Article Text
Abstract
Aim To describe in vivo confocal microscopy (IVCM) findings in patients with limbal stem cell deficiency (LSCD).
Methods 23 eyes of 17 consecutive patients suffering from LSCD were included in this study. A detailed examination by IVCM was performed in addition to a routine slit-lamp biomicroscopy. Size and density of corneal epithelial and conjunctival epithelial cells on cornea were measured and statistically analysed using SPSS version 8.0 software. Results were compared with histology in select cases.
Results Anatomical and morphological differences were observed between normal corneal cells and conjunctival epithelial cells on cornea. Size and density differences reached statistically significant levels between the normal corneal cells and the conjunctival epithelial cells on cornea (p<0.01). Goblet cells were visible throughout the conjunctivalised corneal epithelium in eight eyes. Several IVCM features could be correlated with histology in six of our patients.
Conclusions A number of features were demonstrated by laser IVCM in patients presenting clinically with LSCD. Some of these features were corroborated with features observed on histological examination of tissue samples.
- Limbal stem cell deficiency
- confocal microscopy
- cornea
- inflammation
- wound healing
- imaging
- diagnostic tests/investigation
- treatment surgery
- stem cells
- degeneration
- treatment surgery
- medical education
- eye (tissue) banking
Statistics from Altmetric.com
- Limbal stem cell deficiency
- confocal microscopy
- cornea
- inflammation
- wound healing
- imaging
- diagnostic tests/investigation
- treatment surgery
- stem cells
- degeneration
- treatment surgery
- medical education
- eye (tissue) banking
Introduction
Limbal stem cell deficiency (LSCD), when moderate or severe, causes considerable ocular morbidity. Common causes include chemical burns and chronic inflammatory diseases such as Stevens–Johnson syndrome and ocular cicatricial pemphigoid.1–4 Following injury or disease, when inflammation is active, the patient suffers from a variety of ocular symptoms.5–9 LSCD leads to loss of corneal transparency with consequent visual impairment or loss and leads to several clinical signs of which conjunctivalisation of the cornea is the hallmark.10 Diagnosis is essentially based on clinical signs and can be complemented by impression cytology.6–8 11
With the relatively recent introduction of the in vivo confocal microscopy (IVCM), it has become possible to study the micro anatomical features of transparent, translucent and semiopaque structures on the ocular surface. A large number of articles have been published on the clinical applications and features of IVCM of the human cornea indicating the rising popularity and importance of this tool for clinical evaluation in ophthalmology.12–18
We studied the IVCM features of patients presenting with LSCD with a view to correlating these with histological specimens where available and establishing definitive IVCM features of LSCD.
Methods
A total 23 eyes of 17 patients, 10 male and 7 female subjects, suffering from partial or total LSCD were included in this study.
All patients attended the Cornea unit of the Department of Ophthalmology, Queens Medical Centre, University Hospital, Nottingham, UK. Ethics approval for this study was obtained from the local Ethics Committee, No. 06/Q203/46.
Detailed clinical examination with slit-lamp biomicroscopy was carried out for signs of LSCD.12 19 These included loss of normal limbal anatomy and conjunctivalisation of the cornea wherein conjunctival/metaplastic cells on the cornea were highlighted with fluorescein stain.
A detailed IVCM examination was performed using the Heidelberg Retina Tomograph II Rostock Corneal Module (Heidelberg Engineering GmBH, Dossenheim, Germany).20 21 The central, peripheral and limbal epithelia (conjunctival or corneal) were scanned in all cases.
The abnormal epithelium covering the cornea in eyes with LSCD was referred to as ‘conjunctivalised corneal epithelial cells (CEC)’.
Cell diameter and density of corneal and CEC in the superficial and basal layers of each eye were measured after taking the average value of three frames for each layer examined. The overall average value for each cell type was also calculated. When the cell body was not visible, the density was measured by the aid of the nuclei and the diameter was assessed by measuring the distance between centre points of two prominent nuclei. Density was ascertained with the original software of the confocal microscope whereas cell diameter was measured by using ImageJ software. (ImageJ V.1.31, Wayne Rasband, Maryland, USA). Data were statistically analysed using SPSS program statistical package V.8.0 (SPSS Inc.). The normality of the data was tested prior to analysis. The t test was used for normally distributed data and the Wilcoxon signed rank test was used when the data were not normally distributed.
IVCM frames of the sub basal layers containing nerves were selected for analysis of the quantitative (density) and qualitative (growth pattern) features of the nerves using image processing and analysis software program ImageJ. A plug-in software called NeuronJ was used with ImageJ to facilitate the tracing and quantification of corneal nerves semimanually.22
In six cases of total LSCD, ocular surface tissue was obtained during surgical reconstruction and examined to demonstrate histological features of the specimen. Samples were spread flat on paper and covered with formaldehyde fixative for half an hour. They were then paraffin embedded, sectioned and stained by H&E and periodic acid Schiff stain for histological examination. The IVCM features were correlated with histological findings.
Results
Of the 23 eyes included in this study, 17 had total LSCD (figure 1A) and six presented with partial LSCD (figure 1B). Age of patients ranged from 19 to 83 years with a mean of 56±18 years. In our series, the aetiology of LSCD included contact lens wear in two patients (four eyes) (11.6%), ocular cicatricial pemphigoid in two patients (three eyes) (11.6%), chemical injury in two patients (11.6%) and aniridia in two patients (11.6%). Other causes included one patient each (6%) of ophthalmia neonatum, radiation, atopic keratoconjunctivitis (two eyes of one patient), multiple eye surgeries (two eyes of one patient), polygrandular autoimmune disease, Stevens–Johnson syndrome and fireworks injury. In two patients (three eyes), the cause for LSCD was unknown.
IVCM of unaffected (non-conjunctivalised) cornea in partial LSCD
In the unaffected cornea the superficial cells showed bright cytoplasm and bright nuclei (figure 2A). The wing cells were smaller than the superficial cells with dark cytoplasm and well defined bright borders (figure 2B). Cells in the basal cell layer appeared dark with bright borders but were even smaller in size (figure 2C). Unlike the superficial cells, nuclei in the wing cells and the basal cells were not seen by IVCM.
Mean diameters of the superficial corneal epithelial cells, wing cells and basal cells were 41.4±5, 23.6±2.8 and 12±2 μm, respectively.
Mean densities of corneal wing and basal cells were 3830±347 and 6329±539 cells/mm2, respectively. Hyper-reflective cells with dendritic morphology were seen among the basal epithelial cells in four eyes (figure 2D).14
In the stroma, keratocyte nuclei were seen as separate bright structures with a dark background (figure 2E).
In the unaffected limbus, the palisades of Vogt presented as hyper-reflective, double contoured linear structures. These alternated with islands of epithelial cells which corresponded with rete pegs (figure 2F).
In vivo confocal microscopy of affected (conjunctivalised) limbus and cornea in partial and total LSCD
The superficial CEC were distinctly visible on the cornea in 21 eyes. These cells were hyper-reflective with bright nuclei and ill-defined borders between cells (figure 3A). Basal CEC were visible in the conjunctivalised epithelium in 17 eyes. Two distinct morphological patterns of the basal cells were observed. In nine eyes the cells had bright nuclei with ill-defined borders (figure 3B) and in eight eyes the cells had bright borders with dark cytoplasm and no visible nuclei (figure 3C). The cells were smaller than the superficial cells. The mean diameter was 30.4±3.5 μm for the superficial CEC and 12.5±1.9 μm for the basal CEC. The mean densities of superficial and basal CEC were 2245±248 cells/mm2 and 3785±410 cells/mm2 respectively.
The stromal collagen beneath CEC could be visualised clearly by IVCM as a dense network of hyper-reflective linear and curvilinear strands (figure 3D). This pattern was seen in 17 eyes (three with partial LSCD and 14 with total LSCD).
Goblet cells were visible among CEC in eight eyes of which the reasons for LSCD were aniridia, radiation, contact lens wear, atopic keratoconjunctivitis, polyglandular autoimmune disease and multiple ocular surgeries in two eyes. In one eye, the reason of LSCD was not identified. In six eyes, the goblet cells had the distinct rosette pattern of arrangement (figure 3E). These cells were between 10 and 30 μm in size, hyper-reflective and round to oval in shape. The density of these cells was variable with a maximum of 841±44 cells/mm2 (figure 3E). Goblet cell crypts17 23 were found in one eye which had partial LSCD (figure 3F) and in another eye which presented with total LSCD (figure 3G). Other IVCM features that were consistently present in CEC included intraepithelial cystic changes in 18 eyes (13 total LSCD and five partial LSCD) with a variable diameter ranging from 12 to 52 μm (figure 3E,F,H) and superficial and deep vessels in 19 eyes (16 total LSCD and three partial LSCD) (figure 3I). Some of the cysts contained hyper-reflective spots within the lumen (figure 3H). Bright ‘goblet’ cells were seen in the vicinity of or surrounding the smaller cystic spaces. Intraepithelial dendritic cells were visible among the CEC in 11 eyes (figure 3J); in four of these there were also dendritic cells among the basal epithelial cells of central unaffected cornea.
In two eyes, neither conjunctival nor corneal epithelial cells could be delineated by IVCM (figure 3K). Both had presented with total destruction of the ocular surface. The Bowman's zone was identified in only three eyes which had partial LSCD.
IVCM of the transitional zone
The conjunctivalised epithelium on the cornea could easily be distinguished from the corneal epithelium, as the cells were hyper-reflective with bright nuclei and ill-defined borders. Conversely, the wing and basal corneal epithelial cells presented as dark structures with extremely well-defined margins (figure 4A).
At places, small islands of corneal epithelial cells were seen within the sheet of conjunctival epithelium (figure 4B–E). In oblique views, the corneal epithelium could be seen extending from the Bowman's zone to the surface (figure 4E). Bright hyper-reflective cells, brighter than the conjunctivalised epithelium, could be seen immediately surrounding the corneal epithelial islands (figure 4D).
In 16 eyes (10 total LSCD and six partial LSCD), cells with hyper-reflective nuclei were seen in the corneal epithelium adjacent to the CEC at the transition zone interpreted as metaplastic cells. Such nuclei extended through all layers of the epithelium. The cells were not as bright as the hyper-reflective conjunctivalised corneal epithelium. The superficial cells measured from 15 to 35 μm with a mean of 26±9 μm compared with basal cells which measured 11±2.6 μm (figure 4F).
Statistically the variation of the cells sizes between the corneal and CEC cells was significant (p<0.05) with the corneal wing and basal cells being smaller than the corresponding CEC cells.
Differences in density reached statistically significant levels with p values of <0.03 for superficial epithelial cells and <0.01 for basal epithelial cells, the corneal epithelial cells being more densely distributed than the CEC cells.
IVCM of the limbus did not show any features of normal limbus as described above and by others previously.20 24
Corneal nerve changes in LSCD
Sub basal nerve plexuses were evaluated both qualitatively and quantitatively in all cases. Sub basal nerves were detected in 22.22% of patients with total LSCD (four out of 17) and in 100% of patients with partial LSCD (all six cases). The density of sub basal nerves was 4.26±0.66 mm/mm2 in cases of total LSCD and 9.70±6.32 mm/mm2 in cases of partial LSCD. The difference was statistically significant (p<0.05).
The majority of sub basal nerves were found in areas covered with a normal corneal epithelium (figure 4G) except in one case of partial LSCD where the sub basal nerves were seen running under a superficially conjunctivalised area (figure 4H). In addition, in another case of partial LSCD the sub basal nerves presented an abnormal tortuous configuration (figure 4I).
Histological results
All samples showed multilayered epithelium with occasional intraepithelial lymphocytes. Cystic changes were seen in three samples (figure 5A) and goblet cells in two (figure 5A,B) of which a crypt of goblet cells was visible in one sample (figure 5B). In the two eyes where no clear details of the cells were visible by the IVCM, histology showed a multilayered stratified epithelium with superficial keratinisation (figure 5C).
Discussion
It is important that a diagnosis of LSCD and its extent (partial or total) is ascertained before considering treatment options. The diagnosis of LSCD is essentially clinical. The clinical features include loss of limbal anatomy, irregular and thin epithelium with stippled fluorescein staining of the area covered by abnormal epithelium (conjunctivalisation), unstable tear film, filaments and erosions, superficial and deep vascularisation, persistent epithelial defects leading to ulceration, melting, and perforation and invasion of corneal surface by a fibrovascular pannus, scarring, keratinisation and calcification.1 4 7 8 19 24–29 At times one has to resort to impression cytology to confirm the diagnosis.19
IVCM has been used previously to evaluate the signs of ocular surface diseases including keratoconjunctivitis sicca, ocular rosacea, vernal keratoconjunctivitis and LSCD.30 IVCM would have a distinct advantage if it could substitute impression cytology and at the same time provide objective evidence of cellular details in suspected LSCD cases. In this study, we were able to correlate the IVCM features of LSCD with histology in six of our patients who had surgical reconstruction. Not all patients had surgical intervention and hence tissue for histological examination was not available for correlation with IVCM findings in every case.
IVCM of the central cornea in partial LSCD showed features that were essentially compatible with the description of normal corneal epithelium, normal corneal stroma and normal limbus except for the central location of dendritic cells in four eyes.31 The more central location of dendritic cells in cases with partial LSCD suggests that the morphologically ‘normal’ surviving epithelium is not entirely normal. The dendritic cells have most probably responded to the original insult and migrated centrally as has been previously demonstrated.14 Similarly, in partial LSCD where islands of ‘normal’ corneal epithelium were observed among CEC the islands were surrounded by bright cells, which are likely to represent inflammatory cells. These could not be verified by histology as tissue sampling was considered unethical in such cases.
The anterior two to three layers (‘superficial cells’) of conjunctivalised epithelium of cornea (CEC) were made up of hyper-reflective cells with ill-defined borders and prominent nuclei. The cell diameter was variable with a mean of 30.4±3.5 μm and so was the cell shape. Compared with normal corneal epithelium the most striking difference was the low reflectivity of the anterior corneal cells, which presented dark cell bodies and bright cell borders. The basal cell ‘footprint’ was significantly larger in CEC compared with basal cells of normal cornea. The basal cells of the CEC presented two distinct IVCM features, one showed prominent nuclei with ill-defined borders and the other showed dark cell bodies with bright cell borders. The latter appearance has been described with normal conjunctival basal epithelium.32 33 The former type of basal cell appearance seems to be a feature of CEC. It was not noted in the other studies on normal human conjunctiva.32 33
Efron et al33 reported conjunctival basal cell densities of 2368±741 cells/mm2 which are lower than the density reported in our study. Messmer et al reported a larger size and lower density of the basal cells in the normal human bulbar conjunctiva. They reported cell diameters of 21.3 μm, 22 μm and 23.1 μm for the non-desquamating superficial, intermediate and basal cells respectively with no statistical difference. They commented on the subepithelial hyper-reflective layer and claimed that it may represent the basement membrane zone.32 Another study suggested that the hyper-reflectivity under the conjunctival epithelium seen just prior to visualisation of the stroma could represent the subepithelial adenoid layer of the conjunctiva, which is the superficial layer of the conjunctival stroma containing mast cells and lymphocytes.33 Kobayashi et al have reported dendritic cells and presumed goblet cells with relatively homogeneous brightness in normal conjunctiva.17
The aforementioned features of normal conjunctival cells and LSCD CEC suggest that despite similarities, conjunctival cells migrating on to the denuded cornea as part of LSCD acquire different morphological characteristics.
We were able to demonstrate goblet cells throughout the CEC in only six eyes which means that although pathgnomic of LSCD they are not always present. The low number may be related to the severity of the original injury or disease. Extensive damage could be associated with complete destruction of all goblet cells. Reduced goblet cell density has been reported in patients with aqueous tear deficiency.34 35 LSCD is often associated with dry eye and this too could explain a low goblet cell density. However, it is also possible that IVCM did not detect goblet cells as it is not possible to scan every millimetre of the surface examined. Nevertheless, in patients for whom we had histology the IVCM presence or absence of goblet cells correlated accurately with histological presence or absence suggesting a degree of correlation. In two of our patients we noted goblet cell crypts, which have been reported in human colonic mucosa.36
Intraepithelial cysts were also a consistent feature of conjunctivalised epithelium. The diameter was variable ranging from 12 to 52 μm. Goblet cells were consistently seen surrounding the smaller cysts. The smallest cysts appeared as dark spaces in the centre of a rosette of goblet cells. With larger cysts, the number of goblet cells was few or absent. This suggests that these cysts start as the space between rosettes of goblet cells and eventually enlarge at the expense of the surrounding goblet cells. Presence of hyper-reflective spots in the lumen of the cysts may represent mucus secretions from goblet cells.33 Histological examination of area demonstrating rosette patterns on IVCM confirmed that the images were produced by goblet cells that stained with periodic acid Schiff stain. A direct morphological comparison of cross sections with en face images is not ideal but was sufficient to suggest that the structures examined were indeed goblet cells. Similarly, cysts were also seen on histological examination proving that they were not an artefact of confocal imaging.
To our knowledge, the sub basal nerves have not been evaluated in eyes with LSCD. We studied the changes in sub basal nerves morphologically and quantitatively. The densities of sub basal nerves in cases of total and partial LSCD were far below published densities observed in normal subjects using laser scanning IVCM.37 In 13 of our cases with total LSCD they were not detectable. This could be due to actual damage to the nerves as their absence was most notable in cornea with absent Bowmans zone. It has been reported that the sub-Bowmans nerve trunks penetrate the Bowmans zone predominantly at the mid-periphery of the cornea38 and this region was affected in all cases with total and some with partial LSCD. Alternatively, the ‘absence’ may just be due to a failure to visualise the nerves through the thickened CEC. This has been observed in patients with conjunctival intraepithelial neoplasia where sub basal nerves underlying dysplastic epithelium could not be demonstrated by IVCM but ‘reappeared’ when the lesion was treated with mitomycin eye drops.39
In 20 eyes, we were unable to demonstrate Bowmans zone in enface and oblique sections of areas with CEC. Destruction of this layer may be related to the injury or disease causing LSCD or occur secondary to the invasion by the fibrovascular panus. The loose network of hyper-reflective fibres with blood vessels in the sub-CEC stroma is also a feature of the fibrovascular pannus of LSCD.
This study has demonstrated a number of features detectable by IVCM, some corroborated by histology, which would enable one to make a diagnosis of LSCD without resorting to impression cytology. These include the presence on the cornea of goblet cells, cysts, large basal epithelial cells with bright borders and dark cytoplasm (without nuclei) and/or ill-defined borders with bright nuclei, loose network of vascularised subepithelial stroma, absent Bowmans zone and attenuated or absent subepithelial basal nerve plexus. Absence of the palisade architecture at the limbus also contributes to the IVCM diagnostic criteria.
Although making a clinical diagnosis of stem cell deficiency is relatively straightforward, evaluating the extent (partial or total) and identifying areas of the limbus which might still retain normal characteristics is often difficult at the slit-lamp. The presence of goblet cells on the cornea is taken as a pathognomic feature of LSCD. This is impossible to ascertain by clinical examination alone. IVCM offers a simple, non-invasive method of confirming the diagnosis and evaluating the extent by examination of the limbus and peripheral cornea. It provides more detail than the slit-lamp as structures at the cellular and subcellular level can be identified. It also allows examination through relatively opaque media, which is also an advantage. In making diagnosis and monitoring progression or response to therapy objective evidence is always considered superior to subjective evidence. IVCM offers objective evidence, which is its greatest advantage. Nevertheless, there are some limitations. The equipment is expensive and not all models allow detailed examination of the limbus and peripheral cornea. It requires technical expertise and a cooperative patient for good images to be obtained. Furthermore, it is not possible to scan every millimetre of the surface and hence some normal areas or areas with pathology may be missed. Due to some of the above reasons, it is not readily available in all centres and reliance still has to be placed on clinical diagnosis supplemented by impression cytology in some instances.
References
Footnotes
Linked article 300550.
Competing interests None.
Patient consent Obtained.
Ethics approval Ethics approval was provided by the Nottingham research ethics committee, No. 06/Q203/46.
Provenance and peer review Not commissioned; externally peer reviewed.
Linked Articles
- Clinical science