Comparison of mean normal prothrombin time (PT) with PT of fresh normal pooled plasma or of a lyophilized control plasma (R82A) as denominator to express PT results: collaborative study of the International Federation of clinical chemistry.
The Verband der Deutschen Gerate-Hersteller has prepared a lyophilized normal pooled plasma from 101 healthy blood donors for in-house calibration of commercial lots of normal control plasma . This plasma, named R82A, was calibrated in a study involving the manufacturing companies (Baxter, Behring, Immuno) against fresh normal plasma pools (FNPP) with the use of the reference thromboplastin BCT/099 and a series of additional PT systems . The PT of this lyophilized plasma was 2% longer than that of FNPP with the combined thromboplastins and 6% longer with the plain thromboplastins. Lang et al.  concluded that plasma R82A could be used as a substitute for FNPP with the use of different correction factors for plain and combined thromboplastins. Plasma R82A was also calibrated against fresh individual normal plasmas (MNPT) with the use of a combined reference thromboplastin (OBT/79) and a small series of plain reference thromboplastins or candidate reference thromboplastins . In this study, there was no difference between the PT of plasma R82A and the MNPT with the combined thromboplastin, but there was an excess 2% PT prolongation of the lyophilized plasma with plain thromboplastins, suggesting adoption of a correction factor with plain thromboplastin only . In addition to the conflicting results obtained by comparison with FNPP and MNPT, an absolute requirement for a reagent-dependent correction factor has not been established by these studies because the well-known effect of instruments on the PT ratio [11,12] could not be investigated. In addition, the recently introduced thromboplastin reagents containing recombinant human tissue factor were not evaluated in either study.
A collaborative project of the IFCC was started with the aim to prove the principle of calibration of plasma R82A against the MNPT and the corresponding (obtained from the same donors) FNPP with the use of a variety of plain, combined, and recombinant thromboplastins and of endpoint detectors.
Materials and Methods
DESIGN OF THE STUDY
Fifteen laboratories participated in the study. In each laboratory, at least two different thromboplastins and endpoint detectors were used. Overall, 16 different PT reagents (11 plain, 3 combined, and 2 recombinant thromboplastins) and 18 endpoint detectors were evaluated, for a total of 58 PT systems. Plasma samples from 30 apparently healthy subjects were tested to calculate the MNPT. Within 3 h after blood collection by venipuncture in Vacutainer Tubes containing 32 or 38 g/L trisodium citrate, storage, and centrifugation at room temperature, each laboratory tested 6 individual plasma samples (from healthy volunteers, 3 women and 3 men), a fresh plasma pool obtained from the individual normal plasmas (FNPP), and the lyophilized plasma R82A on 5 different days with each PT system. Plasma R82A was tested within 15 and 30 min after reconstitution in 1 mL of distilled water. The 30 individual normal plasmas were tested in duplicate; the 5 FNPPs and the R82A plasma were tested in triplicate on each day. Reagents and endpoint detectors are listed in Table 1.
Aliquots of each FNPP obtained on the 5 days in all laboratories were snap-frozen and stored at -30 [degrees]C or lower until being shipped to one of the authors (H.L.) for determination of the citrate content (Citronen-Saure UV-Test, Boehringer Mannheim).
Within-day, between-day, and total variations in the PT of the individual normal plasma samples, corresponding FNPPs, and plasma R82A were calculated for each PT system . CVs were compared by Kruskal-Wallis analysis of variance and the Mann-Whitney test.
Arithmetic mean, median, and geometric mean of the PT of individual normal plasmas over all days (MNPT), of the fresh plasma pools over all days (mean FNPP-PT), and of plasma R82A over all days (mean R82A-PT) were calculated for each reagent-method combination and compared by two-way analysis of variance.
PT values of the 30 apparently healthy individuals tested in each laboratory with the different PT systems were expressed as PT ratios with the use of either the MNPT, the grand mean (arithmetic) PT of the 5 separate FNPPs, or the grand mean (arithmetic) PT of plasma R82A as the denominator term.
The separate effects of reagents, endpoint detectors, plasma citrate concentrations, and laboratories on the mean PT ratios expressed with the different modalities were evaluated by analysis of variance and covariance with repeated measures with the use of the BMDP[R] software program.
Comparison of the CVs obtained with the 58 PT systems is reported in Table 2. As expected, total and between-day variations were significantly higher with the FNPPs than with plasma R82A (P = 0.0004 and P = 0.0002, respectively). With FNPPs, the between-day variation was greater than the within-day variation (P = 0.03). Significant between-laboratory differences in components of variations were observed for the individual plasma samples (P ranging from 0.03 to 0.0001) and the FNPPs (between-day variation, P = 0.004; total variation, P = 0.002), but in no case for plasma R82A.
For the expression of the normal PT, the geometric mean of as many individual determinations as possible should be calculated [141. Comparison of the overall geometric mean, arithmetic mean, and median of the PT values determined in plasma R82A and in FNPP showed no statistically significant differences (Table 3). With the individual normal samples, the arithmetic mean was greater than the geometric mean with 23 combinations (P <0.0001), although in no case by >0.1 s or >1%.
Box plots of PT ratios obtained with the different PT systems in the apparently healthy subjects according to the different modality of expression of PT ratios are shown in Fig. 1. As expected, with use of the MNPT as the denominator term, the mean PT ratios of apparently healthy subjects did not differ from 1.00, but they were significantly >1.00 (mean value 1.02) when the FNPP-PT was used as the denominator term and significantly <1.00 (mean value 0.98) when the R82A-PT was used as the denominator term.
Ninety-five percent confidence limits of the mean citrate concentration in FNPPs ranged from 15.3 to 22 nmol/L, and there was no significant difference between the citrate concentration of plasma R82A (17.6 nmol/L) and the average FNPP citrate concentration (18.7 nmol/L, range 15.4-22.5 nmol/L).
There was no effect of reagents, endpoint detectors, laboratories, and plasma citrate concentration on PT results of apparently healthy subjects expressed when the MNPT was used as the denominator term. Conversely, the endpoint detectors significantly affected the PT ratios of the apparently healthy subjects expressed with the use of the FNPP-PT as the denominator term (F-value = 1.9; df = 14, P = 0.024), which were also significantly different in the two laboratories where the same PT system was tested at a similar plasma citrate concentration (F-value = 7.13, df = 1, P = 0.008). Reagents (F-value = 27.7, df = 14, P = 0.00001), endpoint detectors (F-value = 7.8, df = 14, P = 0.00001), and the plasma citrate concentration (Fvalue = 7.9, df = 1, P = 0.005) exerted significant separate effects on PT ratios expressed with the use of the R82A-PT as the denominator term.
[FIGURE 1 OMITTED]
Because it has been reported that combined and plain thromboplastins behave differently with plasma R82A, analysis of the effects of reagents, endpoint detectors, and plasma citrate concentration on PT ratios of apparently healthy subjects expressed with the use of plasma R82A-PT as the denominator term was conducted separately for plain, combined, and recombinant thromboplastin reagents (Fig. 2). Mean PT ratios of apparently healthy subjects ranged from 0.91 to 1.05 (mean 0.98) with plain thromboplastin reagents, from 0.98 to 1.05 (mean 1.02) with the combined thromboplastin reagents, and from 0.87 to 0.97 (mean 0.93) with the recombinant thromboplastin reagents. With the plain thromboplastins, reagents (F-value = 13.2, df = 9, P = 0.00001), endpoint detectors (F-value = 6.0, df = 13, P = 0.00001), and plasma citrate concentration (F-value = 9.2, df = 1, P = 0.0024) all significantly affected PT ratios, also influenced by between-laboratory differences (F-value = 10.2, df = 1, P = 0.014). With the combined thromboplastins, the endpoint detectors, but not the reagents, had a significant effect on PT ratios (F-value = 4.2, df = 4, P = 0.0025). The influence of the plasma citrate concentration and between-laboratory differences could not be explored. With the two recombinant thromboplastins, reagents (F-value = 64.6, df = 1, P = 0.00001), and endpoint detectors (F-value = 15.1, df = 4, P = 0.00001) affected PT ratios of apparently healthy subjects. The effect of the citrate concentration could not be explored, but there was a significant between-laboratory difference (F-value = 12.9, df = 1, P = 0.0004).
[FIGURE 2 OMITTED]
This field study evaluated the results of substituting the PT of the primary reference lyophilized plasma R82A for that of fresh normal plasma as the denominator term in the expression of PT ratios of apparently healthy subjects. We first evaluated the imprecision of PT determinations carried out in plasma R82A with the use of a variety of systems and compared them with the within-day and day-to-day variations of PT determinations carried out in 30 normal plasma samples collected in each laboratory and in fresh normal pooled plasmas obtained from the same donors. The median day-to-day imprecision in the PT of plasma R82A was 1.23%, lower than the dayto-day variation of FNPPs obtained from 6 different donors on each day (2.13%). Given the relatively narrow total variation observed for the normal plasma samples (5.27%), these results indicate acceptable reproducibility of PT determinations carried out in plasma R82A.
Performance of plasma R82A in routine laboratory practice was evaluated by direct comparison of the PT ratios of the normal plasma samples expressed with the use of either the MNPT, the mean PT of the 5 FNPPs, or the mean PT of plasma R82A as the denominator term. Underestimation of PT ratios with plasma R82A and overestimation of PT ratios with FNPP were observed as compared with PT ratios calculated with the use of the MNPT as the denominator term. Reagents had no effect on PT ratios expressed with either the MNPT or the FNPP-PT, but reagent-dependent differences in PT ratios expressed with the PT of plasma R82A as the denominator term were observed. In line with the previous reports [8,10], PT ratios were higher with the combined than with the plain thromboplastins, a finding expected because of the slight but significant loss of factor V activity during the lyophilization procedure of plasma R82A . The largest deviation from unity was observed for PT ratios obtained with the two recombinant thromboplastins, which ranged from 0.87 to 0.97. However, analysis of the influence of reagents on PT results obtained with plasma R82A as the denominator term revealed highly significant reagent effects even within the groups of plain and recombinant thromboplastins, arguing against the adoption of correction factors for the above types of reagents to normalize PT results. Further arguments against the introduction of correction factors for plasma R82A resulted from the observation of a significant influence on PT ratios of endpoint detectors with all types of thromboplastin reagents and of the plasma citrate concentration whenever its effect could be explored. Given the range of deviations from unity of PT ratios with the 58 PT systems evaluated, normalization of PT results may be improved only by the adoption of system-specific correction factors for plasma R82A.
The results obtained with one of the recombinant reagents evaluated (Recombiplastin) are at variance with those of a study reporting that the MNPT could be substituted by the PT of a lyophilized normal plasma produced by the manufacturer of Recombiplastin . Differences in the preparation of lyophilized plasmas may explain the apparently conflicting results. However, full commutability  of PT ratios with the Ortho lyophilized control plasma remains to be demonstrated, because only one additional plain thromboplastin reagent of the same brand was evaluated by Poller et al. .
We observed a significant difference between the MNPT and the FNPP-PT obtained from the same donors contributing to the calculation of the MNPT. This observation is in line with a report that the fresh pooled plasma gives a PT that is shorter than the MNPT obtained with the individual plasmas constituting the pooled plasma . The cause of the discrepancy is unknown, but it is noteworthy that PT ratios calculated with the FNPP-PT as the denominator term were influenced by endpoint detectors and not by reagents.
An a priori requirement of the design of our study was that the number of apparently healthy subjects evaluated in each laboratory was sufficiently representative of the normal population. According to accepted protocols the MNPT should be determined by the measurement of PTs of at least 20 apparently healthy individuals . Because significant between-laboratory differences emerged in our analysis when the PTs of FNPP or of plasma R82A were used as denominator terms, the possibility of differences in the selection of the normal plasma samples among the participating laboratories cannot be ruled out. Although it cannot be excluded that a higher number of apparently healthy subjects may be required for correct calculation of the MNPT, between-laboratory differences might also be explained by the different analytical performance of endpoint detectors of the same brand.
In conclusion, full commutability of PT ratios obtained with a large series of PT systems with the use of the PT of the lyophilized plasma R82A or of FNPP was not achieved in this collaborative study, nor can it be obtained with the use of reagent-specific correction factors because of the separate effects of endpoint detectors and plasma citrate concentration. Although the differences observed may hardly prevent diagnosis of clinically relevant deficiencies of the extrinsic clotting pathway, they may affect commutability of INR values in patients on oral anticoagulant treatment. The use of lyophilized control plasmas, like the one evaluated in this study, has merit as a measure of interlaboratory variability and intralaboratory QC. However, the MNPT should be used as the denominator term in the expression of INR values.
Received June 9, 1997; revision accepted July 29, 1997.
[1.] Peters RHM, van den Besselaar AMHP, Olthuis FMFG. Determination of the mean normal prothrombin time for assessment of international normalized ratios. Thromb Haemostasis 1991;66: 442-5.
[2.] NCCLS. One-stage prothrombin time (PT) test and activated partial thromboplastin time (APTT) test; approved guideline. NCCLS document H47-A (ISBN 1-56238-301-9). Wayne, PA: NCCLS, 1996.
[3.] NCCLS. How to define and determine reference intervals in the clinical laboratory; approved guideline. NCCLS document C-28A (ISBN 1-56238-269-1). Villanova, PA: NCCLS, 1995.
[4.] Van den Besselaar AMHP. Arguments for the recommended mean normal prothrombin time to calculate the international normalized ratio. Thromb Haemostasis 1994;72:487-8.
[5.] van den Besselaar AMHP, Evatt BL, Bogan DR, Triplett DA. Proficiency testing and standardization of prothrombin time: effect of thromboplastins, instrumentation and plasma. Am J Clin Pathol 1984;82:688-99.
[6.] Poller L. INR and the therapeutic range. Biol Clin Hematol 1987; 9:203-13.
[7.] Lang H, Scheer B, Moritz B, Legenstein E, Kaiser E, Fischer M. International normalized ratio (INR) proficiency tests by OQUASTA for the prothrombin time: use of plasma from donors under oral anticoagulation. Hamostaseologie 1995;15:41-8.
[8.] Lang H, Spaethe R, Beeser H, Becker U, Kolde HJ, Spanhut E, et al. Calibration of a lyophilized pooled plasma as a candidate reference plasma for standardization of the prothrombin time ratio. Hamostaseologie 1993;13:96-105.
[9.] Owren PA, Aas K. The control of dicoumarol therapy and quantitative determination of prothrombin and proconvertin. Scand J Clin Lab Invest 1951;3:201-8.
[10.] van den Besselaar AMHP. Multi-center study of replacement of the International Reference Preparation for thromboplastin, rabbit, plain. Thromb Haemostasis 1993;70:794-9.
[11.] D'Angelo A, Seveso MP, Vigano'D'Angelo S, Gilardoni F, Macagni A, Manotti C, Bonini PA. Comparison of two automated coagulometers and the manual tilt-tube method for the determination of prothrombin time. Am J Clin Pathol 1989;92:321-8.
[12.] Poggio M, van den Besselaar AMHP, van der Velde EA, Bertina RM. The effect of some instruments for prothrombin time testing on the international sensitivity index (ISI) of two rabbit tissue thromboplastin reagents. Thromb Haemostasis 1989;62:86874.
[13.] Krouwer JS, Rabinowitz R. How to improve estimates of imprecision. Clin Chem 1984;30:290-2.
[14.] Kirkwood TBL. Calibration of reference thromboplastins and standardisation of the prothrombin time ratio. Thromb Haemostasis 1983;49:238-44.
[15.] Poller L, Houghton D, Carroll J. The reliability of the mean normal prothrombin time of fresh plasmas and of the normal value from a lyophilized "normal" plasma in prothrombin ratio determination. Br J Haematol 1994;88:866-73.
[16.] D'Angelo A. Prothrombin time standardization: the problem of the control plasma. Eur J Clin Chem Biochem 1995;33:1019-22.
[17.] Burgess-Wilson ME, Burri R, Woodhams BJ. Important differences encountered in the normal plasma pools used for the control of oral anticoagulation [Abstract]. Thromb Haemostasis 1993;69: 1124
[18.] Poller L, Hirsh J. A simple system for the derivation of international normalized ratios for the reporting of prothrombin time results with North American thromboplastin reagents. Am J Clin Pathol 1989;92:124-6.
ARMANDO DANGELO, (1)* LAURA GALLI, (2) and HARTMUT LANG (3) ON BEHALF OF THE IFCC WORKING GROUP STANDARDIZATION OF COAGULATION TESTS (4)
(1) Coagulation Service and (2) Department of Epidemiology, Scientific Institute H S. Raffaele, Milano, Italy. (3) Immuno AG, Vienna, Austria. (4) List of participants (in alphabetical order): J. Amiral, Gennevilliers, France; H. Beeser, Freiburg LB., Germany; P. J. Braun, Durham, NC; P. C. Comp, Oklahoma City, OK; A. D'Angelo, Milano, Italy; F. Datii, Marburg, Germany; A. Duncan, Atlanta, GA; A. Girolami, Padova, Italy; F. Keller, Wdrzburg, Germany; H. Lang, Vienna, Austria; M. M. Miiller, Vienna, Austria; H. Pelzer, Miami, FL; P. Sic, Toulouse, France; D. A. Triplett, Muncie, IN; J. W. J. van Wersch, Heerlen, The Netherlands. (5) Nonstandard abbreviations: PT, prothrombin time; MN PT, mean normal prothrombin time; FNPP, fresh normal plasma pool; INR, International Normalized Ratio.
* Author for correspondence. Servizio di Coagulazione, Istituto Scientifico H S. Raffaele, Via Olgettina 60, 20132 Milano Italy. Fax 39-2-26432640.
Table 1. Reagents and endpoint detectors evaluated in the collaborative study. Lab Thromboplastin reagent Clot detection method 1 Neoplastin (Stago) Plain ST4 (Stago) Neoplastin C (Stago) Plain Manual 2 Thromboplastin IS Plain Schnitger Grob (Amelung) (Dade) OBT 79 (IRP) Combined Manual 3 Simplastin L (Organon Plain MDA 180 (Organon Teknika) Teknika) Simplastin Excel Plain Coag-mate X2 (Organon (Organon Teknika) Teknika) 4 Simplastin Excel Fibrometer (Becton Dickinson) Thromboplastin IS ACL 300 (Instrumentation Laboratory) 5 Recombiplastin (Ortho) Recombinant ACL 3000 (Instrumentation Laboratory) Innovin (Dade) Recombinant Electra 1000C (MLA) 6 Chromoquick (Behring) Plain Chromotimer (Behring) Thromborel S (Behring) Plain Schnitger Grob KC 10 (40) (Amelung) Fibrintimer A (Behring) 7 Thromborel S Fibrintimer A Immunoplastin HIS Plain KC 10 (Immuno) 8 Immunoplastin HIS Schnitger Grob Thrombotest (Nyegaard) Combined Thrombotrack 4 (Behnk) 9 Normotest (Nyegaard) Combined CA 5000 (Sysmex) Thrombotest KC 10 10 Innovin Electra 900 (MLA) Thromboplastin IS Fibrometer 11 Neoplastin STA (Stago) Thromborel S KC 10 12 Thromboplastin C Plus Plain Electra 900C (MLA) (Dade) Simplastin Excel STA 13 Chromoquick Chromotimer Thromborel S Schnitger Grob 14 Thromborel S Clot-timer (Mechrolab) Recombiplastin Electro-mechanic 15 IL-PT reagent (Instru- Plain ST4 mentation Laboratory) Thromboplastin D Plain ACL 3000 Plus (Instrumen- (Pacific Hemostasis) tation Laboratory) Table 2. Within-day, between-day, and total variation in the PT of the normal plasma samples, FNPPs obtained from the same individuals, and plasma R82A. CV, %, median (and range) (a) n Within-day Between-day Total Normal 60 4.89 (2.08-8.31) 1.24 (0.00-6.33) 5.27 (2.05-8.64) plasma samples FNPPs 15 1.47 (0.53-7.06) 2.13 (0.38-9.24) 2.83 (1.05-10.77) R82A 15 1.65 (0.33-6.57) 1.23 (0.00-8.54) 2.16 (0.81-8.94) P (b) <0.0001 <0.000 <0.0001 (a) Calculated for the 58 PT systems. Normal plasma samples were tested in duplicate (6 individuals on each day for 5 days). Fresh normal plasma pools and plasma R82A were tested in triplicate on each of the 5 days. (b) Kruskal-Wallis analysis of variance. Table 3. Grand means of arithmetic mean, median, and geometric mean of the PT of the normal plasma samples, FNPPs, and plasma R82A with the 58 PT systems. PT, s (SD) Normal plasma samples FNPP R82A Arithmetic mean 15.40 (7.6) 15.12 (7.3) 15.62 (7.4) Median 15.39 (7.6) 15.14 (7.4) 15.59 (7.4) Geometric mean 15.36 (7.6) 15.11 (7.3) 15.61 (7.4) P (a) 0.02 ns ns (a) Two-way analysis of variance. With normal plasma samples, the geometric mean of the PTs was significantly lower than the arithmetic mean of the PTs (P <0.00001). ns, not significant.
|Printer friendly Cite/link Email Feedback|
|Author:||D'Angelo, Armando; Galli, Laura; Lang, Hartmut|
|Date:||Nov 1, 1997|
|Previous Article:||Comparison of NCEP performance specifications for triglycerides, HDL-, and LDL-cholesterol with operating specifications based on NCEP clinical and...|
|Next Article:||Biological variation of international normalized ratio for prothrombin times, and consequences in monitoring oral anticoagulant therapy: computer...|