High-Throughput Screening of Abused Steroids in Urine Using Direct TOF-MS and LC–High-Resolution TOF-MS with Comprehensive Ion Fragmentation

2 Current Trends in Mass Spectrometry March 2012 www.spectroscopyonline.com

A high-throughput screening and confirmation workflow for the analysis of steroids and metabolites in urine was tested. Two urine extracts were analyzed per minute by flow injection electrospray-ionization ultrahigh-resolution time-of-flight mass spectrometry (TOF-MS). Confirmation by liquid chromatography (LC) with high-resolution TOF-MS with comprehensive ion fragmentation achieved sub-nanogram-per-milliliter sensitivity and yielded rich, repeatable fragment spectra that allowed confident analyte verification. Ultrahigh-resolution TOF-MS was chosen for screening, as this technique can easily resolve well over 10,000 chemically attributable signals in a single measurement of 1 s or less. LC with high-resolution TOF-MS was chosen for confirmation because it provides chromatographic and structural selectivity orthogonal to the screening technique.

Kevin Siek, Jeffrey S. Patrick, Erica A. Guice, Stephanie Amaya, and Joe Binkley

orkflows for monitoring abused chemical substances abused steroids in urine samples collected from an antidop- often use a high-throughput screening technique ing surveillance program. Urine samples were extracted and Wfollowed by an orthogonal technique to confirm screened using direct flow injection electrospray ionization positive results. This screening-confirmation strategy maxi- (ESI)ultrahigh-resolution time-of-flight mass spectrometry mizes sample throughput while minimizing the chance of false (TOF-MS). Two analyses per minute were recorded without positive results. These workflows typically use two independent advanced programming of the autosampler. technologies: immunoassay, planar chromatography, or a simi- Using exactly the same equipment, positive results were larly low-cost option for screening; and gas chromatography– confirmed by liquid chromatography–high resolution time- mass spectrometry (GC–MS) or liquid chromatography–mass of-flight mass spectrometry (LC–high resolution TOF-MS). spectrometry (LC–MS) for confirmation. A short LC column took the place of a union between the However, if a hyphenated chromatography–mass spectrom- autosampler and the ESI source, and the electrospray sol- etry system that is necessary for confirmation is purchased, vent became the mobile phase. Spectra deconvolved from maximum return on this investment could be achieved by using comprehensive collision-induced dissociation allowed ana- the same instrument for the initial high-throughput screening. lyte confirmation and enabled post hoc analyte-specific and Here, we describe a proof-of-concept application for detecting class-specific searches of the data. www.spectroscopyonline.com March 2012 Current Trends in Mass Spectrometry 3

Theory Screening with Flow-Injection 1.25 1.25 Ultrahigh-Resolution MS (a) Expected proﬁle at R = 52000 (FWHM) (a) Expected proﬁle at R = 88000 (FWHM) Scaled 1st derivative Scaled 1st derivative 1.00 1.00 High resolution MS can most likely re- 14 N 14 N 13 CH 13 CH solve signals from more analytes in a 0.75 0.75 single measurement than any other rapid 0.50 0.50 direct analysis technique. Nevertheless, 0.25 0.25 m/z m/z using this technique alone to screen for 0.00 0.00 499.9000 499.9850 499.9900 499.9950 500.0000 500.0050 500.0100 500.0150 500.0200 500.0250 500.0300 target formulas demands high perfor- 025 499.9800 499.9850 499.9900 499.9950 500.0050 500.0050 500.0100 500.0150 500.0200 500.0250 500.0300 025 mance of the mass spectrometer. Iso- baric interferences present risks for false Figure 1: Resolution of isobars modeled near m/z 500: (a) An instrument operating at 52,000 positive and also false negative results. resolving power would be unable to resolve the illustrated isobars present in a 1:1 ratio; (b) Ultrahigh resolving power, approaching an instrument operating at 88,000 resolving power would be able to resolve the same isobars 100,000 measured at 50% peak height present in a 1:10 ratio. (FWHM), is the minimum requirement for practical implementation. Isobaric interference is not limited (a) to monoisotopic peaks with the same 1000 5 4 nominal mass-to-charge ratio (m/z), but an isotope of one formula may obscure 800 the monoisotopic peak of another formula. Consider the formula exchange N 1 vs. 13CH. Such isobaric signals would dif- 600 fer by about 8 mDa. An instrument with 2 400 50,000 resolving power could not distin- 3 guish these signals if they were present in a 1:1 ratio near m/z 500. However, an 200 instrument with 90,000 resolving power is expected to resolve both of these signals 0 275 300 325 350 375 400 425 450 475 500 even if one is 10 times more abundant Time(s) than the other. Expected profile spectra (b) 1.6e51.6e5 generated using a Gaussian peak model Stanozolol and hydrhydroxystanozololoxystanozolol are shown in Figure 1. 1.4e51.4e5

1.2e5 Confirmation by LC–MS with THG and hydroxylated THG Comprehensive Collision-Induced 1.0e5 Dissociation Unambiguous formula assignment 8.0e4 achieved by high-resolution MS is not suffi- 6.0e4 cient for confident identification; orthogo- 4.0e4 nal information is required to discriminate amongst isomers and unresolved isobars. 2.0e4

Chromatographic separation and colli- 0.0e0 275 300 325 350 375 400 425 450 475 500 sion-induced dissociation (CID) provide Time(s) complementary structural selectivity to enable such discrimination. However, iso- Figure 2: A segment of the ultrahigh resolution screening analysis: (a) extracted ion traces: 1 lation of a single nominal m/z before CID = THG, 2 = hydroxylated THG, 3 stanozolol, 4 = hydroxylated stanozolol, 5 = testosterone;. diminishes two of the touted advantages (b) multiple signal correlation traces: greater than zero only if signals for an unmetabolized associated with high resolution MS: the steroid and its hydroxylated metabolite correlate within a flow-injection peak. ability to mine data post hoc for analytes of interest and the ability to leverage accurate Comprehensive CID must be accom- Post Hoc Signal-of-Interest isotope abundance in formula generation. panied by accurate spectral deconvolu- Extraction Comprehensive CID, in which all coeluted tion that correctly reports product and Selected reaction monitoring experi- ions are simultaneously fragmented and precursor signals to be useful. Deconvo- ments performed with essentially 100% analyzed, maintains the advantage of ret- lution algorithms have been developed duty cycle are the benchmark for sensi- rospective analysis and takes full advantage that are capable of this; examples are tive target analysis, but have no capability of a high-resolution mass analyzer. shown with the results. for retrospective data analysis. Further- 4 Current Trends in Mass Spectrometry March 2012 www.spectroscopyonline.com

(Morristown, New Jersey). [AUTHOR- correct location for Burdick & Jackson?] 319.22660 313.19842 329.21095 LC–MS-grade formic acid and ammonia 329.22208 329.22984 1.4e4 were obtained from Sigma-Aldrich (St.

328.19452 Louis, Missouri). Polyethylene glycol was 313.21577 313.23615 1.2e4 313.19085 obtained from Alfa Aesar (Ward Hill,

1.0e4 Massachusetts). Urine samples were ob- 313.2 313.22 313 329.20 329.21 329.22 329.23 329.24 tained from an antidoping surveillance 8.0e3 program. The samples were blinded; Intensity (counts) no sample history was received, and no 314.17522

6.0e3 329.00500 327.00786 identification other than a sample num- 323.25796 331.20921

4.0e3 321.24214

320.23004 ber was given. 312.15696

2.0e3 Sample Preparation

0.0e0 Samples were prepared using conven- 332 332 332 332 332 332 332 332 332 332 m/z tional procedures of hydrolysis followed by solid phase extraction. Details have Figure 3: Ultrahigh resolution spectrum of a flow injection peak showing signals for THG (m/z been given by Guice (2). 313.216) and hydroxylated THG (m/z 329.211). Flow-Injection Ultrahigh-Resolution MS An Agilent 1290 series pump and autosampler were used for flow injection sam- (a) 323.17677 (b) 329.25774 2.0e4 pling (Agilent Technologies, Santa Clara, 1.0e4 1.8e4 8.0e3 1.5e4 1.3e4 California). 6.0e3 1.0e4 4.0e3 7.5e3 5.0e3 The injection volume was 3 μL and 2.0e3 2.5e3 0.0e0 0.0e0 the flow rate was 300 μL/min; the injec- 100 150200 250 300 100 150 200 250 300 350 Peak True (Cid) - sample “3002KS2011w43-049 150 ng_mL”, Peak 49, at 262.183 s Peak True (Cid) - sample “3002KS2011w43-069 150 ng_mL”, 19 Stanozolol, at 278.75 s tion solvent consisted of a volume of 10%

143.02559 329.25776 Intensity (counts) Intensity (counts) Intensity (counts) mobile phase A and 90% mobile phase B 4.0e3 7.0e3 3.5e3 6.0e3 as described in the section discussing LC

3.0e3 4 5.0e3 9 9 131.0256 0 81.10444 4.0e3 81.10444 8 323.1768 2 2.0e3 3.0e3 1.5e3 procedures. 2.0e3 345.1587 345.1587 4 113.0706 113.0706 9 95.06012 175.1479 1.0e3 175.1479 9 157.0412 0 135.0913 3 121.0756 121.0756 7 5.0e2 1.0e3 149.110700 0.0e0 0.0e0 Detection was achieved using a LECO 100 150 200 250 300 100 150 200 250 300 350 m/zm/z m/m/zz Citius LC-HRT high-resolution TOF-MS system with an ESI source operated in Figure 4: Deconvolved spectra: (a) clostebol at collision offsets of 10 V (top) and 60 V (bottom); positive-ion mode (LECO Corporation, (b) stanozolol at collision offsets of 10 V (top) and 75 V (bottom). St. Joseph, Michigan). The MS system was operated in ultrahigh-resolution mode, more, Thurman and Ferrer (1) discussed theoretically improve the signal-to-noise which yielded a mass resolving power the concept of a crossover point between ratio (S/N) as long as the second m/z has of about 100,000 measured at FWHM. multiple reaction monitoring (MRM) at least 41% abundance with respect to Spectral data were acquired from m/z experiments and high-resolution experi- the first. Although the assumption of 160 to m/z 600 at a rate of one spectrum ments. In an MRM experiment, each in- homogenous, random noise is not valid every 0.3 s. The m/z axis was externally dependently monitored reaction within in high-resolution MS, this still provides calibrated with a dilute (0.01%) solution a time segment dilutes sensitivity in a some guidance for selecting signals to of polyethylene glycol in methanol before linear fashion. At some point, high reso- co-add. analysis of the samples. lution measurements become more sen- Consecutive duplicate injections of sitive than multiple reaction monitoring Experimental each sample were made. An injection was experiments. Materials made approximately every 30 s, allowing Comprehensive CID retains full capa- Methandrostenolone, stanozolol, 3-hy- sufficient time for flushing of the injec- bility of post hoc data interrogation and droxystanosolol, and 3-hydroxystanozo- tion system between injections. allows some strategies for signal enhance- lol-d3 (deuterated internal standard) were ment that are not available in an isolated obtained from Cerilliant (Round Rock, LC–MS with Comprehensive CID ion CID experiment. For example, signals Texas). Clostebol, 6β-hydroxyboldenone, An Agilent 1290 series pump, autosam- from different adduct species or frag- and oxandrolone were obtained from pler, and thermostated column compart- ments may be combined mathematically Alltech (Deerfield, Illinois). [AUTHOR- ment were used. A 50 mm × 2.1 mm, 2.2- to enhance the signal. If noise is homog- correct location for Alltech?] LC–MS- μm dp Restek Ultra II C18 column was enous and random, adding a second m/z grade water and methanol were obtained used (Restek Corporation, Bellefonte, to an extracted ion chromatogram will from Honeywell Burdick & Jackson Pennsylvania). Mobile phase A was an www.spectroscopyonline.com March 2012 Current Trends in Mass Spectrometry 5 aqueous solution of 26 mM formic acid Table I: Quantitative results and 1 mM ammonia that was prepared by Concentration 6-β Hydroxy- 3-Hydroxy- adding 1 mL formic acid and 0.1 mL 25% (ng/mL) boldenone stanozolol aqueous ammonia to 1 L of water. Mobile 300.00 402390 197530 phase B was methanol. The flow rate was 360 μL/min. The mobile-phase composi- 30.00 48481 22498 tion was 10% B at time zero, followed by a 3.00 4792 1488 linear gradient to 40% B at 0.25 min, and 1.00 2292 327 a linear gradient to 100% B at 6.25 min. 100% B was held isocratically for 1.5 min 0.30 1609 ND before re-equilibration at 10% B for 1.25 Slope 2668 1314 min. The injection volume was 5 μL and the column temperature was 40 °C. r2 0.999653 0.999992 Detection was achieved using a LECO Concentration Oxandrolone Methandro- Clostebol Stanozolol Citius LC-HRT high-resolution TOF-MS (ng/mL) stenolone system with an ESI source operated in 150.00 437306 772486 519461 1897309 positive-ion mode. The mass spectrome- 15.00 49787 84442 56905 216595 ter was operated in high-resolution mode, which yielded a mass resolving power of 1.50 9674 10629 5345 22791 about 50,000 measured at FWHM. Spec- 0.50 3783 5264 900 5519 tral data were acquired from m/z 75 to 0.15 107 939 87 1523 m/z 525. For one sequence of injections, MS with CID was performed at two al- Slope 2895 5131 3460 12615 ternating collision offsets: 10 V and 50 r2 0.999935 0.999907 0.999905 0.999986 V. In a second sequence of injections, the alternating collision offsets were 15 V and 75 V. One spectrum at low collision offset and one spectrum at high collision off- 1.6e4 Methandrostenolone Stanozolol set were acquired every 0.3 s. During the 15 ng/mL 15 ng/mL 1.5 ng/mL 1.5 ng/mL 1.4e4 0.5 ng/mL 0.5 ng/mL first 20 s of the isocratic 100% methanol 6β-Hydroxyboldenone Oxandrolone 1.2e4 30 ng/mL 15 ng/mL segment of the gradient, a dilute (0.01%) 3.0 ng/mng/mL 1.5 ng/mng/mL Clostebol 1.0 ng/mng/mLL solution of polyethylene glycol in metha- 1.0e4 0.5 ng/mL 15 ng/mL 1.5 ng/mng/mLL nol was coinfused with the eluent, using 0.5 ng/mL 8.0e3 a time-programmable syringe pump that is integrated with the mass spectrometer. 6.0e3 3-Hydroxystanozolol 4.0e3 This facilitated postacquisition recali- 30 ng/mL 3.0 ng/mL bration of the m/z axis for each sample. 2.0e3 1.0 ng/mL

Calibrant solution was not coinfused 0.0e0 100 150 200 250 300 350 while any peaks of interest were being Time(s) eluted, and background signals from the calibrant ions were reduced to negligible Figure 5: Overlaid enhanced extracted ion chromatograms for six steroids.. levels by the time the 100% methanol isocratic wash and gradient re-equilibration Figure 2a shows extracted ion traces for There are several advantages to this were complete. testosterone, THG, stanozolol, and hy- screening approach. Direct analysis droxylated metabolites of the latter two combines all isomeric hydroxylated me- Data Processing steroids. THG was designed specifically tabolites into one signal, enhancing sen- All data was acquired and processed to evade detection by targeted analytical sitivity for analytes that have multiple using LECO ChromaTOF-HRT soft- methods, and was used as an illicit ergo- significant hydroxylation sites. However, ware. Expected profile spectra shown genic aid for several years before antidop- the the acknowledged limitation is that in the theory section were generated in ing authorities discovered it. However, xenobiotic substances cannot be analyzed Microsoft Excel using a Gaussian model. untargeted, direct, ultrahigh-resolution in the presence of isomeric endogenous MS analysis indicates the presence of compounds. Results and Discussion THG and a hydroxylated metabolite in at Another advantage is illustrated in Screening with Flow-Injection Ultra- least one pair of replicate injections. Hy- Figure 2b. Here, settings in the data high-Resolution MS droxylated stanozolol and some unme- processing software were configured Screening indicated the possible presence tabolized stanozolol also are clearly seen to return a nonzero trace only if sig- of exogenous steroids in several samples. in another pair of replicate injections. nals from both an unmetabolized 6 Current Trends in Mass Spectrometry March 2012 www.spectroscopyonline.com

tern of two major fragments differing by exactly 12 Da is commonly observed in Methandrostenolone 301.21566 steroid-derived drugs including clostebol, 6.0e3 8.0e3 6.0e3 oxandrolone, testosterone, and others.

5.0e3 4.0e3 283.20521

2.0e3 149.13218 121.06460 323.19771 Thus, this spectral feature could be used 4.0e3 Oxandrolone 0.0e0 100 150 200 250 300 350 as a class-specific screen for other steroids 3.0e3 Intensity (counts) 121.06455 8.0e3 with similar structures. 2.0e3 7.0e3 3-Hydroxystanozolol 6.0e3 5.0e3 4.0e3 Figure 4b shows the deconvolved 1.0e3 3.0e3 149.13223 301.21`559

2.0e3 93.06979 283.20509 107.08539 135.08027 173.09581 323.19766 1.0e3 161.13232 0.0e0 0.0e0 precursor ion and product ion spec- 180 190200 210 220 230 240 250 100 150 200 250 300 350 m/z m/z tra of stanozolol. Fragmentation of 307.22665 345.25325 4.0e3 2.0e3 3.5e3 1.8e3 the presucor in a manner such as that 3.0e3 289.21613 1.5e3 2.5e3 1.3e3 2.0e3 1.0e3 1.5e3 7.5e2

324.25313 observed here (and common among 271.20559 1.0e3 339.25267 5.0e2 367.23535 257.18986 5.0e2 361.23486 2.5e2 0.0e0 0.0e0 100 150 200 250 300 350 100 150 200 250 300 350 sterols) can make derivation of selec- m/z m/z

Intensity (counts) Intensity (counts) tive MRM transitions challenging. 93.06979121.10106 345.25328 2.3e3 1.2e3 1.8e3 1.0e3 289.21606 Comprehensive fragmentation allows 8.0e2 1.3e3 147.11670 346.25668 307.22671

6.0e2 257.19000 271.20563 229.19500 239.17937 161.13232 7.5e2 173.13243 4.0e2 213.16320 367.23533

361.23484 all product ions to be monitored simul- 85.064469 111.08031 153.09109 179.10658 189.16365 193.12201 243.21079 2.0e2 200.15137 339.25206 2.5e2 0.0e0 0.0e0 100 150 200 250 300 350 100 150 200 250 300 350 taneously, then combined together by Time(s) Time(s) data processing to yield an enhanced Figure 6: Deconvolved spectra of coeluting steroids, each at collision offsets of 10 V (top) and signal. This is not time-dependent nor 60 V (bottom). is there a limit to the number of fragments to be correlated with a precursor ion. In addition, unlike MRMs, the produict ions may in fact be correlated 250,000 with one another to create signal en- 20,000 hancement in the chromatogram. Fur- ther enhancement in S/N is achieved 200,000 10,000 by requiring selected precursor and 0 product ions to be concurrently pres- 150,000 012 ent before a nonzero signal is returned. Stanozolol Figure 5 shows the resultant enhanced extracted ion chromatograms for six 100,000 Linear (Stanozolol) steroids. Figure 6 shows the deconvolved spectra of three coeluting steroids: 50,000 R2 = 0.9999 oxandrolone, which fragments exten- sively at 60 V collision offset; meth- 0 androstenolone, which yields a few in- 0 5 101520 tense fragments at 60 V collision offset; and 3-hydroxystanozolol, which yields Figure 7: Unweighted least squares regression for the four lowest levels of stanozolol. no measureable fragments at 60 V collision offset. Note that the oxandrolone steroid and its hydroxylated metabo- Confirmation by LC–MS with CID and methandrostenolone signals are lite correlate within a flow-injection Steroid standards were obtained com- completely deconvolved from the 3-hy- peak. Requiring the presence of both mercially and diluted as shown in Table droxystanozolol spectra. Also, note that parent and metabolite increases the I. The most concentrated solution of each both oxandrolne and methandrosteno- confidence of positive results, and can analyte was analyzed using MS with lone yield intense fragments of nominal allow these positive results to be eas- comprehensive CID (MSc2), as described m/z 121. In a typical QqQ analysis, this ily spotted amongst a long series of in the experimental section. could lead to cross-talk. However, high- injections. Processing the acquired data yielded resolution MS easily distinguishes these Figure 3 shows a signal-rich spec- deconvolved precursor ion spectra that two fragments that represent a formula trum that contains signals for THG and were aligned with corresponding de- difference of CH4 vs. O. hydroxylated THG. Average resolving convolved product ion spectra. Such Quantitative results are listed in power within the displayed segment is spectra are complete with isotopes for all Table I. Figure 7 shows the unweighted more than 90,000. The analyte signals signals and adducts in some cases. Fig- least squares model for the lowest four were assigned with relative m/z errors ure 4a shows the MSc2 of clostebol. Note levels of the stanozolol standard curve. of -1.4 ppm for THG and -0.5 ppm for that the two principal fragments differ Similar to a typical MRM analysis, the hydroxylated THG. by one carbon. This characteristic pat- enhanced extracted ion chromatogram www.spectroscopyonline.com March 2012 Current Trends in Mass Spectrometry 7 shows high selectivity and a robust linear correlation. Comprehensive CID also yields 4.5e4 5 5 4.0e4 more information for confirmation 4.0e4 3.5e4 4 of analyte identity by facilitating the 3.0e4 3.5e4 monitoring several qualifier ions. A 2.5e4 3 4 3.0e4 2.0e4 sample positive for stanozolol and 1.5e4 hydroxylated stanozolol by the di- 2.5e4 1.0e4 3 5.0e4 rect analysis screening method was 2.0e4 0.0e0 2 220 225 230 235 240 1 analyzed using the MSc technique for 1.5e4 Time(s) confirmation. Two chromatographic 6 peaks were observed in the resulting 1.0e4 2 enhanced extracted ion chromatogram 5.0e3 for stanozolol, and multiple hydroxyl- 0.0e0 100 150 200 250 300 350 ated stanozolol peaks were observed. Time(s) Figure 8 shows enhanced extracted ion chromatograms for these analytes. Figure 8: Stanozolol-related analytes detected in a sample: 1 = stanozolol; [M+H]+ = 329.259, Stanozolol, 3-hydroxystanozolol, and principal fragment = 81.045; 2 = stanozolol-related; [M+H]+ = 329.259, principal fragment

3-hydroxystanozolol-d3 (the internal = 81.045; 3 = 3-hydroxystanozolol-d3 (internal standard); [M+H]+ = 348.272, principal standard) are confirmed by compari- fragment = 97.040; 4 = 3-hydroxystanozolol; [M+H]+ = 345.254, principal fragment = son with authentic reference material. 97.040; 5 = hydroxystanozolol-related; [M+H]+ = 345.254, principal fragment = 81.045; 6 = Basesd on the principal fragments, the hydroxystanozolol-related; [M+H]+ = 345.254, principal fragment = 81.045. other peaks may be an epimer of stanozolol and metabolites hydroxylated in become more facile. In silico libraries provide both rapid screening and high- positions other than the 3 position. will provide much of the information performance quantitative opportunities. needed to make these effective. Conclusions MSc2 suits high-resolution analysis References As we discussed here, significant ad- because it takes full advantage of a high (1) E.M. Thurman and I. Ferrer, in Liq- vances continue to be made in high- resolution mass analyzer. MSc2 main- uid Chromatography Time-of-Flight performance MS. The gains achieved tains the highest capability for retro- Mass Spectrometry: Principles, Tools, may allow new workflow strategies to be spective data analysis and makes com- and Applications for Accurate Mass considered in the future. plete precursor and product ion spectra Analysis (Wiley, Hoboken, New Jersey Flow-injection ESI high-resolution MS that are available for confirmation of an- 2006). Author- page numbers? can process sample at rates that are com- alyte identity. The utility of these spectra (2) E.A. Guice, “Anabolic Androgenic petitive with conventional robotic immu- depends on the accurate deconvolution Steroid Testing by Liquid Chroma- noassay platforms. However, because of algorithms. Algorithms that are capable tography Quadrupole Time-Of- the need to distinguish background ions of accurately distingushing precursor Flight Mass Spectrometry in Urine,” from analyte ions, and from one another, and product ions of near-dead coelu- presented at Mass Spectrometry a resolving power of about 100,000 is the tions have been developed. Signals from Applications to the Clinical Lab, San minimum requirement to make direct deconvolved spectra can be adapted Diego, California, 2011. MS analysis a viable alternative for high- and combined to retrospectively extract throughput screening applications. Ide- highly selective quantitative information Erica A. Guice is with Western Slope ally, direct high-resolution MS should regarding any known or unknown ana- Laboratory, LLC, in Troy, Michigan. be preceded by sample cleanup proce- lyte in the sample. Sensitivity remains dures that are selective for particular independent of the number of signal Kevin Siek, Jeffrey S. Patrick, compound classes. In addition, system combinations extracted. Stephanie Amaya, and Joe stability (mass drift over short times) in The data described above demonstrate Binkley are with separation science the low parts-per-million range and in- the overall utility of high-performance division at LECO Corporation, in St. Joseph, herent root-mean-square mass accuracies TOF-MS in the qualitative and quantita- Michigan. Please direct correspondence to: of approximately 1 ppm are required. tive analysis of drugs of abuse using one [email protected]. ◾ As the resolving power of commer- of the more challenging examples — ste- cial instruments continues to increase, rols. The lower ionization efficiency and this approach will become more prac- complex fragmentation make these less tical for general screening. Compiling desirable and more challenging in MRM For more information on this topic, libraries of ions expected in certain analyses. However, results indicate that please visit our homepage at: matrices will become faster and dis- high-performance TOF and the lever- www.spectroscopyonline.com covery of unexpected substances will aging of high-resolution fragment ions