Abstract
Background: The anti-VEGF drug, bevacizumab (Bev), has been associated with arterial thromboembolism in colorectal cancer patients. However, the mechanism of this remains poorly understood, and preclinical testing in mice failed to predict thrombosis. Prevailing opinion on the molecular mechanism behind Bev-associated bleeding and thrombosis is that tissue factor driven coagulation, secondary to vascular endothelial cell dysfunction, may cause thrombosis due to VEGF suppression by Bev. Bev forms immune complexes (IC) with VEGF (vascular endothelial growth factor), a heparin-binding protein. In our previous in vitro studies we showed that, in the presence of heparin, Bev+VEGF immune complexes activate platelets via the IgG receptor FcγRIIa —a mechanism similar to that observed with antibodies from patients with heparin-induced thrombocytopenia (HIT).
Objectives: First, we investigated whether Bev-associated thrombosis might be replicated in mice. Because mouse platelets do not carry FcγRIIa, we used mice transgenic for this human IgG receptor (hFcR mice) in order to enable the signaling pathway identified above. Second, using human platelets in vitro, we studied the functional roles of heparin and platelet surface localization of IC in Bev-induced FcγRIIa activation.
Methods: Bev+VEGF IC were preformed using VEGF165 or VEGF121 (similar to VEGF165 but lacking the heparin-binding domain). Platelet dense granule release and aggregation were measured by the serotonin release assay (SRA) and Chrono- Log aggregometers, respectively. Platelet surface localization was assessed by flow cytometry (50,000 events/test condition) and fluorescence microscopy using Alexa488- labeled Bev (Bev488). For in vivo studies, Bev+VEGF+Heparin IC (60–500 nM) or control reagents were injected intravenously into wild-type (WT) or hFcR mice. Platelet counts were measured 10–60 minutes following IC injection after obtaining blood (0.45 ml) by cardiac puncture. Immediately afterward, lungs were processed for hematoxylin and eosin staining and analyzed microscopically for evidence of thrombosis.
Results: IC consisting of Bev+VEGF165+Heparin (0.2U/ml) caused thrombotic thrombocytopenia in hFcR but not WT mice, showing a requirement for FcγRIIa. Injection of Bev+VEGF121+Heparin (0.2U/ml) into hFcR mice did not cause thrombocytopenia, suggesting a requirement for the VEGF165 heparin binding domain. Bev+VEGF165 was without effect in the absence of heparin or in the presence of excess (200 U/ml) heparin demonstrating that a limited range of heparin concentrations enable Bev-induced thrombocytopenia and thrombosis. This mechanism is similar to that observed in HIT and our in vivo results were consistent with SRA and aggregation in vitro studies. By flow cytometry, maximal Fab-dependent Bev488 platelet surface binding occured only with VEGF165+0.2U/ml heparin. Saturating IV.3 (anti-FcγRIIa antibody) concentrations, present in all samples, excluded Bev-Fc binding to FcγRIIa. Furthermore, binding of Bev488+VEGF121+0.2 U/ml heparin was not detected, suggesting the VEGF heparin binding domain is required for heparin-enhanced surface binding.
Conclusions: In the presence of heparin, Bev can induce platelet aggregation, degranulation and thrombosis through complex formation with VEGF and activation of FcγRIIa receptor. This mechanism may be relevant to the thromboembolic complications observed in patients receiving Bev therapy.
Binding of lipoprotein lipase to α2-macroglobulin