Pitfalls of onco-metabolomics: impact of sample integrity on metabolomic investigations in more than 4500 human serum samples from ten different cohorts

Introduction: Metabolomics such as nuclear magnetic resonance spectroscopy or mass spectrometry (MS) in different body fluids are considered potentially useful diagnostic techniques for various diseases including cancer. One of the most important prerequisites of metabolomics is a high sample quality, for which reason explicit care must be taken during pre-analytical/analytical handling. Methods: In the present study, we investigated the influence of pre-processing (PPT), and pre-centrifugation time (PCT), sample storage time (SST), and sample texture on NMR-based metabolite levels in 4,658 long-term and short-term stored retrospectively and prospectively collected serum samples from breast and prostate cancer patients as well as from healthy men. Results: We found that the majority of the metabolites were highly stable with regard to variations in PCT, PPT, or SST. PCT and PPT significantly affected the concentrations of only a few individual metabolites, including ascorbic acid, asparagine, glucose, glutamic acid, glutamine, lactate, phenylalanine, pyruvic acid, and serine, indicating that pre-analytical protocol variations need to be considered for the quantitative analysis of metabolites. Notably, the glucose:lactate and glutamine:glutamic acid ratios were found to be suitable to assess sample quality in case of high PCT or PPT. Importantly, the highest sample quality was detected in prospectively collected serum samples with strict protocol adherence and a total PPT of only 1.2 h. Specific care must also be taken with the analysis of lipemic samples, in which strong variations in the concentrations of lipid metabolites, albumin, and valine were observed. Discussion: In summary, our data show that the majority of metabolites are mostly stable with regard to variations in pre-analytical processing, indicating that retrospective biobank samples are suitable for metabolomics studies. However, individual metabolites are strongly dependent on PCT and PPT, suggesting that a short PPT may be mandatory for clinical diagnosis, depending on the individual metabolite to be measured.