Abstract
Objectives
Frozen and freeze-dried plasmas may be used for local prothrombin time system calibration, for direct international normalized ratio (INR) determination, and for quality assessment. The purpose of the present study was to evaluate the usefulness of INRs assigned with various types of thromboplastins to frozen and freeze-dried pooled plasmas obtained from patients treated with vitamin K antagonists.
Methods
INRs were calculated according to the international sensitivity index (ISI) model using various thromboplastins and instruments, i.e. International Standards for thromboplastin as well as six commercial reagents prepared from rabbit and bovine brain, and recombinant human tissue factor. The uncertainty of the INRs was assessed using the standard deviations of clotting times and ISI values. Commutability of the plasmas was assessed according to the approved Clinical and Laboratory Standards Institute (CLSI) Guideline EP30-A. Validation of a set of six frozen plasma pools for direct INR determination was performed according to the Subcommittee on Control of Anticoagulation of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis (SSC/ISTH) guidelines.
Results
For all frozen and freeze-dried plasmas, the INRs calculated with bovine thromboplastin Thrombotest were lower than the INRs assigned with other thromboplastins. With a few exceptions, the frozen and freeze-dried pooled plasmas were commutable. When the set of six frozen plasma pools was used for local calibration, the analytical bias of the INR was less than ±10% for all commercial reagents except Thrombotest.
Conclusions
Processing of fresh plasmas to prepare pooled frozen plasmas and freeze-dried plasmas may lead to different INR assignments depending on the thromboplastin used. Despite minor INR differences, a set of six frozen plasma pools could be used for local calibration by direct INR determination.
Introduction
Monitoring of treatment with vitamin K antagonists (VKA) is performed primarily with the prothrombin time (PT), transformed to the international normalized ratio (INR). Reliable INR determination depends on accurate values for international sensitivity index (ISI) and mean normal prothrombin time (MNPT). Local ISI calibration or ‘direct’ INR determination may be performed by means of frozen or freeze-dried plasmas to which certified values of the PT or INR have been assigned [1]. Many European laboratories perform a local calibration procedure using a set of calibrator plasmas either of commercial origin or obtained from an External Quality Assessment (EQA) organization [2]. It is assumed that the calibrator plasmas are commutable, but it has been reported that local INR calibration with lyophilized calibrator plasmas may not be valid for some thromboplastin-instrument combinations [3], [4].
Lyophilized plasmas for local calibration have been certified using recombinant human and rabbit thromboplastins [5]. Relatively few studies have been performed with frozen plasmas. In one study, frozen plasmas were studied with human and rabbit thromboplastins [6]. In the present study, we describe a set of six frozen plasma pools that were prepared specifically for ‘direct’ INR determination. In addition, we describe 10 freeze-dried plasma pools that were prepared for EQA of the PT/INR by the Section Coagulation of the Dutch Foundation for Quality Assurance in Medical Laboratories. Commutability is a requirement for the application of EQA results in the evaluation of the performance of participating laboratories in terms of standardization of their measurements [7]. Only very few studies have addressed the commutability of EQA materials with regard to INR [8], [9].
The purpose of the present study is to assign INRs to the frozen and freeze-dried plasma pools using a set of thromboplastin reagents prepared from rabbit brain, bovine brain, and recombinant human tissue factor and to assess the uncertainties of the assigned values. In addition, we assessed the commutability of the frozen and freeze-dried plasmas in order to determine their suitability as potential trueness verification materials. Commutability assessment was performed according to the Clinical and Laboratory Standards Institute (CLSI) guidelines approved in 2010 [7]. The present study is the first to apply the CLSI guideline EP30-A for commutability assessment of frozen and freeze-dried plasma pools to be used with laboratory systems for PT/INR determination. Finally, we evaluated the usefulness of the set of frozen plasma pools for ‘direct’ INR determination, according to guidelines issued by the Subcommittee on Control of Anticoagulation of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis (SSC/ISTH) [1].
The experiments described in this article were performed in the period 2011–2014. The present study is of historical interest because we included an evaluation of the reagent Thrombotest (Axis Shield, Oslo, Norway), a bovine brain thromboplastin combined with adsorbed bovine plasma, developed originally by Owren [10]. Thrombotest has been used extensively by clinical laboratories in Europe and Japan [11], [12], [13], [14], [15]. In the past, two batches of Thrombotest were established successively as international reference preparations for thromboplastin, bovine, combined, but have been discontinued several years ago [16], [17]. The production of Thrombotest was terminated in 2013. The other commercial thromboplastin reagents evaluated in this study are still available on the market.
Materials and methods
The study has been approved by the Leiden University Medical Center Ethics Committee.
Reagents and instruments
The International Standard for thromboplastin recombinant human (rTF/09) and the International Standard for thromboplastin rabbit plain (RBT/05) were obtained from the National Institute for Biological Standards and Control (Potters Bar, UK). The following commercial thromboplastin reagents were used. Recombiplastin 2G was obtained from Instrumentation Laboratory (Bedford, MA, USA), Dade Innovin from Siemens Healthcare Diagnostics Products GmbH (Marburg, Germany), STA Hepato Quick, STA-SPA Plus and STA-Néoplastine R from Diagnostica Stago (Asnières, France), and Thrombotest from Axis Shield (Oslo, Norway). The following instruments were used for PT measurements: Sysmex CA-1500 (Sysmex Corporation, Kobe, Japan), ACL Top 700 (Instrumentation Laboratory, Bedford, MA, USA), and STA-Rack (Diagnostica Stago, Asnières, France). One freeze-dried plasma (#35) was obtained from Technoclone (Vienna, Austria). All reagents and instruments were used according to the manufacturers’ instructions.
Preparation of freeze-dried plasma pools
Venous blood of patients receiving VKA was collected in Sarstedt Monovette tubes containing 0.109 mol/L sodium citrate. The blood was centrifuged at 2800×g for 15 min. The platelet-poor plasma was removed and centrifuged for a second time at 2800×g for 15 min to ensure the thorough removal of residual platelets. The individual plasma donations were stored at −70 °C. The frozen plasma was thawed and the individual donations were pooled. The pools were buffered by dissolving N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES) in the plasma to a concentration of 0.02 mol/L. The pools were frozen and stored at −70 °C.
The frozen plasma pools were shipped on dry ice to the MCA Laboratory (Winterswijk, The Netherlands) for freeze-drying. Siliconized glass vials (diameter 22 mm) were obtained from Thijssens Packaging (Purmerend, The Netherlands). Siliconized rubber bungs were obtained from Aluglas (Uithoorn, The Netherlands). After thawing and homogenizing of the frozen plasmas, each plasma pool was dispensed in 1 mL aliquots in the vials. The precision of filling was <0.5% according to the MCA laboratory specifications. After freeze-drying of the plasma, the vials were closed with rubber bungs under vacuum. The rubber bungs were fixed onto the vials with aluminum caps. The vials were stored at −20 °C. Before use, the vials were equilibrated at room temperature, and each vial was reconstituted with 1 mL of water. Reconstituted plasmas were kept at room temperature, gently swirled, and used between 20 min and 3 h after reconstitution.
Preparation of deep-frozen plasma pools
Six deep-frozen citrate plasma pools were prepared. One frozen plasma pool was prepared out of 35 healthy adult individual donations (code #12-0). The other five plasma pools were prepared of patients receiving VKA, essentially as described previously [9]. These pools were made from samples collected according to the following INR ranges: 1.5–2.0 (code #12-1), 2.0–2.4 (code #12-2), 2.4–2.8 (code #12-3), 2.8–3.2 (code #12-4), and 3.2–3.6 (code #12-5). Each VKA plasma pool was prepared from 175 individual patient samples. The plasma pools were divided into 0.7 mL aliquots in capped 2 mL polypropylene tubes and stored at −70 °C. Before use, each sample was thawed in a water bath at 37 °C for 4 min and homogenized. Thawed samples were kept at room temperature and used within 3 h.
ISI calibration and INR assignment
ISI calibrations of commercial thromboplastin-instrument combinations (Table 1) were performed according to the WHO Guidelines in the years 2011–2014 [17]. In that time period, the previous International Standards for thromboplastins (i.e. RBT/05 and rTF/09) were used. Commercial rabbit tissue reagents were calibrated against RBT/05 and recombinant human tissue factor reagents against rTF/09. Each ISI calibration was performed on at least 5 working days. On each working day, venous blood samples were collected in Sarstedt Monovette tubes containing 0.109 mol/L sodium citrate. The blood was centrifuged at 2800×g for 15 min. Plasma was taken off the red-cell layer with a plastic pipette, kept in a polystyrene tube at room temperature, and analyzed within 5 h after blood collection. Single PT determinations were performed with each fresh plasma sample with each thromboplastin reagent. PT determinations with either International Standard were performed with the manual tilt tube technique. Different sets of plasmas were collected and analyzed on each working day. For each reagent’s ISI calibration, a total number of 20 fresh plasma samples of healthy adult individuals and 60 fresh samples of patients receiving long-term anticoagulation with VKA were analyzed. The standard deviation (SD) of the ISI (SDISI) for the International Standards rTF/09 and RBT/05 was the between-lab SD as reported previously [18], [19]. The SDISI for each commercial thromboplastin-instrument combination was calculated according to the propagation of errors [20]. The MNPT was estimated as:
Thromboplastin reagents, instruments, and calibration parameters.
| Thromboplastin reagent name | Reagent type | Instrument | International Standard used for ISI calibrationa | ISI | SDISI | MNPT, s | SD loge (MNPT) |
|---|---|---|---|---|---|---|---|
| rTF/09 | Recombinant, human | Manual (Op. #1) | – | 1.08 | 0.045 | 13.1 | 0.008 |
| rTF/09 | Recombinant, human | Manual (Op. #2) | – | 1.08 | 0.045 | 13.3 | 0.005 |
| RBT/05 | Rabbit brain, plain | Manual (Op. #1) | – | 1.15 | 0.057 | 16.4 | 0.005 |
| RBT/05 | Rabbit brain, plain | Manual (Op. #2) | – | 1.15 | 0.057 | 16.4 | 0.005 |
| Recombiplastin 2G | Recombinant, human | ACL-Top 700 | rTF/09 | 0.93 | 0.040 | 10.8 | 0.016 |
| Dade Innovin | Recombinant, human | Sysmex CA1500 | rTF/09 | 1.00 | 0.043 | 10.4 | 0.009 |
| STA Néoplastine R | Recombinant, human | STA-R | rTF/09 | 0.91 | 0.039 | 13.8 | 0.011 |
| STA Hepato Quick | Rabbit, combined | STA-R | RBT/05 | 0.84 | 0.042 | 27.8 | 0.019 |
| STA SPA Plus | Rabbit, combined | STA-R | RBT/05 | 0.89 | 0.045 | 21.9 | 0.014 |
| Thrombotest | Bovine, combined | Sysmex CA1500 | RBT/05 | 0.88 | 0.046 | 33.8 | 0.016 |
-
aInternational Standard used for ISI calibration of commercial thromboplastin reagent-instrument combination. SDISI is the standard deviation of the ISI used in the authors’ laboratory. SD loge MNPT is the standard deviation of the mean loge MNPT used in the authors’ laboratory. Op., operator.
in which PTnormal, i is the PT of a healthy adult individual i and n the number of healthy adult individuals [21]. For all commercial thromboplastins, we used fresh plasmas from 20 healthy adult individuals to estimate the MNPT (n=20). The SD of the mean loge MNPT was estimated as:
PTs of the frozen and freeze-dried plasma pools were determined in duplicate on 3 days. The mean loge PT was calculated for each plasma pool on each reagent-instrument combination. The mean loge PT was transformed to INR using ISI and the mean loge MNPT for each thromboplastin-instrument combination [17]:
The uncertainty of the INR, expressed as the SD of loge INR, was estimated using the SDISI, the SD of loge PT, and the SD of loge MNPT according to the formula:
in which m is the number of PT determinations in the pooled plasma sample for which the mean INR is to be calculated [22]. According to our experimental design, we used m=6. The lower and upper limits of the 95% confidence interval for the INR are exp(loge INR–2×SD loge INR) and exp(loge INR+2×SD loge INR), respectively.
Commutability assessment of frozen and freeze-dried plasma pools
Commutability assessment was performed according to the CLSI guideline EP30-A [7]. The same sets of 60 patients’ PTs were used for commutability assessment as had been used for ISI calibrations (see above). In this section, reagent #1 is the International Standard (either RBT/05 or rTF/09) and reagent #2 the commercial thromboplastin-instrument combination that had been calibrated against the corresponding International Standard. Log-transformed PTs determined with thromboplastin reagent #1 (loge PT1) were plotted against log-transformed PTs with thromboplastin reagent #2 (loge PT2). Orthogonal regression lines for the functional relationship loge PT1=a+b×loge PT2 and the SD about the line (SDL) were calculated from 60 native fresh VKA plasma samples, as described previously [17]. The perpendicular distance of a plasma pool point (x, y) to the orthogonal regression line was calculated with the formula d=(y−bx–a)/√(b 2+1). The residual of a plasma pool, i.e. its distance d to the line, was normalized by calculating the ratio of the distance d to the SDL. A plasma pool was considered commutable if the absolute value of its normalized residual was not greater than 2. This is equivalent to stating that the plasma pool is within the 95% prediction interval around the regression line [7].
Validation of certified plasma pools
Validation of a set of six deep-frozen plasma pools (nr. 12-0, 12-1, 12-2, 12-3, 12-4, and 12-5) for direct INR determination was performed according to the SSC/ISTH guidelines as described previously [1]. Briefly, certified INR values for this set were obtained with RBT/05 and rTF/09 by two experienced operators using the manual tilt tube technique. The set of six deep-frozen plasmas was used for direct INR determination with each of the commercial reagents. An orthogonal regression equation was calculated relating the assigned INR to the PT determined with the commercial reagent:
The INRs for fresh plasmas of 60 VKA patients obtained with the direct INR procedure were compared to the corresponding INRs for the same plasmas obtained with RBT/05 or rTF/09. The relative difference ∆ (in %) was calculated between the mean value with the direct INR procedure (INRC) and the mean value with either RBT/05 or rTF/09 (INRR), according to the formula
Results
ISI and MNPT for all reagents and instruments are shown in Table 1. In general, there was good agreement among the INRs for the frozen plasma pools assigned with the various reagents (Table 2). The INRs assigned with RBT/05 were (slightly) higher than the INRs assigned with the other reagents. Figure 1 shows the scatterplots of INRs calculated for patients’ fresh plasmas as well as the six frozen plasma pools. The frozen pools were close to the fresh plasmas in the scatterplots with rTF/09, but more distant in the plots with RBT/05.
Mean INRs for six deep-frozen plasma pools.
| Thromboplastin reagent | Plasma code number |
|||||
|---|---|---|---|---|---|---|
| #12-0 | #12-1 | #12-2 | #12-3 | #12-4 | #12-5 | |
| rTF/09 (operator #1) | 1.03 (0.99–1.07) | 1.52 (1.45–1.59) | 2.17 (2.02–2.33) | 2.55 (2.34–2.78) | 2.94 (2.68–3.22) | 3.35 (3.02–3.71) |
| rTF/09 (operator #2) | 1.02 (1.00–1.05) | 1.58 (1.52–1.65) | 2.14 (2.01–2.29) | 2.51 (2.32–2.72) | 2.87 (2.62–3.15) | 3.27 (2.95–3.61) |
| RBT/05 (operator #1) | 1.02 (1.00–1.04) | 1.76 (1.66–1.87) | 2.42 (2.21–2.64) | 2.72 (2.46–3.02) | 3.01 (2.69–3.36) | 3.38 (2.99–3.82) |
| RBT/05 (operator #2) | 1.05 (1.03–1.07) | 1.75 (1.65–1.86) | 2.31 (2.11–2.52) | 2.76 (2.49–3.06) | 2.97 (2.66–3.31) | 3.38 (2.99–3.82) |
| Recombiplastin 2G | 1.00 (0.97–1.03) | 1.53 (1.46–1.61) | 2.09 (1.94–2.24) | 2.44 (2.24–2.64) | 2.81 (2.56–3.09) | 3.21 (2.89–3.56) |
| Dade Innovin | 1.02 (1.00–1.04) | 1.56 (1.49–1.63) | 2.12 (1.98–2.26) | 2.48 (2.29–2.69) | 2.92 (2.66–3.21) | 3.29 (2.96–3.65) |
| STA Néoplastine R | 1.02 (1.00–1.05) | 1.68 (1.60–1.76) | 2.26 (2.10–2.43) | 2.59 (2.38–2.82) | 2.95 (2.68–3.25) | 3.33 (3.00–3.70) |
| STA Hepato Quick | 1.00 (0.97–1.04) | 1.65 (1.55–1.75) | 2.25 (2.07–2.45) | 2.61 (2.37–2.89) | 2.93 (2.63–3.28) | 3.37 (2.98–3.81) |
| STA SPA Plus | 0.99 (0.97–1.01) | 1.54 (1.47–1.62) | 2.10 (1.94–2.28) | 2.43 (2.21–2.66) | 2.75 (2.48–3.06) | 3.14 (2.79–3.54) |
| Thrombotest | 0.96 (0.93–0.99) | 1.56 (1.48–1.65) | 2.11 (1.94–2.29) | 2.40 (2.18–2.64) | 2.68 (2.41–2.99) | 2.97 (2.64–3.34) |
| Reference interval | 0.93–1.13 | 1.49–1.82 | 2.03–2.49 | 2.37–2.90 | 2.65–3.24 | 3.01–3.68 |
-
The 95% confidence intervals are given in brackets. The INR reference interval (lower row) is obtained as the mean of rTF/09 and RBT/05 values ±10%.
Scatterplots of the natural logarithms of INRs calculated for 60 fresh patient plasma samples (open symbols) and natural logarithms of mean INRs calculated for six frozen plasma pools coded 12-0, 12-1, 12-2, 12-3, 12-4, and 12-5, respectively (closed symbols).
INRs were calculated with MNPT and ISI shown in Table 1. Panel A: Innovin (horizontal axis) versus rTF/09; Panel B: Innovin (horizontal axis) versus RBT/05; Panel C: STA Hepato Quick (horizontal axis) versus rTF/09; Panel D: STA Hepato Quick (horizontal axis) versus RBT/05.
The pattern of INRs for the freeze-dried plasma pools was different (Table 3). The INRs assigned with Hepato Quick were all higher than the INRs with RBT/05 and with rTF/09. The INRs assigned with Thrombotest were lower than the INRs assigned with all other reagents.
Mean INRs for 10 freeze-dried plasma pools.
| Thromboplastin | Plasma code number |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| #25 | #26 | #28 | #29 | #30 | #31 | #32 | #33 | #34 | #35 | |
| rTF/09 (operator #1) | 3.89 (3.46–4.37) | 2.88 (2.63–3.15) | 3.03 (2.76–3.33) | 3.94 (3.51–4.42) | 2.36 (2.19–2.55) | 2.77 (2.54–3.02) | 3.65 (3.27–4.07) | 2.35 (2.18–2.53) | 3.39 (3.06–3.76) | 3.52 (3.17–3.92) |
| RBT/05 (operator #1) | 3.67 (3.23–4.18) | 2.91 (2.62–3.24) | 3.02 (2.71–3.38) | 3.85 (3.36–4.40) | 2.52 (2.29–2.77) | 2.93 (2.63–3.26) | 3.52 (3.11–3.99) | 2.25 (2.07–2.44) | 3.22 (2.86–3.62) | 3.45 (3.04–3.90) |
| Recombiplastin 2G | 3.79 (3.36–4.27) | 2.83 (2.58–3.11) | 2.99 (2.71–3.30) | 3.87 (3.43–4.37) | 2.39 (2.21–2.60) | 2.82 (2.56–3.09) | 3.63 (3.23–4.07) | 2.28 (2.11–2.46) | 3.29 (2.96–3.66) | 3.46 (3.10–3.87) |
| Dade Innovin | 3.83 (3.41–4.31) | 2.89 (2.63–3.17) | 3.12 (2.82–3.45) | 4.22 (3.72–4.78) | 2.50 (2.30–2.71) | 2.93 (2.67–3.22) | 3.81 (3.39–4.28) | 2.31 (2.14–2.49) | 3.46 (3.10–3.86) | 3.70 (3.30–4.15) |
| STA Néoplastine R | 3.85 (3.41–4.34) | 3.00 (2.72–3.31) | 3.20 (2.88–3.55) | 4.09 (3.61–4.63) | 2.51 (2.31–2.72) | 2.94 (2.66–3.24) | 3.73 (3.32–4.19) | 2.33 (2.15–2.52) | 3.35 (3.00–3.74) | 3.52 (3.15–3.93) |
| STA Hepato Quick | 4.05 (3.51–4.67) | 3.04 (2.71–3.40) | 3.32 (2.93–3.75) | 4.11 (3.56–4.74) | 2.62 (2.37–2.90) | 3.12 (2.78–3.50) | 3.87 (3.38–4.44) | 2.47 (2.24–2.72) | 3.58 (3.14–4.07) | 3.70 (3.24–4.22) |
| STA SPA Plus | 3.88 (3.37–4.46) | 2.78 (2.50–3.10) | 3.02 (2.69–3.39) | 3.91 (3.40–4.49) | 2.39 (2.18–2.62) | 2.82 (2.53–3.14) | 3.64 (3.19–4.16) | 2.21 (2.04–2.41) | 3.27 (2.89–3.69) | 3.39 (2.98–3.85) |
| Thrombotest | 3.36 (2.95–3.83) | 2.59 (2.34–2.88) | 2.73 (2.44–3.04) | 3.38 (2.96–3.85) | 2.15 (1.97–2.34) | 2.61 (2.36–2.90) | 3.21 (2.83–3.64) | 2.11 (1.94–2.29) | 3.02 (2.68–3.41) | 2.80 (2.50–3.13) |
| Reference interval | 3.40–4.16 | 2.61–3.18 | 2.72–3.33 | 3.51–4.28 | 2.20–2.68 | 2.57–3.13 | 3.23–3.94 | 2.07–2.53 | 2.97–3.64 | 3.14–3.83 |
-
The 95% confidence intervals are given in brackets. The INR reference interval (lower row) is obtained as the mean of rTF/09 and RBT/05 values ±10%.
The assessment of commutability according to CLSI showed a few frozen plasma pools with a normalized residual greater than 2 in various reagent comparisons (Table 4). In several cases, the residuals of all frozen pools had the same sign.
Normalized residuals for deep-frozen plasma pools.
| Thromboplastin reagent #1 | Thromboplastin reagent #2 | SDL | Normalized residuals for plasma code number |
|||||
|---|---|---|---|---|---|---|---|---|
| #12-0 | #12-1 | #12-2 | #12-3 | #12-4 | #12-5 | |||
| RBT/05 | Thrombotest | 0.057 | −0.51 | 0.61 | 1.04 | 1.02 | 0.93 | 1.20 |
| RBT/05 | STA Hepato Quick | 0.036 | 1.22 | 2.16 | 2.18 | 1.58 | 1.23 | 0.86 |
| RBT/05 | STA SPA Plus | 0.025 | −2.44 | 1.11 | 1.18 | 2.80 | 1.68 | 2.08 |
| RBT/05 | rTF/09 | 0.036 | −3.06 | 1.02 | 1.26 | 0.91 | 0.63 | 0.65 |
| rTF/09 | Innovin | 0.028 | 0.07 | −0.61 | 0.72 | 0.85 | 0.37 | 0.66 |
| rTF/09 | STA Néoplastine R | 0.029 | 2.42 | −1.00 | −0.12 | 0.28 | 0.26 | 0.25 |
| rTF/09 | Recombiplastin 2G | 0.022 | 3.54 | 2.89 | 2.20 | 1.96 | 1.43 | 1.07 |
-
SDL is the standard deviation about the orthogonal regression line.
All freeze-dried plasmas had normalized residuals smaller than 2 with one exception, i.e. plasma #35 which had a normalized residual of 2.12 for Thrombotest when compared to RBT/05 (Table 5).
Normalized residuals for freeze-dried plasma pools.
| Reagent #1 | Reagent #2 | Freeze-dried plasma code number |
|||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| #25 | #26 | #28 | #29 | #30 | #31 | #32 | #33 | #34 | #35 | ||
| RBT/05 | TT | 0.78 | 0.92 | 0.81 | 1.30 | 1.33 | 0.91 | 0.80 | 0.12 | 0.37 | 2.12 |
| RBT/05 | HQ | −1.21 | −0.07 | −1.08 | −0.60 | 0.02 | −0.48 | −1.15 | −1.06 | −1.34 | −0.65 |
| RBT/05 | SPA Plus | −0.71 | 0.97 | −0.01 | 0.45 | 0.67 | 0.80 | −0.32 | −0.75 | −0.18 | 0.88 |
| RBT/05 | rTF/09 | −0.21 | 0.24 | 0.12 | 0.45 | 0.68 | 0.98 | 0.01 | −1.41 | −0.49 | 0.17 |
| rTF/09 | Innovin | 0.66 | 0.12 | −0.48 | −1.31 | −1.15 | −1.17 | −0.73 | 0.60 | −0.23 | −0.88 |
| rTF/09 | Neo R | 0.15 | −0.67 | −1.05 | −1.10 | −0.73 | −1.02 | −0.55 | 0.99 | 0.41 | 0.10 |
| rTF/09 | R2G | 0.71 | 0.95 | 0.74 | 0.40 | 0.37 | −0.07 | 0.16 | 1.82 | 1.03 | 0.59 |
-
TT, Thrombotest; HQ, STA Hepato Quick; Neo R, STA Néoplastine R; R2G, Recombiplastin 2G.
For validation of the direct INR procedure with the set of six frozen pools, several sets of fresh patient plasmas were used. As there were differences in the assigned INR between the two International Standards rTF/09 and RBT/05, the direct INR equations for each commercial reagent were established using the assigned INR with rTF/09 and with RBT/05 separately. As expected, the mean INRs for the fresh plasmas calculated with the RBT/05 direct INR equation were higher than the corresponding mean INRs calculated with the rTF/09 direct INR equation (Table 6). In all cases, better agreement was observed with the direct INR procedure based on rTF/09 assigned INRs. There was a wide range of the SD of the slope of the direct INR regression line: the lower value (0.009) was observed in the relation Recombiplastin 2G-rTF/09, and the higher value (0.057) in the relation Innovin-RBT/05.
Equations for direct INR determination and validation of frozen pooled plasmas.
| Reagent | International Standard used for certification of frozen plasma pools | Equation for direct INR determination |
Mean INR of 60 fresh plasmas |
Relative difference ∆, % | |||
|---|---|---|---|---|---|---|---|
| a | b | S b | INRC | INRR | |||
| Recombiplastin 2G | rTF/09 | −2.217 | 0.940 | 0.009 | 2.52 | 2.42 | 4.1 |
| Recombiplastin 2G | RBT/05 | −2.158 | 0.940 | 0.047 | 2.67 | 2.42 | 10.0 |
| Innovin | rTF/09 | −2.354 | 1.005 | 0.013 | 2.81 | 2.81 | 0.1 |
| Innovin | RBT/05 | −2.296 | 1.005 | 0.057 | 2.99 | 2.81 | 6.1 |
| STA Néoplastine R | rTF/09 | −2.431 | 0.915 | 0.032 | 2.74 | 2.81 | −2.5 |
| STA Néoplastine R | RBT/05 | −2.373 | 0.915 | 0.021 | 2.91 | 2.81 | 3.6 |
| STA Hepato Quick | rTF/09 | −2.817 | 0.844 | 0.026 | 2.79 | 2.81 | −0.6 |
| STA Hepato Quick | RBT/05 | −2.759 | 0.845 | 0.020 | 2.94 | 2.81 | 4.7 |
| STA SPA Plus | rTF/09 | −2.790 | 0.910 | 0.016 | 2.79 | 2.74 | 2.1 |
| STA SPA Plus | RBT/05 | −2.731 | 0.910 | 0.034 | 2.94 | 2.74 | 7.3 |
| Thrombotest | rTF/09 | −3.200 | 0.916 | 0.040 | 2.95 | 2.78 | 5.8 |
| Thrombotest | RBT/05 | −3.143 | 0.917 | 0.013 | 3.13 | 2.78 | 11.8 |
-
The parameters a and b refer to Equation 5. S b is the standard deviation of b. INRC and INRR refer to the mean INR of 60 fresh patients’ samples obtained with the direct INR determination and with an International Standard, respectively.
Discussion
The present study was performed by a single laboratory, in which International Standards for recombinant human and rabbit brain thromboplastins were used. The International Standards were used with a manual tilt tube technique by two experienced operators who obtained very similar results for the MNPT (Table 1) as well as the mean INRs for the set of frozen plasma pools (Table 2). The good agreement between our two operators contrasts with the wide variation between operators observed in multi-center studies [23]. It is not surprising that there is good agreement between the two operators because they used the same equipment and the same harmonized technique. It has been recommended by the SSC/ISTH guidelines that three to five laboratories should be involved in the certification process for each set of plasmas [1]. The reason for this recommendation was to reduce the effect of considerable between-laboratory variation of ISI and MNPT and the resulting INR values. One weakness of the present study is that INR assignment was performed by only two operators in one laboratory, which may have resulted in a certain bias when compared to other laboratories. The uncertainty of the INR was estimated with Equation 4 which includes the interlaboratory SDISI derived from multicenter calibration studies [18], [19]. The mean INRs for the frozen plasmas obtained with RBT/05 were slightly higher than those obtained with rTF/09. In addition to the International Standards, six commercial reagents were used which had been calibrated against the International Standards by the same experienced operators. The INRs for the frozen plasmas obtained with the commercial reagents were similar to those assigned with the International Standards, but minor differences were observed. The mean INRs determined with SPA Plus and Thrombotest were lower than those obtained with the International Standard RBT/05. Although we did not perform a statistical test, 95% confidence intervals for the mean INR were overlapping in most cases. No overlap of 95% confidence intervals was observed for plasmas 12-0 and 12-1 with Thrombotest and RBT/05. Similar INR differences in lyophilized plasmas between bovine and rabbit thromboplastins have been observed in earlier studies [20], [24]. These results suggest that the clotting times determined with Thrombotest were shortened during the processing of fresh plasma to frozen pools and freeze-dried pools, resulting in lower INRs. In previous studies, we observed shortening of the Thrombotest clotting times of freshly collected blood during the first 6 h [25]. Apparently Thrombotest is more sensitive than other thromboplastin reagents to the pre-analytical activation of plasma in the tissue factor pathway. We did not analyze the individual clotting factors in the frozen and freeze-dried plasmas. It cannot be excluded that part of coagulation factor VII is activated during the preparation of frozen and freeze-dried plasma pools. Bovine thromboplastin is more sensitive to activated factor VII than rabbit and human thromboplastins in factor VII assays [26].
HEPES was added to plasma before freeze-drying to control the pH of the product. The concentration of added HEPES was relatively low, but it has been shown that HEPES can have differential effects on various coagulation reactions [27]. It cannot be excluded that part of the INR differences between reagents as shown in Table 3 are due to the effects of HEPES. Also the freeze-drying procedure itself may prolong normal and coumarin PTs [28].
For the commutability assessment according to the CLSI guideline, it was essential that the PTs of all plasmas (i.e. fresh clinical samples as well as the frozen and freeze-dried plasma pools) were determined with the same laboratory methods. As the manual tilt tube PTs of the fresh clinical samples were determined by either one of our two experienced operators, the manual tilt tube PTs of the frozen and freeze-dried plasma pools had to be determined by the same operators. It is surprising that application of the CLSI procedure showed that practically all freeze-dried plasmas were commutable (Table 5). Only one freeze-dried plasma (#35) was non-commutable when used with Thrombotest. The mean INR values assigned with the International Standards should be considered as the reference values, and deviations amounting to 10% are allowed according to the SSC/ISTH guidelines [1]. Applying the 10% deviation rule to the INRs obtained with the commercial systems, we observed that all values were within the reference interval, except Thrombotest values for six out of 10 freeze-dried plasmas which were lower than the reference interval. In other words, the values assigned with rTF/09 and RBT/05 were not suitable for use with Thrombotest. Although Thrombotest is no longer available, other bovine thromboplastin reagents are on the market and may be used with frozen or freeze-dried plasmas. Discrepant results obtained with certain reagents should be kept in mind when evaluating the results of EQA schemes [2]. Our results suggest that the CLSI procedure for assessment of commutability is not sensitive enough to identify plasmas with discrepant results.
Recently, a new process has been described to assess the commutability of a reference material as a calibrator in the calibration hierarchy of clinical laboratory measurement procedures [29], [30], [31], [32]. The new process for commutability assessment involves requirements for the experimental design and the statistical approach that are different from the guidelines published by CLSI [7]. Specifically, the new process requires all measurements to be performed in one run and at least two replicate measurements on the clinical samples [31]. The present study did not meet these requirements and could only be evaluated according to the CLSI document published in 2010 [7]. Future commutability assessments should use the design and statistical analysis described in references [30], [31], [32].
Mean INRs for fresh patient samples were used to assess the validity of the direct INR determination with the set of six frozen plasma pools (Table 6). We assessed the validity for rTF/09 assigned values and for RBT/05 assigned values separately. The relative differences between the direct INR procedure and the International Standard were acceptable for all commercial reagents if rTF/09 assigned values were used.
In 2016 the International Standards RBT/05 and rTF/09 were replaced by their successors RBT/16 and rTF/16 [23]. It is not known whether INRs of our frozen and freeze-dried plasmas, if determined with RBT/16 and rTF/16, would be the same as the values obtained with RBT/05 and rTF/09. Nevertheless, using a different set of freeze-dried plasmas, similar INRs were obtained with all four International Standards [23].
It is obvious that freeze-dried plasmas are more convenient for shipment than frozen plasmas, but we dispatched frozen plasma pools on dry ice successfully to other centers in the Netherlands for experimental local calibration [33].
We conclude that the set of six frozen plasma pools can be used for ‘direct’ INR determination, but agreement was better for values assigned with rTF/09 than with RBT/05.
Acknowledgments
The authors thank Ms. Charmane Abdoel and Ms. Evelina Witteveen for technical assistance and Mr. Jeroen Kraaij for assistance with the calculations. Fresh plasmas were kindly provided by the Thrombosis Service Leiden (Leiden, The Netherlands). We thank Dr. Dirk de Korte (Department of Product and Process Development, Sanquin Blood Bank, Amsterdam, The Netherlands) for his assistance in the preparation of the normal plasma pool. Freeze-dried plasma samples were kindly provided by the Section Coagulation of the Dutch Foundation for Quality Assurance in Medical Laboratories (SKML, Nijmegen, The Netherlands).
-
Author contributions: All authors have accepted responsibility for the entire content of this manuscript and approved its submission.
-
Research funding: This work was supported in part by the Federation of Netherlands Thrombosis Services and in part by the Section Coagulation of the Dutch Foundation for Quality Assurance in Medical Laboratories.
-
Employment or leadership: None declared.
-
Honorarium: None declared.
-
Competing interests: Authors state no conflict of interest.
-
Ethical approval: The study has been approved by the Leiden University Medical Center Ethics Committee.
References
1. Van den Besselaar AM, Barrowcliffe TW, Houbouyan-Reveillard LL, Jespersen J, Johnston M, Poller L, et al. On behalf of thesubcommittee on control of anticoagulation of the Scientific and Standardization Committee of the ISTH. Guidelines on preparation, certification, and use of certified plasmas for ISI calibration and INR determination. J Thromb Haemost 2004;2:1946–53.10.1111/j.1538-7836.2004.00970.xSearch in Google Scholar
2. Meijer P, Kynde K, van den Besselaar AM, van Blerk M, Woods TA. International normalized ratio (INR) testing in Europe: between-laboratory comparability of test results obtained by Quick and Owren reagents. Clin Chem Lab Med 2018;56:1698–703.10.1515/cclm-2017-0976Search in Google Scholar
3. Fattorini A, Pattarini E, Viganò S, Crippa L, D’Angelo A. Use of INR calibrator plasmas in the routine coagulation laboratory: a study of two thromboplastin reagents. Thromb Res 2012;130:e158–62.10.1016/j.thromres.2012.05.015Search in Google Scholar
4. Van den Besselaar AM. Artificially depleted plasmas are not necessarily commutable with native patient plasmas for International Sensitivity Index calibration and International Normalized Ratio derivation. J Thromb Haemost 2012;10:303–5.10.1111/j.1538-7836.2011.04582.xSearch in Google Scholar
5. Poller L, Keown M, Ibrahim S, van den Besselaar AM, Roberts C, Stevenson K, et al. Comparison of local International Sensitivity Index calibration and ‘Direct INR’ methods in correction of locally reported International Normalized Ratios: an international study. J Thromb Haemost 2007;5:1002–9.10.1111/j.1538-7836.2007.02452.xSearch in Google Scholar
6. Van den Besselaar AM, Houbouyan-Reveillard LL, Aillaud MF, Denson KW, Droullé C, Johnston M, et al. Multicenter evaluation of lyophilized and deep-frozen plasmas for assignment of the International Normalized Ratio. Thromb Haemost 1999;82:1451–5.10.1055/s-0037-1614854Search in Google Scholar
7. Clinical and Laboratory Standards Institute. Characterization and qualification of commutable reference materials for laboratory medicine; Approved Guideline 2010, CLSI document EP30-A (formerly C53-A). Wayne, PA: CLSI.Search in Google Scholar
8. Stavelin A, Rikshelm BO, Christensen NG, Sandberg S. The importance of reagent lot registration in external quality assurance/proficiency testing schemes. Clin Chem 2016;62:708–15.10.1373/clinchem.2015.247585Search in Google Scholar
9. Van den Besselaar AM, Abdoel CF, van Rijn CJ, van der Meer FJ, Cobbaert CM. Accuracy assessment of consecutive test strip lots for whole blood INR point-of-care instruments: clarifying the role of frozen plasmas. Clin Chem Lab Med 2019;57:1349–57.10.1515/cclm-2018-1054Search in Google Scholar
10. Owren PA. Thrombotest. A new method for controlling anticoagulant therapy. Lancet 1959;2:754–8.10.1016/S0140-6736(59)90857-8Search in Google Scholar
11. Laharrague P, Carrières J, Cambus JP, Boneu B, Biermé R. Évaluation du thrombotest d’Owren dans la surveillance des traitements par antivitamine K. Ann Biol clin 1983;41:123–7.Search in Google Scholar
12. Morrison M, Caldwell A, McQuaker G, Fitzsimons EJ. Discrepant INR values: a comparison between Manchester and Thrombotest reagents using capillary and venous samples. Clin Lab Haemat 1989;11:393–8.10.1111/j.1365-2257.1989.tb00238.xSearch in Google Scholar PubMed
13. Van den Besselaar AM. Use of lyophilized calibration plasmas and control blood for international normalized ratio calculation in external quality assessment of the prothrombin time. Am J Clin Pathol 1994;102:123–710.1093/ajcp/102.1.123Search in Google Scholar PubMed
14. Poller L, Barrowcliffe TW, van den Besselaar AM, Jespersen J, Tripodi A, Houghton D. The European Concerted Action on Anticoagulation (ECAA) evaluation of a set of lyophilized normal plasmas to establish the normal prothrombin time for coagulometer systems. Thromb Haemost 1998;79:122–8.10.1055/s-0037-1614246Search in Google Scholar
15. Uetsuka Y, Sakuragawa N, Wada H, Aizawa Y, Aosaki M. ISI/INR system in Japan: experience from simultaneous measurement of the same plasma at four different laboratories. Semin Thromb Hemost 2000;26:109–1210.1055/s-2000-9812Search in Google Scholar PubMed
16. Van den Besselaar AM, Tripodi A. Is there a need for replacement of the International Reference Preparation for thromboplastin, bovine, combined (OBT/79)? J Thromb Haemost 2005;3:2365–6.10.1111/j.1538-7836.2005.01594.xSearch in Google Scholar PubMed
17. WHO Expert Committee on Biological Standardization. Guidelines for thromboplastins and plasmas used to control oral anticoagulant therapy with vitamin K antagonists. WHO Technical Report Series No. 979, 2013:271–316.Search in Google Scholar
18. Tripodi A, Chantarangkul V, van den Besselaar AM, Witteveen E, Hubbard AR. International collaborative study for the calibration of a proposed international standard for thromboplastin, human, plain. J Thromb Haemost 2010;8:2066–8.10.1111/j.1538-7836.2010.03980.xSearch in Google Scholar PubMed
19. Chantarangkul V, van den Besselaar AM, Witteveen E, Tripodi A. International collaborative study for the calibration of a proposed international standard for thromboplastin, rabbit, plain. J Thromb Haemost 2006;4:1339–45.10.1111/j.1538-7836.2006.01910.xSearch in Google Scholar PubMed
20. Van den Besselaar AM, Bertina RM. Multi-center study of thromboplastin calibration precision – influence of reagent species, composition, and international sensitivity index (ISI). Thromb Haemost 1993;69:35–40.10.1055/s-0038-1651544Search in Google Scholar
21. Van den Besselaar AM, Lewis SM, Mannucci PM, Poller L. Status of present and candidate international reference preparations (IRP) of thromboplastin for the prothrombin time. Thromb Haemost 1993;69:85.10.1055/s-0038-1651553Search in Google Scholar
22. Van den Besselaar AM, Witteveen E, van der Meer FJ. Uncertainty of International Sensitivity Index and International Normalized Ratio. J Thromb Haemost 2013;11:1615–7.10.1111/jth.12311Search in Google Scholar PubMed
23. Van den Besselaar AM, Chantarangkul V, Angeloni F, Binder NB, Byrne M, Dauer R, et al. International collaborative study for the calibration of proposed International Standards for thromboplastin, rabbit, plain, and for thromboplastin, recombinant, human, plain. J Thromb Haemost 2018;16:142–9.10.1111/jth.13879Search in Google Scholar PubMed
24. Van den Besselaar AM. Comparison of lyophilized plasmas with fresh plasmas for calibration of thromboplastin reagents in oral anticoagulant control. Br J Haematol 1996;93:437–44.10.1046/j.1365-2141.1996.5011041.xSearch in Google Scholar PubMed
25. Van den Besselaar AM, van Halem-Visser LP, Loeliger EA. The use of evacuated tubes for blood collection in oral anticoagulant control. Thromb Haemost 1983;50:676–7.10.1055/s-0038-1665285Search in Google Scholar
26. Poggio M, Tripodi A, Mariani G, Mannucci PM. Factor VII clotting assay: influence of different thromboplastins and factor VII-deficient plasmas. CISMEL Study Group. Thromb Haemost 1991;65:160–4.10.1055/s-0038-1647476Search in Google Scholar
27. Roberts PS, Hughes HN, Fleming PB. The effects of hepes buffer on clotting tests, assay of factors V and VIII and on the hydrolysis of esters by thrombin and thrombokinase. Thromb Haemost 1976;35:202–10.10.1055/s-0038-1647945Search in Google Scholar
28. Poller L, Keown M, Shepherd SA, Shiach CR, Tabeart S. The effects of freeze drying and freeze drying additives on the prothrombin time and the international sensitivity index. J Clin Pathol 1999;52:744–8.10.1136/jcp.52.10.744Search in Google Scholar PubMed PubMed Central
29. Young IS. The enduring importance and challenge of commutability. Clin Chem 2018;64:421–3.10.1373/clinchem.2017.284216Search in Google Scholar PubMed
30. Miller WG, Schimmel H, Rej R, Greenberg N, Ceriotti F, Burns C, et al. IFCC Working Group Recommendations for Assessing Commutability Part 1: General experimental design. Clin Chem 2018;64:447–54.10.1373/clinchem.2017.277525Search in Google Scholar PubMed PubMed Central
31. Nilsson G, Budd JR, Greenberg N, Delatour V, Rej R, Panteghini M, et al. IFCC Working Group Recommendations for Assessing Commutability Part 2: Using the difference in bias between a reference material and clinical samples. Clin Chem 2018;64:455–64.10.1373/clinchem.2017.277541Search in Google Scholar PubMed PubMed Central
32. Budd JR, Weykamp C, Rej R, MacKenzie F, Ceriotti F, Greenberg N, et al. IFCC Working Group Recommendations for Assessing Commutability Part 3: Using the calibration effectiveness of a reference material. Clin Chem 2018;64:465–74.10.1373/clinchem.2017.277558Search in Google Scholar PubMed
33. Van den Besselaar AM. Adjustment of ISI calibration of Innovin. Ned Tijdschr Klin Chem Labgeneesk 2007;32:224–7.Search in Google Scholar
©2020 Antonius M.H.P. van den Besselaar and and Christa M. Cobbaert, published by De Gruyter, Berlin/Boston
This work is licensed under the Creative Commons Attribution 4.0 International License.
