2.1. Chronic Fatigue Syndrome (CFS)
CFS has been associated with errors in the immune system, central nervous system, autonomic nervous system, and energy metabolism, however, these findings are not consistent enough to provide satisfactory sensitivity and/or specificity for a diagnostic test for CFS. As such, the cause of CFS remains unknown [38
]. The condition has long been recognized, and has been described by many different terms (e.g., Da Costa's syndrome, effort syndrome, soldier's heart, neurasthenia, myalgic encephalitis /encephalomyelitis, Iceland disease, Akureyri disease, Royal Free disease, immune dysfunction syndrome) [39
], and most recently, systemic exertion intolerance disease [21
]. Clinical characteristics include at least six months of moderate to severe intensity symptoms at least 50% of the time chiefly dominated by fatigue. Criteria include persistent fatigue, post-exertional malaise, sleep problems, cognitive impairment, and orthostatic related symptoms amongst others [41
Recent metabolomics analysis of plasma from CFS subjects and healthy controls revealed significant variations of intermediate metabolites from the tricarboxylic acid and urea cycles [42
]. These findings allowed CFS patients to be discriminated from healthy controls based on variances in the ornithine/citrulline and pyruvate/isocitrate ratios (P < 0.0001 and P = 0.0069). Further plasma metabolic profiling of CFS subjects and healthy controls quantified 832 metabolites, which identified significantly different concentrations of 14 metabolites related to redox imbalances in CFS subjects [43
]. More recently, 361 plasma metabolites were compared between CFS patients and matched controls [44
]. 74 of these metabolites were originally found to be significantly altered (P < 0.05), but only 35 were significant after statistical correction. The 35 metabolites were associated with various biological pathways, including taurine metabolism, glyoxylate and dicarboxylate metabolism, purine and amino acid metabolism, and energy metabolism. Another analysis of plasma metabolites assessed 612 metabolites from 63 various biochemical pathways in CFS patients and controls using hydrophilic interaction liquid chromatography, electrospray ionization (ESI), and tandem mass spectrometry (MS/MS) [45
]. The majority of these metabolite concentrations were decreased in CFS patients compared to controls, but significant abnormalities were identified in 20 unique metabolic pathways. These include an increase in pyrroline-5-carboxylate and decreases in sphingolipids, phospholipids, purines, cholesterol, and microbiome amino acids, to name a few. This finding is consistent with CFS as a hypometabolic state, but the pathways linked to these abnormal metabolite concentrations vary. One pathway consistently affected in CFS subjects is mitochondrial metabolism. Since abnormal mitochondrial function is associated with fatigue, a recent study sequenced mitochondrial DNA (mtDNA) of 93 CFS patients and evaluated the sequences for mutations [47
]. No clinically proven mtDNA mutations were realized in the CFS patients, suggesting that mtDNA mutations are not a common element in the etiology of CFS. However, a more recent comparison of mtDNA mutations in CFS subjects and controls identified a significant difference in the percentage of CFS patients lacking a deleterious variant of mtDNA compared to the control group [48
]. These findings are interesting, but the importance of these findings on the etiology and pathogenesis of CFS necessitates further study.
A 2003 systematic review analyzed 58 articles that were both directly concerned with CFS and had sample sizes larger than 10. After sorting and reviewing the articles according to quality of methodology, no consistent pattern of immunological abnormalities was identified for CFS patients [49
Urine samples from CFS children and controls were collected before and after exercise to identify possible variances in metabolic profiles between the groups [50
]. Pre and post-exercise metabolites were assessed using liquid chromatography-mass spectrometry (LC-MS), Principal Component Analysis (PCA), and orthogonal projections to latent structures-discriminant analysis. Prior to exercise, the CFS group was found to have significantly decreased creatine, indole-acetaldehyde, phytosphingosine, and pyroglutamic acid compared to the control group. Eleven metabolites (nonanedioic acid, methyladenosine, acetylcarnitine, capric acid, corticosterone, creatine, levonorgestrel, pantothenic acid, pyroglutamic acid, xanthosine, and xanthurenic acid) were selected for analysis after exercise. The CFS group demonstrated significantly increased methyladenosine and creatine (P < 0.05), while the other metabolites were significantly decreased (P < 0.05) compared to the control group. The MetPA database was then used to associate the metabolites to their respective metabolic pathways. Prior to exercise, the CFS group exhibited a disordered arginine-proline metabolism pathway. Furthermore, three disordered metabolic pathways (marker metabolites) were identified in the CFS group post-exercise, which were arginine-proline metabolism (creatine), biosynthesis of pantothenic acid and CoA (pantothenic acid), and steroid hormone biosynthesis (corticosterone).
Analysis of peripheral blood mononuclear cells (PBMCs) from 52 CFS patients and 35 healthy controls to assess for consistent cellular patterns of oxidative phosphorylation (OXPHOS) and glycolysis of CFS patients. The data revealed consistently decreased markers of OXPHOS in PBMCs taken from CFS patients compared with healthy controls [51
]. These markers result from various pathways, such as ATP production and non-mitochondrial respiration. Of note, the markers that best differentiated CFS patients from controls are associated with maximal respiration. Maximal respiration was consistently impaired in CFS patients (P ≤ 0.003). This lower maximal respiration implies that PBMCs in CFS patients cannot respond to extracellular stressors as effectively as healthy PBMCs and, therefore, cannot meet cellular energy needs.
Recently, a reproducible, broad-spectrum metabolomics technique was developed that allows for the simultaneous analysis of over 600 plasma metabolites from many biochemical pathways via a single injection [53
]. This technique shows promise as a diagnostic tool for CFS and other multifaceted diseases.
Ventricular cerebrospinal lactate levels were compared in patients with CFS patients, Fibromyalgia (FM) patients, patients with overlapping CFS/FM, and controls using proton nuclear magnetic resonance spectroscopy (1HNMR) [54
]. Although the mean lactate levels did not significantly differ between diseased groups, all three groups had markedly higher lactate levels compared to the healthy volunteers. As such, lactate levels in cerebrospinal fluid (CSF) cannot be used to distinguish FM patients from CFS patients. Contrarily, the increased lactate concentration in patients with central sensitivity syndromes requires further study to determine what role this metabolite has in the various disease states.
Analysis of fecal and plasma samples from 24 CFS/IBS patients, 26 CFS patients, 1 IBS patient, and 49 controls identified decreased levels of choline, carnitine, and phosphatidylcholine in CFS patients. Additionally, increased levels of phosphatidylethanolmines, triglycerides, and ceramides were identified in CFS patients with IBS comorbidity [55
]. Combined analysis of fecal metagenomic and plasma metabolomic data provides a more reliable predictive model for CFS than either data sets alone.
This term describes a non-cancerous, hormone-dependent, inflammatory disease that can affect females of all ages and hormonal stages, although it typically presents in reproductive years. It involves the development of endometrial tissue outside the uterus. This ectopic endometrial tissue typically develops in the pelvis, but can also appear in the bowel, bladder, and other anatomical locations. The presence of this tissue in abnormal areas can cause inflammation, dysmenorrhea, chronic pain, infertility, and dyspareunia [62
Recently, the metabolic composition of follicular fluid (FF) was analyzed in women undergoing fertility treatment for tubal diseases (n = 10), unexplained causes of infertility (n = 13), male infertility (n = 10), endometriosis (n = 8), or PCOS (n = 12) was analyzed using NMR [65
]. The FF profiles of male infertility, tubal disease, and unexplained cause subjects were not distinguishable by metabolic profiles. Contrarily, the FF profiles of subjects with endometriosis and PCOS could be reliable differentiated from the FF of male infertility subjects due to significantly altered levels of metabolites associated with infertility. A second study compared the metabolic profiles of FF from 17 endometriosis subjects to FF from 16 unaffected subjects via sequential window acquisition of all theoretical fragment-ion spectra (SWATHTM) [66
]. In the subjects with endometriosis, significantly different concentrations of phytosphingosine and two lysophosphatidylcholine isomers were identified. These metabolites may contribute to infertility in patients with endometriosis due to their role in essential cellular processes. Another study compared FF metabolites collected from 12 women with ovarian endometriosis to FF from 12 matched controls using NMR [67
]. Analysis revealed that endometriosis subjects could be consistently distinguished from controls due to significantly elevated levels of lactate, β-glucose, pyruvate, and valine in the FF. A fourth study utilized 1HNMR and biochemical analyses to develop a metabolic profile of FF from women with stage I-II endometriosis (n = 8), stage III-IV endometriosis (n = 8), and unaffected controls (n = 7) using [68
]. Metabolite concentrations from glycolysis, lipid, and amino acid pathways were found to be significantly altered in both endometriosis groups compared to the control group.
MS/MS was recently used to assess metabolic variation in FF and endometrial tissue from women with stage III-IV endometriosis (n = 10) and healthy controls (n = 10) [69
]. Significantly increased concentrations of sphingolipids and phosphatidylcholines were found in both the FF and the endometrial tissue of endometriosis patients compared to control samples. The variation between the metabolomic profile of endometrial tissue located in normal anatomical positions of 29 women with minimal to mild endometriosis patients was compared to similar tissue in 37 healthy women using ultra-high-performance liquid chromatography (UHPLC), high resolution mass spectrometry (HRMS), and ESI [70
]. Statistical analyses of the differentiating metabolites revealed that levels of uric acid, hypoxanthine, and lysophosphatidylethanolamine could be used to diagnose minimal to mild endometriosis with 66.7% sensitivity and 90.0% specificity. Further analysis of eutopic endometrial tissue via 1
H-NMR revealed differences in the concentration of metabolites associated with energy, ketogenesis, and gluconeogenesis consistent with various stages of endometriosis [72
]. Alanine concentration was found to diagnose Stage 1 endometriosis with 90% sensitivity and 58% specificity. Construction of a regression model of the serum concentrations of alanine, leucine, lysine, proline, and phenylalanine allowed for the diagnosis of Stage II endometriosis with 100% sensitivity and 83% specificity. A third study compared the metabolic profile of eutopic endometrial tissue between 21 patients with either stage I or stage II endometriosis with eutopic endometrial tissue from 20 unaffected women using UHPLC-ESI-HRMS [73
]. Analyses revealed significantly decreased concentrations of various phosphatidylcholine, (18:1/22:6), (20:1/14:1), (20:3/20:4), and phosphatidyl-serine (20:3/23:1). Analyses also revealed increased levels of phosphatidic acid (25:5/22:6). Using these metabolites, early endometriosis could be predicted with 90.5% sensitivity and 75.0% specificity.
Metabolic profiles of peripheral blood samples from laparoscopically confirmed endometriosis subjects (n = 25) and controls (n = 19) were generated using UPLC-ESI-Quadrupole (Q)-TOF techniques [74
]. Various lipid, fatty acid, signaling, and cholesterol metabolites were investigated; however, a group of acylcarnitines was identified that consistently and objectively distinguished endometriosis subjects from controls. Unfortunately, the curve analysis outcome was insignificant due to data limitations.
The metabolic profiles of plasma from early-stage endometriosis subjects (n = 6), advanced stage endometriosis subjects (n = 44), and healthy controls (n = 23) were analyzed using 1
H-NMR and statistical analyses [75
]. Metabolic profiles differed significantly between endometriosis patients and controls with subjects from both endometriosis groups having altered metabolites relating to creatine metabolism, branched chain amino acids (BCAA) metabolism, cellular needs, and the syntheses of cell membranes. A second quantification of plasma metabolites in endometriosis patients and controls was conducted using ESI-MS/MS [76
]. Over 140 targeted analytes were detected using this technique, including glycerophospholipids, sphingolipids, and acylcarnitines. Eight lipid metabolites were identified at significantly elevated levels in the endometriosis group. A BMI/age-adjusted model was developed using the identified metabolites that allowed for differentiation of endometriosis subjects from control subjects with 90.0% sensitivity and 84.3% specificity.
Serum from women with stage II-III endometriosis (n = 31) and healthy women (n = 15) was analyzed for predictive metabolites using 1HNMR and pattern recognition. Following metabolite detection, computational methods were utilized to develop a diagnostic model [77
]. Results showed significantly altered metabolites from steroid hormone biosynthesis, primary bile acid biosynthesis, biotin metabolism, and taurine/hypotaurine metabolisms in endometriosis subjects compared to control subjects. Positive and negative predictive values for these groups were 71% and 78%, respectively. A second 1
H-NMR metabolic analysis of serum from laparoscopically confirmed endometriosis subjects (n = 75) compared to healthy controls (n = 60) was performed [78
]. Analyses revealed significantly altered concentrations of metabolites associated with glucose metabolism, the Krebs cycle, and mitochondrial deficiencies. These metabolites allowed for classification of endometriosis patients with 92.83% sensitivity and 100% specificity. Further 1
H-NMR analysis of serum from women with stage I-II endometriosis (n = 22) and age/BMI matched healthy women (n = 23) demonstrated altered metabolite concentrations that correspond to the immune response to endometriosis, increased aerobic glycolysis, faulty pyruvate metabolism, and altered amino acid composition [79
]. These metabolite variances allowed endometriosis patients to be discriminated from controls with a sensitivity of 81.8% and a specificity of 91.3%. In a recent case-control study, the serum metabolomic profile was developed in women with stage III-IV endometriosis (n = 50) and controls (n = 50) using MS to test the diagnostic capabilities of serum metabolomics [80
]. Using this profile, 10 metabolites were identified based on their significance to the predictive model. The metabolites allowed endometriosis subjects to be distinguished from controls with 84% accuracy.
Peritoneal fluid (PF) from women with endometriosis of the ovaries (n = 29) was compared to PF from healthy women (n = 36) and assessed for invasive metabolites [81
]. Of the 148 metabolites quantified, 10 metabolites were found to have significantly decreased concentrations and 125 metabolite ratios were found to be altered in the endometriosis group. The affected metabolites include various forms of carnitine, acylcarnitines, sphingomyelins, and phosphatidylcholines. After adjusting for age, two of the altered metabolite ratios could differentiate endometriosis subjects from controls with 82.8% sensitivity, 94.4% specificity, and an area under the curve (AUC) of 0.944.
H-NMR analysis of metabolites in the urine of women with early-stage endometriosis (n = 6), advanced stage endometriosis (n = 39), and healthy controls (n = 36) was performed [82
]. After statistical analyses, women with endometriosis had consistently altered concentrations of metabolites associated with protein catabolism, inflammatory processes, oxidative stress, and immune response, compared to healthy women.
2.4. Fibromyalgia (FM)
FM is a significant member of a class of disorders termed central sensitivity syndromes and overlapping chronic pain conditions [83
]. These conditions have shown significant diagnostic and therapeutic challenges to date. Due to the lack of diagnostic methods for FM, the condition goes undiagnosed in up to 75% of individuals resulting in postponed care [84
]. Similarly, the lack of effective treatment options results in substandard care and poor management of patient symptoms. Research suggests that the syndrome arises due to aberrant neuroendocrine function, increased sensitivity to the body’s threat response, and/or dysfunction in the body’s pain receptors [88
]. FM presents as general pain and tenderness in consistent locations throughout the body, sleep disturbances relating to this tenderness, and assorted comorbidities [84
]. Fibromyalgia disproportionately affects women and is the leading cause of chronic widespread pain in the United States [85
]. Until a reliable biomarker can be identified, the diagnosis of FM will require in-depth clinical evaluation. The difficult and ambiguous identification of FM persists despite the ability to eliminate other potential causes of pervasive pain through laboratory testing, such as vitamin D deficiency and hypothyroidism.
A persistent effort in FM research is the search for unambiguous and quantifiable biomarkers that may allow for the identification of at-risk individuals, assist in FM diagnosis, and facilitate more dependable and beneficial treatment of FM patients. Previous studies have shown that metabolomics are especially useful in the study of the brain due to the key role of metabolites in neurological signaling. Similarly, recent studies have reported on the metabolomic disruptions in the brains of fibromyalgia patients [93
]. These studies suggest that discrepancies in the metabolic profile of the brain may contribute to neurological syndromes. Recently, the importance of subcategorizing FM patients according to symptoms was realized [94
]. These findings demonstrate that FM presents in distinct forms which future studies may distinguish as biochemically distinct. Furthermore, the differentiation of FM types may allow for improved evaluation and care for affected patients.
The comprehensive metabolite profile of urine from subjects with FM (n = 18), first generation family members of the FM subjects (n = 11), age-matched healthy controls (n = 10), and healthy 18–22 year-old subjects (n = 20) was recently analyzed using NMR [96
]. The analyses revealed increased concentrations of metabolites related to the gut microbiome, which allowed for successful differentiation of FM individuals from all three control groups with an AUC of 90%. Further NMR analysis of the urine of fibromyalgia patients and matched controls found significantly altered metabolites in the urine of fibromyalgia patients suggesting significant muscle damage compared to controls [36
]. Further analysis of urinary metabolites in subjects with persistent FM (n = 18), first generation family members of the FM subjects (n = 11), age-matched healthy controls (n = 10), and healthy 18–22 year-old volunteers (n = 41) [97
] via GC-MS was executed. This analysis identified 196 metabolites in the urine; however, the FM subjects could be distinguished from the control groups using 14 significantly increased metabolites that are broadly associated with energy metabolism, carbohydrate degradation and utilization, and host metabolites.
The metabolic profile of plasma from women with FM (n = 22) and healthy women (n = 21) was analyzed using liquid chromatography-Q-TOF/MS) [98
]. Many lipid compounds were identified, but the metabolites responsible for discriminating the FM subjects from controls were 1-tetradecanoyl-sn-glycero-3-phosphocholine and 1-hexadecanoyl-sn-glycero-3-phosphocholine. Recent evidence suggests that these metabolites may play a critical role in pain signaling and sensitivity.
Anti-dense fine speckled 70 (anti-DFS70) antibodies were collected from the blood of FM subjects (n = 39), subjects with systemic lupus erythematosus (SLE) (n = 17), and healthy subjects (n = 19) [99
]. These antibodies were subsequently quantified via enzyme-linked immunosorbent assay. The FM and healthy groups had higher levels of anti-DFS70 antibodies compared to the SLE group. Additionally, FM subjects that had either self-endorsed arthralgia or sleep disturbances were found to have significantly higher anti-DFS70 antibody values than those without arthralgia or sleep disturbance.
The first study of metabolomics in FM examined blood from 14 FM patients, 15 osteoarthritis (OA) patients, and 12 rheumatoid arthritis (RA) patients [100
] using mid-infrared microspectroscopy (IMS). This analysis allowed for differentiation of RA, OA, and FM patients without error. This finding demonstrates that FM spectral patterns can be consistently separated from other diseases using IMS. Further analysis of group metabolic profiles found that the profiles of the RA and OA groups were similar, while significant differences in metabolites associated with tryptophan catabolism were identified in the FM group.
More recently, vibrational spectroscopy was performed on blood from various disease groups to differentiate FM patients from RA, OA, and SLE patients (n = 50, n = 29, n = 19, n = 23, respectively) [35
]. Distinctive spectral signatures were identified via pattern recognition analysis and patients were organized into disease classes without error. UHPLC, MS/MS, and photodiode array revealed that protein backbones and pyridine-carboxylic acids were significant metabolites in disease discrimination. Furthermore, spectra from FM patients correlated with reported patient reported pain severity. This suggests that these techniques may be capable of differentiating disease severity using biological samples and patient reports.
2.7. Painful Bladder Syndrome (PBS)/Chronic Prostatitis (CP)/Interstitial Cystitis (IC)
These are common chronic pain syndromes that have poorly understood pathophysiology [114
], but it is likely that the respective symptoms of these diseases result from the contribution of multiple factors; however, these factors are difficult to isolate because these disorders are often accompanied by other chronic pain conditions, such as FM. Recent epidemiological research suggests that 25–30 million women in the US alone may suffer from Bladder Pain Syndrome (BPS)/IC [115
] Another epidemiological study of US women found that around 3-6% of women meet diagnostic criteria for BPS/IC, but only about 10% of these women have been diagnosed with BPS/IC [116
]. In addition to the persistent pain, dysuria, and nocturnal disturbances reported by individuals with IC, relationships and work-life may be significantly strained. Furthermore, there are substantial financial impacts from IC, such as frequent physician visits, medications, and decreased productivity. Recent studies found discovered that the average yearly cost of IC ranges from $
3,631 to $
7,043, which is similar or more costly than other chronic pain disorders [117
The understanding of IC has been notably inhibited because reliable, diagnostic tests have not been developed and identified. The discovery of a consistent and specific reliable biomarker would modernize IC diagnosis, allow for more accurate prognosis, and enhance treatment capabilities. This finding would also serve as a novel contribution to the increasing literature addressing the etiology and pathogenesis of IC. Studies have identified possible markers in the urine, bladder tissue, and serum of IC patients [118
]. For example, significantly altered levels of antiproliferative factors and growth factors have been identified in the urine of IC patients compared to the urine of healthy controls that show promise as a diagnostic marker [119
]. Unfortunately, tests for these biomarkers are not currently available in clinical settings.
A 2012 review of metabolomic, genomic, proteomic, and other ‘omics’ techniques for IC highlighted a substantial lack of research on the use of ‘omics’ techniques to better understand IC’s pathogenesis, to aid in diagnosing and quantifying the severity of disease, and to allow for the development of more effective treatments for IC [120
]. A more recent review on the use of metabolomics in IC supports the aforementioned finding that the use of metabolites in the diagnosis, treatment, and research of IC is underutilized and endorses that the lack of biomarkers in IC is impeding the understanding of this disease, etiology, and possible responses to therapies [121
]. GC-MS analysis of urine from IC subjects and controls examined over 450 metabolites and 30 compounds were significantly altered in IC subjects [122
]. Unfortunately, the majority of these compounds are unknown, which suggests that the understanding, treatment, and diagnosis of IC may be improved by studying these unknown compounds.
Analysis of urine samples from IC patients (n = 10), bacterial cystitis patients (n = 10), and healthy controls (n = 10) was recently conducted using UPLC-MS-NMR-MS/MS to identify any metabolites that could be used to discriminate IC subjects [123
]. Results showed that phenylacetylglutamine (PAGN) was a significantly altered marker in the urine of IC subjects. Further quantitative methods showed altered ratios of PAGN/creatine ratios that could be used to differentiate IC patients from other patients. Further metabolic analysis of urine from IC patients (n = 43) and healthy controls (n = 20) identified 140 metabolites that were significantly different in the urine of IC patients compared to the urine of healthy controls, with a false discovery rate of only 5% [124
]. Based on significant correlation, 15 NMR spikes associated with metabolites in IC were identified as the most reliable markers of IC patients. These metabolites were found to be associated with cellular repair after oxidative stress and neuromodulation of pain. A third study utilized MS to assess the global metabolite profile of urine of patients with IC/bladder pain syndrome (BPS) (n = 40) compared to healthy controls (n = 40) [125
]. This analysis allowed for the identification of noteworthy metabolites that are capable of distinguishing IC/BPS subjects from controls. Specifically, etiocholan-3α-ol-17-one sulfate (Etio-S) could identify female IC/BPS subjects with 90% specificity and sensitivity. This metabolite is a reduced isomer of testosterone and its concentration was closely associated with patient reported symptoms. Because of this correlation to symptom scores, Etio-S could be used to separate IC/BPS patients based on severity of disease. Analysis of volatile metabolites in the urine from IC subjects (n = 10) and healthy controls (n = 10) was conducted using solid-phase microextraction-GC-TOF-MS [126
]. This analysis found that volatile metabolites had significantly reduced concentrations in the urine of IC patients compared to control urine. Furthermore, these metabolites were associated with inflammatory responses, which indicates they may be used to objectively assess disease severity.
A bioinformatics approach looked at the urine of IC subjects (n = 50), subjects with bacterial cystitis (n = 30), and healthy controls (n = 47) via MS and 1
H-NMR to assess for metabolic or protein profiles unique to IC patients [127
]. Using this approach, subjects could be discriminated into the respective groups with a success rate of 84%. The metabolic profile of urine from an animal model with nonbacterial prostatitis (NBP) (n = 8), a control group without NBP (n = 8), and a group of NBP animals after treatment with limonin (n = 8) was constructed using UHPLC-ESI-TOF, multivariate analysis (MVA), and pathway analysis [128
]. Results identified 20 possible biomarkers for NBP and found significantly altered metabolites from amino acid metabolism and glycerophospholipid metabolism in the NBP group that were normal in the group treated with limolin. This suggests that diagnosis and understanding of NBP may be improved via metabolomics.
The feasibility of using serum to diagnose IC in humans and domestic cats was assessed via infrared microspectroscopy [129
]. Data was obtained from healthy cats (n = 11), cats with feline IC (n = 22), healthy humans (n = 19), and humans with IC (n = 25). This data was used to develop a model using SIMCA. This study was able to successfully predict whether the subject was healthy or had IC with 100% sensitivity using this data.
2.12. Polycystic Ovary Syndrome (PCOS)
PCOS is the most common hormone disorder and cause of infertility in women across the world, affecting between 6.5–8% of all women [166
]. This disorder often presents during adolescence due to ovary dysfunction and typically has symptoms of hyper-androgenism, such as excessive hair growth, acne, and alopecia [168
A 2018 systematic review of various metabolites in the diagnosis and treatment in PCOS identified promising metabolites from multiple metabolic pathways. These findings support the use of metabolomics in PCOS research and reveal that metabolomics may allow for an improved understanding of the disease, its etiopathology, and improved treatment and diagnosis [169
Recently, the serum levels of two metabolites (ghrelin and leptin) were recently analyzed in PCOS subjects (n = 130) and healthy controls (n = 121) to assess for metabolic, hormonal, and biochemical variances between groups and between lean/obese subjects in these groups [170
]. No significant differences in ghrelin or leptin concentrations were identified between PCOS subjects and healthy controls; however, both of these metabolites were significantly altered in obese PCOS subjects compared to lean PCOS subjects.
Another study generated serum metabolic profiles from PCOS patients (n = 145) and controls (n = 687) to assess the impact of testosterone levels and obesity on metabolic variances [171
] via NMR association analyses. As expected, PCOS women exhibited more metabolic disturbances than controls and several very low-density lipoproteins were altered in this group. These variances in lipid metabolites persisted after adjusting for obesity measures. High-density lipoproteins, apolipoprotein A1, and albumin concentrations were significantly decreased in obese PCOs subjects compared to controls. Furthermore, testosterone levels inversely correlated to insulin levels in the PCOS group, but not in the control group.
H-NMR analysis of serum from PCOS subjects (n = 74) and controls (n = 68) identified 8 significantly altered amino acids and four altered metabolites of energy metabolism [172
]. Pathway analysis of these metabolic variances via MetPA implicate dysfunction in aminoacyl-tRNA biosynthesis, amino acid biosynthesis, and pyruvate metabolism, among others. Lactate, threonine, proline, acetate, and alanine were the most accurate predictors of PCOS. Overall, these findings demonstrate altered carbohydrate, lipid, and amino acid metabolism in PCOS subjects. Further analysis of the serum metabolic profile was conducted on PCOS women (n = 30) and age/BMI matched controls (n = 30) using GC-LC-MS [173
]. The subjects in the PCOS group displayed significantly altered metabolites from multiple disrupted metabolic pathways, such as AA synthesis, steroid hormone metabolism, and lipid metabolism. Specifically, PCOS subjects had significantly increased concentrations of phospholipids, organic acids, hormones, and aromatic AA in serum samples compared to the control group and significantly decreased cholesterol levels. Notably, total cholesterol, phenylalanine, uric acid, and lactic acid were among the metabolites used to differentiate PCOS subjects from the control subjects. Further analysis compared the serum metabolites of PCOS subjects (n = 10) and healthy controls (n = 10) via LC-MS to improve the understanding of this disease, its etiopathology, and treatment options [174
]. Six unique differentiating metabolites from lipid, androgen, carnitine, and bile acid metabolism pathways were identified. These differentiating metabolites had crucial and distinct roles in disease development.
In an interventional study, PCOS patients (n = 15) and healthy controls (n = 15) were treated with myo-inositol, D-chiro-inositol, and glucomannan over a three month period [175
]. The concentrations of various metabolites and blood components were measured in the serum before and after treatment. Comparing groups before and after treatment showed significantly altered concentrations of 15 relevant metabolites from 12 metabolic pathways
Recently, PCOS subjects (n = 12) and age/BMI matched controls (n = 10) participated in an eight week exercise program at 60% maximal oxygen consumption [176
]. Participants had maximal stimulation of insulin resistance and, subsequent, insulin sensitivity measured before and after exercise. The metabolic profile of both groups was measured at baseline and at stimulated max insulin resistance, both before and after exercise. Results found significantly increased concentrations of 8 amino acids in PCOS subjects compared to the control group prior to exercise. Of note, these amino acids were not significantly different between groups following exercise.
Previous studies have noted a higher prevalence of decreased insulin sensitivity, altered mitochondrial function, and increased insulin resistance (IR) in subjects with PCOS [177
]. Furthermore, adults with IR and either obesity or diabetes typically exhibit altered amino acids (AA), free fatty acids (FFA), and acylcarnitines (AC) profiles. A recent study sought to assess whether these same metabolites are associated with IR/androgens in adolescent PCOS [180
]. Teenage, obese girls with PCOS (n = 15) and without PCOS (n = 6) provided plasma samples during a fasting insulin state and hyperinsulinemic state. LC-MS of plasma samples revealed significantly altered AA, AC, and FFA profiles in the PCOS group in both insulin states compared to the control group. Androgen levels had a negative correlation to AC, but no correlation was identified with FFA or AA. These results reveal that young, obese individuals with PCOS exhibit a unique metabolic profile during fasting and hyperinsulinemia that may be associated with IR and metabolic diseases, such as diabetes mellitus.
UPLC-Q-TOF-MS analysis of plasma from PCOS patients (n = 49) and controls (n = 50) was performed to identify variances in metabolic profiles between these groups [181
]. This technique identified abnormal metabolites from lipid and hormone metabolism pathways, such as testosterone, estradiol, low-density lipoprotein, and apolipoprotein B. Significantly different metabolites allowed for the differentiation of PCOS subjects from controls with 100% sensitivity and 86% specificity. These findings support the role of disordered hormone and lipid pathways in the occurrence of PCOS.
The urinary metabolic profile was analyzed in PCOS subjects (n = 21) and healthy controls (n = 16) using GC-MS to assess for differentiating metabolites [182
]. 35 significantly altered metabolites were identified in the PCOS group compared to the control group. Lactose, stearic acid, palmitic acid, and succinic acid were found to be most useful in the differentiation of PCOS subjects. Furthermore, PCA allowed for the grouping of PCOS subjects and controls into two unique units. Of the differentially expressed metabolites, stearic acid, palmitic acid, benzoglycine, and threonine were the most promising differentiating metabolites.
A UPLC-MS/MS study recently identified 59 altered metabolite concentrations between PCOS patients (n = 22) and controls (n = 15) [183
]. After analysis, six metabolites of steroid hormone biosynthetic pathways were determined to be the most promising differentiating metabolites. While the identification of these differentiating metabolites is encouraging, further research must be performed to elucidate their role in the pathogenesis of PCOS and their value in PCOS diagnosis.