Thrombotic Dysfibrinogenemia

Thrombosis Research 99 (2000) 187–193

ORIGINAL ARTICLE

Thrombotic Dysfibrinogenemia: Fibrinogen “Caracas V” Relation between Very Tight Fibrin Network and Defective Clot Degradability Rita Marchi1,2, Shah Soltan Mirshahi2,3, Claudine Soria3, Manouchehr Mirshahi2,3,4, Mishal Zohar5, Jean Philippe Collet2, Norma B. de Bosch6, Carmen Luisa Arocha-Pin˜ango1 and Jeannette Soria2 1 Laboratorio de Fisiopatologı´a, Instituto Venezolano de Investigaciones Cientı´ficas, IVIC, Caracas, Venezuela; 2Laboratoire de Biochimie A et Laboratoire Sainte Marie, Hoˆtel Dieu, Paris, France; 3Laboratoire Difema, Faculte´ de Medecine et Pharmacie, Rouen, France; 4Universite´ Tarbiat Modares, Te´he`ran, Iran; 5 Laboratoire de Cytometrie de Flux, CNRS, Villejuif, France; and 6Banco Municipal de Sangre, Caracas, Venezuela. (Received 24 November 1999 by Editor S. Bellucci; revised/accepted 1 March 2000)

Abstract Fibrinogen Caracas V is a thrombotic dysfibrinogenemia with an A␣ 532 Ser→Cys mutation characterized by a tight fibrin network formed of thin fibers responsible for a less porous clot than a normal one. In the present work, fibrinogen Caracas V is further characterized in order to understand the relationship between the structural defect and thrombophilia. This thrombotic disorder has been attributed to a tight fibrin network responsible for a decreased permeation of flow through the clot, leading to defective thrombus lysis due to a diminished availability of fibrinolytic enzymes to the inner fibrin surface. Correction of clot structure anomaly, by addition of dextran 40 to fibrinogen before clotting, induces an improvement in fibrin degradation that was attributed to an increase in porosity. The pulmonary embolism observed in this family has been related to an hyper rigidity of the clot, an anomaly that is also corrected by dextran. Furthermore, this abnormal fibrinogen binds more Abbreviation: t-PA, plasminogen activator. Corresponding author: Dr. Jeannette Soria, Laboratoire Sainte Marie, Hoˆtel Dieu. 1 Place, Parvis de Notre Dame, 75181, Paris 4, France. Tel: ⫹33 (1) 4234 8234 (ext. 2558); Fax: ⫹33 (1) 4326 2673, E-mail: ⬍[email protected]⬎.

albumin than does normal fibrinogen, a phenomenon attributed to the mutation of serine in A␣532 by cysteine. Therefore, this fibrinogen shows a striking similarity to the fibrinogen Dusart, allowing us to confirm that the ␣C-terminal part of fibrinogen plays an important role in fibrin structure, and to conclude that the anomaly of fibrin network observed in fibrinogen Caracas V is responsible for a deficient thrombus lysis.  2000 Elsevier Science Ltd. All rights reserved. Key Words: Fibrinogen; Dysfibrinogenemia; Clot architecture; Fibrinolysis; Thrombotic disorder

I

n blood, fibrinolysis is regulated by specific molecular interactions between fibrin, plasmin(ogen), plasminogen activator (t-PA), and ␣2-antiplasmin [1,2]. When fibrin is formed, t-PA and plasminogen bind to the clot, resulting in local plasmin generation. The structure of the fibrin network depends on both the clotting potential and the fibrinogen concentration: Enhancement of the plasma clotting potential or increased plasma fibrinogen concentration is thus likely to favor formation of tight, rigid, and space-filling network structures with small pores [3]. The fibrinolytic rates of fibrin degradation depends on the architecture of the fibrin

0049-3848/00 $–see front matter  2000 Elsevier Science Ltd. All rights reserved. PII S0049-3848(00)00235-8

188

R. Marchi et al./Thrombosis Research 99 (2000) 187–193

clots [4,5]. Plasma clots composed of thin fibrin fibers are relatively resistant to plasmin degradation, whereas clots made of thick fibers are degraded faster [6]. Congenital dysfibrinogenemias, in which the clots formed are made up of a tight network, are associated with recurrent thrombosis [5,7]. It seems that a network composed of thin fibers is responsible for a reduced porosity and an increased rigidity, favoring thrombotic events. In some patients suffering from thrombotic disorders, defective thrombolysis related to abnormal clot structure has been reported [8,9]. The thrombotic events in dysfibrinogenemias might be explained by two different mechanisms: firstly, defective binding of thrombin to fibrin leading to the presence of free thrombin available for further fibrinogen clotting or for platelet aggregation [9–14]; and secondly, defective fibrinolysis due to a decreased activation of plasminogen by t-PA in the fibrin clot [9,10,14–18]. Fibrinogen Caracas V is a thrombotic dysfibrinogenemia with an A␣ 532 Ser→Cys mutation characterized by a tight fibrin network formed of thin fibers responsible for a less porous clot than the normal one [10,19]. In the present work, other biophysical studies are reported showing that the architecture of the clot, i.e. tight network with highly branched thin fibers, is responsible for a decreased rate of fibrin degradation, which might explain the recurrent thrombotic episodes.

1. Materials and Methods 1.1. Measurement of Viscoelastic Properties of Plasma Clots To plasma samples with and without Dextran 40 (40.000 MW-Assistance Publique, Paris, France) (30 mg/mL, final concentration) were added CaCl2 25 mM (final concentration) and thrombin (Diagnostica Stago, Asnie`res, France) (0.8 NIH units/ mL final concentration). Before clotting, 120 ␮L of the mixture was placed in a 1-mm height chamber formed between 14-mm diameter glass coverslips connected to a Plazek torsion pendulum. The dynamic storage modulus (G⬘) and loss modulus (G⬘⬘) were calculated from records of the free oscillations induced by the application of a momentary stan-

dardized impulse to the pendulum at room temperature [20–22]. G⬘ allows the evaluation of clot stiffness, and G⬘⬘ represents energy dissipated by nonelastic, viscous processes.

1.2. Analysis of Fibrin Clot by Confocal Laser 3D Microscopy Briefly, the plasma was clotted by adding 25 mM CaCl2 and 0.1 NIH units/mL of thrombin (final concentrations) in the absence or in the presence of 30 mg/mL Dextran 40 (final concentration) in a special chamber as described by Blomba¨ck et al. [23]. After overnight incubation in a moist atmosphere to allow complete coagulation, the clot was carefully washed using 10 times its volume with 0.15 M NaCl. Then, it was perfused successively with 1% of glutaraldehyde (3 times its volume) and 0.15 M NaCl (10 times its volume). The fibrin network was observed in a confocal scanning laser microscope (ACAS 570, Meridian Instruments, Okemos, MI, USA) with a laser power of 20 W (wavelength excitation: 488 nm, emission: 525 nm).

1.3. Clot Lysis and Clot Permeability Clot lysis was evaluated by determination of fibrin degradation products (D-dimers) released from the clot perfused by fibrinolytic enzymes. The D-dimer assay was performed by enzyme-linked immunosorbant assay (ELISA) using Asserachrom D-dimer (Diagnostica Stago). Clot perfusion was done essentially as described elsewhere [24]. Briefly, thrombin and calcium chloride (1.25 NIH units/mL and 20 mM final concentrations, respectively) were added to 100 ␮L of plasma in the absence or presence of Dextran 40 (from 10 to 30 mg/mL). Before clotting, the mixture was transferred to a preetched tube of 2.5 mm diameter. After 4 hours of incubation in a moist atmosphere to achieve coagulation, the clot was carefully washed by perfusing with 300 ␮L of 0.15 M NaCl containing hirudin at 10 ␮g/mL (Knoll Laboratories, Paris, France) in order to elute thrombin bound to fibrin and to avoid further clotting during perfusion. Then, degradation was carried out by perfusing the clot with 200 ␮L of r-tPA 1.5 ␮g/mL (Boehringer, Ingelheim, Germany) dissolved in normal plasma, diluted 1/5 with 0.15 M NaCl (as a source of plasminogen), and

R. Marchi et al./Thrombosis Research 99 (2000) 187–193

189

Table 1. Viscoelastic properties of plasma clots from caracas V dysfibrinogenemia and normal volunteers—effect of dextran (30 mg/mL final concentration)a Without dextran

With dextran

Parameter

Control

Caracas V

Control

Caracas

Fibrinogen (g/L) G⬘ (Dynes/cm2) G″ (Dynes/cm2)

2.7 412⫾41 45⫾9

3.0 1507⫾206 135⫾19

2.7 290⫾6 41⫾4

3.0 514⫾110 72⫾12

a

Mean of three experiments⫾standard deviation. G⬘⫽storage modulus; G″⫽loss modulus.

thereafter with 0.15 M NaCl. The eluate was collected at different times for D-dimer measurement. The results were expressed as a percentage of fibrin degraded, taking into account plasma fibrinogen concentration.

1.4. Binding of Albumin Plasma fibrinogen was captured to anti-human fibrinogen immunoglobulin, immobilized to a 96well polystyrene plate (primary antibody). The amount of albumin bound to the fibrinogen was then quantified by a secondary peroxidase-labeled anti-human albumin antibody. The experimental conditions were as follows:

1.4.1. Immobilization of the Antibodies to the Polystyrene Multiwell Plates One hundred microliters of anti-human fibrinogen antibody (Dako, AS Troppes, France) dissolved in carbonate/bicarbonate buffer pH 9.6 at a final concentration of 5 ␮g/mL was added to the first 48 wells of a 96-well polystyrene microtiter plate. In order to quantify albumin bound to fibrinogen, polyclonal anti-human albumin antibody (Nordic, Tilburg, Netherlands) was immobilized to the remaining 48 wells of the plate under the same conditions. After 2 hours of incubation at 37⬚C, the plates were carefully washed with 0.15 M NaCl containing 0.06% Tween 20. 1.4.2. Capture of Fibrinogen and Albumin to the Immobilized Antibodies Fibrinogen concentrations in the plasma samples (1.8–4.4 mg/mL) were adjusted to 1 mg/mL with a 4% human albumin solution. The resulting normalized plasma was diluted with PBS containing 0.05%

Tween 20 and 0.5% gelatin (PBS-T-G) in order to obtain fibrinogen concentrations of 0.5 and 0.25 ␮g/mL, respectively. Pilot experiments done using immobilized Ig anti-fibrinogen as capture antibody and peroxidase labeled anti-fibrinogen as Tag antibody had shown that these fibrinogen concentrations were in the linear part of the binding curve to the immobilized anti-human fibrinogen antibody. One hundred microliters of patient and control plasma diluted as described above were added to the wells coated with anti-human fibrinogen antibody. In parallel, a standard curve of albumin was prepared by adding 100 ␮L of albumin solution at concentrations ranging from 50 to 0.75 ng/mL in PBS-T-G to the wells coated with anti-human albumin antibody. After an incubation period of 2 hours at 37⬚C, the plates were washed.

1.4.3. Detection of Albumin One hundred microliters of peroxidase-labeled anti-human albumin was added to each well of the plate. After 2 hours of incubation at 37⬚C, the plates were washed and immunoperoxidase was quantified using o-phenylene diamine and hydrogen peroxide as described by Wolters et al. [25]. The results were then expressed in moles of albumin bound per M of fibrinogen.

2. Results 2.1. Viscoelastic Properties of the Clots The clot obtained from the patient’s plasma was four times more rigid than that of the normal clot. This anomaly of rigidity was corrected when the abnormal plasma was clotted in the presence of Dextran 40 at 30 mg/mL (Table 1). The loss tangent modulus (G⬘⬘/G⬘) of the patient was not signifi-

190

R. Marchi et al./Thrombosis Research 99 (2000) 187–193

cantly different from control, regardless of whether the experiment was done in the presence or in the absence of dextran.

The fibrin network obtained by plasma clotting is presented in Figure 1. It is clearly evidenced that the abnormal fibrin, as compared to control (Figure 1a), forms a very tight network with thin, short, and highly branched fibers (Figure 1b). Furthermore, the pores in the abnormal fibrin are very small compared to those present in the normal fibrin clot. Addition of Dextran 40 (30 mg/mL final concentration) to plasma (normal or abnormal) induces the formation of thicker fibrin fibers and a looser network (Figure 1c,d, respectively).

(nine times less). However, when clotting the abnormal plasma in the presence of dextran (10 mg/ mL), there is a dramatic increase of the permeability of the patient’s clot, rendering the flow rate similar to that found with a control clot formed without dextran (Figure 2). Caracas V fibrin is very resistant to plasmin degradation when perfused with r-tPA and plasminogen: The control clot is totally degraded after 45 minutes of perfusion, whereas only 54% of the Caracas V fibrin is degraded after 2 hours of perfusion. When clotting is done in the presence of Dextran 40 (10 mg/mL), the fibrinolytic rate of the Caracas V clot is greatly improved. Under this conditions, the clot is completely degraded after 45 minutes (same as the control clot formed in the absence of dextran). The results are summarized in Table 2.

2.3. Clot Permeability and Clot Lysis

2.4. Binding of Albumin

The permeation through fibrin Caracas V is greatly reduced in comparison to that of a control clot

The quantity of albumin bound to fibrinogen Caracas V is shown in Table 3. The amount of albumin

2.2. Confocal Laser 3D Microscopy

Fig. 1. Confocal microscopy of fibrin Caracas V compared to a normal fibrin clot. Effect of dextran. (a) Normal fibrin clot. (b) Caracas V fibrin clot. (c) Normal fibrin clot obtained by clotting plasma in the presence of dextran 40 (30 mg/mL final concentration). (d) Fibrin Caracas V obtained by clotting plasma in the presence of Dextran 40 (30 mg/ml final concentration).

191

R. Marchi et al./Thrombosis Research 99 (2000) 187–193

Fig. 2. Permeability of normal and Caracas V fibrin clots. Effect of Dextran 40 (10 mg/mL final concentration).

bound per M of fibrinogen is 0.4 M, whereas no significant amount of albumin is bound to control fibrinogen.

3. Discussion In previous studies, it was found that polymerization of fibrinogen Caracas V is abnormal, leading to a clot characterized by a tight network of thin fibers. The structural anomaly has been localized to the ␣C domain (Ser→Cys substitution at the A␣-532 position) [10,19]. This substitution, probably because of the presence of free cysteine, may be responsible for the albumin binding in fibrinogen Caracas V. Increased binding of albumin to fibrinogen has already been reported for other fibrinogens in which arginine is replaced by cysteine in the ␣C

Table 2. Degradation of clot perfused by plasminogen and r-tPA Percentage of fibrin degraded Time of perfusion (minutes) Without dextran Control Caracas V With dextran Control Caracas V

18

45

60

90

120

34 0

100 0

— 2

— 21

— 54

65 19

100 100

— —

— —

— —

domain, i.e. fibrinogens Dusart and Chapel Hill III, both with an A␣ 554 Arg→Cys substitution [7,26–28]. The amount of albumin bound to fibrinogen Caracas V is in good agreement with that found to fibrinogen Dusart, using an electron microscopy analysis [28]. Furthermore, as our patient is heterozygous, it can be assumed that each molecule of abnormal fibrinogen binds approximately one molecule of albumin. This might lead to a steric hindrance of some epitopes in the ␣C domain of this fibrin(ogen) and, thus, be responsible for the impairment in lateral aggregation of the fibers. It has already been described that this domain contributes to lateral association of fibrin fibers [29,30]. The recurrent thrombotic pulmonary embolisms observed in our patients may be related to the increase in clot rigidity, since rigid clots are easy to break under the circulatory stress [8,28,31,32]. This increased clot rigidity was also seen in other dysfibrinogenemias associated with thromboembolism (Dusart and Chapel Hill III) [7,27]. Decreased permeability of the abnormal clot is easily explained by the anomaly of fibrin network

Table 3. Amount of albumin bound to fibrinogen Albumin (in moles) bound per mole of fibrinogen Fibrinogen Caracas V Control fibrinogen

0.4 0

192

R. Marchi et al./Thrombosis Research 99 (2000) 187–193

since by confocal microscopy, fibrin Caracas V is very tight compared to a normal fibrin clot. These anomalies lead to a poor accessibility of fibrinolytic enzymes to the inner fibrin surface, and as a consequence to a defective thrombus lysis. This is supported by the fact that, by correcting fibrin structure by clotting plasma in the presence of dextran, there is a parallel increase in clot permeation and in fibrin degradability. In conclusion, the thrombotic disorder associated with pulmonary embolism observed in Caracas V dysfibrinogenemia may be explained by the anomaly of clot structure. This anomaly, close to Dusart syndrome, confirms the observation that a poorly permeable clot leads to undegradable thrombi, and that a clot with high rigidity is easily embolized. This observation also confirms the beneficial effect of Dextran 40 by improving clot structure. Therefore, infusion of Dextran 40 in these patients may be of interest in preventing thrombus formation in risk situations. This work was partially supported by Consejo Nacional de Investigaciones Cientı´ficas y Tecnolo´gicas, Venezuela, CONICIT Grants S-2697 and 97000669 to C.L.A.P. and R.M. We wish to thank Dr. Ulf Lundberg for useful comments, and Mrs. L. Rybak for excellent secretarial work.

8.

9.

10.

11.

12.

References 13. 1. Wiman B, Collen D. Molecular mechanism of physiological fibrinolysis. Nature 1978;272: 549–50. 2. Collen D. On the regulation and control of fibrinolysis. Thromb Haemostas 1980;43:77–89. 3. Blomba¨ck B, Okada M. Fibrin gel structure and clotting time. Thromb Res 1982;25:51–70. 4. Gabriel DA, Muga K, Boothroyd EM. The effect of fibrin structure on fibrinolysis. J Biol Chem 1992;267:24259–63. 5. Collet JP, Soria J, Mirshahi M, Hirsch M, Dagonnet FB, Caen J, Soria C. Dusart syndrome: a new concept of the relationship between fibrin clot architecture and fibrin clot degradability: hypofibrinolysis related to an abnormal clot structure. Blood 1993;82:2462–9. 6. Carr ME Jr, Alving BM. Effect of fibrin structure on plasmin mediated dissolution of plasma clots. Blood Coagul Fibrinolysis 1995;6:567–73. 7. Carrel N, Gabriel DA, Blatt PM, Carr ME,

14.

15.

McDonagh J. Hereditary dysfibrinogenemia in a patient with thrombotic disease. Blood 1983; 62:439–47. Fatah K, Hamsten A, Blomba¨ck B, Blomba¨ck M. Fibrin gel network characteristics and coronary heart disease: relations to plasma fibrinogen concentration, acute phase protein, serum lipoproteins and coronary atherosclerosis. Thromb Haemost 1992;68:130–5. Soria J, Soria C, Collet JP, Mirshahi Mc, Lu H, Caen JP. Dysfibrinogenemia and thrombosis. Nouv Rev Fr Hematol 1991;33:457–9. Haverkate F, Samama MM. Familial dysfibrinogenemia and thrombophilia. Report on a study of the SSC Subcommitee on fibrinogen. Thromb Haemostas 1995;73:151–61. Liu CY, Wallen P, Handley DA. Fibrinogen New York I: the structural, functional, and genetic defects and a hypothesis of the role of fibrin in the regulation of coagulation and fibrinolysis. In: Lane DA, Henschen A, Jasani MK, editors. Fibrinogen-Fibrin Formation and Fibrinolysis. Berlin: Walter de Gruyter; 1986. p. 79–90. Liu CY, Koehn JA, Morgan FJ. Characterization of fibrinogen New York I. A dysfunctional fibrinogen with a deletion of Bbeta (9–72) corresponding to exon II of the gene. J Biol Chem 1985;260:4390–6. Koopman J, Haverkate F, Lord ST, Grimbergen J, Mannucci PM. Molecular basis of fibrinogen Naples associated with defective thrombin binding and thrombophilia. Homozygous substitution of B␤ 68 Ala→Thr. J Clin Invest 1992;90:238–44. Fernandez J, Paramo JA, Cuesta B, Aranda A, Rocha E. Fibrinogen Pamplona II. A new congenital dysfibrinogenemia with abnormal fibrin enhanced plasminogen activation and defective binding of thrombin to fibrin. In: Mu¨ller-Berghaus G, Scheefers-Borchel U, Seylmar E, Henschen A, editors. Fibrinogen and its Derivatives. Biochemistry, Physiology and Pathophysiology. Amsterdam: Excerpta Medica, Elsevier Science Publishers, 1986. p. 25–30. Lijnen HR, Soria J, Soria C, Collen D, Caen JP. Dysfibrinogenemia (Fibrinogen Dusard) associated with impaired fibrin-enhanced plasminogen activation. Thromb Haemostas 1984; 51:108–9.

R. Marchi et al./Thrombosis Research 99 (2000) 187–193

16. Koopman J, Haverkate F, Grimbergen J, Lord ST, Mosesson M, Diorio JP, Siebenlist KS, Legrand C, Soria J, Soria C, Caen JP. Molecular basis for fibrinogen Dusart (A␣ 554Arg→Cys) and its association with abnormal fibrin polymerization and thrombophilia. J Clin Invest 1993;91:1637–43. 17. Engesser L, Koopman J, De Munk G, Haverkate F, Nova´kova´ Y, Verheijen JH, Brie¨t E, Brommer EJP. Fibrinogen Nijmegen: congenital dysfibrinogenemia associated with impaired t-PA mediated plasminogen activation and decreased binding of t-PA. Thromb Haemostas 1988;60:113–20. 18. Ieko M, Sawada KI, Sakamura S, Yamagishi Y, Isogawa S, Nakagawa S, Satoh M, Yasukuochi T, Matsuda M. Fibrinogen Date: Congenital hypodysfibrinogenemia associated with decreased binding of tissue-plasminogen activator. Am J Hematol 1991;37:228–33. 19. Marchi R, Arocha-Pin˜ango CL, Escobar W, Torres A, Haverkate F. Fibrinogen Caracas V with an A␣ 532 Ser→Cys substitution associated with a thrombotic tendency. 14th International Congress on Thrombosis of Mediterranean League. Montpellier, France; 1996. 20. Plazek DJ, Vrancken MN, Berge JW. A torsion pendulum for dynamic and creep measurements of soft viscoelastic materials.Trans Soc Rheol 1958;2:39–51. 21. Gerth C, Roberts WW, Ferry JD. Rheology of fibrin clots. II Linear viscoelastic behavior in shear creep. Biophys Chem 1974;2:208–17. 22. Janmey A. A torsion pendulum for measurement of the viscoelasticity of biopolymers and its application to actin networks. J Biochem Biophys Methods 1991;22:41–53. 23. Blomba¨ck B, Carlsson K, Hessel B, Liljeborg A, Procyk R, Aslund N. Native fibrin gel networks observed by 3D microscopy, permeation and turbidity. Biochem Biophys Acta 1989;997: 96–110. 24. Blomba¨ck B, Carlsson K, Fatah K, Hessel B, Procyk R. Fibrin in human plasma: Gel archi-

25.

26.

27.

28.

29.

30.

31.

32.

193

tectures governed by rate and nature of fibrinogen activation. Thromb Res 1994;75:521–38. Wolters G, Kuijpers L, Kacabi J, Schuurs A. Solid phase enzyme immunoassay for detection of hepatitis B surface antigen. J Clin Path 1976;29:873–9. Wada Y, Lord ST. A correlation between thrombotic disease and a specific fibrinogen abnormality (A␣ 554 Arg→Cys) in two unrelated kindred Dusart and Chapel Hill III. Blood 1994;84:3709–14. Siebenlist KR, Mosesson MW, DiOrio JP, Soria J, Soria C, Caen JP. The polymerization of fibrinogen Dusard (A␣ 554 Arg→Cys) after removal of carboxy terminal regions of the A␣ chains. Blood Coagul Fibrinolysis 1993;4:61–5. Collet JP, Woodhead JL, Soria J, Soria C, Mirshahi Mc, Caen JP, Weisel JW. Fibrinogen Dusart: Electron microscopy of molecules, fibers and clots, and viscoelastic properties. Biophys J 1996;70:500–10. Gorkun OV, Veklich YI, Medved LV, Henschen AH, Weisel JW. Role of the ␣C domains of fibrin in clot formation. Biochemistry 1994; 33:6986–97. Veklich YI, Gorkun OV, Medved LV, Nieuwenhuizen W, Weisel JW. Carboxi-terminal portions of the ␣ chains of fibrinogen and fibrin. Localization by electron microscopy and the effects of isolated ␣C fragments on polymerization. J Biol Chem 1993;268:13577–85. Scrutton MC, Ross-Murphy SB, Bennett GM, Stirling V, Meade TW. Changes in clot deformability: A possible explanation for the epidemiological association between plasma fibrinogen concentration and myocardial infarction. Blood Coagul Fibrinolysis 1994;5:719–23. Blomba¨ck B, Banerjee D, Carlson K, Hamsted A, Hessel B, Zarchaski L. Native fibrin gel networks and factors influencing their formation in health disease. In: Liu CY, Chien S, editors. Fibrinogen, Thrombosis, Coagulation & Fibrinolysis. New York: Plenum, 1992; p. 1–23.