Endovascular Coiling of the Ophthalmic Artery in Pigs to Induce Retinal Ischemia

Abstract

Purpose.: The authors recently showed that the retinal circulation can be accessed by transfemoral endovascular catheterization. The purpose of this study was to examine whether endovascular coiling can be used to induce different degrees of ischemic injury. The possibility of creating occlusions at different sites in the vasculature to cause retinal ischemia with different degrees of severity was investigated.

Methods.: The ophthalmic artery was catheterized through the external carotid system using a fluoroscopy-monitored, transfemoral, endovascular approach in 12 pigs (mean weight, 70 kg). The effects were evaluated using angiography and multifocal electroretinography.

Results.: Occlusion of arteries supplying the retina was established using endovascular coiling. Coiling in the proximal part of the ophthalmic artery caused no or little ischemia, presumably because of collateral blood supply. Coiling in the distal part of the ophthalmic artery, over the branching of the main ciliary artery, caused more severe retinal ischemia. Multifocal electroretinography recordings, which reflect retinal function in an area close to the visual streak, showed decreased amplitudes and increased implicit times after distal occlusion, but not after proximal occlusion of the ophthalmic artery. The responses were similar 1 hour and 72 hours after coiling, indicating that a permanent ischemic injury was established.

Conclusions.: The porcine ophthalmic artery can be occluded using an endovascular coiling technique. This provides an experimental animal model of retinal ischemia in which occlusion at different sites of the vasculature produces different degrees of severity of the ischemic damage.

Retinal ischemia ensues when the retinal circulation is insufficient to meet the metabolic demands of the retina. This is most commonly caused by local circulatory failure resulting from diabetes, vein thrombosis, or arterial occlusion. ¹ Experimental animal models of retinal ischemia are required to develop treatments that can limit the extent and severity of the ischemic injury. A number of animal models have been developed to study retinal ischemia, although many of these have limitations. Ischemia is often induced by elevating the intraocular pressure (IOP), which produces global ischemia identical to that seen in central retinal artery occlusion (CRAO). ^{1 –3} However, retinal injury may result from both ischemia and pressure. Vascular ligation, another common method of causing retinal ischemia, is achieved in its simplest form by placing a suture around the optic nerve bundle. ⁴ This occludes blood flow, elevates the IOP (because of pressure on the globe), and constricts the optic nerves, which damages the axons. ¹ Ligating the posterior ciliary vessels independently of the optic nerve can be performed in rats, although it is technically more demanding. This produces features similar to those seen after CRAO but also causes uveal ischemia. Incomplete ischemia can be produced by ligating more proximal arteries in the neck; the degree of ischemia produced is dependent on the number of vessels ligated. ^{5 –8} This intervention may mimic carotid insufficiency but is hampered by optic nerve ischemia and cerebral infarction. ⁸ Retinal vessels can be occluded by intravenous injection of rose bengal, a photosensitive dye, followed by intense retinal illumination, resulting in thrombosis of retinal vessels, which can mimic branch retinal artery occlusion (BRAO). ^{9 –11} However, this may cause retinal injury resulting from phototoxicity, in addition to the ischemic component. BRAO may also be produced by laser photocoagulation or transvitreal diathermia, ^12,13 but these interventions may have direct effects on the retina, such as the rupture of Bruch's membrane. 

Taken together, many of the commonly used animal models for retinal ischemia have limitations. The clinical relevance of results obtained in the laboratory depends on the nature of extrapolation. If an experimental animal model of retinal ischemia can replicate human pathology and pharmacologic treatment can ameliorate this pathology, then it is a logical assumption that such treatment may be effective in the clinical setting and, thus, merits further investigation. Clearly, the ability to extrapolate the results obtained using an animal model to the clinical situation requires an experimental model that closely resembles retinal ischemia in humans. Pigs have a retinal anatomy virtually identical to that of humans. ¹⁴ The porcine eye appears to have a typical primate-like architecture and is similar to the human eye in both size and retinal blood supply. ¹⁴ The pig has also proven to be a suitable animal for experimental analysis of the retina and retinal arteries. ¹⁵ If unambiguous retinal ischemia, without confounding factors, can be created in the pig, this may provide an experimental model that closely resembles the pathology in humans. In 1992, Scheurer et al. ^{16 –18} were successful in catheterizing the pig external maxillary artery and injecting microparticles before the branching of the ophthalmic artery. However, since the injections were administered in the maxillary artery, which is a large artery supplying large parts of the head, ischemia was presumably not produced only in the retina. We recently accessed the retinal circulation by transfemoral endovascular catheterization. ¹⁹  

One major advantage of occlusion of the retinal circulation using a transfemoral endovascular approach is that it affects only the blood supply and should not have any other, undesirable, effects on the eye. Occlusion of the vasculature supplying the retina was performed using a liquid embolic agent. ¹⁹ However, it was difficult to control the administration of the embolic agent, and the result was inadvertent clogging of the vasculature, leading to variation in the severity of ischemia. ¹⁹  

Our ambition was thus to develop a technique of inducing stable, controllable ischemia of different degrees of severity. We envisaged that endovascular coiling could be a suitable technique because it is known to allow a precise, repeatable location of the desired occlusion. This technique also decreases the risk for inadvertent clotting in arteries outside the intended location, leading to undesirable and adverse effects. This study was conducted to ascertain whether the vasculature of the porcine retina was accessible for endovascular coiling and to investigate the possibility of creating occlusions at different sites in the vasculature (proximally and distally in the ophthalmic artery) so as to create retinal ischemia of different degrees of severity. The effects of occlusion on the retina were examined using angiography and multifocal electroretinography (mfERG) 1 hour and 72 hours after the intervention to examine whether a temporary or permanent ischemic injury was created. 

Materials and Methods

Animals and Anesthesia

Twelve domestic landrace pigs of both sexes, with a mean body weight of 70 kg, were used in the study. Before the surgical procedure, the animals were fasted overnight and had free access to water. An intramuscular injection of ketamine (Ketaminol vet, 100 mg/mL; Farmaceutici Gellini S.p.A., Aprilia, Italy), 15 mg/kg body weight, in combination with xylazine (Rompun vet, 20 mg/mL; Bayer AG, Leverkusen, Germany), 2 mg/kg, was used for premedication. Anesthesia was maintained with thiopental (Pentothal, 50 mg/mL; Abbott, Stockholm, Sweden), 1 to 2 mL when necessary, in combination with fentanyl (Fentanyl B. Braun; B. Braun Melsungen AG, Melsungen, Germany) at approximately 3.5 μg/kg/h until catheterization was initiated and a vascular sheath was inserted in the femoral artery. Anesthesia was then switched to continuous intravenous infusion of propofol (Diprivan 20 mg/mL; Astra Zeneca, Södertälje, Sweden) at a dosage of 0.1 to 0.2 mg/kg/min in combination with fentanyl at approximately 3.5 μg/kg/h. Mechanical ventilation was established with a ventilator (Servo Ventilator 900B; Siemens-Elema, Göteborg, Sweden) in the volume-controlled mode. Continuous monitoring of the animal was performed during the experiment, using electrocardiography and arterial pH, PO₂, and PCO₂. After completion of the experimental procedure on day 1, before waking the animals, each animal was given an intramuscular injection of antibiotics at a dosage of 1 mL/10 kg (Streptocillin vet, 250 mg/mL dihydrostreptomycin + bensylpenicillinprocain 200 mg/mL; Boehringer Ingelheim Vetmedica, Malmö, Sweden), a subcutaneous injection of nonsteroidal anti-inflammatory drugs at a dosage of 1.4 mg/kg (Rimadyl Bovis vet, 50 mg/mL; karprofen; Orion Pharma Animal Health, Sollentuna, Sweden), and eye ointment antibiotics at a dosage of 0.1 g/eye (Fucithalmic 1%; LEO Pharma, Malmö, Sweden). The animals were wakened and monitored for adverse events. Three days later the animals were re-anesthetized using the same protocol as on day 1. After the experiment, each animal was given a lethal dose of potassium. 

Ethics

All procedures and animal treatments were conducted in accordance with the guidelines of the Ethics Committee of Lund University, the US National Institutes of Health Guide for the Care and Use of Laboratory Animals, and the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. 

Experimental Procedure

Catheterization was performed in each of the pigs. A 6-F (French) vascular sheath was inserted into the right femoral artery using a percutaneous approach (Radiofocus Introducer II, 6 F; Terumo Europe N.V., Leuven, Belgium). A 5-F angiographic catheter (Tempo 5 Headhunter Catheter; Cordis, South Ascot, UK) was then inserted through the external carotid artery into the maxillary artery using fluoroscopic guidance. Cerebral angiography of the external carotid system was performed using frontal and lateral projections. The 5-F active tracking catheter was then replaced with a 6-F guide catheter (Envoy Guide catheter, 6F; Cordis), which was inserted into the external carotid artery. The ophthalmic artery was catheterized with a microcatheter (Excelsior SL10; Boston Scientific, Cork, Ireland), advanced over a guidewire (Mirage 0008-inch, Ev3; Neurovascular, Irvine, CA). The guidewire was removed once the microcatheter was positioned, and a coiling catheter was subsequently inserted. Coils were used to occlude the artery. A coil is a thin metallic thread contained in a catheter. The coils are threaded through the catheter and deployed into the blood vessel. On exiting the catheter at the desired site, the coil takes on a 3-dimensional structure because of an intrinsic inclination to coil. The coils do not constitute an absolute barrier but lower the blood flow to the point of clot formation, which produces complete occlusion. Typically, 2 to 10 coils were needed to achieve occlusion. The coils used were the ultrasoft (GDC-US) or soft (GDC-10-Soft-SR) 2 mm × 3 cm, 2 mm × 4 cm, 2 mm × 6 cm, and 2.5 mm × 4 cm (Boston Scientific, Natick, MA). 

The coils were detached from the coiling catheter using an electric device. After coiling, occlusion was verified by performing a local angiography proximal to the coiling site. If the passage of contrast medium was seen, further coils were positioned until no leakage was observed. Coiling was performed at two different locations in the vascular tree; in the proximal part of the ophthalmic artery, before the branching of the main ciliary artery (which supplies the retina), and in the distal part of the ophthalmic artery, over the branching of the main ciliary artery. 

Multifocal ERG

The eyes were dilated with topical cyclopentolate hydrochloride (Cyclogyl 1%; Alcon Laboratories, Inc., Fort Worth, TX) to a diameter of 8 to 10 mm. mfERG was performed 1 hour and 72 hours after endovascular artery occlusion to evaluate the retinal function, in line with the ISCEV standard mfERG, ²⁰ as described here. The animals were kept in normal room light for 1 hour before and during stimulation. A Burian-Allen bipolar contact lens electrode with built-in infrared emitters (Hansen Ophthalmic Development Laboratory, Iowa City, IA) lubricated with 2% hydroxypropyl-methylcellulose (Methocel) was applied to the eye, and a ground electrode needle was inserted into the skin behind the ear. Recordings were made with a visual evoked response imaging system (VERIS Science 4.3; EDI, San Mateo, CA). The stimulus was a picture of 103 geometric patches, unscaled hexagons, delivered by a miniature cathode ray tube. The pattern seemed to flicker randomly, but each element followed a fixed, predetermined sequence (the m-sequence). The equipment was appropriately calibrated according to the manufacturer's instructions, regarding both the grid and the luminance. The light intensity in the recording area was 0.110 lux. Two additional blank, dark frames were inserted in every m-sequence. The signal gain was 100,000, and the filter range was 3 to 300 Hz with no additional notch filtering. The luminance flickered between light and dark according to a pseudorandom binary m-sequence of 75 Hz, with a mean stimulus luminance of 16.6 cd/m² and a flash intensity of 1.33 cd s/m². Spatial averaging was set to 17%, as in the settings for the visual evoked response imaging system used in the clinical setting (VERIS Clinic 4.3; EDI). One iteration of the artifact rejection system included in the visual evoked response imaging system (VERIS Science 4.3; EDI) software was used. The fundus was visualized in the infrared camera by means of the infrared light from the recording electrode, allowing continual visualization of the retina during the examination. The stimulus pattern was consistently positioned with the optic nerve head at the lower central part of the recording area. 

mfERG traces from the visual streak area were analyzed using maximum amplitude and implicit time. The visual streak area is referred to as area 1 in our previous study. ¹⁹  

Statistical Analysis

Calculations were performed using statistical and graphing software (Prism 5.0; GraphPad San Diego, CA). Statistical analysis was performed using the Mann-Whitney U test when comparing two groups and the Kruskal-Wallis test with Dunn's posttest for multiple comparisons when comparing three or more groups. Significance was defined as P < 0.05. All differences referred to in the text were statistically significant. Results are presented as mean ± the SEM. 

Results

Transfemoral Access of the Retinal Vasculature That Supplies the Retina

Angiography of the carotid system revealed the truncus bicaroticus, which splits into a left and a right common carotid artery. Faint internal but prominent external carotid circulation was seen. The common carotid artery has a small side branch, the ascending pharyngeal artery, which is the equivalent of the internal carotid artery in humans. The ascending pharyngeal artery divides into tiny infrabasal arteries that anastomose into an arteriolar network at the base of the skull, the rete mirabilis. The rete mirabilis prevents catheterization of the intracerebral arterial territory in the pig. 

The external carotid artery proved to be a continuation of the common carotid artery, and this vascular system was easily accessed by catheterization. The maxillary artery branches off from the external carotid artery. The maxillary artery gives off the temporal, the lingual, the auricular, the facial, and the buccinator arteries. The maxillary artery gives rise to the infraorbital artery. The ophthalmic artery branches off the infraorbital artery. After having accessed the ophthalmic artery, injection of a contrast medium produced the characteristic crescent-shaped outline of the retina. The ophthalmic artery gives off the main ciliary artery, from which the retinal artery branches as a single artery or several arteries. Figure 1 shows an angiogram of the internal carotid artery system, which branches into the ophthalmic artery and supplies the retina. 

Figure 1.

 

(A) Angiogram of the proximal region of the left common carotid artery in a pig (lateral view) before coiling. (B) Enlargement of the angiogram of the distal region of the ophthalmic artery of the same pig in the same projection. The ascending pharyngeal artery (1) originates as a small side branch from the common carotid artery and feeds the rete mirabilis, (2) which then converges to form the intracranial carotid artery. The external carotid artery (3) is a continuation of the common carotid artery. The maxillary artery (4) branches from the external carotid artery. The maxillary artery gives off the lingual, (5) the auricular, (6) and the buccinator (7) arteries. The maxillary artery gives rise to the infraorbital (8) artery. The ophthalmic artery (9) branches off the infraorbital artery. Black arrows: position of the intra-arterial microcatheter in the ophthalmic artery. After having accessed the ophthalmic artery, the injection of a contrast medium will produce the characteristic crescent-shaped outline of the retina (white arrows). The ophthalmic artery gives off the main ciliary artery, (10) from which the retinal artery branches.

Coiling Experiments

Coiling was performed to occlude the ophthalmic artery at two different levels of the vascular tree in the proximal part of the ophthalmic artery, before the branching of the main ciliary artery (which supplies the retina), and in the distal part of the ophthalmic artery, over (to occlude) the branching of the main ciliary artery. 

Proximal Occlusion of the Ophthalmic Artery

Coiling was performed in the proximal part of the ophthalmic artery. Angiography of the ophthalmic artery showed that blood flow through the artery was completely inhibited. However, the late angiogram frames revealed a faint retinal contour, suggesting that collateral arteries, entering the vasculature distal to this site, supplied the retina with blood (Fig. 2). The mfERG recording showed a nonsignificant tendency toward reduced amplitudes and prolonged implicit times, suggesting no or little ischemia (Figs. 3, 4). 

Figure 2.

 

Angiograms of the porcine ophthalmic artery (lateral projection) during coiling. Upper (A–C): proximal occlusion. Lower (D–F): distal occlusion. Left (A–D): angiograms before coiling. Middle: positioning of the coils (arrows) (B) in the proximal part of the ophthalmic artery, and (E) in the distal part of the ophthalmic artery, over the branching of the main ciliary artery. Right: angiograms after coiling of the ophthalmic artery. After proximal coiling (C), the retina shows weak contrast, indicating a collateral blood supply. After distal coiling (F), no noticeable contrast can be seen in the retina.

Figure 3.

 

Representative examples of mfERG responses in the pig eye subject to ischemia for 72 hours and the fellow control eye. Left: individual recordings. Right: topographic maps in which the optic nerve head (white arrow) and visual streak (black arrows) are visualized. Upper: mfERG recordings from a pig in which one eye was used as control (A) and the other eye was subject to coiling in the proximal part of the ophthalmic artery to induce ischemia (B). Lower: mfERG recordings were obtained from a pig in which one eye was used as the control (C) and the other eye was subject to coiling of the distal part of the ophthalmic artery, over the branching of the main ciliary artery, to induce ischemia (D). It can be seen that occlusion results in attenuation of the mfERG recordings, especially when the coil is placed distally in the vasculature.

Figure 4.

 

Top: amplitude and (bottom) implicit times of mfERG recordings obtained from eyes in which a coil had been placed in either the proximal or the distal part of the ophthalmic artery, over the branching of the main ciliary artery, to induce ischemia, compared with fellow control eyes. Results are shown as mean ± SEM calculated from the visual streak area. Statistical analysis was performed using the Mann-Whitney U test. It can be seen that the amplitude is decreased and the implicit time is increased, especially after distal occlusion of the ophthalmic artery, indicating ischemia.

Distal Occlusion of the Ophthalmic Artery

Coiling was performed in the distal part of the ophthalmic artery, over the branching of the main ciliary artery. Angiography of the ophthalmic artery showed that blood flow through the artery was completely inhibited and the retina could not be visualized, even in late angiogram frames, suggesting ischemia (Fig. 2). The mfERG recordings showed significant increases in implicit times and decreased amplitudes (Figs. 3, 4). The responses were similar 1 hour and 72 hours after coiling, indicating that a permanent ischemic injury was established. 

Discussion

To the best of our knowledge, this is the first study to demonstrate that endovascular coiling can be used to occlude the ophthalmic artery. Coiling may be a suitable method of occluding the vasculature because it permits precise, repeatable occlusion of the artery at a specific location. Furthermore, it is possible to induce different degrees of ischemia by placing the coil at different locations in the vascular tree, in this case in the proximal part of the ophthalmic artery, which produces no or little ischemia, or in the distal part of the ophthalmic artery, which produced significant ischemia. 

In a previous study, we characterized the vascular anatomy to reach the retinal circulation using a transfemoral endovascular approach. ¹⁹ It has been shown by others ²¹ that the cerebrovascular anatomy of domestic pigs differs from that of humans. In the pig, the rete mirabilis, which is a network of fine arterioles through which a selective angiographic catheter cannot pass, gives rise to the internal carotid artery. ^22,23 It is thus disappointing that the carotid circulation of the pig does not allow intracerebral catheterization and occlusion for cerebral infarction research. ²⁴ On the other hand, the pig has an extensive external carotid system that can be catheterized, ^{16 –18} allowing the ophthalmic artery to be accessed. ¹⁹ The ophthalmic artery in the pig gives rise to the main ciliary artery from which the retinal artery branches. ^19,25  

We believe that endovascular access to the retinal circulation is useful for creating experimental animal models for retinal ischemia. Experimental animal models for retinal ischemia without limitations and confounding factors are scarce. ¹ One major advantage of occlusion of the retinal circulation using a transfemoral endovascular approach is that it affects only the blood supply and probably has no undesirable side effects on the eye, such as pressure in high-IOP ischemia models, or nerve effects in optic nerve bundle ligation. ¹  

In previous studies by Scheurer et al., ^{16 –18} the external maxillary artery was catheterized, and microparticles were injected before the branching of the ophthalmic artery. ^{16 –18} However, since the injections were administered in the maxillary artery, which is a large artery supplying major parts of the pig's head, ischemia was presumably not restricted to the retina. We previously performed occlusion of the ophthalmic artery and the main ciliary artery in the pig to produce retinal ischemia using a liquid embolic agent. ¹⁹ However, this was difficult to control and resulted in inadvertent clogging of the vasculature, leading to variations in the severity of ischemia in the different animals. The liquid state of the agent makes it difficult to predict the exact location of occlusion, and the repeatability is therefore limited. The agent may travel along the artery and produce a more distal occlusion than intended. It is also difficult to locate the agent precisely, e.g., over a bifurcation. Further disadvantages include the risk for accidental embolization at other locations. 

In this study, coiling was used to occlude the ophthalmic artery. We found that the technique allows precise, repeatable occlusion at specific locations on the vascular tree. The experimental model is thus more reliable, and the repeatability of the experiments in individual animals has been found to be good. The more precise the location of the occlusion, the less variability and the better quality of the information obtained. This technique also decreases the risk for inadvertent clotting in arteries outside the intended location, leading to undesirable and adverse effects. 

The degree of ischemia depends on the proximity of the vascular occlusion to the retina. Proximal occlusion of the ophthalmic artery, before the branching of the main ciliary artery, produced little or no retinal ischemia, whereas distal occlusion of the ophthalmic artery produced a significant degree of ischemia. The probable reason for this is that arteries anastomose to distal parts of the vasculature, allowing collateral blood supply to the retina through, for example, the lingual artery, which may prevent retinal ischemia. The technique of endovascular coiling provides unique opportunities to vary the degree of ischemic injury to create an optimal experimental model of retinal ischemia. The ischemia persisted over 3 days, indicating that a permanent ischemic injury to the retina was indeed established. 

In conclusion, we report, for the first time, the technique of endovascular coiling to occlude the ophthalmic artery in the pig so as to provide an experimental model of retinal ischemia. Coils can be inserted proximally to occlude the ophthalmic artery or, more distally, at the branching of the main ciliary arteries to produce different degrees of ischemia. Proximal occlusion resulted in incomplete or no ischemia, presumably because of collateral blood supply to the retina. Distal occlusion, over the branching of the main ciliary artery, resulted in significant ischemia. This study is one step in the development of an experimental model of retinal ischemia that may be useful in future research aimed at developing pharmaceutical agents for the treatment of retinal ischemia. 

Footnotes

 Supported by the Magn Bergvall Foundation, the Anna Lisa and Sven-Eric Lundgren's Foundation for Medical Research, the Crown Princess Margaret's Foundation (KMA), the Visually Impaired in the County of Malmöhus, the Anna och Edvin Berger's Foundation, the Märta Lundqvist Foundation, the Swedish Medical Association, the Royal Physiographic Society in Lund, the Swedish Medical Research Council, the Crafoord Foundation, the Medical Faculty at Lund University, Lund University Hospital Research Grants, the Swedish Government Grant for Clinical Research, the Per-Eric and Ulla Schyberg's Foundation, and the Eye Foundation.

Footnotes

 Disclosure: H. Morén, None; B. Gesslein, None; P. Undrén, None; S. Andreasson, None; M. Malmsjö, None

The authors thank Jaak Roosa (Boston Scientific Nordic AB), Gisela Håkansson, and Ronald Carpio for valuable contributions.