Blood Lipids and First-Ever Ischemic Stroke/Transient Ischemic Attack in the Bezafibrate Infarction Prevention (BIP) Registry
High Triglycerides Constitute an Independent Risk Factor
Background— Despite unclear associations between blood lipids, including fractionated cholesterol and triglycerides, and stroke, recent evidence demonstrates that lipid-modifying agents decrease the risk of stroke in patients with coronary heart disease (CHD).
Methods and Results— Patients with documented CHD who were screened for but not included in the Bezafibrate Infarction Prevention study and had no history of stroke or transient ischemic attack (TIA) (n=11 177) were followed up. At baseline, medical histories were obtained and blood lipids assessed at a central study laboratory. During a 6- to 8-year follow-up period, 941 patients were identified as having nonhemorrhagic cerebrovascular disease, of whom 487 had verified ischemic stroke (per clinical findings and brain CT) or TIA. Patients experiencing an ischemic stroke/TIA had higher mean levels of triglycerides, lower levels of HDL cholesterol, and lower percentages of cholesterol contained in the HDL cholesterol moiety (%HDL; P<0.01 for all). In a logistic regression model, the adjusted ORs for developing an ischemic stroke/TIA were 1.27 (95% CI 1.01 to 1.60) associated with triglycerides >200 mg/dL and 0.87 (95% CI 0.78 to 0.97) associated with a 5% decrease in %HDL. The increased risk associated with high triglycerides was found across subgroups of age, sex, patient characteristics, and cholesterol fractions.
Conclusions— High triglycerides constitute an independent risk factor for ischemic stroke/TIA across subgroups of age, sex, patient characteristics, and cholesterol fractions, whereas high %HDL was an independent protective factor among patients with CHD. These findings support the role of blood lipids, including triglycerides, as important modifiable stroke risk factors.
Received October 3, 2001; accepted October 8, 2001.
Stroke is a leading cause of death and disability worldwide.1,2 Several conditions and lifestyle factors have been well established as risk factors for stroke.2,3 Despite unclear associations between blood lipids and stroke, studies in patients with coronary heart disease (CHD) demonstrate that lipid-modifying agents decrease the risk of stroke.4–7 Subsequent to these findings, there is a great deal of interest in the role of blood lipids in the pathogenesis of stroke.
Few studies have examined the relationship between fractionated cholesterol and stroke. We recently assessed the role of cholesterol and its fractions in prediction of ischemic cerebrovascular disease (CVD).8 Low levels of HDL cholesterol (HDL-C) or low percentage of cholesterol contained in the HDL-C moiety (%HDL) and high levels of total or LDL cholesterol (LDL-C) were each found to predict ischemic CVD.
A meta-analysis of prospective studies indicates that elevated triglycerides are an independent risk factor for developing a first manifestation of CHD.9 Prospective data on the association between serum triglycerides and stroke are scarce, with conflicting results. The third report of the National Cholesterol Education Program expert panel categorized triglyceride levels ≥200 mg/dL as high.10 We undertook the present study within a large cohort of male and female patients (n=11 177) with established CHD to assess (1) the specific role of high serum triglyceride in the prediction of ischemic stroke or transient ischemic attacks (TIA) and (2) the significance of triglycerides in the context of cholesterol and its fractions and within subgroups of lipid levels.
Patients with documented CHD (n=15 524) were screened to examine their eligibility for inclusion in the Bezafibrate Infarction Prevention (BIP) study, a placebo-controlled secondary prevention randomized clinical trial assessing the effect of bezafibrate retard.11 The design and rationale of the BIP study have been published.12 Patients were 40 to 74 years old, with a diagnosis of CHD. Patients were eligible for inclusion in the BIP randomized clinical trial if they had evidence of myocardial infarction occurring ≥6 months but ≤5 years before enrollment, or coronary insufficiency observed either at rest or during effort as manifested by typical pain and dynamic ECG changes, or both. Episodes of coronary insufficiency had to have occurred between 6 months and 2 years before enrollment. In addition, specific serum lipid ranges were required, because the study examined the efficacy of intervening to reduce serum triglycerides and increase HDL-C.
Patients with total cholesterol ≤7.0 mmol/L (270 mg/dL), HDL-C ≤1.17 mmol/L (45 mg/dL), and serum triglycerides ≤3.39 mmol/L (300 mg/dL) at the time of the first visit were invited for a second visit after a 2-month diet. We excluded patients with a history of prior stroke/TIA so as to assess the risk of first-ever stroke associated with blood lipids. We also excluded patients who enrolled to participate in the BIP study to avoid the possible modification of bezafibrate treatment on the association between blood lipids and stroke. The participants in the BIP study (3122 patients) fulfilled strict inclusion and exclusion criteria for entry into this randomized clinical trial.12 Their ages were similar to those of the other BIP screenees, but they included a higher proportion of men (91.5%) and had a lower proportion of diabetes mellitus (9.7%) and a higher proportion of prior myocardial infarction (78%). Because of the lipid selection criteria (serum total cholesterol between 180 and 250 mg/dL, LDL-C ≤180 mg/dL, HDL-C ≤45 mg/dL, and triglycerides ≤300 mg/dL), these patients had lower total and HDL-C levels.
The total number of patients in the present analysis was 11 177. During the first physician visit, records were obtained on medical history, conventional risk factors, and medications used, and a complete physical examination was carried out. Mortality data were obtained through January 1999 from the Israel Population Registry, with cause of death coded according to International Classification of Diseases (ICD)-9 codes.
Laboratory measurements were performed in a central laboratory (Physiological and Hygiene Laboratory at the Wolfson Medical Center, Holon, Israel). All analyses were performed with a Boehringer-Hitachi 704 random-access analyzer with Boehringer diagnostic kits. Accuracy and precision were under periodic surveillance by the Centers for Disease Control and Prevention service in Atlanta, Ga. Blood samples were taken after 12 hours of fasting. Between February 1990 and January 1994, we determined triglyceride levels by subtracting the free glycerol level (determined with a separate enzymatic kit [Sigma Chemical Co]) from the total triglyceride value. Since January 1994, we calculated triglyceride levels by subtracting 4.5 mg/dL (mean value of free glycerol level). HDL-C was determined by precipitation of LDL-C and VLDL with phosphotungstate.
Adjudication of CVD
We obtained computerized data files of hospitalizations with a diagnosis of CVD (ICD-9 codes 430 to 438 or code 38.1). We matched the patients (based on national ID number and name) against a registry of the Clalit Health Services, which insures >60% of Israelis (Israel has an obligatory 100% National Health Insurance), as well as all hospitals except one participating in the BIP screening process. A total of 1100 patients were identified, and attainable medical records and hospital discharge summaries were systematically reviewed. Data were collected on history, findings on neurological examination, brain CT, and ancillary examinations as available to verify the diagnosis and to determine stroke type and ischemic stroke subtypes. Two investigators, including a stroke neurologist (D.T.), reviewed all classifications.
Stroke was defined according to World Health Organization criteria. Events that resolved completely within <24 hours were diagnosed as a TIA. Patients with subarachnoid hemorrhage or subdural hemorrhage and those who did not fulfil the criteria for stroke or TIA after review were excluded. Ischemic stroke and intracerebral hemorrhage were differentiated by the results of brain CT performed at the acute stage. Ischemic stroke was diagnosed if the patient had an appropriate clinical event and had a brain CT that showed a compatible low-density lesion, or was normal, or had findings compatible with hemorrhagic conversion of a cerebral infarction. Only 24 cases were identified with verified intracerebral hemorrhage by clinical findings and brain CT, and thus, associations with intracerebral hemorrhage were not sought.
For the purpose of this study, we assessed 941 cases with nonhemorrhagic CVD according to ICD-9 codes, excluding cases considered after chart review to have had a nonvascular pathogenesis. Of those, review of medical records showed that 487 patients had a verified ischemic stroke (350 cases) or TIA (137 cases). For the remaining patients, brain CT was not performed or medical records were not available for review, so that stroke type could not be determined.
Data were analyzed with SPSS software. Multivariable analyses for prediction of ischemic stroke/TIA were performed with logistic regression models. Because the distribution of triglycerides was skewed, a natural logarithm of triglyceride was introduced into the models. The significance levels for entering and removing an explanatory variable were set at 0.05 and 0.10, respectively. The goodness of fit of the model to the observed event rates was evaluated by the Hosmer-Lemeshow statistic.13 The ability of the model to classify patients correctly (ie, its predictive performance) was evaluated by use of the C statistic, a term equivalent to the area under a receiver operating characteristic curve for dichotomous outcomes. Interactions were sought between selected variables and triglycerides for prediction of ischemic CVD.
A single measurement of triglycerides is subject to random fluctuation because of laboratory measurement and biological fluctuations. Because 6739 patients attended the second screening visit, at which another triglyceride measurement was performed, we estimated the reliability of serum triglyceride measurements using the values of the first and second visits.14 The effect of incorporating the regression dilution bias is to provide an estimate of ischemic stroke/TIA risk associated with triglyceride increment, correcting for regression to the mean.
Of the 11 177 patients with CHD who were included in the present study (78% male, 12% female), 941 developed a nonhemorrhagic CVD during a 6- to 8-year follow-up. Of those, 487 had a verified ischemic stroke or TIA. Patients experiencing an ischemic stroke/TIA had higher mean levels of triglycerides, lower levels of HDL-C, and lower %HDL (P<0.01 for all; Table 1).
Elevated Triglyceride Levels and Other Comorbid Conditions
Patients with high triglycerides (>200 mg/dL; 2.3 mmol/L) tended to have a higher proportion of diabetes mellitus (29% versus 19%), hypertension (35% versus 32%), or a New York Heart Association classification ≥2 (32% versus 27%). They were somewhat younger (mean±SD, 58±7 versus 60±7 years), smoked cigarettes more frequently (14% versus 9%), and had a higher mean body mass index (27.6±3.7 versus 26.4±3.5 kg/m2). Mean levels of total cholesterol (245±46 versus 219±46 mg/dL) and LDL-C (158±42 versus 154±36 mg/dL) were higher and those of HDL-C (32.0±7.8 versus 40.9±10.8 mg/dL) considerably lower among patients with triglycerides >200 mg/dL.
Ischemic Stroke/TIA in Relation to Lipids
Age-adjusted incidence rates of ischemic stroke/TIA by quintiles of serum triglyceride concentrations are depicted in Figure 1. Age-adjusted rates of ischemic stroke decreased with increasing tertile of %HDL, from 5.1% to 3.7% in patients with serum triglyceride ≤200 mg/dL and from 6.4% to 3.8% in patients with triglycerides >200 mg/dL (Figure 2a). Rates rose with increasing tertiles of LDL-C among patients with serum triglyceride below and above 200 mg/dL (Figure 2b).
Adjusted for traditional risk factors, triglycerides >200 mg/dL were associated with an OR for ischemic stroke/TIA of 1.47 (95% CI 1.19 to 1.80) compared with lower triglyceride levels (Table 2). An increase of 1 natural log unit in triglyceride levels (the magnitude of change depends on the actual concentration of triglycerides, eg, an increase from 75 to 204 mg/dL or from 100 to 272 mg/dL) was associated with an adjusted OR of 1.39 (95% CI 1.16 to 1.68). The computed regression dilution factor for triglycerides was 1.31. Correction for this factor (see Methods) would modify the OR for ischemic stroke/TIA, associated with an increase of 1 natural log unit in triglyceride levels, from 1.39 to 1.54 (95% CI 1.22 to 1.98).
At baseline, 518 patients (4.6%) were treated with lipid-modifying drugs. Of these, 203 were treated with lovastatin, 294 were treated with bezafibrate, and 37 were treated with colestipol. Triglyceride levels at baseline differed slightly between the patients treated with lipid-modifying drugs at baseline (172±102 mg/dL) and counterparts not treated with any lipid-modifying drugs (165±102 mg/dL). A logistic regression model excluding the patients treated with lipid-modifying drugs at baseline did not have any meaningful effect on the results of the model (the adjusted OR for ischemic stroke/TIA associated with triglycerides >200 mg/dL was 1.42; 95% CI 1.15 to 1.78).
In separate models adjusting for total cholesterol or LDL-C or HDL-C in addition to conventional cerebrovascular risk factors, the adjusted ORs associated with triglycerides >200 mg/dL were 1.40, 1.43, and 1.39, respectively. In a further model adjusting for total and HDL-C in addition to conventional risk factors, the association was attenuated, but high triglycerides remained an independent predictor of ischemic stroke/TIA (OR 1.28; 95% CI 1.01 to 1.62).
In a final logistic regression model for the prediction of ischemic stroke/TIA, traditional risk factors, namely increasing age, diabetes mellitus, current smoking, hypertension, and prior myocardial infarction, as well as male sex, all predicted future ischemic stroke/TIA (Table 3). Triglycerides >200 mg/dL emerged as an independent risk factor with an OR of 1.27, whereas high %HDL was a protective factor (OR 0.87 per 5% increase). The Hosmer and Lemeshow goodness-of-fit test (P=0.15) did not exhibit a meaningful departure of the observed from the expected rates according to the model. The C statistic for the model was 0.66.
Adjusted OR associated with serum triglycerides >200 mg/dL were further estimated within subgroups of patients (Table 4). ORs were nominally higher for women, for patients ≥60 years old, for patients with body mass index <25 kg/m2, and for those with concomitant high levels of LDL-C (>130 mg/dL) or high levels of HDL-C (>35 mg/dL). Tests for interaction did not substantiate significant synergism between triglycerides and any of these variables in predicting CVD.
The main findings in this prospective study were that high triglycerides constitute an independent risk factor for ischemic stroke/TIA across subgroups of age, sex, patient characteristics, and cholesterol fractions, whereas high %HDL was an independent protective factor among patients with CHD.
Multiple studies and a meta-analysis showed that elevated triglyceride levels were associated with increased risk for CHD.9,15,16 In some of these studies, the positive relation between triglyceride levels and CHD risk persisted after adjustment for possible confounders as well as for HDL-C. Triglycerides are carried in virtually all plasma lipoproteins and present a different risk profile in the fasting and postprandial states, making triglyceride-rich lipoproteins highly heterogeneous. This heterogeneity is a major contributing factor to the complexity of the relationship between triglycerides and cardiovascular disease. Concurrent hypertriglyceridemia, hypertension, diabetes mellitus, obesity, and other dyslipidemias are well known. This aggregate of metabolic and clinical abnormalities was called the “metabolic syndrome,” and it was suggested that a single abnormality (insulin resistance and hyperinsulinemia) is the underlying cause of this cluster.17 These associations, in particular the inverse association with HDL-C, complicate analyses designed to evaluate the independent contribution of elevated triglyceride levels to subsequent morbidity and mortality in these subjects. Triglyceride and HDL-C levels are usually (including in the present study) inversely related, implying multicolinearity. Therefore, adjusting for HDL-C may result in underestimation of the true risk inherent in elevated levels of triglycerides. Nevertheless, high triglycerides remained an independent predictor of ischemic stroke/TIA even after adjustment for HDL-C. The increased risk was evident primarily among patients with serum triglycerides >200 mg/dL, corresponding approximately to the upper quintile of triglyceride levels among our study population and to levels above normal by authoritative recommendations.10
Several case-control studies have found an association between high triglycerides and ischemic stroke.18 Few prospective studies assessed the association between triglycerides and stroke, with inconsistent results. In the Copenhagen City Heart Study, triglyceride levels were positively and significantly associated with risk of nonhemorrhagic CVD, with a relative risk per 1 mmol/L increase of 1.12.19 In the Framingham and British Regional Heart studies, no such associations were identified,20,21 and in the Finnmark Study, a weak association was identified in women only.21 None of these studies assessed a cohort of patients with CHD. In the Atherosclerosis Risk in Communities study, triglycerides were associated only weakly with carotid atherosclerosis but were associated strongly with CHD incidence.22 To the best of our knowledge, this prospective study is the first to indicate that high triglycerides expose CHD patients to increased risk of ischemic stroke beyond cholesterol and its fractions.
In addition to a direct atherogenic effect of triglyceride-rich lipoproteins, high triglycerides appear to be a marker of a series of other potentially atherogenic and prothrombotic changes. Elevated triglycerides have been associated with several abnormalities of the clotting-fibrinolytic systems,23 which may contribute further to their association with ischemic CVD.
High total cholesterol and LDL-C and low HDL-C are well-established risk factors for a first or recurrent event of CHD. We have recently found, in our cohort of patients with chronic CHD, similar trends for prediction of ischemic CVD,8 and in the present analysis, we demonstrate the additive risk conferred by cholesterol fractions and high triglycerides. CHD and ischemic stroke share multiple risk factors and a common pathophysiological antecedent, namely atherosclerosis.24 Incorporating triglycerides and fractionated cholesterol, the best predictive model for future ischemic stroke/TIA beyond traditional risk factors included high triglycerides (>200 mg/dL) and %HDL. Increasing %HDL was an independent protective factor for ischemic stroke. These findings add to evidence relating low HDL-C to risk of ischemic stroke.19,25,26 They are also in agreement with the recently reported reduction in ischemic stroke among men with CHD and low HDL-C by pharmacologically modifying lipid levels with gemfibrozil in the Veterans Affairs HDL Intervention Trial (VA-HIT).27
The stronger associations we have found for incident ischemic stroke/TIA than for all ischemic CVD demonstrate the importance of case-by-case ascertainment, rather than relying on less accurate ICD codes.28 Because of repeat lipid measurements in a large proportion of patients, we were able to estimate and correct for the regression dilution bias, thus providing more accurate estimates of risks.14
Our study is prospective but limited by its post hoc observational design. Because ≈F3100 patients included in our registry were included in a randomized clinical trial of a lipid-modifying drug, we excluded them from all analyses. Second, a common limitation to many observational studies is the absence of information concerning potential spontaneous or therapy-induced changes in triglyceride level during the follow-up period. Third, CHD tends to occur at a substantially higher rate and earlier age than ischemic stroke. Because of shared risks between these conditions, blood lipids may be underestimated as predictors of ischemic stroke. These shortcomings may have, if anything, diluted the reported associations between blood lipids, including triglycerides and ischemic CVD. Finally, the predictive role of blood lipids in this study was found in a group of patients 40 to 74 years old at baseline with established CHD based on evidence of myocardial infarction occurring ≥6 months but ≤5 years before enrollment, or coronary insufficiency, or both. Generalization of these results to broader populations, such as those without clinically manifest CHD, awaits further studies.
In conclusion, blood lipids improve the prediction of ischemic stroke/TIA beyond traditional risk factors. Specifically, this study shows for the first time that high triglycerides, in addition to fractionated cholesterol, are associated with increased risk of ischemic stroke/TIA in a large cohort of men and women with CHD. In view of the positive results of lipid-modifying drugs in stroke prevention among patients with CHD and the renewed interest in the role of lipids for stroke, further research should focus on blood lipids, including serum triglycerides, as part of the global risk assessment and potentially modifiable risk factors for ischemic stroke.
Wolf PA, D’Agostino RB. Epidemiology of stroke.In: Barnett HJM, Mohr JP, Stein BM, et al, eds. Stroke: Pathophysiology, Diagnosis and Management. New York, NY: Churchill Livingstone; 1998: 3–28.
Goldstein LB, Adams R, Becker K, et al. Primary prevention of ischemic stroke: a statement for healthcare professionals from the Stroke Council of the American Heart Association. Stroke. 2001; 32: 280–299.
Plehn JF, Davis BR, Sacks FM, et al. Reduction of stroke incidence after myocardial infarction with pravastatin: the Cholesterol and Recurrent Events (CARE) study. The Care Investigators. Circulation. 1999; 99: 216–223.
Rubins HB, Robins SJ, Collins D, et al. Gemfibrozil for the secondary prevention of coronary heart disease in men with low levels of high-density lipoprotein cholesterol. Veterans Affairs High-Density Lipoprotein Cholesterol Intervention Trial Study Group. N Engl J Med. 1999; 341: 410–418.
Koren-Morag N, Tanne D, Graff E. et al. Cholesterol in low and high density lipoproteins and ischemic cerebrovascular disease: the Benzafibrate Infarction Prevention (BIP) Registry. Arch Intern Med. In press.
Secondary prevention by raising HDL cholesterol and reducing triglycerides in patients with coronary artery disease: the Bezafibrate Infarction Prevention (BIP) study. Circulation. 2000; 102: 21–27
Goldbourt U, Behar S, Reicher-Reiss H, et al. Rationale and design of a secondary prevention trial of increasing serum high-density lipoprotein cholesterol and reducing triglycerides in patients with clinically manifest atherosclerotic heart disease (the Bezafibrate Infarction Prevention Trial). Am J Cardiol. 1993; 71: 909–915.
Lemeshow S, Hosmer DW Jr. A review of goodness of fit statistics for use in the development of logistic regression models. Am J Epidemiol. 1982; 115: 92–106.
Jeppesen J, Hein HO, Suadicani P, et al. Triglyceride concentration and ischemic heart disease: an eight-year follow-up in the Copenhagen Male Study. Circulation. 1998; 97: 1029–1036.
Reaven GM. Banting lecture 1988. Role of insulin resistance in human disease. Diabetes. 1988; 37: 1595–1607.
Lindenstrom E, Boysen G, Nyboe J. Influence of total cholesterol, high density lipoprotein cholesterol, and triglycerides on risk of cerebrovascular disease: the Copenhagen City Heart Study. BMJ. 1994; 309: 11–15.
Njolstad I, Arnesen E, Lund-Larsen PG. Body height, cardiovascular risk factors, and risk of stroke in middle-aged men and women: a 14-year follow-up of the Finnmark Study. Circulation. 1996; 94: 2877–2882.
Sharrett AR, Sorlie PD, Chambless LE, et al. Relative importance of various risk factors for asymptomatic carotid atherosclerosis versus coronary heart disease incidence: the Atherosclerosis Risk in Communities Study. Am J Epidemiol. 1999; 149: 843–852.
Byberg L, Siegbahn A, Berglund L, et al. Plasminogen activator inhibitor-1 activity is independently related to both insulin sensitivity and serum triglycerides in 70-year-old men. Arterioscler Thromb Vasc Biol. 1998; 18: 258–264.
Tanne D, Goldbourt U, Zion M, et al. Frequency and prognosis of stroke/TIA among 4808 survivors of acute myocardial infarction. The SPRINT Study Group. Stroke. 1993; 24: 1490–1495.
Tanne D, Yaari S, Goldbourt U. High-density lipoprotein cholesterol and risk of ischemic stroke mortality: a 21-year follow-up of 8586 men from the Israeli Ischemic Heart Disease Study. Stroke. 1997; 28: 83–87.
Bloomfield Rubins H, Davenport J, Babikian V, et al. Reduction in stroke with gemfibrozil in men with coronary heart disease, and low HDL cholesterol. The Veterans Affairs HDL Intervention Trial (VA-HIT). Circulation. 2001; 103: 2828–2833.
Benesch C, Witter DM Jr, Wilder AL, et al. Inaccuracy of the International Classification of Diseases (ICD-9-CM) in identifying the diagnosis of ischemic cerebrovascular disease. Neurology. 1997; 49: 660–664.