In their recent report, Bougie et al1 demonstrate the presence of drug-dependent antibodies to platelet glycoprotein IIb/IIIa (GPIIb/IIIa) in patients who developed thrombocytopenia following treatment with the GPIIb/IIIa inhibitors tirofiban and eptifibatide. In an effort to locate the epitope(s) recognized by those antibodies, the authors performed blocking experiments with abciximab. The latter is the Fab fragment of a chimeric IgG construct (human IgG with murine antigen-binding region) derived from the murine monoclonal antibody 7E3, which binds with high affinity to an epitope close to the fibrinogen binding site on GPIIb/IIIa. The degree of blocking of antibody binding was assessed by measuring the effect of incubation with abciximab on the amount of test serum immunoglobulin (Ig) bindable to control platelets pretreated with fiban (tirofiban or eptifibatide). In their description of the blocking experiments, the investigators do not provide data on antibody binding to abciximab-treated platelets not previously exposed to fiban, and therefore, in this setting, recruitment of Ig onto the platelet surface could have been caused by the presence of either antifiban or antiabciximab antibodies. Although elsewhere in the paper it is stated that “[n]one of the antibodies reacted with abciximab-treated platelets,” 1(p2073)no data or methods descriptions are given to support this. As a matter of fact, this statement appears to be at variance with previously published work showing that a statistically significant elevation of platelet-bound IgG occurred in patients with coronary heart disease (CHD) receiving infusions of Fab fragments of chimeric 7E3, and that 14 of 21 control sera from CHD patients contained varying amounts of IgG bindable in vitro to 7E3-Fab–coated platelets.2 These phenomena might be related to the existence of “naturally occurring” antibodies to the Fab fragments derived by enzymatic cleavage of the IgG molecule, which are commonly found in sera from healthy individuals.3,4 More evidence will be therefore required before the great differences in the extent of blocking of antibody binding by abciximab observed by Bougie et al are ascribed to the existence of multiple antifiban epitopes, as it is possible that they were merely a reflection of differences in the titers of “naturally occurring” anti-Fab (ie, antiabciximab) in the test sera.
Immune thrombocytopenia associated with fiban therapy
Dr Christopoulos refers to our finding1-1 that eptifibitide and tirofiban-dependent antibodies differ in the extent to which their binding is blocked by abciximab, providing evidence that they recognize different epitopes on GPIIb/IIIa. He points out that naturally occurring antibodies specific for the papain cleavage site on Fab fragments such as abciximab are commonly found in patients and normal sera and correctly notes that, if such antibodies were present in any of the patient samples used in our study, it might have been impossible to know whether abciximab blocked drug-dependent antibody binding or not.
The naturally occurring antibodies to which Dr Christopoulos refers are usually weak1-2 and would be undectable at the dilutions used in our studies (generally 1:20 or more). But we specifically stated that “[n]one of the antibodies reacted with abciximab-treated platelets.” 1(p2073) Probably, we should have reiterated this in paragraph 5 of “Discussion” where the abciximab blocking studies are specifically described.
We thank Dr Christopoulos for the opportunity to clarify this issue.