HIGHLIGHTS
-
•
GLP-1 RA use offsets the high morbidity from vascular complications in T2D with DR.
-
•
Patient with T2D and DR on GLP-1 RAs experienced fewer microvascular complications.
-
•
Patient with T2D and DR on GLP-1 RAs experienced fewer macrovascular complications.
-
•
Risk of proliferative DR, RVOs, and neovascular glaucoma were lower with GLP-1 RAs.
Purpose
To evaluate the association of glucagon-like peptide-1 receptor agonists (GLP-1 RAs) use on macrovascular and microvascular outcomes in patients with type 2 diabetes (T2D) and diabetic retinopathy (DR)—a high-risk group often excluded from clinical trials.
Design
Retrospective, population-based cohort study.
Participants
Adults aged ≥18 years with T2D (with hemoglobin A1c of ≥6.5%) and a pre-existing diagnosis of DR from the TriNetX research network database between January 1, 2015, and December 31, 2022.
Methods
The study included 173,216 adults with T2D, all of whom had DR, adjusted for baseline characteristics through propensity score matching (PSM) based on whether the individuals received at least 2 prescriptions of a GLP-1 RA (semaglutide, dulaglutide, liraglutide, exenatide, tirzepatide, or lixisenatide) at least 6 months apart.
Main outcome measures
Cox proportional hazard regression models were used to evaluate the association between GLP-1 RAs and the risk of incident macrovascular and microvascular complications over a 2-year follow-up period.
Results
After PSM, 30,613 individuals (mean [SD] age, 61.6 [11.4] years; 53.2% were females) were prescribed GLP-1 RAs. Patients on GLP-1 RAs had a decreased risk of myocardial infarctions (MIs; hazard ratio [HR], 0.65; 95% CI, 0.61-0.69), coronary artery revascularization procedures (HR, 0.75; 95% CI, 0.67-0.84), heart failure exacerbations (HR, 0.78; 95% CI, 0.76-0.81), ischemic strokes (HR, 0.78; 95% CI, 0.74-0.83), lower extremity amputations (HR, 0.78; 95% CI, 0.69-0.88), acute kidney injuries (AKI; HR, 0.68; 95% CI, 0.66-0.71), or the need for renal replacement therapy (RRT; HR, 0.40; 95% CI, 0.36-0.43). Fewer individuals also progressed to proliferative diabetic retinopathy (HR, 0.78; 95% CI, 0.71-0.86), experienced retinal vein occlusions (RVOs; HR, 0.70; 95% CI, 0.61-0.80), or developed neovascular glaucoma (HR, 0.65; 95% CI, 0.47-0.89); no association was observed for retinal artery occlusions (RAOs; HR, 0.85; 95% CI, 0.57-1.26) or cases of non-arteritic ischemic optic neuropathy (NAION; HR, 0.88; 95% CI, 0.54-1.44).
Conclusions
In patients with T2D and pre-existing DR, the use of GLP-1 RAs was associated with a reduced risk of major macrovascular and microvascular complications, including those directly affecting the retina. Future studies are needed to assess the extent to which GLP-1 RAs benefit long-term outcomes.
INTRODUCTION
G lucagon-like peptide-1 receptor agonists (GLP-1 RAs) are an emerging class of antihyperglycemic medications proven to be effective in helping individuals with type 2 diabetes (T2D) meet glycemic targets and improve cardiometabolic and renal outcomes in high-risk subgroups. ,,,,,,, Multiple large clinical trials have demonstrated that GLP-1 RAs reduce the risk of major adverse cardiovascular events (MACE), including all-cause and cardiovascular mortality, stroke, myocardial infarction (MI), and the progression of diabetic kidney disease.
Diabetic retinopathy (DR), one of the most common microvascular complications of diabetes, is a well-established marker of systemic vascular injury. Individuals with DR have substantially elevated risk of macrovascular complications—including MI, ischemic stroke, and congestive heart failure—as well as microvascular complications such as nephropathy and neuropathy. , Moreover, the presence and severity of DR are independently associated with higher all-cause mortality. Retrospective cohort studies have shown that GLP-1 RA use is associated with a modestly increased risk of incident DR; however, fewer patients experienced sight-threatening DR complications, including blindness, even among those with preexisting DR. Despite these associations, patients with DR—particularly those with unstable or advanced disease—are often systematically excluded from randomized controlled trials of GLP-1 RAs. ,, As a result, the impact of GLP-1 RAs on systemic and ophthalmic macrovascular and microvascular outcomes in this high-risk subgroup remains understudied.
This retrospective cohort study addresses this critical knowledge gap by evaluating the role of GLP-1 RAs on macrovascular, microvascular, and retinal vascular outcomes in a large, real-world population of patients with T2D and pre-existing DR. Specifically, we examined whether GLP-1 RA use is associated with reduced incidence of MIs, coronary artery revascularization procedures, heart failure exacerbations, ischemic strokes, new-onset diabetic foot disease, lower extremity amputations, and renal complications including acute kidney injury or the need for renal replacement therapy (RRT). In parallel, we assessed sight-threatening ophthalmic outcomes, including progression to proliferative-stage DR or the development of retinal vein and artery occlusions, neovascular glaucoma, or non-arteritic ischemic optic neuropathy (NAION), to explore potential benefits beyond glycemic control and systemic outcomes.
METHODS
This study used the TriNetX Research Network database (TriNetX, LLC), a multicenter federated health research network of more than 120 participating healthcare organizations, including academic medical centers, specialty physician practices, and community hospitals; this anonymized data set aggregates the electronic health records (EHRs) from a population of approximately 275 million patients. , The study adhered to the principles of the Declaration of Helsinki and was deemed to meet the criteria for exemption from human subjects research review and the requirement for informed consent, as it was a secondary analysis of existing health record data that had been de-identified in accordance with the HIPAA Privacy Rule (§164.514[a]) by the Lahey Hospital & Medical Center Institutional Review Board.
Population and Design
Between January 1, 2015 and December 31, 2022, we conducted a retrospective cohort study of patients aged 18 years or older with T2D and DR and a recent hemoglobin A1c of 6.5% or higher identified from the TriNetX database. Patients were split into 2 cohorts based on whether they had received a prescription for GLP-1 RA (semaglutide, dulaglutide, liraglutide, exenatide, tirzepatide, or lixisenatide): those with GLP-1 RA prescriptions were considered the treatment group, while those without GLP-1 RA prescriptions were considered the reference group. We included only individuals with at least 2 GLP-1 RA prescriptions issued at least 6 months apart. The index event was defined as the date of the second GLP-1 RA prescription for the treatment group and as the date of meeting the inclusion criteria for the control group. Individuals with more than 2 prescriptions were included in the treatment group based on the query criteria. Those with only one prescription or 2 prescriptions less than 6 months apart were excluded. Cohorts were matched based on propensity scores for clinically-relevant variables, including the severity of DR (Supplementary Appendix; Supplementary Table 1). Outcomes were assessed within a 2-year period after the index event. The analysis was conducted on February 22, 2026.
Endpoints
The study quantified the incidence of macrovascular outcomes (MI, coronary artery revascularization, heart failure exacerbation, ischemic stroke, new-onset diabetic foot disease, any lower extremity amputation, foot or below amputation, or debridement procedure), microvascular outcomes (acute kidney injury [AKI], RRT, or renal transplantation), and ophthalmic outcomes (new-onset proliferative diabetic retinopathy, incident retinal vein occlusion [RVO], retinal arterial occlusions [RAO], cases of neovascular glaucoma, or NAION) by using the International Classification of Diseases, Tenth Revision (ICD-10) or Current Procedural Terminology (CPT) codes in the 2 cohorts: individuals prescribed GLP-1 RAs and those not prescribed these agents (Supplementary Table 2). Finally, to strengthen the reliability of our observational data, we performed a sensitivity analysis by evaluating falsification outcomes, such as the incidence of appendicitis, nephrolithiasis, optical coherence tomography imaging of the macula (retina), as well as the rate of all-cause emergency department visits or hospitalizations within the same 2-year follow-up timeframe for the GLP-1 RA and comparator groups.
Statistical Analysis
Continuous variables are presented as mean (SD), and categorical variables are presented as a number (%). One-to-one propensity score matching (PSM) was performed using greedy nearest-neighbor matching with a caliper of 0.1 times the pooled SD of the linear propensity scores to control for baseline differences between the study groups. We used the standardized mean difference post-PSM to quantify the variation in baseline characteristics between the 2 groups in terms of SD units, thereby assessing the balance in measured variables in a sample weighted by the inverse probability of treatment. The variables were selected based on their potential role in the outcomes examined in our study and were specifically designed to mirror our previous analysis of DR outcomes.
After PSM, outcomes were compared between the 2 cohorts using absolute and relative risk differences. Kaplan-Meier curves and Cox-proportional hazard models were used for survival analysis. Statistical significance was set at a 2-sided p <.05. Statistical analyses were performed using an integrated R (R Project for Statistical Computing, Version 4.0.2) on the TriNetX platform. We calculated the E-value as part of a sensitivity analysis to assess robustness against bias from unmeasured confounding or omitted covariates for both primary and secondary outcomes. A high E-value implies that unmeasured confounders with a greater influence on the outcome of interest would be required to negate the observed association between exposure and outcome.
Results
Baseline Characteristics
The study cohort included 173,216 adults with T2D and pre-existing DR. Among these, 49,361 were on GLP-1 RAs, and 123,855 were not. Before PSM, patients on GLP-1 RA were younger females with higher rates of hypertension, hyperlipidemia, and ischemic heart diseases, but lower rates of chronic kidney disease (CKD) and ischemic stroke ( Table 1 ). After PSM, 30,613 individuals were prescribed GLP-1 RAs. These patients had a mean age (SD) of 61.6 (11.4) years and included 16,279 (53.2%) females and 14,334 males (46.8%), with 1930 (6.3%) individuals who identified as Asian, 8198 (26.8%) as Black or African American, 3979 (13.0%) as Hispanic or Latino, 21,073 (68.8%) not Hispanic or Latino (69.7%) and 15,968 (52.2%) as White ( Table 1 ). The matched control group of individuals without GLP-1 RA prescriptions, had a mean (SD) age of 61.6 (12.7) years and include 16,303 (53.3%) females and 14,310 (46.7%) males, with 1902 (6.2%) individuals who identified as Asian, 8224 (26.9%) as Black or African American, 3947 (12.9%) as Hispanic or Latino, 21,154 (69.1%) not Hispanic or Latino and 16,004 (52.3%) as White. The ICD-10 and Veterans Affairs health care–specific codes for these baseline characteristics are presented in Supplementary Table 1.
Table 1
Demographic Data of the Patient Population With T2D With Pre-Existing DR Before and After Propensity Matching.
| Patient Characteristics | Before PSM | Standard Difference | After PSM | SMD | ||
|---|---|---|---|---|---|---|
| On GLP-1 RAs (n = 49,361) | NOT on GLP-1 RAs (n = 123,855) | On GLP-1 RAs (n = 30,613) | NOT on GLP-1 RAs (n = 30,613) | |||
| Demographics (%) | ||||||
| Age, years | 60.7 ± 11.3 | 63.6 ± 13.1 | 0.241 | 61.6 ± 11.4 | 61.6 ± 12.7 | 0.004 |
| Sex | ||||||
| Female | 26,765 (54.2) | 60,922 (49.2) | 0.101 | 16,279 (53.2) | 16,303 (53.3) | 0.002 |
| Race | ||||||
| Black or African American | 13,363 (27.1) | 28,477 (23.0) | 0.094 | 8198 (26.8) | 8224 (26.9) | 0.002 |
| Asian | 2827 (5.7) | 11,387 (9.2) | 0.132 | 1930 (6.3) | 1902 (6.2) | 0.004 |
| White | 26,297 (53.3) | 61,850 (49.9) | 0.067 | 15,968 (52.2) | 16,004 (52.3) | 0.002 |
| Hispanic or Latino | 5960 (12.1) | 15,538 (12.5) | 0.007 | 3979 (13.0) | 3947 (12.9) | 0.003 |
| Not Hispanic or Latino | 34,677 (70.2) | 82,864 (66.9) | 0.072 | 21,073 (68.8) | 21,154 (69.1) | 0.006 |
| Comorbidities (%) | ||||||
| Mild Nonproliferative Diabetic Retinopathy | 23,989 (48.6) | 43,131 (34.8) | 0.282 | 13,572 (44.3) | 13,554 (44.3) | 0.001 |
| Moderate Nonproliferative Diabetic Retinopathy | 7667 (15.5) | 11,995 (9.7) | 0.177 | 4039 (13.2) | 4153 (13.6) | 0.011 |
| Severe Nonproliferative Diabetic Retinopathy | 2711 (5.5) | 5012 (4.0) | 0.068 | 1460 (4.8) | 1537 (5.0) | 0.012 |
| Proliferative Diabetic Retinopathy | 8270 (16.8) | 27,264 (22.0) | 0.133 | 5414 (17.7) | 5460 (17.8) | 0.004 |
| Comorbidities (%) | ||||||
| Hypertension | 45,697 (92.6) | 106,266 (85.8) | 0.219 | 28,072 (91.7) | 28,120 (91.9) | 0.006 |
| Dyslipidemia | 45,720 (92.6) | 95,407 (77.0) | 0.445 | 27,735 (90.6) | 27,919 (91.2) | 0.021 |
| Ischemic Stroke | 5091 (10.3) | 13,487 (10.9) | 0.019 | 3368 (11.0) | 3300 (10.8) | 0.007 |
| Ischemic Heart Diseases | 18,770 (38.0) | 46,291 (37.4) | 0.013 | 11,766 (38.4) | 11,819 (38.6) | 0.004 |
| Atrial Fibrillation and Flutter | 5523 (11.2) | 14,931 (12.1) | 0.027 | 3574 (11.7) | 3514 (11.5) | 0.006 |
| CKD | 18,004 (36.5) | 50,148 (40.5) | 0.083 | 11,816 (38.6) | 11,853 (38.7) | 0.002 |
| Peripheral Arterial Disease | 3656 (7.4) | 9234 (7.5) | 0.002 | 2448 (8.0) | 2489 (8.1) | 0.005 |
| Peripheral Vascular Disease | 7915 (16.0) | 19,329 (15.6) | 0.012 | 5176 (16.9) | 5092 (16.6) | 0.007 |
| Chronic Obstructive Pulmonary Diseases | 16,178 (32.8) | 28,664 (23.1) | 0.216 | 9382 (30.6) | 9451 (30.9) | 0.005 |
| Malignant Neoplasms | 23,933 (48.5) | 42,491 (34.3) | 0.291 | 13,801 (45.1) | 14,008 (45.8) | 0.014 |
| Benign Neoplasms | 19,414 (39.3) | 28,085 (22.7) | 0.366 | 10,757 (35.1) | 10,879 (35.5) | 0.008 |
| History of Nicotine Dependence | 12,624 (25.6) | 22,974 (18.5) | 0.170 | 7229 (23.6) | 7254 (23.7) | 0.002 |
| Tobacco Use | 4136 (8.4) | 5999 (4.8) | 0.143 | 2309 (7.5) | 2249 (7.3) | 0.007 |
| Obesity (BMI ≥ 30 kg/m 2) | 33,958 (68.8) | 55,104 (44.5) | 0.506 | 19,575 (63.9) | 19,750 (64.5) | 0.012 |
| Medications (%) | ||||||
| Statin | 44,458 (90.1) | 86,125 (69.5) | 0.529 | 26,740 (87.4) | 26,906 (87.9) | 0.016 |
| Angiotensin-Converting Enzyme Inhibitors | 31,453 (63.7) | 60,977 (49.2) | 0.295 | 19,134 (62.5) | 19,248 (62.9) | 0.008 |
| Angiotensin II Receptor Blockers | 21,009 (42.6) | 36,617 (29.6) | 0.273 | 11,893 (38.9) | 12,014 (39.2) | 0.008 |
| Angiotensin Receptor–Neprilysin Inhibitors | 1180 (2.4) | 831 (0.7) | 0.140 | 425 (1.4) | 401 (1.3) | 0.007 |
| Loop Diuretics | 17,049 (34.5) | 41,939 (33.9) | 0.014 | 11,033 (36.0) | 11,155 (36.4) | 0.008 |
| Potassium-sparing Diuretics | 5884 (11.9) | 8903 (7.2) | 0.161 | 3176 (10.4) | 3194 (10.4) | 0.002 |
| Insulin | 42,223 (85.5) | 84,367 (68.1) | 0.422 | 25,397 (83.0) | 25,527 (83.4) | 0.011 |
| Metformin | 40,853 (82.8) | 58,669 (47.4) | 0.799 | 23,288 (76.1) | 23,405 (76.5) | 0.009 |
| Sitagliptin | 13,333 (27.0) | 15,681 (12.7) | 0.366 | 6704 (21.9) | 6681 (21.8) | 0.002 |
| Empagliflozin | 13,777 (27.9) | 3274 (2.6) | 0.750 | 2729 (8.9) | 2546 (8.3) | 0.021 |
| Canagliflozin | 5498 (11.1) | 1698 (1.4) | 0.412 | 1361 (4.4) | 1250 (4.1) | 0.018 |
| Dapagliflozin | 5639 (11.4) | 1839 (1.5) | 0.413 | 1409 (4.6) | 1313 (4.3) | 0.015 |
| Glipizide | 16,452 (33.3) | 23,813 (19.2) | 0.325 | 9191 (30.0) | 9303 (30.4) | 0.008 |
| Glyburide | 4958 (10.0) | 9044 (7.3) | 0.098 | 3153 (10.3) | 3180 (10.4) | 0.003 |
| Antineoplastic agents | 7904 (16.0) | 14,268 (11.5) | 0.131 | 4466 (14.6) | 4504 (14.7) | 0.004 |
| Lab values | ||||||
| BNP (pg/mL) | 310.9 ± 2037.0 | 737.2 ± 2543.3 | 0.185 | 334.3 ± 2458.0 | 528.9 ± 2150.1 | 0.080 |
| NT-proBNP (pg/mL) | 1628.1 ± 5197.2 | 5010.5 ± 10 370.2 | 0.412 | 1980.0 ± 6082.5 | 3711.0 ± 8267.7 | 0.239 |
| LDL cholesterol (mg/dL) | 82.4 ± 37.3 | 88.7 ± 39.5 | 0.164 | 82.3 ± 37.1 | 90.2 ± 39.9 | 0.205 |
| Hemoglobin A1c (%) | 8.4 ± 1.9 | 8.1 ± 2.1 | 0.130 | 8.3 ± 1.9 | 8.4 ± 2.1 | 0.036 |
| CRP (mg/L) | 31.1 ± 56.4 | 37.8 ± 63.8 | 0.110 | 31.8 ± 57.3 | 34.3 ± 59.4 | 0.042 |
| Iron (ug/dL) | 65.8 ± 33.1 | 62.3 ± 36.2 | 0.100 | 65.4 ± 33.1 | 62.2 ± 35.8 | 0.093 |
| LVEF (%) | 58.1 ± 12.2 | 55.4 ± 14.3 | 0.201 | 58.8 ± 11.7 | 55.5 ± 14.2 | 0.249 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
