FIBTEM provides early prediction of massive transfusion in trauma

Blood samples for both ROTEM^® analysis and standard coagulation tests were collected in 3 mL tubes containing 0.3 mL buffered 3.2% trisodium citrate (volume ratio 1:9). Thromboelastometric analyses were typically performed at the bedside within minutes of sample collection by the attending anesthetist or intensivist. The more severe the trauma, the more quickly the analyses were undertaken (minimum set-up time approximately two minutes). Three ROTEM^® tests were performed: extrinsically activated assay with tissue factor (EXTEM), intrinsically activated test using kaolin (INTEM), and extrinsically activated test with tissue factor and the platelet inhibitor cytochalasin D (FIBTEM).

For the EXTEM and INTEM assays, the following variables were measured: clotting time (CT (s)); clot formation time (CFT (s)); A10 (clot amplitude 10 minutes after the end of CT); and maximum clot firmness (MCF (mm)). For the FIBTEM assay, A10 and MCF were investigated. The platelet component was calculated as MCE_EXTEM - MCE_FIBTEM, where maximum clot elasticity MCE = (MCF*100)/(100-MCF) [23]. Finally, the lysis index at 60 minutes (LI60 (%), clot firmness 60 minutes after CT as percentage of MCF) was collected for the EXTEM and INTEM assays. Hyperfibrinolysis was defined as complete breakdown of the clot, in accordance with previous work by our group [24].

In parallel, standard laboratory analyses were performed: Quick value and aPTT (determined on Sysmex XE-2100 (Roche Diagnostics GmbH, Mannheim, Germany)); pH, base deficit (BD) and lactate (analyzed using Roche OMNI^® S Blood Gas Analyzer (Roche Diagnostics GmbH); normal range for BD, −3.0 to +3.0 mmol/L; normal range for lactate 0.5 to 2.2 mmol/L). Fibrinogen concentration was measured by the Clauss method (STA-Fib^® assay (Roche Diagnostics GmbH); optical read-out), using a STA-Compact^® machine (Roche Diagnostics GmbH, Vienna, Austria). Hemoglobin, hematocrit and platelet count were analyzed using blood samples anti-coagulated with ethylenediamine tetra-acetic acid, with an SF 3000 analyzer (Sysmex Corporation, Kobe, Japan).

For patients with ongoing bleeding, coagulation therapy was based on ROTEM^® test results with administration of coagulation factor concentrates as previously described [20, 21]. Red blood cells (RBCs) were administered as required to maintain a target intra-operative hemoglobin concentration of 10 g/dL; later during intensive care unit (ICU) therapy, a lower cut-off value of 7 g/dL was accepted.

Since 2006, electronic documentation has been implemented in our hospital where the type, amount and timing of allogeneic blood products transfused are recorded. For 2005, the amount and timing of allogeneic blood transfusion were reviewed from anesthesia and ICU records. Two groups of patients were defined according to RBC transfusion: non-massive transfusion (non-MT) group (0 to 9 U RBC transfused in 24 hours), and massive transfusion (MT) group (≥10 U RBC transfused in 24 hours). This definition of MT is consistent with that used in previous publications [4, 5, 12, 25]. There was no expectation that the two patient groups would be well matched, since patients with more severe injury upon admission are more likely to undergo MT. Also, the number of patients in the two groups was unlikely to be equal, as in our centre <50% of patients undergo MT; the inclusion of all eligible patients minimizes the risk of bias.

Data are presented as mean ± standard deviation or median and interquartile range (IQR) for continuous variables, and as percentages for categorical variables. Continuous variables were analyzed for normal distribution by the Kolmogorov-Smirnov test. To detect differences between patient groups, either the Student's t-test or the Mann-Whitney U-Test was performed, depending on the underlying distribution. Group differences were compared by ANOVA with the Kruskal-Wallis test. For categorical variables, Fisher's exact test was used. Correlation between parameters was analyzed by Spearman's correlation coefficient rho. As a measure for discrimination not depending on a certain cut-off point, the area under the receiver operating characteristic (ROC) curve was determined, together with its 95% confidence interval. A P-value <0.05 was considered significant for all statistical tests. Statistical calculations were performed using GraphPad Prism 5.03 (GraphPad Software, La Jolla, CA, USA) and IBM SPSS 19 (SPSS Inc., Chicago, IL, USA).

ROC curves for ROTEM^® parameters showed that the best predictive value for MT was provided by FIBTEM MCF (area under the curve (AUC) 0.84), with a similar outcome for FIBTEM A10 (AUC 0.83) (Table 4). For A10, a threshold of ≤4 mm provided the best values for sensitivity (63.6, 95% confidence interval (CI) 51.9 to 74.3) and specificity (82.9, CI 77.5 to 87.5). The corresponding threshold for MCF was ≤7 mm (sensitivity 78.2 (CI 67.4 to 86.8); specificity 74.7 (CI 68.7 to 80.1)). AUCs for prediction by the EXTEM and INTEM test parameters were slightly lower than those for FIBTEM MCF and FIBTEM A10 (Table 4). A statistically significant difference was observed between AUCs for FIBTEM MCF and INTEM MCF (P = 0.047), but there was no significant difference between the AUCs for FIBTEM A10 and INTEM MCF or EXTEM MCF. For all ROTEM^® parameters, the ROC-AUC remained unchanged if clot elasticity was used instead of clot firmness (for example, maximum clot elasticity (MCE) instead of MCF). The platelet component had an AUC for MT prediction of 0.74 (CI 0.69 to 0.79).

Considering standard laboratory coagulation analyses, the highest predictive values for MT were provided by hemoglobin, Quick value and aPTT (ROC-AUC values shown in Table 4). Fibrinogen concentration provided predictive value similar to that of the FIBTEM parameters.

Quantities for RBC transfusion and coagulation therapy within the first 24 hours are summarized in Table 5. Patients in the MT group (by definition receiving more units of RBC) also received higher amounts of fibrinogen concentrate (FC), prothrombin complex concentrate (PCC) and platelet concentrate. Fresh frozen plasma (FFP) administration is not outlined in the table as it was used in only 10 out of 323 patients.

Higher amounts of RBC transfusion were observed among patients with lower FIBTEM A10 values (P < 0.01) and lower fibrinogen concentration (Figure 2). Using a cut-off value for FIBTEM A10 of ≤4 mm, patients received a median of 10 (IQR 5 to 15) units of RBC, compared to a median of 2 (IQR 0 to 6) units in patients with a FIBTEM A10 >4 mm, representing five-fold higher RBC transfusion among those with the lower A10 values. Among the 31 patients with a FIBTEM A10 ≤3 mm, 26 (84%) were in the MT group. Unsurprisingly, significant correlations were observed between FIBTEM A10 and fibrinogen concentration (Spearman's correlation coefficient rho 0.78, P < 0.0001), and between MCF and fibrinogen concentration (Spearman's correlation coefficient rho 0.75, P < 0.0001). Twenty-nine patients had FIBTEM A10 values of zero and a median fibrinogen concentration of 64 mg/dL (IQR 54 to 94 mg/dL; Figure 2c). In 17 of these patients, 'spindle' traces (LI values up to 100%) for the INTEM and EXTEM assays indicated the presence of hyperfibrinolysis.

The clinical value of the MT definition we used in this study has been questioned: mortality increases with increasing red blood cell transfusion (no apparent threshold level of transfusion), so the criterion of 10 units in 24 hours does not allow specific selection of patients with worsened clinical outcomes [37]. An alternative definition of MT has been explored (at least five units of RBC during the first four hours) [38], but at present there is not enough evidence to support the use of such criteria.

MT patients in our study may not be directly comparable with those studied at other centres, due to our use of coagulation factor concentrates for coagulation therapy. Use of the standard FIBTEM assay also represents a possible limitation of the study: it has been shown that cytochalasin D does not provide complete inhibition of platelet activity [39]. As a result, in our study, values for FIBTEM could be higher than they should be and values for platelet component could be lower. These effects would vary between patients depending on platelet count. The retrospective nature of the study was another limitation - for example, we were not able to identify co-medications or co-morbidities that could have influenced test results. As expected, the separation of patients according to whether they underwent MT led to groups of patients that were not balanced in relation to severity of injury. This imbalance could possibly be considered as a study weakness, though on the other hand it could be considered simply as a reflection of clinical reality. The difference in patient numbers between the two groups, which is attributable simply to the number of patients meeting the inclusion criteria for each group during the study period, might also be considered as a limitation of the study. We proceeded with the analysis of the unbalanced groups because this approach avoids the bias that might potentially arise from the 'artificial' exclusion of a proportion of the non-MT patients. Similar between-group differences, both in patient numbers and severity of injury, were evident in the study published by Leemann et al. [5]. Lastly, the exclusion of patients for whom therapy was withheld due to non-survivable injuries could be considered as a limitation potentially introducing bias, as no patient with the maximum ISS score of 75 was included in the study. However, the inclusion of patients not receiving standard medication would certainly have introduced bias to the study, because they died without receiving any transfusion and ROTEM^® test results would not have been followed by any therapy.