A Robust Workflow And Quality Control Procedure To Analyze The Human Metabolome By High-Resolution Mass Spectrometry In Epidemiological Studies


1 Biomarkers Group, Nutrition and Metabolism Section, International Agency for Research on Cancer, Lyon, France

Purpose: Blood and urine samples contain thousands of metabolites that can be analyzed in prospective epidemiological studies by high resolution mass spectrometry (MS) to provide new information on disease risk factors and on mechanisms leading to cancer. A major challenge to apply these analytical methods to cancer epidemiology has been the control of the technical variability associated with sample processing and MS analysis to make possible the statistical comparison of levels of thousands metabolites in several hundred subjects. Unlike in quantitative analysis, calibration standards for the metabolites specific cannot be employed, and thus drifts in the instrument performance over time may compromise data quality.   
Methods: Plasma/serum samples (20-30 µL) were prepared by protein precipitation in well plates and analysed by high resolution MS (Agilent 6550 Q-TOF) coupled with UHPLC. Two orthogonal chromatographic methods were used (HILIC, RP) with electrospray ionization in both positive and negative polarities. Data-dependent MS/MS spectra were acquired from a pool of study samples as part of the analytical batch. Data preprocessing was performed using Agilent feature extraction workflow. Rigorous quality control (QC) was employed by monitoring of selected metabolites in QC samples.
Results: Sub-15 minute analysis time and straightforward sample preparation enabled throughput of 500 samples per week. Representative QC results for a batch of 404 plasma samples showed 2900 features with CV<20% in QCs (n=36), with CV% for 10 known compounds 5.9-15.8% without loss in overall signal intensity along the whole batch. The workflow has been successfully used for cross-sectional, nested case-control, and nutritional intervention studies with preliminary results presented.
Conclusions: An untargeted metabolomic workflow for epidemiological investigations is presented that enables fast analysis of large sets of samples without commonly observed analytical drifts, with a robust QC procedure for in-depth assessment of the data quality.
Funding source: IARC Postdoctoral Fellowship, EU-FP7 Cofund, Eurocan.