Inclusion of Ischemic Stroke in Cardiovascular Risk Prediction Instruments - Why We Should Care

The AHA/ASA Scientific Statement, “Inclusion of Stroke in Cardiovascular Risk Prediction Instruments,¹” addresses several situations in which stroke has not been historically considered along with coronary heart disease, either as a cardiovascular disease outcome or predictor. Perhaps one of the motivating factors for this Statement is how the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) guidelines^2,3 identify high risk populations for the most aggressive LDL target. These guidelines recommend a LDL target, <100 mg/dL or an optional goal of <70 mg/dL, for persons with coronary heart disease risk (CHD) or a CHD risk equivalent defined as a predicted 10 year CHD risk >20%. Conditions named as CHD risk equivalents include peripheral vascular disease, abdominal aortic aneurysm, and diabetes. Symptomatic carotid artery disease is also considered a CHD risk equivalent but not the other stroke subtypes that comprise more than 75% of all ischemic strokes.

Two questions can be raised about the NCEP ATP III guidelines.⁴ First, why is CHD the appropriate outcome for identifying a high-risk population for intensive treatment and not any ischemic cardiovascular event, including stroke? NCEP ATP III stated that their primary aim was to match the intensity of LDL-lowering therapy to absolute risk within the constraints of currently accepted standards for cost-effectiveness.² In the Stenting and Aggressive Medical Management for Preventing Recurrent Stroke in Intracranial Stenosis (SAMMPRIS) trial, the medical management group had a 14.9% probability of stroke at one year,⁵ a rate likely to exceed 20% at 10 years. Kaplan-Meier estimates of the probabilities of recurrent stroke five years after first ischemic stroke have been estimated at 40.2% for large artery cervical or intracranial atherosclerotic stroke, 31.7% for cardioembolic stroke, 24.8 for lacunar stroke, and 33.2% for ischemic stroke of uncertain cause.⁶ Thus, it is likely that 10-year recurrent stroke rates exceed the absolute risk magnitude of 20%. If both the CHD and stroke risk were considered, the 10-year absolute event rate would be even higher.

The argument could be advanced that the reason that NCEP ATP III guidelines focused on CHD or CHD risk equivalent was that available data suggested only atherosclerotic disease outcomes would be prevented by statins. If so, subsequent data have called this assumption into question. The Stroke Prevention by Aggressive Reduction in Cholesterol Levels (SPARCL) trial⁷ tested statin therapy in a group composed primarily of ischemic stroke (67%) or transient ischemic attacks (TIAs; 30%) without CHD with LDL levels100-190 mg/dL at baseline. In the SPARCL trial, with a median follow-up period of 4.9 years, 13.1% of the placebo group had a fatal or nonfatal stroke, with Kaplan Meier curves suggesting that the 10-year stroke rate approximated 20%. Among the almost 2,400 participants in the group randomized to atorvastatin 80 mg daily, 46 strokes and 39 major coronary events were prevented. While not powered to look at differences in benefit by stroke subtype, the point estimate for treatment effect supported a benefit not only for large artery stroke but also for TIAs, small vessel stroke and stroke of unknown cause.⁸

Second, even if a CHD equivalent defined as a predicted 10-year CHD risk >20% were acknowledged to be the appropriate definition of the high risk category for intensive treatment, what about CVD risk of other ischemic stroke subtypes such as intracranial atherosclerosis or even ischemic stroke in general? In the SAMMPRIS trial of secondary prevention in symptomatic intracranial atherosclerosis, the probability of myocardial infarction at 1 year exceeded 2%.5 A meta-analysis of studies with 65,996 TIAs or ischemic strokes found predicted annual risks of 2.1% for nonstroke vascular death and 2.2% for nonfatal or fatal myocardial infarction.⁹ While it is possible that these high rates of future CHD were due to a high prevalence of CHD among the studies in this meta-analysis, the median proportion of patients with previous symptomatic CAD at baseline was 11% and there was no correlation between that proportion and the subsequent risk of MI or nonstroke vascular death.⁹ In the Northern Manhattan Study (NOMAS), among stroke cases ≤ 70 years without CHD, the 5-year risk of MI or vascular death was 9.7%.¹⁰

Beyond the specific example of cholesterol treatment targets, the issue of the inclusion of stroke in cardiovascular risk prediction instruments and global cardiovascular outcomes has broad implications for assessing the global burden of vascular disease.¹¹ Stroke makes a disproportionate contribution to the burden of vascular disease both among African-Americans in the U.S.¹² and internationally within China, Africa, and South America.¹³ A heightened awareness that even non-atherosclerotic ischemic stroke is a strong predictor of future cardiovascular events could also have implications for the design of other prevention programs. For example, a 2006 Science Advisory from the American Heart Association recommends influenza vaccine for patients with coronary heart disease and other forms of atherosclerotic disease.¹⁴ It is ambiguous whether all forms of stroke or only atherosclerotic stroke are included in this recommendation. In short, vascular disease prevention would be strengthened by including ischemic stroke as a CHD risk equivalent and in other aggregate measures of the burden of vascular disease.