Quality control in GC–MS analysis of amino acids in human urine and plasma: Possible implications for targeted and untargeted metabolomics

In metabolomics, LC-MS(/MS) is currently used about two times more frequently than GC–MS(/MS) since about 2005, perhaps the year of appearance of metabolomics as an individual analytical approach in the life sciences. LC-MS(/MS) and GC–MS(/MS) share many common challenges in targeted and untargeted metabolomics, the Janus face of metabolomics. Especially the importance of the key issue of quality assurance (QA) and quality control (QC) has been recognized and is increasingly addressed by individual researchers and consortia. In previous work, our group has proposed a QC system for the quantitative GC–MS analysis of amino acids in human plasma samples. In the present study, we investigated the utility of such a QC approach for the quantitative (targeted) GC–MS analysis of amino acids in human urine samples by using previously validated methods. Endogenous (unlabeled) amino acids were analyzed in 10-μL aliquots of study urine samples and in QC urine samples as methyl ester pentafluoropropionyl derivatives (d₀Me-PFP) using a mixture of in-situ prepared trideuteromethyl esters for use as internal standards, which were then converted into their PFP derivatives (i.e., d₃Me-PFP). GC–MS analysis of 38 study urine samples and 8 QC urine samples was performed in the negative-ion chemical ionization (NICI) mode by selected-ion monitoring (SIM) of characteristic ions of d₀Me-PFP and d₃Me-PFP within a single run by using an oven temperature program. For direct comparison, analysis of 35 study plasma samples and 8 QC plasma samples of the same clinical study was performed. Closely comparable experimental and instrumental conditions were used in the analyses, and the same staff was involved in the entirely analytical process. Chromatographic H/D isotope effects and peak area values were determined and examined with respect to qualitative and quantitative GC and MS parameters including accuracy and precision. Study and QC plasma behaved similarly. On a molar basis, the amino acid derivatives have different peak area values. Yet, this does not affect the accuracy of the GC–MS method. Our study suggests that untargeted GC–MS metabolomics studies on amino acids in biological samples are inappropriate for quantitative GC–MS analysis. Targeted metabolomics, i.e., use of isotopologs are indispensable for reliable quantitative GC–MS analysis of amino acids in biological samples. It is reasonable to assume that our findings will also apply to other classes of analytes and types of biological samples.