Optimization of Plasma Fluorogenic Thrombin-Generation Assays

(1)

Optimization of Plasma Fluorogenic Thrombin-Generation

Assays

Wayne L. Chandler, MD, and Mikhail Roshal, MD, PhD

Key Words: Thrombin generation; Fluorogenic; Tissue factor; Plasma; Coagulation

DOI: 10.1309/AJCP6AY4HTRAAJFQ

A b s t r a c t

We optimized fluorogenic thrombin-generation assays with regard to sample volume, calibration, analytic corrections, and activation reagents. Lower sample volumes (40 vs 80 µL) were associated with better recovery of thrombin activity, lower interference due to absorbance of light, and higher total thrombin generation (area under the curve), even using internal standards to calibrate plasma samples. With lower sample volumes, there was no advantage to internal calibration of samples without obvious interference (hemolysis). Previously developed corrections for measured vs expected fluorescence units, residual thrombin–α₂-macroglobulin activity, and hemolysis improved the analytic accuracy of the assay. An optimized assay with a 40-µL sample volume, analytic corrections, and a corn trypsin inhibitor to block contact activation showed that 0.6 pmol/L tissue factor activator was better than 5 pmol/L at differentiating healthy subjects from patients with sepsis while demonstrating good reproducibility (area under the curve, 4% within-run and 7% between-run coefficient of variation).

Thrombin-generation assays measure the amount of active thrombin produced in plasma or whole blood after recalcifica-tion. Thrombin generation can be initiated by tissue factor– bearing microparticles or other substances in the sample itself (native thrombin generation) or by addition of exogenous acti-vators such as tissue factor and phospholipids. Older methods measured the formation of thrombin-antithrombin complexes or other thrombin markers as a measure of thrombin genera-tion. This method worked but was time-consuming, manually intensive, and not suited for the evaluation of large numbers of clinical samples. Subsequently, methods based on thrombin-sensitive fluorogenic or chromogenic peptide substrates have been developed that offer greater applicability to clinical stud-ies. Fluorogenic assays have the advantages that they do not require addition of fibrin polymerization inhibitors and can be calibrated to absolute thrombin-generation rates (nmol/L of active thrombin per minute) rather than a percentage of nor-mal plasma. Prospective studies have shown that these assays have the potential for predicting the risk of thrombosis.1,2

Currently, there are 2 commercially available fluoro-genic thrombin-generation assays, the Calibrated Automated Thrombogram, developed by Hemker et al3_{and marketed}

by Thrombinoscope, Maastricht, the Netherlands, and the Technothrombin TGA method marketed by Technoclone, Vienna, Austria. Although similar in overall design, the details of these 2 methods are different, and many modifica-tions have been described in the literature.2-8_{The purpose of}

this study was to optimize fluorogenic thrombin-generation assays with regard to sample volume, calibration, activation reagents, and corrections for hemolysis and other problems. We did not attempt to repeat all prior evaluations of these assays. When an extensive evaluation had already been done,

by guest on January 29, 2016

http://ajcp.oxfordjournals.org/

(2)

we confirmed the results and evaluated whether it should be incorporated in an optimized method.

Materials and Methods Human Subjects

The study was approved by the University of Washington (Seattle) Human Subjects Review Committee. Informed consent was obtained from subjects. Blood samples were anticoagulated with 0.109 mol/L citrate. Citrated blood was centrifuged at 4,400g for 2 minutes at room temperature. Platelet-poor plasma was decanted and frozen at –80°C. Materials

Lipidated recombinant human tissue factor (RecombiPlasTin) was obtained from Instrumentation Lab, Bedford, MA. Corn trypsin inhibitor was obtained from EMD Biosciences, San Diego, CA. L-α-phosphatidylcholine and L-α-phosphatidylserine were obtained from Avanti Polar Lipids, Alabaster, AL. Phospholipid vesicles com-posed of 0.8 mmol/L phosphatidylcholine and 0.2 mmol/L phosphatidylserine in 20 mmol/L HEPES, 100 mmol/L sodium chloride, 0.2 g/L sodium azide, pH 7.5, were pre-pared by drying the phospholipids under dry nitrogen, then strong vacuum, followed by suspension in HEPES buffer and sonication. Pooled normal plasma anticoagulated with 0.109 mol/L citrate was obtained from Precision Biologics, Dartmouth, Canada. Thrombin calibrator (865 nmol/L) and thrombin-sensitive fluorogenic substrate (Z-Gly-Gly-Arg-AMC) calcium reagent were obtained from Technoclone through DiaPharma Group, Columbus, OH. Thrombin–α₂ -macroglobulin calibrator (equivalent to 625 nmol/L active thrombin)3_{and Z-Gly-Gly-Arg-AMC calcium reagent were}

obtained from Thrombinoscope through Diagnostica Stago, Parsippany, NJ.

Thrombin-Generation Assay

Both fluorogenic thrombin-generation assays combine 3 solutions, a plasma sample, an activation reagent, and a combined fluorogenic substrate and calcium solution to start the reaction. When only buffer is added as the activation reagent (termed native thrombin generation), initiation of thrombin generation is dependent on procoagulants present in the plasma.

All samples and reagents were warmed to 37°C before use. The activation reagent consisted of thrombin-generation buffer (20 mmol/L HEPES, 140 mmol/L sodium chloride, 5 mg/mL bovine serum albumin, 0.02% sodium azide, pH 7.35) containing variable amounts of lipidated tissue factor, phospholipid vesicles, and corn trypsin inhibitor to block

contact activation. In the Technoclone assay, 40 µL of plasma was added to 10 µL of activation reagent followed by 50 µL of calcium–fluorogenic substrate reagent (7.5 mmol/L calcium and 0.5 mmol/L substrate final reaction concentra-tions) to start the reaction. In the Thrombinoscope assay, 80 µL of plasma was added to 20 µL of activation reagent followed by 20 µL of calcium–fluorogenic substrate reagent (16.7 mmol/L calcium and 0.42 mmol/L substrate final reac-tion concentrareac-tions). Fluorescence units were monitored at 37°C at 1-minute intervals for 90 minutes using a Synergy HT microplate reader (Bio-Tek Instruments, Winooski, VT) set at an excitation wavelength of 360 nm and an emission wavelength of 460 nm. Results included determination of lag time in minutes, defined as the interval from the start of substrate/calcium addition until formation of 10 nmol/L of thrombin,3_{peak height for thrombin generation in nmol/L,}

peak height time in minutes, velocity index or peak rate of thrombin generation [peak thrombin/(peak time – lag time)], and area under the curve (AUC), defined as the sum of thrombin concentrations from 1 to 90 minutes (nmol/L × min). The AUC has also been referred to as the endogenous thrombin potential (ETP).3_{Correction of measured}

fluores-cence units for the inner filter effect and substrate exhaus-tion and correcexhaus-tion for residual thrombin–α₂-macroglobulin complex formed during the reaction were as previously described.3_{Absorbance in plasma at 360 and 460 nm was}

determined in a clear microtiter plate using the same sample volume used in the thrombin-generation assay.

Statistics

Results are given as the mean ± SD unless otherwise indicated. Statistical comparisons used paired t tests for paired samples and 2-way unpaired t tests for unpaired samples.

Results

Comparison of the Technoclone and Thrombinoscope Methods

zFigure 1z shows a comparison of the AUC results for

the Technoclone method run per manufacturer’s recommen-dations (thrombin calibrator in buffer, no fluorescence unit correction, no residual thrombin–α₂-macroglobulin correc-tion) vs the Thrombinoscope method run per manufacturer’s recommendations (thrombin–α₂-macroglobulin calibrator added to plasma samples, correction for inner filter effect, correction for residual thrombin–α₂-macroglobulin) in sam-ples without obvious hemolysis or other interference. On average, the Technoclone method gave higher AUC results for most samples.

(3)

Effect of Plasma Volume

The 2 assays use different amounts of plasma in their final reactions. The Technoclone assay uses 40 µL of plasma in a 100-µL total volume (40% plasma), whereas the Thrombinoscope assay uses 80 µL of plasma in a 120-µL total volume (67% plasma). zFigure 2z shows that as the percentage of plasma volume in the assay decreases, thrombin generation goes up, peaking when the plasma volume is about 20% to 30% and then falling for lower plasma volumes. The pattern was similar for substrates from either company.

Calibration of Thrombin Generation

zTable 1z shows a comparison of the purified active

thrombin calibrator from Technoclone with the thrombin–α₂ -macroglobulin complex calibrator from Thrombinoscope. The specific activity of the 2 calibrators was similar in both assays, but each company’s calibrator showed slightly more activity in its own assay. In the Thrombinoscope assay, the Thrombinoscope calibrator in buffer had a specific activity 6% higher than the Technoclone calibrator in buffer. In the Technoclone assay, the Thrombinoscope calibrator in buf-fer had a specific activity 20% lower than the Technoclone calibrator in buffer.

Effect of Corrections

The Thrombinoscope assay uses 3 corrections to improve the accuracy of estimated active thrombin concentrations gen-erated in plasma that are not used in the Technoclone method: (1) a correction for inner filter effect and substrate exhaus-tion, (2) a correction for residual thrombin–α₂-macroglobulin activity in the sample, and (3) a correction for hemolysis and other interfering factors.3,4_{We evaluated the effect of these}

corrections on the Technoclone method.

When thrombin generation is high, measured fluorescence unit levels may be underestimated owing to a combination of substrate exhaustion and the inner filter effect.3_{We found}

that as substrate was consumed, measured fluorescence units underestimated expected fluorescence in the Technoclone assay zFigure 3z. Based on this observation, we determined the fluorescence unit correction for our fluorescence reader and confirmed that the correction was a function of the

0 1,000

0 1,000 2,000 3,000 Technoclone AUC (nmol/L × min)

Thrombinoscope AUC (nmol/L × min) 4,000 5,000 2,000 3,000 4,000 5,000 0 20 0 10 20 30 40 50 60 70 80 Percent Plasma in Final Reaction Volume

Technoclone assay standard volume Thrombinoscope assay standard volume Percent of Maximum AUC 40 60 80 100 120

zFigure 1z Comparison of thrombin-generation area under the

curve (AUC) for the Technoclone and Thrombinoscope assays in 38 samples. Stimulated thrombin generation was initiated by addition of 0.6 pmol/L human recombinant tissue factor, 1 µmol/L phospholipid, and 40 µg/mL corn trypsin inhibitor (final reaction mixture).

zFigure 2z Effect of the percentage of plasma in the assay

on the area under the curve (AUC). Filled circles and solid line, Technoclone assay; open squares and dashed line, Thrombinoscope assay. Arrows show the standard plasma volumes used in each commercial assay. Stimulated thrombin generation was initiated by addition of 0.6 pmol/L human recombinant tissue factor, 1 µmol/L phospholipid, and 40 µg/mL corn trypsin inhibitor (final reaction mixture).

zTable 1z

Technothrombin vs Thrombinoscope Calibrator

Assay*

Calibrator Thrombinoscope Technoclone

Technoclone: purified thrombin in buffer 7.41 ± 0.66 8.80 ± 0.78

Thrombinoscope: thrombin–α2-macroglobulin in buffer 7.86 ± 0.46 7.05 ± 0.43 *_{Specific activity of each calibrator given as mean ± SD fluorescence units per minute per nanomole per liter of thrombin in the final reaction mixture based on the reported}

equivalent active thrombin concentration in the calibrator.

(4)

measured fluorescence units only, not the thrombin concen-tration.3_{Figure 3 shows that on average, the peak thrombin}

generation was 22% ± 10% higher when the fluorescence unit correction was applied.

During the thrombin-generation assay, part of the throm-bin formed throm-binds to α₂-macroglobulin in plasma and accumu-lates as thrombin–α₂-macroglobulin complexes that remain active against the fluorogenic substrate but do not activate coagulation, leading to a stable level of apparent residual thrombin activity at the end of the reaction.4_{We determined}

that this residual thrombin–α₂-macroglobulin activity was present in the Technoclone assay zFigure 4z, leading to an overestimate of the AUC. The correction for residual thrombin–α₂-macroglobulin activity had only a minor effect on peak thrombin generation, producing results that were

on average 2% lower than the uncorrected values with high correlation (r2_{= 0.99), but had a greater effect on the AUC,}

which was, on average, 26% ± 7% lower (r2_{= 0.91).}

Correction for Hemolysis

Hemolysis, icterus, and other factors that increase the absorbance in the sample can result in a falsely low AUC owing to absorbance of the excitation and emission light in the sample. Hemker et al3_{proposed a method for}

correct-ing absorbance interference to the AUC or ETP by addcorrect-ing a known amount of exogenous thrombin–α₂-macroglobulin to each sample as an internal calibrator. The Thrombinoscope method recommends calibrating every sample using exog-enous thrombin–α₂-macroglobulin, even the samples showing no evidence of hemolysis or other interference.3

0 5,000 0 20 40 60 80 100 Time (min) Fluorescence Units 10,000 15,000 20,000 25,000 30,000 0 200 0 200 400 600 800

Peak Thrombin Generation Measured Fluorescence Units (nmol/L)

Peak Thrombin Generation

Corrected Fluorescence Units (nmol/L)

400 600 800 0 5,000 0 5,000 10,000 15,000 20,000 Measured Fluorescence Units

Corrected Fluorescence Units

10,000 15,000 20,000 25,000 30,000 A B

C zFigure 3z Measured vs predicted fluorescence generation.

A, Increasing amounts of thrombin calibrator (sample concentrations, 54, 108, 216, and 432 nmol/L) were added to substrate and buffer in the Technoclone assay followed by measurement of fluorescence for 90 minutes at 37°C. Solid lines show the initial slope for the first 5 minutes representing the theoretical slope if inner filter effect and substrate depletion were not occurring. The filled circles represent measured fluorescence units over time. B, Plot of measured fluorescence units vs expected or corrected fluorescence units based on the slope of the reaction during the first 5 minutes for all data shown in A. C, Comparison of tissue factor–stimulated peak thrombin generation (PTG) values based on measured vs corrected fluorescence unit values: Corrected PTG = (1.35 × Measured PTG) – 20; r2_{= 0.983; n = 98. Stimulated thrombin}

generation was initiated by addition of 0.6 pmol/L human recombinant tissue factor, 1 µmol/L phospholipid, and 40 µg/mL corn trypsin inhibitor (final reaction mixture).

(5)

In the Thrombinoscope assay, background absorbance was twice as high as in the Technoclone assay (80- vs 40-µL sample volumes), resulting in more interference in that assay, even in samples with no apparent hemolysis or other prob-lems. In the Thrombinoscope assay (80-µL sample volume), the thrombin–α₂-macroglobulin calibrator showed 71% ± 5% as much activity in normal plasma as it did in buffer (n = 12). In the Technoclone assay (40-µL sample volume), the thrombin–α₂-macroglobulin calibrator showed 96% ± 4% as much activity in normal plasma as it did in buffer (n = 12). In the Technoclone assay using pooled normal plasma (PNP), the within-run coefficient of variation was 5% for the AUC and 5% for the thrombin–α₂-macroglobulin calibrator slope but 8% for the AUC calibrated using thrombin–α₂ -macroglobulin added to each plasma sample (n = 8 for each).

In plasma samples without evidence of hemolysis, icterus, or other color changes, calibration in each sample with exog-enous thrombin–α₂-macroglobulin did not improve the pre-cision and had only minor effects on the accuracy of AUC measurements using the Technoclone assay with a 40-µL sample volume but was useful in the Thrombinoscope assay to improve calibration accuracy.

To evaluate the effect of hemolysis on the assay, increas-ing amounts of hemoglobin up to 700 mg/dL were added to normal plasma. The AUC was reduced in direct proportion to the concentration of hemoglobin in the sample. Addition of thrombin–α₂-macroglobulin as an internal calibrator in the hemolyzed plasma samples corrected the AUC on aver-age to 98% ± 5% of the original unhemolyzed value (n = 6). We compared the correction predicted by calibration with 0

20

0 20 40 60 80 100

Raw thrombin generation Corrected thrombin generation Estimated thrombin–α2 -macro-globulin Time (min) Thrombin (nmol/L) 40 60 80 100 120 140 0 200 0 200 400 600 800

Peak Thrombin Generation Original Uncorrected (nmol/L) Peak Thrombin Generation Corrected for

Thrombin-α2 -Macroglobulin (nmol/L) 400 600 800 0 2,000 0 2,000 4,000 6,000 8,000 10,00012,000 AUC Original Uncorrected (nmol/L × min)

AUC Corrected fo r Thrombin – α2 -Macroglobulin (nmol/L × min) 4,000 6,000 8,000 10,000 12,000 A B

C zFigure 4z Correction for residual thrombin–α2-macroglobulin

complex. A, Typical thrombin-generation curve showing correction for residual thrombin–α2-macroglobulin complex;

dashed line, original thrombin-generation data showing residual thrombin activity at the end of the assay due to thrombin–α2-macroglobulin complex formed in plasma; solid

line, final data after correction to remove contribution by thrombin–α2-macroglobulin complex formed in plasma; dotted

line, estimated thrombin–α2-macroglobulin complex formed

during the assay. B, Effect of correction for residual thrombin–

α2-macroglobulin complex formed in plasma on peak thrombin

generation in the Technoclone assay (r2_{= 0.999; n = 92).}_C_,

Effect of correction for residual thrombin–α2-macroglobulin

complex formed in plasma on the area under the curve (AUC) in the Technoclone assay (Corrected AUC = [0.836 × Original AUC] – 513; r2_{= 0.91; n = 92). Stimulated thrombin}

generation was initiated by addition of 0.6 pmol/L human recombinant tissue factor, 1 µmol/L phospholipid, and 40 µg/ mL corn trypsin inhibitor (final reaction mixture).

(6)

thrombin–α₂-macroglobulin with absorbance measurements at the excitation (360 nm) and emission (460 nm) wavelengths in 56 samples containing varying levels of plasma hemoglo-bin and biliruhemoglo-bin. Sample absorbance at 360 nm (Abs 360 nm) and 460 nm (Abs 460 nm) showed correlations of r2₌

0.72 and 0.80, respectively, with the correction predicted by the thrombin–α₂-macroglobulin. The best correlation with the thrombin–α₂-macroglobulin correction was with what we termed the absorbance ratio (r2_{= 0.92)}_z_{Figure 5}_z_.

0 2

0 1 2 3 4 5

Thrombin–α2-Macroglobulin Correction Absorbance Ratio 4 6 8 10 12 14 16

zFigure 5z Correcting the area under the curve in 56 samples

with varying levels of hemolysis and icterus. Comparison of correction based on individual sample calibration using exogenous thrombin–α2-macroglobulin vs absorbance

ratio described in the text using sample absorbance measurements at 360 and 460 nm.

Corn Trypsin Inhibitor (�g/mL) 0 20 40 >40 Lag T ime (s) 0 20 40 60 80 100 120 0 20 0 20 40 >40 Corn Trypsin Inhibitor (�g/mL)

Peak Thrombin Generation (nmol/L)

40 60 80 100 120 A B

zFigure 6z Effect of corn trypsin inhibitor concentration on native thrombin generation (n = 8). A, Peak generation. B, Lag time,

defined as the interval from the start of substrate addition until the formation of 10 nmol/L of thrombin.3_{Boxes indicate the}

mean; error bars, SD.

Absorbance Ratio = Sample Abs 360 nm + Sample Abs 460 nm PNP Abs 360 nm PNP Abs 460 nm By using the best fit between the absorbance ratio and the thrombin–α₂-macroglobulin correction, we estimated an absorbance-based correction. In hemolyzed samples, the absorbance correction adjusted the AUC to 100% ± 6% of the original unhemolyzed value (n = 6).

Sample Activation

Both companies offer a variety of activation reagents ranging from physiologic buffer only for native thrombin generation to increasing amounts of tissue factor and phos-pholipid. Prior studies have shown that contact activation is occurring in the Thrombinoscope assay; it can be blocked by using corn trypsin inhibitor, and in the presence of corn trypsin inhibitor, healthy subjects showed little or no sponta-neous thrombin generation.8

We evaluated the effect of contact system inhibition with corn trypsin inhibitor on the Technoclone assay. Addition of corn trypsin inhibitor to plasma to block con-tact activation during native thrombin generation (no tissue factor or phospholipids added) prolonged the lag time and reduced the peak thrombin generation (0 vs 40 µg/mL; P = .012; paired t test) zFigure 6z. The amount of peak throm-bin generation due to contact activation (as estimated by percentage of activity blocked by corn trypsin inhibitor) varied widely among different healthy subjects, ranging from 16% to nearly 100%. With corn trypsin inhibitor present, little or no native thrombin generation occurred in normal plasma. When relatively high concentrations of tissue factor were added, the effect of contact system inhi-bition decreased as the tissue factor in the activator became

(7)

thrombin generation with no activator and tissue factor– stimulated thrombin generation with relatively low levels of tissue factor (0.6 pmol/L) and phospholipid (1 µmol/L). Reference Range and Imprecision

Native thrombin generation and tissue factor–stimulated thrombin generation were evaluated in samples from 25 healthy subjects zTable 3z. The majority of healthy subjects showed little or no native thrombin generation with corn trypsin inhibitor present. zTable 4z shows the within-run and between-run imprecision for tissue factor–stimulated peak thrombin generation and the AUC.

Discussion

Thrombin-generation assays have multiple potential uses, including estimation of hemorrhagic and thrombotic risk, monitoring of therapy, and detection of circulating procoagu-lants and microparticles.3,9_{Different protocols may be needed}

depending on the specific use. Two types of thrombin-gener-ation assays are currently available.5_Chromogenic

thrombin-generation assays use high concentrations of tissue factor or partial thromboplastin time reagents as activators, produce empiric results reported as a percentage of normal, require fibrin polymerization inhibitors, and have other disadvan-tages.6,10_{Fluorogenic thrombin-generation assays use lower}

levels of tissue factor activator and are calibrated to give an estimate of the actual concentration of active thrombin formed in the assay. We optimized fluorogenic thrombin-generation the primary initiator of thrombin generation. Owing to the

variable nature of contact activation, at intermediate levels of tissue factor and phospholipid, it became difficult to separate whether the tissue factor was the primary activator or the contact system. The effect of adding phospholipid was highly dependent on whether contact activation was blocked. Peak thrombin generation increased nearly 6-fold as phospholipid was increased from 0 to 4 µmol/L when no corn trypsin inhibitor was present and contact activation was occurring, but when corn trypsin inhibitor was present and contact activation blocked, addition of up to 4 µmol/L exogenous phospholipids increased peak thrombin genera-tion only 88% zFigure 7z.

Based on the preceding results, we selected a candidate assay for further testing of tissue factor and phospholipid: 40-µL sample volume to increase thrombin generation and reduce interference, correction for inner filter effect and residual thrombin–α₂-macroglobulin to improve accuracy, and addition of 40 µg/mL corn trypsin inhibitor to block contact activation. With this candidate assay, peak thrombin generation rose as more tissue factor was added to the reaction

zFigure 8z. The most rapid change in peak thrombin

genera-tion occurred between 0 and 1 pmol/L tissue factor in the final reaction, but peak thrombin generation continued to rise even at the maximum tissue factor concentration tested, 6 pmol/L. Similarly the AUC rose as more tissue factor was added, but the AUC appeared to plateau at about 1 pmol/L tissue factor with only a slow rise with higher tissue factor concentrations. With corn trypsin inhibitor present, addition of phospholipids increased peak thrombin generation only at the highest tis-sue factor concentrations. The maximum effect occurred at a phospholipid concentration of approximately 1 µmol/L.

By using 5 normal samples and 5 samples from patients with sepsis, we evaluated the effect on thrombin generation of 3 tissue factor–phospholipid concentrations (with corn trypsin inhibitor present): (1) no tissue factor or phospho-lipid (native thrombin generation), (2) 0.6 pmol/L tissue factor and 1 µmol/L phospholipid, and (3) 5 pmol/L tissue factor and 4 µmol/L phospholipid zTable 2z. Peak throm-bin generation and the AUC were significantly different between healthy subjects and patients with sepsis when native thrombin generation was used (no activator, contact activation blocked). Peak thrombin generation was also significantly different between healthy subjects and patients with sepsis when relatively low concentrations of tissue factor (0.6 pmol/L) and phospholipid (1 µmol/L) were used to activate thrombin generation. Peak thrombin generation and AUC were not significantly different between healthy subjects and patients with sepsis when a relatively high con-centration of activator was used (5 pmol/L tissue factor and 4 µmol/L phospholipid). Based on these results, we selected 2 options for further thrombin-generation testing: native

0 50

0 1 2 3 4 5

Phospholipid Concentration (�mol/L)

100 150 200 250

zFigure 7z Effect of phospholipid concentration on tissue

factor–stimulated peak thrombin generation. Thrombin generation was initiated by addition of 0.6 pmol/L human recombinant tissue factor. Filled circles, no corn trypsin inhibitor added; open circles, 40 µg/mL corn trypsin inhibitor added.

(8)

calibrators in buffer was similar but not identical. Substantial between-batch variability in calibrators has been reported.11

Further work is needed to define an international standard for calibration of thrombin generation. The thrombin–α₂ -macroglobulin calibrator has the advantage that it can be used as a calibrator and as an internal standard for samples with absorbance interference problems, as discussed later.

assays with regard to sample volume, calibration, activation reagents, and corrections for thrombin-generation assays with regard to hemolysis and other problems.

Thrombin-generation assays can be calibrated by using purified thrombin standards in buffer or thrombin–α₂ -macroglobulin standards added to plasma. The apparent specific activity of the 2 commercial thrombin-generation

0 200

0 1 2 3 4 5 6 7

400 600 800 0 2,000 0 1 2 4 �mol/L PL 3 �mol/L PL 2 �mol/L PL 1 �mol/L PL 0.6 �mol/L PL 0.3 �mol/L PL 0.1 �mol/L PL 0.0 �mol/L PL 4 �mol/L PL 3 �mol/L PL 2 �mol/L PL 1 �mol/L PL 0.6 �mol/L PL 0.3 �mol/L PL 0.1 �mol/L PL 0.0 �mol/L PL 6 pmol/L TF 4 pmol/L TF 3 pmol/L TF 2 pmol/L TF 1 pmol/L TF 0.6 pmol/L TF 0.3 pmol/L TF 0.0 pmol/L TF 6 pmol/L TF 4 pmol/L TF 3 pmol/L TF 2 pmol/L TF 1 pmol/L TF 0.6 pmol/L TF 0.3 pmol/L TF 0.0 pmol/L TF 3 4 5 6 7 AUC (nmol/L × min)

AUC (nmol/L

×

min)

Tissue Factor (pmol/L) 4,000 6,000 8,000 0 2,000 0 1 2 3 4 5 Phospholipid (�mol/L) Tissue Factor (pmol/L) Phospholipid (�mol/L)

4,000 6,000 8,000 0 200 0 1 2 3 4 5 400 600 800

zFigure 8z Effect of tissue factor (TF) and phospholipid (PL) concentrations on peak thrombin generation and area under the

curve (AUC). All reactions contained corn trypsin inhibitor, 40 µg/mL final concentration. The first column shows TF on the x-axis, and curves show PL concentrations. The second column shows PL concentration on the x-axis, and curves show TF concentrations.

(9)

require calibration in each sample, whereas lower plasma volume assays may use a single assay calibrator if the sample does not show absorbance interference (eg, hemolysis or icter-us). Running the sample in duplicate is still required because occasional samples showed aberrant curves owing to bubbles or other problems.3

As expected, use of higher sample volumes (80 µL) in the Thrombinoscope assay was associated with greater interference when hemolysis was present. We determined that hemolysis affects the Technoclone assay and that the Lower sample volumes in both assays were

associ-ated with increased total thrombin generation (AUC or ETP), down to a plasma volume of about 20% to 30% of the total reaction volume. This effect has been reported in another study and has been attributed to lower thrombin inhibition due to reduced antithrombin levels as the plasma volume in the assay is reduced.12

Prior studies reported that owing to absorbance of the excitation and emission light by substances in plasma, recov-ery of thrombin–α₂-macroglobulin calibrator added to plasma was approximately 30% lower than calibrator added to buf-fer and was highly variable between difbuf-ferent samples, even in the absence of hemolysis.13_{Based on this observation, it}

was suggested that thrombin generation must be calibrated in each plasma sample analyzed using a separate reaction with thrombin–α₂-macroglobulin calibrator added to plasma and that each plasma sample and plasma plus calibrator be run in duplicate to quadruplicate to reduce variability, result-ing in 4 to 8 reactions per sample analyzed.3_{We determined}

that when using an 80-µL sample volume in samples without obvious interference (hemolysis), the recovery of standard in normal pooled plasma was indeed reduced approximately 30% and showed up to 18% variability between different samples. However when a 40-µL volume was used in samples without obvious interference, calibration was similar in buf-fer or plasma with less variability between plasma samples. Thrombin-generation assays using higher plasma volumes

zTable 3z

Reference Ranges*

Native TG TF-Stimulated TG

Peak thrombin generation (nmol/L) 13 (1-72) 240 (143-369)

Lag time (min) 79 (32-90) 8 (6-10)

Peak time (min) 89 (49-90) 16 (13-22)

Velocity index (nmol/L/min) 2 (0-5) 29 (14-51)

Area under the curve (nmol/L × min) 179 (12-2,075) 4,522 (3,229-5,641)

*_{Range based on 25 healthy donors (12 women and 13 men; mean ± SD age, 32 ± 10 years; range, 19-56 years). Native thrombin generation (TG) run with corn trypsin inhibitor}

(40 µg/mL). Tissue factor (TF)-stimulated TG run with 0.6 pmol/L recombinant human tissue factor, 1.0 µmol/L phospholipid, and 40 µg/mL corn trypsin inhibitor. Data are shown as the median followed by the 4th to 96th percentile of the distribution. Area under the curve determined over 90 minutes.

zTable 4z

Imprecision for Tissue Factor–Stimulated

Thrombin-Generation Parameters*

Peak Thrombin Area Under the Curve

Generation (nmol/L) (nmol/L × min)

Within-run (n = 8) Mean 454 5,159 SD 18 204 CV (%) 4 4 Between-run (n = 15) Mean 404 4,553 SD 38 334 CV (%) 9 7 CV, coefficient of variation.

*_{Tissue factor–stimulated thrombin-generation run with final reaction concentrations}

of 0.6 pmol/L recombinant human tissue factor, 1.0 µmol/L phospholipid, and 40 µg/mL corn trypsin inhibitor. Area under the curve determined over 90 minutes.

zTable 2z

Effect of Activator Concentrations on Thrombin Generation in Clinical Samples*

Peak Thrombin Generation (nmol/L) Area Under the Curve (nmol/L × min)

Tissue Factor Phospholipid Healthy Patients With Healthy Patients With

(pmol/L) (µmol/L) Subjects Sepsis P† _Subjects _Sepsis _P†

0 0 58 ± 22 276 ± 90 .001 1,617 ± 719 4,181 ± 1,055 .003

0.6 1.0 198 ± 34 362 ± 107 .02 3,672 ± 394 4,587 ± 1,052 NS

5.0 4.0 534 ± 46 504 ± 132 NS 4,565 ± 349 5,228 ± 1,764 NS

NS, not significant.

*_{Data are given as mean ± SD.}

†_Unpaired_t_{test, healthy subjects vs patients with sepsis; n = 5 for each group.}

(10)

corrections for inner filter effect and residual thrombin–α₂ -macroglobulin to improve accuracy and addition of 40 µg/mL corn trypsin inhibitor to block contact activation. Calibration with thrombin–α₂-macroglobulin has the advantage that it can be used as an internal standard for samples with absorbance interference problems. As an alternative, the AUC can be corrected by using an absorbance ratio. Like others, we found that in the presence of corn trypsin inhibitor and lower dose tissue factor activator seemed to be more sensitive to differ-ences in thrombin generation between plasma samples.7,8,14

Tissue factor and phospholipid concentrations can be adjusted to suit specific clinical questions.

This assay depends on fluorescence measurements that are less standardized than absorbance measurements. Fluorescence can vary with the light source and detectors used in the instrument. An important question is, What aspects of this study apply to other microtiter plate fluorescence instru-ments? In setting up this assay on any instrument from any company, including the model used in this study, it is most important to calibrate thrombin generation using an active thrombin standard. If inner filter effect and substrate exhaus-tion correcexhaus-tions will be used, the correcexhaus-tion factors will need to be determined. The correction for residual thrombin–α₂ -macroglobulin activity is independent of the fluorescence instrument used. If exogenous thrombin–α₂-macroglobulin calibrator is used to correct the AUC for hemolysis or other interference, this is also independent of the fluorescence instrument. If an absorbance-based hemolysis correction is used, the correction will need to be determined for the instru-ment used because it is dependent on the original fluorescence measurements. Once all of these have been established, refer-ence range and imprecision should be determined based on the final reagents and instrument selected.

From the Department of Laboratory Medicine, University of Washington, Seattle.

Address reprint requests to Dr Chandler: Dept of Laboratory Medicine, Box 357110, University of Washington, Seattle, WA 98195.

References

1. Hron G, Kollars M, Binder BR, et al. Identification of patients at low risk for recurrent venous thromboembolism by measuring thrombin generation. JAMA. 2006;296:397-402. 2. van Hylckama Vlieg A, Christiansen SC, Luddington R, et al.

Elevated endogenous thrombin potential is associated with an increased risk of a first deep venous thrombosis but not with the risk of recurrence. Br J Haematol. 2007;138:769-774. 3. Hemker HC, Giesen P, Al Dieri R, et al. Calibrated automated

thrombin generation measurement in clotting plasma.

Pathophysiol Haemost Thromb. 2003;33:4-15.

4. Hemker HC, Beguin S. Thrombin generation in plasma: its assessment via the endogenous thrombin potential. Thromb Haemost. 1995;74:134-138.

thrombin–α₂-macroglobulin calibrator added to plasma could be used to correct the AUC, similar to prior reports for the Thrombinoscope assay.3_{In addition, we developed a sample}

absorbance ratio that could be used in place of an internal standard to provide a similar correction to the AUC in hemo-lyzed and icteric samples.

We determined that inner filter effect, substrate exhaus-tion, and residual thrombin–α₂-macroglobulin activity occur in the Technoclone assay, as previously described for the Thrombinoscope assay.3_{We further determined that}

cor-rections for measured vs expected fluorescence units and residual thrombin–α₂-macroglobulin activity can be used in the Technoclone assay to improve analytic accuracy.

Native thrombin generation is potentially useful for detecting circulating procoagulants in plasma. Variable levels of contact activation occurred in samples during thrombin generation in the Technoclone assay and have been reported in the Thrombinoscope assay.8_{To assess native thrombin}

gen-eration, contact activation must be blocked,8_{which has}

typi-cally been done by using corn trypsin inhibitor at 40 µg/mL.7,8

In the Technoclone assay, healthy subjects showed little or no thrombin generation in the presence of corn trypsin inhibitor.

Because most patients show little or no native thrombin generation, it cannot be used to assess the overall response of the system to a given stimulus. Most studies have used a fixed amount of activator, usually tissue factor and phos-pholipid, to stimulate thrombin generation. The optimal tissue factor–phospholipid concentration for clinical studies is still being evaluated; higher tissue factor concentrations produced less variability in the AUC and less sensitivity to contact activation, whereas lower concentrations were more sensitive to changes in thrombin generation.1,7,14_Prior

studies reported no difference in the plasma AUC and lower peak thrombin generation in control subjects vs patients with sepsis using 5 pmol/L tissue factor and no corn trypsin inhibitor.15_{By using corn trypsin inhibitor, we found that}

native and low-level tissue factor (0.6 pmol/L)–stimulated thrombin generation could be used to differentiate healthy subjects from patients with sepsis, but higher levels of tissue factor stimulation (5 pmol/L), which overwhelmed differ-ences in normal vs septic samples, could not be used. Others have reported that corn trypsin inhibitor plus low tissue factor stimulation could be used to differentiate pathologic from normal samples.7,8_{At present, tissue factor}

prepara-tions are highly variable in source material and specific activity. International standards are needed for tissue factor activators.

Both of the commercial assays had disadvantages. Based on the results, we developed an optimized assay. We selected a sample volume of 40 µL to increase thrombin generation and reduce interference and still allow use of commercial fluoro-genic substrate–calcium reagents. The optimized assay uses

(11)

11. van Veen JJ, Gatt A, Cooper PC, et al. Between-batch variation of calibrator activity can significantly influence fluorogenic measurement of thrombin generation. J Thromb Haemost.

2006;4:2514-2516.

12. Schols SE, Feijge MA, Lance MD, et al. Effects of plasma dilution on tissue-factor-induced thrombin generation and thromboelastography: partly compensating role of platelets.

Transfusion. 2008;48:2384-2394.

13. Hemker HC, Al Dieri R, De Smedt E, et al. Thrombin generation, a function test of the haemostatic-thrombotic system. Thromb Haemost. 2006;96:553-561.

14. Chantarangkul V, Clerici M, Bressi C, et al. Thrombin generation assessed as endogenous thrombin potential in patients with hyper- or hypo-coagulability. Haematologica.

2003;88:547-554.

15. Collins PW, Macchiavello LI, Lewis SJ, et al. Global tests of haemostasis in critically ill patients with severe sepsis syndrome compared to controls. Br J Haematol. 2006;135:220-227. 5. Devreese K, Wijns W, Combes I, et al. Thrombin generation

in plasma of healthy adults and children: chromogenic versus fluorogenic thrombogram analysis. Thromb Haemost.

2007;98:600-613.

6. Baglin T. The measurement and application of thrombin generation. Br J Haematol. 2005;130:653-661.

7. Dargaud Y, Luddington R, Gray E, et al. Effect of

standardization and normalization on imprecision of calibrated automated thrombography: an international multicentre study.

Br J Haematol. 2007;139:303-309.

8. Luddington R, Baglin T. Clinical measurement of thrombin generation by calibrated automated thrombography requires contact factor inhibition. J Thromb Haemost.

2004;2:1954-1959.

9. Keuren JF, Magdeleyns EJ, Govers-Riemslag JW, et al. Effects of storage-induced platelet microparticles on the initiation and propagation phase of blood coagulation. Br J Haematol.

2006;134:307-313.

10. Hemker HC, Giesen P, AlDieri R, et al. The calibrated automated thrombogram (CAT): a universal routine test for hyper- and hypocoagulability. Pathophysiol Haemost Thromb.

2002;32:249-253.