• platelet activation;
  • ITP;
  • protein A immunoadsorption;
  • P-selectin;
  • platelet-associated immunoglobulin


  1. Top of page
  2. Abstract
  6. References

Adults with chronic relapsing ITP present a difficult therapeutic challenge. The ongoing antibody- mediated platelet destruction in this group might be expected to be associated with increased expression of platelet surface membrane activation antigens.

We have studied a group of 10 patients with refractory ITP and 35 healthy controls. Using an immediate, sensitive, unfixed, whole blood, flow cytometric method to detect platelet surface P-selectin and GP53, we have detected markedly increased platelet activation in the ITP group compared with the controls (P-selectin; patient median 24.5% v control median 2.0%, GP53 median 6.5% v 2.1%, P < 0.01 for both).

Five patients underwent protein A immunoadsorption therapy. The effect of protein A immunoadsorption on platelet activation before, during and after 18 treatments in these patients was studied and patients were followed-up to assess clinical outcome. Platelet-associated immunoglobulin measurements were made before and at the end of six treatments.

Platelet activation decreased after immunoadsorption. P-selectin expression fell significantly; pre- and post-treatment median values differed by 15.5%, P < 0.01, for GP53 the difference was 2.5.%, P = NS. A reduction in both platelet-associated IgG (median reduction of 11.8 ng/106 platelets, P = 0.08) and IgM (7.6 ng/106 platelets, P = 0.06) was recorded.

ITP in adults is frequently resistant to first- and second-line therapy (steroids and intravenous immunoglobulin) ( Newland et al, 1983 ; Manoharan, 1991) and patients may proceed to splenectomy. Splenectomy, although useful ( McMillan et al, 1972 ), is not universally successful, and some patients, having responded, relapse. For these patients the next treatment options are often cytotoxic or immunosuppressive therapies ( Andersen, 1994) which are associated with significant side-effects. Such patients may respond to immunoadsorption with protein A columns ( Snyder et al, 1992 ).

In ITP the patients' platelets are frequently coated with immunoglobulin, some of which is directed against specific platelet surface antigens such as GPIIb/IIIa ( Brighton et al, 1996 ), although the majority is non-specifically bound to the platelet membrane. This immunoglobulin may be derived from within the platelet alpha granules ( George et al, 1985 ). The binding of immunoglobulin and immune complexes to the platelet surface has been associated with platelet activation ( Yanabu et al, 1991 ). Platelet activation is associated with alterations in platelet size and shape, alterations in transmembrane molecules, such as GPIIb/IIIa and the release of intraplatelet substances to the cell membrane and plasma. Two such molecules, normally held intracellularly in alpha and lysosomal granules respectively, are P-selectin (CD62P) ( McEver & Martin, 1984; Hsu-Lin et al, 1984 ) and GP53 (CD63) ( Nieuwenhuis et al, 1987 ). These molecules are expressed solely on the surface of activated platelets and are useful markers of platelet activation. P selectin is an adhesion molecule which functions as a ligand for platelet–neutrophil binding ( Larsen et al, 1989 ). The function of GP53 is currently unknown.

We have previously reported a strong positive correlation between P-selectin and platelet-associated IgG (PAIgG) ( Morris et al, 1991 ). Platelet activation and immunoglobulin binding appear to be closely related. Platelet destruction is likely to result from this bound autoantibody or from immune complexes binding to platelet Fc receptors. Platelets, opsonized in this way, are destroyed in liver, spleen and bone marrow ( Imbach, 1995).

We hypothesized that sufficient immunoglobulins might be removed by protein A immunoadsorption to reduce platelet activation. This might lead in turn to longer platelet survival and translate to improvement in the platelet count.

To investigate these hypotheses we examined the platelets of 10 adults with severe refractory ITP. We studied platelet activation flow cytometrically. One previous study reported elevated P-selectin in a minority (17%) of children with ITP ( Semple et al, 1996 ) and this is the first study to report elevated P-selectin and GP53 in adults with ITP.


  1. Top of page
  2. Abstract
  6. References

Protein A columns

Protein A columns are therapeutic immunoadsorption devices which fit into a plasmapheresis circuit. The column removes circulating immune complexes and immunoglobulin G ( McDougal et al, 1979 ) from the plasma and all other components are returned to the patient. The sterile columns (PROSORBAc (Imré)) contain 200 mg of highly purified protein A, covalently bound to a silica matrix, a small amount of which may leech into the patient's circulation without clinical effect ( Balint & Jones, 1995).

Patients and controls

Following ethical approval and informed consent, patients were recruited from the haematology outpatients department. Patients with platelet counts < 85 × 109/l and a diagnosis of ITP requiring active treatment were selected. 10 consecutively attending patients were studied. Of these 10 patients, five were offered treatment with protein A columns. These five patients had severe refractory ITP, having undergone three or more treatment modalities without lasting remission. These five patients were studied serially throughout their treatment with protein A columns. A further two patients had protein A immunoadsorption with measurement of platelet associated IgG and IgM before and at the end of column treatment, but not P-selectin or GP53; this data is included in Figs 1 and 2. Patient characteristics are given in Table I. A group of 35 healthy controls of similar age, on no medication, were used to delineate the normal ranges of activation antigens.


Figure 1. 06 platelets).

Download figure to PowerPoint


Figure 2. Fig 2. PAIgM before and after protein A immunoabsorption (normal values 0–1.5 ng/106 platelets).

Download figure to PowerPoint


Samples for baseline activation studies were obtained from all 10 patients at entry into the study. Patients were treated with protein A columns receiving six treatments each. The gap between treatments varied from 3 to 9 d but in general took place two days per week for three consecutive weeks.

Blood for assessment of the effects of protein A columns on the platelet activation were obtained immediately before and after each treatment, i.e. when the patient was connected to and disconnected from the plasmapheresis machine. The assay for P-selectin and GP53 was commenced within a maximum of 10 min of collection.

Platelet surface expression of P-selectin and GP53 was studied using an unfixed whole blood flow cytometric method as previously reported ( Cahill et al, 1993 ). Briefly, peripheral venous samples were obtained with minimal trauma. 4.5 ml of blood was collected into 3.8% sodium citrate. Whole blood was diluted 1:10 in filtered Tyrode's solution and incubated with antibody within a maximum of 10 min of sampling. No fixation was used. Fluorescent conjugated anti-P-selectin (CD62P) and anti-GP53 (CD63) were used (Immunotech) with fluorescent conjugated F(ab)2 goat anti-mouse immunoglobulin (Dako) as a negative control. Samples were incubated for 10 min and then analysed by flow cytometer (Becton Dickinson FACScan) after further dilution with Tyrode's solution. The methodology was subjected to rigorous analysis before clinical studies commenced. Alterations of assay parameters such as time to analysis, washing, fixation and antibody and control concentration were repeatedly examined ( Cahill, 1995) before this study. The coefficient for inter-assay variation in the measurement of P-selectin and GP53 expression was between 10% and 15%.

Total platelet associated IgG and IgM was also measured flow cytometrically as previously described ( Goodall & Macey, 1994). Briefly, blood was collected into sodium ethylenediaminetetra-acetic acid (EDTA) then centrifuged at 100 g for 15 min. The platelet-rich plasma was removed to a sterile tube and washed three times at 400 g with phosphate-buffered saline containing 9 m m EDTA, pH 6.8. The platelets were then resuspended in PBS containing 1% azide and the count adjusted to 200 × 109/l. Aliquots (100 μl) were incubated with either 5 μl of fluorescein isothiocyanate conjugated (FITC) F(ab′)2 anti-human IgG, IgM isotype negative control or CD41 positive control (Dako, High Wycombe). After incubation at room temperature for 15 min the samples were diluted to 1 ml and analysed by flow cytometry. The PAIg fluorescence was converted to  ng/106 platelets from a constructed standard calibration curve of antibody concentration versus fluorescence intensity.

Assessment of response to columns

Response was assessed by serial platelet counts at the end of therapy with six column treatments per patient and follow-up was continued for 5 years or until death.


  1. Top of page
  2. Abstract
  6. References

The expression of P-selectin and GP53 before immunoadsorption treatments was significantly higher in patients compared to controls ( 1Table II).

Table 1. Table II. Median and interquartile range (IQR) for P-selectin and GP53 in 10 patients with ITP before protein A column treatments.Thumbnail image of

Column results

Platelet-associated IgG (median reduction of 11.8 ng/106 platelets, P = 0.08) and IgM (7.6 ng/106 platelets, P = 0.06) was reduced after immunoadsorption in comparison to levels recorded at the start of treatment ( Figs 1 and 2), although this was of borderline significance only (Wilcox test) most likely due to lack of power. Data on PAIgG and PAIgM from one patient is missing. Therefore in Figs 1 and 2 six data sets are shown (four patients in whom platelet activation was studied and two additional patients who underwent column treatment but did not have platelet activation studied).

Table III summarizes the clinical responses to the column treatment.

Of the 18 column treatments (in five patients) which were fully evaluated, antigen expression of P-selectin was 22% (10.9–38.5%, median and IQR) pre-treatment and 6.5% (2.8–13.7%) post-treatment (P = 0.01). Evaluation of 17 episodes for platelet GP53 expression showed 8.1% (1.7–17%) pre-treatment and 5.6% (1.8–13.2%) post-treatment (P = ns) ( Figs 3 and 4). There was an increase in activation between column treatments, although P-selectin showed a progressive fall throughout treatment. This trend is demonstrated in the data from one patient (Fig 5).


Figure 3. after protein A immunoadsorption.

Download figure to PowerPoint


Figure 4. Fig 4. Changes in P selectin after protein A immunoadsorption.

Download figure to PowerPoint


Figure 5. January 1993 and 22 January 1993. ‘Pre’ samples were taken just before the pheresis commenced and ‘post’ samples as the patient was being disconnected from the machine on each occasion.

Download figure to PowerPoint

There was a negative correlation between the expression of P-selectin and platelet count r = −0.46, n = 10, P = 0.22 (see Fig 6), although this did not reach statistical significance. There was no correlation between the expression of GP53 and platelet count (data not shown). P-selectin was positively correlated with PAIgG (r = 0.6, P = 0.09, n = 5). Serum levels of IgG and IgM were also measured. In two patients plasma IgG (but not IgM) fell after column treatment. These patients also had reductions in PAIgG (both patients) and IgM (one patient). Levels were stable in the remaining patients.


Figure 6. Fig 6. Correlation of P-selectin and platelet count in patients with chronic ITP.

Download figure to PowerPoint

Clinical responses

Follow-up of the five protein A treated patients revealed only one sustained remission. However, this patient subsequently relapsed and died due to massive gastrointestinal haemorrhage. This patient was simultaneously treated with chemotherapy ( Bernard et al, 1993 ) during the column treatment. There was one death from thrombocytopenic intracerebral haemorrhage in a patient who failed to respond to immunoadsorption therapy, and there were three other treatment failures. One patient responded to splenectomy after column treatment and remains well.


  1. Top of page
  2. Abstract
  6. References

These results support the conclusion that platelet activation is markedly increased in autoimmune thrombocytopenic purpura. The presence of platelet activation in ITP is not unexpected. Indeed, the results of this study support those of other workers whose data associated platelet surface immunoglobulin with platelet activation, measured by a variety of techniques. Immunoglobulin binding to the platelet surface has been linked to alpha granule release ( George, 1990; George et al, 1985 ), and F(ab′)2 fragments have also been associated with activation ( Sugiyama et al, 1987 ). The binding of platelet-specific immunoglobulin to GPIIb/IIIa has been shown to cause platelet aggregation and granule release in ITP ( Yanabu et al, 1991 ). Platelet activation increases Fcγ RII expression ( McCrae et al, 1990 ) and some platelet activation may be mediated via Fcγ RII ( Worthington et al, 1990 ).

Therefore we propose a positive feedback loop promoting platelet activation and enhancing platelet destruction. Activated platelets may be more susceptible to immune destruction due to increased Fcγ expression ( McCrae et al, 1990 ). Immune complex binding to Fc receptors is associated with increased platelet destruction which can be blocked by infusion of Fcγ fragments in vivo ( Debre et al, 1993 ).

In summary: initiating event — platelet alpha granule release — increased platelet associated IgG and IgM binding — increased Fcγ RII expression — further platelet activation — alpha granule release — platelet destruction.

Our hypothesis, substantiated by previous data ( Morris et al, 1991 ), predicts also that the most severely affected patients should have highest expression of platelet surface markers of activation. Our data also demonstrate this, showing a negative correlation between P-selectin and platelet count (r = −0.46, Fig 6). Although not statistically significant (most probably because of lack of power), this data confirms previous findings in larger numbers of patients ( Morris et al, 1991 ) which did show statistical significance.

The data on GP53 show that it too was expressed in increased amounts on the platelet surface in ITP. Protein A immunoadsorption also decreased the expression of this molecule, although the results were not significant. Both were sensitive markers of platelet activation, although the data on GP53 were more variable than that on P-selectin. We have previously shown that P-selectin and GP53 behave differently and that their expressions are not normally correlated ( Cahill et al, 1996 ).

Documenting platelet activation is difficult because measurement techniques may cause in vitro activation ( Cahill et al, 1993 ). Before this study was undertaken, an extensive analysis of analytical and pre-analytical variables was undertaken in our laboratory (see Methods). As a result, the risks of ex vivo activation were minimized by using the whole blood technique and rapid analysis.

Protein A binds the Fc fragment of IgG with an increased affinity for complexed IgG ( McDougal et al, 1979 ). The binding capacity of the columns used in this study (PROSORBAc (Imré)) is about 1 g of IgG per elution cycle. The small absorption capacity of the column and the variable volume of plasma processed with success (250–2000 ml with no apparent difference in outcome ( Snyder et al, 1992 )), make it unlikely that bulk removal of IgG and complexes is important. However, removal of small amounts of immunoglobulin may be enough to restore the idiotype/anti-idiotype control networks ( Berchtold et al, 1989 ), or to reduce platelet activation. The data demonstrate a reduction in both IgG and IgM bound to platelets which cannot be explained simply by immunoglobulin removal by the column since protein A does not bind IgM ( Goding, 1978). It may be that reductions in both PAIgG and PAIgM are brought about because of the reduction in platelet activation, reduction in alpha granule secretion and therefore reduction in granule-derived PAIgM, after column treatment. In addition, reduction in platelet activation may make it more likely that plasma immunoglobulins do not become platelet associated or surface bound.

Our data showed an overall decrease in platelet activation after protein A immunoadsorption which accompanied most (but not all) column treatments. It is possible that the most activated platelets remained bound to the columns, leaving the least activated in the circulation to be analysed. However, in the small surface area available, bound platelets were unlikely to significantly influence the characteristics of the circulating platelet mass. Attempts to remove column bound platelets for analysis would possibly result in significant spurious platelet activation ( Goodall, 1991).

In a few cases activation antigen expression appeared to increase. A number of explanations for this phenomenon include host factors on that day (e.g. acute reactions associated with protein A), chance and within-assay variation where the ‘increases’ are small. One patient (patient 5) had a moderate hypersensitivity reaction while undergoing immunoadsorption and his GP53 rose from 4.7% to 13.2% after treatment. P-selectin was unchanged. Another patient (patient 4) showed a rise in P-selectin (GP53 unchanged) from 17.6% to 30% on his fourth column treatment. Five other treatments in this patient were accompanied by a fall in expression. This patient also had hypersensitivity reactions to protein A immunoadsorption after a number of treatment episodes.

The failure of the protein A columns to produce clinical benefit was probably due to the failure of the therapy to reverse either the initiating cause of the platelet activation (usually attributed to immune dysregulation with or without a viral trigger) or the platelet destruction. Our data are consistent with a temporary beneficial effect on platelet surface membrane factors which if prolonged might, arguably, lead to remission.

Another view of platelet activation in ITP, for which there is no explicit evidence, might be that platelet activation is protective in ITP. Increased baseline platelet activation is noted in conditions associated with increased thrombosis, e.g. inflammatory bowel disease ( Collins et al, 1994 ), diabetes ( Tschoepe et al, 1995 ) and coronary artery disease ( Cahill et al, 1994 ). It could be postulated that in thrombocytopenic patients a degree of platelet activation may be haemostatically desirable in order to offset low platelet numbers and therefore it may be a physiological response to thrombocytopenia. It is observed clinically that patients with chronic thrombocytopenia bleed less at the same platelet count than those whose thrombocytopenia is acute. One observation which supports this hypothesis comes from a recent study of acute ITP in children which failed to demonstrate significant platelet activation (P-selectin expression) in this condition ( Semple et al, 1996 ). These questions cannot be resolved, however, on the basis of the data in this paper.

In conclusion, our data suggest that increased expression of platelet surface activation antigens is present in chronic ITP and that protein A immunoadsorption decreases platelet activation. This might provide an additional mechanism of action of the protein A columns, although we did not observe clinically significant benefits of this treatment.


  1. Top of page
  2. Abstract
  6. References
  • 1
    Andersen, J.C. (1994) Response of resistant idiopathic thrombocytopenic purpura to pulsed high-dose dexamethasone therapy. New England Journal of Medicine, 330, 1560 1564.
  • 2
    Balint, J.P. & Jones, F.R. (1995) Evidence for proteolytic cleavage of covalently bound protein A from a silica based extracorporeal immunoadsorbent and lack of relationship to treatment effects. Transfusion Science, 16, 85 94.
  • 3
    Berchtold, P., Dale, G.L., Tani, P., McMillan, R. (1989) Inhibition of autoantibody binding to platelet glycoprotein GPIIb/IIIa by anti-idiotypic antibodies in intravenous gamma globulin. Blood, 74, 2414 2417.
  • 4
    Bernard, T.B., Kelsey, S.M., Macey, M.G., Cahill, M.R., Howe, T., Newland, A.C. (1993) Protein A column plasma immunoadsorption in refractory neutropenia and thrombocytopenia. Lancet, 341, 1657 1658.
  • 5
    Brighton, T.A., Evans, S., Castaldi, P.A., Chesterman, C.N., Chong, B.H. (1996) Prospective evaluation of the clinical usefulness of an antigen-specific assay (MAIPA) in idiopathic thrombocytopenic purpura and other immune thrombocytopenias. Blood, 88, 194 201.
  • 6
    Cahill, M.R. (1995) Platelet activation: measurement and clinical significance. M.D. thesis, University of London.
  • 7
    Cahill, M.R., Macey, M.G., Dawson, J.R., Newland, A.C. (1994) Prevention of restenosis after angioplasty. Lancet, 343, 136.
  • 8
    Cahill, M.R., Macey, M.G., Newland, A.C. (1993) Fixation with formaldehyde induces expression of activation-dependent platelet membrane glycoproteins. British Journal of Haematology, 84, 527 529.
  • 9
    Cahill, M.R., Macey, M.G., Newland, A.C. (1996) Resting platelet P-selectin and GP53 fail to correlate in healthy subjects but are closely linked in myeloproliferative disorders. Blood Coagulation and Fibrinolysis, 7, 169 171.
  • 10
    Collins, C.E., Cahill, M.R., Newland, A.C., Rampton, D.S. (1994) Circulating platelets are activated in inflammatory bowel disease: increased platelet aggregation in inflammatory bowel disease (IBD). Gastroenterology, 106, 840 845.
  • 11
    Debre, M., Bonnet, M.C., Fridman, W.H., Carosella, E., Philippe, N., Reinert, P., Vilmer, E., Kaplan, C., Teillaud, J.L., Griscelli, C. (1993) Infusion of Fcγ fragments for treatment of children with acute immune thrombocytopenic purpura. Lancet, 342, 945 947.
  • 12
    George, J.N. (1990) Platelet immunoglobulin G: its significance for the evaluation of thrombocytopenia and for understanding the origin of alpha granule proteins. Blood, 76, 859 870.
  • 13
    George, J.N., Saucerman, S., Levine, S.P., Knieriem, L.K., Bainton, D.F. (1985) Immunoglobulin G is a platelet alpha granule secreted protein. Journal of Clinical Investigation, 76, 2020 2025.
  • 14
    Goding, J.W. (1978) Use of staphylococcal protein A as an immunological reagent. Journal of Immunol Methods, 20, 241 253.
  • 15
    Goodall, A.H. (1991) Platelet activation during preparation and storage of concentrates: detection by flow cytometry. Blood Coagulation and Fibrinolysis, 2, 377 382.
  • 16
    Goodall, A.H. & Macey, M.G. (1994) Platelet associated molecules and immunoglobulins. Flow Cytometry: Clinical Applications (ed. by M. G. Macey), pp. 148 151. Blackwell Scientific Publications, Oxford.
  • 17
    Hsu-Lin, S.-C., Berman, C.L., Furie, B.C., August, D., Furie, B. (1984) A platelet membrane protein expressed during platelet activation and secretion: studies using a monoclonal antibody specific for thrombin activated platelets. Journal of Biological Chemistry, 259, 9121 9126.
  • 18
    Imbach, P. (1995) Immune thrombocytopenia in children: the immune character of destructive thrombocytopenia and the treatment of bleeding. Seminars in Thrombosis and Hemostasis, 21, 305 312.
  • 19
    Larsen, E., Celi, A., Gilbert, G.E. (1989) PADGEM protein: a receptor that mediates the interaction of activated platelets with neutrophils and monocytes. Cell, 59, 305 312.
  • 20
    Manoharan, A. (1991) Treatment of refractory idiopathic thrombocytopenic purpura in adults. British Journal of Haematology, 79, 143 147.
  • 21
    McCrae, K.R., Shattil, S.J., Cines, D.B. (1990) Platelet activation induces increased Fc gamma receptor expression. Journal of Immunology, 144, 3920 3927.
  • 22
    McDougal, J.S., Redecha, P.B., Inman, R.D., Christian, C.L. (1979) Binding of immunoglobulin G aggregates and immune complexes in human sera to staphylococci containing protein A. Journal of Clinical Investigation, 63, 627 636.
  • 23
    McEver, R.P. & Martin, M.N. (1984) A monoclonal antibody to a membrane glycoprotein binds only to activated platelets. Journal of Biological Chemistry, 259, 9799 9804.
  • 24
    McMillan, R., Longmire, R.L., Yelenosky, R., Smith, R.S., Craddock, C.G. (1972) Immunoglobulin synthesis in vitro by splenic tissue in idiopathic thrombocytopenic purpura. New England Journal of Medicine, 286, 681 684.
  • 25
    Morris, A., Macey, M., Newland, A. (1991) Platelet immunoglobulins, glycoproteins and activation antigens in autoimune thrombocytopenic purpura (ATP). British Society for Haematology (meeting supplement), 105.
  • 26
    Newland, A.C., Treleaven, J.G., Minchinton, R.M., Waters, A.H. (1983) High-dose intravenous IgG in adults with autoimmune thrombocytopenia. Lancet, i, 84 87.
  • 27
    Nieuwenhuis, H.K., van Oosterhout, J.J.G., Rozemuller, E., van Iwaarden, F., Sixma, J.J. (1987) Studies with a monoclonal antibody against activated platelets: evidence that a secreted 53,000 molecular weight lysosome-like granule protein is exposed on the surface of activated platelets in the circulation. Blood, 73, 838 845.
  • 28
    Semple, J.W., Milev, Y., Cosgrave, D., Mody, M., Hornstein, A., Blanchette, V., Freedman, J. (1996) Differences in serum cytokine levels in acute and chronic autoimmune thrombocytopenic purpura: relationship to platelet phenotype and anti-platelet T-cell reactivity. Blood, 87, 4245 4254.
  • 29
    Snyder, H.W., Cochran, S.K., Balint, J.P., Bertram, J.H., Mittelman, A., Githrie, T.H., Jones, F.R. (1992) Experience with protein A-immunoadsorption in treatment resistant adult immune thrombocytopenic purpura. Blood, 79, 2237 2245.
  • 30
    Sugiyama, T., Minoru, O., Ushikubi, F., Sensaki, S., Kanaji, K., Uchino, H. (1987) A novel platelet aggregating factor found in a platelet with defective collagen induced platelet aggregation and autoimmune thrombocytopenia. Blood, 69, 1712 1720.
  • 31
    Tschoepe, D., Driesch, E., Schwippert, B., Nieuwenhuis, H.K., Gries, F.A. (1995) Exposure of adhesion molecules on activated platelets in patients with newly diagnosed IDDM is not normalised by near-normoglycaemia. Diabetes, 44, 890 894.
  • 32
    Worthington, R.E., Carroll, R.C., Boucheiux, C. (1990) Platelet activation by CD9 monoclonal antibodies is mediated by the Fc gamma II receptor. British Journal of Haematology, 74, 216 222.
  • 33
    Yanabu, M., Nomura, S., Fukuroi, T., Soga, T., Kondo, K., Sone, N., Kitada, C., Nagata, H., Kokawa, T., Yasunaga, K. (1991) Synergistic action in platelet activation induced by antiplatelet autoantibody in ITP. British Journal of Haematology, 78, 87 93.