THROMBIN GENERATION KINETICS ARE PREDICTIVE OF RAPID TRANSFUSION IN TRAUMA PATIENTS MEETING CRITICAL ADMINISTRATION THRESHOLD

Abstract

Introduction:

We hypothesize that a patient (pt) with accelerated thrombin generation, time to peak height (ttPeak), will have a greater odds of meeting critical administration threshold (CAT) criteria (> 3 packed red blood cell [pRBC] transfusions [Tx] per 60 min interval), within the first 24 h after injury, independent of international normalized ratio (INR).

Methods:

In a prospective cohort study, trauma patients were enrolled over a 4.5-year period and serial blood samples collected at various time points. We retrospectively stratified pts into three categories: CAT+, CAT− but receiving some pRBC Tx, receiving no Tx within the first 24 h. Blood collected prior to Tx was analyzed for thrombin generation parameters and prothrombin time (PT)/INR.

Results:

A total of 484 trauma pts were analyzed: injury severity score = 13 [7,22], age = 48 [28, 64] years, and 73% male. Fifty pts met criteria for CAT+, 64 pts CAT−, and 370 received no Tx. Risk factors for meeting CAT+: decreased arrival systolic blood pressure (OR 2.82 [2.17, 3.67]), increased INR (OR 2.09, [1.66, 2.62]) and decreased time to peak OR 2.27 [1.74, 2.95]). These variables remained independently associated with increased risk of requiring Tx in a multivariable logistic model, after adjusting for sex and trauma type.

Conclusions:

Pts in hemorrhagic shock, who meet CAT+ criteria, are characterized by accelerated thrombin generation. In our multivariable analysis, both ttPeak and PT/INR have a complementary role in predicting those injured patients who will require a high rate of Tx.

INTRODUCTION

Trauma-induced coagulopathy (TIC) is incompletely understood, and existing therapies to treat TIC are limited. The two faces are: the hyper- and the hypocoagulable states. The latter results in hemorrhage and has a mortality rate of up to 40% to 60% (1–3). Uncontrolled bleeding is still the primary cause of preventable death after trauma (4). Higher plasma and platelet ratios early in resuscitation are associated with decreased mortality in patients who received transfusions of at least three units of blood products during the first 24 h after admission (5). While early resuscitation from hemorrhagic shock can be life-saving, we need effective tools to identify at-risk patients, before overt clinical signs of hemorrhagic shock become evident. Early recognition of patients with potential for significant bleeding would be a valuable tool for earlier aggressive transfusion, decreasing risk of mortality (6). Current models to predict massive transfusion are based on studies with limited retrospective data, with inadequate power, lack of mechanistic explanation for the observed clinical outcome, and lack of validation of the model used. Attempts to validate a published model to predict massive transfusion have resulted in poor performance (7–9). Additionally, the traditional definition of massive transfusion (MT) of 10 units packed red blood cells (pRBCs) over 24-h period, does not reflect the varied cadence of resuscitation amongst patients, even though same number of units may have been transfused during this period. The critical administration threshold (≥ 3 critical red blood cell units per 60 min interval) within the first 24 h of injury, as defined by Savage et al. (10), reduces the survival bias which exists in many transfusion prediction models that defines abnormal hemostasis as total number of units transfused over the initial 24 h; as patients who die early from bleeding cannot have ongoing transfusions (11).

A recent systemic review and meta-analyses of 3,548 records by Tran et al. (12) revealed that existing published studies were of poor quality, as assessed by the Prognosis Research Strategy recommendations and Critical Appraisal and Data Extraction for Systematic Reviews of Modeling Studies checklist. However, they were able to identify seven candidate predictors of massive transfusion, six of which were clinical variables, and one being laboratory variable, the international normalized ratio (INR) > 1.5.

The objective of this study is to assess thrombin generation kinetics in bleeding trauma patients and assess its predictive role, along with INR, in identifying those who require any or many transfusions. This is the first study to analyze the differences of thrombin generation kinetics in those requiring rapid transfusion, i.e., meeting (critical administration threshold [CAT]) criteria from those who do not. The study hypothesis is that a patient with accelerated time to peak thrombin generation (ttPeak) will have greater odds of meeting CAT criteria, within the first 24 h after injury, and that this association will be independent of INR and other risk factors.

METHODS

Study design, setting, and population

This is a subcohort analysis of a parent study, which was a prospective case-cohort study that was designed to assess thrombin generation kinetics as predictors of symptomatic venous thromboembolism (VTE) after trauma. Over the 4.5-year period, February 2011 to September 2015, 2,954 patients with acute trauma were screened for study enrollment; 1,960 met study inclusion criteria of which 1,234 (63%) consented to participate. Of the 726 patients (37%) of eligible patients who were not consented: 423 (or their surrogates) could not be reached for consent and a total of 303 patients declined to participate. These 726 patients were of similar median age (48[28,68]), sex (69% men), and mechanism of injury (93% blunt) as compared with the 1,234 enrolled. Of those enrolled, 487 trauma patients had plasma analyzed, due to the case-cohort design of the parent study (see Discussion). Three patients died within 24 h of injury and were excluded from analysis, leaving a total of 484 patients in our analysis population.

In this study, all trauma patients transported to the Mayo Clinic Emergency Department (ED) by ambulance or air transport, as a Level 1 or 2 trauma activation (13), from February 2011 to June 2014 were considered for study inclusion. Exclusion criteria were prior history of VTE, age < 18 years, anticoagulation (e.g., heparin, warfarin) or antithrombotic therapy (excluding aspirin or non-steroidal anti-inflammatory drugs), pre-existing coagulopathy, more than 12 h from time of injury, no blood drawn within first 12 h after injury, active cancer, sepsis, renal failure, burn injuries, or declined consent by the patient or legal guardian. The time of injury (TOI) was determined by the prehospital medical providers based on information at the injury scene. If the TOI was unclear, the prehospital medical providers estimated the time and relayed this information to the emergency communication center. Data collected include demographics, baseline, and time-dependent clinical characteristics, including injury severity score (ISS), patient age and sex, body mass index, hospital length of stay, all-cause mortality, and start and stop of anticoagulant-based chemoprophylaxis and other medications affecting coagulation, trauma injury severity score (TRISS), all injury codes from Mayo Clinic Trauma Registry, transfusion data in the prehospital and for entire index hospitalization, and standard admission laboratory data. Transfusion therapy was mainly based on Mayo Clinic Trauma Center transfusion guidelines at the discretion of the medical provider. This study was approved by the Mayo Clinic Institutional Review Board. When patients were unable to provide consent at the time of the trauma, consent was obtained from the patient or legal guardian within 30 days of hospital discharge; samples were discarded when consent could not be obtained.

Transfusion categories

Patients were stratified into three categories:

1. CAT+ as defined by ≥ 3 packed red blood cell units within 60 min interval in the first 24 h after injury.

2. CAT− but receiving some transfusion within the first 24 h after injury.

3. No transfusions within the first 24 h after injury.

Sample collection and processing

Blood samples were collected at baseline (0–2 h), 6 h, 12 h, and day 1 after injury. A total of 18 mL of whole blood was collected by antecubital venipuncture or via existing indwelling catheters into 4.5 mL citrated Vacutainer tubes (0.105 M buffered sodium citrate, 3.2% Becton Dickinson, Plymouth, UK) and processed to platelet-free plasma by double centrifugation (3,000 g, 15 min) as recommended by the ISTH vascular biology SSC Collaborative Workshop (14) and stored in multiple aliquots at −80°C until analysis. All samples were processed within 1 h of collection.

Calibrated automated thrombogram analyses

Thrombin generation was measured with the Calibrated Automated Thrombogram (Thrombinoscope BV, Maastricht, The Netherlands), utilizing a Fluoroskan Ascent plate reader (390 nm excitation, 460 nm emission, Thermo Electron Corp, Vantaa, Finland), as previously described by Hemker et al. and our group (15–17). Assays of trauma patient samples were performed in triplicate. Corn trypsin inhibitor (50 ug/mL final concentration) was added to each plasma sample prior to thrombin analyses. Thrombin generation was initiated with addition of 20 uL of “PPP” reagent (5 pM tissue factor and 4 uM phospholipid, Stago, US) reagent. Then, 80 uL of citrated platelet-free plasma was added to each well of U bottom 96-well microtiter plates (Nunc, Thermo Fischer Scientific, Rochester, NY) using a single-channel pipette. After 10 min of incubation at 37°C, 20 uL of warmed FLUCA reagent (Fluca kit, TS50, Thrombinoscope, BV), which contains the fluorogenic substrate and CaCl₂ was added to each well via an automated dispenser. Thrombin generation curves were recorded continuously for 90 min at a rate of three readings per minute. Separate wells containing the thrombin calibrator, which corrects for inner filter effects and quenching variation among individual plasmas, were also measured in parallel. A dedicated program, Thrombinoscope, was used to calculate thrombin activity over time. The parameters derived were: Lagtime (LT), peak height (PH), and ttPeak. LT (minutes) is defined as the moment that the signal deviates by more than two standard deviations from the horizontal baseline, PH (nM) is greatest level of thrombin reached on thrombin generation curve, which is generated over time, and ttPeak is time from beginning of the assay reaction (minutes) to reach PH.

Statistical analyses

Categorical data is presented as counts (percent). Continuous data is presented as median [interquartile range (IQR)], unless otherwise specified. The outcome (no transfusion, CAT− with some transfusion, and CAT+) formed an ordinal scale, and the ordinal logistic regression model, also known as the proportional odds model, assumes a common odds ratio for each comparison of “higher versus lower” need for transfusion. Univariate and multivariable ordinal logistic regression was used to evaluate the simple and joint associations of demographic and clinical variables with risk of requiring transfusion and risk of meeting CAT+ criteria. In the logistic analysis, INR and peak height were winsorized (outlier values were replaced by the nearest nonoutlier value, minimizing their influence) (18), while Lagtime and ttpeak were both winsorized and log-transformed. To facilitate comparison between continuous risk factors, odds ratios were calculated and reported for an increase or decrease of one standard deviation. To assess the separate and joint contributions of INR and time to peak thrombin generation in predicting the need for massive transfusion, the model was refit for the two-level variable (CAT+ yes/no) and the area under the ROC curve (AUC) was calculated for the models with and without each biomarker. Data analysis was performed using SAS, version 9.4 (SAS Institute Inc, Cary, NC) and R version 3.2.0.

RESULTS

Among the 484 patients analyzed, the demographic data are: ISS median=13 [7,22], hospital length of stay=5 [2, 11] days, age=48 [28, 64] years, and 73% male. For mechanism of injury, 22 patients had penetrating, 461 had blunt, and one had an unknown mechanism. For this study, thrombin data obtained prior to any transfusions were utilized, given that each patient had samples drawn at multiple set time points after injury. The study group included 50 patients who met CAT+ criteria, 64 who received some pRBC transfusions (CAT−), and 370 who received no transfusions (Table 1). Overall, CAT+ patients received a median of 6 U pRBC [IQR 4,10] versus CAT− patients who received a median of 2 U pRBC [IQR 1,3] during first 24 h of injury.

Table 1.

Comparison of the transfusion groups (total number of patients, whose blood samples were analyzed = 484)^*

Parameter	CAT + (n = 50)	CAT− (some pRBC) (n = 64)	CAT− (no pRBC) (n = 370)	Total (n = 484)
pRBC Tx (units)	6 [4, 10]	2 [1, 3]	0 [0, 0]	0 [0, 0]
FFP Tx (units)	5 [2, 7]	0 [0, 1]	0 [0, 0]	0 [0, 0]
PLT Tx (units)	1 [0, 2]	0 [0, 0]	0 [0, 0]	0 [0, 0]
Cryo Tx (units)	0 [0, 0]	0 [0, 0]	0 [0, 0]	0 [0, 0]
ISS	26.5 [18.0, 38.0]	22.0 [11.0, 29.0]	10.0 [5.0, 17.0]	13.0 [7.0, 22.0]
Hospital length of stay	14.0 [10.0, 21.0]	11.5 [7.5, 17.5]	4.0 [1.0, 8.0]	5.0 [2.0, 11.0]
Age	47.5 [27.0, 61.0]	52.0 [32.0, 65.0]	48.0 [28.0, 64.0]	48.0 [28.0, 64.0]
Sex
Male	41 (82%)	48 (75%)	262 (71%)	351 (73%)
Female	9 (18%)	16 (25%)	108 (29%)	133 (27%)
Lagtime (minutes)	2.38 [1.98, 2.96]	2.47 [2.23, 2.89]	2.67 [2.30, 2.96]	2.60 [2.27, 2.96]
Peak height (nM)	346 [279, 448]	396 [330, 458]	357 [313, 413]	360 [311, 424]
ttPeak (minutes)	3.93 [3.43, 4.52]	4.12 [3.59, 4.60]	4.61 [4.06, 5.33]	4.50 [4.00, 5.14]
INR	1.1 [1.0, 1.3]	1.1 [1.0, 1.1]	1.0 [1.0, 1.1]	1.0 [1.0, 1.1]
PTT (minutes)	29 [26, 34]	27 [25, 28]	26 [24, 29]	26 [24, 29]

^*Median (IQR), where applicable.

Cryo indicates cryoprecipitate; FFP, fresh frozen plasma; ISS, injury severity score; PLT, platelets; pRBC, packed red blood cells; ttPeak, time to peak; Tx, transfusion.

When INR was winsorized (outlier values were replaced by the nearest nonoutlier value, minimizing their influence), and ttpeak was both winsorized and log-transformed, almost all of the patients who met CAT criteria are concentrated at the low end of the distribution (shorter ttPeak), even when conditioned on the value of the INR (please refer to Supplement, http://links.lww.com/SHK/B119). Without winsorizing the data, all the data points are clustered together with relationship between INR and ttPeak obscured. In univariate analysis (Table 2), clinical factors associated with increased risk of requiring transfusion or meeting CAT criteria included: decreased ED arrival systolic blood pressure (SBP) (OR 2.82 [2.17, 3.67]), increased INR as continuous variable (OR 2.09, [1.66, 2.62]) and decreased time to peak (log-transformed OR 2.27 [1.74, 2.95]). The two trauma severity scores (ISS and TRISS), were also associated with increased bleeding risk in univariate analysis, but as those scores are not readily available at time of hospital arrival, they were excluded from multivariable analysis. The three variables (ED SBP, INR, and ttPeak) all remained strongly and independently associated with the need for transfusion and for high rate of transfusion (CAT+), after adjusting for sex and trauma type (Table 2). INR and ttPeak independently and jointly contributed in predicting need for massive transfusion. Retaining both INR and ttPeak in the model led to greatest area under the ROC curve of 0.868 (Table 3).

Table 2.

Univariate and multivariable analysis

Parameter	Univariate odds ratio* (95% CI)	P Value	Multivariable odds ratio* (95% CI)	P value
Male sex (vs. female)	1.50 (0.91, 2.46)	0.11	1.80 (0.97, 3.34)	0.062
Penetrating trauma (vs. blunt)	1.62 (0.66, 3.99)	0.29	0.97 (0.29, 3.26)	0.97
ED SBP	2.82 (2.17, 3.67)	<0.0001	2.46 (1.80, 3.34)	<0.0001
INR (winsorized)	2.09 (1.66, 2.62)	<0.0001	1.79 (1.39, 2.29)	<0.0001
ttPeak (log, winsorized)	2.27 (1.74, 2.95)	<0.0001	1.61 (1.21, 2.14)	0.0011
Lagtime (log, winsorized)	1.24 (0.99, 1.54)	0.058
Peak height (winsorized)	0.96 (0.78, 1.19)	0.73
ISS	3.13 (2.47, 3.97)	<0.0001
Age	1.00 (0.81, 1.24)	0.98
TRISS	1.91 (1.59, 2.30)	<0.0001

^*Odds ratio associated with one-SD increase in variable for continuous variables.

ED SBP indicates systolic blood pressure on arrival to emergency department; TRISS, trauma and injury severity score (probability of survival).

Table 3.

Receiver operating characteristics

Model		AUROCC*
1	Male sex + penetrating trauma + ED SBP	0.814
2	Male sex + penetrating trauma + ED SBP +admission INR	0.847
3	Male sex + penetrating trauma + ED SBP + ttPeak†	0.841
4	Male sex + penetrating trauma + ED SBP +admission INR + ttPeak	0.868

^*Area under the ROC curve (AUROCC).

^†ttPeak–time to peak (minutes).

DISCUSSION

There is a paucity of laboratory tests that have been validated to assess the severity of trauma-induced coagulopathy (12). Numerous other studies have evaluated clinical and laboratory tools to predict the need for MT in trauma patients. This includes the “ABC” scoring system, which has been shown to be a helpful adjunct to clinical judgment in predicting need for massive transfusion (19–21). Umemura et al. (22) showed low fibrinogen and base deficit to be independent predictors of massive transfusion, and Shackelford et al. (23) generated a computer-based MT prediction algorithm based on vital signs and point of care labs. These and other studies highlight the important clinical need for early identification of patients who will require MT, so that the need for transfusions can be anticipated early, before the development of severe coagulopathy.

We initially expected to observe patients who are CAT+ to have an increased (prolonged) ttPeak, indicating an overall hypocoagulable state. Instead, the CAT+ patients have accelerated clot initiation (decreased ttPeak and Lagtime) but overall lower maximum thrombin generation (decreased peak height), consistent with consumptive coagulopathy. As such, the body’s attempt to obtain hemostasis is attenuated during hemorrhage. Previously, our group found that a decreased ttPeak, within 12 h after injury, is an independent predictor of VTE after traumatic injury (13). Hence, in reponse to injury, ttPeak appears to shorten (quicker to reach peak thrombin generation) in patients who require rapid transfusion and those who go onto develop VTE.

Unlike standard assays such as prothrombin time (PT), PT/INR that spot checks clotting potential at one time point, the Calibrated Automated Thrombogram assay evaluates an individual’s thrombin generation profile. As such, this assay has the potential to become a simple clinical tool to augment decision making at the bedside. In our multivariable analysis, both ttPeak and PT/INR were highly independently associated with transfusion rate (CAT+). That is, both added independent information to predicting CAT+. Both assays quantify distinct areas of the coagulation cascade. A PT is a test reflective of the extrinsic pathway and INR is calculated from a PT result and is used to monitor how well the blood-thinning medication (anti-coagulant) warfarin (Coumadin) is working to prevent blood clots. Thrombin (FIIa) cleaves plasma fibrinogen to produce insoluble fibrin and activates platelets which adhere with fibrin to produce a thrombus (clot). Its generation is the culmination of extrinsic pathway (Tissue Factor-dependent) during initiation phase, and also the propagation phase, during which small amount of thrombin, generated during initiation, augments the intrinsic pathway, via activation of FXI.

There are several limitations of this study. This is a subcohort analysis of a parent study, which was a prospective case-cohort study of 1,234 patients that had been designed to assess thrombin generation kinetics as predictors of symptomatic VTE after trauma. Hence, the number of patients requiring transfusions was limited by the enrollment design of the parent study. The parent study had a consent rate of 63%, so as with all observational studies sampling bias is of concern in the original data set. However, the excluded patients were largely similar to those that were enrolled with a similar median age (48 [28,68]), sex (69% men), and mechanism of injury (93% blunt) which should mitigate any selection bias. In terms of injury mechanism, 93% of our patients were blunt traumas, reflective of our center’s rural level one population, so we have less data on the efficacy of CAT in penetrating trauma victims and we did not stratify by injury mechanism to elucidate any differences in our patient population.

The CAT criteria utilizes the pRBC transfusion rate, and not that of plasma, which will have an impact on thrombin generation. However, the thrombin generation assay was from samples collected prior to any transfusions. The thrombin assay is currently used for resarch purposes only. Standard coagulation tests, which include PT and activated thromboplastin time assays, are measurements of the canonical extrinsic and intrinsic clotting pathways. These tests use excess concentrations of tissue factor or contact activator, which makes these tests relatively insensitive to perturbation of the clotting cascade, because the endpoints of these assays coincide with the onset of the propagation phase of clotting, where >95% of thrombin is still to be generated. Thrombin generation assay measures the rate of thrombin generation and the entire thrombin that can be generated in plasma. (24, 25). Lastly, viscoelastic hemostatic assays such as thromboelastograms were not performed at the same time as these sample blood draws. Hence, whole blood fibrinolysis marker as a predictor of MT was not evaluated.

CONCLUSION

The coagulopathy of injured patients in hemorrhagic shock, undergoing a high rate of transfusion (CAT+), is characterized by accelerated clot initiation (decreased Lagtime and ttPeak). The kinetic data afforded by the thrombin assay provides a more dynamic portrayal of clotting than the traditional clotting assays, such as the PT which yield only one time point and early in the process of thrombin generation. In our multivariable analysis, both ttPeak and PT/INR were highly independently associated with need for transfusion and transfusion rate (CAT+). These two assays have a complementary role in predicting those injured patients who will require a high rate of transfusions.