Close Approximation of Two Platelet Factor 4 Tetramers by Charge Neutralization Forms the Antigens Recognized by HIT Antibodies
Objective— Heparin-induced thrombocytopenia (HIT) is a prothrombotic drug reaction caused by antibodies that recognize positively charged platelet factor 4 (PF4), bound to the polyanion, heparin. The resulting immune complexes activate platelets. Unfractionated heparin (UFH) causes HIT more frequently than low-molecular-weight heparin (LMWH), whereas the smallest heparin-like molecule (the pentasaccharide, fondaparinux), induces anti-PF4/heparin antibodies as frequently as LMWH, but without exhibiting cross-reactivity with these antibodies. To better understand these findings, we analyzed the molecular structure of the complexes formed between PF4 and UFH, LMWH, or fondaparinux.
Methods and Results— By atomic force microscopy and photon correlation spectroscopy, we show that with any of the 3 polyanions, but in the order, UFH>LMWH≫fondaparinux—PF4 forms clusters in which PF4 tetramers become closely apposed, and to which anti-PF4/heparin antibodies bind. By immunoassay, HIT antibodies bind strongly to PF4/H/PF4 complexes, but only weakly to single PF4/heparin molecules.
Conclusion— HIT antigens are formed when charge neutralization by polyanion allows positively charged PF4 tetramers to undergo close approximation. Whereas such a model could explain why all 3 polyanions form antibodies with similar specificities, the striking differences in the relative size and amount of complexes formed likely correspond to the observed differences in immunogenicity (UFH>LMWH≈fondaparinux) and clinically relevant cross-reactivity (UFH>LMWH≫fondaparinux).
Heparin is the most frequently used parenteral anticoagulant, but can induce a life-threatening adverse drug effect, heparin-induced thrombocytopenia (HIT).1,2 HIT is caused by antibodies against platelet factor 4 (PF4), a tetrameric chemokine secreted by human platelets.3 These antibodies recognize multimolecular complexes of positively charged PF4 and polyanions4,5 such as heparin.3,6–9 Whereas highly sulfated polyanions form this antigen, low-sulfated glycosaminoglycans such as heparan sulfate or dermatan sulfate10 do not, although they also bind PF4 and can interact with the PF4/heparin complexes, as shown by their capacity to inhibit antigen formation.10,11 When HIT antibodies bind to PF4/heparin complexes, the immune complexes activate platelets via FcγIIa receptors,2 leading to increased thrombin generation. HIT is therefore a prothrombotic disorder, with 50% or more of affected patients developing thrombosis.12 HIT can occur during treatment with either unfractionated heparin (UFH) or low-molecular-weight heparin (LMWH).13,10 Antibodies from these patients typically show strong in vitro cross-reactivity with either UFH or LMWH regardless of which heparin initiated the immune response.13,10,14 However, in postoperative patients, UFH induces HIT ≈10 times more often than does LMWH.15 In contrast to UFH (15 000 Da) and LMWH (4000 to 6000 Da), the smaller (1728 Da) pentasaccharide, fondaparinux,16 consistently lacks in vitro cross-reactivity with anti-PF4/heparin antibodies, ie, when PF4 is incubated with fondaparinux, HIT antibodies do not bind to these complexes.14,17,18 This indicates that PF4/fondaparinux complexes do not form the HIT antigen. Indeed, fondaparinux has even been used successfully to anticoagulate patients with HIT.19 Based on its small size, the lack of in vitro cross-reactivity with HIT antibodies, and early favorable experience for treating HIT, we and others believed that fondaparinux was unlikely to induce anti-PF4/heparin antibodies and/or HIT.20 However, in large clinical trials,21,22 fondaparinux induced anti-PF4/polyanion antibodies as often as did LMWH.23 Although, to date, no patient has been reported to have developed HIT with fondaparinux, this would be likely to occur if a patient immunized through fondaparinux treatment were then treated with UFH or LMWH.
We investigated potential explanations for this apparent dissociation of immunogenicity and cross-reactivity among these 3 anticoagulants (UFH, LMWH, fondaparinux), namely why UFH induces more HIT than LMWH, and why fondaparinux forms little antigen in vitro while apparently being as immunogenic as LMWH in vivo. We used atomic force microscopy (AFM) and photon correlation spectroscopy (PCS) to search for structural similarities among these different PF4/polyanion complexes to explain their common property of being able to trigger formation of anti-PF4/heparin antibodies. We also used quantitative immunoassays to compare HIT antibody reactivity against the different PF4/polyanion complexes. Our findings suggest that close approximation of 2 PF4 tetramers by polyanion-induced charge neutralization is crucial for forming the antigen site(s) recognized by HIT antibodies. Differences in the relative amounts and size of these complexes could be the basis for dissociation of immunogenicity and cross-reactivity observed with these anticoagulants in clinical studies. These features could represent a model for other autoimmune diseases, in which a small molecule might induce an immune response without necessarily evincing pathogenicity unless other circumstances expose the antigen in quantities sufficient to produce disease.
Materials and Methods
Detailed methods are available in the online data supplement, at http://atvb.ahajournals.org.
Atomic Force Microscopy
PF4, UFH, LMWH, and fondaparinux alone or in combination were incubated with freshly cleaved mica. Mica coated with PF4-UFH, PF4-LMWH, or PF4-fondaparinux was incubated with IgG-fractions (5 μg/mL) from either HIT or control serum.
AFM measurements were performed in tapping mode to avoid mechanical alterations of the proteins in air.
Particle Size Measurements by Photon Correlation Spectroscopy
Particle size was determined by PCS as mean hydrodynamic diameter (MHD). Titrations with fondaparinux were performed after preincubation of 10 μg/mL PF4 with 3.3 μg/mL UFH.
Consecutive Adsorption of PF4 and Heparin in the Enzyme-Linked Immunosorbent Assay
To distinguish whether the antigens on PF4 form when heparin binds to a single PF4 molecule (ie, by a conformational change within the PF4 tetramer), or whether at least 2 PF4 molecules form the antigen, microtiter plates were coated in 4 ways. First, PF4 was coated covalently alone via a spacer. Second, PF4 was covalently coated, washed, and then incubated with UFH and washed again. Third, PF4 was covalently coated, washed, incubated with UFH, washed, and again incubated with PF4 to allow formation of a PF4-heparin/PF4 “sandwich,” and compared with PF4 incubated with UFH in solution, with the complexes then coated. Antibody binding was assessed by enzyme-linked immunosorbent assay (ELISA).24
Experiments Using Charge Neutralization of PF4 by Small Molecules
PF4 coated to polystyrene beads in the absence of heparin but in the presence of charge neutralizing glycine/boric acid and control beads (both beads were kindly provided by Dr Ray Akers; Akers Biosciences, Thorofare, NJ) were used to test whether HIT antibodies recognize PF4 when its charge had been neutralized.
14C-serotonin release assay was performed as described.25,26
Solid-phase PF4-polyanion ELISA was performed as described.5,24 In some experiments, fondaparinux (25 μg/mL) was substituted for heparin.
Comparisons between groups were made using the Wilcoxon rank-test.
Atomic Force Microscopy
UFH and LMWH adsorbed to the mica as small particles (Figure 1a and 1b), with a maximal height (1 nm) similar to individual glucose molecules (0.7 nm). These particles resembled other sulfated linear polyanions such as polystyrene sulfonate, which adsorb to mica as collapsed flat coils.27,28 This suggests that the heparin molecules also adsorbed as collapsed flat coils to the mica. As the resolution of the Z-axis is 1000-fold higher than the X-axis and Y-axis, the flat coils appear as peaks. The lengths of the heparin molecules are reflected by the diameters of the coils, corresponding to the diameters of the peaks at half-height (Figure 1). The diameters of the peaks were 15 to 20 nm for UFH molecule coils and 12 to 15 nm for LMWH coils. Calculated by Chem3D-Ultra 9.0 software, this corresponds to 8 to 64 nm length for UFH molecules and to a rather more constant length of ≈10 nm for LMWH molecules. This agrees with the theoretical length of heparin molecules (based on ≈350 Da mass and 0.7 nm diameter/glucose moiety). However, as we dried heparin onto the mica, structural artifacts are possible. We could not find fondaparinux molecules bound to mica using the same technique, most likely because of their uniformly very small size.
PF4 tetramers alone rapidly covered the mica surface, with individual structures 4 nm high, separated by an average of 20 nm (<5% of molecules had an intervening distance of 10 to 20 nm, and in none of >250 measurements were 2 PF4 molecules closer than 10 nm; Figures 1c and 3⇓).
PF4 (20 μg/mL) preincubated with heparin (3.3 μg/mL; this concentration showed optimal binding in the PF4/heparin ELISA) adsorbed within 1 hour to mica. Nearly all PF4 tetramers were organized as large complexes (160 to 200 complexes/1 μm2-field) with a uniform height (2.9±0.3 nm) and a constant width (26±5 nm). The individual complexes appeared as continuous ridges with slight height variations (±0.3 nm peak-to-valley between sequential PF4 tetramers) (Figures 2a and 3⇑). We term this characteristic profile “ridge-like.” The complexes attached to the mica without overlapping, and were thus typically separated along their main axis by an intervening space. The complexes were up to 200 nm long. The large individual complexes chosen in each panel for further analysis (columns C and D) show typical examples from 3 identical experiments. When PF4 was incubated with UFH at very high concentrations (660 μg/mL), individual structures 4-nm high were observed, consistent with the expected appearance of single PF4 tetramers saturated by heparin (Figure 1d).
When PF4/LMWH complexes adsorbed to the mica, ≈50% of PF4 molecules formed ridges (65 to 110 ridges/1 μm2-field), resembling the structure of PF4/UFH complexes (height 3.4±0.3 nm; width 26±1 nm) but were shorter (Figures 2b and 3⇑).
PF4/fondaparinux covered the mica surface rapidly (5 minutes) with small particles, with a slightly larger uniform height (4.0±1 nm) than the PF4/UFH complexes (Figure 2c). Most of the particles appeared loosely associated in a nonstructured manner, but still closer together than PF4 molecules alone. To better visualize the close approximation, we show by pseudocolor the upper height range of 3.6 to 4.0 nm, ie, the tips of the PF4 molecules in the presence and absence of fondaparinux (Figure 4). Although the molecules began to cluster (supplemental Figure I, available online at http://atvb.ahajournals.org), in contrast to PF4/UFH and PF4/LMWH complexes, most PF4 molecules retained their individual appearance as separate peaks with intervening grooves (≥0.5 nm) (Figure 2c, column A). When the cross-sectional profiles of individual PF4 tetramers are superimposed on the longitudinal section (line 1 and 2 in Figure 2c, column C), each peak can be attributed to a single PF4 tetramer (Figure 2c, column C). This is clearly different from the longitudinal profile of PF4/UFH complexes (Figure 2a, column D).
Importantly, we observed a few PF4 ridges in the presence of fondaparinux (2 to 3/1 μm2-field) with very minor height variation of ±0.3 nm (Figure 2c, column D, line 3 and 4, and Figure 3) that resembled the structure of PF4/UFH complexes (Figure 2a, column D). These ridge-like structures were never observed with PF4 coated alone. Thus, the ridge-like structures consisting of closely-packed PF4 tetramers are a unique feature of PF4 presentation in the presence of any of the three polyanions (UFH, LMWH, or fondaparinux).
Size Distribution of PF4/Polyanion Complexes
The PF4/UFH and PF4/LMWH complexes of 100 nm are longer than individual heparin molecules (UFH: 8 to 65 nm, peak 25 to 30 nm; LMWH: mean ≈10 nm). However, the straight segments within the individual complexes corresponded more to the expected length distribution of UFH molecules. Thus, several PF4/UFH or PF4/LMWH complexes can attach end-to-end to each other (supplemental Figure I). Determination of the sizes of the PF4/fondaparinux complexes was difficult as the molecules attached so densely on the mica. Thus the complexes >30 nm given in supplemental Figure I may represent in part artifacts or they are aggregates of PF4 clustering together in a less structured way than PF4/UFH complexes.
Binding of HIT Antibodies to PF4/UFH or PF4/LMWH Complexes
IgG from HIT sera bound to the ridge-like complexes. Structures 10 to 12 nm high and ≈100 nm wide presumably represent IgG molecules aligned along the PF4/UFH (Figure 5A to 5C) and PF4/LMWH ridges (Figure 5D to 5E). They did not bind to loosely-clustered PF4 molecules between the ridge-like structures. When PF4 was incubated with fondaparinux, structures consistent with bound antibodies were occasionally observed (not shown), but could not be clearly distinguished from controls (Figure 5F).
Photon Correlation Spectroscopy
Photon correlation spectroscopy (PCS) also showed the presence of large complexes of PF4 with UFH, LMWH, and fondaparinux. PF4 alone had a MHD of ≈5 nm. Addition of polyanion (UFH, LMWH, fondaparinux) in increasing amounts to a fixed concentration of PF4 resulted in progressively increasing MHDs. However, beyond the optimal stoichiometric ratio (UFH:PF4=0.7; LMWH:PF4=0.3; fondaparinux:PF4=0.9 to 5.5), the MHDs progressively decreased (supplemental Figure IIa). High concentrations of UFH, LMWH, and fondaparinux (>100 μg/mL; supplemental Figure IIb) completely disrupted PF4/UFH complexes that had already formed at optimal stoichiometric concentrations. In line with the AFM experiments this indicates that the multimolecular complexes between polyanion and PF4 form best with UFH, to an intermediate extent with LMWH, and in only a few cases between fondaparinux and PF4.
Solid-Phase ELISA and Consecutive Adsorption Studies
We next addressed the question why multimolecular complexes are required to express the HIT antigen. If the antigen is expressed only by a conformational change in PF4 induced by heparin, it should be present on single PF4 tetramers after heparin binding. Our hypothesis was that the HIT antigen is exposed on the ridge-like structures generated when 2 PF4 tetramers are brought into close approximation rather than by a conformational change induced when a single PF4 tetramer binds to heparin. We compared antibody binding to PF4 alone, to PF4/UFH, and to the PF4/UFH/PF4 “sandwich” (supplemental Figure III). We found minimal antibody binding to PF4 alone. If PF4 was covalently linked to a microtiter plate, then washed, and then UFH added, HIT antibody binding to these single “PF4/UFH” complexes increased minimally (<2-fold), compared with PF4 alone (n=10; mean optical density [OD], 0.471±0.224 versus 0.271±0.168; P=0.06). In contrast, strong binding of antibodies occurred when the above PF4–UFH complexes were again incubated with PF4 to create PF4/UFH/PF4 “sandwich” complexes (mean OD 1.272±0.525 versus 0.271±0.168 for PF4 alone; P<0.001; and versus PF4/UFH 0.471±0.224; P<0.001). Whereas the total amount of PF4/well was the same in all experiments, strong antibody binding was only seen when the conditions allowed close approximation of 2 PF4 molecules (ie, via formation of the PF4/UFH/PF4 sandwiches). Control sera gave ODs <0.1. This indicates that complexes of at least 2 PF4 tetramers are crucial for marked antibody binding.
Experiments Using Charge Neutralization of PF4 by Small Molecules
When HIT serum which gave an OD of ≈1.0 in the PF4/UFH ELISA was preadsorbed by PF4 (coated on polystyrene beads under charge neutralizing conditions) or control beads, OD after adsorption with PF4 coated beads decreased by 49% (±16%; n=4 experiments; P=0.021).
Fondaparinux Interaction With PF4-UFH Complexes and With PF4
Solid-phase ELISA showed a small increase in binding (n=26; OD 0.373±0.23 versus OD 0.268±0.151; P=0.03) of HIT antibodies to immobilized PF4/fondaparinux, compared with PF4 alone. However, the changes in OD were always below the usual cut-off of this assay [OD 0.5]), again indicating that very few antibody binding sites are exposed by fondaparinux. Additional evidence for an interaction of fondaparinux with the HIT antigen is that heparin-dependent platelet activation by HIT antibodies was inhibited by high concentrations (>100 μg/mL) of fondaparinux (supplemental Figure IV), paralleling the disruption of PF4/heparin complexes by excess fondaparinux in the PCS experiments (supplemental Figure IIb). This inhibitory effect was specific for HIT antibodies and was not seen with other antibodies26 that activate platelets via their Fc receptors.
This study of the structure of PF4/polyanion complexes provides potential explanations for the observed differences in immunogenicity (ie, frequency of induction of anti-PF4/polyanion antibodies) and in cross-reactivity (ie, the in vitro ability of antibodies to bind to PF4/polyanion complexes, and the corresponding in vivo capacity to cause thrombocytopenia) among the 3 anticoagulant polyanions, UFH, LMWH, and fondaparinux.
In the presence of UFH, nearly all PF4 molecules formed large linear complexes (Figure 2, upper panel, and Figure 3) in which individual PF4 tetramers could no longer be readily discerned. In contrast, with LMWH, PF4 formed shorter linear complexes, but also smaller less-organized clusters (Figures 2b and 3⇑). Purified HIT-IgG antibodies bound to these PF4/UFH and PF4/LMWH complexes (Figure 5) demonstrating directly that they express the HIT antigens. PF4 also formed a few of these ridge-like structures in the presence of fondaparinux (Figure 2c, column D; Figure 3), but not in the absence of polyanions (Figures 1 and 3⇑). As the interaction of several PF4 molecules to form these ridge-like complexes, in which the contours of the individual PF4 tetramers become less distinct (Figure 3) is the common structural feature observed when PF4 interacts with either UFH, LMWH, or fondaparinux, we suggest that underlying close approximation of at least 2 PF4 tetramers could be a key event in leading to formation of the antigen(s) recognized by HIT antibodies.
The smaller PF4/LMWH/IgG complexes should cross-link fewer of the platelet FcγRIIa receptors, thereby activating platelets less than the larger PF4/UFH/IgG complexes. This is a likely reason for the reduced breakthrough of HIT-related thrombocytopenia and thrombosis with LMWH treatment among immunized patients.
The few PF4/fondaparinux/IgG complexes are most likely insufficient to induce activation of many platelets via cross-linking of the platelet Fc receptors. This could explain why patients immunized either with UFH, LMWH, or fondaparinux remain clinically asymptomatic when treated with fondaparinux.19 However, this also indicates that HIT might occur rapidly if a patient who is immunized through fondaparinux treatment is then treated with UFH or LMWH. Disruption of PF4/heparin complexes (Figure 1b and supplemental Figure IIb) and inhibition of the platelet-activating effects of anti-PF4/heparin antibodies in the presence of UFH (supplemental Figure IV) at very high concentrations of fondaparinux is further evidence that fondaparinux interacts with PF4.
In vivo, fondaparinux induces HIT antibodies as often as LMWH.23 However, in vitro, the antibodies that are formed only bind to PF4/LMWH and to PF4/UFH, but not to PF4/fondaparinux complexes. Our findings provide a potential explanation for this paradox.
The small noncomplexed PF4 tetramers have the larger diffusion coefficient, and therefore adsorb faster to a surface than larger complexes (Stokes-Einstein equation29). They rapidly covered the mica surface, which made it very difficult to find these large ridge-like PF4/fondaparinux complexes as separate structures as shown in Figure 3. Also, in solid-phase ELISA assays, the small, non-complexed PF4 molecules competitively coat the microtiter plate much more rapidly than the few large PF4/fondaparinux complexes and therefore HIT antibodies usually do not show cross-reactivity. Another explanation for the discrepant results between AFM experiments showing few large PF4/fondaparinux complexes and other methods could be that in the AFM experiments the mica served as a negatively-charged template facilitating formation of the PF4/fondaparinux complexes. This would not occur in ELISAs where the antigen is generated in the fluid phase23 or attached via a spacer to the plate and would also explain why these complexes were not seen by electron microscopy (EM).30
Rauova et al30 recently assessed PF4/heparin complexes using EM, while we used AFM. AFM is 10 times superior to EM in height resolution. This allowed us to provide new insights into the structure of the HIT antigen.3,6,8,9,30 We show that PF4 molecules attach to each other forming ridge-like structures in the presence of negatively-charged polyanions, including fondaparinux. AFM allowed us to show that within these ridge-like structures, PF4 molecules lose their minimal distance from one another, which results in the loss of the 0.2 to 0.3 nm deep indentations between the single PF4 molecules. As antibodies usually recognize epitopes consisting of less than 10 amino acids, it is very unlikely that those recognized by HIT antibodies on the multimolecular PF4/heparin complexes are formed by more than 2 PF4 tetramers. Based on the AFM experiments, we suggest that the HIT antigen is primarily generated by close approximation of 2 PF4 tetramers (<10 nm), which we only observed in the presence of polyanions (Figure 2). This, however, also implicates that all strongly negatively charged drugs should have the potential to interact with PF4 in a way to induce the HIT antigen. Close approximation of PF4 is unlikely to occur in vivo in the absence of polyanions, due to the strong electrostatic repulsion (zeta potential) of the positively charged PF4 tetramers. The experiments comparing HIT antibody binding to PF4 under four conditions of consecutive adsorption of PF4 and heparin showed that the HIT antigen is formed by PF4-PF4 proximity mediated by polyanion binding (supplemental Figure III). This is different than the current model of exposure of the antigen by a conformational change within a single PF4 tetramer after heparin binding. This should also occur when heparin binds to one PF4 tetramer, which was obviously not the case in the ELISA experiments using sequential incubation of PF4 and heparin. Adsorption of PF4/heparin antibodies by beads coated with PF4 in the presence of charge neutralizing small molecules (glycine/boric acid) further corroborates our model of generating the HIT antigen by charge dependent approximation of PF4. Whether in addition to close approximation of PF4 tetramers, additional antigenically-relevant conformational changes also occur in the PF4 tetramers is unresolved. The slight reduction in the height profile of PF4 tetramers (4.0 nm) after forming ridge-like structures with UFH (2.9±0.3 nm), which was less evident with PF4 binding to LMWH (3.4±0.3 nm) or to fondaparinux (3.4 to 3.5 nm; Figure 2), indicates that in all likelihood both conformational changes and close approximation of 2 PF4 tetramers are important for exposing the HIT antigen(s).
The proposed concept fits recent findings31 showing enhanced binding of a mouse monoclonal anti-PF4/heparin antibody to platelets expressing high amounts of PF4 on their surface and previous findings identifying several amino acids in the PF4 molecule as essential for the HIT antigen.6–8 Amino acid substitutions at these positions might very well disturb formation of the ridge-like structures we observed, as they are located within the upper and lower polar region of the PF4 tetramer, which are expected to be important for the upper surface of the complexes.
In summary, our studies provide insights into why UFH, LMWH, and fondaparinux produce the same types of antibodies (via inducing close approximation of PF4 tetramers) while at the same time showing clear differences in the frequency of clinically manifest HIT (UFH > LMWH > fondaparinux) based on the different sizes and amounts of their respective complexes formed with PF4. Potentially, this is also a model for other autoimmune diseases, in which a small molecule might induce an immune response without necessarily being pathogenic, unless other circumstances lead to exposure of the antigen in quantities sufficient to lead to disease.
Sources of Funding
This work was supported by German Federal Ministry for Education and Research (CAN04/006) (NBL3 program, reference 01-ZZ0403); Deutsche Forschungsgesellschaft (Graduiertenkolleg GRK-840), Landesförderungsprogramm EFRE; Fellow-Program “Life Sciences” of the Alfried Krupp Wissenschaftskollegs Greifswald and by the Alfried Krupp von Bohlen und Halbach-Stiftung; by Heart and Stroke Foundation of Ontario (grant #T-5207) and by the Department of Cardiovascular Medicine of the Ernst-Moritz-Arndt-University-Greifswald. Deutsche Forschungsgemeinschaft funded the work on AFM imaging of nanosized clusters and the development of image analysis techniques used in this study under TR24.
Original received February 27, 2006; final version accepted June 27, 2006.
Greinacher A, Michels I, Schafer M, Kiefel V, Mueller-Eckhardt C. Heparin-associated thrombocytopenia in a patient treated with polysulphated chondroitin sulphate: evidence for immunological crossreactivity between heparin and polysulphated glycosaminoglycan. Br J Haematol. 1992; 81: 252–254.
Rosenthal MA, Rischin D, McArthur G, Ribbons K, Chong B, Fareed J, Toner G, Green MD, Basser RL. Treatment with the novel anti-angiogenic agent PI-88 is associated with immune-mediated thrombocytopenia. Ann Oncol. 2002; 13: 770–776.
Li ZQ, Liu W, Park KS, Sachais BS, Arepally GM, Cines DB, Poncz M. Defining a second epitope for heparin-induced thrombocytopenia/thrombosis antibodies using KKO, a murine HIT-like monoclonal antibody. Blood. 2002; 99: 1230–1236.
Visentin GP, Liu CY, Aster RH. Molecular immunopathogenesis of Heparin-induced thrombocytopenia. In: Warkentin TE, Greinacher A, eds. Heparin-induced Thrombocytopenia. New York: Marcel Dekker; 2004: 179–196.
Ziporen L, Li ZQ, Park KS, Sabnekar P, Liu WY, Arepally G, Shoenfeld Y, Kieber-Emmons T, Cines DB, Poncz M. Defining an antigenic epitope on platelet factor 4 associated with heparin-induced thrombocytopenia. Blood. 1998; 92: 3250–3259.
Chong BH, Ismail F, Cade J, Gallus AS, Gordon S, Chesterman CN. Heparin-induced thrombocytopenia: studies with a new low molecular weight heparinoid, Org 10172. Blood. 1989; 73: 1592–1596.
Martel N, Lee J, Wells PS. Risk for heparin-induced thrombocytopenia with unfractionated and low-molecular-weight heparin thromboprophylaxis: a meta-analysis. Blood. 2005; 106: 2710–2715.
Amiral J, Lormeau JC, Marfaing-Koka A, Vissac AM, Wolf M, Boyer-Neumann C, Tardy B, Herbert JM, Meyer D. Absence of cross-reactivity of SR90107A/ORG31540 pentasaccharide with antibodies to heparin-PF4 complexes developed in heparin-induced thrombocytopenia. Blood Coagul Fibrinolysis. 1997; 8: 114–117.
Savi P, Chong BH, Greinacher A, Gruel Y, Kelton JG, Warkentin TE, Eichler P, Meuleman D, Petitou M, Herault JP, Cariou R, Herbert JM. Effect of fondaparinux on platelet activation in the presence of heparin-dependent antibodies: a blinded comparative multicenter study with unfractionated heparin. Blood. 2005; 105: 139–144.
Warkentin TE, Cook RJ, Marder VJ, Sheppard JA, Moore JC, Eriksson BI, Greinacher A, Kelton JG. Anti-platelet factor 4/heparin antibodies in orthopedic surgery patients receiving antithrombotic prophylaxis with fondaparinux or enoxaparin. Blood. 2005; 106: 3791–3796.
Greinacher A, Amiral J, Dummel V, Vissac A, Kiefel V, Mueller-Eckhardt C. Laboratory diagnosis of heparin-associated thrombocytopenia and comparison of platelet aggregation test, heparin-induced platelet activation test, and platelet factor 4/heparin enzyme-linked immunosorbent assay. Transfusion. 1994; 34: 381–385.
Horsewood P, Hayward CP, Warkentin TE, Kelton JG. Investigation of the mechanisms of monoclonal antibody-induced platelet activation. Blood. 1991; 78: 1019–1026.
Netz RR, Andelman D. Neutral and charged polymers at interfaces. Physics Reports-Review Section of Physics Letters. 2003; 380: 1–95.
Rauova L, Poncz M, McKenzie SE, Reilly MP, Arepally G, Weisel JW, Nagaswami C, Cines DB, Sachais BS. Ultralarge complexes of PF4 and heparin are central to the pathogenesis of heparin-induced thrombocytopenia 2. Blood. 2005; 105: 131–138.
Rauova L, Zhai L, Kowalska MA, Arepally GM, Cines DB, Poncz M. Role of platelet surface PF4 antigenic complexes in heparin-induced thrombocytopenia pathogenesis: diagnostic and therapeutic implications. Blood. 2006; 107: 2346–2353.