Article Text

Download PDFPDF

Prevalence of vitreomacular interface abnormalities on spectral domain optical coherence tomography of patients undergoing macular photocoagulation for centre involving diabetic macular oedema
  1. I Akbar Khan1,
  2. M D Mohamed2,
  3. S S Mann2,
  4. P G Hysi2,
  5. D A Laidlaw2
  1. 1Guy's Kings’ Thomas’ Medical School, University of London, London, UK
  2. 2Department of Ophthalmology, St Thomas’ Hospital, Guys & St Thomas’ Hospital NHS Foundation Trust, London, UK
  1. Correspondence to Alistair Laidlaw, Department of Ophthalmology, St Thomas Hospital, Westminster Bridge Road, London SE1 7EH, UK; Alistair.laidlaw{at}


Aim Macular traction may influence the formation and response to treatment of diabetic macular oedema (DME). The aim of this study was to determine the prevalence and associations of spectral domain optical coherence tomography (SD-OCT) evident epiretinal membrane (ERM) and/or partial vitreomacular separation (pVMS) in consecutive patients undergoing macular photocoagulation for centre involving DME.

Methods A single-centre retrospective cross-sectional observational study.

Results 198 eyes of 198 patients were included. Twelve per cent of eyes demonstrated pVMS and 14% ERM. All cases of pVMS had vitreoretinal adhesion located in the Early Treatment Diabetic Retinopathy Study grid central 1 mm subfield. In 2/3 of ERM cases, ERM was either found in the central subfield or the thickening associated with ERM was contiguous with the thickening in the central subfield. Patients with signs of ERM or pVMS were significantly older and had significantly worse acuity than those without (mean age 67.2 vs 62.8 years (p=0.02); 0.49 vs 0.31 logMAR, p=0.0006). Macular thickness was similar in both groups. The prevalence of pVMS and/or ERM were 31% in Caucasian, 5% in Asian and 24% in Afro-Caribbean subjects (p=0.11).

Conclusions ERM or pVMS was found on SD-OCT scanning in 25% of patients undergoing laser for centre involving DME. In 20% of all patients, these potentially tractional elements were either present in the central subfield scan or the traction was contiguous with the central macular thickening, suggesting a possible role for surgical or enzymatic relief of traction in their management. This requires targeted investigation.

  • Vitreous
  • Diagnostic tests/Investigation
  • Imaging
  • Macula
  • Treatment Surgery

Statistics from


Macular surface traction is recognised as one of many possible factors in the aetiology of diabetic macular oedema (DME).1–3 Such traction may arise from an epiretinal membrane (ERM) or partial vitreomacular separation (pVMS); these have been jointly termed vitreomacular interface abnormalities (VMIA).4

There are data to suggest that vitrectomy and relief of traction induced by VMIA may be an effective treatment for DME. Many of these studies were performed on patients with biomicroscopically evident distortion of the macular surface.2 ,3 ,5–8 Furthermore, at least three groups have retrospectively identified that time domain (TD) optical coherence tomography (OCT) evidence of an ERM or pVMS even in the absence of biomicroscopically evident VMIA predicts a beneficial outcome of vitrectomy for DME.2 ,9 ,10 The effectiveness of relief of VMIA identified on spectral domain (SD)-OCT scanning in patients with centre involving DME and a biomicroscopically normal vitreomacular interface has not been evaluated prospectively. Such VMIA may also be associated with, rather than necessarily the cause of, the centre involving DME.

There are only limited data on the prevalence of VMIA in patients with DME.2 ,4 ,10–14 Published studies have used varying diagnostic criteria and retinal imaging methodologies. Selection bias may also have favoured patients undergoing vitrectomy for DME or those with refractory DME, and there may also be racial differences affecting prevalence. We are only aware of two studies in which SD-OCT was performed on an unselected group of patients with DME.11 ,12 These studies have respectively reported prevalences of VMIA of 75% and 6.7%.

The introduction and standard use of antivascular endothelial growth factor (anti-VEGF) agents as the gold standard in management of centre involving DME has been shown to improve clinical outcomes but also to increase the treatment burden and cost.15 There are data to suggest that anti-VEGF agents may however be less effective in patients with VMIA.16 Anti-VEGF agents are also known to induce an angiofibrotic switch, which may alter the vitreoretinal interface and in turn increase the prevalence of VMIA.17 The introduction of enzymatic agents to facilitate induction of a posterior vitreous detachment (PVD) may also conceivably provide a non-surgical means of chemically manipulating the diabetic vitreoretinal interface.18 ,19 The study described below was performed at a time when routine anti-VEGF therapy had not been introduced in the English National Health Service (NHS); standard of care for patients with centre involving DME was therefore laser therapy at that time.

Investigation of the effectiveness of alternative treatment modalities, and in particular relief of VMIA, to augment or replace standard anti-VEGF therapy for DME is therefore timely, and this study forms a part of such a programmed investigation.

The aim of this retrospective cross-sectional observational study was therefore to determine the prevalence and associations of SD-OCT signs of VMIA in a consecutive series of patients undergoing macular laser for centre involving DME.


A retrospective review of medical records and pretreatment SD-OCT scans from 198 eyes of 198 consecutive patients who underwent macular photocoagulation for centre involving DME at a single centre over a 15-month period.

Inclusion criteria were a confirmed clinical diagnosis of diabetic retinopathy, centre involving DME with a mean SD-OCT central subfield grid thickness of 280 µm or greater. Patients who had undergone previous laser treatments were not excluded. No exclusions were made based on type or duration of diabetes or on the presence of proliferative versus non-proliferative diabetic retinopathy.

In patients with both eyes meeting these criteria, one eye was selected at random using a computer-based pseudo-random number generator with outputs 1 and 2 corresponding to left or right eye, respectively (Research Randomizer,

OCT was routinely performed prior to laser treatment using commercially available equipment (Topcon 3D-OCT 1000 or 2000) by trained ophthalmic technicians. The scanned area was 6  × 6 mm2, centred on the fovea. Each horizontal B scan comprised 512 A scans, and 128 sequential vertically arranged B scans completed the grid image.

Grey-scale B scan OCT images were analysed to classify the presence, type and zone of VMIA. Analysis was confined to the 6 mm diameter Early Treatment Diabetic Retinopathy Study (ETDRS) subfield grid20 centred on the fovea.21 ,22

The OCT appearance of the inner retinal surface was classified as one of (a) normal, (b) ERM, (c) pVMS or (d) complete pre-macular PVD.

Scans showing no abnormalities of the inner retinal signal or evidence of vitreomacular separation were classified as normal.

ERMs were identified based on the presence of an internal high signal layer, with or without localised separation from the inner retina or macular pseudo hole formation and no persistent vitreoretinal adhesion within the scan field. This is a modification of the previously described definitions of ERMs to apply to higher resolution images.23

pVMS was identified as persistent vitreomacular adhesion, with localised separation of the posterior hyaloid from the macular surface, with or without an ERM involving the anterior or posterior vitreoschisis lamellae.24

Complete macular PVD was the presence of a normal inner macular signal with a visualised posterior hyaloid face and no identified VMIA. The extent of any possible eccentric or papillary adhesion was not evaluated in such cases.12 ,22

Areas of ERM or pVMS were further analysed to assess the relationship of the area of traction with the central OCT subfield thickening. This was classified as contiguous or discontiguous traction. Contiguous traction was defined as pVMS or ERM within the thickened central subfield, or macular thickening with the traction being located in ETDRS grid zone 2 or 3, but with continuity of macular thickening between the central subfield and eccentric areas in which traction was present. Discontiguous traction was defined as an area of traction and underlying macular thickening in zone 2 or 3, which was separated by an area of normal thickness macula from the central subfield. The assumption being that discontiguous traction may not be contributing to the central macular thickening.

Comparisons between the groups were done using Fisher's exact test for categorical analysis and Mann–Whitney–Wilcoxon test and multivariable logistic regressions models. All statistics were calculated using STATA V.12.1 (StataCorp, College Station, Texas, USA).


The study cohort included 115 men and 83 women with a median age of 65.5 years at the time of laser treatment (age range 25–89 years). The self-reported ethnic mix reflected the local population with 48% reporting Caucasian, 29% African and 10% Asian ethnicity, the remainder being either undefined in the medical record or of mixed background. A total of 59 of 198 patients (30%) had undergone macular photocoagulation for DME (range 1–3 times), the remainder were treatment naive. logMAR transformed Snellen visual acuity (VA) measurements preoperatively ranged from −0.08 log MAR to counting fingers, with a median of 0.30 logMAR. The median central subfield thickness of the whole group was 337.5 µm (range 280–901 µm).

No SD-OCT VMIA was visible on the scans of 93 patients, and a further 55 showed a pre-macular PVD with no persistent adhesion within the macular area.

ERM was found in 27 patients (14%) and pVMS in 23 (12%). VMIA was therefore found in 50 of the 198 patients studied (25%).

In 40/50 cases with VMIA (80%), the area of abnormality either involved zone 1 or the retinal thickening in zone 1 was contiguous with the eccentric area of thickening and VMIA.

Of the 23 patients exhibiting pVMS, all had VMIA located within the central subfield.

In 11/27 patients with ERM, the membrane was present in zone 1. A further 6/27 ERMs were contiguous but not central. In 10/27 cases, ERM was discontiguous: that is, the area affected by VMIA was separated from the thickened central subfield by an area of normal thickness macular.

Patients with VMIA were significantly older than those without (mean age 67.2 years, range 29.4–89.3 years; and mean age 62.8 years, range 25.3–87.5 years(Mann–Whitney p=0.02)).

The mean pre-laser VA of eyes demonstrating VMIA was worse than that of eyes without VMIA 0.49 logMAR (range 0–1.85 logMAR) vs 0.31 logMAR (range −0.08 to 1.85 logMAR, p=0.0006).

There was no significant difference in the mean central subfield thicknesses of patients with VMIA and those without (380 µm, range 280–901 µm and 363 µm, range 280–840 µm in the VMIA and non-VMIA groups, respectively).

The ethnic mix of patients with VMIA largely reflected that of the sample population; however, there was a non-significant (two-sided Fisher's exact p=0.11) over-representation of Caucasian patients with VMIA (58% vs 45% of non-Caucasians), and only one Asian patient exhibited VMIA (two-sided Fisher's exact p=0.03).

VMIA were present in 37/138 (27%) of patients undergoing their first macular laser treatment compared with 13/60 (22%) in those undergoing second or subsequent laser therapy (p=0.51).


Relief of clinically apparent macular traction has been shown to be an effective treatment in patients with DME.2 ,3 ,5–8

Biomicroscopic assessment of the inner macular surface to identify such patients is complex and now routinely complemented by OCT imaging.4 TD-OCT was found to be twice as sensitive as ophthalmoscopy, fluorescein angiography and colour photography combined in detecting VMIA in patients with DME.4 The evolution from TD to SD-OCT scanning has resulted in improved axial resolution from 12 to 6 µm and more complete scanning of the macular surface, suggesting that biomicroscopically occult VMIA may be even more prevalent if SD-OCT scanning is routinely performed.25 ,26

We are not aware of a comprehensive standardised classification of the SD-OCT vitreoretinal interface.11 ,27 ,28 All patients in this study had centre involving DME with a mean central subfield thickness of at least 280 µm. Accordingly, all patients would be expected to have an abnormal macular profile. This study concentrated on evidence of VMIA (pVMS or ERM) that might be providing a mechanical contribution to the central macular thickening. This was to inform further studies on the effectiveness of relief of such OCT-defined traction in patients with DME.

pVMS arises from partial separation of the posterior hyaloid. In cases where there is secondary tractional change, it is referred to as vitreomacular traction27 with evidence on OCT scanning of ‘tissue elevation and deformity’ at the site of separation.22 In cases where the macular profile is not altered, it is referred to as vitreomacular adhesion. Vitreoschisis may underlie the persistence of adhesion in the latter cases. We have used the term pVMS because in the presence of concurrent DME the cause of retinal thickening cannot be assumed to be tractional. The classification of pVMS included patients in whom there was partial vitreoretinal separation in the presence of an ERM in the posterior lamella.

ERM is usually employed to describe a macular surface membrane occurring in the absence of partially separated posterior hyaloid.27 This appears as a ‘thin hyper-reflective band’ seen anterior to the retina on OCT.4 ,12 Histological studies reporting persistent vitreous cortex within ERMs, OCT studies demonstrating persistent vitreopapillary attachment in patients with ERM and a greater understanding of vitreoschisis suggest that persistent vitreoretinal adhesion may however be important in both conditions.12 ,22 ,24 ,29

As described in the introduction, anti-VEGF agents now represent the standard of care for patients with centre involving DME. Such treatment has recently been made available in England on the NHS, but was not available at the time of this study; macular photocoagulation was therefore the contemporaneous standard of care. The study cohort were therefore not selected or potentially biased by exclusion of a contemporaneous group receiving anti-VEGF therapy.

In this study, evidence of VMIA within the central 6 mm ETDRS grid area of the macula was present in 25% of patients. In 17% of cases, VMIA were present in the central OCT subfield. In a further 3% of cases, VMIA occurred in zone 2 or 3 but thickening was contiguous with that within the central subfield. A further 5% of cases had VMIA that were discontiguous with areas of central subfield thickening.

It might be suggested from these results that one in five of this consecutive series of patients undergoing treatment for centre involving DME would be potential candidates for a study of the effectiveness of relief of SD-OCT-defined VMIA: that is, those with contiguous oedema.

These prevalence data are considerably at odds with the 75% prevalence reported by Ophir and Martinez from a North African and Middle Eastern population and considerably more than the 6.7% prevalence reported by Chang et al11 in a Chinese population. 12 Ophir and Martinez included scans of the disc and consequently sampled a larger area but employed the same definition of contiguous traction. In our study, we found a prevalence of VMIA of 31% in Caucasian patients, 5% in Asian and 24% in Afro-Caribbean subjects (p=0.11). The significance of these differences weakens when adjustments are made for age and gender, although this may be explained in part by the fact that the sample used is too small for complex logistic regression models. It is conceivable that racial factors influence the vitreomacular interface. Further investigation of this is however required to allow conclusions to be drawn.

The finding of worse acuity in those with VMIA compared with those without is new. The groups were similar in both age and macular thickness. There was a median difference equivalent to nine ETDRS letters between the groups (0.31 logMAR vs 0.49 logMAR, p=0.0006). It may be that traction has a more deleterious effect on acuity than exudation.

There was no significant difference found in the prevalence of SD-OCT signs, suggesting VMIA in patients undergoing their first (27%) or subsequent (22%) laser treatments. This could be interpreted to suggest that such macular traction does not affect the outcome of laser treatment; laser might however result in changes in the vitreoretinal interface, and it is likely that there is a multifactorial aetiology of DME of which traction is only a part.

In conclusion, OCT evidence of potentially tractional VMIA was found to be present in 20% of patients undergoing the contemporaneous standard of care, macular photocoagulation, for centre involving DME. Systematic studies on the effectiveness of surgical or pharmacological manipulation of the vitreomacular interface and relief of any resulting traction in such patients with DME are required in order to understand the significance of these OCT-defined abnormalities.


View Abstract


  • Competing interests None.

  • Ethics approval Not required.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Linked Articles

  • At a glance
    Keith Barton James Chodosh Jost Jonas