ABSTRACT
Aims
To investigate the potential of sodium–glucose cotransporter 2 inhibitors (SGLT2i) as a less invasive treatment option for mild diabetic macular edema (DMO) via post hoc subgroup analysis. Previous findings were derived exclusively from observational studies, and no interventional trials based on protocol‐defined criteria had been conducted.
Materials and Methods
Based on the COMET trial data, this post hoc analysis targeted mild DME cases, defined by baseline central retinal thickness (CRT) ≤375 μm and best‐corrected visual acuity (BCVA) ≤0.155 (logMAR). Patients were randomized to SGLT2i (luseogliflozin) or sulfonylurea (SU, glimepiride) groups. Mild subgroups were compared with the corresponding non‐mild cases. The primary outcome was the 48‐week frequency of intravitreal ranibizumab injections (IVRs).
Results
In the CRT ≤375 μm subgroup, IVR frequency was significantly lower with SGLT2i vs SU (mean 0.5 ± 1.0 vs 4.3 ± 0.8; P = 0.005, adjusted ancova. Cohen's d of 2.1). The cumulative mean number of IVRs was significantly lower in the SGLT2i group than in the SU group (P = 0.004, hazard ratio 0.29, 95% confidence interval 0.12–0.67).
In the BCVA ≤ 0.155 subgroup, whose mean CRT exceeded 400 μm, no significant difference in IVR frequency was observed between the SGLT2i‐treated and SU‐treated patients. In addition, no intergroup differences were found in patients outside the mild subgroups.
Conclusions
In this exploratory analysis, SGLT2 inhibition was associated with fewer intravitreal injections in patients meeting the CRT‐based mild criteria. These findings are hypothesis‐generating and warrant confirmation in future prospective studies.
Keywords: Anti‐VEGF injection, Diabetic macular edema, Sodium–glucose cotransporter 2 inhibitor
Combining oral sodium‐glucose cotransporter 2 inhibitors with reduced intravitreal injections of anti‐vascular endothelial growth factor showed potential efficacy for mild diabetic macular edema.
INTRODUCTION
The prevalence of proliferative retinopathy and the incidence of blindness caused by diabetes have been decreasing in recent years. 1 Diabetic macular edema (DMO) occurs at any stage of diabetic retinopathy and threatens the patients' quality of life by reducing the ability to read or hindering the acquisition of a driver's license. Therefore, the need for DMO treatment is becoming increasingly important. Currently, the first‐line treatment for DMO is intravitreal injection of anti‐vascular endothelial growth factor (VEGF) agents. 2 , 3 However, repeated administration increases drug costs, burdens the patient with frequent hospital visits, and leads to insufficient effectiveness in 30–60% of cases. 4 Furthermore, no consensus has been reached regarding the treatment methods for early or mild DMO. 5
In clinical practice, we encountered cases where sodium–glucose cotransporter 2 inhibitors (SGLT2i), one of the glucose‐lowering drugs, showed remarkable efficacy in DMO, 6 and we have focused on investigating the usefulness of SGLT2i in the treatment of DMO. A retrospective study of 25 eyes demonstrated the improvement effect of SGLT2i on central retinal thickness (CRT) in DMO eyes. 7 In database studies, the use of SGLT2i reduced the frequency of anti‐VEGF and steroid injections, as well as the number of vitreous surgeries. 8 , 9 Moreover, in our COMET trial, an interventional study using SGLT2i for DMO treatment, 10 the primary evaluation criterion—the number of additional anti‐VEGF administrations up to 48 weeks—showed no significant difference between the SGLT2i and sulfonylurea (SU) groups. However, significantly more patients in the SGLT2i group achieved the desired treatment effects by the third intravitreal ranibizumab injection (IVR), including the initial IVR, and the rate of fourth reinjections was significantly lower in the SGLT2i group. In addition, for the fellow eye with mild DMO, the number of IVRs was significantly fewer in the SGLT2i group. 11 , 12 These results suggest potential efficacy of SGLT2i in early stage or less severe DMO cases.
In this study, we performed a comparative analysis of relatively mild cases from the COMET trial 10 , 11 to explore the usefulness of SGLT2i as a noninvasive treatment method for early or mild DMO cases. The term ‘mild DMO’ in this study refers to an exploratory subgroup defined by baseline CRT, rather than a formal clinical classification.
MATERIALS AND METHODS
The trial protocol and statistical analysis plan were also provided in supporting material.
Ethical approval and registration
The COMET trial was approved by Chiba University Hospital's clinical research review committee (approval number: 5‐380) and registered in the University Hospital Medical Information Network Clinical Trial Registry (UMIN000033961) and the Japan Registry of Clinical Trials (jRCTs031180210). Written informed consent was obtained from all participants at the time of the COMET trial enrollment.
This study, a post hoc analysis of the COMET trial, was approved by the Institutional Review Board of the Chiba University Graduates School of Medicine (HK202408‐01 approved September 25, 2024) and registered in the University Hospital Medical Information Network Clinical Trial Registry (UMIN000057674). Participants were offered the opportunity to opt out and did not receive any compensation.
Study design
This post hoc subgroup analysis focused on mild DMO cases based on the data obtained from the COMET trial in Japanese patients. The protocol and results of the COMET trial have already been reported in previous studies. 10 , 11 The study included patients with type 2 diabetes and DMO eligible for anti‐VEGF therapy. The participants were allocated to two groups at a 1:1 ratio: the SGLT2i group, those receiving luseogliflozin as the study agent, and the SU group, those receiving glimepiride as the control agent. Glycated hemoglobin, urinary albumin, and BCVA were used as the allocation factors. At baseline, all participants underwent a comprehensive ophthalmic examination by retina specialists, including assessment of diabetic retinopathy severity and documentation of prior ocular treatments.
All participants received an initial IVR, followed by additional IVR administrations every 4 weeks up to 48 weeks according to the following protocol: (1) optical coherence tomography (OCT)‐measured CRT >350 μm; (2) OCT‐measured CRT increase by >100 μm from the reference CRT obtained post‐enrollment; and (3) OCT‐measured CRT increase relative to the previous observation point and BCVA increase by 0.2 (logMAR) from that obtained post‐enrollment. CRT measurements were performed using OCT machines manufactured by Carl Zeiss AG (Oberkochen, Germany) or Heidelberg Engineering GmbH (Heidelberg, Germany). No calibration was performed between the models. In cases of bilateral DMO, the more severely affected eye was selected as the study eye, whereas the fellow eye received usual medical care. For this exploratory subgroup analysis, mild DMO was defined using baseline CRT <375 μm and BCVA <0.155 logMAR. These thresholds were not prespecified in the original COMET protocol but were selected as clinically plausible cutoffs within the range generally considered mild or early‐stage DMO (approximately 400–450 μm). The fellow eye findings from the COMET trial supported the plausibility of these thresholds but were not used to derive them. 11 , 12 These mild groups were compared with the remaining severe cases that were not included in each respective subgroup. Because the COMET trial required baseline CRT ≥350 μm for enrollment, no patients with CRT <350 μm were included.
Outcomes
The primary outcome was the difference in the number of intravitreal anti‐VEGF injections between groups through week 48, including the first injection. Secondary outcomes included changes in CRT and BCVA.
Data collection and analysis
The full analysis set (FAS) obtained from the COMET trial included the following data: age, sex, physical examination findings, medical and ophthalmologic history, treatment history, interventions performed during the study period, and longitudinal blood, urine, and fundus imaging parameters. Primary and secondary endpoints were analyzed using the FAS, excluding patients with severe protocol violations. All subgroup analyses were performed according to the intention‐to‐treat principle, with patients analyzed in their originally assigned treatment groups. Baseline characteristics are summarized as frequencies/proportions for categorical data and summary statistics for continuous data. Categorical data were compared using Pearson's Chi‐square or Fisher's exact test as appropriate, whereas continuous data were compared using Welch's t‐test. Missing data for nonvisits were imputed using the last observation carried forward method. To evaluate repeated intravitreal ranibizumab injections, we conducted a recurrent‐event analysis using the Andersen–Gill extension of the Cox proportional hazards model. Each injection was treated as an independent recurrent event, and patients contributed multiple at‐risk intervals defined by the time elapsed since the initial injection (day 0). Because repeated events within the same patient may be correlated, we applied robust (sandwich) variance estimators to account for intraindividual correlation. Event order was not stratified, consistent with the independent‐increment assumption of the Andersen–Gill model. The proportional hazards assumption was assessed using log–log survival plots and Schoenfeld residuals. In the subgroup with baseline CRT <375 μm, where a significant treatment effect was detected, the log–log curves were approximately parallel, supporting the proportional hazards assumption. Hazard ratios represent the relative rate of reinjection over time, and cumulative event curves are presented for visualization only. A sensitivity analysis was additionally performed using alternative thresholds based on the mean baseline CRT (450 μm) and BCVA (0.275 logMAR) of the study population. These cutoffs were selected as independent, non‐data‐driven reference points to evaluate the robustness of the subgroup findings. Statistical significance was set at a P‐value <0.05. Statistical analysis was performed by a third‐party entity using SAS version 9.4. (SAS Institute, Cary, NC) under the supervision of independent biostatisticians.
RESULTS
Study population
The FAS dataset of the COMET trial included 29 patients in the SGLT2i group and 30 patients in the SU group. One patient in the SGLT2i group discontinued visits. All the 59 patients received at least one additional anti‐VEGF treatment. Of these, 27 patients from each group, who were followed for 48 weeks and whom endpoint data were available, were included in the analysis. In the SGLT2i group, one patient was excluded from the subgroup analysis of CRT and BCVA because data on CRT and BCVA at week 0 were missing. In the SU group, one patient was excluded from the subgroup analysis of BCVA because data on BCVA at week 0 were missing. The subgroup with a baseline CRT of ≤375 μm included six patients from the SGLT2i group and four from the SU group, while the subgroup with a baseline CRT of >375 μm included 22 patients from the SGLT2i group and 26 from the SU group. The subgroup with BCVA of <0.155 (logMAR) included 14 patients from the SGLT2i group and 13 from the SU group while the subgroup with BCVA of ≥0.155 (logMAR) included 14 patients from the SGLT2i group and 16 from the SU group (Figure 1). The background characteristics for patients stratified by CRT are shown in Table 1, and for those stratified by BCVA are shown in Table 2. Across the four subgroups, no significant differences were observed in baseline patient background characteristics—internal medical data, including body mass index, glycated hemoglobin, and estimated glomerular filtration rate, or fundus treatment history or status. HbA1c reduction from baseline to week 48 was comparable between groups. Renal function was monitored throughout the study, and although modest changes were observed over 48 weeks, the trajectory did not differ between groups. No differences in adverse events were observed, and no systemic safety signals were identified that could explain the difference in injection frequency.
Figure 1.
CONSORT flow diagram for patient selection. Using the FAS dataset obtained from the COMET trial, patients were divided into four groups based on CRT and BCVA. BCVA, best‐corrected visual acuity; CRT, central retinal thickness; DMO, diabetic macular edema; FAS, full analysis set; SGLT2i, sodium–glucose cotransporter 2 inhibitors; SU, sulfonylurea.
Table 1.
The background characteristics for patients stratified by CRT
| CRT < 375 μm | CRT ≥ 375 μm | |||||
|---|---|---|---|---|---|---|
| SGLT2i | SU | P‐value | SGLT2i | SU | P‐value | |
| N | 6 | 4 | 22 | 26 | ||
| Age (years) | 67.0 ± 10.3 | 59.0 ± 12.4 | 0.33 | 64.3 ± 10.3 | 63.0 ± 9.3 | 0.64 |
| Sex | 1.00 | 0.77 | ||||
| Male | 5 | 3 | 14 | 15 | ||
| Female | 1 | 1 | 8 | 11 | ||
| Diabetes | 6 | 4 | 22 | 26 | ||
| Duration of diabetes (years) | 16.7 ± 10.5 | 8.9 ± 11.2 | 0.31 | 13.9 ± 10.4 | 13.3 ± 10.8 | 0.86 |
| Insulin | 3 | 2 | 1.00 | 7 | 6 | 0.53 |
| GLP‐1 RA | 0 | 1 | 0.40 | 2 | 2 | 1.00 |
| DPP4i | 5 | 3 | 1.00 | 16 | 19 | 1.00 |
| Metformin | 5 | 3 | 1.00 | 15 | 21 | 0.34 |
| Thiazolidine | 0 | 0 | – | 0 | 1 | 1.00 |
| Glinide | 0 | 0 | – | 0 | 1 | 1.00 |
| Alpha GI | 2 | 1 | 1.00 | 8 | 6 | 0.36 |
| Hypertension | 2 | 2 | 1.00 | 15 | 18 | 1.00 |
| Antihypertensive | 2 | 2 | 1.00 | 14 | 14 | 0.57 |
| RAS inhibitors | 2 | 2 | 1.00 | 11 | 11 | 0.77 |
| Diuretic | 0 | 0 | – | 3 | 1 | 0.32 |
| Dyslipidemia | 4 | 2 | 1.00 | 15 | 18 | 1.00 |
| Lipid‐lowering | 4 | 2 | 1.00 | 14 | 14 | 0.57 |
| Cardiovascular disease | ||||||
| Stroke | 0 | 0 | – | 0 | 0 | – |
| Myocardial infarction | 0 | 0 | – | 0 | 0 | – |
| Chronic heart failure | 0 | 1 | 0.40 | 2 | 0 | 0.20 |
| Chronic kidney disease | 0 | 0 | – | 2 | 1 | 0.59 |
| Antiplatelet | 0 | 0 | – | 5 | 0 | 0.02 |
| Anticoagulant | 0 | 0 | – | 3 | 1 | 0.32 |
| Diabetes eye | ||||||
| Diabetic retinopathy | 6 | 4 | 1.00 | 21 | 26 | 0.44 |
| Non‐proliferative retinopathy | 5 | 4 | 16 | 23 | ||
| Proliferative retinopathy | 1 | 0 | 5 | 3 | ||
| Treatment | ||||||
| Panretinal photocoagulation | 2 | 1 | 1.00 | 10 | 8 | 0.37 |
| Diabetic macular edema | ||||||
| Target eye | 6 | 4 | – | 22 | 26 | – |
| Spongy edema | 5 | 3 | 1.00 | 18 | 17 | 0.33 |
| Cystoid edema | 3 | 3 | 0.57 | 15 | 15 | 0.56 |
| Serous detachment | 0 | 1 | 0.40 | 3 | 8 | 0.19 |
| Fellow eye | 4 | 1 | 0.52 | 17 | 18 | 1.00 |
| Spongy edema | 2 | 1 | 1.00 | 13 | 12 | 0.71 |
| Cystoid edema | 2 | 0 | 1.00 | 4 | 9 | 0.16 |
| Serous detachment | 0 | 0 | – | 3 | 3 | 1.00 |
| Treatment | ||||||
| Anti‐VEGF agent | 3 | 3 | 0.57 | 12 | 13 | 0.78 |
| Steroid (triamcinolone acetonide) | ||||||
| Sub‐Tenon's | 0 | 0 | – | 8 | 6 | 0.36 |
| Intravitreal | 0 | 0 | – | 1 | 1 | 1.00 |
| Retinal arteriole aneurysm direct photocoagulation | 2 | 1 | 1.00 | 3 | 9 | 0.18 |
| Macular subthreshold laser | 0 | 1 | 0.40 | 1 | 0 | 0.46 |
| Photocoagulation | 3 | 2 | 1.00 | 11 | 12 | 1.00 |
| Cataract | 4 | 2 | 1.00 | 12 | 17 | 0.56 |
| Post‐surgery | 2 | 1 | 1.00 | 8 | 8 | 0.76 |
| Glaucoma | 1 | 0 | 1.00 | 1 | 1 | 1.00 |
| Under eye drop treatment | 0 | 0 | – | 1 | 0 | 0.46 |
| BMI (kg/m2) | 23.3 ± 2.0 | 24.5 ± 6.8 | 0.76 | 24.4 ± 3.6 | 25.9 ± 3.4 | 0.16 |
| SBP (mmHg) | 146.7 ± 12.3 | 146.3 ± 14.9 | 0.96 | 139.6 ± 19.8 | 141.0 ± 12.7 | 0.78 |
| DBP (mmHg) | 84.5 ± 14.9 | 85.8 ± 12.1 | 0.89 | 78.6 ± 13.3 | 76.7 ± 12.6 | 0.62 |
| Pulse (bpm) | 79.8 ± 9.4 | 84.0 ± 17.3 | 0.73 | 74.9 ± 14.0 | 77.5 ± 16.3 | 0.62 |
| HbA1c (%) | 8.1 ± 0.6 | 8.0 ± 1.2 | 0.86 | 7.6 ± 1.1 | 7.6 ± 1.0 | 0.98 |
| Fasting BG (mg/dL) | 189.0 ± 45.1 | 222.3 ± 40.4 | 0.26 | 156.8 ± 33.6 | 157.1 ± 42.2 | 0.98 |
| eGFR (mL/min/1.73 m2) | 72.1 ± 13.0 | 89.8 ± 21.2 | 0.20 | 66.9 ± 18.3 | 75.4 ± 17.2 | 0.11 |
| U‐Alb (mg/g. Cre) | 60.1 ± 64.0 | 151.4 ± 252.5 | 477.0 ± 821.9 | 227.0 ± 346.0 | ||
| (Logarithmic conversion value) | 3.37 ± 1.60 | 3.92 ± 1.63 | 0.61 | 4.58 ± 2.01 | 4.49 ± 1.41 | 0.86 |
| Hct (%) | 43.2 ± 2.3 | 43.0 ± 2.1 | 0.87 | 40.7 ± 4.4 | 40.7 ± 3.6 | 0.96 |
| AST (IU/L) | 21.5 ± 4.5 | 23.3 ± 6.8 | 0.67 | 26.1 ± 12.2 | 23.7 ± 7.1 | 0.43 |
| ALT (IU/L) | 24.3 ± 7.9 | 33.8 ± 18.4 | 0.39 | 26.2 ± 14.4 | 25.6 ± 11.6 | 0.87 |
| γGTP (IU/L) | 21.3 ± 8.6 | 52.3 ± 33.5 | 39.0 ± 46.2 | 29.9 ± 20.7 | ||
| (Logarithmic conversion value) | 2.97 ± 0.49 | 3.76 ± 0.78 | 0.14 | 3.33 ± 0.73 | 3.25 ± 0.51 | 0.68 |
Patient characteristics were compared using Pearson's Chi‐square or Fisher's exact test for categorical variables and Welch's t‐test for continuous variables. Data are presented as mean ± standard deviation. Data were compared between the two groups using the one‐sample t‐test. Urinary albumin levels were compared using the logarithmic conversion value. Alpha GI, alpha glucosidase inhibitor; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BG, blood glucose; BMI, body mass index; CRT, central retinal thickness; DBP, diastolic blood pressure; DPP4i, dipeptidyl peptidase‐4 inhibitor; eGFR, estimated glomerular infiltration rate; GLP‐1 RA, glucagon‐like peptide‐1 receptor agonist; GTP, glutamyl transpeptidase; HbA1c, glycated hemoglobin; Hct, hematocrit; RAS, renin‐ angiotensin system; SBP, systolic blood pressure; SGLT2i, sodium–glucose cotransporter 2 inhibitor; SU, sulfonylurea; U‐Alb, urinary albumin; VEGF, vascular endothelial growth factor.
Table 2.
The background characteristics for patients stratified by BCVA
| BCVA <0.155 | BCVA ≥0.155 | |||||
|---|---|---|---|---|---|---|
| SGLT2i | SU | P‐value | SGLT2i | SU | P‐value | |
| N | 14 | 13 | 14 | 16 | ||
| Age (years) | 62.4 ± 10.4 | 59.8 ± 11.7 | 0.55 | 67.5 ± 9.5 | 64.2 ± 7.5 | 0.30 |
| Sex | 1.00 | 0.12 | ||||
| Male | 8 | 8 | 12 | 9 | ||
| Female | 6 | 5 | 2 | 7 | ||
| Diabetes | 14 | 13 | 14 | 16 | ||
| Duration of diabetes (years) | 12.9 ± 10.5 | 11.2 ± 10.6 | 0.68 | 17.2 ± 10.4 | 14.2 ± 11.4 | 0.49 |
| Insulin | 5 | 4 | 1.00 | 5 | 4 | 0.69 |
| GLP‐1 RA | 1 | 2 | 0.60 | 1 | 1 | 1.00 |
| DPP4i | 10 | 8 | 0.69 | 12 | 13 | 1.00 |
| Metformin | 12 | 10 | 0.65 | 9 | 13 | 0.42 |
| Thiazolidine | 0 | 0 | – | 0 | 1 | 1.00 |
| Glinide | 1 | 0 | 1.00 | 0 | 1 | 1.00 |
| Alpha GI | 3 | 3 | 1.00 | 8 | 4 | 0.13 |
| Hypertension | 8 | 10 | 0.42 | 9 | 10 | 1.00 |
| Antihypertensive | 7 | 7 | 1.00 | 9 | 9 | 0.72 |
| RAS inhibitors | 7 | 6 | 1.00 | 6 | 7 | 1.00 |
| Diuretic | 1 | 0 | 1.00 | 2 | 1 | 0.59 |
| Dyslipidemia | 9 | 9 | 1.00 | 10 | 11 | 1.00 |
| Lipid‐lowering | 8 | 7 | 1.00 | 10 | 9 | 0.47 |
| Cardiovascular disease | ||||||
| Stroke | 0 | 0 | – | 0 | 0 | – |
| Myocardial infarction | 0 | 0 | – | 0 | 0 | – |
| Chronic heart failure | 0 | 1 | 0.48 | 2 | 0 | 0.21 |
| Chronic kidney disease | 1 | 0 | 1.00 | 1 | 1 | 1.00 |
| Antiplatelet | 1 | 0 | 1.00 | 4 | 0 | 0.04 |
| Anticoagulant | 1 | 1 | 1.00 | 2 | 0 | 0.21 |
| Diabetes eye | ||||||
| Diabetic retinopathy | 14 | 13 | 0.33 | 13 | 16 | 0.57 |
| Non‐proliferative retinopathy | 10 | 12 | 11 | 15 | ||
| Proliferative retinopathy | 4 | 1 | 2 | 1 | ||
| Treatment | ||||||
| Panretinal photocoagulation | 5 | 3 | 0.68 | 7 | 6 | 0.71 |
| Diabetic macular edema | ||||||
| Target eye | 13 | 13 | – | 14 | 16 | – |
| Spongy edema | 10 | 7 | 0.41 | 12 | 12 | 0.66 |
| Cystoid edema | 7 | 8 | 1.00 | 10 | 9 | 0.47 |
| Serous detachment | 1 | 4 | 0.32 | 2 | 5 | 0.40 |
| Fellow eye | 8 | 10 | 0.17 | 12 | 8 | 0.11 |
| Spongy edema | 4 | 6 | 1.00 | 10 | 6 | 1.00 |
| Cystoid edema | 4 | 5 | 1.00 | 2 | 3 | 0.35 |
| Serous detachment | 1 | 1 | 1.00 | 1 | 2 | 0.54 |
| Treatment | ||||||
| Anti‐VEGF agent | 5 | 5 | 1.00 | 9 | 10 | 1.00 |
| Steroid (triamcinolone acetonide) | ||||||
| Sub‐Tenon's | 3 | 2 | 1.00 | 4 | 3 | 0.67 |
| Intravitreal | 1 | 0 | 1.00 | 0 | 0 | – |
| Retinal arteriole aneurysm direct photocoagulation | 3 | 1 | 0.60 | 2 | 8 | 0.06 |
| Macular subthreshold laser | 1 | 0 | 1.00 | 0 | 1 | 1.00 |
| Photocoagulation | 6 | 6 | 1.00 | 8 | 8 | 0.73 |
| Cataract | 8 | 8 | 1.00 | 8 | 11 | 0.71 |
| Post‐surgery | 5 | 4 | 1.00 | 5 | 4 | 0.69 |
| Glaucoma | 1 | 0 | 1.00 | 1 | 1 | 1.00 |
| Under eye drop treatment | 0 | 0 | – | 1 | 0 | 0.47 |
| BMI (kg/m2) | 24.8 ± 3.3 | 25.0 ± 4.6 | 0.88 | 23.8 ± 3.4 | 26.1 ± 3.3 | 0.08 |
| SBP (mmHg) | 142.8 ± 20.6 | 140.8 ± 12.9 | 0.77 | 137.8 ± 19.2 | 140.6 ± 11.5 | 0.63 |
| DBP (mmHg) | 80.6 ± 15.5 | 80.2 ± 13.9 | 0.94 | 77.1 ± 12.7 | 75.3 ± 11.6 | 0.70 |
| Pulse (bpm) | 78.9 ± 9.8 | 85.0 ± 14.3 | 0.28 | 73.2 ± 15.3 | 72.5 ± 16.0 | 0.91 |
| HbA1c (%) | 7.8 ± 1.1 | 7.6 ± 0.9 | 0.66 | 7.6 ± 1.0 | 7.7 ± 1.2 | 0.85 |
| Fasting BG (mg/dL) | 168.7 ± 47.3 | 166.8 ± 52.2 | 0.92 | 157.1 ± 30.9 | 167.6 ± 46.8 | 0.49 |
| eGFR (mL/min/1.73m2) | 70.6 ± 20.7 | 78.4 ± 16.5 | 0.30 | 65.8 ± 13.6 | 76.8 ± 20.2 | 0.09 |
| U‐Alb (mg/g. Cre) | 236.1 ± 429.9 | 158.0 ± 252.0 | 516.1 ± 946.7 | 275.0 ± 392.9 | ||
| (Logarithmic conversion value) | 3.74 ± 2.08 | 4.23 ± 1.25 | 0.48 | 4.65 ± 1.93 | 4.59 ± 1.61 | 0.92 |
| Hct (%) | 41.4 ± 3.2 | 41.5 ± 3.1 | 0.99 | 41.3 ± 4.8 | 40.5 ± 3.8 | 0.61 |
| AST (IU/L) | 25.7 ± 12.1 | 24.8 ± 7.3 | 0.82 | 24.2 ± 10.5 | 21.9 ± 5.7 | 0.47 |
| ALT (IU/L) | 28.3 ± 14.1 | 29.8 ± 15.6 | 0.79 | 23.1 ± 12.3 | 23.0 ± 8.5 | 0.99 |
| γGTP (IU/L) | 29.5 ± 19.0 | 34.9 ± 24.8 | 39.5 ± 55.4 | 32.1 ± 23.4 | ||
| (Logarithmic conversion value) | 3.23 ± 0.58 | 3.35 ± 0.65 | 0.61 | 3.23 ± 0.83 | 3.32 ± 0.51 | 0.74 |
Patient characteristics were compared using Pearson's Chi‐square or Fisher's exact test for categorical endpoints and Welch's t‐test for continuous variables. Data are presented as mean ± standard deviation. Data were compared between the two groups using the one‐sample t‐test. Urinary albumin levels were compared using the logarithmic conversion value. Alpha GI, alpha glucosidase inhibitor; ALT, alanine aminotransferase; AST, aspartate aminotransferase; BCVA, best‐corrected visual acuity; BG, blood glucose; BMI, body mass index; DBP, diastolic blood pressure; DPP4i, dipeptidyl peptidase‐4 inhibitor; eGFR, estimated glomerular infiltration rate; GLP‐1 RA, glucagon‐like peptide‐1 receptor agonist; GTP, glutamyl transpeptidase; HbA1c, glycated hemoglobin; Hct, hematocrit; RAS, renin‐ angiotensin system; SBP, systolic blood pressure; SGLT2i, sodium–glucose cotransporter 2 inhibitor; SU, sulfonylurea; U‐Alb, urinary albumin; VEGF, vascular endothelial growth factor.
Subgroup with baseline CRT of <375 μm
The number of IVR up to 48 weeks, including the first dose, was 2.0 ± 1.1 in the SGLT2i group and 4.5 ± 1.3 in the SU group. Comparison using analysis of covariance, with allocation adjustment factors as covariates, showed that the number of IVR was 0.5 ± 1.0 in the SGLT2i group and 4.3 ± 0.8 in the SU group (P = 0.005, Cohen's d of 2.1). The adjusted mean difference in the number of IVRs between groups was −3.8 injections (95% confidence interval [CI] −5.8 to −1.7). The cumulative mean number of IVRs after the start of the intervention was significantly lower in the SGLT2i group than in the SU group (P = 0.004, hazard ratio [HR] 0.29, 95% CI: 0.12–0.67) (Figure 2a). CRT significantly decreased in the SGLT2i group from a baseline of 363.8 ± 7.5 to 314.0 ± 18.6 μm at week 48 (P = 0.002), whereas no significant change was observed in the SU group (CRT of 362.8 ± 8.9 μm at baseline to 364.8 ± 37.6 μm at week 48, P = 0.92) (Figure 2b). BCVA (logMAR) showed no significant changes in either group, with the SGLT2i group changing from a baseline of 0.13 ± 0.20 to 0.09 ± 0.08 at week 48 (P = 0.61) and the SU group changing from a baseline of 0.09 ± 0.24 to 0.11 ± 0.16 at week 48 (P = 0.80) (Figure 2c).
Figure 2.
Changes in ophthalmological indicators of the subgroup with baseline CRT of <375 μm. The cumulative mean number of IVRs after the start of the intervention was significantly lower in the SGLT2i group than in the SU group (a). CRT significantly decreased in the SGLT2i group, whereas no significant change was observed in the SU group (b). BCVA showed no significant changes in either group (c). Panels b and c show mean ± standard deviation. *P < 0.05 vs. 0 w is the P‐value of a paired comparison test with the 0 w data within the same group. #P < 0.05 vs. SGLT2i is the P‐value of the intergroup comparison test. BCVA, best‐corrected visual acuity; CRT, central retinal thickness; IVR, intravitreal injection of ranibizumab; SGLT2i, sodium–glucose cotransporter 2 inhibitors; SU, sulfonylurea.
Subgroup with baseline CRT of ≥375 μm
The number of IVR up to 48 weeks, including the first dose, was 5.9 ± 3.5 in the SGLT2i group and 6.1 ± 2.9 in the SU group. Comparison using analysis of covariance, with allocation adjustment factors as covariates, revealed no significant differences, with the number of IVR being 5.9 ± 0.8 in the SGLT2i group and 5.7 ± 0.8 in the SU group (P = 0.80). The adjusted mean difference in the number of IVRs between groups was 0.2 injections (95% CI: −1.7 to 2.2). The cumulative mean number of IVRs after the start of the intervention were not significantly different between the two groups (P = 0.80, HR 0.95, 95% CI: 0.65–1.39) (Figure S1a). In the SGLT2i group, CRT significantly decreased from a baseline of 466.6 ± 71.2 to 359.8 ± 57.4 μm at week 48 (P < 0.001). Similarly, in the SU group, CRT significantly decreased from a baseline of 496.5 ± 108.7 to 379.2 ± 104.0 μm at week 48 (P < 0.001). However, the amount of change in CRT was not significantly different between the two groups (P = 0.33) (Figure S1b). BCVA (logMAR) significantly improved in both groups, decreasing from a baseline of 0.32 ± 0.28 to 0.23 ± 0.27 at week 48 in the SGLT2i group (P = 0.004) and from 0.31 ± 0.27 to 0.24 ± 0.21 at week 48 in the SU group (P = 0.018). Similarly, no significant difference was observed in the amount of change in BCVA between the two groups (P = 0.91) (Figure S1c).
Subgroup with BCVA of <0.155 (logMAR)
The number of IVR up to 48 weeks, including the first dose, was 3.5 ± 3.2 in the SGLT2i group and 4.7 ± 2.2 in the SU group. Comparison using analysis of covariance, with allocation adjustment factors as covariates, revealed no significant differences in the number of IVR between the two groups (3.5 ± 0.8 in the SGLT2i group and 4.7 ± 1.1 in the SU group, P = 0.36). The adjusted mean difference in the number of IVRs between groups was −1.1 injections (95% CI −3.6 to 1.4). The cumulative mean number of IVRs after the start of the intervention was not significantly different between the groups (P = 0.39, HR 0.72, 95% CI 0.34–1.51) (Figure S2a). CRT significantly decreased in the SGLT2i group from a baseline of 416.5 ± 74.2 to 346.9 ± 44.5 μm at week 48 (P = 0.003). Similarly, in the SU group, CRT significantly decreased from a baseline of 455.0 ± 117.9 to 411.2 ± 118.8 μm at week 48 (P = 0.043). The two groups showed no significant difference in the amount of change in CRT (P = 0.77) (Figure S2b). BCVA (logMAR) showed no significant changes in either group, with the SGLT2i group changing from a baseline of 0.05 ± 0.06 to 0.04 ± 0.12 at week 48 (P = 0.57) and the SU group changing from a baseline of 0.03 ± 0.08 to 0.05 ± 0.10 at week 48 (P = 0.53) (Figure S2c).
Subgroup with BCVA of ≥0.155 (logMAR)
The number of IVR up to 48 weeks, including the first dose, was 5.9 ± 2.6 in the SGLT2i group and 6.5 ± 3.0 in the SU group. Comparison using analysis of covariance, with allocation adjustment factors as covariates, revealed no significant differences between the groups, with the number of IVR being 5.2 ± 0.9 in the SGLT2i group and 5.8 ± 0.8 in the SU group (P = 0.58). The adjusted mean difference in the number of IVRs between groups was −0.6 injections (95% CI: −2.7 to 1.6). The cumulative mean number of IVRs after the start of the intervention was not significantly different between the two groups (P = 0.45, HR 0.86, 95% CI: 0.59–1.26) (Figure S3a). CRT significantly decreased in the SGLT2i group from a baseline of 464.5 ± 72.0 to 348.6 ± 65.2 μm at week 48 (P = 0.003). Similarly, in the SU group, CRT significantly decreased from a baseline of 493.6 ± 108.1 to 359.8 ± 76.5 μm at week 48 (P < 0.001). The amount of change in CRT was not significantly different between the two groups (P = 0.49) (Figure S3b). BCVA (logMAR) significantly improved in both groups, decreasing from a baseline of 0.49 ± 0.21 to 0.34 ± 0.23 at week 48 in the SGLT2i group (P = 0.002) and from 0.48 ± 0.19 to 0.34 ± 0.17 at week 48 in the SU group (P = 0.007). The two groups showed no significant difference in the amount of change in BCVA (P = 0.67) (Figure S3c).
Proportional hazards diagnostics
The proportional hazards assumption was evaluated using log–log survival plots. In the subgroup with baseline CRT <375 μm, where a significant treatment effect was observed, the curves for the two treatment groups were approximately parallel (Figure S4A). In the other subgroups, the curves overlapped or crossed slightly, consistent with the absence of a treatment effect (Figure S4B–D).
Sensitivity analysis using alternative cutoffs
To assess the robustness of the subgroup findings, additional analyses were performed using independent thresholds based on the mean baseline values of the study population (CRT 450 μm and BCVA 0.275 logMAR). In the subgroup with baseline CRT <450 μm (SGLT2i: n = 17; SU: n = 13), the adjusted mean number of IVR injections through week 48 was 3.5 vs. 5.0, with an adjusted mean difference of −1.5 injections (95% CI: −3.7 to 0.7; P = 0.18). In the subgroup with CRT ≥450 μm (SGLT2i: n = 9; SU: n = 14), the adjusted mean number of injections was 7.6 vs. 5.8, with an adjusted mean difference of +1.8 injections (95% CI: −0.5 to 4.0; P = 0.12). In the subgroup with baseline BCVA <0.275 (SGLT2i: n = 14; SU: n = 12), the adjusted mean number of injections was 3.5 vs. 4.6, with an adjusted mean difference of −1.1 injections (95% CI: −3.3 to 1.2; P = 0.33). In the subgroup with BCVA ≥0.275 (SGLT2i: n = 12; SU: n = 14), the adjusted mean number of injections was 5.1 vs. 6.2, with an adjusted mean difference of −1.1 injections (95% CI: −3.2 to 1.1; P = 0.31). Across all alternative thresholds, the direction of the treatment effect consistently favored the SGLT2 inhibitor group, although statistical significance was not reached due to limited sample size. These findings support the robustness of the primary subgroup results.
DISCUSSION
In this study, the number of anti‐VEGF injections in the mild group with CRT of <375 μm at the start of treatment was significantly reduced by the administration of SGLT2i. In contrast, for cases with CRT of ≥375 μm, as well as those with either good or poor BCVA, the number of IVR injections did not change. Among patients with CRT < 375 μm, the baseline visual acuity was favorable, with 0.13 (logMAR) for the SGLT2i group and 0.09 (logMAR) for the SU group. However, among patients with BCVA <0.155 (logMAR), the baseline CRT was over 400 μm in both groups, indicating that favorable BCVA did not necessarily align with mild edema. Our results show that the subgroup with CRT < 375 μm represented a milder group compared with the subgroup with BCVA <0.155 (logMAR). These findings are consistent with existing knowledge that CRT and visual acuity in DMO do not always correlate. 13 The reduction in the number of IVR injections due to SGLT2i depends on baseline CRT. This finding is consistent with previous results for fellow eyes with mild edema in the COMET trial, emphasizing the utility of SGLT2i in cases of mild or early‐stage DMO 11 , 12 while contrasting with the outcomes of our retrospective study on chronically advanced severe cases. 6 , 7 On the other hand, in the subgroups with CRT < 375 μm and BCVA <0.155 (logMAR), both groups had good BCVA at baseline, thus BCVA was not affected at 48 weeks. Furthermore, in the subgroups with CRT ≥375 μm and BCVA ≥0.155 (logMAR), BCVA recovered from baseline in both groups by repeating IVR, which further suggests the importance of anti‐VEGF therapy as a basic treatment. Although the systemic effect of IVR on the fellow eyes cannot be completely ignored, its impact on the present results is thought to be limited because ranibizumab has less effect on the whole body. 14 The target eye was treated first, and the number of IVR sessions for fellow eyes in the COMET trial was relatively low (<4 sessions) and significantly lower in the SGLT2i group.
Anti‐VEGF therapy is not necessarily recommended for patients with good vision and mild edema. Patients with visual acuity of <0.1 (logMAR) require follow‐up observations, as reported in a previous study that showed no change in visual prognosis after observation for over 2 years. 5 In contrast, patients with significant fluctuations in central foveal retinal thickness have a poor visual prognosis. 15 In such cases, regardless of BCVA, clinicians tend to choose early treatment with anti‐VEGF therapy. In early DMO, SGLT2i administration can help maintain CRT stability, consistent with the COMET trial findings, in which fellow eyes—those with milder disease—showed a significantly reduced number of anti‐VEGF injections in the SGLT2i group. 11 , 12 The present study, based on protocol‐driven analysis, showed similar results, even in target eyes with mild disease. Moreover, although evidence is currently limited to animal models, early administration of SGLT2i is reported to suppress initial pathological changes in diabetic retinopathy, further supporting its potential role in modifying disease progression. 16 Therefore, it is not appropriate to uniformly choose observation for all cases with good vision, and early therapeutic intervention may be warranted based on worsening CRT. However, considering the cost, mode of administration, and evidence from past treatments using the current first‐line anti‐VEGF drug, no consensus has been established with regard to anti‐VEGF injection at the stage of good vision and mild CRT.
The robustness of this trend was further supported by sensitivity analyses using independent, mean‐based thresholds (CRT 450 μm and BCVA 0.275 logMAR). Although statistical significance was not reached due to the limited sample size, the direction of the treatment effect consistently favored the SGLT2 inhibitor group in eyes with relatively mild anatomical involvement. These findings indicate that the observed reduction in injection burden is unlikely to be an artifact of the specific subgroup cutoffs or their proximity to the retreatment criteria, and instead reflect a reproducible pattern across multiple clinically plausible thresholds.
Diabetic retinopathy and DMO target both blood vessels and neuronal cells. 17 Apoptosis of neuronal cells in the retina starts from the early stages of diabetes. However, current treatments for DMO are administered when DMO has already developed and progressed to a certain degree 18 , 19 and irreversible changes in neuronal cells have occurred. Under such circumstances, our study suggests that the combination of oral SGLT2i with fewer intravitreal injections of anti‐VEGF drugs is effective in achieving therapeutic effects at the early stage of DMO. The effectiveness of SGLT2i in treating DMO can be attributed to several mechanisms. From a systemic perspective, SGLT2i contribute to improving blood glucose levels, blood pressure, weight reduction, fluid retention, and renal protection. 20 , 21 Locally within the eye, SGLT2 has been detected in the retinal vascular plexus, external limiting membrane, retinal pigment epithelium, and microglia. 16 In the early stages of diabetes, SGLT2i suppress the swelling and loss of pericytes 22 while inhibiting microglial activation and VEGF production, thereby exerting vascular and neuroprotective effects. 16 , 23 Furthermore, SGLT2i have been shown to suppress mTOR activation during the proliferative stages of diabetic retinopathy and reduce oxidative stress. 24 These mechanisms collectively suggest that SGLT2i provide significant protective effects against early and mild diabetic retinopathy and DMO.
This study has a few limitations. This subgroup analysis was purely exploratory, with no statistical power calculation to determine the ideal sample size. The small number of cases with CRT < 375 μm may have exaggerated the observed differences, whereas the relatively short follow‐up period of 48 weeks may have diminished the differences. The effect size (Cohen's d = 2.1) indicated a large magnitude of difference, although the subgroup size was small. These factors could have introduced bias into the study results. The reduction in HbA1c during the study period did not differ between groups, indicating that glycemic control was unlikely to account for the observed differences in injection frequency. Adjusting for HbA1c change was therefore not included to avoid overadjustment. In addition, due to ethical considerations, no dedicated period was established to observe the therapeutic effects of SGLT2i alone. Despite these limitations, the findings of this sub‐analysis highlight the potential of oral administration of SGLT2i as a less invasive treatment option for DMO. However, these findings underscore the need for future placebo‐controlled trials specifically focused on mild DMO.
In summary, the administration of SGLT2i significantly reduced the number of IVR in the subgroup with CRT of <375 μm at the start of treatment. These exploratory findings suggest that SGLT2 inhibition may reduce treatment burden in selected mild cases, although confirmation in dedicated prospective trials is required.
DISCLOSURE
R.I. received research funding from Taisho Pharmaceutical Co., Ltd.; Astellas Pharma, Inc.; Otsuka Pharmaceutical Co., Ltd.; AstraZeneca K.K.; and Novo Nordisk Pharma Ltd. M.K. and K.Y. received research funding from Astellas Pharma, Inc. and Taisho Pharmaceutical Co., Ltd. T.T. and S.Y. received research funding from Bayer Yakuhin Ltd. None of the funding agencies played a role in the study design, data collection and analysis, decision to publish, or manuscript preparation.
Approval of the research protocol: This study, a post hoc analysis of the COMET trial, was approved by the Institutional Review Board of the Chiba University Graduates School of Medicine (HK202408‐01 approved September 25, 2024).
Informed consent: Written informed consent was obtained from all participants at the time of the COMET trial enrollment, were offered the opportunity to opt out, and did not receive any compensation.
Registry and the registration no. of the study: Registered in the University Hospital Medical Information Network Clinical Trial Registry (UMIN000057674 approved April 25, 2025).
Animal studies: N/A.
Supporting information
Figure S1 Changes in ophthalmological indicators of the subgroup with baseline CRT of ≥375 μm.
Figure S2 Changes in ophthalmological indicators of the subgroup with baseline BCVA of <0.155 (logMAR).
Figure S3 Changes in ophthalmological indicators of the subgroup with baseline BCVA of ≥0.155 (logMAR).
Figure S4 Log–log survival plots for recurrent intravitreal injections.
ACKNOWLEDGMENTS
The authors would like to thank the patients for their participation in this study. We would like to thank Soiken Inc. and EviPRO Co., Ltd. for managing this study and Editage (www.editage.com) for English language editing. This study was supported by Taisho Pharmaceutical Co., Ltd. The sponsor did not influence the study design, data collection and analysis, publication decision, or manuscript preparation.
Clinical Trial Registry
University Hospital Medical Information Network Clinical Trial Registry
UMIN000057674.
DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.
References
- 1. Morizane Y, Morimoto N, Fujiwara A, et al. Incidence and causes of visual impairment in Japan: The first nation‐wide complete enumeration survey of newly certified visually impaired individuals. Jpn J Ophthalmol 2019; 63: 26–33. [DOI] [PubMed] [Google Scholar]
- 2. Diabetic Retinopathy Clinical Research Network , Wells JA, Glassman AR, et al. Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema. N Engl J Med 2015; 372: 1193–1203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Yoshida S, Murakami T, Nozaki M, et al. Review of clinical studies and recommendation for a therapeutic flow chart for diabetic macular edema. Graefes Arch Clin Exp Ophthalmol 2021; 259: 815–836. [DOI] [PubMed] [Google Scholar]
- 4. Bressler NM, Beaulieu WT, Glassman AR, et al. Persistent macular thickening following intravitreous aflibercept, bevacizumab, or ranibizumab for central‐involved diabetic macular edema with vision impairment: A secondary analysis of a randomized clinical trial. JAMA Ophthalmol 2018; 136: 257–269. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Baker CW, Glassman AR, Beaulieu WT, et al. Effect of initial management with aflibercept vs laser photocoagulation vs observation on vision loss among patients with diabetic macular edema involving the center of the macula and good visual acuity: A randomized clinical trial. JAMA 2019; 321: 1880–1894. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Takatsuna Y, Ishibashi R, Tatsumi T, et al. Sodium‐glucose cotransporter 2 inhibitors improve chronic diabetic macular edema. Case Rep Ophthalmol Med 2020; 2020: 8867079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Tatsumi T, Oshitari T, Takatsuna Y, et al. Sodium‐glucose co‐transporter 2 inhibitors reduce macular edema in patients with diabetes mellitus. Life 2022; 12: 692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Ishibashi R, Inaba Y, Koshizaka M, et al. Sodium‐glucose co‐transporter 2 inhibitor therapy reduces the administration frequency of anti‐vascular endothelial growth factor agents in patients with diabetic macular oedema with a history of anti‐vascular endothelial growth factor agent use: A cohort study using the Japanese health insurance claims database. Diabetes Obes Metab 2024; 26: 1510–1518. [DOI] [PubMed] [Google Scholar]
- 9. Ishibashi R, Koshizaka M, Takatsuna Y, et al. Sodium‐glucose cotransporter 2 inhibitor therapy reduces the administration frequency of steroid injection in patients with diabetic macular edema: A cohort study using the Japanese health insurance claims database. J Diabetes Investig 2024; 15: 1231–1238. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Ishibashi R, Takatsuna Y, Koshizaka M, et al. Safety and efficacy of ranibizumab and luseogliflozin combination therapy in patients with diabetic macular edema: Protocol for a multicenter, open‐label randomized controlled trial. Diabetes Ther 2020; 11: 1891–1905. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Ishibashi R, Takatsuna Y, Koshizaka M, et al. Ranibizumab with luseogliflozin in type 2 diabetes with diabetic macular oedema: A randomised clinical trial. Diabetes Obes Metab 2025; 27: 2473–2484. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Tatsumi T, Oshitari T, Takatsuna Y, et al. Sodium‐glucose co‐transporter 2 inhibitors reduce the intravitreal anti‐VEGF treatment necessity in eyes with early diabetic macular oedema: A post hoc analysis focusing on the fellow eye of the COMET trial. Diabetes Obes Metab 2025; 27: 5812–5821. [DOI] [PubMed] [Google Scholar]
- 13. Wang P, Hu Z, Hou M, et al. Relationship between macular thickness and visual acuity in the treatment of diabetic macular edema with anti‐VEGF therapy: Systematic review. J Vitreoretin Dis 2023; 7: 57–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Avery RL, Castellarin AA, Steinle NC, et al. Systemic pharmacokinetics and pharmacodynamics of intravitreal aflibercept, bevacizumab, and ranibizumab. Retina 2017; 37: 1847–1858. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Starr MR, Salabati M, Mahmoudzadeh R, et al. Fluctuations in central subfield thickness associated with worse visual outcomes in patients with diabetic macular edema in clinical trial setting. Am J Ophthalmol 2021; 232: 90–97. [DOI] [PubMed] [Google Scholar]
- 16. Yamato M, Kato N, Yamada KI, et al. The early pathogenesis of diabetic retinopathy and its attenuation by sodium‐glucose transporter 2 inhibitors. Diabetes 2024; 73: 1153–1166. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Duh EJ, Sun JK, Stitt AW. Diabetic retinopathy: Current understanding, mechanisms, and treatment strategies. JCI Insight 2017; 2: e93751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Barber AJ, Lieth E, Khin SA, et al. Neural apoptosis in the retina during experimental and human diabetes. Early onset and effect of insulin. J Clin Invest 1998; 102: 783–791. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Sohn EH, van Dijk HW, Jiao C, et al. Retinal neurodegeneration may precede microvascular changes characteristic of diabetic retinopathy in diabetes mellitus. Proc Natl Acad Sci U S A 2016; 113: E2655–E2664. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Dailey GE. Empagliflozin: A new treatment option for patients with type 2 diabetes mellitus. Drugs Today 2015; 51: 519–535. [DOI] [PubMed] [Google Scholar]
- 21. Neal B, Perkovic V, Mahaffey KW, et al. Canagliflozin and cardiovascular and renal events in type 2 diabetes. N Engl J Med 2017; 377: 644–657. [DOI] [PubMed] [Google Scholar]
- 22. Wakisaka M, Nagao T. Sodium glucose cotransporter 2 in mesangial cells and retinal pericytes and its implications for diabetic nephropathy and retinopathy. Glycobiology 2017; 27: 691–695. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Hanaguri J, Yokota H, Kushiyama A, et al. The effect of sodium‐dependent glucose cotransporter 2 inhibitor tofogliflozin on neurovascular coupling in the retina in type 2 diabetic mice. Int J Mol Sci 2022; 23: 1362. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Zaini LM, Kartasasmita AS, Gondhowiardjo TD, et al. Potential molecular mechanism of action of sodium‐glucose co‐transporter 2 inhibitors in the prevention and management of diabetic retinopathy. Expert Rev Ophthalmol 2022; 17: 199–210. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1 Changes in ophthalmological indicators of the subgroup with baseline CRT of ≥375 μm.
Figure S2 Changes in ophthalmological indicators of the subgroup with baseline BCVA of <0.155 (logMAR).
Figure S3 Changes in ophthalmological indicators of the subgroup with baseline BCVA of ≥0.155 (logMAR).
Figure S4 Log–log survival plots for recurrent intravitreal injections.
Data Availability Statement
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.