The first step in this study showed that the cytological impression technique is very suitable for collecting conjunctival epithelial cells. It can collect many cells while preserving intercellular junctions. In contrast, in healthy subjects, it retrieves few corneal epithelium cells, the pauci-desquamative character of which partly explains the difficulty of collecting them. A rich cytologic impression is therefore a characteristic specific to conjunctival epithelium and the high retrieval rate of cells by corneal IC represents in itself a criterion of conjunctival differentiation and thus a diagnosis of LSCD.
Among the markers that we selected in accordance with current data in the literature, K12 has been shown to be an excellent marker for corneal differentiation, and K13 an excellent new marker for conjunctival differentiation, as recently confirmed by the work of Ramirez-Miranda et al.
26 Moreover, K13 is a more reliable marker for distinguishing between the corneal and conjunctival epithelia compared with K19. Indeed, we should carefully consider the conjunctival specificity of K19 due to its staining in the basal cells of the peripheral corneal EC, as reported too in several recent studies.
14,17,26 –28
Moreover, K3, the most widely used marker, has not been shown to be specific to the corneal epithelium.
One hypothesis for this K3 nonspecificity is that, in this study, we sampled from the bulbar conjunctiva, far from the cornea to avoid contamination of our conjunctival impressions by corneal cells, whereas the bibliographic data indicating the specificity of K3 tend to focus on the study of the near-limbus conjunctiva.
29 It has, in fact, been reported that K3 was expressed by noncorneal tissues, including bulbar conjunctiva,
11,14,30 nose, and oral mucosa.
31 As shown in
Figure 1, strong K3 expression was observed in the suprabasal layers of the conjunctival EC (
Fig. 1A) but absent of the limbus (
Fig. 1B).
A second hypothesis on the lack of K3 specificity is that the expression of our antibody AE5 in the conjunctiva is due to the detection of K76. Indeed, today, the specificity of the AE5 clone that we used is in question, because the latter would recognize the K3 and K76 isoforms. K76 is a keratin that is specific to suprabasal cells of the masticatory epithelium (gum and palate),
31 and its presence in conjunctival cells still remains to be proven.
The second step in this study showed that IC produces strong returns for LSCD with 90% of samples suitable for analysis. Unfavorable anatomic conditions, however, are a contraindication to sampling.
This diagnostic test has proved very sensitive, with 8 positive samples of 9 tested here, and capable of diagnosing LSCD at a mild form of the disease (
Fig. 2, patient 4). The negative predictive value of the test was confirmed on healthy histologic tissue.
In the case of patient 3, we identified a mixed corneal and conjunctival epithelium (K12, K13, and K19+). This patient had a herpetic keratitis with major corneal opacification and advanced corneal neovascularization, which are not specific of LSCD but can result from the viral infection. After reading the diagnostic test, we questioned our clinical diagnosis of severe LSCD and accepted, despite biomicroscopic findings, the possibility of a mild form of the disease. Thus, K12 could be used to quantify the severity of the disease, its detection defining mild LSCD. However, we must accept that significant sampling error does exist in the frequency of K12+ cells sampled, as shown in healthy subjects, where few corneal epithelial cells were retrieved by IC. Perhaps the quantification of K13 and/or K19+ cells could be useful as a measure to determine the degree of LSCD. More mild LSCD should be sampled to answer this question.
Finally, K7 has shown to be another new specific marker of conjunctival differentiation.
17 The immunocytochemical search for both K7 and K13 by corneal IC could be more efficient than the K13/K19 couple in diagnosing LSCD.