Bindesh Shrestha (Presenter)
Waters Corporation
Authorship: Kirill Veselkov(1), Emrys Jones(2), Don Cooper(2), Emmanuelle Claude(2), James Langridge(2)
(1) Division of Computational and Systems Medicine, Department of Surgery and Cancer, Imperial College London, London, UK, (2) Waters Corporation, Manchester, UK
| Short Abstract Mass Spectrometry Imaging (MSI) is an emerging technology in pathology research that generates hundreds of gigabytes of detailed molecular data of potential importance. Effective translation of MSI data into biologically or clinically useful information requires advanced computational solutions. Here, an integrated bioinformatics research platform is presented that allows intuitive histology-directed interrogation of MSI datasets for tissue-specific biomarker recovery, automated tissue classification and entropy driven tumour heterogeneity assessment. |
Long Abstract
Introduction:
Cancer incidence in Europe was recently estimated at 3.45 million cases, with 1.75 million deaths, and costing the EU 124 billion Euros every year. With such incidence rates, fundamental understanding of cancer biology is required to prevent (prognosis), identify (diagnosis), stratify (theranosis) and treat cancer. Mass Spectrometry Imaging (MSI) is an emerging technology in pathology research that generates hundreds of gigabytes of detailed molecular data of potential diagnostic and prognostic importance. The current need for mass spectrometry-based bioinformatics can be summarized as follows:
• A critical requirement to rapidly process and visualize large amounts of data that is generated even for small-scale studies due to the non-targeted nature of MSI
• High demand for precise and personalized molecular biomarkers for clinical research, and translation
• Overall high demand for test automation and standardisation
Recent advances in MS technologies, including Matrix Assisted Laser Desorption/Ionization (MALDI) and Desorption ElectroSpray Ionization (DESI), allows the generation of comprehensive molecular information, should allow the widespread adoption of MSI technologies in the near- to midterm. However, the major impediment to this progress currently centres on the lack of a complete analytical solution(s) based on chemo-informatics strategies. Here, we outline current roadblocks in translational MSI and introduce a comprehensive workflow designed to address current methodological limitations [1,2].
Methods:
An integrated bioinformatics solution is presented that allows intuitive histology-directed interrogation of MSI datasets for tissue specific biomarker recovery, automated tissue classification and tumour heterogeneity assessment. The proposed bioinformatics pipeline includes a series of designated steps covering (i) robust pre-processing workflow, capable of reducing data complexity while simultaneously retaining disease-relevant information; (ii) image-fusion algorithms for automated co-registration of biochemical and histological datasets; (iii) putative molecular ion annotation tools; (iv) computational statistical learning approaches for amalgamation of MSI with other diagnostics and prognostics techniques v) entropy driven tumour heterogeneity assessment. This computational strategy aims to transform the data generated into biologically and clinically useful information, e.g. to recover MSI-driven molecular signatures of patho-physiological significance across a variety of pathological or biomedical applications.
Results:
We show that raw analytical signal processing, image co-registration methods and statistical learning tools for the recovery of clinically relevant molecular patterns can influence one another, and thus ideally must be considered within an integrated platform. Using the proposed bioinformatics solution, we have investigated region-specific lipid biochemistry in colorectal and liver cancer tissue sections generated from MSI data acquired on Waters mass spectrometry platforms (SYNAPT G2-Si and/or Xevo G2-S). MSI profiles of the studied tissue sections consisted predominantly of complex lipids including phosphatidylinositols, phosphatidylserines, and phosphatidylethanolamine plasmalogens, among others. Unique lipid patterns were observed using this approach according to tissue type, and a tissue recognition system using multivariate molecular patterns allowed highly accurate identification of pixels according to the morphology of the different tissue types. The ontology driven annotation of lipidomic signatures for clinical research is demonstrated. In the case of tumour tissues, the quantitative metric based on entropic framework was developed to assess the heterogeneity of molecular phenotypes within a tumour and its surrounding tissues. We show that this strategy offers unique insights into tumour micro-environmental biochemistry, tumour induced heterogeneity of molecular phenotypes and should facilitate the compilation of a large-scale tissue morphology-specific MSI spectral database to pursue next-generation, fully automated histological approaches.
Conclusions:
Without a step-wise increase in the computational interpretation of the highly complex MSI data sets, it is unlikely that MSI technologies will be able to realize its significant potential in translational and clinical applications. The proposed bioinformatics solution facilitates the validation of Mass Spectrometry Imaging data across a variety of clinical applications in translational cancer research settings.
References & Acknowledgements:
[1] Veselkov KA, et al. (2014), Chemo-informatic strategy for imaging mass spectrometry-based hyperspectral profiling of lipid signatures in colorectal cancer. PNAS, 111: 1216-122
[2] Veselkov KA* & Abbassi-Ghadi N*, et al. (2014). Discrimination of lymph node metastases using desorption electrospray ionisation-mass spectrometry imaging. Chem Comm, 1359-7345 (*Equal Contribution).
| Description | Y/N | Source |
| Grants | no | |
| Salary | yes | Waters |
| Board Member | no | |
| Stock | no | |
| Expenses | yes | Waters |
IP Royalty: no
| Planning to mention or discuss specific products or technology of the company(ies) listed above: | yes |