RT Journal Article SR Electronic T1 Network analysis of mass spectrometry imaging data from colorectal cancer identifies key metabolites common to metastatic development JF bioRxiv FD Cold Spring Harbor Laboratory SP 230052 DO 10.1101/230052 A1 Paolo Inglese A1 Nicole Strittmatter A1 Luisa Doria A1 Anna Mroz A1 Abigail Speller A1 Liam Poynter A1 Andreas Dannhorn A1 Hiromi Kudo A1 Reza Mirnezami A1 Robert D Goldin A1 Jeremy K Nicholson A1 Zoltan Takats A1 Robert C Glen YR 2018 UL http://biorxiv.org/content/early/2018/01/09/230052.abstract AB A deeper understanding of inter-tumor and intra-tumor heterogeneity is a critical factor for the advancement of next generation strategies against cancer. The heterogeneous morphology exhibited by solid tumors is mirrored by their metabolic heterogeneity. Defining the basic biological mechanisms that underlie tumor cell variability will be fundamental to the development of personalized cancer treatments. Variability in the molecular signatures found in local regions of cancer tissues can be captured through an untargeted analysis of their metabolic constituents. Here we demonstrate that DESI mass spectrometry imaging (MSI) combined with network analysis can provide detailed insight into the metabolic heterogeneity of colorectal cancer (CRC). We show that network modules capture signatures which differentiate tumor metabolism in the core and in the surrounding region. Moreover, module preservation analysis of network modules between patients with and without metastatic recurrence explains the inter-subject metabolic differences associated with diverse clinical outcomes such as metastatic recurrence.Significance Network analysis of DESI-MSI data from CRC human tissue reveals clinically relevant co-expression ion patterns associated with metastatic susceptibility. This delineates a more complex picture of tumor heterogeneity than conventional hard segmentation algorithms. Using tissue sections from central regions and at a distance from the tumor center, ion co-expression patterns reveal common features among patients who developed metastases (up of > 5 years) not preserved in patients who did not develop metastases. This offers insight into the nature of the complex molecular interactions associated with cancer recurrence. Presently, predicting CRC relapse is challenging, and histopathologically like-for-like cancers frequently manifest widely varying metastatic tendencies. Thus, the methodology introduced here more robustly defines the risk of metastases based on tumor biochemical heterogeneity.Author contributions P.I., Z.T., R.C.G.: designed the study, developed the workflow, analyzed the data, interpreted the results, wrote the paper; N.S. collected the MS, performed the H…E staining, wrote the paper; L.D.: interpreted the results, wrote the paper; A.M.: collected the MS; A.S.: histological assessment; L.P.: collected the tissue specimens and clinical metadata; A.D.: collected the MS; H.K.: performed the H…E staining; R.M.: collected the tissue specimens and clinical metadata. R.G.: histological assessment; J.K.N: designed the study, edited the paper.