Blunt thoracic trauma occurs in 10% to 17% of all trauma patients and accounts for 20% to 50% of the approximately 200,000 annual trauma-related deaths in the United States (1–3). Approximately one-third of blunt thoracic trauma admissions include pulmonary contusion, associated with 10% to 25% mortality in the days to weeks after injury (4). Pulmonary contusion secondary to thoracic trauma gives rise to both localized and systemic derangements, including infections, acute respiratory distress syndrome (ARDS), multisystem organ failure (MOF), and thromboembolic events (2). Thromboembolic events such as pulmonary embolism (PE) in fact have been demonstrated to have an incidence which is approximately 42% higher in patients with severe chest trauma defined as an Abbreviated Injury Scale score of 3 or higher when compared with trauma patients without severe chest trauma (5).
PE is a troublesome consequence of severe trauma with an associated mortality as high as 30% depending on size and location within the pulmonary system (4, 6). Pulmonary thromboembolic events are almost exclusively considered to be embolic, originating from thrombus within a systemic deep vein. Based on the assumption that PE is always the sequel of systemic venous thrombosis, it follows that patients with a newly diagnosed PE should also show evidence of recent DVT (4, 7). However, repeatedly clinical trials have shown that this is not always true, particularly in the blunt thoracic trauma population (4, 7–11). A recent review of the National Trauma Data Bank (NTDB) by Knudson et al. examined the injury data of 888,652 trauma patients and reported a substantial increase in the incidence of PE without an increase in the incidence of DVT (5). Furthermore, reports have shown CT scans with filling defects of the pulmonary arteries obtained within hours after blunt thoracic trauma, challenging the appropriateness of the default diagnosis “pulmonary embolism” and suggesting that perhaps it should be referred to as a “pulmonary thrombosis” because it is very likely that these clots arise in situ via de novo arterial thrombosis (7, 12, 13). Based on this inference coupled with the improbability of early lower extremity DVT after thoracic trauma, we hypothesize that blunt thoracic trauma precipitates early events leading to de novo pulmonary arterial thrombosis.
To address this issue, we examined lung tissue pathology in a murine model of blunt unilateral thoracic trauma. Murine models are shown to exhibit hallmark characteristics of human lung contusion, including immune cell recruitment, inflammation, and edema beginning within the first few hours after injury (14–19). Furthermore, experimental murine models demonstrate that de novo thrombosis begins within minutes to hours after venous occlusion (20) or arterial injury (21, 22). Therefore, murine models are well suited for investigating early events in de novo thrombosis in the wake of severe blunt thoracic trauma.
In the present study, we describe a medium velocity blunt lateral thoracic trauma model that is relevant for sports-, automobile-, or falling-related injuries. This simplified weight-drop model is convenient, easy to perform and can be coupled with other forms of experimental polytrauma. Using the weight-drop model, we induce a reproducible lung contusion, including alveolar hemorrhage, edema, and immune cell recruitment, with minimal extrapulmonary injury. Furthermore, we find histological evidence for de novo pulmonary arterial thrombosis in both injured coup and contrecoup areas of lung tissue, within 24 h after traumatic injury.
MATERIALS AND METHODS
All procedures in this study were conducted in accordance with the Institute for Laboratory Animal Research Guide for the Care and Use of Laboratory Animals, and the University of California Davis Animal Welfare Assurance on file with the US Public Health Service; IACUC approved UC Davis animal use and care protocol no. 19038. Adult male C57 black/6 mice (retired breeders aged 7–10 months; weight ranged 30–45 g; Charles River) were used in this study. Before and during experimentation, both injured and uninjured (sham) mice were anesthetized with 5% isoflurane in oxygen gas, and maintained at 2% via nose cone. Both sham and injured mice received buprenorphine (s.c. 0.05 mg/kg body weight) postoperatively, and were placed in a heated cage to maintain core body temperature at 37°C for at least 1 h after the injury procedure. Mice were housed individually and given free access to food and water both pre- and postoperatively.
The weight-drop blunt thoracic trauma model
The blunt thoracic trauma model consists of an anesthetized mouse rested in the left lateral decubitus position on a solid metal platform (Fig. 1A). The platform was positioned horizontally using a carpenter's level, and stabilized with modeling clay on a solid surface. The weight-drop device consists of a 50 g brass weight (19.6 mm diameter) attached via a swivel to a fishing line, dropped from a specified height through a guide tube (20.5 mm i.d.) mounted on a ring stand. The guide tube was positioned vertically and leveled with the open end immediately over the mouse's right lateral rib cage, centered on the axillary line inferior to the axilla. Immediately after impact, the mouse was rested on a 37°C heating pad in the supine position to observe recovery of reflexes and normal breathing behavior.
Weight-drop impact velocity was determined from video (240 frames/s) frame-by-frame analysis of displacement distance. Displacement data were fit to a second-order polynomial function with free-fall acceleration set at 9.81 m/s2. The first-order derivative was used to calculate velocity at the impact position (x = 0 cm) based on the extrapolated time of exit from the guide tube (Fig. 1B). Using this approach, impact energy (E = ½ mv2) was determined at each drop height in this study.
A total of 92 mice were included in the present study. Mice were divided into five groups, including a control sham group and four experimental injury groups: 35 cm, 40 cm, 45 cm, and 50 cm drop heights. Mean (±SD) body weights in each group were 38.5 ± 6.5 g (sham), 37.0 ± 5.4 g (35 cm), 36.4 ± 2.9 g (40 cm), 38.6 ± 3.2 g (45 cm), and 38.4 ± 2.5 g (50 cm). Individual mice were randomly allocated into their respective experimental groups, with 5 to 10 mice in each group designated either for lung pathology/histology, or for fluid (bronchoavleolar lavage and blood/plasma) analysis as described in the following sections. All live animal experiments were performed in the UC Davis Surgery Department surgical suite during daytime hours (9:00 AM to 6:00 PM, US PST), and mice were otherwise housed in the Teaching Research and Animal Care Services vivarium at UC Davis Medical Center. Postexperimental euthanasia consisted of exsanguination under isoflurane gas anesthesia.
Blood collection and analysis
Blood was collected via left ventricular puncture, into heparinized tubes, and arterial PO2, PCO2, pH, and base excess (BE) were determined using a Stat Profile Critical Care Express (Nova Biomedical) point-of-care blood analyzer. Plasma was separated from formed elements by successive centrifugation steps at 800 × g and 10,000 × g, then flash frozen, and saved at −80°C for future immunoassays. Clarified plasma was thawed once and used for biomarker analysis using mouse ELISA kits for Myoglobin (Abcam #ab210965), Creatine Kinase MB Isozyme (MyBiosource, Inc. #MBS762450), or Cardiac Troponin I (Life Diagnostics, Inc. #CTNI-1-US).
After aortic exsanguination, bronchoalveolar lavage (BAL) was performed in two sequential 1 mL phosphate-buffered saline (PBS) washes via a tracheostomy cannula. For one-sided analyses, BAL was similarly performed with 600 μL (right) or 400 μL (left) with the opposite side bronchus temporarily occluded. BAL fluid was stored temporarily on ice. Total leukocyte count was assessed by hemocytometer, and differential counts were performed via light microscopy after cytocentrifugation (Cytospin 4; Thermo) and Wright-Giemsa staining. Remaining BAL fluid was stored at −80°C for further analysis.
Frozen BAL fluids were thawed once and analyzed for total protein using a Pierce BCA Protein Assay Kit (Thermo Scientific), and for chemokine/cytokine content using a Cytometric Bead Array Mouse Inflammation Kit (Becton-Dickinson Life Sciences #552364) and BD Canto II flow cytometer (UC Davis Flow Cytometry Core).
After aortic exsanguination, lungs were gravity perfused (30 cm H2O) via intratracheal cannula for 10 min with 10% neutral-buffered formalin. Excised tissues were immersed in 10% neutral formalin for 12 to 18 h and paraffin embedded for sectioning at 5 μm thickness. Tissue sections were subject to routine hematoxylin and eosin (H&E) staining, or immunofluorescence (described below). Ten randomly acquired light microscopy (200×) images of H&E stained sections were analyzed and scored by two independent blinded evaluators for acute lung injury score (three criteria: alveolar thickness, hemorrhage, and cellularity; 0–3-point severity scale), as previously described (23, 24).
Cardiac and lung tissue paraffin sections prepared as described above were subjected to terminal deoxynucleotidyl transferase (TdT) dUTP Nick-End Labeling (TUNEL) performed according to the manufacturer's specifications (In Situ Cell Death Detection Kit, TMR-red, Roche Biochemical Reagents, Waltham, Mass). Whole tissue cross sections were imaged (50× magnification) by fluorescence microscopy. Representative images and individual randomly selected arteries were imaged at 200× magnification. For arterial cross sections, individual data points represent mean values from 10 randomly selected fields of view. In each instance, the Apoptotic Index was determined from the ratio of TUNEL positive cells to DAPI positive nuclei, and expressed as a percent.
Hydrated deparaffinized tissue sections were blocked in PBS with 5% BSA, followed by incubation overnight with the following primary antibodies: fibrin/fibrinogen (1:1,000, rabbit polyclonal IgG; Lifespan BioSciences #C150799), CD41 (1:200, Rat IgG1κ; BioLegend #133901). Fluorophore-conjugated secondary antibodies were from Thermo Fisher Scientific: Goat anti-rat Alexa Fluor 647 (A-21247; 1:1,000), Donkey anti-rabbit Alexa Fluor 546 (A-10040; 1:1,000). After immunolabeling and DAPI counterstaining, coverslips were mounted with Prolong Gold Antifade Mountant (Thermo Fisher).
Microscopy imaging and analysis
Microscope images acquired were performed with an Axio Observer 200 M inverted microscope (Zeiss, Jena, Germany) and a digital color camera for light microscopy, or monochromatic camera and arc lamp illumination using selective filters (DAPI/eGFP/Cy3/Cy5) for fluorescence. Images were analyzed using Image J software (W.S. Rasband, Image J, U.S. National Institutes of Health, Bethesda, Md, http://rsb.info.nih.gov/ij/). For each individual mouse, arterial luminal fluorescence was quantified as the integrated fluorescence intensity per linear micrometer of the arterial luminal surface from at least 10 randomly acquired microscopy (200×) images of immunostained arterial cross sections (from 5 μm thick lung tissue sections). Gross pathological images were acquired using a dissection microscope equipped with a standard digital camera.
Statistical analyses and graphical presentation were performed using GraphPad Prism version 5.00 (GraphPad Software, Inc.). Data are presented in whisker plots as median ± interquartile range (mean values are denoted by + symbols). Replicates (n) represent numbers of individual mice. Survival data comparisons were performed by Mantel–Cox log-rank test. Timed series data series were analyzed by two-way analysis of variance (ANOVA), followed by Bonferroni posttests to compare mean values at individual time points. Mean comparisons versus sham were analyzed by one-way ANOVA, followed by Dunnett's multiple comparison test. Paired analyses presented as bar graph data were performed using Student t test. For mean differences, the level of significance was set at P < 0.05 (denoted as 0.05*, 0.01**, or 0.001***).
Physical characteristics of the weight-drop injury model
Our aim was to produce a severe, yet nonfatal traumatic thoracic contusion injury using a simple weight-drop model. To ascertain impact velocity and reproducibility of the model, a 50-g (19.6 mm diameter) brass weight was dropped from various fixed heights (25 cm, 30 cm, 35 cm, 40 cm, 45 cm, and 50 cm). High-speed video frame analysis of vertical displacement demonstrated measured impact velocities of 1.88, 2.04, 2.22, 2.45, 2.76, and 2.97 m/s (Fig. 1B), which all were within 12% of predicted theoretical velocities (velocity equals the square root of 2 × gravitational force × height), corresponding to impact energy densities of 305, 358, 426, 517, 657, and 761 J/m2, respectively. For comparison with other models, we also estimated the time of contact and depth of tissue compression with a mouse in place. Using video analysis at the 45 cm drop height (n = 6), the maximum recorded compression depth into the mouse's thorax ranged 3.6 mm to 5.6 mm, which occurred at approximately 8.3 ± 4.2 ms (2 ± 1 frames) after initial contact.
After impact at 35 cm, 40 cm, 45 cm, or 50 cm, mice were observed for indications of injury severity. We observed brief periods (<15 s) of apnea and labored breathing in injured mice at all drop heights. For all mice included in the study, acute survival was monitored over a period of 1 h. One hundred percent of the mice (10/10) survived acutely at at most 35 cm drop heights (Fig. 2A). Most of the mice (88%–90%) also survived at the 40 cm and 45 cm drop heights (9/10 and 22/25, respectively), with a higher acute death rate (72% survival; 13/18) at the 50 cm drop height (Fig. 2A). In all groups, acute deaths occurred within the first 5 min, coinciding with hemothorax. For mice that died acutely, necropsy revealed various sources of intrathoracic bleeding that included atrial appendage rupture, bleeding from the superior surface of the heart (possibly large blood vessel rupture), and three instances (at 50 cm) of rib fracture with right cardiac ventricular puncture.
Blood chemistry analysis after blunt thoracic trauma
At 24 h after blunt thoracic trauma, we analyzed arterial blood gases and pH. Arterial PCO2, pH, and extracellular BE (BEECF) were consistent with reported normal values for isoflurane anesthetized mice (25), and were not significantly different in mice injured at 45 cm or 50 cm (Fig. 2B–D). As expected, arterial PO2 was elevated for all mice due to 100% O2 included with the gas anesthesia. Together the data indicate no significant alterations in gas exchange, blood pH, or base deficit at 24 h after blunt thoracic trauma compared with uninjured sham mice.
Gross pathological characteristics of blunt thoracic trauma
To characterize pulmonary injury, we performed gross necropsy at 24 h after injury. At the site of impact, we found little or no evidence of superficial tissue injury. In mice subjected to a 50 cm drop height, an occasional area of redness was observed on the underlying interior thoracic wall, as well as evidence of rib fracture in 3 of 19 mice examined. Similarly, gross cardiac or vascular tissue damage was evident in many of the mice injured at 50 cm (not shown). In contrast, no overt signs of cardiac or other extrapulmonary thoracic injury were found in mice injured at drop heights at most 45 cm (Fig. 3A and B). Despite an apparent lack of extrapulmonary thoracic injury in mice injured at 45 cm, we observed large areas of pulmonary contusion (Fig. 3A and B) compared with uninjured sham mice (Fig. 3C). Areas of contusion were present to a lesser extent in mice injured at 35 cm or 40 cm drop heights (not shown). Gross examination of mice injured at drop heights at most 45 cm revealed no visible extrathoracic abdominal injury, i.e., liver, spleen, intestines, or kidneys (not shown), suggesting that the energy of blunt force trauma was transferred primarily to the lung tissues.
To more quantitatively evaluate possible skeletal or cardiac muscle injury, we performed enzyme-linked immunosorbent assays (ELISA) for plasma biomarkers: myoglobin, creatine kinase (CK)-MB, and cardiac troponin I (cTnI) at 24 h after thoracic trauma. Plasma myoglobin, CKMB, and cTnI levels remained virtually unchanged irrespective of injury at 45 cm or 50 cm, reflecting no measureable skeletal or cardiac muscle injury in injured versus sham mice (P > 0.05; n ≥ 6 each) (Fig. 3D). Fixed cardiac tissue sections were also more carefully examined for injury using a TUNEL apoptosis assay. Gross visual inspection of cardiac tissue section TUNEL staining or H&E staining revealed no obvious areas of contusion (not shown). Similarly, using the TUNEL assay, we found no significant increases in Apoptotic Index (percent apoptotic cells) in myocardial tissues from mice injured at 45 cm or 50 cm compared with uninjured mice (Fig. 3E and F).
Histological evidence for lung contusion injury
To further support our conclusion of lung contusion injury after blunt thoracic trauma, H&E stained lung tissue sections were examined by light microscopy at 24 h after injury. Visual inspection of (stitched images from a 5× objective) whole lung tissue sections revealed areas of redness in the alveolar tissues, most apparent in the coup injury (right) side (Fig. 4A). Higher magnification (200×) images were examined and scored for alveolar hemorrhage, membrane thickening, and cellularity on the injured (coup) and contrecoup sides of the lungs (Fig. 4B and C), compared with uninjured sham lungs (Fig. 4D). Blinded histopathological evaluation yielded significantly higher composite lung injury scores in injured mice (compared with sham), consistent with lung contusion injury (Fig. 4E). Lung injury scores increased with increasing drop height, especially on the injury coup side (significant linear trend; P < 0.05) compared with contrecoup or sham (Fig. 4E).
To assess more widespread indications of lung injury, TUNEL assays were performed on fixed lung tissue sections in tissue areas with obvious contusion (as in Fig. 4A inset), as well as distal sites on both coup and contrecoup sides of the lungs. Multiple TUNEL positive cells were observed within areas of obvious contusion (Fig. 5A, row 2, right column) compared with relatively few TUNEL positive cells observed in areas outside the contusion areas on either the coup (Fig. 5A, row 3, right column) or contrecoup (Fig. 5A, bottom row, right column) sides of lungs from injured mice, or sham mice (Fig. 5A, top row, right column). Apoptotic indices from TUNEL stained lung tissues revealed no significant increases in Apoptotic Index in whole lung tissues on either the coup or contrecoup sides of lungs from mice injured at 45 cm or 50 cm drop heights compared with sham mice (Fig. 5B). Suggestive though not significant (P > 0.05; estimated 9- to 10-fold; n = 5 each coup/contrecoup) increases in apoptotic cells might be inferred from the lungs of mice injured at 50 cm, but not whatsoever from mice injured at 45 cm (n = 5). Such a difference might be predicted based on more widespread areas of obvious alveolar tissue damage in mice injured at 50 cm compared with 45 cm. In support of this interpretation, directly contused areas of tissue (as identified in corresponding H&E stained sections) showed a mean 70-fold significant (P < 0.001) increase in the local densities of apoptotic cells (Fig. 5B, con). Together this suggests that increases in apoptotic cell counts are elevated markedly in areas of direct tissue injury and bleeding (predominantly on the injury coup side; Fig. 5A, row 2, right column), whereas apoptotic cells are virtually nonexistent outside of focal injury areas (Fig. 5A, rows 3–4, right side). Similarly, randomly acquired images of pulmonary artery cross sections were analyzed from sham and from injury coup or contrecoup sides of lungs from injured mice (Fig. 5C). Very few TUNEL positive cells were identified in the arterial walls of any of these images (Fig. 5C and D). Likewise, apoptotic indices were very low in pulmonary arterial tissues from coup or contrecoup sides of lungs from mice injured at 45 cm or 50 cm drop heights with no significant differences found compared with sham mice (Fig. 5D). Together this suggests that apoptotic cell death is mainly confined to alveolar and microvascular tissues located specifically in areas of obvious bleeding or focal tissue injury. Moreover, pulmonary arteries did not contain appreciable numbers of apoptotic cells after injury at any drop height used in our model.
BAL leukocyte and fluid analysis
To further investigate the nature and extent of lung injury, bronchiolar lavage (BAL) fluid was analyzed for total and differential leukocyte counts, 24 h after injury at 45 cm or 50 cm. Total leukocyte counts were significantly higher in BAL samples from mice injured at 45 cm (2.1-fold; P < 0.01) or 50 cm (2.7-fold; P < 0.001) compared with sham (n ≥ 8 each) (Fig. 6A). Differential counting revealed significantly elevated macrophage (P < 0.01), neutrophil (P < 0.01), and lymphocyte (P < 0.01) counts in samples from injured mice compared with sham mice (Fig. 6A), consistent with immune cell recruitment to the alveolar tissues after blunt thoracic trauma. In particular, neutrophil counts were dramatically elevated, 30-fold and 44-fold, for 45 cm and 50 cm drop heights, respectively, consistent with lung tissue inflammation. In addition to leukocyte counts, BAL fluid protein content also increased approximately 3-fold in lungs from mice injured at 45 cm or 50 cm drop heights (Fig. 6B). To further understand the basis for this increase, in a subset of mice injured at 45 cm (n = 6), BAL fluids were collected specifically from the right versus left side of the lungs. In these mice, BAL fluid protein was significantly increased on the right side, similar to the whole lung levels, though no such increase was seen on the left side (Fig. 6B). Consistent with accompanying histopathology results (Fig. 4), this unilateral protein accumulation suggests that alveolar tissue injury mostly occurs on the injury coup side of the lungs.
Inflammatory processes were further investigated by examining chemokine and cytokine levels in whole lung BAL fluids. Specifically, the chemokine monocyte chemoattractant protein (MCP)-1 was significantly elevated in BAL fluids from the lungs of mice injured at 45 cm or 50 cm (Fig. 6C), suggesting that monocytes are actively recruited to the injured lung tissues as part of the inflammation process. BAL cytokine levels were also assayed, demonstrating significant elevation of proinflammatory cytokines tumor necrosis factor (TNF) and interleukin (IL)-12 (active heterodimer, p70) in BAL fluids from mice injured at 45 cm and at 50 cm, relative to that of sham mice (Fig. 6D). Notably, in the same BAL fluid samples, no significant increases in IL-6 or IL-10 were found (not shown).
Histological evidence for pulmonary thromboembolism
Historically, pulmonary blood clots develop as detached emboli that arise from systemic DVT. More recent human trauma studies suggest that pulmonary arterial clots appear early (<24 h) in the absence of DVT, and may represent de novo pulmonary arterial thrombosis. To investigate this in our model, we examined lung tissue histology at 24 h after injury. Light microscopy examination of H&E stained lung tissues revealed visual evidence of eosin positive acellular proteinaceous material associated with the arterial walls in an eccentric pattern, suggestive of thrombosis (Fig. 7A). To further support this conclusion, additional fixed tissue sections were immunofluorescently labeled with antibodies raised against fibrin/fibrinogen and platelets (CD41). Corresponding images of arterial cross sections showed eccentric patterns of fibrin/fibrinogen and CD41 positive fluorescence, confirming the presence of characteristic thrombus materials (Fig. 7B).
To determine the extent of thrombosis, randomly selected medium–large arteries were imaged (200×) in lung tissue sections from mice injured at 40 cm, 45 cm, or 50 cm, and examined for arterial eccentric fibrin immunofluorescence. Fluorescence images revealed fibrin accumulation in pulmonary arteries in both the injury coup (Fig. 7C) and contrecoup (Fig. 7D) sides of the lungs, whereas arterial fibrin was much less apparent uninjured sham lung tissues (Fig. 7E). Quantitative evaluation of circumferential fibrin/fibrinogen fluorescence per micrometer of the arterial wall revealed significant (P < 0.05) 2.4-fold and 2.5-fold increased fibrin/fibrinogen accumulation in pulmonary arteries on the injured coup side of lungs injured at 45 cm and 50 cm, respectively, compared with sham. Surprisingly, similarly significant (P < 0.05) 2.4-fold increases in arterial fibrin/fibrinogen immunofluorescence were also present in contrecoup lung tissues from mice injured at 45 cm and 50 cm, suggesting that early de novo thrombotic events occur to a similar extent in arteries both proximal and distal from the site of blunt thoracic injury (Fig. 7F).
This study provides evidence of de novo pulmonary arterial thrombosis in a translational mouse model of blunt thoracic trauma and lung injury. Given that current VTE prophylaxis and thrombolytic interventions are ineffective at preventing pulmonary thromboembolism after trauma, great need exists for a simplified in vivo translational model in which to test novel intervention strategies aimed at improving patient outcomes after severe blunt thoracic trauma and polytrauma. Our mouse model is ideal for this purpose in that we produce severe pulmonary injury in absence of measurable mortality or extrapulmonary injury, and that we observe pulmonary arterial thrombosis that mimics PE seen in clinical trauma patients.
Here we developed a simplified weight-drop model of lateral blunt thoracic trauma in adult male C57BL/6 mice, a commonly used strain for mechanistic studies with genetically modified mice. A convenient feature of the weight-drop model is that drop height and weight parameters can be varied to adjust impact velocity and energy, to produce injury of varied severity. Using high-speed video capture, we confirmed that velocity at the time of impact is highly reproducible at each given drop height in our study. Using a 50-g weight, our model consistently produces lung contusion pathology, with little or no extrapulmonary injury at at most45 cm drop height.
Similar to our mouse blunt thoracic trauma model, a variety of other rodent chest trauma or pulmonary contusion models have been reported. For example, cortical impactor models are implemented to create highly controlled penetrating or focal lung injury over a limited area (16, 26) including a mouse model of unilateral blunt chest injury used by Hoth et al. using a 1 cm diameter cortical impactor. Additional blunt trauma models use a contusive pendulum or a controlled blast to create a more broadly distributed thoracic injury (19, 27, 28). Previously reported rat ventral thoracic trauma models were shown to create both lung contusion injury and extensive injury to the heart (27). To circumvent this, Wang et al. (29) developed a dorsal injury weight-drop model in rats using a protective shield to direct the impact energy specifically toward the lungs. The rat weight-drop model is used to deliver a specified impact energy by raising or lowering the height of a metal weight of known mass (Energy = mass × gravitational force × height). This type of contusion model is convenient for creating a controlled injury of known severity, and with good translational representation of sports injury or falling-related accidents, in which medium velocity impacts transfer energy across a large surface area of the rib cage. Our mouse thoracic trauma model is similar in these respects. In contrast, our mouse model is unique in using a lateral impact approach to circumvent cardiac injury, with a larger impact surface area to create a more broadly distributed injury across the thoracic surface. To our knowledge, a full-surface lateral thoracic impact model has not previously been developed for mice.
In our thoracic trauma model, we identified hallmark characteristics of lung contusion injury: vascular congestion, tissue edema, alveolar fluid accumulation, and leukocyte recruitment. In addition, we found widespread evidence of alveolar hemorrhage, especially at the interface of the superior and middle lobes of the right lung, directly under the area of impact. Based on this, we infer that our model produces a predominantly right-side lung contusion injury at drop heights of at most 45 cm. At 50 cm, we observed more overt gross evidence of left-side lung injury coinciding with cardiac and skeletal injury (including obvious rib fractures in three mice), and hemothorax. Although skeletal and cardiac muscle injury plasma biomarkers (cTnI; CKMB; myoglobin) were not elevated, gross pathology suggests that the greater impact energy at 50 cm is sufficient to cause deeper and more widespread tissue injury. In contrast, impact at at most 45 cm creates a consistent right-side lung contusion injury without significant damage to the heart or other neighboring tissues, with moderate injury severity comparable to a human thoracic Abbreviated Injury Score (AIS-90) of approximately 3 to 4.
Using video frame-by-frame analysis at 240 fps, we estimated the depth of tissue penetration in our model to be 3.6 to 5.6 mm (a likely underestimation due to frame capture rate of 240 fps), and with a contact duration of 8 ± 4 ms. This is similar to the cortical impactor mouse blunt chest injury model described by Hoth et al. (16) who reported a tissue penetration depth of 6.3 mm at a constant impact velocity of 5.8 m/s, delivering an estimated energy density of 152 J/m2. At the conditions in our study (45 cm, 50 g), video frame-by-frame analysis demonstrated an impact velocity of 2.76 m/s, corresponding to an impact energy of 657 J/m2, which is 3 to 4 times that reported for murine contusion impactor models of blunt thoracic trauma (30, 31), and without cardiac injury as reported in rodent models of blunt chest trauma (19, 27). Therefore, our model may offer some advantages over other murine models for translational studies of pulmonary injury in blunt thoracic trauma.
Similar to other thoracic trauma models, we observed indications of inflammatory processes. Our model included elevated BAL levels of the chemokine MCP-1 and proinflammatory cytokines TNF and IL-12 at 24 h after blunt thoracic trauma. We found no increase in BAL levels of IL-6 or IL-10 in our model (data not shown). This latter observation is consistent with Hoth et al. (14), who found that levels of cytokines including IL-1 (BAL) and IL-6 (plasma) that were elevated at 3 h were no longer elevated at 24 h after blunt chest trauma. Similarly, Suresh et al. (31) found no significant elevation of the BAL cytokines IL-1β, IL-6, or IL-10 at 24 h after blunt chest trauma in mice. This is in contrast to the blast model of blunt chest trauma in which Niesler et al. (32) found significant elevation of BAL cytokines (IL-6 and IL-10) that persisted at 24 h after injury. Similar to our findings, both Hoth (14) and Niesler (32) reported increased BAL neutrophils at 24 h after blunt chest trauma in mice. Together, these results suggest that the BAL cytokine expression changes over time (3 vs. 24 h) after blunt chest trauma, and that the cytokine expression profile may be slightly different after blast injury compared with impact-based blunt thoracic trauma models.
A major goal of our study was to produce a clinically relevant blunt thoracic trauma model in which to study the development of de novo pulmonary arterial thrombosis. After blunt thoracic trauma, we identified eccentric aggregates of fibrin and platelets adherent to the luminal surface of pulmonary arteries in injured mice. Moreover, eccentric fibrin accumulation occurred similarly in both injured coup and contrecoup sides of the lungs, including areas of lung tissue distinctly void of hemorrhage or other visible injury. This suggests that de novo pulmonary arterial thrombosis begins within 24 h after blunt thoracic trauma, and that primary thrombotic events occur ubiquitously within the pulmonary arteries irrespective of the location of primary tissue injury.
In our study, we demonstrate early evidence of eccentric fibrin and platelet accumulation, the beginning events of thrombosis (33, 34). Clinically, fully developed pulmonary thromboemboli are identified in only a small percentage (albeit a large number) of clinical patients with severe traumatic injury. In contrast, our experimental evidence shows a high frequency of early thrombotic events in the pulmonary arteries, suggesting that these events are pervasive after blunt thoracic trauma in mice. Given investigative limitations, it is not possible to determine how many of these early thrombotic events would ultimately fully develop into obstructive clots in our model. However, this supports the interpretation that early thrombotic events are ubiquitous in the pulmonary arteries and extremely common after blunt trauma, and that in most cases, would resolve naturally over time due to protective fibrinolytic mechanisms. Such early subclinical thrombotic events would be impossible to detect using clinically available imaging modalities, and may only present once an occlusive thrombus is formed. The prevalence of early subclinical pulmonary thrombotic events may be an important factor for increasing the likelihood of thromboembolytic events leading to obstructive and lethal pulmonary arterial clots after trauma.
An improved understanding of the mechanism behind pulmonary thromboembolic events is imperative for both its treatment and prevention. Current prophylactic strategies such as early ambulation and intermittent lower extremity mechanical compression are measures that aim to prevent blood stasis in the deep veins (35). Furthermore, high-risk patients are often subjected to inferior vena cava (IVC) filters based on the theory that all thromboembolic events in the lungs are secondary to DVT (36). These thromboembolic events are frequently reported as markers of quality care as they are considered preventable complications. Interestingly, despite the increased utilization of prophylactic measures the incidence of PE in the trauma population continues to rise (7). Although this increased incidence is likely multifactorial, our failure to prevent this complication should make us question if we are treating the right disease and if PE should be considered a complication at all. In the setting of trauma, pulmonary thrombus may have a different natural history, morbidity, and effective treatment, necessitating a diagnostic distinction and calling into question its use as an indicator of quality.
The authors thank Jubi Lin, Zach Paxton, Andrew Escobar, and Zoe Saenz for excellent technical assistance, as well as the UC Davis Center for Genomic Pathology Laboratory staff for histological sectioning services, Dr. A.D. Borowsky for pathology consultation, and Jonathan Van Dyke for flow cytometry consultation.
1. Kochanek KD, Murphy SL, Xu J, Tejada-Vera B. Deaths: final data for 2014. Natl Vital Stat Rep
65 4:1–122, 2016.
2. Raghavendran K, Notter RH, Davidson BA, Helinski JD, Kunkel SL, Knight PR. Lung contusion
: inflammatory mechanisms and interaction with other injuries. Shock
32 2:122–130, 2009.
3. Gayzik FS, Martin RS, Gabler HC, Hoth JJ, Duma SM, Meredith JW, Stitzel JD. Characterization of crash-induced thoracic loading resulting in pulmonary contusion. J Trauma
66 3:840–849, 2009.
4. Ganie FA, Lone H, Lone GN, Wani ML, Singh S, Dar AM, Wani NU, Wani SN, Nazeer NU. Lung contusion
: a clinico-pathological entity with unpredictable clinical course. Bull Emerg Trauma
1 1:7–16, 2013.
5. Knudson MM, Ikossi DG, Khaw L, Morabito D, Speetzen LS. Thromboembolism
after trauma: an analysis of 1602 episodes from the American College of Surgeons National Trauma Data Bank. Ann Surg
240 3:490–496, 2004.
6. Abad Rico JI, Llau Pitarch JV, Rocha E. Overview of venous thromboembolism
70 (Suppl. 2):3–10, 2010.
7. Bandle J, Shackford SR, Sise CB, Knudson MM, Group CS. Variability is the standard: the management of venous thromboembolic disease following trauma. J Trauma Acute Care Surg
76 1:213–216, 2014.
8. Velmahos GC, Spaniolas K, Tabbara M, Abujudeh HH, de Moya M, Gervasini A, Alam HB. Pulmonary embolism
and deep venous thrombosis in trauma: are they related? Arch Surg
144 10:928–932, 2009.
9. Paffrath T, Wafaisade A, Lefering R, Simanski C, Bouillon B, Spanholtz T, Wutzler S, Maegele M, Trauma Registry of DGU. Venous thromboembolism
after severe trauma: incidence, risk factors and outcome. Injury
41 1:97–101, 2010.
10. Menaker J, Stein DM, Scalea TM. Pulmonary embolism
after injury: more common than we think? J Trauma
67 6:1244–1249, 2009.
11. Menaker J, Stein DM, Scalea TM. Incidence of early pulmonary embolism
after injury. J Trauma
63 3:620–624, 2007.
12. Benns M, Reilly P, Kim P. Early pulmonary embolism
after injury: a different clinical entity? Injury
45 1:241–244, 2014.
13. Van Gent JM, Zander AL, Olson EJ, Shackford SR, Dunne CE, Sise CB, Badiee J, Schechter MS, Sise MJ. Pulmonary embolism
without deep venous thrombosis: de novo or missed deep venous thrombosis? J Trauma Acute Care Surg
76 5:1270–1274, 2014.
14. Hoth JJ, Wells JD, Jones SE, Yoza BK, McCall CE. Complement mediates a primed inflammatory response after traumatic lung injury. J Trauma Acute Care Surg
76 3:601–608, 2014.
15. Suresh MV, Ramakrishnan SK, Thomas B, Machado-Aranda D, Bi Y, Talarico N, Anderson E, Yatrik SM, Raghavendran K. Activation of hypoxia-inducible factor-1 alpha in type 2 alveolar epithelial cell is a major driver of acute inflammation following lung contusion
. Crit Care Med
42 10:E642–E653, 2014.
16. Hoth JJ, Hudson WP, Brownlee NA, Yoza BK, Hiltbold EK, Meredith JW, McCall CE. Toll-like receptor 2 participates in the response to lung injury in a murine model of pulmonary contusion. Shock
28 4:447–452, 2007.
17. Hoth JJ, Wells JD, Brownlee NA, Hiltbold EM, Meredith JW, McCall CE, Yoza BK. Toll-like receptor 4-dependent responses to lung injury in a murine model of pulmonary contusion. Shock
31 4:376–381, 2009.
18. Hoth JJ, Wells JD, Hiltbold EM, McCall CE, Yoza BK. Mechanism of neutrophil recruitment to the lung after pulmonary contusion. Shock
35 6:604–609, 2011.
19. Knoferl MW, Liener UC, Seitz DH, Perl M, Bruckner UB, Kinzl L, Gebhard F. Cardiopulmonary, histological, and inflammatory alterations after lung contusion
in a novel mouse model of blunt chest trauma. Shock
19 6:519–525, 2003.
20. Brill A, Fuchs TA, Chauhan AK, Yang JJ, De Meyer SF, Kollnberger M, Wakefield TW, Lammle B, Massberg S, Wagner DD. von Willebrand factor-mediated platelet adhesion is critical for deep vein thrombosis in mouse models. Blood
117 4:1400–1407, 2011.
21. Cooley BC. In vivo fluorescence imaging of large-vessel thrombosis in mice
. Arterioscler Thromb Vasc Biol
31 6:1351–1356, 2011.
22. Vu TT, Zhou J, Leslie BA, Stafford AR, Fredenburgh JC, Ni R, Qiao S, Vaezzadeh N, Jahnen-Dechent W, Monia BP, et al. Arterial thrombosis is accelerated in mice
deficient in histidine-rich glycoprotein. Blood
125 17:2712–2719, 2015.
23. Mikawa K, Nishina K, Tamada M, Takao Y, Maekawa N, Obara H. Aminoguanidine attenuates endotoxin-induced acute lung injury in rabbits. Crit Care Med
26 5:905–911, 1998.
24. Sun C, Beard RS Jr, McLean DL, Rigor RR, Konia T, Wu MH, Yuan SY. ADAM15 deficiency attenuates pulmonary hyperpermeability and acute lung injury in lipopolysaccharide-treated mice
. Am J Physiol Lung Cell Mol Physiol
304 3:L135–L142, 2013.
25. Iversen NK, Malte H, Baatrup E, Wang T. The normal acid-base status of mice
. Respir Physiol Neurobiol
180 (2–3):252–257, 2012.
26. Hoth JJ, Stitzel JD, Gayzik FS, Brownlee NA, Miller PR, Yoza BK, McCall CE, Meredith JW, Payne RM. The pathogenesis of pulmonary contusion: an open chest model in the rat. J Trauma
61 1:32–44, 2006.
27. Wang ND, Stevens MH, Doty DB, Hammond EH. Blunt chest trauma: an experimental model for heart and lung contusion
. J Trauma
54 4:744–748, 2003.
28. Jaffin JH, McKinney L, Kinney RC, Cunningham JA, Moritz DM, Kraimer JM, Graeber GM, Moe JB, Salander JM, Harmon JW. A laboratory model for studying blast overpressure injury. J Trauma
27 4:349–356, 1987.
29. Wang S, Ruan Z, Zhang J, Zheng J. A modified rat model of isolated bilateral pulmonary contusion. Exp Ther Med
4 3:425–429, 2012.
30. Hoth JJ, Hudson WP, Brownlee NA, Yoza BK, Hiltbold EM, Meredith JW, McCall CE. Toll-like receptor 2 participates in the response to lung injury in a murine model of pulmonary contusion. Shock
28 4:447–452, 2007.
31. Suresh MV, Yu B, Machado-Aranda D, Bender MD, Ochoa-Frongia L, Helinski JD, Davidson BA, Knight PR, Hogaboam CM, Moore BB, et al. Role of macrophage chemoattractant protein-1 in acute inflammation after lung contusion
. Am J Respir Cell Mol Biol
46 6:797–806, 2012.
32. Niesler U, Palmer A, Radermacher P, Huber-Lang MS. Role of alveolar macrophages in the inflammatory response after trauma. Shock
42 1:3–10, 2014.
33. Diaz JA, Obi AT, Myers DD Jr, Wrobleski SK, Henke PK, Mackman N, Wakefield TW. Critical review of mouse models of venous thrombosis. Arterioscler Thromb Vasc Biol
32 3:556–562, 2012.
34. Yamashita A, Furukoji E, Marutsuka K, Hatakeyama K, Yamamoto H, Tamura S, Ikeda Y, Sumiyoshi A, Asada Y. Increased vascular wall thrombogenicity combined with reduced blood flow promotes occlusive thrombus formation in rabbit femoral artery
. Arterioscler Thromb Vasc Biol
24 12:2420–2424, 2004.
35. Geerts WH, Bergqvist D, Pineo GF, Heit JA, Samama CM, Lassen MR, Colwell CW, American College of Chest P. Prevention of venous thromboembolism
: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines (8th Edition). Chest
133 (6 Suppl.):381S–453S, 2008.
36. Barrera LM, Perel P, Ker K, Cirocchi R, Farinella E, Morales Uribe CH. Thromboprophylaxis for trauma patients. Cochrane Database Syst Rev