207

CZinica Chimica Acta, 74 (1977) 207-215 @ ~l~evier/North-Holland Biomedical Press

CCA 8216

THE DETERMINATION OF TOTAL CHOLESTEROL IN SERUM BY GASLIQUID CHROMATOGRAPHY COMPARED WITH TWO OTHER METHODS

F.R. HINDRIKS,

B.G. WOLTHERS

and A. GROEN

*

University Hospital Groningen, Central Laboratory for Clinical Chemistry, Oostersingel59, Groningen (The Netherlands) (Received

July 9th, 1976)

Summary A gas-liquid chromatographic method is described for the measurement of total cholesterol in serum. The method has been found to be simple, specific and precise. The results have been compared with the results of the enzymatic method of Roschlau, P., Bernt, E. and Gruber, W. (1974) (Z. Klin. Chem. Klin. Biochem. 12, 403) and with the results of an Auto-Analyzer method based on the manual method of Huang, T.C., Chen, C.P., Wefler, V. and Rafter-y, A. (1961) (Anal. Chem. 33, 1405) and Ness, A.T., Pastewka, J.V. and Peacock, A.C. (1964) (Clin. Chim. Acta 10, 229). The results of the gas-liquid chromatographic and of the enzymatic method show a high degree of correlation. The results of the Auto-Analyzer method are about 0.75 mmolfl higher than those of the other two methods. The conclusion is drawn that the gas-liquid chromatographic method should be given consideration as a reference method for the measurement of total cholesterol in serum. It is a viable alternative for the generally accepted calorimetric reference method of Abell, L.L., Levy, B.B., Brodie, B.B. and Kendall, F.E. (1952) (J. Biol. Chem. 195,357).

Introduction The method of Abel1 et al. [I] is still generally accepted today as the reference method ** for the estimation of total cholesterol in serum. This method can be criticized as to its accuracy and precision. It has a number of critical steps and the recovery of cholesterol from the sample after extraction may be incomplete. Therefore it requires considerable skill to perform the assay. This led us to search for methods with a high accuracy and precision which are more easy to perform. -* To whom correspondence should be addressed. ** A reference method is one which after exhaustive inaccuracy in comparison with its impwxision.

investigation

has been shown

to

have negligible

Over the last few years gas-liquid chromatographic: (GLC) 121 and enzymatic [3] methods have become increasingly popular for the estimation of serum cholesterol. The methods are claimed to be very specific and precise. This study was undertaken to search for a more suitable refcrencc method for total cholesterol in serum. Therefore the GLC and enzymatic method were investigated. They could possibly serve as a substitut,e for the method of Abel1 et al. [ 11. In addition an automated routine method according to Levine et al. [4] was included in this study. Materials and methods Blood Blood samples were drawn from adult patients of our hospital after an overnight fast and were centrifuged at 1500 X g for 10 min. Serum was separated from the cellular elements within 2 h after collection of the blood and stored at -20°C or used immediately.

Total cholesterol was determined with a commercial kit (Boehringer M;annheim, G.F.R.) according to Rijschlau et al. [3]. The results were calculated using the molar extinction coefficient of 3,5-diacetyl-1,4_dihydrolutidine. Auto-Analyzer method Total cholesterol was determined with the tinuous-flow system (Technicon Instrument method is based on the work of Levine et al. method of Huang et al. [ 5] and Ness et al. (Technicon Instrument Corp., Tarrytown, N.Y

SMA-C, a third generation conCorp., Tarrytown, N.Y.). This [4] who automated the manual [6]. SMA-C Reference I serum .) was taken as the standard.

GLC method Total cholesterol was determined according to a modification of the method of Blomhof [2]. 3a,5P-Cholestanol is chosen as the internal standard for silylation purposes and because it resembles cholesterol much more than 5a-cholestane. The silylation procedure is carried out because it diminishes peak tailing. To 1.0 ml of a freshly prepared mixture, consisting of 6 ml 33% (w/v) KOH, 94 ml 96% (v/v) ethanol and 0.5 ml of an internal standard solution containing 200 mg 3~,5~-cholestal~ol per 100 ml of 96% (v/v) ethanol, is added 50 ~1 serum in a reaction tube. This mixture is incubated for 15 min at 55°C in a waterbath. After incubation, 1.0 ml distilled water and 2.0 ml hexane is added and the reaction tube is shaken for 30 set on a Vortex mixer. After centrifugation for 5 min at 1500 X g 1.0 ml of the hexane phase is separated from the water phase and pipetted in a reaction tube. The hexane is evaporated under a stream of nitrogen gas at 55°C. The residue is dissolved in 0.1 ml of a freshly prepared silylation mixture. This mixture consists of 4.0 ml bis-trimethylsilylacetamide, 1 .O ml trichloroethane and 5.0 ml pyridine. After incubation for 60 min at 37°C a 3-~1 sample of this reaction mixture is injected into a gaschromatograph. The gaschromato~aph was a Becker model 409 equipped with a flame ionisation detector {conditions: 2 m column 3% OV-225 on Chromosorb

209

W-HP 100-120 mesh, temperature of the oven 245°C and carrier gas nitrogen with a flow rate of 20 ml per min). A solution of 200 mg cholesterol in 100 ml 96% (v/v) ethanol (5.181 mmol/l) is taken as the standard. In every batch 2 standards are included which are treated in the same way as the serum samples. The concentration of cholesterol can be expressed as: peakheight cholesterol of sample ~~~ peakheight cholesterol of standard

x peakheight cholestanol of standard peakheight cholestanol of sample

x

X 5.181 mmol/l. The peakheights are measured in mm. 3a,5P-Cholestanol was purchased from Steraloids and all other chemicals were purchased from Merck and were of analytical quality. Results and discussion Chroma tograms Fig. 1 shows typical chromatograms of cholestanol and cholesterol acquired by gas-liquid chromatography. The peaks are almost symmetrical without tailing. No additional peaks were detected in the chromatograms of the standard solutions. Precision The within-run variation for the enzymatic method is about 2% (N = 9; Mean = 3.32 mmol/l; S.D. = 0.080 mmol/l; C.V. = 2.4% and N = 9; Mean = 7.25 mmol/l; S.D. = 0.155 mmol/l; C.V. = 2.1%). The day-to-day variation is less than 3% (N = 12; Mean = 7.28 mmol/l; S.D. = 0.181 mmol/l; C.V. = 2.5%). The determinations were carried out on 12 consecutive working days. These results agree well with those obtained elsewhere [ 31. The within-run variation for the Auto-Analyzer method is approx. l-2% (N = 9; Mean = 3.50 mmol/l; S.D. = 0.065 mmol/l; C.V. = 1.9% and N= 9; Mean = 7.25 mmol/l; S.D. = 0.062 mmol/l; C.V. = 0.9%). The day-to-day variation is less than 3% (N = 20; Mean = 7.20 mmol/l; S.D. = 0.194 mmol/l; C.V. = 2.7%). The determinations were carried out on 20 consecutive working days. These results agree well with those obtained elsewhere [ 71. The within-run variation for the GLC method is about 2% (N = 10; Mean = 4.84 mmol/l; S.D. = 0.119 mmol/l; C.V. = 2.5% and N = 10; Mean = 11.09 mmol/l; S.D. = 0.179 mmol/l; C.V. = 1.6%). The instrumental contribution of the GLC method to the overall within-run variation is approx. 1% as determined by repeated injection of the same sample (N = 10; Mean = 4.84 mmol/l; S.D. = 0.039 mmol/l; C.V. = 0.8%). Accuracy To several serum samples different known amounts of cholesterol dissolved in ethanol were added. The recovery of cholesterol by the GLC method was complete. The results are presented in Table I. Linearity For the enzymatic

and Auto-Analyzer

method

there exists a good linearity

210

r

I 15

1

I

L

/

30

25

20

I5

time

(mln

IO

i

I

5

0

1

Fig. 1, Chromatograms of cholestanol (1) and cholesterol (2) for standard solution (A) and for serum obtained by the GLC method. Conditions were as described in Materials and methods.

(B)

between the measured and the relative cholesterol concentration to approx. 11 mmol/l under the test conditions. The graphical representation of this data is given in Figs. 2 and 3. To test the linearity of the GLC method, dilutions are made of serum samples and of solutions of cholesterol in ethanol. In both cases there exists a good linearity between the measured and the relative cholesterol concentration to approx. 11 mmol/l under the test conditions. The graphical representation of these data is given in Fig. 4. Comparison

study

GLC us. e~zyrn~~ic rnet~~~ (?‘&le II; Fig. 5) The slope is 0.9988,

which indicates

a proportional

error of only 0.1%. Con-

211 TABLE I ACCURACY

OF THE CHOLESTEROL

DETERMINATION

BY GAS-LIQUID CHROMATOGRAPHY

Concentrations and recovery percentages of serum cholesterol spiked with an equivalent of 2.59 and 5.18 mmol cholesterol per litre are given. Original

Addition

COllCll.

2.59 mmol/l

(mmol/l)

4.82 4.87 5.18 5.18

5.18 mmol/l

-~

Concn. (mmolli)

Recovery Wf

Concn. (mmolfl)

Recovery (%)

7.46 7.56 7.82 7.88

101 104 101 102

10.03 9.95 10.36 10.36

100 98 100 100

Mean

100

102

stant error is estimated at only 0.03 mmoi/l by the y intercept. The correlation coefficient is 0.9965 or nearly ideal. The standard deviation of the differences (S,,) is estimated at 0.23 mmol/l. For a cholesterol concentration of 6.00 mmol/l determined by the GLC method, the enzymatic method will give an average value of 6.03 mmol/l, and there exists 95% certainty that the value will be between 5.56 and 6.50 mmol/l (+2S,, kO.47 mmol/l). GLC us. A~~o~A~~~y~e~ ~e~~o~ (Table II; Fig. 6) The slope is 0.8948, indicating a proportional error of 10%. Constant ,4

14 i

0

20

rekwe

do

60

IW

80

cholesterol concentratmn

(%I

error is

4

0

20

relatwe

Fig. 2. Linearity between the measured and the refative cholesterol enzymatic method.

60

40

cholesterol

concentration

Fig. 3. Linearity between the measured and the relative cholesterol concentration Analyzer method.

80

concentratton

100

(%I

obtained by the

obtained by the Auto-

20

0

40

relative

60

cholesterol

80

100

concentration

(%I

Fig. 4. Linearity between the measured and the relative cholesterol concentration obtained by the GLC method. :)-------‘) , standard solution of cholesterol; lB, cholesterol in serum.

estimated at 1.12 mmol/l by the y intercept which is ascribed to a non-specific contribution to the cholesterol values of the Auto-Analyzer method of bilirubin, proteins and phenolic compounds and is also found by other authors [ 3,8].

T-

2

0

total

4

cholesterol

6

by

8

GLC

IO

method

T

12

Cm mot/I

)

Fig. 5. Comparison of the cholesterol concentration obtained by the GLC method vs. the enzymatic method. See Table II for statistical parameters of the regression line.

213

The correlation coefficient is 0.9677. S, is estimated at 0.60 mmol/l. For a cholesterol concentration of 6.00 mmol/l by the GLC method, the Auto-Analyzer method will give an average value of 6.48 mmol/l, and there exists 95% certainty that the value will be between 5.28 and 7.68 mmol/l (?2S,, 51.20 mmol/l). The high value of S, reflects the magnitude of the random error introduced by the Auto-Analyzer method probably owing to a variable nonspecific contribution of interfering substances. Enzymatic vs. Auto-Analyzer method (Table II; Fig. 7) The slope is 0.9291, indicating a proportional error of 7%. Constant error is estimated at 0.72 mmol/l by the y intercept. The correlation coefficient is 0.9708. s,, is estimated at 0.60 mmol/l. For a cholesterol concentration of 6.00 mmol/l determined by the enzymatic method, the Auto-Analyzer method will give an average value of 6.29 mmol/l, and there exists 95% certainty that the value will be between 5.09 and 7.49 mmol/l (+2S,, 51.20 mmol/l). Here too exists a great random error owing to the Auto-Analyzer method reflected by S,,. In our opinion the GLC method is a viable alternative for the method of Abel1 et al. [l] as a reference method for the estimation of cholesterol in serum. The GLC method is simple and accurate, separates cholesterol from interfering substances, can be standardised with a primary standard of cholesterol and losses during the extraction procedure are corrected by the internal standard. The precision of the GLC method would be even better using the integrated areas of the peaks for calculation rather than the ratios of the peakheights that were used in the present study. The purity of the cholesterol standard was only tested by the GLC method described above under Materials and methods. No additional peaks were detected on the chromatogram. The reasonable assumption is made that cholestanol and cholesterol behave in the same way during the extraction procedure although slight differences cannot be excluded. This uncertainty can be overcome using a gas chromaIn this way an isotope dilution tograph-mass spectrometer combination. method can be set up in which an isotope derivative of cholesterol can be taken as the internal standard. The fact that the GLC method showed almost iden-

TABLE

II

STATISTICAL TEROL

PARAMETERS

OBTAINED

N = number correlation calculated Parameters

of results; coefficient; from

D = slope Sy,

the least-squares GLC

of

the

= standard

COMPARISON

69

least-squares deviation

equation

method

Auto-Analyzer N

FROM

OF

THREE

DIFFERENT

CHOLES-

METHODS

(Y

vs. method

= aX

line;

b = the Y intercept

of the differences + b), X method GLC

method

Enzymatic

49

of the least-squares

of the actual vs. Y method vs. method

Y value

from

the

line;

r =

Y value

is given. Enzymatic Auto-Analyzer 71

D

0.8948

0.9988

b

1.1157

0.0329

0.9291 0.7182

r

0.9677

0.9965

0.9708

SYX

0.5992

0.2325

0.6013

method

vs.

method

2

0

total

4

6

cholesterol

8

by GLC

IO method

12 (m mol/l

1

Fig. 6. Comparison of the cholesterol concentrations obtained by the GLC method method. See Table II for statistical parameters of the regression line.

0

2

total

cholesterol

6

4

by

enzymatic

8

IO

method

vs. the Auto-Analyzer

I2

( m mol/l 1

Fig. 7. Comparison of the cholesterol concmtration obtained by the enzymatic Analyzer method. See Table II for statistical parameters of the regression line.

method

vs. the Auto-

215

tical cholesterol values to the enzymatic method strongly supports the accuracy of these methods. Peaks never appeared other than those of cholesterol and 3a,5/3-cholestanol in the GLC chromatograms. The conclusion is drawn that other substances, e.g. drugs and sterols, interfere in the GLC method to a negligible degree. For routine purposes the GLC method is hardly suitable because one technologist can carry only out 50 to 75 tests per working day. However, when the reaction mixtures are automatically shaken and with an automatic sample injecting system and automatic digital integration of peak areas one can easily increase the workload to 100-150 tests per working day. Here the enzymatic method can serve as an accurate method. Its results show an excellent correlation with the GLC method. In addition, it is possible to carry out a great workload with the enzymic method since it can be automated [9,10]. The enzymatic method can be standardised using standards analysed by the GLC method. The present Auto-Analyzer method will always give inaccurate results owing to the non-specific contribution of interfering substances [3,8] and a serum blank as correction for this contribution is therefore insufficient. The proportional error of the results of the Auto-Analyzer method in comparison with both other methods can be omitted by correcting the calibration value of the reference serum which is taken as the standard. Acknowledgements The authors wish to express their gratitude to Miss C.A. de Bruyne, Mr. J. van der Meulen, Mr. G.J. Nagel and Mr. E.E. Ligeon for their technical assistance and Mr. J.J. Hoks for drawing and photographing of the figures. References 1 2 3 4 5 6 7 8 9 10

Abell, L.L., Lew, B.B., Brodie. B.B. and Kendall, F.E. (1952) J. Biol. Chem. 195, 357 Blomhoff, J.P. (1973) Clin. Chim. Acta 43, 257 Riischlau, P., Bernt, E. and Gruber, W. (1974) Z. Klin. Chem. Klin. Biochem. 12, 403 S. and Vlastelica, D. (1968) in Automation in Analytical Chemistry, Levine, J., Morgenstern, Technicon Symposia 1967, Vol. 1, p. 25, Mediad Inc.. White Plains, New York Huang, T.C., Chen. C.P., Wefler. V. and Raftery. A. (1961) Anal. Chem. 33.1405 Ness, A.T., Pastewka, J.V. and Peacock, A.C. (1964) Clin. Chim. Acta 10. 229 Schwartz, M.K., Bethune, V.G., Fleisher. M., Pennachia. G.. Menendez-Botet, C.J. and Lehman, D. (1974) Clin. Chem. 20.1062 Tanks, D.B. (1967) Clin. Biochem. 1.12 Zoppi, F. and Fenili, D. (1976) Clin. Chem. 22,690 Witte, D.L.. Barret. D.A. and Wycoff, D.A. (1974) Clin. Chem. 20.1282