Background: Many studies show that the intake of raspberries is beneficial to immune-metabolic health, but the responses of individuals are heterogeneous and not fully understood.
Methods: In a two-arm parallel-group, randomized, controlled trial, immune-metabolic outcomes and plasma metabolite levels were analyzed before and after an 8-week red raspberry consumption. Based on partial least squares discriminant analysis (PLS-DA) on plasma xenobiotic levels, adherence to the intervention was first evaluated. A second PLS-DA followed by hierarchical clustering was used to classify individuals into response subgroups. Clinical immune and metabolic outcomes, including insulin resistance (HOMA-IR) and sensitivity (Matsuda, QUICKI) indices, during the intervention were assessed and compared between response subgroups.
Results: Two subgroups of participants, type 1 responders (n = 17) and type 2 responders (n = 5), were identified based on plasma metabolite levels measured during the intervention. Type 1 responders showed neutral to negative effects on immune-metabolic clinical parameters after raspberry consumption, and type 2 responders showed positive effects on the same parameters. Changes in waist circumference, waist-to-hip ratio, fasting plasma apolipoprotein B, C-reactive protein and insulin levels as well as Matsuda, HOMA-IR and QUICKI were significantly different between the two response subgroups. A deleterious effect of two carotenoid metabolites was also observed in type 1 responders but these variables were significantly associated with beneficial changes in the QUICKI index and in fasting insulin levels in type 2 responders. Increased 3-ureidopropionate levels were associated with a decrease in the Matsuda index in type 2 responders, suggesting that this metabolite is associated with a decrease in insulin sensitivity for those subjects, whereas the opposite was observed for type 1 responders.
Conclusion: The beneficial effects associated with red raspberry consumption are subject to inter-individual variability. Metabolomics-based clustering appears to be an effective way to assess adherence to a nutritional intervention and to classify individuals according to their immune-metabolic responsiveness to the intervention. This approach may be replicated in future studies to provide a better understanding of how interindividual variability impacts the effects of nutritional interventions on immune-metabolic health.
Keywords: clustering; machine learning; metabolic health; metabolomics; precision nutrition; raspberry.
Copyright © 2023 Barbe, de Toro-Martín, San-Cristobal, Garneau, Pilon, Couture, Roy, Couillard, Marette and Vohl.