Determination of Guanidinoacetate and Creatine in Urine and Plasma by Liquid Chromatography–Tandem Mass Spectrometry

In the last 10 years, three new inborn errors of creatine metabolism and transport, called creatine deficiency syndromes (CDS), have been described (1). These are deficiencies of arginine:glycine amidinotransferase (EC 2.1.4.1), S-adenosyl-l-methionine:guanidinoacetate N-methyltransferase (GAMT; EC 2.1.1.2), and creatine transporter. Patients with CDS have mental retardation, severe speech disturbance, and depending on the disorder, epilepsy and/or extrapyramidal signs. Biochemical detection of CDS relies on the determination of two main metabolites in biological fluids: guanidinoacetate (GAA) and creatine. Few patients with CDS have been reported to date, probably because of underdiagnosis, most likely attributable to the limited availability of quantitative GAA and creatine assays. This underdiagnosis is particularly harmful because some patients improve with oral creatine supplementation. Several analytical methods for GAA and creatine have been described, including use of the Sakaguchi reaction (2), stable-isotope-dilution gas chromatography–mass spectrometry (GC/MS) (3)(4), and HPLC (5). Recently, a liquid chromatography–tandem mass spectrometry (LC/MS/MS) method has been reported for analysis of these metabolites in plasma and dried blood on filter cards (6). Here we describe the validation of LC/MS/MS for the simultaneous quantification of GAA and creatine in urine and plasma and compare the method with a stable-isotope-dilution GC/MS method. To investigate the molecular fragmentation and to prepare calibrators, we prepared 5 mmol/L stock solutions and successive dilutions of creatine, GAA (Sigma-Aldrich), N-methyl-d3-creatine (CDN Isotopes), and [13C2]GAA (Dr. H. Ten Brink, VU University Medical Center, Amsterdam, The Netherlands) in distilled water. We evaluated MS/MS conditions in the positive-ion mode by infusion of aqueous solutions into the API 2000 tandem mass spectrometer (Sciex Applied Biosystems) as described previously (7). Nitrogen was used as both the curtain and collision gas. The transitions chosen for the quantification of GAA, creatine, d3-creatine, and [13C2]GAA (multiple-reaction monitoring mode) were m/z 174→101 [declustering potential (DP), 15V; collision energy (CE), 18V]; 188→90 (DP, 20V; CE, 26V); 191→93 (DP, 20V; CE, 26V); and 176→103 (DP, 15V; CE, 20V), respectively. We mixed 50 μL of samples (0, 50, and 500 μmol/L aqueous calibrators; plasma; and 10-fold-diluted urines) with 50 μL of water, 30 μL of 50 μmol/L d3-creatine, and 30 μL of 50 μmol/L [13C2]GAA. We added 400 μL of ethanol (Carlo-Erba) and 400 μL of n-hexane (Merck) to each sample. The vials were then mixed for 2 min. The hexane was removed, and vials were centrifuged for 5 min at 17 530g. The supernatants were transferred to new vials and evaporated to dryness under nitrogen at room temperature. The remaining residues were subsequently derivatized with 200 μL of 3 mol/L HCl in n-butanol (Regis Technologies Inc.) for 15 min at 65 °C. Samples were dried again under nitrogen at 45 °C, and 500 μL of mobile phase was added. We injected 5 μL in the LC/MS/MS system. Liquid chromatography was performed with a 2.1 × 50 mm Symmetry C18 HPLC column (3.5-μm particle size; Waters). The column flow rate was 0.150 mL/min, and the mobile phase consisted of acetonitrile (200 mL/L in water) acidified with formic acid (0.5 mL/L). A Series 200 micropump and autosampler (Perkin-Elmer) were used for solvent delivery and sample introduction. The elution time was 3 min. The eluate was injected in the TurboIonSpray probe (350 °C) of the mass spectrometer, and results were acquired with Analyst (Ver. 1.3.1) software (Sciex Applied Biosystems). GAA and creatine were quantified relative to the [13C2]GAA and the d3-creatine internal standards, respectively. Calibration curves were constructed by linear regression analysis of the ratios of the GAA or creatine calibrator to the respective internal standard. The calibration curves (0–500 μmol/L) covered the range of GAA and creatine concentrations typically found in urine (when diluted 10-fold) or plasma (8). We monitored intraday (n = 5) precision of the calibration curves. The mean slope, intercept, and coefficient of linear regression for creatine were 0.039 (95% confidence interval, 0.038–0.040), 0.047 (0.017–0.077) μmol/L, and 0.998, respectively. The mean slope, intercept, and coefficient of linear regression for GAA were 0.037 (95% confidence interval, 0.035–0.038), −0.003 (−0.039 to 0.033) μmol/L, and 0.998, respectively. The intraday imprecision (CV) of the global procedure (extraction and quantification) was estimated on one sample extract diluted 10-fold. The CVs were 10% for GAA and 8% for creatine in urine and 20% for GAA and 3% for creatine in plasma. The intraday CVs for the single LC/MS/MS quantification step (10 measurements of the same sample extract in the same run) were 4% for GAA and 3% for creatine in urine and 9% for GAA and 2% for creatine in plasma. Interday imprecision was assessed on the same urine and plasma samples extracted and injected on 5 days. The CVs were 10% for GAA and 9% for creatine in urine and 13% for GAA and 9% for creatine in plasma. The limit of detection (signal-to-noise >3) and limit of quantification (signal-to-noise >10) were measured on successive aqueous dilutions of GAA and creatine solutions. The limits of detection were 0.01 μmol/L for creatine and 0.025 μmol/L for GAA. The limits of quantification were 0.05 μmol/L (CV = 7%; n = 5) for creatine and 0.1 μmol/L (CV = 6%; n = 5) for GAA. To estimate linearity, we added 25, 50, 125, 250, 500, 1250, and 2500 μmol/L GAA and creatine to controls. For urine, the assay was linear for creatine between 0 and 1250 μmol/L and for GAA between 0 and 2500 μmol/L in a 10-fold diluted sample, corresponding to 12 500 and 25 000 μmol/L, respectively, in the original sample. In plasma, the assay was linear between 0 and 1250 μmol/L for creatine and between 0 and 500 μmol/L for GAA. To estimate recovery, we added two concentrations of GAA and creatine (n = 5 for each concentration) to urine and plasma. The recovery in urine of 250 and 500 μmol/L GAA was 85% (CV = 14%) and 78% (CV = 14%), respectively. The recovery in urine of 250 and 500 μmol/L creatine was 96% (CV = 7%) and 89% (CV = 5%), respectively. In plasma, the recovery of 20 and 40 μmol/L GAA was 131% (CV = 8%) and 120% (CV = 8%), respectively, and recovery of 100 and 200 μmol/L creatine was 115% (CV = 4%) and 124% (CV = 8%), respectively. We analyzed 35 urine and 21 plasma samples with the described LC/MS/MS method and a stable-isotope-dilution GC/MS method (3). Deming regression and Bland–Altman plots (Fig. 1 ) revealed excellent agreement be-tween the two methods for creatine and GAA in urine and creatine in plasma. For GAA in plasma, the less satisfying results are probably attributable to a lack of sensitivity of our tandem spectrometer. Comparison of the LC/MS/MS method with a GC/MS method for urine and plasma. (A), creatine in urine (mmol/mol creatinine; n = 35); (B), creatine in plasma (μmol/L; n = 21); (C), GAA in urine (mmol/mol creatinine; n = 35); (D), GAA in plasma (nmol/L; n = 21). The solid lines indicate the Deming regression; the dashed lines indicate the lines of unity. The insets show the Bland–Altman plots of the differences vs the means of paired values for the LC/MS/MS method and the GC/MS method. For creatine in urine and plasma, the regression equations were as follows (values in parentheses are the SE): urine, y = 1.11 (0.06)x + 1.58 (4.88) mmol/mol creatinine (r = 0.98); plasma, y = 1.13 (0.08)x − 1.91 (4.85) μmol/L (r = 0.91). The mean differences were 67.3 mmol/mol creatinine for urine and 7.81 μmol/L for plasma. For GAA in urine and plasma, the regression equations were as follows: urine, y = 1.13 (0.11)x + 0.92 (2.6) mmol/mol creatinine (r = 0.96); plasma, y = 0.87 (0.36)x + 0.79 (0.47) μmol/L (r = 0.42). The mean differences were 951 mmol/mol creatinine for urine and 0.58 μmol/L for plasma. To assess the ability to detect CDS patients, we analyzed a urine sample from a 21-year-old patient with GAMT deficiency and compared the results with the values for controls. Reference values were estimated by use of control urines (n = 47; age range, 5–23 years of age) with nonparametric analysis (5th and 95th percentiles). As expected, urinary GAA was increased (386 mmol/mol creatinine; values for control group, 7–88 mmol/mol creatinine), and urinary creatine was in the range of values obtained for the controls (34 mmol/mol creatinine; values for control group, 12–585 mmol/mol creatinine). In summary, we have described and validated a LC/MS/MS method for the diagnosis of CDS-affected patients that quantifies GAA and creatine in urine and plasma by use of stable-isotope-labeled creatine and GAA as internal standards. This analytical method shows advantages of speed, specificity, linearity over large concentration ranges, and comparability with a stable-isotope-dilution GC/MS method. We thank Dr. Maria Luis Cardoso from Instituto de Genetica Medica Jacinto de Magalhaes (Porto, Portugal) for providing a urine sample from a patient with GAMT deficiency.