Introduction

Globally, cataract is the leading cause of preventable blindness [1]. Modern phacoemulsification techniques yield superior visual outcomes with reduced complication rates. Despite this, pseudophakic cystoid macular oedema (PCMO) remains the most common sight-threatening complication following cataract surgery [2].

PCMO is characterised by retinal swelling due to a disruption of the blood-retina barrier following a cataract operation. The majority of PCMO occurs within 90 days of the cataract operation and peaks at 4–6 weeks [3, 4]. PCMO can be further subcategorised as angiographic (as seen on FFA) or clinically significant (clinical evidence of PCMO on examination or OCT, with an associated decrease in visual acuity).

There is currently no consensus on a prophylactic protocol for the prevention of PCMO, however topical NSAIDs are most commonly used. Several systemic reviews have highlighted that there is inadequate evidence for the use of prophylactic NSAIDs, in that there is low certainty of the risk reduction obtained and that the size of this reduction is likely exaggerated [5, 6].

Routine prescription of prophylactic NSAID drops adds a high societal cost to cataract surgery (£467 000 per case of BCVA of 20/200 or worse prevented) [7]. The use of NSAIDs adds a further eye drop to the post-operative regime and often causes pain on administration, hindering patient adherence [8]. There are also infrequent but significant side effects of impaired corneal wound healing, corneal melt, and perforation in patients who have had otherwise uncomplicated cataract surgeries [9,10,11,12]. Recently, a group at Moorfields Eye Hospital observed a pattern of increased atypical keratopathy following cataract surgery when a commonly used combination antibiotic and steroid eye drop, neomycin/polymyxin B sulphate/dexamethasone (Maxitrol), was used in conjunction with ketorolac (Acular). The authors propose that the combined use of a topical NSAID and agents such as neomycin and benzalkonium further compromises the corneal epithelium and therefore should be avoided [13]. Subconjunctival steroids (SCS) may have the potential to provide a cheaper, drop-free alternative to NSAID prophylaxis.

The use of a combined subconjunctival injection of antibiotic and steroid was commonplace in the 1950s however has now become omitted with the introduction of intracameral antibiotics [14, 15].

To date, there are no studies which have been adequately powered to assess a treatment effect for the use of subconjunctival steroids and PCMO prophylaxis [15,16,17,18]. This is due to the relatively low incidence of clinically significant PCMO, which the most recent literature reports as approximately 1.5% [2, 4, 19]. When studying an outcome of such a low incidence, an appropriately large sample size must be obtained to avoid a type II error. The required sample size to detect a 50% decrease in PCMO from the reported rate of 1.5% was calculated to be 3103 per group (alpha 0.05, power 80%).

The question of whether simply a higher dose of steroids, administered intraoperatively to arrest the inflammatory cascade, would achieve the same or better results as topical NSAID drops has not been answered [20]. This study investigates the effect of subconjunctival steroids (betamethasone or dexamethasone sodium phosphate) on rates of clinically significant PCMO.

Methods

All eyes undergoing cataract surgery at an NHS University Hospital Trust between December 2007 and November 2017 were identified through an electronic medical record (EMR) (Medisoft OphthalmologyTM, Medisoft Ltd, Leeds, UK).

Structured clinical data were anonymised and extracted for each eye undergoing cataract surgery including demographic data (age and sex), ophthalmic data (laterality and history of known risk factors), and surgical data (surgeon grade, surgical complications, and administration of SCS). ‘High-risk cases’ were defined as eyes with a co-pathology that is known to increase the risk of PCMO, including the presence of diabetic retinopathy, previous pars plana vitrectomy (PPV), retinal vein occlusion (RVO), epiretinal membrane (ERM), full-thickness macular hole (FTMH) or ‘complex surgery’, which was defined as any surgery requiring iris stretching, hooks or pupil expander and those that were complicated by iris trauma or prolapse [19, 21]. The main outcome measure was the diagnosis of clinically significant PCMO within 90 days of surgery.

The inclusion criteria were ages 18 years or older with documented phacoemulsification surgery and intraocular implant. Exclusion criteria were concurrent administration of topical NSAID drops, previous history of uveitis, previous dexamethasone (Ozurdex™) implant, combined surgery (e.g. phaco-vitrectomy, phaco-trabeculectomy), surgery complicated by posterior capsular rupture and/or vitreous loss and patients who have opted out of having their confidential data accessed for the purpose of research. Patients with second eye surgery were included in the analysis. No bilateral cases of PCMO were noted, indicating that inclusion of both eyes from a single subject is unlikely to confound the reported findings.

Either betamethasone sodium phosphate 4 mg/ml (RPH Pharmaceuticals Ltd, Lancashire, United Kingdom) or dexamethasone sodium phosphate 3.3 mg/ml (Hameln Pharmaceuticals Ltd, Gloucester, United Kingdom) was used. When used, a small amount of steroid was injected to produce a visible blister (approximately 0.2–0.5 ml), typically in the inferior fornix.

There were no significant changes to surgical technique or equipment over the data collection period. All cases received a standard course of topical antibiotics and steroid eye drops, regardless of SCS injection. A post-operative visit occurred 3–4 weeks after surgery. Post-operative patients did not routinely undergo OCT or FFA imaging unless the visual acuity did not improve as expected. Once a diagnosis of PCMO was made, it was recorded on the EMR as a post-operative complication.

Statistical analyses were performed using Stata (v16, StataCorp, College Station, TX, USA). Non-parametric tests were used. Mann–Whitney U was used for between-group comparisons and Pearson’s chi-squared analyses were used to compare frequency distributions. Simple and multivariate logistic regression models were used to measure associations. The study was carried out according to the tenets of the Declaration of Helsinki and was approved by the local Research and Development Department.

Results

An initial dataset of 20,066 cataract operations was collected. 2718 were excluded for concurrent NSAID use, 773 were excluded for combined or complicated surgery, 114 were excluded for treated uveitis, 42 were excluded where the patient was less than 18 years of age, and 32 were excluded for previous use of Ozurdex. 16,387 eyes from 11,357 subjects met the criteria for inclusion in the final analysis.

In this study, there were a total of 75 surgeons. Consultants performed 10,753 (66%) of the operations and used SCS in 43% of operations. Non-consultants performed 5634 operations (34%) and used SCS in 42% of operations. Chi-squared tests revealed that the grade of surgeon was not associated with a significant difference in the use of SCS (Table 1).

Table 1 Subject characteristics.
Full size table

The overall incidence of PCMO across the cohort was 0.7%. In the 6885 (42%) eyes that received SCS, betamethasone sodium phosphate was used in 90.3% and dexamethasone sodium phosphate in 9.7%. The odds of PCMO occurring in the group receiving SCS [40/6885 (0.58%)] was significantly lower than the group that did not receive SCS [82/9502 (0.86%)], odds ratio (OR): 0.67; 95% confidence interval (95%CI): 0.46–0.98, p = 0.039), despite a significantly higher proportion of patients in the former group with documented risk factors for PCMO (p < 0.001) (Table 1). This finding indicates an overall relative risk reduction for PCMO of 33% in the treated group (Fig. 1). The number needed to treat to prevent one case of PCMO with SCS was 357. Whilst the greatest relative risk reduction in PCMO with SCS was observed in ‘high-risk’ subjects (45% relative risk reduction) (Fig. 1) (Table 2), this study was not sufficiently powered to detect a statistically significant difference in this subgroup.

Fig. 1: Bar chart depicting the incidence of pseudophakic macular oedema (PCMO) with the use of subconjunctival steroids (SCS) versus no SCS : A reduction of incidence of PCMO is seen with the use of SCS across all groups.
figure 1

The number of subjects within each group are shown within each bar.

Full size image
Table 2 Effect of SCS on PCMO.
Full size table

Measured variables, including younger age, presence of any ‘high-risk’ factor and lack of SCS injection were all statistically significantly associated with the incidence of PCMO on simple logistic regression, while subject sex and surgeon grade was not. All remained independently associated on multivariate logistic analysis (Table 3). The mean age in those with PCMO was significantly less than those without (73.2 ± 11.2 years, compared to 75.5 ± 10.3 years, p = 0.018).

Table 3 Measures of association with development of PCMO.
Full size table

While the incidence of raised IOP (defined as any event of IOP above 21 mmHg) may have been expected to be higher in the SCS group, the post-hoc analysis revealed that raised IOP events were significantly greater in those who did not receive SCS (p = 0.001). This significance persisted even once those with known pre-existing glaucoma and other risk factors were removed (p = 0.030).

Discussion

This study demonstrates that a reduced incidence of PCMO is associated with the intraoperative administration of SCS injection. To the authors’ knowledge, this is the first adequately powered study investigating the effect of SCS on the rate of clinically significant PCMO.

There is currently considerable interest in finding effective prophylaxis for PCMO. Recently, a randomized controlled multicentre clinical trial; the ESCRS PREMED study found that eyes that had received subconjunctival triamcinolone acetonide had lower macular thickness and volume compared to those who had not. However, the incidence of raised IOP was significantly higher in the triamcinolone group [17].

In this study, there was no evidence of an increased risk of raised IOP with the use of SCS. Most reports of raised IOP follow the use of triamcinolone acetonide, which is practically insoluble in water and its crystalline form releases over a long period of time [22,23,24,25,26]. Studies investigating water-soluble SCS injections such as betamethasone sodium phosphate have not found adverse effects on IOP [27, 28]. Conversely, in a prospective randomised controlled trial, IOP lowering was observed with water-soluble SCS injection, as was observed in our study [28]. We hypothesise that both the short-acting nature of the water-soluble steroids used in this study and possible reduced trabecular meshwork inflammation may explain the reduction of raised IOP events seen in patients receiving water-soluble SCS.

Steroid injected into the subconjunctival space can enter the eye via the sclera or by diffusing via the conjunctival sac towards the cornea, achieving rapid direct penetration into the aqueous and vitreous humour [29]. Weijtens et al. showed that subconjunctival injection of dexamethasone disodium phosphate was the most effective method of delivering steroid into human aqueous and vitreous humour, giving the highest concentration of steroid when compared to frequent steroid eye drops, peribulbar injection and systemic steroids [30]. Another study comparing the penetration of topical and subconjunctival corticosteroids into human aqueous humour showed that subconjunctival dexamethasone sodium phosphate 0.4% (Dexona, 0.5 ml of 0.4% solution) achieved over 8000 times the mean peak concentration, up to 12 times quicker, and lasting significantly longer than topical dexamethasone alcohol 0.1% (Maxidex, Alcon) [31]. The half-life of subconjunctival dexamethasone sodium phosphate 0.4% in human aqueous is 24 h [32]. Given these properties, subconjunctival steroids have been investigated for use in “dropless” cataract surgery, with promising results [33, 34].

These properties may make water-soluble SCS an ideal PCMO prophylaxis as it quickly delivers an extremely high dose of steroid in the early postoperative period, with the potential to arrest the initiation of the inflammatory cascade, yet dissipates quickly enough to avoid side effects such as raised IOP. Dexamethasone and betamethasone sodium phosphate are readily available in most ophthalmology theatres and are low-cost at £2.40 and £7.72 per ampoule respectively [35, 36].

A noteworthy drawback of SCS is subconjunctival haemorrhage and pain. Intracameral steroid co-administration with intracameral cefuroxime may be an alternative to consider in future studies. A small number of studies have been done to assess intracameral steroid injection, with or without a drug delivery device, which show efficacy in reducing post-operative anterior chamber inflammation [37]. Chang et al showed that in normal and glaucomatous eyes, intracameral dexamethasone did not cause any significant effect on IOP [38]. However, in another study, intracameral triamcinolone showed a small rise in IOP [39]. The challenge with using intracameral steroids for PCMO prophylaxis is maintaining a therapeutic concentration. Without a drug delivery device, the intraocular half-life of dexamethasone is approximately 3 h and is further limited by rapid aqueous volume turnover (approximately 100 min) [40, 41]. Concerns regarding intracameral steroid delivery devices include additional cost, damage to the corneal endothelium, extrusion of the implant from the anterior chamber, and difficulty removing the implant in a case of endophthalmitis or raised IOP [37]. So far, intracameral steroids have not yet been investigated for use in PCME prophylaxis specifically.

In this study, the number needed to treat is similar to that of NSAID drops (357 vs 320) (ref. [4]). The risk reduction of PCMO with SCS was significant, despite an overrepresentation of ‘high-risk’ cases in the treated group. This indicates that the observed effect may be greater with control for these confounding variables. However, the administration of SCS remained independently associated with a lower rate of PCMO, when adjusted for age and risk factor status, supporting a role in low-risk cases. The low reported incidence of PCMO and limited subgroup sample sizes meant that this study was not sufficiently powered to detect an effect from SCS use in low or high-risk cohorts, but it is likely that, from a cost-benefit viewpoint, SCS may, as with NSAID drops, be most beneficial in high-risk cases [42].

In terms of study limitations, this study was a retrospective, observational study, therefore association can be measured but causality cannot. Some data, which may contribute to the risk of PCMO, were not reliably documented, such as previous use of prostaglandins or rare co-pathologies such as retinitis pigmentosa. The overall incidence of PCMO in this study is lower than that reported in the literature, raising the possibility of underreporting at the point of contemporaneous documentation. A large-scale, prospective study is required to clarify the precise role of intraoperative SCS injection, and avoid the risk of confounding.

Regardless, this large cohort study is appropriately powered to detect the effect of SCS injection on PCMO incidence. Due to the large sample size required to investigate PCMO, many other studies investigating the prophylactic effect of therapeutics use a proxy metric, such as central subfield mean macular thickness on OCT [17]. Studies also tend to exclude high-risk cases, where the benefit of prophylaxis most likely lies [43]. This study, however, directly addresses clinically significant PCMO in a real-world patient cohort, which is of primary concern to physicians and patients alike [44].

Conclusion

The study of prophylactic measures for clinically significant PCMO has so far been limited by small sample sizes. Our large, real-world cohort study, with adequate, a priori power calculation, has demonstrated that intraoperative SCS injection is both a safe and effective preventative treatment for PCMO in routine, uncomplicated cataract surgery. The potential cost-saving of treating high-risk cases with SCS instead of topical NSAIDS may also be considerable.

Further prospective study is warranted to identify the optimal route of steroid delivery with the lowest clinical risk and maximum benefit of PCMO prophylaxis, targeting those most at risk.

Summary

What was known before

  • The current evidence for the use of NSAID drops as PCMO prophylaxis is of low certainty. There is still a need to discover an inexpensive yet effective prophylaxis for PCMO. The study of prophylactic measures for PCMO has so far been limited by small sample sizes due to the low incidence of clinically significant PCMO (1.5%).

What this study adds

  • This is the first adequately powered study to show that SCS significantly reduced the odds of developing PCMO (odds ratio [OR] 0.67; 95% confidence interval [95% CI] 0.46–0.98, p = 0.039, n = 16,387). Multivariate logistic regression showed that SCS was independently associated with reduced PCMO rates, following correction for age and other risk factors. There was no increase in rate of raised IOP events with the use of water-soluble SCS.