An eight-hour ECMO model was successfully established in rats and can be used to examine differences in circuit thrombus formation using different anticoagulation methods. The model is simple, economical, safe, and reproducible. Unlike a large animal model, a single operator can perform experiments with this model independently, including multiple experiments at the same time. In addition, this model could be produced in the same way in different laboratories using standardized rat strains, allowing better comparison between different laboratories.
To our knowledge, there is no previous rat model designed for investigating ECMO coagulopathy. Several studies have reported the establishment of an animal model of cardiopulmonary bypass (CPB) and venoarterial (V-A) ECMO in rats [19–32]. (Table 2). However, these models were developed to investigate the pathophysiology, complications, and effectiveness of methods of cardiac arrest specific to CPB and V-A ECMO. CPB is usually performed by central cannulation through median sternotomy because its purpose is complete replacement of cardiopulmonary function during short-term cardiac surgery. The CPB circuit requires an open reservoir, which requires anticoagulation therapy with high-dose heparin. In contrast, ECMO provides long-term cardiac or pulmonary support for severe cardiac or respiratory failure in the intensive care unit by peripheral cannulation. The ECMO circuit is a closed system without a reservoir. During ECMO, bleeding and thrombus formation are managed by titration of minimal anticoagulation. Therefore, previous rat models have certain limitations for investigating ECMO coagulopathy. First, they require a venous reservoir and large surface area of the oxygenator, which strongly promote thrombus formation. Second, the dosage of heparin used in most previous models is 500 units/kg, which greatly exceeds that used in ECMO (Table 2). Even in the V-A ECMO model, a venous reservoir and high doses of heparin (285–400 IU/kg) were used [31]. Third, they require median sternotomy or a custom-designed cannula and application of transesophageal echocardiography (TEE) [21] to avoid drainage failure. Finally, all of them depend on a large priming volume of between 4 and 40 mL, which represents approximately 10 to 200% of the circulating blood volume of a 250 to 550 g male Sprague-Dawley rat (blood volume 64 mL/kg). This problem is compounded by the need to take blood samples during support for the assessment of coagulation and other physiological variables. The combination of a large circuit volume plus blood sampling can quickly result in excessive hemodilution. As a result, some of the models require donor blood to prime the circuit [19, 21, 22, 25]. These disadvantages limit the use of these models for coagulation research in ECMO.
Table 2
Previously published rat cardiopulmonary bypass and venoarterial ECMO models
|
Weight
(grams)
|
Oxygenator
|
Surface area
of oxygenator (m2)
|
Priming volume (mL)
|
Donor priming
|
Duration (hours)
|
Dosage of heparin
|
Popovic et al. [19]
|
250–300
|
Homemade
|
0.048–0.064
|
13.7
|
yes
|
3
|
?
|
Alexander and Al Ani [20]
|
250
|
Homemade
|
0.017
|
12
|
no
|
6
|
1 mg/kg
|
Grocott et al. [21]
|
350–400
|
Homemade
|
0.33
|
40
|
yes
|
1
|
150 IU
|
Fabre et al. [22]
|
475–550
|
Homemade
|
?
|
35
|
yes
|
2
|
500 IU/kg
|
Gourlay et al. [23]
|
350–450
|
Homemade
|
0.05
|
12
|
no
|
1
|
1000 IU/kg
|
Hamamoto et al. [24]
|
420 − 350
|
HPO-003
|
0.03
|
9
|
no
|
1
|
200 IU/kg
|
Dong et al. [25]
|
450–550
|
Micro-1
|
0.05
|
4
|
yes
|
1
|
500 IU/kg
|
You et al. [26]
|
430–475
|
Homemade
|
0.02
|
9
|
no
|
1
|
500 IU/kg
|
Modine et al. [27]
|
422.9 ± 32
|
Homemade
|
?
|
10
|
no
|
1
|
500 IU/kg
|
Gunzinger et al. [28]
|
400–500
|
Homemade
|
0.063
|
8
|
no
|
1
|
500 IU/kg
|
Cresce et al. [29]
|
250–350
|
Homemade
|
0.045
|
6.3
|
no
|
2
|
500 IU/kg
|
Zhang et al. [30]
|
250–300
|
Micro-1
|
0.05
|
< 12
|
no
|
0
|
500 IU/kg
|
Ali et al. [31]
|
250–350
|
Homemade
|
?
|
8
|
no
|
1
|
100 IU
|
Chang et al. [32]
|
450–550
|
Micro-1
|
0.05
|
20
|
?
|
1
|
500 IU
|
Present study
|
450–550
|
3D-printed
|
0.0015
|
2.5
|
no
|
8
|
15–50 IU/kg/h
|
Data are expressed as mean ± standard deviation [27]. Duration: Duration of cardiopulmonary bypass or venoarterial ECMO. |
These problems were addressed in the rat ECMO model performed in this study. First, unnecessary aspects of CPB that are not present in ECMO were eliminated, including the venous reservoir and median sternotomy. Second, a pumped A-V configuration was chosen instead of the more common venovenous (V-V) or V-A configurations to avoid complications that occur when the cannula tip is not perfectly placed near the right atrium. This requires TEE, a custom-designed cannula and gravity blood drainage to avoid inserting the cannula too little, resulting in poor venous drainage, or too far, resulting in right ventricular damage and/or arrythmias. Furthermore, we decided against pumpless A-V support to avoid changes in circuit flow that occur due to arterial blood pressure changes under anesthesia. Third, the oxygenator was manufactured using 3D printing, which allows a smaller prime volume and greater reproducibility. The commercially available Micro1 (Kewei Medical Instrument Inc., China) has a surface area of 0.05 m2 and requires a 3.5 mL priming volume [25, 30, 32, 33]. Similarly, commercially available HPO-003 (Senko Medical Co., Ltd, Japan) has a surface area of 0.03 m2 and a 3.3 mL priming volume but utilizes blood flow inside fibers, unlike current ECMO oxygenators [24]. Other custom-built oxygenators used in these studies required a priming volume between 6.3 and 40 mL [19–23, 26–29, 31]. Unlike these oxygenators, the 3D-printed microfluidic device in this study has a 0.3 mL priming volume.
As a result of all these changes, the total priming volume of the circuit was reduced to 2.5 mL. This enabled a small animal model with hemodynamic stability and reduced hemodilution, which is crucial for investigating coagulopathy during ECMO. In previous CPB animal models, the MAP fell to 44–66 mmHg during CPB [21–23, 26, 27]. In this study, MAP decreased when ECMO was initiated but was maintained at approximately 80 mmHg thereafter. The reason for hypotension at the beginning of ECMO was due to blood being pulled from the drainage cannula and hemodilution with the circuit prime volume. The need for more vasopressors at the end of the experiment was likely due to progressive vasodilation due to prolonged anesthesia and repeated blood sampling. Lactate levels were normal throughout the experiment in all rats except one, and thus, adequate tissue perfusion and oxygenation were maintained.
Hemodilution was also minimal. Previous studies have reported that Hct and Hb concentrations were decreased by 35–50% from baseline in CPB without donor blood due to hemodilution from priming and laboratory sampling [23, 26, 28, 30]. However, Hct and Hb concentrations were decreased by only 20–30% from baseline at the end of the study in this current model. The final values of Hct and Hb were approximately 30% and 10 g/dL, respectively, which are still within clinically acceptable limits.
Most importantly, our results provide a guide to heparinization methods in the rat model. Most of the previous CPB and V-A ECMO models did not measure clotting times throughout the study to evaluate the anticoagulation level [19–24, 26–31]. Other experimental studies have maintained high ACT levels of over 300 seconds [25, 32]. However, these are different from the clinical practice of ECMO and are not suitable for studying anticoagulation therapy in the setting of clinical ECMO. To our knowledge, there are no previous studies that showed the relationship between ACT level and heparin dosage in a rat model. In this study, ACT was controlled within the range of 152 to 210 seconds in the HD group. The majority of institutions utilize an ACT between 140 and 220 seconds [5]. Therefore, this same dose of heparin should serve as a control group for future studies examining new anticoagulation regimens, including surface coatings and systemic anticoagulants. In addition, this study demonstrated lower thrombus weight and longer bleeding times with an increasing level of heparin anticoagulation. In a meta-analysis by Sy et al. [5] patients with a higher ACT target (> 180 seconds) bled more than patients with lower targets (< 180 seconds). However, a greater level of anticoagulation was associated with lower thromboembolic events [5]. Our results are consistent with clinical studies and suggest that mock oxygenators could be useful in investigating thrombus formation.
In contrast, the platelet count decreased over time in both groups, with no significant difference between groups (p = 0.94). Thrombocytopenia during ECMO has been well documented during in vivo and clinical ECMO studies [8, 20, 34]. This observation is typically attributed to platelets adhering to the circuit and oxygenator surfaces [35] and damage or activation by the pump [36]. However, emerging studies suggest that this effect may not be due to gross thrombus formation in the ECMO circuit [8, 10]. This rat ECMO study is thus consistent with these studies’ suggestion that the degree of thrombocytopenia does not appear to be related to the degree of thrombus formation.
Despite the positive results, this study has several potential limitations. First, circuit inflammation was evaluated in this study only by white blood cell count. To understand this aspect of blood biocompatibility, a more complete study is needed to examine complement activation and cytokine generation. Second, our model does not use the polymethylpentene (PMP) hollow fibers that form clinical ECMO oxygenator gas exchange membranes. The simulated fibers in our devices are solid urethane acrylate (UA) posts, and thus, one cannot study gas exchange. Urethane acrylate and PMP are both hydrophobic polymers, and thus, the thrombusting process is similar. However, the rate of thrombus formation may be slightly different due to the different chemical structures. Last, rats have a platelet count three to six times higher than that of humans, and thus, the rate of thrombus formation and thrombus composition may be different, even at the same ACT [37].