Introduction

The seventh report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-7),¹ the World Health Organization International Society of Hypertension (WHO-ISH) 2003 statement,² and the British Hypertension Society 2004 guidelines (BHS IV)³ consider hypertensive retinopathy target-organ damage. The 2007 European Society of Hypertension–European Society of Cardiology guidelines for the management of arterial hypertension (ESH-ESC 2007)⁴ stated that grade 3 and 4 hypertensive retinopathy as defined with the Keith-Wagener-Barker classification⁵ was associated with cardiovascular events. In contrast, the ESH-ESC guidelines argued that the ability of grade 1 and 2 retinal changes detected by fundal analysis to be used for prognosis was questionable on the basis of results of several cross-sectional studies.^6–8 However, those cross-sectional studies had relatively small sample sizes and did not directly examine a temporal relationship.

Clinical Perspective on p 2511

Several population-based studies^9–12 have provided evidence that hypertensive retinopathy or optic fundus abnormalities were associated with risks for cardiovascular disease independently of blood pressure levels in the general population. Results of these studies imply that mild hypertensive retinopathy might be associated with a risk for cardiovascular disease even in normotensive individuals.

Examining the association between mild hypertensive retinopathy and risk of cardiovascular disease among a population with and without hypertension can yield important evidence to identify high-risk individuals. However, limited data are available regarding this association.

The present study was conducted to examine whether mild hypertensive retinopathy graded by the Keith-Wagener-Barker classification could be a risk factor for cardiovascular mortality in individuals with and without hypertension.

Methods

Study Cohort and Population

In 1993, the Ibaraki prefectural government initiated a community-based large cohort study, known as the Ibaraki Prefectural Health Study, to obtain information on health status for the purpose of health education and policy making. The cohort included 97 042 individuals (33 130 men and 63 912 women) 40 to 79 years of age living in Ibaraki Prefecture who completed an annual health checkup, including an ophthalmoscopic examination in 1993.

We excluded 4067 persons (1133 men and 2934 women) from the analyses because of incomplete health checkup data. We also excluded 4996 persons (2043 men and 2953 women) because of a history of stroke or heart disease at baseline. Furthermore, we excluded 89 individuals (37 men and 52 women) because of grade 3 or 4 retinopathy, which was not considered mild retinopathy. Thus, 87 890 individuals (29 917 men and 57 973 women) were enrolled in the present study. Informed consent was obtained from community representatives to conduct an epidemiological study according to guidelines of the Council for International Organizations of Medical Science. The Ethics Committee of Ibaraki Prefecture approved this study.

Baseline Measurements

At baseline, nonmydriatic retinal photography of right eye was taken by a trained medical technologist using a CR6-45VAF device (Canon, Tokyo, Japan) and FUJICHROME TREBI 100C reversal film (Fujifilm, Tokyo, Japan). A photographic slide on the roll film was interpreted by a grader with information on the subject's sex, age, and systolic and diastolic blood pressures. In addition, a previous grade of Keith-Wagener-Barker classification⁵ was included in the information if it was completed in the last health checkup. Approximately 20 trained graders evaluated all photographic slides for the presence or absence of microvascular abnormalities using the Keith-Wagener-Barker classification.⁵

Height in stocking feet and weight in light clothing were measured, and body mass index was calculated as weight (kg) divided by height squared (m²). Blood pressure was measured on the right arm of seated subjects by trained observers using standard mercury sphygmomanometers.

Blood samples were drawn from seated subjects. Fasting was not needed. Serum total cholesterol and triglycerides were measured by enzymatic methods, and high-density lipoprotein cholesterol levels were measured by phosphotungstic acid magnesium methods. Plasma glucose level was measured by means of a glucose oxidase electrode method. A standard 12-lead ECG was obtained. Trained physicians evaluated the ECG for the presence or absence of atrial fibrillation and ST-T abnormality.

An interview was conducted to ascertain smoking status, number of cigarettes smoked per day, usual weekly intake of alcohol in go units (a Japanese traditional unit converted to grams of ethanol per day by 22 g ethanol per go unit), and history of stroke, heart disease, hypertension, dyslipidemia, and diabetes mellitus.

Follow-Up Surveillance

To ascertain deaths in the cohort, investigators conducted a systematic review of death certificates. Data from health checkups and date of death or moving were obtained from local governments. It is believed that all deaths that occurred in the cohort were ascertained, except for subjects who died after they had moved from their original community, in which case the subject was withdrawn from the study. Data on death are centralized at the Ministry of Health and Welfare, where the underlying causes of death are coded for the National Vital Statistics according to the International Classification of Diseases (ICD), 9th (1993–1994) and 10th (1995–2008) revisions. The underlying causes of death of this cohort were obtained from the Ministry of Health and Welfare. Record matching between the residence certificates and the underlying causes of death was done by residential area, sex, birthday, and death date.

Definitions

Subjects were classified with regard to their retinal photography based on the Keith-Wagener-Barker classification⁵: normal, grade 1, or grade 2. According to the classification, grade 1 is defined as mild narrowing or sclerosis of the retinal arterioles. Grade 2 is defined as moderate to marked sclerosis of the retinal arterioles, moderate narrowing of the retinal arterioles, or arteriosclerotic retinopathy or thrombosis of retinal veins. Grade 3 is defined as angiospastic retinopathy, characterized by edema, cotton-wool patches, and hemorrhages in the retina, in addition to marked sclerosis of the retinal arterioles. Grade 4 is defined as measurable edema of the disks in addition to grade 3 pictures. In the present study, subjects without retinopathy were graded as normal, and subjects with grade 3 or 4 retinopathy were excluded because these categories were not considered mild retinopathy.

Blood glucose level was divided into 3 categories: normal (6.1 mmol/L fasting or <7.8 mmol/L nonfasting), prediabetes (6.1–6.9 mmol/L fasting or 7.8–11.0 mmol/L nonfasting), and diabetes mellitus (≥7.0 mmol/L fasting or ≥11.1 mmol/L nonfasting). Smoking habits were divided into 4 categories: never smoker, ex-smoker, currently smoking <20 cigarettes a day, and currently smoking ≥20 cigarettes/d. Alcohol intake was divided into 4 categories: never, sometimes (not almost every day), almost every day (<44 g/d), and almost every day (≥44 g/d).

Cause-specific mortality was classified according to the ICD code of the underlying cause of death. Total cardiovascular disease deaths were identified as codes 393 through 459 in ICD-9 and as codes I00 through I99 in ICD-10. Total stroke deaths were identified as code 430 through 438 in ICD-9 and as code I60 through I69 in ICD-10. Cerebral infarction deaths were identified as codes 433 through 434 and 437.7 in ICD-9 and as code I63 and I69.3 in ICD-10. Intracerebral hemorrhage deaths were identified as codes 431 through 432 in ICD-9 and as codes I61 and I69.1 in ICD-10. Ischemic heart disease deaths were identified as codes 410 through 414 in ICD-9 and as codes I20 through I22 and I24-I25 in ICD-10.

Statistical Analysis

The P values for difference of baseline characteristics according to the Keith-Wagener-Barker classification were calculated by an ANOVA for age, body mass index, systolic blood pressure, diastolic blood pressure, serum total cholesterol level, and serum high-density lipoprotein cholesterol level and by χ² text for antihypertensive medication use, antidyslipidemic medication use, diabetes mellitus, antidiabetic medication use, atrial fibrillation, ST-T abnormality, smoking status, and alcohol intake.

Risk ratios and 95% confidence intervals (CIs) for all-cause and cause-specific mortality according to the Keith-Wagener-Barker classification were calculated with reference to the normal grade by use of Cox proportional hazards regression models. Covariates included age, body mass index, systolic blood pressure, antihypertensive medication use (yes or no), serum total cholesterol level, serum high-density lipoprotein cholesterol level, antidyslipidemic medication use (yes or no), blood glucose level (normal, prediabetes, and diabetes mellitus), antidiabetic medication use (yes or no), atrial fibrillation (yes or no), ST-T abnormality (yes or no), smoking status (never smoker, ex-smoker, currently smoking <20 cigarettes a day, currently smoking ≥20 cigarettes/d), and alcohol intake (never, sometimes, <44 g/d almost every day, and ≥44 g/d almost every day). The analysis was also stratified by hypertension status: hypertensive subjects (systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, and/or use of hypertensive medication) and normotensive subjects (systolic blood pressure <140 mm Hg, diastolic blood pressure <90 mm Hg, and no use of hypertensive medication). The P values for trend were calculated by multivariable Cox proportional hazard models adjusted for the variables described above. All statistical tests were 2 sided, and values of P<0.05 were considered statistically significant. All statistical analyses were conducted with SAS version 9.1.3 (SAS Institute, Inc, Cary, NC).

Results

Sex-specific baseline characteristics of this study subjects according to the Keith-Wagener-Barker classification of ophthalmoscopy are shown in Table 1. Statistically significant differences according to the Keith-Wagener-Barker classification were found for all covariables at baseline except antidyslipidemic medication use in men. Mean age was higher with higher grades of the Keith-Wagener-Barker classification in both sexes. The means of systolic and diastolic blood pressures and the proportion of antihypertensive medication use were also higher with the higher grades in both sexes.

During follow-up through 2008, a mean of 14.1 years (13.6 years in men and 14.3 years in women), 12 946 total deaths (7001 in men and 5945 in women), 3697 cardiovascular deaths (1801 in men and 1896 in women), 1746 total stroke deaths (841 in men and 905 in women), 990 cerebral infarction deaths (546 in men and 444 in women), 453 intracerebral hemorrhage deaths (206 in men and 247 in women), and 957 ischemic heart disease deaths (511 in men and 446 in women) were observed.

Table 2 shows hazard ratios of all-cause and cause-specific mortality according to the Keith-Wagener-Barker classification among all study subjects. The multivariable hazard ratios for all-cause mortality, total cardiovascular disease, total stroke, and cerebral infarction were higher with higher grades of the Keith-Wagener-Barker classification in both sexes (P for trend <0.01). In addition, the multivariable hazard ratios for intracerebral hemorrhage were higher with higher grades of the Keith-Wagener-Barker classification among women (P for trend <0.01). Compared with normal grade, the multivariable hazard ratios for all-cause mortality were significantly high for grades 1 and 2 among men and for grade 2 among women. The multivariable hazard ratios for total cardiovascular disease mortality, total stroke mortality, and cerebral infarction mortality were significantly higher for grades 1 and 2 among men and women. In addition, a significantly high hazard ratio for intracerebral hemorrhage mortality for grade 2 was found among women but not among men. The multivariable hazard ratios for ischemic heart disease mortality were not statistically significant in both sexes.

Table 3 shows the multivariable hazard ratios of all-cause and cause-specific death according to Keith-Wagener-Barker classification of ophthalmoscopy, stratified by hypertensive status. Compared with the normal grade, both hypertensive and normotensive subjects with grade 1 or 2 showed a significantly higher multivariable hazard ratios for all-cause mortality, total cardiovascular mortality, and total stroke mortality for both sexes. In hypertensive subjects, the multivariable hazard ratio for total cardiovascular mortality was significantly high for grade 1 among men and for grades 1 and 2 among women. The multivariable hazard ratios for total stroke mortality were significantly high for grades 1 and 2 among hypertensive men and women. In nonhypertensive subjects, the multivariable hazard ratios for total cardiovascular mortality and total stroke mortality were significantly high for grade 2 in both sexes. In addition, a significantly high multivariable hazard ratio for ischemic heart disease mortality was found among nonhypertensive men with grade 2 disease.

Discussion

To the best of our knowledge, the results of the present study are the first to show that mild hypertensive retinopathy classified by the Keith-Wagener-Barker classification is a risk factor for cardiovascular mortality independently of other cardiovascular risk factors in both sexes with and without hypertension.

In 1939, Keith et al⁵ reported a graded decline in survival during an 88-month follow-up period from grade 1 to grade 4 among 219 hypertensive patients. Their cumulative all-cause mortality rates at 5 years after baseline survey were 30% for grade 1, 46% for grade 2, 80% for grade 3, and 99% for grade 4. Several follow-up studies^13–16 confirmed the association between hypertensive retinopathy classified by the Keith-Wagener-Barker classification and all-cause mortality among hypertensive patients. Kato et al¹⁷ reported that the cumulative all-cause mortality rates at 5 years after baseline survey were 2.4% for grade 1 and 16.6% for grade 2 among 170 Japanese patients with fundus hypertonicus.

The ESH-ESC 2007⁴ states that patients with grade 3 and 4 retinopathy are at high risk for cardiovascular events, and the WHO-ISH 2003 statement² and the BHS IV³ regard retinopathy of grades 3 and 4 as target-organ damage. The JNC-7¹ also indicates that retinopathy (without grading) is target-organ damage. However, the ESH-ESC 2007⁴ questioned the ability of grade 1 and 2 retinopathy detected by fundal analysis to predict prognosis on the basis of several cross-sectional studies.^6–8 Those cross-sectional studies^6–8 had a relatively small sample size and did not directly examine a temporal relationship. In contrast, the present large prospective cohort study supports that mild degrees of retinopathy can predict the risk of stroke mortality in individuals with hypertension.

The Atherosclerosis Risk in Communities (ARIC) Study,^10,12 a population-based cohort study of ≈10 000 men and women 51 to 72 years of age in 4 US communities, provided evidence that the multivariable hazard ratio of retinopathy (no use of the Keith-Wagener-Barker classification) was 2.58 (95% CI, 1.59–4.20) for incident total stroke and 2.60 (95% CI, 1.55–4.34) for incident ischemic stroke.¹⁰ In that study, a 6-year mean arterial blood pressure was included as a covariate. The Shibata Study,⁹ a cohort study of 2302 residents ≥40 of age in a Japanese provincial city, provided evidence that an association between optic fundus abnormality and incident stroke in men (multivariable hazard ratio, 3.42 [95% CI, 1.03–11.31] for all stroke, 4.54 [95% CI, 1.05–19.57] for cerebral infarction, and 8.53 [95% CI, 1.04–69.21] for intracerebral hemorrhage) but not in women. The Beaver Dam Eye Study,¹¹ a nested case-control study of 413 cases and 1198 controls 43 to 84 years of age, provided evidence that the multivariable odds ratio of retinopathy was 1.8 (95% CI, 1.2–2.7) for total cardiovascular mortality after adjustment for cardiovascular risk factors, including systolic blood pressure. The present study is consistent with previous findings regarding the association between hypertensive retinopathy and risk for stroke mortality.

The potential mechanism of the independent association between hypertensive retinopathy and stroke disease is unclear. However, a significant association between “masked hypertension” and risk of cardiovascular mortality and stroke incidence has been reported.¹⁸ Masked hypertension, which may be considered nonhypertension, may yield both hypertensive retinopathy and cardiovascular disease. Retinal arteriolar narrowing has been shown to precede the development of hypertension, which was defined as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg, or the combination of a self-reported diagnosis of high blood pressure and use of antihypertensive drugs.¹⁹ Hypertensive retinopathy may predict future hypertension, and future hypertension may lead to cardiovascular disease. Ophthalmoscopy may identify masked hypertension or future hypertension among a normotensive population. Furthermore, retinal microvascular changes have been reported to be associated with future cerebral ventricular enlargement after adjustment for 6-year mean arterial blood pressure.²⁰

The strength of the present study is the use of a large population-based cohort in which sex-stratified and cause-specific analyses were available, compared with previous studies.^5,9–17 In addition, all blood samples were measured by the same device, reagents, and quality control program.

On the other hand, our study had several limitations. First, we did not have incidence data for cardiovascular disease. Previous studies, however, provided evidence that death certificate diagnosis of stroke subtypes is valid as a result of the high prevalence of computed tomography scans or magnetic resonance imaging use in hospitals in Japan.^21,22 Moreover, in the present study sample, a validation study confirmed the validity of stroke.^22,23 In addition, previous studies reported that three fourths of death certificate diagnoses of ischemic heart disease have been found to be correct in Japan.^24,25 Second, subjects of this study were participants of health checkup for residents with an ≈40% response rate, so a “healthy” participant effect cannot be negated. Third, direct ophthalmoscopy was not performed to identify the grade of retinopathy. Retinopathy beyond the 45 ° photograph might have been missed. However, the nonmydriatic camera is the preferred method of screening because retinal photography with interpretation by an ophthalmologist is more sensitive than direct ophthalmoscopy by general physicians.²⁶ Fourth, despite a previous study that provided the predictive value of arteriovenous nicking and focal arteriolar narrowing,¹⁰ those variables were not examined in the present study. Finally, parental history of cardiovascular disease was not available. Further study is warranted to clarify the generalizability and effect of family history of cardiovascular disease.

Conclusion

Even a mild degree of hypertensive retinopathy classified by the Keith-Wagener-Barker classification is a risk factor for cardiovascular mortality independently of cardiovascular risk factors among men and women with and without hypertension.

Sources of Funding

This research was supported by grant-in-aids from the Ministry of Health, Labor and Welfare, Health and Labor Sciences Research Grants, Japan (Research on Health Services: H17-Kenkou-007; Comprehensive Research on Cardiovascular and Life-Style Related Diseases: H18-Junkankitou[Seishuu]-Ippan-012; Comprehensive Research on Cardiovascular and Life-Style Related Diseases: H20-Junkankitou[Seishuu]-Ippan-013; Intractable Diseases Conquest Research: H21-Nanchi-Ippan-059; and Comprehensive Research on Cardiovascular and Life-Style Related Diseases: H23-Junkankitou[Seishuu]-Ippan-005).

Disclosures

None.

Footnotes