In Vitro Biocompatibility of Silicone Oil Siluron Xtra® on Porcine Retina in a Perfusion Culture System

Purpose: To examine the biocompatibility of Siluron Xtra® (Fluoron, Ulm, Germany), a silicone oil used as a vitreous substitute in vitreo-retinal surgery on porcine retinal tissue in vitro. Methods: Retinal tissue of 37 porcine eyes was immediately prepared post mortem and placed into Minucell perfusion systems (Minucell, Bad Abbach, Germany) to be perfused with liquid culture medium over a period of 4-8 days. 23 retinal tissue samples including retinal pigment epithelium (RPE) were covered by silicone oil (group 1) during the perfusion and compared to a control group of 7 samples (group 2) without silicone oil. Group 3 included 7 single RPE tissues without retinal tissue layer in direct contact with silicone oil, thus imitating a retinal tear, during the perfusion. The morphology of the retina and RPE was examined by light microscopy and stained with immunohistochemical markers to determine Müller cell damage in the retina with Glial fibrillary acidic protein (GFAP) and proliferation in the RPE with Ki67. Results: Ki67 staining showed significantly less proliferation in the tissue covered by silicone oil (group 1) compared to the control samples in group 2 (p=0.001). Direct contact of silicone oil and RPE (group 3) showed no significant increase in proliferation compared to the controls (p=1). GFAP staining also did not show any significant Müller cell damage related to Siluron Xtra® (p=0.9). No structural changes in the retinal tissue were observed related to silicone oil by HE staining. Conclusion: The results of our in vitro examination verified good structural biocompatibility of silicone oil (Siluron Xtra®) on porcine retina and RPE in vitro. Furthermore, silicone oil may exert a protective layer preventing proliferation on retinal tissue. Further examination of other silicone oils and alternative vitreous substitutes, such as gas and water are necessary to determine the protective advantage of Siluron Xtra® on retinal tissue proliferation.


Introduction
Following their invention in the 60 ties by Paul A.Cibis, silicone oils have been used routinely as a substitute to replace vitreous humour post vitrectomy surgery [1,2]. While most commonly gas is used as an endotamponade during standard vitrectomy surgery, silicone oils have shown good post-operative results in the treatment of persistent macular holes [3] and especially in more complicated procedures such as giant retinal tear [4] and patients with retinal detachment due to proliferative diabetic retinopathy [5] and proliferative vitreopathy [6]. Usually, air or expanding gases are used as an endotamponade for macular holes in vitrectomy surgery [7][8][9].
Porcine retina is often used as a model for in vivo and in vitro experimentation because of its similar micro and macroscopic features to the human retina. Previous studies have used porcine eyes to test new endotamponades [10] and autologeous RPE-Choroid grafts [11] that had not been approved for use in humans.
This in vitro examination was dedicated to examine the biocompatibility of the silicone oil Siluron Xtra © (Fluoron, Ulm, Germany) on porcine retina. To current date, the interaction of porcine retinal tissue and silicone oil has never before been examined histologically in an in vitro experiment although clinical use of silicone oil has shown good long term results in patients [12]. Purpose of the examination was to verify positive results shown in patients undergoing vitrectomy surgery using silicone oil tamponades with objective histological findings.
The perfusion system used in our investigation (Minucells and Minutissue, Bad Abbach, Germany) has previously been used for various organ cultures, including embryonic retinal epithelial cells [13,14], RPE [15] and tissue cultivation [16]. The pump driven system enables a continuous supply of nutritious culture medium and oxygen to the specimens whilst simultaneously deporting waste materials and carbon dioxide out of the perfusion chambers. Hereby imitating the in vivo blood supply to the retinal tissue. Previous work has shown good preservation of porcine retina and RPE using this in vitro perfusion model before [17].
Immunohistochemical markers GFAP and Ki67 were used to identify destruction in muller cells and retinal proliferation in retinal tissue.
Glial fibrillary acidic protein (GFAP) is a protein that is expressed in the central nervous system as well as in healthy ganglion cells of the retina. The protein has shown to play a vital part during repair mechanisms in the central nervous system [18]. Increased vitreous GFAP expression has also recently been observed in correlation to retinal gliosis [19]. Healthy retinal tissue does not express GFAP in its muller cells tissue layer. Thus a correlation is suggested between GFAP expression and damaged muller cell tissue in the retina [20,21].
The Ki-67 protein is an antigen expressed in proliferating cells [22] during all active phases of replication (G1, S, G2 and mitosis) and thus is suitable as a marker to detect proliferation in the retinal tissue [23].

Tissue preparation
Fresh porcine eyes were collected from a local abattoir and transported to our laboratory suspended in 0.9% NaCl solution. Assisted by a surgical microscope (Zeiss Universal S3, Jena, Germany) the globes were dissected under sterile conditions within 1 h post mortem. After removal of remaining muscle tissue, the anterior section, including cornea, iris, lens and vitreous, was removed by a circular incision into the pars plana region (Figure 1.1). The empty globes were carefully halved and the larger half pinned down through the sclera onto a styrofoam surface for fixation (Figure 1.2). Using a forceps and a microsurgical knife, the retinal tissue including RPE was slowly separated from the sclera without rupturing the fragile tissue with the sharp instruments. Both tissues were clamped between a double ring tissue carrier with an outer diameter of 13 mm and an inner diameter of 9 mm (Minucells and Minutissue, Bad Abbach, Germany). The retinal tissue and RPE were carefully lifted off the sclera and placed onto the smaller white ring, covering the inner hole completely ( Figure  1.3). The slightly larger black ring was placed on top and firmly pressed downward to enclose the tissue inside the double ring carriers ( Figure  1.4).

Perfusion culture
The tissue carriers were placed into double compartmented perfusion containers by Minucell containing 3 consecutive spaces per compartment with the retinal tissue facing upwards.
Each compartment was connected to two gas-permeable silicone tubes with a 1 mm inner diameter supplying the chambers with medium from a sterile supply of Dulbecco's modified Eagle's medium (DMEM, Sigma-Aldrich, Seelze, Germany), supplemented with 15% porcine serum, 2.5% Hepes buffer solution and 1% penicillin/ streptomycin (Gibco Life Technologies, Eggenstein, Germany). Two additional silicone tubes for each compartment transferred the consumed medium out of the containers and into empty bottles.
The culture medium was kept cool using ice packets and pumped through the silicone tubes and into the perfusion containers at a speed of 1 ml/h using a peristaltic pump (IPC N8, Ismatec, Wertheim, Germany).
The containers were kept on a warming plate (Medax, Kiel, Germany) at a constant temperature of 37°C. The retinal tissues enclosed in the sealed chambers were perfused for 24 h over a period of 4-7 days, in a room kept under a physiological light/dark rhythm.
The perfusion system was checked 3 times per day to ensure a steady flow of medium running through the perfusion chambers.
Overall 37 specimens were randomly selected into 3 different groups (Table 1). 23 retinal tissues attached to RPE were coated with a drop of silicone oil Siluron Xtra © in group 1. 7 control samples (group 2) were perfused without silicone oil. An additional set of 7 specimens containing only RPE tissue was also coated with silicone oil during the perfusion (group 3). Thus imitating a retinal tear situation in which RPE would be in direct contact with silicone oil (Figures 2 and 3).

Histology
Following the perfusion the tissue samples were carefully removed from the double ring carriers and fixated with paraformaldehyde 4% in phosphate buffered saline (PBS, Sigma-Aldrich) over night. The samples were washed in PBS 0.1 M, dehydrated in ascending alcohol and embedded in paraffin. Xylol was used for deparaffinisation and rehydration.
Finally, the embedded tissues were cut into 2-5 μm slices and placed onto slides (Superfrost Plus, Menzel, Braunschweig, Germany). A sample of each specimen was stained with haematoxylin-eosin and examined using a light microscope (Axiotech, Zeiss). The visible results at a 10X, 20X and 40X magnifications were documented by digital camera (Canon, Tokio Japan) and transferred to the computer.

Immunohistochemistry
All samples were stained using GFAP and Ki-67 markers. Immunohistochemical markers GFAP and Ki-67 were used to examine muller cell damage and proliferation within the specimens further. Therefore the paraffin embedded samples were de-paraffinised and rehydrated. Incubation in H 2 O 2 3% was used to disable endogenous peroxide in the tissue section. The samples were then incubated by specific primary and secondary antibodies and horseradish peroxidase conjugation (HRP) consecutively. Chromogen solution (AEC) was added to the specimens in the final step, causing an enzymatic reaction with peroxide. AEC caused visible red colored staining at primary antibody binding sites as can be seen in Figures 4 and 5.

Statistical analyses
SPSS © version 22.0 for Windows (IBM Corporation, Armonk, NY, USA) was used for statistical analyses. The non-parametric Mann-Whitney U Test was used to determine the correlation between positive GFAP stained muller cells and silicone oil coverage. Mann-Whitney U Test was also used to compare the number of proliferative sites visible through Ki67 staining and silicone oil coverage. P values of less than 0.05 were considered statistically significant. The results if the statistical analysis can be seen in Table 2 for GFAP and Table 3

Tissue preparation and perfusion
The perfusion was tolerated well by all 38 specimens, showing no visible damage to the tissues upon removal from the perfusion chambers after duration of 4-8 days. Silicone oil was visible on all specimens, thus ensuring total silicone oil coverage of the samples throughout the entire perfusion. 8 retinal tissues in group 1 and 2 were irreparably damaged during the preparation of the paraffin cuts post perfusion. However, all RPE samples stayed intact and could be used for further evaluation.

Structural evaluation in HE staining
All test groups showed intact structure of retinal and RPE tissue in HE staining at 10X, 20X and 40X magnification using a light microscope. There was no silicone oil imprinting visible in any of the samples. Figure 6 pictures the intact retinal structures within all retinal layers in HE staining of a sample taken from group 1 post perfusion.

Immunohistochemistry
GFAP was used to identify damage in muller cells. Only the samples containing retinal tissue could be used, as muller cells are only found in the retinal tissue. Therefore, group 3 was excluded from GFAP staining and evaluation. Damage to the muller cells was unspecific to silicone oil perfusion with 71% of the specimens in the silicone oil coated group showing positive muller cell staining versus 80% in the control group respectively (p=0.9). The results of the GFAP staining are pictured in Figure 7 showing that there was no correlation between GFAP expression and silicone oil coverage of retinal tissue.   Ki67 was used as a marker for proliferation. All samples were photographed in 20X magnification and the number of proliferation sites was counted on the computer using a grid of evenly distributed quadrants. Overall only 33% of the retinal samples coated with silicone oil showed positive Ki67 staining at all, whereas the control groups without silicone oil coverage showed significantly higher proliferation with Ki67 staining in 87% of the samples (p=0.001). Figure 8 pictures the significantly lower expression of Ki67 in retinal tissue covered by silicone oil during the perfusion.
RPE in direct contact with silicone oil showed 87% proliferation as well (p=0.9). Figure 9 pictures that the expression of Ki67 in RPE covered by silicone oil is equal to the expression of Ki67 in retina The length of post mortem time, caused by transport and preparation of the specimens, the number of days of perfusion and relative oxygen and nutrient supply, depending on the position of specimens in the perfusion chambers were recorded for each specimen. However, none of the variants showed significant correlation to the tissue damage evaluated with GFAP and Ki67 immunohistochemistry.
GFAP enhancement, confirming structural tissue impairment could be identified in both groups 1 and 2 and could not be correlated to contact with silicone oil.

Air bubbles in the perfusion chamber
Occasionally air bubbles occurred in the silicone tubes and travelled into the chambers, compressing the tissue samples and restricting medium supply to the specimens. Most of the bubbles appeared only temporarily and eventually dissolved on their own accord. However, in 4 of the overall 37 samples that were perfused it was not possible to get rid of larger air bubbles. Nonetheless, there seemed to be no impairment to the specimens that were tamponaded by air bubbles for a period of time extending up to 48 h during the perfusion.

Clinical results of silicone oil endotamponades in pars plana vitrectomy
Silicone oils have been used as ocular endotamponades in ophthalmological surgery since their introduction in the 60 ties by Paul A. Cibis [1,2].
As verified in previous studies, the use of silicone oil has been successfully applied in the treatment of difficult retinal detachment, showing good anatomical results, improved vision and reduced metamorphopsia in patients [3,5,6].
In the post-surgical stage it has been shown that there are reversible effects to the biomechanics of the anterior chamber associated with silicone oil. These may result in an increase of anterior chamber depth and thickening of the central cornea. Although, these effects are most likely linked to the fact that silicone oil injected eyes are more complicated cases in the first place, the surgery itself or changes in intraocular pressure [24,25].
In 2014 Caramoy, A. et al. published a study comparing retinal tissue in healthy and silicone oil tamponaded eyes. Using SD-OCT (spectral domain optical coherence imaging) and ImageJ software they determined thinning of the inner retinal layers, especially affecting ganglion cells and inner plexiform layer, that were associated with silicone oil use [26].
Whilst silicone oils have never before been tested on porcine models in an in vitro perfusion system, in vivo injection of silicone oils in rabbits post pars plana vitrectomy has been performed in current literature. ELISA and immunohistochemical staining of hypoxia inducing factor-1-alpha (HIF-1α) and growth factor VEGF (vascular endothelial growth factor) did not show any pathological vascular or hypoxic signs in the retina [26]. In addition no retinal toxicity has been detected in silicone oil tamponaded eyes in past clinical studies [27].
However, unlike gaseous and air filled endotamponades, silicone oil needs to be removed from the patients eye within several months in a second surgical procedure. This is necessary to avoid silicone oil emulsification and toxic effects on the retina [28][29][30]. Although showing good anatomical results, the removal of silicone oil has been associated with several complications including cataracts in phakic eyes, keratopathy, hypotonia and most frequently, re-detachment of the retina [28,31].
indicated that silicone oil may exhibit protective, anti-proliferative effect on the retinal tissue. Silicone oil coating of the RPE did not reduce proliferation in the RPE compared to the control group that was protected by retinal tissue; however, no evidence of increased proliferation in the RPE related to silicon oil could be determined either. This makes silicone oil a very tolerable agent in the treatment of retinal holes, where tamponade oil can leak through the retinal fissure and onto the RPE. In 1992 Heidenkummer et al. investigated proliferative activity in epiretinal membranes using Ki67 too. Here, highly proliferative activities were observed in 6 of 11 recurrent membranes after intraocular silicone oil tamponade [32].
In the same year, Shikishima et al. investigated the effects of intravitreally injected silicone oil in rabbit retina and detected silicone oil in the inner retinal layer and in phagocytes in the vitreous cavity as well as prominent subretinal and epiretinal proliferation with abundant phagocytes containing silicone oil particles. They hereby suggested that silicone oil may be the cause of excessive subretinal and epiretinal proliferation [33].
Both studies contradict our current findings that have in fact shown decreased proliferation in porcine retina post silicone oil exposure.

Muller cell affection/GFAP expression
GFAP enhancement in the muller cells could not be correlated to silicon oil, suggesting that it must have been caused by other factors. A previous study examining healthy, grown up, human eyes embedded in 2% PFA (paraformaldehyde) determined that GFAP expression is not correlated to the time of preservation in PFA. 4% PFA was used as an agent to fixate our specimens. The study showed that GFAP enhancement significantly rose with prolonged post mortem time exceeding 30 h [34].
Post mortem decay during transportation from the abattoir to the lab and preparation of the tissue before the perfusion with culture medium varied between 1-4 h in our examination but may have contributed to the enhanced GFAP expression in our specimens.
Although it was impossible to identify structural damage associated to post mortem decay using light microscopy, this does not rule out retinal damage on an ultramicroscopic level.

Air bubbles in the perfusion
The air bubbles that occured in our perfusion system during the experiment did not have an effect on the specimens. In fact, air is successfully being used clinically on patients as an intravitreal tamponade after retinal detachment [35] and pars plana vitrectomy and has shown an increased closure rate of macular holes [36]. Some findings have considered air to be equally effective as a intravitreal tamponade whilst even reducing complications compared to the most commonly used gas tamponade [7,37].
Post-operative findings have shown that air bubbles in the human eye tend to dissolve within 7-10 days [35] showing no clinical damage whereas none of our air bubbles even lasted for more than 48 h.
Both air and silicone oil have shown good properties as an intravitreal tamponade for retinal detachment in previous studies [35,36].
In terms of our examination we can conclude that for a period of 48 h porcine retina can survive without showing signs of structural damage if tamponaded by air.

Conclusion
In conclusion, silicone oil Siluron Xtra © has shown good histological biocompatability for porcine retina and retinal pigment epithelium. In fact, protective features, decreasing proliferation in retinal tissue could potentially be linked to Siluron Xtra © .
Siluron Xtra © has also shown good biocompatibility towards retinal pigment epithelium. In case of post-operative retinal tear or insufficient closure of retinal tear, this suggests that silicone oil does not exhibit damaging effects on retinal tissue.
The question which arises in this context is whether the antiproliferative properties of silicone oils are possibly superior to other retinal tamponades such as water or gas. Thus, further investigation of the biocompatibility of other silicone oils and alternative retinal tamponades is of interest, to determine a significant protective advantage of Siluron Xtra © and other silicone oils on retinal tissue.