A comparison of artificially-depleted, lyophilized coumarin and fresh coumarin plasmas in thromboplastin calibration


Professor LeonPoller Department of Pathological Sciences, University of Manchester, Stopford Building, Oxford Road, Manchester M13 9PT.


Artificially-depleted lyophilized plasmas and lyophilized coumarin plasmas were prepared and compared with fresh coumarin plasmas to assess their comparative reliability in local thromboplastin calibration using the manual prothrombin time (PT) technique. Their certified PT values were inserted in turn on the vertical axis in place of the PT obtained with fresh coumarin plasmas. PT results were obtained at eight ECAA national laboratories (‘test centres’) and inserted on the horizontal axis. The resulting thromboplastin calibration slopes were compared with conventional fresh coumarin plasma calibration slopes at the same ‘test centres’. When 60 artificially-depleted plasmas were substituted for 60 fresh plasmas, the mean calibration slopes with the human plain International Reference Preparation (IRP) were 4.2% higher. For comparison with 20 lyophilized coumarins, three sets of 20 artificially-depleted plasmas were selected in sequential order from the 60. The lyophilized coumarin plasmas gave a mean deviation of 9.6% from the fresh plasma calibration slopes compared with values of 2.0%, 6.1% and 11.7% for the three sets of 20 depleted plasmas.

Although both types of lyophilized plasma calibration slopes give measurable differences from conventional fresh plasmas, these may be regarded as acceptable in clinical terms.

Prothrombin time (PT) standardization using International Normalized Ratios (INR) ( W.H.O. Expert Committee on Biological Standardisation, 1983) is based on the manual PT technique. Coagulometers have been shown to affect the International Sensitivity Index (ISI) of thromboplastin reagents on which the INR are based ( D'Angelo et al, 1989 ; Poller et al, 1989 ; van Rijn et al, 1989 ; Ray & Smith, 1990; Clarke et al, 1992 ). Lyophilized plasmas certified with the manual PT values in terms of International Reference Preparations (IRP) for thromboplastin have been proposed for this purpose as local ISI calibrants ( Clarke et al, 1992 ; van den Besselaar, 1994). Local ISI calibration of coagulometer systems using lyophilized plasmas artificially depleted of vitamin K dependent clotting factors in collaborative studies gave satisfactory INR correction ( Poller et al, 1994 , 1995a, b; Johnson & Bridgen, 1998 ). On the other hand, it has been suggested that only lyophilized plasmas from coumarin-treated patients may be satisfactory for local INR correction ( Houbouyan & Goguel, 1993) because of the presence of PIVKAs (proteins induced by vitamin K antagonism). Stevenson et al (1997 ) claimed better INR correction with lyophilized coumarin plasmas than with their own lyophilized-depleted plasmas over a range of thromboplastin reagents.

The value of lyophilized plasmas as substitutes for fresh coumarin plasmas in thromboplastin calibration can, however, only be reliably decided by comparing results with those of conventional full fresh plasma calibrations using the established W.H.O. protocol with the manual PT technique in a multicentre calibration study. We have used this multicentre approach in the present study and believe this is the first such objective assessment of the two types of lyophilized plasmas in thromboplastin calibration.


Thromboplastin calibration with PT certified lyophilized plasmas was compared with conventional fresh plasma calibrations, performed according to the conventional W.H.O. recommended procedure ( W.H.O. Expert Committee on Biological Standardisation, 1983) using 60 plasmas from long-term stabilized coumarin-treated patients. A spectrum of lyophilized artificially-depleted plasmas was prepared for the study to provide a wide spread of INR values over the therapeutic range. The results have been analysed in two ways. For valid comparison with the 60 fresh plasmas in a W.H.O. type calibration, 60 artificially-depleted plasmas were included in the calibrations. With the second approach, reduced sets of 20 depleted plasmas were considered in order to be comparable with the restricted number of 20 lyophilized coumarin plasmas available to the study. To avoid bias, the 60 depleted plasmas were divided into three consecutive series of 20. These were certified with mean PT values by the European Concerted Action on Anticoagulation (ECAA) national laboratories. In a previous ECAA report ( European Concerted Action on Anticoagulation, 1998a) it was shown that 20 of the ECAA certified artificially-depleted plasmas combined with seven lyophilized normal plasmas was sufficient to provide a reliable estimate of the ISI in manual PT calibration of the ECAA human thromboplastin included in the present report. The effect of substituting in turn the two types of lyophilized plasmas (coumarin and depleted) for the calibration slopes of the 60 fresh coumarin plasmas have been compared. For simplicity, the study has been limited to the effect of abnormal plasmas only on these slopes.

Lyophilized plasmas

Both types of lyophilized plasma were dispensed in 0.5 ml volumes in rubber-capped vials and vacuum sealed. They were reconstituted in 0.5 ml volumes with distilled water.

(a) Artificially-depleted plasmas

The lyophilized depleted plasmas were manufactured at the ECAA Central Facility in Manchester by artificial depletion of normal human plasma by selective adsorption of vitamin K-dependent clotting factors with barium sulphate to provide a range of values which spanned the therapeutic interval of 1.5–4.5 INR when tested with the International Council for Standardization in Haematology (ICSH) thromboplastin IRP (BCT/441 human plain). Different INR levels were achieved by variable adsorption. Hepes buffer (0.26 g%), glycine (2.18 g%) and sucrose (2.18 g%) were added to the plasmas as protectives prior to lyophilization. Because of possible complicating effects of depletion of the non-coumarin-dependent clotting factors, factor V and fibrinogen on the PT results, factor V assays ( Thomson, 1970) and fibrinogen determinations ( Clauss, 1957) were performed on all the artificially-depleted plasmas to ensure adequate levels. Samples containing less than a minimum level of 50% factor V or 1.5 g/l fibrinogen were excluded. These minimum levels are regarded as adequate for prothrombin time measurement and only when the individual clotting factors V or fibrinogen fall below these levels do they prolong the PT test. The factor V of all the artificially-depleted plasmas included in the study ranged from 50% to 100% and fibrinogen content from 1.5 g/l to 4.0 g/l. It has been shown that when factor V levels are > 40% they do not influence INR results ( Tripodi et al, 1995 ). Inter-vial variation, accelerated degradation and long-term stability studies were also performed on each of the depleted lyophilized plasmas. All plasmas gave a coefficient of variation (CV) of < 3% in inter-vial studies and minimum heat degradation stability of 7 d at 40°C. Long-term stability studies also proved satisfactory up to a minimum of 12 months.

(b) Coumarin plasmas

These also were lyophilized at the Central Facility in the same manner as above for the artificially-depleted plasmas. Single donations obtained from each long-term stabilized patient were separately lyophilized after the addition of Hepes buffer, glycine and sucrose as protectives in the same concentrations as in the depleted plasmas. Owing to the difficulty of obtaining adequate volumes of plasma from anticoagulant-treated patients, the difficulty of duplication of INR and the problems of obtaining a wide spectrum of INR values from these patients, only 20 of these plasmas were regarded as suitable for inclusion.

PT testing of lyophilized plasmas was performed using the manual PT technique. Harmonization of the manual PT technique had been attempted at a preliminary ‘wet workshop’ held at the Central Facility.


(a) Reference thromboplastin: International Council for Standardization in Haematology (ICSH) human plain IRP (BCT/441).

(b) Thromboplastin for local calibration: ECAA Human Recombinant Thromboplastin (low ISI human recombinant thromboplastin donated by Ortho DS, Raritan, N.J., U.S.A.) calibrated previously against the ICSH human plain IRP in a multicentre exercise ( Poller et al, 1996 ).

Certification of lyophilized plasmas

The certification of the lyophilized plasmas in PT terms of the human plain IRP (BCT/441) by the manual PT technique was performed by six of the 14 participating national laboratories (‘certifying centres’). The plasmas were tested in quadruplicate. The overall mean PT of each lyophilized plasma from all six centres was regarded as its certified value.

Thromboplastin calibrations with certified lyophilized plasmas

Evaluation of the certified lyophilized plasma calibrations was performed on the results from the remaining eight ECAA national centres (‘test centres’). The ECAA human reference thromboplastin was used to determine PT values for the sets of lyophilized depleted plasmas and lyophilized coumarin plasmas. Four replicate tests were performed on each plasma with the ECAA human thromboplastin.

Thromboplastin calibrations with fresh plasmas

The eight ‘test centres’ also undertook parallel fresh plasma calibrations on the human ECAA reference thromboplastin versus the human plain IRP according to the W.H.O. protocol on 60 plasmas collected locally from patients on oral anticoagulant treatment for at least 6 weeks.

Statistical analysis

Thromboplastin calibration with fresh plasmas

The calibration procedure previously adopted in collaborative studies for thromboplastin IRP was used ( van der Velde, 1984; Kirkwood, 1983). PTs were plotted on a double logarithmic scale with the ICSH human plain IRP (BCT/441) on the vertical axis, and ECAA human on the horizontal axis, and the method of orthogonal regression employed to determine slopes and intercepts of the calibration line. The study, to compare the effects of the two types of abnormal plasma, was confined to the effects on the slope of the thromboplastin calibration line. ISI were not calculated as this would necessarily have introduced the variable effects of the different numbers of normal subjects on the full (60 plasma) and reduced (20 plasma) calibrations. The number of normals would have had to be reduced proportionately and this would necessarily have introduced a degree of uncertainty for which the abnormal plasmas were not responsible. Intra-laboratory precision of the calibration line was estimated from the CV of the slope (b) of the orthogonal regression line (CV(b) = 100 × SD(b)/b).

Thromboplastin calibration with lyophilized plasmas

Lyophilized plasma calibrations were performed for each of the eight ‘test centres’ by placing the certified BCT/441 values for the log PT of the lyophilized plasmas on the vertical axis and the locally determined log PT results for the lyophilized plasmas with the ECAA human reagent on the horizontal axis.

In the present report, for valid comparison with conventional orthogonal regression analysis, it was assumed throughout that the same degree of error was present in both axes and was not reduced by using certified values. The orthogonal regression model was therefore still required.


Thromboplastin calibration slopes

The results presented in 1Table I and Fig 11 show mean values for thromboplastin calibration slopes of the 60 fresh coumarin patient plasmas and the 60 artificially depleted plasmas determined at the eight ‘test centres’. The mean artificially-depleted plasma calibration slope of 1.0228 was 2.7% higher with the depleted plasmas than the mean fresh plasma calibration slope. The mean CV of the slope was higher with both types of abnormal calibration slopes than would be expected in a combined calibration of normals and abnormals. The CV of the calibration slopes of fresh and lyophilized plasmas was, however, of similar magnitude.

Table 1. Table I. Calibration equations for fresh coumarin plasmas and certified lyophilized depleted plasmas. The coefficients of the equations a (intercept) and b (slope) are provided together with the CV of the slope and the number of abnormals (P) prior to exclusions. Thumbnail image of
Figure 1.

Fig 1. Mean abnormal-only calibration slopes of the eight ‘test centres’ for the 60 fresh coumarin plasma samples and the 60 artificially-depleted lyophilized plasmas.

Table II gives the thromboplastin calibration slopes using the three sets of 20 artificially-depleted plasmas. These gave slopes of 0.9762, 1.0567 and 1.1124 respectively. Fig 2 shows the mean calibration lines for these three sets of depleted plasmas.

Table 2. Table II. Calibration equations for sets of lyophilized depleted plasmas. The coefficients of the equations a (intercept) and b (slope) are provided together with the CV of the slope and number of abnormal (P) plasmas prior to exclusions. The number of outliers (observations in excess of 2 SD from the line) are also provided. Thumbnail image of
Figure 2.

0 artificially-depleted lyophilized plasmas.

The 20 lyophilized coumarins gave a mean slope of 1.0915 (see 3 Table III).

Table 3. Table III. Calibration equations for sets of lyophilized coumarin plasmas. The coefficients of the equations a (intercept) and b (slope) are provided together with the CV of the slope and number of abnormal (P) plasmas prior to exclusions. The number of outliers (observations in excess of 2 SD from the line) are also provided. Thumbnail image of


A new calibration step to control coagulometer effects on INR is required to implement the W.H.O. prothrombin time standardization scheme. This is because coagulometers have largely replaced the manual technique on which the W.H.O. scheme was based and may affect the ISI of thromboplastins. Various additional procedures involving lyophilized artificially-depleted or coumarin plasma calibrations have been proposed ( Clarke et al, 1992 ; van den Besselaar, 1994; Houbouyan & Goguel, 1993; Poller et al, 1994 ). The best way to assess the reliability of a new method of thromboplastin calibration is to compare the results in a parallel multicentre calibration using the established W.H.O. manual method. This is the approach we have employed and which we believe has been used for the first time for the assessment of lyophilized plasma calibrations.

The established procedure for thromboplastin ISI calibration ( W.H.O. Expert Committee on Biological Standardisation, 1983) depends on parallel testing of 60 fresh plasmas from stabilized coumarin-treated patients and 20 normal subjects with a thromboplastin IRP. The local reagent is tested in parallel using the manual PT technique. In the present study the calibration slopes resulting from the use of a comparable number of 60 depleted lyophilized plasmas were compared with those of 60 fresh plasmas from stabilized coumarin-treated patients. Sets of 20 depleted plasmas were also compared with the same number of lyophilized coumarins. The mean slopes with both types of lyophilized plasma were generally higher than that of fresh coumarin plasmas when the lyophilized plasmas were substituted but the differences were not great in clinical terms.

The small but measurable differences from fresh plasma calibrations observed may have been due to either protective additives (Hepes, glycine and sucrose) or the freeze-drying process or a combination of these. The degree of deviation of the calibration slopes which ranged from 2% with the 20 depleted plasmas (set 1) and 9.6% with the lyophilized coumarins could be compared with the possibly more marked and uncontrolled effects of local variables including coagulometers on ISI ( D'Angelo et al, 1989 ; Poller et al, 1989 , 1994; van Rijn et al, 1989 ; Ray & Smith, 1990; Clarke et al, 1992 ; van den Besselaar, 1994). The 2.7% deviation with the full set of 60 artificially-depleted plasmas may be regarded as the best overall representation of this type of calibrant. The effect of lyophilized plasmas would be therefore to slightly increase the ISI.

For simplicity and specificity, the study of the effect of the substitution of the two types of abnormal lyophilized plasmas (artificially-depleted and coumarin) for fresh plasmas from coumarin-treated patients, was limited to the abnormal-only calibration slopes. The exclusion of normals undoubtedly increased the per cent deviations and the CV of slopes of the thromboplastin calibration lines because normal subjects showed much less variability than the PT plasma samples taken from coumarin-treated patients. The precision (CV of the slope) with the full complement of 60 of lyophilized depleted plasmas was in fact slightly greater than the abnormal-only calibration based on 60 fresh coumarin plasmas. The higher CV with both sets of 20 lyophilized plasmas is therefore likely to be a factor of the smaller number in the abnormal-only calibrations. A lower CV of the calibration slope was obtained in a previous study to establish the minimum requirements of lyophilized plasmas for local ISI calibration ( European Concerted Action on Anticoagulation, 1998a) using 27 plasmas (the 20 lyophilized plasmas used in this study plus seven lyophilized normal plasmas).

The present results appear to support previous studies where artificially-depleted lyophilized plasmas certified with PT values in terms of an IRP have provided a considerable but not absolute degree of correction of local effects of coagulometers on ISI and INR ( Clarke et al, 1992 ; Poller et al, 1995a , b; Johnston & Brigden, 1998 ).

In addition to variables introduced by lyophilization, the method of certification of lyophilized plasmas must also be carefully standardized, using a sufficient number of replicates tested by multiple operators. In the present study the calibrants were certified in quadruplicate tests by six centres experienced in the manual PT technique. Finally, the route of calibration cannot be disregarded, as it has been shown that with lyophilized plasmas there is a quantitative difference between INR when calibrating human and rabbit thromboplastins ( European Concerted Action on Anticoagulation, 1998b). Certified values cannot therefore be used for dissimilar thromboplastins. The present results can only be related to the sets of artificially-depleted and coumarin-lyophilized plasmas prepared and certified at the ECAA Central Facility.

Although both types of lyophilized plasma calibration slopes gave measurable differences from conventional fresh plasma calibration slopes, the deviations from the conventional fresh plasma calibration may be regarded as acceptable in clinical terms for the purposes of local ISI calibration.


The following participated in the collaborative exercise. J. Arnout, Thrombosis and Vascular Research, Campus Gasthuisberg, Leuven, Belgium; H. Beeser, Department of Transfusion Medicine and Coagulation, Universität Freiburg, Germany; A. M. H. P. van den Besselaar, Haemostasis and Thrombosis Research Centre, Leiden University Hospital, Netherlands; N. Egberg, Department of Clinical Chemistry, Karolinska Hospital, Stockholm, Sweden; J. A. Iriarte, Hospital Civil de Basurto, Bilbao, Spain; J. Jespersen, Department of Clinical Biochemistry, Ribe County Hospital, Esbjerg, Denmark; I. Kontopoulou-Griva, First Regional Transfusion Centre, Hippocration Hospital, Athens, Greece; B. Lämmle, Central Haematology Laboratory, University Hospital, Bern, Switzerland; K. Lechner, 1. Medizinische Klinik der Universität Wien, Austria; P. M. Mannucci, A. Bianchi Bonomi, Haemophilia and Thrombosis Centre, Milan, Italy; B. Otridge, Mater Misericordiae Hospital, Dublin, Ireland; J. Pina Cabral, Centro de Fisiologia da Hemostase, Universidade do Porto, Portugal; L. Poller, ECAA Central Facility, Department of Pathological Sciences, University of Manchester, U.K.; M. Samama, Service D'Hématologie, Hôtel-Dieu de Paris, France.

This work was supported by grant PL931349 of the EC Biomed programme.