- Email: [email protected]
Quantification of Retinal Tangential Movement in Epiretinal Membranes
Quantification of Retinal Tangential Movement in Epiretinal Membranes Mads Kofod, MD, Morten la Cour, MD, DMSc Objective: To describe a technique of quantifying retinal vessel movement in eyes with epiretinal membrane (ERM) and correlate the retinal vessel movement with changes in best-corrected visual acuity (BCVA), central macular thickness (CMT), and patients’ subjective reports about experienced symptoms (symptoms). Design: Retrospective, observational, comparative case series. Participants: A total of 206 eyes of 113 patients: 142 eyes with ERM and 64 healthy fellow eyes. Methods: All patients were examined as part of a screening protocol for a randomized clinical trial on ERM. Heidelberg Spectralis (Heidelberg Engineering Inc, Carlsbad, CA) optical coherence tomography (OCT) scanning had been performed twice or more in all patients. Eyes with ERM and healthy fellow eyes were examined. For each eye, the 2 fundus images were aligned using Heidelberg’s AutoRescan feature. The macular area was divided into 9 subfields, and retinal vessel movements were calculated. For each eye, the total length of the 9 vectors was summed to describe the total retinal vessel movement (retinal tangential movement [RTM]). Main Outcome Measures: To quantify retinal vessel movements associated with ERM. The secondary outcome was to correlate measured retinal vessel movement with changes in BCVA, CMT, and patients’ subjective symptoms. Results: The study found significantly greater RTM in ERM eyes compared with healthy eyes (P⬍0.001). Among ERM eyes, the RTM was significantly greater in patients with worsening of symptoms compared with ERM eyes with unchanged assessment of symptoms. There were statistically significant correlations between increased RTM and reduction in BCVA (P ⫽ 0.024), increased CMT (P⬍0.001), and time between visits (P⬍0.001). Conclusions: This study showed that ERM is not a static retinal disease but a dynamic condition in which retinal vessel movement associated with ERM was measureable, even in patients who had stable BCVA and CMT. The retinal vessel movements correlated to worsening of BCVA and increased CMT, and were more pronounced in patients with worsening of symptoms. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article. Ophthalmology 2012;119:1886 –1891 © 2012 by the American Academy of Ophthalmology.
Epiretinal membrane (ERM) is a macular disease in which patients commonly have blurred central vision and metamorphopsia.1 The symptoms are associated with the opacity of the membrane and the amount of macular distortion induced by the contracting ERM. Decreased best-corrected visual acuity (BCVA) has been associated with increased central macular thickness (CMT),2– 4 but the natural history of tractional force on the retina or retinal vessel movement in ERM is rarely described.5,6 Although some patients with ERM adapt to their distortions and learn to ignore them, a subset of patients describe the inability to adapt. We hypothesize that such patients might have continual small shifts in the retinal tangential movement (RTM) on the retina making continual small movement of the central cones. The small cone movements could make distortions more dynamic and more difficult to adapt to. Contraction of retinal vessels in fundus imaging has been used to evaluate the natural course of ERM in a smaller study and to describe correlation with metamorphopsia scores using the M-chart.5,6 Arimura et al5 found a correlation between severity of metamorphopsia and retinal con-
1886
© 2012 by the American Academy of Ophthalmology Published by Elsevier Inc.
traction, but no significant correlation to BCVA. Weinberger et al6 reported measureable contraction and release of ERM. Aligned fundus imaging on the Heidelberg Spectralis has been used to examine changes in multiple sclerosis and geographic atrophy.7,8 A new study by Lo et al9 describes foveal dystopia in aligned fundus images. They describe preoperative movement of fovea by traction from an ERM in 2 patients.9 Advances in optical coherence tomography (OCT) technology leading to improvements in scan speed and software options allow for new and innovative methods of quantifying retinal images. Heidelberg Engineering’s Spectralis OCT scanner allows for measuring the same part of the retina over time through its TruTrack and AutoRescan functionalities. Progression scanning provides aligned fundus images and the aligned OCT scan. These aligned fundus photographs enable easy evaluation of retinal vessel movements associated with ERM. Our study aimed to quantify retinal vessel movements over time in healthy eyes and eyes with ERM. Secondary ISSN 0161-6420/12/$–see front matter http://dx.doi.org/10.1016/j.ophtha.2012.03.022
Kofod and la Cour 䡠 Retinal Tangential Movement in ERM goals were to correlate the retinal vessel movement with changes in symptoms, BCVA, and CMT. We also wanted to investigate whether the retinal vessel movement correlated to worsening of symptoms. Retinal vessel movement associated with the ERM could be a quantifiable predictor of disease progression in ERM. Our hypothesis was that excessive retinal vessel movements exist in ERM eyes, and that these movements were greatest in patients with worsening of symptoms.
Materials and Methods The study population was screened for a randomized clinical trial (RCT) on early surgery for ERM. All patients screened for the trial did not require surgery according to department standards (no binocular symptoms, BCVA ⬎0.4 decimal, no disturbing metamorphopsia) at the initial examination. Patients requiring surgery at any visit were referred accordingly and left the screening program. For the remaining screening patients, the need for surgery was reevaluated at subsequent visits. Patients included in the RCT and patients screened for the RCT did not require surgery according to department standard. This study was conducted as chart and OCT scan reviews of patients undergoing protocolled screening for an RCT and who had been scanned on a Heidelberg Spectralis OCT scanner twice or more using the follow-up (AutoRescan) procedure. Inclusion criteria were as follows: 2 or more visits to our department, diagnosis of ERM, and scanned on the Heidelberg Engineering Spectralis scanner using AutoRescan. Exclusion criteria were as follows: prior surgery to the posterior segment, obvious misaligned follow-up scans, and pathology other than ERM in the retina. All BCVA measurements were recorded by trained optometrists. Refraction was performed on Snellen charts, and BCVA was recorded on Early Treatment Diabetic Retinopathy Study charts at 4 m distance. Ophthalmic examinations and history were recorded by the same examiner for all patients. The focus of the examination was to identify changes in subjective symptoms and quantifiable worsening in BCVA to allow participation in the clinical trial. Visual acuity, gender, age, and phakic/pseudophakic lens data were recorded from patient records. In patients with OCT-verified binocular ERM, both eyes were recorded as being ERM eyes. Asymptomatic ERM fellow eyes were recorded as ERM eyes. In patients with monocular ERM, the OCT-verified healthy fellow eyes were included as the control group. No eye in the study required surgery at the initial visit. At the screening and subsequent examinations, patients were presented with standardized questions regarding distortions and if they had experienced a change in distortions since the prior examination. Patients were divided into subgroups according to description of symptoms. The ERM eyes were divided into 2 groups. Patients who did not describe changes in metamorphopsia were noted as having unchanged symptoms. Patients who described worsening of metamorphopsia were recorded as reporting worsening of symptoms. Patients with monocular or binocular ERM were included. The image grader was unaware of patients’ symptoms at the time of grading. The Spectralis OCT AutoRescan feature gave both an OCT scan and a corresponding high-quality fundus picture. In essence, the AutoRescan depends on active eye tracking provided through TruTrack. The eye tracking creates a detailed retinal map and corrects for eye movements. The images are improved through multisampling and noise reduction through the feature called artificial real time (ART). Each follow-up OCT scan was registered
and locked to the baseline scan. AutoRescan then accurately superimposes the new OCT scan on top of the baseline scan. The OCT image can be combined with several different fundus image modalities. The AutoRescan uses the same fundus image modality as recorded in the baseline scan. The finer points of fundus picture and OCT alignment are not disclosed, because they are protected under Heidelberg Engineering copyrights. To measure retinal vessel movements, a baseline and a follow-up fundus picture were selected for each eye. All retinal vessel movement measurements presented in this study were recorded on fundus pictures obtained with the Infrared ⫹ OCT setting. If patients had been scanned with the AutoRescan feature more than once, the images spaced furthest apart in time were compared. We used the standard 30-degree single line scans though the fovea (using central fixation) and a 30-degree infrared fundus picture. These scans used 100 ART setting, enabling multisampling and noise reduction over 100 fundus images and OCT B-scans. The fundus photographs were exported as 768⫻768 pixel TIFF images. Central macular thickness was measured through a 20⫻20degree volume scan with 49 scan lines and 20 ART. Central 1-mm subfield thicknesses are reported in this study. Foveal dystopia was also identified through volume OCT scans. The foveola was identified on 1 of the 49 B-scans, and the corresponding coordinate was noted on the fundus image. In the follow-up scan, the foveola was once again identified and marked on the fundus image. Foveal dystopia was calculated by Pythagoras theorem and converted from pixels to microns. Pictures were imported into a freeware image manipulation software (GIMP 2.6, www.gimp.org downloaded February 12, 2011) and placed in separate layers. The fundus image was divided into 5 equal parts horizontally and vertically. Retinal vessel movement was measured in the central 9 squares (Fig 1).
Figure 1. Infrared fundus picture with overlay showing divisions into subfield grids. In the numbered central 9 of the 25 subfields, retinal tangential movement (RTM) was recorded. Presented are the vectors for healthy eyes, all epiretinal membrane (ERM) eyes, ERM eyes with stable subjective symptoms, and ERM eyes with worsening of subjective symptoms.
1887
Ophthalmology Volume 119, Number 9, September 2012 Table 1. Basic Characteristics of Epiretinal Membrane and Healthy Fellow Eyes
Number of patients Age in years: mean (range) Days between baseline and follow-up: median (q25–q75) BCVA baseline in ETDRS letters: median (q25–q75) BCVA follow-up in ETDRS letters: median (q25–q75) BCVA change between visits in ETDRS letters: median (q25–q75) OCT central thickness baseline in microns: median (q25–q75) OCT central thickness follow-up in microns: median (q25–q75) OCT central thickness change in microns: median (q25–q75) RTM in microns: median (q25–q75)
Healthy Eyes
ERM Eyes, All ERM Eyes
ERM Eyes, Unchanged Symptoms
64 69.7 (29–88) 224 (125–294)
142 70.0 (50–88) 174 (118–267)
103 70.2 (60–86) 171 (111–267)
39 69.4 (52–88) 213 (124–352)
85 (82–90.3)
80 (75–85)
80 (75–86.5)
78 (75–83.5)
86 (83–90)
81 (74–86)
82 (75–87)
77 (69–83)
0.5 (⫺2–⫹3)
0 (⫺3–⫹3)
1 (⫺2–⫹3)
ERM Eyes, Worsening of Symptoms
⫺1 (⫺5.5–⫹1.5)
279 (256–296)
416 (341–470)
394 (315–456)
454 (400–482)
278 (254–295)
420 (338–472)
399 (316–457)
456 (402–501)
⫺1 (⫺3.5–⫹2) 140.2 (112–168)
0 (⫺7–⫹13) 301.5 (191–495)
⫺1 (⫺8–⫹9) 274.4 (171–483)
1 (⫺5.5–⫹26.5) 350.8 (252–574)
BCVA ⫽ best-corrected visual acuity; ERM ⫽ epiretinal membrane; ETDRS ⫽ Early Treatment Diabetic Retinopathy Study; q25 ⫽ lower 25% quartile; q75 ⫽ upper 75% quartile; OCT ⫽ optical coherence tomography; RTM ⫽ retinal tangential movement.
In each of the 9 squares, retinal vessel branch points were selected in the baseline picture, and x and y coordinates were recorded. Recognizable branch points with the greatest movement between the baseline and follow-up pictures were selected. Toggling between the layers allowed for identification of movement in retinal vessel branch points. The same branch points were selected in the follow-up picture, and x and y coordinates (pixels) were once again recorded. Vector lengths were calculated with Pythagoras’ theorem and converted to microns according to the manufacturer’s description of pixel-to-micron ratio. We added the 9 vector lengths in an eye into a single vector length to represent the total retinal vessel movement. To estimate the reliability of image alignment, choroidal vessel movement was measured. Retinal vessel movement was expected in ERM eyes, but choroidal vessel movement should not occur in healthy fellow eyes or ERM eyes. To confirm that we were examining retinal vessel movement and not optical artifacts, we controlled our retinal vessel movement against choroidal vessel movement. This was done in the same 9 subfields in a subset of patients with visible choroidal vessels. The current study obtained required approval from the Danish Data Protection Agency. This study was conducted from the screening material of the RCT approved by the regional ethical committee of the capital region of Copenhagen, Denmark, under H-C-2008-026 (registered at Clinicatrial.gov: NCT00902629). Statistic testing was performed on SigmaPlot 10.0 (2006 Systat Software Inc, Chicago, IL). Descriptive statistics were used to describe means and quartiles. Kruskal–Wallis nonparametric analysis of variance tests were used to compare anatomic subgroups. Ranked sum tests were used to compare groups, and linear regression models were used to describe correlations.
Results A total of 206 eyes in 113 patients were included distributed among 53 men and 60 women. Among the analyzed eyes, 64 were healthy and 142 had ERM. Three patients reported relief of symptoms at the follow-up examination and were pooled with the
1888
unchanged group. Subjective symptoms were unchanged in 103 ERM eyes, and 39 ERM eyes experienced worsening of symptoms between the 2 visits. Basic characteristics and BCVA and CMT measurements are shown in Table 1. The calculated retinal vessel movement vector lengths are shown in Figure 2 (available online-only at http:// aaojournal.org) for each anatomic subfield. To localize whether retinal movement was more predominant in 1 macular subfield compared with another, Kruskal–Wallis nonparametric analysis of variance on ranks was performed. This showed that in ERM eyes, the central subfield had a significantly lower ranking (greater retinal vessel movement) compared with the inferior nasal subfield (P⬍0.05) and the 3 superiorly located subfields (P⬍0.05). The same test in healthy eyes found no statistically significant differences (P ⫽ 0.138). To achieve a better estimate of the total retinal vessel movement, we used the sum of the length of the 9 recorded retinal vectors and denoted this sum as retinal tangential movement (RTM), which disregards the direction of the vector and only sums the vector lengths. The total RTM was statistically significantly greater in ERM eyes compared with healthy eyes (P⬍0.001) (Fig 3). No choroidal vessel movement was expected in any eyes. To control fundus image alignment, choroidal vessel movement was measured in 10 healthy and 13 ERM eyes. These were minimal with vector lengths between zero and 45.6 m in both healthy and ERM eyes. Kruskal–Wallis 1-way analysis on ranks found no statistically significantly difference between anatomic subgrids (P ⫽ 0.348) in choroidal vessel movement. The 9 anatomic subfield vector lengths were summed into a single total choroidal vessel movement vector. Ranked sum showed no statistically significant difference between choroidal vessel movement in healthy eyes and ERM eyes (P ⫽ 0.077), as shown in Figure 3. There was no difference between choroidal vessel movement in all eyes and retinal vessel movement in healthy eyes (P ⫽ 0.866). Ranked sum tests between choroidal vessel movement and retinal vessel movement in healthy eyes did not show significant differences (P ⫽ 0.87). Ranked sum test in ERM eyes comparing choroidal vessel movement and retinal vessel movement found significantly greater retinal vessel movements (P⬍0.001). There
Kofod and la Cour 䡠 Retinal Tangential Movement in ERM central thickness at both examinations. Among ERM eyes with worsening of symptoms, 25.6% lost more than 5 Early Treatment Diabetic Retinopathy Study letters, whereas ERM eyes with unchanged symptoms lost more than 5 letters in 9.7% of cases. Ranked sum tests showed that BCVA was significantly worse in patients demonstrating worsening at the follow-up (P ⫽ 0.002), and decrease in BCVA was also statistically significantly greater in ERM eyes with worsening of symptoms (P ⫽ 0.011). Correlation analysis was also performed using only the retinal vessel movement instead of RTM. This produced outcomes on par with RTM but with more uncertain estimates.
Discussion
Figure 3. Comparison of retinal tangential movement (RTM) in healthy eyes and epiretinal membrane (ERM) eyes.
was statistically greater retinal movement in ERM eyes compared with choroidal vessel movement measurements in both healthy and ERM eyes (P⬍0.001). The comparison of choroidal and retinal vessel movement indicated that the measured retinal vessel movement is indeed retinal movement and does not stem from image alignment problems. The RTM was further analyzed in ranked sum tests and linear regressions. Among healthy eyes, no significant correlation was found between RTM and BCVA, change in BCVA, CMT, or change in CMT (Table 2, available online-only at http://aaojournal.org). Comparison of all ERM eyes with healthy eyes showed significantly greater RTM in ERM eyes (P⬍0.001). The study found significant correlations between decreasing BCVA as CMT thickness increased at both baseline (coefficient ⫺4.66, P⬍0.001) and follow-up (coefficient ⫺4.84, P⬍0.001) among ERM eyes. Regression analysis between BCVA and CMT in patients with ERM showed a significant correlation between decline in BCVA and increase in CMT (P⬍0.001). A correlation was found between increasing RTM and increasing CMT at both baseline (coefficient ⫹0.02, P⬍0.001) and follow-up (coefficient ⫹0.084, P⬍0.001). The RTM increased statistically significantly as loss of BCVA between visits increased (coefficient ⫺1.49, P⬍0.001). The RTM correlated to time between visits but only in ERM eyes (P⬍0.001). The gain in RTM was 47.7 m/month in ERM eyes and 5.3 m/month in healthy fellow eyes (P ⫽ 0.073). Foveal dystopia was 27.4 m on average in healthy fellow eyes and 55.5 m in ERM eyes. There was a linear correlation between central grid retinal vessel movement and foveal dystopia with a coefficient of ⫹0.398 (P⬍0.001). There was no correlation between the foveal dystopia and the RTM (P ⫽ 0.751). Foveal dystopia did not correlate to BCVA change (P ⫽ 0.835), CMT change (P ⫽ 0.685), or time between examinations (P ⫽ 0.686). The ERM eyes were divided into a group with unchanged symptoms and a group experiencing worsening of symptoms. The RTM was significantly greater in patients with worsening of symptoms compared with patients with unchanged symptoms (P ⫽ 0.024) (Fig 4). In patients with worsening of symptoms, the CMT increase was 7.95 m (⫾29.60) compared with 2.28 m (⫾21.03) CMT increase in eyes with unchanged symptoms (P ⫽ 0.209). Ranked sum test showed a statistically significant difference in CMT at both baseline (P ⫽ 0.001) and follow-up (P⬍0.001), where patients who developed worsening of symptoms had thicker
Although commercially available OCT scanners can measure retinal thickness and view B-scan morphologic features in great detail, the presented option to view aligned fundus images adds a new method to evaluate ERM cases. When viewing aligned fundus images in toggle mode, it was apparent whether tractional changes associated with the ERM were present or not. This quantification of the retinal vessel movement illustrates the possibility of using aligned retinal images to track and describe RTM in ERM eyes over time. The study describes a method to quantify the retinal traction to enable comparison between ERM eyes and healthy fellow eyes. Significantly greater RTM was found in ERM eyes compared with healthy fellow eyes. The RTM was correlated to decreased BCVA and increased CMT. All findings were more excessive in patients who described subjective worsening of symptoms between the 2 visits. The automated alignment process of fundus images is not perfected, and minute changes were present even in healthy eyes. The type or method of image manipulation used in the process of creating aligned fundus images is known only by the manufacturer. To determine whether our measurements of dynamic retinal traction was indeed a measure of traction or a part of the AutoRescan process, we compared retinal and choroidal blood vessel movements. One could speculate that the AutoRescan alignment algorithm would rotate the image and induce artificial vessel
Figure 4. Comparison of retinal tangential movement (RTM) in epiretinal membrane (ERM) eyes with and without worsening of symptoms.
1889
Ophthalmology Volume 119, Number 9, September 2012 movement. If this were true, the choroidal vessels would move as much as the retinal vessels. Our results indicate that the choroidal vessel movements were equal to retinal movements in healthy eyes. Because the choroidal movements were statistically significantly less than the retinal movement in an ERM eyes, we believe that we are indeed measuring the retinal tangential traction associated with ERM on the retinal surface. The patients were asked about changes in metamorphopsia, but they may confuse changes in metamorphopsia with changes in visual acuity. This makes the assessment of changes in patients’ symptoms an uncertain measure. We present our best estimate of changes in symptoms because all examinations were performed by the same examiner using standardized questions during screening for an RCT, and the focus of the examination was to find changes in metamorphopsia and BCVA to allow participation in the RCT. A systematic questionnaire quantifying patients’ symptoms and a prospective design would have been preferable. Similar to Weinberger et al,6 we measured minor shifts in the retinal vessels of healthy eyes, which we contribute to the AutoRescan algorithm. We calculated the uncertainty of the alignment algorithm to be 242.8 m across 9 vessel measurements equal to the 95% fractile of RTM in healthy eyes. Movement in excess of this represented true retinal vessel movement. This was present in 58% of patients with unchanged symptoms and in 76% of patients describing worsening of symptoms. Observing substantial retinal vessel movement in ERM eyes even without changes to BCVA or CMT shows that ERM is a dynamic disease. Weinberger et al6 measured vessel displacement movements between major and minor vessel junctions in the upper and lower temporal arcades, as well as movement between disc and vessel junction temporal to the fovea. In contrast with their study, our applied subfield grid was placed automatically in each image, eliminating the need for manual corrections of grid location. The aligned pictures made it evident that multiple centers of traction existed in most eyes. The retinal movement did not have a consistent relationship to the fovea. Simple modeling of the observed complex retinal movement is a challenge. The assumption that the central grid retinal vessel movement would give a better estimation for changes in BCVA and CMT was tested. This showed equal correlations as presented above with RTM, but with a greater uncertainly of the estimate. This is perhaps not too surprising because stereopsis is not isolated to the fovea. A recent study found depth perception to be almost as accurate in 7 degrees eccentricity as centrally.10 The assumptions in ERM are that a centripetal focus exists over the fovea and that there is a central contraction of retinal tissue. Gupta et al11 have shown that multiple foci of retinal contractions exist in 20 of 44 examined eyes. These multiple centers of centripetal retinal vessel movement were also present in this material. With several centripetal foci located away from the fovea, the traction can seem to be centrifugal. Quantification of metamorphopsia by binocular correspondence perimetry in ERM and macu-
1890
lar holes finds metamorphopsia in ERM to be more chaotic and widespread.12,13 For this reason and the findings presented by Gupta et al, we elected to use the total length of the 9 subfield vectors, disregarding the direction of the individual vector. The retinal vessel movement associated with ERM can occur in all subfields, and the central grid will almost always be involved. This could explain the significantly lower ranking (greater vector lengths), but not mean vector lengths in the central subfield. Lo et al9 described severe foveal dystopia and found that the fovea returned 32.8%⫾22.1% toward its expected anatomic position after surgery. In the 2 cases they measured foveal movement preoperatively, the centripetal movement was smaller than the centrifugal movement after surgery. This indicates a continual contraction, which in the study by Lo et al must have had foci away from the fovea to allow such severe foveal dystopia. Foveal movement and retinal vessel movement in the central grid could be assumed to be more representative of changes in BCVA or changes in patients’ symptoms than retinal vessel movement in adjacent subfields. This was not the case in the present study in which exclusively analyzing the foveal retinal vessel movement gave a more uncertain estimate compared with RTM. We hypothesized that RTM played a significant role in progression of disease as the ERM convolutes the neurosensitive tissue. This was most evident in patients with worsening of symptoms in whom excessive RTM was present in the absence of increased CMT or decreased BCVA. This was the case in 41% of patients with subjective worsening. A weakness in this study was our selection of measuring points in which we chose to record traction in 9 central subfields in the macula. We found strong correlations between CMT and RTM and assumed this to be caused by the ERM, but the increase in CMT could be caused by other factors, such as axoplasmic flow obstruction.14 The study is also limited by its retrospective nature and the fact that the control group consisted of fellow eyes to ERM eyes. In patients with multiple rescans, only the 2 spaced furthest in time were compared. Patients experiencing worsening of symptoms are more likely to continue participating in controls, leading to a selection bias. Describing RTM evolution over several visits could strengthen the conclusions, but this was not performed because the majority of patients had only 2 examinations to compare. Despite many patients having stable BCVA and CMT, there was a strong correlation between time between examinations and RTM, suggesting that a continual, and in some patients a fast tractional, change is present in ERM eyes. This strengthens our hypothesis that patients with changes in the retina (i.e., greater RTM) will have a harder time adjusting to the constant changes and be more likely to experience distortions or want to undergo surgery. The symptomatology of ERM consists of visual acuity loss and metamorphopsia. For patients with good visual acuity but marked metamorphopsia, we lack an objective measurement that correlates with the patients’ subjective symptoms of metamorphopsia. This study describes a method to quantify retinal movement accompanying ERM and correlates this with the more traditional measures of
Kofod and la Cour 䡠 Retinal Tangential Movement in ERM ERM traction. We think this adds to the repertoire of our objective means of studying ERM. We do not think our method is immediately applicable for clinical use. However, the principle that retinal movements over a wider macular area are correlated with both function and symptoms is worthwhile to explore further. Future prospective studies may benefit from correlating retinal movement in ERM to symptomatic metamorphopsia or binocular symptoms. There is reason to believe that this will be a more accurate and objective correlate to these symptoms than CMT, which is more likely to be correlated with visual acuity. In conclusion, we present a method to quantitatively describe the retinal movement that accompanies ERM and correlate this to the more traditional measures of ERM traction. This quantification of RTM correlated well with changes in BCVA and CMT in ERM eyes, especially in patients with worsening of symptoms. Even in patients with unchanged symptoms, we found increased RTM over time.
References 1. Ting FS, Kwok AK. Treatment of epiretinal membrane: an update. Hong Kong Med J 2005;11:496 –502. 2. Massin P, Paques M, Masri H, et al. Visual outcome of surgery for epiretinal membranes with macular pseudoholes. Ophthalmology 1999;106:580 –5. 3. Michalewski J, Michalewska Z, Cisiecki S, Nawrocki J. Morphologically functional correlations of macular pathology connected with epiretinal membrane formation in spectral optical coherence tomography (SOCT). Graefes Arch Clin Exp Ophthalmol 2007;245:1623–31. 4. Treumer F, Wacker N, Junge O, et al. Foveal structure and thickness of retinal layers long-term after surgical peeling of idiopathic epiretinal membrane. Invest Ophthalmol Vis Sci 2011;52:744 –50.
5. Arimura E, Matsumoto C, Okuyama S, et al. Retinal contraction and metamorphopsia scores in eyes with idiopathic epiretinal membrane. Invest Ophthalmol Vis Sci 2005;46: 2961– 6. 6. Weinberger D, Stiebel-Kalish H, Priel E, et al. Digital red-free photography for the evaluation of retinal blood vessel displacement in epiretinal membrane. Ophthalmology 1999;106: 1380 –3. 7. Sayegh RG, Simader C, Scheschy U, et al. A systematic comparison of spectral-domain optical coherence tomography and fundus autofluorescence in patients with geographic atrophy. Ophthalmology 2011;118:1844 –51. 8. Serbecic N, Aboul-Enein F, Beutelspacher SC, et al. High resolution spectral domain optical coherence tomography (SD-OCT) in multiple sclerosis: the first follow up study over two years. PLoS One [serial online] 2011;6:e19843. Available at: http://www.plosone.org/article/info%3Adoi%2F10. 1371%2Fjournal.pone.0019843. Accessed February 29, 2012. 9. Lo D, Heussen F, Ho HK, et al. Structural and functional implications of severe foveal dystopia in epiretinal membranes. Retina 2012;32:340 – 8. 10. Devisme C, Drobe B, Monot A, Droulez J. Stereoscopic depth perception in peripheral field and global processing of horizontal disparity gradient pattern. Vision Res 2008;48:753– 64. 11. Gupta P, Sadun AA, Sebag J. Multifocal retinal contraction in macular pucker analyzed by combined optical coherence tomography/scanning laser ophthalmoscopy. Retina 2008;28: 447–52. 12. Kroyer K, Christensen U, Larsen M, la Cour M. Quantification of metamorphopsia in patients with macular hole. Invest Ophthalmol Vis Sci 2008;49:3741– 6. 13. Kroyer K, Jensen OM, Larsen M. Objective signs of photoreceptor displacement by binocular correspondence perimetry: a study of epiretinal membranes. Invest Ophthalmol Vis Sci 2005;46:1017–22. 14. Arroyo JG, Irvine AR. Retinal distortion and cotton-wool spots associated with epiretinal membrane contraction. Ophthalmology 1995;102:662– 8.
Footnotes and Financial Disclosures Originally received: August 15, 2011. Final revision: March 12, 2012. Accepted: March 13, 2012. Available online: May 26, 2012.
Grant Support: Danish Agency for Science Technology and Innovation, Synoptik Foundation, Velux Foundation, Bagenkop Nielsen Eye Foundation. Manuscript no. 2011-1232.
Department of Ophthalmology, Glostrup Hospital, University of Copenhagen, Denmark. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.
Correspondence: Mads Kofod, MD, Department of Ophthalmology, Glostrup Hospital, Nordre Ringvej 57, DK-2600 Glostrup, Denmark. E-mail: madkof01@glo. regionh.dk.
1891
Epiretinal membrane (ERM) is a macular disease in which patients commonly have blurred central vision and metamorphopsia.1 The symptoms are associated with the opacity of the membrane and the amount of macular distortion induced by the contracting ERM. Decreased best-corrected visual acuity (BCVA) has been associated with increased central macular thickness (CMT),2– 4 but the natural history of tractional force on the retina or retinal vessel movement in ERM is rarely described.5,6 Although some patients with ERM adapt to their distortions and learn to ignore them, a subset of patients describe the inability to adapt. We hypothesize that such patients might have continual small shifts in the retinal tangential movement (RTM) on the retina making continual small movement of the central cones. The small cone movements could make distortions more dynamic and more difficult to adapt to. Contraction of retinal vessels in fundus imaging has been used to evaluate the natural course of ERM in a smaller study and to describe correlation with metamorphopsia scores using the M-chart.5,6 Arimura et al5 found a correlation between severity of metamorphopsia and retinal con-
1886
© 2012 by the American Academy of Ophthalmology Published by Elsevier Inc.
traction, but no significant correlation to BCVA. Weinberger et al6 reported measureable contraction and release of ERM. Aligned fundus imaging on the Heidelberg Spectralis has been used to examine changes in multiple sclerosis and geographic atrophy.7,8 A new study by Lo et al9 describes foveal dystopia in aligned fundus images. They describe preoperative movement of fovea by traction from an ERM in 2 patients.9 Advances in optical coherence tomography (OCT) technology leading to improvements in scan speed and software options allow for new and innovative methods of quantifying retinal images. Heidelberg Engineering’s Spectralis OCT scanner allows for measuring the same part of the retina over time through its TruTrack and AutoRescan functionalities. Progression scanning provides aligned fundus images and the aligned OCT scan. These aligned fundus photographs enable easy evaluation of retinal vessel movements associated with ERM. Our study aimed to quantify retinal vessel movements over time in healthy eyes and eyes with ERM. Secondary ISSN 0161-6420/12/$–see front matter http://dx.doi.org/10.1016/j.ophtha.2012.03.022
Kofod and la Cour 䡠 Retinal Tangential Movement in ERM goals were to correlate the retinal vessel movement with changes in symptoms, BCVA, and CMT. We also wanted to investigate whether the retinal vessel movement correlated to worsening of symptoms. Retinal vessel movement associated with the ERM could be a quantifiable predictor of disease progression in ERM. Our hypothesis was that excessive retinal vessel movements exist in ERM eyes, and that these movements were greatest in patients with worsening of symptoms.
Materials and Methods The study population was screened for a randomized clinical trial (RCT) on early surgery for ERM. All patients screened for the trial did not require surgery according to department standards (no binocular symptoms, BCVA ⬎0.4 decimal, no disturbing metamorphopsia) at the initial examination. Patients requiring surgery at any visit were referred accordingly and left the screening program. For the remaining screening patients, the need for surgery was reevaluated at subsequent visits. Patients included in the RCT and patients screened for the RCT did not require surgery according to department standard. This study was conducted as chart and OCT scan reviews of patients undergoing protocolled screening for an RCT and who had been scanned on a Heidelberg Spectralis OCT scanner twice or more using the follow-up (AutoRescan) procedure. Inclusion criteria were as follows: 2 or more visits to our department, diagnosis of ERM, and scanned on the Heidelberg Engineering Spectralis scanner using AutoRescan. Exclusion criteria were as follows: prior surgery to the posterior segment, obvious misaligned follow-up scans, and pathology other than ERM in the retina. All BCVA measurements were recorded by trained optometrists. Refraction was performed on Snellen charts, and BCVA was recorded on Early Treatment Diabetic Retinopathy Study charts at 4 m distance. Ophthalmic examinations and history were recorded by the same examiner for all patients. The focus of the examination was to identify changes in subjective symptoms and quantifiable worsening in BCVA to allow participation in the clinical trial. Visual acuity, gender, age, and phakic/pseudophakic lens data were recorded from patient records. In patients with OCT-verified binocular ERM, both eyes were recorded as being ERM eyes. Asymptomatic ERM fellow eyes were recorded as ERM eyes. In patients with monocular ERM, the OCT-verified healthy fellow eyes were included as the control group. No eye in the study required surgery at the initial visit. At the screening and subsequent examinations, patients were presented with standardized questions regarding distortions and if they had experienced a change in distortions since the prior examination. Patients were divided into subgroups according to description of symptoms. The ERM eyes were divided into 2 groups. Patients who did not describe changes in metamorphopsia were noted as having unchanged symptoms. Patients who described worsening of metamorphopsia were recorded as reporting worsening of symptoms. Patients with monocular or binocular ERM were included. The image grader was unaware of patients’ symptoms at the time of grading. The Spectralis OCT AutoRescan feature gave both an OCT scan and a corresponding high-quality fundus picture. In essence, the AutoRescan depends on active eye tracking provided through TruTrack. The eye tracking creates a detailed retinal map and corrects for eye movements. The images are improved through multisampling and noise reduction through the feature called artificial real time (ART). Each follow-up OCT scan was registered
and locked to the baseline scan. AutoRescan then accurately superimposes the new OCT scan on top of the baseline scan. The OCT image can be combined with several different fundus image modalities. The AutoRescan uses the same fundus image modality as recorded in the baseline scan. The finer points of fundus picture and OCT alignment are not disclosed, because they are protected under Heidelberg Engineering copyrights. To measure retinal vessel movements, a baseline and a follow-up fundus picture were selected for each eye. All retinal vessel movement measurements presented in this study were recorded on fundus pictures obtained with the Infrared ⫹ OCT setting. If patients had been scanned with the AutoRescan feature more than once, the images spaced furthest apart in time were compared. We used the standard 30-degree single line scans though the fovea (using central fixation) and a 30-degree infrared fundus picture. These scans used 100 ART setting, enabling multisampling and noise reduction over 100 fundus images and OCT B-scans. The fundus photographs were exported as 768⫻768 pixel TIFF images. Central macular thickness was measured through a 20⫻20degree volume scan with 49 scan lines and 20 ART. Central 1-mm subfield thicknesses are reported in this study. Foveal dystopia was also identified through volume OCT scans. The foveola was identified on 1 of the 49 B-scans, and the corresponding coordinate was noted on the fundus image. In the follow-up scan, the foveola was once again identified and marked on the fundus image. Foveal dystopia was calculated by Pythagoras theorem and converted from pixels to microns. Pictures were imported into a freeware image manipulation software (GIMP 2.6, www.gimp.org downloaded February 12, 2011) and placed in separate layers. The fundus image was divided into 5 equal parts horizontally and vertically. Retinal vessel movement was measured in the central 9 squares (Fig 1).
Figure 1. Infrared fundus picture with overlay showing divisions into subfield grids. In the numbered central 9 of the 25 subfields, retinal tangential movement (RTM) was recorded. Presented are the vectors for healthy eyes, all epiretinal membrane (ERM) eyes, ERM eyes with stable subjective symptoms, and ERM eyes with worsening of subjective symptoms.
1887
Ophthalmology Volume 119, Number 9, September 2012 Table 1. Basic Characteristics of Epiretinal Membrane and Healthy Fellow Eyes
Number of patients Age in years: mean (range) Days between baseline and follow-up: median (q25–q75) BCVA baseline in ETDRS letters: median (q25–q75) BCVA follow-up in ETDRS letters: median (q25–q75) BCVA change between visits in ETDRS letters: median (q25–q75) OCT central thickness baseline in microns: median (q25–q75) OCT central thickness follow-up in microns: median (q25–q75) OCT central thickness change in microns: median (q25–q75) RTM in microns: median (q25–q75)
Healthy Eyes
ERM Eyes, All ERM Eyes
ERM Eyes, Unchanged Symptoms
64 69.7 (29–88) 224 (125–294)
142 70.0 (50–88) 174 (118–267)
103 70.2 (60–86) 171 (111–267)
39 69.4 (52–88) 213 (124–352)
85 (82–90.3)
80 (75–85)
80 (75–86.5)
78 (75–83.5)
86 (83–90)
81 (74–86)
82 (75–87)
77 (69–83)
0.5 (⫺2–⫹3)
0 (⫺3–⫹3)
1 (⫺2–⫹3)
ERM Eyes, Worsening of Symptoms
⫺1 (⫺5.5–⫹1.5)
279 (256–296)
416 (341–470)
394 (315–456)
454 (400–482)
278 (254–295)
420 (338–472)
399 (316–457)
456 (402–501)
⫺1 (⫺3.5–⫹2) 140.2 (112–168)
0 (⫺7–⫹13) 301.5 (191–495)
⫺1 (⫺8–⫹9) 274.4 (171–483)
1 (⫺5.5–⫹26.5) 350.8 (252–574)
BCVA ⫽ best-corrected visual acuity; ERM ⫽ epiretinal membrane; ETDRS ⫽ Early Treatment Diabetic Retinopathy Study; q25 ⫽ lower 25% quartile; q75 ⫽ upper 75% quartile; OCT ⫽ optical coherence tomography; RTM ⫽ retinal tangential movement.
In each of the 9 squares, retinal vessel branch points were selected in the baseline picture, and x and y coordinates were recorded. Recognizable branch points with the greatest movement between the baseline and follow-up pictures were selected. Toggling between the layers allowed for identification of movement in retinal vessel branch points. The same branch points were selected in the follow-up picture, and x and y coordinates (pixels) were once again recorded. Vector lengths were calculated with Pythagoras’ theorem and converted to microns according to the manufacturer’s description of pixel-to-micron ratio. We added the 9 vector lengths in an eye into a single vector length to represent the total retinal vessel movement. To estimate the reliability of image alignment, choroidal vessel movement was measured. Retinal vessel movement was expected in ERM eyes, but choroidal vessel movement should not occur in healthy fellow eyes or ERM eyes. To confirm that we were examining retinal vessel movement and not optical artifacts, we controlled our retinal vessel movement against choroidal vessel movement. This was done in the same 9 subfields in a subset of patients with visible choroidal vessels. The current study obtained required approval from the Danish Data Protection Agency. This study was conducted from the screening material of the RCT approved by the regional ethical committee of the capital region of Copenhagen, Denmark, under H-C-2008-026 (registered at Clinicatrial.gov: NCT00902629). Statistic testing was performed on SigmaPlot 10.0 (2006 Systat Software Inc, Chicago, IL). Descriptive statistics were used to describe means and quartiles. Kruskal–Wallis nonparametric analysis of variance tests were used to compare anatomic subgroups. Ranked sum tests were used to compare groups, and linear regression models were used to describe correlations.
Results A total of 206 eyes in 113 patients were included distributed among 53 men and 60 women. Among the analyzed eyes, 64 were healthy and 142 had ERM. Three patients reported relief of symptoms at the follow-up examination and were pooled with the
1888
unchanged group. Subjective symptoms were unchanged in 103 ERM eyes, and 39 ERM eyes experienced worsening of symptoms between the 2 visits. Basic characteristics and BCVA and CMT measurements are shown in Table 1. The calculated retinal vessel movement vector lengths are shown in Figure 2 (available online-only at http:// aaojournal.org) for each anatomic subfield. To localize whether retinal movement was more predominant in 1 macular subfield compared with another, Kruskal–Wallis nonparametric analysis of variance on ranks was performed. This showed that in ERM eyes, the central subfield had a significantly lower ranking (greater retinal vessel movement) compared with the inferior nasal subfield (P⬍0.05) and the 3 superiorly located subfields (P⬍0.05). The same test in healthy eyes found no statistically significant differences (P ⫽ 0.138). To achieve a better estimate of the total retinal vessel movement, we used the sum of the length of the 9 recorded retinal vectors and denoted this sum as retinal tangential movement (RTM), which disregards the direction of the vector and only sums the vector lengths. The total RTM was statistically significantly greater in ERM eyes compared with healthy eyes (P⬍0.001) (Fig 3). No choroidal vessel movement was expected in any eyes. To control fundus image alignment, choroidal vessel movement was measured in 10 healthy and 13 ERM eyes. These were minimal with vector lengths between zero and 45.6 m in both healthy and ERM eyes. Kruskal–Wallis 1-way analysis on ranks found no statistically significantly difference between anatomic subgrids (P ⫽ 0.348) in choroidal vessel movement. The 9 anatomic subfield vector lengths were summed into a single total choroidal vessel movement vector. Ranked sum showed no statistically significant difference between choroidal vessel movement in healthy eyes and ERM eyes (P ⫽ 0.077), as shown in Figure 3. There was no difference between choroidal vessel movement in all eyes and retinal vessel movement in healthy eyes (P ⫽ 0.866). Ranked sum tests between choroidal vessel movement and retinal vessel movement in healthy eyes did not show significant differences (P ⫽ 0.87). Ranked sum test in ERM eyes comparing choroidal vessel movement and retinal vessel movement found significantly greater retinal vessel movements (P⬍0.001). There
Kofod and la Cour 䡠 Retinal Tangential Movement in ERM central thickness at both examinations. Among ERM eyes with worsening of symptoms, 25.6% lost more than 5 Early Treatment Diabetic Retinopathy Study letters, whereas ERM eyes with unchanged symptoms lost more than 5 letters in 9.7% of cases. Ranked sum tests showed that BCVA was significantly worse in patients demonstrating worsening at the follow-up (P ⫽ 0.002), and decrease in BCVA was also statistically significantly greater in ERM eyes with worsening of symptoms (P ⫽ 0.011). Correlation analysis was also performed using only the retinal vessel movement instead of RTM. This produced outcomes on par with RTM but with more uncertain estimates.
Discussion
Figure 3. Comparison of retinal tangential movement (RTM) in healthy eyes and epiretinal membrane (ERM) eyes.
was statistically greater retinal movement in ERM eyes compared with choroidal vessel movement measurements in both healthy and ERM eyes (P⬍0.001). The comparison of choroidal and retinal vessel movement indicated that the measured retinal vessel movement is indeed retinal movement and does not stem from image alignment problems. The RTM was further analyzed in ranked sum tests and linear regressions. Among healthy eyes, no significant correlation was found between RTM and BCVA, change in BCVA, CMT, or change in CMT (Table 2, available online-only at http://aaojournal.org). Comparison of all ERM eyes with healthy eyes showed significantly greater RTM in ERM eyes (P⬍0.001). The study found significant correlations between decreasing BCVA as CMT thickness increased at both baseline (coefficient ⫺4.66, P⬍0.001) and follow-up (coefficient ⫺4.84, P⬍0.001) among ERM eyes. Regression analysis between BCVA and CMT in patients with ERM showed a significant correlation between decline in BCVA and increase in CMT (P⬍0.001). A correlation was found between increasing RTM and increasing CMT at both baseline (coefficient ⫹0.02, P⬍0.001) and follow-up (coefficient ⫹0.084, P⬍0.001). The RTM increased statistically significantly as loss of BCVA between visits increased (coefficient ⫺1.49, P⬍0.001). The RTM correlated to time between visits but only in ERM eyes (P⬍0.001). The gain in RTM was 47.7 m/month in ERM eyes and 5.3 m/month in healthy fellow eyes (P ⫽ 0.073). Foveal dystopia was 27.4 m on average in healthy fellow eyes and 55.5 m in ERM eyes. There was a linear correlation between central grid retinal vessel movement and foveal dystopia with a coefficient of ⫹0.398 (P⬍0.001). There was no correlation between the foveal dystopia and the RTM (P ⫽ 0.751). Foveal dystopia did not correlate to BCVA change (P ⫽ 0.835), CMT change (P ⫽ 0.685), or time between examinations (P ⫽ 0.686). The ERM eyes were divided into a group with unchanged symptoms and a group experiencing worsening of symptoms. The RTM was significantly greater in patients with worsening of symptoms compared with patients with unchanged symptoms (P ⫽ 0.024) (Fig 4). In patients with worsening of symptoms, the CMT increase was 7.95 m (⫾29.60) compared with 2.28 m (⫾21.03) CMT increase in eyes with unchanged symptoms (P ⫽ 0.209). Ranked sum test showed a statistically significant difference in CMT at both baseline (P ⫽ 0.001) and follow-up (P⬍0.001), where patients who developed worsening of symptoms had thicker
Although commercially available OCT scanners can measure retinal thickness and view B-scan morphologic features in great detail, the presented option to view aligned fundus images adds a new method to evaluate ERM cases. When viewing aligned fundus images in toggle mode, it was apparent whether tractional changes associated with the ERM were present or not. This quantification of the retinal vessel movement illustrates the possibility of using aligned retinal images to track and describe RTM in ERM eyes over time. The study describes a method to quantify the retinal traction to enable comparison between ERM eyes and healthy fellow eyes. Significantly greater RTM was found in ERM eyes compared with healthy fellow eyes. The RTM was correlated to decreased BCVA and increased CMT. All findings were more excessive in patients who described subjective worsening of symptoms between the 2 visits. The automated alignment process of fundus images is not perfected, and minute changes were present even in healthy eyes. The type or method of image manipulation used in the process of creating aligned fundus images is known only by the manufacturer. To determine whether our measurements of dynamic retinal traction was indeed a measure of traction or a part of the AutoRescan process, we compared retinal and choroidal blood vessel movements. One could speculate that the AutoRescan alignment algorithm would rotate the image and induce artificial vessel
Figure 4. Comparison of retinal tangential movement (RTM) in epiretinal membrane (ERM) eyes with and without worsening of symptoms.
1889
Ophthalmology Volume 119, Number 9, September 2012 movement. If this were true, the choroidal vessels would move as much as the retinal vessels. Our results indicate that the choroidal vessel movements were equal to retinal movements in healthy eyes. Because the choroidal movements were statistically significantly less than the retinal movement in an ERM eyes, we believe that we are indeed measuring the retinal tangential traction associated with ERM on the retinal surface. The patients were asked about changes in metamorphopsia, but they may confuse changes in metamorphopsia with changes in visual acuity. This makes the assessment of changes in patients’ symptoms an uncertain measure. We present our best estimate of changes in symptoms because all examinations were performed by the same examiner using standardized questions during screening for an RCT, and the focus of the examination was to find changes in metamorphopsia and BCVA to allow participation in the RCT. A systematic questionnaire quantifying patients’ symptoms and a prospective design would have been preferable. Similar to Weinberger et al,6 we measured minor shifts in the retinal vessels of healthy eyes, which we contribute to the AutoRescan algorithm. We calculated the uncertainty of the alignment algorithm to be 242.8 m across 9 vessel measurements equal to the 95% fractile of RTM in healthy eyes. Movement in excess of this represented true retinal vessel movement. This was present in 58% of patients with unchanged symptoms and in 76% of patients describing worsening of symptoms. Observing substantial retinal vessel movement in ERM eyes even without changes to BCVA or CMT shows that ERM is a dynamic disease. Weinberger et al6 measured vessel displacement movements between major and minor vessel junctions in the upper and lower temporal arcades, as well as movement between disc and vessel junction temporal to the fovea. In contrast with their study, our applied subfield grid was placed automatically in each image, eliminating the need for manual corrections of grid location. The aligned pictures made it evident that multiple centers of traction existed in most eyes. The retinal movement did not have a consistent relationship to the fovea. Simple modeling of the observed complex retinal movement is a challenge. The assumption that the central grid retinal vessel movement would give a better estimation for changes in BCVA and CMT was tested. This showed equal correlations as presented above with RTM, but with a greater uncertainly of the estimate. This is perhaps not too surprising because stereopsis is not isolated to the fovea. A recent study found depth perception to be almost as accurate in 7 degrees eccentricity as centrally.10 The assumptions in ERM are that a centripetal focus exists over the fovea and that there is a central contraction of retinal tissue. Gupta et al11 have shown that multiple foci of retinal contractions exist in 20 of 44 examined eyes. These multiple centers of centripetal retinal vessel movement were also present in this material. With several centripetal foci located away from the fovea, the traction can seem to be centrifugal. Quantification of metamorphopsia by binocular correspondence perimetry in ERM and macu-
1890
lar holes finds metamorphopsia in ERM to be more chaotic and widespread.12,13 For this reason and the findings presented by Gupta et al, we elected to use the total length of the 9 subfield vectors, disregarding the direction of the individual vector. The retinal vessel movement associated with ERM can occur in all subfields, and the central grid will almost always be involved. This could explain the significantly lower ranking (greater vector lengths), but not mean vector lengths in the central subfield. Lo et al9 described severe foveal dystopia and found that the fovea returned 32.8%⫾22.1% toward its expected anatomic position after surgery. In the 2 cases they measured foveal movement preoperatively, the centripetal movement was smaller than the centrifugal movement after surgery. This indicates a continual contraction, which in the study by Lo et al must have had foci away from the fovea to allow such severe foveal dystopia. Foveal movement and retinal vessel movement in the central grid could be assumed to be more representative of changes in BCVA or changes in patients’ symptoms than retinal vessel movement in adjacent subfields. This was not the case in the present study in which exclusively analyzing the foveal retinal vessel movement gave a more uncertain estimate compared with RTM. We hypothesized that RTM played a significant role in progression of disease as the ERM convolutes the neurosensitive tissue. This was most evident in patients with worsening of symptoms in whom excessive RTM was present in the absence of increased CMT or decreased BCVA. This was the case in 41% of patients with subjective worsening. A weakness in this study was our selection of measuring points in which we chose to record traction in 9 central subfields in the macula. We found strong correlations between CMT and RTM and assumed this to be caused by the ERM, but the increase in CMT could be caused by other factors, such as axoplasmic flow obstruction.14 The study is also limited by its retrospective nature and the fact that the control group consisted of fellow eyes to ERM eyes. In patients with multiple rescans, only the 2 spaced furthest in time were compared. Patients experiencing worsening of symptoms are more likely to continue participating in controls, leading to a selection bias. Describing RTM evolution over several visits could strengthen the conclusions, but this was not performed because the majority of patients had only 2 examinations to compare. Despite many patients having stable BCVA and CMT, there was a strong correlation between time between examinations and RTM, suggesting that a continual, and in some patients a fast tractional, change is present in ERM eyes. This strengthens our hypothesis that patients with changes in the retina (i.e., greater RTM) will have a harder time adjusting to the constant changes and be more likely to experience distortions or want to undergo surgery. The symptomatology of ERM consists of visual acuity loss and metamorphopsia. For patients with good visual acuity but marked metamorphopsia, we lack an objective measurement that correlates with the patients’ subjective symptoms of metamorphopsia. This study describes a method to quantify retinal movement accompanying ERM and correlates this with the more traditional measures of
Kofod and la Cour 䡠 Retinal Tangential Movement in ERM ERM traction. We think this adds to the repertoire of our objective means of studying ERM. We do not think our method is immediately applicable for clinical use. However, the principle that retinal movements over a wider macular area are correlated with both function and symptoms is worthwhile to explore further. Future prospective studies may benefit from correlating retinal movement in ERM to symptomatic metamorphopsia or binocular symptoms. There is reason to believe that this will be a more accurate and objective correlate to these symptoms than CMT, which is more likely to be correlated with visual acuity. In conclusion, we present a method to quantitatively describe the retinal movement that accompanies ERM and correlate this to the more traditional measures of ERM traction. This quantification of RTM correlated well with changes in BCVA and CMT in ERM eyes, especially in patients with worsening of symptoms. Even in patients with unchanged symptoms, we found increased RTM over time.
References 1. Ting FS, Kwok AK. Treatment of epiretinal membrane: an update. Hong Kong Med J 2005;11:496 –502. 2. Massin P, Paques M, Masri H, et al. Visual outcome of surgery for epiretinal membranes with macular pseudoholes. Ophthalmology 1999;106:580 –5. 3. Michalewski J, Michalewska Z, Cisiecki S, Nawrocki J. Morphologically functional correlations of macular pathology connected with epiretinal membrane formation in spectral optical coherence tomography (SOCT). Graefes Arch Clin Exp Ophthalmol 2007;245:1623–31. 4. Treumer F, Wacker N, Junge O, et al. Foveal structure and thickness of retinal layers long-term after surgical peeling of idiopathic epiretinal membrane. Invest Ophthalmol Vis Sci 2011;52:744 –50.
5. Arimura E, Matsumoto C, Okuyama S, et al. Retinal contraction and metamorphopsia scores in eyes with idiopathic epiretinal membrane. Invest Ophthalmol Vis Sci 2005;46: 2961– 6. 6. Weinberger D, Stiebel-Kalish H, Priel E, et al. Digital red-free photography for the evaluation of retinal blood vessel displacement in epiretinal membrane. Ophthalmology 1999;106: 1380 –3. 7. Sayegh RG, Simader C, Scheschy U, et al. A systematic comparison of spectral-domain optical coherence tomography and fundus autofluorescence in patients with geographic atrophy. Ophthalmology 2011;118:1844 –51. 8. Serbecic N, Aboul-Enein F, Beutelspacher SC, et al. High resolution spectral domain optical coherence tomography (SD-OCT) in multiple sclerosis: the first follow up study over two years. PLoS One [serial online] 2011;6:e19843. Available at: http://www.plosone.org/article/info%3Adoi%2F10. 1371%2Fjournal.pone.0019843. Accessed February 29, 2012. 9. Lo D, Heussen F, Ho HK, et al. Structural and functional implications of severe foveal dystopia in epiretinal membranes. Retina 2012;32:340 – 8. 10. Devisme C, Drobe B, Monot A, Droulez J. Stereoscopic depth perception in peripheral field and global processing of horizontal disparity gradient pattern. Vision Res 2008;48:753– 64. 11. Gupta P, Sadun AA, Sebag J. Multifocal retinal contraction in macular pucker analyzed by combined optical coherence tomography/scanning laser ophthalmoscopy. Retina 2008;28: 447–52. 12. Kroyer K, Christensen U, Larsen M, la Cour M. Quantification of metamorphopsia in patients with macular hole. Invest Ophthalmol Vis Sci 2008;49:3741– 6. 13. Kroyer K, Jensen OM, Larsen M. Objective signs of photoreceptor displacement by binocular correspondence perimetry: a study of epiretinal membranes. Invest Ophthalmol Vis Sci 2005;46:1017–22. 14. Arroyo JG, Irvine AR. Retinal distortion and cotton-wool spots associated with epiretinal membrane contraction. Ophthalmology 1995;102:662– 8.
Footnotes and Financial Disclosures Originally received: August 15, 2011. Final revision: March 12, 2012. Accepted: March 13, 2012. Available online: May 26, 2012.
Grant Support: Danish Agency for Science Technology and Innovation, Synoptik Foundation, Velux Foundation, Bagenkop Nielsen Eye Foundation. Manuscript no. 2011-1232.
Department of Ophthalmology, Glostrup Hospital, University of Copenhagen, Denmark. Financial Disclosure(s): The author(s) have no proprietary or commercial interest in any materials discussed in this article.
Correspondence: Mads Kofod, MD, Department of Ophthalmology, Glostrup Hospital, Nordre Ringvej 57, DK-2600 Glostrup, Denmark. E-mail: madkof01@glo. regionh.dk.
1891