Comparison of fundus changes following silicone oil and sterilized air tamponade for macular-on retinal detachment patients

This retrospective, single-center study included 21 patients of the inpatients Department of Ophthalmology, Shanghai General Hospital Affiliated to Shanghai Jiao Tong University, from October 2017 to March 2019. This study adhered to the tenets of the Declaration of Helsinki and was approved by the Ethics Committee of Shanghai General Hospital affiliated to Shanghai Jiao Tong University. All patients suffered from primary RRD without macular involvement (Fig. 1). Exclusion criteria included macular involvement in detachment, proliferative vitreoretinopathy over B level, previous history of intraocular surgery, sign of epiretinal membrane, macular hole, macular degeneration or other macular disorders, high myopia (axial length>26 mm), secondary glaucoma, lens turbidity or massive vitreous hemorrhage, persistent subretinal fluid and intraretinal cystic space. Patients with diabetes mellitus, poorly controlled hypertension or other severe systemic vascular diseases were also excluded.

Standard three-port 23-gauge PPV was performed using the Alcon Constellation system (Alcon Laboratories, Inc., Fort Worth, TX, USA) under retrobulbar anesthesia by the same surgeon (Dr. Haiyun Liu). Preoperative macular status, extent of RRD (number of quadrants) and number of retinal tears were recorded before surgery, and double confirmed during the procedure. Before surgery, adequate doctor-patient communications were carried with each of our patients. The choice for type of tamponade was first discussed during the conversation, mainly based on intraocular pathological condition, and was finally decided by the surgeon during the operation. Intravitreal tamponade was used after drainage of sub-retinal fluid during air-fluid exchange. 7 patients with intravitreal SO tamponade were defined as Silicone Oil (SO) group, and 14 patients with intravitreal sterilized air tamponade were defined as Gas group. Usually, sterilized air could be spontaneously absorbed in 2 weeks. Upon the first postoperative follow-up visit (at 2 weeks), patients of the Gas group actually had no remaining intravitreal air tamponade. But for ease of expression, we denominated the ‘Gas/SO group’ and ‘Gas/SO tamponade’ in the present study, based on intravitreal tamponade choices during surgery. No patient underwent combined cataract surgery during the procedure. SO removal was performed around 4 months after retinal reattachment surgery. No patient suffered from recurrent retinal detachment.

OCT-angiogram imaging was performed using the RTVue XR Avanti device (AngioVue software, version 2017.1.0.155; Optovue Inc., Fremont California, USA) with the Angio Retina mode (6 × 6 mm, blood flow density) by the same operator (H, Zhou). The technique including the SSADA method has been previously described [26, 27]. The software automatically segmented the vasculature into three layers: superficial capillary plexus (SCP), deep capillary plexus (DCP), and choriocapillaris plexus (CCP) (Fig. 2). The segmentation of the capillary plexus was checked before any data measurement was recorded. An experienced doctor (S, Zhang) and technician (H, Zhou) independently reviewed the images and occasionally corrected the segmentation lines when necessary. For each layer, flow density (FD) was calculated separately in three annulus regions: fovea (1 mm), parafovea (1–3 mm) and perifovea (3-6 mm) according to the ETDRS grids. The images with poor image quality were excluded based on one or more of the following criteria: low signal strength index (< 60), presence of blink artifacts, poor fixation leading to motion or doubling artifacts, and media opacities obscuring the view of the vasculature.

With the automatic segmentation of retinal layers by RTVue XR Avanti device, we got seven sets of retinal layer thickness data for analyzation in the present study: NFL (nerve fiber layer), GCL + IPL (ganglion cell layer and inner plexiform layer), INL (inner nuclei layer), OPL (outer plexiform layer), ONL + IS (outer nuclei layer and inner segment), OS+RPE (outer segment and retinal pigment epithelium) and BRM (Brunch’s membrane) (Fig. 3). Choroidal thickness was obtained with the automatic built-in software of the SS-OCT device (Topcon FastMap, version 10.13.003.06, Topcon Medical Systems). All patients underwent three times of postoperative examinations at 2 weeks, 6 weeks and 12 weeks after surgery.

From 2 weeks to 6 weeks, we found opposite changing trends of SCPFD in Gas group and SO group, which led to significant intergroup differences (Table 2). SO tamponade led to more decrease in SCPFD, while the blood flow had a slight increase in Gas tamponade eyes. Similarly, foveal DCPFD also had opposite changing trends between Gas group and SO group (Gas: 1.28 ± 3.02, SO: − 2.62 ± 4.92, p = 0.036, Table 2). Furthermore, parafoveal CCPFD showed more increase in SO tamponade eyes which also led to significant intergroup difference (Gas: − 0.21 ± 1.55, SO: 3.35 ± 5.27, p = 0.028, Table 2).

From 6 weeks to 12 weeks, we found decreased foveal SCPFD in both SO and Gas tamponade eyes. SO tamponade eyes had more decrease in blood flow which led to significant intergroup difference (Gas: − 1.03 ± 1.12, SO: − 6.28 ± 6.39, p = 0.006, Table 2). Compared to gas tamponade, SO tamponade also resulted in more decrease in parafoveal DCPFD, which led to significant difference (Gas: 1.28 ± 1.26, SO: − 0.62 ± 2.78, p = 0.042, Table 2).

No significant result was found in other analysis of retinal or choroidal capillary plexus flow density changes during observation.

From 2 weeks to 6 weeks, NFL thickness was found to decrease in both Gas tamponade and SO tamponade eyes (Table 3). SO tamponade resulted in more pronounced decrease which led to significant intergroup differences. Opposite changing trends were found in GCL + IPL and ONL + IS thicknesses due to different means of tamponade. SO tamponade led to more decrease in these two segmented layers, which also led to significant intergroup differences. Foveal INL thickness had more decrease in SO tamponade eyes as well (p = 0.006, Table 3).

From 6 weeks to 12 weeks, SO tamponade led to more decrease in INL and OPL thicknesses, while gas tamponade led to a slight increase, which led to significant intergroup difference (Table 3). Foveal ONL + IS thickness also showed opposite changing trends due to different tamponades. SO tamponade led to more decrease in ONL + IS thickness, while Gas tamponade led to an increase (p = 0.001, Table 3). OS+RPE thickness had more increase in Gas tamponade eyes compared to SO tamponade eyes (p = 0.032, Table 3).

No significant result was found in other analysis of retinal or choroidal layer thickness changes during observation.

The macular perfusion changes after RRD surgery have not reached a commitment according to previous researches, which is due to different study designs and observation length. Wu et al. found significantly lower blood flow in both SCP and DCP than fellow eyes after retinal repair [26]. The surgical operation included vitrectomy and scleral buckling. The follow-up time was relatively longer than our research (3.6 ± 2.4 months, ranging 2–9 months). Wang et al., found significant increase in retinal blood flow in RRD eyes with successful PPV [22], however, the operation was combined with cataract surgery in their study. Phacoemulsification operation has been demonstrated to increase macular perfusion after surgery for up to 3 months [28, 29]. Furthermore, these two studies enrolled RRD patients with macular involvement. Detachment of macula before retinal repair could possibly influence macular perfusion even after successful reattachment [25]. Thus, the finding of differences in the alterations of various retinal blood flow among our macular-on RRD patients is more likely to present the exact fundus vasculature changes after surgery. For retinal blood flow, from 2 weeks to 6 weeks, we found an increasing trend in gas tamponade eyes (Tables 2 and 4). The short-term increase of blood flow might be related to postoperative inflammation [22]. On the other hand, we found a decrease of blood flow in SO tamponade eyes. More importantly, such opposite changing trend led to significant intergroup difference (Tables 2 and 4). From 6 weeks to 12 weeks, we could still found that SO tamponade led to more blood flow decrease in both SCPFD and DCPFD than Gas tamponade (Table 2). Thus, the present study demonstrated that SO tamponade may result in poorer macular perfusion in both superficial and deep retinal capillary plexus.

For postoperative choroidal status after retinal repair, previous studies used choroidal thickness as an indicator. Akkoyun et al. reported increased subfoveal choroidal thickness (SFCT) at 1 week after retinal repair, which they thought might represent inflammatory reaction [30]. Karimi et al. reported time-related reduce of SFCT in SO tamponade eyes which may possibly be due to tamponade effect of silicone oil [18]. However, change of choroidal thickness is not necessarily associated with choroidal vasculature change [31, 32]. We failed to find any significant result from choroidal thickness comparison during observation. On the other hand, we found significant intergroup difference in parafoveal CCPFD between Gas tamponade and SO tamponade eyes (Table 2). From 2 weeks to 6 weeks, parafoveal CCPFD had more increase in SO tamponade eyes. We suspect the increased of choroidal blood flow in SO group may indicate more severe postoperative choroidal inflammation, which could be related to the relatively greater amount of cryopexy or laser photocoagulation during procedure following SO use. We postulate that choroidal blood flow change could be a more sensitive indicator than choroidal thickness change.

In previous studies on retinal structure change after retinal repair, most researchers divided the retinal layer in a relatively general way. In the present study, the retinal layers were strictly segmented into seven layers as introduced in the method section. As components of inner retina, both NFL and GCL + IPL have shown significant intergroup differences in their thicknesses changes. From 2 weeks to 6 weeks, SO tamponade led to more pronounced thickness decrease of these layers than gas tamponade (Tables 3 and 5). Such finding is consistent to the hypothesis of the toxic effect of SO on ganglion cell [16]. Kamila et al. also reported more pronounced decrease of thickness in GCL-IPL complex after SO tamponade in PPV, compared to gas (C3F8&SF6) tamponade [16]. As macular was not involved in our RRD patients, our findings may strongly indicate negative tamponade effects of silicone oil on inner retinal structures (NFL&GCL + IPL), which could possibly lead to poorer final visual acuity [23].

We found relatively more decrease of INL and OPL thicknesses in SO tamponade eyes than in Gas tamponade eyes during our observation (Table 3). Such findings seemed to contradict with results from certain previous studies. Yasin et al. reported thicker central INL and OPL in SO tamponade eyes than in fellow eyes [17]. Marcel et al. also reported a time related increase of INL-OPL thickness after successful retinal reattachment [23]. However, their patients all suffered from macular-off detachment. Theoretically, the inflammatory reactions after retinal detachment and reattachment repair may be observed in the pattern of increase in the volume of INL and OPL thickness [33]. However, in the present study, we found more decrease of these two retinal layers thicknesses in SO tamponade eyes than in Gas tamponade eyes, which led to significant intergroup difference (Table 3). We thought that in the process of structural remodeling of Müller cells after retinal detachment and reattachment, the INL and OPL layer thickness might also be influenced by factors other than postoperative inflammation. SO may have potential negative effects on INL and OPL in the undetached retinal area. Whether our finding indeed revealed variable impacts of different intravitreal tamponades deserve further studies.

Photoreceptor lies in the ONL + IS and OS+RPE layers. The integrity and thickness of IS/OS has been reported to be an important predictor of postoperative VA after RD surgery [34, 35]. On the other hand, due to ischemia in external layer and neuron cell loss, reduction of ONL thickness after retinal repair has been demonstrated, and maintenance of ONL thickness is also found to be correlated with postoperative VA [23, 24, 36‐38]. We found that SO tamponade lead to more decrease in both ONL + IS and OS+RPE thicknesses (Table 3). We hypothesized that compared to gas tamponade, SO tamponade may lead to more thickness loss in ONL, and it might suppress photoreceptor recovery, which could both be related to poorer VA prognosis. Such finding might indicate potential negative impact of SO tamponade on visual prognosis.

During recent years, numerous researches has been performed for the underneath mechanism of Silicone Oil-Related Visual Loss (SORVL), and several pathophysiologic hypotheses have been proposed [39]. The following factors may all play their roles: photo-toxicity [40], fat soluble elements from the retina dissolved by SO [41], lost buffering capacity of the vitreous cavity and presence of intravitreal SO leading to impaired retinal homeostasis [5, 42, 43]. However, the exact underlying mechanism remains unknown.

Several remarks should be made with respect to the results of this study. On account of tamponade choice and timely SO removal clinical practice in our ophthalmic center, the observation time was relatively short and only comparison between gas and SO tamponade were analyzed. Extended observation period for SO tamponade impact would be necessary in future studies. Due to practical routine of intravitreal tamponade choice in our center, we only compared sterilized air and SO. We obviously lacked fundus vasculature and structure data before surgery as baseline, and we did not take fellow eyes data as control like several previous studies [16, 17, 22, 24, 26]. Unaffected eyes with asymptomatic feature could have vascular abnormalities in fundus diseases like retinal vein occlusion and primary open-angle glaucoma [44, 45]. Retinal detachment could also be bilateral in quite a number of patients. Thus, we did not take the fellow eye data of our patients as baseline or control to avoid potential bias. After all, the present study emphasized on the changes of fundus vasculature and structure during follow-up visits. Furthermore, the sample size of our study was relatively small.