primidone, phenobarbital, phenytoin, carbamazepine, theo- phylline, procainamide, propranolol, quinidmne, diazepam, chlordiazepoxide, acetaminophen, and ...
CLIN. CHEM. 27/3, 441-443
(1981)
Fluorometric Liquid-Chromatographic Determination of Serum Cortisol George R. Gotelli, Jeffrey H. Wall, Polar M. Kabra, and Laurence J. Marton1 We describe a sensitive and specific liquid-chromatographic assay for serum cortisol, exploiting acid-induced fluorescence. Each analysis requires only 50 zL of serum, and chromatography is complete in 7 mm. Analytical recovery of cortisol added to serum ranged from 90 to 105% and between-run precision (CV) from 6 to 8%. The lower limit of detection for cortisol is 10 g/L, and linearity extends to 1600 tgIL. Numerous steroids and drugs tested did not interfere.
were from Burdickand Jackson MI 49442.
The technique of sulfuric acid-induced fluorescence of steroids was first applied to the measurement of plasma corticosteroids by Sweat in 1954 (1). In 1962, Mattingly (2) modified the technique and developed a more practical method for the measurement of plasma cortisol. Subsequent modifications, to develop simpler but more specific assays, were reviewed by Rubin (3). Although cortisol is the principal adrenal corticosteroid found in plasma, other steroids, drugs, and several normal blood constituents can interfere with fluorometric methods (1, 4-6). Liquid chromatography, however, can provide the necessary specificity needed for the determination of serum cortisol. Recently, Reardon et al. (7) and Kabra et al. (8) have described specific liquid-chromatographic methods; however, these methods involve spectrophotometric detection, a technique generally less sensitive than fluorometric detection. Fluorometric liquid-chromatographic methods have been described by Kawasaki et al. (9) and Goehl et al. (10), but these methods involve derivatization with dansylhydrazine and thus require additional time. We describe a simple, specific, and sensitive fluorometric liquid-chromatographic method for serum cortisol in which corticosterone, 11-deoxycortisol, and numerous other steroids and drugs do not interfere. Only 50 1zL of serum is required foranalysis, making the method suitable for pediatric use.
equilenin
Materials and Method Equipment
We used a Model 601 liquid-chromatograph witha Model 023 recorder and a Model 650-1OLC fluorescence spectrophotometer (all from Perkin-Elmer Corp., Norwalk, CT 06856). Excitation and emission wavelengths were 366 and 488 nm, respectively. A reversed-phase column (-Bondapak C18; Waters Associates, Inc., Milford, MA 01757) was mounted in a 50#{176}C temperature-controlled oven. The column was eluted with a mobile phase consisting of water/acetonitrile/tetrahythofuran and 1 mol/L pH 4.4 phosphate buffer (63/28/8/1 by vol). A flow rate of 2 mL/min was used. Reagents
and Materials Centrifuge tubes. We used polypropylene 1.5-mL centrifuge tubes, with caps. Acetonitrile,
methylene
chloride,
and
conical-bottom tetrahydrofuran
Division of Clinical Chemistry, Department of Laboratory Medicine, School of Medicine, University of California, San Francisco, CA 94143. 1 L.J.M. is also a member of the Department of Neurosurgery. Received Aug. 28, 1980; accepted Dec. 30, 1980.
Ethanolic of concentrated
Laboratories,
Inc., Muskegon,
sulfuric acid was prepared by adding 70 parts sulfuric acid to 30 partsof absoluteeth-
anol. Steroids were purchased Louis,MO 63178. Injection
standard
from
(optional).
Sigma The
Chemical injection
equilenin(Sigma Chemical Co.), is prepared Evaporate to dryness 1.0mL of a 1000 mgfL in methanol.
To the dried
Co., St.
residue
standard, as follows: solution of add 1 mL of
ethanolicsulfuric acid,vortexmix, and heat for 1 h at 70 #{176}C, then cool and add 9 mL of water. Extract with 30 mL of methylene chloride, centrifuge, and aspirate the aqueous (upper) phase. Wash the methylene chloride once with 10 mL of water, aspirate, and discard the aqueous phase. Dry the methylene chloride with 2 to 3 g of sodium sulfate, decant the organic layer, and evaporate to dryness. Dissolve the residue in 200 mL of acetonitrile and store at 4 #{176}C. This solution is stable for at least two months at 4 #{176}C. Standard. A primary cortisol standard is prepared by dissolving 100 mg cortisol in 1 L of methanol, then diluting this 500-fold with water to make a final concentration of 200 zg of cortisol per liter. The primary cortisol standard is used to determine, by replicate analysis, the cortisol concentration of a serum pool, which is subsequently used as a daily working serum standard. This daily working serum standard is stored frozen at -20 #{176}C and is stable for at least two months.
Method Transfer 50 iL of unknown serum and 50 iL ofworking serum standard into separate polypropylene conical-bottom centrifuge tubes. Add 1.0 mL of methylene chloride to each tube, cap, shake for 5 mm, centrifuge for 1 mm, and carefully decant all of the methylene chloride from each tube into a clean centrifuge tube. (As shown by the recovery and precision studies, the methylene chloride can be quantitatively decanted, without the serum phase, if the recommended conical-bottom polypropylene small-volumecentrifuge tubesare used.)Then add 50 sL of ethanolicsulfuric acid to the de-
cantedmethylene chloride, cap,and shake for5 mm, centrifugefor1 mm, and aspirateand discardthe methylenechloride(upper)layer. Incubatethetubesforexactly2 mm at70 remove the tubes from the heating bath, add 50 iL of the preparedinjection standardto each,vortexmix, and immediately place the tubesintoa -20 #{176}C freezeruntilready to inject
into the liquid
chromatograph.
Results Figure 1 illustrates the chromatogram for a reference standard and for serum samples containing cortisol. Sensitivity. Cortisol can be detected and measured at a concentration of 10 ig/L when 50 iL of serum is extracted. Linearity. Cortisol was added to a serum pool to give concentrations from 100 to 1600 ig/L and aliquots of this pool were processed. The concentration of assayed cortisol was linearly related over the stated range. Precision. We evaluated(Table 1) within-run and between-run precision by processing replicate aliquots of a
serum sample atlow and highcortisol concentration. CLINICALCHEMISTRY,Vol. 27, No. 3, 1981 441
Table 2. AnalytIcal Recovery of Cortisol Added to A
I
Serum
ic
Cortleol, gig/L
100 200
90-95 198-206
Recovery, % 90-95 99-103
300 400 500
287-288 377-383 500-525
95-96 94-96 100-105
750
747-768
99-102
Add.d
a
a (
‘C 0
z
‘C
‘C
z
a (
I,-
U,
z
U,
z
‘C U,
z
0
0
0
V
U
z
z
I I
I
6
9
3
6
9
TIME (Minutes)
Fig. 1. Typical chromatograms for (A) the reference standard and (B and C) serum samples containing, respectively, 80 and 300 rg of cortisol per liter Recovery. Known amounts of cortisol were added to a pool of known cortisol concentrations and processed. At least three determinations were done at each indicated concentration. Recoveries are tabulated in Table 2. Additionally, we assayed a set of lyophilized cortisol standards (Quantimmune Cortisol Radioimmunoassay Standards, Control 99/7357; Bio-Rad Laboratories, Richmond, CA serum
94804). Results are tabulated in Table 3.
Discussion During development of this method we observed that sulfuric acid treatment of a cortisol standard resulted in numerous chromatographic peaks and that the number and amplitude of these peaks related directly to the duration and temperature of incubation. Minor peaks could be minimized
Table 1. PrecIsIon of Assay for Serum Cortisol Low cortleol concn
and the major cortisol peak maximized by reacting the cortisol/ ethanolic sulfuric acid mixture for 2 to 5 mm at 70 #{176}C. The resulting cortisol fluorphor was stable for 24 h if it was promptly cooled to -20 #{176}C after incubation. However, if the reaction time exceeded 5 mm and (or) if the reaction tube was not cooled, the cortisol fluorphor was unstable. To form the fluorescent derivative of equilenin, however, required a 60-mm incubation at 70 #{176}C. Because the equilenin and cortisol required different incubation times to form their respective flurophores, we were unable tousethe equilenin as an internal standard, but it can be added optionally at the end of the assay to control the injection volume. We also noted that the fluorescent properties of the sulfuric acid-induced cortisol fluorophor changed when added to the mobile phase; thus we used excitation and emission wavelengths quite different from those published for cortisol in sulfuric acid. The mobile phase we used has a pH of 3.4 and the column effluent has a pH of 2.7. Nevertheless, we have experienced no loss in column resolution after 500 analyses. Nonspecificity has been a major problem associated with sulfuric acid-induced fluorometric procedures for serum cortisol. Numerous drugs, other corticosteroids, and nonsteroidal material in normal serum can contribute to nonspecificity and have been the subject of intensive investigation. Potential interference with our method by other steroids and drugs was studied by chromatographing selected steroids and drugs at a concentration of 1000 zg/L after their reaction with ethanolic sulfuric acid as described. The following steroids were not detectable: dexamethasone, prednisone, prednisolone, deoxycorticosterone, 11-dehydrocorticosterone, testosterone, androsterone, 17a-hydroxyprogesterone, aldosterone, 11-deoxycorticosterone, tetrahydrocorticosterone, a-cortolone, 3-cortol, cortisone, 11-deoxycortisol, estrone, estradiol, estriol, and progesterone. Metyrapone was not detected, and corticosterone eluted at 8.4 mm. Both spironolactone and canrenone yielded two peaks each, which emerged
Table 3. Fluorometrlc Cortisol Assay: Results for Lyophlllzed Cortisol Standards a
High cortisol concn
Standard cortisol concn
Within-run precision Mean,
ig/L
SD, g/L
CV, % n
64 4.9 7.6
10 tg/L
SD, .zg/L
CV,% n
442
76 7.1 9.3
11.0
3.3 10
10 CLINICALCHEMISTRY.Vol. 27. No. 3, 1981
377 13.1 3.5
10
CorUsol concn determined
g/L
331
Between-run precision Mean,
Found
0
0
10
8.2 41
50 100
101
200
207
400
398
800
aQuantimmune Cortisol Radlolmmunoassay das.
791 Standards. Bio-Rad Laborato-
from the column at 7.5 mm and 9.1 mm. None of these peaks interfered with the analysis, though they appeared on the chromatogram. The following drugs were not detectable: ethosuximide, primidone, phenobarbital, phenytoin, carbamazepine, theophylline, procainamide, propranolol, quinidmne, diazepam, chlordiazepoxide, acetaminophen, and acetylsalicylic acid. Interference from nonspecific serum fluorogens was studied by examining the serum of subjects suppressed with dexamethasone. Dexamethasone is a cortisol analog that causes a dramatic decrease in circulating plasma cortisol. Baseline and post-suppression serum samples were collected from 13 subjects and the cortisol concentration was determined by the described fluorometric method and by two radioimmunoassay methods (Quantimmune I’ Cortisol Radioimmunoassay Kit from Bio-Rad Laboratories, Richmond, CA 94804; and Gammacoat I’ Cortisol Radioimmunoassay Kit from Clinical Assays, Division of Travenol Laboratories, Inc., Cambridge, MA 02139). The regression data on comparing the fluorometric method and the Bio-Rad radioassay method were: r = 0.987, m = 0.895, and they-intercept was -11.1. The regression analysis comparing the fluorometric method and the Clinical Assays radioassay method was: r = 0.992, m = 0.974, and they-intercept was -6.3. The 26 samples compared had a range of concentrations from less than 10 ig to 350 tg of cortisol per liter. We further assessed our method by comparing the cortisol concentration of 50 randomly chosen sera with the spectrophotometric method of Kabra et al. (8). The regression data for this comparison: r = 0.983, m = 1.09, andy-intercept = -4.0. The range of cortisol measured was between 10 g and 250 g of cortisol per liter. The results of these three correlation studies indicate that
the fluorometric method does not measure any nonspecific fluorogenic serum component. This liquid-chromatographic fluorometric method has now been routinely used in our laboratory to determine the cortisol concentration of 500 serum precision (CV) of 6.8%.
samples,
with
a between-run
References 1. Sweat, M. L., Sulfuric acid induced fluorescence of corticosteroids. Anal. Chem. 26, 773-776 (1954). 2. Mattingly, D., A simple fluorometric method for the estimation of free Il-hydroxycorticosteroids in human plasma. J. Clin. Pat hol. 15,374-379 (1962). 3 Rubin, M., Fluorometry and phosphorimetry in clinical chemistry. Adu. Clin. Chem. 13, 161-269 (1970). 4. Rado, J. P., Falsely high fluorescence in cortisol determination due to carbamezepine. Horm. Metab. Res. 5,63-66 (1973). 5. Verjee, Z. H., A note on the fluorometric determination of plasma cortisol. Clin. Chim. Acta 33, 268 (1971). 6. Lever, M., Repurification of dichloromethane used for extraction in fluorometric methods. Clin. Chim. Acta 31, 291-293 (1971). 7. Reardon, G. E., Caldarella, A. M., and Canalis, E., Determination of serum cortisol and il-deoxycortisol by liquid chromatography. Clin. Chem. 25, 122-126 (1979). 8. Kabra, P. M., Tsai, L. L., and Marton, L. J., Improved liquidchromatographic method for determination of serum cortisol. Clin. Chein. 25, 1293-1295 (1979). 9. Kawasaki, T., Maeda, M., and Tsuji, A., Determination of plasma and urinary cortisol by high-performance liquid chromatography using fluorescence derivatization with dansyl hydrazine. J. Chromatogr. 163, 143-150 (1979). 10. Goehl, T. J., Sundaresan, G.M.,and Vadlameni, P.K., Fluorometric high-pressure liquid chromatographic determination by hydrocortisone in human plasma. J. Pharm. Sd. 40, 1374-1376 (1979).
CLINICALCHEMISTRY,Vol. 27, No. 3, 1981 443
