Tear Lipid Layer Thickness and Variability Both Impact Tear Film Stability

Abstract

Purpose : An ocular surface interferometer such as LipiView measures in vivo tear lipid layer thickness (LLT). However, little information about its diagnostic capability is available to clinicians. According to thin film theory, a thicker film is expected to be more stable. Further, a healthy tear lipid layer is thought to inhibit aqueous evaporation. In this study, we model non-invasive tear breakup time (NITBUT) as a function of LLT and the coefficient of variation (CV) of LLT, and estimate the sensitivity and specificity of LipiView as a tool for diagnosing tear film instability.

Methods : Data were extracted from a database, with n=546 records meeting eligibility criteria, which included having completed both NITBUT (Medmont) and tear LLT (LipiView) measurements. Subjects ranged in age from 18 to 92yrs. Mixed effects models were employed to account for repeated measures.

Results : Longer NITBUT was significantly associated with thicker minimum LLT (p=0.010). The model estimates 3.0s longer NITBUT for the thickest (100nm) vs. the thinnest (21nm) minimum LLT. The CV was also significantly related to NITBUT (p=0.027), with greater variability in LLT (range: 0.01 to 0.21) corresponding to 3.9s shorter NITBUT. Using the common clinical criterion for tear film instability of NITBUT<10s, LLT had diagnostic sensitivity and specificity of 54.3 and 64.8%, respectively, at a threshold of 52nm. The sensitivity and specificity of CV were 62.5 and 51.6%, respectively, at a threshold of 0.052. Based on those thresholds, among subjects with less variable LLT, those with thinner LLT had a mean (SD) NITBUT of 10.1 (6.4)s, and those with thicker LLT had 9.9 (6.3)s. In contrast, among those with more variable LLT, those with thinner LLT had a mean (SD) NITBUT of 7.8 (4.7)s, and those with thicker LLT had 9.6 (5.9)s.

Conclusions : Tear film (TF) stability depends on both the thickness and the variability of the lipid layer. If the lipid layer is less variable (more uniform) the TF can remain stable for thick or thin LLT. In contrast, if the lipid layer is more variable (less uniform), the TF can remain stable only if the LLT is thick. These contributions of the lipid layer to TF stability were both clinically significant (3-4s). We speculate that the variability reflects some aspect of lipid layer biochemical composition, which is important to TF stability in addition to the physical property of LLT.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.