Application of the international normalized ratio in the scoring system for disseminated intravascular coagulation

The International Society on Thrombosis and Haemostasis (ISTH) scoring system for the diagnosis of overt disseminated intravascular coagulation (DIC) utilizes four widely available routine hemostatic tests, i.e. prothrombin time (PT), platelet count, fibrinogen and a fibrin-related marker. PT prolongation of over 3 and 6 s would score 1 and 2 points, respectively. The mechanism of PT prolongation in patients with DIC is different from those on vitamin K antagonists (VKA). However, as most requests received by coagulation laboratories are for oral anticoagulant monitoring, the international normalized ratio (INR) has often been applied inappropriately to samples aimed at investigating DIC. The present study sets out to examine whether an INR system that is alternatively calibrated could standardize PT results in patients with DIC. First, the pattern of coagulation factor deficiency between DIC and warfarin treatment was investigated. Second, an alternative International Sensitivity Index (ISI) specific for DIC (ISIDIC) was determined and compared with conventional ISI for vitamin K antagonism (ISIVKA). Third, sensitivity to PT-based coagulation factors (II, V, VII, X) in six thromboplastins was explored to understand differences between ISIDIC and ISIVKA. Finally, the applicability of ISIDIC and ISIVKA to the ISTH overt DIC scoring system was examined. For conventional calibration of ISIVKA, 60 stable anticoagulated patients on warfarin were selected. Alternative calibration of ISIDIC was performed using data derived from the study of 60 patients with overt DIC as defined by a score of 5 or greater using the ISTH criteria [1]. Forty healthy subjects were selected to determine mean normal prothrombin time (MNPT) and 20 controls were used in the calibration procedure. Six commercial thromboplastins were used (Supplement Table S1) with one lot (N10628, ISI 0.82) of Recombiplastin (Instrumentation Laboratory, Lexington, MA, USA) utilized as reference thromboplastin to assign ISI values. ISIVKA and ISIDIC of the six thromboplastins were determined according to World Health Organization guidelines. To examine for differences in INRVKA and INRDIC between the six thromboplastins, plasma from 37 patients with overt DIC that were different from those selected for calibration were used. Each INR was calculated using the following equations: INRVKA = (patient PT/MNPT)ISIVKA or INRDIC = (patient PT/MMPT)ISIDIC. The DIC score was calculated in these 37 patients according to the six INR values obtained with each thromboplastin using the INR, instead of the PT. Detailed information about patients, reagents and procedures are described in the Supplement (Methods). In the DIC group, mean levels of vitamin K-dependent factors (II, VII, X) were higher but FV levels were lower than those on warfarin. For a given INR, levels of vitamin K-dependent factors (II, VII and X) were usually higher in DIC than in patients on warfarin. FV levels were high in the oral anticoagulation group, while fibrinogen levels were relatively high in the DIC group (Supplement Table S2 and Fig. S1). Calibration parameters of the six commercial thromboplastins are shown in Supplement Table S1. The ISIDIC of all thromboplastins, except Recombiplastin, were lower than the corresponding ISIVKA. The ISIDIC value of PT fib was highest among the thromboplastins. The largest percentage difference between ISIVKA and ISIDIC was observed with the rabbit brain thromboplastins. Human recombinant thromboplastins (Recombiplastin and Innovin) were more sensitive to factors II, V, VII and X than the tissue-derived thromboplastins, especially at a lower concentration of 10% or less (Supplement Fig. S2). Mean values of PT, PT prolongation and INRVKA were significantly different between the six different thromboplastins. However, these differences were largely ameliorated when INRDIC was used (Supplement Table S3). The dispersion of PT prolongation and INR values in patients with DIC (n = 37) between the different thromboplastins was estimated by % coefficient of variation (CV). Between-thromboplastin CV for PT prolongation was 26.7%, which was highest among the variables. The between-thromboplastin CV of 11.7% for INRVKA was reduced to 3.8% when INRDIC was applied (Supplement Fig. S3). The individual DIC score for patients with DIC was calculated with PT, INRVKA or INRDIC for each thromboplastin. We set the INR equivalent to PT prolongation for assigning 1 or 2 points as [(MNPT+3)/MNPT]ISI or [(MNPT+6)/MNPT]ISI, respectively. Accordingly, equivalent INR values were different across the six thromboplastins as they were dependent on the MNPT and ISI of each thromboplastin. Irrespective of whether PT, INRVKA or INRDIC was used, no significant difference in the mean DIC score was observed across all six thromboplastins (Supplement Fig. S4). The overt DIC score is therefore not affected by the INR value because conversion into 0, 1 or 2 points nullified any differences through being encapsulated within a discrete and limited system. However, if the DIC scoring formula were to be changed from a discrete into a variable/continuous PT-INR scoring system, the score differences between thromboplastins would be larger with INRVKA than INRDIC (Supplement Fig. S5). The present study has demonstrated that standardization of PT reporting based on the INR system can be achieved by modification to an alternative ISI that is specific for DIC, using calibration plasmas from DIC patients. Thromboplastin reagents have different sensitivities to coagulation defects induced by DIC and vitamin K antagonists. In liver cirrhosis, two papers [2, 3] have reported that alternative calibration using plasmas from patients with cirrhosis instead of from vitamin K antagonist patients resolve the variability of thromboplastins. This strongly indicates that conventional ISI is not suited to standardize PT results for other patients than those on oral anticoagulation. In theory, this could mean that several kinds of ISI would have to be determined using several kinds of calibration plasma for PT standardization. Practically, it would be difficult and manufacturers would need to provide several ISI values for the different clinical conditions. Alternatively, Tripodi [4] has suggested the use of human recombinant thromboplastins such as Recombiplastin and Innovin as these yield negligible differences between laboratories and between ISIVKA and ISI for liver disease. Our results are concordant in also displaying small differences between ISIVKA and ISIDIC with the human recombinant thromboplastins. With tissue-derived thromboplastins, we found significant variability in INRVKA that could be normalized through the application of INRDIC. Despite these differences, we found that they did not translate into similar differences when applied to scoring for DIC. This appears to be because conversion into a limited point range of 0–2 within the present ISTH overt DIC scoring system minimized any variability between thromboplastins. If, however, a continuous rather than a discrete point scoring system was applied; i.e. higher INR values directly resulting in higher DIC scores, considerable variability between the thromboplastins would then be obvious. While such a system is not being proposed, the findings in this study are very important and should be considered in the event of any future discussions or attempts to improve the current DIC diagnostic system. In its present format, the magnitude of prolonged PT accounts for up to 30% of the maximal DIC score achievable and has been shown to be prognostically meaningful in several studies [1, 5, 6]. With due consideration for the limitations of this study, as discussed in the Supplement, our conclusion is that alternative calibration using plasma from patients with DIC provides better comparability of INR results between different commercial thromboplastins in patients with DIC. Nonetheless, the present format of diagnosing overt DIC whereby changes in the PT-INR are assigned discrete scores is not significantly influenced by thromboplastin differences. The authors state that they have no conflict of interest. Table S1. Calibration parameters of six commercial thromboplastins. Table S2. The levels of coagulation factors according to the clinical conditions. Table S3. Means of PT, PT ratio, prolonged PT in seconds, INRVKA and INRDIC for patients with overt DIC (n = 37) using six commercial thromboplastins. Fig. S1. The levels of vitamin K dependent coagulation factors (II, VII, X) and non-vitamin K dependent coagulation factors (I, V) for a given INR value according to the clinical condition [open circle, warfarin treatment (n = 60); closed circle, DIC (n = 55)]. This plot showed natural logarithm (Ln) of each factor level (Y-axis) versus natural logarithm of INR (X-axis). Fig. S2. Sensitivity of six commercial thromboplastins to deficiencies in factor II, V, VII and X. Prothrombin time (PT) ratio was determined at various factor levels (1%, 3%, 10%, 25%, 50% or 100% normal plasma). *P < 0.01 at each factor level using ANOVA for repeated measures (for comparison of mean PT ratios of six thromboplastins at each factor level). Fig. S3. The dispersion of PT prolongation and INR values in patients with DIC (n = 37) among different thromboplastins. (A) Prolongation in PT (seconds) above the upper limit of normal reference range as measured by 6 different thromboplastins. (B) INRVKA obtained by different thromboplastins when calibrated by conventional methodology using plasma from patients stabilized on vitamin K antagonists. (C) INRDIC was determined by calibration with plasma from patients with DIC. In all figures, mean values are shown as open circle and the bar limits represent 95 percent confidence interval. Abbreviations: CV, coefficient variation. Fig. S4. The dispersion of DIC scores calculated with discrete score of PT or INR variable in patients with DIC (n = 37) among different thromboplastins. (A) DIC scorePT was calculated with the PT variable. (B) DIC scoreINRVKA was calculated with the INRVKA. (C) DIC scoreINRDIC was calculated with INRDIC. In all figures, mean values are shown as open circle and the bar limits represent 95 percent confidence interval. Fig. S5. The dispersion of DIC scores calculated with continuous score of INR variable in patients with DIC (n = 37) among different thromboplastins. (A) DIC scoreINRVKA was calculated with the INRVKA (B) DIC scoreINRDIC was calculated with INRDIC. The mean values are shown as open circle and the bar limits represent 95 percent confidence interval. Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.