Thomas J. Wolfensberger MD, PD, MER Department of Vitreretinal Surgery Jules-Gonin Eye Hospial University of Lausanne 15, Avenue de France CH-1004 Lausanne Switzerland Tel: + 41 21 626 8307 Fax: + 41 21 626 8144 Email: firstname.lastname@example.org
Macular recovery after surgery for retinal detachment (RD) depends on preoperative and postoperative predictive factors. Preoperative visual acuity is the main preoperative factor correlating positively with good macular recovery. Preoperative factors, which influence macular recovery negatively, include duration of macular detachment, height of macular detachment and vitreomacular traction. Postoperative factors, which influence macular recovery negatively, include cystoid macular oedema, epiretinal membranes, retinal folds, subretinal retinal pigment epithelium (RPE) migration and persistent subretinal fluid on optical coherence tomography (OCT). According to the latest available data, a detached macula has to be reattached within 5 days to optimize functional recovery. However, new therapeutic options such as exposure to hyperoxia or different growth factors may help to improve the final visual outcome in the presence of an already detached macula.
Incomplete visual recovery following surgery for retinal detachment (RD) has been a matter of debate for decades. Algernon Reese was the first to propose a hypothesis attributing poor postoperative central vision to the presence of cystic macular degeneration in the detached macula (Reese 1937) (Fig. 1). Over the ensuing years, several pre- and postoperative factors as well as different surgical techniques have been evoked as crucial factors that may influence visual recovery after macula-off RD. Some recent experimental models of RD have further contributed to our understanding of the pathophysiology of macular recovery by identifying biochemical changes as potential targets for pharmacological treatments. On the clinical side, new tools, such as optical coherence tomography (OCT), have also become very useful in elucidating the cause of delayed or incomplete visual acuity (VA) recovery after successful surgery for RD.
This review summarizes the current literature on visual recovery after macula-off RD, as well as advances in the identification of anatomical and biochemical changes after RD. The major clinical pre- and postoperative factors correlated with final macular recovery are viewed in the light of recent advances in the pharmacological enhancement of macular recovery. For the review of the literature, we performed a MEDLINE search for all published studies on visual or macular recovery after RD surgery, and selected those that provided statistical analyses of predictive factors of postoperative functional recovery in large groups of patients.
The first experimental studies on RD were performed by Robert Machemer, who published the results of a detailed analysis of experimental RD performed on owl monkeys (Kroll & Machemer 1968; Machemer 1968a, 1968b; Machemer & Norton 1968). Histological changes included the formation of cystoid spaces in the retina, which were first observed in the inner plexiform and nuclear layers with subsequent migration towards the outer layers as the RD persisted. Similar cysts were also observed at the transition zone between the outer and inner segments of the photoreceptors, with a degeneration of the outer segments (Fig. 2). The cyst size increased with the duration of the RD, thus increasing total retinal thickness. Interestingly, the height of the detachment was important to the extent that a shallowly detached retina presented less marked, albeit similar, anatomical modifications. It was proposed that diffusion of nutritive products from the choroid through the subretinal fluid remains possible when the detachment is shallow. With the techniques available at the time, histologically visible cell death in the retina was not observed until 14 weeks after detachment, while photoreceptor outer segments showed irregularities at only 1 day after RD. Thus, the photoreceptors appeared to be the first structure to be affected by the loss of contact with the retinal pigment epithelium (RPE) (Machemer 1968a), and both the duration and the height of the detachment affected the severity of the morphological changes in the detached retina of both owl monkeys and cats (Machemer 1968a; Anderson et al. 1983). The RPE did not appear to degenerate outright, but it did show a change in the shape of its cells, which became rounded at their apex. This appeared to happen as a consequence of the retraction of the apical microvilli and was also observed in rabbits (Immel et al. 1986) and cats (Anderson et al. 1983)(Fig. 3). In addition, as early as 1 day after detachment, a proliferation of RPE cells was observed in the cat model of RD at the interface between the RPE and photoreceptors (Anderson et al. 1981). Both shape modification and the proliferation of RPE cells may thus threaten functional recovery of the macula, emphasizing the importance of an intact interface between the photoreceptor outer segments and the RPE. After the retina had been reattached surgically, absorption of retinal oedema occurred after 1 day in owl monkeys. Recovery of the outer segments and of RPE morphology was more protracted, as was the restoration of the vertical arrangement of the photoreceptors (Machemer 1968b; Anderson et al. 1983). This may explain the gradual rather than acute improvement in VA and the persistence of metamorphopsia after successful reattachment of the macula.
The exact mechanism of retinal cell death after RD has not yet been established in detail. However, there are some data that indicate that it is not the process of necrosis, but rather the programmed cell death of apoptosis with specific recognition and phagocytosis of selected cells that plays a crucial role. These data stem from experimental animal models (Cook et al. 1995; Berglin et al. 1997) and from human tissue (Chang et al. 1995), and the findings were based upon DNA nick end labelling and electron microscopy. Chang et al. (1995) were the first to report apoptosis in human eyes enucleated after traumatic RD. Using the same methods of analysis, Berglin et al. (1997) presented evidence of apoptosis of retinal cells in an experimental model of RD in the rabbit, where a viscous sodium hyaluronate gel was injected into the subretinal space to create and maintain the RD over a prolonged period of time. There was a linear decline in the number of photoreceptor cell nuclei during the time period of RD, leading to a decrease of 90% in the photoreceptor nuclei count after 4 weeks (Fig. 4). This observation contrasts quite considerably with previous data in owl monkeys, where up to 14 weeks had to pass before cell death occurred (Machemer 1968a). However, this difference can be attributed to the more viscous solution of sodium hyaluronate used to induce the RD in the rabbit. The gel was solid enough to inhibit the passage of oxygen, glucose and other nutrients from the RPE to the photoreceptor, thus accelerating the process of cell death.
As visual function relies on complex biochemical interactions, the analysis of retinal amino acid metabolism after RD may also be of particular interest to elucidate short and longterm VA recovery after RD. Recently, biochemical analysis of a detached cat retina showed that the amino acid contents, such as glutamate, taurine and gamma-aminobutyric acid, of Müller cells, RPE and neural cells showed marked changes as early as 3 days after RD (Marc et al. 1998). These observations may help to pave the way for further understanding of the biochemical effects of RD. Nevertheless, the extent of signature restoration after RD has still to be determined.
Clinical appraisal of macular recovery
Several tools have been developed over the last few decades to optimize the clinical appraisal of macular recovery after macula-off RD. These are listed in Table 1 and further described below. Visual acuity and a clinical examination of the macula constitute the most important indicators of macular integrity. Both direct and indirect ophthalmoscopy with contact or non-contact lenses are valuable tools for macular evaluation. Indirect ophthalmoscopy provides a better fundus visualization through media opacities, and red-free light may be used to better delineate epi- and intraretinal lesions such as membranes or cystoid macular oedema. However, the diagnosis of macular lesions is often difficult to establish solely by fundus examination, and ophthalmoscopic assessment of the macula is examiner-dependent.
Table 1. Clinical and technical investigations useful in the appraisal of macular recovery after macula-off retinal detachment.
Additional technical tools for the appraisal of visual recovery have therefore been developed. Examination using the Amsler grid has, for example, allowed a qualitative evaluation of persistent macular defects after RD surgery (Kreissig et al. 1981). Similarly, contrast sensitivity and colour vision testing can be useful to monitor functional and qualitative recovery; both may be altered by macular oedema or previous RD and show slower recovery than VA (Chisholm et al. 1975; Kreissig et al. 1981; Liem et al. 1994). Furthermore, Stiles−Crawford function measurements can complete the functional analysis of macular recovery (Fitzgerald et al. 1980). This examination evaluates the directional sensitivity of photoreceptors, showing a peak in the measured function if the photoreceptors have a central orientation in the area tested. Stiles−Crawford function is a very sensitive test as it samples thousands of cells, while Snellen acuity tests a range of merely a hundred or so cells. As previously pointed out in an experimental model of the reattached retina, photoreceptors may show a change in their spatial orientation by becoming more oblique; Fitzgerald et al. (1980) showed decreased Stiles−Crawford function in humans after surgery for RD with progressive improvement over time, corroborating previous observations made in owl monkeys. Fundus reflection densitometry has also been used to assess visual recovery by monitoring the macular photopigment density, an indicator of the functional recovery of the photoreceptor−RPE complex (Liem et al. 1994). Indeed, VA improvement tends to slacken several months after surgery for RD, while colour discrimination and foveal densitometry continue to improve over a more extended period of time (Chisholm et al. 1975; Liem et al. 1994). Electroretinograms performed after successful surgery for RD can show specific changes, suggesting a more profound decay of the S-cone system than of the L,M-cone system (Hayashi & Yamamoto 2001). Clinically, persistent pseudoprotanomaly was observed using the Nagel type 1 anomaloscope up to 1 year after RD surgery (Liem et al. 1994), implying a selective impairment of the L-cone system. Among the different psychophysical measurements, perception of the Haidinger phenomenon is impaired after RD surgery and this impairment appears to be related to the duration of RD (Zygulska-Mach et al. 1979).
Since its discovery by Huang et al. (1991), OCT technology has become an invaluable tool for rapid non-contact imaging of the eye, and in particular of the retina (Hee et al. 1995). It has for the first time allowed quantitative cross-sectional analysis of the retinal layers to be obtained. Although fluorescein angiography can confirm the diagnosis of macular oedema in most instances, only OCT can definitively measure the retinal thickness, thus enabling a more precise and reproducible assessment to be made. The rate of detection of macular oedema by means of OCT has been described as higher than that of fluorescein angiography (Hee et al. 1998; Brancato et al. 2002). More importantly, OCT can be useful in detecting shallow subfoveal postoperative fluid accumulation after macular-off RD that cannot be seen clinically or on fluorescein angiography (Wolfensberger & Gonvers 2002; Hagimura et al. 2002). It is not yet clear whether or not these lesions seen on OCT predispose to lower VA recovery. Optical coherence tomography has also been used to measure the height of the macular detachment and a correlation between this and final VA have been sought (Hagimura et al. 2000; Lecleire-Collet et al. 2005).
Factors influencing macular recovery
Several large studies of final VA after successful surgery for macula-off RD have suggested that, in approximately 39% of cases, postoperative vision will be around ≥ 20/50 (Burton 1977; Tani et al. 1981). Visual recovery after successful surgery for macula-off RD thus appears to be incomplete in the majority of patients, and this observation has been attributed to several pre- and postoperative factors.
Factors influencing macular recovery after RD
Relationship to visual acuity
Pre-operative Visual Acuity (VA)
Low pre-operative VA predisposes to reduced post-operative VA.
Gundry and Davies 1974 Burton 1977 Burton and Lambert 1978 Tani et al. 1981 Friberg and Eller 1992 Ross and Kozy 1998
Duration of macular detachment
Very early reports suggested that surgical attachment of the retina within up to 6 weeks after macula-off RD would not compromise final visual acuity recovery.
Latest large studies conclude that delay in surgical repair within the first week does not preclude good visual recovery after macula-off RD.
Reese 1937 Jay 1965 Davies 1972 Gundry and Davies 1974 Grupposo 1975 Tani et al. 1981 Burton 1982 Ross and Kozy 1998 Hartz et al. 1992
Height of macular detachment
Clinical and OCT-based RD height estimation showed that the extent of macular elevation correlates with impaired functional recovery.
Machemer 1968 Davidorf et al. 1975 Tani et al. 1981 McPherson et al. 1982 Kreissig 1977 Hagimura et al. 2000 Lecleire-Collet 2005
Vitreoretinal traction negatively influences the success of retinal detachment surgery.
Wilkinson and Bradford 1984
Cystoid macular edema
Cystoid macular edema appears to be one of the most frequent post-operative macular complications although it generally regresses spontaneously within two years after surgery.
Gundry and Davies 1974 Cleary and Leaver 1978 Burton and Lambert 1978 Meredith et al. 1980 Bonnet et al. 1983 Sabates et al. 1989
Epiretinal membranes appear to be a more long-standing cause of post-operative incomplete recovery than cystoid macular edema.
Gundry and Davies 1974 Cleary and Leaver 1978 Burton and Lambert 1978 Meredith et al. 1980 Bonnet et al. 1983
Retinal folds have been identified as negative predictive factors of visual recovery, occurring mostly after vitrectomy and complete gas fill for a bullous RD.
Meredith et al.1980 Sabates et al. 1989
Subretinal RPE migration
Macular subretinal pigmentary changes have been identified as a negative predictive factor for visual recovery.
Burton and Lambert 1978 Cleary and Leaver 1978 Sabates et al. 1989
Persistent subretinal fluid
Longstanding but clinically invisible small amounts of subfoveal fluid can be diagnosed with OCT and may be associated with reduced visual recovery after macula-off RD.
Retinal oedema in the detached macula has for many decades been implicated in limited visual recovery after macula-off RD. The earliest effort to resolve this was undertaken in 1937 by Algernon Reese. Using the direct ophthalmoscope, he studied 12 eyes with simple RD including the macula, and observed in all eyes cystic spaces in the macula at the level of what he thought to be the external plexiform layer (Fig. 1). Reese described these lesions as cystic degeneration of the macula and concluded that they were responsible for reduced vision after RD surgery. Many decades later, Machemer (1968b) confirmed the importance of such cyst formation in the retina in experimental RD in the monkey. However, an OCT-based pilot study (Wolfensberger & Gonvers 2002) was unable to show any statistically significant correlation between the extent of preoperative macular oedema, as measured on OCT, and final postoperative VA.
Preoperative VA appears to be a significant predictive factor of visual recovery (Gundry & Davies 1974; Burton 1977; Burton & Lambert 1978; Tani et al. 1981; Friberg & Eller 1992; Ross & Kozy 1998). In a large series of 473 cases of macula-off rhegmatogenous RD, for which the anatomical success rate was 90%, favourable functional results, defined as postoperative VA ≥ 20/50, were positively correlated with preoperative VA. When the preoperative VA was ≥ 20/50, postoperative VA was ≥ 20/50 in 75% of cases (Tani et al. 1981). Furthermore, preoperative VA appears to correlate well with the anatomical reattachment success rate (Burton 1977; Tani et al. 1981). In slight contrast, Friberg & Eller (1992) recorded VAs of patients with poor to unmeasurable Snellen VAs using the potential acuity meter (PAM) and described a highly significant correlation between preoperative PAM VA and postoperative Snellen VA.
Duration of the RD is another significant predictive preoperative factor of retinal recovery, although the exact duration after which visual prognosis may be compromised is matter of debate (Reese 1937; Jay 1965; Davies 1972; Gundry & Davies 1974; Grupposo 1975; Tani et al. 1981; Burton 1982). The cut-off point after which visual recovery is thought to be worse has fallen dramatically over the last century. In the early 1930s, it was thought that operating a macula-off RD for up to 6 weeks after the initial event would not compromise VA recovery (Dunnington & Macnie 1934). Reese (1937) reduced this time to 1 month as he observed that 38% of cases of macular-off RD lasting less than 1 month had a postoperative VA of 20/30, while this percentage was much lower if the detachment lasted longer than 1 month. Davidorf et al. (1975) and Tani et al. (1981) also observed more favourable results when the RD had lasted less than 1 month. Other hypotheses included the notion that there was a first cut-off point in visual recovery after 1 day of detachment and a second after 6 months (Gundry & Davies 1974). Davies (1972) had previously estimated the critical duration to be 1 week, while others drew the line at 2 weeks (Jay 1965; Grupposo 1975). However, based on the most elaborate study on the subject, which included several hundred patients, Burton (1982) observed that no patient recovered VA of 20/20 if the duration of the RD exceeded 5 days. If the RD persisted for longer than 5 days, one line of vision was lost for each additional week before surgery until 27 days. If the RD persisted for 4 weeks, one line was lost every 10–11 additional days until 70 days. The decline in visual recovery appears to take place in an exponential fashion.
In a large, prospective study of 100 cases with macula-off RD, Ross & Kozy (1998) also addressed the question of whether or not surgical delay within the first week would alter final postoperative VA. They compared three groups of patients operated on between 1 and 2 days, 3 and 4 days and 5 and 7 days after macular detachment. The average follow-up in this study was 10.5 months. There was no statistical difference in visual recovery between the three groups. The authors concluded that a delay in surgical repair within the first week does not preclude good visual recovery. In view of these data and the high costs of out-of-hours emergency surgery, a well planned surgical intervention can reasonably be advocated if performed within the first week of macular detachment (Hartz et al. 1992).
The height of macular detachment may also interfere with visual recovery in a negative way. In accordance with observations made on experimental models, in which photoreceptor degeneration increased with greater distance of the detached retina from the RPE (Machemer 1968a), the extent of macular elevation correlates with impaired functional recovery. Indeed, several clinical studies have confirmed this relationship (Davidorf et al. 1975; Kreissig 1977; Tani et al. 1981; McPherson et al. 1982).
Recently, OCT has allowed objective and quantitative measurements of RD height. In a prospective study of 25 patients with macula-off RD, preoperative VA was negatively affected by the height of the RD as measured by OCT. Furthermore, in patients with a highly detached macula, morphological changes such as retinal splitting and undulation of the outer retina could be observed. This was thought to be a risk factor for impaired postoperative visual recovery (Hagimura et al. 2000). To further determine whether or not OCT can help predict postoperative macular outcome, another study examined 16 patients with OCT preoperatively and 1, 6 and 10–12 months after surgery for macula-off RD (Wolfensberger & Gonvers 2002). Although four categories of morphological macular changes were observed (Fig. 5), including two categories of macular oedema, none of them was clearly correlated with a worse postoperative VA. In accordance with data presented previously (Hagimura et al. 2000), Lecleire-Collet et al. (2005) very recently reported a negative correlation between the final postoperative VA and both the height of the RD and the extent of structural changes in the detached retina. This prospective study included 20 eyes with macula-off RD examined with OCT. It concluded that no statistically significant correlation existed between the extent of retinal oedema in the detached retina and final VA, confirming previous reports (Wolfensberger & Gonvers 2002). The study by Lecleire-Collet et al. (2005) also reported that the final postoperative VA was negatively correlated with preoperative distance from the central fovea to the nearest undetached retina on OCT. Although Friberg & Eller (1992) reported a similar correlation, it should be noted that this factor is closely linked to the height of RD at the fovea, limiting its predictive value for the calculation of final VA.
Finally, the presence of preoperative vitreomacular traction may influence the postoperative macular recovery to an important degree. Vitreoretinal traction can, for example, be responsible for anatomical failure of RD surgery (Wilkinson & Bradford 1984).
It has been known for many decades that several macular abnormalities, such as cystoid macular oedema, epiretinal membrane formation, retinal folds and pigment migration, can occur after successful surgery for RD.
In some cases, no clinically detected macular changes are observed and reduced postoperative VA remains unexplained. Recently, a possible cause of unexplained postoperative visual loss was identified. Using OCT, a foveal detachment with residual subretinal fluid, not visible clinically or on fluorescein angiography, was found in up to two-thirds of patients after buckle surgery (Wolfensberger & Gonvers 2002)(Fig. 6). This trend implies that persistent subfoveal fluid accumulation is correlated with a worse visual outcome. Indeed, eight eyes (50%) showed persistent subfoveal fluid accumulation at 6 months after surgery and one eye continued to do so at 12 months. A second prospective study published a few months later also described this phenomenon of clinically silent subfoveal fluid accumulation on OCT in seven of 15 cases at 1 month after surgery (Hagimura et al. 2002). The residual foveal detachment resolved within 12 months in most cases but was correlated with delayed visual recovery. The presence of this foveal detachment after RD surgery was later confirmed by others (Lecleire-Collet et al. 2005).
To determine whether or not the type of surgical procedure could influence the persistence of postoperative subfoveal fluid, a prospective comparative study was performed on 33 patients, of whom nine were treated with episceral buckle, while 24 were operated on using vitrectomy, cryotherapy and fluid−gas exchange (Wolfensberger 2004). Interestingly, 1 month after RD surgery, postoperative persistent subretinal fluid was observed only in cases operated on with episceral buckles; all cases operated on with vitrectomy showed an attached fovea both clinically and on OCT. The reason for this difference is still unexplained and others have since reported subfoveal fluid even after vitrectomy and gas injection for RD. Further studies are needed to elucidate this issue.
Non-surgical enhancement of macular recovery
Apart from the surgical apposition of the photoreceptor to the RPE, little could be done until now to speed up retinal recovery after macula-off RD. However, recent experimental evidence suggests that additional measures may be taken to hasten photoreceptor recovery. It is well known that the outer layers of the retina obtain their nutrition from the choriocapillaris. Retinal detachment disrupts the RPE−photoreceptor complex and thus prevents the diffusion of important metabolic resources, such as oxygen and glucose. The separation of the retina from its nutritive support leads to the degenerative and apoptotic responses described earlier. A recently published study using a feline model of RD compared two groups of adult cats (Mervin et al. 1999). The first group was placed in a hyperoxic chamber (70% oxygen) for 3 days after inducing RD, while the second was placed in a normoxic chamber (21% oxygen). In the hyperoxic group, the retina showed fewer degenerative changes and a reduction in the frequency of apoptotic cells (Fig. 7). These results suggest that hypoxia and hypoglycaemia caused by the RD are among the causes of the degenerative apoptotic retinal modifications. In a parallel article, it was reported that hyperoxic conditions after inducing RD in cats allowed the reduction of proliferation and morphological changes of Müller cells as well as glutamate cycling deregulation (Lewis et al. 1999). As the feline retina is dominated by rods, a similar experiment was later conducted in ground squirrels, whose retina is cone-dominated (Sakai et al. 2001). This latter study showed a similarly protective effect of oxygen supplementation on photoreceptor degeneration. It has thus been suggested that oxygen supplementation between diagnosis of RD and surgery in humans may help to improve VA recovery after surgery.
Pharmacological modulation of photoreceptor cell apoptosis may represent another target for novel therapies. As previously mentioned, apoptosis is an important process of cell death after RD and is probably regulated by the Bcl-2 protein family. This family is divided into antiapoptotic and proapoptotic members. Bax is a proapoptotic protein and an experimental Bax-deficient mice model of RD has shown that this protein is a major inducer of RD-related apoptosis. Because Bax-deficient mice were to some extent protected from apoptotic cell death, Bax may be a potential new target for pharmacological enhancement of macular recovery (Yang et al. 2004).
Finally, several different experimental animal models of RD have highlighted the significant role of signalling pathways influenced by RD, such as fibroblast growth factor (FGF) and fibroblast growth factor receptor 1 (FGFR1). These factors and perhaps other protective factors may be targeted soon for pharmacological treatment (Ozaki et al. 2000; Geller et al. 2001). Experimental gene therapy by adeno-associated virus vector has been described in rat RD models for glial cell line-derived neurotrophic factor delivery (Wu et al. 2002), and results showed that treated eyes had a reduction both of apoptotic cells and of Müller cell proliferation, with preservation of the outer segments and outer nuclear layer structure. However, the best tools of delivery of other growth factors and their longterm effects have yet to be established.
The conundrum of macular recovery after RD has occupied the ophthalmic research community for over a century. Early ideas about how long a detached macula could be left to its own devices before a successful operation could still guarantee some useful vision were very lenient. More detailed research towards the end of the last century has narrowed this period down to a few days rather than weeks or months. Clinically, the two most important factors correlated with final macular recovery are preoperative VA and the duration of the RD. Recent studies have also shown that the height of the RD at the fovea, as measured with OCT, may also be an important indicator for postoperative macular recovery. Moreover, cystoid macular oedema and preretinal membranes are the main postoperative factors that can impair final VA recovery in the longterm. On a cellular level, metabolic pathway analysis has shown significant modifications in neurotransmittors as a consequence of RD, and it appears that even an RD of a very short duration may lead to degenerative retinal changes and to programmed cell death such as apoptosis. In the future, it may be possible, as some recent studies of animal models suggest, that novel therapeutic agents may help to enhance functional recovery after RD by limiting cellular death during the macular detachment.