IgG Anticardiolipin Antibody Titer >40 GPL and the Risk of Subsequent Thrombo-occlusive Events and Death
A Prospective Cohort Study
Background Anticardiolipin antibodies (aCL) have been associated with an increased risk of stroke and thrombo-occlusive events. Little is known about the influence of aCL on recurrent thrombo-occlusive events.
Methods Consecutively identified patients (n=132) with focal cerebral ischemia [stroke=112, transient ischemic attack (TIA)=20] harboring aCL of at least 10 GPL units at the time of their index event were prospectively followed to estimate the effect of aCL titer on time to and risk of subsequent thrombo-occlusive events (stroke, TIA, deep venous thrombosis, pulmonary embolism, myocardial infarction) and death. On the basis of prior literature, we divided patients into those with aCL ≤40 GPL (n=111; mean age, 63±14 years; mean follow-up, 1.95 years) and those with aCL >40 GPL (n=21; mean age, 54±20 years; mean follow-up, 1.50 years).
Results There was no difference between groups for prevalence of hypertension, diabetes mellitus, cigarette smoking, atrial fibrillation, prior TIA, or sex. The GPL >40 group was younger (54±20 versus 63±14 years; P=.055), had more prior strokes [9/21 (48%) versus 27/111 (20%); P=.030], more frequent subsequent thrombo-occlusive events and death [15/21 (71%) versus 51/111 (48%); P=.030], and a shorter median time (years) to event (0.15 versus 0.61, log rank P=.005). The risk ratio for recurrent event and death with GPL >40 obtained from Cox proportional hazards models, adjusted for prior strokes, prior TIAs, hypertension, diabetes mellitus, atrial fibrillation, and cigarette smoking was 1.9 (95% confidence interval, 1.0 to 3.5; P=.051).
Conclusions Our data suggest that subsequent thrombo-occlusive events and death after focal cerebral ischemia associated with IgG aCL may occur sooner and more frequently with GPL >40.
- antibodies, anticardiolipin
- antibodies, antiphospholipid
- lupus coagulation inhibitor
There is evidence supporting an association (if not pathogenicity) of aPL with thrombosis1 and thrombo-occlusive events including focal cerebral ischemia.2 3 4 aCLs, a subgroup of aPL, are a quantifiable aPL5 6 7 8 and have been shown to be an independent risk factor for stroke.9 Preliminary data suggest that recurrent thrombo-occlusive events may be relatively common in patients with aPL and first thrombo-occlusive events.10 11 Furthermore, data based on relatively small sample sizes suggest that patients with the highest aCL titers may be at greatest risk for subsequent thrombo-occlusive events.10 12 13 14 A recently completed prospective multicenter follow-up study did not show a significant increase in recurrent thrombo-occlusive events and death in the Antiphospholipid Antibodies and Stroke Study (APASS) prevalence cohort when the IgG aCL positivity cutoff was >10 GPL compared with aCL negative patients.15 On the basis of preliminary data from systemic lupus erythematosus patients12 and obstetric patients,13 we divided our study cohort into two groups based on IgG aCL titer obtained within 7 days of the index events. The GPL cutoffs were ≤40 (n=111) and >40 (n=21). This dichotomy has been used to stratify risk based on preliminary data for miscarriage in obstetric patients13 and thrombotic events in patients with systemic lupus erythematosus (SLE).12
Our specific aims of this study were (1) to assess the influence of the IgG aCL titer in patients at the time of their index event of focal cerebral ischemia on the occurrence and timing of subsequent thrombo-occlusive events and death and (2) to estimate the risk ratio for recurrent thrombo-occlusive events and death associated with IgG aCL>40 GPL in patients with index focal cerebral ischemia.
We prospectively studied a cohort of consecutively identified patients at Henry Ford Hospital with focal cerebral ischemia, either ischemic stroke or TIA who had sera drawn by venipuncture for IgG aCL determination within 7 days of their index event. The study was approved by the Institutional Review Board of our institution, and subjects or their next of kin gave informed consent. Specific exclusion criteria included American Rheumatism Association criteria for SLE,16 human immunodeficiency virus infection previously documented or documented during the index event hospitalization, syphilis, myocardial infarction within the previous 6 weeks, and blood drawn for IgG aCL more than 7 days after the index event.
aCL was assayed for the IgG isotope by enzyme-linked immunosorbent assay as previously described9 and standardized via the Second International Standardization Workshop7 and reported as GPL (IgG phospholipid) units. One GPL unit is equivalent to 1 μL/mL of IgG aCL immunoreactivity.
Index Event of Focal Cerebral Ischemia
Stroke and TIA were diagnosed and defined on the basis of the Classification and Diagnosis of Cerebrovascular Diseases.17 All patients underwent head CT scanning (GE 8800 or 9800 models) and were examined by a neurologist.
End points were subsequent thrombo-occlusive events and death. They were all assessed prospectively and blinded to aCL titer as were the head CT and MRI scans. We defined recurrent thrombo-occlusive events as ischemic stroke, TIA, DVT, PE, MI, peripheral or systemic visceral/arterial or venous thrombo-embolism as previously published.15 End points were determined using both clinical and investigative criteria and standardized forms at the time of patient follow-up, telephone contact, and medical record review. Patients were telephoned every 6 months from the time of the index event. If they could not be reached, a mailing with standard questions was sent to the last known mailing address to be completed and returned.
If patients were unable to be contacted, searches were carried out using Equifax to ascertain if death had occurred or to locate an individual who may have moved. Equifax is a nationwide death search with more than 61 million deaths archived since 1955. Systematic efforts were used in all cases to review as much data as possible from other hospitals when recurrent events occurred and patients were hospitalized outside our institution. The final determination of an end point was made with all identifiers and aCL titer data removed. All causes of death were ascertained from death certificates, family, and medical records.
The GPL groups were compared with respect to three prespecified outcomes: (1) time until first recurrence of any thrombo-occlusive event, (2) time until first recurrence of ischemic stroke, and (3) time until first recurrence of either ischemic stroke, myocardial infarction, or death. Initially, unadjusted Kaplan-Meier survival curves were generated for each outcome and compared using log-rank tests. Cox proportional hazards models were performed to adjust for prior strokes, prior TIAs, hypertension, diabetes mellitus, history of cigarette smoking, and atrial fibrillation. Adjusted risk ratios were derived from these analyses. Pearson’s χ2 tests and Student’s t tests were used to compare the GPL groups with respect to various demographic and medical history variables.
In an attempt to refine the prediction of the occurrence of subsequent thrombo-occlusive events based on GPL data, we used ROC curves18 19 20 to find the IgG aCL value that “best” discriminates between the occurrence of subsequent events and no further events during follow-up. ROC curves are used to describe and compare the performance of diagnostic technology (in this case aCL titer). ROC curves are obtained by considering each unique value of an outcome to be a possible cut point, and calculating sensitivities and specificities for that cut point. These sensitivities and specificities use a gold standard measure as the true measure of the disease (recurrent thrombo-occlusive events). After each cut point is evaluated, a plot of the sensitivities versus (1–specificities) is generated and the area under the ROC curve, θ, is calculated. This area can range from 0 to 1 and reflect the discriminatory value of an outcome. An area of 0.5 (a straight line on the diagonal) suggests the outcome has no apparent accuracy in discriminating between higher aCL titer and lower aCL titer subjects, whereas an area of 1 indicates perfect accuracy. The area under the curve is compared with 0.5 to determine if the outcome has any discriminatory value.
The mean± SD age of the study population (n=132) was 61.5±15.3 years (range, 20 to 89 years). Table 1⇓ summarizes the demographic characteristics and stroke risk factor profiles of the two groups based on index GPL cutoffs. The mean age of the GPL >40 group was approximately 10 years younger (54±20 years) than the GPL ≤40 group (63 ±14 years), P=.055 (two-sample t test). There were no differences between groups for the prevalence of hypertension, diabetes mellitus, cigarette smoking, atrial fibrillation, prior TIA, or sex. The GPL >40 group had more prior strokes (48% versus 20%, P=.030).
IgG aCL Results
There were 21 patients with GPL >40 and 111 patients with GPL ≤40. The frequency distribution of GPL values for the cohort with GPL >40 was as follows: 40 to 45 GPL, n=4; 46 to 50 GPL, n=2; 51 to 55 GPL, n=2; 56 to 60 GPL, none; 61 to 65 GPL, n=1; 66 to 70, n=1; 71 to 75 GPL, n=1; 76 to 80 GPL, n=1; 81 to 95 GPL, none; 96 to 100, n=1; and >100 GPL, n=8.
Mean length of prospective follow-up for the entire cohort was 686 days (1.9 years) (±436 days [1.2 years]; range, 0 to 1827 days [5.0 years]). There was no statistically significant difference in the mean length of follow up between the two groups: 550±500 days (1.5±1.4 years) for the GPL >40 group versus 711±420 days (1.9±1.2 years) for the GPL ≤40 group, P=.121 (two-sample t test).
Recurrent thrombo-occlusive events (ischemic stroke, TIA, DVT, PE, MI, and death) were documented in 15 (71%) of the GPL >40 group and in 51 (46%) of the GPL ≤40 group (P=.030). The median time to subsequent event for the GPL >40 group was 0.16 year (1.9 months) compared with 0.61 year (7.4 months) for the GPL ≤40 group (log-rank P=.005). The mean time (and range for recurrent events was 0.67 year (1 day to 2.8 years) for the GPL >40 group and 1.02 years (0 days to 3.3 years) for the GPL ≤40 groups. Table 2⇓ summarizes the type and frequency of recurrent thrombo-occlusive events that occurred in the cohort on the basis of the GPL cutoffs.
Male and female patients did not differ with respect to time to first recurrent event.
Among all patients with GPL >40, 29% suffered only subsequent TIAs compared with 5% of the GPL ≤40 group, P=.003. This represented approximately a sixfold increased risk of subsequent TIA in the GPL >40 group. The absolute increased risk of ischemic stroke (fatal or nonfatal, with or without other thrombo-occlusive events) in the GPL >40 group was 4% (20% versus 24%). Only 2% of the entire cohort suffered a subsequent DVT during follow-up (only in the IgG aCL ≤40 group) and only 4% suffered an MI (again, only in the IgG aCL ≤40 group). Mortality (with or without prior thrombo-occlusive events) occurred in 27% of the GPL ≤40 and 24% of the GPL >40 group. The risk ratio for any recurrent thrombo-occlusive event with GPL >40 was 1.9 (95% CI, 1.0 to 3.5; P=.051).
Ischemic Stroke Subgroup
Table 3⇓ summarizes the frequency of events and time to subsequent thrombo-occlusive events for the subgroup of patients with index stroke based only on baseline GPL status. In the subgroup of index ischemic stroke (n=112), recurrent ischemic stroke occurred almost twice as frequently in the higher titer group: in 5 (39%) of the GPL >40 group and 21 (21%) of the GPL ≤40 group. Median time to recurrent ischemic stroke in this subgroup was an order of magnitude sooner for the GPL >40 group than for the GPL ≤40 group (log-rank P=.082). The mean time (years) to recurrent ischemic stroke was approximately 5 months sooner for the GPL>40 group than for the lower titer group.
Using ROC curves for the entire cohort to refine a prediction for occurrence of subsequent thrombo-occlusive events based on GPL data (ie, the best GPL value that discriminates between subsequent events and no subsequent events), we were not able to find a specific IgG aCL value (GPL unit) that was better than any other IgG aCL value in accurately discriminating between patients who experienced subsequent events and those who did not experience subsequent events for three different analyses of outcome variables: (1) any thrombo-occlusive event minus area under the ROC curve=0.595 (SE=0.050), (2) ischemic stroke minus area under ROC curve=0.536 (SE=0.060), and (3) ischemic stroke or death minus area under the ROC curve=0.546 (SE=0.052). (The closer the area under the ROC curve is to 1, the more likely there will be a discriminating IgG aCL value.) These analyses did not take into consideration the time to event.
The Figure⇓ shows the Kaplan-Meier survival curves generated for the two groups based on the a priori GPL cutoff of 40.
Patients with GPL >40 were more likely to have been on aspirin (most common dose was 325 mg/d) with or without other therapeutic agents at the time of their index focal cerebral ischemic event (40% versus 18%, P=.038, likelihood ratio, χ2). Twenty-six percent of patients with GPL ≤40 were on relevant medications at the time of their index event compared with 40% of patients with GPL >40 (P=.217). Five percent of patients with GPL≤40 were on anticoagulants at baseline compared with none in the GPL >40 group. Only one subject (in the GPL ≤40 group) was on both Coumadin and aspirin at the time of the index event. Corticosteroids (alone or in combination) were taken by 3% of the GPL ≤40 group and 5% of the GPL >40 group. Forty percent of the entire cohort was on aspirin at the time of their index event. Of this 40%, 15% were from the GPL ≤40 group and 25% were from the GPL >40, P=.307.
Table 4⇓ summarizes the therapies the patients were prescribed at the time they suffered their recurrent thrombo-occlusive events. Of the entire cohort, only 13% were not on any relevant medications at the time of a recurrent event. Significantly more patients in the GPL >40 group were administered anticoagulants in combination with other therapeutic agents (76% versus 37%, P=.001, likelihood ratio, χ2). The GPL ≤40 group was more likely to be on aspirin alone (46% versus 14%, P=.003, likelihood ratio, χ2). Combinations of antithrombotics, anticoagulants, and/or corticosteroids were commonly prescribed after the index event: in 25% of the GPL ≤40 group and in 43% of the GPL >40 group.
The adjusted risk ratios (RR; adjusted for prior strokes, prior TIAs, hypertension, diabetes mellitus, atrial fibrillation, cigarette smoking, age, and sex) and the 95% CIs for the entire cohort and the index stroke subgroup were calculated. Cox proportional hazard models were built to compare recurrent event/death functions defined by GPL >40. For the entire cohort, the adjusted RR for GPL >40 for any recurrent thrombo-occlusive event was 1.82 (95% CI, 0.92 to 3.60; P=.085). The adjusted RR for recurrent stroke was 1.36 (95% CI, 0.49 to 3.76, P=.551). For the ischemic stroke subgroup, the adjusted RR for GPL>40 for any recurrent thrombo-occlusive event was 1.64 (95% CI, 0.74 to 3.62, P=.221), and for stroke alone was 1.98 (95% CI, 0.68 to 5.76; P=.212).
The event-free survival estimates based on time to first recurrent event for the GPL ≤40 and GPL >40 groups were as follows: At 1 year post index event, the survival estimate for the GPL ≤40 group was 0.70 (95% CI, 0.62 to 0.79) and for the GPL >40 was 0.48 (95% CI, 0.26 to 0.69); by 2 years post index event, the survival estimate for the GPL ≤40 group was 0.63 (95% CI 0.54 to 0.72) and for the GPL >40 was 0.33 (95% CI, 0.13 to 0.53). By 4 years post index event, the survival estimate for the GPL ≤40 group was 0.36 (95% CI, 0.17 to 0.54) and for the GPL >40 group was 0.17 (95% CI, 0.00 to 0.42).
Our data suggest that subsequent thrombo-occlusive events and death after focal cerebral ischemia may occur sooner and more frequently in patients found to have IgG aCL immunoreactivity >40 GPL units at the time of their index focal cerebral ischemic event when compared with patients who harbor 10 to 40 GPL at the time of their index event. Therefore GPL >40 may be an important prognostic marker for subsequent thrombo-occlusive events in patients with index focal cerebral ischemia. Patients with >40 GPL also appear to have more features of the antiphospholipid syndrome: younger age at onset and more recurrent strokes at the time of diagnosis.21 Furthermore, being in the GPL >40 group appears to confer a sixfold-increased risk of developing subsequent TIAs only compared with being in the GPL ≤40 group. TIAs do not generally have as equally bad an outcome as ischemic stroke, although many hemispheric TIAs are minor infarcts by persistent subtle signs or neuroimaging abnormalities, and cognitive function may be subtly or slightly impaired after TIAs. In the ischemic stroke subgroup of the entire cohort, recurrent strokes occurred approximately twice as quickly in the higher titer group, with a median time to recurrence approximately 10 times sooner.
We did not find any clinically meaningful or prognostic features based on IgM aCL titers in this cohort. Our data support the importance of aCL isotype specificity, specifically IgG, in predicting clinical complications associated with aCL as others have reported.22 23 In our clinical experience, elevated IgM aCL titers alone are more variable and transient than IgG aCL titers.
There are several limitations to this study. Despite prospectively identifying a consecutive group of patients with GPL >40, we had a relatively small sample size group (n=21). We systematically only measured aCL among all aPL. Antiphosphatidylserine antibodies may also be important5 24 and may have greater specificity for identifying patients with the antiphospholipid syndrome.
There were a small number of specific types of subsequent thrombo-occlusive events, which leads to low statistical power for any subgroup analyses. Although some patients were followed for up to 5 years, mean follow-up was less than 2 years for both groups. In building adjusted Cox proportional hazards models there may be a regression dilution effect associated with the measurement of continuous variables with error,25 thereby underestimating the strength of association of outcome. Despite these considerations, there is evidence to support the hypothesis that patients with higher titers of IgG aCL may be at increased risk for subsequent events, even with the best empirical medical therapy (risk factor modification, antiplatelets, anticoagulants, corticosteroids, immunosuppressants, immunomodulation, plasmapheresis).1 10 14
Based on the equal distribution of other stroke risk factors (hypertension, diabetes mellitus, cigarette smoking, atrial fibrillation, prior TIA, or sex) and mean length of follow-up, it is unlikely that an imbalance in these factors played a significant role in determining prognosis in our study. There was actually somewhat fewer hypertensives, diabetics, men, cigarette smokers and less mean length of follow-up in the GPL >40 group. However, the higher titer group had more prior strokes identified, further emphasizing the risk of recurrent events in patients with IgG aCL >40 GPL. Over two thirds of the GPL >40 group suffered a recurrent event during prospective follow-up compared with less than half of the ≤40 GPL group, despite only a 4% absolute risk of stroke in the higher titer group. The majority of cerebral nonfatal events in the ≤40 GPL group were strokes, whereas they were TIAs in the >40 GPL group. Occurrences of DVT, MI, and PE were uncommon, supporting the previous observation of a similar vascular bed for recurrent events as the index event.11 A titer >40 GPL conferred almost a twofold-increased risk for the occurrence of any further thrombo-occlusive event or death. Ginsburg et al1 found that a titer of >33 GPL identified a group at increased risk of subsequent DVT or PE. Escalante et al26 and Finazzi et al27 also found GPL titer to be important in identifying a higher risk group of patients for subsequent thrombo-occlusive events. Others have found that the higher GPL titers may confer an increased risk of fetal loss,28 29 whereas lower titers may not be as significant a risk factor.30
We could not refine our precision of estimating a better titer cutoff for predicting subsequent events, although our sample size may have precluded this. ROC curves allow a more refined estimate of positive predictive values and improved test accuracy that may improve upon a simple dichotomization of a laboratory test. There is generally very good interlaboratory and interassay agreement/reliability for detecting higher positive GPL titers (APASS [R.L. Brey, S.R. Levine], unpublished data).
While patients were treated empirically during the study, more patients in the higher titer group were receiving aspirin (alone or in combination with other antiplatelets, anticoagulants, or corticosteroids) at the time of their index event (possible due to a higher prevalence of prior stroke). Furthermore, more patients were on anticoagulants (alone or in combination) in the GPL >40 group at the time of subsequent thrombo-occlusive events and death and significantly fewer were on aspirin. This suggests that recurrent events frequently still occur despite the prescription of empirical medical therapy with antiplatelet agents or anticoagulants or some combination. Therefore, controlled, randomized, standardized therapy to assess future risk in aCL positive patients will be important to further clarify the effects of treatment on the natural history of the condition.31
In summary, although our data suggest that the patients with index focal cerebral ischemia are at greater risk for recurrent thrombo-occlusive events and death if they harbor GPL>40, further study in larger cohorts will be needed to confirm and extend our results. Currently, such a study is under way, the WARSS-APASS collaboration.31 As we did not assay for β2-glycoprotein 1 (the aPL cofactor), further studies will also need to determine whether cofactor dependency is important in refining the prognostic value of aCL isotype and titer.32 33 34
Selected Abbreviations and Acronyms
|DVT||=||deep venous thrombosis|
|ROC||=||receiver operating characteristic|
|TIA||=||transient ischemic attack|
This study was supported in part by NIH grants NS30896 and NS23393 and an American Heart Association Grant-in-Aid from the Michigan Affiliate. We thank Yolanda Washington and Shirley Sian for expert secretarial assistance.
Presented in part at the 19th International Joint Conference on Stroke and Cerebral Circulation, San Diego, Calif, February 17-19, 1994, and published in abstract form (Stroke. 1994;25:258).
- Received February 20, 1997.
- Revision received July 3, 1997.
- Accepted July 3, 1997.
- Copyright © 1997 by American Heart Association
Ginsburg KS, Liang MH, Newcomer L, Goldhaber SZ, Schur PH, Hennekens CH, Stampfer MJ. Anticardiolipin antibodies and the risk for ischemic stroke and venous thrombosis. Ann Intern Med. 1992;117:997-1002.
Camerlingo M, Casto L, Censori B, Drago G, Frigeni A, Ferraro B, Servalli MC, Radice E, Mamoli A. Anticardiolipin antibodies in acute non-hemorrhagic stroke seen within six hours after onset. Acta Neurol Scand. 1995;92:60-71.
Brey RL, Hart RG, Sherman DG, Tegeler CT. Antiphospholipid antibodies and cerebral ischemia in young people. Neurology. 1990;40:1190-1196.
Gharavi AE, Harris EN, Asherson RA, Hughes GRV. Anticardiolipin antibodies: Isotope distribution and phospholipid specificity. Ann Rheum Dis. 1987;46:1-6.
Antiphospholipid Antibodies in Stroke Study (APASS) group. Anticardiolipin antibodies are an independent risk factor for first ischemic stroke. Neurology. 1993;43:2069-2073.
Rosove MH, Brewer PMC. Antiphospholipid thrombosis: clinical course after the first thrombotic event in 70 patients. Ann Intern Med. 1992;117:303-308.
Verro P, Levine SR, Tietjen GE. Cerebrovascular ischemic events with high positive anticardiolipin antibodies. Stroke. 1995;26:160. Abstract.
Antiphospholipid Antibodies in Stroke Study (APASS) group. Anticardiolipin antibodies and the risk of recurrent thrombo-occlusive events and death. Neurology. 1997;47:91-94.
Special Report from the National Institute of Neurologic Disorders and Stroke (Whisnant JP, chairman). Classification of cerebrovascular diseases, III. Stroke. 1990;21:637-676.
Henderson AR. Assessing test accuracy and its clinical consequences: a primer for receiving operating characteristic curve analysis. Ann Clin Biochem. 1993;30:521-539.
Gharavi AE. IgG subclass and light chain distribution of anticardiolipin and antiDNA antibodies in systemic lupus erythematosus. Ann Rheum Dis. 1988;47:286-290.
Sammaritano LR. Significance of aPL IgG subclasses. Lupus. 1996;5:436-439.
Finazzi G, Brancaccio V, Moia M, Ciavarella N, Mazzucconi G, Schinco PC, Ruggeri M, Pogliani EM, Gamba G, Rossi E, Baudo F, Manotti C, D’Angelo A, Palareti G, De Stefano V, Berrettini M, Barbui T. Natural history and risk factors for thrombosis in 360 patients with antiphospholipid antibodies: a four year prospective study from the Italian registry. Am J Med. 1996;100:530-536.
Aoki K, Hayashi Y, Hirao Y, Yagami Y. Specific antiphospholipid antibodies as a predictive variable in patients with recurrent pregnancy loss. Am J Reprod Immunol. 1993;29:82-87.
The Warfarin-Aspirin Recurrent Stroke Study (WARSS), The Antiphospholipid Antibodies and Stroke Study (APASS), The Patent Foramen Ovale in Cryptogenic Stroke Study (PICSS), The Hemostatic System Activation Study (HAS), The Genes in Stroke (GENESIS) Study. The feasibility of a collaborative double-blind study using an anticoagulant. Cerebrovasc Dis. 1997;7:100-112.
Matsuura E. Heterogeneity of anticardiolipin antibodies defined by the anticardiolipin cofactor. J Immunol. 1992;148:3885-3891.
McNally T, Mackie IJ, Machin SJ, Isenberg DA. Increased levels of B2-glycoprotein I antigen and B2-glycoprotein I binding antibodies are associated with a history of thromboembolic complications in patients with SLE and primary antiphospholipid syndrome. Br J Rheumatol. 1995;34:1031-1036.