Integrating metabolomic signatures and psychosocial parameters in responsivity to an immersion treatment model for adolescent obesity

With childhood obesity tripling in the past two decades alone, developing effective treatment to promote long-term weight loss and maintenance is critical. Unfortunately, while many children respond favorably to behavioral treatments, failure to lose and maintain weight loss is common. There is a critical need to determine non-invasive markers that can serve to predict an individual’s likelihood of having a positive response to treatment, and to monitor the treatment progress for early identification of those who may be at risk for relapse (e.g., weight gain) in order to provide more early, targeted and ongoing intervention of these high risk individuals. An exploratory investigation was conducted to determine: (a) the relevance of using metabolomics in delivering non-invasive markers to predict an individual’s responsiveness to an immersion treatment program, and (b) to integrate psychosocial and metabolomic profiles that can further serve to predict an adolescent’s positive response to weight loss. Obese adolescents (12–18 years old, BMI >95th percentile) attending a 3 weeks immersion healthy lifestyle camp, Take Off 4-Health (TO4-H), were recruited to provide first morning void urine samples and to complete psychosocial and health-behavior inventories at baseline and at the end of the 3 weeks program. Subjects were categorized as responders (decline of ?0.5 BMI units) or non-responders (decline of <0.5 BMI units) based on weight loss between baseline and the end of the 3 weeks program. Subjects were additionally classified as having either a healthy or impaired self-esteem (impaired ?20 or healthy <20) and depression (impaired ?15 or healthy <15) at baseline and the end of the 3 weeks camp-based on their responses to psychosocial inventories. NMR based metabolomics was used to generate signatures of low molecular components in urine. Using multivariate analysis of signals, findings demonstrated a unique pattern for treatment responders vs. non-responders. Moreover, inspection of loadings plots and variable importance plots enabled the identification of a subset of metabolites and psychosocial variables that may assist in our understanding and prediction of adolescent responsivity to treatment for weight loss. For the psychosocial variables, improvement in self-esteem and depression did not correlate with weight loss, multivariate analysis of urinary metabolomics data using healthy and impaired classifications for self-esteem and depression did enable the determination of subsets of metabolites that best associated with impaired self-esteem and depression. Integrating the metabolomic and psychosocial data provided a marker profile pointing to both biochemical (metabolite) and psychosocial indexes (self-esteem and depression) that are most relevant to determining an individual’s overall response to treatment. This study demonstrates the utility of metabolomics in providing non-invasive markers that upon validation in a larger sample may be an extremely important tool in the early detection and clinical management of adolescent obesity.