Corresponding Author: Kyu Hyung Park, MD, PhD. Department of Ophthalmology, Seoul National University Hospital, Seoul National University College of Medicine, 101 Daehak-ro, Jongno-gu, Seoul 03080, Korea. Diabetic macular edema (DME) is a leading cause of vision impairment in patients with diabetic retinopathy and a significant global healthcare burden [1,2]. The increasing prevalence of diabetes worldwide has contributed to a rising incidence of DME, necessitating effective treatment strategies to prevent visual deterioration and maintain quality of life [3]. Current treatment options for DME primarily include intravitreal anti-vascular endothelial growth factor (anti-VEGF) injections, corticosteroids, and laser photocoagulation, each with distinct efficacy and safety profiles [4-7].
While randomized controlled trials have demonstrated the effectiveness of anti-VEGF agents in improving visual and anatomical outcomes, real-world evidence suggests that treatment patterns and patient adherence often differ from clinical trial settings [8,9]. Variability in treatment regimens, injection frequency, and patient response in real-world practice may impact long-term outcomes [10]. Additionally, the economic burden of DME management, including treatment costs, clinic visits, and patient adherence, highlights the importance of understanding healthcare resource utilization [11].
This study aims to evaluate real-world treatment patterns for DME, assess functional and anatomical outcomes, and analyze healthcare resource utilization trends in South Korea. By providing insights into real-world clinical practice, this research seeks to bridge the gap between clinical trial data and real-world data, ultimately guiding more effective and sustainable management strategies for DME.
Materials and Methods
Ethics statement
This prospective cohort study was approved by the Institutional Review Board of Seoul National University Bundang Hospital (No. B-1601-330-307). All participants provided written informed consent. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki.
Study design and setting
This prospective, observational study was conducted at 17 hospitals in South Korea from January 2017 to June 2022. The key inclusion criteria required participants to be aged 20 years or older with a diagnosis of type 1 or type 2 diabetes and evidence of DME confirmed by spectral-domain optical coherence tomography (OCT) and fluorescein angiography. Patients were categorized into center-involving DME (CI-DME) and non-center-involving DME (non-CI-DME) groups based on OCT findings. The CI-DME group included eyes with retinal thickening or hard exudates within 500 μm of the foveal center and a central subfield thickness (CST) ≥300 μm within the 1-mm central subfield on OCT. The non-CI-DME group included eyes with retinal thickening of at least one disc diameter located ≥500 μm from the foveal center, CST ≤300 μm within 1-mm central subfield on OCT, and retinal thickening of ≥400 μm located ≥500 μm from the foveal center. The Spectralis OCT (Heidelberg Engineering) was used for 110 patients, 80 with the DRI Triton OCT (Topcon), and 19 with the Cirrus OCT system (Carl Zeiss Meditec). Segmentation errors on OCT were reviewed and manually corrected when present to ensure accurate assessment of DME status and retinal thickness measurements. Key exclusion criteria included prior intraocular surgery, and other ocular or systemic conditions that could interfere with study outcomes. A full list of inclusion and exclusion criteria, along with the criteria for group classification, is provided in the Supplementary Fig. 1.
Ocular examinations, visits, and treatments were conducted at the discretion of retinal specialists at each institution. Data were recorded at baseline screening (visit 1) and at 6 months (visit 2), 12 months (visit 3), 18 months (visit 4), and 24 months (visit 5). At visit 1, demographic data, history of ocular treatments, best-corrected visual acuity (BCVA) in letters, and OCT findings including CST, were collected. Demographic data were collected, including age, sex, smoking, and comorbidities. During follow-up visits, the number of visits, treatment history including anti-VEGF, steroid, focal laser, and panretinal photocoagulation, and findings of ophthalmic examinations, including BCVA, and CST were documented.
Statistical analysis
Serial changes in variables were compared between the two groups using linear mixed models. The group and visit were included as fixed effects, with patient-specific random intercepts and an autoregressive correlation structure of order 1. Paired t-tests were used for comparisons between visits within each group. Bonferroni correction was applied for multiple comparisons of the values. All statistical analyses were performed using R ver. 4.4.2 (R Foundation for Statistical Computing). We considered p < 0.05 as statistically significant throughout our analysis.
Results
A total of 209 participants were enrolled (81 women [38.8%] and 128 men [61.2%]), with a mean age of 60.2 ± 10.8 years. CST was higher in the CI-DME group than in the non-CI-DME group (419.0 ± 115.9 μm vs. 279.4 ± 45.1 μm, p < 0.001) (Table 1). Of the 209 participants at visit 1, 165 remained for visit 2, 149 for visit 3, 124 for visit 4, and 124 for visit 5, and data on these patients were included in the analysis (Table 2). Among the 68 participants with non-CI-DME at baseline, 10 (14.7%) developed CI-DME during the follow-up (Fig. 1A-1F).
A linear mixed model analysis showed no significant interaction between visit and group for BCVA (p = 0.085), while significant interaction for CST (p < 0.001), indicating different longitudinal trends between the two groups (Table 3). In the CI-DME group, the mean BCVA changed from 61.1 ± 19.3 letters at baseline to 64.0 ±17.3 letters at visit 5 (p > 0.999). In the non-CI-DME group, the mean BCVA changed from 66.7 ± 18.2 letters at baseline to 66.8 ± 17.9 letters at visit 5 (p = 0.661) (Fig. 2). In the CI-DME group, 89 patients were treatment-naive and 52 were previously treated group. The mean BCVA did not significantly change in either subgroup (treatment-naive: 59.2 ± 20.9 at visit 1 vs. 63.6 ± 17.7 letters at visit 5, p > 0.999; previously treated: 64.7 ± 15.5 vs. 64.5 ± 17.1 letters, p > 0.999). A similar pattern was observed in the non-CI-DME group (treatment-naive [n = 45]: 65.3 ± 20.1 vs. 66.7 ± 19.5 letters, p = 0.703; previously treated [n = 23]: 69.5 ± 13.5 vs. 66.9 ± 15.9 letters, p > 0.999). In the CI-DME group, the mean CST significantly decreased from 419 ± 116 μm at baseline to 338 ± 101 μm at visit 2 (p < 0.001), 340 ± 125 μm at visit 3 (p < 0.001), 332 ± 121 μm at visit 4 (p < 0.001), and 343 ± 126 μm at visit 5 (p = 0.001) (Figs. 3, 4A-4F).
Over the 2-year period, the total number of anti-VEGF injections was 3.1 ± 3.6, while steroid injections were 0.7 ± 1.5 (Table 4). The CI-DME group received significantly more anti-VEGF injections than the non-CI-DME group (3.8 ± 3.9 vs. 2.0 ± 2.8, p = 0.004). Among patients with CI-DME, 10 of 141 (7.1%) received three to five anti-VEGF loading injections, while in the non-CI-DME group, 2 of 68 (2.9%) received the same treatment. Focal laser treatment was performed more frequently in the non-CI-DME group (14 [20.6%] vs. 5 [3.5%], p < 0.001). The CI-DME group had more frequent visits (15.8 ± 7.3 vs. 11.3 ± 8.1, p = 0.017) and underwent OCT imaging (9.7 ± 4.2 vs. 7.4 ± 3.9, p = 0.023) more often over the 2-year period.
The number of anti-VEGF injections decreased over the follow-up period only in the CI-DME group, while steroid injections showed no change in either group (Fig. 5A, 5B, and Supplementary Table 1). The number of visits significantly decreased over the follow-up period in both groups, while the number of OCT examinations decreased only in the non-CI-DME group (Fig. 6A-6C and Supplementary Table 2).
Discussion
In this study, we reported the functional and anatomical outcomes of DME patients in multiple institutions in Korea, as well as their treatment and examination patterns. The CI-DME group showed a significant reduction in CST compared to baseline, which was well maintained over the 2-year period. The BCVA slightly improved in the CI-DME group, although the change did not reach statistical significance. This functional outcome is somewhat disappointing, as it is poorer than not only the results from clinical trials but also those from other real-world studies. The DRCR.net Protocol T trial showed gains of +12.8, +12.3, and +10.0 letters at 2 years for eyes treated with aflibercept, ranibizumab, and bevacizumab, respectively [12]. Similarly, the VISTA/VIVID trial reported gains of +10.3 to +11.7 letters at 3 years for eyes treated with aflibercept [13]. Recent real-world studies on DME treatment have reported visual gains of approximately 3.2 to 8.2 letters at 1 year [10,14-16]. The primary reason for this difference is that the number of injections in our cohort was significantly lower compared to other studies. In clinical trials, anti-VEGF injections were typically administered on an every 4- or 8-week schedule, resulting in a significantly higher number of injections compared to real-world studies [13,17,18]. Real-world injection numbers tend to be lower than clinical trials, often 3.9 to 7.9 injections in year 1, then declining in subsequent years [10,14-16,19]. In our cohort, the number of anti-VEGF injections was 1.8 in the first year and 3.1 over 2 years. Specifically, in the CI-DME group, the injection frequency was 2.2 in the first year and 3.8 over 2 years. This number is lower than the injection frequency reported in other studies, indicating the possibility that patients in our cohort were undertreated. Similar to our result, a real-world data from Japan showed that a group receiving anti-VEGF therapy for DME had a mean of 3.8 anti-VEGF injections over 2 years, with CST improvement but no significant visual acuity gain [20]. One of the reasons for the lower injection frequency in real-world data may be the time and financial burden on patients [21]. Additionally, in South Korea, insurance criteria for DME treatment (a limit of 14 injections per patient and a CST ≥300 μm) may encourage physicians to prefer a as-needed regimen. Future studies incorporating socioeconomic data and healthcare system are warranted to better understand the factors underlying low treatment and monitoring frequencies.
Secondly, our study included a more heterogeneous patient population than clinical trials or other real-world studies, which may have contributed to the poorer visual outcomes. Since one-third of patients are known to have an incomplete response to anti-VEGF therapy, the 75 of 209 patients (36%) in our cohort with a history of previous treatment likely influenced the poor outcomes [22]. Subgroup analysis within the treatment-naive patient group also showed no significant improvement in visual acuity, which may be attributed to undertreatment and the heterogeneity of the patient population.
Steroid injections were not frequently administered but were consistently used over the 2-year period (0.7 ± 1.5 times). This is more than a previous study in the United States (0.3 injections in the 24-month cohort) [10] but less frequent than in a Japanese study (2.0 ± 1.3 for 2 years) [20]. The use of local steroids is likely to vary depending on the healthcare environment of each country, such as accessibility to cataract surgery.
Over the 24-month period, patients visited an average of 14.2 times, equating to more than every other month. OCT was performed approximately nine times, roughly every 2 to 3 months. While this frequency is lower than the monthly visits and examinations seen in clinical trials, it is relatively reasonable considering the nature of real-world data. However, we found that the number of visits and OCT examinations decreased over 2 years. In a previous study investigating DME treatment patterns and healthcare resource utilization in Korea, the number of clinical visits, treatment visits, and visits for OCT and fluorescein angiography were highest in the first 6 months, followed by a decline by year 3 [23]. This suggests the need for measures to ensure continuous follow-up for DME patients.
This study has several limitations. The small sample size reduced the reliability of the statistical analysis. Additionally, incomplete follow-up, with not all baseline patients reaching visit 5, and missing data at intermediate visits further hindered the analysis. One of the major factors contributing to the high rate of follow-up loss in our study may be the large proportion of patients treated at tertiary referral hospitals, where limited accessibility due to economic, geographic, and time constraints is common. Furthermore, reduced adherence to regular intravitreal injections or examinations over time likely played a role. This trend has also been noted in previous studies, which reported a gradual decline in compliance among patients with DME, often leading to poorer visual outcomes [24,25]. Furthermore, patients underwent examinations and treatments based on the judgment of retinal specialists at each institution, rather than following a standardized treatment and examination protocol. Lastly, although subgroup analyses based on insurance status, treatment cost, and the type of anti-VEGF agent could potentially provide important insights, these analyses were not feasible due to the lack of collected data or the heterogeneity of treatment patterns. Despite these limitations, this study remains valuable as it prospectively collected data from multiple South Korean medical institutions, providing insight into overall treatment and examination patterns for South Korean patients with DME.
In conclusion, DME patients recruited from multiple centers in South Korea showed poor visual outcome despite anatomical gains, likely due to the low number of injections. Encouraging continuous follow-up and treatment adherence in DME patients may help improve visual outcomes.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Full text via PMC
Download Citation
Supplement1
Print
