Discrimination of healthy and neoplastic human colon tissues by ex-vivo HR-MAS NMR spectroscopy and chemometric analysis.

Colorectal cancer is the fourth most common cancer with high morbidity and mortality. Here, we report a spectroscopic study of healtny and neoplastic colon tissues in combination with multivariate analysis with the aim to find metabolic markers differentiating colon tumors. The characterization of the metabolic profiles was performed through High Resolution Magic-Angle Spinning (HR-MAS) NMR spectroscopy on biopsies or chirurgical samples of human colorectal tissues. Principal Component Analysis (PCA) was carried out as explorative analysis in order to obtain an overview on the whole data set without forcing any model and to extract relevant information. Partial Least Squares-Discriminant Analysis (PLS-DA) was used to build a classification model able to separate the classes of healthy, macroscopically normal and neoplastic human colon tissues based on their ex vivo HR-MAS NMR spectra. Samples of 23 patients, 14 with colon cancer, and 9 healthy were evaluated. From all the patients affected by colon cancer, macroscopically normal colon specimens at least 15 cm from the adenocarcinoma were also collected. NMR spectra of samples were acquired using a Bruker Avance400 NMR spectrometer (4 kHz spinning, 4° C). PCA and PLS-DA analyse were carried out using PLS-Toolbox 4,1 for MATLAB. 1D and 2D HR-MAS NMR experiments provide a complete and unambiguous identification of the metabolic pattern characterizing the examined tissues. The NMR spectra of neoplastic and macroscopically normal colon tissues present different metabolic profiles with respect to thse of the healthy ones. Furthermore, the NMR spectra display variations in the metabolic profile for neoplastic tissues with different histological diagnosis, and also for those classified identical from the histological analyses. This findings indicates that the family of colon carcinoma is characterized by metabolic heterogeneity also within the same subclass of tumours. The statistical multivariate approach applied to HR-MAS NMR spectra permits to find metabolic markers of the healthy and the neoplastic state of colon tis¬sues, and to correctly classify the samples. The combination of spectroscopic data with statistical analyses offers a useful tool to better understand some molecular aspects of colon tumour biochemistry, as well as to identify tumoral markers (taurine, myo-inositol, choline containing compounds, carboxylic acids and lipids), and could represent the molecular basis for an accurate and non invasive clinical applications if in vivo NMR