Inflammatory cytokines as mediators of retinal endothelial barrier dysfunction in non‐infectious uveitis Authors: Lisia Barros Ferreira, Keryn A Williams, Giles Best, Cameron D Haydinger, Justine R Smith Published: 2023 ABSTRACT AbstractCharacterised by intraocular inflammation, non‐infectious uveitis includes a large group of autoimmune and autoinflammatory diseases that either involve the eye alone or have both ocular and systemic manifestations. When non‐infectious uveitis involves the posterior segment of the eye, specifically the retina, there is substantial risk of vision loss, often linked to breakdown of the inner blood‐retinal barrier. This barrier is formed by non‐fenestrated retinal vascular endothelial cells, reinforced by supporting cells that include pericytes, Müller cells and astrocytes. Across the published literature, a group of inflammatory cytokines stand out as prominent mediators of intraocular inflammation, with effects on the retinal endothelium that may contribute to breakdown of the inner blood‐retinal barrier, namely tumour necrosis factor (TNF)‐α, interleukin (IL)‐1β, IL‐6, IL‐8, IL‐17 and chemokine C‐C motif ligand (CCL)2. This article reviews the function of each cytokine and discusses the evidence for their involvement in retinal endothelial barrier dysfunction in non‐infectious uveitis, including basic laboratory investigations, studies of ocular fluids collected from patients with non‐infectious uveitis, and results of clinical treatment trials. The review also outlines gaps in knowledge in this area. Understanding the disease processes at a molecular level can suggest treatment alternatives that are directed against appropriate biological targets to protect the posterior segment of eye and preserve vision in non‐infectious uveitis. Introduction Non‐infectious uveitis is a cluster of diseases defined by intraocular inflammation. 1 These conditions include autoimmune and autoinflammatory disorders that either involve the eye alone or have both ocular and systemic manifestations. 2 In populations with a high prevalence of the human leukocyte antigen (HLA)‐B27, non‐infectious uveitis is frequently an HLA‐B27‐associated disease and may be accompanied by spondyloarthropathy. Other forms of non‐infectious uveitis that are localised to the eye include pars planitis, birdshot chorioretinopathy and multifocal choroiditis with panuveitis. Systemic disorders associated with non‐infectious uveitis include sarcoidosis, multiple sclerosis, tubulointerstitial nephritis, juvenile idiopathic arthritis, Vogt‐Koyanagi‐Harada syndrome and Behçet disease. If uveitis does not have a specific clinical phenotype, it is termed idiopathic or undifferentiated. While not the focus of this review, uveitis also may result from infections, such as toxoplasmosis, onchocerciasis, cysticercosis, tuberculosis, syphilis, leprosy and numerous viral infections, or it may represent an inflammation masquerade, most commonly vitreoretinal lymphoma. 1 , 2 In 2005, the Standardization of Uveitis Nomenclature (SUN) Working Group established a uveitis classification that has set the standards for uveitis description in the literature ever since. 3 Considering the primary site of the inflammation, uveitis is described as anterior if involving the anterior chamber of the eye, affecting the iris (iritis), anterior ciliary body (anterior cyclitis) or both (iridocyclitis); as intermediate if the vitreous is inflamed (pars planitis, posterior cyclitis and hyalitis); as posterior if the retina (retinitis), choroid (choroiditis) or both (retinochoroiditis) are affected; and as panuveitis, if the inflammation is based across all the ocular compartments. These anatomical groupings have clinical relevance: when uveitis involves the posterior segment of the eye, specifically the retina, there is a substantial risk of vision loss. 4 Subsequently, in 2008, representatives of the International Uveitis Study Group suggested a simple clinical classification that was based on etiological criteria: infectious, non‐infectious and masquerade. 5 The SUN Working Group followed up on this initiative in 2021, when they published detailed classification criteria for 25 of the most common uveitides, developed using a machine‐learning approach. 6 The retina is supplied by blood vessels that are lined by a monolayer of non‐fenestrated endothelial cells that are tightly interconnected by junctional complexes. 7 Reinforced by supporting cells that include pericytes, Müller cells and astrocytes, the retinal vascular endothelium forms the inner blood‐retinal barrier, which limits the movement of circulating cells, micro‐organisms, fluid and solutes into the retina, and thereby protects vision. 7 , 8 In non‐infectious uveitis, an array of inflammatory molecules causes the inner blood‐retinal barrier to malfunction, resulting in extravasation of fluid and proteins, and transendothelial leukocyte migration. 9 In addition to generalised leaking of the retinal vasculature, the accumulation of fluid in the central retina – termed macular oedema – is a primary cause of vision loss in patients. 10 , 11 A plethora of cytokines and their receptors have been examined for their putative roles in the pathophysiology of uveitis in humans, in experimental animals and in in vitro models. In this review, our focus is on those inflammatory cytokines that may mediate retinal endothelial barrier dysfunction, leading to the development of retinal vascular leakage and macular oedema in non‐infectious uveitis. Inflammatory cytokines in non‐infectious uveitis Inflammatory cytokines enact a central role in the pathogenesis of non‐infectious uveitis, with activities that include chemotaxis and activation of T‐helper (Th) cells, neutrophils and monocytes, as well as effects on resident ocular cells. 9 , 12 Some of these proteins are also vasoactive and likely play key roles in breakdown of the inner blood‐retinal barrier. 9 , 12 Much knowledge of disease mechanisms derives from the study of animal models of non‐infectious uveitis. 9 The assessment of the intraocular fluids (aqueous and vitreous humours) from patients is also of considerable interest, since the inflammatory cytokine load is a proxy for the disease process affecting the ocular tissues. 13 , 14 , 15 , 16 , 17 , 18 , 19 , 20 , 21 , 22 , 23 , 24 , 25 , 26 , 27 , 28 , 29 Intraocular cytokine profiles appear to vary with disease activity, 26 treatment status, 17 extraocular manifestations 25 and ocular complications. 24 , 27 , 30 While it is difficult to draw conclusions about relative importance, specific inflammatory cytokines stand out as mediators of retinal endothelial cell dysfunction during non‐infectious uveitis: tumour necrosis factor (TNF)‐α, interleukin (IL)‐1β, IL‐6, IL‐8, IL‐17 and chemokine C‐C motif ligand (CCL)2. As key protagonists of inflammation, TNF‐α, IL‐1β and IL‐6 all participate in the pathogenesis of experimental non‐infectious uveitis. Across different animal models, TNF‐α is increased in the inflammatory cell infiltrate, and TNF‐α blockade substantially alleviates the severity of the inflammation. 31 Differentiation of Th17 cells relies on IL‐1β expression and is crucial for initiation of autoimmune uveitis in mouse models, 32 while uveitis is greatly mitigated in IL‐6‐deficient mice and with IL‐6 antagonism. 33 , 34 Interleukin‐8, IL‐17 and CCL2 also play major roles in non‐infectious uveitis; intravitreal injection of IL‐8 induces uveitis in rabbits, whereas IL‐8 blockade reduces disease severity. 35 , 36 , 37 Participating indirectly through the IL‐23/IL‐17 signalling pathway, IL‐17 is necessary for the expansion and pathogenicity of Th17 cells and promotes the development of experimental autoimmune uveitis. 38 Both the local vasculature and infiltrating leukocytes express CCL2 in the inflamed eyes of mice, and intraocular injection of CCL2 elicits monocyte infiltration. 39 , 40 In addition to their involvement in experimental disease, TNF‐α, IL‐1β, IL‐6, IL‐8, IL‐17 and CCL2 are upregulated in intraocular fluids of patients with non‐infectious uveitis (Tables 1 and 2). Table 1Findings of studies that assessed changes in the concentration of selected inflammatory cytokines in ocular fluids of patients with non‐infectious uveitisCytokineChange in concentration, study a AqueousVitreousIncreaseDecreaseNo change or not detectableIncreaseDecreaseNo change or not detectableTNF‐α 13, 15, 16, 19, 23, 146, 228, 229, 230 – 15, 17, 26, 28, 231, 232, 233 21, 24 – 24, 234 IL‐1β 13, 16, 23, 26, 29, 228 – 14, 15, 17, 28, 29, 146, 235 –– 21, 24 IL‐6 14, 15, 16, 17, 22, 23, 26, 27, 146, 231, 233, 236 – 17, 28, 235 21, 24, 234, 237 ––IL‐8 15, 17, 20, 22, 26, 28, 146, 230, 231, 238, 239, 240 – 15, 23 18, 21, 24 – 234 IL‐17A 16, 19, 23, 146 – 15, 19, 28 21 – 24, 234 CCL2 13, 17, 26, 146, 239, 241 – 13, 15, 23, 28, 230 21, 24 ––a30 studies included patients with active uveitis (n = 16 studies), inactive uveitis (n = 5 studies), active or inactive uveitis (n = 6 studies), and patients whose uveitis activity was not specified (n = 3 studies). The studies included patients whose uveitis was treated (n = 2 studies), was not treated (n = 9 studies), was treated or not treated (n = 12 studies), and patients whose treatment status was not specified (n = 7 studies). Findings of studies that assessed changes in the concentration of selected inflammatory cytokines in ocular fluids of patients with non‐infectious uveitis 30 studies included patients with active uveitis (n = 16 studies), inactive uveitis (n = 5 studies), active or inactive uveitis (n = 6 studies), and patients whose uveitis activity was not specified (n = 3 studies). The studies included patients whose uveitis was treated (n = 2 studies), was not treated (n = 9 studies), was treated or not treated (n = 12 studies), and patients whose treatment status was not specified (n = 7 studies). Table 2Findings of studies that assessed changes in the concentration of selected inflammatory cytokines in ocular fluids of patients with different types of non‐infectious uveitisDisease (Ocular fluid)Cytokine change in concentration, study a IncreaseDecreaseNo change or not detectableBehçet disease (Aqueous)TNF‐α, 13 , 15 , 19 IL‐1β, 13 IL‐6, 15 IL‐8, 15 , 17 , 20 , 28 , 238 CCL2 17 –TNF‐α, 17 , 28 IL‐1β, 15 , 17 , 28 IL‐6, 17 , 28 IL‐17A, 15 , 28 CCL2 13 , 15 , 28 Birdshot chorioretinopathy (Aqueous)TNF‐α, 23 IL‐1β, 23 IL‐6 23 –IL‐8, 23 CCL2 23 HLA‐B27‐positive uveitis (Aqueous)TNF‐α, 13 , 19 IL‐1β, 13 , 29 IL‐8 20 –CCL2 13 Juvenile idiopathic arthritis‐associated uveitis (Aqueous)IL‐6, 22 IL‐8 22 –CCL2 22 Sarcoid uveitis (Aqueous)TNF‐α, 13 IL‐1β, 13 IL‐8 20 –CCL2 13 Sarcoid uveitis (Vitreous)TNF‐α, 21 IL‐6, 21 IL‐8, 21 IL‐17A, 21 CCL2 21 –IL‐1β 21 Vogt‐Koyanagi‐Harada syndrome (Aqueous)TNF‐α, 13 , 19 IL‐1β, 13 IL‐6, 15 , 236 IL‐8, 20 , 28 CCL2 13 –TNF‐α, 15 , 28 IL‐1β, 15 , 28 IL‐6, 28 IL‐8, 15 IL‐17A, 15 , 28 CCL2 15 , 28 Undifferentiated uveitis b (Aqueous)TNF‐α, 146 IL‐6, 17 , 146 IL‐8, 17 , 28 , 146 CCL2 17 , 146 –TNF‐α, 17 , 28 IL‐1β, 17 , 28 , 29 , 146 IL‐6, 28 IL‐17A, 28 CCL2 28 a13 studies included patients with active uveitis (n = 9 studies), inactive uveitis (n = 2 studies), active or inactive uveitis (n = 1 study), and patients whose uveitis activity was not specified (n = 1 study). The studies included patients whose uveitis was treated (n = 0 studies), was not treated (n = 5 studies), was treated or not treated (n = 3 studies), and patients whose treatment status was not specified (n = 5 studies).bUndifferentiated uveitis includes those types described as ‘idiopathic’ or those without an etiological diagnosis. Findings of studies that assessed changes in the concentration of selected inflammatory cytokines in ocular fluids of patients with different types of non‐infectious uveitis 13 studies included patients with active uveitis (n = 9 studies), inactive uveitis (n = 2 studies), active or inactive uveitis (n = 1 study), and patients whose uveitis activity was not specified (n = 1 study). The studies included patients whose uveitis was treated (n = 0 studies), was not treated (n = 5 studies), was treated or not treated (n = 3 studies), and patients whose treatment status was not specified (n = 5 studies). Undifferentiated uveitis includes those types described as ‘idiopathic’ or those without an etiological diagnosis. Tumour necrosis factor‐α Tumour necrosis factor‐α is a transmembrane protein that can be cleaved by matrix metalloproteases such as TNF‐α‐converting enzyme (TACE) – also known as a disintegrin metalloproteinase (ADAM) 17 – into soluble TNF homotrimers. 41 It is secreted by activated macrophages, microglia, astrocytes, T cells and mast cells. 42 Tumour necrosis factor‐α binds to two TNF‐α receptors (TNFRs): TNFR1 (p55), which is widely expressed throughout the body, and TNFR2 (p75), which is more restricted to the different cells of the immune system, and is also expressed by vascular, muscle and brain cells. 41 Importantly in the context of this review, both these receptors are expressed by human retinal endothelial cells. 43 Following TNF‐α binding to TNFR1, the TNF receptor 1‐associated protein with death domain (TRADD), receptor‐interacting protein kinase 1, TNF receptor‐associated factor (TRAF) 2 and cellular inhibitor of apoptosis proteins (cIAPs) 1 and 2 are recruited to the receptor, resulting in the activation of the nuclear factor kappa B (NF‐κB), p38 mitogen‐activated protein kinase (MAPK) and c‐Jun N‐terminal kinase (JNK) pathways. 41 The activation of TNFR2 results in the clustering of TRAF2, cIAP1/cIAP2 and HOIP, a component of the linear ubiquitin assembly complex, to form a signalling complex that activates and induces the non‐canonical NF‐κB pathway and protein kinase B (PKB) phosphorylation. Actions of TNF‐α include cytotoxicity, host defence against intracellular microorganisms, activation of endothelial cells and T cells, stimulation of natural killer (NK) cells, and modulation of leukocyte function. 44 Levels of TNF‐α are higher in the eyes of patients with non‐infectious uveitis than in the eyes of controls. 13 , 15 , 16 , 19 , 21 , 23 , 24 Specifically, TNF‐α is increased in the aqueous and/or vitreous humours of patients suffering from the forms of uveitis that mainly affect the posterior segment of the eye, such as Behçet's disease, 13 , 15 , 19 sarcoidosis, 13 , 21 Vogt‐Koyanagi‐Harada syndrome 13 , 19 and birdshot chorioretinopathy. 23 Elevated intraocular levels of TNF‐α correlate with inflammatory activity 19 and with the development of glaucoma secondary to uveitis. 30 The effect of TNF‐α on the human retinal endothelium has been evaluated in a number of studies. 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 Tumour necrosis factor‐α has been demonstrated to decrease cell proliferation, 50 induce apoptosis 52 and prompt the production of IL‐6 50 and matrix metalloproteinase 3 51 in these cells. Decreased transcellular electrical resistance 46 , 47 , 48 , 53 and disruption of junctional molecules have also been reported. 46 , 47 , 48 , 49 However, other authors have noted no impact of TNF‐α on endothelial barrier function, adherens junction component or expression of junctional molecules. 45 , 48 In studies using human brain endothelial cells, downregulation of junctional molecules was reported, but contradictory findings persisted regarding the effects of TNF‐α on endothelial barrier function. 54 , 55 , 56 In bovine retinal endothelial cells, TNF‐α induced endothelial barrier breakdown, downregulation of junctional complexes and apoptosis, 57 , 58 , 59 , 60 whereas in rodents and rabbits, TNF‐α was associated with increased transendothelial vesicular transport, leukocyte adhesion and disruption of tight junctions. 61 , 62 , 63 , 64 , 65 There is convincing evidence for the clinical efficacy of TNF‐α blockers, such as infliximab 66 , 67 (chimeric monoclonal anti‐TNF‐α antibody) and adalimumab 68 , 69 , 70 , 71 and golimumab 72 (human monoclonal anti‐TNF‐α antibodies) in non‐infectious uveitis. Two large randomised‐controlled trials showed a delayed time to treatment failure in patients with active and inactive uveitis, through the use of adalimumab, 70 , 73 with a 50% and 43% reduction in the risk of treatment failure in the adalimumab group compared with placebo at 18 months. In a randomised, placebo‐controlled study, 60% of children with juvenile idiopathic arthritis and uveitis failed treatment with methotrexate by 18 months versus 27% of those who took the combination of methotrexate and adalimumab. 71 The development of anti‐drug antibodies to adalimumab or infliximab therapy may impact drug efficacy, and TNF‐α blockers have been associated with paradoxical intraocular inflammation. 74 , 75 , 76 Some studies have focused on the use of anti‐TNF‐α drugs for the treatment of persistent uveitic macular oedema, with an overall response rate ranging from 22% to 85% after 6–12 months of follow‐up. 66 , 68 , 69 , 77 , 78 The endothelia in different vascular beds respond differently to cytokines such as TNF‐α and consequently, the efficacy of different anti‐TNF biologics may also vary in disparate disease contexts. This has been apparent in investigations into the utility of anti‐TNF‐α biologics in disorders of the central nervous system, 79 where, for example, the TNF‐α blocking agent lenercept was shown to trigger relapses and worsen neurological deficits in multiple sclerosis. 80 For systemically administered TNF‐α biologics designed to modulate uveitis, the possibility of off‐target neurotoxicity must be considered. The intraocular use of anti‐TNF‐α monoclonal antibodies has been explored for the treatment of recalcitrant non‐infectious uveitis. 81 , 82 , 83 , 84 , 85 , 86 , 87 Temporary improvement of inflammation has been reported with the use of intravitreal infliximab; however, monthly treatment over a 9‐month period was largely unsuccessful, and 20% of patients had worsening of intraocular inflammation. 82 , 86 , 87 Despite improving uveitis in patients with Behçet's disease in some studies, macular oedema resolved in fewer than 20% of patients. 81 , 83 Furthermore, intravitreal infliximab administration has been reported as being ineffective for the treatment of diabetic macular oedema and choroidal neovascularization, and concerns have been raised over retinotoxicity and the development of intraocular inflammation. 88 , 89 , 90 Intravitreal adalimumab controlled uveitis in 75% of patients, and resolution of macular oedema was achieved in about 60% of cases, although a small case series showed that intravitreal adalimumab was ineffective in non‐infectious uveitis. 84 , 85 Data from existing studies of local TNF‐α blockade are insufficient to determine efficacy for the management of uveitis and uveitic macular oedema, and reports of toxicity demand careful consideration of clinical risk–benefit ratios. In summary, TNF‐α plays a major role in the development of non‐infectious uveitis. 91 It is upregulated intraocularly in several forms of posterior uveitis, 13 , 15 , 19 , 21 , 23 and blockade has proven successful in many clinical studies, 70 , 71 , 73 including cases of macular oedema and retinal vascular leakage. 66 , 68 , 69 , 77 , 78 However, the use of intravitreal TNF‐α antagonists appears to only partially control uveitis. 81 , 82 , 83 , 84 , 85 , 86 , 87 The role of TNF‐α in retinal endothelial dysfunction seems to involve increased permeability, secondary disruption of junctional complexes, increased leukocyte adhesion and trafficking and upregulation of vesicular transport and possibly cytotoxicity. 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , 57 , 58 , 59 , 60 , 61 , 62 , 63 , 64 , 65 However, much knowledge of the role of TNF‐α in blood‐retinal barrier impairment has stemmed from publications focused on diabetic retinopathy and other vitreoretinal diseases. 45 , 46 , 47 , 48 , 49 , 50 , 51 , 52 , 53 A debate lingers concerning the effects of TNF‐α on retinal endothelial permeability and damage of junctional complexes. 45 , 46 , 47 , 48 , 49 , 53 Furthermore, the effects of TNF‐α on retinal endothelial cell viability have not been extensively researched. Interleukin‐1β Interleukin‐1β is produced by dendritic cells, epithelial cells, endothelial cells, monocytes, NK cells, B cells and T cells. 92 , 93 Interleukin‐1β acts through the IL‐1 receptor, type 1 (IL1R1), which is blocked by the naturally occurring IL‐1 receptor antagonist (IL1RA). 94 Interleukin‐1 receptor, type 1, is widely expressed in immune cells, such as macrophages, neutrophils, eosinophils, T cells and endothelial cells. 95 Transcriptomic profiling has shown that human retinal endothelial cells also express IL1R1. 43 The production of IL‐1β is elicited by cytokines or activation of toll‐like receptors via pathogen‐associated molecular patterns. An inactive precursor of IL‐1β is cleaved by caspase‐1 following activation by danger‐associated molecular changes. 96 After binding to its receptor, the co‐receptor, IL‐1 receptor accessory protein (IL1RAcP), is engaged to the IL‐1β/IL1R1 complex, forming a ternary structure. This complex then attaches to the Toll/IL1R1 intracellular domains and recruits the myeloid differentiation primary response gene 88 (MYD88), toll‐interacting protein and IL1R1‐associated kinase 4. 97 Down‐stream events activate the NF‐κB or MAPK/JNK/extracellular signal‐regulated kinase (ERK) pathways; the former is involved in immune responses and regulation of cell survival and proliferation, while the latter regulates tumour suppression, apoptosis and cell growth. 96 , 97 Interleukin‐1β has pleiotropic actions in autoinflammatory diseases, including the activation of various cells involved in immune responses and induction of endothelial adhesion molecules. 94 , 97 Many studies have reported higher concentrations of IL‐1β in the aqueous humour of uveitic patients than in that of controls. 13 , 14 , 16 , 23 , 26 , 29 Two groups of investigators described raised levels of IL‐1β in the aqueous humour of patients with HLA‐B27‐associated uveitis. 13 , 29 Elevated concentrations of IL‐1β were also found in the aqueous humour of patients with uveitis that affects the posterior segment, such as Behçet's disease, 13 birdshot chorioretinopathy, 23 sarcoid uveitis 13 and Vogt‐Koyanagi‐Harada syndrome. 13 While IL‐1β upregulation in the aqueous humour during non‐infectious uveitis has been reported by multiple studies, none have reported that it is reduced, suggesting an important role of IL‐1β in uveitis. In the human retinal endothelium, IL‐1β has been found to trigger oxidative stress, mitochondrial dysfunction and production of TNF‐α, IL‐6, CCL2, intercellular adhesion molecule (ICAM)‐1, vascular adhesion molecule (VCAM)‐1, vascular endothelial growth factor (VEGF)‐A and matrix metalloproteinases‐2 and ‐9. 50 , 98 , 99 , 100 Some reports have shown that IL‐1β can induce the breakdown of the human retinal endothelial barrier 46 , 48 , 53 , 101 and downregulate the junctional molecules 46 ; however, one study reported induction of retinal endothelial hyperpermeability by IL‐1β under hyperglycaemia but only when human retinal endothelial cells were in co‐culture with pericytes, via induction of pericyte apoptosis and impairment of tight junctions. 102 Conversely, progressive strengthening was observed in human brain endothelial cells following an acute reduction in transcellular electrical resistance after IL‐1β administration. 56 No changes in the apoptosis rate and cell viability have been reported after IL‐1β treatment of human retinal endothelial cells, 102 but disparate results have been reported for other species, including bovine and rodent retinal endothelial cells. 59 , 61 , 103 , 104 Interleukin‐1β has also been shown to promote opening of endothelial tight junctions, leukocyte migration, vesicular transport and endothelial permeability when injected into the eyes of rodents and rabbits. 65 , 105 , 106 , 107 Downregulation of junctional molecules and upregulation of genes related to apoptosis and leukocyte migration have been observed in human umbilical vein endothelial cells exposed to IL‐1β. 108 Interleukin‐1β blockers, such as anakinra (recombinant human IL1RA) and canakinumab (a human anti‐IL‐1β monoclonal antibody), were shown to be effective in treating non‐infectious uveitis. 109 , 110 , 111 , 112 , 113 In seven of nine patients with Behçet's disease, in whom uveitis had failed to respond to TNF‐α blockade, anakinra improved control of the disease. 111 In another study of 19 patients with Behçet's disease, a decrease in the number of uveitis flares was observed after commencement of anakinra or canakinumab, either as monotherapy or combined with other immunomodulatory agents. 109 The efficacy of IL‐1β antagonists as a treatment for patients with Behçet's disease correlates with disease duration, potentially related to the severity of the inflammation. 110 However, another study of an initially promising anti‐IL‐1β drug, gevokizumab (humanised anti‐IL‐1β monoclonal antibody), showed no decrease in the risk of uveitis flares in patients with Behçet's disease, although vision was maintained, and macular oedema was reduced in the first 6 months of treatment in these patients. 114 Although intravitreal blockade of IL‐1β has only been attempted in animal models, this approach did reduce neutrophilic infiltration of the retina, without preventing mononuclear cell trafficking or protein leakage. 115 As a chief cytokine associated with autoinflammatory conditions, IL‐1β is implicated in the mechanisms driving uveitis, 46 , 48 , 53 , 65 , 105 , 106 , 107 as elevated intraocular concentrations have been identified in different posterior uveitis conditions. 13 , 23 Clinically, the antagonism of IL‐1β has not been specifically assessed in the treatment of uveitic macular oedema and retinal vascular leakage, although some studies have shown IL‐1β blockade can control uveitis and resolve macular oedema. 109 , 110 , 111 , 112 , 113 The disruption of the inner blood‐retinal barrier triggered by IL‐1β appears to be secondary to tight and adherens junction damage, enhanced transendothelial vesicular transport, promotion of leukocyte migration, mitochondrial dysfunction and oxidative damage. 46 , 48 , 50 , 53 , 56 , 59 , 61 , 65 , 98 , 99 , 100 , 103 , 104 , 105 , 106 , 107 , 108 As in the case of TNF‐α, the majority of the studies concerning the effects of IL‐1β on human retinal endothelial cells have been in the context of diabetic retinopathy and other retinal diseases. 46 , 48 , 50 , 53 , 98 , 99 , 100 , 102 The influence of IL‐1β on expression of junction molecules has not been extensively assessed, and controversies remain regarding the effects of IL‐1β on barrier permeability and cell viability in human retinal endothelial cells. 46 , 48 , 53 , 102 Interleukin‐6 Interleukin‐6 is secreted by multiple cell types including monocytes, T cells, B cells, endothelial cells, and fibroblasts. IL‐6 is a pleiotropic cytokine that signals through the IL‐6 receptor (IL‐6R), which is also known as IL‐6Rα, CD126 and glycoprotein (gp)80. 116 , 117 The IL‐6R exists in two forms: a membrane‐bound IL‐6R (mIL‐6R), which is limited to a few cell types, and a soluble IL‐6R (sIL‐6R), which is generated via proteolytical cleavage of the mIL‐6R by proteases, or by alternative splicing, forming a protein that lacks the transmembrane domain of the full length receptor. 116 , 117 When IL‐6 binds to mIL‐6R, a signalling pathway termed classical signalling is elicited, while IL‐6/sIL‐6R ligation triggers the trans‐signalling pathway. 116 , 117 Both pathways require the association of the IL‐6/IL‐6R complex with the widely expressed signal‐transducing gp130, activating tyrosine kinase Janus kinases (JAK) 1 and 2, and tyrosine‐protein kinase 2. Janus kinase 1 phosphorylates tyrosine residues in the intracellular portion of gp130, activating the MAPK, phosphoinositol‐3 kinase (PI3K)/AKT and the signal transducer and activator of transcription (STAT)3 pathways. Downstream effects of this signalling include cell proliferation, differentiation and survival, immune regulation stress responses and angiogenesis. 118 Increased levels of IL‐6 have been found in the aqueous 14 , 15 , 16 , 17 , 22 , 23 , 26 and vitreous humours 21 , 24 of patients with non‐infectious uveitis, compared with controls. The levels of IL‐6 in the aqueous humour of uveitic patients were over 130 times higher than controls; in one study, IL‐6 was identified as the cytokine that was most significantly elevated. 16 Raised intraocular IL‐6 levels have been assessed in the context of several uveitides, including those affecting the posterior segment, such as Behçet's disease, Vogt‐Koyanagi‐Harada syndrome, sarcoidosis and birdshot chorioretinopathy. 15 , 17 , 21 , 22 , 23 Intraocular concentration of IL‐6 correlated with the number of cells, including neutrophils, in the anterior chamber of patients with idiopathic anterior uveitis, Behçet's disease and Vogt‐Koyanagi‐Harada syndrome. 15 , 17 Furthermore, IL‐6 levels in the aqueous humour have been shown to correlate with inflammatory activity and the grade of secondary cataracts. 14 , 27 Overall, IL‐6 plays a major role in non‐infectious uveitis, with several studies reporting increased levels in the aqueous and vitreous humours, and no reports of IL‐6 downregulation, suggesting this cytokine is involved in both anterior and posterior segment inflammation. Interleukin‐6 enhances cell proliferation and viability in human retinal endothelial cells. 50 , 119 A couple of studies have reported that IL‐6 decreases the transcellular electrical resistance and increases the permeability of human retinal endothelial cells, 48 , 120 with diminished expression of some components of tight junctions. 120 Others have reported no change in human retinal endothelial barrier function or tight junctional molecule expression, 121 with one study noting no effect of the cytokine. 46 Co‐administration of sIL‐6R and IL‐6 to human retinal endothelial cells in vitro has been shown to lead to phosphorylation of STAT3, which prompts ICAM‐1 expression, mitochondrial dysfunction and oxidative stress, and barrier function impairment. 122 Endothelial barrier breakdown and disruption of junctional complexes have also been observed after simultaneous administration of IL‐6 and sIL‐6R in human vein umbilical endothelial cells, but not with administration of IL‐6 alone. 123 However, one study reported that IL‐6 alone was sufficient to reduce transcellular electrical resistance of human vein umbilical endothelial cells in a dose‐dependent manner. 124 Downregulation of junctional molecules and enhanced permeability has also been reported in human brain microvascular endothelial cells. 54 , 55 In mice, uveitis was attenuated in IL‐6‐deficient animals or with intravitreal anti‐IL‐6 treatment. 33 , 34 Interleukin‐6 is also crucial for Th17 differentiation, an important cell lineage in the pathogenesis of uveitis. 33 Retinal endothelial cells were thought to lack the transmembrane IL‐6R 121 , 122 , 125 while expressing the glycoprotein gp130, 122 , 125 which would allow for IL‐6 trans‐signalling. Although IL‐6R transcript expression was reported in human retinal endothelial cells, 126 contemporary reports suggested that these cells did not express IL‐6R protein. 121 , 122 , 125 Low levels of sIL‐6R were identified in the medium of cultured human retinal endothelial cells but were insufficient to result in endothelial barrier breakdown with exogenous IL‐6 administration. 121 In addition to increased IL‐6 intraocular levels, raised sIL‐6R levels have also been observed in uveitis. 127 Therefore, it has been suggested that during inflammatory states, sIL‐6R is generated by proteolysis or alternative splicing from mIL‐6R expressed by infiltrating leukocytes, enabling IL‐6 to activate the retinal endothelium through the IL‐6/sIL‐6R/gp130 complex. 122 , 128 In recent work, we used flow cytometry to demonstrate the presence of mIL‐6 on multiple primary human retinal endothelial cell isolates and demonstrated this protein to be functional in experiments tracking electrical impedance of IL‐6‐treated retinal endothelial cell monolayers. 129 Clinical experience has shown that broad IL‐6R blockade improves non‐infectious uveitis, particularly in refractory cases of macular oedema and retinal vasculitis, both associated with the impairment of the retinal endothelial barrier. 130 , 131 , 132 Tocilizumab is a humanised monoclonal antibody against sIL‐6R and mIL‐6R. Intraocular inflammation can be controlled using tocilizumab in patients with a range of non‐infectious uveitides and macular oedema refractory to conventional immunomodulatory therapy, anti‐TNF‐α drugs and interferons. 131 , 132 , 133 , 134 , 135 In contrast, a study of 21 patients with juvenile idiopathic arthritis‐associated refractory uveitis – a condition mostly featuring anterior uveitis – demonstrated that tocilizumab was effective in only one‐third of patients after 12 weeks. 136 In the same study, coexistent macular oedema resolved in three of four patients. Treatment for 16 weeks with sarilumab, a human anti‐IL‐6 receptor monoclonal antibody, achieved a reduction in vitreous haze and/or a decrease in corticosteroid dose in 64% of 38 treated patients compared with 35% of 20 patients in the placebo group, by investigator assessment. 137 In mice, 75% of those administered with an intravitreal IL‐6 blockade had reduced signs of inflammation and vascular leakage. 34 As a key mediator of inflammation, the role of IL‐6 in uveitis and other retinal diseases has been the subject of growing interest. 138 , 139 Elevated intraocular levels of IL‐6 have been reported in cases of posterior uveitis, and IL‐6 blockade has been particularly effective in treating retinal vascular leakage and macular oedema in uveitis. 131 , 132 , 133 , 134 , 135 However, there has been an ongoing debate on the effect of IL‐6 on retinal endothelial permeability, with several authors describing decreased barrier impedance in the presence of the cytokine, 48 , 120 , 129 and others reporting no effect. 46 , 121 Some investigators have suggested that endothelial permeability is mediated only by trans‐signalling, since co‐administration of sIL‐6R and IL‐6 was necessary to elicit hyperpermeability. 122 , 123 There also remains some controversy regarding expression of the IL‐6R by the retinal endothelium. 121 , 122 , 125 , 126 , 129 Interleukin‐17 Interleukin‐17, or IL‐17A, is one of the six members of the IL‐17 family and the hallmark of Th17 cells. 140 , 141 , 142 Besides being produced by Th17 cells, IL‐17 is also secreted by γδ T cells, cytotoxic CD8+ T cells, NK T cells, neutrophils and microglia. Interleukin‐17 receptors are heterodimeric structures composed of the IL‐17 receptor A (IL‐17RA) plus a second chain, IL17‐RC, which confers ligand and signalling specificity. 143 Interleukin‐17RA is widely expressed, while IL17‐RC is expressed by non‐immune cells, which in the retina include Müller cells, retinal endothelial cells and photoreceptors. 144 , 145 Interleukin‐17 binds to the IL‐17RA/IL17‐RC receptor complex, initiating an interaction between the cytoplasmic SEFIR domains of the receptor and the signalling adaptor protein Act1, which recruits TRAF6 and activates signalling via the NF‐κB, MAPK, ERK, JAK/PI3K, JAK/STAT, p38, JNK and CCAAT‐enhancer‐binding protein (C/EBPs) pathways. 140 , 143 Increased activity of these pathways leads to the transcription of inflammatory cytokines, chemokines and antimicrobial peptides. In addition, a noncanonical pathway, dependent on I kappa B (IκB) kinase and phosphorylation of Act1, can be induced via recruitment of TRAF2 and TRAF5, which stabilise mRNAs involved in the transcription of cytokines and chemokines. 143 Interleukin‐17 is increased in the aqueous humour 16 , 19 , 23 , 146 and vitreous humour 21 of patients with non‐infectious uveitis, in comparison with healthy controls. Raised intraocular levels of IL‐17 have been observed in Behçet's disease, 19 birdshot chorioretinopathy, 23 sarcoidosis 21 and Vogt‐Koyanagi‐Harada syndrome. 19 In the study of patients with birdshot chorioretinopathy, IL‐17 levels in the aqueous humour and serum were compared, and intraocular levels were higher than systemic levels. 23 Studies have also demonstrated a correlation between aqueous levels of IL‐17 and the clinical severity of uveitis. 16 , 19 In multiple studies using human retinal endothelial cells, increased IL‐17 expression has been associated with monolayer hyperpermeability, downregulation of junctional components, induction of IL‐6, IL‐8, ICAM‐1 and VEGF‐A, secretion of neutrophil elastase, promotion of cell migration, angiogenesis, increased apoptosis and reduced cell proliferation. 144 , 147 , 148 , 149 , 150 However, contradictory studies suggest that IL‐17 enhances cell proliferation and has no effect on human retinal endothelial barrier function. 46 , 148 Interleukin‐17 also appears to have pro‐angiogenic effects on human choroidal endothelial cells, enhancing cell migration and tube formation by inducing chemokine expression by the retinal pigment epithelium. 151 , 152 In human extra‐ocular endothelial cells, IL‐17 upregulates IL‐6, IL‐8 and other chemokines, fosters neutrophilic recruitment and migration, and induces endothelial cell apoptosis. 142 , 153 , 154 , 155 The barrier function of human retinal pigment epithelial cells and mouse brain endothelial cells is also decreased by IL‐17 through disruption of junctional proteins. 156 , 157 Blockade of IL‐17 was shown to ameliorate uveitis in animal models but was toxic towards photoreceptors. 158 , 159 However, injection of IL‐17 has also been shown to reduce the severity of experimental autoimmune uveitis. 160 Systemic IL‐17 blockade has been used therapeutically as a treatment for spondyloarthropathies. 161 Data from three phase III clinical trials, which enrolled a total of 274 patients with non‐infectious uveitis, suggest that subcutaneous secukinumab – a human monoclonal anti‐IL‐17 antibody – did not reduce uveitic recurrences during withdrawal of immunomodulatory therapy, compared with placebo. 162 However, a similar study of 24 patients showed that intravenous secukinumab controlled intraocular inflammation in over two‐thirds of the patients with non‐infectious uveitis. 163 Subcutaneous administration of secukinumab was also found to control intraocular inflammation in a case series of four patients with HLA‐B27‐associated uveitis. 164 There are studies that have reported onset of Behçet's‐like and Behçet's disease, and uveitis following introduction of anti‐IL‐17 drugs. 165 , 166 , 167 However, pooled data from three trials suggested that there was no increase in the incidence of uveitis in patients with active ankylosing spondylitis who were treated with secukinumab. 168 Interleukin‐17 is an important cytokine in the pathogenesis of uveitis, 169 although only a few studies have evaluated the activity of IL‐17 in human retinal endothelial cells. 46 , 144 , 147 , 148 , 150 Findings suggest that IL‐17 promotes the production of other inflammatory cytokines and may drive cytotoxicity 142 , 144 , 147 , 148 , 149 , 150 , 153 , 154 , 155 ; however, it is unclear whether IL‐17 has any effect on retinal endothelial cell impedance or cell proliferation due to contradictory reports. Moreover, the effects of IL‐17 on retinal endothelial junctional complexes have not been assessed. Interleukin‐8 Interleukin‐8 – or C‐X‐C motif ligand (CXCL)8 – is a member of the CXC chemokine family characterised by an amino acid between two cysteine residues. Interleukin‐8 is produced by macrophages, lymphocytes, neutrophils, γδ T cells, mast cells, endothelial cells and epithelial cells. 170 , 171 , 172 Two receptors from the rhodopsin‐like receptor family, which are coupled to guanosine‐5′‐triphosphate (GTP) binding proteins (G‐protein), are known to mediate IL‐8 activity. These are known as C‐X‐C motif chemokine receptor (CXCR)1 (IL‐8RA) and CXCR2 (IL‐8RB) and are composed of α, β and γ subunits. CXCR1 interacts only with IL‐8 and CXCL‐6, whereas CXCR2 binds more broadly to CXC chemokines. 173 Multiple cell types express CXCR1 and CXCR2, including T cells, neutrophils, fibroblasts, epithelial cells and vascular endothelial cells, 174 while human retinal endothelial cells express CXCR2 but not CXCR1. 175 Following binding of IL‐8 to its receptor, several signalling pathways are activated, including pathways mediated by PI3Kγ, phospholipase C, and tyrosine‐protein kinase HCK, small GTPases Ras/Rac/Rho/cdc42/Rap1, protein kinase C (PKC) and AKT. Interleukin‐8 has been shown to promote cell adhesion, cytoskeletal changes, membrane protrusion, cell migration, exocytosis of lysosomal enzymes and endothelial permeability. 173 Interleukin‐8 can also activate signalling via the ERK1/2, MAPK and JAK/STAT pathways, which regulate cell proliferation, growth, differentiation and survival. 171 , 172 Additionally, IL‐8 has been shown to be a chemotactic agent for granulocytes and can induce phagocytosis, oxidative stress and angiogenesis. Several studies have reported raised levels of IL‐8 in the aqueous 15 , 22 , 26 , 28 , 146 and vitreous 18 , 24 humours of patients with non‐infectious uveitis. Elevated levels of IL‐8 in the aqueous 26 and vitreous humours 24 have been associated with uveitic macular oedema, with levels in the aqueous humour correlating with disease activity. 26 Intraocular levels of IL‐8 also correlate with the total number of cells and neutrophils in the anterior chamber during uveitis. 15 , 17 High levels of IL‐8 have been detected in the eyes of patients with juvenile idiopathic arthritis‐associated uveitis, 22 acute anterior uveitis, 28 idiopathic uveitis, 17 HLA‐B27‐associated uveitis, 20 Behçet's disease, 15 , 17 , 20 , 28 Vogt‐Koyanagi‐Harada syndrome, 20 , 28 sarcoidosis 20 , 21 and idiopathic posterior, intermediate or panuveitis. 146 The biological activity of intravitreal IL‐8 was confirmed in a study that demonstrated neutrophil chemotaxis was significantly reduced by a monoclonal anti‐IL‐8 antibody. 18 Additionally, T‐cell clones isolated from the intraocular fluid of patients with Behçet's uveitis were shown to spontaneously produce higher amounts of IL‐8 than controls. 176 In human retinal endothelial cells, IL‐8 is induced by lipopolysaccharide (LPS), TNF‐α, hyperglycaemia and low shear stress. 177 , 178 , 179 , 180 The addition of IL‐8 to human non‐ocular endothelial cells increases endothelial permeability via phosphorylation and downregulation of junctional molecules, and by transactivation of the VEGF receptor. 181 , 182 , 183 , 184 Intravitreal injection of IL‐8 in rabbits induced transient uveitis, 35 whereas blockade of IL‐8 reduced the severity of endotoxin‐induced uveitis and suppressed leukocyte infiltration but did not reduce protein levels in the aqueous compartments. 37 Furthermore, intravitreal blockade of IL‐8 decreased neutrophilic infiltration by 66% in endotoxin‐induced uveitis in rabbits. 36 To date, IL‐8 blockers have not been implemented for the treatment of inflammatory diseases in humans, but in animal models, administration of an anti‐IL‐8 antibody has been shown to reduce the severity of endotoxin‐induced uveitis. 37 HuMax‐IL8, a humanised monoclonal antibody directed against IL‐8, has been investigated in clinical trials as an adjuvant or single therapy for metastatic solid tumours, 185 lung and liver cancer, 186 and cancers of the pancreas, 187 and prostate 188 due to its potential anti‐neoplastic and anti‐inflammatory effects. A CXCR2 blocker has also been evaluated for the treatment of chronic obstructive pulmonary disease and influenza infection, with a good safety profile, while CXCR1/CXCR2 antagonism has been shown to improve the response of NK and T cells in cancer patients treated with immunotherapies. 189 , 190 , 191 The effects of IL‐8 on the human retinal endothelial barrier have not been investigated, but there have been a few studies that have reported increased levels of IL‐8 in the human retinal endothelium under different conditions. 177 , 178 , 179 , 180 From the limited number of published studies on retinal endothelial and extra‐ocular endothelial cells and in relevant animal models, it appears that IL‐8 promotes neutrophilic infiltration and alters the constituents of junctional complexes 35 , 36 , 37 , 181 , 182 , 183 , 184 ; however, it is unclear whether IL‐8 per se impairs the endothelial barrier or whether this is a result of its effects on leukocyte migration. Interleukin‐8 is an important mediator of intraocular inflammation, and raised levels have been found intraocularly during posterior uveitis 15 , 17 , 20 , 21 , 28 , 146 and macular oedema. 26 Despite improving intraocular inflammation in in vivo animal studies, 37 IL‐8 inhibition has not been investigated as a treatment for non‐infectious uveitis in human clinical trials. C‐C motif chemokine ligand 2 Chemokine C‐C motif ligand 2, also known as monocyte chemoattractant protein (MCP)‐1, is a member of the C‐C chemokine superfamily, which is characterised by two adjacent cysteine residues next to the N‐terminus. 192 The chemokine is synthesised by monocytes, macrophages, fibroblasts, astrocytes, microglia, epithelial cells, endothelial cells and smooth muscle cells. 192 , 193 Several receptors recognise CCL2: C‐C motif chemokine receptor (CCR) 5, CCR10, CCR11 and atypical chemokine receptor 1 (ACKR1). However, CCR2 is the main receptor through which CCL2 exerts its actions. 192 , 193 C‐C motif chemokine receptor‐2 is expressed by monocytes, macrophages, neutrophils, activated T cells and NK cells and endothelial cells, 194 , 195 , 196 and has been identified in the retina. 197 , 198 Upon ligation to CCR2, CCL2 activates signalling pathways, including those mediated by PI3K/AKT, NF‐ĸB, MAPK/p38, ERK and JAK/STAT3, which enhance cell survival, migration, angiogenesis and tumour growth. 192 , 194 , 199 C‐C motif chemokine ligand 2 is a crucial mediator of inflammation, Th1, Th2 and Th17 responses, recruitment of effector cells to the site of inflammation, regulation of cell adhesion via expression of integrins, modulation of macrophage polarisation, activation of monocytes and maturation of dendritic cells. 192 , 193 , 194 , 199 Studies of a diverse range of diseases have implicated CCL2 in their pathogenesis, including cancer, cardiovascular diseases, neuroinflammatory disorders and autoimmune diseases. 199 Various studies have reported raised levels of CCL2 in the aqueous humour 13 , 17 , 26 , 146 and the vitreous humour 21 , 24 of patients with non‐infectious uveitis, as well as uveitic macular oedema. Increased concentrations of this chemokine have been found in the aqueous humour of patients with Behçet's disease and uveitis, 17 idiopathic uveitis, 17 , 146 Vogt‐Koyanagi‐Harada syndrome 13 , 15 and in the vitreous of patients with sarcoid uveitis, 21 all disorders associated with posterior segment intraocular inflammation. Elevated levels of levels of CCL2 were also observed in the aqueous humour in patients with active versus quiescent non‐infectious disease. 26 One study of patients with non‐infectious uveitis and rubella virus‐associated uveitis determined that CCL2 was associated with the development of secondary glaucoma. 200 The expression of CCL2 may be induced by IL‐1β, LPS, VEGF‐A and hyperglycaemia in the human retinal endothelium, 40 , 98 , 201 and angiogenic effects of CCL2 have been observed in non‐ocular human endothelial cells, linked to upregulation of VEGF‐A. 202 In one report, CCL2 had no effect on the impedance of a commercial human retinal endothelial cell line; however when these cells were exposed to conditioned medium from activated human macrophages, a decrease in impedance was observed. 40 In rat retinal endothelial cells, CCL2 induced an upregulation of TNF‐α, IL‐1β and interferon (IFN)‐γ, 203 but no impact on cell viability, migration or the adherens junctions was observed in murine retinal endothelial cells. 61 C‐C motif chemokine ligand 2‐induced endothelial hyperpermeability has been observed in brain endothelial cells from mice, via an alteration of tight junctional complexes. 204 , 205 , 206 Across the course of experimental autoimmune uveitis, CCL2 was detected during peak disease, evident in retinal and choroidal vessels and in infiltrating cells. 39 Intraocular injection of CCL2 in mice elicited monocyte infiltration and microglial activation in the retina, 206 while intravitreal antagonism of CCL2 decreased leucocytic infiltration and protein leakage in endotoxin‐induced uveitis. 36 There have been no attempts to block CCL2 as a treatment for uveitis in patients. Although a CCR2/CCR5 antagonist has been proposed as an approach to reduce diabetic macular oedema and retinal vascular leakage, the results of a clinical trial demonstrated the efficacy of this treatment were inferior to standard treatment with the anti‐VEGF‐A drug, ranibizumab. 207 Furthermore, the CCL2 blockers, carlumab and ABN912, have been studied in humans with rheumatoid arthritis, cancer and idiopathic pulmonary fibrosis but had no apparent benefit 208 , 209 , 210 ; in a randomised clinical trial of an anti‐CCL2 antibody for the treatment of rheumatoid arthritis, no favourable clinical or immunopathological outcomes were observed, despite the drug being well tolerated. 208 There have been only two published studies that have assessed the biological functions of CCL2 in the human retinal endothelium, both focused on diabetic retinopathy. 40 , 201 Data from studies using animal models, non‐human retinal endothelial cells and non‐ocular endothelial cells suggest that CCL2 promotes angiogenesis, leukocyte transendothelial infiltration and impairment of junctional complexes. 36 , 202 , 204 , 205 , 206 Chemokine C‐C motif ligand 2 is an important mediator in uveitis development, 36 , 39 , 204 , 205 , 206 and high levels of CCL2 have been found intraocularly in patients with non‐infectious uveitis affecting the posterior segment, 13 , 15 , 17 , 21 , 146 associated with active disease. 26 Although they may reduce the severity of experimental intraocular inflammation, 36 CCL2 inhibitors have not yet been used to treat patients with non‐infectious uveitis. Other inflammatory cytokines Other cytokines, such as IFN‐γ, IL‐2, IL‐12 and IL‐23, are also implicated in the pathogenesis of non‐infectious uveitis. Interferon‐γ is produced mainly by activated Th1 and NK cells and is essential in immune defence against viral infections. 211 Interferon‐γ has been shown to have activity towards human non‐ocular endothelial cells, 212 , 213 and it may worsen 169 , 214 or attenuate inflammation. 215 However, there is little published evidence that IFN‐γ affects the retinal endothelial barrier during uveitis. CD4‐positive T cells are the main source of IL‐2, a cytokine that fosters the expansion of all T‐cell populations that express the IL‐2 receptor, including Th17 cells. 216 Raised intraocular levels of IL‐2 have been found in patients with uveitis, 15 , 17 , 21 , 23 and IL‐2 blockade with daclizumab improved cases of refractory non‐infectious uveitis, 217 , 218 although clinical development of the drug was discontinued due to its association with meningitis and encephalitis. 219 There is some evidence of IL‐2 activity towards non‐ocular endothelial cells, 220 , 221 but this cytokine has not been studied in relation to the retinal endothelial barrier. Interleukin‐12 is produced by dendritic cells and macrophages and promotes the expansion of Th1 cells. 222 Interleukin‐12 levels are raised in the intraocular humour of uveitic patients, 17 , 24 , 26 , 146 but IL‐12 appears to have opposing effects depending on the phase of experimental autoimmune uveitis development. 223 , 224 Interleukin‐23 is part of the IL‐12 superfamily and is produced mainly by macrophages and dendritic cells, and it plays an essential role in the differentiation of pathogenic Th17 cells. 225 An imbalance in the IL‐23/IL‐17 axis has been associated with autoimmune diseases, such as Vogt‐Koyanagi‐Harada syndrome, Behçet's disease and sarcoidosis. 38 Although IL‐23 was required for development of experimental autoimmune uveoretinitis in mice, its antagonism during the course of the disease did not decrease uveitis severity, 226 and intraocular levels of IL‐23 were not detected in patients with non‐infectious uveitis. 14 , 16 , 23 , 146 Transforming growth factor (TGF)‐β and IL‐10 are also important mediators of uveitis, but these cytokines are mostly associated with the induction of regulatory T cells and suppression of inflammatory responses. 227 Although IL‐12, IL‐23, TGF‐β and IL‐10 appear to be important in uveitis, there is no published evidence to suggest that they are involved in retinal endothelial barrier dysfunction. A summary of the studies in which changes in the concentration of cytokines in ocular fluids of patients with non‐infectious uveitis were assessed is presented in Supplementary tables 1 and 2. Conclusion The management of retinal vascular leakage and macular oedema in non‐infectious uveitis continues to defy clinicians. Understanding the pathophysiology at a molecular level may suggest treatment alternatives that are directed against appropriate biological targets. In this review, we have assembled the evidence for the involvement of specific inflammatory cytokines in the breakdown of the inner blood‐retinal barrier in this group of diseases. Levels of six cytokines – TNF‐α, IL‐1β, IL‐6, IL‐8, IL‐17 and CCL2 – have been shown to be increased in the intraocular fluids of patients with non‐infectious uveitis; laboratory investigations support the activity of all of these cytokines on the functioning of retinal endothelium; and the effectiveness of biologic drugs targeting TNF‐α, IL‐1β, IL‐6 and IL‐17 has been demonstrated in some clinical trials (Table 3). Deficiencies and controversies in the literature remain, so further clarification will be essential to provide a basis for drug selection in humans. However, biologic drugs targeting all six cytokines are already available and therefore may represent immediate therapeutic options for patients with non‐infectious uveitis. Table 3Biologic immunomodulatory drugs targeting selected inflammatory cytokinesDrug a Drug typeUsed in uveitisNotesTumour Necrosis Factor‐alpha (TNF‐α)AdalimumabHuman mAbYesReduced risk of recurrence in phase III randomised controlled clinical trials conducted in adults with varying types of uveitis 70 , 73 and children and adolescents with juvenile idiopathic arthritis‐associated uveitis. 71 CertolizumabPegylated humanised FabYesImproved control in open‐label and retrospective clinical studies 242 , 243 and decreased acute anterior uveitis recurrence in phase IV clinical study of treatment for spondyloarthritis. 244 EtanerceptReceptor‐Fc fusion proteinYesIneffective in controlling inflammation in randomised clinical trials 245 , 246 , 247 and may be associated with paradoxical uveitis. 248 , 249 GolimumabHuman mAbYesTreatment responses in open‐label and retrospective clinical studies 72 , 243 , 250 , 251 and systematic review with metanalysis 252 and decreased acute anterior uveitis occurrence in prospective clinical study of treatment for ankylosing spondylitis. 253 InfliximabChimeric mAbYesEffective treatment for refractory inflammation in phase II clinical trial 254 and retrospective clinical studies. 255 , 256 JNJ‐61178104Human bispecific AbNoInterleukin‐1beta (IL‐1β)AnakinraReceptor antagonistYesTreatment effects in retrospective clinical studies of patients with Behçet disease and uveitis 109 , 111 plus reports of individual patients with autoinflammatory types of uveitis.CanakinumabHuman mAbYesSimilar treatment effects to anakinraGevokizumabHumanised mAbYesControl of Behçet disease‐associated uveitis in phase II clinical trials of subcutaneous or intravenous drug, 257 , 258 but not in phase III randomised controlled clinical trials of subcutaneous drug. 114 RilonaceptReceptor & Accessory protein‐Fc fusion proteinNoInterleukin 6 (IL‐6)ClazakizumabHumanised mAbNoLevilimabHuman anti‐receptor mAbNoOlokizumabHumanised mAbNoSarilumabHuman anti‐receptor mAbYesImprovement in severe inflammation in phase II clinical trial. 137 SatralizumabHumanised anti‐receptor mAbNoSiltuximabChimeric mAbNoSirukumabHuman mAbNoTocilizumabHumanised anti‐receptor mAbYesEffective treatment in retrospective clinical studies and phase I/II clinical trial, 259 , 260 , 261 but with exceptions. 136 Used to resolve macular oedema including pathology resistant to TNF‐α blockade. 132 , 262 Interleukin 8 (IL‐8)BMS‐986253Human mAbNoInterleukin 17 (IL‐17)BimekizumabHumanised mAbYesLower incidence in drug‐ versus placebo‐treated patients in phase III randomised controlled clinical trials for ankylosing spondylitis. 263 BrodalumabHuman anti‐receptor mAbNoCJM112Human mAbNoCNTO6785Human mAbNoHB0017Humanised mAbNoIxekizumabHumanised mAbNoIzokibepSmall molecule inhibitorYesPhase II clinical trials in progress (NCT04706741, NCT05384249).NetakimabHumanised mAbNoSecukinumabHuman mAbYesNo effect on recurrence in phase III randomised controlled clinical trials of subcutaneous drug, 162 but treatment response in phase II clinical trial of intravenous drug in comparison to subcutaneous. 163 SonelokimabCamelid nanobodyNoSSGJ‐608Humanised mAbNoChemokine (C‐C motif) Ligand 2 (CCL2)ABN912Human mAbNoCarlumabHuman mAbNoLeronlimabHumanised mAbNoPF‐04634817Small molecule receptor inhibitorNoAb, antibody; Fab, fragment, antibody binding; Fc, fragment, crystallizable; mAb, monoclonal antibody.aEtanercept targets TNF‐α and lymphotoxin (LT)‐α. JNJ‐61178104 targets TNF‐α and IL‐17A. Bimekizumab and sonelokimab target IL‐17A and IL‐17F. PF‐04634817 targets CCR2 and CCR5. Biologic immunomodulatory drugs targeting selected inflammatory cytokines Ab, antibody; Fab, fragment, antibody binding; Fc, fragment, crystallizable; mAb, monoclonal antibody. Etanercept targets TNF‐α and lymphotoxin (LT)‐α. JNJ‐61178104 targets TNF‐α and IL‐17A. Bimekizumab and sonelokimab target IL‐17A and IL‐17F. PF‐04634817 targets CCR2 and CCR5. Author contributions Lisia Barros Ferreira: Writing – original draft; writing – review and editing. Keryn A Williams: Supervision; writing – original draft; writing – review and editing. Giles Best: Supervision; writing – original draft; writing – review and editing. Cameron D Haydinger: Visualization; writing – review and editing. Justine R Smith: Supervision; writing – original draft; writing – review and editing. Conflict of interest The authors declare no conflict of interest. Supporting information Supplementary table 1 Supplementary table 2 Click here for additional data file.