ABSTRACT
PURPOSE: In limbal stem cell deficiency, both the Ocular Surface Disease Index (OSDI) questionnaire and tear break-up time (BUT) are comparable between traditional methods and the Keratograph 5M. In this study, we aimed to correlate OSDI with Keratograph 5M interviewed OSDI, as well as slit-lamp tear BUT with Keratograph 5M noninvasive tear break-up time (NIKBUT) in limbal stem cell deficiency. PATIENTS AND METHODS: Thirty-eight limbal stem cell-deficiency patients (76 eyes) from Virgen Macarena-Rocio Hospital (Seville, Spain) underwent this diagnostic test study. All patients completed the traditional OSDI. We measured the BUT, performed a Keratograph 5M analysis of NIKBUT first (employed for the analysis) followed by the average NIKBUT, the level of dryness, and conducted the OSDI questionnaire through an interview. For each pair of tests, we analyzed the means and applied an intraclass correlation coefficient (r), creating a Bland-Altman plot for data dispersion. RESULTS: Average values were 47.5 points (±25.8), and 47.3 points (±27.5) for traditional OSDI and Keratograph OSDI, respectively (P =0.87); the r value indicates good agreement (0.72). The Bland-Altman plot followed a linear pattern, and the results were similarly distributed. The NIKBUT mean was shorter than the BUT mean (P = 0.007); the r value indicates moderate agreement (0.574). The Bland-Altman plot formed an almost horizontal line, with almost all values between the mean and two standard deviations. CONCLUSION: Keratograph 5M is useful for the evaluation of the ocular surface in limbal stem cell deficiency. NIKBUT can substitute BUT based on its advantages of being noninvasive, objective, with intraobserver and interobserver repeatability and reliability. The Keratograph 5M OSDI is comparable to the traditional questionnaire.
ABSTRACT
BACKGROUND/AIMS: Limbal stem cell deficiency (LSCD) is characterised by a marked decrease in limbal stem cells. It is classified primarily using subjective slit-lamp observations. In vivo confocal microscopy (IVCM) can non-invasively provide objective information on the condition of the limbal niche, the corneal epithelial basal cell density and the corneal sub-basal nerve plexus density (SND). We here used IVCM to evaluate changes in SND to improve LSCD classification. METHODS: We evaluated and classified 38 patients (76 eyes, 44 with LSC and 32 control eyes) using the Rama, López-García and Deng (clinical and confocal) classifications and evaluated the concordance of the confocal and clinical classifications. We constructed a logistic regression model using multivariate analysis to correlate different degrees of conjunctivalisation with IVCM parameters and used receiver operating characteristic (ROC) curve analysis to establish the SND cut-off value with maximum diagnostic sensitivity and specificity. RESULTS: The classification systems correlated moderately at best (kappa, 0.449). The corneal SND of cases (6469±6295 µm/mm2) was less (p<0.001) than in controls (20911±4142 µm/mm2). The SND, but not basal cell density, played a protective role against conjunctivalisation (OR, 0.069; 95% CI 0.008-0.619; p=0.01). An SND cut-off value of 17 215 µm/mm2 yielded a sensitivity and specificity of 95.5% and 90.6%, respectively, for LSCD diagnosis. CONCLUSION: The density of the corneal sub-basal nerve plexus was inversely related to conjunctivalisation in LSCD. Further studies are needed to verify this and to elucidate the directionality between these factors.
