TAFI antigen and D-dimer levels during normal pregnancy and at delivery


Professor Philippe de Moerloose, Haemostasis Unit, Geneva University Hospital, 1211 Geneva 14, Switzerland. E-mail: philippe.deMoerloose@hcuge.ch


We have investigated whether the levels of thrombin-activatable fibrinolysis inhibitor (TAFI) were correlated with d-dimer levels during pregnancy and at delivery. From the 10th week of pregnancy to delivery, 519 samples from 144 women (mean age 29·3 ± 5, range 19–43) were obtained. We confirm the gradual increase of d-dimer levels, and provide reference intervals for d-dimer measurements throughout normal pregnancy. TAFI levels increased moderately during pregnancy but no inverse correlation with d-dimer levels was observed.

In normal pregnancy there is an enhanced thrombin generation and decreased fibrinolytic activity which may play a role in the higher incidence of venous thromboembolic events observed. Impaired fibrinolysis has been related to progressive increase in plasminogen activator inhibitors (PAI)-1 and PAI-2 levels throughout pregnancy, whereas there is controversy about the changes in tissue plasminogen activator (t-PA) activity (Wright et al, 1988). In addition, t-PA release upon venous occlusion has been shown to be impaired during pregnancy (Ballageer et al, 1987). Paradoxically, d-dimer levels increase gradually, which indicates an increased fibrin deposition consecutive to enhanced thrombin generation, but also a sustained plasmin generation and activity (Cadroy et al, 1993; Lindoff et al, 1993; Nolan et al, 1993; Francalanci et al, 1995; Bellart et al, 1997; Kjellberg et al, 1999; Sattar et al, 1999).

Increased thrombin generation, as observed in normal pregnancy by prothrombin fragment F1+2 and thrombin–antithrombin complex measurements, may further depress fibrinolytic activity by raising the activity of the recently described thrombin-activatable fibrinolysis inhibitor (TAFI) (Bajzar et al, 1995). TAFI levels have been reported to remain stable over months and only to be correlated with age in women (Chetaille et al, 2000). Considering the aforementioned changes in fibrinolytic activity during normal pregnancy, we were interested to know whether they were associated with changes in TAFI antigen levels. In addition, we measured d-dimer levels in a large pregnant population in order to determine whether there was an inverse correlation between TAFI and d-dimer levels.

Patients and methods

The study was performed at the women's clinic of the Centre Hospitalier Universitaire Vaudois at Lausanne. The protocol was approved by the ethics committee. Samples (n = 519) were collected from 144 pregnant women with uncomplicated pregnancies. Data concerning past history of thromboembolic disease, previous pregnancies, medications and weight before pregnancy and at delivery were collected.

The data were divided into six time periods, namely 10–14 weeks (n = 27), 15–19 weeks (n = 123), 20–24 weeks (n = 41), 25–29 weeks (n = 96), 30–34 weeks (n = 98), 35 weeks-delivery (n = 94) and at delivery (n = 40). The mean age of the women was 29·3 ± 5 (range 19–43).

Plasmas were stored at −80°C until assayed. TAFI levels were measured using enzyme-linked immunosorbent assay (ELISA) using reagents from Milan Analytica (La Roche, Switzerland). Briefly, microplates were coated with affinity purified sheep anti-TAFI in carbonate buffer overnight. After blocking with bovine serum albumin-phosphate buffered saline (BSA-PBS) buffer, samples were applied after 1/200 dilution in HEPES–NaCl–BSA–Tween 20 buffer. Bound antigen was quantified with horseradish peroxidase conjugated affinity purified sheep anti-TAFI. A standard curve was run with dilutions of normal pooled plasma. d-dimer levels were measured with VIDAS d-dimer NEW assay (de Moerloose et al, 2001).

Data are expressed as box plots for each time period considered. Comparison of pregnancy periods was performed using the Kruskal–Wallis analysis of variance and the Mann–Whitney test. Correlation between TAFI and d-dimer levels was calculated using linear regression.


The distribution of TAFI values during pregnancy is shown in Fig 1. Variance analysis indicated differences in TAFI levels between the pregnancy periods (P < 10−4). This was caused by the increase observed from the 10–14-week period to the last four pregnancy periods and from the 15–19-week period to the three last pregnancy periods (P < 0·02 at least). For the last four pregnancy periods, variance analysis showed no difference (P = 0·58), the median TAFI levels ranging between 126 and 133%. At delivery, they ranged from 54 to 179% (median value 94%).

Figure 1.

Box plots of TAFI values as a function of pregnancy periods. The box represents 50% of the values, the horizontal bar inside the median, and the lower and the upper bars the 10th and 90th percentiles respectively. *, significantly different from the 10–14-week period (P < 0·02 at least); #, significantly different from the 15–19-week period (P < 0·02 at least).

d-dimer levels increased gradually during the course of pregnancy (Fig 2). In the period 35 week-delivery, all values were > 500 ng/ml. The 5th and 95th percentiles of d-dimer were 139–602 ng/ml for the first trimester of pregnancy, 291–1231 for the second trimester and 489–2217 for the last trimester. At delivery, median d-dimer value was 1581 ng/ml and the distribution ranged between 678 and 5123 ng/ml.

Figure 2.

Box plots of d-dimer values as a function of pregnancy periods. The box represents 50% of the values, the horizontal bar inside the median, and the lower and the upper bars the 10th and 90th percentiles respectively.

There was no correlation between TAFI and d-dimer levels during pregnancy (r2 = 0·018) or at delivery (r2 = 0·153). Within the different gestational age groups, correlation ranged from 0·0002 (35 week-delivery) to 0·05 (30–34-week period).


Our data indicate that the concentration of TAFI antigen does not appear to increase to the same extent as that of plasminogen activator inhibitors, factor VIII or fibrinogen levels during pregnancy. In contrast to our results, Chetaille et al (2000) did not find any difference in TAFI mean levels between a small group of 12 women in the third trimester of pregnancy and age-matched non-pregnant women. We have no explanation for this discrepancy, but it is possible that the number of women was not sufficient to reveal the difference we observed. However, their interesting observation that TAFI antigen was related to age only in women suggests hormonal involvement, and could explain in part the modest increase we observed during pregnancy.

The gradual increase in d-dimer levels during pregnancy has been reported previously by other groups using various assays. The numerical results are of course not comparable because of the differences in assays. Using the new Vitek Imuno Diagnostic assay system (VIDAS), VIDAS d-dimer NEW (de Moerloose et al, 2001), we propose reference ranges according to the three trimesters of pregnancy.

On an individual level, we found no inverse correlation between TAFI antigen and d-dimer values. Therefore, it does not seem that increased fibrinolytic response resulting in the high d-dimer levels observed in pregnancy is mediated by TAFI. However, we cannot exclude that such a relationship may occur with TAFI activity in vivo.