| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
Athletic Drug Testing and Toxicology Laboratory, Indiana University Medical Center, Medical Science Bldg. A-128, 635 Barnhill Dr., Indianapolis, IN 46202-5120.
a Author for correspondence. Fax 317-274-3223; e-mail lbowers{at}iupui.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Key Words: indexing terms: quadrupole ion trap • quadrupole mass filter • GC-MS • abused drugs • clenbuterol
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Screening for anabolic agents in urine is a difficult analytical problem, given the large number of compounds sought, their low concentration, and their similarity to endogenous steroid compounds. Under routine SIM GC/MS screening conditions, >95 ions are monitored during a 20-min chromatographic run. Selected storage of ions of interest should dramatically affect detection limits under these circumstances. We report here on the application of selected-ion storage (SIS) techniques to the detection of trimethylsilyl (TMS) derivatives of anabolic agents in urine.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
99%; Aldrich, Milwaukee, WI)
were used as received. Steroid standards were prepared in
chromatography-grade methanol (CMS, Houston, TX). The solutions were
dried, and the TMS-enol-TMS ether derivatives of the
steroids were formed by reaction for 15 min at 60 °C with 100 µL
of a solution of MSTFA containing 3 mL of 1-propanethiol and 2 g
of NH4I per liter. We extracted steroids from urine by a modification of the method of Donike et al. (8). Briefly, 3 mL of urine was extracted with solid-phase extraction columns, deconjugated at pH 7.0 with ß-glucuronidase from Escherichia coli (Boehringer Mannheim Diagnostics, Indianapolis, IN), and extracted with tert-butylmethyl ether (EM Science, Gibbstown, NJ). The extract was dried and derivatized with 100 µL of MSTFA/1-propanethiol/NH4I as above.
The derivatization mixtures were injected directly without further treatment into a Varian (Walnut Creek, CA) Saturn III GC/MS equipped with a Model 3400 GC for all QIT studies. Injections were made with a Model 8200 autosampler into a Model 1078 split/splitless injector (also from Varian) operated in the splitless mode. The split vent was closed for the first 0.7 min after injection and the injection rate was 0.2 µL/s. A polysilphenylene-polydimethylsiloxane bonded-phase capillary column (DB-5 ms, 0.25 mm x 30 m, 0.25 µm df; J & W Scientific, Folsom, CA) was used throughout. The flow rate of the He carrier gas was adjusted to 1 mL/min at 250 °C. The temperature program for analysis of tetrachlorobenzene began at 80 °C, ramped to 175 °C at 15 °C/min, and finally ramped to 300 °C at 25 °C/min. The GC temperature program for anabolic steroids began at 170 °C, ramped to 260 °C at 20 °C/min; ramped to 305 °C at 2.7 °C/min; and finally ramped to 320 °C at 20 °C/min.
The Saturn GC/MS system was equipped with a Wave~Board and operated with Saturn GC/MS (Version 5.2) software. The Ion Trap Toolkit for Selected Ion Storage software was used to create frequency-modulated SIS (FM-SIS) ion preparation methods (IPMs). The Ion Trap Toolkit for MS/MS software was used to create unit mass resolution SIS (µSIS) IPMs. IPMs were programed to be applied at specified times during a chromatographic run.
Because of the width of the mass isolation bands, FM-SIS IPMs were constructed by entering the m/z value of the ions to be selectively stored. No more than four ion ranges for the tetrachlorobenzene experiments or more than two ion ranges for the anabolic agents were isolated. The automatic gain control (AGC) target value was set to either 5000 or 10 000 counts. Default software values were used for the remainder of the SIS settings.
The µSIS IPMs were constructed by entering the m/z values of each ion to be selectively stored. The collision-induced dissociation amplitude value was set to 0 V and the time to 0 ms. The AGC target value was set to 5000 counts. Default software values were used for the remainder of the µSIS settings.
For the QMF comparison data, we used a Hewlett-Packard (Little Falls, DE) 5972 Mass Selective Detector equipped with a 5890 Series II GC and a Model 7920 autosampler. Injections were made in the splitless mode. Temperatures and program rates were the same as above. We used a Hewlett-Packard Ultra 2 polydimethylsiloxane capillary column (0.25 mm x 15 m; 0.25 µm df). For data acquisition we used Hewlett-Packard's ChemSystem UNIX-based data system with autotune settings and the multiplier gain increased 200 V. For the anabolic steroid analyses, we used our routine GC/MS method, with GC conditions nearly identical to those reported above. The SIM acquisition, however, was performed in nine groups of ions, ranging from 6 to 16 ions per group, with dwell times of 10–60 ms during a 21-min chromatographic run.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
, top). Note that the S/N profile remains constant for the QIT
operated in the multiple ion storage mode (FM-SIS), while the S/N ratio
for the QMF decreases with increasing numbers of ions monitored, the
crossover point occurring at about four ions. The decrease in S/N for
the QMF is primarily related to the required reduction in dwell time; a
similar decrease in S/N ratio can be observed when monitoring a single
ion and varying the dwell time (Fig. 1
, bottom).
|
To assess whether the advantage of the QIT could be extended to urine
samples, we added four anabolic steroids (norandrosterone;
5
-androstan-17-methyl-3,17ß-diol;
5ß-androst-1-en-17-methyl-3,17ß-diol; and
3'-hydroxystanozolol) and the ß-agonist/anabolic agent clenbuterol to
a blank urine matrix and analyzed the resulting mixture by both
ion-monitoring techniques. Results from full-scan and SIS acquisition
approaches on the QIT are shown in Fig. 2
. Panels A and B display the total ion chromatographic
data. Using SIS leads to substantial increases in AGC ionization time
and decreased numbers of chromatographic peaks. Panels C and D show the
ion traces of only the m/z 405 ions of norandrosterone,
acquired during the various acquisition approaches. The first large
peak in the m/z 405 ion trace (8.7 min) is from ~1.5 ng of
norandrosterone. Clearly, there is much less improvement in detection
limit when only the extracted ions of interest are monitored. Note also
the number of peaks in the urine matrix that have the same unit mass as
norandrosterone.
|
To better understand the limitations of the ion storage algorithms, we
looked at the improvements in storage of specific ions. Using FM-SIS,
we found that the fraction of ions in the trap at the m/z of
interest increased substantially (Table 1
). In most cases the improvement in storage efficiency
correlated well with improved detection limits. The second factor in
determining the efficiency of storage is the ionization time used by
the AGC, which can be adjusted to produce the targeted number of ions
in the trap. A decrease to <25 000 µs indicates the presence of a
large amount of material entering the QIT and thus a reduction
in absolute signal from any analyte being ionized. The benefit of a SIS
algorithm is to maximize the storage of the ions of interest. As can be
seen, in many cases the QIT has been filled with compounds other than
the one of analytical interest and therefore has not maximized the
analytical signal.
|
Although µSIS potentially has some advantages with regard to mass
storage range, operation requires that one mass range be isolated at a
time, the default value for the isolation range being ± 1.5 Da.
This approach loses the duty-cycle (decreasing observation time with
increasing numbers of ions) advantage of multiple ion storage on the
QIT. The operation of the QIT in the µSIS mode results in a S/N vs
number of ions profile (Fig. 1
, top, 
) similar to that of the QMF.
Thus, although an improved S/N was observed for as many as four ion
ranges stored, the advantage was lost when additional ions were
monitored.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The stable ion trajectories required for storage in a QIT are a function of the mass-to-charge (m/z) ratio of the ion, the radiofrequency (RF) storage voltage, the frequency applied to the ring electrode, and the trap geometry (7). The stable motion of an ion of particular m/z occurs at a characteristic secular frequency. Application of an oscillating waveform to the endcap electrodes with a frequency identical to the secular frequency causes that ion to absorb energy efficiently, overcome the trapping field constraints, and be ejected from the QIT. Thus a combination of endcap waveforms and RF voltage changes can selectively store (or eject) ions from the QIT. Recognition of this fact has resulted in several new approaches to SIS (11)(12)(13)(14)(15).
Two fundamentally different approaches to ion storage have emerged
commercially. The first approach applies a variety of frequencies to
the endcap electrodes, omitting frequencies that correspond to the
secular frequencies of the ions of interest. Variations in the ring
electrode voltage may (11) or may not
(12)(13)(14) be used to facilitate ion ejection. Thus,
multiple ions of interest can be stored in the trap simultaneously and
scanned out sequentially. This would remove the limitations of duty
cycle observed for sequential ion transmission in the QMF (Fig. 1
). The
difficulty inherent in this approach is that secular frequencies of
ions become increasingly similar as mass increases. Thus, under the
frequency and geometry conditions reported for QIT operation,
m/z 100 and 150 differ by 58 kHz but m/z 450 and
500 by only 4 kHz (7). Accordingly, the selection and
frequency width of the ejection waveform becomes more critical at
higher masses.
The second approach uses a combination of endcap waveforms and ring
electrode RF voltage changes to isolate a single m/z in the
trap during each scan function. This approach is used in the Finnigan
SIM and Varian µSIS modes of operation. Because the scan function is
computer-controlled, a different ion can be isolated on each successive
scan. This mode of operation is similar to SIM with the QMF, but with
an important difference: After ion isolation, the ions are ejected from
the QIT by the scan function and detected externally. At present, the
QIT has no ability to adjust dwell time, as a QMF does, to improve the
S/N ratio. Any improvements in S/N ratio are the result of an increase
in the number of ions of interest stored in the QIT relative to
background. Thus, our observation that the µSIS mode of operation is
similar to the SIM mode (Fig. 1
, top) with respect to increasing
numbers of monitored ions is not
surprising.
We have utilized two approaches in the present report: FM-SIS, a frequency-modulated approach for isolating a range of ions (11), and µSIS, a two-step ejection scheme capable of isolating a single mass unit (15). Our initial hypothesis with respect to selected ion operation was that the QIT would provide lower limits of detection than the QMF. This conjecture was based on the fact that the QMF must sequentially focus a single ion on the detector, whereas multiple ions can be stored simultaneously if the appropriate waveform is applied to the QIT. Because at least 10 scans must be obtained across the width of the chromatographic peak to adequately quantify ion ratios (16), an increase in the number of ions monitored must result in a decrease in QMF dwell time. The fraction of time available for any single ion decreases not only because of shorter dwell times, but also because of system overhead, e.g., quadrupole rod settling and data management operations (9). As expected, when more than four ions are acquired simultaneously, the QIT has a S/N ratio, and thus a limit of detection, advantage. This advantage could be very important for anabolic steroid screening because 10–20 ions are acquired simultaneously in each SIM group during the chromatographic run. Given that three or four ions must be acquired to meet compound identification criteria for coeluting natural and deuterium-labeled internal standards monitored in forensic workplace testing, one would expect the observed improvement in limit of detection with increasing numbers of ions to be relevant for workplace testing as well.
We also evaluated hexachlorobenzene (m/z 284) as a model compound but achieved no improvement in S/N on the QIT because of ions trapped from column bleed (m/z 281). This interference decreased the total number of analyte ions in the QIT and decreased the S/N ratio. Thus interferences from either the instrument background or matrix components within the ion storage range continue to present a problem for the QIT. An example matrix ion is the endogenous steroid androsterone (m/z 434 for the di-TMS derivative), which is present in urine at concentrations 100-fold greater than some of the anabolic steroids screened for. Because the width of the ion isolation band for this mass in FM-SIS is ~±18 Da, this compound would be acquired in the isolation window for norepiandrosterone (m/z 420), decreasing the ion accumulation time in AGC and therefore also decreasing the absolute signal from the compound of interest.
Our observations with the TMS-enol-TMS ether
derivatives of four model anabolic steroid metabolites and the
ß-agonist clenbuterol extracted from a urine matrix mirrored those of
the hexachlorobenzene. Ideally, a SIS algorithm would allow
simultaneous storage of an unlimited number of selected-mass ions with
100% efficiency while having very sharp-edged mass discrimination
against adjacent masses. Because of the method in which the ions are
ejected in the FM-SIS approach (11), however, the minimum
isolation range increases with mass (Fig. 3
). In the mass range of 450 Da observed for anabolic steroids, a
±19 Da window will be isolated. This broad mass isolation range limits
mass selectivity and results in a large number of different mass ions
being formed with concomitant activation of AGC (Table 1
).
|
The RF storage voltage is involved in both the efficiency of mass
storage and the secular frequency spacing of ions of different
m/z (Fig. 3
). Unfortunately, the present version of SIS
software allows storage voltages corresponding to between 32 and 60 Da
only. All of the studies presented here were acquired at the default
storage voltage corresponding to 48 Da. We were unable to explore the
use of the storage voltage systematically because of the inability of
the commercial software to apply waveform corrections to compensate for
shifts in the calculated secular frequencies. Moreover, given that the
isolation range is an important aspect of improved S/N, the
"sharpness" of the edges of the isolation band is of some concern.
When we changed the lower edge of the FM-SIS window from m/z
405 to 406, the intensity of the m/z 405 ion from
norandrosterone decreased 300-fold—an indication that the isolation
band edges are quite sharp.
The mass isolation range limitations in the two-step ejection scheme used in µSIS (15) are somewhat different. The default isolation range is ± 1.5 Da. If the range is decreased to ± 0.5 Da, the intensity of the selected mass ion is decreased several orders of magnitude. Increasing the isolation range to ± 2.5 Da increases the signal to about twice that of the default value. The decision with respect to mass isolation range in µSIS is related to the source of the limiting noise—electronic vs chemical (e.g., matrix, column bleed, etc.). Again, the sequential nature of µSIS removes the duty-cycle advantage observed with multiple ion storage; therefore, we did not explore µSIS further.
The stored waveform inverse Fourier transform (11) and
filtered noise (12) SIS approaches are reported to have
more ideal characteristics with respect to storage efficiency and unit
mass storage than the FM-SIS algorithm reported here. Careful
examination of the unit mass display of m/z 405 (Fig. 2
, C
and D) suggests that, in screening for anabolic steroids, unit mass
storage and resolution may not be sufficient. The baseline noise, which
limits the S/N ratio, is the result of the complex matrix and
derivatization chemistry. This suggests that, although improvements in
ion isolation could improve detection limits by perhaps fivefold,
improved sample preparation and derivatization schemes or improved
chromatographic resolution could result in more impres-sive
improvements. The other potential solution for the problem
described here is the use of higher mass resolution. High mass
resolution has been reported for the QIT
(17)(18)(19) but has not been tested under
analytical conditions as difficult as those discussed here.
Estimates of the S/N ratio for a urine extract of 10 anabolic steroid
metabolites by using the full-scan–extracted ion profile on the QIT
compared with SIM on the QMF indicated 2- to 10-fold better performance
by the QMF. For the five model anabolic agents studied, the
improvements in S/N ratio for SIS vs full scan were two- to fivefold.
Although the response is compound-dependent, this suggests that limits
of detection for SIS on the QIT are comparable with those for SIM on
the QMF, but not significantly better—primarily because of the complex
urine matrix, which includes endogenous steroids and fatty acids.
Similar findings were reported for norandrosterone by de Boer et al.
(5), using an earlier version of single-ion storage. Wu et
al. have reported two- to fivefold improved detection limits for
9-carboxytetrahydrocannabinol, methamphetamine, benzoylecgonine,
morphine, and phencyclidine from the use of SIM relative to full scan
on the Finnigan ITS40 (4), also in agreement with our
study. The fact that several SIS algorithms fail to improve the limits
of detection is, in our opinion, an indication that no unit mass
resolution ion-isolation algorithm will provide significantly improved
results.
One of the advantages of the QIT is the variety of MS techniques that can be used. We reported on our experiences with MS/MS on the QIT earlier (20). Although we obtained detection limits of 1 pg on-column for norandrosterone by using temperature-programed injection and MS/MS, this technique is best suited for confirmation. The use of MS/MS for screening has been suggested. However, the necessity for application of a large number of sequential ion isolation and optimized collision-induced dissociation conditions will result in the loss of the duty-cycle advantage and suggests that satisfactory detection limits will be difficult to achieve. Nevertheless, the prospect of having a single mass analyzer that can perform computer-selectable high mass resolution, rapid scanning, SIS, and tandem mass spectroscopy at a price comparable with that of bench-top GC/MS instruments is exciting.
In summary, the QIT has demonstrable advantages over the more widely used QMF under specific SIS conditions. The QIT offers many alternative methods for storing and manipulating ions, which makes true optimization and comparison of results from various instruments difficult. As regards screening for urinary anabolic steroids, we observed no significant improvement in limits of detection relative to the QMF for either mixtures of standards in urine or extracted urine metabolic studies. The presence of either instrumental interferences (e.g., column bleed) or matrix constituents that produce large numbers of ions in the mass isolation range can degrade QIT performance despite the use of SIS. Unfortunately, the similar masses of steroids likely to be present in the urine matrix presents exactly this situation.
|
Acknowledgments |
|---|
. We
also acknowledge many useful scientific exchanges with Carl Feigel, Bob
Brittain, and Terry Sheehan of Varian Associates. |
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  L. D. Bowers Analytical advances in detection of performance-enhancing compounds Clin. Chem., July 1, 1997; 43(7): 1299 - 1304. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. L. Fitzgerald, D. A. Herold, and A. L. Yergey Trade-offs in Mass Spectrometry Clin. Chem., June 1, 1997; 43(6): 915 - 915. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) reflect the fact that a single mass isolation range
contained four, three, and two of the monitored ions such that the same
S/N ratio was determined in all cases. All data points represent the
average of six experiments.
