Decoding the Metabolome and Lipidome of Child Malnutrition by Mass Spectrometric Techniques: Present Status and Future Perspectives

Abstract

Child malnutrition (CM) is a global public health problem. It contributes to poor health in one in four children under five years worldwide and causes serious health problems in children including stunted, wasted, and overweight growth. These serious public health issues lead to a higher chance of living in poverty in adulthood and consequently reduced economic productivity and increases the serious national and international burden. Currently, there is no meaningful therapeutic intervention of CM, and the use of different therapeutic foods have shown exceedingly poor outcomes among supplemented malnourished children. The role of metabolites and lipids has been extensively recognized as early determinants of child health, but their contribution in CM and its pathobiology are poorly understood. This perspective provides a most recent update on these aspects. After briefly introducing the disciplines of metabolomics and lipidomics, we describe a mass spectrometry-based metabolic workflow for analysis of both metabolites and lipids and summarize several recent applications of metabolomics and lipidomics in CM. Finally, we discuss the future directions of the field towards the development of meaningful interventions for CM through metabolomics and lipidomics advances.

Graphical Abstract

An overview of malnutrition

Malnutrition is a life threating condition for children in developing countries, which includes undernutrition (stunting and wasting) and overnutrition (obesity). According to a recent combined report by the United Nations Children’s Fund (UNICEF), the World Health Organization (WHO) and the World Bank Group, stunting affected approximately 22.2% (151 million) of children under 5 years of age, while wasting continued to threaten the lives of approximately 7.5% (501 million) of children under the age of 5. Furthermore, an estimated 5.6% (38 million) of children under the age of 5 globally were overweight in 2017 (UNICEF/WHO/World Bank Group, 2018)¹. Moreover, Child malnutrition (CM) is directly or indirectly associated with 35% of deaths among children under five years worldwide². Survivors with malnutrition suffer from numerous health problems including growth failures, cardiovascular complications, neurological disorders, impaired cognitive and motor development, social-emotional development, and increased risk of early morbidity and mortality^3–5. Numerous factors are associated with CM including poor nutrition, maternal health, parental history, socio-economic status, microbiome and other pathogenic infections, education, healthcare, food insecurity, and other environmental variables (Figure 1)^6–11. Low- and middle-income countries people who developed childhood malnutrition tending to become prone to many health related complexity including obesity and diabetics in adulthood (Ivana 2012 & Tzioumis 2014). Long-lasting impacts of malnutrition exacerbates the costs for the public health system and the poverty of the nation (Martins 2011).

Figure 1.

Some key factors associated with child malnutrition and its consequences in children health.

Recently, nutrient-based food supplementation, such as lipid-based nutrient supplements or Ready-To-Use Therapeutic Food (RUTFs), micronutrients such as vitamin A, zinc, iron individually or in combinations have been conducted on diverse CM studies^9,12–15. Unfortunately, these interventions resulted in poor outcomes such as suboptimal growth (Choudhury 2016) among supplemented malnourished children^15–18. Current knowledge regarding poor outcome by food supplementation in malnourished children is not quite understood (Uchiyama 2018 & Bhutta 2017). However, several studies recently emphasis that dysregulated epigenetics marks that were established at germ cells under malnourished condition could be transmitted and maintained in the subsequent offspring via the paternal lineage and impacts metabolism including lipids, thereby influencing health and disease risk (Uchiyama 2018 and Martinez 2014). Furthermore, failure of food supplementation in malnourished systems is expected because critical health problems due to malnutrition onset in most cases within the first two years of life and can persist even if nutrition becomes adequate later in life (Uchiyama 2018, Carlon SE 2009, Martinez 2014). Therefore, further research and analytical innovation is needed to improve our understanding of the pathobiology of severe malnutrition towards the development of meaningful therapeutic interventions for malnourished children, especially the molecular mechanisms underlying childhood health in the malnourished state.

Recently, discovery of metabolite or lipid-based biomarkers through metabolomics or lipidomics has increased our understanding of pathophysiology and improved therapeutic strategies for the development of precision medicine for many diseases with unmet challenges^19–24. For example, significantly upregulated levels of corticosterone²⁵ and sarcosine in prostate cancer²⁰, 27-hydroxycholesterol levels in hypercholesterolemia and breast cancer²⁶, and elevated levels of branched chain amino acids (BCAA) in pancreatic adenocarcinoma²¹ and myeloid leukemia²⁷ are now linked with cancer progression and metastasis. Recently mass spectrometry-based metabolomics and lipidomics studies demonstrated that malnutrition results in serious dysfunction of the metabolome and lipidome profiles in malnourished children compared to healthy children and thus opens an emerging avenue to develop a new interventions for improved management of CM through identification of metabolic fingerprints^7,23,28–30.

An overview of metabolomics, lipidomics and Mass Spectrometry (MS) techniques

Metabolomics and lipidomics are newly emerging ‘Omics approaches in bioscience towards the development of precision therapeutic and clinical interventions (Figure 2). Metabolomics launched in 1999 and Lipidomics emerged in 2003 and both ‘Omics approaches have greatly advanced since that initial work, largely due to the development of mass spectrometry and its allied techniques^19,31. Metabolomics involves the holistic characterization of small molecules, called metabolites, such as amino acids, organic acids, nucleotides, carbohydrates, and secondary metabolites in biological systems^19,31. On the other hand, lipidomics elucidates the complete set of lipids such as free fatty acids (FFAs), saturated fatty acids (SFAs), monounsaturated fatty acids (MUFA), polyunsaturated fatty acids (PUFA), glycerolipids, glycerophospholipids, sphingolipids, and sterols in a given biological system³².

Figure 2.

Model of precision medicine development through multi-omics approaches. Multi-omics includes genomics (study of gene), transcriptomics (study of gene expression), proteomics (study of protein), and metabolomics (study of small molecules metabolites or lipids). Systems biology followed by precision medicine incorporates comprehensive knowledge of multi-omics and external factors towards suitable therapy development.

Metabolites and lipids represent both the input and the output of cellular and physiological processes and their levels are therefore exquisitely sensitive to a wide range of perturbations linked to disease, genetic modification and environmental conditions^19,33. Precision medicine or personalized medicine is an active component of today’s medical practice for protecting health and treating disease through understanding individual patient information. Metabolite or lipid levels represent both the downstream output of the genomics (study of the genome), epi-genomics (study of the chromatin occupancy), transcriptomics (study of the gene expression or transcript), and proteomics (study of protein expression), and the upstream input from the life style and/or environment, and therefore the metabolome and lipidome demonstrate an exciting inclusive tool for precision medicine initiatives (Figure 2). Recently, systems biology or integrated ‘Omics approaches have evolved as a scientific discipline to obtain a holistic understanding of global metabolism for the development of personalized medicine^19,34.

Recent applications of Mass Spectrometry-based techniques such as direct MS analysis and chromatography coupled to MS analysis in metabolomics and lipidomics have made tremendous advances in biomedical research^35–37. Numerous techniques and tools have been incorporated in modern mass spectrometry for metabolomics and lipidomics studies^32,35,36,38. For example, direct MS technologies such as Time-of-Flight mass spectrometers (TOF-MS), orbitrap MS, Fourier transform ion cyclotron mass spectrometers (FT-ICR-MS), paper spray ionization mass spectrometry (PSI-MS) have broaden the metabolomics and lipidomics research by providing high spectral resolution and mass accuracy as well as simplified sample collection. Although direct MS techniques have been used for the analysis of large number of analytes, they have shown several limitations including susceptibility to ion suppression or enhancement, less efficient to distinguish unique metabolite ions from adduct and product ions as well as has inability to differentiate isomers^39–42. On the other hand, chromatography coupled with MS provides high level of confidence, accuracy, and advantages such as reduction of matrix effects and ion suppression, isomer separation, retention time index for metabolite annotation, and accurate quantification of metabolite or lipid. For example, Gas Chromatography coupled to Mass Spectrometry (GC-MS) is highly advantageous for both volatile and nonvolatile complex mixture analyses following derivatization. GC interfaced to several mass analyzers including single quadrupole MS detectors, TOF-MS, or triple quadrupole MS and has been efficiently employed to study metabolome and lipidome of diverse biological studies^32,35–38. Unfortunately, GC-MS is less efficient to detect the large number of polar metabolites expected. Therefore, efficient separation and detection of these compounds require liquid phase separation tools such as liquid chromatography (LC) coupled to MS.

Among all, Ultra High Performance Liquid Chromatography coupled with High Resolution Mass Spectrometry (UHPLC-HRMS) is a gold standard analytical tool for metabolomics or lipidomics studies due to its high-throughput, reproducibility, dynamic range of metabolome or lipidome coverage, and high mass accuracy^24,43,44. UHPLC-HRMS-based metabolomics and lipidomics can be either untargeted, in which case global metabolites or lipid profiling is usually performed, or targeted, in which metabolites of interest are already known and directly analyzed^45,46. The decision to perform an untargeted analysis versus a targeted analysis is typically decided based on whether there is a hypothesis. Untargeted metabolomics is often considered hypothesis generating because it aims to analyze as many metabolites as possible and then utilize bioinformatics to identify metabolites associated with the study population. This type of analysis requires defining the study population very well to ensure a well-represented sample set that then builds a hypothesis based on the differential metabolites. On the other hand, a targeted analysis typically begins with a hypothesis that suggests a key set of metabolites is altered and then uses the study population to investigate whether that smaller set of metabolites is different.

CM is a vast area of research and onset of CM is very complex. UHPLC-HRMS based metabolomics or lipidomics can be utilized to investigate this serious global public health problem towards the discovery of metabolic fingerprints for the early detection and management of CM. Recently, limited scale targeted metabolomics investigations into CM have been conducted using LC-MS and uncovered the remarkable potential of metabolites and lipids as a therapeutic interventions for malnourished children^7,28,29,47. Future studies examining the untargeted global metabolome and lipidome of malnourished and healthy children or corresponding animal models will be warranted to address CM.

Metabolomics and lipidomics workflow

A typical workflow for metabolomics or lipidomics study includes four major steps such as study design (1), sample preparation (2), data acquisition (3), and statistical analysis (4) (Figure 3). Prior to any metabolomics or lipidomics study, proper study design is critically important. Metabolites and/or lipids change very rapidly in response to subtle changes in the environment. Therefore, it is essential to minimize any biological, technical or experimental variability while conducting metabolomics research towards the development of precision medicine. Consistency is one key to minimizing variation. To achieve reproducibility of metabolomics result, performing every single step the same as the first time is crucial. Prior to any metabolic profile analysis, understanding study types (basic or translational or clinical studies), study question, sample replication (at least 5 or more for basic translational study and at least 10 or more clinical study), sample randomization and stratification, sample categories (cell, microbes, parasites, tissue, organ, biofluids such as whole blood, plasma, serum, urine, saliva, seminal fluids, amniotic fluids, cerebrospinal fluid, synovial fluid, etc), sampling conditions (solid, suspension, adherent, trypsinizing, scratching, etc), sample preprocessing and storage conditions (temperature, biobanking, aliquoting, etc) are very important. Several extensive resources and reviews that have been published recently on study design of metabolomics and lipidomics^19,48,49. In this section, we summarized key considerations during experimental design for global metabolomics and lipidomics using UHPLC-HRMS in biomedical research (Figure 3).

Figure 3.

High-throughput metabolomics and lipidomics workflow using UHPLC-HRMS. (Step 1) study design, which includes selection of study subjects for example human or animal, sample types and replications. (Step 2) sample preparation includes most commonly used biphasic (metabolite and lipid) extraction method following modified Folch methods (chloroform/methanol/water = 2:1:1). (Step 3) data collection and data matrix generation. (Step 4) Statistical analysis and data visualization includes clustering analysis by multivariate score plot and heatmap analysis, fold change analysis, metabolic pathway, and biomarker discovery.

The collected samples under proper experimental design are then prepared for metabolite/lipid extraction. Extraction methods for global metabolites and lipids have been extensively reviewed recently³⁸. Most MS-based metabolomics or lipidomics techniques utilize biological extracts except imaging MS which utilize tissue slices^50,51. In this workflow, we suggest most commonly used biphasic (metabolite and lipid) extraction method following modified Folch methods (chloroform/methanol/water = 2:1:1) (Figure 3) that our lab and others successfully utilized for diverse research in bioscience^43,52,53. Quality control (QC) methods are critically needed for sample preparation that include proper internal standards (isotopically labeled) to account for extraction variations and measurement of total protein concentration (cells and tissue samples), total DNA/RNA concentration (cell and tissue), wet/dry tissue weight, or fluid volume are commonly utilize for pre/post-data normalization⁴³.

The properly extracted metabolome or lipidome are then prepared for suitable MS analysis. During MS data acquisition, global LC-MS based metabolomics and lipidomics are commonly utilized for the analysis of metabolome or lipidome solutions^36,38. On the other hand, tissue slices can be directly submitted to MS imaging^45,51. The most popular MS ionization techniques for metabolomics or lipidomics are Electrospray ionization (ESI), Atmospheric Pressure Chemical Ionization (APCI), Atmospheric Pressure Photoionization (APPI), Matrix-assisted Laser Desorption/Ionization (MALDI), Desorption ESI (DESI), and Paper Spray Ionization Mass Spectrometry (PSI-MS). Of these approaches, ESI is by far the most common approach when coupled with LC for both metabolite and lipid analyses^35,36,38. After MS data acquisition, spectral data are processed for deisotoping in order to generate a data matrix, followed by identification and quantification of metabolites or lipids species using data processing software or tools compatible with analytical techniques (eg. global, LC-MS, or imaging) (Mzmine, XCMS, Lipidsearch)^54,55. The data matrix which contains both qualitative and quantitative information for metabolites or lipids are then ready for further statistical analysis to explore metabolite or lipid-based biomarker discovery and its underlying mechanism related to disease state (Figure 3).

Although, this section mainly focused UHPLC-HRMS technology, but, GC-MS or NMR (Nuclear Magnetic Resonance) technology-based metabolomics or lipidomics can be equally applied in children malnutrition following this workflow. Considering range of sensitivities and detection capacities of metabolomic technologies, LC-MS is highly sensitive and show higher detection capacity than GC-MS or NMR (Wishart 2011). However, elucidating some lipid isomers in complex mixtures by LC-MS is still problematic. Recently, LC-IMS-MS, integrated ion mobility spectrometry (IMS) between LC and MS separations, emerge as an efficient technique to identify isomeric changes of lipids occurring under complex mixtures (Kyle 2016 Analyst). Therefore, IMS could be a potentially important tools coupled with LC-MS to discern lipid analysis of malnutrition studies. On the other hand, although NMR is less sensitive but it is highly specific, reproducible and has ability to detect low concentrations of metabolites from a distinct specimens (Mahmud 2015, 2014 & Emmanuel 2018).

Applications in Children malnutrition

CM is a complex public health problem worldwide due to its severity and heterogeneity; however, MS-based metabolomics and lipidomics approaches recently offer an exciting opportunity to understand this complexity (Table 1). Moreau et al. recently reported an association of targeted metabolic signatures with growth and neurocognitive outcomes of nine-month-old children. Study identified that higher concentrations of hydroxy-sphingomyelin and essential amino acids and lower levels of acylcarnitines and bile acid conjugation are associated with improved health outcome of childen⁴⁷. Semba et al. report a targeted metabolomics approach to measure 139 serum metabolites including amino acids, glycerophospholipids, sphingolipids, and other metabolites in 313 children based on Height-for-age Z-scores (HAZ) and weight-for-height Z-scores, aged 12–59 months, from rural Malawi, of whom 60% were stunted. Using LC-MS/MS, this study identified lower serum concentrations of all nine essential amino acids (tryptophan, isoleucine, leucine, valine, methionine, threonine, histidine, phenylalanine, lysine) and also lower serum concentrations of conditionally essential amino acids (arginine, glycine, glutamine), non-essential amino acids (asparagine, glutamate, serine), and six different sphingolipids compared with non-stunted children. This study also found an association of stunting with alterations in serum glycerophospholipid concentrations. Overall, Semba et al findings support the idea that children with a high risk of stunting may not be receiving an adequate dietary intake of essential amino acids and choline, an essential nutrient for the synthesis of sphingolipids and glycerophospholipids²⁹. It was recognized that amino acids particularly leucine, tryptophan, and lysine are found very critical for childhood early developmental stages, where reduced levels of these amino acids was associated with stunted growth of children. Therefore, the metabolome findings from Semba et al could provide important insight towards the understanding of the management strategies for CM.

Table 1.

Metabolomic studies performed in children malnutrition studies.

Study Design	Technique	Results and implications	Resources
Plasma samples were collected from children aged at 9 and 36 months (n=130)47	LC-MS/MS, targeted	Revealed for the first time that metabolic signature is associated with neurodevelopment in an undernourished child cohort. Study identified 34 metabolites were associated with Height-for-Age z-score (HAZ), 37 with ΔHAZ (average HAZ), and 34 with full-scale IQ (e intelligence quotient).	7,23,28–30,47,57
Serum samples of children (stunted vs non-stunted) aged <5.0 years23,28,29	LC-MS, targeted	Uncover the decreased concentrations of circulating amino acids including essential amino acids, conditionally essential amino acids, non-essential amino acids, phosphotidylcholine, and sphingolipids in children with stunted growth compared with nonstunted (Semba 2016). Unravel lower level of w3 and w6 long-chain PUFAs, carnitine, nerosteroids, and glutathione metabolism in stunted children (Semba 2017).
Serum samples of children (Kwashiorkor vs marasmus vs community controls) aged <5.0 years56	LC-MS, targeted	Clustering analysis identified children with SAM (kwashiorkor) were metabolically distinct from those with marasmus, and were more prone to severe metabolic disruptions. Several amino acids and biogenic amines, including those of the kynurenine-tryptophan pathways were highly reduced in SAM compared to other study groups.
Serum samples of children (aged <5.0 years) with and without environmental enteric dysfunction (EED)30	LC-MS, targeted	New insights into the relationship between gut permeability and alteration of serum metabolites (Semba 2016).
Plasma samples of Children (aged <2.0 years)7 Low income countries vs high-income countries children 1. Gut microbes of malnourished vs well-nourished children Healthy vs stunted children	LC-MS, targeted	For the first time, this study identified reduced metabolite production capabilities in children from low income countries compared with a high-income country. Consistent with Semba et al this study also identified that stunted children had reduced levels of essential amino acids as well as lower ratio of tryptophan to other neutral amino acids compared to the healthy group. Moreover, this study also identified that gut microbiota is associated with malnutrition and reduced amino acid levels. (kumar 2018)
Fecal samples of children: malnourished with cholera, malnourished without cholera, and healthy57	HPLC	Various short chain fatty acid metabolites (SCFA) such as acetate, propionate, valerate, and butyrate concentration were significantly lower in malnourished with cholera than that in the malnourished without cholera and healthy condition (Monira 2010)

Semba et al. also evaluated 677 targeted serum metabolites in a cross-sectional study of 400 Malawian children aged 12–59 month, of whom 62% were stunted. Using LC-MS/MS following volcano plot based differential statistical analysis tools, this study revealed a potential association of stunted growth of children with lower serum concentrations of 1) ω−3 (n–3) and ω−6 (n–6) polyunsaturated fatty acids (PUFAs), 2) sulfated neurosteroids, which play a role in brain development, 3) carnitine, a conditionally essential nutrient with an important role in the carnitine shuttle for the metabolism of fatty acids and energy production, and 4) gamma-glutamyl amino acids, which represent an altered gamma-glutamyl cycle of glutathione metabolism. This study also identified significantly higher serum concentrations of five biomarkers such as cotinine, catechol sulfate, 4-vinylphenol sulfate; trigonelline (N’-methylnicotinate), and N-(2-furoyl) glycine associated to cigarette smoke exposure²⁸. Remarkably, this metabolomics and lipidomics finding steered toward a clinical trial in order to explore efficacy of metabolic intervention towards the management of CM.

Furthermore, targeted LC-MS/MS was used by Giovanni et al in a study of 40 severe acute malnourished (SAM) Malawian children aged 9–59 month, of whom 21 with kwashiorkor (characterized by nutritional edema and metabolic disturbances, including hypoalbuminemia and hepatic steatosis) and 19 with marasmus (characterized by severe wasting) or living in the community (n = 157; 78 stunted and 79 non-stunted). This study identified significantly lower levels of several amino acids and biogenic amines, including those of the kynurenine-tryptophan metabolic pathway. This study summarized that children with kwashiorkor were metabolically distinct from those with marasmus, and were more prone to severe metabolic disruptions. Both SAM conditions (kwashiorkor and marasmus) showed distinct metabolic profiles from stunted and non-stunted controls, even after clinical stabilization, therefore, suggesting critical outcomes for future a SAM treatment plan⁵⁶.

Moreover, Kumar et al. examined genome-scale metabolic modeling to understand the metabolic functions and interactions between individual species in the gut microbiota during healthy and malnutrition states of Bangladeshi children under the age of 5. Previously, it was reported persistent gut microbiota immaturity in Bangladeshi malnourished children⁸. Using targeted LC-MS/MS, Kumar et al. identified the gut microbial dysbiosis, which was associated with malnutrition and reduced plasma amino acid levels in Bangladeshi malnourished children. Therefore, this study provided a framework for future studies towards further characterization of gut microbial metabolic capabilities and their contribution to malnutrition⁷.

Conclusions and future perspectives

Nutrition has intense effects on human health throughout a person’s lifetime. Metabolome and lipidome at suitable physiological levels represent the early determinants of child health, and most importantly their levels are associated with cognitive and social development of children^4,5. Metabolomics in a nutritional context has already been employed to investigate and characterize relevant features of the metabolic phenotype. As we discussed, LC-MS based targeted metabolic profiling studies of child malnutrition revealed some potential interventions for the reduction of child stunting and SAM^7,28,47. Currently Ready-to-Use therapeutic food (RTUF), mostly based on metabolite or lipid supplementation, is a commonly used intervention for malnourished children^12,58. Unexpectedly, RTUF led to exceedingly poor outcomes among supplemented malnourished children^7,16–18,29. Although targeted metabolic studies based on LC-MS in child malnutrition provide some potential interventions for severe malnutrition, however, global profile of metabolome or lipidome in children with malnutrition is poorly studied and its expand application in CM could reveal more rigorous output on the pathophysiology and the management of SAM.

The metabolome and lipidome are the major molecular components of biological systems. According to The Human Metabolome Database (HMDB), there are over 40,000 metabolites and lipids that have already been identified or are likely to be found in the human body⁵⁹. UHPLC-HRMS is an emerging analytical technique in biomedical sciences and has a unique strength to profile and identify the metabolic fingerprints in a wide variety of samples including biofluids and tissues¹⁹. In the recent years, untargeted metabolomics or lipidomics using UHPLC-HRMS technologies have enabled to detection over 3000 metabolic features, but have not yet been employed in CM based studies^43,44. Notably, LC-MS based metabolomics or lipidomics technologies are often used to screen metabolite-based biomarker in a wide variety of bioscience disciplines such as inborn errors in metabolism⁶⁰, genotyping⁶¹, pharmacometabolomics⁶², gut microbiome activity⁶³, and nutrition-associated metabolic phenotyping^64,65. Evidently, CM is a major health and wellness problem in developing countries where relatively rapid analytical technologies with little or no sample preparation and cost effective features are in high demand. Our group recently reported the use of paper spray ionization mass spectrometry (PSI-MS) for untargeted metabolomics research⁴⁰. PSI-MS is a relatively new analytical technique allowing for rapid mass spectrometric analysis of biological samples with little or no sample preparation, incredibly faster data acquisition time (~1min), and much cheaper analysis⁴⁰. There is a considerable interest in the clinic for the inexpensive storage and transportation of liquid biopsy samples such as blood, urine, saliva, and other bodily fluid as dried spots on paper. We believe PSI-MS could be considered the future bedside screening tool for high-throughput screening of global metabolome and lipidome of children with malnutrition.

In summary, emerging application of MS technologies-based metabolomics or lipidomics in CM could open exciting research avenue in global malnutrition research and may provide global public health researchers with novel platforms for investigating severe children malnutrition, uncovering new mechanistic insights through metabolic network that may lead to meaningful management strategy and cure.

Key messages.

1. Child malnutrition (CM) is a life threating global public health problem

2. Outcome from current management strategies for malnutrition are exceedingly poor

3. Emerging metabolomics and lipidomics approaches based on Mass Spectrometry techniques recently offer an exciting opportunity to understand the etiology and management of severe CM.