Thrombosis Research 129 (2012) 591–597

Contents lists available at SciVerse ScienceDirect

Thrombosis Research

j ourna l homepage: www.e lsev ie r .com/ locate / thromres

Regular Article

Circulating microparticles and risk of venous thromboembolism

Paolo Bucciarelli a,⁎, Ida Martinelli a, Andrea Artoni a, Serena M. Passamonti a, Emanuele Previtali a,Giuliana Merati a, Armando Tripodi a, Pier Mannuccio Mannucci b

a A. Bianchi Bonomi Hemophilia and Thrombosis Center, Department of Medicine and Medical Specialties,Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico and University of Milan, Milan, Italyb Scientific Direction, Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico and University of Milan, Milan, Italy

Abbreviations:MPs, microparticles; VTE, venous thro95%CI, 95% confidence interval.⁎ Corresponding author at: Hemophilia and Thromb

Ca’ Granda Ospedale Maggiore Policlinico, Via Pace, 9–02 55035274; fax: +39 02 50320723.

E-mail address: bucciarelli@policlinico.mi.it (P. Bucc

0049-3848/$ – see front matter © 2011 Elsevier Ltd. Alldoi:10.1016/j.thromres.2011.08.020

a b s t r a c t

a r t i c l e i n f o

Article history:

Received 3 June 2011Received in revised form 16 August 2011Accepted 17 August 2011Available online 9 September 2011

Keywords:Blood coagulationMicroparticlesPlateletsVenous thrombosisRisk factors

Introduction: Circulating microparticles (MPs) may trigger a hypercoagulable state, leading to thromboticcomplications. Data on their association with venous thromboembolism (VTE) are few and inconsistent.Materials and methods: To investigate whether or not high levels of MPs are associated with an increased riskof VTE, we carried out a case-control study on 186 patients with a first, objectively diagnosed, episode of VTEand 418 healthy controls. Plasma levels of circulating MPs were measured by flow cytometry.Results: Patients had higher median plasma levels of total MPs than controls (2184 per μL vs 1769 per μL,pb0.0001). The risk of VTE increased progressively with increasing MPs, with a linear dose-response effectin the log odds. Individuals with MPs above the 90th percentile of the controls’ distribution (P90=3263per μL) had a 5-fold increased risk of VTE than those with MPs below the 10th percentile of controls(P10=913 per μL), independently of sex, age, body mass index, thrombophilia, and plasma factor VIII levels[adjusted odds ratio: 5.30 (95%CI: 2.05-13.7)]. Using the 95th percentile of controls as cut-off (P95=4120 per

μL), the adjusted odds ratio was 2.20 (1.01-4.79) for individuals with MPsNP95 compared with those havingMPs≤P95. After exclusion of individuals with antiphospholipid antibodies and hyperhomocysteinemia, theinteraction between MPsNP95 and thrombophilia increased the VTE risk from 1.63 (0.60-4.50) to 6.09(1.03-36.1).Conclusions: High levels of circulating MPs are a possible independent risk factor for VTE.

© 2011 Elsevier Ltd. All rights reserved.

Introduction

Microparticles (MPs) are circulating, phospholipid-rich particles ofb1 μm diameter released from the membranes of platelets, endothelialcells, leucocytes and erythrocytes [1–3]. They are formed by the exocyticbudding of cell membranes, after which the asymmetry of plasmamem-brane lipid bilayer is altered resulting in the exposure of a surface rich innegatively charged phospholipids, particularly phosphatidylserine [4–6].The release of MPs from the cell surface is the consequence of cell activa-tion or apoptosis by chemical (e.g. cytokines, thrombin, endotoxin) orphysical (e.g. shear stress, hypoxia) stimuli [7–9]. These changes resultin cytoskeletal reorganisation, membrane blebbing and formation of

mboembolism; OR, odds ratio;

osis Center, Fondazione IRCCS20122 Milan, Italy. Tel.: +39

iarelli).

rights reserved.

MPs. The protein composition of MPs reflects that of the cell membranefrom which they are released [2]. It is likely that both the cell origin andthe nature of the trigger influence the number and phenotype of MPsand their mechanistic effects. For instance, the MPs released from acti-vated platelets or from endothelial cells following apoptosis (represent-ing together the vast majority of total MPs) express high levels ofphosphatidylserine (detected by Annexin V binding), and the anionicsurface of this phospholipid can bind coagulation factors, promote theactivity of tenase and prothrombinase complexes and ultimately en-hance fibrin clot formation [10]. Platelet glycoprotein receptors, such asGPIb (that mediates adhesion to von Willebrand factor), and P-selectinmay also be involved in platelet MP formation under shear stress condi-tions [11]. MPs expressing tissue factor can also be identified both inphysiological and in some pathological conditions, thus providing anoth-er suitable environment to initiate and support coagulation. In physio-logical conditions, tissue factor-positive MPs are mostly derived frommonocytes and to a less extent from platelets, while their sheddingfrom endothelial cells occurs mainly upon cellular activation. They aremore procoagulant than tissue factor-negative MPs, being rapidly accu-mulated in a developing thrombus and increased in some prothromboticdisorders [12]. MPs have been studied in different clinical conditions, in-cluding cardiovascular disease [13–16], the antiphospholipid antibodysyndrome [17,18], sepsis [19], thrombotic thrombocytopenic purpura

592 P. Bucciarelli et al. / Thrombosis Research 129 (2012) 591–597

[20], hemolytic anemias [21,22], inflammatory states [23,24], and cancer[25]. Many of these conditions share a hypercoagulable state leading tothrombosis.

Venous thromboembolism(VTE) is amultifactorial diseasewith an in-cidence in the general population of about 1–2 cases per 1000 person-years [26], that develops with no apparent cause in a substantial propor-tion of cases [27]. Data on the potential mechanistic role of highMP levelsare scanty and contrasting. Four case-control studies of small sample sizein selected patients with antiphospholipid antibodies [18], acute VTE [28]or cancer-related VTE [29,30] showed an association between these con-ditions and high MP levels, but a recent study on patients with recurrentVTE [31] failed to show an association.

With this as background, we designed a case-control study to inves-tigate whether or not high levels of circulating MPs are associated withan increased risk of a first VTE.

Methods

Study population

Two hundred and two patients consecutively referred to our Throm-bosis Center between January 2004 and July 2009 for a thrombophiliawork-up after a first episode of VTEwere included in the study. Their de-mographic data, medical history, and exposure to risk factors for VTEwere collected. Sixteen patients were then excluded because VTE wasprovokedby cancer (n=9) or hadoccurred a long time (N5 years) beforestarting recruitment (n=7). VTE included objectively documented firstepisodes of lower-limb deep vein thrombosis (diagnosed by compressionultrasound or venography) and/or pulmonary embolism (diagnosed byV/P lung scan, CT scan or pulmonary angiography). Surgery, prolongedbed rest (N1 week), pregnancy/puerperium, oral contraceptive use andhormone replacement therapy were considered risk factors for VTE. Theevents that had occurred in the absence of the aforementioned risk fac-tors were considered unprovoked. Patients with surgery-related VTEwere investigated at least 3 months after surgery, and those withpregnancy-related VTE at least 3 months after delivery.

Four hundred and twenty-two healthy people who were partners orfriends of the patients formed the control group. They were recruitedduring the same inclusionperiod of patients. Previous episodes of throm-bosis were excluded by a validated questionnaire [32]. Four controlswere excluded because samples had been inadequately stored. Hence,186 patients and 418 controls formed the study population. None ofthem was on anticoagulant or antiplatelet therapy at the time of bloodcollection, had overt cancer, myeloproliferative disorders, diabetes,liver, kidney or chronic inflammatory diseases. The Hospital InstitutionalReview Board approved the study, that was carried out and is reportedaccording to STROBE guidelines [33]. All patients and controls gavewrit-ten informed consent to participate to the study.

MP measurement

MPs were measured on 3.2% citrated plasma by one of us (GM)who was unaware of the case/control status. When drawing blood,prolonged use of a tourniquet was avoided and a needle of 19 G di-ameter was used to reduce cell activation or damage. The bloodwas collected into vacuum tubes containing sodium citrate, centri-fuged within 15 min at (controlled) room temperature for 20 minat 2880×g to pellet the cells, providing an optimal yield of MPswith minimal platelet contamination. The plasma obtained was ali-quoted and snap-frozen in liquid nitrogen, and then stored at−80 °C until analysis. APC-conjugated Annexin V (BD Pharmingen,San Diego, CA), a FITC-conjugated monoclonal antibody (mAb) toplatelet glycoprotein GPIIb (FITC-CD41, clone P2, Beckman Coulter,Marseille, France) and a PE-conjugated mAb to tissue factor (PE-CD142, clone HTF-1, BD Pharmingen) were used to identifyphosphatidyl-serine positive MPs, platelet-derived MPs and MPs

expressing tissue factor, respectively. For direct MP quantification25 μl of freshly thawed plasma was incubated in a polystyrene tube(BD Pharmingen) with 2.5 μl of annexin V-APC, 10 μl of CD142-PEand 10 μl CD41-FITC for 30 minutes in Hepes buffer containing5 mM CaCl2. Proper concentrations of isotype antibodies (antimouse IgG-FITC, Sigma, Buchs, Switzerland; mouse IgG1K-PE, BDPharmingen) or Annexin V-APC in Hepes buffer without calciumwere used to set up background fluorescence levels. Then 500 μl ofa stirred suspension of 7 μm counting beads (Fluka Buchs, Switzer-land) in Annexin V-binding buffer at a concentration of 350 beadsper μl was added to each sample. Samples were then immediatelyanalysed on a FACS Calibur (Becton Dickinson) flow cytometer, andMPs were identified according to their size and fluorescence. Briefly,on a log-forward scatter (FSC) versus log-side scatter (SSC) plot, theMP upper size limit was defined using standard 1 μm beads (Fluka).A gate was drawn to include particles b1 μm. Only the events includ-ed in this gate were further analyzed on a log-SSC versus log-Fl plot.Another gate including the 7 μm counting beads population was de-fined and, by using the absolute value of this calibrant, MP count wasexpressed as number per μL of plasma, as previously described [21].Throughout this paper MPs are going to be identified as total MPs(Annexin V positive), platelet-derived MPs (Annexin V and CD41positive) or tissue factor-bearing MPs (Annexin V and CD142 posi-tive), according to whether they refer to the total numbers, tothose derived from platelets or those bearing tissue factor. Beingaware of the many methodological issues associated with MP mea-surement, plasma of cases and controls were handled in the sameway. In order to avoid systematic errors, a maximum of 20 sampleswere run in each set of determinations, with a case:control ratio of1:2. In each analysis we inserted two quality control samples withdifferent levels of MPs, one obtained from 50 healthy lab workersand the other from 50 healthy blood donors. When one of the qualitycontrol samples gave results out of range (i.e., less or more 2 SD), theentire set of results was not considered and the analysis was repeat-ed. This situation occurred in 8 of the 41 total sets of determinations(19%). The inter-assay coefficient of variation measured on the inter-nal standards in terms of MP number was 6.2% for total MPs, 5.6% forplatelet-derived MPs and 12% for tissue factor-bearing MPs. Residualplatelet concentration in plasma was monitored by counting eventsin the platelets gate in all the samples analyzed. The median (min-max values) residual platelet number was 88 per μL (15 – 732) inVTE patients and 77 per μL (14 – 619) in controls.

Other laboratory tests

Patients and healthy controls were tested for such thrombophilic ab-normalities as antithrombin, protein C or protein S deficiency, G1691Amutation in factor V gene (factor V Leiden), G20210A mutation in pro-thrombin gene, antiphospholipid antibodies, hyperhomocysteinemia (di-chotomous variables) and high coagulation factor VIII levels (continuousvariable). All these tests were performed on citrated plasma or DNA aspreviously described [34]. A subject was defined as having thrombophiliaif one or more of these defects was present. Hyperhomocysteinemia wasdefined as either fasting or post-methionine load homocysteine levels ex-ceeding the 95th percentile of distribution among controls. D-dimer plas-ma levels were measured as a marker of coagulation activation and VTErisk in a subsample of patients (n=173) and controls (n=274), usinga highly sensitive test (HemosILTM D-Dimer HS performed on ACL TOP,Instrumentation Laboratory, Bedford, MA, USA).

Statistical analysis

Assuming a hypothetical 5% prevalence of high plasma levels of cir-culating MPs in the control group and a hypothetical 2.5 relative riskof VTE for individuals with circulating MPs above the 95th percentile,with a case:control ratio of 1:2, a two-tailed α error of 0.05 and a 80%

Table 1Demographic and clinical characteristics of the study population.

patients (n=186) controls (n=418)

sex (M/F) 78/108 120/298age at blood sampling (years) 45 (11 – 83) 41 (14 – 80)age at thrombosis (years) 44 (11 – 78) NAtime between thrombosisand blood sampling (months)

13 (1 – 169) NA

body mass index (Kg/m2) 24.9 (13.3 – 45.8) 23.4 (16.1 – 45.2)type of VTE event

- DVT [n (%)] 126 (68)- DVT+PE [n (%)] 27 (14) NA- PE [n (%)] 33 (18)

risk factors of VTE- none [n (%)] 75 (40) —

- surgery [n (%)] 41 (22) —

- bed rest [n (%)] 38 (20) —

- pregnancy/puerperium [n (%)] * 15 (38) —

- oral contraceptive intake [n (%)] † 41 (63) 44 (22)- hormone replacement therapy[n (%)] ‡

3 (11) 3 (1)

thrombophilia §- none [n (%)] 110 (59) 351 (84)- AT, PC, PS deficiency [n (%)] 15 (8.1) 2 (0.5)- factor V Leiden [n (%)] 22 (12) 19 (4.5)- prothrombin G20210A [n (%)] 16 (8.6) 20 (4.8)- antiphospholipid antibodies [n (%)] 5 (2.7) 6 (1.4)- hyperhomocysteinemia [n (%)] 28 (15) 25 (6.0)

factor VIII (IU/dL) 130 (60 – 300) 111 (51 – 252)

Continuous variables are expressed as median (minimum and maximum valuesbetween brackets).NA=not applicable, VTE=venous thromboembolism, DVT=deep vein thrombosis,PE=pulmonary embolism.AT=antithrombin, PC = protein C, PS = protein S.* % calculated on women on reproductive age (80 patients), after exclusion of those onoral contraceptives.† % calculated on women on reproductive age (80 patients and 202 controls), afterexclusion of pregnant women.‡ % calculated on women on menopausal age (28 patients and 96 controls).§ some individuals carry more than one defect.

593P. Bucciarelli et al. / Thrombosis Research 129 (2012) 591–597

power, a theoretical sample size of 215 cases and 430 controls was cal-culated. Continuous variables were expressed asmedianwithminimumand maximum values, and categorical variables as counts. Because thedistribution of MPs and D-dimer was left-skewed, all parametric statis-tical analyses were performed after a logarithmic transformation. Vari-ables potentially affecting MP levels were examined in controls.Comparison between groups was made by a Student's t test or one-way ANOVA (with Bonferroni's post-hoc contrast analysis) for continu-ous variables, and by a χ2 test for categorical variables. Correlation be-tween two variables was made by the Spearman's correlation test. Toassess the risk of VTE associated with different plasma levels of MPs, aunivariable logistic regression model with MPs as a continuous variablewas firstfitted. The possible presence of non-linear effects of theMP dis-tribution on the risk of VTE (expressed as log odds) was evaluated usinga restricted cubic spline function with 3 knots, which was the one thatmaximized the Akaike's information criterion [(model likelihood ratioχ2 –2p), with p equal to the number of parameters in the model asidefrom the intercept (i.e., the number of knots – 1)] [35]. The non-lineareffect was visualized by plotting the model-predicted log odds againstMP values, and tested by using the Wald test. Then we chose the 10th(913 per μL), the 50th (1769 per μL) and the 90th (3263 per μL) percen-tiles of the MP distribution among controls to divide the study popula-tion into 4 categories (b913 per μL, 913–1769 per μL, 1770–3263 perμL, and N3263per μL) andcalculated theodds ratio (OR) and its 95% con-fidence interval (95%CI) as a measure of the relative risk of VTE for indi-viduals in each MP category, taking the lowest one as reference. The95th percentile of MP distribution among controls (P95=4120 per μL)was also used as a cut-off point and the odds ratio of VTE for individualswith MPs above P95 was calculated, compared with those havingMPs≤P95. Unconditional logistic regression analysis was performed tocontrol for such possible confounders as sex (0 = female, 1 = male),age (continuous variable), bodymass index (continuous variable), pres-ence of thrombophilia (0 = no, 1 = yes), and factor VIII plasma levels(continuous variable). Interaction between high levels of circulatingMPs (above P95) and presence of thrombophilia on the risk of VTE wasalso investigated. This analysis was repeated after exclusion of individ-uals with antiphospholipid antibodies or hyperhomocysteinemia, twothrombophilic conditions that are known to cause endothelial damageand to increase circulating MPs, thus leading to a possible overestima-tion of the risk of VTE associated with both thrombophilia and highlevels of MPs. P≤0.05 was chosen as cut-off level for statisticalsignificance.

All analyses were performed with the statistical software R (release2.9.1; R Project for Statistical Computing, Vienna, Austria).

Results

Demographic and clinical data of patients and controls are shown inTable 1.Womenweremore represented thanmen both among patientsand controls. Patients were slightly older than controls (median age: 45vs 41 yrs; p=0.005), had higher body mass index (24.9 vs 23.4 Kg/m2;pb0.0001), factor VIII plasma levels (130 vs 111 IU/dL; pb0.0001), andmore frequent thrombophilic abnormalities (41% vs 16%; pb0.0001). In75 patients (40%) VTE was unprovoked.

The number of total, platelet-derived and tissue factor-bearing MPsis shown in Table 2. A strong correlation was found between total MPsand platelet-derived MPs (ρ=0.99, pb0.0001) and between platelet-derived MPs and tissue factor-bearing MPs (ρ=0.94, pb0.0001). Theratio between platelet-derived MPs and total MPs was 0.91 (min-maxvalues: 0.47 – 0.98) for VTE patients and 0.90 (0.59 – 0.97) for controls(p=0.176). The median (min-max values) percentage of subjects withMPs expressing both CD41 and CD142 was 23.9% (14.4 – 44.4%), andwas equally distributed between patients and controls. Because of theabove-mentioned high correlations, we chose to show only results per-taining to total MPs.

Table 3 shows plasma levels of circulating MPs related to differentvariables in controls. No statistically significant difference was foundaccording to sex or presence or absence of thrombophilia, nor amongdifferent age, body mass index or factor VIII categories. In a subgroupof control women of reproductive age, no statistically significant differ-ence in MP levels was found between oral contraceptives users andnon-users.

Median plasma levels of circulatingMPs were significantly higher inVTE patients than in controls [2184 per μL (min-max: 367–8783) vs1769 per μL (302–7356); pb0.0001] (Fig. 1). Patients tested within3 months fromVTE (n=22) had higher circulatingMPs than those test-ed later [2485 per μL (1012–8783) vs 1829 per μL (367–8449) at 4–12months from VTE (n=62) and 2032 per μL (562–6594) after12 months from VTE (n=102); p=0.075 for the overall test, with themain difference found between ≤3 months and 4–12 months in theBonferroni's post-hoc contrast (p=0.070)]. Hence, the following ana-lyses on the risk of VTE associated with MP levels were performed ex-cluding 22 patients tested within 3 months from VTE.

Fig. 2 represents the splinefit estimating the relationship between thelog odds of VTE and MP plasma levels. The risk of VTE linearly increasedwith increasing MPs. The linear rise on a log odds scale corresponds toan exponential rise on a non-logarithmic scale. The contribution of non-linear effects on the log odds of VTE was insignificant (Wald test:χ2=0.63 with 1 degree of freedom; p=0.428). For every 500 per μL in-crease in MPs, there was a 16% increase in the VTE risk [adjustedOR=1.16 (95%CI: 1.06-1.26)]. Table 4 shows the risk of VTE associatedwith different categories of MPs based on the 10th, the 50th and the90th percentiles among controls. After adjustment for age, sex, bodymass index, presence of thrombophilia and factor VIII plasma levels, indi-viduals with MPs above 90th percentile of MP distribution had a 5-fold

Table 2Plasma levels of microparticles in VTE patients and healthy controls.

patients (n=186) controls (n=418) p-value

Total MPs (No. per μL)(Annexin V+)

2184 (367 – 8783) 1769 (302 – 7356) b0.0001

Platelet-derived MPs (No. per μL) *(Annexin V+and CD41+)

1942 (339 – 8074) 1519 (242 – 7032) b0.0001

Tissue factor-bearing MPs (No. per μL) *(Annexin V+and CD142+)

579 (98 – 2269) 454 (76 – 2097) b0.0001

Values are expressed as median (minimum and maximum values between brackets).MPs = microparticles.* In 24% of the total subjects (equally distributed between VTE patients and controls), MPs co-expressed CD41 and CD142 antigens.

594 P. Bucciarelli et al. / Thrombosis Research 129 (2012) 591–597

increased risk of VTE compared with those having MPs b10th percentile.Using the 95th percentile of MP distribution in controls as cut-off point,individuals with MPs above the 95th percentile had a 2.2-fold increasedrisk of VTE than those with MPs≤95th percentile [adjusted odds ratio:2.20 (95%CI: 1.01-4.79)], and after the exclusion of individuals with anti-phospholipid antibodies or hyperhomocysteinemia, the adjusted OR was2.13 (95%CI: 0.99-4.66). Table 5 shows the interaction between high plas-ma levels ofMPs (N95th percentile of controls) and thrombophilia on therisk of VTE, thatwas increased 11-fold in the presence of both, indicating amultiplicative interaction. The risk was 6-fold increased after exclusion ofindividuals with antiphospholipid antibodies or hyperhomocysteinemia.The excess relative risk (6.09-1.00=5.09) was higher than that expectedby the sumof the two separated excess relative risks in an additivemodel[(1.63-1.00)+(3.37-1.00)=3.00], thus suggesting a synergistic effect.

Finally, considering D-dimer as a marker of coagulation activation,we observed higher median levels in VTE patients than in controls[124 ng per mL (min-max: 17–3415) vs 85 ng per mL (min-max: 5–2188), respectively; pb0.0001]. A mild positive correlation betweenD-dimer and total MPs was found (ρ=0.135, p=0.004), almost totallyattributed to controls (ρ=0.115, p=0.058) but not to patients(ρ=0.064, p=0.406), even after exclusion of the 22 patients testedwithin 3 months from VTE.

Table 3Plasma levels of total microparticles in controls related to different variables.

Variable No. of subjects Total MPs (No. per μL) p value

sex- females 298 1769 (302 – 7168)- males 120 1767 (368 – 7356) 0.250

age (years)- b25 38 1793 (462 – 4374)- 25 – 40 174 1783 (302 – 7168) 0.616- 41 – 60 156 1711 (368 – 7356)- N60 50 1835 (407 – 6797)

body mass index (Kg/m2)- b22 147 1721 (302 – 7168)- 22 – 26 151 1813 (532 – 6797) 0.973- N26 111 1768 (407 – 7356)

thrombophilia *- no 351 1771 (302 – 7356)- yes 67 1732 (707 – 7168) 0.202

factor VIII (IU/dL)- b93 102 1793 (813 – 6493)- 93 – 110 101 1784 (302 – 7356) 0.320- 111 – 129 114 1663 (368 – 5121)- N129 101 1747 (407 – 6797)

oral contraceptive intake †

- no 158 1820 (302 – 7168)- yes 44 1774 (462 – 5775) 0.907

MPs = microparticles.Values of MPs are expressed as median (minimum and maximum values betweenbrackets).* AT, PC or PS deficiency, factor V Leiden, prothrombin G20210A, antiphospholipidantibodies, hyperhomocysteinemia, N1 abnormality.† calculated in the subgroup of control women on reproductive age (n=202).

Discussion

This retrospective case-control study shows an association betweenhigh plasma levels of total MPs and the risk of a first VTE, with a lineardose-response effect in the log odds. Compared with individuals having

Fig. 1. Box-plots of the distribution of total microparticles (MPs) in controls and pa-tients with VTE. Each box-plot represents the median, interquartile range, and 95% in-tervals. Open circles and asterisks identify outliers and extreme values, respectively.

Fig. 2. Restricted cubic spline curve showing the model-predicted log odds of VTEagainst plasma levels of total microparticles (MPs). Dashed lines represent 95% confi-dence intervals. Black triangles identify the observed log odds related to the midpointsof the four categories based on the 10th, 50th and 90th percentiles of MP distributionamong controls.

Table 4Risk of VTE associated with different categories of plasma levels of total microparticles.

MPs(No. per μL) *

n (%) OR (95%CI) ORadj(95%CI) ‡

patients †(n=164)

controls(n=418)

b913 10 (6) 41 (10) 1 (Ref.) 1 (Ref.)913 – 1769 59 (36) 168 (40) 1.44 (0.68 – 3.06) 2.10 (0.91 – 4.88)1770 – 3263 61 (37) 168 (40) 1.49 (0.70 – 3.15) 2.02 (0.87 – 4.66)N3263 34 (21) 41 (10) 3.40 (1.49 – 7.78) 5.30 (2.05 – 13.7)

MPs = microparticles.Ref. = reference category.* Categories based on the 10th, the 50th and the 90th percentiles of MPs distributionamong controls.† 22 patients tested within 3 months from VTE were excluded.‡ OR adjusted for sex, age, body mass index, thrombophilia and plasma levels of factorVIII.

595P. Bucciarelli et al. / Thrombosis Research 129 (2012) 591–597

MPs b10th percentile, those with MPs N90th percentile had a 5-fold in-creased risk of having had a previous VTE, independently of otherknown risk factors for VTE. Considering the 95th percentile of controlsas cut-off point, the risk of VTE was 2.2-fold higher in individuals withMPs above this point than in those with MPs ≤95th percentile.

To date, few case-control studies (most of them of small size) inves-tigatedwhether or not highMP levels are a risk factor for VTE [18,28–31].A study in patients with antiphospholipid antibodies showed higher MPlevels in thosewith orwithout VTE than in controls or VTE patientswith-out antiphospholipid antibodies [18]. These findings were supported byin vitro analyses showing that plasma from patients with antiphospholi-pid antibodies (independently of the presence or absence of VTE) in-creased the release of endothelial MPs from cultured endothelial cells.Another study found an association between acute VTE and high endo-thelial but not platelet-derived MPs [28], however most of the eventswere triggered by surgery or underlying malignancy that may per se af-fectMP levels [36–38]. In two studies on patients with cancer-associatedVTE, the MP-associated tissue factor activity measured with a functionalmethod was higher in cancer patients with VTE than in healthy controlsor in patients with cancer without VTE [29,30]. Using another functionalassay, other authors failed to find an association between MP levels andrecurrent VTE [31]. This is a phosphatidylserine-equivalent assay thatmeasures the amount of thrombin generated from phosphatidylserineexposed on the surface ofMPs in plasma. Functional assays give informa-tion on functionalMP activity but nodetails onnumber, size and origin ofMPs. Moreover, there are some publications in which there was no cor-relation between MP functional assay results and the number of MPsidentified by flow cytometry in the same sample, possibly becausethese methods may also detect the activity of large apoptotic bodies

Table 5Interaction between high levels of total microparticles and thrombophilia on the risk of VT

patients *

MPsN4120 per μL (95th percentile) thrombophilia ‡

no no 93 (57)yes no 6 (4)no yes 54 (33)yes yes 11 (7)

MPsN4120 per μL (95th percentile) thrombophilia §no no 93 (68)yes no 6 (4)no yes 32 (24)yes yes 5 (4)

MPs = microparticles.Ref. = reference category.* 22 patients tested within 3 months from VTE were excluded.† Adjusted for sex, age, body mass index and plasma levels of factor VIII.‡ AT, PC or PS deficiency, factor V Leiden, prothrombin G20210A, antiphospholipid antibod§ AT, PC or PS deficiency, factor V Leiden, prothrombin G20210A or N1 abnormality.

[3,29,39]. Therefore the difference between studies using functional orflow cytometric methods may simply be related to the type of assayused to measure MPs. We chose to measure MPs by flow cytometrywith noparallelmeasurement ofMPprocoagulant activity.We recognizethe limitations and possible pitfalls of flow cytometry in detecting MPs,and that a wide variety of methods has been used up to now for MPquantification, each giving results not directly comparable to the others,thus explaining the inconsistency of findings often observed in differentstudies [40]. However, in this studywe used plasma samples for patientsand controls which were collected, processed and submitted to flowcytometry testing exactly in the same manner. Therefore, any effectdue to the pre- and analytical variability should be equally distributedamong patients and controls, an important prerequisite to draw mean-ingful conclusions from a case-control study. At variance of a body ofpublished works, the method of centrifugation that we used in thisstudy may leave a small amount of platelets into the samples, thus lead-ing to a possible overestimation of MP number due to the activation ofresidual platelets. However, the contamination of our samples by plate-lets was minimal and equally represented in patients and controls; wetherefore are confident that it did not affect the difference in MP plasmalevels found between VTE patients and controls. In a Dutch study [29]that measured MP plasma levels with flow cytometry (besides a func-tional method measuring their tissue factor activity), and with pre- andanalytical conditions nearly close to those of this study, themedian num-ber (range) of total MPs per μL in healthy controls was very similar tothat found in the present study [1600 (720–9000) vs 1769 (302–7356),respectively].

Despite the high correlation between platelet-derived MPs and tis-sue factor-bearing MPs, only about one fourth of subjects showed aco-expression of CD41 and CD142, thus suggesting a common underly-ingmechanism of MP generation. The co-expression of a platelet mem-brane antigen and tissue factor in a fraction of MPs can be explained byeither exchange of vesicles between different cell types [39] or a directexpression of tissue factor by platelets [41].

Another strength of this study is the accurate selection of patients,who were consecutively referred to a tertiary Thrombosis Center aftera first episode of VTE, were not on anticoagulant or antiplatelet ther-apy and had no overt cancer, all factors that may influence plasmalevels of MPs [42,43]. Moreover, patients with VTE provoked bysuch other factors potentially affecting MP count as pregnancy or sur-gery were tested far (N3 months) from these triggering events. Thepossibility that some patients with unprovoked VTE had occult cancer(thus having MP count higher than normal) cannot be completelyruled out.

The selection of patients from a tertiary center where they are re-ferred for thrombophilia screening is a limitation, because they do not

E.

n (%) OR (95%IC) ORadj (95%IC) †

controls

334 (80) 1 (Ref.) 1 (Ref.)17 (4) 1.27 (0.49 – 3.31) 1.63 (0.60 – 4.50)64 (15) 3.03 (1.97 – 4.65) 2.89 (1.82 – 4.57)3 (1) 13.2 (3.60 – 48.2) 11.0 (2.81 – 43.5)

334 (86) 1 (Ref.) 1 (Ref.)17 (4) 1.27 (0.49 – 3.31) 1.63 (0.60 – 4.50)34 (9) 3.38 (1.98 – 5.77) 3.37 (1.90 – 6.00)2 (1) 8.98 (1.71 – 47.0) 6.09 (1.03 – 36.1)

ies, hyperhomocysteinemia or N1 abnormality.

596 P. Bucciarelli et al. / Thrombosis Research 129 (2012) 591–597

represent VTE patients from the general population. Although our resultscan not be generalized to all VTE patients, the absence in our patients ofcomorbidities associated with both high levels of MPs and increased riskof VTE may have underestimated the magnitude of the association be-tween high levels of MPs and the risk of VTE. Another limitation of thisstudy is its retrospective design, since patients were investigated afterthe occurrence of thrombosis, and the possibility that the acute phasehave enhanced MP number cannot be ruled out. Indeed, patients testedclose to the episode of VTE had higher levels of MPs than those testedlater. However, after the former were excluded from the analysis onVTE risk, an increased relative risk of VTE in association with high MPlevels was still observed, thus suggesting a possible causal relationshipbetween high circulating MPs and VTE.

Onemay argue thatMPs are not a true risk factor for VTE but simply amarker of hypercoagulability in the pathway between the known riskfactor(s) and thrombosis. This may be particularly the case for such ac-quired risk factors as antiphospholipid antibodies or hyperhomocystei-nemia, that might lead to endothelial cell activation and heightenedproduction of endothelial MPs which in turn stimulate the release ofboth platelet-derived and tissue factor-expressing MPs [17,18,44]. How-ever, the exclusion from this study of individuals with antiphospholipidantibodies or hyperhomocysteinemia did reduce only slightly the riskof VTE associated with high levels of MPs. On the other hand, inheritedthrombophilia abnormalities (such as deficiencies of naturally occurringanticoagulants and gain-of-function genemutations) do not cause endo-thelial damage and, whatever their origin, MPs seem to interact sinergis-tically with inherited thrombophilia in the pathogenesis of thrombosis.The composition of MPs, rich in negatively-charged phospholipids thatcan bind coagulation factors and promote the formation and activity oftenase and prothrombinase complexes, gives biological plausibility toMPs as triggers of clot formation, supporting the hypothesis of their po-tential prothrombotic property. Furthermore, by measuring D-dimerplasma levels a more pronounced coagulation activation was observedin VTE patients than in controls. At variancewith themild correlation be-tween D-dimer and MPs observed in controls (that supports the role ofMPs in the clot formation), the lack of correlation in patients is not sur-prising, because after VTE (whatever the cause) D-dimer levels increasebut MPs do not necessarily. This observation is in favour of the hypothe-sis that high levels ofMPs are a risk factor for VTE rather than amere epi-phenomenon. However, possible subtle changes in low levels of MPsafter VTE cannot be completely ruled out due to the higher variabilityof MP measurement for low than high plasma levels of MPs.

In conclusion, this study shows that elevated levels of circulatingMPs are associated with an increased risk of VTE, which is independentof other known risk factors for VTE. Our findings of high MP levels aspossible independent risk factor for VTE remain to be confirmed in pro-spective cohort studies.

Source of funding

The study has been supported by funding from the FondazioneIRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy.

Conflict of interest statement

The authors do not have any conflict of interest or financial rela-tionship to disclose.

Authorship

P.B., I.M., A.A. and A.T. were responsible for study design andcoordination;

P.B., I.M., S.M.P., E.P. were involved in data collection of patientsand controls;

A.A., G.M. and A.T. were responsible for laboratory analysis;

P.B. was responsible for statistical analysis and wrote the initialdraft of the manuscript;

P.B., I.M., A.A., A.T. and P.M.M. were responsible for revisions of themanuscript;

all authors were responsible for approval of the final manuscript.

References

[1] Freyssinet JM. Cellular microparticles: what are they bad or good for? J ThrombHaemost 2003;1:1655–62.

[2] Lynch SF, Ludlam CA. Plasma microparticles and vascular disorders. Br J Haematol2007;137:36–48.

[3] Davizon P, López JA. Microparticles and thrombotic disease. Curr Opin Hematol2009;16:334–41.

[4] Fox JEB, Austin CD, Boyles JK, Steffen PK. Role of the membrane skeleton in pre-venting the shedding of procoagulant-rich microvesicles from the platelet plasmamembrane. J Cell Biol 1990;111:483–93.

[5] Basse F, Gaffet P, Bienvenue A. Correlation between inhibition of cytoskeleton pro-teolysis and antivesiculation effect of calpeptin during A23187-induced activationof human platelets: are vesicles shed by filopod fragmentation? Biochim BiophysActa 1994;1190:217–24.

[6] Flaumenhaft R. Formation and fate of platelet microparticles. Blood Cells Mol Dis2006;36:182–7.

[7] van Wijk MJ, van Bavel E, Sturk A, Nieuwland R. Microparticles in cardiovasculardiseases. Cardiovasc Res 2003;59:277–87.

[8] Wiedmer T, Sims PJ. Participation of protein kinases in complement C5b-9-induced shedding of platelet plasma membrane vesicles. Blood 1991;78:2880–6.

[9] Miyazaki Y, Nomura S, Miyake T, Kagawa H, Kitada C, Taniguchi H, et al. Highshear stress can initiate both platelet aggregation and shedding of procoagulantcontaining microparticles. Blood 1996;88:3456–64.

[10] Zwaal RFA, Schroit AJ. Pathophysiologic implications of membrane phospholipidasymmetry in blood cells. Blood 1997;89:1121–32.

[11] Reininger AJ, Heijnen HF, Schumann H, Specht HM, Schramm W, Ruggeri ZM.Mechanism of platelet adhesion to von Willebrand factor and microparticle for-mation under high shear stress. Blood 2006;107:3537–45.

[12] Lechner D, Weltermann A. Circulating tissue factor–exposing microparticles.Thromb Res 2008;122(Suppl 1):S47–54.

[13] Bonderman D, Teml A, Jakowitsch J, Adlbrecht C, Gyöngyösi M, Sperker W, et al.Coronary no-reflow is caused by shedding of active tissue factor from dissectedatherosclerotic plaque. Blood 2002;99:2794–800.

[14] Preston RA, JyW, Jimenez JJ, Mauro LM, Horstman LL, Valle M, et al. Effects of severehypertension on endothelial and platelet microparticles. Hypertension 2003;41:211–7.

[15] Leroyer AS, Isobe H, Lesèche G, Castier Y, Wassef M, Mallat Z, et al. Cellular originsand thrombogenic activity of microparticles isolated from human atheroscleroticplaques. J Am Coll Cardiol 2007;49:772–7.

[16] Sabatier F, Darmon P, Hugel B, Combes V, Sanmarco M, Velut JG, et al. Type 1 andtype 2 diabetic patients display different patterns of cellular microparticles. Dia-betes 2002;51:2840–5.

[17] Combes V, Simon AC, Grau GE, Arnoux D, Camoin L, Sabatier F, et al. In vitro gen-eration of endothelial microparticles and possible prothrombotic activity in pa-tients with lupus anticoagulant. J Clin Invest 1999;104:93–102.

[18] Dignat-George F, Camoin-Jau L, Sabatier F, Arnoux D, Anfosso F, Bardin N, et al.Endothelial microparticles: a potential contribution to the thrombotic complica-tions of the antiphospholipid syndrome. Thromb Haemost 2004;91:667–73.

[19] Nieuwland R, Berckmans RJ, McGregor S, Böing AN, Romijn FP, Westendorp RG,et al. Cellular origin and procoagulant properties of microparticles in meningococ-cal sepsis. Blood 2000;95:930–5.

[20] Jimenez JJ, Jy W, Mauro LM, Horstman LL, Ahn YS. Elevated endothelial micropar-ticles in thrombotic thrombocytopenic purpura: findings from brain and renal mi-crovascular cell culture and patients with active disease. Br J Haematol 2001;112:81–90.

[21] Shet AS, Aras O, Gupta K, Hass MJ, Rausch DJ, Saba N, et al. Sickle blood containstissue factor-positive microparticles derived from endothelial cells and mono-cytes. Blood 2003;102:2678–83.

[22] Simak J, Holada K, Risitano AM, Zivny JH, Young NS, Vostal JG. Elevated circulatingendothelial membrane microparticles in paroxysmal nocturnal haemoglobinuria.Br J Haematol 2004;125:804–13.

[23] Pamuk GE, Vural O, Turgut B, Demir M, Umit H, Tezel A. Increased circulatingplatelet-neutrophil, platelet-monocyte complexes, and platelet activation in pa-tients with ulcerative colitis: a comparative study. Am J Hematol 2006;81:753–9.

[24] Sheremata WA, Jy W, Horstman LL, Ahn YS, Alexander JS, Minagar A. Evidence ofplatelet activation in multiple sclerosis. J Neuroinflammation 2008;27:5–27.

[25] Zwicker JI. Tissue factor-bearing microparticles and cancer. Semin ThrombHemost 2008;34:195–8.

[26] Nordström M, Lindblad B, Bergqvist D, Kjellström T. A prospective study of the in-cidence of deep-vein thrombosis within a defined urban population. J Intern Med1992;232:155–60.

[27] Bezemer ID, Bare LA, Doggen CJ, Arellano AR, Tong C, Rowland CM, et al. Gene var-iants associated with deep vein thrombosis. JAMA 2008;299:1306–14.

[28] Chirinos JA, Heresi GA, Velasquez H, Jy W, Jimenez JJ, Ahn E, et al. Elevation of en-dothelial microparticles, platelets, and leukocyte activation in patients with ve-nous thromboembolism. J Am Coll Cardiol 2005;45:1467–71.

597P. Bucciarelli et al. / Thrombosis Research 129 (2012) 591–597

[29] Tesselaar MET, Romijn FPHTM, van der Linden IK, Prins FA, Bertina RM, Osanto S.Microparticle-associated tissue factor activity: a link between cancer and throm-bosis? J Thromb Haemost 2007;5:520–7.

[30] TesselaarMET, Romijn FPHTM, van der Linden IK, Bertina RM, Osanto S.Microparticle-associated tissue factor activity in cancer patients with and without thrombosis. JThromb Haemost 2009;7:1421–3.

[31] Ay C, Freyssinet JM, Sailer T, Vormittag R, Pabinger I. Circulating procoagulantmicro-particles in patients with venous thromboembolism. Thromb Res 2009;123:724–6.

[32] Frezzato M, Tosetto A, Rodeghiero F. A validated questionnaire for the identifica-tion of previous personal or familial venous thromboembolism. Am J Epidemiol1996;143:1257–65.

[33] von Elm E, Altman DG, Egger M, Pocock SJ, Gøtzsche PC, Vandenbroucke JP. STROBEInitiative. Strengthening the Reporting of Observational Studies in Epidemiology(STROBE) statement guidelines for reporting observational studies. BMJ 2007;335:806–8.

[34] Lane DA, Mannucci PM, Bauer KA, Bertina RM, Bochkov NP, Boulyjenkov V, et al.Inherited thrombophilia: Part 1. Thromb Haemost 1996;76:651–62.

[35] Harrell Jr FE. Regression modeling strategies. New York: Springer; 2001. p. 16–26.[36] Hron G, Kollars M, Weber H, Sagaster V, Quehenberger P, Eichinger S, et al. Tissue

factor-positive microparticles: cellular origin and association with coagulation ac-tivation in patients with colorectal cancer. Thromb Haemost 2007;97:119–23.

[37] Tilley RE, Holscher T, Belani R, Nieva J, Mackman N. Tissue factor activity is in-creased in a combined platelet and microparticle sample from cancer patients.Thromb Res 2008;122:604–9.

[38] van den Goor JM, van den Brink A, Nieuwland R, van Oeveren W, Rutten PM,Tepaske R, et al. Generation of platelet-derived microparticles in patients under-going cardiac surgery is not affected by complement activation. J Thorac Cardio-vasc Surg 2003;126:1101–6.

[39] Zahra S, Anderson JAM, Stirling D, Ludlam CA. Microparticles, malignancy andthrombosis. Br J Haematol 2011;152:688–700.

[40] Jy W, Horstman LL, Jimenez JJ, Ahn YS, Biró E, Nieuwland R, et al. Measuring circu-lating cell-derived microparticles. J Thromb Haemost 2004;2:1842–51.

[41] Brambilla M, Camera M, Colnago D, Marenzi G, De Metrio M, Giesen PL, et al. Tissuefactor in patients with acute coronary syndromes: expression in platelets, leuko-cytes, and platelet-leukocyte aggregates. Arterioscler Thromb Vasc Biol 2008;28:947–53.

[42] Nomura S, Takahashi N, Inami N, Kajiura T, Yamada K, Nakamori H, et al. Probucoland ticlopidine: effect on platelet and monocyte activation markers in hyperlipi-demic patients with and without type 2 diabetes. Atherosclerosis 2004;174:329–35.

[43] Sustar V, Jansa R, Frank M, Hagerstrand H, Krzan M, Iglic A, et al. Suppression ofmembrane microvesciculation - a possible anticoagulant and antitumor progres-sion effect of heparin. Blood Cells Mol Dis 2009;42:223–7.

[44] Mohan IV, Jagroop IA, Mikhailidis DP, Stansby GP. Homocysteine activates plate-lets in vitro. Clin Appl Thromb Hemost 2008;14:8–18.