Introduction

Tight fluid balance regulation is crucial in patients receiving extracorporeal membrane oxygenation (ECMO) support. Fluid overload is a significant risk factor for early mortality among critically ill patients receiving ECMO1. Consequently, a method for precisely assessing body fluid status is essential to ensure optimal hydration levels.

For patients receiving ECMO, various methods are employed to evaluate the body fluid status; nevertheless, no single, independent measure exists to accurately determine hydration levels. Typically, an integrated approach is used, considering the patient’s vital signs, daily fluid balance, weight fluctuations, central venous pressure, and pulse wave velocity2,3. However, these assessments may lack precision and are subject to inter-evaluator variability, which can undermine the consistency of therapeutic interventions. Moreover, assessing body water status in patients undergoing ECMO is more complex than in those not receiving ECMO because of factors such as hemorrhage, frequent bleeding at ECMO insertion sites, and additional fluid requirements needed for maintaining the ECMO circuit volume.

Bioelectrical impedance analysis (BIA) is being utilized as a valuable tool for assessing body fluid volumes. This method is simple, easy to employ, noninvasive, and cost-effective, allowing for an accurate assessment of body fluid volumes in a wide range of patients – healthy and critically ill4,5,6,7,8. However, BIA’s utility in measuring body fluid volumes in patients receiving ECMO has not yet been established. Also, BIA measurements in ECMO patients may differ from non-ECMO individuals due to factors such as systemic inflammatory response and hemodilution effect.

We have developed a rat ECMO model to monitor changes in BIA with fluid administration. This study aimed to investigate whether BIA effectively reflects changes in body fluid status associated with fluid infusion during ECMO support.

Methods

Ethical statement

All animals used in this study received care per the Principles of Laboratory Animal Care as outlined by the National Society for Animal Research and the Guide for the Care and Use of Laboratory Animals. The study received ethical approval from the Ethics Committee of Chonnam National University Medical School (approval number: CNUH-IACUC-24013). Furthermore, all methods are reported in accordance with the ARRIVE guidelines to ensure comprehensive and transparent reporting of animal research.

Animal preparation

We utilized 26 male Sprague–Dawley rats, each weighing 400 g, sourced from Samtako Bio Korea Co., Ltd in Osan City, Korea. These rats were accommodated in an environment with a controlled temperature (20 °C) and subjected to a 12-h light–dark cycle, with unrestricted access to standard food and water.

All animals were anesthetized with an intramuscular injection of ketamine (80 mg/kg) and xylazine (8 mg/kg), followed by the administration of isoflurane inhalation to maintain anesthesia throughout the surgical procedures. Each rat was placed in the supine position on a surgical table with heating pads to prevent hypothermia. After induction of anesthesia, each rat was intubated with a 16-gauge catheter. Mechanical ventilation was provided with 90% (v/v) oxygen and 2.5–2.0% (v/v) isoflurane via a rodent respirator (Harvard Apparatus Inc., Holliston, MA, USA). The ventilation settings were as follows: tidal volume, 8 mL/kg; respiration rate, 55 breaths/min. The left femoral artery was cannulated with a 24-gauge catheter to monitor systemic arterial blood pressure, and the right femoral vein was cannulated for the intravenous administration of fluids. All experimental rats were euthanized by CO₂ asphyxiation at the end of the study.

Establishment of the rat ECMO model

We developed two ECMO modes in a rat model: venoarterial (VA, n = 13) and venovenous (VV, n = 13; Fig. 1). In the VA configuration, the drainage cannula was inserted into the right external jugular vein, and the perfusion cannula into the right common carotid artery (Fig. 2(1). Conversely, in the VV setup, the drainage and perfusion cannulas were placed in the right and left external jugular veins, respectively. A 24-gauge angiocatheter served as the perfusion catheter. For drainage, we employed a custom-made catheter crafted from a neonatal feeding tube (5-French, 50 cm, with a functional length of 10 cm) that included eight lateral holes to improve perfusion efficiency. This catheter was positioned at the junction of the right atrium and superior vena cava (Fig. 2(2). Right after inserting the drainage cannula, 500 IU/kg of heparin sodium was administered. Continuous heparin infusion was not administered, as it was considered a potential confounding factor that could affect the interpretation of BIA parameter changes following fluid administration. The ECMO circuit, comprising an oxygenator (Micro-1; Senko Medical Instrument Mfg. Co., Ltd., Tokyo, Japan), peristaltic pump (Watson-Marlow Pumps; Falmouth, Cornwall, United Kingdom), and interconnected tubing, had configurations consistent with those established and reported in earlier experimental studies9. Each ECMO circuit was primed with a total of 14 mL of solution, consisting of 7 mL of 20% (w/v) albumin and 7 mL of normal saline. The priming volume was identical across all rat models.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Rat ECMO model diagram. BIA, bioelectrical impedance analysis; ECMO, extracorporeal membrane oxygenation.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Venoarterial (1) and venovenous (2) rat ECMO models with tetrapolar electrodes for BIA. A, Fluid injection line at the right femoral vein; B, A-line monitor at the left femoral artery; C, Right external jugular vein for drainage; D(a), Right common carotid artery (A), D(b) left external jugular vein (B) for perfusion; E, Bipolar electrodes for BIA at the upper trunk; F, Bipolar electrodes for BIA at the lower trunk. BIA, bioelectrical impedance analysis; ECMO, extracorporeal membrane oxygenation.

Rat ECMO model experimental procedures

ECMO lasted for 120 min, with a flow rate maintained at 23–27 mL/min (30 rpm), 1 mL of fluid was administered six times: pre-cannulation, post-cannulation, thrice during the ECMO maintenance phase, and post-ECMO cessation. These administrations were spaced every 30 min throughout the 120-min ECMO support period, including before and after ECMO. The infusion solution consisted of a mixture containing one-third albumin and two-thirds normal saline (Fig. 3). When the mixture is prepared as above, the resulting osmolality is estimated at 300 mOsm/L, which falls within the osmolality range of blood, classifying it as an isotonic solution. Reportedly, using a mixture of albumin and normal saline for the priming volume in an ECMO rat model is associated with a higher survival rate than solely using crystalloid solutions9.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Fluid infusion timeline. ECMO, extracorporeal membrane oxygenation.

BIA

InBody M20 (Inbody Co., Seoul, Korea) was employed to assess the body water status. The rat ECMO models were positioned supine, with their tails extended distally. Tetrapolar electrodes were attached to the upper and lower trunk of rats and connected to InBody M20. Specifically, four 1 cm x 26-gauge needles were inserted 5 mm subcutaneously. The insertion sites were as follows: two needles were placed 1 cm to the right of the ventral midline at the upper trunk level (aligned with the upper limb), and two needles were placed 1 cm to the left of the ventral midline at the lower trunk (aligned with the iliac crest). Electrode clamps were then attached to each needle according to the manufacturer’s instructions, with red clamps for the current-injecting electrodes and black clamps for the voltage-sensing electrodes. The distance between the two sets of needle electrodes (upper and lower trunk) was 16 cm for all models (Fig. 2). This device primarily measures four parameters: impedance (Z), resistance (R), reactance (Xc), and phase angle (θ) across three distinct frequencies: 5, 50, and 250 kHz. Impedance (Z) is calculated through vector analysis of resistance (R) and reactance (Xc). As electrical current is transmitted through bodily tissues, resistance (R) is encountered, which varies according to body composition, including water status, opposing the current flow. Reactance (Xc) represents the opposition to changes in the current. Consequently, as the current traverses the cell membrane, reactance (Xc) induces a time delay, resulting in a phase shift between voltage and current, from which the phase angle (θ) is derived10.

To summarize what these measurements reflect in the condition of the rats, resistance at infinite frequency (Rinf), resistance at zero frequency (R0), and Intracellular resistance (Ri) are inversely related to total body water (TBW), extracellular water (ECW), and intracellular water (ICW), respectively. The phase angle is a measure associated with cell membrane integrity. Thus, a decrease in Rinf, R0, and Ri would indicate an increase in TBW, ECW, and ICW in the rats, and a higher phase angle is generally associated with better cell membrane integrity11,12.

BIA was performed at one-minute intervals using the InBody M20 throughout the entire period from before ECMO cannulation to its termination. From the continuous data obtained, eight predefined time points were selected for analysis of the corresponding BIA parameters. Each ECMO model, corresponding to the same time points, was classified into time point 1 through 8. This classification and the measurement process are detailed in Fig. 3. The median values of the measurements for each time point were calculated.

Statistical analysis

To examine changes in extracellular and intracellular fluid volumes over time with the administration of fluids, the Friedman rank sum test was employed comparing the median values across time points. We conducted a post-hoc analysis using the Wilcoxon signed-rank test to determine which time points showed significant differences, following the significant result from the Friedman test. To account for the multiple comparisons that arise when comparing several pairs, we adjusted the p-values using the Bonferroni correction method. To investigate how the fluid administration at regular intervals during ECMO support affected these fluid volumes and whether the ECMO configurations influenced these changes, a linear mixed-effects model analysis was conducted. Statistical significance was set at P < 0.05. All statistical analyses were conducted using R version 4.2.2 (R Core Team, 2020).

Results

From the original cohort of 26 rats receiving ECMO, 3 succumbed during the experiment: 1 in the VV mode and 2 in the VA mode. These incidents necessitated their exclusion from the final data analysis. Thus, findings from 23 rats were evaluated.

Overall changes in BIA parameters

The changes in all BIA parameters (Rinf, R0, Ri, and phase angle) across the time points were found to be significant. (Rinf: Friedman chi-squared = 50.146, df = 7, p < 0.001; Ri: Friedman chi-squared = 39.536, df = 7, p < 0.001; R0: Friedman chi-squared = 82.464, df = 7, p < 0.001; phase angle: Friedman chi-squared = 22.679, df = 7, p = 0.002) (Table 1). Further post-hoc analysis is included in supplementary material 1.

Table 1 Changes in BIA parameters using Friedman rank sum test.
Full size table

Changes in BIA parameters before ECMO initiation and after ECMO termination

Before ECMO initiation, Rinf, R0, and phase angle showed a decreasing trend, but Ri displayed variability. Following ECMO discontinuation, Rinf, R0, and Ri decreased. Additionally, an increase in phase angle from time points 7 to 8 was observed (Table 1) (Fig. 4).

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

Changes in Rinf (A), R0 (B), Ri (C), and the phase angle (D).

Changes in BIA parameters during ECMO support

During ECMO support, all resistance parameters, including Rinf, R0, and Ri, showed a decreasing trend. The phase angle consistently increased, with the rise immediately following ECMO initiation (between time points 4 and 5) (Table 1) (Fig. 4).

Between time points 4 and 7, Rinf exhibited a significant decrease associated with fluid accumulation during ECMO support (T = − 5.438, P < 0.001). However, mode of ECMO (whether VA or VV ECMO) did not significantly influence Rinf (Table 2) (Fig. 5). R0 demonstrated a significant decrease related to fluid accumulation during ECMO support (T = − 3.407, P = 0.002). Different mode of ECMO did not significantly affect R0, as shown in Table 3. Similarly, fluid accumulation over time had a significantly negative effect on Ri (P < 0.001), with this decrease being unaffected by ECMO configuration variations. Notably, a significant interaction between fluid accumulation over time and ECMO configuration on Ri was found, detailed in Table 4. Furthermore, the phase angle showed a significant increase associated with fluid accumulation during ECMO support (T = 2.436, P = 0.023). Alterations in ECMO configuration did not significantly affect the phase angle, as reported in Table 5.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Effect of fluid accumulation by time on Rinf changes. ECMO, extracorporeal membrane oxygenation; VA, venoarterial; VV, venovenous.

Table 2 Linear Mixed-effects model showing the effects of fluid accumulation by time on Rinf changes during ECMO Support.
Full size table
Table 3 Linear Mixed-effects model showing the effects of fluid accumulation by time on R0 changes during ECMO Support.
Full size table
Table 4 Linear Mixed-effects model showing the effects of fluid accumulation by time on Ri changes during ECMO Support.
Full size table
Table 5 Linear Mixed-effects model showing the effects of fluid accumulation by time on phase angle changes during ECMO Support.
Full size table

Discussion

Our research validates BIA could be an effective tool for assessing changes in body water status and distribution during ECMO support. In our study, we utilized a rat ECMO model and administered consistent volumes of fluids intravenously at regular intervals. We observed a significant reduction in Rinf, indicative of changes in TBW. By employing a linear mixed-effects model, we confirmed that Rinf decreased significantly over time with regular fluid administration. These findings are consistent with previous BIA studies on healthy human participants, where fluid infusion was associated with a substantial reduction in electrical resistance13. Moreover, our experiment enabled monitoring changes in body water distribution, as evidenced by Rinf, Ri, and R0 measurements, corresponding to TBW, ICW, and ECW, respectively. Consequently, BIA has proven to be a valuable method for real-time monitoring of body water content, including its distribution, during the application of ECMO.

During ECMO support, isotonic fluid infusion was observed to increase ICW – a finding that contrasts with typical expectations. Normally, isotonic fluid, which matches the osmotic pressure of body fluids, predominantly augments ECW without altering intracellular volumes14. This unusual increase might stem from an inflammatory response induced by ECMO, particularly due to prolonged contact with the synthetic surface used in ECMO, which can activate phagocytes and affect endothelial cells15. Inflammation typically causes edema by increasing ECW, specifically within interstitial tissues. However, ECMO support may compromise cell membrane integrity through both endothelial and cellular membrane dysfunction, ultimately leading to cellular edema. In particular, inflammatory cytokines released during ECMO have been shown to downregulate aquaporin expression on the cell membrane, thereby reducing water efflux and promoting intracellular water retention16. Additionally, these cytokines may inhibit Na⁺/K⁺ ATPase activity, resulting in intracellular sodium accumulation, osmotic imbalance, and water influx. Although no studies have directly demonstrated ECMO-induced inhibition of Na⁺/K⁺ ATPase activity or the development of cellular edema, it is well established that exposure of blood to the extracorporeal circuit triggers the release of inflammatory cytokines, and elevated serum levels of these cytokines have been observed in patients undergoing ECMO support17,18. Furthermore, in vitro studies have shown that inflammatory mediators such as TNF-α can reduce Na⁺/K⁺ ATPase activity19 and the inevitable hemodilution associated with extracorporeal life support has been reported to activate neutrophils and contribute to the development of systemic inflammatory response syndrome16,20. Our findings are supported by an increased phase angle during ECMO – a parameter positively related to cell membrane integrity. A higher phase angle is generally associated with improved cell membrane integrity. However, since there is no consensus on an appropriate cut-off value for phase angle, it is difficult to conclude that an excessively high phase angle always indicates better cell membrane integrity. Previous studies have reported that the normal range of the phase angle in healthy adults is 5.82–6.86, while in porcine models, it is 6.25–7.021,22. Based on these references, in our study, the phase angle during ECMO application exceeded 10.5 and showed a rising trend afterward. This suggests that ECMO application may have influenced cell membrane integrity, leading to significant changes in the phase angle. However, since the normal reference range for phase angle in rats has not yet been established, further research is needed to draw definitive conclusions. Contrarily, previous studies on baboons did not show significant changes in body water content or distribution under ECMO, likely because of differences in experimental protocols. Those studies applied VA ECMO for approximately 8 h without regular fluid infusions, unlike our approach23. Future investigations could benefit from continuously monitoring BIA at time points with and without fluid infusions during extended ECMO support to delineate the specific effects of ECMO on body water dynamics more clearly.

Shortly after initiating ECMO, between time points 3 and 4, significant increases were observed in Rinf, Ri, and R0, reflecting substantial reductions in TBW, ECW, and ICW, respectively. Although fluid was administered twice before starting ECMO, it is likely that the reductions in intravascular volume due to cannulation-related bleeding was more pronounced. This factor might have contributed to the observed decreases in ECW and TBW. While the possibility of an ICW influx due to ECMO-induced changes cannot be entirely excluded, time point 4 reflects a measurement taken immediately after ECMO initiation. Therefore, it is unlikely that sufficient time had elapsed for significant transmembrane water shifts to occur. The observed decrease in ICW is more plausibly explained by the loss of red blood cells secondary to cannulation site bleeding.

To the best of our knowledge, this study is the first to employ BIA to continuously monitor in real-time parameters indicative of TBW content and distribution during ECMO in a rat model. This method enabled the detailed analysis of trends associated with these parameters under ECMO, enhancing our comprehension of how fluid distribution is altered when fluids are administered during ECMO support. Unlike previous research, which primarily involved patients or healthy human participants not receiving ECMO, our experiment stands out as both unique and pioneering13,24. BIA measurements can be affected by demographic factors such as ethnicity, age, and sex, necessitating careful interpretation of the results of prior studies11,25,26. By utilizing a consistent rat model, our research minimized confounding variables, thereby bolstering the validity of our findings and providing a clearer insight into the effects of ECMO on body fluid dynamics.

Our study has some limitations. First, the use of small animals such as rats introduces potential artifacts in BIA parameter results due to their susceptibility to minor stimuli. When preparing our rat ECMO model, even slight disturbances during electrode attachment and ECMO cannulation caused significant parameter fluctuations. Despite the challenges of setting up the ECMO model, we attempted to mitigate these errors using a relatively large number of animals (n = 26). Second, our results were reflective of resistance changes, rather than absolute values of TBW, ECW, and ICW. While formulas exist to calculate these absolute values, they do not apply to rats, which limits the interpretation of our data. Third, the duration of BIA parameter measurement was relatively short, lacking information on fluid distribution and changes during prolonged ECMO application. Although our team has extensive experience with the ECMO rat model and was the first to implement BIA in this context, maintaining ECMO for an extended duration in the rat model remains challenging. The non-ECMO period was also brief, making it difficult to compare fluid-induced changes during ECMO application adequately. Fourth, we acknowledge that fluid balance at each time point, including urine output, was not calculated and presented alongside BIA parameters. Future studies should include a control group of rats without ECMO support and, ideally, extend the duration of BIA measurements before, during, and after ECMO application to analyze changes in body water content and distribution thoroughly.

Conclusion

In conclusion, our study utilizing a rat ECMO model has demonstrated that BIA could be a potent method for monitoring changes in body water status and distribution during ECMO application. Independent of ECMO configuration, fluid infusion was associated with decreases in Rinf, R0, and Ri along with an increase in the phase angle. As BIA is a noninvasive, cost-effective, and simple method, its application in clinical settings may offer potential advantages. Thus, the application of BIA for volume management in clinical settings involving patients receiving ECMO could be considered a promising approach, warranting further clinical investigation.