Anton M.H.P. van den Besselaar 1 and Wim P.M. Houdijk 2* Thrombosis. Clinical Chemistry 49: (2003) Hemostasis and

Transcription

1 Clinical Chemistry 49: (2003) Hemostasis and Thrombosis Use of Lyophilized Calibrant Plasmas for Simplified International Normalized Ratio Determination with a Human Tissue Factor Thromboplastin Reagent Derived from Cultured Human Cells Anton M.H.P. van den Besselaar 1 and Wim P.M. Houdijk 2* Background: For monitoring of treatment with oral anticoagulants, the clotting time obtained in the prothrombin time (PT) test is transformed to the International Normalized Ratio (INR) with use of a systemspecific International Sensitivity Index (ISI). The calibrant plasma procedure (CPP) is an alternative approach to INR calculation based on the use of a set of lyophilized plasmas with assigned INRs. Methods: With the CPP, a linear relationship is established between log(pt) and log(inr), using orthogonal regression. CPP was validated for Simplastin HTF, a new human tissue factor reagent derived from cultured human cells. CPP precision was assessed as the CV of the slope of the regression line. The accuracy of the CPP was determined by comparing the INR obtained with the CPP with that obtained with the established ISIbased reference method. INRs of the calibrants were assigned by different routes: by manufacturer (consensus labeling) or by use of Simplastin HTF or International Reference Preparations (IRPs; rtf/95 or RBT/90). Results: The mean CV of the CPP regression slope ranged from 1.0% (Simplastin HTF reagent-specific INR) to 2.4% (INR assigned with rtf/95). INRs calculated with the CPP were similar to those obtained with the reference method, but when the routes for assigning INRs to the calibrant plasmas were compared, the mean difference in INR between CPP and the reference method was smaller with Simplastin HTF reagent-specific values. In several (but not all) cases, this difference was significant (P <0.05, t-test). Conclusion: CPP can be used for local INR determination, but better precision and accuracy are obtained with reagent-specific INRs compared with INR assignment by consensus labeling or IRP American Association for Clinical Chemistry For the monitoring of oral anticoagulant treatment, the results of the prothrombin time (PT) 3 test should be expressed as the International Normalized Ratio (INR). The INR system was originally developed for the manual determination of PTs and was envisaged for the assignment of a single value of the International Sensitivity Index (ISI) by calibration of each batch of thromboplastin reagent with the manual method. In practice, the manual PT has been almost universally replaced by coagulometers, and several studies have shown that the ISI may differ according to the type of instrument used (1, 2). Some reagent manufacturers have introduced instrumentspecific ISIs, but this does not totally prevent the potential for bias because of the many possible instrument reagent combinations and because ISIs may differ even among instruments of the same type. A simplified alternative approach is for laboratories to calibrate their own local systems, using calibrated plasmas (3 5). For this purpose, lyophilized plasmas with assigned INRs have become commercially available. With such calibrant plasmas a direct relationship between log(pt) and log(inr) is established by regression analysis. This procedure is also known as the calibrant plasma procedure (CPP) (6). This approach offers the advantage of simple direct conversion of clotting times into INR 1 Haemostasis and Thrombosis Research Center, Department of Haematology, Leiden University Medical Center, Leiden, The Netherlands. 2 biomérieux bv, P.O. Box 84, 5280 AB Boxtel, The Netherlands. *Author for correspondence. Fax ; wim.houdijk@ eu.biomerieux.com. Received June 23, 2003; accepted September 16, DOI: /clinchem Nonstandard abbreviations: PT, prothrombin time; INR, International Normalized Ratio; ISI, International Sensitivity Index; CPP, calibrant plasma procedure; MNPT, mean normal prothrombin time; IRP, International Reference Preparation; and ÖQUASTA, Austrian Society of Quality Assurance and Standardization of Diagnostic Medical Investigations. 2006

2 Clinical Chemistry 49, No. 12, without the need for a system-specific ISI and mean normal PT (MNPT) (7). Several studies have shown that local calibration can considerably reduce interlaboratory variability in INR determination (5, 8 10), but little is known about the accuracy of the INR obtained by local calibration using lyophilized plasmas (6, 11). The conventional WHO reference method for ISI determination is based on testing of fresh plasma samples from healthy individuals and patients on long-term oral anticoagulant therapy (12). This method has been used in a multicenter ISI calibration study of Simplastin HTF, a new human tissue factor thromboplastin reagent derived from cultured human cells (13). The purpose of the present study was to compare INRs obtained with the conventional WHO reference method (i.e., using instrument-specific ISIs and MNPTs based on fresh plasmas) with INRs obtained with a local lyophilized CPP. Although the CPP does not require the use of an ISI, the slope of the CPP regression line corresponds to the ISI. We compared the slope of the CPP regression line with the ISI obtained by the conventional WHO reference method. Materials and Methods thromboplastin reagents Ampoules of rtf/95 [International Reference Preparation (IRP) for human thromboplastin] and RBT/90 (IRP for rabbit thromboplastin) were obtained from a WHO designated laboratory for biological standards (Sanquin Blood Supply Foundation, Amsterdam, The Netherlands). Simplastin HTF (lot no ) was provided by biomérieux, Inc. (Durham, NC). plasma calibrants AK Calibrant (lot no. OU96000) was obtained from Progen Immuno Diagnostika. This was a set containing four lyophilized calibrant plasmas. Calibrant A was a pooled normal plasma obtained from at least 100 healthy donors and calibrated by the German Institute for Standardisation and Documentation in the Medical Laboratory (INSTAND), using the reference plasma of the German Association of Manufacturers of Diagnostic Instruments and Reagents (VDGH) (14). Calibrants B, C, and D were prepared from the plasmas of patients on long-term oral anticoagulant treatment and calibrated using the results of 200 laboratories participating in an Austrian external quality-assessment scheme [Austrian Society of Quality Assurance and Standardization of Diagnostic Medical Investigations (ÖQUASTA) (15)]. clotting time determination Each of the seven participating laboratories used a manual method for clotting time determination with the three thromboplastins. The tilt tube and the hooking methods were used by five and two centers, respectively. Simplastin HTF was also used with six different automated coagulometers. The measurement principles included photooptical [MDA and Coag-A-Mate MTX (biomérieux); CA6000 (Sysmex Company)], optomechanical [Coag-A- Mate MAX (biomérieux)], mechanical [STA (Diagnostica Stago)], and nephelometric [ACL (Instrumentation Laboratory)]. In agreement with WHO guidelines, each method (i.e., type of coagulometer) was used by at least two laboratories. design of the study All participating centers received detailed instructions for blood collection, handling of thromboplastins and lyophilized plasmas, and determination of PTs. Simplastin HTF was calibrated according to WHO guidelines with rtf/95 used as the reference thromboplastin (12). The calibration was performed on 10 days. On each day, a different set of fresh plasmas (two healthy donors and six patients on long-term oral anticoagulation) was used, together with the same set of four lyophilized calibrant plasmas. Blood was collected with each center s routine system, i.e., commercial evacuated tubes containing mol/l sodium citrate (Becton Dickinson Vacutainer System). Nine volumes of blood were mixed with one volume of sodium citrate solution. Citrated blood was centrifuged at a minimum force of 2500g for 10 min at a controlled room temperature. Plasma was transferred to plastic tubes, which were capped and stored at room temperature. Single PT determinations were performed on each plasma within 5 h from the start of blood collection. To minimize the effect of instability of plasma samples and reconstituted thromboplastin reagents, the testing order of reagents was changed from day to day. statistical analysis For each local system (i.e., combination of Simplastin HTF and clot detection method), an ISI and MNPT were calculated from the fresh plasma results as described in the WHO guidelines, with rtf/95 used as the reference thromboplastin (12). For the CPP, an orthogonal regression line was calculated based on log-transformed mean PTs for the four lyophilized calibrant plasmas from the 10 days of testing (x axis) and log-transformed assigned INRs (y axis). Orthogonal rather than linear regression was used because there was error in both log(pt) and log(inr). The assigned INRs were the manufacturer s values, the mean INR determined with the manual method and an IRP (rtf/95 or RBT/90), or the mean INR determined with Simplastin HTF and various methods (manual and coagulometers). The latter are referred to as reagent-specific INRs. INRs for Simplastin HTF were calculated with the fresh plasma MNPT and ISI from the same laboratory. INRs for rtf/95 and RBT/90 were calculated using the established ISI (0.94 and 1.00, respectively) and the freshplasma MNPTs for these reagents from the same laboratory. The mean INRs are shown in Table 1. Differences in mean INR between manual measurements with the three thromboplastin reagents (i.e., Simplastin HTF, rtf/95,

3 2008 van den Besselaar and Houdijk: Simplified INR Determination with Calibrant Plasmas Plasma calibrant Manufacturer-assigned Table 1. Apparent INRs of calibrant plasmas. a Simplastin HTF Mean (SD) INR All methods (n 19) b Manual (n 7) b rtf/95 manual method (n 7) b RBT/90 manual method (n 7) b A (0.04) 1.15 (0.02) 1.13 (0.03) c 1.07 (0.02) d B (0.07) 2.08 (0.04) 2.03 (0.05) c 2.01 (0.12) C (0.14) 3.11 (0.12) 3.01 (0.09) c 3.30 (0.26) D (0.25) 4.32 (0.16) 4.04 (0.17) c 4.80 (0.48) d a INRs were determined with three different reagents: Simplastin HTF, rtf/95, and RBT/90. Only manual methods were used for rtf/95 and RBT/90. The Student t-test was used to assess the differences between the manual methods. b n number of methods. c Significant difference (P 0.05) between rtf/95 and Simplastin-HTF (manual). d Significant difference (P 0.05) between RBT/90 and Simplastin-HTF (manual). and RBT/90) were tested by Student t-test on paired observations from each laboratory. For each orthogonal regression line, the within-laboratory CV of the slope was calculated with formulas described in the WHO Guidelines (12). The orthogonal regression lines were used to transform PTs determined with Simplastin HTF to INRs. The same PTs were also transformed to INRs by use of the ISI and MNPT obtained with the conventional WHO calibration procedure based on 60 fresh patient samples, 20 fresh normal samples, and rtf/95, performed by the same laboratory. The INRs obtained with the two procedures, i.e., the lyophilized calibrant procedure (INR CPP ) and the conventional WHO procedure (INR WHO ), were compared and the relative difference (as a percentage) was calculated as: 100(INR WHO INR CPP )/INR WHO. Relative INR differences obtained for the various routes of calibrant INR assignment were compared with use of the Student t-test on paired observations. Results Table 1 shows the apparent INRs for the four lyophilized calibrant plasmas. There were considerable differences in the mean INRs for plasmas A and D. The errors in the INRs assigned by the manufacturer were not known. The INR assigned to plasma A by the manufacturer was 10% lower than the mean value obtained with Simplastin HTF. The mean INR for plasma D measured with rtf/95 was 10% lower than the value obtained with Simplastin HTF (manual) and 16% lower than the value determined with RBT/90 (P 0.01). For Simplastin HTF there was a tendency for higher INRs on the ACL compared with the manual determinations with this reagent. The betweenlaboratory variation in the INRs measured with RBT/90 was greater than the variation in the INRs determined with rtf/95 and with Simplastin HTF (with the exception of plasma A). The mean values shown in Table 1 were used for the calibration of each local method for Simplastin HTF. Although there were differences in INR between the manual and automated methods for Simplastin HTF, the mean INR for all methods (manual automated) was used. Table 2 shows the CV for the slopes of the orthogonal regression lines. With a few exceptions, the CV for the lines based on Simplastin HTF reagent-specific values were smaller than the CV for the lines based on values assigned with other thromboplastins. The mean CV were all 3%. The relative differences in INR between the conventional calculation using each center s ISI and MNPT based on fresh-plasma testing and the simplified calibration Table 2. Imprecision of regression lines obtained with calibrant plasmas. a CV of the slope, % Center Method Manufacturer Simplastin HTF rtf/95 RBT/90 1 Manual STA MDA MTX Manual MDA MAX Manual ACL CA MTX Manual MAX Manual ACL Manual STA Manual CA Mean (n 19) a Within-laboratory imprecision of the slope of the orthogonal regression line (CV of the slope, %) relating log(pt) for Simplastin HTF to log(inr) for four lyophilized calibrant plasmas. INRs were assigned by four different routes: by the manufacturer and by use of Simplastin HTF (all methods), rtf/95 (manual method), or RBT/90 (manual method).

4 Clinical Chemistry 49, No. 12, based on the four lyophilized plasmas are shown in Table 3. The relative differences were calculated for two intensities of oral anticoagulation: one corresponding to low intensity (INR 2) and the other to high intensity (INR 4.5). These differences ranged from 9.5% to 11% or from 7.5% to 12.7% for calibration lines with reagent-specific and manufacturer-assigned INRs, respectively. In contrast, the maximum differences for rtf/95- and RBT/90- based calibration lines amounted to 18.1% and 17.7%, respectively. At both intensities, the mean differences were smallest for the Simplastin HTF reagent-specific calibration lines; i.e., 2.5% and 4.2% for low and high intensity, respectively. At INR 2, the absolute values of the relative INR differences were significantly smaller for Simplastin HTF than for the manufacturer- and rtf/95- assigned values [2.5% for Simplastin HTF vs 6.6% for the manufacturer-assigned value (P 0.01) and 4.1% for rtf/95 (P 0.05)]. At INR 4.5, the absolute values of the relative INR differences were significantly smaller for Simplastin HTF than for the rtf/95- and RBT/90-assigned values [4.2% for Simplastin HTF vs 8.1% for rtf/95 (P 0.01) and 7.9% for RBT/90 (P 0.01)]. The slope of the CPP regression line corresponds to the ISI. The slopes of the CPP regression lines based on the INRs assigned by the manufacturer were compared with the ISIs obtained by the previously reported conventional WHO reference method (13). The mean ISIs obtained for the manual method and the various instruments are shown in Fig. 1. The CPP slopes showed a systemdependent pattern similar to that observed for the ISIs determined by the WHO reference method. In all cases, the mean CPP regression line slopes were greater than the corresponding ISIs obtained by conventional calibration against rtf/95. The data used to construct Fig. 1 are not shown here, but the data for the method-specific ISIs obtained by calibration against the IRPs are presented in a table to be published elsewhere (13). Discussion In the determination of INR values for the control of oral anticoagulant therapy, lyophilized calibrant plasmas have been proposed as an alternative for the calibration of local combinations of thromboplastin reagent and instrumentation (3 5). In the present study, this simplified procedure was based on the use of a commercial lyophilized plasma calibrant set consisting of one pooled normal plasma (A) and three coumarin plasmas (B, C, and D) with assigned INR values. In a previous study it was proposed that the use of reagent-specific INRs for lyophilized calibrant plasmas would provide relatively homogeneous INRs for fresh coumarin plasmas, but the accuracy of the CPP remained to be assessed (6). We validated the accuracy of the lyophilized CPP for a single thromboplastin reagent (Simplastin HTF). INRs obtained with the CPP method were compared with INRs obtained by the conventional WHO reference method. In the latter method, the reagent instrument combination was calibrated with use of fresh plasmas and an appro- Table 3. Comparison of the CPP and conventional WHO procedure. a INR difference at INR 2.0, % INR difference at INR 4.5, % Center Method Manufacturer Simplastin HTF rtf/95 RBT/90 Manufacturer Simplastin HTF rtf/95 RBT/90 1 Manual STA MDA MTX Manual MDA MAX Manual ACL CA MTX Manual MAX Manual ACL Manual STA Manual CA Mean absolute difference a The differences between INRs obtained with the conventional calibration procedure (using ISI and MNPT, according to the WHO Guidelines) and the lyophilized CPP were calculated. INRs for the CPP were assigned by four different routes: by the manufacturer and by use of Simplastin HTF (all methods), rtf/95 (manual method), or RBT/90 (manual method).

5 2010 van den Besselaar and Houdijk: Simplified INR Determination with Calibrant Plasmas Fig. 1. ISI by clotting endpoint method and route of calibration. ISI values [mean and between-laboratory SD (error bars)] for calibration of Simplastin HTF by different clotting endpoint methods and routes of calibration. Testing was performed using the manual method or six types of automated coagulometers against two different IRPs: rtf/95 (human) and RBT/90 (rabbit). The ISI was also obtained as the slope of the CPP regression line obtained using the manufacturerassigned INR values (AK Calibrant). Method-specific ISI values are ranked from high (left) to low (right) based on calibration against rtf/95. priate IRP to obtain a system-specific ISI and MNPT. Simplastin HTF is a human thromboplastin; consequently, calibration with the human IRP, rtf/95, is considered as the recommended procedure (12). The multicenter ISI determination for this reagent will be reported elsewhere (13). INRs for the CPP were assigned by four different routes: (a) manufacturer-assigned values; (b) Simplastin HTF reagent-specific values (all methods included); (c) values obtained with rtf/95 (human thromboplastin IRP); and (d) values obtained with RBT/90 (rabbit thromboplastin IRP). Our study showed considerable differences among the mean INRs assigned to the calibrant plasmas by the various routes. In agreement with previous observations (16), our study also indicated that the INR for a lyophilized pooled normal plasma is influenced by the thromboplastin reagent used. According to the manufacturer of this calibrant set, the assigned INR for plasma A (1.04) was obtained by calibration with a German normal reference plasma. Although the error was not given, this value was 10% lower than the mean INR obtained with Simplastin HTF and rtf/95 (Table 1). Furthermore, for the manual method, the value obtained with RBT/90 (1.07) was significantly lower than the values obtained with Simplastin HTF and rtf/95 (Table 1). The present study also showed significant differences in INR for the lyophilized coumarin plasmas B, C, and D measured by the three reagents (Table 1). It is likely that the reagents respond differently to lyophilization-induced changes in the plasmas (i.e., matrix effects). This phenomenon of significant differences in INR values assigned by different reagents was observed previously in a study on INR calibration of lyophilized reference plasmas (17). The authors of that study indicated that single assigned INR values may not be applicable for use with all thromboplastin reagents. The differences with the manufacturerassigned INRs for the coumarin plasmas (Table 1) can be explained by the use of a single consensus value. The INRs for these plasmas were determined for the manufacturer by ÖQUASTA from the results reported by 200 laboratories (15). Although the laboratories participating in the ÖQUASTA external quality-assessment scheme used different reagents and methods, the overall mean was used as a single INR target value. We also observed that the precision of the slope of the calibration line relating log-transformed PTs to log-transformed INRs was better with reagent-specific INR values than that with the manufacturer-assigned values or the values obtained with each IRP (Table 2). The decreased precision for dissimilar comparisons is most probably caused by an interaction between plasma factors and the thromboplastin reagents. If reagent-specific INRs are used, a total of four pooled plasmas appear to be adequate for good precision of the orthogonal regression line. In this case, the CV of the slope was always 3%, the upper limit recommended for the established WHO procedure for ISI calibration with fresh plasmas (12). Finally, for each route of INR assignment, we calculated the difference in INR between the CPP and the established WHO method (Table 3). The mean differences were smaller with the reagent-specific values than the differences seen with the manufacturer-assigned or the IRP values. In a few cases, the differences for automated methods were greater than for the manual method, e.g., the ACL (Table 3). This may be caused by an interaction between plasma factors and the instrument. The assigned INRs for the lyophilized plasmas with Simplastin HTF varied slightly according to the coagulometer used (Table 1). The slope of the CPP regression line corresponds to the ISI. In all cases, the CPP-derived ISIs were greater than the ISIs obtained by calibration with the established WHO reference method and RTF/95 (Fig. 1). The cause of this discrepancy is related to the INR differences between the manufacturer (ÖQUASTA) and rtf/95 (Table 1). In conclusion, we believe that a simple CPP using a commercial set of lyophilized plasmas consisting of a limited number of pooled plasmas can be used for local INR determination. However, both the precision and accuracy of such a CPP are better when reagent-specific INR values are used than when INRs are assigned by use

6 Clinical Chemistry 49, No. 12, of overall mean consensus labeling (ÖQUASTA) and by use of IRPs. This observation may have practical implications, and further field trials to assess the validity of using lyophilized calibrant plasmas with single assigned INR values with a variety of thromboplastin reagents are required. The following laboratories participated in the study: Hamilton Civic Hospitals Research Centre, Hemostasis Reference Laboratory, Hamilton, Ontario, Canada (M. Johnston); Kingston General Hospital, Department of Pathology, Kingston, Ontario, Canada (Dr. D. Rapson and L. Dwyre); Leiden University Medical Center, Hemostasis and Thrombosis Research Center, Leiden, The Netherlands (Dr. A.M.H.P. van den Besselaar); University Hospital Leuven, Center for Molecular and Vascular Biology, Leuven, Belgium (Prof. J. Arnout); A. Bianchi Bonomi Hemophilia and Thrombosis Centre, University and IRCCS Maggiore Hospital, Milan, Italy (Dr. A. Tripodi); University Hospital Robert Debré, Laboratoire Central d Hématologie, Reims, France (Prof. G. Potron and Dr. C. Droulle); and Royal Hallamshire Hospital, Department of Hematology, Sheffield, United Kingdom (Dr. S. Kitchen). Excellent technical assistance was given by H. Schaefervan Mansfeld and E. Witteveen. The study was financially supported by biomérieux, Inc. (Durham, NC). References 1. 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 Haemost 1989;62: Ray MJ, Smith IR. The dependence of the international sensitivity index on the coagulometer used to perform the prothrombin time. Thromb Haemost 1990;63: Poller L, Triplett DA, Hirsh J, Carroll J, Clarke K. The value of plasma calibrants in correcting coagulometer effects on international normalized ratios. Am J Clin Pathol 1995;103: Hubbard AR, Margetts SML, Barrowcliffe TW. International normalized ratio determination using calibrated reference plasmas. Br J Haematol 1997;98: Adcock DM, Duff S. Enhanced standardization of the international normalized ratio through the use of plasma calibrants: a concise review [Review]. Blood Coagul Fibrinolysis 2000;11: Van den Besselaar AMHP, Houbouyan-Réveillard LL. Field study of lyophilized calibrant plasmas for fresh plasma INR determination. Thromb Haemost 2002;87: Craig S, Stevenson KJ, Duffy JM, Taberner DA. Local INR correction: justification for a simplified approach. J Clin Pathol 1997;50: Houbouyan LL, Goguel AF. Long-term French experience in INR standardization by a procedure using plasma calibrants. Am J Clin Pathol 1997;108: Adcock DM, Johnston M. Evaluation of frozen plasma calibrants for enhanced standardization of the international normalized ratio (INR): a multi-center study. Thromb Haemost 2002;87: Chantarangkul V, Tripodi A, Cesana BM, Mannucci PM. Calibration of local systems with lyophilized calibrant plasmas improves the interlaboratory variability of the INR in the Italian external quality assessment scheme. Thromb Haemost 1999;82: Van den Besselaar AMHP. International normalized ratio: towards improved accuracy [Comment]. Thromb Haemost 1999;82: WHO Expert Committee on Biological Standardization. Guidelines for thromboplastins and plasma used to control oral anticoagulation therapy. World Health Organ Tech Rep Series 1999;889: Houdijk WPM, van den Besselaar AMHP. International multicenter ISI calibration of a new human tissue factor thromboplastin reagent derived from cultured human cells. J Thromb Haemost; in press. 14. Lang H, Spaethe R, Beeser H, Becker U, Kolde HJ, Spanuth E, et al. Calibration of a lyophilized pooled plasma as candidate reference plasma for standardization of the prothrombin time ratio. Hämostaseologie 1993;13: Lang H, Scheer B, Moritz B, Legenstein E, Kaiser E, Fischer M. International Normalized Ratio (INR) proficiency tests by ÖQUASTA for the prothrombin time. Hämostaseologie 1995;15: D Angelo A, Galli L, Lang H. 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. Clin Chem 1997;43: Hubbard AR, Margetts SM, Weller LJ, Macnab J, Barrowcliffe TW. An international collaborative study on the INR calibration of freeze-dried reference plasmas. Br J Haematol 1999;104: