Anticardiolipin Antibodies and the Risk of Recurrent Thrombo-occlusive Events and Death

The Antiphospholipid Antibodies and Stroke Study Group (APASS), a consortium of 15 medical centers (see Appendix), previously reported that the presence of anticardiolipin antibody (aCL) is an independent risk factor for first ischemic stroke. ^[1] Two-hundred forty-eight unselected stroke patients and age- and sex-matched hospitalized control subjects demonstrated anticardiolipin immunoreactivity in 9.7% of patients and 4.3% of control subjects. After adjustment for conventional stroke risk factors, aCL positivity conferred an odds ratio for stroke of 2.3. IgG was the only aCL isotype that attained a statistically significant association with ischemic stroke (odds ratio = 2.8).

APASS has now extended its observations in a prospective follow-up study of individuals with a first ischemic stroke and IgG aCL positivity. We sought to define the likelihood of death or a thrombo-occlusive event in the stroke cohort identified in the earlier APASS prevalence study. ^[1]

Methods.

Patients or their surrogates were monitored by follow-up visits or interviewed by telephone after their index stroke. Data obtained included information about death or thrombo-occlusive events and treatment. Information as to the cause of death was obtained from patient records or death certificates, whenever possible. For individuals who died, we have no information as to whether a thrombo-occlusive event occurred before death, unless such an event precipitated the patient's demise.

All aCL assays were performed in one central laboratory (Dr. Robin Brey, University of Texas Health Science Center, San Antonio), as previously described. ^[1] Patients were defined as being aCL positive if their titer was >or=to10 IgG phospholipid (GPL) units. The aCL-positive patients were compared with the aCL-negative patients to characterize their clinical status during follow-up. Historic information reflected risk factor status at the time of the index stroke.

To test the hypothesis that an aCL titer of >or=to10 GPL at the time of an index ischemic stroke is associated with an increased risk of subsequent thrombo-occlusive events or death, the survival distributions of the aCL groups were evaluated and compared using Kaplan-Meier life-table estimates and log-rank tests. A Welch's two-sample t test was used to compare the groups with respect to age, whereas Fisher's exact tests were used to analyze dichotomous variables.

Results.

No follow-up information was available for 29 individuals (12%). Therefore, this analysis reports on the data from 219 patients. Of this group, 17 patients (8%) were positive for IgG aCL.

The aCL-positive group had more men than the aCL-negative group (82% versus 56%) (Table 1). Otherwise, there were no significant differences in race, mean age, or smoking status between the aCL-positive and -negative groups. There was also no difference between the groups for a history of hypertension, medication-dependent diabetes mellitus, angina or myocardial infarction, deep venous thrombosis, arterial embolism, pulmonary embolism, atrial fibrillation, valvular heart disease, and congestive heart failure.

We were unable to demonstrate an increased risk of subsequent death, stroke, myocardial infarction, TIA, pulmonary embolus, or any other thrombo-occlusive event in the aCL positive cohort (Table 2 and Figure 1). Even if the groups were compared regarding any nonfatal event, no differences were statistically significant. The median time to each end point was also analyzed, and there were no significant differences between the groups. The odds ratios were adjusted for age, gender, hypertension, diabetes, smoking, atrial fibrillation, congestive heart failure, and valvular heart disease. Our cohort's risk of recurrent stroke was 11.1 to 12.4% (depending on aCL status) over a mean of approximately 3 years.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Survival distributions of the aPL status groups. Any thrombo-occlusive event or death is considered an end point.

We performed several exploratory subgroup analyses in an effort to detect differences among aCL-positive and -negative patients and our defined end points: no differences between patients with lacunar and nonlacunar stroke subtypes; no differences between individuals with a normal or abnormal transthoracic echocardiogram among the 132 patients (124 aCL negative and 8 aCL positive) who had this test; no association with a reactive RPR, positive ANA, low platelet count, prolonged activated partial thromboplastin time (aPTT), or echocardiogram findings; no difference between patients treated with aspirin or warfarin; and no difference as to stroke pathophysiology according to atherosclerotic, cardiac or transcardiac, other causes, or indeterminate categories. ^[1] Finally, the study group was analyzed comparing individuals with either aCL IgG >or=to 10 or IgM >or=to 10 to aCL-negative individuals. No differences between groups were found in this analysis (data not shown).

Discussion.

We could not demonstrate an increased risk of recurrent thrombo-occlusive events or death in aCL-positive first ischemic stroke patients compared with aCL-negative first ischemic stroke patients, with adjustment for other stroke risk factors. However, there are several limitations to our study. Most importantly, the sample size limited the power of the study to only 13% to detect the observed odds ratio of 1.4. It is desirable in studies of this type to have a power of at least 80%. Thus, with our sample sizes, we would need an odds ratio of at least 4.9. Therefore, our results cannot exclude a clinically important effect of aCL on stroke recurrence risk. We also did not determine aCL status at the time of a thrombo-occlusive end point. Telephone follow-up and lost patients were additional limitations. Finally, treatment may have influenced the clinical course, negating any aCL effect.

We could not demonstrate an excess risk of achieving an end point in the presence of aCL, using a cutoff of IgG aCL positivity of >or=to10 GPL. Sletnes et al. ^[2] found an analogous situation among aCL patients after myocardial infarction. The associated conditions that our patients had may be such robust risk factors for the defined end points that the aCL status added little additional risk. Low titer aCL may represent a transient and nonspecific phenomenon analogous to an acute phase reactant and may have diluted the effect of higher titers of aCL. Manoussakis et al. ^[3] found IgG aCL in 51.6% of healthy elderly individuals. On the other hand, in a series of 81 consecutively studied patients with antiphospholipid antibodies, nine had IgG aCL titers greater than 100 GPL units. Of these nine patients, seven (78%) had a thrombo-occlusive event including TIA or stroke over a mean follow-up of 1.14 years. ^[4] Thus, low aCL titers have a high frequency, especially in the elderly, whereas high aCL titers probably constitute a risk factor for recurrent thrombo-occlusive events.

The risk of recurrent stroke in our cohort is similar to the Oxfordshire Community Stroke Project's 13.2% cumulative risk of recurrent stroke at 1 year and 19.9% at 2 years ^[5] and the finding of Lai et al. ^[6] of a cumulative stroke recurrence rate of 13% after 2 years. Among a select group of patients with antiphospholipid antibodies, the yearly stroke incidence rate was 6.75%; 3 of 23 patients (13%) sustained a recurrent stroke within approximately 1 year. ^[7] Rosove and Brewer ^[8] reviewed 70 patients with antiphospholipid antibodies: "recurrences were in the carotid distribution in nine (13%) patients." Finally, Tohgi et al. ^[9] found a stroke recurrence rate of 8.9% per person-year in their series of aCL-positive patients older than 50 years. Interestingly, 33% of their aCL-positive patients had no other known stroke risk factor, compared with the 16% of their aCL-negative patients who had no other stroke risk factor.

Our findings have some practical implications. It is probably not appropriate to screen every ischemic stroke patient, especially in the geriatric age range, for aCL because it did not confer significant prognostic value. ^[10,11] We advocate assaying for aCL in young or elderly patients with no obvious stroke cause and patients with a history suggestive of the antiphospholipid syndrome (deep venous thrombosis, repeated abortions, livedo reticularis, thrombocytopenia, false positive syphilis serology, positive ANA, or a prolonged aPTT) because the finding of aCL in these settings may alter therapy. ^[12]

We need to better define in which group of patients aCL plays a primary pathogenetic role and whether various molecular forms of aCL have different clinical implications. ^[12] For instance, are there differing co-factor dependencies among different aCLs, with resulting differences in thrombo-occlusive risk? ^[12] What is the importance of the lupus anticoagulant as a risk factor for subsequent stroke? The APASS-WARSS (Warfarin Aspirin Recurrent Stroke Study) collaborative study ^[13] will remedy the major limitations of the current study by providing information on lupus anticoagulants, serial annual antibody titers, and aCL titers at the time of a thrombo-occlusive endpoint with a stroke cohort sample size approximately nine times the size of our cohort.

Appendix.

Biostatistical Group: Steven Kittner, MD, MPH, University of Maryland, Baltimore, MD; Christine Joseph, PhD, Sue Havstad, MS, Lonni Schultz, PhD, Horace Spencer, MS, Division of Biostatistics, Henry Ford Hospital, Detroit, MI; Barbara C. Tilley, PhD, Statistical Coordinating Center, Henry Ford Health Science Center, Detroit, MI. Clinical Centers: Steven R. Levine, MD, Leeza Salowich-Palm, BS, John Carey, MD, and Michael Deegan, MD, Henry Ford Hospital, Detroit, MI; Robin Brey, MD, William Davis, MD, Johanna Kaste, BA, David McGlasson, MA, and Robert McAtee, BS, University of Texas Health Science Center, San Antonio, TX; Edward Feldmann, MD, and Karen Fredberg, MD, Rhode Island Hospital, Providence, RI; Barney Stern, MD, and Cherie Sewell, RN, Sinai Hospital, Baltimore, MD; Percy Karanjia, MD, Daniel Jacobson, MD, and Sandra Lobner, RN, Marshfield Clinic, Marshfield, WI; David Hess, MD, and Robert Adams, MD, Medical College of Georgia, Augusta, GA; Jesse Weinberger, MD, Michelle Dunn, RN, Deborah Horowitz, MD, and Stanley Tuhrim, MD, Mt. Sinai Medical Center, New York, NY; Pierre Fayad, MD, Lawrence M, Brass, MD, and Frank Pavalkis, PA-C, Yale University, New Haven, CT; Te-Long Hwang, MD, and Antonio Trujillano, MD, University of South Carolina, Columbia, SC; Viken Babikian, MD, and Harry Hersey, PA, Boston University, Boston, MA; Michael Sloan, MD, and Dianne Eden-Fetzer, RN, University of Maryland, Baltimore, MD; Bruce Coull, MD; Dennis Briley, MD, and Patricia DeGarmo, RN, Portland, OR; Camilo Gomez, MD, and Melissa Jedicks, RN, St. Louis University Medical Center, St. Louis, MO; Marc Fisher, MD, and Elizabeth Doyle, RN, Medical Center of Central Massachusetts-Memorial, Worcester, MA; and L. Dana DeWitt, MD, Tufts University, Boston, MA.