Retinal photographs. A, Note large retinal venular caliber (black arrows) and normal arteriolar caliber (white arrows). B, Note normal retinal venular caliber (black arrows) and narrowed arteriolar caliber (white arrows).
Wong TY, Kamineni A, Klein R, Sharrett AR, Klein BE, Siscovick DS, Cushman M, Duncan BB. Quantitative Retinal Venular Caliber and Risk of Cardiovascular Disease in Older PersonsThe Cardiovascular Health Study. Arch Intern Med. 2006;166(21):2388-2394. doi:10.1001/archinte.166.21.2388
Small vessel disease may contribute to the risk of cardiovascular disease in older persons. We describe the relation of retinal vascular caliber to incident coronary heart disease (CHD) and stroke in elderly persons.
Prospective population-based cohort study composed of 1992 men and women aged 69 to 97 years living in 4 US communities. Retinal arteriolar and venular calibers were measured from retinal photographs using a computer-assisted method. Incident CHD and stroke events were ascertained using standardized methods.
After 5 years of follow-up, there were 115 incident CHD events and 113 incident stroke events. Participants with larger retinal venular caliber had a higher incidence of CHD (11.7%; 95% confidence interval [CI], 8.7%-15.8%, vs 8.1%; 95% CI, 5.7%-11.6%), comparing largest with smallest venular caliber quartiles, and stroke (8.4%; 95% CI, 6.0-11.7, vs 5.8%; 95% CI, 3.9-8.4). At multivariable analysis, controlling for age, sex, race, arteriolar caliber, systolic and diastolic blood pressure, diabetes, glucose concentration, cigarette smoking, pack-years of smoking, and high-density-lipoprotein and low-density lipoprotein cholesterol levels, larger retinal venular caliber was associated with incident CHD (rate ratio, 3.0; 95% CI, 1.6-5.7, comparing largest with smallest venular caliber quartiles; Ptrend = .001) and incident stroke (rate ratio, 2.2; 95% CI, 1.1-4.3; Ptrend = .02). Additional adjustment for C-reactive protein and common and internal carotid artery intimal-media thickness had minimal effect on these associations. At multivariable analysis, smaller retinal arteriolar caliber was associated with incident CHD (rate ratio, 2.0; 95% CI, 1.1-3.7, comparing largest with smallest arteriolar caliber quartiles; P = .03) but not stroke (rate ratio,1.1; 95% CI, 0.5-2.2; P = .73).
Larger retinal venular caliber is independently associated with risk of cardiovascular disease in elderly persons.
Cardiovascular disease is a leading cause of death in persons aged 65 years or older in the United States.1 Traditional risk factors do not have the same predictive value in older persons2- 8 and, thus, identification of new risk factors is important. There is increasing recognition that small vessel disease (microvascular disease) may have a possible pathophysiologic role in the development of both subclinical and clinical cardiovascular disease.9- 11 Microvascular disease affecting cerebral arterioles, for example, is linked to a large proportion of subclinical and lacunar strokes defined at magnetic resonance imaging,12- 14 whereas coronary microvascular dysfunction may explain the occurrence of myocardial ischemia in persons without overt coronary artery occlusion.15,16 Most studies of microvascular disease have been conducted in small numbers of patients with symptomatic disease in specialized laboratory settings,10,11 and few have been conducted in older persons.
The retinal microcirculation is accessible to noninvasive visualization, providing a unique opportunity to investigate the relationship between microvascular changes and risk of cardiovascular disease.17 Changes in retinal arteriolar and venular caliber may represent structural damage or functional alterations.18 Recent data from the Atherosclerosis Risk in Communities (ARIC) study, a population-based cohort study in middle-aged (51-72 years) persons, showed that a smaller arteriolar-venular caliber ratio (AVR), reflecting narrowed arterioles or enlarged venules, was associated with risk of incident stroke events and coronary heart disease (CHD) events, independent of standard risk factors.19,20 Whether retinal arteriolar narrowing and venular dilation have similar predictive values in older persons is less clear and has not been consistently demonstrated in 3 cohort studies with older participants (the Beaver Dam Eye Study, the Blue Mountains Eye Study, and the Rotterdam Study).21- 24
In the Cardiovascular Health Study (CHS), a population-based study of persons 65 years or older, we have previously described cross-sectional associations of narrowed retinal arterioles with higher blood pressure and prevalent cardiovascular disease.25,26 In the current study, we examine prospectively the association of retinal vascular caliber to incident CHD and stroke.
The CHS is a population-based, longitudinal investigation of CHD and stroke in adults 65 years of age and older.27 The study design and methods are described in detail elsewhere.28 In brief, 5201 eligible persons were recruited in 1989 through 1990 in 4 field centers: Allegheny County (Pennsylvania), Forsyth County (North Carolina), Sacramento County (California), and Washington County (Maryland). An additional 687 eligible African Americans aged 65 years or older were recruited from Forsyth County, Sacramento County, and Allegheny County in 1992 through 1993. Differences between those recruited and those not recruited are described elsewhere.28 At the study clinic, all participants underwent a standardized interview, physical examination including blood pressure and anthropometric measurements, carotid ultrasonography, and phlebotomy for blood chemistry studies.29- 32
Photographic examination of the retina was first offered to participants during the 1997-1998 clinic visit.25,26 Of the 4249 participants (95.5% of the survivors) who were contacted for this examination, 2824 (66.5%) underwent retinal photography, which was performed at the clinic. Of these, 832 (29.5%) had photographs that were ungradeable for retinal vascular caliber, leaving 1992 persons for this study. Comparisons of persons with and without gradable photographs have been previously reported.26 Persons who did not undergo retinal photography or who had ungradeable photographs were older and more likely to be men and to be African Americans; were more likely to have hypertension and diabetes, higher systolic and diastolic blood pressure, and higher fasting glucose and total cholesterol levels; and were more likely to be current cigarette smokers.
Retinal photography procedures in the CHS have been reported in detail.25,26 In brief, after 5 minutes of dark adaptation, a 45° retinal photograph centered between the optic disc and the macula was obtained of 1 randomly selected eye. Photographs were evaluated at the Fundus Photograph Reading Center in Madison, Wis, by 2 trained and certified graders who were blinded to subject characteristics, according to a standardized protocol.
For the evaluation of retinal vascular caliber, photographs were digitized with a high-resolution scanner, and the diameters of all arterioles and venules coursing through an area ½ to 1 disc diameter from the optic disc margin were measured on a computer monitor.25,26 The results from this image analysis were summarized as central arteriolar and venular equivalents, using formulas by Hubbard et al,33 representing the average central arteriolar and venular caliber in that eye, respectively. These measurements were also expressed as an AVR, which represents the relative caliber of arterioles to venules and introduces some adjustment for variable magnification owing to differences in refractive errors. In addition to retinal vascular caliber, the presence of focal retinal microvascular signs (retinopathy, focal arteriolar narrowing, and arteriovenous nicking) was graded from photographs using a standardized light box protocol.25,26
Quality control procedures were implemented throughout the grading process.26 In the CHS, there were 71 intragrader and 69 intergrader rereadings. The intragrader and intergrader intraclass correlation coefficients for retinal vascular caliber equivalents ranged from 0.67 to 0.91.
Prebaseline and incident cardiovascular diseases were ascertained in the CHS using a detailed standardized protocol.34,35 Participants, family members, or other previously identified informants reported new cardiovascular events during semiannual contacts by telephone or at a clinic visit. Medical records were obtained to confirm the diagnosis, and events were adjudicated by a committee.34,35 The 2 cardiovascular outcomes of interest in this study were incident CHD, including fatal and nonfatal myocardial infarction, fatal CHD, and other CHD-attributed deaths (eg, sudden cardiac death),34 and incident stroke, including fatal and nonfatal stroke.35 Adjudicated events occurring through June 30, 2002, were available for analysis, providing a maximum of 5 years of follow-up for the present study.
For the analysis of incident CHD, we excluded 477 subjects with prevalent CHD (fatal and nonfatal myocardial infarction, fatal CHD, other CHD-attributable deaths, angina pectoris, coronary bypass surgery, and coronary angioplasty), leaving 1515 subjects for analysis. For incident stroke, we excluded 104 subjects with prevalent stroke, leaving 1888 subjects for analysis.
Participants underwent an assessment of cardiovascular risk factors during the course of the study.29- 32 Blood pressure was measured according to a standardized protocol,29 and the means of the first and second readings were computed for the systolic and diastolic values. Hypertension was defined as systolic blood pressure of 140 mm Hg or higher, diastolic blood pressure of 90 mm Hg or higher, or the combination of self-reported hypertension and use of antihypertensive medications. Diabetes was defined as fasting glucose concentration of 126 mg/dL (7.0 mmol/L) or greater or self-reported use of insulin or oral hypoglycemic agents. Medical history, medication use, cigarette smoking, and alcohol consumption were ascertained from questionnaires. Technicians assessed weight and standing height. Body mass index was calculated as weight in kilograms divided by height in meters squared. Blood sample collection, and processing and laboratory methods for fasting glucose concentration; total, low-density-lipoprotein, and high-density-lipoprotein cholesterol levels; and C-reactive protein (CRP) levels are described elsewhere.31,32,36 Vascular ultrasonography of the internal and common carotid arteries was used to determine the intima-media thickness (IMT).30 The common carotid artery IMT was determined to be the mean of the maximum wall thickness for near and far walls on both the left and right sides. We used the mean of the internal and common carotid IMT for adjustment.
All variables used in analyses were based on the 1997-1998 clinic examination, concurrent with retinal photography, except fasting glucose concentration, diabetes medications, and standing height, which were taken from the 1996-1997 examination, and data on high-density-lipoprotein cholesterol level, CRP level, and internal and common carotid IMT, which were obtained from the 1992-1993 examination.
Retinal arteriolar and venular calibers were categorized into quartiles. We defined large retinal venular caliber as present if caliber was in the fourth quartile of the population distribution.
We estimated the 5-year cumulative incidence of CHD or stroke, defined as 100 × (1 − Kaplan-Meier cumulative CHD- or stroke-free survival at 5 years). Follow-up time was defined as the number of days from the time of the 1997-1998 clinic visit to the date of the first cardiovascular event of interest (CHD or stroke), death, last contact, or June 30, 2002, whichever occurred first. We used Cox proportional hazards regression to estimate hazard rate ratios and 95% confidence intervals (CIs) for CHD and stroke in association with retinal vascular caliber and generalized retinal venular dilatation. We initially controlled for age, sex, race (white or black), and field center, and, further, for standard cardiovascular risk factors including systolic and diastolic blood pressure (in millimeters of mercury), diabetes (present or absent), fasting plasma glucose concentration (in milligrams per deciliter [millimoles per liter]), cigarette smoking status (never, former, or current), the number of pack-years of smoking, and low-density-lipoprotein and high-density-lipoprotein cholesterol (in milligrams per deciliter [millimoles per liter]) (model 1). We constructed 2 additional multivariable models adjusting for variables in model 1 plus body mass index and CRP level (model 2) and variables in model 1 plus mean common and internal carotid artery IMT (model 3). In these analyses, because retinal calibers were highly correlated (r = 0.61), arteriolar and venular calibers were included in the same model.37 We have shown that this approach accounts for confounding between arterioles and venules and produces fewer biased results.37 Analyses in which arteriolar and venular calibers were included in separate models were also conducted. Tests for trend across quartiles of retinal vascular diameter were conducted by fitting the grouped linear model; the P value was derived from a 2-sided Wald test.
To examine for interaction, analyses were repeated, stratifying the population by age group, sex, and the presence of diabetes and hypertension. All data analyses were performed using STATA version 8.0 (StataCorp, College Station, Tex).
Table 1 gives participant characteristics according to distribution of retinal vascular caliber, comparing first with fourth quartiles of arteriolar and venular caliber. Persons with smaller retinal arteriolar caliber (first quartile) were older; more likely to have hypertension, and higher systolic and diastolic blood pressure; and less likely to be current cigarette smokers. Persons with larger retinal venular caliber (fourth quartile) were younger and more likely to be African American; to have lower systolic but higher diastolic blood pressure, higher fasting glucose concentration, and total cholesterol level; and to be current cigarette smokers. Larger arterioles and venules were both associated with higher CRP levels.
There were 115 incident CHD events. Table 2 gives the 5-year cumulative incidence and rate ratio of CHD by quartiles of retinal caliber. Participants with larger venular caliber had higher cumulative incidence of CHD (11.7%; 95% CI, 8.7%-15.8%, vs 8.1%; 95% CI, 5.7%-11.6%, comparing largest with smallest venular caliber quartiles). After controlling for age, sex, race, and field center, participants with larger retinal venular caliber were more likely to develop incident CHD (rate ratio, 2.7, comparing largest with smallest venular caliber quartiles). This association was not substantially changed with additional adjustment for standard cardiovascular risk factors (rate ratio, 3.0; model 1), body mass index and CRP (rate ratio, 2.9; model 2), or carotid artery IMT (rate ratio, 3.0; model 3). Smaller retinal arteriolar caliber was associated with incident CHD (rate ratio, 2.0, comparing largest with smallest arteriolar caliber quartiles).
There were 113 incident stroke events. Table 3 shows associations of retinal vascular caliber and incident stroke. Participants with the largest retinal venular caliber had the highest incidence of stroke (8.4%; 95% CI, 6.0%-11.7%, vs 5.8%; 95% CI, 3.9%-8.4%, comparing fourth with first venular caliber quartiles). After controlling for age, sex, race, and field center, participants with larger retinal venular caliber were more likely to develop incident stroke (rate ratio, 2.1, comparing fourth with first venular caliber quartiles), which did not change after additional multivariable adjustment. Retinal arteriolar caliber was not associated with incident stroke.
In analyses in which arteriolar and venular caliber were entered separately in the models, associations of larger venular caliber with incident CHD (rate ratio, 2.0; 95% CI, 1.2-3.5, comparing fourth with first venular caliber quartiles; Ptrend = .01, adjusting for variables in model 1) and incident stroke (rate ratio, 2.1; 95% CI, 1.2-3.9; Ptrend = .02) were largely similar. A smaller retinal AVR was also associated with incident CHD (rate ratio, 1.9; 95% CI, 1.1- 3.5, comparing first with fourth AVR quartiles; Ptrend = .005) and stroke (rate ratio, 1.7; 95% CI, 0.9-3.1; Ptrend = .06). Focal retinal signs (retinopathy, focal arteriolar narrowing, and arteriovenous nicking) were not associated with incident CHD or stroke (data not shown).
Table 4 shows the association of large retinal venular caliber (venular caliber in the fourth quartile of the population) with incident CHD and stroke in the total cohort and stratified by age, sex, and diabetes and hypertension status. There were no substantial sex differences in these associations. For incident CHD, somewhat stronger associations were noted in older persons, persons with diabetes, and persons without hypertension. For incident stroke, stronger associations were noted in persons with diabetes and persons with hypertension. None of these interactions were statistically significant.
In this prospective, population-based study in older persons with a mean age of 79 years, we describe an association between larger retinal venular caliber (Figure, A), as quantified from retinal photographs, and 5-year risk of CHD and stroke. Participants with larger venular caliber were at higher risk of CHD and stroke, after adjusting for age, sex, race, blood pressure, glucose concentration, cigarette smoking status, low-density-lipoprotein and high-density-lipoprotein cholesterol levels, and other cardiovascular risk factors including CRP and common and internal carotid artery IMT. In addition, smaller retinal arteriolar caliber (Figure, B) was associated with 5-year risk of CHD but not stroke. Focal retinal microvascular signs, however, were not associated with either CHD or stroke.
Prevention of CHD and stroke in older persons requires an improved understanding of pathogenic mechanisms and predictive factors in this age group. Previous studies have principally focused on large vessel atherosclerosis and its risk factors (eg, lipid levels), but our findings, along with other evidence, suggest that a substantial proportion of cardiovascular events in older persons might relate to or share risk factors with changes in the microvasculature.7- 11
Our findings should be compared with those of other cohort studies that have recently examined the association of retinal vascular caliber, measured from retinal photographs using similar methods, and risk of various cardiovascular diseases. In the middle-aged (51-72 years) population from the ARIC study, a smaller AVR, representing either smaller arteriolar caliber or larger venular caliber, was associated with incident CHD19 and stroke.20 In the Beaver Dam Eye Study, a smaller AVR was associated with cardiovascular mortality only in younger (43-74 years) but not older (75-84 years) participants.21 Similarly, the Blue Mountains Eye Study found an association of larger venular caliber with incident CHD in younger (49-75 years) but not older (>75 years) men and women, and smaller arteriolar caliber with CHD in younger women only.23 The Rotterdam Study (participant age, >55 years) was the first to demonstrate significant associations of larger venular caliber with incident stroke in an older cohort.24 The current CHS extends these findings to incident CHD, supports the importance of studying retinal arteriolar and venular caliber separately, and provides further evidence that retinal vascular caliber may also provide predictive information about cardiovascular risk in older persons.
Our initial observations of these associations were interpreted as reflecting ischemic effects of arteriolar disease.38 In the ARIC study, the association of smaller AVR with CHD19 and stroke20 was thought to reflect smaller retinal arterioles, presumably from chronically elevated blood pressure, although associations for arteriolar and venular caliber were not reported separately. The findings from our study and others showing associations of enlarged retinal venules with cardiovascular diseases are not so readily explained on the basis of ischemia. There may be other explanations. Retinal venular dilatation has been hypothesized to reflect the effects of hypoxia and has been associated with inflammation and endothelial dysfunction.39,40 Animal studies showed that lipid hydroperoxide administration in the vitreous of rats results in infiltration of leukocytes in the retinal microvasculature and an increase in retinal venular but not arteriolar caliber.39 Dilation of the retinal vessels has been further attributed to increased nitric oxide production and release of cytokines resulting from activated endothelial cells.40 Recent clinical and epidemiologic studies support these observations, showing that larger venular caliber is associated with systemic biomarkers of inflammation or endothelial dysfunction.41- 44 In the Rotterdam Study,42 the Beaver Dam Eye Study,43 and the Multi-Ethnic Study of Atherosclerosis,44 larger venular caliber was associated with a range of inflammatory markers, including CRP and interleukin 6 concentrations. Thus, inflammation may underlie part of the association of larger venular caliber with CHD and stroke in the CHS. Further research into mechanisms underlying retinal venular dilation will clearly be useful in understanding the importance of these observations.
In the CHS, we found no association of focal retinal signs (retinopathy, focal arteriolar narrowing, and arteriovenous nicking) with either CHD or stroke, whereas other studies have found stronger associations of these focal retinal signs with both subclinical and clinical stroke,20- 22,45 particularly in younger persons.20,21,45 These differences may be related to selective mortality in the CHS (discussed in the following paragraph), a higher prevalence of comorbid conditions in older persons that may mask the cardiovascular risks associated with retinal signs, or weaker predictive associations of retinal signs in older persons compared with younger persons. However, we found that the association of larger venular caliber and incident CHD was not weaker but stronger in older (≥80 years) compared with younger (<80 years) participants.
The strengths of the CHS include a community-sampled study population, quantitative measurement of retinal vascular caliber, and standardized identification of incident CHD and stroke events. There are several important limitations. First, retinal photography was performed approximately 10 years after the start of the study, and a significant proportion of photographs were ungradeable because of media opacity or poor pupil dilation in this older population. Biases related to nonattendance for photography, ungradeable photographs, and selective mortality may lead to an underestimation of the true risks. For example, the lack of association of retinopathy (eg, microaneurysms and hemorrhages) and incident cardiovascular events may be related to higher mortality among participants with retinopathy at baseline who did not attend subsequent follow-up examinations. Second, although we adjusted for cardiovascular risk factors, residual confounding may still be present because some factors (eg, CRP and high-density-lipoprotein cholesterol levels) were measured 5 years before retinal photography was performed.
Our study has 2 potential clinical applications. First, our findings support the value of specifically targeting the microcirculation in reducing cardiovascular morbidity and mortality in older persons.7,8,46 There is increasing evidence that some antihypertensive agents (eg, angiotensin-converting enzyme inhibitors) may have direct beneficial effects on microvessel structure and function beyond their primary effect of lowering blood pressure.46 Such agents may, therefore, have added therapeutic value in preventing and treating cardiovascular disease. Second, the collective data from our study and others suggest that a quantitative measurement of retinal vascular caliber from retinal photographs may provide additional information for cardiovascular risk prediction.
In conclusion, we demonstrate an association of larger venular caliber and future risk of CHD and stroke and of smaller arteriolar caliber with risk of CHD in older persons. It is not known whether similar processes affecting venules occur in the heart and brain, but studying the mechanisms underlying venular dilation may provide important insights into various pathophysiologic processes of CHD and stroke in older persons.
Correspondence: Tien Yin Wong, MD, PhD, Centre for Eye Research Australia, University of Melbourne, 32 Gisborne St, Victoria 3002, Australia (email@example.com).
Accepted for Publication: August 29, 2006.
Author Contributions:Study concept and design: Wong and B. E. Klein. Acquisition of data: Wong, R. Klein, Siscovick, and Cushman. Analysis and interpretation of data: Wong, Kamineni, Sharrett, Siscovick, and Duncan. Drafting of the manuscript: Wong and Kamineni. Critical revision of the manuscript for important intellectual content: Wong, Kamineni, R. Klein, Sharrett, B. E. Klein, Siscovick, Cushman, and Duncan. Statistical analysis: Wong, Kamineni, and Siscovick. Obtained funding: Wong and R. Klein. Administrative, technical, and material support: Wong, R. Klein, B. E. Klein, and Cushman. Epidemologic and conceptually favorable content: Duncan.
Financial Disclosure: None reported.
Funding/Support: This study was supported by contract NHLBI HC-97-06 and grant R21-HL077166 from the National Heart, Lung, and Blood Institute. Dr Wong was also supported by the Sylvia and Charles Viertel Clinical Investigator Award, and Dr Duncan was also supported by a Centers of Excellence Grant of the Brazilian National Council for Scientific and Technological Development.