Towards a fasting-mimicking diet for critically ill patients: the pilot randomized crossover ICU-FM-1 study

Abstract

Background: In two recent randomized controlled trials, withholding parenteral nutrition early in critical illness improved outcome as compared to early up-to-calculated-target nutrition, which may be explained by beneficial effects of fasting. Outside critical care, fasting-mimicking diets were found to maintain fasting-induced benefits while avoiding prolonged starvation. It is unclear whether critically ill patients can develop a fasting response after a short-term nutrient interruption. In this randomized crossover pilot study, we investigated whether 12-h nutrient interruption initiates a metabolic fasting response in prolonged critically ill patients. As a secondary objective, we studied the feasibility of monitoring autophagy in blood samples.

Methods: In a single-center study in 70 prolonged critically ill patients, 12-h up-to-calculated-target feeding was alternated with 12-h fasting on day 8 ± 1 in ICU, in random order. Blood samples were obtained at the start of the study, at the crossover point, and at the end of the 24-h study period. Primary endpoints were a fasting-induced increase in serum bilirubin and decrease in insulin requirements to maintain normoglycemia. Secondary outcomes included serum insulin-like growth factor I (IGF-I), serum urea, plasma beta-hydroxybutyrate (BOH), and mRNA and protein markers of autophagy in whole blood and isolated white blood cells. To obtain a healthy reference, mRNA and protein markers of autophagy were assessed in whole blood and isolated white blood cells of 23 matched healthy subjects in fed and fasted conditions. Data were analyzed using repeated-measures ANOVA, Fisher's exact test, or Mann-Whitney U test, as appropriate.

Results: A 12-h nutrient interruption significantly increased serum bilirubin and BOH and decreased insulin requirements and serum IGF-I (all p ≤ 0.001). Urea was not affected. BOH was already increased from 4 h fasting onwards. Autophagic markers in blood samples were largely unaffected by fasting in patients and healthy subjects.

Conclusions: A 12-h nutrient interruption initiated a metabolic fasting response in prolonged critically ill patients, which opens perspectives for the development of a fasting-mimicking diet. Blood samples may not be a good readout of autophagy at the tissue level.

Trial registration: ISRCTN, ISRCTN98404761. Registered 3 May 2017.

Keywords: Autophagy; Critical illness; Intensive care; Intermittent fasting; Metabolism; Nutrition.

Fig. 1

Enrollment and randomization. Reasons for ineligibility and non-inclusion of eligible patients are listed. Ultimately, 35 patients per randomization group completed the study protocol, hence 70 patients in total. *Vital organ support was defined as dependency on mechanical ventilation and mechanical and/or pharmacological hemodynamic support. **Medical reasons include high glucose need to treat hypernatremia (n = 2) or multiple hypoglycemic episodes between randomization and start of the protocol (n = 1). ICU, intensive care unit; RCT, randomized controlled trial; DNR, do not resuscitate

Fig. 2

Calories administered. Daily total calorie administration and calories administered via the parenteral and enteral route separately are shown for both randomization groups starting 7 days before the intervention day. On the intervention day, caloric intake is shown for the two intervention intervals separately. In the top panels, administered calories are expressed as kilocalories per kilogram of body weight per hour. In the bottom panels, administered calories are expressed relative to the caloric target. Kilocalories from parenteral nutrition include kilocalories from total parenteral nutrition (glucose, proteins, and lipids), infused glucose-containing fluids, and propofol. Boxes represent interquartile ranges, and horizontal lines within the boxes represent the medians. Lines connect the medians from subsequent days or study intervals. Int., interval

Fig. 3

Metabolic fasting parameters. Results of serum total bilirubin, insulin requirements to maintain normoglycemia, serum urea, serum IGF-I, and plasma BOH are shown on the study day for both randomization groups. For bilirubin, urea, IGF-I, and BOH measurements, blood samples were obtained at the start of the protocol, at the end of the first study interval, and at the end of the second study interval. For insulin requirements, insulin administration was averaged per hour for the 24 h before the study day to obtain a baseline value and averaged per hour for both study intervals. Reported p values were obtained with repeated measures ANOVA. Lines represent means, and bars represent standard errors of the means. IGF-I, insulin-like growth factor I; BOH, beta-hydroxybutyrate

Fig. 4

POC ketone meters. Results of BOH measurements with use of a point of care ketone meter during the study day. BOH was measured every 4 h. Results are shown for both randomization groups. 0 h marks the start of the study day, 12 h marks the end of the first study interval, and 24 h marks the end of the second study interval. The reported p value was obtained using repeated measures ANOVA. Lines represent means, and bars represent standard errors of the means. BOH, beta-hydroxybutyrate

Fig. 5

Markers of autophagy. mRNA and protein expression of autophagy-related genes in whole blood and isolated white blood cells are shown for healthy controls and patients. Results are expressed relatively to the median of 23 fed healthy controls. Patient results are shown as lines representing the means and bars representing the standard errors of the means. For controls, the grey areas represent the standard errors of the means in fed (solid grey) and fasted (patterned grey) conditions. Interaction p values are higher than 0.05 for all studied genes (see the “Results” section for exact p values)