Heparin-like anticoagulant associated with systemic candidiasis

American Journal of Hematology 3 5 3 7 4 2 (1990) Heparin-Like Anticoagulant Associated With Systemic Candidiasis McDonald K. Horne 111, Elizabeth S. Chao, and Olga J. Wilson Hematology Section, Clinical Pathology Department, Warren G. Magnuson Clinical Center, National institutes of Health, Bethesda, Maryland A 15 year old girl with apladic anemia developed a heparin-like anticoagulant during the course of systemic candldiasis. This was inltlally detected In the laboratory by an elevation of the thrombin clotting time which corrected with toluidine blue but not by mixing with normal plasma. In vivo and in vitro the anticoagulant was remarkably resistant to neutralization by protamine sulfate. Nevertheless, its heparin-like nature was confirmed by its sensitivity to heparinaae and its dependence on antithrombin 111. zyxwvutsrq zyxwvuts zyxwvutsr zyxwvutsrq Key words: anticoagulants, glycosaminoglycans, heparin, candldlasis INTRODUCTION been overlooked as the patient’s coagulopathy became progressively complex in the terminal phase of her illThe appearance of endogenous heparin-like activity in ness. the circulation is reported to be extremely rare. Although The case we are presenting was also remarkable in that there is evidence that a heparinoid may normally be the anticoagulant was refractory to the usual heparin anpresent in the plasma of newborns [ 11, in older individ- tidote, protamine sulfate, both in vivo and in vitro. Aluals such anticoagulants have only been described in though protamine has been used with variable success to pathological states. Usually they have been associated treat other heparin-like anticoagulants [5,10,12], it was with a neoplastic disease, either multiple myeloma [2- completely ineffective in our patient even in a very large 51, acute monoblastic leukemia [ 6 ] , systemic mastocy- dose. tosis [7], or carcinoma [8-lo]. We have documented their appearance in patients with malignancies treated CASE REPORT with suramin [ 111. In this report we describe a heparin-like anticoagulant The patient was a 15 year old Caucasian female who which arose in a different clinical setting-during the was transferred to the National Institutes of Health for course of systemic candidiasis in an adolescent girl with treatment of aplastic anemia that had developed approxsevere aplastic anemia. Only one other case has been imately 6 weeks earlier. On admission she was febrile reported during an infectious complication of immuno- and was found to have pneumonia. Her hemoglobin was suppression [ 121; this patient had toxoplasmosis, sus- 9.0 g/dL, white count <lOo/pL, and platelet count pected drug-induced marrow hypoplasia, and the ac- 5,OOO/pL. Blood cultures grew Candidu albicans. Amquired immunodeficiency syndrome. The true incidence photericin, which had been started prior to admission, of heparin-like anticoagulants in such settings is unclear, was continued. Because of persistent fever other antibisince they could be easily obscured by the effects of otics were subsequently added (gentamicin, vancomydisseminated intravascularcoagulation and hepatic insufficiency, which often complicate overwhelming infections. The clue to the anticoagulant in the case we are for publication December 1 1 , 1989; accepted March 15, presenting was a prolongation of the thrombin clotting Received 1990. time which normalized with toluidine blue, a cationic dye that binds to highly anionic substances including Address reprint requests to Dr.Horne, Rm 2C390, Building 10, NIH heparin. Without this test, the anticoagulant might have 9OOO Rockville Pike, Bethesda, MD 20892 zy Published 1990 by Wiley-Liss, Inc. 38 zyxwvutsr zyxwvut zyxw Case Report: Horne et al. TABLE 1. Coagulation Parameters, Hospital Day 27 zyxwvutsrqpo zyxwvutsrq Measurement ~ ~ _ _ _ _ Result (normal) ~ Prothrombin time (sec) Activated partial thromboplastin time (sec) Thrombin time (sec) Thrombin time. I:l mix with normal plasma (sec) Thrombin time with toluidine blue (sec) Thrombin time 15 min after 100 mg prutamine sulfate, i.v. (sec) Reptilase time (sec) Fibrin D-dimer ( ~ g / m L ) Fibrinogen, method of von Clauss (mg/dL) Fibrinogen, radial immunodiffusion (mg/dL) Factor VII (%) Factor VIII (8) Factor IX (%) Factor X (Yo) Antithrombin 111 (mg/dL) cin, metronidazole, acyclovir). The patient also had recurrent gastrointestinal and vaginal hemorrhage for which she received multiple transfusions of packed red cells and platelets. With the support of transfusions her platelet count fluctuated between 10,000 and 70,000/ kL. In addition she was given anti-thymocyte globulin on hospital days 4 through 8. Routine coagulation tests during the first 3 weeks of the hospitalization were normal. On the 20th hospital day, however, the patient’s thrombin time rose to 41 seconds (normal, 28-35). By day 27 the prothrombin time (PT) and activated partial thromboplastin time ( A m ) had also become abnormal and additional coagulation studies were performed (Table I). The onset of the coagulopathy coincided with new abnormalities in liver function. There was no apparent response to multiple doses of intravenous vitamin K. On the basis of the data in Table I, it appeared that the coagulopathy was partially due to a heparin-like anticoagulant. For this reason a trial of protamine sulfate was given, first 10 mg and then 100 mg (approximately 3 mg/kg) intravenously over 10 minutes. Coagulation tests after the protamine showed no improvement (Table I), nor was there any obvious reduction in the patient’s hemorrhage. Although her coagulopathy remained stable, the patient expired 3 days later with a respiratory arrest. Autopsy revealed extensive hemorrhage in the lungs, gastrointestinal tract, and endometrium. Postmortem cultures of the liver grew C. ulbicuns. 16.3 (9.4-13.7) 40 (23-35) 84 (28-35) 48 (C35) 25 (<30) 73 23 (23-26) 8-16 (<0.2) 450 (160-350) 670 (10-30% greater than von Clauss) 25 (50-150) 200 (50-150) 80 (50-150) 100 (50-150) 21 (24-36) suring fibrin D-dimer were supplied by Diagnostic Stago (Asnieres-Sur-Seine, France). Bovine thrombin was purchased from Parke-Davis (Moms Plains, NJ), reptilase from Pentapharm, Ltd. (Basel, Switzerland), toluidine blue from Roboz Surgical Instruments (Washington, DC),and radial immunodiffusion plates for fibrinogen and for antithrombin LII (ATIII) from Behring Diagnostics, Inc. (Somerville, NJ). Chromogenic substrates for factor Xa (S2222) and for thrombin (S2238) as well as a heparin assay kit (Coatest), which included ATIII, were from KabiVitrum (Stockholm, Sweden). Protamine sulfate was the product of LyphoMed, Inc. (Rosemont, IL). Beef lung heparin was from Upjohn Co. (Kalamazoo, MI), and heparan sulfate, dermatan sulfate, and chondroitin 6-sulfate were from Sigma Chemical Co. (St. Louis, MO). DEAE-Sephacel was manufactured by Pharmacia Fine Chemicals (Uppsala, Sweden), and Spectra-Por dialysis membrane by Spectrum Medical Industries (Los Angeles, CA). Papain was obtained from Kodak (Rochester, NY), and Fluvobacterium heparinum heparitinase (EC 4.2.2.8) and heparinase (EC 4.2.2.7) and Proteus vulguris chondroitin ABC lyase (EC 4.2.2.4) were from Seikagaku Kogyo Co. (Tokyo, Japan). zyxwvutsr MATERIALS AND METHODS Materials Reagents for routine coagulation studies were obtained from Organon Teknika Corporation (Durham, NC), Ortho Diagnostic Systems (Raritan, NJ), and George King Bio-Medical, Inc. (Overland Park, KS). Kits for mea- Coagulation Tests PTs, AFTTs, and fibrinogen concentrations (von Clauss method [ 131) were measured with a Coag-A-Mate X2 (General Diagnostics, Moms Plains, NJ). Thrombin clotting times were determined with a Fibrometer (BBL FibroSystem, Becton-Dickinson, Orangeburg, NJ) by mixing 0.1 mL Tris-buffered saline (0.1 M NaCl, 0.05 M tris, pH 7.4), 0.1 mL test plasma, and 0.1 mL bovine thrombin to give a final thrombin concentration of 1 NIH unit/mL. The activity of the patient’s isolated anticoagulant was tested with the thrombin time assay by including the sample in the buffered saline component of - Case Report: Heparin-Like Anticoagulant the mixture and using pooled normal plasma as the test plasma. Similarly the effect of protamine sulfate on the patient’s anticoagulant and on authentic heparin was studied by adding the protamine and the anticoagulants to the Tris-buffered saline fraction of the thrombin time mixture. Reptilase times and coagulation factor assays were performed by published methods [14]. The antifactor Xa and antithrombin activities of the patient’s anticoagulant were quantitated using chromogenic substrates and purified antithrombin 111, according to directions supplied by KabiVitrum. The antithrombin activity was also measured with the thrombin clotting time. In both assay systems the effect of the patient anticoagulant was compared with that of authentic heparin. 39 zy .-E I- zyxwvutsrq zyxw zyxwvuts zy zyxwvutsr Purificationof the Patient Anticoagulant Nine mL of fresh-frozen patient plasma (obtained on day 28 of hospitalization and anticoagulated with 11 mM sodium citrate) was applied at 0.25 mL/min to a 10 mL column of DEAE-Sephacel which had been equilibrated with 0.3 M NaCl, 0.01 M Tris, pH 8 [ 111. The A280 of the effluent was monitored and the flow-through material was saved. When the A280 approached zero, the column was eluted with 2 M NaCl, 0.01 M Tris, pH 8; two column volumes were collected. Both the flowthrough and eluate were concentrated by ultrafiltration (nominal molecular weight cut-off, 3,000) to the original sample volume, dialyzed against Tris-buffered saline (nominal molecular weight cut-off, 2,000), and tested for anticoagulant activity with the thrombin time. Only the material eluting with 2 M NaCl had activity. This was digested with 2 mg/mL papain in Tris-buffered saline including 5 mM Na2EDTA, 5 mM cysteine, pH 6, at 56°C for 48 hours. After the pH had been adjusted to -8, the digest was reapplied to a 10 mL column of DEAE-Sephacel in 0.3 M NaCl, 0.01 M Tris, pH 8. The column was washed with this buffer until the A280 reached zero; then the bound material was eluted with 2 M NaCl, 0.01 M Tris, pH 8. The eluate was dialyzed extensively against distilled water, then lyophilized and redissolved in 0.8 mL water for electrophoresis, enzyme digestions, and studies of anticoagulant activity. Similar procedures were followed using 50 mL samples of plasma from two normal donors and a 50 mL column of DEAE-Sephacel. Incubation of Patient Anticoagulant With Glycosaminoglycan-SpeclflcLyases A portion of the isolated and lyophilized patient anticoagulant was dissolved in Tris-buffered saline in a concentration sufficient to give reproducible prolongations in the thrombin clotting time. Aliquots of the anticoagulant in this concentration were then incubated at 37°C for 2 hours with the following enzymes: 1 unit/mL F. Protamine Sulfate (pg/ml) Fig. 1. Effect of protamine sulfate (pglmL) on the thrombin clotting time (seconds) of patient plasma (M) and of nor). mal plasma containing 0.25 units/mL heparin (uThe shaded area indicates the normal range. heparinum heparitinase; 8 units/mL F. heparinum heparinase; or 0.5 units/mL P. vufgaris chondroitin ABC lyase (units defined by the manufacturer). Enzyme activities were tested by measuring their effect on the anticoagulant activity of appropriate substrates (authentic heparan sulfate, heparin, or dermatan sulfate) using the thrombin time. Electrophoresis One to 15 (I.L samples of the isolated patient anticoagulant and the material similarly purified from normal plasma were applied to 9.4 X 7.6 cm supported cellulose acetate membranes (Optiphor-10; Gelman Sciences, Inc., Ann Arbor, MI) and electrophoresed at 9 mAmp for 1 hour at 20°C in 0.2 M ZnS04 [15]. Standards of chondroitin 6-sulfate and heparan sulfate were included with each run. Following electrophoresis the membranes were stained for 10 minutes with 0.1% alcian blue, 0.03 M MgC12,O. 1% acetic acid, 10% ethanol, and destained with the same solution lacking the alcian blue. RESULTS Results of the clinical coagulation tests on day 27 are summarized in Table I. The effect of protamine sulfate on the thrombin time of the patient’s plasma is shown in Figure 1, where it is compared with the effect of protamine sulfate on the anticoagulant activity of heparinized 40 Case Report: Horne et al. zyxwvut zyxwvut zyxw zyxwvu zyxwvuts TABLE II. Effect of Enzymes on Anticoagulant Purified From Patient Plasma and on Authentlc Glycosaminoglycans* Anticoagulant Patient Dermatan sulfate, 20 p@mL Heparan Sulfate, 20 pg/mL Heparin, 0.1 U/mL Enzyme Thrombin time (28-35 sec) None Chondroitin ABC lyase, 0.5 U/mL Heparitinase, 1 U/mL Heparinase, 8 U/mL None Chondroitin ABC lyase, 0.5 UlmL None Heparitinase, 0.5 U/mL None Heparinase, 8 U/mL 77; 97; 93 84; 88 65; 65 32; 32 76; 1 3 32; 32 >120; >I20 40; 39 >120; >120 33; 33 *The anticoagulants were incubated alone or with the indicated enzymes for 2 hours at 37°C in Tris-buffered saline. pH 7.4. Then 0.1 mL of the incubation mixtures was mixed with 0. I mL normal plasma before 0.1 mL thrombin was added and the clotting time measured. Enzymes units are those defined by the manufacturer (Seikagaku). normal plasma. A 15- to 20-fold higher concentration of protamine was necessary to normalize the prolonged thrombin time of the patient’s plasma than was needed to counteract 0.25 units/mL heparin. Based upon this study, to achieve the plasma protamine concentration necessary to normalize the patient’s prolonged thrombin time in vivo would have required a bolus injection of 250-350 mg of the drug. The anticoagulant activity of the patient’s plasma was bound to DEAE in 0.3 M NaCl and was recovered in the material eluted with 2 M NaCl. After incubation with papain the anticoagulant was again retained by DEAE in 0.3 M NaCl, whereas the majority of the material with A280 passed through. The anticoagulant activity was once again recovered with 2 M NaCl. The effect of the isolated anticoagulant on the thrombin time is shown in Table 11. Assuming no loss during the purification process, the anticoagulant material used in the studies shown in Table I1 was in a concentration approximately one-third of that found in the patient plasma from which it was extracted. Yet at this reduced concentration the anticoagulant produced thrombin times in normal plasma equivalent to those observed with the original patient plasma. In other words, the anticoagulant appeared to be approximately three times more potent in normal plasma than it was in the patient’s plasma. The anticoagulant activity was not affected by incubation with chondroitin ABC lyase or heparitinase, although these enzymes had marked effects upon the anticoagulant activity of their respective substrates, dermatan sulfate and heparan sulfate (Table 11). In contrast, the patient anticoagulant activity was eliminated by digestion with heparinase. Fractions extracted by the same techniques from two normal plasma samples did not prolong the thrombin time, even in a concentration calculated to be six-fold that found in the original plasma. The purified patient anticoagulant was shown to have both antithrombin and anti-factor Xa activity of approximately equal potency when tested in the presence of ATIII (Table 111). However, when anti-Xa activity was assayed in the absence of antithrombin 111, none was found. The results of electrophoresis of the patient anticoagulant material and the material identically isolated from normal plasma are shown in Figure 2. Both samples contained components staining with alcian blue, which reacts with glycosaminoglycans. The alcian blue-positive material in the patient sample appeared divided between a major component which co-migrated with heparan sulfate and a minor component which migrated close to chondroitin 6-sulfate. The normal material, in contrast, appeared more heterogeneous, although two dominant components could be discerned, one co-migrating with heparan sulfate and the other migrating between heparan sulfate and chondroitin 6-sulfate7 approximately where dermatan sulfate migrates in this system. Attempts to quantitate the glycosaminoglycan content of the samples by measuring uronic acid [I61 were confounded by the presence of other substances, presumably strongly anionic glycopeptides, which interfere with the usual colorimetric assays [17]. DISCUSSION We have described the case of an adolescent girl with aplastic anemia who developed a heparin-like anticoagulant in the setting of progressive hepatic candidiasis. Although hemorrhage was a major complication of her illness, this preceded the appearance of the anticoagulant zy zyx Case Report: Heparin-Like Anticoagulant TABLE 111. Purlfied Patient Anticoagulant Activity Against Factor Xa and Thrombin ~~ 41 [20]. Two additional observations nevertheless con- firmed that the anticoagulant was heparin-like: it was zyxwvutsrqponmlkjihgfedcbaZYXWVUTSRQPONMLKJIHGFEDCBA Heparin units destroyed by heparinase (Table II) and its anticoagulant activity depended upon antithrombin I11 (Table 111). The latter characteristic perhaps explains why the purified anticoagulant appeared to be more potent in normal plasma than in the patient’s own plasma, which was deficient in antithrombin 111 (Table I). After electrophoresis in ZnS04, staining with alcian blue confirmed that the isolated anticoagulant material contained glycosaminoglycans, but the migration pattern of these components differed somewhat from that of glycosaminoglycans purified from normal plasma (Fig. 2). Despite these characteristics, the refractoriness of the anticoagulant to protamine sulfate suggested that it was not actually heparin but rather a similar glycosaminoglycan, heparan sulfate [21]. The antagonism of heparin by protamine depends upon charge interactions between the cationic protein and the sulfate groups of the glycosaminoglycan [20,22]. The molecular spacing of these sulfates resembles the spacing of the phosphates in DNA, to which protamine is bound in its natural state in fish sperm nuclei [23]. Polyanions with charge distributions which closely complement that of protamine bind readily to the protein, whereas polyanions like heparan sulfate, which are less sulfated than heparin, bind with lower affinity (221. Therefore the relative resistance of our patient’s anticoagulant to neutralization by protamine indicated that its pattern of sulfation was different from that of heparin. Nevertheless, this non-heparin anticoagulant activity was degraded by heparinase but not by heparitinase (heparan sulfate lyase). This apparent inconsistency is explained by the fact that the antithrombin activity of both heparin and heparan sulfate depends upon an octasaccharide which always includes a bond (linking Nsulfated, 6-O-sulfated glucosamine and 2-O-sulfated iduronate) that is cleaved by heparinase [21,24]. Heparitinase, on the other hand, cleaves glucosamine-glucuronate bonds that are not present in this region but apparently may exist in flanking oligosaccharides which, although not yet well defined, are also known to be necessary for antithrombin activity. The effect of heparitinase on these flanking oligosaccharides was evident in the elimination of the antithrombin activity of the authentic heparan sulfate used in our studies (Table 11). The heparan sulfate anticoagulants we reported in suramintreated patients were also sensitive to heparitinase [ 1 13. Failure of this enzyme to affect the anticoagulant we are now describing indicates that in this glycosaminoglycan heparitinase-sensitivebonds did not occur within the oligosaccharide domains subserving antithrombin activity. It does not mean, however, that bonds in other regions of the molecule were not cleaved. Such cleavage would zyxwvutsrqpo Anti-thrombin activity By thrombin clotting time With S2238 + AT111 Anti-factor Xa activity With S2222, + AT111 - ATIII 0.8 u n i t s l d 0.6 units/mL 0.7 uniWmL 0 -C6S C6S - -HS HS - -Origin Origin - Patient Normal Fig. 2. Migration pattern of patient anticoagulant material and material identically purified from normal plasma, after electrophoresis in 0.2 M ZnSO, and staining with alcian blue. The migration distances of authentlc heparan sulfate (HS) and chondroltin &sulfate (C6S) are indicated. and was largely due to thrombocytopenia. During the terminal phase of her hospitalization, however, the bleeding may have been accelerated by the anticoagulant and a degree of disseminated intravascular coagulation. The latter is suggested by the relatively high concentrations of fibrin degradation products in the patient’s plasma and the low antithrombin 111, although interpretation of these abnormalities is difficult in the face of hepatic disease [ 181. As the coagulopathy progressed the prothrombin time rose as a reflection of factor VII deficiency, presumably due to liver failure. Other procoagulant factors, however, remained normal (Table I). The most unusual component of this coagulopathy was the anticoagulant that arose in the midst of this very complicated clinical setting. The heparin-like nature of this substance was initially suggested by the prolongation of the thrombin clotting time, which normalized after the addition of toluidine blue (a cationic dye that binds to heparin) [19] but not by mixing with normal plasma (Table I). Paradoxically, however, both in vivo and in vitro the anticoagulant was remarkably insensitive to protamine sulfate, a potent heparin antagonist (Fig. 1) zy zy 42 zyxwvutsrqpo zyxwvutsrqp zyxwvu zyxwvutsrqp Case Report: Horne et al. have been undetected by our assays, which only measured anticoagulant activity (Table 11). Unfortunately insufficient material was available to study the effect of heparitinase on the molecular size of the anticoagulant. The origin of this anticoagulant can only be conjectured. Cundidu species have never been described to synthesize heparan sulfate (personal communication: Dr. Errol Reiss, Centers for Disease Control, Atlanta). On the other hand, heparan sulfate is ubiquitous on plasma membranes and in extracellular matrices of vertebrates [2 I] and is also detectable in normal human plasma [ 171. Our studies suggest, however, that the glycosaminoglycans in normal plasma, even when concentrated six-fold, do not have anticoagulant activity. It seems likely, therefore, that the heparan sulfate observed in our patient was either shed from damaged endothelial surfaces or released into her circulation from extravascular sources, most likely her injured hepatic or pulmonary tissues. REFERENCES 1. Miller AD, van Doorm JM, Hemker HC: Heparin-like inhibitor of blood coagulation in normal newborn. Nature 267:616, 1977. 2. Khoory MS, Neshein ME, Bowie EIW, Mann KG: Circulating heparan sulfate proteoglycan anticoagulant from a patient with a plasma cell disorder. J Clin Invest 65:666, 1980. 3. Palmer RN, Rick ME, Rick PD, Zeller JA, Gralnick HR: Circulating heparan sulfate anticoagulant in a patient with a fatal bleeding disorder. N Engl J Med 310:1696, 1984. 4. Chapman GS, George CB, Danley DL: Heparin-like anticoagulant associated with plasma cell myeloma. Am J Clin Pathol83:764, 1985. 5 . Kaufman PA, Gockerman JP. Greenberg CS: Production of a novel anticoagulant by neoplastic plasma cells: Report of a case and review of the literature. Am J Med 86612, 1989. 6. Bussel JB, Steinherz PG, Miller DR, Hilgartner MW: A heparin-like anticoagulant in an 8-month-old boy with acute monoblastic leukemia. Am J Hematol 16:83, 1984. 7. Nenci GG,Berrettini M, Parise P, Agnelli G: Persistent spontaneous heparinaemia in systemic mastocytosis. Folia Haematol 109:453, 1982. 8. Watts E, Chisholm M, Hickton M, Smith AG: Circulating heparinlike anticoagulants (letter). N Engl J Med 311:1055, 1984. 9. Rodgers GM, Corash L: Acquired heparinlike anticoagulant in a patient with metastatic breast carcinoma. West J Med 143:672, 1985. LO. Tefferi A, Owen BA, Nichols WL, Witzig TE, Owen WG: Isolation of a heparin-like anticoagulant from the plasma of a patient with metastatic bladder carcinoma. Blood 74:252, 1989. 11. Horne MK. Stein CA, LaRocca RV, Myers CE: Circulating glycosaminoglycan anticoagulants associated with suramin treatment. Blood 71:273, 1988. 12. De Prost D, Katlama C, plaloux G, Karsenty-Mathonnet F, Wolff M: Heparin-like anticoagulant associated with AIDS. Thromb Haemost 57:239, 1987. 3. Clauss A: Gerinnungs physiologische Schnell methode zur Bestimming dis Fibrinogens. Acta Haematol (Basel) 17:237, 1957. 4. Gralnick HR. Givelber HM, Shainoff JR, Finlayson JS: Fibrinogen Bethesda: A congenital dysfibrinogenemia with delayed tibrinopep tide release. J Clin Invest 50:1819, 1971. 5. Haruki F, Kirk JE: A method for separation of chondroitin sulfate B from isomers A and C by paper electrophoresis with zinc sulphate or zinc acetate solution as electrocyte medium. Biochim Biophys Acta 136:391, 1967. 16. Blumenkrantz N, Asboe-Hansen G: New method for quantitative determination of uronic acids. Anal Biochem 54:484, 1973. 17. Staprans I, Felts JM: Isolation and characterization of glycosaminoglycans in human plasma. J Clin Invest 76:1984, 1985. 18. Wilde JT, Kitchen S, Kinsey S, Greaves M, Preston FE: Plasma D-dimer levels and their relationship to serum fibrinogen/fibrin degradation products in hypercoagulable states. Br J Haematol 71:65, 1989. 19. Peden JC. McFarland JA: Use of the plasma thrombin time to assess the adequacy of in vivo neutralization of heparin: Comparative studies following operations employing extracoporeal circulation. Blood 14: 1230, 1959. 20. Jaques LB: Protamine-antagonist to heparin. Can Med Assoc J 108: 1291-1 297, 1973. 21. Gallagher JT, Lyon M, Steward WP: Structure and function of heparan sulfate proteoglycans. Biochem J 236:313-325, 1986. 22. Hubbard AR, Jennings CA: Neutralisation of heparan sulfate and low molecular weight heparin by protamine. Thromb Haemost 53:86-89, 1985. 23. Felix K: Protamines. In Anfinsen CB, Anson ML, Bailey K, Edsall JT (4s):“Advances in Protein Chemistry.” New York: Academic Press, 1960, p I . 24. Rosenberg RD: The heparin-antithrombin system: A natural anticoagulant mechanism. In Colman RW, Hirsh J, Marder VJ, Salzman EW (eds): “Hemostasis and Thrombosis,” 2nd Ed. Philadelphia: JB Lippincott Co, 1987, p 1373. zyxwvu zyxwvutsrqp zyxwvutsrq