Fluocinolone acetonide vitreous insert for chronic diabetic macular oedema: a systematic review with meta-analysis of real-world experience

We conducted a meta-analysis of real-world studies on the 0.19 mg Fluocinolone Acetonide (FAc) intravitreal implant for chronic diabetic macular oedema (DMO), comparing these findings with the Fluocinolone Acetonide for Diabetic Macular Edema (FAME) study. The primary outcome was mean change of best corrected visual acuity (BCVA) at 24 months. Secondary outcomes were 36-month mean BCVA, mean central macular thickness (CMT) change, rates of eyes receiving supplementary intravitreal therapy, cataract surgery, intraocular pressure (IOP)-lowering drops and glaucoma surgery. Mean differences (MDs) with 95% confidence intervals (CIs) were calculated. Nine real-world studies were included. The FAc implant yielded a significantly improved BCVA at 24 and 36 months (24-month MD = 4.52; 95% CI 2.56–6.48; 36-month MD = 8.10; 95% CI 6.34–9.86). These findings were comparable with the FAME study. The FAc implant yielded significantly reduced 24- and 36-month CMT. Pooled proportions of cataract surgery, IOP-lowering drops and glaucoma surgery were 39%, 27% and 3%, respectively, all lower than the FAME study. Pooled estimate of supplementary intravitreal therapy was 39%, higher than the 15.2% of the FAME study. This meta-analysis of real-world studies confirms favorable visual and anatomical outcomes following FAc insert for chronic DMO. In real-life studies more than one third of patients received supplementary intravitreal therapy, an issue that needs to be further explored.


Results
shows the study selection process. A total of 1001 articles were identified. Following removal of duplicates, 638 articles were screened, of which 39 articles were deemed potentially eligible and were full-text evaluated. A total of 11 articles met eligibility criteria and were included, of which two reported the 24-and 36-month result of the same RCT, the FAME study 6,7 . Study characteristics. A total of 9 real-world studies [8][9][10][11][12][13][14][15][16] and one RCT 6,7 were included in this systematic review. The only RCT was the FAME study, which consisted of two parallel, multicenter, 36-month randomized clinical trials comparing the 0.2 µg/day FAc implant, the 0.5 µg/day FAc implant and a sham. A total of 375 eyes were enrolled in the 0.2 µg/day FAc implant group, of which 301 and 270 eyes were included in the 24-month and 36-month analyses, respectively 6,7 . Patient enrollment started in 2007 and the study was completed in 2010. The 24-month and 36-month results were published in 2011 and 2012, respectively. Foveal thickness evaluation was based on a time-domain OCT, namely the Stratus 3 OCT instrument. Baseline phakic eyes were 235 in the 0.2 µg/day FAc implant group 6,7 .
Risk of bias assessment. All real-world studies were given a MINORS score ≥ 11 (Supplementary Table   S1 online). Funnel plots inspection revealed a nearly symmetrical shape for each outcome explored. Egger's test confirmed no significant publication bias ( Supplementary Fig. S1-4  Articles included in the qualitative synthesis (n = 11 ) Articles included in the quantitative synthesis (n = 11 )    Author, year

Number of eyes with supplementary intravitreal therapy Drug (number of eyes) Injection number (mean) Mean time from FAc
Panos et al. 9 13 out of 24

Discussion
The present meta-analysis explored for the first time real-world outcomes of 0.2 µg/day FAc intravitreal implant for chronic DMO, primarily showing that long-term visual improvement is comparable with the FAME study. The FAME study reported the results of two parallel phase III randomized controlled trials comparing three different groups, namely the 0.2 µg/day FAc intravitreal implant, the 0.5 µg/day FAc intravitreal implant and  Randomized clinical trials have a primary role in evidence-based practice since the highest level of evidence is based on these type of trials 18 . Phase III trials investigate efficacy and safety of a drug and the results of these trials may allow drug licensing 19 . Phase IV trials and post-marketing studies evaluate the effectiveness and safety of the drug during the post-commercialization phase in a real-world setting 8,19 . Real-world studies report findings produced outside the context of RCTs. As such they are a reliable indicator of what should be expected in clinical practice 20 . These studies are of great value because clinical practice could fail to reach the same results obtained in RCTs. This is particularly relevant when it comes to chronic diseases that require continuous and intensive treatment, such as DMO.
A very recent real-world study including more than 28 thousand eyes with DMO demonstrated that realworld outcomes of intravitreal anti-VEGF therapy are worse than those of RCTs, irrespective of the anti-VEGF agent 4 . In real-world practice, patients received a mean of 6.3, 6.3 and 6.7 injections of bevacizumab, aflibercept and ranibizumab, respectively, over 1 year 4 , while these figures were 9.7, 9.2 and 9.4, respectively, in the DRCR.net Protocol T trial 21 . In real-world practice, 1 year visual gain was + 4.5, + 4.3 and + 3.4 letters for bevacizumab, aflibercept and ranibizumab, respectively 4 , while DRCR.net Protocol T reported a visual improvement of + 9.7, + 13.3 and + 11.2 letters, respectively, at 1 year 21 . Undertreatment mainly accounts for worse visual  The injection burden could be reduced by using sustained-release corticosteroid intravitreal implants, which present a long-lasting action as well as a proved clinical effectiveness for DMO treatment 22,23 .
A real-world study of 128 eyes on the use of the intravitreal 0.7 mg dexamethasone implant for DMO 24 reported even better outcomes compared to the MEAD trial 25 : 25% of eyes achieved a 15-letter gain at 36 months 24 versus 22% recorded in the MEAD trial 25 ; a mean of 3.6 implants were administered over 36 months 24 versus 4.1 recorded in the MEAD trial 25 .
Our meta-analysis demonstrated a visual gain of 4.52 letters and 7.89 letters at 24 months and 36 months, respectively, in agreement with a 24-month gain of 4.40 letters and a 36-month gain of 8.10 letters reported by the FAME study 6,7 . The fact that real-world evidence confirmed the visual improvement demonstrated by the RCT could be related to the long-term efficacy of the insert. It requires less frequent follow-ups and treatments compared with anti-VEGF. Thus, there is less chance of undertreatment. Our results confirmed the FAME finding of a better visual gain at 36 months compared with 24 months 6 . The authors speculated that such long-term benefits could be related to a trophic effect secondary to adequate control of inflammation 6 .
As concerns central retinal thickness, this meta-analysis yielded a CMT reduction of 127 µm and 170 µm at 24 months and 36 months, respectively. The 24-month result was characterized by higher heterogeneity and was less similar to the FAME finding compared to 36 months 6,7 . This comparison has to be considered cautiously because our outcome was mean CMT change, while the FAME study evaluated mean change in foveal thickness, defined as center point thickness, which is assumed to be the mean thickness at the crossing point of the 6 radial scans 6,7 . An average difference of approximately 30 µm has been reported between the 1-mm diameter CMT and the central point thickness 26 . Additionally, the FAME study used a time-domain OCT, while almost all studies included in the present meta-analysis were based on spectral-domain OCT. However, when evaluating mean changes, these minor differences are averaged out.
Cataract progression and IOP rise have been reported as the main adverse events following intravitreal FAc implant and intravitreal steroid use in general 6,7,23 . The included studies defined IOP rise applying different threshold values. Therefore, we chose to investigate the rate of patients requiring IOP-lowering drops and glaucoma surgery, which are two reliable measures of this complication in clinical practice. The pooled estimate rate of phakic eyes having cataract surgery was 39% in the real-world studies. This is much lower compared to the 80% rate seen at the 36-month follow-up of the FAME study 6 . Likewise, rates of eyes receiving IOP-lowering drops and glaucoma surgery were lower in the real-world setting compared with the 36-month follow-up of the FAME study (27% vs 38.4% and 3% vs 4.8%, respectively) 6 . Possible reasons to explain this difference could be an underestimation of these events because of the shorter follow-up of the included real-world studies compared to the 36-month follow-up of the FAME study as well as other biases and confounders related to a real-world setting. In particular, in clinical practice the FAc insert is often administered after treatment with other intravitreal steroids, such as a dexamethasone implant, which could help to select non-steroid responders. Moreover, the dexamethasone intravitreal implant showed a lower real-world 24 rate of IOP rise compared with those of the MEAD trial 25 (IOP ≥ 25 mmHg: 10.2% vs 32% at 36 months). Similarly, rates of cataract surgery following dexamethasone intravitreal implant were lower in real-world practice 24 compared with the MEAD trial 25 (47% vs 59%). Cataract surgery was associated with better outcomes when performed at the same time as intravitreal dexamethasone implant administration because of a reduction of postoperative inflammation 27 . This could be assumed to be valid also for the FAc implant.
The main advantage of the FAc intravitreal implant is long-term efficacy and, as a consequence, a reduced treatment frequency. In turn, this could result in a reduction of complications related to intensive anti-VEGF intravitreal injections. These include endophthalmitis with repeated intravitreal injections as well as a theoretical risk of death and cardiovascular events 28,29 . However, increased risk of mortality and cardiovascular events Figure 6. Pooled estimates of rates of eyes requiring cataract surgery, supplementary intravitreal treatment, intraocular pressure (IOP) lowering drops, glaucoma surgery. Blue histograms refer to pooled estimates from real-world studies; red histograms refer to rates from the 36-month FAME study.  30 . Metaanalysis studies have also shown no higher risk overall 28,31 , raising only a possible warning signal of increased risk in subjects with the highest-level of exposure (i.e. high risk diabetic patients receiving long-term intensive intravitreal anti-VEGF therapy) 28,30 .
Our results showed that 39% of eyes treated with the FAc implant in real-world practice received additional intravitreal therapy for DMO treatment, much greater than 15.2% reported by the 36-month FAME study 6 . Even if these two figures cannot be directly compared given the difference in methodology and design between real-world studies and RCTs, our analysis suggests that more than one third of patients treated with a FAc insert for chronic DMO could require an additional intravitreal therapy. It would have been informative to know how many treatments and with which frequency they were given in clinical practice, but such analyses were not conducted because of lack of evidence. This discrepancy between real-world practice and RCT could be explained by the fact that the FAME study was conducted in the period 2007-2010 7 , when the treatment of DMO was mainly performed with macular laser or off-label steroids. At that time, intravitreal anti-VEGF therapy was licensed for age-related macular degeneration, but not yet approved and scarcely used for DMO treatment 32 . In the FAME study, intravitreal anti-VEGF and triamcinolone were not deemed as allowable rescue treatment and were administered only in cases not experiencing any improvement 7 . Furthermore, the FAME study was based on time domain OCT imaging 7 , while most of the included real-world studies adopted spectral domain OCT imaging. All these factors probably contributed to the lower percentage of patients receiving additional treatment in the FAME study.
Additionally, 13% of patients enrolled in the FAME study received an additional 0.2 µg/day FAc implant during the follow-up period and this additional therapy could have reduced the need for other intravitreal agents. Conversely, only three real-world studies recorded the use of additional 0.2 µg/day FAc implants, with few patients (1-8.6%) having such a retreatment during the follow-up period.
This study had the following limitations. First, we conducted a meta-analysis of real-world studies, which, by definition, have different designs compared with RCTs. Therefore, no formal analysis could statistically compare real-world findings with RCT findings, but visual comparison of pooled estimates and 95% CIs with those obtained by RCT allowed us to assess whether real world outcomes matched those obtained with the RCT or not. Furthermore, while BCVA and CMT outcomes were reported by the included studies at the different time points of interest, namely 24 and 36 months, this was not the case for proportions of eyes receiving cataract surgery, IOP-lowering drops, glaucoma surgery and additional intravitreal therapy. All these proportions were provided throughout the study follow-up, which differed among the included studies. Nonetheless, in all cases mean follow-up exceeded 12 months. Ultimately, included studies might have been influenced by different clinical variables due to their real-world setting and bias could have been introduced. However, all meta-analyses were characterized by low heterogeneity across studies, except the one on the 24-month CMT. Included studies were deemed as low-to-moderate risk of bias. Funnel plots inspection revealed no significant risk of publication bias. All these support a good quality level of evidence. Moreover, a meta-analysis has more accurate confidence and higher power than a single report 33,34 .
In conclusion, our study revealed favorable outcomes in terms of visual improvement and macular thickness reduction following an intravitreal FAc implant for chronic DMO, which is in line with the findings reported by the FAME study. While the pooled proportion of cataract surgery and eyes experiencing requiring-treatment IOP rise are not concerning, the pooled estimate rate of eyes requiring additional intravitreal therapy is significant and further studies are warranted to better investigate this issue.

Materials and methods
Search method. This study was based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (PRISMA checklist available as Supplementary Table S2 online).
Systematic search of studies on FAc intreavitreal implant use for chronic DMO was conducted on Pubmed, Embase and Medline databases, from their inception up to 16th October 2020. The search strategy was performed including the terms 'fluocinolone acetonide' , 'diabetic macular edema' , 'diabetic macular oedema' , 'macular edema' , 'macular oedema' , 'diabetic retinopathy' , connected by and/or in various combinations. Reference lists of included studies and potentially eligible studies were also screened. Eligibility criteria. The following inclusion criteria had to be meet: (1) to report on the use of 0.2 µg/day FAc intravitreal implant for chronic DMO; (2) to report outcomes at 24-month follow-up or longer; (3) to report data on the primary outcome of this meta-analysis; (4) to include a minimum of 10 patients for the primary outcome measure of this meta-analysis. No restriction on study design was imposed. Only articles published in peerreviewed journals and in English were considered. Abstracts and conference posters were excluded. Reports including only vitrectomized eyes were excluded as well. Real-world studies were defined as those reporting data collected outside the context of RCTs 20 .
The primary outcome of the present study was the mean change in best corrected visual acuity (BCVA) following FAc intravitreal implant at 24 months. Secondary outcomes included BCVA change at 36 months, the mean change in optical coherence tomography (OCT) central retinal thickness (CMT), the rate of supplementary intravitreal therapy, and the rate of adverse events, such as cataract surgery, rates of eyes requiring IOP lowering drops and glaucoma surgery. Central macular thickness was the average value of the fovea-centered area with 1 mm diameter 35 37 for RCTs and non-randomized studies, respectively. A MINORS score ≥ 9 was considered as low-to-moderate risk of bias.
Statistical analysis. We first meta-analyzed effect sizes for primary and secondary outcomes obtained from real-world studies. Specifically, primary outcome was BCVA change at the 24-month follow-up, reported as mean differences (MDs) between post-treatment and baseline values and their 95% Confidence Interval (95% CI). Similarly, 36-month BCVA change and CMT change were reported as MD with 95% CIs. Further secondary outcomes included the proportion of eyes receiving supplementary intravitreal therapy, cataract surgery (i.e. exclusively among phakic eyes), IOP lowering drops and glaucoma surgery. For each individual study, the score confidence intervals were constructed and proportions were pooled using the Metaprop command on Stata (version 16) 38 . Heterogeneity across studies was tested using the Q-statistics and the I 2 index. A fixed effect model was applied in the absence of significant heterogeneity, while a random effect model with the DerSimonian-Laird method was applied if p value for Q-statistics < 0.1 and I 2 > 50%. The extent of publication bias was explored by Funnel plots and tested using Egger's test.
Pooled effect sizes with their 95%CI were compared with those obtained by the RCT. All the statistical analyses were carried out on STATA (version 16) with significance level α < 0.05 if not otherwise stated.
Ethical approval and informed consent. Since this is a systematic review, ethical approval and informed consent are not required. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.