Obesity is Associated with Postinjury Hypercoagulability

Abstract

Background:

Obesity is linked to hypercoagulability with an increased risk of venous thromboembolic events (VTE) in the uninjured population. Therefore, we hypothesize that obesity (Body Mass Index (BMI) ≥30 kg/m²) is associated with a hypercoagulable state postinjury characterized by increased clot strength and resistance to fibrinolysis.

Methods:

Our prospective Trauma Activation Protocol database includes all trauma activations patients for whom a rapid thrombelastography (TEG) is obtained within 60 minutes postinjury prior to any transfusions. The dataset was then stratified by BMI and subjects with BMI ≥ 30 kg/m² were compared to those with BMI<30 kg/m²). The following TEG measurements were obtained: activated clotting time (ACT), clot formation rate (angle), maximum clot strength (MA), and % clot lysis 30 min after MA (LY30, %). Fibrinolysis shutdown (SD) was defined as LY30 < 0.6% and hyperfibrinolysis (HF) as LY30 > 7.6%. Continuous variables are expressed as median (IQR).

Results:

Overall, 687 patients were included of whom 161 (23%) had BMI ≥ 30kg/m² (BMI30). The BMI30 group was older, had a lower proportion of males and of blunt trauma, and were less severely injured. After adjustment for confounders, BMI30 was independently associated with lower odds of MA<55mm (OR 0.28; 95% CI 0.130.60) and of HF (OR 0.31; 95% CI 0.10–0.97) and higher odds of SD (OR 1.82; 95% CI 1.09–3.05). No independent association was observed with angle<65⁰ (OR 0.57 95% CI 0.30–1.05). While VTEs were more frequent among BMI30 patients (5.0 vs 3.3%), this did not reach significance after confounding adjustment (p=0.11).

Conclusion:

Obesity was protective against diminished clot strength and hyperfibrinolysis, and obesity was associated with an increased risk of fibrinolytic shutdown in severely injured patients. These findings suggest a relative hypercoagulability. Although no difference in VTEs was noted in this study, these findings may explain the higher rate of VTEs reported in other studies.

Background:

The prevalence of obesity continues to increase worldwide and has reached epidemic levels in the United States with 40 percent of Americans now suffering from obesity.(1, 2) Several studies have shown that the presence of obesity increases the risk of morbidity and mortality following trauma; however, the underlying reason for this increased risk is not well known.(3–6) One possible explanation is the presence of a hypercoagulable state and consequent thrombotic morbidity associated with obesity.(7) With this hypercoagulable state, obesity has been linked to both arterial and venous thrombotic complications such as myocardial infarction and pulmonary emboli, respectively.(8–11) In trauma patients, obesity has been associated with a higher rate of venous thromboembolic (VTE) events, cardiovascular and pulmonary complications, and an increased risk of multiple organ failure (MOF). (12–15)

The mechanism of the obesity-related hypercoagulability involves a pro-thrombotic state secondary to chronic inflammation as well as decreased clot breakdown via inhibition of the fibrinolytic pathway.(16) Prior studies have shown that the obesity-induced hyperinflammatory state triggers upregulation of procoagulant factors in vascular endothelium and triggers increased platelet activity and thrombin generation.(17) Furthermore, levels of the antifibrinolytic protein plasminogen activator inhibitor-1 (PAI-1) are increased in subjects with obesity, especially those with high levels of visceral fat.(18) However, the interplay of these prothrombotic pathways in the setting of trauma-induced coagulopathy after severe injury in obese patients is poorly described.

Despite existing literature on obesity’s effects on coagulation, the effect of obesity on trauma-induced coagulopathy and fibrinolysis phenotypes after severe injury is not well known. Our group has previously shown that fibrinolysis shutdown is associated with an increase in mortality following trauma.(19) Kornblith et al. documented several effects of obesity on trauma-induced coagulation, yet they did not detect a correlation between BMI and thrombelastography-measured fibrinolysis as a continuous variable measured, and no comparison with specific fibrinolysis phenotypes (i.e., hyperfibrinolysis, physiologic fibrinolysis, and fibrinolysis shutdown) was done.(12)

Understanding the contribution of obesity to fibrinolysis phenotypes in trauma is of critical importance given the growing interest in early empiric therapies for coagulopathy in trauma and the increasing prevalence of obesity in trauma populations. Therefore, the aim of this study is to evaluate the effects of obesity on trauma-induced coagulation and fibrinolytic phenotypes in trauma using TEG. Given the hyperinflammatory and hypercoagulable state that occurs with obesity, we hypothesize that obesity is associated with a hypercoagulable state characterized by increased clot strength and resistance to clot lysis (fibrinolysis shutdown).

Methods:

Setting

This is a retrospective analysis of the Trauma Activation Protocol (TAP) dataset, a prospective study of risk factors for trauma-induced coagulopathy in consecutive trauma activation patients at the Ernest E Moore Shock Trauma Center at Denver Health (DH), an American College of Surgeons verified and Colorado State certified Level-1 trauma center. Inclusion criteria for this study were adults (age ≥ 18) meeting criteria for the highest trauma designation and who had a rapid thrombelastography (TEG) collected in the first hour after arrival and prior to any blood transfusions. All subjects were enrolled in this study after approval by the Colorado Multiple Institutional Review Board (COMIRB #13–3087) under waiver of consent due to minimal risk. To conform to the guidelines of minimal risk, only a single time point blood draw is obtained.

Participants:

Criteria for trauma activation at DH are: 1) any injury with systolic blood pressure (SBP) < 90 mm Hg, 2) respiratory compromise, obstruction and/or intubation with mechanism attributed to trauma, 3) Glasgow Coma Scale (GCS) <9 with mechanism attributed to trauma, 4) mechanically unstable pelvic injury, 5) gunshot wound penetrating the neck/torso or stab wounds to the neck, torso, or extremity proximal to elbow or knee, 6) amputation proximal to the ankle or wrist or 7) fall greater than or equal to 30 feet, 8) the emergency medicine attending or chief surgical resident suspects the patient is likely to require urgent operative intervention. Patients were excluded if they were transferred from an outside hospital, pregnant, prisoner, minor (<18 years old), with known chronic liver disease, an inherited coagulation dysfunction, or were taking anticoagulants prior to injury. Obesity was defined as body mass index (BMI) ≥ 30 kg/m².(20)

Procedures:

Whole blood samples were collected in citrated vacuum tubes (3.5 mL, 3.2% sodium citrate, Greiner Bio-One, Monroe, North Carolina) either in the field by paramedics or upon arrival to the emergency department from May 2014 to April 2018. Citrated samples were recalcified and TEG was performed using the TEG 5000 Thrombelastography Hemostasis Analyzer (Haemonetics, Niles, IL) by trained professional research assistants. The TEG tracing provides the following measurements: activated clotting time (ACT [sec]), angle (°), maximum amplitude (MA [mm]), and percent lysis at 30 minutes after MA (LY30 [%]). ACT and angle represent clot formation while MA correlates with clot strength. Lastly, LY30 provides a measure of fibrinolysis (clot dissolution). Coagulopathy following trauma in this study was defined as ACT >128 seconds, MA <55 mm, and angle <65 degrees.(21) To determine fibrinolysis phenotypes, we used cutoffs derived for the trauma population as follows: fibrinolysis shutdown (SD) LY30 < 0.6% and hyperfibrinolysis (LY30 >7.7).(22) Functional Fibrinogen was performed as previously described.(23)

The TAP dataset prospectively collects transfusion data and clinical lab data including hemoglobin, platelet count, blood gas parameters, and conventional coagulation tests prothrombin time/International Normalized Ratio (PT/INR), and partial thromboplastin time (PTT). While research TEGs are completed prior to any transfusions received, clinical labs such as PT/INR and PTT may occur after the administration of transfusions or other interventions.

Outcomes and covariates

The primary outcome was the hemostatic profile by TEG (ACT, angle, MA, and LY30). Secondary outcomes included ventilator- and ICU-free days, clinical VTE, and mortality.(24) While guidance protocols exist for both, timing of extubation and ICU discharge was determined by the providing team. According to the guidelines by the American College of Chest Physicians our institution reserves routine VTE surveillance for high-risk trauma patients (e.g. pelvic factures and head injuries).(25)

Statistical analysis

Univariate analysis was conducted using t-tests, Wilcoxon tests, ANOVA or Kruskal-Wallis for numeric variables depending on whether the variable was normally distributed or skewed. Chi-square and Fisher exact tests were used for categorical variables. Multivariate analysis was conducted using logistic regression, with risk expressed with odds ratios (95% confidence intervals) for dichotomous outcomes and multinomial logistic regression for outcomes with greater than 2 categories (e.g. fibrinolysis). Model performance was assessed with AUROC (95%CI) for logistic regression models and with deviance and Pearson goodness of fit tests (for which higher p-values indicate better fit) for multinomial models. Effect modification by BMI30 on trauma-induced coagulopathy (TIC) indicators was tested by including interactions between TIC indicators and BMI30 into logistic regression models of death and VTE.

Results

A total of 687 (85%) patients out of 807 consecutive patients had available data on BMI and were included, of whom 161 (23%) had BMI ≥30 kg/m² (BMI30). Patients with and without BMI data did not differ regarding the outcomes or variables of interest suggesting data were missing at random. A consort diagram detailing patient exclusions can be found in Figure 1. The characteristics and outcomes of the study sample for subjects with and without BMI≥ 30 kg/m² are shown in Table 1. The BMI30 group had a lower proportion of men and of blunt trauma, was less severely injured (lower new injury severity score, NISS) and had a higher Glasgow Coma Scale (GCS). There were no significant differences in hospital arrival systolic blood pressure (SBP), base deficit, hemoglobin, or platelet count. The BMI30 group had shorter, albeit not statistically different, prothrombin time/International Normalized Ratio (INR) and partial thromboplastin time (PTT) (Table 2). However, no difference was seen with the proportion of patients with INR > 1.3 or PTT > 35sec.

Figure 1:

CONSORT Diagram

Table 1:

Study sample characteristics stratified by BMI. Data presented as median (25–75 interquartile range) or sample size (%) as appropriate.

	BMI< 30 kg/m2	BMI ≥ 30 kg/m2	P value *
Demographics
Age (Years)	33.4 (25.5 – 48.6)	38.0 (27.2 – 50.1)	0.08
Male (%)	432 (82)	121 (75)	0.04
Blunt mechanism (%)	306 (58)	82 (50)	0.01
NISS	22.0 (11 – 34.0)	18.0 (9.0 – 32.0)	0.1
Arrival GCS	14.0 (6.5 – 15.0)	15.0 (9.0 – 15.0)	< 0.01
Severe TBI**	146 (28%)	26 (16%)	< 0.01
Physiology upon hospital arrival
SBP (mmHg)	117 (91 – 140)	116 (90 – 140)	0.52
Base Deficit (mEq/L)	6.8 (4 – 10)	6.9 (3.6 – 10)	0.99
Hemoglobin (g/dL)	13.9 (12.4 – 15.2)	13.8 (12.6 – 15.5)	0.50
Platelet count (1,000/dL)	250 (201 – 305)	258 (217 – 305)	0.13
Transfusions within 24 hours postinjury
Required RBC (%)	204 (39%)	69 (43%)	0.37
Required plasma (%)	199 (38)	67 (42)	0.41
Required platelet transfusion (%)	108 (21)	29 (18)	0.47
Required cryoprecipitate (%)	53 (10)	17 (11)	0.87
Outcomes
Venous Thromboembolic Events	17 (3.2)	8 (5.0)	0.31
Deep Venous Thrombosis (%)	11 (2.1)	3 (1.9)	0.85
Pulmonary Embolus (%)	6 (1.2%)	5 (3.1%)	0.08
Deaths (%)	45 (8.6)	13 (8.1)	0.84
ICU Days	3 (1 – 7)	3 (1.5 – 6)	0.98
Ventilator Days	1 (0 – 4)	1 (0 – 3)	0.98
ICU free days	23 (4.5 – 26)	23 (13 – 25)	0.84
Ventilator free days	26 (11 – 28)	26 (18 – 28)	0.88

TBI – traumatic brain injury, BMI – body mass index, NISS – New Injury Severity Score, GCS – Glasgow Coma Scale, SBP – Systolic Blood pressure, RBC – red blood cell units, ICU – intensive care unit.

^*Wilcoxon test

^**Severe TBI defined as AIS head > 2

Table 2:

Differences in coagulation measurements between subjects with BMI ≥ 30 kg/m² and BMI < 30 kg/m² following injury. Data presented as median (25–75 interquartile range) or sample size (%) as appropriate.

	BMI< 30 kg/m2	BMI ≥ 30 kg/m2	P value *
INR	1.12 (1.03 – 1.25)	1.08 (1.01 – 1.21)	0.08
INR>1.3 (%)	99 (19)	25 (16)	0.91
PTT	27.2 (24.4 – 31.0)	26.3 (24.0 – 30.0)	0.13
PTT > 35	76 (15)	24 (15)	0.92
ACT (seconds)	121 (113 – 128)	121 (113 – 128)	0.60
ACT <128 seconds	197 (39)	51 (34)	0.22
MA (mm)	62.5 (56.8 – 66.3)	66.0 (61.0 – 69.4)	<0.01
MA<55 mm (%)	92 (18)	9 (6)	<0.01
Angle (degrees)	72.2 (66.7 – 75.7)	74.3 (67.8 – 77.7)	<0.01
Angle<65 degrees (%)	94 (19)	15 (10)	0.01
LY30 (%)	1.8 (0.9 – 3.0)	1.7 (0.5 – 2.9)	0.17
Fibrinolysis shutdown (LY30<0.6%) (%)	63 (14)	31 (23)	0.01
Physiologic fibrinolysis	344 (77)	99 (74)	0.44
(LY30 0.6–7.6%) (%)
Hyperfibrinolysis (LY30>7.7%) (%)	39 (9)	4 (3)	0.03
Functional fibrinogen (mg/dL)	410(347–475)	452 (383 – 502)	<0.01
Functional fibrinogen <356mg/dL (%)	141 (29)	27 (18)	<0.01

BMI- Body Mass Index - INR – International Normalized Ratio, PTT – Partial Thromboplastin Time, ACT – Activated clotting time, MA – Maximal Amplitude, LY30 – lysis 30 minutes after MA

Thrombelastography (TEG)

The TEG phenotypes of the two groups differed in several ways, with the BMI30 group having a higher median angle and MA and a lower proportion with a diminished angle (< 65°) and MA (< 55 mm) (Table 2). The two groups did not differ in ACT, and this lack of difference persisted after adjustment for confounders (age, sex, blunt mechanism, NISS, SBP and GCS, Table 3). Similarly, after adjustment for confounders, there was no association between BMI30 and angle< 65°; however, the BMI30 group was significantly less likely to have a low MA. The BMI30 group also had a higher median functional fibrinogen and a lower proportion with hypofibrinogenemia; although, this difference did not persist after adjustment for confounders.

Table 3:

Independent Effect of BMI>30 kg/m2 on thrombelastography (TEG) values

Dependent variable	Adjusted Odds Ratios (95% confidence interval)	Model performance ***
ACT>128 (seconds)	0.95 (0.60 – 1.51)	0.62 (0.57 – 0.68)
MA<55 (mm)	0.28 (0.13 – 0.60)	0.74 (0.69 – 0.79)
Angle<65 (degrees)	0.57 (0.30 – 1.05)	0.73 (0.68 – 0.78)
FLEV<365 (mg/dL)	0.63 (0.39 – 1.02)	0.74 (0.70 – 0.78)
Hyperfibrinolysis (LY30>7.7%) **	0.31 (0.10 – 0.97)	Pearson Goodness of fit = 0.72 *
Fibrinolysis shutdown (LY30<0.6%) **	1.82 (1.09 – 3.05)

All models adjusted for age, sex, blunt mechanism, New Injury Severity Score, arrival systolic blood pressure and Glasgow Coma Scale.

^*higher p-values indicate better fit;

^**Compared to physiologic fibrinolysis;

^***Area under the receiver-operating characteristics curve (AUROC) with 95% CI unless otherwise indicated ACT – Activated Clotting Time, MA – Maximal Amplitude, LY30- lysis 30 minutes after MA FLEV – Functional Fibrinogen Level.

The BMI30 group had a significantly higher proportion with fibrinolysis shutdown and lower proportion with hyperfibrinolysis (Table 2). After adjustment for the aforementioned confounders, BMI30 remained significantly associated with a higher likelihood of fibrinolysis shutdown and was protective against hyperfibrinolysis (Table 3).

Outcomes

There were no significant differences between groups in the rate of VTE, length of stay, ventilator free days, or ICU free days (Table 1). After adjustment for confounders, BMI30 was not associated with VTE or death (p=0.10 and p=0.71 respectively, both AUROCs>0.70). Interactions between BMI and TEG abnormal values in logistic regression models for death and VTE (adjusted for the confounders) were all non-significant (p>0.05), suggesting no effect modification, although a type 2 error could not be ruled out.

Discussion

Our study aimed to evaluate the contribution of obesity to the frequency of trauma-induced coagulopathy and fibrinolysis phenotypes following injury. There is limited data describing how fibrinolysis phenotypes are altered by the presence of obesity. Similar to prior studies by Kornblith et al., we found that BMI30 provided a protective effect against trauma-induced coagulopathy with less frequent diminished clot formation and low clot strength. Furthermore, BMI30 was significantly associated with a higher risk of fibrinolysis shutdown and with a lower risk of hyperfibrinolysis. This, however, did not translate to significant differences in thrombotic events such as VTEs or to worse outcomes such as decreased ICU free days or increased mortality.

The association of obesity with increased incidence of fibrinolytic shutdown and decreased incidence of hyperfibrinolysis following trauma is congruent with the mechanistic studies evaluating obesity’s effects on fibrinolysis. Obesity has been linked to an increased rate of VTE events in several studies.(8, 26–28) This risk has been attributed to an increased circulating level of PAI-1. PAI-1, a potent inhibitor of fibrinolysis, is increased in visceral adipose tissue, and subsequently subjects with obesity have higher levels of plasma-based PAI-1.(18, 29, 30). In contrast, hemorrhagic shock triggers a release of tissue plasminogen activator (tPA) stimulating a fibrinolytic response.(31, 32) However, it is likely that trauma patients with obesity have a higher capacity to counteract this hyperfibrinolytic response to hemorrhagic shock. While Kornblith et al. did not identify a correlation with BMI and fibrinolysis, by stratifying by fibrinolytic phenotypes (e.g. fibrinolytic shutdown or hyperfibrinolysis), we found that obesity alters a patient’s propensity for fibrinolytic shutdown.(12)

Our work further supports the findings of prior studies that obesity invokes a protective effect against trauma-induced coagulopathy. Previous work has shown that in animal models of trauma and human cohorts, obesity prevents prolongation of clot initiation, enhances fibrin cross-linking, and promotes clot strength following hemorrhagic shock. (12, 33) Several studies have also shown an interaction with obesity and platelet activity due to alterations in several pathways that invoke platelet activation.(34) Leptin, a signaling hormone produced by adipocytes has been shown to enhance platelet aggregation, promoting thrombosis.(35, 36) Similarly, tumor necrosis factor α (TNFα), which circulates at increased concentrations in obesity, has been shown to stimulate platelet activity as well.(37, 38) Preserved clot strength may also be due to increases in fibrin formation and cross-linking, since obesity has been associated with an increase in fibrinogen concentration and factor XII (which cross-links fibrin and enhances clot strength).(39, 40) Our finding of a protective effect of obesity against a diminished MA is consistent with previous studies. However, we failed to demonstrate an association with obesity and decreased risk of hypofibrinogenemia or altered fibrin cross-linking as demonstrated by a diminished TEG angle.

The findings in this study have several implications. First, the finding of increased clot strength in severely injured subjects with obesity suggests a more conservative platelet transfusion strategy may be appropriate in this population. Following the PROPPR trial, many trauma centers have adopted a 1:1 red blood cell to platelet transfusion strategy.(41) However, the PROPPR trial did not account for or report BMI or body weight, leaving unanswered how BMI may alter outcomes in the setting of a ratio-based platelet transfusion strategy. Furthermore, the association with obesity and clot strength may enhance the risk of VTEs. Although we did not identify differences in VTE rates, other studies have found an association with increased clot strength an VTE formation in trauma and non-trauma cohorts.(42, 43)

Interestingly, we did not identify an association with obesity and increased clot formation or functional fibrinogen. Several studies examining coagulation changes due to obesity in a non-trauma population have found an association with obesity and a higher fibrinogen level.(16, 40, 44) Similarly, our group recently published a prospective study comparing a healthy control group to an un-injured cohort with severe obesity, and we found the cohort with obesity had a significantly higher TEG angle and MA compared to controls, and this persisted six months after bariatric surgery.(45) One possible explanation is that we controlled for the presence of TBI. Our group previously published an association with severe TBI and a diminished angle and decreased fibrinogen.(46) By adjusting for TBI, we likely lost any association between obesity and alterations in these two variables.

This study has several limitations. First, as obesity frequently increases the risk of several comorbidities including hypertension, hypercholesterolemia, and type 2 diabetes mellitus, it is possible that a related disease process that accompanies obesity is the true underlying cause for these findings. However, several studies, including work by this group have shown that obesity, independent of other comorbidities is associated with a hypercoagulable state. (47–49) Also, we may be underestimating the effect of obesity on fibrinolysis in treating obesity as a categorical variable, instead of evaluating BMI as a continuous variable. It is possible that changes in fibrinolysis in response to weight gain occur in linear fashion with diminished fibrinolysis with any increase in BMI. Lastly, at our center, routine surveillance for venous thrombosis is only done for high risk patients and at the discretion of the attending physician. As a result, the true incidence of VTEs may be underreported. This resulted in a lack of power for assessing differences in VTEs and could account for the lack of a difference seen between the two groups.

Conclusion

This study further supports obesity’s protective effect against trauma-induced coagulopathy. Importantly, we found that obesity is associated with an increased risk of fibrinolytic shutdown and decreased risk of hyperfibrinolysis.