MALDI Imaging Combined with Hierarchical Clustering as a New Tool for the Interpretation of Complex Human Cancers

Abstract

Proteomics analyses have been exploited for the discovery of novel biomarkers for the early recognition and prognostic stratification of cancer patients. These analyses have now been extended to whole tissue sections by using a new tool, that is, MALDI imaging. This allows the spatial resolution of protein and peptides and their allocation to histoanatomical structures. Each MALDI imaging data set contains a large number of proteins and peptides, and their analysis can be quite tedious. We report here a new approach for the analysis of MALDI imaging results. Mass spectra are classified by hierarchical clustering by similarity and the resulting tissue classes are compared with the histology. The same approach is used to compare data sets of different patients. Tissue sections of gastric cancer and non-neoplastic mucosa obtained from 10 patients were forwarded to MALDI-Imaging. The in situ proteome expression was analyzed by hierarchical clustering and by principal component analysis (PCA). The reconstruction of images based on principal component scores allowed an unsupervised feature extraction of the data set. Generally, these images were in good agreement with the histology of the samples. The hierarchical clustering allowed a quick and intuitive access to the multidimensional information in the data set. It allowed a quick selection of spectra classes representative for different tissue features. The use of PCA for the comparison of MALDI spectra from different patients showed that the tumor and non-neoplastic mucosa are separated in the first three principal components. MALDI imaging in combination with hierarchical clustering allows the comprehensive analysis of the in situ cancer proteome in complex human cancers. On the basis of this cluster analysis, classification of complex human tissues is possible and opens the way for specific and cancer-related in situ biomarker analysis and identification.

Data-driven identification of prognostic tumor subpopulations using spatially mapped t-SNE of mass spectrometry imaging data

“…MSI has several key characteristics that make it well suited to this task; it is an untargeted analysis that can simultaneously analyze hundreds of molecular ions, it can be directly applied to tissue sections, it is inexpensive, and it is fast. Several previous studies have reported MSI's ability to uncover tumor subpopulations in histologically identical regions of tumor tissue (5,20,21). Here we have used dimensionality reduction based on t-SNE followed by bisecting k-means clustering to automatically segment the tumor-specific MSI data from a patient series into an optimum number of subpopulations.…”

“…Deininger et al (5) were the first to report that MSI may reveal the biomolecular intratumor heterogeneity associated with a tumor's clonal development. A hierarchical cluster analysis of the MSI data revealed a patchwork of molecularly distinct regions, which were postulated to reflect the tumor's clonal evolution.…”

Протеоміка Та Мас-Спектрометрія З Матрично-Активованою Лазерною Десорбцією/Іонізацією Як Сучасний Діагностичний Інструмент При Захворюваннях Нирок

“…This molecular information was then combined with clinical data taken from patients in order to predict the survival rate of patients based upon the number of observed molecular tumour sub-populations. This is a prime example of not only how MALDI-MSI can provide information that is confirmatory, or complimentary to histological information, but also how it can provide further information that was not previously possible [15,16].…”

Serum Protein and Genetic Tumor Markers of Gastric Carcinoma

“…Lots of studies on gastric cancer have been reported recently (Deininger et al, 2008;Li et al, 2008;Khoder et al, 2009;Zhang et al, 2010), and the results showed some clinical significance in the diagnosis of gastric cancer. Lu et al (2010) compared the difference of proteomic analysis between gastric cancer (n = 34) and normal control (n = 30) and found five different protein peaks, by which a diagnostic model was developed and showed a sensitivity of 94.3% and a specificity of 93.3% in the diagnosis of gastric cancer.…”