AlpsNMR: an R package for signal processing of fully untargeted NMR-based metabolomics.

SUMMARY: Nuclear magnetic resonance (NMR)-based metabolomics is widely used to obtain metabolic fingerprints of biological systems. While targeted workflows require previous knowledge of metabolites, prior to statistical analysis, untargeted approaches remain a challenge. Computational tools dealing with fully untargeted NMR-based metabolomics are still scarce or not user-friendly. Therefore, we developed AlpsNMR (Automated spectraL Processing System for NMR), an R package that provides automated and efficient signal processing for untargeted NMR metabolomics. AlpsNMR includes spectra loading, metadata handling, automated outlier detection, spectra alignment and peak-picking, integration and normalization. The resulting output can be used for further statistical analysis. AlpsNMR proved effective in detecting metabolite changes in a test case. The tool allows less experienced users to easily implement this workflow from spectra to a ready-to-use dataset in their routines. AVAILABILITY AND IMPLEMENTATION: The AlpsNMR R package and tutorial is freely available to download from http://github.com/sipss/AlpsNMR under the MIT license. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Biomarker-based validity of a food frequency questionnaire estimating intake in Brazilian children and adolescents.

This study evaluated the validity of nutrient and food group intakes estimated by an FFQ against biomarkers. A 71-item semiquantitative FFQ was administered to 210 Brazilian children and adolescents aged 9-13 years. Intakes were correlated with biomarkers in plasma and red blood cells. Correlations between nutrients and their biomarkers were presented for animal protein, myristic acid (C14:0), EPA, DHA, ß-carotene, folate, and vitamins B3, B5 and B6. Food groups and biomarkers were correlated as follows: fish products with EPA and DHA; milk and dairy with C14:0, pyridoxal 5'-phosphate and vitamin B12; total vegetables and dark green and orange vegetables with ß-carotene; 5-methyltetrahydrofolate with green vegetables; and flour products with para-aminobenzoylglutamic acid. This FFQ is a valid tool for ranking Brazilian children and adolescents according to their intake of several nutrients and food groups.

High-throughput and simultaneous quantitative analysis of homocysteine-methionine cycle metabolites and co-factors in blood plasma and cerebrospinal fluid by isotope dilution LC-MS/MS.

The methionine cycle is a key pathway contributing to the regulation of human health, with well-established involvement in cardiovascular diseases and cognitive function. Changes in one-carbon cycle metabolites have also been associated with mild cognitive decline, vascular dementia, and Alzheimer's disease. Today, there is no single analytical method to monitor both metabolites and co-factors of the methionine cycle. To address this limitation, we here report for the first time a new method for the simultaneous quantitation of 17 metabolites in the methionine cycle, which are homocysteic acid, taurine, serine, cysteine, glycine, homocysteine, riboflavin, methionine, pyridoxine, cystathionine, pyridoxamine, S-adenosylhomocysteine, S-adenosylmethionine, betaine, choline, dimethylglycine, and 5-methyltetrahydrofolic acid. This multianalyte method, developed using ultra-performance liquid chromatography-tandem mass spectrometry (UPLC-MS/MS), provides a highly accurate and precise quantitation of these 17 metabolites for both plasma and cerebrospinal fluid metabolite monitoring. The method requires a simple sample preparation, which, combined with a short chromatographic run time, ensures a high sample throughput. This analytical strategy will thus provide a novel metabolomics approach to be employed in large-scale observational and intervention studies. We expect such a robust method to be particularly relevant for broad and deep molecular phenotyping of individuals in relation to their nutritional requirements, health monitoring, and disease risk management.

Genome-wide association study of metabolic traits reveals novel gene-metabolite-disease links.

Metabolic traits are molecular phenotypes that can drive clinical phenotypes and may predict disease progression. Here, we report results from a metabolome- and genome-wide association study on (1)H-NMR urine metabolic profiles. The study was conducted within an untargeted approach, employing a novel method for compound identification. From our discovery cohort of 835 Caucasian individuals who participated in the CoLaus study, we identified 139 suggestively significant (P<5×10(-8)) and independent associations between single nucleotide polymorphisms (SNP) and metabolome features. Fifty-six of these associations replicated in the TasteSensomics cohort, comprising 601 individuals from São Paulo of vastly diverse ethnic background. They correspond to eleven gene-metabolite associations, six of which had been previously identified in the urine metabolome and three in the serum metabolome. Our key novel findings are the associations of two SNPs with NMR spectral signatures pointing to fucose (rs492602, P = 6.9×10(-44)) and lysine (rs8101881, P = 1.2×10(-33)), respectively. Fine-mapping of the first locus pinpointed the FUT2 gene, which encodes a fucosyltransferase enzyme and has previously been associated with Crohn's disease. This implicates fucose as a potential prognostic disease marker, for which there is already published evidence from a mouse model. The second SNP lies within the SLC7A9 gene, rare mutations of which have been linked to severe kidney damage. The replication of previous associations and our new discoveries demonstrate the potential of untargeted metabolomics GWAS to robustly identify molecular disease markers.

Modeling Longitudinal Metabonomics and Microbiota Interactions in C57BL/6 Mice Fed a High Fat Diet.

Longitudinal studies aim typically at following populations of subjects over time and are important to understand the global evolution of biological processes. When it comes to longitudinal omics data, it will often depend on the overall objective of the study, and constraints imposed by the data, to define the appropriate modeling tools. Here, we report the use of multilevel simultaneous component analysis (MSCA), orthogonal projection on latent structures (OPLS), and regularized canonical correlation analysis (rCCA) to study associations between specific longitudinal urine metabonomics data and microbiome data in a diet-induced obesity model using C57BL/6 mice. (1)H NMR urine metabolic profiling was performed on samples collected weekly over a period of 13 weeks, and stool microbial composition was assessed using 16S rRNA gene sequencing at three specific time periods (baseline, first week response, end of study). MSCA and OPLS allowed us to explore longitudinal urine metabonomics data in relation to the dietary groups, as well as dietary effects on body weight. In addition, we report a data integration strategy based on regularized CCA and correlation analyses of urine metabonomics data and 16S rRNA gene sequencing data to investigate the functional relationships between metabolites and gut microbial composition. Thanks to this workflow enabling the breakdown of this data set complexity, the most relevant patterns could be extracted to further explore physiological processes at an anthropometric, cellular, and molecular level.

Impact of breast-feeding and high- and low-protein formula on the metabolism and growth of infants from overweight and obese mothers.

BACKGROUND: The combination of maternal obesity in early pregnancy and high protein intake in infant formula feeding might predispose to obesity risk in later life. METHODS: This study assesses the impact of breast- or formula-feeding (differing in protein content by 1.65 or 2.7 g/100 kcal) on the metabolism of term infants from overweight and obese mothers. From birth to 3 mo of age, infants received exclusively either breast- or starter formula-feeding and until 6 mo, exclusively either a formula designed for this study or breast-feeding. From 6 to 12 mo, infants received complementary weaning food. Metabonomics was conducted on the infants' urine and stool samples collected at the age of 3, 6, and 12 mo. RESULTS: Infant formula-feeding resulted in higher protein-derived short-chain fatty acids and amino acids in stools. Urine metabonomics revealed a relationship between bacterial processing of dietary proteins and host protein metabolism stimulated with increasing protein content in the formula. Moreover, formula-fed infants were metabolically different from breast-fed infants, at the level of lipid and energy metabolism (carnitines, ketone bodies, and Krebs cycle). CONCLUSION: Noninvasive urine and stool metabolic monitoring of responses to early nutrition provides relevant readouts to assess nutritional requirements for infants' growth.

Early metabolic adaptation in C57BL/6 mice resistant to high fat diet induced weight gain involves an activation of mitochondrial oxidative pathways.

We investigated the short-term (7 days) and long-term (60 days) metabolic effect of high fat diet induced obesity (DIO) and weight gain in isogenic C57BL/6 mice and examined the specific metabolic differentiation between mice that were either strong-responders (SR), or non-responders (NR) to weight gain. Mice (n = 80) were fed a standard chow diet for 7 days prior to randomization into a high-fat (HF) (n = 56) or a low-fat (LF) (n = 24) diet group. The (1)H NMR urinary metabolic profiles of LF and HF mice were recorded 7 and 60 days after the diet switch. On the basis of the body weight gain (BWG) distribution of HF group, we identified NR mice (n = 10) and SR mice (n = 14) to DIO. Compared with LF, HF feeding increased urinary excretion of glycine conjugates of ß-oxidation intermediate (hexanoylglycine), branched chain amino acid (BCAA) catabolism intermediates (isovalerylglycine, α-keto-ß-methylvalerate and α-ketoisovalerate) and end-products of nicotinamide adenine dinucleotide (NAD) metabolism (N1-methyl-2-pyridone-5-carboxamide, N1-methyl-4-pyridone-3-carboxamide) suggesting up-regulation of mitochondrial oxidative pathways. In the HF group, NR mice excreted relatively more hexanoylglycine, isovalerylglycine, and fewer tricarboxylic acid (TCA) cycle intermediate (succinate) in comparison to SR mice. Thus, subtle regulation of ketogenic pathways in DIO may alleviate the saturation of the TCA cycle and mitochondrial oxidative metabolism.

A whole-grain-rich diet reduces urinary excretion of markers of protein catabolism and gut microbiota metabolism in healthy men after one week.

Epidemiological studies consistently find that diets rich in whole-grain (WG) cereals lead to decreased risk of disease compared with refined grain (RG)-based diets. Aside from a greater amount of fiber and micronutrients, possible mechanisms for why WGs may be beneficial for health remain speculative. In an exploratory, randomized, researcher-blinded, crossover trial, we measured metabolic profile differences between healthy participants eating a diet based on WGs compared with a diet based on RGs. Seventeen healthy adult participants (11 female, 6 male) consumed a controlled diet based on either WG-rich or RG-rich foods for 2 wk, followed by the other diet after a 5-wk washout period. Both diets were the same except for the use of WG (150 g/d) or RG foods. The metabolic profiles of plasma, urine, and fecal water were measured using (1)H-nuclear magnetic resonance spectroscopy and gas chromatography-mass spectrometry (plasma only). After 1 wk of intervention, the WG diet led to decreases in urinary excretion of metabolites related to protein catabolism (urea, methylguanadine), lipid (carnitine and acylcarnitines) and gut microbial (4-hydroxyphenylacetate, trimethylacetate, dimethylacetate) metabolism in men compared with the same time point during the RG intervention. There were no differences between the interventions after 2 wk. Urinary urea, carnitine, and acylcarnitine were lower at wk 1 of the WG intervention relative to the RG intervention in all participants. Fecal water short-chain fatty acids acetate and butyrate were relatively greater after the WG diet compared to the RG diet. Although based on a small population and for a short time period, these observations suggest that a WG diet may affect protein metabolism.

Specific dietary preferences are linked to differing gut microbial metabolic activity in response to dark chocolate intake.

Systems biology approaches are providing novel insights into the role of nutrition for the management of health and disease. In the present study, we investigated if dietary preference for dark chocolate in healthy subjects may lead to different metabolic response to daily chocolate consumption. Using NMR- and MS-based metabolic profiling of blood plasma and urine, we monitored the metabolic response of 10 participants stratified as chocolate desiring and eating regularly dark chocolate (CD) and 10 participants stratified as chocolate indifferent and eating rarely dark chocolate (CI) to a daily consumption of 50 g of dark chocolate as part of a standardized diet over a one week period. We demonstrated that preference for chocolate leads to different metabolic response to chocolate consumption. Daily intake of dark chocolate significantly increased HDL cholesterol by 6% and decreased polyunsaturated acyl ether phospholipids. Dark chocolate intake could also induce an improvement in the metabolism of long chain fatty acid, as noted by a compositional change in plasma fatty acyl carnitines. Moreover, a relationship between regular long-term dietary exposure to a small amount of dark chocolate, gut microbiota, and phenolics was highlighted, providing novel insights into biological processes associated with cocoa bioactives.

An Open-Access Data Platform: Global Nutrition and Health Atlas (GNHA).

The rapid development of nutrition science is embracing digital transformation to generate large amounts of data. Precision nutrition and "Big Data" place increasing demand for data repositories and visualization, which enhances the digital transformation. We defined the need for an integrated nutrition data platform as a web-based platform that can collect, store, track, analyze, monitor, and visually display key metrics in nutrition and health while allowing users to interact with visuals and download data provided in the platform. Interactive dashboards create new opportunities for scholars and practitioners to generate and test hypotheses. We present the development and implementation of the Global Nutrition and Health Atlas (GNHA; https://sites.tufts.edu/gnha/), an open-access online platform covering nutrition and health data with 26 themes and 500+ indicators from 190+ countries up to 30 y. We view GNHA as an interactive tool aiming to share information and perspectives and foster collaborations and innovations.

Food and Nutrition Systems Dashboards: A Systematic Review.

The rapid expansion of food and nutrition information requires new ways of data sharing and dissemination. Interactive platforms integrating data portals and visualization dashboards have been effectively utilized to describe, monitor, and track information related to food and nutrition; however, a comprehensive evaluation of emerging interactive systems is lacking. We conducted a systematic review on publicly available dashboards using a set of 48 evaluation metrics for data integrity, completeness, granularity, visualization quality, and interactivity based on 4 major principles: evidence, efficiency, emphasis, and ethics. We evaluated 13 dashboards, summarized their characteristics, strengths, and limitations, and provided guidelines for developing nutrition dashboards. We applied mixed effects models to summarize evaluation results adjusted for interrater variability. The proposed metrics and evaluation principles help to improve data standardization and harmonization, dashboard performance and usability, broaden information and knowledge sharing among researchers, practitioners, and decision makers in the field of food and nutrition, and accelerate data literacy and communication.

Metabotyping of Caenorhabditis elegans and their culture media revealed unique metabolic phenotypes associated to amino acid deficiency and insulin-like signaling.

Insulin/IGF-like signaling (IIS) and nutrient sensing are among the most potent regulators of health status and aging. Here, a global view of the metabolic changes in C. elegans with impaired function of IIS represented by daf-2 and daf-16 and the intestinal di- and tripeptide transport pept-1 was generated using (1)H nuclear magnetic resonance spectroscopic analysis of worm extracts and spent culture media. We showed that specific metabolic profiles were significantly associated with each type of mutant. On the basis of the metabonomics data, selected underlying processes were further investigated using proteomic and transcriptomic approaches. The observed changes suggest a decreased activity of the one carbon metabolism in pept-1(lg601) mutants. Higher concentration of branched-chain amino acids (BCAA) and altered transcript levels of genes involved in BCAA metabolism were observed in long-living strains daf-2(e1370) and daf-2(e1370);pept-1(lg601) when compared to wild types and daf-16(m26);daf-2(e1370);pept-1(lg601) C. elegans, suggesting a DAF-16-dependent mechanism.

Metabolic phenotyping of the Crohn's disease-like IBD etiopathology in the TNF(ΔARE/WT) mouse model.

The underlying biochemical consequences of inflammatory bowel disease (IBD) on the systemic and gastrointestinal metabolism have not yet been fully elucidated but could help to better understand the disease pathogenesis and to identify tissue-specific markers associated with the different disease stages. Here, we applied a metabonomic approach to monitor metabolic events associated with the gradual development of Crohn's disease (CD)-like ileitis in the TNF(ΔARE/WT) mouse model. Metabolic profiles of different intestinal compartments from the age of 4 up to 24 weeks were generated by combining proton nuclear magnetic resonance ((1)H NMR) spectroscopy and liquid chromatography-mass spectrometry (LC-MS). From 8 weeks onward, mice developed CD similar to the immune and tissue-related phenotype of human CD with ileal involvement, including ileal histological abnormalities, reduced fat mass and body weight, as well as hallmarks of malabsorption with higher energy wasting. The metabonomic approach highlighted shifts in the intestinal lipid metabolism concomitant to the histological onset of inflammation. Moreover, the advanced disease status was characterized by a significantly altered metabolism of cholesterol, triglycerides, phospholipids, plasmalogens, and sphingomyelins in the inflamed tissue (ileum) and the adjacent intestinal parts (proximal colon). These results describe different biological processes associated with the disease onset, including modifications of the general cell membrane composition, alteration of energy homeostasis, and finally the generation of inflammatory lipid mediators. Taken together, this provides novel insights into IBD-related alterations of specific lipid-dependant processes during inflammatory states.

Dietary modulation of gut functional ecology studied by fecal metabonomics.

A major source of intestinal metabolites results from both host and microbial processing of dietary nutrients. (1)H NMR-based metabolic profiling of mouse feces was carried out over time in different microbiome mouse models, including conventional (n = 9), conventionalized (n = 10), and "humanized" gnotobiotic mice inoculated with a model of human baby microbiota (HBM, n = 17). HBM mice were supplemented with Lactobacillus paracasei with (n = 10) and without (n = 7) prebiotics. Animals not supplemented with prebiotics received a diet enriched in glucose and lactose as placebo. In conventionalized animals, microbial populations and activities converged in term of multivariate mapping toward conventional mice. Both groups decreased bacterial processing of dietary proteins when switching to a diet enriched in glucose and lactose, as described with low levels of 5-aminovalerate, acetate, and propionate and high levels of lysine and arginine. The HBM model differs from conventional and conventionalized microbiota in terms of type, proportion, and metabolic activity of gut bacteria (lower short chain fatty acids (SCFAs), lactate, 5-aminovalerate, and oligosaccharides, higher bile acids and choline). The probiotics supplementation of HBM mice was associated with a specific amino acid pattern that can be linked to L. paracasei proteolytic activities. The combination of L. paracasei with the galactosyl-oligosaccharide prebiotics was related to the enhanced growth of bifidobacteria and lactobacilli, and a specific metabolism of carbohydrates, proteins, and SCFAs. The present study describes how the assessment of metabolic changes in feces may provide information for studying nutrient-microbiota relationships in different microbiome mouse models.

Chemometric strategy for modeling metabolic biological space along the gastrointestinal tract and assessing microbial influences.

Over the past decade, the analysis of metabolic data with advanced chemometric techniques has offered the potential to explore functional relationships among biological compartments in relation to the structure and function of the intestine. However, the employed methodologies, generally based on regression modeling techniques, have given emphasis to region-specific metabolic patterns, while providing only limited insights into the spatiotemporal metabolic features of the complex gastrointestinal system. Hence, novel approaches are needed to analyze metabolic data to reconstruct the metabolic biological space associated with the evolving structures and functions of an organ such as the gastrointestinal tract. Here, we report the application of multivariate curve resolution (MCR) methodology to model metabolic relationships along the gastrointestinal compartments in relation to its structure and function using data from our previous metabonomic analysis. The method simultaneously summarizes metabolite occurrence and contribution to continuous metabolic signatures of the different biological compartments of the gut tract. This methodology sheds new light onto the complex web of metabolic interactions with gut symbionts that modulate host cell metabolism in surrounding gut tissues. In the future, such an approach will be key to provide new insights into the dynamic onset of metabolic deregulations involved in region-specific gastrointestinal disorders, such as Crohn's disease or ulcerative colitis.

Alignment using variable penalty dynamic time warping.

In this article we highlight a novel variation on dynamic time warping (DTW) for aligning chromatogram signals. We are interested in sets of signals that can be aligned well locally, but not globally, by shifting individual signals in time. This kind of alignment is often sufficient for aligning gas chromatography data. Regular DTW often "over-warps" signals and introduces artificial features into the aligned data. To overcome this we introduce a variable penalty into the DTW process. The penalty is added to the distance metric whenever a nondiagonal step is taken. We select our penalty based on a morphological dilation of the two signals. We showcase our method by aligning GC/MS datafiles from 712 blood plasma samples processed in 23 batches over the course of 6 months. The use of variable penalty DTW significantly reduces the number of nondiagonal moves. In the examples presented here, this reduction is by a factor of 30, with no cost to visual quality of the alignment.

Diet Diversity and Micronutrient Adequacy among Filipino School-Age Children.

Previous studies have shown that the dietary diversity of young Filipino children to be limited and that the prevalence of nutrient inadequacies is high. This study extends the current knowledge to examine the relationship between diet diversity and the probability of adequacy of micronutrients among Filipino schoolchildren (aged 6 to 12 years), by the wealth status and dwelling location. The dietary intake data were collected using a single 24-h recall from 6460 children in the Filipino National Nutrition Survey 2013. The diet diversity score (DDS) and the probability of adequacies (PA) of 11 micronutrients were calculated, and further stratified by socio-economic status (SES) and dwelling location. The diet diversity was generally low (mean DDS = 4 out of 9). Children from the lowest SES, and living in rural areas, tended to have a lower DDS. Children with a DDS of 1 were likely to be inadequate in all 11 micronutrients. The higher DDS (≥6) was associated with higher PAs for the B vitamins but not for calcium, folate, iron, vitamin A and to large extent, vitamin C. This suggests that it was difficult for this population to achieve adequacy in these 5 micronutrients. More rigorous research on the topic is needed. Better access to nutrient-rich or fortified staple foods, in tandem with increased education on the importance of dietary diversity, are potential strategies to support children in achieving adequate micronutrient intakes.

In Vitro Gut Metabolism of [U-13 C]-Quinic Acid, The Other Hydrolysis Product of Chlorogenic Acid.

SCOPE: Quinic acid in its free form is broadly abundant in plants, and can accumulate in copious amounts in coffee, tea, and certain fruits. However, it has been mostly studied as chlorogenic acid, an ester of caffeic and quinic acids. When chlorogenic acid reaches the colon, it is hydrolyzed by microbial esterases releasing caffeic and quinic acids. While biotransformation of chlorogenic and caffeic acids have been elucidated by in vitro and in vivo studies, the gut metabolism of quinic acid has been so far overlooked. METHODS AND RESULTS: [U-13 C]-Quinic acid is submitted to a colonic model using human fecal microbiota for assessing its metabolic fate. The metabolite profiles formed along microbial biotransformation are monitored by a combined metabolomics approach, using both 2D GC- and ultra-HPLC-MS. Six metabolic intermediates are identified by incorporation of isotopic label. CONCLUSION: Two parallel degradation pathways could be proposed: (1) an oxidative route, leading to aromatization and accumulation of protocatechuic acid, and a (2) reductive route, including dehydroxylation to cyclohexane carboxylic acid. Elucidating the biotransformation of food bioactives by the gut microbiota is of relevance for understanding nutrition, interindividual variability and potential effects on human metabolism.

Consensus Clustering of temporal profiles for the identification of metabolic markers of pre-diabetes in childhood (EarlyBird 73).

In longitudinal clinical studies, methodologies available for the analysis of multivariate data with multivariate methods are relatively limited. Here, we present Consensus Clustering (CClust) a new computational method based on clustering of time profiles and posterior identification of correlation between clusters and predictors. Subjects are first clustered in groups according to a response variable temporal profile, using a robust consensus-based strategy. To discover which of the remaining variables are associated with the resulting groups, a non-parametric hypothesis test is performed between groups at every time point, and then the results are aggregated according to the Fisher method. Our approach is tested through its application to the EarlyBird cohort database, which contains temporal variations of clinical, metabolic, and anthropometric profiles in a population of 150 children followed-up annually from age 5 to age 16. Our results show that our consensus-based method is able to overcome the problem of the approach-dependent results produced by current clustering algorithms, producing groups defined according to Insulin Resistance (IR) and biological age (Tanner Score). Moreover, it provides meaningful biological results confirmed by hypothesis testing with most of the main clinical variables. These results position CClust as a valid alternative for the analysis of multivariate longitudinal data.

Identification of Pre-frailty Sub-Phenotypes in Elderly Using Metabolomics.

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.