Desorption electrospray ionization mass spectrometry (DESI-MS) imaging provides valuable information on disease state and grade based on lipid profiles in tissue sections, which is valuable in cancer diagnosis and complementary to traditional histopathological examination. An inherent feature of MS image acquisition is the large amount of information contained in each image. Hyperspectral imaging (HSI) and principal component analysis (PCA) provide a comprehensive exploration of the spatial and multivariate chemical information contained in MS images, allowing tissue characterization on the basis of the entirety of chemical information. Specific m/z values characteristic for each tissue type can be easily visualized and correlated to the 2D image. Correspondence between the chemical and spatial information can be achieved by applying an interactive brushing procedure. Normal and tumor tissue sections of kidney, bladder, germ and prostate human tissue were imaged using a commercial LTQ linear ion trap mass spectrometer and a lab-built prototype DESI-MS ion source. Experiments were carried out in the negative ion mode, using a 5 kV spray voltage, a flow rate of 1.5 μL/min, and 150 psi nebulizing gas (N2) pressure. The spray solvent was acetonitrile:water (50:50 v/v). Mass spectra were acquired over the range m/z 150-1000. For each DESI-MS image, the information was coded using the corresponding data hypercube. Reduction of the spectral variables (1:4) by a consecutive-window averaging algorithm, normalization with respect to the total ion current (TIC) and column-centering were performed. Multivariate analyses were carried through in-house MatLab routines. A multivariate approach for exploring spatial and chemical information contained in any DESI-MS image is advantageous compared to that based on absolute intensities in single ion images, since it provides a global visualization of all the spectral variables and their inter-correlations. The strategy followed for normalization with respect to the TIC allows unwanted variations due to instrumental variability and/or irregularities in the sample surface – such as folded tissue – to be minimized. The developed procedure can be used to guide subsequent investigations of the mass spectra and single ion images and maximize the exploratory characterization of different regions of any DESI-MS image. This can be done, as a first step, prior to supervised analysis that requires additional levels of information to be known. Furthermore, with the interactive brushing procedure, pixels with similar chemical profiles can be manually selected from the PCA space, scaling down to the single-pixel level, in order to automatically identify correspondences between the groups of points in PCA space and particular regions of the DESI-MS image. This multivariate strategy applied to DESI-MS images of different types of human tissues showed good ability in characterizing tumor and normal tissue sections by correlating the lipid information with pathological evaluation of the same samples. Some common lipid ions, such as m/z 885.5 and m/z 788.5, identified as PI(18:0/20:4) and PS(18:0/18:1), as well as ions of free fatty acids and their dimers, appeared to be highly discriminating across different types of human cancers, while other ions, such as m/z 217 for normal kidney, m/z 465.5 (cholesterol sulfate) for prostate cancer tissue and m/z 795.5 (seminolipid 16:0/16:0) for germ tissue, appeared to be selective for the type of tissue analyzed. These data indicate that lipid profiles can reflect not only the disease/health state of tissue but also are characteristic of tissue type. Comprehensive and interactive investigation of spatial and chemical information contained in DESI-MS images with automatic correspondence between the information domains.
Interactive hyperspectral approach for exploring and interpreting DESI-MS images of cancerous and normal tissue sections
PIRRO, VALENTINA;
2012-01-01
Abstract
Desorption electrospray ionization mass spectrometry (DESI-MS) imaging provides valuable information on disease state and grade based on lipid profiles in tissue sections, which is valuable in cancer diagnosis and complementary to traditional histopathological examination. An inherent feature of MS image acquisition is the large amount of information contained in each image. Hyperspectral imaging (HSI) and principal component analysis (PCA) provide a comprehensive exploration of the spatial and multivariate chemical information contained in MS images, allowing tissue characterization on the basis of the entirety of chemical information. Specific m/z values characteristic for each tissue type can be easily visualized and correlated to the 2D image. Correspondence between the chemical and spatial information can be achieved by applying an interactive brushing procedure. Normal and tumor tissue sections of kidney, bladder, germ and prostate human tissue were imaged using a commercial LTQ linear ion trap mass spectrometer and a lab-built prototype DESI-MS ion source. Experiments were carried out in the negative ion mode, using a 5 kV spray voltage, a flow rate of 1.5 μL/min, and 150 psi nebulizing gas (N2) pressure. The spray solvent was acetonitrile:water (50:50 v/v). Mass spectra were acquired over the range m/z 150-1000. For each DESI-MS image, the information was coded using the corresponding data hypercube. Reduction of the spectral variables (1:4) by a consecutive-window averaging algorithm, normalization with respect to the total ion current (TIC) and column-centering were performed. Multivariate analyses were carried through in-house MatLab routines. A multivariate approach for exploring spatial and chemical information contained in any DESI-MS image is advantageous compared to that based on absolute intensities in single ion images, since it provides a global visualization of all the spectral variables and their inter-correlations. The strategy followed for normalization with respect to the TIC allows unwanted variations due to instrumental variability and/or irregularities in the sample surface – such as folded tissue – to be minimized. The developed procedure can be used to guide subsequent investigations of the mass spectra and single ion images and maximize the exploratory characterization of different regions of any DESI-MS image. This can be done, as a first step, prior to supervised analysis that requires additional levels of information to be known. Furthermore, with the interactive brushing procedure, pixels with similar chemical profiles can be manually selected from the PCA space, scaling down to the single-pixel level, in order to automatically identify correspondences between the groups of points in PCA space and particular regions of the DESI-MS image. This multivariate strategy applied to DESI-MS images of different types of human tissues showed good ability in characterizing tumor and normal tissue sections by correlating the lipid information with pathological evaluation of the same samples. Some common lipid ions, such as m/z 885.5 and m/z 788.5, identified as PI(18:0/20:4) and PS(18:0/18:1), as well as ions of free fatty acids and their dimers, appeared to be highly discriminating across different types of human cancers, while other ions, such as m/z 217 for normal kidney, m/z 465.5 (cholesterol sulfate) for prostate cancer tissue and m/z 795.5 (seminolipid 16:0/16:0) for germ tissue, appeared to be selective for the type of tissue analyzed. These data indicate that lipid profiles can reflect not only the disease/health state of tissue but also are characteristic of tissue type. Comprehensive and interactive investigation of spatial and chemical information contained in DESI-MS images with automatic correspondence between the information domains.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
