Abstract
Introduction: In adults and children, withholding parenteral nutrition (PN) for 1 week in ICU (late PN), hereby accepting macronutrient deficit early during critical illness, as compared with supplementing insufficient enteral nutrition with PN (early PN), accelerates weaning from mechanical ventilation, reduces infections, and shortens ICU stay1,2. We hypothesized that these benefits are in part mediated by fasting-induced ketogenesis.
Methods: This is a secondary analysis of the Early versus Late Parenteral Nutrition in the Pediatric ICU (PEPaNIC) RCT (N=1440)2. First, for a matched subset of 96 patients with a PICU stay of ≥5 days, daily plasma 3-hydroxybutyrate (3HB) concentrations were determined to identify the time point of maximal effect of late PN versus early PN, if any, on 3HB. Thereafter, for all patients with a plasma sample available on that “maximal effect day” (or last day for shorter stayers), plasma 3HB and insulin concentrations were quantified (N=1142). The independent association between plasma 3HB on that day and outcome was assessed by multivariable Cox proportional hazard analysis for time to live weaning from mechanical ventilation and for time to live PICU discharge and by multivariable logistic regression for incidence of new infection and PICU mortality, adjusted for randomization to late PN versus early PN and baseline risk factors (demographics, diagnosis, illness severity). In a sensitivity analysis, models were further adjusted for key regulators of ketogenesis (plasma insulin, blood glucose, corticosteroids and catecholamines) to assess whether any effect was direct or indirect.
Results: In the matched cohort, late PN increased plasma 3HB as compared with early PN (P<0.0001 for PICU-days 1 to 5), with maximal effect observed on PICU day 2. In the 1142 patients, plasma 3HB concentration on that “maximal effect day” was (mean±SEM) 0.19±0.05 mM in early PN patients and 1.17±0.02 mM in late-PN patients (P<0.0001). Adding these plasma 3HB concentrations to the multivariable models, adjusted for baseline risk factors and randomization, showed that higher plasma 3HB concentrations were independently associated with a higher likelihood of early live weaning from mechanical ventilatory support (P=0.0002) and of early live PICU discharge (P=0.004). As the 3HB concentrations replaced the effect of the randomization, this suggested that the 3HB effect statistically explained these effects of the randomization. Further adjustment for key regulators of ketogenesis did not alter these findings. The effect of late PN versus early PN on plasma 3HB did not explain its impact on infections and was not related to mortality.
Conclusion: Withholding early PN increased plasma 3HB concentrations in critically ill children, a direct effect that mediated an important part of its beneficial impact on recovery.
1Casaer M. et al, N Engl J Med 20112Fivez T. et al, N Engl J Med 2016
Perspective of mechanical ventilation of lungs