Expansion of Schlemm's Canal by Travoprost in Healthy Subjects Determined by Fourier-Domain Optical Coherence Tomography

Abstract

Purpose.: To determine the effect of travoprost 0.004% on Schlemm's canal (SC) in healthy human eyes using Fourier-domain optical coherence tomography (FD-OCT).

Methods.: Twelve healthy volunteers were recruited for a double-blind, placebo-controlled, randomized, and paired comparison study. Right eyes of subjects were randomly assigned to receive either travoprost 0.004% or placebo; the contralateral eye received the other treatment. FD-OCT imaging of SC and measurements of IOP were carried out before and at 1, 2, 4, 6, 8, 12, 24, 36, 48, 60, 72, and 84 hours after eye drop instillation.

Results.: After instillation of travoprost eye drops, IOP gradually reduced, and the SC lumens expanded, while those values remained unchanged in placebo treated eyes. At 8 hours after the travoprost administration, the mean SC area increased 90.30% and 90.20%, respectively, in the nasal and temporal quadrant of the treated eyes as compared with the placebo group. The SC area and IOP showed a similar pattern of changes at most time points examined. In travoprost-treated eyes, a statistically significant correlation between SC area and IOP is observed (r = −0.2817; P = 0.0004). Measurements of the SC area showed sufficient repeatability and reproducibility.

Conclusions.: SC can be noninvasively imaged and quantitatively assessed in the living healthy human eye by FD-OCT. Travoprost treatment leads to SC lumen expansion accompanied by a drop of IOP in the healthy eye, likely as a result of the enhancement of pressure sensitive trabecular meshwork outflow induced by travoprost.

Introduction

Travoprost 0.004% is a prodrug of a prostaglandin F2 alpha (PGF2α), which is used topically to treat elevated IOP. It effectively reduces IOP in glaucoma and normal subjects. ^1–3 It is generally recognized that PGF2α analogs induce the synthesis of matrix metalloproteases (MMPs) in the ciliary body and sclera ^4,5 and increase uveoscleral outflow (“pressure insensitive”). ^2,3,6 However, it remains controversial whether PGs also affect conventional (trabecular) outflow (“pressure sensitive”). While some studies reported an increase of trabecular outflow by 45%, ^3,7–9 most of these results were derived from functional tests such as measurements of aqueous humor flow, tonographic outflow, and episcleral venous pressure by Schiotz tonography or fluorophotometry, and so on. Currently, the in vivo morphological evidence that supports such a notion is still missing. 

Previous studies have suggested that glaucoma patients have narrower Schlemm's canals (SCs) than healthy individuals. ¹⁰ Reduced SC size may be associated with elevated IOP because the dimension of SC is closely related to trabecular outflow. ^10–12 Past observations of reduced SC size and its correlation with outflow facility were obtained in cadaveric eyes. Therefore, characterization of the in vivo morphologic effects of PGF2α analogs on the pressure sensitive outflow pathway, especially SCs, may help us elucidate the ocular hypotensive mechanism of these drugs. 

Noninvasive imaging of the anterior ocular segment has become possible using newly developed optical coherence tomography (OCT) techniques. ¹³ OCT technology has been revolutionized by the development of Fourier-domain (FD) techniques, which provide a significantly increased signal-to-noise ratio and improved resolution as compared with traditional Time-domain OCT. ^14–16 Using FD-OCT, high definition images of SC and its surrounding tissues were successfully acquired from the living eyes and quantitatively evaluated in a noninvasive manner. ^17–20 The purpose of the present study was to take advantage of the technology of FD-OCT for the study of the effects of travoprost 0.004% on SCs in healthy subjects. 

Methods

This was a double-blinded, placebo-controlled, randomized, and paired comparison trial. Human study was approved by the institutional review board of the Eye and Ear, Nose, and Throat (ENT) Hospital of Fudan University (Shanghai, China), and signature of informed consent was obtained from all subjects in accordance with the Declaration of Helsinki. 

Subjects

Twelve healthy volunteers were recruited from the staff and faculty of the ophthalmology department of the Eye and ENT Hospital of Fudan University. Comprehensive ophthalmic examinations were conducted in all subjects. The recruitment criteria includes normal, healthy volunteer, age 21 or older, with IOP between 10 and 21 mm Hg in both eyes and asymmetry of IOP between the two eyes at 2 mm Hg or less, an angle open to grade 3 or grade 4 by the Shaffer method, with no history or evidence of ocular disease, trauma, or surgery. Subjects with myopia or hyperopia more than ±3 diopters (D) were excluded. At the time of recruitment, the SCs must be completely visible in nasal and temporal quadrants in both eyes by FD-OCT that was operated at a central wavelength of 830 nm (RTVue OCT; Optovue, Inc., Fremont, CA). 

Imaging of SC

Seated subjects were examined by a single examiner (HH) under normal indoor illumination. A CAM-L lens (cornea lens adapter; Optovue, Inc.) was mounted over the FD-OCT imaging aperture. The FD-OCT imaging was performed according to the CL Angle protocol (software version 4.0.7.5; RTVue OCT; Optovue, Inc., Fremont, CA). In this scan mode, a rectangular area of 3.0 × 2.3 mm centered on the limbus was imaged by averaging 32 horizontal B-scans and 1024 A-scans, at a duration of 0.04 seconds. For angle imaging, the subject was directed to look away to the side of the instrument at an externally fixed light source so that the iridocorneal angle of the subject was centered in the instrument's field of view. The cross-sectional angle images of FD-OCT were obtained from two sites of each eye: one image scanned the angle at the 3 o'clock position and the other at the 9 o'clock position, representing the nasal and temporal angles of each eye, respectively. The scans of each site were repeated multiple times (>3 times), and three images were chosen for final analysis based on image quality. The scans of SC ostia at collector channel (CC) junctions were excluded from this study. During the examination, the subjects were encouraged to open their eyes as wide as possible on their own without the help of the examiner in order to avoid putting additional pressure onto the eye. The superior and inferior quadrants of the eyes were not scanned because in order to expose the superior and inferior limbus, pulling eyelids during these scan acquisitions would be technically inevitable. 

Image Analysis

All scans were exported in a Joint Photographic Experts Group (JEPG) format and analyzed separately by an examiner (SZ) masked to randomization data using image analysis software, Image-Pro Plus (version 6.0; Media Cybernetics, Inc., Bethesda, MD). According to the previous studies, SC was judged as observable when the thin, black, lucent space was found outside of the trabecular meshwork (TM; Fig. 1). ^18,19 The SC areas of nasal and temporal angle were measured separately. The area of SC in each site was recorded as the arithmetic mean of measurements from three images. The repeatability and reproducibility of the measurements was assessed in all subjects (24 eyes, 48 quadrants). Images were measured for SC area by different observers (SZ and XC). All images were remeasured again in an interval of 1 week by the same observer (SZ). Both of the observers were masked to the results of the other observer. 

Figure 1. 

 

SC in a B-scan image in CL-Angle scan mode. Boundary of SC was sketched out.

Treatment and Follow-Up Protocol

The study was placebo controlled and double blinded. Right eyes of subjects were randomly assigned to receive either travoprost 0.004% (Travatan; Alcon Laboratories, Fort Worth, TX) or placebo (TEARS NATURALE II; Alcon Laboratories); the contralateral eye received the other treatment. To minimize the chance of an instillation error, all eye drops were instilled by a technician based on randomization grouping. Subjects were instructed to blot each eye with a separate tissue and not to touch the eyelid afterward, thus, reducing the chance of transferring traces of the drug from one eye to the other. Neither the subject nor the investigators knew which drug was administered in any particular eye until the end of the study. 

At 8 AM on the baseline visit, subjects first received FD-OCT scan and IOP measurement with Goldmann tonometer. The pressure was measured first in the right eye, then the left eye, and then repeated in the same sequence for three times. The pressure of each eye was recorded as the arithmetic mean of its three measurements. After the baseline evaluations, travoprost or placebo eye drops were instilled to each eye in same subject according to the results of randomization. All subjects were instructed to come back for FD-OCT scan and IOP measurement at 1, 2, 4, 6, 8, 12, 24, 36, 48, 60, 72, and 84 hours after the instillation of eye drops. The same procedure being conducted at the baseline evaluation was followed for all FD-OCT scans and IOP measurements. 

Statistical Analysis

Data are shown as mean ± SD. χ² test was used to test the distribution of right and left eyes in two groups. Area of SC and IOP of eye treated with travoprost were compared with the same variables of the contralateral eye treated with a placebo at each time point using two-sided paired t tests. Differences were considered significant if P values were less than 0.05. The association between SC areas and IOP were tested using Pearson's correlation and linear regression. One-way ANOVA was performed to evaluate the differences of SC area and IOP within the same group. The repeatability and reproducibility coefficients and intraclass correlation coefficients (ICCs) for the measurements of SC area were also assessed. The coefficient of repeatability was defined as 2 SD of the differences between the measurements obtained from the same subjects in a different session by the same observer (SZ). The coefficient of reproducibility was defined as 2 SD of the differences between the measurements obtained from the same subject by different observers (SZ and XC). All of the tests were performed by using a statistical software package (SPSS for Windows, version 17.0; SPSS, Inc., Chicago, IL). 

Results

The average age of the 12 healthy subjects who completed the study was 27.5 ± 3.9 years (range, 24–36). Five of these subjects were male, and seven were female. All of them were Chinese. 

SC was identified in all subjects throughout the observation. At baseline studies (8 AM, before the eye drops instillation), SC area and IOP showed no significant differences between the groups. The results are summarized in the Table. There were no significant differences in nasal and temporal SC areas within the same group (P = 0.983 and 0.647 in travoprost and placebo treated groups, respectively). 

In placebo group, the areas of SC remained stable in both nasal and temporal quadrant at all time points examined. However, after the instillation of eye drops, the SC lumens (nasal and temporal quadrant) expanded obviously in travoprost treated group (Figs. 2, 3) and reached the peak value at 8 hours after the instillation in both quadrants. As compared with the placebo group (nasal: 2853.76 ± 791.44 μm², temporal: 2862.25 ± 1036.27 μm²), the mean SC areas of travoprost-treated group increased 90.30% (5430.83 ± 1609.47 μm²) and 90.20% (5443.91 ± 1663.84 μm²) in the nasal and temporal quadrant, respectively. The SC area between the placebo and travoprost-treated groups showed significant differences (P < 0.05) at 2, 4, 6, 8, 12, 24, 36, and 48 hours in the nasal quadrant, and at 1, 2, 4, 6, 8, 12, 24, 36, 48, and 60 hours in the temporal quadrant. However, no significant difference was observed for SC areas in the nasal and temporal quadrant of the same eye at all time points (P > 0.05) in both groups. 

Figure 2. 

 

SC area changes after eye drops instillation. In placebo group, SC area remained stable throughout the whole observation period. In travoprost group, SC lumens expanded obviously in both nasal and temporal quadrant in 1 or 2 hours after the instillation of travoprost. The expansion effects had been maintained to 48 or 60 hours after administration. *P < 0.05 travoprost versus placebo in nasal quadrant. #P < 0.05 travoprost versus placebo in temporal quadrant.

Figure 3. 

 

SC lumen changes over time of a travoprost-treated eye. (A) Baseline status; (B) 8 hours after travoprost treatment; (C) 36 hours after travoprost treatment; (D) 84 hours after travoprost treatment.

Table. 

 

Baseline SC Data before Eye Drops Instillation

	Travoprost Group, n = 12 Eyes	Placebo Group, n = 12 Eyes	P Value
Right eyes, n = 12	5	7
Left eyes, n = 12	7	5	0.414
SC area of nasal, μm2	2759.10 ± 823.84	2763.54 ± 664.11	0.983
SC area of temporal, μm2	2909.63 ± 827.30	2848.80 ± 903.07	0.647
IOP, mm Hg	13.41 ± 1.28	13.65 ± 1.69	0.490

The average IOP before eye drop instillation on day 1 (baseline) was 13.41 ± 1.28 mm Hg in the travoprost designated eyes and 13.65 ± 1.69 mm Hg in the placebo designated eyes. Although a decline in IOP was observed during this study in both groups, there were significant differences (P < 0.05) at 2, 4, 6, 8, 12, 24, 36, 48, and 60 hours after instillation between the two groups (Fig. 4). IOP gradually reduced in travoprost group with a mean reduction of 2.23 ± 1.02 mm Hg from the baseline (P < 0.001), while the placebo group showed the lowest level of IOP at 8 hour after instillation with a mean reduction of 1.03 ± 0.87 mm Hg (P = 0.0127). In the above studies, SC area and IOP level of travoprost-treated group showed a similar pattern of changes. 

Figure 4. 

 

IOP changes after eye drops instillation. IOP declined in both groups. However, IOP in the travoprost treatment group decreased more significantly. *P < 0.05 travoprost versus placebo.

By averaging nasal and temporal SC areas taken from each eye at each time point, we observed no significant correlation in the placebo group (r = 0.0056; P = 0.9449), whereas SC area and IOP of travoprost-treated eyes showed a significant correlation (r = −0.2817; P = 0.0004; Fig. 5). The ICC of the SC area measured by the same observer (SZ) was 0.95. The ICC of SC measured by two different observers (SZ and XC) was 0.90. Thus, the repeatability and reproducibility of this parameter was excellent. 

Figure 5. 

 

Graph of SC area versus IOP in travoprost-treated eyes. Also shown is a least squares error regression line.

Discussion

SC serves as the initial collecting system for the majority of aqueous humor outflow. Previous studies suggest that SC dimensions play an important role in influencing the aqueous humor outflow. Strong correlation exists between the outflow capacity and SC dimensions. ^10,21 However, measurements of human SC dimension had been limited to histologic sections until recent years. 

With the development of FD-OCT technology, structures of aqueous outflow facility, including SC and CCs, can be noninvasively assessed in the human eye. ^17–19 Using FD-OCT, SC in healthy subjects can be readily identified. Kagemann et al. reported successful images of SC through nasal and temporal assessments of both eyes in all 21 healthy subjects. ¹⁸ Usui and colleagues found that at no less than 85.0% to 90.0% of times, SC was visible in FD-OCT images (based on observations of 34 normal eyes of 17 subjects and 26 eyes of 13 subjects with a shallow peripheral anterior chamber). ¹⁹ In the study performed by Asrani and coworkers, SC and the TM were visualized in all of the 12 participants (six healthy subjects and six patients with glaucoma). ¹⁷ In the present study, SC could be identified in all eyes nasally and temporally throughout the whole observation. This may not be entirely surprising as our inclusion criteria was that the SC could be identified in nasal and temporal quadrants in both eyes at baseline evaluation. In any case, these studies indicate that FD-OCT is a promising method to observe SC in living human eye, at least in healthy subjects. 

SC cross-sectional areas have been measured in fixed tissues in histologic studies. Following the same protocol that Kagemann et al. used in their study, ¹⁸ we calculated the overall mean and SD of the SC area, and the 95% confidence interval (CI, mean ±1.96 × SD). The 95% CI of baseline SC area (24 eyes) in the present study ranged from 1388 to 4253 μm², if the measurements of nasal and temporal SC areas were averaged to yield the SC area of each eye. At the peak of SC lumen expansion phase (8 hours after travoprost treatment), the 95% CI of SC area (12 eyes) were from 2341 to 8534 μm². When 15% of tissue shrinkage associated with fixation was accounted, ²² the magnitudes of SC area measurements in the present study are in agreement with the published values. ^10,23,24 An important factor that can cause errors of measurements in the present study is that SC area measurements vary greatly (40%–50%) depending on the effect of the CCs according to Kagemann's reports. ¹⁸ In fact, it is almost impossible to perform the scan on the exact same position every time. To circumvent this issue, multiple scans (>3 times) were made at 3 or 9 o'clock at each time point and three high quality images were selected for final analysis. The scans on SC ostia at CC junctions were excluded. 

It is commonly believed that juxtacanalicular connective tissue (JCT) and SC inner wall endothelium are important sites of resistance in the aqueous outflow system. The relationships between SC dimension and outflow resistance were also closely examined. Based on microsurgical and perfusion studies, Grant and coworkers reported that the diameter of the SC lumen was IOP dependent and reached minimum at high pressures, but was enlarged under the low pressures. ²⁵ The aqueous outflow decreased in a linear fashion with increased perfusion pressure. ^11,12 Grant and colleagues ²⁵ proposed that the apposition of SC inner wall and outer wall contributed to the increasing resistance as well as abnormal resistance in glaucoma. Extensive SC wall apposition began to develop at relatively low pressures (20–25 mm Hg) in the living eye. ^12,26 SC inner wall endothelial cells undergo pressure-dependent configuration changes: they progressively separate from the underlying juxtacanalicular space along with the decrease of outflow in response to IOP increases. ^25,27–29 These hydrodynamic changes are likely driven by morphologic changes associated with SC collapse and herniation of meshwork tissue into CC ostia. ²⁷ Progressive herniation of the inner wall and JCT into the CC ostia may be an important additional factor contributing to the decrease in aqueous outflow under elevated IOP. ^27,30 According to the perfusion study in excised human eyes by Van Buskirk, the width of SC decreased nearly 50% when IOP increased from 5 mm Hg to 20 mm Hg. ³¹ In our studies, the maximum SC area increased over 90% with 2.23 mm Hg IOP reduction in travoprost group as compared with the placebo group. Therefore, dilation of SC lumen in travoprost-treated group cannot be attributed by merely a mechanical effect of IOP reduction. In glaucomatous eyes, similar correlation exits between SC dimension and outflow resistance. Allingham and coworkers found that the dimensions of SC in glaucomatous human eyes were significantly smaller than those in the normal eyes in the histologic study. ¹⁰ If the results of normal subjects and patients with POAG are combined, a statistically significant correlation is observed between the outflow capacity and SC area or the length of the SC inner wall. Reduction of SC dimensions alone contributes approximately 50% of the decreased aqueous outflow measured in the POAG eyes. Kagemann et al. measured aqueous outflow structures, including SC and CCs using FD-OCT in the eyes of living humans. ¹⁸ They found that SC areas were significantly smaller in glaucoma patients than in normal subjects, suggesting an important role of SC dimensions in influencing aqueous outflow. The change of SC dimension, thus, may be served as a reliable indicator of TM outflow capacity. 

PG analogues are effective ocular hypotensive agents being used increasingly for the treatment of elevated IOP. In the present study, the maximum IOP reduction by a single dose of travoprost was 2.23 ± 1.02 mm Hg (P < 0.001) comparing with the baseline value. Although the pressure-lowering effects of the PGs have been known for almost two decades, the exact mechanism remains elusive. Several studies showed that the increase in aqueous outflow induced by PGs, at least in primates, is due to increased flow through the uveoscleral pathway. ^32,33 However, not only laboratory investigations but also clinical studies are suggesting that PGs may also affect conventional outflow facility. ^{3,7–9,34,35} In the study of Bahler, ³⁴ latanoprost significantly increased outflow capacity in cultured human anterior segments. This capacity change occurred within one hour of drug treatment, which continued to increase over the first 12 hours. The changes were reversible, requiring approximately 48 hours to return to the pretreament values. All of these results are in accordance with the results showed in the present study. In a clinical trial, the effects of three prostaglandin analogs, bimatoprost, latanoprost, and travoprost, on aqueous dynamics in the same subjects were compared. ³ The results showed that all medications increased outflow capacity as compared with placebo. Together, these results suggest that the hypotensive effects of travoprost are caused partially by the enhancement of TM outflow capacity. Our results provide morphological evidence to support this notion. 

Several cellular and molecular studies suggest the possible mechanism through which PGs increase TM outflow. ³⁶ First, FP receptors have been identified to associate with human TM cells, as determined by the presence of mRNA, protein, and a functional response to PG agonists. ^37–39 MMPs are expressed by human TM. ⁴⁰ PGs have direct effects on MMPs, which initiate the degradation of extracellular matrix and play a major role in regulating the resistance to aqueous flow through the tissues. Second, PGs and selective laser trabeculoplasty may share a common mechanism of pressure-lowering effects. ⁴¹ PGs affect SC endothelial cells directly or indirectly through modulating cytokine levels (IL-1a, IL-1b, TNF-a, and IL-8) produced by TM cells. ^41,42 PGs can induce intercellular junction disassembling and a 4- to 16-fold increase in conductivity of cultured SC cells. Histologic examination in Bahler's study found that PGs can cause focal detachment and loss of SC endothelial cells and extracellular matrix in some areas of the TM. ³⁴ All these phenomena can be explained by cytoskeletal and focal adhesion changes. Cytoskeletal disrupting agents (Y-27) can increase outflow facility significantly. ⁴³ PGs cause disassembling of actin stress fibers and inhibition of phosphorylation of paxillin and other focal adhesion proteins in aortic smooth muscle cells. ⁴⁴ Finally, FP receptor agonists can block endothelin-1–induced contraction of the TM. Evidence indicates this inhibition is mediated by the FP receptor. ⁴⁵  

There are several limitations to the present study. We cannot carry out each scan on the same position every time. Although we have managed to minimize error of SC measurements by averaging data from three images, rapid positional changes may still affect the outcome. This suggests that using a faster scan mode at the junction of CC ostia, like what Kagemann et al. ¹⁸ have done, maybe a good choice. A customized scan pattern will be needed to achieve this purpose for the equipment used in current study. Samples of SC were limited to 3 and 9 o'clock sites. The data of two points we took might not be fully representative of the overall changes in SC. Histologic study proved that there were often appreciable differences in SC morphometric measurements from one quadrant to another within the same eye. ¹⁰ The sample size (12 subjects) was also relatively small. However, the self control, placebo control, and randomization methods of this study may reduce the possibilities of error. Future work shall include a comprehensive 360° survey of SC and more subjects. The mean age of our sample is young (27.5 ± 3.9 years). We do not yet know if the same effects of travoprost would be observed in elderly subjects. However, the difficulty of SC observation will increase correspondingly with age. In order to study the performance of FD-OCT in imaging the fine angle structures in aged people, we scanned the angle structures of 96 healthy Chinese subjects over 50 years old and found relatively lower detection rate of SC, which was 38.81% in nasal angle and 68.66% in temporal angle, respectively (Chen J, Hong J, Huang H, unpublished data, 2011–2012). It is possible that longer wavelength systems are more robust to shadow artifact. In the present study, we have only observed the effects of one dose travoprost. The long-term effects on conventional outflow deserve further examination. 

In conclusion, SC can be noninvasively imaged and quantitatively assessed in living, healthy, human eyes by FD-OCT. With this method, the expansion of SC lumen induced by travoprost was observed. We propose that travoprost may decrease IOP, partially, by enhancing TM outflow. 

Acknowledgments

The authors thank Dong Feng Chen from Schepens Eye Research Institute for critical reviews of the manuscript. They also thank Chunshi Liu and Fengying Mao for excellent technical assistance. 

Footnotes

³  These authors contributed equally to the work presented here and should therefore be regarded as equivalent authors.

Footnotes

 Supported by grants from the Key Clinical Program of the Ministry of Health (2011-36) and the National Natural Science Foundation of China (NSFC81020108017).

Footnotes

 Disclosure: J. Chen, None; H. Huang, None; S. Zhang, None; X. Chen, None; X. Sun, None