Inductively Coupled Plasma Mass Spectrometry for Trace Element Analysis in the Clinical Laboratory*

KERN L. NUTTALL, M.D., Ph.D., WILLIAM H. GORDON, CLS (NCS), and K. OWEN ASH, Ph.D.

Department of Pathology, University of Utah School of Medicine, Salt Lake City, UT 84108

ABSTRACT

Inductively coupled plasma mass spectrometry (ICP-MS) is a relatively new technique for trace element analysis. The basic operating principles of ICP-MS are described and our experience with this technique in a clinical setting is discussed for the analysis of serum, whole blood, and urine. Advantages to ICP-MS include the favorable detection limits (0.01 to 0.1 |xg/L for many elements), simple specimen preparation, high throughput (about 40 specimens per hour), and the ability to measure more than one element simultaneously. A major disadvantage is the high capital cost of the instrumentation. Heavier elements, such as lead, are well-suited for ICP-MS analysis, whereas lighter elements are prone to more interfer ences. Lighter elements which are not amenable to assay by ICP-MS include chromium and iron. The ability to measure isotopes is a major advantage for mass spectrometry methods and has the potential to expand the usefulness of trace element analysis.

Introduction powerful technique is now beginning to find acceptance for analysis of clinical This article is intended to introduce specimens. Our experiences are laboratorians to the potential of induc described with ICP-MS for analysis of tively coupled plasma mass spectrometry trace elements which began in 1989 (ICP-MS) as an emerging technology for when methods for commercial instru analysis of trace elements in biological ments! were developed for analysis of specimens. Until recently, ICP-MS has trace elements in biological specimens, been primarily used in the environmen primarily urine, serum and whole blood. tal field for analysis of water and air. This The basic operating principles are described and both the advantages and limitations to ICP-MS are discussed. * Send reprint requests to: Kern L. Nuttall, M.D., Ph.D., Department of Pathology, University of Utah School of Medicine, % ARUP—500 Chipeta Way, t Perkin-Elemer Elan 500 and, subsequently, Salt Lake City, UT 84108. Elan 5000. 264 0091-7370/95/0500-0264 $01.20 © Institute for Clinical Science, Inc. INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY FOR TRACE ELEMENT ANALYSIS 265 The ICP-MS Instrument men is lost to the drain of the spray cham ber. More efficient direct injection sys Inductively coupled plasma mass spec tems are available,3 but have not to date trometry is a relatively new technique. been implemented in our laboratory. The first analytical mass spectra were Direct injection into the plasma has the obtained from an ICP in 1978, and the potential to increase the throughput sig first commercial instruments were intro nificantly and decrease the specimen vol duced by Sciex in 1983.1 A typical instru ume required for analysis. ment (figure 1) consists of an ICP torch A typical ICP torch consists of three connected via an interface to a mass spec concentric quartz tubes and a coil which trometer. The sample solution to be ana carries high-energy radiofrequency lyzed is transferred by a peristaltic pump power in the range of 1 to 3 kilowatts. to a nebulizer which converts the solu The radiofrequency power acts on the tion into an aerosol. The aerosol is car argon exiting the middle tube to produce ried by argon gas into the center of the a plasma of more than 6000°K. The ICP torch. The high temperature plasma plasma is created by the energetic colli vaporizes and ionizes the sample, and sions among the argon atoms oscillating directs ions into the mass spectrometer, in the high-energy radiofrequency field. where the elements of interest are The outer tube carries argon gas which detected on the basis of the mass- flows at a high rate to act as a barrier or to-charge (m/z) ratio. A major disadvan coolant layer to prevent the torch from tage to ICP-MS is the high capital melting in the intense heat of the plasma. cost, although recently introduced The inner tube carries the aerosolized benchtop instruments are considerably sample into the center of the plasma. less expensive .2 Specimens introduced into the plasma A nebulizer is an easy and inexpensive are rapidly vaporized, atomized, and ion way to introduce a sample solution into ized. The flow rate of the argon carrier the ICP torch. Our instruments use gas positions the plasma for sampling Meinhard-type nebulizers made of boro- into the mass spectrometer. silicate glass, which produce an aerosol The ICP torch can also be coupled to when the argon carrier gas is forced an atomic emission spectrometer (ICP- through a small capillary into a spray AEM), which uses photon emission to chamber. One problem with this design detect trace elements. While ICP-AEM is that more than 95 percent of the speci instruments are less expensive, they may

ICP Torch Interface Mass Spectrometer

Nebulizer Quadrupole F ig u r e 1. Schematic of a typical instrument for inductively coupled plas mas spectrometry. The inductively coupled plasma (ICP) torch, which operates at atmo spheric pressure, is cou pled to the mass spec -Argon Sample *” Gas trometer through an Solution interface. RF Power IVacuuml I Vacuum] Supply 1 Pump 1 1 Pump 1 Computer 266 NUTTALL, GORDON, AND ASH also suffer from greater interferences and mined by the ability of the Channeltron do not have the low detection limits electron multiplier to handle intense ion of ICP-M S.3 beams and is about six orders of magni The ICP-MS interface must transfer tude above the detection limit, thereby ions from the torch at several thousand providing a wide usable range for most degrees Kelvin and atmospheric pressure trace elements. The mass resolution and into a mass spectrometer which operates computer capabilities of the ICP-MS in a vacuum at low temperature. The first often permit multiple trace elements to stage of the interface consists of a sam be analyzed at the same time, thereby pler cone with an orifice of about 0.5 mm significantly increasing efficiency in the positioned to collect ions from the laboratory.4 For example, in our labora plasma. Because of the intense heat gen tory assays for arsenic, cadmium, anti erated by the torch, the sampler cone mony, tellurium, thallium, lead, and bis must be cooled in an efficient manner muth are conducted at the same time. with refrigerated water. The sampler The ICP-MS instruments often achieve cone opens into an interface region, a signal stability of about ± 5 percent rela which is pumped down to approximately tive standard deviation over several 1 torr by a mechanical pump. The second hours, and an internal standard can stage of the interface is the skimmer improve this to about 1 percent. Prob cone, which is placed several millimeters lems with instrumental drift can occur behind the sampler orifice. The orifice of with specimens containing high concen the skimmer cone opens directly into the trations of salts which can deposit on the vacuum chamber of the mass spectrome sampler and skimmer cones. To monitor ter. A quadrupole mass filter is usually drift, control material is run every 15 used to provide low-cost mass discrimi specimens; when the control falls out of nation. (More costly high-resolution mass range, the instrument is recalibrated. spectrometers can circumvent many of Assays for heavy elements, such as lead, the isobaric overlaps that occur with quad- tend to be remarkably stable and seldom rupoles). Ions exiting the quadrupole require recalibration, whereas lighter are detected by a Channeltron electron elements such as aluminum require con multiplier (CEM). siderably more frequent recalibration. Approximately 40 patient specimens can Performance Characteristics be processed per hour with the major determinant of throughput being the Over 50 elements have detection limits time required to pump a specimen in the range of 0.01 to 0.1 (xg/L, often through the nebulizer and into the torch. matching or surpassing those of atomic In general, the instruments have been absorption spectrometry .3 Only eight ele easy to maintain. Routine maintenance ments (carbon, nitrogen, oxygen, fluo requires approximately 1 0 to 2 0 m in per rine, silicon, phosphorus, sulfur, and day and consists primarily of cleaning the chlorine) have detection limits greater torch and sampler cone. Minor problems than 10 (Jig/L owing to their low degree of consist of items such as breaking of the ionization in the plasma torch. Detection torch during cleaning. Twice a year the limits are usually defined as three stan manufacturer schedules preventative dard deviations above the mean of the maintenance, which usually requires a blank and are determined under ideal full day off-line. With instruments less conditions. Practical detection limits are than three years old, approximately one often an order of magnitude poorer. The major failure per instrument/year has upper limit of the linear range is deter occurred where downtime lasted 3 to 7 INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY FOR TRACE ELEMENT ANALYSIS 267 days. Our older instrument (Elan 500) is Interferences now suffering more frequent failures. Unavailability of parts has been the chief Because a mass spectrometer discrimi factor in the length of downtime. nates on the basis of the mass-to-charge ratio, interferences in ICP-MS come pri Specimen Preparation marily through mass overlap from: (1 ) Specimen preparation typically con polyatomic ions, (2 ) elements with the sists of diluting or digesting the speci same isotopic masses (elemental isobaric men in mineral acids and analyzing the overlap), and (3) doublely charged ions. resulting solution. As with all trace ele Matrix effects can also cause signal sup ment analysis, care must be taken to pression and/or enhancement. Since like insure that sample preparation does not charges repel in the ion beam of a mass add exogenous contaminants to the spectrometer, as the number of like- specimen. A dilution step in the prepara charged ions increases, mutual repul tion reduces salt concentrations and sion can expel ions from the beam . 1 The matrix effects and is appropriate when use of an internal standard minimizes sufficient sensitivity is available. An matrix effects. internal standard adds a level of control Polyatomic interferences arise from which compensates for variations in recombination events which occur in the instrumental parameters, such as the rate intense heat of the ICP plasma torch. of sample introduction into the torch as Recombination events can include well as for matrix effects. To use a spe oxides, chlorides, nitrides, and sulfides, cific example, whole blood drawn for as well as argon species arising from the lead determination is treated as follows: argon carrier gas. Ions unique to plasma (1 ) the specimen is mixed well on a tube conditions include species like oxychlo- rocker; (2) 0.50 mL of specimen is added ride (OCl+) and argon chloride (ArCl+). to 0.50 mL of 12 mol/L nitric acid in a Polyatomic interferences are most sig glass tube (containing a yttrium internal nificant for the first transition series of standard); (3) 4.0 mL deionized water is elements. For example, 4 0 A r1 2 C + and added, and the contents mixed by vortex- 4°Ar 16o + interfere with the major iso ing briefly; (4) the tube is centrifuged at topes of chromium and iron. The largest 1000 g for 5 min to remove precipitated polyatomic species which forms in the protein and cellular debris; and (5) the argon-driven plasma torch is the argon supernatant is poured through a 2 |xm fil dimer (ArAr+). Above a mass-to-charge ter into a polyproplylene tube for loading ratio of 85, the background is essentially onto the instrument. In this manner, 50 free of polyatomic interferences. specimens can be prepared for analysis Elemental isobaric overlaps arise in approximately 25 min. when two isotopes of different elements Mercury is an example of an assay have a similar mass-to-charge ratio. For which requires a modification of the typi example, the ions 5 8 F e + and 5 8 N i+ haves cal preparation previously described . 5 masses 57.9333 and 57.9352, respec The standard specimen preparation tively. A typical ICP-MS quadrapole- results in poor recovery of mercury, prob type detector does not have sufficient ably through the loss of volatile nitrates. resolving power to distinguish between Instead of nitric acid, hydrochloric acid is the two, although other types of mass used for specimen digestion in combina spectrometers (e.g., a magnetic sector tion with cysteine and ethylene diamine mass spectrometer) have the potential to tetraacetic acid (EDTA) to trap the mer resolve this type of overlap. cury in solution. Doubly charged ions occur for some 268 NUTTALL, GORDON, AND ASH elements with low ionization potentials, cant memory effect where a residual sig notably calcium, strontium, barium and nal from silver remained elevated for the lanthanides. An ion of mass M and several hours; this is likely due to the charge 1 + will be indistinguishable from deposition of silver onto the sampling an ion of mass 2M and charge 2 +. Thus, and skimmer cones. 4 8 Ca2+ can cause a signal at the 2 4 M g+ peak, and Ba2+ species interfere with Low Weight Elements several isotopes of zinc. In general, the interferences from doubly charged Elements <85 amu are prone to a vari ions are few and can be avoided by ety of interferences, particularly poly- proper positioning of the ICP torch with atomic-type interferences.3 The elements the interface . 1 chromium and iron are not practical to assay by ICP-MS, at least in biological High Atomic Weight Elements specimens, owing to the interferences present. In contrast, the isotopes of man Elements >85 amu are well-suited for ganese (5 5 Mn) and copper (6 3 Cu) are rela ICP-MS analysis being essentially free of tively free of significant interferences. polyatomic interferences. Lead is an For other low atomic mass elements, a example of an assay which has proven to variety of methods are available to be stable and accurate. With the current reduce interferences. Zinc has several emphasis on lead testing, more than natural isotopes, 64Zn being the most 3.000 whole blood specimens are being abundant and the one targeted by ICP- performed per month and well over MS. However, an isobaric-type interfer 100.000 have been performed since 1989. ence arises from nickel (6 4 Ni). Since the The ICP-MS whole blood lead assay isotope ratios for nickel are invariant, the has performed admirably in proficiency amount of 64Ni in a specimen can be testing with the College of American determined by measuring 6 0 Ni. Thus, the Pathologists (CAP), the New York State amount of nickel contributing to the sig Health Department, and the Pennsylva nal at mass 64 can be subtracted to give a nia Department of Hygiene. The only more accurate determination of zinc. problem has come from the bias intro Selenium is an example of an element duced into proficiency surveys by less which suffers from a polyatomic interfer accurate methods. It is worth noting that ence, but for which there is an alternate the Center for Disease Control (CDC) isotope to monitor. The most abundant now uses ICP-MS to determine the lead isotope of selenium (8 0 Se, 49.8 percent concentration in its reference material abundance) has the same mass as the (see page 269). argon dimer (4 0 Ar4 0 Ar+), which forms at Our experience with other heavy ele relatively high concentrations in the ments include, antimony, bismuth, cad argon-based plasma torch. However, the mium, mercury, platinum and thallium. A isotope 82Se (present at 9.19 percent tellurium assay performs well in urine, abundance) is suitable for measuring although there is a significant interfer selenium in serum and urine. Arsenic ence in blood specimens; the nature of provides an example of compensating for this problem has not been investigated as an interference using the method of iso- there is currently little demand for blood baric fractionation . 6 Argon chloride tellurium. A gadolinium assay works well (40 Ar35Cl + ) forms in specimens contain in urine but has not been evaluated in ing high chloride concentrations. Since blood. An initial attempt to assay for sil arsenic has only one isotope (7 5 As), ver was not successful owing to a signifi choosing a different isotope is not possi INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY FOR TRACE ELEMENT ANALYSIS 269 ble to avoid this source of interference. target values for its certified reference However, by monitoring the 160 3 5 C1, the material (lead in bovine blood) supplied chloride interference can be corrected. A under the Blood Lead Laboratory Refer similar adjustment is made for aluminum, ence System (BLLRS). Another useful which also has only one isotope (2 7 A1) aspect of isotopic analysis is in kinetic corresponding to a polyatomic interfer studies which employ enriched nonra ence from 1 3 C 14N + . W hen nitrogen is dioactive isotopes to study trace element present in excess (as w hen the specim en metabolism. Alternatively, when the is prepared using nitric acid precipita natural isotopic composition varies from tion), the signal at 13C is directly propor source to source as it does with lead, tional to the interference, and can be sub the isotopic composition might be used tracted from the signal at mass 27. This to identify the probable source of sort of strategy may prove useful to lead poisonings.8,9 extend the number of elements which can be assayed by ICP-MS. Reference Ranges Isotope Analysis Reference ranges for the most fre The ability to measure trace element quently requested trace elements deter isotopes is a major advantage for mass mined by ICP-MS in a Utah population spectrometry methods. Isotope dilution are shown in tables I and II; reference mass spectrometry assays have the poten ranges published in a recent edition of tial to be reference methods for many ele Tietz Textbook of Clinical Chemistry 10 ments such as lead .7 The CDC uses iso are included for comparison. Other ele tope dilution ICP-MS to establish the ments for which ranges have been devel

TABLEI

Selected Trace Elements in Serum and Whole Blood from Utah Adults Determined by Inductively Coupled Plasma Mass Spectrometry

Utah Published Mean ± SD Range Range Element Specimen n [ig/dL \ig/dL \ig/dL

Aluminum Serum 91 1.2 ± 0.4 0 .4 - 2.0 < 1 . 0 - 14.0 Arsenic Blood 89 2 .6 + 0.7 1 .1 - 4.1 0.2 - 6.2 Cadmium Blood 93 0.1 ± 0.1 0.0 - 0.3 0.1 - 0.5 Copper Serum Male 69 83.6 ± 15.5 53.0- 115.0 7 0 .0 -1 4 0 .0 Female 24 108.0 ± 23.8 60.0- 115.0 8 0 .0 -1 5 5 .0 Lead Blood 55 6.3 ± 3.7 0 .0 - 14.0 <40.0 Manganese Blood 91 1.2 ± 0.2 0 .7 - 1.7 Mercury Blood 57 0.5 ± 0.3 0.0 - 1.2 0.1 - 5.9 Tellurium Blood 93 1.1 ± 0.8 0 .0 - 3.0 Zinc Serum 69 1 1 5 .0 1 2 1 .0 72.0- 157.0 7 0 .0 -1 5 0 .0

Specimens were collected from healthy nonpregnant laboratory employees. Utah ranges were calculated from the mean ± 2 SD. 270 NUTTALL, GORDON, AND ASH

TABLE II

Selected Trace Elements in 24-Hour Urine Collections from Utah Adults Determined by Inductively Coupled Plasma Mass Spectrometry

Utah Published M e an ±S .D . Range Range * Element l-tg/d [ig/L \ig/d v-g/d

Aluminum 3.7 ± 3.1 2.8 ± 2.1 0 - 10.0 Antimony 0.3 ± 0.5 0.3 ± 0.4 0 - 1.3 Arsenic 32.5 + 15.7 24.5 ± 14.1 0 - 64.0 5-50 Bismuth 0.9 + 1.0 0.7 ± 0.4 0 - 2.9 < 2 0 Cadmium 1.3 + 1.0 1.0 ± 0.8 0 - 3.3 < 15 Lead 13.0 ± 8.9 9.4 ± 6.6 0 - 3 1 .0 < 8 0 Manganese 5.3 ± 2.6 4.1 ± 1.9 0 - 11.0 Mercury 5.2 ± 2.6 3.7 ± 1.9 0 - 1 5 .0 < 2 0 Tellurium 2.4 ± 2.0 1.8 ± 1.3 0 - 6.4 Thallium 0.8 ± 0.6 0.6 ± 0.4 0 - 2.1 < 2

Specimens were collected from healthy nonpregnant laboratory employees (25 male and 25 female); nosexbiaswasfound. Utah ranges were calculated from the mean ± 2 SD. Them ean24hour urine volume was 1.4 L (SD 0.4 L).

oped by us include random urine plati erence interval, it may still be well below num (less than 4 ixg/L) 11 and urine the concentration at which toxic symp gadolinium (not detected; limit of detec toms may appear. Defining ranges for tion 0.1 jxg/L). Trace elem ents are per deficiency and toxicity are often complex formed by alternative methodologies and a matter of considerable debate. (excluding common electrolytes) include chromium and iron. Conclusion A number of factors should be consid ered when discussing trace element Inductively coupled plasma mass spec ranges. First, specimens for trace ele trometry is a practical method for many ment determinations are prone to con trace elements in the clinical laboratory tamination, and ranges have fallen over setting. Major advantages to ICP-MS the years as improvements have been include a high specimen through-put made in collection procedures and ana with little specimen preparation, and a lytic techniques. Second, reference inter large dynamic range with low limits of vals are dependant on the population detection. In the future, ICP-MS is studied. For example, a Utah population expected to be able to survey rapidly which eats relatively little fish has sig specimens for a trace element profile. A nificantly lower mercury exposure than a major disadvantage to ICP-MS is the high similar population on the coast of Ore capital cost, although recently introduced gon. Third, the lower limit of a reference benchtop instruments are considerably interval does not define deficiency nor less expensive. The heavier elements does an upper limit define the toxic >85 amu are particularly well suited to range. When a specimen is above the ref ICP-MS analysis, whereas the lighter INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY FOR TRACE ELEMENT ANALYSIS 271 elements suffer from more interferences. 5. Kalamegham R, Ash KO. A simple ICP-MS pro Chromium and iron are two elements cedure for the determination of total mercury in whole blood and urine, J Clin Anal 1992;6: that remain impractical to assay in clini 190-3. cal specimens by ICP-MS. The ability to 6. Kershisnik MM, Kalamegham R, Ash KO, Nixon measure isotopes expands the applica DE, Ashwood ER. Using 160 35C1 to correct for chloride interference improves accuracy of tions of trace element analysis. urine arsenic determinations by inductively coupled plasma mass spectrometry. Clin Chem 1992;38:2197-202. References 7. Aggarwal SK, Kinter M, Herold DA. D eterm i nation of lead in urine and whole blood by sta 1. Houk RS. Mass spectrometry of inductively ble isotope dilution gas chromatography-mass coupled plasmas. Anal Chem 1986;58:97A. spectrometry. Clin Chem 1994;40:1494-502. 2. Potter D. An ICP-MS instrument for the mod 8. Yaffe Y et al. Identification of lead sources in em laboratory. Am Lab 1994;26(ll):35-7. California children using the stable isotope 3. Templeton DM. Inductively coupled plasma- ratio technique. Arch Environ Health 1983;38: atomic emission spectrometry (ICP-AES) and 237-45. inductively coupled plasma-mass spectrometry 9. Ketterer ME. Assessment of overall accuracy of (ICP-MS). In: Seiler HG, Sigel A, Sigel H, edi lead isotope ratios determined by inductively tors. Handbook on Metals in Clinical and Ana coupled plasma mass spectrometry. J Anal lytical Chemistry. New York: Marcel Dekker, Atomic Spectro 1992;7:1125-9. 1994:167-79. 10. Painter PC, Cope JY, Smith JL. Reference inter 4. Vaughan M-A, Andrew DB, Templeton DM. vals. In: Burtis CA, Ashwood ER, editors. Teitz Multielement analysis of biological sample by Textbook of Clinical Chemistry; Philadelphia: inductively coupled plasma-mass spectrome Sanders, 1994:2175-211. try. II. Rapid survey method for profiling trace 11. Nuttall KL, Gordon WH, Ash KO. Breast elements in body fluids. Clin Chem 1991;37: implants and urinary platinum. Clin Chem 210-5. 1994;40:1787.