Risk Factors for Intracranial Hemorrhage and Mortality in Adult Patients with Severe Respiratory Failure Managed Using Veno-Venous Extracorporeal Membrane Oxygenation
The use of veno-venous extracorporeal membrane oxygenation (VV-ECMO) has become an increasingly critical intervention for managing severe respiratory failure (SRF) caused by reversible conditions or as a bridge to lung transplantation. Despite technological advancements, mortality rates remain high, with neurological complications such as intracranial hemorrhage (ICH) representing a leading cause of death. This study explores the risk factors for ICH and mortality in adults undergoing VV-ECMO for SRF, offering insights into optimizing clinical management.
Methodology and Patient Characteristics
A retrospective analysis was conducted on 77 patients who received VV-ECMO at the China-Japan Friendship Hospital from July 2013 to May 2019. Patients were included if they met predefined criteria for SRF: a ratio of partial pressure of arterial oxygen (PaO2) to the fraction of inspired oxygen (FiO2) 3 hours, PaO2:FiO2 6 hours, or pH 6 hours despite optimized mechanical ventilation. Exclusion criteria encompassed patients on veno-arterial ECMO, perioperative cases, or those requiring low-flow extracorporeal CO2 removal.
Clinical data included demographics, pre-ECMO laboratory values, anticoagulation parameters (activated partial thromboplastin time [APTT], fibrinogen, platelet counts), and complications. Neurological monitoring involved daily awakenings, pupil assessments, and prompt computed tomography (CT) for suspected ICH. Multivariable logistic regression identified independent risk factors for ICH and mortality.
Key Findings on Intracranial Hemorrhage
Among 77 patients, 11 (14.3%) developed ICH, with a significantly lower survival rate in the ICH group (18.2% vs. 51.5%, P = 0.040). ICH manifested at a median of 4.3 days post-ECMO initiation, with clinical signs including mydriasis, confusion, coma, and hemiplegia. CT scans revealed multiple parenchymal hemorrhages (9/11 cases), often coexisting with subarachnoid or subdural bleeding. Only one patient with a single parenchymal hemorrhage survived.
Multivariable analysis identified diabetes mellitus (adjusted odds ratio [aOR]: 12.848; 95% confidence interval [CI]: 1.129–146.188; P = 0.040) and minimum fibrinogen levels during ECMO (aOR: 2.557; 95% CI: 1.244–5.252; P = 0.011) as independent risk factors for ICH. Pre-ECMO platelet counts were lower in the ICH group (median 102.0 vs. 155.5 ×10⁹/L; P = 0.040), suggesting baseline coagulation abnormalities may exacerbate hemorrhage risk. Elevated fibrinogen, paradoxically linked to ICH, reflects systemic inflammation and endothelial dysfunction, potentially destabilizing cerebral vasculature.
Mortality Risk Factors
Overall mortality was 46.8% (36 survivors). Independent predictors included acute hepatic failure during ECMO (aOR: 9.205; 95% CI: 1.375–61.604; P = 0.022), pre-ECMO CO2 retention (PaCO2 ≥50 mmHg; aOR: 7.602; 95% CI: 1.514–38.188; P = 0.014), and lowest platelet concentration during ECMO (aOR: 0.130; 95% CI: 0.029–0.577; P = 0.007). Non-survivors exhibited longer pre-ECMO mechanical ventilation (3.0 vs. 1.0 days; P = 0.018) and higher rates of acute kidney injury (78.0% vs. 55.6%; P = 0.035) and hepatic failure (39.0% vs. 8.3%; P = 0.002). These findings underscore multiorgan dysfunction as a critical contributor to poor outcomes.
Pathophysiological and Clinical Implications
Diabetes and ICH Risk: Chronic hyperglycemia in diabetic patients induces endothelial dysfunction, oxidative stress, and proinflammatory states, accelerating cerebrovascular aging. Combined with ECMO-induced coagulopathy, this predisposes to microvascular fragility and hemorrhage. The study advocates stringent anticoagulation targets and vigilant neurological monitoring in diabetic patients.
Fibrinogen Dynamics: Elevated fibrinogen, typically a marker of clotting capacity, may paradoxically increase ICH risk by reflecting systemic inflammation. Fibrinogen consumption during ECMO exacerbates coagulopathy, necessitating balanced transfusion strategies to maintain levels ≥2 g/L while avoiding oversupplementation.
CO2 Retention and Mortality: Hypercapnia before ECMO correlates with poor prognosis, potentially due to cerebral vasodilation increasing intracranial pressure or compensatory mechanisms worsening systemic acidosis. Rapid correction of PaCO2 post-ECMO (median decrease: −14.2 mmHg) did not directly link to ICH but highlighted the need for gradual pH normalization to avoid reperfusion injury.
Thrombocytopenia and Hepatic Failure: Platelet counts <60 ×10⁹/L during ECMO were associated with mortality, emphasizing the role of platelet transfusion thresholds. Acute hepatic failure, linked to synthetic dysfunction and coagulopathy, underscores the need for liver-protective strategies during ECMO.
Limitations and Future Directions
The retrospective design and small sample size limit generalizability. Anticoagulation protocols, platelet function assays, and transfusion impacts were not fully analyzed. Prospective studies with larger cohorts are required to validate these findings and refine risk stratification models.
Clinical Recommendations
- Anticoagulation Adjustment: Tailor heparin regimens for diabetics, considering lower APTT targets (e.g., 50–60 seconds) to mitigate ICH risk.
- Fibrinogen Monitoring: Maintain levels ≥2 g/L but avoid excessive replacement, which may exacerbate hypercoagulable states.
- Neurological Surveillance: Implement hourly pupil checks and daily consciousness assessments, with low thresholds for CT imaging.
- Organ Support: Prioritize early renal and hepatic support to prevent multiorgan failure, a key mortality driver.
Conclusion
ICH remains a devastating complication of VV-ECMO, with diabetes and fibrinogen fluctuations as modifiable risk factors. Mortality is driven by multiorgan dysfunction, emphasizing holistic critical care. These insights guide targeted interventions to improve outcomes in SRF patients requiring ECMO.
doi.org/10.1097/CM9.0000000000001719
Was this helpful?
0 / 0