Prolonged extracorporeal membrane oxygenation for pediatric necrotizing pneumonia due to Streptococcus pneumonia and influenza H1N1 co-infection: how long should we wait for native lung recovery?

A 3-year-old boy (body weight 17 kg) presented to a regional hospital with a 5-day history of cough and febrile illness. He was diagnosed with left-sided pneumonia and started on intravenous antibiotics (Ampicillin). The boy was previously healthy and up to date with his immunizations (including Streptococcus pneumonia). Due to worsening respiratory distress he was transferred to our pediatric intensive care unit (PICU) the next day. He developed acute respiratory failure, followed by endotracheal intubation and invasive ventilation. The chest X-ray (CXR) showed an effusion and pneumothorax on the left side which required the placement of two chest drains. Streptococcus pneumonia was isolated from the chest drain fluid and Influenza H1N1 from a tracheal aspirate via polymerase chain reaction (PCR). A chest computer tomography (CT) showed a severe necrotizing pneumonia of the left lung with destruction of lung parenchyma and formation of a large abscess cavity in the lower lobe, as well as consolidation of the right lung (Fig. 1). A trial of high frequency ventilation and inhaled nitric oxide did not improve oxygenation and was abandoned. Despite ventilation with high peak inspiratory pressure (PIP = 40 cm H₂O), high positive end-expiratory pressure (PEEP = 10 cm H₂O) and FIO₂ 100%, hypoxemia and hypercapnia became worse (oxygen saturation (SpO₂) = 75–80%, arterial blood gas: pH = 7.28, paO₂ = 38 mmHg, paCO₂ = 127 mmHg), thus fulfilling the criteria for severe ARDS (PaO₂/FiO₂ = 38 mmHg, oxygenation index = 52.6) after excluding acute cardiac dysfunction. Consequently, venovenous extracorporeal membrane oxygenation (VV-ECMO) was implemented on day 19 of mechanical ventilation. A 15 French access cannula was placed into the left femoral vein and a 14 French return cannula into the right internal jugular vein. Initial ECMO settings were blood flow of 2 L/min, RPM 3000 and gas sweep of 1 L/min which immediately increased his SpO₂ to 93%. Mechanical ventilation was reduced to lung rest settings (PIP 20 cm H₂O, PEEP 10 cm H₂O, respiratory rate 10/min, FIO₂ 30%). Over the next 4 weeks our patient showed basically no lung aeration, depending completely on extracorporeal support. His tidal volumes (TV) were 5–8 mL, which is less than 0.5 mL/kg (Fig. 2). Additionally, due to insufficiency of the venous cannula we had to reduce flows and accepted SpO₂ around 75% while keeping hemoglobin levels over 14 g/dL. Next, we started several measures to promote lung recovery, namely daily prone positioning for 12–16 h, twice daily inhalation with DNase and 2–3 toilet bronchoscopies per week to remove necrotic endobronchial material and to reduce its viscosity. This led to an increase in TV to 35 mL (2 mL/kg) by week five, but finally to 100 mL (6 mL/kg) on day 41, together with areas of re-aeration on CXR. Forty-eight hours later our patient was successfully decannulated (total VV-ECMO time 43 days). During his entire ECMO run, our patient was in single-organ system respiratory failure without exhibiting any serious complications. Over the next 3 weeks, ventilation was slowly weaned as his native lung function continued to improve (total mechanical ventilation time 88 days). He was subsequently discharged from PICU and home 4 weeks later without additional oxygen. Because we hesitated to expose him to the risk of a both general anesthesia and extended surgery, we decided not to proceed to a surgical resection of the necrotic area, although the large abscess was still visible on repeat chest CT. Our patient is now 1 year post discharge and has normal oxygen saturation in room air and has fully recovered. His latest CXR showed only small residual parenchymal changes and a thickening of the pleura of the left lung (Fig. 3).