Tranexamic acid dose-response relationship for antifibrinolysis in postpartum haemorrhage during Caesarean delivery: TRACES, a double-blind, placebo-controlled, multicentre, dose-ranging biomarker study

Abstract

Background: The optimal dose of tranexamic acid to inhibit hyperfibrinolysis in postpartum haemorrhage is unclear. Tranexamic Acid to Reduce Blood Loss in Hemorrhagic Cesarean Delivery (TRACES) was a double-blind, placebo-controlled, randomised, multicentre dose-ranging study to determine the dose-effect relationship for two regimens of intravenous tranexamic acid vs placebo.

Methods: Women experiencing postpartum haemorrhage during Caesarean delivery were randomised to receive placebo (n=60), tranexamic acid 0.5 g (n=57), or tranexamic acid 1 g i.v. (n=58). Biomarkers of fibrinolytic activation were assayed at five time points, with inhibition of hyperfibrinolysis defined as reductions in the increase over baseline in D-dimer and plasmin-antiplasmin levels and in the plasmin peak time.

Results: In the placebo group, hyperfibrinolysis was evidenced by a mean increase over baseline [95% confidence interval] of 93% [68-118] for D-dimer level at 120 min and 56% [25-87] for the plasmin-antiplasmin level at 30 min. A dose of tranexamic acid 1 g was associated with smaller increases over baseline (D-dimers: 38% [13-63] [P=0.003 vs placebo]; plasmin-antiplasmin: -2% [-32 to 28] [P=0.009 vs placebo]). A dose of tranexamic acid 0.5 g was less potent, with non-significant reductions (D-dimers: 58% [32-84] [P=0.06 vs placebo]; plasmin-antiplasmin: 13% [18-43] [P=0.051]). Although both tranexamic acid doses reduced the plasmin peak, reduction in plasmin peak time was significant only for the 1 g dose of tranexamic acid.

Conclusions: Fibrinolytic activation was significantly inhibited by a dose of intravenous tranexamic acid 1 g but not 0.5 g. Pharmacokinetic-pharmacodynamic modelling of these data might identify the best pharmacodynamic monitoring criteria and the optimal tranexamic acid dosing regimen for treatment of postpartum haemorrhage.

Clinical trial registration: NCT02797119.

Keywords: D-dimer; antifibrinolytic drug; fibrinogen; fibrinolysis; plasmin; plasmin–antiplasmin complex; postpartum haemorrhage; thrombin; tranexamic acid.

Fig 1

Study flow chart. PK, pharmacokinetic; PPH, postpartum haemorrhage; TXA, tranexamic acid.

Fig 2

Biomarkers of fibrinolysis activation in the placebo, low dose (TXA 0.5 g), and standard dose (TXA 1 g) groups over baseline. (a) Box plots showing the median [IQR] D-dimer level increase (ng ml⁻¹) over baseline at 2 h after infusion in the placebo, low dose (TXA 0.5 g), and standard dose (TXA 1 g) groups. A linear mixed model was used to compare the three groups. (b–f) Laboratory data for the placebo group (black diamonds), the TXA 0.5 g group (dark grey squares), and the TXA 1 g group (light grey triangles). Significant differences between T0 and each time point are indicated by an asterisk (for TXA 0.5 g vs placebo) or a star (TXA 1 g vs placebo). (b) The D-dimer level increased less from T0 to T120 in the standard dose group than in the placebo group. The figure shows the percentage increase between baseline and T30, T120, and T360 in the placebo (TXA 0 g), low dose (TXA 0.5 g), and standard dose (TXA 1 g) groups. The groups were compared by applying a mixed linear model of covariance. (c) Plasma PAP complex levels increased less from T0 to T30 and T60 in the standard dose group than in the placebo group. The figure shows the percentage increase between baseline and T30, T60, T120, and T360 in the placebo, low dose (TXA 0.5 g), and standard dose (TXA 1 g) groups. The groups were compared by a mixed linear model of covariance. (d) The PG peak from T0 to T30, T60, T120, and T360 was significantly lower in the low dose group (TXA 0.5 g) and the standard dose (TXA 1 g) group than in the placebo group. The groups were compared by a mixed linear model of covariance. (e) The PG peak time decreased more in the standard (TXA 1 g) group than in the placebo group. The figure shows the percentage decrease between baseline and T30, T120, and T360 in the placebo, low dose (TXA 0.5 g), and standard dose (TXA 1 g) groups. The groups were compared by a mixed linear model of covariance. (f) The time interval between the thrombin generation peak and the PG peak decreased significantly more in the standard dose (TXA 1 g) group than in the placebo group. The figure shows the percentage decrease from baseline (T0) to each time point (T30, T60, T120, and T360). The groups were compared by a mixed linear model of covariance. PAP, plasmin–antiplasmin; PG, plasmin generation; TXA, tranexamic acid.