Decellularized Descemet membrane anterior keratoplasty with allogeneic simple limbal epithelial transplantation for partial limbal stem cell deficiency following partial keratolimbal allograft failure

Abstract

Purpose

To describe the use of Descemet membrane anterior keratoplasty (DMAK) with modified allogeneic simple limbal epithelial transplantation (allo-SLET) to treat a case of partial limbal stem cell deficiency (LSCD) following keratolimbal allograft (KLAL) failure.

Case

A 21-year-old woman with Autoimmune Polyglandular Syndrome-related LSCD presented with pain and decreased vision. There was partial failure and recurrence of LSCD following a severe/acute KLAL rejection that led to persistent epithelial defects refractory to conservative therapy. This was treated with a superficial keratectomy and placement of a DMAK. A modified allo-SLET was performed with overlying amniotic membrane and a temporary tarsorrhaphy. There was epithelialization of the corneal surface by 3–4 weeks with an improved ocular surface. Despite partial recurrence of late staining, the cornea has remained epithelized; vision has improved; and the patient has remained pain free more than 1.5 years following the procedure.

Conclusions

DMAK may be a long-term substrate to help improve and maintain epithelization of the cornea up to 1.5 years. DMAK may be a viable alternative to using amniotic membrane as a scaffold in allo-SLET for treatment of partial LSCD. While late epithelial staining recurred in our patient, DMAK appears to prevent recurrent epithelial defects and reduce ocular surface pain, conveying an improvement in quality of life in patients at high-risk for rejection/failure.

INTRODUCTION

Located between the corneal stroma and endothelium, Descemet’s membrane (DM) is a transparent basement membrane of the corneal endothelium made of collagenous and non-collagenous elements, which include laminins, collagen type IV, nidogens, and perlecan.¹ Previously, DM has been shown to be a viable culture substrate for the ex vivo expansion of limbal stem cells (LSC).² LSCs cultured on the anterior stromal side of DM maintained expression of putative limbal stem cell markers ABCG2 and p63α better than human amniotic membrane (HAM).³ In both outpatient and surgical settings, HAM is commonly used as a scaffold or bandage for treating corneal persistent epithelial defects (PED).^4,5 It is also used as a scaffold in simple limbal epithelial transplantation (SLET) in the treatment of LSCD.⁶ Both the cryopreserved and dehydrated forms of HAM are relatively opaque, decrease vision, and often degrade shortly after transplantation on the ocular surface, limiting their long-term benefits. In contrast, DM is transparent and has been used as a carrier for endothelium to treat endothelial dysfunction via Descemet membrane endothelial keratoplasty. Additionally, DM is very resistant to degradation and can persist on the ocular surface as evidenced by the clinical behavior of descemetoceles. This suggests that DM may be a viable scaffold/substrate for SLET and may be useful as a long-term substrate for supporting the ocular surface.

Limbal stem cell deficiency (LSCD) occurs from deficiency or dysfunction of the LSCs located along the Palisades of Vogt. LSCs are responsible for generating clear corneal epithelium and maintaining a stable ocular surface free from epithelial defects, conjunctivalization, and neovascularization. In cases of total LSCD, an ocular surface stem cell transplantation (OSSTx) can restore the ocular surface. However, allogeneic OSSTx is susceptible to rejection and subsequent failure. We present a case of partially failed KLAL leading to painful PEDs treated with a Descemet membrane anterior keratoplasty (DMAK) with modified allogeneic simple limbal epithelial transplantation (allo-SLET).

CASE PRESENTATION

A 22-year-old woman with LSCD resulting from Autoimmune Polyglandular Syndrome type 1 (systemic diseases included Addison’s disease, hypoparathyroidism, hypogonadism, and recurrent candidiasis) treated with keratolimbal allograft (KLAL) 1.5 years ago presented with pain in her right eye (OD). Her KLAL graft had previously suffered severe acute rejection 6 weeks following surgery. The acute rejection episode was refractory to systemic and topical augmentation of corticosteroids and immunosuppression but responded to IVIG.⁷ Her surface remained stable for 6 months after resolution of the rejection episode, but subsequently developed late fluorescein staining extending from the superior segment (Figure 1A). This developed into a corneal epithelial defect, prompting her presentation due to the pain.

Figure 1.

Slit lamp photograph demonstrating late superior fluorescein staining denoting late failure (1A). Preoperative slit lamp photograph demonstrating a failing ocular surface with hazy epithelium, neovascularization, epithelial defects, and corneal scarring (1B). Fluorescein highlights the epithelial defects, late fluorescein staining, especially superiorly and peripherally (1C). Preoperative intraoperative still photograph demonstrating another view of these features.

She had a complex past ocular history of bilateral (OU) LSCD leading to recurrent corneal PEDs and ulcers, neurotrophic keratoconjunctivitis OU, herpes viral keratitis, and dry eye syndrome OU. Additionally, her ocular surgical history consisted of a living-related conjunctival limbal allograft (lr-CLAL) OD four years prior with her mother as a first-degree ABO (Antibodies Blood Group) and HLA (Human Leukocyte Antigen) compatible donor. She was on extended-release tacrolimus 4 mg daily (target 8–10 ng/mL), and azathioprine 50 mg daily (due to previous intolerance of mycophenolate mofetil).

Due to the recurrence of the partial LSCD, she developed painful PEDs and ulcers over the ensuing 3 months (Figure 1B–D). These were refractory to increased lubrication, chronic bandage contact lens wear, punctal occlusion, serum tears, dehydrated sutureless amniotic membrane, and tape tarsorrhaphy. She had previously been treated with topical cenegermin and sutured amniotic membrane with keratectomy, but they had only afforded her comfort for a few months before painful PEDs recurred following failure of her lr-CLAL, prior to her subsequent KLAL.

Following KLAL failure, best corrected visual acuity (BCVA) declined to 20/200 OD. Given the history of multiple prior rejections and elevated panel reactive antibody (PRA, a measure of potential susceptibility to future rejection), she was considered high-risk for repeat KLAL. In order to delay or prevent the need for repeat KLAL, an alternative procedure was suggested: a DMAK with allo-SLET. Table 1 summarizes the order of procedures/events.

Table 1.

Order of procedures/events

Procedure/event	Preop BCVA	Best Postop BCVA
lr-CLAL	20/400	20/30
lr-CLAL acute rejection
SuperK/AMT (for painful PEDs)	20/50	20/50
KLAL	20/150	20/40
KLAL acute rejection
KLAL failure
DMAK/allo-SLET/tars	20/200	20/40

BCVA: best corrected visual acuity; lr-CLAL: living related conjunctival limbal allograft; superK: superficial keratectomy; PEDs: persistent epithelial defects; AMT: amniotic membrane transplantation; KLAL: keratolimbal allograft; DMAK: Descemet membrane anterior keratoplasty; allo-SLET: allogeneic simple limbal epithelial transplantation; tars: tarsorrhaphy

Surgical technique

The abnormal epithelium and fibrovascular pannus were removed with a keratectomy (Figure 2A). The involved area was measured to determine the graft size. The donor corneal tissue was previously prepared as pre-stripped, decellularized Descemets membrane (BrightMEM^™ Corneal Allograft, Brightstar Therapeutics; Lions Gift of Sight, St. Paul, MN). The donor tissue was trephinated to the desired diameter (8.0 mm, Figure 2B) with care to ensure this diameter was 0.5 mm within the debrided area. The corneal recipient bed was dried thoroughly. Next, the donor tissue was positioned (Figure 2C), with posterior endothelial side in apposition to the recipient bed, within the area of lamellar keratectomy. The orientation is based on our ex vivo data where the fetal layer (rich in vitronectin and osteonectin) of DM supported growth of LSCs in culture. Then this was allowed to dry into place, and the overlying anterior corneal button was peeled away leaving the DMAK in place (Figure 2D). A 30 g cannula with balanced salt solution was used to smooth the DMAK and minimize any wrinkles (Figure 2E). During processing, an S-stamp was placed on the posterior side of the DMAK for orientation (seen as a reverse S-stamp in Figure 2).

Figure 2.

Intraoperative still photographs. The abnormal epithelium and fibrovascular pannus were removed with a keratectomy (2A). The donor cornea was trephinated (2B). The donor tissue was positioned (2C), with posterior endothelial side in apposition to the recipient bed, within the area of lamellar keratectomy. The anterior corneal button was peeled away leaving the DMAK in place (2D). A 30 g cannula was used to smooth any wrinkles from the DMAK (2E, * denotes reversed S-stamp). Multiple slivers of donor limbus were excised with a supersharp blade and vannas scissors (2F). The SLET pieces were placed along the DMAK surface (2G, * denotes reversed S-stamp, white arrows highlight the mega-SLET pieces). A layer of cryopreserved HAM covered the surface including the tissue glued allo-SLET pieces (2H).

The DMAK was allowed to dry in place while the allo-SLET pieces were prepared from an ABO-matched cornea. Multiple slivers of limbus were excised with a supersharp blade and vannas scissors (Figure 2F). Two larger segments (2 mm, mega-SLET pieces) were placed along the periphery nasally and superiorly. Additionally tiny pieces were cut up into traditional SLET sized pieces⁸ and were placed along the DMAK surface (Figure 2G). Tissue glue was used to secure these into place. A layer of cryopreserved HAM (AmnioGraft, Biotissue) was then brought onto the field to cover the cornea, limbus/KLAL segments, and anterior conjunctiva (Figure 2H). This was sutured into place with 10–0 nylon, and a large diameter bandage lens (Kontur) was placed. A temporary tarsorrhaphy was constructed to maximize healing given the neurotrophic status and minimize loss of the allo-SLET pieces. Postoperatively, the patient was administered topical difluprednate 1% 4 times daily (QID), tobramycin 0.3% QID, autologous serum tears 20% QID, cyclosporine 0.05% QID (off-label increased dosing), and preservative free tears.

By approximately 3–4 weeks, there was a restored ocular surface with clear corneal epithelium centrally and no epithelial defects (Figure 3). There was partial late fluorescein staining in the periphery (temporally). BCVA improved to 20/40 OD. At 7 months postoperatively, there was slight increase of the late fluorescein staining peripherally with neovascularization, yet the cornea remained epithelized with a BCVA at 20/70. The patient has been pain-free and maintained her improved ocular surface with no recurrence of PEDs over 1.5 years of follow-up.

Figure 3.

Postoperative slit lamp photograph demonstrating smooth epithelium as well as midperipheral and peripheral allo-SLET pieces (3A). Fluorescein demonstrates mild late staining temporally in the periphery but otherwise clear corneal epithelium (3B). Note the striae in both images from wrinkles in the underlying DMAK.

DISCUSSION

PEDs can occur from a variety of ocular surface conditions such as severe dry eye, neurotrophic keratopathy, medication toxicity (including preservatives), exposure keratopathy, band keratopathy, inflammatory disease, and LSCD. The risk of corneal infections, scarring, keratolysis, vascularization, and vision loss are higher the longer a PED is present.⁹ Initial therapy for a PED often includes addressing underlying causes (if possible), starting aggressive lubrication (preservative free tears, gels, and ointments), and minimizing ocular toxicity while providing anti-microbial prophylaxis. Additional treatments may include punctal occlusion, bandage contact lenses, sutureless HAM placement, and autologous serum tears. These conventional treatments may only help to a certain degree or may be ultimately ineffective when the underlying issue is LSCD. An OSSTx may be necessary to provide PED resolution in those cases.¹⁰ In this patient, prior LSCD (before OSSTx treatment) had originally led to PEDs that demonstrated no improvement with sutureless amniotic membrane placement and only temporary improvement with a superficial keratectomy with amniotic membrane. Similarly, after failure of her first OSSTx (lr-CLAL) due to rejection, recurrence of chronic, painful PEDs did not respond to a superficial keratectomy with amniotic membrane, necessitating a KLAL to provide relief and ocular surface stabilization.

The presented case involved a chronic PED following her second OSSTx due to the partially failed KLAL leading to severe pain. Not only were there previous concerns for patient adherence to systemic immunosuppression and follow-ups, but the severe acute rejection of her last KLAL was refractory to traditional management, necessitating ongoing IVIG infusions. For these reasons, we sought another option in lieu of a repeat KLAL to quickly improve comfort and ocular surface (if possible). The rationale for using DM as a scaffold for allo-SLET (rather than HAM) was multifactorial. Prior studies have provided evidence that the anterior surface of DM is a viable substrate of promoting limbal stem cell expansion ex vivo.² With the potential to help epithelize the corneal surface, DM also has the benefits of being optically transparent, providing an additional protective layer that covers the sub-basal plexus of corneal nerves, and delivering some degree of additional resistance/integrity to the cornea. When submerged in high dose collagenase (5mg/ml), DM can remain intact for at least 24 hours compared to 30 minutes with HAM (unpublished data, JHH). DM has also previously been used as an inlay for stabilization of keratoconus. The transplanted intrastromal DMs were intact at up to 24 months postoperatively and provided structural stability with no signs of keratoconus progression.¹¹ In the presented case, the membrane was still present 1.5 years postoperatively on slit lamp examination.

Allo-SLET pieces from the donor limbus are most commonly cut into tiny pieces and placed in the paracentral and midperipheral cornea. We placed additional longer mega-SLET pieces (thinner, longer in length, and more peripheral than previously described)¹² in the periphery, which is more analogous to our typical KLAL or living related conjunctival limbal allograft procedures. This was performed to provide more stem cells, especially in the areas with greater preoperative conjunctivalization and larger epithelial defects. While they were closer to the typical anatomical location of limbal stem cells, these pieces were still not in proximity to the conjunctival/limbal vasculature with the hope that there would be less risk of rejection.

The patient maintained her systemic immunosuppression under the care of a pediatric transplant nephrologist. During the follow-up, there were no signs of acute rejection, but there was peripheral late fluorescein staining. With signs of partial failure, it is unclear of the long-term efficacy of these allo-SLET pieces. Interestingly, the majority of staining occurred in the two quadrants where the mega-SLET pieces were not placed. Nonetheless, the primary objectives of stabilizing the ocular surface without recurrence of PEDs and improving patient comfort were achieved. On occasion, the authors have previously noted challenges with the limbal pieces sloughing off from the underlying amniotic membrane (traditional SLET/allo-SLET) with or without a HAM sandwich technique. This DMAK may also allow greater initial adherence, but further study is needed.

The authors still prefer lr-CLAL and KLAL for primary OSST compared to allo-SLET given the advantages of providing many more stem cells, creating a barrier along the limbus from conjunctival invasion, and mimicking the anatomic location of stem cells which minimizes loss of stem cells during future keratoplasty. Additionally, a lr-CLAL also provides fresh conjunctival tissue (ideal for conjunctival deficient cases) and goblet cells which can help concomitant dry eye which is often present. All three options (lr-CLAL, KLAL, and allo-SLET) require systemic immunosuppression due to their allogeneic nature.^6,13

Cases such as the one presented seem ideal for DMAK with allo-SLET as the patient already had a limbal OSST and was on systemic immunosuppression. It therefore seemed reasonable to try this approach before submitting the patient to another KLAL with the resulting exposure to another HLA profile and additional foreign antigens. It is unclear at this time how responsive the immune system is to allo-SLET and whether this technique increases one’s PRA and risk of rejection with antigenic exposure. Allo-SLET’s longevity in this setting is also uncertain; we started noticing areas of late fluorescein staining early on postoperatively, so transplanting more allo-SLET pieces may be beneficial for this surgery. However, there was no recurrence of the PED as there was following the failed KLAL segment.

This case utilized a combination of DMAK with allo-SLET to successfully treat persistent epithelial defects in a patient with a partially failed ocular surface following KLAL rejection. An improved ocular surface was noted without any epithelial defect formation during the 1.5-year follow-up. Comfort and visual acuity also improved. This provides an additional surgical strategy for addressing corneal epithelial defects refractory to conservative treatment.