Effect of silicone oil on peripapillary capillary density in patients with rhegmatogenous retinal detachment

Silicone oil (SO) tamponade, widely used in vitreoretinal surgery, is known to be associated with various complications including cataract, keratopathy, glaucoma, and optic neuropathy [1‐3]. Optic neuropathy following SO tamponade, subclinical or visually significant, is reported to occur at an incidence rate varying from 0 to 28% [2, 4]. The pathogenesis of SO related optic neuropathy remains unclear. Possible mechanisms may include high intraocular pressure (IOP) [2] and migration of microdroplets of SO into optic nerve [3, 5]. In patients who underwent vitrectomy and SO tamponade for rhegmatogenous retinal detachment (RRD) repair, optic coherence tomography (OCT) revealed retinal thickness changes, particularly in the retinal nerve fiber layers (RNFL) [6‐10]. Despite the extensive studies on structural changes in optic nerve, the effect of SO on blood flow around optic nerve has not yet been studied.

Radial peripapillary capillaries (RPC), the innermost layer of capillaries that runs parallel to peripapillary RNFL, function to nourish RNFL in its distribution around the optic nerve head. With the advent of new imaging modality like OCT angiography, in vivo imaging of RPC becomes feasible [11], and quantitative evaluation of RPC is also available. Reduced or abnormal blood flow in RPC has been described in glaucoma and other optic nerve diseases [12‐14].

Among the various conditions which need vitreoretinal surgery and SO tamponade, RRD serves as an ideal model to study the effect of SO on RPC, because the optic nerve head at presentation can be absence of intrinsic vascular abnormalities. Also, during the vitreoretinal surgery and SO removal surgery, internal limiting membrane peeling is not always necessary, therefore the likelihood of developing disassociated optic nerve fiber layer appearance [15, 16] or other retinal structural changes [17] is minimized. We thereby enrolled a prospective cohort of patients who underwent successful vitrectomy for RRD repair and subsequent SO removal, and measured RPC vessel density (VD) with SO and after SO removal. We hypothesized that SO filled eyes have transient decreased blood flow in RPC which will recover after SO removal. The purpose of the study was to evaluate the effect of SO on peripapillary vessel density.

This prospective, consecutive cohort study was conducted at Ophthalmology Department, Peking Union Medical College Hospital (PUMCH). This study adhered to the tenets of the Declaration of Helsinki. Ethics committee of PUMCH approved the study protocol. All participants provided written informed consent.

To our knowledge, the current study is the first one that investigated the effect of SO on peripapillary blood flow. We found a mild but definite increase of RPC VD in SO filled eyes following SO removal. Our results suggested that SO may have potential negative effect on microvascular blood flow around optic nerve head.

The increase in RPC VD reflected a mild but true increase in microvascular blood flow around optic nerve head after SO removal. The repeatability of OCT angiography measurements had been well demonstrated in previous studies [19, 20]. Factors that could influence the accuracy of OCT angiography measurements included IOP fluctuation [21], refractive media opacity and insufficient signal strength [20, 22, 23], diurnal variations and blood pressure changes [24]. In our study, we made several efforts to limit confounding factors. First, with healthy fellow eyes serving as controls, the impact of systemic errors, such as diurnal variations and systemic circulatory changes, could be minimized. Second, the main outcome measures were acquired at least 3 months after primary vitrectomy and 3 months after SO removal, which was a long enough period for acute surgical inflammation and IOP fluctuations to subside. Thus, the resulting RPC VD measurements were unlikely to be affected by postoperative corneal edema, anterior chamber flares, hypotonic papilledema, or potentially other unusual conditions. Third, we excluded subjects with severe cataract before SO removal and included only patients with adequate OCT angiography image quality. Also, the sensitivity analysis in patients who did not have cataract surgery during SO removal detected similar increase in RPC VD. Therefore, the increase in peripapillary vessel density was unlikely to be confounded by improvement in optic media clarity due to cataract removal. Fourth, although OCT angiography scan quality index exhibited mild increase after SO removal, its changes were adjusted as covariates when comparing RPC VD changes between affected eyes and controls. Fifth, trend analysis revealed that RPC VD after SO removal manifested a steady increase rather than an irregular fluctuation, supporting the reliability of our results. One might argue that the recovery of peripapillary vessel density could reflect a continuous improvement of optic nerve microcirculation after retinal reattachment. However, peripapillary vessel density increase was not observed from 1 month to 3 months after primary vitrectomy surgery (46.7 ± 4.0% versus 46.8 ± 3.7%, P = 0.92, paired t test), suggesting that the increase of RPC VD after SO removal was not part of a prolonged recovery process from retinal reattachment. One might also argue that SO tamponade and refractive changes after SO removal will affect image magnification which will further influence RPC VD measurements. By comparing the disk diameter on en-face OCT images before and after SO removal (1.518 ± 0.226 mm and 1.518 ± 0.229 mm, P = 1.00), we did not detect significant difference in image magnification with and without SO tamponade. Therefore, SO tamponade and refractive changes are unlikely to significantly influence RPC VD measurements. Others might argue that OCT angiography images under SO could be dark because of masking artifact. However, high quality images acquisition could be achieved in SO filled eyes when a suitable angle of scan could be found to avoid silicone oil reflection. Based on the above considerations, the increase in RPC VD was likely to be secondary to SO removal instead of other confounders. Although the amount of RPC VD increase was small, we are confident that a real improvement of peripapillary capillary density occurred after the removal of SO.

We tried to examine global RPC VD change after SO removal for its association with demographic and clinical characteristics including duration of SO tamponade and full SO tamponade, but none of these factors was associated with RPC VD change. Several reasons might account for the negative findings. First, the judgement of full SO tamponade relied on not only the quantity of SO in eye, but also the skill of examiner and the diameter of dilated pupil. Second, duration of SO tamponade was 3 to 4 months according to our institute’s common practice. If SO tamponade was kept for a longer duration, we might be able to detect a positive correlation between RPC VD change and SO tamponade duration. According to previous studies, longer duration of SO tamponade might be related with retinal structural abnormalities [25]. Third, the small sample size might be insufficient to yield significant results in correlation analysis.

In the current study, we did not evaluate RPC VD changes before and after primary vitrectomy and SO tamponade. Several difficulties existed in such investigation. First, most RRD patients at presentation had low vision, poor fixation, and considerable vitreous opacity which reduced the quality OCT angiography images. In some patients, the peripapillary retina was even detached, resulting in segmentation errors in OCT angiography images which could hardly be fully adjusted. Therefore, the accurate evaluation of RPC VD before primary vitrectomy and SO tamponade was difficult. Second, to evaluate the solitary effect of SO tamponade on peripapillary vessel density and eliminate the confounding effect from surgical trauma of primary vitrectomy, a group of controls who underwent vitrectomy without SO tamponade would be needed. However, legal SF6 or C3F8 product for medical use was unavailable across China until September 2019 when we end the current study. Considering the above difficulties, we studied the effect of SO on peripapillary vessel density by comparing RPC VD before and after SO removal, based on the assumption that SO removal without retinal manipulations caused limited surgical trauma and minimal pathophysiological changes in the posterior segment of eye. Previous studies had compared the effect of different types of tamponade on retinal structures using OCT. A case series by Christensen et al. [7] performed OCT scans in RRD patients and detected inner retinal thinning in SO filled eyes compared with gas filled eyes. On the contrary, a case control study by Bansal et al. [26] suggested that optic neuropathy after vitrectomy was related with intraoperative systemic hypotension and low ocular perfusion, but not related with the type of tamponade. Further studies are needed to compare peripapillary vessel density changes in eyes with SO and gas tamponade.

The strengths of this study included prospective design, longitudinal follow-up, and high-quality data acquisition. We acknowledge several limitations in our study. First, we excluded patients with high myopia whose peripapillary microvasculatures might react differently to the use of SO. Second, our findings should be limited to RRD. We should be careful about generalizing to other vitreoretinal diseases. Third, the duration of SO tamponade was short, and the effect of long-term SO tamponade on peripapillary vessel density remained unknown. Fourth, we did not compare the effect of SO and gas tamponade on RPC VD changes, as was discussed above. Future studies should further examine this issue. Fifth, we only identified subclinical increase in peripapillary vessel density after SO removal, and the increase was not analyzed for its visual correlation.

In patients who underwent vitrectomy and SO tamponade for RRD repair, we detected a mild increase in peripapillary capillary vessel density after SO removal, mainly in the superior hemifield of peripapillary area. Our findings suggested that SO removal was associated with a recovery of peripapillary microvascular blood flow, possibly due to the release of vascular compression. Our results implied that the use of SO should be held with caution due to its potential negative impact on optic nerve microcirculation, even when the duration of SO tamponade does not exceed four months. Our study also added new insights to the complex biologic mechanisms of SO related optic neuropathy.