Retinal cholesterol emboli have been linked to ulcerated atherosclerotic plaques at the carotid artery bulb [1-3], but these emboli can originate from ulcerated atherosclerotic plaques in any artery leading to the retina, including the common carotid artery, the brachiocephalic artery, or the ascending aorta. Ulcerated atherosclerotic plaques do not necessarily produce hemodynamically significant stenosis. The prevalence of extracranial carotid artery stenosis greater than 50% ipsilateral to asymptomatic retinal cholesterol emboli ranges from 0% to 17% [9-11].

Symptomatic ocular ischemia is usually evaluated and treated regardless of the presence of retinal emboli, but the clinical significance of asymptomatic retinal cholesterol emboli has not been the subject of controlled studies. Retinal cholesterol emboli have been associated with increased mortality [12, 13], but previous studies used a cohort of both asymptomatic and symptomatic patients who were grouped together and compared with general population controls, and patients and controls were not matched for some important vascular risk factors. We compare the vascular outcome of a group of 70 patients with asymptomatic retinal cholesterol emboli with that of a matched control group.

Methods

Top Methods Results Discussion Author & Article Info References

All participants had a dilated ocular examination in the Albuquerque Veterans Affairs Medical Center Eye Clinic between January 1989 and February 1991. Approximately 4500 patients annually had a dilated ocular examination in that clinic during this study.

Patients

Persons who were examined in the eye clinic for any reason and in whom an asymptomatic retinal cholesterol embolus was found during examination were considered for this study. A retinal cholesterol embolus was diagnosed if a bright orange, yellow, or copper-colored fragment was seen in a retinal arteriole. The study optometrists drew a diagram of the embolus location and took photographs of the fundus. An embolus was considered asymptomatic if the patient had no history of sudden monocular visual loss and no corresponding scotoma or retinal infarction on examination. Seventy consecutive men with asymptomatic retinal cholesterol emboli were entered into the study; none had had an arteriogram, cardiac surgery, or other vascular surgery within 12 months before entry. The study neurologist interviewed and examined all patients and referred each of them for a total serum cholesterol level test, a carotid Duplex ultrasound test, and an electrocardiographic study.

Controls

We selected the controls retrospectively and randomly in 1993. Beginning with all patients who visited the eye clinic and who had a dilated ocular examination during the study period, a computer generated at random a list of potential participants [11]. Controls were selected from this list if they had no retinal emboli and no evidence of ocular ischemia on examination. The first 40 controls were selected without matching for vascular risk factors. This group was then compared with the 70 patients for age, hypertension, diabetes mellitus, ischemic heart disease, and smoking. The risk factors that differed substantially between the two groups were either sought or avoided in selecting the next group of 10 controls, so that the patient and control groups would be as closely matched as possible. This procedure was repeated twice more to obtain 70 controls that were well matched to the patients. Because all of the patients were men, we selected only men to be controls. When a possible control lacked the qualifying vascular risk factors, we considered the next person who had had a dilated ocular examination in the eye clinic. Approximately 15% of those screened were selected. We used no other selection criteria and the investigators selecting the controls were unaware of any outcome events.

Vascular Risk Factors

The study neurologist interviewed patients and reviewed medical records to identify history of hypertension, diabetes mellitus, and ischemic heart disease. We defined a patient as having hypertension if he was being treated for it or if his blood pressure was greater than 160 mm Hg systolic or 90 mm Hg diastolic on two consecutive clinic visits. We defined a patient as having diabetes mellitus if he was being treated for it or if his fasting blood glucose level was greater than 6.66 mmol/L (120 mg/dL). We accepted a serum cholesterol level if it was measured during the study period or within 1 year before study entry.

We diagnosed ischemic heart disease if it was shown by an objective cardiac test (electrocardiography, echocardiography, stress test, or arteriography) or if the patient had had coronary bypass surgery. We defined current smoking as tobacco smoking within 2 years of study entry.

Follow-up

The study neurologist informed the patients about transient ischemic attack and stroke symptoms and asked to be telephoned if they suspected that a cerebrovascular event had occurred. He also explained the importance of limiting vascular risk factors, recommended 325 mg of aspirin daily to reduce the likelihood of ischemic events, and examined the patients at 6- to 9-month intervals. The patients' primary physicians provided treatment for hypertension, diabetes mellitus, hypercholesterolemia, and ischemic heart disease, and assistance with cessation of smoking.

We defined stroke as a sudden focal brain dysfunction lasting longer than 24 hours. Computed tomographic scans distinguished between cerebral infarction and hemorrhage. We identified the cause of cerebral infarction as carotid artery disease if there was greater than 50% carotid stenosis ipsilateral to carotid territory ischemia; as cardioembolic if there was a high risk for cardiac emboli; as small-vessel disease if the symptoms and neuroimaging test results were consistent with a small (<1.5 cm diameter), deep infarction; as "other" if another cause was found; or as undetermined. High-risk factors for cardiac emboli were atrial fibrillation, valvular heart disease, intracardiac thrombus, myocardial infarction in the preceding 3 months, or severe myocardial hypokinesia.

When a patient developed a stroke, we did computed tomographic scans of the head, blood tests for hypercoagulability, a carotid Duplex test if the event was in a carotid artery territory, and echocardiography if heart disease was present. Additional tests were done on the basis of the results of these standard tests.

The follow-up period for the controls was the same as that for the patients, from selection (January 1989 to February 1991) either to a telephone interview by the study neurologist between October and December 1993 or to death. The study neurologist also reviewed the medical records of the controls. When interview or review of the medical record suggested that a stroke may have occurred, the study neurologist reviewed the pertinent tests. The controls' primary physicians provided treatment for hypertension, diabetes mellitus, hypercholesterolemia, and ischemic heart disease, and assistance with cessation of smoking.

Review of medical records was done to confirm myocardial infarction, death, and cause of death. We classified death as vascular if atherosclerosis was the main cause. When a participant had more than one cerebral or myocardial infarction, we tabulated the first of each type of vascular event for analysis.

Statistical Analysis

We used a two-sided Fisher exact test to compare the baseline characteristics of patients and controls for categorical data and a two-sided t-test to compare the two groups for continuous data. Cox proportional-hazards analysis [14] measured adjusted relative risk, 95% CIs, and P values for outcome events associated with asymptomatic retinal cholesterol embolism. We considered controlling the analysis for stroke history, smoking history, and current smoking because these variables were the least adequately matched between patients and controls. With these three variables included in Cox proportional-hazards analysis, the adjusted relative risk associated with current smoking was minimal for all outcome events (range, 0.7 to 1.3); therefore, we controlled the analysis only for stroke history and smoking history. Annual rates of outcome events were given by the ratio of the number of particular events to person-years of observation for that event.

Results

Top Methods Results Discussion Author & Article Info References

Vascular risk factors of patients and controls at the time of study entry are shown in Table 1. Median ages of patients and controls were 69 and 68 years, respectively. Vascular outcomes for patients and controls are shown in Table 2. Mean duration of follow-up for all participants was 3.4 years (those alive at the end of the study were followed from 2.6 to 4.9 years). We lost 1 patient to follow-up after 2.6 years of observation. Fifty patients (71%) and 37 controls (53%) reported taking between 80 and 650 mg aspirin daily during follow-up. Five patients (7%) and 5 controls (7%) received warfarin during follow-up.

Stroke occurred at an annual rate of 8.5% among patients and 0.8% among controls [adjusted relative risk, 9.9; 95% CI, 2.3 to 43.1; P = 0.002] Figure 1, top). All 19 strokes, 17 of which were in patients and 2 of which were in controls, were nonfatal and ischemic by computed tomographic scans of the head or by magnetic resonance imaging done within 7 days. Ocular infarction or hemorrhagic stroke did not occur in either group. Fifteen of the cerebral infarctions in patients were in the carotid artery territory and two were in the vertebrobasilar territory. Twelve cerebral infarctions occurred in the carotid artery territory ipsilateral to the qualifying retinal cholesterol embolus and five were in another vascular territory. The probable cause of cerebral infarction was small-vessel disease in eight patients, extracranial carotid artery disease in four patients (occluded in two and 50% to 79% stenosis in two), cardioembolic in one patient (valvular heart disease), and undetermined in four patients. Among the controls, one cerebral infarction was in the carotid territory of undetermined cause and the other was a vertebrobasilar lacunar stroke.

View larger version (14K): [in this window] [in a new window]   Figure 1. Event-free survival for cerebral infarction and nonfatal myocardial infarction or vascular death in 70 men with asymptomatic retinal cholesterol emboli and 70 matched controls. Calculated using Cox proportional-hazards analysis controlling for stroke history and smoking history.

Nonfatal myocardial infarction or vascular death occurred at an annual rate of 7.7% among patients and 4.9% among controls [adjusted relative risk, 1.4; 95% CI, 0.7 to 2.9; P = 0.39] Figure 1, bottom). None of the patients had an autopsy. Vascular deaths among the patients were due to cardiac arrest in seven patients, congestive heart failure in three patients, complications of vascular surgery in two patients (coronary artery bypass grafting in one and abdominal aortic aneurysm repair in the other), and acute myocardial infarction in one patient. Vascular deaths among the controls were due to acute myocardial infarction in four, cardiac arrest in three, and congestive heart failure in three.

Discussion

Top Methods Results Discussion Author & Article Info References

In this controlled study, asymptomatic retinal cholesterol embolism was associated with an increased rate of cerebral infarction independent of the commonly recognized vascular risk factors. The average annual rate of cerebral infarction in patients with asymptomatic retinal cholesterol emboli was 8.5%, which is approximately 10 times higher than that in our controls Table 2 and approximately 13 times higher than that expected among men of similar age [15].

Because the controls were recruited retrospectively, two problems need to be considered. First, the controls were not followed periodically by the study neurologist and they may not have been informed about stroke symptoms. Consequently, stroke symptoms may have been unreported or undocumented in the medical record. However, the study neurologist interviewed all living controls about possible undiagnosed stroke during the follow-up period.

Second, carotid artery studies were not done in the controls. Less severe carotid atherosclerosis among the controls may have contributed to the lower rate of stroke in this group. However, the two groups were well matched for the important predictors of carotid atherosclerosis [16-18] Table 1, and only 4 of the 17 (24%) strokes among our patients were attributed to carotid atherosclerosis (>50% stenosis).

Pfaffenbach and Hollenhorst [12] reported the outcome of 205 patients with retinal cholesterol emboli, 174 (85%) of whom were asymptomatic and 175 (85%) of whom were men. They followed 97% of the patients for at least 6 years and reported the combined outcome for patients with symptomatic and asymptomatic retinal emboli. Stroke occurred during follow-up in 19% of patients, but the stroke rate was not reported. The apparently higher rate of stroke in our study could be partly explained by the higher median patient age in our study (69 years compared with 64 years). Prevalences of vascular risk factors were similar in these two studies.

Savino and colleagues [13] reported the outcome of 86 patients with retinal thromboembolic events. Among these patients were 49 who had retinal emboli; 16 of these were asymptomatic and 5 (31%) were men. Savino and colleagues did not describe or classify the retinal emboli, but one of these investigators in a later publication referred to these emboli as bright plaques [19]. During up to 9 years of observation, stroke occurred in 38% of the patients with asymptomatic retinal emboli, but the stroke rate was not reported. These results appear to be similar to ours.

In another study of 18 patients with asymptomatic retinal cholesterol emboli [10], no ocular or cerebral ischemic events occurred during a mean 18-month follow-up. The ages and vascular risk factors of the patients in that study were not stated.

The total mortality rate was similar among our patients and our controls Table 2, but in the studies by Pfaffenbach and Hollenhorst [12] and by Savino and colleagues [13], patients with retinal cholesterol emboli had a higher mortality rate than did age- and sex-matched general population controls. The reason for this difference may be that controls in the other studies had less hypertension, diabetes, and ischemic heart disease than did the patients.

The difference in the rate of nonfatal myocardial infarction or vascular death between our patients and our controls was not statistically significant (Table 2). Although the patients had evidence of ulcerated atherosclerotic plaques in the arteries leading to the eyes, coronary atherosclerosis may have been similar in these two groups during this study. This possible similarity and the nearly equivalent prevalences of vascular risk factors, including ischemic heart disease, in our patients and controls could account for the similar cardiovascular outcome in these two groups.

Recurrent atherosclerotic plaque ulceration and overlying thrombosis are important steps in the pathogenesis of the acute coronary syndromes and progression of atherosclerosis [20]. Because atherosclerosis is a systemic disorder, this may also be true for cerebrovascular disease. We postulate that recurrent atherosclerotic plaque ulceration and thrombosis in the arteries leading to the eyes and brain predispose patients to both asymptomatic cholesterol embolization and symptomatic thromboembolism. Pure cholesterol crystal embolization is asymptomatic and can be seen on ophthalmoscopy if it involves the retina; thromboembolism results in symptomatic ocular or cerebral ischemia. This could explain the strong association between asymptomatic retinal cholesterol embolism and cerebral infarction in our study.

Optimal management of patients with asymptomatic retinal cholesterol emboli needs to be determined. A small percentage of patients with asymptomatic retinal cholesterol emboli have greater than 50% extracranial carotid artery stenosis ipsilateral to the embolus [9-11], but whether carotid endarterectomy benefits these patients remains to be established. The high rate of cerebral infarction in these patients may warrant an aggressive approach to vascular risk factor reduction.