| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Articles |
Divisions of Endocrinology of the Department of Pediatrics (G.P.B.K., N.M.D.) and Internal Medicine (R.P.F.D.), Isotopenlaboratorium (J.J.P.), Central Laboratory for Clinical Chemistry (B.G.W.), University Hospital Groningen, 9700 RB Groningen; and Center for High Performance Computing (R.d.B.), University of Groningen, 9700 AV, Groningen, The Netherlands.
Address all correspondence and requests for reprints to: G. P. B. Kraan, Ph.D., van Houtenlaan 206, 9722 GZ, Groningen, The Netherlands.
Abstract
We have measured the urinary cortisol production rate (uCPR)
simultaneously with the serum cortisol production rate (sCPR) in four
healthy men within a period of 3 days. uCPR, determined by isotope
dilution of 11-oxoetiocholanolone was compared with sCPR, which was
measured in three different ways (a, b, c). Blood was sampled at 10-min
intervals for 24 h, and deconvolution analysis was applied to the
cortisol concentrations. The daily serum cortisol production per liter,
multiplied by the distribution volume yielded sCPR. The measurement
methods are characterized as follows: a) the secretion and elimination
terms were free; b) like method a, but with the input of the rate
constants 
and ß into the elimination function; c) the average
24-h cortisol concentration was multiplied by the metabolic clearance
rate. uCPR was 25.4 ± 4.7 [range: 21.3–31.4]
µmol/(m2·day), sCPR (method a) was 28.8 ± 4.5
[range: 23.5–34.3] µmol/(m2·day), sCPR (method b)
was 27.9 ± 8.1 [range: 18.5–37.7]
µmol/(m2·day), and sCPR (method c) was 29.3 ± 4.8
[range: 22.7–33.2] µmol/(m2·day). uCPR did not
significantly differ from each of the 3 sCPR values
(P > 0.30; >0.46; and >0.06, respectively). The
patterns of the cortisol secretory rates in the present and previous
studies do not necessarily represent the physiological process of the
secretory bursts. We conclude that the estimated CPR, being 25–30
µmol/(m2·day) [9–11 mg/(m2·day)], can
serve as a guideline for glucocorticoid replacement dose and that the
urinary route to measure CPR is preferred because of its relative ease.
GLUCOCORTICOID secretion by the human adrenal cortex can be studied by determining CPR.1 Since the original papers on pCPR (1, 2) and uCPR (3, 4), it has been studied intensively (5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17). An early study showed that uCPR amounted to 32.3 ± 5.5 µmol/(m2·day) in adults (5). More recently, pCPR, determined by infusion of deuterated cortisol, was 15.5 µmol/(m2·day), i.e. about 50% of uCPR (14). Because the trideuteryl cortisol to cortisol ratio (14) was as high as 0.25 between 2000 and 0400 h, the presence of extra cortisol during these hours could have led to a decreased secretion of ACTH and, hence, to a lower cortisol secretion (15).
In 1993, a previously described technique (17, 18) was used to measure sCPR (19). Multiple-parameter deconvolution analysis of the measured cortisol concentrations sampled during 24 h (17, 18, 20) led to sCPR/V in normal pubertal males. Multiplication by a fixed Vd of 5.3 L/m2 (21), resulted in sCPR = 15.8 ± 0.9 µmol/(m2·day) (19), again being 50% of uCPR. The secretory pulses of the hormones described (17, 18, 20) were regarded to be symmetrical with a uniform burst half-width, the elimination of cortisol was defined to be monoexponential (18, 19), and all other parameters were considered to be homogeneous for any particular individual time series (17).
The absence of concomitantly measured values of uCPR in the sCPR studies (13, 14, 19), and the supposition that sCPR/V has to be multiplied with a larger and individually determined volume, led us to measure uCPR and sCPR simultaneously within a period of 3 days in four normal adult volunteers. In a parallel study with the same men, we found that, in agreement with previous studies on the plasma cortisol elimination (22, 23), cortisol disappears biexponentially and that the minimal value of Vd ss averaged 7.1 L/m 2 (24). In this report, we describe and compare the results obtained with the method of determining uCPR, and three methods of determining and calculating sCPR in four men.
Materials and Methods
Subjects
Four healthy men, 55 ± 4 [range: 50–60] yr old, recruited from hospital personnel, participated in the study, which was approved by the Medical Ethical Committee of the Academic Hospital Groningen. Written informed consent was obtained from the participants, who were medically checked on normal hematologic parameters and organ functions, and on serum thyroxine and TSH. Their body surface area (25) was 1.93 ± 0.19 [range: 1.69–2.11] m2.
Exp 1
On day 1, radioactive cortisol was injected iv within 10 sec at 0900 h. During the 3 consecutive days, all urine was collected and stored at -20°C until analysis (24).
Exp 2
On day 2, an indwelling catheter was positioned into a forearm vein at 1215 h; and starting at 1230 h, 1-mL blood samples were taken every 10 min during the next 24 h (n = 145). The serum was frozen at -20°C and stored until analyzed. Meals and fluids were taken freely, and lights were dimmed overnight.
Chemicals
Standard steroids were purchased from Sigma Chemical Company (St. Louis, MO). All other chemicals, solvents, and reagents were used as described previously (15, 16).
uCPR
Radioactive cortisol, [1,2,6,7-3H]F was obtained
from New England Nuclear Company (Dupont, MA). Shortly before use, the
tracer was purified by HPLC (silica column; eluent: 5% ethanol, 0.2%
water in dichloromethane). Under these conditions, easily exchangeable
tritium atoms were removed. The SA fell from 86 to 70 Ci/mmol. On day
1, approximately 5 µCi [3H]F was dissolved in 50 µL ethanol plus
10 mL 0.9% NaCl solution, which was then filtered through a 0.2-µm
Millex filter (Millipore Products Div, Etten-Leur, The Netherlands)
into a closed 10-mL vial. At approximately 0900 h, 9 mL of the
[3H]F solution was administered iv within 5 min. The remaining
solution was used for quantitation of the radioactive dose (26).
Radioactivity and creatinine in 1 or 2 mL of all urine samples were
determined, as published previously (26). uCPR was determined, as
described previously (16). In short, after deconjugation of the urinary
cortisol metabolites in the 3-day urine collections, THE, THF, and the
two cortolones (20- and 20ßHHE) were isolated and highly purified
on reversed- and straight-phase HPLC columns. The combined 4 compounds
were oxidized by NaBiO3 to 11-OET. About 25% of THF
contributed to 11-OET (26). 11-OET was intensively purified by HPLC
(16). uCPR, given as µmol/(m2·day), was calculated from
the equation: uCPR = [3 H]F/(SA·t), where
[3 H]F = dose of [3 H]cortisol (Bq),
SA = specific activity of 11-OET (Bq/µmol) and t = time of
the urine collection (days). The UER of cortisol metabolites, given as
µmol/(m2[chempday), was calculated from the sum of the
neutral cortisol metabolites, THE, THF, 5THF, 20- and 20ßHHE,
11-OET, and 11-hydroxyetiocholanolone and -androsterone. These steroids
were measured by gas chromatography in two samples of the 3-day urine
collection (16).
Serum cortisol
Cortisol concentrations were determined by RIA (24, 27). Note that the serum cortisol concentrations are identical to the corresponding plasma cortisol levels (27).
sCPR
sCPR was estimated in three ways. sCPR/V was determined largely
as described previously (18). It is based on the multiparameter
deconvolution analysis (17, 18, 20) to the 24-h cortisol concentrations
in 145 serum samples. In addition, the model was applied more generally
(see the Appendix and Table 1
). sCPR was calculated from the
twice-determined sCPR/V and the corresponding Vd, which
value was derived from the parallel study (24). In the third method,
sCPR was derived from the average 24-h cortisol concentration
multiplied by the MCR (11, 24).
|
The results are presented as the mean ± SD, with the range within square brackets. Two-sided probability values were derived from Student‘s paired t tests.
Results
sCPR
Deconvolution analysis was applied as described in the
Appendix and in Table 1
. The results are given in Tables 2 and 3,
and in Fig. 1, where the 24-h cortisol
profile and two matching, but different, patterns of the cortisol
secretory rate are shown for one volunteer. The secretory burst
frequency was 19 ± 1 per 24 h, with a half-duration of the
skewed bursts (50% of the secreted mass) of approximately 35–40 min.
Table 2 shows that the skewness of the cortisol secretion pattern is
less than 1 (the increase of the cortisol secretion rate is larger than
its decrease). Without predefined conditions of the elimination term in
the convolution integral, i.e. a free search, the mean
half-life of the cortisol decay was 81 ± 18 min, and sCPR/V was
3.44 ± 0.44 µmol/L. Multiplying sCPR/V by the distribution
volume Vd ext yielded sCPR, which was 29 ± 5
µmol/(m2·day) (Table 3, column c). With predefined
conditions of the elimination term, we have analyzed the convolution
integrals three times with the input of the individual ratio (n =
2) of the two rate constants and ß derived from the parallel
study (24). The first time, the half-duration of the skewed bursts was
36 ± 1 min, the mean half-life of cortisol decay,
t1/2 (ß), was 52 ± 10 min and 72 ± 20 min,
and sCPR/V = 3.86 ± 1.24 µmol/L and 4.02 ± 1.02
µmol/L. The 2 half-lives were not significantly different
(P > 0.06), as were the 2 values of sCPR/V
(P > 0.60). The 2 corresponding values of Vd
ss (24) were also equal (P > 0.60). The
resulting equal 2 values of sCPR (P > 0.80) were
combined to be (Table 3, column d) 28 ± 8
µmol/(m2·day). However, repeated (second- and
third-time) analyses with the same values of and ß were
indicative for too-high sCPR/V values: 7.45 ± 2.04 µmol/L
(second time), and 8.26 ± 1.21 and 15.92 ± 8.81 µmol/L
(third time).
|
|
|
, column e) was
29 ± 5 µmol/(m2·day). Table 3 shows that the
three ways to calculate sCPR led to equal values of CPR. Paired
observations showed that sCPRc was not significantly
different from sCPRd (P > 0.70);
sCPRc not from sCPRe (P >
0.70); and sCPRd not from sCPRe
(P > 0.50). Therefore, the individual values of the
three serum production rates were combined to give the average sCPR,
being 28 ± 6 (22–35) µmol/(m2·day). uCPR
Table 3 shows uCPR to be 25 ± 5
µmol/(m2·day). Table 3 also shows that uCPR was not
significantly smaller than each of the three sCPR values. UER was
19 ± 3 µmol/(m2·day), which is 74 ± 5% of
uCPR. The recovery of the neutral urinary cortisol metabolites,
UER/(uCPR·FODfinal), where FODfinal is the
final fraction of dose, 91 ± 4 (88–97) %. The creatinine
excretion was 190 ± 8 µmol/(kg·day).
Discussion
We tested the statement that sCPR provides a better estimate of CPR than uCPR (13, 14, 19) and found that sCPR is not significantly different from uCPR. To our knowledge, it is the first time that sCPR, obtained via deconvolution analysis and distribution volume, has been compared with uCPR and found to be not lower than uCPR. The overall mean sCPR was 28 ± 6 [22–35] µmol/(m2·day), whereas the mean uCPR was 25 ± 5 [21–31] µmol/(m2·day), which might be slightly less than the reported uCPR (5), being 32 ± 6 µmol/(m2·day). Therefore, the present observations strongly challenge sCPR being smaller than and superior to uCPR (13, 14, 19). The main reason for the discrepancy between the reported sCPR, 17 ± 4 and 15 ± 4 µmol/(m2·day) (19), and that in the present study is the fact that the distribution volume we used is 1/3 larger (24). In fact, sCPR/V was not essentially different from earlier reported data obtained with normal adult subjects (20, 28) and also probably not different from those obtained with pubertal males (19).
Vd
sCPR is higher than 16 µmol/(m2·day), mainly
because the parallel study (24) showed that Vd ss was
7.1 ± 1.0 [6.2–8.3] L/m2, being 1.34 times 5.3 L/m
2, the fixed value used earlier (19). Fig. 1a shows that
the cortisol concentration is not in a steady state. Therefore, the
proper Vd should be the one obtained by extrapolation to
t0: Vd ext or V(ß). According to the theory
on the two-compartment open model (29), we found that V(ß) was, on
average, 18% larger than Vd ss (24). Therefore, Vd
ss is the minimal value to calculate sCPR from sCPR/V. The
larger volume, V(ß), was used in case of free elimination conditions
during deconvolution analysis, because only the slow phase of cortisol
disappearance could have influenced the search of the secretory
profile. In case of input of the ratio /ß into the integrals,
Vd ss was used. The recently published value of Vd
ss in five normal women, of 8 ± 1.5 L/m2,
corresponds quite well with our values in healthy men (30).
Application of the model of cortisol secretion simultaneous with its disappearance
In the original papers, the simplest form of the secretory rate
function was used to estimate the secretory bursts of the studied
hormones (20, 31). These bursts of hormones, including cortisol, were
assumed to consist of a series of random Gaussian distributions of
instantaneous secretory rates, i.e. homogeneous and
symmetrical bursts, which only differ in amplitude (20, 31). The
cortisol disappearance was assumed to be monoexponential (18, 19, 28).
In our study (see Table 1), the 24-h curve of the cortisol
concentration profile was fitted to the convolution integrals using a
nonlinear least-squares optimization routine instead of fast Fourier
analysis, as used previously (18). The secretory rate profile of
cortisol was set free, i.e. each pulse could differ in
height, width, and skewness; whereas the cortisol elimination was
defined to be biexponential. Table 2 shows that the secretory burst
frequency was 19 ± 1 per day and 19 ± 2 in a previous study
(18), but because of the skewed secretory rate profile, the half-life
of these bursts was 35–40 min, which is about two times that found
earlier (18).
Daily production rate of cortisol per liter of distribution volume (sCPR/V)
Without predefined elimination conditions, the mean sCPR/V was
3.44 ± 0.44 [3.09–3.85] µmol/L, which is not essentially
different from published values of 3.92 ± 0.94 µmol/L (18),
3.40 [2.20–5.10] µmol/L (28), and 3.17 ± 0.74 and 2.79
± 0.79 µmol/L (pubertal boys) (19). However, in this way, a rapid
phase of cortisol disappearance was not found and could not have been
found. Related to the small half-life of the rapid phase
(t1/2 = 3–4 min), the frequency of six cortisol
measurements per hour was too low to analyze the fast phase, as only
one data point at t = ti + 10 min provided information
on the rapid phase of elimination of that fraction of cortisol, which
was secreted at the previous time point ti. This was in
contrast to all the consecutive data points, which contributed to the
corresponding slow phase. We showed that if the data of disappearing
cortisol were reanalyzed, taking only the values at 10-min intervals
and omitting all other data, the resulting loss of correct values,
especially of the rapid phase, could be partially restored by input of
the known individual ratio /ß into the biexponential elimination
function (24). Therefore, we included the fast phase, during which
period, 5–8% of cortisol disappeared irreversibly (24), by changing
the elimination term of the convolution integral (see
Appendix) into pe-yßt+
(1-p)e-ßt, where y is the individual ratio of the
rate constants and ß, and p and 1-p the
corresponding weight factors (24). Note that the half-life of the slow
phase, t1/2 (ß), in Table 2 does not characterize the
elimination of cortisol adequately because of these missing weight
factors p and 1-p.
With predefined elimination conditions, i.e. with the input
of these individual ratios of and ß, the mean sCPR/V first was
3.9 and 4.0 µmol/L, respectively. Note that the mean sCPR/V values
from repeated analyses were too high. Apparently, the input of the
ratio /ß, combined with the free secretory rate conditions,
sometimes led to wrong output because of numerical complexity and a
suboptimal solution, and correction of the integrals with data on a
10-min-interval basis did not always yield reliable results. To our
point of view, the latter incorrect results do not invalidate the way
we used the model of deconvolution analysis. Rather, the latter outcome
refers to the complex nature of concurrent secretion and elimination,
which analysis leads to reliable results only if the functions of the
secretory rate and elimination are defined to be restricted and simple,
as reported previously (18, 19, 28). Is a perfect reconvolution fit of
the cortisol profile a sufficient proof for the real secretory profile
and the correct value of sCPR/V? Mathematically, the two functions of
the secretion rate and elimination are interrelated and not
independent. In general, the more parameters the model function
c contains, the better c fits to the data. This
does not imply that the model is relevant in a biological sense;
e.g. if fi (t) =
(t-ti ) and e(t) = (t),
where is the Dirac (spike) function and
ti are the time points at which the data are
collected, then the model will fit perfectly to the data, but its
biological meaning is nil. The same objection holds if
fi are described by very narrow Gaussians with
fast elimination. Fig. 1 shows two significantly different secretory
profiles, which were obtained under the above stated different
conditions. Because both cortisol secretory profiles differ markedly
from those in earlier reports (18, 19, 28), their physiological
significance is low, and hence, that of the published secretory
profiles, because of the assumption of the restricted conditions of the
secretory and the eliminations functions (18, 19, 28).
Daily production rate of cortisol (sCPR and uCPR)
It has been concluded in the past (10), and quoted by others (13, 19), that because of unequal labeling of urinary cortisol metabolites,
uCPR would be of limited value. The procedure of three cases of
simulated cortisol production by adrenalectomized patients (10) led to
significant differences collecting the cortisol metabolites in a 3-day
urinary pool, compared with that in normal humans. For instance, in
case 1, the urinary pool was not a quantitatively complete collection
of each of the cortisol metabolites (10) and, because of the low or
zero cortisol blood levels at the start of dosing of the label in the
cases 1 and 2, too much of the radioactive dose was oxidized into
11-oxo metabolites (10). If, instead of taking the arithmetic mean CPR
from the six used metabolites (10), uCPR is calculated by taking the
weighed mean uCPR from THE, THF, and the combined two cortolones
(+ßHHE), the main 5ß metabolites, which are roughly present in a
ratio 2:1:1, one may find the same published values. Therefore, we used
the method of the weighed mean by determining the specific activity of
11-OET, which was obtained by combined oxidation of THE, THF, and
+ßHHE (16). In another paper, it was assumed that uCPR slightly
overestimates (but pCPR slightly underestimates) the true CPR, with an
error of less than 25% (11). However, in the latter study, uCPR was
obtained using the specific activity of the chemically derived
11-hydroxyetiocholanolone, largely as described originally (9). As
suggested (10), this leads to overestimation of uCPR, indeed.
In conclusion, we report that the daily production of cortisol is 25–30 µmol/(m2·day) or 9–11 mg/(m2·day), and that no significant difference exists between the urinary and the serum values of CPR in men. This estimation of CPR may serve as a guideline for glucocorticoid supplementation therapy. For instance, in children with congenital adrenal hyperplasia, ultimate height can be improved with a cortisol dose of 12 mg/(m2·day), compared with a higher dose (32); and adult men with Addison’s disease, who have been treated for at least 10 yr with a dose of 13.6 mg/(m2·day), showed normal bone mineral density (33). Those treated with higher doses showed inverse correlation between the daily dose and the bone density (33). The urinary route to measure CPR, as described here and elsewhere (15, 16), is simpler and less burdening than that of 24-h blood sampling. Furthermore, the method of deconvolution analysis yields sCPR only under restricted and simple conditions and when a separately determined value of the distribution volume is known. Therefore, this study revalues the use of the urinary method to achieve CPR.
Appendix
Model description
The secretion function s(t) is assumed to consist of a
number of bursts with a daily period, i.e. s(t+1) =
s(t), and an elimination function e(t), which is
monotonously decreasing with e(0) = 1 and e(
) = 0. These
simultaneously operating functions are used in the convolution process.
The convolution integral then reads the plasma concentration
c(t) (20):
 |
The functions s and e can be chosen at will, provided they are physiologically significant.
Usually s is expressed as:
 |
where fi (t) 
0 and
 |
i.e.
 |
is the daily production per liter and s is the weighed sum of m
bursts. The nature of a burst is that its value is positive for a
limited period of time, and zero elsewhere, and it has one maximal
value. In this report a modeling of fi is used by means of
basic spline approximation. Each burst may have its own shape and
placing in time. The elimination function e(t) is chosen to
be monoexponential or biexponential of nature:
be-ßt or ae-t+ be-ßt, respectively, where and
ß are the rate constants, and a and
b corresponding concentrations at the time of the considered
c(t).
Approximation of the model parameters
The model function c contains a number of parameters that can be approximated with some optimization technique, e.g. nonlinear least-squares optimization. Because, in this study, specific conditions (such as uniformity of all bursts) were not appropriate, a regular but computationally intensive optimization was used, which lasted about 20 h on a Cray-J32 computer per set of 145 data points.
Acknowledgments
Thanks are due to Clasine Venker, Jan van der Molen, and Gerrie Stob for their assistance in sample treatment. We are grateful to Mr. K. L. Nijdam for his expert technical assistance in the liquid chromatographic separations of the steroids.
Footnotes
1 Presented, in part (P2–428), at the 10th International Congress of
Endocrinology, San Francisco, CA, 1996. 
Received August 19, 1997.
Revised November 7, 1997.
Accepted December 9, 1997.
References
This article has been cited by other articles:
 |
|
 |
|
  H. Filipsson and G. Johannsson GH replacement in adults: interactions with other pituitary hormone deficiencies and replacement therapies Eur. J. Endocrinol., November 1, 2009; 161(S1): S85 - S95. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. Johannsson, R. Bergthorsdottir, A. G Nilsson, H. Lennernas, T. Hedner, and S. Skrtic Improving glucocorticoid replacement therapy using a novel modified-release hydrocortisone tablet: a pharmacokinetic study Eur. J. Endocrinol., July 1, 2009; 161(1): 119 - 130. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. A. Wudy, M. F. Hartmann, and T. Remer Sexual dimorphism in cortisol secretion starts after age 10 in healthy children: urinary cortisol metabolite excretion rates during growth Am J Physiol Endocrinol Metab, October 1, 2007; 293(4): E970 - E976. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Heckmann, M. F. Hartmann, B. Kampschulte, H. Gack, R.-H. Bodeker, L. Gortner, and S. A. Wudy Cortisol Production Rates in Preterm Infants in Relation to Growth and Illness: A Noninvasive Prospective Study Using Gas Chromatography-Mass Spectrometry J. Clin. Endocrinol. Metab., October 1, 2005; 90(10): 5737 - 5742. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. M. Keenan, F. Roelfsema, and J. D. Veldhuis Endogenous ACTH concentration-dependent drive of pulsatile cortisol secretion in the human Am J Physiol Endocrinol Metab, October 1, 2004; 287(4): E652 - E661. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. J. Inder and P. J. Hunt Glucocorticoid Replacement in Pituitary Surgery: Guidelines for Perioperative Assessment and Management J. Clin. Endocrinol. Metab., June 1, 2002; 87(6): 2745 - 2750. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
N. Picard-Hagen, V. Gayrard, M. Alvinerie, H. Smeyers, R. Ricou, A. Bousquet-Melou, and P. L. Toutain A nonlabeled method to evaluate cortisol production rate by modeling plasma CBG-free cortisol disposition Am J Physiol Endocrinol Metab, November 1, 2001; 281(5): E946 - E956. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. G. Langendonk, J. D. Veldhuis, J. Burggraaf, R. C. Schoemaker, A. F. Cohen, A. Edo Meinders, and H. Pijl Estimation of growth hormone secretion rate: impact of kinetic assumptions intrinsic to the analytical approach Am J Physiol Regulatory Integrative Comp Physiol, January 1, 2001; 280(1): R225 - R232. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Bergendahl, A. Iranmanesh, C. Pastor, W. S. Evans, and J. D. Veldhuis Homeostatic Joint Amplification of Pulsatile and 24-Hour Rhythmic Cortisol Secretion by Fasting Stress in Midluteal Phase Women: Concurrent Disruption of Cortisol-Growth Hormone, Cortisol-Luteinizing Hormone, and Cortisol-Leptin Synchrony J. Clin. Endocrinol. Metab., November 1, 2000; 85(11): 4028 - 4035. [Abstract] [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
