Roland Geyer (Presenter)
Inselspital, Bern University Hospital
Authorship: Roland Geyer (1), Alexander B. Leichtle (1), Christos Nakas (1,2), Anke Tönjes (3) and Martin G. Fiedler (1)
(1) University Institute of Clinical Chemistry, Inselspital, Bern University Hospital, 3010 Bern, SWITZERLAND; (2) Laboratory of Biometry of the University of Thessaly (Greece); 3 Division of Endocrinology, Dept. Internal Medicine, Neurology and Dermatology, University of Leipzig, Leipzig, Germany
| Short Abstract Currently, metabolomics studies start with a broad, non-targeted screening for as many metabolic features as possible. Identified metabolites with potential as biomarker are then analyzed by a targeted, quantitative method(s) to confirm the marker sensitivity and specificity. However, clinical metabolomics often has to deal with very heterogeneous samples. This requires large cohorts to achieve sufficient correlation coefficients for comparison of groups in non-targeted screening and to reveal whether metabolites may be significant for e.g. diagnostics prognostics or patient stratification. Sensitive targeted profiling of metabolites may be suitable to lower heterogeneity in the sample groups and to identify outlier. Here we compared a non-targeted with a targeted workflow for initial characterization of a study group related to pathogenesis of diabetes. |
Long Abstract
Introduction:
An early onset of diabetic stages in a sorbian population shall be evaluated by the analysis of metabolic differences after intervention by an 75g oral glucose tolerance test. We compared a non-targeted with a targeted workflow for initial characterization of a study group related to pathogenesis of diabetes comparing normal glucose tolerance with type 2 diabetes and diabetic pre-stages.
Methods:
Blood samples were obtained 30 and 120min after intake. Frozen serum samples (n=40) were prepared with the AbsoluteIDQ® p180 Kit (Biocrates) [1] to extract acylcarnitines, glycerophospho- and sphingolipids as well as amino acids. The lipids containing eluates were analyzed by triple quadrupole mass spectrometry (QqQ) with 111 lipid specific mass transitions and by high resolutions mass spectrometry (HRMS, QToF) for the mass range m/z 100-1200amu. Using the same chromatography (solvents and column, 12min gradient) enabled an easier comparison of the metabolite peaks obtained by the two screening methods.
Peak picking and alignment was done either with Progenesis QI software (Waters) or MRM-DIFF, a tool for QqQ data [2]. The different sets of raw data were scaled and transformed prior their evaluation with multivariate statistical methods (e.g., PCA, OPLSD-DA) in SIMCA (UMETRICS) or with scripts from the r-project. Samples were grouped either as normal glucose tolerance (NGT) or as type 2 diabetes and diabetic pre-stages based on blood glucose values already determined after intervention [3].
Results and Discussion:
With the 111 mass transitions, covering glycerophosphatidylcholine- and sphingolipids in positive electrospray mode, more than 1800 peaks were extracted while scanning by quadrupole time-of-flight (QToF) HRMS revealed about 3900 features. Peaks from the semi-targeted or non-targeted analysis were annotated by their retention time and mass or by mass transition. After data clean up, QToF data showed about 2.5 times more features (~960 vs. 2300, respectively). However, separation of the two glucose tolerance groups by OPLS-DA algorithm was more pronounced with the MRM data covering choline and sphingomyeline lipids 120min after oGTT. Using the whole set of HRMS features the samples did not show a group separation at all. Therefore, we used a subset of mass features matching compounds with significance in the MRM analysis or related to some acylcarnitines know to be altered with diabetic pre-stages. With these limited number of HRMS features (n=51) a group separation by OPLS-DA was achievable.
However, most features did not relate to the putative identified metabolites from the MRM screening. HRMS values and retention times indicated different metabolites.
The lipids from QqQ measurements with highest significance in the OPLS-DA model were at concentrations in the samples too low for detection by the QToF analysis.
Conclusion:
A targeted profiling by mass transitions (MRM based screening) enabled a more pronounced group separation for the heterogeneous sample set as it is more sensitive and specific to a certain pathway (lipid metabolism).
For a larger sample set MRM data would probably help 1) to control the anticipated grouping of patient categories and 2) to identify outlier samples. Data analysis is much faster and feature assignment more straightforward. Full scan high resolution mass spectrometry is then suited for in-depth analysis of representative samples covering a broader range of pathways.
References & Acknowledgements:
References:
[1] http://www.biocrates.com/products/research-products/absoluteidq-p180-kit
[2] Tsugawa et al. 2015, Front. Genet. 5:471. doi: 10.3389/fgene. 2014.00471
[3] Mai M. et al. 2013, PLoS ONE 8(12): e82459. doi:10.1371
| Description | Y/N | Source |
| Grants | no | |
| Salary | no | |
| Board Member | yes | MSACL EU Scientific Committee |
| Stock | no | |
| Expenses | no |
IP Royalty: no
| Planning to mention or discuss specific products or technology of the company(ies) listed above: | no |