CLIN. CHEM. 31/1, 60-64 (1985)
Equilibrium Dialysis, Ultrafiltration, and Ultracentrifugation Compared for
Determiningthe Plasma-Protein-Binding
Characteristicsof ValproicAcid
J#{233}r#{244}me
Barr#{233},1
Jean Marie Chamouard,1Georges Houin,2 and Jean Paul Tlllement1
Equilibrium dialysis, ultrafiltration, and ultracentrifugation
were compared to determine their reliability and applicability
in the study of binding of an anticonvulsant drug, valproic
acid, by plasma proteins. We studied drug binding with
pooled serum and with solutions of human serum albumin at
physiological concentrations. We compared binding charac-
tenstics such as number of binding sites, affinity constants,
and percent of binding as measured by each method in the
therapeutic range for valproic acid. Results by ultracentrifugation differed from those by equilibrium dialysis and ultrafiltration, which agreed reasonably well with each other.
Addftlonal Keyphrases: anticonvulsant drugs . pharmacokinetics
bound vs free drug
serum vs albumin
effects of
free fatly acids
monitoring therapy
The plasma-protein
binding of drugs continues to become
more important for proper pharmacokinetic evaluations and
in routine clinical monitoring of drugs (1-5). This is because, for many drugs, the therapeutic and (or) toxic response correlates better with the concentration of diffusible,
unbound drug than with the total drug concentration (6). At
present, the reliability of techniques for measuring the
binding of drugs by plasma proteins is difficult to assess,
there being no “standard” in measurements of drug-protein
binding, unlike other routine biochemical tests. Equilibrium dialysis is often regarded as the “reference method” (7),
especially for measurements of a drug that saturates plasma
proteins
even at therapeutic
concentrations.
The availability of new ultraffitration
and ultracentrifugation equipment has encouraged us to evaluate these
methods and to compare results with those obtained by
equilibrium
dialysis.
To our knowledge, there are no reports that directly
compare the reliability of these methods in a single set of
experiments. We therefore decided to use all three methods
to investigate the protein binding of valproic acid (VPA, ndipropylacetic acid), a drug known to supersaturate the
plasma proteins even at therapeutic concentrations (8).
Materials and Methods
Reagents
VPA solutions were prepared in phosphate buffer (67
mmol/L, pH 7.4, ionic strength 0.176) by isotopic dilution of
a constant amount of ‘4C-labeled VPA (58 Ci/mol; Amersham International,
Amersham, Bucks, U.K.) with increas‘Laboratoire de Biochimie I et Pharmacologiedu Centre Hospitalier Intercommunal de Creteil, 40 avenue de Verdun, 94000
Cr#{233}teil,
France.
2Laboratorie de Pharmacologie,Facult#{233}
de Pharmacie, Universite Paul Sabatier, 31400 Toulouse, France.
‘ Nonstandard abbreviations:VPA, valproicacid; HSA, human
serum albumin; FFA, free fatty acids; VLDL, very-low-density
lipoprotein;LDL, low-densitylipoprotein; and HDL, high-density
lipoprotein.
Received February 21, 1984; accepted October4, 1984.
60 CLINICALCHEMISTRY, Vol. 31, No. 1, 1985
ing amounts of unlabeled drug. The purity of the [14CIVPA
was assessed by using thin-layer chromatographic plates of
silica gel (no. 5714; Merck, Darmstadt, F.R.G.) with the
following solvent system: hexane/diethyl ether/acetic acid
(80/20/1 by vol). It was found to exceed 98%.
VPA binding was studied with use of human serum
albumin (HSA) (no. A-1887; Sigma Chemical Co., St Louis,
MO) containing 40 mmol of free fatty acid (FFA, expressed
as palmitic acid) per mole of HSA. HSA was dissolved in the
phosphate buffer to give a concentration of 600 nol/L.
Plasma lipoproterns, very-low (VLDL), low- (LDL), and
high-density (HDL), were isolated by ultracentrifugation
from pooled normolipidemic human serum as described by
Glasson et al. (9). We also used pooled human serum from
healthy subjects. It had the following characteristics:
total
proteins 68 g/L, HSA 690 molfL, and FFA (expressed as
palmitic acid) 520 mol/L.
Procedures
FFA assay. We determined FFA by gas-liquid chromatogby a modification
of the method of Sampson and
Hensley (10). We used a 3000 Series gas chromatograph
(Girdel, Suresnes, France) equipped with a flame ionization
detector and a Servotrace (Sefram, Paris, France) recorder.
Protein assay. We measured total protein concentration
in
the serum pool by using the biuret reagent (Sigma, no. 540A). To quantify HSA, we used an albumin reagent test kit
(no. 662996; Beckman Instruments, Gagny, France) with
the Beckman Immunochemistry
Systems ICS
III nephelometer. The sensitivity of this method is 116 nmolJL.
We detected trace amounts of protein in the supernates by
the method of Lowry et al. (11) after ultracentrifugation
(sensitivity: 2 pg/mL). VLDL, LDL, and HDL concentrations in serum corresponding to apolipoproteins B (VLDL +
LDL) and A (LDL) were measured by radial immunodiffusion on plates (M. Partigen, Behringwerke, F.R.G.).
Binding
techniques.
VPA binding to HSA and serum was
measured by equilibrium dialysis, ultrafiltration, and ultracentrifugation. We investigated two ranges of VPA concentrations, one (4-14 000 tmol/L) to determine the binding
parameters, the other (150-3000 nnol/L) to investigate the
influence of the kind of technique used on the measurement
of VPA binding. Each experiment was run in triplicate and
the results were expressed as the mean ±SD.
For equilibrium dialysis, we used Teflon microcells (Dianorm#{174};
Diachema, Ruschlikon, Zurich, Switzerland) with
two 200-giL chambers separated by a semipermeable membrane (Mr cutoff 12 000; Union Carbide, Chicago, IL). Samples were stirred constantly at 20 rotations per minute, at
37 #{176}C.
We introduced VPA into the buffer compartment
to
check for the absence of binding to lipoproteins and to determine the parameters of VPA binding to serum and HSA.
We added VPA to serum in a minimum volume of phosphate
buffer to facilitate the comparison of percentage binding
with those obtained by the other techniques. The ratio of
this added volume to serum sample volume never exceeded
0.05, so that dilution of serum protein was negligible.
raphy,
In preliminary experiments, we found that equilibrium
analysis of variance for balanced data. The confidence
was attained after 3 h of dialysis in all cases, so all samples
intervals based on the residual variance were computed for
each pair of techniques (13).
were measured after that interval. No degradation of 4Clabeled VPA was found by thin-layer
chromatographic
Results
measurement after completion of dialysis. In the dialysis of
serum, the magnitude of the volume shift from the buffer to
Determination
of the presence of protein in dialysate,
the protein compartment
was negligible (<10%), so we did
ultrafiltrate,
and supernate.
No HSA leakage could be denot correct for it. We measured the concentrations of VPA in
tected in the dialysate, the ultrafiltrate, or the supernate
both compartments after dialysis, accounting for at least
collected after ultracentrifugation. Therefore the HSA con95% of the drug.
centration, if any, was <116 nmolJL, which is negligible.
For ultrafiltration,
we used the EMIT free level ifiter
However, the Lowry assay detected a protein concentration
system (Syva Co., Palo Alto, CA), placing in the sample
of 280 g/mL in the supernate after ultracentrifugation.
reservoir 1-mL portions of protein solutions supplemented
Those proteins were identified as lipoproteins, mainly
with various concentrations of VPA. We obtained ultraffiVLDL and LDL. The presence of lipoproteins in the supertrate containing free VPA by subjecting the system to lownate of the ultracentrifugate does not alter measurement of
speed centrifugation
(900 x g, 37#{176}C).
HSA and serum
the free VPA concentrations, because VPA did not exhibit
samples were centrifuged for 5 and 10 mm, respectively. We
any detectable binding to VLDL, LDL, or HDL in our
then could collect a volume of ultraflltrate ranging between
equilibrium-dialysis
experiments.
10 and 15% of the initial sample volume. Analytical recovHSA binding parameters.
Binding of VPA over the range
ery of VPA in this system was also about 95%.
of 4 to 14 000 moI/L
at an HSA concentration of 600
For ultracentrifi.igation, we used polyallomer tubes (Beckmol/L was studied by equilibrium dialysis, ultraffitration,
man) containing
230 pL of the protein solutions, and we
and ultracentrifugation techniques. In all cases the percentcentrifuged them at 100000 x g for 6 h at 37 #{176}C,
in a 42.2 Ti
age binding decreased from approximately 98% to 10% as
fixed-angle rotor (Beckman). The serum samples used were
concentrations of VPA were increased. In Figure 1, the
supplemented with VPA in the same way as for ultrafiltrathree binding curves show saturation of the VPA-binding
tion. Free valproic acid in the supernate was measured as
sites on HSA. From these results two saturable classes of
indicated in the following section. No degradation of
binding sites can be determined. The relevant binding
[‘4CIVPA was found by thin-layer chromatography after
parameters for each technique are shown in Table 1. Signifiultracentrifugation
was completed. Analytical recovery of
cant differences were observed only for the second class of
VPA in this system was 100%.
binding sites. The number of binding sites determined by
At the end of each kind of experiment, we measured the
equilibrium dialysis and the affinity constant determined by
concentrations of either free or bound drug (or both) in an
ultrafiltration differed significantly (p <0.05) from the corSL 3000 liquid-scintillation counter (Kontron SA, V#{233}lizy- responding parameters determined by ultracentrifugation.
Villacoublay, France).
Serum binding parameters. The range of VPA concentraIn dialysis experiments, we counted the radioactivity of
tions was the same as that previously used for HSA. For the
100-jL portions from each compartment in “Ready-Solv
three techniques, the percent of binding decreased from
MP” (Beckman) scintillation cocktail.
approximately 94% to 15% as the concentration of VPA
In ultrafiltration
experiments,
we measured a 50-FL
increased. These binding curves are also shown in Figure 1.
aliquot of the ultrafiltrate containing free VPA. The concenAs with lISA, two classes of sites can be determined.
tration of bound drug was calculated by subtracting the
Table 1 lists the concentrations of the binding sites for each
concentration of free drug from the total concentration
class and their corresponding affinity constants. The statisintroduced into the system.
tical calculations were performed in the same manner as for
In ultracentrifugation
experiments, counting the radioacHSA. For the first class of binding sites, the afilnity constant
tivity of 50 pL of the supernate allowed us to determine the
as determined
by equilibrium dialysis was higher than
concentration of free VPA. The concentration of the bound
those determined by the other two techniques but was
drug was the same as for ultrafiltration.
significantly
different only from that obtained by ultracenCalculations
of percentage
VPA bound.
In equilibrium
trifugation. For the second class, the affinity constant meadialysis experiments, the percent of bound can be expressed
sured by equilibrium dialysis was significantly higher (p
as
<0.05) than that determined by ultracentrifugation. However,
the FFA concentration in serum was 520 molJL
[(B
A)/B] x 100
(1)
before dialysis and 450 moLIL after dialysis was completed.
where B is the total drug concentration in the protein
VPA binding in serum. The percent of binding in serum
compartmemt and A the free-drug concentration in the
was measured by the three techniques. Samples were supbuffer compartment.
plemented with VPA over a range of 150-3000 mol/L. The
In ultrafiltration and ultracentrifugation techniques, the
results are listed in Table 2. A significant overall difference
percentage binding was determined as follows:
was found among the three techniques (p <0.001). The
confidence intervals showed a significant difference for each
[(T
F)IT] x 100
(2)
pair of techniques (p <0.05). However, it is clear from these
where F denotes the free-drug concentration
determined
in
results that the ultrafiltration values are much closer to
the ultrafiltration or the supernate and T is the total drug
those for equilibrium dialysis than to those obtained by
concentration introduced into the system.
ultracentrifugation.
The results show that the maximum
Computation
of binding parameters. All binding paramedifferences in percentages were 8.4% for equilibrium dialyters were estimated by means of a nonlinear least-squares
sis vs ultrafiltration, 20% for equilibrium dialysis vs ultramethod involving a Gauss-Newton algorithm (12).
centrifugation, and 13% for ultrafiltration vs ultracentrifuStatistical calculations. We compared the binding paramgation.
eters as determined by each method with the other two by
A comparison of the percent of binding obtained by the
calculating Student’s 95% confidence interval.
three methods was also made from another set of experiWe compared the percent binding by using a two-way
ments. In these experiments, VPA in the range 150-3000
-
-
CLINICALCHEMISTRY, Vol. 31, No. 1, 1985 61
-J
I
0
E
3
E
3
0
0
U
U
I)
2
0
2
0
a.
a.
-J
>
>
0
z
0
0
z
0
0
0
2
4
FREE
6
VALPROIC
8
ACID
10
6
mmol/t.
FREE
VALPROIC
ACID
.
mrnol/I.
Fig. 1. Binding of VPA (4-14 000 tmoVL) to HSA (600 jmoVL) (Iefl) orto serum (righi) as determined by equilibrium dialysis (U), ultrafiltration (A)
and ultracentrifugation (#{149})
Each point is the mean of three determinations. Standard deviations of the means fit within the poInts plotted
Table 1. Characteristics of VPA Binding
Equilibriumdialysis
In HSA (600 pmoI/L)
n1 = 1.47 ± 0.13
K, = 52615 ± 19275
=
=
Ultrafiltration
Ultracentritugation
1.50 ±
n1
n,
=
‘(1
=51000
0.14
17000
3.50
±
3.62 ± 0.208
250 ± 60
n2 = 8.11 ±
K2 = 80 ± 508
835 ± 97
28000 ± 85008
1359 ±
610 ± 147a.b
N,
=
1.40
=
43000
±
=
2.15
0.10
70
K2 = 350
0.08
±
±
±
6000
In serum
N,
=
=
N2 =
K2 =
=
=
928 ± 103
15300 ± 4000
N,
K,
N2 = 4188 ± 282
K2 = 173 ± 36
=
=
1138
5800
2298
N2 =
K2 = 75
±
207
2200
± 1428
98
±
±
The VPA concentrations ranged from 4 to 14000 mol/L. n represents the number of binding sites, N the concentration of binding sites, and K the affinity
constant. Subscripts 1 and 2 denote the first and the second class of sites, respectively. Results are expressed as mean ± SD. The calculations are derived froni
the mean value of three binding experiments (see Figure 1).
#{149}p
<0.05 as compared with ultracentrifugation.
bp
<0.05 as compared with ultracentrifugation.
Table 2. Percentages of VPA Binding as
Determined in Pooled Serum Samples
VPA
concn,
pmoi/L
150
300
450
600
750
900
1200
1800
3000
Equilibrium
dialysis
94.4
93.5
92
90.3
87.3
85.2
Ultrafiltration
±
0.1
0.1
0.3
0.2
0.3
0.2
90.7
89.1
87.2
84.6
80.5
78.6
79
±
0.6
69.4
54.3
±
0.2
0.3
±
±
±
±
±
±
Ultracentritugation
,
±
0.2
0.3
0.3
0.2
0.3
0.1
71.8
±
0.1
62.7
48.1
±
0.8
51.9
±
±
1
35.5
±
±
±
±
±
±
80
77.5
75.2
71.0
67.6
66.0
±
0.3
±
0.2
±
58.4
±
0.3
0.2
0.2
0.2
0.2
±
±
±
0.2
0.5
Each result is the mean (± SD) of three determinations. Results by each
method differed significantly from those by the other two (p <0.05).
pmol/L was introduced into the buffer compartment and was
equilibrated with serum. When equilibrium was achieved,
the protein compartment was emptied; aliquots of these
serum samples were used, as usual, for liquid scintillation
counting to determine the bound VPA. The remainder was
used to measure the binding percentages by the other two
techniques. Once again, a significant difference (p <0.001)
was found among the three techniques and between each
pair of methods (p <0.05) (Table 3). Moreover, the results
show differences of 2.5% for equilibrium dialysis vs ultrafiltration, 7% for equilibrium dialysis vs ultracentriftigation,
and 8% for ultrafiltration vs ultracentriftigation.
62
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
Table 3. Percentages of VPA Binding as
Determined by Equilibrium Dialysis and
Subsequent Ultrafiltration and Ultracentrlfugatlon
of Pooled Samples of Serum Dialysate
VPA
1200
Equilibrium
dialysis
95.2 ± 0.3
93.6 ± 0.3
91.1 ±0.4
87.3 ± 0.5
83.7 ± 1.1
77.3 ± 0.7
66.5 ± 0.7
1800
63.1
±
3000
45.9
±
concn,
pmol/L
150
300
450
600
750
900
0.2
0.4
Ultra-
Ultra-
filtration
centrifugation
95.9
93.5
92.2
±
0.7
92.8
±
0.1
±
0.5
89.8
86.9
±
0.2
69.2
±
65.1
±
49.8
±
1
88 ±1.8
84.4 ± 1.5
78.2 ± 1.3
±
0.9
0.5
0.2
1.1
80.9 ± 1.2
78.8 ± 1.2
72.9 ± 1.2
62.2
58.6
42.0
±
±
±
±
0.8
0.2
0.8
Each result is the mean (± SD) of three determinations. After completion of
dialysis of pooled serum supplemented with VPA, the protein compartments
containing free plus bound VPA were collected. Then an aliquot was subjected
to ultrafiltration and ultracentrifugation. Each method differed significantly from
the other two (p <0.05).
Discussion
The characteristics
of the two classesof VPA binding sites
in both serum and HSA solutions as determined by equilibrium dialysis-which
is often regarded as the reference
method-are in good agreement with those found by Unen
et al. (12) For HSA, the parameters of the first class of
binding sites determined by equilibrium dialysis, ultrafil-
tration, and ultracentrifugation
do not show any significant
differences. However, some significant differences appeared
in the second class of binding sites when we compared the
values for n and K obtained by each method. When one
compares the products n1K1 and n2K2, which reflect the
binding capacities of the two classes, it can be seen that the
contribution of the second class of sites is about one hundredth that of the first. Therefore, the discrepancies in the
binding constants of the second class between the three
methods should not affect the interpretation
of the overall
binding characteristics of the drug. For serum, the affinity
constant of the first class (K1) determined by ultracentrifugation is significantly lower than that obtained by equilibrium dialysis. In addition, the K1 obtained with the ultrafiltration method is less than that obtained by equilibrium
dialysis, although the difference is not significant. The
observed decrease in K1 when determined by equilibrium
dialysis, ultrafiltration, and ultracentrifligation, respectively, may be partly due to the differing concentrations of FFA
in the serum at the end of each experiment. In equilibrium
dialysis, a part of the FFA initially present in the protein
compartment is observed to diffuse into the buffer compartment, sothat the final FFA concentration shows a decrease.
Since no such dilution effect occurs in either ultrafiltration
or ultracentrifugation, the FFA concentrations remain constant during these two experimental
procedures. This phenomenon may be of importance,
because it has previously
been shown that FFA competitively inhibits the binding of
VPA to HSA (12). This is also likely to occur with other
unknown endogenous compounds that may interfere with
VPA binding.
In our experiments
with HSA, some significant differences were found in serum when we compared the parameters of the second class of sites measured with the three
techniques. There could be additional reasons for differences
observed in the results between equilibrium dialysis and
ultracentrifugation. The discrepancies could be partly due to
the VPA binding to lipoproteins, namely VLDL and LDL,
which can remain in the supernate and cause a spurious
increase in the free-drug concentration. This may not occur
for VPA because this drug is not bound to lipoproteins. It is
commonly believed that the binding equilibrium is altered
during ultracentrifugation. However, according to Gilbert
and Jenkins (14), equilibrium is not significantly affected if
the drug-protein
complex shifts with the same speed as that
of the protein alone. This may have happened in our
situation,
because the relative
molecular
masses of lISA
and of the complex VPA-HSA are, respectively, 66000 and
66 163. A free VPA concentration gradient developing in
the supernate during ultracentrifugation
and causing sedimentation is unlikely,
because it has been previously shown
(7, 15) that this phenomenon is negligible for low-Me drugs
such as VPA. Rather, drugs of low molecular mass were
shown to diffuse back from the bottom to the top of the tubes,
mainly because of thermal agitation of the molecules (7).
Such a back diffusion of VPA may have occurred in our
experiments, causing an altered equilibrium and subsequently an overestimation of the free-drug concentration.
Unfortunately, these possibilities could not be investigated
because of the inadequate volume (150 1L) and height (9
mm) of the supernate.
To evaluate the validity of the three methods for clinical
purposes, we determined the percent of binding in serum
supplemented with VPA (Table 2). Our results show that,
throughout the range studied, there was a highly significant
difference between equilibrium dialysis and ultrafiltration.
It may be that a dilution effect, due to the diffusion of the
free drug from the protein compartment to the buffer
during the dialysis, causes a disruption of the
equilibrium, thus accounting for this difference. Furthermore, the dilution of FFA may partly explain the higher
percent of binding found in dialysis. At saturating concentrations of drugs the differences are much greater than at
non-saturating
concentrations,
a fact well depicted by our
results, because the difference can reach 7% for high VPA
concentrations,
even within the therapeutic range (200-900
Mmol/L), while it is only 4% for low VPA concentrations.
Ultracentrifugation yields a much lower percent of binding
than the other two methods, probably for the reasons
compartment
already
discussed.
To circumvent the dilution effect of equilibrium dialysis
for the comparison of the three techniques,
ultrafiltration
and ultracentrifugation
were performed sequentially on
serum samples that had already been dialyzed. Although a
significant difference was observed (Table 3) between ultrafiltration and equilibrium dialysis (lower values obtained
with the latter method at higher concentrations), the two
series of values were reasonably close for clinical purposes.
The maximum difference did not exceed 1.1% within the
therapeutic range previously defined, and 2.9% over the
whole range studied. Previous reports (16, 17) have shown
that there is no disturbance of equilibrium during ultrafiltration partition. In ideal situations, when the binding
parameters-that
is, the number of sites and the affinity
constants-remain
unchanged irrespective of the protein
concentration, the concentration of free drug measured by
ultrafiltration
partition is unaltered regardless of the concentration of retained binding protein. However, when one
is dealing with a drug such as VPA, the binding parameters
may vary with the protein concentration (12). In our experiments, free VPA concentrations remained constant even
when the volume of ultrafiltrate ranged from 10 to 60% of
the HSA solutions or serum samples ultrafiltered. Thus,
despite the high variations in retained protein concentrations in the upper reservoir, no disturbance of the protein
binding equilibrium was apparent.
Currently,
there is no “standard”
method for protein
binding measurements. Most of the investigations have
been performed by using equilibrium dialysis, mainly because this method offers the advantage that free and bound
amounts of the drug are not separated and, therefore,
equilibrium
of binding will not be affected. This reason
chiefly accounts for the considerable amount of data now
available with this technique. Therefore, equilibrium dialysis may still be regarded as the reference technique to use in
investigating the binding characteristics of a new drug to
serum or to a particular protein such as the determination of
n and K. One should also refer to this technique to validate
other techniques such as ultrafiltration
and ultracentrifugation.
However, this technique
is limited by several drawbacks,
such as a dilution effect of the drug and the endogenous
compounds in the buffer compartment and the longer time
necessary to determine the binding characteristics. Obviously, this may not be a suitable procedure for drugs prone
to hydrolysis in serum. In light of those limitations, it seems
more judicious to choose either ultrafiltration or ultracentrifugation to evaluate the binding percentage of a drug for
clinical purposes. In our study, ultrafiltration gave results
closer to those obtained with equilibrium dialysis than did
ultracentrifugation.
Further, ultracentrifugation
requires
costly equipment, and some physical phenomena-such as
sedimentation, back diffusion, viscosity, and binding to
plasma lipoproteins in the supernatant fluid-can cause
errors in the estimation of the free-drug concentration.
Ultrafiltration may pose problems if there is significant
CLINICAL CHEMISTRY, Vol. 31, No. 1, 1985
63
nonspecific adsorption onto the membrane. In view of these
potential problems, a careful comparative study should
always be made before one of these two techniques is chosen
for use in routine clinical practice.
J. M. C. is the recipientofan Inserm fellowship.We thank LabazSanofi for their generousgift of radiolabeled valproic acid, and
Syva-Biomerieuxfor kindly supplyingthe ifiters. We are grateful to
Mr. P. D’Athis for his assistancein the statistical calculations, and
to Prof. Ren#{233}
Levy, Dr. Arunkumar Shastri, and Dr. Michael
Brinkley for their fruitful discussions and critical revision of the
English text.
References
1. Jusko WJ, Gretch M. Plasma and tissue protein binding ofdrugs
in pharmacokinetics. Drug Metab Rev 5, 43-140 (1976).
2. Tillement JP, Lhoste F, Giudicelli JF. Diseases and drug protein
binding. Clin Pharm.acoki net 3, 144-154 (1978).
3. Piafsky KM. Disease induced changes in the plasma binding of
basic drugs. Clin Pharrnacokinet 5, 246-262 (1980).
4. Gibson TP. Influence of renal disease on pharmacokinetics. In
Applied Pharmacokinetics,
WE Evans, JJ Schentag, WJ Jusko,
Eds.,Applied Therapeutics Inc., San Francisco, CA, 1980, pp 32-56.
5. Wilkinson GR. Influence of liver disease on pharmacokinetics.
Ibid., pp 57-75.
6. Rowland M. Plasma protein binding and therapeutic drug monitoring. Ther Drug Monitor 2, 29-37 (1980).
7. Kurz H, Trunk H, Weitz B. Evaluation of methods to determine
protein binding of drugs. Arzneim Forsch 27, 1373-1380 (1977).
8. Wilder B,J, Karas BJ. Relation of plasma concentration to
64 CLINICALCHEMISTRY, Vol. 31, No. 1,
1985
seizure control. In Antiepileptic Drugs, DM Woodbury, JK Penry,
CE Pippenger, Eds., Raven Press, New York, NY, 1982, pp 591599.
9. GlassonS, Zini R, Tillement JP. Multiple human serum binding
of two thienopyridinic derivatives, ticlopidine and PCR 2362 and
their distribution between LISA, a,-acid glycoprotein and lipoproteins. Biochem Pharmacol 31, 831-835 (1982).
10. SampsonD, Hensley WJ. A rapid gas chromatographic method
for the quantitation
of underivatised individual free fatty acids in
plasma. Clin Chim Acta 61, 1-8 (1975).
11. Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein
measurements
with the Folin phenol reagent. J Biol Chem 193,
265-275 (1951).
12. Unen S, Albengres E, Tillement JP. Serum protein binding of
valproic acid in healthy subjects and in patients with liver disease.
Int J Clin Pharin 19, 319-325 (1981).
13. Schwarti D. M#{233}thodes
statistisques
a l’usage des m#{233}decinset des
biologistes, F’lammarion Mbdecine-Sciences,Paris, 1969, pp 181205.
14. Gilbert GA, Jenkins RCL. Sedimentation and electrophoresis of
interacting substances. II. Asymptotic boundary shape for two
substances interacting reversibly. Proc R Soc 253, 410-437 (1979).
15. Scholtan W. Bestimmungsmethoden
und gesetzmassigkei
der
serum protein binding von arzneimitteln.
Arzneim Forsch 28,
1037-1047 (1978).
16. Whitlam JB, Brown KF. Ultrafiltration
in serum protein
binding determination. J Pharm Sci 70, 146-150 (1981).
17. Sophianopoulos
JA, Durham SJ, Sophianopoulos
Ad, et al.
Ultrafiltration
is theoretically
equivalent to equilibrium dialysis
but much simpler to carry out. Arch Biochem Biophys 187, 132-137
(1978).