Abstract. Purpose: To develop an animal model of acute deep vein thrombosis (DVT). Methods: In part I of the study nine juvenile domestic pigs were used.
CardioVascular and Interventional Radiology
© Springer-Verlag New York, Inc. 1998
Cardiovasc Intervent Radiol (1998) 21:329 –333 DOI: 10.1007/s002709900270
LABORATORY INVESTIGATIONS
Animal Model of Acute Deep Vein Thrombosis Sumit Roy,1 Frode Lærum,1 Frank Brosstad,2 Knut Kvernebo,3 Kjell S. Sakariassen4* 1
Institute for Surgical Research, National Hospital, N-0027 Oslo, Norway Research Institute for Internal Medicine, National Hospital, N-0027 Oslo, Norway 3 Department of Surgery, Ullevål Hospital, Kirkevien 166, N-0407 Oslo, Norway 4 Nycomed Bioreg A/S, Forskningsparken, Gaustadalleen 21, N-0371 Oslo, Norway 2
Abstract Purpose: To develop an animal model of acute deep vein thrombosis (DVT). Methods: In part I of the study nine juvenile domestic pigs were used. Each external iliac vein was transluminally occluded with a balloon catheter. Thrombin was infused through a microcatheter in one leg according to one of the following protocols: (1) intraarterial (IA): 1250 U at 25 U/min in the common femoral artery (n 5 3); (2) intravenous (IV): 5000 U in the popliteal vein at 500 U/min (n 5 3), or at 100 U/min (n 5 3). Saline was administered in the opposite leg. After the animals were killed, the mass of thrombus in the iliofemoral veins was measured. The pudendoepiploic (PEV), profunda femoris (PF), and popliteal veins (PV) were examined. Thrombosis in the tributaries of the superficial femoral vein (SFVt) was graded according to a three-point scale (0, 1, 11). In part II of the study IV administration was further investigated in nine pigs using the following three regimens with 1000 U at 25 U/min serving as the control: (1) 1000 U at 100 U/min, (2) 250 U at 25 U/min, (3) 250 U at 6.25 U/min. Results: All animals survived. In part I median thrombus mass in the test limbs was 1.40 g as compared with 0.25 g in the controls (p 5 0.01). PEV, PFV and PV were thrombosed in all limbs infused with thrombin. IV infusion was more effective in inducing thrombosis in both the parent veins (mass 1.32–1.78 g) and SVFt (11 in 4 of 6 legs), as compared with IA infusion (mass 0.0 –1.16 g; SFVt 11 in 1 of 3 legs). In part II thrombus mass in axial veins ranged from 1.23 to 2.86 g, and showed no relationship with the dose of thrombin or the rate of infusion. Tributary thrombosis was less extensive with 250 U at 25 U/min than with the other regimens.
* Present address: Department of General Physiology, Institute for Biology, University of Oslo, Kristine Bonnevies hus, Blindern, N-0316 Oslo, Norway Correspondence to: Dr. S. Roy
Conclusion: Slow distal intravenous thrombin infusion in the hind legs of pigs combined with proximal venous occlusion induces thrombosis in the leg veins that closely resembles clinical DVT in distribution. Key words: Thrombosis, experimental—Veins, extremities The development of transcatheter treatment methods for venous thrombosis has been greatly hindered by the lack of a suitable animal model. Out of convenience or to limit costs [1], most of the work to date has been done on rodents [1– 4]. Naturally, this has precluded the satisfactory study of regional thrombolysis protocols. Larger models reported in the literature either provide unsatisfactory replicas of the clinical disease [5– 8] or are difficult to create [9]. While some groups have relied on an ex vivo model[4], whole-animal systems have been the centerpiece of most studies. Stasis either alone [2] or together with a pharmacological [2, 5] or an electrical method [6] for triggering the coagulation cascade has been the procedural choice of most researchers. Another method that has recently regained acceptance after long being out of favor is intraluminal placement of a non-occluding foreign body such as nylon or wool thread [8]. The great variety of methods described in the literature underscores the limitations of the lesions created. These models essentially represent modifications of the Wessler ‘‘test’’ described four decades ago [10], and are characterized by focal thrombosis of a single vein. In human deep vein thrombosis, large masses of thrombus extend in continuity from the tributaries into the axial veins of the legs, and thus represent a pathological entity somewhat different from the in vivo models currently available for research on transcatheter thrombolysis. In light of the current consensus on the etiology of venous thrombosis [11], a strong case can be made for developing a method for creating an animal model that excludes the intravascular insertion of foreign bodies, and causes at most mild endothelial damage. The anatomy of the vascular territory chosen should bear a substantial topographic and di-
330
mensional resemblance to the venous system of the lower limb in humans. As the utility of an animal model ultimately rests on its ease of creation, operative procedures should be substituted by minimally invasive techniques as much as possible. Endoluminal flow obstruction followed by transcatheter peripheral thrombin instillation probably best duplicates the putative pathogenetic scenario in humans, besides satisfying the above criteria. Simple techniques based on this principle were explored in pigs to establish a realistic and reproducible morphological simulation of clinical deep vein thrombosis.
Materials and Methods The study was performed on non-specific pathogen-free juvenile domestic pigs 18 –25 kg in weight. The study protocol was approved by the National Committee for Animal Research.
Part I: Establishment of the Animal Model Nine animals were used for this phase of the study. Following preliminary sedation with intramuscular ketamine (Ketalar, ParkeDavis, Barcelona, Spain) 15 mg/kg and azaperone 2 mg/kg (Stresnil, Janssen Pharmaceutica, Beerse, Belgium), general anesthesia was induced by the intravenous administration of pentobarbitone 10 mg/kg and maintained with bolus doses of 100 mg as required. The animal was intubated, and mechanically ventilated with room air. Bilateral external jugular venotomies were performed. A 7 Fr dilator was placed in the left external jugular vein and physiological saline containing 5 U/ml heparin infused at 1 ml/min. Two Swan– Ganz catheters (Baxter Healthcare, Irvine, CA, USA) were introduced in the right external jugular vein and one advanced into each external iliac vein.
Model 1: Transluminal Occlusion of the External iliac Vein Followed by Thrombin Infusion in Superficial Femoral Artery A left common carotid arteriotomy was performed and an 8 Fr vascular sheath placed in the vessel. Two 3 Fr microcatheters (Cook, Bloomington, IN, USA; Terumo, Tokyo, Japan) were introduced via the sheath into the descending aorta. One microcatheter was negotiated into each superficial femoral artery. The balloons on the iliac venous catheters were then inflated. Bovine thrombin (Thrombostat, Parke-Davis, Cherry Hill, NJ, USA) was infused in the right superficial femoral artery at 0.25 ml/min as a 100 U/ml solution (25 U/min) for up to 50 min. Saline was administered at the same rate in the left limb. Bilateral femoral arteriography was performed after every aliquot of 10 ml. The infusion was halted when flow arrest in the right femoral artery was observed or when 1250 U had been administered. Three animals were used.
Model 2: Transluminal Occlusion of the External Iliac Vein Followed by Retrograde Thrombin Infusion Two 3 Fr microcatheters were coaxially passed through a 8 Fr introducer sheath placed in the right internal jugular vein, and one
S. Roy et al.: Animal Model of Venous Thrombosis
advanced into each popliteal vein. Contrast was injected to confirm that the veins were patent. The balloons were inflated and venography repeated. Thrombin was infused in the right leg through the microcatheter according to one of the two regimens described below, while the opposite limb served as the control. (1) thrombin 5000 U at 500 U/min as a 500 U/ml solution; (2) thrombin 5000 U at 100 U/min as a 100 U/ml solution. At the end of thrombin infusion another set of venograms was obtained. Each regimen was tested on three animals. Blood samples were taken before thrombin infusion was begun and after it was terminated, and during the procedure after every fifth of the dose was administered. Activated partial thromboplastin time, prothrombin time, platelet count, and plasma fibrinogen concentration were determined. Fifteen minutes after the end of thrombin infusion, heparin 2500 IU was intravenously administered and 5 min later the animal was killed with an overdose of pentobarbitone. A prolonged observation period was deemed unnecessary as rapid spontaneous lysis of occlusive venous thrombi does not occur ([5] and unpublished data). The major deep veins in both hind limbs were immediately dissected. The pudendoepiploic and profunda femoris veins were opened and examined for the presence of thrombus, as were the iliofemoral arteries. The extent of thrombosis in the tributaries of the superficial femoral veins was graded according to a three-point scale: 0, few or no tributaries thrombosed; 1, most tributaries thrombosed; 11, all tributaries thrombosed. This subjective, semiquantitative scoring system was used because the variability in the number and distribution of tributaries and their small calibers made gravimetric estimation of thrombosis impractical. The external iliac, common femoral, superficial femoral, and popliteal veins were removed en bloc and fixed in formalin (Fig. 1). The corresponding arteries were opened and examined for the presence of thrombus. Venous thrombi were harvested and weighed.
Part II: Refinement of the Animal Model On the basis of the results obtained from part I of the study, intravenous thrombin infusion was further investigated in a series of nine animals. Essentially the same experimental procedure was retained. In one limb, 1000 U of thrombin was infused in the popliteal vein at 25 U/min as a 25 U/ml solution (40 ml). One of the following regimens was substituted in the opposite leg: (1) 1000 U at 100 U/min (100 U/ml solution; 10 ml); (2) 250 U at 25 U/min (25 U/ml solution; 10 ml); (3) 250 U at 6.25 U/min (6.25 U/ml solution; 40 ml) When the two regimens were of unequal length, the shorter infusion was timed to begin so that both infusions ended simultaneously. Likewise, the external iliac vein was kept occluded only during the duration of the infusion on the ipsilateral side. As the data were not normally distributed, the Wilcoxon test was used to calculate the p value for the differences in thrombus mass between the test limbs and the controls.
Results All the animals survived thrombin infusion. Procedural failures did not occur. Venous rupture due to balloon inflation was not encountered.
S. Roy et al.: Animal Model of Venous Thrombosis
331
Table 1. Extent of thrombosis in the tributaries of the superficial femoral vein observed with the intravenous thrombin regimens tested Thrombin dose
5000 U 1000 U 250 U
Rate of infusion) (U/min)
500 (n53) 100 (n53) 100 (n53) 25 (n59) 25 (n53) 6.25 (n53)
Gradea of SFVt thrombosis (no. of limbs) 0
1
11
0 0 0 0 0 0
2 0 1 3 3 1
1 3 2 6 0 2
SFVt 5 tributaries of the superficial femoral vein Grade: 0 5 few or no tributaries thrombosed; 1 5 most tributaries thrombosed; 11 5 all tributaries thrombosed
a
Fig. 1. Anatomy of the venous territory used to create the model of deep vein thrombosis. With the exception of a large pudenoepiploic vein opening in the external iliac vein, the topography closely resembles its human counterpart. EIV 5 external iliac vein; PEV 5 pudendoepiploic vein; CFV 5 common iliac vein; PFV 5 profunda femoris vein; SFV 5 superficial femoral vein; PV 5 popliteal vein; SFVt 5 tributaries of the superficial femoral vein; il 5 inguinal ligament.
Part I Intraarterial thrombin administration resulted in occlusive thrombosis extending from the external iliac vein to the popliteal vein in two animals, with the harvested thrombus weighing 0.50 g and 1.16 g respectively. Superficial femoral vein thrombosis was rated as grade 1 in two legs and as 11 in the third. The corresponding data for intravenous infusions are incorporated in Table 1 and Figure 2. The pudendoepiploic and profunda femoris veins were invariably thrombosed irrespective of regimen. Some degree of thrombosis was also observed in eight of nine control limbs, but was largely restricted to the external iliac and common femoral veins. Median thrombus mass was 0.25 g (range 0.0 –1.01 g) as compared with 1.40 g (range 0.0 –1.78 g) in the test limbs (p 5 0.01). The superficial femoral vein tributaries were invariably free of thrombus, whereas the pudendoepiploic and profunda femoral veins were only sporadically involved.
Intraarterial thrombin infusion was inevitably followed by ipsilateral arterial thrombosis. It occurred after 1000 U had been administered in two animals, and 1250 U in the third. With intravenous infusion this complication did not occur. With one exception, thrombin administration was associated with a mild prolongation of the activated partial thromboplastin time, with the final value exceeding the baseline by 3 sec in only two animals (range 0 –3.4 sec). The prothrombin time at the end of infusion was never greater than 1.5 times the original value (range 0.8 –1.4; n 5 8). The final fibrinogen concentration normalized to the baseline was less than 0.5 in only one animal (range 0.47– 0.90). The drop in platelet count ranged from 59 3 109 to 240 3 109/L. In none of the animals did the count fall below 300 3 109/L.
Part II Thrombosis of the primary venous channels and the major tributaries was invariably observed even with the minimal dose used. Thrombus mass in the iliofemoral veins ranged from 1.35 to 3.09 g. No correlation was observed between this variable and either the thrombin dose or the rate of infusion (Fig. 2). Thrombosis of the smaller tributaries, albeit of varying severity, occurred in all limbs (Table 1). The results of the hematological tests confirmed that the systemic effects of thrombin infusion in the leg veins after proximal flow arrest are minimal (Table 2). No clear relationship between the dose of thrombin administered and the change in coagulation profile was detected.
Discussion Extension of the lesion from the periphery to the iliofemoral veins, preferably in continuity, is the sine qua non for a useful model of deep vein thrombosis. Venous thrombosis induced by the slow infusion of thrombin in the popliteal vein in the presence of flow arrest seems to fulfill this criterion quite satisfactorily. Thrombosis resembling human deep vein thrombosis distribution was observed in every animal so treated.
332
S. Roy et al.: Animal Model of Venous Thrombosis
Fig. 2. Scattergram of thrombus mass harvested from the limb after intravenous thrombin infusion. The amount of thrombus produced varied little across the wide range of dose and infusion rates tested. Table 2. Change in hematological parameters following intravenous thrombin infusion in the popliteal vein Test
Median
95% confidence interval
Activated partial thromboplastin time (difference between final and baseline values) Prothrombin time (ratio of final to baseline value) Fibrinogen (ratio of final to baseline value) Platelet count (difference between final and baseline values)
0.8 sec
21.0, 1.7
1.2
1.1, 1.3
0.59 131 3 109/L
0.40, 0.71 80, 153
Table 3. Extent of thrombosis in the tributaries of the superficial femoral vein as a function of volume of thrombin infused Volume of thrombin solution
Proportion with SFVt thrombosis (grade 11)
Low (10 ml) (n 5 9) 5000 U (n 5 3) 1000 U (n 5 3) 250 U (n 5 3) High (40 ml or 50 ml) (n 5 15) 5000 U (n 5 3) 1000 U (n 5 9) 250 U (n 5 3)
33% (2, 64)
73% (51, 95)
Figures in parentheses represent 95% confidence intervals. SFVt 5 tributaries of the superficial femoral vein
Impairment of blood flow in the deep veins is probably the single most important factor predisposing to the development of thrombosis in these vessels [11]. In recognition of this fact, surgical flow impairment forms the central component of most techniques used for triggering venous thrombosis in an experimental setting. Our results suggest that endovascular occlusion using a balloon catheter introduced from a distant site can be a suitable alternative. Damage to perivascular nerves modulating fibrinolysis is avoided [12,
13], and reoperation to remove the ligature rendered superfluous [5]. By inflating the balloon in the external iliac vein central to the large pudendoepiploic tributary, complete flow arrest was avoided; some venous flow is believed to be necessary to ensure the formation of a true platelet-containing thrombus rather than a fibrin-rich substitute [3]. Endoluminal occlusion of the external iliac vein was achieved in all 36 limbs in this study. Balloon failure during thrombin infusion was never encountered. Nonetheless, risk of the latter is a potential disadvantage of transluminal occlusion. Hence, periodic fluoroscopic monitoring is recommended because marked movement of the catheter tip in synchrony with respiratory excursions reliably identifies a deflated balloon. Though two groups have reported results seemingly to the contrary, stasis alone [2, 7], unless prolonged, does not consistently induce thrombosis in an intact vein [11, 14]. Hence the potential of thrombin administration was investigated. Unlike many other ubiquitous enzymes, thrombin has retained its immunological identity across species boundaries, allowing the use of freely available bovine thrombin for the experiments [15]. Further, even when blood flow is severely compromised, the enzyme is non-toxic to endothelial cells [3]. Across the range of doses tested, the thrombus mass did not vary appreciably with the dose of thrombin administered (Fig. 2). Neither did the dose influence the extent of small vein thrombosis (Table 1). Unlike the dose, the volume of administration did appear to be an important factor. At a given dose, larger volumes were associated with more widespread thrombosis of superficial femoral vein tributaries (Table 3), probably because a better distribution of thrombin was achieved. From these results 250 IU of thrombin at 6.25 IU/min would appear to be the preferred regimen for inducing thrombosis. Pigs react poorly to stress and hence can succumb to even simple interventions via laparotomy [5, 16]. Our reliance on transcatheter techniques probably helped en-
S. Roy et al.: Animal Model of Venous Thrombosis
sure the survival of all 18 animals in this series. No evidence of systemic hypercoagulability was observed. On the contrary mild prolongation of the partial thromboplastin time and prothrombin time very frequently occurred (Table 2). Stimulation of the protein C and S system by the interaction of thrombin with thrombomodulin is a plausible explanation for this apparent paradox. The consumption of fibrinogen during thrombosis may have also been a contributory factor. A drop in the platelet count was a uniform accompaniment of thrombin infusion. Despite the hypofibrinogenemia and thrombocytopenia, spontaneous bleeding or excessive oozing from the venotomy wound sites did not occur in any of the animals. Ideally, an animal model will reflect not only the morphology of the human lesion, but also its cellular composition and ultrastructural characteristics. Though histological confirmation is not available from this study, it may be pertinent to point out that thrombin instillation in the inferior vena cava combined with partial flow obstruction resulted in histologically realistic thrombi in rats [3]. The large drop in platelet count that invariably followed thrombin infusion in our animals adds indirect support to our belief that bona fide thrombi rather than fibrin-rich clots were formed. Passive entrapment of platelets in the latter would scarcely have altered the platelet count to the degree observed. As the gel structure of the thrombi was not examined, it remains to be seen how closely their fibrillar architecture resembles that of their clinical counterparts. The use of prothrombin time and activated partial thromboplastin times as measures of hypercoagulability could perhaps be questioned. It is possible that the assay of thrombin-antithrombin complex would have been a better choice because of its specificity for thrombin and its sensitivity to prethrombotic states [15]. Despite the limitations discussed above, we believe the model described represents a fresh look at the time-hallowed concept underlying the Wessler’s ‘‘test’’. Distal transcatheter infusion of 250 IU thrombin at 6.25 IU/min combined with proximal endoluminal venous occlusion in the legs of pigs induces peripheral venous thrombosis resembling acute deep vein thrombosis, particularly as regards anatomical distribution. The method is simple and safe: therefore this model of deep vein thrombosis offers a realistic and repro-
333
ducible in vivo environment for evaluating new treatment methods for the condition. References 1. Stassen JM, Lijnen HR, Kieckens L, Collen D (1991) Small animal thrombosis models for the evaluation of thrombolytic agents. Circulation 83 [Suppl IV]: 65–72 2. Monreal M, Silveira P, Monreal L, Monasterio J, Angles AM, Lafoz E, Lorente L (1991) Comparative study on the antithrombotic efficacy of four low-molecular weight heparins in three different models of experimental venous thrombosis. Haemostasis 21:91–97 3. Northeast ADR, Burnand KG (1992) The response of the vessel wall to thrombosis: The in vivo study of venous thrombolysis. Ann N Y Acad Sci 667:127–140 4. Imura Y, Stassen JM, Vreys I, Lesaffre E, Gold HK, Collen D (1992) Synergistic antithrombotic properties of G4120, a RGD-containing synthetic peptide, and agartroban, a synthetic thrombin inhibitor, in a hamster femoral vein platelet-rich thrombosis model. Thromb Haemost 68:336 –340 5. Hosaka J, Roy S, Kvernebo K, Enge I, Laerum F (1996) Induced thrombosis in the pig inferior vena cava: A model of deep venous thrombosis. J Vasc Interv Radiol 7:395– 400 6. Ulutin ON, Tunali H, Ugur MS, Aytis S, Erbengi T, Balkuv-Ulutin S (1986) Effect of defibrotide in electrically induced thrombosis in dogs. Haemostasis 16 [Suppl 1]:9 –12 7. Hamer JD, Malone PC (1984) Experimental deep venous thrombogenesis by a non-invasive method. Ann R Coll Surg 66:416 – 419 8. Korninger C, Matsuo O, Suy R, Stassen JM, Collen D (1982) Thrombolysis with human extrinsic (tissue-type) plasminogen activator in dog with femoral vein thrombosis. J Clin Invest 69:573–580 9. Wakefield TW, Greenfield LJ, Rolfe MW, De Lucia A 3d, Strieter RM, Abrams GD, Kunkel SL, Esmon CT, Wrobleski SK, Kadell AM et al. (1993) Inflammatory and procoagulant mediator interactions in an experimental baboon model of venous thrombosis. Thromb Haemost 69:164 –172 10. Wessler S (1955) Studies in intravascular coagulation. III. The pathogenesis of serum-induced venous thrombosis. J Clin Invest 34:647– 651 11. Thomas DP (1988) Overview of venous thrombogenesis. Semin Thromb Hemost 14:1– 8 12. Iocaviello L, d’Adamo M, de Curtis A, Buczko W, Donati MB (1995) Enhanced vascular plasminogen activator (t-PA) release by epinephrine in aged rats. Thromb Haemost 73:841– 844 13. Bashkov GV, Sergeev IY, Medvedeva NA (1993) Role of sympathetic cholinergic pathway in the neurogenous control of tissue-type plasminogen activator release into the blood. Blood Coag Fibrinol 4:993–998 14. Thomas DP (1996) Venous thrombosis and the ‘‘Wessler Test’’. Thromb Haemost 76:1– 4 15. Ravanat C, Freund M, Dol F, Cadroy Y, Roussi J, Incardona F, Maffrand JP, Boneu B, Drovet L, Legrand CU, et al. (1995) Crossreactivity of human molecular markers for detection of prethrombotic states in various animal species. Blood Coag Fibrinol 6:446 – 455 16. Lundgren TI, Revhaug A, Giercksky KE (1983) The pig as an experimental animal: Selection of animals, handling and housing requirements. Scand J Clin Lab Invest 43:59 – 65
