Background
Hemorrhage is still a common cause of death in trauma. Even when patients are rapidly brought to hospital and are treated according to current trauma guidelines, patients may die because of bleeding that cannot be controlled [
1,
2]. Coagulopathy is associated with worse outcomes and may develop during hemorrhage or transfusion therapy [
3‐
5]. Fibrinogen and prothrombin have been suggested to be the first coagulation factors to become rate limiting for coagulation in such situations [
6]. While the administration of fibrinogen is now recommended to restore fibrinogen levels [
7‐
9], the importance of prothrombin in trauma-induced coagulopathy and patient outcome is unclear.
In experimental models of dilutional coagulopathy, administration of recombinant human (rh) prothrombin alone or in combination with fibrinogen improved survival, reduced blood loss and improved coagulation when measured by ROTEM and PT [
10,
11]. Administration of prothrombin-containing concentrates may therefore be useful also in trauma patients with coagulopathy. Measurement of prothrombin concentration is necessary to identify patients with low levels and address the concern that over supplementation could lead to increased thrombotic risk [
12‐
14]. However, since central lab analyses of plasma samples for prothrombin concentration are too slow, there is a need for surrogate biomarkers to identify patients with low prothrombin concentration. We have therefore conducted analyses of prothrombin levels in a prospective observational cohort from a London Trauma Centre and extended analyses on outcomes in an independent cohort from the Medical University of Innsbruck.
The aims of the study were, firstly, to investigate the consequence of admission prothrombin concentration on massive transfusion (more than 10 PRBC units) and mortality at 24 h. Secondly, to determine the relationship between admission biomarkers (PT, ROTEM EXTEM CT and MCF) and prothrombin concentration, and thirdly, to understand the ability of admission biomarkers to act as a surrogate for low prothrombin concentration and predict outcome. To further evaluate the individual role of prothrombin on PT and ROTEM biomarkers, in vitro studies were performed to investigate the effect of specific prothrombin depletion.
Discussion
In this study we investigated the critical role of prothrombin concentration in coagulopathy and prediction of outcome in coagulopathic trauma patients. In addition, we evaluated currently available coagulation biomarkers as surrogates for detecting critically low plasma prothrombin levels. Data presented indicates that admission prothrombin plasma concentration can be used to predict increased survival and a lower transfusion demand within the first 24 h following admission of the trauma patient and that a cut-off in the range of 50–70 IU/dL at admission is associated with a worse outcome (Fig.
1). During such a coagulopathy many coagulation factors are simultaneously reduced. This has been previously shown where coagulation factors and biomarkers were analysed in coagulopathic patients with INRs of 1.5–3 prior and post FFP administration [
19]. Similar to our findings, these patients had a median prothrombin concentration of 34% of normal before FFP administration and also reduced concentrations of several other coagulation factors such as Factor X and fibrinogen.
Since several coagulation factors are reduced in coagulopathic trauma patients we investigated the specific role of prothrombin concentration in coagulation by in vitro experiments. Our in vitro data comparing specific neutralization of prothrombin by a prothrombin neutralizing antibody and general dilution of all coagulation factors by serial dilution of blood or plasma, showed that reduction in prothrombin concentration had a dramatic effect on both PT and ROTEM CT. Comparable data was found for specific neutralization of prothrombin or general dilution of all coagulation factors suggesting that prothrombin concentration is a rate limiting factor in coagulation (Fig.
2). Our in vitro results also support the usefulness of PT and EXTEM CT as surrogate markers for direct measurement of prothrombin concentration. EXTEM MCF is not appropriate due to the discrepancy between ROTEM MCF obtained during step-wise dilution and neutralization of prothrombin. This is thought to be due to that following neutralization a proportion of prothrombin is still activated and is available to activate platelets and fibrinogen which are at normal concentration. Since fibrinogen and platelets are mainly responsible for building MCF the same clot stability is reached eventually. In the case of step-wise dilution, all coagulation factor concentrations and platelet count are decreased and hence the almost linear decrease in EXTEM MCF.
Extending in vitro findings to the trauma databases, prothrombin concentration in trauma patients was also reflected in the biomarkers PT, EXTEM CT and EXTEM MCF. Furthermore, low prothrombin concentration was correlated with a low fibrinogen concentration (Fig.
3). This suggests that the trauma patients having low fibrinogen and prothrombin concentrations are coagulopathic due to coagulation factor consumption and/or dilution. Since point-of-care (PoC) tests are not available for the fast analysis of fibrinogen or prothrombin, surrogate coagulation markers could be used to identify patients with low prothrombin concentrations. The idea of early identification of patients in need of hemostatic therapy by the use of coagulation assays is not new. ROTEM and TEG, where available, are used for this purpose and for guiding therapy [
7‐
9,
20]. In hemorrhagic patients the PT assay has as far as we are aware been used to analyse retrospective data, although it is sometimes used as a tool to confirm the effectiveness of hemostatic therapy [
21].
In the clinic, a cut-off value to define “low prothrombin” is required in order to decide who will receive prothrombin containing treatment and who will not. A cut-off value for prothrombin is also needed for ROC analysis in order to compare the sensitivity and specificity of biomarkers to predict low prothrombin concentration. From Fig.
1, we observed that there appears to be a change in the proportion of survivors and the total amount of administered PBRC within the prothrombin concentration range of 50–70 IU/dL. Furthermore, in vitro data presented in Fig.
2 shows that decreasing prothrombin below a similar limit has a large impact on PT and EXTEM CT. These two observations together provides evidence for a threshold in prothrombin concentration within the range of 50–70 IU/dL that is critical for coagulation. In order to evaluate biomarker predictivity we chose to evaluate 60 IU/dL as a binary threshold for ROC analyses since a prothrombin concentration below 60 IU/dL is considered to be out of the normal range and therefore is reflective of a clinically relevant cut-off.
ROC analyses suggests that PT is a better predictor of low prothrombin concentration versus EXTEM CT and EXTEM MCF. To check how sensitive the biomarker comparisons were to the choice of cut off level, we also performed the analyses using < 70 IU/dL. The ranking of the coagulation biomarkers remained the same independently of whether the cut off was set to < 60 (Fig.
4) or < 70 IU/dL (data not shown). Furthermore, PT was also the better predictor of massive transfusion and mortality compared to EXTEM CT and EXTEM MCF. We chose not to include the amplitudes at 5 or 10 min (A5, A10) in our analyses since they have been shown to correlate well to MCF [
22‐
24]. In the databases analysed even a small increase in PT is associated with an increased risk. This was observed in both independent databases despite multiple differences in the severity of injury and inclusion/exclusion criteria.
We believe that successful normalization of prothrombin levels requires administration of a concentrate such as the recombinant human prothrombin (MEDI8111) that was recently evaluated in healthy volunteers [
25]. In support of this, administration of recombinant human prothrombin alone was sufficient to reduce bleeding and improve coagulation and survival time in coagulopathic pigs [
11]. In humans, however, the importance of having a sufficiently high fibrinogen concentration has been recognized [
7] and from our database analyses we confirm that there is a strong correlation between admission concentrations of fibrinogen and prothrombin (Fig.
3). We cannot therefore distinguish between the specific contribution of fibrinogen or prothrombin loss alone on prolonged PT and coagulopathy but instead conclude that prothrombin is depleted along with fibrinogen during the development of coagulopathy and that replacement of both factors is required to achieve hemostasis.
PT is a useful biomarker to identify eligible patients for prothrombin replacement therapies although centrally measured PT was used in the current databases which is recognised to have limitations in guiding treatment due to the slow turnaround time [
26]. A number of PoC analysers of PT are available e.g. CoaguChek XS system, Roche; i-STAT system, Abbott, and need to be evaluated further for this purpose. If this can be achieved, it seems possible that in the future massive bleedings can be controlled or even avoided by identification of coagulopathic patients depleted in prothrombin by simple available PoC methodology.
Limitations
This study is based on analyses of two prospective observational cohorts. Since prothrombin concentration was not routinely measured at the Medical University of Innsbruck the number of patients with admission data was limited. All analyses involving prothrombin were therefore restricted to database 1 but database 2 was included and added value by extending observations regarding the predictivity of PT for mortality and transfusion in two independent databases. There are also multiple differences in the databases with regards to severity of injury, inclusion/exclusion criteria, biomarker analyses and importantly admission sampling time from injury. While this variability does add noise to the study we consider that it also adds strength in our conclusions i.e. that despite these differences, PT is still predictive of outcome and the better biomarker in both databases and as a consequence supportive of broader clinical utility.
Acknowledgements
The authors gratefully acknowledge the technical assistance of Dr. Philipp Würtinger for the recalculation of PT/INR data in the Innsbruck database.