BACKGROUND:
Critically ill patients who receive circulatory or respiratory assist using extracorporeal membrane oxygenation (ECMO) may develop hemolysis, which can complicate the delivery of supportive care and be a potential risk factor for increased morbidity and mortality. Clinically, hemolysis is often identified using laboratory markers such as cell-free hemoglobin (CFH) and haptoglobin (Hp). However, such measurements require photometry or enzyme-linked immunosorbent assay (ELISA) and are labor intensive. In contrast, metabolic downstream products of CFH, such as carboxyhemoglobin (CO-Hb) and methemoglobin (Met-Hb), can be regularly monitored via arterial blood gas analyses in the intensive care unit (ICU). We hypothesized that CO-Hb and Met-Hb values measured during ECMO would correlate with the presence of hemolytic events as measured by CFH values exceeding 50mg/dl. We further hypothesized that CO-Hb and Met-Hb levels would correlate with peri-ECMO mortality.
METHODS:
Retrospective analysis of 435 patients with acute respiratory distress syndrome (ARDS) and veno-venous ECMO admitted to a tertiary ARDS referral center. Plasma concentrations of CO-Hb and Met-Hb were correlated with hemolytic events. Cutoff values of mean CO-Hb (mCO-Hb) and mean Met-Hb (mMet-Hb) associated with increased ICU mortality were calculated with recursive binary partitioning. Single and multivariable regression models for HE and ICU mortality were trained and compared.
RESULTS:
Simple and multivariable models including potential confounders identified associations between Met-Hb and hemolytic events (adj. odds ratio [OR] 2.99 [95% confidence interval {CI}, 2.19–4.10], P < .001). A cutoff value with 90% specificity of a hemolytic event was estimated for Met-Hb (1.55%). Both, mean CO-Hb (OR 2.03 [95% CI, 1.60–2.61], P < .001) and Met-Hb (2.78 [1.59–5.09], P < .001) were associated with ICU mortality. Cutoff values for mortality were 2% for mean CO-Hb and 1.25% for mean Met-Hb. The multivariable regression model for mortality including the continuous markers mCO-Hb and mMet-Hb produced an area under the curve (AUC) of 0.803.
CONCLUSIONS:
In patients with ARDS and ECMO, Met-Hb plasma concentrations were independently associated with hemolytic events. Both, mean CO-Hb and Met-Hb levels were associated with ICU mortality. These markers and their associated cutoff values might serve as a risk indicator in clinical practice.