Antibodies to anticardiolipin (acl) and 2 -glycoprotein 1

Transcription

1 Original Contributions 2 -Glycoprotein 1 Dependent Anticardiolipin Antibodies and Risk of Ischemic Stroke and Myocardial Infarction The Honolulu Heart Program Robin L. Brey, MD; Robert D. Abbott, PhD; J. David Curb, MD; Dan S. Sharp, MD, PhD; G. Webster Ross, MD; Christian L. Stallworth, BS; Steven J. Kittner, MD, MPH Background It has been hypothesized that immunoreactivity to 2 -glycoprotein 1 ( 2GP1)-dependent anticardiolipin antibody (acl), but not 2GP1-independent acl, is associated with increased risk of ischemic stroke and myocardial infarction (MI). Methods We performed a nested case-control study examining acl as a risk factor for ischemic stroke and MI by using stored frozen sera obtained from subjects enrolled in the Honolulu Heart Program and followed for up for 20 years. We measured 2GP1-dependent and 2GP1-independent acl and anti- 2GP1 immunoreactivity in 259 men who developed an ischemic stroke, in 374 men who developed an MI, and in a control group of 1360 men who remained free of both conditions. Results Only 2GP1-dependent acl of the IgG class was significantly associated with both incident ischemic stroke and MI. This association was attenuated in the last 5 years of the 20-year follow-up. For stroke, the risk factor adjusted relative odds for men with a positive versus a negative 2GP1-dependent acl of the IgG class were 2.2 (95% CI 1.5 to 3.4) at 15 years and 1.5 (95% CI 1.0 to 2.3) at 20 years. For MI, the adjusted relative odds were 1.8 (95% CI 1.2 to 2.6) at 15 years and 1.5 (95% CI 1.1 to 2.1) at 20 years. Conclusions These data suggest that acl IgG, particularly the 2GP1-dependent variety, is an important predictor of future stroke and MI in men. (Stroke. 2001;32: ) Key Words: antibodies, anticardiolipin case-control studies cerebrovascular disorders myocardial infarction prospective studies risk factors Antibodies to anticardiolipin (acl) and 2 -glycoprotein 1 ( 2GP1) are members of a larger group of autoantibodies directed against phospholipids or phospholipid-protein complexes, termed antiphospholipid antibodies. Many casecontrol studies have shown an association between different types of antiphospholipid antibodies and both stroke 1 8 and myocardial infarction (MI), 9 but some studies have not Few prospective studies examining the association between any type of antiphospholipid antibody and either stroke or MI have been performed, and these have been limited to acl. Two prospective studies 13,14 examining the association between MI and acl have demonstrated an increased risk for MI in patients with acl. In contrast, none of the prospective studies examining the association between acl and ischemic stroke showed significant effects See Editorial, page 1699 Recently, some differentiating features of these different types of antiphospholipid antibodies that may be pathogenic and lead to thrombosis have been described Previous studies have not systematically examined the relative importance of these antibody characteristics on thrombosis risk in the same population. There is also evidence from prior prospective studies 13,14 that the association between acl and MI may be stronger with a shorter duration of follow-up. Using a nested case-control design within a large long-term prospective study, we examined the association between several different types of antiphospholipid antibodies and the risk of both stroke and MI while we also examined the effects on events that occurred early and late in follow-up. Received December 11, 2000; final revision received May 2, 2001; accepted May 3, From the Division of Neurology (R.L.B., C.L.S.), University of Texas Health Science Center at San Antonio; the Division of Biostatistics (R.D.A.), University of Virginia School of Medicine, Charlottesville; the Department of Medicine (R.D.A., J.D.C., G.W.R.), John A. Burns School of Medicine, University of Hawaii, the Honolulu Heart Program (R.D.A., J.D.C., G.W.R.), Kuakini Medical Center, and the Honolulu Department of Veterans Affairs (G.W.R.), Honolulu, Hawaii; the National Heart, Lung, and Blood Institute (D.S.S.), Bethesda, Md; and the Department of Neurology (S.J.K.), University of Maryland School of Medicine, and the Department of Epidemiology and Preventative Medicine (S.J.K.), the Geriatrics Research, Education and Clinical Center, Baltimore Department of Veteran s Affairs Medical Center, Baltimore, Md. Correspondence to Steven J. Kittner, MD, MPH, Department of Neurology, Box 175, University of Maryland School of Medicine, 22 South Greene St, Baltimore, MD skittner@umaryland.edu 2001 American Heart Association, Inc. Stroke is available at

2 1702 Stroke August 2001 Methods Study Population The Honolulu Heart Program is a prospective study of coronary disease and stroke in 8006 American men of Japanese ancestry born from 1900 to 1919 and residing on the island of Oahu in At study enrollment from 1965 to 1968, subjects underwent complete physical examinations. Since enrollment, information on cardiovascular events has been obtained through a comprehensive surveillance of hospital discharges, death certificates, autopsies, and periodic repeat physical examinations. Final diagnoses were reviewed and confirmed by the Honolulu Heart Program Morbidity and Mortality Review Committee. Study methodology has been previously described Assays for acl antibodies were based on serum that was collected at a second examination (1967 to 1970) and stored at 70 C. The second examination was taken by 7498 of the original 8006 originally enrolled in the Honolulu Heart Program ( 95% of the surviving cohort). Among this group, serum was available for Among the remaining men, 1751 had no available sera, and 61 had sera that had been thawed 3 times for other analyses. After removal of prevalent stroke and coronary heart disease and the incident cases of stroke and MI that had occurred over 20 years of follow-up, there were 4145 men, from which 1360 controls were selected. The cases of stroke (n 259) were taken from all stroke cases that occurred over the 20 years of follow-up; the cases of MI (n 374) were randomly selected from the 469 incident cases that were observed. As a result of the high follow-up rate at the second examination and the way in which samples were randomly excluded for special substudies, the available serum used for sampling is likely to be representative of the entire Honolulu Heart Program cohort. Cases of stroke and MI were not preceded by the other event. MI cases and controls were without a history of angina pectoris. Thirteen of the 259 stroke cases had a history of angina pectoris. Serological Assays Five ELISAs were used in the present study: antibodies to cardiolipin in the presence of 2GP1 (dependent acl IgG and IgM), antibodies to cardiolipin in the absence of 2GP1 (independent acl IgG and IgM), and unique IgG antibodies to 2GP1 alone in the absence of cardiolipin. 2GP1-dependent acl IgG and IgM assays were performed according to the manufacturer s instructions by using a commercially available kit with cardiolipin as the detecting antigen (Reaads Medical Products). 2GP1-independent acl IgG and IgM assays were performed by using an in-house method with cardiolipin-coated microtiter plates. The remainder of the procedure was identical to the commercial method, except that BSA was used rather than serum in the blocking and washing steps and in the diluent reagent for the patients serum samples and the detecting antibody. Anti- 2GP1 assays for the IgG isotype were performed by using a commercially available kit (Innova Diagnostics). Cutoff values that defined a positive value in the 2GP1-dependent acl IgG and IgM and the anti- 2GP1 IgG assays correspond to those recommended by the manufacturer ( 2GP1-dependent acl IgG 23 GPL, 2GP1-dependent acl IgM 11 MPL, and anti- 2GP1 16 U). For the 2GP1-independent acl assay, the result was considered positive if the amount of immunoreactivity seen on the 2GP1- independent acl IgG or IgM assay was 20% greater than the amount of immunoreactivity seen on the 2GP1-dependent assay. This cutoff was selected because it exceeds the accepted assay variability in our hands. No systematic effect of 1 or 2 freeze-thaw cycles on immunoreactivity was seen. The within-pair coefficient of variation was always 20% for cases, controls, and standards; otherwise, the assay was repeated. Variables Used in Analysis Risk factor information for diabetes, hypertension, total cholesterol, body mass index, and the use of cigarettes was collected at the second examination (1967 to 1970) concurrent with sera collection for the acl and anti- 2GP1 assays. No information about medication use was available. Likewise, the influence of atrial fibrillation could not be assessed. Atrial fibrillation is rare in Japanese men in the Honolulu Heart Program. Only 13 men in the present study had atrial fibrillation at the second examination, and information about future atrial fibrillation could not be assessed because its collection was too limited. If data were missing, measurements were taken from the initial examination that occurred 2 years earlier (1965 to 1968). Data on alcohol intake and physical activity came from the 1965 to 1968 examination. A diagnosis of diabetes was based on a medical history or on the use of insulin or oral hypoglycemic therapy. A diagnosis of hypertension was made when either a systolic or diastolic blood pressure was 160 or 95 mm Hg, respectively, or when a subject was receiving medication for high blood pressure. Measurement of physical activity was derived from an index to quantify overall metabolic output that occurred during a typical 24-hour period. High levels of the index indicated active lifestyles. Further description of the physical activity index and the other risk factors considered in the present report are provided elsewhere. 26,27 Statistical Methods Estimated age-adjusted risk factor comparisons between cases and controls and according to the presence and absence of a positive isotype were derived from ANCOVA methods. 28 To assess the relationship between a positive isotype and the odds of a stroke or MI, statistical analysis relied on unconditional logistic regression models. 29 Because cases of either event were not individually matched to a specific set of controls, adjustment for confounding variables was accomplished through the use of the unconditional models. Estimated age-adjusted percentages of the men with positive antibody values were also derived for cases of stroke and MI and for controls. 28 Regression coefficients and standard errors from ageadjusted and risk factor adjusted models were used to derive estimates of the relative odds of a stroke or MI (and 95% CIs) by comparing men with a positive versus a negative isotype. Doseresponse effects were assessed by examining isotypes as continuous independent variables. Threshold effects were also explored, as were confounding and interaction with other risk factors. Time effects were assessed by examining the association of the isotype with vascular disease during successive 5-year intervals and during increasing cumulative follow-up periods. Results Table 1 gives the age-adjusted percentage of men with positive antibody assays for cases of stroke and MI and for controls. Because of a lack of serum, 2PG1-dependent acl IgM assays were incomplete in 1 man with a stroke and in 6 controls, and the anti- 2PG1 assay was incomplete in 1 man with an MI. Among the men, the percentages with a positive 2PG1-dependent acl IgM, 2PG1-independent acl IgM, and anti- 2PG1 were similar across the 3 study groups. Although positive 2PG1-independent acl IgG was rare, cases of MI were 3 times more likely to have a positive assay than were controls (2.4% versus 0.8%, P 0.05). Stroke cases also had an excess of positive 2PG1-independent acl IgG (1.5%) relative to controls, but the excess was not statistically significant. Among the assays, positive 2PG1-dependent acl IgG was the most frequent in the study sample. The unadjusted frequency of 2GP1-dependent acl IgG among controls was 10.4% (80 of 770) for ages 48 to 55 years, 13.6% (80 of 590) for ages 56 to 70 years, and 11.8% (160 of 1360) for both age groups combined. The age-adjusted percentage of men with a positive 2PG1-dependent acl IgG was significantly higher in cases of stroke than in controls (16.6% versus12.1%, respectively; P 0.05). Compared with stroke cases, cases of MI had a similar percentage with a positive 2PG1-dependent acl IgG (15.8%), although

3 Brey et al acl Antibodies and Risk of Stroke and MI 1703 TABLE 1. Age-Adjusted Percentage of Men With Positive Antibody Values for Cases of Stroke and MI and for Controls Assay Controls Stroke Cases MI Cases 2PG1-dependent acl IgG, % (n) 12.1 (160/1360) 16.6* (46/259) 15.8 (61/374) 2PG1-dependent acl IgM, % (n) 4.4 (60/1354) 3.6 (9/258) 4.6 (17/374) 2PG1-independent acl IgG, % (n) 0.8 (11/1360) 1.5 (4/259) 2.4* (9/374) 2PG1-independent acl IgM, % (n) 9.0 (120/1360) 8.7 (24/259) 10.2 (39/374) Anti- 2PG1, % (n) 1.9 (26/1360) 2.2 (6/259) 1.3 (5/373) Values in parentheses are number of positive assays/sample size. Positive values are 2PG1- dependent acl IgG 23, 2PG1-dependent acl IgM 11, 2PG1-independent acl assays 20% of dependent assay, and anti- 2PG1 16. *P 0.05 vs controls. it was not significantly different from that of controls. There was no evidence for a dose-response relationship between any of the isotypes and the risk of an event. Use of other cutoffs besides those recommended by the manufacturer did not alter these reported findings appreciably. Table 2 shows the age-adjusted mean values for conventional cardiovascular risk factors for cases of stroke and MI and for controls as well as for those with and without a positive 2PG1-dependent acl IgG value. When compared with controls, cases of stroke and MI were more likely to be hypertensive, to have diabetes, higher total cholesterol levels, and body mass indices, and to smoke more cigarettes. MI cases also tended to be less physically active and consumed less alcohol than did controls. Other than age and total cholesterol, risk factor levels were similar in men with and without a positive assay for 2PG1-independent acl IgG. Men with a positive assay tended to be older and have lower total cholesterol levels. Cardiovascular events early in follow-up were more strongly related to acl than were events that occurred later. Among the stroke cases, the percentage of men with a positive 2PG1-dependent acl IgG did not vary significantly within the first 15 years after blood withdrawal (22.7% overall), although it was significantly reduced to 5.4% in the last 5 years of follow-up (P 0.01). Table 3 further describes the relationship of 2PG1-dependent acl IgG with stroke according to 5-, 10-, 15-, and 20-year follow-up periods. At 15 years of follow-up, the association between 2PG1- TABLE 2. Age-Adjusted Percentage and Mean Risk Factor Levels ( SD) for Cases of Stroke and MI and for Controls and for Those With and Without Positive Dependent acl IgG Values Risk Factor Controls ( y)* Stroke Cases ( y) MI Cases ( y) Hypertension, % 10.8 (145/1360) 27.2 (73/259) 21.0 (80/374) Diabetes, % 13.6 (182/1360) 27.2 (73/259) 23.9 (91/374) Total cholesterol (mmol/l) mg/dl Body mass index, kg/m Alcohol intake, oz/mo Physical activity index Cigarettes smoked/day Dependent acl IgG Positive ( ) Negative ( ) Hypertension, % 12.7 (36/267) 15.3 (262/1726) Diabetes, % 16.8 (47/267) 17.5 (299/1726) Total cholesterol, mmol/l mg/dl Body mass index, kg/m Alcohol intake, oz/mo Physical activity index Cigarettes smoked/day Values in parentheses are number/sample size. *Mean age in years at the time of assay. P 0.01 vs controls; P 0.05 vs controls; P vs those with negative assay; and P 0.05 vs those with negative assay.

4 1704 Stroke August 2001 TABLE 3. Age-Adjusted Percentage of Men With a Positive Dependent acl IgG Value for Cases of Stroke and MI and for Controls by Period of Follow-Up After Assay Follow-Up Period Controls, % (n) Stroke Cases, % (n) MI Cases, % (n) 5 y 12.9 (238/1852) 16.6 (12/65) 21.5* (17/76) 10 y 12.4 (208/1705) 21.4 (27/115) 17.6* (32/173) 15 y 11.7 (178/1553) 20.9 (42/185) 17.8 (47/255) 20 y 12.1 (160/1360) 16.6* (46/259) 15.8 (61/374) Values in parentheses are number positive/sample size. P 0.005, P 0.01, and P for significant excess in cases vs controls. dependent acl IgG and stroke appears strongest, with 20.9% of men with a stroke having a positive 2PG1-dependent acl IgG value versus 11.7% of the controls (P 0.001). Similar findings were also observed in the relationship between 2PG1-dependent acl IgG and MI. Among the MI cases, the percentage of men with a positive 2PG1- dependent acl IgG did not vary greatly within the first 15 years after blood withdrawal (18.4% overall). Although not statistically significant, it was reduced to 15.8% in the last 5 years of follow-up. As seen in Table 3, the effect of 2PG1-dependent acl IgG on MI seems slightly stronger in the first 5 years of follow-up than in the other periods, whereas associations remained significant up to 15 years after blood withdrawal. At 15 years of follow-up, 17.4% of men with an MI had a positive 2PG1-dependent acl IgG value versus 11.7% of the controls (P 0.01). For stroke and MI, adjustment for the effects of other risk factors on the relationship between 2PG1-dependent acl IgG and either event was assessed at 15 and 20 years of follow-up (see Table 4). Data on shorter periods of follow-up are not described because of limited numbers of events. For the 15-year period after blood withdrawal, the overall relative odds ratio of stroke was 2.2 (P 0.001) in the presence versus the absence of a positive 2PG1-dependent acl IgG. In men aged 56 to 70 years, it was 2.7 (P 0.001); in those who were younger, it was slightly weaker (2.1). The relative odds ratios of stroke were similar in men who were neither hypertensive nor diabetic (2.3 and 2.5, respectively; P for each). The presence of a positive 2PG1-dependent acl IgG appeared to increase the odds of stroke by 3-fold in men who were past smokers (4.2, P 0.001) and those who never smoked cigarettes (3.4, P 0.01). For the 15-year follow-up for MI (Table 4), the relative odds ratio of an event in those with a positive versus a negative 2PG1-dependent acl IgG is near 2.0 for the younger and older men (P 0.05). As with stroke, the effect of 2PG1-dependent acl IgG appears strongest in men who had neither hypertension nor diabetes (P 0.01). Unlike stroke, 2PG1-dependent acl IgG had no effect on MI in men who never smoked cigarettes, whereas the odds ratio of MI was more than doubled in past and current smokers (P 0.05). Although the presence or absence of some of the cardiovascular risk factors in Table 4 appeared to alter the observed relative odds of either event, interaction effects were not statistically significant, and any could have occurred by chance alone. Discussion The present findings support a prospective association not only between acl and MI but also between these antibodies TABLE 4. Adjusted Relative Odds of Stroke and MI for Men With Positive vs Negative 2PG1-Dependent acl IgG Value for 15 and 20 y of Follow-Up After Assay Adjusted Relative Odds (95% CI) 15-y Follow-Up 20-y Follow-Up Risk Strata Stroke MI Stroke MI Age, y ( ) 1.9* ( ) 1.3 ( ) 1.6 ( ) ( ) 2.0 ( ) 1.9 ( ) 1.7* ( ) Hypertension Absent 2.3 ( ) 1.9 ( ) 1.5 ( ) 1.5* ( ) Present 2.1 ( ) 1.7 ( ) 2.1 ( ) 1.6 ( ) Smoking status Never 3.4 ( ) 0.8 ( ) 1.5 ( ) 0.7 ( ) Past 4.2 ( ) 2.3* ( ) 2.6 ( ) 2.0* ( ) Current 1.2 ( ) 2.3 ( ) 1.1 ( ) 1.9* ( ) Diabetes Absent 2.5 ( ) 1.9 ( ) 1.7* ( ) 1.6* ( ) Past 1.7 ( ) 1.6 ( ) 1.1 ( ) 1.3 ( ) Overall 2.2 ( ) 1.8 ( ) 1.5* ( ) 1.5* ( ) Relative odds are adjusted for total cholesterol, body mass index, alcohol intake, physical activity index, and the other risk factors in this table. *P 0.05, P 0.001, and P 0.01 for significant odds ratio.

5 Brey et al acl Antibodies and Risk of Stroke and MI 1705 and ischemic stroke in men. Prior published prospective studies of antiphospholipid antibodies and stroke were either negative 15,16 or did not report an independent association. 17 However, these studies were different from the present study in acl assay technique and were limited in statistical power. To our knowledge, the present study is the first to demonstrate a prospective association between acl and stroke independent of other risk factors. Although less consistent, independent acl IgG also appeared to be related to vascular events. Although the percentage of men with positive independent acl IgG was 2 to 3 times higher in those with stroke and MI than in controls, its low prevalence ( 1% in controls) may have limited the power to detect stronger associations with each event. In contrast, positive dependent acl IgG was common (12.1% in controls), and within 15 years of followup, there was a near 2-fold increase in the odds of stroke or MI when it was present compared with when it was absent. Among the studies linking acl to MI, 2 prior prospective studies suggested the possibility of a time-dependent association between acl and MI in men. 13,14 In the first study, 13 the follow-up period was 5 years, and in the second study, 14 a significant association was found only for the first 10 years of follow-up, despite the fact that the second 10 years of follow-up had more than double the number of events. Our analyses show a decreasing association of acl with vascular disease with increasing duration of follow-up for both the MI and stroke end points. Such weakening of the association with time could be due to changes in risk factor status during follow-up, including acl. Our data suggest this possibility, inasmuch as the percentage with acl changed little among controls as the duration of follow-up increased, whereas there was a general trend for decreasing acl levels with time among the MI cases accumulating over time. Except for the initial follow-up period of 5 years, a similar statement can be made for the stroke cases. Although one cannot exclude the possibility that there is no association between acl and stroke in the initial follow-up period, it seems more plausible that this observation was due to the limited number of events early in follow-up. In addition, despite the prior belief that all infection-related antiphospholipid antibodies are nonpathogenic, 30 infection could be a trigger for the development of elevated acl and subsequent stroke or MI, as has been suggested by some 31 but not other 32 clinical studies. In animal models, pathogenic acl has been induced by phospholipid-binding peptides. 33 The finding of a prospective association between acl antibodies of the IgA isotype and MI 13 further supports the potential for infection to induce pathogenic antibodies. A recent prospective study 17 showing an association between acl antibodies of the IgM isotype and stroke of all types is also consistent with this hypothesis. It is unlikely that the observed associations between acl and vascular disease are due to chance or an artifact of study design. Although an indirect association can never be entirely excluded in an observational study, acl is weakly associated with other conventional vascular risk factors in other studies 4,14 and with age and cholesterol in the present report. Other considerations in assessing the causal nature of an association include its temporality, strength, consistency, dose-response relationship, biological plausibility, and experimental evidence. 34 Prior prospective studies of MI 13,14 and the present study of MI and stroke preclude the possibility that the antibodies were a consequence of the vascular event but, as noted above, do not exclude the possibility that the antibodies are a consequence of preclinical disease. 34 Our findings were unchanged when we excluded the 13 stroke cases with angina pectoris at baseline. The association between acl and MI is also relatively consistent across different populations and across both casecontrol 1 9 and prospective findings. It is also possible that technical laboratory differences account for some of the inconsistent findings. 35,36 With regard to the dose-response relationship, the present study and prior data 37 are more suggestive of a threshold effect, although a dose-response relationship at the higher range of the acl distribution may still exist. The association between acl and thrombosis is biologically plausible and supported by experimental data from animal models. Both passive and active immunization of normal laboratory mice with either acl or with 2GP1 results in the induction of an experimental antiphospholipid antibody syndrome, including thrombocytopenia, placental infarction and fetal loss, MI, and neurological dysfunction. 38,39 In addition, mice with pinch-clamp injury to vascular endothelium develop a larger clot size with a longer time to dissolution when they are treated with human antiphospholipid antibody compared with control IgG. 40 Taken together, these studies provide important evidence that antiphospholipid antibodies can cause thrombosis and other antibody-mediated clinical manifestations. In conclusion, the present findings extend prior cohort studies of acl antibodies and MI by showing an association of similar strength for stroke. Given the prevalence of 2GP1-dependent acl IgG, the prospective and independent association between this antibody and both stroke and MI in men, and the experimental evidence supporting a causal mechanism, acl antibodies may represent an important new target for interventions to prevent vascular disease. All prospective studies performed thus far have evaluated the risk for stroke and MI associated with antiphospholipid antibodies only in men; a future challenge is to perform similar studies that include women. Acknowledgment This study was supported by grant R01 NS from the National Institute for Neurological Disorders and Stroke and 5M01 RR from the National Center for Research Resources (Dr Bray), the National Institutes of Health, Bethesda, Md. References 1. Brey RL, Hart RG, Sherman DG, Tegeler CH. Antiphospholipid antibodies in cerebral ischemia in young people. Neurology. 1990;40: Angelini L, Ravelli A, Caporali R, Rumi V, Nardocci N, Martini A. High prevalence of antiphospholipid antibodies in children with idiopathic cerebral ischemia. Pediatrics. 1994;94(pt 1): Chakravarty KK, Al-Hillawi AH, Byron MA, Durkin CJ. Anticardiolipin antibody associated ischemic strokes in elderly patients without system lupus erythematosus. Age Ageing. 1990;19:

6 1706 Stroke August Anticardiolipin antibodies are an independent risk factor for first ischemic stroke: the Antiphospholipid Antibodies in Stroke Study (APASS) group. Neurology. 1993;43: Hess DC, Krauss J, Adams RJ, Nichols FT, Zhang D, Rountree HA. Anticardiolipin antibodies: a study of frequency in TIA and stroke. Neurology. 1991;41: Kushner MJ. Prospective study of anticardiolipin antibodies in stroke. Stroke. 1990;21: Nencini P, Baruffi MC, Abbate R, Massai G, Amaducci L, Inzitari D. Lupus anticoagulant and anticardiolipin antibodies in young adults with cerebral ischemia. Stroke. 1992;23: Nagaraja D, Christopher R, Manjari T. Anticardiolipin antibodies in ischemic stroke in the young: Indian experience. J Neurol Sci. 1997;150: Yilmaz E, Adalet K, Yilmaz G, Badur S, Erzengin F, Koylan N, Ozsaruhan O, Ertem G, Buyukozturk K. Importance of serum anticardiolipin antibody levels in coronary heart disease. Clin Cardiol. 1994;17: Phadke KV, Phillips RA, Clarke DT, Jones M, Naish P, Carson P. Anticardiolipin antibodies in ischaemic heart disease: marker or myth? Br Heart J. 1993;69: Muir KW, Squire IB, Alwan W, Lees KR. Anticardiolipin antibodies in an unselected stroke population. Lancet. 1994;344: Metz LM, Edworthy S, Mydlarski R, Fritzler MJ. The frequency of phospholipid antibodies in an unselected stroke population. Can J Neurol Sci. 1998;25: Vaarala O, Manttari M, Manninen V, Tenkanen L, Puurunen M, Aho K, Palosuo T. Anti-cardiolipin antibodies and risk of myocardial infarction in a prospective cohort of middle-aged men. Circulation. 1995;91: Wu R, Nityanand S, Berglund L, Lithell H, Holm G, Lefvert AK. Antibodies against cardiolipin and oxidatively modified LDL in 50-year-old men predict myocardial infarction. Arterioscler Thromb Vasc Biol. 1997;17: Ginsburg KS, Liang MH, Newcomer L, Goldbaber SZ, Schur PH, Henekens CH, Stampfer MJ. Anticardiolipin antibodies and the risk for ischemic stroke and venous thrombosis. Ann Intern Med. 1992;117: Sletnes KE, Smith P, Abdolnoor M, Arnosen H, Wisloff F. Antiphospholipid antibodies after myocardial infarction and their relation to mortality, reinfarction, and non-haemorrhagic stroke. Lancet. 1992;339: Ahmed E, Stegmayr B, Trifunovic J, Weinehall L, Hallmans G, Lefvert AK. Anticardiolipin antibodies are not an independent risk factor for stroke: an incident case-referent study nested within the MONICA and Vasterbotten cohort project. Stroke. 2000;31: McNeil HP, Chesterman CN, Krilis SA. Immunology and clinical importance of antiphospholipid antibodies. Adv Immunol. 1991;49: Matsuura E, Igarashi Y, Fujimoto M, Ichikawa K, Koike T. Anticardiolipin cofactors and the differential diagnosis of autoimmune disease. Lancet. 1990;336: Galli M, Comfurius P, Maasen C, Hemker HC, de Baets MH, van Breda-Vriesman PJ, Barbui T, Zwaal RF, Bevers EM. Anticardiolipin antibodies (ACA) are directed not against cardiolipin but to a plasma cofactor. Lancet. 1990;335: Nojima J, Suehisa E, Akita N, Toku M, Fushimi R, Tada H, Kuratsune H, Machii T, Kitani T, Amino N. Risk of arterial thrombosis in patients with anticardiolipin antibodies and lupus anticoagulant. Br J Haematol. 1997; 96: Viard JP, Amoura Z, Bach JF. Association of anti- 2 glycoprotein 1 antibodies with lupus-type circulating anticoagulants and thrombosis in systemic lupus erythematosus. Am J Med. 1992;93: Kagan A, Popper JS, Rhoads GG. Factors related to stroke incidence in Hawaii Japanese men: the Honolulu Heart Study. Stroke. 1980;11: Benfante R, Yano K, Hwang L, Curb JD, Kagan A, Ross W. Elevated cholesterol is a risk factor for both coronary heart disease and thromboembolic stroke in Hawaiian Japanese men: implications of shared risk. Stroke. 1994;25: Yano K, Reed DM, McGee DL. Ten-year incidence of coronary heart disease in the Honolulu Heart Program: relationship to biologic and lifestyle characteristics. Am J Epidemiol. 1984;119: Donahue RP, Abbott RD, Reed DM, Yano K. Physical activity and coronary heart disease in meddle-aged and elderly men: The Honolulu Heart Program. Am J Public Health. 1988;78: Kagan A, Harris BR, Winkelstein W Jr, Johnson KG, Kato H, Syme SL, Rhoads GG, Gay ML, Nichaman MZ, Hamilton HB, et al. Epidemiologic studies of coronary heart disease and stroke in Japanese men living in Japan, Hawaii, and California: demographic, physical, dietary and biochemical characteristics. J Chronic Dis. 1974;27: Lane PW, Nelder JA. Analysis of covariance and standardization as instances of prediction. Biometrics. 1982;38: Hosmer DW, Lemeshow S. Applied Logistic Regression. New York, NY: John Wiley & Sons Inc; Campbell AL, Pierangeli SS, Wellhausen S, Harris EN. Comparison of the effects of anticardiolipin antibodies from patients with the antiphospholipid antibody syndrome and with syphilis on platelet activation and aggregation. Thromb Haemost. 1995;73: Ameriso SF, Wong VLY, Quismorio FP, Fisher M. Immunohematologic characteristics of infection-associated cerebral infarction. Stroke. 1991; 22: Grau AJ, Buggle F, Becher H, Zimmermann E, Spiel M, Fent T, Maiwald M, Werle E, Zorn M, Hengel H, et al. Recent bacterial and viral infection is a risk factor for cerebrovascular ischemia: clinical and biochemical studies. Neurology. 1998;50: Gharavi AE, Peirangeli SS. Origin of antiphospholipid antibodies: induction of apl by viral peptides. Lupus. 1998;7:S52 S Hill AB. The environment and disease: association or causation? Proc R Soc Med. 1965;58: Brey RL, Cote SA, McGlasson ML, Triplett DA, Barna LK. Effects of repeated freeze-thaw cycles on anticardiolipin antibody immunoreactivity. Am J Clin Pathol. 1994;102: McGlasson D, Brey RL, Strickland DM, Patterson WR. Differences in kaolin clotting times and platelet counts resulting from variations in specimen processing. Clin Lab Sci. 1989;2: Tuhrim S, Rand JH, Wu XX, Weinberger J, Horowitz DR, Goldman ME, Godbold JH. Elevated anticardiolipin antibody titer is a stroke risk factor in a multiethnic population independent of isotype or degree of positivity. Stroke. 1999;30: Brey RL, Cote S, Barohn R, Jackson C, Crawley R, Teale JM. Model for the neuromuscular complications of systemic lupus erythematosus. Lupus. 1995;4: Ziporen L, Shoenfeld Y. Antiphospholipid syndrome: from patient s bedside to experimental animal models and back to the patient s bedside. Hematol Cell Ther. 1998;40: Pierangeli SS, Harris EN. Antiphospholipid antibodies in an in vivo thrombosis model in mice. Lupus. 1994;3: