Extracorporeal Life Support without Systemic Anticoagulation: Nitric-Oxide-Based Non-Thrombogenic Artificial Placenta Circuit in Fetal Sheep

Abstract

Introduction

Clinical translation of the extracorporeal Artificial Placenta (AP) is impeded by the high risk for intracranial hemorrhage in extremely premature newborns. The Nitric Oxide Surface Anticoagulation (NOSA) system is a novel non-thrombogenic extracorporeal circuit. This study aims to test the NOSA system in the AP without systemic anticoagulation.

Methods

Ten extremely premature lambs were delivered and connected to the AP. For the NOSA group, the circuit was coated with DBHD-N₂O₂/argatroban, 100 ppm nitric oxide (NO) was blended into the sweep gas, and no systemic anticoagulation was given. For the Heparin control group, a non-coated circuit was used and systemic anticoagulation was administered.

Results

Animals survived 6.8±0.6 days with normal hemodynamics and gas exchange. Neither group had any hemorrhagic or thrombotic complications. ACT (194±53 vs. 261±86 seconds; p < 0.001) and aPTT (39±7 vs. 69±23 seconds; p < 0.001) were significantly lower in the NOSA group than the Heparin group. Platelet and leukocyte activation did not differ significantly from baseline in the NOSA group. Methemoglobin was 3.2±1.1% in the NOSA group compared to 1.6±0.6% in the Heparin group (p < 0.001).

Conclusions

The AP with the NOSA system successfully supported extremely premature lambs for 7 days without bleeding or thrombosis.

Introduction

Extreme prematurity is a top public health priority yet remains without an ideal solution. Despite significant advances in the treatment of prematurity, extremely low gestational age newborns (ELGANs), defined as <28 weeks estimated gestational age (EGA), experience significant mortality and substantial respiratory, central nervous system, gastrointestinal, and infectious morbidity ¹. For many critically ill neonates, extracorporeal life support (ECLS) provides crucial gas exchange and hemodynamic support while their organ function recovers from injury. Despite a growing collective experience in providing “preemie ECLS” to neonates born at 29–33 weeks gestation ², ECLS remains contraindicated in ELGANs due to size constraints and risk of intraventricular hemorrhage (IVH). We have utilized ECLS technology to develop a novel system to support extremely premature newborns which we call the Artificial Placenta (AP). The AP recreates fetal physiology by maintaining extracorporeal gas exchange, fetal circulation, fluid-filled lungs, and sustaining a low-oxygen environment. This approach has resulted in over 2 weeks of support for extremely premature sheep and has demonstrated organ protection and ongoing development ^3–9.

Patients supported by ECLS remain in a delicate balance of bleeding and clotting. The entire ECLS circuit is made of an artificial material and thus triggers major cellular activation (particularly platelets) upon contact with blood. This makes the ECLS circuit extremely thrombogenic. Systemic anticoagulation is required to prevent thrombosis of circuit components (e.g., oxygenator, pumps, tubing, connectors). In neonates and pediatric patients, this limits the incidence of thrombosis to 27–60%, but also produces a 25–50% risk of bleeding, including intracranial, gastrointestinal, and pulmonary hemorrhage ¹⁰. This tradeoff is acceptable given the improved survival ECLS confers to these patients. The calculus changes, however, when considering ECLS for premature newborns, who are at high risk of IVH. Hemorrhage in the premature germinal matrix of the brain can result in severe neurologic morbidity or mortality. Very preterm infants (28–32 weeks EGA) have a baseline IVH risk of 1–5% ¹¹. The early collective experience with “preemie ECLS” has found the risk of IVH increases to 21% when this population is supported by ECLS with systemic anticoagulation ². The risk of IVH increases with decreasing gestational age ¹², reaching as high as 35% in ELGANs ¹. Systemic anticoagulation for AP support would make that risk prohibitively high. Therefore, a method for providing ECLS without systemic anticoagulation is critical to make the AP safe for clinical translation in ELGANs.

A promising and revolutionary solution is to provide local anticoagulation that prevents thrombosis of circuit components but does not have any systemic effect on the patient. Such a system could allow the AP and other ECLS circuits to be used without systemic anticoagulation. We have developed one such system—the Nitric Oxide Surface Anticoagulation (NOSA) system. The NOSA system utilizes nitric oxide (NO), which is an endogenous gasotransmitter produced by vascular endothelial cells to inhibit platelet and cellular activation within the native blood vessels. The NOSA system consists of two components: a nitric oxide (NO)-releasing circuit coating and NO gas blended into the sweep gas going to the oxygenator. The circuit coating consists of diazeniumdioated dibutylhexanediamine (DBHD-N₂O₂) NO donor, which releases NO when it comes in contact with blood. The addition of argatroban—a direct thrombin inhibitor—to the coating prevents fibrin deposition and improves thromboresistance of the coated polymer surface. Since it would be extremely difficult to coat the hollow-fiber membrane of the oxygenator without affecting the gas exchange, NO is blended into the sweep gas to prevent platelet and cellular activation on the membrane of the oxygenator. in a previous study, the NOSA system has successfully maintained the patency of a pumpless artificial lung circuit for 24 hours ¹³. While these initial studies of the NOSA system have been promising, AP support would be expected to last several weeks. The long-term performance of the NOSA system has never been evaluated.

The aim of this study was to evaluate the efficacy of the NOSA system in supporting premature sheep with the AP without systemic anticoagulation for 7 days and compare those outcomes to a control group using heparin.

Methods

AP Circuit Construction

The drainage and reinfusion cannulae were connected to a circuit made of 1/4-inch PVC (polyvinyl chloride) Tygon^® tubing (Saint Gobain, Courbevoie, France), oxygenator/heat exchanger (Medos Hilite 800LT, Xenios AG, Heilbronn, Germany [n=8] or Pediatric Quadrox-iD, Maquet, Rastatt, Germany [n=2]), and a passive-filling roller pump (MC3, Ann Arbor, MI) (Figure 1).

Figure 1.

Artificial Placenta (AP) circuit. The system uses jugular vein drainage, umbilical vein reinfusion, a non-occlusive roller pump (MPump, MC3, Ann Arbor, MI), neonatal oxygenator (Medos Hilite 800LT, Xenios AG, Heilbronn, Germany or Pediatric Quadrox-iD, Maquet, Rastatt, Germany), and 1/4-inch Tygon^® tubing (Saint-Gobain, Courbevoie, France). The sheep are intubated and lungs filled with perfluorocarbon to an open meniscus of 5–7 cm H₂O. a) The Heparin control group used the above AP circuit with non-coated circuit tubing and pump chamber, sweep gas without nitric oxide (NO), and a continuous heparin infusion for systemic anticoagulation. b) The NOSA group used circuit tubing and pump chambers coated with NO-releasing diazeniumdiolated-dibutylhexanediamine (DBHD-N₂O₂) and argatroban, sweep gas blended with NO at 100 ppm generated by an electrochemical NO generator (NOgen), and no systemic anticoagulation. JV = jugular vein, SVC = superior vena cava, RA = right atrium, IVC = inferior vena cava, Ao = aorta, DV = ductus venosus.

Circuit Coating

Circuit coating and sterilization:

All the circuit components (tubing, connectors, pump chamber) were coated using a previously reported method ^14,15. Briefly, two layers of 25 wt% diazeniumdiolated dibutylhexanediamine (DBHD-N₂O₂, synthesized in our laboratory as previously described ¹⁶), as the NO donor with a 10 wt% poly(lactic-co-glycolic acid) (PLGA, Sigma-Aldrich, St. Louis, MO) additive and a top layer of 10 μM argatroban (Toronto Research Chemicals, Toronto, Canada)/hexamethlyene diisocyanate (HMDI, Sigma-Aldrich Chemical Co., St. Louis, MO) were coated on the inner surface of the circuit. The NO donor/PEG as well as the argatroban/HMDI were dissolved in a CarboSil/tetrahydrofuran (THF) solution. Each coated layer was allowed to dry for one hour, and the coated and fully assembled circuit was allowed to cure for at least 24 hours. The components of the coated circuit were sterilized by a liquid chemical method (Metricide OPA Plus, Metrex, Romulus, MI) following the manufacturer’s instruction.

NO generator (NOgen):

Nitric oxide (NO) gas was generated by electrocatalytic reduction of nitrite ions (1 M sodium nitrite) using Cu(II)-1,4,7-trimethyl-1,4,7-triazacyclononane mediator ¹⁷. The NO was extracted from the nitrite solution into a nitrogen gas stream (50–200 mL/min) via a silicone hollow-fiber gas separator unit (PermSelect PDMSX-2.1, Ann Arbor, MI), that was blended into the sweep gas of the oxygenator. The NO level in the sweep gas was monitored by electrochemical sensor (AlphaSense, Essex, United Kingdom) and the current for NO generation was feedback regulated in order to maintain the target 100 parts per million (ppm) NO concentration in the sweep gas.

NO release measurement of the coated circuit:

To determine the amount of NO released per square centimeter of the coated circuitry before and at the end of the experiment, a 0.5-cm wide section of the circuit tubing was measured via an ozone chemiluminescent NO analyzer (GE Sievers 280i Nitric Oxide Analyzer (NOA), Boulder, CO). The tube section was placed into a glass measuring cell containing 5 mL phosphate buffered saline (PBS) and thermostated at 37°C. The emitted NO was continuously sparged from that PBS solution with nitrogen gas and the headspace of the NOA cell was analyzed with the NOA at atmospheric pressure with sampling flow rate of 200 mL/min.

Surgical Procedure

The sheep in this experiment were treated in compliance with the Guide for Care and Use of Laboratory Animals (US National Institutes of Health publication No. 85–23, National Academy Press, Washington D.C., revised 2011) and all methods were approved by the University of Michigan Institutional Animal Care and Use Committee.

Extremely premature lambs (n = 10) were delivered by cesarean section between 117–121 days EGA (term = 145 days). This gestational age was chosen because lamb fetuses at 118 days gestation are at a similar stage of lung development as 24-week human fetuses ¹⁸. All lambs were cannulated using 10–14 Fr cannulae (Medtronic, Minneapolis, MN) for drainage via the right jugular vein and reinfusion via the umbilical vein. A 5-Fr arterial line (Medtronic, Minneapolis, MN) was placed in the umbilical artery for hemodynamic monitoring and blood sampling. Patency of the arterial line was maintained in all animals with heparin sulfate 1 U/mL (SAGENT, Schaumburg, IL) and 0.12 mg/mL papaverine (American Regent, Shirley, NY) infused at 1 mL/hr. Prostaglandin E1 (Pfizer, New York, NY) infusion (0.2 μg/kg/min) was administered to maintain patency of the ductus arteriosus. Lambs were intubated and lungs were filled with perfluorodecalin (Origen, Austin, TX) to a meniscus with a pressure of 5–8 cm H₂O which allowed for fetal breathing movement.

Experimental Groups

Nitric Oxide Surface Anticoagulation (NOSA):

The NOSA group (n=5) were connected to an AP circuit that had all blood-contacting surfaces except for the oxygenator (cannulae, circuit tubing, and pump chamber) coated with DBHD-N₂O₂ and argatroban. The sweep gas consisted of 97% O₂, 3% CO₂ blended with the electrochemically generated NO in N₂ (50–200 mL/min) with 100 ppm NO at the inlet of the oxygenator. The animals in the NOSA group did not receive bolused or infused heparin (excluding 1 U/hr for patency of the uncoated arterial line as described above).

Heparinized Controls (Heparin):

The Heparin group (n=5) served as the control group. They used conventional (i.e., non-coated) cannulae and circuits and received systemic therapeutic anticoagulation with heparin titrated to achieve an activated clotting time (ACT) of 240–280 seconds. The sweep gas for the Heparin group was 97% O₂ and 3% CO₂ without NO.

Medical Management

Lambs were monitored continuously and vital signs were recorded every 30 minutes during AP support. Maternal packed red blood cell transfusions were given to the lamb if the hemoglobin was below 10 g/dL. Placental and twin whole blood, colloid solutions, or crystalloid solutions were administered as needed for volume resuscitation. Norepinephrine 0.1–0.5 μg/kg/min (Claris LifeSciences Inc., North Brunswick, NJ), epinephrine 0.1–1.0 μg/kg/min (Hospira Inc., Lake Forest, IL), or dopamine 1–20 μg/kg/min (Baxter Healthcare, Deerfield, IL) were administered as needed for hypotension (mean arterial pressure < 30 mmHg) not responsive to fluid resuscitation. Empiric intravenous piperacillin-tazobactam 100 mg/kg (Hospira Inc., Lake Forest, IL) and fluconazole 6 mg/kg (SAGENT, Schaumburg, IL) were administered every 12 and 24 hours, respectively, for antimicrobial prophylaxis. Methylprednisolone 0.63 mg/kg (Pfizer, New York, NY) was given every 6 hours to protect against adrenal insufficiency and limit inflammatory reaction to the circuit. Buprenorphine 6 mcg/kg (Parr Inc., Spring Valley, NJ) and diazepam 0.5–1 mg/kg (Hospira Inc., Lake Forest, IL) were given for pain and sedation management. Total parental nutrition with SMOFlipid^Ⓡ was administered at 1–4 mL/kg/hr. The total fluid goal was maintained at approximately 80–120 mL/kg/day.

Arterial blood gases (ABL 825 Radiometer, Radiometer, Copenhagen, Denmark) were collected every 1–4 hours and were used to titrate AP support to maintain fetal blood gas values: arterial oxygen saturation (SO₂; 60–75%), arterial partial pressure of oxygen (pO₂; 25–45 mmHg), arterial partial pressure of CO₂ (pCO₂; 40–55 mmHg).

Data collection and laboratory analysis schedules are shown in Table 1. Complete blood count was determined by IDEXX ProCyte Dx (Westbrook, ME) hematology analyzer and ACT were measured with a Hemochron Blood Coagulation System Model 801 (International Technidyne Corp. Edison, NJ).

Table 1.

Data Collection

Category	Parameters	Frequency
Hemodynamics	Heart rate, arterial BP, and temperature	Monitored continuously and recorded every 30 minutes
Circuit data	Circuit blood flow, pre- and post-oxygenator pressure	Monitored continuously and recorded every 30 minutes
Sweep gas	Sweep flow rate and NO concentration	Monitored continuously and recorded every 30 minutes
Acid/base balance	Arterial blood gas	Every 4 hours and as needed
Oxygenator gas exchange	Pre- and post-device blood gases	Every 12 hours
Anti-coagulation monitoring	ACT	Baseline and every 1–4 hours per nomogram (control) or every 4 hours (NOSA)
Laboratory analysis	CBC, CMP, coagulation profile	Baseline and POD 1, 2, 5, and 7
Cellular activation	Flow cytometry for P-selectin, CD11b (granulocyte- and monocyte-specific)	Baseline and POD 1, 2, 5, and 7
Hemolysis	Plasma free hemoglobin	Baseline and POD 1, 2, 5, and 7

Note: Baseline = prior to cannulation, BP = blood pressure, NO = nitric oxide, ACT = activated clotting time, NOSA = Nitric Oxide Surface Anticoagulation, CBC = complete blood count, CMP = comprehensive metabolic panel, POD = post-operative day

Blood chemistry and metabolic panel were assessed using IDEXX Catalyst Dx chemistry analyzer (17-panel clip). For plasma free hemoglobin analysis, a UV Plate reader (Biotek, Ultrospec 500 pro) was used. Coagulation parameters were measured and monitored with a Dade Behring BCS XP analyzer (Siemens Deerfield, IL). The aggregation assay on sheep platelets was performed using a Chrono-Log optical aggregometer (Chrono-log Corporation, Havertown, PA). The percentage of aggregation was determined 3 min after the addition of collagen using Chrono-Log Aggrolink software. Expression of P-selectin (CD62P) as a marker for platelet activation and CD11b as a marker for leukocyte activation were measured using fluorescence-activated cell sorting (FACS). Sample preparation and the FACS assays were performed as previously described ^19,20 using the following antibodies: 1) determination of platelet activation: RPE -mouse anti-human CD62, FITC-mouse anti-pig CD61, FITC-mouse IgG1 negative control, and RPE-mouse IgG1 negative control; 2) determination of leukocyte activation: FITC -mouse anti- bovine CD11b, RPE-mouse anti-human CD14, FITC-mouse IgG2b negative control, and RPE-mouse IgG2a negative control antibodies. All antibodies were obtained from Bio-Rad (Hercules, California). BD FACS Canto II flow cytometer (Becton Dickinson, San Jose, CA) was used for data acquisition and the FlowJo (v10.8.0) (Becton Dickinson, San Jose, CA) software was used for data analysis.

Necropsy and Histopathologic Evaluation

Animals were electively euthanized at the end of each experiment. All relevant organs were collected and formalin fixed. Tissues were sectioned 5 μm thick and slides were stained with hematoxylin and eosin. Slides were reviewed by a pathologist (RR) blinded to experimental groups.

Statistical Analysis

Demographic and clinical characteristics are reported as mean and standard deviation for continuous variables and number of observations and percent for categorical variables. EGA, weight, and histologic scores were compared between the NOSA and Heparin groups using the non-parametric Mann Whitney U test. Linear mixed effects (LME) models with an animal-level random effect were used to assess differences between NOSA and Heparin groups for hemodynamics, blood gas values, and labs while adjusting for within animal repeated measures. Subsequent LME’s which included an interaction term between group and time were used to test for group differences over time. Contrasts were then conducted to test for between group differences at each time point and within group differences over time.

For outcomes regarding oxygenator function (pressure drop, resistance, gas exchange), only two subjects were included for the Heparin group. Two subjects were excluded because the Pediatric Quadrox-iD oxygenator was used instead of the Medos Hilite 800 LT. The Medos had been used in all previous runs (5 NOSA animals and 2 Heparin animals) but was no longer available in the United States. A third subject was excluded due to malfunctioning pre- and post-oxygenator pressure monitoring. Because of this imbalance in sample size for this outcome variable, only between-group differences were tested. All analyses were conducted in STATA 15 (StataCorp, College Station, TX) and significance was set at p<0.05.

Results

Animal Subjects and Clinical Course

Delivery and cannulation for AP support were uneventful and without complications in all lambs in both groups. The experimental groups were similar in terms of weight (NOSA: 3.0±0.6 kg vs. Heparin: 2.6±0.3 kg; p=0.460) and EGA (NOSA: 119.6±1.4 vs. Heparin: 119.0±1.6 days; p=0.732) of the lambs at delivery.

All animals in the NOSA group survived 7 days on AP support. Four of 5 animals in the Heparin group survived 7 days. One animal died after 5 days due to potential adrenal insufficiency. Both groups demonstrated overall hemodynamic stability, with heart rate and blood pressure within normal range for these premature sheep (Table 2). Arterial pH, pCO₂, pO₂, and SO₂ in both groups were within target ranges for fetal physiology. Arterial lactate and hemoglobin values were similar between groups.

Table 2.

Demographics, Hemodynamics, and Laboratory Analysis: NOSA vs. Heparin

	NOSA(n=5)	Heparin(n=5)	p-value
Weight at delivery (kg)	3.0±0.6	2.6±0.3	0.460
EGA (days)	119.6±1.4	119.0±1.6	0.732
Heart rate (bpm)	194.9±34.4	193.4±25.2	0.986
MAP (mmHg)	49.3±7.6	41.4±6.6	≤0.001
pH	7.31±0.08	7.30±0.11	0.871
PaCO2 (mmHg)	48.3±8.6	50.0±15.8	0.680
PaO2 (mmHg)	30.9±20.4	34.6±9.2	0.097
SaO2 (%)	65.5±9.9	68.2±8.6	0.261
Lactate (mmol/L)	1.9±1.5	2.3±1.6	0.236
Hemoglobin (g/dL)	10.9±1.5	10.6±1.3	0.757
Methemoglobin (%)	3.2±1.1	1.6±0.6	≤0.001
BUN (mg/dL)	15.4±8.4	26.6±15.1	0.056
Creatinine (mg/dL)	1.5±2.9	1.4±0.6	0.999
Albumin (g/dL)	2.0±0.4	1.8±0.2	0.260
Aspartate aminotransferase (U/L)	101.1±62.3	120.1±132.5	0.184
Alanine aminotransferase (U/L)	15±8	15.6±10	0.545

Note: Values reported as mean±standard deviation. All laboratory analyses were performed on aortic blood samples. EGA = estimated gestational age, MAP = mean arterial pressure, PaCO₂ = arterial partial pressure of CO₂, PaO₂ = arterial partial pressure of O₂, BUN = blood urea nitrogen

Circuit Performance

There were no oxygenator changes or development of significant circuit clots in either group. AP circuit flow per kg was higher in the NOSA group than the Heparin group (109.0±20.4 mL/kg/min vs. 92.2±17.1 mL/kg/min; p = 0.027) (Figure 2A). Pressure drop (11.3±5.2 mmHg/L/min vs. 8.4±2.0 mmHg/L/min; p = 0.402) and resistance (34.4±9.8 mmHg/L/min vs. 31.3±4.7 mmHg/L/min; p = 0.627) across the oxygenators was similar between the two groups (Figure 2B). Gas exchange remained excellent in the oxygenators of both groups, with post-oxygenator saturations 99% or greater for the duration of all experiments (Figure 2C).

Figure 2.

Circuit and oxygenator performance. a) Artificial Placenta circuit blood flow per kg. b) Oxygenator resistance. Two sheep were excluded because a Quadrox iD was used instead of Medos Hilite 800LT and one sheep was excluded for malfunctioning pre- and post-oxygenator pressure monitoring. c) Pre- and post-oxygenator blood oxygen saturation. (“SO₂ pre” and “SO₂ post”, respectively). Data in panels A&B represent the mean of each group calculated every 6 hours. Data in panel C represent the mean of each group calculated every 24 hours due to less frequent collection of these data points. Error bands (panels A&B) and error bars (panel C) represent standard deviation.

Systemic Effects

There were no hemorrhagic complications in either group. The Heparin group was therapeutically anticoagulated with ACT of 261±86 seconds (target: 240–280 seconds) and aPTT of 69±23 seconds while on AP support (post-operative day 1 to 7). Conversely, the NOSA group did not show evidence of systemic anticoagulation, with average ACT of 194±53 seconds and aPTT of 39±7 seconds (Figures 3A, B). These were both significantly lower than the Heparin group (p < 0.001 for both). International normalized ratio (INR) was similar in both groups (NOSA: 1.6±0.3 vs. Heparin: 1.9±0.6; p = 0.066)

Figure 3.

Coagulation profile. a) Activated clotting time (ACT). Data represent the mean of each group calculated every 6 hours due to the slight variations in the experimental timepoints at which ACT samples were taken according to the specific clinical status of each sheep. b) Activated partial thromboplastin time (aPTT). Mean value for experiment hour 0, 24, 48, 120, and 168. For both panels, time 0 represents sheep baseline prior to cannulation for the AP. Error bands (panel a) and error bars (panel b) represent standard deviation.

Platelet count in both groups decreased significantly from baseline to post-operative day 2 (NOSA from 455±48 × 10³/μL to 138±72 × 10³/μL; Heparin from 321±67 × 10³/μL to 113±40 × 10³/μL; p < 0.001 for both groups) and subsequently gradually increased (Figure 4A). Overall, platelet counts did not differ between groups over the 7-day experiment (p = 0.6761). P-selectin, a marker for platelet activation, did not significantly change over the course of the experiment in either group (p = 0.956) (Figure 4B). Platelet aggregometry showed progressively decreasing aggregation of platelets (measured by amplitude) in both groups, though only the NOSA group reached statistical significance (p < 0.001). There was no difference in platelet aggregometry between groups (p = 0.172) (Figure 4C).

Figure 4.

Platelets. a) Platelet count. b) P-selectin expression, shown as a percent of baseline expression. c) Platelet aggregation percentage (amplitude). Graphs represent the mean values for experiment hour 0, 24, 48, 120, and 168. Time 0 represents sheep baseline prior to cannulation for the AP. Error bars represent standard deviation.

White blood cell count was significantly lower in the NOSA group compared to the Heparin group (3.0±2.0 × 10³/μL vs. 6.2±3.5 × 10³/μL; p = 0.035), although those differences were limited to post-operative day 2 (2.9±2.0 × 10³/μL vs 6.4±3.2 × 10³/μL; p = 0.050) and post-operative day 7 (2.7±0.8 × 10³/μL vs. 7.5±5.1 × 10³/μL; p = 0.014) (Figure 5A). Leukocyte activation was similar between groups as demonstrated by similar expression of monocyte-specific (p = 0.125) and granulocyte-specific (p = 0.240) CD11b. Expression of CD11b in the NOSA group did not significantly change from baseline throughout the course of the experiment (mono: p = 0.882; granulocyte-specific: p = 0.906) (Figure 5B and 5C). In the Heparin group, monocyte-specific CD11b expression changed from baseline over time (p = 0.0188) and granulocyte-specific CD11b changed over time (p = 0.0175).

Figure 5.

Leukocytes. a) White blood cell count. b) Granulocyte-specific CD11b expression, shown as a percent of baseline expression. c) Monocyte-specific CD11b expression, shown as a percent of baseline expression. Data represent the mean values for experiment hour 0, 24, 48, 120, and 168. Time 0 represents sheep baseline prior to cannulation for the AP. Error bars represent standard deviation.

The NOSA group had methemoglobin of 3.2±1.1% of total hemoglobin, which was significantly higher than the Heparin group (1.6±0.6%) (p < 0.001).

Necropsy and Histology

At the time of necropsy, all animals had patent ductus arteriosus and ductus venosus. Gross examination of the tissues revealed no evidence of significant bleeding or thromboembolic events in any of the animals in either group.

Kidneys:

The Heparin group had 3 sheep with intratubular and interstitial hemorrhage and no sheep with thrombosis. The sheep in the NOSA group had no evidence of renal hemorrhage or thrombosis.

Liver:

All sheep in the Heparin group showed evidence of cholestasis. Four had focal parenchymal or subcapsular hemorrhage. Three animals in the NOSA group had cholestasis. Two animals in the NOSA group had areas of focal thrombus, one of which also had areas of focal hemorrhage.

Lungs:

The most common finding in the lungs was focal interstitial hemorrhage, which occurred in four animals in each group. No animals in the Heparin group had evidence of thrombosis, whereas four animals in the NOSA group had focal arterial thrombi or thromboemboli; these tended to be rare and were located in the periphery in all but one instance.

Post-mortem circuit evaluation

Photos of a representative oxygenator and circuit from the NOSA group after 7 days of support are shown in Figure 6A & Figure 6B. There was no major clot burden in the oxygenators or circuits of either group. NO flux from the circuit coating at the beginning of the run was 8.8±3.5×10⁻¹⁰ mol NO/min/cm². After 7 days of support, the flux rates were 4.5±2.2×10⁻¹⁰ mol NO/min/cm² (p = 0.313). The NO flux from all circuits remained within or above the target levels of 0.5–4×10⁻¹⁰ mol NO/min/cm² at the end of the 7-day experiments (lowest = 2.1×10⁻¹⁰ mol NO/min/cm²).

Figure 6.

a) Representative oxygenator and b) circuit tubing from the NOSA group after 7 days of AP support. c) NO flux from NOSA circuit tubing before (Start) and after (End) 7 days of AP support.

Discussion

The Artificial Placenta is a novel extracorporeal life support system that recreates fetal physiology to provide extremely premature infants with an environment that protects and promotes organ growth and development. At this stage of development, the primary barrier to clinical implementation is the requirement for systemic anticoagulation. The goal of this study was to evaluate a novel method of circuit-based anticoagulation to permit the AP to be used without systemic anticoagulation. Application of the NOSA system allowed the AP to support premature lambs for 7 days without systemic anticoagulation. The circuits remained patent and there were no major thrombotic complications for the entire duration of the study. The animals in the NOSA group showed no evidence of systemic anticoagulation, with normal ACTs and aPTTs. They demonstrated only mild elevations in methemoglobin. Gas exchange, hemodynamics, laboratory analysis, and circuit performance were similar to a group of control animals supported by the AP with a standard circuit receiving heparin. To our knowledge, this is the longest successful application of a novel non-thrombogenic extracorporeal circuit in vivo.

The AP is a specialized ECLS system that aims to provide an environment of organ protection and development for ELGANs. It is based on veno-venous ECLS with jugular vein drainage and umbilical vein reinfusion. It is pump-driven, providing the patient with gas-exchange support that runs in parallel with the native circulation. The level of support is titrated to maintain low oxygen tension consistent with fetal blood-gas values. The fetal lambs are intubated and lungs are filled with perfluorocarbons to a meniscus allowing fetal breathing movements to promote lung growth. The AP has shown great promise in the laboratory setting, supporting premature animals for over 2 weeks ³. It has been successful in minimizing injury and promoting development of lungs ^3,4,6, brain ^5,7, liver ²¹, kidney (manuscript in progress), bowel ⁹, and spleen ⁸.

Despite success in the laboratory, one of the major obstacles to clinical translation of the AP is the need for systemic anticoagulation with current ECLS technology. A promising option for eliminating the need for systemic anticoagulation is a non-thrombogenic circuit. Circuit coatings in clinical use today include heparin ²², albumin ²³, phosphorylcholine ²⁴, poly-2-methoxyethylacrylate ²⁵. Of these, heparin coating is the most extensively studied and has been found to be associated with a reduced cellular and complement activation ²⁶, shorter length of stay ²⁶, reduced cerebral complications ²⁶, fewer transfusions ²⁷, and lower systemic heparin dosing ²⁷. Nevertheless, no coating has proven effective at obviating the need for systemic anticoagulation during ECLS.

Several laboratories are making progress on non-thrombogenic circuit coatings using tethered liquid perfluorocarbon ²⁸, Zwitterionic poly-carboxybetaine ²⁹, and cultured endothelial cells ³⁰. The longest successful trial of these coatings was by Roberts et al ²⁸. They achieved a 60% circuit patency rate at 72 hours using a coating of liquid perfluorocarbon, though this was associated with reduced gas exchange. This study of the NOSA system in the AP is the longest published study of a non-thrombogenic circuit being used without major thrombotic events.

Our approach to creating non-thrombogenic surfaces aims to mimic the native vascular endothelium by continuously releasing NO at all artificial surfaces. NO acts by briefly “anesthetizing” platelets that pass very near to the surface, thus inhibiting their activation and adhesion ³¹. Because NO is instantly removed from platelets once they return to the flowing blood, systemic anticoagulation does not occur. The most effective NO donor developed by our laboratory is DBHD-N₂O₂ ¹⁹. Argatroban is added to the coating to limit activation of the coagulation cascade by foreign surface contact by blood proteins ¹⁴. The polymeric DBHD-N₂O₂/argatroban coating cannot be applied to the hollow fibers of the membrane oxygenator because it will impede gas exchange. To address this, we have added NO to the sweep gas, which readily crosses the fibers into the blood phase to produce physiologic NO flux across the membrane without affecting gas exchange ¹³. For potential clinical application, the cost of NO from tanks of gas would be very high; however, the NOSA system uses the NOgen, a novel electrochemical gas phase NO generator, which produces NO at high efficiency and low cost ¹⁷. The NOSA system has previously been shown to maintain the patency of a pumpless artificial lung circuit for 24 hours without anticoagulation ¹³. This study demonstrates its efficacy for 7 days in the AP model. Implementation of the NOSA system as an effective non-thrombogenic circuit is a critical milestone for the safe clinical translation of the AP.

We anticipated that NO release would significantly limit platelet and leukocyte activation. This appeared to be the case, as expression of P-selectin and CD11b were unchanged in the NOSA group throughout the study period. However, platelet aggregation was inhibited in the NOSA group to a degree similar to that of the Heparin group. This discordance of platelet activation assays after exposure to an NO-releasing circuit coating has previously been described ³². One possible explanation is that the NO-releasing circuit coating causes a structural alteration of plasma fibrinogen, a protein that is required for platelets to aggregate. Another possible explanation is that the platelets were affected by the medications administered to the animal subjects, including prostaglandin E1 and steroids. Further studies will be useful to evaluate those possible mechanisms.

The animals in the NOSA group had a decline in platelet counts in the first two post-operative days that was very similar to that seen in the Heparin group. This loss of platelets is commonly seen in ECLS and is generally attributable to platelet consumption by the circuit ³³. It is possible that this could lead to bleeding complications over a longer period of support, though platelet counts did begin to recover later in the study period. Given the circuit and oxygenators used with the NOSA system had no evidence of significant clot burden, the etiology of this decline in platelets among the NOSA group is unclear and will be a focus of future studies of the NOSA system.

The primary potential toxicity of NO and the NOSA system is the formation of methemoglobin. Ferric (Fe³⁺) iron in methemoglobin has lower affinity for oxygen, and due to structural changes in hemoglobin it also increases the affinity in the remaining heme sites with the normal, ferrous (Fe²⁺) state within the heme tetramer, thus impairing its function to deliver oxygen to the tissues, resulting in a functional anemia. The NOSA group did have elevated methemoglobin levels (3.2±1.1%) through, but these were well below the toxic levels of 10–20% of total hemoglobin, the point at which patients typically become cyanotic and symptomatic ³⁴. Although the levels of methemoglobin did not continually increase throughout the study, it will be important to evaluate the trend in methemoglobin levels in future studies over multiple weeks, which would more closely approximate the anticipated duration of an AP course in clinical use.

Bleeding, thrombosis, and anticoagulation monitoring remain unsolved problems in all applications of ECLS. These data in a premature lamb model of the AP are likely generalizable to other modes of extracorporeal support and may have significant clinical impact. In particular, this technology would benefit ‘preemie ECLS’ since they currently have a higher incidence of bleeding due to prematurity ^2,35.

A limitation of this study is the inability to demonstrate an improvement over systemic anticoagulation in terms of bleeding. The 7 days of support were not long enough to allow bleeding complications to manifest in the control group. This is partly due to the fact that sheep are not an ideal model for IVH. They have a well-developed germinal matrix at the gestational age of the sheep used in this study; therefore, the incidence of IVH would likely be extremely low in this study group. The premature germinal matrix of a human fetus at 26–30 weeks gestation resembles that of a fetal sheep at 70 days gestation ^36,37, at which point the sheep is far too premature from a cardiopulmonary standpoint to tolerate delivery and AP support. The only evidence of hemorrhage and thrombosis was on post-mortem histologic evaluation. The animals in the NOSA group had a decreased incidence of microscopic renal and hepatic hemorrhage compared to the Heparin group with a limited incidence of focal hepatic microthrombi. The implications of these histologic findings are not known, as they did not manifest as clinical differences in organ function during the study. Both groups also showed focal pulmonary interstitial hemorrhage and peripheral thrombi. Since the study animals were not weaned from the AP, it is unknown if these findings would affect the lung function of the animals.

The study demonstrated that the NOSA system was effective at maintaining circuit patency for 7 days; however, the full duration of its efficacy was not evaluated. At the end of 7 days, the coating was still releasing NO at physiologic levels, suggesting that the coating would have remained effective for a longer duration. We have found in previous studies that the DBHD-N₂O₂ continues to release NO for over 14 days in a buffering solution ¹⁵. Future studies will be aimed at determining the longevity of the NOSA system.

One other limitation of this study is that two animals in the Heparin group used Pediatric Quadrox iD oxygenators whereas the Medos Hilite 800LT was used for all other animals in the study. This change occurred because the Medos Hilite 800LT became unavailable in the United States while the study was still ongoing. Both oxygenators are designed specifically for pediatric patients that use plasma-tight polymethylpentene hollow fibers. They are used clinically to provide ECLS support to neonates and children at comparable levels of circuit flow and oxygen delivery as in these studies. Both oxygenators have heparin-coated fibers. The Quadrox iD additionally uses a synthetic biopassive polymer coating. Despite these slight differences in fiber coating, they have been shown to be similar in terms of their effects on coagulation profiles and consumption of blood products ³⁸; therefore, the two subjects that used the Quadrox iD were included in most comparisons between the NOSA and Heparin groups. There are significant differences between the oxygenators, however, in terms of priming volume, maximum blood flow rate, and maximum gas flow rate. For that reason, these subjects were excluded from comparisons of oxygenator performance.

Conclusion

The NOSA system is a non-thrombogenic circuit that allowed the Artificial Placenta to support extremely premature sheep for 7 days without systemic anticoagulation. There was no evidence of thrombotic events nor significant systemic toxicity to the animal. The NOSA system may obviate the need for systemic anticoagulation with the AP and thus remove one of the major obstacles to clinical translation. It may also be applied to other forms of ECLS.

Impact:

• The Nitric Oxide Surface Anticoagulation (NOSA) system provides effective circuit-based anticoagulation in a fetal sheep model of the extracorporeal Artificial Placenta (AP) for 7 days.

• The NOSA system is the first non-thrombogenic circuit to consistently obviate the need for systemic anticoagulation in an extracorporeal circuit for up to 7 days.

• The NOSA system may allow the AP to be implemented clinically without systemic anticoagulation, thus greatly reducing the intracranial hemorrhage risk for extremely low gestational age newborns.

• The NOSA system could potentially be applied to any form of extracorporeal life support to reduce or avoid systemic anticoagulation.

Data Availability Statement:

The datasets generated and analyzed during the current study are available from the corresponding author (BPF, bfallon@med.umich.edu) on reasonable request.