Umeå University's logo

BACKGROUND: The Northern Sweden Health and Disease Study is a unique population-based biobank linked to the clinical data registries. The samples originate from over 133 000 individuals living in the northern part of Sweden, and primarily collected during health checkups from the age of 40 years. Our project aims to investigate alterations in metabolite signatures in blood plasma of healthy blood donors that later in life developed a tumor. Brain tumors, especially glioblastoma is associated with poor prognosis. To explore early events of metabolic reprograming linked to future diagnosis, we investigated alterations in metabolite concentrations in plasma collected several years before diagnosis with matched healthy controls.

MATERIALS AND METHODS: In total 392 analytical samples (256 repeated timepoint and 136 single timepoint, case-control samples) were analyzed using GCTOFMS. Constrained randomization of run order was utilized to maximize information output and minimize the false discovery rate. By use of reference databases, we could with high confidence quantify and identify 150 plasma metabolites. We detected metabolites with significant alterations in concertation between pre-clinical glioma cases and healthy controls by the effect projection approach based on orthogonal partial least squares (OPLSEP).

RESULTS AND CONCLUSIONS: For the repeated blood samples, we designed and applied a novel multivariate strategy for high resolution biomarker pattern discovery. We utilize the fact that we have available samples from two repeated time points prior to diagnosis for each future glioma case and their matched controls to construct a small design of experiment (DoE) of four samples for each match pair. The data for each individual DoE was evaluated by OPLS-EP to determine the effect of each individual metabolite in relation to control-case, time and their interaction. Finally, latent significance calculations by means of OPLS were used to extract and evaluate the correct latent biomarker and highlight true significance of individual metabolites. Our study presents an approach to minimize confounding effects due to systematic noise from sampling, the analytical method, as well as take into account personalized metabolic levels over time, enabling biomarker detection within a smaller sample group. We will present and discuss the latest results and biomarkers from this exploratory metabolomics study at the meeting