Article Text
PDF
Abstract
Background Cardiovascular disease is important in rheumatoid arthritis (RA). Tissue factor (TF) is expressed upon platelet activation and initiates coagulation. Anti-tumour necrosis factor-α (TNFα) agents seem to decrease RA-associated cardiovascular events. We investigated whether (1) TNFα activates human platelets and (2) TNFα pharmacological blockade modulates the platelet-leucocyte reciprocal activation in RA.
Design The expression of platelet TNFα receptors has been assessed by flow cytometry and immunogold electron microscopy. Platelet and leucocyte activation has been assessed also in the presence of antibodies against the TNFα receptors 1 and 2 and of infliximab. TF expression, binding to fibrinogen and phosphatidylserine exposure, has been assessed by flow cytometry, TF activity by coagulation time and by endogenous thrombin generation. Markers of platelet and leucocyte activation have been assessed in 161 subjects: 42 patients with RA, 12 with osteoarthritis, 37 age-matched and sex-matched patients with chronic stable angina and 70 age-matched and sex-matched healthy subjects.
Results TNFα elicited the platelet activation and the expression of TF, which in turn prompted thrombin generation and clot formation. Inhibition of the TNFα-induced activation restricted platelet ability to activate leucocytes and to induce leucocyte TF. TNFα inhibition did not influence platelet activation induced by collagen, ADP or thrombin receptor activating peptide-6. Platelets of patients with RA were more activated than those of controls. Activation was reduced in patients treated with TNFα inhibitors.
Conclusions TNFα-dependent pathways control platelet activation and TF expression in RA. Further studies will verify whether the protective effect of TNFα inhibitors on cardiovascular events involves their ability to modulate platelet function.
- Anti-TNF
- Rheumatoid Arthritis
- TNF-alpha
- Cardiovascular Disease
- Atherosclerosis
Statistics from Altmetric.com
Introduction
Cardiovascular disease is a major cause of mortality in patients with rheumatoid arthritis (RA) and accelerated atherosclerosis, and its complications are now considered an extra-articular disease-specific feature.1–4 Inflammation is a key event in atherosclerosis and elevated cardiovascular risk possibly begins in the early stages of RA, with traditional risk factors contributing to further worsen atherosclerosis and its complications. Glucocorticoids increase cardiovascular risk, while a reduction in the incidence of cardiovascular events occurs in patients treated with methotrexate,5–8 even if cardiovascular risk reduction in patients treated with methotrexate and other traditional disease-modifying antirheumatic drugs (DMARDs) has not been confirmed in all studies.9
In recent years, cytokine-specific antagonists have been developed. Anti-tumour necrosis factor-α (TNFα) antagonists are potent anti-inflammatory drugs widely used in patients with RA. Their ability to modulate cardiovascular risk has been evaluated in cohort studies, some of which indicated significant risk reduction in patients with established RA.10 ,11 A recent meta-analysis of the studies confirmed that anti-TNFα agents have protective effects on the cardiovascular morbidity since treated patients had a reduced risk of all cardiovascular events, as well as of strokes, of myocardial infarction and of important adverse cardiac events. In contrast, no effect on heart failure was observed.12 The pathway(s) involved in the protective action of anti-inflammatory agents in RA have not been so far unambiguously identified.
Tissue factor (TF) is the primary cellular initiator of the coagulation cascade. After vessel injury, the TF-FVIIa complex triggers a cascade of protease reactions to form thrombin, eventually leading to fibrin formation and deposition. This mechanism contributes to thrombosis associated with atherosclerosis, with cancer and with antiphospholipid syndrome and other autoimmune diseases.13 ,14 Inflammation is a common substrate of these conditions and plays a crucial role in thrombogenesis and in the persistence of vascular inflammation. Neutrophils express TF in response to formylated peptides14 ,15 and TNFα14 ,16 ,17 and monocytes in response to lipopolysaccharide.15 ,18 Platelets express TF in response to virtually all agonists19 ,20 and in turn amplify the expression of leucocyte TF via the expression of P-selectin and the interaction with PSGL-1, its counter-receptor on neutrophils and monocytes.14 ,15 ,21
Platelets are crucial effectors of haemostasis, coagulation and tissue repair. They play a major role in RA: platelet aggregates are often present in the synovial vasculature of patients with RA and activated platelets and their by-products accumulate in the blood and the synovial fluid.22 ,23 Biologically active inflammatory microparticles released as a consequence of platelet activation accumulate in the joints of patients with RA where they activate fibroblastoid synoviocytes, which in turn secrete chemokines and cytokines and sustain synovitis and bone erosion.24 The event might be important in the natural history of the disease since the depletion of platelets quenches the severity of experimental murine arthritis.24 Platelets prostaglandins and serotonin contribute to the persistent inflammation and vascular permeability that characterises RA synovial vessels.25 ,26 Thus, platelet activation is an early and possibly integral component of the early vascular inflammation that is a hallmark of RA and a counterpart of the cardiovascular risk. The mechanistic events that lead to platelet activation in RA, including the early stimuli that are responsible for platelet activation, have not been extensively characterised. The protective action of anti-TNFα agents on both the articular and cardiovascular involvement of patients with RA raises the possibility that TNFα is involved.
In the present study, we have tested the possibility that human TNFα behaves as a platelet agonist, triggering their activation and consequently affecting the activation state and possibly the function of autologous leucocytes. We also evaluated the effects of treatment with anti-TNFα agents on the degree of activation and expression of TF by platelets and leucocytes to verify the possibility that endogenous TNFα supports the mutual activation of platelets and leucocytes of patients with RA.
Methods
Patients
The study group consisted of 42 consecutive patients with RA, evaluated at the San Raffaele University Hospital, Milano, during a 24-month period. They satisfied both the 1987 American College of Rheumatology (ACR) criteria27 and the 2010 ACR/European League Against Rheumatism criteria.28 Three patients were being treated with aspirin (100 mg die). Thirty-nine patients were not being treated with antiplatelet agents, thienopyridines. The disease activity was assessed via four variables disease activity score on 28 joints.29 Eleven patients with RA were studied before the onset of treatment (naive), 18 received methotrexate either alone (12) or in combination with DMARDs (6) and 12 patients received anti-TNFα agents. Three patients received the anti-TNFα agents alone, seven in combination with methotrexate and two in combination with DMARDs. Individual characteristics of patients with RA are summarised in the online supplementary table S1. In total, 70 healthy volunteers, 12 patients with osteoarthritis (OA) (studied immediately before hip or knee replacement and in the absence of anti-inflammatory treatment for at least three days) and 34 patients with chronic stable angina (CSA), defined as effort angina lasting >3 months with angiographic evidence of coronary artery stenosis (stenosis >50% diameter) in the absence of history of unstable angina or myocardial infarction,30 served as controls. Acute coronary syndromes, other autoimmune diseases, type 1 diabetes mellitus, neoplastic diseases, chronic or acute infections were used as exclusion criteria. Patients’ demographic characteristics are summarised in table 1.
Patients and controls characteristics
Reagents and monoclonal antibodies
Annexin V, monoclonal antibodies (mAbs) against human fibrinogen (clone HYB51-04) were from The Antibody Shop (Italy). Thrombofix, mAbs against CD14 (clone RMO52), CD18 (clone 7E4), CD45 (clone J33), CD61 (clone SZ21), CD66b (clone 80H3), CD62P (P-selectin; clone Thromb-6), anti CD40L (CD154, clone TRAP1), myeloperoxidase (MPO, clone CLB-MPO-1), von Willebrand factor (vWF, clone 4F9) and relevant isotype control antibodies were obtained from Instrumentation Laboratory (Italy). PGE1, ADP and thrombin receptor activating peptide (TRAP-6) were obtained from Sigma (Italy). Recombinant human TNFα, FACs Lysing Solution, mAbs against TF (clone HT142) and its isotype control were obtained from Beckton Dickinson. Goat anti-human TF polyclonal Ab (catalogue no. 4501) for thrombin generation assay was from American Diagnostica, mAb against myeloid-related protein (MRP) 8/14 was obtained from Acris (Star Fish, Italy). Collagen was obtained from Horm (Cabru Italy). mAbs against TNFα receptor I (NBP2-11944) and against TNFα receptor II (NBP2-11945) were obtained from Novus Biological (DBA Italy). The chimeric monoclonal antibody against TNFα, which is thought to function in RA by hindering the binding of TNFα to its receptors (infliximab, Remicade), was obtained from Merck Sharp & Dohme. Zenon IgG Labeling kits (488 or 546) were obtained from Invitrogen. Fix & Perm kit was obtained from Caltag. All reagents for thrombin generation assay were obtained from Stago. The mAb against pentraxin 3 (PTX3) (clone IC8) was kindly provided by Barbara Bottazzi and Alberto Mantovani (IRCCS Humanitas, Rozzano Italy).
Blood sampling and processing
Venous blood was drawn through a 19-gauge butterfly needle. After having discarded the first 3–5 mL of blood, 1.8 mL were carefully collected in tubes containing 0.2 mL Na citrate and a cocktail containing antiproteases to limit in vitro cellular activation as much as possible.30 ,31 Blood samples were immediately fixed with equal volumes of Thrombofix and analysed on a FC500 flow cytometer (Beckman Coulter, Milan, Italy). The activation of platelets and leucocytes and the proportion of platelet-neutrophil heterotypic aggregates were assessed by multiparametric flow cytometry evaluating platelet P-selectin, TF, fibrinogen binding and vWF content and leucocyte MPO and PTX3 content, the expression of MRP8.14, TF, CD40L, CD18 and fibrinogen binding. The interassay coefficient of variations was consistently <8%.30 ,31
Platelet activation
Whole blood samples anticoagulated with Na citrate were centrifuged at 150 g 10 min at 20°C to obtain platelet-rich plasma (PRP). PRP aliquots were stimulated with recombinant human TNFα (R&D: 10–200 ng/mL), ADP (5 µM), collagen (1 µg/mL) or TRAP-6 (25 µM) for 5 min at 37°C in the presence or the absence of infliximab (source, final concentration 250 µg/mL) added 5 min before the agonists. Reactions were stopped by addition of Thrombofix. The expression of P-selectin, TF, binding of Annexin V, CD40L and the content of vWF were determined by flow cytometry as described.30 ,31
Platelets, isolated as described30 ,31 and resuspended in CaCl2-HEPES-Tyrode buffer (pH 7.4) (200 000/μL), were incubated with TNFα (100 ng/mL), ADP (5 µM), collagen (1 µg/mL) or TRAP-6 (25 µM) for 5 min at 37°C and added to autologous platelet-depleted blood samples for 5 min at 37°C. Reactions were stopped by addition of Thrombofix. The formation of platelet-leucocyte heterotypic aggregates, the expression of neutrophil MPO, TF and fibrinogen and monocyte TF and fibrinogen were determined by flow cytometry. When indicated, infliximab (final concentration 250 µg/mL) was added 5 min before platelet stimulation. Reactions were stopped by addition of Thrombofix. Platelet activation was determined by flow cytometry as above. When indicated, whole blood samples aliquots anticoagulated with Na were directly stimulated with recombinant human TNFα (10–200 ng/mL), ADP (5 µM), collagen (1 µg/mL) or TRAP-6 (25 µM) for 5 min at 37°C in the presence or the absence of infliximab (250 µg/mL) added 5 min before the agonists. Reactions were stopped by addition of Thrombofix.
Platelet procoagulant activity
Blood samples were centrifuged at 150 g for 10 min to obtain PRP, which was immediately placed in coagulometer cuvettes and incubated with TNFα (10–200 ng/mL) or collagen (1 µg/mL) for 5 min at 37°C before addition of CaCl2 (100 mM). When indicated, infliximab (final concentration 250 µg/mL) was added 5 min before platelet stimulation. The coagulation time was determined in a semiautomated coagulometer (Option 2, BioMerieux; Marcy l'Etoile, France). The intraassay coefficient of variations was consistently<10%.
Statistical analysis
Results were reported as mean±SEM, unless otherwise indicated. The normal distribution of each continuous variable was assessed with the use of the Kolmogorov–Smirnov test. All patients’ data sets were found to be normally distributed. Statistical analysis was performed by one-way analysis of variance, followed by Bonferroni multiple comparison test to compare all pairs of groups. We assumed a two-tailed p<0.05 after Bonferroni correction for multiple testing as statistically significant. Statistical analyses were performed by GraphPad Prism 6.00.
Results
TNFα induces platelet activation
Treatment of PRP with recombinant human TNFα resulted in platelet activation, as demonstrated by the increased fraction of platelets that express P-selectin on the plasma membrane (see figure 1A and online supplementary table S2). The effect of the cytokine was clearly dose-dependent, reaching a plateau between 50 and 100 ng/mL, and time-dependent (figure 1A, B). Consensually to P-selectin expression, the treatment with TNFα resulted in the reduction of the intracellular content of vWF (see online supplementary table S2), indicating that the α granules are depleted. Moreover, it prompted the expression of platelet CD40L (CD154) and the binding of fibrinogen to platelets (see online supplementary table S2), a marker of platelet GPIIbIIIa activation. Immunogold electron microscopy indicates that human platelets express the TNFα receptors, TNFαR1 and TNFαR2, and reveals that TNFαR1 was preferentially associated to the membrane of the α granules and TNFαR2 to the cell membrane (figure 1C, D). Analysis by flow cytometry confirms the mutually exclusive distribution of the two moieties, with the TNFαR2 mostly expressed at the cell membrane and the TNFαR1 at intracellular domains (figure 1E, F).
Tumour necrosis factor-α (TNFα) induces platelet activation. Purified human platelets (200 000/μL) were treated with increasing concentrations of TNFα (x axis) for 5 min at 37°C. (A) Activation was assessed after treatment by flow cytometry verifying the fraction of platelets expressing P-selectin (y axis). Grey symbols indicate the platelet response of each donor. Red symbols depict the means of five independent experiments, carried out using platelets purified from different healthy donors. (B) Platelet activation peaked at around 3 min after challenge with the cytokine administration. Grey symbols indicate the platelet response of each donor. Red symbols depict the means of five independent experiments, carried out using platelets purified from different healthy donors. (C and D) TNFα receptors 1 and 2 (TNFα-R1 and TNFα-R2) were consistently preferentially expressed in association to the membrane of the platelets α-granules and to the plasma membrane, respectively, as assessed by immunogold electron microscopy. (E and F) Preferential association of TNFα-R1 and TNFα-R2 to intracellular domains or to the plasma membrane has been confirmed by flow cytometry. Black lines corresponded to fluorescence associated to irrelevant (isotypic) mAb, red to extracellular and blue to intracellular lining associated to TNFα receptors.
The effect of TNFα on platelet activation was specific since it was blocked by pretreatment with infliximab, a chimeric monoclonal antibody against TNFα, which is thought to function in RA by hindering the binding of TNFα to its receptors (figure 2A). The magnitude of the response of platelets to TNFα did not substantially differ from the response to the classical agonists, TRAP-6, collagen and ADP (see figures 2A and 3A, B and online supplementary table S2).
Tumour necrosis factor-α (TNFα) specifically upregulates the ability of platelets to interact with autologous neutrophils and monocytes and results in the expression of platelet and leucocyte tissue factor (TF). Platelet-rich plasma samples were treated with TNFα (100 ng/mL) and with the classical agonists thrombin receptor activating peptide (TRAP)-6 (25 µM), collagen (1 µg/mL), ADP (5 µM) for 5 min at 37°C in the absence (open symbols) or the presence of infliximab (red symbols) or of the platelet inhibitor PGE1 (blue symbols) and either analysed immediately by flow cytometry or added back to autologous platelet-depleted blood. The fraction of platelets expressing P-selectin (y axis, A), TF (y axis, B), the presence of heterotypic platelet-neutrophil and platelet-monocyte aggregates (y axis, C and E), the fraction of neutrophils expressing TF (y axis, D) and the fraction of monocytes expressing TF (y axis, F) have been assessed. Symbols depict the individual observations and lines the means±SEM. *Significantly different from platelets treated with infliximab, p<0.01. Correlation analysis indicates the association between neutrophil TF expression and platelet activation: neutrophil TF membrane expression (G and H, x axis) is plotted against platelet P-selectin expression (r=0.66, p<0.0001) and the fraction of platelet neutrophil aggregates (r=0.72, p<0.0001).
Tumour necrosis factor-α (TNFα) specifically induce the pro-coagulant action of platelets. Platelets were treated with TNFα (100 ng/mL) and with thrombin receptor activating peptide (TRAP)-6 (25 µM), collagen (1 µg/mL) and ADP (5 µM) for 5 min at 37°C. The expression of anionic phospholipids has been assessed using FITC-labelled Annexin V (A) and the tissue factor (TF) expression (B) were determined by flow cytometry. Symbols depict the individual observations and lines the mean. (C) Assessment of the thrombin generation capacity of platelets was carried out by the calibrated automated thrombogram (CAT) assay. Curves generated in a representative experiment in which platelets (106) were either left untreated (grey line) or stimulated for 5 min at 37°C with TNFα (200 ng/mL) in the absence (black line) or in the presence of 40 µg/mL anti TNFα R1 and R2 mAb (red line) or 100 µg/mL neutralising anti-TF antibody (blue line). (D) The endogenous thrombin potential (ETP) as assessed in the experimental conditions described in (C) is shown. Symbols depict the individual observations and lines the mean (n=5). *Significantly different from results observed with TNFα stimulated platelets, ***p<0.0003, ****p<0.0001. (E) The clotting times (y axis) of recalcified plasma were assessed to compare the procoagulant activity of platelets either untreated (black symbols) or treated with TNFα (100 ng/mL) or with collagen (1 µg/mL) (open symbols) in the presence or the absence of monoclonal antibodies recognising the TNFαR1 (orange symbols) and the TNFαR2 (green symbols), infliximab (red symbols) or PGE1 (blue symbols). Symbols depict the individual observations and lines the mean. Significantly different from the clotting time observed in basal conditions, ****p<0.0001.
TNFα promotes via P-selectin the ability of platelets to adhere to and to activate leucocytes
Figure 2 and online supplementary table S2 indicate that TNFα-treated platelets form heterotypic aggregates and activate neutrophils and monocytes when added back to platelet-depleted autologous blood samples. Neutrophils and monocytes challenged with autologous TNFα-treated platelets bound to fibrinogen and expressed TF on the cell membrane while neutrophils released MPO from the primary granules. Similar results were obtained using platelet activated with other agonists (TRAP-6, collagen or ADP).
Treatment with infliximab abrogated the ability of TNFα-treated platelets to express P-selectin, to form heterotypic aggregates, to express TF and to elicit TF expression on the membrane of neutrophils. In contrast, platelets activated with other agonists were not influenced by the presence of TNFα blockers. Complete platelet inhibition by PGE1 abrogated the formation of heterotypic aggregates, the cellular activation and the TF upregulation regardless of the agonist used to activate platelets (figure 2A–F). The extent of platelet P-selectin expression and the platelet-neutrophils aggregates correlated with the neutrophil TF expression (figure 2G,H).
The treatment with TNFα also caused the expression of anionic phospholipids, as assessed by flow cytometry evaluating the binding of the phosphatidylserine-binding moiety, Annexin V (figure 3A) and the expression of TF (see figure 3B and online supplementary figure S1), that is, events specifically associated to the platelet pro-coagulant activity. To verify whether platelet TF is functionally active, we assessed the ability of activated platelets to generate thrombin and to accelerate the clotting process in vitro. The overall capacity of platelets to generate thrombin (endogenous thrombin potential) is significantly higher upon TNFα stimulation (p<0.003, see figure 3C,D and supplementary figure S1), as assessed using the calibrated automated thrombogram assay. Thrombin increase was blunted by mAbs against TNFαR1 and TNFαR2 and reverted when the assay was performed in the presence of a neutralising antibody against TF, confirming the contribution of platelet TF to the kinetics of the event. TNFα-treated platelets reduced the clotting time from 117.7±10.7 to 46.3±1.7 s (p<0.0001). Clotting acceleration was prevented using blocking mAbs against TNFαR1 and TNFαR2 and using infliximab (figure 3E), further confirming that the effect of TNFα is specific. The kinetics of the response and the functional inhibition obtained with specific mAbs (see above) are consistent with a direct interaction of TNFα with platelet receptors.
Platelets and leucocytes activation in RA
The results obtained ex vivo with blood of healthy subjects suggest that a TNFα-dependent pathway might be active in conditions in which the cytokine is persistently generated, prompting the activation and the pro-coagulant activity of circulating platelets. In turn, activated platelets could contribute to amplify and maintain leucocyte activation and inflammatory action. We thus characterised platelets and leucocytes of patients with RA, the prototypic chronic rheumatologic disease in which TNFα plays a pivotal role and specific inhibitors are routinely used. Age-matched and sex-matched subjects served as controls. Moreover, since subjects with RA have a higher atherosclerotic burden than the general population, patients with CSA, candidates to percutaneous coronary intervention revascularisation, were also studied in parallel. Patients with OA studied immediately before hip replacement served to verify the effect of joint inflammation in the absence of rheumatoid synovitis and of autoimmunity.
All groups of patients (RA, CSA and OA) were similar in terms of traditional risk factors for atherosclerosis and of age. Characteristics of the groups studied are reported in table 1 and online supplementary table S1. Patients with RA as expected had higher erythrocyte sedimentation rate and C-reactive protein levels than control subjects, reflecting systemic inflammation. Platelets of patients with RA had lower vWF granular content and a substantially higher fraction of P-selectin expressing platelets than platelets of patients with CSA and OA and of healthy donors (figure 4 and table 2). The fraction of circulating neutrophils and monocytes with adherent platelets (platelet-leucocyte heterotypic aggregates), a more sensitive marker of platelet activation than the expression of platelet P-selectin itself,32 was higher in patients with RA than in control subjects (figure 4B,C). Platelet-neutrophils and platelet-monocytes heterotypic aggregates directly correlate with the extent of platelet P-selectin expression (figure 4G). Blood leucocytes of patients with RA were also significantly more activated than those of controls, as assessed by monitoring neutrophils MPO and PTX3 intracellular content, monocyte PTX3 expression, ß2 integrins expression and fibrinogen binding and CD40L surface expression. Moreover, the fraction of neutrophils of patients with RA expressing MRP8.14, an index of recent neutrophil release into the bloodstream, was substantially higher (table 2). Cellular TF represents a bona fide marker of a pro-thrombotic cellular profile, particularly when associated with membrane-bound fibrinogen. Patients with RA have an increased fraction of neutrophils, monocytes and platelets that express both TF and fibrinogen compared with patients with CSA and OA and age-matched and sex-matched healthy donors (figure 4D–F and table 2).
Markers of cellular activation in patients and controls
Platelet activation is associated with leucocyte activation and the presence of heterotypic aggregates in patients with rheumatoid arthritis (RA). Platelet P-selectin (A), the heterotypic aggregates of platelets and neutrophils (B), heterotypic aggregates of platelets and monocytes (C) and the tissue factor (TF) expression by platelets (D), neutrophils (E) and monocytes (F) were assessed by flow cytometry in whole blood samples of healthy donors, of patients with chronic stable angina (CSA), of patients with osteoarthritis (OA) and of patients with RA. Symbols depict the individual observations and lines the mean. Significantly different from the relevant control: *p<0.04, **p<0.003, ***p<0.001 and ****p<0.0001.
Treatment with anti-TNFα agents curbs platelets and leucocytes activation in RA
Patients with RA were recruited cross-sectionally and included untreated patients, patients treated with DMARDs alone, with TNFα blockers either alone or in association with DMARDs (see online supplementary table S1). Parameters reflecting platelet activation were consensually and significantly modulated in patients treated with TNFα blockers, including P-selectin expression, TF expression, fibrinogen binding, vWF granular content, fraction of platelet-leucocyte heterotypic aggregates (table 2). The difference was statistically significant both comparing patients with RA treated with TNFα blockers with those treated with DMARDs only and with patients that were not receiving any treatment. The anti-TNFα treatment also effectively modulated the activation state of neutrophils (table 2). The effect on monocyte TF expression and fibrinogen binding was not statistically significant.
There was an inverse correlation between adhesion of neutrophils to activated platelets and their content of MPO (r=−0.41, p<0.02) and of PTX3 (r=−0.47, p<0.01; figure 4G). The neutrophil TF expression was significantly correlated with (1) the platelet P-selectin expression; (2) platelets’ adhesion to neutrophils (platelet-neutrophil heterotypic aggregates) and (iii) the platelet TF expression (figures 2 and 4G).
Discussion
In this study, we make three significant observations. First, human TNFα is an effective platelet agonists that prompts their inflammatory action as well as their pro-coagulant activity. Second, circulating platelets of patients with RA, a condition in which TNFα plays a crucial non-redundant role, mimic virtually all features of platelets that had been challenged in vitro with TNFα. Third, the treatment with anti-TNFα agents—which have been shown to be associated in recent meta-analyses with reduction in the risk of all cardiovascular events12—not only prevents in vitro platelet activation, but associates in vivo with a significant reduced activation of platelets of patients with RA and with the inhibition of their expression of the pro-coagulant moieties they express, TF in particular.
Heightened risk of cardiovascular events, coronary heart disease and stroke in particular, characterises RA and rheumatological diseases in general. The cardiovascular risk of RA is similar to the risk of patients with diabetes mellitus and is responsible of more than half of the premature deaths due to the disease.33 The underlying mechanisms have not been completely elucidated. Heightened risk has been proposed to involve the composition of atherosclerotic plaques. While the atherosclerotic burden is similar in patients with RA and age-matched and sex-matched control subjects, unstable plaques of patients with RA were twice as many, with prominent medial and adventitial inflammation.34 Plaque vulnerability has been confirmed by ultrasound analysis of carotid arteries of patients with RA3 and 18F-FDG-PET reveals in patients with active RA aortic inflammation suggestive of subclinical vasculitis.35 Vascular inflammation appears to be responsive to anti-TNFα agents.35 Thus, the increased risk of cardiovascular events depends not so much on an increased atherosclerotic burden of patients with RA but on the enhanced inflammation of the vessel wall possibly at least partially dependent on TNFα, with atherosclerotic plaques that become more susceptible to rupture, with ensuing atherothrombosis.1 ,33
Platelets are intriguing candidates to link systemic inflammation, active synovitis and cardiovascular risk in patients with RA. Platelets interact with bacteria, release interleukin 1ß (IL-1ß), express innate receptors of the toll-like receptor family1 ,33 and the prototypic endogenous inflammatory signal (DAMP/alarmin) high-mobility group box-1, 14 ,31 all signals that contribute to the chronic inflammatory response that characterises the prototypic rheumatic disease, RA. Patients with active RA have higher percentage of circulating platelets that express P-selectin compared with healthy controls or to patients with remitting RA36 ,37 and the extent of platelet activation is associated with systemic inflammation.36 Platelet-derived microparticles are present in the joint fluid of patients with RA and elicit the production of IL-1ß by fibroblasts.24 The route of entry of platelets within the inflamed joints of patients with RA is not known, even if they might be transported within the synovial fluid by inflammatory leucocytes to which they adhere. Platelet-leucocyte adhesion depends on platelet P-selectin and P-selectin knockout mice had a significantly lower joint inflammation,38 which could be reverted via the injection of wild-type platelets.39
The data of the present study confirm that circulating platelets of patients with RA are activated, displaying various features of activation including membrane P-selectin and forming heterotypic aggregates with leucocytes. Indeed, platelet activation correlated with leucocyte activation in patients with RA, supporting the contention that their reciprocal activation might contribute to persisting inflammation.14 ,24 ,26 ,31 ,33 ,40 We have also verified that platelet activation in RA results in the upregulation of moieties involved in their procoagulant activity, including exposure of anionic phospholipids, binding to fibrinogen and expression of TF. These properties might well be involved in the enhanced propensity to atherothrombosis of patients with RA.
Our study provides novel hints on the nature of the inflammatory signals that activate platelets in RA. We have observed that, on the one hand, recombinant TNFα activates platelets, through the interaction with membrane TNFαR1 and TNFαR2 receptors, acquiring both the ability to activate leucocytes via a P-selectin-dependent pathway and to trigger a pro-coagulant response leading to the expression of functional TF and to the generation of thrombin. The expression of platelet TNFα receptors has been previously reported,41 ,42 and our observations on their expression in non-overlapping platelet domains (membranes of α granules vs plasma membrane) well agree with a cooperative role of the two moieties in the fine-tuning of the response to the cytokine.43 Functional TNFα plays a non-redundant role in systemic inflammation and in the joint involvement of RA. Our results seem to indicate that it might also be involved in enhancing the platelet pro-coagulant function, an event that could well justify at least in part the enhanced risk of atherothrombosis.
This study has limitations, specifically including the cross-sectional assessment of patients and the relatively small sample sizes. Patients were not assuming drugs known to directly interfere with platelet function with the exception of COX2 inhibitors that, however, have been demonstrated not to interfere with the platelet response in terms of aggregation, TXB2 formation and bleeding time.44 However, we cannot rule out indirect effects due to their action on monocytes. Studies on a substantially larger cohort of patients will be valuable to limit the potentially confounding effects of the patients’ medical treatment. Longitudinal prospective assessment of patients with RA before and at different times after the onset of the anti-TNF therapy will also be important.
In conclusion, we have provided in vitro evidence that TNFα behaves as a platelet agonist, triggering a response that was specific, saturable and that comprised amplification of their inflammatory and pro-coagulant actions. Platelets of patients with active RA are similar to those activated in vitro with the cytokine, and that treatment with anti-TNFα agents curbed both inflammatory and pro-coagulant activities, supporting the physiological relevance of the pathway in inflammatory conditions. Although our findings might have important implications for understanding platelet-leucocyte interactions and their involvement in the heightened thrombotic risk of patients with RA, no firm conclusions can be drawn on whether the inhibition of platelet procoagulant activities justifies the protective effects of anti-TNFα agents. Prospective studies on larger patient cohorts are required to verify this contention.
Acknowledgments
The authors thank Dr Giliola Calori for advice on statistical analysis and Dr Maria Carla Panzeri for the electron microscopy. All microscopies have been carried out in ALEMBIC, an advanced microscopy laboratory established by the San Raffaele Scientific Institute and the Vita-Salute San Raffaele University.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
- Data supplement 1 - Online figures
- Data supplement 2 - Online tables
Footnotes
Handling editor Tore K Kvien
Contributors AAM and NM defined the experimental strategy, performed experiments and wrote the manuscript. MatB elaborated the clinical database and performed the statistical analysis, MaB, EB, GP, AM, PR-Q and MGS selected patients, discussed the experimental strategy and corrected the manuscript. MC, MarB and ET contributed the evidence on thrombin generation and corrected the manuscript.
Funding Our research is supported by the Italian Ministry of Health (Ricerca Finalizzata RF11-14) and by the MIUR (PRIN R0504).
Competing interests None declared.
Patient consent Obtained.
Ethics approval Institutional Review Board approved the blood biobanking.
Provenance and peer review Not commissioned; externally peer reviewed.