A fresh frozen cadaver including head, neck, chest, and abdomen and pelvis was obtained via our hospital’s anatomical bequest program (Mayo Clinic Department of Anatomy, Rochester, Minnesota, USA), which had previously been exsanguinated but not chemically preserved. The specimen was originally approximately 80 kg and 180 cm in height, which yielded an approximately 50 kg cadaveric specimen. In order to simulate cardiac and pulmonary blood flow, a modification was employed to reverse the normal flow in cardiac bypass (Fig. 1), whereby venous return cannula were placed in the aorta and arterial supply cannula in the inferior vena cava (IVC). In comparison, standard cardiopulmonary bypass replaces the heart’s pumping function with forward arterial flow moving through a cannula in the ascending aorta and venous flow exiting the body from cannulas in the right atrium or superior and IVC. This configuration allows the heart to be stopped to permit cardiovascular surgery. Instead of bypassing the heart, we sought to utilize the external pump to allow selective perfusion of the heart and lungs with relative exclusion of the peripheral organs, thus mimicking forward cardiac flow through the right and left heart and pulmonary vasculature in order to allow detection of entrained air.
All major extremity vessels were ligated to prevent loss of intravascular volume after initiation of flow. In the upper extremity, this included bilateral brachial arteries and veins as well as cephalic and basilic veins. In the lower extremity this included bilateral femoral arteries and veins. We then performed a midline abdominal incision from xiphoid to pubic symphysis followed by bilateral subcostal extension. An umbilical hernia with mesh was noted on abdominal entry but other than this, no signs of surgical intervention were noted. The specimen had a fully intact diaphragm and esophagus with a small type 1 paraesophageal hernia. There were grossly normal stomach, spleen, liver, gallbladder, greater omentum, small bowel and colon. In serial, we then preformed 3 dissection techniques, moving from two decreasingly peripheral cannulation approaches to last an intrathoracic approach via thoracotomy.
In order to simulate respiratory variation, the cadaver was intubated with an 8.0 mm endotracheal tube using direct laryngoscopy and connected to the Hamilton G5 mechanical ventilator (Hamilton Medical, Bonaduz, Switzerland), the ventilator was set to Continuous Mandatory Ventilation (CMV) mode ventilation with a tidal volume of 500 mL (approx.. 6.8 mL/kg IBW), a rate of 15 breaths per minute, positive end expiratory pressure (PEEP) of 5 cmH2O and FiO2 of 21%. Prior to initiating flow we were able to ventilate with nominal peak pressures of 21 cmH2O with minimal air leak.
Exposure Techniques
In the initial first extrathoracic approach, colon and small bowel were retracted cephalad, and the rectum was retracted laterally. The midline inframesocolic space and the inferior border of the root of the mesentery was identified and the bifurcation of the aorta was palpable as it approached the sacral promontory. The retroperitoneal fat was dissected down to the level of the aorta and IVC. The dissection was carried inferiorly exposing the bifurcation and iliac arteries and veins. These vessels were ligated with 0 silk sutures. The dissection was then carried cephalad exposing the great vessels up to the level of the crossing of the renal vein. The inferior mesenteric artery and vein were noted at this point and ligated. A double purse string technique with 3 − 0 Prolene suture was used on both the aorta and IVC just superior to the bifurcation. Incisions in the great vessels were performed and the cannulas were placed (see Fig. 2.) A 20 French aortic cannula EZ glide (Edwards Lifesciences, Irvine, California, USA) and 32 French bullet venous cannula (Medtronic Inc., Minneapolis, Minnesota, USA) were selected and the arterial cannula was placed in the IVC, and the venous cannula was placed in the aorta. The cannulae were connected to saline primed 9.5 mm (3/8 in.) tubing which was connected to the CentriMag centrifugal pump (Abbott Laboratories, Abbott Park, Illinois, USA.) After clearance of intravascular debris, and switching to a roller pump we achieved a pump flow of 1.56 L/min, at 249 RPM, with 52 mmHg of positive pressure and − 32 mmHg of negative pressure under vacuum assist with continuous normal saline infusion.
After approximately 10 minutes of bypass, mechanical ventilation pressures increased which necessitated suctioning of fluid in the endotracheal tube. We then clamped the endotracheal tube to limit fluid egress as we hypothesized that the little of oncotic pressure from saline circulation alone in conjunction with tissue destruction from high flow rates allowed rapid flow of fluid transvascularly. We also observed significant periorbital and facial edema, skin edema in the chest and abdomen, visceral edema with a significant increase in size and volume of the liver, stomach, spleen and especially small bowel and colon. We also noted that at this time 15 L of fluid had been lost from circuit based on input volumes into the patient, which was consistent with observed end organ engorgement via venous-to-capillary bed extravasation with reduced return to the bypass circuit.
In our second approach, we undertook more central but still extrathoracic cannulation to reduce volume losses to the periphery with improvement in the qualitative degree of volume loss. We started by eviscerating the abdomen (Fig. 3). A right and left medial visceral rotation were performed and the entire colon was mobilized. The root of the mesentery was dissected, and the superior mesenteric artery and vein were identified and ligated. A Kocher maneuver exposed duodenum. The rectum and third part of the duodenum were ligated and the intestines were removed from the abdomen. The right and left triangular ligaments of the liver were mobilized to the hepatic veins which were ligated and portal structures were identified. The splenic attachments were mobilized, and the esophagus was circumferentially dissected to the level of the aorta. The celiac axis was identified from the left and further dissected through the lesser sac and ligated. The liver, spleen, pancreas, stomach, and omentum were removed from the abdomen leaving retroperitoneal structures including the great vessels, kidneys, adrenals, and stump of the esophagus. At this point, double purse string sutures with 2 − 0 Prolene were placed around the right left and middle hepatic veins in the IVC and the 20 French forward flow cannula was once again inserted and secured with a Rumel tourniquet with the tip of the cannula in the right atrium. The inferior aspect of the IVC was dissected circumferentially and ligated with a 0 silk suture to prevent reverse flow. The 32 French receiving cannula was left in its insertion point, but advanced further into the thoracic aorta and once again secured with a Rumel tourniquet. We again established flow at 1.5 L/min. at the previous settings now in the absence of mechanical ventilation but again with considerable volume losses.
A third exposure was completed to attempt to eliminate volume loss into the superior vena cava (SVC). A midline sternotomy was performed and a sternal retractor was placed. The pericardium was opened exposing the aortic root and pulmonary artery down to the level of the diaphragm. (Fig. 4.) The IVC and aortic cannulas were palpated and found to be in acceptable position. The right and left pleural space were opened revealing multiple liters of extravascular fluid which had extravasated during the previous approaches. To direct the flow of air and fluid through the IVC into the right atrium and into the right ventricle and pulmonary artery (which is the normal path of a VAE), the SVC and IVC (around the instillation cannula) were dissected and snared with Rumel tourniquets which blocked the SVC and retrograde flow back down the IVC pushing all blood into the right atrium. We then moved the venous return canula and inserted it into the distal ascending aorta and secured with a purse string suture. The aorta distal to this was then clamped to isolate flow out of the heart to the cannula alone. Flow was again established at 1.5 L/min in the same fashion as before.
Measuring Venous Air Embolism through Various Pathways
In order to demonstrate viability of this cadaveric simulation model, a digital oscilloscope was used to measure impedance across multiple bipolar montages using needle tips in combinations of the skin, esophagus, diaphragm, and pulmonary artery. Impedance waveforms were recorded after delivering 30 to 60 mL boluses of air during each simulation run via independent vascular access in the IVC. At the end of our assessment period, a hole was noted in the RV which most likely occurred due to the high pulmonary artery resistance from the cadaveric lung tissue and a forward pressure regularly and often above 50 mmHg. At the conclusion of simulation, the volume of normal saline infused through all simulations were 46 L.