Real-time OCT guidance and multimodal imaging monitoring of subretinal injection induced choroidal neovascularization in rabbit eyes

Abstract

Choroidal neovascularization (CNV) is a major cause of vision loss that consists of abnormal growth of new blood vessels from the choroidal vasculature. High resolution in vivo imaging of animal models is essential to better elucidate and conduct research on CNV. This study evaluates a novel multimodal imaging platform combining optical coherence tomography (OCT) and photoacoustic microscopy (PAM). Using real-time OCT guidance subretinal injection to induce and multimodality imaging system to monitor CNV over time in rabbit eyes. The significance of our work lies in providing the optimal setting and conditions to make use of the OCT image guided system to improve the consistency and reproducibility of experimental results in subretinal injection induced CNV model in rabbits. For the first time, this study successfully demonstrated the dual-modality PAM-OCT system, without using exogenous contrast agents, can detect and visualize CNV in the rabbit eye with high resolution. This is promising system for diagnosing and monitoring CNV.

1. Introduction

Choroidal neovascularization (CNV) occurs in several diseases which are leading causes of vision loss and blindness, including age-related macular degeneration (AMD), pathologic myopia, presumed ocular histoplasmosis, angioid streaks, and multifocal choroiditis (Cheung et al., 2017; Cohen et al., 1996; Ferris et al., 1984). CNV consist of new blood vessels growing from the choroid through breaks in Bruch’s membrane and proliferating under the retinal pigmented epithelium (RPE) or retina. These new vessels may leak and bleed, resulting in scar formation and irreversible loss of vision. AMD is the leading cause of irreversible visual impairment and blindness among elderly people in developed countries(Rahmani et al., 1996). Even with modern anti-vascular endothelial growth factor (anti-VEGF) therapy for AMD, the majority of vision loss with AMD remains due to CNV(Nguyen et al., 2018a; Stem et al., 2018).

The exact mechanism of CNV development remains an area of active investigation. Various animal models attempt to mimic CNV. A commonly used animal model is laser-induced CNV, which has been used in mice, rats, monkeys, and pigs(Kiilgaard et al., 2005; Miller et al., 1990; Semkova et al., 2003; Shah et al., 2015). Attempts to replicate laser-induced CNV in rabbits have not been successful. Rabbits have eyes with an axial length relatively similar to that of the human eye. Establishment of a successful, rapid-onset model of CNV in rabbits would be useful to better elucidate the etiology of CNV. Previous research has demonstrated that the accumulation of abnormal extracellular deposits in the space between the RPE and Bruch’s membrane plays an important role in CNV development. Recent studies have also shown that artificially created sub-RPE deposits can induce the development of CNV in rodents and rabbits(Cao et al., 2010; Kimura et al., 1995).

Previous studies have demonstrated CNV induced by subretinal injection of Matrigel and vascular endothelial growth factor (VEGF) in rabbits. Qiu et al. (2006) have illustrated that this model is a reproducible, reliable, and sustainable rabbit model of experimental CNV. Matrigel is a solubilized basement membrane preparation extracted from the Engelbreth-Holm-Swarm mouse sarcoma. The major components of Matrigel are laminin, collagen Ⅳ, heparin sulfate proteoglycans, entactin/nidogen, and growth factors. Matrigel is a liquid at 4 °C, but it forms a solid gel at 37 °C. Once the cold liquid Matrigel is injected into tissue, it solidifies, thus trapping the growth factors, permitting their slow release, and causing sustained formation of new blood vessels. However, this technique is technically challenging with a low success rate due to the difficulty in reliably introducing a subretinal injection in animal eyes. In previous publications, the success rate of this technique is 80% in the mouse eye(Johnson et al., 2008). This technique and animal model could thus be further enhanced through the use of real-time imaging to facilitate the model formation.

Optical coherence tomography (OCT) is a noncontact tomographic imaging modality that achieves micrometer-scale axial resolution via interferometric detection of backscattered near infrared light. OCT has been extended to a wide variety of applications. Intraoperative OCT (iOCT) image-guided surgery has significantly enhanced intraoperative feedback to the surgeon(Ehlers et al., 2011). iOCT has been described in surgeries including macular holes (Ehlers et al., 2014b; Ray et al., 2011), epiretinal membranes (Leisser et al., 2018), vitreomacular traction (Ehlers et al., 2014a), and retinal detachment(Lee and Srivastava, 2011). This gave us inspiration to use iOCT to improve and facilitate an animal model of CNV in rabbits. When doing subretinal injection through a modern microscope without OCT, the surgeon is limited to an en face view. The axial information must be inferred from instrument shadowing and other indirect cues like a slight resistance to the movement of the needle. However, subretinal injection requires precise manipulation of delicate tissue on the submillimeter scale. The introduction of iOCT could enhance manipulation via visual feedback, help refine current techniques, and improve the success rate and ease of performing this animal model.

As a new imaging technique, photoacoustic microscopy (PAM) has been utilized in evaluating retinal and choroidal vascular diseases. PAM is based on energy conversion from light to sound and involves optical excitation and ultrasonic detection. PAM can explore the optical absorption properties and acoustic properties of tissue(Nguyen et al., 2018b, 2018c). Recently, PAM has been demonstrated as a non-invasive, non-ionizing technique to image deep retinal and choroidal structures(Tian et al., 2017). Since 2010, several groups have built ocular PAM systems and worked on imaging of the eye(de la Zerda et al., 2010; Hu et al., 2015; Li et al., 2017; Liu and Zhang, 2016). Most reported work in PAM imaging uses mice or rat eyes, which have an axial length significantly smaller than the human eye. This is particularly important for PAM imaging since the ultrasound signal attenuates with distance, particularly for high-frequency components. Previously, our group has successfully reported that the integrated PAM and OCT dual-modality imaging platform could be applied for evaluating retinal and choroidal blood vessels in living rabbits with high temporal and spatial resolution(Nguyen et al., 2018b, 2018c; Tian et al., 2017, 2018; Zhang et al., 2018). Our system has a lateral spatial resolution of 4.1 μm for PAM and 3.8 μm for OCT at the focal plane of the objective with a high depth of penetration that allows visualization of the retinal and choroidal vasculature.

Here we developed a real-time integrated PAM and OCT dual-modality imaging system for producing CNV in rabbit eyes using a subretinal injection of Matrigel and VEGF (M&V). The dual-modality imaging system is capable both to facilitate and improve the formation of the animal model in addition to monitoring the animal model. We also perform the first PAM evaluation of choroidal neovascularization in vivo in living rabbits. With this novel system, rabbit CNV models can be more easily created and consistently monitored, which paves the way for future studies on their pathophysiology and therapeutic targets.

2. Material and methods

2.1. Animal model preparation

All animals procedures adhered to the ARVO (The Association for Research in Vision and Ophthalmology) Statement for the Use of Animals in Ophthalmic and Vision Research and were approved by the Institutional Animal Care & Use Committee (IACUC) of the University of Michigan (Protocols PRO00008566 and PRO00006486, Photoacoustic & Molecular Imaging of the Eye, PI: Y. Paulus).

Twelve New Zealand rabbits (weight 2.5–3.0 kg, 3–4months of age, both genders) were acquired from the Center for Advanced Models and Translational Sciences and Therapeutics (CAMTrasST) at the University of Michigan Medical School. The animals were housed in an air-conditioned room with a 12-h light-dark cycle, fed standard laboratory food, and allowed free access to water. All the animals received subretinal injection by OCT guidance. Eight adults New Zealand rabbits received a subretinal injection of M&V, and four rabbits were injected with sterile water as a control group. Subretinal injections of Matrigel basement membrane matrix (Corning, NY, USA) and Human VEGF-165 (Shenandoah Biotechnology, Warwick, USA) was reconstituted in 1% bovine serum albumin in sterile water at 0.1 mg/mL in stock solution. M&V suspension was prepared by suspending 750 ng VEGF (100 μg/mL) in 20 μL of Matrigel.

Rabbits were anesthetized with a mixture of Ketamine (40 mg/kg, 100 mg/mL) and Xylazine (5 mg/kg, 100 mg/mL) by intramuscular injection. A vaporized isoflurane anesthetic (1 L/min oxygen and 0.75% isoflurane) (SurgiVet, MN, USA) was provided to maintain anesthesia during the in vivo experiments. Pupillary dilation was achieved by one drop each of 1% tropicamide and 2.5% phenylephrine hydrochloride. Topical tetracaine 0.5% was instilled before treatment for topical anesthesia. Rabbit vital signs were monitored before and throughout the procedures, including mucous membrane color, temperature, heart rate, respiratory rate, and oxygen saturation using a pulse oximeter (V8400D Capnograph & SpO2 Digital Pulse Oximetry, Smiths Medical, MN, USA).

2.2. Dual-modality imaging system

A schematic diagram of the dual-modality imaging system with integrated spectral-domain OCT (SD-OCT) and Photoacoustic Microscopy (PAM), is shown in Fig. 1. The system has been described in depth previously (Tian et al., 2018). OCT detects backscattered photons by low-coherence interferometry. The SD-OCT system was adapted from a commercially available system from Thorlabs (Ganymede-Ⅱ-HR,Thorlabs, Newton, NJ, USA) by adding an ocular lens (OL) after the scan lens and a piece of dispersion compensation glass (DCG) in the reference arm. For PAM, an optical parametric oscillator (OPO) (NT-242, Ekspla, wavelength tunable from 405 to 2600 nm, pulse duration 3–6ns) was used as the illumination source. The generated photoacoustic signal propagates through the eye and is captured by a custom-built needle shaped ultrasonic transducer with a central frequency of 27.0 MHz (Optosonic Inc, Arcadia, CA, USA), which is placed in contact with phosphate-buffered saline (PBS) on the rabbit conjunctiva. The lateral resolution of PAM and SD-OCT was previously quantified to be 4.1 μm and 3.8 μm respectively. Meanwhile, the quantified axial resolution of PAM and OCT was 37.0 μm and 4.0 μm separately. In this study, the laser wavelength 532 nm and the laser energy of 80 nJ before the eye were used. The safety of the laser energy has been discussed at length in our previous work and is half of the ANSI safety limit(Tian et al., 2017).

Fig. 1.

Schematic of integrated PAM and OCT for dual-modality chorioretinal imaging and real-time OCT-guided subretinal injection. PA, photoacoustic; OPO, optical parametric oscillator; SLD, super luminescent diode; BS, beam splitter; DM, dichroic mirror; DAQ, data acquisition.

2.3. In vivo OCT/PAM imaging for CNV

After finishing the fundus camera imaging, the rabbits were transferred to the platform of the system and adjusted to position one eye under the ophthalmic lens. The head and body of the rabbit were placed on two custom-made stabilization platforms, which were used to minimize breathing and other motion artifacts. A water-circulation blanket (TP-700, Stryker Corporation, Kalamazoo, MI) was used to maintain the body temperature of the rabbits. The eyelid was retracted with a pediatric Barraquer wire speculum. The integrated charge-coupled device (CCD) camera was used to visualize the region of interest, and the reference arm was adjusted to optimize the image quality.

PAM detects the absorption of photons by major chromophores, such as oxyhemoglobin and deoxyhemoglobin in blood. To acquire PAM images, the ultrasonic transducer was mounted on a three-dimensional (3D) translation stage and positioned in contact with phosphate-buffered saline (PBS) on the rabbit conjunctiva pointing to the fundus. The photoacoustic signal was amplified by 57 dB low-noise amplifier (AU-1647, L3 Narda-MITEQ, NY). The signals were digitized and recorded using a high-speed digitizer (PX1500–4, Signatec Inc., Newport Beach, CA) with 500 MHz sampling rate. The recorded data was then used to reconstruct two-dimensional (2D) or three-dimensional (3D) images of the eye blood vessels. 2D PAM images were acquired by implementing horizontal scanning lines along the x-axis. In addition, to visualize the structure of blood vessels and the morphology of small blood vessels, 3D image reconstruction was performed using the Amira software (FEI, Hillsboro, OR).

2.4. OCT-guided subretinal injection to induce CNV

Subretinal injection was performed via a sterile technique. The conjunctiva was cleaned with 5% povidone-iodine. A 3-mm inferotemporal conjunctival peritomy was performed in the eye using Westcott scissors and micro-forceps. After a marker was used to measure a distance 3.5 mm posterior to the corneal limbus, a 30-gauge ¹/₂ inch needle was inserted at about the 8:00 to 8:30 position with care to avoid the ciliary vasculature and extraocular muscles. Under the OCT system, a blunt 30-gauge needle (Hamilton, Reno, NV) attached to a 50 μL Hamilton syringe (Hamilton, Reno, NV) was inserted through the 30-gauge ¹/₂ inch needle inserted site into the vitreous cavity, then introduced into the subretinal space about 0.5–1 optic disc diameter inferior to the retinal vessels, to inject 20 μL of Matrigel and 750 ng (7.5 μL) VEGF or 27.5 μL sterile water. The OCT system is used for real time guide and assisting in confirming the injection depth by providing the anatomy of the injected retinal layers with 10.74 Hz frame rate. The surgeon looked at the head-up OCT display monitor. When the needle was in contact with the RPE hyper-reflective layer, the surgeon stopped progressing the needle, maintained the needle position, and then pushed the syringe. The injection resulted in a localized subretinal bleb/neurosensory retinal detachment. During experiments, eyewash (Altaire Pharmaceuticals, Inc., Aquebogue, NY) was applied to the rabbit corneal surface every 2 min to prevent corneal surface keratopathy. Vitals and rectal temperature were monitored and recorded. Topical antibiotic was applied, such as 0.2 mg Neomycin and Polymyxin B Sulfates and Dexamethasone Ophthalmic Ointment and 2 drops of Flurbiprofen, and 0.3 mg/kg Meloxicam were injected subcutaneously at the conclusion of the procedure.

2.5. Fundus photography, fluorescein angiography (FA) and indocyanine green angiography (ICGA) and quantitative intensity level analysis of FA images

Another follow-up examination procedure included color fundus photography, FA, ICGA. These imaging studies were performed on all animals before and after subretinal injection for five consecutive weeks, focusing on the injection site. After anesthesia and pupil dilation, the rabbit fundus was first imaged using a fundus camera (TRC 50EX, Topcon Corporation, Tokyo, Japan). Fluorescein sodium (0.2 mL, 10% solution) (Akorn, Lake Forest, IL, USA) was administered intravenously through the marginal ear vein of the anesthetized rabbit. Photographs were acquired rapidly during the transit phase and then at a minimum each 30-s for 10 min. Indocyanine green (0.5 mg/kg, 5 mg/mL, HUB Pharmaceuticals LLC, Patheon, Italy) was injected in the marginal ear vein, and photographs were subsequently acquired for at least 5 min. Following this, the dual-modality imaging system was used to acquire OCT and PAM images.

The fluorescence intensity of average grey value was calculated using ImageJ. The hyperfluorescent CNV lesion was determined using the free hand tool. Background fluorescence intensity was measured by calculating the average background fluorescence intensity of six representative areas in an unbiased manner. Thus, the fluorescence intensity was the CNV hyperfluorescence intensity subtracting the calculated background value(Shah et al., 2006; Wigg et al., 2015).

2.6. Histology

For histological analysis, the whole eyes were immersion-fixed in Davidson’s (Hartmann’s) fixative for 24 h. The eyes then were rinsed in phosphate buffer and dehydrated in a series of graded alcohols. After dehydration, the anterior cap including the cornea, iris, and lens were removed. A rectangular segment of posterior eye wall (retina, choroid, and sclera), including the entire area of Matrigel deposit, was dissected. The samples were embedded in paraffin, and then sectioned serially and stained with Hematoxylin and Eosin (H&E). The sections were examined and photographed using a Leica DM6000 Microscope (Wetzlar, Germany).

3. Results

Subretinal injection was adequately performed in all 12 eyes with the real-time image-guided system. All the M&V eyes (8 eyes) showed CNV and mild tortuosity of the small retinal vessels near the injection site. None of the sterile water controls (4 eyes) have CNV or vascular tortuosity. Following the subretinal injection, four out of eight M&V eyes showed small subretinal or preretinal hemorrhage, which didn’t affect the progress of CNV and spontaneously resolved within 1–3 weeks. All M&V eyes and all control eyes showed the same trend. No suprachoroidal hemorrhage, deep choroidal hemorrhage, or choroidal detachment was observed. No rabbits developed endophthalmitis, retinal tear, rhegmatogenous retinal detachment of the periphery, or cataract formation.

3.1. Subretinal injection causes subretinal bleb by OCT guidance

Real-time OCT imaging was utilized to improve live assessment and manipulation of the needle location and depth. OCT guidance played an important role in determining the depth of needle injection in order to avoid deep or shallow injections, which could result in either a deep choroid hemorrhage, intravitreal injection, or significant injection leakage into the vitreous cavity. This study is the first to use real-time OCT to guide subretinal injection to induce CNV in rabbit eyes. OCT data was visualized intraoperatively as B-scans displayed on the computer screen. Fig. 2 shows excerpts of real-time CCD camera fundus images and B-scans. The red arrows in A1–D1 are corresponding to the B-scan location in A2–D2. Fig. 2(A1, A2) to Fig. 2(D1, D2) demonstrate the process of injection and changes in retinal morphology over time. Fig. 2 (A1, A2) presents the needle touching the surface of the retina. As the injection progresses, the injection needle reached the RPE layer shown in Fig. 2(B1, B2). In Fig. 2(A1) and Fig. 2 (A2), no obvious change in injection site can be seen. The change in the needle position can be seen by comparing Fig. 2(B1) and Fig. 2(B2). From Fig. 2(C1, C2) and Fig. 2(D1, D2), where one can see the process of retinal detachment and retinal bleb formation after injecting either M&V or control. Supplemental Videos A and B show the real-time OCT and CCD camera video of subretinal injection, respectively.

Fig. 2.

OCT B-scan recording of subretinal injection procedure. The images were viewable during injection by the manipulator with the heads-up display. Excerpts from the time series are shown in fundus images (A1–D1) and corresponding B-scans (A2–D2). The bottom row cross-section OCT images acquired along the scanning red arrows from top row images. The yellow arrow on the images demonstrates the needle of injection. The RPE layer is clearly visualized in B-scans. The OCT shows the retinal changes during the needle contacting the retina, initiation injection, and injecting substance into the subretinal space to induce retinal detachment during and after injection. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Supplementary video related to this article can be found at https://doi.org/10.1016/j.exer.2019.107714.

3.2. Subretinal bleb at early stage

Before and after injection of sterile water and M&V, color fundus photography, ICGA, FA, and OCT imaging were used to observe the changes in the retina, which are shown in Fig. 3. As show in Fig. 3(A–C, G–I), all rabbits were documented as having healthy retinas by fundus photography, FAs, and ICGAs prior to performing subretinal injection. OCT images of rabbit eyes are different from those obtained in humans, primates, and rodents as rabbits have partially myelinated nerve fiber layer at the meridian zone called the medullary ray. From OCT images (Fig. 3(A1–3, G1–3)), retinal detachments were not detected prior to the onset of experimental CNV. The retinal elevation/bleb caused by M&V or sterile water could be detected in all cases after injection. The dotted circles in Fig. 3(D–F, J–L) are the areas of subretinal bleb caused by subretinal injection. At an early stage after injection, all eyes demonstrated mottled subretinal hyperfluorescence, corresponding to the site of the retinal bleb. OCT images (Fig. 3(D1–3, J1–3)) demonstrate retinal detachment post-injection in the cross-section of the retina. In this model, M&V and sterile water all had low reflectivity on OCT. Two eyes injected with M&V caused a small amount of hemorrhage with highly reflective OCT signal on the surface of the retina as shown by red arrows on the OCT images.

Fig. 3.

Color fundus photography, indocyanine green angiography (ICGA), fluorescein angiography (FA), and OCT imaging before and after performing OCT-guided subretinal injection of sterile water and Matrigel & VEGF in rabbit eyes: A–F are images of the control group rabbit eye, which received sterile water injection. G–L are the experimental group images which received Matrigel & VEGF injection. A,D, G, J are color fundus photography of rabbit eyes. B, E, H, K are ICGA images. C, F, I, L are FA late-phase images. A1–A3, D1–D3, G1–G3, J1–J3 are cross-sectional OCT images acquired along the scanning white dotted arrows from figure A, D, G, J, respectively. A–C, A1–A3, G–I, and G1–G3 are normal images of rabbit eyes before subretinal injection. D–F, D1–D3 are images acquired immediately after injecting sterile water. J–L and J1–J3 are images acquired immediately after injection of Matrigel & VEGF. The dotted circles show area of the retinal detachment caused by subretinal injection in images D–F and J–L. In dotted circles region, the choroidal vessels become blurred due to retinal detachment. White short arrows represent the position of the retinal vessels. Yellow arrows indicate the injection site. Red arrows show a small area of hemorrhage on the surface of the retina caused by injection. R, retina; Ch, choroid; S, sclera; Ss, subretinal space; RPE, retinal pigment epithelium. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

3.3. Follow-up of angiography images

In the control group (Fig. 4(A)), all 4 sterile water-injected eyes demonstrated mottled subretinal hyperfluorescence with sharp margins on ICGA and FA images at day 1 at the bleb location. However, the area receded at day 4, and the hyperfluorescence became mild. At 1 week, the bleb disappeared on color fundus photography, and the ICGA and FA images returned to normal at the same time. There was no significant difference in the results from week 1 to week 4 after injection with sterile water. In Fig. 4(C), the left panel show the quantitative grey value of average fluorescence intensity of FA late-phase in control group, which indicate the fluorescence intensity receding at day4.

Fig. 4.

(A) Representative fundus photographs, FAs, and ICGAs at different timepoints of control rabbit following subretinal injection with 27.5 μL sterile water. At day 1, the bleb was caused by sterile water injection. The bleb area demonstrated sharply demarcated hyperfluorescence in both early and late fluorescein frames at 1 day after injection. The bleb receded by day 4 and resolved at 1 week after injection. The FA and ICGA returned to normal after 1 week. (B) Representative images of rabbits after subretinal injection of Matrigel & VEGF. Fundus photographs, ICGA, early, and late phases of the FAs are shown from top to bottom. Results at week 1 showed hemorrhage at the injection site. At week 3, 4, and 5, late leakage can be seen on FA. Arrowheads indicate fluorescein leakage. (C) Representative average fluorescence intensity of FA late-phase was calculated at different time point in control group and M&V group respectively. Values represent average grey value ± SD (in control group, n = 4; in M&V group, n = 8).

In all the M&V injected eyes, fluorescein leakage occurred at week 1 as illustrated in Fig. 4(B). FA demonstrated progressive late hyperfluorescence with blurred margins consistent with leakage from CNV. The subretinal hyperfluorescent leaking lesion increased in the first 4 weeks after injection and decreased slightly in week 5 (shown in Fig. 4(B) arrowhead). During the follow-up imaging, ICGA leakage demonstrated similar late leakage as FA images. In Fig. 4(C), the right panel show the quantitative grey value of average fluorescence intensity of FA late-phase in M&V group, which indicate the fluorescence intensity increasing in first 4 weeks.

3.4. Follow-up of OCT images

OCT imaging was performed after FA and ICGA imaging identified the subretinal CNV lesions. Fig. 5 shows in the control group retinal edema and detachment over time after subretinal injection of sterile water. At early stage (Fig. 3), we can see the retinal bleb appeared around and includes the retinal vessels. The OCT images on Fig. 5(A1–A4) illustrate that the retinal detachment area becomes smaller and the retinal vascular area was attached but with retinal edema. At day 4 (Fig. 5(B1–B4)), the retinal edema was gone, the retinal detachment displaced and moved at the inferior area of the retinal vessels. By 1 week (Fig. 5(C1–C4)), the retinal detachment resolved completely. However, the retina overlying the bleb region demonstrated retinal atrophy. The morphological structure of the retina was stable after 1 week, and was similar between weeks 1 and 4.

Fig. 5.

OCT data at different timepoints from control rabbits with sterile water injection. (A1–A4) corresponding OCT images acquired along the dotted arrows as shown in figure A. The OCT demonstrates retinal detachment and edema on both sides of the retinal vessels in the bleb area at day 1. (B1–B4) is the B-scan OCT images 4 days after injection, acquired at the white dotted arrows in figure B. OCT images illustrate the retinal edema resolved, while the retinal detachment shifted to inferiorly and the extent of retinal detachment was enlarged. No subretinal hyper-reflective substance was observed. (C1–C4), (D1–D4) cross-section B-scan OCT images acquired along the dotted arrows in figure C and figure D. The images were the results of 1 week and 4 weeks after injection, respectively. At 1 week, the retinal detachment resolved. The retina overlying the previous bleb demonstrated atrophy. There was no significant change in the retina from 1 to 4 weeks.

Fig. 6 demonstrates OCT data for five weeks of a rabbit that received subretinal injection of M&V. At week 1 (Fig. 6(A1–A4)) unlike with the injection of sterile water, the retinal detachment remained after injection of M&V. And there is no obvious fibrous or vascular tissue in the subretinal space at week 1. At week 3 (Fig. 6(B1–B4)), the OCT images illustrated medium reflective patches (arrowhead) above the discontinuous band representing the broken RPE/choriocapillaris layer, which represent vascular or fibrotic tissues. At week 4 (Fig. 6(C1–C4)) and week 5 (Fig. 6(D1–D4)), the retinal detachment is smaller than before, and moderately hyper-reflective dots and patches (arrowheads) have appeared in its original position. At week 5, the dots became more hyper-reflective.

Fig. 6.

OCT data at different timepoints of a rabbit with subretinal injection of Matrigel & VEGF. (A1–A4) corresponding OCT images acquired along the dotted arrows as shown in figure A. The OCT images demonstrate retinal detachment remained at week 1. (B1–B4) cross-section B-scan OCT images acquired along the dotted arrows in figure B at 3 weeks after injection. The OCT images illustrated hyper-reflective patches (arrowhead) connected with a discontinuous retinal pigmented epithelial layer in the subretinal space, representing vascular or fibrotic tissues. The retinal detachment is reduced from 1 week. (C1–C4), (D1–D4) are the B-scan OCT images acquired at white dotted arrows in figure C and figure D, at weeks 4 and weeks 5 after injection, respectively. The OCT images at weeks 4 and 5 demonstrated highly reflective dots and patches (arrowhead) in the subretinal space, the extent of retinal detachment was further reduced.

3.5. In vivo PAM imaging

This is the first study that uses photoacoustic microscopy (PAM) imaging to observe CNV in rabbit eyes. Fig. 7 displays PAM imaging of the rabbit model of subretinal injection with sterile water and M&V at week 4. Similar to the FA and OCT results, PAM demonstrated that there were signs of CNV in the M&V group. The color fundus photograph (Fig. 7(A, E)) demonstrates the morphology of the retinal and choroidal vessels. In the fundus photographs, the white dotted rectangle shows the selected scanning region (PAM) of Fig. 7(B) and (F). The PAM image presents the individual structure of the retinal and choroidal vessels with high contrast. Fig. 7(B) presents the visualization with PAM of the major retinal vessels, choroidal vessels, and retinal microvessels, indicating the capacity of PAM for detection of individual blood vessels in vivo. In the control group, these blood vessels appear normal. At 4 weeks after injection of M&V, the retinal vessels in the fundus of the eye became thinner. With fundus imaging alone, shown in Fig. 7(E) arrowhead area, the morphological changes of small blood vessels are not clear. In addition to the clear retinal vascular margin, Fig. 7(F) arrowhead area also shows the tortuous morphological changes of small vessels around the main retinal vessels. Since Fig. 7(B) and (F) are 2D, one cannot visualize the 3D depth of the vascular layers. The three-dimensional (3D) images of the vessels are reconstructed and illustrated in Fig. 7(C, D, G, H). The 3D videos are shown in supplemental videos C and D. Fig. 7(C and G) exhibit coronal plane images, and Fig. 7(D and H) display sagittal plane images. The sagittal plane images can distinguish the retinal and choroidal vessels. Fig. 7(D) shows that the retinal and choroidal vessels are close together, which is consistent with the results of retinal atrophy shown on OCT imaging at 4 weeks after injection of sterile water. Fig. 7(H) depicts the retinal vascular layer is above the choroidal vascular layer, and there is subretinal space between the two layers. Small blood vessels can be seen in the subretinal space, that is choroid neovascularization (CNV). This result not only proves that subretinal injection of M&V can induce CNV in rabbit eyes, but also indicates that PAM can effectively image the rabbit CNV model without the use of an exogenous contrast agent.

Fig. 7.

Photoacoustic Microscopy (PAM) imaging of the retinal and choroidal vessels in rabbit model of subretinal injection sterile water (A–D) and Matrigel & VEGF (E–H) at 4 weeks. A, E: color fundus photography of retina. White dotted rectangle shows the selected scanning region (PAM) of figure B and F. C–D, G–H: 3D reconstruction of PAM images from different angles, C and G are coronal images. D and H are parasagittal images. The PAM images show clearly the structure of individual vessels including RVs, CVs, capillaries and CNV. D: Control group of rabbit vessels. The RVs and CVs are adjacent to one another given the significant retinal atrophy that occurs at 4 weeks. No obvious neovascularization is observed. H: In Matrigel & VEGF group, the RVs layer and CVs layer are separated, and neovascularization (CNV) can be seen in the subretinal space. (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Supplementary video related to this article can be found at https://doi.org/10.1016/j.exer.2019.107714.

3.6. Histology analysis

The histology results shown in Fig. 8 were used to demonstrate the irregular vascular structures, which were visualized with the dual-modality system, were in fact choroid neovascularization induced by subretinal injection with M&V as demonstrated by the presence of vascular endothelial cells under the neurosensory retina. Fig. 8(A) and (B) were normal controls, representing the normal cellular structure of the retina and choroid. In the region of the new subretinal blood vessels, RPE abnormalities were observed with the absence of the RPE and Bruch’s membrane. Blood vessels were observed to extend from the choroid into subretinal space (Fig. 8(C and D) black arrows).

Fig. 8.

Histological analysis of normal and Matrigel & VEGF induced choroidal neovascularization in rabbits. B, D are 40x magnification of regions in figure A and B, respectively. A, B: H &E stained images from normal rabbits. C, D: H& E staining of rabbit retina and choroid 2 months after Matrigel & VEGF injection induced CNV. New blood vessels are visualized. Black arrows indicate regions of choroidal neovascularization with an abnormal increase in vascular cells. R, retina; C, choroid; S, sclera; NFL, nerve fiber layer; IPL, inner plexiform layer; INL, inner nuclear layer; OPL, outer plexiform layer; ONL, outer nuclear layer. Magnification: A, 20 × ; B, 40 × ; C, 20 × ; D, 40 ×.

4. Discussion

This study presents high quality, non-invasive visualization of retinal and choroidal vasculature and neovascularization in New Zealand rabbits in vivo with combined PAM and OCT imaging. This study utilizes real-time OCT image-guidance to facilitate the performing of subretinal injections in rabbit eyes for the induction of CNV. This is also the first study to successfully demonstrate a dual-modality PAM and OCT imaging system for detection and visualization of CNV in the rabbit eye. This study is also the first to evaluate choroid neovascularization in larger eyes with dual-PAM and OCT imaging and demonstrate in 3D the spatial distribution of the choroidal vasculature and neovascularization. In comparison with currently available imaging techniques, the dual-PAM and OCT system allows the detection and visualization of individual capillaries with high contrast and high resolution. The dual-modality imaging system has excellent lateral and axial resolution of 4.1 and 37 μm for PAM and 3.8 and 4.0 μm for OCT. PAM imaging can be achieved noninvasively below the American National Standards Institute (ANSI) laser safety limit (Jiao et al., 2010; Tian et al., 2017).

This study utilizes real-time image guidance with OCT to facilitate subretinal injections to induce CNV in the rabbit eye. Most animal models that successfully induce CNV are physical destruction of RPE/Bruch’s membranes. The laser induced CNV model is currently the most widely used in vivo model, yet it has limitations. No successful cases of laser induced CNV have been reported in rabbit eyes, likely due to the difference in the anatomy of the rabbit eye. We also unsuccessfully attempted to induce rabbit CNV with a 532 nm retinal photocoagulator laser (Vitra, Quantel Medical, data unpublished). In 2006, Qiu G. et al. (Qiu et al., 2006) developed a novel method that using subretinal injection M&V to induce rabbit CNV. They found that this method provides a highly reproducible, reliable, and sustainable rabbit model of experimental CNV. In 2010, Cao J. et al. (Cao et al., 2010) induced CNV in adult Sprague-Dawley rats by subretinal injection of Matrigel. Based on the previous study, VEGF is a major stimulator of new blood vessel formation. Matrigel is rich in growth factors such as IGF, EGF and bFGF. When it at 37 °C,the Matrigel will be a solid gel and slow release the growth factors. The dual effect of VEGF and Matrigel at subretinal space can effectively promote choroidal angiogenesis. Our study uses the OCT guided subretinal injection to optimize induction of CNV. Compared to conventional microscopy manipulation, the image-guide system may be more accurate, convenient, and efficient. No cover glass is required to convert the corneal surface to a planar surface.

Similarly, Dai C. et al.(Dai et al., 2018) have presented that dual-modality PAM and OCT is a valuable tool for in vivo evaluation of the laser-induced CNV model in rats. However, the eyeball size of rats is small (axial length of rat eyeball ~6 mm) compared with those of human (23 mm). The rabbit axial length is 18.1 mm, which is similar to that of humans and three times larger than that of rats. Since PAM signal is attenuated by distance from the detector, the larger axial length can pose a challenge for PAM imaging of rabbit eyes. However, our work demonstrates that PAM could play an important role in the diagnosis of CNV related disease such as AMD, degenerative myopia, angioid streaks, and multifocal choroiditis.

Currently used clinical technologies to image CNV include fundus photography, FA, ICGA, OCT, and OCT angiography (OCTA). All these imaging technologies have their own advantages and limitations in CNV imaging. Fundus photography only provides structural information. FA provides limited information of the choroidal circulation, since fluorescein can permeate the choroidal vascular wall. ICGA imaging of CNV can be ambiguous and difficult to interpret. In addition, both FA and ICGA are invasive and require the injection of an exogenous dye. They also cannot specify the depth of the vasculature. OCT may not distinguish CNV from subretinal fibrosis, hemorrhage, or adjacent retinal tissue, such as retinal pigment epithelium (RPE) and Bruch’s membrane. OCTA provides vascular imaging via motion contrast processing of decorrelation signals(Patel et al., 2018). It provides both vascular and structural information. However, OCTA is unable to show leakage, provides limited view of microaneurysms, has a limited depth of penetration, and has a restricted field of view often with motion artifacts that require careful interpretation (de Carlo et al., 2015; Told et al., 2018).

Unlike OCTA images are essentially motion-contrast images, PAM imaging contrast is defined based on the optical absorption properties of tissue. Hemoglobin is one of the primary endogenous absorbers in the eye, and its visualization can be used to image blood vessels and bleeding non-invasively. Even though the vessel size and blood flow in neovascularization are smaller and slower than normal vessels, PAM is still sensitive enough to detect neovascularization and the micro-vasculature. Furthermore, PAM has better image penetration than OCTA, can provide improved visualization of the deeper layers, like the choroid. While OCTA has difficulty imaging the choroid even using a longer wavelength light at 1050 nm(Spaide et al., 2018). Hence, by using PAM, the retinal and choroidal vessels can be identified, and the neovascularization between the two layers can also be visualized clearly with 3D depth information. In addition, PAM imaging can obtain functional imaging of the eye, such as oxygen saturation and blood velocity, while current OCTA technology can only provide limited quantitative information about the blood flow. PAM imaging can improve imaging sensitivity and specificity by using exogenous contrast agents like gold nanoparticles and targeted molecular probes.

There are some limitations to this study. Performing intraoperative OCT requires some practice and is different from microscope-based surgery. In addition, this study is performed on only 12 rabbits, and thus has a small sample size. This study was performed on New Zealand rabbits, which lack melanin, and thus the impact of RPE melanin on the imaging cannot be evaluated. Some studies have proposed that Near Infrared Spectrum (NIR) is likely to be a solution for PAM imaging of choroidal visualization with pigmented RPE in vivo, but further testing is needed. Therefore, future studies will focus on optimizing the OCT guidance system and finding methods to penetrate the pigmented RPE layer and improve choroidal visualization.

5. Conclusion

This is the first study demonstrating that real-time OCT guidance can be used to perform subretinal injection to induce CNV in rabbit eyes in dual-modality OCT and PAM imaging system. Our findings provide the optimal setting and conditions to make use of the OCT image guided system to improve the consistency and reproducibility of experimental results in subretinal injection induced CNV model in rabbits. This is also the first study to successfully demonstrate a dual-modality PAM and OCT imaging system for detection and visualization of CNV in the rabbit eyes. The results demonstrate this multimodality imaging system can noninvasively visualize choroid neovascularization in larger animal eyes.