BREAST CANCER BIOMARKERS AND IDENTIFICATION METHODS USING NMR AND GAS CHROMATOGRAPHY-MASS SPECTROMETRY

[0001] This invention was made in part with government support under grant reference number NIH/NIDDK 3 R21 DK070290-01 awarded by the National Institutes of Health ("NIH") Roadmap Initiative on Metabolomics Technology. The Government has certain rights in the invention. RELATED APPLICATIONS

[0002] This application claims priority to U.S. Provisional Patent Application

Serial No. 60/977,445 filed October 4, 2007, the disclosure of which is expressly incorporated by reference herein. TECHNICAL FIELD

[0003] The present invention generally relates to small molecule biomarkers comprising a set of metabolite species useful for the early detection of breast cancer, including methods for identifying such biomarkers within biological samples by using a process that combines gas chromatography-mass spectrometry and nuclear magnetic resonance spectrometry. BACKGROUND

[0004] Current breast cancer detection methods often involve mammographic examinations, followed by biopsy procedures. However, mammographies often produce inaccurate results and thereby force many women to undergo unnecessary biopsies, which can be both painful and expensive. To combat the problems related to poor diagnostic accuracy for a number of diseases, research efforts have recently focused on metabolomics to diagnosis diseases through the identification of various biomarkers. ¹ ' ² Metabolomics, which combines high resolution chemical analysis with multivariate statistics, provides a means to identify a subset of metabolites that differentiate sample populations, as well as gives detailed information regarding biochemical status changes. ³ ' ⁴ A variety of standard analytical instruments have been utilized for metabolomics applications, including Fourier transform infrared ("FTIR") spectroscopy, nuclear magnetic resonance ("NMR") spectroscopy, liquid chromatography ("LC") and gas chromatography coupled mass spectrometry ("GC- MS"). ⁵ ' ^6"8 ' ⁹ ' ¹⁰ Moreover, several atmospheric ionization mass spectrometry ("MS") techniques have also been applied to metabolomics-based studies, including desorption electrospray ionization mass spectrometry ("DESI-MS"), ¹¹ extractive electrospray

ionization mass spectrometry ("EESI-MS") and direct analysis in real time ("DART") ^12"15

[0005] Among the analytical techniques employed in metabolomics research,

NMR has been found to be a useful quantitative and reproducible tool for the analysis of biofluids. NMR has been utilized in a number of studies for detecting diseases and identifying putative biomarkers. ^15"18 However, because NMR has relatively poor analytical sensitivity, it is often unable to detect species found in low concentrations within the biofluids. As such, in addition to simple ID ¹ H-NMR, more advanced NMR approaches, such as two-dimensional ("2D") J-resolved spectroscopy and selective total correlation spectroscopy ("TOCSY") experiments, have been explored as alternative analytical techniques within the area of metabolomics research. ^19'21 [0006] Research has shown that chromatographic methods are ideal compliments to NMR-based metabolomics. ¹³ ' ¹⁵ ' ²² In particular, GC-MS has been used for metabolic profiling for over 30 years. ²³ ' ²⁴ GC-MS provides a sensitive and reasonably reproducible analytical platform for analysis. ²⁵ For example, GC-MS has recently been coupled with multivariate analysis for the differentiation of extracts from Arabidopsis plants to explore the effect of silent plant genotypes. ²⁶ ' ²⁷ However, a major disadvantage of GC-MS is its limited ability to resolve and/or even detect many types of metabolites. This is problematic, particularly because complex biological samples often contain thousands of metabolites. Recently, comprehensive two- dimensional GC has been coupled with time-of-flight mass spectrometry (TOF-MS) for the analysis of complex mixtures. ^29"31 The advantage of two-dimensional gas chromatography ("2D GC" or "GCxGC") is its ability to add an additional and complementary second separation to the analysis. Coupled with TOF-MS, 2D GC is capable of identifying exact masses of many compounds within complex mixtures. For instance, Mohler and coworkers have used this technique to analyze metabolites in fermenting and respiring yeast. ³² Principal component analysis ("PCA") and parallel factor analysis ("PARAFAC") ³³ were applied for the purposes of classification and quantification, respectively. The Synovec group has also developed different algorithms for the analysis of the multi-way data generated from GCxGC-MS including the Fisher ratio method, peak alignment, and the DotMap algorithm. ^34"3 [0007] The combination of complementary analytical techniques in metabolomics research opens a number of new opportunities. ¹³ ' ¹⁵ ' ²² The development of

new and emerging technologies in metabolomics has proven to be powerful and promising in a number of cases. For NMR and MS, the combination of two essentially orthogonal analytical techniques, both having extremely high resolving power and the unequivocal ability to identify unknown metabolites, also provides a powerful approach for metabolomics research.

[0008] Data from NMR and MS experiments are generally complex since they contain qualitative/quantitative information on upwards of several hundreds of metabolites. Multivariate statistical analyses are thus used for data reduction and in particular for differentiating biofluids samples into "disease" and "control" populations based on the differences in signals of multiple metabolites. A variety of multivariate statistical methodologies provide extremely helpful tools for filtering the large amounts of data and for accessing the often-subtle biochemical perturbations latent in the spectra. In addition, these approaches can be used to extract sets of biomarker metabolites that have the best properties for the assessment of disease status. However, the use of a single analytical method to uncover useful biomarkers for early disease detection, and in particular early breast cancer detection has so far been unsuccessful. [0009] The present invention is intended to address and/or to improve upon one or more of the problems discussed above. SUMMARY OF THE INVENTION

[0010] The present teachings are generally related to biological sample classification methods, which utilize a combination of MS and NMR techniques. More particularly, the present teachings take advantage of the combination of NMR and two- dimensional gas chromatography-mass spectrometry ("GCxGC-MS") to identify small molecule biomarkers comprising a set of metabolite species. These identified biomarkers have been found to be useful for detecting early forms of breast cancer. [0011] Mass and NMR spectra generated by the present methods can be analyzed with an advanced multivariate statistical pattern recognition approach, which allows sample differentiation by combining many metabolites in parallel. More particularly, the present teachings allow multiple metabolite species to be measured in parallel by using a combination of GCxGC-MS and NMR. The use of two analytical diagnostic methods provides better coverage of possible metabolites in the body, and assists with the cross-validation of the identified biomarkers.

[0012] According to one exemplary embodiment in accordance with present teachings, GCxGC-MS and NMR spectroscopy methods are combined to analyze serum samples from breast cancer patients and healthy controls. The spectral results are then evaluated using multivariate analysis tools, such as PCA (principal component analysis). Compounds showing significantly altered concentrations in breast cancer samples are identified and compared to healthy controls using an established library of exact mass (time of flight, TOF) data and then confirmed using authentic compounds. ¹ H-NMR and statistical analysis of the same samples can then be used to produce additional molecules of interest, as well as used to classify patients in classes, such as patients having "breast cancer" and patients that are "healthy." Moreover, profiles from the NMR analysis are comparable and can be combined and statistically correlated with the GCxGC-MS results. According to this exemplary method, perturbations in the glycerophospholipid metabolism can be implicated as being related to carcinogenesis.

[0013] According to another exemplary embodiment, a method for the parallel identification of one or more metabolite species within a biological sample is provided. The method comprises producing a first spectrum by subjecting the sample to a nuclear magnetic resonance analysis. The first spectrum contains individual spectral peaks that are representative of the one or more metabolite species contained within the sample. A second spectrum is also produced by subjecting the sample to a mass spectrometry analysis. The second spectrum also contains individual spectral peaks representative of the one or more metabolite species contained within the sample. Each of the individual spectral peaks is subjected to a statistical pattern recognition analysis to identify the one or more metabolite species contained within the sample. To identify the one or more metabolite species contained within the sample, the individual spectral peaks of the mass and nuclear magnetic resonance spectra are then analyzed. [0014] In yet another exemplary embodiment, a method for the parallel identification of one or more metabolite species within a biological sample, comprises producing a first spectrum by subjecting the sample to a nuclear magnetic resonance analysis, the first spectrum containing individual spectral peaks representative of the one or more metabolite species contained within the sample; producing a second spectrum by subjecting the sample to a two-dimensional gas chromatography coupled mass spectrometry analysis, the second spectrum containing individual spectral peaks

representative of the one or more metabolite species contained within the sample; performing a principle component analysis on each of the individual spectral peaks of the first and second spectra to identify the one or more metabolite species contained within the sample; and assigning the sample into a defined sample class after the one or more metabolite species within the sample are identified.

[0015] In still yet another exemplary embodiment, a method for detecting breast cancer status within a biological sample comprises measuring one or more metabolite species within the sample by subjecting the sample to a combined nuclear magnetic resonance and mass spectrometry analysis, the analysis producing a spectrum containing individual spectral peaks representative of the one or more metabolite species contained within the sample; subjecting the individual spectral peaks to a statistical pattern recognition analysis to identify the one or more metabolite species contained within the sample; correlating the measurement of the one or more metabolite species with a breast cancer status. According to this exemplary embodiment, the one or more metabolite species are selected from the group consisting of alpha-hydroxy acid, hexadecanoic acid, vaccenic acid, octadecanoic acid, lipid-CH2 signals, glycerol, lactic acid and glucose.

[0016] In yet another exemplary embodiment, a small molecule biomarker for detecting breast cancer status within a biological sample is provided. In accordance with this embodiment, the biomarker comprises one or more metabolite species selected from the group consisting of alpha-hydroxy acid, hexadecanoic acid, vaccenic acid, octadecanoic acid, lipid-CH2 signals, glycerol, lactic acid and glucose. The one or more metabolite species are measurable by a combined nuclear magnetic resonance and mass spectrometry analysis, as well as identifiable by a statistical pattern recognition analysis.

BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above-mentioned aspects of the present teachings and the manner of obtaining them will become more apparent and the teachings will be better understood by reference to the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0018] Figure l(a) shows GC/GC/MS spectra of serum samples from a patient having breast cancer in accordance with the present teachings;

[0019] Figure l(b) shows GC/GC/MS spectra of serum samples from a healthy patient;

[0020] Figure 2 shows PCA score plots from a multivariate analysis of

GC/GC/MS measurements in accordance with the present teachings;

[0021] Figure 3 (a) shows a 2D representation of the PCl loading of exemplary

GCxGC-MS data in accordance with the present teachings;

[0022] Figure 3(b) shows a box plot of five biomarker candidates from PCA loading;

[0023] Figure 4a shows ID NMR spectra of a normal sample in accordance with the present teachings;

[0024] Figure 4b shows ID NMR spectra of a breast cancer sample in accordance with the present teachings;

[0025] Figure 5 shows a PCA score plot in accordance with the present teachings; and

[0026] Figure 6 shows a 3D score plot combining two PC's from NMR and one

PC from GCxGC-MS in accordance with the present teachings.

DETAILED DESCRIPTION

[0027] The embodiments of the present invention described below are not intended to be exhaustive or to limit the invention to the precise forms disclosed in the following detailed description. Rather, the embodiments are chosen and described so that others skilled in the art may appreciate and understand the principles and practices of the present invention.

[0028] The present teachings are generally related to the combination of

GCxGC-MS and NMR for the metabolomics-based analysis of human serum samples from breast cancer patients and matched healthy controls. After GCxGC-MS and NMR measurements, a multivariate statistical method, such as PCA, is applied to reduce the data set size and complexity. While both GCxGC-MS and NMR independently have the ability to differentiate healthy and cancer patients using serum samples, the present inventors have found that combining the score plots of these two techniques results in an improved classification system. More particularly, the present method allows GCxGC-MS spectra to be statistically correlated with NMR using the statistical heterospectroscopy ("SHY") method ³⁷ , thereby allowing important metabolites that correlated with one another across the two spectroscopic methods to be identified. [0029] Advantages and improvements of the processes and methods of the present invention are demonstrated in the following example. This example is

illustrative only and is not intended to limit or preclude other embodiments of the invention.

[0030] Commercial human blood serum samples from breast cancer patients

(n=20) and gender and age-matched healthy controls (n=20) were purchased from Asterand (Detroit, MI). All the serum samples were obtained from female volunteers of ages ranging from 40 to 75 years old. Samples were stored at -80 ⁰ C until the measurements were made.

[0031] NMR measurements were performed on a Bruker DRX 500 MHz spectrometer equipped with a room temperature HCN probe. Samples were prepared by mixing 400 μl serum with a 300 μl solution of 1.5 mM 3-(trimethylsilyl) propionic- (2,2,3,3-d4) acid sodium salt (TSP) in D ₂ O. TSP was used as the frequency standard (δ=0.00). After vortexing and centrifugation, 580 μl aliquots were transferred to standard 5mm NMR tubes for analysis. Samples were measured using a standard ID CPMG (Carr-Purcell-Meiboom-Gill) pulse sequence coupled with water presaturation. For each spectrum, 32 transients were collected resulting in 32k data points using a spectral width of 6000 Hz. An exponential weighting function corresponding to 0.3 Hz line broadening was applied to the free induction decay (FID) before Fourier transformation. After phasing and baseline correction using Bruker' s XWINNMR software, the processed data were saved in ASCII format for further multivariate analysis. The spectral region from 4.5 to 6 ppm, which contains water and urea signals, was removed from each spectrum prior to data analysis.

[0032] For GCxGC-MS measurements, protein precipitation was first performed for each sample by mixing 200 μL serum with 200 μL methanol in a 1.5 mL Eppendorf tube. The mixture was incubated on a thermomixer at 33 ⁰ C for 15 min and then centrifuged at 15000 rpm for 5 min. The upper portion was transferred into another Eppendorf tube for future use. 200 μL chloroform was mixed with the precipitate for 1 hour at room temperature. After centrifugation, the aliquot was transferred and combined with the methanol solution from the previous step. The mixture was then lyophilized to remove the organic solvents. Each dried sample was dissolved in 100 μL pyridine, and 20 μL of this solution was mixed with 10 μL ethylhydroxylamine on a thermomixer at 60 ⁰ C for 30 min. 20 μL of the derivatization reagent MTBSTFA (N-Methyl-N-(tert-butyldimethylsilyl) trifluoroacetamide) was then added to the mixture on the same thermomixer for 60 min, resulting in the addition of a

tert-butyldimethylsilyl group to active functional groups such as the hydroxyl, amino or carboxylic groups.

[0033] Derivatized samples were then transferred to the LECO Pegasus 4D

GCxGC-MS instrument (St. Joseph, MI) for measurements. The automated LECO system facilitates a two dimensional GC capacity using a cryo-trapping unit between the two chromatographic columns. The first dimension chromatographic column was a 10 m DB-5 capillary column with an internal diameter of 180 μm and a film thickness of 0.18 μm, and the second dimension chromatographic column was a 1 m DB- 17 capillary column with an internal diameter of 100 μm and a film thickness of 0.1 μm. High purity helium was used as carrier gas at a flow rate of 1.0 ml/min. The first dimension column oven ramp began at 5O ⁰ C with a hold time of 0.2 min, then increased to 300°C at a rate of 10°C/min and held at this temperature for 5 min. The second dimension column oven began 2O ⁰ C higher than the corresponding first dimension column oven ramp with the same program rate and hold time. The second dimension separation cycle time was set for 4 seconds. 2 μL of the derivatized solutions was injected in a split mode with a ratio of 20:1. The temperatures for the inlet and transfer line were set at 280°C, and the ion source was set at 200°C. The detection and filament bias voltages were set at 1600 V and -70 V, respectively. Mass spectra ranging from 50 to 900 m/z were collected at a rate of 50 Hz, and each sample was measured twice to assure the reproducibility of the instrument. The NIST MS database (NIST MS Search 2.0, NIST/EPA/NIH Mass Spectral Library; NIST 2002) was used for data processing and peak matching.

[0034] All pre-processing and multivariate analyses of the experimental data were carried out using Matlab 7.1.0.246 (Mathworks Inc., Natick, MA) with the PLS toolbox (Eigenvector Research Inc, Wenatchee, WA). Both GCxGC-MS and NMR data were transferred from the instruments in plain text format and then imported into Matlab. Three-way GCxGC-MS data was composed of the first dimension GC retention time (in sec), the second dimension GC retention time (in sec) and corresponding TIC (total ion current). GCxGC-MS data were unfolded by inserting the second dimension data into the first dimension to form a two-column dataset. Prior to data analysis, the spectral region below 252 sec, which contains large peaks from pyridine, tert-butyldimethylsilanol and 2,2,2-trifluoro-N-methyl-acetamide, was removed. In addition, the data containing the tert-butyldimethylsilanol background

peak eluting in the second GC column at about 0.6 sec was also excluded from the data analysis.

[0035] NMR spectra were binned to 1000 frequency buckets of equal width in order to remove the errors resulting from the small fluctuations of chemical shifts due to pH or ion concentration variations while the GCxGC-MS spectra were kept at full resolution. Both NMR and GCxGC-MS data were normalized against the total spectral intensity and then mean-centering was carried out prior to multivariate analysis. PCA was all performed using the PLS toolbox. For GCxGC-MS results, the corresponding loading or weight plot was folded back into the 3 -way presentation format for visualization.

[0036] Typical GCxGC-MS spectra derived from the serum of a cancer patient and a healthy volunteer are shown in Figure Ia and Ib, respectively. The three- dimensional GCxGC-MS spectra are shown in the bottom of each panel, while the corresponding contour plots are shown in the upper panels. Compared with the chromatogram resulting from the first column (shown in white in the three-dimensional spectra), the unique feature of GCxGC-MS is clearly the much improved resolution due to the separation on the second GC column. More than 800 compounds could be identified in these spectra by searching against the NIST database. This complexity made it impossible to differentiate the spectra visually, and thus multivariate analysis was applied.

[0037] PCA was first applied to the 40 GCxGC-MS spectra. Original spectra were projected onto the two-dimensional space of the first two principal components, PCl and PC2. The classification between breast cancer and normal samples is clearly shown in the score plot (Figure 2), where circles indicate 95% confidence limits. Separation between the two sample groups is along PCl while the variation within one sample group is primarily along PC2.

[0038] In the PCA results, breast cancer and normal samples are separated mostly along the first principal component (PCl). The weight of PCl is shown in Figure 3 a, in which orange and blue colors indicate up- and down-regulation, respectively, of metabolite concentrations observed in the breast cancer samples, compared to the normal samples. Identification of the seven most significant peaks was carried out by the NIST database search. Six of the compounds, which are observed to be the most altered in concentration in the cancer samples, are listed in Table 1 , along

with calculated /7-values from the Student's t-test. These p- values are shown before any Bonferroni correction. ⁴⁰ A very conservative estimate of this correction would be 1000, since approximately 800 compounds were identified in the serum samples. Five of these 6 compounds would remain significant or very close to significant (p<0.05) even after a Bonferroni correction of this magnitude, and the corresponding box plot indicating the altered concentrations of these five compounds is shown in Figure 3b. The identities of the six compounds were confirmed by running GCxGC-MS experiments of authentic (commercial) compounds under the same preparation, derivatization and acquisition conditions. [0039]

Table 1: Compounds identified from the PCA of GCxGC-MS Data.

[0040] Among these compounds, a number of alpha-hydroxy acids (AHA) have been reported to be related to the development of skin cancer by affecting the expression of epidermal growth factor receptor (EGFR). ⁴¹ Fatty acids, glycerol and cholesterol are also reported to be general markers for cancer growth. Among them, hexadecanoic acid (palmitic acid), which is up-regulated in breast cancer samples, has been reported to induce cancer cell death. ⁴² ' ⁴³ More generally, it has also been reported that unsaturated acids stimulate the proliferation of human cancer cells and saturated acids inhibit it by causing apoptosis. Certainly, there are more species, which may also explain the difference between breast cancer and normal samples by comparing the

library search and PCA loading; however, even using just these six compounds as a starting point for discrimination yields high classification accuracy. [0041] In addition to the exemplary compounds listed above, it should be understood and appreciated herein that other metabolite species useful as biomarkers may also be identified in accordance with the present invention. Some of these additional metabolite species include, but are not limited to, lipid-CH2 signals (which are detectable by NMR), as well as lactic acid and glucose.

[0042] NMR has been one of the major techniques in metabolomics research because of its reproducible and quantitative features. Figures 4a and 4b compare typical ID CPMG spectra of breast cancer and normal samples, where only the aliphatic region is shown and compared (the intensities of the peaks in the aromatic region were very small). From visual inspection, a number of peaks appear to be clearly different between the cancer and normal samples. For example, the glucose level is decreased in the cancer sample, while the intensity of the doublet belonging to lactate at 1.34 ppm is significantly larger. Several of the broad lipid signals also have higher intensities in the NMR spectra of cancer samples. PCA was applied to the 40 NMR spectra and the score plot is shown in Figure 5. All the samples are classified by projecting corresponding NMR spectra onto the 2D plane of PCl and PC2 where ellipses indicate 95% confidence limits. Using a linear discrimination, a discriminant accuracy of approximately 90% was achieved. In contrast to the results of the GCxGC-MS, the separation occurs along the first two PC's, primarily along PC2, which showed 32% of the variance in the NMR data. Several putative biomarkers responsible for the separation were identified from PCA. The levels of lactate (1.33 ppm) and glucose (3.53, 3.73, 3.86ppm) are observed to increase in breast cancer patients. In addition, lipid signals at 0.92 and 1.31ppm are also more concentrated in the serum samples from cancer patients. The PCl loading is mainly contributed by the variation within the breast cancer patients, for instance, the concentration of sugars and lipids. [0043] Previously conducted research ¹³ has suggested an approach to combine the PCA scores from two independent analytical techniques, DESI-MS and NMR spectroscopy, when the 2D score plot is insufficient to differentiate the samples. As shown in Figure 5, PCl score values from the GCxGC-MS PCA as the third dimension of the 3D score plot were used. Only a single GCxGC-MS measurement for each sample was included in the plot. The classification accuracy of NMR PCA can be

improved by combining with GCxGC-MS resulting in all the samples being correctly classified. This combined approach indicates that inclusion of a second orthogonal analytical technique can be useful for improving the discrimination of similar samples. [0044] The combination of GCxGC-MS and NMR spectroscopy, using a metabolomics approach, provides a novel and powerful approach for analyzing serum samples from breast cancer patients and healthy controls. Advanced GCxGC-MS possesses the ability to resolve and identify many more metabolites than conventional GC-MS. A good discrimination can be achieved from GCxGC-MS measurements after multivariate analysis. NMR spectroscopy also shows the capability of differentiating breast cancer from healthy control samples. More importantly, the composite PCA score plot of GCxGC-MS and NMR data improves the accuracy of classification. [0045] While an exemplary embodiment incorporating the principles of the present invention has been disclosed hereinabove, the present invention is not limited to the disclosed embodiments. Instead, this application is intended to cover any variations, uses, or adaptations of the invention using its general principles. Further, this application is intended to cover such departures from the present disclosure as come within known or customary practice in the art to which this invention pertains and which fall within the limits of the appended claims.

[0046] REFERENCES: The following references are incorporated herein by reference in their entirety:

(1) Lindon, J. C; Holmes, E.; Nicholson, J. K. Pharm. Res. 2006, 23, 1075-1088.

(2) Lindon, J. C; Holmes, E.; Nicholson, J. K. FEBS. 2007, 274, 1140-1151.

(3) Remer, T.; Boye, K. R.; Hartmann, M. F.; Wudy, S. A. J. Clin. Endocr. Met. 2005, 90, 2015-2021.

(4) Valongo, C; Cardoso, M. L.; Domingues, P.; Almeida, L.; Verhoeven, N.; Salomons, G.; Jakobs, C; Vilarinho, L. Clin. Chim. Acta 2004, 348, 155-161.

(5) Oliver, S. G.; Winson, M. K.; KeIl, D. B.; Baganz, F. Trends Biotech. 1998, 16, 373-378.

(6) Beckwith-Hall, B. M.; Nicholson, J. K.; Nicholls, A. W.; Foxall, P. J. D.; Lindon, J. C; Connor, S. C; Abdi, M.; Connelly, J.; Holmes, E. Chem. Res. Toxicol. 1998, 11, 260-272.

(7) Belton, P. S.; Colquhoun, I. J.; Kemsley, E. K.; Delgadillo, I.; Roma, P.; Dennis, M. J.; Sharman, M.; Holmes, E.; Nicholson, J. K.; Spraul, M. Food Chem. 1998, 67, 207-213.

(8) Gibb, J. O. T.; Holmes, E.; Nicholson, J. K.; Weeks, J. M. Comp. Biochem. Physiol. B-Biochem. MoI. Biol. 1997, 118, 587-598.

(9) von Roepenack-Lahaye, E.; Degenkolb, T.; Zerjeski, M.; Franz, M.; Roth, U.; Wessjohann, L.; Schmidt, J.; Scheel, D.; Clemens, S. Plant Physiol. 2004, 134, 548-559.

(10) Zywicki, B.; Catchpole, G.; Draper, J.; Fiehn, O. Anal. Biochem. 2005, 336, 178-186.

(11) Takats, Z.; Wiseman, J. M.; Gologan, B.; Cooks, R. G. Science 2004, 306, 471- 473.

(12) Cody, R. B.; Laramee, J. A.; Durst, H. D. Anal. Chem. 2005, 77, 2297-2302.

(13) Chen, H. W.; Pan, Z. Z.; Talaty, N.; Raftery, D.; Cooks, R. G. Rapid Comm. Mass Spectrom. 2006, 20, 1577-1584.

(14) Gu, H.; Pan, Z.; Zhang, J.; Raftery, D. In preparation 2007.

(15) Pan, Z.; Gu, H.; Talaty, N.; Chen, H. W.; Hainline, B. E.; Cooks, R. G.; Raftery, D. Anal. Bioanal. Chem. 2007, 387, 539-549.

(16) Brindle, J. T.; Antti, H.; Holmes, E.; Tranter, G.; Nicholson, J. K.; Bethell, H. W. L.; Clarke, S.; Schofield, P. M.; McKilligin, E.; Mosedale, D. E.; Grainger,

D. J. Nature Med. 2002, 8, 1439-1444.

(17) Beger, R. D.; Schnackenberg, L. K.; Holland, R. D.; Li, D.; Dragan, Y. Metabolomics 2006, 2, 125-134.

(18) Odunsi, K.; Wollman, R. M.; Ambrosone, C. B.; Hutson, A.; McCann, S. E.; Tammela, J.; Geisler, J. P.; Miller, G.; Sellers, T.; Cliby, W.; Qian, F.; Keitz, B.; Intengan, M.; LeIe, S.; Alderfer, J. L. Int. J. Cancer 2005, 113, 782-788.

(19) Sandusky, P.; Raftery, D. Anal. Chem. 2005, 77, 7717-7723.

(20) Sandusky, P.; Raftery, D. Anal. Chem. 2005, 77, 2455-2463.

(21) Viant, M. R. Biochem. Biophys. Res. Comm. 2003, 310, 943-948.

(22) Pan, Z. Z.; Raftery, D. Anal, and Bioanal. Chem. 2007, 357, 525-527.

(23) Pauling, L.; Robinson, A. B.; Teranish.R; Cary, P. Proc. Nat. Acad. Sci. U.S.A 1971, 68, 2374-2376.

(24) Schulten, H. R.; Beckey, H. D.; Meuzelaa.Hl; Boerboom, A. J. Anal. Chem.1973, 45, 191-195.

(25) van der Greef, J.; Smilde, A. K. J. Chemom. 2005, 19, 376-386.

(26) Weckwerth, W.; Loureiro, M. E.; Wenzel, K.; Fiehn, O. Proc. Nat. Acad. Sci. U.S.A 2004, 707, 7809-7814.

(27) Weckwerth, W.; Wenzel, K.; Fiehn, O. Proteomics 2004, 4, 78-83.

(28) Bino, R. J.; Hall, R. D.; Fiehn, O.; Kopka, J.; Saito, K.; Draper, J.; Nikolau, B. J.; Mendes, P.; Roessner-Tunali, U.; Beale, M. H.; Trethewey, R. N.; Lange, B. M.; Wurtele, E. S.; Sumner, L. W. Trends Plant Sci. 2004, 9, 418-425.

(29) Pierce, K. M.; Hope, J. L.; Hoggard, J. C; Synovec, R. E. Talanta 2006, 70, 797-804.

(30) Sinha, A. E.; Hope, J. L.; Prazen, B. J.; Fraga, C. G.; Nilsson, E. J.; Synovec, R.

E. J. Chrom.. A 2004, 1056, 145-154.

(31) Sinha, A. E.; Hope, J. L.; Prazen, B. J.; Nilsson, E. J.; Jack, R. M.; Synovec, R. E. J. Chrom. A 2004, 1058, 209-215.

(32) Mohler, R. E.; Dombek, K. M.; Hoggard, J. C; Young, E. T.; Synovec, R. E. Anal. Chem. 2006, 78, 2700-2709.

(33) Smilde, A. K.; Doornbos, D. A. J. Chemom.1991, 5, 345-360.

(34) Pierce, K. M.; Hoggard, J. C; Hope, J. L.; Rainey, P. M.; Hoofnagle, A. N.; Jack, R. M.; Wright, B. W.; Synovec, R. E. Anal. Chem. 2006, 78, 5068-5075.

(35) Hope, J. L.; Sinha, A. E.; Prazen, B. J.; Synovec, R. E. J.Chrom. A 2005, 1086, 185-192.

(36) Pierce, K. M.; Wood, L. F.; Wright, B. W.; Synovec, R. E. Anal. Chem. 2005,

77, 1135-1141.

(37) Crockford, D. J.; Holmes, E.; Lindon, J. C; Plumb, R. S.; Zirah, S.; Bruce, S. J.; Rainville, P.; Stumpf, C. L.; Nicholson, J. K. Anal. Chem. 2006, 78, 363-371.

(38) Trygg, J. PhD Thesis, Umea University, 2001.

(39) Rubingh, C. M.; Bijlsma, S.; Derks, E. P. P. A.; Bobeldijk, I.; Verheij, S.; Kochhar, S.; Smilde, A. K. Metabolomics 2006, 2, 53-61.

(40) Shaffer, J. P. Ann. Rev. Psych. 1995, 46, 561-584.

(41) Henson, E. S.; Gibson, S. B. Cell. Sig. 2006, 18, 2089-2097.

(42) El-Assaad, W.; Buteau, J.; Peyot, M. L.; Nolan, C; Roduit, R.; Hardy, S.; JoIy, E.; Dbaibo, G.; Rosenberg, L.; Prentki, M. Endocrin. 2003, 144, 4154-4163.

(43) Hardy, S.; El-Assaad, W.; Przybytkowski, E.; JoIy, E.; Prentki, M.; Langelier, Y. J.Biol. Chem. 2003, 278, 31861-31870.

(44) Xu, R. H.; Pelicano, H.; Zhou, Y.; Carew, J. S.; Feng, L.; Bhalla, K. N.; Keating, M. J.; Huang, P. Cancer Res. 2005, 65, 613-621.

(45) Glunde, K.; Serkova, N. J. Pharmacogenomics 2006, 7, 1109-1 123.