Deep learning in systems medicine.

Systems medicine (SM) has emerged as a powerful tool for studying the human body at the systems level with the aim of improving our understanding, prevention and treatment of complex diseases. Being able to automatically extract relevant features needed for a given task from high-dimensional, heterogeneous data, deep learning (DL) holds great promise in this endeavour. This review paper addresses the main developments of DL algorithms and a set of general topics where DL is decisive, namely, within the SM landscape. It discusses how DL can be applied to SM with an emphasis on the applications to predictive, preventive and precision medicine. Several key challenges have been highlighted including delivering clinical impact and improving interpretability. We used some prototypical examples to highlight the relevance and significance of the adoption of DL in SM, one of them is involving the creation of a model for personalized Parkinson's disease. The review offers valuable insights and informs the research in DL and SM.

Targeting the gut to prevent and counteract metabolic disorders and pathologies during aging.

Impairment of gut function is one of the explanatory mechanisms of health status decline in elderly population. These impairments involve a decline in gut digestive physiology, metabolism and immune status, and associated to that, changes in composition and function of the microbiota it harbors. Continuous deteriorations are generally associated with the development of systemic dysregulations and ultimately pathologies that can worsen the initial health status of individuals. All these alterations observed at the gut level can then constitute a wide range of potential targets for development of nutritional strategies that can impact gut tissue or associated microbiota pattern. This can be key, in a preventive manner, to limit gut functionality decline, or in a curative way to help maintaining optimum nutrients bioavailability in a context on increased requirements, as frequently observed in pathological situations. The aim of this review is to give an overview on the alterations that can occur in the gut during aging and lead to the development of altered function in other tissues and organs, ultimately leading to the development of pathologies. Subsequently is discussed how nutritional strategies that target gut tissue and gut microbiota can help to avoid or delay the occurrence of aging-related pathologies.

PeakForest: a multi-platform digital infrastructure for interoperable metabolite spectral data and metadata management.

INTRODUCTION: Accuracy of feature annotation and metabolite identification in biological samples is a key element in metabolomics research. However, the annotation process is often hampered by the lack of spectral reference data in experimental conditions, as well as logistical difficulties in the spectral data management and exchange of annotations between laboratories. OBJECTIVES: To design an open-source infrastructure allowing hosting both nuclear magnetic resonance (NMR) and mass spectra (MS), with an ergonomic Web interface and Web services to support metabolite annotation and laboratory data management. METHODS: We developed the PeakForest infrastructure, an open-source Java tool with automatic programming interfaces that can be deployed locally to organize spectral data for metabolome annotation in laboratories. Standardized operating procedures and formats were included to ensure data quality and interoperability, in line with international recommendations and FAIR principles. RESULTS: PeakForest is able to capture and store experimental spectral MS and NMR metadata as well as collect and display signal annotations. This modular system provides a structured database with inbuilt tools to curate information, browse and reuse spectral information in data treatment. PeakForest offers data formalization and centralization at the laboratory level, facilitating shared spectral data across laboratories and integration into public databases. CONCLUSION: PeakForest is a comprehensive resource which addresses a technical bottleneck, namely large-scale spectral data annotation and metabolite identification for metabolomics laboratories with multiple instruments. PeakForest databases can be used in conjunction with bespoke data analysis pipelines in the Galaxy environment, offering the opportunity to meet the evolving needs of metabolomics research. Developed and tested by the French metabolomics community, PeakForest is freely-available at https://github.com/peakforest .

Non-targeted metabolomics analyses by mass spectrometry to explore metabolic stress after six training weeks in high level swimmers.

The objective was to compare the metabolic responses of high-level national swimmers to threshold or polarised training. 22 swimmers (n = 12 males and 10 females) participated in a 28-week cross-over intervention study consisting of 2 × 6 period weeks of training. Swimmers were assigned randomly to either training group for the first period: polarised (POL) (81% in energetic zone 1: blood lactate [La]b ≤ 2 mmol.L-1; 4% in zone 2: 2 mmol.L-1 <[La]b ≤ 4 mmol.L-1; 15% in zone 3: [La]b > 4 mmol.L-1) or threshold (THR) (65%/25%/10%). Before and after each training period, urine samples were collected for non-targeted metabolomics analysis. Mixed model analysis was performed on metabolomics data including fatigue class factors and/or training and/or interaction. Ion intensities of 6-keto-decanoylcarnitine (+31%), pregnanediol-3-glucuronide (+81%), P-cresol sulphate (+18%) were higher in the threshold group (P < 0.05) indicating higher glycogenic depletion and inflammation without alteration of the neuroendocrine stress axis. 4-phenylbutanic acid sulphate was 200% higher in less fatigued swimmers (P < 0.01) linking the anti-inflammatory activity at the cell membrane level to the subjective perception of fatigue. This research suggests the importance of replenishing glycogen stores and reducing inflammation during high thresholds training loads.

Mediterranean diet intervention alters the gut microbiome in older people reducing frailty and improving health status: the NU-AGE 1-year dietary intervention across five European countries.

OBJECTIVE: Ageing is accompanied by deterioration of multiple bodily functions and inflammation, which collectively contribute to frailty. We and others have shown that frailty co-varies with alterations in the gut microbiota in a manner accelerated by consumption of a restricted diversity diet. The Mediterranean diet (MedDiet) is associated with health. In the NU-AGE project, we investigated if a 1-year MedDiet intervention could alter the gut microbiota and reduce frailty. DESIGN: We profiled the gut microbiota in 612 non-frail or pre-frail subjects across five European countries (UK, France, Netherlands, Italy and Poland) before and after the administration of a 12-month long MedDiet intervention tailored to elderly subjects (NU-AGE diet). RESULTS: Adherence to the diet was associated with specific microbiome alterations. Taxa enriched by adherence to the diet were positively associated with several markers of lower frailty and improved cognitive function, and negatively associated with inflammatory markers including C-reactive protein and interleukin-17. Analysis of the inferred microbial metabolite profiles indicated that the diet-modulated microbiome change was associated with an increase in short/branch chained fatty acid production and lower production of secondary bile acids, p-cresols, ethanol and carbon dioxide. Microbiome ecosystem network analysis showed that the bacterial taxa that responded positively to the MedDiet intervention occupy keystone interaction positions, whereas frailty-associated taxa are peripheral in the networks. CONCLUSION: Collectively, our findings support the feasibility of improving the habitual diet to modulate the gut microbiota which in turn has the potential to promote healthier ageing.

Untargeted plasma metabolomic profiles associated with overall diet in women from the SU.VI.MAX cohort.

PURPOSE: Dietary intakes are reflected in plasma by the presence of hundreds of exogenous metabolites and variations in endogenous metabolites. The exploration of diet-related plasma metabolic profiles could help to better understand the impact of overall diet on health. Our aim was to identify metabolomic signatures reflecting overall diet in women from the French general population. METHODS: This cross-sectional study included 160 women in the SU.VI.MAX cohort with detailed dietary data (≥ 10 24-h dietary records) selected according to their level of adherence to the French dietary recommendations, represented by the validated score mPNNS-GS; 80 women from the 10th decile of the score were matched with 80 women from the 1st decile. Plasma metabolomic profiles were acquired using untargeted UPLC-QToF mass spectrometry analysis. The associations between metabolomic profiles and the mPNNG-GS, its components and Principal Component Analyses-derived dietary patterns were investigated using multivariable conditional logistic regression models and partial correlations. RESULTS: Adherence to the dietary recommendations was positively associated with 3-indolepropionic acid and pipecolic acid (also positively associated with fruit and vegetable intake and a healthy diet)-2 metabolites linked to microbiota and inversely associated with lysophosphatidylcholine (LysoPC(17:1)), acylcarnitine C9:1 (also inversely associated with a healthy diet), acylcarnitine C11:1 and 2-deoxy-D-glucose. Increased plasma levels of piperine and Dihydro4mercapto-3(2H) furanone were observed in women who consumed a Western diet and a healthy diet, respectively. Ethyl-ß-D-glucopyranoside was positively associated with alcohol intake. Plasma levels of LysoPC(17:1), cholic acid, phenylalanine-phenylalanine and phenylalanine and carnitine C9:1 decreased with the consumption of vegetable added fat, sweetened food, milk and dairy products and fruit and vegetable intakes, respectively. CONCLUSION: This study highlighted several metabolites from both host and microbial metabolism reflecting the long-term impact of the overall diet. TRIAL REGISTRATION: SU.VI.MAX, clinicaltrials.gov NCT00272428. Registered 3 January 2006, https://clinicaltrials.gov/show/NCT00272428.

Multi-block PLS discriminant analysis for the joint analysis of metabolomic and epidemiological data.

INTRODUCTION: Metabolomics is a powerful phenotyping tool in nutrition and health research, generating complex data that need dedicated treatments to enrich knowledge of biological systems. In particular, to investigate relations between environmental factors, phenotypes and metabolism, discriminant statistical analyses are generally performed separately on metabolomic datasets, complemented by associations with metadata. Another relevant strategy is to simultaneously analyse thematic data blocks by a multi-block partial least squares discriminant analysis (MBPLSDA) allowing determining the importance of variables and blocks in discriminating groups of subjects, taking into account data structure. OBJECTIVE: The present objective was to develop a full open-source standalone tool, allowing all steps of MBPLSDA for the joint analysis of metabolomic and epidemiological data. METHODS: This tool was based on the mbpls function of the ade4 R package, enriched with functionalities, including some dedicated to discriminant analysis. Provided indicators help to determine the optimal number of components, to check the MBPLSDA model validity, and to evaluate the variability of its parameters and predictions. RESULTS: To illustrate the potential of this tool, MBPLSDA was applied to a real case study involving metabolomics, nutritional and clinical data from a human cohort. The availability of different functionalities in a single R package allowed optimizing parameters for an efficient joint analysis of metabolomics and epidemiological data to obtain new insights into multidimensional phenotypes. CONCLUSION: In particular, we highlighted the impact of filtering the metabolomic variables beforehand, and the relevance of a MBPLSDA approach in comparison to a standard PLS discriminant analysis method.

Discovery and Validation of Banana Intake Biomarkers Using Untargeted Metabolomics in Human Intervention and Cross-sectional Studies.

BACKGROUND: Banana is one of the most widely consumed fruits in the world. However, information regarding its health effects is scarce. Biomarkers of banana intake would allow a more accurate assessment of its consumption in nutrition studies. OBJECTIVES: Using an untargeted metabolomics approach, we aimed to identify the banana-derived metabolites present in urine after consumption, including new candidate biomarkers of banana intake. METHODS: A randomized controlled study with a crossover design was performed on 12 healthy subjects (6 men, 6 women, mean ± SD age: 30.0 ± 4.9 y; mean ± SD BMI: 22.5 ± 2.3 kg/m2). Subjects underwent 2 dietary interventions: 1) 250 mL control drink (Fresubin 2 kcal fiber, neutral flavor; Fresenius Kabi), and 2) 240 g banana + 150 mL control drink. Twenty-four-hour urine samples were collected and analyzed with ultra-performance liquid chromatography coupled to a quadrupole time-of-flight MS and 2-dimensional GC-MS. The discovered biomarkers were confirmed in a cross-sectional study [KarMeN (Karlsruhe Metabolomics and Nutrition study)] in which 78 subjects (mean BMI: 22.8; mean age: 47 y) were selected reflecting high intake (126-378 g/d), low intake (47.3-94.5 g/d), and nonconsumption of banana. The confirmed biomarkers were examined singly or in combinations, for established criteria of validation for biomarkers of food intake. RESULTS: We identified 33 potentially bioactive banana metabolites, of which 5 metabolites, methoxyeugenol glucuronide (MEUG-GLUC), dopamine sulfate (DOP-S), salsolinol sulfate, xanthurenic acid, and 6-hydroxy-1-methyl-1,2,3,4-tetrahydro-ß-carboline sulfate, were confirmed as candidate intake biomarkers. We demonstrated that the combination of MEUG-GLUC and DOP-S performed best in predicting banana intake in high (AUCtest = 0.92) and low (AUCtest = 0.87) consumers. The new biomarkers met key criteria establishing their current applicability in nutrition and health research for assessing the occurrence of banana intake. CONCLUSIONS: Our metabolomics study in healthy men and women revealed new putative bioactive metabolites of banana and a combined biomarker of intake. These findings will help to better decipher the health effects of banana in future focused studies. This study was registered at clinicaltrials.gov as NCT03581955 and with the Ethical Committee for the Protection of Human Subjects Sud-Est 6 as CPP AU 1251, IDRCB 2016-A0013-48; the KarMeN study was registered with the German Clinical Trials Register (DRKS00004890). Details about the study can be obtained from https://www.drks.de.

Metabolomics Reveals that the Type of Protein in a High-Fat Meal Modulates Postprandial Mitochondrial Overload and Incomplete Substrate Oxidation in Healthy Overweight Men.

Background: A meal rich in saturated fatty acids induces a postprandial metabolic challenge. The type of dietary protein may modulate postprandial metabolism. Objective: We studied the effect of dietary protein type on postprandial changes in the metabolome after a high-fat meal. Methods: In a 3-period, crossover, postprandial study, 10 healthy overweight men with an elevated waist circumference (>94 cm) ingested high-fat meals made up of cream fat (70% of energy), sucrose (15% energy), and protein (15% energy) from either casein (CAS), whey protein (WHE), or α-lactalbumin-enriched whey protein (LAC). Urine collected immediately before and 2, 4, and 6 h after the meal was analyzed for metabolomics, a secondary outcome of the clinical study. We used mixed-effect models, partial least-square regression, and pathway enrichment analysis. Results: At 4 and 6 h after the meal, the postprandial metabolome was found to be fully discriminated according to protein type. We identified 17 metabolites that significantly explained the effect of protein type on postprandial metabolomic changes (protein-time interaction). Among this signature, acylcarnitines and other acylated metabolites related to fatty acid or amino acid oxidation were the main discriminant features. The difference in metabolic profiles was mainly explained by urinary acylcarnitines and some other acylated products (protein type, Ps < 0.0001), with a dramatically greater increase (100- to 1000-fold) after WHE, and to a lesser extent after LAC, as compared with CAS. Pathway enrichment analysis confirmed that the type of protein had modified fatty acid oxidation (P < 0.05). Conclusion: Taken together, our results indicate that, in healthy overweight men, the type of protein in a high-fat meal interplays with fatty acid oxidation with a differential accumulation of incomplete oxidation products. A high-fat meal containing WHE, but not CAS, resulted in this outpacing of the tricarboxylic acid cycle. This study was registered at clinicaltrials.gov as NCT00931151.

Evaluation of oxidized phospholipids analysis by LC-MS/MS.

Phospholipids (PLs) represent a class of metabolites of interest for evaluating the relationship between diet and the development of several metabolic diseases. Given that PLs are rich in unsaturated fatty acids, they can be oxidized. Because of their structure and reactivity, oxidized phospholipids (PLs-Ox) are increasingly recognized as markers of oxidative stress and of various diseases associated with inflammation. Therefore, there is a growing interest in studying PLs-Ox in lipidomics. Because of their limited commercial availability, very little information is currently available in databases to identify these molecules. The aim of this study is to acquire new knowledge about PLs-Ox in order to propose an analytical strategy for their analyses. For this purpose, a synthesis method of PLs-Ox, in auto-oxidation, has been developed and applied on phosphatidylcholine and phosphatidylethanolamine molecular species with various chain lengths, degree, and position of unsaturations. An analysis method based on mass (MS) and tandem mass spectrometry coupled to electrospray ionization was then developed and enabled the identification of a great diversity of long- and short-chain oxidation products. Formation kinetics of oxidation products was evaluated. Results showed that the formation of oxidized compounds was largely influenced by the degree of unsaturation on fatty acid chains. Oxidation time promotes the formation of some biologically important oxidation products. Coupling the MS method with liquid chromatography in flow injection analysis mode enabled the development of a full analytical strategy. Structural analysis of PLs-Ox allowed the enrichment of databases with important information to identify these molecules in biological matrices.

Metabolic adaptations to HFHS overfeeding: how whole body and tissues postprandial metabolic flexibility adapt in Yucatan mini-pigs.

PURPOSE: In the present study, we aimed to metabolically characterize the postprandial adaptations of the major tissues involved in energy, lipids and amino acids metabolisms in mini-pigs. METHOD: Mini-pigs were fed on high-fat-high-sucrose (HFHS) diet for 2 months and several tissues explored for metabolic analyses. Further, the urine metabolome was followed over the time to picture the metabolic adaptations occurring at the whole body level following overfeeding. RESULTS: After 2 months of HFHS consumption, mini-pigs displayed an obese phenotype characterized by high circulating insulin, triglycerides and cholesterol levels. At the tissue level, a general (muscle, adipose tissue, intestine) reduction in the capacity to phosphorylate glucose was observed. This was also supported by the enhanced hepatic gluconeogenesis potential, despite the concomitant normoglycaemia, suggesting that the high circulating insulin levels would be enough to maintain glucose homoeostasis. The HFHS feeding also resulted in a reduced capacity of two other pathways: the de novo lipogenesis, and the branched-chain amino acids transamination. Finally, the follow-up of the urine metabolome over the time allowed determining breaking points in the metabolic trajectory of the animals. CONCLUSIONS: Several features confirmed the pertinence of the animal model, including increased body weight, adiposity and porcine obesity index. At the metabolic level, we observed a perturbed glucose and amino acid metabolism, known to be related to the onset of the obesity. The urine metabolome analyses revealed several metabolic pathways potentially involved in the obesity onset, including TCA (citrate, pantothenic acid), amino acids catabolism (cysteine, threonine, leucine).

Systems Metabolomics for Prediction of Metabolic Syndrome.

The evolution of human health is a continuum of transitions, involving multifaceted processes at multiple levels, and there is an urgent need for integrative biomarkers that can characterize and predict progression toward disease development. The objective of this work was to perform a systems metabolomics approach to predict metabolic syndrome (MetS) development. A case-control design was used within the French occupational GAZEL cohort (n = 112 males: discovery study; n = 94: replication/validation study). Our integrative strategy was to combine untargeted metabolomics with clinical, sociodemographic, and food habit parameters to describe early phenotypes and build multidimensional predictive models. Different models were built from the discriminant variables, and prediction performances were optimized either when reducing the number of metabolites used or when keeping the associated signature. We illustrated that a selected reduced metabolic profile was able to reveal subtle phenotypic differences 5 years before MetS occurrence. Moreover, resulting metabolomic markers, when combined with clinical characteristics, allowed improving the disease development prediction. The validation study showed that this predictive performance was specific to the MetS component. This work also demonstrates the interest of such an approach to discover subphenotypes that will need further characterization to be able to shift to molecular reclassification and targeting of MetS.

Time Course of Molecular and Metabolic Events in the Development of Insulin Resistance in Fructose-Fed Rats.

We aimed to determine the time-course of metabolic changes related to the early onset of insulin resistance (IR), trying to evidence breaking points preceding the appearance of the clinical IR phenotype. The model chosen was the fructose (FRU)-fed rat compared to controls fed with starch. We focused on the hepatic metabolism after 0, 5, 12, 30, or 45 days of FRU intake. The hepatic molecular metabolic changes followed indeed a multistep trajectory rather than a continuous progression. After 5 d of FRU feeding, we observed deep modifications in the hepatic metabolism, driven by the induction of lipogenic genes and important glycogen depletion. Thereafter, a steady-state period between days 12 and 30 was observed, characterized by a switch from carbohydrate to lipid utilization at the hepatic level and increased insulin levels aiming at alleviating lipid accumulation and hyperglycemia, respectively. The FRU-fed animals were only clinically IR at day 45 (altered homeostasis model assessment-estimated insulin resistance and muscle glucose transport). Furthermore, the urine metabolome revealed even earlier metabolic trajectory changes that precede the hepatic alterations. We identified several candidate metabolites linked to the tryptophan-nicotinamide metabolism and the installation of fasting hyperglycemia that suggest a role of this metabolic pathway on the development of the IR phenotype in the FRU-fed rats.

Red meat and colorectal cancer: Nrf2-dependent antioxidant response contributes to the resistance of preneoplastic colon cells to fecal water of hemoglobin- and beef-fed rats.

Epidemiological studies have associated red meat intake with risk of colorectal cancer. Experimental studies explain this positive association by the oxidative properties of heme iron released in the colon. This latter is a potent catalyst for lipid peroxidation, resulting in the neoformation of deleterious aldehydes in the fecal water of heme-fed rats. The toxicity of fecal water of heme-fed rats was associated to such lipid peroxidation. This study demonstrated that fecal water of hemoglobin- and beef-fed rats preferentially induced apoptosis in mouse normal colon epithelial cells than in those carrying mutation on Apc (Adenomatous polyposis coli) gene, considered as preneoplastic. Highlighting the importance of lipid peroxidation and neoformation of secondary aldehydes like 4-hydroxy-2-nonenal (HNE), we optimized the depletion of carbonyl compounds in the fecal water which turned out to abolish the differential apoptosis in both cell lines. To explain the resistance of preneoplastic cells towards fecal water toxicity, we focused on Nrf2, known to be activated by aldehydes, including HNE. Fecal water activated Nrf2 in both cell lines, associated with the induction of Nrf2-target genes related to aldehydes detoxification. However, the antioxidant defense appeared to be higher in preneoplastic cells, favoring their survival, as evidenced by Nrf2 inactivation. Taken together, our results suggest that Nrf2-dependent antioxidant response was involved in the resistance of preneoplastic cells upon exposure to fecal water of hemoglobin- and beef-fed rats. This difference could explain the promoting effect of red meat and heme-enriched diet on colorectal cancer, by initiating positive selection of preneoplastic cells.

Variations in the metabolome in response to disease activity of rheumatoid arthritis.

BACKGROUND: Anti-Tumor Necrosis Factor (TNF) therapies are able to control rheumatoid arthritis (RA) disease activity and limit structural damage. Yet no predictive factor of response to anti-TNF has been identified. Metabolomic profile is known to vary in response to different inflammatory rheumatisms so determining it could substantially improve diagnosis and, consequently, prognosis. The aim of this study was to use mass spectrometry to determine whether there is variation in the metabolome in patients treated with anti-TNF and whether any particular metabolomic profile can serve as a predictor of therapeutic response. METHODS: Blood samples were analyzed in 140 patients with active RA before initiation of anti-TNF treatment and after 6 months of Anti-TNF treatment (100 good responders and 40 non-responders). Plasma was deproteinized, extracted and analyzed by reverse-phase chromatography-QToF mass spectrometry. Extracted and normalized ions were tested by univariate and ANOVA analysis followed by partial least-squares regression-discriminant analysis (PLS-DA). Orthogonal Signal Correction (OSC) was also used to filter data from unwanted non-related effects. Disease activity scores (DAS 28) obtained at 6 months were correlated with metabolome variation findings to identify a metabolite that is predictive of therapeutic response to anti-TNF. RESULTS: After 6 months of anti-TNF therapy, 100 patients rated as good responders and 40 patients as non-responders according to EULAR criteria. Metabolomic investigations suggested two different metabolic fingerprints splitting the good-responders group and the non-responders group, without differences in anti-TNF therapies. Univariate analysis revealed 24 significant ions in positive mode (p < 0.05) and 31 significant ions in negative mode (p < 0.05). Once intersected with PLS results, only 35 ions remained. Carbohydrate derivates emerged as strong candidate determinants of therapeutic response. CONCLUSIONS: This is the first study describing metabolic profiling in response to anti-TNF treatments using plasma samples. The study highlighted two different metabolic profiles splitting good responders from non-responders.

Development of a LC-MS/MS method for the simultaneous screening of seven water-soluble vitamins in processing semi-coarse wheat flour products.

Wheat is the second largest crop cultivated around the world and constitutes a major part of the daily diet in Europe. It is therefore important to determine the content of micronutrient in wheat and wheat-based food products to define the contribution of wheat-based foods to the nutrition of the consumers. The aim of the present work was to develop a simple and rapid method based on liquid chromatography tandem mass spectrometry (LC-MS/MS) for the simultaneous determination of seven water-soluble vitamins in various wheat-based food materials. The vitamins present in the test material were separated in less than 15 min by using a reverse-phase C18 column, and analyzed by positive ion electrospray selected reaction monitoring MS/MS. The MS response for all the vitamins was linear over the working range (0.05 to 9 µg/mL) with correlation coefficients ranging between 0.991 and 1. Limits of quantification in the different food materials ranged from 0.09 to 3.5 µg/g. Intra-day and inter-day precision were found satisfactory. The developed method was applied for the simultaneous analysis of the water-soluble vitamin natural content of different semi-coarse wheat flours and in their corresponding baking products.

Untargeted metabolomics as a screening tool for estimating compliance to a dietary pattern.

There is a growing interest in studying the nutritional effects of complex diets. For such studies, measurement of dietary compliance is a challenge because the currently available compliance markers cover only limited aspects of a diet. In the present study, an untargeted metabolomics approach was used to develop a compliance measure in urine to distinguish between two dietary patterns. A parallel intervention study was carried out in which 181 participants were randomized to follow either a New Nordic Diet (NND) or an Average Danish Diet (ADD) for 6 months. Dietary intakes were closely monitored over the whole study period, and 24 h urine samples as well as weighed dietary records were collected several times during the study. The urine samples were analyzed by UPLC-qTOF-MS, and a partial least-squares discriminant analysis with feature selection was applied to develop a compliance model based on data from 214 urine samples. The optimized model included 52 metabolites and had a misclassification rate of 19% in a validation set containing 139 samples. The metabolites identified in the model were markers of individual foods such as citrus, cocoa-containing products, and fish as well as more general dietary traits such as high fruit and vegetable intake or high intake of heat-treated foods. It was easier to classify the ADD diet than the NND diet probably due to seasonal variation in the food composition of NND and indications of lower compliance among the NND subjects. In conclusion, untargeted metabolomics is a promising approach to develop compliance measures that cover the most important discriminant metabolites of complex diets.

Discovery and validation of urinary exposure markers for different plant foods by untargeted metabolomics.

While metabolomics is increasingly used to investigate the food metabolome and identify new markers of food exposure, limited attention has been given to the validation of such markers. The main objectives of the present study were to (1) discover potential food exposure markers (PEMs) for a range of plant foods in a study setting with a mixed dietary background and (2) validate PEMs found in a previous meal study. Three-day weighed dietary records and 24-h urine samples were collected three times during a 6-month parallel intervention study from 107 subjects randomized to two distinct dietary patterns. An untargeted UPLC-qTOF-MS metabolomics analysis was performed on the urine samples, and all features detected underwent strict data analyses, including an iterative paired t test and sensitivity and specificity analyses for foods. A total of 22 unique PEMs were identified that covered 7 out of 40 investigated food groups (strawberry, cabbages, beetroot, walnut, citrus, green beans and chocolate). The PEMs reflected foods with a distinct composition rather than foods eaten more frequently or in larger amounts. We found that 23 % of the PEMs found in a previous meal study were also valid in the present intervention study. The study demonstrates that it is possible to discover and validate PEMs for several foods and food classes in an intervention study with a mixed dietary background, despite the large variability in such a dataset. Final validation of PEMs for intake of foods should be performed by quantitative analysis.

Mass spectrometry-based metabolomics for the discovery of biomarkers of fruit and vegetable intake: citrus fruit as a case study.

Elucidation of the relationships between genotype, diet, and health requires accurate dietary assessment. In intervention and epidemiological studies, dietary assessment usually relies on questionnaires, which are susceptible to recall bias. An alternative approach is to quantify biomarkers of intake in biofluids, but few such markers have been validated so far. Here we describe the use of metabolomics for the discovery of nutritional biomarkers, using citrus fruits as a case study. Three study designs were compared. Urinary metabolomes were profiled for volunteers that had (a) consumed an acute dose of orange or grapefruit juice, (b) consumed orange juice regularly for one month, and (c) reported high or low consumption of citrus products for a large cohort study. Some signals were found to reflect citrus consumption in all three studies. Proline betaine and flavanone glucuronides were identified as known biomarkers, but various other biomarkers were revealed. Further, many signals that increased after citrus intake in the acute study were not sensitive enough to discriminate high and low citrus consumers in the cohort study. We propose that urine profiling of cohort subjects stratified by consumption is an effective strategy for discovery of sensitive biomarkers of consumption for a wide range of foods.

Global and Partial Effect Assessment in Metabolic Syndrome Explored by Metabolomics.

In nutrition and health research, untargeted metabolomics is actually analyzed simultaneously with clinical data to improve prediction and better understand pathological status. This can be modeled using a multiblock supervised model with several input data blocks (metabolomics, clinical data) being potential predictors of the outcome to be explained. Alternatively, this configuration can be represented with a path diagram where the input blocks are each connected by links directed to the outcome-as in multiblock supervised modeling-and are also related to each other, thus allowing one to account for block effects. On the basis of a path model, we show herein how to estimate the effect of an input block, either on its own or conditionally to other(s), on the output response, respectively called "global" and "partial" effects, by percentages of explained variance in dedicated PLS regression models. These effects have been computed in two different path diagrams in a case study relative to metabolic syndrome, involving metabolomics and clinical data from an older men's cohort (NuAge). From the two effects associated with each path, the results highlighted the complementary information provided by metabolomics to clinical data and, reciprocally, in the metabolic syndrome exploration.