Rapid Analysis of Metanephrine and Normetanephrine in Urine by Gas Chromatography Mass Spectrometry

Clinical Chemistry 48:2 332 337 (2002) Endocrinology and Metabolism Rapid Analysis of Metanephrine and Normetanephrine in Urine by Gas Chromatography Mass Spectrometry David K. Crockett, 1* Elizabeth L. Frank, 2 and William L. Roberts 2 Background: Widely used HPLC methods for quantification of metanephrine and normetanephrine in urine often have long analysis times and are frequently plagued by drug interferences. We describe a gas chromatography mass spectrometry method designed to overcome these limitations. Methods: Metanephrine and normetanephrine conjugates were converted to unconjugated metanephrine and normetanephrine by acid hydrolysis. To avoid the rapid decomposition of the deuterated internal standards (metanephrine-d 3 and normetanephrine-d 3 ) under hydrolysis conditions, the internal standards were added after hydrolysis. Solid-phase extraction was used to isolate the hydrolyzed metanephrines from urine. Samples were concentrated by evaporation, then derivatized simultaneously with N-methyl-N-(trimethylsilyl)trifluoroacetamide and N-methyl-bis-heptafluorobutryamide at room temperature. Results: The assay was linear from 25 to 7000 g/l. The intraassay CVs were <5% and the interassay CVs <12%. Comparison with a routine HPLC method (n 192) by Deming regression yielded a slope of 1.00 0.02 g/l, an intercept of 5.8 7.8 g/l, and S y x 50.6 g/l for metanephrine and a slope of 0.94 0.03, intercept of 19 11 g/l, and S y x 60 g/l for normetanephrine. The correlation coefficients (r) were calculated after log transformation of the data and gave r 0.97 for metanephrine and r 0.97 for normetanephrine. Interference from common medications or drug metabolites was seen in <1% of samples. The time between sequential injections was <7 min. Conclusions: This new gas chromatography mass spectrometry assay for total fractionated metanephrines is rapid, compares well with a standard HPLC assay, and avoids most drug interferences that commonly affect HPLC assays for urine metanephrines. 2002 American Association for Clinical Chemistry Quantification of urine metanephrine and normetanephrine is commonly requested in the diagnosis and monitoring of patients with pheochromocytoma and related neurogenic tumors. The determination of urinary metanephrines is considered a reliable screening test for pheochromocytoma (1, 2). More than 97% of urinary metanephrines are conjugated as either the glucuronide or sulfate forms (3). For measuring total fractionated metanephrines, acid hydrolysis or enzymatic treatment is necessary before the assay. Recent studies indicate that the measurement of plasma metanephrines, either total or free, is an alternative to urinary measurements (4 6). Plasma metanephrines may be a more sensitive indicator of the presence of a tumor, whereas urinary metanephrines may be more specific (4). A literature search revealed that most published laboratory methods for metanephrine and normetanephrine analysis in urine use HPLC. These HPLC methods can accommodate simplified sample preparation, but often have relatively long analysis times and are plagued by drug interferences (7 9). In addition, none of the many methods available are ideally suited for the high volume of samples encountered in a commercial reference laboratory (10 13). Our goal was to overcome these limitations of current HPLC methods. Consequently, we have developed a rapid and specific method for measurement of total urinary metanephrine and normetanephrine that uses gas chromatography mass spectrometry (GC-MS). 3 1 ARUP Laboratories, Inc., 500 Chipeta Way, Salt Lake City, UT 84108. 2 Department of Pathology, University of Utah School of Medicine, Salt Lake City, UT 84132. *Author for correspondence. E-mail crockedk@aruplab.com. Received August 1, 2001; accepted November 13, 2001. 3 Nonstandard abbreviations: GC-MS, gas chromatography mass spectrometry; MSTFA, N-methyl-N-(trimethylsilyl)trifluoroacetamide; and MB- HFBA, N-methyl-bis-heptafluorobutryamide. 332

Clinical Chemistry 48, No. 2, 2002 333 Materials and Methods chemicals and reagents Metanephrine and normetanephrine were obtained from Sigma. The deuterated internal standards metanephrine-d 3 and normetanephrine-d 3 were purchased from Medical Isotopes. All other solvents were reagent grade and obtained from Fisher Scientific. The Lyphocheck urine calibrator and controls were purchased from Bio- Rad Clinical Division. samples Urine samples (n 192), for which metanephrine testing had been performed by HPLC, were chosen at random. Concentrations of metanephrine and normetanephrine ranged from 25 g/l to 7000 g/l. All samples were stored at 2 8 C until analysis. The calibrator and controls were prepared as needed and stored at 2 8 C. All studies using samples from patients were approved by the Institutional Review Board of the University of Utah Health Sciences Center. apparatus Solid-phase extraction was performed with a Speedisk 48 Pressure Processor (SPEware, Inc.). A TurboVap TM LV Evaporator sample concentrator station from the Zymark Corporation was used to evaporate solvents. Clean Screen CSDAU 303 solid-phase extraction columns (C8 mixed-mode cation exchange) were purchased from United Chemical Technologies, Inc. The analysis of metanephrines was performed using a Finnigan Voyager GC-MS instrument equipped with an A220S autosampler and ToxLab TM software (Thermoquest). We performed data acquisition in the selected-ion monitoring mode using electron ionization with the detector at 350 mv. The GC was equipped with a DB-5ms capillary column (length, 15 m; diameter, 0.25 mm; film thickness, 0.25 m; J&W Scientific). The instrument was tuned with perflurotributylamine. assay procedures A 1.0-mL aliquot sample was used for hydrolysis and extraction. HCl (50 L of 6.0 mol/l) was added to each urine sample. The tubes were incubated at 90 C inadry heat block for 25 min and allowed to cool before proceeding. Internal standard (100 L containing 5 ng/ L each of metanephrine-d 3 and normetanephrine-d 3 ) was added. Solid-phase extraction was performed in the mixed-mode Clean Screen column. The extraction column was preconditioned with 3 ml of methanol followed by 3 ml of water. The samples were loaded and the columns washed with 3 ml of water followed by 3 ml of 0.1 mol/l HCl and then 3 ml of methanol. The columns were dried for 5 min before the metanephrines were eluted. The elution solvent for each column was 3 ml of methanol containing 20 ml/l ammonium hydroxide. The samples were evaporated to dryness and derivatized with 50 L of acetonitrile, 25 L of N-methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA), and 25 L of N-methyl-bisheptafluorobutryamide (MBHFBA). The dual derivatization was performed simultaneously at room temperature. A sample volume of 1 L was injected with a split ratio of 40:1. The helium carrier gas was set at a constant flow of 2.0 ml/min. The injection port temperature was 200 C. The initial oven temperature was 120 C, held for 0.2 min, then ramped at 30 C/min to 200 C and held for 0.2 min, then ramped at 70 C/min to 310 C and held for 2.0 min. The detector interface temperature was 300 C. The source temperature was 240 C. The ions monitored were m/z 508, 509, 511, 512, 522, 523, 525, and 526. The urine calibrator was prepared according to the package insert and used for single-point calibration. The calibrator target concentrations were 328 and 718 g/l for metanephrine and normetanephrine, respectively. The negative control was prepared by the addition of 100 L of working internal standard to blank urine. The blank urine was prepared by diluting a nonpathologic urine sample with water at a ratio of 1:5. The Lyphocheck normal control sample was prepared per package insert directions and used as the low control. The target ranges were 54 90 g/l for metanephrine and 204 292 g/l for normetanephrine. The Lyphocheck normal control sample was used as the high control, but prepared at twice the concentration prescribed by the package insert. The target ranges were 400 900 g/l for metanephrine and 1400 2000 g/l for normetanephrine. This allowed the controls to bracket the concentration of the single-point calibrator. The HPLC method uses a 3.0-mL aliquot sample. HCl (2.0 ml of 0.6 mol/l solution) was added to each sample, and the tubes were incubated in a hot oil bath at 107 C for 30 min and allowed to cool before proceeding. Water (5 ml) was added to each tube, and the samples were filtered and centrifuged for 10 min at 2000g. An aliquot of 1 ml was transferred to each autosampler vial, and 40 L was injected for analysis. The same calibrator and Lyphocheck controls described above were used, but no internal standard was used. The HPLC method uses the Coulochem Electronic Array System (ESA Inc.). Mobile phase (no. 70-3067) and HPLC column (no. 70-0340) from ESA Inc. were also used. The limit of detection for this assay was 20 g/l. The overall imprecision of the assay was 12% for the Lyphocheck normal control sample and 8.3% for the Lyphocheck abnormal control sample. The assay was derived from the method described by Gamache et al. (14). Nine concentrations (25, 50, 100, 250, 500, 1000, 1500, 3000, and 7000 g/l) of calibrators were analyzed in triplicate with one run per day for 3 days to assess linearity and imprecision. The long-term stability of the derivatized sample was monitored for 14 days. Sample vials were stored at room temperature. Hydrolysis of the conjugated metanephrine and normetanephrine in urine was optimized by evaluating differing conditions of time,

334 Crockett et al.: Urine Metanephrines by GC-MS temperature, and acid concentration. EP Evaluator-CLIA software (David G. Rhoads Associates) was used for all regression analysis and calculation of S y x and correlation coefficients (r). All drugs known to interfere with the current HPLC method, including several biogenic amines and cardiac medications, were derivatized and injected to determine retention times and prominent mass fragments. For each compound, 10 mg was derivatized at room temperature with 50 L of acetonitrile plus 25 L of MSTFA plus 25 L of MBHFBA. Normetanephrine-d 3 was used to calculate the relative retention times of all other compounds. Results The concentration of HCl is key to successful hydrolysis of the metanephrine and normetanephrine conjugates in urine (Fig. 1). Increasing concentrations of HCl decomposed the deuterated internal standard. Excessive acid will destroy metanephrine and normetanephrine as well. HCl was tested from 0.1 to 6 mol/l. Time and temperature play a secondary, but minor role in the hydrolysis process. Temperatures between 25 and 120 C were evaluated. Hydrolysis times from 10 to 120 min were evaluated as well (data not shown). Hydrolysis in 0.3 mol/l HCl at 90 C for 25 min was optimal. Because of the rapid decomposition of the deuterated internal standards, these compounds were added after the hydrolysis step and before solid-phase extraction. Linearity was evaluated with nine concentrations analyzed in triplicate. One run a day for 3 days was performed. The linearity of metanephrine and normetanephrine after hydrolysis and solid-phase extraction is shown in Fig. 2. The limit of detection of the assay, defined as a signal-to-noise ratio 5:1, was 10 g/l. The limit of quantification, with requirements of accuracy within 85 115% of the target concentration and ion mass ratios within 20% of the target ratio as set by the calibrator, was 25 g/l. The linear range of the assay was 25 7000 g/l. This range is consistent with the more widely used HPLC methods. The total-ion chromatogram and full-scan mass spectra of derivatized metanephrine and normetanephrine are shown (Fig. 3). All CVs for within-run imprecision were 4.2%. All CVs for between-run imprecision were 12%. Total imprecision was 12% (Table 1). By the addition of deuterated analogs, the GC-MS assay imprecision was greatly improved in contrast to the HPLC method without internal standard. A correlation study was performed using 192 patient samples that were assayed previously for metanephrine and normetanephrine by HPLC. Sample concentrations Fig. 1. Evaluation of acid hydrolysis with increasing concentrations of HCl for 25 min at 90 C. (A), recovery of metanephrine ( ) and metanephrine-d 3 ( ) after acid hydrolysis. (B), recovery of normetanephrine ( ) and normetanephrine-d 3 ( ) under the same hydrolysis conditions. Fig. 2. Summary of nine different concentrations (25, 50, 100, 250, 500, 1000, 1500, 3000, and 7000 g/l) analyzed in one run per day for 3 days. Linearity after acid hydrolysis and solid-phase extraction for metanephrine (A; y 0.947x 7.21; r 0.999) and normetanephrine (B; y 0.932x 35.1; r 0.999).

Clinical Chemistry 48, No. 2, 2002 335 Fig. 3. GC-MS analysis of urine metanephrines showing the total ion chromatogram from a representative injection in selected-ion monitoring mode (A) and the full-scan mass spectra of metanephrine (B) and normetanephrine (C) derivatized with MSTFA and MBHFBA. ranged from 25 to 7000 g/l. Results from the new GC-MS method were compared with the routine HPLC method (Fig. 4). Deming regression was performed before log transformation. Because the data were not gaussian, the Pearson correlation coefficients (r) were calculated after log transformation. Because of interference in the original HPLC method, the results for three metanephrine Table 1. Precision a of Bio-Rad urine calibrator and controls. Mean value, g/l Within-run CV, % Between-run CV, % Total CV, % Metanephrine Sample Low control 56 4.1 12 12 High control 648 1.2 1.9 2.3 Calibrator 322 1.2 0.5 1.3 Calibrator, 0.5 187 2.5 7.7 8.1 Normetanephrine Sample Low control 233 1.8 1.4 2.3 High control 1937 2.4 5.2 5.7 Calibrator 673 1.7 0.8 1.9 Calibrator, 0.5 355 3.2 6.9 7.6 a Five runs with each calibrator in triplicate. Fig. 4. Comparison of HPLC and GC-MS methods in 186 samples. Deming regression analysis before log transformation gives a slope of 1.00 0.02, an intercept of 6 8 g/l, S y x 51 g/l, and r 0.97 for metanephrine (A) and a slope of 0.94 0.03, an intercept of 19 11 g/l, S y x 60 g/l, and r 0.97 for normetanephrine (B). Pearson correlation coefficients (r) were calculated after log transformation. samples and five normetanephrine samples were not usable, and the data were excluded from the graphs in Fig. 4. The dual derivative showed excellent stability, with no decrease in the abundance of metanephrine or normetanephrine after 14 days at room temperature. An extensive interference study was performed to evaluate the specificity of the assay. The results are summarized in Table 2. In addition, of the 192 samples analyzed by GC-MS, only 1 showed interference. The sample contained a coeluting peak next to metanephrine with 40% resolution and ion mass ratio failure. Quantification for this sample was likely affected. Overall, the sample failure rate attributable to poor chromatography or drug interference with either metanephrine or normetanephrine was reduced from 11% (21 of 192 by HPLC) to 1% (1 of 192 by GC-MS). An evaluation of potential column carryover found 1.3 g/l metanephrine and 7.7 g/l normetanephrine after a supplemented sample (10 000 g/l) was injected. As an additional safeguard to prevent carryover, the software is programmed to automatically inject a solvent blank when a concentration threshold of 5000 g/l is exceeded.

336 Crockett et al.: Urine Metanephrines by GC-MS Table 2. Summary of drug interferences, relative retention times, and mass fragments. Compound Retention time, min Relative retention time, min a m/z b Ephedrine 1.576 0.629 179, 180, 209, 283 Pseudoephedrine 1.640 0.655 179, 180, 209, 283 Norepinephrine 1.792 0.715 292, 219, 149, 423 3-Methoxytyramine 2.273 0.907 209, 222, 191, 435 Dopamine 2.377 0.949 267, 280, 193, 493 Normetanephrine-d 3 2.505 1.000 298, 511, 512 Normetanephrine 2.513 1.003 297, 508, 509 Norepinephrine 2.633 1.051 355, 356, 265, 476 Metanephrine-d 3 2.649 1.057 298, 525, 526 Metanephrine 2.657 1.061 297, 522, 523 Epinephrine 2.753 1.099 355, 356, 239, 580 Salsolinol 2.849 1.137 504, 505, 519, 415 Serotonin 3.417 1.364 444, 445, 218, 231 Fenfluramine ND c Methamphetamine ND Nicotine ND Phentermine ND Phenylpropanolamine ND a Relative retention time based on normetanephrine-d 3. b Major mass fragments found with MSTFA and MBHFBA double derivatization. c ND, not detected. Discussion Acid hydrolysis is required before solid-phase extraction to determine the total concentration of metanephrine and normetanephrine. The ideal internal standards would be the deuterated sulfate conjugates of metanephrine and normetanephrine. However, these compounds are not commercially available and are difficult to synthesize in sufficient quantities for routine use. Therefore, they were not used in our study. When the optimal conditions for the hydrolysis of the metanephrine and normetanephrine conjugates are used, the unprotected or nonconjugated internal standard is quickly degraded. Therefore, to avoid rapid decomposition of the deuterated analogs under hydrolysis conditions, we added the internal standards after acid hydrolysis. This improved the agreement between GC-MS results and the current well-established reference intervals (1, 9). Some of the advantages of solid-phase extraction technology are well known (15). However, a cleaner final extract is often exchanged for increased sample preparation time and the additional cost of the solid-phase columns and an extraction manifold. After evaluation of several solid-phase methods and extraction columns, the UCT Clean Screen solid-phase extraction column was selected for its cleanliness and relative low cost. In addition, the extraction was moved to a 48-place positive pressure manifold to increase the efficiency of the extraction. Routine HPLC methods may take 20 min or longer per injection (5, 7 9). Our initial experiments showed promise with a much faster GC oven method and a total analysis time of 4 min. However, the ion mass ratios for an occasional specimen would fall outside of the allowed range or an interfering peak would coelute with metanephrine or normetanephrine. Therefore, the oven program was optimized to allow sufficient GC separation of all interfering compounds. This GC-MS method allows a throughput of approximately eight samples per hour in comparison with three per hour by HPLC with a greatly decreased occurrence of interferences. In addition, the cost of a GC-MS system may be as little as 25% more than the HPLC equipped with electrochemical detection. Furthermore, although a higher degree of operator training may be necessary, one GC-MS instrument can replace two HPLC instruments. The limit of quantification of this assay compares favorably with HPLC methods for urinary metanephrine and normetanephrine quantification. The initial sample aliquot volume, concentration of internal standard, final solvent volume in the autosampler vial, GC injection volume, and injection port split ratio settings can all impact the overall sensitivity of the assay. By adjusting these factors in the assay, we found that the sensitivity could be enhanced. It is important to note, however, that clinically significant concentrations of metanephrines are usually 5- to 10-fold higher concentrations than the upper limit of the reference interval (1). The extended linear range of the assay is a tradeoff for slightly reduced low-end sensitivity. The assay and instrumentation, as described, are not sensitive enough to measure free urinary metanephrines, which are present at only 3% of the total concentration (3). The double derivatization has several advantages compared with MSTFA alone. The higher molecular weight of the parent fragments (m/z 522 for metanephrine and m/z 508 for normetanephrine) improved sensitivity and decreased interferences. Single peaks were seen for metanephrine and normetanephrine, in contrast to multiple peaks when biogenic amines are derivatized with MSTFA only. This enhanced sensitivity because only one peak was present for each compound. In addition, when drugs that could interfere with HPLC methods were tested, no interference was found. No acid anhydride byproducts are produced when MBHFBA is used, as are common in derivatization with pentafluoropropionic acid anhydride or heptafluorobutyric acid anhydride. This eliminated a second evaporation and reconstitution step and extended the useful lifetime of the GC column because acid derivatives were not injected. The two derivatization reactions can be performed simultaneously, without heated incubation. This dual derivatization technique not only improves chromatography and decreases interference, but also enhances the sensitivity. In moving the assay from HPLC to GC-MS, several improvements were made. The sample aliquot was decreased from 3 ml to 1 ml, allowing the laboratory sufficient volume to retest the specimen if necessary.

Clinical Chemistry 48, No. 2, 2002 337 Deuterated internal standards were incorporated, leading to improved accuracy and precision. Interferences from drugs and drug metabolites decreased from 11% to 1%. Carryover from samples with increased metanephrine or normetanephrine concentrations was minimal. The assay linearity was extended, thereby decreasing the number of repeat analyses for concentrations above the linear range. Analysis time was decreased from 22 min per sample to 7 min per sample. This assay is fast and simple, is reliable in a high-volume clinical laboratory setting, and to date, analysis of more than 8000 samples has been performed. This work was supported by the ARUP Institute for Clinical and Experimental Pathology. References 1. Rosano TG, Whitley RJ. Catecholamines and serotonin. In: Burtis CA, Ashwood ER, eds. Tietz textbook of clinical chemistry, 3rd ed. Philadelphia: WB Saunders, 1999:1570 600. 2. Graham PE, Smythe GA, Edwards GA, Lazarus L. Laboratory diagnosis of phaeochromocytoma: which analytes should we measure? Ann Clin Biochem 1993;30:129 34. 3. Eisenhofer G. Free or total metanephrines for the diagnosis of pheochromocytoma: what is the difference [Editorial]? Clin Chem 2001;47:988 9. 4. Lenders JW, Keiser HR, Goldstein DS, Willemsen JJ, Friberg P, et al. Plasma metanephrines in the diagnosis of pheochromocytoma. Ann Intern Med 1995;123:101 9. 5. Roden M, Raffesberg W, Raber W, Bernroider E, Niederle B, et al. Quantification of unconjugated metanephrines in human plasma without interference by acetaminophen. Clin Chem 2001;47: 1061 7. 6. Eisenhofer G, Lenders JW, Linehan WM, Walther MM, Goldstein DS, Keiser HR. Plasma normetanephrine and metanephrine for detecting pheochromocytoma in von Hippel-Lindau disease and multiple endocrine neoplasia type 2. N Engl J Med 1999;340: 1872 9. 7. Gupta RN. Improved sample preparation in determination of urinary metanephrines in liquid chromatography with electrochemical detection. Clin Chem 1990;36:538 40. 8. Sample preparation for the analysis of catecholamines and their metabolites in human urine. Liq Chromatogr 1991;14:2779 80. 9. Jouve J, Mariotte N, Sureau C, Muh JP. High-performance liquid chromatography with electrochemical detection for the simultaneous determination of methoxylated amines, normetanephrine, metanephrine and 3-methoxytyramine in urine. J Chromatogr 1985;224:53 62. 10. Canfell C, Binder SR, Khayam-Bashi H. Quantitation of urinary normetanephrine and metanephrine by reversed-phase extraction and mass-fragmentographic analysis. Clin Chem 1982;28:25 8. 11. Gerlo EA, Sevens C. Urinary and plasma catecholamines and urinary catecholamine metabolites in pheochromocytoma: diagnostic value in 19 cases. Clin Chem 1994;40:250 6. 12. Anggard E. Gas chromatography of catecholamine metabolites using electron capture detection and mass spectrometry. Anal Chem 1969;41:1250 4. 13. Wolthers BG, Kema IP, Volmer M, Wesemann R, Westermann J, Manz B. Evaluation of urinary metanephrine and normetanephrine enzyme immunoassay (ELISA) kits by comparison with isotope dilution mass spectrometry. Clin Chem 1997;43:114 20. 14. Gamache PH, Kingery ML, Acworth IN. Urinary metanephrine and normetanephrine determined without extraction by using liquid chromatography and coulometric array detection. Clin Chem 1993;39:1825 30. 15. Jennison T, Jones C, Wozniak E, Urry FM. The reliability of a solid-phase extraction system for the analysis of benzoylecgonine in urine. J Chromatogr Sci 1994;32:126 31.