The increase in retinal vascular permeability which drives the accumulation of liquid within the layers of retina and increases its thickness results in DME. It is associated with the disruption of the blood-retinal barrier and increase in the levels of VEGF [15]. The DME treatment has been revolutionized by the progress in anti-VEGF injections, with a well-documented recovery in BCVA and a reduction of CMT, was confirmed by several RCTs [16, 17]. Nevertheless, there are only a few studies in real-life setting comparing IVR and IVA in DME patients, while most of them have included low number of patients [18–20].
This retrospective, real-life study shows long term visual and anatomical outcomes in patients with sight-impairing, center-involving DME treated with IVR or IVA up to 3 years in routine clinical practice in a tertiary eye center. The visual and anatomical functions improved at year 1 and maintained stable at years 2 and 3 in both groups, compared to the baseline. The amount of BCVA improvement between the groups was similar except at first year. At year 1, BCVA recovery was significantly superior (p = 0.042) in IVA group. The change in CMT was also similar between the groups.
Protocol T is a RCT which compares IVA and IVR in DME patients. Based on the results of protocol T, IVA seemed to be more efficient in BCVA improvement, especially in eyes with BCVA between 20/50 and 20/320 at first year [21]. However, there was no difference in BCVA improvement between two drugs at second and fifth years [4, 22]. In our real-world study, we also measured BCVA differences between IVR and IVA groups at 1st and 2nd years, similar to the Protocol T. However, there were some major differences. First of all, in our real-world study, the baseline mean BCVA of IVA group was significantly lower than IVR group (52.2 letters vs 58.3 letters, p = 0.04), which is probably the most important factor to explain the difference. In Protocol T, the median baseline BCVA were 68 and 69 letters in IVR and IVA groups, respectively. And also, the mean number of injections in IVA group was significantly higher than IVR group (5.32 vs 4.98, p = 0.023) which may also be associated with the BCVA difference. In Protocol T, the median number of injections were 10 and 9 in a monthly period in IVR and IVA groups, respectively. Furthermore, over 2 years, 52% and 41% of eyes treated with IVR and IVA, respectively, received focal/grid laser (p < 0.01). Unlike our study, the macular laser treatment rates were higher and there was a statistically significant difference between two groups.
In general, the number of injections in real-life studies remains to be lower than that of RCTs. In our study, the mean number of injections received in first year was 5.2 similar to other real-life studies in DME patients (between 3.1–7.2 in first year) [23, 13, 24]. However, in RCTs, injection numbers were higher (mostly 8 to 10 in first year) [25, 16, 26]. In 2nd and 3rd years, mean number of injections were 2.19 and 2.07, respectively. After the 1st year, the number of injections was reduced which may be related to disease modifying properties of anti-VEGF treatment in DME patients [27].
The mean BCVA changes in IVR group were 4.5, 4.7 and 6.6 letters at years 1,2 and 3 and in IVA group, it were 8.2, 6.0, and 7.4 letters at years 1,2 and 3. In protocol T it was 11.2 vs 13.3 letters at year 1, 12.3 vs 12.8 letters at year 2 in IVR and IVA groups, respectively [4]. In RESTORE-RESOLVE studies and Protocol I, visual improvement after IVR at year 1 were 6.1, 10.3 and 9 letters, respectively [8, 28, 29]. In RISE and RIDE studies, mean visual improvement after IVR at year 2 were 12.5 and 12 letters, respectively [6]. In VISTA and VIVID studies, improvement after IVA at year 1 were 12.5 and 10.7 letters, respectively [17]. In our real-life population, improvement in BCVA was less evident than Protocol T and other prospective studies as obviously. The most probable reason for the difference is the higher frequency of injections reported by these studies, being up to 12 injections per year. Additionally, patient selection, strict adherence to follow-up periods, early intervention to complications may contribute to the difference.
Plaza-Ramos et al. [18] conducted a real-life study with 213 eyes (122 IVR vs 91 IVA) and there was no difference between groups at 12 months. Conversely to our study, the mean baseline BCVA values were lower in IVR group (0.55 vs 0.48 logMAR) but not statistically significant (p = 0.109). Change in mean BCVA at 1st year was 0.15 logMAR in IVR group and 0.08 logMAR in IVA group which are similar to our year 1 results, and naïve patient ratio were higher in IVR group (70% vs 26%) which may contribute to final BCVA because chronic DME patients who previously treated with anti-VEGF therapy have a lower response to treatment [30]. Eleven eyes (9%) in IVR group and 3 eyes (3.3%) in IVA group were administered focal laser therapy (p = 0.096) which were not significantly different, but the frequency of laser therapy was much lower than our study (37.4% in IVR and 36.2% in IVA groups). In the real-life study of Shimizu et al. [19] with a similar concept, 49 eyes in IVR group and 46 eyes in IVA group are included and followed up for 6 months. The mean baseline BCVA was 0.48 logMAR in IVR group and 0.39 logMAR in IVA group. Change in mean BCVA at month 6 was 0.03 vs 0.09 logMAR in IVR and IVA groups, respectively and they reported that the effectiveness of IVA in improving the BCVA might be better than IVR. In the study of Bhandari et al. [20] which includes 303 treatment naïve eyes (136 IVR vs 167 IVA) of 228 patients with DME who completed 12 months follow-up period, authors founded that both IVR and IVA were effective for DME over 12 months, with aflibercept having somewhat better anatomical outcomes (change in CMT was − 126 𝜇m vs -89 𝜇m, p < 0.01). Larger BCVA gains were observed in IVA group when the initial BCVA was ≤ 0.3 logMAR (change in BCVA was 10.6 letters vs 7.6 letters, p = 0.01). Also 5 eyes (3.7%) in IVR group and 2 eyes (1.2%) in IVA group underwent macular laser treatment (p = 0.24), furthermore 7 eyes (5.1%) in IVR group and 4 eyes (2.4%) in IVA group received steroid injection. Frequency of these additional treatments was much lower than our study.
The main limitation of our study is its retrospective design which are inherent in real world studies. The treated eyes were not divided to stages of non-proliferative diabetic retinopathy. The real-world population was heterogeneous. However, we believe that publishing real-world data is of benefit to the clinicians who treat DME patients in real-life because clinical trial settings are difficult to apply in routine clinical practice. Still, further prospective studies with a larger cohort and longer follow-up time may be required to better understand and compare the output of IVR and IVA for DME. To the best knowledge of the authors, the current study which compares ranibizumab and aflibercept efficiency in DME patients, includes the largest real-world population in literature and have a satisfactory potential to reflect realistic clinical practice in routine.
In conclusion, both ranibizumab and aflibercept treatments achieved a good long-term visual and anatomical response in DME patients. In real-life setting, it may not be simple to comply with the strict follow-up and treatment protocols that were used in RCTs while treating DME. However, the clinical management of the patients should be optimized to have better outcomes in real-life.