Dr Bjørn Bendz, Haematological Research Laboratory, Ullevål University Hospital, N-0407 Oslo, Norway. e-mail: email@example.com
Tissue factor pathway inhibitor (TFPI) is released to circulating blood after intravenous (i.v.) and subcutaneous (s.c.) injections of heparins, and may thus contribute to the antithrombotic effect of heparins. We have recently shown that total TFPI activity, plasma free TFPI antigen, and heparin releasable TFPI were partially depleted during repeated and continuous i.v. infusion of unfractionated heparin (UFH), but not during s.c. treatment with a low molecular weight heparin (LMWH). The difference may be attributed to a different mode of action or the different mode of administration. In the present randomized cross-over study, s.c. administration of therapeutic doses of UFH was compared with s.c. administration of two LMWHs. 12 healthy male volunteers were treated for 3 d with UFH, 250 U/kg twice daily, dalteparin, 200 U/kg once daily, and enoxaparin, 1.5 mg/kg once daily. Six participants were also treated with UFH, 300 U/kg once daily. On day 5 a single dose of either drug was given. Peak levels of total TFPI activity and free TFPI antigen were detected 1 h after injection, whereas maximal prolongation of activated partial thromboplastin time (APTT) and peak levels of anti-factor Xa activity and anti-factor IIa activity were detected after 4 h. On UFH administered twice daily, free TFPI antigen decreased by 44% from baseline level before the first injection on day 1 to pre-injection level on day 5. On UFH administered once daily, basal free TFPI antigen decreased by 50%, 56% and 27% on day 2, 3 and 5 respectively, compared with day 1. Minimal depletion of TFPI was detected during treatment with LMWHs. The study demonstrates the different modes of action of LMWHs and UFH and may help to explain the superior antithrombotic efficacy of LMWHs.
Numerous randomized clinical trials have recently provided clear evidence that low molecular weight heparins (LMWHs) are of superior efficacy when compared with unfractionated heparin (UFH) for the treatment of both venous (Siragusa et al, 1996) and arterial (Cohen et al, 1997) thrombosis. Although a large number of studies in animal and human models have elucidated the main antithrombotic effects of UFH and LMWHs, the underlying antithrombotic mechanism(s) responsible for the difference in therapeutic efficacy are still not fully understood.
A major anticoagulant effect of both UFH and LMWHs is due to the potentiation of antithrombin activity, but the target proteases are different. LMWHs potentiate the inhibitory effect of antithrombin on factor Xa (FXa) activity to inhibit prothrombin activation (Hemker & Beguin, 1990), whereas UFH also inhibit thrombin generated by the prothrombinase complex (Samama et al, 1994). Another effect of both LMWHs and UFH is mediated through release and redistribution of tissue factor pathway inhibitor (TFPI) from the vascular endothelium to circulating blood (Sandset et al, 1988). TFPI inhibits FXa directly, but its main effect is to inhibit factor VIIa/tissue factor (FVIIa/TF) catalytic activity in an FXa-dependent manner (Broze & Miletich, 1987; Rao & Rapaport, 1987). The inhibition involves either the binding of preformed TFPI/FXa complexes to FVIIa/TF complexes (Girard et al, 1989) or the binding of TFPI to preformed FXa/FVIIa/TF complexes (Broze, 1995).
TFPI is a member of the Kunitz family of serine protease inhibitors. The molecule contains 276 amino acid residues with an acidic amino-terminal region followed by three tandem Kunitz-type inhibitory domains and a basic carboxyl-terminal region (Wun et al, 1988). The first and second Kunitz inhibitory domains contain the reactive sites responsible for binding to and inhibition of FVIIa and FXa, respectively (Girard et al, 1989). The third domain may be involved in the association of TFPI with lipoproteins (Abumiya et al, 1995) and is mandatory for the anticoagulant function of TFPI in clotting based (diluted prothrombin time) TFPI assays (Wesselschmidt et al, 1992; Nordfang et al, 1991). The carboxyl-terminal region contains a highly basic region which serves as a high affinity binding site for heparin (Wesselschmidt et al, 1992). A low affinity binding site for heparin is located in the third Kunitz domain (Wesselschmidt et al, 1993).
Vascular endothelium is the primary production site for TFPI (Bajaj et al, 1990). A major pool of TFPI in vivo (50–80% of total) is native TFPI bound to the endothelial surface which may be released into circulating blood after injection of heparin (Hubbard et al, 1994; Novotny et al, 1991; Sandset et al, 1987). A smaller TFPI pool (10–50% of total) circulates in plasma, but is predominantly found as various truncated forms of TFPI bound to lipoproteins (Hansen et al, 1995; Novotny et al, 1989). Only 5–20% of plasma TFPI circulates as carrier-free TFPI molecules. Small amounts of TFPI have also been detected in platelets (Novotny et al, 1988). Heparin-releasable TFPI and plasma-free TFPI, but not lipoprotein-associated TFPI, exert strong anticoagulant effect in diluted prothrombin time assays (Hansen et al, 1997; Lindahl et al, 1992), which indicate that these forms of TFPI are physiologically active.
We have recently shown that plasma-free TFPI antigen and heparin releasable TFPI are partially depleted during repeated and continuous intravenous (i.v.) infusion of therapeutic doses of UFH in man (Hansen et al, 1996, 1998), but not during subcutaneous (s.c.) injections of therapeutic doses of a LMWH (Hansen et al, 1998). We speculated that differential effects of UFH and LMWH on depletion of TFPI could explain the apparent clinical superiority of LMWHs. However, since the mode of administration of UFH and LMWH differed, the possibility remains that the differential effect is related to different pharmacokinetics rather than a real difference of UFH and LMWH on TFPI levels. In the present study we have therefore carefully compared s.c. injections of UFH with two LMWHs, all in therapeutic doses, to determine whether the differences could be attributed to a different mode of action, i.e. a novel property of LMWHs, or to a different mode of administration.
MATERIALS AND METHODS
Unfractionated heparin (Heparin®, 25 000 U/ml) was purchased from Nycomed Pharma AS, Oslo, Norway. Dalteparin (Fragmin®, 10 000 anti-factor Xa U/ml) was a gift from Pharmacia & Upjohn, Oslo, Norway, and enoxaparin (Klexane®, 100 mg/ml) was a gift from Rhône-Poulenc Rorer, Oslo, Norway.
12 (n = 12) healthy, normolipaemic male medical students at the University of Tromsø (mean age 23, range 20–26 years), who were consuming a traditional western diet, were recruited. Before inclusion all participants underwent a clinical consultation including a medical history and a physical examination. Body weight and height, blood pressure, routine haematological and haemostatic variables (including platelet count, activated partial thromboplastin time (APTT), prothrombin time, and primary (Ivy) bleeding time), and routine biochemistry (including blood glucose and serum lipids) were measured. Exclusion criteria were regular use of drugs, use of drugs that might interfere with haemostasis during the month prior to study entry, general or local bleeding diathesis, previous thromboembolic disease, arterial aneurysm, trauma or surgical treatment during the last month before inclusion, hypertension, hypersensitivity to heparin, liver and kidney disease, mental illness, and alcohol or drug abuse. Participants were instructed to refrain from strenuous physical exercise and alcohol for 48 h before each injection. The study was approved by the Regional Board of Research Ethics and written informed written consent was obtained from each individual.
The study was a randomized cross-over design, with a 1 week wash-out period between each treatment allocation, performed on an out-patient basis at the Clinical Research Facilities of the University Hospital in Tromsø. The 12 participants received treatment (A) s.c. UFH, 250 U/kg twice daily, (B) s.c. dalteparin, 200 U/kg kg once daily, and (C) s.c. enoxaparin, 1.5 mg/kg once daily, for 3 d. On day 5 of each treatment cycle a single dose of either drug was given. The 12 participants were randomly allocated in two groups of six subjects. The first group conducted treatment in sequence A–B–C, whereas the second group conducted treatment in sequence C–B–A. Six of the participants were also separately treated with s.c. UFH, 300 U/kg once daily, in the same fashion.
During the first day of treatment, blood samples were collected immediately prior to the first injection and after 0.5, 1, 1.5, 2, 4, 6, 8, 12 and 24 h. During the second and third day of treatment, blood samples were collected immediately prior to and 4 h after the next injection, i.e. after 24 and 28 h on the second day, and after 48 and 52 h on the third day. On the fifth day, blood samples were drawn immediately prior to the final injection, i.e. after 96 h, and then after 1, 2, 4 and 8 h, i.e. 97, 98, 100 and 104 h after the first injection.
Blood was drawn through an 18 Gauge Venflon (BOC, Ohmeda AB, Helsingborg, Sweden) or a 19 Gauge needle into triple siliconized Vacutainer tubes (Becton Dickinson, Meylan Cedex, France) containing 0.129 m sodium citrate (1 vol anticoagulant and 9 vol whole blood). Blood was centrifuged at 2000 g for 15 min at 22°C and plasma was stored in sterile cryovials of 0.5 ml at −70°C. Blood for reference plasma was obtained from blood donors at the Blood Bank, Ullevål University Hospital.
Total TFPI activity (TFPI Ac) was assayed with a two-stage amidolytic assay (Sandset et al, 1991) but with minor modifications. Tissue factor (TF), Dade® Innovin®, was from Baxter Diagnostics Inc., U.S.A. Recombinant factor VIIa was a gift from Novo- Nordisk AS, Biopharmaceutical Division, Gentofte, Denmark. The concentration of recombinant factor VIIa in the final incubation was 15.0 ng/ml. Incubation I and II were 20 and 30 min respectively, both at 37°C. Total TFPI activity in a reference plasma was defined as 1 U/ml.
Free TFPI antigen (free TFPI Ag) was measured with a solid phase two-site enzyme immunoassay (Ostergaard et al, 1997). The method detects carrier-free and heparin-releasable TFPI molecules in plasma, but not lipoprotein-associated TFPI. The assay is insensitive to UFH (0–10 IU) added to plasma in vitro (Hansen et al, 1996). Basal free TFPI Ag was defined as the level of free TFPI Ag in the blood samples collected immediately before an injection of UFH or LMWH. Post-heparin free TFPI Ag was defined as the level of free TFPI Ag in samples collected 4 h after an injection.
Activated partial thromboplastin time (APTT) was measured on an ACL 3000 coagulation analyser (Instrumentation Laboratories, Milan, Italy) with reagents from Nycomed Pharma (Cephotest®). Anti-Xa activity and anti-IIa activity in plasma were measured with commercial chromogenic assays (Coatest Heparin, Chromogenix AB, Mölndal, Sweden, and IL TestTM Heparin (IIa), Instrumentation Laboratories, respectively). The anti-Xa activity was calibrated with serial dilutions (0–0.8 IU/ml) of a commercial calibrator (LMW heparin anti-Xa standard, Chromogenix, which was calibrated against the first international standard for LMWH) in a reference plasma, whereas the anti-IIa activity was calibrated with serial dilutions (0–1.0 IU/ml) of UFH.
The SPSS 8.0 statistical software package for Windows was used. Descriptive statistics on variables revealed normal distribution of recorded data. One-way analysis of variance (ANOVA) was used to identify differences between treatment groups. Repeated measures of variance was used to test for changes of a variable over time within a treatment group. The paired t-test was applied for comparison of a variable before and after treatment and also for comparison of changes between treatment groups in the cross-over design. P-values <0.05 were considered statistically significant. All values are given as mean and standard deviation (SD) if not otherwise stated.
Mean body weight (BW) and body mass index (BMI) were 78.0 (9.9) kg and 23.8 (2.9) kg/m2 respectively. In the subgroup receiving UFH 300 U/kg once daily, BW was 77.4 (5.0) kg and BMI was 23.2 (1.3) kg/m2. All the participants exhibited normal haemostasis assessed by the medical history and laboratory screening (see Methods). All individuals completed the study according to the protocol and no adverse events or bleeding complications were observed.
Activated partial thromboplastin time, anti-Xa activity, and anti-IIa activity
In the group receiving UFH, maximal prolongation of APTT, and peak levels of anti-Xa and anti-IIa activities were achieved 4 h after injection (Fig 1A). On UFH administered twice daily progressive increases of anti-Xa activity (Fig 2A) and anti-IIa activity (data not shown) were seen in the pre-injection samples and in the samples collected 4 h after injection (Fig 2B) suggesting accumulation of UFH. In the pre-injection sample on day 3 peak anti-Xa activity (Fig 2A) and anti-IIa activitiy (data not shown) were 0.73 (0.40) U/ml and 0.61 (0.20) U/ml, respectively. In the post-injection samples anti-Xa activity (Fig 2B) and anti-IIa activity (data not shown) increased from 0.40 (0.18) U/ml and 0.41 (0.14) U/ml on day 1 to 1.17 (0.46) U/ml and 0.79 (0.11) U/ml on day 3, respectively (P < 0.001).
Due to the observed accumulation of heparin on the twice daily regimen, six of the 12 participants were also given UFH, 300 U/kg, once daily. In this group, both APTT (data not shown), and anti-Xa activity (Fig 2A) and anti-IIa activity (data not shown) reached pretreatment levels before the next injection. However, also on this regimen, APTT, anti-Xa activity and anti-IIa activity increased significantly during the first 3 d in the samples collected 4 h after injection (Fig 2B).
The LMWHs, dalteparin and enoxaparin, had only minimal effect on APTT and anti-IIa activity (data not shown). Peak anti-Xa activity was observed 4 h after injection (Fig 1) and reached 1.13–1.22 U/ml (days 1–3) in the dalteparin group and 1.21–1.26 U/ml in the enoxaparin group. There was no accumulation of dalteparin, but in the enoxaparin group a minor accumulation increase was seen in the pre-injection samples to 0.16 (0.04) U/ml (P < 0.001) (Fig 2A).
The change in anti-Xa activity after each injection (post-injection value − pre-injection value) increased progressively on UFH once daily, but remained stationary on UFH twice daily, dalteparin, and enoxaparin (Fig 2).
Free TFPI antigen
UFH administered twice daily yielded a progressive increase in basal free TFPI Ag levels (Fig 3), which coincided with accumulation of UFH as judged by APTT, anti-Xa and anti-IIa activities (Fig 2A). However, on day 5, i.e. 36 h after the last injection, basal free TFPI Ag was significantly decreased by 44% (P < 0.001) compared with the pretreatment level. At this time-point no heparin was detectable in the plasma samples. Because of the accumulation of UFH observed in the group given UFH twice daily, a subgroup of six individuals was then given UFH once daily. In this group the heparin concentration decreased to pretreatment levels prior to each injection (see above). Basal free TFPI Ag levels decreased progressively by 56% (P < 0.001) from 20 (5) ng/ml at baseline on day one to 9 (2) ng/ml on day 3. On day 5, TFPI was still reduced by 27% (P = 0.02).
On dalteparin and enoxaparin, minimal changes were seen in basal free TFPI Ag, although a marginal 14% decrease in the dalteparin group on day 3 (P = 0.01) and a 12% decrease on day 5 in the enoxaparin group (P = 0.03) reached statistical significance (Fig 3).
Peak levels of free TFPI Ag were detected only 1 h after the injection of UFH and the two LMWHs in spite of maximum prolongation of the APTT and peak anti-Xa and anti-factor IIa activities after 4 h (Fig 1). Analysis of variance indicated significant differences between the four treatment modalities on the levels of free TFPI Ag in the post-heparin samples (P < 0.01). Dalteparin, 200 U/kg, released slightly more TFPI than both enoxaparin, 1.5 mg/kg (P < 0.01), and more than UFH, 300 U/kg (Fig 3B).
Post-heparin free TFPI Ag was unchanged within all the four treatment groups, despite the fact that APTT, anti-Xa and anti-IIa activities increased progressively in the two groups receiving UFH, but not on LMWHs. Thus, progressive increased levels of UFH may have released more TFPI and masked a true reduction of post-heparin free TFPI Ag during UFH treatment. However, when the participants were exposed to a single injection of either drug and dose on day 5, UFH administered twice daily was associated with a 20% reduction in the post-heparin sample compared with day 1 (P = 0.01).
Total TFPI activity
Pretreatment levels of total TFPI Ac increased progressively during treatment with UFH administered twice daily (Fig 4). Again, the effect was most probably attributed to accumulation of UFH. Nonsignificant changes, both in basal and post heparin samples (Fig 4), were seen on UFH administered once daily and in the groups receiving dalteparin and enoxaparin.
We have previously shown that intravascular TFPI is partially depleted during repeated i.v. injections (Hansen et al, 1996) and continuous i.v. infusion (Hansen et al, 1998) of UFH, but not during s.c. treatment with a LMWH (Hansen et al, 1998). The difference is of potential importance for the understanding of the apparent superior efficacy of LMWHs in the prophylaxis and treatment of thrombosis. Since the mode of administration differed, our study did not allow us to conclude whether this was a novel property of LMWHs or whether the difference simply was related to different mode of administration (Hansen et al, 1998). In the present study we have therefore compared the effects of s.c. treatment with UFH and two LMWHs on TFPI levels in plasma.
The study was designed to make administration of UFH and LMWH as similar and comparable as possible with a high peak plasma concentration and only trace amounts of UFH and LMWH detectable before the next injection. The bioavailability of therapeutic, but not prophylactic, dosages of s.c. UFH is high and comparable to that of LMWHs (Bergqvist et al, 1983). We initially considered a twice daily regimen of UFH to be comparable to a once daily regimen of LMWH. Our results showed, however, that twice daily injections of UFH caused significant accumulation as judged by prolongation of APTT and levels of anti-Xa activity and anti-IIa activity in the pre-injection and post-injection samples. Since accumulation of heparin in plasma might release more TFPI from the vessel wall to plasma and interfere with interpretation of the test results, a second series was performed with UFH injected only once daily. In this experimental group no accumulation of UFH was observed as judged by the pre-injection levels, but there was an interesting progressive increase in post-injection anti-Xa activity levels. This increase may not be explained by accumulation of heparin, since heparin did not increase in the pre-injection samples, but could be attributed to depletion of heparin-neutralizing proteins on repeated injections of heparin. We believe that the present experiments compared the different heparins as close as possible.
Our findings confirm and extend our previous observations on depletion of TFPI during i.v. UFH therapy (Hansen et al, 1996, 1998) that TFPI is also is depleted during s.c. treatment with UFH. The effect was most evident from the progressive decrease in basal free TFPI antigen levels by > 50% during s.c. UFH once daily. During UFH given twice daily we believe that the large accumulation of heparin may have released more TFPI from the vessel wall to mask a true depletion of free TFPI antigen in plasma. However, also in this group depletion was evident in the sample collected on the fifth day, i.e. 36–48 h after the last injection of UFH. At this time-point no heparin was detectable in plasma and basal free TFPI antigen was reduced by nearly 50%.
In the post-heparin samples, accumulation of heparin on UFH given twice daily may similarly have masked a true reduction of heparin-releasable TFPI. Again, the reduced response after the injection on the fifth day indicates that depletion of heparin-releasable TFPI indeed took place. The results obtained on the release of TFPI after injection of UFH once daily indicated no significant depletion of TFPI (Fig 3B). However, this regimen was associated with a peculiar progressive increase in the post-injection heparin concentration not related to accumulation (see above) and may have masked a true depletion of free TFPI antigen. The present results may therefore be taken as evidence that both plasma TFPI and endothelial-associated or heparin-releasable TFPI were depleted during s.c. UFH therapy.
The present study also confirms that LMWH therapy is associated with no or only minimal depletion of TFPI (Hansen et al, 1998). Similar responses were seen for dalteparin and enoxaparin, which suggests that the effect may be a class effect of LMWHs. Release of TFPI was slightly higher for the recommended therapeutic dose of dalteparin (200 U/kg) compared with that of enoxaparin (1.5 mg/kg). The clinical significance remains questionable, since there were no signs of depletion in any of the TFPI pools and the role of different TFPI pools for the antithrombotic effect of heparins is unknown.
Our study failed to show significant effects of either UFH or LMWH on TFPI activity, which may again in part be explained by the accumulation or progressive increase in plasma concentration of UFH. TFPI activity was assayed with a chromogenic substrate assay (Sandset et al, 1991), which detects both truncated and full-length free TFPI and lipoprotein-associated TFPI (Sandset et al, 1991; Lindahl et al, 1992; Hansen et al, 1996; Sandset, 1999). Since the assay was calibrated against pooled normal plasma, which contains mostly lipoprotein-associated TFPI, it was rather insensitive to the assay of free or heparin-releasable TFPI and may therefore have not detected depletion of TFPI in the present study. Also in our previous study, depletion of TFPI was much less prominent in the activity assay compared with the free antigen assay (Hansen et al, 1996). Finally, it must be borne in mind that free TFPI and heparin-releasable TFPI exert much stronger anticoagulant activity when compared with lipoprotein-associated TFPI (Lindahl et al, 1992; Hansen et al, 1996, 1997; Sandset, 1999). These data indicate that free TFPI and heparin-releasable TFPI are biologically active in vivo, and that the assay of free TFPI antigen may be more relevant than the assay of TFPI by an end-point activity assay (Sandset, 1999).
Two recent studies have indicated increased survival in patients with venous thromboembolism and metastatic carcinoma who initially received treatment with LMWH compared with UFH (Hull et al, 1992; Prandoni et al, 1992; Leizorovicz et al, 1994). High levels of TFPI have been detected in some patients with cancer (Iversen et al, 1998), which may indicate an important physiological role of TFPI in cancer. Depletion of TFPI on UFH, but not on LMWH, could be an important mechanism to explain the increased survival, although depletion of several other plasma and vascular wall proteins of potential importance for tumour cell growth or angiogenesis, e.g. lipoprotein lipase, superoxide dismutase, and platelet factor 4, could also be involved.
Maximal release of TFPI after injection of UFH and LMWH was observed within only 1 h, in contrast to maximal prolongation of APTT and peak levels of anti-IIa and anti-Xa activities after 3–4 h. Similar discordant effects of s.c. heparin on TFPI and anti-Xa activity have also been observed by other investigators (Falkon et al, 1998; Holst et al, 1997). After i.v. injections of either UFH (Sandset et al, 1988) or LMWH (Samama et al, 1994), release of TFPI is closely related to anti-Xa activity. The rapid release of TFPI after s.c. injections of UFH or LMWH may involve rapid absorption of heparin fragments without anti-IIa or anti-Xa activities, but with retained ability to release TFPI. It could also be an effect secondary to high binding affinity of heparin to TFPI, which fits with the low concentration of heparin needed to release TFPI in cell culture studies (Iversen et al, 1996). Alternatively, rapid release of TFPI could be mediated by other components released or generated by the absorption of UFH and LMWH.
The differential effect on intravascular TFPI may reflect a real pharmacological difference between UFH and LMWH related to differences in chemical composition. Valentin et al (1994) demonstrated that the affinity of full-length TFPI for heparin increased with the charge density and chain length of heparin species. One possibility is that LMWHs, unlike UFH, do not bind with high affinity to, and hence do not deplete, plasma TFPI. Since also in this study the pharmacokinetic profiles of UFH and LMWH were not identical, we may not completely rule out that the differential effect was related different pharmacokinetics (Hansen et al, 1998). However, we may conclude that treatment with i.v. and s.c. UFH, but not with s.c. LMWH, was associated with partial depletion of plasma TFPI and heparin-releasable TFPI, which may help to explain the apparent clinical superior efficacy of LMWHs in patients with acute thrombosis.
The authors thank medical students Christine Nitter and Karsten Myhre, and the staff of the Clinical Research Centre, University Hospital of Tromsø, for excellent technical assistance. The study was supported by grants from the Norwegian Council on Cardiovascular Diseases, Rhône-Poulenc Rorer, Norway, and Pharmacia & Upjohn, Norway.