Article Text

Download PDFPDF
Systematic correlation of morphologic alterations and retinal function in eyes with uveitis-associated cystoid macular oedema during development, resolution and relapse
  1. Marion R Munk1,
  2. Christopher G Kiss1,
  3. Irene Steiner2,
  4. Florian Sulzbacher1,
  5. Phillipp Roberts1,
  6. Maria Kroh1,
  7. Alessio Montuoro3,
  8. Christian Simader1,3,
  9. Ursula Schmidt-Erfurth1
  1. 1Department of Ophthalmology, Medical University of Vienna, Vienna, Austria
  2. 2Section for Medical Statistics, Center for Medical Statistics, Informatics and Intelligent Systems, Medical University of Vienna, Vienna, Austria
  3. 3Vienna Reading Center, Medical University of Vienna, Vienna, Austria
  1. Correspondence to Dr Christopher Kiss, Department of Ophthalmology, Medical University of Vienna, Waehringer Guertel 18-20, Vienna A-1090, Austria; christopher.kiss{at}


Purpose To evaluate morphological changes due to uveitis-associated cystoid macular oedema (uvCME) and their impact on central retinal sensitivity (CRS) before and after intravitreal triamcinolone-acetonide (IVTA).

Methods 28 eyes with uvCME were examined with microperimetry and spectral-domain optical-coherence-tomography (SD-OCT) before and after IVTA. Microperimetry-maps were superimposed on SD-OCT and morphological-alterations were correlated point to point with CRS and followed-up for 3 months. The effects of morphological-alterations on CRS over time were evaluated with a linear mixed-model.

Results Mean-CRS increased significantly after IVTA (p=0.009). Proportion of cysts correlated negatively with corresponding CRS (estimate/95% CI −3.8dB/−6.6 to −0.9, p=0.011). Proportion of diffuse macular-oedema (DifME) had no significant effect on mean-CRS (−0.76dB/−4.9 to 3.3, p=0.71). The proportion of serous retinal detachment (SRD) had a borderline significant effect on mean-CRS (−9.5dB/−19.1 to 0.1, p=0.052), however the initial presence of SRD at baseline had no significant negative effect on mean-CRS (−1.3dB/−4.9 to 2.3, p=0.46). Patients with epiretinal-membrane showed lower mean-CRS than patients without (−3.3dB/−6.5 to −0.008, p=0.05). The lowest percentage of morphological-alterations was achieved 30 days post IVTA concordant to best visual-acuity (logMAR 0.16±0.26), while best mean-CRS was achieved 90 days post IVTA (16.9±1.8dB). Fixation-stability showed no significant improvement.

Conclusions UvCME Morphological-alterations were associated with specific CRS-decreases. DifME showed no significant- and SRD only a borderline effect on mean-CRS, which implicates that their presence should be considered when interpreting SD-OCT and making treatment-decisions.

  • Imaging
  • Macula
  • Treatment Medical

Statistics from

Request Permissions

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


Inflammatory retinal diseases often trigger the release of proinflammatory mediators, leading to increased vascular permeability, especially in the inner blood–retina barrier, resulting in cystoids macular oedema (CME) associated with visual function decrease.1 ,2

Some patients regain normal visual function. Others, despite total CME resolution, suffer further subjective visual impairment, which is sometimes hard to quantify with the distance visual function tests commonly used in clinical practice.3 Morphological and functional criteria are needed to evaluate efficacy and treatment outcome and to guide retreatment decisions. Currently, morphological features responsible for visual performance are the main basis for evaluation. Changes in distance visual acuity (VA) and central retinal thickness (CRT) are still used to assess treatment efficacy in clinical trials and daily practice. However, these two variables alone may not adequately reflect function and disease activity or indicate retreatment need.4–6 Defining which morphological alterations can cause visual impairment and respond to therapeutic intervention requires a point-to-point evaluation with microperimetry. Here, stimuli are placed in defined macular regions and exact retinal sensitivity is measured at each test point. Previous studies analysed retinal diseases using microperimetry and compared the results with those from optical coherence tomography (OCT) and different visual functional tests. These studies adjudged microperimetry as a useful tool for evaluating visual function.3 ,7–10

In this study, spectral domain (SD) OCT images were overlaid on retinal sensitivity maps to evaluate the impact of pathological alterations. The treatment response after intravitreal triamcinolone acetonide (IVTA) injection was examined at a point-to-point level by monitoring the influence of morphological changes in a tight longitudinal clinical setting. Looking at morphological alterations and their impact on vision impairment may help to improve understanding of uveitic CME (uvCME) by comparing morphological alterations and corresponding retinal functions and determining morphological variables associated with functional defects.


Patient selection and setting

This prospective observational study enrolled 28 eyes from 23 patients who had chronic recurrent uvCME for ≥3 months. Patients with media opacities, active uveitis (more than +1 anterior chamber or vitreous haze/cells according to the Standardization of Uveitis Nomenclature working group classification), optic nerve pathology, maculopathies other than CME and disintegrated retinal pigment epithelium, photoreceptor layer and external limiting membrane were excluded to assure good image quality, accurate grading and to eliminate confounding factors.11 Patients were recruited at the Department of Ophthalmology, Medical University of Vienna. The study adhered to the tenets of the Declaration of Helsinki, was registered at (NCT 01299129) and approved by the local ethics committee. Benefits and possible risks of study-related procedures were discussed with each patient who then gave informed consent. At every visit, patients were tested using 4 m Early Treatment Diabetic Retinopathy Study (ETDRS) best-corrected VA (BCVA), microperimetry, slit-lamp examination, intraocular pressure measurement, SD-OCT and indirect ophthalmoscopy. Visits were at baseline before therapy and 1 day, 1 week and 1–3 months after injection of 4 mg IVTA, following current standards.12

Imaging procedures

Fundus-controlled microperimetry was performed with MP-1 Microperimeter (Nidek Technologies, Navis, Austria). A Cartesian grid consisting of 41 stimulation loci (12° × 12°) was used (figure 1).9 The fixation mark presented as a 3° circle was adjusted for stable fixation and a 4-2-1 staircase strategy was used. The stimulus intensity ranged from 0 to 20 dB (0 dB refers to the strongest signal intensity of 127 cd/m2) in 1 dB steps. At the outset the light was dimmed, the fellow eye was covered and patients ran through a short test examination to familiarise themselves with the procedure. At the end of each examination a colour fundus picture was taken with a MP-1-associated xenon flash fundus camera. Follow-up function was used at subsequent visits to ensure each stimulation point was positioned at the same location.

Figure 1

Microperimetry grid used in the study, consisting of a central locus and three concentric box-shaped stimulation areas with 41 stimulation loci (12°×12°). Test points were numbered consecutively (1–41). After superimposing spectral domain optical coherence tomography B scans, underlying morphological alteration and corresponding central retinal sensitivity were recorded for each test point. Microperimetric data were evaluated using proprietary software calculating the mean central retinal sensitivity by weighting these data by area. Fixation stability within 2° and 4° according to the MP-1 microperimetry setup was also recorded.

Spectralis HD-OCT (Heidelberg Engineering, software was performed after microperimetry with dilated pupils. Volume scans with 37 B scans with 30 ART-frames of 20°×15° centred foveolarly were performed. The distance between each of the 37 scans was 120 µm. Using the follow-up function, scans were generated at the same retinal location at every visit.

Planimetric evaluation of morphological alterations and point-to-point correlation with retinal sensitivity

We used a software program (OCT toolkit, V.1.7), developed at the Vienna Reading Center, to relate functional sensitivity maps to SD-OCT. This software is able to display exported Spectralis XML data files and upload bmp and jpeg image files of imaging devices (in this case microperimetry maps with corresponding recognised stimulation points and an ETDRS grid).13 By displaying either the marked B scans or the ETDRS grid, stimulation points were evaluated according to their exact position on a section scan, underlying morphological alteration and location in the ETDRS subgrid (figure 2A–D). The following morphological alterations were graded: diffuse macular oedema (DifME), outer nuclear layer/Henle layer (ONL/HL) cysts, inner nuclear layer (INL) cysts, ganglion cell layer cysts (GCLCs) and serous retinal detachment (SRD). DifME was defined as CRT>250 µm or parafoveal/extrafoveal thickness >300 µm according to the ETDRS grid with hyporeflectivity in the ONL/HL and/or INL without cysts. Only stimulation points directly positioned on the B scans were included in the analyses. Each single test point of the microperimetry test grid was targeted consecutively and the corresponding central retinal sensitivity (CRS) together with the underlying morphological alterations and the exact location in the ETDRS grid (central (=1 mm), inner (=3 mm) and outer (=6 mm) circle) were registered for each point (figure 2A–D).

Figure 2

(A) The optical coherence tomography (OCT) toolkit software is able to display exported Spectralis XML data files and mark the area of interest. Morphological alterations were marked manually with one colour in each OCT section scan and then projected by the OCT toolkit software en face to the infrared (IR) picture. (B) With the OCT toolkit software the microperimetry map was uploaded and projected onto the IR picture at exactly the same location by marking the corresponding landmark points on the original IR picture and on the colour fundus and the sensitivity map. In this representative picture test point #11 with a threshold value of 9 dB was targeted by a single B scan representing the x-axis and by a y-slide. (C) Underlying morphological alterations are easily detectable by overlaying all marked areas on the IR picture again. In this case, test point #11 reveals an outer nuclear layer/Henle's layer cyst and serous retinal detachment, provoking a retinal sensitivity decrease down to 9 dB at this locus. (D) The ETDRS subgrid was projected over the IR picture and the microperimetry map. The location of the test point according to the ETDRS subgrid was then recorded. ETDRS, Early Treatment Diabetic Retinopathy Study.

The mean CRS increase was compared in relation to the ETDRS grid location and to the reduction in morphological alterations. Further, mean CRS weighted by area, fixation stability and distance VA was recorded for each visit. The mean CRS weighted by area was evaluated using a software program developed by one of the coauthors (CS) by summarising mean sensitivity values and weighting these data by area.14 ,15

Statistical methods

Statistical analysis was carried out with SAS V.9.2; diagrams were plotted with R 2.14.0 ( To analyse the effects of IVTA, morphological changes and the effect of an epiretinal membrane (ERM) and SRD at baseline on mean CRS, a linear mixed effect model was calculated. The mean CRS per patient was taken as the dependent variable. The independent variables time (continuous variable), proportion of cysts (proportion that at least one of the following morphological alterations occurred: inner nuclear layer cysts, ONL/HL cysts, GCLCs), proportion of DifME, proportion of SRD and presence of ERM and SRD at baseline were included in the model. Patient number was taken as a random factor (n=20). The mean CRS per patient and proportions were calculated over all investigated points per visit (maximal 41 points or 82 points if both eyes were enrolled). Further, the effect of IVTA on mean CRS weighted by area, on fixation stability within 2° and 4° and on distance VA (logMAR) (dependent variables) were analysed with a mixed model, including time (taken as a categorical variable) as an independent factor and the study eye within patients as a random factor. In the case of significant results, a comparison with baseline was done using contrasts. Multiple testing was adjusted by Dunnett's correction. Descriptive statistics are reported as mean±SD. p Values<0.05 were deemed statistically significant.


Description of the study sample

The study included 23 eyes of 20 patients. In five eyes, the microperimetry examination or an exact image overlay was not possible (9 men, 11 women, average age 47.8±17.5 years). Fifteen eyes had intermediate uveitis and eight anterior uveitis. Mean CME duration was 34.5±33.3 months. The majority did not have systemic involvement. Three patients had proven systemic inflammatory disease and received systemic immunosuppressant therapy (eg, anti-tumour necrosis factor α, methotrexate) before and during the study. Two patients received additional IVTA injections 2 and 3 months after the first injection, respectively. In these two cases, only the data collected up to the second IVTA injection were included in analyses. No patient showed anterior chamber relapse during the 3-month period or required additional systemic anti-inflammatory medications. Of the 23 eyes, 6 had an ERM and 14 SRD. All eyes had intact external limiting membrane, photoreceptor layer and retinal pigment epithelium.

The mean CRS weighted by area as well as distance VA increased significantly during the observational period. However, 2° and 4° fixation stability showed no significant improvement (tables 13).

Table 1

Mean central retinal sensitivity (CRS) weighted by area before and after intravitreal triamcinolone acetonide (IVTA) using a mixed model

Table 2

Mean fixation stability in microperimetry within 2° and 4° before and after intravitreal triamcinolone acetonide (IVTA) using a mixed model

Table 3

Mean distance visual acuity (logMAR) before and after intravitreal triamcinolone acetonide (IVTA) using a mixed model

The mean CRS increased significantly (estimate (95% CI) 0.03 dB (0.007 to 0.04), p=0.009). The proportion of cysts correlated negatively with mean CRS (−3.8 dB (−6.6 to −0.9), p=0.011). Thus, the more cysts that were present, the more the mean CRS decreased. The proportion of SRD had a borderline significantly negative effect on the mean CRS (−9.5 dB (−19.05 to 0.07), p=0.052), whereas the proportion of DifME did not significantly decrease the mean CRS (p=0.71). However, patients with initial presence of SRD on a test point at baseline did not differ significantly in mean CRS from patients without SRD (p=0.46). Patients with ERM at baseline, however, showed a lower mean CRS than patients without (−3.3 dB (−6.5 to −0.008), p=0.050). Mean CRS decreases due to different morphological alterations are shown in table 4.

Table 4

Morphological changes seen on HD-OCT and associated retinal sensitivity values (dB)

The proportions of affected test loci are shown in tables 5, 6 and figure 3A,B, respectively. The lowest proportion of affected test points was found 30 days after IVTA, concordant to highest distance VA (table 3). The best mean CRS, however, was achieved 90 days after IVTA (tables 1, 5 and 6, figure 3A,B).

Table 5

Mean central retinal sensitivity (CRS)±SD and percentage of affected test points presenting with at least one intraretinal morphological alteration (including outer nuclear layer/Henle layer, inner nuclear layer and ganglion cell layer cysts and diffuse macular oedema) or serous retinal detachment (SRD) at each visit

Table 6

Mean central retinal sensitivity (CRS)±SD and percentage of affected test points according to time point and 1°, 3° and 6° ETDRS subgrid

Figure 3

(A,B) Course of mean central retinal sensitivity (CRS)±1.96×SE (A) and mean proportion of intraretinal cysts (proportion of at least one of the following morphological alterations occurring: inner nuclear layer cysts, outer nuclear layer/Henle layer cysts and ganglion cell layer cysts, reflecting the mean number of respective cysts on respective test points)±1.96×SE (B) at baseline and during follow up after intravitreal triamcinolone injection.


This study evaluated the treatment response during 3 months following IVTA in uvCME on a point-to-point level, focusing on the initial impairment and on therapy-induced improvement of CRS correlated to different morphological alterations.

The initial mean CRS was comparable to published values in previous studies of CME.3 The mean differential threshold value of 13.63 dB in loci without pathology in our study was similar to previous studies.3 ,9 All patients in our study showed intact hyperreflective bands and no comorbidities such as microangiopathy or increased hydrostatic pressure. This may explain the slightly better mean threshold values in the non-pathological test points compared with previous studies in diabetic macular oedema (DME) or retinal vein occlusion. A previous study found a range of 13–17 dB threshold value in healthy subjects.13 This is similar to the mean CRS values of loci without pathologies in this study. Interestingly, loci with DifME had comparable mean sensitivity values to loci without underlying pathologies and to mean CRS values previously measured in healthy subjects.13 Thus, the presence of DifME in uvCME had no negative effect on retinal sensitivity. Similarly, ONL/HL swelling was not observed to have an impact on retinal sensitivity in a point-to-point study in DME.9 This finding and those of our study are also supported by a report showing no significant CRS decrease in DifME.16 Likewise VA was found to be nearly normal in patients with uvCME and DifME.17 As DifME is presumed to be one of the first steps in the development of macular oedema ,18 ,19 it seems possible that DifME does not cause a CRS decrease. Only when ME progresses and cysts develop will CRS be impaired. As already suggested by a previous study, intraretinal fluid itself does not seem to induce visual impairment. However, due to intraretinal cystoids spaces, retinal thickness increases and bipolar axons, connecting photoreceptors with ganglion cells, may be stretched and break.20 DifME made of either diffuse fluid accumulation or very small cysts, non-educible for HD-OCT, leads only to a moderate retinal thickness increase and therefore only to little traction on axonal interconnections.20

Whether CME with SRD causes a greater decrease in visual function than intraretinal cysts alone is uncertain. Some studies in this field have found a lower VA when SRD is present compared with intraretinal cysts alone,17 while others have found no differences.21 Actually, SRD seems to be associated with short disease duration and may be a positive predictive factor for morphological and functional outcome in uvCME.22–25 A similar point-to-point correlation study in DME demonstrated a mean CRS decrease of 2 dB with SRD lesions compared with loci without any pathology.9 In contrast, another study found no independent association between SRD and subnormal retinal sensitivity.16 In this study, the proportion of SRD had only a borderline significant effect on CRS and the initial presence of SRD had no effect on mean CRS at baseline. However, the descriptive evaluation on test points with SRD revealed lower CRS values compared with test points with no pathology. We believe that the non-significant effect of SRD on CRS in this evaluation may be an indirect effect and may be due to the fact that retinal thickness excluding subretinal fluid in CME is significantly greater in eyes with intraretinal cysts lacking SRD compared with eyes with intraretinal cysts and SRD. The greater the intraretinal thickness and the greater and more frequent the intraretinal cysts, the more axonal connections between the photoreceptors and the ganglion cells may be stretched and break and also the more the interstitial pressure will rise and intraretinal perfusion will be impaired, which will lead to visual function decrease.20 ,26–28 The effect of SRD on visual function will therefore be much smaller than the effect of larger intraretinal cysts and intraretinal thickness in CME without SRD. These suggestions may be further strengthened by a previous finding, which demonstrated that CRT excluding serous retinal fluid correlated better with CRS decrease than CRT including the subretinal space.28 Taken together, the cause and severity of CRS decrease due to SRD and its prognostic value remain uncertain. This in turn implicates that the presence of SRD should be considered when analysing and interpreting retinal thickness data.

One study found the CRS decrease to be dependent on ONL/HL cyst sizes.9 In our study, the proportion of cysts on respective test points had a negative effect on mean CRS decrease. Thus, the more intraretinal fluid there is, the more retinal sensitivity decreases. Soliman et al16 concluded that vision impairment in CME is induced by impediment of light due to extracellular fluid, whereas axonal disruption or a compartment syndrome due to intraretinal fluid accumulation may also explain these findings.20 ,27 ,28

Concordant to the best distance VA, morphological alterations were least prevalent 30 days post IVTA. However, despite cyst relapses, mean CRS was best at the last visit, 90 days after IVTA. These findings are in line with previous observations. Best distance VA was also found 30 days post IVTA in previous studies.12 ,29 In addition, former studies already demonstrated further CRS improvement despite CME relapses in retinal vein occlusion and after BCVA stabilisation in neovascular age-related macular degeneration.30 ,31 All these findings suggest delayed retinal sensitivity improvement compared with distance VA and morphological recovery. As previously discussed, CME may lead to expansion and disruption of axonal connections.20 Successive regeneration of impaired structures and therefore delayed retinal sensitivity improvement on respective loci seem reasonable. In distance VA, however, single letters can be projected to an arbitrary area of the macula, thus if angular resolution fails in one area, an adjacent area can substitute for respective dysfunction. Distance VA would therefore increase concordant to morphological recovery. CRS, however, is tested with a fixation mark, preventing alternative fixation. Therefore CRS improvement resembling the local functional status may lag behind. Thus, assessing CRS may be a valid outcome parameter for treatment efficacy in retinal diseases.

In summary, cystic morphological alterations in uvCME were associated with specific retinal sensitivity decreases, whereas DifME had no significant effect and SRD only a borderline significant effect on retinal sensitivity. This implicates that the presence of SRD and DifME should be considered when interpreting morphological and corresponding retinal thickness data and when making treatment decisions.


We want to thank Professor Lee Jampol and Dr Rene Rückert for their careful reviewing and editing of the manuscript. The authors have obtained written permission from all persons named in the acknowledgement.



  • Contributors Conception and design: MM, CK, FS, IS and CS. Data acquisition: MM, FS, PR, MK and AM. Data analysis and interpretation: MM, IS, MK, AM, CS and USE. Drafting the article: MM. Revising it for intellectual content: MM, CK, CS, IS and USE. Contributing to statistical analysis: MM, IS, FS, AM and CS. Administrative, technical or material support: MM, IS, FS, PR, CS, CK and USE. Supervision: CK and USE.

  • Competing interests None.

  • Patient consent Obtained.

  • Ethics approval Ethics commission of the Medical University of Vienna.

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