In vivo monitoring of venous thrombosis in mice


Mark D. Blostein, Lady Davis Institute for Medical Research, Sir Mortimer B. Davis-Jewish General Hospital, 3755 Côte-Ste-Catherine Road, Montreal, PQ H3T 1E2, Canada.
Tel.: +1 514 340 8214; fax: +1 514 340 8281.


Summary.  Background:  Venous thrombosis (VT) is an important cause of morbidity and mortality in clinical medicine. Animal models studying venous thrombosis are scarce and, in most cases, very crude and rely on sacrificing the animals to excise formed thrombi. Developing an in vivo murine model of venous thrombosis can be a powerful tool for studying venous thrombosis.

Objectives:  We sought to use a high-frequency ultrasound system (HFUS) to dynamically and non-invasively monitor thrombus formation in the inferior vena cava (IVC) of mice.

Methods:  We developed a murine model of venous thrombosis using, for detection, the Vevo 770®, a micro-imaging HFUS. Two different thrombosis models were used to generate thrombi in the IVC of C57Bl/6NCr mice: (i) ligation and (ii) application of ferric chloride (FeCl3). We then assessed venous thrombosis by HFUS.

Results:  In both models, measurements of the clot pathologically correlated favorably with measurements acquired with HFUS. Thrombus develops less than an hour after ligation or FeCl3-induced injury of the IVC and the size of the clot increases over time for up to 24 h. Importantly, we demonstrate that HFUS can be used to monitor the effect of an anticoagulant such as dalteparin until complete resolution of the thrombus.

Conclusions:  These data show that HFUS assesses venous thrombosis in mice reliably and non-invasively. Developing a murine model of thrombosis using more accurate, and clinically more relevant, techniques such as ultrasonography, is a step towards a better understanding of the pathophysiology of venous thromboembolism.


Venous thromboembolism (VTE) afflicts 117 people per 100 000 each year [1]. It is an important cause of morbidity and mortality [2] in clinical medicine. VTE comprises deep vein thrombosis (DVT) and pulmonary embolism, whose complications can lead to the post-thrombotic syndrome [3], pulmonary hypertension [4], and even death. VTE is a multicausal disease. Classically a triad of risk factors for this disease was described by Virchow over a hundred years ago; they include venous stasis, endothelial damage and hypercoagulability [5].

Given the importance of VTE in human disease, animal models are very useful to understand its pathophysiology and aid in the development of therapies. Mice are commonly used to study human disease because they recapitulate many analogous biologic functions found in humans, they reproduce and grow quickly, and they can be genetically manipulated. Murine models of venous thrombosis are scarce and mainly rely on surgically disturbing one of the axes of Virchow’s triad. In a recent review of murine models of thrombosis, only one of 50 studies reviewed examined the venous system [6]. For our study, we focused on two models that have been described to study venous thrombosis in mice. One model surgically decreases blood flow in the inferior vena cava (IVC) to cause stasis [7] and the second one uses ferric chloride (FeCl3) to chemically injure the endothelium and activate coagulation [8]. In previous studies, mice were sacrificed after induction of thrombosis to excise the clot and measure parameters such as clot weight or length to evaluate clot formation [7,9,10]. However, in these studies, venous thrombus formation was not monitored in the same mouse over time using ultrasound.

Developing a murine model that uses non-invasive methods for real-time imaging of thrombus progression would be useful in studying the disease in animal models. We used the Vevo770®, a micro-imaging high-frequency ultrasound system (HFUS) that has been developed to study murine arterial circulation [11–13]. The objective of this study is to demonstrate that HFUS can be used to dynamically and non-invasively monitor thrombus formation in the IVC of mice.


Animal models of venous thrombosis

All procedures were approved by the institutional Animal Care Committee of McGill University. C57Bl/6NCr male mice, between 8 and 12 weeks of age obtained from Charles River laboratories, were used in all experiments.

Venous thrombosis was induced in mice using two different models; the stasis model and the endothelial injury model. Mice were anesthetized by inhalation of 2% isoflurane gas. A midline laporotomy incision was made. The IVC was exposed and separated from the aorta at the level of the renal veins. A 6–0 silk suture was tied around the IVC below the renal veins (IVC ligation). Control animals underwent the same surgical procedures, except the IVC was not ligated (sham ligation). For the endothelial injury model, the IVC was exposed and the surface of the vein was cleared by blunt dissection between the renal and iliolumbar veins. Whatman filter paper (2 × 4 mm), saturated in 0.37 m (10%) FeCl3, was applied to the IVC for 3 min, and then removed. For both models, after inducing thrombosis, the abdomen was closed and HFUS was used to monitor thrombus formation. In order to compare HFUS measurements with those obtained pathologically, mice subjected to the above thrombosis-induction models were sacrificed and thrombi in the IVCs were dissected and measured with a ruler.

Use of anticoagulation

To demonstrate that HFUS can be used for monitoring the effects of antithrombotic agents on thrombus formation in vivo, mice were systemically administered dalteparin (200 IU kg−1), via a tail injection prior to the induction of thrombosis using the above models. In another set of experiments, the presence of a thrombus at 1 h was verified using HFUS, after which dalteparin was injected subcutaneously and clot resolution was monitored over time.

HFUS methodology

The Vevo770® micro imaging system (Visualsonics, Toronto, Canada), used to visualize the IVC and for all acquisitions, consists of a single element probe of center frequency of 40-MHz. The transducer has an active face of 3 mm, a lateral resolution of 68.2 μm and an axial resolution of 38.5 μm [14,15]. During all experiments, mice were sedated using 2% isoflurane gas. Heart rate of the animal was monitored and kept at 500 beats per mins. Temperature was monitored using a rectal probe and regulated with a heating pad. The abdomen of the mouse was shaved and warm ultrasound transmission gel was applied to enable visualization and optimize image quality. First, a long axis view was used to visualize the IVC, the ligation or the injury site, and the formed thrombus. An optimal freeze-frame image was taken manually and, using the Vevo770® software, the cross-sectional area of the clot was traced to obtain the measurement. The length and width of clots were measured using the software as well. The software has different modes. The pulse-wave Doppler mode senses the movement of blood cells (Doppler effect) and is used to measure the direction, intensity and speed of blood cells. The color Doppler mode quantifies blood flow velocity by color encoding the flow.

Statistical analysis

Data are presented as mean ± SEM. Within-group differences were assessed with one-way anova. Post hoc comparisons were tested with the Newman–Keuls test. A value of P < 0.05 was considered statistically significant.


Visualizing venous thrombosis by ultrasonography

HFUS uses high-frequency sound waves to measure differences in matter density to visualize the body’s internal organs (ultrasonography) [15]. In the absence of ligation in the sham group, we could visualize the vascular wall of the vein, because it is denser than the surrounding fat tissue. In the experimental group of mice, thrombosis was induced by ligation of the IVC. Because in veins thrombi are mainly composed of fibrin, which is denser than flowing blood, we could appreciate the formed thrombus inside the IVC using ultrasonography. HFUS can also visualize the aorta and the IVC in both transverse and axial views (Fig. 1A).

Figure 1.

 Monitoring of inferior vena cava (IVC) thrombosis with high-frequency ultrasound system (HFUS). (A) Typical images of IVC in the axial and transversal views in the sham group (left panel) and 5 h after ligation (right panel). The transversal view shows the aorta and the IVC adjacent to each other in the sham and ligated groups. A thrombus is visualized in both the axial and transversal views. (B) HFUS also allows quantification of blood flow velocity. Using pulse-wave Doppler function of HFUS, no flow was observed in the ligated group. Velocity of blood flow can also be depicted by color Doppler using color processing. The scale of color-coded blood flow ranges from red or high flow to yellow or low flow. In the ligated animals the absence of flow is depicted by the absence of color.

HFUS allows for quantification of the velocity of blood flow in the vessels using either pulse-wave Doppler or color Doppler. In the sham group, normal blood flow is shown in Fig. 1(B). IVC ligation reduces blood flow (Fig. 1B). Velocity of blood flow can also be depicted by color Doppler using color processing. The ultrasonography instrument is assigning color values that depend on whether blood is moving away from or towards the transducer. In addition to showing the direction of flow, the colors vary in intensity depending on the flow velocity, red being high flow velocity and yellow or no color depicting low or no flow, respectively. Thus, in sham animals low flow velocity is depicted in orange-yellow and no color depicted absence of flow in a ligated IVC (Fig. 1B).

Correlation between thrombus parameters obtained by HFUS and histopathology

In order to demonstrate that HFUS can be used to precisely measure thrombus formation, we correlated thrombus length values given by the ultrasonography instrument with those obtained histopathologically. Using the stasis model, we measured thrombus formation at 5 and 24 h after ligation of the IVC. We showed a correlation between the two measurements at 5 h (R2 = 0.9785) and 24 h (R2 = 0.9116) (Fig. 2A,B). We obtained similar results using the endothelial injury model. Seventy-five minutes after the application of FeCl3 to the outer vessel layer, the clot length obtained histopathologically correlated favorably with HFUS (R2 = 0.962) (Fig. 2C). These data demonstrate that measurement of venous thrombosis by HFUS correlates favorably with thrombi obtained pathologically.

Figure 2.

 Positive correlation between high-frequency ultrasound system (HFUS) and histopathology measurements. (A) In the stasis model, correlation of clot length at 5 h, n = 10, R2 = 0.96. (B) In the stasis model, correlation of clot length at 24 h, n = 20, R2 = 0.98. (C) Correlation of clot length in the FeCl3 injury model 75 min after injury, n = 6, R2 = 0.962. In both models, HFUS and histopathology measurements are positively correlated.

Clot size progression over time

In the context of venous thrombosis formation, one of the principal advantages of HFUS is the ability to monitor thrombus progression over time. Thus, using the ligation model we measured the size of thrombi formed after 5 and 24 h. Clot length and clot surface area were measured with HFUS. We found that the thrombus length and surface area increased significantly at 24 h compared with 5 h after ligation (Fig. 3).

Figure 3.

 Measurement of thrombus size at 5 and 24 h. Thrombus length and cross-sectional area are significantly increased at 24 h compared with 5 h after ligation in the ligation model of thrombosis. Data are presented as means ± SEM. n = 20, *P < 0.0001 for clot length and *P < 0.00005 for cross-sectional area.

As ultrasonography allows us to follow thrombus development in the same animal, we sought to monitor thrombus cross-sectional area over time. Thus, we ligated the IVC of six mice, and monitored these mice continuously after surgery using HFUS for up to 4 h after ligation. We found that mice developed a visible thrombus less than an hour after ligation and that thrombus size reached a plateau 2½ h after ligation (Fig. 4A). We repeated this experiment using the FeCl3 injury model. Thrombus develops more quickly after endothelial injury compared with the ligation model. The size of clots started increasing at 5 min after injury to reach a plateau at 30 min (Fig. 4B). However, because FeCl3 is very toxic for the animal, the majority of the mice died after 90 min of monitoring.

Figure 4.

 Continuous monitoring of thrombus formation by high-frequency ultrasound system (HFUS). HFUS was used to continuously monitor clot formation over time in the ligation model (A) and in the endothelial cell injury model (B). Clot size increased progressively over time to reach maximum size at 2.5 h in the ligation model (A) or 30 min in the injury model (B). Data are presented as means ± SEM. n = 6, *P < 0.05 vs. time 0.


Low-molecular-weight heparins are effective in treating venous thromboembolism in humans. As a proof of principle for our model, we injected a therapeutic dose of dalteparin (200 units kg−1) into mice immediately after IVC ligation. We then verified clot formation 5 h (data not shown) and 24 h after ligation (Fig. 5A). As expected, the transversal view obtained with HFUS shows an absence of clot formation, at both time-points, in mice that undergo IVC ligation treated with dalteparin as compared with control mice that were given only phosphate-buffered saline (PBS) (Fig. 5A). To dynamically follow the effect of dalteparin, we first induced thrombosis using the ligation model and then administered the anticoagulant after the formation of a clot. Thus, 1 h after IVC ligation, the mice were monitored using HFUS, and the ones with a prevalent thrombus were then randomly chosen to be injected with PBS or dalteparin (200 IU kg−1). Each group of mice was continuously monitored for up to 3 h after injection. Mice were then allowed to regain consciousness, returned to the animal facility, and were monitored again at 24 and 48 h after ligation. We found that about 1½ h after injection, the size of thombi from mice injected with dalteparin started decreasing, to completely resolve at 24 h. In the PBS injected mice, size of thrombi continued to increase until 48 h (Fig. 5B).

Figure 5.

 Continuous monitoring of the anticoagulant effect on thrombus resolution by high-frequency ultrasound system. (A) Mice were administered phosphate-buffered saline (PBS) or 200 Units kg−1 of dalteparin before ligation in the stasis model. No thrombus developed in animals treated with dalteparin. (B) PBS or dalteparin were injected 1.5 h after induction of thrombosis using the ligation model. Resolution of the thrombus was monitored from the injection (1.5 h) to 48 h. Dalteparin significantly reduced thrombus size from 3.5 h to complete disappearance at 24 h. n = 5 in each group, *P < 0.05 vs. 1.5 h PBS and #P < 0.05 vs. 1.5 h dalteparin. §P < 0.05 PBS vs. dalteparin.


In this study, we demonstrate that HFUS reliably monitors venous thrombosis in mice. Several models have been described for studying venous thrombosis in animals. The models are reviewed by Day et al. [9]. They include photochemical endothelial injury [16], IVC stasis [7,17], and chemical or mechanical trauma to the endothelium [8,18]. In all these models, the animals are sacrificed to study the characteristics of thrombi, such as their progression over time or the effect of anticoagulants. The characterization of thrombi in these models mostly relies on crude or imprecise measurements such as clot weight or length. These techniques are insufficient given that thrombus progression cannot be monitored in the same mouse over time. Establishing a non-invasive, in vivo model for monitoring thrombus formation, progression and regression in the venous system would be a powerful tool.

Various non-invasive imaging technologies are available to study physiological and disease status in animal models [19–21]. Some of these techniques include ultrasound, magnetic resonance imaging, positron emission tomography and X-ray computed tomography scans [22]. Among these techniques, ultrasound is the most versatile, portable, cost efficient and available for small animal research [15]. Furthermore, ultrasonography is used clinically to diagnose venous thromboembolism in humans [23–25]. Therefore, we developed a model using ultrasonography to study venous thrombosis in mice. Vevo770® is an HFUS, designed specifically to examine the vasculature of small animals in vivo. This system operates at high frequency (40 MHz) and high resolution (30 μm), and therefore can be used to visualize and measure the vasculature of mice. For example, HFUS has been shown to precisely measure and monitor atherosclerotic events. The ultrasound measurements in these studies show a positive correlation with histologic data [26,27].

Classically, according to Virchow’s triad, venous thrombosis results from three different pathophysiologic entities: stasis, hypercoagulability of the blood and endothelial dysfunction. In this study, we used two different methods to induce venous thrombosis that take advantage of two components of Virchow’s triad, stasis and endothelial injury. We observed with HFUS that thrombi formed using these models have different characteristics [7,28,29]. The ligation model generated small thrombi whose size increased progressively over time. After endothelial cell injury, thrombi were larger and their size increased more quickly than for the stasis model. However, because ferric chloride (FeCl3) is a toxic agent, animals do not survive long after exposure. This is consistent with other studies using this model [28]. In contrast, animals survived for many days after the surgery in the stasis model [7,30]. Even though HFUS alone does not provide insights about the underlying mechanisms of thrombus formation, it allows us to appreciate the kinetics of clot formation in the endothelial cell injury versus the stasis model in mice.

Our study demonstrated that HFUS presented several advantages for detecting and monitoring thrombus development in the small animal. First, this method is non-invasive and non-terminal. Therefore, mice can recover and thrombus formation can be monitored in the same animal over time, which may be more physiologically relevant than scarifying animals at different time-points to evaluate thrombus size. Hence, this method would decrease potential biologic variability. Second, fewer animals need to be used. Finally, precise measurements, surface area, length and blood velocity, can be obtained for the same animal at the same time-point. Because ultrasonography can be used to follow thrombus size over time, it appears to be useful to study the effects of antithrombotic or antifibrinolytic agents. We took advantage of this property to demonstrate that the administration of dalteparin immediately after the ligation of the IVC prevents thrombus formation. More importantly, we followed clot resolution dynamically for up to 48 h when dalteparin was injected after establishment of thrombosis.

There are limitations to this model. For instance, HFUS does not give any insight about clot composition, such as fibrin content [15]. HFUS is a non-invasive detection method but, to induce thrombosis, we still have to rely on surgical techniques, which are not entirely physiologically relevant. However, because, unlike in humans, mice do not spontaneously develop thrombi [9], surgical induction of venous thrombosis is the only described methodology to study venous thrombosis in mice [31].

In conclusion, HFUS is a reliable, non-invasive and relatively simple technique for monitoring venous thrombosis in mice. Given our knowledge and ability to manipulate the murine genome, a better understanding of the pathophysiology of venous thrombosis in humans may result from using this model.


We are grateful to V. Micheau for excellent technical assistance. This work was supported by funding from the Canadian Institutes of Health Research (CIHR).

Disclosure of Conflict of Interests

The authors state that they have no conflict of interest.