Venous and arterial thrombosis in patients with VEXAS syndrome Free

VEXAS (vacuoles, E1 enzyme, X-linked, autoinflammatory, somatic) syndrome is a newly described autoinflammatory disorder caused by somatic mutations in the ubiquitin-like modifier activating enzyme 1 (UBA1) gene.¹ Most commonly involving substitutions of methionine-41 (p.Met41), these loss-of-function mutations result in decreased cytoplasmic ubiquitylation and increased misfolded proteins and endoplasmic reticulum stress, leading to upregulation of inflammatory pathways.^2-5 

VEXAS syndrome is characterized by a diverse range of multisystem inflammatory manifestations, including constitutional symptoms, skin involvement, ear and nose chondritis, vasculitis, joint involvement, and pulmonary and cardiac manifestations.⁶ Prevalent hematologic abnormalities include cytopenias (commonly macrocytic anemia), myelodysplastic syndrome (MDS), and plasma cell dyscrasias (PCDs).^1,7 Rates of thrombosis appear strikingly increased in patients with VEXAS syndrome, with reported incidence of venous thromboembolism (VTE) from 35% to 56% and arterial events from 1% to 25% in different case series.^7,8 Many reported VTE events are unprovoked and recurrent and can occur while on therapeutic anticoagulation.⁹ 

Mortality is high in VEXAS syndrome, up to 50% depending on the underlying UBA1 mutation.^5,8,10 Unfortunately, effective therapies are lacking, and most patients require lifelong corticosteroids to prevent disease flares, resulting in significant toxicities.⁸ The contribution of thrombosis to disease morbidity and mortality in VEXAS syndrome has not been well established, and the drivers of increased thrombotic risk remain unclear. Two potential factors identified in our initial small report of 16 patients were an increase in factor VIII (FVIII) levels and D-dimers, suggesting inflammation plays a role, and lupus anticoagulant (LA), suggesting a potential antibody-mediated mechanism.⁷ 

In the current study, we describe the incidence and characteristics of thrombosis in a large cohort of patients with VEXAS syndrome, correlate its presence with clinical and survival outcomes, and assess for potential risk factors.

Patients were enrolled in clinical research protocols (NCT02257866, NCT00001373, NCT05012111) approved by the institutional review board at the National Institutes of Health (NIH) or were seen as part of standard clinical care at the Mayo Clinic (MC) (institutional review board 21-009151). All gave informed consent in accordance with the Declaration of Helsinki. Clinical data were obtained from NIH or MC electronic medical records or from referred outside records.

Patients had a confirmed pathogenic UBA1 somatic mutation and clinical features of VEXAS syndrome. Time of symptom onset (either inflammatory or hematologic) was used as time of disease onset since VEXAS syndrome clinical symptoms preceded genomic diagnosis by years in most cases. Disease characteristics collected included patients’ genotype, inflammatory phenotype, and associated hematologic conditions including cytopenia, PCD, or MDS. Thrombosis was considered the disease onset of VEXAS syndrome if it occurred within 2 years of other inflammatory or hematologic symptoms. Patients with a thrombosis date >2 years prior to VEXAS syndrome symptom onset (n = 5) were excluded from analysis as their event was unlikely VEXAS syndrome related and could not be included in statistical analysis as their thrombosis date preceded symptom onset date. Five patients were excluded due to preceding VTE (n = 2) 11 and 16 years prior to symptom onset and myocardial infarction (MI) (n = 3) occurring 14, 12, and 4 years earlier.

Thrombosis was defined as any venous or arterial thromboembolism excluding superficial vein thrombosis (SVT). VTE was defined as pulmonary embolism (PE) or deep vein thrombosis (DVT). Below-knee DVT was defined as distal to the popliteal vein. Arterial thrombosis included ischemic stroke, MI, or occlusive peripheral arterial disease. SVT was defined as thrombosis of the axial veins (great saphenous, anterior accessory saphenous, and small saphenous).

Provoking factors were defined as per American Society of Hematology 2020 guidelines.¹¹ Inflammatory and autoimmune diseases were not considered risk factors and were assessed separately as related to VEXAS syndrome.

VEXAS syndrome inflammatory symptoms were defined as follows: constitutional symptoms (fever, chills, weight loss, night sweats, or fatigue), Sweet syndrome (neutrophilic dermatosis with or without fever), skin manifestations (rashes not attributable to another cause), musculoskeletal symptoms (joint, muscle, or connective tissue inflammation), pulmonary involvement (pulmonary infiltrates or pleural effusions), and cardiac manifestations (cardiomyopathy, myocarditis, pericarditis, or pericardial effusion). Data cutoff date was March 2023.

Coagulation assays were Clinical Laboratory Improvement Amendments (CLIA)-certified and performed in a clinical laboratory at the NIH Clinical Center. LA was collected in the majority (41 of 42) of NIH patients at time of initial clinic visit, not at the time of acute thrombosis. Other coagulation testing included D-dimer, fibrinogen, factor assays (II, V, VII, VIII, IX, X, XI), von Willebrand factor (VWF) antigen and activity, anti-phospholipid antibodies (APLAs) (B2 glycoprotein [B2GP] and anti-cardiolipin antibodies [ACL]), antithrombin III, protein C and S, thromboelastography (TEG), P-selectin, thrombomodulin, plasminogen, and plasminogen activator inhibitor type 1. These specialized assays were performed on up to 27 NIH patients with VEXAS syndrome prospectively regardless of the presence of thrombosis; some patients did not have every individual test performed (supplemental Table 1, available on the Blood website). In the remaining 15 NIH patients, testing for APLAs was performed but otherwise they underwent minimal coagulation testing. Coagulation data on MC patients were not included due to low numbers, bias for collection only in patients with thrombosis, and differences in assays. Patients undergoing prospective coagulation assay collection had documented suspension of anticoagulation at the time of bone marrow biopsy, part of standard evaluation for cytopenias, MDS, and risk of hematologic malignancy, and not for research coagulation testing specifically. Coagulation testing was performed the day of the marrow biopsy due to optimal timing. Patients were risk-assessed preprocedure in regards to their anticoagulation management. Factor activity assays were prothrombin time or activated prothrombin time based. VWF antigen was assessed using enzyme-linked immunosorbent assay and activity using the ristocetin cofactor assay. LA was dilute Russell's viper venom time (dRVVT)-based and, if necessary, both dRVVT and activated prothrombin time based, in an algorithmic approach in which both types of assays consisted of a screening and confirmatory step. Full details of all coagulation assays performed can be found in supplemental Table 2. High or low values were determined as outside normal laboratory ranges.

Patient characteristics are reported as median with range for continuous variables and count with proportion for categorical variables. Univariate and multivariate logistic regression (LR) was performed to quantify the association of baseline variables with clinical outcomes. Kaplan-Meier curves and the Cox proportional hazards (PH) models were used to estimate overall survival (OS), with the log-rank test to compare survival distributions and determine statistical significance. Univariate Fine-Gray competing risk regression models and cumulative incidence curves were generated to analyze time to first thrombotic event in the presence of death, to account for this as a competing risk. For all time to event analyses, 3 time points were considered: the onset of inflammatory symptoms, the date of thrombotic event (if applicable), and the time to event date for censoring or death. The Pearson correlation coefficient and scatterplots were used to assess the correlation between coagulation factors and C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), and LA; statistical significance was examined using the Student t test. The data were analyzed using R version 4.0.2 and GraphPad Prism. Statistical significance was measured as P < .05.

A total of 119 patients were included in this study (Table 1). All patients were male (100%) with a median age of 65 years (range, 39-86) at diagnosis. Somatic mutations in UBA1 included p.Met41Thr (c.122T>C) in 59% (n = 69), p.Met41Val (c.121A>G) in 23% (n = 27), p.Met41Leu (c.121A>C) in 14% (n = 16), and splice motif mutation in 4% (n = 5). The commonest clinical symptoms were constitutional (fever, fatigue, weight loss; n = 95, 80%), skin involvement (n = 89, 75%), and musculoskeletal (arthritis, tenosynovitis; n = 76, 64%). Macrocytic anemia was the most frequent cytopenia (90%, n = 107); 36 patients (30%) were diagnosed with MDS and 21 (18%) had a PCD.

Forty-nine percent (58 of 119) of patients had a thrombotic event, either VTE or arterial thrombosis. VTE was more common (n = 49, 41%) than arterial thrombosis (n = 15, 13%). Of the patients with VTE, 41 (84%) had a DVT, 17 (35%) had a PE, and 9 (18%) had both DVT and PE (Table 1). Twelve patients had upper limb DVT, 3 of which were line associated. Of the 49 patients who had a VTE, most (n = 32, 65%) were unprovoked and 41% (n = 20) were recurrent; 10 patients (20%) had a VTE while on therapeutic anticoagulation (Table 1). Major transient risk factors for provoked VTE were hospitalization >3 days (n = 8), surgery (n = 1), and significant lower limb injury (n = 1). Minor transient risk factors were travel >4 hours (n = 3) and minor leg injury (n = 2). One patient with hospitalization >3 days had a line-associated upper limb DVT. Two patients had both an unprovoked VTE and subsequent line-associated DVT.

The median time from symptom onset to the first VTE was 410 days (range, 0-6636) and to the first arterial event was 349 days (range, 0-2780). The cumulative incidence of VTE (Figure 1A) at 1 year from diagnosis was 17%, which increased to 40% by 5 years, higher for DVT (cumulative incidence of 9% at 1 year, 35% at 5 years) than PE (cumulative incidence 4% at 1 year, 13% at 5 years) (Figures 1C-D). Cumulative incidence for VTE was 7% and 46% for UBA1 M41L, 22% and 37% for M41V, and 18% and 41% for M41T at 1 and 5 years after symptom onset, respectively (P = .37) (Figure 1E). No patients with splice site variants had VTE. Patients with UBA1 M41T and M41V tended to have events early in their disease course compared with those with M41L.

The cumulative incidence of arterial thrombosis was lower, 6% at 1 year and 11% at 5 years (Figure 1B). Many patients had arterial thrombotic risk factors (Table 1), and 20% had a history of atrial fibrillation. Of the 15 patients with arterial thrombosis, 5 (33%) had a stroke and 7 (47%) had an MI. Other arterial events occurred in 3 patients: 1 with critical limb ischemia disease requiring amputation and 2 with acute thrombosis of the lower limb. One patient had more than 1 arterial event, an MI and splenic infarct. Of patients who had a stroke, 2 (40%) had a history of atrial fibrillation, and of the 3 patients with arterial limb ischemia, 1 had atrial fibrillation and 1 had aortic mural thrombus. No patients had known carotid stenosis.

Cardiac manifestations of VEXAS syndrome were strongly associated with development of VTE. This was true using univariate LR (odds ratio [OR] 9.63, 95% confidence interval [CI] 1.57-185.15, P = .04) and the Cox PH and Fine-Gray models, (hazard ratio [HR] 3.13, P < .001 and subdistribution hazard 3.22, P < .0001, respectively) (Figure 2A, supplemental Table 9). No other factors were associated with increased risk of VTE (Figure 1A, Table 2, and supplemental Table 9).

In subgroups of VTE, cardiac manifestations remained associated with both DVT (univariate LR: OR 5.28, 95% CI 1.08-38.1, P = .05; Cox PH: HR 2.96, P = .03; Fine-Gray subdistribution hazard: 3.04 P < .001) and PE (univariate LR: 5.25; 95% CI 0.95-26.34, P = .04; Cox PH: HR 4.56, P = .02; Fine-Gray subdistribution hazard: 4.4, P < .02). Pulmonary involvement (LR: OR 2.25; 95% CI 1.05-4.94, P = .04; Cox PH: HR 2.08, P = .03; subdistribution hazard: 1.94, P < .04,) was associated only with DVT (Figure 1B-C, supplemental Tables 5 and 6, 10, and 11). No other clinical factors were associated with increased risk of DVT or PE.

UBA1 M41L variant was associated with PE on both univariate (supplemental Table 5; OR 4.58, 95% CI 1.28-16.21, P = .02) and multivariate (OR 16.94, 95% CI 1.99-144.30, P = .01) LR; this was not true for DVT (supplemental Table 6; OR 1.05, 95% CI 0.27-4.19, P = .94) or arterial thrombosis (supplemental Table 7; OR 0.27, 95% CI 0.02-3.43, P = .31).

Arterial thrombosis was associated with ear or nose chondritis on univariate (supplemental Table 7; OR 3.2, 95% CI 1.06-10.91, P = .046) but not multivariate analysis (OR 3.59, 95% CI 0.77-16.64, P = .10). No other factors were associated with arterial thrombosis (supplemental Table 7).

The OS at median follow-up time of 4.8 years was 88%; there was no significant difference in OS between patients with (88%) or without (89%) a thrombotic event (Figure 3A). Using Cox PH model, neither VTE (DVT or PE) nor arterial thrombosis was associated with worse survival (Figure 3B, supplemental Table 8). However, age at diagnosis (HR 1.07, 95% CI 1.01-1.13, P = .01) and pulmonary involvement (HR 3.05, 95% CI 1.18-7.86, P = .02) were predictive for increased risk of mortality. There was no significant difference in OS between UBA1 variants (P = .2) (supplemental Figure 1).

Twenty-five patients died, and cause of death was reported in 14. Infection was most frequent (n = 6), followed by progression of VEXAS syndrome (n = 4). No patients died due to thrombosis. One patient had an intracranial hemorrhage, had been receiving rivaroxaban, and had mild thrombocytopenia (96 K/μL) at last follow-up.

Of the 49 patients who developed VTE, data on anticoagulation were available for 47. Forty-five patients received at least 1 anticoagulant and 2 received none, 1 due to severe thrombocytopenia and 1 who had isolated below-knee DVT with SVT. Indication for anticoagulation was VTE in 38 patients, atrial fibrillation and VTE in 7 patients, and VTE with coronary artery disease in 1 patient. Anticoagulants received included rivaroxaban (n = 20), apixaban (n = 16), warfarin (n = 9), enoxaparin (n = 9), and edoxaban (n = 3). A total of 11 patients received >1 anticoagulant; of these, 9 had >1 VTE. Ten patients had recurrent VTE while on therapeutic anticoagulation; most (n = 5) had been on direct oral anticoagulants or warfarin (n = 3). Of those with VTE, 30 (61%) remained on anticoagulation at last follow-up.

Sixteen patients were on antiplatelet therapy, 4 for arterial events, 3 for primary prevention, 2 for VTE, and 7 for other reasons. Of the 15 patients with arterial events, most received anticoagulation (n = 9), antiplatelets (n = 3), or both (n = 1); 2 had unavailable data.

At time of first VTE, data on corticosteroid and immunosuppressive therapy were available for 36 (73%) and 35 (71%) patients, respectively. Twenty-three (64%) patients were not on corticosteroids, and of the 13 patients who were, prednisone doses ranged from 5 mg to 60 mg at the time of VTE. Two-thirds (n = 23; 66%) were not on any steroid-sparing immunosuppression at the time of VTE. Other immunosuppressants included methotrexate (n = 5), JAK inhibitors (upadacitinib [n = 1] and ruxolitinib [n = 1]), mycophenolate (n = 1), tocilizumab (n = 1), azathioprine (n = 1), infliximab (n = 1), hydroxychloroquine (n = 1), and rituximab (n = 1). A total of 37 patients had available data to assess for acute disease activity (acute inflammatory symptom with or without high CRP/ESR) or chronic poorly controlled inflammation (persistent inflammatory symptom or high CRP/ESR) at time of VTE; 47% had either acute disease activity (n = 12, 32%) or had poorly controlled inflammation (n = 6, 15%). Only 1 patient had active multiple myeloma requiring therapy, without VTE or arterial thrombosis.

Coagulation assay testing was available in a subset of patients tested consecutively at the NIH (Figure 4 and supplemental Table 1). The majority of patients with VEXAS syndrome had high FVIII levels (26 of 27; 96%) and VWF activity (16 of 27; 59%), and many had high VWF antigen (11 of 25; 44%), FIX levels (12 of 27; 44%), and protein C activity (11 of 27; 41%). D-dimers were >0.5 μg/mL in 17 of 27 (63%) and >1 μg/mL in 6 of 27 (22%). Fibrinogen was high in 8 of 26 (31%) patients. Other factor levels, antithrombin III activity, protein S activity, plasminogen, plasminogen activator inhibitor type 1, P-selectin, and thrombomodulin were within normal range in the majority (supplemental Table 1). CRP and ESR positively correlated with D-dimer (R = 0.54, P = .003 and R = 0.64, P < .001, respectively) and negatively correlated with VWF activity (R = −0.42, P = .04 and R = −0.45, P = .03, respectively) (supplemental Figure 2).

There were no differences in coagulation factor levels between patients with and without thrombosis (Figure 4A-O), except for plasminogen, where patients without thrombosis had significantly higher levels (P = .03; Figure 4E). No patients had paroxysmal nocturnal hemoglobinuria (PNH) clones nor pathogenic mutations in the FV Leiden or prothrombin G20210A genes.

LA was performed in 41 (98%) NIH patients and was positive in 16 of 41 (40%); this was persistent in 9 of 10 patients with >1 LA performed, 7 of whom had thrombosis and 2 of whom did not. All patients with thrombosis who had persistent LA positivity remained on anticoagulation at last follow-up. LA positivity correlated significantly with higher CRP (P < .01; supplemental Figure 3). Four patients with positive LA had concurrent low-level positive ACL immunoglobulin M antibodies (range 21-37 IgM phospholipid units), and none had positive ACL immunoglobulin G or B2GP antibodies.

TEG testing was within normal parameters for most patients (supplemental Table 1 and supplemental Figure 4). TEG reaction time ranged from 4.2 to 9.1 minutes (normal 4.6-9.1 minutes), TEG clot time ranged from 0.8 to 2.3 minutes (normal 0.8-2.1 minutes), TEG alpha angle ranged from 61° to 78° (normal 63° to 78°), and TEG MA (maximal amplitude) ranged from 47 to 78 mm (normal 52-69 mm) (supplemental Table 1). Most abnormalities were seen in TEG MA with 4 patients <52 mm and 2 patients >69 mm. Three patients with low TEG MA had thrombocytopenia.

Patients with VEXAS syndrome have high rates of VTE, as confirmed in this large cohort. Incidence of VTE is similar to some of the highest-risk secondary prothrombotic disorders. For example, patients with Behçet syndrome, a form of vasculitis, have a reported incidence of ∼40% VTE or arterial thrombosis, and those with PNH have a 10-year cumulative incidence of thrombosis of 44% (without primary thromboprophylaxis),¹² both similar rates to our VEXAS cohort. Additionally, the risk for VTE recurrence was very high in our study, 20% while on anticoagulation. However, despite the high incidence in VEXAS syndrome, thrombosis was not associated with poorer overall survival, and no deaths were directly attributed to a thrombotic complication. Common causes of deaths in our cohort included infection and progression of VEXAS syndrome, as has previously been reported.^1,5 

Although VTE rates were clearly striking, rates of arterial thrombosis, although high, were less so when accounting for the age and sex demographic of VEXAS syndrome, which is older men. The prevalence of atherosclerosis is high, ∼75%, in the general American population aged 60 to 79.¹³ Additionally, many of our patients had pertinent atherosclerotic risk factors, and we found a high rate of atrial fibrillation (21%) that may have contributed. Therefore, the rate of arterial events may be reflective of factors other than underlying VEXAS syndrome; further longitudinal data will be needed to further delineate this.

The mechanisms underlying increased thrombotic risk in VEXAS syndrome remain incompletely understood but likely relate to the high degree of inflammation driving dysregulation between coagulant and anticoagulant factors.^14,15 Loss of ubiquitination due to UBA1 mutations leads to cellular stress with accumulation of vacuoles, activation of immune pathways, and significant inflammatory cytokines, which ultimately results in a prothrombotic profile, including increased platelet activation and clotting factor activity. Increased FVIII activity has been confirmed in our large cohort as the predominant coagulation abnormality present in patients with VEXAS syndrome.⁷ Both elevated FVIII and VWF activity have been associated with increased risk of VTE, though the association with VWF may be due to its role in binding to FVIII. FVIII is an acute phase reactant; studies have shown that risk of VTE is independently associated with FVIII level in a dose-dependent manner, with levels >175 IU/dL a particular risk of recurrent thrombosis.^16,17 In our study, almost all patients had FVIII levels >175 IU/dL, regardless of whether they had suffered a thrombotic event or not. As our coagulation assay data were not collected at the time of thrombosis, we were not able to determine whether patients had higher levels at the time of the acute event; however, it has been shown that persistently elevated FVIII levels, remote from thrombosis, increase risk of recurrence, and it is therefore likely that persistent elevation of FVIII in VEXAS syndrome contributes to overall thrombotic risk.¹⁸ 

VEXAS syndrome is a myeloid cell–driven disease. Analysis of the transcriptome of VEXAS syndrome neutrophils has revealed upregulation of gene networks relevant to thromboinflammation, such as tumor necrosis factor α, interleukin-6, interferon gamma, and others.¹ Additionally, neutrophils from patients with VEXAS syndrome have demonstrated exaggerated spontaneous NETosis, an inflammatory cell death where neutrophils expunge their DNA as a weblike scaffold of extracellular traps (NETs), driving inflammation and thrombosis. Monocytes may also be playing a role in the pathogenesis of thromboinflammation, as activated monocytes express tissue factor, which plays an essential role in coagulation.¹⁹ Additionally, within monocytes, an increased percentage of spliced XBP1 and activated unfolded protein response lead to cellular stress and increased inflammatory response; elevated levels of tumor necrosis factor α, interleukin-6, interferon gamma, interleukin-8, and interferon-inducible protein 10 have been described, where a prothrombotic milieu arises from this cytokine-mediated interaction between endothelial cells, leukocytes, and platelets.²⁰ 

Vasculitis may also contribute to thrombotic risk in patients with VEXAS syndrome, resulting in direct inflammation and injury to blood vessels along with alterations in the endothelial lining that promote prothrombotic conditions, particularly in smaller vessels.²¹ Vasculitis has been reported in between 26% and 64% of patients with VEXAS syndrome, particularly with neutrophilic infiltration.^8,22 The pathogenesis of thrombosis in relation to chronic inflammation is complex, with multiple proinflammatory mediators and activation of various cell-cell interactions, which may lead to endotheliopathy. In Behçet syndrome, a model of thromboinflammation, an increase in neutrophil activation leads to endothelial dysfunction and higher levels of reactive oxygen species as well as increased NETs, activating platelets and contributing to thrombogenesis.²³ NETs, then, activate macrophages, leading to increased inflammatory cytokines, and are able to impact coagulation cascades and production of thrombin.²³ In Behçet syndrome, increased levels of endothelial injury markers including VWF and thrombomodulin have been reported.^24,25 In our cohort, although 44% of patients with VEXAS syndrome had higher than normal levels of VWF, levels of thrombomodulin were within normal limits, and there was no difference between those with thrombosis. Interestingly, venous wall thickening detected on Doppler ultrasound was shown to be a sensitive and specific method for diagnosis of Behçet syndrome, independent of thrombosis or active inflammation.²⁶ Using the same method, 2 patients with VEXAS syndrome have been found to have significantly greater venous wall thickening compared with Behçet syndrome or antiphospholipid syndrome (APS) patients, suggesting that the level of vessel inflammation in VEXAS syndrome may also be a contributing factor.²⁷ 

Furthermore, although we noted decreased plasminogen levels among those who had thrombosis in our study, it is difficult to ascertain whether this is a direct consequence of thrombosis and fibrinolysis in these patients. Plasminogen also plays nonfibrinolytic roles, such as removal of misfolded proteins and regulation of macrophages and neutrophil apoptosis. Therefore, the reduction in plasminogen seen in patients with VEXAS syndrome may indeed relate to underlying pathophysiology of the disease rather than be due to the fibrinolysis pathway.²⁸ 

At time of thrombosis, most patients were not on steroids or other immunosuppressive therapy, suggesting that lack of disease control may play a role in thrombotic risk, with higher rates of thrombosis occurring earlier in the disease course before patients are established on immunosuppression. Although no current consensus exists on how to define severity of disease or flares in VEXAS syndrome, approximately half of patients had some evidence of either acute disease activity or chronic inflammation at the time of thrombotic event. Timing of the VTE early in the disease, disease activity at time of VTE, and the high FVIII levels all suggest that inflammation is a driver of thrombosis in VEXAS syndrome. However, given the study’s retrospective design, it is not possible to directly correlate risk to disease activity, and further prospective studies are needed to fully elucidate this.

A high incidence of LA has been reported in VEXAS syndrome and confirmed within our cohort.⁷ LA is an in vitro assay. Interference with coagulation tests is attributed to APLAs, a heterogeneous group of antibodies directed against phospholipids such as cardiolipin, or phospholipid-binding proteins such as B2GP. APLAs make up the laboratory criteria for APS, a systemic autoimmune condition associated with thrombosis (venous, arterial, or microvascular). Positive APLA is a feature of many autoimmune disorders, in particular systemic lupus erythematosus, where positive LA and ACLs are strongly associated with increased thrombotic risk; persistently positive LA has also been associated with thrombosis in healthy individuals.^29,30 In our cohort, the presence of LA did not correlate with thrombosis, and we observed very few patients who had B2GP or ACL positivity. LA positivity may also be due to other factors such as inflammation, infection, high CRP, or assay interference from anticoagulation, as opposed to representing a true APS. In our cohort we did see a correlation between high CRP and LA positivity, suggesting possible assay interference and false-positive results. Additionally, we only assessed for ACLs and B2GP, not prothrombin, phosphatidylserine, or phosphatidylinositol antibodies, other antibodies that could result in a positive LA. Finally, 14 of the LA assays were collected in historic patients prior to prospective collection of samples, and although we do have some documentation that anticoagulation was held, it is possible that some unknown contamination with anticoagulation occurred.

Risk factors for thrombosis included cardiac manifestations and pulmonary manifestations. Cardiac and pulmonary manifestations may represent a more severe disease phenotype or select for those requiring more hospitalizations. Leucine UBA1 genotype was specifically associated with PE but not with other thromboses; the reason is not clear and will require further study. Although thrombosis was not associated with increased mortality, pulmonary involvement was a poor prognostic marker in VEXAS syndrome with an HR of >3. MDS has previously been identified as a poor prognostic feature of VEXAS syndrome by our group but was not seen in this present study.⁸ As we were not focused on MDS, diagnosis was based on clinical records, and patients did not undergo specific review and reclassification of their bone marrow findings and cytopenias as performed previously³¹ in which a rigid definition of MDS was applied, which may account for the discrepancy.

Whether patients with VEXAS syndrome should be treated with prophylactic anticoagulation is an ongoing area of discussion. In some other disorders with similarly high thrombotic risk, such as PNH, or in high-risk cancers, prophylactic anticoagulation is utilized.^11,32 In inflammatory bowel disease, thrombotic risk is threefold higher than the general population and is associated with acute disease flares. Current guidelines suggest pharmacologic prophylaxis in hospitalized patients with inflammatory bowel disease without evidence of bleeding and in outpatients with history of provoked VTE, but not in outpatients experiencing flares without a VTE history.³³ As thrombosis in VEXAS syndrome is likely driven by inflammation, the disease itself represents an ongoing nonmodifiable risk factor, and patients may also be less mobile or require hospitalization during disease flares, further increasing thrombotic risk. However, patients with VEXAS syndrome also have multiple risk factors that put them at increased bleeding risk, including older age, disease-related comorbidities, chronic steroid use, and thrombocytopenia. In the setting of malignancy, scoring systems have been utilized, and such a strategy may be useful in VEXAS syndrome to identify patients with high thrombotic risk but lower bleeding risk in whom preemptive anticoagulation may be appropriate. In the absence of contraindications, patients should receive thromboprophylaxis in high-risk situations such as hospitalization or surgery and be considered for thromboprophylaxis with individual risk assessment for other situations such as immobilization outside of hospitalization, immobilization of a limb, or administration of certain medications such as JAK inhibitors or lenalidomide. Patients with VEXAS syndrome who develop VTE should be strongly considered for extended anticoagulation due to their risk of recurrence. Periodic bleeding risk assessments are necessary given the risk of progressive marrow failure. Our study was not designed to compare different anticoagulants, therefore choice of drug is recommended per standard practices. Prospective studies on outpatient thromboprophylaxis and efficacy of specific anticoagulants in this population are needed.

In summary, we found that in VEXAS syndrome VTE contributes significantly to morbidity. With the increasing awareness and thus increasing incidence of VEXAS syndrome, patients and physicians should be educated regarding this progressive disease and its high risk of VTE.

This research was supported by intramural funding from the National Heart, Lung, and Blood Institute, National Institutes of Health. The visual abstract was created with BioRender.com.

Contribution: Y.K., A.G., and E.M.G. designed the study, analyzed data, and wrote and edited the manuscript; R.B. and C.O.W. performed statistical analysis; P.E.A. collected and analyzed data; A.D.-F., K.N., S.R.P., Y.K., K.R.C., and R.C.B. performed and analyzed coagulation assays; M.A.F., A.H., K.A.Q., A.K.O., K.R., I.D., W.G., L.W., H.O., T.L., D.L.K., K.J.W., A.M., R.S.G., D.B.B., M.M.P., N.S.Y., A.I.C., P.C.G., M.J.K., D.E.H., B.A.P., and J.D. provided patient care; and all authors edited the paper and approved the final manuscript.

Conflict-of-interest disclosure: N.S.Y. has a cooperative research and development agreement with Novartis. M.M.P. has research funding from Kura, Stem Line, Epigenetix, and Polaris and is on the advisory board for CTI Pharmaceuticals. The remaining authors declare no competing financial interests.

Correspondence: Emma M. Groarke, 10 Center Drive, Building 10-CRC, Room 3E-5140, Bethesda, MD 20892; email: emma.groarke@nih.gov.