Association between external limiting membrane status and visual acuity in diabetic macular oedema
- 1Department of Ophthalmology, Kobe City Medical Centre General Hospital, Kobe, Hyogo, Japan
- 2Department of Ophthalmology, Institute of Biomedical Research and Innovation, Kobe, Hyogo, Japan
- Correspondence to Dr Noriko Miyamoto, Department of Ophthalmology, Kobe City Medical Centre, General Hospital, 2-1-1 Minatojima-minamimachi, Chuo-ku, Kobe Hyogo 6500047, Japan;
- Received 20 December 2011
- Revised 24 August 2012
- Accepted 23 October 2012
- Published Online First 21 November 2012
Aim To evaluate the association between the foveal external limiting membrane (ELM) status and visual acuity (VA) in diabetic macular oedema (DMO).
Methods We retrospectively reviewed the spectral domain optical coherence tomography images of 127 eyes from 127 patients with DMO and evaluated the correlation between the logarithm of the minimal angle of resolution (logMAR) VA and the statuses of the foveal ELM, inner segment/outer segment (IS/OS) and cone outer segment tips (COST); foveal macular thickness (FMT); and presence or absence of hard exudates (HE), serous retinal detachment (SRD) and vitreous adhesion. The integrity of the ELM, IS/OS, and COST was classified into three categories (absent, disrupted and complete).
Results There was a strong correlation between VA and the statuses of the ELM (r=0.699, p<0.001), IS/OS (r=0.716, p<0.001) and COST (r=0.471, p<0.001). There was no correlation between FMT and logMAR VA (r=−0.036, p=0.687). However, when we analysed the correlation between FMT and VA by dividing patients into those with FMT ≤250 μm and those with FMT >250 μm, there was a positive correlation between FMT and VA in eyes with FMT ≤250 μm (r=−0.601, p<0.0001) and a negative correlation in eyes with FMT>250 μm (r=0.290, p<0.01). Other factors HE, SRD and vitreous adhesion did not correlate with VA.
Conclusions In DMO, the ELM status may be as closely related to VA as the IS/OS status.
Diabetic macular oedema (DMO) may occur at any stage in the progression of diabetic retinopathy and can often cause significant acquired vision loss. In the early stages of DMO, breakdown of the blood–retinal barrier (BRB) may occur, resulting in accumulation of plasma proteins, lipids, and extracellular fluid in the macula.1–3 The BRB consists of inner and outer components: the inner BRB consists of tight junctions between retinal capillary endothelial cells, and the outer BRB consists of tight junctions between retinal pigment epithelial cells (RPE). Astrocytes, Müller cells and pericytes contribute to the proper functioning of the inner BRB.4 ,5 In diabetic retinopathy, retinal hypoxia results in a breakdown of the inner BRB and induces accumulation of fluid in the outer plexiform layer and the inner nuclear layer, resulting in cystoid macular oedema. Serous retinal detachment (SRD) and sponge-like retinal swelling are also observed with or without cystoid macular oedema in optical coherence tomography (OCT) findings of DMO patients.6 The pathogenesis of the different types of DMO is not well known. Outer BRB dysfunction was observed in diabetic rodents.7 The RPE plays an important role as an ion and fluid pump for the reabsorption of subretinal fluid, and impairment of these functions may also be related to the pathogenesis of DMO.
OCT has been used to quantify macular thickness, a characteristic always considered to be an indicator of visual acuity (VA). Currently, spectral domain OCT is used to detect previously unseen detailed morphological changes of the retina such as in the inner segment/outer segments (IS/OS) and external limiting membrane (ELM). It is believed that the presence of these structures is crucial for good vision. Indeed, it has been previously reported that the IS/OS status may be related to VA in DMO8; however, the association between the ELM status and VA is not well known. Furthermore, it has also been reported that the ELM status is more useful than the IS/OS status in the evaluation of age-related macular degeneration (AMD).9 In this study, we investigated the association between the ELM status and VA in DMO.
Patients and methods
We retrospectively reviewed medical records of 127 eyes of 127 patients with diabetic retinopathy with current or previous clinically significant DMO. These patients were examined via spectral domain OCT (Spectralis Heidelberg retina angiograph+optical coherence tomography; Heidelberg Engineering, Dossenheim, Germany) between October 2009 and February 2011 at Kobe City Medical Centre General Hospital. All procedures conformed to the Declaration of Helsinki for research involving human subjects, and Institutional Review Board and Ethics Committee approval was obtained. When patients had DMO in both eyes, only the right eye was included in this study.
Spectral domain OCT images were obtained using Spectralis (Spectralis Family Acquisition Module, V.184.108.40.206, Heidelberg, Germany) and Heidelberg Eye Explorer (V.220.127.116.11, Heidelberg, Germany) equipment. Horizontal and vertical scans of the macula were recorded for each eye. The measurements were performed under pupillary dilation. The eye tracking system of the device was used to ensure that the scans were in the correct position; the position of the fovea was manually detected during examination. An average of >40 scans were recorded for each eye. Images with 40 scans or fewer, or without appropriate eye tracking were excluded from this study.
We determined the presence or absence of SRD, hard exudates (HE) and vitreous adhesion for each image (figure 1). This procedure was performed within a foveal area 1.8 mm in diameter, which is considered to be the macula.10 The status of the foveal ELM, IS/OS and cone outer segment tips (COST) on each image was categorised into three groups (complete, disrupted and absent) within this area. The complete group included patients who had 100% preservation of structures in ELM, IS/OS and COST status. The disrupted group included patients who had disrupted structures from nearly 100% loss to almost 100% preservation; the absent group included patients who had 100% loss of these structures (figure 2). The percentage disruption was averaged to generate a number between 0% (complete) and 100% (absent). Foveal macular thickness (FMT) was defined as the distance between the vitreoretinal border and the edge of the RPE. All of the measurements were performed by two of the authors (SI and NM), who were blinded to the VA of the participants. Clinical characteristics, including age and sex, were also reviewed and analysed.
The statistical program SPSS (V.11.0.1, SPSS Japan, Tokyo, Japan) was used to analyse the sample data. Descriptive analyses are reported as mean±SD unless otherwise specified. VA was measured with a Landolt C chart and then converted to a logarithm of the minimal angle of resolution (logMAR) equivalent. Bivariate relationships were examined using the Spearman correlation coefficient. Measurement values of the groups were compared using one-way analysis of variance with the Bonferroni correction. Multivariable regression analysis was used to investigate which parameter was important for VA, setting VA as the dependent variable. A p-value of <0.05 was considered statistically significant.
A total of 127 eyes of 127 patients (69 men and 58 women) with DMO were included this study. The mean age was 66.1±8.3 years and their mean logMAR VA was 0.428±0.403 (range, −0.18–2.0). Of the 127 patients, spectral domain OCT revealed SRD in 25 patients (19.7%), HE in 62 patients (48.8%) and vitreous adhesion in 50 patients (39.4%). The mean FMT was 381.5±186.3 μm (range, 115–914 μm).
Spectral domain OCT images in the photoreceptor layer were categorised into three groups: complete, disrupted and absent. Of the 127 patients, the ELM was complete in the fovea of 59 patients (46.5%), disrupted in 43 patients (33.9%) and absent in 25 patients (19.7%); the IS/OS was complete in 32 patients (25.2%), disrupted in 60 patients (47.2%) and absent in 35 patients (27.6%); and the COST was absent in 86 patients (67.7%), disrupted in 38 patients (29.9%) and complete in only three patients (2.4%) (table 1). The distribution of patients based on the integrity of the IS/OS junction and the ELM was observed: complete IS/OS and ELM (32 patients, 25.2%), disrupted IS/OS but intact ELM (27 patients, 21.3%), disrupted IS/OS and ELM (33 patients, 26%), absent IS/OS but disrupted ELM (10 patients, 7.9%), and absent IS/OS and ELM (25 eyes, 19.7%).
Table 2 shows the correlation coefficient between each factor and VA. There was a strong correlation between VA and ELM status (r=0.699, p<0.001), IS/OS status (r=0.716, p<0.001) and COST status (r=0.471, p<0.001). Other factors (HE, SRD and vitreous adhesion) had no correlation with VA.
The most important multivariate predictor of VA was ELM status (r2=0.460) and IS/OS status was second in importance (ELM status and IS/OS status: r2=0.497). The correlation was not changed by adding the other factors.
Figure 3 shows the comparisons between VA and ELM status (figure 3A), IS/OS status (figure 3B) and COST status (figure 3C). The absence or disruption of ELM and IS/OS lines was associated with decreased VA (p<0.05). A significant difference was also found between absent and disrupted COST groups (p<0.05).
Figure 4 shows the correlation between VA and FMT. FMT was not significantly correlated with logMAR VA (r=−0.036, p=0.687, figure 4A); however, there was a strong positive correlation between FMT and VA in eyes with FMT≤250 μm (r=−0.601, p<0.0001) and a negative correlation in eyes with FMT>250 μm (r=0.290, p<0.01, figure 4B).
In this study, we evaluated the associations between VA and IS/OS, ELM and COST integrity in DMO, between VA and FMT, and between VA and the presence or absence of HE, SRD and vitreous adhesion in DMO. Other studies have shown that the statuses of the IS/OS and ELM are also useful to evaluate the effect of pars plana vitrectomy on VA in various retinal diseases, such as macular holes,11 ,12 retinal detachment13 ,14 and DMO.15 We classified the integrity of the ELM and IS/OS into three categories (absent, disrupted and complete), whereas other groups have used disrupted ELM and IS/OS lengths16 or the percentage of IS/OS disruption.8 However, it was difficult to evaluate ELM and IS/OS disruption objectively, because not all OCT scan lines were used. Furthermore, since the disrupted category could include patients who had almost 100% preservation and other patients who had nearly 100% loss of these structures, this category is too broad to evaluate the association with VA accurately. To verify this category, we also assigned a percentage disruption to each patient. There was a strong correlation between VA and ELM status (r=0.716 p<0.001), IS/OS status (r=0.747, p<0.001) and COST status (r=0.483, p<0.001). There was not much difference between three categories and a percentage disruption. Although the classification used in this study is rather simple, it is very useful for clinical practice.
A previous report showed a modest correlation between OCT-measured centre point thickness and VA in DMO.17 Our data showed no significant correlation between VA and FMT. However, when we analysed the correlation between FMT and VA by dividing patients into one group with FMT ≤ 250 μm and those with FMT >250 μm, there was a strong positive correlation between FMT and VA in eyes with FMT≤250 μm (r=−0.601, p<0.0001) and a negative correlation in eyes with FMT >250 μm(r=0.290, p<0.01). We included patients with current or previous clinically significant DMO. FMT is probably just one of several variables affecting VA. Both thickening and thinning of the retina may contribute to poor VA. There were some cases with thinning retinas that had been damaged by chronic thickening that showed significantly reduced VA. Retinal thickening may not have an immediate effect on VA. Indeed, we often observe that some patients maintain good VA with retinal thickening and an intact outer retinal layer in the early stage of DMO. When the resolution of older OCT devices was limited, FMT was the only quantifiable factor associated with VA.18 Now, with the improved resolution of the current generation of OCT machines, visualisation of intraretinal structures is greatly enhanced, particularly at the level of the ELM and IS/OS junction, which describes the integrity of the photoreceptors.19–21
It has been previously reported that the integrity of the photoreceptor IS/OS junction and ELM signals detected by OCT may be correlated with visual function in various retinal diseases such as AMD9 and retinal vein occlusion.16 In addition, disruption of the photoreceptor IS/OS junction is an important predictor of VA in DMO.8 Our results show that the ELM status may be as closely related to VA as the IS/OS status in DMO. In our study, disrupted and absent ELMs were seen in 43 patients (33.9%) and 25 patients (19.7%), respectively, whereas disrupted and absent IS/OSs were seen in 60 patients (47.2%) and 35 patients (27.6%), respectively. Disappearance of the IS/OS was more frequently observed compared to disappearance of the ELM. When the distribution of patients based on the integrity of the IS/OS junction and the ELM was observed, the status of the IS/OS was always same or worse than status of the ELM. This suggests that the IS/OS may be more sensitive than the ELM, and that there may be differences in the process of retinal damage between these two structures. It was reported that the ELM status was more highly correlated with VA than was the IS/OS junction; a complete or disrupted ELM was seen in 65.2% of patients, whereas a complete or disrupted IS/OS was seen in only 37.3% of patients and was absent in 67.2% of patients in AMD.9 In our study, absent ELMs and IS/OSs were seen in 19.7% and 27.6% of patients, respectively. The incidence of absent ELM and IS/OS, especially IS/OS, was much higher in AMD than in DMO. The difference in the status of ELM and IS/OS between AMD and DMO may be due to the process of retinal structural damage. In the early stages of DMO, breakdown of BRB may occur and this process may be related to damage to the ELM and IS/OS. It is known that the BRB is a physiological barrier that regulates ion, protein and water flux into and out of the retina, and it consists of inner and outer components. Diabetic retinopathy is initiated by an alteration of the inner BRB, and neovascular AMD is a result of an alteration of the outer BRB.22 The alteration of the BRB may make a difference in the process of retinal structural damage between AMD and DMO.
Although the classification system in this study is rather simple, it is easy to use in clinical practice, and is applicable to patients who undergo other therapies, such as laser photocoagulation, intravitreal injection of antivascular endothelial growth factor agents, and intravitreal or sub-Tenon's injection of triamcinolone acetonide. However, when we observe the statuses of the ELM and IS/OS before and after treatment and examine whether they can recover from disruption or absence, this classification is insufficient and further detailed evaluations, such as the percentage of disruption, are needed.
In conclusion, the ELM status is likely be as closely related to VA as is the IS/OS status in DMO and it may be useful in the evaluation of prognosis for the treatment of DMO.
Contributors (1) substantial contributions to conception and design, acquisition of data, and analysis and interpretation of data: SI and NM; (2) drafting the article or revising it critically for important intellectual content: KI and YK; (3) final approval of the version to be published: NM.
Competing interests None.
Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed.