Pathogenic Antibodies in Heparin-Induced Thrombocytopenia Specifically Target an Immunodominant Region on Platelet Factor 4

Introduction: Heparin-induced thrombocytopenia (HIT) is an adverse drug reaction that occurs when heparin binds to platelet factor 4 (PF4) forming immunogenic complexes. Anti-PF4/heparin IgG antibodies bind PF4/heparin complexes, leading to cross-linking of FcγRIIa receptors on platelets and FcγRI on monocytes, resulting in platelet activation, thrombocytopenia, and thrombosis. The current diagnostic challenge is that the majority of patients suspected of HIT yield false-positive results in immunoassays, since up to 50% of patients will make anti-PF4/heparin antibodies but will not develop HIT. The antibody response in HIT patients is polyclonal, making it difficult to identify a common pathogenic epitope. The disparities between anti-PF4/heparin antibodies that activate platelets (pathogenic HIT antibodies) and those that do not (non-pathogenic anti-PF4/heparin antibodies) present a significant challenge in diagnosing HIT. The objective of this study was to map and characterize the critical immunodominant region on PF4 for the binding of pathogenic antibodies in confirmed HIT patients. Methods: We used sera with anti-PF4/heparin antibodies from patients with confirmed HIT (n=10). Post-cardiopulmonary bypass patients (CPB; n=10) and healthy individuals (n=10) were used as controls. Confirmed HIT patients met clinical criteria (4Ts ≥ 4) and tested positive in both the anti-PF4 IgG/A/M immunoassay (OD > 0.4; range 2.33 - 3.90) and in the serotonin release assay (SRA release > 20%; range 88-100%). CPB patients all received heparin but did not develop HIT, tested positive in the anti-PF4 IgG/A/M immunoassay (OD > 0.4; range 0.42 - 2.73), and tested negative in the SRA (SRA release < 20%; range 0-18%). We previously used alanine scanning mutagenesis and identified 30 amino acids that were on the surface of PF4 and were likely a part of the region essential for the binding of pathogenic HIT antibodies. From those results, we used the panel of 30 PF4 mutants and tested their ability to bind to HIT, CPB, and healthy control sera. Loss of binding to PF4 mutants was applied to in-silico structural analysis to determine binding regions specific for pathogenic and non-pathogenic antibodies. We also determined binding affinities of pathogenic and non-pathogenic anti-PF4/heparin antibodies using biolayer interferometry (BLI). Results: When 30 PF4 mutants were used to test the effect of the amino acid changes on the binding of HIT and CPB patient sera, an average of 8 different PF4 mutants resulted in more than 35% loss of binding to confirmed HIT sera when compared to wild-type PF4. None of the 30 PF4 mutants resulted in more than 35% loss of binding to CPB sera. Structural analysis demonstrated that the amino acids of PF4 that significantly affected the binding of HIT sera, but not CPB sera, were clustered to a specific region on PF4, similar to the region of KKO, but with varying epitopes. Using BLI, anti-PF4/heparin antibodies of confirmed HIT patients had a stronger binding response to PF4 and PF4/heparin than that of CPB patients and healthy controls. Overall, we were able to show a significant difference between confirmed HIT sera and the false-positive antibodies of CPB patients that did not develop HIT (P < 0.01). Conclusion: This work shows that among the polyclonal response in HIT, pathogenic HIT antibodies must bind to the critical immunodominant region on PF4 with high affinity. This ensures the proper spatial configuration of the antibodies for Fc-receptor cross-linking, platelet activation, and subsequently HIT. This study has implications for the development of novel epitope-targeted diagnostic and therapeutic approaches for HIT. Disclosures Arnold: Novartis: Honoraria, Research Funding; Rigel: Consultancy, Research Funding; Principia: Consultancy; Bristol-Myers Squibb: Research Funding.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.