Abnormal carnitine concentrations in critical illness associated with compromised outcome

BACKGROUND: Carnitine deficiency affects mitochondrial and muscle function, but its relevance during critical illness remains unknown. Our aim was to investigate the relationship between plasma free carnitine concentrations and outcome in prolonged critical illness. METHODS: In this secondary analysis of the EPaNIC randomised controlled trial, abnormal plasma free carnitine concentrations, measured on ICU-day-6 (N = 1600), were defined by their association with a lower likelihood of an earlier alive ICU discharge (the primary endpoint) in a Cox proportional hazards model. Subsequently, they were binned based on their distribution and partial residuals in the Cox-model. Adjusted multivariable Cox-model and logistic regression analysed both association of abnormal carnitinemia with acute and long-term morbidity and mortality, and predictive risk factors. RESULTS: The median plasma free carnitine concentration on ICU-day-6 was 34.8 (IQR 24.4-49.8 µmol/L). Surprisingly, higher concentrations associated with a lower likelihood of an earlier alive ICU discharge (HR [95% CI] (per 10 µmol/L increase): 0.97 [0.95-0.99]). Yet, the partial residuals plot revealed this likelihood to be lower for patients with concentrations corresponding to the lowest (< 24 µmol/L; N = 374) and highest quartiles (> 50 µmol/L; N = 395) as compared to intermediate quartiles (24-50 µmol/L; N = 831). Both low and high carnitine concentrations were associated with a prolonged ICU and hospital dependency, a prolonged need for life-supporting therapies, and increased mortality at 90-days. Carnitine concentrations above 50 µmol/L also associated with muscle weakness and increased two and five year-mortality. CONCLUSION: On ICU-day-6, both low and high free carnitine concentrations associated with delayed ICU-recovery, and excess morbidity and mortality, suggesting a U-shaped relationship.