Background
Diabetic macular edema (DME) is a complex disease of multifactorial origin in which fluid accumulates in the retinal layers due to the disruption of blood retinal barrier and increased vascular permeability [
1‐
3]. Treatment options to manage DME include intravitreal injections of anti-vascular endothelial growth factor (anti-VEGF) or intravitreal corticosteroids (i.e. dexamethasone, fluocinolone acetonide and triamcinolone acetonide) [
4‐
6]. The intravitreal dexamethasone implant (IDI, 0.7 mg, Ozurdex®: Allergan, Inc., CA, USA) is a biodegradable, sustained-release drug delivery system that releases dexamethasone into the vitreous for up to 6 months, and is currently approved for the treatment of macular edema secondary to retinal vein occlusion [
7], non-infectious posterior uveitis [
8] and DME, based on the results of the MEAD trial [
9]. This study pooled the data from 2 randomized, multicenter, masked, sham-controlled, phase 3 clinical trials (
ClinicalTrials.gov identifiers NCT00168337 and NCT00168389), that demonstrated visual and anatomic improvements in DME eyes, which were confirmed in subsequent trials [
10,
11].
However, significant concerns appear with regards to the applicability of clinical trials results to real world scenarios. First, in clinical practice, the selection criteria are less strict, usually limited to failure of other therapies. This fact is especially relevant as eyes treated with IDIs in routine clinical practice are often those in which other therapies have primarily failed (laser, anti-VEGFs, etc.) and are at risk of developing chronic macular edema, limiting their potential for greater visual gains. Second, the visit and treatment schedule applied in clinical trials rarely reflects real world clinical conditions. In particular, in the MEAD trial reinjections of the IDI could not be performed prior to 6-months, limiting the potential of the implant to achieve greater visual gains in poor responsive eyes. Third, the potential loss of follow-up visits can produce an overestimation of the benefits and/or an underestimation of its side effects, affecting either way the outcomes reported in comparison to clinical trials. For these different reasons, it is important to evaluate the IDI performance in real-life scenarios.
Currently, anti-VEGF therapies are considered the first-line therapy for DME, and meanwhile the importance of corticosteroid therapy has been recognized it is mainly employed as a second-line therapy. As suggested by the EURETINA guidelines for the management of DME [
12], IDIs are only considered as first-line therapy in patients whose medical history excludes the use of anti-VEGF therapies or in specific conditions: history of major cardiovascular events, unwillingness to receive monthly injections, or pseudophakic patients. Nevertheless, there is a growing body of evidence in the last years supporting the benefits of IDIs in naïve DME patients, and several studies have reported better visual outcomes compared to refractory DME eyes [
13,
14]. In contrast, very few studies have evaluated specifically the treatment frequency and the number of injections in naïve DME eyes compared to previously treated eyes.
Thus, the purpose of the present study is to audit the use of the IDI in a large series of DME eyes treated in real-life clinical conditions, to identify different treatment patterns in naïve versus previously treated eyes. The study was performed over a 5-year period at two tertiary referral retinal units from a well-defined geographic area that covered a population of 1.8 million individuals. In addition, a specific sub-analysis was carried out to identify differences in baseline characteristics, VA, anatomical outcomes, number of injections and reinjection frequency in naïve eyes (defined as eyes with no prior intravitreal therapies) and previously treated eyes (which previously received intravitreal drugs). The results obtained were compared to those reported in the literature in clinical trials (e.g., MEAD, CHAMPLAIN, BEVORDEX) and previous real world published studies, to address the performance of IDI in a large cohort of unselected DME eyes in real-life conditions.
Methods
Study design
This study was approved by the Institutional Review Board (IRB) at the Hospital Clinic of Barcelona and it was conducted in accordance with the Tenets set forth in the Declaration of Helsinki (HCB/2016/0905). Clinical data were collected retrospectively from 2 specialized tertiary referral retina clinics in Barcelona (Spain): Institut Clínic de Oftalmología (ICOF) at Hospital Clinic of Barcelona and Hospital Vall d’Hebrón. No written informed consent was required as data was retrospectively collected from routine clinical care, as approved by the reference IRB. All eyes receiving IDI injections for DME between October 2010 and May 2015 were included in the study. A comprehensive dataset was distributed and completed in both study centers within the predetermined timeframe. Patient identifiers were removed to anonymize the data, and data from the individual centres were collated and merged into a centralized database for analysis.
Clinical data collection
Data collected included demographics (age, gender, etc.), laterality, previous local treatments, number of previous injections, number of injections, surgical details, complications, current topical treatments, visual acuity (VA), intraocular pressure (IOP) and central retinal thickness (CRT) assessed by optical coherence tomography (OCT). This data was collected at all the study timepoints: baseline, 1–2 weeks, 6–8 weeks, and 3, 6, 9, 12, 18 and 24-months post-injection of the first IDI. Additional data was collected at each individual repeated injection during the study, including VA, IOP and CRT data prior to the procedure, and 1–2 weeks, 6–8 weeks, and 3 and 6-months post-procedure.
Data sources/outcome measurements
All original data were gathered in routine clinical care visits. All injections data (i.e. injection date, number of injections, pre and post-injection data) were collected as described above. At each time point, VA was determined as the best VA with habitual correction or pinhole, rather than as the best-corrected refracted VA and presented in logarithm of the mínimum angle of resolution (logMAR) units. The analysis of eyes with a low VA was undertaken by substituting counting fingers (CF) and hand movement (HM) with 2.0 and 2.3 logMAR, respectively [
15]. IOP measurements were obtained by Goldmann tonometry and presented in mmHg. The CRT was determined by OCT imaging using one of 2 different devices depending on the participating center (Hospital Clínic, Cirrus HD-OCT
®, Dublin, CA, USA and Hospital Vall de Hebrón, Spectralis OCT
®, Heidelberg Engineering, Germany). No research softwares were employed to control for inter-device measurement differences in CRT. No missing value substitutions were performed in patients where data were not available for a particular visit or were lost during follow-up.
Statistical analysis
Descriptive, frequency statistics and the chi-squared test were used to assess the qualitative variables. The normality of quantitative variables was examined in histograms, and inter-group differences where evaluated with an independent Student’s T-test and Mann-Whitney U-test, when appropriate. A paired t-test was used to compare pre- and post-treatment changes. For VA change analysis, VA values are converted and presented in ETDRS letters. The cumulative probability of IOP events occurring after IDI injection are presented as survival curves using the Kaplan Meier (KM) method [
16]. High IOP was defined as an IOP greater than 21 mmHg, 25 mmHg or 35 mmHg, as described elsewhere [
17]. The probability of IOP elevation was evaluated for naïve and previously treated DME eyes subgroups, and KM survival curves were compared with the log-rank test. A
p-value ≤0.05 was considered significant.
Discussion
This study reports better visual outcomes and reduced treatment burden in naïve DME eyes compared to previously treated DME eyes, based on a large cohort of DME patients treated with the IDI in routine clinical care. The results of this study support the use of the IDI in early DME stages in patients who do not qualify for anti-VEGF therapy and provide further evidence of better IDI response when used as first-line therapy compared to its use as second-line therapy, after previous failed intravitreal treatments.
Our real-world cohort of DME patients presented different demographics and baseline clinical characteristics from those reported in clinical trials, overall with a worse mean baseline VA [
9,
10,
13,
18,
19] and CRT [
9,
10,
18,
19]. In our series, a significant number of study eyes were treated in routine clinical care with VA levels that fall outside the inclusion criteria used in the MEAD trial (33.5% of the study cohort, 6% with VA better than 0.4 logMAR and 27.5% with VA worse than 1.0 logMAR, as presented in Fig.
1). This data is of particular relevance, as reflects more closely the situation in which such a therapy is to be employed in routine clinical care: patients that require a therapeutic option even though they might not conform to the ideal profile, either in terms of specific disease related parameters, prior treatment failure or co-morbidities. In case of our cohort of previously treated eyes, their basal clinical characteristics are generally worse than those reported previously in real-world studies, particularly with lower baseline VA [
13,
14,
20‐
24] and greater CRT [
20,
25]. Moreover, more eyes had received prior treatment, particularly with anti-VEGF and/or IVTA, than the study cohorts in these previous series [
14,
20,
22,
24‐
26].
The overall results presented here are somehow comparable with those from previous smaller cohorts. Indeed, previous assessments of the effect of IDI on refractory and treatment-naive patients showed similar decreases in CRT and improvement in VA in both refractory and treatment-naive groups. Other real-life studies have also shown similar improvements in VA and CRT without serious adverse events, mainly in smaller series over shorter follow-up periods [
27‐
32]. In our population of DME patients, IDI considerably improved the VA relative to baseline in both treatment-naive and previously treated patients, suggesting that IDI therapy offers benefits to both types of patient. Nevertheless, naive eyes maintained a better mean VA than previously treated eyes at all time points studied. A particularly significant improvement was observed in naïve patients during the first 3 months relative to previously treated eyes, suggesting they respond better in terms of VA, consistent with earlier data [
13,
14]. Moreover, in our series the percentage of eyes with good VA levels was consistently higher in naïve eyes vs previously treated eyes at all study timepoints, as graphically presented in Fig.
1. Regarding anatomical changes, IDI treatment improves the CRT at all time points during the follow-up, both in treatment-naïve and previously treated eyes, with no significant differences between these subgroups. This discrepancy between functional and anatomical outcomes in DME has extensively been reported in previous studies, that suggest that retinal thinning may also be related with outer retinal layers atrophy preventing visual improvement, more common in chronic DME eyes [
33].
The need for frequent injections in DME represents a considerable burden for patients, especially with antiVEGF drugs [
34‐
36]. In different studies, fewer IDI injections have been shown to be necessary to achieve similar visual and anatomical outcomes in DME patients [
10,
11,
37], although the loss of vision mainly due to cataract must be controlled, which may be more common when IDIs are used. Further head-to-head trials will be needed to compare the efficacy of IDI and anti-VEGF therapy based on the patients’ clinical characteristics at baseline and their prior treatments, tailoring the treatment choice individually in a case-by-case basis. Towards this personalized medicine approach, several attempts have been recently reported to shed some light on predictive biomarkers for IDI response, based in retinal imaging (i.e. OCT) or aqueous samples [
38‐
41]. Likewise, recent studies have been directed to identify those patients who don’t respond to anti-VEGF treatment, [
42] as well as to determine the synergistic and beneficial effect of IDIs in combination with other treatments [
43,
44].
Significantly fewer IDIs were administered to treatment-naïve eyes than to previously treated eyes and indeed, many more treatment-naïve eyes received just 1 injection than previously treated eyes. However, when additional implants were required, the time to reinjection did not differ significantly between the two groups. This is an important point, as suggest that reinjections were timely performed in both groups when required. Such differences have not always been detected when this parameter has been compared between naïve and non-naïve eyes [
14,
20]. Our results raise the interesting hypothesis that early treatment may reduce the treatment burden associated to IDI therapy in naïve DME, beyond the benefit already reported for refractory DME eyes when used as second line therapy. It is well recognized that managing DME with IDIs is generally associated with a need for fewer injections than anti-VEGF therapies, as well as longer periods between the need for treatment. The benefits to be gained from this need for fewer injections have already been recognised in the EURETINA guidelines for DME, whereby IDI use is recommended as a first-line therapy only in specific subgroups of patients. If confirmed in future studies, this finding may offer an additional reason to support the use of IDI as first-line therapy in a wider spectrum of DME eyes.
As found elsewhere, IDI therapy was well tolerated by DME patients. The main adverse event was high IOP but that could typically be managed with medication. Indeed, while nearly a third of patients developed a high IOP, it was controlled with topical antihypertensive drugs. There is no evidence of a previously cumulative effect of multiple injections on increased IOP, irrespective of pre-existing glaucoma or ocular hypertension (OHT) [
17,
45]. However, the probability of taking IOP-lowering medication increased considerably from 12 months to 24 months, an important consideration when using the IDI in routine clinical care [
46].
Our study has some limitations, such as the retrospective design of the study. In addition, a detailed comparison of the data obtained with that from previous clinical trials and other retrospective studies is complicated by the fact that they involve cohorts with different characteristics at baseline (Table
4). This is particularly the case of DME severity and the inclusion/exclusion criteria, and notably in terms of the baseline VA that differs substantially between trials, as well as the type and length of the prior treatments, and the intervals between the repeated IDI injections. Despite the absence of detailed comparisons, the results of the present study in a clinical setting are somehow comparable to the data from clinical trials. Furthermore, as the data is obtained from a cohort of patients not included in a clinical trial it better reflects patients seen in clinical practice, both treatment-naive and previously treated patients with DME. Therefore, similar outcomes can be expected of IDI treatment at other clinical centres for DME patients, as reported by previous smaller cohorts. Moreover, it is likely that the patients included in this study are patients that might otherwise not be recommended for such therapy as they do not conform to the criteria employed in clinical trials. It should also be noted that some of the patients considered to be naïve to treatment in this study had received prior laser therapy but not any prior intravitreal therapy. This may contribute to explain differences in treatment response and disease progression (and/or adverse events) between this cohort of naïve patients and those studied elsewhere. Finally, two different OCT machines from two different manufacturers were employed in this study (one in each participating center), but CRT measurements were not adjusted for inter-device differences and this data should be interpreted with caution, especially when comparing these results with other clinical practice studies and trial outcomes.
Table 4
Comparison of dexamethasone intravitreal implant clinical trials and real-life studies of in diabetic macular edema (series ≥30 eyes)
MEAD Boyer DS et al., 2014 [ 9] | DME | 36 | 351 | 56.1 | 463.0 | 65.8% | 7.1% | 16.5% | 58.6 | 351.4 |
MEAD-Treated Augustin AJ et al., 2015 [ 19] | DME-Treated | 36 | 247 | 55.2 | 478 | 93.5% | 25% | 23.5% | 58.4 | 352 |
CHAMPLAIN Boyer DS et al., 2011 [ 18] | DME-PPV | 6 | 55 | 54.5 | 403.4 | 69.6% | 48.2% | 57.1% | 57.5 | 364.5 |
BEVORDEX Gillies MC et al., 2014 [ 10] | DME | 24 | 88 | 55.5 | 474.3 | – | – | – | 62.4 | 287.3 (*) |
Escobar-Barranco et al., 2015 [ 13] | DME-Naive | 6 | 40 | 59.6 | 568 | 0% | – | – | 71.1 | 323 |
DME-Treated | 6 | 36 | 51.3 | 600 | 100% | – | – | 59.0 | 281 |
| DME-Treated | 6 | 58 | 52* | 543.24 | 74.1% | 75.9% | 67.2% | 58.5* | 420.16 |
| DME-Treated | 6 | 30 | 57* | 517 | 56.7% | 100% | – | 64* | 411 |
| DME | 14.53 | 52 | 44* | 514.2 | 100% | 67.2% | – | 51* | 419.9 (6 months) |
| DME | 4 | 39 | 51.5* | 559 | 44% | 49% | 36% | 81.5* | 477 |
| DME treated and naive | 6 | 78 | 53.9 | 537.6 | 42.3% | 52.6% | 17.9% | 60.1 | 384.6 |
CHROME | DME (subgroup analysis) | 36 | 34 | 55* | 450.4 | 55.9% | 55.9 | 38.2 | 53 (6 months) | 259.5 |
RELDEX Malclès et al., 201 7[ 20] | DME naive and treated | 36 | 128 | 50.5 | 450 | 16.4% | 70.3% | 15.6% | 60.6 | 280 |
IRGREL-DEX Iglicki et al., 2019 [ 14] | DME naive and treated | 24 | 130 | 55* | 575 | 15% | 7.4% | – | 65.5* | 294.4** |
This study: Zarranz-Ventura et al. 2020 [ 6] | DME-All | 24 | 203 | 39 | 498.7 | 75.8% | 53.6% | 25.6% | 45 | 402.6 |
Naive | 24 | 67 | 42.5 | 482.1 | 56.7% | 0% | 0% | 47 | 340.5 |
Previously treated | 24 | 136 | 37.5 | 506.5*** | 85.2% | 80.1% | 38.2% | 44 | 427.4*** |
Acknowledgements
The authors would like to thank the patients that attended the visits described in this series, as well as the help provided by the administrative and support staff in the daily work at the Retinal services of both study centers.
Hospital Clínic1 - Hospital Vall de Hebron2 Intravitreal Dexamethasone Implant study group (in alphabetical order):
- Alfredo Adan1
- Socorro Alforja1
- Carolina Bernal-Morales1
- Anna Boixadera2
- Ricardo P. Casaroli-Marano1
- Sergio Copete2
- Laura Distefano2
- Marc Figueras-Roca1
- Jose García-Arumi2
- Joan Giralt1
- Victor Llorens1
- Marina Mesquida1
- Manuel J. Navarro-Angulo1
- Laura Pelegrin1
- Barbara Romero-Núñez1
- Anna Sala-Puigdollers1
- Daniel Velazquez-Villoria2
- Miguel Angel Zapata2
- Javier Zarranz-Ventura1
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.