Multivariate Pattern Recognition Analysis Research Articles

A proton nuclear magnetic resonance-based metabonomics study of metabolic profiling in immunoglobulin a nephropathy

OBJECTIVES:Immunoglobulin A nephropathy is the most common cause of chronic renal failure among primary glomerulonephritis patients. The ability to diagnose immunoglobulin A nephropathy remains poor. However, renal biopsy is an inconvenient, invasive, and painful examination, and no reliable biomarkers have been developed for use in routine patient evaluations. The aims of the present study were to identify immunoglobulin A nephropathy patients, to identify useful biomarkers of immunoglobulin A nephropathy and to establish a human immunoglobulin A nephropathy metabolic profile.METHODS:Serum samples were collected from immunoglobulin A nephropathy patients who were not using immunosuppressants. A pilot study was undertaken to determine disease-specific metabolite biomarker profiles in three groups: healthy controls (N = 23), low-risk patients in whom immunoglobulin A nephropathy was confirmed as grades I-II by renal biopsy (N = 23), and high-risk patients with nephropathies of grades IV-V (N = 12). Serum samples were analyzed using proton nuclear magnetic resonance spectroscopy and by applying multivariate pattern recognition analysis for disease classification.RESULTS:Compared with the healthy controls, both the low-risk and high-risk patients had higher levels of phenylalanine, myo-Inositol, lactate, L6 lipids ( = CH-CH2-CH = O), L5 lipids (-CH2-C = O), and L3 lipids (-CH2-CH2-C = O) as well as lower levels of β-glucose, α-glucose, valine, tyrosine, phosphocholine, lysine, isoleucine, glycerolphosphocholine, glycine, glutamine, glutamate, alanine, acetate, 3-hydroxybutyrate, and 1-methylhistidine.CONCLUSIONS:These metabolites investigated in this study may serve as potential biomarkers of immunoglobulin A nephropathy. Point scoring of pattern recognition analysis was able to distinguish immunoglobulin A nephropathy patients from healthy controls. However, there were no obvious differences between the low-risk and high-risk groups in our research. These results offer new, sensitive and specific, noninvasive approaches that may be of great benefit to immunoglobulin A nephropathy patients by enabling earlier diagnosis.

1H NMR-based metabolomic study of metabolic profiling for systemic lupus erythematosus

Systemic lupus erythematosus (SLE) is a chronic inflammatory disease characterized by multi-system involvement, diverse clinical presentation, and alterations in circulating metabolites. In this study, a (1)H NMR spectroscopy-based metabolomics approach was applied to establish a human SLE serum metabolic profile. Serum samples were obtained from patients with SLE (n = 64), patients with rheumatoid arthritis (RA) (n = 30) and healthy controls (n = 35). The NOESYPR1D spectrum combined with multi-variate pattern recognition analysis was used to cluster the groups and establish a disease-specific metabolites phenotype. Principal component analysis (PCA) and orthogonal partial least-squares discriminant analysis (OPLS-DA) models were capable of distinguishing SLE or RA patients from healthy subjects. The OPLS-DA model was able to predict diagnosis of SLE with a sensitivity rate of 60.9% and a specificity rate of 97.1%. For diagnosing RA, the model has much higher sensitivity (96.7%) and specificity (91.4%). The SLE serum samples were characterized by reduced concentrations of valine, tyrosine, phenylalanine, lysine, isoleucine, histidine, glutamine, alanine, citrate, creatinine, creatine, pyruvate, high-density lipoprotein, cholesterol, glycerol, formate and increased concentrations of N-acetyl glycoprotein, very low-density lipoprotein and low-density lipoprotein in comparison with the control population. The results not only indicated that serum NMR-based metabolomic methods had sufficient sensitivity and specificity to distinguish SLE and RA from healthy controls, but also have the potential to be developed into a clinically useful diagnostic tool, and could also contribute to a further understanding of disease mechanisms.

Metabolic profiling of bile in cholangiocarcinoma using in vitro magnetic resonance spectroscopy

ObjectivesCholangiocarcinoma (CCA) has a poor prognosis and its aetiology is inadequately understood. Magnetic resonance spectroscopy (MRS) of bile may provide insights into the pathogenesis of CCA and help identify novel diagnostic biomarkers. The aim of this study was to compare the chemical composition of bile from patients with CCA with that of bile from patients with benign biliary disease. MethodsMagnetic resonance spectra were acquired from the bile of five CCA patients and compared with MRS of control bile from patients with benign biliary disease (seven with gallstones, eight with sphincter of Oddi dysfunction [SOD], five with primary sclerosing cholangitis [PSC]). Metabolic profiles were compared using both univariate and multivariate pattern-recognition analysis. ResultsUnivariate analysis showed that levels of glycine-conjugated bile acids were significantly increased in patients with CCA, compared with the benign disease groups (P= 0.002). 7β primary bile acids were significantly increased (P= 0.030) and biliary phosphatidylcholine (PtC) levels were reduced (P= 0.010) in bile from patients with CCA compared with bile from gallstone patients. These compounds were also of primary importance in the multivariate analysis: the cohorts were differentiated by partial least squares discriminant analysis (PLS-DA). ConclusionsThese preliminary data suggest that altered bile acid and PtC metabolism play an important role in CCA aetiopathogenesis and that specific metabolites may have potential as future biomarkers.

NMR‐based metabolomic analysis of cerebrospinal fluid and serum in neurological diseases – a diagnostic tool?

We sought to evaluate the diagnostic accuracy of metabolomic biomarker profiles in neurological conditions (idiopathic intracranial hypertension (IIH), multiple sclerosis (MS) and cerebrovascular disease (CVD) compared to controls with either no neurological disease or mixed neurological diseases). Spectra of CSF (n = 87) and serum (n = 72) were acquired using (1)H NMR spectroscopy. Multivariate pattern recognition analysis was used to identify disease-specific metabolite biomarker profiles. The metabolite profiles were then used to predict the diagnosis of a second cohort of patients (n = 25). CSF metabolite profiles were able to predict diagnosis with a sensitivity and specificity of 80% for both IIH and MS. The CVD serum metabolite profile was 75% sensitive and specific. On analysing the second patient cohort, the established metabolite biomarker profiles generated from the first cohort showed moderate ability to segregate patients with IIH and MS (sensitivity:specificity of 63:75% and 67:75%, respectively). These findings suggest that NMR spectroscopic metabolic profiling of CSF and serum can identify differences between IIH, MS, CVD and mixed neurological diseases. Metabolomics may, therefore, have the potential to be developed into a clinically useful diagnostic tool. The identification of disease-unique metabolites may also impart information on disease pathology.

Characterization of Inflammatory Bowel Disease With Urinary Metabolic Profiling

Distinguishing between the inflammatory bowel disease (IBD), Crohn's disease (CD), and ulcerative colitis (UC) is important for both management and prognostic reasons. Discrimination using noninvasive techniques could be an adjunct to conventional diagnostics. Differences have been shown between the intestinal microbiota of CD and UC patients and controls; the gut bacteria influence specific urinary metabolites that are quantifiable using proton high-resolution nuclear magnetic resonance (NMR) spectroscopy. This study tested the hypothesis that such metabolites differ between IBD and control cohorts, and that using multivariate pattern-recognition analysis, the cohorts could be distinguished by urine NMR spectroscopy. NMR spectra were acquired from urine samples of 206 Caucasian subjects (86 CD patients, 60 UC patients, and 60 healthy controls). Longitudinal samples were collected from 75 individuals. NMR resonances specific for metabolites influenced by the gut microbes were studied, including hippurate, formate, and 4-cresol sulfate. Multivariate analysis of all urinary metabolites involved principal components analysis (PCA) and partial least squares discriminant analysis (PLS-DA). Hippurate levels were lowest in CD patients and differed significantly between the three cohorts (P<0.0001). Formate levels were higher and 4-cresol sulfate levels lower in CD patients than in UC patients or controls (P=0.0005 and P=0.0002, respectively). PCA revealed clustering of the groups; PLS-DA modeling was able to distinguish the cohorts. These results were independent of medication and diet and were reproducible in the longitudinal cohort. Specific urinary metabolites related to gut microbial metabolism differ between CD patients, UC patients, and controls. The emerging technique of urinary metabolic profiling with multivariate analysis was able to distinguish these cohorts.

Validation of metabolomics for toxic mechanism of action screening with the earthworm Lumbricus rubellus

One of the promises of environmental metabolomics, together with other ecotoxicogenomic approaches, is that it can give information on toxic compound mechanism of action (MOA), by providing a specific response profile or fingerprint. This could then be used either for screening in the context of chemical risk assessment, or potentially in contaminated site assessment for determining what compound classes were causing a toxic effect. However for either of these two ends to be achievable, it is first necessary to know if different compounds do indeed elicit specific and distinct metabolic profile responses. Such a comparative study has not yet been carried out for the earthworm Lumbricus rubellus. We exposed L. rubellus to sub-lethal concentrations of three very different toxicants (CdCl2, atrazine, and fluoranthene, representing three compound classes with different expected MOA), by semi-chronic exposures in a laboratory test, and used NMR spectroscopy to obtain metabolic profiles. We were able to use simple multivariate pattern-recognition analyses to distinguish different compounds to some degree. In addition, following the ranking of individual spectral bins according to their mutual information with compound concentrations, it was possible to identify both general and specific metabolite responses to different toxic compounds, and to relate these to concentration levels causing reproductive effects in the worms.

FPA micro spectral imaging of non-Hodgkin lymphomas

A FT-IR microspectroscopy study on reactive lymph nodes and non-Hodgkin lymphomas is reported. Mid infrared absorption spectra collected at diffraction limit spatial resolution from reactive and neoplastic lymph nodes resulted sufficiently different once analysed by multivariate pattern recognition analysis to distinguish tumoral from non tumoral samples. The potential of infrared spectroscopy as a post-operative screening is gained by the use of a multielement Focal Plane Array detector. Spectral differences between normal and malignant spectra were mainly in the methyl stretching and in the low frequency region.

Liquid Chromatography−Mass Spectrometric Analysis of Urinary Metabolites and Their Pattern Recognition for the Prediction of Drug-Induced Hepatotoxicity

Multivariate pattern recognition (PR) analysis combined with LC/MS was utilized to evaluate the feasibility of predicting chemical-induced toxicity in rats. Urine samples were collected from rats treated with vehicles or four hepatotoxins, alpha-naphthyl isothiocyanate (ANIT), carbontetrachloride (CCl4), acetaminophen, and diclofenac, and analyzed by HPLC coupled with electrospray mass spectrometry. Chromatographic data were normalized using modified Z score transformation with those of the control group, to remove the vehicle effects for a further enhanced multivariate analysis. The LC/MS-based profiles of the urine samples showed different levels of endogenous metabolites, which were characteristic of each hepatotoxin. In the principal component (PC) map of the urinary spectra from rats treated with ANIT, the metabolic trajectory moved away from the predose position, reaching a maximum separation at the 32-48 h time period. The metabolic profiles partially recovered to the basal conditions on 7 days postdose. A principal component analysis was performed on the urinary spectra of rats treated with the vehicles or four hepatotoxins. Each group formed a distinct and isolated cluster in the PC map, indicating drug-induced perturbation in the urine profiles. To construct mathematical models for predicting drug-induced hepatotoxicity, supervised analyses, such as linear discriminant analysis and soft independent modeling of class analogy with residual distance (SIMCA-RD), were performed. The SIMCA-RD showed high predictability, over 95%, in the results of cross-validation using the leaving-one-out method. The developed LC/MS-PR approach might be a useful tool for the prediction of drug-induced hepatotoxicity and for the understanding of hepatotoxic mechanisms.

Antemortem Identification of Bovine Spongiform Encephalopathy from Serum Using Infrared Spectroscopy

Since 1986, more than 180 000 clinical cases of bovine spongiform encephalopathy (BSE) have been observed in the U.K. alone. Most of these cases were confirmed by postmortem examination of brain tissue. However, BSE-related risk assessment and risk management would greatly benefit from antemortem testing on living animals. A serum-based test could allow for screening of the cattle population; thus, even a BSE eradication program would be conceivable. Here we report on a novel method for antemortem BSE testing, which combines infrared spectroscopy of serum samples with multivariate pattern recognition analysis. A classification algorithm was trained using infrared spectra of bovine sera from more than 800 animals (including BSE-positive, healthy controls and animals suffering from classical viral or bacterial infections). In two validation studies, sensitivities of 85 and 84% and specificities of 86 and 91% were achieved, respectively. The combination of classification algorithms increased the sensitivity and specificity of BSE detection to 96 and 92%, respectively.