Statistics from Altmetric.com
When scleral entry sites over the pars plana were first used for vitrectomy access, concerns were raised that anterior vitreous complications might compromise surgical outcomes.1 2 After all, it had long been recognised that, in chronic uveitis or in inflammation complicating a penetrating injury, the non-pigmented ciliary epithelium might undergo proliferation and fibrous metaplasia, accompanied by a stroma derived vascular component, to form a “cyclitic” membrane using the anterior surface of the vitreous as a scaffold; this could lead to traction on the ciliary processes and peripheral retina, hypotony, and phthisis.3 The retrolental membrane might also include glial and retinal pigment epithelial cells in cases of concomitant rhegmatogenous retinal detachment or an “ingrowth” of episclera derived fibrovascular tissue at sites of penetration. Happily, such foreboding over pars plana vitrectomy has proved largely unfounded, presumably reflecting in part the dispersion of any inflammatory mediators involved. But vitrectomy for the ischaemic retinopathies, especially for proliferative diabetic retinopathy (PDR), has been a notable exception.
Following vitrectomy for PDR, any residual new vessels located posterior to the vitreous base usually undergo regression,4 but some eyes develop florid basal neovascularisation as a supplement to the normal wound healing process at the sclerotomies.2 5 6 Such entry site neovascularisation (ESNV) may arise as an isolated event after otherwise successful surgery when it represents one of a number of possible sources of “delayed” diabetic vitreous cavity haemorrhage (DVCH)—that is, bleeding after an initial haemorrhage-free period of 3 or more weeks post-vitrectomy. Other possible culprits for delayed DVCH include “secondary haemorrhage” from shedding of thrombus (for example, from a sclerotomy or from a retinal vein from which a neovascular outgrowth had been avulsed during vitrectomy) or detachment of residual attached post-basal cortex and avulsion of fibrovascular tissue contained therein. The mechanism of bleeding from ESNV (that is, whether through some mechanical event or simply as a manifestation of friability) is uncertain as is the nature of its spontaneous regression.6 In other circumstances, however, ESNV forms part of a more widespread uveal proliferation involving both the iris and ciliary body and frequently associated with retinal detachment as a cause or consequence of the vascularised cyclitic membrane.1 2 7 8
The term “fibrovascular ingrowth” has been used inadvisedly2 6 to denote the intraocular vascularised proliferation at sclerotomies for it implies an episcleral contribution to, or source of, ESNV9 (which is seldom the case). There has also been a dearth of information regarding the incidence of ESNV post-vitrectomy. Michels4 discovered only one example of ESNV within 18 months of successful vitrectomy in an early series of 107 eyes with PDR, most of which must be presumed to have been aphakic; many such aphakic vitrectomised eyes nevertheless developed rubeosis iridis, ostensibly though removal of the hyalolenticular barrier to anterior diffusion of cytokines and growth factors.4 7The source of DVCH in eyes kept phakic in later years was seldom identified,10 11 albeit, many surgeons believed that ESNV was a major contributor to DVCH with consequential fears that any additional entry sites would increase the potential for rebleeding. However, two recent studies have provided new insights into the contribution of ESNV to DVCH. Firstly, Hotta and colleagues12 found ESNV at one or more sclerotomies in six out of 13 eyes (46%) undergoing vitreous cavity washout (VCWO) for DVCH. Of the 12 affected sclerotomies, 11 had shown a “low reflective trapezoid image” on ultrasound biomicroscopy (though the overall predictive value of this ultrasonic feature for ESNV was only 50%). Secondly, West and Gregor, reporting in this issue of theBJO (p 822), found ESNV in 11 out of 19 eyes (58%) undergoing VCWO for DVCH. The presence or absence of an episcleral vessel at a sclerotomy (putatively a sentinel of ingrowth) was an unreliable predictor of ESNV and no instance of more widespread fibrovascular proliferation in the anterior vitreous, whether of cyclitic or retinal origin, was discovered during VCWO. Vessels derived from the anterior retina are the distinctive and distinguishing feature of so called anterior hyaloid fibrovascular proliferation (AHFP) which complicated vitrectomy for PDR in a high proportion (13%) of a particular series of cases.13 14However, such rampant “retrolental neovascularisation” (RLNV),15 apparently arising independently of ESNV,14 15 was probably a consequence of concomitant scleral buckling (causing superimposed ischaemia).
Three factors appear to be important in the pathogenesis of ESNV (and indeed RLNV/AHFP). Firstly, continuing post-vitrectomy secretion of ischaemic retina derived angiogenic growth factors is essential since ESNV is not seen after vitrectomy for non-ischaemic pathologies. Secondly, the presence of an intact anterior hyaloid barrier to egress of these growth factors via the aqueous humour is postulated,15 at least for RLNV/AHFP. Thirdly, a requirement for a vitreous scaffold for de novo growth of vessels is suggested by past experience of extension of iris neovascularisation onto the anterior hyaloid face in rubeotic, intracapsular aphakic eyes and by analogy with preretinal neovascularisation where the absence of a cortical vitreous substrate either precludes16 or restricts17 new vessel proliferation. Various preventative measures thus arise whereby the incidence of ESNV (and DVCH) might be reduced. The most obvious and currently applicable approach is to undertake scatter endophotocoagulation of all previously untreated ischaemic retina at the time of vitrectomy for PDR; this might logically be supplemented by post-oral cryotherapy behind the sclerotomies and wherever basal haemorrhagic residues or the danger of lens damage prevents fill-in laser up to the ora serrata through 360 degrees.9 18 19However, the evidence base for the effectiveness of such treatment is not substantial. Using surrogate measures for ESNV such as the incidence of DVCH, rubeosis, and VCWO historical comparisons of vitrectomy outcomes have been made before and after xenon arc19 and later argon laser endophotocoagulation18 20 became available. Liggett and colleagues20 obtained a significant reduction in delayed DVCH and VCWO using an average of only 338 endolaser burns at the time of vitrectomy but, unless there is evidence of “burn out” or complete scatter laser of the retinopathy, much more extensive treatment is generally recommended as prophylaxis.18
A second preventative approach to the phakic or pseudophakic diabetic eye is to deliberately remove a small portion of the anterior hyaloid during vitrectomy. This creates the state of “pseudo-aphakia” whereby free exchange of cells and solutes takes place between the anterior chamber and vitreous cavity.18Just as aphakia accelerates the spontaneous clearance (via the trabecular meshwork) of both immediate7 21 and delayed7 DVCH, so does pseudo-aphakia, whether induced intentionally or arising incidentally (by surgical interference with the anterior hyaloid face at one or more entry sites). A full scatter photocoagulation (to prevent rubeosis) and thorough clearance of inferior basal gel haemorrhage (to prevent immediate DVCH and erythroclastic glaucoma22) are essential to this strategy. Chemical modulation or disruption of the anterior vitreous scaffold for vessel proliferation is a potential future extension of this approach.23
So far as management of established ESNV (and RLNV/AHFP) is concerned, further “indirect” treatment using additional scatter endolaser during VCWO8 13 15 18 is the current mainstay is also promulgated by West and Gregor; an attempt to establish pseudo-aphakia should also be considered. Direct dissection of fibrovascular membranes (necessitating lensectomy) has previously been advocated for RLNV/AHFP complicated by peripheral traction retinal detachment,9 13 and a similar method of managing extensive or persistent ESNV was employed by West and Gregor even in eyes without such detachment. Others have recommended “lens sparing surgery” for ESNV using silicone oil to obstruct access of growth factors to the sclerotomy sites,8 but reparative epiretinal fibrosis and retro-silicone oil neovascularisation15 24 are a danger. Such measures remain to be justified by longer term follow up of individual cases and formal comparison with more modest approaches to DVCH uncomplicated by cyclitic traction.