A meta-analysis of the risk of venous thromboembolism in inflammatory rheumatic diseases

Venous thromboembolism (VTE) is a vascular phenomenon that includes clinical entities such as deep vein thrombosis (DVT) and pulmonary embolism (PE). These venous coagulopathies usually occur in the setting of Virchow's triad, which describes conditions in which thromboses develop as a consequence of stasis, endothelial injury and innate hypercoagulability [1],[2]. It is increasingly becoming recognized that active inflammation is an important process that increases coagulability and leads to thrombosis [3]. Active inflammation is a prothrombotic state characterized by upregulation of tumor necrosis factor α (TNF-α) and activation of endothelial cells. It is thought that upregulation of TNF-α increases tissue factor in the serum, which is a natural procoagulant, while downregulating protein C, which is a natural anticoagulant [3]. Also, activation of endothelial cells promotes platelet activation, which is important for thrombus formation.

Rheumatologic conditions are often inflammatory by nature. However, despite this mechanistic link between rheumatologic diseases and VTEs, these highly inflammatory conditions may be under-recognized as risk factors for hypercoagulability. The only exception is the well-known association between antiphospholipid antibodies in systemic lupus erythematosus (SLE) and both venous and arterial thromboses. There are several case reports, retrospective cohort studies and prospective observational analyses highlighting the increased risk of VTE in patients with rheumatologic diseases [4]-[9]. Most of the data in the literature reveal this concern in patients with rheumatoid arthritis (RA) and SLE, and the SLE studies are focused mostly on increased risks associated with positive antiphospholipid status rather than on the innate hypercoagulability nature of this inflammatory disease [8],[10]. VTEs seem to be linked to disease activity and/or inflammation in many of the inflammatory rheumatologic diseases.

Patients who develop VTEs have high rates of morbidity and mortality [2]. The incidence of first-time VTE in the United States is about 1 in 1,000 person-years [1]. Therefore, it is important to understand the excess magnitude of this issue in patients with inflammatory rheumatologic diseases. Ideally, the modifiable risk factors would be known and altered and poor outcomes mitigated. To investigate these issues, we conducted a meta-analysis of the risk of developing DVT and/or PE in patients with inflammatory arthritis, vasculitis and connective tissue diseases (CTDs) such as SLE, Sjögren's syndrome, inflammatory myositis and systemic sclerosis.

Patients with inflammatory rheumatologic diseases are at increased risk of developing VTEs. To our knowledge, our present meta-analysis is the first to show this increased inherent risk in patients across all inflammatory rheumatologic diseases. This increased risk was seen in studies of hospitalized and nonhospitalized patients. Specifically, patients with RA, SLE, inflammatory myositis, Sjögren's syndrome, ANCA-associated vasculitis and systemic sclerosis appear to have this increased risk. There was insufficient data in the literature for us to comment on risks for other inflammatory conditions.

Certainly, patients with various rheumatologic illnesses may be at increased risk of developing VTEs for several different reasons. Patients with inflammatory joint diseases are more likely to be immobile because of pain, especially during times of active joint inflammation. Also, these patients are more likely to undergo surgeries such as joint arthroplasty procedures. We attempted to account for these confounders in this meta-analysis by excluding VTE rates reported in the perioperative setting. The inherent hypercoagulable nature of inflammatory autoimmune diseases is often overlooked [1],[2]. This is despite recent evidence which suggests that active inflammation is a strong driving force that leads to local and systemic imbalances in the coagulation system [3].

The data presented in our present study are clinically significant when considering the morbidity and mortality associated with VTEs [1],[2]. In a recently published epidemiological study by Tagalakis et al. [2], the 1-year survival rate of unprovoked VTEs was estimated to be only 93%. For patients who developed a VTE secondary to a major risk factor such as recent surgery or hospitalization, the 1-year survival rate was only 84%. If the patient survives a VTE episode, the subsequent cost of rehabilitation and long-term anticoagulation is great. Also, by recognizing active inflammation as a risk factor for VTEs, the duration of anticoagulation may be adjusted once an event has occurred. This may be an area of future research in both rheumatology and hematology.

Recognizing a risk factor is only part of the battle. The natural follow-up to the results reported here is to determine factors important in prevention and when, if at all, it is appropriate to use prophylaxis. There are limited and scant data in the literature regarding this issue. Most of the literature regarding prophylaxis against VTEs in rheumatology patients focuses on antiphospholipid antibody status. Specifically, hydroxychloroquine has shown promise in reducing the incidence of VTEs in patients with SLE [34],[35]. Aspirin also seems to be beneficial in the setting of antiphospholipid antibodies, particularly during pregnancy [36]. However, there are no studies in the current literature in which the benefit of aspirin in rheumatologic patients without antiphospholipids in the serum has been assessed.

We believe that the increased VTE risk is associated with the activity of the inflammatory diseases rather than with the treatments used for controlling the disease. There are two major reasons for this hypothesis. First, some of these studies show that the risk of VTE is highest in the first year of disease, with a significant drop-off in subsequent years. Some postulate that this is because the disease is most active in the first year, with improvement occurring once inflammation is controlled by medications. Second, studies of vasculitis, such as ANCA-associated vasculitides and giant cell arteritis show that the risk of thrombosis is lower when the disease is treated with glucocorticoids and disease-modifying agents. Therefore, we believe that the risk of thrombosis is associated with the disease. However, further studies are required to confirm this hypothesis.

To our knowledge, this meta-analysis is the first comprehensive study to recognize the hypercoagulability risk in patients with many inflammatory rheumatologic conditions. Notable limitations of this meta-analysis include possible publication bias. Due to underrecognition of this issue in the published literature and in practice, negative studies may not have been available for inclusion. However, under the same principle, rates of VTE in the published literature may underestimate true rates. Clinical signs of VTEs, especially DVTs, are often subtle. Patient-reported symptoms may be vague and may even be misattributed to their rheumatologic disease. Another potential limitation of this meta-analysis involves lack of prospective data that include disease activity, other comorbidities and confounders. Particularly in studies without control comparators, it may be difficult to isolate patients' rheumatologic diseases as the sole cause for VTEs.

With regard to the increased risk in patients with SLE- and ANCA-associated vasculitis, we also recognize that their risk is substantially higher compared to the other disease populations. For ANCA-associated vasculitides, we felt that the risk was attributable to a combination of (1) direct vessel injury from inflammation (vasculitis), (2) greater local edema and vascular narrowing secondary to vascular inflammation and remodeling and (3) overall inflammatory state, as in all other inflammatory diseases. Of course, further investigations are required to prove this theory. With regard to the increased risk in SLE, again it is likely a multifactorial issue, including renal involvement (such as nephrotic syndrome, which can increase hypercoagulability by an imbalance in excreting antithrombotic factors), an increased prevalence of antiphospholipid antibodies and an overall inflammatory state, as in all other inflammatory diseases leading to endothelial dysfunction and via other mechanisms. Specifically with regard to antiphospholipid antibody status, we tried to mitigate this confounder as much as possible by excluding studies that directly tested for hypercoagulability in the setting of known antiphospholipids in the serum. However, we recognize that the studies included did not specifically test for or exclude antiphospholipid status, so it is probable that at least some of the increased VTE risk in SLE patients is due to antiphospholipid antibodies. The prevalence of antiphospholipid syndrome in SLE patients can vary by cohort, so part of the usual lupus cohort studied would have one-fourth with antiphospholipid antibodies and some with renal lupus. In general, none of the included studies reported disease activity in relation to the VTE events, so we were unable to perform subanalyses in this regard.

In our present meta-analysis, we pooled the studies using a random-effects model. However, high heterogeneity was still observed, owing to the wide differences in patient populations in each cohort. These differences included ethnicity, age, duration of disease, activity of disease, setting of VTE (that is, inpatient vs. outpatient), environmental factors and other unknown variables. We tried to minimize the variance by including large studies with clearly defined patient characteristics. Still, there were a limited number of studies from which data were pooled. However, for all inflammatory rheumatic diseases for which data were extracted, the rate of VTEs was higher than expected, with a pooled increased odds ratio of 3.4.

In this study, we quantified the elevated risk of venous thromboembolic events in patients across the included inflammatory rheumatologic diseases. Although we provide strong evidence for this elevated baseline risk in the rheumatology patient population, identifying high-risk patients within each disease and reducing the risk by treatment of disease activity or antiplatelet prophylaxis cannot be determined on the basis of our meta-analysis.