Cbemistry

Clinical 42, No. 6, 1996 979

show around 8-10-fold higher sensitivities for both the intact drome. The assay of other steroids such as 11 -deoxycortisol-
and des-3 1,32-proinsulin assays compared with thosepossible in useful in the 11 (3-hydroxylasedeficiency form of congenital
methods utilizing the aminocellulose-bound antibody, despite adrenal hyperplasia [1]and after giving metyrapone or ketocon-
improvements in sensitivity in the latter methods. The assays azole, which blocks cortisol synthesis [2]-prednisolone and
can measure fasting concentrations of these molecules in non- methylprednisolone (for elucidating pharmacokinetics in organ-
diabetic subjects as well as in clinical conditions where their transplant patients and people with asthma), and dexamethasone
concentrations are higher [6]. In the intact proinsulin assay the (which is metabolized in Liddle’s test) is also of importance.
cross-reactivity with des-64,65-promnsulin was slightly higher These steroids can all be measured separately by RIA but not
than published for the cellulose IRMA (78% vs 66%, respec- without significant potential for cross-reactivity [3]. HPLC
tively). On the other hand, in the des-31,32-proinsulin assay the methods have been described for these steroids, but no tech-
cross-reactivity with intact proinsulin was lower (59% vs 84%, nique has yet been described for the analysis of all these steroids
respectively), whereas that with des-64,65 was similar(54% and in one run. We have developed an HPLC procedure with
60%, respectively). fludrocortisone asan internalstandard that both simultaneously
This methodology is very easy to set up and allows the measures these steroids at clinically useful concentrations and
processing of large numbers of samples. It also lends itself to eliminates the problem of cross-reactivity.
automation. The detection system can be modified by replacing We purchased cortisol, prednisolone, methylprednisolone,
the radiolabel with an enzyme label. This may be particularly 11-deoxycortisol, dexamethasone, and fludrocortisone from
relevant for the des-64,65-proinsulin assay, for which improve- Sigma Chemicals (St. Louis, MO). Tetrahydrofuran from
ments in sensitivity are necessary. Merck (Sydney, Australia) was of HPLC grade, and water was
The comparison of results obtained by the microplate IRMA processed by a water purification system (Nanopure Barnstead,
with those by the cellulose IRMA and HPLC methods, and in Sydney, Australia). All other chemicals were of analytical re-
fastingas well as stimulated samples, shows good correlation. agent grade.
Some of the discrepancies may be attributable to differing Stock calibration solutions (500 molil) of cortisol and
lengths of storage of the samples: The microplate IRMA was fludrocortisone were prepared in methanol. The working inter-
carried out a year after the cellulose IRMA and HPLC assays. nal standard solution, prepared by diluting the stock fludrocor-
The effects of prolonged freezing on microplate IRMA results tisone solution 200-fold with mobile phase, was stored at 4 #{176}C.
are not known, but do not appear to have adversely affected the Urine cortisol calibrators were prepared from a urine sample in
correlation. which the concentration of endogenous cortisol had been
In conclusion, this microplate IRMA for proinsulin and determined by the use of mobile phase-based calibrators. The
des-3 l,32-proinsulin is simple and sensitive and appears to be working solutions of the urine-based calibrators were prepared
applicable for use with large numbers of samples. by adding the appropriate volume of the stock solution to this
urineto give concentrations of 500 and 1500 nmoVL; aliquots (2
References mL) were then stored frozen at -20 #{176}C until analyzed. Calibra-
1. Temple RC, Clark PMS, Nagi DK, Schneider AE, Yudkin JS, Hales CN.
Radioimmunoassay may overestimate insulin in non-insulin-dependent dia- tors for the other steroids were prepared in similar manner.
betics. Clin Endocrinol 1990;32:689-93. For HPLC we used a Waters (Milford,MA) Model 712
2. Dhahir F, Cook DB, Self CH. An amplified enzyme-linked immunoassay for WISP autosampler, a Model 510 solvent-delivery system, and a
human proinsulin. Clin Chem 1992;38:227-32.
3. Lundkjerns L. R#{248}der
ME, Dinesen B, Hartling 5G. J#{248}rgensen
PN, Binder C.
Model 481 UV/VIS detector. The system was controlled by
Highly sensitive enzyme immunoassay of proinsulin immunoreactivity with Millennium 2010 Chromatography Manager (Waters). The
use of two monoclonal antibodies. Clin Chem 1993;10:2146-50. analytical column was 250 X 4.6 mm (i.d.) Spherex C18 5-gm
4. Sobey WI, Beer SF, Carrington CA, Clark PMS, Frank BH, Gray IP, et al.
Sensitive and specific two-site immunoradiometric assays for human insu-
particles Phenomenex (Torrance, CA). The mobile phase was
lin, proinsulin, 65-66 split, and 32-33 split proinsulin. Biochem J 1989; methanol, tetrahydrofuran, and water (3:25:72 by vol), which we
260:535-41. filtered and degassed by passing the mixture through a 0.45-sm
5. Ostrega D, Polonski K, Nagi DK, Yudkin J, Cox U, Clark PMS, Hales CN.
Measurement of proinsulin and intermediates: validation of immunoassay
pore-size Millipore (Bedford, MA) filter before use. The system
methods by high-performance liquid chromatography. Diabetes 1995;44: was run isocratically at a flow rate of 1.0 mL/min and the
437-40. absorbance peaks were detectedat 254 nm. Total analysis time
6. Mohamed-Ali V. Gould MM, Goubet S. Yudkin JS. The relationship between
for allsteroids was 32 mm.
insulin, intact proinsulin and des 31,32 proinsulin and other cardiovascular
risk factors in non-diabetics [Abstract]. Diabet Med 1993;10(Suppl 3):P54. To analyze serum samples, we added 500 j.tL of 0.2 moltL
acetate buffer (pH 3.85) to 1 mL of serum before extraction;
urinerequired no pretreatment. Fludrocortisone (2.5 p.moVL in
mobile phase) was added (400 L) to all samples before extrac-
tion. If sample volume was limited (e.g., pediatric), we used a
Improved HPLC Method for Simultaneous Analysis of smaller sample volume with a corresponding reduction in the
Cortisol, I 1-Deoxycortisol, Prednisolone, Methylpred- amount of internal standard added. For example, reducing
msolone, and Dexamethasone in Serum and Urine, BrettC. sample volume to 0.5 mL increases the detection threshold to
McWhinney, * Gregory Ward, and Peter E. Hickman (Dept. of <20 nmolil, and the CV increases from 4.6% to 7.1%. We
Chem. Pathol., Princess Alexandra Hosp., Woolloongabba, then centrifuged the serum and urine samples and extracted the
QLD, Australia 4102; *author for correspondence: fax 61 7 steroids with Sep-Pak Vac 3-mL (500 mg) C18 cartridges
32407070, e-mail PHickman@gpo.pa.uq.edu.au) (Waters), which had been preactivated with 3 mL of methanol
followed by 3 mL of water. After the cartridges were positioned
Assay of steroid hormones is a major part of any general on a 24-port vacuum elution manifold (Waters), we washed
endocrinology laboratory. Routine analysis of serum and urine them under reduced pressure with 3 mL of each in the following
cortisol is used in the differential diagnosis of Cushing syn- order: acetone:water (20:80 by vol), water, and hexane. The
980 Technical Briefs

steroids were subsequently eluted with 3 mL of diethyl ether were 0.21% and 0.22%, respectively. The retention-time ratios
into 16 x 100 mm glass tubes (Corning Glass Works, Corning, for selected corticosteroids to fludrocortisone was calculated by
NY) containing 1 mL of 0.2 mol/L NaOH. The tubes were then the Millennium software and was used as a secondary peak
capped, and their contents were vortex-mixed and centrifuged. identification technique.
The diethyl ether layer was transferred to another glass tube Correlation studies on both urine and serum cortisol were
with glasspipettes, dried under nitrogen, reconstituted in 250 performed against the Incstar Clinical Assays (Stillwater, MN)
p.L of mobile phase, and placed on a rotary mixer for 5 min.We GammaCoat Cortisol method (urinary free cortisol with dichlo-
then injected 60 .tL of the reconstituted sample intothe HPLC romethane extraction). Results of regression analysis were as
system. The retention times for prednisolone, cortisol, methyl- follows:
prednisolone, 1 l-deoxycortisol, and dexamethasone were 12.00, urine HPLC = 0.57 RIA - 4 (n = 46, r2 = 0.897)
12.85, 21.00, 22.10, and 29.85 mm, respectively. Fig. 1 is a serum HPLC = 1.09 RIA - 40 (n = 30, r2 = 0.950)
chromatogram of a serum sample supplemented with 500 One explanation for the lower urine HPLC comparison
nmol/L of the steroidsof interest. cortisol values is that RIA detects both cortisol and other
The intra- and interassay (20 runs over 3 months) precision structurally related compounds. Our current experience sug-
studies for the different corticosteroids are shown in Table 1. In gests that peaks that elute within the 9-21 mm range (cortisol,
analytical recovery studies performed on 30 urines with 500 12.85 mm) cross-react in the RIA method (prednisolone 77%,
nmol/L of added cortisol, recovery was 97.7% ± 4.22% (mean 6-methylprednisolone 43%, fludrocortisone 33%, and il-de-
± SD). The linearity of the method, evaluated by analyzing a oxycortisol 6%, according to the GammaCoat Cortisol kit
urine with increasing amounts of added cortisol, extended over insert). Because the urinary free cortisol results by HPLC were
the range of 25-3000 nmol/L for cortisol (linear regression lower than those by RIA, we have had to determine new
coefficient of r2 =0.995). Similar results were obtained for referenceranges for HPLC analyses. Preliminary ranges from
prednisolone, 11 -deoxycortisol, methylprednisolone, and dexa- 50 healthy laboratorypersonnel (ages 18-55 years; 25 men, 25
methasone. The detection limit for all analytes (signal-to-noise women) were (<200 nmoll24 h) for 24 h urinary cortisol and
ratio of 3:1) was 5 nmolfL. The lowest concentration for each <25 nmol/mmol for the random early morning cortisol/creat-
analyte that gave an interassay CV of s20% was 10 nmol/L over mine ratio, the latter aiding in the diagnosis of cyclical Cushing
20 separate runs. disease [4].
We included two urine-based cortisol calibrators (500 and The HPLC method appears useful for the simultaneous
1500 nmoVL) with each run. Although the assay was linear to at analysis of 11 -deoxycortisol and cortisol in patients before and
least 3000 nmol/L, any sample with a concentration exceeding after metyrapone administration. In one representative patient,
that of the highest calibrator was diluted and rerun. The before metyrapone treatment, the agreement between RIA (680
Sep-Pak Vac Cartridges can be used repeatedly for as many as 10 nmollL) and HPLC (570 nmoVL) cortisol results was consistent
times by washing with 3 mL of tetrahydrofuran between runs. with our above-stated correlation data. After the metyrapone
The retention-time instabilities (CVs) over a run of 20 administration, however, the expected decrease in cortisol asso-
samples for cortisol (12.78 mm) and fludrocortisone (15.90 mm) ciated with an increased 1l-deoxycortisol (528 nmollL) was
observed in the HPLC-generated results(390 nmolfL), whereas
the measured RIA cortisol concentration (1050 nmol/L) was
increased. We observed this artifactual increase in RIA cortisol
5 0 in allpatients (n = 12) who received metyrapone. Because the
35.0 o_: ‘-I
o overnight metyrapone test has recently been suggested as an
5 5 0
I1J
.1
jIi 0 ,.
alternative to either Synacthen stimulation or insulin hypogly-
Ok cemia for assessment of the hypothalamic-pituitary-adrenal axis
30.0 o_: ‘-40 o - #{149}
ri -i
[2], we and others [5, 6] point out that RIA cortisol results might
_I 14 14
#{149}. ‘ n.
0
o be misleading if used as part of this test.
‘-I
25 .0 0_ o , 0
S 0 0
Table 1. Intra- and Interassay Imprecision for selected

Ift
-I Ii ‘d
0, 0
0 S glucocorticosterolds In two In-house-prepared controls.
20 .0 0_
Intraassay Interassay

14 Analyte Mean CV Mean CV

0
nmol/L nmol/L
Cortisol 59 3.0 61 10.5

5.00 _
J Prednisolone

11-Deoxycortisol
2595
95
964
51
312
2.9
3.1
2.8
4.3
3.9
2635
96
953
49
318
4.6
4.5
3.2
7.6
5.5
0. 0 20.00
Minute 8 Methylprednisolone 91 5.6 93 9.5
905 4.1 896 5.0
Fig. 1. Chromatogram of a serum supplemented with 500 nmol/L each
of prednisolone, cortisol, methylprednisolone, 11-deoxycortisol, and Dexamethasone 52 6.8 55 10.0
dexamethasone. 310 5.4 320 7.8
Fludrocortisone (IS) is the internal standard. n = 20 each.
 Clinical Chemistry 42, No. 6, 1996 981

In summary, the present method allows the separation of tions might be helpfulin improving the accuracy of detectionof
several important steroids simultaneously and resolves cortisol, these proteins [12].
prednisolone, and methylprednisolone sufficiently to allow their All steps of sample preparation devisedin our laboratory were
accurate determination.The procedure is straightforward and carried out at 4 #{176}C, and all reagents were purchased from Sigma
robust, and therefore is suitable for any laboratory interested in Chemical Co. (St. Louis, MO). We used tissues excised from
the analysis of these steroids.Assays are usually run overnight patients with benign tumors (n = 13)-serous cystadenoma (n =
forconvenience, but they can be completed within4 h if needed, 3), benign ovariancysts (n = 4), endometrial cysts (n = 2), and
which is quite sufficient for the clinical decision-making re- mucinous cystadenoma, sclerosing stromal tumor, cystic tera-
quired for diagnosis of patients with congenital adrenal hyper- toma, and thecoma (n = 1 each)-and from patients with the
plasma. most commonly seen malignant ovarian tumor serous cystade-
nocarcinoma (n = 18). Tissues were immediately washed in
ice-cold saline and homogenized on ice in 10 mmol/L Tris
References buffer EpH 7.5; containing 100 mLIL glycerol, 0.4 mollL KCI,
1. Miller WL. Congenitaladrenal hyperplasia. Endocrinol Metab Clin North Am 10 mmol/L EDTA, 5 mmol/L benzamidine, 10 mmol/L 2-mer-
1991:20:721-50.
2. Fiad TM, Kirby JM, Cunningham SK, McKenna TJ. The overnight single-dose
captoethanol, 0.39 mmollL phenylmethylsulfonyl fluoride
metyrapone test is a simple and reliable index of the hypothalamic-pituitary- (PMSF), and 5 mg/L aprotinin] with an Ultraturax T-25
adrenal axis. Clin Endocrinol 1994:40:603-9. homogenizer for three bursts of 55 s each separated by a pause
3. Young OS, ed. Effects of drugs on clinical laboratory tests, 3rd ed. for 1 mm. The homogenate was filtered and then centrifuged at
Washington: AACC Press, 1990:117.
4000g for 15 mm with a Beckman Instruments (Brea, CA)
4. Atkinson AB, Kennedy AL, Carson Di, Hadden OR, Weaver JA, Sheridan B.
Five cases of cyclical Cushing’s syndrome. Br Med J 1985:291:1453-6. CS-oR centrifuge to obtain the nuclear pellet. The pellet was
S. Reardon GE, Caldarella AM, Canalis E. Determination of serum cortisol and dissolvedin phosphate-buffered saline and sonicated for three
11.deoxycortisol by liquid chromatography. Clin Chem 1979:25:122-6. bursts of 30 s each.Afterquantifying the proteinconcentration
6. Canalis E, Caldarella AM, Reardon GE. Serum cortisol and 11.deoxycortisol in the pellet by using Bradford’s method [13] with bovine serum
by liquid chromatography: clinical studies and comparison with radioimmu-
noassay. Clin Chem 1979:25:1700-3.
albumin as the calibrator, we applied the samples directly to
ELISA plate wells for assay of p53 with an EIA kit from
Oncogene Science (Uniondale, NY).
The supernatant fluid remaining after the above centrifuga-
tion was recentrifuged at 25 000g for 1 h with a Beckman L7
ultracentrifuge to obtainthe cytosol; the membrane fraction was
Quantitative Measurement of c-erbB-2 p185 and Mutant obtained by redissolving the pellet obtained in this second
p53 Expression in Ovarian Neoplasms by Enzyme Immu- centrifugation in 10 mmol/L HEPES buffer (pH 7.5; containing
noassay, Garnal M. Mabrouk, * Sahar A.A. He/al, KhalilI. El- 10 mmol/L EDTA, 5 mmolfL benzamidine, 10 mL/L Triton
Lamie,’ and A/i Khalifa (Oncology Diagnostic Unit, Dept. of X-100, 10 mmol/L 2-mercaptoethanol, 0.39 mmolfL PMSF,
Biochem. and ‘Dept. of Gynecol. & Obstet., Am Shams Faculty and 5 mg/L aprotinin) and recentrifugingat the same speed for
of Med., Cairo, Egypt; *author for correspondence: fax 20-2- an additional 30 mm. After measuring the protein concentration
2 85-9928) in the membrane fraction, we diluted the samples in sample
diluent provided by the kit and analyzed them for c-erbB-2 pl85,
Ovarian cancer is currently considered a leading cause of death using another EIA kit from Oncogene Science.
among gynecologicalmalignanciesin the US [1,2].In Egypt, We observed an overexpression of c-erbB-2 and mutant p53
according to the National Cancer Institute records,ovarian proteinsin malignant cases compared with benign ones (Fig. 1).
cancer is the second most common malignancy in gynecological The respective cutoff values for these analytes-0.5 fmol/g
tumors (0.8%) after cervical cancer [3].Prognostic markers (nmol/g) protein and 1.8 fmol/mg (mol/g) protein-repre-
c-erbB-2 p185 and p53 are overexpressed in ovarian malignan-
cies, as shown by immunohistochemical studies [4-8]. Most of
these studies generally require tedious and time-consuming 25 12
preparationsof samples and must be interpreted by experienced 3
a,
personnel. Some other studies have quantified mutant p53 0
20 0 0
protein by immunofluorometry in sera, breast cancer tissue 0
cytoplasmic extracts, malignant cell lines [9], and ovarian cancer C)

[10];however, results are expressed in arbitrary units without 0 0

8
15
referral to the expression level of p53 in benign tumors. Other E
investigators have reported a high expression of c-erbB-2 in 14)
malignant breast tissue whole homogenate, as quantified by 10
0 0
enzyme immunoassay (EIA) [11].
In this study we have tried to develop an easierand less
time-consuming technique that would enable us to have cutoff
values for p53 and c-erbB-2 assayed in their subcellular
tions, i.e., nucleus and membrane, respectively.
loca-
The use of such
c.J

P
0
5

0
6
oI
numerical valuesforlaboratory diagnosticand prognostic assays
would increase their sensitivities. Because EIA is considered an Benign Malignant Benign Malignant
easy routine laboratory test,we thought that itsuse in detection Fig. 1. Expression of c-erb8-2 and mutant p53 in ovarian neoplasms at
of onco gene and suppressor gene products in subcellular frac- stage I(n = 4; A) and stages II & Ill (n = 14; 0).