Evaluating the Treatment of Diabetic Macular Edema with Aflibercept Based on a Regional Network of Ophthalmologic Care Givers

Purpose In Austria, anti-VEGF therapies are reimbursed only in clinical settings. This study aimed to describe the outcome of a treat and extend regimen (TER) with aflibercept for diabetic macular edema (DME) in a network of practitioners. Methods In a prospective study over 36 months, patients with DME were treated with a loading dose of aflibercept and further on with adjusted treatment intervals based on optical coherence tomography (OCT) findings. All patients were monitored in an outpatient setting by regional ophthalmologists, and the treatment was administered in the clinic. Main outcome parameters were best-corrected visual acuity (BCVA) from baseline to the last regular visit. Number of visits at the practitioner's office as well as the number of injections were secondary outcome parameters. Results Thirty-three patients completed the study at their final visit. BCVA improved significantly by 5.8 letters between baseline and the final visit from 70.4 letters at baseline (p=0.004). Patients visited the practitioner's office 12.8 times in the observation period of 36 months. 3.7, 5.1, and 3.9 visits were performed, respectively, in the first, second, and third years, and 25.5 ± 7.9 injections were performed. The mean interval of injections over the observation period was 6.2 ± 2.2 in weeks. Conclusion The treat and extend regimen was valuable for treating patients with DME in this specific setting. The functional results of this study were comparable to those of other real-world evaluations. Adherence to the same treating institution seems to be important to avoid differences in therapeutic decision making and may also increase patient's compliance.


Introduction
Diabetic macular edema (DME) is one of the major causes of vision loss in developed countries and the leading cause of vision loss in patients with diabetes type 1 and type 2, afecting 1 in 15 people with diabetes [1]. Anti-VEGF (antivascular endothelial growth factor) agents are the gold standard in the treatment of DME. Compared to laser monotherapy, anti-VEGF intravitreal therapy showed superior BCVA (best-corrected visual acuity) achievements in multiple studies [2,3]. Afibercept, bevacizumab, and ranibizumab showed improvement of BCVA depending on initial vision loss, whereas afibercept was more efective than the other two agents in patients with severe vision loss [4]. Te two main concepts to administer anti-VEGF therapy in clinical practice are either PRN (pro re nata) or TER (treat and extend regimen). In PRN, treatment is only conducted in case of recurring macular edema while visits are performed on a regular basis, usually every four weeks, whereas in TER, the interval of the visits will be extended or reduced individually for each patient depending on the activity of the disease. Te advantage of TER is reducing the number of visits and avoiding overand under-treatment.
In Austria, anti-VEGF therapy is only reimbursed in a clinical setting and not in out-clinic ofces, leading to a signifcant organizational burden for ophthalmological departments. Even more, the overall requirement of intravitreal injection is exponentially growing throughout the last decade. For this reason, a network of caretakers of a university clinic and of ophthalmologic practitioners were established to avoid redundant examinations and defne a disease management protocol for the anti-VEGF treatment of patients with DME. Based on this protocol, patients were followed by practitioners indicating the individual treatment interval, and all the treatments were performed at the university clinic.
Tis study aimed to evaluate a TER with afibercept in this network of intra and extramural caregivers. TER was chosen to lower disease burden of the patients and for evaluation of patients' adherence to this specifc regimen. Morphologic and functional results and the efcacy of the treatment protocol were analyzed in detail.

Study Design.
Te study was a monocentered, prospective, open-labeled phase 4 trial investigating a standard TER with afibercept (Eylea, Bayer Pharma AG, 2 mg/0.05 ml per injection). Te study followed the tenets of the Helsinki agreement and was approved by the local Ethics Committee.

Inclusion and Exclusion Criteria.
Patients with DME were recruited from the outpatient clinic of the Dpt. of Ophthalmology of the Kepler University Clinic, Linz, Austria as participants of the study. Tey were either referred by external ophthalmologists or by self-referral. Main inclusion criteria were as follows (1) patients with diabetes type 1 or 2; (2) the presence of intra or subretinal fuid within the central 6 mm and central subfeld macular thickness ≥300 μm, as determined by the optical coherence tomography (OCT; Spectralis OCT ® , Heidelberg Engineering, Dossenheim, Germany); (3) best-corrected visual acuity (BCVA) >30 letters based on the ETDRS charts. If both eyes were afected, the eye with worse BCVA was included in the study. Both proliferative and nonproliferative stages of diabetic retinopathy were included. Te need for or previous laser coagulation was not exclusion criteria. Before being admitted to this study, the subject must have consented to participate after the nature, scope, and possible consequences of the clinical study have been explained in an understandable way.
Key exclusion criteria were any prior treatment of DME within 3 months before inclusion, other reasons for macular edema or visual impairment except for ametropia, inability to communicate in German or English, and active intra or periocular infection in the study of eye or hypersensitivity to afibercept.

Treatment and Assessment.
Treatment with afibercept was initiated with a loading dose consisting of 5 monthly injections with afibercept at the department of ophthalmology, Kepler University Clinic, Linz, Austria. Following this loading dose, treatment intervals were extended or shortened continuously according to TER based on individual OCT scans. If there was no sign or no change of intraretinal fuid of more than 10% in two follow-up visits in the central 6 millimeters as assessed by the EDTRS grid in macular OCT reports, the interval was extended in steps of 4 weeks to a maximum of 16 weeks. If a follow-up investigation revealed the presence of increasing or constant DME involving the central 6 millimeters, the interval was reduced by 4 weeks. Intervals were either shortened or extended in steps of 4 weeks (minimum 4 weeks and maximum 16 weeks). If the maximum interval of 16 weeks was achieved twice, the treatment was ceased (Figure 1). Following the loading dose, depending on the presence of DME, the practitioner decided whether the treatment interval should be shortened or extended. To evaluate the individual treatment decision in the network of caregivers in ofces and in the clinical centers, each follow-up visit was performed twice, at the outpatient practitioner's ofce and at the study center of the Kepler University clinic. All the treatments were performed at the university clinic.
At each follow-up, a full-ophthalmologic examination was performed with fundus examination, OCT scan, and BCVA. Fluorescein angiography (FA) was performed at the baseline, after 6, 12, and 24 months using 10% injectable solution fuorescein sodium (SERB, Paris, France). Standardized OCT measurements were performed using the Spectralis HRA + OCT system (Heidelberg Engineering, Heidelberg, Germany).

Outcome Measures.
Te change in the BCVA from the baseline to the fnal visit was the primary end point. Te secondary outcome measures were CRT injection interval, number of visits at the clinic, and the practitioner's ofce.

Statistical
Analysis. Wilcoxon signed-rank or Mann-Whitney U nonparametric tests were used to compare the measured values. Categorical variables were analyzed using the χ 2 test or Fisher's exact test. Te statistical signifcance was defned as p < 0.05. IBM SPSS Statistics for Windows (v23.0, IBM Corp., Armonk, NY, USA) was used to perform statistical analyses.

Results
Forty-one patients were enrolled, of which 33 fnished the study according to the study criteria. Twenty-fve patients completed the study on the fnal visit at month 36 (Table 1). Eight patients had their fnal visit before month 36, because the injection interval was extended twice to 16 weeks. Eight patients could not complete the study (death n � 1, termination of patient n � 3, cardiorespiratory diseases that lead to study dropout n � 4).
Te mean interval between the follow-up visits was 6.2 ± 2.2 weeks over the observation period. In the frst year, the mean interval (weeks) between injections following the loading dose was 5.0 ± 1.5 (median 4, range 4 to 8), 7.2 ± 3.9 (median 5.3, range 4 to 16) in the second year and 8.1 ± 4.4 (median 6.0, range 4 to 16) in the third year. Te interval increased signifcantly from the frst year to the third year by 3.2 weeks (±4.1, 95% confdence interval 1.7 to 4.8, p < 0.001).
Patients visited the practitioner's ofce 12.8 ± 10.6 times in the observation period. In the frst year, 3.7 ± visits were  Figure 1: After a patient was included in the study, 5 monthly injections were admitted as loading dose. 2 months after the ffth injection, the patient went to the practitioner's ofce for OCTscan and funduscopic examination. If a macular edema is present, the patient was referred to our clinic for reevaluation and injection if indicated. Table 1: Age, sex, and baseline-values for BCDVA and CRT for the whole study population and for the subgroups who had their fnal visit at month 36 or before the month 36. Te p value describes the diference of the visual acuity and the central retinal thickness at the baseline and at the fnal visit between the two groups. Seventeen patients visited the practitioner's ofce more than 6 times a year in the observation period. Tose patients had altogether 431 visits to their ophthalmologist. Te decision of the external ophthalmologist was congruent with the decision of the clinic in 86% to admit an intravitreal injection (n � 372). In 14% (n � 59), the decision difered. In 32% (n � 19) of the incongruent decisions, the practitioner indicated a therapy, as opposed to the clinic. In 68% (n � 40) of the incongruent decisions, the clinic indicated an injection, as opposed to the practitioner. However, the evaluation of diferences in treatment decisions was not the study aim.

Subgroup Analysis.
Patients were divided into two subgroupsthose who completed the study at visit month 36 (n � 27, "36 months complete") and those who fnished the observation before month 36 (n � 6, "36 months incomplete") because their injection interval was extended to 16 weeks twice. At the baseline, BCVA, age, and CRT showed no statistically signifcant diference between both groups (Table 1). In group "36 months complete", there was no statistically signifcant increase in BCVA from the baseline (70.8 ± 12.2) to month 36 (74.4 ± 13.7; p � 0.165). However, CRT decreased signifcantly from 480 ± 156.2 μm at baseline to 314.7 ± 97.3 μm at month 36 (p < 0.001). In the group "36 months incomplete", BCVA increased signifcantly by 15.2 ± 9.9 letters from the baseline (68.7 ± 11.5) to the fnal visit (83.8 ± 4.6, p � 0.014). Te CRT in this group decreased accordingly (p < 0.001) from 399.8 ± 22.4 μm at the baseline to 298.8 ± 29.1 μm at the fnal visit. In terms of BCVA and CRT, no diference was observed between both groups, neither at the baseline nor at the fnal visit (see Table 1).

Discussion
Tis study evaluated the morphologic and functional outcomes of TER with afibercept in patients with DME in a network of university clinics and outpatient practitioners. Te main outcome parameter, BCVA, from the baseline to the fnal visit, increased signifcantly by 5.8 letters (±13.4, p � 0.019), with the highest gain of letters from the baseline to month 6 with 6.3 letters (±10.0, p � 0.001). Accordingly, the CRT decreased by 154.1 (±151.3 μm, p < 0.001). As these results are comparable to other TER trials, this individual disease management program proved to be efcient in a healthcare setting, providing reimbursement for an intravitreal anti-VEGF treatment only in a clinical setting [5][6][7].
To further evaluate the characteristic diferences according to the individual retreatment intervals, in a post hoc analysis, all patients could be divided mainly into 2 subgroups. In subgroup 1 (n � 6), following the loading dose, the retreatment interval could be extended twice to 4 months. According to the protocol, these patients fnished their study participation before month 36. In this group, BCVA increased signifcantly by 15 letters (±9.9, p � 0.014)  from the baseline to the individual fnal visit. In subgroup 2 (n � 27), all patients received frequent retreatments throughout the study period of 36 months. In this group, BCVA increased by 4 letters (p � 0.165), showing no statistical signifcance from the baseline to month 36. Tose results indicate a worse visual outcome in patients with a persistent need for retreatment for DME. Tis is according to the fndings of a study by Bressler et al. showing the persistence of DME throughout a treatment period of 3 years in some patients, correlating with a worse visual outcome [8]. Any comparison to other studies based on TER in clinical settings is confned by diferent inclusion criteria. In our study, representing a real-world setting, patients were included with clinically signifcant DME without prior treatment in the last three months. Frequently cited phase 3 studies, such as VIVID and VISTA, included DME patients if BCVA was between 73 and 24 letters [3]. Apart from that, in these studies, a fxed dose regimen was used with fxed retreatment intervals. Another trial by Pak and co-authors based on TER with afibercept revealed a signifcant and even higher increase in BCVA of 8.9 letters by month 12 compared to our study [5]. Tis could be due to an adjustment of the retreatment interval by 2 weeks, as opposed to 4 weeks in our study. All in all, in this optimized phase 3 study setting, the mean BCVA increased by 5.8 throughout the 3 years. Tus, the results were comparable compared to the VIBIM, although our observation period was longer (one year in VIBIM study).
In our study, we chose a maximum treatment interval of 16 weeks due to several reasonsFirst, the burden of disease is reduced by lowering the number of clinical visits and injections. Second, the vitreous half-life of afibercept is longer than that of ranibizumab, which should reduce the number of injections [9]. As presented in the ALTAIR study, the interval of afibercept injection can be extended to more than 12 weeks in 57-60% of the patients sufering from age related macular degeneration [10]. In eyes with DME that received bevacizumab and switched to afibercept as a second-line therapy the treatment interval was extended to a maximum of 10 weeks, as shown in the TADI study [11]. Tose patients improved visual acuity in the one year follow up.
Te mean injection interval was 6.2 weeks over the observation period of 3 years, ranging from 4 to 12 weeks. Te mean interval between injections signifcantly raised from 5 (±1.5) weeks in the frst year to 8 (±4.4) weeks in the third year (p < 0.001). Te mean number of injections per year in our study decreased signifcantly from 11 (±1.5) injections in the frst to 7.9 (±3.3) in the third year (p < 0.001).
Apart from functional and morphological outcomes, this study aimed also to evaluate the efcacy of a network consisting of a university clinic and ophthalmologic practitioners. In Austria, the intravitreal anti-VEGF treatment is only reimbursed in clinical settings. Tis leads to an overload of follow-up visits and necessary re-treatments. Te idea of the network was to reduce the burden caused by the clinical examinations and additional waiting time for patients and to outsource them to partners in external ophthalmologic ofces. Following their examination and advice, patients were deferred to the university clinic for the anti-VEGF treatment. Only to evaluate the efcacy of the network in the study setting, OCT examinations were repeated in the university clinic. Retreatment criteria were increasing or still decreasing DME involving the central 6 millimeters. In 86%, the decision for retreatment was identical in the ophthalmologic ofces and the university clinic. In the remaining 14%, decisions difered with the clinical retina experts tending to indicate more re-treatments compared to their partners in ophthalmological ofces.
Patients' adherence to the external ophthalmologist was not as high as expected within the study group. Only 17 patients visited the external ophthalmologist more than 6 times a year in the observation period. Terefore, the incongruence of the decisions between the clinic and practitioner is rather inconclusive. Not visiting the practitioner was not an exclusion criterion. In the area where the study was conducted, the distance to the treating clinic could be rather long in individual cases, causing an additional treatment burden. It has been shown previously that the distance of the external ophthalmologist to the patient's residence has an infuence on the patient's adherence to the treatment [12]. Nevertheless, it can be concluded that the treating institution, whether a centralized university clinic or the decentral practitioner, should be responsible for the whole diagnostic and therapeutic process to avoid diferences in the treatment approach.
As shown in a retrospective analysis in which 28,658 eyes with DME were screened, the gain of VA in the frst year correlated with the number of injections [13]. Tis implicates that the adherence is crucial for maintaining visual acuity in patients with DME. Research shows that patient's adherence to the anti-VEGF therapy is associated with better visual outcome in eyes with DME and nAMD [14]. Apart from that, it is known that nAMD patients show a greater compliance to the therapy than patients with DME [15]. In the VIOLET study, it was shown that in patients with DME, both TER and PRN regimens achieve similar functional results. However, fewer injections were administered in the TER group, making higher patient adherence more likely [16].
Tere are several limitations to this studyMorphologic fndings in OCT scans were not described in detail by the ophthalmologic practitioners. It had to be determined if the criteria for a reduction or elongation of the retreatment interval were met. Tis could have helped to avoid inconsistency between the decisions of ophthalmic practitioners and clinical ophthalmologists. Furthermore, extending and reducing the retreatment interval by 2 instead of 4 weeks as it was conducted in this study could have improved the treatment outcomes. Nevertheless, previous studies have shown that a 4-week interval extension is a reasonable strategy to treat patients with diabetic macular edema [17,18]. Te study was performed as an uncontrolled noncomparative study. Terefore, it cannot be compared with other administration regimens such as monthly injections or pro re nata use. Also, the sample size was relatively small, for this reason, the signifcance of the therapeutic efect is reduced.
Journal of Ophthalmology 5

Conclusion
Treat and extend regimen is valuable and widely used for treating patients with diabetic macular edema, since the results are comparable with fxed dosing intervals. Patient's treatment burden can be lowered by extending the intervals, and the healthcare system can be relieved by decreasing the frequency of treatment and controls. Adherence to the same treating institute is important to avoid diferences in therapeutic decisions and may increase patient's compliance.

Data Availability
Te data used to support the fndings of this study are available from the corresponding author upon request.

Conflicts of Interest
Haidar Khalil received lecture fees from Bayer, Austria, and other authors declare that they have no conficts of interest.