SCIENTIFIC SUBCOMMITTEE SESSION
6 July 2007 Palexpo, Geneva, Switzerland 

Platelet Immunology

Chair: T. Warkentin (Canada)
Co-Chairs: B. Chong (Australia), A. Greinacher (Germany), Y. Gruel (France), V. Kiefel (Germany), H. Kroll (Germany)

Committee Co-Chairs (not in attendance): None

The program was divided into several parts: (I) Autoimmune Thrombocytopenia, (II) Alloimmune Thrombocytopenia, (III) Drug-Dependent Thrombocytopenias, (IV) Autoimmune HIT, and (V) Heparin-induced Thrombocytopenia..

Information item : Our committee received as an information item a proposed standard for human antibody against HPA-1a (minimum potency preparation) NIBSC code 05/106 (report of Paul Metcalfe & Mathew Breirley) and this information was transmitted to Dr. Koen Mertens (Chair, WHO-ISTH Standards Liaison Group). This item is for information only as the ISBT is primarily responsible for this study.

AUTOIMMUNE THROMBOCYTOPENIA (Chairs: B. Chong, V. Kiefel) 15:45-16:15h

P. Imbach: Update: Activities of the Intercontinental Childhood ITP Study Group (ICIS). An update of the activities of ICIS were presented. Several unresolved issues in ITP diagnosis and management were listed. Four prospective studies were described, two completed (Registries I and II), and two ongoing (Splenectomy Registry, PARC-ITP). Certain interesting results were highlighted (e.g., marked male predominance in very young ITP children <1 yr of age). Various substudies ("Trees in the PARC") were listed, e.g., Genetics/Single Nucleotide Polymorphisms, Severity of Bleeding, Quality of Life, Refractory ITP, Secondary ITP, and ITP during Pregnancy. Publications of ICIS were listed. The ICIS ITP Expert Meeting (Sep 2006) in Yverdon Switzerland was discussed, with six categories of ITP defined (primary, prolonged, chronic, intermittent, secondary, Evan's syndrome), with staging and response criteria summarized.

F. Rodeghiero Standardization of terminology and definitions in ITP: an update. A summary of the recent efforts to standardize terminologies and definitions for ITP was presented. The heterogeneity of the current literature in this regard was emphasized. Outcomes of the recent meeting of the European Hematology Association (Vienna, June 2007) and information on a planned consensus meeting (Vicenza, October 2007) of ~25 experts with predefined program and structured questionnaires were provided. A relevant website regarding these activities was provided (www.tcpeha.org). In the Discussion, the Platelet Immunology SSC Chair commented upon the potential value of including a laboratory component (autoantibody detection) within the framework of definitions in order to help spur relevant research into laboratory testing for these disparate conditions.

ALLOIMMUNE THROMBOCYTOPENIA (Chairs: Y. Gruel, V. Kiefel) 16:15-17:15h

J.A. Peterson, M.L. Gitter, B.C. Pietz, S.C. Johnson, B. Curtis, R.H. Aster: Simultaneous testing by Multicode â -PLx of DNA for all known platelet alloantigens. After listing the various known platelet alloantigen systems, a technique known as Multicode â -PLx was described, which is a 3-step assay for multiple alleles in a single tube. The technique utilizes multiplex PCR, allele-specific extension, with adherence to "addressed" Luminex beads for detection. The use of a synthetic nucleotide (iC) allows for complementary binding of biotin-iG, for detection by streptavidin. Forty-two different alleles (21 allele systems) could be tested using just 3 wells per patient assayed, with 100 patient samples per day being analyzed. The technique allows for clear discrimination of homozygous and heterogygous allele status.

B. Curtis, RH. Aster: Do ABO blood group alloantigens ever cause NAIT? The authors presented a case in which considerable evidence was provided indicating that the cause of neonatal thrombocytopenia might well have been maternal anti-B alloantibodies. A minority of blood group A and B individuals are "high expressors" of A and B antigens, including on platelets (on PECAM, GPIIb/IIIa). In particular, the "type II" hyperexpressors have a sharp peak with greatly elevated blood group antigen on platelets. In the case presented, the data supported the concept that maternal anti-B resulted in thrombocytopenia in the affected neonate (3rd pregnancy) who was a high expressor of blood group B antigen on platelets. Evidence presented included high-titer anti-B in mother, reactivity of mother's serum with father's GPIIb/IIIa by MACE, absorption of antibody using B red cells, low H expression in father and children, absence of identifiable "specific" platelet alloantibodies, and lack of thrombocytopenic in a subsequent (4th) pregnancy in whom the baby was blood group A, and so forth. The authors suggest considering ABO incompatibility as a possible explanation for NAIT when no other cause can be found, and a relevant ABO discrepancy exists.

I. Socher, C. Andrei-Selmer, G. Bein, H. Kroll, S. Santoso: Severe neonatal alloimmune thrombocytopenia caused by low-avidity anti-HPA-1a alloantibodies: detection by surface plasmon resonance . In 15-25% of cases of suspected NAIT due to anti-HPA-1a alloantibodies, such antibodies cannot be detected. The technique of surface plasmon resonance (SPR) technology permits a rapid assessment in real time without the need for monoclonal antibodies and washing techniques, as it relies on detection of antibodies based upon kinetic (association/dissociation) and affinity considerations. In this technique, HPA-1a and -1b alloantigens are immobilized using HPA-1a-specific monoclonal antibodies. Antibodies can be distinguished by SPR based upon their avidity of binding. The authors propose that antibody avidity may represent a parameter to predict bleeding complications in NAIT.

V. Kiefel, H. Kroll, A. Reil, J. Bux: External quality assessment in platelet serology—conclusions from the German experience. The results of an external quality assessment (EQA) in German labs of platelet serology were reported. Beginning in 2007, two assessments per year are planned (2002-2006, once/yr), involving 2 sera (obligatory), 2 DNA samples (HPA-1,2,3,5 obligatory, HPA-15 optional), and 2 platelet samples (optional). Various problems were described (e.g., difficulty in obtaining certain test samples in sufficient quantities, the expectation of laboratories that only antibodies that are easily detectable using commercial reagents would be tested). Specific frequencies of correct reporting of results for various serologies (including presence of anti-HLA alloantibodies) and DNA testing in various years were shown. The presentation listed the reactivities that are expected to be correctly assessed in the future (HPA-1, 2, 3, 5, 15; presence of anti-HLA alloantibodies; platelet autoantibodies; platelet cross-matching, antigen testing (DNA and serological for HPA-1), and quality of counseling by lab staff).

C. Ghevaert, P. Stafford, K. Campbell, C. Proulx, G. Smith, L. Williamson, E. Ranasinghe, N. Watkins, J. Huntington, W. Ouwehand: Immunological and structural analysis of ten novel domain-deletion β 3 integrin probes designed for detection of HPA-1 and HPA-rare antibodies . Given the importance of the GPIIIa (β 3) domain as a target for several platelet alloantibodies implicated in NAIT (both common and rare) the strategy of using domain-deletion β 3 integrin probes was described. Of note, some alloantibodies reacted with all 4 peptides, whereas some reacted only against the longer fragments. This approach could be used for detecting anti-HPA-1a alloantibodies, though it remained unclear to what extent antibodies not detected by routine assays would be detected using this strategy. Other peptides were generated with rare alloantigen targets.

C. Kaplan and G. Bertrand :NAIT and rare platelet alloantigens. This presentation listed some of the rare platelet alloantigens and the various pitfalls involved in establishing the role of putative rare platelet alloantigens in cases of NAIT.

DRUG-DEPENDENT THROMBOCYTOPENIA (Chairs: H. Kroll, T.E. Warkentin) 17:15-17:45h

R. Aster, D. Bougie: Sensitivity and specificity of lab testing for drug-induced immune thrombocytopenia: quinine and vancomycin as models . Dr. Aster noted that the J. George criteria for evidence for drug-induced thrombocytopenia do not include lab testing. There is little information on sensitivity and specificity of lab testing for drug-dependent antibodies in the literature. Results of screening the normal population for quinine-, sulfamethoxazole-, and ceftriaxone-dependent antibodies showed that antibodies were rare with quinine and sulfa, but 11/466 were positive (4 strong) with ceftriaxone. Some of the difficulties in obtaining such data from hospitalized patients receiving drugs (e.g., vancomycin) were listed. However, of 25 patients who received vancomycin without developing thrombocytopenia, none had detectable antibodies. Patients suspected of having vancomycin-induced thrombocytopenia but in whom antibodies were not detected generally had other explanations for thrombocytopenia found. The authors suggest that sensitivity and specificity of testing for vancomycin-dependent antibodies is probably high. Brief mention was made of a murine model under development utilizing SCID mice in which it may be possible to show in vivo pathogenicity of human drug-dependent antibodies (thus avoiding re-challenge in patients).

D. Bougie, B. Curtis, R. Aster: Naturally-occurring anti-GPIIb/IIIa antibodies. The speaker reviewed the frequencies of thrombocytopenia with six different GPIIb/IIIa receptor antagonists. The frequency of naturally-occurring antibodies with three clinically-used agents (abciximab = 1.6%; tirofiban = 1-2%; eptifibatide = 0.5-4%) correlates with the approximate frequencies of abrupt-onset thrombocytopenia seen in clinical practice. For abciximab, the situation is somewhat more complex, in that 20% of normal individuals have antibodies that react with abciximab-coated platelets, of which 92% of these react against the antibody's papain cleavage site in abciximab, and the remainder (1.6% overall) react against murine sequences incorporated into abciximab that confer specificity for GPIIb/IIIa.

A. Greinacher: Amegakaryocytic thrombocytopenia occurring with anti-GPIIb/IIIa inhibitors . This presentation was waived by the presenter (Committee Co-Chair, A. Greinacher) in the interests of keeping the session on schedule.

AUTOIMMUNE HIT (Chair: A. Greinacher) 17:55-18:15h

T.E. Warkentin, R. Jay, M. Makris, J.G. Kelton: Spontaneous HIT . Three cases were presented that appeared to have clinical HIT (thrombocytopenia, thrombosis or other known sequelae of HIT, presence of strong platelet-activating anti-PF4/heparin antibodies) yet without a plausible history of preceding heparin treatment (either in the recent or remote past). Of note, the patients had one or more recent (or concurrent) inflammatory events. The concept that on exceptionally rare occasions patients could develop "spontaneous" HIT—perhaps as a result of associated inflammation—was suggested.

T. Warkentin, B.T. Maurer, R.H. Aster: Fondaparinx-associated HIT? A case was presented of HIT with thrombosis (bilateral adrenal necrosis, DVT, moderately-severe thrombocytopenia) that occurred on about day 7 of fondaparinux thromboprophylaxis post-orthopedic surgery, in the absence of apparent perioperative exposure to heparin. The serologic feature was the presence in patient serum of platelet-activating antibodies that caused 90% serotonin release in the absence of heparin, although serotonin release increased to 96% release in the presence of therapeutic heparin (with 0% release in the presence of high heparin), i.e., serological features of "delayed-onset HIT". Since fondaparinux therapy is known to be associated with formation of anti-PF4/heparin antibodies (some with platelet-activating properties), the concept is that on rare occasions a patient could form antibodies in association with fondaparinux therapy that might cause HIT irrespective of whether any anticoagulant (including fondaparinux) is given/continued some days later. Some implications of the case, and of this model, were presented, i.e., low-dose fondaparinux might not prevent thrombotic events in the setting of severe HIT-induced hypercoagulability; the occurrence of fondaparinux-associated thrombocytopenia should not be taken to infer that fondaparinux might not be an effective treatment of HIT –indeed, by the model proposed, even such a case of apparent fondaparinux-associated HIT might have been treated effectively by use of (therapeutic-dose) fondaparinux.

R.H. Aster: Delayed-onset HIT: musings on pathogenesis . The speaker reviewed the literature on delayed-onset HIT, emphasizing that administration of further heparin (after presenting with thrombosis/thrombocytopenia) can be associated with catastrophic thrombi, and the presence of unusually high titers of HIT antibodies, many of which showed considerable platelet activation in the absence of added heparin. Investigators have shown that residual heparin is not the explanation for the lack of heparin requirement. Animal studies by the Philadelphia group show that the monoclonal antibody KKO binds to human platelets without requirement for heparin when high levels of platelet-associated PF4 are present. It remains uncertain whether delayed-onset HIT is due to high-titer antibodies in the setting of high PF4 expression by platelets, or whether other unusual characteristics of these antibodies are present.

HEPARIN-INDUCED THROMBOCYTOPENIA (Chairs: T. E. Warkentin) 18:15-19:45h

When should HIT be suspected after cardiac surgery?

C. Pouplard, Y. Gruel: Post-cardiac surgery HIT: Temporal profiles . Two presentations of HIT in post-cardiac surgery patients were presented: type 1 pattern (biphasic), and type 2 pattern (monophasic). The previous evidence that the type 1 profile is more common than the type 2 pattern, and more predictive of HIT (as judged by a positive serotonin-release assay [SRA}), was summarized. New data were shown indicating that the predictive values of these patterns for HIT remain true: 6/7 type 1 pattern patients were shown to have HIT, and 1/8 type 2 pattern patients were shown to have HIT (data from May 1/2005 – Jul 1/2007). The importance of a functional test (e.g., SRA) was emphasized for diagnosis in this patient population.

S. Selleng, A. Greinacher: Post-cardiac surgery HIT: Greifswald study . A study of HIT in post-cardiac intensive care unit (ICU) patients was presented. Both type 1 and type 2 patterns were recognized, though the type 1 pattern was not as specific for HIT as seen in the French study (perhaps reflecting the ICU patient composition in the Greifswald study). A new finding was that a further fall in platelet count in the day 5-10 "window" increased the specificity for HIT in patients with the type 2 pattern.

What tests for HIT are being performed?

J. Zehnder, E.A. Price, C. Hayward, T. Warkentin, K. Moffat, J. Moore: NASCOLA Survey of laboratory practices regarding testing for heparin-induced thrombocytopenia . In this survey of 44 specialized labs that perform testing for HIT antibodies, two objectives were stated: to determine current practice in laboratory diagnosis of HIT, and to identify areas in need of standardization and improvement. A wide variety of specimen types/pre-analytic variables were accepted, with two "concerns" being heparin contamination within samples (78%) and that the timing of the blood draw could be too early for antibody detection (75%). Antigen assays (e.g. EIA [ELISA]) are most commonly performed, especially the assay from GTI (71%) versus the one from Stago (27%), with a wide variety of reporting strategies and test cutoffs utilized. A minority of labs had tried the "rapid" assays (particle gel immunoassay and particle immunofiltration assay). Platelet activation assays were performed in only about 1/3 of labs, with similar numbers using "washed platelets" as platelet-rich plasma. Most labs used prescreened donors known to react well in the activation assays. Again, a wide variety of test conditions (e.g., heparin concentrations) were used. The most common pattern of practice (55% of labs) involved screening with a commercial EIA, and referring out of samples to a reference lab for platelet activation assays. A number of items that might benefit from standardization were presented.

Rapid assays for HIT: DiaMed

Y. Gruel, C. Pouplard: Prospective evaluation of 4T's score and rapid particle gel immunoassay specific to H/PF4 complexes (Pa GIA â ID H/PF4 to exclude heparin-induced thrombocytopenia . The aim of the study was to combine the 4T's score and a rapid assay for HIT—the particle gel immunoassay (PaGIA) from DiaMed—in excluding the diagnosis in emergency situation. This prospective study evaluated 213 patients in 4 hospitals with suspected HIT over 11 months. HIT was diagnosed based upon positive EIA/SRA. About one-third (34.7%) of patients had low 4T's score: none had HIT, though 5/74 had a positive PaGIA. Among the 22 patients with HIT, 21 were positive in the PaGIA, and all 22 in the EIA. In contrast, among the 191 non-HIT patients, 8.4% had a positive PaGIA, whereas 18.3% of 191 non-HIT patients had a positive EIA. The presenters also showed Bayesian approach, and concluded that a negative PaGIA with a low/intermediate score essentially excludes HIT. Also, a high probability 4T's score with a positive PaGIA had a high probability (>90%) of HIT. In other combinations (e.g., high 4T's score, negative PaGIA; intermediate 4T's score, positive PaGIA), the diagnosis of HIT required the SRA. The authors conclude that a low 4T's score is useful for ruling out HIT, and that a negative PaGIA is useful in ruling out HIT in a patient with an intermediate 4T's score. The SRA is useful in many situations for ruling in the diagnosis of HIT.

S.J. McRae, M. Al Muslahi, E.M. Duncan, R. Tadros, J.V. Lloyd, T.E. Warkentin: Evaluation of a pretest clinical score for the diagnosis of HIT . Consecutive patients referred over a 3-yr period with suspected HIT in Adelaide, Australia, were investigated. The DiaMed PaGIA and the platelet aggregation assay (using citrated platelet-rich plasma) were performed prospectively. The 4T's scoring system was applied retrospectively by 2 hematology registrars. Three additional assays—the SRA, in-house EIA-IgG, and EIA-GTI were performed retrospectively by the McMaster Platelet Immunology Laboratory. Presence of HIT antibodies was defined as SRA>50% release plus positive EIA-GTI. Of 115 patients studied, 24 (21%) met the criteria for HIT antibodies. The mean platelet count was 48. Using the 4T's scoring system, almost half the patients (48%) had a low score, 35% had an intermediate score, and 20% had a high score. The incidence of HIT antibodies by test score was 0% (low score), 15% (intermediate score), and 90% (high score). The PaGIA was positive in 17/24 (71% sensitivity) patients with HIT antibodies, with 98% specificity. The platelet aggregation assay was positive in 20/24 (83% sensitivity), with 100% specificity. The EIA-IgG was positive in 24/24 (100% sensitivity), with 94% specificity. Using a Bayesian diagnostic approach, a low 4T's score essentially ruled out HIT, an intermediate score with a positive and negative PaGIA had 85% and 5% probability of HIT, respectively, and a high 4T's score with a positive and negative PaGIA had 99% and 73% probability of HIT. The authors concluded that a low 4T's score ruled out HIT (thus, further lab testing may not be required), that a high 4T's score plus positive PaGIA essentially ruled in HIT, but that in all other situations, confirmatory testing beyond that of the PaGIA was required. Other conclusions were that the EIA-IgG was very sensitive for HIT, and the platelet aggregation test quite specific for HIT.

Rapid assays for HIT: PIFA

J. Francis, A. Drexler, M. Duncan, H. Desai, M. Amaya, T. Robson, T. Meyer, E. Reuyes, A. Amirkhosravi: Performance of the PIFA test for anti-PF4/H antibodies in a prospective study of patients with suspected HIT . This study prospectively compared a Particle ImmunoFiltration Assay (PIFA), from Akers BioSciences, that is FDA-approved based upon >90% agreement with the EIA (from GTI). However, these authors did not find any significant correlation between the GTI and the PIFA among consecutive samples tested using both assays: 32/73 sera testing positive in the EIA were positive in the PIFA, and 68/101 sera testing negative in the EIA were negative in the PIFA, resulting in 57.5% level of agreement (essentially random). Low agreement levels with the EIA from Stago (57.4%) and SRA (59.1%) were also seen. Problems identified with the assay included difficulty distinguishing pos (white) vs neg (light blue) and a very high false-positive rate. The presenter concluded that the assay cannot be recommended.

A. Greinacher, R. Raschke, J.I. Sheppard, T.E. Warkentin: The operating characteristics of the PIFA test for HIT antibodies . Two centres evaluated the PIFA: the first (McMaster Platelet Immunology, Hamilton, Canada) utilized 93 previously-frozen sera that had been evaluated in a prospective study of the 4T's scoring system and 3 assays (EIA-GTI, EIA-IgG, SRA); the SRA (>50% release) was used to define presence of HIT antibodies. Three different volumes of serum (20, 10, and 5 uL) were used so as to mimic the effect of 2- and 4-fold sample dilution. The second study (Greifswald, Germany) tested fresh (never frozen) serum from 199 consecutive patients referred for diagnostic testing for HIT antibodies, as well as 137 fresh (never frozen) samples obtained in ongoing prospective studies of anti-PF4/heparin antibody formation during heparin therapy. For these studies, 20 uL of patient serum was used. A positive heparin-induced platelet activation (HIPA) test with lag time <30 min was used to define the presence of HIT antibodies. All samples were processed within 24 hr of collection (21.4% within 5 hr). Both the McMaster and Greifswald studies showed that the sensitivity/specificity tradeoffs—plotted as receiver-operating characteristic (ROC) curves—were that of a non-informative assay. In contrast, the EIA's assessed in both centers (EIA-GTI and EIA-IgG in Hamilton, EIA-IgG/A/M in Greifswald) showed excellent operating characteristics, with optimal cutoffs of approximately 1.40 and 0.60 for the EIA-GTI and EIA-IgG/A/M in Hamilton and Greifswald, respectively. The presenter concluded that the assay cannot be recommended.

EIA Controversies

T.E. Warkentin, J.I. Sheppard: Relationship of EIA positivity in relation to onset of HIT . The authors addressed an interesting paradox, namely that the EIA is claimed to have a very high sensitivity for clinical HIT (~99%), yet some advocate testing a later blood sample when an initial EIA result is negative. The authors studied 10 patients who developed clinical HIT during prospective clinical studies to determine to what extent EIA's were positive at the beginning of the platelet count fall marking an episode of HIT. The authors found that the median EIA was strongly positive (~2.20 OD units), with the lowest measuring about 0.70 units, at the onset of HIT. These data were interpreted as indicating that the view that an EIA is initially negative at the time of onset of thrombocytopenia due to HIT is likely to be incorrect.

T.E. Warkentin, J.I. Sheppard: What is the appropriate cut-off for a positive EIA? The authors reviewed the EIA OD distributions of heparin-treated patients, both with and without clinically-significant HIT antibodies (as per SRA >50%), to determine what OD cutoff(s) were appropriate for diagnosis of HIT. Considerable overlap in OD values between the two groups was shown. The authors proposed a reporting system in which the physician is given data on the usual distributions/ranges of OD values for patients testing positive and negative in the SRA, e.g., 90% and 95% of patients with HIT have OD values >1.5 and >0.95, respectively; whereas, for heparin-treated patients, 90% and 95% of patients without HIT have OD values <0.70 and <1.10, respectively (these values are shown for illustration purposes, and do not necessarily reflect the experience of a particular lab).

R.H. Aster, B.R. Curtis: Diagnostic utility of high concentrations of heparin in "confirming" a positive EIA—pro . From >5000 patient samples tested, 5.3% showed a positive EIA that was not inhibited by high heparin. The authors compared the 4'Ts score among 21 patients whose sera tested positive in the EIA and was inhibited by high heparin, versus 31 patients with a positive EIA that was not inhibited. The distributions of probability for HIT (as indicated by "low/intermediate/high" 4T's score) were: heparin-inhibited (2/4/15) vs non-heparin-inhibited (22/6/2). These data suggest that a sample that is not inhibited by high heparin is unlikely to be HIT, and thus useful diagnostic information is provided by this maneuver.

J. Francis: Diagnostic utility of high concentrations of heparin in "confirming" a positive EIA—con . The presenter observed that 12/75 samples were not inhibited by high heparin concentrations, of which 7 and 5 were low and intermediate probability for HIT, respectively. However, 2/12 were positive in the SRA. The presenter also stated that the term "confirmatory" testing was misleading, as it might suggest "confirmation" of the diagnosis of HIT rather than for the detection of anti-PF4/polyanion antibodies. The presenter also observed that performing the high heparin assay will either double the test costs (by requiring high heparin with each sample tested) or increase the turnaround time (by requiring a second assay with high heparin in case the initial test is not performed with high heparin). The lack of published data in support of performing the high heparin step was also commented upon.

J. Amiral: Evaluation of a new EIA for detecting heparin-dependent antibodies . A new assay for pathologic heparin-dependent antibodies (Zymutest) was described that utilizes a heparin-coated surface to which patient serum/plasma as well as platelet-leukocyte lysate is added. The expectation is that non-PF4-dependent (e.g., anti-IL-8/heparin antibodies) antibodies as well as anti-PF4/heparin antibodies would be detected. The assay allows for separate detection of IgG, IgA, and IgM antibodies. A high correlation with antibody detection by standard EIA was reported (r 2=0.890).