development of a liquid chromatographic time-of-flight mass spectrometric method for the...
TRANSCRIPT
RAPID COMMUNICATIONS IN MASS SPECTROMETRY
Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/rcm.1263
Development of a liquid chromatographic time-of-flight
mass spectrometric method for the determination of
unlabelled and deuterium-labelled a-tocopherol in blood
components
Wendy L. Hall1, Yvonne M. Jeanes1, Jonathan Pugh2 and John K. Lodge1*1Centre for Nutrition and Food Safety, School of Biomedical and Molecular Sciences, University of Surrey, Guildford GU2 7XH, UK2Micromass UK Ltd., Atlas Park, Simonsway, Manchester M22 5PP, UK
Received 6 August 2003; Revised 10 October 2003; Accepted 10 October 2003
A method is described for the analysis of deuterated and undeuterated a-tocopherol in blood com-
ponents using liquid chromatography coupled to an orthogonal acceleration time-of-flight (TOF)
mass spectrometer. Optimal ionisation conditions for undeuterated (d0) and tri- and hexadeuter-
ated (d3 or d6) a-tocopherol standards were found with negative ion mode electrospray ionisation.
Each species produced an isotopically resolved single ion of exact mass. Calibration curves of pure
standards were linear in the range tested (0–1.5mM, 0–15pmol injected). For quantification of d0
and d6 in blood components following a standard solvent extraction, a stable-isotope-labelled
internal standard (d3-a-tocopherol) was employed. To counter matrix ion suppression effects, stan-
dard response curves were generated following identical solvent extraction procedures to those of
the samples. Within-day and between-day precision were determined for quantification of d0- and
d6-labelled a-tocopherol in each blood component and both averaged 3–10%. Accuracy was
assessed by comparison with a standard high-performance liquid chromatography (HPLC) method,
achieving good correlation (r2¼ 0.94), and by spiking with known concentrations of a-tocopherol
(98% accuracy). Limits of detection and quantification were determined to be 5 and 50 fmol injected,
respectively. The assay was used to measure the appearance and disappearance of deuterium-
labelled a-tocopherol in human blood components following deuterium-labelled (d6) RRR-a-toco-
pheryl acetate ingestion. The new LC/TOFMS method was found to be sensitive, required small
sample volumes, was reproducible and robust, and was capable of high throughput when large
numbers of samples were generated. Copyright # 2003 John Wiley & Sons, Ltd.
Vitamin E is a term used to describe a number of tocopherols
(a-, b-, d-, and g-) and tocotrienols (a-,b-, d-, and g-) of which a-
tocopherol is the most biologically active comprising over
90% of vitamin E in the body.1,2 The most widely researched
function of a-tocopherol is that of the major lipophilic chain-
breaking antioxidant; however, recent evidence suggests
a-tocopherol to have further roles in cellular signalling which
are independent of its antioxidant action.3 These actions have
led to clinical trials involving vitamin E in coronary disease
prevention, some of which have shown vitamin E to have a
beneficial effect.3,4 Thus it is important to gain a greater
understanding of the absorption, transport and distribution
of vitamin E.
Deuterium-labelled a-tocopherol can be a useful tool with
which to measure the rate of absorption, transport and
uptake in various blood components and excretion as urinary
metabolites. Indeed, the use of deuterated tocopherols has
greatly increased the knowledge of vitamin E transport.2,5 For
example, such studies confirmed that vitamin E isoforms are
absorbed indiscriminately in the intestine but that the
naturally occurring RRR form of a-tocopherol is retained on
the plasma,6,7 and plasma recycling of a-tocopherol is in
itself rapid.8 As well as absorption and transport, deu-
terated tocopherols can be used for biokinetic analyses of
a-tocopherol which relate to its antioxidant function; for
example, using cigarette smokers as models of in vivo
oxidative stress.9,10
The use of stable-isotope-labelled compounds requires
mass spectrometry as a quantitation tool. Previous methods
for the analysis of deuterated vitamin E have required gas
chromatography/mass spectrometry (GC/MS).2,5,8–11 How-
ever, the disadvantage of GC/MS is the need for sample
clean-up and prior derivatisation of the samples which can
greatly increase experimental error, especially as the
trimethylsilyl derivatives are unstable and rapidly hydrolyse
Copyright # 2003 John Wiley & Sons, Ltd.
*Correspondence to: J. K. Lodge, School of Biomedical and Mole-cular Sciences, University of Surrey, Guildford GU2 7XH, UK.E-mail: [email protected]/grant sponsors: British Heart Foundation; The RoyalSociety.
in the presence of moisture. Recently, a number of LC/MS
methods for the analysis of tocopherols have been estab-
lished.12–14 These methods involve either single14 or triple
quadrupole12,13 instruments. Until now, no time-of-flight
mass spectrometry (TOFMS) methods for vitamin E analysis,
or indeed analysis of biomolecules, have been available. TOF
mass spectrometers offer increased sensitivity for the gen-
eration of full mass range spectra compared with scanning
instruments such as quadrupoles. This offers the advantage
of allowing quantitation to be performed on any mass
observed in the acquired range along with simultaneous
exact mass confirmation of the analyte elemental composi-
tion. Additionally, the higher resolution associated with TOF
spectra confers greater specificity of analyte ion selection in
the presence of potential nominally isobaric matrix inter-
ferences. The aim of this work was to develop a rapid,
quantitative method for the determination of deuterium-
labelled and unlabelled a-tocopherol in blood components
using liquid chromatography (LC) TOFMS.
MATERIALS AND METHODS
TocopherolsRRR-[5,7-Methyl-(2H3)2]-a-tocopheryl acetate (d6) and all
racemic-[5-methyl-2H3]-a-tocopheryl acetate (d3) were kind
gifts from Cognis Nutrition and Health. Purities of the acetates
were 98.8% for both species. Isotopic purity was determined to
be >99.9%. d6-RRR-a-Tocopherol acetate was encapsulated
(150 mg) for human consumption. RRR-a-Tocopherol (d0;
purity� 99.0%) was purchased from Fluka (Poole, UK).
Reagents and solventsLC/MS grade methanol (LC/MS Chromasolv), butylated
hydroxytoluene (BHT), ascorbic acid and lithium perchlorate
were purchased from Sigma-Aldrich Chemical Co. (Poole,
UK). Sodium dodecyl sulphate (SDS) and potassium hydro-
xide (KOH) was from BDH, hexane (HPLC grade) was from
Fisher (Loughborough, UK), absolute ethanol from Hayman
Ltd. (Witham, UK), and phosphate-buffered saline (PBS)
tablets from Oxoid Ltd. (Basingstoke, UK).
Samples for analysisHuman plasma, erythrocytes, platelets and lymphocytes
were isolated from blood donated by healthy, normolipidae-
mic male and female volunteers, aged 23–54 years, at various
times following supplementation with a capsule containing
150 mg d6-RRR-a-tocopheryl acetate taken with a standard
breakfast. These procedures were approved by the Univer-
sity of Surrey Advisory Committee on Ethics.
Platelets were isolated from whole blood by first centrifu-
ging whole blood at 280 g for 14 min at 28C to obtain platelet-
rich plasma (PRP). The PRP was then centrifuged at 1120 g for
15 min at 28C to pellet the platelets. After removal of the
platelet-poor plasma the platelets were washed with Tris
buffer (15 mM Tris, 134 mM NaCl, 5 mM KCl, pH 7.4 with
HCl) before reconstitution in 1 mL of the Tris buffer. Samples
were then aliquoted and snap-frozen in liquid nitrogen and
stored at �808C prior to analysis.
Lymphocytes were isolated from whole blood using
Histopaque-1077 (Sigma-Aldrich) according to the manufac-
turer’s protocol. After washing the lymphocytes three times
with PBS, pH 7.4, the lymphocytes were reconstituted into 0.5
mL PBS, aliquoted, and snap-frozen in liquid nitrogen and
stored at �808C prior to analysis.
Erythrocytes were carefully removed from whole blood
after centrifugation at 1550 g for 10 min and washed three
times with saline. Haematocrit was then measured and the
cells aliquoted, snap-frozen in liquid nitrogen, and stored at
�808C prior to analysis.
Plasma was collected following platelet removal (platelet-
poor plasma). Plasma was then aliquoted, snap-frozen in
liquid nitrogen, and stored at �808C prior to analysis.
Sample preparation and vitamin E extractionTo ensure that signals fell into the dynamic range of the instru-
ment, and to reduce matrix ion suppression effects (discussed
later), aliquots of blood components (100mL) were diluted
with PBS prior to extraction, and the final dried extracts
were reconstituted into a relatively large volume of ethanol
(500mL). It was found that plasma and erythrocyte aliquots
required diluting 1:5 and 1:10 with PBS, respectively, while
platelet and lymphocyte aliquots did not require dilution.
Total vitamin E was extracted from the blood components
by a combination of SDS, ethanol and hexane, in the presence
of BHT (10mL of a 1 mg/mL solution in ethanol), as described
previously.15 For LC/TOFMS analysis an internal standard
(d3-labelled a-tocopherol, 50 pmol) was added to the aliquot
prior to extraction. Lymphocyte and platelet aliquots were
sonicated for 3 min prior to extraction to obtain a homo-
genous solution. In separate experiments plasma samples
underwent saponification to extract vitamin E. The method
used was that of Podda et al.16 using KOH and ascorbic acid.
Following extraction and hexane evaporation, extracts
were reconstituted into 500 mL of ethanol ready for
LC/TOFMS analysis. All samples were prepared and
analysed in duplicate.
Standard preparationStandard stock solutions of d0-, d3- and d6-a-tocopherol were
prepared in ethanol. The hexadeuterated RRR-[5,7-methyl-
(2H3)2]-a-tocopheryl acetate (d6) and trideuterated all race-
mic-[5-methyl-2H3]-a-tocopheryl acetate (d3) were first
hydrolysed using KOH as described.16 The concentrations
of the stock solutions were then determined using the molar
extinction coefficient (e292¼ 327016) and diluted accordingly.
For quantitation, standards underwent the same extraction
procedure as that of the blood components (described above).
Aliquots of d0 and d6 were added to PBS (100mL) containing
BHT (10 mL of 1 mg/mL solution in ethanol). A range of d0
and d6 concentrations (0–400 pmol) with a constant concen-
tration of internal standard (d3, 50 pmol) was used.
Following extraction and evaporation, extracted standards
were reconstituted in 500mL ethanol. Final concentrations of
standard curves therefore typically ranged from 0–2 mM (for
the analysis of plasma and erythrocytes) or 0–0.8 mM (for the
analysis of platelets and lymphocytes), containing 0.1 mM
internal standard, and consisted of at least five calibration
points. Standards were also analysed in duplicate.
Liquid chromatographyThe HPLC system used was a Waters Alliance system
(Waters Corporation, Milford, MA, USA). The Waters 2695
Copyright # 2003 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803
2798 W. L. Hall et al.
separations module comprises a solvent delivery system,
online degasser, peltier-cooled autosampler (set at 48C), con-
troller, and column oven (set at 308C). The mobile phase con-
sisted of 100% methanol (LC/MS grade) with a total run time
of 3.5 min. The tocopherols were separated on a reverse-
phase Waters Symmetry1 column (2.1� 50 mm, C18,
3.5 mm particle size) at a flow rate of 0.3 mL/min. a-Tocopher-
ols eluted at 2.2 min.
Mass spectrometrySamples were analysed using a Micromass LCTTM (Micro-
mass UK Ltd., Manchester, UK) orthogonal acceleration
TOF mass spectrometer, with negative ion mode electro-
spray ionisation (ESI). The LCT consists of an orthogonal
ion extraction (Z-Spray) source equipped with an electro-
spray inlet. The ions are transferred via a dual hexapole to
a pusher region where the ions are deflected into the analy-
ser flight tube at a rate of 20 kHz (50 ms flight time). Ion arri-
val times are recorded using a multichannel plate detector
(MCP) linked to a 3.6 GHz time-to-digital converter
(TDC). The individual 50 ms digitised spectra are summed
over a defined interval (1 s) to produce the full mass range
spectra. To optimise conditions the instrument was tuned
on a 10 mM solution of a-tocopherol in ethanol continuously
infused into the ion source. Optimum ionisation parameters
were found to be: negative ion electrospray ionisation
(ESI–); desolvation gas temperature, 2508C; source tem-
perature, 1208C; capillary voltage, 2800 V; sample cone vol-
tage, �35 V; extraction cone voltage 3 V; RF lens, 350.
Typically ions in the m/z range 350–500 Da were collected
from 1.5–3.5 min following injection, and acquired at 4000
FWHM resolution. Spectra were acquired at a rate of 1 spec-
trum/s (0.95 s average, 0.05 s delay). The target masses of
the ions of interest were used to identify each tocopherol
species by producing an extracted mass chromatogram
from the total ion chromatogram (TIC) over a 0.1 Da mass
window, thereby utilising the specificity gained by acquir-
ing spectra at high resolution. The identified peaks were
then integrated and amounts calculated from response
curves generated from a range of standards using the stan-
dard quantitation processing function with Mass Lynx ver-
sion 4.0 software. Ions generated by deuterium-labelled and
unlabelled a-tocopherol were: m/z 429.3742 for d0-RRR-a-
tocopherol, m/z 432.3973 for d3-all rac-a-tocopherol, and
m/z 435.4116 for d6-RRR-a-tocopherol.
HPLC/electrochemical detection (ECD)The same LC system to that described above was used but
incorporating a Bioanalytical Systems (BAS) electrochemical
detector with a glassy carbon electrode operating at an
applied potential of þ0.5 V. Tocopherols were separated on
a Waters Spherisorb ODS-2 column (4.6� 250 mm, 5mm par-
ticle size) using a mobile phase consisting of 99% methanol,
1% water and 0.1% lithium perchlorate at a flow rate of
1.5 mL/min. Tocopherols were quantitated using an internal
standard (d-tocopherol, 0.5 mM final concentration) with
responses calculated from a range of pure a-, g- and d-
tocopherol standards in ethanol. Data were analysed using
Waters MilleniumR32 software.
StatisticsData are expressed as means� standard deviations. Within-
day and between-day precision are shown as coefficients of
variance (standard deviation/mean� 100¼CV%). In addi-
tion to using the Pearson correlation coefficient to assess the
association between HPLC/ECD and TOFMS, Bland-Altman
plots17,18 were made to compare the difference between
methods. Bland-Altman tests are sensitive to detecting bias
between two methods estimating the same variable.17,18 Since
there was a large range in concentrations tested (plasma
and platelet unlabelled a-tocopherol), the between-method
variation was larger as concentrations increased. Therefore,
the percentage difference ((HPLC/ECD�TOFMS/(HPLC/
ECDþTOFMS/2))� 100) was used rather than the differ-
ence between methods.
RESULTS AND DISCUSSION
Method developmentAs this is the first method to employ LC/TOFMS to analyse
tocopherols, the appropriate method of ionisation was first
established. Ionisation was tested on both positive and nega-
tive ion mode electrospray (ESI) and atmospheric pressure
chemical ionisation (APCI) by tuning on a 10mM solution of
d0-RRR-a-tocopherol constantly infused into the spectro-
meter and observing the TIC following a 1-s acquisition.
Negative ion ESI was found to give the optimum response
(data not shown); hence this was used in subsequent method
development.
Typical TICs and spectra over the mass range of interest for
the tocopherol species following an injection of 0.5 pmol of
either d0, d3 or d6 standard are shown in Fig. 1. TOF ESI–
produces an isotopically resolved single ion of exact mass
(Fig. 1(B)). This mass for d0 (429.3742) is nominally isobaric
with the [M–H]� calculated mass of a-tocopherol (429.3733).
Obtaining exact mass greatly increases the confidence in the
correct assignment of the compound of interest. For example,
a nominal mass of 429.3 has the possibility of thousands of
possible empirical formulae, while a mass within�5 ppm can
only result from a very limited number of elemental
compositions. In this case the closest match corresponds to
C29H49O2, which is the formula for a-tocopherol. The clear
and monoisotopically resolved spectrum obtained by ESI-
TOF contrasts with the spectrum obtained in previous
methods employing either single14 or triple12,13 quadrupole
mass spectrometers. In these methods, APCIþ MS/MS,12
APCIþ MS,14 and ESþ MS/MS13 produced a cluster of ions
for a-tocopherol in them/z range 423–432, the expected target
protonated ion not being the predominant species. The
presence of these additional ions has been attributed both to
the oxidation of a-tocopherol, and to the protonation
followed by dehydrogenation of a-tocopherol, during ionisa-
tion12 and represents novel ionisation characteristics. How-
ever, this appears to occur only in quadrupole spectrometers
since only the expected target ions were observed during all
ionisation methods with TOFMS.
The principle of quantitation from TOF full ‘scan’ spectra
by integrating extracted mass chromatograms over a narrow
(0.1 Da) window allows greater specificity than typically
observed in selected ion recording (SIR) experiments
LC/TOFMS analysis of tocopherols in blood components 2799
Copyright # 2003 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803
employed on single quadrupole mass spectrometers,
although the overall sensitivity and linear dynamic range
are generally lower. The optimum specificity and selectivity
and linear dynamic range are usually obtained from multiple
reaction monitoring experiments performed on triple quad-
rupole instruments where quantitation is performed on the
most abundant fragment ions generated from the selected
precursor ion. However, in this example where the target
concentrations are relatively high and the dynamic range less
than three orders of magnitude, the benefits of exact mass and
high-resolution full spectrum acquisition conferred by TOF
data outweigh the limitations. With a-tocopherol the main
fragment observed during quadrupole MS was found to be at
m/z 165,12–14 which corresponds to the loss of the phytyl
chain. TOFMS does not produce product ions if the cone
voltage is kept relatively low. However, in-source fragmen-
tation can be induced by increasing the sampling cone
voltage. This was assessed in our experiments. a-Tocopherol
fragmentation occurred when the cone voltage was increased
to �70 mV. Among others, a fragment at m/z 163.3 was
observed (data not shown) which corresponds to an ion 1 Da
lower in mass than the major fragment ion observed in
previous methods.12–14
Typical peak area versus concentration curves for d0-, d3-
and d6-a-tocopherol standards demonstrated a linear
response in the range 0–1.5 mM (0–15 pmol injected). Con-
centrations in excess of 5 mM a-tocopherol (150 pmol injected)
resulted in non-linearity. TOF instruments generally have a
relatively low dynamic range, which is offset by the increased
sensitivity of these machines, so increasing concentrations of
analyte will eventually saturate the detector and produce a
non-linear response.
Quantification of a-tocopherolAn internal standard, d3-labelled all racemic a-tocopherol
(50 pmol added, 0.1 mM final concentration in extract),
was used for the quantification of unlabelled (d0) and d6-
labelled a-tocopherol in biological samples. The d3-labelled
a-tocopherol was added to the blood components prior to
extraction thereby accounting for any losses in the extraction
procedure.
First, we established the concentration range of tocopher-
ols that did not exceed the dynamic range of the detector. By
increasing concentrations of either d0 or d6 (0–10mM) and
maintaining a constant concentration of internal standard d3
(0.1mM), we observed saturation of the detector at concentra-
tions of d0 and d6 in excess of 5mM (as described above),
which also resulted in suppression of the internal standard
signal. Thus a maximum concentration of 4mM was deemed
appropriate. Therefore, biological samples, such as plasma,
had to be diluted prior to extraction, or reconstituted in larger
volumes, in order not to exceed the dynamic range of the
instrument. This has the added benefit of requiring very low
volumes of sample for quantitation. Indeed<10mL of plasma
can be used (standard HPLC methods generally use 100 mL).
Therefore, sample dilution into the dynamic range of the
instrument is an important and necessary prerequisite for
quantitation with TOFMS.
During initial spiking and recovery experiments with
increasing concentrations of d6-labelled a-tocopherol into
blood components, we noticed that quantitating using
calibration curves prepared in ethanol resulted in calculated
amounts much smaller than that added. This was first
attributed to matrix ion suppression effects due to the
biological matrix as previously noted.13 However, further
Figure 1. Typical LC/TOFMS extracted mass chromatograms (A) and mass spectra (B) of a-tocopherolsused in the method. The structures of the species d0 (RRR-a-tocopherol, unlabelled), d3 (all rac-a-5(CD3)-tocopherol) and d6 (RRR-a-5, 7-(CD3)2-tocopherol) are also shown. Tocopherols elute at 2.2 min.
The mass spectra shown in (B) demonstrate the presence of a single main species with accurate mass.
2800 W. L. Hall et al.
Copyright # 2003 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803
experiments revealed this to be mainly due to matrix effects
arising from the extraction procedure itself. To counter this,
standards underwent identical extraction to those of the
biological samples. Thus, in order to quantitate d0- and d6-
labelled a-tocopherol in samples, calibration curves were
generated using increasing concentrations of d0 and d6 and a
constant concentration of internal standard, then treated in
the same way as samples (solvent extraction, as described in
the previous section). The same concentration of internal
standard was added to unknown samples. A response curve
is generated to determine amounts from the relationship:
Response ¼ Areaunknown
� ðInternal standardconcentration= Internal standardareaÞ
(where unknown is either d0 or d6 in the standard or
unknown sample).
A typical response curve for d0- and d6-a-tocopherol is
shown in Fig. 2. The curve is non-linear and fits a polynomial
(coefficient of determination >0.995).
Typical TICs and extracted mass chromatograms of the
tocopherol species in representative plasma and platelet
extracts taken from an individual 3 h following ingestion of
d6-labelledRRR-a-tocopheryl acetate are shown in Fig. 3. The
ion pattern of the peak eluting at 2.2 min (not shown) showed
the presence of the three tocopherol species (d0, d3 and d6)
and no other ions in the range, comparable with mass spectra
of pure standards (Fig. 1).
It has been suggested previously that saponification
of biological samples may be required in the analysis of
a-tocopherol using LC/MS without multiple quadrupole
instruments, in order to remove potential interfering species
Figure 2. Representative area response curve used for the
determination of d0- and d6-labelled a-tocopherol in blood
components using d3-labelled a-tocopherol as an internal
standard. This concentration range was used for determina-
tion of d0 and d6 in plasma. The coefficient of determination
for each species was >0.995. The inset shows an expanded
region from 0–0.3mMwith response for both d0 and d6 on the
y-axis.
Figure 3. Typical LC/TOFMS mass chromatograms of vitamin E extracts from plasma and
platelets taken 3 h following supplementation of an individual with 150mg d6-labelled RRR-a-tocopheryl acetate. The total ion chromatograms (TICs) demonstrate the elution of the tocopherol
peak at 2.2min. Extracted mass chromatograms of the tocopherol species d0, d3 and d6 are shown
over a mass window of 0.1Da.
LC/TOFMS analysis of tocopherols in blood components 2801
Copyright # 2003 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803
such as phospholipids and their oxidation products with m/z
similar to that of a-tocopherol.12 To investigate any potential
interfering compounds with the TOFMS method, a number
of plasma samples (n¼ 8) underwent extraction by saponi-
fication16 and this analysis was compared with those
extracted using the solvent (ethanol and hexane) method
(n¼ 8) performed in parallel. The spectrum of the tocopherol
peak eluting at 2.2 min was similar for both methods (data not
shown) and the d0 and d6 concentrations determined were
also similar (solvent: [d0] mM 20.8� 1.5, [d6] mM 8.0� 4; cf.
saponification: [d0] mM 21.3� 2, [d6] mM 8.3� 0.5). These
observations, coupled with the exact mass measurement of
a-tocopherol (which substantially increases confidence as
described above), suggest that both methods of vitamin E
extraction can be used in the analysis of tocopherols using
TOFMS.
Limits of quantification and detection (LOQ andLOD)LOD and LOQ were calculated using pure standards and
blood components spiked with varying concentrations of
d6-a-tocopherol prior to extraction. Values for LOQ and
LOD are presented in Table 1. The LOD was determined
from the amount injected which produced a peak with a sig-
nal/noise ratio of <5:1. This was found to be in the region of
5 fmol injected for standards, but 2-3-fold higher in blood
components, presumably because of either matrix effects or
losses during the extraction process. The LOQ was deter-
mined from the lowest point in the standard response curve
in which a reliable amount can be obtained. The values
between 50 and 100 fmol are approximately 3-fold higher
than the LOD for each blood component. The LOQ and
LOD are similar for plasma and platelets, but were higher
in erythrocytes presumably due to matrix ion suppression.
It should be noted though that these values for LOD and
LOQ using TOFMS represent an overestimate since: (i) they
were performed on diluted samples and (ii) the MCP gain
and TDC threshold can be altered to vary sensitivity if
required. Thus the LOQ is potentially lower. However, in
our experiments, these conditions were sufficient to detect
d6-labelled a-tocopherol in all blood components at various
times following 150 mg d6-RRR-a-tocopheryl acetate supple-
mentation in humans.
Performance of the TOFMS methodPerformance of the new LC/TOFMS method was evaluated
by accuracy and precision. To assess quantitative accuracy
the TOFMS method was compared with a HPLC/ECD meth-
od.16 Although HPLC/ECD is an established method it can-
not be considered a gold standard. However, comparison
of these two methods would indicate the accuracy of
LC/TOFMS to some extent. Non-deuterated samples
were obtained from volunteers taking part in a separate a-
tocopherol supplementation study. Plasma and platelet sam-
ples (n¼ 13) were used so that a wide range of a-tocopherol
values could be compared and from different matrices. d3-a-
Tocopherol was used as an internal standard for the TOFMS
method, while d-tocopherol was used as an internal standard
for the HPLC/ECD method. The use of different internal
standards required separate extraction for the two methods,
which may have added another source of error to the varia-
tion resulting from the method of detection. The LC/TOFMS
method correlated closely with HPLC/ECD (r2¼ 0.939). The
% differences between HPLC/ECD and TOFMS determined
with Bland-Altman plots17,18 showed that 9 out of 13 data
points were scattered randomly between the upper and
lower 20% range (data not shown). The upper and lower lim-
its of agreement (�2 SD) were 41% different due to two out-
lying data points. The mean percentage difference between
methods was �6%.
Plasma was also spiked with increasing amounts of d6-
labelled a-tocopherol (20–300 pmol added, 0.2–3 mM final
Table 1. Limit of detection (LOD) and limit of quantification
(LOQ) of a-tocopherol in blood components analysed by LC/
TOFMS
Standard Plasma RBCs Platelets
LODa 5 15 30 15S/Nb 5:1 3:1 3:1 3:1LOQa,c 50 100 50
a LOD and LOQ values in fmol injected.b Signal-to-noise ratio at LOD.c LOQ values extrapolated from lowest point in standard curve.
Table 2. Precision of LC/TOFMS method in the
determination of unlabelled (d0) and deuterium-labelled
(d6) a-tocopherol. Values shown are coefficients of
variation (CV) following a number of determinations either
within- or between-day
Variation(CV %) Iona Curveb Standardc Plasmad rbcsd Plateletsd
Within-daye d0 0.4 5 3 4 4d6 0.4 6 3 7 4
Between-dayf d0 0.4 4 7 5 7d6 0.4 4 10 6 3
a Variation in analysis of either d0 or d6, with an approximately 10-fold difference in concentration between d0 and d6.b Variation in the coefficient of determination of the standardresponse curve used for determination of either d0 or d6 amounts.Two concentrations of internal standard (d3) were tested (final con-centration 0.1 or 0.5 mM) and gave similar results.c Based on peak areas of pure standards in ethanol (0.5 pmol injec-tions).d Based on quantification of either d0 or d6 using d3 as internal stan-dard as described.e Based on n¼ 8 determinations of the same sample during the courseof a day (except curve; n¼ 3).f Based on n¼ 8 determinations of the same sample over the course of4 weeks.
Table 3. Concentrations of unlabelled (d0) a-tocopherol inblood components determined by LC/TOFMS
Blood componenta-Tocopherol determined
by LC/TOFMS (n¼ 6)
Plasma (mmol/L) 27.1� 8.6Erythrocytes (mmol/mL packed cellhaematocrit)
5.2� 0.4
Platelets (mmol/g protein) 2.0� 0.6Lymphocytes (mmol/g protein) 1.7� 0.8
2802 W. L. Hall et al.
Copyright # 2003 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803
concentration) and, following extraction, the d6 concentra-
tion was determined by the LC/TOFMS method. The
accuracy of this quantitation was found to be 98� 8%
(n¼ 12).
To assess precision, plasma, erythrocyte and platelet
samples were extracted and analysed either on a number of
occasions during a single day, or a number of times on
separate days over a 4-week period. Table 2 shows the
precision of the LC/TOFMS method based on the coefficient
of variation from within-day and between-day quantitation.
Good precision was found with CVs ranging from 3–10%.
Higher CVs for between-day variation were observed for
erythrocytes, and in general CVs were higher for d6 than d0,
presumably because of the smaller amounts.
Application of the LC/TOFMS method tobiological samplesIn order to apply the LC/TOFMS method to biological
samples, d0- and d6-labelled a-tocopherol were analysed in
blood components taken from a range of individuals various
times following ingestion of a capsule containing 150 mg
d6-labelled RRR-a-tocopheryl acetate. Table 3 shows
baseline concentrations of d0-a-tocopherol in plasma, ery-
throcytes, platelets and lymphocytes in a range of individuals
(n¼ 6), determined by LC/TOFMS. Values are similar to
published values for a-tocopherol in blood components.19,20
Profiles of d6-labelled a-tocopherol in blood components in a
single individual are shown in Fig. 4. Following ingestion of
d6-labelled RRR-a-tocopheryl acetate, there is a rapid ap-
pearance of d6-a-tocopherol in the plasma (peak at 9 h) and
erythrocytes (peak at 24 h), followed by a decrease in d6 con-
centration. In contrast, platelet and lymphocyte d6-a-
tocopherol increases gradually over the time period. Similar
a-tocopherol turnover curves in plasma8,11 and erythro-
cytes21 following deuteratedRRR-a-tocopheryl acetate inges-
tion have been observed. No such data exists for platelets and
lymphocytes.
CONCLUSIONS
This study demonstrates that LC/TOFMS can be applied in
the analysis of labelled and unlabelled tocopherols in blood
components. The method is rapid, sensitive, accurate and
reproducible.
AcknowledgementsWe are grateful to Dr. Christine Gartner and Dr. James Clark
of Cognis Nutrition and Health (Dusseldorf, Germany and
LaGrange, IL, USA, respectively) for the gift of the deuterated
(d3 and d6) tocopheryl acetates, and to the Pharmacy Depart-
ment of St. Thomas’ Hospital for encapsulation. We are grate-
ful to the British Heart Foundation and The Royal Society
for funding and to the Medical Research Council for sup-
porting YMJ.
REFERENCES
1. Brigelius-Flohe R, Traber MG. FASEB J. 1999; 13: 1145.2. Burton GW, Traber MG. Annu. Rev. Nutr. 1990; 10: 357.3. Brigelius-Flohe R, Kelly FJ, Salonen JT, Neuzil J, Zingg JM,
Azzi A. Am. J. Clin. Nutr. 2002; 76: 703.4. Pryor WA. Free Rad. Biol. Med. 2000; 28: 141.5. Burton GW, Ingold KU, Cheeseman KH, Slater TF. Free Rad.
Res. Comm. 1990; 11: 99.6. Traber MG, Kayden HJ. Am. J. Clin. Nutr. 1989; 49: 517.7. Traber MG, Rader D, Acuff R, Brewer HB, Kayden HJ.
Atherosclerosis 1994; 108: 27.8. Traber MG, Ramakrishnan R, Kayden KJ. Proc. Natl. Acad.
Sci. USA 1994; 91: 10005.9. Traber MG, Winklhofer-Roob BM, Roob JM, Khoschsorur G,
Aigner R, Crocc C, Ramakrishnan R, Brigelius-Flohe R. FreeRad. Biol. Med. 2001; 31: 1368.
10. Munro LH, Burton G, Kelly FJ. Clin. Sci. 1997; 92: 87.11. Traber MG, Rader D, Acuff RV, Ramakrishnan R,
Brewer HB, Kayden HJ. Am. J. Clin. Nutr. 1998; 68: 847.12. Lauridsen C, Leonard SW, Griffin DA, Liebler DC,
McClure TD, Traber MG. Anal. Biochem. 2001; 289:89.
13. Mottier P, Gremaud E, Guy PA, Turesky RJ. Anal. Biochem.2002; 301: 128.
14. Kalman A, Mujahid C, Mottier P, Heudi O. Rapid Commun.Mass Spectrom. 2003; 17: 723.
15. Burton GW, Webb A, Ingold KU. Lipids 1985; 20: 29.16. Podda M, Weber C, Traber MG, Packer L. J. Lipid Res. 1996;
37: 893.17. Bland JM, Altman DG. Lancet 1986; 1: 307.18. Bland JM, Altman DG. Stat. Methods Med. Res. 1999; 8: 135.19. Lehmann J, Rao DD, Canary JJ, Judd JT. Am. J. Clin. Nutr.
1988; 47: 470.20. Saito M, Nakatsugawa K, Oh-hashi A, Nishimuta M,
Kodama N. J. Clin. Biochem. Nutr. 1992; 12: 59.21. Roxborough HE, Burton GW, Kelly FJ. Free Rad. Res. 2000;
33: 437.
Figure 4. Concentration of d6-labelled a-tocopherol deter-mined by LC/TOFMS in blood components in a single
individual following ingestion of a capsule containing 150
mg d6-labelled RRR-a-tocopheryl acetate.
LC/TOFMS analysis of tocopherols in blood components 2803
Copyright # 2003 John Wiley & Sons, Ltd. Rapid Commun. Mass Spectrom. 2003; 17: 2797–2803