An evidence-based analysis of surgical interventions for uncomplicated rhegmatogenous retinal detachment
Assistant Professor Seang-Mei Saw
Department of Community, Occupational
and Family Medicine
Yong Loo Lin School of Medicine
National University of Singapore 16 Medical
Tel: + 65 6516 4989
Fax: + 65 6779 1489
Purpose: To evaluate the various surgical interventions available for uncomplicated rhegmatogenous retinal detachment.
Methods: Reports of controlled clinical trials of surgical interventions (pneumatic retinopexy, scleral buckling and vitrectomy) for uncomplicated rhegmatogenous retinal detachment indexed in MEDLINE from 1968 to January 2006 were included. The primary outcomes evaluated included single-operation reattachment rates, multiple reoperation reattachment rates and improvements in visual acuity (VA).
Results: We found five controlled trials (two randomized) comparing the efficacy of pneumatic retinopexy versus scleral buckling. The single-operation reattachment rates were higher for scleral buckling, but the final reattachment rates were similar. We found nine controlled trials (four randomized) evaluating vitrectomy. There were no statistically significant differences between retinal reattachment rates or final visual acuities, except in one randomized and one non-randomized controlled trial in which the VAs were significantly better in the vitrectomy than the scleral buckling group.
Conclusions: Pneumatic retinopexy is a possible alternative to scleral buckling in the treatment of uncomplicated rhegmatogenous retinal detachment. The rates of missed or new retinal breaks after pneumatic retinopexy, however, are higher than following scleral buckling. The clinical outcomes of vitrectomy for rhegmatogenous retinal detachment compare favourably.
Rhegmatogenous retinal detachment (RRD) occurs when there is a separation of the neurosensory retina from the retinal pigment epithelium (RPE) with accumulation of subretinal fluid (SRF) in the presence of one or more retinal breaks. If left untreated, RRD will often progress to involve the entire retina and cause severe visual loss. The reported incidence rates of RRD vary from 14 per 100 000 persons per year in Sweden (Algvere et al. 1999), 12.6 per 100 000 persons per year in Minnesota (Rowe et al. 1999) to 7.98 per 100 000 persons per year in Beijing (Li & the Beijing Rhegmatogenous Retinal Detachment Study Group 2003). The incidence rates of RRD have been increasing over the past few decades, partly because of the rising number of cataract operations (Ah-Fat et al. 1999).
Surgical management of RRD has evolved around finding all retinal breaks, sealing the retinal breaks by creating a chorioretinal scar around the breaks (‘retinopexy’), drainage of SRF in some cases and relief of vitreoretinal traction. In the late 1930s, scleral buckling (SB) was introduced with an encircling element to achieve the apposition of neurosensory retina and RPE while relieving the vitreous traction on the retina. Refinements of SB techniques including SB without drainage and SB with a variety of SRF drainage techniques have been tested in various clinical trials. The multitude of available surgical options belies the complexity of choice of technique for SB surgery.
Two decades ago, Hilton & Grizzard (1986) introduced the technique of pneumatic retinopexy (PR) to treat RRDs with retinal breaks in the superior eight clock hours of the retina. A gas bubble is injected into the vitreous cavity and the patient's head positioned so that the bubble tamponades the retinal breaks, while laser photocoagulation or cryotherapy is used for retinopexy. This technique has been further modified and its use extended to selected complicated RRDs (Holz & Mieler 2003). More recently, with rapid advances in instrumentation and improved success rates of vitrectomy, there has been a growing trend towards primary vitrectomy for RRD; the reasons cited for this are greater success in locating small or anterior breaks and a more complete resolution of vitreous traction.
In response to these recent trends in RRD surgery, we summarized the best available evidence for the effectiveness and safety of PR, SB and vitrectomy for uncomplicated RRD. We also report the success of various techniques for drainage of SRF and retinopexy.
Search strategy for the identification of studies
Electronic searches of the literature published from 1968 to January 2006 were conducted using PubMed and the Cochrane Collaboration database to identify articles evaluating surgical techniques for the treatment of uncomplicated RRD. The surgical approaches evaluated included PR, SB and primary vitrectomy. Only randomized and non-randomized controlled clinical trials were included. Articles were excluded if the studies were not controlled, the RRD was complicated (examples include giant retinal tears and proliferative vitreoretinopathy greater than stage B), evaluation of the efficacy of surgical treatment was not the main objective of the study or if original data were not reported. The year of publication of the article, sample size of the study and duration of follow-up were not among the exclusion criteria. Search strategies employed included key words and Boolean operators described as follows: ‘treatment and RD’ limited to clinical trials (301 abstracts found), ‘PR’ (161 abstracts found) and ‘SB and RD’ (1297 abstracts found), ‘’vitrectomy and RDs (1761 abstracts found). This search was augmented by a hand search of the reference lists of relevant articles included in the literature review. Three different investigators performed the search (S-MS, JL, AMW) and the abstracts were independently reviewed for possible inclusion.
Data extraction and study appraisal
A total of 83 potentially relevant articles were retrieved. A content assessment form was developed to extract relevant information in a standardized fashion. Information extracted included the design of the study, sample size, nature of the interventions, randomization, masking, study population characteristics, length of follow-up, loss to follow-up rate, confounding, biases and adverse effects. The outcome measures reported include the rates of single-operation reattachment at 6 months, overall reattachment with reoperations and final visual acuity (VA). There were 14 articles describing controlled clinical trials of PR, SB or vitrectomy, six randomized controlled trials (RCTs) describing SRF options during SB, and four RCTs of retinopexy options for SB in our review. The entire team of one epidemiologist and four ophthalmologists reviewed all manuscripts using the content assessment form. If there were disagreements, the team met to resolve differences to achieve a final consensus.
Pneumatic retinopexy and scleral buckling
There were five articles with sample sizes ranging from 20 to 1000 which described controlled clinical trials evaluating the efficacy of PR versus that of SB for the treatment of uncomplicated RRD (Table 1) (McAllister et al. 1988; Kreissig et al. 1989; Tornambe et al. 1991; Mulvihill et al. 1996; Han et al. 1998). Only three of the studies had been conducted after 1990 (Han et al. 1998; Tornambe et al. 1991; Mulvihill et al. 1996) and only two of the five were randomized (Tornambe et al. 1991; Mulvihill et al. 1996). Three studies had been conducted in the USA (McAllister et al. 1988; Tornambe et al. 1991; Han et al. 1998) and two in Europe (Kreissig et al. 1989; Mulvihill et al. 1996). Two of the five trials compared PR with temporary balloon buckle (McAllister et al. 1988; Kreissig et al. 1989). The trials evaluated eyes with RRD and a single break or group of breaks not larger than one clock hour in size located in the superior eight clock hours of the retina. Outcomes assessed included anatomical success defined as 6-month overall reattachment with reoperation rates, single-operation reattachment rates and postoperative VA. None of the trials were masked, except the Tornambe et al. (1991) study, where a masked observer performed best-corrected VA measures at the follow-up visit. Often, the final outcomes were not measured at predefined follow-up times and the final outcomes were often assessed at different set time-points in the various trials. The follow-up periods ranged from 6 months to 7.5 years.
Table 1. Interventions comparing pneumatic retinopexy with scleral buckle procedures for uncomplicated rhegmatogenous retinal detachment.
|Randomized controlled trials|
|Tornambe et al. (1991) (USA)|
|RCT||PR (n = 93) SB (n = 86) [N = 179]||Single break no larger than 1 clock hour or a group of small breaks within 1 clock hour located in the superior 8 clock hours of the fundus, with macula-on or off||Single for VA||24||Single-operation reattachment at 6 months (PR: 73% vs. SB: 82%; p > 0.05) Overall reattachment with reoperations (PR: 99% vs. SB: 98%; p > 0.05) 24-month VA ≥ 20/50 in eyes with macular detachment for ≤ 14 days (PR: 89% vs. SB 67%; p < 0.05)|
|Mulvihill et al. (1996) (Ireland)|
|RCT||PR (n = 10) SB (n = 10) [N = 20]||1. Single retinal break or group of breaks ≤ 1 clock hour of retina in size 2. Break or breaks located within the superior 8 clock hours of the retina||None||Mean = 16.7 (PR) (5–27 mo) Mean = 16.0 (SB) (8–23 mo)||Single-operation reattachment (PR: 70% vs. SB: 80%) p = 0.5 Overall reattachment with reoperations (PR: 90% vs. SB: 100%) p = 0.5|
|Non-randomized controlled trials|
|McAllister et al. (1988) (USA)|
|Non-randomized controlled study||PR (n = 56) SB (n = 78) Lincoff balloon (n = 28) [N = 162]||Uncomplicated RRD||None||> 6||Overall reattachment rates in at least 6 months (PR: 71% vs. SB: 96% vs. Lincoff balloon: 64%) (p < 0.001)*|
|Kreissig et al. (1989) (Germany)|
|Non-randomized study (comparative analysis with previous reports on PR)||Temporary balloon buckle (n = 500) PR (n = 500) [N = 1000]||Single break or group of breaks close together that do not subtend > 6–8 mm for about 1 clock hour at the equator (majority uncomplicated RRD)||None||Balloon buckle: Mean = 25 (6 months to 7.5 years)||Single-operation reattachment (PR: 79.6% vs. SB: 91%) Overall reattachment with reoperations (PR: 99% vs. SB: 99%)|
|Han et al. (1998) (USA)|
|Non-randomized controlled retrospective study||PR (n = 50) SB (n = 50) [N = 100]||Primary RRD with retinal breaks limited to the superior 8 clock hours of the fundus and all breaks within 1 clock hour of the fundus||None||≥ 6||Single-operation reattachment (PR: 62% vs. SB ± postoperative laser or 84%; p = 0.01) Overall reattachment with reoperations (PR: 98% vs. SB: 98%; p > 0.05) Final VA not significantly different|
The final reattachment rates for PR and SB in the two RCTs (Tornambe et al. 1991[n = 179]; Mulvihill et al. 1996[n = 20]) were not significantly different. In the Tornambe et al. (1991) 2-year follow-up study, single-operation reattachment rates were 73% for PR eyes and 82% for SB eyes after 6 months, and multiple-operation reattachment rates at 2 years were 99% for PR eyes and 98% for SB; however, 89% of PR eyes had 24-month VA ≥ 20/50 compared with 67% of SB eyes with macular detachment of 14 days or less. In the Mulvihill et al. (1996) study of 20 patients followed for 5–27 months, 70% of the PR group were reattached after a single operation at 6 months compared with 80% of the SB group. The overall reoperation reattachment rates were not significantly different for the PR (90%) and SB groups (100%). The small sample size of the Mulvihill et al. (1996) study precludes any definitive conclusions of the effect of PR on final VA and unfortunately the time-point of the final follow-up visit was not clearly predefined.
Similar results were found in the three non-randomized controlled trials: the final retinal reattachment rates for PR were either similar to or lower than those for SB, and there was no significantly increased risk of late redetachment after PR compared with SB (McAllister et al. 1988; Kreissig et al. 1989; Han et al. 1998). The study by Han et al. (1998) combined data from many surgeons over many years, and the majority of the SB cases were macula-on, thus biasing the SB VA outcomes in favour of SB. The VAs at specified follow-up times were not reported and only final VA was presented. Complications observed in PR eyes in the five trials included new or missed retinal breaks. In the study by Tornambe et al. (1991), the rate of missed or new retinal breaks was higher, although only to a point of borderline significance, in PR eyes (23%) compared with SB eyes (13%) (p = 0.05); while in the Han et al. (1998) study this rate was significantly higher (PR 10% versus SB 1%).
Given that only two RCTs and three controlled studies have examined the efficacy of PR, the current evidence is weak. The most recent RCT (Mulvihill et al. 1996) had a sample size of only 20 and the other trial was conducted 15 years ago (Tornambe et al. 1991). Thus, there is a need for future RCTs evaluating the short- and long-term efficacy of PR using appropriate outcomes of interest including initial reattachment rates, reattachment rates at predefined follow-up intervals and postoperative VA.
Subretinal fluid management options during scleral buckling
External drainage of SRF can be achieved with a variety of techniques. Unfortunately, there is a lack of uniformity and consistency in the terminology used with many individual variations. For the purpose of this review we have categorized these techniques as conventional subretinal drainage (CSD), hypodermic needle drainage (HND), suture needle drainage (SND) and argon laser drainage (ALD). Conventional subretinal drainage involves making a sclerotomy followed by perforation of the exposed choroid with a diathermy electrode. Hypodermic needle drainage involves drainage of the SRF with a hypodermic needle, using a 25-gauge, 15-mm disposable needle attached to a 1-cc syringe with the plunger removed. The needle is placed against the sclera with the bevel facing away from the scleral surface. While observing the fundus with an indirect ophthalmoscope, the SRF is drained by obliquely penetrating the sclera and entering the subretinal space. A modified HND uses a 26-gauge, 12.5-mm disposable needle, which is introduced perpendicularly through the sclera to a maximum length of 2 mm till a bead of fluid is seen in the hub of the needle. A one-stage SND is performed with a 5/0 ethibond spatulated needle, held in the jaws of Castro−Viejo forceps 2–3 mm from its tip; the sclera and choroid are punctured in one smooth motion to drain the SRF. A two-stage SND is similar to a CSD except that a curved spatulated or tapered suture needle is used to puncture the exposed choroid. Argon laser drainage uses an endolaser probe held adjacent to the exposed choroid following a sclerotomy to puncture the choroid by photocoagulation.
There are six RCTs on the various management options of subretinal fluid during SB among patients with uncomplicated RRD (Hilton 1981; Burton et al. 1993; Das & Jalali 1994; Ibanez et al. 1994; Aylward et al. 1995; Azad et al. 2004). Hilton (1981) evaluated SB with and without drainage in the USA; Burton et al. (1993) compared HND with two-stage SND during SB in Australia, while Azad et al. (2004) compared CSD with modified HND in India. Argon laser drainage was compared with two-stage SND in studies conducted in India and the USA and with one-stage SND in the UK (Das & Jalali 1994; Ibanez et al. 1994; Aylward et al. 1995). The sample sizes ranged from 50 to 175. None of the studies were masked except for that by Hilton (1981), where double-masking to VA outcomes was performed. The follow-up ranged from 24 hours to 6 months.
In the study by Hilton (1981), 120 patients with RRD who had no contraindications to drainage were randomized to SB with two-stage SND (average duration of operation = 65 mins) and SB with no drainage (average duration of operation = 67 mins). The initial reattachment rate after a single operation was 87% in the drainage group and 83% in the non-drainage group, and overall reoperation reattachment rates were identical (97%) in both groups. There was no significant difference in postoperative VA between the two groups. The complication rate was significantly higher in the drainage group (p = 0.005) and complications included peripheral subretinal haemorrhage and retinal break at the drainage site. Another RCT was conducted amongst RRD patients in Australia: 45 patients were randomized to HND and 55 to two-stage SND (Burton et al. 1993). The single-operation reattachment rates were 84.5% in the HND group and 90.9% in the two-stage SND group (p = 0.46). In India, 40 patients were randomized to CSD and 40 patients to modified HND; complication rates were 32.5% in the CSD group and 15% in the modified HND group (p = 0.76), but 100% drainage was achieved in both groups (Azad et al. 2004).
Ibanez et al. (1994), Das & Jalali (1994), and Aylward et al. (1995) have all evaluated the efficacy of the argon laser endoprobe for ALD. In the study by Das & Jalali (1994) (n = 50), most of the complications, including subretinal haemorrhages, retinal incarceration and incomplete drainage, occurred in the two-stage SND group. Likewise, the Ibanez et al. (1994) study (n = 175), carried out in the USA, also showed more complications in the two-stage SND group compared with the ALD group (intraoperative complications: 9.8% versus 7.6%; postoperative complications: 6.2% versus 5.4%), although these differences were not statistically significant. In the study by Aylward et al. (1995) (n = 93), ALD patients had significantly higher drainage success rates (97.9% versus 84.8%) and lower haemorrhage complication rates compared with one-stage SND patients, but neither of these differences were statistically significant.
Retinopexy options for scleral buckling
Four RCTs evaluated the method of retinopexy in patients with RRD managed with SB: Avitabile et al. (2004) (n = 703, SB with transpupillary frequency-doubled Nd:YAG laser, compared with SB with cryopexy); Figueroa et al. (2002) (n = 60, SB with transscleral cryotherapy, compared with SB without retinopexy); Steel et al. (2000) (n = 120, SB with transscleral cryotherapy, compared with SB with diopexy), and Veckeneer et al. (2001) (n = 46, SB with intraoperative transscleral cryotherapy, compared with SB with postoperative slit-lamp delivery of 532-nm laser). The 6-month reattachment rates in the Avitabile et al. (2004) trial (6 months follow-up), which randomized 730 patients to SB with laser retinopexy (83%) or SB with cryopexy (83.1%), were similar. In the Figueroa et al. (2002) study (22–33 months follow-up), the overall reattachment rates were similar: 87% in the SB with cryotherapy group and 90% for SB without retinopexy. In the Steel et al. (2000) study (median follow-up 24 months in the cryopexy group and 21 months in the diode laser group), the single-operation reattachment rates in the cryopexy and diode laser groups were not significantly different (92% versus 83%). In the Veckeneer et al. (2001) study (10 weeks follow-up), the postoperative flare values were significantly higher in the cryotherapy (mean = 10.8) compared with the laser group (mean = 6.6) (p < 0.001).
Another RCT carried out in the UK evaluated whether cryotherapy before drainage of SRF was more effective compared with cryotherapy after SRF drainage in 80 patients (Pearce et al. 1997). The single and final operation reattachment rates for cryotherapy before SRF drainage compared with cryotherapy after SRF drainage were 86.9% versus 82.4% (p = 0.38), and 100% versus 97.1%, respectively.
Several RCTs have been conducted to refine various SB techniques. However, the number of studies evaluating each technique is small and often clinically relevant conclusions cannot be drawn. There is scope for the conduct of large-scale, masked RCTs evaluating the different SB methods.
We summarize here the results of four RCTs and five non-RCTs evaluating primary vitrectomy for uncomplicated RRD. Combined vitrectomy and SB (mean follow-up 100 days) were compared with SB alone (mean follow-up 96 days) in an RCT carried out in 44 eyes of Indian patients with uncomplicated RRD and preoperatively unseen retinal breaks (Tewari et al. 2003). The single-operation reattachment rates were similar in the combined surgery (80%) and the SB group (70%) (p = 0.72) and VA improvements did not differ between the groups (p = 0.4).
Pseudophakic RRDs were evaluated in three RCTs. Fifty patients with pseudophakic primary RRD were randomized to SB (n = 25) or vitrectomy (n = 25) and followed for 6 months. The final reattachment rates were 100% in both groups, but the vitrectomy group had a VA that was at least two lines better than that of the SB group at 6 months (p = 0.034) (Sharma et al. 2005). A multicentre clinical trial of 225 patients with pseudophakic or aphakic RRD randomized to SB or primary vitrectomy without any buckle found no statistically significant differences in single-operation retinal reattachment rates and postoperative VA at 1, 2, 4 and 6 months (Ahmadieh et al. 2005). In another clinical trial of 150 patients with pseudophakic primary RRD and proliferative vitreoretinopathy stage B or less randomized to SB (75 eyes) or vitrectomy (75 eyes), reduced operating time, more accurate diagnosis of retinal breaks, a higher single-operation reattachment rate, and less postoperative axial length change were reported for the vitrectomy group. However, the final reattachment rates (94.7% in the SB group versus 98.7% in the vitrectomy group) and mean logMAR best-corrected VA at 1 year were similar (0.40 versus 0.33; p = 0.26) (Brazitikos et al. 2005).
A non-randomized controlled trial of 102 Japanese patients with primary macula-off RRD compared vitrectomy with SB (minimum follow-up 24 months) (Oshima et al. 2000). The single-operation reattachment rates were similar in the vitrectomy (91%) and SB (91%) groups (p = 0.72). The improvement in logMAR VA was greater (mean = 0.95) in the vitrectomy group compared with the SB group (mean 0.54, p = 0.03). Patients with poorer preoperative VA, ocular hypotony or macular detachment of more than 7 days reported better visual recovery after primary vitrectomy compared with SB. In another non-randomized controlled study of 225 Japanese patients with superior RD caused by flap tears (with follow-up of at least 6 months), the initial and final reattachment rates did not vary between groups (92% versus 92%; 100% versus 100%) (Miki et al. 2001). In a non-randomized, controlled trial, outcomes for 24 patients treated with vitrectomy, cryotherapy and fluid−gas exchange were compared with those for nine patients treated with episcleral buckle and cryotherapy (Wolfensberger 2004). Optical coherence tomography examinations at 1 month after surgery showed subfoveal fluid in 0% of patients in the vitrectomy group and 67% in the episcleral buckle group. A further non-randomized, controlled, prospective trial of vitrectomy alone and vitrectomy plus encircling SB for pseudophakic RRD was conducted in 71 eyes of 68 patients. After surgery, the initial reattachment rates were 97.8% for vitrectomy alone and 92.3% for vitrectomy with SB, while improvement of three or more lines in VA were achieved in 60% of patients with vitrectomy alone, and 69% of patients with vitrectomy with SB (Stangos et al. 2004). Moreover, an additional buckle may prolong the operating time, contribute to postoperative myopic shift and predispose the eye to other buckle-related complications. A retrospective case note review of 86 patients with inferior break RDs compared 41 patients treated with vitrectomy and gas, with 45 patients treated with vitrectomy, gas and scleral buckle. The difference between the two groups in final reattachment rates was not statistically significant: 95% (39) in the vitrectomy group and 93% (42) in the vitrectomy and SB group (p = 1.0) (Wickham et al. 2004). The interpretation of trials evaluating vitrectomy are limited by study design issues, such as lack of randomization in the trials by Oshima et al. (2000), Miki et al. (2001), Stangos et al. (2004), Wolfensberger (2004) and Wickham et al. (2004), and small sample sizes (n = 44 eyes) in the trials by Tewari et al. (2003) and Wolfensberger (2004).
The most important advantages of primary vitrectomy over SB have been itemized as its direct approach to the release of vitreous traction, the ability to deal with media opacities, improved detection of retinal holes and controlled internal drainage of SRF. In addition, the absence of risk of cataract formation itself is another reason for considering vitrectomy for pseudophakic and aphakic RRDs. However, one of the greatest disadvantages of primary vitrectomy concerns the possibility of causing new retinal breaks and lens damage. Others have suggested that the advantages may be offset by the higher complication rate of this more invasive intraocular procedure and its relatively higher costs.
Current evidence does not suggest that primary vitrectomy holds any significant benefit over SB in the management of uncomplicated RRD with respect to final retinal reattachment rates and visual outcomes. However, faster foveal reattachment and better final VAs are potential advantages of vitrectomy. We await the results of a large RCT underway in six European countries, the Scleral Buckling versus Primary Vitrectomy in Rhegmatogenous Retinal Detachments Study (SPR Study), which is currently being conducted in patients with RRD that is not complicated by proliferative vitreoretinopathy (Heimann et al. 2001).
Rhegmatogenous retinal detachment is a relatively infrequent, acute, heterogeneous condition compared to other ocular conditions such as glaucoma, in which very large clinical trials have been possible, albeit at great cost. In choosing the optimal treatment strategy for RRD, several clinical factors such as the location, size and number of retinal breaks may influence technique and success rate. In addition, research into the treatment of RRD is complicated by large variations in the multitude of surgical techniques and the large influence of surgeon-dependent factors (such as skills and experience) that have a bearing on outcome. Differences in surgical technique, training and experience between surgeons may be more important than clinical factors. The clinical diagnostic criteria for study inclusion and definitions of single-operation and overall reattachment rates are not uniform across trials. Moreover, VA must be reported at specific time-points to allow for meaningful comparison between studies, for example at 6 months, 1 year and 2 years after surgery. Visual recovery in RRD surgery is as important as the reattachment of the retina. Limitations to the results from many previous studies include the non-availability of control groups, short duration of follow-up and lack of randomization. Large, masked RCTs are required to control for such variations, although they have attendant difficulties of recruitment and cost. There are great opportunities for more research: large, well conducted, double-masked RCTs should be conducted to evaluate PR, SB and vitrectomy. Clinically important outcomes that need to be determined include reattachment rates and VA at specific follow-up times.
In conclusion, PR may be performed for uncomplicated RRD with retinal breaks within one clock hour area of the superior two-thirds of the retina. In patients with uncomplicated RRD with inferior breaks or patients who cannot tolerate the PR procedure, SB is the preferred procedure. The role of vitrectomy for uncomplicated RRD, a surgical procedure with greater operator control, is evolving and may be indicated especially when there is poor visualization of the retina. Recent results are promising and better final VAs may be attained. As there is currently no single preferred mode of treatment, the surgeon should choose the technique in which he or she is most experienced in the management of RRD.