Real-Life Results after the Administration of a Single 0.19 mg Fluocinolone Acetonide (ILUVIEN^®) Implant in Patients with Refractory Diabetic Macular Edema

In emerging nations, diabetic macular edema (DME) is the main cause of diabetes mellitus related visual impairments [1]. In DME, the breakdown of the blood-retina barrier leads to increased vascular permeability, which induces an exudation and accumulation of extracellular fluid and proteins in the macula [2]. Further, the DME associated chronic, low-grade inflammation in DME fosters vascular changes [3].

Current guidelines for the first-line treatment of DME recommend intravitreal injections (IVIs) of anti-vascular endothelial growth factor (VEGF) [4], which result in a rapid decrease of macular edema and improvement of visual acuity (VA) [5, 6]. However, treatment regimen with anti-VEGF requires a large number of injections that may be disruptive to the patient’s quality of life. This might explain the poorer functional outcomes in real-life practice compared to clinical trials [1, 7].

Moreover, up to 30–40% of diabetic patients respond inadequately to anti-VEGF therapy [8] and may develop persistent DME [9]. Then, corticosteroids as a valid alternative treatment can reduce the expression of VEGF and the synthesis of pro-inflammatory mediators and activation of retinal immune cells that are supposed to be responsible for the retinal cell degeneration [10, 11]. Thus, to reduce treatment burden and improve patient adherence, the use of a sustained-delivery intravitreal corticosteroid implant may offer longer lasting treatment effects.

The 0.19-mg fluocinolone acetonide (FAc) implant (Iluvien^®; Alimera Sciences, Berlin, Germany) is a small, cylindric, and nonbiodegradable implant that releases submicrogram doses of corticosteroid (initially released at 0.25 µg/day [average, 0.2 µg/day]) into the vitreous for up to 36 months [12], while offering a low burst and near zero-order kinetics due to the pharmacokinetic dosing profile [13]. Randomized clinical trials of FAc have indicated both efficacy and safety in DME [14, 15]. However, we lack real-life data.

Since these real-life data are not consistently in line with the pivotal studies, our primary aim was to evaluate the functional and anatomical efficacy of the FAc implant in real-life practice over 3 years. The secondary aim was to determine prognostic anatomical biomarkers of treatment response, the need for additional treatment, and the safety profile.

We retrospectively included 44 eyes from 29 patients treated between January 2017 and June 2020 at Eye Hospital Sulzbach. The follow-up lasted from January 2017 to August 2023. We considered consecutive type 1 or type 2 diabetes mellitus patients, with a minimum age of 18 years, who received a FAc implant for chronic refractory DME. Refractory DME was defined by a central retinal thickness (foveal thickness, FT) ≥250 µm (Stratus-equivalent values) over 24 weeks that was previously treated with intravitreal dexamethasone (DEX), triamcinolone or anti-VEGF. Exclusion criteria were other causes of macular edema or previous FAc implant. The study was conducted in accordance with the Declaration of Helsinki. The Ethical Committee of Saarland (243/14) approved the study and written consent was obtained from all patients.

Patient demographic, medical, and ophthalmological data were collected from patient records, which included age, gender, ophthalmological history, lens status, previous DME treatments and history of panretinal photocoagulation treatment. A complete ophthalmological examination, i.e., measurement of best-corrected VA (BCVA), intraocular pressure (IOP), slit lamp examination, dilated fundus examination, and spectral-domain optical coherence tomography (SD-OCT, Spectralis; Heidelberg Engineering, Heidelberg, Germany), was performed at the initial visit and repeated at each follow-up visit. Macular thicknesses were automatically calculated by the SD-OCT software, and manually adjusted when necessary. The central maximum thickness (CMT) was defined as the maximum thickness of the retina from Bruch’s Membrane to Internal Limiting Membrane, measured within the 1 mm-Early Treatment Diabetic Retinopathy Study (EDTRS) grid centered on the fovea. The FT was defined as mean retinal thickness within the 1 mm-EDTRS grid.

Treatment with FAc was initiated according to international guidelines. The FAc implant was injected through the pars plana superotemporal quadrant 3.5–4 mm posterior to the limbus, using a 25-gauge injector under topical anesthesia. In case of no additional treatment, we monitored patients every 3 months after FAc injection. In contrast, patients that received additional treatment were monitored monthly. Additional treatment with DEX, triamcinolone, or anti-VEGF was indicated at the ophthalmologist’s discretion and generally administered in the event of anatomical DME recurrence or a significant drop in BCVA. The follow-up data were documented until loss of FAc efficacy, defined as recurrence of DME in SD-OCT (FT increase >10%).

The primary outcome measures were FT, CMT, and BCVA before FAc injection (defined as baseline) and at the follow-up visits and their variation to baseline. Secondary outcome measures included IOP and recording of adverse events.

Qualitative variables were recorded as numbers and percentages, quantitative variables are presented as mean, standard deviation (±SD), median, range, and subsections with their corresponding p values. In order to consider that DME can occur asymmetrically, we computed ANOVA-based linear mixed-effects models with SPSS^® (IBM SPSS Statistics^®, Version 28.0.0.0) which allow to differentiate between the left and right eye of 1 and the same patient. All pairwise ANOVA comparisons were adjusted with Bonferroni corrections.

The mean interval from the last IVI (anti-VEGF or steroids) was 5.73 ± 4.24 months. 22/44 (50%) eyes had additional pathologies relevant to VA excluding cataract. The patient demographics are summarized in Table 1.

The therapeutic loss of FAc effect occurred on average after 21.34 ± 12.74 months. The supplemental free probability after 6 months was 91% (n = 40), after 12 months 80% (n = 35), after 24 months 55% (n = 24) and after 36 months still 50% (n = 22) (Fig. 1). One patient died after 24 months due to other medical reasons and his follow-up ended at 24 months.

The number or type of therapeutic procedures prior to FAc injection had no influence on the time until loss of efficacy during the course (p = 1.000). Half of the cases had secondary diagnoses that could influence impaired VA, but this influence was not significant compared to patients without secondary diagnoses (p = 0.067). Further, these patients’ CMT was not significantly higher in the course (p = 0.07). However, they had a significantly higher FT (p = 0.02) which could explain the poorer VA. Neither the stage of diabetic retinopathy nor the systemic therapy had a significant influence on FT, CMT, or BCVA (p > 0.05). However, the subsample of preoperatively vitrectomized eyes (n = 19) showed a significantly shorter maintenance of FAc effect (20.2 ± 12.8 vs. 23.1 ± 12.7 months, p < 0.001). There was no significant difference in anatomical effect on FT (p = 0.403) or CMT (p = 0.875), though.

The clinical outcome (Table 2) showed a significantly reduced FT and CMT in all eyes for the first time 6 months after FAc implant compared to baseline (Fig. 2; FT p < 0.01, CMT p = 0.01) which persisted until month 24. At month 24, macular edema of 5 patients increased and a new IVI was performed. The remaining 22 eyes had a significantly lower CMT than baseline (p < 0.001), and this effect lasted until month 36. Nevertheless, despite postoperative visual improvement over time, BCVA did not significantly change from baseline (p = 0.568).

Subdividing the eyes according to FT and CMT at baseline showed that eyes >400 µm respond with a greater reduction in retinal thickness (Fig. 3; p < 0.001). However, regarding BCVA, eyes with low preoperative BCVA (cut-off ≥0.6 logMAR) also retained poorer BCVA after FAc (Fig. 3; p < 0.001). There was no influence of the cut-off BCVA on post-implant outcome of FT and CMT (p = 0.531 and p = 0.415, respectively).

Mean IOP was stable during follow-up (p = 1.000). Only four eyes required IOP-lowering procedures. Two patients showed increased IOP that did not respond to medication, so that 1 patient who had received FAc in both eyes underwent canaloplasty after 16 months (first eye) and 27 months (second eye), respectively. Another patient with proliferative diabetic retinopathy with rubeosis iridis underwent cyclophotocoagulation after 37 months. One patient temporarily received topical monotherapy that controlled the IOP after additional 12 months (Table 3).

No patient showed endophthalmitis, retinal detachment, or vasculitis after injection. Two eyes (both with scleral fixed intraocular lenses) required vitrectomy due to implant migration into the anterior chamber. Before possible consecutive corneal decompensation, in one eye the implant was fixed via pars plana with retained therapeutic effect until 36 months. In the other case, upon patient’s demand, the implant was removed 1 month after injection. Nevertheless, this patient still showed a therapeutic effect for 12 months. He has been excluded from the further follow-up analysis after removal of the implant.

As a safe and effective treatment, anti-VEGF or combined anti-VEGF and anti-angiopoietin-2 therapy is considered the gold standard for the treatment of DME. However, intravitreal biologics are a relatively short-acting treatment option that can lead to important limitations such as high injection burden, nonadherence, tachyphylaxis and retinal thickness variability as the injection effect decreases. Besides, despite monthly anti-VEGF therapy, the Diabetic Retinopathy Clinical Research Network (DRCR.net) protocol T showed that up to 65.6% of treated eyes with DME still had persistent DME at 24 weeks, which might be a potential consequence of not treating the underlying chronic low-grade inflammation [16].

It is evident that the pathophysiology of DME is multifactorial, including numerous inflammatory proteins [3]. Hyperglycemia-induced chronic low-grade inflammation and associated inflammatory cytokines (e.g., VEGF, interleukin-6) are key drivers of neurodegeneration and vascular dysfunction in diabetic retina [17]. In combination with anti-VEGF therapy, corticosteroids may provide a broader mechanism of action. They are used to reduce the breakdown of the inner blood-retinal barrier and extravasation from leaky vessels [18] and also downregulate the production and activity of inflammatory factors including cytokines and growth factor (e.g., VEGF, angiopoietin 2) [19‒21]. Deuchler et al. [22] showed that the FAc implant led to rapid and sustained reduction of some cytokines with improvement of the overall clinical presentation. Also, OCT-angiography analysis revealed improvements in macular perfusion after treatment with FAc. This may be associated with corticosteroid related beneficial effects on leukostasis [21].

The final PALADIN data, a 3-year, phase 4, nonrandomized, open-label, observational study, at 36 months showed that FAc provided stable or improved VA, decreased macular thickness, reduced treatment burden, and predictable and manageable IOP changes [14]. In this real-life retrospective study monitored over 3 years, we could also demonstrate efficacy of the FAc implant in the treatment of chronic DME.

First, evidence showed that after FAc implant patient’s VA improves, accompanied by reduced needs for additional treatments und consecutively a lower mean number of treatments per year [14, 23]. Other real-life studies have also shown an average gain in VA and a functional response to FAc in more than 85% of eyes [24]. Although in our study BCVA improved after FAc, we found that eyes with low preoperative BCVA (cut-off 0.6 logMAR) retained significantly worse BCVA after FAc. Mathis et al. [25] also showed that the strongest baseline characteristic associated with final BCVA was baseline BCVA, i.e., poor baseline BCVA is associated with worse final BCVA but greater BCVA gain. This association had been demonstrated in further studies [24, 26]. Consequently, an early treatment of DME affected eyes before severe vision loss due to neurodegeneration is crucial. This is further emphasized by better final BCVA in eyes with a shorter duration of DME [24].

The functional results were previously associated with anatomical improvement displayed as decrease in retinal thickness over time [25, 27]. Mathis et al. [25] showed that the gain in BCVA is negatively associated with a decrease in macular thickness: mean central macular thickness decrease continued until 18 months and at 21 months reversed to a subsequent increase, while BCVA continuously increased until 21 months and then decreased. Our real-life data confirmed an improvement in FT, CMT, and BCVA with a shift after an average of 24 months. However, the correlation between initial morphological recovery (macular thickness) followed by the consecutive functional recovery (BCVA) confirms that DME treatment must be initiated based on anatomical recurrence, not waiting for functional deterioration.

Moreover, FAc implant is assumed to be more effective in mild DME than in moderate or severe DME and it is suggested to primarily treat with agents (e.g., anti-VEGF, DEX, triamcinolone) and afterward use FAc implant for maintenance control over longer periods of time [28]. However, in our study, subdividing the patients according to their FT and CMT showed that eyes with a macular thickness >400 µm at baseline respond with significantly greater reduction in thickness than eyes ≤400 µm. It should be noted that other SD-OCT biomarkers, such as the presence of subretinal fluid, hyperreflective foci or hard exudates, were shown to be predictors of a better response to steroids than anti-VEGF. Therefore, an SD-OCT-based decision as included in the recent algorithm for DME management is recommended [7].

Treatment burden measured by time to first additional treatment is an important side effect in the treatment of chronic diseases such as DME. Although comparisons cannot be made directly because of differences in study design and patient population, multiple studies showed that FAc implant leads to reduced treatment frequencies and burden compared to other DME treatments [14, 25, 29]. For example, patients in the PALADIN study received a mean of three adjunctive treatments over 36 months, i.e., one treatment per year [14]. The incidence of remaining free of additional treatment over 36 months was 25.53% in the group of eyes followed up for 36 months [14]. These results are consistent with previous real-life studies: Patients received a mean of 4.7 injections in the year before FAc in contrast to 1.4 injections in the year after FAc injection. Similarly, these patients had a 63% probability of remaining free of additional treatment at 12 months [30]. Merrill et al. [29] demonstrated a 51% probability of remaining treatment free at 12 months. In our study, the therapeutic loss of effect of FAc occurred after a mean of 21.34 ± 12.74 months and after 12 months the probability of remaining treatment free was 80%. Mathis et al. [25] also showed that almost 2 thirds of patients do not need additional treatments during the period of follow-up. The mean timepoint of retreatment changed from 3.7 months to 10.3 months after FAc injection. Besides, eyes without requirement for additional treatment showed better functional and anatomical results, even though the VA did not reach significance level [25]. Likewise, in the US Retrospective Chart Review in Patients Receiving ILUVIEN (USER) study, among patients with DME, the treatment frequency improved from 4.1 injections per year before FAc to 0.8 injections per year after FAc implant while VA maintained or improved [31]. Furthermore, the Fluocinolone Acetonide in Diabetic Macular Edema (FAME) study, in which DME patients gained a mean of 8.1 letters over 36 months after FAc, showed that fewer supplemental DME treatments were required compared to control (13.4% vs. 34.8%; p < 0.001) [23]. In sum, these data are in alignment with our study and support FAc as a therapy that slows disease progression and controls the DME over a mean of 21 months.

Unlike other currently available therapy options for DME such as anti-VEGF (bevacizumab, aflibercept, ranibizumab, brolucizumab, faricimab), triamcinolone or DEX, the FAc implant continuously provides a very low dose of fluocinolone into the vitreous over a mean period of 18–36 months, according to different real-life and randomized controlled trials [32‒34]. Thus, a retreatment after FAc cannot be equated with a failure since the treatment frequency and thereby the risk of complications such as endophthalmitis in nearly all cases is still reduced. It should be noted that some eyes seem to require higher doses of steroid to maintain their anatomical stability, though. In these cases, supplemental treatments with other agents may work as a bolus dosing to maintain dryness. In order to work out biomarkers for differentiating the patients, further studies are necessary.

In other studies, the subsamples of vitrectomized and nonvitrectomized eyes showed no significant difference in efficacy regarding retinal thickness or BCVA [25, 35]. However, in our study the subsample of preoperatively vitrectomized eyes (n = 19) showed a shorter effect of FAc. There was no difference in anatomical effect on FT or CMT, though. Nonetheless, according to Pessoa et al. [36] the presence of vitreous seems to play a role in the response pattern to FAc, with a more constant, predictable, and stable effect observed in vitrectomized eyes.

Along with improved disease control, FAc has displayed a good safety profile across prior studies [14, 29, 37]. Although it is well known that steroids can lead to ocular hypertension and induce glaucoma in the long term, throughout the PALADIN study IOP remained predictable and manageable (IOP >30 mm Hg in 10.89% of eyes). In addition, the rate of IOP-lowering surgery due to FAc implant was only 1.49% [14]. Also Mathis et al. [25] showed that only 11% of the eyes experienced ocular hypertension during follow-up while 17.7% of these required IOP-lowering treatment, including one eye requiring surgery. Their results regarding IOP incidences seem to be similar to other reported real-life studies but poorer than reports of randomized controlled trials [24, 38, 39]. Our real-life data includes 2 patients (4.5%) with ocular hypertension and 3 patients (7%) with glaucoma requiring local therapy who were previously stable and controllable after DEX implants. During follow-up after FAc, 9% of eyes developed an increased IOP which was manageable in all cases. The requirement of undergoing a steroid challenge before implantation according to the guidelines of FAc helps improve the safety profile and reduce the necessity of additional IOP-lowering therapy. Singer et al. [14] show that an IOP of <25 mm Hg after the steroid challenge predict that 96.92% of eyes would have a similar IOP outcome to FAc implant at the last visit.

Our results show that the FAc implant was well tolerated during follow-up. There was no case of endophthalmitis, retinal detachment or vasculitis. However, two eyes required vitrectomy due to migration of the implant into the anterior chamber. In general, eyes with posterior capsule defect have an increased risk for the anterior chamber migration of intravitreal implants [40, 41]. Therefore, it is crucial to consider communication between the anterior and posterior eye compartment before injection of FAc implant. In addition to IOP concerns, another important consideration is the risk of cataract formation and progression. According to a cross analysis between the DRCR.net protocol I (ranibizumab) and the FAME trial (FAc implant), the cataract formation is indeed higher after FAc versus ranibizumab. In the FAME 3-year final readout, the rate of eyes requiring cataract extraction was 80% in the subgroup of phakic eyes, compared to 14% 2 years after treatment with ranibizumab in DRCR.net protocol I [23, 42]. Furthermore, in the PALADIN study, 18 out of 29 phakic eyes required cataract extraction [14]. In our real-life data, 91% of eyes were already pseudophakic. All four remaining phakic eyes showed cataract progression after FAc, three of them requiring cataract extraction. Patient education regarding the need for cataract surgery after FAc injection is therefore mandatory.

It should be mentioned that the exposure and outcome assessments could not be controlled or modified. Among other things, the precise follow-up time with strictly defined retreatment criteria was missing. Therefore, we are, inter alia, unable to specify the chronological order of functional and anatomical changes based on our data. Moreover, different factors such as patient age, disease severity and HbA1c value, secondary diagnoses, or diet might influence the outcome. However, the aim of our real-life study was to obtain data on treatments and their outcomes in routine clinical practice and thus complement randomized clinical trials. Besides, despite limitations due to retrospective design and small sample size, our data are consistent with other studies by demonstrating functional benefit of FAc treatment.

In this 3-year real-life retrospective study, the FAc implant improved in FT, CMT, and BCVA in DME patients. However, the therapeutic loss of FAc occurred after a mean of 21 months. The injection frequency improved for the patients as the probability of remaining treatment free was 80% after 12 months.

Although the number of cases is small, it was shown that eyes with FT/CMT >400 µm at baseline responded with a greater reduction in retinal thickness and eyes with a low preoperative BCVA (cut-off 0.6 logMAR) maintained significantly worse after FAc. It was also shown that preoperative vitrectomized eyes had a significantly shorter maintenance of the FAc effect, but there was no significant difference in FT or CMT.

Even if FAc is effective and can potentially contribute to reducing the treatment burden, patients must be clearly informed about the development of a cataract. The migration of the implant into the anterior chamber should also be explained to patients with, e.g., posterior capsule defect. Due to the potential increase in IOP, which was also shown in this study, patients without preexisting glaucoma should also be monitored in this respect and informed about possible pressure-lowering therapy or even follow-up surgery.

The study was conducted in accordance with the Declaration of Helsinki. This study was reviewed and approved by Ärztekammer Saarland, Approval No. 243/14. Written informed consent was obtained from all patients.

The authors have no conflicts of interest to declare.

No funding was received.

Warda Dariwsch and Annekatrin Rickmann: conceptualization, data curation, formal analysis, investigation, methodology, validation, writing – original draft, and writing – review and editing; Maria della Volpe-Waizel: conceptualization, formal analysis, validation, and writing – review and editing; Philipp K. Roberts, Karl T. Boden, and Peter Szurman: validation and writing – review and editing.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.