Findings from a recent study published in American Journal of Ophthalmology indicate that increasing severity of diabetic retinopathy (DR) in type 2 diabetes (T2M) patients is tied to progressively higher risk of all-cause dementia and vascular dementia (VD).
Give me some background first.
Some context: Diabetes mellitus (DM) and dementia share several proposed pathological links, with impaired insulin signaling and oxidative stress among the mechanisms thought to connect them.
On the other hand, DR is a microvascular—and one of the most common—complication of DM.
- Because it reflects cumulative vascular damage, its presence may act as a clinical marker for broader systemic conditions.
The vascular complications associated with DM2, including DR, share features with VD: chronic vascular damage and endothelial dysfunction.
But here’s the distinction: The connection between DR and AD is less clear. AD involves biochemical pathways beyond microvascular disease, and prior research on whether DR independently contributes to AD risk has produced conflicting results.
Now, talk about the study.
Investigators conducted a retrospective cohort study using the TriNetX Global Collaborative Network, a federated health research platform that pulls real-world de-identified patient data from 153 healthcare organizations worldwide, representing over 184 million patients.
The goal: Determine whether DR stage (proliferative vs. nonproliferative) is associated with different levels of dementia risk—and whether that risk varies across dementia subtypes.
Who was included in the study?
The analysis drew from 769,930 individuals aged 65 and older who underwent an eye exam or optical coherence tomography (OCT) between Jan. 1, 2010, and Jan. 1, 2020.
Four cohorts were identified:
- 14,034 with proliferative diabetic retinopathy (PDR)
- 29,188 with nonproliferative diabetic retinopathy (NPDR)
- 208,640 with DM2 without DR
- 447,054 without diabetes
Individuals under 65 and those with diabetic macular edema (DME) were excluded. All cohorts underwent 1:1 propensity score matching to account for baseline differences.
Findings?
Compared to nondiabetic controls, all three diabetic groups carried a higher risk of all-cause dementia.
- PDR had the highest hazard ratio at 1.583 (p < .0001), followed by NPDR at 1.405 (p < .0001) and DM2 without DR at 1.262 (p < .0001).
VD showed the steepest gradient while PDR patients faced more than double the risk compared to controls: HR: 2.077 (p < .0001), with
- NPDR at 1.917 (p < .0001)
- DM2 without DR at 1.384 (p < .0001)
AD risk was elevated across all diabetic groups compared to controls, but it did not track with DR severity. The HRs were similar across all three groups:
- PDR at 1.175 (P = .0419)
- NPDR at 1.233 (p < .0001
- DM2 without DR at 1.117 (p < .0001).
Tell me more.
When PDR and NPDR patients were compared directly to those with DM2 without DR, both groups had higher risk for all-cause dementia and VD, but not AD.
- PDR carried an HR of 1.202 (p < .0001) for all-cause dementia and 1.504 (p < .0001) for VD
- NPDR showed a similar pattern: 1.113 (p < .0001) for all-cause dementia and 1.322 (p < .0001) for VD
Go on ..
When stratified by DR severity directly: PDR was associated with higher risk of all-cause dementia (HR: 1.121, P = .0003) and VD (HR: 1.177, P = .0126) compared to NPDR.
- AD risk showed no difference between the two groups.
Why that matters: The stepwise increase in dementia and VD risk with DR severity suggests that progressive retinal microvascular damage parallels cerebral small vessel disease.
The lack of an AD-severity link also points to a different mechanism, one driven more by diabetes itself than by the degree of microvascular injury.
Any limitations?
As this was a retrospective study relying on ICD coding across multiple institutions, variability was introduced in how diagnoses were recorded.
Other considerations:
- Education and socioeconomic status, both known dementia risk factors, could not be captured
- The average follow-up of 6.7 years may have been too short for younger participants who hadn’t yet reached the typical age of dementia onset
- Potential survivorship bias: the mean age at PDR diagnosis was 69.1 years, while AD typically presents after age 80
- Patients with diabetic macular edema were excluded, which may limit generalizability
Expert opinion?
The authors argued that their findings underscore the role of systemic microvascular injury in cognitive decline. DR severity did not appear to confer additional AD risk beyond that associated with diabetes itself, which they interpreted as further evidence that DR is tracking vascular pathology specifically.
- Their position: DR serves as a potential observable biomarker for systemic vascular injury and a practical tool for identifying patients at higher risk of cognitive decline.
Anything else?
One finding worth calling out: Incidence rates across the matched cohorts showed a clear gradient. All-cause dementia incidence was 15.9% in the PDR group vs. 10.2% in matched controls. For VD specifically, the gap was even more pronounced: 3.6% in PDR patients vs. 1.7% in controls.
The authors flagged that patients with DR often experience years of microvascular damage before any cognitive symptoms surface.
- Routine ophthalmologic examinations—already standard of care in diabetes management—could represent an accessible screening platform for dementia risk assessment and early intervention.
Take home.
For clinicians managing diabetic patients with retinopathy: the severity of DR may signal more than ocular disease alone. The stepwise link between DR progression and dementia risk, particularly VD, makes a case for considering cognitive screening in patients with advancing retinopathy.
As such: Routine eye exams may be a practical entry point for earlier identification of patients at risk for cognitive decline, and a reason to loop in primary care for additional dementia screening.