Acute Immune Response in Venoarterial and Venovenous Extracorporeal Membrane Oxygenation Models of Rats

Abstract

Although experimental extracorporeal membrane oxygenation (ECMO) animal models have been reported, there are few studies on the immune response to ECMO. We developed the venoarterial (VA) and venovenous (VV) model in rats and serially investigated the changes in the distribution of immune cells. Forty rats underwent both VA and VV modes of ECMO, and blood samples were collected at 1 day before ECMO (D-1), at the end of ECMO run (D+0), and 3 days after the ECMO (D+3). Flow cytometry was used to characterize surface marker expression (CD3, CD4, CD8, CD43, CD45, CD45R, CD161, and His48) on immune cells. Granulocytes were initially activated in both ECMO types and were further reduced but not normalized until 3 days of decannulation. Monocyte and natural killer cells were decreased initially in VA mode. B lymphocytes, helper T lymphocytes, and cytotoxic T lymphocytes also significantly decreased in VA modes after ECMO, but this phenomenon was not prominent in the VV modes. Overall immune cells proportion changed after ECMO run in both modes, and the immunologic balance altered significantly in the VA than in VV mode. Our ECMO model is feasible for the hemodynamic and immunologic research, and further long-term evaluation is needed.

Introduction

Extracorporeal membrane oxygenation (ECMO) is an accepted modality of mechanical circulatory support for patients with cardiac and respiratory failure unresponsive to conventional therapy.¹ While the utilization of ECMO is increasing, the survival rates vary between 30% and 70%, depending on the configuration and cause of illness that has necessitated the ECMO.² The causes of death include sepsis, neurologic injuries, thromboembolic complications, and bleeding, associated principally with activation of inflammatory mediators and changes in immune function during ECMO support.³ To better understand the interactions between extracorporeal circuits/oxygenators and the body, animal models have been developed by many researchers. ECMO models in large animals (dogs, sheep, and pigs) allow simple placement of cannula and mimic the human better than do small animals.⁴ However, such models are costly. Although ECMO models in small animals (rabbits, rats, and mice) have also been developed, these do not allow long-duration ECMO and cannulation is difficult.^5–13

The venoarterial (VA) model played a decisive role in conducting the study of ECMO-related pathophysiology, while until recently, the study of small animal models for venovenous (VV) ECMO was limited.

Here, the aim of our study was to develop reproducible models of both VA and VV ECMO in rats, which can be used to explore the acute changes in immune cell profiles.

Materials and Methods

General Procedure

All animals received human-quality care in compliance with the Principles of Laboratory Animal Care formulated by the National Society for Animal Research and the Guide for the Care and Use of Laboratory Animal Resources. The study was approved by the Ethics Committee of Chonnam National University Medical School (approval no. CNU IACUC-H-2018-36). We used 40 male Sprague–Dawley rats (500–600 g, purchased from Samtako Bio Korea Co., Ltd., Osan City, Korea). All animals were housed at a controlled temperature (20°C) under a 12 hours/12 hours light-dark cycle with standard food and water available ad libitum.

All animals were anesthetized via intramuscular injection of ketamine (80 mg/kg) and xylazine (8 mg/kg). Isoflurane inhalation was maintained during all surgical procedures. Each rat was secured supine on a surgical table, and heating pads positioned to avoid hypothermia. Each rat was intubated with a 16 gauge catheter and mechanically ventilated with 90% (v/v) oxygen and 1–1.5% (v/v) isoflurane via a rodent respirator (Harvard Apparatus Inc.; Holliston, MA). The tidal volume was 8 ml/kg, and the respiratory rate was 55 breaths/min. The right femoral artery was cannulated with a 24 gauge catheter to monitor the systemic arterial blood pressure (BP) and to collect arterial blood samples during the experiment.

Establishment of Rat ECMO Model

In the VA mode, the drainage cannula was inserted into the right external jugular vein and the perfusion cannula into the right carotid artery (Figure 1A). In the VV mode, the drainage cannula was inserted into the right external jugular vein and the perfusion cannula into the left jugular vein (Figure 1B). To create a drainage catheter, a neonatal feeding tube (5-French, 50 cm) was used with 10 cm lengths and the end of catheter was placed at the junction of right atrium and superior vena cava. We made eight side-holes to enhance drainage (Figure 1C). A 24 guage angiocatheter was used as a perfusion catheter. Heparin sodium (1,000 IU/kg) was administered before cannulation.

Figure 1.: The photographs of cannulation in both ECMO modes and drainage catheter. A: Venoarterial mode, blood drains from right jugular vein (blue arrow) and perfused to right carotid artery (red arrow). B: Venovenous mode, blood drains from right jugular vein (blue arrow) and perfused to left jugular vein (red arrow). C: Modified drainage catheter with eight side holes (black arrows) to enhance drainage. ECMO, extracorporeal membrane oxygenation.

The ECMO circuit consisted of a small, polypropylene animal membranous oxygenator (Micro-1) with a gas exchange surface of 0.05 m² and a priming volume of 3.5 ml (Dongguan Kewei Medical Instrument Co. Ltd.; Dongguan City, China); a polyvinyl chloride tube (internal diameter: 2.5 mm); a venous reservoir (a 10 ml syringe containing 5–6 ml of priming solution); and a peristaltic pump (Watson-Marlow Pumps; Falmouth, Cornwall, United Kingdom). The cannulas, tubes, and venous reservoir were consecutively interconnected via a “three-way stopcock.” The venous reservoir supplied the additional fluid required to maintain pump flow. Each ECMO circuit was primed with 14 ml of solution: 7 ml albumin solution and 7 ml normal saline (NS). Each ECMO run was 2 hours in duration at a flow rate of 50 ml/min (20 rpm). The priming volumes and circuit assemblies were identical in all rats. During circulation, a heating lamp was focused on the primed circuit to prevent hypothermia (Figure 2).

Figure 2.: The schema of rat venoarterial ECMO model. A: Intubation with mechanical ventilation, B: hemodynamic monitor, C: drainage catheter through jugular vein and perfusion catheter through carotid artery, D: peristaltic pump and oxygenator, E: reservoir. ECMO, extracorporeal membrane oxygenation.

Classification of an Experimental Models

The experimental groups were classified in two ways; first by the mode of ECMO (VA and VV) and, second, by blood sampling time. Blood samples (2.5 ml) were collected at 1 day before the experiment (D-1), at the end of 2 hours ECMO run just before decannulation (D+0), and 3 days after the ECMO decannulation (D+3). Frequent blood sampling during extracorporeal circulation may result in fatal hypovolemia, so we did not perform three times samples sequentially in one experimental rat. Instead, an indirect comparison was performed by the following method: experiment 1 (comparisons of data between D-1 vs. D+0) and experiment 2 (comparisons of data between D-1 vs. D+3). We ultimately categorized all rats into four subgroups: VA (n = 10) and VV (n = 10) for experiment 1; and VA (n = 10) and VV (n = 10) for experiment 2 (Figure 3).

Figure 3.: Experimental classifications and blood sampling time in each group. The experimental groups were classified in two ways: first by the mode of ECMO and second by blood sampling time. Blood samples (2.5 ml) were collected at 1 day before the experiment (D-1), at the end of ECMO run just before decannulation (D+0), and 3 days after the ECMO decannulation (D+3). ABGA, arterial blood gas analysis; ECMO, extracorporeal membrane oxygenation; FACS, fluorescence activated cell sorting; VA, venoarterial; VV, venovenous.

Hemodynamic Monitoring and Blood Gas Analysis

During ECMO, the heart rate and systolic/diastolic BP were continuously monitored and recorded at 3, 15, 30, 60, 90, and 120 minutes from ECMO initiation till decannulation. Then, we compared the hemodynamic results between VA to VV model. After ECMO weaning, the remained priming solution (approximately 5 ml) was returned to each rat just before decannulation. After decannulation, the hemodynamic parameters were observed for 30 minutes on ventilator support after all surgical procedures had been completed. After general condition (movement and respiration) recovered, the rats were moved to cages. To confirm the stability of the model, we continued assessment of the rats for 30 days after the end of the experiment or until death. Arterial blood gas analysis (ABGA) was measured using the GEM Premier 3000 system (Werfen; Bedford, MA) using disposable cartridges that measure the pH, PCO₂, PO₂, lactate level, and hematocrit. Then, we compared between VA and VV models.

Immunological Assessment in Experimental Design

The following monoclonal antibodies and other reagents were used for flow-cytometric analysis: FITC-conjugated anti-CD3, APC-conjugated anti-CD4, PE-conjugated anti-CD8, BV421-conjugated anti-CD3 (all from Becton Dickinson Pharmingen; San Diego, CA); PerCP/Cy5.5-conjugated anti-CD45, APC-conjugated anti-CD161, PE-conjugated anti-CD43 (all from BioLegend; San Diego, CA); and PE-cyanine7-conjugated anti-CD45R and FITC-conjugated anti-His48 (both from eBioscience; San Diego, CA). BD Pharm Lyse (a lysis solution) is a buffered, concentrated (10×) ammonium chloride–based reagent (Becton Dickinson Bioscience; San Diego, CA) and was used to lyse erythrocytes. Cells were washed with phosphate-buffered saline without calcium and magnesium (Lonza; Walkersville, MD) before flow cytometry on a Navios instrument (Beckman Coulter; Brea, CA).

Plasma samples were obtained by centrifuging whole blood samples with 1,500 rpm for 15 minutes. Then after, the assays were coated with the following buffers and antibodies: 25 µl universal assay buffer 1×, 25 µl standard mix A, 25 µl 5-plex det. AB, 50 µl streptavidin-PE, 120 µl reading buffer (all from Bender Medsystems GmbH Vienna, Austria). Samples were incubated for two hours before applying antibody. The results of interleukins 6 (IL-6) and tumor necrosis factors α (TNF-α) were obtained with the “Luminex xMAPTM 200 (USA) machine.”

Statistical Analyses

Continuous variables are expressed as means ± standard deviations. The independent t-test was used to compare continuous variables between the groups. The paired t-test was used to compare the differences between two consecutive measurements. Repeated measures analysis of variance was used to compare serial values measured during ECMO (BPs and HR) between the groups. In all analyses, a p value < 0.05 was considered to indicate statistical significance. All statistical analyses were performed using MedCalc Statistical Software version 19.2.1 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org; 2020)

Results

Hemodynamic Monitoring and Blood Gas Analysis

Hemodynamic parameters (systolic/diastolic BP and heart rate) were significantly higher in VA than in the VV mode at all measurement times (Figure 4). All deaths recorded occurred within 5 days after decannulation, which led to overall survival of 85% in the VA mode and 60% in VV mode. Although there was no statistical significance in the 30 day survival comparison, the rats in VA mode showed a higher tendency to survive (Kaplan-Meier survival analysis; p = 0.07).

Figure 4.: Serial hemodynamic measurements. The hemodynamic parameters (systolic blood pressure, diastolic blood pressure, and heart rate) were significantly higher in the VA than the VV mode at all times. VA, venoarterial; VV, venovenous.

The results of ABGA in both ECMO modes of experiment 1 (D-1 vs. D+0) and experiment 2 (D-1 vs. D+3) are summarized in Table, Supplemental Digital Content 1, https://links.lww.com/ASAIO/A534. In experiment 1 (D-1 vs. D+0), after ECMO run, lactate levels were significantly elevated in both ECMO modes, and this phenomenon prevailed in the VV mode over the VA mode (p = 0.003) (Figure 5A). In experiment 2 (D-1 vs. D+3), lactate levels returned to pre-ECMO levels in the VA mode (p = 0.38) but still significantly higher in the VV mode (p = 0.03) after 3 days (D+3). Hyperlactatemia was associated with early mortality in experiment 1. The lactate level after ECMO run (D+0) was 2.5 ± 2.2 mg/dl in rats that survived and 6.0 ± 3.6 mg/dl in those that did not survived (p = 0.03) (Figure 5B).

Figure 5.: Lactate levels. A: The lactate level was significantly elevated in both groups after ECMO run (D+0), and it was especially higher in VV group than VA group. B: The lactate level after ECMO run (D+0) was 2.5 ± 2.2 mg/dl in rats that survived and 6.0 ± 3.6 mg/dl in those that did not. ECMO, extracorporeal membrane oxygenation; VA, venoarterial; VV, venovenous.

Changes of Immune Response After ECMO

Changes in Innate Immune Cells.

In experiment 1 (D-1 vs. D+0), the number of granulocytes (His48+) was significantly increased after ECMO run in both VA and VV modes of ECMO (p < 0.001 in the VA mode and p = 0.005 in the VV). In experiment 2 (D-1 vs. D+3), it remained high after 3 days in both modes but gradually stabilized (p = 0.009 in VA mode; p < 0.001 in VV) (Figure 6A and D).

Figure 6.: Innate immune cells change. The upper rows (A, B, and C) are comparisons between pre-ECMO (D-1) and right after ECMO run (D+0). The lower rows are comparisons between pre-ECMO (D-1) and 3 days after decannulation (D+3). ECMO, extracorporeal membrane oxygenation; NK, natural killer; VA, venoarterial; VV, venovenous.

In experiment 1 (D-1 vs. D+0), the numbers of classical monocytes (low-CD43 high-CD48) in the VA and VV modes decreased significantly (p < 0.0001 and p = 0.0001, respectively), but recovered after 3 days in experiment 2 (D-1 vs. D+3) (Figure 6B and E). The numbers of natural killer (NK) cells (CD161+) were initially decreased right after the ECMO run in VA (D-1 vs. D+0) but were increased late for the next 3 days in both modes (D-1 vs. D+3) (Figure 7C and F). All flow-cytometric data of innate immune cells according to the blood sampling time are summarized in Table, Supplemental Digital Content 2, https://links.lww.com/ASAIO/A534.

Figure 7.: Adaptive immune cells change. The upper rows (A, B, and C) are comparisons between pre-ECMO and right after ECMO run in experiment 1. The lower rows (D, E, and F) are comparisons between pre-ECMO and 3 days after decannulation. The helper T cell and cytotoxic T cell were both suppressed significantly in VA mode and the B cells were initially suppressed and restored on 3 days after decannulation in both modes types. ECMO, extracorporeal membrane oxygenation; VA, venoarterial; VV, venovenous.

Changes in Adaptive Immune Cells.

The numbers of helper T lymphocytes (CD4+) and cytotoxic T lymphocytes (CD8+) were significantly decreased in VA modes immediately after ECMO run (D-1 vs. D+0; p = 0.03 and 0.01, respectively). B lymphocytes (CD45R+) were also decreased significantly in both VA and VV modes p = 0.009 and 0.04, respectively, immediately after ECMO run (D-1 vs. D+0, Figure 7A–C). The numbers of helper T lymphocytes (CD4+) were still low in both modes even after 3 days of ECMO run (D-1 vs. D+3, p = 0.0002 and 0.0038, respectively). Whereas, cytotoxic T lymphocytes (CD8+) were still lower in VA modes (D-1 vs. D+3, p = 0.004) but not in VV modes. Changes of B lymphocytes (CD45R+) after 3 days of ECMO run were not significantly different in both modes (D-1 vs. D+3, Figure 7D–F). All flow-cytometric data of adaptive immune cells according to the blood sampling time are summarized in the Table, Supplemental Digital Content 2, https://links.lww.com/ASAIO/A534.

Changes in Cytokines.

IL-6 and TNF-α levels did not change significantly after ECMO runs in experiment 1. However, TNF-α increased gradually after ECMO run in both modes of experiment 2 (D-1 vs. D+3; p = 0.002 in VA mode, p = 0.0273 in VV mode), and there were distinct changes of IL-6 in only VA mode in 3 days after ECMO run during experiment 2 (D-1 vs. D+3, p = 0.002 in VA mode, p = 0.06 in VV mode). More detailed data of cytokine changes are summarized in Table, Supplemental Digital Content 3, https://links.lww.com/ASAIO/A534.

The levels of IL-6 increased gradually after ECMO circulation in VA mode in experiment 1 (p = 0.046), and then slightly decreased in 3 days after ECMO run during experiment 2 (D-1 vs. D+3, p = 0.02). Additionally, TNF-α also increased significantly after ECMO circulation in experiment 1 (D-1 vs. D+0, p = 0.05), especially in VA mode. All cytokine data are shown in the Table, Supplemental Digital Content 3, https://links.lww.com/ASAIO/A534.

Discussion

Establishment of a Rat ECMO Model

We sought to establish a stable rat model and to explore changes in immune cell numbers in both VA and VV modes of ECMO. Pigs, sheep, and dogs have been mainly used to study ECMO-related pathophysiology, but such models are costly, and require multiple trained personnel for one experiment.⁴ On the other hand, ECMO model in rodents has been considered as low cost, and a single trained experimenter can perform one experiment. Several ECMO models in rodents have already been described^5–13; these vary in terms of the cannulation sites, circuits, priming volumes, priming materials, and flow rates. However, ECMO duration was maintained for <1 hour in most of previous studies, which may be too short to observe changes in immune cell numbers. VV mode also have described in previous studies.^14,15 As VV support generally have longer ECMO duration compared with VA support in clinical settings, to have more meaningful VV ECMO model that mimics human closely, longer duration of ECMO run should be achieved in this rat VV ECMO model. Although in our pilot study, some rats survived in 4 hours of ECMO, the ideal ECMO duration was 2 hours.

To make more stable and unique models, our experiments have four characteristics: (1) less invasive instrumentation, (2) mixture of priming solution, (3) effective drainage catheter, and (4) experiments on both modes of ECMO. First, we performed less invasive way for all procedures from intubation to cannulation. The rats were orally intubated rather than tracheostomy and were peripherally cannulated with open technique, rather than central cannulation through sternotomy. Such less invasive instrumentations facilitated rat survival even after decannulation and allowed for more prolonged evaluation of immune cell changes for up to 3 days after decannulation.

Second, in our early experience, we primed the circuit with only crystalloid solution, such as phosphate-buffered saline or NS. This was associated with a high mortality within our pilot studies. Some investigators have used homologous rat blood and colloid solutions for priming of ECMO circuit. However, we believed that blood priming would have significant impact on immune response and thus decided to avoid the transfusion. Furthermore, using blood prime necessitates the sacrifice of the donor rat. Thus, we used the strategy of an albumin mixed priming by 20% (w/v) albumin to half of the priming solution (7 ml albumin solution and 7 ml NS), which was associated with a higher rate of survival. As the priming volume of 14 ml is less than half the blood volume of an adult rat (70 ml/kg),¹⁶ the hematocrit dropped significantly as the ECMO was started and cause pathologic hemodilution. In future studies, we need to overcome hemodilution by reducing the priming volume and shorten the circuit, as the hemodilution can also affect the response of immune cells. Third, to create a more effective drainage catheter, we modified a neonatal feeding tube by adding eight side holes. A “16 gauge catheter,” which was used in other rat ECMO model, may potentially cause vena cava or atrial injury with the sharp tip during cannulation. Apart from these catheters, neonatal feeding tube, which we chose to use, has a round closed tip which can prevent the injury. To reduce the priming volume and for easier handling, we cut a 5 Fr neonatal feeding tube into 10 cm length and allowed the tip to be placed in the right atrium cavity, as verified by several autopsy. Additionally, we have made eight side holes to enhance drainage which resulted in a hemodynamically stable model that could extend the ECMO duration. In future studies, we have plans to examine the hemolysis from these drainage catheters compared with standard angio-catheters.

Fourth, to mimic clinical scenarios, we formed both VA and VV ECMO modes to seek for the differences according to the modes. As described above, we found no significant difference in the ABGA findings between the VA and VV mode, except that the lactate level, which was twofold higher in the VV mode after ECMO weaning. For stressed rats under anesthesia, instrumentations and blood loss, VA ECMO effectively supported the hemodynamics and organ perfusion. In addition, the higher level of lactate in the VV mode is attributable to organ hypoperfusion caused by hemodynamic impairments during VV ECMO run, which lacked circulatory support.

Changes in Immunity

We performed flow-cytometric analyses and cytokine measurements at 3 time points: one day before ECMO, after ECMO run (just before decannulation) and 3 days after decannulation in both modes to study ECMO-induced immune dysfunction and inflammation and also to obtain baseline data for further studies. To the best of our knowledge, no prior report has addressed serial investigation of immune response in a rat ECMO model.

Granulocytes generally have been activated in response to infection, foreign bodies, hypovolemia, or stress in the bloodstream or elsewhere in the body.^17,18 In this current study, the numbers of granulocyte significantly increased in both the VA and VV modes immediately after ECMO run, and we could think as the inflammatory response is caused by the ECMO circulation itself even this is not a disease model. Millar et al.¹⁹ explained granulocytes (including neutrophil) exposed to the extracorporeal circulatory environment are known to be activated due to various causes, and this plays an important role in organ damage during ECMO support. Oxygenator may have a significant effect on neutrophil activation during ECMO run.^11,20 It was known that neutrophil function is maintained even if the number of neutrophils increases after the initiation of ECMO.^21,22

Monocytes and NK cells engage in phagocytosis, antigen presentation, and cytokine production.²³ Monocytes are also reported to be activated in response to ECMO,²⁴ however, respond more slowly than activation of neutrophils.²⁵ In our study, the monocytes and NK cells dropped significantly after ECMO run and the reason for this phenomenon can only be assumed. Levels of both monocytes and NK cells were restored at 3 days post decannulation. Further studies are required to determine the reason and the impact of lowered monocytes and NK cells after ECMO initiation.

T lymphocytes play a central role in cell-mediated immunity.²⁶ Interestingly, both helper and cytotoxic T lymphocytes were decreased significantly after ECMO run only in the VA mode but not the VV mode. Both remained lower even after 3 days of ECMO initiation in both modes. One could assume that T lymphocytes need more time to be restored after VA ECMO. Further studies are required to determine the relevance of reduced numbers of T lymphocytes during and after ECMO. B lymphocytes produce antibodies, present antigens, and secrete cytokines such as interleukins, lymphokines, and tumor necrosis factors.²⁷ The B lymphocytes decreased significantly in both the VA and VV modes after ECMO run. The differences were only short term, however, and had normalized by the 3 days after decannulation.

Although, cautious interpretation is needed when we correlate these results to clinical setting, we believe that this study revealed valuable insights on the trend and changes of cellular immunity in both human and animal models in response to ECMO support. Furthermore, these data will guide us as a baseline for our further studies with disease models.

There are several limitations in our study, which we need to overcome in future studies. First, as the rats could not cope with multiple blood samples (2.5cc for 3 times) and with ECMO run (14 ml of priming volume), we had to divide the rats into smaller subgroups, which resulted in small sample size study. Second, we only observed for a short duration which was 2 hours of ECMO run, but this can be extended to 4–5 hours, which we successfully extended during our more recent pilot studies. Third, ECMO was instituted in healthy rats that was done to determine the effect of ECMO only, rather than the effect of ECMO and acute illness. Fourth, no functional studies of immune cells were performed, and we only evaluated the distribution of each immune cells. We plan to proceed with functional immune cell analysis in future studies. Fourth, only the distribution of each immune cell was evaluated without functional studies of immune cells which we will be performing in future studies. Finally, the immune cells of rats may differ from human or larger mammalian studies, but these results provide baseline data and as first step in understanding changes in specific immune cells in further disease model studies.

We conclude that our ECMO model featuring a novel drainage catheter and an albumin-based priming solution may allow prolonged observation of ECMO in both VA and VV modes. We confirmed that overall immune proportion changed after ECMO initiation in both VA and VV modes, and the immunologic balance altered more significantly in the VA mode than in VV mode, and restoration of immune cell distribution was faster in VV mode than VA mode. Our ECMO model in the rats may be feasible for the hemodynamic and immunologic research, and further long-term evaluation is needed in the disease-specific models.