A modified international normalized ratio as an effective way of prothrombin time standardization in hepatology

Authors


  • Potential conflict of interest: Nothing to report.

Abstract

International Normalized Ratio (INR), which standardizes prothrombin time (PT) during oral anticoagulation, has been extended to standardize PT in liver diseases and is included in prognostic models such as the Model for End stage Liver Disease (MELD). However, mechanisms of PT prolongation in liver diseases differ from those involved in oral anticoagulation, and the thromboplastin reagents differ in their sensitivities to these 2 mechanisms. Our aim was to determine whether, in the calibration model for thromboplastins proposed by the World Health Organization, the use of plasmas from patients with liver diseases instead of plasmas from patients on oral anticoagulation could lead to a new INR specific for liver diseases (INR “LD”), achieving a real standardization of PT. First, 5 thromboplastins were calibrated against an international reference using 60 plasmas of patients with liver failure and, in a second step, the variation of PT reported as seconds, the ratio of patient PT to normal PT, INR, and INR“LD” was assessed in 34 other patients. MELD scores were calculated with the INR values obtained with the 5 thromboplastins. Only INR“LD” eliminated variability in PT results observed with the different thromboplastins. The discrepancy between MELD scores were up to 4 and 7 points in 52% and 17% of the patients, respectively. Conclusion: INR “LD” may provide a common international scale of PT reporting in hepatology. Its adoption would be an important step because of the significant impact on MELD score induced by interlaboratory variability in INR determination. (HEPATOLOGY 2007.)

Prothrombin time (PT) is the usual test for monitoring oral anticoagulation by vitamin K antagonists1 and is widely adopted to assess the severity of acute and chronic liver injury.2 PT is the recalcification time of citrated plasma in the presence of a thromboplastin reagent.3 The term thromboplastin refers to a complex mixture of tissue factor and phospholipids prepared from tissue extracts of animal or human origin. Recent thromboplastins contain recombinant tissue factor. PT is influenced by the levels of the following coagulation factors: I (fibrinogen), II, V, VII, and X, all factors being synthesized by the liver and, among them, 3 (II, VII, and X) being vitamin K dependent. Thromboplastins can vary markedly in their responsiveness to the defects induced by vitamin K antagonist therapy. As a result, the poor agreement between PT in different laboratories whatever the methods of reporting results4 has led to important differences in oral anticoagulant dosing.5, 6 Standardization of PT reporting in patients on oral anticoagulation has been improved by the calibration model for thromboplastins proposed by the World Health Organization (WHO).7 This system defines, for a specific thromboplastin reagent, the International Sensitivity Index (ISI) obtained by a calibration using plasmas of patients on stable oral anticoagulant doses, against a WHO international reference thromboplastin.7 The ISI is a numerical value that reflects the responsiveness of a given thromboplastin to reduction of the vitamin K–dependent coagulation factors; the lower the ISI value, the greater the sensitivity of the reagent. The recommended scale of reporting PT results is the International Normalized Ratio (INR), calculated as follows: INR = (patient PT/mean normal PT)ISI.7, 8 This standardization allows safe and effective dosing of vitamin K antagonists, independently of the sensitivity of the thromboplastin used.1

Thromboplastin responsiveness also varies widely with respect to the coagulation defects induced by liver failure. This variability may lead to large interlaboratory differences in PT results.9, 10 Because of its omnipresence for monitoring oral anticoagulation, the INR/ISI system has been inadvertently extended to standardize PT in liver failure.9 The INR/ISI mode of reporting is now widely recommended for the evaluation of acute liver failure11 and is incorporated in different models to predict significant fibrosis in patients with hepatitis C12, 13 and survival in patients with severe liver disease, such as the Model for End stage Liver Disease (MELD) score.14–16

However, the INR/ISI system has been meant and validated for monitoring oral anticoagulant therapy only. Some studies have previously reported that the use of INR for reporting PT in patients with liver disease fails to yield standardization.10, 17–19 A recent study has shown that variability between laboratory methods in determination of INR has a significant impact on the MELD score in patients listed for liver transplantation.20 In 2000, both the National Academy of Clinical Biochemistry and the Practice Guidelines Committee for the American Association for the Study of Liver Disease (AASLD) have made the following recommendations: “PT in seconds rather than the INR should be used to express results of PT in patients with liver disease, however this does not standardize results between laboratories” and “Additional research into standardization of PT reagents and use of derived indices in liver disease is needed.”21 We have previously demonstrated that, in liver failure, reporting PT in terms of percentage of normal could provide a common international scale of PT reporting.10 However, this mode of PT reporting has gained little popularity outside Europe, presumably because of the difficulties of the calibration system.

We hypothesized that the use of plasmas of patients with liver disease instead of plasmas from patients on oral anticoagulation in the calibration model for thromboplastins proposed by the WHO7 could lead to a PT standardization specific for liver disease. The aim of our study was to determine whether, in a new INR/ISI “liver disease”(“LD”) system, INR“LD” would achieve a real standardization of PT in liver diseases compared with INR or other modes of reporting.

Abbreviations

INR, international normalized ratio; ISI, international sensitivity index; LD, liver disease; MELD, Model for End stage Liver Disease; PT, prothrombin time; WHO, World Health Organization.

Patients and Methods

Blood Samples.

Blood samples were collected into Vacutainer tubes containing 0.109 M tri-sodium citrate (Becton Dickinson, Plymouth, UK). Platelet-poor plasma was prepared by centrifugation at 2,500g for 15 minutes at 20°C, aliquoted and stored at −70°C until assayed.

Calibration Procedure of Thromboplastins for the Definition of the Parameter ISI“LD”.

Calibration was performed according to the WHO guidelines for thromboplastins and plasma used to control oral anticoagulant therapy,22 with the following differences: the use of frozen instead of fresh plasmas and the use of plasmas from patients with liver disease instead of plasmas from patients on stable oral anticoagulation.

Sixty patients with chronic or acute liver failure were selected on the basis of their PT report as a ratio of the patient's PT to that of normal plasma and their level of factor V obtained with the routine thromboplastin of the laboratory (Thromborel S, Dade Behring). To cover the whole range of PT in liver failure, the selection of patients was as follows: 20 patients with a 1.16 ≤ PT ratio < 1.45 and with a factor V level < 75% (level 1); 20 patients with a 1.45≤ PT ratio < 1.90 and with a factor V level < 65% (level 2); 20 patients with a PT ratio ≥1.9, and with a factor V level < 50% (level 3). Twenty healthy subjects were selected among volunteers from hospital staff.

The 3rd International Standard for human, recombinant plain thromboplastin (rTF/95, ISI = 0.94) was obtained from a WHO laboratory for biological standard (Central laboratory of the Netherlands Red Cross blood transfusion service, Amsterdam, The Netherlands). Calibration of the 5 commercial thromboplastins described in Table 1 was performed according to the WHO guidelines22 with the rTF/95 as the reference thromboplastin on 10 working sessions, using for each session a different set of 8 plasmas (2 healthy subjects and 6 patients with liver failure).

Table 1. Characteristics of Thromboplastin Reagents
NumberNameManufacturerSourceISI*ISI (“local”)ISI“LD”
  • *

    ISI values were those provided by the manufacturers for photo-optical instrument except for thromboplastin 1, for which no instrument was mentioned.

1Neoplastin CIDiagnostica Stago (Asnières, France)Rabbit brain1.721.670.98
2Simplastin ExcelBioMerieux (Durham, NC, USA)Rabbit brain1.691.540.94
3SimplastinExcel SBioMerieux (Durham, NC, USA)Rabbit brain1.071.050.70
4Thromborel SDade Behring (Marburg, Germany)Human placenta0.981.030.84
5RecombiPlasTinInstrumentation Laboratory (Lexington, MA, USA)Human recombinant0.800.830.85

PT of each plasma was determined with rTF/95 by the manual tilt tube technique and with the 5 commercial thromboplastins on an automated photo-optical coagulometer ACL TOP (Instrumentation Laboratory). For each commercial thromboplastin, an ISI“LD” was calculated as described in the WHO guidelines.22 PT of the 20 healthy subjects plus the 60 patients with liver disease were plotted on a double logarithmic scale with rTF/95 on the vertical axis, and the commercial thromboplastin on the horizontal axis. The slope of the orthogonal regression line was used as ISI“LD.”

Study of the Influence of PT Reporting on the Test Standardization.

Thirty-four patients with liver failure different from those selected for the calibration procedure were included. Thirteen, 11, and 11 of these patients belonged to the previously described levels 1, 2, and 3, respectively.

PT were performed for each plasma sample with the 5 commercial thromboplastins on the coagulometer ACL TOP.

The mean normal prothrombin time was determined for each thromboplastin reagent by calculating the geometric mean of the PT results of 40 plasma samples collected from healthy individuals.

PT results were reported in seconds (PTs), in the ratio of the patient's PT to mean normal prothrombin time (PTr), in INR, and in INR“LD”. To take into account the thromboplastin/coagulometer combination in our laboratory, “local” ISI were determined for the 5 commercial thromboplastins. This was achieved by means of a lyophilized plasma calibration set (AK-Calibrant; Technoclone, Vienna, Austria) consisting of 1 normal and 3 anti–vitamin K pooled plasmas with assigned INR values. PT of the calibrant plasmas measured on the ACL TOP with 1 of the commercial thromboplastins were plotted against the assigned INR values according to the guidelines on preparation, certification, and use of certified plasma for ISI calibration and INR determination.23 The derived orthogonal regression line was used to determine local ISI of the 5 commercial thromboplastins.23 INR“LD” was calculated as follows: INR“LD” = (patient PT/mean normal prothrombin time)ISI “LD”.

MELD Score.

The MELD score was calculated for the 34 patients tested in the standardization study using the following formula: 9.57 × loge (creatinine mg/dL) + 3.78 × loge (bilirubin mg/dL) +11.2 × loge (INR) + 6.43.14 For each patient, MELD scores were calculated according to the 5 different INR values obtained with the 5 commercial thromboplastins.

Statistical Analysis.

Data were expressed as means, standard deviations, and ranges.

For each mode of PT reporting, statistical differences between mean values of PT results across the different thromboplastins used were determined by a 1-way analysis of variance (ANOVA) for repeated measures followed by the post-hoc Bonferroni's multiple comparison test. The same statistical analysis was used to determine statistical difference between mean values of MELD score calculated with the INR obtained with the 5 different thromboplastins.

Coefficients of variation (CV%) were calculated to represent within-subject variation of the PT results obtained with the 5 different thromboplastins.

Bland and Altman plots24 of differences between INR and mean INR and differences between INR“LD” versus mean INR“LD” obtained with 2 different thromboplastins were used to identify agreement between the methods. The 2 thromboplastins used in each graph were the most and the least sensitive to the coagulation defect induced by liver failure as assessed by the mean INR and INR“LD” obtained with the 34 patients with liver disease. The same plot analysis was used to test the agreement between the MELD scores calculated with the INR obtained with the most and least sensitive thromboplastin to the coagulation defect induced by liver failure as assessed by the mean INR obtained with the 34 patients with liver disease.

Graphic and statistical data were analyzed using Analyse-it for Microsoft Excel, Leeds, UK (http://www.analyse-it.com) and GraphPad Prism version 4.02 for Windows (GraphPad Software, San Diego, CA; http://www.graphpad.com).

A P value of less than 0.05 was considered to indicate statistical significance.

Results

Characteristics of the Patients.

There was no difference between the patient group selected for the calibration procedure and that for the standardization study, regarding demographic characteristics and liver disease causes, with alcohol being the most frequent cause of liver damage (Table 2).

Table 2. Characteristics of the Patients with Liver Failure Selected for the Calibration Procedure and for the Study of Prothrombin Time Standardization
CharacteristicsPatients Selected for
Calibration Procedure (n = 60)Standardization Study (n = 34)
Males n, (%)41 (68)23 (68)
Age (y), mean (SD)54 (16)49 (12)
Cause of liver disease  
 Alcohol n, (%)31 (52)19 (56)
 Viral hepatitis(B and/or C) n, (%)11 (18)8 (23)
 Alcohol and viral hepatitis n, (%)9 (15)3 (9)
 Other, n (%)9 (15)4 (12)

Sensitivities of the Thromboplastins to Defects Induced by Vitamin K Antagonists and by Liver Disease.

When the “local” ISI (Table 1), determined to take into account the thromboplastin/coagulometer combination in our laboratory, were compared with the ISI provided by the manufacturers (Table 1), the values of both were similar except for a difference of 0.15 for thromboplastin 2.

The ISI“LD” calculated by the calibration procedure were equal or below 0.94 (which is the ISI of the reference preparation rTF/95) for 4 of 5 thromboplastins. This means that these 4 reagents have the same or a greater sensitivity to coagulation defects induced by liver failure than the reference thromboplastin. The variability in thromboplastin responsiveness to these defects, represented by the range of ISI“LD” from 0.98 to 0.70, was smaller than that observed for defects induced by vitamin K antagonist for which the “local” ISI of the most sensitive thromboplastin was 2 times greater than that of the less sensitive thromboplastin (Table 1). Finally, among the 5 reagents, the most sensitive thromboplastin to defects induced by liver failure was a rabbit brain thromboplastin (no. 3), whereas the 2 human thromboplastins (nos. 4 and 5) were the most sensitive to defects induced by vitamin K antagonist.

Agreement Among PT Results Obtained With 5 Thromboplastins in Patients With Liver Failure According to the Methods of Reporting.

The mean PT values obtained with the 5 different thromboplastins were significantly different (P < 0.0001) for 3 modes of reporting: PTs, PTr, and INR (Table 3). The multiple comparison test showed significant differences between most pairs of thromboplastins except the following: thromboplastins 1 versus 4 and 1 versus 5 in the PTs group; thromboplastins 1 versus 2 and 4 versus 5 in the PTr group; thromboplastins 1 versus 2 in INR group. Only the INR“LD” mode of PT reporting led to nonsignificant differences (P = 0.09) between mean PT values obtained with the 5 different thromboplastins (Table 3).

Table 3. Prothrombin Time (PT) Results Obtained for Patients with Liver Failure (n = 34) Using 5 Different Thromboplastins and Expressed as PTs, PTr, INR, and INR“LD”
 Thromboplastin Number
12345
  1. PTs, PT results reported in seconds; PTr, PT results reported in the ratio of the patient's PT to MNPT.

PTs     
 Mean (SD)22.5 (5.4)20.3 (5.0)31.2 (9.5)23.9 (6.1)22.1 (6.5)
 Range16.3–34.914.4–31.519.7–55.216.6–37.715.2–37.1
PTr     
 Mean (SD)1.8 (0.4)1.8 (0.4)2.2 (0.7)1.9 (0.5)1.9 (0.6)
 Range1.3–2.71.3–2.81.4–3.81.3–3.01.3–3.2
INR     
 Mean (SD)2.7 (1.1)2.5 (1.0)2.3 (0.7)2.0 (0.5)1.7 (0.4)
 Range1.5–5.41.4–4.81.4–4.11.3–3.11.3–2.6
INR“LD”     
 Mean (SD)1.8 (0.4)1.7 (0.4)1.7 (0.4)1.7 (0.4)1.7 (0.4)
 Range1.3–2.71.2–2.61.2–2.61.3–2.51.3–2.7

The INR“LD” mode of reporting led to the smallest mean within-patients CV observed for PT results performed with the 5 thromboplastins, whereas the highest CV were observed for INR modes of expression as indicated in Table 4.

Table 4. Within-Patients (n = 34) Variation of Prothrombin Time (PT) Results Obtained with the 5 Different Thromboplastins According to the Mode of PT Reporting
 Mode of PT Reporting
PTrINRINR“LD”
Mean CV (%)9163
Range3–164–291–7

The Bland-Altman difference plots illustrate the great discordances for PT results reported as INR (Fig. 1). The scatter of the differences between results increased according to the prolongation of the INR, and the discrepancy for INR values >3 was as large as 75% of the mean INR. In contrast, the narrow scatter of difference data points around the value zero observed when PT results were reported as INR“LD” (Fig. 1) confirmed the excellent agreement between the results when using this mode of reporting.

Figure 1.

Agreement between the INR (A) or the INR“LD” (B) obtained with 2 different thromboplastins in patients with liver failure (n = 34). For each plasma sample, the differences between the INR (A) or INR“LD” (B) obtained with the most and the least sensitive thromboplastin to the coagulation defect induced by liver failure were plotted against the mean of the INR (A) and INR“LD” (B) generated by the 2 reagents. The thromboplastins were as follows: (A) reagent numbers 1 and 5 and (B) reagents 1 and 3.

Agreement Among MELD Scores Calculated With the Different INR Values Obtained With the 5 Commercial Thromboplastins.

The mean MELD scores values calculated with the INR obtained with 5 different thromboplastins were significantly different (P < 0.0001), and post hoc analysis showed significant differences between all pairs of thromboplastins. The Bland-Altman difference plots (Fig. 2) show the great discordances between the MELD scores calculated with the INR values obtained with the 2 most different thromboplastins in terms of INR values (thromboplatins 1 and 5). The discrepancy between MELD scores was more than 4 points in 52% of the patients and more than 7 points in 17% of the patients (Fig. 2).

Figure 2.

Agreement between the MELD scores calculated with the INR values obtained with 2 different thromboplastins in patients with liver failure (n = 34). For each plasma sample, the differences between the scores calculated with the INR obtained with the most (reagent no. 1) and the least sensitive thromboplastin (5) to the coagulation defect induced by liver failure were plotted against the mean of the scores generated by the 2 reagents.

Discussion

This study shows that thromboplastin reagents have different sensitivities to coagulation defects induced by vitamin K antagonists and to those induced by liver failure. As a consequence, in this latter situation, a standardization of PT reporting based on the ISI/INR system can only be achieved by a modification of the existing system, the ISI“LD”/INR“LD,” making it specific for the defects induced by liver disease.

We demonstrate, in keeping with previous studies of our group10 and others17–19 that, in patients with liver disease, poor agreement is observed among PT results reported as PTs, PTr, and INR. This disagreement worsens markedly according to the prolongation of PT as shown by the Bland-Altman plots. Moreover, the within-patients variation observed for INR results being twice greater than that observed for PTr results clearly shows that the correction of the PTr by the ISI value of the correspondent thromboplastin worsens the discrepancy of the PT results obtained with different reagents for patients with liver disease. In other words, the rough “standardization” brought out by the use of the PTr mode of reporting was abolished by the use of INR expression. These results can be explained by the inappropriate use of an ISI value, which reflects the thromboplastin sensitivity to the effects of oral anticoagulation and not its sensitivity to the effects of liver disease, to correct, in the INR formula, the PTr obtained in patients with liver disease. Oral anticoagulation antagonizes the vitamin K–dependent carboxylation of coagulation factors II, VII, and X. This results in the synthesis of non-carboxylated or partially carboxylated forms of these factors, named proteins induced by vitamin K absence or antagonist, which are biologically inactive or even inhibitory.25, 26 In most liver diseases, the level of proteins induced by vitamin K absence or antagonist is low and the prolonged PT is rather due to reduced synthesis of coagulation factors II, VII, X, as well as factor V and fibrinogen.27 The different sensitivities of the thromboplastins to the defects induced by vitamin K antagonists and liver failure are illustrated in the current study by the differences observed between ISI and ISI“LD” values and could be explained, at least in part, by the differences in the origin and the nature of the different commercial and reference thromboplastins. Indeed, human thromboplastins are known to have a greater sensitivity to proteins induced by vitamin K absence or antagonist inhibitor effect than rabbit brain thromboplastins.18, 28

A main result of our study is the achieved PT standardization in liver disease provided by the use of the INR“LD” mode of reporting. This standardization has been obtained by one main and several minor modifications of the WHO calibration model for thromboplastins in the ISI/INR system.22 The main modification was the use of plasmas of patients with liver disease instead of plasmas from patients receiving oral anticoagulation for the calibration procedure. Three other points deserving comments. First, deep-frozen instead of fresh plasmas were used because of the difficulty of obtaining, on the same day, the 6 fresh plasmas samples from patients with liver failure needed for a working session of calibration.22 However, studies have shown that INR for patients on anticoagulation therapy or PT in other clinical situations can be reliable even if determined on frozen plasmas.29, 30 Because both our calibration and standardization studies were performed on frozen plasmas, we do not believe that sample freezing could have influenced our data. Second, the WHO calibration model uses international reference thromboplastin preparation with a great responsiveness to the defect induced by the vitamin K antagonist with an ISI value close to 1. The results of our calibration study showed that the human recombinant reference thromboplastin rTF/95 had a similar or lower responsiveness to coagulation defects induced by liver failure than the thromboplastins tested because the ISI“LD” of these 5 reagents was equal to or below the ISI of the reference thromboplastin. Therefore, a thromboplastin with a greater sensitivity to the defects induced by liver failure would have been, perhaps, a better choice as a reference material. With the use of a rabbit brain thromboplastin as a reference material, the ISI“LD” of the commercial thromboplastins tested would have been similar to or greater than the ISI of the reference material. However, considering the close agreement among the INR“LD,” the choice of a rabbit brain reference material would be unlikely to improve the PT standardization in patients with liver failure. Third, the WHO guidelines recommend for the calibration a selection of patients who have been on oral anticoagulants for at least 6 weeks with an INR value in the range 1.5 to 4.5.22 These criteria could not be applied to the patients with liver failure selected to calibrate the ISI“LD”/INR“LD” system. Lastly, we chose to select patients with a large range of PT/INR results to ensure a PT standardization whatever the liver disease severity.

The harmonization of PT results in liver failure, obtained in our laboratory, by the use of the INR“LD” remains to be confirmed in a multicenter study. For this purpose, a “local” ISI“LD” calibration taking into account thromboplastin/coagulometer combination appears necessary for each laboratory.31 However, the manual determination of PT, the requirement for WHO thromboplastin preparation, and the need for 60 plasmas from patients with liver failure for the ISI“LD” calibration according to the WHO recommended procedure22 is not possible in routine hospital laboratories. To avoid these constraints, laboratories could calibrate their own local system with supplied frozen plasmas from patients with liver failure to which certified values of INR“LD” would have been assigned by following, with some modifications, the guidelines on preparation, certification, and use of certified plasmas for ISI calibration and INR determination elaborated by the Subcommittee on Control of Anticoagulation of the Scientific and Standardization Committee of the International Society on Thrombosis and Haemostasis.23

Our study provides evidence that the adoption of the INR“LD” expression as a common scale of PT reporting in hepatology is an important goal, especially in view of the striking discrepancies that we and others20 found between MELD scores calculated with INR values obtained with different thromboplastins. Moreover, our results, illustrated in the Bland-Altman plot, show that the higher the MELD scores the more discrepant. This effect likely results from the deterioration of the INR agreement observed with prolongation of PT. Therefore, PT standardization with INR”LD” appears particularly important in view of the policy prioritizing patients with a MELD score of 15 or greater.32 However, even if INR “LD” is accepted as a better standardized prognostic parameter than INR in liver diseases, its appropriate weight as a variable in indexes such as MELD score remains to be determined. Although the MELD score variability is a key issue, the need for PT standardization by the use of INR“LD” instead of INR does not have to be overlooked in screening, diagnosis, and monitoring acute and chronic liver injury.2

Besides liver diseases, the INR/ISI system has been also extended to standardize PT from patients whose PT is prolonged because of reasons other than oral anticoagulation, such as sepsis or disseminated intravascular coagulation.33, 34 We can hypothesize that, as for liver diseases, INR-based criteria are proposed inadvertently for those patients and that INR“LD” expression could represent a common international scale of PT reporting.

Ancillary