To determine if whole blood viscosity (WBV), a rheologic variable contributing to risk of myocardial infarction and stroke in the general population, is elevated in patients with systemic lupus erythematosus (SLE), particularly SLE patients with a history of thrombotic or atherothrombotic events. Because the high rates of arterial and venous thrombosis in lupus cannot be explained by traditional risk factors, elevated WBV may be an easily measurable nontraditional risk factor to identify SLE patients at high risk for thrombotic events.
Sixty SLE patients (30 with a history of a thrombotic event) and 20 matched controls were recruited into the study. The thrombosis group was further subdivided into an arterial thrombosis group (n = 17). WBV values were determined at 9 different shear rates (1, 2, 5, 10, 50, 100, 150, 300, and 1,000 seconds−1). WBV was then compared between groups by repeated-measures analysis of variance.
SLE patients with a history of arterial events had significantly elevated WBV relative to either controls (P = 0.022) or SLE patients without arterial events (P = 0.014). WBV in the total SLE group did not differ from controls. Differences in WBV were most prominent at lower shear rates (1, 2, 5, 10, 50, and 100 seconds−1). Anticoagulation, prednisone dose, and antiphospholipid antibodies did not significantly impact WBV.
Our study demonstrated that WBV is selectively elevated in patients with SLE with a history of arterial events. Although this association is striking, longitudinal studies are needed to assess the positive predictive value of WBV for atherothrombotic events in SLE.
Systemic lupus erythematosus (SLE) is a chronic inflammatory condition associated with an increased arterial and venous thrombotic risk. The estimated 5-year risk of arterial thrombosis ranges from 5.1% to 8.5%, while 5-year venous thrombosis risk ranges from 3.7% to 10.3% (1). Consequently, patients with SLE have a significantly increased risk of premature cerebrovascular accident (CVA) and myocardial infarction (MI) compared with age-matched controls (2–5). Although most studies document a role of traditional cardiovascular risk factors, the increased incidence of cardiovascular disease in SLE cannot be explained by traditional risk factors alone, suggesting that there are additional factors contributing to atherothrombotic disease in SLE (3, 4, 6–8). With modern appreciation of the inflammatory nature of atherogenesis, it is increasingly likely that lupus directly contributes to atherothrombotic diseases by an immune-mediated inflammatory process. However, the specific mechanisms by which lupus-related inflammation may promote atherosclerosis have yet to be delineated.
Whole blood viscosity (WBV) is a major rheologic variable associated with risk of atherothrombotic disease within the general population. Numerous studies have documented increased rates of stroke and MI in individuals with elevated blood viscosity (9–13). Because multiple traditional cardiovascular risk factors such as age, male sex, obesity, hypertension, diabetes, and smoking are all associated with increased viscosity, the correlation between viscosity and atherothrombotic events decreases after adjustment for these risk factors (9, 14). However, there may be an additional, independent increase in risk for such events associated with high blood viscosity itself (12, 13, 15).
WBV is predominantly determined by hematocrit, plasma viscosity (primarily driven by acute-phase reactants such as fibrinogen), and red blood cell aggregation and deformability (16). In contrast to serum or plasma viscosity, measurement of WBV does not involve separation of the constituents of blood, and therefore most closely resembles the natural rheologic conditions in the bloodstream. Whole blood is a non-Newtonian fluid, which exhibits higher viscosity at low shear rates relative to high shear rates. It is thought that at low shear rates, rouleaux formation and aggregation by erythrocytes are a major determinant of WBV in normal human blood. At high shear rates, normal erythrocytes deform and cell–cell aggregates are broken apart. Excessive rigidity of erythrocytes, such as seen in hypertension and diabetes (17–19), results in increased WBV at high shear rates, due to the failure of erythrocytes to deform in high velocity conditions (16, 20). In contrast, at low shear rates, excessive erythrocyte rigidity may lower WBV due to impairment in the formation of erythrocyte aggregates. Thus, WBV may vary in specific ranges of shear rates based on the underlying pathologic condition.
The actual measurement of WBV has been technically difficult, introducing potential artifact by the addition of anticoagulants or by indirectly estimating WBV by measurement of hematocrit-adjusted plasma viscosity, and researchers have not been able to measure WBV at a wide range of shear rates. Recent technological advances now allow the simultaneous direct measurement of WBV at a wide range of shear rates with or without anticoagulants (21, 22) through the use of a scanning capillary viscometer, making WBV measurement more practical in the clinical setting.
There has been little research on blood viscosity in SLE. Yet, a relationship seems likely because the products of chronic inflammation are some of the major determinants of WBV. Elevations in interleukin-6, a cytokine characteristic of SLE (23–25), are associated with higher blood viscosity in the general population and are thought to drive the hepatic production of acute-phase reactants (9, 26, 27). Several small studies have found that patients with SLE have elevated plasma or blood viscosity compared with healthy controls (28–30). We hypothesized that WBV would be higher in patients with SLE compared with matched controls, and that WBV may vary between patients with SLE with and without a history of thrombotic events.
PATIENTS AND METHODS
Between September 2004 and July 2005, we studied 30 patients with SLE without a history of a thrombotic event (SLE group), 30 patients with SLE with a history of a thrombotic event (SLE-T group), and 20 controls who were matched for age, sex, body mass index (BMI), race, and tobacco usage. All patients were >18 years of age, and all SLE patients met the American College of Rheumatology criteria for the classification of SLE (31). A thrombotic event was defined as an MI, CVA, transient ischemic attack (TIA), deep vein thrombosis (DVT), pulmonary embolism (PE), other significant thrombotic event, and pregnancy complication attributed to thrombotic event per antiphospholipid antibody syndrome criteria (32). The patients were recruited during routine visits to the adult rheumatology clinic at the University of Chicago and were not paid for their participation. Controls were recruited from rheumatology clinic patients without inflammatory diseases or from healthy university personnel, and were frequency-matched to the SLE group for BMI, race, age, and tobacco use. The study was approved by the Institutional Review Board at the University of Chicago and all participants gave written informed consent for the study.
Information on patients with SLE was obtained through a structured interview, a review of the patient's medical records, laboratory tests, and WBV measurements. Information on controls was obtained through structured interview and WBV measurements. Data on age, race, sex, BMI, significant medical history including history of thrombotic events, general medical diagnoses (hypertension, diabetes, hyperlipidemia), smoking history, and medication history were collected for all SLE patients. Prednisone dosage at the time of WBV measurement was dichotomized as none, ≤10 mg/day, or >10 mg/day. For the SLE-T patients, antiphospholipid status was determined, with present defined as a positive lupus anticoagulant (dilute partial thromboplastin time or Russell's viper venom time with positive confirmatory test), anticardiolipin, or β2-glycoprotein I antibody by enzyme-linked immunosorbent assay (Inova Diagnostics, San Diego, CA). A Systemic Lupus Erythematosus Disease Activity Index (SLEDAI) score, a summary measure of current SLE activity, was obtained for all SLE patients at a recent clinic visit (33).
Nine simultaneous WBV measurements were obtained for each participant at shear rates of 1, 2, 5, 10, 50, 100, 150, 300, and 1,000 seconds−1 using a scanning capillary viscometer (Rheolog; Rheologics, Exton, PA), which was developed in 2001 and has a Food and Drug Administration Investigational Device Exemption status. The Rheolog has been validated to accurately determine WBV (mPa × second) at incremental shear rates from 1 to 1,000 seconds−1 (21, 22). It comprises a central scanning unit and U-shaped disposable cartridge used to hold the unadulterated blood sample during the test procedure. For this measurement, ∼4–6 cc of blood were collected with a standard venipuncture. Samples were either run immediately or kept warm in a hot water bath at 37°C for up to 30 minutes prior to running the sample through the viscometer. The raw data were then presented graphically and numerically on an attached computer.
The size of the study was determined by a power calculation extrapolated from 2 sources, because there have not been measurements of WBV in patients with SLE with this rheologic technique. We used previous mean blood viscosity and standard deviations, which were obtained from an ongoing cardiovascular study, using the Rheolog at a shear rate of 1,000 seconds−1 (Holper E: personal communication) along with previous data indicating a mean difference of 13% in plasma viscosity of SLE patients and controls (30). Using 20 controls and 60 SLE patients, we achieved ∼90% power to detect a true difference of this magnitude in WBV between patients with SLE and healthy controls.
Statistical analysis was performed using Stata software, version 8.0 (StataCorp, College Station, TX). Descriptive statistics, chi-square tests, t-tests, and analyses of variance (ANOVAs) were used to assess the group characteristics and matching success between SLE, SLE-T, and the control group. WBV values were log transformed to make standard deviations uniform across a variety of shear rates. Because 9 separate measures of viscosity, corresponding to 9 shear rates, were obtained for each patient's sample, repeated-measures ANOVAs and a Greenhouse-Geisser adjustment were used throughout to compare WBV among the SLE, SLE-T, and control groups. The interaction between shear rate and group was also evaluated to determine whether there was evidence for group differences at particular shear rates, and the data were also reassessed using one-way ANOVA at each shear rate. WBV was first compared in all lupus patients (SLE plus SLE-T) versus control and SLE-T versus control. The SLE-T group was then subdivided into arterial (SLE-ART) versus nonarterial (SLE-NA) thrombotic events, because arterial events have been most strongly associated with WBV in the general population. Pregnancy losses (n = 2) were not classified as arterial due to the complexity of mechanisms active in antiphospholipid-mediated fetal death. WBV in SLE-ART was then compared again by repeated-measures ANOVA with the control group, SLE without arterial thromboses (SLE plus SLE-NA), and SLE-NA alone. In addition, a variety of variables thought to affect WBV (age, BMI, hypertension, tobacco, prednisone dose, SLEDAI score, anticoagulation status, antiphospholipid status) were investigated in the entire study group or appropriate subgroups by repeated-measures analysis of covariance (ANCOVA) to assess their effect on WBV and rule out potential confounding influences.
The demographic characteristics of the 30 patients with SLE without a history of thrombosis (SLE), 30 patients with SLE with a history of a thrombotic event (SLE-T), and 20 controls are presented in Table 1. Patients and controls were successfully matched for age, sex, BMI, tobacco usage, and race. Tobacco usage was highest in the control group at 20% and was somewhat lower in the other groups (17% in SLE, 7% in SLE-T; P = not significant [NS]).
Values are the percentage unless otherwise indicated. SLE = systemic lupus erythematosus patients without a history of thrombosis; SLE-T = SLE patients with a history of thrombotic events; BMI = body mass index.
Because the comorbidities hypertension, diabetes, and hyperlipidemia have been known to affect WBV, we attempted to compare those variables as well. There were too few patients in either group to compare hyperlipidemia or diabetes (data not shown). Twenty-five percent of the control group was hypertensive, compared with 33% in both the SLE group and SLE-T group (P = NS).
There were a total of 35 thrombotic events within the 30 patients in the SLE-T group. Antiphospholipid antibodies were present in 13 of the 30 SLE-T patients. Of the total events, 18 were arterial events, with 1 patient having 2 arterial events (CVA/MI). Cerebrovascular events (CVA/TIA) were the most common arterial event (n = 12), followed by MI (n = 4) and peripheral vascular disease (n = 2). PE/DVT had occurred in 14 patients, portal vein thrombosis in 1, and spontaneous late abortions fulfilling the Sapporo criteria for antiphospholipid syndrome in 2. Of the 30 patients in the SLE-T group, 13 received anticoagulation therapy with warfarin, 2 with heparin or heparin products, and 11 received aspirin or clopidogrel.
The SLE and SLE-T groups differed in mean SLEDAI score. The SLE patients without a history of thrombosis had a higher mean SLEDAI score compared with the SLE patients with history of thrombosis (mean SLEDAI score 6.7 versus 3.6; P = 0.0090 by Wilcoxon's rank sum test). The glucocorticoid dose between the 2 SLE groups was comparable, with 73% in the SLE group and 66% in the SLE-T group receiving ≤10 mg/day of prednisone or its equivalent. In the SLE group, 33% were receiving no glucocorticoids, whereas 20% of the thrombosis group were receiving no glucocorticoids (P = NS).
Not surprisingly, the SLE-ART group was somewhat older than the overall SLE group (mean age 46.7 years versus 38.1 years; P = 0.021) (Table 2). However, other variables associated with atherothrombotic risk such as hypertension, obesity, and tobacco did not vary between the SLE and SLE-ART groups.
Table 2. Characteristics of systemic lupus erythematosus patients with arterial events (SLE-ART)*
Repeated-measures ANOVA did not reveal a statistically significant overall effect of group (SLE, SLE-T, and control) on WBV at the 9 measured shear rates, nor any significant interaction between group and shear rate. The total SLE group (SLE plus SLE-T) had a significantly higher WBV compared with controls (mean ± SD 3.23 ± 0.35 versus 3.05 ± 0.31; P = 0.038) at a shear rate of 1,000 seconds−1, but because the overall repeated-measures ANOVA was not significant, this may represent a spurious finding.
The subgroup of SLE-T that had arterial thrombotic events (SLE-ART) appeared to have the highest WBV (Table 3). SLE-ART had significantly higher WBV compared with controls, with an overall group-by-shear-rate interaction significant at P = 0.022 (Figure 1). WBV was also significantly higher at 6 individual shear rates (1, 2, 5, 10, 50, and 100 seconds−1; all P < 0.05). WBV in the SLE-ART group was also elevated relative to the other SLE patients, with an overall group-by-shear-rate interaction significant at P = 0.014 (Figure 1). WBV in the SLE-ART group was significantly elevated relative to other SLE patients at the individual low shear rates 1, 2, 5, and 10 seconds−1 (all P < 0.05). SLE-ART patients differed from SLE-NA patients at all shear rates except 1,000 seconds−1, with an overall interaction term significant at P = 0.029 (data not shown).
Table 3. Whole blood viscosity (mPa · seconds) at various shear rates in systemic lupus erythematosus patients with arterial events (SLE-ART) versus other SLE patients and healthy controls*
Other specific variables previously associated with WBV were examined by ANCOVA. Despite the increased mean age of the SLE-ART group, which had the highest WBV, the age variable failed to be significantly associated with WBV by ANCOVA (all P > 0.350) and therefore did not appear to be a confounding factor. BMI, hypertension, and tobacco use also did not affect WBV. In addition, within the total SLE group (n = 60), prednisone dose, BMI, tobacco use, and SLEDAI score had no significant group effects on WBV. Among the SLE-T patients, there was no significant difference in WBV values between those who received anticoagulation therapy with warfarin or heparin versus those who did not receive anticoagulation therapy, or between groups with or without antiphospholipid antibodies (data not shown).
Despite a high level of scientific interest in blood rheology and cardiovascular disease in the general population, little work has focused on the role of rheologic variables in lupus, a disease characterized by premature atherothrombotic events (6, 7, 34, 35). Because traditional risk factors do not fully explain the high rate of atherosclerotic disease in patients with SLE, nontraditional risk factors warrant greater scrutiny.
There were several interesting findings in our study. Overall, WBV across 9 shear rates did not seem to be increased simply by the presence of SLE compared with a healthy age- and sex-matched control group in this cross-sectional study. In contrast, the subset of SLE patients with arterial thrombotic events did have significantly elevated WBV. This is in concordance with population studies, which suggest that high blood viscosity is associated with arterial events such as cardiovascular and cerebrovascular disease (11–13). It is possible that WBV is a nontraditional risk factor linked to atherothrombotic disease in SLE. Although some studies have linked venous thrombosis to elevated WBV (36), this association was not found in our study group. However, our study power was relatively low on subgroup analysis, and it is possible that modest differences in WBV were missed in our venous thrombosis group due to the scale of our study.
However, our finding of no significant elevation of WBV in the total SLE group conflicts with the sparse published literature in SLE (28–30). It is possible that there is a minor increase in WBV in patients with SLE, as measured by this rheologic technique, detectable only in a larger study. Other possible factors contributing to the discrepancy between our findings and other SLE viscosity studies include differences in technique of viscosity measurement, publication bias, and our decision to not correct for hematocrit. Differences in study population may also have contributed to the disparity between our findings and previous SLE literature. Confounding by associated parameters such as age, BMI, hypertension, or cigarette smoking seems unlikely because these parameters did not exert significant group effects on WBV and did not vary markedly between groups. We also did not measure other factors that may contribute to WBV, such as fibrinogen, C-reactive protein, or the presence of antibodies against erythrocytes, as reflected in a positive direct antiglobulin test. Further studies using the new technique of scanning capillary viscometry will be needed to clarify these issues.
We did not find an association between SLEDAI score and WBV. Reid and de Ceulaer (29) and Rosenson et al (30) also did not find an association between high SLEDAI score and high blood viscosity, whereas Rosenson et al found that high blood viscosity correlated with high damage index scores in SLE. Damage index scores were not performed in the current study.
Elevation in WBV among SLE patients with arterial events was most prominent at low shear rates, an environment in which erythrocyte aggregation and rouleaux formation are thought to be major determinants of WBV (16, 20). Ernst et al (28) documented increased erythrocyte aggregates and normal erythrocyte deformability in patients with SLE relative to controls, which is consonant with our findings of elevated WBV at low shear rates in patients with SLE.
Although it might be tempting to believe that patients who were receiving warfarin might have a lower viscosity, WBV did not vary between SLE patients receiving or not receiving anticoagulation. Our sample size was small in this regard, however, because only 10 SLE patients were receiving anticoagulant therapy. Furthermore, because antiphospholipid syndrome plays a significant role in thrombosis in patients with SLE by not yet fully elucidated mechanisms, it is notable that the presence of these antibodies did not have obvious effects on blood rheology among the thrombotic SLE subgroup.
We propose that high WBV levels may be a nontraditional risk factor for arterial events in patients with SLE, as has been reported prospectively in the general population. Because this is a cross-sectional study, we cannot assess the positive predictor value of elevated WBV. Prospective studies to determine the predictive value of high WBV are indicated to determine the utility of WBV in identifying patients with lupus at high risk for cerebrovascular or cardiovascular events.
Several studies indicate that statin therapy may lower blood viscosity, and that this may be an important mode of action in the prevention of cardiovascular events (37, 38). Viscosity might also be lowered by dietary changes (39) and pentoxifylline (40–42). A new tool for identifying patients with SLE at high risk for catastrophic atherothrombotic events could improve patient care and potentially open new therapeutic avenues for cardiovascular disease prevention in SLE.
Dr. Utset had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study design. Booth, Utset.
Acquisition of data. Booth, Chohan, Curran, Schmitz.
Analysis and interpretation of data. Booth, Karrison, Utset.
We thank Dr. Elizabeth Holper, University of Chicago, for sharing her experience with the Rheolog, and Dr. Susan Manzi, University of Pittsburgh, for her review of this manuscript. Additionally, we thank Rheologics for their assistance in this project.