Introduction

Venous thromboembolism remains one of the main causes of death worldwide, occurring with the rate of 1—2 persons per 1,000 of population annually, reaching 300,000 to 600,000 cases in the USA and up to 1,000,000 total cases in the USA and Europe [1, 2]. In the United States alone, up to 60 thousand people die every year as a result of verified pulmonary embolism (PE) [3]. Despite a large number of clinical and experimental studies, the pathogenesis of PE has not been understood completely, which leaves many questions unanswered, including the role of floating venous thrombi and the risk of embolism of the pulmonary artery (PA) branches with these thrombi, as well as the possibility for development of thrombotic events in the PA with no evident primary source.

Morphological studies indicate significant structural differences between the thrombi in arterial and venous vessels, based on the specific features of the pathogenesis of thrombus formation [3]. At the same time, comparison of the histological structure of thromboemboli from the PA and venous thrombi demonstrates differences in the ratio of their structural components, which may indicate additional non-obvious mechanisms inherent in the pathogenesis excluding embolism itself [3, 4]. The results of experimental studies demonstrate a variety in the clinical presentation, inflammatory response and morphological structure of a thrombus in PE, depending on the conditions of its induction. The scattered data of the available experimental studies do not allow to summarize the results obtained and to formulate unambiguous conclusions. However, their comparison with each other can shed some light on the specific features of the pathogenesis of PE. This literature review answers the questions on the grade of impact of experimental PE conditions on its course and outcome.

Clinical and morphological studies

One single study compared the ultrastructure of endovascularly extracted coronary artery thrombi from the patient with an acute coronary syndrome, a floating head of a venous thrombi taken by deep vein open thrombectomy, and a thromboembolus removed from the PA trunk and its branches postmortem [3]. Fibrin and erythrocytes (RBC) appeared to be predominant components of venous thrombus (VT) and thromboembolus (TE), thus distinguishing them from the arterial thrombus (AT), a significant proportion of which was represented by platelets (PLT) (Table 1) [3]. RBC in the thromboembolus predominantly had the polyhedral forms (polyhedrocytes) as a result of clot retraction, comprising on average 81% of all RBC. VT was inferior to TE in the number of polyhedrocytes, as well as in the amount of fibrin presented in the form of single fibers. TE was also superior to VT in the number of leukocytes (LEU) and in the number of microparticles (MP). In its turn, VT was characterized by a strong positive correlation between the PLT and MP counts, as well as by a strict negative correlation between MP and total fibrin [3]. No correlation of this kind was registered for TE.

Table 1. Content of different structural components in venous thrombus, thromboembolus and arterial thrombus

Thromboemboli may distinguish in their structure even in the conditions seeming similar in their formation and location. It was showed in one of the histological studies of autopsy material that fibrin in the TE from the PA trunk can be distributed evenly, with LEU “immured” in it, or have a zonal location along the peripheral parts of the thrombus with LEU scattered along the borderline between fibrin and RBC accumulated in the core of the thrombus [4]. It seems difficult to draw unambiguous conclusions from the data presented for the heterogeneity of the morphological material. The structural differences between the floating head of a thrombus and an embolus in the PA suggest the possibility of primary formation of a thrombus with different morphological characteristics within the PA, or of the embolus secondary modification under the influence of PA vascular wall specific properties.

Studying human blood clots is a technical and ethical challenge, therefore most of the data on the pathogenesis and morphology of PE have been obtained from experimental studies on animals. A number of PE models have been suggested for this purpose.

Systemic administration of platelet agonists

Systemic administration of PLT agonists causes massive PA thrombosis within a short period of time. The agents with such effect are epinephrine with collagen, adenosine diphosphate (ADP), thromboplastin, and thrombin [5—8]. The models are mostly used in acute experiments with high animal mortality rates. After the injection of epinephrine with collagen into the mice tail or jugular vein, mortality can reach 90% based on a dose-dependent effect [5—7]. Tracking radiolabeled PLT demonstrates their rapid accumulation in the PA system after the administration of their agonists, with the following decrease to the baseline levels [9]. An extremely rapid increase in the concentration of labelled PLT is registered with administration of ADP with a return to normal levels 20—30 seconds later. With administration of thrombin a decrease was registered 2 minutes later. Collagen injection results in reaching the peak labelled PLT concentration in 1 minute, which drops to the baseline values within 20 minutes [9]. Thrombi rich in PLT and fibrin are discovered in the pulmonary blood vessels larger than 50 μm in diameter 3 minutes after the injection of epinephrine and collagen combination [5]. Immunohistochemical test reveals CD-41-positive PLT in aggregates with LEU in the PA branches after the administration of collagen and thrombin [9]. Lysis of the pulmonary thrombi induced by the administration of collagen with epinephrine is significantly reduced compared to exogenous thrombus injection [10].

Intravenous injections of thrombin or thromboplastin to animals cause activation of PLT with coagulation cascade and are accompanied by vasoconstriction [8]. The injection of thrombin and thromboplastin into the inferior vena cava, femoral veins or retro-orbital venous plexus of mice and rats potentiates PLT aggregation in the PA [9, 11—13]. As a result of the injection, more than 70% of mice die 7.5 minutes later due to occlusive thrombosis of 72% of large and small PA vessels of more than 40 μm in diameter [11]. An important mechanism of thrombus formation in the thrombin injection model is the activation of the PLT PAR receptors [11]. 85% survival rate is registered in PAR3- (Par3-/-) and PAR4- (Par4-/-) deficient mice after administration of the same doses of thromboplastin, with preserved PA patency [11].

PE with a formed exogenous thrombus

Two methods were used to provoke PE with exogenous thrombus. The first one is ex vivo thrombus preparation with its subsequent intravenous injection, and the second one is thrombus induction in the venous system followed by mechanical stimulation of thrombus movement to provide thromboembolism. Exogenous thrombi can be prepared from blood or plasma by mixing them with thrombin and other components [14]. Both models have been created for small laboratory animals, mice and rats, as well as for large ones, including dogs and pigs [15—17].

A specific feature of the experimental PE induced by exogenous thrombus injection is the difficulty of maintaining PA thrombotic obstruction due to intense thrombolysis [10]. The PA endothelium has a high fibrinolytic ability, which decreases when exposed to the negative factors, including hypoxia and inflammation [18, 19]. In an experiment with the injection of thrombi prepared in vitro or induced in the venous system, the intact PA endothelium that has not been previously exposed to the negative agents retains the ability of effective thrombolysis.

In one of murine PE model exogenous thrombi were prepared by mixing human plasma, 125I-labeled fibrinogen and thrombin, and injected them into the jugular vein [20]. The animals were euthanized 4 hours later. tPA (tissue plasminogen activator) expression was identified in the endothelium of embolized PA branches. tPA was also identified in the endothelium of the non-embolized PA branches and in the bronchial arteries of the same mice. In addition to tPA, positive staining for uPA (urokinase type plasminogen activator) was registered in the PA endothelium, as well as in the extravascular pulmonary tissue and in the embolus. Neutrophils (NEU) were discovered inside a thrombus and along its peripheral parts, and in the absence of an embolus — in the parenchyma and, quite rarely, in the non-thrombotic blood vessels. uPA expression was also determined in the NEU. Plasminogen activity in the embolized vessel was 10-fold higher as compared to the intact one.

An experiment with the injection of exogenous thrombi prepared ex vivo showed dependence of their distribution in the PA system upon the size of the thrombotic particles [10]. Large ¹²⁵I-labeled emboli measuring 5 mm in diameter and microemboli in the form of a homogeneous suspension were injected intravenously to mice and rats [10]. The injection of large emboli resulted in death of 40% of the animals within 5 minutes. Along with that, about 60% of the radioactive isotope was found in the heart and in the PA trunk. In survived rodents 90% of ¹²⁵I were detected in the PA system and the emboli were distributed randomly with a tendency towards localization of 70—90% in one lobe. 10 minutes after the injection of the microemboli 50—60% of the isotope were localized in the lungs. About 5% were found in other areas, including blood. Microthrombi were distributed evenly in the lobes of both lungs. 70% of the pulmonary emboli in mice and 50% of the pulmonary emboli in rats were lysed within the first hour after embolism. Nearly complete thrombolysis was observed in both rodent species in 5 hours with no increase in the plasma concentration of radioactive iodine, which indicated its rapid elimination.

Exogenous thrombi injection with fibrinolytic agent (tranexamic acid) infusion pretreatment in Sprague-Dawley and Copenhagen rats was accompanied by near complete lysis of the TE in a short time with a minor difference in the process intensity between the rodent lines [21]. In Sprague-Dawley rats, 95±1.0% of the TE were lysed at 24 hours, and 97±0.8% — at 5 days. In Copenhagen rats the process appeared to be less active and ranged within 69.8±7.4% and 87.3±7.6% at the same time points. Bronchoalveolar lavage tests revealed no signs of inflammation, no increases in myeloperoxidase activity and LEU presence, and no increase in the concentrations of chemokines (MCP-1, CINC-1, CINC-2, CINC-3) as compared to the control rats.

The inflammatory response to exogenous clots PE induction was studied in rabbits. The increased content of the inflammatory biomarkers (TNF-α, IL-8, CXCL5) and encoding — mRNAs were registered in the pulmonary endothelium [22]. Tissue factor expression of the PA endothelium was affected by exogenous thrombi injection [23]. No changes in its expression were discovered in the embolized PA segment in 3, 8 and 24 hours. However, a significant decrease in the TF expression occurred in 3 and 8 hours in the segments distal to the embolized ones. In 24 hours this parameter reached the reference values corresponding to those registered in the control animals.

PE provoked by mechanical stimulation of venous thrombus migration

In a rat model, venous thrombosis was induced by left femoral vein blocking by microvessel clip [24]. The largest and most stable thrombi were formed by the end of the 1^st day. On Days 1, 4, or 7 (D1, D4, D7) thrombus was aspirated from the left femoral vein and injected into the right femoral vein. The grade of thrombotic obstruction of the pulmonary arteries was assessed on days 1, 4, and 7 (P1, P4, P7) after the injection. The rate of its development turned out to be the lowest with the infusion of 7-day thrombi: D7 — 44%, D4 — 83%, D1 — 100%. Residual PA stenosis as a result of thrombolysis was lower in D1 subgroup: D1 — 39%, D4 — 73%, D7 — 100%. Thus, embolism with fresh thrombi was associated with the maximum incidence of PE and the most complete subsequent recanalization. Swollen PA endothelium, enlargement of the nucleus and increased number of pinosomes in the endotheliocytes were representative characteristics on the next day after D1 thrombus injection; fibroblasts hyperplasia and elastic fibers thickening were revealed on days 4 and 7 after D1 injection. Swollen endothelial mitochondria and partial cytoplasm dissolution, elastic fibers necrosis and disappearance occurred on the 1^st day after D7 injection. Endotheliocytes dissociation, media thickening, lymphocytic infiltration, and fibroblasts hyperplasia were registered on the 4^th and 7^th days. The results demonstrate dependence of the PA wall reaction to the thrombus maturity. The structure of the thrombi themselves has not been described in the study.

In the Sprague-Dawley rats model with inferior vena cava (IVC) thrombosis induced by its clipping with tributaries ligation, PE was stimulated 48 hours later by IVC massage, provoking thrombus proximal migration [25]. Prominent thrombolysis appeared to be characteristic for this model as well: PA obstruction resolution was 40% on the 2 day, 90% on the 4 day and 100% on the 6 day. NEU were identified along the peripheral parts of the thrombus. The total LEU count accompanying thrombotic obstruction of the PA was 38 times higher than in the control sham-operated rats in 3 hours, and increased 320-fold 2 days after PE compared to controls. NEU predominated over monocytes in the cell populations. LEU concentration decreased along with the lysis of the embolus. Histological examination on the 4 day revealed intima thickening and an increase in its cellularity as well as smooth muscle cells proliferation. The rise of intima/media ratio persisted from day 4 to day 14. KC/GRO (keratinocyte-derived chemokine) and IL-10 proinflammatory biomarkers levels grew in the walls of the affected PA branches and vessels of the contralateral lung as compared to the baseline in response to embolism. MCP-1 increased in the embolized and contralateral PA branches. It was important that levels of P- and E-selectins in the PA wall remained unchanged.

Advanced thrombolysis in PE models is characteristic not only for rodents. In a dog model infusion of tranexamic acid prior to embolism provoked by a mechanical stimulation of IVC thrombus migration was used to suppress thrombolysis [26, 27]. The model was based on thrombin injection into IVC segment restricted between two double-balloon catheters. Thrombi in the phase of organization were discovered in the branches of the PA 8 days later. However, discontinuation of tranexamic acid in 10 days to dogs was followed by a partial or complete recanalization of the PA branches by the 30^th day [27].

Spontaneous PE based on IVC stenosis model

Partial ligation of the IVC in mice results in embolic events, which 21 days later can be detected as presence of fibrinogen- and CD-41-positive emboli in the PA branches [28]. The IVC stenosis model with 80% of the vein lumen narrowing is characterized by thrombus formation within 6—12 hours, and within 24—48 hours occlusive thrombus can be revealed in 60% of mice [29]. Rolling and adhesion of LEU occurs 1 hour after IVC ligation, and after 6 hours these immune cells cover the endothelium completely. The population is dominated by NEU (up to 70%), while monocytes account for up to 30% of cells. A large number of neutrophil extracellular traps (NETs), significant participants in thrombus formation, appear in the thrombus 3 hours later, and an active fibrin formation occurs within 6 hours. PLT are discovered in the thrombus 2 hours later, and 6 hours later they can be observed adhered to the endothelial surface or to the LEU. PLT are recruited in a thrombus as individual cells or in small aggregates, which is in contrast with the arterial thrombus and venous thrombi induced by other methods. Inhibition of PLT and NEU activity as well as DNase addition for NETs destruction suppresses thrombus formation in the model of IVC stenosis.

Spontaneous PE based on photochemical venous thrombosis model

The histological structure of a thrombus in the jugular vein induced by exposure to laser irradiation (1.5 W, 540 nm) followed by Rose Bengal (photoreactive substance) infusion was studied in mice [30, 31]. Free radicals released as a result of a photochemical reaction constitute the main mechanism of endothelial injury resulting in PLT-rich thrombus development with its structure similar to that of an arterial blood clot [32, 33]. Small diameter PA branches are involved in thrombotic process in such model. However, the structure of the emboli has not been described [31].

Spontaneous PE based on venous thrombosis model induced by ligation combined with exposure to irradiation

An exclusion of the photoreactive substance from the previous model with the vein ligation instead, changes the structure of the forming thrombus [33]. A model of femoral vein ligation proximally to the confluence with the saphenous vein accompanied by exposure to fluorescence microscopy light waves (475/35 nm, 60 s) was proposed to study PE in vivo [33]. RBC appeared to be among the main components of thrombus structure in this model. PLT and NEU played a less prominent role. Thrombus formation started within 15 seconds from the beginning of exposure. PLT were identified in the thrombus 30 seconds later without a fixation to the wall. The thrombus layers rich in PLT were discovered between the layers rich in RBC. Neither suppression of the PLT and NEU activity nor administration of DNase and free radical scavengers affected the thrombus formation process. The entire thrombus disappeared within 30 minutes after removal of the ligature as a result of its gradual transformation into a large number of emboli. Thrombotic masses rich in fibrin and inflammatory cells were discovered in the PA branches.

Spontaneous PE based on electrolytic venous thrombosis model

A different process of embolism and a different histological structure of the venous thrombus were discovered with the use of the electrolytic model [34]. The vein wall was exposed to a direct current of 1.5 V for 30 seconds with immediate thrombus formation and reaching the largest clot size by the 10^th minute. The basic elements of such thrombus are the PLT and fibrin. PLT tended to accumulate near the injured wall of the blood vessel in a homogeneous mass. Fibrin was deposited in the form of a shell around the perimeter of the lesion, growing and forming a borderline, stabilizing the thrombus and preventing embolism. Venous thrombi were characterized by embolism with microemboli coming from the apex of the thrombus, while the main body of the thrombus remained fixed to the vein wall. Intravital fluoromicroscopy with staining for glycoprotein αIIb/ß3 reveals small PLT aggregates sliding along the surface of the thrombus and turning into microemboli at its tail. The difference with the electrolysis-induced arterial thrombus consisted in the ability of near-total embolization with vessel emptying and subsequent growth of a new thrombus in its place. The rate of PA Involvement was not assessed in the study.

Spontaneous PE based on Ferric Chloride venous thrombosis model

Histological and quantitative assessment of embolic processes in the model with application of 4% FeCl₃ solution was carried out in a study with mice [35, 36]. Similarly to the electrolytic model, chemical induction resulted in rapid thrombus formation within 12 minutes [34]. Video microscopy 2 hours after the induction showed that the thrombus is the least stable at the beginning of the formation [35]. The size of the thrombus did not decrease during the entire monitoring period despite the microemboli detaching from the thrombus [35]. The basic elements of the thrombus are the PLT and fibrin [35]. The emboli were represented by fragments varying in the form, size and formation regularity. The number of general embolic events and massive embolisms was 40.2±3.0 and 4.8±1.1 minute^–1 [36]. The size of the thrombus did not correlate with the number of embolic events but there was a correlation between the incidence of emboli formation in a vein and the number of thrombi discovered in the PA, which consisted mostly of fibrin [35].

When analyzing the models of venous thrombosis as a source of experimental PE, one should take into account the fact that the method of thrombus induction affects the reactivity of the venous wall [37]. Significant differences were shown by direct comparison of the models [37]. An increase in the expression of the uPA gene is characteristic of the experiment with complete ligation of the IVC but absent in the models with temporary clipping of the IVC, application of 10% FeCl₃ solution and implantation of a silicone tube into the IVC. The same pattern was registered for TNF-α and TGF-ß. MCP-1 increases in the stenosis model and in the silicone insert model but does not change in other models. Changes are also discovered for MMP-2 that is identified in the active state only in the models of IVC ligation and FeCl₃ application. The models of stasis and chemical induction with FeCl3 differ from the others in high-grade cell proliferation.

Thus, differences in the local response of the venous wall to thrombus induction may have a potential effect on the clot composition and systemic inflammatory response, resulting in unequal changes in the pulmonary artery system, which should be taken into account when interpreting the results of experimental studies.

PE induced by artificial microspheres

The use of artificial microspheres makes it possible to assess the specific features of the reaction of the pulmonary vascular system in response to embolization of the PA branches by emboli varying in diameter [38]. This model can be used for acute and chronic experiments. It has been described for rats, rabbits, pigs and dogs [38—41]. Reactive changes in response to injections of microspheres measuring 45 μm in diameter at different dosages appeared unequal in the rabbit model [38]. The highest mortality and the most advanced systemic changes occurred at a dosage of 32 μl (microspheres)/ml (blood) with the reduction of the thrombus formation time based on the results of thromboelastography, decrease in the fibrinogen levels and prothrombin time 5 minutes after the injection. The injection at a dosage of 16 μl/ml also caused the death of a large proportion of rabbits but was accompanied by a decrease in fibrinogen levels only. Other parameters appeared unchanged. No changes in the parameters listed above were registered with the low dosages of microspheres (0.3 and 6 μl/ml). The levels of fibrin degradation products and thrombin-antithrombin complexes did not change in any of the 4 groups. No fibrin was formed around the microspheres.

Provocation of moderate and severe PE by injection of polystyrene microspheres into the jugular vein in rats allowed to update the data on the inflammatory response to this kind of impact [42] (4. A 6-fold increase of the NEU counts in the bronchoalveolar lavage (BAL) fluid was observed in case of severe embolism at 18 hours with a more than 100-fold increase in their chemotactic activity. No increase of a similar grade in the NEU activity occurred in the group of rats with a moderately severe course of PE. Among chemokines, significant increases have been registered for CINC, MIP-2, MIP-1α, MIP-1ß, MCP-1, CINC2, IP-10. A 50-fold increase in protein was also registered in BAL fluid with severe PE and was not registered in the second group of rats. These changes were not the result of the effect of the microspheres themselves on the endothelium as evidenced by their joint cultivation.

PE induced by stem cells injection

The model that uses stem cells for PE induction stands somewhat apart [43]. The model was created to study the possible causes of high mortality from PE in hematological patients. The use of stem cells has broad prospects in hematology but is associated with the risk of embolic complications. Murine mesenchymal stem cells (MSCs) from adipose tissue were injected into the tail vein. Mortality rate 24 hours later reached 85%. Histological examination revealed fibrin thrombi in the PA and in the right ventricle. The cells were inside these thrombi. No thrombi were discovered in other organs. Evaluation of the available MSC lines used in practice showed that 97% of cells carry TF on their surface. MSC just isolated from the murine adipose tissue without subsequent cultivation rarely have TF on their surface, which determines their low procoagulant activity. Injection of this kind of MSC into the murine vein is not accompanied by embolism and high mortality.

Studies comparing PE models

Comparison of embolism by a thrombus from the IVC induced by its clipping with ligation of tributaries (TE group) to the model with injection of silicone microspheres (SM group) made it possible to identify differences in the cytokine expression [39]. The SM group was characterized by a significant increase in the levels of the profibrotic cytokine IL-13, both in the affected and contralateral lung, 1 day after PE induction; a higher MCP-1 value was registered on the 4^th day; an increased IL-13 level was recorded repeatedly on the 14^th day, while TGFβ and MCP-1 level did not differ across the groups. Histological examination revealed intimal hyperplasia, which was more pronounced in the SM group: 3-fold exceeding on day 4 and 4-fold exceeding on day 21 as compared to the TE group. Thus, embolism with silicone microspheres caused a more advanced proliferative reaction and sclerosis in the pulmonary artery system.

Comparison of the murine model of the IVC stenosis with reduction of its lumen by up to 88% without ligation of its tributaries to the murine model of retro-orbital thrombin administration also demonstrates significant differences in the response of the pulmonary vascular system [44]. Thrombi were discovered in the small PA branches with no systemic hemodynamic changes 24 hours after blood flow restriction. After thrombin injection, thrombi were formed in large PA branches, which was accompanied by dilation of the right ventricle and a decrease in the PA blood flow. A decrease in the sensitivity of the PA wall to acetylcholine and an increase in the expression of the source of superoxides, gp91^phoxNADPH oxidase, were registered in this group of mice. The increased expression of PAI-1 mRNA was discovered in the PA homogenates. None of the changes described above were found in the IVC stenosis group.

Conclusions

Results of the basic research studies demonstrate heterogeneity in the pathogenetic mechanisms of PE development. It can vary in the specific features of its course and in the prognosis depending on the conditions and causes of its development (Table 2). Intact pulmonary endothelium has obviously an extensive thrombolytic potential, allowing rapid and effective dissolution of migrated clots. At the same time, negative systemic effects may result in the loss or perversion of these properties. The possible variants are a decrease in fibrinolytic activity, the formation of a pro-inflammatory phenotype and the formation of superoxides. These deviations of the endothelial function can be determined by the conditions of thrombosis induction, maturity, size and number of clots, as well as frequency of their migration with embolic events and the global status of blood coagulation, fibrinolysis and inflammation systems. The inflammatory response is an integral component of initiation of thrombus formation and the local response of the pulmonary vascular system to embolism, being a mechanism for clot resorption. The sources of PE can be not only completely formed clots that migrate from the thrombosed vein with vessel emptying but also separate fragments of a thrombus containing cellular elements and fibrin, including activated platelets. The latter play an important role in the development of primary PA thrombosis and PE. PLT activation can occur both immediately in the pulmonary vascular system or in the primary venous thrombus with the subsequent migration and formation of distant secondary foci of thrombus formation in PA. These data can be an explanation for the insufficient efficacy of mechanical PE prevention methods, including cava filter implantation [45—47]. Microaggregates of activated PLT can easily pass through the filter elements and activate secondary thrombus induction immediately in the branches of the PA. Local pulmonary thrombosis cannot be ruled out as well in the setting of altered systemic inflammatory and coagulation reactions. It should be taken into consideration that the structure of the primary venous thrombus determines the mechanism for PE development. Thus, the migration of the entire mass with emptying of the vessel is more characteristic of clots rich in fibrin and RBC, while embolism with microfragments and platelet aggregates is more characteristic of platelet-fibrin thrombi. It is important that any of these variants can be induced in the venous system. Translation of experimental studies allows us to explain the inconsistency of data related to efficacy of mechanical prevention of recurrent PE that have been obtained in a number of clinical studies, which could have enrolled patients with different pathogenetic variants of PE but similar clinical presentation [46—49].

Table 2. PE models

Note. NETs — Neutrophil Extracellular Traps.

Thus, the pathogenesis of PE has not been studied sufficiently yet. The disease seems to have a heterogeneous nature that affects the prognosis. The available fundamental data are not yet reflected in the clinical practice; therefore, further clinical and experimental studies with uniform comparable parameters assessment are still required.

The authors declare no conflicts of interest.