The pathogenic role of VEGF in the complications of uveitis

Through its promotion of angiogenesis and increased vascular permeability, VEGF plays an important role in the inflammatory process. As such, the expression of VEGF is intimately linked to that of major cytokines in the inflammatory cascade. Nuclear factor-κβ (NF-κβ) is a transcription factor that is crucial in the inflammatory response. In addition to inducing the expression of numerous inflammatory mediators, NF-κβ also induces the expression of VEGF.14 Several inflammatory cytokines, including interleukin (IL)-1β,15 16 tumour necrosis factor (TNF)-α,15 17 IL-6,18 IL-819 and TGF-β₂15 can upregulate VEGF production through various molecular pathways. In patients with uveitis, aqueous humour VEGF levels are increased in eyes with CMO.20 In patients with neovascular AMD, aqueous humour VEGF levels are related to CNV disease activity.21 Similarly, the role of VEGF in promoting RNV has been well established.22 The importance of VEGF in the development of CMO, CNV and RNV, as well as its involvement in the inflammatory cascade, suggests that its inhibition may have therapeutic potential when these complications occur in the setting of uveitis. While VEGF inhibition seems reasonable to treat uveitis complications such as CMO, CNV and RNV (which are secondary to angiogenesis and increased vascular permeability), the role of anti-VEGF therapy to treat inflammation is less clear. Whether inhibition of VEGF has any anti-inflammatory effect in addition to its anti-angiogenic and anti-permeability actions is not known and needs to be explored.

The corticosteroid triamcinolone acetate has been widely used as a local therapy for treatment of uveitis and its complications such as CMO, CNV and RNV. Its mechanism of action remains to be fully elucidated, but it is thought to involve the inhibition of prostaglandin and leukotriene synthesis, as well as downregulation of cell adhesion and major histocompatibility molecules.23 In animal models of hypoxia triamcinolone decreases retinal levels of IL-1β, TNF-α, NF-κβ and VEGF.24 In contrast to anti-VEGF agents that are very specific, corticosteroids are more non-specific with respect to their mechanism of action. Which therapeutic strategy, specific versus non-specific, is superior in the treatment of uveitic complications is not clear. It can be speculated that when only one mediator is specifically inhibited, compensatory mechanisms may exist to adapt to its removal by increasing the production of other mediators. Compensatory elevations of angiogenic and inflammatory factors following VEGF inhibition have been described.25 26 Recently, elevations of several inflammatory cytokines following intravitreal bevacizumab injection have been described in the setting of proliferative diabetic retinopathy.27 The significance of this in the setting of uveitis and the treatment of its complications is not known, but deserves further investigation.

VEGF inhibition for the treatment of uveitic CMO

To date, nine reports regarding the use of intravitreal bevacizumab or ranibizumab for the treatment of uveitic CMO have been published.28–36 Seven of these were case series, three prospective28 30 35 and five retrospective,29 32–34 36 and one was an isolated case report.31 All published reports included small numbers of patients, ranging from one to 27, with the follow-up period ranging from 1 week to 1 year. The majority of cases reported were of non-infectious uveitis, although two reports included patients with infectious uveitis.34 Only one of the articles described the use of ranibizumab for uveitic CMO.30 36 Only one study28 reported the use of Early Treatment of Diabetic Retinopathy Study (ETDRS) charts to record visual acuity. Four articles reported a statistically significant improvement in visual acuity at the end of the follow-up period,28 30 34 35 and four articles reported no change in visual acuity.29 32 33 36 Five of the case series found a statistically significant decrease in central retina thickness as measured by optical coherence tomography (OCT),28–30 33 34 while three series observed no change in central retinal thickness.32 35 36 Both visual acuity and OCT thickness were noted to improve in the isolated case report.31 Several reports included patients requiring multiple injections to maintain the transient response.28–31 34–36 When reporting retrospective data from clinical series of patients with variable follow-up, the reporting of proportions or means can be misleading. It is more appropriate to report rates, or use statistical analyses that incorporate the variable follow-up times.37 However, this was only done in two of three retrospective case series where variable follow-up was present.33 36

Although there is no consensus on the best treatment for uveitic CMO, most studies support aggressive treatment of the underlying inflammatory disease as an important component of the treatment regiment.38 39 In patients with multifocal choroiditis, use of systemic immunosuppressive therapy is associated with a reduction in the development of CMO.40 Adjunctive treatment with anti-VEGF treatments may be appropriate in cases of uveitis where CMO persists despite adequate control of the inflammatory process. In some reports, the level of uveitis activity was not defined.28 31 Standardised criteria for defining disease activity have been established by the Standardization of Uveitis Nomenclature (SUN) working group.41 In many reports where anti-VEGF therapy was used for refractory CMO in the setting of inactive uveitis, the definition of ‘inactive’ was not stated.32 33 35 Hence, it is difficult to know if the inflammation was truly inactive in these cases. The SUN criteria define inactive uveitis as grade 0 cells in the anterior chamber, or less than 1 cell per high power field (HPF).41 CMO is considered a structural complication of uveitis, and is not considered in the definition disease activity.41 The presence of grade 0.5–1+ cells implies activity, and thus does not represent inactive disease. However, in two reports describing the use of anti-VEGF in the setting of inactive uveitis, the definition of inactive allowed up to grade 0.5–1+ cells to be present.29 30 One study included patients with active uveitis, but ‘active’ was not defined.36 Furthermore, in some studies patients were concurrently treated with intravitreal triamcinolone.28 34 The ability to determine if anti-VEGF therapy was beneficial is thus confounded by these issues, which suggest that there may have been active inflammation still present that, if treated with corticosteroids and/or immunosuppression, would have resulted in resolution of the CMO.

Based on the GRADE approach,13 the level of quality of evidence for anti-VEGF therapy in the treatment of uveitic CMO would be rated as very low. Given the multiple confounding issues noted amongst the various studies in the literature, there is no clear-cut evidence supporting the routine use of intravitreal anti-VEGF therapy in uveitic CMO. Treatment of uveitic CMO should be focused on controlling the underlying inflammation, and steroids and/or immunosuppressive agents are the gold standard at this time.12 Although the relative efficacy of intravitreal anti-VEGF compared with intravitreal triamcinolone as adjunctive therapies for uveitic CMO is not known, cases that were unresponsive to intravitreal bevacizumab but responded to intravitreal triamcinolone have been described.28 Based on the available data in the literature, further data from prospective controlled trials are warranted in order to fully determine the role of intravitreal anti-VEGF agents in the treatment of uveitic CMO. At this point, however, it seems reasonable to consider intravitreal anti-VEGF therapy on a case-by-case basis in patients with refractory uveitic CMO in the setting of inactive uveitis and in which intravitreal triamcinolone is contraindicated.

VEGF inhibition for the treatment of uveitic CNV

Multiple reports have been published on the use of intravitreal anti-VEGF agents in the management of uveitic CNV. To date, there have been 11 reports describing intravitreal anti-VEGF in the treatment of CNV secondary to non-infectious uveitis; one study was prospective,42 whereas the remaining were retrospective, non-comparative case series.36 43–51 Sample sizes varied from two to 96 patients, while follow-up periods ranged from 1 to 24 months. In addition to non-infectious uveitis, many series also included patients with CNV secondary to infectious uveitis or other non-uveitic causes.36 42 48–51 Five of these studies were restricted to only patients with non-infectious uveitic CNV.43–47 Most studies involved bevacizumab, with only one report describing the use of ranibizumab.45 Only six studies performed a statistical analysis, and all of these reported an improvement in visual acuity and/or a decrease in OCT thickness.36 42 47 49–51 In addition, two of the 10 reports used ETDRS charts for measurement of visual acuity, but these charts were used only on some of the patients.50 51 Several reports described the need for multiple injections to maintain the therapeutic response.36 42 45–51 In some retrospective studies where statistical analysis was performed, the analyses did not account for the variable follow-up of the patients.47 49–51

Analogous to the treatment of uveitic CMO, the treatment of CNV secondary to uveitis should include control of the underlying inflammation. This approach is supported by evidence suggesting that treatment with systemic prednisone or immunosuppressive therapy may be of benefit in the treatment of CNV secondary to uveitis.40 52–54 Some of the reports describing the use of intravitreal anti-VEGF therapy for uveitic CNV included patients with active uveitis,36 50 51 while others did not comment on or define the activity of uveitis at the time of CNV presentation using standardised criteria.42–49 Furthermore, in some studies patients were treated with concurrent intravitreal triamcinolone.36 46 49 Thus, it is not possible to know how much the activity of the underlying inflammatory disease was contributing to the development of CNV and the need for intravitreal bevacizumab/ranibizumab. Interestingly, one of the studies reported that the angiographic regression pattern correlated with the presence of active uveitis, with complete regression occurring in a higher percentage of patients with inactive uveitis.50

Based on the GRADE approach,13 the level of quality of evidence for anti-VEGF therapy in the treatment of uveitic CNV would be rated as very low. However, given what is known about the important contribution of VEGF to the pathophysiology of CNV development, the use of intravitreal bevacizumab/ranibizumab may be a viable option in cases of uveitic CNV. However, more data from prospective controlled studies will be required before the role of these agents in treatment of uveitic CNV can be fully known. Until that time, a two-pronged approach to treatment of uveitic CNV may be considered; while the acute complication of CNV is being treated with anti-VEGF therapy, attention should also be focused on treatment of the underlying inflammatory disease. It should be noted that treatment of the underlying disease based on published guidelines12 will likely reduce the likelihood of recurrent CNV and reduce the need for repeated anti-VEGF treatment.

VEGF inhibition for the treatment of uveitic RNV

RNV can occur in uveitis secondary to peripheral retinal ischaemia from occlusive vasculitis. Cases of RNV (as well as neovascularisation of the disc, or NVD) were included in reported case series that included predominantly cases of uveitic CNV.50 51 In addition, there has been one report of RNV included with a case of CNV as well as two reports describing three patients with Eales' disease.44 55 56 Only one report included a subanalysis of eyes with RNV/NVD, and this report did not demonstrate any statistically significant gain in visual acuity following intravitreal bevacizumab.51 In three cases the authors reported complete regression of the neovascular complex, which had persisted despite retinal laser photocoagulation, following a single injection of intravitreal bevacizumab.44 55 56 Other series reported complete regression of some cases and partial regression of others.50 51 Based on the available reports, the quality of data supporting a role for intravitreal bevacizumab in the treatment of uveitis RNV would be classified as very low using the GRADE approach,13 and more data are needed. The treatment of uveitic RNV should focus on treatment of the underlying inflammatory disease, as well as laser photocoagulation to areas of ischaemic retina. If RNV persists despite these efforts, intravitreal bevacizumab may be an option. However, the limited data concerning use of intravitreal bevacizumab in the setting of uveitic RNV, as well as reports of tractional retinal detachment following intravitreal bevacizumab in patients with proliferative diabetic retinopathy,57 need to be carefully considered.