Recurrent Arterial Thrombosis Linked to Autoimmune Antibodies Enhancing von Willebrand Factor Binding to Platelets and Inducing FcγRII Receptor-Mediated Platelet Activation

A 37-year-old woman was referred to our hospital for further investigation of a history of thrombosis. At the age of 22 and 23 years, both pregnancies of this patient resulted in intrauterine death at 26 and 28 weeks, respectively, with evidence of fetal growth retardation and pathologically documented multiple placental infarcts on both occasions. At the age of 36, she developed an ischemic cerebrovascular accident, confirmed by computer tomography and magnetic resonance imaging. At the age of 37, she developed a myocardial infarction; both arterial thrombotic episodes were preceded by flu-like symptoms: moderate fever, arthralgia, and headache. Echocardiography showed hypokinesia of the interventricular septum, but no valvular abnormalities; she was not taking oral contraceptives. Angiography performed 1 month after the myocardial infarction showed normal coronary arteries. Antinuclear factor, lupus anticoagulant, anticardiolipin antibodies, and other thrombophilia parameters (antithrombin, protein C, protein S, activated protein C resistance, and hyperhomocysteinemia) were negative. The patient had moderate hyperthyroidism, with positive thyroid peroxidase antibodies. Her platelet-rich plasma (PRP) manifested spontaneous platelet aggregation, which disappeared with aspirin intake; however, aggregation with low-dose ristocetin (0.5 mg/mL) persisted (Table1). Ristocetin cofactor activity was 100% at the start of the observation, but decreased to 50% over less than 6 months. Factor VIII activity, on the contrary, remained normal throughout the study (Table 1). Treatment with thiamazol (antithyroid drug), temporary corticosteroids, aspirin, and ticlopidine failed to correct the hypersensitivity to ristocetin, but the patient is currently doing well.

Cryoprecipitate was made from normal and patient plasma following thawing overnight at 4°C. vWF was isolated from cryoprecipitate by reprecipitation and bentonite-mediated fibrinogen depletion, followed by gel filtration on a Sepharose 4B-CL column (2.6 × 95 cm, Pharmacia, Uppsala, Sweden) as described.13 On reducing sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE), the isolated vWF showed a single band. Monoclonal antibodies were isolated from ascites by protein A-chromatography. Human immunoglobulins (Igs) were also purified by protein A-chromatography, taking care to collect acid-eluted antibodies in 1 mmol/L Tris-HCl buffer, pH 8. To maximally avoid Ig aggregation during antibody preparation, patient Igs were also isolated via ion-exchange chromatography on a mono-Q Sepharose column, equilibrated in 20 mmol/L Tris-HCl buffer, pH 8.5. Bound Igs were eluted from this column via a linear NaCl gradient (0 to 250 mmol/L), pooled, dialyzed against phosphate-buffered saline (PBS), and concentrated via reverse osmosis against polyethylene glycol with an average M_r of 30,000.

Platelet aggregations were performed on a dual-channel Chrono-Log Aggregometer (Chronolog Corp, Havertown, PA). PRP was prepared from normal or patient blood collected on 0.38 mol/L citrate, by centrifugation at 150g for 10 minutes. Washed platelets were prepared by collecting blood from normal donors on 10% (vol/vol) of acid-citrate-dextrose (ACD) solution pH 6.2 (93 mmol/L Na₃-citrate, 7 mmol/L citric acid, 0.14 mmol/L dextrose) and by centrifugation at 150g for 20 minutes. The PRP was then mixed with an equal volume of ACD and platelets were pelleted by centrifugation at 1,000g for 10 minutes. Platelets were carefully resuspended in Hanks balanced salt solution (HBSS), pH 7.4 (0.44 mmol/L KH₂PO₄, 0.34 mmol/L Na₂HPO₄, 136 mmol/L NaCl, 5.4 mmol/L KCl, and 5.6 mmol/L d-glucose), plus a one-third volume fraction of ACD and platelets were washed again via centrifugation at 1,000g for 10 minutes. Platelets were then resuspended either in plasma or plasma mixtures, or in HBSS and mixed with vWF (0 to 10 μg/mL) or cryoprecipitate, in which the vWF concentration was adjusted to 10 μg/mL, and aggregation was induced by ristocetin. Neutralizing murine monoclonal antibodies, such as the anti-vWF monoclonal antibody AJvW-2 (Ajinomoto, Yokohama, Japan) up to 20 μg/mL,14 an antiglycoprotein Ib (GPIb) monoclonal antibody (G19H10, raised in our laboratory) up to 20 μg/mL, or the anti-FcγRII receptor monoclonal antibody IV.3 (Medarex, Annandale, NJ) up to 10 μg/mL, were added to PRP or to washed platelets before induction of aggregation by ristocetin. Spontaneous platelet aggregation was studied in an ELVI 840 dual-channel aggregometer (Pabisch, Brussels, Belgium) in citrated PRP at 1,000 rpm. Following resuspension of washed normal platelets in patient plasma, the spontaneous aggregation was investigated at 1,500 rpm in the presence of the indicated concentrations of G19H10, AJvW-2, and IV.3.

Washed normal platelets were resuspended in either normal plasma or patient plasma and stirred in an aggregometer at 1,000 rpm in the absence or the presence of 0.5 mg/mL ristocetin. After 2 minutes, 5 μL PRP was removed and fixed by the addition of 45 μL 1% formol in PBS for 1 hour, following which the remaining formaldehyde was neutralized by the addition of 50 μL 1-mol/L Tris-HCl buffer, pH 8. Platelet-associated human Igs were then revealed via addition of 400 μL of a 100-fold diluted fluorescein isothiocyanate (FITC)-labeled goat antihuman Ig antiserum (Becton-Dickinson, San José, CA) and acquisition on a FACS Calibur flow cytometer.

Blood (blood group O) anticoagulated with low-M_r heparin (2 IU/mL) was recirculated at 1,300 s⁻¹ through parallel-plate flow chambers with a height of 0.4 mm, in the presence of added normal or patient plasma (5% to 10%). Perfusions were performed for 2 minutes over thermonox coverslips coated overnight with calf-skin collagen (1 mg/mL in 50 mmol/L acetic acid) or vWF (25 μg/mL in PBS). Coverslips were removed from the perfusion chamber, rinsed in HEPES buffer, and fixed with 1% glutaraldehyde in PBS before staining with May-Grünwald-Giemsa as described previously.15 To minimize nonspecific platelet activation due to the recirculation of the anticoagulated blood, monoperfusion experiments were also performed through rectangular glass capillaries (0.2 × 2 mm internal diameters; Vitro Dynamics, Rockaway, NJ) at a shear rate equal to 1,000 s⁻¹. These ethanol-rinsed glass capillaries had been coated overnight at 4°C with calf-skin collagen or vWF as described earlier. The glass surface was blocked via perfusion with 1% bovine serum albumin dissolved in HEPES-buffered saline (HBS: 10 mmol/L HEPES, 150 mmol/L NaCl, pH 7.5) for 10 minutes. Blood was then aspirated through these capillaries by means of a Harvard Apparatus (Holliston, MA) 22 pump at a constant reverse-flow rate of 0.75 mL/min for 2 minutes. The capillaries were then perfused with HBS for 10 minutes and stained using May-Grünwald (10-minute perfusion)-Giemsa (1-hour perfusion), all by aspiration through the capillaries. Following final rinsing with HBS and fixation with methanol (10-minute perfusion), the capillaries were dried.

Quantitation of platelet adhesion was done by image analysis of the surface covered using en face light microscopy (Dialux 20 EB, E; Leitz, Wetzlar, Germany) at a low magnification (×40) and the TCL-Image image-processing software (Multihouse TSI, Amsterdam, The Netherlands).

Before treatment, stirring of the patient's PRP induced spontaneous platelet aggregation, with this aggregation reaching a plateau after 40 minutes (Fig 1). Platelet activation by stirring responded well to high-dose aspirin: 4 days after the daily intake of 160 mg, aggregation was reduced, and had disappeared almost entirely after 4 additional days of 1,000 mg/d (Fig 1). At the start of the study, patient PRP aggregations were normal when induced by ADP, collagen, the thromboxane A₂ analog U46619, and arachidonic acid (AA). In view of the treatment with aspirin, the AA-induced platelet aggregation was abolished entirely after 7 weeks. Initially, a full-blown aggregation was observed with ristocetin concentrations as low as 0.5 mg/mL, but was somewhat tempered following aspirin intake for 7 weeks, albeit the slope of the initial agglutination step remained unaltered (Table 1 and Fig 2A).The low-dose ristocetin aggregation remained strong throughout the entire follow-up period (Table 1), even at concentrations of ristocetin as low as 0.3 mg/mL (Fig 2B) and despite the fact that the ristocetin cofactor activity, ie, vWF antigen levels, decreased to approximately 50% at the end of the observation period (Table1).

Platelet aggregation induced by 0.3 and 0.5 mg/mL ristocetin was completely blocked by the neutralizing anti-vWF antibody AJvW-2 and by the neutralizing anti-GPIb antibody G19H10 (Fig3), with both monoclonal antibodies being specific inhibitors of the vWF-mediated platelet activation. The anti-FcγRII receptor antibody IV.3, although it had no effect on the ristocetin-induced agglutination of normal PRP, surprisingly reduced the weak aggregation induced by 0.3 mg/mL ristocetin and had a partial effect on the stronger aggregation observed with 0.5 mg/mL ristocetin (Fig 3B). Aggregations of patient PRP induced by 0.5 mg/mL ristocetin and performed in the presence of the GPIIb/IIIa antagonist G4120 also reduced the second wave of the aggregation curve, confirming that the addition of 0.5 mg/mL ristocetin to PRP induced both an initial agglutination (first wave), followed by a subsequent platelet activation step and platelet aggregation (second wave).

To exclude that the patient's platelets were responsible for the observed sensitivity to vWF, washed platelets were isolated from normal donors and resuspended in either normal or patient plasma. Ristocetin induced a dose-dependent agglutination/aggregation reaction in the reconstituted normal PRP (Fig 4A),comparable to that in normal PRP. In the reconstituted mixture of platelets and patient plasma, a normal response to AA was noted, but the increased sensitivity toward ristocetin could be confirmed. The partial inhibition by IV.3 of the platelet aggregation induced by 0.5 mg/mL of ristocetin could also be reproduced using normal platelets in combination with patient plasma (Fig 4B). When washed platelets were resuspended in mixtures of normal and patient plasma, maximal aggregation by 0.5 mg/mL of ristocetin was observed with 25% to 50% of patient plasma (Fig 5).

To evaluate the involvement of patient Igs, ristocetin-induced aggregation was studied using washed platelets, purified normal vWF, and protein A–purified Igs isolated from patient or normal plasma. At 0.15 mg/mL ristocetin, no agglutination developed, but the addition of purified patient Igs induced a dose-dependent increase of platelet activation (not shown). Optimal stimulation was observed between 0.25 and 1 mg/mL of added antibody, but at 2 mg/mL of antibody, virtually no aggregation was observed, suggestive of a bell-shaped dose-response curve. Control aggregations performed in the presence of Igs isolated from different batches of normal plasma would yield considerably weaker stimulation, although occasionally some aggregation was found, indicative of weak Ig aggregate-induced platelet activation. However, with vWF concentrations less than 5 μg/mL, aggregations gradually became weaker.

In an effort to avoid Ig aggregate-induced platelet activation, patient Igs were isolated via ion-exchange chromatography and platelet aggregation studies were performed with these antibodies. As shown in Fig 6, at 0.5 mg/mL, these antibodies enhanced the weak aggregation of washed platelets induced by 0.17 mg/mL ristocetin in the presence of vWF. Furthermore, this aggregation could be inhibited almost completely by the anti-vWF monoclonal antibody AJvW-2, the anti-GPIb monoclonal antibody G19H10, and the anti-FcγRII receptor antibody IV.3. The double involvement of vWF and Igs in the aggregation process was further confirmed by the lack of aggregation in the absence of ristocetin or of vWF. The low residual aggregation observed after 10 minutes in the absence of ristocetin is compatible with the formation of vWF-Ig complexes with some affinity for the platelet Fc receptor. However, the lack of platelet aggregation in the absence of vWF excludes Ig aggregates as a cause of platelet activation. When the preincubation time of platelets, vWF, and antibody was prolonged from 0 to 10 minutes, a clear enhancement of the initial platelet response to ristocetin was observed (not shown), also compatible with immune complex formation between vWF and Igs.

When normal platelets were resuspended in patient plasma and stirred at 1,500 rpm, a mild spontaneous aggregation could be induced, with approximately 40% aggregation after 45 minutes. In the presence of IV.3, this aggregation was blocked entirely, but the slope of the aggregation was also reduced by 62% by the anti-GPIb antibody G19H10 and by 78% by the anti-vWF antibody AJvW-2. These experiments confirm that the weak aggregation observed after 10 minutes, shown in Fig 6 for mixtures of vWF and patient Igs, is caused by the simultaneous binding to platelets of vWF and Igs, even in the absence of further mediators of vWF binding. The presence on the platelet membrane surface of such complexes was further investigated via flow cytometry. Stirring washed platelets resuspended in patient plasma for 2 minutes reduced single platelet numbers to 27% of initial values. In the presence of 0.5 mg/mL ristocetin, this number dropped further to 19%. Resuspension in normal plasma and stirring, on the contrary, reduced single platelet numbers only to 52% and 53%, respectively, in agreement with a weaker platelet activation. In addition, whereas the median antihuman Ig antibody-bound fluorescence for the remaining single platelets in normal plasma did not change in the absence (relative value, 28) and the presence (relative value, 26) of ristocetin, the median fluorescence for the single platelets resuspended in patient plasma (27 in the absence of ristocetin) increased to 36 following stirring in the presence of ristocetin. These data indicate that the patient vWF binding induced by 0.5 mg/mL ristocetin is accompanied by binding of patient Igs to the platelet.

Cryoprecipitate was made from normal and patient plasma and adjusted to 10 μg vWF/mL. The cryoprecipitate from the patient was much more capable of supporting low-dose ristocetin-induced aggregation of washed platelets than normal cryoprecipitate, and this effect was inhibited by IV.3. Because both the patient vWF antigen levels and ristocetin cofactor activity were normal at the start of the study, these findings suggested that the patient cryoprecipitate was enriched in vWF-IgG immune complexes, supporting aggregation.

Reperfusion experiments under shear stress, more representative of physiologic vWF-dependent platelet activation than ristocetin, were performed in low-M_r heparin anticoagulated blood with normal or patient plasma added to 5% and 10% of the volume. There was increased platelet adhesion to a collagen surface in the presence of 5% but not 10% patient plasma; platelet adhesion to vWF was enhanced in the presence of 10%, but not yet at 5% patient plasma (Table2).

To investigate these observations further, monoperfusions of anticoagulated blood were performed at 1,000 s⁻¹ through glass capillaries coated with collagen or vWF. These monoperfusions at moderately high shear forces through rectangular capillaries coated with calf-skin collagen resulted in the adhesion of single platelets and of small to moderately sized platelet aggregates (Fig7a) that contained up to 20 platelets, as evaluated by area analysis. In this setup, substitution of 5% normal plasma by 5% patient plasma resulted in the formation of larger platelet aggregates (Fig 7b), without increase in the degree of adhesion (normal plasma, 18.3% ± 1.8%; patient plasma, 20% ± 5.1%). Although monoperfusions performed over vWF yielded a less homogeneous platelet adhesion, substitution of 10% normal plasma by patient plasma resulted in an increase of the degree of platelet adhesion from 7.9% ± 0.6% to 14.1% ± 2.6% (Fig 7c and d, P < .05), but no significant differences in the size of the aggregates could be detected (∼50% of platelets adhered either as single platelets or as very small aggregates consisting of two to three platelets only). Substitution of 20% normal plasma with patient plasma no longer showed any difference either in the degree of adhesion or in the size of the aggregates formed.

The patient reported in this study was referred to our hospital following a history of recurrent fetal loss and arterial thrombosis. The usual risk factors associated with fetal loss or thrombosis such as activated protein C resistance, antiphospholipid antibodies, and hyperhomocystinemia could all be excluded. However, the patient's platelets aggregated spontaneously on stirring; spontaneous aggregation has previously been linked to thrombocythemia,16 which was not present in this patient; it disappeared with aspirin, although initially only when given at high doses.

Platelet aggregation by low ristocetin concentrations persisted throughout aspirin treatment; mutations in vWF enhancing the binding of vWF to its receptor on the platelet, the GPIb/IX/V complex, and giving rise to von Willebrand disease (vWD) type IIb,17,18 could be excluded as a basis for the observed phenomenon, because the patient had no bleeding tendency and had a normal ristocetin cofactor, as well as a normal platelet count; furthermore, the enhanced aggregation was reproduced in a purified system that contained normal, not patient, vWF. Alternatively, mutations in GPIb itself, equally giving rise to enhanced vWF binding and a platelet type vWD,19 were also excluded, since low-dose ristocetin-induced aggregation could be reproduced with normal platelets when resuspended in patient plasma.

Evidence is provided that vWF-antibody complexes may explain the observed low-dose ristocetin-induced aggregations. Aggregations performed in patient plasma could be inhibited completely by neutralizing antibodies reactive with vWF or with GPIb and could be blocked partially by the monoclonal antibody IV.3. By binding to the platelet FcγRII receptor, this monoclonal antibody prevents Fc-mediated interactions and signal transduction leading to platelet activation.20 The finding that IV.3 primarily blocked the second phase of the low-dose ristocetin-induced aggregation curves is compatible with this interpretation. Normal plasma enhanced the response to patient plasma, according to a bell-shaped curve, a finding that is compatible with optimization of immune complex formation as a consequence of plasma mixing. The data with patient cryoprecipitate and normal washed platelets further support the concept of vWF-antibody complexes being responsible for enhanced response to low-dose ristocetin.

The interpretation of results obtained with Igs isolated via acid elution was somewhat more difficult, since the isolated Igs in some batches tended to aggregate and promote aggregation by themselves. However, ion-exchange isolated patient antibodies supported the aggregation of washed platelets only in the presence of vWF and low concentrations of ristocetin. Furthermore, aggregations were dependent on the concentration of added vWF and could almost completely be inhibited by the vWF neutralizing antibody AJvW-2, the GPIb neutralizing antibody G19H10, or the Fc receptor neutralizing antibody IV.3.

It appears that the patient's autoimmune antibodies facilitate vWF binding to platelets. This occurs to some extent even in the absence of further mediators of vWF binding, a finding that may explain the spontaneous aggregation also observed following resuspension of normal platelets in patient plasma. The inhibition by IV.3 of the spontaneous aggregation confirms the involvement of immune complexes; in addition, the substantial inhibition by G19H10 and AJvW-2 confirms that aggregation not only depends on the binding of vWF-Ig complexes to the FcγRII receptor, but that it involves vWF-GPIb–dependent interactions. We have observed previously that a murine monoclonal anti-vWF antibody is indeed capable of enhancing the binding of vWF to its receptor on the platelet,21 a finding that may explain why in the present study antibody binding to vWF facilitates vWF binding to GPIb. Once the vWF-antibody complexes have become bound to platelets, the Fc part of the Ig would then interact with the FcγRII receptor and activate the platelet, as outlined in Fig8. Alternatively, antibody bound to vWF may first interact with the Fc receptor, following which the binding of vWF to GPIb will be facilitated, as evidenced by the faint and slow aggregation of washed platelets observed in the absence of ristocetin. Finally, as recently proposed,22 a quaternary complex is formed between platelet receptors, vWF, and antibodies, stabilized by multiple interactions, and in which platelet activation occurs primarily via the Fc receptor (Fig 8). During the “spontaneous” platelet activation, stirring in the presence of antibody alone would then suffice to prompt binding of vWF-Ig complexes, followed by subsequent antibody-mediated platelet activation, which can be suppressed by aspirin.

The anti-vWF antibodies would have low affinity, which would not allow the direct demonstration of immune complex formation by enzyme-linked immunosorbent assay (ELISA), Western blotting, or surface plasmon resonance interaction studies. This low affinity is compatible with the aforementioned model in which both vWF and antibody need to be stabilized via secondary interactions on platelet receptors to form a more stable quaternary complex. This interpretation would support our results obtained via flow cytometry, the only approach that enabled us to show that increased vWF binding was associated with human Ig binding to the platelet. The low affinity is compatible with the normal plasma vWF levels found in the patient, at least at the beginning of the follow-up period, and also would explain her higher platelet numbers in comparison to other situations of immune-mediated thrombocytopenia. Shortly after the onset of treatment, platelet numbers tend to increase, in parallel with an improved prevention of Fc-mediated platelet activation. These findings are compatible with a diminished platelet consumption following initiation of treatment. In view of the continued sensitivity to low-dose ristocetin-induced aggregation and the persistence of anti-vWF antibodies, higher platelet numbers in turn may have an impact on the clearance of vWF via platelet-bound immune complexes, explaining the small drop in vWF antigen levels and ristocetin cofactor activity toward the end of the follow-up period.

Thrombosis would be facilitated by optimal proportions of antigen and antibody, as was demonstrated by varying the antigen/antibody ratios during the ristocetin-induced platelet aggregation studies. In this regard, it is remarkable that both episodes of arterial thrombosis were immediately preceded by flu-like symptoms, possibly suggesting circulating immune complexes at that moment. At moderate shear stress, the antibody could be expected to somewhat stimulate platelet adhesion to vessel wall–exposed vWF, as was observed during the perfusion experiments over collagen and vWF-coated surfaces. Furthermore, it could also be expected to promote platelet thrombus growth. Within limited dose ranges, this was observed.

These observations suggest the existence of a new type of antibody-mediated thrombosis, characterized by the production of autoantibodies to vWF, in which the platelet FcγRII receptor plays a central role both by stabilizing platelet-bound vWF-Ig complexes and by participating in platelet activation. This type of thrombosis bears a high mechanistic similarity to other types of antibody-mediated thrombosis, such as heparin-induced thrombocytopenia and perhaps the antiphospholipid syndrome.

We thank M. Vanrusselt for performing the initial platelet aggregations reported in this study, and I. Vreys for doing the perfusion studies. J. Vermylen is holder of the “Dr J. Choay Chair in Haemostasis Research.”