Quantification of urinary AICAR concentrations as a matter of doping controls
- First Online:
- Cite this article as:
- Thomas, A., Beuck, S., Eickhoff, J.C. et al. Anal Bioanal Chem (2010) 396: 2899. doi:10.1007/s00216-010-3560-8
- 438 Views
Influencing the endurance in elite sports is one of the key points in modern sports science. Recently, a new class of prohibited substances reached in the focus of doping control laboratories and their misuse was classified as gene doping. The adenosine monophosphate activated protein kinase activator 5-amino-4-imidazolecarboxyamide ribonucleoside (AICAR) was found to significantly enhance the endurance even in sedentary mice after treatment. Due to endogenous production of AICAR in healthy humans, considerable amounts were present in the circulation and, thus, were excreted into urine. Considering these facts, the present study was initiated to fix reference values of renally cleared AICAR in elite athletes. Therefore a quantitative analytical method by means of isotope-dilution liquid chromatography (analytical column: C6-phenyl) coupled to tandem mass spectrometry, after a sample preparation consisting of a gentle dilution of native urine, was developed. Doping control samples of 499 athletes were analysed, and AICAR concentrations in urine were determined. The mean AICAR value for all samples was 2,186 ng/mL with a standard deviation of 1,655 ng/mL. Concentrations were found to differ depending on gender, type of sport and type of sample collection (in competition/out of competition). The method was fully validated for quantitative purposes considering the parameters linearity, inter- (12%, 7% and 10%) and intraday precision (14%, 9% and 12%) at low, mid and high concentration, robustness, accuracy (approx. 100%), limit of quantification (100 ng/mL), stability and ion suppression effects, employing an in-house synthesised 13C5-labelled AICAR as internal standard.
KeywordsGene dopingSports drug testingLC-MS/MMass spectrometry
The adenosine monophosphate activated protein kinase (AMPK) activator 5-amino-4-imidazolecarboxyamide ribonucleoside (AICAR) has attracted the spotlight of sports drug testing due to attestable enhancement of physical performance even without exercise . The orally active agent alone or in combination with a PPARδ agonists (e.g. GW1516) mimics exercise by promoting oxidative processes in the myofibers of the skeletal muscle and increases the endurance by activation of the AMPK signalling pathway . Various mechanisms of these processes were described by genetic up-regulation of transcriptional regulators and increased gene expression in skeletal muscle. Untrained mice that were treated with doses of 500 mg kg−1 day−1 of AICAR over 4 weeks significantly improved their running endurance in a treadmill experiment by 44% . These results confirm former studies with 5 days of AICAR treatment of rats and subsequently increased levels of glycogen, GLUT4 and mitochondrial enzymes [3, 4]. Generally, as drug compound AICAR has reached phase II of clinical trials and with the described effects, it is proposed to treat metabolic disorders, obesity or related diseases [3, 4].
Considering these facts, the World Anti-Doping Agency (WADA) prohibited the substances AICAR and GW1516 recently and classified their misuse as gene doping . In contrast to the entirely artificial and synthetic compound GW1516, a simple qualitative determination of AICAR in biological fluids is not sufficient for doping control purposes due to the endogenously produced amounts appearing in the circulation [6–8]. Earlier studies for determination from urine were based on capillary electrophoresis or 2D-HPLC-UV with preconcentration on boronic acid substituted silica [8, 9]. Unfortunately, those procedures suffer from the incompatibility to mass spectrometry, which represents the state-of-the-art analyzer in doping controls. Owing to the fact that considerable amounts of AICAR were renally cleared and excreted into urine in healthy humans, the scope of the present study was the development of a simple and fast analytical assay to quantitatively determine urinary concentrations of AICAR and establish reference values for healthy humans and athletes. These reference levels will serve as basis to find threshold values that potentially uncover the exogenous application of AICAR. Quantitative determination was realised by liquid chromatography coupled to tandem mass spectrometry after a sample preparation procedure based on gentle dilution of the native urine. For quantification, a 13C-labelled internal standard was chemically synthesised that ideally compensates all possible matrix effects.
Materials and methods
Reagents and chemicals
Acetic acid (glacial), acetonitrile (analytical grade), ammonium acetate (p.a.), sodium dihydrogenphosphate dihydrate (p.a.), disodium hydrogenphosphate dodecahydrate (p.a.) and sodium chloride (p.a.) were obtained from Merck (Darmstadt, Germany). Tris(carboxyethyl)phosphine hydrochloride, AICAR and urea were from Sigma (Deisendorf, Germany). All solutions, solvents and buffers were prepared with pure water in MilliQ quality. Products for synthesis of labelled ISTD were used as p.a. quality from different commercial providers: [13C5]-d-ribose, acetic anhydride, pyridine, ethanol, hypoxanthine, ammonium acetate, N-methyl-N-trifluoroacetamide, dichloroethane, trimethylsilyl trifluoromethanesulfonate, sodium hydrogen carbonate, inosine, 2-methoxyethoxymethyl chloride and diisopropylethylamine, sodium hydroxide and dichloromethane.
Synthesis of labelled ISTD
Under Ar atmosphere, 370 mg hypoxanthine (2.7 mmol) and 50 mg of ammonium acetate were refluxed in 5 mL N-methyl-N-trifluoroacetamide until dissolved and then concentrated under reduced pressure to yield bis(trimethylsilyl)hypoxanthine. According to Saito et al. , the persilylated hypoxanthine and 460 mg 1 (1.4 mmol) were dissolved in 5 mL dichloroethane, 600 µL trimethylsilyl trifluoromethanesulfonate (3.1 mmol) were added, and the mixture refluxed for 2.5 h under Ar. Quenching with saturated aqueous sodium hydrogencarbonate solution and extractive workup yielded [13C5]-2′,3′,5′-tri-O-acetylinosine (Fig. 1: 2). Yield, 200 mg (0.5 mmol, 35%).
The labelled inosine derivative was converted to [13C5]-AICAR following the method described by Kohyama et al. . Briefly, protected inosine 2 was treated with 2-methoxyethoxymethyl chloride and diisopropylethylamine to yield [13C5]-2′,3′,5′-tri-O-acetyl-1[(2-methoxyethoxy)methyl]inosine, which was deacetylated by alkaline hydrolysis using aq. NH3 (28%)/methanol (1:2, v/v). Refluxing the product in aqueous sodium hydroxide (0.2 N) yielded [13C5]-5-amino-1-β-d-ribofuranosylimidazole-4-carboxamide ([13C5]-AICAR) by intramolecular rearrangement. The product was purified by column chromatography (silica gel, CH2Cl2/MeOH, 3:1). Overall yield, 51 mg (0.194 mmol). Purity and identity were tested for possible impurities, and especially non-labelled AICAR was not found with analysis by means of liquid chromatography coupled to high-resolution/high mass-accuracy tandem mass spectrometry.
Stock and working solutions
Five milligrams of AICAR and labelled AICAR (ISTD) reference substance was dissolved in 5 mL of acetonitrile and used as stock solution. Appropriate dilution with phosphate-buffered saline (PBS) was performed for the preparation of AICAR working solutions. The PBS buffer (pH 7.4) was prepared by dissolving 8 g of sodium chloride, 2.9 g of disodium hydrogenphosphate dodecahydrate and 0.2 g of sodium dihydrogenphosphate dihydrate in 1 L of water.
ISTD working solution was diluted to a final concentration of 50 µg/mL with PBS.
A solution containing 10 mg/mL of urea was used as artificial urine matrix and fortified with ISTD. Respective volumes of AICAR working solution diluted with artificial matrix were prepared to obtain calibration solutions in the aimed working range of the method (100 to 5,000 ng/mL).
A set of 499 urine samples from elite athletes were analysed in order to fix reference values for healthy humans (373 men and 126 women). These samples were drawn as 360 in competition (IC) and 141 out of competition (OOC) specimens and enclose various disciplines. All samples were routinely analysed in common doping control procedures with negative test results. Urine samples after administration of AICAR were not analysed.
Aliquots of urine samples (150 µL) were centrifuged for 3 min at 17,000×g, and 100 µL of the supernatant was transferred into a LC vial. Subsequently, 5 µL of ISTD working solution and 95 µL of PBS were added and gently mixed. A volume of 2 µL of this solution was used for injection into the LC-MS/MS system.
In order to consider the influence of the specific gravity of each sample on quantification, the concentrations were corrected to a standard urine density of 1.020 g/mL according to the WADA Technical Document TD2004NA . The measurements were performed on a PAAR DMA38 resonance frequency transducer with automatic sampling.
Liquid chromatography–mass spectrometry
Mass spectrometric parameters
Precursor ion M + H+ (m/z)
Fragment ion (m/z)
Collision offset (V)
Declustering potential (V)
Due to the lack of available blank urine, artificial urine matrix (10 mg/mL of urea) was used for the preparation of the calibration solutions.
The unambiguous identity of the analyte occurring in the urine was proven by measurements of native urine samples and comparison of MRM transition peak area ratios and retention time to the respective reference compound. Additionally, measurements with liquid chromatography coupled to high-resolution/high-accuracy mass spectrometry were performed.
The linearity of the method was tested in the range between 100 and 7,500 ng/mL (100, 500, 1,000, 1,500, 2,000, 3,000, 5,000 and 7,500 ng/mL) by fortifying the artificial urine matrix by the respective amounts of AICAR reference compound.
The intraday precision was demonstrated at low (native urine with approximately 500 ng/mL endogenously produced AICAR), middle (fortified urine at 1,500 ng/mL) and high (fortified urine at 4,500 ng/mL) concentration levels by analysing ten replicates for each concentration level. Another set of ten replicates were prepared and analysed on another day by another person in order to determine the interday precision.
The stability of the analyte in urinary matrix was tested at 23 °C (room temperature), 4 °C and −20 °C with real fortified urine samples (concentration, 3,000 ng/mL) and analysis of all samples after storage for 1, 2, 3, 6, 7 and 12 days.
Calculation of the theoretical amount and comparison to the actually quantified value was performed in order to determine the accuracy of the method by a threefold determination at three concentration levels (750, 1,500 and 4,000 ng/mL) with fortified urine samples. Quantification was performed by linear regression of the peak area ratios from the external calibration curve (100 to 5,000 ng/mL) and the ISTD.
Limit of detection and limit of quantification
The limit of detection (LOD) was estimated by the evaluation of the signal-to-noise (S/N > 3) in the chromatogram of the respective reference compound. The limit of quantification (LOQ) was evaluated in the same manner at first (S/N > 9), and additionally, a tenfold determination at the estimated level (100 ng/mL) with fortified urine samples was performed in order to show the precision at the LOQ level.
Ion suppression effects were examined by infusing AICAR reference compound (10 µg/mL) via post column split and injecting four regularly prepared urine samples containing ISTD .
AICAR values were summarised using standard descriptive statistics in terms of means and standard deviations. Analysis of variance was used to compare AICAR values between males and females, sport types and sampling types (IC vs. OCC). Residual plots and histograms were used to evaluate model assumptions. The Levene test was used to compare the variability between groups . A regression analysis approach was used to determine quantile values of AICAR, which could be used to identify “abnormally high” (presumably due to doping) cases. The AICAR quantile values were computed as the upper 95% and 99% confidence interval limits of the corresponding predicted values. Since the distribution of AICAR values was skewed to the left, the log-transformed AICAR values were tested for log-normal distribution (Shapiro, Kolmogorov–Smirnov and Anderson tests).These tests suggested (p > 0.00001) that a log-normal distribution represents not a valid model for the dataset, but a skew-normal distribution obviously fits to the experimental AICAR values . A complete case analysis was performed. Values less than LOQ were not imputed. All data analyses were performed using SAS statistical software, version 9.1 (SAS Inc., Cary, NC, USA). p values less than 0.05 were considered to be statistically significant. All p values were two-sided.
Experimental masses, elemental composition and calculated errors of protonated molecules after collision-induced fragmentation measured with high-resolution/high-accuracy mass spectrometry
M + H+ (m/z)
Analysis of gender
Analysis of sampling type
Table 4 illustrates the means and standard deviations of AICAR levels for IC and OOC sampling types.
The comparison of means between IC and OOC resulted in p = 0.0002 and comparison of standard deviations between males and females was determined with p = 0.0062. Thus, there is a significant difference in both mean and variability between IC and OOC samples.
Analysis of type of sport
There are more than 30 different types of sport, some with a low frequency. The sport types were summarised into three main categories: A, soccer (representing endurance or gaming sports); B, weightlifting (representing strength sports); C, all other sport types.
This categorisation was based on performing a regression analysis, and results were illustrated in Table 4. Comparison of means between sport types was calculated with p < 0.0001, and comparison of standard deviations between sport types yield p = 0.0053. There is a significant difference in both mean variability between sport types .
Determination of quantile 95 and 99 values for AICAR
A regression analysis approach with skew normally distributed error terms was used to determine quantile 95 and 99 values of AICAR, which could be used to identify “abnormally high” (presumably due to doping) cases. Based on univariate analysis, it was determined that gender, probe type and certain sport types (soccer and weightlifting) may influence AICAR excretion . Results were summarised in Table 4.
Considering the fact that there were no female soccer players, the values were computed as the upper 95% and 99% quantiles of the corresponding predicted values.
If the assumptions are correct (i.e. random samples from a normal population), then the values listed in Table 4 are the values to achieve 95% and 99% specificity for the various sub-groups. To evaluate the sensitivity, there is a need of samples deriving from an “abnormal” (i.e. doping cases or post administration samples) population (see “Doping control aspects”).
Especially for quantitative methods, a proper validation is of utmost importance. Although a standard addition procedure would provide best results for each sample, an external calibration strategy was chosen in order to increase the effectiveness of the method. The utilisation of the labelled ISTD ensures best possible performance for quantification via calibration curve considering peak area ratios of the target analyte to the ISTD. Main results for validation are summarised in Table 2.
Qualitative identification of urinary AICAR by rel. abundances of the peak areas from diagnostic fragment ions and retention time
Rel. abundance (%)
Rel. abundance (%)
Linearity was tested in the concentration ranges between 100 and 7,500 ng/mL and permitted the approximation with a simple linear model with a coefficient of correlation R = 0,998. Urine samples with values greater than 7,500 ng/mL (even after density correction) should be diluted with artificial urine matrix adequately.
This parameter indicates the systematic errors in the method that would falsify the results. Accuracy was determined by comparison of the measured concentration to the theoretically calculated values. Analysis at three different levels (∼750, 1,500 and 4,000 ng/mL) yielded a mean accuracy of 103% with a mean relative standard deviation of below 10%.
A standard addition procedure provides the best option for quantification if a blank matrix is not available. Unfortunately, the procedure is rather laborious, and at least a five-fold preparation for each sample is required to obtain reliable results. The comparison of the two different quantification strategies for three samples with different concentrations provided no significant differences (<10%) for the obtained results.
Precision at low, medium and high concentration was tested in order to verify that the relative standard deviations of the measured concentrations were in acceptable ranges and that the results are independent from concentration and day of preparation. Intraday precision was calculated below 12%, and interday precision was below 14%.
Limit of detection and limit of quantification
The LOD was estimated via signal-to-noise (S/N) with approximately 30 ng/mL. As the more important parameter, the LOQ was estimated with 100 ng/mL considering a sufficient S/N of more than 10. Additionally, the precision, determined at this concentration level (100 ng/mL), was calculated with less than 12%.
The common storage condition of urinary doping control samples is refrigeration, and, thus, stability was tested by storage at room temperature, 4 °C and −20 °C in order to cover all possible scenarios.
For the chosen conditions, there was no significant decrease of AICAR observed after 3 days at RT, 7 days at 4 °C and more than 12 days at −20 °C.
Although the labelled ISTD would ideally compensate possible matrix effects, the method was tested for ion suppression impacts during electrospray ionisation. The visual evaluation of the chromatogram in the diagnostic MRM ion traces provided no significant ion suppression at the respective retention time. This was also valid for the evaluation of the MRM ion trace for the ISTD in real urine samples.
Doping control aspects
AICAR has not yet reached the status as a approved pharmaceutical product, but availability for cheating sportsmen in certified purity is not hindered due to a widespread distribution by the chemical industry or even by low-scale synthesis and purification in small underground laboratories.
Renal clearance of endogenously produced AICAR is described only barely, and to date, no reference values for excretion in healthy humans were reported. Elevated amounts of excretion are known in association with vitamin B12 and folic acid deficiencies due to impairing the AICAR transformylase, as well as in leukaemia patients and patients with hypoxanthine–guanine phosphoribosyl-transferase deficiency . Despite the lack of secured data for the performance enhancement after treatment in healthy humans, the mouse model experiments provided a reasonable relevancy for potential misuse in elite sports. The amounts of AICAR applied to the animals were 500 mg kg−1 day−1, and endurance increases significantly . The metabolic fate is described via a common purine pathway including phosphorylation to AICAR-5`monophosphate, degradation to hypoxanthine, xanthine and finally to uric acid. Pharmacokinetic studies on disposition and metabolism of AICAR after intravenous (25, 50 or 100 mg/kg) application to healthy humans showed a fast plasma clearance and renal excretion into urine of the intact analyte in ranges of 5% to 8% of the applied dose [16, 17]. Thus, expected concentrations in urine will peak at more than 100 µg/mL, at least for the time period of approximately 10 h after application. These levels are approximately tenfold higher than the highest concentrations found in the athletes samples (Fig. 2). For the orally applied drug compound, a lower bioavailability is described, but this is compensated by five- to tenfold higher doses [1, 16, 17]. Based on the presented data and the statistical evaluation, analytical findings of AICAR values above 20 µg/mL would be considered inconsistent with an endogenous production in healthy humans.
The endogenous nature of the analyte impedes the direct identification as exogenous compound, and reliable analytical methods are required to provide more characteristics for the unambiguous determination. Pilot studies by means of gas chromatography/isotope-ratio mass spectrometry after potential sample preparation, including a solid phase extraction with subsequent liquid chromatographic fractionation and derivatisation, have outlined the potential to differentiate the origin of AICAR by the 13C/12C ratios of the chemically synthesised reference compound and the endogenously produced AICAR. This hyphenated analytical approach provides more detailed information and supports the identification of artificial AICAR in urine with an independent analytical method.
AICAR has a reasonable relevancy as performance enhancing agent, and thus, it was prohibited by the WADA according to their actual list of banned compounds. The illicit use of this compound was classified as gene doping, but reference values for endogenous AICAR excreted into urine were not evaluated yet. The present study provides data of AICAR levels in urine of elite athletes, in order to fix reference values for further investigations. The data clearly show that AICAR excretion depends on gender, type of sport and type of sampling (IC, OOC), but will never reach concentrations, that occur in urine after i.v. application.
Highly sensitive mass spectrometers enabled the determination without laborious purification, desalting or concentration steps during sample preparation and allow analytics in a high-throughput approach. Implementation into existing doping control procedures is not hindered due to commonly used instrumentation and simple sample preparation procedure.
The study was carried out with the support of the Manfred Donike Institute for Doping Analysis, Cologne, Germany, the Federal Ministry of the Interior of the Federal Republic of Germany and the Federal Office of Sports, Bern, Switzerland.