Anal Bioanal Chem (2014) 406:771–784 DOI 10.1007/s00216-013-7506-9
Determination of urine caffeine and its metabolites by use of high-performance liquid chromatography-tandem mass spectrometry: estimating dietary caffeine exposure and metabolic phenotyping in population studies Michael E. Rybak & Ching-I Pao & Christine M. Pfeiffer
Received: 18 September 2013 / Revised: 8 November 2013 / Accepted: 11 November 2013 / Published online: 4 December 2013 # Springer-Verlag Berlin Heidelberg (outside the USA) 2013
Abstract We have developed and validated a highperformance liquid chromatography-tandem mass spectrometric (LC-MS/MS) method for determining urine caffeine and 14 caffeine metabolites suitable for estimating caffeine exposure and metabolic phenotyping in population studies. Sample preparation consisted solely of a series of simple reagent treatments at room temperature. Stable isotopelabeled analogs were used as internal standards for all analytes. We developed rapid LC-MS/MS separations for both positive and negative ion mode electrospray ionizations to maximize measurement sensitivity. Limits of detection were 0.05–0.1 μmol/L depending on the analytes. Method imprecision, based on total coefficients of variation, was generally 1 μmol/L. Analyte recoveries were typically within 10 % of being quantitative (100 %), and good agreement was observed among analytes measured across different MS/MS transitions. We applied this method to the analysis of a convenience set of human urine samples (n =115) and were able to detect a majority of the analytes in ≥99 % of samples as well as calculate caffeine metabolite phenotyping ratios for cytochrome P450 1A2 and N-acetyltransferase 2. Whereas existing LC-MS/MS methods are limited in number of caffeine metabolites for which they are validated, or are designed for studies in which purposely elevated caffeine levels are expected, our method is the first of its kind designed specifically for the rapid, sensitive, accurate,
M. E. Rybak (*) : C.200 mg/day in the US for individuals >20 years . The effects of caffeine on mental alertness and topics such as caffeine tolerance, addiction, and withdrawal have been studied extensively [2–8]. Caffeine consumption has also been studied as a risk factor for diseases and conditions such as hypertension [9, 10], cardiovascular disease [11, 12], various cancers [13, 14], reproduction and developmental abnormalities , and mental and behavioral disorders . Almost all of these studies used dietary intake data (e.g., food frequency questionnaires, 24-h dietary recalls) to quantify caffeine exposure. Relying upon dietary intake data can be problematic because of challenges related to accurately establishing the caffeine content of dietary sources . The use of biologic indicators such as caffeine and/or its metabolites has
been proposed as an alternative means of assessing caffeine intake , and the measurement of these compounds in serum and urine has been explored for this purpose [19, 20]. In addition to determining dietary caffeine exposure, the quantitation of urine caffeine and its metabolites can be used to assess differences in metabolic activity. Caffeine undergoes an intricate series of reactions, primarily in the liver, to yield a mixture of N-methylated xanthines, uric acids, and an acetylated uracil (Fig. 1) [21, 22]. The enzymes involved in caffeine metabolism, such as cytochrome P450 1A2 (CYP1A2) and other P450 enzyme systems, N -acetyltransferase 2 (NAT2) and xanthine oxidase (XO) [23–25], are of interest from a phenotyping standpoint due to their roles of the activation or detoxification of various xenobiotic compounds. The phenotyping of CYP1A2, NAT 2, and XO enzyme activities has been accomplished by measuring urine caffeine and caffeine metabolites in subjects administered a controlled dietary caffeine challenge [26–33] as well as in subjects where dietary caffeine consumption was uncontrolled [34, 35]. Several high-performance liquid chromatography-tandem mass spectrometry (LC-MS/MS) methods for measuring caffeine and its metabolites in urine have been reported [22, 36–42]. None of these methods, however, are ideal for generating population-based data for the combined purpose of estimating dietary consumption and phenotyping. Enzyme activity phenotyping applications [22, 36, 37, 42] typically quantify only those compounds required in the phenotype calculation at concentrations purposely elevated from a caffeine challenge. Methods designed for doping analysis [40, 41] only quantify a limited number of analytes, and method performance is optimized for specified concentration thresholds. Existing methods also lacked complete internal standardization with stable isotope-labeled analogs for each respective analyte. Our goal was to develop an LC-MS/MS for quantifying urine caffeine and its metabolites with the intended purpose of providing reference biologic data for estimating caffeine consumption and detoxification enzyme–activity classifications (i.e., phenotypes) in population studies such as the National Health and Nutrition Examination Survey (NHANES) . Specifically, we sought to develop an LC-MS/MS method that would favorably combine a high degree of measurement sensitivity, specificity, accuracy, and precision with short analysis times to accommodate the throughput burden of thousands of samples of potentially wide-ranging analyte concentrations.
Material and Methods Samples, standards, and reagents A convenience set of urine samples (n =115) was collected from anonymous on-site volunteers in a manner consistent
M.E. Rybak et al.
with an internal review board-approved human subject protocol. Once collected, samples were assigned random identification numbers, and aliquots of these samples were generated and stored at −70 °C prior to analysis. Single analyte stock solutions for each analyte and internal standard were prepared by dissolving solid material in aqueous solution using 0.45-μm-filtered reverse osmosis deionized water from a laboratory source. The following compounds were obtained from Sigma-Aldrich (St. Louis, MO, USA): 1methylxanthine (1X), 3-methylxanthine (3X), 7methylxanthine (7X), 1,3-dimethylxanthine (theopylline, 13X), 1,7-dimethylxanthine (paraxanthine, 17X), 3,7dimethylxanthine (theobromine, 37X), 1,3,7trimethylxanthine (caffeine, 137X), 1-methyluric acid (1U), 1,3-dimethyluric acid (13U), 1,7-dimethyluric acid (17U), 3, 7-dimethyluric acid (37U), and 1,3,7-trimethyluric acid (137U). Toronto Research Chemicals (Toronto, ON, Canada) provided 3-methyluric acid (3U), 7-methyluric acid (7U), 5acetylamino-6-amino-3-methyluracil (AAMU), and 5acetylamino-6-formylamino-3-methyluracil (AFMU). All analyte materials were at least 98 % chemical purity. The following internal standards were custom synthesized by IsoSciences (King of Prussia, PA, USA) and are presently available as catalog items 1X, 3X, 7X, 17X, 1U, 3U, 7U, 13U, 17U, 137U (2,4,5,6-13C4; 1,3,9-15N3); 7X, 37X (2,4,5, 6-13C4; 3,9-15N2); 37U (2,4,5,6-13C4; 9-15N); and AAMU (2, 4,5,6-13C4; 1,3-15N2; 6-amino-15 N). Internal standards 13X [1,3-(methyl-2H3)2] and 137X [1,3,7-(methyl-2H3)3] were obtained from CDN Isotopes (Pointe-Claire, QC, Canada). All internal standards contained