Introduction

Macular edema is the major cause of visual loss in patients with nonproliferative diabetic retinopathy (NPDR) [1]. The pathogenesis of DR is complex, and several vascular, inflammatory and neuronal mechanisms are involved. Inflammation mediates structural and molecular alterations associated with DR. Tissue hypoxia and dysregulation of immune responses associated with diabetes mellitus can induce increased expression of numerous vitreous mediators (cytokines, chemokines and other factors) responsible for DR development [2]. The breakdown of the tight junctions of endothelial retinal cells (which constitute the inner blood–retinal barrier) and of the pigment epithelial cells (which constitute the outer blood–retinal barrier) is responsible for the accumulation of plasma within the retina. The leakage of fluid and plasma constituents is due to oxidative damage caused by hyperglycemia. The thickening of the retina (namely the diabetic macular edema or DME) and subsequent ischemia causes the progressive loss of vision [3]. DME is stimulated and supported both by inflammatory factors that accumulate in the vitreous [2,3,4] and by vitreomacular traction forces exerted by contraction of posterior hyaloid over the internal limiting membrane (ILM) [5]. ILM is the site of adhesion between retina and vitreous; thus, the role exerted by ILM in the pathogenesis of tractional DME is surely worthwhile. Cortical vitreous of diabetic patients contains enzymes able to cross-link vitreous fibers, with subsequent stiffening of collagen over the ILM. The posterior hyaloid stiffens and becomes shorter, causing retinal traction and the loss of the physiological concavity of the macula [6]. The thickening of ILM in diabetic patients depends on the accumulation of extracellular matrix (i.e., collagen type I, III, IV, V, proteoglycans and adhesion molecules such as laminin and fibronectin) and immune cells, such as macrophages and fibroblasts [7]. The thickened ILM impedes the outflow of fluids accumulating into the retina and hinders the diffusion of oxygen from the vitreous. The accumulation of inflammatory cells increases inflammation and provokes the breakdown of blood–retinal barrier, whereas the accumulation of fibroblasts may be responsible of the epiretinal membranes formation.

The treatment for DME involves rigorous glycemic control, macular laser treatment, as indicated by the Early Treatment Diabetic Retinopathy Study, intravitreal pharmacotherapy, especially with anti-VEGF, and surgical treatment [8–10]. Also considering the inflammatory impact of diabetic retinopathy, currently the use of slow-release dexamethasone (Ozurdex; Allergan, Irvine, CA) or fluocinolone acetonide (Iluvien®) intravitreal implant has been investigated in the literature as second-line therapy for DME refractory to anti-VEGF therapy [11, 12]. Otherwise surgical treatment with pars plana vitrectomy is indicated when other approaches have not been found useful in reducing the edema. The main goal of vitrectomy in NPDR is the reduction in retinal vitreous traction. Other possible benefits are that the removal of the vitreous gel increases the availability of oxygen, with a partial reduction in the inner retina ischemia. Moreover, vitrectomy removes the inflammatory mediators, which further increases the leakage. Finally, the removal of the ILM improves exchanges with the oxygen-rich vitreous and promotes the discharge of liquid from the retina, ensuring the absence of any residual traction. Although the atraumatic removal of ILM seems to play a role also in chronic diabetic macular edema without evidence of traction [13], its efficacy in the improvement of visual acuity is controversial. In fact, some studies have reported that reduction in macular edema does not lead to noticeable functional improvement in vision [14, 15]. However, despite the removal of macular ILM has yielded good anatomical outcome in the surgical treatment of macular hole and epiretinal membrane [17–19], the adverse effects potentially related to this procedure could interfere with good functional recovery in medium- and long-term follow-ups.

The objective of this study was to investigate the functional and anatomical effect of ILM peeling in nontractional diabetic macular edema.

Methods

An extensive Medline, Embase and Cochrane search was performed including the following words “diabetic macular edema,” “internal limiting membrane peeling,” “diabetic retinopathy,” “vitrectomy” and “Muller cells.” The search accrued data from January 1, 1988, up to November 31, 2016. In addition, completed but still unpublished RCTs evaluating the effects of ILM peeling on diabetic macular edema on different outcomes were identified through a search on the www.clinicaltrials.gov Web site. The identification of relevant studies was performed independently by two of the authors (MR, FC), and conflicts resolved by the third investigator (CC).

Eligible studies included randomized or nonrandomized studies comparing the visual outcomes with and without ILM peeling in the management of DMR. There was only one RCT from Poland (NCT01218750), without study results posted. A period of up to 12 months was considered an appropriate duration of follow-up for the evaluation of surgical outcomes. We extracted data on outcome measures including best-corrected visual acuity (BCVA) before and after surgery, and central macular thickness (CMT) before and after surgery, along with the study design, diagnosis, surgical procedures and laser treatment. The primary outcome of our study was BCVA measure, whereas CMT was the secondary outcome.

The quality of trials was assessed using the Cochrane criteria (Cochrane Handbook for Systematic Reviews of Interventions, available at: www.cochrane-handbook.org). For each study, we also assessed how the population was selected, the duration and the adequacy of study follow-up [20].

Statistical analysis

Meta-analysis was performed using CMA software (comprehensive meta-analysis). For each study, the odds ratio and 95% confidence interval were considered. In the Forest plot, the studies are represented by symbols whose area is proportional to the study’s weight in the analysis. However, a random-effects model was applied for all analyses and used to estimate the difference in the outcomes between the vitrectomy group with ILM peeling and that without ILM peeling. The Z value and p value were calculated for each study and for the cumulative effect.

Results

The first database query yielded 507 articles. Of these, we screened titles and abstracts for relevant reports and selected 9 articles for which we obtained full articles. Within these reports, we excluded 3 articles because, to our opinion, were of lower level evidence. Two other reports were also excluded because they were noncomparative studies [21, 22]. The remaining four studies selected evaluated the visual and anatomical outcomes of patients affected by diffuse, nontractional diabetic macular edema, who underwent pars plana vitrectomy with and without ILM peeling [23, 24]. Among these studies, three were randomized trials and one was a retrospective observational study. Table 1 shows the main characteristics of the 4 studies included. The prevalence of proliferative diabetic retinopathy, duration of diabetes mellitus type 2, previous macular laser therapy were well balanced preoperatively between the groups, along with the main baseline characteristics of patients (age, preoperative lens status, preoperative absence of PVD, etc.). Between studies ILM peeling was performed using indocyanine green (ICG) in just one [23] out of three studies, while Patel el al did not specify if ICG was used during surgery. Posterior hyaloid face was always removed by surgeons in all eyes, even in the groups where ILM peeling was not performed.

Table 1 Characteristics of the studies included for review
Full size table

The results obtained comparing the postoperative BCVA between the ILM peeling and nonpeeling groups did not show a significant statistical superiority between the two groups on a long-term follow-up. The meta-analysis evaluating the change in BCVA before and after surgery showed that there was no statistical difference (p = 0.16) in terms of visual improvement between the groups analyzed (Fig. 1). Unfortunately it was not possible to perform a meta-analysis over the CMT data due to lack of the pre- and postoperative CMT values per group, along with the indispensable standard deviation. Anyhow the results obtained from each study did not show a statistical difference in macular thickness reduction between vitrectomy with and without ILM peeling even on a long-term follow-up.

Fig. 1
figure 1

BCVA before and after surgery is showed. The columns include the odds ratio and 95% confidence interval for each study. The last row in the spreadsheet shows the summary data. The same information is captured by the Forest plot. This plot shows each study as a point estimate with its lower and upper limit and provides a sense of the study-to-study dispersion. In the Forest plot, the studies are represented by symbols whose area is proportional to the study’s weight in the analysis. A non-ILM peeling, B ILM peeling

Full size image

The differences in the BCVA between the two groups were not statistically significant.

Discussion

The role of ILM peeling in the treatment of nontractional diabetic macular edema is controversial, and no sufficient data are available to clarify the utility of this procedure in this condition.

Data obtained in this review showed that vitrectomy with ILM peeling does not offer clear advantages. There was no significant difference in visual acuity outcomes and reduction in macular edema among patients undergoing vitrectomy with or without ILM peeling.

The role of vitrectomy in the treatment of nontractional DME has not been fully addressed, although represents a therapeutic choice for several surgeons.

Vitrectomy is performed to release the pathological influence of the vitreous on the retina and is useful in restoring the normal anatomical shape of the macula and improving visual acuity.

ILM peeling during the vitrectomy was suggested to yield a better anatomical and functional result [27,28,29,30,31,32,33], and its effectiveness was sustained long term with less recurrence [34]. However, the surgical removal of the ILM is also challenging because it can compromise the already affected retinal tissue. Poor visibility of the thin transparent membrane, the sensitivity of the macular tissue and the small dimensions must be considered during the procedure. Anyway several studies demonstrated an accelerated absorption of DME, improved anatomy but not necessarily improved VA [16, 29]. The addition of ILM peeling to vitrectomy was based on the benefit obtained by the removal of the scaffold used by astrocytes to proliferate on the retinal surface and by the elimination of all tractional forces at the vitreoretinal interface. When no tractional adherences are observable, such as in nontractional DME, vitrectomy with ILM peeling does not offer a clear advantage from the clinical data now available. However, there are many variables that can affect the data in each study. Treated patients hardly are similar and the general conclusions are difficult. The main problem today is that it is not yet possible to assess the real vitreous morphology, its relationship with the macula and the presence of pockets of liquid in front of the retina. The state of the vitreous can only be assessed during surgery using vital dyes suitable. OCT instruments can evaluate the thickness of the retina, but are not yet able to distinguish intracellular edema from that extra cellular one; thus, they cannot determine whether the damage is irreversible or not. Furthermore, the OCT frames are not able to evaluate the impairment of retinal micro circulation. The surgery is only effective if performed before irreversible damage occurs to the photoreceptors layer. Other variables useful to determine the reversibility of visual impairment are the duration of macular edema, the presence of glucose control, the extent of ischemic damage and the age of patients. Because the asset of ILM peeling might reveal only later after vitrectomy, in this study we analyzed postoperative results from 12 up to 80 months.

In conclusion, the visual acuity and CMT outcomes using pars plana vitrectomy with or without ILM peeling in patients affected by nontractional diabetic macular edema were not statistically different. In our opinion, the better way to treat a nontractional diabetic macular edema is a personalized approach involving an increased collaboration between ophthalmologists and primary care providers, considering the ILM anatomy together with data on the microcirculatory state of the retina to perform the better surgical strategy, customized on the personal risk factor. Also a larger prospective and randomized study would be necessary to appropriately address the effectiveness of ILM peeling on visual acuity outcomes.