Extracorporeal circulation impairs microcirculation perfusion and organ function

Abstract

Extracorporeal membrane oxygenation (ECMO) is a procedure used to aid respiratory function in critical patients, involving extracorporeal circulation (ECC) of blood. There is a limited number of studies quantifying the hemodynamic effects of ECC procedures on the microcirculation. We sought to mimic veno-arterial-ECMO flow conditions by use of a scaled-down circuit primed with either lactate Ringer (LR) or 5% human serum albumin (HSA). The circuit was first tested using benchtop runs with blood, and subsequently used for in vivo experiments in Golden Syrian hamsters instrumented with a dorsal window chamber to allow for quantification of microvascular hemodynamics and functional capillary density (FCD). Results showed significant impairment in FCD, and a reduction of arteriolar and venular blood flow, with HSA providing significant higher blood flows and FCD compared with LR. Changes in hematocrit and RBC labeling after ECC reflected a shift in plasma volume, which may stem from a loss in intravascular oncotic pressure due to priming fluids. The distribution of hemoglobin oxygen saturation in the microvasculature showed a significant decrease in venules after ECC. In addition, major organs such as the kidney and heart showed increases in both inflammatory and damage markers. These results suggest that ECC impairs microvasculature function and promotes ischemia and hypoxia in the tissues, which can be vital to understanding comorbid clinical outcomes from ECC procedures such as acute kidney injury and multiorgan dysfunction.

NEW & NOTEWORTHY ECC reduces microvascular perfusion, with no full recovery 24 h after ECC. HSA performed better as compared with LR in terms of FCD and venule flow, as well as venule oxygen saturation. Increases in inflammatory and damage markers in key organs were observed within all organs analyzed.

INTRODUCTION

According to the Extracorporeal Life Support Organization (ELSO, Ann Arbor, MI) Registry 2021 International Summary, the mortality rate from extracorporeal membrane oxygenation (ECMO) procedures since 1990 is more than 30% (1). Prior research investigating this high rate has pointed to development of coagulopathies, hemorrhages, systemic inflammatory responses, and multiorgan dysfunction. A third of complications reported are hemorrhages (2), which mostly occur at cannulation sites; balancing bleeding issues with coagulopathies (3, 4) makes anticoagulant management one of the main concerns. Beyond maintaining hemostasis, inflammation due to foreign surface contact increases cytokine and activated neutrophil levels (5, 6). This inflammatory response permeates, resulting in multiorgan failures (7) that are especially prevalent in the kidneys and brain, with rates as high as 50% and 13% respectively in adults (8). Despite these observed downsides, for many critical patients, it remains their only hope of survival.

ECMO is utilized to provide lung and, in some cases, cardiac support to treat underlying pathophysiological conditions or for acute recovery (e.g., failure to wean from open-heart surgery, congenital defects requiring reconstructive surgeries, and acute lung injury) (9). Setup of the ECMO circuit can vary, but the chief components remain the same. Blood is drained from a venous cannula and supplies either a peristaltic or centrifugal pump, driving fluid into an oxygenator. The semipermeable hollow fiber membrane housed inside the oxygenator allows for oxygen (O₂) diffusion into blood and carbon dioxide (CO₂) removal, which is the chief goal of the circuit. A heat exchanger is used to warm the blood back to body temperature, before being reinfused in either the arterial (VA for cardiac support) or venous (VV without cardiac support) side (10). Once blood is reinfused, it travels through the arterial highway until it reaches small arterioles and capillaries, where the oxygen is delivered to parenchymal cells in organs that use it to undergo aerobic respiration to perform critical functions (11). Without oxygen, tissues become hypoxic by one of the following three main methods: 1) hypoxemia, 2) abnormal hemodynamics, or 3) ischemia. All three routes lead to the tissues succumbing to anaerobic metabolism, which has significant ramifications. Glycolysis can occur, but the resultant pyruvate undergoes lactic acid fermentation; coupled with reduced or complete cessation of blood flow from ischemia, this leads to the buildup of waste products over time, making the process toxic and unsustainable (12, 13) especially in vital organs such as the heart, kidney, and brain. Therefore, in order for the oxygenation occurring within the ECC circuit to be effective and have an impact, O₂ needs to be delivered and offloaded to the tissues. Since exchange of gases and metabolites occurs at the microcirculation level (11), it is imperative to understand the implications of extracorporeal circulation (ECC) on microcirculatory flow.

Prior work in large mammal and human studies investigating the microcirculation and ECC have used sublingual imaging and technologies such as orthogonal polarization spectral, sidestream dark field, and laser Doppler systems (14). However, these methods are limited since they rely heavily on quality of image acquisition, vary based on applied pressure, cannot directly quantify flow, and only provide semiquantitative scaled measures/rankings (15). Small rodent models have the advantage of being less expensive, require less volumes, and have been developed to mimic ECC conditions (16); however, there are a limited number of models that can also analyze the microvascular changes. One such model was utilized by Kamler et al. to investigate the effects of ECC on microvascular inflammation and flow (17–20). Flow was shunted between the arterial side to the venous side using implanted catheters, while the microvasculature was examined using a dorsal skinfold window chamber implanted on Golden Syrian hamsters. These studies concluded that ECC increases oxygen-free radical within the microcirculation (21) and venular leukocyte rolling in preparations devoid of inflammation (22), which increased with time (21). However, the flow direction used in these studies are arterial-to-venous, which is retrograde compared with the direction used in cardiopulmonary bypass as well as both traditional modes of ECMO.

Therefore, the goal of this study was to analyze the changes in the perfusion of microvessels as a result of the hemodynamics due to a scaled-down ECC circuit to mimic VA-ECMO conditions. Intravital microscopy was utilized to directly quantify the hemodynamic changes due to ECC, while monitoring changes in localized oxygen saturation.

METHODS

The first part of this study was to characterize the circuit consisting of a peristaltic pump and a bubble trap with ex vivo benchtop testing (Fig. 1) to see the effects on blood. The second and central part was to quantify the impact on the microcirculation in vivo (Fig. 3) when the circuit was primed with lactate Ringer (LR), or 5% human serum albumin (HSA) as compared with healthy Sham control animal.

Figure 1.

Schematics for the ex vivo benchtop tests. PE50, polyethylene 50 tubing; 3-FR PU, 3 French polyurethane tubing.

Ex Vivo Benchtop Testing

Donor male Golden Syrian hamsters were anesthetized using 2.5% inhaled isoflurane. Blood was drained via cardiac puncture using a 22-gauge needle and subsequently anticoagulated by adding 0.72 IU heparin/mL blood. Blood (4 mL) was put into a separate vial, while the rest was placed on a tube rotator. The vial was connected to the circuit (Fig. 1), and primed with blood, making sure that all bubbles were flushed out via the vent line in the bubble trap. Flow was ramped up for 15 min, held at 2 milliliters per minute for 1 h, and ramped down for 15 min. Samples were taken via the vent line at 0, 15, 45, 75, and 90 min; the line was flushed with heparinized saline (concentration of 30 IU/mL) in between samples to prevent coagulation. Blood constitution was analyzed with 80 μL of blood using a blood gas analyzer, whereas hematocrit (Hct) and plasma hemoglobin (pHb) levels were measured via centrifugation and a Hb 201+ machine (see Supplemental Table S1 for details on equipment; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.c.5672074).

In Vivo Study

Surgical preparation.

Healthy male Golden Syrian hamsters were purchased from Charles River Laboratory, kept in a 12-h light-dark cycle, and provided free access to water and ENVIGO Teklad 8406 rodent diet. Animals were cared for in line with the NIH Guide for the Care and Use of Animals and all experiments were approved by the University of California, San Diego (UCSD) Institute Animal Care and Use Committee (IACUC). Hamsters weighing 60–75 g were first instrumented with a dorsal skinfold window chamber under subcutaneous ketamine-xylazine (200 mg/kg and 10 mg/kg) anesthesia, as has been described earlier (23). To recapitulate, the animal’s fur is removed, and dorsal skin stretched before the top layers of skin and adipose tissue are removed, revealing the underlying muscle, connective tissue, and microvasculature. A titanium frame is sutured to the dorsal skinfold, saline layer applied to the preparation, and a glass coverslip measuring 1.2 cm in diameter is secured to the frame to allow for intravital microscopy. To keep track of systemic changes, a PE50 catheter with a PE10 tip (via silicone) was implanted the next day into the animal’s right carotid artery, exteriorized between the shoulder blades and through the anterior side of the dorsal skinfold. Animals were allowed to recover for 24–48 h before starting any experimental procedure. During the experiment, the hamster was placed in a restraining tube without any anesthetic, and all measurements were taken in the awake, unanesthetized state; the one exception being the ECC procedure, during which the hamster was under isoflurane anesthetic and on a surgical stage. After the experimental procedures were finished, animals were euthanized with an overdose of Euthasol (pentobarbital sodium) at a dosage of 300 mg/kg. Key organs such as the spleen, kidney, liver, heart, and lung were collected along with plasma and, if present, urine for analysis by the UC San Diego Histology Core (see Supplemental Table S2 for commercial kits used). A timeline of the surgeries and experiment is provided in Fig. 2.

Figure 2.

Timeline of the in vivo study. Note that day 1 and day 2 surgeries were done under subcutaneous ketamine-xylazine anesthetic, whereas extracorporeal circulation (ECC) was conducted under inhaled isoflurane anesthesia. BL, baseline; CO, cardiac output.

Systemic parameters.

Mean arterial pressure (MAP) was measured by connecting the arterial catheter to the MP150 differential pressure transducer, feeding into a BIOPAC data acquisition board. Heart rate (HR) was calculated from the pressure signal using AcqKnowledge software. Blood constitution was measured as previously mentioned (see Ex Vivo Benchtop Testing).

Microcirculation flow.

Using an Olympus microscope fitted with a ×40 water immersion lens, the dorsal skinfold microcirculation (area of 0.46 cm²) was visualized to quantify changes in flow. The first step was to map the window chamber, by finding at least five arterioles and five venules per preparation. Functional capillary density (FCD) was then measured by counting the capillaries with RBC flow along two lines in the chamber. Finally, diameter and velocity were measured using techniques previously used (24, 25). Diameter was measured using an image shearing device, while velocity was measured using a two-photodiode system coupled with a correlogram. The photodiodes captured the shadows of blood cells as light shone through the preparation. The cross-correlation between the two photodiode signals was calculated using the correlogram, and the voltage output was converted to velocity, based on the phase shift of the signal. Volumetric flow rate through each vessel was subsequently calculated from both these measurements.

Baseline acceptance criteria.

All experiments were done at least one day after surgery. Preparations in which there was low visibility, inflammation, air bubbles were immediately excluded from the study. To be considered for the study, the following conditions at baseline must have been met for each hamster: 1) MAP = 130–100 mmHg, 2) HR = 400–500 beats/min, 3) Hct > 40%.

Hyperspectral imaging.

After flow was characterized via intravital microscopy, a hyperspectral camera was used to obtain oxygen saturation distribution within the chamber. Specifics about the technique have been outlined elsewhere (26). At each timepoint, a Pika-L hyperspectral camera mounted with a VariMagTL telecentric lens was used to take a 2,900 × 900 pixel image, where each pixel contains 300 wavelength bands that range from 390 to 1,020 nm. Each image was then truncated, filtered twice (mean and Savitzky-Golay filters), mask applied, and saturation calculated based on reference deoxyhemoglobin and oxyhemoglobin spectra. Within each vessel, the mean saturation of a selected region was calculated and recorded.

Technetium-99m RBC labeling.

To determine whether there is significant loss of whole blood after the ECC procedure, RBCs were labeled in two animals for both experimental groups (LR and HSA) using the following in vivo method. UltraTag RBC solution (100 μL) was injected into the arterial line before the start of ECC. Blood (40 μL) was taken as a baseline sample 15 min after the injection and subsequent samples of the same volume were taken at 2 h and 24 h after the ECC procedure was finished. Samples were quantified for radioactivity on a Cobra II γ counter (Packard Instrument Co., Meriden, CT) with a signal-to-noise ratio of at least 10 at the same time so counts reported are independent of sample time and only representative of the still-circulating radiolabeled RBCs.

Extracorporeal circulation.

All hamsters were randomly assigned to one of three groups before baseline: Sham, lactate Ringer (LR), or 5% human serum albumin (HSA). Healthy Sham animals served as a control and underwent the surgery, catheterization, heparinization, 1.5 h of isoflurane, but did not receive extracorporeal circulation (ECC); both experimental groups (LR and HSA) underwent the full ECC procedure (Fig. 3). After baseline measurements were taken, extracorporeal circulation was initiated. Hamsters were anesthetized with 5% isoflurane flowing through an induction chamber. The animal was then placed prone with his nose inside the surgical station nosecone, isoflurane turned down to 2.5%, and eye lubricant applied to both eyes. The hamster was flipped supine on a custom surgical stage with a central, vertical slit to account for the presence of the chamber window. After disinfecting the neck, a lateral incision was made, left jugular vein isolated, and tied off distally. The vessel was clamped, a small slit made, and the drainage catheter was filled with heparinized saline (30 IU/mL). The drainage catheter (3-Fr PU catheter with 3 fenestrations in the distal 0.3–0.5 cm and total length of 30 cm) was inserted to a depth of 2–2.2 cm. Verification of blood flow was made qualitatively by pulling on the syringe and checking ease of back flow before fixing the position. To prevent coagulation, a bolus of 50 IU heparin/mL solution at a dose of 0.05 IU/g animal weight was provided into the exteriorized carotid artery catheter, which served as the infusion cannula. To preserve the animal core body temperature during ECC, the animal was covered with a heating blanket connected to a water recirculation pump set at 37°C to reduce the effect of heat loss during the procedure.

Figure 3.

Schematics for the in vivo study. PE50, polyethylene 50 tubing; 3-FR PU, 3-French polyurethane tubing.

The circuit was primed with either LR or HSA at a flowrate of 0.1 milliliters per minute to minimize bubble formation and any bubbles were flushed out of the vent port within the bubble trap before the two cannulas were attached to the circuit. The flow rate was ramped up for 15 min to 15% of the animal’s cardiac output (CO), maintained for 1 h, then ramped down. CO was determined based on previous work on hamster cardiac index via thermodilution technique (27). Sham animals were used to compare the effects of isoflurane and the surgery on microcirculatory changes.

After 1.5 h of ECC, the pump was stopped, and catheters were flushed after being disconnected and removed from the circuit. The drainage catheter was removed, and the vein was sutured shut. The surgical site was closed, and a cautery was used to prevent microbleeds at the neck. A subcutaneous bolus of 0.02 mL buprenorphine as analgesic and 1 mL of saline to help account for fluid loss was provided and the hamster was removed from isoflurane anesthetic. To aid recovery, the hamster and his cage was placed on a heating blanket and wet food was provided. Ample time (at least 30 min) was provided for the hamster to awake from the anesthetic plane, and subsequent measurements were taken at 1, 2, and 24 h after the completion of ECC.

Statistical analysis.

Sample size was estimated based on α = 0.05 and 1-β = 0.9 for microcirculatory flow measurements. Microvascular perfusion measurements (FCD, velocity, diameter, and flow) are normalized to baseline. Extreme values were identified as outside three times the interquartile range (IQR) from Q1 or Q3 and were subsequently removed. Data are presented as box-and-whisker plots based on results from Tukey’s honest significance difference analysis. Statistically significant results are denoted for each figure if P < 0.05. All statistical analyses were completed using RStudio Version 1.3.1093 with R Version 4.0.3.

RESULTS

Ex Vivo Benchtop Testing

Evaluation of serum markers after ex vivo circulation of whole blood revealed a statistically significant increase in interleukin-6 (IL-6) and interleukin-10 (IL-10) after 90 min ECC (Fig. 4). Ex vivo circulation of whole blood for 90 min also showed that clotting factors, such as fibrinogen, soluble glycoprotein V, P-selectin, and platelet factor 4 were all statistically significantly decreased after 90 min ECC (Fig. 5).

Figure 4.

Changes in inflammatory markers from ex vivo whole blood recirculation for 90 min. A: interleukin-6; B: interleukin-10. ****P < 0.0001 between groups. For both pre-extracorporeal circulation (ECC) and post-ECC, n = 6.

Figure 5.

Changes in clotting factors over 90 min of ex vivo whole blood recirculation. A: fibrinogen; B: soluble glycoprotein V; C: P-selectin; D: platelet factor 4. ****P < 0.0001 between groups. For both pre-extracorporeal circulation (ECC) and post-ECC, n = 6.

Hematological changes during the ex vivo ECC are summarized in Supplemental Table S3. Total hemoglobin (tHb) and Hct were significantly lower at 75 and 90 min during ECC compared with before ECC, while pHb was significantly higher at 45, 75, and 90 min compared with before ECC. The only significant changes in electrolytes were seen in calcium and chloride. Calcium levels decreased as early as 15 min during ECC, whereas chloride levels increased as early as 45 min during ECC. Blood glucose was statistically significantly reduced after 90 min of ECC compared with before ECC.

In Vivo Implications of ECC

A total of 18 animals were randomly assigned to three groups: Sham (n = 6), lactate Ringer (LR) (n = 6), and 5% human serum albumin in saline (HSA) (n = 6). Measurements were taken at baseline (BL), 1 h after ECC (1Hr), 2 h after ECC (2Hr), and 24 h after ECC (24Hr), to record acute (1Hr and 2Hr) and chronic (24Hr) responses to ECC. All animals survived the ECC procedure and for over 24 h after ECC.

Systemic parameters.

The MAP and HR before and after the ECC are summarized in Fig. 6. All groups had lower MAP as compared with BL at 2-h post-ECC, and the MAP was recovered at 24 h. Interestingly, MAP was significantly lower in LR as compared with Sham at 2 h and 24 h. HR was lowered in both experimental groups at 1-h post-ECC compared with BL; however, only LR remained statistically lower at 2-h post-ECC compared with BL. HR was lowered in HSA and LR at 1-h post-ECC compared with Sham, with only LR remaining statistically lower than Sham at 2 h. There was no statistical significance at 24-h post-ECC in HR between groups or compared with BL.

Figure 6.

Systemic parameters. A: mean arterial pressure; B: heart rate. *P < 0.05; B: **P < 0.01 between groups, at each timepoint. †P < 0.05 compared with baseline (BL), within each group. n = 6 hamsters for all groups.

Electrolytes and hematological parameters.

Hematology measurements of samples taken from the carotid artery catheter are summarized in Table 1. The changes in electrolytes observed ex vivo were partially observed at 24-h post-ECC during the in vivo experiments. Chloride was statistically higher in both experimental groups compared with Sham at 1 h and 2-h post-ECC and were increased from BL chloride levels. Unlike the ex vivo experiments, calcium showed no changes at any timepoint. Interestingly, there was a decrease in potassium ions observed in both experimental groups at 1-h and 2-h post-ECC, which was not observed in the ex vivo circuit benchtop tests. Lactate levels for both LR and HSA were higher at 1 Hr than BL. In addition, LR showed significantly higher lactate compared with Sham and HSA at both 1 Hr and 2 Hr post-ECC; we attribute this increase in LR partially to the amount of lactate inherently present. All groups exhibited acidosis acutely as shown by the extremely low pH, with LR being significantly reduced compared with Sham at 1 h. pH recovered back to BL levels at 24-h post-ECC in all groups.

Table 1.

Arterial hematological parameters

	Hct, %	pHb, g/dL	pH	PCO2, mmHg	HCO3−, mmol/L	PO2, mmHg	tHb, g/dL	SO2, %	K+, mmol/L	Na+, mmol/L	Ca2+, mmol/L	Cl−, mmol/L	Glu, mg/dL	Lac, mmol/L
BL
Sham	52 (3)	0 (0)	7.397 (0.039)	54.3 (5.2)	32.24 (2.45)	56.6 (4)	15.5 (1.1)	89.4 (3.9)	5 (0.7)	138 (1)	1.19 (0.2)	99 (4)	160 (38)	1.1 (0)
LR	50 (2)	0 (0)	7.382 (0.039)	54.5 (6.3)	31.26 (3.28)	60.7 (13.4)	15.6 (1)	89.6 (5.7)	5 (0.5)	138 (1)	1.21 (0.13)	100 (4)	164 (21)	1.3 (0)
HSA	50 (2)	0 (0)	7.392 (0.031)	55 (2.1)	32.34 (1.75)	57 (10.3)	14.8 (0.6)	88.3 (4.4)	4.9 (0.5)	138 (1)	1.13 (0.13)	98 (1)	194 (37)	1.5 (0)
1 h
Sham	53 (5)	0 (0)	7.287 (0.045)†	51.6 (2.5)	23.91 (2.81)†	65.6 (2.6)†	15.7 (1.3)	91.1 (0.8)	3.6 (0.5)†	137 (1)	1.15 (0.11)	105 (3)	203 (36)	1.4 (0)
LR	45 (2)†@	0.1 (0.1)†@#	7.149 (0.106)†@	46 (6.3)	14.61 (0.85)†@#	97.3 (4.9)†@#	14 (0.6) †@#	95.5 (0.9)†@	4 (0.4)†	138 (2)	1.13 (0.1)	111 (2)†@	128 (41)@	5.6 (0.1)†@#
HSA	42 (3)†$	0 (0)	7.222 (0.051)†	49 (6.5)	19.54 (3.07)†$	82.7 (11.7)†$	12.1 (0.6) †$	93.6 (2.5)	3.9 (0.4)†	138 (1)	1.1 (0.08)	109 (3)†$	181 (54)	2.9 (0)†
2 h
Sham	51 (3)	0 (0)	7.263 (0.017)†	52.1 (1.6)	22.76 (1.05)†	60.8 (5.7)	15 (1)	89.5 (2.5)	4 (0.7)	137 (1)	1.1 (0.21)	107 (2)†	184 (35)	0.9 (0)
LR	44 (3)†@	0 (0.1)	7.232 (0.103)†	38.1 (7.8)†@	15.36 (2.39)†@#	108.4 (10.8)†@#	13 (0.8) †@#	97.5 (0.9)†@	3.9 (0.5)†	136 (2)	1.08 (0.05)	112 (3)†@	162 (51)	2.7 (0.1)@#
HSA	40 (2)†$	0 (0)	7.257 (0.032)†	43.2 (4.2)†$	18.61 (1.90)†$	80 (11.8)†$	11.4 (0.4) †$	94.8 (2.1)†$	4 (0.3)†	137 (1)	1.19 (0.15)	110 (1)†$	162 (41)	1 (0)
24 h
Sham	47 (3)	0 (0)	7.373 (0.034)	57.2 (4.9)	32.21 (3.01)	56.8 (3.4)	14.3 (0.8)	89.9 (3)	4.9 (0.4)	139 (2)	1.16 (0.19)	101 (5)	170 (69)	1.6 (0)
LR	32 (1)†@	0 (0)	7.404 (0.047)	48.6 (5.1)@	29.46 (4.28)	67.1 (12.3)	10.3 (0.5)†@	93.3 (3.2)	4.7 (0.2)	139 (2)	1.19 (0.06)	103 (3)	174 (27)	2.7 (0)
HSA	34 (3)†$	0 (0)	7.398 (0.016)	53.1 (1.5)	31.62 (1.06)	67.3 (9.4)	10.2 (0.9)†$	92.3 (4.1)	4.8 (0.4)	139 (1)	1.29 (0.1)	100 (1)	179 (33)	2.6 (0)

Table showing mean and standard deviation of hematological parameters from arterial samples during in vivo experiments. Glu, glucose; Hct, hematocrit; HSA, human serum albumin; Lac, lactate; LR, lactate Ringer; pHb, plasma free hemoglobin; tHb, total hemoglobin; SO₂, oxygen saturation. HCO₃⁻ was calculated using the Henderson–Hasselbalch equation (28). n = 6 hamsters for all groups.

†P < 0.05 as compared with BL, within each group; @P < 0.05 LR vs. sham, within each timepoint; #P < 0.05 LR vs. HSA, within each timepoint; $P < 0.05 HSA vs. sham, within each timepoint.

Hematological parameters showed statistically significant changes in Hct and tHb in both LR and HSA. Both experimental groups were significantly reduced compared with Sham; we attribute this to the hemodilution induced by the ECC priming fluids. However, the Hct and the tHb continue to decrease after the ECC relative to BL and compared with Sham.

Microcirculatory hemodynamics.

Hemodynamics for arterioles are shown in the left column of Fig. 7. Arteriole diameter was significantly lower in the HSA group at all post-ECC timepoints compared with BL, whereas the diameter of the LR group was significantly lower only at 24 h compared with BL. Arteriole diameter in the HSA group was reduced compared with the Sham group at all timepoints post-ECC, and lower than LR at 2 Hr. Velocity for both experimental groups was significantly reduced at all timepoints post-ECC compared with BL and the Sham group. The HSA group had statistically significantly reduced arteriolar velocity compared with LR at 2 Hr post-ECC. Ultimately, arteriolar blood flow for both experimental groups post-ECC was statistically significantly reduced compared with BL and the Sham group; no significant difference in arteriolar blood flow between experimental groups was observed at any timepoint.

Figure 7.

Microcirculatory hemodynamics. Diameter, velocity, and volumetric blood flow are shown from top to bottom, while arterioles [Sham: n = 38, lactate Ringer (LR): n = 33; human serum albumin (HSA): n = 33] and venules (Sham: n = 38, LR: n = 30, HSA: n = 33) are shown from the left to right columns. All values are normalized to baseline [BL, before extracorporeal circulation (ECC)] measurements for each vessel. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 between groups at each timepoint. †P < 0.05 compared with BL, within each group.

Flow measurements for venules are shown in the right column of Fig. 7. Venule diameter showed little change, apart from HSA having a significant increase over Sham and LR at 2 h. Velocity and volumetric flow rate for both experimental groups were statistically lower at all post-ECC timepoints as compared with BL and significantly lower than Sham within each post-ECC timepoint. However, unlike arterioles, HSA displayed higher flow rates compared with LR at both acute timepoints.

The FCD is shown in Fig. 8. The FCD of the LR group was significantly reduced compared with the Sham group at all timepoints after ECC. The FCD of the HSA group was significantly increased compared with the LR group, but lower than the Sham group at 1- and 2-h post-ECC. As compared with BL, the FCD of the LR group was significantly reduced up to 24-h post-ECC, whereas the FCD of the HSA group was only reduced at 1-h post-ECC.

Figure 8.

Functional capillary density (FCD). *P < 0.05, ***P < 0.001, ****P < 0.0001 between groups. †P < 0.05 compared with baseline (BL). n = 6 for all groups.

Oxygen saturation distribution.

Hemoglobin oxygen (HbO₂) saturations (example images in Fig. 9) within arterioles and venules are shown in Fig. 10. At 1 h, arteriole HbO₂ saturation in the LR group was significantly increased compared with the HSA group, while at 2 h, the HbO₂ arteriole saturation in the LR group was statistically significantly lower than the Sham group. The more drastic change occurred on the venule side of the microcirculation. Both experimental groups had statistically lower venule saturation as compared with Sham at all post-ECC timepoints. HSA displayed higher venule saturation as compared with LR at 1-h and 24-h post-ECC. In addition, the venule HbO₂ saturation of the LR group was significantly lower at all timepoints after ECC compared with before ECC, whereas the HbO₂ saturation of the HSA group significantly decreased at 1 and 2 h compared with before ECC, and to the Sham group.

Figure 9.

Representative images depicting the changes in hemoglobin oxygen saturation. Human serum albumin (HSA), lactate Ringer (LR), and Sham are shown from top to bottom. Baseline (BL) is shown on left, while 1 h post-extracorporeal circulation (ECC) is shown on the right.

Figure 10.

Microvascular hemoglobin oxygen saturation changes. The top row shows arterioles [Sham: n = 19, lactate Ringer (LR): n = 13, human serum albumin (HSA): n = 16], whereas the bottom row shows venules (Sham: n = 21, LR: n = 19, HSA: n = 22). *P < 0.05, **P < 0.01, ****P < 0.0001 between groups. †P < 0.05 compared with baseline (BL).

Postinfusion recovery of labeled RBCs.

The Technetium-99m (Tc-99m) count and the Hct were normalized to baseline values, and the mean and standard deviation of both were calculated for each corresponding timepoint. The normalized Hct and Tc-99m mean and standard deviations at 2-h post-ECC were 0.81 [SD 0.06] and 0.84 [SD 0.07] respectively, showing similar values. However, the normalized Hct and Tc-99m mean and standard deviations at 24-h post-ECC were 0.63 [SD 0.03] and 0.52 [SD 0.06].

Systemic inflammation.

Serum inflammatory marker results are shown in Fig. 11, A–C. IL-6, IL-10, and CXCL1 were statistically significantly increased in the LR group compared with Sham and the HSA groups, respectively. In addition, serum epinephrine and norepinephrine (Fig. 11, D and E) show elevated levels in both LR and HSA as compared with the Sham control, with LR being statistically higher as compared with HSA.

Figure 11.

Serum inflammatory markers and catecholamine levels at 24-h post-extracorporeal circulation (ECC). A: interleukin-6; B: interleukin-10; C: CXCL1; D: epinephrine; E: norepinephrine. *P < 0.05, **P < 0.01, ****P < 0.0001. n = 6 for all groups. HSA, human serum albumin; LR, lactate Ringer.

The changes between serum markers from the circuit 0-h post-ECC and at the end of the experiment (24 h) are summarized in Supplemental Table S4. At 0-h post-ECC, creatinine, IL-6, IL-10, CXCL1, bilirubin, epinephrine, norepinephrine, AST, ALT, and atrial natriuretic peptide (ANP) were all statistically significantly lower in the LR group compared with the HSA group. In contrast, ferritin was higher in the LR group than the HSA group 0-h post-ECC. Serum from 24-h post-ECC revealed higher levels of IL-6, IL-10, CXCL1, bilirubin, norepinephrine, ferritin, AST, ALT, ANP, and FABP6 in the LR group compared with the HSA group. Apart from ferritin, all serum markers for the LR group showed significant increases at 24-h post-ECC as compared with 0-h post-ECC. CXCL1, ferritin, AST, FABP2, and FABP6 serum markers in the HSA group was raised at 24-h post-ECC as compared with 0-h post-ECC.

Markers of iron transport and processing were analyzed to see how the hemolysis due to the pump and foreign surface contact affect vital organs. Serum bilirubin (Fig. 12A) and ferritin levels in the serum, spleen, liver, kidney, and heart (Fig. 12, B–F, respectively) all presented a similar trend. The bilirubin and ferritin of the LR group were statistically significantly elevated compared with both Sham and HSA, with bilirubin and ferritin increased in HSA compared with Sham.

Figure 12.

Iron transport markers. A: bilirubin; B: serum ferritin; C: splenic ferritin; D: hepatic ferritin; E: renal ferritin; F: cardiac ferritin. **P < 0.01, ***P < 0.001, ****P < 0.0001. n = 6 for all groups. HSA, human serum albumin; LR, lactate Ringer.

Markers of renal injury, which is one of the most common organ dysfunctions associated with ECMO, are presented in Fig. 13. Serum creatinine, serum blood urea nitrogen (BUN), and urinary neutrophil gelatinase-associated lipocalin (NGAL) were increased in LR as compared with both HSA and Sham, with HSA displaying elevated levels compared with Sham; renal IL-1, IL-6, IL-10, and CXCL1 also showed the same trend as serum markers. Urinary creatinine was statistically significantly higher in LR than both HSA and Sham but did not show any differences between Sham and HSA. Markers for cardiac damage and inflammation are shown in Fig. 14. Cardiac troponin, TNF-α, monocyte chemoattractant protein-1 (MCP-1), C-reactive protein (CRP), CXCL1, IL-1, IL-6, and IL-10 were elevated in LR as compared with HSA and Sham, with HSA being significantly increased compared with Sham. Serum ANP was statistically significantly increased in LR as compared with Sham and HSA, with no difference between the two.

Figure 13.

Markers for renal damage and inflammation after 24-h of extracorporeal circulation (ECC). A: serum creatinine; B: blood urea nitrogen; C: urinary neutrophil gelatinase-associated lipocalin; D: urinary creatinine; E: renal interleukin-1; F: renal interleukin-6; G: renal interleukin-10; H: renal CXCL1. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 between groups. n = 6 for all groups. HSA, human serum albumin; LR, lactate Ringer.

Figure 14.

Cardiac markers for inflammation and myocarditis. A: cardiac troponin; B: cardiac TNF-α; C: cardiac monocyte chemoattractant protein-1; D: cardiac C-reactive protein; E: serum atrial natriuretic peptide; F: cardiac CXCL1; G: cardiac interleukin-1; H: cardiac interleukin-6; I: cardiac interleukin-10. **P < 0.01, *** P < 0.001, **** P < 0.0001 between groups. n = 6 for all groups. HSA, human serum albumin; LR, lactate Ringer.

Splenic CXCL1, serum FABP2, serum FABP6, serum AST, serum ALT, hepatic CXCL1, lung CXCL1, and lung myeloperoxidase (MPO) (Fig. 15, A–H) were statistically significantly elevated in LR as compared with Sham and HSA, with HSA increased compared with Sham. Lung Neu+ count (Fig. 15I) was also elevated in LR compared with Sham but was not increased compared with HSA; Neu+ count for HSA was statistically significantly elevated compared with Sham.

Figure 15.

Markers for splenic (A), small intestine (B and C), liver (D–F), and lung (G–I) damage. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001 between groups. n = 6 for all groups. HSA, human serum albumin; LR, lactate Ringer.

DISCUSSION

ECMO procedures offer a backstop for physicians, providing time to treat underlying pathophysiological conditions that affect respiratory and/or cardiac function. This therapy allows for diffusion of O₂ into the patient’s blood via an oxygenator with the goal of increasing the amount of oxygen delivered to the tissues. Despite the microcirculation’s role in metabolic and gas transport, limited studies have been conducted to investigate the impact on perfusion after these procedures. The present study revealed impaired blood flow in arterioles and venules, with the drop in velocity contributing more to the changes in flow as compared with the changes in diameter. The ECC circuit requires an increase in blood flow, and this additional demand may lead to shunting of flow away from the source of highest vascular resistance: the microcirculation.

Despite both LR and HSA failing to fully recover blood flow at 24-h post-ECC, observed differences between these two priming fluids in the microcirculation indicate that the volume expander properties of the priming fluid partially determine the response to ECC. Using HSA as a priming fluid improved venular blood flow and FCD compared with LR acutely. These differences could stem from the differences in oncotic properties between LR and HSA, as changes in plasma colloidal osmotic pressure (COP) determine the capillary filtration rate, following Starling’s law (29). When ECC starts up, the prime fluid mixes with native blood, causing hemodilution and shifts in plasma protein content. LR lacks colloidal properties, which causes a drop in plasma oncotic pressure. This decrease in plasma oncotic pressure increases net fluid flux from the capillary lumen into the interstitial space. The HSA at 5 g/dL has a similar COP to plasma and a higher COP compared with LR; this difference results in less fluid filtration in the capillaries. Over time, the fluid balance rectifies, causing fluid to shift back from the interstitial space into the capillaries, where it flows into venules. Results from cardiopulmonary bypass (CPB) patients also indicate this, as patients with HSA prime showed positive fluid balance as compared with those subjected to crystalloid primes (30). Thus, the observed difference between LR and HSA may boil down to differences in oncotic pressure post-ECC, but more experiments are required to qualify this.

Pervasive fluid imbalance in the microvasculature led to notable changes in Hct and MAP. A drastic difference between the normalized Hct and Tc-99m counts at 24 h was observed, indicating a shift in the proportions making up blood volume, causing the drop in Hct. The loss of RBCs from hemolysis or increased spleen activity to filter out damaged RBCs (31) is one plausible scenario. However, one of the tell-tale signs, splenomegaly, was not qualitatively observed during organ excision at 24 h, and the levels of hemolysis observed ex vivo were not drastic enough to explain the precipitous Hct drop. The second, and more likely scenario, is an increase in plasma volume due to the shift in fluid volume from the interstitial space into the bloodstream post-ECC, which is also supported by the microcirculation perfusion changes. MAP showed significant drop in all groups acutely; we attribute this partially to the effects of isoflurane, since the Sham control group also experienced this drop in pressure. However, LR was statistically significant as compared with the Sham at 2- and 24-h post-ECC, suggesting that this fluid shift may aggravate this hypotensive state. Additional experiments to measure plasma volume shifts, such as using indocyanine green, are required to confirm this theory.

Hypovolemia and lowered microvascular perfusion cause hypoxia and ischemia. Microvascular distribution of HbO₂ saturation post-ECC show minimal changes in arteriole HbO₂ saturation, whereas venous HbO₂ saturation reduced for both experimental groups after ECC and failed to recover at 24-h post-ECC. By using average microvascular perfusion, average saturation levels, and hemoglobin levels, one can estimate that the oxygen extraction, relative to baseline, was 10% for LR and 25% for HSA at 2-h post-ECC (disregarding the oxygen dissolved in plasma). Both groups demonstrated increased oxygen extraction at 24 h, but neither group reached even 50% of baseline oxygen extraction. These results indicate a repayment of oxygen debt accumulated during and acutely after ECC. Without full repayment of this oxygen debt, the organism can quickly succumb to a state of shock and production of reactive oxygen species, causing a perfect storm for proinflammatory responses that may cause organ deterioration.

Organ markers demonstrate pervasive dysfunction in the spleen, kidney, liver, heart, lung, and small intestine due to ECC. First, iron transport is upregulated in all organs that were analyzed, as shown by elevated ferritin levels (32). Along with elevated ferritin levels, creatinine, BUN (33), and uNGAL (34) point to excessive renal damage, which is one of the most common organs to fail in patients with ECMO, causing development of acute kidney injury that requires renal replacement therapies. Second, bilirubin, AST, ALT elevation in both experimental groups points to significant liver damage (35, 36). Third, elevation of cardiac damage markers such as ANP, cardiac troponin, TNF-α, and C-reactive protein all increases the risk of myocardial infarct, dilated cardiomyopathy, and LV pressure overload (37–40). This is especially noteworthy, since VA-ECMO can increase afterload due to reinfusion of blood on the arterial side, which may lead to LV dilation (41). The most surprising change was the increase in FABP2 and FABP6, due to the scarcity of studies looking at intestinal impact from ECC as compared with most other organs. The presence of these proteins has been shown to be an indicator of epithelial dysfunction and increased permeability of the small intestine primarily due to ischemia (5, 42, 43). This can allow bacterium and other pathogens to cross into the bloodstream, which in severe cases can lead to sepsis. Inflammation with the lungs was apparent, as shown by elevated MPO and neutrophil count. This is important to note, since acute lung injury (ALI) accounts for a significant portion of ECMO cases, which may be exacerbated due to the increase in these markers. Finally, IL-1, IL-6, IL-10, and CXCL1 were elevated in multiple organ systems and plasma, showing perfuse upregulation of inflammatory pathways (37, 44–46). These results demonstrate the dysfunction and upregulation of inflammatory responses within these organs, which can aggravate any number of underlying pathophysiological conditions.

The upregulation of inflammation observed in organs was also seen in the ex vivo whole blood recirculation trials, which increased both IL-6 and IL-10. In vivo, the increase of acute phase response cytokines leads to activation of the complement system that can cascade into systemic inflammatory response syndrome (SIRS) (7). The ex vivo whole blood ECC also revealed significant depletion of coagulation factors such as fibrinogen, disrupting the hemostatic balance. Both this increase in inflammation and decrease in coagulation factors observed in the present study can be attributed to foreign surface contact. The decrease in coagulation factors can stem from plasma proteins adhering to foreign surfaces that then get replaced by higher affinity proteins (e.g., replacement of fibrinogen by kininogen), a phenomenon known as the Vroman effect (47). Foreign surfaces in extracorporeal circuits affect inflammation by binding of C3b to foreign surfaces, activating the complement pathway (48). To alleviate some of these detrimental effects in clinical settings, ECC surfaces are often protein coated with albumin. The presence of albumin partially explains the reduced organ damage and inflammation in the HSA group as compared with the LR group. However, the use of an albumin solution to prime rather than coat could improve microvascular perfusion by maintaining colloidal oncotic pressure, while limiting inflammation. Various other approaches to minimize the effects of foreign surface contact have been investigated, such as NO matrices (49), superhydrophobic surfaces (50), and heparin-bonded surfaces (51). Further research into these areas is needed to help reduce inflammation and maintain hemostasis in ECC procedures.

Despite these findings, there are several limitations in the current study. First, we used healthy subjects for this study, which can be seen a detriment. Although use of healthy animals provides more consistency as compared with clinical studies that involve patients with varying pathophysiological conditions, one cannot speculate as to whether microcirculatory flow would have improved with ECC if the animals had comprised cardiopulmonary function. The hematocrit as measured at 1-h post-ECC was closer to 40%, whereas clinical levels are often closer to 30% for clinical VA-ECMO configurations. Hemodilution lowers blood viscosity and does aid in microvascular perfusion to an extent, but the difference between a hematocrit of 30% and 40% does not fully explain the decreased microvascular flow observed in the present study. Third, we were only able to achieve 15% of the CO through the circuit, which is not enough flow in comparison to VA-ECMO. This is most likely due to limited blood volume of the animal (between 4–5 mL), in contrast with an average adult human (∼1,000 times the hamster volume). In addition, the CO has a relatively higher contribution from HR for hamsters (resting ∼400–500 beats/min) as compared with humans (resting ∼60 beats/min). The heparin dosage used was intended to limit clot formation. Although heparin can also attenuate inflammation, further studies are required to explore optimization of heparin dose to reduce inflammation. Markers of organ function and inflammation were quantified with ELISA assays with reactivity to Golden Syrian hamsters, but not calibrated specifically for hamsters by manufactures. Thus, the actual protein concentration may not precisely be the level measured with the assay. Finally, the dorsal skinfold model analyzes perfusion in subcutaneous tissue and striated skeletal muscle and does not account for microcirculation changes in other critical organs; this is an important distinction, since upstream shunting effects during and after ECC cannot be considered.

Conclusions

This study quantifies the changes in microcirculatory flow after extracorporeal circulation. ECC was induced on hamsters fitted with a dorsal window preparation using a circuit to pull blood from the venous side, pump it through a bubble trap using a peristaltic pump, and reinfuse it into the arterial side. At the microvascular scale, a significant drop in perfusion as well as increase in oxygen extraction post-ECC was observed both acutely and after 24 h. HSA priming fluid performed better than LR in terms of venule and capillary flow post-ECC. These results suggest that both hypoxia and hypoperfusion are present, leading to ischemia. If persistent for an extended period, cells switch to anaerobic metabolism, leading to the interruption of key cellular processes and the buildup of toxic waste byproducts. Therefore, this study asserts that microcirculation compromise is the vehicle driving ischemia, causing inflammation and dysfunction in critical organs, and promoting multiorgan failure in ECC. Future work is needed to improve upon this model and investigate mechanisms to help alleviate these hypoxic microcirculatory conditions post-ECC.

SUPPLEMENTAL DATA

Supplemental Tables S1–S4: https://doi.org/10.6084/m9.figshare.c.5672074.

GRANTS

This work was supported by National Institutes of Health Grants R01HL159862 and R01HL138116, and the Department of Defense under Grants W81XWH-18-1-0059.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

K.G. and P.C. conceived and designed research; K.G. performed experiments; K.G. and P.C. analyzed data; K.G. and P.C. interpreted results of experiments; K.G. prepared figures; K.G. drafted manuscript; K.G. and P.C. edited and revised manuscript; P.C. approved final version of manuscript.