Pulmonary Hemorrhage Associated with Negative-pressure Pulmonary Edema

An otherwise healthy 33-yr-old woman underwent elective outpatient septoplasty at our institution for relief of chronic nasal obstructive symptoms. The patient quit smoking 2 yr before the surgery. She had previously undergone general anesthesia once, without complications. There was no family history of difficulties with anesthesia. The patient’s height was 160 cm, she weighed 70 kg, and results of her physical examination were normal.

After premedication with midazolam, anesthesia was induced with use of propofol, lidocaine, and fentanyl intravenously, followed by rocuronium to facilitate tracheal intubation. Anesthesia was maintained with sevoflurane in 70% nitrous oxide and 30% oxygen. The surgery was uneventful. As the procedure concluded, nasal packing was inserted, neostigmine and glycopyrrolate were administered, and administration of inhalational agents was discontinued. The patient initiated spontaneous ventilation but began coughing vigorously. She reached for the endotracheal tube and was extubated subsequently. Her ventilatory efforts resulted in chest wall retraction, without apparent air movement for approximately 45 s. Continuous positive airway pressure with 100% oxygen was applied using a face mask. This failed to relieve the laryngospasm, and 30 mg succinylcholine was administered intravenously. The patient was reintubated without trauma to the trachea. Neither gastric contents nor blood was present in the oropharynx. Oxygen saturation measured by pulse oximetry (Spo²), which had decreased to approximately 70% before administration of succinylcholine, improved after intubation; however, 8 cm H²O positive end-expiratory pressure (PEEP) and 100% oxygen were needed to increase Spo²to 100%. Blood pressure decreased to 77/50 mmHg but returned to 100 mmHg systolic after administration of 5.0 mg intravenous ephedrine and a total of 900 ml crystalloid. Bilateral rhonchi were audible during auscultation and a frothy serosanguinous fluid was suctioned from the endotracheal tube. Within 5 min, this fluid became progressively more bloody. A chest roentgenogram obtained in the recovery room showed perihilar interstitial and alveolar opacification consistent with pulmonary edema, with a normal-sized heart (fig. 1). Peribronchial cuffing and air bronchograms were noted in the upper lung fields. The first arterial blood gas measurement obtained within 20 min of the initial event showed a pH of 7.34, an arterial carbon dioxide tension (Paco²) of 39 mmHg, and an arterial oxygen tension (Pao²) of 61 mmHg, with a fractional inspired oxygen tension (Fio²) of 1.0. As neuromuscular blockade regressed, the patient experienced paroxysmal coughing, which seemed to exacerbate pulmonary bleeding. She was administered 25–35 μg · kg⁻¹· min⁻¹propofol by continuous infusion, 4 mg pancuronium bromide for neuromuscular blockade, and 10 mg furosemide. Mechanical ventilation was reinitiated. She was transferred subsequently to the intensive care unit with an Fio²of 0.6, a PEEP of 8.0 cm H²O, and a respiratory rate of 10 breaths/min. With these settings, arterial blood gas measurements showed a pH of 7.35, a Paco²of 40 mmHg, and a Pao²of, again, 61 mmHg. During the next 4 h, respiratory and cardiovascular conditions worsened, necessitating increased Fio²to 1.0, a PEEP of 15 cm H²O, and vasopressor support with use of 8.0 μg · kg⁻¹· min⁻¹dopamine. The patient’s condition and course were consistent with alveolar hemorrhage (fig. 2). Pulmonary and radial arterial monitoring were performed. Initial pulmonary arterial readings included cardiac output, 4.9 l/min; pulmonary arterial pressure, 29/17 mmHg; pulmonary arterial occlusion pressure, 9 mmHg; and systemic vascular resistance, 950 dyne · s · cm⁻⁵. Transthoracic echocardiography showed an ejection fraction of 63% and normal chamber size. Empiric antibiotic coverage with alatrofloxacin was initiated. The patient continued to experience pulmonary hemorrhage. Her hemoglobin concentration decreased by 3.0 g/dl to 7.7 g/dl over 7 h. Bronchoscopy and open-lung biopsy were thought to be contraindicated because of the patient’s respiratory status, which necessitated a PEEP of 15 cm H²O and an Fio²of 0.85 to yield a Pao²of 61 mmHg. She continued to bleed extensively, resulting in transfusion of 2 units of packed erythrocytes. On the second day in the hospital, she continued to have extensive pulmonary bleeding. Decreasing arterial oxygenation necessitated an increase in Fio²to 1.0. In addition to general supportive measures, administration of cyclophosphamide, corticosteroids, and plasmapheresis was started for presumed Wegener vasculitis, capillaritis, or Goodpasture syndrome. On the fourth postoperative day, the patient’s condition began to show marked improvement. Oxygenation improved dramatically, and Fio²was reduced to 0.4. Neuromuscular blockade was discontinued, sedation was decreased, and ventilatory support was reduced. On the sixth postoperative day, she was extubated. Tests for human immunodeficiency virus, lupus anticoagulant, antinuclear antibodies, antineutrophil cytoplasmic antibodies, and anti–glomerular basement membrane antibodies all yielded negative results, and multiple blood and urine cultures had no microbial growth. Plasmapheresis and immunosuppressants were discontinued. She was discharged from the hospital after 2 weeks, with normal results of chest roentgenographt. Follow-up at 21 months showed no functional deficit, and, therefore, no pulmonary function tests were performed.

We describe the case of a 33-yr-old woman whose routine surgery with general anesthesia was complicated by laryngospasm, followed by NPPE and alveolar hemorrhage. Aspiration, blood from the surgical site, fluid overload, and anaphylactoid reaction were ruled out as possible causes. The initial chest roentgenogram, arterial blood gas, and presentation after nasal surgery were characteristic of NPPE, except for the presence of pulmonary hemorrhage. Although a possible cause could have been disruption of bronchial vessels, which has been implicated as the cause of frank hemorrhage, 1,2worsening oxygenation, continued hemorrhage with a decreasing hematocrit concentration, and alveolar consolidation shown by chest roentgenography indicate alveolar hemorrhage.

Laryngospasm is a common complication after tracheal extubation, occurring in as many as 8–237 of every 1,000 anesthetic procedures. 3–5NPPE has been reported in 11% of healthy young patients who experience laryngospasm. 6These data suggest that the incidence of NPPE may be as high as 1 in 1,000 anesthetic cases. Despite these statistics, the presence of pulmonary hemorrhage complicating NPPE has not been extensively reported.

Frank pulmonary hemorrhage associated with NPPE is rare, with only one report in the English literature. 7Two other case reports describe bleeding from bronchial vessels. 1,2Goldenberg et al.  8reported the development of tricuspid and pulmonary insufficiency in two of six patients with postlaryngospasm NPPE. The current patient had only isolated, mild tricuspid regurgitation shown by echocardiographic examination. This may have been the result of higher pulmonary artery pressure development secondary to hypoxemia because she required increasingly higher levels of Fio²and PEEP.

Negative-pressure pulmonary edema is the result of three pathophysiologic processes: negative intrathoracic pressure, 6increase of systemic vascular and pulmonary capillary hydrostatic pressure, and mechanical stress at the alveolar–capillary level. 9First, negative intrathoracic pressure increases venous return by reduction in right atrial pressure, with a concomitant increase in pulmonary arterial and pulmonary capillary hydrostatic pressures, complicated by a reduction in perivascular interstitial hydrostatic pressure. 9This causes an increased transcapillary pressure gradient, favoring the transudation of fluid into the interstitial space. In addition, increased central venous pressures impede passive lymphatic blood flow. 10Second, pulmonary blood volume is augmented by increases in systemic pressure secondary to the release of norepinephrine in response to hypoxia, hypercapnia, and anxiety. 7The effect is distention of the right ventricle, leading to interventricular septal shift and a resultant reduction in cardiac output. 7The increase of systemic vascular resistance further increases left ventricular wall tension, which contributes to impediment of left ventricular ejection. 11However, the classic description of NPPE includes serosanguinous or pink, frothy fluid, which implies some capillary leak of proteinaceous material. Therefore, the third invoked process indicates that this capillary leak could be a result of mechanical failure of the alveolar–capillary membrane, resulting in alveolar edema or frank hemorrhage. 12,13The actual negative-pressure threshold for pulmonary microvasculature disruption is not known in humans. In a rabbit model, this stress failure occurs at a microvasculature pressure of approximately 40 mmHg. 14Marked negative intrapleural pressures have caused increased capillary permeability in a rabbit model of reexpansion pulmonary edema. 15 

Laryngospasm after tracheal extubation was observed in the patient described, and the subsequent development of NPPE was not unexpected. However, the dramatic and prolonged pulmonary hemorrhage forced us to consider other causes. These included Wegener granulomatosis, Goodpasture syndrome, capillaritis, vasculitis, and infection. Each of these potential causes was treated empirically in this patient, although each was ruled out subsequently by studies that yielded negative results. This patient shows that frank pulmonary hemorrhage can complicate NPPE after laryngospasm.