RESUMEN
BACKGROUND: Long-term treatment-related toxicities, such as neurological and metabolic toxicities, are major issues in breast cancer. We investigated the interest of metabolomic profiling to predict toxicities. METHODS: Untargeted high-resolution metabolomic profiles of 992 patients with ER+/HER2- breast cancer from the prospective CANTO cohort were acquired (n=1935 metabolites). A residual-based modeling strategy with a discovery and validation cohort was used to benchmark machine learning algorithms, taking into account confounding variables. RESULTS: Adaptive LASSO has a good predictive performance, has limited optimism bias, and allows the selection of metabolites of interest for future translational research. The addition of low-frequency metabolites and non-annotated metabolites increases the predictive power. Metabolomics adds extra performance to clinical variables to predict various neurological and metabolic toxicity profiles. CONCLUSIONS: Untargeted high-resolution metabolomics allows better toxicity prediction by considering environmental exposure, metabolites linked to microbiota, and low-frequency metabolites.
RESUMEN
Aging is a dynamic process depending on intrinsic and extrinsic factors and its evolution is a continuum of transitions, involving multifaceted processes at multiple levels. It is recognized that frailty and sarcopenia are shared by the major age-related diseases thus contributing to elderly morbidity and mortality. Pre-frailty is still not well understood but it has been associated with global imbalance in several physiological systems, including inflammation, and in nutrition. Due to the complex phenotypes and underlying pathophysiology, the need for robust and multidimensional biomarkers is essential to move toward more personalized care. The objective of the present study was to better characterize the complexity of pre-frailty phenotype using untargeted metabolomics, in order to identify specific biomarkers, and study their stability over time. The approach was based on the NU-AGE project (clinicaltrials.gov, NCT01754012) that regrouped 1,250 free-living elderly people (65-79 y.o., men and women), free of major diseases, recruited within five European centers. Half of the volunteers were randomly assigned to an intervention group (1-year Mediterranean type diet). Presence of frailty was assessed by the criteria proposed by Fried et al. (2001). In this study, a sub-cohort consisting in 212 subjects (pre-frail and non-frail) from the Italian and Polish centers were selected for untargeted serum metabolomics at T0 (baseline) and T1 (follow-up). Univariate statistical analyses were performed to identify discriminant metabolites regarding pre-frailty status. Predictive models were then built using linear logistic regression and ROC curve analyses were used to evaluate multivariate models. Metabolomics enabled to discriminate sub-phenotypes of pre-frailty both at the gender level and depending on the pre-frailty progression and reversibility. The best resulting models included four different metabolites for each gender. They showed very good prediction capacity with AUCs of 0.93 (95% CI = 0.87-1) and 0.94 (95% CI = 0.87-1) for men and women, respectively. Additionally, early and/or predictive markers of pre-frailty were identified for both genders and the gender specific models showed also good performance (three metabolites; AUC = 0.82; 95% CI = 0.72-0.93) for men and very good for women (three metabolites; AUC = 0.92; 95% CI = 0.86-0.99). These results open the door, through multivariate strategies, to a possibility of monitoring the disease progression over time at a very early stage.