Primary tracheal tumors are rare and comprise only 1% of all neoplasms. Among these only 25% are benign.8 Schwannoma is an uncommon benign tracheal tumor, and thus tracheal schwannomas are extremely rare.8 Since schwannomas are slow growing tumors of the trachea, they produce symptoms only when they are large enough to cause obstructive symptoms. Since they present with dyspnea and dry cough, they are often mistreated for years as asthma 9 or chronic obstructive pulmonary disease, until an endotracheal lesion is suspected due to unresponsiveness to inhaled medications or by the appearance of the flow volume loop or when the symptoms progress to stridor, like it did in our patient.
When near total obstruction of the tracheal lumen occurs, attempts at inserting an endotracheal tube or flexible fiberoptic bronchoscope may cause complete occlusion of the airway and respiratory arrest. Subsequent recovery from this acute deterioration may be very difficult with catastrophic outcomes.
The use of extracorporeal circuits to support pulmonary function during resection of a primary tracheal tumor was initially reported in 1961 but other reports with successful outcomes followed.4–7,10−13 This technique provides a method of maintaining normal gas exchange allowing time for resection and good surgical access even if complete tracheal obstruction occurs.11,14 CPB is well described as an adjunct to tracheal tumor resection but is rarely instituted under local anaesthesia and mild intravenous sedation.15,16 Two cases of thyroid lymphoma causing near complete airway obstruction have been reported where the use of a portable CPB circuit was used in one case and conventional femorofemoral VA CPB under local anaesthesia in the other to assist airway management and provide a safe resection.17,18 Chiu et al. and Misra et al. reported the use of temporary CPB using femoral artery and vein cannulation under spinal anaesthesia for tracheal resection and reconstruction.7,19 Also, Kar et al. reported the use of temporary CPB using right femoral artery and vein cannulation, for tracheal tumor resection, after a right femoral nerve block was administered.5
Despite the fact that CPB ensures adequate gas exchange, concerns have been raised regarding CPB-related complications, such as lower limbs ischemia, swelling, and the propensity to develop deep vein thrombosis. Major complications include intrapulmonary haemorrhage, coagulopathy, and neurological deficits especially if surgical dissection and bypass times are extensive.19–21
Our patient presented with progressively worsening dyspnea and audible stridor due to an intraluminal tracheal schwannoma causing near total obstruction. The use of cross-field ventilation, ventilation through a low-lying tracheostomy, or endobronchial ventilation was ruled out by the ENT surgeon, due to the site of the tumor near the carina and near total degree of tracheal lumen obstruction. Alternate methods of oxygenation like jet ventilation were also not considered because this may cause dislodgement of a part of the tumor or may change the position of the tumor leading to further luminal obstruction. Moreover, high‑frequency jet ventilation may cause inadequate time for expiration leading to further air trapping, pneumothorax, and worsening pneumomediastinum.5 Laser resection was not opted as any bleeding or edema during tumor manipulation can be catastrophic.
Therefore, it was decided to institute femorofemoral bypass in the semi-sitting position because of absolute inability of the patient to lie flat. Femorofemoral bypass in the semi-sitting position can have several problems like kinking of the venous cannula and inadequate venous return.5 In our case, even after making the patient’s position supine, we could not achieve full flow because of the preserved native cardiac output of the patient rather than a problem in the venous cannula resulting in inadequate venous return.
CPB provides the ENT surgeon sufficient time for adequate tumor resection. Most cases described in literature have a short bypass time. Establishing early ventilation once under CPB leads to early weaning from bypass.5
Currently, there is no clear consensus on the management of critical subglottic airway obstruction. In these cases, CPB and extracorporeal membrane oxygenation (ECMO) are reasonable methods to use.4 Two types of extracorporeal support are present, which are VA and venovenous (VV). VA extracorporeal support is described as the extracorporeal oxygenator is being parallel to the patient’s lungs and involves drainage of venous blood followed by oxygenation and return to a peripheral or central artery.4 VV extracorporeal support is described as the oxygenator is being in series with the patient’s lungs and involves taking venous blood either from the superior vena cava (SVC) or the inferior vena cava (IVC) followed by oxygenation and return to the other vena cava.4 VV extracorporeal support in the form of VV ECMO can be used in cases where there is adequate function of the heart and a need for ventilatory support only (e.g., in cases of respiratory failure).4 On the other hand, VA ECMO support can be used in cases where there is inadequate function of both the heart and lungs (e.g., in cases of cardiogenic shock).4,22
In our patient, VV ECMO support was initially considered using the femoral vein-right internal jugular vein (IJV) approach. However, the cardiothoracic surgeon estimated that our adolescent patient was very anxious, and it was believed that she would not cooperate with the insertion of a 19- or 21-Fr venous cannula in her neck using local anesthetic infiltration and mild intravenous sedation. Furthermore, the greatly increased cost of the VV ECMO circuit as compared to the traditional VA CPB circuit was another obstacle for its use in her case.
Our patient developed differential hypoxemia (Harlequin, or North-South, syndrome) while on femorofemoral VA CPB support after general anesthesia was induced. This is rare, as most cases reported in the literature describe Harlequin syndrome while on femorofemoral VA ECMO support, not on femorofemoral VA CPB support.7 An important difference between the two techniques is the venous reservoir, which is present in VA CPB support. It allows for complete heart emptying, ensuring that only the CPB circuit pumps all the cardiac output.7
Harlequin syndrome develops when native left ventricular (LV) function is preserved or is recovering while the lungs’ function is still poor (e.g., in respiratory failure) or mechanical ventilation is temporarily paused.22–24, 27 If peripheral femorofemoral VA ECMO is instituted in cardiogenic shock patients, oxygenated blood will be delivered mainly retrograde up to the aortic root by the arterial cannula in the common femoral artery. With native LV function improving, deoxygenated blood exiting the LV because of inadequate pulmonary gas exchange will be pumped back into the arterial circulation. This will create a watershed zone, with an upper body getting deoxygenated blood exiting the LV and a lower body getting oxygenated blood from the arterial cannula.22–24 The watershed zone site will be determined by the native LV function, with better native LV function resulting a watershed zone being more distal 25 (Fig. 3). In fact, Harlequin syndrome is a sign of recovery of native LV function while on femorofemoral VA ECMO support.22
Depending on the magnitude of native LV output, the location of the meeting zone or interface between retrograde flow and antegrade flow will vary, thus oxygenated blood exiting the femoral arterial cannula may not reach the aortic root and aortic arch. This results in coronaries and carotid arteries being perfused with deoxygenated blood, which leads to heart and brain hypoxia.7 This is characterised by normal oxygen saturations in the lower body and low oxygen saturations in the upper body. A right radial artery catheter can reliably determine the location of the watershed zone being distal or proximal to the brachiocephalic artery for adequacy of brain oxygenation by serving as a sampling site for arterial blood gas values. Furthermore, near infrared spectroscopy (NIRS) may be used to monitor brain oxygenation.4
Based on SpO2 values, ST segment depressions, and arterial blood gas findings in our patient we concluded that Harlequin syndrome is occurring with a watershed zone located in the aortic arch, somewhere between the innominate artery and the left subclavian artery. Despite the desaturation, a healthy circle of Willis in our young patient could have contributed to adequate bilateral cerebral oxygenation through her left common carotid artery. Although previous case reports described attempts of augmenting pump flows to overcome the desaturation,5 achieving full (100%) flows with peripheral femorofemoral VA CPB is not possible in the beating heart, even with the use of a multi-stage venous cannula designed to augment venous drainage. This is because beating hearts continue to receive and pump deoxygenated venous blood as the vena cavae are not clamped. In addition, augmenting pump flows, even for a brief period, trying to improve upper body oxygen saturations can be detrimental because this can significantly increase LV end-diastolic pressures leading to LV dilatation with subsequent acute pulmonary edema.22
In addition, when we use a single multi-stage femoral venous cannula in a normally functioning beating heart, venous drainage most often will be insufficient. In this situation, oxygenation of the deoxygenated venous blood passing through the heart can be done by placing an additional venous cannula (most often in the right IJV) that delivers oxygenated blood. This describes a venoarterial-venous (VA-V) CPB configuration, which necessitates precise regulation of pressures by clamps and flow sensors 29 and sometimes the use of two separate pumps.30 On the other hand, venous drainage is most often adequate in a normally functioning beating heart if two venous cannulae are properly positioned (most often in the femoral vein and IJV).7 This describes a venovenous-arterial (VV-A) CPB configuration. However, IJV manipulation was not feasible in our fully heparinized patient, with the neck draped and positioned for rigid bronchoscopy. Accurate positioning of two different wires for two venous cannulae requires either fluoroscopy or transesophageal echocardiography (TEE), both of which were not possible in the setting of severe respiratory compromise.
High-dose beta-blocker therapy with IV esmolol could have reduced her native LV output. This is useful in her situation by ensuring that the oxygenated blood pumped from the femoral arterial cannula can now reach the aortic arch and aortic root.22
Another possible way to ameliorate Harlequin syndrome was to convert to central arterial cannulation of the right axillary artery or ascending aorta, in order for oxygenated blood to be pumped directly into the aortic arch and aortic root.7 However, this requires surgical exposure of right axillary artery for direct cannulation or performing a sternotomy, respectively. In our patient, the safest and quickest approach to correct Harlequin syndrome was endotracheal intubation and lung ventilation. At this point, this approach was possible since the tracheal schwannoma was greatly downsized by the ENT surgeon.
VV ECMO support, such as with the femoral vein-right IJV approach, can prevent Harlequin syndrome from happening. In addition, VV ECMO support requires minimal anticoagulation and has lower incidences of retrograde arterial atheromatous embolization and lower limb ischemia, which are more commonly seen with VA extracorporeal support, whether by VA ECMO or VA CPB. However, VV ECMO is not capable of providing hemodynamic support 26 and may be complicated by recirculation syndrome, where oxygenated blood delivered to the IVC or SVC may be drained by the other venous cannula if both venous cannulae are misplaced too close to each other or if there is significant TR.24 In addition, VV ECMO circuits entail a much higher cost than traditional VA CPB circuits.
Other approaches for VV ECMO support also exist. In femorofemoral VV ECMO support, both femoral veins are percutaneously cannulated. One femoral vein is cannulated using a long single-stage venous cannula with its tip ending in the RA lumen, acting as a return cannula for oxygenated blood. The other femoral vein is cannulated with a long multi-stage venous cannula with its tip ending just beneath the RA-IVC junction. This approach is successfully used in patients requiring VV ECMO support but have no IJV access.7 Another approach for VV ECMO support would be to insert a dual-lumen cannula in the right IJV, under fluoroscopy or TEE guidance.7 All these approaches are safe and excellent for oxygenation support in such cases.
Termination of CPB was after tumor biopsy, downsizing, and safe endotracheal tube insertion, as the consensus was that it is safer to continue with CPB support for the whole surgical time period. Although continued CPB support requires full heparinization, fortunately this did not lead to further surgical bleeding. However, whenever possible, all efforts must be made to terminate CPB support after achieving a definitive airway to avoid further surgical bleeding and postoperative coagulopathy.7 Normothermia is maintained, if possible, during the whole duration of CPB support in an attempt to terminate CPB quickly and ameliorate postoperative coagulopathy.7
Spinal anesthesia can be used for cannulation of the femoral vessels in CPB surgeries.7,19 However, this is controversial and uncommonly used by most anesthesiologists who prefer to avoid neuraxial procedures in these cases. Guidelines by the American Society of Regional Anesthesia and Pain Medicine recommend that 1 hour at least should pass before full systemic heparinization for CPB after an atraumatic neuraxial procedure has been performed.28 Also, it recommends limiting unfractioned heparin doses after performing a neuraxial procedure; therefore, VV ECMO or VA ECMO support are preferred over VA CPB support, as they require only partial heparinization.