Limbal Epithelial and Mesenchymal Stem Cell Therapy for Corneal Regeneration

ABSTRACT

Corneal pathologies are a major cause of blindness and visual impairment, especially in the developing world. However, not only is there a global shortage of donor corneal tissue, but a significant proportion of these blinding pathologies also carry an unfavourable long-term prognosis for conventional corneal transplantation. In the last few decades, there has been a spurt of research on developing alternate approaches to address corneal blindness, including stem cell therapy. After the discovery of epithelial stem cells at the limbus, successful cell-based approaches to treat severe ocular surface disease were developed and have subsequently become widely practised across the world. More recently, mesenchymal stem cells were identified near the epithelial stem cells at the limbus, providing a unique opportunity to develop regenerative therapies for both corneal epithelial and stromal pathologies. This review firstly emphasises on qualifying limbal stem cells as either epithelial or mesenchymal and then summarises all the existing knowledge on both cell types and their individual roles in corneal regeneration.  The review describes the history, indications, techniques, and outcomes of the different methods of limbal epithelial stem cell transplantation and elaborates on the potential applications of limbal mesenchymal stem cell therapy.

Keywords:

• Limbus
• Cornea
• Stem Cell
• Limbal Epithelial Stem Cell
• Limbal Mesenchymal Stem Cell

Introduction

The cornea is a very delicate part of the anterior eye that focusses incident light onto the light-sensitive retina by acting as a transparent refractive lens (Figure 1(a-c)). Since it is exposed to the external environment, the cornea is constantly at risk of getting damaged. Although there are redundant layers of physical and biological defences to prevent such damage, even small changes in the clarity, smoothness or shape of the cornea can adversely impact its optical functions and result in visual symptoms. Corneal diseases are an important cause of visual impairment and blindness that affect millions worldwide, particularly in the developing world.¹

Figure 1. Normal Appearance of the Cornea and Limbus. The healthy cornea is optically transparent (a) and surrounded by the annular limbus which separates it from the opaque sclera. The limbus contains the palisades of Vogt (b) which are finger-like projections of the stroma. The optical coherence tomography (OCT) image of the normal cornea shows a smooth uniform stratified epithelium (c) with underlying compact stroma. The OCT angiography image of the normal limbus shows linear anastomosing hair-pin loops of the limbal capillaries (d) with adjacent avascular corneal stroma.

The corneal surface is especially delicate and consists of tightly packed stratified squamous epithelium arranged in a regular fashion (Figure 1(c)). Underneath the epithelium is the corneal stroma made of regularly arranged collagen lamellae, interspersed by proteoglycans and scantily populated by mesenchymal cells called keratocytes. The epithelial surface has an extremely high turnover (4–7 days) as mature superficial cells exposed to the environment are continuously shed and replaced instantly by younger cells. The corneal stroma, on the other hand, being relatively acellular has a very slow turnover rate.² Although their symptoms may overlap, corneal epithelial and stromal diseases have traditionally been considered to have distinct pathophysiology. The burden of visual impairment and blindness is also shared unequally by these two entities, stromal opacification being by far the more common cause. In the last three decades much attention has been paid to the limbus (Figure 1(b,d)), the annular transition zone between the clear cornea and the opaque sclera, and its role in maintaining corneal homeostasis. Two distinct populations of adult stem cells, epithelial and mesenchymal, have been discovered at the limbus and their role in corneal regeneration has been extensively studied. This review discusses the known role and future potential of limbal epithelial and mesenchymal stem cells in corneal regeneration and in addressing the problem of blindness and visual impairment due to corneal diseases.

Limbal epithelial stem cells

The mechanisms of corneal epithelial renewal or the reasons for its dysfunction, termed ocular surface disease, were not well understood until about three decades ago. The concept that stem cells could be present peripherally in the limbal palisades of Vogt (Figure 1(b)) was first suggested by Davanger and Evensen in 1971.³ It was becoming clear at that time that corneal transplantation alone did not work in eyes with severe ocular surface disease and Thoft proposed peri-limbal conjunctival transplantation as an alternative.^4,5 Conjunctival transplantation actually showed better results than conventional penetrating keratoplasty (PK) in a series of eyes with chemical burns.⁴ At that point in time, it was proposed and widely believed that the corneal surface epithelialized from conjunctival cells.⁶

However, subsequently, Thoft and colleagues realized that the limbal epithelium was much more similar to the corneal epithelium than to the conjunctival epithelium.⁷ This prompted them to propagate the x-y-z hypothesis of corneal epithelial homeostasis. This hypothesis proposed that epithelial cells from the limbus/peripheral cornea migrated centripetally and differentiated to replace the more mature cells in the central cornea.⁸ Finally in 1986, Schermer et al. in a seminal paper demonstrated the presence of putative stem cells in the basal layers of the limbal epithelium which were slow-cycling, label-retaining and were capable of proliferation in vitro. They proposed that these basal epithelial cells of the limbus transform into transient amplifying cells which in turn give rise to terminally differentiated cells of the corneal epithelium.⁹ Cotsarelis et al. further elaborated that injury to the central cornea leads to rapid proliferation of otherwise slow-cycling limbal basal epithelial cells.¹⁰ Thus, the limbus was established as the central dogma of corneal epithelial regeneration, a paradigm that holds true till this day.¹⁰ These important discoveries in the late 1980s led to the coinage of the term ‘limbal stem cells’ to denote the corneal epithelial stem cells and the clinical term ‘limbal stem cell deficiency’ to denote severe ocular surface disease due to limbal stem cell failure.

Limbal epithelial stem cell deficiency

Limbal epithelial stem cell deficiency or LSCD occurs because of direct damage to the limbal epithelial stem cells or to their microenvironment, the niche. As a consequence, the corneal epithelium is unable to regenerate or heal and the corneal surface gets gradually replaced with conjunctival epithelium. This process of conjunctivalization is usually accompanied by neovascularization and scarring with eventual loss of corneal clarity and deterioration of vision (Figure 2(a)). The compromised cornea in eyes with LSCD is also at risk of developing infections, persistent epithelial defects, and even stromal melting and perforation. The most common method for clinical diagnosis of LSCD is slit-lamp biomicroscopy.¹¹ The clinical features suggestive of LSCD are superficial corneal vascularization, diffuse fluorescein staining of the corneal surface with or without persistent epithelial defects, conjunctivalization of the corneal surface and absence of limbal palisades of Vogt (Figure 2(b)).

Figure 2. Clinical Appearance of Limbal Epithelial Stem Cell Deficiency (LSCD). LSCD is characterized clinically by loss of corneal clarity because of superficial neovascularization (a), positive fluorescein staining (b), presence of conjunctival cytokeratin (CK) markers on impression cytology (CK19) and immunohistochemistry (c), altered epithelial cell morphology with sub-basal fibrosis on in vivo confocal microscopy (d), replacement of the dark hyporeflective corneal epithelial phenotype with bright and thick conjunctival phenotype on optical coherence tomography (OCT, E) and loss of normal limbal vascular architecture on OCT angiography (F).

More objective methods that have been used to confirm the clinical diagnosis are impression cytology and immunohistochemistry, in vivo confocal microscopy and anterior segment optical coherence tomography (AS-OCT, Figure 2(c–f)).¹² LSCD can be clinically classified as partial or total depending on the amount of surface conjunctivalization of the cornea. Partial LSCD is characterized by incomplete limbal damage with focal conjunctivalization and presence of healthy-looking limbus in other quadrants. Total LSCD is characterized by conjunctivalization of the entire corneal surface and complete loss of the limbal palisades in all quadrants. A staging system for LSCD based on clinical presentation has also been devised recently taking the total surface area of corneal conjunctivalization and the limbal involvement into account.¹³ Depending on laterality, LSCD can be further classified into unilateral or bilateral. Here, it is important to consider that an affected individual may have a unilateral total/partial LSCD or bilateral LSCD with either total/partial LSCD in both the eyes or total LSCD in one eye and partial LSCD in the other eye.

Causes of LSCD

The aetiologies for LSCD are varied and can be summarized under the broad categories of traumatic, inflammatory, immunological, hereditary, and idiopathic (Table 1). Of all the listed aetiologies, ocular burns are the commonest cause of both unilateral and bilateral LSCD.¹⁴ Whereas immunological and hereditary diseases usually cause bilateral LSCD, although the presentation may be asymmetrical. There have been few reports on medical management of limited or partial LSCD,^15,16 but total LSCD is typically a condition that requires surgical management when visually significant.

Table 1. Aetiological classification of limbal stem cell deficiency.

Category	Clinical Conditions
Traumatic/Inflammatory	Ocular (chemical, thermal or combined) burns, Multiple ocular surgeries, Radiation or Chronic contact lens wear
Immunological	Stevens-Johnson syndrome (SJS), Mucous membrane pemphigoid (MMP), Chronic ocular allergy, Graft-versus-host disease (GvHD) and Keratoconjunctivitis sicca (primary and secondary Sjogren’s syndrome
Hereditary	Congenital aniridia, Epidermolysis bullosa, Xeroderma pigmentosum, Multiple endocrine deficiency and Ectodermal dysplasia
Idiopathic	Diseases with unknown cause

Limbal epithelial stem cell transplantation (LSCT)

Following the discovery of the limbal epithelial stem cells (LESCs), Tseng and colleagues conducted the first pre-clinical trial in rabbits to show that limbal transplantation was far more effective than conjunctival transplantation in improving corneal clarity in experimentally induced LSCD.¹⁷ This paved the way for the “proof-of-concept” human clinical studies and set the ball rolling for an interesting two decades of technical evolution in the surgical approaches to LSCT. Depending on the source of the donor limbal tissue, LSCT can either be autologous (source: healthy fellow eye in unilateral LSCD) or allogeneic (source: healthy eye of one or more living related/unrelated donors or cadaveric donors). The clinical outcomes of LSCT can be classified as anatomical and functional. The anatomical outcome is usually subjectively defined as the regeneration of a stable, epithelialized and avascular corneal surface. But the anatomical outcome can also be objectively defined in terms of restoration of the normal corneal phenotype seen on impression cytology, in vivo confocal microscopy or AS-OCT. On the other side, the functional outcome is more objectively assessed with respect to the improvement in visual acuity.¹² Different techniques of limbal epithelial transplantation (e.g. conjunctival-limbal autografting (CLAu), cultivated limbal epithelial transplantation (CLET), simple limbal epithelial transplantation (SLET)), their comparison, limitations, and future potential are described in the following sections.

Conjunctival-limbal autografting (CLAu)

In 1989, Kenyon and Tseng described the original technique of autologous LSCT in a series of cases with acute and chronic unilateral chemical burns.¹⁸ Initially called limbal autograft transplantation (LAT), the current universally accepted terminology for this technique is conjunctival-limbal autografting (CLAu). This involves harvesting up to six clock-hours of donor limbal tissue from the healthy contralateral donor eye and transplanting the two limbal lenticules of three clock-hours each to the affected eye.¹⁸ The outcomes of this technique are summarized in Table 2.^18–28 Although CLAu is largely effective in reversing LSCD and establishing a stable corneal surface, there are concerns about the impact of such extensive dissection on the healthy eye as there have been reports of iatrogenic LSCD in the donor areas.^29,30 However, Cheung et al. in their recent study dismissed these concerns by demonstrating no evidence of LSCD in 51 donor eyes from which 4 clock-hours (2 segments of 2 clock-hours each) of limbal tissue had earlier been harvested for transplantation.³¹ Busin et al. also demonstrated that limbal epithelial stem cells repopulated the donor site as early as 1-year post-operatively.³² Interestingly, Kheirkhah et al. showed the restoration of the entire corneal surface in a case of total LSCD with a modified technique of minimal CLAu using only 60 degrees (2 clock-hours) of limbal tissue instead of the usual 4–6 clock-hours.³³ This interesting report not only questioned the need for large-sized grafts but also supported the notion that there is some redundancy in the limbal stem cell reserve of a normal eye and that a minimal number of limbal stem cells can sustain the entire corneal surface in the long term.³³

Table 2. Clinical outcomes of conjunctival-limbal autografting (CLAu) for unilateral limbal epithelial stem cell deficiency (LSCD).

						Follow-up (years)
Author	Year	Eyes	Clinical Success (%)	2-line vision gain (%)	Complications (n)	Mean	Range
Kenyon et al^Citation18	1989	26	77	65	Corneal perforation (1), PED (1)	1.6	0.2–3.75
Rao et al^Citation19	1999	16	94	82	None	1.6	0.3–3.75
Shimazaki et al^Citation20	2004	11	91	NA	PED (1)	1.3	NA
Ozdemir et al^Citation21	2004	15	87	80	PED (1)	1.2	0.3–2
Santos et al^Citation22	2005	10	80	61	None	2.75	NA
Wyelgala et al^Citation23	2008	21	71	NA	Not mentioned	2.6	NA
Miri et al^Citation24	2010	12	100	81.3	None	3.9	1–9.9
Baradar an-Rafii et al^Citation25	2012	34	88	NA	Graft dislodgement (4), thick graft (4), progressive LSCD (2)	1.4	0.5– 2.75
Burcu et al^Citation26	2013	16	69	69	LSCD recurrence (3), corneal melt (1)	7.6	4.3–15.5
Barreiro et al^Citation27	2014	23	87	79	NA	1.7	0.7–3
Moreira et al^Citation28	2015	15	33	NA	PED (2), Corneal melt (1)	1.5	NA

Only studies with a minimal sample size of 10 and mean follow-up of 1 year have been included in this analysis. NA = data not available; LSCD: Limbal stem cell deficiency; PED: persistent epithelial defect; LSCT: limbal stem cell transplantation.

Cultivated limbal epithelial transplantation (CLET)

In 1997, Pellegrini and associates described autologous CLET, a technique in which a transplantable epithelial sheet is cultivated ex vivo on a fibrin scaffold in the laboratory by using less than one clock-hour of limbal tissue.³⁴ Following this, several groups have published their outcomes of this technique, however the cultivation protocols are quite variable across groups as summarized in Table 3.^35–48 Broadly there are two techniques of limbal cultivation: a) the suspension culture, where the biopsied limbal tissue is first digested enzymatically to separate the cells from the extracellular matrix and then the cell suspension is placed in culture over a suitable substrate; and b) explant culture where the limbal tissue fragment is sectioned into smaller pieces and placed directly on the substrate without enzymatic digestion. Additionally, the constituents of the culture medium may or may not contain animal-derived products or xenobiotic materials. Xenogeneic constituents of a limbal culture system can be in the form of murine feeder cells, bovine serum, or animal-derived growth factors.⁴⁹ This is one of the major limitations of CLET because use of xenogeneic material carries some risk of disease transmission due to both known and unknown pathogens.⁴⁹ The use of xenogeneic materials makes regulatory approvals difficult and when considered along with the costs of establishing and maintaining a clinical grade laboratory, together these factors make CLET quite an expensive procedure.⁵⁰

Table 3. Techniques and clinical outcomes of autologous-cultivated limbal epithelial transplantation (CLET) for unilateral Limbal Stem Cell Deficiency (LSCD).

								Follow-up (years)
Author	Year	Culture Technique	Subs-trate	Culture Time (days)	Eyes	Clinical Success (%)	2-line visual gain (%)	Mean	Range
Feeder-Free and Xeno-Free Cell Cultures
Sangwan et al^Citation35	2011	Explant	hAM	10 − 14	200	71	60.5	3	1– 7.6
Feeder-free but not Xeno-free Cell Cultures
Sangwan et al^Citation36	2006	Explant	hAM	11–15	88	73	37	1.5	0.3– 3.3
Shimakazi et al^Citation37	2007	Explant	hAM	14.6	16	50	37.5	2.5	0.5– 7.1
Pauklin et al^Citation38	2010	Explant	hAM	14	30	77	73	2.4	0.8– 6
Neither Feeder-Free nor Xeno-Free Cell Cultures
Schwab et al^Citation39	1999	Suspension	hAM	28–35	17	76	16	0.9	0.2– 2
Schwab et al^Citation40	2000	Suspension	hAM	21–28	10	60	36	1.1	0.5– 1.6
Rama et al^Citation41	2001	Suspension	Fibrin	14–16	18	74	33	1.5	1– 2.2
Rama et al^Citation42	2010	Suspension	Fibrin	14–16	107	68	54	2.9	1– 10
Di Iorio et al^Citation43	2010	Suspension	Fibrin	NA	166	80	NA	NA	NA
Prabhasawat et al^Citation44	2012	Suspension	hAM	14–21	12	75	58.3	2.2	0.5– 3.9
Sejpal et al^Citation45	2013	Explant	hAM	10–14	107	46.7	54.2	3.4	1 − 9.3
Zakaria et al^Citation46	2014	Explant	hAM	14	12	58	17	1.6	0.3– 3.6
Ganger et al^Citation47	2015	Explant	hAM	NA	54	72	23	1.8	0.3 − 3.3
Fasolo et al^Citation48	2017	Suspension	Fibrin	NA	59	42	NA	6	1– 13

Only studies with a minimal sample size of 10 and mean follow-up of 1 year have been included in this analysis. hAM = human amniotic membrane; CL = contact lens; NA = data not available.

However, Mariappan et al. have developed completely xeno-free protocol for limbal epithelial cell cultivation.⁵¹ In this protocol, the donor limbal tissue was transported to the laboratory in human corneal epithelium (HCE) medium and shredded into small pieces under strict aseptic conditions. A human amniotic membrane (hAM) prepared and preserved by the eye bank was used as a carrier. A 3 cm x 4 cm hAM sheet was de-epithelialized using TrypLE® for 30 minutes. The shredded bits of limbal tissue were explanted over the centre of de-epithelized hAM with the basement membrane side-up (Figure 3(a)). A submerged explant culture system without a feeder-cell layer was used under standard culture conditions using autologous serum. A confluent monolayer of cells, ready for clinical transplantation, usually formed on the hAM in 10 to 14 days (Figure 3(b-c)).

Figure 3. Limbal Explant Culture on the amniotic membrane: After attaching the explants to the hAM membrane, slow growth can be seen around the explant till the whole membrane is fully covered. Day 2 (a), Day 9 (b), Day 15 (c). Phenotypic marker expression by LESCs. LESCs showing positive expression of the CK3 + 12, their native phenotypic marker and ABCG2, stem cell marker, on day 9 of culture.

Simple limbal epithelial transplantation (SLET)

Sangwan et al., in 2012 described an innovative approach called SLET.⁵² In this procedure, a one-clock-hour limbal tissue is harvested from the donor eye, cut into multiple small pieces and secured on the recipient eye over an hAM graft.⁵² This is akin to in vivo cultivation, as the cornea re-epithelializes using the natural environment of the ocular surface instead of laboratory reagents in a petri-dish. Several large studies have subsequently established the long-term safety and efficacy of SLET in the treatment of severe ocular surface disease and LSCD, particularly in chemical burns.^53–56 The clinical outcomes of SLET for unilateral LSCD are summarized in Table 4.^53–56 Since SLET works very differently from CLAu or CLET it may be worthwhile to discuss its mechanism of action. Serial photographic imaging of eyes using fluorescein staining has clearly shown the multi-directional growth of epithelial cells from each transplant within the first few days which gradually coalesced to form a confluent stratified sheet of epithelium on the corneal surface by 10–14 days.⁵⁷ Amescua et al. used high-resolution AS-OCT to show that this epithelial proliferation occurs on the hAM scaffold and the hAM itself becomes the new basement membrane for the regenerated epithelium (Figure 4(a–d)).⁵⁸ Histopathological and immunohistochemical analysis of excised corneal buttons from eyes undergoing PK after SLET also confirmed that the corneal surface epithelium was of corneal phenotype (CK3⁺, CK12⁺, CK19⁻, MUC5AC⁻) and also that there was focal retention of stem cells (ABCG2⁺ and Δp63α⁺) in the basal epithelial layer of the newly regenerated epithelium.⁵³ A recent study used a combination of impression cytology and in vivo confocal microscopy to show that clinically successful post-SLET cases did indeed demonstrate features suggestive of a normal corneal epithelial phenotype.⁵⁹

Table 4. Clinical outcomes of autologous simple limbal epithelial transplantation (SLET) for Unilateral Limbal Stem Cell Deficiency (LSCD).

						Follow-up (years)
Author	Year	Number of eyes	Clinical Success (%)	2-line visual gain (%)	Complications (%)	Mean	Range
Basu et al^Citation53	2016	125	76	75	Recurrence (24) Keratitis (6) Perforation (1)	1.5	1–4
Vazirani et al^Citation54	2016	68	84	65	Recurrence (31) Infection (7) Glaucoma (1) Granuloma (1)	1	0.5–4.9
Gupta et al^Citation55	2018	30	70	50	Recurrence (30)	1.1	0.5–3.4
Basu et al^Citation56	2018	30	80	NA	Recurrence (20) Hemorrhage (7) PED (3)	2.3	0.8– 3.8

Only studies with a minimal sample size of 10 and mean follow-up of 1 year have been included in this analysis.

Figure 4. Clinical Outcome of Simple Limbal Epithelial Transplantation (SLET). The top row shows the pre-operative appearance of affected eye with opacification and vascularization of the corneal surface (a) and a thick bright hyperreflective conjunctival epithelial phenotype seen on optical coherence tomography (OCT, b). The same eye one-year post-operative shows remarkably improved corneal clarity, absence of vascularization and opacification (c) and smooth dark hyporeflective corneal epithelial phenotype on OCT imaging (d).

Comparison of CLAu, CLET and SLET in unilateral LSCD

There are no randomised controlled clinical trials that have compared the outcomes of CLAu, CLET and SLET. However, a recent systematic review compared the different techniques of autologous LSCT in patients with unilateral LSCD due to mostly ocular surface burns.⁶⁰ The review included 22 non-comparative case series comprising 1023 eyes. At 1.75 years post-operatively, autologous LSCT for unilateral LSCD showed overall anatomical and functional success rates of 69% and 60%, respectively. However, the review reported that the anatomical and functional success rates of SLET (78%; 68.6%) and CLAu (81%; 74.4%) were significantly better than those of CLET (61.4%; 53%). The review concluded that although all techniques of autologous LSCT are safe and effective for the treatment of unilateral LSCD, existing evidence clearly indicates that clinical outcomes are better with SLET and CLAu as compared to CLET.⁶⁰

Allogeneic LSCT for bilateral LSCD

The regenerative techniques that are performed for bilateral LSCD include CLAL (conjunctival-limbal allograft), KLAL (kerato-limbal allograft), allogeneic CLET, and allogeneic SLET. The clinical outcomes of CLAL/KLAL for patients with bilateral LSCD are mentioned in Table 5.^61–66 The clinical outcomes of allogeneic CLET for bilateral LSCD are mentioned in Table 6.^37,38,67,68 Unlike autologous LSCT, it is difficult to compare the different techniques of allogeneic LSCT because the indications are diverse and systemic immunosuppression regimens used are not uniform and vary widely across most studies. Like autologous CLET, different groups have used diverse cell cultivation protocols, which is an additional confounder that makes this comparison difficult. A recent review summarized the outcomes of allogeneic SLET for bilateral LSCD in Stevens-Johnson syndrome, mucous membrane pemphigoid, ocular burns and severe chronic allergy in 30 eyes of 29 patients.⁵⁷ The patients underwent either living-related or cadaveric allogeneic SLET with an identical postoperative immunosuppression regimen. The median follow-up was 28 months (range: 13–66 months) with no significant difference at baseline between the living-related and cadaveric groups. One year post-operatively, the median best-corrected visual acuity had improved from hand-motions to 20/80 and more than 60% of operated eyes had a visual recovery of 20/60 or better, irrespective of the source of donor tissue. Overall allogeneic SLET was successful in 83.3% eyes at final follow-up. Kaplan Meier survival analysis showed statistically similar 5-year cumulative survival probability of 90% in the living related and 82% in the cadaveric donor group.⁵⁷

Table 5. Clinical outcomes of conjunctival limbal allografts (CLAL)/Keratolimbal allograft (KLAL) for Bilateral Limbal Stem Cell Deficiency (LSCD).

						Follow-up (Years)
Author	Year	Number of eyes	Clinical Success (%)	Visual Gain (%)	Complications (%)	Mean	Range
Holland^Citation61	1996	25	72	66	Rejection (40) PED (40)	2.2	0.5–5.3
Tsubota et al^Citation62	1999	43	51	60	PED (60) Glaucoma (30)	3.2	1–6.3
Ilari and Daya^Citation63	2002	23	27.3	82.6	Rejection (40) PED (24) Glaucoma (26) Infection (13) Melting/Perforation (13)	5	1.3–8
Solomon et al^Citation64	2002	39	23.7	44.6	PED (36) Glaucoma (26) Infection (8)	2.8	1–9.8
Maruyama-Hosoi et al^Citation65	2006	85	65.9	NA	Rejection (13) PED (34) Glaucoma (33) Infection (8)	3.9	NA
Shi et al^Citation66	2008	23	56.5	43.5	Rejection (48) PED (9)	2.7	2–3.2

Only studies with a minimal sample size of 10 and mean follow-up of 1 year have been included in this analysis. PED = persistent epithelial defect; NA = data not available.

Table 6. Clinical Outcomes of Allogeneic Cultivated Limbal Epithelial Transplantation (CLET) for Bilateral Limbal Stem Cell Deficiency (LSCD).

								Follow-Up (Years)
Author (Year)	Donor Source (n)	Culture Technique	Feeder Cell	Serum	Air-lifting	Xeno-Free	Success (n/N)	Mean	Range
Basu et al (2011)^Citation67	LR (28)	Explant	No	AS	No	Yes	71.4%	4.83	1–9.5
Pauklin et al (2010)^Citation38	LR (4), Cadaveric (10)	Explant	No	AS	No	No	50%	2.375	0.8–6
Shimazaki et al (2007)^Citation37	LR (8), Cadaveric (12)	Suspension	Yes	FBS	Yes	No	50%	2.42	0.5–7.1
		Explant	No	AS	Yes
Koizumi et al (2001)^Citation68	Cadaveric (13)	Explant	Yes	FBS	Yes	No	100%	0.93	0.9–1.1

Only studies with a minimal sample size of 10 and mean follow-up of 1 year have been included in this analysis. LR = Live Related; AS = Autologous Serum; FBS = Fetal Bovine Serum.

Summary, limitations, and future of LSCT for LSCD

The main limitation of LSCT for unilateral and bilateral LSCD is that irrespective of the surgical technique, source of donor tissue, or post-operative care, about 20–30% of all cases develop a recurrence of LSCD over the long term. The risk factors of failure have been studied in detail and are usually related to the severity of the pathology.^35,42,53 Those with extensive conjunctival fibrosis and with co-morbidities like corneal stromal damage tend to do poorly. It is also well recognized that dryness, exposure and eyelid pathologies (entropion, ectropion and trichiasis) are relative contraindications for LSCT. Corneal epithelial regeneration also does not always translate into visual gain, as evident from the data presented in the tables, because the primary pathology may have rendered the underlying stromal matrix irregular or severely scarred. Those with bilateral LSCD particularly need concomitant or sequential keratoplasty for optimal visual restoration. In such situations, it is recommended that the surface reconstruction is performed first followed by the corneal transplantation in a sequentially staged manner.^67,69 Overall, significant advances have been made in the last 2–3 decades in the management of severe ocular surface disease due to LSCD. What was once considered an incurable cause of corneal blindness due to the poor outcomes with keratoplasty, is now treatable with a variety of different techniques of LSCT with 70–80% success rates.⁶⁰ This is a phenomenal example of bench-to-bedside translation that hardly has any parallels in other branches of medicine.

With a better understanding of the limbal stem cell niche using advanced imaging techniques and development of markers specific for the epithelial stem cells, it is likely that clinical outcomes will only improve with time. Since limbal epithelial stem cells (LESCs) are morphologically indistinct from transiently amplifying cells or corneal epithelial cells, much attention has been devoted to developing markers that will help in identifying them. The LESCs were initially identified in suspension cultures as holoclones (clones with properties of long-term proliferation, self-renewal, and clonogenicity).⁷⁰ Later, p63α (more specifically termed as deltaNp63α) was reported to be essential for maintaining migration and proliferation of LESCs.⁷¹ Further studies reported the expression of ATP-binding cassette transporter super-family G member 2 (ABCG2) in side population of limbal epithelium,⁷² whereas CCAAT enhancer binding protein delta (C/EBPδ) was co-expressed along with deltaNp63α in mitotically inactive limbal stem cells and reported to regulate cell cycle and auto-renewal of human LESCs.⁷³ Most recently in 2014, ATP-binding cassette, sub-family B, member 5 (ABCB5) was discovered as a limbal stem cell gene which is essential for the development and repair of cornea and specifically identifies label-retaining stem cell population in limbus and not in cornea.⁷⁴ While the jury is still out on whether a single marker or a combination of the above-mentioned markers should be used to selectively identify the actual LESCs, the ability to quantify the amount of stem cells being transplanted will help standardize LSCT procedures, particularly CLET. It is also likely that predictable gene-editing combined with CLET may help treat the rare hereditary forms of LSCD using the patient’s own cells.^75,76

Limbal mesenchymal/stromal stem cells

Mesenchymal stem/stromal cells (MSCs) belong to a diverse population of non-hematopoietic fibroblast-like cells which are plastic adherent and phenotypically defined as CD90⁺ CD105⁺ CD73⁺ CD29⁺ CD34⁻ CD45^−.77 These cells are mostly derived from the adult tissues and therefore have exceptional genetic stability and fewer ethical concerns compared to induced pluripotent stem cells (iPSCs) and embryonic stem cells, respectively.⁷⁸ They can repair injured tissue either by direct differentiation or by secretion of trophic factors.⁷⁹ MSCs can be easily isolated from different sources including bone marrow (BM-MSCs), dental pulp (DPSCs), adipose tissue (ADSCs), and umbilical cord (UC-MSCs) and cultured in vitro due to their plastic adherence property. The MSCs induce the modulation of immune response and tissue repair through inflammatory cytokines and growth factors.⁷⁹

Recently MSCs have been identified and successfully isolated from the corneal stroma and limbus (Figure 5).^80–83 A distinct population of cells with mesenchymal characteristics (e.g., multipotent differentiation, clonal growth and expression of MSC-specific markers) sub-adjacent to the limbal basement membrane has been reported to support the viability and potency of the limbal epithelial stem cells. They were initially identified as ‘side population’ in the flow cytometry and termed as corneal stromal stem cells (CSSCs) in 2005.⁸⁰ These limbal MSCs were also thought to be present at the anterior peripheral (limbal) stroma in the cornea and therefore were termed as peripheral and limbal corneal stromal cells (PLCSCs).⁸¹ These were reported to conform to all the criteria (plastic adherent, CD105⁺, CD73⁺, CD90^+, CD45⁻, CD34⁻, CD14⁻, CD11b⁻, CD79α⁻, and HLA-DR⁻, differentiate into osteoblasts, adipocytes and chondroblasts in vitro) specified by the International Society for Cell Therapy (ISCT) for MSCs.^77,78 Limbal niche cells isolated from the limbal stroma were reported to generate progenitors with MSC-like characteristics, indicating their potential role in healing of the wounded cornea.^82–84 Later, in 2014, these cells were isolated from the limbal stroma and termed as limbus-derived stromal/mesenchymal stem cells (LMSCs).⁸⁵ These LMSCs have the potential to: a) differentiate into various cell types including keratocytes, b) secrete normal corneal extra-cellular matrix (ECM) components like collagen I, lumican and keratocan in a regular lamellar pattern, c) remodel the diseased corneal stroma, d) suppress inflammation and angiogenesis and e) restore normal corneal transparency.^82,85 Owing to the fact that these LMSCs can be obtained by biopsy and expanded in vitro to give large number of cells, these possess potential therapeutic application for cell-based therapy for corneal stromal pathologies.⁸³

Figure 5. Expansion of Limbal Mesenchymal Stem Cells (LMSCs). Phase-contrast images of the cultured LMSCs on Day 3 (a) and Day 15 (b) of primary culture; end of passages 1 (c) and 2 (d). Phenotypic expression of LMSCs. The LMSCs show positive expression of the markers for mesenchymal-lineage: Vimentin, CD90 and CD105; and are also positive for stem cell markers ABCG2 and PAX-6.

CSSC/LMSCs and corneal stromal wound healing

Corneal stromal opacification can either be due to genetic abnormalities or acquired factors like traumatic corneal injuries and ulceration. The role of CSSC/LMSCs have been evaluated in experimental animal models of both congenital opacities (Lumican null mice)⁸⁴ and traumatic injury (ulceration with a mechanical rotating burr).⁸⁶ Du et al. in 2009 isolated the CSSCs and injected them directly into the corneal stroma of the Lumican-null mice.⁸⁴ Lumican, being a critical proteoglycan component of the corneal stromal ECM, helps in stromal organization and in maintaining corneal transparency. Lumican null mice develop corneal opacity (due to disorganization of the stromal collagen) resembling corneal scars.⁸⁴ The injected cells survived for months in wild-type mice without inducing a T-cell-mediated response. Remarkably, in Lumican null mice, thickness of the corneal stroma and defects of collagen fibril were restored to that of the wild-type mice, post injection of human CSSCs.⁸⁴

Acquired corneal stromal opacities are more common and develop due to a fibrotic response to injury or inflammation. Boote et al. in 2012 reported that penetration of epithelial basement membrane during wounding is a decisive factor for inducing suitable ultrastructural alterations in the stroma leading to opacity and scars.⁸⁶ They used a rotating mechanical burr to mechanically induce the injury. OCT and x-ray scattering were used to measure corneal opacity, diameter of collagen and order of matrix; electron microscopy to image proteoglycans; and q-PCR to evaluate transcript expression of quiescent and fibrotic genes. They observed that debridement of corneal epithelium while leaving the basement membrane intact resulted in a significant increase in matrix disorder but minimal corneal opacity at 8 weeks. Whereas epithelial debridement along with basement membrane resulted in presence of atypically large proteoglycans in the subepithelial stroma, enhanced disorder in matrix, scattering of light, collagen diameter and upregulation of fibrotic markers 2 to 4 weeks post-injury. Using the similar mouse model of corneal scarring, Basu et al. in 2014 transplanted the LMSCs (expanded from clinically repeatable, small and superficial limbal biopsies obtained from human cadaveric corneo-scleral rims) into wounded mouse corneas which stopped the formation of scars containing abnormal ECM components.⁸⁵ Moreover, these LMSCs prompted replacement of the depleted stroma with structured and organized lamellar layers resembling the native stroma.⁸⁵

Further studies in this direction revealed the significance of secreted factors like hepatocyte growth factor (HGF). HGF is critical for potentiating the tissue repair function of MSCs and is itself able to restore corneal transparency post corneal injury.⁸⁷ This suggests direct implication of HGF in development of HGF-based therapy for corneal injuries.⁸⁸ HGF is a pleiotropic growth factor secreted by MSCs and reported to induce proliferation of epithelial cells, morphogenesis, motility, and angiogenesis in different organs through phosphorylation of its receptor c-Met.⁸⁸ This results in mitogenic, motogenic, and morphogenic activities in different types of cells.^88,89 HGF treatment has been reported to improve and accelerate corneal epithelial repair by suppressing corneal inflammation and enhancing the proliferation of corneal epithelial cells (CECs) in mouse model of corneal injury.⁹⁰ MSC-conditioned medium containing HGF along with other growth factors and interleukins has been shown to significantly increase the wound closure rate of corneal stromal cells, in vitro. This further results in accelerated corneal wound healing due to enhanced viability and migration of stromal keratocytes and formation of desirable ECM.⁹¹ Most recently, Mittal et al. have reported that MSCs modulate corneal alloimmunity and promote graft survival via secretion of HGF.⁹² Previous studies have also shown that other growth factors like transforming growth factor-β (TGFβ) and platelet-derived growth factor (PDGF) also play important roles in corneal stromal wound healing. Bone marrow cells can also move to wounded site for repair along with limbal stromal cells and may help to prevent fibrosis.^93–95

Clinical applications of LMSCs

A pilot clinical study by Basu et al. has been examining the safety and efficacy of both autologous and allogeneic LMSCs in corneal stromal regeneration in superficial corneal pathologies like scars, ulcers and burns (https://clinicaltrials.gov/ct2/show/NCT02948023).^96–98 The LMSCs were derived from the limbal biopsies obtained from human cadaveric corneo-scleral rims using a well-established xeno-free ex vivo cultivation technique.⁸⁵ Briefly, the procedure involved collection of the limbal ring from serologically tested and clinically accepted cadaveric corneas from the eye bank, followed by dissection of the limbal ring into small pieces and overnight enzymatic digestion. The cells thus obtained are cultured in low serum (2%) medium and passaged thrice for elimination of epithelial cells from the culture. These LMSCs were delivered using an onlay technique using fibrin glue. During transplantation, the corneal epithelium was removed with a surgical sponge and 100 µl of stromal cells (5 x 10³ cells/µl, diluted with the thrombin component of fibrin glue (TISSEL, Baxter)) were applied to the debrided corneal stroma, and a bandage contact lens was placed over the cornea. Post-operatively, the patient received topical antibiotics and comprehensive ophthalmic evaluation including scheimpflug imaging, anterior segment OCT scanning and slit-lamp biomicroscopy was performed.^96,97 The patients did not receive any topical corticosteroids but when examined at 6-weeks, 3-months, 6-months, and 1-year, post-operatively; the eyes receiving LSSCs demonstrated significantly better uncorrected and best-corrected visual acuity with improved corneal clarity and reduced opacification and vascularization as compared to the controls.^96,97 The LMSCs have certain inherent advantages over the MSCs derived from other sources like bone marrow or dental pulp, including but not limited to less invasive and painful extraction procedures as compared to bone marrow and dental pulp extraction.⁹⁸

Controversies and future challenges

It has been around three decades since the role of LESCs in corneal epithelial renewal was established, but some questions continue to remain unanswered. Clinicians have often wondered how some eyes continue to maintain a clear central cornea despite having obvious clinical signs of peripheral LSCD. These observations have called into question the role of LESCs in corneal epithelial homeostasis.⁹⁹ There have been controversial reports suggesting that oligopotent epithelial stem cells may be distributed throughout the mammalian ocular surface,¹⁰⁰ giving rise to the concept of corneal stem cells (CSCs) as opposed to LESCs. Nasser et al., recently reported that committed corneal cells have the potential to dedifferentiate, repopulate the stem cell pool, and correctly re-construct the corneo-scleral junction in the presence of intact stroma.¹⁰¹

The complexity and costs of CLET as a technique of LSCT have also been called into question, particularly with recent reports suggesting that it may not offer any clinical benefit above and beyond that of other techniques like CLAu and SLET.⁶⁰ The proposed advantage that CLET results in amplification of not only corneal epithelial cells but also LESCs in culture may not be the case. Perhaps the discovery of a specific marker for LESCs will help in this cause. While ABCB5 has been reported as a limbal stem cell gene⁷⁴ and ABCG2⁺/ABCB5⁺double-positive LESCs have been reported to be highly proliferative,¹⁰² the search for a suitable marker specifically identifying LMSCs is also ongoing. Since a single marker is not able to discern between LESCs and LMSCs, using a possible combination of established markers (e.g., ABCG2, ΔNp63α, CEBPδ, and ABCB5) for LESCs and an additional subset of mesenchymal markers for LMSCs (e.g., CD90, CD73, and CD166) is recommended. Finally, the major advantage of limbal stem cells (both epithelial and stromal) lies in their potentially autologous clinical applications which substantially reduce ethical concerns and risk of graft rejection. However, the ethical issues associated with the use of allogeneic tissue, particularly concerning living-related donors need to be addressed appropriately with specific guidelines.¹⁰³

Conclusions

The limbus harbours both epithelial and mesenchymal stem cells. The LESCs reside in the basal layer of the limbal epithelium in the palisades of Vogt whereas LMSCs are present in the limbal stroma sub-adjacent to the limbal basement membrane. The LESCs, and perhaps also the LMSCs, play a vital role in maintaining normal corneal homeostasis and their dysfunction leads to a clinical state of stem cell deficiency. This is perhaps the only known clinical stem cell-deficient condition other than cancer/chemotherapy-induced bone marrow stem cell deficiency. Great progress has been made in this area, with limbal epithelial stem cell therapy being successfully practised worldwide using different surgical approaches. Unlike LESCs, the role of LMSCs in normal homeostasis is not yet completely understood. There is some speculation that the mesenchymal cells are vital to the stromal niche that maintains or modulates the function of the epithelial stem cells. There is also no clinically known state of limbal or corneal mesenchymal stem cell deficiency, and therefore the clinical application of LMSCs differs principally from that of LESCs. However, the therapeutic potential of the LMSCs along with other sources of MSCs in the treatment of blinding corneal stromal pathologies is being enthusiastically explored through pre-clinical and early clinical studies. Since both these cell types are concurrently the subject of intense research and clinical applications, the authors feel that the term ‘Limbal Stem Cells’, should, therefore, be qualified by adding the cell type, epithelial or mesenchymal.

Sources of support

Department of Science and Technology (No. IFA14-LSBM-104), and Science and Engineering Research Board (No. SERB/EMR/2017/005086), Govt. of India.

Acknowledgments

Sachin Shukla acknowledges support from Department of Science and Technology, Govt. of India (No. IFA14-LSBM-104). Sayan Basu acknowledges support from Science and Engineering Research Board (No. SERB/EMR/2017/005086), Govt. of India.

Disclosure statement

Authors declare that there is no conflict of interest.

Additional information

Funding

This work was supported by the Hyderabad Eye Research Foundation, Department of Science and Technology (No. IFA14-LSBM-104), Science and Engineering Research Board (No. SERB/EMR/2017/005086), Govt. of India.