Multiplex Cytokine Detection versus ELISA for Aqueous Humor: IL-5, IL-10, and IFNγ Profiles in Uveitis

Abstract

purpose. The purpose of this study was to determine levels of IL-2, -4, -5, -10, TNF-α, and IFN-γ in aqueous humor (AH) from patients with active panuveitis, anterior uveitis (AU), and noninflammatory controls by using a flow cytometric mutiplex array (CBA) and to compare with results from ELISA.

methods. Pooled normal AH was spiked with six cytokines at decreasing concentrations for evaluating the CBA. AH was also obtained from 10 controls (cataract patients) and 36 patients with active uveitis. Cell-free supernatants were added to a cocktail of capture beads and detector antibodies or to antibody-coated wells for CBA and ELISA determination, respectively.

results. CBA demonstrated greater sensitivity for detecting IL-4, IL-10, and TNF-α than with ELISA. Increased IFN-γ was detected in both AU and panuveitis groups compared with controls (P < 0.01). IL-10 was higher in the panuveitis group on steroids (P < 0.01). IL-5 was detected in the control (P < 0.01) and AU groups (P < 0.05) but was undetectable in the panuveitis group (n = 10). Correlations between IFN-γ and IL-10 were found in all uveitis groups (P < 0.01) but not in controls, whereas TNF-α correlations with IL-4/IFN-γ were obtained in controls but not in the uveitis groups (P < 0.01).

conclusions. It was possible to measure cytokines titrated into normal AH specimens by CBA, and a greater number of cytokines were detected with increased sensitivity than with ELISA. Elevated IFN-γ in active uveitis and decreased IL-5 in posterior uveitis suggest Th1 polarity is more marked, with greater uveal tract involvement. The increased IL-10 in the steroid treated group suggests glucocorticoid-induced IL-10 upregulation.

Uveitis is an umbrella term for intraocular inflammation and encompasses a wide variety of clinical disorders, including both mild and severe types of potentially sight-threatening intraocular inflammation. ¹ Anterior uveitis (AU) is an inflammatory disorder of the anterior structures of the uveal tract, including the anterior chamber (AC), the iris, and the ciliary body. It can be part of a well-defined clinical syndrome, such as sarcoidosis, or a localized ocular disorder, such as idiopathic AU (IAU) or Fuchs’ heterochromic cyclitis (FHC). ² Acute anterior uveitis (AAU) is the most common form, accounting for 75% of all uveitis cases, with a yearly incidence of 8 per 100,000 in the general population. ^{3 4} Panuveitis affects the entire uveal tract. 

Immunostaining of aqueous humor (AH) has demonstrated infiltration mainly by neutrophils but also lymphocytes, plasma cells, and fewer numbers of eosinophils and mast cells. Despite these similarities in cellular infiltration, the response to treatment varies with disease. The phenotype of the infiltrating cells and the cytokines they secrete have been studied by using various methodologies, including ELISA and intracellular cytokine staining. ^{2 5 6 7} AH from patients with IAU, compared with noninflammatory controls, had significantly increased levels of CD4⁺ T cells and decreased IL-10 but not if associated with a systemic disease. ⁷ AH cytokine profiling has also shown higher levels of both IFN-γ and IL-10 but with decreased IL-12 in FHC compared with IAU. ² Cytokines in vitreous humor of patients with intermediate uveitis were investigated via T-cell line expansion and have been shown to parallel those in the AH of IAU, thus validating AH for investigating basic inflammatory mechanisms occurring in panuveitis. ⁵ In addition, AH-derived T-cell expression of IFN-γ and IL-10 in AAU and panuveitis has been detected by intracellular cytokine staining and flow cytometry. ⁶ However, these techniques require the T cells to be stimulated ex vivo to express cytokines that may or may not reflect what is occurring in vivo. 

One of the limitations of testing AH is that only a small sample volume can be obtained (typically 50 to 150 μL), thus only one to two cytokines can be assayed by using traditional ELISA techniques. Cytokines found in AH from patients with uveitis include IL-2, -6, -10, -12, IFN-γ, TGF-β2, TNF-α, and macrophage migration inhibitory factor (MIF). ^{2 7 8 9 10 11 12 13} Even when using sequential ELISA, levels of only two or three cytokines per sample can be measured. The alternative is to use pooled samples, which can mask individual patient differences. With the advent of flow cytometric bead-based technology, multiple cytokine analytes can now be more rapidly quantified simultaneously in individual samples, with better reproducibility and sensitivity than ELISA. ¹⁴ Up to 15 cytokines in a 100-μL volume have been accurately measured. ¹⁵ The cytometric bead array (CBA) system has been tested by using cell-culture supernatant, human serum, tears, and nasal lavage. ^{14 16} In this study, we first evaluated the use of CBA for cytokine determinations (IL-2, -4, -5, -10, TNF-α, and IFN-γ) in AH of normal controls, comparing levels of IL-4, -10, TNF-α, and IFN-γ with ELISA. Second, cytokine levels in AH from controls versus those with AU and panuveitis were assayed, investigating the effects of steroids and the presence of systemic disease. 

Methods

Patients

Control AH was collected from 10 patients who were undergoing routine phacoemulsification, none of whom were receiving ocular medication or had previous ocular problems. Samples of AH were also taken from 36 patients with active uveitis who had at least 2+ cells in the AC. Twenty-two patients with idiopathic uveitis (mean age, 54.4 years) were diagnosed as either having panuveitis (n = 4) or anterior uveitis (AU, n = 18) after detailed questioning, and appropriate investigations were all negative. Of the 22, 10 were on topical steroids and 10 were untreated. The remaining 14 patients with uveitis had an associated systemic disease (mean age, 39.2 years), comprising 8 AU (5 ankylosing spondylitis, 2 sarcoidosis, and 1 Reiter’s disease) and 6 panuveitis (4 Behçets disease, 1 sarcoidosis, and 1 Vogt-Koyanagi-Harada syndrome). Two of the patients with Behçets were on systemic steroids, one of whom also received cyclosporin A. Of the other 12, 11 were on topical steroids and one was receiving no treatment. All the studies involving human subjects were conducted according to the tenets of the Declaration of Helsinki. The protocol had ethics committee approval, and informed consent was given by all patients. 

Sampling Procedure

AH samples (50–150 μL) from patients with active uveitis or cataracts were collected by anterior chamber paracentesis, as previously described. ² Samples were immediately aliquoted into EDTA-coated microtubes, centrifuged at 400g for 5 minutes, and the cell-free supernatant was stored frozen (−70°C) until assayed. 

CBA

The CBA was used according to the manufacturers’ protocol (Becton Dickinson, Oxford, UK). ¹⁶ Briefly, 10-μL samples (standards or test) were added to 50 μL of a cocktail of capture beads and detector antibodies, and the mixture was incubated for 18 hours at room temperature in the dark. ¹⁷ Excess unbound detector antibody was washed off before data acquisition. 

Two-color flow cytometric analysis was performed using a flow cytometer (FACScan; Becton Dickinson). A total of 1800 events were acquired following the protocol supplied. Analysis was performed using CBA dedicated analysis software (CellQuest; Becton Dickinson). The minimum detection levels for each cytokine were: IL-2 and IL-4, 2.6 ρg/mL; IL-5, 2.4 ρg/mL; IL-10 and TNF-α, 2.8 ρg/mL; and IFN-γ, 7.1 ρg/mL. In all samples where the calculated cytokine concentration was below the given sensitivity, it was treated as undetectable. For spiking experiments, standards were solubilized in assay diluent before subsequent dilutions in pooled AH (resulting in a final concentration of 98.75% AH in the highest cytokine standard) or assay buffer. 

ELISA

Where there was sufficient AH remaining (100 μL), the levels of IL-4, -10, TNF-α, and IFN-γ were determined by using commercially available kits (OptEIA; BD Biosciences Pharmingen). The minimum detection level for all cytokines was 11 ρg/mL. In all assays, mean OD_450nm readings obtained from duplicate wells of diluent negative controls were subtracted from all readings (samples and controls) before making standard curves from which to calculate sample concentrations. In all samples where the calculated cytokine concentration was below the given sensitivity, it was treated as undetectable. 

Statistics

Comparisons of mean cytokine levels as detected by CBA or ELISA were performed by using a paired t-test, with P < 0.01 being significant. The Mann–Whitney test was used to compare results from different patient groups. Correlation of cytokines within individual AH samples in different groups used Spearmann’s rank coefficient test, with R ² > 0.75 being significant. 

Results

Evaluating Cytokine Detection in AH by CBA and ELISA

Because of the high albumin concentration in AH, which may affect binding of antibodies, we initially determined the ability of CBA to detect cytokines in control AH by spiking cytokine standards in pooled samples of control AH (n = 7) and by comparing the levels with those in the kit diluent. The cytokine standards were added to 50-μL AH aliquots and serially diluted to final concentrations of 625, 156, 80, 40, 20, and 0 ρg/mL, to be within the range of cytokines previously detected in AH by ELISA. No significant differences in cytokine detection were seen between those diluted in AH or the buffer supplied, as shown for TNF-α (Fig. 1A) . 

Uveitis AH specimens were used to compare CBA with ELISA results, and the CBA technique was more sensitive than ELISA in detecting IL-4, TNF-α (P < 0.01; Fig. 1B 1C ), and IL-10 but only at lower concentrations (Fig. 1D) . Overall, there were no statistically significant differences between CBA and ELISA in the detection of IL-10 (Fig. 1D)and IFN-γ (Fig. 1E) . To detect six cytokines by CBA, 10 μL of AH was required, whereas 50 μL AH was needed for ELISA. 

Effect of Disease Location on AH Cytokine Profile

CBA was used for simultaneously measuring IL-2, -4, -5, -10, TNF-α, and IFN-γ in each AH (Fig. 2) . There was a significant increase in levels of IFN-γ in both AU and panuveitis groups compared with controls (P < 0.01; Fig. 2A ). Interestingly, IL-5 was not detected in the panuveitis group (n = 10) compared with control (P < 0.01) and AU groups (P < 0.05; Fig. 2B ). For IL-4, -10, and TNF-α, no differences in levels were found between the groups, and IL-2 was not detectable in any of the samples. 

Significant correlations between TNF-α and IL-4 (R ² = 0.77; P < 0.01), and TNF-α with IFN-γ (R ² = 0.77; P < 0.01) were detected in individual AH samples in the control group but not in any from the uveitis groups (data not shown). 

Effect of Steroid Therapy on AH Cytokine Profile

Cytokine levels in the controls were compared with the panuveitis group receiving steroids (n = 8, 6 receiving topical steroids) and the AU group receiving steroids (n = 16, all received topical steroids) or not (n = 10). There were only 2 of 10 patients in the panuveitis group who did not receive steroids, and these data were not included. 

The panuveitis group on steroids (n = 8) had no detectable levels of IL-5 compared with AU not taking steroids or control groups (P < 0.01; Fig. 3A ). There were no differences in IL-5 levels comparing the panuveitis group receiving steroids with the AU group receiving steroids because of the relatively low and variable levels (Fig. 3A) . Compared with the control group, there were no differences in IL-5 or IL-10 in either AU group, but a significant increase in IL-10 was detected in the panuveitis group that received steroids (P < 0.01; Fig. 3B ). 

There were significant increases in IFN-γ levels in both AU groups and the panuveitis group receiving steroids compared with controls (P < 0.01; Fig. 3C ). There was a marked positive correlation between IFN-γ and IL-10 levels within the uveitis subgroups, whether with or without steroids, as shown for the AU group with steroids (R ² = 0.76; Fig. 3D ). 

Effect of Systemic Disease on Cytokine Profile

The correlation of cytokines between the uveitis groups with and without systemic disease and controls was investigated. A significant positive correlation in individual AH samples between IFN-γ and IL-10 was found in all uveitis subgroups (0.86 < R ² < 0.97; P < 0.01; data not shown), regardless of whether there was an association with a systemic disease, which was not found in the control group. 

IL-5 was detectable in the AU groups and did not differ significantly compared with the control group (Fig. 4A) . In contrast, no IL-5 was detectable in either panuveitis with (n = 6) or without systemic disease (n = 4) groups (Fig. 4A) . The IL-5 levels in the panuveitis group without systemic disease were significantly lower than in the corresponding AU group (P < 0.05; Fig. 4A ). 

The IFN-γ levels in the AU group, with or without an associated systemic disease, were increased compared with the control group (P < 0.01; Fig. 4B ). Similarly in the panuveitis groups, with (P < 0.05) or without (P < 0.01) an associated systemic disease, IFN-γ levels were significantly higher than the control group. There were no significant differences in IFN-γ levels between the uveitis subgroups, with or without a systemic disease (Fig. 4B) . 

A significant correlation was found between TNF-α and IL-4 (R ² = 0.78; P < 0.01) in the AU group with systemic disease. No cytokine correlations were seen in the AU group as a whole, in the AU without systemic disease, or in the panuveitis groups, whether associated with a systemic disease or not (data not shown). 

Discussion

In this study, we first evaluated the CBA technique for detecting cytokines in pooled human AH and did not observe any difference in levels whether cytokine standards were diluted in assay buffer or in AH, confirming that components of AH did not interfere with cytokine detection. CBA was comparable with ELISA in the detection of IFN-γ but proved to be more sensitive than ELISA in the detection of low levels of the cytokines IL-4, -10, and TNF-α. Previously, it was found necessary to dilute AH samples for cytokine detection by ELISA, because of the presence of inhibitory proteins. ⁷ This is likely to be more of a problem with two-dimensional detection systems, for example, ELISA, whereas, in CBA, the antibodies are in the fluid phase and might be more accessible. The other major advantage of CBA was as a rapid method of measuring multiple cytokines from a single, small volume, usually 10 μL of AH. At this volume, it is not possible to perform ELISA without a minimum one in five dilution, thus decreasing the sensitivity of ELISA even further. 

Uveitis describes a variety of clinical entities with differing clinical phenotypes and it is possible that different mechanisms are involved in each clinical subtype. However, T lymphocytes have been shown to dominate the cellular immune response in the flow cytometric analysis of lymphocytes from ocular fluids during noninfectious uveitis ^{18 19} The magnitude and the pattern of the cytokine response in different types of uveitis, however, remains unclear. The second aim of this study, therefore, was to investigate the cytokine profiles in subgroups of uveitis, using CBA technology and focusing on a panel of T-cell-related cytokines, some of which have previously been detected in AH in uveitis. 

Of the six cytokines, IL-2 was not detectable in any of the AH, in agreement with others. ^{7 9} IL-4 was detected by CBA at the lower end of the concentration range but not at all by ELISA, and levels did not differ between uveitis and control groups. The differences observed in IL-5 in the uveitis groups might suggest that Th1 polarity is more marked, with greater uveal tract involvement in panuveitis than in AAU. Similarly, peripheral blood lymphocyte supernatants of patients with presumed sarcoid uveitis demonstrated high IL-2:IL-5 and IFN-γ:IL-5 ratios, suggesting a Th1 polarization of the systemic immune response. ²⁰ IL-10 was detected in all groups, but levels did not differ in the uveitis groups with or without systemic disease, compared with the controls. This agrees with a previous study in which IL-10 was only increased in 3 of 22 uveitis patients. ¹⁰ In contrast, in 5 patients with FHC thus far investigated, we detected significant increases in IL-10 (P < 0.05, data not shown), in agreement with previous reports. ^{2 6 7} Both IL-10 and TGF-β are immunoregulatory cytokines, because of their inhibitory effects on T-cell function and on TNF-α. ²¹ IL-10 negatively regulates IL-12-induced inflammation ^{22 23} and is produced by many cell types, including T-cell subsets, B cells, eosinophils, mast cells, and cells of the monocyte/macrophage lineage, the latter being an important source in humans. ^{21 24 25 26 27} The increased level of IL-10 found in the panuveitis group that was taking steroids is in agreement with the work of Barrat et al., ²⁸ who showed that production of IL-10 by T cells was upregulated in vitro by glucocorticoids. 

By using the CBA method, significant increases were found in IFN-γ levels in all uveitis groups in comparison with controls. This was not dependent on steroid treatment or on associated systemic disease. Interestingly, no significant increases in IFN-γ have been detected in AH by ELISA ⁷ or in AC cells by using intracellular staining, ⁶ further supporting CBA as a more sensitive method for cytokine detection. 

No increase in TNF-α was found in uveitis AH in comparison with controls, in agreement with our previous study that used ELISA, ⁷ despite CBA being more sensitive for detection. Although the overall levels of TNF-α in AH did not increase, a positive correlation was observed between TNF-α and IL-4 in the AU group with systemic disease, including those with Behςet’s disease, for whom it has been shown that anti-TNF-α therapy is clinically effective. ²⁹ Others have detected increases in TNF-α in uveitis AH, which was found to be associated with HLA-B27, ^{30 31} supporting the use of anti-TNF-α therapies in ankylosing spondylitis. ³² Unfortunately, there were too few HLA-B27–positive AH specimens in our study to provide statistically meaningful data. 

The location of the disease did not affect the levels of IFN-γ, because both AU and panuveitis groups had significantly increased levels in comparison with controls, regardless of whether there was an associated systemic disease. In contrast, there was a striking absence of IL-5 in the panuveitis groups but not in the AU groups, perhaps because of a more polarized Th1 response in panuveitis. This was the same whether there was an associated systemic disease or not and is unlikely to be influenced by steroid therapy, because, in the two patients with panuveitis who did not receive steroids, we were similarly unable to detect IL-5 (unpublished results). 

Based on the cytokine profiles in the uveitis subgroups, steroid therapy was mainly associated with an upregulation of IL-10 in the panuveitis group in comparison with controls. However, because IL-10 was not increased in the AU group that received steroids, it would suggest that the upregulation in IL-10 was related to the subtype of disease as opposed to an overall effect of steroid therapy, as previously reported. ⁶ It should be noted that the majority of the patients in this study were receiving topical steroids, and it is not clear if this route of steroid administration upregulates intraocular levels of IL-10. Certainly, the other cytokines investigated were unaffected by steroid therapy. 

One advantage of CBA is the ability to compare relative cytokine levels within single samples. Positive correlations were seen between TNF-α and IL-4, TNFα and IL-10, TNFα and IFN-γ, as well as IL-10 and IL-5 within the control AH, suggesting a balance of Th1 and Th2, pro- and anti-inflammatory cytokines. Positive correlations were seen in all uveitis groups between IFN-γ and IL-10, perhaps suggesting an immunoregulatory response is occurring with the upregulation of IL-10 in an attempt to counter the pro-inflammatory effects of IFN-γ. 

The ability of CBA to provide information within individual AH samples on both quantities of several cytokines and correlations between cytokines suggests that this technique is an important tool for further understanding the localized cellular responses occurring during uveitis. This has enabled us to more accurately define the interplay of cytokines between AU and panuveitis groups, and to suggest a more skewed Th1 cytokine profile in panuveitis compared with AU. 

 Supported by Fight for Sight, the Arthritis Research Campaign and the Faculty of Health Sciences of the University of Adelaide.

 Submitted for publication June 22, 2005; revised September 15, 2005; accepted November 21, 2005.

 Disclosure: K.G.-J. Ooi, None; G. Galatowicz, None; H.M.A. Towler, None; S.L. Lightman, None; V. L. Calder, None.

 The publication costs of this article were defrayed in part by page charge payment. This article must therefore be marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.

 

Figure 1.

 

Mean fluorescence intensity (MFI) of TNF-α by CBA, spiking TNF-α into either assay diluent (solid line) or AH (hatched line) (A), showing no significant difference in levels detected between the two diluents. Cytokine levels (ρg/mL) in AH of patients with uveitis as determined by CBA (solid line) vs. ELISA (dotted line). CBA technique was more sensitive in detecting IL-4 (n = 39; P < 0.01) (B) and TNF-α (n = 19; P < 0.01) (C) at the lower end of the concentration range. There were no statistically significant differences between CBA and ELISA in the detection of IL-10 (n = 21) (D) and IFN-γ (n = 36) (E).

Figure 2.

 

Cytokine levels (ρg/mL) in AH of patients with acute AU (AAU, n = 26; solid bars), panuveitis (PAN, n = 10; hatched bars) and controls (control, n = 10; open bars). Increased levels of IFN-γ were detected in the AAU and PAN groups compared with controls (P < 0.01) (A). There was a decrease in production of IL-5 in the PAN group compared with controls (P < 0.01), and the AU group (P < 0.05) (B). For IL-4 (C), IL-10 (D), and TNF-α (E), no differences in levels between uveitis groups and controls were found.

Figure 3.

 

Cytokine levels (ρg/mL) in AH comparing controls with AU either on or off steroids (AU steroid+, AU steroid−, respectively), and panuveitis taking steroids (PAN steroid+), for detecting IL-5 (A), IL-10 (B) and IFN-γ (C). * P < 0.01 compared with PAN steroid+. Correlation of IFN-γ with IL-10 levels within individual AH samples, with controls (open circles; R ² = 0.32) and AU steroid+ (closed circles; R ² = 0.76) (D).

Figure 4.

 

Cytokine levels (ρg/mL) in AH of controls compared with AU and panuveitis patients with or without systemic disease (AU systemic+, AU systemic−, PAN systemic+, PAN systemic−). * P < 0.05 compared with control. ** P < 0.01 compared with control. ^† P < 0.01 compared with AU systemic−.