| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
Institute for Clinical Chemistry and Pathobiochemistry, Medical Faculty, University of Technology Aachen, Pauwelsstr. 30, D-52057 Aachen, Germany.
a Author for correspondence. Fax 49-241-88-88-512.
 |
Abstract |
|---|
 Top Abstract IntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
Here we describe our new liquid chromatography–isotope dilution mass spectrometry (LC IDMS)1 method for the determination of cholesterol in serum. The novelty of this method is the separation of the analyte by HPLC previous to mass spectrometry. The evaporation of the eluent is done in a particle-beam interface used for coupling the liquid chromatograph and the mass spectrometer. Finally selective ion monitoring (SIM) is performed after electron impact (EI) ionization. A derivatization prior to analysis on the LC-MS instrument is not required. The results obtained for pooled human sera were compared with those obtained by a slightly modified variant of the gas chromatography (GC) IDMS method used by Siekmann (11). It was not intended to propose a candidate reference method, but to present an analytical procedure, which may be worked out to a candidate reference method for the determination of total cholesterol in serum.
|
Principles |
|---|
|
TopAbstractIntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
In the case of the GC IDMS method, a derivatization of the alcoholic groups to trimethylsilyl ethers has to be done prior to analysis. The GC separation is performed on a nonpolar fused-silica capillary GC column. In this case, the principal isotope ratio measurements are made from the ion abundances of the molecular mass ion of trimethylsilyl cholesterol, m/z = 458, and the respective molecular mass ion of the internal standard, m/z = 461. Standards are made by combining pure unlabeled cholesterol and [25,26,27-13C3]cholesterol to give one with an unlabeled/labeled ratio of ~1.0, one standard somewhat lower, and one somewhat higher. These mixtures are evaporated without previous liquid–liquid extraction and processed further like the serum samples. To get the best possible precision, a bracketing technique was applied for both methods (6).
|
Materials and Methods |
|---|
|
TopAbstractIntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
samples
The control materials used for checking accuracy of the methods
were human serum SRM 909 with a certified cholesterol concentration of
1415 ± 46 mg/L; SRM 909b, grade 1 with a certified cholesterol
concentration of 1464 ± 18 mg/L; and SRM 909b, grade 2 with a
certified cholesterol concentration of 2353 ± 30 mg/L from NIST.
The other control materials were Precinorm U from Boehringer Mannheim
with a target value for the concentration of total cholesterol of 1200
mg/L and Kontrollogen-LP from Behringwerke with a target value of 1330
mg/L. Both control materials were based on processed human sera, and
the target value for total cholesterol was determined by use of a GC
IDMS method that was not further specified.
For the method comparison pooled sera were used (n = 28). The pools were prepared with serum samples taken after analysis in the Central Laboratory of the Institute of Clinical Chemistry and Pathobiochemistry at the University Hospital of the Technical University Aachen. Samples were obtained from the outpatients' departments as well as the patient care units of the University Hospital. All specimens were collected in Sarstedt monovettes with separation gel. Serum was obtained after centrifugation.
instruments and settings
LC IDMS.
The analysis was performed on a Waters Integrity
system (Waters), consisting of an Alliance 2690 chromatography module,
a column bypass module, a photodiode array detector 996, and a Waters
Thermabeam mass detector, equipped with ion-source working unchangeable
in EI- mode. System controlling, data acquisition, and
integration were performed with the Waters Millennium Software, Rel.
2.21.
For chromatography a 150 mm x 2 mm Novapak C18 analytical minibore column (Waters Chromatography) was used. The eluent consisted of acetonitrile and isopropyl alcohol (65:35 by vol). The flow rate was 0.3 mL/min. The injection volume was 5 µL for all samples. The duration of one chromatographic run was 7 min.
The helium flow in the particle-beam liquid chromatography–mass
spectrometry interface was 30 mL/min. The nebulizer was heated to
70 °C, and the temperature of expansion region was set to 90 °C.
The pressure in the interface was constantly at 67 Pa. The temperature
of the ion source was set to 220 °C, the EI energy was set to 70 eV,
and the pressure in the ion chamber was 
0.026 Pa. The voltage
settings for the ion optic were 5 V for the ion volume, -42 V for the
extraction lens, -20 V for the prequad, and -101 V for the exit lens.
The multiplier voltage was set to 1980 V.
The measurement was performed in SIM mode at m/z = 386 (cholesterol) and m/z = 389 ([25,26,27-13C3]cholesterol) with a frequency of 1 scan/s; within one chromatography 420 scans were performed, and the duration of the elution of the cholesterol peak was about 50 scans.
GC IDMS.
The instrument used was a Fisons MD-800 combined gas
chromatograph–quadrupole mass spectrometer (Fisons Instruments),
equipped with an EI source and a GC8000 series gas chromatograph and
autosampler AS800. For instrument controlling and data acquisition the
Fisons MassLab Software Rel. 1.30 was used.
The gas chromatography was performed on a Hewlett- Packard Ultra 1 [0.33 µm, 12 m x 0.32 mm (i.d.)] capillary column (IAS, Leipzig, Germany). The carrier gas was helium at 690 kPa (100 psi) at a flow rate of 1 mL/min, the split exit was set at 50 mL/min (1:50), the injector temperature was 320 °C, the oven temperature was isothermal 280 °C, and the interface temperature was set to 290 °C. The sample size injected by the AS800 was constantly set to 1.0 µL.
The temperature in the ion source was set to 200 °C, the EI energy was set to 70 eV, and the emission current was set to 235 mA. The voltage settings for the ion optic were 1.4 V for the ion energy, 0.7 V for the repeller, 8 V for lens 1, 79 V for lens 2, 7.4 V for low mass resolution, and 12.2 V for high mass resolution. The multiplier voltage was set to 500 V.
For SIM mode measurements mass detection was set at m/z = 458 ± 0.25 (cholesterol) and m/z = 461 ± 0.25 ([25,26,27-13C3]cholesterol), the dwell time was set to 0.15 s, and the channel delay to 20 ms, leading to a measurement frequency of 3 scans/s. The data acquisition delay time was set to 2.45 min. Within one chromatography 450 scan were performed, and the duration of the elution of the peak of the cholesterol derivate was about 30 scans.
procedures and measurements
Weighing and pipetting procedures.
Cholesterol (SRM 911b) and
the isotopically labeled
[25,26,27-13C3]cholesterol were weighed on a
microbalance (Mikrowaage 708501, Fa. Sartorius). This balance has a
weight range of 15 mg, and the certified accuracy is 0.5% at 1 mg and
0.15% at 10 mg, respectively, which was checked with calibrated weight
prior to each use. All other weighing procedures, including all
required calibrations of volumetric devices, were done on a
semimicrobalance (Halbmikrowaage AC 211 S-OCE, Fa. Sartorius). This
balance has a single measuring range up to 210 g with a certified
reproducibility of ±0.1 mg and certified linearity deviation of
±0.2 mg. All pipetting procedures were performed with Digital
Syringe Series 1700 syringes (Fa. Hamilton). Every volume setting used
was calibrated gravimetrically before use.
Sample preparation for calibrators, control materials, and pooled
sera.
The reconstitution of lyophilized control sera was performed
as described previously (1). The calibrators of
cholesterol and [25,26,27-13C3]cholesterol
with a concentration of 1 mg/mL were prepared fresh every day by
dissolving 10 mg of solute in 10 mL of ethanol. To minimize the effects
of the varying accuracy of the syringe with varying pipetting volume,
in all cases 100 µL (corresponding to 100 µg of
[25,26,27-13C3]cholesterol) of internal
standard were pipetted, so only the volume of the unknown sample or
calibrator had to be varied.
Internal standards.
Aliquots of 100 µL of the internal
standard, the labeled
[25,26,27-13C3]cholesterol solution, were
placed in Reacti-vial test tubes. Then aliquots of 75 µL (standard 1)
or 125 µL (standard 2) of unlabeled cholesterol were added, and the
tubes were gently swirled. Two samples were required for the
determination of the isotope ratio in the pure unlabeled and the
labeled cholesterol. For this we placed 200 µL of unlabeled
cholesterol in one vial and 200 µL of the labeled
[25,26,27-13C3]cholesterol in another. The
ethanol was removed under a stream of nitrogen at 60 °C. For LC IDMS
measurements the residue was dissolved in 100 µL of ethanol. For GC
IDMS measurements we dissolved the residue in 50 µL of
N-methyl-N-(trimethylsilyl)trifluoroacetamide/pyridine,
and the derivatization was performed for 30 min at 60 °C.
Control materials, pooled sera.
Aliquots of 100 µL of the
internal standard, the labeled
[25,26,27-13C3]cholesterol solution, were
placed in test tubes. Then appropriate aliquots of the control material
or pooled serum were added volumetrically to give an isotope ratio of
~1.0, and the tubes' contents were gently swirled.
To prepare a set for the total cholesterol determination, we then added to each of the test tubes 150 µL of an aqueous potassium hydroxide solution (8.9 mol/L) and 1 mL of ethanol. This mixture was gently swirled and then heated at 50 °C for 3 h. To check for complete hydrolysis, the hydrolysis was performed in a separate experiment by adding 300 µL of the aqueous 8.9 mol/L potassium hydroxide solution and 1 mL of ethanol, swirling, and finally heating at 50 °C for 6 h. After hydrolysis, 1 mL of deionized water and 2 mL of cyclohexane were added. After continuous shaking for 5 min the cyclohexane phase was transferred to Reacti-vials. The samples were dried and derivatized as described for the internal standard.
Calibration and calculation for the determination of cholesterol in
serum with the GC IDMS and LC IDMS methods.
For the measurement of
the unknown samples, each sample of a control material or serum was
measured in triplicate bracketed by triplicate measurements of standard
1 (isotope ratio ~0.75) and standard 2 (isotope ratio ~1.25) in
either the order: lower weight ratio standard, sample, higher weight
ratio standard. This measurement was then repeated in the reversed
order. The three observed intensity ratios were acceptable only if the
CV was <0.5%, then they were averaged. If this could not be achieved,
the measurement of the standard, control, or unknown sample was
discarded. The quantity of analyte in the sample was calculated by
linear interpolation of the measured ratio of the sample between the
measured ratios of the standards with the known weight ratios as
described elsewhere (6). In every series two values for
each control material or serum sample were obtained by this procedure,
and these two values were averaged.
|
Results |
|---|
|
TopAbstractIntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
shows the UV-chromatogram at a wavelength of 210 nm as well as
the SIM chromatograms at the molecular mass ions
m/z = 386 and m/z = 389 for an extract
of a serum sample. The interferences observable in the UV-chromatogram
are not visible in the SIM measurements.
|
The EI ionization was sensitive enough for detection of the underivatized cholesterol, and the peak of the molecular ion was the base peak in the mass spectrum; in the case of the trimethylsilyl-derivatized cholesterol the base peak was at m/z = 329 for the unlabeled cholesterol as expected from previous publications (12), but the detection limit at m/z = 458 was nearly identical to that at m/z= 329. The signal-to-noise ratios were comparable, the retention time of the cholesterol in the LC IDMS method was 5.44 min, and the retention time of the cholesterol derivative in the GC IDMS method was 4.15 min.
We determined the isotope ratios of the pure unlabeled cholesterol and the pure [25,26,27-13C3]cholesterol from the two samples containing only one type of cholesterol, which were required in the further calculations. The pure unlabeled cholesterol had a peak intensity ratio (A389/A386) of 0.00242, the trimethylsilyl derivative a peak intensity ratio (A461/A458) of 0.0344; the pure labeled cholesterol had a peak intensity ratio (A386/A389) of 0.00646, the trimethylsilyl derivative a peak intensity ratio (A458/A461) of 0.0175.
interference of endogenous and exogenous steroids
We tested the GC IDMS method and the LC IDMS method for the
possible interference of 7-dehydrocholesterol,
5
-cholest-7-en-3ß-ol (lathosterol), lanosterol, ß-sitoserol,
ergosterol, cholest-4,6-dien-3-one, coprostan-3-ol,
25-hydroxycholesterol, cholesterol-5,6-epoxide,
4-cholesten-3-one, 5-cholesten-3-one, and dihydrocholesterol. In the GC
IDMS method none of the steroids studied interfered, as was expected
from earlier studies (12). In the LC IDMS only lathosterol
with a retention time of 5.49 min and a molecular mass of 386.7 could
interfere with the determination of the unlabeled cholesterol, the
actual signal of lathosterol at m/z = 386 being 30.1%
of that of an equal amount of cholesterol. Coprostan-3-ol with a
retention time of 5.28 min and a molecular mass of 388.7 could
interfere with the determination of the labeled cholesterol, the signal
of coprostan-3-ol at m/z = 386 being 1.9% and at
m/z = 389 8.9% compared with that of an equal amount
of cholesterol. All other steroids tested could not interfere, because
they eluted well separated from the cholesterol in HPLC.
memory effects
We tested the analytical systems used for LC IDMS and GC IDMS for
the existence of memory effects. If an unexpected memory effect were
present, then the measured isotope ratio of a sample would be
influenced by the history of samples measured previously. So we
routinely measured on both systems sequences of five determinations of
a sample of unlabeled cholesterol, followed by five determinations of
the isotope-labeled cholesterol and repeated this all five times. We
never observed a drift in the isotope ratios for the unlabeled or the
labeled cholesterol with both IDMS methods. If a memory effect were
present, then it should be detectable at least in this situation,
measuring a sequence of samples with the most extreme isotope ratios
possible.
linearity of the lc idms method
We tested the linearity of the relationship between the mass
ratios (c/c) of unlabeled and labeled cholesterol and the isotope
ratio, calculated from the area A386 under the peak
obtained in the SIM chromatogram at m/z = 386 and from
the area A389 under the peak obtained in the SIM
chromatogram at m/z = 389. The isotope ratio was
corrected for the isotope ratio f1 = 0.00646 of
the pure labeled cholesterol and the isotope ratio
f2 = 0.00242 of the pure unlabeled cholesterol,
both determined in the previous section, by the formula
 | (1) |
, demonstrate that linearity is given for isotope ratios between
0.25 and 2.0. The relationship is Y = (-0.0005 ±
0.0116) + (1.1404 ± 0.0094) x (c/c) with a correlation
coefficient of r2 = 0.9998, the intercept of the
regression line is not significantly deviating from the origin, and
there was no intrinsic nonlinearity observable in the range of mass
ratios tested.
|
standard stability
The stability of the mass ratio between days was calculated from
the measured, corrected isotope ratios Y found for the lower
and for the higher standard. The statistic was done for the first
measurement of the standards on each day. For the lower standard we
found for 20 days a mean of 0.7514 and a SD of 0.0021 (CV =
0.28%); for the higher standard we found under the same conditions a
mean of 1.2529 and a SD of 0.0033 (CV = 0.26%).
standard consistency
For the assessment of the standard consistency we used every day
two sets of two standards, one set for the calibration of the method
and an independent second set with the same isotope ratios for
measuring standard recovery. The second set of standards was treated
like unknown samples, and the results were calculated as for unknown
samples. The data were evaluated as described elsewhere
(5). In the LC IDMS we never found differences between the
calculated mass ratio (c/c) of unlabeled and labeled cholesterol and
the weighed-in mass ratio (c/c) of unlabeled and labeled cholesterol
>0.20%.
sample preparation
The hydrolysis procedure used in this work is similar to that
described elsewhere (6)(10). The main
difference is the much larger excess of hydrolyzing reagent used here.
So we only checked whether the simultaneous doubling of the hydrolyzing
reagent and the hydrolysis time had an impact on the results obtained.
Each sample was hydrolyzed in a separate experiment with a doubled
amount of hydrolyzing reagent and was incubated for 6 instead of 3
h. In no case could a difference be observed between the results
obtained for both hydrolysis procedures.
We also checked for the possible influence of the sample preparation method on the standards. We performed experiments in which we prepared standards like unknown samples. No differences could be observed between the measured isotope ratios of the standards used for calibration and prepared normally and the standards prepared like serum samples or control materials.
precision and accuracy of the lc idms method
Table 1
shows the day-to-day precision for five different serum-based
control materials, obtained from 10 independent series on separate
days. For the three control materials obtained from NIST, the CVs are
<1.0%. For the other control materials the imprecision is slightly
>1.0%. This could be caused by the vial-to-vial variabilities of the
control material rather than by LC IDMS method itself. The control
materials obtained from NIST were also used as controls for the GC IDMS
method. In that case CVs <1.0% were obtained also.
|
The accuracy of the LC IDMS method for the determination of total cholesterol in serum was checked with the five different serum-based control materials, for which the target values were based on Definitive Methods in the case of the NIST materials and on GC IDMS Reference Method-based target values supplied by the manufacturer. For all tested control materials the bias (from 10 independent series on separate days) was <1.0% and in all cases lower than the CV. In the GC IDMS method, the bias of the three control materials SRM 909, SRM B1, and SRM B2 was <1.0% as well.
For the method comparison 28 pooled human sera were used and measured
independently with the LC IDMS method presented and GC IDMS as a
Reference Method. Fig. 3
shows the CVs, obtained for each serum pool from five
determinations with both methods. The mean imprecision was 0.66%
(range 0.26–1.21%) for the GC IDMS method and 0.72% (range
0.31–1.17%) for the LC IDMS method. The mean results for the 28
pooled human sera, obtained by our LC IDMS method, were compared with
the chosen GC IDMS Reference Method. Fig. 4
a shows the correlation of the results including the graph of
the following linear relationship obtained by the method of Passing and
Bablok (13): CLC IDMS = 0.993
x CGC IDMS - 0.15 mg/L; 95% confidence
interval for the slope: 0.978–1.008; 95% confidence interval for the
intercept: -25.8 mg/L to +25.9 mg/L; correlation coefficient
r: 1.000; standard error of the estimates
Sy|x: 1.375.
|
|
In Fig. 4b
the relative deviations of the results obtained with the LC
IDMS method from those obtained with the GC IDMS Reference Method
(11) are presented graphically. For 32% of the samples
the relative deviation between the LC IDMS and the GC IDMS method was
<0.5%; for 93% of the samples the relative deviation did not exceed
2.0%.
|
Discussion |
|---|
|
TopAbstractIntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
As has been stated (10), the use of an isotope-labeled internal standard differing only by 2 amu from the unlabeled cholesterol leads to a nonlinearity between the corrected peak intensity ratio Y and (c/c). The use of an internal standard differing by 3 amu from the unlabeled compound prevents this effect nearly completely.
We chose two standards and bracketing as the calibration method, as
described previously (6). The use of two standards proved
to be sufficient, the standards were consistent within-run, and the
isotope ratios obtained for the standards were stable for at least 1
month with a CV <0.3% in the measured isotope ratios. The sample
preparation technique did not influence the measurement of isotope
ratios of standards prepared like serum samples. As stated earlier
(6), even shorter periods of time for hydrolysis and lower
concentrations of the hydrolyzing reagent lead to a total hydrolysis of
cholesterol esters. We did not observe any differences between aliquots
hydrolyzed under different conditions. The error introduced by the
uncertainty of the used standard materials as well as the error
introduced by the weighing procedure are in the order of magnitude of
0.1–0.2%. Overall the imprecision of the LC IDMS method was 1.2%
for all control materials studied, <1% for NIST materials. For all
control materials the bias was lower than the CV of the results.
Furthermore in the case of the NIST control sera with certified
uncertainty of the target value, the bias found was lower than this
uncertainty, which represents the highest bias allowed (Table 1
).
The new method was tested against an GC IDMS Reference Method for the determination of total cholesterol in serum. For 28 pooled human sera we found a very good correlation between the results obtained with the new method and the Reference Method. For 93% of the samples the difference of the results for both analytical methods did not exceed 2.0%. Neither the imprecision nor the bias between the methods depended on the concentration of the pooled serum samples used. Even for the highest concentration sample (4351 mg/L) measured by LC IDMS, the bias was 2.04% and was therefore not excluded from the statistical analysis; the relevance of a precise measurement in this range of concentrations is nevertheless limited. The between-run imprecisions for these serum pools did not exceed those obtained for the control materials.
In this study a new analytical technique was used for the development of an isotope dilution method. The liquid chromatographic separation method is applicable to a wide variety of dissolved organic molecules (14), and no derivatization procedure is required previous to analysis, making the sample preparation more simple than in the GC-MS method. The comparison of accuracy and precision of the new LC IDMS method with an GC IDMS method chosen as reference did not show any substantial advantage of the GC IDMS method. The method presented in this study may be considered as a prototype of isotope dilution methods applicable to analytes for which the sample preparation required for gas chromatography is complicated or a volatile derivative does not exist.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionPrinciplesMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  D. W. Johnson, H. J. ten Brink, and C. Jakobs A rapid screening procedure for cholesterol and dehydrocholesterol by electrospray ionization tandem mass spectrometry J. Lipid Res., October 1, 2001; 42(10): 1699 - 1706. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
