Not so fast: Paradoxically increased variability in the glucose tolerance test due to food withdrawal in continuous glucose-monitored mice

OBJECTIVE: This study was performed to determine the effect of fasting on reproducibility of the glucose tolerance test. Due to individual variation in animal feeding behaviors, fasting animals prior to metabolic and behavioral experiments is widely held to reduce inter-subject variation in glucose and metabolic parameters of preclinical rodent models. Reducing variability is especially important for studies where initial metabolite levels can influence the magnitude of experimental interventions, but fasting also imposes stress that may distort the variables of interest. One such intervention is the glucose tolerance test (GTT) which measures the maximum response and recovery following a bolus of exogenous glucose. We sought to investigate how fasting affects the response of individual mice to a GTT. METHODS: Using simultaneous continuous glucose monitoring (CGM) and indirect calorimetry, we quantified blood glucose, physical activity, body temperature, metabolic rates, and food consumption levels on a minute-to-minute basis in adult male mice for 4 weeks. We tested the effects of a 4-h or 18-h fast on the GTT to examine the effect of food withdrawal in light or dark photoperiods. Studies were also performed with 4-h fasting in additional mice without implanted CGM probes. RESULTS: Contrary to our expectations, a 4-h fast during the light photoperiod promotes a paradoxical increase in inter-animal variation in metabolic rate, physical activity, body temperature, glycemia, and glucose tolerance. This hyperglycemic and hyper-metabolic phenotype promotes increased corticosterone levels and is consistent with a behavioral stress response to food deprivation, even in well-fed mice. We find that mice undergoing an 18-h fast entered torpor, a hibernation-like state. In addition to low body temperature and metabolic rate, torpor is also associated with glucose levels 56 mg/dl lower than those seen in mice with ad libitum access to food. Moreover, the time spent in torpor affects the response to a GTT. CONCLUSION: Our results suggest fasting mice before glucose tolerance testing, and perhaps other experiments, can have the opposite of the intended effect where fasting can increase, rather than decrease, experimental variability.