We studied healthy men who underwent blood sampling for plasma
nandrolone, testosterone and inhibin measurements before and for 32 days after a single i.m. injection of 100 mg of nandrolone ester in
arachis oil. Twenty-three men were randomized into groups receiving
nandrolone phenylpropionate (group 1, n = 7) or
nandrolone decanoate (group 2, n = 6) injected into
the gluteal muscle in 4 ml of arachis oil vehicle or nandrolone
decanoate in 1 ml of arachis oil vehicle injected into either the
gluteal (group 3, n = 5) or deltoid (group 4, n = 5) muscles. Plasma nandrolone, testosterone and
inhibin concentrations were analyzed by a mixed-effects indirect
response model. Plasma nandrolone concentrations were influenced
(P < .001) by different esters and injection sites, with higher
and earlier peaks with the phenylpropionate ester, compared with the
decanoate ester. After nandrolone decanoate injection, the highest
bioavailability and peak nandrolone levels were observed with the 1-ml
gluteal injection. Plasma testosterone concentrations were also
influenced (P < .001) by the ester and injection site, with the
most rapid, but briefest, suppression being due to the phenylpropionate
ester, whereas the most sustained suppression was achieved with the
1-ml gluteal injection. Plasma inhibin concentrations were also
significantly influenced by injection volume and site, with the lowest
nadir occurring after the nandrolone decanoate 1-ml gluteal injection.
Thus, the bioavailability and physiological effects of a nandrolone
ester in an oil vehicle are greatest when the ester is injected in a
small (1 ml vs. 4 ml) volume and into the gluteal
vs. deltoid muscle. We conclude that the side-chain
ester and the injection site and volume influence the pharmacokinetics
and pharmacodynamics of nandrolone esters in an oil vehicle in men.
 |
Introduction |
For decades, administration of
androgens such as testosterone and 19-nor-testosterone has been most
frequently via depot i.m. injections of steroid esters
dissolved in a vegetable oil vehicle (Junkman, 1957
; Behre et
al., 1990). Such i.m. injections provide sustained androgen
release into the circulation and have remained the mainstay of androgen
replacement therapy for the last few decades (Nieschlag and Behre,
1990), although the basic pharmacological mechanisms are complex and
only partially understood (Zuidema et al., 1988). The basic
pharmacology of this depot androgen formulation differs among species
(van der Vies, 1965) but has been little studied in humans. The current
understanding is that the rate-limiting mechanism governing the
appearance of active steroid in the bloodstream is the retention of
steroid esters from the oil vehicle depot due to oil/water
partitioning, with gradual release into the extracellular fluid, where
esters are rapidly hydrolyzed to liberate biologically active steroid.
Other physiological and physico-chemical factors that could influence
steroid appearance in the bloodstream include the chemistry of the
side-chain ester (hydrophobicity, steric hindrance of hydrolysis and
solubility), injection factors (depth, site and volume, pH and
osmolarity of the solution), exercise and systemic illness. The
influence of site and volume of injection on the release kinetics of
androgen esters from oil vehicle depots has, however, not been
systematically investigated in humans.
This study compared the pharmacokinetics and pharmacodynamics of two
currently available esters of nandrolone, the decanoate and
phenylpropionate, as well as the influence of i.m. injection sites
(gluteal vs. deltoid) and injection volumes (4 ml
vs. 1 ml). In addition to measuring plasma nandrolone to
investigate pharmacokinetics, we measured plasma testosterone and
inhibin by radioimmunoassay to determine the pharmacodynamic effects of nandrolone-induced inhibition of pituitary gonadotrophin secretion, as
reflected in LH-dependent Leydig (testosterone) and FSH-dependent Sertoli (inhibin) cell function in healthy men. We analyzed these data
using an indirect pharmacodynamic response model, which has demonstrated, for the first time, prominent pharmacological differences between esters differing in only a single carbon in the side-chain, as
well as systematic differences attributable to injection site and
volume in humans.
| |
Materials and Methods |
Experimental design.
Twenty-three healthy volunteers were
randomly allocated into four groups in two balanced blocks. The first
stratum of the study involved comparing two different nandrolone esters
while controlling for dose, injection site and volume. In this stratum, volunteers received either nandrolone phenylpropionate (Durabolin; Organon) (group 1) or nandrolone decanoate (DecaDurabolin;
Organon) (group 2), administered as a single deep i.m. injection of 100 mg nandrolone ester into a single injection site (gluteal) in a fixed
volume (4 ml of arachis oil vehicle). In the second stratum, a single
ester (nandrolone decanoate) with fixed dose (100 mg) and volume (1 ml
of arachis oil vehicle) was injected into two different sites, gluteal
(group 3) or deltoid (group 4) muscles. This design also allowed
comparison between different injection volumes (4 ml vs. 1 ml) for a single nandrolone ester with fixed dose and injection site
(i.e., group 2 vs. group 3).
Subjects.
Healthy nonobese men, 18 to 40 years of age, with
normal reproductive function, not competing in sports requiring
International Olympic Committee-sanctioned urinary drug screening for
anabolic steroids, not allergic to peanuts, free from chronic medical
illness, not requiring regular prescribed medication and without
abnormalities in routine clinical examination and biochemical screening
were recruited. Fully informed written consent was obtained from
volunteers, and the study had approval from the University of Sydney
Human Ethics Committee within National Health and Medical Research
Council Guidelines on Human Experimentation, which conform to the
Declaration of Helsinki.
Procedures.
Blood (5 ml) was sampled before injection
(8:00-9:00 A.M.) and at 1, 2, 4, 6, 8 and 10 hr and 1, 2, 3, 4, 7, 9, 11, 14, 16, 18, 21, 23, 25, 28, 30 and 32 days after
injection. During the first 10 hr, the participants were in a metabolic
ward facility and were allowed to undertake normal daily activities.
Subsequently, participants returned for blood sampling on an ambulatory
basis while resuming their habitual daily activities.
To calculate absolute bioavailability, two additional subjects received
a single i.v. bolus of 1 mg of 19-nor-testosterone, with sampling
before injection and at 2, 4, 6, 8, 10, 15, 20, 30, 45, 60, 120, 180,
240, 360, 480, 720, 900, 1440, 1980 and 2880 min. For this study,
crystalline nandrolone (Steraloids Inc., Wilton, NH) was dissolved in
dehydrated ethanol BP (David Bull Laboratories, P/L, Mulgrave,
Australia) at a concentration of 40 g/liter, from which a nandrolone
stock solution (2 mg/ml in 15% ethanol) was produced by appropriate
dilution with sterile saline. The nandrolone stock was filtered through
a 0.2-µm cellulose acetate filter (Minisart NML; Sartorius GmBH,
Gottingen, Germany), and the first 1 ml was discarded. Preliminary
experiments showed that >80% of tracer nandrolone was recovered after
such filtration. At the start of the study, the stock solution was
diluted 10-fold with sterile normal saline in a 20-ml syringe (final
ethanol concentration, 1.5%) and 5.0 ml was infused rapidly (within 10 sec) via an indwelling cannula in a forearm vein (nominal
dose, 1.0 mg), followed by a flush of the syringe and cannula with an
additional 5 ml of sterile saline solution. Aliquots of the injection
solution were kept for subsequent assay of the exact amount of
nandrolone administered. Blood sampling was from another cannula placed
in the other arm.
Assays.
Plasma samples were stored at 
20°C until assay
in a single batch per analyte. Steroids were measured by
radioimmunoassay after extraction from plasma by a modification of a
solid-phase method (Vining, 1980). Serum (200 µl for the nandrolone
assay and 75 µl for the testosterone assay) was applied to a 5-cm
mini-column of Extrelut (kieselguhr; Merck, Kilsyth, Australia) packed
in a Pasteur pipette. Steroids were eluted by four washes of 750 µl
of hexane/ethyl acetate (3:2) at 5-min intervals, the combined eluate
was dried and the extract was reconstituted in assay buffer. Extraction
efficiency was 97 ± 5% (n = 29) for testosterone
and 90 ± 5% (n = 25) for nandrolone; therefore,
no corrections were made for extraction losses. Radioactivity was
measured with a Wallac 1410 liquid scintillation counter (Wallac Oy,
Turku, Finland) and a LKB 1261 Multigamma gamma counter (Wallac Oy),
using RIACALC data reduction software on an IBM-compatible computer.
All steroid assay reagents were of analytical or higher grade.
Plasma nandrolone was measured by radioimmunoassay using a rabbit
antibody to 19-nor-testosterone 17-hemisuccinate-bovine serum albumin
(CER, Marloie, Belgium), tritiated tracer
(19-[19-3H]nor-testosterone; specific activity, 1.37 TBq/mmol; Amersham, North Ryde, Australia), nandrolone standard
(Steraloids), extracts of plasma equivalent to 100 µl/tube and a
dextran-charcoal separation of bound and free steroid.
Cross-reactivities (expressed as molar ratios at the ED50)
with testosterone (0.04%), dihydrotestosterone (0.7%),
androstenedione (0.3%), estradiol (0.02%), nandrolone phenylpropionate (6.3%) and nandrolone decanoate (1.3%) were
negligible in relation to circulating levels. The assay detection limit
(B/Bo = 0.90) was 8 pg/tube (equivalent to
0.25 nM), and the ED50 was 100 pg/tube (equivalent to 3.5 nM). Coefficients of variation at low (
ED80), medium
(ED50) and high (ED20) levels of the standard curve (n = 16-24 assays) were 11.7, 9.8 and
11.9% (between-assay) and 6.2, 3.5 and 5.3% (within-assay),
respectively.
Plasma testosterone was measured by radioimmunoassay using a rabbit
antibody to testosterone-3-O-carboxymethyloxime-bovine serum
albumin (SGT-1, supplied by Dr. B. Caldwell, Yale University), [1,2,6,7,16,17(N)-3H]testosterone (specific activity,
5.00-6.66 TBq/mmol; DuPont, North Ryde, Australia), testosterone
standard (Steraloids), extracts of plasma equivalent to 12.5 µl/tube
and a dextran-charcoal separation of bound and free steroid. Correction
was made for cross-reactivity with nandrolone (20.5%), but
cross-reactivities with dihydrotestosterone (21.1%), androstenedione
(2.5%), estradiol (0.17%), nandrolone phenylpropionate (0.04%) and
nandrolone decanoate (<0.0002%) were negligible in relation to
circulating levels. The assay detection limit
(B/Bo = 0.90) was 2 pg/tube (equivalent to
0.6 nM), and the ED50 was 22 pg/tube (equivalent to 6 nM).
Coefficients of variation at low (ED80), medium
(ED50) and high (ED20) levels on the
standard curve (n = 8-32 assays) were 12.6, 17.1 and
12.9% (between-assay) and 8.3, 3.7 and 6.0% (within-assay),
respectively.
Plasma inhibin was measured by a heterologous double-antibody
radioimmunoassay established in our laboratory (Handelsman et al., 1990; Crawford and Handelsman, 1994) and validated for humans (McLachlan et al., 1990; Burger, 1992; Dong et
al., 1992; Wallace et al., 1993). Reagents, provided by
Dr. G. Bialy (Contraceptive Development Branch, National Institute of
Child Health and Human Development), were rabbit antiserum to bovine
inhibin (rAs-1989), purified 31-kDa inhibin from bovine follicular
fluid for iodination (bINH-R-90/1) and bovine inhibin standard
(bINH-R-90/1). Duplicate plasma samples (100 µl) were preincubated
overnight with antibody before addition of 125I-labeled
inhibin. The standard curve was blanked with castrate human serum to
equalize serum protein concentrations in the assay tubes. The detection
limit (B/B0 = 0.90) was 19.5 pg/ml,
and coefficients of variation at inhibin standard levels of 90, 200 and
390 pg/ml were 3.1, 7.3 and 7.4% (between-assay) and 1.8, 3.0 and
5.0% (within-assay), respectively.
Data analysis.
Plasma nandrolone, testosterone and inhibin
levels were initially analyzed to evaluate the full time course for
each hormone by repeated-measures analysis of variance with BMDP 5V
software (Dixon, 1992), testing main effects by the Wald
2 test and using suitable linear contrasts to define
effects of ester, injection site and volume. Due to the very different
time courses of the phenylpropionate and decanoate esters, the three groups (groups 2-4) receiving the decanoate ester were also analyzed separately.
Plasma nandrolone levels were also analyzed by standard pharmacokinetic
methods involving polyexponential curve fitting (Gibaldi and Perrier,
1982), with a weighted, nonlinear, least-squares, curve-fitting
algorithm, by BMDP 3R software (Dixon, 1992). Standard pharmacokinetic
parameters (time of peak, peak concentration, mean residence time,
apparent half-times for absorption and clearance, systemic clearance
and area under the curve) were derived empirically from the plasma
nandrolone concentrations and as mathematical functions of the
coefficients of the best-fit curve (Gibaldi and Perrier, 1982).
Estimates of absolute bioavailability of nandrolone were calculated
from the disappearance curves of plasma nandrolone after i.v.
administration, by fitting to a variance-weighted, triexponential curve
with corrections for the molecular weight of nandrolone (274.4) and its
decanoate (428.7) and phenylpropionate (406.6) esters.
This pharmacokinetic analysis was subsequently extended by building a
population pharmacokinetic model using the approach of Mandema et
al. (1992). The plasma concentration of nandrolone was considered
to be the convolution of a monoexponential or biexponential absorption
function with a biexponential unit disposition function
|
(1)
|
The input rate (IR) for the monoexponential absorption model is
given by
|
(2)
|
where D is the dose, F is the fraction
absorbed (bioavailability) and k is the first-order
absorption rate constant. The input rate for the biexponential
absorption model is given by
![IR<IT>=D · F · </IT>[<IT>P · k<SUB>1</SUB> · e</IT><SUP>−<IT>k<SUB>1</SUB> · t</IT></SUP><IT>+</IT>(<IT>1−P</IT>)<IT> · k<SUB>2</SUB> · e</IT><SUP>−<IT>k<SUB>2</SUB> · t</IT></SUP>]](/content/vol281/issue1/fulltext/93/img003.gif)
|
(3)
|
where k1 and k2
are first-order absorption rate constants, P is the
proportion of the bioavailable dose absorbed by
k1 and (1 P) is the
proportion of the bioavailable dose absorbed by k2. The convolution of these input functions
with a biexponential disposition function yielded equations 4 and 5,
which describe the plasma concentrations over time after i.m. injection
for the monoexponential and biexponential absorption models,
respectively.
|
(4)
|
|
(5)
|
The population pharmacokinetic model did not include the data
from the two subjects who received nandrolone i.v. Thus, parameter F was not identifiable and was incorporated into the models
as F·A1 and
F·A2. The interindividual error on
each of the model parameters was modeled using a logarithmic-normal
variance model
where Pi is the value of the parameter in
the individual, 
TV is the typical value of the parameter
in the population and 
is a random variable with mean 0 and variance
2. The estimates of obtained with NONMEM are similar
to the coefficient of variation for the parameter, which we report as
the population variability, expressed as a percentage. We used a
"constant coefficient of variation" model for the variance of the
intraindividual residual error. Empirical Bayesian estimates of the
individual pharmacokinetic parameters were obtained based on the
typical values of the structural model parameters and on the variances
of the interindividual errors. These population pharmacokinetic models
were implemented with the computer program NONMEM (Beal and Sheiner,
1992).
We used a GAM (Chambers and Hastie, 1993) to permit identification of
linear and nonlinear relationships between the Bayesian estimates of
individual model parameters and the volunteer covariates (ester, site
and volume of injection, group, age, weight, height, body surface area
and lean body mass). The model that resulted in the biggest decrease in
the Akaike information criterion was returned by the GAM function as
the best model. The structural model incorporating covariates
identified by the GAM analysis was then evaluated with NONMEM to
develop the final parameter estimates. The final model was examined for
parsimony by exclusion of individual covariates and demonstration of a
statistically significant increase in the NONMEM objective function and
by analysis of the standard errors of the estimated parameters.
Pharmacodynamics.
The pharmacodynamic effects of nandrolone
on plasma testosterone and inhibin were estimated by nadir
concentration, time of nadir, net secretion and duration of
suppression. The duration of suppression was defined as the time when
plasma testosterone levels were below normal (<10 nM) or inhibin
levels were reduced by 50% of base line. The effects of ester,
injection site and volume were determined by parametric (testosterone)
and nonparametric (inhibin) analysis of variance.
This pharmacodynamic analysis was extended by building a population
pharmacodynamic model, again using the approach of Mandema et
al. (1992). The pharmacodynamic model was based on one of the four
basic indirect response models recently proposed by Dayneka et
al. (1993). Because testosterone and inhibin secretion are both
suppressed by nandrolone, model I of Dayneka et al. was
physiologically the most appropriate to describe the pharmacodynamic
effects of nandrolone and is shown in equation 6.
|
(6)
|
R represents the measured response variable (either
testosterone or inhibin concentrations),
kin0 is a zero-order rate
constant (the base-line testosterone or inhibin daily input rate),
Cp(t) is the plasma concentration of the
inhibiting drug (nandrolone) as a function of time and IC50 is the drug concentration that results in 50% of maximum inhibition of
the production rate. Under steady-state base-line conditions, it is
noted that
kin0 = kout · R0,
where R0 is R(t) at
t = 0, reducing the number of parameters in the model.
We have modified equation 6 to allow for incomplete inhibition of
testosterone and inhibin synthesis by nandrolone and have included a
parameter (
) to describe the steepness of this relationship
|
(7)
|
where, for either testosterone or inhibin,
R(t) is the measured concentration,
R0 is the base-line concentration,
Rmin is the minimum concentration when the input
rate is maximally suppressed by nandrolone, kout
is the first-order elimination rate constant, Cp(t) is the predicted plasma concentration for
nandrolone (based on individual dosing and Bayesian pharmacokinetic
parameter estimates) and IC50 is the concentration of
nandrolone associated with 50% suppression of synthesis. The parameter
was implemented as = 1 + , to enable a comparison of the
full model ( > 1) and reduced model ( = 1, = 0) using
the likelihood ratio test.
Alternatively, partial inhibition of the input rate can be modeled with
an additional term expressing the fractional inhibition (Imax), as shown in equation 8.
|
(8)
|
We used the parameterization shown in equation 7, because we
were interested in estimating the base-line and maximally suppressed concentrations of testosterone and inhibin (and the interindividual variability in these parameters) directly. Based on the
parameterization in equation 7, Imax is readily
calculated as
|
(9)
|
The interindividual errors in the model parameters
(Ro, Rmin,
kout, IC50 and ) were assumed to
have a logarithmic-normal distribution, and the variance of the
residual errors was assumed to be homoscedastic. As described for the
pharmacokinetic analysis, a GAM was used to identify significant
covariates, and NONMEM was used to develop the final pharmacodynamic
model. All data are expressed as mean ± S.E.M.
| |
Results |
Volunteers randomized into the four groups were comparable in
anthropometric and hormonal variables (table 1). There
were no significant differences between groups in mean
dehydroepiandrosterone sulfate, LH, FSH, prolactin, insulin-like growth
factor-I, hemoglobin, urea or creatinine concentrations (data not
shown), which were normal for all men.
Global statistical analysis.
Considering all four groups,
global statistical analysis demonstrated significant differences in the
time course of plasma nandrolone concentrations (group,
2 = 84.6, 3 dF, P < .001 by Wald test; group × time interaction, 2 = 643, 66 dF, P < .001).
These systematic differences were attributable to differences between
different nandrolone esters (table 2). Similarly the
time course of plasma testosterone concentrations varied significantly
by group (group × time interaction, 2 = 266, 66 dF, P < .001) due to effects of both ester and injection site
(table 2). To adjust for the dominating effect of differences between
esters, a global analysis conducted for the three groups receiving
nandrolone decanoate (groups 2-4) demonstrated significant effects of
injection site on plasma nandrolone levels, as well as effects of
injection volume and site on plasma testosterone and inhibin levels
(table 3).
View this table:
[in this window]
[in a new window]
|
TABLE 2
Global statistical analysis of plasma nandrolone and testosterone after
nandrolone ester injection by time course, ester, injection site and
volume
|
|
View this table:
[in this window]
[in a new window]
|
TABLE 3
Global statistical analysis of plasma nandrolone, testosterone and
inhibin after nandrolone decanoate injection by time course, injection
site and volume
|
|
Pharmacokinetic analysis.
Plasma nandrolone concentrations
reached higher and earlier peak concentrations and had a shorter mean
residence time after injection of the phenylpropionate ester (group 1;
n = 7), compared with the other three decanoate ester
groups (fig. 1; tables 2 and 4). Analysis
of the concentration data obtained from the two subjects who received
i.v. nandrolone gave the following values: area under curve/unit
dose = 1.3224 × 10
3 days/liter, mean residence
time = 25.65 ± 5.22 min, volume of distribution = 11.46 ± 0.30 liters and systemic clearance = 31.52 ± 7.26 liters/hr. From the optimal triexponential curve fit, the following parameters were estimated: A = 153.3 ± 2.7 nM, 
= 0.4132 ± 0.0030 min1,
B = 8.8659 ± 0.4575 nM, 
= 0.0098 ± 0.0022 min1, C = 0.9708 ± 0.0066 nM
and = 0.00043 ± 0.00045 min1. Based on the
area under the curve estimates for these two subjects, the absolute
bioavailability of nandrolone from i.m. injections of esters was
significantly higher for nandrolone decanoate injected into gluteal
muscle with a 1-ml volume (73%), compared with the other three groups
(53-56%).
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 1.
Time course of plasma nandrolone concentrations in
23 healthy men over 32 days after i.m. injection of 100 mg of
nandrolone phenylpropionate in 4 ml of arachis oil vehicle into the
gluteal muscle (group 1) ( ) or injection of 100 mg of nandrolone
decanoate into the gluteal muscle in 4 ml of arachis oil vehicle (group 2) ( ), into the gluteal muscle in 1 ml of arachis oil vehicle (group
3) ( ) or into the deltoid muscle in 1 ml of arachis oil vehicle
(group 4) ( ). Results are expressed as mean and S.E.M., unless the
S.E. is smaller than symbol.
|
|
The final population pharmacokinetic model incorporated ester, site and
volume of injection and height as significant covariates. Height was
significantly superior to weight, body surface area or lean body mass
as a covariate. The type of ester influenced the absorption profile of
nandrolone, such that the phenylpropionate ester was best described by
a one-compartment absorption model and the decanoate ester was best
described by a two-compartment absorption model. This was implemented
in the model with parameter P (table 5). The
interpretation of this parameter was that, effectively, the total dose
of the phenylpropionate ester is administered into the "fast"
compartment characterized by the rapid absorption rate constant
(k1), whereas only ~14% of the total dose of
the decanoate ester is administered into this compartment, with the
remaining ~86% of the total dose being administered into the
"slow" compartment characterized by a slower absorption rate
constant (k2). This basic difference in the
profiles of the nandrolone concentration data is shown in figure
2. In figure 2, the individual in each of the four
groups with the median mean absolute prediction error was selected to
represent the Bayesian predictions based on the individual
pharmacokinetic parameters. In addition, the rate of absorption from
the fast compartment (k1) was greater for the deltoid muscle than the gluteal muscle.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 2.
Observed and model-predicted time course of plasma
nandrolone concentrations in four healthy men over 32 days after i.m.
injection of 100 mg of nandrolone ester. The four men were selected
from each of the treatment groups according to the median predicted error in nandrolone concentrations, so that they were most
representative of that group. Note the logarithmic vertical scale. ,
observed data;  , individual Bayesian predictions.
|
|
A two-compartment disposition function performed significantly better
than a one-compartment disposition function. Site and volume of
injection were important covariates for the two parameters, with
bioavailability as a component
(F·A1 and
F·A2).
F·A1 was greater in the gluteal
muscle, compared with the deltoid muscle, and
F·A2 was greater with a smaller
injection volume (1 ml vs. 4 ml). Although the slow hybrid
rate constant 
2 was difficult to estimate accurately, it
describes the very slow terminal elimination phase (fig. 2) and
significantly improved the logarithmic-likelihood objective function of
the model, as exemplified by the improved fit of the two-compartment,
compared with one-compartment, disposition function.
Pharmacodynamic analysis.
Plasma testosterone concentrations
were most rapidly and completely suppressed within the first week after
injections of the phenylpropionate ester (fig. 3; tables
2 and 6), but this suppression was sustained for the
shortest time. The duration of suppression was significantly longest
after the gluteal 1-ml injection. Plasma testosterone concentrations
returned to base line by day 13 after the phenylpropionate ester but
required >20 days to return to base-line levels after the decanoate
ester. Among the decanoate ester injections, both injection volume and
site significantly influenced plasma testosterone concentrations
(tables 3 and 6). Plasma inhibin levels after decanoate ester
injections were suppressed to significantly lower nadir levels after
1-ml gluteal injection (fig. 3; table 6). Plasma inhibin was not
assayed after nandrolone phenylpropionate (group 1) injection.
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 3.
Time course of plasma testosterone concentrations
in 23 healthy men over 32 days after i.m. injection of 100 mg of
nandrolone phenylpropionate in 4 ml of arachis oil vehicle into the
gluteal muscle (group 1) () or injection of 100 mg of nandrolone
decanoate into the gluteal muscle in 4 ml of arachis oil vehicle (group 2) (), into the gluteal muscle in 1 ml of arachis oil vehicle (group
3) () or into the deltoid muscle in 1 ml of arachis oil vehicle
(group 4) (). Results expressed as mean and S.E.M., unless the S.E.
is smaller than the symbol.
|
|
The GAM analysis detected no statistically significant covariates for
the testosterone pharmacodynamic model (table 7). The inclusion of a parameter to estimate the nadir concentration of testosterone resulting from maximal suppression of testosterone synthesis by nandrolone and of a slope parameter describing the steepness of the relationship between the nandrolone concentration and
the testosterone output rate significantly improved the model. Figure
4 shows the predicted testosterone concentrations for
the same individuals as shown in figure 2. These predictions were calculated using the Bayesian estimates of the individual's nandrolone pharmacokinetic parameters and testosterone pharmacodynamic parameters.
View larger version (20K):
[in this window]
[in a new window]
|
Fig. 4.
Observed and model-predicted time course of plasma
testosterone concentrations in four healthy men over 32 days after i.m. injection of 100 mg of nandrolone ester. The four men were the same
individuals as illustrated in figure 2, who were originally selected
from each of the treatment groups according to the median predicted
error of nandrolone concentrations, so that they were most
representative of that group. , observed data; , individual Bayesian predictions.
|
|
No statistically significant covariates were detected in the GAM
analysis for the inhibin pharmacodynamic model (figs. 5
and 6; table 7). Unlike testosterone, the inclusion of a
parameter to estimate the nadir inhibin concentration did not improve
the model, although a slope parameter did significantly improve the model fit. Figure 6 shows the predicted inhibin concentrations for the
same individuals as shown in figure 2. Plasma inhibin was not assayed
after nandrolone phenylpropionate (group 1) injection. These
predictions were calculated using the Bayesian estimates of the
individual's nandrolone pharmacokinetic and inhibin pharmacodynamic parameters.
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 5.
Time course of plasma inhibin concentrations in 23 healthy men over 32 days after i.m. injection of 100 mg of nandrolone
decanoate into the gluteal muscle in 4 ml of arachis oil vehicle (group 2) (), into the gluteal muscle in 1 ml of arachis oil vehicle (group
3) () or into the deltoid muscle in 1 ml of arachis oil vehicle
(group 4) (). Results are expressed as mean and S.E.M. unless the
S.E. is smaller than the symbol.
|
|
View larger version (14K):
[in this window]
[in a new window]
|
Fig. 6.
Observed and model-predicted time course of plasma
inhibin concentrations in three healthy men over 32 days after i.m.
injection of 100 mg of nandrolone decanoate. No inhibin data were
available for the group that received nandrolone phenylpropionate
injections. The three men were selected from three of the four
treatment groups according to the median predicted error, so that they
were most representative of that group. , observed data; ,
individual Bayesian predictions.
|
|
| |
Discussion |
The present study demonstrates that, in addition to the chemistry
of the side-chain ester, both injection site and volume can
systematically influence blood nandrolone levels after i.m. injection
of nandrolone esters in an oil vehicle formulation. Corresponding to
the patterns of blood nandrolone concentrations, pharmacodynamic
indices reflecting androgen-induced inhibition of pituitary-testicular
function, namely blood testosterone and inhibin concentrations, are
also systematically influenced by these factors. Crucially, the
mixed-effects pharmacodynamic modeling demonstrated that essentially
all of the pharmacodynamic variability in plasma testosterone and
inhibin concentrations was accounted for by the variability between
esters and the site and volume of injection of the nandrolone
injections. The present study extends knowledge of the clinical
pharmacokinetics of nandrolone esters, which were reported in two
previous studies concerning nandrolone decanoate kinetics in humans
(Belkien et al., 1985; Wijnand et al., 1985);
there are no reports of the pharmacokinetics of the phenylpropionate
ester.
One feature of this study is the use of the population pharmacokinetic
approach to integrating pharmacokinetic and pharmacodynamic data.
Whereas the global statistical analysis indicates the significance of
some key variables, a structural model allows more physiological interpretation of the findings, especially identifying the
relationships between the pharmacokinetic and pharmacodynamic effects.
We have used an indirect response model (Dayneka et al.,
1993), which allows for more physiologically meaningful models, more
realistic interpretation of derived estimates and statistically valid
testing for categorical covariables, while more efficiently using all of the experimental data (Jusko and Ko, 1994). The indirect
physiological model-based approach also readily allows incorporation,
into a general model, of data from different subpopulations where the kinetics and dynamics had differing shapes for the effective
dose-response relationships. It is a striking validation of the
model-based estimates that the nadir concentrations of testosterone and
inhibin correspond very accurately to the known concentrations of these hormones in castrated men (1-3 nM and 0 pg/ml, respectively)
(McLachlan et al., 1990; Handelsman, 1994).
An interesting feature of this analysis of testosterone suppression is
that the first-order rate constant for the response compartment
(kout) does not correspond to the metabolic
clearance rate for testosterone. Using either bolus injection or
steady-state infusion, the metabolic clearance rate for testosterone in
men is ~540 liters/m2/day (Gandy, 1977) and the volume of
distribution is 10 to 20 liters, which is similar to the 11.5 liters
estimated in this study for nandrolone, a molecule almost identical to
testosterone. These estimates would indicate a
kout value of ~50 day1, whereas
our observed estimate for kout was 0.708 day1; the latter is consistent with a much slower rate
(half-time, ~1 day). This discrepancy is attributable to the fact
that the overall kinetics of suppression of testosterone are dominated by the slow negative feedback system, rather than the much faster metabolic clearance of testosterone. This negative feedback is mediated
via inhibition of pulsatile gonadotropin-releasing hormone secretion from hypothalamic neurons into the pituitary portal system
and then pituitary LH secretion from gonadotropes. For example, a
highly potent and specific gonadotropin-releasing hormone antagonist
that causes immediate cessation of gonadotropin-releasing hormone
action leads to castrate testosterone concentrations within 12 hr
(Behre et al., 1992), compared with 5 to 10 days in this study. This illustrates the need for physiological insight when interpreting indirect pharmacodynamic models because, in this instance,
the relationship between circulating nandrolone concentration and the
input function to the model may itself be indirect. Our paradigm
exemplifies a paradigm where the kout parameter
may accurately predict the time course of overall behavior of a system
without corresponding to the metabolic clearance rate of the drug or
the pharmacodynamic endpoint under study.
In the present study we have used specific radioimmunoassays to measure
the nandrolone, testosterone and inhibin concentrations, with the
latter two representing effective markers of endogenous pituitary
gonadotropin (LH and FSH, respectively) secretion. This reflects the
physiological fact that pituitary LH acts exclusively upon testicular
Leydig cells, due to their unique expression of cell surface membrane
LH receptors. In healthy men, virtually all circulating testosterone
originates from Leydig cells, with an absolute requirement for trophic
influence from LH derived from the bloodstream. Similarly, pituitary
FSH acts exclusively upon testicular Sertoli cells, which uniquely
express FSH receptors on their cell surface membranes, and virtually
all circulating immunoreactive inhibin originates from the gonads
(Burger, 1992). As a result, blood levels of these two hormones are
useful integrated bioassay indicators of endogenous pituitary
gonadotropin secretion, as reflected by the testicular hormonal
response to ambient blood LH and FSH levels. In the present study,
these two pharmacodynamic indices showed physiologically meaningful
distinctions between the esters and the effects of injection site and
volume.
Variations in side-chain ester chemistry are important in the
pharmacokinetics of androgen esters in oil vehicle (Behre et al., 1990). Experimental studies suggest that absorption rates are
predicted by the oil/water partition coefficients (or hydrophobicity) and that the oil vehicle is absorbed more slowly than the androgen ester (Tanaka et al., 1974). In humans, the very short
propionate (three-carbon aliphatic) ester of testosterone has
distinctly shorter duration of action than esters with longer (seven-
or eight-carbon) side-chains (Nieschlag et al., 1976;
Schulte-Beerbuhl and Nieschlag, 1980; Schurmeyer and Nieschlag, 1984;
Belkien et al., 1985; Fujioka et al., 1986). More
subtle changes in side-chain ester structure have proven ineffective in
altering human clinical pharmacokinetics, because substitution of a
linear aliphatic side-chain of seven carbons (enanthate) with either a
saturated, cyclized, seven-carbon aliphatic chain
(cyclohexanecarboxylate) (Schurmeyer and Nieschlag, 1984) or a linear,
aliphatic, eight-carbon chain (cypionate) (Schulte-Beerbuhl and
Nieschlag, 1980) resulted in virtually unchanged kinetics. Wider
variation in ester side-chain chemistry to include greater chain length
and/or aromatic ring structures is a more effective determinant of
ester pharmacokinetics, because nandrolone hexoxyphenylpropionate ester
(aromatic ring with 18 carbons) had far better depot properties, with a
prolonged and retarded release profile, compared with the decanoate
(aliphatic chain with 10 carbons) (Belkien et al., 1985).
The present study indicates that a side-chain ester consisting of a
10-carbon aliphatic chain has better depot properties than a
nine-carbon chain including an aromatic ring. Because the vehicle
(arachis oil) was unchanged during this study and because of the
experimental observation that the oil vehicle influences local reaction
to the oil injection (Brown et al., 1944), as well as
androgen ester pharmacology (Ballard, 1980; Al-Hindawi et
al., 1986), the present conclusions may be extrapolated to other
vegetable oil injection vehicles only with caution.
Injection technique, including injection site, volume and
concentration, as well as the nature of the vehicle, could
theoretically be important for androgen ester release rate. Injection
site may be important because of differences in tissue composition
(Cockshott et al., 1982) and blood flow (Bederka et
al., 1971); indeed, i.m. oil-based injections may more accurately
be termed intermuscular (Ballard, 1968) or intralipomatous (Cockshott
et al., 1982). The former reflects the tendency of oil
vehicle to distribute along intermuscular fascial planes (Ballard,
1968), whereas the latter depends upon the amount of fat at the
injection site (including systematic gender differences) (Modderman
et al., 1983) together with needle geometry and anatomy of
the injection depot. Intralipomatous deposition of injections with a
larger vehicle volume may explain the slower release kinetics of
nandrolone decanoate in the gluteal region, as well as the differences
from the deltoid site, which has a lower fat content. The higher blood
flow in the deltoid, compared with the gluteal, muscle (Evans et
al., 1975) may also be important. Analogous site-dependent
differences in absorption rate and physiological effects have been
described for a variety of drugs in aqueous solution (Greenblatt and
Koch-Weser, 1976). To our knowledge, there are no previous reports
examining the systemic pharmacokinetic and pharmacodynamic effects of
injection site and volume for androgen esters in oil vehicle in men.
One possible clinical impact of these observations may lie in recent
observations of differences between population groups in the efficacy
of regular i.m. injections of testosterone enanthate in an oil vehicle
to suppress testicular function for male contraception (World Health
Organization Task Force on Methods for the Regulation of Male
Fertility, 1990). In that and related (World Health Organization Task
Force on Methods for the Regulation of Male Fertility, 1993) studies,
interethnic differences in susceptibility to androgen-induced azoospermia were not due to differences in overall body size or related
differences (Handelsman et al., 1995). Evaluation of the possibility of ethnopharmacological differences, however, required a
greater understanding of the rate-determining mechanisms of androgen
release from androgen ester depots in oil vehicles. The present
findings suggest that differences in absorption of androgen esters may
contribute to such interethnic differences through possible local
mechanical factors (e.g., exercise, compression and muscle
and fat mass) at the injection site, and this issue warrants further
study. Analogous variations in the pharmacokinetics of steroid esters
have been reported among women from different countries using
long-acting contraceptive steroids (Garza-Flores, 1994), although no
explanation has been advanced. Further analysis of the present
observations may facilitate such ethnopharmacological studies, as well
as clinical applications of androgen esters in oil vehicle
formulations.
The authors thank Paul Mutton for his help in conducting this
study. The authors are also grateful to Dr. G. Bialy of the Contraceptive Development Branch, National Institute of Child Health
and Human Development, for kindly supplying inhibin kits and to
Christine Young and Jennifer Spaliviero for assisting with assays.
FSH, follicle-stimulating hormone;
GAM, generalized additive model;
LH, luteinizing hormone.
![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
Top
Abstract
Introduction
