| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Automation and Analytical Techniques |
1 Academic Medical Center, University of Amsterdam, Laboratory Genetic Metabolic Diseases, Emma Children’s Hospital and Department of Clinical Chemistry, Amsterdam, The Netherlands.
2 Erasmus Medical Center, Department of Clinical Genetics, Rotterdam, The Netherlands.
aAddress correspondence to this author at: Laboratory Genetic Metabolic Diseases, F0-224, PO Box 22700, 1100 DE Amsterdam, The Netherlands. Fax 31-20-6962596; e-mail w.kulik{at}amc.uva.nl.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: Urine sample size was adjusted to the equivalent of 100 nmol of creatinine. After addition of 2-keto-3-deoxy-D-glycero-D-galactonononic acid as internal standard, samples were diluted with water to an end volume of 250 µL. We used 10 µL for HPLC–tandem mass spectrometric analysis in the negative electrospray ionization mode, monitoring transitions m/z 308.3
m/z 86.9 (sialic acid) and m/z 267.2m/z 86.9 (internal standard). The overall method was validated and studied for ion suppression, interfering compounds, and pH effects. Samples from controls (n = 72) and SSD patients (n = 3) were analyzed.
Results: The limit of detection was 3 µmol/L. Intraassay imprecision (CV; n = 10) was 6%, 3%, and 2% at 30, 130, and 1000 mmol/mol creatinine, respectively; corresponding interassay CV (n = 10) were 5%, 5%, and 2%. Recovery was 109% (100–1000 mmol/mol creatinine). The mean (SD) [range] excretion rates (mmol/mol creatinine) were 31.3 (16.6) [0.7–56.9] at 0–1 year (n = 20), 21.2 (9.8) [6.3–38.3] at 1–3 years (n = 15), 14.4 (8.2) [1.7–32.9] at 3–10 years (n = 25), and 4.6 (2.6) [0–9.8] above age 10 years (n = 12). SSD patients 1.2, 3.9, and 12 years of age had concentrations of 111.5, 54.2, and 36.1 mmol/mol creatinine, respectively.
Conclusions: The HPLC-tandem MS method for free sialic acid in urine is more rapid, accurate, sensitive, selective, and robust than earlier methods and may serve as a candidate reference method for free sialic acid in diagnosis of SSD.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The known forms of SSD are primarily caused by a transport defect across the lysosomal membrane (3)(7), which leads to the accumulation of free sialic acid [N-acetylneuraminic acid (NANA); Fig. 1
] in tissues, fibroblasts, and urine. The origin of the transport defect lies in a mutation in the SLC17A5 gene, which encodes for sialin protein (8)(9).
|
Defective free sialic acid transport can be established by quantitative analysis of NANA in urine relative to the concentration of creatinine. Several types of analysis are available to quantify NANA, including colorimetric methods (10)(11), thin-layer chromatography (12), gas chromatography–mass spectrometry (12)(13), HPLC with fluorescence detection (14), HPLC with ultraviolet detection (15), enzymatic assays (14)(16), high-performance anion-exchange chromatography with pulsed amperometric detection (HPAE-PAD) (17)(18), and HPLC with mass spectrometry (HPLC-MS) (19)(20). The main disadvantages of these methods are a lack of selectivity, a lack of speed, or both. For example, enzymatic methods appear to be more accurate than colorimetric methods (16)(21), but neither method can differentiate among different neuraminic acids; HPAE-PAD, used mainly for the analysis of sialic acids released from glycoconjugates, can reduce the possibility of interferences by relatively long gradient separations. HPLC-MS approaches (19)(20) are more selective and sensitive, but they are still time-consuming, multistep methods that have not been validated as a rapid diagnostic tool.
In this report we describe and validate a simple, rapid, selective, and sensitive method for the analysis of NANA in urine samples based on HPLC combined with tandem MS (HPLC-tMS). The deaminated sialic acid 2-keto-3-deoxy-D-glycero-D-galactonononic acid (KDN) was used as an internal standard (IS) (17)(22) (see Fig. 1
).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
patient and control samples
Samples were collected from control individuals (n = 72) who were examined in our institute for exclusion of inherited metabolic disease. All had normal NANA excretion, as determined by thin-layer chromatography. In addition, three patients with previously established increased free NANA excretion were investigated. These patients were confirmed to have SSD. The condition in patient 1 was detected at birth because he had an affected older brother. Both were severely mentally retarded and had impaired speech and ataxia. Patient 1 was 14 months of age when his urine was collected (23). Patient 2 was born to a consanguineous couple. He was referred at the age of 4 years because of psychomotor retardation and a mild facial dysmorphism. Patient 3 was originally studied at the age of 2 years because of psychomotor retardation, aggressive behavior, spastic/atactic gait, and a slight facial dysmorphism. Urine sialic acid analysis was not performed then. Reevaluation at the age of 12 years gave the correct diagnosis.
Samples from all of the patients were initially analyzed for urinary free sialic acid and sialic acid-containing oligosaccharides by thin-layer chromatography with Bial staining. Confirmation was obtained by measurement of the sialic acid content of cultured fibroblasts and mutation screening (1).
All samples were collected according to the institutional guidelines for sampling, including obtaining informed consent from the persons involved or their representatives.
preparation of is and calibrators
A NANA stock solution was prepared by dissolving NANA (0.81 mmol/L) in distilled water. The IS stock solution was prepared by dissolving KDN (0.075 mmol/L) in distilled water. Calibrators were prepared by adding 50 µL of IS and a range of volumes of NANA stock solution to 100 µL of a large urine sample containing 1 mmol/L creatinine (100 nmoles of creatinine).The mixtures were diluted with distilled water to a volume of 250 µL.
sample preparation
Sample size was adjusted to the equivalent of 100 nmol of creatinine (initial creatinine concentrations were established by colorimetric Jaffe assay). IS (50 µL of a 0.075 mmol/L solution) was added, and samples were diluted with distilled water to an end volume of 250 µL. A 10-µL aliquot of the final sample was loaded on the HPLC column for tMS analysis.
instrumentation
HPLC.
The HPLC system consisted of a HP1100 series binary gradient pump, a vacuum degasser, a column temperature controller (all from Hewlett Packard), and a Gilson 231 XL autosampler (Gilson). Sample (10 µL) was loaded on a LiChrospher amino propyl analytical HPLC guard column [20 x 2 (i.d.) mm; 5-µm particle size; Supelco, Sigma-Aldrich Chemie B.V.], and the column temperature was maintained at 22 °C. Both NANA and KDN were eluted with a flow rate of 0.3 mL/min and a linear gradient between solution B (acetonitrile) and solution A (water containing 0.114 g/L aqueous ammonia). The gradient was programmed as follows: 0–2 min, 100% B; 2–2.5 min, gradient to 10% B; 2.5–6 min, hold at 10% B; 6–6.1 min, gradient back to 100% B; 6.1–10 min, 100% B with a flow rate of 0.5 mL/min to equilibrate the column. All gradient steps were linear, and the total analysis time, including the equilibration, was 10 min. A splitter between the HPLC column and the mass spectrometer allowed introduction of eluate at a flow rate of 30 µL/min. An electronically operated valve allowed the eluate to enter the mass spectrometer between 4 and 8 min only.
MS.
A Quattro II triple-quadrupole mass spectrometer (Micromass) was used in the negative electrospray ionization (ESI) mode. Nitrogen was used as nebulizing gas, and argon was used as collision gas at a pressure of 0.25 Pa. The capillary voltage used was 3 kV. The source temperature was set at 80 °C, and the optimum cone voltage was 20 V. Both NANA and KDN were measured by multiple reaction monitoring (MRM) in the negative-ion mode, using transitions m/z 308.3m/z 86.9 for NANA and m/z 267.2m/z 86.9 for KDN with an optimal collision energy of 20 eV.
validation
The HPLC and MS settings were optimized by use of a 20 µmol/L solution of NANA and KDN in water. The linearity of the method and the detection limit for NANA were established by injection of calibrators. For the analysis of all control and patient samples, the equivalent of 100 nmol of creatinine was used in the assay.
Independent quality-assurance samples from the European Research Network for Evaluation and Improvement of Screening, Diagnosis and Treatment of Inherited Disorders of Metabolism (ERNDIM) were used to assess the standardized procedure for quantification of NANA in urine.
We determined the intraassay (within-day) variation of the method by measuring 10 times a sample enriched with NANA at 97 mmol/mol creatinine (urine+) and a sample enriched with 970 mmol/mol creatinine (urine++). We determined the interassay (between-day) variation by measuring blank samples and samples enriched with NANA at 97 and 970 mmol/mol creatinine during 3 separate weeks.
We evaluated the recovery of the method by measuring 10 different samples before and after enrichment with a known amount of NANA and by measuring external quality-control samples.
Ion suppression by interfering substances was investigated by comparison of the peak intensities of both NANA and KDN in the enriched samples relative to NANA and KDN dissolved in water at the same concentrations.
We studied the effect of pH by measuring urine samples 10 times to establish the mean (SD) NANA values in those samples. Acidic or basic solutions were then added to the urine samples to change the pH, and NANA was again measured as described.
In all experiments, no special precautions (e.g., cleaning of the high-voltage lens and sample cone) were taken to optimize the detection limit of the MS system for this method.
statistics
The data were analyzed by statistical tools that are part of the Microsoft Excel 2000 and SPSS 11.5 packages. Subgroup distributions were tested with the Kolmogorov–Smirnov and Shapiro–Wilk tests for normality. For the comparison of gaussian-distributed groups, we used a t-test. A P value <0.05 was considered significant. Data are expressed as means (SD). A sample concentration was assigned as increased if it exceeded the mean + 3 SD of the age-related group.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
).
|
The daughter ion at m/z 86.9 was the most abundant product for both NANA and KDN; therefore, NANA and KDN were measured in the MRM mode through the transitions m/z 308.3m/z 86.9 and m/z 267.2m/z 86.9, respectively (Fig. 3
).
|
Using directly injected (no column) urine samples, we analyzed urine and water samples to which NANA had been added at the same concentration as in the calibrators. Comparison of the detector response showed a reduction of >90% for NANA in the enriched urine samples compared with the enriched water samples. To minimize signal quenching of the analytes, we used a short in-line aminopropyl silica-based guard column to partially purify the urine samples. The analytes were eluted with a linear gradient of acetonitrile and water containing 0.114 g/L ammonia (6.9 mmol/L); NANA and KDN coeluted at a retention time of 
5.7 min.
validation
Background.
We investigated the presence of a sizeable background signal for NANA/KDN by analyzing the chemicals and solutions used to prepare the calibrators and samples. No significant signal was obtained for KDN in nonenriched urine samples. We investigated the possible mutual interference of NANA and KDN by analyzing separately NANA and KDN solutions in distilled water. No significant interferences were observed.
Suppression.
We evaluated the suppression of the analyte response by analyzing a calibration mixture of NANA and KDN in water as well as urine samples enriched with NANA and KDN at the same concentration as in the calibration mixture but with various creatinine concentrations (range, 0–2.5 mmol/L creatinine). Using the HPLC guard column, we compared the response for the urine samples with the response for the calibration mixture in water (see Fig. S-1 in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol50/issue2/). NANA and KDN were suppressed equally. Urine with a creatinine of 0.4 mmol/L showed a signal reduction of 25%.
Linearity.
The ratio of the responses for NANA and KDN (areas under the curves) vs the added amount of NANA in a urine sample containing 0.4 mmol/L creatinine showed a good linear relationship. Linear least-squares regression for the range 0–1.5 mmol NANA/mol creatinine gave the equation: y = 29.9 (0.2)x + 0.3 (0.1) mmol NANA/mol creatinine, where the values are the mean (SE); n = 3; R2 >0.999 (Fig. S-2 in the online Data Supplement).
Limits of detection (LOD).
The LOD were established and defined as the lowest signals detected with a signal-to-noise ratio of 3. The lowest NANA concentrations in the urine samples (range, 10–15 µmol/L) were above the LOD, which was calculated to be 3 µmol/L. The LOD for NANA in water was 2 µmol/L. We determined that a urine sample containing the equivalent of at least 50 nmol of creatinine was needed to prepare control samples to measure NANA with a signal-to-noise ratio >3.
Precision and accuracy.
We assessed the intra- and interassay variation and the recovery of the method at different enrichment concentrations (see Table 1
) and found that NANA could be determined within the limits of ± 2.5 mmol/mol creatinine (95% confidence interval).
|
Quality-control samples.
Comparison of the measured and theoretical values of the ERNDIM quality-control samples (urine accurately enriched with 100, 200, 300, and 400 nmol/L NANA) showed a good linear relationship (R2 = 0.997) and a mean recovery of 108.7% (Fig. S-3 in the online Data Supplement). The NANA concentration in the nonenriched sample was calculated as 23.8 nmol/L.
pH.
We investigated the effect of pH by analyzing urine samples at pH values of 2, 5, 10, and 12. It appeared that the pH of the urine samples had no significant effect on the outcome of the NANA analysis. The measured NANA/KDN ratio did not exceed the limits of the mean ratio ± 2 SD (obtained by analyzing the same sample 10 times).
Interferences.
NANA was analyzed in selected urine samples with a specific high content of endogenous contaminants, such as blood, protein, or ketones, or with high pH, or in urine samples that were visually dark or otherwise looked abnormal. Urines containing drug metabolites were also tested. The results were compared with the controls, and no significant differences (P >0.35) were observed.
controls and patients
After validation of the method, we measured NANA in 72 control urine samples, and the results were categorized by age. Group I was 0–1 year of age (n = 20), group II was 1–3 years (n = 15), group III was 3–10 years (n = 25), and group IV was 
10 years (n = 12; Table 2
). Statistical tests showed gaussian distributions for all four groups. Increased concentrations of NANA were observed in urine samples from three patients with SSD. The results are summarized in Fig. 4
, which shows the NANA concentration as a function of age.
|
|
Generally, the means and ranges for NANA decreased with increasing age, stabilizing at 10–20 years. The patients with SSD had a 3.8- to 7.8-fold higher NANA concentration than the (mean value of) controls in the same age category and exceeded the interval of (mean + 3 SD) in the corresponding age category.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
12 min/sample (batch of 10 samples), including sample preparation and tMS measurement; requires a typical sample size of only 20 µL; and has selectivity that is not hindered by interfering components.
On the basis of the calibration curves and the calculated inter- and intraassay CV, the method can measure NANA concentrations in urine within the limits of ± 2.5 mmol/mol creatinine (95% confidence interval). However, creatinine concentrations are generally determined by the colorimetric Jaffe assay, which implies that the creatinine value assigned to a urine sample has an imprecision of 3%. As a result, the actual values expressed in mmol/mol creatinine will have an overall CV that is higher than the CV established for the described HPLC-tMS method. Therefore, the values relative to creatinine can be expressed as ± 4.5 mmol/mol creatinine (95% confidence interval), taking into account the imprecision of the reported creatinine value.
Suppression of the signal by salts or other sample components can be a serious problem in ESI MS analyses. As an online purification step, we introduced HPLC with a short aminopropyl column, which allowed interference-free detection of NANA and KDN. However, signal suppression still appeared to be dependent on the creatinine concentration (Fig. S-1 in the online Data Supplement). Creatinine concentrations >0.5 mmol/L suppressed the signal by 50%. Diluting the samples to decrease the creatinine concentration had a net positive effect: it reduced the effect of suppression more than it decreased the NANA/KDN signal as a result of lower concentrations. The best results, in terms of signal-to-noise ratios, were obtained in samples diluted to a creatinine concentration between 0.3 and 0.4 mmol/L. In these diluted samples, the suppression was <30% with a typical signal-to-noise ratio of 30 in controls.
The method appears to be very robust because the overall results were not affected by a range of different variables, such as pH (from pH 2 to pH 12) or substantial amounts of blood protein, ketones, or medications, particularly high concentrations of metabolites from anticonvulsants such as valproate, clonazepam, phenytoin, or diazepam, in the samples.
We observed a slight decrease in absolute NANA concentration in urine samples left at room temperature for a few days or at 4 °C for more than 4 weeks. We therefore recommend keeping samples at -20 °C.
As shown in Fig. 4
and Table 2
, NANA concentrations are age dependent: NANA concentrations decrease with age, and the concentration range becomes narrower with age (11). We analyzed 72 different controls and subdivided them in four different age categories to create possible reference values. P values calculated by comparison of the different age categories (t-test) showed significant differences among the groups, although the overlap was obvious (Table 3
). Among these age categories, we observed a significant difference between controls and patients with SSD (n = 3; Table 2
). In fact, all patients had a NANA concentration at least threefold higher than the mean of the corresponding age-matched controls (Fig. 4
and Table 3
). These data are in good agreement with previously reported data on the urinary excretion of NANA (11)(15). Our data, however, showed systematically lower values (25% lower) than those reported by other research groups. The reliability and accuracy of the HPLC-tMS method is supported by the good agreement between the measured and added amounts of NANA in the ERNDIM quality-control samples. The differences between our measured values and values in the literature might be attributable to the relatively high selectivity of tMS compared with other detection methods used for the analysis of NANA; tMS may exclude interferences that might add to the measured values.
|
In summary, the overall method described compares favorably with the existing methods. The method is rapid, accurate, sensitive, selective, and robust and qualifies as a high-throughput method. We believe that the described HPLC-tMS method for quantification of sialic acids in urine could be a candidate reference method for the diagnosis of SSD.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  E. Roura, C. Andres-Lacueva, R. Estruch, and R. M. Lamuela-Raventos Total Polyphenol Intake Estimated by a Modified Folin-Ciocalteu Assay of Urine Clin. Chem., April 1, 2006; 52(4): 749 - 752. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R Froissart, D Cheillan, R Bouvier, S Tourret, V Bonnet, M Piraud, and I Maire Clinical, morphological, and molecular aspects of sialic acid storage disease manifesting in utero J. Med. Genet., November 1, 2005; 42(11): 829 - 836. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Ma, F. M. Vaz, Z. Gu, R. J. A. Wanders, and M. L. Greenberg The Human TAZ Gene Complements Mitochondrial Dysfunction in the Yeast taz1{Delta} Mutant: IMPLICATIONS FOR BARTH SYNDROME J. Biol. Chem., October 22, 2004; 279(43): 44394 - 44399. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
, patients with SSD. Error bars, SD; solid line, trend; dashed line, trend + 3 SD.