The cornea is the clear window at the front of the eye and is the eye's main refractive medium. Its transparency is essential for vision. Corneal neovascularisation is a common clinical problem with serious consequences for vision; it can compromise corneal transparency and plays a major role in corneal graft rejection by breaching corneal immune privilege. In this review, we formulate a consensus on the unmet medical needs in the management of corneal neovascularisation and outline a framework for the clinical research that is needed to identify suitable agents to meet these needs.
- Corneal neovascularisation
- vascular endothelial growth factor
- clinical trials
- public health
- treatment medical
Statistics from Altmetric.com
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.
- Corneal neovascularisation
- vascular endothelial growth factor
- clinical trials
- public health
- treatment medical
Corneal neovascularisation (CN) is a common clinical feature in different corneal diseases including keratitis from infectious (eg, viral, bacterial, fungal) and from immunological origin, keratoplasties and trauma (eg, burns).1–4 CN is part of the natural healing and defence processes after exposure of the cornea to pathogens or trauma, and is not necessarily ‘harmful’ when limited in extent and duration. In the long term and under certain conditions, however, CN can surpass a threshold, invading the corneal centre and/or stroma, endangering vision and, in case of penetrating keratoplasty (PK) or lamellar keratoplasty (LK), threatening corneal graft (CG) survival.5–9 Nonetheless, since the normal cornea is avascular and maintains its avascularity by multiple mechanisms, any CN is per definition ‘pathological’.10–12 This review summarises the authors' opinion on the current unmet medical needs in the management of CN and on the scientific understanding of disease mechanisms involved in CN. We propose a consensual approach for the development of new drugs in this field in terms of inclusion criteria and clinical endpoints that we consider to be clinically relevant.
Overall prevalence of and health burden caused by CN
CN is a common problem that not only reduces visual acuity (VA) but also results in the loss of the immune privilege of the cornea, thereby worsening the prognosis of subsequent keratoplasty.13 In a study in the US in 1996, CN was present in 35 (4.14%) of 845 consecutive eye clinic patients and in 12% of these cases it was associated with a decrease in VA.3 Of 1278 full thickness and lamellar corneal buttons randomly retrieved between 1992 and 1996, 254 (about 20%) showed CN.14 Furthermore, a recent meta-analysis found that graft failure and rejection risk increased with the number of corneal quadrants affected by neovascularisation before keratoplasty.6 The published evidence therefore suggest that although CN can occasionally serve a beneficial role in the clearing of infections, wound healing and arresting stromal melts, its disadvantages are numerous and it often leads to tissue scarring, oedema, lipid deposition and persistent inflammation that may significantly alter VA.1 15 16
Causes of CN
CN occurs as a non-specific response to a variety of clinical insults.2 Many infections can cause CN with the patterns of response being different. Herpes simplex virus (HSV) infection is likely to cause extensive vascularisation and lipid keratopathy whereas, with Acanthamoeba infection, vascularisation tends to appear in the later stages. Studies indicate that the continued presence of HSV–DNA and HSV–immune complexes contribute to angiogenesis and inflammation in herpetic stromal keratitis (HSK) through increased levels of vascular endothelial growth factor (VEGF) and matrix metallopeptidase (MMP)-9.17 Evidence indicates a close link between location (ie, peripheral or central) and extent (ie, superficial or stromal) of infections or lesions, and the location and extent of CN.
Hypoxia related to contact lens wear is a common cause where CN is usually superficial, involving only the corneal periphery.18 19 However, if contact lens wear is not reduced or discontinued, central corneal and deep stromal invasion can occur.
Limbal stem cell deficiency (LSCD) is a severe corneal abnormality that occurs in a variety of ocular diseases both acquired (eg, following chemical burns, contact lens use, drugs, etc) and congenital (eg, aniridia) due to partial or total loss of limbal stem cells (LSC).20 Such LSCD leads to conjunctivalisation of the cornea with vascularisation, appearance of goblet cells, and an irregular and unstable epithelium.21
Chemical (alkaline and acidic) substances can damage and penetrate the cornea and anterior chamber of the eye, with alkali burns being most frequent.3 Neovascularisation may become harmful in the late reparative phase after severe burns. Conjunctivalisation of the corneal surface with massive neovascularisation may develop, leading to severe reductions in corneal transparency and VA through the formation of a pannus in the superficial corneal tissue.3 22
Ocular surface neoplasia, such as papilloma and conjunctival and corneal intraepithelial neoplasia, can cause CN as part of the tumour angiogenic response. Initially the vessels can be confined to the lesion but eventually invade the entire cornea. Other specific aetiologies of CN include corneal allograft rejection and persistent corneal oedema as in bullous keratopathy and chronic hydrops of keratoconus.
Degenerative conditions such as pterygium and diffuse keratoconjunctival proliferation are associated with CN that usually is accompanied with a fibrovascular pannus situated on, rather than in, the corneal stroma. Chronic irritation of the ocular surface such as in perennial atopic keratoconjunctivitis can cause aggressive CN. Other less common causes of CN are exposure to sun, radiation or chemical toxins such as mustard gas.23–25
Common mechanisms of CN
Most vessels that appear in response to inflammatory stimuli tend to have an active stage during which they appear bright red and aggressive in nature. Observations based on a study of more than 200 patients have demonstrated that clinically corneal vascularisation can be classified as active young, active old, mature, partially regressed and regressed (L Faraj and HS Dua, unpublished observations). This often corresponds with the stage of activity or chronicity of the disease. Although clinically invisible, lymphangiogenesis almost always accompanies haemangiogenesis in the cornea.26
The upstream molecular pathway mechanisms leading to CN and the clinical manifestations of CN differ in the various underlying conditions. Nonetheless, core molecular pathways governing the processes of corneal haemangiogenesis and lymphangiogenesis seem to be shared among different underlying diseases leading to the active stage of CN. Table 1 provides a summary of these upstream manifestations (when known) linked to CN for the specific contexts that we define as unmet medical needs with regards to anti-angiogenic therapy in the following section. It seems highly likely from a series of observations that there is a central step involving pro-angiogenic processes that invariably include VEGF signalling, making this event a logical therapeutic target. However, additional molecules are involved in CN (eg, β-fibroblast growth factor (β-FGF) and interleukin-1 (IL-1)), so a multi-pronged approach is desirable.
Unmet medical needs in the management of CN
Treatment with corticosteroids can control inflammation and consequently CN in numerous disease contexts. However, steroids have only limited anti-(lymph)angiogenic effects.42 Furthermore, long-term steroid use is problematic since it can cause corneal or scleral thinning (with risk of perforation), increased intra-ocular pressure (with risk of damage to the optic nerves and resultant defects in visual fields) and posterior subcapsular cataract formation. Therefore, steroid-sparing treatment approaches to CN are needed. In addition to this pharmacological approach, surgical procedures, such as fine needle diathermy or laser ablation, are useful particularly to eliminate established and large corneal vessels.43 The argon laser and the 577-nm yellow dye lasers have also been used to obliterate corneal vascularisation, with the yellow dye laser being the most effective.44–51 However, many treatment centres lack the needed laser equipment. Fine needle diathermy therefore becomes the most useful approach to occlude mature corneal vessels.43 In the following discussion, we are reviewing the available treatment approaches for CN and their limitations.
Novel specific anti-(lymph)angiogenic drugs such as anti-VEGF biologicals (bevacizumab, ranibizumab, pegaptanib) have so far only been licensed for use in oncology or vitreoretinal diseases. Currently, only limited experience using anti-VEGFs on the cornea is available, and only in an off-label setting.52 Antisense oligonucleotide eye drops against insulin receptor substrate 1 (IRS-1) are currently being evaluated in clinical trials.53
Keratitis from infectious origin
In infectious keratitis (viral, bacterial, fungal or parasitic), the primary goal for treatment is to control the pathogen itself, using appropriate (eg, antiviral, antibacterial) medication (table 2).
Herpes simplex and varicella zoster virus keratitis
The two human α-herpes viruses (α-HHVs), varicella zoster virus and, in particular, HSV type 1 are the main herpes viruses that cause herpetic keratitis (HK).64 The presentation of HK ranges from a superficial infection to a chronic inflammatory disease, HSK, which is the leading cause of corneal blindness in developed countries. While HSV infection leads to recurring keratitis, appropriate control of the viral replication with antiviral agents (eg, acyclovir) and, if necessary, subsequent anti-inflammatory treatment with steroids, can in most cases completely avoid CN. When CN develops, however, a close and inverse interrelation between viral load and grade of corneal vascularisation has been shown for patients with a history of HK.54
We consider the following treatment algorithm to be the most appropriate:
In patient with viral keratitis (without CN), the first goal is to control infection with antiviral agents and subsequently the inflammation with steroids, while retaining antiviral therapy.
Patient presenting with untreated HSV with CN:
a. The first goal is to control the viral infection and the inflammation.
b. The next goal is to treat new vessels in the following circumstances, in which they are defined as ‘harmful’:
i. Presence of large stromal feeder vessel(s) or active large stromal vessel(s);
ii. Vessels leaking lipids causing lipid keratopathy;
iii. As preparation for keratoplasty, when required (‘pre-conditioning of the host cornea’).
When keratoplasty is required and performed, antiviral and preventive anti-angiogenic adjuvant treatment may be indicated to prevent CN especially if the keratitis is stromal and involved more than one quadrant. However, this notion requires formal verification. Experimental data in the murine model indicate a significant role for CN in the disease process leading to HSK.28 65 Accordingly, in the murine model of HK, anti-angiogenic therapy can ameliorate the disease course.66 67 Whether this holds true for human HK remains to be studied.
Bacterial, fungal and amoebic keratitis
These types of infectious keratitis are non-recurring and CN is not frequent. Indeed, with appropriate anti-pathogen treatment, bacterial corneal infections often heal without CN.1 Similarly, Acanthamoeba keratitis is rarely, if ever, associated with CN even in relatively severe and long-standing cases.57 Treatment of CN is needed only, despite specific anti-pathogenic and anti-inflammatory treatment:
When the infection is progressing, not controlled or chronic, or
If the CN endangers central vision, or
For pre-keratoplasty conditioning,
as described further below.
A recent meta-analysis of 19 studies including a total of 24 944 grafts undergoing PK demonstrated a statistically robust association between the presence of CN and CG failure with the risk increasing as more quadrants were affected by CN.6 This study confirms previous reports that associated the risk of graft rejection with increasing vascularisation.59–61 68
Furthermore, ample experimental data from the murine model of corneal transplantation demonstrate that inhibition of CN both prior to as well as after PK can significantly promote graft survival both in the low- and high-risk situations.69–71
CN is therefore clearly a risk factor for failure and rejection of CG and should be treated in the following situations:
As ‘pre-conditioning’ treatment prior to PK, if stromal vessels are in the area of the graft bed.
As ‘post-conditioning’ treatment after PK:
a. As prophylaxis, if stromal neovessels were in the area of the CG, or
b. If active stromal neovessels develop, or
c. In active immune rejection of the graft and in high-risk corneal grafts.
Lamellar keratoplasties: deep anterior lamellar keratoplasty/Descemet stripping automated endothelial keratoplasty/Descemet membrane endothelial keratoplasty
Pre- and post-conditioning treatment could be helpful in deep anterior lamellar keratoplasty (DALK), Descemet stripping automated endothelial keratoplasty (DSAEK) and Descemet membrane endothelial keratoplasty (DMEK) to avoid neovascularisation along the lamellar plane or immune rejections. Furthermore, treatment of CN could be helpful in patients undergoing posterior LK to avoid stromal CN and reduction in VA.72 73 The presence of CN that extends into the zone to be trephined can cause postoperative bleeding or neovascularisation into the lamellar interface. That in our experience is difficult to treat. In addition, treatment of CN in diseases potentially amenable to DALK (such as progressive contact lens-induced CN in keratoconus patients) could maintain the option of performing lamellar instead of full-thickness grafting, thus avoiding prolonged visual recovery periods.72–74 Therefore it is likely that anti-angiogenic therapy could be very beneficial in both pre-and postoperative contexts.
LSCD and chemical burns
Treatment of LSCD
Current treatment strategies consist of direct transplantation of either autologous limbus containing LSC from the contralateral normal eye (unilateral LSCD) or allogeneic limbal epithelium in the form of a keratolimbal allograft to regain a stable corneal epithelium.75 76 Auto- and allografts of ex vivo expanded sheets of limbal epithelium are also viable options.76–78 Anti-angiogenic treatment options are considered desirable for patients with severe LSCD. After transplantation and restoration surgery, preventive treatment of CN should be considered as soon as a stable corneal epithelium is in place. In this context as well, preventive anti-angiogenic medication administered prior to LSC transplantation could be beneficial.
Where possible, a patient suffering from a chemical burn could be treated immediately to prevent CN and the formation of the fibrovascular pannus. In the late stage management, after removal of the fibrovascular pannus, all remaining CN should be aggressively treated. Burns may cause LSCD and anti-angiogenic treatment may be needed as described in the section above.
Indeed, CN is the earliest sign of failure of LSC transplantation in patients with a chemical burn and topical anti-angiogenic therapy after stem cell transplantation may improve long-term success of limbal grafting.79 80 Conjunctivalisation of the cornea, a significant and vision-threatening hallmark of chemical burns, is at least partly mediated via the VEGF signalling pathway, so that experimentally early anti-angiogenic therapies can improve the long-term disease course after chemical burns.81 82
Therefore, there are unmet medical needs for anti-angiogenic therapies both in the early phase after chemical burns and prior to and/or after transplantation of LSC.
A number of inflammatory conditions such as Meibomian gland dysfunction, atopic keratitis, ocular eczema, pterygium, Stevens–Johnson syndrome and graft-versus-host disease may lead to variable degrees of CN. Steroid treatment is generally the primary intervention, within the limitations discussed above. Additional benefit may be derived in these conditions from a novel anti-angiogenic therapy.83
Treatment of CN: clinical study inclusion criteria and endpoints
Because of the unmet medical needs described above, the authors consider that it is important to discuss the key elements of clinical trials aimed at identifying new approaches for the treatment of CN in the different associated diseases. Indeed, clinically meaningful and clear inclusion criteria and endpoints are currently lacking and a consensus on these variables will help to investigate and to compare the safety and efficacy of novel anti-angiogenic therapies specifically aimed at CN treatment and/or prevention. It was the authors' choice to address in this section only the most common contexts of CN, that is, infectious keratitis and keratoplasty.
CN in the context of infectious keratitis
In all patients with infectious keratitis considered for inclusion in clinical studies, primary pathogen control with anti-infective agents must have been established, or have failed. Inflammation should be controlled with steroids as appropriate.
In patients with HSK, neovascularisation associated with large stromal feeder vessels, active stromal new vessels and/or new vessels leaking lipids should be present and documented.
In patients with infectious keratitis other than HSK, a progressive, uncontrolled or chronic infection should be present and documented, or a risk to central vision by CN should be established.
Clinically relevant endpoints
Inhibition of CN is the most relevant clinical primary endpoint to demonstrate the efficacy of any novel anti-angiogenic treatment of CN in keratitis from infectious origin.
Evaluation of regression of CN (becoming ghost vessels) is also a desirable clinical outcome that should be considered as a clinically useful secondary endpoint. However, regression is currently challenging to measure, as corneal fluorescein angiography with its associated risks may be required. Impairment of VA may be the result of pathological CN. The effect of a novel anti-angiogenic treatment on VA is expected to derive from its direct effect on CN itself. Therefore, VA could be considered only as a relevant secondary endpoint.
CN in the context of keratoplasty
The inclusion criteria will depend on the setting of the study, that is, pre- or post-keratoplasty. In patients considered for inclusion in the pre-keratoplasty setting, stromal CN should be present and documented in the intended graft area. In studies in the post-keratoplasty setting, inclusion criteria depend on whether the intervention is intended to treat or to prevent CN and CG rejection (CGR). If the intention is treatment, post-keratoplasty patients who develop active stromal neovascularisation or with active CGR should be included. If the intention is prevention, recipients at HR of CGR should be included in initial studies. Subsequently, a broader population (mixed risk for CGR) may be chosen for inclusion in clinical studies. Inflammation should be controlled with steroids as appropriate.
Clinically relevant endpoints
The main goal after keratoplasty is to maintain clarity of the graft and to avoid its rejection. However, when studying a novel anti-angiogenic product, clear corneal graft survival, even if being the final desired outcome, is not considered as an appropriate primary endpoint as many variables other than CN (eg, active inflammation, irritating sutures, uncontrolled intraocular pressure, etc) may affect graft clarity. However, investigations of clear graft survival as a secondary endpoint is highly recommended, possibly in an open label follow-up context. In this context, a clear identification of the causes and associated factors of CGR that may arise should be ensured, to clearly separate the effect of CN from other possible confounding variables.
Inhibition of CN is a clinically valid surrogate primary endpoint for CGR, as the causal relationship of pre- and post-graft CN and CGR has been abundantly documented in the literature (see ‘Penetrating keratoplasty’). As discussed above, regression of CN and improvement of VA could be considered as relevant secondary endpoints.
This expert roundtable outcome report defines several very serious unmet medical needs in the treatment of CN associated with underlying corneal diseases. We propose a framework for the clinical research that is needed to identify suitable agents, their optimal dosing and treatment duration, the appropriate time points at which to start treatment for each underlying condition, and potential treatment side-effects. We anticipate that the anti-angiogenic strategy will become an integral part of the treatment regimen to address the unmet medical needs in the field of CN and to avoid the dramatic complications of CN.
Of note, it will be necessary to further define standard measurement techniques for the evaluation of CN. Imaging techniques (with associated analysis software) aimed at morphometrically evaluating the corneal area covered by CN, such as analysis of standardised corneal slit-lamp pictures, have been developed that allow an overall scoring of the vascularised surface.53 84 Further refinement of imaging techniques that would allow assignment of vascular depth and quantification of vascular dynamics would be highly welcome.
This Roundtable was conducted during the 5–9 June 2010 World Ophthalmology Congress in Berlin, Germany.
Funding IZKF Erlangen (A9) and Voisin Consulting Life Sciences, France. The planning of this study, interpretation of findings, writing and conclusions of manuscript were undertaken entirely independently of the company interests.
Competing interests CC is a consultant to Gene Signal; all other authors except RD, LAF and DM received travel reimbursement and honoraria from Voisin Consulting Life Sciences, France.
Provenance and peer review Not commissioned; externally peer reviewed.