance of rejection. KLAL is also a better alternative in those patients with LSCD but normal conjunctiva as seen in aniridia, as there is no need to include donor conjunctiva.50 However, in patients with diseases also affecting the conjunctiva, leading to a combined conjunctival and limbal deficiency, for example, in chemical burns or Stevens–Johnson syndrome Ir-CLAL might be more advantageous than KLAL as it includes healthy donor conjunctiva.50

KLAL

Historically, Kenyon and Tseng51 first reported CLAU to transplant autologous limbal SC for unilateral or focal LSCD. As stated above, KLAL has become one of the more performed surgical procedures devised to restore limbal SC population in bilateral LSCD. Reported variations of KLAL include transplantation of various sizes of grafts,44 the use of AM as a substrate,52,53,54 grafts from different donors to increase the amount of transplanted SC29,44 and different immunosuppressive regimens50,55 to improve the outcome.

KLAL outcomes

The final aim of KLAL is to relieve symptoms of photophobia, improve visual acuity, and restore a normal corneal surface without vascularization. Based on these clinical parameters, earlier studies with a short follow-up of 1–2 years revealed that the outcome of success ranged from 82 to 100%44,45,46,55 (Table 2, I: short-term follow-up). However, when the follow-up was extended to 3–5 years, the overall success rate declined to around 50%29,56,57,58 (Table 2, II. long-term follow-up). Based on these studies, we now can formulate the following strategies to improve the outcome of KLAL in the future.

Table 2 Summary of literature reports on keratolimbal allografts
Full size table

First strategy is to restore ocular surface defense

Several risk factors have been identified that may worsen the outcome of KLAL.58,59,60,61 They are severe dryness, uncorrected lid and lid margin abnormalities, keratinization, and chronic inflammation. Since these factors reflect deficiencies in the mechanisms necessary to maintain a healthy and sound ocular surface (for review see Tseng,62 Solomon et al,63 Tseng and Tsubofa64), without prior correction they present as risk factors threatening the well-being of transplanted KLAL. The high prevalence of the aforementioned conditions in patients with Stevens–Johnson syndrome limits the long-term outcome of KLAL in this disease.56,60,61 Therefore, it is a prerequisite that a sound ocular surface defense be restored prior to performing the KLAL.

Identification of these unwanted risk factors by external and biomicroscopic examinations and by the use of dynamic tear function test, dye staining and impression cytology are crucial steps before planning any transplantation. Measures taken to correct severe sicca include punctal occlusion and application of frequent autologous serum drops.65,66,67,68 The lid margin and lash abnormalities such as trichiasis, entropion, or meibomian gland orifice metaplasia will need to be corrected by appropriate plastic surgeries and scleral contact lens. Symblepharon if causing deficiencies in ocular surface defense can be corrected before or at the time of KLAL by mucous membrane graft or AM graft. Exposure problems can also be corrected by Botox-induced ptosis or tarsorrhaphy. These measures are summarized in Figure 2. If ocular surface defenses are not corrected, they represent major contraindications for KLAL. In the future, we believe it is necessary to establish an objective grading of the severity of dryness as a criterion for being an absolute contraindication for KLAL. Furthermore, new therapies will have to be developed to reverse the untreatable severe squamous metaplasia and to control relentless inflammation of the ocular surface in order to improve the KLAL outcome.

Figure 2
figure 2

Summary of several measures that could be taken to augment ocular surface defense so that surgical reconstruction may be considered.

Full size image

Second strategy is to stage KLAL and penetrating keratoplasty

When LSCD is accompanied with deep stromal opacity, penetrating keratoplasty (PKP) becomes necessary. Penetrating keratoplasty is performed at the same time of KLAL or 3–6 months after KLAL. When PKP was performed at the same time as KLAL,53 a rate of 64% of PKP rejection was noted, of which 44% were reversible with institution of immunosuppression. Shimazaki et al69 also reported a 35.6% incidence of endothelial rejection in eyes receiving simultaneous PKP and KLAL. Subsequently, 62% of these eyes with endothelial rejection developed endothelial decompensation. Solomon et al61 during a longer follow-up period noted that 100% of patients with simultaneous PKP and KLAL presented irreversible PKP rejection in a long-term follow-up (5 years). Performing PKP concomitantly with KLAL, during the first KLAL procedure, was found to decrease the visual outcome compared to performing KLAL alone. However, after 3 years, no difference in ambulatory vision survival was noted between eyes that had simultaneous KLAL and PKP (43.1±30.8%) and eyes that had KLAL alone (39.1±11.4%). Solomon et al concluded that performing PKP simultaneously with KLAL is a significant variable associated with reduced survival of the ambulatory vision. However, Ikari and Daya58 performed 14 PKPs with a mean interval between KLAL and PK of 15 months and still noted a 93% PKP failure rate at mean time of 9.5 months.

A poor prognosis of corneal graft survival might be partially explained by the increased exposure of the host immune system to the donor corneal antigens through the recognition and sensitization to limbal allograft antigens when the limbal and corneal graft have originated from the same donor. Furthermore, inflammation generated by KLAL might put the eye susceptible to concomitant PKP sensitization. Based on these data, we concur with the advice proposed by Holland and Schwartz59 that PKP be performed 3–6 months after KLAL. It is noteworthy that when combined with AM transplantation to resurface the cornea, KLAL alone sometimes improved the corneal clarity to the extent that PKP can be avoided or postponed as illustrated in Figure 3(a–d). If remaining corneal opacity needs to be removed, it is advised that deep lamellar keratoplasty be considered at all cost to lessen the chance of rejection. If endothelial dysfunction is noted with stromal opacity, PKP will have to be performed and is best to do so 3–6 months later as illustrated in Figure 3 (e–h).

Figure 3
figure 3

Case presentation to illustrate how KLAL can improve the corneal surface in patients with total LSCD. (a) A 48-year-old man with chemical injuries to his right eye leading to persistent epithelial defect and stromal ulceration, which was associated with deep lipid keratopathy and vision of counting fingers. (b) At 4 months after KLAL and AM transplantation to cover the cornea and perilimbal sclera together with systemic immunosuppression as described in Table 3, the corneal surface has improved without defect and vision of 20/80 even in the absence of PKP. (c) A 51-year-old woman with Stevens–Johnson syndrome presented with decreased visual acuity (counting fingers) and vascularized cornea in the right eye even after lamellar keratoplasty for total limbal deficiency. (d) After correcting sicca by punctal occlusion, autologous serum drops, lid margin eversion, and high permeability contact lenses, a KLAL with AM transplantation was performed without PKP. Subsequently, she received extracapsular cataract extraction and intraocular lens implantation. This photograph taken 22 months after the KLAL shows an avascular smooth corneal surface and improved visual acuity to 20/25. (e) A 37-year-old male with a severe alkali burn lost his right eye by evisceration and his left eye had two failed PKPs. (f) The reconstruction was staged using a two-step procedure beginning with KLAL and AM transplantation, resulting in a more quiescent surface. (g) PKP, extracapsular cataract extraction and intraocular lens implantation were performed four months later resulting in 20/25 vision. (h) This enlarged photography illustrates the donor's palisades of Vogt of KLAL and quiescent surface.

Full size image

Third strategy is to restore limbal stromal environment

A healthy limbal stromal environment is essential for maintaining and supporting proper limbal SC function based on the premise that SC tends to reside in a specialized niche (for review see Tseng70). Therefore, restoration of limbal stromal environment is another important adjunctive strategy. AM is an ideal substrate to achieve this goal by providing a thick basement membrane and an avascular stroma, which suppresses inflammation, scarring and unwanted vascularization (for reviews see Tseng and Tsubota,64 Kruse et al,71 Dua and Azuara-Blanco,72 Sippel et al73). For total LSCD, AM transplantation has been used in combination with KLAL for a variety of corneal diseases such as chemical burns, Stevens–Johnson syndrome, atopic diseases, neoplasias, and contact lens-induced keratopathy.52,53,61,74,75

The benefit of using AM for KLAL is supported by our recent study of a case which should that the resultant epithelial phenotype is corneal as evidenced by the expression of K3 keratin and connexin 43, the basement membrane-containing substrate of AM is preserved, and basal epithelial cells express integrins α3β1 and α6β4 to form hemidesmosomes with laminin 5 (Espana et al79). Furthermore, recent data from the laboratory of Kinoshita also demonstrate that AM can suppress mixed lymphocyte responses in the production of Th1 and Th2 cytokines, an index supporting its role in downregulating alloreactive responses to KLAL.76

Since AM transplantation alone is not sufficient to suppress the severe inflammation derived from Stevens–Johnson syndrome, further research is needed to identify the cause of such severe inflammation and new therapeutic modalities for its control.

Fourth strategy is to employ effective immunosuppressants

The final outcome of KLAL is ultimately determined by the survival of transplanted limbal epithelial SC. Previous studies have produced conflicting results as to the long-term survival of these cells. Henderson et al77 did not recover donor cells from the ocular surface at 3–5 years post-KLAL using impression cytology and DNA fingerprinting, despite the patient showing improved symptoms, visual acuity and surface healing. Nevertheless, Shimazaki et al57 detected donor-derived epithelial cells in the paracental cornea of eyes following KLAL using the fluorescence in situ hybridization and polymerase chain reaction restriction fragment length polymorphism (RFLP) analysis.

As mentioned above, poor ocular surface defense will increase the demand of transplanted SC and may eventually exhaust their potentials. The other major cause of KLAL failure is acute and chronic allograft rejection (see Figure 4). The transplantation of an allograft to the limbus, a highly vascularized area relatively abundant in Langerhans cells, theoretically increases the risks of SC rejection. Uncontrolled inflammation due to the underlying disease in the case of Stevens–Johnson syndrome will help elicit allograft sensitization and hence rejection. Topical and systemic steroids, cyclosporin A and tacrolimus (FK506) have been used in preventing KLAL rejection in several reports with various outcomes (Table 2) (Figure 4). Dua and Azuara-Blanco55 reported a short-term success in using an immunosuppression regimen based on tacrolimus FK506, analogous to cyclosporin A in action, but more potent immunosuppressant, for KLAL in six patients with LSCD. Despite the continuous administration of systemic cyclosporin A, the success rate of KLAL declines from 75–80% in 1 year44,45,55,78 to 50% in 3 years of follow-up.46,58,59,60,61

Figure 4
figure 4

Two different patients with episodes of acute KLAL rejection successfully treated by enhancing the immunosuppression. (a) A 77-year-old male underwent KLAL elsewhere and was treated with mycophenolate mophetyl (CellCept®) 1 g/day. (b and c) He presented on a routine follow-up visit with perilimbal redness, vessel engorgement with hemorrhage, and progressive ulceration. (d) Postoperative view of a quiescent eye following a repeat PKP and institution of mycophenolate, prednisone, and cyclosporine A using the regimen described in Table 3. (e) A 22-year-old male with chemical burn received KLAL elsewhere and presented with reduced vision, redness, and photophobia while under oral mycophelonate alone. He was noted to have vessel engorgement and ulceration with a 3 × 4 mm2 corneal epithelial defect (see inset). (f) Impression cytology showed lymphocyte infiltration at the border of ulceration (see arrows) and no conjunctivalization, supporting the diagnosis of immune rejection and not LSCD. Lymphocyte infiltration × 100 magnification (inset). (g and h) At 2 weeks after addition of cyclosporine A and prednisone, the limbal inflammation and redness was reduced and the ulceration ceased.

Full size image

The major concern of continuous use of systemic immunosuppression is the potential side effects to patient's general health. It is advised that patients who are to receive systemic immunosuppressive therapies be screened and periodically monitored by physicians dealing with immunosuppression and by laboratory tests directed to detecting their side effects and therapeutic trough levels. Pregnancy tests, tuberculosis, HIV status, and appropriate vaccination for immunosuppressed patients are among the ‘new fields’ that the treating ophthalmologist must address in conjunction with a primary doctor. Although the length of immunosuppression in KLAL has not been established, we believe it should be administered for a prolonged, if not indefinite, period of time. Our current personal immunosuppression protocol is listed in Table 3, and is used for patients shown in Figure 3.

Table 3 Personal recent immunosuppressive regimen used for KLAL
Full size table

Surgical technique

The surgical technique involves the transfer of a corneoscleral rim from a cadaver eye to the affected eye. For the recipient eye (Figure 5a), our surgical technique starts with a 360° conjunctival peritomy adjacent to the limbus, and removal of subconjunctival fibrovascular tissue if present. A plane between the fibrovascular pannus covering the cornea and the remaining healthy stroma can be identified by lifting the pannus at the limbus and peeling toward the cornea with a fine forceps. The entire fibrovascular pannus can be removed in this manner and aided by the closed tips of the sharp Wescott scissors (Figure 5b). Once the fibrovascular pannus is removed, superficial keratectomy might be performed to remove frank scar. If the remaining surface is smooth and the Bowman's membrane is intact, AM can be used to cover the limbal and peripheral cornea. However, if the remaining corneal surface is irregular, AM (basement membrane side up) can be laid to cover the entire corneal surface and secured to the perilimbal sclera with 8-0 vicryl or 10-0 nylon sutures (Figure 5c and d). For the donor tissue, a KLAL is then prepared from a cadaver fresh eye, preferably no more than the next day after harvesting. The rim is obtained by trephination of the cornea, preserving the peripheral cornea, the limbus, and the adjacent sclera for transplantation. The graft is thinned with Wescott scissors, removing the posterior stroma and endothelium to avoid steps in the junction between the cornea and the rim, allowing a better distribution of the tear film and preventing continuous abrasion by the overlying eyelid when blinking (Figure 5e).

Figure 5
figure 5

Key surgical steps of KLAL, (a) A 37-year-old woman with a completely vascularized right eye secondary to a long history of severe vernal keratoconjunctivitis unresponsive to topical treatment. (b) A plane between the pannus and the transparent corneal stroma is found. The corneal pannus and conjunctival scar tissue are carefully removed after lifting the pannus at the limbus and peeling toward the cornea with a fine forceps. (c) Corneal clarity was improved after pannus removal. (d) A layer of AM was used to cover the corneal stroma and to provide a new healthy basement membrane for the donor's corneal cells. (e) A cadaveric corneoscleral rim was used for transplantation after the uveal tissue and endothelium were removed, and the graft was thinned with Wescott scissors to avoid steps in the junction between the cornea and the rim, allowing a better distribution of the tear film and a continuous rubbing with the eyelid. (f) Final appearance after a second AM as a patch was placed over the cornea to improve corneal healing.

Full size image

The KLAL is secured with interrupted 9-0 nylon sutures to the episclera and underlying AM that will serve as a new basement membrane. Finally, we cover the surface with a second AM as a temporary patch (Figure 5f). Postoperative evaluation with fluorescein is the routinely used test to evaluate the epithelium migration over the AM. Postoperative steroids and antibiotics must be used until no epithelial defect is observed.

Other groups use different KLAL techniques with promising results. The corneoscleral crescent technique of Holland/Schwartz30 uses two stored corneoscleral rims that will be cut into two halves, providing four crescents of limbal tissue for transplant. The crescents are placed end to end to cover the entire limbus. They argue that an increased number of SC is transplanted using this technique. Reinhard et al78 have described a technique where an eccentrically trephined button containing 30–40% of limbus is transplanted to a recipient corneal bed.