Nuclear magnetic resonance metabolomics and human liver diseases: The principles and evidence associated with protein and carbohydrate metabolism (Review)
- Authors:
- Published online on: March 3, 2017 https://doi.org/10.3892/br.2017.868
- Pages: 387-395
Abstract
Introduction
Outcomes for patients suffering from liver diseases are particularly difficult to predict (1). Biochemical liver markers facilitate with disease diagnosis (2), however, there is a lack of useful biomarkers that allow physicians to identify patients at higher risk of developing these diseases or that enable physicians to predict treatment response. In order to accurately determine patient prognoses and to adapt therapeutic procedures to each patient's situation, three solutions are proposed as follows: i) Interpretation of novel data from clinical or biological tests; ii) identification of novel biomarkers; and iii) novel methods for processing of biochemical and clinical data. The identification and validation of biomarkers is currently a major focus of liver disease research, and aims to refine liver disease prognosis and to adapt therapeutic procedures to individual patient features. The characteristics defining these biomarkers are required to be: i) Applicable in routine practice, ii) discriminant, and iii) accurate, robust and universal. Proteomics and metabolomics are new approaches to the identification and characterization of novel biomarkers. High-throughput analysis of metabolism in a biological fluid now allows the evaluation of a large number of metabolites simultaneously [for example, in the case of liver cancer (3)]. It is widely accepted that the concept of identifying a single biomarker in routine clinical practice for the management of patients suffering from a liver disease is probably obsolete (4). Furthermore, it may be more informative to correlate biomarker levels with clinical events and to analyse its variation during follow-up than to have a single measurement of a given compound in a biological fluid. Metabolomic approaches necessitate a translational bridge between clinical and basic research. The aim of the present review is to provide a state-of-the-art overview of nuclear magnetic resonance (NMR) metabolomic studies in the field of human liver diseases using the Pubmed database (from 1994 up until May 2016). The review briefly presents the different biofluids or tissues that have been investigated, the NMR experiments that are recorded and the multivariate analyses that were performed in the NMR metabolomic liver studies. In addition, the dysfunctions associated with protein and carbohydrate metabolism that have been illustrated in these studies are reviewed.
Data acquisition and statistical analysis
Data acquisition
Typical data acquisition has previously been reviewed (5). The majority of the studies investigated samples obtained from serum or plasma (Tables I and II) (6–25), which are usually the most pertinent biofluids for hepatological disease. Various studies sought to identify urine NMR fingerprints (22,26) and one study combined urine and plasma (18). Notably, using endoscopic retrograde cholangiopancreatography, Sharif et al (25) collected and evaluated bile samples by NMR. To limit inter-individual variability and to optimize metabolite concentration variations that correspond to intra-individual variability, previous studies collected sequential sera (19,27). The second type of sample investigated consisted of intact tissues that were evaluated by high-resolution magic-angle spinning (HR-MAS) spectroscopy, which is performed on whole tissue from liver biopsies and preserves the majority of the cellular architecture (6,12,14). The typical tissue mass required is between 15 and 25 mg, however, even 1 to 10 mg is sufficient when using smaller sample vessels. Concerning biopsies or tissues, a dual chloroform/methanol extraction may be performed to separate water-soluble and lipid-containing extracts (19). In all publications, classical NMR experiments (one-dimensional nuclear Overhauser effect spectroscopy and/or Car-Purcell-Meiboom-Gill) and 2D experiments (total correlation spectroscopy and J-resolved) were recorded. A study by Duarte et al (28) using heteronuclear multiple quantum coherence 1H-13C spectra proposed to easily identify metabolites from chemical shifts. Notably, this study also measured T1 and T2 relaxation times for CH2-CH2-CO lipid protons, which are correlated with lipid mobility. Another promising approach is to combine data from various-omics (metabolomics, proteomics, transcriptomics and genomics) techniques or from clinical or biological data. Andersson et al (20) combined NMR with mass spectrometry metabolomics and proteomics results. Cobbold et al (12) combined the results of in vitro 1H NMR spectroscopy with in vivo magnetic resonance spectroscopy to specifically determine lipid compositions.
Statistical analysis
The multivariate analysis methods, which are most commonly used in the studies discussed in the current review, are presented.
Unsupervised analysis
Analyzing spectroscopic data information with unsupervised Principal Component Analysis (PCA) is generally the first step of the statistical investigation (29). The aim of this technique is to reduce the multidimensional data space by preserving the majority of the variance (30). Usually, PCA is used for the detection of outliers or to observe clustering of the dataset according to important variations in metabolite profiles.
Supervised analysis
The second step in most of the statistical analyses for metabolomics is to use a regression-based method termed partial least squares (PLS) analysis. PLS analysis is the most widely used statistical method to interpret NMR multivariate data (31). PLS models attempt to establish a linear association between an X predictor matrix (e.g., spectrometric data of biological samples) and a Y response matrix (e.g., clinical results or treatments). Based on PLS analysis, PLS-discriminant analysis (DA) was introduced specifically for classification problems (32). Orthogonal projections to latent structures DA, which is an extension of PLS-DA, is based on applying an integrated orthogonal signal correction filter to remove variability not relevant to class separation (33). This statistical method has been increasingly used. The validation is essential, particularly from a clinical perspective. The only method of reliably estimating the ability of the model to predict Y values of novel individuals is to predict individuals from an independent dataset (i.e., those that were not used to build the model). The cross-validation method is predominantly used when no independent dataset is available. The principle of this technique consists of splitting the cohort into training and validation sets. The training set is used to build a model that is posteriorly tested with the validation set. The cross-validation process is completed when all samples in the cohort have been used at least once in the training and validation sets. Considering this principle, many cross-validation procedures have been proposed. These procedures may differ in the method by which the training and validation sets are constructed. A recent publication evaluated the impact of rows order on cross-validation quality parameters (34).
Further statistical methods
Additional methods will be cited, however a full explanation of the statistical basis of these methods is outside the scope of the present review. The significance analysis of microarrays and metabolites approach was used by Munshi et al (18). This method recognizes the chemical shifts (metabolites) with statistically significant differences between experimental groups by integrating data from a set of metabolite-specific t-tests wherein each chemical shift is given a score, calculated on the basis of changes in its level relative to the standard deviation of repeated measurements for those chemical shifts. The chemical shifts with scores above a defined threshold are considered as potentially significant.
Variable selection was used by Godoy et al (35). The n-1 components of the PCA (where n is the number of patients) were calculated. The components associated significantly with the question (discrimination between disease group and controls) were used to build a function predicting assignment to one group or the other.
Serkova et al (19) followed metabolite trajectories. Three blood samples (2, 24 and 48 h after liver transplantation) were used to measure and follow metabolite concentrations. An additional method used by Westerhuis et al (36) was paired analysis. Briefly, the between subject variation, which is described by the average of two observations from each subject (for example before and after treatment) is separated from the within subject variation, which is the difference between the two observations. This split provides the advantage of specifically discriminating the metabolic changes within individuals caused by the intervention and allows evaluation of the large variability between human subjects due to factors such as age, body mass index, genetics or environmental differences [for an example, refer to Goossens et al (27)].
Random forests were used by Liu et al (37). This algorithm is particularly efficient in cases of large datasets. It has been widely used in genomic and proteomic studies, and it may be suitable for metabolomic studies. In addition to the above-mentioned types of analyses, non-linear methods were used (38).
Metabolism in the liver
This section of the review presents the metabolic transformations described in the NMR studies that have attempted to characterize liver diseases. The liver is considered to be the major body organ for metabolism, storage and detoxification. The nutriments and xenobiotics that cross the intestinal barrier reach the liver before being delivered through the general circulation. In the liver, these compounds undergo metabolic transformations that may activate or deactivate the molecules. In the majority of cases, metabolic transformation produces metabolites that are directly usable by other organs.
In addition, the current review describes the major metabolic pathway impairments that are highlighted by NMR metabolomic studies of hepatic disease. The xenobiotic detoxification process that takes place in the liver will not be considered here. A recent review focused on hepatocellular carcinoma (HCC) (39) and lipid metabolism was recently investigated (40). Protein and carbohydrate metabolism modulations that are the result of liver pathologies are presented. Furthermore, a summary of these metabolic pathway modifications is provided in Tables I and II (6–25).
Protein metabolism
The liver is where the synthesis and degradation of proteins occur. Amino acids (AA) absorbed in the intestine are used in the liver for the synthesis of structure and transport proteins, as well as enzymes. AAs may also be deaminated for direct energy supply or gluconeogenesis. The degradation of proteins that takes place in the liver may also be used to supply energy needs. In the liver, various AAs enter the detoxification process by conjugation reactions. The result of AA catabolism is the production of ammonia, which is eliminated as urea and creatinine in the urine through the kidneys. All these metabolic pathways involve a large number of enzymatic activities, which are frequently impaired in hepatic diseases. Metabolomic studies of these liver impairments frequently report modifications of the AA content in serum, plasma, tissues and bile.
In serum and plasma NMR spectra, and in tissue spectra, alanine is the easiest AA to detect. Alanine participates in protein metabolism and may be considered as an anabolite and a catabolite of protein. In addition, alanine enters the energy metabolic pathways following deamination into lactate and this process occurs, for the most part, in the liver. The enzyme responsible for this deamination, alanine aminotransferase is routinely assayed as a marker of hepatic impairment. Consequently, variations in alanine have been frequently mentioned in hepatic pathologies (13). According to trends observed in the variation of alanine concentrations in liver diseases, it appears that alanine levels increase in the presence of pathological samples when compared with non-pathological samples, and this increase is enhanced when the disease worsens. This is the case for tumours (6), cirrhosis (8,13) and for their complications, such as hepatic encephalopathy or cirrhosis decompensation (10,21). In the case of acetaminophen toxicity, elevated alanine levels are identified in urine (11).
Glutamine and glutamate are AAs involved in nitrogen metabolism. Glutamine is the most abundant AA and one of the entry points into the tricarboxylic acid cycle (TCA; citric acid or Kreb's cycle) following its deamination into ketoglutarate. Glutamate is a precursor of glutamine. In liver pathologies, the levels of glutamine are increased in plasma and serum. As observed for alanine, the levels of glutamine and glutamate increase in plasma and serum, and in liver tissue, in cases of HCC (6), cirrhosis [when compared with control subjects or with fibrosis in tissue (14) or serum (12)], hepatic transplantation failure (7,19), and in fulminant hepatic failure (17). The same increases may also be identified in urine samples in cases of fulminant hepatitis (17). The only cases where a decrease in serum glutamine levels were reported were during the discrimination between viral and ethanol aetiologies of cirrhosis (8) and when comparing the severity of cirrhosis (13).
Glycine is generated from serine, which is derived from 3-phosphoglycerate formed by glycolysis. Glycine participates directly in the urea cycle through its degradation into ammonia and CO2. Consequently, variations in glycine levels follow those of glutamine and glutamate in hepatic diseases. However, hepatoencephalopathy is notable; when this neurologic side effect of cirrhosis is present, glycine levels in serum decrease (21). This may be associated with the role of this AA as an inhibitory neurotransmitter.
In NMR spectra, essential AAs are also detectable and frequently reported as branched-chain AAs (BCAAs) and aromatic AAs (AAAs). BCAAs include valine, leucine and isoleucine, which have resonances located in the same region of the NMR spectrum that frequently overlap. BCAAs are essential AAs that must be contained in the diet and they are involved in the regulation of protein synthesis, glucose metabolism and oxidation.
In liver tissue, only leucine has been found to be increased in HCC compared to adjacent non-tumour tissue and this increase is associated with a higher tumour grade (6). In plasma and serum, a BCAA decrease signals symptoms of encephalopathy in cirrhotic patients (21) and cirrhosis of viral or alcoholic aetiology when compared with non-cirrhotic patients. Conversely, an increase in leucine was reported in the hepatitis E infection when compared with controls, while the concentration of valine decreased. Similarly, with severity of alcoholic cirrhosis, the concentrations of leucine and isoleucine, but not valine are increased (13). Low levels of BCAA have been associated with neuropathological disorders associated with cirrhosis (21).
The AAA group of essential AAs produces signals in the region >6 ppm in NMR spectra. This region is generally less crowded in serum and urine spectra than the regions <5 ppm, thus, these AAs are easy to detect when increased in fluids and tissues. The primary AAs involved are tyrosine, phenylalanine and histidine. Phenylalanine and tyrosine are metabolically associated, as tyrosine is synthesized from phenylalanine by phenylalanine hydroxylase, and a genetic deficit in this enzyme is responsible for a metabolic disease termed phenylketonuria, which is characterized by abnormal excretion of phenylalanine in the urine. This disease causes neurologic disorders. The concentrations of these AAA have been reported to decrease in plasma (and urine) of patients with viral hepatitis infections (18). By contrast, AAA are reported to be increased in cirrhosis (8) and in cases of hepatic transplantation failure (7). The end point of catabolism for AAA is composed of fumarate and acetoacetate. Their increase may also be associated with energy supply modulation in the liver (8,18).
Downstream of AA metabolism, the aminated molecules are considered to be direct catabolites of AAs and proteins, as their role is to eliminate nitrogen in the form of ammonia produced by deamination of AAs. Ammonia is detoxified in the urea cycle, which occurs in the liver. However, to the best of our knowledge, urea levels were not referred to in the metabolomic studies of hepatic disease. Conversely, various nitrogen-containing compounds were found to be increased in urine, including trimethylamine N-oxide and creatine (22).
Glucose metabolism and the TCA cycle
The liver is directly involved in glucose metabolism and the consequent energy supply to the organism. The liver is the organ containing the highest levels of glycogen, the storage form of glucose, and this compound may be synthesized through hepatic gluconeogenesis. Other sugars may be transformed into glucose.
The major metabolites directly implicated in glucose metabolism and detected with NMR spectroscopy are glucose (detected in blood and tissues), glycogen (detected in liver tissue), lactate and pyruvate. Various other compounds participating in the TCA cycle are also present in biofluids and tissues in sufficient concentrations for detection by NMR spectroscopy. Those reported in the literature concerning hepatic diseases are citrate, fumarate and succinate (6,8). Their concentration may vary in opposite directions, depending on the pathology.
Glycaemia is a relatively stable biological value, as the majority of the hepatic pathways, including glucogenesis, glycolysis and gluconeogenesis attempt to maintain constant glucose levels. However, three different studies have shown an increase of glucose levels in serum in cases of cirrhosis. The first study compared viral and alcoholic cirrhosis to control subjects (8) and the second showed that levels of serum glucose increase with cirrhosis severity (13). The third study reported higher glycaemia for cirrhotic patients with encephalopathy than those without encephalopathy (21). Glycogen NMR signals are detected in liver tissues under resolution conditions offered by the HR-MAS technique. The glycogen content of liver tissues is decreased for high-grade HCC when compared with low-grade disease and to adjacent non-tumour tissue (6). Similarly, glycogen is decreased in cirrhotic tissues when compared with mild and moderate fibrosis (12). These lower glucose and liver glycogen levels appear to be associated with pathology grade for cirrhosis and carcinoma, and should be associated with lactate content, which is increased in high-grade HCC, when compared with low-grade HCC tissues (6), and in the sera of cirrhotic patients with hepatoencephalopathy when compared with cirrhotic patients without encephalopathy (21). Subsequent to liver transplantation, lactate in the blood is also increased in cases of a non-functional transplant when compared with successful transplants (19) and in more severe cirrhosis (13). This pattern favours a deviation toward an anaerobic glycolysis pathway utilizing more glucose molecules for energy supply. However, lactate in the blood was found to be lower in cirrhotic patients than in control patients (8) and in cases of viral infection comparing the hepatitis B virus with the hepatitis E virus (18), indicating an indirect association between glucose, glycogen and lactate. It can be noted that pyruvate content is also modulated (8,10), as well as citrate (18,19) and the levels of other TCA compounds (including formate and succinate) (8,10,18).
It is apparent that impairments in liver glucose and energy metabolism are detected in liver tissue and have systemic consequences, which are detected in serum and plasma as the imbalanced profile of the metabolites participating in these pathways.
Conclusion
The overview of literature results presented in the current review indicates that NMR metabolomic studies provide important data for improving the diagnosis and prognosis of human liver diseases. This technique is non-invasive, simple, inexpensive and fast, and has high sensitivity and good specificity. The question that must be addressed now is how metabolomics will impact the healthcare of patients with liver diseases. It is tempting to speculate that intra-individual analysis of changes during follow-up of treated patients may provide essential information regarding therapeutic responses. Various conditions associated with a rigorous clinical methodological approach need to be addressed in this context. First, these research programs should be developed in large prospective cohorts of well-defined patients undergoing standardized therapeutic procedures. Second, in order to avoid biases that could influence metabolomic profiling of patients, attention should be paid with respect to the stage and cause of the underlying liver diseases. Third, repeated metabolomic profiling should be performed in samples drawn at different time points corresponding to specific clinical situations as follows: Before treatment, after treatment at the time of judgement of efficacy, and subsequently at interesting end-points (41,42). Fourth, combining datasets from different metabolomics (such as NMR and mass spectrometry) or omics (such as genomics and transcriptomics) techniques will improve the understanding of the different metabolic pathways. The development of statistical models for accurate interpretation of variations in metabolomic fingerprints may then be used to provide pivotal information for patient management. Integration of metabolomic-based diagnostic principles will likely be an essential tool for the development of personalized medicine for decision-making in liver disease management.
Glossary
Abbreviations
Abbreviations:
|
BCAA |
branched-chain amino acids |
|
AAA |
aromatic amino acids |
|
HCV |
hepatitis C virus |
|
HCC |
hepatocellular carcinoma |
|
HR-MAS |
high-resolution magic-angle spinning |
|
MHE |
minimal hepatic encephalopathy |
|
NAFLD |
non-alcoholic fatty liver disease |
|
NMR |
nuclear magnetic resonance |
|
PCA |
principal component analysis |
References
|
Kumar R Shalimar, Sharma H, Goyal R, Kumar A, Khanal S, Prakash S, Gupta SD, Panda SK and Acharya SK: Prospective derivation and validation of early dynamic model for predicting outcome in patients with acute liver failure. Gut. 61:1068–1075. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Gowda S, Desai PB, Hull VV, Math AA, Vernekar SN and Kulkarni SS: A review on laboratory liver function tests. Pan Afr Med J. 3:172009.PubMed/NCBI | |
|
Nie W, Yan L, Lee YH, Guha C, Kurland IJ and Lu H: Advanced mass spectrometry-based multi-omics technologies for exploring the pathogenesis of hepatocellular carcinoma. Mass Spectrom Rev. 35:331–349. 2016. View Article : Google Scholar : PubMed/NCBI | |
|
Chaiteerakij R, Addissie BD and Roberts LR: Update on biomarkers of hepatocellular carcinoma. Clin Gastroenterol Hepatol. 13:237–245. 2015. View Article : Google Scholar : PubMed/NCBI | |
|
Smolinska A, Blanchet L, Buydens LM and Wijmenga SS: NMR and pattern recognition methods in metabolomics: from data acquisition to biomarker discovery: a review. Anal Chim Acta. 750:82–97. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Yang Y, Li C, Nie X, Feng X, Chen W, Yue Y, Tang H and Deng F: Metabonomic studies of human hepatocellular carcinoma using high-resolution magic-angle spinning 1H NMR spectroscopy in conjunction with multivariate data analysis. J Proteome Res. 6:2605–2614. 2007. View Article : Google Scholar : PubMed/NCBI | |
|
Tripathi P, Bala L, Saxena R, Yachha SK, Roy R and Khetrapal CL: 1H NMR spectroscopic study of blood serum for the assessment of liver function in liver transplant patients. J Gastrointestin Liver Dis. 18:329–336. 2009.PubMed/NCBI | |
|
Qi S, Tu Z, Ouyang X, Wang L, Peng W, Cai A and Dai Y: Comparison of the metabolic profiling of hepatitis B virus-infected cirrhosis and alcoholic cirrhosis patients by using (1) H NMR-based metabonomics. Hepatol Res. 42:677–685. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Cheng S, Wiklund P, Autio R, Borra R, Ojanen X, Xu L, Törmäkangas T and Alen M: Adipose Tissue Dysfunction and Altered Systemic Amino Acid Metabolism Are Associated with Non-Alcoholic Fatty Liver Disease. PLoS One. 10:e01388892015. View Article : Google Scholar : PubMed/NCBI | |
|
Qi SW, Tu ZG, Peng WJ, Wang LX, Ou-Yang X, Cai AJ and Dai Y: 1H NMR-based serum metabolic profiling in compensated and decompensated cirrhosis. World J Gastroenterol. 18:285–290. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Winnike JH, Li Z, Wright FA, Macdonald JM, O'Connell TM and Watkins PB: Use of pharmaco-metabonomics for early prediction of acetaminophen-induced hepatotoxicity in humans. Clin Pharmacol Ther. 88:45–51. 2010. View Article : Google Scholar : PubMed/NCBI | |
|
Cobbold JF, Patel JH, Goldin RD, North BV, Crossey MM, Fitzpatrick J, Wylezinska M, Thomas HC, Cox IJ and Taylor-Robinson SD: Hepatic lipid profiling in chronic hepatitis C: An in vitroin vivo proton magnetic resonance spectroscopy study. J Hepatol. 52:16–24. 2010. View Article : Google Scholar : PubMed/NCBI | |
|
Amathieu R, Nahon P, Triba M, Bouchemal N, Trinchet JC, Beaugrand M, Dhonneur G and Le Moyec L: Metabolomic approach by 1H NMR spectroscopy of serum for the assessment of chronic liver failure in patients with cirrhosis. J Proteome Res. 10:3239–3245. 2011. View Article : Google Scholar : PubMed/NCBI | |
|
Martínez-Granados B, Morales JM, Rodrigo JM, Del Olmo J, Serra MA, Ferrández A, Celda B and Monleón D: Metabolic profile of chronic liver disease by NMR spectroscopy of human biopsies. Int J Mol Med. 27:111–117. 2011.PubMed/NCBI | |
|
Embade N, Mariño Z, Diercks T, Cano A, Lens S, Cabrera D, Navasa M, Falcón-Pérez JM, Caballería J, Castro A, et al: Metabolic Characterization of Advanced Liver Fibrosis in HCV Patients as Studied by Serum 1H-NMR Spectroscopy. PLoS One. 11:e01550942016. View Article : Google Scholar : PubMed/NCBI | |
|
Nahon P, Amathieu R, Triba MN, Bouchemal N, Nault JC, Ziol M, Seror O, Dhonneur G, Trinchet JC, Beaugrand M, et al: Identification of serum proton NMR metabolomic fingerprints associated with hepatocellular carcinoma in patients with alcoholic cirrhosis. Clin Cancer Res. 18:6714–6722. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Saxena V, Gupta A, Gowda GA Nagana, Saxena R, Yachha SK and Khetrapal CL: 1H NMR spectroscopy for the prediction of therapeutic outcome in patients with fulminant hepatic failure. NMR Biomed. 19:521–526. 2006. View Article : Google Scholar : PubMed/NCBI | |
|
Munshi SU, Taneja S, Bhavesh NS, Shastri J, Aggarwal R and Jameel S: Metabonomic analysis of hepatitis E patients shows deregulated metabolic cycles and abnormalities in amino acid metabolism. J Viral Hepat. 18:e591–e602. 2011. View Article : Google Scholar : PubMed/NCBI | |
|
Serkova NJ, Zhang Y, Coatney JL, Hunter L, Wachs ME, Niemann CU and Mandell MS: Early detection of graft failure using the blood metabolic profile of a liver recipient. Transplantation. 83:517–521. 2007. View Article : Google Scholar : PubMed/NCBI | |
|
Andersson U, Lindberg J, Wang S, Balasubramanian R, Marcusson-Ståhl M, Hannula M, Zeng C, Juhasz PJ, Kolmert J, Bäckström J, et al: A systems biology approach to understanding elevated serum alanine transaminase levels in a clinical trial with ximelagatran. Biomarkers. 14:572–586. 2009. View Article : Google Scholar : PubMed/NCBI | |
|
Jiménez B, Montoliu C, MacIntyre DA, Serra MA, Wassel A, Jover M, Romero-Gomez M, Rodrigo JM, Pineda-Lucena A and Felipo V: Serum metabolic signature of minimal hepatic encephalopathy by (1)H-nuclear magnetic resonance. J Proteome Res. 9:5180–5187. 2010. View Article : Google Scholar : PubMed/NCBI | |
|
Shariff MI, Gomaa AI, Cox IJ, Patel M, Williams HR, Crossey MM, Thillainayagam AV, Thomas HC, Waked I, Khan SA, et al: Urinary metabolic biomarkers of hepatocellular carcinoma in an Egyptian population: A validation study. J Proteome Res. 10:1828–1836. 2011. View Article : Google Scholar : PubMed/NCBI | |
|
McPhail MJ, Shawcross DL, Lewis MR, Coltart I, Want EJ, Antoniades CG, Veselkov K, Triantafyllou E, Patel V, Pop O, et al: Multivariate metabotyping of plasma predicts survival in patients with decompensated cirrhosis. J Hepatol. 64:1058–1067. 2016. View Article : Google Scholar : PubMed/NCBI | |
|
Ranjan P, Gupta A, Kumar S, Gowda GA, Ranjan A, Sonker AA, Chandra A and Ramakant: Detection of new amino acid markers of liver trauma by proton nuclear magnetic resonance spectroscopy. Liver Int. 26:703–707. 2006. View Article : Google Scholar : PubMed/NCBI | |
|
Sharif AW, Williams HR, Lampejo T, Khan SA, Bansi DS, Westaby D, Thillainayagam AV, Thomas HC, Cox IJ and Taylor-Robinson SD: Metabolic profiling of bile in cholangiocarcinoma using in vitro magnetic resonance spectroscopy. HPB Oxf. 12:396–402. 2010. View Article : Google Scholar | |
|
Shariff MI, Ladep NG, Cox IJ, Williams HR, Okeke E, Malu A, Thillainayagam AV, Crossey MM, Khan SA, Thomas HC, et al: Characterization of urinary biomarkers of hepatocellular carcinoma using magnetic resonance spectroscopy in a Nigerian population. J Proteome Res. 9:1096–1103. 2010. View Article : Google Scholar : PubMed/NCBI | |
|
Goossens C, Nahon P, Le Moyec L, Triba MN, Bouchemal N, Amathieu R, Ganne-Carrié N, Ziol M, Trinchet JC, Sellier N, et al: Sequential Serum Metabolomic Profiling after Radiofrequency Ablation of Hepatocellular Carcinoma Reveals Different Response Patterns According to Etiology. J Proteome Res. 15:1446–1454. 2016. View Article : Google Scholar : PubMed/NCBI | |
|
Duarte IF, Stanley EG, Holmes E, Lindon JC, Gil AM, Tang H, Ferdinand R, McKee CG, Nicholson JK, Vilca-Melendez H, et al: Metabolic assessment of human liver transplants from biopsy samples at the donor and recipient stages using high-resolution magic angle spinning 1H NMR spectroscopy. Anal Chem. 77:5570–5578. 2005. View Article : Google Scholar : PubMed/NCBI | |
|
Jackson JE: A User's Guide To Principal Components. John Wiley & Sons, Inc.; Hoboken, NJ: 1991, View Article : Google Scholar | |
|
Holmes E, Foxall PJ, Nicholson JK, Neild GH, Brown SM, Beddell CR, Sweatman BC, Rahr E, Lindon JC, Spraul M, et al: Automatic data reduction and pattern recognition methods for analysis of 1H nuclear magnetic resonance spectra of human urine from normal and pathological states. Anal Biochem. 220:284–296. 1994. View Article : Google Scholar : PubMed/NCBI | |
|
Wold S, Sjöström M and Eriksson L: PLS-regression: A basic tool of chemometrics. Chemom Intell Lab Syst. 58:109–130. 2001. View Article : Google Scholar | |
|
Barker M and Rayens W: Partial least squares for discrimination. J Chemometr. 17:166–173. 2003. View Article : Google Scholar | |
|
Trygg J and Wold S: Orthogonal projections to latent structures (O-PLS). J Chemometr. 16:119–128. 2002. View Article : Google Scholar | |
|
Triba MN, Le Moyec L, Amathieu R, Goossens C, Bouchemal N, Nahon P, Rutledge DN and Savarin P: PLS/OPLS models in metabolomics: The impact of permutation of dataset rows on the K-fold cross-validation quality parameters. Mol Biosyst. 11:13–19. 2015. View Article : Google Scholar : PubMed/NCBI | |
|
Godoy MM, Lopes EP, Silva RO, Hallwass F, Koury LC, Moura IM, Gonçalves SM and Simas AM: Hepatitis C virus infection diagnosis using metabonomics. J Viral Hepat. 17:854–858. 2010. View Article : Google Scholar : PubMed/NCBI | |
|
Westerhuis JA, van Velzen EJ, Hoefsloot HC and Smilde AK: Multivariate paired data analysis: Multilevel PLSDA versus OPLSDA. Metabolomics. 6:119–128. 2010. View Article : Google Scholar : PubMed/NCBI | |
|
Liu Y, Hong Z, Tan G, Dong X, Yang G, Zhao L, Chen X, Zhu Z, Lou Z, Qian B, et al: NMR and LC/MS-based global metabolomics to identify serum biomarkers differentiating hepatocellular carcinoma from liver cirrhosis. Int J Cancer. 135:658–668. 2014. View Article : Google Scholar : PubMed/NCBI | |
|
Fonville JM, Bylesjö M, Coen M, Nicholson JK, Holmes E, Lindon JC and Rantalainen M: Non-linear modeling of 1H NMR metabonomic data using kernel-based orthogonal projections to latent structures optimized by simulated annealing. Anal Chim Acta. 705:72–80. 2011. View Article : Google Scholar : PubMed/NCBI | |
|
Kimhofer T, Fye H, Taylor-Robinson S, Thursz M and Holmes E: Proteomic and metabonomic biomarkers for hepatocellular carcinoma: A comprehensive review. Br J Cancer. 112:1141–1156. 2015. View Article : Google Scholar : PubMed/NCBI | |
|
Amathieu R, Triba MN, Goossens C, Bouchemal N, Nahon P, Savarin P and Le Moyec L: Nuclear magnetic resonance based metabolomics and liver diseases: Recent advances and future clinical applications. World J Gastroenterol. 22:417–426. 2016. View Article : Google Scholar : PubMed/NCBI | |
|
Toyoda H, Kumada T, Tada T, Niinomi T, Ito T, Kaneoka Y and Maeda A: Prognostic significance of a combination of pre- and post-treatment tumor markers for hepatocellular carcinoma curatively treated with hepatectomy. J Hepatol. 57:1251–1257. 2012. View Article : Google Scholar : PubMed/NCBI | |
|
Wang X, Zhang A and Sun H: Power of metabolomics in diagnosis and biomarker discovery of hepatocellular carcinoma. Hepatology. 57:2072–2077. 2013. View Article : Google Scholar : PubMed/NCBI |
