A pilot study of alternative substrates in the critically Ill subject using a ketogenic feed

Abstract

Bioenergetic failure caused by impaired utilisation of glucose and fatty acids contributes to organ dysfunction across multiple tissues in critical illness. Ketone bodies may form an alternative substrate source, but the feasibility and safety of inducing a ketogenic state in physiologically unstable patients is not known. Twenty-nine mechanically ventilated adults with multi-organ failure managed on intensive care units were randomised (Ketogenic n = 14, Control n = 15) into a two-centre pilot open-label trial of ketogenic versus standard enteral feeding. The primary endpoints were assessment of feasibility and safety, recruitment and retention rates and achievement of ketosis and glucose control. Ketogenic feeding was feasible, safe, well tolerated and resulted in ketosis in all patients in the intervention group, with a refusal rate of 4.1% and 82.8% retention. Patients who received ketogenic feeding had fewer hypoglycaemic events (0.0% vs. 1.6%), required less exogenous international units of insulin (0 (Interquartile range 0-16) vs.78 (Interquartile range 0-412) but had slightly more daily episodes of diarrhoea (53.5% vs. 42.9%) over the trial period. Ketogenic feeding was feasible and may be an intervention for addressing bioenergetic failure in critically ill patients. Clinical Trials.gov registration: NCT04101071.

Fig. 1. Ketone body formation.

Plasma Beta-hydroxybutyrate (A) and Acetoacetate (B) concentrations during the 10-day intervention. Data are mean (95%CI). Red lines represent ketogenic feeding, and blue lines controls. *p < 0.05; **p < 0.01 for two-tailed Mann-Whitney-U test. n = 14 subjects in the ketogenic arm and n = 15 subjects in the control arm.

Fig. 2. Targetted metabolic parameters.

Co-efficient of Variation of serum glucose (A); Nutritional adequacy (B); Plasma pyruvate (C) and Lactate (D) concentrations during the intervention. Data are mean (95% CI). Red represents ketogenic feeding, blue controls. *p < 0.05; **p < 0.01; ***p < 0.001 between arms (two-tailed Mann–Whitney U test). N = 14 subjects in the ketogenic arm and n = 15 subjects in the control arm, (2 A 102 glucose readings in ketogenic arm vs 128 glucose readings in the control arm, p < 0.001).

Fig. 3. Sparse Partial Least Squares Discriminant analysis of patient randomised to ketogenic feeding on Day 10 (red triangle) and control feeding on day 10 (blue sphere).

Clockwise from top right: A polar positive, B non-polar positive, C non-polar negative, D polar negative. Error rates are <20% (12%, 15%, 14% and 17% respectively), suggesting plot is a result of real variation. n = 14 subjects in the ketogenic arm and n = 15 subjects in the control arm.

Fig. 4. A-H clockwise: Log abundance in Arbitrary Units (AU) of metabolites driving differential pathway analysis.

Data are mean (95%CI, Minimum-Maximum) and controls vs ketogenic: A Beta alanine [16 (16–17, 16–17) vs 17 (17–18, 17–18); B Ureidopropionic acid [15(15–16,14–17) vs16 (16–16, 15–17)]; C Lithocholate 3-O-glucuronide [15(15–15, 15–16) vs 15(15–15, 15–15)]; D PC (15:0/18:1(11Z))[19]18–19–18–20] vs. 13(13–13,12–14)]; E LysoPE (0:0/20:4(5Z,8Z,11Z,14Z))[11(11–12, 10–12)vs 14(13–14, 13–14)]; F PC (14:0/15:0) [14(14–15, 14–15)vs 15(14-15, 13–16)]; G PE-NMe (14:0/14:1(9Z)) [16 (15–16, 15–16)vs. 17(16–18,15–19)]; H PE (18:4(6Z,9Z,12Z,15Z)/p-18:1(11Z)) [15 (14–15, 14–16) vs 13(13–13, 12–14)];.