Ion Spray Tandem Mass

EDTA in Human Plasma and Urine. Robin L. Sheppard† and Jack Henion*. Analytical Toxicology, Cornell University, 927 Warren Drive, Ithaca, New York 1...
0 downloads 0 Views 265KB Size
Download PDF
Anal. Chem. 1997, 69, 2901-2907

Quantitative Capillary Electrophoresis/Ion Spray Tandem Mass Spectrometry Determination of EDTA in Human Plasma and Urine Robin L. Sheppard† and Jack Henion*

Analytical Toxicology, Cornell University, 927 Warren Drive, Ithaca, New York 14850

A quantitative method has been developed for the determination of EDTA in human plasma and urine. The samples are prepared with automated anion-exchange solid-phase extraction using 100 µL of human plasma. The extracts are analyzed by capillary electrophoresis/ mass spectrometry using selected reaction monitoring in the negative ion mode. Large-volume injections (10% of the CE capillary volume) are used to improve the concentration level of detection via field-amplified sample injection. The first reported validation of a CE/MS/MS technique was carried out for this method. Using a 13C stable-label isotope for the internal standard, the lower level of detection and lower level of quantitation were determined to be 7.3 and 14.6 ng/mL EDTA in human plasma, respectively. The injection precision had a relative standard deviation (RSD) of 6.1%. The intra-assay precision was less than 15% RSD. The intra-assay accuracy was less than (12% bias from the nominal concentration. The interassay precision was less than 18% RSD and the interassay accuracy was less than (9% bias from the nominal concentration. Ethylenediaminetetraacetic acid (EDTA) is a metal complexing agent that has been popular for many purposes since its commercialization in the early 1950s.1 The free acid structure with molecular weight 292.1 is shown in Figure 1A. EDTA has four acidic protons with pK values of 2.0, 2.67, 6.16, and 10.26, respectively.2 The free acid is only slightly soluble in water. The disodium salt, which is moderately soluble in water, is more commonly used. The real power of EDTA is its ability to form 1:1 complexes with 62 different metal cations.3 The complexforming species is the 2- charge state corresponding to the ionization of the first two protons.2 In solution, EDTA is thought to be a hexadentate ligand structure via two nitrogen atoms and four oxygen atoms. A traditional three-dimensional representation using nickel(II) EDTA as an example is shown in Figure 1B.2 Nevertheless, crystal structure studies of many EDTA-metal complexes show that the metal is actually pentadentate with the sixth coordination site being occupied by a water molecule.4-7 In †

Current address: Xerox Corp., 800 Phillips Rd., 139-64A, Webster, NY 14580. (1) Bersworth Chemical Co. The Versenes. Powerful Organic Chelating Agents for the Exacting Chemical Control of Cations; Framingham, MA, 1954. (2) Welcher, F. J. The Analytical Uses of Ethylenediaminetetraacetic Acid; Van Nostrand: Princeton, NJ, 1958; Chapter 1. (3) Dean, J. A. Langes’s Handbook of Chemistry, 14th ed.; McGrw-Hill: New York, 1992; Table 8.13, pp 8.93-8.94. (4) Mizuta, T.; Yamamoto, T.; Miyoshi, K.; Kushi, Y. Inorg. Chim. Acta 1990, 175, 121-126. S0003-2700(97)00068-1 CCC: $14.00

© 1997 American Chemical Society

Figure 1. (A) Structure of the free acid of EDTA, (B) threedimensional structure of doubly charged Ni-EDTA, and (C) structure of singly charged Ni-EDTA.

addition, ion spray mass spectrometry of Ni-EDTA indicates that this complex and other EDTA complexes with 2+ oxidation state metals retain one carboxylate proton giving a net charge of 1-. (5) Douglas, B. E.; Radavonic, D. J. Coord. Chem. Rev. 1993, 128, 139-165. (6) Solans, X.; Font-Altaba, M.; Oliva, J.; Herrera, J. J. Acta Crystallogr. 1983, C39, 435-438. (7) Solans, X.; Galı´, S.; Font-Altaba, M.; Oliva, J.; Herrera, J. J. Acta Crystallogr. 1983, C39, 438-440.

Analytical Chemistry, Vol. 69, No. 15, August 1, 1997 2901

A representation of this structure for Ni-EDTA is shown in Figure 1C. The free acid of EDTA and its various salt forms are used extensively in our food and in our environment.8-10 Its uses are extremely diverse. In medicine it is used to treat metal poisoning11 and to act as a blood anticoagulant.12 In foods, it is added to preserve color and flavor in many items like canned or pickled vegetables, mayonnaise, and salad dressings.8,10 It is a common additive to detergents as a water-softening agent. It is used to deliver trace minerals in animal feeds8 and some human foods.9 EDTA is also used widely in decontamination operations for heavy metals and radioactive wastes.13 In these various ways, EDTA finds its way into the food we eat, the clothes we wear, rivers, groundwater, and even our drinking water.14,15 In 1954, a metabolism study in humans was carried out using 14C -labeled calcium EDTA in combination with calcium disodium 4 EDTA.16 The study included intravenous, intramuscular, oral, and skin absorption routes of administration. It was determined that, in blood, the majority of the activity was present in the plasma; therefore, it was not necessary to use whole blood in the activity assay. Radioactivity (5% total) was detectable in the urine for ∼18 h after oral administration. On average, 95% of the oral dose was recovered in the urine or feces within three days of administration, leaving very little possibility for the presence of EDTA in the plasma. More recent metabolism studies studying the NaFeIIIEDTA complex report that the complex dissociates during digestion and confirm that ∼5% of the EDTA is absorbed and excreted in urine.17 Since the EDTA-metal complexes have a permanent negative charge in solution, we selected capillary electrophoresis as the separation technique for this method. The determination of EDTA in human plasma was used as an example to explore the analytical ruggedness of CE/MS/MS as a routine bioanalytical technique. The CE/MS/MS method was rigorously validated and, to our knowledge, represents the first complete report of method validation by a CE/MS/MS technique. Mass spectrometry is a desirable mode of detection for capillary electrophoresis since it is more universal than any other available CE detector and it has the advantage of positive analyte identification. However, it must be interfaced to the outlet end of the capillary. It is this interfacing that has caused CE/MS to be considered nonroutine by many. Nonetheless, there are a significant number of qualitative CE/MS applications that have been published. In contrast, very few quantitative methods using CE/MS have been described in the literature.18,19 In practice, the more rugged CE/MS interfaces utilize makeup liquids that mix with the CE effluent and bring the total flow (8) Hunter, B. T. Consumers’ Res. 1988, 71, 8-9. (9) Whittaker, O.; Vanderveen, J. E.; Dinovi, M. J.; Kusnesof, P. M.; Dunkel, V. C. Reg. Toxicol. Pharm. 1993, 18, 419-427. (10) Fishbein, L.; Flamm, W. G.; Falk, H. L. Chemical Mutagens, Environmental Effects on Biological Systems; Academic Press: New York, 1970; pp 249251. (11) Markowitz, M. E.; Bifur, P. E.; Ruff, H.; Rosen, J. F. Pediatrics 1993, 92, 265-271. (12) Seegers, W. H. Blood Clotting Enzymology; Academic Press: New York, 1967; pp 364-366. (13) Means, J. L.; Crerar, D. A.; Duguid, J. O. Science 1978, 200, 1383-1385. (14) Xue, H.; Sigg, L.; Kari, F. G. Environ. Sci. Technol. 1995, 29, 59-68. (15) Kari, F. G.; Giger, W. Environ. Sci. Technol. 1995, 29, 2814-2827. (16) Foreman, H.; Trujillo, T. T. J. Lab., Clin. Med. 1954, 43, 566-571. (17) Hurrell, R. F.; Ribas, S.; Davidsson, L. Br. J. Nutr. 1994, 71, 85-93. (18) Henion, J. D.; Mordehai, A. V.; Cai, J. Anal. Chem. 1994, 66, 2103-2109. (19) Hsieh, F. Y. L.; Cai, J.; Henion, J. J. Chromatogr., A 1994, 679, 206-211.

2902 Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

Figure 2. Detailed schematic drawing of the self-aligning liquid junction (SALJ) ion spray CE/MS interface.

up to 0.5-10 µL/min.20 More recently, CE/MS interfaces that do not require make-up liquids have been investigated.20-23 Alternatively, we recently modified the liquid junction CE/MS interface24 with the intention of making it easier to use.25,26 A detailed schematic drawing of an ion spray version of the resulting “self-aligning liquid junction” (SALJ) CE/MS interface is shown in Figure 2. CE/MS experiments demonstrated a three-fold increase in signal-to-noise ratio in head-to-head comparisons with the popular sheath-flow interface.26 The SALJ CE/MS interface which incorporates ion spray conditions was used in all of the experiments in this paper.

EXPERIMENTAL SECTION Chemicals. The (13C4)EDTA internal standard was generously provided by Dr. K. Ballard of the Baylor School of Medicine (Houston, TX). Control and other samples of human heparinpreserved plasma and Na2EDTA-preserved plasma were a gift from Advanced Bioanalytical Services (Ithaca, NY). Disodium EDTA dihydrate, copper(II) chloride dihydrate, 88% formic acid, glacial acetic acid, sodium hydroxide, and HPLC-grade water were obtained from Fisher Chemical (Pittsburgh, PA). Iron(II) chloride hexahydrate, calcium chloride, and magnesium chloride hexahydrate were obtained from Sigma Chemical (St. Louis, MO). Nickel(II) nitrate hexahydrate, ammonium hydroxide, trifluoroacetic acid, and bromothymol blue, sodium salt, were obtained from Aldrich Chemical (Milwaukee, WI). Ammonium formate was obtained from Fluka Chemical Corp. (Ronkonkoma, NY). HPLC-grade methanol was obtained from J. T. Baker (Phillipsburg, NJ). Instrumentation. The capillary electrophoresis separations were carried out on a Hewlett-Packard 3DCE instrument (Palo Alto, CA) with a diode array UV detector that was used at 200 nm in some of the preliminary experiments. CElect-Amine bonded capillaries were from Supelco, Inc. (Bellefonte, PA). The CE (20) Cai, J.; Henion, J. J. Chromatogr., A 1995, 703, 667-692. (21) Wahl, J. H.; Gale, D. C.; Smith, R. D. J. Chromatogr., A 1994, 659, 217222. (22) Wahl, J. H.; Smith, R. D. J. Capillary Electrophor. 1994, 1, 62-71. (23) Kriger, M. S.; Ramsey, R. S.; Cook, K. D. Anal. Chem. 1995, 67, 385-389. (24) Lee, E. D.; Mu ¨ ck, W.; Henion, J. D.; Covey, T. R. Biomed. Environ. Mass Spectrom. 1989, 18, 844-850. (25) Wachs, T.; Sheppard, R. L.; Henion, J. U.S. Patent 5,587,582, December 1996. (26) Wachs, T.; Sheppard, R. L.; Henion, J. J. Chromatogr., B 1996, 685, 335342.

running buffer used was 30 mM ammonium formate/0.15% formic acid at pH 3.5. The separation was performed at -30 kV with 50 mbar of inlet pressure applied throughout the run. Since the SALJ interface essentially consists of a splitting tee, this positive inlet pressure from the CE instrument ensures that the make-up liquid does not enter the CE separation capillary. The amine-bonded separation capillary has a positively charged surface which causes anions to migrate faster than cations under these negative voltage conditions. All mass spectrometry was performed on a PE-Sciex API 300 triple quadrupole instrument (Concord, ON, Canada). The CE/ MS interface was a homemade self-aligning liquid junction interface25,26 with ion spray ionization and a sprayer needle made of 100 µm i.d. × 400 µm o.d. platinum tubing (Goodfellow Cambridge Limited, Berwyn, PA). Stainless steel tubing was obtained from Small Parts, Inc. (Miami Lakes, FL), and PEEK tubing and fittings were obtained from Upchurch Scientific (Oak Harbor, WA). A Harvard Apparatus (Natick, MA) syringe pump was used to deliver the make-up liquid (5 mM ammonium formate in 95% methanol) to the interface at 10 µL/min. Strong anion-exchange (SAX) solid-phase extraction disks were obtained from SPEC (Division of Ansys, Irvine, CA). The solidphase extractions were performed on a Gilson ASPEC sample preparation instrument (Middleton, WI). The Gilson ASPEC instrument uses positive pressure to push solutions through the extraction disk. All extraction steps were performed at a flow rate of ∼1 mL/min. Routine tuning and calibration of the mass spectrometer were performed using solutions of poly(propylene glycol)s, 2.5 µM PPG 425 and PPG 1000, in 33:66 2 mM ammonium acetate/acetonitrile, which produce numerous mass peaks in the mass range of 501000 Da. These solutions were used to verify that the mass axis calibration of the instrument was correct. In addition, the resolution of the instrument was tuned so that each singly charged mass peak had a peak width at half-height between 0.5 and 0.7 Da in both Q1 and Q3. This requirement assured that the instrument was operating with at least unit mass resolution, which means that a mass peak would be resolved from another mass peak that is separated by 1 Da. Generally, the mass calibration and resolution the API 300 were very stable and did not need to be changed even over several months of use, although these parameters were routinely verified. The mass spectrometer was used in the selected reaction monitoring (SRM) by monitoring one precursor-product ion transition for Ni-EDTA and the corresponding transition for the internal standard. These transitions are as follows: m/z 347 fragmenting to m/z 329 for Ni-EDTA and m/z 351 fragmenting to m/z 333 for Ni-(13C4)EDTA. The mass spectrometer was operated with a dwell time of 300 msec for each transition, resulting in a scan rate of more than one data point/s. Standard Preparation. Serial dilution standards were prepared from two different stock solutions made from separate weighings of Na2EDTA. The calibration standards were serially diluted from a 5 mM stock solution of Na2EDTA, prepared by dissolving 0.0188 g of the dihydrate in water with dilution to the mark in a 10.0 mL volumetric flask. The quality control (QC) samples were serially diluted from a 3.93 mM Na2EDTA, prepared by dissolving 0.0147 g of the dihydrate in water with dilution to the mark in a 10.0 mL volumetric flask. Human heparin-preserved plasma samples were fortified at 7.3, 14.6, 29.2, 58.4, 117, 234,

467, and 1460 ng/mL EDTA for the plasma calibration standards and 22.2, 88.2, and 314 ng/mL for the plasma QC samples. A stock solution of the (13C4)EDTA internal standard was prepared by dissolving 1.0 mg of the free acid in 1.0 mL of 3.5 mM sodium hydroxide to form the soluble sodium salts. A 1 ng/ µL solution was made by serial dilution from the stock solution. Using this 1 ng/µL solution, 50 ng of internal standard was added to each of the fortified plasma calibration standards and QC samples. Choice of Metal Chelate. EDTA that might be present in a blood sample would be primarily chelated to calcium, magnesium, and iron(III). To obtain the highest possible electrospray CE/ MS/MS signal intensity, it is advantageous to convert all available EDTA to a single metal complex that is highly stable. There are many competing phenomena that determine which EDTA chelate is most readily formed.2 While copper and iron have both been used for the analytical determination of EDTA in water,27-34 nickel has a stability constant similar to copper.2 Preliminary experiments were carried out to determine the most suitable metal for this application. Although iron(III) has the highest formation constant of these cations,3 it was not considered a strong candidate for this method for three reasons: (1) Iron(III) hydroxide forms readily at pH >5, thereby reducing the iron available for EDTA complexation,2 (2) Fe(III)-EDTA is slow to form at room temperature,2 and (3) Fe(III)-EDTA readily photodegrades, which means that sample solutions might not be stable over time.15 To study the effectiveness of Cu-EDTA and Ni-EDTA complex formation, synthetic solutions containing varying levels of calcium, magnesium, copper, nickel, and iron were prepared in a limiting concentration of Na2EDTA at pH 8.5. The results indicated that when used in excess at pH 8.5, both nickel and copper effectively chelate all available EDTA (data not shown). Since it appeared that either nickel or copper could be used, the MS and MS/MS sensitivity of Ni-EDTA and Cu-EDTA were investigated. Infusion ion spray mass spectrometry analysis of the Ni-EDTA and Cu-EDTA solution complexes showed that Ni-EDTA gave a more abundant ion current signal at equal concentrations. Therefore, nickel was used in this procedure to convert all available EDTA to the Ni-EDTA complex. Extraction Procedure. Anion-exchange chromatography has been used successfully as a sample preparation procedure for the free acid of EDTA.35-37 All of these references used hand-packed, large-volume columns in which large volumes of sample were processed. Since the free acid of EDTA exists only under extremely acidic conditions, which are not ideal for either CE or MS, it was desirable to use anion-exchange solid-phase extraction (SPE) to isolate the Ni-EDTA. There are no known reports that (27) Unger, M.; Mainka, E.; Konig, W. Fresenius J. Anal. Chem. 1987, 329, 5054. (28) Hall, L.; Takahashi, L. J. Pharm. Sci. 1988, 77, 247-250. (29) Parkes, D. G.; Caruso, M. G.; Spradling, J. E., III Anal. Chem. 1981, 53, 2154-2156. (30) Bauer, J.; Heathcote, D.; Krogh, S. J. Chromatogr. 1986, 369, 422-425. (31) Silanpa¨a, M.; Kokkonen, R.; Sihvonen, M.-L. Anal. Chim. Acta 1995, 303, 187-192. (32) Tran, G.; Chen, C.; Miller, R. B. J. Liq. Chromatogr. Relat. Technol. 1996, 19, 1499-1508. (33) Venesky, D. L.; Rudzinski, W. E. Anal. Chem. 1984, 56, 315-317. (34) Harmsen, J.; van der Toorn, A. J. Chromatogr. 1982, 249, 379-384. (35) Schu ¨ rch, S.; Du ¨ bendorfer, G. Mitt. Gebiaete Lebensm. Hyg. 1989, 80, 324334. (36) Ito, Y.; Toyoda, M.; Suzuki, H.; Iwaida, M. J. Assoc. Off. Anal. Chem. 1980, 63, 1219-1223. (37) Schaffner, C.; Giger, W. J. Chromatogr. 1984, 312, 413-421.

Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

2903

Figure 3. Summary of plasma extraction procedure.

use anion-exchange SPE to isolate EDTA metal chelates. Use of ion-exchange SPE for the extraction of EDTA from plasma had several benefits. If the optimum conditions were found, the isolated extract could be concentrated to some extent. In addition, the extract would be free of interfering anions so that CE analyte focusing techniques could be more easily optimized. The strong anion-exchange SPE method that was developed to isolate the Ni-EDTA complex from plasma is outlined in Figure 3. The Ni-EDTA complex is rather difficult to elute from the strong anion-exchange media. Therefore, the elution solvent uses a displacing anion, bromothymol blue, in addition to trifluoroacetic acid at pH 1, which dissociates the complex. Bromothymol blue was chosen for two reasons. First, it contains sulfonate groups, which are very effective at displacement of Ni-EDTA in anion exchange. Second, it has a large charge-to-mass ratio, which causes it to migrate more slowly than Ni-EDTA in the CE separation. Therefore, it stacks behind the Ni-EDTA during fieldamplified injection. Sample Injection. Using the principles of field amplification,38 the samples were dissolved in water and large sample volumes were introduced onto the capillary for focusing and separation in one step. Pressure injections of 0.1 min at 950 mbar inlet pressure were successfully used. This corresponds to ∼13% (∼150 nL) of the capillary length being filled with sample solution. However, these large sample injections occasionally led to capillary restriction from sample particulates and/or precipitation. In this case, the capillary was rinsed under high pressure by manually pushing liquid through the capillary with a syringe. Injection Sequence. The precision of the CE injection process requires careful control of experimental conditions. The capillary inlet and outlet must be level to ensure that siphoning (38) Li, S. F. Y. Capillary electrophoresis principles, practice and applications; Elsevier: Amsterdam, 1992.

2904

Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

in either direction is negligible. The nebulizing gas is turned off at the end of the CE run and is not turned back on until after the next sample injection. This is to prevent the coaxial nebulizing gas from forming a vacuum and siphoning liquid from the capillary outlet. To obtain reproducible injections, the same injection sequence is followed for each sample: (1) With make-up liquid and nebulizing gas turned off, the capillary is rinsed at 950 mbar for 3 min with run buffer, (2) the sample is injected for 0.1 min at 950 mbar, (3) the CE separation is started, (4) at 0.2 min, the make-up liquid, nebulizing gas, and electrospray voltage are turned on, and (5) at 1 min, the MS data acquisition commences. This sequence can easily be automated by allowing the API 300 mass spectrometer to send the start/stop signals to the capillary electrophoresis instrument and the make-up syringe pump, and to time-program the nebulizing valve and data acquisition. Method Validation. Once a satisfactory extraction and CE/ MS/MS method were developed for the analysis of human plasma, a complete validation of the method was undertaken. In general, method validation consists of demonstrating or establishing method specificity, linearity, accuracy, linear range, precision, detection limit, limit of quantitation, stability, and robustness.39 For the determination of EDTA by CE/MS/MS, the main goal was to test the general robustness of the methodology for bioanalytical applications. To date, there have been no published reports of a CE/MS method validation performed according to modern accepted protocols.39-41 Therefore, the expected precision and accuracy can only be compared to existing LC and LC/MS methods, which are generally less than 15% for all standard concentrations and 20% for the lower level of quantitation.39 These values may or may not be realistic for CE/MS, and realistic acceptance criteria cannot be established until after many more CE/MS method validations are reported. The method validation was carried out by fortifying human heparin-preserved plasma with disodium EDTA at six standard concentration levels in duplicate and 5-fold replicate quality control samples at each of three concentration levels. These standards were prepared by serial dilution from one EDTA stock solution. Quality control samples were prepared by serial dilution from a different (separately weighed) EDTA stock solution. In these experiments, a “run” is defined as an entire set of standards and QC standards described above, which are prepared fresh and are analyzed in one continuous session under identical conditions. These method validation runs were completed with five replicates of each QC concentration level within each run. The intra-assay precision and accuracy were calculated using the QC samples within each complete validation run. The interassay precision and accuracy of the method were calculated by comparing the results of the QC samples from three separate validation runs. RESULTS AND DISCUSSION Specificity. While there are many instances where UV detection may be sufficient for CE analyses, the determination of trace analytes in complex biological samples may best be trusted (39) Green, J. M. Anal. Chem. 1996, 68, 305A-309A. (40) Shah, V. P.; Midha, K. K.; Dighe, S.; McGilveray, I. J.; Skelly, J. P.; Yacobi, A.; Layloff, T.; Viswanathan, C. T.; Cook, C. E.; McDowell, R. D.; Pittman, K. A.; Spector, S. Pharm. Res. 1992, 9, 588-592. (41) Braggio, S.; Barnaby, R. J.; Grossi, P.; Cugola, M. J. Pharm. Biomed. Anal. 1996, 14, 375-388.

Figure 4. Demonstration of method specificity: (A) CE/UV of blank plasma (200 nm), (B) CE/UV of 1 µM Ni-EDTA-spiked plasma, (C) SRM-CE/MS of blank plasma (m/z 347 fragmenting to m/z 329), and (D) SRM-CE/MS of 1 µM Ni-EDTA-spiked plasma.

to a more specific and selective method of detection such as SRM mass spectrometry. Even with the best extraction procedures, matrix constituents can still interfere with other modes of detection. With SRM, the probability of interference is significantly reduced, which greatly improves the specificity of the method and the confidence level of the final results obtained. To illustrate this point, a blank plasma sample and a plasma sample spiked with 1 µM EDTA were diluted 1:1 with water. These samples were not cleaned by the SPE procedure. They were simply filtered and injected. The samples were analyzed using both CE/UV and SRM-CE/MS to demonstrate the selectivity of the mass spectrometer in this mode.42 The results are shown in Figure 4. The blank plasma in Figure 4A and the Ni-EDTAspiked plasma in Figure 4B are indistinguishable by CE/UV because of the excessive amount of chemical background detected by UV. In contrast, when analyzed by SRM-CE/MS, the same blank plasma sample in Figure 4C is free of all matrix peaks and the Ni-EDTA-spiked plasma in Figure 4D displays only the targeted Ni-EDTA peak. This example clearly shows the advantages of using tandem mass spectrometry for targeted analyses. Lower Level Detection (LOD) and Lower Level of Quantitation (LLQ). The LOD is generally defined as an analyte signal that is at least 3 times the average noise level.39,43 For the determination of EDTA in plasma, the LOD was found to be 7.3 ng/mL (∼3 fmol injected). Similarly, the LLQ is defined as the lowest concentration with acceptable accuracy and precision, or a signal that is 10 times the noise.39,43 For this method, the LLQ was determined to be 14.6 ng/mL (∼6 fmol injected). If this method was being used to determine whether a forensic blood stain had been “planted”, this LLQ corresponds to “planting” between 1 and 3 nL of EDTA-preserved blood (using the typical concentration of EDTA in blood samples to be 4.5 mM or 1300 ppm). Since it would be physically difficult to manipulate such a small volume, any such forensic sample would probably contain at least 1 µL. This hypothetical scenario illustrates the excellent sensitivity and potential forensic usefulness of the method. (42) Vessman, J. J. Pharm. Biomed. Anal. 1996, 14, 867-869. (43) Lindstedt, J. Met. Finishing 1993, (April), 64-70.

Injection Precision. Two aspects of injection precision were addressed: (1) reproducibility of repetitive injections of the same sample and (2) sample carryover from previous injections. A 197 ng/mL plasma extract was injected five times in sequence and demonstrated a 6.1% RSD for the peak area ratio that is similar to the individual variability of the raw peak areas. To address the possibility of carryover from sample injection, four samples were injected in the following sequence: (1) double blank plasma extract with no internal standard, (2) blank plasma extract with 50 ng of the (13C4)EDTA internal standard, (3) a very high concentration fortified plasma extract (4380 ng/mL), and (4) blank plasma extract. No injection carryover was observed. Although a very small analyte peak was detected in the blank sample electropherograms, the size does not increase after the high concentration injection, verifying that there is negligible injection carryover. Extraction Recovery. The amount of EDTA recovered from the plasma was estimated by preparing a double blank plasma extract and then spiking the extraction eluate after extraction with 50 ng of internal standard and 8.8 ng of EDTA. Since 100 µL of plasma is extracted, this amount of EDTA corresponds to a concentration of 88 ng/mL in plasma. A separate sample was spiked to the same levels of EDTA and internal standard prior to extraction. Both the extract and the postextract spiked samples were then evaporated to dryness and reconstituted in 30 µL of 10 mM Ni(NO3)2. The SRM-CE/MS area ratio of the postextract spiked sample was compared to that of the preextract spiked sample. The results showed that the sample extraction procedure recovers ∼88% of the added EDTA from the plasma matrix. At 88%, the recovery achieved was quite good. Since a stable-label isotope was used for the internal standard, any variance in the extraction recovery would be compensated by using the analyte/ internal standard area ratio. Method Linearity. The area ratio is calculated by dividing the analyte peak area by the internal standard peak area. The response factor39 is calculated by dividing the area ratio by the sample concentration in nanograms per milliliter. Although standards were analyzed in the range of 7.3-1460 ng/mL, the best fit by linear regression using 1/x weighting indicated a linear range from the 14.6 to 467 ng/mL standard concentrations. The concentration levels for the QC samples were therefore prepared at 22.2, 88.2, and 314 ng/mL with five replicates at each level. A representative calibration plot from one of the three separate runs is shown in Figure 5. The data from the QC runs were also plotted to visualize how well they relate to the calibration curve, but they were not used in the curve-fitting calculation. The linear regression results for the three method validation runs showed that the slope remained constant with a range of 0.0020-0.0021, while the y-intercept varied from 0.020-0.027. The curves were linear and demonstrated correlation coefficients in a range of 0.984-0.997. As a further test of linearity, the response factors for the three validation runs were plotted against concentration. Ideally, this plot should be linear with a slope of zero. Most of the data fulfilled this prediction within 15%, with higher variance at the LLQ. Precision and Accuracy. The precision and accuracy of the data were calculated by statistically analyzing the calculated concentration values obtained for the QC samples. The intraassay precision was determined at each QC level within each run by calculating the relative standard deviation (RSD) of the found concentrations. The interassay precision is determined at each Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

2905

Figure 5. Representative calibration curve from a method validation run.

QC level by calculating the RSD of the found concentrations from all three runs combined. The RSD is obtained by dividing the standard deviation by the mean and multiplying by 100. Similarly, the intra-assay and interassay accuracy are calculated as the percent bias of the found concentrations of the three QC levels. The percent bias from the nominal concentration is calculated by subtracting the nominal concentration from the found concentration, dividing by the nominal concentration, and multiplying by 100. The intra-assay precision ranged from 4.1 to 23.8% RSD, with all but one data point within 15% RSD. The intra-assay accuracy ranged from -11.7 to 22.5% bias, with only one data point greater than (12% bias. The interassay precision ranged from 7.9 to 18.4% RSD. The interassay accuracy ranged from -7.9 to 8.6% bias. Overall, these values are slightly higher than would be expected for a typical LC/MS method. However, typical ranges have not yet been established for CE/MS methods, and the values obtained for this method may indeed be representative of CE/MS. Representative electropherograms of a double blank, blank, LOD, LLQ, and internal standard taken from the third method validation run are shown in Figure 6. Both the analyte SRM transition of m/z 347 fragmenting to m/z 329 and the internal standard SRM transition of m/z 351 fragmenting to m/z 333 are shown. It should be pointed out that the ratio between the analyte and the internal standard is the significant measurement, not the absolute signal abundance. Although the internal standard concentration was the same in every sample, it became clear that the peak areas obtained for the internal standard varied widely throughout these experiments. Several observations suggested that variable injection volumes were the main cause of this variance. These observations included gradual or sudden increases in migration time that were usually reversible upon high-pressure rinsing of the CE separation capillary. This was an indication that capillary restriction was occurring over time. A further observation that suggested capillary restriction was that, in many cases, a marked increase in migration time was accompanied by a decrease in internal standard peak area. This capillary restriction was experienced even when using CE/UV, i.e., without a CE/MS interface. Other laboratory members have also noticed variable migration times and frequent capillary restriction with the amine-bonded capillar2906 Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

Figure 6. Representative SRM-CE/MS electropherograms from the third method validation run: (A) Double blank plasma, (B) blank plasma, (C) LOD in plasma (7.3 ng/mL), (D) LLQ in plasma (14.6 ng/mL), and (E) internal standard (500 ng/mL). CE conditions: 50 µm × 60 cm CElect-Amine capillary, 30 mM ammonium formate/ 0.15% formic acid at pH 3.5, separated with -30 kV plus 50 mbar inlet pressure.

ies. In general, these problems were more prevalent with the amine-bonded capillaries than with bare-fused-silica capillaries having the same inner diameter (50 µm). Determination of EDTA in Plasma Samples. To determine whether average samples of human heparin-preserved plasma contain any nominal level of EDTA from dietary or environmental sources, six additional sources of human plasma were analyzed. One double blank and one blank sample of each of these plasma samples were prepared. The double blank contains no internal standard, while the blank contains 50 ng (500 ng/mL) of the (13C4)EDTA internal standard. While some of the plasma samples contained detectable EDTA, all of the samples were below the LLQ (14.6 ng/mL) (data not shown). Considering the metabolism of EDTA, these results are not surprising.16 There is very little possibility of finding significant nominal levels of EDTA in the plasma.16 In agreement with these results, Ballard and colleagues also reported no detectable EDTA in unpreserved human blood using their GC/MS method.44 Determination of EDTA in Human Urine Samples. After the method was validated using human plasma, the applicability of the method to urine was investigated. The following examples demonstrate feasibility, not rigorous validation. Using exactly the same sample preparation as with plasma, a calibration curve was constructed using EDTA-fortified human urine with the same standard solutions prepared for the plasma. At the same time, three other human urine samples were prepared. Large-volume (estimated to be 13% of the CE column) pressure injections of 0.1 min at 950 mbar were again used. When the data were analyzed, it was observed that all of the urine samples contained detectable EDTA. Therefore, the quanti(44) Ballard, K. D.; Orkiszewski, R. S.; Taylor, M. S.; Johnson, E. A.; Ragle, L. Determination of EDTA in Forensic Samples by Capillary GC-MS and GCMS/MS. Proceedings of the 44th ASMS Conference on Mass Spectrometry and Allied Topics; ASMS: Portland, OR, May 12-16, 1997.

Figure 7. Representative SRM-CE/MS electropherograms of two human urine extracts and the internal standard.

ties of EDTA in the urine samples were calculated using the method of standard addition. Again, good linearity was demonstrated with a similar slope (0.0019) to the plasma calibrations. The detection level and lower level of quantitation were determined to be the same as with the plasma extraction. Representative SRM-CE/MS electropherograms of the human urine samples are shown in Figure 7. The estimated levels of EDTA in four urine samples ranged from 50 to 80 ng/mL. To determine whether the values obtained for the urine samples were valid, an estimate of the expected concentration was calculated using the following assumptions: (1) The estimated maximum daily dose of EDTA is 15 mg,9 (2) 5% of that dose, or 0.75 mg, is excreted in the urine within 24 h,16 and (3) average urinary excretion is 1200 mL/24 h.45 The result would be a maximum urinary concentration of 625 ng/mL. As with most estimates used in toxicity studies, the 15 mg level is probably much higher than the actual daily intake to ensure safe margins. Therefore, the EDTA concentrations found in these urine samples are certainly within reasonable expectations. Since the human urinary levels of EDTA are at least 10 times higher than plasma levels, urine could be used for monitoring EDTA exposure in humans. CONCLUSIONS After exhaustively searching the literature, we found no report for nominal levels of EDTA in human blood, let alone a method for accurately determining EDTA in blood at trace levels. The need for a specific, quantitative method for the determination of EDTA in human blood became evident. Since capillary electrophoresis is well suited for the separation of anionic EDTA metal chelates and CE/MS/MS can provide a unique combination of

separation and identification ability, we selected this application to demonstrate the utility and analytical robustness of the CE/ MS/MS technique. During the course of this research, a unique anion-exchange extraction procedure was developed for the isolation of EDTA from human plasma. The relatively clean extracts provided by the sample preparation allowed us to overcome the traditionally poor concentration detection levels of CE by using large-volume, fieldamplified sample injection. We also had the opportunity to evaluate amine-bonded CE capillaries since they were essential to the techniques employed. The amine columns afforded fast, high-resolution separation of the EDTA chelates and had remarkable ruggedness, as demonstrated by being usable for several months of daily use. However, the amine-bonded capillaries were marked by constantly varying migration times, which would have been a major problem if it were not for the high specificity afforded by mass spectrometric detection. Because of the migration time and injection volume differences, the use of a stable-label isotope as internal standard was necessary. It is predicted that the best results for quantitative CE/MS/MS will require a labeled internal standard just as this has become the norm in pharmaceutical LC/ MS analyses. Overall, it was found that CE/MS/MS could be practiced on a daily, routine basis with analytical ruggedness similar to LC/MS analyses. ACKNOWLEDGMENT The authors are grateful to Hewlett-Packard for the generous loan of the CE instrument that was used throughout the course of this work. We also thank PE-Sciex for providing the API 300 mass spectrometer and for partial funding for the laboratory. We also thank Hoffmann-LaRoche and Merck Research Laboratories for partial financial support, Monsanto for the loan of the Gilson APSEC instrument, and Supelco and Xerox for providing CElectAmine capillaries. The (13C4)EDTA internal standard was generously provided by Dr. K. Ballard of the Baylor College of Medicine. Thanks are due to the Henion research group, especially Dr. T. Wachs and Dr. M. Anderson. R.L.S. especially thanks Xerox for funding all of the tuition and many of the expenses incurred along the way. SUPPORTING INFORMATION AVAILABLE Table summarizing intra- and interassay precision and accuracy and figures showing SRM-CE/MS electropherograms and response factor vs EDTA concentration (3 pages). Ordering information is given on any current masthead page. Received for review January 21, 1997. Accepted April 24, 1997.X AC9700686

(45) Lehninger, A. L. Principles of Biochemistry; Worth Publishers: New York, 1982; p 703.

X

Abstract published in Advance ACS Abstracts, June 15, 1997.

Analytical Chemistry, Vol. 69, No. 15, August 1, 1997

2907