The biochemical detection of catecholamine-secreting tumors relies on the determination of several metabolites, among them vanillylmandelic acid (VMA), a metabolite derived from epinephrine and norepinephrine via the intermediate 4-hydroxy-3-methoxyphenylglycol (1)(2). VMA may also be produced by the oxidative deamination of normetanephrine and metanephrine or by O-methylation of dihydroxymandelic acid, a minor metabolite of norepinephrine (2)(3).
Our laboratory is actively involved in the application of stable-isotope-labeled internal standards and tandem mass spectrometry (MS/MS) to the determination of a wide variety of biochemical markers currently being tested with more conventional analytical platforms. This effort has substantially improved the efficiency and effectiveness of our services and reduced turnaround time in a high-volume testing environment. The following is a report of our development of a MS/MS method for the determination of VMA in urine.
VMA was purchased from Sigma, and d,l-vanillylmandelic acid, ring-2H3 98% (d3-VMA; GC/MS for isotopic enrichment 99.7%) was purchased from Cambridge Isotope Laboratories. OasisTM HLB solid-phase extraction columns were obtained from Waters. All other chemicals and solvents were of the highest purity available from commercial sources and were used without further purification.
Stock solutions of VMA and d3-VMA were prepared by dissolving 5 or 10 mg, respectively, in 10 mL of 0.1 mol/L HCl. A working solution of VMA (20 mg/L) was prepared by diluting stock solutions with 0.1 mol/L HCl.
Calibrators were prepared in deionized water by the addition of working solution (20 mg/L) to VMA concentrations of 0 (blank), 1.04, 2.08, 4.17, 8.33, 12.50, 16.67, and 20.00 mg/L. Prepared calibrators were then treated as specimens for the rest of the analysis. In the selected-reaction monitoring mode, VMA was quantified relative to the internal standard peak area by linear regression analysis.
Sample preparation involved dilution of 0.6 mL of urine sample with 0.6 mL of reverse osmosis water and 10 μL of internal standard solution (d3-VMA; 5 μg) with use of an automated sample processor (Gilson ASPECTM). Solid-phase extraction using this robotic device proceeded as follows: 1 mL (30 mg of packing) Oasis HLB columns were preconditioned with 1 mL each of methanol and water, in that order, before application of 1 mL of the diluted urine sample. After a washing step (0.5 mL of deionized water), VMA and d3-VMA were eluted with 1 mL of methanol and evaporated to dryness under nitrogen in a water bath at 30 °C. The remaining residues were reconstituted in mobile phase (1 mL) consisting of 150 mL of methanol and 850 mL of aqueous formic acid (0.5 g/L formic acid) per liter. The choice of mobile phase was important because acidification is required to separate VMA from the specimen matrix but also causes some signal suppression. The amount of acid added to the isocratic mobile phase was the best compromise that in our estimation allowed adequate separation of VMA without an excessive loss of signal intensity.
An API 2000 tandem mass spectrometer (Applied Biosystems/MDS SCIEX) operated with a pneumatically assisted electrospray ionization source voltage at −5000 V (negative ion mode) was used. Peripherals included a Perkin-Elmer Series 200 quaternary pump and an autosampler. Instrument conditions were as follows: injection volume, 10 μL; source assist gas (zero-grade air) at setting 40; source drying gas (zero-grade air) at setting 40; source temperature, 280 °C; orifice, −51 V; Q1 ion energy, 1 V; collision energy, 27 V; collision gas (nitrogen) at setting 3; Q1 and Q3 resolutions at unit mass (0.7 full width at half maximum). Product ion scans of VMA and d3-VMA calibrators were collected in continuous flow mode as described previously (4)(5). In the selected-reaction monitoring mode, the instrument was optimized to monitor the m/z 197 to m/z 137 and m/z 200 to m/z 140 transitions for VMA and d3-VMA, respectively, likely reflecting the loss of CHO3 from the carboxylate moiety and an additional loss of oxygen. Data were acquired and processed by MassChrom (Ver. 1.1.1) software, including Multiview (Ver. 1.4) for chromatographic and spectral interpretation and TurboQuan (Ver. 1.0) for quantification. Separation of VMA and d3-VMA from the specimen matrix was achieved by use of a Discovery® RP Amide C16 column [50 × 4.6 mm (i.d.); 5-μm bead size; Supelco] with a split of the column effluent flow set to deliver 200 μL/min to the electrospray source. VMA and d3-VMA coeluted with a retention time of 1.2 min; the run time was 3 min/sample.
We monitored intraassay (n = 4) and interassay (n = 5) variability of the calibration data obtained over concentrations from 1.04 to 20.0 mg/L. The average slope, intercept, and coefficient of linear regression (r2) were 1.109 (95% confidence interval, 1.000–1.217), −0.006 (−0.019 to 0.007) mg/L, and 0.999, respectively. Urine specimens with VMA concentrations up to 56 mg/L were analyzed accurately without dilution. At low concentrations, a urine specimen with a calculated VMA concentration of 0.30 mg/L exhibited a signal-to-noise ratio for the VMA extracted single-reaction monitoring signal of 25:1 (injected amount equal to 3 ng).
Recovery experiments were conducted using leftover human urine submitted for VMA determination. We added 1.25 mg/L VMA to 14 urine aliquots with VMA values of 1.2–5.7 mg/L and analyzed them in single determinations alongside paired aliquots without added VMA. VMA recovery was 80–144%. The mean recovery and CV were 102.5% and 16%, respectively. Intra- and interassay precision was assessed by replicate analysis of specimen aliquots on a single day or on successive days and is shown in Table 1⇓ . Good precision was obtained, demonstrating the accuracy of the method for the quantitative determination of VMA.
The stability of a prepared specimen was investigated by repeat injections of multiple aliquots (n = 5) on the same day (day 1) and subsequent days (days 2 and 5) with interim storage at 4 °C. All aliquots gave a value of 1.1 mg/L, which indicates the stability of VMA and d3-VMA on refrigeration for at least 5 days after sample preparation.
For the purpose of method comparison, we compared the MS/MS method against a commercial HPLC assay (Bio-Rad Laboratories) previously used in our laboratory (6)(7). The acquisition time for the HPLC method is relatively long (6 min/specimen) and is prone to several problems: shift of retention time, interference by coeluting compounds (an average of 5% of specimens were rejected after analysis), detector instability, and the necessity of frequent maintenance.
Unused portions of 324 specimens successfully analyzed by the HPLC method were retested using the new method. The correlation between the new (y) and old (x) methods was: y = 1.05x + 0.06 mg/L, with an average difference of 0.2 mg/L. Bland–Altman and Deming (inset; range, 0–20 mg/L) plots are shown in Fig. 1⇓ . More than 99% of compared values differed by less than 2 SD. Results above our adult reference interval (<8 mg/24 h) were assessed on the basis of clinical sensitivity; for example, for a specimen with an average VMA value of 57.5 mg/L, a difference of 7.4 mg/L corresponded to a markedly increased VMA value: 680 and 598 mg/24 h by the new and old methods, respectively. Analysis of proficiency testing specimens carried out during the method comparison gave VMA values in agreement at low concentrations [5.2 and 5.2 mg/L; mean (SD), 5.4 (0.4) mg/L] and increased concentrations [22.1 and 23.7 mg/L; mean (SD), 22.8 (0.7) mg/L].
In summary, we have developed and validated a MS/MS method for the determination of VMA in urine that uses a stable-isotope-labeled internal standard. Our method includes automated solid-phase extraction, an isocratic mobile phase, and quantification against a stable-isotope-labeled internal standard. Sample preparation is fully automated, and a rapid analytical time (3 min/sample) is achieved with little or no need to repeat analyses.
- © 2003 The American Association for Clinical Chemistry