Diabetic Retinopathy and Risk of Stroke: A Secondary Analysis of the ACCORD Eye Study

Abstract

Background and Purpose

Diabetic retinopathy (DR) is a common microvascular complication of diabetes which causes damage to the retina and may lead to rapid vision loss. Previous research has shown that the macrovascular complications of diabetes, including stroke, are often comorbid with DR. We sought to explore the association between DR and subsequent stroke events.

Methods

This is a secondary analysis of patients enrolled in the ACCORD Eye study. The primary outcome was stroke during follow-up. The exposure was presence of DR at study baseline. We fit adjusted Cox proportional hazards models to provide hazard ratios for stroke and included interaction terms with the ACCORD randomization arms.

Results

We included 2,828 patients, in whom the primary outcome of stroke was met by 117 (4.1%) patients during a mean (SD) of 5.4 (1.8) years of follow-up. DR was present in 874/2,828 (30.9%) of patients at baseline and was more common in patients with than without incident stroke (41.0% vs 30.5%, p=0.016). In an adjusted Cox regression model, DR was independently associated with incident stroke (HR 1.52, 95% CI 1.05–2.20, p=0.026). This association was not affected by randomization arm in the ACCORD glucose (p=0.300), lipid (p=0.660), or blood pressure interventions (p=0.469).

Conclusion

DR is associated with an increased risk of stroke, which suggests that the microvascular pathology inherent to DR has larger cerebrovascular implications. This association appears not to be mediated by serum glucose, lipid and blood pressure interventions.

Introduction

The macrovascular complications of diabetes, including stroke, are often comorbid with shared and, possibly, synergistic vascular pathology. However, whether microvascular disease plays an important role in diabetic strokes remains unknown.^1,2 Diabetic retinopathy (DR) is a common microvascular complication of diabetes and remains a leading cause of blindness in working-age adults in the United States, affecting almost one-third of diabetic patients over 40 years of age.³ Because diabetic patients have a higher risk of developing stroke than the general population, DR could be an important microvascular risk factor for the macrovascular sequelae of stroke.^4,5 The aim of the present study was to determine the association of retinopathy with the risk of stroke in diabetic patients.

Methods

Study Population:

This is a secondary analysis of the Action to Control Cardiovascular Risk in Diabetes (ACCORD) Eye study, a sub-study of the ACCORD trial which is publicly available at https://www.ninds.nih.gov/Current-Research/Research-Funded-NINDS/Clinical-Research/Archived-Clinical-Research-Datasets. The ACCORD Eye study enrolled 3,472 patients without a history of photocoagulation or vitrectomy for DR and consisted of eye examinations conducted by a study ophthalmologist or optometrist, as well as seven-field stereoscopic color fundus photographs at baseline and at year 4 of follow-up. The eye exam included visual acuity measurement, slit lamp examination, and dilated fundus examination.⁶ We included patients with non-missing data on the study exposure, demographics, and outcome. Because the data was completely de-identified, local IRB approval was not required.

Assessment of DR and Stroke:

The ACCORD Eye study utilized a modified version of the Early Treatment Diabetic Retinopathy Study (ETDRS) scale. Severity levels from both eyes were combined, for a final score up to 17 points (see Supplementary Table I for the full score).⁷ Our study’s exposure was the presence of any DR on the baseline examination, which we defined as an ETDRS score of ≥4, corresponding to at least mild non-proliferative DR. The primary outcome was stroke during follow-up, which included ischemic and hemorrhagic events.⁸ Stroke was adjudicated by two primary reviewers and an experienced stroke adjudicator, who were blinded to the study interventions.

Statistical Analysis:

We fit Cox models to the primary outcome to provide hazard ratios for the risk of stroke in relation to DR. Covariates in the adjusted Cox model, chosen with backwards stepwise selection set to a p value <0.05, included baseline patient age, sex, history of cardiovascular disease, ACCORD treatment arm, and total cholesterol level (Table 1). We performed a sensitivity analysis where we adjusted for all baseline covariates and a second sensitivity analysis where we fit our adjusted Cox model with the predictor of no DR (ETDRS 1–3), mild DR (ETDRS 4–5), or moderate to severe DR (ETDRS ≥6).

Table 1.

Baseline demographics for the full cohort, and stratified by the outcome of stroke.

Variable	Full Cohort (n=2,828)	Stroke (n=117)	No stroke (n=2,711)	p-value*
Age (years, mean±SD)	62.1±6.3	63.8±6.2	62.0±6.3	0.004
White (n, %)	1,917, 67.8%	86, 73.5%	1,831, 67.5%	0.176
Male sex (n, %)	1748, 61.8%	88, 75.2%	1,660, 61.2%	0.002
Current cigarette smoking (n, %)	318, 11.2%	13, 11.3%	305, 11.3%	0.963
Hemoglobin A1c (%, mean±SD)	8.5±0.9	8.5±0.9	8.5±0.9	0.961
Years since diabetes diagnosis (mean±SD)	9.9±7.0	10.0±7.2	9.9±7.0	0.887
Total cholesterol (mg/dL, mean±SD)	179.4±40.8	185.5±43.6	179.2±40.7	0.100
History of prior stroke (n, %)	125, 4.4%	12, 10.3%	113, 4.2%	0.002
History of cardiovascular disease (n, %)	886, 31.3%	67, 57.3%	1,875, 69.2%	0.007
History of antihypertensive medications (n, %)	2,375, 84.0%	103, 88.0%	2,272, 83.8%	0.222
History of hyperlipidemia medications (n, %)	2,263, 80.0%,	92, 78.6%	2,171, 80.1%	0.701
ETDRS score (median, IQR)	2, 1–4	2, 1–5	2, 1–4	0.031
Diabetic retinopathy (n, %)	874, 30.9%	48, 41.0%	826, 30.5%	0.016

^*p value calculated with Student’s t-test for continuous variables, the Wilcoxon rank-sum test for ordinal variables, and the chi-square test for binary variables.

Results

Of the 3,472 patients that were enrolled, we included 2,828 (Table 1, Supplementary Table I). Of the 2,828 patients, the primary outcome of stroke was met by 117 (4.1%) patients during a mean (SD) of 5.4 (1.8) years of follow-up. DR was present in 874/2,828 (30.9%) of patients at baseline and was more common in patients with than without incident stroke during follow-up (41.0% vs 30.5%, p=0.016).

Figure 1 shows a Kaplan-Meier curve fit to the outcome of stroke (unadjusted HR 1.55, 95% CI 1.07–2.24, p=0.020, Log-rank p=0.019). In the Cox regression model, DR was independently associated with incident stroke (HR 1.52, 95% CI 1.05–2.20, p=0.026) (Table 2). The interaction terms between DR and ACCORD treatment arm were not significant in unadjusted (data not shown) and adjusted models: DR*glucose intervention, HR 0.67, 95% CI 0.32–1.42, p=0.300; DR*lipid intervention, n=1,582, HR 1.29, 95% CI 0.42–4.00, p=0.660; DR*blood pressure intervention, n=1,246, HR 0.69, 95% CI 0.25–1.91, p=0.469.

Figure 1.

Kaplan-Meier curves showing stroke events in patients without diabetic retinopathy at baseline (ETDRS 1–3) versus those with diabetic retinopathy (ETDRS ≥4).

Table 2.

Multivariate Cox regression model fit to incident stroke.

	Hazard Ratio (95%CI)	p value
Baseline age	1.05 (1.02–1.08)	0.001
Total cholesterol	1.01 (1.00–1.01)	0.004
History of cardiovascular disease	1.60 (1.10–2.33)	0.013
Male sex	2.03 (1.31–3.13)	0.001
ACCORD randomization arm	1.09 (1.00–1.18)	0.039
Diabetic retinopathy (ETDRS ≥4)	1.52 (1.05–2.20)	0.026

In the sensitivity analysis adjusted for all covariates in Table 1, DR remained associated with incident stroke (HR 1.76, 95% CI 1.16–2.65, p=0.007). In the second sensitivity analysis, we found that compared to no DR (n=1,954), the subgroups of mild (n=563) and moderate-severe DR (n=311) remained associated with stroke (adjusted HR per ordinal shift: 1.32, 95% CI 1.04–1.67, p=0.022). Compared to the patients with no DR, those with moderate-severe DR had an even higher risk of stroke (adjusted HR 1.69, 95% CI 1.01–2.80, p=0.044).

Discussion

Baseline DR in the ACCORD Eye trial was associated with an increased risk of stroke during follow-up. These findings are similar to an analysis performed in the ARIC cohort⁹ and also in the Wisconsin Epidemiological Study of DR, which found that only proliferative DR was associated with stroke.^10,11 Conversely, in the United Kingdom Prospective Diabetes Study, DR was not a significant risk factor for stroke (Supplementary Table II).¹² However, these studies did not have strict control of vascular risk factors or intensive treatment cohorts. We did not, however, find that the glucose, lipid, or blood pressure interventions in ACCORD had an impact on the association between diabetic retinopathy and stroke, although we may have been underpowered to detect such an association.

This study strengthens the association between DR and stroke. While the pathogenesis of DR is not well-understood, inflammatory and hematologic changes from prolonged hyperglycemia ultimately lead to microvascular endothelial injury and pericyte loss.¹³ Intraretinal capillary occlusion can lead to progressive retinal nonperfusion and sequelae from retinal ischemia. Additional studies into these microvascular mechanisms may yield insight into potential treatment approaches to mitigate the risk of diabetic stroke.

The main limitation of our study is that it is a secondary analysis of a subgroup from a randomized-controlled clinical trial, which introduces selection bias. We also do not have enough stroke events to evaluate the potential association between DR with incident ischemic and hemorrhagic strokes, which have shared, but also distinct pathophysiology. We also were not able to delve into ischemic stroke subtypes. DR, for example, could lead to a higher risk of lacunar stroke than large artery atherosclerotic stroke. The strengths of our study are that ACCORD Eye had exceptional rigor in the measurements of exposure, outcome, and potential confounders.

Conclusion

We found an independent association between DR and stroke, suggesting that the microvascular pathology inherent to DR has larger cerebrovascular implications.

Non-standard abbreviations and acronyms

DR

Diabetic Retinopathy

ACCORD

Action to Control Cardiovascular Risk in Diabetes

ETDRS

Early Treatment Diabetic Retinopathy Study