The opinions and assertions contained herein are private views of the authors and are not to be construed as official or as reflecting the views of the Army, Navy, or the Department of Defense.
The prevalence, onset, and clinical significance of antiphospholipid antibodies prior to diagnosis of systemic lupus erythematosus †
Article first published online: 5 APR 2004
Copyright © 2004 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 50, Issue 4, pages 1226–1232, April 2004
How to Cite
McClain, M. T., Arbuckle, M. R., Heinlen, L. D., Dennis, G. J., Roebuck, J., Rubertone, M. V., Harley, J. B. and James, J. A. (2004), The prevalence, onset, and clinical significance of antiphospholipid antibodies prior to diagnosis of systemic lupus erythematosus . Arthritis & Rheumatism, 50: 1226–1232. doi: 10.1002/art.20120
- Issue published online: 5 APR 2004
- Article first published online: 5 APR 2004
- Manuscript Accepted: 11 DEC 2003
- Manuscript Received: 28 MAR 2003
- NIH. Grant Numbers: AI-31584, AR-45084, AR-45231, AR-42460, AR-48940, AI-24717, RR-15577, RR-14467
- Department of Veterans Affairs
To determine whether antiphospholipid antibodies (aPL) occur before the diagnosis of systemic lupus erythematosus (SLE) and before initial clotting events, and whether their presence early in the disease course influences clinical outcome.
Serum samples obtained from 130 lupus patients before and after SLE diagnosis were screened for IgG and IgM aPL using an anticardiolipin (aCL) enzyme-linked immunosorbent assay. Medical records of all patients were carefully reviewed for data on the time of onset of SLE features meeting clinical criteria and on disease manifestations.
Twenty-four patients (18.5%) were positive for IgG and/or IgM aCL prior to SLE diagnosis. Anticardiolipin antibodies appeared from 7.6 years prior to SLE diagnosis to within the same month as SLE diagnosis, with a mean onset occurring 3.0 years before SLE diagnosis. Additionally, aCL presence early in the disease process seemed to predict a more severe clinical outcome; these patients eventually met an average of 6.1 of the 11 classification criteria for SLE, compared with 4.9 criteria for other patients (P < 0.001). The early aCL-positive population also had more frequent renal disease, central nervous system disease, thrombocytopenia, and clotting events. In this population, aCL preceded initial thrombotic events by a mean of 3.1 years.
Anticardiolipin antibodies in SLE patients tend to precede initial clotting events by several years. Furthermore, the presence of early, prediagnosis aPL seems to herald a more varied, severe clinical course with earlier onset in patients with SLE.
Systemic lupus erythematosus (SLE) is a complex autoimmune disease, the etiology of which is largely unknown. The evolution of lupus after diagnosis has been extensively documented and studied for >50 years. Conversely, only a few case reports of the presence of autoantibodies prior to the establishment of criteria sufficient for an SLE diagnosis have been published, likely due to the dearth of appropriate early serum samples (1, 2). Only recently have the resources been available to examine on a large scale certain immunologic events that occur prior to the development of clinical SLE (3).
Although SLE is a heterogeneous disease, a common feature is the presence of antibodies that target a number of self antigens. These range from antibodies that are very specific for lupus, such as anti-Sm and anti–double-stranded DNA (anti-dsDNA), to antibodies that are also commonly present in a number of other autoimmune diseases, such as anti-Ro and anti–nuclear RNP. Antiphospholipid antibody (aPL) is another autoantibody that can occur in either the presence of or the absence of SLE, or even in healthy individuals. Antiphospholipid antibodies also occur in primary antiphospholipid syndrome (APS), viral infections (most notably human immunodeficiency virus), malignancy, and other autoimmune diseases (4–6).
Antiphospholipid antibodies have been shown to be quite diverse, targeting a number of different antigens. Thus, these antibodies are directed not only against negatively charged phospholipids, but also against portions of lipid–protein complexes. The lupus anticoagulant and anticardiolipin antibodies (aCL) have traditionally been shown to be associated with thrombotic symptoms of APS and SLE (7–15). However, aPL have also been shown to bind antigens such as prothrombin, annexin V, and β2-glycoprotein I (β2GPI). Antibodies reacting with these antigens may be directly involved in the pathogenesis of aPL-associated symptoms (16–21).
Clinical symptoms associated with aPL include, but are not limited to, arterial and venous thrombosis, recurrent fetal loss, thrombocytopenia, hemolytic anemia, and aPL-associated nephropathy (7–10, 22). The presence of aPL has been shown to variably increase risk for the development of these clinical symptoms from 2-fold to 10-fold, consistent with the notion that they play an important role in disease pathogenesis (7–10, 22). They have further been demonstrated to often precede these clinical events, although by an unspecified amount of time (12–15). The exact mechanisms of this complex procoagulant pathogenesis remain partially a mystery, although it has been shown that some aCL can bind directly to platelet surfaces and cause thromboagglutination in vitro (18). Furthermore, some aPL may promote prothrombotic events through effects on the vascular endothelium (20, 23).
Reports of the prevalence of aPL in SLE are myriad and have varied widely, depending on the antigen source and method used. In fact, studies have demonstrated that anywhere from 5% to 70% of lupus patients produce some type of aPL (9, 10). However, very little is known about the actual prevalence and onset of aPL in patients prior to SLE diagnosis. In this study, we analyzed serum samples collected before the onset of SLE that fulfilled the American College of Rheumatology (ACR) criteria for SLE classification (24, 25). We evaluated the samples for aCL and determined the relationship to disease onset and course.
PATIENTS AND METHODS
Computer databases of military hospital records were used to identify 130 individuals who had been diagnosed as having SLE (24, 25) according to the ACR criteria during active military duty, had at least 1 retrievable serum sample prior to disease diagnosis, and had available medical records. For each case, 1 normal control was randomly selected; controls were matched to the SLE patients for sex, ethnicity, age (within 1 year), length of military service, and availability of stored serum samples. Demographic characteristics of the studied patient population are shown in Table 1. The mean ± SD age at diagnosis in the SLE patients was 30.4 ± 6.8 years (range 18.5–46.9 years).
|aCL+ (n = 24)||aCL− (n = 106)||P|
|Male (n = 46)||10 (41.7)||34 (33.3)|
|Female (n = 84)||14 (58.3)||68 (66.7)|
|African American (n = 80)||11 (45.8)||68 (66.7)||0.01 (χ2 = 7.5, OR = 3.5)‡|
|White (n = 34)||12 (50.0)||21 (20.5)|
|Other (n = 16)||1 (4.2)||13 (12.7)|
|Age at SLE diagnosis, mean years||27.1||31.6||<0.005|
A total of 633 serum samples were retrieved from the 130 SLE patients identified, with a mean ± SD of 4.9 ± 2.5 serum samples available per patient (median 5.0, range 1–12). There were 4.7 samples per case in aPL-positive patients and 5.0 samples per case in aPL-negative SLE patients (median 5.0 in both subgroups). The earliest serum sample available for each patient was one obtained a mean ± SD of 4.4 ± 2.5 years prior to diagnosis (median 4.8 years, range 9.4 years prior to diagnosis to within the same month as diagnosis). Again, availability of early samples was not significantly different among aPL-positive and aPL-negative patients (average time between first serum sample and SLE diagnosis 4.5 years and 4.3 years, respectively). While all patients were required to have at least 1 prediagnosis sample available, 54% also had available serum samples that had been obtained after diagnosis. For 3 patients, samples from up to 6 years after diagnosis were available.
Clinical chart review.
Information regarding cumulative clinical and laboratory features was obtained by retrospective chart review. Clinical and laboratory variables were evaluated, including each feature of the ACR criteria for SLE classification (24, 25). Medical records were also reviewed for evidence of additional medical features frequently associated with APS, such as the presence of stroke, venous or arterial thrombosis, and spontaneous abortions. The initial observation time for each feature was recorded.
Enzyme-linked immunosorbent assays were used to evaluate patient and control sera for aCL. Cardiolipin in ethanol at a concentration of 0.25 μg/well was coated onto 96-well plates using a previously published protocol (26). Seven standard dilutions were used on each plate to determine a standard curve. Based on calibration standards, IgG or IgM antibody concentrations (an average between 2 duplicates) were determined in units. Samples that were scored as moderate (20–80 units) or high-positive (> 80 units) on a preexisting scale (26) were considered to be positive for purposes of analysis.
The chi-square test was used to analyze differences between categorical variables. Earliest time positive was determined as the first serum sample for which a positive antibody result was obtained. All P values were 2-tailed (except where otherwise indicated), and P values less than 0.05 were considered significant.
Relationship of race and sex to aPL prevalence prior to SLE diagnosis.
Moderate or high levels of aPL (IgG and/or IgM aCL) were found prior to SLE diagnosis in 24 of 130 patients (18.5%) (Figure 1). These individuals first produced aCL from as early as 7.6 years prior to SLE diagnosis to within a month of SLE diagnosis (in 1 case only; all others >6 months prior to diagnosis), with a mean ± SD aCL onset time of 3.0 ± 0.31 years (median 3.2 years) prior to SLE diagnosis. Four additional patients (3%) developed moderate or high levels of aCL only after SLE diagnosis. These individuals first exhibited aCL 0.85 ± 0.18 years after fulfilling the diagnostic criteria for SLE (median 0.75 years). IgM aCL and IgG aCL were found at low levels in 1 (0.8%) and 2 (1.5%) of the 130 controls, respectively.
The prevalence of autoantibodies prior to SLE diagnosis was compared between African American and white patients (Table 1). In this prediagnosis patient population, an increased prevalence of aCL (IgG and/or IgM) was found in white patients (12 of 34 [35%], versus 11 of 80 African American patients [14%]) (P < 0.01). The difference was even more dramatic for IgG aCL alone (white patients 10 of 34 [29%]; African American patients 3 of 80 [4%]) (P < 0.00005).
An increased prevalence of IgG aCL prior to diagnosis was also found in men (22%) compared with women (6%) (P < 0.02). The difference in prevalence of IgG aCL between men and women was detected both in African American patients (8.7% versus 0%) and in white patients (43% versus 20%). In African American patients, the number of individuals positive for IgG aCL was too small for statistical evaluation (3 men, 2 women). While significant differences were seen in the prevalence of aCL among white compared with African American patients, there was no difference in the time at which these antibodies first appeared in relation to SLE diagnosis.
Earlier onset and more complex course of SLE in patients with aCL prior to diagnosis.
Review of the medical records of the 24 patients who developed aCL prior to SLE diagnosis revealed a number of differences between the aPL group and the other 106 lupus patients in the study. The 24 patients who developed aCL prior to SLE diagnosis appeared to eventually have a more severe, complex disease course. On average, these 24 patients met 6.1 of the 11 ACR criteria, compared with only 4.9 in the aCL-negative group (P < 0.001), without inclusion of aPL as a criterion (Table 2) (median 6.0 and 5.0 criteria, respectively). Specifically, patients with aCL had a number of clinical symptoms in greater frequency and/or with earlier onset. Thrombocytopenia and hemolytic anemia were more common in the aCL group (thrombocytopenia was >4 times more common) and occurred much earlier in the disease course (>1.4 years earlier for both criteria) (Table 3). Lupus patients with aCL prior to SLE diagnosis also had an earlier onset of malar rash, photosensitivity, serositis, neurologic symptoms, and nephropathy (the latter of which was more than twice as common in the aPL group than in the other SLE patients) (Tables 2 and 3). Furthermore, these 24 patients had an earlier disease onset overall, with an average age at diagnosis of 27.1 years, compared with 31.6 years in the aCL-negative lupus patients (P < 0.005).
|aCL+ (n = 24)||aCL− (n = 106)||P|
|Malar rash||10 (42)||28 (26)||<0.005|
|Discoid lesions||2 (8)||21 (20)||<0.01|
|Photosensitivity||8 (33)||39 (37)||<0.01|
|Oral ulcers||11 (46)||39 (37)||NS|
|Arthritis||21 (88)||89 (84)||NS|
|Serositis||13 (54)||49 (46)||NS|
|Renal disorder||17 (74)||40 (36)||<0.001|
|Neurologic disorder||6 (25)||6 (6)||<0.001|
|Hemolytic anemia||7 (29)||18 (17)||<0.01|
|Thrombocytopenia||9 (37)||10 (9)||<0.001|
|Anti-dsDNA||19 (79)||59 (55)||<0.001|
|Anti-Sm||12 (50)||40 (38)||<0.05|
|Antinuclear antibody||24 (100)||103 (97)||NS|
|Mean no. of criteria met†||6.1||4.9||<0.001|
|Mean age at SLE diagnosis‡||27.1||31.6||<0.005|
|aCL+ (n = 24)†||aCL− (n = 106)|
Presence of IgG aCL prior to the onset of thrombotic events in SLE.
Of the 24 patients who developed aCL prior to SLE diagnosis, 12 (50%) had sufficient clinical and serologic data available to meet the Sapporo classification criteria for APS (27). The associated clinical thrombotic events (venous or arterial thrombosis or recurrent spontaneous abortions) occurred prior to SLE diagnosis in all but 1 of these patients, an average of 0.7 years prior to diagnosis. Of the 12 patients, 7 had available serum samples stored prior to the time the thrombosis criterion for APS was met. In each of these 7 patients, IgG aCL were present prior to (mean 3.1 years [range 0.9–4.9 years] before) the development of the initial clinical thrombotic event. In the remaining 5 patients, IgG aCL were present in the initial serum sample available (obtained at the same time as the thrombotic event).
Most of the patients with documented thrombotic events (10 of 12 [83%]) had high-titer aCL, while most of the patients with no documented thrombosis had moderate-titer aCL (10 of 12 [83%]), creating a significant difference in mean titer between the 2 groups (80.2 versus 50.2 units; P < 0.05). None of the 3 patients with only IgM aCL developed clots during the time of observation. In addition, only 2 aCL-negative SLE patients (1.5%) had a recorded thrombotic event. No significant differences were noted between the aCL-positive group with APS and the non-APS patients, with regard to the time of onset of aPL (mean 3.14 years and 3.03 years, respectively, prior to SLE diagnosis), eventual number of SLE criteria met (6.31 versus 5.91; P = 0.21), or age at diagnosis of SLE (26.8 years versus 27.5 years; P = 0.31).
Correlation between early aCL and early onset of other autoantibodies.
The 24 patients who developed aCL prior to SLE diagnosis developed additional specific autoantibodies earlier than was common among the aCL-negative population (Table 3). Perhaps the greatest difference was seen with anti-Sm antibodies, which developed at a similar prevalence among aCL-positive and aCL-negative subjects (50% and 38%, respectively) but occurred almost 2 years earlier in the aCL group (2.98 years prior to diagnosis versus 1.27 years prior; P < 0.001). Differences were also seen with anti-dsDNA antibodies, which were more common in the aCL group compared with the other lupus patients (79% versus 55%). Anti-dsDNA antibodies in the aCL group began to appear, on average, nearly a full year before they did in the aCL-negative group (2.83 years versus 2.03 years prior to diagnosis; P < 0.05). Not all autoantibodies displayed this difference, however: antinuclear antibodies appeared in the 2 groups at almost the same time in relation to SLE diagnosis (3.40 and 3.32 years, respectively, prediagnosis).
IgG versus IgM aCL.
When the aCL data were stratified by isotype, most of the detected aCL effect was found to be due to the presence of IgG aCL rather than IgM aCL. The increased rates of hemolytic anemia, thrombocytopenia, and autoantibodies were even more pronounced in the subgroup of 21 patients with IgG aCL (of whom 8 additionally had IgM aCL and 13 did not). The average number of SLE criteria met by the IgG aCL–positive group increased to 6.5 (P < 0.001), and the average age at lupus onset decreased almost a full year, to 26.2 years (P < 0.001). This stratification revealed an even earlier time of onset of several individual criteria, especially malar rash and nephritis, in the IgG aCL group. The IgG aCL–positive individuals exhibited diminished rates or earlier onset time for 2 criteria (photosensitivity and oral ulcers), but none of the differences specifically noted above as being significant in the total aCL group were decreased in the IgG-positive subset.
A number of studies have previously demonstrated that aPL can increase the risk of thrombotic events, suggesting an important role for these autoantibodies in the pathogenesis of APS (28, 29). However, their role in the development of lupus has largely remained unknown. Analysis of prediagnosis serum samples has revealed for the first time that aPL are commonly present for years prior to the eventual diagnosis of SLE. Furthermore, these aPL were present for years prior to the development of venous and/or arterial thrombosis in this SLE population.
Antiphospholipid antibodies, measured by IgG and/or IgM antibodies against cardiolipin, were detected in 18.5% of the SLE patients prior to diagnosis. Studies investigating the prevalence of aPL in SLE clinic populations, using similar methodology, have shown prevalence rates of between 5% and 70%, compared with up to 4% in healthy controls (9, 10, 30, 31). Interestingly, in the present study, we found a higher frequency of aPL among white patients (35%) compared with African American patients (14%) in prediagnosis serum samples, although this finding may be influenced by the unusually high proportion of male patients in our analysis. In the evaluation of all samples from patients with sera available after diagnosis, these differences appeared to be restricted to antibodies of the IgG isotype.
In contrast to our results, several previous studies evaluating the prevalence of aPL in SLE patients have not demonstrated any significant differences among ethnic groups (32, 33). In fact, in a study in which IgG and IgM aPL and/or the lupus anticoagulant were considered as a whole, an increased (but not statistically significantly increased) prevalence of aPL was found among African Americans (10.9%) compared with European Americans (4.8%) (33). Although the significance of our finding that IgG aPL was specifically responsible for the aPL results is still unknown, antibody isotype may be one of the features determining the development of symptoms commonly associated with APS (6). Indeed, significant associations between elevated IgG aCL and spontaneous abortion, thrombocytopenia, and livedo reticularis, which are not necessarily found with IgM aCL, have been demonstrated (31).
Antiphospholipid antibodies in the APS have also been previously associated with several common clinical symptoms of SLE. Lupus patients who develop aPL have an increased risk of developing thrombocytopenia (range 11–40%, versus ∼4–17% in aPL-negative SLE), hemolytic anemia (range 5–16% versus 2–8%), and symptoms traditionally thought of as being associated with APS rather than SLE, such as arterial or venous thrombosis and recurrent fetal loss (6–10, 28–31). In our study, thrombocytopenia and hemolytic anemia were found to be significantly more common among aPL-positive patients, as was the risk of thrombotic events usually associated with APS.
Interestingly, aPL-positive SLE patients evaluated over the time period of the study met a significantly higher number of ACR SLE criteria (mean 6.1) than the other lupus patients analyzed (4.9; P < 0.001). Furthermore, when the patients who were positive for IgG aCL were analyzed separately, this difference became even more pronounced (6.5 versus 4.8 criteria; P < 0.001). This distinctive difference, which was not dependent on race and was calculated without inclusion of aPL as a criterion, seems to indicate that the early presence of aCL is a harbinger of a more severe, complex SLE disease course.
The use of a cohort for this study that included multiple serum samples collected years prior to the onset of SLE enables us to make unique observations about the origins and progression of aPL in lupus patients. Not only did aPL-positive patients develop certain clinical events more commonly than the other lupus patients, they developed them much earlier in the course of their disease and hence, on average, at an earlier age. For thrombocytopenia and hemolytic anemia, onset occurred >1.4 years earlier, on average, in aPL-positive patients than in the remainder of SLE patients studied. Other symptoms that occurred ∼1 year earlier in the aPL group included serositis, malar rash, and nephropathy. Moreover, the overall age at disease onset, defined as the age at which the individual first met 4 of the diagnostic criteria for SLE, was >4 years younger in the patients identified as having had aPL prior to diagnosis (27.1 years versus 31.6 years). This association seems to be specific to aPL; younger-onset disease has not been demonstrated in individuals with other prediagnosis autoantibodies, even those antibodies that commonly occur earlier than aPL, such as anti-Ro (3).
Furthermore, aPL preceded initial clotting events by an average of 3.1 years in this SLE patient population. While several studies have identified that aPL often precede aPL-associated events (12–14), this is the first to measure the time between antibody onset and thrombotic events. The number of patients for whom sufficient data were available to make this determination was small; however, the fact that these antibodies preceded thrombotic events in every patient studied supports the idea that these antibodies may play a causative role in the development of a prothrombotic environment, rather than arising as a result of thrombotic events (12).
As further evidence of the relationship between aPL and SLE onset and course, the 24 patients who developed aPL prior to lupus also developed several lupus-specific autoantibodies earlier than has previously been reported (3) (Table 3). Anti-dsDNA appeared in these patients almost a year earlier, while anti-Sm antibodies arose almost 2 years earlier, than in the other lupus patients. The fact that these temporal differences were not seen with all autoantibodies reinforces the contention that aCL tend to define a distinct subset of SLE. Thus, the presence of aPL in individuals who are also positive for anti-dsDNA and/or anti-Sm antibodies may in itself suggest that early treatment for SLE should be initiated, even in the absence of positivity for 4 of the SLE diagnostic criteria. Whether early examination of other targets of aPL, such as β2GPI, will augment this type of analysis remains to be seen.
Such examination of early autoantibody profiles and their relationships to clinical outcome can provide insight into disease processes that has heretofore been largely unavailable. This study demonstrates that aCL are present both years before the diagnosis of SLE and years before the development of clinical features of APS. Nearly all identified clotting events occurred in SLE patients with IgG aCL, and these clotting events usually occurred before SLE diagnosis. Within this lupus population, early-onset aCL seemed to signal a more complex, severe disease course developing at a younger age.
The authors would like to thank Tara Bruner, Suzanne Lapolla, and James Longton for their technical assistance with this work as well as Kimmie Kohlhase and all of the referring and cooperating military rheumatologists and rheumatology clinics for their assistance in case identification. We would also like to thank Drs. Christie Burgin and Don Parker of the Oklahoma General Clinical Research Center for their aid in statistical analysis.
- 25Diagnostic and Therapeutic Criteria Committee of the American College of Rheumatology. Updating the American College of Rheumatology revised criteria for the classification of systemic lupus erythematosus [letter]. Arthritis Rheum 1997; 40: 1725., for the