Low tissue factor pathway inhibitor (TFPI) together with low antithrombin allows sufficient thrombin generation in neonates

Authors


Gerhard Cvirn, Ludwig Boltzmann Institut für pädiatrische Hämostaseologie, University Klinik für Kinder- und Jugendheilkunde, Gerinnungslabor, Auenbruggerplatz 30, A-8036 Graz, Austria. Tel.: +43 3163854031; fax: +43 3163854024; e-mail: gerhard.cvirn@klinikum-graz.at

Abstract

Summary.  Neonates have an excellent hemostasis despite, in comparison to adults, markedly decreased and delayed ability to generate thrombin. Only 30–50% of peak adult thrombin activity can be produced in neonatal plasma by means of conventional in vitro assays. We show that in contrast to conventional activation, activation with small amounts of lipidated tissue factor (<10 pmol L−1) results in shorter clotting times and faster activated factor X- and thrombin generation in neonates compared with adults due to the concomitant action of low tissue factor pathway inhibitor and antithrombin. The concentrations of both inhibitors in cord plasma are approximately 50% of the respective adult values. After addition of 2.5 pmol L−1 lipidated tissue factor, cord plasma clotted ∼90 s earlier than adult plasma and the amount of free thrombin generated was ∼90% of adult value (291 ± 14 vs. 329 ± 16 nmol L−1 min−1, P < 0.01). Our results might help to explain the clinically observed excellent hemostasis of neonates despite low levels of procoagulant factors.

The hemostatic system of neonates is different to that of children and adults. Neonates have lower levels of vitamin K-dependent coagulation factors and of contact factors. This is reflected in prolonged prothrombin time (PT) and activated partial thromboplastin time (APTT [1]). A marked impaired ability of neonatal plasma from the healthy neonate to generate thrombin, the pivotal enzyme in hemostasis, compared with that of adult plasma has been reported [2]. From these laboratory findings one would expect a bleeding tendency in neonates, but neonates have an excellent hemostasis. Healthy neonates show no easy bruising, no increased bleeding during surgery, and good wound healing. This discrepancy between in vitro and clinical findings so far has not been understood completely.

Physiological low levels of anticoagulant proteins in neonatal plasma might compensate for low levels of coagulation factors and allow sufficient thrombin generation. Neonatal plasma levels of antithrombin (AT) are reduced to approximately 60% of adult values [1]. It has been shown that these physiological low AT levels result in an impaired ability of neonatal plasma to inhibit thrombin compared with adult plasma [3,4]. Free thrombin generation dose-dependently increases when the AT content is successively decreased [5]. However, variations of the AT content do not significantly alter clotting times. Additionally, neonatal plasma levels of the inhibitor protein C (PC) are approximately 35% of adult values [1]. We have shown recently that these low PC levels in neonates result in decreased thrombin inhibition [6]. However, despite of low AT and PC levels, only 30–50% of peak adult thrombin activity can be produced in neonatal plasma activated via standard assay systems such as APTT or PT [7]. Thus, the physiological low AT and PC levels alone do not explain the clinical observed excellent hemostasis in neonates.

Several issues might be important in this context: for a number of reasons the in vitro clotting assays, where relatively high amounts of initiator are applied, give a very simplified picture of the relative importance of various components of the coagulation process. It has been shown that coagulation activation via the extrinsic pathway with low amounts of tissue factor (TF) is probably more physiological than the plasma activation commonly used in standard assay systems [8]. Additionally, in contrast to the high amounts of initiator conventionally applied and the marked dilution of the plasma samples in routine determinations of clotting parameters, low concentrations of lipidated TF as trigger for coagulation should be applied. These low concentrations have been shown to be suitable for sensitive detection of the effects of different levels of pro- and anticoagulants on thrombin generation [9]. Full-length tissue factor pathway inhibitor (TFPI) is the major inhibitor of TF-initiated thrombin formation [10]. TFPI is a reversible, active site-directed inhibitor of factor (F)Xa, which regulates coagulation by inhibiting FVIIa/TF in a FXa-dependent manner. The TFPI/FXa complex binds to the FVIIa/TF complex, resulting in the formation of a TF/FVIIa/TFPI/FXa-quaternary complex [11]. So far data about the effects of TFPI on thrombin generation in neonates have not been reported.

We therefore investigated whether low levels of TFPI combined with low AT levels facilitate thrombin generation in cord plasma activated with small amounts of lipidated TF. The effects of different levels of TFPI and AT on FXa and thrombin generation, prothrombin activation, and clotting time were determined.

Materials and methods

Reagents

Buffer A contained 0.05 mol L−1 Tris-HCl at pH 7.4, 0.1 mol L−1 NaCl and 0.5 mol L−1 bovine serum albumin. Buffer B was analogous to buffer A but contained in addition 20 mmol L−1 EDTA. Buffer C contained 50 mmol L−1 Tris-HCl and 7.5 mmol L−1 EDTA at pH 8.4. Full-length lipidated recombinant human tissue factor (a stock solution was prepared by dissolving 50 ng of the lyophilized lipoprotein in 4.5 mL buffer A, aliquots of 500 µL were stored at −70 °C, addition of 10 µL to 300 µL plasma resulted in a final lipidated TF concentration of 10 pmol L−1), full-length recombinant human TFPI activity standard (5 µg of the glycoprotein were dissolved in 500 µL buffer A, after subsequent 1 : 10 dilution in buffer A 50-µL aliquots were stored at −70 °C), rabbit anti-TFPI (250 µg were dissolved in 500 µL buffer A, 40 µL aliquots were stored at −70 °C), Imubind™ Total TFPI ELISA Kit, Actichrome™ TFPI activity assay, and ActichromTM TF activity assay were obtained from American Diagnostica Inc. (Greenwich, CT, USA). ATenativ, an antithrombin concentrate for clinical application, was from Pharmacia & Upjohn GmbH, Vienna, Austria. To 1000 IU of the lyophilized protein 20 mL of distilled water were added, aliquots of 500 µL were stored at −70 °C. Immobilized sheep antihuman antithrombin (AT), class IgG, coupled to sepharose, was purchased from Affinity Biologicals, Inc. (Ontario, Canada). Rabbit antihuman tissue factor, IgG, was purchased from American Diagnostica Inc., Greenwich, CT, USA. The fibrin polymerization-inhibitor H-Gly-Pro-Arg-Pro-OH (GPRP, Pefabloc™ FG) was purchased from Pentapharm LDT, Basel, Switzerland. The chromogenic substrate used for FXa determination was Suc-Ile-Glu(γ-Pip)-Gly-Arg.pNA (S-2732) from CoaChrom Diagnostics, Vienna, Austria. Fifteen milligrams of the lyophilized substrate were reconstituted with 6.5 mL buffer C to a concentration of 2.9 mmol L−1. The chromogenic substrate used for thrombin determination was H-d-Phe-Pip-Arg-pNA.2HCl (S-2238) from CoaChrom Diagnostics, Vienna, Austria. Twenty-five milligrams of the lyophilized substrate were dissolved in 7 mL of distilled water to a concentration of 5.7 mmol L−1. Testkit prothrombin fragment 1 + 2 (F1 + 2) micro for determination of prothrombin fragment 1 + 2 was purchased from Behring Diagnostics GmbH, Marburg, Germany. Stopping solution for F1 + 2-determination consisted of Trasylol™/EDTA/Na-citrate 8 : 1 : 1 and 110 µmol L−1 PPACK (d-Phe-Pro-Arg chloromethyl ketone) from Sigma, Vienna, Austria. Trasylol™ from Bayer, Vienna, Austria, contained the protease inhibitor aprotinin.

Collection and preparation of plasma

Cord blood was obtained immediately following the delivery of 28 full term infants (38–42 weeks gestational age). Newborns with Apgar scores of nine or less 5 min after delivery were excluded from the study. Blood was collected into 0.1 mol L−1 citrate using a two syringe technique, centrifuged at room temperature for 15 min at 2800 × g, pooled and stored at −70 °C in propylene tubes until assayed. Pro- and anticoagulant factors were in the normal range for neonates. In the same way, plasma from 18 healthy adults was collected from the antecubital vein, prepared, and checked.

Determination of the TF plasma concentrations

TF procoagulant activity was quantitated by means of the Actichrome™ TF activity assay.

Preparation of plasma with different TFPI levels

The TFPI level of the pooled cord plasma was increased by addition of 100 µL buffer A containing 0–11 µL of human TFPI standard to 900 µL plasma. The TFPI level of the pooled adult plasma was decreased by addition of 100 µL buffer A containing 0–55 µL of rabbit anti-TFPI to 900 µL plasma.

Determination of the TFPI plasma concentrations

TFPI antigen levels were determined by means of the Imubind™ Total TFPI ELISA Kit, TFPI activity was determined by means of the chromogenic assay Actichrome™ TFPI activity assay.

Preparation of plasma with different AT levels

AT levels of pooled cord and adult plasma were increased by addition of 100 µL buffer A containing 0–7.5 µL of antithrombin concentrate to 900 µL plasma. The AT content was decreased by using antihuman antithrombin. To 1 mL of plasma different amounts of the antibody poured in buffer A in a ratio of 1 : 1 were added. After 20 min of shaking, the plasma was centrifuged at room temperature for 10 min at 2800 × g. The immobilized antibody/AT complex gathered on the surface and was discharged.

Determination of the AT content

Determination of the AT content of cord plasma was performed by means of a standard chromogenic method on a BM/Hitachi 917 from Boehringer Mannheim, Vienna, Austria.

Activation of plasma

Three hundred microliters of plasma with different TFPI and/or AT levels were incubated with 100 µL of buffer A containing lipidated recombinant human TF (0–100 pmol L−1 final concentration) for 1 min at 37 °C. After subsequent incubation with 20 µL buffer A containing H-Gly-Pro-Arg-Pro-OH (Pefabloc™ FG, 1.0 mg mL−1 final concentration) to inhibit fibrin polymerization [12], plasma samples were activated by addition of 12 µL 0.5 mol L−1 CaCl2.

Determination of clotting time

Clotting times were determined by means of the optomechanical coagulation analyzer Behring Fibrintimer II from Dade Behring Marburg GmbH, Marburg, Germany, which applies the turbodensitometric measuring principle.

Determination of FX activation

Plasmas were prepared and activated as described above. The reagents were prewarmed to 37 °C. At timed intervals 25 µL aliquots were withdrawn from the activated plasma and subsampled into 300 µL buffer C containing 0.97 mmol L−1 S-2732. Amidolysis of S-2732 was stopped after 8 min by addition of 250 µL of 50% acetic acid. The amount of FXa generated was quantitated by measuring the absorbance by double wave length (405–690 nm) in the Anthos microplate-reader 2001, from Anthos Labtec Instruments GmbH, Salzburg, Austria.

Determination of thrombin generation

We used a subsampling method derived from a recently described technique [13,14]. Plasmas were prepared and activated as described above. At timed intervals, 10 µL aliquots were withdrawn from the activated plasma and subsampled into 490 µL buffer B containing 255 µmol L−1 S-2238. Amidolysis of S-2238 was stopped after 6 min by addition of 250 µL 50% acetic acid. The amount of thrombin generated was quantitated by measuring the absorbance by double wave length (405–690 nm) in the Anthos microplate-reader 2001, from Anthos Labtec Instruments GmbH, Salzburg, Austria. The total amidolytic activity measured is caused by the simultaneous activity of free thrombin and the alpha 2-macroglobulin (α2-M)/thrombin complex [15]. Free thrombin generation curves were calculated by mathematical treatment of total amidolytic activity curves using a method developed by Hemker et al. [13].

Determination of prothrombin fragment 1 + 2

Plasmas were prepared and activated as described above. At timed intervals 10 µL aliquots were withdrawn from the plasma and subsampled into 490 µL stopping solution. After subsequent 1 : 10 dilution in stopping solution, the amount of F1 + 2 generated was quantitated by using a standard immuno enzymatic test kit.

Statistical analysis

Results obtained in cord and adult plasma were compared by means of Mann–Whitney U-test. The effects of different concentrations of TFPI and AT on clotting time, FXa-generation, TP, and F1 + 2 activation were analyzed using paired t-test. The significance level of P-values was set at 5%. Calculations were performed using winstat 3.1 (Kalmia Co. Inc., USA).

Results

Physiological TF levels in cord and adult plasma

TF procoagulant activity levels were significantly higher in adult plasma (22.6 ± 1.8 pmol L−1) than in cord plasma (9.5 ± 1.1 pmol L−1, P < 0.001).

Effect of varying TF concentrations on clotting time in cord and adult plasma

Cord plasma, physiologically containing approximately 9.5 pmol L−1 TF, clotted within 10 min after addition of CaCl2 (Fig. 1). When the TF content of the pooled cord plasma was decreased by applying antihuman TF, clotting did not occur within 20 min after addition of CaCl2 at TF concentrations below 6 pmol L−1. When the TF concentration in cord plasma was successively elevated by addition of the purified protein, the clotting time was dose-dependently shortened. In adult plasma, physiologically containing approximately 22.6 pmol L−1 TF, clotting did not occur within 20 min after addition of CaCl2. After addition of 2.5 pmol L−1 TF, the plasma clotted within 10 min. Further elevation of the TF resulted in dose-dependently shortened clotting times. After addition of less than 10 pmol L−1 lipidated TF, cord plasma clotted significantly earlier than adult plasma (323 ± 6 vs. 411 ± 7 s, P < 0.05, 2.5 pmol L−1 lipidated TF were added). After addition of more than 10 pmol L−1 lipidated TF, adult plasma clotted slightly, but not significantly earlier than cord plasma (114 ± 4 vs. 121 ± 4 s, 20 pmol L−1 lipidated TF were added).

Figure 1.

Effect of addition of various amounts of lipidated TF on clotting time in cord (▪) and adult plasma (●). Results shown are expressed as means (n = 5). SDs were not shown for clarity of graph reading but represented less than 10% of the mean.

Effect of varying TF concentrations on thrombin generation in cord and adult plasma

Thrombin generation curves in both cord and adult plasma were monitored after addition of different amounts of lipidated TF. Table 1 shows that the TP (thrombin potential, the area under the thrombin generation curve) was significantly higher in adult compared with cord plasma. However, the differences between the TP of adult and cord plasma decreased with decreasing concentrations of lipidated TF added. After simultaneous rise of TFPI and AT in cord plasma to adult levels the TP became significantly lower compared with untreated cord plasma, whereby the proportional differences between the TPs increased with decreasing amounts of lipidated TF added (Table 1).

Table 1.  Effects of lipidated TF on the thrombin potential in adult, cord, and cord plasma containing TFPI and AT at adult levels, respectively
 Added lipidated TF (pmol L−1)
100202.5
  • *

    P < 0.01 compared with adult plasma.

  • **

    P < 0.01 compared with cord plasma.

Thrombin potential (nmol L−1 min−1) in
 Adult plasma694 ± 28468 ± 19329 ± 16
 Cord plasma347 ± 22*298 ± 15*291 ± 14*
 Cord plasma + TFPI + AT246 ± 17**197 ± 10**180 ± 11**

Physiologic TFPI levels in cord and adult plasma

TFPI activity levels determined by activity assay were significantly higher in adult plasma than in cord plasma (1.24 ± 0.16 vs. 0.81 ± 0.09 U mL−1; P < 0.01). Accordingly, TFPI antigen levels determined by ELISA were also significantly higher in adult plasma (54.53 ± 10.63 ng mL−1) than in cord plasma (23.24 ± 3.96 ng mL−1; P < 0.01).

Effect of varying initiator concentrations on clotting time in cord plasma in the presence of elevated TFPI and/or AT levels

Raising the TFPI level in cord plasma to adult value resulted in significantly prolonged clotting times for a wide range of lipidated TF concentration applied to initiate clotting (0–100 pmol L−1) compared with cord plasma containing TFPI at neonatal level (e.g. –t = 230 s, P < 0.001, after addition of 20 pmol L−1 lipidated TF, Fig. 2). The anticoagulant effect of raised TFPI was most distinct when plasmas were activated by small amounts of lipidated TF. In contrast to spiking with TFPI, when cord plasma was spiked to contain AT at adult levels, clotting times were only significantly prolonged when lipidated TF concentrations below 5 pmol L−1 were applied (–t = 132 s, P < 0.05, after addition of 1.25 pmol L−1 lipidated TF). The simultaneous raise of both inhibitors resulted in the most pronounced effect on clotting time: prolongation of clotting time due to addition of AT was more distinct in cord plasma containing TFPI at adult levels compared with cord plasma containing physiological amounts of TFPI (–t = 80 ± 4 s vs. 39 ± 3 s, P < 0.01, when 2.5 pmol L−1 lipidated TF were added, Fig. 2). This effect increased with decreasing amounts of added lipidated TF.

Figure 2.

Effect of lipidated TF on prolongation of clotting time in cord plasma spiked to contain TFPI at adult level (▪), AT at adult level (●), and cord plasma spiked to contain both inhibitors at adult levels (▴), respectively, compared with cord plasma containing physiological amounts of both inhibitors. Results shown are expressed as means (n = 5). SDs were not shown for clarity of graph reading but represented less than 10% of the mean.

Effect of raising TFPI and/or AT to adult value on FXa generation in cord plasma activated by 2.5 pmol L−1 lipidated TF

Raising TFPI, AT, and both TFPI and AT to adult levels, respectively, resulted in significantly suppressed FXa-peak values compared with cord plasma containing physiological amounts of inhibitors (6.94 ± 0.33, 5.99 ± 0.44, and 5.06 ± 0.24 vs. 9.02 ± 0.58 nmol L−1, P < 0.01, Fig. 3).

Figure 3.

FXa generation in cord plasma with physiological amounts of TFPI and AT (▪), TFPI at adult level (▴), AT at adult level (●), and both inhibitors at adult levels (▾), respectively. Lipidated TF (2.5 pmol L−1) were applied to initiate coagulation. Results shown are expressed as means (n = 5). SDs were not shown for clarity of graph reading but represented less than 10% of the mean.

Effect of raising TFPI and/or AT to adult value on thrombin generation in cord plasma activated by 2.5 pmol L−1 lipidated TF

The raise of TFPI, AT, and of both inhibitors, respectively, resulted in significant suppression of the TP (227 ± 9, 197 ± 10, and 180 ± 11 vs. 291 ± 14 nmol L−1 min−1, P < 0.01, Fig. 4).

Figure 4.

Thrombin generation in adult plasma (▪), in adult plasma containing TFPI at neonatal level (●), in cord plasma containing physiological amounts of TFPI and AT (▴), in cord plasma containing TFPI at adult level (◆), in cord plasma containing AT at adult level (●), and in cord plasma containing both inhibitors at adult levels (▾), respectively, in the presence of 2.5 pmol L−1 lipidated TF as initiator of coagulation. Results shown are expressed as means (n = 5). SDs were not shown for clarity of graph reading but represented less than 10% of the mean.

Effect of reducing the TFPI level in adult plasma to neonatal value on thrombin generation and clotting time after addition of 2.5 pmol L−1 lipidated TF

Anti-TFPI antibody was added to adult plasma to decrease the TFPI level to neonatal level. In the prepared plasma samples significantly more thrombin was generated (384 ± 19 vs. 329 ± 16 nmol L−1 min−1, P < 0.01), and the clotting time was significantly shortened (280 ± 5 vs. 411 ± 7 s, P < 0.001) compared with adult plasma with physiological TFPI content (Fig. 4).

Effect of TFPI and AT on prothrombin fragment 1 + 2 generation in cord plasma

Raising the TFPI concentration to adult values resulted in significantly suppressed F1 + 2-generation (0.470 ± 0.02 vs. 0.712 ± 0.03 µmol L−1, P < 0.001). In accordance, raising AT to adult levels also resulted in significantly suppressed prothrombin fragment 1 + 2 generation (0.356 ± 0.02 vs. 0.712 ± 0.03 µmol L−1, P < 0.001). Whereby, F1 + 2-generation was significantly higher in the presence of TFPI at adult levels compared with F1 + 2-generation in the presence of AT at adult levels (P < 0.001). Raising both TFPI and AT to adult levels resulted in the most pronounced suppression of F1 + 2-generation (0.328 ± 0.02 vs. 0.712 ± 0.03 µmol L−1, P < 0.001).

Discussion

In the present study we found increased ability to coagulate of cord plasma compared with adult plasma in the presence of small amounts (<10 pmol L−1) of the initiator lipidated TF [16,17]. Cord plasma clotted earlier, and, correspondingly, the lag-phase until the onset of FXa and thrombin generation was markedly shortened compared with adult plasma. We show that the high ability to coagulate of cord plasma in the presence of small amounts of initiator is attributable to the physiological low levels of the natural inhibitors TFPI and AT present in cord plasma.

High efficacy of TFPI and AT together in the regulation of thrombin inhibition has already been shown by Van't Veer and Mann [18] in a purified system adjusted to contain pro- and anticoagulant factors at adult levels. The effects of TFPI and AT acting in concert result in a synergistic inhibition of thrombin generation at low concentrations of the initiator lipidated TF. The authors divided thrombin formation into two phases, an ‘initiation phase’, in which FV and FVIII are quantitatively cleaved and trivial amounts of FXa and FIXa are produced, and a ‘propagation phase’, in which prothrombin is quantitatively activated [19,20]. TFPI has been shown to extend the initiation phase of thrombin generation and to reduce the rate of thrombin generation during the propagation phase [21]. FVa has been reported to protect FXa from inactivation by AT [22,23]. TFPI significantly delays FV activation, thereby increasing the ability of AT to inhibit FXa. Thus, TFPI potentiates the action of AT, and TFPI and AT together are ∼70-fold more potent at inhibiting thrombin formation than the inhibitors acting separately. The two inhibitors, acting in concert, provide a synergistic inhibitory effect [18]. However, this claim of synergism is made with caution without full knowledge of all the mechanisms involved [24].

Van't Veer and Mann [18] have shown in this context that the amount of initiator necessary to induce explosive thrombin generation depends on the concentrations of TFPI and AT in the reconstituted system. For the adult system this threshold (10–20 pmol L−1 TF/FVIIa) defines a ‘go’ or ‘no go’ switch for the blood coagulation reaction. However, the height of this threshold probably is not valid in cord plasma due to markedly lower concentrations of TFPI and AT present. In addition, we applied defined amounts of lipidated TF and not of TF/FVIIa to cord or adult plasma to initiate clotting in order to take into account that cord plasma physiologically contains less FVII/FVIIa than adult plasma [1,25]. Therefore, our experiments were designed to provide enough time for the complex formation between TF and FVIIa. We found that the initiator threshold to inducing clot formation within 10 min after applying the initiator in cord plasma was approximately 6 pmol L−1, and in adult plasma approximately 25 pmol L−1. As a consequence, this low threshold resulted in increased ability to coagulate of cord plasma when clot formation was induced by addition of small amounts (<10 pmol L−1) of lipidated TF compared with adult plasma.

The effect of TFPI on prolongation of the lag time of thrombin generation decreased with increasing amounts of exogenous initiator. Thus, under a ‘high coagulant challenge’, where the effect of TFPI is negligible, the lag phase of thrombin generation is shorter in adult than in cord plasma due to the physiological higher levels of procoagulant proteins present in adult plasma. FVIIa has been shown to shorten the duration of the initiation phase [26], and elevated levels of prothrombin compared with cord plasma result in enhanced thrombin generation [7].

In conclusion, despite of low concentrations of procoagulant proteins, physiological low levels of TFPI together with physiological low levels of AT allow sufficient thrombin formation in cord plasma activated with small amounts of lipidated TF. These findings seem to offer a good explanation for the efficient clotting in neonates despite physiological low prothrombin concentrations.

Acknowledgements

This study was supported by a grant from the ‘Gesellschaft zur Förderung der Gesundheit des Kindes (INVITA)’.

Ancillary