BIOMEDICAL CIIROMATOGKAPHY, VOL. 6 , 109-114 (1992)

Determination of Urinary Free Cortisol by High Performance Liquid Chromatography with Sulphuric Acid-ethanol Derivatization and Column Switching Osamu Nozaki,* Tsuneko Ohata, and Yasuhiro Ohba Department of Clinical Pathology, Kinki University School of Medicine. 377-2. Ohno-Higashi, Osaka-Sayama, Osaka 589, Japan

Hiroyuki Moriyama and Yoshio Kato Scientific Instrument Division, TOSOH Corporation, 1-11-39. Akasaka, Minato-ku, Tokyo 107. Japan

A rapid and highly sensitive determination method for urinary free cortisol has been developed using reversed phase high performance liquid chromatography (HPLC) with a precolumn for sulphuric acid-ethanol fluorescence derivatizationand column switching. Urinary cortisol, eluted from the octadecylsilane-bondedsilica (ODS) minicolumn with 90% aqueous ethanol, was derivatized with the addition of sulphuric acid only at ambient temperature. Cortisol derivatives injected directly onto the ODS precolumn were purified on-line. After switching the columns, the cortisol derivative was separated on an ODS analytical column with a retention time of 15.3 min and monitored at an emission wavelength of 520 nm (exitation wavelength of 365 nm) to decrease the detection limit to 0.26 pg/dL (signal-to-noise ratio =3). The automated HPLC operation resulted in good reproducibility and recovery of the stable cortisol derivative at 5°C.

INTRODUCTION

Serum-free cortisol, an important physiological substance, is excreted from the kidney after conjugation in the liver. However, a small quantity of unconjugated material is excreted into the urine. The urinary free cortisol level indirectly reflects the value of serum-free cortisol as well as adrenocortical function (Kobberling and von zur Muhlen, 1974; Nakamura and Yakata, 1989). Immunoassays for urinary free cortisol requiring no previous extraction have been mainly adopted by clinicians because of convenience. However, direct immunoassays have problems with interference and cross-reaction with coexisting conjugated glucocorticoids and other hydrophilic substances in urine samples (Beverley et al., 1981). High-performance liquid chromatography (HPLC) is one method which avoids the cross-reactivity associated with urinary cortisol immunoassays. Some investigators have separated urinary cortisol by HPLC (Canalis et al., 1982; Kabra, 1988), but these methods required inconvenient prior liquid-liquid (Canalis at al., 1982) and solid-liquid (Diamandis and D’Costa, 1988) extractions. Schoneshofer assayed urinary cortisol by radioimmunoassay after separation by HPLC (Schoneshiifer, 1980). Thercfore, a chromatographic method for the highly sensitive detection of urinary cortisol without need for prior extraction is needed for clinical applications. Shihabi and coworkers developed a highly sensitive assay method for urinary cortisol by precolumn for fluorescence derivatization with aqueous sulphuric acid * Author to whom correspondence should be addressed, 0269-3879/92/030109-06 $05.OO @ 1992 by John Wiley & Sons, Ltd.

and heating at 70°C and then by reversed phase HPLC with fluorimetry (exitation wavelength 390 nm and emission wavelength 475 nm) (Shihabi et al., 1982; Kabra, 1988). However, their method required two extractions with dichloromethane of urinary cortisol and a derivatized reaction aliquot, as well as manual sample injection due to storing the derivatives at -20°C. In this paper, we describe a rapid and highly sensitive determination method for urinary free cortisol by precolumn for sulphuric acid-ethanol derivatization and automated reversed phase HPLC with column switching. Urinary cortisol was derivatized fluorescently by adding sulphuric acid to a 90% aqueous ethanol eluate from an octadecylsilane-bonded silica (ODS) gel minicolumn at ambient temperature. The stability of the cortisol derivative at 5°C in the dark enabled automatic sample injection onto an ODS precolumn. The cortisol derivative was concentrated and purified on-line, and separated on an ODS analytical column after switching columns.

EXPERIMENTAL Chemicals. Ethanol. acetonitrile and sulphuric acid were obtained from Kanto Chemical Co. (Tokyo, Japan). Potassium dihydrogen phosphate and dipotassium hydrogen phosphate were from Wako Pure Chemical Industries (Osaka, Japan). Trifluoroacetic acid was from Tokyo Kasei Kogyo Co. (Tokyo, Japan). All these reagents were of analytical reagent grade. Cortisol( llB,17a,21trihydroxypregn-4-ene-3,20-dione), allotetrahydrocortiso!(3u,ll~,17u,21 - tetrahydroxy - 5a - pregnan - 20 - one), Received 26 June 1991 Accepted 4 October I991

110

OSAMU NOZAKI E T A L .

prednisolone(11~,17cr,21 - trihydroxy - 1,4- pregnadiene- 3,20dione), cortolone(3a,17~,20a,21-tetrahydroxy-5~-pregnan11-one), cortisone( 17a,21-dihydroxy-l,4-pregnadiene3,11,20-trione), dexarnethasone(9a-fluoro-16a-methylllp,17a,21 - trihydroxy - 1,4 - pregnadiene - 3,20 - dione), betamethasone(9u - fluoro - 1613 - methyl - llb,17a,21 trih ydroxy- 1,4-pregnadiene-3,2O-dione), cortol(3a ,116, 17u,20u,2 1-pentahydroxy-56-pregnan) , tetrahydrocortisone(3~.17a,21-trihydroxy - 5b - pregnan - 11,20dione), corticosterone(11~,21-dihydroxy-4-pregncne-3,20dione), androstenediol(3~,17~-dihydroxy-5-androstene), androstenedione(4-androstene-3,17-dione), testosterone(17~-hydroxy-4-androsten-3-one), dehydroepiandrosterone(3b-hydroxy-5-androsten- 17-one), estradio1(3,17&dihydroxy1,3,5(10)-estratriene), estriol(3,16a, 176-trihydroxy1,3,5(10)-estratriene), estrone(3-hydroxy-1,3,5(10)estratrien-17-one) and progesterone(4-pregnene-3,20-dione) were from Sigma Chemical Co. (St. Louis, MO, USA). These reagents were used as received. Distilled water was further purified by passing through an ion exchange column (Mini-Q Laboratories, Millipore, Bedford, MA, USA). Preparation of solutions. Phosphate buffer, 57 mmol/L (pH 8.0), was prepared with potassium dihydrogen phosphate and dipotassium hydrogen phosphate. Aqueous potassium dihydrogen phosphate solutions of 20, 23.5 and 36.4 mmol/L were adjusted to pH 1.85 with trifluoroacetic acid. Stocks of cortisol were prepared to 100 pg/dL with ethanol and stored at 4°C in the dark for up to one week. Working solutions of cortisol were prepared by diluting the stocks with water, stored at 4°C and used within the day. Sample collection. Spot urine specimens collected from patients and healthy volunteers were stored at 4°C and assayed within 24 h. Urine specimens of patients and cortisol data were kindly donated by Mitsubishi Yuka Biochemical Laboratories (‘l’okyo, Japan), for comparison with storage at -20°C for one month after assay with a fluorescence polarization immunoassay (FPIA) kit (TDX system, Abbott Laboratories, North Chicago, IL, USA) (Shipchandler et al., 1987). Specimens were thawed only once, immediatcly prior to HPLC assay. Extraction of urinary free cortisol. Urinary cortisol was extracted by passing 2.0 mL urine, 15% ethanol in 57 mmol/L phosphate buffer (pH 8.0; 2.0 mL) and water (2.0 mL x 2) through a bovine serum albumin (BSA) coated ODS minicolumn ( l m L , 20ym, TSK gel BSA-ODs; TOSOH Co., Tokyo, Japan) using an aspiration box ( V A C - E L U Y ; Analytichem International Co., Harbor, CA, USA) connected with a vacuum pump. BSA-ODS minicolumns were previously washed by passing 4.0 mL methanol and water under reduced pressure. Urinary cortisol was eluted from the minicolumn with 2.0 mL 90% aqueous ethanol. Fluorescence derivatization. Sulphuric acid (480 pL) was added to 320p.L of the eluate from the BSA-ODS minicolumn. The mixture was allowed to stand in the dark at ambient temperature for 10 min after vortex mixing for 1min. The mixture was cooled with crushed ice for 15min in the dark, then 20 mmol/L aqueous phosphate solution (pH 1.85, 700 pL) was added and the mixture was cooled with crushed ice in the dark for 15min, once more. The samples were stored at 5°C in the autosampler before assay by HPLC within three hours of preparation.

n

q-F*

waste

Figure 1. Block diagram of the column switching HPLC. S1, S2 and 53: mobile phase; P1 and P2: HPLC pump; V1 and V2:

electric high pressure column switching valve; AS: autosampler; C1: precolumn; C2: analytical column; D: detector; INT: integrator; FDD: floppy disk drive.

HPLC system and chromatographic conditions. Figure 1 shows a block diagram of the column switching HPLC system. It consists of two HPLC pumps, P1 and P2 (CCPM, TOSOH), electric switching valves, V1 and V2(PT-8WO, Tosoh), an autosampler with l0OOpL sample loop volume (AS-8000, TOSOH), a fluorimeter with 16yL flow cell volume (,FS-8010, TOSOH), an integrator (CP-8000, TOSOH) and a floppy disk drive (FD-8021, TOSOEI). The precolumn (Cl) was ODS-80Tm, 5 pm, TSK; 10x4.6mm i.d.. The analytical column (C2) was ODS-80Tm, 5 p m , TSK; 150 X 4.6 mm i.d.. The column temperatures were both ambient. The mobile phase S1 consisted of acetonitrile and 23.5 mmollL phosphate solution (15:85, v/v; pH 1.85 with trifluoroacetic acid). Solvent S2 consisted of acetonitrile and water (5050, v/v). Solvent S3 was acetonitrile and 36.4 mmol/L phosphate (4555, v/v; pH 1.85 with trifluoroacetic acid). Solvents S1, S2 and S3 were delivered isocratically at flow rates of 1.0, 2.0 and l.Oml/min, respectively. Initially, solvent S l was discarded after passing through the column switching valve V2 and a precolumn C1. On the other hand, solvent 53 flowed through the switching valve, an analytical column C2 and a detector to waste in that order. Aliquots of the crude cortisol derivatives, 1000 pL, were injected onto the precolumn C1 with an autosampler, where they were trapped and rinsed with S1. Five minutes after injection, samples were transferred from the C1 column to C2 with S3 by turning the valve V2, and monitored with a fluorimeter (exitation wavelength 365 nm, emission wavelength 520 nm) after separation in the C2 column. Valve V2 was returned to the original position at 9 min. For 5-13 min after sample injection, the C1 column was rinsed with S2 and for 13-25min, the solvent in the C1 column was replaced with S1 to the initial condition. On the other hand, the solvent S3 passed through the V2 and C2 column during analysis, but flowed via V2, C1 to C2 during 5-9 min (Fig. 2). The electric signals from a detector were integrated and recorded on a floppy disk. The calculation of peak areas on a chromatogram was carried out with the valley-to-valley mode of the integrator. The samples were injected with an autosampler every 25 min and the assays were completed within 3 h after the sample preparation. Three kinds of authentic cortisol solutions as an external standard were assayed for a calibration line at the same time as assays of urinary cortisol.

IIPLC OF URINARY FREE CORTISOL

111

n

c 100>r

c,

*r

vl

80-

c,

c

.c

60aJ

x

RESULTS

Influence of sulphuric acid concentration on the derivatization The optimum volume of sulphuric acid required to derivatize cortisol(5.0 pg) in an 800 pL reaction aliquot consisting of sulphuric acid and 90% aqueous ethanol was determined. Figure 3 shows the yields of the cortisol derivative when the sulphuric acid ratios were 30,40,50,60,70 and 80% (v/v) of the reaction aliquot, respectively. The maximum yield of the derivative was obtained with 60%, v/v, or 480 pL of sulphuric acid. Influence of ethanol concentration on the derivatization The optimum ratio of aqueous ethanol required to derivatize cortisol (5.0 pg) was determined using 10,20, 30, 40, 50, 60, 70, 80, 90 and 100%(v/v) aqueous ethanol(320 pL) and 480 pL sulphuric acid. The results are shown in Fig. 4. The maximum cortisol derivative yield was obtained with 90% ethanol (36% in the reaction aliquot). Stability of the fluorescent derivative The time course of the fluorescence intensity of the cortisol derivative was investigated. When the cortisol derivatives were kept in the dark in an autosampler at either ambient temperature or at 1O"C, the intensity decreased quickly. Figure 5 shows the time course of the fluorescence intensity of the cortisol derivative after storage at 5°C in the dark. The changes of the relative fluorescence intensity of the derivative of cortisol in water was 100, 98.5, 102, 101, 102, 99.6, 99.6, 99.6, 95.8, 92.8 and 92.7'30 at 0, 1, 2, 3, 4, 5 , 6, 7, 8, 9 and 10 h after the reaction, respectively. On the other hand, those in urinary cortisol changed to 103, 98.9, 98.4, 97.8, 92.0, 92.5, 93.2, 90.5,93.0, 87.8 and 86.6% at the same time, respectively. Therefore, we finished the assays of the urinary cortisol samples 3 h after the reaction with the aqueous authentic cortisol as an external standard. Specificity of the derivatization reaction for several steroids Table 1 shows the retention times and relative fluorescence intensities of 5 pmol/L glucocorticoids, glucocorticoidal drugs, corticosterone, androgens, estrogens and progesterone. The derivatives of cortisol, corticosterone and testosterone emitted intense fluorescence, whereas the others emitted very weak fluorescence or

v-7-l-l 0

20

40

60

80

Sulfuric acid ( % I Figure 3. Influence of the volumes of sulphuric acid on the cortisol derivatization.

none at all. The retention times of the derivatized cortisol, corticosterone and testosterone were 15.3, 32.0 and 96.9 min, respectively. Chromatogram

Figure 6(a) shows a chromatogram of the authentic cortisol derivative (21.3 ng/injection) which had a retention time of 15.3 min. Figure 6(b) shows a chromatogram of a typical urinary cortisol derivative assayed as 15,4ng/injection. The retention time is the same as that of the authentic cortisol derivative. The peak corresponding to cortisol in the urinary specimens was identified as follows: (1) an increase of peak area at a retention time of 15.3 min with no n

w

Y

>r

2 100vl c:

a, c,

.4 80. a,

u c

60-

vl

a, L

0 1

G

40.

aJ w '$ 20a a, 8

20

w

8

8

40 60 80 Ethanol ( % I

d

100

Figure 4. Influence of the concentrations of aqueous ethanol on the cortisol derivatization.

OSAMU NOZAKI ET A L

lo

b)

a)

(mV)

1 I

f

Time ( h )

0

Figure 5. Stability of fluorescence intensity of derivatized cortisol. The solid line and the broken line show the time courses of the fluorescence intensities of authentic and urinary cortisol derivatives, respectively.

changes in the shape of the peak relative to the added amount of authentic cortisol and (2) no or small peaks at the retention time in the blank urine where cortisol was not identified by FPIA. Calibration

Working stocks of authentic cortisol (0, 0.05, 0.1, 0.3, 0.5, 1 .O, 5.0, 10, 50, 100 and 200 pg/dL) were assayed by HPLC. The amounts of cortisol and the peak areas were compared to result in correlation of the coefficients P = 1.00 for n = 36 and the regression line Table 1. Specificity of the fluorescent derivatives. The concentration of the steroids was 5 pmol and the relative fluorescence intensity of the cortisol derivative was 100 Retention Steroid

Cortisol Allotetra hydrocortisol Prednisolone Cortolone Cortisone Prednisone Dexarnethasone Betarnethasone Cortol T-+-.. h .,A--..-

A:..-..-

I CLI~II~UIUL.UILI~UIIC

corticdsterone Androstenediol Androstenedione Testosterone Dehydroepiandrosterone Estradiol Estriol Estrone Progesterone a not detected

time (min)

Relative fluorescence intensity ( % I

15.3 15.4 15.3

100 1.4 0.7

-a

32.0 13.9

707 0.2

96.9

-

13.9 12.9

-

426 0.5 0.7

1

1

10 15.3 20

, 0

I

10 15.3 20

Retention time ( m i n 1 Figure 6. Chromatograms of authetic (a) and Urinary (b) cortisol derivatives. The amounts of authentic and urinary cortisol injected were 21.3 and 15.4 ng assayed respectively. The retention time of the cortisol derivative was 15.3 min.

as y = 17.6x+0.11 0, axis; peak area of the cortisol derivative, lo3mV). Therefore, the regression line was taken as the calibration curve for cortisol at 0.1-200 &dL. The detection limit was 0.10 pg/dL and 0.26 pg/dL for authentic and urinary cortisol, respectively, with a signal-to-noise ratio of 3 .

Precision Intra-assay reproducibility. Three different urine speci-

mens were assayed for intra-assay reproducibility. The values of cortisol were 5.81 +0.11,32.5 k 1.0 and 117 t2.9 pg/dL (means k S.D., n = 10). Therefore, the coefficients of variation (C.V) were 1.9, 3.2 and 2.5”/0, respectively. Inter-assay reproducibility. Three different urine samples were also assayed for inter-assay reproducibility. The values of cortisol were 5.44 k 0.27, 28.3 k 0.73 and 1 2 4 k 4 . 2 pg/dL (means+S.D., n = 10). Therefore, the C.V. were 4.9, 2.6 and 3.4%, respectively.

Recovery

Authentic cortisol (5.0, 20.0 and 30.0 pg/dL) was added to the urine samples containing 8.9, 58.0 and 118 pg/dL cortisol, the predicted results being 13.9, 78.0 and 148 pg/dL, respectively. The actual results after extraction of urinary cortisol with the BSA-ODS minicolumns were 13.3, 76.8 and 146 pg/dL cortisol, respectively. Therefore, the recoveries were 88,94 and 93%, respectively (each n = 5).

HPLC OF URINARY FREE CORTISOL

Comparison with FPIA method

The cortisol values of patients' urine samples assayed by HPLC and FPIA were compared, resulting in a coefficient of correlation r=0.76 and a regression line ~ 5.4 0,axis-HPLC) for n = 45 as shown of y = 0 . 5 2 + in Fig. 7.

DISCUSSION Extraction of urinary cortisol with a BSA-ODS minicolumn

The prior extraction of cortisol from urine samples is undesirable for its rapid assay. When sulphuric acid and ethanol were directly added to urine samples, the cortisol derivative could not be well separated on the chromatograms because of interference by various unknown substances. Thus, prior extraction of urinary cortisol was required to eliminate contaminants. We used a BSA-ODS minicolumn for the simple and rapid prior extraction of urinary free cortisol. The BSA-ODS gel was not damaged by urinary protein because of its biphasic characteristics; e.g. it is hydrophilic on its surface and hydrophobic in the pores. The BSA-ODS minicolumn was also economical because it could be reused at least 10 times. The eluate from the minicolumn containing 90% aqueous ethanol could be used as a part of the sulphuric acid-ethanol reaction without :he drying which was required in previous methods (Canalis et al., 1982; Diamandis and D'Costa, 1988; Shihabi et al., 1982). Fluorescence derivatization with sulphuric acidethanol

Although ethanol is needed to derivatize cortisol in the assay method for 11-hydroxycorticosteroids (De Moor et al., 1960; Sweat, 1954). Shihabi and coworkers deri-

-

'O01

113

vatized cortisol by heating with sulphuric acid and water at 70°C. In our study, ethanol was advantageous in yielding the cortisol derivative as shown in Fig. 4. The optimum ratio of aqueous ethanol(36% v/v) in the reaction mixture consisted of 320 pL 90% aqueous ethanol and 480 pL sulphuric acid at ambient temperature. That is, 90% aqueous ethanol was both effective in eluting urinary cortisol from the BSA-ODS minicolumn mentioned above and in derivatizing cortisol. This derivatization method for cortisol has another advantage of not requiring the elimination of fluorescence labelling reagents since these were not used. Cortisol, corticosterone and testosterone were unique among the steroids examined (Table 1). Although the corticosterone and testosterone derivatives were also transferred from C1 to C2 with S3, there was no interference of the cortisol derivative peak with the those peaks. The reason seems to be that the peak of the corticosterone derivative appeared at the retention time of 7 min of the next chromatogram. The peak of urinary testosterone did not seem to be detected on the chromatogram because of the low concentrations in the urine samples and the broad peak due to its very long retention time. On-line extraction and purification

The eluates of urine samples from the BSA-ODS minicolumns still contained some contaminants derivatized by the sulphuric acid-ethanol reaction, which interferred with the cortisol derivative peak on the chromatogram. The reactant therefore required further purification. The precolumn liquid-liquid extraction of the cortisol derivative by manual operation (Eechaute, 1966) is complex and time-consuming. The cortisol derivative was extracted on-line by direct injection of the reactant onto the ODS precolumn. Then, the cortisol derivative was separated in the ODS analytical column and monitored fluorimetrically. This resulted in good separation of the cortisol derivative peak from contaminants. There was no loss of the cortisol derivative during purification from the C1 column with the solvent S1. This contributed to good reproducibility and recovery of urinary free cortisol. Detection Limits

7

U

\

0

01 a

The detection limit of the cortisol derivative decreased by changing the emission wavelength from 475 (Kabra et al., 1988) to 520 nm. This was caused by altering the pH of the mobile phase from 6.8 (Gotteli et al., 1981; Shihabi et al., 1982) to 1.85 (Nozaki et al., 1991). The HPLC detection limit was 0.26 pg/dL, which was lower than that of the FPIA (about 10pg/dL) and the Gammacoat cortisol radioimmunoassay kit for serum cortisol (about 1.0 pg/dL) (Silver et a/., 1983).

0

v

0

2

50-

I 01

r +

0

Stability of the fluorescence product 50

The FPIA (&dl

100

1

Figure 7. Comparison between urinary cortisol values determined by HPLC and FPIA. The coefficient of correlation was r=0.76. The regression line was y=O.52x+5.4 for n = 4 5 .

The fluorescence intensities of the cortisol derivative decreased quickly at ambient temperature in the previous method (Eechaute, 1966). Gotelli et af. stored cortisol derivative samples at -20°C before HPLC

114

OSAMU NOZAKI E T A L .

assay with manual injection (Gotelli et al., 1981), but failed in consecutive and automatic assays of the samples. For consecutive assays of urinary cortisol, the fluorescence intensity must be stable until all the samples have been assayed, and an autosampler which can store the samples at low temperatures is needed. When samples were stored at ambient temperature or at 10°C in an autosampler, the fluorescence intensities decreased at the same rate as those previously reported. When stored at 4-5°C in the dark, however, the fluorescence intensity of both urinary and authentic cortisol was stable, showing no decrease until 3 h had passed (Fig. 5 ) . Therefore, we used authentic cortisol as an external standard, and all samples were assayed within the time. The cortisol derivative could therefore be assayed automatically and consecutively using HPLC with an autosampler. Precision and comparison with FPIA

The results of both the reproducibility (under 5 % ) and recovery (about 90%) showed that this HPLC method for urinary cortisol would be suitable for clinical application. A comparison of the cortisol values in the patients' urine samples between HPLC and the FPIA with no prior extraction revealed that the coefficient of correlation was r = 0.76, but 5 among the 45 cases in total were seriously dissociated as shown in Fig. 7. Furthermore, the HPLC values were about half those obtained by FPIA. These results were the same as those previously reported (Silver et al., 1983) for similar reasons (Huang and Zweig, 1989; Lantto, 1982), namely cross-reaction

between anticortisol antibody and other steroids coexisting in the samples and interference by other hydrophilic contaminants. The further examination of recovery test for the urine samples with seriously dissociated results will be carried out in the next study.

Automated column switching HPLC

Automated column switching HPLC simplified urinary cortisol assays by direct injection of the derivative samples onto the precolumn, on-line purification and fluorimetry. Therefore, this assay method enabled good reproducibility, recovery and a low detection limit for urinary cortisol. The precolumn C1 withstood 100-200 times sample injections and the analytical column C2 was not damaged during this study. This was due to rapid replacement of strong sulphuric acid in the sample aliquots to the solvent S1 in the C1 column, and endcapping of the ODs-80Tm gel. This HPLC method for urinary cortisol is suitable for clinical application.

Acknowledgements The authors are very grateful to Mitubishi Yuka BCL for providing the urinary samples and the FPIA data on cortisol. The authors also thank Dr. Hiroko Kawamoto of Tottori University College of Medical Care Technology (Yonago, Japan) for her advice during this study.

REFERENCES Beverley, E., Murphy, P., Okouneff, L. M., Klein, G. P. and Ngo, S. C. (1981). J. Clin. Endocrinol. Metab. 53, 91. Canalis, E., Reardon, G. E. and Caldarella, A. M. (1982). Clin. Chem. 28,2418. De Moor, P., Steeno, O., Raskin, M. and Hendrikx, A. (1960).Acta Endocrinol. 33, 297. Diarnandis, E. P. and D'Costa, M . (1988). J. Chromatogr. 426,25. Eechaute, W. (1966). Steroids 8, 633. Gotelli, G. R., Wall, J. H., Kabra, P. M. and Marton, L. J. (1981). Clin. Chem. 27, 441. Huang, C. M. and Zweig, M. (1989). Clin. Chem. 35, 125. Kabra, P. M. (1988). J. Chromatogr. 429,155. Kobberling, J. andvon zur Muhlen, A. (1974).J. Clin. Endocrinol. Metab. 38, 313.

Lantto, 0. (1982). Clin. Chem. 28, 1129. Nakamura. J. and Yakata, M. (1989). Acta Endocrinol. 120, 277. Nozaki, O., Ohata,T., Ohba, Y., Moriyarna, H. and Kato,Y. (1991). J. Chromatogr. 570, 1. Schoneshofer, M., Fenner. A., Altinok, G. and Duke, H. J. (1980). Clin. Chim. Acta 106, 63. Shihabi, 2. K., Andrews, R. I. and Scaro, J. (1982). Clin. Chim. Acta 124, 75. Shipchandler, M. T., Fino, J. R., Klein, L. D. and Kirkemo, C. L. (1987). Anal. Biochem. 162, 89. Si1ver.A. C.. Landon, J.. Smith, D. S. and Perry. L. A. (1983). Clin. Chem. 29, 1869. Sweat, M. L. (1954). Anal. Chem. 26, 773.