Integrated Proteomic and Computational Approach Reveals Determinants of Bleeding Risk in Factor XI Deficiency.

Reitsma, S.; Sindi S.; Kissau D.; Burthem J.; Bolton-Maggs P.; Gidley G.; Hansen K.; Leiderman K.; Wolberg A.

Integrated Proteomic and Computational Approach Reveals Determinants of Bleeding Risk in Factor XI Deficiency.

All Authors

Reitsma, S.

Sindi S.

Kissau D.

Burthem J.

Bolton-Maggs P.

Gidley G.

Hansen K.

Leiderman K.

Wolberg A.

LTHT Author

Gidley, Gillian

LTHT Department

Haematology

Publication Date

2023

Item Type

Conference Abstract

Abstract

Background: Individuals with factor (F)XI deficiency have variable bleeding risk that is not predicted by FXI activity, posing clinical challenges. Plasma-based clot formation and fibrinolysis assays can differentiate bleeding risk in FXI-deficient individuals, suggesting differences in the plasma proteome determines the phenotype. Aim(s): Determine if analysis of the plasma coagulome in a computational simulation of thrombin generation can predict bleeding phenotype. Method(s): Citrated plasmas from healthy individuals (N = 49) and individuals with FXI deficiency (N = 57; non-bleeders [N = 34] and bleeders [N = 23] defined by bleeding after tonsillectomy and/or dental extraction before diagnosis of FXI deficiency) were analyzed by mass spectrometry (FXI, FX, FIX, FVII, FII, antithrombin) or ELISA (FVIII, FV, TFPI). Values were converted to % using the control group median. "Coagulome avatars" of each individual were analyzed in an experimentally-validated computational model of thrombin generation initiated with 1-15 fmol/cm2 tissue factor (TF). Result(s): Median FVIII was significantly higher in non-bleeders than bleeders, but logistic regression of FXI and FVIII levels, alone or together, did not separate non-bleeders from bleeders (area under the receiver operator characteristic curve [AUROC] 0.582, 0.780, and ~0.781, respectively). Principal component analysis of the measured plasma coagulome separated controls from FXI-deficient individuals, but did not separate non-bleeders from bleeders. Thrombin simulations of the coagulome triggered with high TF (15 fmol/cm2) did not differentiate non-bleeders and bleeders; however, the lag time, time to inflection, and thrombin peak from simulations with low TF (3 fmol/cm2) distinguished non-bleeders and bleeders. Non-bleeders produced more FVIIIa and prothrombinase than bleeders. Thrombin simulations using only patient levels of FXI and FVIII and normal levels of all other factors produced parameters that when integrated with FXI and FVIII yielded additional power to classify bleeders and non-bleeders (AUROC 0.866; Figure 1). Conclusion(s): Integration of patient FXI and FVIII levels and a computational model of plasma coagulation separates FXI-deficient non-bleeder and bleeder phenotypes. [Figure presented]