Perioperative intraretinal fluid observed using optical coherence tomography in the epiretinal membrane

The pathogenesis of epiretinal membrane (ERM) involves glial cell proliferation over the internal limiting membrane (ILM) of the retina. The prevalence of ERM is estimated to be between 7 and 11.8%, and it increases with an increase in age [1, 2]. The ERM frequently involves the macula, and pars plana vitrectomy (PPV) together with membrane peeling, with or without ILM peeling, is commonly performed for symptomatic ERM, resulting in favorable visual prognoses in most cases [3, 4].

Previous studies have reported the incidence of macular edema after PPV. Particularly for ERM, the incidence rates of macular edema after PPV varied between 13 and 64% [5‐7]. Previously, macular edema was synonymous with macular thickening detected by ophthalmoscopic examinations. However, in the era of optical coherence tomography (OCT), macular edema can be differentiated from an increase in macular thickness by mechanical deformation; this term is reserved for macular thickening due to the presence of intraretinal fluid (IRF) or intraretinal cystoid changes shown by OCT. The presence of IRF was associated with a delay in postoperative visual improvement [5, 8]. Additionally, cystoid changes were also commonly observed before ERM surgery [5, 7]. In these cases, persistent preoperative IRF would introduce a bias in interpreting the impacts of postoperative IRF, which includes both persistent IRF and newly developed IRF.

The current study comprehensively investigated the clinical implications of perioperative IRF in idiopathic ERM by analyzing functional outcomes according to the presence of preoperative and postoperative IRF.

Medical records of patients who underwent surgery, at Pusan National University Hospital, for idiopathic ERM between January 2014 and January 2017 were reviewed retrospectively. The institutional review board of Pusan National University Hospital approved the study protocol, which complied with the tenets of the Declaration of Helsinki.

Patients with secondary ERM (diabetic retinopathy, venous occlusion, uveitis, retinal detachment, trauma, etc.) were excluded from the study. Patients who had undergone intraocular surgery within 3 months before ERM surgery, had severe cataract before operation, had a postoperative follow-up period of less than 6 months, had undergone surgery performed by surgeons with less than 2 years of experience, or had an intraretinal space because of dehiscence of the retinal layers similar to retinoschisis were also excluded to prevent bias.

All patients underwent a complete ophthalmic examination, including slit-lamp examination with a 90D lens, fundus photography, and spectral domain OCT (Cirrus HD-OCT; Carl Zeiss Meditec, Inc., Dublin, CA, USA) or swept-source OCT (DRI-OCT1 Atlantis; Topcon Inc., Tokyo, Japan). Best-corrected visual acuity was measured before surgery and 1, 3, and 6 months postoperatively using the Snellen chart and was converted to the logarithm of minimum angle of resolution (logMAR) value for statistical analysis. Demographic data such as age, sex, lens status, and surgical factors (staining method, ILM peeling, concurrent cataract surgery, and duration of surgery) were reviewed.

All the surgeries were performed by two expert surgeons (JE Lee and SW Park). Three-port 25G PPV was performed using the Constellation (Alcon Laboratories, Inc., Fort Worth, TX, USA) with a RESIGHT 700 non-contact wide-angle viewing system (Carl Zeiss Meditec AG, Jena, Germany). Phacoemulsification was performed concurrently at the surgeon’s discretion. A clear corneal incision was made at the 12 O’clock position for phacoemulsification, following which vitrectomy was performed. Triamcinolone acetonide (TA) was applied to visualize the vitreous and ERM. The fibrous membrane was removed with intraocular forceps. The ILM was removed at the surgeon’s discretion after staining it with Brilliant Blue G (BBG). The sclerotomy site remained without a suture in most cases.

The IRF associated with ERM was frequently observed before and after surgery. We assessed the presence of IRF in the perioperative period and highlighted the various impacts on preoperative and postoperative visual outcomes. Generally, perioperative IRF was related to a lower preoperative visual acuity, but it did not prevent visual recovery after the surgery.

Although the literal definition of macular edema is the accumulation of fluid in the macula, it has been inconsistently defined in the literature. Traditionally, macular edema refers to thickening of the macula. However, the macula may be thickened by mechanical deformation in ERM, and the entirety of the thickening may not necessarily represent fluid accumulation. It has been reported that the CSMT could change even after 12 months of follow-up, but it does not reach the normal value [9]. Moreover, IRF observed using OCT has been associated with cystoid leakage detected using FA [10]. McDonald et al. [11] reported that the preoperative incidence rate of cystoid macular edema (CME) in idiopathic ERM was 71% (12 out of 17 eyes), as assessed using FA. However, FA findings did not always correspond to those observed using OCT, and the positioning of the fluid in the retinal layers is not obvious in FA [12, 13].

Accordingly, we used the term IRF instead of macular edema in the current study, as that is defined more specifically and consistently in the literature. OCT allows detailed imaging of the microstructures of the retinal layers and visualization of cystoid structures not seen using FA or an ophthalmoscope. Each cross-sectional image of 3-D volume scans was critically assessed to find even small IRFs.

The results from this study appear to be conflicting with those reported by previous studies. The incidence rate of postoperative IRF including persistent preoperative IRF was 29.1% at 6 months, and the incidence rate of new IRF was 41% (49 out of 119) in the eyes without preoperative IRF. Previous studies have reported widely varying postoperative IRF incidence rates, 13 to 64% [5‐7], and new IRF incidence rates, 1 to 15.6% [5, 7, 8]. It is likely that the highest incidence rate [6] would represent thickening of the macula by both mechanical deformation and presence of fluid, as postoperative macular edema was assessed based on macular thickness measured by OCT and not based on the presence of IRF. The lowest incidence rate of about 1% indicated a proportion of delayed onset inner nuclear layer cystoid changes and not entirely that of the postoperative new IRF [8]. Frisina et al. [5] reported an incidence rate of about 10% for postoperative CME and new CME, as assessed by six radial scans (not a volume scan). Dolz-Marco et al. [7] compared the incidence rates of new IRF depending on whether combined phacoemulsification was performed or not, using macular cube scans, and reported the incidence to be 6.3% in the vitrectomy only group and 25% in the combined surgery group. In the current study, when patients were similarly divided into two subgroups, similar incidence rates, i.e., 6.9% among patients who underwent vitrectomy alone and 37.3% among those who underwent combined surgery, were observed. Overall, although the incidence rates of postoperative IRF appeared to show a wide variation in the literatures, they were comparable to ours considering the various factors associated.

Our study revealed that about one-fourth of patients had preoperative IRF and that these patients had significantly worse preoperative and postoperative visual acuity than other patients. Several previous studies reported the incidence rate of preoperative IRF to range from 13.6 to 40.6% using OCT [5, 7]. However, those reports did not analyze the impact of preoperative IRF on postoperative visual acuity. In the current study, visual acuity improved significantly after the surgery although preoperative IRF persisted in about half of the eyes for 6 months. The functional outcomes of the preoperative IRF group were worst among three groups despite the postoperative improvement. Our results were consistent with those of a recent prospective study that evaluated the risk factors of intraretinal cystoid changes after ERM surgery, which demonstrated that preoperative cystoid changes resulted in poorer postoperative visual outcomes [14].

The observations in the present study propose that preoperative factors are important in the development of IRF in ERM. This is supported by the differences in the baseline characteristics as well as the interesting findings of postoperative functional recovery. In general, visual function of the new IRF group was found to be intermediary to that of the other two groups. At baseline, visual acuity of the new IRF group was not different to that of the preoperative IRF group, but it was worse than that of the IRF(−) group. After surgery, the new IRF group demonstrated the most prominent visual improvement, which was not different from that of the IRF(−) group at 6 months. The preoperative IRF group had significantly worse visual acuity than the other two groups at 6 months postoperatively; however IRF was resolved in about half of them. Differences in the postoperative improvement between the new IRF and preoperative IRF groups suggests that preoperative IRF would indicate a more advanced stage with irreversible change, whereas new IRF group was at a less advanced stage with a relatively reversible damage. Furthermore, a more prevalent defect in DRIL indicated the irreversible structural change in the preoperative IRF group and supported our hypothesis.

The prominent improvement in the visual acuity of the new IRF group might be biased because it involved a higher rate of concurrent phacoemulsification. However, we excluded the cases with significant preoperative cataract, and visual acuity improved gradually during the 6-month follow-up period. The postoperative improvement in vision accordingly would be because of ERM surgery rather than because of cataract removal in the new IRF group. Furthermore, considering that more than 97% of the patients were pseudophakic postoperatively, the key findings of the current study, namely the worse outcomes of preoperative IRF group were still valid regardless of the functional effects of concurrent phacoemulsification.

Our results are somewhat contradictory to those of previous studies, which reported that postoperative IRF was related to worse functional outcomes. Sigler et al. [8] reported that new IRF was visually significant and led to a poor postoperative visual outcome in a small number of cases. However, their report was specifically about a finding of delayed onset IRF that had developed in the inner nuclear layer similar to the second case of FA in the present study and did not consider all cases of postoperative IRF. Another study did not differentiate the postoperative new IRF cases and the persistent preoperative IRF cases [5], which would be biased towards lowering the postoperative visual outcomes. Our novel findings of visual acuity changes related to perioperative IRF were obtained as we distinguished postoperative IRF into persistent IRF and new IRF to avoid the bias.

Several mechanisms have been suggested to underlie perioperative macular edema in idiopathic ERM. Glial cell proliferation on the retinal surface could cause damage to the vasculature by mechanical traction, which may result in vascular stasis and subsequent blood-retinal barrier (BRB) breakdown [15]. Postoperative inflammation is also one of the well-known causes. Frequent development of new IRF during the early postoperative period suggests that postoperative newly developed IRF would be associated with postoperative inflammation. Our study also demonstrated that new IRF developed most frequently at 1 month postoperatively. Release of inflammatory mediators triggered by surgery can break down the BRB and lead to an increase in vascular permeability, as observed in the case of pseudophakic CME [16]. This mechanism would be represented in this study by FA findings corresponding to the first case.

Macular edema is not an event influenced by a single factor, but it is rather a status of balance between many factors. Mechanical traction, venous stasis, and damaged BRB can cause vascular leakage; however, accumulation of IRF can be countered by physiological mechanisms that maintain the equilibrium of fluids. After ERM surgery, intraoperative damage and postoperative inflammation could put weights on the arm of IRF accumulation, whereas removal of membrane could attenuate the leakage caused by the preoperative factors. A combination of the factors noted above could tilt the balance towards either the development or resolution of IRF. Therefore, newly developed IRF represents a more advanced stage than the IRF(−) group does, but it still maintains a reversible status that could allow for a prominent improvement in vision after surgery. In contrast, preoperative IRF represents degenerative changes that exceed the issue of physiological balance even before surgery and would be a predictor of poorer functional outcomes than the other groups. These dynamic features of many participating factors were well-demonstrated in the two cases for which FA was performed in this study.

This study has some limitations. TA was used in some cases to enhance the visualization of ERM/IRF. TA is a steroid with anti-inflammatory properties, and it may affect the development of postoperative IRF. However, in our previous study, postoperative outcomes of ERM surgery were compared between two groups classified according to intravitreal or posterior subtenon TA injection at the end of the operation [17, 18]. No significant differences were found in terms of visual acuity and foveal thickness. In the present case series, a much smaller amount of TA was used just as an adjuvant for membrane peeling, and most of it was washed out immediately. Furthermore, intravitreal TA in a vitrectomized eye is known to be cleared very rapidly [19]. Consequently, the impact of TA on the results would be minimal.

The other limitations of the current study were its retrospective design, short follow-up period, the use of two types of OCT machines, and inclusion of a small cohort of patients in whom FA was performed. However, comprehensive analysis of perioperative IRF highlighted its various impacts on postoperative functional outcomes. Future studies are required to identify the specific OCT findings that may be useful for predicting the development of IRF and functional outcomes.

In summary, perioperative IRF was commonly observed in ERM surgery, and the preoperative status would be an important factor involved in its development. IRF was resolved in many cases without any treatment other than ERM surgery, and visual acuity improved significantly regardless of IRF. Preoperative IRF had a clinical significance as it reflected a more advanced stage of ERM with an irreversible damage to the macula, and it is correlated to worse visual outcomes.