Michael J. Koss, MD Department of Ophthalmology, Section of Vitreo-Retinal Surgery Hospital of the Goethe University, Frankfurt am Main Theodor Stern Kai 7, 60590 Frankfurt am Main Germany Tel: + 49 69 6301 5649 Fax: + 49 69 6301 5621 Email: Michael.Koss@me.com
Purpose: To compare cytokines in undiluted vitreous of treatment-naïve patients with macular oedema without vitreomacular traction secondary to branch (BRVO), central (CRVO) and hemi-central (H-CRVO) retinal vein occlusion.
Methods: Ninety-four patients (median age 72 years, 42 men) underwent an intravitreal combination therapy, including a single-site 23-gauge core vitrectomy and the application of bevacizumab and dexamethasone due to vision-decreasing macular oedema. Among these were 43 patients with BRVO, 35 with CRVO and 16 patients with hemi-CRVO, which were distributed in a fresh or old retinal vein occlusion type (seven or more months after onset). Undiluted vitreous samples were analysed for interleukin 6 (IL-6), monocyte chemoattractant protein-1 (MCP-1) and vascular endothelial growth factor (VEGF-A) with cytometric BEAD assay. Vitreous samples from patients with idiopathic epiretinal membrane served as controls (n = 14).
Results: The mean cytokine values were highest in the CRVO group with IL-6 = 64.7 pg/ml (SD ± 115.8), MCP-1 = 1015.8 pg/ml (±970.1) and VEGF-A = 278.4 pg/ml (±512.8), followed by the H-CRVO group with IL-6 = 59.9 pg/ml (SD ± 97.5), MCP-1 = 938.8 pg/ml (±561.1) and VEGF-A = 211.5 pg/ml (±232.4). The BRVO group had IL-6 = 23.2 pg/ml (SD ± 48.8), MCP-1 = 602.6 pg/ml (±490.3) and VEGF-A = 161.8 pg/ml (±314.4). The values of MCP-1 and VEGF-A were significantly different for CRVO or H-CRVO versus BRVO. All values were significantly higher than in the control samples, which had 6.2 ± 3.4 pg/ml (IL-6), 253 ± 74 pg/ml (MCP-1) and 7 ± 4.9 pg/ml (VEGF-A). Within the old RVO type, only MCP-1 was significantly different for CRVO or H-CRVO versus BRVO.
Conclusions: Both inflammatory markers and VEGF-A were higher in CRVO and H-CRVO than in BRVO undiluted vitreous samples. It seems that monocyte recruitment to the vessel wall, which might underlie the importance of eosinophils in tissue remodelling after RVO, is of special interest owing to the significant difference in MCP-1 in the older RVO types.
Retinal vein occlusion (RVO) is the second most common and important retinal vascular disease after diabetic retinopathy. In most cases, RVO is caused by a reduction in venous flow, which manifests itself either as a central (CRVO) or branch retinal vein occlusion (BRVO; Shahid et al. 2006; The Eye Disease Case-Control Study Group1993, 1996).
Associated macular oedema (ME) can appear after multifactorial pathophysiologic changes, with a breakdown of the blood–retinal barrier owing to the compromised capillary flow, which disturbs the balance of angiogenic and inflammatory cytokines in ocular fluid (Funk et al. 2009; Noma et al. 2009, 2010). Cytokines therefore play a crucial role in the functioning of endothelial cells and leucocytes, which contribute to cytokine secretion (Jo et al. 2003; Ryan 2006). The long-term exposure of ECs to proinflammatory cytokines can lead to leucocyte extravasation and thrombosis, in addition to the originating retinal venous obstruction. Thus, inflammation and vascular dysfunction interact with each other and are stimulated predominantly by vascular endothelial growth factor (VEGF), a known chemoattractant cytokine for macrophages and leucocytes. An upregulation of VEGF mRNA has previously been described in the context of CRVO (Pe’er et al. 1998).
It was previously reported that anti-VEGF monotherapy influences the expression of various molecules, such as VEGF and inflammatory markers, including interleukin 6 (IL-6) and monocyte chemoattractant protein 1 (MCP-1; Funk et al. 2009). This is important because IL-6, the main stimulator of most acute-phase proteins, itself promotes VEGF upregulation (Cohen et al. 1996). IL-6 is essential for the transition from acute to chronic inflammation and thus links the inflammatory process with angiogenesis. In RVO, elevated vitreous levels of stimulatory cytokines, such as VEGF and IL-6, and inhibitory cytokines, such as PEDF, are correlated with the severity of macular oedema (Noma et al. 2011b). Noma et al. (2009) have demonstrated elevated levels of IL-6 and VEGF and recently investigated the role of the soluble VEGF receptor in patients with CRVO (Noma et al. 2011a). It is noteworthy that all of their vitreous samples were acquired by complete three-port vitrectomy procedures due to the relief of vitreomacular traction and were analysed by ELISA.
Recent advances in spectral domain optical coherence tomography (SD-OCT) have allowed more detailed observations of macular oedema and have contributed to our understanding of the pathomorphology of RVO-associated ME (Yamaike et al. 2008; Tsujikawa et al. 2010).
A new, innovative technique, the cytometric bead array (CBA) technology, allows quantitative analysis of multiple markers in various specimens, requires a smaller sample volume and is time- and cost-effective. Cytometric bead array technology is based on flow cytometry, which is an analytical tool that allows for the discrimination of different particles on the basis of size and/or colour in various fluids (Morgan et al. 2004; Eickmeier et al. 2010) and has been utilized in ophthalmology before (Maier et al. 2006; Funk et al. 2009; Yoshimura et al. 2009; Kaneda et al. 2011). Kaneda et al. (2011) characterized with CBA the expression of IL-6 in BRVO but correlated these with VEGF and MCP-1 expression results analysed with ELISA.
The purpose of this study was to assess inflammatory and angiogenic cytokine levels using a CBA on undiluted vitreous samples of treatment-naïve patients with ME secondary to RVO without vitreomacular traction. To the best of our knowledge, this is the first quantitative study on intraocular cytokine levels performed with CBA comparing BRVO, CRVO and hemi-CRVO.
Samples were collected after the approval from local institutional review board (57/08) of the Goethe University Frankfurt am Main (Germany) in accordance with the European Guidelines for Good Clinical Practice and the Declaration of Helsinki of 1975 (sixth revision, 2008). Informed consent was obtained from each patient before the start of therapy.
The inclusion criteria were clinically significant macular oedema involving the fovea due to a RVO; compromised visual acuity of not worse than 2.0 logMAR and a central retinal thickness of not more than 1000 μm, measured with standard SD-OCT with typical clinical signs of a RVO in fundus photography/fluorescein angiography (intraretinal flame-shaped bleeding).
Exclusion criteria included presence of vitreomacular traction or neovascular complications defined as rubeosis iridis, vascularization of the anterior chamber and neovascularization on the disc or elsewhere; previous intravitreal drug injections; photolaser coagulation; previous vitrectomy; other intraocular surgery on both eyes, including cataract surgery on the partner eye in the last 6 months; signs of glaucoma, diabetic retinopathy, intraocular inflammation or trauma; participation in any clinical trial; use of immunosuppressive drugs or history of malignant tumours of any location.
Subjective clinical parameters
Best-corrected visual acuity was measured with the Early Treatment Diabetic Retinopathy Study charts at 5 m with the false choice technique, with a minimum of three of five optotypes. If less than thirty letters were read, the study chart was read at a 1-m distance. Best-corrected visual acuity was indicated as the logarithm of the minimum angle of resolution (LogMAR). Slit lamp examination, measurement of intraocular pressure, ophthalmoscopy and fundus photography were performed before treatment.
Objective clinical parameters
The macular region was examined in detail to evaluate morphologic changes in the retina using (SD-OCT; 3D OCT-2000®; Topcon, Tokyo, Japan). The SD-OCT performs measurements at an A-scan speed of 27.000 A-scans/second with a scan depth of 2.3 mm, a horizontal resolution of 20 μm and a longitudinal resolution of 5–6 μm.
Fluorescein angiography was asses- sed by a certified medical photographer to assess ischaemic retinopathy with >5 disc areas of nonperfusion in BRVO and with >10 disc areas in CRVO and H-CRVO (each eye was thus classified as ischaemic or nonischaemic with OIS WinStation 11K™; CCS Pawlowski GmbH, Jena, Germany).
Sample collection and preparation
A limited core pars plana vitrectomy was performed using a single-site 23-gauge vitrector (Intrector®; Insight Instruments, Stuart, FL, USA), which has two separate channels for aspiration and infusion. After conjunctival displacement, an oblique sclerotomy was performed to illuminate the tip of the vitrector with a headset and a magnifying 28-dioptre lens in the midvitreous cavity, in an antiseptic operating room environment. An assistant then aspirated a total of 0.5–0.7 ml of undiluted vitreous fluid and cut midvitreous and posterior vitreous, as instructed by the surgeon, who controlled against clinically relevant perioperative hypotonia. Thus, a minimum sample volume of 0.5 ml of vitreous was aspirated for all patients. At the end of the limited posterior core vitrectomy, subsequent isovolumetric substitution of balanced salt solution (BSS®; Alcon, Freiburg, Germany), 1.25 mg (0.1 ml) of bevacizumab (Avastin®; Genentech, San Francisco, CA, USA) and 0.8 mg (0.2 ml) of dexamethasone (Dexa-ratiopharm®, Ulm, Germany) was injected, in adherence with the principle of combination therapy (Koss et al. 2010).
Undiluted vitreous samples from 3-port vitrectomies performed for macular peeling due to idiopathic epiretinal membrane served as controls. A volume of 0.8 ml was aspirated before the infusion line was set active. The undiluted vitreous samples were isolated before drug application and rapidly frozen at −80°C. There is no significant influence of intermittent frozen storage on the retest reliability of CBA measurements (data under submission).
Cytometric BEAD assay (CBA)
The amounts of IL-6, MCP-1 and VEGF-A were determined using the cytometric bead array system with flex sets (BD™, Heidelberg, Germany). Experiments were carried out according to the manufacturer’s instruction manual. In brief, 50 μl of each vitreous body sample was incubated for 1 hr with the appropriate amounts of detection beads, which were specific for each investigated factor. Afterwards, samples were incubated for 2 hr with detection reagent, which again was specific for each detection bead used. Samples were measured on a FACSArray™ Bioanalyzer and analysed by FCAP array software (both by BD™). The amounts of VEGF, IL-6 and MCP-1 were calculated using a specific standard curve (in pg/ml).
Data were stored in EXCEL® (Microsoft Office 2010, Redmond, VA, USA) and analysed using Bias® software (Version 8.3.8; Epsilon, Darmstadt, Germany) for Windows. All tests were performed at an error level of 5%. Data are expressed as the mean ± standard error and a 95% confidence interval. David’s test was used to check the distribution of the data. These data were qualified as nonparametrically distributed.
Comparisons within one RVO group were analysed with the Wilcoxon–Mann–Whitney test. Comparisons between the RVO groups were made with the Kruskal–Wallis test for independent groups with multiple Dunn comparisons. A p value of <0.05 was considered significant.
Table 1 summarizes the demographic and clinical characteristics of the RVO groups and the epiretinal membrane (ERM) control group. The control group consisted of 14 patients with idiopathic epiretinal membrane, which was not associated with diabetic retinopathy, uveitis or previous laser/cryocoagulation. The only statistically significant difference between the control group and the study subgroups was the visual acuity for CRVO/hemi-CRVO and the control groups (p < 0.05 Kruskal–Wallis test).
Table 1. Epidemiology: Values in mean (±standard deviation), blood pressure in mmHg, duration of RVO in months, ischemic signs, including Cotton Wools, massive intraretinal haemorrhage, enlarged foveolar avascular zone, capillary dropouts, area of non-perfusion >5 disc areas for branch retinal vein occlusion (BRVO) and >10 disc areas for central retinal vein occlusion (CRVO), old type of RVO = seven or more months after onset.
* Significant difference was tested with the Kruskal–Wallis test with p < 0.05.
Age in years
69.7 ± 12.9
71.1 ± 11.7
70.2 ± 12.5
66.2 ± 7.9
144 ± 17
141 ± 18
143 ± 13
82 ± 17
83 ± 14
84 ± 12
VA in logMar
0.85 ± 0.51
1.28 ± 0.59*
1.1 ± 0.42*
0.51 ± 0.22*
Duration of RVO
6.7 ± 4.0
7.4 ± 3.5
5.7 ± 1.2
The mean IL-6 value for all patients with BRVO was 23.2 ± 48.8 pg/ml (CI 8.2–38.2), for all patients with CRVO was 64.7 ± 115.8 (CI 24.9–104.4, Table 2) and for all patients with hemi-CRVO was 59.9 ± 97.5 (8–112). The mean IL-6 value for the control patients was 6.2 ± 3.4 (CI 4.9–7.5), which was significantly higher in the CRVO (p < 0.004) and in the hemi-CRVO groups (p < 0.0009). IL-6 in the BRVO samples was smaller as in the CRVO group (p = 0.06) and significantly smaller than in the hemi-CRVO group (p < 0.02). There was no difference between the CRVO and the hemi-CRVO groups (p = 0.33).
Table 2. Characteristics of the spectral domain optical coherence tomography (SD-OCT) measurements (please see Fig. 1). All measurements are depicted as mean values ± standard deviation (SD) in micrometre.
The 95% confidence interval values are depicted in brackets; CMT, central macular thickness, TRT, total retinal thickness; TNeuro, thickness neurosensorium; SRT, subretinal thickness; dIS/OS, discontinued inner and outer photoreceptor segment band in the fovea; dELM, discontinued external limiting membrane band; Cysts, intraretinal cysts; HR spots, hyperreflective spots. The absolute mean values of the vitreous cytokines ± standard deviation (SD) are in pg/ml with the 95% confidence interval values in brackets. W–M–W = Wilcoxon–Mann–Whitney test.
23.2 ± 48.8 (8–38)
6.2 ± 3.4 (4.9–7.5)
64.7 ± 115.8 (25–104)
6.2 ± 3.4 (4.9–7.5)
59.9 ± 97.5 (8–112)
6.2 ± 3.4 (4.9–7.5)
BRVO versus CRVO; p = 0.06; BRVO versus hemi-CRVO; p < 0.02; CRVO versus hemi-CRVO; p = 0.33
602.6 ± 490.3 (451–754)
253 ± 74 (225–282)
1015.8 ± 970 (683–1349)
253 ± 74 (225–282)
939 ± 561.1 (640–1238)
253 ± 74 (225–282)
BRVO versus CRVO; p < 0.001; BRVO versus hemi-CRVO; p < 0.05; CRVO versus hemi-CRVO; p = 0.97
161.8 ± 314.3 (65–259)
7 ± 4.9 (5.1–8.9)
278.4 ± 512.8 (102–455)
7 ± 4.9 (5.1–8.9)
212 ± 232 (88–335)
7 ± 4.9 (5.1–8.9)
BRVO versus CRVO; p < 0.05; BRVO versus hemi-CRVO; p < 0.05; CRVO versus hemi-CRVO; p = 0.67
The mean MCP-1 value was 602.6 ± 490.3 pg/ml (CI 451.7–753.5) in the BRVO group, 1015.8 ± 970.1 (CI 682.5–1349) in the CRVO group and 939 ± 561.1(CI 640–1238) in the hemi-CRVO group. It was 253 ± 74 (CI 225–282) in the control group, which was significantly smaller than in the three study groups. Both comparisons of the BRVO with the CRVO values and the hemi-CRVO values were significantly different (p < 0.001 and p < 0.05, respectively). No significant difference was noticed for the comparison of the CRVO with the hemi-CRVO values.
In the BRVO group, the VEGF-A level was 161.8 ± 314.3 pg/ml (CI 65.1–258.6); in the CRVO group, it was 278.4 ± 512.8 (CI 102.2–454.5) and in the hemi-CRVO group, it was 212 ± 232 (88–335), which was significantly higher in all three groups than in the control group (VEGF-A 7 ± 4.9; CI 5.1–8.9). The BRVO VEGF-A values were both significantly smaller than the CRVO and the hemi-CRVO VEGF-A values (p < 0.05). CRVO and hemi-CRVO values were without significant difference.
Old versus fresh RVO type
The mean absolute values (±standard deviation) distributed in fresh (onset of RVO earlier than 7 months) and older RVO types are depicted in pg/ml in Table 3. There was no significant difference comparing the fresh with the old RVO type values, and no significant difference comparing the cytokine values of the fresh RVO types. Comparing the old RVO cytokine values, a significant difference was noticed for MCP-1, which was smaller in the BRVO group versus the CRVO (p < 0.03) and the hemi-CRVO (p < 0.001) groups.
Table 3. All values are depicted as mean values with minimum and maximum in brackets, W–M–W is the abbreviation for Wilcoxon–Mann–Whitney test, to analyse statistical differences between the fresh and the old absolute values within the branch retinal vein occlusion (BRVO), central retinal vein occlusion (CRVO) or hemi-CRVO groups. The Kruskal–Wallis test was applied to test for differences for cytokine values between the three study groups within the fresh or old RVO type
28.9 ± 60.3 (5–53)
13.8 ± 14.3 (6–21)
63.8 ± 115 (4–123)
66 ± 119.5 (6–125)
62 ± 109 (−3–128)
49 ± 29 (−22–120)
No statistically significant difference in the IL-6 values comparing the fresh or the old subgroups of each RVO type
643.4 ± 576 (416–871)
534 ± 299.9 (374–694)
875 ± 648 (541–1208)
1149 ± 1203 (551–1748)
799 ± 500 (497–1101)
1545 ± 428 (482–2607)
No statistically significant difference in the MCP-1 values comparing the fresh or the old subgroups of each RVO type; within the old subgroups, the BRVO value was significantly smaller in comparison with CRVO (p < 0.03) and with hemi-CRVO (p < 0.001)
195 ± 372 (48–342)
107 ± 179 (12–202)
320 ± 629 (−4–643)
239 ± 387 (47–432)
197 ± 204 (73–320)
276 ± 385 (−681–1234)
No statistically significant difference in the VEGF-A values comparing the fresh or the old subgroups of each RVO type
The levels of interleukin-6 (IL-6) and vascular endothelial growth factor (VEGF) in vitreous fluid have previously been evaluated with ELISA and were correlated with the severity of macular oedema secondary to branch and central retinal vein occlusion (Noma et al. 2008, 2009). Intravitreal anti-VEGF monotherapy clinically appears to be a safe and effective treatment in these patients, despite its short-term effectiveness and high recurrence rate (Stefansson 2008; Prager et al. 2009). Combination therapy modalities may thereby be more effective in addressing pathophysiologic components (neovascularization and inflammation), resulting in red-uced retreatment rates (Ehrlich et al. 2010). We have previously suggested a combination therapy with anti-VEGF agents and steroids, including a single-site limited vitrectomy (Koss et al. 2010). Such an approach provides the opportunity to evaluate vitreous cytokine levels in treatment-naïve patients with RVO. Using the CBA technique, we were able to efficiently test for three cytokines in 0.5-ml undiluted vitreous fluid samples, which would allow for a maximum of one to two cytokine analyses when using ELISA.
We found that both the inflammatory markers, IL-6 and MCP-1, and VEGF were higher in patients with RVO than in control patients with idiopathic ERM without diabetic retinopathy (Figs 1–3). This is important as the aetiology of ERM is until now unclear but seems to be rather associated with a mismatch of glial cell and fibroblast expression (Mandelcorn et al. 2003). In comparison with the study by Noma et al. (2008), who previously evaluated IL-6 and VEGF in BRVO with ELISA, the mean vitreous level of IL-6 and of VEGF-A was slightly lower (Δ−17 and Δ−69 pg/ml) in our patients with BRVO. In CRVO, they previously evaluated IL-6 and VEGF with ELISA, and the mean vitreous level of IL-6 was slightly higher (difference of Δ + 14 pg/ml) and the VEGF-A level was lower (Δ−157 pg/ml) than in our patients with CRVO (Noma et al. 2009). This may be explained by the shorter duration of CRVO in our study and is probably not explained by the differences in the detection methods. There is only one study, which demonstrated that the CBA results were comparable to the ELISA results in vitreous samples from one cohort of patients with diabetic retinopathy (Maier et al. 2006). Due to restrictions in their study design, which was not an escalation study with known protein concentrations, a concentration-based study with known concentrations would have been more appropriate. Comparisons between CBA and ELISA need to be analysed with caution, and more research should be performed in this aspect. Yoshimura et al. (2009) examined earlier the values of undiluted vitreous of BRVO and CRVO patients with CBA but different BEADS (LUMINEX).They demonstrated higher cytokine values for IL-6 Δ + 43 pg/ml and similar ones for VEGF-A Δ−5 pg/ml in BRVO. Our MCP-1 results were in contrast higher with a Δ + 1891 pg/ml than in the study of Yoshimura et al. In CRVO, they demonstrated higher cytokine values for all three cytokines (for IL-6 Δ +920 pg/ml; for MCP-1 Δ +1891 pg/ml and for VEGF-A Δ +1357 pg/ml). It is questionable whether their group of patients, who also qualified for a three-port pars plana vitrectomy, is comparable to ours, as they totally examined less patients and these yielded cytokine values with high confidence intervals. Very little is known about the role of MCP-1 in the pathophysiology of RVO. MCP-1 is a potent eosinophilic chemotactic cytokine and plays an important role in monocyte recruitment to the vessel wall after vascular injury, which might underlie the importance of eosinophils in tissue remodelling after RVO (Schober & Zernecke 2007). Additionally, MCP-1 induces VEGF-A gene expression by upregulating hypoxia-inducible factor 1 (HIF-1 alpha; Hong et al. 2005). Funk et al. (2009) examined aqueous samples with CBA at baseline prior to intravitreal anti-VEGF therapy in five patients 16 months after the onset of nonischaemic CRVO. IL-6 (Δ−19 pg/ml), MCP-1 (Δ−913 pg/ml) and VEGF levels (Δ−627 pg/ml) were all higher than in our study, which might be explained by the following differences: (i) the longer duration of CRVO, as MCP-1 is particularly important for tissue remodelling (Wong et al. 1991; Hoshino et al. 2001), and (ii) the different fluid samples taken (aqueous versus undiluted vitreous).
To the best of our knowledge, there has been no evaluation, neither with ELISA nor with CBA, of intraocular cytokines in hemi-CRVO. We could demonstrate cytokine values that were more similar to the CRVO values than to the BRVO values. The difference in the BRVO values was significant, even for IL-6, which was slightly not significant between CRVO and BRVO. If we consider that in hemi-CRVO, clinically half of the retinal tissue volume than in CRVO is compromised, we may expect a higher cytokine load in the vitreous. Thus, it seems that cytokine expression is driven not only in a tissue-dependent way but takes into consideration the duration of RVO, lens status, rest-vascularization of the tissue and various other factors like the age of the patient. A multivariate analysis warrants a higher number of RVO samples than in our study and should be performed in the future with a standardized protocol. Therefore, we looked into the influence of duration of the RVO and distributed the three RVO groups in fresh and old types, when there were more than 7 months after onset of disease. We could not find any significant difference. Comparing the values of only the three old RVO types with each other, however, we could find a statistical difference for MCP-1, which was in BRVO significantly smaller than in the CRVO or in the hemi-CRVO type. It seems that monocyte recruitment to the vessel wall, which might underlie the importance of eosinophils in tissue remodelling after RVO, is of special interest owing to the significant difference in MCP-1 in the older RVO types.
The shortcomings of our study are undoubtedly the small sample size, which may have allowed us to overlook significant differences, especially in the hemi-CRVO groups. Further studies at early and later time-points after RVO onset and after different interventions (monotherapy versus combination therapy) are therefore warranted to optimize our knowledge of the interrelation of intraocular pharmacology and pathophysiology (Stefansson & Loftsson 2006).
The comparison of three signature cytokines in undiluted vitreous of three subgroups of treatment-naïve RVO demonstrated that both inflammatory markers and VEGF-A were higher in CRVO and H-CRVO than in BRVO. It seems that monocyte recruitment to the vessel wall, which might underlie the importance of eosinophils in tissue remodelling after RVO, is of special interest owing to the significant difference in MCP-1 in the older RVO types.
The Adolf Messer Foundation, Königstein, Germany, supports the research group.