Clinical and Epidemiologic Research
The Clinical Spectrum of Microcystic Macular Edema Marloes C. Burggraaff,1 Jennifer Trieu,1 Willemien A. E. J. de Vries-Knoppert,1 Lisanne Balk,2 and Axel Petzold2 1Department 2Department
of Ophthalmology, VU University Medical Center, Amsterdam, The Netherlands of Neurology, VU University Medical Center, Amsterdam, The Netherlands
Correspondence: Marloes C. Burggraaff, Department of Ophthalmology, VU University Medical Center Amsterdam, De Boelelaan 1117, PO Box 7057, 1007 MB Amsterdam, The Netherlands; [email protected].
PURPOSE. Microcystic macular edema (MME), originally described in British literature as microcystic macular oedema (MMO), defines microcysts in the inner nuclear layer (INL) of the retina. Microcystic macular edema was described in multiple sclerosis (MS), but can be found in numerous disorders. The presence of MME has important prognostic and therapeutic implications; however, the differential diagnosis is unknown. This study aimed to describe the clinical spectrum of MME.
Submitted: July 25, 2013 Accepted: December 17, 2013
METHODS. A single-center, retrospective cohort study. A bootstrap analysis was performed to reduce the 5865 patients (22,376 scans), who had undergone OCT imaging between January 2010 and February 2013, to a representative dataset. The presence of MME was rated by independent observers.
Citation: Burggraaff MC, Trieu J, de Vries-Knoppert WAEJ, Balk L, Petzold A. The clinical spectrum of microcystic macular edema. Invest Ophthalmol Vis Sci. 2014;55:952–961. DOI:10.1167/iovs.13-12912
RESULTS. The dataset consisted of 1368 patients (mean age 62, range, 4–101 years), 2589 eyes and 6449 scans. Microcystic macular edema was present in 133/1303 (10%) of patients and 0/ 65 (0%) of healthy controls. The interrater agreement for detecting MME was substantial (kappa 0.6) and could be further improved after refining the criteria (kappa 0.8). The clinical spectrum included age-related macular degeneration, epiretinal membranes, postoperative lesions, diabetic retinopathy, vascular occlusion, MS (with/without optic neuritis), optic neuropathy, central serous chorioretinopathy, medication, and miscellaneous causes. The longitudinal pattern of MME was transient (84%) or static. Microcystic macular edema could be associated with an increase or decrease in INL thickness and was predominantly located nasally (48%) and/or temporally (50%). CONCLUSIONS. This study substantially widened the clinical spectrum of MME. Diagnostic criteria were refined and validated. The associated phenotype may imply M¨ uller cell dysfunction within the watershed zone. The longitudinal data and evidence from previous studies suggest follow-up of these patients and their visual function. Keywords: MME, microcysts, pseudocysts, optical coherence tomography, inner nuclear layer thickness, microcystic macular changes
M
icrocystic macular edema (MME) consists of retinal microcysts which are predominantly located in the inner nuclear layer (INL).1,2 The optically empty spaces are more or less square-shaped with at least one of the borders appearing concave or straight. They have no obvious wall; therefore, the term ‘‘pseudocysts’’ is preferred by some authors.3 Recent studies predominantly have focused on MME in multiple sclerosis1,4,5 and neuromyelitis optica (NMO).6,7 It was suggested that MME originated from breakdown of the bloodretinal barrier or focal inflammation.1 However, microcystic changes in the INL were also described in compressive optic neuropathy (due to glioma),8 Leber’s hereditary optic neuropathy, dominant optic atrophy,9 and endemic optic neuropathy,10 suggesting retrograde trans-synaptic degeneration from optic neuropathy as a causative factor for degeneration of the INL with formation of cystic spaces.8 Microcystic macular edema is not restricted to neurologic etiologies and has been described in ARMD, group 2A idiopathic juxtafoveolar retinal telangiectasis, and tamoxifen retinopathy.3,11–14 Following the original description of MME,1 it might be suggested that because of the almost exclusive localization of MME within the poorly vascularized perimacular rim, a
relationship may exist with M¨ uller cell function. Anatomically, M¨ uller cells transverse through the entire retina from the inner limiting membrane down to the basal membrane, with the bulk of their cell bodies residing in the INL. The function of M¨ uller cells are to maintain the neurochemical and water homeostasis of the retina, to support neuronal activity, and to regulate the integrity of the blood-retina barrier. It has been known that edema formation is related to pathologies affecting M¨ uller cell function. In animal models of various retinopathies (retinal ischemia, ocular inflammation, retinal detachment, and diabetes), M¨ uller cells decreased the expression of their major potassium channel Kir4.1, which impaired the rapid water transport across M¨ uller cell membranes and resulted in cellular swelling.15 Interestingly, Kir4.1 is also the dominant potassium channel of the human brain and spinal cord, against which recently autoantibodies were detected, suggesting an acquired channelopathy might be related to the development of MME in MS.2,16,17 Since the clinical spectrum of MME is not yet known, it remains difficult to formulate focused and testable hypotheses on the underlying pathophysiology. Furthermore, the presence of microcysts was associated with a poor long-term functional outcome both in patients with ARMD and patients with MS,
Copyright 2014 The Association for Research in Vision and Ophthalmology, Inc. www.iovs.org j ISSN: 1552-5783
952
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 953 good clinical practice. The patient flow of this retrospective cohort study is summarized in Figure 1. Optical coherence tomography (OCT) scans in this study were derived from patients who underwent spectral-domain OCT (SD-OCT) imaging at the VU University Medical Center in Amsterdam, The Netherlands, between January 1, 2010, and February 15, 2013. During this period, 22,376 macular volume scans were derived from 5865 patients. To reduce the large number of scans and patients to a manageable sample size, bootstrap was used. In total, 6548 macular scans from 1370 patients (2593 eyes, Fig. 1) were randomly selected. Sequential OCT scans were available in 100 patients. These were included to determine if MME was a transient phenomenon or remained static over time.1,4 Because poor quality scans can give rise to artefacts resembling MME, such scans were excluded. Control subjects were included if they had macular volume scans of sufficient quality,18 and no ophthalmologic or neurologic abnormalities. The SD-OCT images were screened by one of the authors (JT) for the presence of MME. As a sensitivity analysis, 3520 (54%) scans were rated by an independent second observer (MCB). The interrater agreement was determined. The reasons for disagreement between the two raters were discussed in a consensus revision of all relevant scans (JT, MCB, WAEJdV, AP). This led to refinement of the diagnostic criteria for MME. These criteria were then reapplied to the dataset. All four authors individually rated the presence of MME in a substantial part of the original data. If at least three raters agreed on the presence of MME, the patient was included in the MME group. Clinical details of these patients were retrieved by a retrospective case note analysis. Because of the retrospective nature of the study, visual acuity (VA) was not measured systematically and the decimal notation was used in this paper.
FIGURE 1. Study design and patient flow.
Optical Coherence Tomography compared with these patients without MME.1,3,4 Therefore, detection of MME might have important prognostic and therapeutic implications. The present study aimed to carefully describe the extent, location, sequential pattern, and clinical spectrum of MME as observed in a mixed neurological and ophthalmological patient population.
METHODS Study Design and Patients This study complied with the tenets of the Declaration of Helsinki and was in accordance with the Dutch guidelines for
Retinal OCT images were obtained with SD-OCT (Spectralis software version 1.1.6.3.; Heidelberg Engineering, Inc., Heidelberg, Germany).
Microcystic Macular Edema Based on previous publications, MME was defined as cystic, lacunar areas of hyporeflectivity with clear boundaries in the INL.1,19 Gelfand et al.1 also suggested to only include those cases where MME was seen on at least two adjacent scans. Because of the retrospective nature of the present study, the distance between two sections (or B-scans) was not standardized, and the number of sections ranged from 10 to 193
FIGURE 2. Two examples of OCT macular volume scans with interrater disagreement on presence of MME. (A) In this patient, there was doubt if the hyporeflective lesion was a cyst or an artefact, and if the lesion was not to close to the foveola, which does not have an INL. (B) The cyst might be in the INL or in the adjacent layer. There might be confluence of the cyst in both layers.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 954
Microcystic Macular Edema TABLE 1. Refined and Validated Diagnostic Criteria for MME Inclusion Criteria
Exclusion Criteria
Configuration Location
Small hyporeflective lesions in the INL Within or outside the perimacular rim
Confluence Coexisting lesions
Permitted within the INL Small cysts in other retinal layers or subretinal fluid is permitted but should be documented as such
Longitudinal pattern
MME fulfilling all of the above criteria needs to be seen on two macular volume scans taken at separate time points, at least 1 week apart Transient/dynamic: MME fulfilling all of the above criteria needs to be seen on at least one macular volume scan, whereas MME is not present in a previous or follow-up volume scan MME fulfilling all of the above criteria needs to be seen on all macular volume scans of one patient, taken at separate time points, at least 1 week apart
Transient/dynamic pattern Static pattern
(median 25). Therefore, patients presenting with MME in only one section/B-scan were included as well. The macular volume scans evaluated in this study represent scans from a routine clinical setting.
Inner Nuclear Layer Segmentation Image post-processing was performed using three different algorithms incorporated in automated segmentation software: first, a purpose build algorithm from Heidelberg Engineering, Inc. (beta software, version 5.9.0.3)20; second, the OCT segmentation and evaluation graphical user interface (OCTSEG) algorithm that allows us to suppress retinal blood vessels and void signal artefacts21; and third, a graph-based multisurface segmentation combining aspects of the two other algorithms with machine learning.22 Of the three algorithms, only the Heidelberg beta software permitted isolated segmentation of the INL. The two other algorithms provided data on a combined INL and outer plexiform layer (OPL) segmentation. For every layer, the OCT software generated a thickness map on a 1-, 3-, and 6-mm grid. After the automated segmentation process, all scans and all layers were carefully
Cysts with a clearly visible cell wall Small cysts directly adjacent to or within the foveola, where there is no INL Confluence with cysts in adjacent retinal layers MME needs to be distinguished from blood vessel artefacts which can give the impression of hyporeflectivity
revised for algorithm failures (LB, AP). Manual correction was performed using the Heidelberg software. All three automated retinal layer segmentation algorithms used in this study were previously shown to be reliable in analyzing the retinas of healthy control subjects20 and patients with drusen.22
Statistical Analysis Statistical analyses were performed using statistical software (SPSS version 17.0; SPSS, Inc., Chicago, IL); programming language (R; R Development Core Team, University of Auckland, Auckland, New Zealand); and analytical software (SAS version 9.3; SAS Institute, Cary, NC). Bootstrap analyses were performed with analytical software (SAS Institute). Kappa statistics were calculated to assess the interrater agreement, Cohen’s kappa for data rated by two raters, and Fleiss’ multirater kappa for data from more than two raters. The level of agreement was rated as slight (0–0.2), fair (0.2–0.4), moderate (0.4–0.6), substantial (0.6–0.8), or almost perfect (0.8–1).23 The overall prevalence of MME was estimated by dividing the number of eyes with MME in the INL, by the total number of eyes scanned.3 Proportions were presented as total
TABLE 2. The Clinical Spectrum of MME Disease* Age-related macular degeneration Epiretinal membrane/vitreomacular traction Iatrogen† Diabetic retinopathy Vascular occlusion Uveitis Multiple sclerosis‡ Optic neuropathy§ Central serous chorioretinopathy Retinitis pigmentosa Medicationjj Other/unknown Healthy subjects
n (%) 36 25 27 9 9 9 4 1 3 1 2 7
(27.1) (18.8) (20.3) (6.8) (6.8) (6.8) (3.0) (0.8) (2.3) (0.8) (1.5) (5.3) 65
Age (y) 84.1 72.7 72.7 64.6 75.2 57.8 57.8
6 6 6 6 6 6 6
51.0 6 53.5 6 72.6 6 50.5 6
7.0 (61–95) 9.0 (52–93) 10.6 (50–88) 10.0 (44–80) 9.0 (64–91) 18.8 (23–85) 7.8 (51–69) 78.0 17.0 (34–68) 30.0 6.4 (49–58) 13.8 (52–85) 7 (28–63)
Sex, % F 69.4 48.0 55.6 33.3 55.6 77.8 50.0 0.0 (1 male) 0.0 (3 male) 0.0 (1 male) 100.0 71.4 65.1
The age (mean 6 SD), sex (% female), and total number (%) of patients with MME in each disease category, as well as the age and sex of the healthy control subjects (no MME), are presented. N/A, not applicable. * Cause of MME according to the patient’s ophthalmologist, extracted from the patient’s chart. † Following vitrectomy (n ¼ 16), cataract extraction (n ¼ 8), and intraocular lens exchange (n ¼ 3). ‡ In three cases MME was observed in the eye with a previous episode of optic neuritis. In one case MME was observed in a 52-year-old male patient without a history of optic neuritis, but with severe atrophy of the brain involving the optic pathways. § Posterior ischemic optic neuropathy (PION). jj Hydroxychloroquine.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 955
FIGURE 3. An example of a normal retina compared with the retina of a patient with MME. (A) A normal retina from a healthy control subject. The pseudocolored surface image is the infrared surface photo; the vertical stacked gray image an OCT B-scan. (B) Optical coherence tomography image taken from a 76-year-old woman with a 3-year history of ARMD, for which she received regular injections with ranibizumab and bevacizumab. At time of imaging, her best corrected VA OS was 0.7. (C) Illustrates layer segmentation for the OCT B-scan shown in (A). The INL is bordered by the blue and orange lines. (D) The same for the patient with ARMD. Clearly, there is focal thickening of the INL were MME is observed. (E) Pan-retinal thickness map from the healthy control subject and the corresponding INL thickness map (F). (G) In the patient with ARMD, there is overall thinning of the retina. (H) There is marked heterogeneity of the INL thickness map in the ARMD patient compared with the healthy control subject.
number (%). Parametric tests (t-tests) were used for Gaussian data. Categorical data was analyzed using the v2 test. Statistical significance was accepted for two-sided testing with P < 0.05.
RESULTS Patients A total of 22,376 macular volume scans were obtained from 5865 patients between January 1, 2010, and February 15, 2013. After bootstrap, a representative dataset of 6548 macular scans from 1370 patients (2593 eyes) was subject to in-depth analyses; 99 scans of 13 patients failed quality control criteria. Two patients were excluded from further analyses because all their scans were of poor quality. Therefore, the final dataset consisted of 6449 scans from 1368 patients (2589 eyes; see Fig. 1).
revealed that interrater disagreement was due to: (1) size/ configuration—could the cysts be identified as MME/pseudocysts or did they have cell walls and could be classified as cystic macular edema? (2) location: presence of cysts within or just outside the INL; (3) confluence of cysts with cyst in other retinal layers; (4) coexisting lesions in other layers; and (5) the longitudinal pattern of presumed MME (i.e., the presence of cysts could be transient or static if macular volume scans of different dates were available). Representative examples of OCT scans which caused interrater disagreement are illustrated in Figure 2. Taking the reasons for disagreement into account, we refined the diagnostic criteria for MME as summarized in Table 1. Based on the refined diagnostic criteria for MME, the overall interrater agreement improved to kappa 0.8. Of note, two patients who were thought to have MME during the first rating were unanimously excluded using our refined diagnostic criteria.
Validation of Diagnostic Criteria for MME The screening set was split into two parts. In the first part, the observed percentage of agreement between the two raters was 92%. The interrater agreement for identifying MME was substantial (kappa 0.6). The consensus revision of the scans
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Clinical Spectrum of MME In 133/1303 (10.2%) of patients from our cohort, MME was present according to the refined diagnostic criteria (Table 1). There was a slight female predominance (57.1%) in the MME
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 956
FIGURE 4. Representative OCT images of patients with MME from the clinical spectrum are shown. The infrared surface photo and OCT image are presented to the left and the manually segmented INL to the right. (A) Microcystic macular edema in the right eye of a 70-year-old male patient with a history of proliferative diabetic retinopathy treated with panretinal photocoagulation (VA OD 0.3). (B) Optical coherence tomography image showing MME 8 months after occlusion of the vena temporalis superior OD in a 66-year-old female patient (VA OD 0.7). Microcystic macular edema was located in the temporal superior quadrant of the inner 3-mm EDTRS grid. However, INL thickening extended to the periphery. In addition, hyperreflective spots were observed in all inner retinal layers. (C) Microcystic macular edema in a 66-year-old female patient with sarcoidosis and posterior uveitis in the right
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 957
eye. Fluorescein angiography showed an ischemic maculopathy (VA OD 1.0). Serum ACE was 29, CT-thorax showed mediastinal and hilar lymphadenopathy. In this case, hyperreflective spots were restricted to the retinal nerve fiber layer, ganglion cell layer, and INL. (D) Microcystic macular edema in a 58-year-old female patient with a newly diagnosed pucker in the right eye (VA OD 0.2). Again, there were multiple hyperreflective spots in the inner retinal layers. (E) Microcystic macular edema 3 months after vitrectomy, in a 71-year-old female patient with a retinal detachment in the left eye (VA OS 0.05).
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 958
Microcystic Macular Edema
FIGURE 5. Microcystic macular edema in a patient with multiple sclerosis and a history of optic neuritis and branch retinal vein occlusion in the right eye. This 55-year-old female patient was diagnosed with clinical definite MS in 1992. Magnetic resonance imaging showed multiple periventricular brain lesions, and spinal T2 hyperintense lesions. In 2001, she experienced one episode of optic neuritis OD; coincidentally, an occlusion of the vena temporalis superior was found (VA 0.4). This case illustrates coexistence of MME with both a thinned (green box) and thickened (blue box) aspect of the INL. (A) The OCT B-scan clearly shows MME in the INL, which is thickened in the center and thinned near the peripheral parts of the image. The OCT image was taken 11 years after conduction of the fundus photograph in 2001, which showed superficial intraretinal hemorrhages inline with occlusion of the vena temporalis superior (B). There was thinning of the nerve fiber layer of the right optic disc, with normal appearance on the left (not shown). (C) The corresponding fluorescein angiography does not reveal leakage. Furthermore, there seemed to be a preocclusion of the vena temporalis inferior. (D) The thickness map of the INL showed more prominent thickening along the course of the inferior temporal vein, with relative thinning along the course of the superior temporal vein. (E) The destruction of the INL related to the thickness difference can be best appreciated by comparing the temporal part of the inferior retina (blue box) with the corresponding area shown on top of the image (green box). Interestingly, hyperreflective spots of the inner retinal layers were found, which were also seen in other entities of the clinical spectrum of MME.
group. Patients with MME were significantly older (72.8 years, SD 13.8, range, 23–95 years) compared with patients without MME (63.2 years, SD 17.8, range, 4–101 years; P ¼ 0.002), and healthy control subjects (50.5 years, SD 7.2, range, 28–63 years; P < 0.001). The clinical spectrum and demographic data of patients with MME is summarized in Table 2. Most frequently, MME was observed in patients with ARMD (27.1%) followed by patients with preceding ophthalmic surgery (20.3%) or presence of an epiretinal membrane (18.8%). Furthermore, MME was identified following a retinal vascular occlusion (6.8%), uveitis (6.8%), diabetic retinopathy (6.8%), MS (3.0%), posterior ischemic optic neuropathy (0.8%), and miscellaneous/other retinal conditions (5.3%). Three of the four patients with MS had a history of optic neuritis in the affected eye. Importantly, the patient who never suffered from optic neuritis had a long disease duration (14 years), was severely disabled (EDSS 6.0), showed severe white and gray matter atrophy involving the optic pathways on magnetic resonance imaging (MRI), and had bilaterally delayed visual evoked potentials (this patient never received fingolimod). Figure 3 shows an example of a normal retina of a healthy control subject compared with the retina of a patient with MME based on ARMD. Other examples of OCT images of patients found in the different disease categories are presented in Figure 4. Figure 5 shows an example of a patient presenting both with an optic neuritis and a branch retinal vein occlusion at the same time (MME was seen 11 years after onset).
Bilateral MME Bilateral MME was present in 9/133 (6.8%) of patients. The diagnosis was ARMD in four, an epiretinal membrane in three,
iatrogenic (postoperative) in one, and central serous chorioretinopathy in one.
Longitudinal Pattern of MME Longitudinal data was available from 100 out of 133 patients with MME (75.2%). Microcystic macular edema was transient in 84% of patients. It remained static for the available observation period (mean 233 days, SD 157 days; range, 46–494 days) in the remaining 16% of patients. Of note, 42% of patients with a transient pattern of MME developed cystic macular edema (n ¼ 35).
Localization of MME The perimacular rim was affected in all cases with MME, but the localization could be within any quadrant or combination of quadrants of the superimposed Early Treatment Diabetic Retinopathy Study (EDTRS) grid (Table 3). Most frequently a single quadrant was affected (67.7%). Two quadrants were affected in 15.0% of patients, three quadrants in 9.0%, and all four quadrants in 8.3% (Fig. 6). The nasal (48.1%) and temporal (49.6%) quadrants were more frequently involved compared with the superior (29.3%) and inferior (30.1%) quadrants.
MME and INL Thickness The failure rate of automated segmentation of the INL and other retinal layers was above 90% of scans from patients with MME for all three algorithms used. Therefore, no statistical analyses were performed on this data. Manual segmentation was performed in scans representative for the main disease
TABLE 3. Location of MME With Respect to the EDTRS Grid Disease Age-related macular degeneration Epiretinal membrane/vitreomacular traction Iatrogenic Diabetic retinopathy Vascular occlusion Uveitis Multiple sclerosis Optic neuropathy Central serous chorioretinopathy Retinitis pigmentosa Medication Other/unknown Total
Superior, n (%) 6 12 7 3 2 5 2 1 1 0 0 1 39
(16.7) (18.0) (25.9) (50.0) (22.0) (55.6) (50.0) (100) (25.0) (0.0) (0.0) (14.3) (29.3)
Inferior, n (%) 12 11 6 1 0 4 2 0 0 0 1 2 40
(33.3) (44.0) (22.2) (20.0) (0.0) (44.4) (50.0) (0.0) (0.0) (0.0) (50.0) (28.6) (30.1)
Nasal, n (%) 19 12 15 4 2 4 2 0 3 0 0 2 64
(52.8) (48.0) (55.6) (44.4) (22.2) (44.4) (50.0) (0.0) (75.0) (0.0) (0.0) (28.6) (48.1)
Temporal, n (%) 14 15 14 2 5 7 2 0 0 1 1 4 66
(38.9) (60.0) (51.9) (40.0) (56.0) (77.8) (50.0) (0.0) (0.0) (100) (50.0) (57.1) (49.6)
Localization in the nasal and temporal quadrants dominated in age-related macular degeneration and iatrogenic conditions. For all other conditions MME was more equally distributed within the perimacular rim. In a proportion of patients, MME was observed in more than one quadrant, explaining why the percentages for each entity can add up to more than 100%. This is further illustrated by the proportional distribution of MME in Figure 3.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 959
Microcystic Macular Edema
FIGURE 6. Sectoral localization and perimacular rim extension of MME. One quadrant was affected in the vast majority of cases (67.7%). Within this group, MME was most often found in the temporal and nasal quadrants. In a small proportion of patients, MME was extensive and lesions were found in two (15.0%), three (9.0%), or four quadrants (8.3%).
categories. Noteworthy, the INL thickness was increased in most, but not all areas with MME (Fig. 3).
DISCUSSION In this large, retrospective, single-center study, the prevalence of MME was 10.2% (n ¼ 133). To put our data into context, thus far only approximately 127 patients with MME have been reported in literature. The present study more than doubled this number, and substantially widened the clinical spectrum of MME (Table 4). Furthermore, the diagnostic criteria for MME were refined such that a substantial level of agreement (kappa 0.8) could be achieved. Microcystic changes were most frequently observed in ARMD, postoperatively after vitrectomy, cataract extraction, or intraocular lens exchanges, and in epiretinal membranes or
vitreomacular traction. The percentage of patients with MME due to MS (optic neuritis) in our study was relatively small, which might be due to our study population (i.e., a random sample of all patients receiving an OCT scan at an ophthalmologic department). In the present study, MME occurred in a large number of underlying disorders, which is in line with the uller cells become activated finding of Bringmann et al.24 that M¨ upon virtually all pathogenic stimuli. Therefore, our study supports M¨ uller cell dysfunction as part of the pathophysiology of MME. This interpretation is in line with the literature.5–7 Interestingly, MME was most often found in the nasal and temporal quadrants of the perimacular rim. It can be hypothesized that the end-arterial nature of the choroidal vasculature and the existence of watershed zones in the choroid may have a role in the location of MME. Watershed zones are borderline areas between the territories of distribu-
TABLE 4. Number of Patients (Eyes) With MME Found in Literature Disease ARMD Retinal telangiectasis Tamoxifen MS
Relapsing isolated optic neuropathy Chronic relapsing inflammatory optic neuropathy NMO Leber’s hereditary optic neuropathy/dominant optic atrophy Endemic optic neuropathy Optic nerve glioma Retrograde optic neuropathy (compressive, n ¼ 9 eyes; vascular, n ¼ 3; hereditary, n ¼ 2; not clear, n ¼ 2) Autosomal dominant optic atrophy Autosomal dominant optic atrophy Total cases of MME* This study
MME Cases
Reference(s)
22 eyes of 22 patients 24 eyes 20 eyes (of 10 patients) 1 eye of 1 patient 20 eyes of 15 patients 12 eyes of 10 patients 1 eye of 1 patient 1 eye of 1 patient 1 patient 10 patients 7 eyes of 5 patients 10 patients 25 eyes of 16 patients 2 eyes of 1 patient
Querques3 Cohen11 Gaudric12 Park33 Gelfand1 Saidha4 Balk5 Balk5 Petzold and Plant34 Sotirchos7 Gelfand1 Barboni9 Shalchi10 Abegg8
16 eyes of 9 patients 4 eyes of 2 patients 2 eyes of 1 patient Approximately 178 eyes of 127 patients 142 eyes of 133 patients
Abegg30 Gocho31 Lujan32
Patients are categorized by disease entity. * In some of the publications, the number of patients and/or the number of eyes was not well specified. Therefore, the total number of eyes and patients was estimated based on the available information.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 960
Microcystic Macular Edema tion of two end-arteries, and these areas with a relatively poor blood flow are most vulnerable to hypoxia-ischemia.25,26 The main watershed zone of the choroid is located between the nasal edge of the optic disc and the fovea and represents the area between territories supplied by the temporal and nasal posterior ciliary arteries.27 Furthermore, hyperreflective lesions were anecdotally observed in some scans of patients presenting with MME. They were distributed throughout all retinal layers. Similar foci were found in patients with diabetic macular edema, which were thought to represent extravasated (lipo)proteins preceding hard exudates or blood-retina barrier breakdown.28 In patients with ARMD, such lesions were thought to represent activated microglial cells from early biological inflammatory reactions (resolution was associated with a better visual outcome).29 A particularly salient limitation of the study was that, because the SD-OCT scans were acquired without a systematic protocol, the macular volume scans consisted of different rasters with a median of 25 sections. Consequently, resolution was limited and cases with MME in between the sections may have been missed.3 In addition, it proved to be difficult to identify MME. In previous studies, problems with the identification of MME were not reported. To our knowledge, there are only few criteria applying to MME. Microcystic macular edema was defined as lacunar areas of hyporeflectivity with clear boundaries and absence of a clearly visible cell wall, in the INL, present on at least two adjacent scans.1,3,19 These criteria were used in the first part of the screening-set, by two independent raters. Of note, patients presenting with MME on only one of the sections of the macular volume scan were also included because of the limited number of sections. The interrater agreement was kappa 0.6, which can be considered substantial.23 However, including almost exclusively patients (and just a few healthy controls) with most OCT scans showing absence of MME (n ¼ 2937) might have caused the interrater agreement to be somewhat smaller than expected (whereas the observer agreement was 0.9). Clearly, when layers become difficult to separate from each other (e.g., due to poor contrast) or presence of (pseudo)cysts in adjacent layers, it can be expected that difficulties arise with localizing MME. Therefore, the present study refined the criteria for MME such that they considered abnormalities found in the INL and adjacent layers; differences in the configuration and location of cysts; confluence of cysts with cysts in other retinal layers; the presence of coexisting lesions; artefacts; and the heterogeneous longitudinal pattern. Despite the good interrater agreement (kappa 0.8), there might be other problems in rating MME that we did not account for. Therefore, we would be keen to learn how these refined criteria perform in other hands. Furthermore, we did not consider degenerative pseudocysts located just below the internal limiting membrane (50%), in the outer nuclear layer (36%), or in all retinal layers (27%), as described by Querques et al.3 for patients with ARMD. Finally, it needs to be discussed that current automated retinal layer segmentation was not feasible in patients with MME. Three different algorithms were used, which were reliable analyzing retinas of healthy control subjects and patients with drusen.20–22 The first limitation to mention was the time necessary to segment the retinal layers, which took 2 to 3 weeks for each algorithm. Second, two of the algorithms did not permit a separate segmentation of the INL.21,22 Third, the algorithm failure rate of accurate layer segmentation in eyes with MME was almost 100%, requiring extremely timeconsuming manual corrections. The automated layer segmentation could not be completed for those eyes in which the primary pathology affected the structure of the layers, resulting
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
in a change of contrast of the layers. However, local pathology (such as highly reflective spots and shadow artefacts) also led to difficulty segmenting the layers. Therefore, retinal pathology proved to be a severe limitation to the routine application of currently available retinal layer segmentation software. It has been proposed that machine learning might help to overcome this problem.22 Acknowledging the failure to provide quantitative data on INL thickness, the qualitative revision of the scans (manually corrected INL thickness maps) presented in this study suggested that it might not be appropriate to pool data from the four sectors composing the perimacular rim. Areas with visible thinning coexisted to areas with visible thickening. A more sophisticated and detailed analysis of the INL thickness may need to be performed in future studies, requiring segmentation software reliable in pathological conditions and statistical models feasible for refined retinal area surface analyses. We would be keen to hear of such efforts and discuss whether the data compiled in this study, an anonymous dataset of MME patients, may contribute to advancing the field. Further research is needed to determine the clinical significance of MME. Because MME was not present in healthy controls, further examination is suggested when microcysts are present on OCT images in (new) patients presenting in a (neuro)ophthalmologic service. Especially since imaging of the central nervous system might be needed in some cases (4% of patients with MME in the present study suffered from a neurologic condition). Furthermore, the presence of microcysts may negatively predict long-term functional outcome both in patients with ARMD and patients with MS.1,3,4 These studies showed that visual acuity decreased when microcysts were present. On the other hand, a relation between the presence of MME, visual acuity and overall disease disability was not found in patients with NMO. However, the sample size of the latter study might have been too small for determining such associations.6 In the present study, visual function was not systematically investigated. Follow-up of patients with MME in a neuro-ophthalmological service may be advised to better understand the potential prognostic and therapeutical implications of this new clinical sign.
Acknowledgments Disclosure: M.C. Burggraaff, None; J. Trieu, None; W.A.E.J. de Vries-Knoppert, None; L. Balk, None; A. Petzold, None
References 1. Gelfand JM, Nolan R, Schwartz DM, Graves J, Green AJ. Microcystic macular oedema in multiple sclerosis is associated with disease severity. Brain. 2012;135:1786–1793. 2. Petzold A. Microcystic macular oedema in MS: T2 lesion or black hole? Lancet Neurol. 2012;11:933–934. 3. Querques G, Coscas F, Forte R, Massamba N, Sterkers M, Souied EH. Cystoid macular degeneration in exudative agerelated macular degeneration. Am J Ophthalmol. 2011;152: 100–107. 4. Saidha S, Sotirchos ES, Ibrahim MA, et al. Microcystic macular oedema, thickness of the inner nuclear layer of the retina, and disease characteristics in multiple sclerosis: a retrospective study. Lancet Neurol. 2012;11:963–972. 5. Balk LJ, Killestein J, Polman CH, Uitdehaag BMJ, Petzold A. Microcystic macular oedema confirmed, but not specific for multiple sclerosis. Brain. 2012;135:e226. 6. Gelfand JM, Cree BA, Nolan R, Arnow S, Green AJ. Microcystic inner nuclear layer abnormalities and neuromyelitis optica. JAMA Neurol. 2013;70:629–633.
IOVS j February 2014 j Vol. 55 j No. 2 j 961
Microcystic Macular Edema 7. Sotirchos E, Saidha S, Byraiah G, et al. In vivo identification of morphologic retinal abnormalities in neuromyelitis optica. Neurology. 2013;80:1406–1414. 8. Abegg M, Zinkernagel M, Wolf S. Microcystic macular degeneration from optic neuropathy. Brain. 2012;135:e225. 9. Barboni P, Carelli V, Savini G, Carbonelli M, La Morgia C, Sadun A. Microcystic macular degeneration from optic neuropathy: not inflammatory, not trans-synaptic degeneration. Brain. 2013;136(pt 7):e329. 10. Kisimbi J, Shalchi Z, Mahroo OA. Macular spectral domain optical coherence tomography findings in Tanzanian endemic optic neuropathy. Brain. 2013;136(pt 11):3418–3426. 11. Cohen SY, Dubois L, Nghiem-Buffet S, et al. Retinal pseudocysts in age-related geographic atrophy. Am J Ophthalmol. 2010;150:211–217. 12. Gaudric A, Ducos de Lahitte G, Cohen SY, Massin P, Haouchine B. Optical coherence tomography in group 2A idiopathic juxtafoveolar retinal telangiectasis. Arch Ophthalmol. 2006; 124:1410–1419. 13. Cohen SM, Cohen ML, El-Jabali F, Pautler SE. Optical coherence tomography findings in nonproliferative group 2a idiopathic juxtafoveal retinal telangiectasis. Retina. 2007;27:59–66. 14. Gualino V, Cohen SY, Delyfer MN, Sahel JA, Gaudric A. Optical coherence tomography findings in tamoxifen retinopathy. Am J Ophthalmol. 2005;140:757–758. 15. Reichenbach A, Wurm A, Pannicke T, Iandiev I, Wiedemann P, Bringmann A. Muller cells as players in retinal degeneration and edema. Graefes Arch Clin Exp Ophthalmol. 2007;245: 627–636. 16. Srivastava R, Aslam M, Kalluri SR, et al. Potassium channel KIR4.1 as an immune target in multiple sclerosis. N Engl J Med. 2012;367:115–123. 17. Udagawa T, Tatsumi N, Tachibana T, et al. Inwardly rectifying potassium channel Kir4.1 is localized at the calyx endings of vestibular afferents. Neuroscience. 2012;215:209–216. 18. Tewarie P, Balk L, Costello F, et al. The OSCAR-IB consensus criteria for retinal OCT quality assessment. PLoS One. 2012;7: e34823. 19. Brar M, Yuson R, Kozak I, et al. Correlation between morphologic features on spectral-domain optical coherence tomography and angiographic leakage patterns in macular edema. Retina. 2010;30:383–389. 20. Traber G, Oberwahrenbrock T, Gabilondo I, Nolan R, Songster C, Balk L. Multicenter inter-rater reliability of retinal layer segmentation using spectral-domain OCT. Poster presented at: The North American Neuro-Ophthalmology Society (NANOS)
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
21.
22.
23. 24. 25. 26.
27.
28.
29.
30.
31.
32.
33.
34.
Annual Meeting Neuro-Ophthalmology Collection; February 12, 2013; Snowbird, UT. Mayer M, Hornegger J, Mardin C, Tornow R. Retinal nerve fiber layer segmentation on FD-OCT scans of normal subjects and glaucoma patients. Biomed Opt Express. 2010;1:1358–1383. Dufour PA, Ceklic L, Abdillahi H, et al. Graph-based multisurface segmentation of OCT data using trained hard and soft constraints. IEEE Trans Med Imaging. 2013;32:531–543. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. Bringmann A, Pannicke T, Grosche J, et al. M¨ uller cells in the healthy and diseased retina. Prog Retin Eye Res. 2006;25:397–424. Hayreh SS. In vivo choroidal circulation and its watershed zones. Eye. 1990;4:273–289. Hayreh SS. Submacular choroidal vascular bed watershed zones and their clinical importance. Am J Ophthalmol. 2010; 150:940–941. Giuffre G. Main posterior watershed zone of the choroid. Variations of its position in normal subjects. Doc Ophthalmol. 1989;72:175–180. Bolz M, Schmidt-Erfurth U, Deak G, Mylonas G, Kriechbaum K, Scholda C. Optical coherence tomographic hyperreflective foci: a morphologic sign of lipid extravasation in diabetic macular edema. Ophthalmology. 2009;116:914–920. Coscas G, De Benedetto U, Coscas F, et al. Hyperreflective dots: a new spectral-domain optical coherence tomography entity for follow-up and prognosis in exudative age-related macular degeneration. Ophthalmologica. 2013;229:32–37. Abegg M, Dysli M, Wolf S, Kowal J, Dufour P, Zinkernagel M. Microcystic macular edema, retrograde maculopathy caused by optic neuropathy. Ophthalmology. 2014;121:142–149. Gocho K, Kikuchi S, Kabuto T, et al. High-resolution en face images of microcystic macular edema in patients with autosomal dominant optic atrophy. Biomed Res Int. 2013; 2013:676803. Lujan B, Horton JC. Microcysts in the inner nuclear layer from optic atrophy are caused by retrograde trans-synaptic degeneration combined with vitreous traction on the retinal surface. Brain. 2013;136:1–4. Park SS, Zawadzki RJ, Truong SN, Choi SS, Werner JS. Microcystoid maculopathy associated with tamoxifen use diagnosed by high-resolution Fourier-domain optical coherence tomography. Retin Cases Brief Rep. 2009;3:33–35. Petzold A, Plant GT. Chronic relapsing inflammatory optic neuropathy: a systematic review of 122 cases reported. J Neurol. 2014;261:17–26.
The Clinical Spectrum of Microcystic Macular Edema Marloes C. Burggraaff,1 Jennifer Trieu,1 Willemien A. E. J. de Vries-Knoppert,1 Lisanne Balk,2 and Axel Petzold2 1Department 2Department
of Ophthalmology, VU University Medical Center, Amsterdam, The Netherlands of Neurology, VU University Medical Center, Amsterdam, The Netherlands
Correspondence: Marloes C. Burggraaff, Department of Ophthalmology, VU University Medical Center Amsterdam, De Boelelaan 1117, PO Box 7057, 1007 MB Amsterdam, The Netherlands; [email protected].
PURPOSE. Microcystic macular edema (MME), originally described in British literature as microcystic macular oedema (MMO), defines microcysts in the inner nuclear layer (INL) of the retina. Microcystic macular edema was described in multiple sclerosis (MS), but can be found in numerous disorders. The presence of MME has important prognostic and therapeutic implications; however, the differential diagnosis is unknown. This study aimed to describe the clinical spectrum of MME.
Submitted: July 25, 2013 Accepted: December 17, 2013
METHODS. A single-center, retrospective cohort study. A bootstrap analysis was performed to reduce the 5865 patients (22,376 scans), who had undergone OCT imaging between January 2010 and February 2013, to a representative dataset. The presence of MME was rated by independent observers.
Citation: Burggraaff MC, Trieu J, de Vries-Knoppert WAEJ, Balk L, Petzold A. The clinical spectrum of microcystic macular edema. Invest Ophthalmol Vis Sci. 2014;55:952–961. DOI:10.1167/iovs.13-12912
RESULTS. The dataset consisted of 1368 patients (mean age 62, range, 4–101 years), 2589 eyes and 6449 scans. Microcystic macular edema was present in 133/1303 (10%) of patients and 0/ 65 (0%) of healthy controls. The interrater agreement for detecting MME was substantial (kappa 0.6) and could be further improved after refining the criteria (kappa 0.8). The clinical spectrum included age-related macular degeneration, epiretinal membranes, postoperative lesions, diabetic retinopathy, vascular occlusion, MS (with/without optic neuritis), optic neuropathy, central serous chorioretinopathy, medication, and miscellaneous causes. The longitudinal pattern of MME was transient (84%) or static. Microcystic macular edema could be associated with an increase or decrease in INL thickness and was predominantly located nasally (48%) and/or temporally (50%). CONCLUSIONS. This study substantially widened the clinical spectrum of MME. Diagnostic criteria were refined and validated. The associated phenotype may imply M¨ uller cell dysfunction within the watershed zone. The longitudinal data and evidence from previous studies suggest follow-up of these patients and their visual function. Keywords: MME, microcysts, pseudocysts, optical coherence tomography, inner nuclear layer thickness, microcystic macular changes
M
icrocystic macular edema (MME) consists of retinal microcysts which are predominantly located in the inner nuclear layer (INL).1,2 The optically empty spaces are more or less square-shaped with at least one of the borders appearing concave or straight. They have no obvious wall; therefore, the term ‘‘pseudocysts’’ is preferred by some authors.3 Recent studies predominantly have focused on MME in multiple sclerosis1,4,5 and neuromyelitis optica (NMO).6,7 It was suggested that MME originated from breakdown of the bloodretinal barrier or focal inflammation.1 However, microcystic changes in the INL were also described in compressive optic neuropathy (due to glioma),8 Leber’s hereditary optic neuropathy, dominant optic atrophy,9 and endemic optic neuropathy,10 suggesting retrograde trans-synaptic degeneration from optic neuropathy as a causative factor for degeneration of the INL with formation of cystic spaces.8 Microcystic macular edema is not restricted to neurologic etiologies and has been described in ARMD, group 2A idiopathic juxtafoveolar retinal telangiectasis, and tamoxifen retinopathy.3,11–14 Following the original description of MME,1 it might be suggested that because of the almost exclusive localization of MME within the poorly vascularized perimacular rim, a
relationship may exist with M¨ uller cell function. Anatomically, M¨ uller cells transverse through the entire retina from the inner limiting membrane down to the basal membrane, with the bulk of their cell bodies residing in the INL. The function of M¨ uller cells are to maintain the neurochemical and water homeostasis of the retina, to support neuronal activity, and to regulate the integrity of the blood-retina barrier. It has been known that edema formation is related to pathologies affecting M¨ uller cell function. In animal models of various retinopathies (retinal ischemia, ocular inflammation, retinal detachment, and diabetes), M¨ uller cells decreased the expression of their major potassium channel Kir4.1, which impaired the rapid water transport across M¨ uller cell membranes and resulted in cellular swelling.15 Interestingly, Kir4.1 is also the dominant potassium channel of the human brain and spinal cord, against which recently autoantibodies were detected, suggesting an acquired channelopathy might be related to the development of MME in MS.2,16,17 Since the clinical spectrum of MME is not yet known, it remains difficult to formulate focused and testable hypotheses on the underlying pathophysiology. Furthermore, the presence of microcysts was associated with a poor long-term functional outcome both in patients with ARMD and patients with MS,
Copyright 2014 The Association for Research in Vision and Ophthalmology, Inc. www.iovs.org j ISSN: 1552-5783
952
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 953 good clinical practice. The patient flow of this retrospective cohort study is summarized in Figure 1. Optical coherence tomography (OCT) scans in this study were derived from patients who underwent spectral-domain OCT (SD-OCT) imaging at the VU University Medical Center in Amsterdam, The Netherlands, between January 1, 2010, and February 15, 2013. During this period, 22,376 macular volume scans were derived from 5865 patients. To reduce the large number of scans and patients to a manageable sample size, bootstrap was used. In total, 6548 macular scans from 1370 patients (2593 eyes, Fig. 1) were randomly selected. Sequential OCT scans were available in 100 patients. These were included to determine if MME was a transient phenomenon or remained static over time.1,4 Because poor quality scans can give rise to artefacts resembling MME, such scans were excluded. Control subjects were included if they had macular volume scans of sufficient quality,18 and no ophthalmologic or neurologic abnormalities. The SD-OCT images were screened by one of the authors (JT) for the presence of MME. As a sensitivity analysis, 3520 (54%) scans were rated by an independent second observer (MCB). The interrater agreement was determined. The reasons for disagreement between the two raters were discussed in a consensus revision of all relevant scans (JT, MCB, WAEJdV, AP). This led to refinement of the diagnostic criteria for MME. These criteria were then reapplied to the dataset. All four authors individually rated the presence of MME in a substantial part of the original data. If at least three raters agreed on the presence of MME, the patient was included in the MME group. Clinical details of these patients were retrieved by a retrospective case note analysis. Because of the retrospective nature of the study, visual acuity (VA) was not measured systematically and the decimal notation was used in this paper.
FIGURE 1. Study design and patient flow.
Optical Coherence Tomography compared with these patients without MME.1,3,4 Therefore, detection of MME might have important prognostic and therapeutic implications. The present study aimed to carefully describe the extent, location, sequential pattern, and clinical spectrum of MME as observed in a mixed neurological and ophthalmological patient population.
METHODS Study Design and Patients This study complied with the tenets of the Declaration of Helsinki and was in accordance with the Dutch guidelines for
Retinal OCT images were obtained with SD-OCT (Spectralis software version 1.1.6.3.; Heidelberg Engineering, Inc., Heidelberg, Germany).
Microcystic Macular Edema Based on previous publications, MME was defined as cystic, lacunar areas of hyporeflectivity with clear boundaries in the INL.1,19 Gelfand et al.1 also suggested to only include those cases where MME was seen on at least two adjacent scans. Because of the retrospective nature of the present study, the distance between two sections (or B-scans) was not standardized, and the number of sections ranged from 10 to 193
FIGURE 2. Two examples of OCT macular volume scans with interrater disagreement on presence of MME. (A) In this patient, there was doubt if the hyporeflective lesion was a cyst or an artefact, and if the lesion was not to close to the foveola, which does not have an INL. (B) The cyst might be in the INL or in the adjacent layer. There might be confluence of the cyst in both layers.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 954
Microcystic Macular Edema TABLE 1. Refined and Validated Diagnostic Criteria for MME Inclusion Criteria
Exclusion Criteria
Configuration Location
Small hyporeflective lesions in the INL Within or outside the perimacular rim
Confluence Coexisting lesions
Permitted within the INL Small cysts in other retinal layers or subretinal fluid is permitted but should be documented as such
Longitudinal pattern
MME fulfilling all of the above criteria needs to be seen on two macular volume scans taken at separate time points, at least 1 week apart Transient/dynamic: MME fulfilling all of the above criteria needs to be seen on at least one macular volume scan, whereas MME is not present in a previous or follow-up volume scan MME fulfilling all of the above criteria needs to be seen on all macular volume scans of one patient, taken at separate time points, at least 1 week apart
Transient/dynamic pattern Static pattern
(median 25). Therefore, patients presenting with MME in only one section/B-scan were included as well. The macular volume scans evaluated in this study represent scans from a routine clinical setting.
Inner Nuclear Layer Segmentation Image post-processing was performed using three different algorithms incorporated in automated segmentation software: first, a purpose build algorithm from Heidelberg Engineering, Inc. (beta software, version 5.9.0.3)20; second, the OCT segmentation and evaluation graphical user interface (OCTSEG) algorithm that allows us to suppress retinal blood vessels and void signal artefacts21; and third, a graph-based multisurface segmentation combining aspects of the two other algorithms with machine learning.22 Of the three algorithms, only the Heidelberg beta software permitted isolated segmentation of the INL. The two other algorithms provided data on a combined INL and outer plexiform layer (OPL) segmentation. For every layer, the OCT software generated a thickness map on a 1-, 3-, and 6-mm grid. After the automated segmentation process, all scans and all layers were carefully
Cysts with a clearly visible cell wall Small cysts directly adjacent to or within the foveola, where there is no INL Confluence with cysts in adjacent retinal layers MME needs to be distinguished from blood vessel artefacts which can give the impression of hyporeflectivity
revised for algorithm failures (LB, AP). Manual correction was performed using the Heidelberg software. All three automated retinal layer segmentation algorithms used in this study were previously shown to be reliable in analyzing the retinas of healthy control subjects20 and patients with drusen.22
Statistical Analysis Statistical analyses were performed using statistical software (SPSS version 17.0; SPSS, Inc., Chicago, IL); programming language (R; R Development Core Team, University of Auckland, Auckland, New Zealand); and analytical software (SAS version 9.3; SAS Institute, Cary, NC). Bootstrap analyses were performed with analytical software (SAS Institute). Kappa statistics were calculated to assess the interrater agreement, Cohen’s kappa for data rated by two raters, and Fleiss’ multirater kappa for data from more than two raters. The level of agreement was rated as slight (0–0.2), fair (0.2–0.4), moderate (0.4–0.6), substantial (0.6–0.8), or almost perfect (0.8–1).23 The overall prevalence of MME was estimated by dividing the number of eyes with MME in the INL, by the total number of eyes scanned.3 Proportions were presented as total
TABLE 2. The Clinical Spectrum of MME Disease* Age-related macular degeneration Epiretinal membrane/vitreomacular traction Iatrogen† Diabetic retinopathy Vascular occlusion Uveitis Multiple sclerosis‡ Optic neuropathy§ Central serous chorioretinopathy Retinitis pigmentosa Medicationjj Other/unknown Healthy subjects
n (%) 36 25 27 9 9 9 4 1 3 1 2 7
(27.1) (18.8) (20.3) (6.8) (6.8) (6.8) (3.0) (0.8) (2.3) (0.8) (1.5) (5.3) 65
Age (y) 84.1 72.7 72.7 64.6 75.2 57.8 57.8
6 6 6 6 6 6 6
51.0 6 53.5 6 72.6 6 50.5 6
7.0 (61–95) 9.0 (52–93) 10.6 (50–88) 10.0 (44–80) 9.0 (64–91) 18.8 (23–85) 7.8 (51–69) 78.0 17.0 (34–68) 30.0 6.4 (49–58) 13.8 (52–85) 7 (28–63)
Sex, % F 69.4 48.0 55.6 33.3 55.6 77.8 50.0 0.0 (1 male) 0.0 (3 male) 0.0 (1 male) 100.0 71.4 65.1
The age (mean 6 SD), sex (% female), and total number (%) of patients with MME in each disease category, as well as the age and sex of the healthy control subjects (no MME), are presented. N/A, not applicable. * Cause of MME according to the patient’s ophthalmologist, extracted from the patient’s chart. † Following vitrectomy (n ¼ 16), cataract extraction (n ¼ 8), and intraocular lens exchange (n ¼ 3). ‡ In three cases MME was observed in the eye with a previous episode of optic neuritis. In one case MME was observed in a 52-year-old male patient without a history of optic neuritis, but with severe atrophy of the brain involving the optic pathways. § Posterior ischemic optic neuropathy (PION). jj Hydroxychloroquine.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 955
FIGURE 3. An example of a normal retina compared with the retina of a patient with MME. (A) A normal retina from a healthy control subject. The pseudocolored surface image is the infrared surface photo; the vertical stacked gray image an OCT B-scan. (B) Optical coherence tomography image taken from a 76-year-old woman with a 3-year history of ARMD, for which she received regular injections with ranibizumab and bevacizumab. At time of imaging, her best corrected VA OS was 0.7. (C) Illustrates layer segmentation for the OCT B-scan shown in (A). The INL is bordered by the blue and orange lines. (D) The same for the patient with ARMD. Clearly, there is focal thickening of the INL were MME is observed. (E) Pan-retinal thickness map from the healthy control subject and the corresponding INL thickness map (F). (G) In the patient with ARMD, there is overall thinning of the retina. (H) There is marked heterogeneity of the INL thickness map in the ARMD patient compared with the healthy control subject.
number (%). Parametric tests (t-tests) were used for Gaussian data. Categorical data was analyzed using the v2 test. Statistical significance was accepted for two-sided testing with P < 0.05.
RESULTS Patients A total of 22,376 macular volume scans were obtained from 5865 patients between January 1, 2010, and February 15, 2013. After bootstrap, a representative dataset of 6548 macular scans from 1370 patients (2593 eyes) was subject to in-depth analyses; 99 scans of 13 patients failed quality control criteria. Two patients were excluded from further analyses because all their scans were of poor quality. Therefore, the final dataset consisted of 6449 scans from 1368 patients (2589 eyes; see Fig. 1).
revealed that interrater disagreement was due to: (1) size/ configuration—could the cysts be identified as MME/pseudocysts or did they have cell walls and could be classified as cystic macular edema? (2) location: presence of cysts within or just outside the INL; (3) confluence of cysts with cyst in other retinal layers; (4) coexisting lesions in other layers; and (5) the longitudinal pattern of presumed MME (i.e., the presence of cysts could be transient or static if macular volume scans of different dates were available). Representative examples of OCT scans which caused interrater disagreement are illustrated in Figure 2. Taking the reasons for disagreement into account, we refined the diagnostic criteria for MME as summarized in Table 1. Based on the refined diagnostic criteria for MME, the overall interrater agreement improved to kappa 0.8. Of note, two patients who were thought to have MME during the first rating were unanimously excluded using our refined diagnostic criteria.
Validation of Diagnostic Criteria for MME The screening set was split into two parts. In the first part, the observed percentage of agreement between the two raters was 92%. The interrater agreement for identifying MME was substantial (kappa 0.6). The consensus revision of the scans
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Clinical Spectrum of MME In 133/1303 (10.2%) of patients from our cohort, MME was present according to the refined diagnostic criteria (Table 1). There was a slight female predominance (57.1%) in the MME
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 956
FIGURE 4. Representative OCT images of patients with MME from the clinical spectrum are shown. The infrared surface photo and OCT image are presented to the left and the manually segmented INL to the right. (A) Microcystic macular edema in the right eye of a 70-year-old male patient with a history of proliferative diabetic retinopathy treated with panretinal photocoagulation (VA OD 0.3). (B) Optical coherence tomography image showing MME 8 months after occlusion of the vena temporalis superior OD in a 66-year-old female patient (VA OD 0.7). Microcystic macular edema was located in the temporal superior quadrant of the inner 3-mm EDTRS grid. However, INL thickening extended to the periphery. In addition, hyperreflective spots were observed in all inner retinal layers. (C) Microcystic macular edema in a 66-year-old female patient with sarcoidosis and posterior uveitis in the right
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
Microcystic Macular Edema
IOVS j February 2014 j Vol. 55 j No. 2 j 957
eye. Fluorescein angiography showed an ischemic maculopathy (VA OD 1.0). Serum ACE was 29, CT-thorax showed mediastinal and hilar lymphadenopathy. In this case, hyperreflective spots were restricted to the retinal nerve fiber layer, ganglion cell layer, and INL. (D) Microcystic macular edema in a 58-year-old female patient with a newly diagnosed pucker in the right eye (VA OD 0.2). Again, there were multiple hyperreflective spots in the inner retinal layers. (E) Microcystic macular edema 3 months after vitrectomy, in a 71-year-old female patient with a retinal detachment in the left eye (VA OS 0.05).
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 958
Microcystic Macular Edema
FIGURE 5. Microcystic macular edema in a patient with multiple sclerosis and a history of optic neuritis and branch retinal vein occlusion in the right eye. This 55-year-old female patient was diagnosed with clinical definite MS in 1992. Magnetic resonance imaging showed multiple periventricular brain lesions, and spinal T2 hyperintense lesions. In 2001, she experienced one episode of optic neuritis OD; coincidentally, an occlusion of the vena temporalis superior was found (VA 0.4). This case illustrates coexistence of MME with both a thinned (green box) and thickened (blue box) aspect of the INL. (A) The OCT B-scan clearly shows MME in the INL, which is thickened in the center and thinned near the peripheral parts of the image. The OCT image was taken 11 years after conduction of the fundus photograph in 2001, which showed superficial intraretinal hemorrhages inline with occlusion of the vena temporalis superior (B). There was thinning of the nerve fiber layer of the right optic disc, with normal appearance on the left (not shown). (C) The corresponding fluorescein angiography does not reveal leakage. Furthermore, there seemed to be a preocclusion of the vena temporalis inferior. (D) The thickness map of the INL showed more prominent thickening along the course of the inferior temporal vein, with relative thinning along the course of the superior temporal vein. (E) The destruction of the INL related to the thickness difference can be best appreciated by comparing the temporal part of the inferior retina (blue box) with the corresponding area shown on top of the image (green box). Interestingly, hyperreflective spots of the inner retinal layers were found, which were also seen in other entities of the clinical spectrum of MME.
group. Patients with MME were significantly older (72.8 years, SD 13.8, range, 23–95 years) compared with patients without MME (63.2 years, SD 17.8, range, 4–101 years; P ¼ 0.002), and healthy control subjects (50.5 years, SD 7.2, range, 28–63 years; P < 0.001). The clinical spectrum and demographic data of patients with MME is summarized in Table 2. Most frequently, MME was observed in patients with ARMD (27.1%) followed by patients with preceding ophthalmic surgery (20.3%) or presence of an epiretinal membrane (18.8%). Furthermore, MME was identified following a retinal vascular occlusion (6.8%), uveitis (6.8%), diabetic retinopathy (6.8%), MS (3.0%), posterior ischemic optic neuropathy (0.8%), and miscellaneous/other retinal conditions (5.3%). Three of the four patients with MS had a history of optic neuritis in the affected eye. Importantly, the patient who never suffered from optic neuritis had a long disease duration (14 years), was severely disabled (EDSS 6.0), showed severe white and gray matter atrophy involving the optic pathways on magnetic resonance imaging (MRI), and had bilaterally delayed visual evoked potentials (this patient never received fingolimod). Figure 3 shows an example of a normal retina of a healthy control subject compared with the retina of a patient with MME based on ARMD. Other examples of OCT images of patients found in the different disease categories are presented in Figure 4. Figure 5 shows an example of a patient presenting both with an optic neuritis and a branch retinal vein occlusion at the same time (MME was seen 11 years after onset).
Bilateral MME Bilateral MME was present in 9/133 (6.8%) of patients. The diagnosis was ARMD in four, an epiretinal membrane in three,
iatrogenic (postoperative) in one, and central serous chorioretinopathy in one.
Longitudinal Pattern of MME Longitudinal data was available from 100 out of 133 patients with MME (75.2%). Microcystic macular edema was transient in 84% of patients. It remained static for the available observation period (mean 233 days, SD 157 days; range, 46–494 days) in the remaining 16% of patients. Of note, 42% of patients with a transient pattern of MME developed cystic macular edema (n ¼ 35).
Localization of MME The perimacular rim was affected in all cases with MME, but the localization could be within any quadrant or combination of quadrants of the superimposed Early Treatment Diabetic Retinopathy Study (EDTRS) grid (Table 3). Most frequently a single quadrant was affected (67.7%). Two quadrants were affected in 15.0% of patients, three quadrants in 9.0%, and all four quadrants in 8.3% (Fig. 6). The nasal (48.1%) and temporal (49.6%) quadrants were more frequently involved compared with the superior (29.3%) and inferior (30.1%) quadrants.
MME and INL Thickness The failure rate of automated segmentation of the INL and other retinal layers was above 90% of scans from patients with MME for all three algorithms used. Therefore, no statistical analyses were performed on this data. Manual segmentation was performed in scans representative for the main disease
TABLE 3. Location of MME With Respect to the EDTRS Grid Disease Age-related macular degeneration Epiretinal membrane/vitreomacular traction Iatrogenic Diabetic retinopathy Vascular occlusion Uveitis Multiple sclerosis Optic neuropathy Central serous chorioretinopathy Retinitis pigmentosa Medication Other/unknown Total
Superior, n (%) 6 12 7 3 2 5 2 1 1 0 0 1 39
(16.7) (18.0) (25.9) (50.0) (22.0) (55.6) (50.0) (100) (25.0) (0.0) (0.0) (14.3) (29.3)
Inferior, n (%) 12 11 6 1 0 4 2 0 0 0 1 2 40
(33.3) (44.0) (22.2) (20.0) (0.0) (44.4) (50.0) (0.0) (0.0) (0.0) (50.0) (28.6) (30.1)
Nasal, n (%) 19 12 15 4 2 4 2 0 3 0 0 2 64
(52.8) (48.0) (55.6) (44.4) (22.2) (44.4) (50.0) (0.0) (75.0) (0.0) (0.0) (28.6) (48.1)
Temporal, n (%) 14 15 14 2 5 7 2 0 0 1 1 4 66
(38.9) (60.0) (51.9) (40.0) (56.0) (77.8) (50.0) (0.0) (0.0) (100) (50.0) (57.1) (49.6)
Localization in the nasal and temporal quadrants dominated in age-related macular degeneration and iatrogenic conditions. For all other conditions MME was more equally distributed within the perimacular rim. In a proportion of patients, MME was observed in more than one quadrant, explaining why the percentages for each entity can add up to more than 100%. This is further illustrated by the proportional distribution of MME in Figure 3.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 959
Microcystic Macular Edema
FIGURE 6. Sectoral localization and perimacular rim extension of MME. One quadrant was affected in the vast majority of cases (67.7%). Within this group, MME was most often found in the temporal and nasal quadrants. In a small proportion of patients, MME was extensive and lesions were found in two (15.0%), three (9.0%), or four quadrants (8.3%).
categories. Noteworthy, the INL thickness was increased in most, but not all areas with MME (Fig. 3).
DISCUSSION In this large, retrospective, single-center study, the prevalence of MME was 10.2% (n ¼ 133). To put our data into context, thus far only approximately 127 patients with MME have been reported in literature. The present study more than doubled this number, and substantially widened the clinical spectrum of MME (Table 4). Furthermore, the diagnostic criteria for MME were refined such that a substantial level of agreement (kappa 0.8) could be achieved. Microcystic changes were most frequently observed in ARMD, postoperatively after vitrectomy, cataract extraction, or intraocular lens exchanges, and in epiretinal membranes or
vitreomacular traction. The percentage of patients with MME due to MS (optic neuritis) in our study was relatively small, which might be due to our study population (i.e., a random sample of all patients receiving an OCT scan at an ophthalmologic department). In the present study, MME occurred in a large number of underlying disorders, which is in line with the uller cells become activated finding of Bringmann et al.24 that M¨ upon virtually all pathogenic stimuli. Therefore, our study supports M¨ uller cell dysfunction as part of the pathophysiology of MME. This interpretation is in line with the literature.5–7 Interestingly, MME was most often found in the nasal and temporal quadrants of the perimacular rim. It can be hypothesized that the end-arterial nature of the choroidal vasculature and the existence of watershed zones in the choroid may have a role in the location of MME. Watershed zones are borderline areas between the territories of distribu-
TABLE 4. Number of Patients (Eyes) With MME Found in Literature Disease ARMD Retinal telangiectasis Tamoxifen MS
Relapsing isolated optic neuropathy Chronic relapsing inflammatory optic neuropathy NMO Leber’s hereditary optic neuropathy/dominant optic atrophy Endemic optic neuropathy Optic nerve glioma Retrograde optic neuropathy (compressive, n ¼ 9 eyes; vascular, n ¼ 3; hereditary, n ¼ 2; not clear, n ¼ 2) Autosomal dominant optic atrophy Autosomal dominant optic atrophy Total cases of MME* This study
MME Cases
Reference(s)
22 eyes of 22 patients 24 eyes 20 eyes (of 10 patients) 1 eye of 1 patient 20 eyes of 15 patients 12 eyes of 10 patients 1 eye of 1 patient 1 eye of 1 patient 1 patient 10 patients 7 eyes of 5 patients 10 patients 25 eyes of 16 patients 2 eyes of 1 patient
Querques3 Cohen11 Gaudric12 Park33 Gelfand1 Saidha4 Balk5 Balk5 Petzold and Plant34 Sotirchos7 Gelfand1 Barboni9 Shalchi10 Abegg8
16 eyes of 9 patients 4 eyes of 2 patients 2 eyes of 1 patient Approximately 178 eyes of 127 patients 142 eyes of 133 patients
Abegg30 Gocho31 Lujan32
Patients are categorized by disease entity. * In some of the publications, the number of patients and/or the number of eyes was not well specified. Therefore, the total number of eyes and patients was estimated based on the available information.
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
IOVS j February 2014 j Vol. 55 j No. 2 j 960
Microcystic Macular Edema tion of two end-arteries, and these areas with a relatively poor blood flow are most vulnerable to hypoxia-ischemia.25,26 The main watershed zone of the choroid is located between the nasal edge of the optic disc and the fovea and represents the area between territories supplied by the temporal and nasal posterior ciliary arteries.27 Furthermore, hyperreflective lesions were anecdotally observed in some scans of patients presenting with MME. They were distributed throughout all retinal layers. Similar foci were found in patients with diabetic macular edema, which were thought to represent extravasated (lipo)proteins preceding hard exudates or blood-retina barrier breakdown.28 In patients with ARMD, such lesions were thought to represent activated microglial cells from early biological inflammatory reactions (resolution was associated with a better visual outcome).29 A particularly salient limitation of the study was that, because the SD-OCT scans were acquired without a systematic protocol, the macular volume scans consisted of different rasters with a median of 25 sections. Consequently, resolution was limited and cases with MME in between the sections may have been missed.3 In addition, it proved to be difficult to identify MME. In previous studies, problems with the identification of MME were not reported. To our knowledge, there are only few criteria applying to MME. Microcystic macular edema was defined as lacunar areas of hyporeflectivity with clear boundaries and absence of a clearly visible cell wall, in the INL, present on at least two adjacent scans.1,3,19 These criteria were used in the first part of the screening-set, by two independent raters. Of note, patients presenting with MME on only one of the sections of the macular volume scan were also included because of the limited number of sections. The interrater agreement was kappa 0.6, which can be considered substantial.23 However, including almost exclusively patients (and just a few healthy controls) with most OCT scans showing absence of MME (n ¼ 2937) might have caused the interrater agreement to be somewhat smaller than expected (whereas the observer agreement was 0.9). Clearly, when layers become difficult to separate from each other (e.g., due to poor contrast) or presence of (pseudo)cysts in adjacent layers, it can be expected that difficulties arise with localizing MME. Therefore, the present study refined the criteria for MME such that they considered abnormalities found in the INL and adjacent layers; differences in the configuration and location of cysts; confluence of cysts with cysts in other retinal layers; the presence of coexisting lesions; artefacts; and the heterogeneous longitudinal pattern. Despite the good interrater agreement (kappa 0.8), there might be other problems in rating MME that we did not account for. Therefore, we would be keen to learn how these refined criteria perform in other hands. Furthermore, we did not consider degenerative pseudocysts located just below the internal limiting membrane (50%), in the outer nuclear layer (36%), or in all retinal layers (27%), as described by Querques et al.3 for patients with ARMD. Finally, it needs to be discussed that current automated retinal layer segmentation was not feasible in patients with MME. Three different algorithms were used, which were reliable analyzing retinas of healthy control subjects and patients with drusen.20–22 The first limitation to mention was the time necessary to segment the retinal layers, which took 2 to 3 weeks for each algorithm. Second, two of the algorithms did not permit a separate segmentation of the INL.21,22 Third, the algorithm failure rate of accurate layer segmentation in eyes with MME was almost 100%, requiring extremely timeconsuming manual corrections. The automated layer segmentation could not be completed for those eyes in which the primary pathology affected the structure of the layers, resulting
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
in a change of contrast of the layers. However, local pathology (such as highly reflective spots and shadow artefacts) also led to difficulty segmenting the layers. Therefore, retinal pathology proved to be a severe limitation to the routine application of currently available retinal layer segmentation software. It has been proposed that machine learning might help to overcome this problem.22 Acknowledging the failure to provide quantitative data on INL thickness, the qualitative revision of the scans (manually corrected INL thickness maps) presented in this study suggested that it might not be appropriate to pool data from the four sectors composing the perimacular rim. Areas with visible thinning coexisted to areas with visible thickening. A more sophisticated and detailed analysis of the INL thickness may need to be performed in future studies, requiring segmentation software reliable in pathological conditions and statistical models feasible for refined retinal area surface analyses. We would be keen to hear of such efforts and discuss whether the data compiled in this study, an anonymous dataset of MME patients, may contribute to advancing the field. Further research is needed to determine the clinical significance of MME. Because MME was not present in healthy controls, further examination is suggested when microcysts are present on OCT images in (new) patients presenting in a (neuro)ophthalmologic service. Especially since imaging of the central nervous system might be needed in some cases (4% of patients with MME in the present study suffered from a neurologic condition). Furthermore, the presence of microcysts may negatively predict long-term functional outcome both in patients with ARMD and patients with MS.1,3,4 These studies showed that visual acuity decreased when microcysts were present. On the other hand, a relation between the presence of MME, visual acuity and overall disease disability was not found in patients with NMO. However, the sample size of the latter study might have been too small for determining such associations.6 In the present study, visual function was not systematically investigated. Follow-up of patients with MME in a neuro-ophthalmological service may be advised to better understand the potential prognostic and therapeutical implications of this new clinical sign.
Acknowledgments Disclosure: M.C. Burggraaff, None; J. Trieu, None; W.A.E.J. de Vries-Knoppert, None; L. Balk, None; A. Petzold, None
References 1. Gelfand JM, Nolan R, Schwartz DM, Graves J, Green AJ. Microcystic macular oedema in multiple sclerosis is associated with disease severity. Brain. 2012;135:1786–1793. 2. Petzold A. Microcystic macular oedema in MS: T2 lesion or black hole? Lancet Neurol. 2012;11:933–934. 3. Querques G, Coscas F, Forte R, Massamba N, Sterkers M, Souied EH. Cystoid macular degeneration in exudative agerelated macular degeneration. Am J Ophthalmol. 2011;152: 100–107. 4. Saidha S, Sotirchos ES, Ibrahim MA, et al. Microcystic macular oedema, thickness of the inner nuclear layer of the retina, and disease characteristics in multiple sclerosis: a retrospective study. Lancet Neurol. 2012;11:963–972. 5. Balk LJ, Killestein J, Polman CH, Uitdehaag BMJ, Petzold A. Microcystic macular oedema confirmed, but not specific for multiple sclerosis. Brain. 2012;135:e226. 6. Gelfand JM, Cree BA, Nolan R, Arnow S, Green AJ. Microcystic inner nuclear layer abnormalities and neuromyelitis optica. JAMA Neurol. 2013;70:629–633.
IOVS j February 2014 j Vol. 55 j No. 2 j 961
Microcystic Macular Edema 7. Sotirchos E, Saidha S, Byraiah G, et al. In vivo identification of morphologic retinal abnormalities in neuromyelitis optica. Neurology. 2013;80:1406–1414. 8. Abegg M, Zinkernagel M, Wolf S. Microcystic macular degeneration from optic neuropathy. Brain. 2012;135:e225. 9. Barboni P, Carelli V, Savini G, Carbonelli M, La Morgia C, Sadun A. Microcystic macular degeneration from optic neuropathy: not inflammatory, not trans-synaptic degeneration. Brain. 2013;136(pt 7):e329. 10. Kisimbi J, Shalchi Z, Mahroo OA. Macular spectral domain optical coherence tomography findings in Tanzanian endemic optic neuropathy. Brain. 2013;136(pt 11):3418–3426. 11. Cohen SY, Dubois L, Nghiem-Buffet S, et al. Retinal pseudocysts in age-related geographic atrophy. Am J Ophthalmol. 2010;150:211–217. 12. Gaudric A, Ducos de Lahitte G, Cohen SY, Massin P, Haouchine B. Optical coherence tomography in group 2A idiopathic juxtafoveolar retinal telangiectasis. Arch Ophthalmol. 2006; 124:1410–1419. 13. Cohen SM, Cohen ML, El-Jabali F, Pautler SE. Optical coherence tomography findings in nonproliferative group 2a idiopathic juxtafoveal retinal telangiectasis. Retina. 2007;27:59–66. 14. Gualino V, Cohen SY, Delyfer MN, Sahel JA, Gaudric A. Optical coherence tomography findings in tamoxifen retinopathy. Am J Ophthalmol. 2005;140:757–758. 15. Reichenbach A, Wurm A, Pannicke T, Iandiev I, Wiedemann P, Bringmann A. Muller cells as players in retinal degeneration and edema. Graefes Arch Clin Exp Ophthalmol. 2007;245: 627–636. 16. Srivastava R, Aslam M, Kalluri SR, et al. Potassium channel KIR4.1 as an immune target in multiple sclerosis. N Engl J Med. 2012;367:115–123. 17. Udagawa T, Tatsumi N, Tachibana T, et al. Inwardly rectifying potassium channel Kir4.1 is localized at the calyx endings of vestibular afferents. Neuroscience. 2012;215:209–216. 18. Tewarie P, Balk L, Costello F, et al. The OSCAR-IB consensus criteria for retinal OCT quality assessment. PLoS One. 2012;7: e34823. 19. Brar M, Yuson R, Kozak I, et al. Correlation between morphologic features on spectral-domain optical coherence tomography and angiographic leakage patterns in macular edema. Retina. 2010;30:383–389. 20. Traber G, Oberwahrenbrock T, Gabilondo I, Nolan R, Songster C, Balk L. Multicenter inter-rater reliability of retinal layer segmentation using spectral-domain OCT. Poster presented at: The North American Neuro-Ophthalmology Society (NANOS)
Downloaded From: http://iovs.arvojournals.org/ on 06/11/2015
21.
22.
23. 24. 25. 26.
27.
28.
29.
30.
31.
32.
33.
34.
Annual Meeting Neuro-Ophthalmology Collection; February 12, 2013; Snowbird, UT. Mayer M, Hornegger J, Mardin C, Tornow R. Retinal nerve fiber layer segmentation on FD-OCT scans of normal subjects and glaucoma patients. Biomed Opt Express. 2010;1:1358–1383. Dufour PA, Ceklic L, Abdillahi H, et al. Graph-based multisurface segmentation of OCT data using trained hard and soft constraints. IEEE Trans Med Imaging. 2013;32:531–543. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics. 1977;33:159–174. Bringmann A, Pannicke T, Grosche J, et al. M¨ uller cells in the healthy and diseased retina. Prog Retin Eye Res. 2006;25:397–424. Hayreh SS. In vivo choroidal circulation and its watershed zones. Eye. 1990;4:273–289. Hayreh SS. Submacular choroidal vascular bed watershed zones and their clinical importance. Am J Ophthalmol. 2010; 150:940–941. Giuffre G. Main posterior watershed zone of the choroid. Variations of its position in normal subjects. Doc Ophthalmol. 1989;72:175–180. Bolz M, Schmidt-Erfurth U, Deak G, Mylonas G, Kriechbaum K, Scholda C. Optical coherence tomographic hyperreflective foci: a morphologic sign of lipid extravasation in diabetic macular edema. Ophthalmology. 2009;116:914–920. Coscas G, De Benedetto U, Coscas F, et al. Hyperreflective dots: a new spectral-domain optical coherence tomography entity for follow-up and prognosis in exudative age-related macular degeneration. Ophthalmologica. 2013;229:32–37. Abegg M, Dysli M, Wolf S, Kowal J, Dufour P, Zinkernagel M. Microcystic macular edema, retrograde maculopathy caused by optic neuropathy. Ophthalmology. 2014;121:142–149. Gocho K, Kikuchi S, Kabuto T, et al. High-resolution en face images of microcystic macular edema in patients with autosomal dominant optic atrophy. Biomed Res Int. 2013; 2013:676803. Lujan B, Horton JC. Microcysts in the inner nuclear layer from optic atrophy are caused by retrograde trans-synaptic degeneration combined with vitreous traction on the retinal surface. Brain. 2013;136:1–4. Park SS, Zawadzki RJ, Truong SN, Choi SS, Werner JS. Microcystoid maculopathy associated with tamoxifen use diagnosed by high-resolution Fourier-domain optical coherence tomography. Retin Cases Brief Rep. 2009;3:33–35. Petzold A, Plant GT. Chronic relapsing inflammatory optic neuropathy: a systematic review of 122 cases reported. J Neurol. 2014;261:17–26.
