Study setting

King's College Hospital (Denmark Hill site) is a 900-bed tertiary care hospital, with around 5,400 babies delivered each year, including 1,500 caesarean sections. Nulliparous and multiparous women prescribed antenatal enoxaparin were eligible for entry to the study. Women who did not wish to participate or were unlikely to comply with directions, those with impaired renal function and those <18 years of age were excluded. Following informed written consent, subjects attended the thrombophilia clinic monthly during their pregnancy for ongoing monitoring, as well as drawing blood samples for determination of renal and liver function, full blood count, and D-dimer concentrations. Once in each trimester, women had additional blood drawn for the assessment of thrombin generation. Samples for assessment of D-dimer and thrombin generation samples were also drawn at the time of delivery and at ≥8 weeks post-partum; the latter sample providing a baseline thrombin generation and D-dimer concentration for subjects, acting as their own control.

Sample collection and plasma preparation: D-dimers

Samples were obtained through antecubital venepuncture using a butterfly Terumo® Surflo® winged infusion set (21G × ¾”, 0.8 × 19 mm UTW tubing, REF: SV-21BL, Terumo Europe N.V., Belgium). For determination of D-dimer, 2.7 mL blood sample was collected in 0.109 M (3.2% trisodium citrate) Becton-Dickinson Vacutainer®, centrifuged at 2500 g for 7 minutes on a Rotina 38 R centrifuge (Hettich Zentrifugen), within 1 hour of collection. The resultant plasma was analysed using the STA-Liatest® D-DI kit (immune-turbidimetric assay of D-dimer (Diagnostica Stago, France)). Analysis was conducted using the STA-R evolution (Diagnostica Stago) analyser.

Sample collection and plasma preparation: thrombin generation

Samples were obtained through antecubital venepucture using the aforementioned infusion set, without the use of a vacuum and with minimal suction into a 20 ml syringe, with the first 5 ml of blood drawn discarded. The blood was then transferred to 0.109 M (3.2% trisodium citrate) Becton-Dickinson Vacutainer®. Platelet poor plasma (PPP) was obtained by centrifugation at 4750 g for 10 minutes at room temperature. The supernatant was decanted into a polypropylene tube and centrifugation repeated. The plasma was decanted into a plastic tube, capped and immediately frozen. This procedure was performed within 60 minutes of venepuncture. The samples were stored at -35°C and analysed using calibrated automated thrombinography (CAT), as developed by Hemker [21].

Calibrated automated thrombinography

Thrombin generation was measured with the thrombinoscope assay (Thrombinoscope BV, The Netherlands), according to the manufacturer's instructions, as has previously been described by our group [6],[7],[22]. Concentrations of tissue factor and phospholipid used in the experiments were 5 pmol/L and 4 μmol/L, respectively.

Five thrombin generation parameters were computed: lag time, time to peak, peak height, start-tail and endogenous thrombin potential (ETP).

Intra- and inter-assay variability was assessed using plasma from a healthy control (male subject, aged 36-37 years). This volunteer was bled every eight weeks from the 16^th April 2010 to the 24^th January 2012. Eight plasma samples were obtained from the volunteer on each occasion and one sample was run in each thrombin generation experiment, from the 20^th April 2010 (exp 17) to the 21^st February 2012 (exp 125). For the primary thrombin generation parameters, lag-time, ETP, peak height, time to peak and start tail, the following coefficients of variation (%) were obtained: 4.5, 1.6, 3.9, 4.0, 2.1 for intra-assay variation and 7.6, 11.8, 13.7, 7.5 and 5.1 for inter-assay variation.

Statistical analysis

Data was grouped into five time points: first, second and third trimester, within a week of delivery and ≥8 weeks post-partum.

Continuous variables are presented as means and standard deviations or median and inter-quartile ranges for normally and non-normally distributed variables respectively. For thrombin generation, differences in the means of variables were compared between the five time points, using the repeated measures ANOVA with a Greenhouse-Geisser correction. Post-hoc analysis was conducted using Bonferroni correction. For D-dimer concentrations, the Friedmans mean rank test was used to assess if statistically significant differences existed between the different time-points, with the Wilcoxon pair test used to assess significance between specific pairs of study time points.

As the Caucasian and African-Caribbean ethnic groups comprised the majority of the subjects recruited, differences in the thrombin generation and D-dimers results at each of the time points were compared for women injecting prophylactic doses of enoxaparin, using the independent t-test for thrombin generation and the Mann–Whitney U independent rank-sum test for D-dimer concentrations.

Further, to assess the impact of enoxaparin dose on the thrombin generation parameters and D-dimer concentrations, women were grouped into those injecting treatment doses and those injecting prophylactic doses. Descriptive statistics was used to report these results.

Statistical significance was considered at the 0.05 α level. Analysis was performed using SPSS version 18.0 (SPSS Inc., Chicago, Ilinois, USA).

The study received ethical approval from the Isle of White, Portsmouth and South East Hampshire Ethics Committee (REC reference: 09/H0501/57).

Study participants and pregnancy outcomes

The range of enoxaparin doses prescribed for the subjects recruited are presented in Additional file 1: Table S1. Furthermore, details on the 19 women prescribed treatment doses of enoxaparin (9 women managed for antenatal VTE and the 10 women switched from long-term warfarin during pregnancy) are presented in Additional file 2: Table S2 and Additional file 3: TableS3 respectively.

Pregnancy outcomes

Thrombin generation

The 123 subjects provided 498 individual thrombin generation profiles; 414 (83%) of the profiles were drawn from women prescribed prophylactic doses of enoxaparin with the remaining 84 profiles from women prescribed treatment doses.

The samples drawn at ≥8 weeks post-delivery provide an insight into the cohort of women's baseline thrombin generation was prior to their pregnancy by which time-point women had discontinued their anticoagulation therapy. Using the ≥8 week post-partum values as a reference point; a rise in ETP is seen over the course of pregnancy, despite an enoxaparin prescription.

When the thrombin generation results are separated according to women injecting treatment and prophylactic doses of enoxaparin, a difference in changes are observed between the group of women injecting treatment doses (84 profiles) and those injecting prophylactic doses (414 profiles) (Figure 1).

The results demonstrate a dose dependent enoxaparin effect. The ETP rise over pregnancy for women on treatment doses of enoxaparin is more attenuated, relative to those women prescribed prophylactic doses, as observed by differences in peak height. An influence of enoxaparin on the lag time, time to peak, and start tail is also observed. Examining the results from the women prescribed prophylactic doses, where all anticoagulant therapy had been discontinued by eight weeks postpartum, one can see that ETP is already higher by the first trimester, with a further rise seen during the second trimester, after which it settles through to delivery, where it peaks.

Eight women prescribed treatment doses of enoxaparin, had thrombin generation measurements at all study time-points. No significant difference in results for the individual thrombin generation parameters were found, across these five time-points. This was not the case, for women prescribed prophylactic doses of enoxaparin. In this group, 22 women had thrombin generation assayed at all study time-points. The following results were found for each thrombin generation parameter;

Lag-time: significant differences were found between the time points (F(2.684, 56.261) = 3.732, p < 0.05). Post-hoc analysis demonstrated significant differences between the 8 weeks postpartum samples and the samples drawn in the first, second and third trimester, where a prolongation of the lag-time was observed.

ETP: the mean ETP concentrations did statistically differ between the five time points (F(2.312, 48.561) = 9.072, p < 0.05). Post-hoc analysis revealed that specific differences were between the 8 weeks post-partum time point and the second and third trimesters and at the time of delivery, mirroring the result seen for the cohort of women as a whole.

Peak height: statistically significant differences between the five time points (F(2.375, 49.877) = 5.209, p = <0.05) were found. A post-hoc analysis however, was unable to identify at which time points the differences were.

Time to peak: significant differences were found (F(2.715, 57.021) = 6.181, p = <0.05). A post-hoc analysis revealed these differences between the ≥8 weeks post-partum time point and the first, second and third trimesters. No significant differences were found between the samples drawn at delivery and the ≥8 week post-partum time point.

Start-tail: significant differences were found (F(2.252, 47.291) = 11.953, p = <0.05). A post-hoc analysis revealed that the specific differences were between the ≥8 weeks post-partum time point and the other four time points.

Variation in thrombin generation by ethnicity

There are significant differences between the two ethnic groups for the mean ETP, and peak parameters with the African-Caribbean group exhibiting higher values for these relative to their Caucasian counterparts, during the first and second trimesters.

D-dimers concentrations

Although considerable inter-patient variability is exhibited, a clear pattern was observed in line with gestational age.

Significant differences were found when each study time-point was compared, except when the first trimester was compared with the ≥8 weeks post-partum time point.

Women injecting prophylactic versus treatment doses of enoxaparin: D-dimer concentrations

Twenty-six women injecting prophylactic doses of enoxaparin had serial D-dimer concentrations measured consecutively from the first trimester through to ≥8 weeks post-partum. Statistically significant differences between the five different time points were found (n = 26, df = 4, p = <0.05). When the different study time points were compared with one another, significant differences were found between each study pair, except when the first trimester was compared with the ≥8 weeks post-partum time point.

Ten women prescribed treatment doses of enoxaparin had serial D-dimer concentrations consecutively measured from the first trimester through to ≥8 weeks of delivery. Statistically significant differences were found between the five study time points (n = 10, df = 4, p = <0.05); specifically between the second and third trimesters and within a week of delivery and ≥8 weeks postpartum.

D-dimer concentrations

D-dimer concentrations reflect active coagulation and fibrinolysis. Our large set of D-dimer results agree with previously published studies, that D-dimer concentrations increase exponentially during pregnancy, peaking at the time of delivery. To date, little work has been published on how the D-dimer concentrations might be altered in pregnant women who are concurrently injecting LMWH. Hoke and colleagues have previously explored this [19] and assessed 61 women prescribed LMWH thromboprophylaxis throughout pregnancy due to a history of VTE, hereditary thrombophilia and/or previous pregnancy related complications. D-dimer concentrations were measured during their pregnancy, having been prescribed dalteparin or enoxaparin at a dose range of 4000 to 7500 anti-factor Xa IU daily; commenced following a positive pregnancy test. The D-dimer concentrations were then compared with a control group comprising 113 healthy pregnant women not requiring thromboprophylaxis. The author's found that D-dimer concentrations in both subjects and controls increased significantly from the first to the second trimester, as well as from the second to the third trimester. The authors also describe the D-dimer concentrations being significantly higher among subjects compared with controls. Despite being prescribed prophylactic LMWH, substantial activation of coagulation was seen during pregnancy in this study, i.e. the prophylactic doses of LMWH did not dampen down the increase in D-dimers one might expect to observe. Furthermore, Abou-Nassar and colleagues [20], as part of the Thrombophilia in Pregnancy Prophylaxis Study (TIPPS), patients at high risk of pregnancy complications with confirmed thrombophilia were randomised to receive either dalteparin 5000 units daily until 20 weeks gestation, then 5000 units twice a day until 37 weeks gestation or onset of labour, or no treatment. Samples for assessment of TAT, prothrombin fragment 1 + 2, and D-dimer concentrations were drawn in each trimester. The authors report that the levels of TAT, prothrombin fragment 1 + 2 and D-dimer concentrations increased significantly in the dalteparin and control group, again illustrating that the prophylactic LMWH does not impact on the D-dimer concentrations. The results from our study confer with the findings from both these studies.

The increase in D-dimer concentrations during pregnancy, appear relatively predictable and the utility of using D-dimer concentrations for the diagnosis of antenatal VTE should be explored further, with a view of developing a normal D-dimer pregnancy reference range, as has been suggested previously [31]-[33]. The UK National Institute for Health Research recently made a Health Technology Assessment call for researchers to define the optimal algorithm to diagnose antenatal pulmonary embolism. The results from this commissioned research [34] will be eagerly awaited in years to come and perhaps D-dimer concentrations will form part of the final diagnostic algorithm recommended.

Differences in TG and D-dimers between the African-Caribbean and Caucasian populations

When we compared D-dimer concentrations in women injecting prophylactic doses of enoxaparin, between the African-Caribbean and Caucasian populations, no differences were found. This confirms findings previously reported by our group from a non-pregnant population [7]. Differences in thrombin generation were found between the African-Caribbean women compared to the Caucasian women. These differences were particularly prominent during the first trimester. The last UK confidential enquiry into maternal deaths report [35] reported that 3 of the 5 deaths during the first trimester were of women from an African-Caribbean origin. It could well be, that significantly higher thrombin generation in these women, pre-dispose them to VTE during early pregnancy more so, than the Caucasian population. In part, the increased thrombin generation observed may be related to higher factor VIII levels in the African-Caribbean population [36], and further work is required comparing factor VIII concentrations in these two groups of pregnant women would be useful and might help explain the differences observed.

Limitations

The results of our work should be considered in the context of their limitations. Not all women had sequential thrombin generation and d-dimer concentrations measured from the first trimester to 8 weeks post-partum. This was largely due to the women booking after the first trimester had passed, and then being referred for haematology input. Furthermore, the sampling of thrombin generation was done at times when women were attending for outpatient appointments; ideally sampling would have been completed in defined trimester `windows', to minimise any impact this may have. Finally, the thrombin generation and ethnicity results were not controlled for weight, which is known to impact significantly on thrombin generation [10]. Future studies specifically assessing thrombin generation differences in different ethnic groups, should be mindful to control for this.

Future work

There is clear need for a study which measures thrombin generation longitudally during uncomplicated pregnancy. Once the pattern of normal thrombin generation is known, studies can then begin to evaluate if global coagulation assays can detect other pregnancy co-morbidities, e.g. pre-eclampsia or intrauterine growth restriction. Given the differences in TG between the African and Caucasian population, a study formally evaluating TG differences between these two groups would be of value, particularly since the African-Caribbean women represent a significant proportion of women in the last confidential enquires report.