Whole-blood coagulation and platelet dynamics identify differential prothrombotic states in patients with rheumatoid arthritis treated with JAK or TNF-ɑ inhibitors

Abstract

Background

Rheumatoid arthritis (RA) is associated with inflammation-driven hypercoagulability and increased venous thromboembolic risk. Post hoc analyses of safety trials have raised concerns regarding a potential differential thrombotic risk with Janus kinase inhibitors (JAKi) compared with tumour necrosis factor inhibitors (TNFi).

Objectives

To compare viscoelastic coagulation profiles and platelet reactivity in patients with RA treated with JAKi or TNFi, and in healthy controls (HC).

Methods

In this single-centre observational study, patients with RA with stable JAKi or TNFi therapy (>3 months) underwent whole-blood rotational thromboelastometry (ROTEM) and impedance aggregometry (MULTIPLATE). The primary outcomes were evaluation of: (a) whole-blood thrombogenic potential and (b) platelet reactivity. Multivariable linear regression adjusted for age, body mass index, ln-transformed Simple Disease Activity Index, treatment group and ln-transformed prednisone dose.

Results

Sixty-one patients with RA (30 JAKi, 31 TNFi) and 34 HC were included. Compared with HC, patients with RA exhibited significantly higher maximum clot firmness (MCF) across all ROTEM assays (INTEM 63.4±4.0 vs 57.2±4.1 mm; EXTEM 66.2±4.1 vs 55.9±4.7 mm; FIBTEM 15.7±3.6 vs 12.4±3.7 mm; all p<0.001) and enhanced platelet aggregation (all p<0.001). In multivariable analyses, JAKi therapy (vs TNFi) was independently associated with higher EXTEM MCF (β=2.32, 95% CI 0.42 to 4.23; p=0.018) and FIBTEM MCF (β=1.72, 95% CI 0.07 to 3.37; p=0.041). Prednisone dose independently predicted INTEM and EXTEM MCF, while SDAI independently predicted ADP-induced aggregation (β=11.34, 95% CI 4.31 to 18.37; p=0.002).

Conclusions

RA is characterised by enhanced clot firmness and platelet reactivity. JAKi therapy was independently associated with higher viscoelastic clot strength compared with TNFi, while inflammatory burden and glucocorticoid exposure were key determinants of prothrombotic signatures.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Rheumatoid arthritis (RA) is associated with a prothrombotic state and an increased risk of venous thromboembolism driven by chronic systemic inflammation.

• Safety signals from clinical trials and observational studies have suggested a potentially higher risk of thrombotic events in some patients treated with Janus kinase inhibitors (JAKi) compared with tumour necrosis factor inhibitors (TNFi).

• Conventional coagulation biomarkers provide limited insight into global haemostatic dynamics and may not fully capture inflammation-driven hypercoagulability.

WHAT THIS STUDY ADDS

• Using whole-blood viscoelastic testing, patients with RA demonstrated increased clot firmness and platelet reactivity compared with healthy controls, confirming a systemic procoagulant phenotype.

• Treatment with JAKi was independently associated with higher EXTEM and FIBTEM maximum clot firmness compared with TNFi after adjustment for clinical covariates.

• These findings suggest that JAK inhibition may be associated with distinct haemostatic signatures consistent with enhanced thrombo-inflammatory activation in RA.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Viscoelastic whole-blood testing may represent a useful translational tool to investigate therapy-specific haemostatic modulation in inflammatory diseases.

• Understanding how targeted therapies influence thrombo-inflammatory pathways may help refine cardiovascular and thrombotic risk stratification in patients with RA.

• Future longitudinal studies integrating viscoelastic biomarkers with clinical thrombotic outcomes are needed to clarify the clinical relevance of these haemostatic signatures.

Introduction

Rheumatoid arthritis (RA) is associated with significant mortality and cardiovascular morbidity, most notably an increased risk of major adverse cardiovascular events (MACE)¹ and venous thromboembolism (VTE).² Patients with RA exhibit a nearly twofold higher risk of deep vein thrombosis (DVT) and pulmonary embolism (PE) compared with the general population, yet this risk is strictly related to disease activity control.² This prothrombotic state is driven by thrombo-inflammation, where systemic cytokines—such as tumour necrosis factor-alpha (TNF-α) and interleukin-6 (IL-6)—trigger endothelial dysfunction, enhance tissue factor expression and promote reactive megakaryopoiesis.³

The introduction of targeted therapies, including Janus kinase inhibitors (JAKi) and TNF-α inhibitors (TNFi), has largely improved RA outcomes and reduced its associated cardiovascular disease (CVD) risk.⁴ However, safety concerns have emerged regarding a potential differential risk of VTE. Post hoc analyses of the Oral Rheumatoid Arthritis Trial (ORAL) Surveillance trial indicated an increased risk of MACE and VTE in high-risk patients treated with the JAKi tofacitinib compared with TNFi (etanercept and adalimumab), leading to global safety alerts and prescribing restrictions.^{5 6}

As demonstrated by longitudinal data from patients treated with JAKi or TNFi,^{7 8} conventional coagulation tests are insufficient to determine whether various disease-modifying antirheumatic drugs (DMARDs) may differentially modulate thrombo-inflammatory pathways in RA.

Whole-blood (WB) viscoelastic coagulation testing (VET), specifically rotational thromboelastometry (performed with ROTEM apparatus), offers a comprehensive assessment of global WB evaluation of clot initiation, propagation and stability, capturing dynamic changes in haemostasis that are not detectable with traditional tests.⁹ Impedance aggregometry (performed with MULTIPLATE apparatus) evaluates platelet reactivity to specific agonists^{10 11} (figure 1).

Figure 1. (A–C) Rotational thromboelastometry (ROTEM) is a viscoelastic laboratory test used to assess whole-blood coagulation (including the full contribution of blood cells) in real time. Using an electromechanical transducer and a viscoelastic measurement method, it provides a graphical representation of the clot formation process. (D–F) Aggregometry is a set of tests that assess platelet function by measuring their ability to activate and aggregate. As platelets adhere to two electrodes in the sample, impedance rises and generates a curve. Its area (AUC) is the key parameter, capturing both how much and how fast platelets aggregate. Increased values in these tests indicate global platelet hyper-reactivity, suggesting a prothrombotic coagulative state. ASPI-test, arachidonic acid-induced platelet aggregation test; AUC, area under the curve; COX-1, cyclo-oxygenase-1; PAR-1, protease-activated receptor 1; TRAP-test, thrombin receptor-activating peptide-induced platelet aggregation test.

Emerging evidence indicates that ROTEM can detect hypercoagulability in RA,¹² characterised by shorter clot formation time (CFT) and increased maximum clot firmness (MCF). These viscoelastic alterations appear to reflect inflammatory activity, supporting the presence of inflammation-associated hypercoagulability across immune-mediated conditions.^{12 13} The JAnus Kinase and KLOTting (JAKKLOT) study was designed to assess whether JAKi and TNFi differentially modulate the prothrombotic state and platelet function in an actively treated RA cohort.

Methods

Study population and design

JAKKLOT is a single-centre observational study. The present analysis is a cross-sectional evaluation of prospectively enrolled patients with RA aimed at comparing the effect of active treatment with DMARDs on coagulation parameters. Inclusion criteria are: diagnosis of RA according to the 2010 American College of Rheumatology/European Alliance of Associations for Rheumatology (EULAR) criteria; disease duration >12 months; stable treatment (>3 months) with a JAKi (tofacitinib, baricitinib, upadacitinib, filgotinib) or TNFi (adalimumab, etanercept, certolizumab pegol). Exclusion criteria were: age <18 years and >80 years; current use of antiplatelet agents or anticoagulants; platelet counts <150 000/μL or >450 000/μL, abnormal haematocrit; history of MACE (myocardial infarction, stroke or transient ischaemic attack); history of VTE (DVT, PE or both); recent (<3 months) infection; active cancer. Consecutive eligible patients attending the rheumatology outpatient clinics during the study period were screened for enrolment according to the predefined inclusion and exclusion criteria. Patients were enrolled and sampled between March 2024 and January 2025. Consenting healthy volunteers, referred in the same study period, matched for age (±3 years) and sex with the study cases, served as controls.¹⁴

Procedures

Each enrolled participant underwent a blood withdrawal for global coagulation, full blood counts, inflammatory markers and lipids. For each enrolled participant, global coagulation was assessed fasting in the morning (08:00–11:00 hours) by WB ROTEM (Instrumentation Laboratory-Werfen) and WB impedance aggregometry (Multiplate Analyser, Roche Diagnostics) tests. The patients were instructed not to smoke in the 12 hours before blood sampling. To minimise pharmacokinetic variability, the date and timing of the last drug administration were prospectively recorded for all participants. Blood sampling was performed during predefined trough windows whenever feasible (table 1). For patients receiving JAKi, samples were collected approximately 24 hours after the last drug intake, whereas for patients receiving TNFi, blood collection was scheduled immediately before the next planned administration according to the dosing interval of the specific agent. All patients were receiving stable treatment and had reached steady-state exposure at the time of sampling. On the completion of blood collection, an independent joint assessor conducted a 28-joint count examination on each patient and demographics, vitals, cardiovascular risk factors and medical history were all recorded.

Table 1. Sampling window relative to last administration according to each drug exposure.

Treatment class	Agent	Approximate terminal half-life	Sampling window relative to last administration
JAK inhibitor	Tofacitinib	3 hours	Approximately 24 hours after the last dose
JAK inhibitor	Baricitinib	12 hours	Approximately 24 hours after the last dose
JAK inhibitor	Upadacitinib	9–14 hours	Approximately 24 hours after the last dose
JAK inhibitor	Filgotinib	7 hours (active metabolite: 19–27 hours)	Approximately 24 hours after the last dose
TNF-ɑ inhibitor	Etanercept	70–100 hours	Immediately before the next scheduled administration
TNF-ɑ inhibitor	Adalimumab	10–20 days	Within 24–48 hours before the next scheduled administration
TNF-ɑ inhibitor	Certolizumab pegol	14 days	Within 24–48 hours before the next scheduled administration

JAK, Janus kinase; TNFα, tumour necrosis factor-alpha.

Whole-blood laboratory tests

Rotational thromboelastometry (ROTEM) is a viscoelastic WB test that evaluates changes in the viscoelastic strength of blood by applying a constant rotational force and comprehensively assesses the coagulation process from clot formation and stabilisation to lysis. ROTEM analysis was performed according to the standard protocols provided by the manufacturer, as previously reported.¹⁵ The following tests were performed: INTEM (to assess the intrinsic coagulation pathway), EXTEM (to assess the extrinsic coagulation pathway) and FIBTEM (to assess the contribution of fibrinogen to clot formation and stability). For EXTEM and INTEM tests, the following ROTEM parameters were considered: (i) clotting time (CT, s), the elapsed time from the start of coagulation analysis to an increase in amplitude of 2 mm; (ii) MCF (mm), the maximum amplitude achieved in the thromboelastogram. For the FIBTEM assay, MCF was considered. CT reflects the initiation phase of the coagulation process, the shorter the time the greater the coagulation potential of the analysed sample. MCF evaluates clot stiffness. Platelet aggregation was tested with the Multiplate functional analyser (Roche Diagnostics, Mannheim, Germany). Impedance aggregometry is a platelet function test in WB used to diagnose platelet abnormalities, in monitoring antiplatelet therapy and as a potential predictor of the need for transfusions and bleeding risk during major surgical procedures. The instrument continuously measures changes in electrical resistance (called ‘impedance’) between two copper wires. The greater the area, the greater the platelet aggregation. Briefly, 300 μL of WB citrate was added to an equal amount of saline solution, prewarmed to 37°C and platelet aggregation was tested after specific activation with arachidonic acid (ASPI-test), ADP-test and a thrombin analogue (TRAP-test). Platelet aggregation was measured electronically for 6 min and expressed as units of area under the curve (AUC) plotted over time in arbitrary units (U) (1 U=10 AU×min).¹⁶

Outcomes

The primary outcomes were to investigate the effects of JAK or TNFi on global haemostasis using WB rotational thromboelastometry (ROTEM) and impedance aggregometry (Multiplate)

Covariates

Demographics (age, sex), inflammatory markers (erythrocyte sedimentation rate, C reactive protein), disease activity (Simple Disease Activity Index (SDAI) and Disease Activity Score-28) and cardiovascular risk (lipid profile, smoking status, body mass index (BMI), diabetes mellitus) were prospectively assessed and considered as covariates in multivariable analysis.

Statistical analysis

Continuous variables were summarised as mean (SD) or median (IQR) as appropriate. Analysis of variance (ANOVA) with Scheffé test, independent-samples t-tests, Mann-Whitney U test, or Fisher’s exact test were employed to conduct univariate analyses, contingent on the nature of the comparisons. Pearson’s correlation tests were implemented to conduct correlation analyses.

Associations between clinical covariates and outcomes (INTEM MCF, EXTEM MCF, FIBTEM MCF and ADP-test) were investigated using multivariable linear regression models. The primary model was prespecified and included age, BMI, ln-transformed SDAI, treatment group (JAKi vs TNFi) and ln-transformed prednisone dose. Regression coefficients (β) with 95% CIs and two-sided p values were reported. Model fit was quantified using R².

Given the exploratory and mechanistic nature of the study, no formal a priori sample size calculation was performed; enrolment was based on feasibility and availability of eligible participants during the study period.

All analyses were conducted with IBM SPSS Statistics (V.26.0). Two-sided p values <0.05 were considered statistically significant.

Results

Clinical characteristics of the study population

Two patients were excluded because of a history of VTE. Out of 102 blood samples collected, five had abnormal platelet counts and two were lost. Thus, during the study period, we enrolled 61 patients with RA (31 patients receiving TNFi, 30 receiving JAKi) and 34 age-matched and sex-matched healthy controls (HC) (years (SD) 57.60 (10.34), p=0.932 vs JAKi, p=0.361 vs TNFi; females n (SD): 25 (71.4%), p=0.165 for χ² test). All participants had normal standard coagulation tests and platelet counts. The JAKi group had shorter treatment group duration (p=0.001), a higher number of previous failed biological and targeted synthetic disease-modifying antirheumatic drug classes (p<0.001) and greater concomitant use of conventional synthetic disease-modifying antirheumatic drugs (csDMARDs) (p=0.039), compared with the TNFi group. Most patients in both groups (85%) were in remission or low disease activity. No significant differences were observed between treatment groups in disease activity, joint counts, inflammatory markers, lipid profile, current use of non-steroidal anti-inflammatory drugs (NSAIDs) or glucocorticoid dose or cardiovascular risk factors (table 2). All participants had normal activated partial thromboplastin time, prothrombin time, fibrinogen levels and platelet counts.

Table 2. Characteristics of the study population.

Mean (SD%)	JAKi (n=30)	TNFi (n=31)	JAKi versus TNFi p value
Demographics
Age, years (SD)	56.63 (11.02)	61.28 (9.86)	0.223
Female sex, n (%)	27 (90.0%)	23 (74.2%)	0.165
Disease activity
Disease duration, months	181 (107)	161 (108)	0.480
ACPA >7 kU/L, n (%)	19 (63.3%)	16 (51.6%)	0.355
ACPA titre, kU/L	286 (231)	260 (279)	0.799
RF >14 kU/L, n (%)	24 (80.0%)	21 (67.7%)	0.384
RF titre, kU/L	162 (158)	145 (228)	0.778
ESR mm/hour (SD)	22.33 (19.16)	16.81 (13.30)	0.194
CRP, mg/L (SD)	3.52 (4.50)	2.27 (4.05)	0.256
TJC28 (SD)	1.5 (2.3)	2.8 (5.3)	0.210
SJC28 (SD)	0.7 (1.3)	0.9 (1.8)	0.617
DAS28CRP (SD)	2.29 (1.04)	2.53 (1.21)	0.409
SDAI (SD)	8.14 (8.11)	10.10 (10.40)	0.414
Remission	23 (76.7%)	21 (67.7%)	0.900
Low disease activity	3 (10.0%)	4 (12.9%)
Moderate disease activity	3 (10.0%)	4 (12.9%)
High disease activity	1 (3.3%)	2 (6.5%)
Medications
Treatment duration, months (SD)	38.10 (23.15)	80.06 (67.42)	0.002
Prior b/tsDMARD classes, n (SD)	1.53 (1.04)	0.42 (0.50)	<0.001
Prednisone dose, mg/day (SD)	1.23 (2.08)	1.15 (2.10)	0.870
NSAID, n (%)	14 (46.7%)	14 (45.2%)	0.906
Current csDMARDs (%)	20 (66.7%)	11 (35.5%)	0.039
Methotrexate	18 (60.0%)	11 (35.5%)	0.069
Leflunomide	1 (3.3%)	1 (3.2%)
Sulfasalazine	1 (3.3%)	0 (0.0%)
CVD risk profile
Hypercholesterolaemia, n (%)	16 (55.2%)	15 (48.4%)	0.617
Total cholesterol, mg/dL (SD)	205.50 (27.32)	202.15 (33.97)	0.645
LDLc, mg/dL (SD)	117.96 (29.34)	121.74 (35.61)	0.654
People who have ever smoked	7 (23.3%)	7 (22.6%)	0.865
People who have never smoked, n (%)	23 (76.7%)	24 (77.4%)	0.892
People who previously smoked, n (%)	4 (13.3%)	3 (9.7%)
People who smoke, n (%)	3 (10.0%)	4 (12.9%)
Diabetes mellitus, n (%)	1 (3.3%)	3 (9.7%)	0.612
Prior CVD, n (%)	4 (13.3%)	5 (16.1%)	0.758
BMI, kg/m2 (SD)	24.60 (3.80)	25.08 (4.47)	0.650
Hypertension, n (%)	8 (26.7%)	12 (38.7%)	0.316

P values refer to Mann-Whitney U test for non-parametric continuous variables and Fisher’s exact test for categorical variables.

ACPA, anticitrullinated protein antibodies; BMI, body mass index; b/tsDMARDs, biological and targeted synthetic disease-modifying antirheumatic drugs; CRP, C reactive protein; csDMARDs, conventional synthetic DMARDs; CVD, cardiovascular disease; DAS28CRP, Disease Activity Score-28 with CRP; ESR, erythrocyte sedimentation rate; JAKi, Janus kinase inhibitors; LDLc, low-density lipoprotein cholesterol; NSAID, non-steroidal anti-inflammatory drug; RA, rheumatoid arthritis; RF, rheumatoid factor; SDAI, Simple Disease Activity Index; SJC28, 28-joint swollen joint count; TJC28, 28-joint tender joint count; TNFi, tumour necrosis factor inhibitors.

RA increases ROTEM and aggregometry

Patients with RA (n=61) exhibited a prothrombotic ROTEM profile (INTEM MCF 63.4±4.0 vs 57.2±4.1 mm; EXTEM MCF 66.2±4.1 vs 55.9±4.7 mm; FIBTEM MCF 15.7±3.6 vs 12.4±3.7 mm; all comparisons p<0.001 vs HC), and enhanced platelet aggregation (ADP-test 81.1±24.4 vs 57.8±17.0 AUC; ASPI-test 71.9±23.1 vs 52.0±18.9 AUC; TRAP-test 117.8±25.4 vs 89.9±21.5 AUC; all comparisons p<0.001 vs HC) as compared with HC (figure 2). In univariable analysis (ANOVA), there were no significant differences in ROTEM and aggregometry parameters between JAKi and TNFi groups (figure 2).

Figure 2. Comparisons of ROTEM MCF and aggregometry tests between HC, TNFi and JAKi patients. In the upper panels, comparisons of INTEM, EXTEM and FIBTEM tests between HC, TNFi and JAKi patients are illustrated. In the lower panels, comparisons of aggregometry tests (ADP-test, ASPI-test, and TRAP-test) between HC, TNFi and JAKi patients are shown. ASPI-test, arachidonic acid-induced platelet aggregation test; EXTEM, extrinsically activated rotational thromboelastometry assay; FIBTEM, fibrin-based extrinsically activated rotational thromboelastometry assay; HC, healthy controls; INTEM, intrinsically activated rotational thromboelastometry assay; JAKi, Janus kinase inhibitors; MCF, maximum clot firmness; ROTEM, rotational thromboelastometry; TNFi, tumour necrosis factor inhibitors; TRAP-test, thrombin receptor-activating peptide-induced platelet aggregation test.

Correlation between inflammation and disease severity with ROTEM and aggregometry

Within the entire RA cohort (n=61), significant correlations (r=0.2–0.5, weak to moderate strength; figure 3) emerged for MCF values with joint counts, ESR, age, prednisone dose and BMI and for aggregometry tests with joint counts, disease activity and prednisone dose. Use of NSAIDs, csDMARD, methotrexate, time in treatment or disease duration were not correlated with any VET parameters. To mitigate overfitting and address collinearity across the extensive clinical, inflammatory and metabolic predictor set, multivariable models were constructed using a data-driven selection strategy. The final selection yielded the following predictors: age, BMI, ln-transformed SDAI, treatment group (JAKi vs TNFi) and ln-transformed prednisone dose.

Figure 3. Significant correlations between ROTEM, aggregometry and patient and disease characteristics. ASPI-test, arachidonic acid-induced platelet aggregation test; BMI, body mass index; CRP, C-reactive protein; ESR, erythrocyte sedimentation rate; EXTEM, extrinsically activated rotational thromboelastometry assay; FIBTEM, fibrin-based extrinsically activated rotational thromboelastometry assay; INTEM, intrinsically activated rotational thromboelastometry assay; MCF, maximum clot firmness; PDN, prednisone; ROTEM, rotational thromboelastometry; SDAI, Simplified Disease Activity Index; SJC28, 28-joint swollen joint count; TJC28, 28-joint tender joint count; TRAP-test, thrombin receptor-activating peptide-induced platelet aggregation test.

Multivariable analysis

In the multivariable model, prednisone dose was positively associated with INTEM MCF (β=1.54, 95% CI 0.24 to 2.84, p=0.021), while BMI showed a moderate positive effect (β=0.24, 95% CI 0.01 to 0.48, p=0.044). Age, disease activity and treatment group were not independently associated after adjustment (figure 4A).

Figure 4. Multivariable analysis of ROTEM MCF and ADP-test. In the upper left panel, INTEM MCF as an outcome; in the upper right panel, EXTEM MCF; in the lower left panel, FIBTEM MCF; and in the lower right panel, ADP-TEST. BMI, body mass index; EXTEM, extrinsically activated rotational thromboelastometry assay; FIBTEM, fibrin-based extrinsically activated rotational thromboelastometry assay; INTEM, intrinsically activated rotational thromboelastometry assay; JAKi, Janus kinase inhibitors; MCF, maximum clot firmness; PDN, prednisone; ROTEM, rotational thromboelastometry; SDAI, Simplified Disease Activity Index; TNFi, tumour necrosis factor inhibitors.

Treatment with JAKi (vs TNFi) was independently associated with higher EXTEM MCF values in adjusted analyses (β=2.32, 95% CI 0.42 to 4.23, p=0.018). Prednisone dose (β=1.33, 95% CI 0.02 to 2.63, p=0.046) also showed positive associations of smaller magnitude (figure 4B).

Age was strongly associated with FIBTEM MCF (β=0.17, 95% CI 0.09 to 0.25, p<0.001), as well as treatment with JAKi (vs TNFi; β=1.72, 95% CI 0.07 to 3.37, p=0.041), after multivariable adjustment (figure 4C).

Disease activity (SDAI) was strongly associated with ADP-test values in multivariable models (β=11.34, 95% CI 4.31 to 18.37, p=0.002). In contrast, treatment group, BMI, age and prednisone dose were not consistently associated with ADP-test after adjustment (figure 4D).

Sensitivity analysis

Given the higher frequency of concomitant csDMARD use among patients receiving JAKi, a sensitivity analysis was performed to evaluate its potential confounding effect. Concomitant csDMARD therapy was not associated with EXTEM MCF (β=0.03, p=0.978), FIBTEM MCF (β=−0.38, p=0.663) or INTEM MCF (β=0.26, p=0.798). Furthermore, inclusion of csDMARD use as a covariate in the multivariable models did not materially modify the association between JAKi treatment and clot firmness parameters. JAKi exposure remained independently associated with higher EXTEM MCF (β=2.32, 95% CI 0.32 to 4.31, p=0.024) and FIBTEM MCF (β=1.83, 95% CI 0.10 to 3.55, p=0.039), supporting the robustness of the primary findings. Concomitant csDMARD therapy was not associated with platelet reactivity as assessed by ADP-test, either in unadjusted analyses (p=0.414) or after multivariable adjustment (p=0.289). In contrast, disease activity remained the main independent determinant of ADP-induced platelet aggregation.

Time from the last drug administration to blood sampling was explored as a continuous variable using Pearson’s correlation analyses for JAKi-treated patients and according to the interval before the next scheduled administration for TNFi-treated patients. No significant associations were observed between sampling time and any ROTEM or impedance aggregometry parameter (all p>0.10), suggesting that the observed haemostatic differences were unlikely to be driven by short-term pharmacokinetic variability.

Discussion

This study expands the prothrombotic landscape of patients with RA undergoing treatment with JAKi and TNFi. Our primary findings add further evidence that patients with RA exhibit a significant prothrombotic state and enhanced platelet reactivity compared with HC, regardless of the treatment class. However, the most critical finding is the identification of distinct prothrombotic signatures by WB coagulopathy testing associated with JAKi versus TNFi therapy independent of disease activity control. Specifically, active treatment with JAKi was independently associated with a more pronounced prothrombotic profile, characterised by increased EXTEM MCF and FIBTEM MCF, compared with TNFi therapy.

An important finding of our study is that patients with RA exhibited a more procoagulant haemostatic profile than HC irrespective of treatment class. This observation was consistently reflected by increased clot firmness and altered platelet function parameters, supporting the concept that thrombo-inflammatory activation remains detectable even in the era of targeted therapies. These findings are consistent with epidemiological data demonstrating an increased risk of both arterial and venous thrombotic events in RA and suggest that normalisation of clinical disease activity does not necessarily translate into complete restoration of haemostatic homeostasis. However, this prothrombotic phenotype is likely multifactorial. In addition to persistent inflammatory activation, contributing factors may include glucocorticoid exposure, age-related changes, cardiovascular comorbidities, metabolic risk factors, reduced physical activity and treatment-related effects. Therefore, the observed differences between patients with RA and HC should be interpreted as the result of complex interactions between inflammation, comorbidity burden and therapeutic exposures rather than a direct consequence of RA alone.

RA is a chronic systemic inflammatory disease associated with excess CVD morbidity and mortality, largely attributed to accelerated atherothrombosis^{4 17 18} and inflammation-driven prothrombotic pathways.¹⁹ Systemic inflammation enhances tissue factor expression, amplifies thrombin generation, increases fibrinogen synthesis and suppresses fibrinolysis.²⁰ Potent anti-inflammatory therapy is believed to revert the prothrombotic imbalance of RA. For instance, a 24-week study compared the effects of bDMARDs (certolizumab pegol, abatacept, tocilizumab) and csDMARDs improved prothrombotic biomarkers while reducing disease activity, but biological agents, particularly the IL-6 antagonist tocilizumab, were most effective in normalising coagulation and fibrinolysis markers, suggesting a significant reduction in RA-associated procoagulant burden due to their potent anti-inflammatory effects.⁸

Mechanistically, JAK-STAT signalling influences cytokine networks central to coagulation activation, including IL-6²¹ and interferon pathways. Recent experimental work provides a mechanistic framework for the procoagulant signatures observed with JAKi. In patients with RA who are actively treated with DMARDs, JAK inhibition (including tofacitinib, baricitinib and upadacitinib) increased toll-like receptor 4-driven innate immune activation in vitro and accelerated tissue factor-dependent clot formation, a process that is consistent with immunothrombosis.²² In our study, patients receiving JAKi exhibited higher EXTEM and FIBTEM MCF on WB rotational thromboelastometry, indicating increased fibrin-dependent clot strength. These viscoelastic findings are biologically consistent with enhanced thrombo-inflammatory activation and suggest that JAK inhibition may influence haemostatic balance beyond inflammation control alone. Notably, this mechanistic and laboratory evidence aligns with clinical safety signals from randomised trials and observational cohorts reporting increased VTE risk in selected RA populations treated with tofacitinib^{5 6 23} and baricitinib.^{23 24} Unfortunately, the low number of patients treated with JAKi included in this study prevented us from stratifying based on their type.

While our study was not designed to evaluate clinical thrombotic outcomes, the viscoelastic differences observed may represent intermediate biological signals linking targeted therapies to haemostatic modulation. VET provides a dynamic, WB assessment integrating platelet function, fibrin polymerisation and thrombin activity.¹⁰ ROTEM-derived MCF, in particular, reflects clot strength determined by platelet-fibrin interaction and fibrinogen concentration.¹⁰

The association between inflammatory states and increased MCF has been described in infectious and inflammatory conditions, including periprosthetic joint infection and sepsis, where EXTEM MCF correlates with hypercoagulability and impaired fibrinolysis.^25–27 In a diagnostic study of periprosthetic joint infection, EXTEM MCF demonstrated high discriminatory accuracy and outperformed D-dimer.²⁸ Similarly, prothrombotic profiles detected by ROTEM have been observed in inflammatory and neoplastic states.^{12 13 16 29 30} Our findings extend this evidence to RA, suggesting that chronic immune activation may produce measurable viscoelastic changes even outside acute infection.

Recent experimental evidence suggests that JAK inhibition may enhance thromboxane A₂ biosynthesis and thrombo-inflammatory activation in stimulated WB models.³¹ However, these studies assessed innate immune-driven thromboxane A₂ (TXA₂) production following lipopolysaccharide stimulation rather than ADP-dependent platelet aggregation pathways. The absence of an independent association between JAKi exposure and ADP-test values in our cohort may therefore reflect differences in the biological pathways assessed, variability in pharmacodynamic exposure at the time of sampling or limited statistical power. Unfortunately, TXA₂ biosynthesis and platelet-leucocyte interactions were not directly evaluated in our study.

Notably, glucocorticoid exposure emerged as a consistent determinant of clot firmness in our models, particularly for INTEM and EXTEM MCF. This finding aligns with the known prothrombotic effects of glucocorticoids on haemostasis. Experimental and clinical studies have shown that glucocorticoids can rapidly enhance thrombin generation, increase circulating levels of factor VIII and von Willebrand factor and impair fibrinolysis through upregulation of plasminogen activator inhibitor-1. These effects may preferentially amplify intrinsic pathway-related and extrinsic pathway-related clot propagation and stabilisation, which is consistent with the observed associations with INTEM and EXTEM clot firmness in our cohort. Importantly, these haemostatic effects may occur independently of inflammatory control and may partly contribute to the dose-dependent cardiovascular and thrombotic risk associated with glucocorticoid therapy in RA. Our findings therefore reinforce the concept that even low-dose glucocorticoid exposure may exert measurable effects on global WB coagulation dynamics, providing a potential mechanistic explanation for the dose-dependent CVD risk previously reported in RA cohorts.^32–34

Finally, ADP-dependent platelet reactivity was primarily associated with disease activity. As active RA is characterised by increased circulating activated platelets and platelet hyper-reactivity, increased soluble P-selectin and enhanced platelet-leucocyte interactions,^35–39 our data suggest that platelet-driven viscoelastic changes parallel inflammatory burden, reinforcing the interplay between immune activation and thrombosis.

Study strengths and limitations

This study has limitations. Its cross-sectional design precludes causal inference, and viscoelastic changes cannot be directly equated with clinical thrombotic risk. Residual confounding by unmeasured cardiovascular risk factors remains possible. Furthermore, although LDL and metabolic variables were considered in exploratory analyses, lipid-inflammation interactions in RA are complex and may require longitudinal modelling. Sensitivity analyses including concomitant csDMARD use did not materially alter the association between JAKi exposure and increased EXTEM or FIBTEM MCF, and csDMARD therapy itself was not associated with these viscoelastic parameters. Thus, although residual confounding cannot be completely excluded, concomitant csDMARD use is unlikely to explain the observed treatment-associated clot firmness signal. Patients receiving JAKi exhibited a more treatment-refractory profile, including longer disease duration and higher prior b/tsDMARD exposure. Although disease duration and treatment duration were not associated with viscoelastic parameters in exploratory analyses, treatment-selection bias and residual confounding remain possible and should be considered when interpreting these findings. Although individual JAK inhibitors may exert differential haemostatic and cardiovascular effects due to differences in selectivity and pharmacodynamic profiles, the limited sample size precluded adequately powered agent-specific analyses. Nevertheless, the persistence of the association between JAKi exposure and increased clot firmness despite within-class heterogeneity suggests a biologically consistent prothrombotic signal at the class level. Finally, clinical thrombotic outcomes were not assessed, and future studies should integrate ROTEM profiles with prospective cardiovascular event data.

Conclusion

In conclusion, RA is characterised by distinct viscoelastic signatures reflecting the combined effects of inflammation, age and immunomodulatory therapy. Glucocorticoid exposure and inflammatory burden appear to exert measurable effects on clot firmness and platelet-dependent parameters. Viscoelastic testing may therefore represent a valuable mechanistic tool to investigate the prothrombotic milieu in RA and to explore therapy-specific haemostatic modulation. Prospective longitudinal studies evaluating haemostatic changes before and after treatment initiation will be necessary to clarify causality and temporal dynamics of therapy-specific coagulation modulation.

Transparency declaration

The lead author (the manuscript’s guarantor) affirms that the manuscript is an honest, accurate and transparent account of the study being reported; that no important aspects of the study have been omitted and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.