Shin Nishiumi, Yoshihiro Izumi, Akiyoshi Hirayama, Masatomo Takahashi, Motonao Nakao, Kosuke Hata, Daisuke Saigusa, Eiji Hishinuma, Naomi Matsukawa, Suzumi M Tokuoka, Yoshihiro Kita, Fumie Hamano, Nobuyuki Okahashi, Kazutaka Ikeda, Hiroki Nakanishi, Kosuke Saito, Masami Yokota Hirai, Masaru Yoshida, Yoshiya Oda, Fumio Matsuda, Takeshi Bamba
Metabolites, 12(2), Feb 1, 2022
In mass spectrometry-based metabolomics, the differences in the analytical results from different laboratories/machines are an issue to be considered because various types of machines are used in each laboratory. Moreover, the analytical methods are unique to each laboratory. It is important to understand the reality of inter-laboratory differences in metabolomics. Therefore, we have evaluated whether the differences in analytical methods, with the exception sample pretreatment and including metabolite extraction, are involved in the inter-laboratory differences or not. In this study, nine facilities are evaluated for inter-laboratory comparisons of metabolomic analysis. Identical dried samples prepared from human and mouse plasma are distributed to each laboratory, and the metabolites are measured without the pretreatment that is unique to each laboratory. In these measurements, hydrophilic and hydrophobic metabolites are analyzed using 11 and 7 analytical methods, respectively. The metabolomic data acquired at each laboratory are integrated, and the differences in the metabolomic data from the laboratories are evaluated. No substantial difference in the relative quantitative data (human/mouse) for a little less than 50% of the detected metabolites is observed, and the hydrophilic metabolites have fewer differences between the laboratories compared with hydrophobic metabolites. From evaluating selected quantitatively guaranteed metabolites, the proportion of metabolites without the inter-laboratory differences is observed to be slightly high. It is difficult to resolve the inter-laboratory differences in metabolomics because all laboratories cannot prepare the same analytical environments. However, the results from this study indicate that the inter-laboratory differences in metabolomic data are due to measurement and data analysis rather than sample preparation, which will facilitate the understanding of the problems in metabolomics studies involving multiple laboratories.