Background: Moderate-to-vigorous intensity exercise acutely suppresses appetite and the orexigenic hormone acylated ghrelin, increases satiety hormones (e.g., peptide YY), and reduces postprandial triacylglycerol and insulin concentrations. Meal consumption also induces appetite suppression alongside corresponding fluctuations in appetite-related hormones. Most investigations focus on mean group-level responses, yet individuals can vary considerably in their appetite and metabolic responses to exercise and eating. Robust quantification of individual differences requires methodological and statistical approaches that distinguish true inter-individual variability from natural within-subject variability. In a series of controlled laboratory experiments, we aimed to determine: (1) whether individual appetite and metabolic responses to exercise and eating are consistent; (2) whether true inter-individual variability exists; and (3) what factors might contribute to such variability.

Methods: Using replicate crossover designs, appetite outcomes (acylated ghrelin, total PYY, hunger, fullness) were assessed in an exercise (study 1: n=15 men) and a meal (study 2: n=18 men) study, while postprandial metabolic outcomes (triacylglycerol, glucose, insulin) were examined in an exercise study (study 3: n=20 men). Each study comprised two crossover cycles of paired intervention (study 1 and 3: exercise; study 2: meal) and control (study 1 and 3: rest; study 2: no meal) conditions. Between-cycle correlation coefficients quantified the consistency of individual differences between the replicates of control-adjusted intervention responses. Within-participant linear mixed-models and between-participant, random-effects meta-analyses of the replicate-averaged condition effect estimated treatment response heterogeneity. In study 2, the moderating influence of the fat mass and obesity-associated (FTO) gene was also examined.

Results: In study 1 and 2, exercise and eating suppressed mean hunger and acylated ghrelin concentrations and increased mean fullness and peptide YY concentrations versus control (main effect condition P ≤ 0.001). Moderate-to-large positive correlations were observed between the two replicates of control-adjusted exercise responses (r range = 0.55 to 0.82, P ≤ 0.035) and meal responses (r range = 0.41 to 0.86, P ≤ 0.091). Participant-by-condition interactions (study 1: P ≤ 0.077; study 2: P ≤ 0.056), and treatment effect heterogeneity estimates from the meta-analyses (e.g., tau statistic [95% CI] for acylated ghrelin: study 1 28.1 [17.1, 48.8] pg/mL; study 2 16.6 [0, 34.1] pg/mL) indicated meaningful inter-individual variability in appetite responses. In study 2, FTO genotype-by-condition interactions showed no evidence of moderation on the magnitude of post-meal responses (all P ≥ 0.192).

In study 3, exercise reduced mean postprandial triacylglycerol and insulin concentrations versus control (main effect condition P ≤ 0.022), but between-condition differences were trivial for glucose (P = 0.126). Between-cycle correlations were small-to-moderate and not statistically significant (r range = -0.42 to 0.11, P ≥ 0.066), participant-by-condition interactions were trivial (all P ≥ 0.137), and treatment effect heterogeneity estimates were negligible (e.g., tau statistic [95% CI] for triacylglycerol: 0 [0, 0] mmol/L h).  

Conclusion: Meaningful inter-individual variability was detected in appetite responses to exercise and eating, but not in postprandial metabolic responses to exercise. Further research is required to identify moderators responsible for the individual variability in appetite responses to exercise and eating, and to determine the clinical implications for weight control.