Multivariate modeling and prediction of breast cancer prognostic factors using MR metabolomics.

Axillary lymph node status together with estrogen and progesterone receptor status are important prognostic factors in breast cancer. In this study, the potential of using MR metabolomics for prediction of these prognostic factors was evaluated. Biopsies from breast cancer patients (n = 160) were excised during surgery and analyzed by high resolution magic angle spinning MR spectroscopy (HR MAS MRS). The spectral data were preprocessed and variable stability (VAST) scaled, and training and test sets were generated using the Kennard-Stone and SPXY sample selection algorithms. The data were analyzed by partial least-squares discriminant analysis (PLS-DA), probabilistic neural networks (PNNs) and Bayesian belief networks (BBNs), and blind samples (n = 50) were predicted for verification. Estrogen and progesterone receptor status was successfully predicted from the MR spectra, and were best predicted by PLS-DA with a correct classification of 44 of 50 and 39 of 50 samples, respectively. Lymph node status was best predicted by BBN with 34 of 50 samples correctly classified, indicating a relationship between metabolic profile and lymph node status. Thus, MR profiles contain prognostic information that may be of benefit in treatment planning, and MR metabolomics may become an important tool for diagnosis of breast cancer patients.

Classification of wild and farmed salmon using bayesian belief networks and gas chromatography-derived fatty acid distributions.

In this study, we present the use of Bayesian Belief Networks (BBN) for the classification of wild versus farmed Atlantic salmon (Salmo salar L.). Using a data set of 131 salmon samples from several geographical origins and the gas chromatography-derived distributions of 12 fatty acids (FAs), a Bayesian Belief Network was constructed, ultimately using only the three most important FAs (16:1n-7, 18:2n-6, and 22:5n-3). The training data set yielded a prediction error of 0% (68/68 farmed; 20/20 wild correct) while the validation data set prediction error was 4.65% (32/32 farmed; 9/11 wild correct). Different randomly chosen validation sets yielded similar prediction accuracies. This model was then applied to 30 market (store-bought) samples where predictions were compared with the product labels.

Predicting survival and early clinical response to primary chemotherapy for patients with locally advanced breast cancer using DCE-MRI.

PURPOSE: To evaluate dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) as a tool for early prediction of response to neoadjuvant chemotherapy (NAC) and 5-year survival in patients with locally advanced breast cancer. MATERIALS AND METHODS: DCE-MRI was performed in patients scheduled for NAC (n = 24) before and after the first treatment cycle. Clinical response was evaluated after completed NAC. Relative signal intensity (RSI) and area under the curve (AUC) were calculated from the DCE-curves and compared to clinical treatment response. Kohonen and probabilistic neural network (KNN and PNN) analysis were used to predict 5-year survival. RESULTS: RSI and AUC were reduced after only one cycle of NAC in patients with clinical treatment response (P = 0.02 and P = 0.08). The mean and 10th percentile RSI values before NAC were significantly lower in patients surviving more than 5 years compared to nonsurvivors (P = 0.05 and 0.02). This relationship was confirmed using KNN, which demonstrated that patients who remained alive clustered in separate regions from those that died. Calibration of contrast enhancement curves by PNN for patient survival at 5 years yielded sensitivity and specificity for training and testing ranging from 80%-92%. CONCLUSION: DCE-MRI in locally advanced breast cancer has the potential to predict 5-year survival in a small patient cohort. In addition, changes in tumor vascularization after one cycle of NAC can be assessed.

Combining clinical assessment scores and in vivo MR spectroscopy neurometabolites in very low birth weight adolescents.

OBJECTIVE: Very low birth weight (VLBW) survivors are at increased risk of neurological impairments that may persist into adolescence and adulthood. The aims of this study were to identify the most important clinical assessments that characterize differences between VLBW and control adolescents, and to look at the relationship between clinical assessments and the metabolites in in vivo MR spectra. METHODS: At 14-15 years of age, 54 VLBW survivors and 64 term controls were examined clinically. Several neuropsychological and motor assessments were performed. The magnetic resonance (MR) brain spectra were acquired from volumes localized in the left frontal lobe and contained mainly white matter. RESULTS: Probabilistic neural networks and support vector machines demonstrated that clinical assessments rendered a possibility of the classification of VLBW versus control adolescents. The most important clinical assessments in this classification were visual-motor integration, motor coordination, stroop test, full scale IQ, and grooved pegboard. Through the use of outer product analysis-partial least squares discriminant analysis on a subset of adolescents (n=36), the clinical assessments found to most strongly correlate with the spectral data were the global assessment scale, Wisconsin card sorting test, full scale IQ, grooved pegboard test, and motor coordination test. Clinical assessments that relate to spectral data may be especially dependent on an intact microstructure in frontal white matter.

Characterization of bright tobaccos by multivariate analysis of 13C CPMAS NMR spectra.

Univariate and multivariate statistics were applied to characterize cured bright tobacco samples on the basis of their 13C CPMAS NMR spectra and leaf constituent analysis. NMR spectra were obtained for 55 samples selected from a set of 134 samples of graded bright tobacco leaves from crop year 1999. Historical leaf constituent analyses were available for total alkaloids, reducing sugars, total nitrogen, and insoluble ash. In addition, we applied HPLC to quantify the two abundant plant polyphenols, chlorogenic acid, and rutin. Principal component analysis (PCA) and partial least squares (PLS) of the NMR spectra revealed systematic relationships between groups of samples related to these substances and afforded predictive quantitative models for the analyzed constituents. Analysis of the PLS significant variables showed that leaf polysaccharides, alkaloids, and minerals are major determinants influencing the grading of cured bright tobacco leaves.

(13)C NMR pattern recognition techniques for the classification of Atlantic salmon (Salmo salar L.) according to their wild, farmed, and geographical origin.

(13)C nuclear magnetic resonance (NMR) in combination with multivariate data analysis was used to (1) discriminate between farmed and wild Atlantic salmon ( Salmo salar L.), (2) discriminate between different geographical origins, and (3) verify the origin of market samples. Muscle lipids from 195 Atlantic salmon of known origin (wild and farmed salmon from Norway, Scotland, Canada, Iceland, Ireland, the Faroes, and Tasmania) in addition to market samples were analyzed by (13)C NMR spectroscopy and multivariate analysis. Both probabilistic neural networks (PNN) and support vector machines (SVM) provided excellent discrimination (98.5 and 100.0%, respectively) between wild and farmed salmon. Discrimination with respect to geographical origin was somewhat more difficult, with correct classification rates ranging from 82.2 to 99.3% by PNN and SVM, respectively. In the analysis of market samples, five fish labeled and purchased as wild salmon were classified as farmed salmon (indicating mislabeling), and there were also some discrepancies between the classification and the product declaration with regard to geographical origin.

Simple dual tasking recruits prefrontal cortices in chronic severe traumatic brain injury patients, but not in controls.

The ability to carry out two tasks simultaneously, dual tasking, is specifically impaired after traumatic brain injury (TBI). The aim of the present study was to investigate the neuronal correlates to this increased dual cost in chronic severe TBI patients (n = 10) compared to healthy controls (n = 11) using functional magnetic resonance imaging (fMRI) at 3 Tesla (T). The tasks were a visual search and a simple two-fingers button press motor task. Performance data demonstrated similar and significant dual task interference in both TBI patients and controls using a linear mixed model. However, principal component analysis showed that TBI patients and controls could be classified into different categories based on motor activity in the single compared to the dual task condition, thus reflecting the increased variability in the performance in the TBI group. Random effects between-group analysis demonstrated significantly reduced activation in the TBI group in both single task conditions in the occipital and posterior cingulate cortices, and for the visual task also in the thalami. This pattern was reversed in the dual task condition with significantly increased activation of a predominantly left lateralized prefrontal-anterior midline-parietal network in the TBI group compared to the controls. The increase in activation occurred within regions described to be engaged in healthy volunteers as dual task cost increases. This finding points to substitution, functional reorganization within the primary network subserving the task, following TBI, and demonstrates more effortful processing. Recruitment of these additional prefrontal resources may be connected to serial rather than parallel processing in low level dual tasking in TBI. Thus, in severe TBI, low level dual task performance depends on increased attentional and executive guidance.

Water distribution in brine salted cod (Gadus morhua) and salmon (Salmo salar): a low-field 1H NMR study.

Low-field (LF) (1)H NMR T 2 relaxation measurements were used to study changes in water distribution in lean (Atlantic cod) and fatty (Atlantic salmon) fish during salting in 15% NaCl and 25% NaCl brines. The NMR data were treated by PCA, continuous distribution analysis, and biexponential fitting and compared with physicochemical data. Two main water pools were observed in unsalted fish, T 21, with relaxation times in the range 20-100 ms, and T 22, with relaxation times in the range 100-300 ms. Pronounced changes in T 2 relaxation data were observed during salting, revealing changes in the water properties. Salting in 15% brine lead to a shift toward longer relaxation times, reflecting increased water mobility, whereas, salting in saturated brines had the opposite effect. Water mobility changes were observed earlier in the salting process for cod compared to salmon. Good linear correlations ( F

Classification of gilthead sea bream (Sparus aurata) from 1H NMR lipid profiling combined with principal component and linear discriminant analysis.

The combination of (1)H NMR fingerprinting of lipids from gilthead sea bream (Sparus aurata) with nonsupervised and supervised multivariate analysis was applied to differentiate wild and farmed fish and to classify farmed specimen according to their areas of production belonging to the Mediterranean basin. Principal component analysis (PCA) applied on processed (1)H NMR profiles made a clear distinction between wild and farmed samples. Linear discriminant analysis (LDA) allowed classification of samples according to the geographic origin, as well as for the wild and farmed status using both PCA scores and NMR data as variables. Variable selection for LDA was achieved with forward selection (stepwise) with a predefined 5% error level. The methods allowed the classification of 100% of the samples according to their wild and farmed status and 85-97% to geographic origin. Probabilistic neural network (PNN) analyses provided complementary means for the successful discrimination among classes investigated.

High-resolution (13)C nuclear magnetic resonance spectroscopy pattern recognition of fish oil capsules.

13C NMR (nuclear magnetic resonance) spectroscopy, in conjunction with multivariate analysis of commercial fish oil-related health food products, have been used to provide discrimination concerning the nature, composition, refinement, and/or adulteration or authentication of the products. Supervised (probabilistic neural networks, PNN) and unsupervised (principal component analysis, PCA; Kohonen neural networks; generative topographic mapping, GTM) pattern recognition techniques were used to visualize and classify samples. Simple PCA score plots demonstrated excellent, but not totally unambiguous, class distinctions, whereas Kohonen and GTM visualization provided better results. Quantitative class predictions with accuracies >95% were achieved with PNN analysis. Trout, salmon, and cod oils were completely and correctly classified. Samples reported to be salmon oils and cod liver oils did not cluster with true salmon and cod liver oil samples, indicating mislabeling or adulteration.

MR-determined metabolic phenotype of breast cancer in prediction of lymphatic spread, grade, and hormone status.

The purpose of the study was to evaluate the use of metabolic phenotype, described by high-resolution magic angle spinning magnetic resonance spectroscopy (HR MAS MRS), as a tool for prediction of histological grade, hormone status, and axillary lymphatic spread in breast cancer patients. Biopsies from breast cancer (n = 91) and adjacent non-involved tissue (n = 48) were excised from patients (n = 77) during surgery. HR MAS MR spectra of intact samples were acquired. Multivariate models relating spectral data to histological grade, lymphatic spread, and hormone status were designed. The multivariate methods applied were variable reduction by principal component analysis (PCA) or partial least-squares regression-uninformative variable elimination (PLS-UVE), and modelling by PLS, probabilistic neural network (PNN), or cascade correlation neural network. In the end, model verification by prediction of blind samples (n = 12) was performed. Validation of PNN training resulted in sensitivity and specificity ranging from 83 to 100% for all predictions. Verification of models by blind sample testing showed that hormone status was well predicted by both PNN and PLS (11 of 12 correct), lymphatic spread was best predicted by PLS (8 of 12), whereas PLS-UVE PNN was the best approach for predicting grade (9 of 12 correct). MR-determined metabolic phenotype may have a future role as a supplement for clinical decision-making-concerning adjuvant treatment and the adaptation to more individualised treatment protocols.

Bioactive compounds in cod (Gadus morhua) products and suitability of 1H NMR metabolite profiling for classification of the products using multivariate data analyses.

This work investigates the suitability of (1)H NMR spectroscopy to identify the fate of some bioactive compounds in seafood submitted to several processing conditions and examines the possibility of using (1)H NMR spectroscopy profiling to classify such products. Perchloric acid extracts of cod white muscle from newly killed and (i) unprocessed, (ii) boiled, and (iii) fried fillets and from (iv) frozen fillets, (v) the frozen fillets after thawing, and (vi) their drip loss and from (vii) rehydrated cod klippfish (n = 5) were analyzed by 500 MHz (1)H NMR spectroscopy. It was possible to identify taurine, betaine, anserine, creatine, and trimethylamine oxide (TMAO) in all extracts examined, and frozen fish was recognizable by the presence of dimethylamine (DMA). None of the heating procedures seemed to induce the loss of bioactive compounds from the fillet, but freezing and thawing did: the compounds were lost in what is known as drip loss. About 80% of the samples were correctly classified using a probabilistic neural network procedure having as inputs the scores of the first 20 principal components of the principal component analysis of a selected region of the NMR spectra.