The aim of glaucoma surgery is to lower the intraocular pressure in order to reduce the risk of further glaucomatous progression, particularly in cases refractory to topical therapy. Although effective in reducing intraocular pressure, these procedures are not without complications, with endophthalmitis being one of the most serious. A PubMed review of the literature was performed for trabeculectomy, glaucoma drainage device procedures (Ahmed, Baerveldt and Molteno implants) and non-penetrating glaucoma surgery (deep sclerectomy and viscocanalostomy) for reports of postoperative infection, including blebitis and endophthalmitis. The literature on infections relating to non-penetrating glaucoma surgery is sparse compared with penetrating surgery, but this may be a reflection of the relatively shorter follow-up duration and comparatively smaller body of data available on non-penetrating procedures. Overall, there is not enough evidence, in terms of well-constructed randomised clinical trials with sufficiently large sample sizes and long follow-up durations, to be able to make informed comparisons of the risk of postoperative endophthalmitis and infection between the various glaucoma operations. This review article summarises the incidences of endophthalmitis from the literature and discusses the major risk factors for postoperative infection.
- Non-penetrating glaucoma surgery
- deep sclerectomy
- glaucoma drainage device
- intraocular pressure
- treatment surgery
Statistics from Altmetric.com
- Non-penetrating glaucoma surgery
- deep sclerectomy
- glaucoma drainage device
- intraocular pressure
- treatment surgery
The aim of glaucoma surgery is to reduce the intraocular pressure (IOP) in order to reduce the risk of progression of glaucomatous optic neuropathy, especially in cases that are refractory to treatment with medical therapy. Since it was described in 1968 by Cairns,1 trabeculectomy has undergone multiple modifications, and is now generally considered to be the gold standard for glaucoma surgery. Glaucoma drainage devices (GDD) are comparatively more recent, but are gaining in popularity particularly in cases where the conjunctiva is scarred and not in a healthy state. Currently, the IOP-lowering outcomes and complication rates for both techniques are being evaluated in the ongoing Trabeculectomy versus Tubes (TVT) prospective randomised trial.2 That these penetrating procedures are effective in lowering IOP is not in question. The problem is the associated complications that may occur, such as hypotony and flat anterior chamber. In 1984, non-penetrating glaucoma surgery (NPGS) was proposed as an effective method of lowering IOP, but with less risk of complications compared with trabeculectomy.3
The techniques for NPGS have since evolved, and at present the two most common NPGS procedures are deep sclerectomy and viscocanalostomy. Several small prospective randomised trials have been conducted to compare the efficacy and side effect profile of NPGS with that of trabeculectomy.4–12 Although these trials were mostly of limited follow-up, they did generally demonstrate that postoperative IOP reduction was equivalent or slightly worse compared with that of trabeculectomy, but with less frequency of complications. Indeed, a major review of NPGS found NPGS to be significantly safer than trabeculectomy, while retaining good IOP control.13 While most publications have generally focused on complications such as hypotony, wound leak and cataract, the issue of postoperative infection seems to have been largely overlooked. The review aims to address the issue of the incidence and risk factors for postoperative infection following both penetrating and non-penetrating glaucoma surgery.
A PubMed search was carried out for the three major glaucoma surgical procedures currently in use: trabeculectomy, GDD implantation (Ahmed, Baerveldt and Molteno aqueous shunts) and NPGS (deep sclerectomy and viscocanalostomy). We included various search terms in the keywords that may indicate infections, such as conjunctivitis, keratitis, scleritis, blebitis, bleb-related infection, endophthalmitis, orbital cellulitis, and also mitomycin-c (MMC) and 5-fluorouracil (5-FU).
There were 64 published articles on infections after glaucoma surgery. Forty-eight papers were pertaining to infections following trabeculectomy, with 18 isolated case reports of various infections, 23 case series of bleb-related infection, and five reviews of bleb-related infection, including endophthalmitis. For GDD procedures, there were 15 publications, 12 of which were isolated case reports, two were case series of postoperative endophthalmitis, and one was a literature review. For NPGS, there have only been three published case reports of postoperative infection, including one case of blebitis.14–16 There have not been any publications describing postoperative infective endophthalmitis following NPGS. The findings from the literature review are outlined in table 1.
Glaucoma procedures with filtering blebs are at risk of developing infections at the site of the filtering bleb. Bleb-related infections are comprised of a spectrum of conditions, which range from blebitis to bleb-related endophthalmitis. Blebitis is defined as presumed infection of the filtering bleb when there is mucopurulent infiltrate in the bleb, and this may be associated with mild to moderate anterior chamber activity. The absence of vitritis is the key differentiating factor from endophthalmitis.17–20 Bleb-related endophthalmitis occurs when there is vitreous involvement, often presenting with hypopyon and severe visual loss.17–20
It is important to bear in mind that bleb-related endophthalmitis can occur at any time following surgery. Some have classified bleb-related endophthalmitis to early onset or late onset, with 4 weeks after surgery being one of the arbitrary cut-off points.19 21 22 Similarly, endophthalmitis from GDD may also develop at any time after surgery.23–29 Endophthalmitis from GDD exposure has been described as late as 6 years after surgery.30
Outcomes of infection
The visual outcomes following prompt and successful treatment of blebitis are generally good, with all patients maintaining vision 20/25 or better in one study.31 If not diagnosed and managed appropriately, blebitis may progress to become endophthalmitis and result in blindness. Late bleb-related endophthalmitis, in particular, has very poor visual prognosis despite aggressive intervention with topical, intravitreal and systemic antibiotics and vitrectomy.17 19 21 31 32 One series found the final vision to be worse than 20/200 in 94%, while others have reported the final vision to be less than 20/400 in up to 74% of cases.19 21 31–33 As for GDD, the visual prognosis following endophthalmitis is also similarly poor; with none in a series of Ahmed valves,34 and only one of four (25%) eyes in a series of Baerveldt devices,35 achieving a final vision better than 20/200.
Incidence of infection
When considering endophthalmitis following any intraocular surgery, the overall incidence is reported to be about 0.05–0.19%.36–39 In general, large hospital-based series tend to demonstrate a higher incidence of endophthalmitis following glaucoma procedures compared with the overall incidence after intraocular surgery: 0.12% versus 0.09% in one series, and 0.20% versus 0.05% in another.36 37
A recent national longitudinal cohort study of postoperative adverse events after glaucoma surgery utilising the US Medicare administrative database found the rate of endophthalmitis to be 0.3–0.7% for trabeculectomy.40 Previous national studies have reported post-trabeculectomy endophthalmitis to occur in 0.2% over a 1-year period to 1.2% over a 6-year period.41 42 The incidence of postoperative infection and other complications is likely to be reduced with the advances in surgical and antimetabolite application techniques, such as the safe surgery system for trabeculectomy, although there is no long-term or large sample data to support this at present. For GDD, the longitudinal Medicare cohort found an endophthalmitis rate of 2.0%,40 while in hospital-based studies, this ranged from 0.3% to 5.0% (approximate mean 1.9%). There appears to be no significant differences in the reported rates of endophthalmitis for the various GDD types.34 Table 2 summarises the incidence of endophthalmitis of glaucoma filtering procedures from the larger published studies.
For NPGS on the other hand, no cases of endophthalmitis have been reported so far. There have not been any published case series looking specifically at postoperative bleb-related infections following NPGS. Only three isolated case reports of infections have been published: fungal keratitis, bacterial keratitis and blebitis.14–16 The case of fungal keratitis from Aspergillus occurred in a 63-year-old patient 1 week after viscocanalostomy.14 A separate case of Staphylococcus aureus keratitis was reported in a 68-year-old woman 2 weeks after deep sclerectomy without antimetabolites.15 The single case report of blebitis so far was in a 59-year-old man 13 days after deep sclerectomy without antimetabolites.16 There was no vitreous involvement, and the patient was successfully treated with intensive topical medication. Conjunctival culture was positive for Staphylococcus aureus. After 6 months, the eye achieved a vision of 6/12 and IOP of 15 mm Hg without topical therapy.
Even in the larger case series of NPGS with longer follow-up, the occurrences of infections could be considered as rare. In more than 2000 cases of NPGS, Roy and Mermoud only noted one case of early blebitis, and this was in a 75-year-old patient receiving treatment for pulmonary tuberculosis.68 In their series of 171 eyes undergoing deep sclerectomy with collagen implant (58 with 5-FU and 53 with MMC), Detry-Morel and Detry observed two cases of blebitis (1.2%) over a mean follow-up of 39.6 months.69 In a series of 64 deep sclerectomy eyes augmented with MMC followed for 24 months, blebitis occurred in one eye (1.6%).70 Blebitis may also develop after goniopuncture, with two cases (1.2%) in Anand and Pilling's series of 173 eyes.71 It is worth noting that none of the large published series of NPGS of at least 100 eyes and over 4 years of follow-up (cumulative total of 677 eyes and cumulative mean follow-up of 71.8 months) have reported any cases of bleb-related infection.72–74
There have been randomised trials comparing NPGS with trabeculectomy, but not with GDD.4–12 While the results of these studies can be conflicting, in general, they demonstrate that the IOP-lowering effect is either the same or better with both augmented and non-augmented trabeculectomy compared with NPGS. The trade-off for better IOP control was more intensive postoperative management and complications, although not from infection. However, with their relatively small sample size and short follow-up duration, these trials were not sufficiently powered to be able to demonstrate any meaningful difference in the rates of complications as rare as endophthalmitis. While the literature may seem to point towards less infection following NPGS compared with penetrating surgery, this apparent finding may be due to the relatively shorter follow-up duration and comparatively smaller body of NPGS literature.
Risk factors for infection
In the literature, various risk factors have been identified for the development of bleb-related infection and endophthalmitis following penetrating glaucoma surgery. Some of these risk factors for bleb-related infection may also apply for NPGS, particularly the use of antimetabolites.
Intraoperative 5-FC or MMC are recognised as important risk factors for bleb-related infection.22 51 75 A Cochrane review concluded that intraoperative MMC reduced the risk of surgical failure, but complications such as infection could not be evaluated due to inadequate power.76 However, drawing from institution-based case series, the incidence of bleb-related endophthalmitis in non-augmented trabeculectomy ranges from 0.2% to 1.5%.45–50 With intraoperative 5-FU and MMC, this incidence increases to 1.0% to 5.7% and 0.3% to 4.9%, respectively.2 33 48 49 51–58 60–62 The risk ratios for bleb-related infection were calculated to be 1.31 with 5-FU and 2.48 with MMC. The conjunctiva of blebs after augmented trabeculectomy shows epithelial irregularity and cell loss, as well as breaks in the basement membrane.77 This compromises the barrier function of the bleb and predisposes to late bleb leaks, which in itself is also thought to be an important predisposing factor for bleb-related infection.18 22 78 In their case–control study, Soltau and colleagues found that bleb leak was 26 times more likely to be detected in eyes with bleb-related infection than eyes without.18 In addition, the RR for subsequent bleb-related infection following a history of bleb leak was calculated to be 3.7.22 Antimetabolites may also affect the conjunctival innate immunity, thus further increasing the susceptibility to infection.75 It is highly likely that these effects on the filtering bleb will also be present for augmented NPGS. It has already been shown that intraoperative MMC allows better IOP control, but causes an increased incidence of avascularity and transconjunctival oozing.70 79 From institutional-based studies, the highest rate of blebitis (not endophthalmitis) following NPGS was 1.6%, and this was in a series of 64 eyes undergoing deep sclerectomy with collagen implant and MMC.70 However, most of the other case series of NPGS that included intraoperative use of 5-FU and MMC have not observed any cases of blebitis during follow-up.73 80–82 It would seem that antimetabolite-related complications are less of a problem following GDD procedures. Adjunctive antimetabolite use during Ahmed and Molteno valve insertion has been investigated in two randomised trials; no significant differences in IOP or complication outcomes were demonstrated.83 84 Other case series describing MMC augmentation during the various GDD procedures have not reported any increase in postoperative complications including exposure or infection.85–89
Inferior bleb location
Another point to consider is the location of the filtering bleb. Inferiorly located blebs are at a significantly higher risk of developing infection compared with superior blebs.33 53 55 Greenfield and co-authors calculated the incidence of bleb-related infection to be 7.8% per patient year for inferior blebs, while this rate was only 1.3% per patient year for superior blebs.33 Other series have noted the infection rate of inferior blebs to be up to 9.4%.53 55 Various reasons for this have been proposed, including repeated mechanical irritation by the lower lid and exposure to the endogenous bacterial flora in the tear lake.33 55 The literature for infections and the positioning of GDD is mixed. Inferiorly placed Ahmed valves have been shown to carry a higher risk of complications including endophthalmitis, although this did not reach statistical significance (0.0% superior versus 2.1% inferior).28 In contrast, the 0.5% rate of endophthalmitis in a series describing the inferonasal insertion of Baeveldt implants was well within the 0.3–5.0% range for GDD endophthalmitis.29 The issue of inferior positioning is perhaps not as relevant for NPGS as it is usually performed superiorly unless in exceptional situations.
For GDDs, a major risk factor for endophthalmitis is tube exposure following erosion of the overlying conjunctiva.34 35 Al-Torbak and associates observed that in the nine eyes that developed endophthalmitis after Ahmed valve implantation, six (66.7%) had exposure of the seton.34 This association was subsequently confirmed to be statistically significant from multiple regression analysis. A case series of late-onset endophthalmitis associated with Baerveldt devices reported that all four cases (100%) had exposed setons.35 It has been postulated that the exposed tube may act as a direct channel for the intraocular passage of conjunctival flora from the ocular surface.35 While perhaps not so relevant for trabeculectomy or NPGS, one can still draw similarities to the bleb thinning and leakage that does occur in both trabeculectomy and NPGS.
Younger age during GDD implantation also seems to confer an increased risk of endophthalmitis; this was found to be statistically significant (4.4% in patients <18 years of age versus 0.9% in those above 18 years; p<0.05%).34 Other case series of GDD in the paediatric population have found the incidence of endophthalmitis to range from 2.9% to 5.0%.63 64 This is likely to be partly related to the higher rate of conjunctival erosion and implant exposure in children (up to 13.3%).64 90 Younger age has also been implicated as a risk factor for endophthalmitis after trabeculectomy,22 with the incidence of late bleb-related endophthalmitis as high as 8.3% in the paediatric population.56 62 91 92 As for NPGS in the paediatric population, the literature is somewhat more limited, and no case of bleb-related endophthalmitis has been reported so far.93 94
Other associations with bleb-related infections include blepharitis,33 diabetes,75 limbus-based conjunctival flaps,95 silk conjunctival sutures,22 persistent use of topical antibiotics past the immediate postoperative phase,22 early postoperative complications such as flat anterior chamber and suprachoroidal haemorrhage,22 and bleb manipulation from revision or needling.22 33 Interestingly, bleb-related infection has been reported to occur less frequently when trabeculectomy was combined with cataract surgery.22
Trabeculo-Descemet membrane: barrier to endophthalmitis?
The trabeculo-Descemet membrane (TDM) comprises the anterior trabecular meshwork and peripheral Descemet membrane, and is the barrier that offers resistance to aqueous humour outflow following NPGS.96 So far, there have not been any microstructural studies evaluating the ‘permeability’ of the TDM to pathogens, nor have there been any specific microbiological research to assess the role of the TDM in the prevention of intraocular infections. It is likely that the TDM provides a physical barrier of some form, such as that conferred by the posterior lens capsule.97 It is also possible that there may also be other immunological or biochemical factors that further provide a protective effect against invasion by microorganisms. Despite the presence of the TDM in blebitis associated with NPGS, there may still be sufficient inflammation in and around the bleb to also induce activity in the anterior chamber although there is no actual intraocular microbial contamination. Nevertheless, it is likely that there is still a risk of intraocular infectious spread, and therefore blebitis should always be managed appropriately in a timely fashion.
Although publications on endophthalmitis after both trabeculectomy and GDD are easily found, the literature on infections associated with NPGS is sparse. So far, no cases of endophthalmitis have been reported following NPGS, even when augmented with intraoperative antimetabolites, and this is likely to be related to the barrier provided by the TDM. However, there is currently no evidence in the form of adequately powered randomised trials with sufficiently long follow-up periods to conclusively demonstrate that NPGS is associated with significantly less risk of endophthalmitis and infection when compared to trabeculectomy and GDD implantation. Regardless of the surgical procedure, it is of utmost importance to educate the patient about the symptoms of bleb-related infection so as to achieve the best possible outcome with early intervention.
Funding Cheltenham Ophthalmology Scholarship (GSA).
Competing interests None.
Provenance and peer review Not commissioned; externally peer reviewed.
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.