Patients and methods

Vitreous and serum specimens from 23 patients with uveitis undergoing vitrectomy for treatment of vitreous opacity or retinal detachment were investigated. Eight of the patients presented with idiopathic uveitis while the remaining 15 patients presented with uveitis associated with systemic diseases. Seventeen of the patients were on corticosteroids, either prednisolone or dexamethasone, one patient was on cyclophosphamide, and five were on no medication before surgery. The disease status of uveitis assessed by a clinician at the time of surgery (that is, active or quiet disease), other diseases associated with the condition, and retinal features, are listed in Table 1. All individuals in the study had uveitis before development of retinal detachment and none of the patients had proliferative retinopathy at the time of surgery. Control vitreous samples were obtained from 36 patients without uveitis or systemic disease undergoing surgery for RRD and from 10 patients undergoing surgery for macular holes. Cadaveric vitreous samples obtained within 7–18 hours post mortem from 10 individuals with no known ocular or systemic inflammatory disease were used as normal controls. Vitreous and serum samples were stored at −70°C until use.

MEASUREMENT OF TNFα AND ICAM-1

Commercially available ELISA kits (R&D Systems) were used for the detection of soluble ICAM-1 and TNFα. Vitreous and serum dilutions (1 in 10 and 1 in 20 respectively) in kit diluent were used for the assay, which was performed according to the manufacturer’s instructions. The coloured product was measured photometrically with test and reference readings of 450 nm and 620 nm respectively on a Dynatech MR5000 plate reader. Unknown concentrations of sICAM-1 and TNFα were determined from dose-response curves of standard preparations.

DETECTION OF BIOLOGICALLY ACTIVE TNFα

Biologically active TNFα was measured by a cytotoxic assay using the TNFα sensitive murine fibrosarcoma cell line WEHI-164–32.23 Briefly, 2 × 10⁴ cells in 100 μl of RPMI + 5% FCS medium were incubated overnight in a 96 well plate. This was followed by addition of 100 μl of standard or sample dilutions in the same medium containing 200 ng actinomycin D and further incubation for 24 hours at 37°C. MTS (3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulphophenyl)-2H-tetrazolium solution) with 0.05% phenazine methosulphate (Promega, UK) was then added to produce a colour reaction as a direct measure of the number of viable cells. Optical density was measured with test and reference readings of 490 nm and 620 nm respectively on a Dynatech MR5000 plate reader and the concentration of TNFα in individual samples was calculated by extrapolation from dose-response curves of standard preparations (National Institute for Biological Standards and Control, NIBSC, England).

STATISTICAL ANALYSIS OF THE RESULTS

The significance of difference between corresponding groups of observations was evaluated by the Mann–Whitney U test and by χ² analysis, after constructing 2 × 2 contingency tables. Acceptable significance was recorded when p values were < 0.05. Spearman’s rank correlation coefficient on the logged data was used to determine the relation between vitreous levels of sICAM-1 and TNFα.

Results

Figure 1 shows that vitreous levels of sICAM-1 as a whole were significantly higher in the 23 patients with uveitis (median, 17.19; range, 4.49–150.78 ng/ml) when compared with normal cadaveric vitreous (median, 5.18; range, 4.17–7.35 ng/ml; Mann–Whitney U test p< 0.0001) or vitreous from eyes with macular holes (median, 4.83; range, 4.41–7.68 ng/ml, p< 0.0001). Vitreous specimens from eyes with uveitis complicated by RRD (13 rhegmatogenous RD and two traction RD (TRD)) (median, 40.87; range, 5.94–150.78 ng/ml) contained significantly higher levels of sICAM-1 than those from non-uveitic eyes with RRD alone (median, 8.69; range, 4.89–23.53 ng/ml; Mann–Whitney U test, p< 0.0001) or uveitic eyes with other complications (median, 13.06; range, 4.49–60.76 ng/ml, p< 0.04) (Fig 2A). Comparison of the vitreous levels of this molecule between active and quiet disease showed that the highest sICAM-1 concentrations were present in eyes with active uveitis (Mann–Whitney U test, p< 0.02) (Fig 2B). All seven vitreous samples from eyes with active uveitis complicated by RRD contained >20 ng/ml sICAM-1 ( >four times the normal levels), in contrast with two of six vitreous from eyes with quiet uveitis complicated by RRD or traction retinal detachment (Fisher’s test, p= 0.02) (Fig 2B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

Levels of sICAM-1 in vitreous from eyes with uveitis complicated or uncomplicated by RD are significantly higher than in control cadaveric vitreous (Mann–Whitney U test, *p<0.0001) and vitreous from eyes with macular holes (Mann–Whitney U test, *p<0.0001). The bars represent median values.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

(A) Levels of sICAM-1 in vitreous from patients with uveitis complicated by retinal detachment (•); patients with other uveitis complications but without retinal detachment (▴ ) and patients with rhegmatogenous retinal detachment alone (• ). (Mann–Whitney U test: uveitis with RD v uveitis without RD *p < 0.02; uveitis with RD v RRD alone *p < 0.0001). The bars represent median values. (B) Levels of sICAM-1 in vitreous from eyes with active uveitis compared with vitreous from eyes with quiet disease. (Mann–Whitney U test *p < 0.02). Uveitis patients with RD (•) and without RD (▴). The dotted line represents 20 ng/ml sICAM-1.

For analysis of the data, patients were divided into subgroups consisting of two possible contributory factors towards raised levels of sICAM-1 as follows: (1) active uveitis with and without RD; (2) quiet uveitis with and without RD; (3) idiopathic uveitis with and without RD; and (4) uveitis with systemic disease with and without RD. As shown in Table 2, although the median values and ranges appear to be higher in the groups with RD, the only statistically significant difference was observed in the group of patients with idiopathic uveitis complicated by RD when compared with patients with idiopathic uveitis without RD (Fisher’s test, p = 0.02).

Figure 3A shows that serum levels of sICAM-1 in patients with uveitis were significantly lower at the time of vitrectomy (Mann–Whitney U test, p < 0.02) than those in serum of healthy subjects matching sex and age of the patients. There was no relation between vitreous and serum levels of sICAM-1 (r = −0.39, p < 0.25) in the whole group of patients (Fig 3B).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

(A) Serum levels of sICAM-1 in patients with uveitis (at the time of vitrectomy) compared with those of sex and age matched controls. (Mann–Whitney U test, *p< 0.02). The bars represent median values. (B) Relation between sICAM-1 levels in serum and vitreous from uveitis patients (Mann–Whitney U test, p<0.25, r = −0.39). The solid line represents the linear regression curve of best fit.

Biologically active TNFα was not detected in any of the vitreous samples investigated. However, immunologically detectable TNFα was found in concentrations ranging between 2 and 33pg/ml in all vitreous from uveitic eyes (Table 3). Interestingly, there was an association between the vitreous levels of sICAM-1 and those of immunologically detectable TNFα (Spearman’s correlation coefficient; r= 0.60, p < 0.006) (Fig 4). Vitreous levels of sICAM-1 were not related to disease status, medication at the time of vitrectomy, or association of uveitis with systemic disease.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Relation between levels of sICAM-1 and immunologically detectable TNFα in vitreous from 19 patients with uveitis. The solid line represents the linear regression curve of best fit.

Discussion

The present observations showed that vitreous from eyes with uveitis complicated or uncomplicated by RD contained higher levels of sICAM-1 than disease control vitreous (RRD alone or macular holes) and normal cadaveric vitreous (p<0.0001). Higher vitreous levels of this molecule were found in eyes with active disease and retinal detachment than in those with quiet disease and other complications (p<0.02), as well as in eyes with idiopathic uveitis complicated by RD when compared with patients with idiopathic uveitis without RD. Interestingly, all seven vitreous specimens from eyes with active uveitis complicated by RD contained >20 ng/ml sICAM-1 (four times the normal levels), in contrast with two of six samples of vitreous from eyes with quiet disease complicated by RRD or TRD (Fisher’s test, p= 0.02). Immunologically detectable but not biologically active TNFα was also observed in all the vitreous specimens investigated and immunoreactive TNFα levels correlated with those of sICAM-1.

Expression of ICAM-1 by various cells of the retina, choroid, and leucocytes has been shown in vivo and in vitro, and activation of these cells with pro-inflammatory cytokines such as TNFα induces increased expression and release of this molecule. Since upregulation of ICAM-1 is often accompanied by release of its extracellular portion into the surrounding tissues and fluids, it is likely that the sICAM-1 found in vitreous from eyes with uveitis is derived from various retinal cells, including retinal vascular endothelium, RPE cells, and leucocytes among others. The findings that a higher proportion of uveitic eyes complicated by RD contained raised levels of sICAM-1 suggests that mechanical breakdown of the blood-retinal barrier may amplify the inflammatory process in the uveitic eye, in a similar way to that seen in proliferative vitreoretinopathy, a complication of rhegmatogenous retinal detachment.18 The high vitreous levels of sICAM-1 observed in uveitic eyes with RD can not be ascribed entirely to the retinal detachment itself, as non-uveitic eyes with RRD alone contained similar vitreous sICAM-1 levels than those of uveitic eyes without RD. Since upregulation of ICAM-1 on activated endothelium constitutes an important mechanism by which leucocytes are recruited into the inflammatory site, it is possible that high expression of sICAM-1 on the retinal endothelium may promote the recruitment and accumulation of leucocytes and local inflammatory cells and hence may be responsible for the chronicity of this process.

Unlike a report by others that serum levels of sICAM-1 are increased in patients with active uveitis,19 we did not observe enhanced levels of this molecule in the serum of patients with uveitis undergoing vitrectomy. This might be due to decreased inflammation caused by systemic steroids routinely administered before retinal surgery.

TNFα is a pro-inflammatory cytokine with important autocrine and paracrine functions. It stimulates mononuclear phagocytes to produce other cytokines, induces and enhances the expression of adhesion molecules in various cell types, and increases vascular permeability.16 The biologically active form of TNFα consists of a trimer of three 17 kD subunits which is inactivated by enzymatic fragmentation into monomers and dimers. Since all forms of active and inactive TNFα can be detected by immunoassay, the present findings that immunoreactive but not biologically active TNFα was detected in vitreous suggest that TNFα may have been produced locally during the initiation of the uveitic process but later inactivated by natural inhibitors of inflammation.24 The presence of TNFα in vitreous from eyes with uveitis is of special significance as this cytokine may be largely responsible for the initiation of the inflammatory cascade that leads to the accumulation of inflammatory cells into the eye and hence to the complications often seen in this condition such as vitreous opacity, vitreous detachment, and retinal detachment.25 Implication of TNFα in the pathogenesis of uveitis is supported by findings that intravitreal administration of this cytokine in rabbits increases aqueous protein concentration and induces influx of polymorphonuclear granulocytes into the anterior chamber,26 and that intravitreal but not systemic administration of this cytokine causes uveitis in rats.8Further evidence for the implication of TNFα in uveitic disease is given by observations that aqueous cells from human uveitic eyes express mRNA coding for TNFα and that ocular tissues from eyes with this condition stain strongly for TNFα.3 8 In addition, lipopolysaccharide induced TNFα production in whole blood is significantly enhanced in patients with active uveitis compared with normal subjects.12

Increased vitreous levels of sICAM-1 and the relation between the presence of this molecule and that of immunoreactive TNFα in uveitic eyes clearly indicate the operation of cytokine mediated vascular reactions at the blood-retinal barrier during the development of this condition. Persistence of high vitreous levels of sICAM-1 in uveitic eyes complicated by RRD, despite administration of immunosuppressive drugs, may indicate both a low rate of clearance of this molecule from the eye cavity, and an exacerbation of the existing inflammatory process by the retinal detachment itself. Better understanding of the mechanisms that promote the release of soluble adhesion molecules within the retinal microenvironment may aid in the design of new therapeutic approaches for the control of the inflammatory process caused by uveitic disease and its complications.