Evaluation of Suprachoroidal Injection in Cadaveric and Preclinical Models

Abstract

Purpose: The suprachoroidal space (SCS) is a new route for delivering therapeutics to the posterior eye. Reliably reaching the SCS is difficult in humans and animal models and necessitates thorough validation of drug delivery techniques. This study quantified SCS coverage in human cadaveric eyes using micro-computed tomography (μCT) and developed injectors optimized for preclinical animal anatomy resulting in reliable SCS access.

Methods: Dynamic μCT captured volumetric images during a 100 µL SCS injection in 10 cadaveric human eyes from five donors. Quantitative measurements of injection coverage, thickness, distribution, and center of volume were calculated. To improve preclinical SCS delivery, our novel TS-Micro injector was scaled to accommodate non-human primate (NHP) anatomy. Devices were tested in vivo (30 total injections, 14 animals) where delivery success was evaluated via direct visualization, and intraoperative imaging.

Results: Delivery to the SCS was successful in 100% of uncompromised human cadaver eyes (80% of all eyes). Injections of 100 µL in human SCS covered 33.3% ± 5.9%. Four-dimensional μCT showed that infusate initially spreads circumferentially from the injection point then posteriorly. Delivery to NHP eyes was successful in 95% of procedures resulting in coverage consistent to human cadaveric testing.

Conclusions: Results provide the first quantifiable measurement of SCS injection spread in human anatomy with validation of suprachoroidal drug delivery in the NHP that mirrors coverage in human.

Translational Relevance: Cell and gene therapies require precise delivery for therapeutic efficacy, necessitating quantification of delivery in preclinical and human anatomy, as inaccurate delivery impacts efficacy of drug candidates and confounds toxicology/dose range studies.

Introduction

Suprachoroidal injection (SCI) is a new route for delivering therapeutics to the back of the eye, specifically to the central posterior retina. Briefly, the suprachoroidal space (SCS) is a potential space which lies between the sclera and the choroid of the eye, extending approximately from the ora serrata to the posterior pole.¹ Access to the SCS can be gained via a sclerotomy or needle puncture and readily opens when fluid is injected, offering a non-surgical route of administration to the posterior retina, alternative to the intravitreal route. 

Several studies have investigated the distribution of infusates injected into the SCS and have found the spread and clearance of fluid to be dependent on particle/molecule size and infusate viscosity.^2–5 As SCI is an attractive delivery route for gene therapy, much research has been spent identifying capsids with the highest tropism in the SCS to maximize transgene expression.⁶ However, much of the research focused on evaluating the extent of SCS coverage uses qualitative methods such as infrared or fluorescent imaging and histology to characterize spread rather than using quantitative methods such as micro-computed tomography (μCT), which produces calibrated volumetric imaging data. 

Additionally, the majority of studies investigating access to the SCS rely on preclinical animal models as a surrogate for human anatomy. Although necessary, species differences in ocular anatomy and physiology have been found to impact drug delivery and development.⁷ Specifically, reliably reaching the SCS is not trivial even in human anatomy. In clinical trials, failure to reach the SCS on first attempt occurred in 29% of patients when using a perpendicular injection.⁸ Preclinical SCS injections are even more challenging because of thinner sclera, smaller orbit size, and limited eye exposure. From our experience even when using published techniques, such as depth-limited needles, success rates are as low as 50% where inadvertent intravitreal or subretinal injections are common and not easily identified until tissue collection.^9–11 This lack of injection reliability and ability to target the desired tissue greatly impacts the ability to effectively evaluate the efficacy of drug candidates and can confound toxicology/dose range studies. 

The purpose of this investigation was first to quantitatively measure the SCS coverage of a 100 µL suprachoroidal injection in human cadaveric eyes using μCT. Second, we aimed to develop suprachoroidal injectors optimized for nonhuman primate (NHP) resulting in reliable access to the SCS and to assess whether coverage was qualitatively similar to that measured in human cadaveric eyes using μCT. 

Methods

Quantification of Suprachoroidal Spread in a Human Cadaveric Model

Eye Preparation

Ten whole human eyes from five donors, not eligible for corneal transplantation, were procured from a partner eye bank (VisionGift, Portland, OR, USA). Eyes were stored refrigerated in Optisol or Life4C storage media to preserve tissue integrity until use. If the eye had become deflated during transport, the intraocular pressure was restored to at least 10 mm Hg by injecting balanced saline solution intravitreally with a 30-gauge needle and syringe. The globe was placed in a custom fixture that supported and fixed the eye in position by clamping the optic nerve. Additionally, this custom fixture was designed to support the TS-Micro Device and Integra Viaflo micropipette allowing a suprachoroidal injection to be automatically performed in the μCT chamber. Eyes were kept moist throughout testing with a balanced saline solution drip. 

Suprachoroidal Injection Procedure

The TS-Micro device was primed with contrast agent, Optiray 320 (Guerbet LLC, Princeton, NJ, USA) to ensure there was no air in the fluid line. The scleral conjunctival surface was dried with a cellulose spear in the desired injection quadrant. The marker of the TS-Micro device was aligned with the limbus of the eye and gently pressed into the sclera to mark the injection location, 3.5 mm posterior to the limbus. The needle of the TS-Micro device was inserted tangentially through the sclera into the suprachoroidal space (Fig. 1, Supplementary Video S1) and connected to the Viaflo (Integra Biosciences, Hudson, NH, USA) electronic micropipette for automatic injection. The eye, with the TS-Micro device and micropipette, was placed into the μCT chamber and the micropipette was set to automatically inject 100 µL of Optiray 320 at a rate of 8 µL/sec. 

Figure 1.

 

Schematic of SCS Injection Procedure. (1) Caliper marking. (2) Engage tissue with needle tip. (3) Pivot device anteriorly. (4) Slide needle into SCS. (5) Inject infusate.

The μCT Imaging and Analysis

Dynamic (four-dimensional: three dimensions over time) μCT was used to capture volumetric images of the eye and infusate throughout the SCI and up to 20 minutes after injection. Twenty minutes was chosen because preclinical data reports clearance of saline solution in the SCS to occur in 19 minutes.³ The μCT data was acquired using continuous imaging at 80 frames per second with an X-50 CT System (North Star Imaging Inc., Rogers, MN, USA). The μCT scans were reconstructed as discrete isotropic volumes with a voxel size of 139 µm. Each volume consisted of 4.5 seconds worth of continuous CT images. 

The volumetric data was aligned to the world coordinate system and prepared for quantitative measurement analysis of infusate SCS spread, thickness, distribution, and center of mass. Voxels that contained high density metals (TS-Micro device needle and the saline solution drip) were omitted from the measurement analysis. The unitless gray value of all remaining voxels were used to approximate the volume of contrast agent present in each individual voxel. Voxels with a similar gray value to the eye tissue were assumed to have no contrast agent, voxels that exceeded a threshold defined by a phantom scan were assumed to have 100% contrast agent, and voxels with gray values in the middle were assumed to be diluted contrast agent or contrast agent in a layer thinner than the voxel size (Fig. 2). 

Figure 2.

 

Dynamic CT Data - Voxel-Based Optiray Saturation.

The infusate SCS percent coverage was calculated by summing the voxel contrast agent volumes within bins defined by a spherical coordinate system with a fixed 2.5° interval for phi and theta. A calculation was performed to determine whether each bin contained more than 0.0015 µL of infusate; this cutoff value was chosen to ensure that scan noise did not contribute to the overall SCS percent coverage. All bins that contained more than 0.0015 µL of infusate were assigned a boolean value 1 indicating that infusate is present; all other bins were assigned a boolean value 0 indicating that no infusate is present. The partial surface area of an idealized sphere was calculated for each bin to correct for variable bin sizes, and bins surrounding the cornea were omitted from the calculation. Finally, the boolean values were multiplied by the partial surface area for each bin; the sum of infusate surface area was compared back to the total SCS surface area to define the SCS percent coverage for all critical datasets throughout the duration of the dynamic scan. 

The infusate center of mass was calculated directly from the voxel contrast agent volumes in a cylindrical coordinate system (Z-height and Theta-angle) for all critical datasets throughout the duration of the dynamic scan. A distribution table of infusate volume was generated for all critical datasets throughout the duration of the dynamic scan by summing the voxel contrast agent volumes within bins defined by a cylindrical coordinate system with a ∼1.5 mm interval for Z-height and 10° interval for Theta-angle. The equator of the eye was aligned with 0 mm Z-height and the injection location aligned with 0° Theta-angle for all trials. 

In addition, volumetric voxel-data was surfaced using an Advanced Surface Determination in VG Studio MAX (VolumeGraphics, Charlotte, NC, USA) to generate a polygonal model of the Optiray 320 contrast agent. The infusate thickness was calculated around the entire polygonal model for three datasets; one dataset represents the infusate thickness when ∼50% of the total infusate volume is present, the second dataset represents the infusate thickness when ∼100% of the total infusate volume is present, and the third dataset represents the last dataset captured in the dynamic scan. The infusate thickness results were displayed as a colormap. 

Unfortunately, μCT data for donor 2 OS was lost because of an acquisition file saving error. However, this eye was used to compare the impact of injection volume on suprachoroidal space coverage. An additional 200 µL of Optiray 320 was injected into the suprachoroidal space for a total injection volume of 300 µL, and a static volumetric scan was acquired and suprachoroidal space coverage was calculated. 

Development and Testing of Preclinical Suprachoroidal Injectors

Gyroscope Therapeutics (wholly owned by Novartis Pharma Corporation) developed a novel SCS injector, TS-Micro, designed to tangentially access the human SCS via a simple injection technique with one needle length. However, the size of the device and needle length was not appropriate for preclinical animal models because of the smaller eye diameter and thinner sclera of NHP.^12–14 To create an SCS injector for the smaller anatomy of NHPs, the eye contacting portion of the TS-Micro handpiece was scaled down to match the scleral curvature of the NHP eye, and the needle exposure was shortened. The proximal portion of the handpiece, fluid line, and syringe did not require adaptation (Fig. 3). 

Figure 3.

 

Comparison of clinical TS-Micro Device to those scaled for NHP injections.

NHP TS-Micro devices were piloted in vivo in adult cynomolgus monkeys. Briefly, 100 µL of fluorescein spiked balanced salt solution (BSS) or indocyanine green (ICG) was injected bilaterally into the suprachoroidal space of three adult cynomolgus monkeys during a nonsurvival surgery. Fluorescein spiked BSS injections were performed on one animal and both eyes were injected twice to confirm access. The remaining two animals received a single bilateral injection with ICG. Injections were considered successful if no back pressure was felt on the syringe during the injection, no efflux was observed from on or around the injection site, and no fluorescein or ICG was seen intravitreally or subretinally on wide field color fundus imaging. For ICG injections, confocal scanning laser ophthalmoscopy (cSLO) images were also acquired to visualize spread of the injection within the SCS. This spread was qualitatively compared to coverage previously gathered in human cadaveric eyes using anatomic landmarks as a reference. 

Based on this pilot analysis, the NHP TS-Micro was further used in a toxicity study where 11 NHP received bilateral 100 µL suprachoroidal injections. Similarly to the pilot study injections were considered successful if no back pressure was felt on the syringe during the injection, no efflux was observed from on or around the injection site, and no evidence of subretinal delivery/retinal tissue disturbance was visible on wide-field color fundus or cSLO imaging. 

All animals were treated in accordance with the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research. Studies conducted in accordance with all applicable sections of the Final Rules of the Animal Welfare Act regulations (Code of Federal Regulations, Title 9), the Public Health Service Policy on Humane Care and Use of Laboratory Animals from the Office of Laboratory Animal Welfare, and the Guide for the Care and Use of Laboratory Animals from the National Research Council. 

Results

Quantification of Suprachoroidal Spread in a Human Cadaveric Model

Donor information is summarized in the Table. Briefly, the five eye donors in this study were aged 71 ± 13 years (range 48–80 years), two female and three male. The average time from donor death to tissue testing was 100 ± 27 hours (range 67–125 hours). Eye dimensions were collected for all but one eye (Donor 2 OS) as dynamic imaging data was lost due to an acquisition file saving error. The average equatorial diameter and apical height for the remaining 9 eyes was 26 ± 1 mm (range 25–27) and 26 ± 1 mm (range 25–27), respectively. 

Table.

 

Donor Information, Ocular Dimensions, and Injection Coverage for All Trials

 

Suprachoroidal injections were successful in 8/10 eyes. The two unsuccessful injections were performed on eyes of the same donor (Donor 5) and the infusate pooled in the anterior portion of the suprachoroidal space, near the injection site. This pooling was likely due to choroidal tissue damage because the donor had a history of cerebrovascular small vessel disease, idiopathic hydrocephalus and whose cause of death was subdural hematoma from a fall. These two trials were excluded from further dynamic imaging analysis, as was the successful trial of Donor 2 OS because of the aforementioned μCT file-saving error. 

Injection testing results are summarized in the Table. The injection thickness, which is equal to the opening of the suprachoroidal space, ranged from 0.1–1.0 mm for all trials. The suprachoroidal coverage increased throughout the entire 20 minutes post injection start, but began to slow at approximately 9 minutes post injection (Fig. 3). The infusate first spread circumferentially away from the injection site, staying mainly in the anterior portion of the suprachoroidal space then spreading posteriorly (Supplementary Video S2). This can be visualized both via injection thickness imaging (Fig. 5) and via the motion of the center of mass of the injection (Fig. 6). When approximately half of the infusate was injected (∼50 µL) the average suprachoroidal space coverage was 11.2% ± 3.3%. Coverage increased to an average of 21.8% ± 6.9% when the full volume was injected (∼100 µL). The final coverage at 20 minutes after injection initiation was 33.3% ± 5.9% (Table). 

For the single trial of Donor 2 OS, a larger injection volume did increase the suprachoroidal coverage, but the percent coverage increased was not equal to the percent of increase in volume injected. Specifically, a 100 µL injection resulted in suprachoroidal space coverage of 32.1% approximately six minutes post injection on the contralateral eye and a 300 µL injection resulted in suprachoroidal space coverage of 48.3% (Fig. 7). 

Evaluation of Preclinical TS-Micro Devices

Delivery to the suprachoroidal space utilizing the preclinical TS-Micro devices was successful in both NHPs. All four NHP injections of 100 µL BSS + fluorescein were successful. There was no back pressure felt on the syringe, and no efflux was seen on or around the injection site. On fundus imaging there was no evidence of fluorescein in the vitreous or subretinal space (Fig. 8A). 

Similarly, all four NHP injections of 100 µL ICG were successful with no back pressure felt on the syringe and no efflux observed on or around the injection site. On fundus imaging there was no evidence of ICG in the vitreous or subretinal space. Additionally, ICG was easily visible on cSLO imaging and qualitatively had similar suprachoroidal coverage to that observed in our ex vivo cadaveric work (Figs. 8B, 8C) extending circumferentially and posteriorly from the site of injection reaching within 3 optic disc diameters from the fovea. 

Injection success in the subsequent toxicity study was similarly high to the pilot study. Twenty-one of the 22 injections successfully delivered the infusate to the suprachoroidal space (95% success rate), and suprachoroidal procedures were well tolerated by all animals. One injection delivered to the left eye of the first animal dosed was inadvertently a subretinal injection, as evidenced on postoperative imaging. This was likely due to animal movement during the procedure because this animal needed additional sedation. 

Discussion

In this study we successfully quantified infusate coverage of the suprachoroidal space when injected in human cadaveric eyes using μCT, developed a reliable suprachoroidal injector specific for NHP anatomy, and confirmed NHP SCS coverage was similar to that of our cadaveric work. 

All studies have limitations and ours is no exception. Although using human cadaveric tissue is beneficial as a result of clinical and anatomical relevancy, the tissue is not perfused, which affects fluid flow in the SCS and likely tissue mechanics and behavior.¹⁵ Additionally, the contrast medium Optiray 320 used in this study has a viscosity of 9.9 cps at 25°C, which is more viscous than aqueous-based injections and decreases spread in the SCS.³ However, both the lack of perfusion and higher viscosity of infusate decrease SCS coverage, bounding our results as a worst case. Additionally, although the NHP TS-Micro device was easy to use, had a high delivery success rate, and based on in vivo imaging provided similar SCS coverage to that seen in cadaveric tissue, this data was qualitative, limited to imaging shortly after injection, and based on a small sample size. To fully characterize the NHP SCS spread in vivo, a larger study with post-injection imaging for at least 20 minutes and histologic sectioning would be needed. 

Even with these limitations, this work provides valuable insights for drug delivery to the suprachoroidal space. The circumferential then posterior spread aligns with previous studies in porcine, NHP, and rabbits but does not immediately fill the space as some studies report.^3–5,16 Rather it spreads circumferentially away from the injection site in the first three to five minutes, rapidly increasing the SCS coverage. The rate of SCS coverage begins to slow five to nine minutes after injection as the fluid continues to move posteriorly. This slower coverage of the SCS could be due to the differences in viscosity and perfusion but is more likely due to the volume of the injection in comparison to the size and anatomy of the injected eye. For example, in rodents the lens comprises much of the anterior of the eye, pushing the limbus and the start of the SCS to almost the equator of the eye.⁷ This decreases the total surface area of the SCS and allows the injection to be much more posterior than in humans, filling the posterior space more quickly. 

One may argue that simply increasing the injection volume would counteract this and increase the coverage of the SCS. However, we found that increasing the injection volume led to increased SCS coverage, but not in a 1:1 ratio, because tripling of the injection volume only resulted in roughly a 50% increase in SCS coverage. 

Similarly, we found that the distribution of a 100 µL injection to be anatomically specific. Interestingly, contralateral eyes from the same donor had similar SCS coverage (Fig. 4) and similar injection distribution as seen via the final border and motion of the center of volume of the injection (Fig. 6). However, SCS coverage varied much more across donors. To our knowledge we are the first to observe and report on such a finding. The same trend occurred in the cSLO imaging of ICG injections in NHP (Fig. 8) where ICG appeared to be influenced by the choroidal vasculature of the eye. This donor-to-donor variability infusate spread is important to keep in mind when evaluating drug safety and efficacy in clinical trials, particularly if a specific area of the posterior eye is the target. 

Figure 4.

 

SCS coverage of 100 µL injection over time.

Figure 5.

 

Injection thickness of Donor 1 OD during injection at injection volume of ∼50 µL, ∼100 µL, and final spread 20 minutes after injection.

Figure 6.

 

Average infusate distribution across eye for all trials at ∼20 minutes. The equator of the eye is located at a height of 0 mm where positive values indicate locations anterior to the equator and negative values indicate locations posterior to the equator. The injection site was consistent for all trials and is located at a circumferential angle of 0°. Darker blue indicates more volume. Solid lines indicate infusate boundary of each trial. Dotted lines indicate movement of infusate center of volume away from the injection location halfway through injection (∼50 µL), at full injection (100 µL), and ∼20 minutes after injection start.

Figure 7.

 

SCS coverage of 100 µL and 300 µL injections in contralateral eyes of same donor.

Figure 8.

 

RetCam fundus images of right and left NHP eyes from one animal 10 minutes after 100 µL BSS + fluorescein suprachoroidal injections showing no evidence of intravitreal or subretinal injection (A). Bilateral 100 µL ICG injections visualized via cSLO IRAF (colorized), overlayed on cSLO IR imaging taken five minutes after injection confirmed SCS delivery with coverage consistent with human cadaveric testing (B). cSLO IRAF imaging on same animal as inset B, but with field of view centered toward periphery at injection quadrant (C).

Appropriately evaluating such variability is also important in preclinical studies assessing tissue toxicity, dosing, and early drug candidate selection. The NHP TS-Micro devices developed in this study allowed for easy and reliable access to the SCS. Having the ability to use a preclinical device that is essentially identical to the clinical device significantly derisks and speeds drug development. Additionally, it can provide valuable preclinical data for combination products, which is being increasingly requested by regulatory bodies. 

In summary, we quantified the SCS coverage of a 100 µL injection in cadaveric eyes using μCT and found there to be more interdonor variability in the SCS coverage than intradonor variability. Additionally we developed a suprachoroidal injector customized for the anatomy of the NHP eye, which reliably delivers infusate to the SCS with a coverage comparable to that observed in human cadaver tissue and is similarly dependent on tissue anatomy. These findings provide insight to the importance of thoroughly evaluating and characterizing the ability of a device to correctly deliver to the tissue of interest which in turn allows for better assessment of drug candidates, dosing, and toxicity, speeding drug development. 

Acknowledgments

Disclosure: M. Hedgeland, Novartis (E, P); S. Camacho Gonzalez, Novartis (E, P); D. Stampoulis, Novartis (E); V. Lee, Novartis (E); K. Powell, Novartis (E); Y. Qiu, Novartis (E); R. Hodge, Novartis (F); G. Bell, Novartis (F); A. Fago, Novartis (F); R. Atkinson-Dell, Novartis (F); N. Choti, Novartis (E); K. Stoner, Novartis (E, P)