---

Abstract

Optical coherence tomography (OCT) provides information akin to real-time, in-vivo histology of the retina. Diabetic retinopathy results in structural changes within the retina that can be observed by OCT imaging, and which correlates with the severity of vision loss. Previously, the second most inner highly reflective outer band seen on OCT was believed to represent the junction between the inner and outer segments (the IS/OS junction) of retinal photoreceptors. In addition to macular volume, the extent of inner/outer photoreceptor segment disruption was found to be a reliable predictive factor accounting for visual acuity loss in patients with diabetic retinopathy. More recent reports have actually demonstrated only a weak correlation between central retinal thickness and visual acuity in these patients, implying that vision may be more readily affected by disruption of the microstructural retinal architecture or direct photoreceptor damage. It has since been discovered, however, that the second most inner highly reflective band more accurately corresponds to the microstructural ellipsoid zone of the photoreceptor inner segmentsthat can beseen on retinal histologic analysis. Accurate evaluation of this highly significant microstructural layer can provide valuable information regarding disease state and visual prognosis in diabetic retinopathy and diabetic macular edema. We report on the most current and accepted information regarding this newly re-identified layer and its accurate clinical evaluation.

---

Introduction

The prevalence of diabetes mellitus is attaining epidemic proportions worldwide [1,2]. As of 2013, it is estimated that 382 million people suffer from diabetes mellitus. This number is expected to rise to 592 million by 2035 [3]. Macular edema is the most common cause of vision loss in diabetic retinopathy [4]. Of patients living with diabetes mellitus for a duration of 20 year or more, The Wisconsin Epidemiological Study of Diabetic Retinopathy reported a prevalence of diabetic macular edema of 29% in younger onset diabetics and 28% in older onset diabetics [5].
 
Diabetic retinopathy can result in structural changes within the retina which have been shown to correlate with severity of retinopathy. Optical Coherence Tomography has come to be an exceedingly useful investigative tool for classifying macular edema and quantifying macular thickness in diabetic retinopathy [6-8]. Photoreceptor dysfunction may also be a significant predictor of visual acuity in these patients [9-12]. Inner segment-ellipsoid (ISel) band (previously believed to represent the inner-segment/outersegment, or IS/OS, junction) disruption has been demonstrated to be an important predictor of visual acuity in diabetic macular edema [13]. The integrity of this anatomical microstructural layer appears to be of high clinical importance, as the absence or disruption of this band has been associated with vision loss in a number of other diseases affecting the photoreceptors and/or choroid [14-20]. Accurate evaluation of this bandwilltherefore be of particular interest to the clinicianthat is faced with treating diabetic retinal disease processeson a daily basis. As such, understanding the evaluative parameters of this microstructural layer will be of high clinical importance to the practicing retinal physician.

---

Anatomy of the Inner Segment Ellipsoid Band

High-resolution spectral domain optical coherence tomography (SD-OCT) provides a nearly cellular level resolution of the retina, invivo. In the foveal region, SD-OCT imaging has conventionally been resolved into four bands. The innermost band has been attributed to the external limiting membrane (ELM), a linear confluence of junctional complexes between Muller cells and photoreceptors [21]. In the past, the second line of high reflectivity has been ascribed to the boundary between the inner segments (IS) and outer segments (OS) of the photoreceptors, often described as the IS/OS junction [22,23]. The third band is commonly referred to as the cone outer segment tips (COST) line, or Verhoeff’s membrane (believed to represent ensheathment of cone outer segments by RPE cell apical processes) [24-26]. The outermost highly-reflective band is believed to represent he retinal pigment epithelium (RPE) itself, although Spaide and Curcio suggest that this fourth band is actually contributed by Bruch’s membrane and the choriocapillaris [27].
 
It has been demonstrated that for foveal cones and rods alike, the length of the IS microstructure is roughly the same as that of the OS microstructure [28-32]. Therefore, one might deduce that the IS/OS junction would form a plane midway between the ELM and RPE, however, in-vivo SD-OCT imaging demonstrates the second reflective band to lie significantly closer to the ELM than to the RPE.
 
Using ultra high resolution SD-OCT, Fernandez et al. were able to demonstrate the anatomy of human foveal cone photoreceptors, including visualization and identification of the distal inner-segment ellipsoid layer [33]. A scale model drawing of the outer retina was constructed by Spaide and Curcio, which was based on published histology, in order to enable visual comparisons [27]. The conclusion of this analysis on the origin of the first and fourth band was the same as previously suggested.
 
Anatomical correlation of the second band, however, has been more controversial. By comparative alignment of OCT bands with an anatomically correct model of the outer retina (using freshly isolated living retinas from the leopard frog), Curcio et al. suggest the inner segment ellipsoid layer to be responsible for contributing this secondmost inner highly reflective band [34]. The third band seemed to have originated from the posterior tips of cone outer segments, called the contact cylinders.
 
When analyzing statistical relationships between visual acuity and macular volume, history of prior treatment, and percent disruption of the second-most inner highly-reflective band seen on OCT, multivariate analysis demonstrated a statistically significant relationship with the latter element [41]. Initial studies found a good correlation between macular thickness, assessed by OCT, and visual acuity [35-38], however, recent reports show only a weak correlation between central retinal thickness and visual acuity in patients with diabetic macular edema (DME) [39]. This might further support the notion of alternative influences more readily impacting visual acuity in these patients, including direct photoreceptor damage and/or disruption of the microstructural retinal architecture (i.e. the inner segment ellipsoid layer) [39,40].

---

Classification Systems for ISel Disruption

A) Maheshwary et al., obtained horizontal and vertical SD-OCT images through the fovea with scanning laser SD-OCT using either an OPKO-OTI (Toronto, Canada) or a Heidelberg Spectralis (Vista, CA) in sixty-two eyes from 38 patients with diabetic macular edema (DME) [41]. An average foveal thickness value was obtained after reviewing both scans. Next, the photoreceptor IS/OS junction band (now the inner segment ellipsoid layer (ISel) band) was evaluated 500 microns in either direction of the fovea. [ISel] disruption was gradedfrom grade 0–2.
 
• Grade 0: Intact [ISel]
 
• Grade 1: Focal [ISel] disruption of 200 microns or less in length
 
• Grade 2: [ISel] disruption greater than 200 microns in length
 
Grades from each patient’s horizontal and vertical scans were added to yield a global disruption scale. Lastly, the percentage of disruption along the [ISel] band, again measured 500 microns in either direction from the foveal center, was recorded. These percentages were then averaged to generate a number between 0% (no ISel disruption) and 100% (total loss of the ISel band in both horizontal and vertical scans). A statistically significant correlation between percentage of disruption of the [ISel] band and visual acuity was found (P = 0.0312). Additionally, there was a strong trend suggesting a relationship between macular volume and visual acuity, although only borderline significance was demonstrated (P = 0.07). It was concluded that percentage disruption of the [ISel] band is an important predictor of visual acuity among diabetic macular edema patients [41].
 
B) Our earlier study showed for the first time that disruption of the ELM occurred even before disruption of the ISel band [10]. The ELM is considered to be part of the blood-retinal barrier that can become disrupted by pathologic conditions like diabetic retinopathy, and thus result in fluid accumulation within the macula [42]. Foreshortening of the photoreceptor inner segments might be a secondary consequence of such ELM insult and fragmentation [43].
 
ISel and ELM band disruption were graded as follows:
 
• Grade 0: No disruption of ELM and ISel
 
• Grade 1: ELM disruption, intact ISel
 
• Grade 2: Both ELM and ISel disrupted
 
A statistically significant positive correlation was found between logMAR visual acuity and grade of disruption (r=0.85, p<0.0001) [10].
 
C) Our most recent study demonstrated a simplified, comprehensive and physician friendly approach to grading ISel disruption based on OCT findings. The inner segment-ellipsoid band was studied using horizontal and vertical SD-OCT scans through the fovea. The integrity of this layer was evaluated by two experienced observers, masked to the status of diabetic retinopathy. ISel disruption was graded as follows:
 
• Grade 0: intact ISel
 
• Grade 1: focal disruption (localized, subfoveal ISel disruption)
 
• Grade 2: global disruption (generalized ISel disruption throughout the macular cube)
 
Above two classifications mentioned by us are physician friendly and easy to recall.
 
Eyes without retinopathy were found to havea normal ISel. Normal ISel was observed in 7.5% eyes with NPDR (Figure 1). Focal ISel disruption was observed in 62.5% eyes with NPDR and 27.5% eyes with PDR. Global ISel disruption was observed in 25% eyes with NPDR and 72.5% eyes with PDR. Increased severity of retinopathy was associated with increased ISel disruption, graded ‘focal’ to ‘global’ (r = 0.49, p < 0.001) (Figure 2). Furthermore, decreased visual acuity was associated with increased grade of ISel disruption (r = 0.67, p < 0.001), and a marked decrease in visual acuity was found to be associated with global disruption, specifically [44].
 
  D) Recently, a new classification system for clinically significant diabetic macular edema (CSDME) was developed utilizing a ratio between the vertical height of the largest macular cyst to the maximum measured macular thickness, as observed on OCT imaging.
 
The four classification groups were as follows:
 
CME I – Cysts less than (30%) of macular thickness
 
CME II – Cysts between 30% and 60% of macular thickness
 
CME III – Cysts between 60% and 90% of macular thickness
 
CME IV – Cyst more than 90% of macular thickness.
 
Each grade was then subdivided according to certain morphologicfeatures using labels ‘A’, ‘B’, ‘C’, and ‘D’:
 
A – Cysts without any disruption to the ELM or ISel bands
 
B – Cysts with ELM disruption
 
C – Cysts with ISeldisruption
 
D – Cysts with disruption of both ELM and ISel bands
 
Additionally, the presence of hyper-reflective foci in the outer retinal layers from the ELM to the retinal pigment epithelium within the 1 mm scanned area centered on the fovea was designated with “+” signs [45].
 
Small hyper-reflective foci can be less than 30 μm in diameter and might not be identified on fundus biomicroscopy, fluorescein angiography, or infrared imaging. They may represent tiny intraretinal protein and/or lipid deposits as they have hyper-reflectivity similar to that of the accumulated spots in an area of hard exudates [46]. The accumulation of hyper-reflective foci in the subretinal spaces indicates a poor prognosis in diabetic macular edema [47]. The presence of hyper-reflective foci in the outer retina is closely associated with ELM and ISel band disruption, and decreased visual acuity in diabetic macular edema [48]. It might also predict photoreceptor damage and a poor prognosis after vitrectomy for DME [49]. Inclusion of hyperreflective foci provided valuable information on the degree of bloodretinal barrier breakdown and its influence on visual acuity.
 
E) Another classification system was proposed to overcome the limitation of variability in the reflectivity levels of continuous [ISel] bands, seenon SD-OCT. For this, quantification of the status of the transverse length of each segment of the [ISel] band was performed. Classifications were as follows:Intact line– highly reflective and continuous line (physiologic)Faint line – presence of a continuous line with lower reflectivity (less than half of the OCT reflectivity of the RPE)
 
Disrupted line – absent or discontinuous line
 
A measurement of the transverse length of each disrupted [ISel] section was also obtained.
 
SD-OCT reflectivity in the ELM band was similarly analyzed. Relatively homogeneous and continuous ELM bands were described as intact. Absent of discontinuous ELM bands were described as disrupted.
 
A measurement of the transverse length of each ELM band section was also obtained.
 
Statistical analysis demonstrated that the total additive transverse length of disrupted [ISel] within the fovea correlated more closely with logMAR visual acuity than did foveal thickness measurements (p = 0.002) [50].

---

ISel Band Integrity and Pharmacological/Surgical Intervention

Few interventional studies have been conducted to evaluate the relationship between pharmacologic and surgical intervention and ISel band integrity. In one study consisting of eleven eyes with resolved macular edema following pars plana vitrectomy with internal limiting membrane peel for diabetic macular edema, the integrity of the ELM and photoreceptor ISel band was evaluated by SD-OCT. On linear regression analysis, a statistically significant association was found betweenvisual acuity and intact ELM and ISel bands [51].
 
In a retrospective interventional study, 61 eyes of 52 patients with DME who were successfully treated with intravitreal triamcinolone were studied. Based on the restoration of photoreceptor ISel and ELM bands on SD-OCT at final visit, the eyes were divided into three groups:
 
(1) Group A- completely visible ISel band and ELM
 
(2) Group B- disruption of ISel band and intact ELM
 
(3) Group C- disruption or loss of the ISel band and ELM.
 
It was concluded that integrity of the photoreceptor ISel band correlated well with final best corrected visual acuity (BCVA). Final BCVA of both Group A and Group B was better than that of Group C. They also concluded that shorter lengths of both ISel band and ELM band disruption were both associated with better BCVA. On multivariate analysis, however, it was found that mean disrupted ISel and ELM band length showed greater correlation to final BCVA thanlength of either disrupted ISel band or ELM band taken alone. Another significant conclusion was made that initial visual acuity (prior to treatment) and photoreceptor status can both assist in predicting the post-treatment recovery of the photoreceptor ISel band, and subsequent visual recovery [52].
 
Another retrospective interventional study was conducted on 61 patients with DME who received intravitreal bevacizumab. Pre- and post-injection assessment of visual acuity, fundus autofluorescence (FAF) and SD-OCT imaging was performed. Foveal FAF was graded n a scale of 1 to 4. Better pre-injection visual acuity and intact photoreceptor ISel band and ELM were associated with better visual acuity after treatment. Significantly increased FAF was seen in cystoid macular edema. FAF was 5.6 times more likely to occur in DME with associated ISel band disruption. It was concluded that correlation between FAF with SD-OCT and visual acuity can help predict restoration of photoreceptor ISel band integrity [53].
 
Wakabayashi retrospectively reviewed 41 eyes of 41 patients with DME who underwent vitrectomy. Associations between axial length, post-operative BCVA and visibility of photoreceptor ISel band on SD-OCT were studied. Wakabayashi observed that median axial length was longer in those with a visible photoreceptor ISel band at 12 months after surgery, and that longer axial length predictedbetter visual outcomes after vitrectomy in patients with DME [54]. The mechanism of this protective effect was not fully elucidated, but a similar relationship has been established between axial length and DME, as well as diabetic retinopathy itself [55]. This protective effect is presumed to be secondary to an inverse relationship between axial length and ocular blood flow, a parameter attributed to stretchinduced microvascular anatomical changes, such as vessel thinning, that appear to result in diminished pressure against the vascular wall, and a resultant decrease in microvascular leakage and damage to the retinal microstructure [56].