Hyperlactatemia in Critically Ill Children: Modeling Early Recovery Kinetics After Initiation of Extracorporeal Membrane Oxygenation

Objectives: Blood lactate concentration ([Lac] b ) reflects the balance among production, clearance (C l [Lac] ), and volume of distribution. We have observed dramatic improvement in [Lac] b in critically ill patients after starting support with extracorporeal membrane oxygenation (ECMO) and discontinuing vasopressors. Here, we evaluated such [Lac] b profiles to develop a mathematical model of recovery kinetics. We then examined the interrelationships between maximum [Lac] b and model-derived parameters of lactate production, endogenous lactate transfer, and C l [Lac] . Design: Mathematical modeling using a convenience sample. Setting: Quaternary U.S. academic children’s hospital. Participants: A retrospective sample of 25 ECMO patients (from birth to < 18 yr) with serial [Lac] b measurements during the first 30 hours after initiation of ECMO. Interventions: None. Measurement and Main Results: The median (interquartile range [IQR]) age of ptients was 17 days (IQR 3–152 d), and the median weight was 3.3 kg (IQR 2.7–4.7 kg). At the initiation of ECMO, the mean peak [Lac] b was 16.7 mmol/L (95% CI, 14.3–20.0 mmol/L). Recovery in [Lac] b could be described using a one-compartment, bi-exponential, open model of kinetics. Solving the model equation showed starting lactate load was 17.7 mmol/kg (95% CI, 14.6–20.7 mmol/kg) and C l [Lac] was 19.7 mL/min (95% CI, 3.0–36.4 mL/min). The interrelationship between maximum [Lac] b and model-derived parameters in children requiring ECMO at the limits of cardiopulmonary survival showed: 1) lactate production ranged from 2.3 to 6.4 µmol/kg/min (95% CI), 2) initial endogenous lactate transfer velocity, 82.5–1301.0 µmol/kg/min, 3) high initial [Lac] b levels suggested severely impaired C l [Lac] , 4) a strong correlation was observed between model-derived velocity and transfer parameters (rho 0.75; p < 0.0001), at levels exceeding those seen in high-intensity endurance exercise, and 5) upon achieving steady state, lactate production and C l [Lac] were balanced. Conclusions: At the time of maximal cardiopulmonary instability requiring ECMO initiation, our model of [Lac] b recovery indicated that high initial [Lac] b reflected severely impaired and reduced C l [Lac] . This modeling approach may also be applicable to assessing changes in lactate kinetics in other forms of critical illness.