Measurement of the rate of thrombin production in human plasma in contact with different materials Gordon Rollason and Michael V. Sefton“ Department of Chemical Engineering and Applied Chemistry and Centre for Biomaterials, University of Toronto, Toronto, Ontario, Canada M5S IA4 Thrombin production in plasma in contact with various materials was consistent with a first-order autocatalytic model (d[T]/dt= k,[T]; [TI = thrombin concentration, t = time, k , = thrombin production rate constant) since the initial portion of a semilogarithmic plot of thrombin concentration against time was linear. Thrombin concentration was measured i n clotting plasma (phospholipid enhanced or platelet-rich plasma) using a fluorogenic substrate (BMCA) by aliquot sampling at various intervals or more conveniently by monitoring cumulative fluorescence. The latter was generated by the action, on BMCA incubated in the clotting plasma, of the thrombin as it was generated. The thrombin concentration was determined from the first derivative of the S-shaped cumulative fluorescence curve. k , was greater for glass (7.92 x cm/s) than for the other materials (polypropyl-
ene, polystyrene, polyethylene and PVA; k , - 3 . 1 x lo-’ cm/s) i n plasma w i t h cephalin without flow. A k , for heparinPVA could not be determined since the thrombin concentration was too low to be quantified. A larger difference between polyethylene and PVA was noted with platelet-rich plasma without flow while lower values (1.0 x cm/s) were noted in a flow system but at a higher surface to volume ratio. The first-order rate constant can be used i n simple models relating production of thrombin at a wall of a tube to its mass transfer away from the wall in flowing blood. One such model predicts that the concentration of thrombin at the wall should become infinite at the point in the tube when the mass transfer coefficient equals k,. According to this model, k , on the order of cm/s would be a useful target for a nonthrombogenic material.
INTRODUCTION
Assessing the in nitro thrombogenicity of a biomaterial frequently involves the use of a clotting test. A delay in fibrin formation, in whole blood, plasma, or a specific clotting factor mixture, is used as evidence of reduced thrombogenicity in comparison with a control material. Beyond this, however, interpretation of clotting times in more fundamental terms is virtually impossible. The presence or absence of platelets, whether the blood/plasma is fresh or recalcified, the preexposure to other materials and degree of the consequent preactivation, the coagulation initiator, the presence of an air interface, temperature, and the flow field can all affect the measured time. Furthermore “To whom correspondence should be addressed.
Journal of Biomedical Materials Research, Vol. 26, 675-693 (1992) CCC 0021-9304/92/050675-19$4.00 0 1992 John Wiley & Sons, Inc.
ROLLASON AND SEFTON
676
the precision of clotting time measurements is compromised by operatorassociated differences in detecting the endpoint, whether it be the initial appearance of fibrin strands or the complete gelation of the mixture. Although, the use of clotting factor deficient plasmas or particular arrays of clotting tests can result in more information, the resulting set of rankings is still of limited value. We encountered this problem when we tried to model the bulk and surface inactivation of thrombin in flowing blood in heparinized and nonheparinized tubes.' To calculate the concentration of thrombin at the wall of a tube under different flow or inactivation conditions, we required the rate at which thrombin was generated at the surface. This parameter was not available. Instead it was estimated from platelet adhesion values and thrombin turnover kinetics or from an assumption of the time in which threshold levels of thrombin (-0.5 Fg/mL) would be developed in a glass tube. Clotting times (whole blood or plasma recalcification), although well known were not helpful. The estimated thrombin production rate was unfortunately the weakest part of our analysis. The definition of thrombogenicity established at the Chester, UK Consensus Conference for Biomaterials, was: "the ability of a material to induce or promote the formation of thrombus."2Much of the associated discussion2revolved about the implicit assumption that such an ability i s a kinetic parameter, i.e., it is the vnte at which thromboeinboli are produced that is the key parameter. Furthermore, the intrinsic property of the material has to be distinguished from the apparent thrombogenicity that might reflect the influence of flow on mass transfer toward or away from the surface. The difference between platelet and clotting factor based thrombogenicity is also important. Here we describe the development of a new biomaterial assay system in which we measure the rate at which thrombin is generated by contact between plasma and the test material. The resulting parameter is consistent with the proposed definition of thrombogenicity and represents a more fundamental characteristic of the material. As such it can be used in a modified version3 of the thrombin inactivation model that was published previously. Thrombin generation (i,e., prothrombin conversion) has been measured by Hemker et al."' and used to further understand the mechanism of action of antithrombin I11 and heparin.' In these studies, Hemker et al. devised a computer-assisted method for correcting the apparent thrombin generation rate for the a-2-macroglobulin and antithrombin 111 associated inactivation, so as to determine the true rate of prothrombin conversion and hence the effect of heparin on the prothrombinase complex; such correction was not attempted here. KINETIC EXPRESSIONS
Thrombin is produced by activation of prothrombin through the action of the platelet-bound prothrombinase complex. The prothrombinase complex is in turn produced by initiation of the intrinsic coagulation cascade and activa-
THROMBIN PRODUCTION IN HUMAN PLASMA
677
tion of the appropriate clotting factors, along with presentation of the appropriate platelet binding sites. Thrombin is known to be a potent platelet activating agent and therefore thrombin production is autoaccelerated through the enhancement of the prothrombinase complex. Even in the absence of platelets, thrombin production can be accelerated through the effect of thrombin on other clotting factors. The detailed kinetics are complex, but can be simplified by assuming that thrombin production can be represented as an autocatalytic process first order in thrombin with an apparent rate constant k;,
where [TI = thrombin concentration in test medium (e.g., plasma) at time t . kb = first-order thrombin production rate constant (s-’) t = time(s) kb as defined by Eq. (1) is dependent on the surface to volume ratio of the experimental system, since it is assumed that all the prothrombinase complex is present at the test material surface. Hence larger surface-to-volume ratios are expected to increase the rate at which thrombin is generated. Thus a mass balance on thrombin in particular experimental systems gives:
V-d[ TI = k,A[T] dt
where k , = intrinsic thrombin production rate constant (cm/s) A = exposed area (cm’) V = volume of test mixture (cm’). Comparison of (1)and (2) yields
k , = kb(V/A)
(3)
For a surface reaction system k, (and not k b ) is the fundamental quantity, which is expected to be independent of surface-to-volume ratio. In our studies thrombin concentration was measured either by aliquot sampling at periodic intervals for subsequent measurement by f luorogenic substrate assay or by simultaneous measurement of fluorescence produced by action of the thrombin on a fluorogenic substrate present at the beginning of the plasma/material incubation (t = 0). In the former case (aliquot measurement), integration of Eq. (2) yields
where [T]” is the concentration of thrombin at t = 0. At least a trace of thrombin is needed initially to begin the autocatalytic process assumed here; its actual value is unimportant. A semilogarithmic plot of [TI against time is expected to be linear with a slope of k p A / V In the second case, the fluorescence produced by the as-generated thrombin is monitored. The instantaneous rate of fluorescence production is related directly to thrombin concentration through a calibration curve which
ROLLASON AND SEFTON
678
is linear over at least some portion of the thrombin concentration region, provided the fluorogenic substrate is in excess. Hence:
dF = B [ T ] + Fo dt
(5)
where F is the fluorescence at time t ; Fo is the [TI = 0 intercept of the calibration curve and B is the appropriate calibration constant. Combining Eq. (5) with Eq. (4) indicates that a semilogarithmic plot of the instantaneous slope of the fluorescence production curve (rate of fluorescence increase) against time is also expected to be linear with a slope of k,A/V (independent of B). MATERIALS A N D METHODS
Materials
Test materials were used in the form of fluorescence cuvettes or capped tubing segments. Cuvettes (12 mm I.D. x 7.5 mm) were glass (SciCan, Cobourg, Ont.), polystyrene and polypropylene (both Falcon, Oxford, CA). None of these were subject to bulk or surface analysis, so no detailed attempt was made to attribute the thrombin production rate to the surface properties of these materials. Polyethylene (PE) tubes, coated or not, (9.6 mm I.D. x 60 mm) were prepared from low-density tubing (Cole Parmer, Chicago, IL) by capping the ends with a 9.6-mm diameter x 2 mm thick disk of low-density PE (Warehouse Mastics, Toronto). Tubes were coated with polyvinyl alcohol (PVA, 7.6% w/w PVA) and heparin-PVA hydrogels (2% w/w heparin) as described e l s e ~ h e r eCapped .~ tubes, af ter chromic acid etching and glow discharge cleaning, were filled with gel solution and then allowed to drain freely after inversion and dried overnight; this was repeated once. Plasma preparation
Pooled citrated human plasma was prepared by mixing equal volumes of three lots of fresh frozen plasma (Canadian Red Cross, Toronto). Aliquots (0.5 mL) were incubated in the test material tube for 2 min at 37"C, after which 1.0 mL of rabbit brain cephalin (Sigma) in 0.85% (w/w) saline was added and the sample incubated for 6 min at 37°C. Platelet rich plasma (PRP) was prepared by centrifugation of freshly drawn citrated human blood and diluted with 1 mL of saline without cephalin. Thrombin production
(a) Cumulative f luovescence method A fluorogenic substrate, BOC-val-pro-arg-7-amido-4-methylcoumarin (BMCA, Sigma, St. Louis, MO) previously dissolved in DMSO (20 pL, 10 mM)
THROMBIN PRODUCTION IN HUMAN PLASMA
679
was added to the test plasma. This was followed by the addition of 0.5 mL of 0.025M CaCl,, after which the sample tube was placed in the cell holder of the spectrofluorometer (SPEX Nova, Baird Atomic, Braintree, UK). The increase in fluorescence was recorded every 30-60 s until a plateau was reached. Five replicates were performed for each test material. All measurements were made at room temperature. BMCA is specific for thrombin with a of that for thrombin and no activity for Factor Xa or plasmin activity 4% kallikrein or urokinase.x The fluorescence was plotted against time and the slope of the curve was calculated as the secants between data points. The fluorescence was determined at different instrument sensitivities for different parts of the curve and then adjusted to a common sensitivity so a single curve could be drawn. A small correction factor (0.96-1.18) for different materials was estimated by comparing the fluorescence at the end of a thrombin production experiment to that for the same mixture in a polystyrene cuvette. The slopes (dF/dt) were converted to thrombin concentrations using a calibration constant and plotted against time to yield the thrombin production curve. (b) Aliquot method Here the CaClz was added to the plasma/cephalin mixture in the test tube without BMCA. Aliquots (0.1 mL) were removed periodically during clotting and added to a polystyrene cuvette containing 2.0 mL of a pH 7.5 solution of 50 mM Tris, 1.3 mM EDTA, 0.7% polyethylene glycol 8000 (to minimize thrombin adsorption) and 0.01 mM BMCA (previously dissolved in DMSO). The production of fluorescence was measured over time in the standard fashion and the initial rate of fluorescence production was used to determine the thrombin concentrations using the standard curve. All measurements were made at room temperature. The clotting time was also recorded and used to create a normalized time scale, to compensate for the variability in clotting time. The difference between the actual clotting time and aliquot time was calculated (At,) and then subtracted from the mean clotting time to give the normalized aliquot time. The thrombin concentration was then plotted against this normalized time.
Flow system To determine the effect of flow (and surface-to-volume ratio) on the measured thrombin production rate, a flow system was devised (Fig. l).A peristaltic pump (Orion Sage 375A, 1.9 mL/min or Masterflex, Cole-Parmer, 15 mL/min) was used to circulate the plasma/BMCA mixture through a 119cm length of Intramedic PE tubing (PE 160,1.14 mm I.D., Clay-Adams, Parsippany, NJ), a 27-cm length of Silastic tubing (1 mm I.D.) and a PE flowcell (made from a 50-mm length of 9.6-mm I.D. PE tubing). The total surface area was 70.1 cm’, of which 88% was PE; the volume was 5 mL so that the total PE
680
ROLLASON AND SEFTON
polyethylene tubing
Silastic
J
tubing
7 I
polyethylene "cuvette"
Figure 1. Schematic illustration of flow system for measurement of thrombin production rate.
surface to volume ratio was 12.3 cm-'. Plasma (1.25 mL) and cephalin (2.5 mL) were incubated within the tubing/flow cell at 37°C after which the flowcell was placed in the sample beam of the spectrofluorometer. CaClz (1.25 mL) and BMCA (50 pL, 10 mM) were injected, the tubing was connected and the pump was started to mix the reactants, at which time the increase in fluorescence was monitored.
Standard curves Standard curves relating the rate of fluorescence production to thrombin concentration were determined by adding 20 p L thrombin (purified bovine, 1500-2500 NJH units/mg, Sigma) at different concentrations in PBS to polystyrene cuvettes containing 2 mL of 50 mM Tris, 1.3 mM EDTA, 0.7% PEG 8000, and 0.1 mM BMCA. Errors in calibration constant were of minimal importance since the slopes of log [TI against t plots (k,) were independent of the absolute thrombin concentration. RESULTS
Cumulative fluorescence The cumulative fluorescence curves are shown in Figure 2. The data cover more than three orders of magnitude in fluorescence values, but this is masked by the linear scale used in Figure 2. The slopes of each segment of
THROMBIN PRODUCTION IN HUMAN PLASMA
681
2 0 [ " " "T T" ' ' ' ' ' I 1 I I ~
Time (minutes)
Figure 2. Cumulative production of fluorescence by thrombin amidolysis of BMCA in recalcified cephalin enhanced plasma in contact with different materials in the absence of flow. 0,glass; 0 , polypropylene; V,polystyrene; T, polyethylene; 0,PVA; B,heparin PVA. Mean fluorescence +f - standard deviation ( n = 5). All data obtained using same pool of plasma. Data corrected for the effect of material on absolute fluorescence and adjusted to a common instrument sensitivity.
the curve (the secant) was calculated from the data, converted to thrombin concentration using the calibration constant, and plotted against the midpoints of the appropriate time intervals to generate the thrombin production curves (Fig. 3). Thrombin concentration increased slowly and then more rapidly as more thrombin was produced to augment its own production rate. However, simultaneous with production it was being inactivated by reaction with antithrombin I11 so that eventually its concentration decreased as the rate of inactivation became greater than the rate of production. Eventually the plasma presumably became exhausted of prothrombin and production was essentially turned off. This was well after a clot had already formed. The mixture was not substrate limited (BMCA) since adding more substrate did not result in increased fluorescence. It is interesting to note that for heparinPVA, [TI was essentially constant at 0.99), but slightly less so for PVA and PE ( r > 0.96). The initial slopes were used to calculate the first-order rate constants according to Eq. (4) as listed in Table I. For these results, V = 2 mL and A was calculated to be 9.3 cmz for the cuvettes and 9.0 cm2for PE and PVA, assuming that only the surface in contact with the stagnant test fluid was involved in thrombin production. In Figure 2, the results were obtained by averaging five fluorescence values (from separate experiments) at each time point, and then calculating a single slope for Figures 3. Individual curves where the data were not averaged prior to conversion to thrombin concentration were still reasonably linear (as semilogarithmic plots) although the correlation constants were not as high as those based on averaged concentrations. The high variability among individual curves for PE accounts for the large error bars seen in Figure 2. Repeating the PE results with a different pooled plasma gave more reproducible results although the k , value was slightly higher
THROMBIN PRODUCTION IN HUMAN PLASMA
683
TABLE I Thrombin Production Rate Constants (Without Flow)
Material Cephalin enhanced plasma Glass Polypropylene Polystyrene Polye thylene PVA Heparin-PVA Platelet rich plasma Polyethylene PVA
Cumulative (Average F)"
Cumulative (Separate F)b
7.92 3.07 3.24 2.92 3.64 2.87
8.50 3.63 3.89 2.78 3.06 2.73
5 0.98 2 0.09
0.09 0.35 5 0.23" & 0.25 ? &
+- 0.65 5 0.42 0.16 5 0.86 4 0.28d 5 0.34
*
Aliquot' 9.08 2 2.09 5.24 2 0.72 6.28 ? 0.98
Not detectable 5.27 5 0.48 2.97 2 0.29
"Mean fluorescence value ( n = 5) plotted against t and slopes used to calculate k,; standard deviation from linear correlation. hFluorescenceplotted against t for each replicate run and slopes used to calculate k , which were then averaged ( n = 5); & standard deviation of replicate runs. 'Thrombin concentration plotted against a normalized time to correct for difference in clotting time; 5 standard deviation from correlation. "Second value for PE obtained using a different pooled plasma than all the other values. _t
(Table I). The mean rate constants calculated from the individual curves are also listed in Table I. The larger standard deviations for the rate constants determined from the separate slopes (except for glass), reflects the difference between intersample variation and that due to the correlation. For glass, only four points in the linear region were obtained leading to a poorer correlation than the others. Using PRP (Fig. 4), the production rate constants (Table I) were similar for PVA and twofold higher for PE, relative to those determined in plasma with cephalin. The difference for PE but not for PVA presumably reflects differences in the extent of platelet adhesion or platelet involvement in thrombin production with the two surfaces. Aliquot method Results from the aliquot method are shown in Figure 5 as the semilogarithmic plot of thrombin concentration against normalized time. The thrombin concentrations were much higher here. The resulting thrombin production rates are listed in Table I, and they were larger (but with poorer correlation) than those determined by the cumulative method. Because of experimental constraints, only a limited number (3-4) of aliquots could be taken from each plasmahube sample. Therefore a full thrombin production curve had to be composed of aliquots taken at different times from different samples, each with a slightly different clotting time. Normalizing the time scale with respect to the clotting time lead to greater conformity in the data and better correlation coefficients.
ROLLASON AND SEFTON
684 25
20 G 0
zx c
.-0
15
3
TI
g
n
8c
P
10
g! 0
3
ic
/ i
5
/a
0 0
2
4
6
8
10
I
I
12
14
Time (minutes)
(a)
Figure 4. Cumulative thrombin production. Production of fluorescence by platelet rich plasma in contact with polyethylene ( 0 ) or PVA (V)without flow. (a) Cumulative mean fluorescence (? SD, n = 5), (b) thrombin concentration.
Effect of flow Using the flow system at two different flow rates, corresponding to wall shear rates of 210 and 1700 s-', gave rise to semilogarithmic plots of thrombin concentration against time with similar slopes [Fig. 6(b) and Table I]. However, the k, values were about half those determined without flow, although the k; were not much different (Table 11). The difference in thrombin concentrations at the two flow rates (although not initial rate constants) presumably reflects differences in mixing, although why the higher flow rate should lead to poorer mixing is not clear; perhaps the two different pumps that were used played a role.
DISCUSSION
Two methods were devised for measuring the time course of thrombin production in plasma (diluted 1:4) in contact with different materials. The
THROMBIN PRODUCTION IN HUMAN PLASMA
685
I 10
1
Y
c
.-
0
c
2 c. c Q
0
0.1
c
0
0
c .-
n E E c
k
0.01
0.001
I
I
0
2
I
I
4
6
I
I
I
I
8
10
12
14
Time (minutes)
(b) Figure 4. (continued)
aliquot method, devised originally, has the advantage that the material need not be in a form of a fluorescence-transparent material or in a form that can be placed in the sample holder of a spectrofluorometer. Unfortunately, many more samples need to be run and the time scale corrected in order to enhance the conformity of the data. The cumulative method is preferred because of the higher level of precision, although material preparation (e.g., coating onto a polyethylene tube) is more involved. Wojiechowski and Brash' have used this method recently in a study of silanized glasses. The data manipulation required with the cumulative method can easily be automated with the appropriate digital interface to the spectrofluorometer. To some extent, a major assumption relates to the calibration. For the cumulative method, it is necessary to assume that the thrombin activity toward f luorogenic substrate is the same in tris-EDTA as it is in recalcified, phospholipid enhanced, clotting plasma, or PRP. It is implicitly assumed that the calibration curve is the same for "free" thrombin as well as for thrombin bound to fibrinogen or the other components in clotting plasma."' It is also assumed that the fluorogenic substrate activity of surface adsorbed thrombin is the same as for solution phase thrombin or at least that the ratio of the two
686
ROLLASON AND SEFTON
Figure 5. Thrombin concentration in recalcified cephalin enhanced plasma in contact with different materials, measured by taking aliquots. Time scale was normalized to account for differences in clotting times. Lines shown are regression lines. o, glass; 0 , polypropylene; V, polystyrene.
doesn’t change from surface to surface. It should be noted, however, that the key question may be the relationship between the small-molecular-weight amidolytic activity and the fibrinogen cleavage activity.” We are not so much interested in thrombin concentrations as much as activity toward fibrinogen. If free and bound (e.g., to fibrin) thrombin have different activities toward fibrinogen, then the real assumption is that the ratio of the activities is identical with the ratio of the activities toward the f luorogenic substrate. These appear to be unavoidable assumptions. The much higher thrombin concentrations in the aliquot method may reflect this issue. In plasma much of the thrombin may be inactive or inaccessible toward BMCA because the thrombin is bound to a protein. However, this bound, hitherto inactive, thrombin may be released upon dilution in Tris/ EDTA, enabling it to be accounted for in the aliquot method but not in the cumulative method. On the other hand, calibration may not be quite as important as it appears. Provided all calibration curves are linear over the appropriate concentration ranges, the particular calibration constant is not important in terms of the final calculated k , value. Only the actual thrombin concentration is affected while the slope of the semilogarithmic plot is inde-
687
THROMBIN PRODUCTION IN HUMAN PLASMA
pendent of calibration constant. Hence, the similarity in k , values despite the large difference in thrombin concentration between the cumulative and aliquot methods is not surprising. The small difference in the two could easily be accounted for by a nonconstant ratio between amidolytic activity in plasma and in tris/EDTA or by other methodological differences. Thrombin production was measured classically by aliquot sampling and clotting time measurement of thrombin concentration,” but without the sensitivity to examine the early stages of production. Hemker’s method‘ differs from that used here in that thrombin production at 37°C in defibrinated plasma initiated by brain thromboplastin was monitored by taking aliquots, using a chromogenic substrate (S2238) to measure thrombin concentration. The data were further analyzed to separate thrombin production from thrombin inactivation, in order to discern the effect of heparin on thrombin production. In initial experiments (without thromboplastin) we found that a chromogenic substrate/aliquot method was only usable for glass. For the plastics studied here, the rate of thrombin production was too low to produce I
I
I
I
I
5
6
7
8
16 14
-
s 0 2 X
Y
C .-c 0 0
3
-0
en 8 c 8 tn 2 0
2
i
0
1
2
3
4
9
Time (minutes)
(4 Figure 6. Thrombin production in the flow system in recalcified cephalin enhanced plasma in contact with polyethylene (a) cumulative mean fluorescence (k SD, n = 5), (b) thrombin concentration. Note similarity of slopes in (b).
10
ROLLASON A N D SEFTON 10
1
0.1
0.01
0.001
0
I
I
I
I
I
1
I
I
I
1
2
3
4
5
6
7
8
9
10
k,(cm/s x
lo-')
Time (minutes) (b)
Figure 6. (continued) TABLE I1 Thrombin Production Rate Constants (with Flow) Polyethylene/cephalin Enhanced Plasma Flow Rate (mL/min)
k;, (s-' x 10-3)
1.9 15 Stagnant
13.4 + 1.5 12.6 -t 2.9 13.1 ? 1.5
1.08 rt 0.13 1.04 +- 0.25 2.92 t 0.35
k standard deviation from correlation; five replicates averaged together to yield a single cumulative fluorescence curve.
detectable levels of thrombin. The f luorogenic substrate was much more sensitive so that the early part of the production curve could be quantified. The thrombin production rate constant as presented here reflects the difference between production and inactivation by antithrombin 111 and other plasma inhibitors. Hemker's results' suggest that the inactivation processes are more important at the later stages of the thrombin production curve near and beyond the maximum. Since the rate constants were largely derived from the initial part of the curve (the pseudo-first-order regime) correcting for inactivation processes would not substantially affect the calculated value. Thus
THROMBIN PRODUCTION IN HUMAN PLASMA
689
it would be necessary to account separately for inactivation processes that occur at later times. It is not clear, however, what relevance these thrombin production values have at later times, since experimentally the first-order regime applies only to the period prior to clotting. Beyond that time, thrombin is presumed to be lost to the analysis because of its incorporation in the clot, because of the inactivation processes that occur at higher rates at later times or because of the consumption of prothrombin. Neglecting inactivation processes, the thrombin concentration at the peak (aliquot method) is, however,
ene, polystyrene, polyethylene and PVA; k , - 3 . 1 x lo-’ cm/s) i n plasma w i t h cephalin without flow. A k , for heparinPVA could not be determined since the thrombin concentration was too low to be quantified. A larger difference between polyethylene and PVA was noted with platelet-rich plasma without flow while lower values (1.0 x cm/s) were noted in a flow system but at a higher surface to volume ratio. The first-order rate constant can be used i n simple models relating production of thrombin at a wall of a tube to its mass transfer away from the wall in flowing blood. One such model predicts that the concentration of thrombin at the wall should become infinite at the point in the tube when the mass transfer coefficient equals k,. According to this model, k , on the order of cm/s would be a useful target for a nonthrombogenic material.
INTRODUCTION
Assessing the in nitro thrombogenicity of a biomaterial frequently involves the use of a clotting test. A delay in fibrin formation, in whole blood, plasma, or a specific clotting factor mixture, is used as evidence of reduced thrombogenicity in comparison with a control material. Beyond this, however, interpretation of clotting times in more fundamental terms is virtually impossible. The presence or absence of platelets, whether the blood/plasma is fresh or recalcified, the preexposure to other materials and degree of the consequent preactivation, the coagulation initiator, the presence of an air interface, temperature, and the flow field can all affect the measured time. Furthermore “To whom correspondence should be addressed.
Journal of Biomedical Materials Research, Vol. 26, 675-693 (1992) CCC 0021-9304/92/050675-19$4.00 0 1992 John Wiley & Sons, Inc.
ROLLASON AND SEFTON
676
the precision of clotting time measurements is compromised by operatorassociated differences in detecting the endpoint, whether it be the initial appearance of fibrin strands or the complete gelation of the mixture. Although, the use of clotting factor deficient plasmas or particular arrays of clotting tests can result in more information, the resulting set of rankings is still of limited value. We encountered this problem when we tried to model the bulk and surface inactivation of thrombin in flowing blood in heparinized and nonheparinized tubes.' To calculate the concentration of thrombin at the wall of a tube under different flow or inactivation conditions, we required the rate at which thrombin was generated at the surface. This parameter was not available. Instead it was estimated from platelet adhesion values and thrombin turnover kinetics or from an assumption of the time in which threshold levels of thrombin (-0.5 Fg/mL) would be developed in a glass tube. Clotting times (whole blood or plasma recalcification), although well known were not helpful. The estimated thrombin production rate was unfortunately the weakest part of our analysis. The definition of thrombogenicity established at the Chester, UK Consensus Conference for Biomaterials, was: "the ability of a material to induce or promote the formation of thrombus."2Much of the associated discussion2revolved about the implicit assumption that such an ability i s a kinetic parameter, i.e., it is the vnte at which thromboeinboli are produced that is the key parameter. Furthermore, the intrinsic property of the material has to be distinguished from the apparent thrombogenicity that might reflect the influence of flow on mass transfer toward or away from the surface. The difference between platelet and clotting factor based thrombogenicity is also important. Here we describe the development of a new biomaterial assay system in which we measure the rate at which thrombin is generated by contact between plasma and the test material. The resulting parameter is consistent with the proposed definition of thrombogenicity and represents a more fundamental characteristic of the material. As such it can be used in a modified version3 of the thrombin inactivation model that was published previously. Thrombin generation (i,e., prothrombin conversion) has been measured by Hemker et al."' and used to further understand the mechanism of action of antithrombin I11 and heparin.' In these studies, Hemker et al. devised a computer-assisted method for correcting the apparent thrombin generation rate for the a-2-macroglobulin and antithrombin 111 associated inactivation, so as to determine the true rate of prothrombin conversion and hence the effect of heparin on the prothrombinase complex; such correction was not attempted here. KINETIC EXPRESSIONS
Thrombin is produced by activation of prothrombin through the action of the platelet-bound prothrombinase complex. The prothrombinase complex is in turn produced by initiation of the intrinsic coagulation cascade and activa-
THROMBIN PRODUCTION IN HUMAN PLASMA
677
tion of the appropriate clotting factors, along with presentation of the appropriate platelet binding sites. Thrombin is known to be a potent platelet activating agent and therefore thrombin production is autoaccelerated through the enhancement of the prothrombinase complex. Even in the absence of platelets, thrombin production can be accelerated through the effect of thrombin on other clotting factors. The detailed kinetics are complex, but can be simplified by assuming that thrombin production can be represented as an autocatalytic process first order in thrombin with an apparent rate constant k;,
where [TI = thrombin concentration in test medium (e.g., plasma) at time t . kb = first-order thrombin production rate constant (s-’) t = time(s) kb as defined by Eq. (1) is dependent on the surface to volume ratio of the experimental system, since it is assumed that all the prothrombinase complex is present at the test material surface. Hence larger surface-to-volume ratios are expected to increase the rate at which thrombin is generated. Thus a mass balance on thrombin in particular experimental systems gives:
V-d[ TI = k,A[T] dt
where k , = intrinsic thrombin production rate constant (cm/s) A = exposed area (cm’) V = volume of test mixture (cm’). Comparison of (1)and (2) yields
k , = kb(V/A)
(3)
For a surface reaction system k, (and not k b ) is the fundamental quantity, which is expected to be independent of surface-to-volume ratio. In our studies thrombin concentration was measured either by aliquot sampling at periodic intervals for subsequent measurement by f luorogenic substrate assay or by simultaneous measurement of fluorescence produced by action of the thrombin on a fluorogenic substrate present at the beginning of the plasma/material incubation (t = 0). In the former case (aliquot measurement), integration of Eq. (2) yields
where [T]” is the concentration of thrombin at t = 0. At least a trace of thrombin is needed initially to begin the autocatalytic process assumed here; its actual value is unimportant. A semilogarithmic plot of [TI against time is expected to be linear with a slope of k p A / V In the second case, the fluorescence produced by the as-generated thrombin is monitored. The instantaneous rate of fluorescence production is related directly to thrombin concentration through a calibration curve which
ROLLASON AND SEFTON
678
is linear over at least some portion of the thrombin concentration region, provided the fluorogenic substrate is in excess. Hence:
dF = B [ T ] + Fo dt
(5)
where F is the fluorescence at time t ; Fo is the [TI = 0 intercept of the calibration curve and B is the appropriate calibration constant. Combining Eq. (5) with Eq. (4) indicates that a semilogarithmic plot of the instantaneous slope of the fluorescence production curve (rate of fluorescence increase) against time is also expected to be linear with a slope of k,A/V (independent of B). MATERIALS A N D METHODS
Materials
Test materials were used in the form of fluorescence cuvettes or capped tubing segments. Cuvettes (12 mm I.D. x 7.5 mm) were glass (SciCan, Cobourg, Ont.), polystyrene and polypropylene (both Falcon, Oxford, CA). None of these were subject to bulk or surface analysis, so no detailed attempt was made to attribute the thrombin production rate to the surface properties of these materials. Polyethylene (PE) tubes, coated or not, (9.6 mm I.D. x 60 mm) were prepared from low-density tubing (Cole Parmer, Chicago, IL) by capping the ends with a 9.6-mm diameter x 2 mm thick disk of low-density PE (Warehouse Mastics, Toronto). Tubes were coated with polyvinyl alcohol (PVA, 7.6% w/w PVA) and heparin-PVA hydrogels (2% w/w heparin) as described e l s e ~ h e r eCapped .~ tubes, af ter chromic acid etching and glow discharge cleaning, were filled with gel solution and then allowed to drain freely after inversion and dried overnight; this was repeated once. Plasma preparation
Pooled citrated human plasma was prepared by mixing equal volumes of three lots of fresh frozen plasma (Canadian Red Cross, Toronto). Aliquots (0.5 mL) were incubated in the test material tube for 2 min at 37"C, after which 1.0 mL of rabbit brain cephalin (Sigma) in 0.85% (w/w) saline was added and the sample incubated for 6 min at 37°C. Platelet rich plasma (PRP) was prepared by centrifugation of freshly drawn citrated human blood and diluted with 1 mL of saline without cephalin. Thrombin production
(a) Cumulative f luovescence method A fluorogenic substrate, BOC-val-pro-arg-7-amido-4-methylcoumarin (BMCA, Sigma, St. Louis, MO) previously dissolved in DMSO (20 pL, 10 mM)
THROMBIN PRODUCTION IN HUMAN PLASMA
679
was added to the test plasma. This was followed by the addition of 0.5 mL of 0.025M CaCl,, after which the sample tube was placed in the cell holder of the spectrofluorometer (SPEX Nova, Baird Atomic, Braintree, UK). The increase in fluorescence was recorded every 30-60 s until a plateau was reached. Five replicates were performed for each test material. All measurements were made at room temperature. BMCA is specific for thrombin with a of that for thrombin and no activity for Factor Xa or plasmin activity 4% kallikrein or urokinase.x The fluorescence was plotted against time and the slope of the curve was calculated as the secants between data points. The fluorescence was determined at different instrument sensitivities for different parts of the curve and then adjusted to a common sensitivity so a single curve could be drawn. A small correction factor (0.96-1.18) for different materials was estimated by comparing the fluorescence at the end of a thrombin production experiment to that for the same mixture in a polystyrene cuvette. The slopes (dF/dt) were converted to thrombin concentrations using a calibration constant and plotted against time to yield the thrombin production curve. (b) Aliquot method Here the CaClz was added to the plasma/cephalin mixture in the test tube without BMCA. Aliquots (0.1 mL) were removed periodically during clotting and added to a polystyrene cuvette containing 2.0 mL of a pH 7.5 solution of 50 mM Tris, 1.3 mM EDTA, 0.7% polyethylene glycol 8000 (to minimize thrombin adsorption) and 0.01 mM BMCA (previously dissolved in DMSO). The production of fluorescence was measured over time in the standard fashion and the initial rate of fluorescence production was used to determine the thrombin concentrations using the standard curve. All measurements were made at room temperature. The clotting time was also recorded and used to create a normalized time scale, to compensate for the variability in clotting time. The difference between the actual clotting time and aliquot time was calculated (At,) and then subtracted from the mean clotting time to give the normalized aliquot time. The thrombin concentration was then plotted against this normalized time.
Flow system To determine the effect of flow (and surface-to-volume ratio) on the measured thrombin production rate, a flow system was devised (Fig. l).A peristaltic pump (Orion Sage 375A, 1.9 mL/min or Masterflex, Cole-Parmer, 15 mL/min) was used to circulate the plasma/BMCA mixture through a 119cm length of Intramedic PE tubing (PE 160,1.14 mm I.D., Clay-Adams, Parsippany, NJ), a 27-cm length of Silastic tubing (1 mm I.D.) and a PE flowcell (made from a 50-mm length of 9.6-mm I.D. PE tubing). The total surface area was 70.1 cm’, of which 88% was PE; the volume was 5 mL so that the total PE
680
ROLLASON AND SEFTON
polyethylene tubing
Silastic
J
tubing
7 I
polyethylene "cuvette"
Figure 1. Schematic illustration of flow system for measurement of thrombin production rate.
surface to volume ratio was 12.3 cm-'. Plasma (1.25 mL) and cephalin (2.5 mL) were incubated within the tubing/flow cell at 37°C after which the flowcell was placed in the sample beam of the spectrofluorometer. CaClz (1.25 mL) and BMCA (50 pL, 10 mM) were injected, the tubing was connected and the pump was started to mix the reactants, at which time the increase in fluorescence was monitored.
Standard curves Standard curves relating the rate of fluorescence production to thrombin concentration were determined by adding 20 p L thrombin (purified bovine, 1500-2500 NJH units/mg, Sigma) at different concentrations in PBS to polystyrene cuvettes containing 2 mL of 50 mM Tris, 1.3 mM EDTA, 0.7% PEG 8000, and 0.1 mM BMCA. Errors in calibration constant were of minimal importance since the slopes of log [TI against t plots (k,) were independent of the absolute thrombin concentration. RESULTS
Cumulative fluorescence The cumulative fluorescence curves are shown in Figure 2. The data cover more than three orders of magnitude in fluorescence values, but this is masked by the linear scale used in Figure 2. The slopes of each segment of
THROMBIN PRODUCTION IN HUMAN PLASMA
681
2 0 [ " " "T T" ' ' ' ' ' I 1 I I ~
Time (minutes)
Figure 2. Cumulative production of fluorescence by thrombin amidolysis of BMCA in recalcified cephalin enhanced plasma in contact with different materials in the absence of flow. 0,glass; 0 , polypropylene; V,polystyrene; T, polyethylene; 0,PVA; B,heparin PVA. Mean fluorescence +f - standard deviation ( n = 5). All data obtained using same pool of plasma. Data corrected for the effect of material on absolute fluorescence and adjusted to a common instrument sensitivity.
the curve (the secant) was calculated from the data, converted to thrombin concentration using the calibration constant, and plotted against the midpoints of the appropriate time intervals to generate the thrombin production curves (Fig. 3). Thrombin concentration increased slowly and then more rapidly as more thrombin was produced to augment its own production rate. However, simultaneous with production it was being inactivated by reaction with antithrombin I11 so that eventually its concentration decreased as the rate of inactivation became greater than the rate of production. Eventually the plasma presumably became exhausted of prothrombin and production was essentially turned off. This was well after a clot had already formed. The mixture was not substrate limited (BMCA) since adding more substrate did not result in increased fluorescence. It is interesting to note that for heparinPVA, [TI was essentially constant at 0.99), but slightly less so for PVA and PE ( r > 0.96). The initial slopes were used to calculate the first-order rate constants according to Eq. (4) as listed in Table I. For these results, V = 2 mL and A was calculated to be 9.3 cmz for the cuvettes and 9.0 cm2for PE and PVA, assuming that only the surface in contact with the stagnant test fluid was involved in thrombin production. In Figure 2, the results were obtained by averaging five fluorescence values (from separate experiments) at each time point, and then calculating a single slope for Figures 3. Individual curves where the data were not averaged prior to conversion to thrombin concentration were still reasonably linear (as semilogarithmic plots) although the correlation constants were not as high as those based on averaged concentrations. The high variability among individual curves for PE accounts for the large error bars seen in Figure 2. Repeating the PE results with a different pooled plasma gave more reproducible results although the k , value was slightly higher
THROMBIN PRODUCTION IN HUMAN PLASMA
683
TABLE I Thrombin Production Rate Constants (Without Flow)
Material Cephalin enhanced plasma Glass Polypropylene Polystyrene Polye thylene PVA Heparin-PVA Platelet rich plasma Polyethylene PVA
Cumulative (Average F)"
Cumulative (Separate F)b
7.92 3.07 3.24 2.92 3.64 2.87
8.50 3.63 3.89 2.78 3.06 2.73
5 0.98 2 0.09
0.09 0.35 5 0.23" & 0.25 ? &
+- 0.65 5 0.42 0.16 5 0.86 4 0.28d 5 0.34
*
Aliquot' 9.08 2 2.09 5.24 2 0.72 6.28 ? 0.98
Not detectable 5.27 5 0.48 2.97 2 0.29
"Mean fluorescence value ( n = 5) plotted against t and slopes used to calculate k,; standard deviation from linear correlation. hFluorescenceplotted against t for each replicate run and slopes used to calculate k , which were then averaged ( n = 5); & standard deviation of replicate runs. 'Thrombin concentration plotted against a normalized time to correct for difference in clotting time; 5 standard deviation from correlation. "Second value for PE obtained using a different pooled plasma than all the other values. _t
(Table I). The mean rate constants calculated from the individual curves are also listed in Table I. The larger standard deviations for the rate constants determined from the separate slopes (except for glass), reflects the difference between intersample variation and that due to the correlation. For glass, only four points in the linear region were obtained leading to a poorer correlation than the others. Using PRP (Fig. 4), the production rate constants (Table I) were similar for PVA and twofold higher for PE, relative to those determined in plasma with cephalin. The difference for PE but not for PVA presumably reflects differences in the extent of platelet adhesion or platelet involvement in thrombin production with the two surfaces. Aliquot method Results from the aliquot method are shown in Figure 5 as the semilogarithmic plot of thrombin concentration against normalized time. The thrombin concentrations were much higher here. The resulting thrombin production rates are listed in Table I, and they were larger (but with poorer correlation) than those determined by the cumulative method. Because of experimental constraints, only a limited number (3-4) of aliquots could be taken from each plasmahube sample. Therefore a full thrombin production curve had to be composed of aliquots taken at different times from different samples, each with a slightly different clotting time. Normalizing the time scale with respect to the clotting time lead to greater conformity in the data and better correlation coefficients.
ROLLASON AND SEFTON
684 25
20 G 0
zx c
.-0
15
3
TI
g
n
8c
P
10
g! 0
3
ic
/ i
5
/a
0 0
2
4
6
8
10
I
I
12
14
Time (minutes)
(a)
Figure 4. Cumulative thrombin production. Production of fluorescence by platelet rich plasma in contact with polyethylene ( 0 ) or PVA (V)without flow. (a) Cumulative mean fluorescence (? SD, n = 5), (b) thrombin concentration.
Effect of flow Using the flow system at two different flow rates, corresponding to wall shear rates of 210 and 1700 s-', gave rise to semilogarithmic plots of thrombin concentration against time with similar slopes [Fig. 6(b) and Table I]. However, the k, values were about half those determined without flow, although the k; were not much different (Table 11). The difference in thrombin concentrations at the two flow rates (although not initial rate constants) presumably reflects differences in mixing, although why the higher flow rate should lead to poorer mixing is not clear; perhaps the two different pumps that were used played a role.
DISCUSSION
Two methods were devised for measuring the time course of thrombin production in plasma (diluted 1:4) in contact with different materials. The
THROMBIN PRODUCTION IN HUMAN PLASMA
685
I 10
1
Y
c
.-
0
c
2 c. c Q
0
0.1
c
0
0
c .-
n E E c
k
0.01
0.001
I
I
0
2
I
I
4
6
I
I
I
I
8
10
12
14
Time (minutes)
(b) Figure 4. (continued)
aliquot method, devised originally, has the advantage that the material need not be in a form of a fluorescence-transparent material or in a form that can be placed in the sample holder of a spectrofluorometer. Unfortunately, many more samples need to be run and the time scale corrected in order to enhance the conformity of the data. The cumulative method is preferred because of the higher level of precision, although material preparation (e.g., coating onto a polyethylene tube) is more involved. Wojiechowski and Brash' have used this method recently in a study of silanized glasses. The data manipulation required with the cumulative method can easily be automated with the appropriate digital interface to the spectrofluorometer. To some extent, a major assumption relates to the calibration. For the cumulative method, it is necessary to assume that the thrombin activity toward f luorogenic substrate is the same in tris-EDTA as it is in recalcified, phospholipid enhanced, clotting plasma, or PRP. It is implicitly assumed that the calibration curve is the same for "free" thrombin as well as for thrombin bound to fibrinogen or the other components in clotting plasma."' It is also assumed that the fluorogenic substrate activity of surface adsorbed thrombin is the same as for solution phase thrombin or at least that the ratio of the two
686
ROLLASON AND SEFTON
Figure 5. Thrombin concentration in recalcified cephalin enhanced plasma in contact with different materials, measured by taking aliquots. Time scale was normalized to account for differences in clotting times. Lines shown are regression lines. o, glass; 0 , polypropylene; V, polystyrene.
doesn’t change from surface to surface. It should be noted, however, that the key question may be the relationship between the small-molecular-weight amidolytic activity and the fibrinogen cleavage activity.” We are not so much interested in thrombin concentrations as much as activity toward fibrinogen. If free and bound (e.g., to fibrin) thrombin have different activities toward fibrinogen, then the real assumption is that the ratio of the activities is identical with the ratio of the activities toward the f luorogenic substrate. These appear to be unavoidable assumptions. The much higher thrombin concentrations in the aliquot method may reflect this issue. In plasma much of the thrombin may be inactive or inaccessible toward BMCA because the thrombin is bound to a protein. However, this bound, hitherto inactive, thrombin may be released upon dilution in Tris/ EDTA, enabling it to be accounted for in the aliquot method but not in the cumulative method. On the other hand, calibration may not be quite as important as it appears. Provided all calibration curves are linear over the appropriate concentration ranges, the particular calibration constant is not important in terms of the final calculated k , value. Only the actual thrombin concentration is affected while the slope of the semilogarithmic plot is inde-
687
THROMBIN PRODUCTION IN HUMAN PLASMA
pendent of calibration constant. Hence, the similarity in k , values despite the large difference in thrombin concentration between the cumulative and aliquot methods is not surprising. The small difference in the two could easily be accounted for by a nonconstant ratio between amidolytic activity in plasma and in tris/EDTA or by other methodological differences. Thrombin production was measured classically by aliquot sampling and clotting time measurement of thrombin concentration,” but without the sensitivity to examine the early stages of production. Hemker’s method‘ differs from that used here in that thrombin production at 37°C in defibrinated plasma initiated by brain thromboplastin was monitored by taking aliquots, using a chromogenic substrate (S2238) to measure thrombin concentration. The data were further analyzed to separate thrombin production from thrombin inactivation, in order to discern the effect of heparin on thrombin production. In initial experiments (without thromboplastin) we found that a chromogenic substrate/aliquot method was only usable for glass. For the plastics studied here, the rate of thrombin production was too low to produce I
I
I
I
I
5
6
7
8
16 14
-
s 0 2 X
Y
C .-c 0 0
3
-0
en 8 c 8 tn 2 0
2
i
0
1
2
3
4
9
Time (minutes)
(4 Figure 6. Thrombin production in the flow system in recalcified cephalin enhanced plasma in contact with polyethylene (a) cumulative mean fluorescence (k SD, n = 5), (b) thrombin concentration. Note similarity of slopes in (b).
10
ROLLASON A N D SEFTON 10
1
0.1
0.01
0.001
0
I
I
I
I
I
1
I
I
I
1
2
3
4
5
6
7
8
9
10
k,(cm/s x
lo-')
Time (minutes) (b)
Figure 6. (continued) TABLE I1 Thrombin Production Rate Constants (with Flow) Polyethylene/cephalin Enhanced Plasma Flow Rate (mL/min)
k;, (s-' x 10-3)
1.9 15 Stagnant
13.4 + 1.5 12.6 -t 2.9 13.1 ? 1.5
1.08 rt 0.13 1.04 +- 0.25 2.92 t 0.35
k standard deviation from correlation; five replicates averaged together to yield a single cumulative fluorescence curve.
detectable levels of thrombin. The f luorogenic substrate was much more sensitive so that the early part of the production curve could be quantified. The thrombin production rate constant as presented here reflects the difference between production and inactivation by antithrombin 111 and other plasma inhibitors. Hemker's results' suggest that the inactivation processes are more important at the later stages of the thrombin production curve near and beyond the maximum. Since the rate constants were largely derived from the initial part of the curve (the pseudo-first-order regime) correcting for inactivation processes would not substantially affect the calculated value. Thus
THROMBIN PRODUCTION IN HUMAN PLASMA
689
it would be necessary to account separately for inactivation processes that occur at later times. It is not clear, however, what relevance these thrombin production values have at later times, since experimentally the first-order regime applies only to the period prior to clotting. Beyond that time, thrombin is presumed to be lost to the analysis because of its incorporation in the clot, because of the inactivation processes that occur at higher rates at later times or because of the consumption of prothrombin. Neglecting inactivation processes, the thrombin concentration at the peak (aliquot method) is, however,
