Type 2 diabetes mellitus (T2DM) is a large public health concern in the world and a significant contributor to long-term microvascular morbidity.[1] Diabetic retinopathy (DR) is one of its complications, as worsening retinal vascular damage may progress to diabetic macular edema, proliferative diabetic retinopathy, irreversible visual loss, and blindness.[2] The burden is significant in Pakistan, with a recent systematic review and meta-analysis of 60 studies that included 45,646 T2DM patients reporting a pooled prevalence of DR of 32.9% (95% CI: 27.67–38.40%).[3] There were significant variations among regions; the prevalence estimates ranged from 47.55% in Khyber Pakhtunkhwa to 54.65% in Balochistan.[3]

Oral glucose-lowering drugs are widely used in T2DM, and there are two drugs currently used that have different mechanisms of action: sodium-glucose cotransporter-2 (SGLT2) inhibitors and dipeptidyl peptidase-4 (DPP-4) inhibitors, which might have different effects on retinal outcomes.[4] In addition to glucose reduction, SGLT2 inhibitors also cause osmotic diuresis, slight weight loss, and improvements in blood pressure, metabolic and vascular parameters, which may impact pathways that lead to retinal microvascular injury.[5] DPP-4 inhibitors are metabolically neutral, as they increase insulin secretion by incretins and lower glucagon levels.[6]

The available evidence has raised an important, but unresolved, debate. A systematic review and meta-analysis of 16 randomized trials and 19 real-world studies involving more than 2.3 million patients found that SGLT2 inhibitors were associated with a lower risk of DR incidence, DR progression, and sight-threatening DR in real-world studies, although the randomized-trial evidence did not demonstrate a statistically significant reduction in DR incidence.[7] Likewise, in a large, propensity-matched study of over 3.5 million patients with newly diagnosed T2DM, use of an SGLT2 inhibitor was also found to provide a statistically significantly reduced risk for Sight-Threatening Retinopathy (STRs) compared with use of DPP-4 inhibitors (adjusted HR 0.57, 95% CI 0.51–0.63).[7] In another large comparative-effectiveness study, 371,698 patients were included, and the researchers observed a reduced risk of treatment for sight-threatening retinopathy in SGLT2 inhibitor users compared to DPP-4 inhibitor users (HR 0.79, 95% CI 0.64–0.97).[8]

However, these findings are not entirely consistent. A meta-analysis of RCTs containing 188,463 patients revealed no significant difference between the odds of DR events in those using SGLT2 inhibitors and placebo (pooled OR: 1.02, 95% CI: 0.76–1.37), or between DPP-4 inhibitors and placebo (pooled OR: 1.10, 95% CI: 0.84–1.42).[9]

In a country like Pakistan with a high burden of DR and limited availability of routine screening and advanced DR treatment, this question is especially relevant. Although SGLT2 and DPP-4 inhibitors have been widely adopted, there is a limited body of evidence to directly compare their effect on retinal status and DM progression, especially in South Asian racial groups. A side-by-side comparison might then provide clues to the potential for a difference between these two widely used treatments other than on the level of glycemic control. The current study aimed to compare the retinal outcomes and the course of diabetic retinopathy in T2DM patients who used SGLT2 inhibitors with those who used DPP-4 inhibitors. The purpose was to assess whether there was an advantage in terms of retinal endpoints and reduced risk of DR progression for the SGLT2 inhibitor group versus the DPP-4 inhibitor group, and to evaluate clinical and metabolic factors associated with DR.

A comparative retrospective cohort study was carried out for six months in the Department of Medicine and the Department of Ophthalmology of a tertiary care teaching hospital in Pakistan from October, 2025 to March, 2026. The sample size was calculated using OpenEpi version 3.01 for comparison of two independent proportions, based on previous evidence showing lower risks of sight-threatening retinopathy with SGLT2 inhibitors than DPP-4 inhibitors (adjusted HR 0.57 and HR 0.79, respectively).[10] A sample size of 62 participants per group was the minimum sample size required using a two-sided confidence level of 95% and 80% power. In order to obtain about 20% inexhaustive follow-up or missing ophthalmological documents, the sample was extended to 150 (75 patients per treatment group). A consecutive sampling technique was employed. Patients who were eligible and attended the selected medical and ophthalmology departments during the study period were approached and recruited consecutively until the sample size of 150 patients was reached. Participants were split into two groups based on the drugs documented for their antidiabetic therapy: 75 patients in the SGLT2 inhibitor group and 75 patients in the DPP-4 inhibitor group. The data were collected using the structured proforma which was designed based on the objectives of the study. Demographic data such as sex and age were obtained. The clinical data comprised the duration of diabetes, BMI, blood pressure, smoking status, hypertension, dyslipidaemia, cardiovascular disease, diabetic nephropathy, and other relevant comorbidities. Diabetes-related parameters were baseline and follow-up glycated hemoglobin (HbA1c), fasting blood glucose, duration of antidiabetic therapy, and concomitant glucose-lowering drugs. Data on antidiabetic medications was collected from clinical charts and prescriptions. The type of SGLT2 inhibitor or DPP-4 inhibitor, treatment duration, dose, and other antidiabetic medications were recorded. Ophthalmological evaluation consisted of visual acuity, fundus examination and retinal imaging when available. Ophthalic coherence tomography was used when available and clinically indicated. The diabetic retinopathy was classified as: no retinopathy, mild non-proliferative diabetic retinopathy, moderate non-proliferative diabetic retinopathy, severe non-proliferative diabetic retinopathy and proliferative diabetic retinopathy. Diabetic Macular Edema (DME) was also documented. Retinal progression was considered to be advancement in a higher category of diabetic retinopathy severity or the occurrence of a clinically important retinal complication during the follow-up period. Data collected were entered and analyzed by using IBM SPSS Statistics. All continuous variables were tested for normality, and presented as mean ± standard deviation or median (interquartile range). Frequencies and percentages were used to present categorical variables. Independent-samples t-test, Mann–Whitney U test for continuous variables and chi-square test and Fisher's exact test for categorical variables were used to compare the baseline characteristics between the SGLT2 inhibitor and DPP-4 inhibitor groups. The main outcome was the progression of diabetic retinopathy in the 6-month observation period. Secondary outcomes included the development of diabetic retinopathy (DR) in patients who were not previously diagnosed with DR, deterioration of the severity of DR, the onset or deterioration of diabetic macular edema (DME), and the onset of proliferative diabetic retinopathy (PDR). Numbers of patients with retinal outcomes were compared between the two treatment groups, and odds ratios with 95% confidence intervals were calculated. To identify whether the use of SGLT2 inhibitor therapy was associated with reduced odds of diabetic retinopathy progression after adjusting for clinically relevant confounding factors by multivariable logistic regression analysis, such as baseline severity of diabetic retinopathy, age, sex, duration of diabetes, baseline HbA1c level, hypertension, dyslipidemia, body mass index, renal function, and concomitant antidiabetic therapy. A p-value of <0.05 (two-sided) was regarded as statistically significant.

A total of 150 patients with T2DM were included, 75 patients with SGLT2 inhibitors and 75 with DPP-4 inhibitors. There were no significant differences between the two groups concerning age, sex, duration of diabetes, BMI, blood pressure, smoking status, hypertension, dyslipidaemia, cardiovascular disease, diabetic nephropathy or baseline glycemic parameters (Table 1).

Concomitant glucose-lowering therapy was similar among treatment groups. At follow-up, the mean change in HbA1c and fasting blood glucose levels were not statistically different between groups, but the mean levels of both were lower in the SGLT2 inhibitor group (Table 2).

Ophthalmological findings at baseline were also similar between the groups. The percentage of patients with DR in the SGLT2 inhibitor group was 64.0%, and for the DPP-4 inhibitor group was 68.0%. There was no significant difference between the groups in the distribution of severity of retinopathy, visual acuity, DME, and retinal imaging findings at baseline (Table 3).

Over the 6-month follow-up, 9.3% of those taking SGLT2 inhibitors and 18.7% of those taking DPP-4 inhibitors had developed progression of DR. SGLT2 inhibitor users also had a lower prevalence of worsening by at least one retinopathy category, development or worsening of diabetic macular edema and proliferative diabetic retinopathy, but none of these differences between groups were statistically significant (Table 4).

Patients with progressive retinopathy were significantly older, had a greater duration of diabetes, higher baseline HbA1c, higher frequency of renal impairment, and a more advanced baseline retinopathy. There was no significant relationship between baseline sex, BMI, hypertension, and dyslipidaemia and progression. The association was not statistically significant even though progression was less frequent among those on SGLT2 inhibitors (Table 5).

The association between the use of an SGLT2 inhibitor and lower odds of DR progression was not statistically significant compared to DPP-4 inhibitor therapy (adjusted OR 0.48; 95% CI: 0.20–1.15; p=0.098). Increased odds of progression were independently associated with longer diabetes duration, higher baseline HbA1c, and moderate/severe baseline retinopathy (Table 6).

Table 1. Baseline demographic and clinical characteristics of study participants (n=150)

Table 2. Antidiabetic treatment characteristics and changes in metabolic parameters during follow-up

Table 3. Baseline ophthalmological characteristics of study participants

Table 4. Retinal outcomes and progression of diabetic retinopathy during six-month follow-up

Table 5. Association of clinical and metabolic factors with diabetic retinopathy progression

Table 6. Multivariable logistic regression analysis of factors associated with diabetic retinopathy progression

In this retrospective comparative cohort study, the risk of progression of DR in SGLT2 inhibitor users was lower than in DPP-4 inhibitor users during the six-month follow-up (9.3% vs. 18.7%). The odds were not statistically significant after adjustment for clinically relevant variables; however, the association was still favorable when looking at a 52% reduced adjusted odds (OR 0.48, 95% CI 0.20–1.15; p=0.098) for progression in favor of the SGLT2 inhibitors. This discovery is clinically significant as it may take longer to observe retinal outcomes than the six-month duration of this study. Our results were similar to the observations of Yang et al., who studied over 3.5 million patients newly diagnosed with T2DM in a population-based study in Taiwan. Following propensity score matching, SGLT2 inhibitor therapy was significantly associated with a lower risk of sight-threatening retinopathy (ST-RP) than DPP-4 inhibitor therapy (adjusted HR 0.57, 95% CI 0.51–0.63). This stronger association in that study may be due to its larger sample size and median follow-up of over 2 years compared to our study with only 150 patients and a median follow-up of 6 months.[10] A similar Japanese retrospective cohort study conducted in 2024 directly comparing SGLT2 and DPP-4 inhibitors also found that there were significantly fewer diabetic retinopathy cases in the SGLT2 inhibitor group (HR 0.83; 95% CI 0.75–0.92). However, the investigators did not observe statistically significant differences in diabetic macular edema or ophthalmic procedures related to DR. This is similar to our results, where the number of cases with DME development or progression was lower with SGLT2 inhibitors but not statistically significant, possibly because the benefit of SGLT2 inhibitors is more evident when used for overall DR incidence rather than specific complications such as DME.[11] The comparative-effectiveness study by Barkmeier et al. from 2024 of 371,698 patients also found that aural rehydration was superior. The risk of treatment for sight-threatening retinopathy was lower with SGLT2 inhibitor use than with DPP-4 inhibitor use (HR 0.79, 95% CI 0.64–0.97). The absolute probability of five-year treatment for sight-threatening retinopathy was also lower with SGLT2 inhibitors than with DPP-4 inhibitors (0.7% vs. 0.9%). Our study showed a similar direction of effect, but for reasons of the much smaller sample size and limited follow-up, we did not see a significant difference.[8] The present results were also consistent with those of a 2026 meta-analysis by Akande and Albert, which incorporated 16 comparative human studies that included 1,785,409 patients. The risk of progression of DR (RR 0.77, 95% CI 0.72–0.82) was significantly lower with SGLT2 inhibitors than with other glucose-lowering agents, as was the risk of progression of DME (RR 0.75, 95% CI 0.69–0.82). The authors noted that this benefit was maintained when DPP-4 inhibitors were used as a comparator. Thus, the same direction of observed decreases in the progression of overall DR and DME was observed, but in the absence of adequate precision in our study, these decreases were not statistically significant.[12] Ma et al. 2022 performed a systematic review and meta-analysis of randomized controlled trials, which yielded more tempered evidence. Treatments with SGLT2 inhibitors were not significantly correlated with the development of retinal disease or DR, but specific agents, empagliflozin and ertugliflozin, had potentially favorable signals. The results of this study may account for why the protective association we measured in our real-world cohort was not statistically significant, when the numeric rate of progression was lower in the SGLT2 group. There may be differences in study populations, retinal outcome definitions, duration of follow-up, and ascertainment methods that explain the discrepancy between randomized and observational evidence.[13, 14] The results were also similar to this systematic review and meta-analysis of 61 randomized trials with 188,463 participants published in 2024 by Małyszczak et al. SGLT2 inhibitors or DPP-4 inhibitors had no significant effects on the risk of DR when compared with placebo (pooled OR 1.02 and 1.10, respectively). In this context, our study indicated a possible beneficial effect of SGLT2 inhibitors on reducing progression, but this should not be interpreted as a direct retinal protective effect, as we observed only for a short duration and in an observational study.[15] Later studies have likewise highlighted the lack of clarity about the retinal effect of a class. In a network meta-analysis published in 2025, which included 30 randomized trials with 70,310 participants, none of the individual SGLT2 inhibitors showed a significant difference in risk for DR compared to placebo. Empagliflozin performed relatively well, but the differences between agents were not statistically significant. This is in contrast to the observational data that we have, which show that the advantage is smaller for those taking SGLT2 inhibitors, but does confirm that the apparent benefit might be affected by baseline risk, concomitant treatment, glycemic control, and other patient characteristics.[16] The evidence for DPP-4 inhibitors has also been mostly inconclusive. In a network meta-analysis of 43 randomized trials, involving 74,546 participants, published in 2026, there was no significant difference in the risk of DR between any DPP-4 inhibitor and placebo, with all credible intervals crossing 1. This helps support the use of DPP-4 inhibitors as a reasonable comparator in this study and indicates that the lower progression seen with the SGLT2 inhibitors is not a result of the DPP-4 inhibitors themselves raising the risk of DR, but rather a relative advantage.[17] In our study, longer duration of diabetes was also independently associated with DR progression. This is biologically possible since the chronic hyperglycemic environment leads to endothelial dysfunction, loss of pericytes, nonperfusion of capillaries, and progressive microvascular damage of the retina. Our analysis also showed that higher baseline HbA1c was an independent risk factor for progression, as has been previously reported, indicating that there is a link between the cumulative glycemic exposure and retinal disease. These findings are also in line with the current evidence that the severity of diabetes continues to have a significant effect on retinal outcomes, even after taking into account the effects of different drug classes used to lower blood glucose.[12, 15] In our analysis, the strongest predictor of the retina was advanced baseline retinopathy, as patients with moderate and severe NPDR or PDR had significantly higher odds of progression. This is not surprising, as patients with well-established advanced microvascular damage would be expected to have a higher level of biological substrate for further vascular damage. Importantly, the 2026 meta-analysis also focused on patients with established DR or DME, and found that there were lower rates of progression when using an SGLT2 inhibitor, which might be more relevant in patients with established retinal disease.[12] The relatively improved glycemic profile experienced by SGLT2 inhibitor users could also have helped drive the positive retinal outcome. Findings from our study show reduced follow-up HbA1c and fasting glucose levels, but the change in both of these levels was not statistically significant. SGLT2 inhibitors also have effects on body weight, blood pressure, renal hemodynamics and vascular function, all of which may theoretically alter the health of retinal microvessels. The present study, however, is observational and treatment groups were not randomized, so that a direct retinal pharmacological effect cannot be differentiated from any difference in glycemic control and other treatment-related factors.[10] Together, the results of the present study suggest that SGLT2 inhibitors provide a potentially more favorable profile in the retina than DPP4 inhibitors, with all progression events and worsening of DR severity being numerically less common in the SGLT2 group and the incidence of DME progression and PDR development being numerically lower in the SGLT2 group. However, the primary outcome was not statistically significant, and the wide range of the adjusted estimate suggests that the results are suggestive rather than conclusive. The retrospective design, small sample size, and the short six-month follow-up, and even residual confounding, all restrict causal inferences. Further larger, prospective studies with standardized retinal image capture and extended follow-up are thus needed to determine if apparent DR reductions are a class effect of SGLT2 inhibitors.[16, 18] Limitations There were several limitations on this research. Because of its retrospective, comparative cohort design, it was not possible to infer causality between the various antidiabetic therapies and retinal outcomes and residual confounding could not be ruled out despite adjustment for multiple variables. The number of participants (150) and duration of follow-up (6 months) were relatively small, and may have limited statistical power to find significant differences among relatively rare retinal outcomes. Retinal evaluations were performed by using clinical data and imaging information available, and for some of the participants, no standardized ophthalmological assessment or optical coherence tomography was available. Retinal outcomes may also have been affected by variations in treatment duration, in addition to background glucose-lowering therapy, and in compliance and disease parameters. Lastly, the study was carried out in one hospital, and the results may not be applicable to other populations.

SGLT2 inhibitor therapy showed a smaller number of patients with progression of DR, increased DR severity, progression of DME, and PDR than DPP-4 inhibitor therapy. The adjusted odds of progression of retinopathy were lower in the group who used the SGLT2 inhibitors, but this difference did not achieve statistical significance. Factors that were important predictors of progression were longer duration of diabetes, higher level of baseline HbA1c, and advanced baseline retinopathy. Further larger, longitudinal, and well-validated retinal assessment studies are required to confirm the clinically meaningful retinal benefit of SGLT2 inhibitors.