| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
1
PE Sciex, 71 Four Valley Dr., Concord, Ontario L4K 4V8, Canada.
a Author for correspondence. Centers for Disease Control, Cham 17/2424 F-19, 4770 Buford Highway, NE, Atlanta, GA 30341-3724. Fax 770-488-4609; e-mail jtb2{at}cdc.gov
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
Furthermore, the adverse effects of smoking are not necessarily limited to active users of tobacco products. In recent years, concern about the health risks experienced by nonsmokers who are involuntarily exposed to tobacco smoke, e.g., fetuses exposed in utero to nicotine and other compounds from maternal smoking or nonsmokers through passive smoking, i.e., exposure to environmental tobacco smoke (ETS),1 has increased. ETS is similar, but not identical, to the mainstream smoke inhaled during active smoking, and many of the hazardous substances known to be present in mainstream smoke are also present in ETS. In fact, because of differences in combustion and aging, some of these substances are actually more prevalent in ETS than in mainstream smoke. Strong evidence indicates that ETS represents a serious and substantial public health problem in the US (4). ETS has been classified as a Group A (known human) carcinogen by the Environmental Protection Agency, is believed to be responsible for ~3000 lung cancer deaths per year among nonsmokers, and has been causally linked to many other health problems as well (4). Because of these serious health concerns, an assessment of exposure to tobacco products was included in the design of the Third National Health and Nutrition Examination Survey (NHANES III). The objective of these analyses was to determine the extent of both active tobacco use and ETS exposure in a statistically representative sample of the entire US population (5).
In general, both active smoking and exposure to ETS have been assessed either by interview or by biochemical analysis. In addition, ETS exposure has sometimes been inferred from environmental air analysis. Biochemical measurements of appropriate markers are often a useful adjunct to interviews in the classification of smokers and other active users of tobacco products. Such measurements are particularly valuable in assessing exposure to ETS because individuals may differ in their awareness of the extent and duration of such exposures. Markers of exposure to cigarette smoke include carbon monoxide (carboxyhemoglobin), thiocyanate ion, nicotine, and cotinine, a primary metabolite of nicotine.
At present, cotinine (Fig. 1
) is generally regarded as the best marker for monitoring
tobacco exposure in either actively or passively exposed individuals
(6). Nicotine, the most tobacco-specific component of
cigarette smoke that is present in relatively abundant amounts (~1–2
mg/ cigarette), is clearly absorbed and is measurable in both active
and passive smokers (7)(8). However, nicotine
is rapidly metabolized and has a half-life of only 1–2 h, so it is
rather poorly suited to monitoring chronic exposure. By contrast,
cotinine has a much longer half-life of ~18–20 h, making it more
appropriate for use as an exposure marker (7). Cotinine
can be reliably measured in blood, saliva, and urine, and all three
sources are generally regarded as acceptable for monitoring nicotine
exposure in people (6)(9)(10). For
the quantitative assessment of both active and passive smoking in the
NHANES III population, we selected serum cotinine.
|
Many methods have been described for measuring cotinine in blood (10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22)(23)(24)(25)(26)(27)(28)(29)(30). These assays are generally based on GLC with flame ionization, nitrogen–phosphorus, or mass spectral detection, HPLC with UV detection, or immunoassays. Such methods have been shown to be reliable for quantitating serum cotinine concentrations >10–15 µg/L, the range normally encountered in active smokers (9)(10)(31), and in some cases, these methods may also be used for measuring concentrations in the range of ~2–10 µg/L, which is consistent with relatively substantial passive exposure to ETS. However, some of these methods are not sensitive or selective enough to reliably measure serum cotinine at the lower passive exposure concentrations, which may extend to well below 1 µg/L. Because our objective for NHANES III was to monitor the entire range of nicotine exposure, from that of active smokers to those who are only minimally exposed to ETS, we needed a procedure for these analyses that provided both excellent sensitivity and high selectivity. In addition, because of the large number of samples involved in this study, the new method needed to be capable of maintaining high sample throughput over an extended period of time.
To meet these requirements for a very sensitive and selective method of analyzing serum cotinine that is also capable of continual high throughput, we developed a new method based on HPLC linked to atmospheric pressure chemical ionization tandem mass spectrometry (APCI MS/MS). This method, which has been in routine operation in our laboratory for >4 years, combines the sensitivity and selectivity for serum cotinine assays required for NHANES III and similar studies with the rapid analysis and robust performance needed for such large-scale surveys (32).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
Calibration standards.
One complete set of cotinine
calibration standards was prepared for use throughout the entire NHANES
study. All glassware was washed and silanized before use. A group of 12
standards with cotinine concentrations ranging from 0 to 25 µg/L
(serum equivalents) and an internal standard (ISTD),
cotinine-D3, at 5 µg/L were prepared in toluene. The
stock ISTD solution used for supplementing samples was prepared in
isopropanol, 40 mL/L in toluene. Standards were analyzed to confirm
their suitability and then aliquoted into washed, silanized glass
ampules, flame-sealed, and stored at -20 °C until used. Generally,
a new set of ampules was opened every 2 weeks.
QC materials.
Four serum pools were prepared for use in this
study. Two were bench QC pools with nominal cotinine concentrations,
determined during characterization assays, of ~202 and ~1.84
µg/L; these were used in high and low cotinine sample assays,
respectively. Two aliquots of the bench QC pools at the appropriate
concentration were included in each assay, one at the beginning of the
assay and the other at the end. The other two pools were blind QC
materials. After the blind pools were prepared, we aliquoted them into
vials with labels indistinguishable from those of standard NHANES
sample vials and randomly inserted them into assays according to a
computer-generated protocol. The high and low blind QC pools had
nominal cotinine concentrations (as determined during the initial
characterization screening) of ~212 and ~0.262 µg/L,
respectively. The number of blind QC samples varied for each assay, but
averaged between 2 and 3 vials per assay of 50 samples. In addition, we
required that at least one blind QC sample be present in every assay of
50 samples; when at least one sample was not designated at random by
the assay-generating software, an additional blind QC position was
inserted manually.
Preparation of materials.
Although background contamination
from environmental cotinine is far less of a problem than is commonly
observed with nicotine, very small amounts of cotinine are present in
ETS in both the gas and particulate phases (34)(35)(36), and
background cotinine from environmental sources can potentially
interfere in very sensitive analyses. Therefore, the careful and
rigorous control of possible interference sources is very important in
these assays. We sought to minimize background interferences by first
permanently banning any smoking in the entire building in which the CDC
cotinine analysis laboratories are located; this ban went into effect
more than a year before any NHANES analyses were carried out. In
addition, all analysts working on any sample-handling aspect of this
project were required to be nonsmokers. The initial sample extraction
was carried out in reusable 10-mL Oak Ridge-type Teflon tubes that were
cleaned by hand with Liqui-Nox laboratory detergent (Alconox, New York,
NY), sonicated, thoroughly rinsed with deionized water, and then dried
and sent to the central CDC automated glassware washing facility to be
cleaned again. On return, the tubes were rinsed once more with
methanol, air-dried, and stored in sealed containers over desiccant
until used. The samples used for predesignated high and low cotinine
concentration assays were kept separate at all times and were
periodically spot-checked for any evidence of residual contamination.
Glassware that contacted samples was disposable and was precleaned and
silanized before use. This glassware was stored in sealed packages
protected from the environment until it was used and then discarded.
All prepared aqueous reagents were made up in the laboratory according
to designated schedules with the use of ORGANICpure water and protected
from atmospheric exposure at all times except when they were being
used. All prepared reagents and pure solvents were routinely tested for
evidence of cotinine contamination.
Instrumentation.
Analyses were conducted with a PE Sciex API
III atmospheric pressure ionization triple quadrupole mass spectrometer
(Perkin-Elmer Sciex Instruments) with heated nebulizer interface
installed. The mass spectrometer was interfaced with a short length of
fused-silica tubing (~0.10 mm x 0.5 m) to a Perkin-Elmer
base-deactivated C18 column (4.6 mm x 3 cm; 3-µm
particle size) mounted in a Hewlett-Packard Model 1090L liquid
chromatograph equipped with an autosampler. The entire system from
sample injection to data acquisition was computer-controlled with
standard Sciex software.
Sample preparation.
The range of serum cotinine concentrations
can encompass >4 orders of magnitude when both smokers and nonsmokers
are included. Therefore, all NHANES samples were first prescreened by
an enzyme-linked immunoassay (EIA) for cotinine (STC Diagnostics,
Bethlehem, PA) and placed into one of two categories, high or low, on
the basis of a nominal cutoff concentration of ~25 µg/L as measured
by the EIA. Individual LC/MS/MS assays of 50 samples each were then
built from these preclassified groups. Each NHANES assay included 2
water blanks, 2 bench QC pools at the appropriate concentration, and 46
unknown samples, including a variable number of blind QC pools. Because
the blind samples were indistinguishable to the analysts from normal
study samples, they were treated as unknowns and were passed through
the preliminary EIA screening procedure and subsequently allocated into
LC/MS/MS assays like all other samples.
Extraction and cleanup.
Normally, 100 samples were prepared
and analyzed each day in 2 assays of 50 samples each. Each run
consisted of a water blank in positions 1 and 50, a bench QC sample in
positions 2 and 49, and a variable number of blind QC samples
interspersed among the unknowns in positions 3–48. All unknown and
blind QC samples carried coded identifications that were not known to
the analysts. Separate bench and blind QC pools with cotinine
concentrations in the appropriate range were selected for use with high
and low assays.
In each run, printed labels were applied to a set of silanized culture tubes (13 x 100 mm), samples were thawed and carefully mixed, and 10 µL of the ISTD and the appropriate sample volume were placed in each tube. Undiluted serum samples (1 mL) were used for the low concentration assays, whereas samples classified as high concentration were assayed as 1:20 dilutions (50 µL of serum + 950 µL of water) or greater, as necessary. The tubes were capped and placed on an Eberbach orbital shaker (Eberbach, Ann Arbor, MI) at low speed for 20–30 min to allow the ISTD to equilibrate with the sample.
We then added 1 mL of 100 g/L trichloroacetic acid to each tube with an automated pipetter. The tubes were recapped, vortex-mixed for 30 s on a VWR MultiTube Vortexer (VWR Scientific), and centrifuged at ~1900g for 20 min. The trichloroacetic acid supernatants were transferred to labeled Oak Ridge-type Teflon tubes (16 x 80 mm), and 0.5 mL of 5 mol/L KOH was added with an Eppendorf Repeater pipet to each sample and mixed briefly. We then added 6 mL of methylene chloride to each tube with the use of an automated pipet, capped the tubes with Teflon screw-caps, and mixed them vigorously for 30 min on the MultiTube Vortexer. After vortex-mixing, the samples were briefly centrifuged again, and the upper (aqueous) layers were carefully removed by a water aspirator.
A set of sodium sulfate columns containing ~1.2 g of high-purity Na2SO4 in 10-mL polypropylene columns (prepared by Varian Sample Preparation Products) were placed in support racks and rinsed with ~4 mL of methylene chloride. The methylene chloride layers from the samples were then passed through the washed columns and collected in clean, labeled silanized glass tubes (13 x 100 mm). The tubes were placed in a Savant Model SC200 SpeedVac (Savant Instruments) equipped with an RT4104 trap and a VN100 VaporNet and taken just to dryness. We added 200 µL of methylene chloride to each residue, swirling gently to ensure that the bottom region of the tube was well-mixed, and transferred the contents to prerinsed, glass-lined autosampler microvials. The vials were placed uncapped in a bench-top hood, and the solvent was allowed to evaporate at room temperature overnight. The next morning, 20 µL of toluene was added to each sample, and the vials were capped, briefly vortex-mixed, and placed in autosampler cartridges for analysis.
Sample analysis.
We injected 10 µL of the sample into the
HPLC, which was operated isocratically at 1 mL/min with a mobile phase
of methanol/ammonium acetate, 2 mmol/L (80:20 by vol). Mass
spectrometric detection was carried out by positive-ion APCI with
corona discharge ionization with the Sciex API III heated nebulizer
interface. The temperature of the nebulizer probe was maintained at
500 °C, and we used multiple reaction monitoring (MRM) with argon as
the collision gas at a nominal thickness of 700 x
1012 molecules/cm1
. We used the m/z
177
80 transition ion of native cotinine, in a ratio to the
corresponding m/z 18080 ion of the ISTD for
quantification; we also monitored the somewhat weaker m/z
17798 ion of native cotinine for confirmation. Dwell times were
normally set at 100 ms. The retention time for cotinine in this system
was <1 min, and the total assay time per sample was ~2 min.
Standards analysis.
The complete set of 12 standards was
evaluated four times each day; twice before and twice after the sample
analyses. Before any samples were measured, the standards set was
assayed twice, from 0 to 25 µg/L and then from 25 to 0 µg/L. The 24
results from this series were then immediately integrated and evaluated
on-line on the Macintosh controller with the use of a custom Pascal
program that checked instrument sensitivity from the mean ISTD area
counts, checked the two blank (0 µg/L) standards to confirm the
absence of contamination, and compared the back-calculated
concentration for each standard with the expected value. The program
printed the results of this preliminary evaluation accompanied by a
suitability recommendation on the basis of preprogrammed acceptable
limits for each parameter. Only after acceptable instrument performance
had been documented in this way were the day's sample analyses begun.
After the analysis of unknowns, the standard set was again assayed
twice in both the forward and reverse directions. Thus, a total of 48
standards were analyzed each day along with the unknown samples.
Data analysis.
Sciex system software (MacQuan) was used to
carry out the initial peak localization and integration for each
sample; then, all sample identification, MRM ion areas, and retention
time data were transferred as ASCII files to SAS data sets (Statistical
Analysis System, Cary, NC) where they were merged electronically with
the sample cleanup and related information that had been entered
earlier into a sample database. All data processing and further
analyses were then conducted off-line with the use of SAS procedures.
The calibration curve for this assay was slightly nonlinear throughout
the concentration range from 0 to 25 µg/L; therefore, all
quantitations were derived as part of a SAS program with the use of a
moving 5-point regression technique as described in Results and
Discussion.
Additional analyses.
For comparisons with gas chromatography
(GC)/low-resolution MS, we used a Hewlett-Packard MSD consisting of an
HP 5890 gas chromatograph interfaced to an HP5970B mass-selective
detector. Analyses were made by splitless injection of 2 µL of the
sample in toluene onto a 0.2 mm x 12 m Ultra-1 column
(Hewlett-Packard). Selected-ion monitoring was used, and quantitation
was based on the area ratios of the native and ISTD molecular ions at
m/z 176 and 179, respectively. High-resolution GC/MS was
carried out by injection of 1 µL of the sample in toluene onto a 0.25
mm x 30 m DB-1701 column (J&W Scientific) mounted in an
HP5890 gas chromatograph, which was interfaced to a VG Autospec mass
spectrometer operated at 10 000 resolution. Two ions were monitored in
these assays at m/z 176.0950 and 179.1138 for native
cotinine and the ISTD, respectively.
|
Results and Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
Calibration curve.
A representative calibration curve is given
in Fig. 2
. This curve is based on 302 replicate analyses of the standard
set (3624 standards). New curves were generated whenever a substantial
change was made to the instrument (e.g., repairs, preventive
maintenance service, or a new state file), and the appropriate
calibration was applied to all samples analyzed within that time
interval. Although not readily apparent in Fig. 2
, these curves are
slightly curvilinear over the three orders of magnitude covered, as
indicated by a Fowlis–Scott test of linearity (39) and by
the overestimation of back-calculated values in the lowest
concentration ranges when a single, linear calibration was used. To
correct for this behavior, we used a moving 5-point regression for
calibration. This approach was favored by the large number of standards
included in each calibration series, which helped to compensate for the
smaller number of concentration points (i.e., 5) used for each
regression curve.
|
For the moving regression analysis, the calibration curve is subdivided into eight overlapping 5-point regions with concentration ranges of 0–0.25, 0.025–0.5, 0.05–1 µg/L, and so forth; a regression analysis is carried out for each region, generating a set of 8 separate regression equations. When unknowns are analyzed, the ion ratio for the unknown is calculated and compared with the mean ratio at the midpoint of each of the 8 regression curves. The curve with a midpoint ratio closest to that of the unknown sample's ratio is then selected for quantitation. The entire processing, selection, and quantitation algorithm is embedded in a SAS program and proceeds automatically.
The inset to Fig. 2
shows the SD for the back-calculated values from
the lower-concentration standards plotted vs the standard
concentration. The limiting SD (S0) calculated
from these data (i.e., the value of S as the native cotinine
concentration approaches 0) was 0.0085. Thus a limit of detection
(LOD), defined as 3 x S0(40), of ~26 ng/L can be estimated from these
data. This value reflects only the instrumental analysis, however, and
does not address other contributions from the entire assay. Therefore,
we calculated the method LOD from the long-term SD of the blank
(41)(42). The mean ± SD of the water
blank was 0.0225 ± 0.016886 µg/L (n = 455), from which we
calculated an estimated method LOD of 50 ng/L.
The high sensitivity (low detection limit) of this method is important, since our primary interest is in monitoring ETS exposure among nonsmokers. Analytes such as pyridine with relatively high gas-phase basicities tend to be measured with very high sensitivities by atmospheric pressure ionization instruments because of the favorable proton transfer kinetics from the hydronium ion reagents (43), and cotinine was found to be ionized and detected very efficiently in this system. Previous methods for measuring cotinine in serum by immunoassays, HPLC, or GC, including GC/MS procedures, have generally been limited to concentrations of a few micrograms per liter or greater. GC with nitrogen–phosphorus detection is potentially capable of detection limits on the order of a few hundred nanograms per liter (10)(11)(12)(13)(14), but the specificity of that GC is much lower than that of MS/MS, and isotopically labeled ISTDs cannot be used. Conversely, the use of deuterated cotinine as an ISTD and the clean ion chromatograms obtained from tandem MS analysis contributed to the low detection limits that are attainable with our method.
The importance of this enhanced sensitivity was evident in the results from Phase 1 of NHANES III, in which the geometric mean serum cotinine concentration among nonsmoking adults was 0.205 µg/L (32). More than one-half of the ~10 500 people examined in Phase 1 had serum cotinine concentrations <1 µg/L (32); thus, a substantial amount of information would have been lost in this study if a less sensitive analytical technique had been used.
Ion chromatograms.
MRM ion chromatograms from a typical serum
sample are given in Fig. 3
. In each case, the response has been normalized by the system
software to 100%. Quantitation in these assays is based on the ratio
of the area counts from Ion A (17780) to that of the ISTD ion (Ion
B, 18080), whereas the confirmation ratio is determined from the
ratio of Ion C (17798) to Ion A. As noted above, these assays
typically produce very clean MRM chromatograms with relatively few
interferences. This reduction in the chemical noise of the assay also
makes a substantial contribution to the sensitivity of the analysis.
The calculated cotinine concentration of the serum sample in Fig. 3
was
0.72 µg/L.
|
Recoveries.
Preliminary evaluations showed good agreement
between recoveries estimated from the ISTD area count for each sample
normalized to the mean ISTD area count for that day's standards vs
recoveries estimated by adding 5-methylcotinine as an additional
external standard. Therefore, we routinely estimated recoveries from
the ISTD area counts. Overall, recovery for the water blanks and for
the (diluted) high cotinine samples estimated in this way averaged
~70% through the entire procedure. Some cotinine is lost during
drying of the extracts, although losses of cotinine from methylene
chloride extracts are much less than those of the more volatile
nicotine, and such losses tend to be minimized when centrifugal vacuum
evaporation is used and care is taken not to overdry the sample.
For the samples with lower cotinine concentrations, from which 1 mL of undiluted serum was taken for analysis, the recoveries were somewhat lower, averaging ~60%. This difference probably reflected additional losses in the protein pellet at the trichloroacetic acid precipitation step. This interpretation is in agreement with preliminary experiments in which serum pools supplemented and equilibrated with [G-3H]cotinine were extracted by this procedure. About 10–15% of the radioactivity was retained in the pellet in those studies, and this residual activity was not appreciably recovered by additional extractions of the pellet. Benowitz et al. (38) have reported that ~2.6% of plasma cotinine is bound to protein. Although the precipitation of serum proteins may have contributed to somewhat lower cotinine recoveries, it completely eliminated problems with emulsions, which was essential to maintaining the high daily throughput of this method. In all cases, native cotinine losses during sample preparation are associated with compensatory losses of the ISTD and thus should not alter the sample quantitation on the basis of the ratio of the two analytes.
Confirmation ratio.
Two native cotinine ion-transition pairs
are monitored in this method. Quantitation is based on the
17780/18080 ratio with the primary progeny ions being used for
both the native and labeled cotinine. In addition, a confirmation ratio
is calculated for each sample with the ratio of the two progeny ions
that are recorded for native cotinine (i.e., 17798/17780). Fig. 4
is a plot of this ratio as a function of (log10)
cotinine concentration for 21 439 standards (1949 analysis sets) at
concentrations ranging from 0.025 to 25 µg/L. These standards were
analyzed over a period of 26 months.
|
Unlike confirmation ratios that are based on elemental isotopic distributions, MRM confirmation ratios are not fundamental and may vary as a result of small differences in the analytical conditions, particularly as they pertain to the operation of Q2 (the collision cell). However, when the analytical conditions remain fixed, this variation is relatively limited, and within-day CVs tend to be reasonably small (~3.5–4%). Therefore, each confirmation ratio for individual samples was evaluated relative to the value calculated from that day's four sets of standards.
Because the m/z 17798 transition ion pair is less
abundant than the 17780 ion pair, the confirmation ratio becomes
unreliable at the lowest cotinine concentration values. As indicated in
Fig. 4
, the ratio remained relatively uniform down to a concentration
of ~250 ng/L, but tended to diverge downward from the expected value
at the lowest concentrations—probably a result of the very low ion
counts for m/z 98 at these low concentrations. Consequently,
the confirmation ratio tends to be less valuable at sample
concentrations <~250 ng/L.
Accuracy evaluations.
The accuracy of this method was
evaluated in part by analyzing a series of aqueous standards prepared
by weight from a pure cotinine perchlorate stock. Standards were
prepared with expected concentrations (corrected for perchlorate)
ranging from 6.2 µg/L to 120 ng/L. Aliquots (1 mL) of these standards
were then taken for analysis in the same manner as unknown serum
samples. As indicated in Table 1
, in each case the calculated concentrations for these samples
agreed well with the expected values. The accuracy of this method was
also evaluated by the analysis of aliquots of NIST Reference Material
8444 (cotinine in freeze-dried human urine). The urine
reference material was used because no serum-based materials were
available. In the RM 8444 set, only the blank sample with a recommended
concentration of 0.8 ± 0.3 ng/g had a target within the low
concentration range of primary interest to us. Our result for the
analysis of this blank sample was 0.898 ± 0.048 ng/g (n =
4). These materials were also analyzed as routine unknowns according to
our regular procedures.
|
Supplemented serum analyses.
Cotinine perchlorate was also
used to supplement a serum sample at four different nominal
concentrations ranging from ~0.5 µg/L to 230 µg/L. The serum
sample used in this case was from a nonsmoker who had no known source
of ETS exposure during the 5 days before donation. After the serum was
supplemented, it was well-mixed and then aliquoted into vials with
coded labels and frozen. On four separate occasions over an interval of
several months, vials were selected for analysis by a chemist with
knowledge of the codes, but with no other involvement in this study,
and analyzed as true unknowns by our method. The results from this
series of analyses are given in Table 2
.
|
Comparisons with GC/MS.
To further evaluate the performance of
our LC/MS/MS method, we carried out split-sample assays with a group of
serum samples by both LC/MS/MS and GC/MS. As indicated in Fig. 5
A, very good agreement was observed throughout a range of
concentrations for a set of serum samples analyzed first by LC/MS/MS
and then reanalyzed by capillary GC/low-resolution MS. Because of the
relatively limited sensitivity of the low-resolution GC/MS analysis,
this comparison was restricted to samples with a serum cotinine
concentration 
10 µg/L. For comparisons at the very low
concentration range of primary interest to us, it was necessary to use
high-resolution GC/MS. The results from split-sample analyses of 28
serum samples by both LC/MS/MS and by high-resolution GC/MS at 10 000
resolving power is given in Fig. 5B
. These data were in very close
agreement down to the detection limit with results obtained by the
LC/MS/MS method and provided further assurance of the accuracy of our
identification of cotinine in the LC/MS/MS assays, even at very low
sample concentrations. Although LC/MS/MS and high-resolution capillary
GC/MS are both mass spectrometric procedures with chromatographic
inlets, the nature and extent of the chromatographic separation is
different in the two techniques, and high-resolution MS is regarded as
the most selective analytical technique available for analyses of this
type. Thus we believe that the close agreement in results obtained by
these two approaches when analyzing common samples provides strong
confirmation for the analytical validity of our LC/MS/MS method.
|
Assay precision.
As noted in Materials and Methods,
both bench and blind QC materials were routinely included in each assay
of 50 samples. Results from the analysis of the first 200 assays of the
low-sample bench QC from the NHANES III, Phase 1 series
(32), with a target value of 1.84 µg/L based on an
initial set of 20 characterization assays, were 1.88 ± 0.122
µg/L (mean ± SD), with a CV of 6.46%. Each sample run also
contained a variable number of blind QC samples that were unknown to
the analysts as QC samples. The target value for this pool on the basis
of the initial set of characterization assays was 0.262 µg/L, and the
result obtained from the NHANES III, Phase 1 series, was 0.268 ±
0.031, with a CV of 11.6% (n = 624). The high assays also
contained both bench and blind QC samples that were inserted into the
assays in a similar manner. The results observed for these materials in
NHANES III, Phase 1, were 207 ± 12.7 µg/L, with a CV of 6.14%
(n = 136) for the bench QC sample and 212 ± 14.8 µg/L,
with a CV of 6.98% (n = 233) for the blind QC pool. Thus the
long-term precision for these analyses was ~6% in each case except
for the very low-concentration blind QC pool, which had a CV of
~12%.
Interferences.
No endogenous substances in serum that could
interfere in these analyses have yet been identified. To interfere, a
substance would need to be extractable from serum by methylene
chloride, be relatively nonvolatile, have the same retention time as
cotinine during HPLC, be ionizable by APCI under our conditions of
analysis, have a molecular mass of 176 ± 1, and fragment in Q2 to
form a product ion at m/z 80. In addition, any significant
interference should be detectable by a shift in the confirmation ratio
for that sample. Both ascorbic acid and serotonin have a molecular mass
of 176 and are present in blood; however, analysis of standards
confirmed that neither of these compounds would interfere in this
analysis. At this time, the only substance known to be potentially
capable of interfering in our assay is the pharmaceutical Pemoline
(CylertTM, Abbott Labs.), a central nervous system
stimulant; it can be resolved from cotinine if necessary by slightly
altering the chromatographic conditions.
Cotinine stability.
Because it is often necessary to store
serum samples for an extended time before the analysis, the stability
of cotinine during storage is an important consideration. Several
investigators have reported that cotinine in serum is quite stable
during storage (10), and our results are in agreement.
Fig. 6
shows the results from the analysis of two serum pools that
were held either at room temperature or at 37 °C for up to 6 weeks.
In all cases, the serum cotinine concentration remained essentially
constant throughout the period of analysis. Fig. 7
, which shows the monthly mean values observed for the
low-concentration bench and blind QC pools over a period of 20 months,
illustrates that both pools maintained a constant concentration during
this period. The same results have been observed with the
high-concentration QC pools. After >4 years, we have seen no evidence
of instability in any of these serum pools during storage at
-60 °C.
|
|
In conclusion, this method for the analysis of cotinine in serum by LC/APCI MS/MS combines the very high sensitivity and analytical specificity of APCI MS/MS in MRM mode with the ruggedness and speed of HPLC. This method has enabled us to undertake the analysis of large numbers of samples in a semiautomated manner while maintaining high sensitivity with a concomitant high degree of confidence in the correct identification of the analyte. Over the past 4 years, >32 000 samples have been analyzed on a single instrument by this method. The sensitivity, accuracy, and ruggedness of this method make it very well-suited to the demands of high-sample-volume epidemiologic investigations of ETS exposure in selected populations.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
A preliminary account of this work was presented at the 46th AACC Annual Meeting, New Orleans, LA, July 1994.
Use of trade names is for identification only and does not constitute endorsement by the Public Health Service, or by the US Department of Health and Human Services.
1 Nonstandard abbreviations: ETS, environmental tobacco smoke; NHANES, National Health and Nutrition Examination Survey; APCI MS/MS, atmospheric pressure chemical ionization tandem mass spectrometry; MRM, multiple reaction monitoring; GC, gas chromatography; LOD, limit of detection; EIA, enzyme-linked immunoassay; ISTD, internal standard. 
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResults and DiscussionReferences |
|---|
-(3-pyridyl)--oxo-N-methylbutyramide. Methods Enzymol 1982;84:628-640.
[ISI][Medline]
[Order article via Infotrieve]
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  R. C. Gwynn, R. K. Garg, B. D. Kerker, T. R. Frieden, and L. E. Thorpe Contributions of a Local Health Examination Survey to the Surveillance of Chronic and Infectious Diseases in New York City Am J Public Health, January 1, 2009; 99(1): 152 - 159. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. L. Benowitz, J. T. Bernert, R. S. Caraballo, D. B. Holiday, and J. Wang Optimal Serum Cotinine Levels for Distinguishing Cigarette Smokers and Nonsmokers Within Different Racial/Ethnic Groups in the United States Between 1999 and 2004 Am. J. Epidemiol., November 19, 2008; (2008) kwn301v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. E. Matt, J. T. Bernert, and M. F. Hovell Measuring Secondhand Smoke Exposure in Children: An Ecological Measurement Approach J. Pediatr. Psychol., March 1, 2008; 33(2): 156 - 175. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. W. Vesper, J. T. Bernert, M. Ospina, T. Meyers, L. Ingham, A. Smith, and G. L. Myers Assessment of the Relation between Biomarkers for Smoking and Biomarkers for Acrylamide Exposure in Humans Cancer Epidemiol. Biomarkers Prev., November 1, 2007; 16(11): 2471 - 2478. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. E. Wilson, R. S. Kahn, J. Khoury, and B. P. Lanphear The Role of Air Nicotine in Explaining Racial Differences in Cotinine Among Tobacco-Exposed Children Chest, March 1, 2007; 131(3): 856 - 862. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. O'Connor, G. A. Giovino, L. T. Kozlowski, S. Shiffman, A. Hyland, J. T. Bernert, R. S. Caraballo, and K. M. Cummings Changes in Nicotine Intake and Cigarette Use Over Time in Two Nationally Representative Cross-Sectional Samples of Smokers Am. J. Epidemiol., October 15, 2006; 164(8): 750 - 759. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Culea, O. Cozar, S. Nicoara, and R. Podea Exposure Assessment of Nicotine and Cotinine by GC-MS Indoor and Built Environment, June 1, 2005; 14(3-4): 293 - 299. [Abstract] [PDF]
|
|
|
|
|
|
W. Q. Gan, S.F. P. Man, and D. D. Sin The Interactions Between Cigarette Smoking and Reduced Lung Function on Systemic Inflammation Chest, February 1, 2005; 127(2): 558 - 564. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 X. Xu, M. M. Iba, and C. P. Weisel Simultaneous and Sensitive Measurement of Anabasine, Nicotine, and Nicotine Metabolites in Human Urine by Liquid Chromatography-Tandem Mass Spectrometry Clin. Chem., December 1, 2004; 50(12): 2323 - 2330. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. D. Kellogg, J. Behaderovic, O. Bhalala, and N. Rifai Rapid and Simple Tandem Mass Spectrometry Method for Determination of Serum Cotinine Concentration Clin. Chem., November 1, 2004; 50(11): 2157 - 2159. [Full Text] [PDF]
|
|
|
|
 |
|
 D. M. Nondahl, K. J. Cruickshanks, D. S. Dalton, C. R. Schubert, B. E. K. Klein, R. Klein, and T. S. Tweed Serum Cotinine Level and Incident Hearing Loss: A Case-Control Study Arch Otolaryngol Head Neck Surg, November 1, 2004; 130(11): 1260 - 1264. [Abstract] [Full Text] |
