Metabolomics in Endocrinology: The Way Forward

Abstract

Metabolomics is a type of laboratory science used to understand the cellular and metabolic defects in any disease process. It comprehensively identifies endogenous and exogenous low-molecular-weight (<1 kDa) molecules or metabolites in a high-throughput manner. Mass spectrometry-based methods are used for metabolomics which can be targeted and non-targeted. Metabolomics workflow consists of sample acquisition, its preparation and extraction, separation, ionisation, data analysis, and metabolite detection and identification. Some of the commonly elevated metabolomes are branched-chain amino acids like isoleucine, leucine, and valine in diabetes, carnitine and glutamate in osteoporosis, deoxycholic acid and betahydroxybutyrate in pituitary tumours, glutamic acid, valine, isoleucine for malignant thyroid nodules, L-asparagine, L-glutamine, dimethylarginine for hyperparathyroidism, tetrahydro-11-doxycortisol for adrenal tumours, and oxidised glutathione for hypogonadism. Knowing metabolomics can help us formulate personalised treatment choices for precision medicine in endocrinology. The main challenge ahead of metabolomics is its technical complexity and cost-benefit issues.

INTRODUCTION

Endocrinology is heavily dependent upon laboratory evaluations and techniques. The accuracy of such measurements is very important for proper diagnosis and treatment. The omics sciences include proteomics, transcriptomics, and metabolomics. Their application to understanding the cellular metabolic defects in endocrinology has been amplified for the past decade. They have given unique insights into the pathology of an endocrinological state such as tumorigenesis, osteoporosis, and diabetes.[1]

Metabolomics is an upcoming diagnostic laboratory science that aims to detect small molecules in blood, urine, and tissue samples. The term “metabolome” was first used in 1998, and “metabolomics” was coined in 1999 to quantify the dynamic multiparametric metabolic response of living systems.[2,3] It comprehensively identifies endogenous and exogenous low-molecular-weight (<1 kDa) molecules or metabolites in a high-throughput manner. The composition of endogenous compounds is affected by proteome and genome as well as environmental and lifestyle factors, medication, and underlying disease.[4] Intermediate metabolites reflect the functions of the cells and organisms. Diet, medicines, exercise, gut microbiome, gender, and age affect their concentrations.[5] The metabolome is the final downstream product, and it can reflect the interactions between genes, proteins, and the environment.[6] It represents the molecular signature of a particular phenotype.[7]

Chromatographic techniques in metabolomics

Liquid or gas chromatography mass spectrometer-based assays can detect multiple analytes from a single aliquot. They are the most common methods used for metabolome measurement. Other methods, such as nuclear magnetic resonance (NMR) and capillary electrophoresis (CE), have also been used for metabolomics measurement.[8]

Basic types of metabolomics

Metabolomics platforms are assessed in two types: “untargeted-discovery-global” and “targeted-validation-tandem.” Untargeted discovery metabolomics tests a particular hypothesis, scans the full genome, and identifies the pattern of the metabolic alterations particular to that phenotype. Targeted metabolomics is performed to validate the untargeted metabolomics. Non-targeted metabolomics applies three technologies: (i) non-targeted profiling, (ii) fingerprinting, and (iii) footprinting. Targeted metabolomics uses quantitative tandem/targeted analysis and diagnostic analysis of a known clinically associated compound/biomarker.[9]

Workflow of metabolomics assays

Sample acquisition

Biological samples, including tissue, biofluids (blood, urine, feces, seminal fluid, saliva, bile, cerebrospinal fluid), and cell culture, are the usual samples for metabolomic analysis. Proper sample preparation and labelling are pivotal for optimum, reproducible, and high-throughput analysis.

Sample preparation and extraction

Optimised methanol–water chloroform combinations are used to extract hydrophilic and hydrophobic compounds during sample preparation. After centrifugation, a biphasic mixture of the upper (aqueous) and lower (organic) layers is separated. Aqueous extraction is used for polar organic solvents (e.g., methanol or acetonitrile), and organic extraction is carried out for non-polar solvents (lipids).[9]

Sample separation

Along with liquid chromatography (LC-MS) and gas chromatography (GC-MS), direct infusion mass spectrometry, direct analysis in real-time MS (DARTMS), capillary electrophoresis, and hydrophilic interaction chromatography (HILIC) are the usual sample separation techniques employed for polar metabolites.[10] Reversed-phase LC using C18 columns is used for separating the non-polar metabolite.[11] GC-MS is preferable for analysing less polar biomolecules like eicosanoids, esters, carotenoids, flavonoids, and lipids compared to LC-MS, which is preferred for separating alkaloids, amino acids, catecholamines, fatty acids, phenolics, prostaglandins, and steroids.[12]

Ionisation and detection

The ionisation procedure involves heating and evaporation, followed by charge transfer to the analytes, which ionises them in both the positive and negative modes. Mass analysers, including time of flight (TOF), quadrupole time of flight (QTOF), orbitrap ion trap, or triple quadrupole (QQQ) with multiple reaction monitoring (MRM), are used for ionised analyte detection with high sensitivity and accuracy.[13]

Data analysis and metabolite identification

Large amounts of complex raw data involving specific metabolic signals are extracted from mass spectrometry, followed by analysis in specialised software to interpret the data and identify the metabolite of interest. Peak selection, assessment, and relative quantitation were done in commercially available free software bioinformatic analysis tools. eXtensible Computational Mass Spectrometry, Metaboanalyst, Progenesis, MetaCore, and 3 Omics, with different analysis capabilities, are the different tools for metabolome data analysis.[14] Compound-specific databases are available in PubChem, Chemical Entities of Biological Interest (ChEBI), ChemSpider, The KEGG GLYCAN, Kyoto Encyclopedia of Genes and Genomes (KEGG), Model SEED, and MetaCyc.[15]

Metabolomics versus metabonomics

Metabonomics is a subset of metabolomics and defined as the quantitative measurement of the multiparametric metabolic responses of living systems to pathophysiological stimuli. This metabolic response can be against genetic modifications or environmental and nutritional stress. Metabonomics usually include intracellular molecules along with extracellular biofluids. The novelty of metabonomics lies in the precise and quantitative measurements carried out rapidly, enabling the temporal evolution of physiological states to be monitored. Metabonomics has been integrated with proteomics and genomics in recent times. This will create a database of intense genome, proteome, and metabolome interrelationships.[16]

Application of metabolomics in endocrinology

Role of metabolomics in diabetes and metabolic syndrome diagnosis

Diabetes remains the most common metabolic disease in the world, with the number of patients living with diabetes expected to rise to 780 million by 2045.[17] Early diagnosis and judicious management are key to preventing microvascular and macrovascular complications. A metabolomic study can help immensely in this regard.

Metabolomic alteration in type 1 diabetes mellitus (T1DM)

T1DM, a chronic autoimmune disease, is caused by a decrease in pancreatic β-cells, leading to decreased insulin production and severe hyperglycaemia. Odd-chain triglycerides and phospholipids containing polyunsaturated fatty acids were higher in autoantibody-positive children with T1DM compared to healthy children. Children with T1DM who developed autoantibodies before age 2 had methionine concentrations two times lower than those who developed autoantibodies in later childhood autoantibody-negative. It was proposed that the methionine pathway is involved in autoantibody production.[18] The levels of lysophosphatidylcholine and methionine decrease, and the level of ceramides increases before the onset of T1DM.[19]

Metabolomic alteration in type 2 diabetes mellitus (T2DM)

Both amino acid metabolism and phospholipid metabolism are altered in T2DM. The classical metabolomic variations in type 2 diabetes patients are increased alanine, tyrosine, glutamate, phenylalanine, methionine, and lysine. Increased glycine and glutamine have been shown to decrease the risk of diabetes.[20] Ahola-Olli AV et al.[21] conducted a large study comparing 244 diabetes cases and controls and found isoleucine, leucine, and valine to be the most significant metabolite changes in 10 years of follow-up. Various other studies have also elucidated the associations of branched-chain amino acids (BCAA) and diabetes.[21] These metabolic alterations precede the onset of diabetes for an average of 10 years. Retinol, pantothenate, and bile acid plasma levels have also been shown to correlate with diabetes onset, while low glycine levels predict insulin resistance. Palmitic acid, linoleic acid, and 2-hydroxybutyric acid can play a significant role in diagnosing T2DM.[22] These metabolomes were derived from Guasch-Ferré M et al.[23] study, where they used gas chromatography/time-of-flight mass spectrometry (GC´GC-TOFMS) combined with pattern recognition to analyse the plasma of diabetic patients and normal people in a control experiment. N-acetylaspartic acid, C20:0 lysophosphatidylethanolamine, and C16:1 sphingomyelin were associated with lower risks of T2DM. N-acetylaspartic acid is a predominant constituent in walnuts, and such consumption has been shown to prevent T2DM in the landmark PREDIMED trial.

Metabolomic alteration in Gestational Diabetes Mellitus (GDM)

GDM has significant adverse effects on pregnant women and fetuses. Nowadays, early diagnosis of GDM is advocated, and metabolomic studies can aid in this regard. Sun et al.[24] found statistically significant differences in the concentrations of 41 metabolites between pregnant women with GDM and those without. The differences were observed mainly in the lysine degradation pathway and aminoacyl transfer RNA biosynthesis pathway. Interestingly, such metabolomic differences are mostly observed in the first trimester of pregnancy.

Role of metabolomics in osteoporosis diagnosis

Osteoporotic fractures are major health-care related issues in post-menopausal women, and significant health care-related costs and resources are directed towards fracture treatment.[25] In today’s clinical endocrinology practice, we focus on osteoporosis prevention in post-menopausal women and men and women with other certain risk factors for osteoporosis.[26] The usual risk factors for secondary osteoporosis are long-term steroid intake, malnutrition, malabsorption diseases like celiac disease, chronic renal or liver disease, previous chemo or radiation therapy, diabetes mellitus, long-term hypogonadism, and smoking.[27] The study of metabolomics is now emerging as an effective tool to detect osteoporosis at the earliest stages. The amino acid lysine was found to be a discriminating metabolite between osteoporosis and normal bone mass.[28] Similar findings were noted in one of the largest Chinese series encompassing 364 participants.[29] In a matched incident case control study, lysine had the strongest association with increased fractures. Similarly, tryptophan, threonine, and phenylalanine had significantly high associations with fracture rates among essential amino acids.[30] Carnitine and glutamate alterations have also been found to correlate with osteoporosis.[31]

Variations in sphingomyelin levels have been seen to correlate with the abnormal proliferation of bone marrow-derived mesenchymal stem cells. Sphingomyelin levels are inversely correlated with bone mineral density, and alterations in the sphingomyelinase gene can cause recurrent fragility fractures and low bone mass.[32] Hydroxybutyric acid, glycerophospholipids, phosphatidylcholine, hyocholic acids, taurine, glycerophospholipids, and glycine levels have found to be altered in OP.[33]

Role of metabolomics in pituitary tumour diagnosis

Pituitary adenoma is often detected late due to its indolent clinical features, especially the non-functional pituitary adenomas (NFPA). Acromegaly also has a long time between onset of symptoms and the actual biochemical and imaging diagnosis. They can present with throbbing headaches along with visual field defects.[34] The metabolism of glucose (particularly the downregulation of the pentose phosphate pathway), linoleic acid, sphingolipids, glycerophospholipids, arginine/proline, and taurine/hypotaurine is mostly altered in acromegaly.[35] There are deoxycholic, 4-pyridoxic acids and 3-methyladipate alterations in a small cohort of ACTH-secreting Cushing’s diseases patients by Oklu et al.[36] Another study had documented short-chain fatty acids, including heptanoic acid, octanoic acid, nonanoic acid, and hexanoic acid upregulation along with downregulation of glucose-6-phosphate in pituitary tumours.[37] Ijare et al.[38] performed a metabolic analysis of NFPA and prolactinomas and found out prolactinomas showed elevation in beta-hydroxybutyrate (BHB) levels. Alanine, tyrosine, and formate elevations were seen in NFPA’s.

Role of metabolomics in thyroid malignancy diagnosis

Thyroid carcinoma is the most common endocrine cancer in the world, with a total number of cases expected to reach around 1 million in 2050.[39] The most common histological subtype is papillary thyroid carcinoma (PTC), which is followed by other varieties such as follicular carcinoma of thyroid (FTC), medullary carcinoma of thyroid (MTC), and anaplastic ones.[40] A thyroid gland ultrasound followed by fine needle aspiration cytology (FNAC) is the usual diagnostic algorithm for any thyroid nodule suspicious of malignancy. FNAC fails to differentiate a malignant tumour from a benign tumour in about 15–30% of cases.[41] Repeating FNAC frustrates the patient and wastes health resources. By integrating metabolic profiling techniques, it is now possible to detect malignancy in such equivocal cases and bypass the need for subsequent interventional techniques.[42]

The researchers have mostly focussed on NMR techniques in non-targeted ways to detect thyroid cancers at the earliest stages. The usual metabolic alterations in thyroid cancers are increased lactate content and amino acids, including glutamic acid, valine, and isoleucine, and decreased choline, myoinositol, scyllo-inositol, and lipid metabolism.[43,44] Elevated amounts of choline were found to be a consistent factor in malignant lesions compared to benign ones.[45,46] Accelerated alanine, aspartate, glutamate, and inositol phosphate metabolism were found in metastatic cancers. PTC downregulated carbohydrate metabolism, whereas lipid metabolism was upregulated.[47] Lower gluconic acid and higher citric acid levels were efficient metabolic pointers toward PTC compared to FTC [Table 1].[48]

Table 1.

Main metabolomes of interest in various diseases

Disease	Metabolome upregulated	Metabolome downregulated
Diabetes and metabolic syndrome	• Isoleucine, Leucine and Valine	• Glycine
	• Alanine, Tri-/Diacylglycerol-Fragments	• N-Acetylaspartic acid
	• Short-chain acylcarnitines	• C20:0 Lysophosphatidylethanolamine
	• Phosphatidylethanolamines	• C16:1 Sphingomyelin
Osteoporosis	• Lysine
	• Carnitine and glutamate
	• Sphingomyelin
	• Hyocholic acids
Pituitary tumours
Cushing’s disease	• Deoxycholic	• Glucose-6-Phosphate
	• 4-Pyridoxic acids
	• 3-Methyladipate
Prolactinoma	• Betahydroxybutyrate (BHB)
NFPA	• Alanine
	• Tyrosine
	• Formate
Thyroid malignancies
Malignant thyroid nodules	• Glutamic acid	• Choline
	• Valine	• Myoinositol
	• Isoleucine	• Scyllo-Inositol
PTC	Citric acid	• Long-chain fatty acids • Galactose metabolites • Gluconic acid
MTC	Glutamine and glutamate	3-Hydroxybutyric acid
Hyperparathyroidism	• L-asparagine	• Fatty acyl carnitine
	• L-glutamine
	• L-histidine
	• L-cysteine
	• Gamma-glutamyl amino acids
	• Dimethylarginine (ADMA)
Adrenal tumour and adrenocortical carcinoma	• Tetrahydro-11-doxycortisol (THS)
	• Creatine riboside
	• Choline-containing compounds
Pheochromocytoma and paraganglioma	• Lactic acid
	• Acetic acid and succinate acid
	• Tyrosine
Hypogonadism	• Oxidized glutathione (GSSG)	• Triacylglycerols
		• Sphingomyelin
		• Phosphatidylethanolamine
		• Lysophosphatidylethanolamine
		• Acyl-carnitine

Ultra-high-performance liquid chromatography coupled with triple quadrupole mass spectrometry (UHPLC/QqQ/MS) techniques have found that alteration of galactose metabolism can influence the PTC’s expression and metastasis.[49] Changes in linoleic acid, phenylalanine, arachidonic acid, glycine, D-glutamine, and D-glutamate can also affect tumorigenesis. Cyclohexanone, 4-hydroxybutyric acid, phenol, 2,2-dimethyldecane, and ethylhexanol level variations can explain the therapeutic response in PTCs.

Variations in glutamine and glutamate levels are the usual offenders in the MTC pathogenesis. Linear ion trap quadrupole (LTQ) orbitrap mass spectrometry has revealed such alteration in Yao et al.[50] from 140 study samples. The malignant MTCs can be differentiated from the benign ones by virtue of 3-hydroxybutyric acid levels.

Role of metabolomics in hyperparathyroidism diagnosis

Marta Wielogórska-Partyka et al.[51] evaluated untargeted metabolomics in their study of primary hyperparathyroidism patients and compared it to control groups. The authors noticed a significant increase in L-asparagine (22.1%), L-glutamine (15.5%), L-histidine (6.5%), and L-cysteine (38.3%) in patients with hypercalcemia due to primary hyperparathyroidism compared to controls. These amino acids can modulate the calcium-sensing receptors (CaSR) activity.[52] CaSR is a type of G protein-coupled receptor that modulates serum calcium levels. Higher levels of L-amino acids are observed in the gastric lumen, which induces gastrin secretion followed by CaSR activation.[53] L-histidine is converted to L-histamine by L-histidine decarboxylase, and L-histamine is one of the primary mediators of hydrochloric acid secretion from the gastric parietal cells. This mechanism explains the potential link between primary hyperparathyroidism and peptic ulcers.[54] Primary hyperparathyroidism is associated with altered levels of gamma-glutamyl transferase (GGT), which metabolises gamma-glutamyl compounds, including glutathione, an important antioxidant. Gamma-glutamyl amino acids were found to be elevated in PHPT cases, and gamma-glutamyl residue attached to a glycine molecule showed the highest percentage of change in a metabolomic study conducted by Marta Wielogórska-Partyka et al.[51,55]

Dimethylarginine (ADMA), a subtype of dimethyl-l-arginine (DMA), is indirectly associated with calcium metabolism through endothelial function and nitric oxide production. ADMA inhibits nitric oxide synthase and leads to impaired endothelial function, which results in cardiovascular and metabolic dysfunctions.[56] Increased levels of ADMA have been found in PHPT patients and can explain the pathogenic associations between PHPT and cardiovascular and metabolic dysfunctions.[57] Interestingly, ADMA levels fail to come down to normal after parathyroidectomy, which can explain the irreversible nature of cardiovascular disease found in PHPT.[51] In summary, increased L-amino acids, gamma-glutamyl amino acids, and ADMA are associated with peptic ulcer disease, osteoporosis, and cardiovascular manifestations, respectively.

Additionally, the researchers have found significant reductions in the levels of fatty acyl carnitine, which can be associated with mitochondrial dysfunction and fragility fractures. L carnitine is involved in increasing kinase activity and controlling osteoblast differentiation, upregulating the expression of osteogenic-related genes, such as Runt-related transcription factor 2 (RUNX2), osterix (OSX), bone sialoprotein (BSP), and osteopontin (OPN).[58]

Role of metabolomics in adrenal tumour management

Glucocorticoids, mineralocorticoids, sex steroids, and catecholamines regulate carbohydrate, protein, and fat metabolisms intricately. Adrenocortical carcinoma (ACC) is the most common malignancy of adrenal gland, and its survival rate beyond 5 years is very low.[59] The most difficult part of our clinical and histological endocrinology is differentiating between benign adrenal adenoma (atypical type) and malignant ACC. The Weiss scoring, which recognises three or more morphological patterns for diagnosing malignancy, cannot effectively differentiate between benign and malignant in borderline cases.[60] LC-MS/MS-based metabolomics techniques have been used extensively in the past decade to differentiate between the two.

Arlt et al.[61] found tetrahydro-11-doxycortisol (THS) to be the most discriminating marker of malignant adrenal tumours in their study of 24-hour urinary steroid metabolomic profile. Creatine riboside levels were found to be significantly elevated in patients of ACC in the study of Patel et al.,[62] where they used unbiased ultra-performance liquid chromatography/mass spectrometry (UPLC/MS). Lactate, acetate, and total choline-containing compounds were found to be part of the discriminatory metabolome in the 1H-high-resolution magic-angle spinning nuclear magnetic resonance (HRMAS NMR) spectroscopy technique [Table 1].[63] One study used LC-MS/MS equipped with a flame ionisation detector (FID) to study 19 major steroids and their metabolites in 58 urine samples from patients with non-functioning adenomas. Cortisol, tetrahydrocorticosterone, tetrahydrocortisol, allotetrahydrocortisol, and etiocholanolone were found to be discriminatory biomarkers.[64]

Lactic acid, acetic acid, and succinate acid were found to be discriminatory biomarkers in diagnosing pheochromocytomas and paragangliomas. Gluconeogenesis, pyruvate metabolism, ammonia recycling, porphyrin, and aspartate metabolism were additionally altered.[65]

Role of metabolomics in hypogonadism management

Metabolomes of purine metabolism and amino acid metabolites triacylglycerols (TGs), diacylglycerol (DG), sphingomyelin (SM), phosphatidylethanolamine (PE), lyso-phosphatidylethanolamine (LPE), and phosphatidylcholines (PCs) were found to be the differentiating metabolomes in congenital hypogonadotropic hypogonadism (cHH).[66] Triglyceride metabolism is found to affect spermatogenesis and sperm quality in men. Grande G et al.[67] found low triacylglycerol and phosphatidylcholine in patients of cHH. Low concentrations of such metabolism affect the lipid peroxidation pathway and harm sperm function and quality. Sphingomyelin (SM) has been found to regulate sperm cell function and prevent immune-mediated destruction in the female genital tract. The pentose phosphate pathway (PPP), Krebs cycle, and β-alanine metabolism pathway are also involved in hypogonadism. Increase in oxidized glutathione (GSSG), reduction in acyl-carnitine, and downregulation of fatty acid β-oxidation were noted due to alteration of these pathways.[68] Testosterone replacement therapy in hypogonadism significantly increased the production of acetyl-CoA and triacylglycerols. Glycolysis is increased and gluconeogenesis is decreased with testosterone replacement therapy.

Challenges in metabolomics assay in endocrinology

Even with extensive databases, untargeted metabolomics suffers from identification dilemmas due to heterogeneous throughput. It needs comprehensive bioinformatics and computational techniques to identify all metabolomes and complete human metabolome data. Chromatographic resolution and compound identification can be affected by polarity, ion sources, ion suppression, flow rates, sample preparation strategies, purity and temperature of reagents and solvents, different laboratory staff, and techniques during the ionisation process. Isomers with identical masses and highly similar spectra can complicate distinguishing and differentiation during metabolite assignment.[69] Quality control (QC) is essential for optimum standardisation during calibration, sample preparation, and batch analysis. Each analysis should be carried out under optimum performance monitoring conditions.[70] Currently, metabolomic assays require liquid chromatography technology at this stage, which is not cost-effective. That is why further validation studies are required to establish their consistent association with the disease. Then, they can be used as biomarkers for early screening and diagnosis.

Metabolomics in endocrinology – the way forward

As metabolomics assay can detect the smallest molecules involved in the disease process, it can create a sufficiently large database to understand the disease process. This is most important as it can guide us in personalised treatment choices in precision medicine. As various pathophysiological markers are being discovered to unravel the disease process in metabolomics, it can help us to achieve more stringent treatment targets in various endocrinological diseases. This can be achieved in a better and more tolerant manner. However, further researches are warranted to specify the metabolomes more and make it a cost-effective diagnostic tool. Till now, its use as a biomarker or diagnostic tool to diagnose a disease process is quite premature and impetuous.

Author contributions

SP and DD conceptualized the study, searched all relevant articles and curated data, did formal analysis, and wrote the original draft. SK provided intellectual content in the manuscript. All the authors have read and approved the final version of the manuscript. As stated earlier in this document, the requirements for authorship have been met, and all authors confirm that the manuscript represents honest work.

Conflicts of interest

There are no conflicts of interest.

Use of artificial intelligence

The authors did not use artificial technology to write and prepare the manuscript.

Funding Statement

Nil.