Statistics from Altmetric.com
Trepidations of ICG use in macular hole surgery
At the close of the 20th century the medical world was witness to a most remarkable advance when therapy was developed for a previously incurable disease. Macular hole surgery is one of the great success stories in ophthalmology, if not modern medicine. Predicated upon an increased awareness of the role of vitreous in the pathogenesis of retinal disorders, this achievement results from the work of pioneering surgeons who had the ingenuity and courage to devise and attempt a new surgical approach that now restores vision to many grateful patients. However, recent modifications of the surgical technique may jeopardise visual outcomes and “lead us astray” from early successes.
Vitreous is an extended extracellular matrix, whose molecular composition and supramolecular organisation result in a clear gel that firmly adheres to the retina in youth.1–3 Ageing induces liquefaction and vitreoretinal dehiscence, which occur concurrently in the overwhelming majority of individuals, resulting in innocuous posterior vitreous detachment (PVD).4 Anomalous PVD5 results from vitreous liquefaction without sufficient vitreoretinal dehiscence. This may preclude the posterior vitreous cortex from separating cleanly from the internal limiting lamina (ILL) of the retina. The untoward consequences of anomalous PVD vary depending upon where in the vitreous body the gel is most liquefied and where on the retina there is greatest vitreous adherence.5 Anomalous PVD in the periphery, for example, results in retinal tears. Along blood vessels, liquefaction without vitreoretinal dehiscence induces vitreous haemorrhage. Another important effect of anomalous PVD is vitreoschisis, a split in the posterior vitreous cortex that has been identified with biomicroscopy,6 ultrasonography,7 and optical coherence tomography.8 Vitreoschisis has been confirmed by histopathological studies9 and has also been documented during surgery by intravitreal triamcinolone injection.10
Anomalous PVD is hypothesised to have a role in the pathogenesis of macular holes via vitreoschisis.5,11,12 Studies have identified premacular membranes by histopathology in 73% of cases (n = 22)13 and by stereoscopic fundus photography in 65% of eyes (n = 224).14 The origin of this membrane is postulated to be the outer wall of a vitreoschisis cavity in the posterior vitreous cortex, since this tissue has been identified as prefoveal vitreous by histopathology.15 In a recent clinical study16 of 69 cases of macular hole, optical coherence tomography detected a prefoveal membrane that could be an advanced form of the outer wall of a posterior vitreoschisis cavity. Migration of cells from the retina, such as fibrous astrocytes and Mueller cells, and recruitment of cells from the circulatory system by hyalocytes result in some degree of cellularity. New collagen (type I) synthesis further alters the appearance of this tissue that, when it began as the outer wall of a vitreoschisis cavity, was thin, composed primarily of type II collagen, and only contained hyalocytes. Centrifugal (outward from the fovea) traction forces are induced by the detached vitreous body, which is still attached to the peripheral circumference of the vitreoschisis cavity, where the inner and outer walls fuse into an intact posterior vitreous cortex. The extent of this anomalous PVD from the fovea has been shown to correlate with the stage of macular hole—that is, stage 2 holes have a less extensive PVD than stage 4 holes.17 Traction by this tissue is augmented by various contractile cells, especially myofibroblasts,14 resulting in a dehiscence of the central macula.5 The importance of the posterior aspect of the posterior vitreous cortex in the pathogenesis of macular holes is underscored by the salubrious outcome when this tissue is successfully removed in its entirety.
And missing thee, I walk unseenOn the dry smooth-shaven green…Like one that had been led astrayThrough the heav’n’s wide pathless way…John Milton, Il Penseroso 
Although previously praised in print,18 the seminal contributions of Kelly and Wendell19 to the treatment of macular holes cannot be overemphasised, as their pioneering work paved the way for the relatively high success rate experienced by many patients. The outer wall of the vitreoschisis cavity is usually “unseen” in their procedure until it is elevated off the retinal surface. While vitreous invisibility3 is critical to its physiological function,1–3 this poses challenges for clinical imaging.20 Echography and optical coherence tomography often fail to identify the outer layer of a vitreoschisis cavity because it is usually thinner than the level of resolution of these techniques. In an attempt to assure that the “unseen” pathogenic tissues are removed and thereby increase the rate of hole closure, surgeons began to dissect farther posteriorly and tried to remove what was thought to represent the ILL. The results from one large retrospective study,21 comparing no ILL peel in 417 cases with ILL peeling in 175 cases found that ILL peeling increased the initial closure rate from 81% to 92% and decreased the reopening rate from 7% to 0.6%. However, there did not appear to be any difference in visual outcomes when the hole was closed by either technique. Complete ILL removal would damage Mueller cells and negatively impact upon retinal neurophysiology and vision. Thus, it is highly unlikely that the entire ILL is removed in patients who experience improved vision. Rather, the deeper dissection undertaken during attempted ILL removal most probably creates a surgical plane between the three laminae of the ILL,22 leaving the innermost layer, the lamina rara externa, intact and the underlying neural retina undamaged. In surgery, however, it is often difficult to accurately assess whether the ILL is being removed in part, in total, or at all.
Intraoperative efforts to enhance the visualisation of pathogenic tissues in macular holes led to the use of indocyanine green (ICG) dye to stain the tissue. Unfortunately, this was undertaken without any preclinical studies to determine safety and efficacy. Thus, while this “smooth-shaven green” approach did increase the rate of hole closure, it was associated with untoward effects on postoperative visual acuity.23–25 That ICG was the cause of poor visual acuity in spite of hole closure was, to a degree, substantiated in a subsequent study where one of these same surgeons found that a short exposure to a lower dose of ICG was associated with improved visual acuity.26 Why ICG was not found to be associated with poor postoperative visual acuity in other series27 may relate to differences in surgical technique that probably employed a more shallow plane of dissection. However, a true understanding of these discrepant findings requires a better understanding of the mechanism of ICG toxicity.
ICG may have untoward effects via several mechanisms that are not mutually exclusive. As alluded to above, the use of ICG could result in a deeper surgical plane of dissection with damage to neural retinal elements. Histopathological analysis of tissues removed at surgery support this postulate.28 There may a direct toxic effect upon retinal neurons by a chemical interaction. In postmortem human eyes, ICG alone was associated with rupture of Mueller cells and detachment of the ILL.29 Apoptosis was induced in human RPE cells in culture with ICG.30 Since ICG is a photosensitiser, there is potential for light toxicity via a photodynamic effect. Studies31 have shown that in the presence of ICG, light from a standard endoilluminator has a dose dependent toxicity on retinal ganglion cells in vitro. Experiments in postmortem human eyes identified wavelengths longer than 620 nm as phototoxic, determined by light and electron microscopy.29 However, other postmortem studies32 in pig eyes found no such effects. The results of postmortem studies are often difficult to interpret, however, as they sometimes lead “through the heav’n’s wide pathless way,” and thus in vivo experimentation is needed to properly address this issue.
In this issue of the BJO (p 897) Kwok and associates in Hong Kong report the results of in vivo studies on the effects of ICG plus endoillumination in rabbits, assessed by electroretinography (preoperatively and postoperatively) and histopathological analyses. At 1 week after surgery, there was significant reduction in the light-adapted a-wave amplitude and significant delays in the light and dark adapted b-wave latencies. Histopathological findings included focal loss of photoreceptor outer segments, some foci of photoreceptor absence, focal oedema of the inner and outer nuclear layers, and localised areas of RPE irregularities. In the absence of a retinal break, it is surprising to find RPE and outer retinal abnormalities. Since these findings were focal in distribution, the abnormalities may have resulted from mechanical trauma (retinal elevation off the RPE) during the experimental surgery. It is well known that the rabbit vitreous is very firmly adherent to the retina. Subsequent studies must rule out any mechanical effects that might have been induced during dissection of the posterior vitreous cortex off this very adherent interface. One possible solution would be to undertake pharmacological vitreolysis33,34 with agents intended to lyse the vitreoretinal interface, making dissection of the posterior vitreous cortex easier with less traction upon the retina. However, as these enzymes might introduce other effects, perhaps even artefacts, it would be simpler to employ a species with less adhesion at the vitreoretinal interface, such as the mini-pig, whose vitreoretinal interface more closely resembles that of humans.
The authors are to be thanked for contributing to our understanding of the effects of ICG upon retinal physiology and structure. As their studies were conducted in the absence of a retinal hole, the findings may also help interpret the observations of ICG toxicity in surgery for macular pucker35 and diabetic macular oedema.36
Note in Proof
Trepidations of ICG use in macular hole surgery