In Vivo Assessment of Thickness and Reflectivity in a Rat Outer Retinal Degeneration Model with Ultrahigh Resolution Optical Coherence Tomography

Abstract

Purpose.: To provide in vivo quantitative assessment of sodium iodate–induced retinal damage in a rat model of outer retinal degeneration using ultrahigh resolution optical coherence tomography (UHR-OCT).

Methods.: Outer retinal degeneration was induced in four female Long Evans rats via tail vein injection of sodium iodate (40 mg/kg). Changes in the thickness and optical reflectivity of individual retinal layers were extracted using a semi-automatic segmentation algorithm and were assessed in vivo at 6 hours, days 1, 3, and 7, and up to 3 months post injection with UHR-OCT. Hematoxylin and eosin (H&E) histology was used to confirm the morphologic changes observed in the UHR-OCT images.

Results.: UHR-OCT tomograms showed progressive structural damage in the rat retina over time, such as swelling, thinning, complete disintegration of individual retinal layers, and clustering of highly reflective cellular debris. Photoreceptor swelling was observed 6 hours after injection of sodium iodate, followed by progressive structural decomposition of the outer retina. At 3 months post injection, the outer retina was completely disintegrated, and the inner nuclear layer (INL) was in direct contact with the choroid. Changes in the reflectivity of individual retinal layers were observed over time and correlated well with the morphologic changes.

Conclusions.: UHR-OCT permits in vivo, noninvasive, longitudinal, quantitative assessment of the progressive changes in retinal morphology and optical reflectivity in a sodium iodate rodent model of outer retinal degeneration.

Introduction

Photoreceptor (PR) and RPE degeneration, typical for neurodegenerative diseases such as retinitis pigmentosa ¹ and AMD ² can lead to severe visual impairment and eventually blindness. Despite extensive research in human subjects ^3–10 and animal models, ^11–19 the exact origins and mechanisms of the development of these diseases are still not well understood. Accurate in vivo evaluation of the structural changes in the retina is essential for the better understanding of the onset, progression, and response to therapy of these retinal diseases. 

Sodium iodate is a retinotoxin that is used in animal models to induce outer retinal degeneration. ^18–26 This compound is known to selectively affect the RPE cells as the primary site of damage. ²⁰ Since normal RPE function is essential for PR survival, PR functional and structural degradation follows the RPE degeneration. It has been shown that the RPE and PR cell death occurs through necrosis and apoptosis respectively. ^21,22 Although the mechanism of the RPE cell death in this model differs from that of human outer retinal degenerative diseases, the route of PR cell degeneration and the overall pattern on the degenerated retina are similar (i.e., patchy loss of RPE and PR layers^19,22,23). Therefore, sodium iodate–induced retinal degeneration can be used as a model for some human retinal degenerative diseases involving RPE and PR cell death. Until recently, all reported studies that investigated the morphologic effects of sodium iodate on different retinal layers have been conducted ex vivo using histologic and immunohistochemical methods. ^18–26 In these longitudinal studies, retinas were evaluated both qualitatively and quantitatively for morphologic changes at different time points after drug administration. The quantitative data were acquired by manual thickness measurement of different layers along with the number count of nuclei rows in the outer nuclear layer (ONL). ²⁴ Although histology and immunohistochemistry provide subcellular level resolution images of the retinal structure and very valuable information as to retinal physiology and biochemistry, these methods have a common disadvantage—they require termination of multiple animals at each time point of a longitudinal study to allow for statistical evaluation of the research results and do not permit in vivo examination of the retinal structure, function, or physiology. 

We recently reported that high-speed, ultrahigh resolution optical coherence tomography (UHR-OCT) can be used to visualize, in vivo and in 3-D, morphologic changes in the rat retina in a sodium-iodate model of outer retinal degeneration, ²⁷ which eliminates the need to terminate research animals at each stage of a longitudinal study. Compared with other optical imaging techniques such as fundus photography and scanning laser ophthalmoscopy (SLO), UHR-OCT provides real-time, cross-sectional morphologic tomograms of the retina that permit morphometric analysis in terms of individual layer thickness and shape. Furthermore, UHR-OCT can provide information about the optical properties of biological tissue in terms of wavelength-dependent absorption and scattering that can be used to infer information regarding retinal biochemistry. ^28–30 This additional information could be useful in retinal degeneration studies as a noninvasive tool for investigating cell death in vivo since it has been demonstrated recently that cell apoptosis and necrosis cause changes in the tissue optical reflectivity. ^31–32 In a recent study, UHR-OCT was used for in vivo quantification of the reflectivity of retinal layers in healthy human retina. ³³ Over the past decade, UHR-OCT has been used to image both human retina ^34,35 and rodent retina ^{14–16,36–38} in health and disease. 

In this study we used a high-speed UHR-OCT system and a custom, semi-automatic layer segmentation algorithm to image in vivo and quantify morphologic changes in the rat retina in a longitudinal study of sodium iodate–induced outer retinal degeneration. 

Materials and Methods

The Imaging System

The UHR-OCT system used in this study was recently developed by our group for imaging human ³⁹ and animal retinas. ^38,40,41 In brief, the system is based on spectral domain OCT technology and utilizes a broad bandwidth superluminescent diode (SLD, Superlum, λ_c = 1020 nm, Δλ = 108 nm, and P_out = 10 mW). The interference signal is detected by a high-efficiency spectrometer (P&P Optica Inc., Waterloo, ON, Canada), interfaced to a 1024-pixel linear array charge-coupled device (CCD) camera (SUI Goodrich, Princeton, NJ) with a 47-kHz readout rate. The system provides 3-μm axial resolution in the rat eye and approximately 99-dB sensitivity for 1.6-mW power of the imaging beam. Three-dimensional in vivo imaging of the rat eyes was carried out with a custom-designed eye imaging probe consisting of three achromatic lenses (Edmund Optics, Barrington, NJ) and a pair of galvanometric scanners (Cambridge Technologies, Lexington, MA). The probe provides a 1.2-mm diameter imaging beam, incident on the rat pupil resulting in about 5–10 μm lateral resolution at the retina. The imaging probe provides an approximately 2-mm diameter field of view (FOV) at the retina. Animals were placed in a custom holder, designed to keep the rat head immobile during imaging and provide easy alignment of the animal's eye with respect to the imaging beam. 

Sodium Iodate–Induced Outer Retinal Degeneration Rat Model

Mature, 11- to 16-week-old, female, Long Evans rats were used in this study. All animal procedures were conducted in accordance with the regulations of the University of Waterloo Animal Care Committee and in compliance with the ARVO statement for ethical use of animals in ophthalmic and vision research. Animals were anesthetized with isoflurane (2.5%), and pupil dilation was achieved by administering one to two drops of tropicamide (Mydriacyl, 1%, Alcon Canada Inc., Mississauga, ON, Canada) in each eye. Outer retinal degeneration was induced by IV administration of 40 mg/kg of sodium iodate solution. ²⁶ During imaging, artificial tears were administered frequently to keep the corneas hydrated and optically clear. Four animals were imaged before drug administration (baseline) as well as at 6 hours, on days 1, 3, and 7, and at months 1, 2, and 3 post sodium iodate injection. An additional three rats were imaged and then euthanized, one at baseline, one at day 7, and another one at month 3 post drug injection for histologic processing to confirm the morphologic changes observed in the UHR-OCT images. 

Image Acquisition, Processing

Multiple 3-D UHR-OCT tomograms were acquired from the right eye of each rat at the central region, close to the optic nerve head. The unique blood vessel pattern at the optic nerve head was used to acquire images from the same location in the retina at different time points of the longitudinal study. All UHR-OCT images were composed of 1000 A-scans × 256 B-scans × 512 pixels. Each 3-D set corresponds to a 1-mm × 1-mm physical area on the tissue. A semi-automated retinal layer segmentation algorithm developed in our research group ⁴² was used to segment individual retinal layers from the UHR-OCT cross-sectional tomograms and to calculate the layer thickness and reflectivity. The segmentation results were also used to generate en face thickness and reflectivity maps over 3-D volumes of retinal tissue demonstrating spatial variations and overall time changes in the retina morphology. 

Statistical Analysis

A 3-D set of data (1000 A-scans × 256 B-scans) was selected for each time point from each rat and was used for segmentation. Twenty equally spaced B-scans (approximately every 60 μm) were selected for segmentation while each B-scan was sampled at 50 equally spaced intervals (approximately every 24 μm), resulting in 1000 sampling points for one 3-D set of OCT images of rat retina. These data were used to acquire the thickness and reflectivity (intensity) values for each retinal layer. For statistical analysis, the missing data module of the statistical package (SPSS version 16, IBM, Armonk, NY) was used to impute five missing values (out of the total 32 values: four animals and eight time points for each animal) due to random occurrences of rats not cooperating. One-way repeated measures ANOVAs on the values were used to determine changes as a function of time. Bonferroni-corrected multiple comparison tests were used as the post-hoc test. For all statistical tests, changes were considered significant at P ≤ 0.05. 

Histology

Both eyes were enucleated and fixed in freshly prepared 4% paraformaldehyde (20 minutes) before being rinsed with 0.1 (wt/vol) mM Sorenson's phosphate buffer (SB, pH 7.5; 3 × 10 minutes). Eyes were cryoprotected with 30% (wt/vol) sucrose in SB overnight. The whole eyes were embedded in optimal cutting temperature embedding medium (Fisher Scientific, Pittsburgh, PA), then frozen. The frozen samples were sectioned at 12 μm onto clean glass slides using a cryostat and were allowed to dry before hematoxylin and eosin (H&E) staining. Cover slips were mounted onto the slides using Permount (Fisher Scientific), and images were collected using a light microscope and a color camera. 

Results

Representative UHR-OCT cross-sectional tomograms of the central retina, located approximately 1 mm away from the optic nerve head in a rat eye and imaged over a time span of 3 months, are shown in Figure 1, thus corresponding to different stages of outer retinal degeneration. H&E-stained histologic sections of a similar location in the retina, prepared at baseline, day 7, and 3 months post injection are presented as well (Figs. 1B, 1G, and 1K, respectively). All intraretinal layers and the choroid are clearly visible in the healthy retina (Fig. 1A). Six hours after drug administration (Fig. 1C), the PR outer segment (OS) structure is disrupted and the OS low reflective band is replaced by highly reflective material (Fig. 1C, bracket). A low reflective layer is observed to appear between the PR layer and RPE (Fig. 1C, arrow). One day post injection (Fig. 1D), the highly reflective material in the OS and the low reflective layer located between the PR and RPE have disappeared and the external limiting membrane (ELM) and the inner segment (IS) to OS junction (IS/OS junction) are less visible. The ordered inner and outer segment structure disappears at day 3 post sodium iodate injection (Fig. 1E), while the layered structure of the PRs is completely disrupted and replaced by highly reflective material (Fig. 1E, black arrow), which is most likely debris from PR and RPE cells. At this time point, the ELM is no longer visible, the outer plexiform layer (OPL) has become irregular in shape (Fig. 1E, gray arrow), and disruptions in the RPE layer are visible (Fig. 1E, white arrow). The image from day 7 (Fig. 1F) shows complete decomposition of the PR layer structure (Fig. 1F, black arrow) and severe irregularity in the OPL (Fig. 1F, gray arrow). One month after drug administration, the outer retina appears collapsed, and a thin, irregularly shaped layer of highly reflective material separates the inner nuclear layer (INL) from the choroid (Fig. 1H). Images acquired 2 and 3 months post injection of sodium iodate (Figs. 1I and 1J, respectively) show progressive and complete disintegration of the outer retinal structure. The INL appears to be in direct contact with the choroid due to the absence of the outer retina. Histologic sections acquired from retinal locations within the OCT-imaged 3-D volume correlated well with the morphologic changes observed with the UHR-OCT tomograms (Figs. 1B, 1G, and 1K, representing baseline, day 7, and month 3 post sodium-iodate injections). 

Figure 1.

 

Cross-sectional UHR-OCT and histology images of rat retina at different stages of the outer retina degeneration. Healthy retina at baseline (A) show all layers of a normal rat retina. Disruption of the OS structure (bracket) and formation of a low reflective layer between the PR and RPE (arrow) at 6 hours after sodium-iodate administration (C) and OS decomposition at day 1 (D) are the early changes taking place. At day 3, the PR layer is replaced by highly reflective material (black arrow), OPL is less visible (gray arrow), and there are gaps (white arrow) in the RPE layer (E). At day 7, the PR layer is fully degenerated (black arrow), and OPL (gray arrow) is disrupted (F). Images acquired 1, 2, and 3 months post injection, respectively, show complete loss of outer retina (H, I, and J). Histologic cross-sections acquired at baseline (B), day 7 (G), and month 3 (K) post drug administration correlate well with the corresponding OCT images of the same time point. NFL, nerve fiber layer; GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer; ELM, external limiting membrane; IS/OS, inner segment outer segment junction; RPE, retinal pigment epithelium; C, choroid; S, sclera.

Quantitative analysis of the structural changes over time in the outer retinal degeneration model was carried out by segmenting the individual retinal layers from the UHR-OCT tomograms and measuring their thickness and optical reflectivity as functions of time. A representative cross-sectional retinal tomogram and the corresponding segmented image are presented in Figure 2. The region between each two lines is considered one layer. In Figure 2A, all retinal layers as well as the choroid and sclera are labeled. Brackets in Figure 2B mark the regions used for quantitative assessment of the thickness and reflectivity changes. The segmentation algorithm was able to segment the choroid and nine layers in the healthy rat retina, four of which were the sublayers of the PR (dark and bright bands, a pair of each corresponding to the IS and OS of the PR layer, respectively). The number of segmented retinal layers changed over time with the outer retinal degeneration. 

Figure 2.

 

Original (A) and segmented (B) UHR-OCT tomogram acquired at baseline (day 0, before sodium-iodate injection) highlighting different retinal layers (A) and regions used for extracting thickness and reflectivity values (brackets in B).

Statistical results from the retinal layer thickness and intensity (reflectivity) measurements are presented in Figures 3 and 4 for different time points of the longitudinal study. The total retinal thickness initially increased at 6 hours post injection (P = 0.0371), before gradually decreasing at later time points of the study, with significant changes at day 7 and months 1, 2, and 3 post injection (P < 0.0001 for all time points; Fig. 3A). No significant change was observed in the choroid over time (P = 0.1055; Fig. 3B). The inner plexiform layer (IPL; Fig. 3C) thickness did not change until month 3, when it became significantly thinner (approximately 3 μm change; P = 0.0096) compared with baseline. In contrast, ONL thicknesses increased (Fig. 3D) beginning at day 1 (P = 0.0198) and remained thicker at both 3 (P = 0.0018) and 7 (P = 0.0037) days post injection. Owing to loss of the OPL, the ONL could not be segmented with high confidence at the time points past day 7 of the study. The thickness of the PR layer structure, between the IS/OS junction and the RPE, increased significantly at 6 hours (P = 0.0016) before decreasing significantly at days 1 (P = 0.0017) and 3 (P = 0.0117) (Fig. 3E). After day 3 post injection, the PR structure appeared so irregular that it could not be segmented as a cohesive layer. 

Figure 3.

 

Results from statistical thickness measurements over four rat eyes. Graphs show results for total retina thickness (A), choroid (B), IPL (C), ONL (D), and IS/OS junction to RPE (E). Asterisks at each column in the graphs indicate the statistically significant deviation of thickness from the baseline.

Figure 4.

 

Results from statistical intensity (reflectivity) measurements over four rat eyes. Graphs show results for total retina thickness (A), choroid (B), IPL (C), ONL (D), and IS/OS junction to RPE (E). Asterisks at each column in the graphs indicate the statistically significant deviation of the reflectivity from the baseline.

Figure 4 summarizes statistical results for the observed changes in the tissue reflectivity of the rat retina as a whole, as well as for individual retinal layers, as a function of time. Since the image contrast varied somewhat during the different imaging sessions owing to a slight variation in the system performance, light coupling in the animal eye, and possible differences in the image quality in different animals, the intensity of each layer was normalized to that of the corresponding IPL. This normalization assumed that the IPL reflectivity was not affected by the sodium iodate–induced outer retinal degeneration. The IPL was chosen for the normalization procedure because it appeared to remain unaffected by the drug-induced retinal degeneration. As reported in previous studies ^22,24,26 and confirmed in our study, the IPL did not change significantly over time (P = 0.4185; Fig. 4C). 

The reflectivity changes are presented in percentage by setting the baseline value (day 0, images before drug administration) equal to 100% for convenient comparison. The total retinal reflectivity decreased over time, which correlates with the initial formation of the low reflective layer, followed by gradual outer retinal degeneration (Fig. 4A). The changes were significant at 6 hours (P = 0.0113), and intensities remained lower throughout the rest of the experiment (P ≤ 0.0016 for all time points). Choroidal reflectivity did not change significantly until month 2 (P = 0.0194; Fig. 4C). The ONL reflectivity increased significantly by day 3 (P = 0.0036), indicating the collapse of the photoreceptor and redistribution of highly reflective material throughout the outer retina (Fig. 4D). The reflectivity of the PR layer structure between the IS/OS junction and RPE decreased at 6 hours (P = 0.0543; Fig. 4E) as a result of the formation of the extra, low reflective layer (see Fig. 1B, black arrow); day 1 (P = 0.0029) and day 3 (P = 0.0121) reflectivity values were also lower in this retinal layer. 

To map out spatial variations in the total retinal thickness and reflectivity over time, all B-scans in the 3-D sets acquired from one animal at all time points of the longitudinal study were segmented. Figures 5 and 6 show representative en face views of the total retina thickness and reflectivity maps (respectively) of the central region in the rat retina at different time points of the longitudinal study. Each en face map consists of 256 segmented B-scans and represents a 1-mm × 1-mm retinal surface area created using a 1000- × 256- × 512-pixel 3-D volume. Figure 5A shows the thickness distribution of a rat normal retina, where slight variation in the thickness is observed because of the presence of surface blood vessels (Fig. 5A, arrow). Six hours after drug administration, swelling of the retina is observed by thickness increase (Fig. 5B). At day 1 post injection, the degeneration is manifested by decrease in overall thickness due to decomposition of the PR outer segment (Fig. 5C). At day 3, no significant change in the thickness is observed compared with day 1 (Fig. 5D). An almost monotonic decrease in the retina thickness is observed at day 7 and month 1 in Figures 5E and 5F, respectively. 

Figure 5.

 

Topologic thickness maps reconstructed from segmenting 3-D OCT image volumes acquired from the central region of the retina at different stages of the degeneration. Each map represent an en face view of a tissue surface area of approximately 1 mm × 1 mm measured at baseline (A), 6 hours (B), day 1 (C), day 3 (D), day 7 (E), and 1 month (F) post drug injection. The color bar indicates the color coding values in micrometers. The arrow in Figure 5A points at one of the surface blood vessels.

Figure 6.

 

Topologic reflectivity maps reconstructed from segmenting 3-D OCT image volumes acquired from the central region of the retina at different stages of the degeneration. Each map represents an en face view of a tissue surface area of approximately 1 mm × 1 mm at baseline (A), 6 hours (B), day 1 (C), day 3 (D), day 7 (E), and 1 month (F) post drug injection. The color bar indicates the camera intensity values in color coding. The arrow in Figure 6A points to one of the surface blood vessels.

Figure 6 presents the total retina reflectivity (intensity) map generated from the same data sets as used for the generation of the thickness data shown in Figure 5. Each figure is a 2-D en face view reconstructed from a 3-D set by adding up all pixel values in each A-scan from nerve fiber layer (NFL) to RPE. Figure 6A shows the intensity distribution in a healthy retina as well as some changes related to the presence of surface blood vessels (arrow). Six hours after drug injection, the reflectivity distribution of the retina was altered due to the structural changes taking place in the PRL (Fig. 6B). At day 1, the overall retinal reflectivity decreased due to partial loss of the OS (Fig. 6C). At day 3, a slight reflectivity increase was observed (Fig. 6D) followed by further increase at day 7 (Fig. 6E). Because of the presence of a severely irregular distribution of the high reflective material, the surface blood vessels could not be visualized clearly at day 7 (Fig. 6E). Figure 6F shows the retinal reflectivity map at 1 month post injection. The outer retina decomposition resulted in significant reflectivity decrease at this time point. 

Discussion

The statistical thickness measurements in Figure 3 illustrate the changes in all layers at different stages of the outer retina degeneration. The IPL thickness did not vary significantly over the 3-month period of this study, as expected, indicating spearing of the inner retina, similar to what was reported in previous studies using histopathology. ^22,24,26 The ONL thickness increased by about 30% at day 3 and 25% at day 7 (with respect to baseline), most likely as a result of collapse of the PR layer. The PR structure from the IS/OS junction to the RPE was the location of the earliest observed changes in the retina. At 6 hours post injection, the PR layer thickness increased by 38% due to the appearance of an additional, low reflective band between the PR OS and the RPE. The high transmissivity and short period of time during which this layer appeared and then disappeared together suggest that the layer is most likely composed of fluid. The thickness of the PR structure decreased by 40% at day 1 and 30% at day 3, suggesting progressive PR decomposition. The statistical reflectivity measurements in Figure 4 showed intensity variations at different stages of the degeneration. As expected, the IPL showed insignificant variations in the reflectivity for all time points, which were most likely due to a slight change in the system performance, coupling of the beam into the animal eye at different imaging sessions. The ONL reflectivity increased by 65% at day 3 and 40% at day 7 with respect to the baseline most likely due to uneven distribution of highly reflective debris within the outer retina. The formation of the low reflective layer below the PR layer at 6 hours post injection led to an approximately 50% decrease in the reflectivity of the PR region between the IS/OS junction and RPE. The decrease in the intensity of this layer was followed by a change of 80% at day 1 and 65% at day 3 (with respect to baseline) most likely owing to decomposition and loss of PR cell bodies, as seen in the histology cross-section by day 7. At later time points, the PR layer could not be segmented as a continuous layer due to severe irregularities. 

The observed retinal swelling and formation of a low reflective region below the PR layer at 6 hours post-injection have not been reported in other studies. The only other method previously used to examine the retinal cross-sectional morphology in detail is histology. The procedure involved in processing tissue for histology can lead to artifacts, including the separation of the retina from the choroid, which could be a reason why this particular layer has never been observed in histopathological studies. However, some studies report swelling of RPE cytoplasmic organelles at 6 hours, a decreased number of RPE cell nuclei at 12 hours, ²¹ and breakdown of the blood retina barrier at 24 hours post drug administration, ¹⁸ all of which could be related to an influx and efflux of fluid; these movements of fluid could explain the appearance and subsequent disappearance of the low reflective region between the PR and RPE in our study. Further investigation with other noninvasive techniques may be required in order to determine the origins and composition of this newly observed layer. 

Conclusion

In this study, a state-of-the-art, high-speed, UHR-OCT system along with a semi-automatic segmentation algorithm were used to visualize, quantify, and monitor the structural and reflectivity changes of the rat retina over time in a sodium-iodate model of outer retinal degeneration. Our results showed that outer retinal damage can be characterized by changes in the thickness, shape, integrity, and optical properties of different retinal layers. To the best of our knowledge, this is the first published quantitative study of OCT imaging of retinal damage caused by sodium iodate. Results from this study could potentially lead to better understanding of neurodegenerative retinal diseases. 

Acknowledgments

The authors thank Akshaya Mishra and D. Li, University of Waterloo, for assistance with the image processing; Nancy Gibson for assistance with animal care and tail vein injections; and Shelley Boyd, St. Michael's Hospital, Toronto, and Melanie Campbell, University of Waterloo, for helpful discussions about the NaIO3 retina degeneration model. 

Footnotes

 Disclosure: S. Hariri, None; A.A. Moayed, None; V. Choh, None; K. Bizheva, None

Footnotes

 Supported by grants from the Natural Sciences and Engineering Research Council of Canada (NSERC).