Dynamics and function of the tear film in relation to the blink cycle

Abstract

Great strides have recently been made in quantitative measurements of tear film thickness and thinning, mathematical modeling thereof and linking these to sensory perception. This paper summarizes recent progress in these areas and reports on new results. The complete blink cycle is used as a framework that attempts to unify the results that are currently available. Understanding of tear film dynamics is aided by combining information from different imaging methods, including fluorescence, retroillumination and a new high-speed stroboscopic imaging system developed for studying the tear film during the blink cycle. During the downstroke of the blink, lipid is compressed as a thick layer just under the upper lid which is often released as a narrow thick band of lipid at the beginning of the upstroke. “Rippling” of the tear film/air interface due to motion of the tear film over the corneal surface, somewhat like the flow of water in a shallow stream over a rocky streambed, was observed during lid motion and treated theoretically here. New mathematical predictions of tear film osmolarity over the exposed ocular surface and in tear breakup are presented; the latter is closely linked to new in vivo observations. Models include the effects of evaporation, osmotic flow through the cornea and conjunctiva, quenching of fluorescence, tangential flow of aqueous tears and diffusion of tear solutes and fluorescein. These and other combinations of experiment and theory increase our understanding of the fluid dynamics of the tear film and its potential impact on the ocular surface.

Introduction

The human tear film is a very thin layer of fluid, approximately 3 microns thick (King-Smith et al., 2000). Although thin, it provides a critical function in the eye's optical system and serves to nourish, protect and enhance the differentiation of surface epithelial cells (Bron et al., 2004, Govindarajan and Gipson, 2010, Montes-Mico et al., 2010, Tutt et al., 2000). When unstable, the tear film may stress and potentially alter the underlying ocular surface, resulting in a common condition known as dry eye (DE), which affects millions in the U.S and elsewhere (DEWS, 2007, Schaumberg et al., 2003, Uchino et al., 2011, Viso et al., 2009). However, despite the common occurrence of the DE condition, knowledge of tear film dynamics remains inadequate to understand fluid dynamics during the complete blink cycle, perhaps due to the rapidity of the changes taking place in the very thin film. This review combines recent imaging of tear film dynamics and other experimental measures with mathematical modeling of tear film parameters to deepen our understanding of the dynamic processes taking place in the tear film during the complete blink cycle.

Tear film instability, which includes both rapid tear thinning and tear break-up (TBU), is considered a core mechanism of DE, along with tear film hyperosmolarity (DEWS, 2007). It is thought that tear film instability occurs due to increased evaporation and leads to increased tear film osmolarity, which stresses the ocular surface and leads to a vicious cycle of inflammation and hyperalgesia. Ultimately this repeated stress can lead to the surface alterations and neural and cellular damage seen in severe dry eye (Baudouin et al., 2013, DEWS, 2007).

Traditionally, the lipid layer has been thought to retard evaporation of the tear film, so that a more stable lipid layer should theoretically increase tear film stability. Mishima and Maurice (1961) provided evidence that the tear film lipid layer of the rabbit can reduce the rate of evaporation by a factor of 15. However, some in vitro studies show that meibum (meibomian lipid) has little or no effect on evaporation when spread on the surface of saline (Borchman et al., 2009, Brown and Dervichian, 1969, Cerretani et al., 2013). This discrepancy between in vivo and in vitro studies of evaporation may be due to differences between the tear film lipid layer and meibum spread on saline. In support of this idea, in vivo studies show an inverse correlation between lipid layer thickness and both tear film thinning rates and tear break-up time. As discussed at the beginning of Section 5, there is additional extensive evidence for the role of evaporation through the lipid layer in tear film instability and dry eye; however, the altered viscoelastic properties in dry eye disorders (Georgiev et al., 2014) may also contribute to tear film instability directly, or by alterations in the lipid layer which cause increased evaporation.

Recent imaging of TBU combined with sensory measures have linked TBU to discomfort, pain and increased DE-like symptoms and suggested that tear film osmolarity may reach 800–900 mOsM within local areas of TBU (Liu et al., 2009). However, tear film osmolarity is difficult to measure directly over the ocular surface and is currently estimated by sampling tears from the lower meniscus (Bron et al., 2014, Lemp et al., 2011). Mathematical modeling of changes in tear film fluorescence within areas of tear film instability has provided a method to estimate tear film hyperosmolarity within the same regions over the ocular surface (Braun et al., 2014). In this paper, we add to previous results from mathematical models of TBU (Peng et al., 2014) and we present new results for osmolarity all over the exposed ocular surface. We also include mathematical models for the fluorescent intensity, in order to help understand this process quantitatively and how fluorescence is related to the osmolarity. Since solute concentrations are not measured outside the meniscus, these quantitative estimates may currently be the best source of concentrations available.

The tear film is usually studied in the inter-blink period, where an unstable tear film quickly becomes evident as TBU progresses rapidly (Liu et al., 2006). This is measured clinically by fluorescein tear break-up time (TBUT), or the first appearance of a dark or “dry spot” after a blink (Norn, 1969). However, conditions within areas of TBU and predisposing factors in the tear film have been difficult to measure experimentally, although theoretical mechanisms for TBU have been proposed (Fatt, 1991, Holly, 1973, Sharma and Ruckenstein, 1985). Recent experimental evidence, combined with mathematical modeling, points to an unequal distribution or composition of lipids and evaporation as driving mechanisms for TBU and thinning (Braun et al., 2014, King-Smith et al., 2013a, King-Smith et al., 2009, King-Smith et al., 2008).

The fullness of the blink may affect the stability of the tear film and its distribution in the inter-blink period (Harrison et al., 2008). Theoretically, the tear film is “painted on” by the upper lid with each blink. Surface tension acts to smooth the tear film, improving it as an optical surface. However, many blinks are partial, leaving a visible line or groove at the fullest extent of the blink (Himebaugh et al., 2009), which may affect the optics of the tear film if the groove occurs within the pupil. Recent experimental data and modeling of tear film thickness above and below the groove has shown that thickness is decreased above the groove compared to below. Theoretically, the groove is smoothed by leveling, which occurs faster with a thicker tear film.

Aside from the inter-blink period, the dynamics of the tear film is more difficult to study, presumably due to the difficulty in imaging the ocular surface and tear film surrounding a blink. Just after a blink, the lipid layer and tear film move upwards for approximately 1–2 s, theoretically driven by the unequal distribution of lipids following the blink (Berger and Corrsin, 1974, King-Smith et al., 2009, Owens and Phillips, 2001). Stroboscopic imaging of the lipid layer and tear film, presented here, has confirmed that tear film lipids tend to be relatively stable from blink to blink (Bron et al., 2004), and shown that a rippling pattern is visible in the tear film during the downstroke and upstroke of the blink. These findings underscore the need to investigate the tear film during the complete blink cycle to provide a broader understanding of tear film dynamics under normal conditions and pathologies such as DE.

This review of recent work presents a combination of experimental imaging of the tear film with mathematical modeling designed to provide additional understanding of tear film dynamics. Recently, there are increasing efforts in this direction resulting in increasing synergy between theoretical and experimental approaches. This review will touch on previous results from both approaches, and will introduce a number of new results. These new results highlight what we believe are profitable interactions of the theory and experiment so that each is better understood. The bulk of the review will present experimental results and the corresponding results and interpretation from mathematical models. Some speculation regarding less well-understood phenomena as well as possible future directions is included.

The principles of the Declaration of Helsinki were followed for all studies in this paper and the studies were approved by the respective Biomedical Institutional Review Boards of the Ohio State University and Indiana University. Informed consent was obtained from all subjects after explanation of the procedures involved in each study.

The blinking process and tear film dynamics have been imaged with a variety of techniques. Doane (1980) hid a high speed camera behind a one-way mirror and filmed subjects relaxing while waiting for what they thought was the upcoming test. The images were subsequently hand processed to reveal lid motion and speed as well as other aspects of blinking and tear supply and drainage (Doane, 1981). Progress in optics and computing has led to more automated approaches; we only cite a few of the possible studies (Cruz et al., 2011, Wu et al., 2014). Results from a new stroboscopic imaging system are presented here to illustrate both the blink process and the dynamics of the tear film related to the blink cycle.

Fig. 1 is a “blink composite” showing images of the tear film surface using the stroboscopic imaging system during a partial blink. An area of 6 mm in diameter was recorded with a microscopic system based on previous designs (Doane, 1989, King-Smith et al., 2006). A high performance video camera, Basler (Ahrensberg, Germany) avA1600-gc, was used with a video rate of 67 frames per second at a resolution of 1400 × 1100 pixels. Stroboscopic illumination with a flash duration of 0.04 ms was used to eliminate blur from eye and tear film movements. Images on the left and right correspond to the downstroke and upstroke of the blink respectively. The labels at the edges of the panel indicate the following: L, M and S correspond to large, medium and small exposed areas of the ocular surface, respectively, while B and A correspond to before and after the blink. Thus LB corresponds to a large area imaged just before any lid movement in the downstroke. MB and SB are the medium and small areas before the blink (downstroke), respectively; SA and MA are defined similarly (upstroke). LA was recorded at a long time after the blink, just before the next blink.

Colors are generated by optical interference in the lipid layer (Doane and Lee, 1998, King-Smith et al., 1999, Millar and King-Smith, 2012). The contrast of the images has been enhanced by a factor of 2, and an approximate scale of lipid thickness in nm is given in the key on the right (Millar and King-Smith, 2012). Lipid thickness less than about 100 nm generates little color, but thicker lipid generates stronger colors. It should be noted that, in addition to recording lipid thickness, the optical system is also sensitive to any distortions of the tear film surface. The meniscus is generally not visible in these images because light is reflected at angles that do not pass into the camera.

The following aspects may be noted. First, during the downstroke, there are distortions or “ripples” on the tear film surface – yellow asterisks in Panels MB and SB. The reason for using the term “ripples” will be described later. These Panels can be compared to Panel LB where the surface is smoother and the image shows the lipid layer without obvious surface distortions. Second, some thicker (more colored) lipid is seen to accumulate under the lid in the downstroke. This indicates a downwards movement of the lipid layer in the downstroke, particularly just under the lid. This lipid movement is much slower than the downward movement of the lid; this implies that lipid accumulates just under the lid, particularly in the meniscus and perhaps on the lid margin which are not visible in the image (Doane, 1994). Third, at the position of lowest descent of the lid, the “turning point”, a groove is formed in the tear film surface which is indicated between small yellow arrows in Panels SA and MA (Deng et al., 2013, Heryudono et al., 2007). Fourth, the thick lipid which has accumulated under the lid is released at the beginning of the upstroke – shown by the strong colors in Panel SA. Fifth, the final image, LA, is very similar to the initial image LB (Bron et al., 2004).

The following caricature of the blink cycle is useful for this review of the tear. The blink cycle is considered to be in four parts and is illustrated in Fig. 2. The downstroke is the motion of the superior lid in the inferior direction to where it stops; where it stops is the turning point. The turning point appears to have some separate phenomena from both the preceding downstroke and the succeeding upstroke. For this review, we take the upstroke to be the superior lid motion in the superior direction plus the events directly tied to this motion that occur in the first roughly two seconds following the lid motion. The interblink is the subsequent time until the next downstroke begins. The collection of all of these events is a blink cycle.

We note that the separation of the different parts of the blink cycle is not always clean or easy. For example, for some subjects with MGD, TBU may actually occur just as the superior lid motion is ending, even though we tend to place TBU firmly in the interblink in this review and in our thinking of the blink cycle in general. Some events that are often thought of as occurring early in the interblink, we placed in the upstroke; optical quality measurements often span those events and well into the interblink, and we view that as a method that straddles the two parts of the blink cycle.

In this cycle, the tear film is both disturbed and replenished. For example, during the upstroke, fluid from the meniscus and the tears under the lid are distributed onto the ocular surface, which replenishes the tears there. The process is analogous to a brushing paint onto a wall; a first quantitative approach using this concept was given by Wong et al. (1996). A number of mathematical modeling papers have further refined the approach to the problem by merging the upstroke and the interblink into a single model (Aydemir et al., 2011, Jones et al., 2006, Jones et al., 2005, Jossic et al., 2009, Maki et al., 2008) or merging the four parts of the blink cycle into a single model (Braun and King-Smith, 2007, Deng et al., 2014, Deng et al., 2013, Heryudono et al., 2007). It is also the case that the deposited tear film is not perfect, and the tear film may have imperfections or so-called “disturbances.” Among a number of possibilities, we mention two here. The lipid layer should ideally be spread uniformly over a uniform aqueous layer. However, the meibomian lipids are secreted from orifices spaced about every 1 mm around the lid margins, and the lipids often do not spread to a completely uniform state (e.g., Goto and Tseng, 2003, King-Smith et al., 2011). Such imperfections may lead to TBU (King-Smith et al., 2013c). A new example of imperfections, or disturbances, will be given below in connection with the downstroke and upstroke: we present evidence that ripples form on the tear film surface during lid motion.