Platelet Autoantibodies in Immune Thrombocytopenic Purpura

Pergamon

Transfus. Sci. Vol. 19, No. 3, pp. 237±244, 1998 # 1998 Published by Elsevier Science Ltd. All rights reserved Printed in Great Britain PII: S0955-3886(98)00037-X 0955-3886/98 $Ðsee front matter

Platelet Autoantibodies in Immune Thrombocytopenic Purpura Diana S. Beardsley, M.D., Ph.D.*y Mehmet Ertem, M.D.*

INTRODUCTION

Immune thrombocytopenic purpura has many similarities to autoimmune hemolytic anemia (AIHA) as a hematologic autoimmune disease (Table 1). However, the search for a laboratory assay for antiplatelet autoantibodies analogous to the Coombs' test for erythrocyte autoantibodies has been a long and dif®cult undertaking. This report will discuss a number of key experimental approaches to this problem over the past several decades and will summarize the current understanding of the pathogenic mechanisms of ITP. The target antigens of antiplatelet autoantibodies will be discussed, and the currently available laboratory tests for ITP autoantibodies will be summarized.

Idiopathic thrombocytopenic purpura (ITP) has been recognized as a serious acquired hemorrhagic tendency for more than 200 years. First known as ``Werlof's Disease,'' this condition is characterized by the sudden, transient development of bruising and bleeding from mucous membranes. In the early Twentieth Century, Werlof's Disease was shown to be associated with a marked decrease in the number of circulating blood platelets. In the 1950s, Harrington et al.1 demonstrated that plasma from patients with ITP destroyed platelets within hours after administration to healthy volunteers, including Dr Harrington himself! It had previously been recognized that a pregnant woman with ITP could give birth to an infant with transient thrombocytopenia, suggesting that the agent of platelet destruction was a soluble plasma component capable of traversing the placenta to the fetal circulation. Thus, it was established that the platelet destruction in ITP is mediated by some component of the globulin fraction of plasma, implicating IgG antiplatelet autoantibodies as the culprit etiologic agents of ITP.

PLATELET ASSOCIATED IMMUNOGLOBULIN AND ANTIPLATELET ANTIBODIES IN ITP A number of methods have been employed to detect and characterize the antiplatelet autoantibodies that cause platelet destruction in ITP. Agglutination tests analogous to the Coombs' test were not satisfactory. This approach employs an antiglobulin reagent to determine whether the target cells are coated with IgG However, normal platelets carry surface IgG (approximately 500-5000 molecules per cell), unlike red cells (<50 molecules per cell), so `false positives' can occur. Furthermore, platelets aggregrate as a part of

*Departments of Pediatrics and Internal Medicine, Yale University School of Medicine, New Haven, CT 06510, U.S.A. y Author for correspondence at 333 Cedar Street, New Haven, CT 06510, U.S.A.

237

238 Transfus. Sci. Vol. 19, No. 3

Table 1.

Comparison of Immune Thrombocytopenic Purpura (ITP) and Autoimmune Hemolytic Anemia (ATHA)

Target cell Predominant antigen Bone marrow production Measure of incr. production Measure of autoantibody

ITP

ATHA

Platelet GPIIb/IIIa incr. Megakaryocytes Reticulated Platelets incr. PAIgG (?) Antigen capture test

Erythrocyte Rhesus antigens incr. Erythroid Precursors Reticulocyte Count

their normal function. This is dif®cult to distinguish from agglutination, the endpoint for the Coombs' test, further increasing the rate of false positive results. Platelet dependent processes such as aggregation, granule release, coagulation, and complement dependent lysis have also been used as indicators to detect antiplatelet antibodies. These tests, referred to as ``Phase I'' tests by Kelton,2 are all indirect assays. They have low sensitivity and speci®city for the diagnosis of ITP and are no longer in general use as clinical laboratory assays for autoantibodies. An increase in the amount of IgG associated with platelets from patients with ITP (Platelet Associated IgG or ``PAIgG'') can be quantitatively determined. Initially reported by Dixon and Rosse3 and by McMillan and colleagues4 and then con®rmed in literally hundreds of subsequent studies, these are the ``Phase II'' assays compiled by Kelton2. Methods included radioimmunoassays, enyzme-linked immunoassays, and ¯ow cytometry utilizing antiglobulin reagents. Reproducibility is generally excellent within a single laboratory using a single technique, instrument, operator, and the same reagents; however, laboratory-to-laboratory variability is high. It soon became clear however, that PAIgG is also elevated in many other conditions that involve increased platelet destruction, such as sepsis, disseminated intravascular coagulation, viral infection, or collagen vascular diseases.5,6 The high sensitivity of elevated

Coombs test

PAIgG in patients with ITP (91 %) contrasts with poor speci®city (27%) for these tests as an aid in the diagnosis of ITP6. A feature shared by all platelet destructive conditions is a reduction in the platelet lifetime. Therefore, the average age of the remaining circulating platelets is decreased, i.e. they are ``younger'' than normal. Platelet age can be quantitated by determining the fraction or absolute number of ``reticulated'' platelets, those platelets that have recently been produced from megakaryocytes. These platelets contain a small amount of RNA and may be identi®ed by ¯ow cytometry after staining with the ¯uorescent dye, thiazole orange.7,8 Blanchette and colleagues9 have shown that reticulated platelet counts are increased in children with destructive thrombocytopenias, and Peterec et al.10 reported high reticulated platelet counts in newborns with immune thrombocytopenia. As demonstrated by George and colleagues,11 the IgG in platelets is contained in their alpha granules in an amount proportional to the IgG concentration in the patient's plasma. The IgG is probably taken up by megakaryocytes during maturation and gradually decreases during the lifetime of the platelets. Thus, younger platelets have more platelet associated IgG than do platelets of average age. The increase in PAIgG cannot be completely explained by the larger size of young platelets.12 The vast majority of patients with ITP have platelets with increased IgG, although the ®nding of increased

Platelet Autoantibodies in ITP 239

PAIgG is not speci®c for a diagnosis of ITP. Most of the PAIgG measured in ``antiplatelet antibody'' tests is not true antiplatelet antibody, but is simply a feature of the younger platelets typical of all platelet destructive states. In this regard, PAIgG is more analogous to the erythrocyte reticulocyte count than to the Coombs' test.

PLATELET GLYCOPROTEINS ARE THE TARGETS FOR ITP ANTIBODIES The platelet surface contains a number of glycoproteins that mediate cellular interactions involving platelets. These proteins are named by Roman numerals I to IX according to their apparent molecular weights as determined by mobility in gel electrophoresis. Individual proteins originally named as a single band but identi®ed as multiple polypeptides by further separation techniques are also noted with letter (a, b, etc.) designations. A number of these proteins are members of the intergrin family, complexes containing alpha and beta subunits. The process by which platelets are linked to macromolecules in the subendothelial tissues is termed adhesion while binding of platelets to each other is known as aggregation. Table 2 lists platelet speci®c receptor complexes involved in these events. The most numerous and functionally important of these protein complexes are GPIbIX and GPIIbIIIa. Under high shear conditions, adhesion involves Table 2.

binding of platelet GPIbIX to the subendothelium via the molecular ``bridge'', von Willebrand factor. After platelet activation, aggregation occurs as platelets are linked to each other by ®brinogen (or von Willebrand factor) binding to GPIIb/IIIa. Attempts to characterize the molecular details of immune platelet destruction began with the identi®cation of the surface glycoproteins that are the targets for platelet autoantibodies. In 1982, van Leeuven et al.13 reported that antibodies from most patients with ITP reacted with all normal platelets but not with platelets from patients with Glanzmann's thrombasthenia. This indicated that ITP target antigens are most commonly located on the glycoprotein (GP) IIb/IIIa complex, which is absent from thrombasthenic platelets. This was con®rmed by immunobead testing.14,15 lmmunoblotting studies further indicated that the antiplatelet antibodies usually bind to the GPIIa part of the complex,16 although antiGPIIb antibodies have also been detected.17 As antiplatelet antibodies have been studied in more depth, it has been determined that a number of other platelet surface glycoproteins can also be the targets for antiplatelet antibodies. Major platelet glycoprotein complexes that have been identi®ed as targets for autoantibodies include GPIb/IX, GPV, GPIa/IIa, and GPIV. In addition, other antigenic proteins have been identi®ed only by their apparent molecular weights when separated by electro-

Platelet glycoproteins

Glycoprotein

Ligand

Function

GpIb/IX* GPIIb/IIIa GPIa/IIa GPIc/lIa Alpha6/GPIIa GPV* GPIV, GPVI*

von Willebrand factor Fibrinogen, vWF, ®bronectin Collagen Fibronectin Laminin Thrombin Collagen

Adhesion Aggregation Adhesion Adhesion Adhesion Enzyme target Adhesion

*denotes nonintegrin.

240 Transfus. Sci. Vol. 19, No. 3

phoretic techniques. Characterization of many of these autoantibody targets used laboratory methods applicable to the study of any antigenic protein. Techniques that have been most successful in this endeavor include immunoblotting and immunoprecipitation. Both of these approaches are technically demanding and are not easily applicable to use as routine clinical laboratory testing. An advantage of immunoprecipitation is that the autoantibodies are exposed to potential target antigens on intact platelets. Therefore, antigens that depend upon multimeric protein complexes and native three dimensional conformation are preserved. Immunoblotting, on the other hand, utilizes denatured, separated, and renatured proteins as the potential targets for autoantibodies. Conformation-dependent antigens are more likely to be lost in this process. Somewhat surprisingly, many antigens are preserved, and this technique allows the identi®cation of the particular antigenic polypeptide target (e.g. the GPIIIa portion of the GPIIb/IIIa complex).16

ASSAYS FOR ANTIPLATELET AUTOANTIBODIES Direct assays for platelet antibodies are those that test for antibodies present on a patient's own platelets. If the platelet count is extremely low, this type of testing may not be possible. Indirect assays involve reaction of a patient's serum with normal target platelets. Antiplatelet antibodies detected by this type of testing could be autoantibodies or alloantibodies (against ABO, HLA, or platelet speci®c antigens). A clinical history regarding transfusion and pregTable 3.

nancy may therefore be helpful in interpreting the results. The most useful clinical assays for antiplatelet autoantibodies are antigen immobilization assays, also referred to ``Phase III'' assays.2 Platelet glycoproteins, puri®ed from platelet extracts using murine monoclonal antibodies, are used as the target for antiplatelet antibodies in this approach. The individual methods have been described and illustrated in three excellent recent reviews 2,18,19 and will not be detailed here. The most clinically valuable tests are those that allow for incubation of patient serum or plasma with intact platelets that can present surface glycoproteins in the native, complexed state before the immune complex is ``captured'' by the monoclonal antibody. These have been referrd to as ``immune complex capture'' assays.18 Tests using platelet glycoproteins that have been ``captured'' from solubilized platelets and immobilized prior to immobilization give a lower sensitivity because complex-dependent epitopes may be destroyed. Furthermore, cytoplasmic epitopes are exposed which can lead to false positive results. Table 3 lists these three approaches to antiplatelet antibody testing that utilize antigen immobilization techniques. One antigen immobilization method currently used in many diagnostic platelet immunology laboratories is the ``MAIPA'' (monoconal antibody immobilization of platelet antigens), developed by Keifel and colleagues.20 In this particular technique, intact platelets are incubated with serum or plasma simultaneously with the monocional antibody that will be used to ``capture'' the glycoprotein complex onto a micro-

Antigen Immobilization Assays Employed by Diagnostic Laboratories for the Detection of Antiplatelet Antibodies

Assay MAIPA Immunobead ACE/MACE

Technique

Incubation State

Detection

Immune complex immobilization Immune complex immobilization Antigen capture ELISA

Intact platelets Intact platelets Isolated GPs

ELISA Radiolabel ELISA

Platelet Autoantibodies in ITP 241

titer well. Therefore, antibodies against internal or cytoplasmic antigens will not be detected. Such antibodies are unlikely to be a cause of ITP, although they are frequently found in the sera of patients with platelet destructive disorders, presumably resulting from an immune reaction to previously inaccessible platelet antigens (``cryptic antigens''). It should also be noted that an autoantibody may not be detected if it is directed against the same epitope as the monoclonal antibody used to capture the glycoprotein complex, because the autoantibodies tend to be weaker than the murine monoclonal antibodies. For this reason, it is wise to use two capture monoclonal antibodies with different speci®cities as the capture reagent in separate parallel assays. A positive assay with either reagent (or, of course, both reagents) is interpreted as a positive result. As it was ®rst described, the MAIPA assay was susceptible to false positive results if the patient had preformed antibody against mouse immunoglobulin. A modi®ed MAIPA assay that includes a preincubation step appears to solve this potential problem.21 It should be noted that the antigen capture techniques can only differentiate platelet autoantibodies from alloantibodies if the antibody is shown to bind to autologous platelets. This can be accomplished by the direct assay (i.e. using the patient's own test platelets.) However, thrombocytopenia may preclude this approach in the acute setting. An alternative is to repeat the indirect assay with stored plasma using the patient's own platelets after an increase in the platelet count allows enough platelets to be obtained. Antigen immobilization techniques were assessed in an international workshop organized by Berchtold et al.22 A group of 8 laboratories participated in this blinded study, performing both direct and indirect antiplatelet antibody assays for anti-GPIIb/IIIa and anti-GPIb/ IX for 22 patients with ITP. Direct antibody assays had excellent agreement among the laboratories (82%), although

sensitivity and degree of agreement was only 39% for the indirect assays. Negative controls were negative for >98% of tests. Antigen immobilization methods are the most promising assays to date for the clinical diagnosis and management of ITP. McMillan18 found that them helpful in the diagnosis of immune thrombocytopenia in patients with collagen vascular diseases or lymphoproliferative syndromes. Furthermore, antiplatelet antibody levels were shown to correspond to the response to therapeutic interventions.18

TARGET EPITOPES OF PLATELET AUTOANTIBODIES The platelet GPIIb/IIIa complex is recognized as the most frequent target of antiplatelet antibodies found in patients with ITP. There is limited information about the actual submolecular binding sites for these antibodies. Some data suggest that the usual ITP epitopes are ``conformational'', depending upon the three dimensional folded structure of an intact GPIIb/IIIa complex in order to allow antibody binding.14,15,23,24 Other studies report binding to the individual GPIIb and GPIIIa proteins after separation on denaturing polyacrylamide gels.16,17 Digestion of platelet GPIIIa with proteolytic enzymes has been used in an attempt to identify the speci®c epitope regions of the molecule. Kekomaki and colleagues25 determined that a 50kDa chymotryptic digestion fragment retained the binding site for 33 of 39 anti-GPIIIa antibodies, localizing these antigens to the disul®de rich, immediately extracellular domain of the molecule. Recombinant GPIIIa peptides have been tested for reactivity with ITP antibodies (see Table 4). Fujisawa et al.23 reported that 5 of 13 antibodies reacted with the cytoplasmic carboxyterminal domain. The investigators suggested that these antibodies may

242 Transfus. Sci. Vol. 19, No. 3

have resulted from the platelet destruction in ITP rather than being the cause of the immune thrombocytopenia. Further investigations by this laboratory suggested that sequence speci®c external antigens are unusual targets of eluted ITP antibodies (only 1 of 33 positive), and that complex dependent, conformational epitopes are more common.24 In contrast, Beardsley et al.26 reported the creation of extracellular recombinant GPIIIa peptides that bound anti-GPIIIa ITP antibodies. Fusion proteins containing peptide sequences from the disul®de rich region of GPIIIa were shown to bind the majority of anti-GPIIIa antibodies from a group of 24 ITP patients. These epitopes could not be complex dependent, since no GPIIb sequence was present. The GPIIIa peptides varied in length from 56 to 691 residues (see Table 4). The antibody binding was demonstrated by immunoblotting, suggesting that the epitopes are not conformation dependent, but sequence de®ned. These antigens were peptide, rather than glycopeptide; since they were expressed in E. coli. Further studies will be needed

to determine whether both conformation dependent and sequence de®ned platelet autoantigens are found in ITP patients or whether particular antigens may be typical of a unique subset of patients.

SUMMARY In summary, the search for a useful clinical laboratory diagnostic assay for the antiplatelet antibodies has been long and dif®cult. Measurement of platelet associated IgG (PAIgG) has been disappointing as a way to detect autoantibodies. This is primarily due to the fact that platelets normally contain IgG in their alpha granules in an amount that varies with plasma IgG levels and age of the platelets. Furthermore, the amount of platelet associated IgG is affected by the presence of circulating immune complexes, platelet activation, and drug dependent antibodies. The newer, platelet antigen capture techniques are promising, but further testing will be needed to con®rm their value to the clinician. Methods that allow incubation of patient serum or plasma with

Table 4. Glycoprotein IIIa Peptides Target peptide *

sGPIIIa : sGPIIIa: sGPIIIa: sGPIIIa: sGPIIIa: sGPIlIa: sGPIIIa: sGPIIIa: sGPIIIa: rGPII1a**: rGPlIla: rGPIlIa: rGPIIIa: rGPIIIa: rGPIIIa: rGPIIIa: rGPIIIa: rGPIIIa: *

93-103 118-128 154-165 400-417 561-573 636-654 669-680 721-744 742-762 1-691 468-690 446-501 593-691 1-200 150-400 350-550 450-700 715-762

Frequency 0/5 0/5 0/5 0/5 0/5 0/5 0/5 3/5 315 416 16/24 5/16 5/16 0133 0133 1133 0/33 0/33

Detection technique Radioimmunoassay

Immunoblotting

26

Radioimmunoassay

sGPlIla=synthetic peptide; **rGPlIla=recombinant fusion protein.

23

24

Platelet Autoantibodies in ITP 243

intact platelets (MAIPA and immunobead) have greater sensitivity than techniques in which the patient antibody is tested against previously isolated platelet glycoproteins. These assays are currently available in a only a limited number of platelet immunology laboratories. Platelet autoantibodies are directed against a number of glycoprotein antigens on the platelet surface. Most studies have shown that anti GPIIb/IIIa antibodies are the most common, although antibodies against GPIb/IX and other targets are frequently detected. Many patients have multiple antiplatelet antibodies circulating simultaneously. The clinical signi®cance of antibodies with different speci®city is under investigation. The precise epitopes on GPIIIa that bind antiplatelet autoantibodies have been studied to a limited extent. Some investigators report that the vast majority of platelet antigens are conformation dependent, being destroyed by treatment with EDTA (separation of GPIIb and GPIIIa) or denaturation with detergents. Others report sequence speci®c peptide antigens. Further investigation promises to better de®ne the targets for platelet autoantibodies; improved clinical management of patients with ITP is the long term goal of these studies.

REFERENCES 1. Harrington WJ, Sprague CC, Minnich V,et al.: Immunologic mechanisms in idiopathic and neonatal thrombocytopenic purpura. Ann Intern Med 1953;38: 433±469. 2. Kelton JG: The measurement of platelet-bound immunoglobulins: an overview of the methods and the biological relevance of platelet-associated IgG. Prog Hematol 1983;8:163±169. 3. Dixon R, Rosse W, Ebbert L: Quantitative determination of antibody in idiopathic thrombocytopenic purpura: correlation of serum and platelet-bound antibody with clinical response. N Engl J Med 1975;292:230±236.

4. McMillan R, Smith RS, Longmire R, et al.: Immunoglobins associated with human platelets. Blood; 37:316±322. 5. Mueller-Eckhardt C, Kayser W, et al.: The clinical signi®cance of plateletassociated IgG: a study on 298 patients with various disorders. Br J Haematol 1980;46:123±131. 6. Kelton J, Powers PJ, Carter CJ: A prospective study of the usefulness of the measurement of platelet-associated IgG for the diagnosis of idiopathic thrombocytopenic purpura. Blood 1982;60:1050± 1053. 7. Kienast J, Schmitz G: Flow cytometric analysis of thiazole orange uptake by platelets: a diagnostic aid in the evaluation of thrombocytopenic disorders. Blood 1990;75:116. 8. Ault KA, Rinder HM, Michell J, Carmody MB, Vary CP, Hillman RS: The signi®cance of platelets with increased RNA content (reticulated platelets). A measure of the rate of thrombopoiesis. Am J Clin Path 1992;98:637. 9. Saxon BR, Blanchette VS, Butchart S, Lim-Yin J, Poon AO: Reticulated platelet counts in the diagnosis of acute immune thrombocytopenic purpura. Am J Pediatr Hematol Oncol 1998;20:44. 10. Peterec SM, Brennan SA, Rinder RM, Wnek JLL, Beardsley DS: Reticulated platelet values in normal and thrombocytopenia neonates. J Ped 1996;129:269± 274. 11. George JN: A platelet immunoglobulin G: its signi®cance for the evaluation of thrombocytopenia and for understanding the origin of a-granule proteins. Blood 1990;76:859±870. 12. Kelton JG, Neame PB, Bishop J, et al.: The direct assay for platelet-associated IgG (PAIgG): Lack of association between antibody level and platelet size. Blood 1979;53:73±80. 13. van Leeuwen EF, vander Ven JTH, Engelfriet CP, et al.: Speci®city of autoantibodies in autoimmune thrombocytopenia. Blood 1982;59:23±26. 14. Woods VL, Oh EH, Mason D, et al.: Autoantibodies against the platelet glycoprotein IIb/IIa complex in patients with chronic ITP. Blood 1984;63:368± 375. 15. Woods VL, Kurata Y, Montgomery RR, et al.: Autoantibodies against platelet glycoprotein Ib in patients with chronic

244 Transfus. Sci. Vol. 19, No. 3

16.

17.

18. 19. 20.

21.

idiopathic thrombocytopenic purpura. Blood 1984;64:156±160. Beardsley DS, Spiegel JE, Jacobs MM, et al.: Platelet membrane glycoprotein IIIa contains target antigens that bind antiplatelet antibodies in immune thrombocytopenias. J Clin Invest 1984;74: 1701±1707. Tomiyama Y, Kurata Y, Mizutani H, et al.: Platelet glycoprotein IIb as a target antigen in two patients with chronic immune thrombocytopenic purpura. Br J Haematol 1987;66:535±538. McMillan R: Antigen-speci®c assays in immune thrombocytopenia. Transfus Med Rev 1990;4:136±143. Warner M, Kelton JG: Laboratory investigation of immune thrombocytopenia. J Clin Path 1997;50:5. Kiefel V, Santoso S, Weisheit M, et al.: Monoclonal antibody-speci®c immobilization of platelet antigens (MAIPA): a new tool for the identi®cation of platelet-reactive antibodies. Blood 1987;70: 1722±1726. Kiefel V, Santoso S, Mueller-Eckhardt C: Serological, biochemical, and molecular aspects of platelet autoantigens. Sem Hematol 1992;29:26±33.

22. Berchtold P, Muller D, Beardsley D, Fugisawa K, Kaplan C, Kekomaki R, Lipp E, Morell-Kopp MC, Kiefel V, McMillan R, von dem Borne AEGKr, Imbach P: International study to compare the measurement of platelet autoantibodies. Blood 1997;96:477. 23. Fujisawa K, O'Toole TE, Tani P, et al.: Autoantibodies to the presumptive cytoplasmic domain of platelet glycoprotein IIIa in patients with chronic immune thrombocytopenia purpura. Blood 1991;77:2207. 24. Bowditch RD, Tani P, McMillan R: Reactivity of autoantibodies from chronic ITP patients with recombinant glycoprotein IIIa peptides. Br J Haematol 1995;91:178. 25. Kekomaki R, Dawson B, McFarland J, Kunicki TJ: Localization of human platelet autoantigens to the cysteine-rich region of glycoprotein IIIa. J Clin Invest 1991;88:847. 26. Beardsley DJS, Tang C, et al.: Platelet autoantibodies react with a recombinant glycoprotein (GP) IIIa peptide containing residues 593±690 of GPIIIa. Blood 1995; 86:227a.