Surgical procedure
Time-dated pregnant ewes were used at gestational ages of 104 to 135 days (term is ∼145 days). Animals were treated according to approved protocols by the institutional animal care and use committee of The Children’s Hospital of Philadelphia Research Institute.
Ewes were anaesthetized with 15 mg kg−1 of intramuscular ketamine, with maintenance of general anaesthesia with inhaled isoflurane (2–4% in O2) and propofol (0.2–1.0 mg kg−1 min−1). Intraoperative haemodynamic monitoring included pulse oximetry, with a constant infusion of isotonic saline administered via a central venous line placed in a jugular vein to maintain maternal fluid balance. A lower midline laparotomy was created to expose the uterus, with a small hysterotomy performed to expose the fetal sheep head and neck (CA/JV) or umbilical cord (UA/UV). Experimental lambs undergoing cannulation of the neck vessels (CA/JV and CA/UV) underwent creation of a small right neck incision to expose the jugular vein and/or carotid artery. Fetuses received one intramuscular dose of buprenorphine (0.005 mg kg−1). After determination of the maximal cannula size accommodated by each vessel, ECMO cannulae were placed (8–12 Fr, Medtronic, Minneapolis, MN, USA), with stabilizing sutures placed along the external length of cannulae at the neck. Cannulas were customized with a silicone sleeve over the external portion of the cannulas to permit increased tension of the stabilizing sutures in CA/JV and CA/UV experiments. Experimental lambs undergoing cannulation of the umbilical vessels were positioned to expose the umbilical cord, with connective tissue sharply dissected to expose the umbilical arteries and veins. Umbilical cannulae were placed in one umbilical vein (CA/UV) as well as two umbilical arteries (UA/UV) (12 Fr, Medtronic, or modified 8–12 Fr custom-made cannulas), with stabilizing sutures placed at the insertion sites.
Following construction and blood priming of the oxygenator circuit as described below, connection of the cannulas to the circuit was performed under continuous ultrasonographic visualization of the fetal heart. Occlusion of the umbilical cord was performed immediately following establishment of blood flow through the circuit, with administration of additional blood volume in a subset of animals (CA/JV) demonstrating poor cardiac filling immediately after establishment of circuit flow. Subsequently, fetal lambs were weighed and transferred to a sterile fluidic incubator for further management as described.
To generate baseline data of the ovine fetus in utero, two time-dated pregnant ewes at 118 days of GA underwent laparotomy for implantation of fetal vascular catheters and electrodes41. Insulated multi-stranded stainless steel wire ocular EMG electrodes were implanted subcutaneously in the superior and inferior margins of the muscle overlying the orbit of one eye41. Briefly, after induction of general anaesthesia and exteriorization of the fetus, catheters were implanted in the fetal carotid artery and jugular vein, in addition to a reference catheter placed in the amniotic sac, followed by placement of EMG wire electrodes as described above. Fetal catheters and electrodes were exteriorized through the maternal flank, and the uterus and abdomen were closed. Following a 48–72 h recovery period, ewes were transferred to a holding cage for fetal monitoring. Monitoring of catheterized fetal lambs in utero was started 48–72 h after surgery and continued in 24 h intervals on alternating days until completion of the experimental protocol at 140 days of GA. Fetal ocular EMG (1,000 Hz) and arterial pressures (400 Hz) corrected for amniotic fluid pressure were continuously recorded (LabChart 5, ADInstruments Inc., Colorado Springs, CO, USA).
Circuit
The pumpless circuit consisted of a low-resistance hollow fibre oxygenator (Quadrox-ID Pediatric Oxygenator, Maquet) connected to ECMO cannulae (Medtronic) or custom-made umbilical cannula via 3/16′ ID × 1/16′ wall thickness BIOLINE-coated tubing (Maquet). In later studies utilizing smaller lambs, a smaller oxygenator was utilized (Quadrox-ID Neonatal Oxygenator, Maquet). Connections were established as an arterial–venous extracorporeal oxygenation circuit, with the carotid artery or umbilical arteries providing inflow to the oxygenator (CA/JV or UA/UV) connected to the oxygenator inflow port and the jugular vein or umbilical vein (CA/JV or UA/UV) providing outflow from the oxygenator and connected to the oxygenator outflow port. Total priming volume was 81 ml of maternal blood for the large oxygenator and 38 ml for the smaller oxygenator. Circuit flow was continuously measured (HT110 Bypass Meter and HXL Tubing Flowsensor, Transonic Systems Inc., Ithaca, NY, USA) and sweep gas supplied to the oxygenator was a blended mixture of medical air, nitrogen and oxygen titrated to achieve fetal blood gas values (target PaO2 20–30 mm Hg, target PaCO2 35–45 mm Hg).
Fluid incubation
The first CA/JV studies were performed in a 30-litre heated stainless steel reservoir filled with sterile synthetic amniotic fluid (‘still reservoir’, pilot study), later expanded to a 40-litre polycarbonate tank with continuous recirculation of fluid through a series of sterile filters. Subsequent fluidic incubators were based on a model of continuous exchange of warmed sterile fluid (temperature 38.5–40.5 °C), with inflow tubing mounted on a double-head peristaltic pump and gravitational outflow to facilitate continuous fluid turnover. Fetal lamb enclosures within this system included a 60-litre customized glass tank (CA/JV series one) (NDS Technologies, Vineland, NJ, USA), and individually customized bag enclosures of 2- to 4-litre total volume initially comprised silver-based antimicrobial polyethylene film (CA/UV and UA/UV studies) (Wiman Custom Films & Laminates, Sauk Rapids, MN, USA) and later the same film without silver impregnation. Synthetic amniotic fluid was composed of a balanced salt solution containing Na+ (109 mM), Cl− (104 mM), HCO3− (19 mM), K+ (6.5 mM), Ca2+ (1.6 mM), pH 7.0–7.1, osmolarity 235.8 mOsm kg−1 water. The rate of fluid inflow (HT110 Bypass Meter and HXL Tubing Flowsensor, Transonic Systems Inc.) and internal fluid temperature (MLT415/A, ADInstruments) were monitored continuously.
Fetal lamb maintenance on circuit
Following stabilization and transfer of animals to the fluid incubator, a continuous infusion of heparin (10–400 USP units per hour) and prostaglandin E1 (0.1 μg kg−1 min−1) were administered intravenously. Heparin dosing was titrated to reach a target activated clotting time of 150–180 s.
Arterial and venous blood were analysed every 1–8 h for blood gas, electrolyte and coagulation values (i-Stat System, Abbott Point of Care Inc., Princeton, NJ, USA) and oxygen saturation (Avoximeter 1000E, Accriva Diagnostics, San Diego, CA, USA). Stored whole maternal blood was transfused as required (10–20 ml kg−1) to maintain fetal Hgb levels above 9 g dl−1. In a subset of lambs (Prototype IV lambs 4–6), erythropoietin (400 U kg−1) was administered intravenously once daily to promote fetal erythropoiesis, and was held for Hgb >16 g dl−1.
Analgesics (buprenorphine, 0.005 mg kg−1 intravenously every 3–5 h as needed) and anxiolytics (propofol, 0.1–0.5 mg kg−1 min−1) were administered during periods of perceived fetal agitation (restless repetitive fetal movements, excessive swallowing, tachycardia and hypertension). This was primarily required in the CA/JV and CA/UV animals with markedly reduced sedation requirement in the UA/UV animals.
Total parenteral nutrition was administered throughout the duration of fetal incubation as described: (CA/JV: amino acids (TrophAmine 10%, 3.5 g kg−1 per day), lipids (Intralipid 20%, 2–3 g kg−1 per day) and dextrose (10.0–12.5 g kg−1 per day) to a total caloric goal of 80 kcal kg−1 per day; CA/UV: amino acids (TrophAmine 10%, 3 g kg−1 per day), lipids (Intralipid 20%, 1–2 g kg−1 per day for lambs 1–3, and 0.1–0.2 g kg−1 per day for lambs 4 and 5), dextrose (titrated to blood glucose target 30 mg dl−1) and iron (1 mg kg−1 per day); UA/UV: amino acids (TrophAmine 10%, titrated to blood urea nitrogen target level 30 mg dl−1), lipids (Intralipid 20%, 0.1–0.2 g kg−1 per day), dextrose (titrated to blood glucose target 30–40 mg dl−1)and iron (1.0–1.5 mg kg−1 per day, titrated to plasma iron target 200–300 μg dl−1)).
Cardiac ultrasound was performed one to two times daily by a fetal echocardiographer. Measured parameters included right ventricular (RV)/left ventricular (LV)/combined cardiac outputs (CCO), ductus arteriosus flow and proximal right pulmonary artery pulsatility index. Control echocardiography data were obtained in pregnant anaesthetized ewes at 109 days and 135 days of GA.
Data acquisition and formulas
Fetal blood pressure, heart rate, circuit blood flow rates, transmembrane pressure differential, sweep gas flow and incubator fluid temperature were continuously recorded (LabChart 7, ADInstruments Inc.).
Post-membrane oxygen content=(1.34 × Hgb × post-membrane oxygen saturation)+(0.0031 × post-membrane PaO2).
Pre-membrane oxygen content=(1.34 × Hgb × pre-membrane oxygen saturation)+(0.0031 × pre-membrane oxygen PaO2).
Weight-adjusted circuit flow=absolute circuit flow/estimated daily weight.
Oxygen delivery (ml kg−1 min−1)=weight-adjusted circuit flow × post-membrane oxygen content.
Oxygen consumption (ml kg−1 min−1)=weight-adjusted circuit flow × (post-membrane oxygen content−pre-membrane oxygen content).
Oxygen extraction (%)=(oxygen consumption/oxygen delivery) × 100%.
Estimated daily weight. Growth rate was assumed to be exponential and derived from measured body weight at the start and end of each run (according to the formula y=aebx, where ‘a’ is starting weight and ‘b’ is growth rate in g kg−1 per day). Estimated daily weights (for weight-adjusted calculations) were extrapolated from the exponential growth rate calculated for each lamb.
Control growth rate. Initial body weights (at time of delivery from uterus) of Prototype III/IV experimental lambs and late-gestation control lambs were plotted against gestational age. Exponential regression analysis was used to determine control growth rate in utero as well as the estimated fetal weight for the expected biparietal diameter calculations.
Decannulation and mechanical ventilation
Following completion of the incubation period, animals were transitioned from the fluid bath, with endotracheal intubation and suctioning to remove excess fluid from the lungs. Surgical decannulation of the carotid artery and/or jugular vein (CA/JV, CA/UV) was performed under general anaesthesia with inhaled isoflurane (2–4% in O2) and propofol (0.2–1.0 mg kg−1 min−1). Patent umbilical vessels were clamped and divided (CA/UV, UA/UV and control lambs). One umbilical artery or carotid artery was catheterized to enable blood gas measurement. Anaesthesia was then reversed and animals were maintained on mechanical ventilation with intermittent arterial blood gas sampling using an i-Stat System (Abbott Point of Care Inc.). A group of normally grown control lambs (N=4) were delivered via hysterotomy at 140–141 days of GA, arterial and venous cannula placed for serial blood sampling and fluid/drug administration, respectively, orally intubated and ventilated.
Lambs were maintained on synchronized intermittent mandatory ventilation, with FiO2 titrated to PaO2>60–80 mm Hg, peak inspiratory pressure titrated to tidal volume 6–8 ml kg−1 body weight, respiratory rate titrated to pH 7.4 (if possible) and positive end-expiratory pressure (PEEP) maintained between 5 and 7 mm Hg. The general goal was to wean ventilator support, and hence lambs were sedated only as needed to optimize respiratory performance, and were often allowed to breathe spontaneously in addition to receiving mandatory ventilator-triggered breaths. Arterial blood gases were obtained every 1–4 h to assess pulmonary gas exchange function.
We could not reliably quantify the degree to which spontaneous, unsupported breaths contributed to oxygenation index (OI=(FiO2× mean airway pressure)/PaO2) and ventilation efficiency index (VEI=3,800/(peak inspiratory pressure × respiratory rate × PaCO2)), and hence absolute values of OI and VEI were invalid measures for comparing pulmonary function between lambs in this study. We used relative OI and VEI calculations to determine which blood gas values (at given ventilator settings) corresponded to each animal's ‘peak’ oxygenation and ventilation (lowest OI was point of peak oxygenation, and highest VEI was point of peak ventilation). Peak oxygenation and ventilation did not necessarily occur at the same time point.
Post-mortem
Body, brain and lung weights were recorded. The lungs were inflation fixed (30 cm H2O) via the trachea with 10% formalin. When the fixation pressure had reached a plateau, typically within 20 min, the trachea was occluded and the lungs were submerged in buffered fixative stored at room temperature for 7–10 days. Brains were submerged in 10% formalin for 7–10 days.
Stereological analysis
Lung volume (VL) was estimated using the water displacement58. Two sections of lung tissue were obtained from the lower lobes, dehydrated through a series of graded alcohol solutions, embedded in paraffin, sectioned at 3 μm and stained with haematoxylin and eosin. Lung images (N=10 per animal) devoid of major airways and blood vessels were visualized using a Toshiba 3CCD camera interfaced with a Leica DMRD microscope and an Apple G4 computer. A transparent multipurpose test lattice consisting of 216 test points and a discontinuous series of line probes was placed over the screen images. The number of test points overlying tissue, and number of line probe intercepts with the luminal surface were used to calculate lung tissue fraction and luminal surface density (SV) respectively. Luminal surface area of the lungs (SL) was calculated according to the equation: SL=SV× VL. Septal wall thickness (TW) was calculated according to the following equation: TW=Vpa,tiss/SL; where Vpa,tiss is parenchymal tissue volume25,26.
Histologic analysis and myelin quantification
Post-therapy and control brain tissues were sectioned into anterior and posterior regions for each hemisphere. Haematoxylin and eosin and Kluver and Barrera Luxol fast blue staining was performed as detailed in ‘Laboratory Methods in Histotechnology’ (Armed Forces Institute of Pathology, Washington, DC)38,39. Slides were digitally scanned at 20 × magnification and evaluated using both Aperio Imagescope Version 12.3 (Leica Biosystems Pathology Imaging, Buffalo Grove, IL, USA) and standard light microscopy. Individual gyri were digitally measured at maximum width on each slide. Myelin analysis and the calculation of myelin-positive pixel ratios was performed using Positive Pixel Count Version 9 (Aperio Technologies, Leica Biosystems) A uniform algorithm adjusted to identify pixels consistent with positive staining was applied to experimental and control slides. Maximum positivity by region was determined by limiting density analysis to multiple areas 1 mm2 in size where the Luxol fast blue staining was strongest40.
Statistical analysis
Haemodynamic parameters (heart rate, mean pre-membrane pressure and circuit flow) and arterial blood gas parameters were averaged over 12 and 24 h, respectively, and analysed using one within-group (time) and one between-group (treatment) analysis of variance. When significant differences between means were detected, multiple comparison analysis was performed using least significant difference test (SPSS Version 23, IBM Corp., Armonk, NY, USA). Epithelial SP-B cell density, biparietal diameter, brain-to-body weight ratios, myelination density and fetal breathing responsiveness to PaCO2 were compared using Student’s unpaired t-test (SPSS). Body weight data were fitted to nonlinear regression curves, the rate constants of which were analysed between groups (GraphPad Prism Version 6, GraphPad Software, San Diego, CA, USA). Significance was accepted at P<0.05. All data are presented as mean±s.e.m.
Data availability
The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information Files or from the corresponding author on reasonable request. The in utero control data in Figs 3, 4, 5, 6 were obtained from published work18,20,24,36,37,59.