Systemic associations of retinal microvascular signs: a review of recent population-based studies


  • Tien Yin Wong,

    1. Centre for Eye Research Australia, University of Melbourne, Australia, 32 Gisborne Street, East Melbourne 3002, Australia
    2. Singapore Eye Research Institute, National University of Singapore, Singapore
    Search for more papers by this author
  • Rachel McIntosh

    1. Centre for Eye Research Australia, University of Melbourne, Australia, 32 Gisborne Street, East Melbourne 3002, Australia
    Search for more papers by this author

Correspondence and reprint requests to: Tien Wong.
Tel.: +61 3 9929 8352; Fax: +61 3 9662 3859.
E-mail address:


Retinal microvascular signs, such as generalized retinal arteriolar narrowing, focal arteriolar narrowing, arteriovenous nicking and retinal haemorrhages, microaneurysms and cotton wool spots, are common fundus findings in the general population, even in individuals without hypertension or diabetes. Recent population-based studies have provided new insights into the systemic associations and clinical significance of these retinal signs. Studies show that these retinal microvascular signs are strongly associated with elevated blood pressure (BP). Generalized retinal arteriolar narrowing may be associated with markers of inflammation and risk of diabetes and hypertension. Retinal haemorrhages, microaneurysms and cotton wool spots are associated with risk of subclinical and clinical stroke, cognitive impairment, renal dysfunction and cardiovascular mortality, independent of BP and cardiovascular risk factors. A consistent pattern of association between retinal microvascular signs and ischaemic heart disease has not been demonstrated. This suggests that patients with some retinopathy signs (retinal haemorrhages, microaneurysms and cotton wool spots) may benefit from a careful systemic evaluation and, if supported by further research, appropriate risk reduction therapy.


Hypertension remains one of the leading causes of morbidity and mortality worldwide (Wolf-Maier et al., 2003). Despite increasing recognition of its importance and development of new therapeutic agents to control blood pressure (BP), suboptimal treatment of hypertension remains a major public problem (Hajjar and Kotchen, 2003). Studies suggest that approximately 40% of individuals with hypertension are not on treatment and more than 60% of hypertensive patients are not being controlled to appropriate BP levels (<140/90 mmHg) (World Health Organization/International Society of Hypertension Writing Group, 2003).

Retinal microvascular signs are common fundus findings in the general adult population (Wong et al., 2001b). These signs, which include generalized and focal arteriolar narrowing, arteriovenous (AV) nicking, isolated retinal haemorrhages, microaneurysms and cotton wool spots, are often associated with elevated BP, and traditionally referred to as ‘hypertensive’ retinopathy (Tso and Jampol, 1982). However, many individuals with these retinal vascular signs do not have a known history of hypertension. In fact, epidemiological studies indicate that retinal microvascular signs can be detected from fundus photographs in 3–13% of the general adult population above 40 years (Wong et al., 2001b).

What is the clinical significance of these hypertensive retinopathy signs? How should these signs be classified? Do patients with certain retinopathy signs need physician referral and further cardiovascular assessment? Since the journals’ last review of hypertensive retinopathy in 2001 (Hurcomb et al., 2001), clinical guidelines for hypertension management have been updated, and there has also been a series of new population-based studies describing the relationship of retinal microvascular signs, as quantified from fundus photographs, to various systemic diseases in the general population.

The purpose of this article is to review recent studies on the systemic associations of retinal microvascular signs, and to discuss their implications in the context of current hypertension management.

New hypertension guidelines

In the United States, the Seventh Report of the Joint National Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-VII) updated its recommendation of hypertension management in 2003 (Chobanian et al., 2003; Jones and Hall, 2004). The key messages of JNC-VII included the following: first, in individuals older than 50 years of age, systolic BP of >140 mmHg is a more important cardiovascular risk factor than diastolic BP. Secondly, individuals who are normotensive at 55 years of age will have a 90% lifetime risk of developing hypertension. Thirdly, individuals with systolic BP 120–139 mmHg or diastolic BP 80–89 mmHg are defined as ‘pre-hypertensive’ and require appropriate management (e.g. lifestyle modifications) to reduce cardiovascular risk and prevent progression to hypertension. Fourthly, in uncomplicated hypertension, thiazide diuretic should be used as the first line treatment for most individuals, either alone or in combination with other drugs; in specific circumstances, other drug classes, such as angiotensin-converting enzyme (ACE) inhibitors, angiotensin II receptor blockers, beta-blockers or calcium channel blockers, may be preferred. Frequently, two or more medications are required to achieve target BP of <140/90 mmHg. Finally, beginning at 115/75 mmHg, cardiovascular risk doubles for each increment of 20/10 mmHg in systolic and diastolic BP. Therefore, in individuals whose BP is 20/10 mmHg above the systolic and diastolic BP target, initiation of therapy using two agents should be considered. Updated guidelines from the European Society of Hypertension-European Society of Cardiology (ESH/ESC) (Guidelines Committee, 2003), and the British Hypertension Society (BHS) (Williams et al., 2004) have a similar message, ‘the lower the pressure, the greater the benefit’.

The new guidelines suggest that physicians should aim for adequate treatment of BP traditionally considered as within the ‘normal’ range. Physicians should also aim for a more aggressive approach to lowering BP in patients with concomitant cardiovascular risk factors (such as diabetes) and those with evidence of target organ damage.

Systemic associations

There have been several population-based studies evaluating the relationship of retinal microvascular signs with systemic diseases in the general population (Table 1). In general, these studies suggest that retinal microvascular signs can be more reliably defined from a standardized assessment of fundus photographs, with high reproducibility for retinal haemorrhages and microaneurysms (kappa values ranged from 0.80 to 0.99) and fair to moderate reproducibility for AV nicking and focal arteriolar narrowing (0.40–0.79) (Couper et al., 2002; Sherry et al., 2002; Wong et al., 2003c). In these studies, generalized arteriolar narrowing was estimated from an assessment of retinal vessel diameters using an imaging technique based on Hubbard et al. (1999). This technique appears to have substantial reproducibility (intra-class correlation coefficient of 0.80–0.99) (Leung et al., 2003a; Wong et al., 2003a,c; Ikram et al., 2004).

Table 1.  Summary of population-based studies
Atherosclerosis Risk in Communities (ARIC) StudyFour US communities: North Carolina; Mississippi; Minneapolis and Maryland12 642 persons aged 51–72 years, white (77.4%) and black (22.6%) ethnicity
Cardiovascular Health Study (CHS)Four US communities: North Carolina; Pennsylvania; California and Maryland2824 persons aged 69–97 years, white (84.2%) and black (15.8%) ethnicity
Beaver Dam Eye Study (BDES)Beaver Dam, Wisconsin, USA4926 persons aged 43–86 years, white ethnicity
Blue Mountains Eye Study (BMES)West of Sydney, Australia3654 persons aged 49 and older, white ethnicity
Rotterdam Eye Study (Rotterdam)Rotterdam, Holland5674 persons aged 55 and older, white ethnicity

Associations with blood pressure

The association of retinal microvascular signs with BP is strong, graded, consistent and seen even in individuals without a known history of hypertension (Klein et al., 1994, 1997, 2000; Sharp et al., 1995; Yu et al., 1998; Sharrett et al., 1999; Wong et al., 2002a; Leung et al., 2003b; Wang et al., 2003). Data from the Beaver Dam Eye Study among 4929 adults aged 43–86 years living in Wisconsin showed that both the prevalence (Klein et al., 1994) and the 5-year incidence (Klein et al., 1997) of various retinal microvascular signs were significantly higher in hypertensive than normotensive participants, with hypertensive individuals 50–70% more likely to have retinal haemorrhages and microaneurysms, 30–40% more likely to have focal arteriolar narrowing, and 70–80% more likely to have AV nicking than normotensive people. The study further showed that hypertensive participants whose BP was elevated despite use of anti-hypertensive medications, an indication of ‘poorer’ control, had higher risk of developing retinopathy than individuals whose BP was controlled with medication.

Generalized retinal arteriolar narrowing has long been regarded as an early but characteristic sign in hypertension (Tso and Jampol, 1982; Walsh, 1982). Previous studies assessed this sign qualitatively using clinical ophthalmoscopy, with poor reproducibility (Aoki et al., 1977; Dimmitt et al., 1989; Fuchs et al., 1995). More recent population-based studies have used imaging techniques to quantify generalized arteriolar narrowing from digitized photographs using a formula first described by Parr and Spears (1974) and later modified by Hubbard et al. (1999) and Knudtson et al. (2003). Using this method, the Atherosclerosis Risk in Communities (ARIC) study reported that retinal arteriolar diameter is strongly and inversely related to increasing levels of BP (Sharrett et al., 1999), a finding confirmed in four other populations (Wong et al., 2002a; Leung et al., 2003b; Wong et al., 2003a; Ikram et al., 2004). The Beaver Dam Eye Study showed that after adjusting for age, gender, diabetes, smoking and other vascular risk factors, each 10 mmHg increase in mean arterial BP was associated with a 6 μm (or 3%) decrease in retinal arteriolar diameters (Wong et al., 2003a). While this supports the strong link between generalized arteriolar narrowing and hypertension, it also suggests that a clinical examination is unlikely to detect such subtle degrees of arteriolar narrowing.

An important issue is whether retinal microvascular signs are markers of cumulative, long-term BP damage, or only reflect a transient effect of acutely elevated BP. Previous studies have already demonstrated strong correlations of retinopathy grades with left ventricular hypertrophy (Dahlof et al., 1992a; Shigematsu et al., 1995; Saitoh et al., 1998) and microalbuminuria (Pontremoli et al., 1995), two other indicators of long-term target organ damage, in persons with hypertension. Recent studies have investigated this issue further by analysing the associations of specific retinopathy signs with both concurrent BP levels (measured at the time of the retinal assessment) and historical BP levels (measured 5–8 years prior to the retinal assessment) (Sharrett et al., 1999; Wong et al., 2002a; Leung et al., 2004). These studies show that generalized arteriolar narrowing and AV nicking are independently related to past BP levels, suggesting that they reflect persistent arteriolar damage from long-term hypertension (Sharrett et al., 1999; Wong et al., 2002a; Leung et al., 2004). In contrast, focal arteriolar narrowing, retinal haemorrhages, microaneurysms and cotton wool spots are related to only concurrent but not past BP levels, and may therefore reflect more fleeting changes related to acute BP elevation (Sharrett et al., 1999; Wong et al., 2002a).

Three new analyses have also provided new insights into the relation of generalized retinal arteriolar narrowing to subsequent development of systemic hypertension (Smith et al., 2004; Wong et al., 2004c,e). Prospective data from the ARIC study suggest that normotensive participants, who had generalized arteriolar narrowing at baseline, were 60% more likely to be diagnosed with hypertension over a 3-year period than individuals without arteriolar narrowing (relative risk 1.62, 95% CI 1.21–2.18) (Wong et al., 2004c). Those individuals with the most severe arteriolar narrowing had the largest 3-year change in BP, and this association was not related to pre-existing BP, body mass index and other known hypertension risk factors. Data from the Beaver Dam and the Blue Mountains cohorts have shown essentially identical results (Smith et al., 2004; Wong et al., 2004e), providing strong evidence that generalized arteriolar narrowing, possibly reflecting similar peripheral arteriolar changes elsewhere in the body, is a pre-clinical marker of hypertension.

Finally, there have been several small clinical reports that have indicated the possibility that hypertensive retinopathy signs may regress with control of BP (Bock, 1984; Dahlof et al., 1992b). It is unclear whether anti-hypertensive medications with putative direct beneficial effects on microvascular structure (e.g. ACE inhibitors) would reduce retinopathy damage beyond the effects of lowered BP (Morishita et al., 1992; Ohta et al., 1995). In a small study of 28 mildly hypertensive patients randomized to treatment with enalapril or hydrochlorothiazide, retinal arteriolar wall opacification (although not other retinopathy signs) was significantly reduced after 26 weeks of enalapril treatment (Bock, 1984). In contrast, hydrochlorothiazide did not have any effect on the retinopathy signs. As yet, however, there have been no controlled clinical trials demonstrating that specifically reducing hypertensive retinopathy signs also decreases cardiovascular risk.

Associations with atherosclerosis

In contrast to the strong association with BP, a consistent link between retinal microvascular signs with direct measures of atherosclerosis have not been found. Carotid artery intima-media thickening and arterial wall plaque, as defined from carotid ultrasound, are well-established measures of atherosclerosis. In the ARIC study, while controlling for BP, generalized arteriolar narrowing was associated with carotid artery plaque but not intima-media thickening; AV nicking was associated with carotid artery intima-media thickening but not carotid plaque; and focal arteriolar narrowing was not related to either carotid artery measure (Klein et al., 2000). In the Rotterdam Eye study, generalized arteriolar narrowing was associated with carotid artery intima-media thickening but not other features of systemic atherosclerosis (Ikram et al., 2004). The Cardiovascular Health Study (CHS) found no consistent independent association of hypertensive retinopathy signs with carotid atherosclerosis (Wong et al., 2003c).

Possible relationships of retinal microvascular signs with earlier indicators of atherosclerosis and novel vascular risk factors have also been examined. These include arterial stiffness, inflammation and endothelial dysfunction. Arterial stiffness (the loss of elasticity) in large and medium-sized arteries has been associated with stroke and is a risk factor for hypertension development (Arnett et al., 1994). In the ARIC study, carotid arterial stiffness was strongly correlated with generalized retinal arteriolar narrowing, although it is unclear if this simply reflects the association between arterial stiffness and elevated BP.

Inflammation plays a key role in the pathogenesis of both large atherosclerosis and microvascular diseases, including diabetic retinopathy (Ridker et al., 1997, 2000; Ross, 1999). The ARIC study reported that generalized retinal arteriolar narrowing was related to elevated white cell counts, elevated fibrinogen levels and reduced albumin levels independent of BP, indicating that systemic inflammatory processes may also contribute to the development of this retinal vascular change (Klein et al., 2000). Other studies, however, have been unable to confirm associations of hypertensive retinopathy with C-reactive protein, a more specific inflammatory marker (Wong et al., 2003c; Ikram et al., 2004).

New studies suggest that endothelial dysfunction, another pathogenic factor in atherosclerosis, may also contribute towards the development of hypertensive retinopathy. In a controlled clinical study, patients with hypertension had impaired endothelial function in the retinal circulation when compared with normotensive controls (Delles et al., 2004). The retinal endothelial function was restored when hypertensive subjects were treated with angiotensin II receptor blocker. Data from the ARIC study showed that AV nicking was associated with elevated levels of von Willebrand factor and Factor VIII, two haematological markers of endothelial dysfunction (Klein et al., 2000).

These data suggest that further research is needed to understand the link between atherosclerosis inflammation, endothelial function and retinal microvascular signs.

Associations with diabetes and the metabolic syndrome

The metabolic syndrome is an entity comprising abdominal obesity, dyslipidaemia [low high-density lipoprotein (HDL) cholesterol levels, high triglyceride levels], hyperglycaemia and elevated BP (Ford et al., 2002). Because of its close association with elevated BP, retinal microvascular signs may also be linked with other features of the metabolic syndrome. In the Hoorn Study in the Netherlands, hyperglycaemia, hypertension and abdominal obesity were independent risk factors for development of retinal haemorrhages, microaneurysms, hard exudates and cotton wool spots in the general population (van Leiden et al., 2003). In an ARIC study analysis, after controlling for BP, AV nicking was related to larger waist circumference and low HDL-cholesterol, focal arteriolar narrowing was related to larger waist circumference, generalized arteriolar narrowing was related to larger waist circumference, and retinal haemorrhages, microaneurysms and cotton wool spots were related to hyperglycaemia (Wong et al., 2004a). The ARIC study showed that generalized retinal arteriolar narrowing may in fact be a pre-clinical marker of overt diabetes, even amongst people at lower risk of developing diabetes, such as individuals without a family history of diabetes (Wong et al., 2002e).

Other studies, however, have not demonstrated a consistent pattern of associations between retinopathy signs and specific metabolic syndrome components. The Rotterdam Eye study showed that generalized arteriolar narrowing was associated with increased body mass index, but not with increased waist hip ratio or elevated serum total cholesterol (Ikram et al., 2004). The Blue Mountains Study has not been able to demonstrate associations of retinal vascular signs with dyslipidaemia (Leung et al., 2005).

Associations with stroke and other cerebrovascular diseases

Previous studies in people with hypertension have shown that retinopathy signs are associated with both subclinical and clinical stroke (Okada et al., 1976; Svardsudd et al., 1978; Tanaka et al., 1985; Nakayama et al., 1997). New studies, using photographs to define retinopathy signs, have confirmed their strong association with cerebrovascular disease. In the ARIC study, individuals with retinal haemorrhages, microaneurysms and cotton wool spots defined from photographs, were two to four times more likely to develop an incident clinical stroke within 3 years, after factoring in the effects of BP, cigarette smoking, lipids and other risk factors (Wong et al., 2001a). Among participants without stroke or transient ischaemic attack, these retinopathy signs were also related to cognitive decline, as defined from standardized neuropsychological tests over a 6-year period (Wong et al., 2002d), and cerebral white matter lesions (Wong et al., 2002b) and atrophy (Wong et al., 2003d), as defined from cranial magnetic resonance imaging (MRI).

The ARIC study further demonstrated that individuals who had both cerebral MRI lesions and retinopathy signs were at substantially higher risk of incident clinical stroke than those without either abnormality (relative risk 18.1, 95% CI 5.9–55.4) (Wong et al., 2002b).

The strong link between certain retinopathy signs and stroke is supported by data from the CHS, in which retinopathy signs were related to prevalent clinical stroke (Wong et al., 2003c), and data from the Beaver Dam and Blue Mountains cohort, in which retinopathy signs were related to stroke mortality. (Mitchell et al., 2002; Wong et al., 2003b). In a smaller clinical series of 179 patients with symptomatic atherosclerotic disease (recent ischaemic stroke, myocardial infarction or peripheral arterial disease), retinal arteriolar narrowing and sclerosis, defined from photographs, were significantly associated with MRI-defined white matter lesions and lacunar infarcts (Kwa et al., 2002). Other studies have linked functional and pathological abnormalities of the retinal blood vessels to stroke. For example, retinal microvascular flow appears to be reduced in the presence of subclinical white matter lesions and lacunar infarction (Schneider et al., 1993), and in stroke descendents, retinal and cerebral arterioles share similar histopathological features (Goto et al., 1975).

Associations with ischaemic heart disease and heart failure

In contrast to cerebrovascular disease, there are less consistent data regarding whether hypertensive retinopathy signs are related to ischaemic heart disease (IHD). Previous studies, again based on clinical ophthalmoscopy, have linked hypertensive retinopathy signs with ischaemic T-wave changes on electrocardiogram (Breslin et al., 1966a,b), and with severity of coronary artery stenosis on angiography (Michelson et al., 1979). In a dyslipidaemia clinical of high-risk men conducted in the 1970s, hypertensive retinopathy signs predicted a doubling of the risk of IHD events (relative risk 2.1, 95% CI 1.0–4.2), independent of vascular risk factors (Duncan et al., 2002).

Two studies have suggested a possible gender difference in the relationship between retinal microvascular signs and risk of IHD. In the National Health Examination Survey, a stronger association between retinal arteriolar signs and IHD was found in women than men (Gillum, 1991). Similarly, in the ARIC study, generalized retinal arteriolar narrowing predicted incident IHD only in women (relative risk 2.2, 95% CI 1.0–4.6) but not in men (relative risk 1.1, 95% CI 0.7–1.8) (Wong et al., 2002b). These results suggest that systemic microvascular processes may play a greater role in the development of heart disease in women than men (Maguire, 2003). If confirmed by further studies, this finding may have important implications for both the prevention and treatment (e.g. decisions regarding revascularization surgery vs medical therapy) of heart disease in women (Cannon and Balaban, 2000; Natarajan et al., 2003).

A common cause of morbidity, hospitalization and mortality in hypertensive patients is congestive heart failure (Jessup and Brozena, 2003). In the ARIC study, after controlling for pre-existing IHD, BP, diabetes and other risk factors, retinal haemorrhage, microaneurysms and cotton wool spots were associated with a twofold higher risk of incident congestive heart failure (relative risk 1.96, 95% CI 1.52–2.56) (T.Y. Wong, unpublished data). Among participants without pre-existing IHD, diabetes or hypertension, these retinal signs predicted a threefold higher risk of heart failure events (relative risk 2.97, 95% CI 1.49–5.91). These findings suggest that microvascular damage to the myocardium from hypertension and other processes may play an important aetiological role in the development of heart failure.

Associations with renal disease

The close association between diabetic retinopathy and nephropathy is well recognized and documented (Chavers et al., 1994; Klein et al., 1999). There have been fewer studies that have investigated a possible link between hypertensive retinopathy signs and kidney disease. Experimental investigations show that pathological changes in the retinal and renal microcirculation are highly correlated in spontaneously hypertensive rats (Nag et al., 1980).

In the ARIC study, individuals with AV nicking, retinal haemorrhages, microaneurysms and cotton wool spots were 1.4–2.7 times as likely to develop renal dysfunction compared with those without these retinopathy signs (Wong et al., 2004b). This association was independent of BP, diabetes and other risk factors and was seen in persons without diabetes or hypertension. It is possible that this association reflects common processes (e.g. endothelial dysfunction, inflammation) in the development of hypertensive damage in the eye and kidneys.

Classification and conclusion

This review suggests that the retinal microvascular signs differ in their associations with systemic diseases (Table 2). Isolated retinal haemorrhages, microaneurysms and cotton wool spots appear to be associated with risk of subclinical and clinical stroke, other cerebrovascular outcomes, renal dysfunction and cardiovascular mortality. In contrast, the systemic associations for generalized retinal arteriolar narrowing, focal arteriolar narrowing and AV nicking appear to be weaker and less consistent.

Table 2.  Systemic associations of retinal microvascular signs, selected population-based studies
Retinal signsSystemic associationsStrength*Studies and references†
  1. *Strength of association: Relative risk/odds ratio >2.0 (Strong), 1.5 to 2.0 (Moderate), <1.5 (Weak).

  2. †See Table 1 for abbreviations.

Retinal haemorrhagesCurrent blood pressureStrongARIC (Sharrett et al., 1999), BMES (Wang et al., 2003), BDES (Wong et al., 2003a), CHS (Wong et al., 2002a)
MicroaneurysmsCarotid artery diseaseStrongARIC (Klein et al., 2000), CHS (Wong et al., 2003c)
Cotton wool spotsIncident clinical strokeStrongARIC (Wong et al., 2001a, 2002b), BMES (Mitchell et al., 2002)
Subclinical cerebral diseaseStrongARIC (Wong et al., 2002b, 2003d)
Cognitive impairmentStrongARIC (Wong et al., 2002d)
Renal dysfunctionStrongARIC (Wong et al., 2004b)
Cardiovascular mortalityStrongBDES (Wong et al., 2003b), BMES (Mitchell et al., 2002)
Arterio-venous nickingCurrent blood pressureStrongARIC (Sharrett et al., 1999), BMES (Wang et al., 2003), BDES (Klein et al., 1994; Klein, at al, 1997) CHS (Wong et al., 2002a)
Past blood pressureStrongARIC (Sharrett et al., 1999), CHS (Wong et al., 2002a)
Inflammatory markersWeakARIC (Klein et al., 1999)
Endothelial dysfunctionWeakARIC (Klein et al., 1999)
Metabolic syndromeWeakARIC (Wong et al., 2004a), BMES (Leung et al., 2005)
Incident clinical strokeModerateARIC (Wong et al., 2001a, 2002b), BMES (Mitchell et al., 2002)
Subclinical cerebral diseaseModerateARIC (Wong et al., 2002b)
Renal dysfunctionWeakARIC (Wong et al., 2004b)
Focal arteriolar narrowingCurrent blood pressureStrongARIC (Sharrett et al., 1999), BMES (Wang et al., 2003), BDES (Klein et al., 1994, 1997), CHS (Wong et al., 2002a)
Incident hypertensionModerateARIC (Wong et al., 2004c)
Metabolic syndromeWeakARIC (Wong et al., 2004a)
Generalized arteriolar narrowingCurrent blood pressureStrongARIC (Sharrett et al., 1999), BMES (Wang et al., 2003), BDES (Klein et al., 1994), CHS (Wong et al., 2002a), Rotterdam (Ikram et al., 2004)
Past blood pressureStrongARIC (Sharrett et al., 1999), BMES (Leung et al., 2004), CHS (Wong et al., 2002a)
Incident hypertensionModerateARIC (Wong et al., 2004c), BDES (Wong et al., 2004e), BMES (Smith et al., 2004)
Inflammatory markersWeakARIC (Klein et al., 1999), Rotterdam (Ikram et al., 2004)
Carotid atherosclerosisModerateARIC (Klein et al., 1999; Liao et al., 2004), Rotterdam (Ikram et al., 2004)
Metabolic syndromeWeakARIC (Wong et al., 2004a), BMES (Leung et al., 2005)
Incident clinical strokeWeakARIC (Wong et al., 2001a, 2002b), BMES (Mitchell et al., 2002)
Incident heart diseaseModerateARIC (Wong et al., 2002c)
Cardiovascular mortalityWeakBDES (Wong et al., 2003b)

Table 3 shows a simple classification system of retinal microvascular signs based on the strength of the reported associations of various retinopathy signs (Wong and Mitchell, 2004). This classification groups retinopathy signs into mild, moderate and malignant. There is no clear evidence that patients with mild retinopathy (or retinal arteriolar signs only) need either physician referral or follow-up at this time. Individuals with moderate retinopathy signs (retinal haemorrhage, microaneurysm, cotton wool spots) may benefit from a systemic assessment of vascular risk and, if supported by further research, appropriate risk reduction therapy. Patients with malignant retinopathy (bilateral disk swelling in the presence of moderate hypertensive retinopathy) will continue to need urgent anti-hypertensive treatment.

Table 3.  Classification of retinal microvascular signs
Grades*DescriptionSystemic associations†Suggested management
  1. *Note that these grades do not imply a sequential temporal relationship of mild to moderate, and moderate to malignant.

  2. †Strength of association: relative risk/odds ratio >2.0 (Strong), 1.5–2.0 (Moderate), <1.5 (Weak).

  3. ‡Anterior ischaemic optic neuropathy, characterized by unilateral optic disc swelling, visual loss and sectoral visual field loss, should be excluded.

MildGeneralized arteriolar narrowing, focal arteriolar narrowing, arterio-venous nicking, arteriolar wall opacity (silver-wiring)Moderate and weak associations with stroke, ischaemic heart disease, cardiovascular mortalityRoutine care
Closer monitoring of blood pressure
Better control of blood pressure in persons with known hypertension
ModerateRetinal haemorrhage (blot, dot or flame-shaped), microaneurysm, cotton wool spot, hard exudatesStrong association with stroke, cognitive decline, congestive heart failure, renal dysfunction, and cardiovascular mortalityMay require physician referral
Need to exclude diabetes
Possible indication for treatment of hypertension and other risk factors in persons without known hypertension
Better control of blood pressure and risk factors in persons with known hypertension
MalignantModerate retinopathy signs plus optic disc swelling‡Associated with mortalityUrgent treatment of hypertension

A number of issues relating to the proposed classification deserve comment. First, while based on new population-based data, this classification system has not been validated in prospective clinical studies. Secondly, the studies reviewed in this article used retinal photography to quantify retinal microvascular signs and computer-imaging techniques to define generalized retinal arteriolar narrowing in highly standardized research settings (Wong et al., 2004d). It remains unclear if a careful clinical ophthalmoscopic examination is similar to photographic techniques in detecting subtle retinopathy signs and if current research methods can be translated to clinical use. Thirdly, the natural history of these signs is not fully understood, and the proposed grades in Table 3 do not imply a sequential temporal relationship. For example, some patients may present with malignant retinopathy without first having mild or moderate retinopathy. Fourthly, some of the findings were not consistent across studies. Therefore, it is possible that some associations could have been due to either chance or residual confounding.

Finally, the recent studies support the concept of specifically targeting the microcirculation for risk reduction in hypertension management. There is an increasing array of anti-hypertensive agents, such as ACE inhibitors, that may have direct beneficial effects on micro-vessel structure and function beyond its primary effect in lowering BP (Levy et al., 2001). Such agents may have added therapeutic value in treating hypertensive complications. Further research evaluating whether these treatment approaches may either reverse or reduce retinopathy signs, and whether this will ultimately reflect in a reduced cardiovascular risk for the patient, may be valuable (Wong et al., 2005).