This material may be protected by Copyright law (Title 17 U.S. Code)
`
`BIOPHARMACEUTICS & DRUG DISPOSITION, VOL. 11, 93~105 (I990)
`
`ABSORPTION OF DOXYCYCLINE FROM A
`CONTROLLED RELEASE PELLET
`FORMULATION:' THE INFLUENCE OF FOOD ON
`
`BIOAVAILABILITY
`
`DESMOND B. WILLIAMS*, WILLIAM J. O’REILLYT, GARTII BOEHM* AND MICHAEL J. STORY*
`
`*F.H. Faulding and Co. Ltd, Adelaide, South Australia and
`TSchool of Pharmacy, South Australian Institute of Technology
`
`ABSTRACT
`
`A three-way crossover study was performed to compare the bioavailability of a new
`pelletised doxycycline product administered either with food or without food and a
`reference product taken without food.
`Four different methods were used to calculate pharmacokinetic parameters from the
`data. The sums of squares, Akaike’s Information Criterion (AIC), and the ranges for the
`parameters obtained were used for comparison. Good fits to the data were obtained when
`all four methods were used, each with a lag time. The two compartment open model was
`the most efficient method for describing the data. The one compartment open model was
`the least efficient, particularly with respect to predicting the peak concentration of doxy-
`cycline in plasma. All the models gave similar rank order results with respect to bioavail-
`ability differences between the three treatments.
`Analysis of the data by different methods suggests that pelletised doxycycline is
`bioequivalent to the reference product when taken in the absence of food. A standardized
`feeding regimen affected the rate, but not extent of absorption of doxycycline from the
`pelletised formulation.
`
`KEY WORDS Doxycycline Absorption Bioavailability Food
`
`INTRODUCTION
`
`This study had three objectives. The first was to examine the bioavailability of
`two doxycycline products in fasted subjects and the second was to compare the
`effects of food on the bioavailability of one product. The third objective was to
`use the plasma doxycycline data to compare different pharmacokinetic methods
`for evaluation of bioavailability.
`
`*Reprint requests to: Dr Desmond B. Williams, F. H. Faulding and Co. Ltd, PO. Box 746,
`Salisbury, South Australia, 5108, Australia.
`
`0142~2782/90/020093~l3$06.50
`© 1990 by John Wiley & Sons, Ltd.
`
`ReceivedZ October 1987
`Revised 7 April 1989
`
`Amneal 1068
`
`Amneal v. Supernus
`|PR2013-00372
`
`‘
`
`

`

`.0.__
`
`94
`
`D. B. WILLIAMS ETAL.
`
`Doxycycline1s a lipid soluble antibiotic andis generally considered to be well
`absorbed (greater
`than 90%).1’2 When compared with other
`tetracycline
`derivatives, the absorption of doxycycline from the gastrointestinal tract appears
`to be less affected by food. Welling et (11.3 found that ingestion with test meals
`reduced the absorption of tetracycline and doxycycline by about 50 and 20 per
`cent, respectively.
`Other comparative bioavailability studies have been performed with doxycy—
`cline.4_6 Insufficient time (up to 24 h) was allowed for collection of blood
`following administration of doxycycline for pharmacokinetic analysis and no
`attempt was made to fit a pharmacokinetic model to the data1n these studies
`When pharmacokinetic analysis has been performed, a one compartment open
`model has usually been assumed.3'78 However, a two compartment open model
`was assumed when doxycycline was administered by intravenous infusion.9
`Numerous methods have been proposed to compare the bioavailability of
`various formulations of the same drug0The measurement of relative bioavail-
`abilityis normally determined by two variables the rate of absorption, and the
`extent of absorption of drug from the product.
`to the
`These are often determined by fitting a compartmental model
`concentration of drug in blood versus time data. Some of the model parameters
`(e. g. absorption rate constant) are used as the basis for comparing the rate aspect
`of drug absorption. The extent of absorption is measured by a comparison of
`areas enclosed by the concentration of drug in blood versus time profiles (AUC).
`More recently, techniques for evaluation of bioavailability known as either
`non—compartmental or model-independent (i.e. independent of compartment
`models) methods have been developed.10 Differences in the extent of absorption
`are evaluated from area measurements and the rates of absorption are compared
`by determining the time of peak concentration of drug in the blood. Useful
`parameters, such as the mean residence time (MRT), have been developed which
`allow some combination of rate and extent of absorption into a combined
`concept.11
`In this paper, a number of these methods were applied to the extensive data
`sets generated in the study. It was hoped to show that a variety of methods, if
`properly applied, will provide useful and equivalent information about the
`relative bioavailability of different pharmaceutical products.
`
`EXPERIMENTAL
`
`Subject protocol
`
`the study was performed by Biodecision
`The experimental part of
`Laboratories, Pittsburg, Pennsylvania. Twenty-four healthy male subjects
`entered the study. Four subjects (subjects 1, 12, 18, and 22) withdrew for reasons
`
`

`

`DOXYCYCLINE ABSORPTION
`
`95
`
`unrelated to the study. The twenty subjects who completed the study were from
`19 to 35 years (25-5 i 5-1 years, mean i SD) and weighed between 609 and
`864 kg (72-5 i 8-0 kg). Each subject was given a physical examination, a medical
`history was taken, and informed consent was obtained. Before each dose, the
`subjects were restricted to at least a 10 h fast, followed by a high protein, low fat
`FDA recommended diet for 48 h after each dose. The subjects were not
`permitted to take either antibiotics for at least 15 days or other drug products for
`at least 7 days, prior to the start of the study.
`One hundred milligrams of doxycycline was administered as doxycycline
`hyclate to each subject during each of three phases. Water (180 ml) was taken
`with each dose. Doxycycline was administered as the pelletised product (Doryx
`Capsules®, doxycycline hyclate delayed-release capsules, USP, F.H. Faulding &
`Co. Ltd, Batch 3ED157) with a standard breakfast for the first treatment.
`The pelletised product and the reference product (Vibramycin Capsules®,
`Pfizer Laboratories, Batch 23091) were administered 2h before a standard
`breakfast for the second and third treatments, respectively. All three treatments
`were administered during each phase according to a Latin square design and
`each phase was separated by 1 week.
`Blood samples were collected (Vacutainer®, Beckton Dickinson and Co.) by
`venepuncture immediately prior to drug administration and at 0-5, 1, 1-5, 2, 3, 4,
`6, 8, 12, 16, 24, 36, and 48 h following drug administration. The volume of each
`collection was
`10 ml, except
`for
`the first collection, which was 20 ml.
`Immediately after blood collection, the samples were stored in an ice bath and
`protected from light. Following centrifugation, the plasma was collected and
`stored at —20° until analysis. Urine was collected at —1 to 0, 0 to 1, l to 2, 2 to 4, 4
`t0 6, 6 to 8, 8 to 12, 12 to 24, and 24 to 48h following drug administration.
`During the collection period, all urine samples were stored at 4°. After
`measurement of the urine volume and pH, a 15 ml aliquot was taken from each
`sample and frozen until required for analysis.
`
`Determination of doxycycline in biologicalfluids
`
`Doxycycline concentrations in plasma and urine were measured by a micro—
`biological agar diffusion assay using Bacillus cereus ATCC 11778 as the test
`organism.12
`.52
`Standards were prepared from USP doxycycline hyclate reference material
`using 01 M potassium phosphate buffer, pH 4-5, as diluent. Plasma standards
`were prepared by spiking buffer standards with either 1:5 or 1:10 plasma which
`was free from doxycycline. Low concentration and high concentration controls
`were prepared by spiking plasma and urine with doxycycline. The controls were
`divided into aliquots and stored with the subjects’ samples.
`A set of standards and controls were processed on each analysis day. The
`samples were diluted with 0-1 M potassium phosphate buffer, pH 45, to fall
`
`

`

`96
`
`D. B. WILLIAMS ETAL.
`
`within the linear portion of the standard curve. The concentrations of
`doxycycline in plasma samples were determined by interpolation on curves
`obtained from standards containing the same proportion of plasma as the
`diluted sample. Urine samples were interpolated on a curve obtained from buffer
`standards. The measured concentrations were corrected for dilution.
`The lowest concentration of doxycycline which could be quantified with an
`acceptable degree of precision was 001% mg 1‘1. Any sample below this
`concentration was reported as zero. Day—to-day reproducibility of the assay was
`determined by analysing control samples. The coefficient of variation for the
`plasma assay was 5-5 per cent at a concentration of 0-60 mg 1’1 and 9-9 per cent at
`a concentration of 1.19 mg 1‘1. In urine, the coefficient of variation was 11-3 per
`cent at a concentration of 0'51 mg 1'1 and 13-7 per cent at a concentration of
`1.13 mg 1“.
`Recovery was determined by comparing the measured concentration and the
`expected concentration of each standard. Recovery on a typical analysis day in
`plasma ranged from 86 to 112 per cent and averaged 100 per cent. In urine,
`recovery ranged from 86 to 113 per cent and averaged 100 per cent.
`The assay was linear over the concentration range 000 to 0-70 mg 1‘1.
`
`Data analysis
`
`The data were subjected to curve peeling techniques by standard graphical
`methods and the use of ESTRIP.13 Two and three term polyexponential
`equations were fitted to the data. The ESTRIP progam generated estimates of
`the coefficients and indices of the equations and also estimates of the lag time
`(nag) before absorption occurs.
`_
`Areas under the curves of concentration of drug in plasma versus time
`(AUC38) up to the last sampling time were estimated by the trapezoidal method.
`The extrapolated areas to infinity (AUCK) were calculated by dividing the last
`measured concentration by the terminal phase rate constant. The MRT was
`calculated from
`
`MRT = AUMC/AUCii>
`
`(1)
`
`where AUMC is the area under the first moment curve.”’14 The AUMC was
`calculated by the method of Riegelman and Collier.14
`When the values of the coefficients, indices, and nag had been determined for
`each subject, the equation describing absorption and biexponential elimination
`was used to estimate the maximum concentration and the time at which the
`maximum concentration occurred. This was achieved by entering the equation
`into a computer program (available on re’quest) which calculated the
`concentration at increasing increments of time near the peak.
`The renal clearance of doxycycline was calculated from
`
`

`

`DOXYCYCLINE ABSORPTION
`
`A48
`CIR : _
`
`AUC. BW
`
`g
`
`'
`
`97
`
`(2)
`
`where A4,? is the amount of docycycline excreted in urine in 48 h and BW is the
`body weight.
`A number of compartmental models were fitted to the data as described
`below.
`
`One compartment model (model I). The simple one compartment model with
`first order input and output10 and including a lag time for absorption was fitted
`to the data. In the fitting procedure, the absorption rate constant, k3, elimination
`rate constant, kc,
`tag, and FDo/ Vb were entered as parameters, where F is the
`fraction of dose absorbed, D0 is the dose administered and Vb is the apparent
`volume of distribution of the drug. Initial values of these parameters were
`calculated using ESTRIP. Final values of these parameters were obtained from
`the best fit
`to the data as described under ‘Model fitting and statistical
`procedures’.
`The maximum concentration and the time at which,
`
`the maximum
`
`concentration occurred for each subject were calculated from standard
`equations. 10
`
`Two compartment open model with first order input (model 2). A standard
`triexponential equation10 with a nag incorporated was fitted to the data. The
`values obtained for slope constants, A1 and A2, were used as initial estimates of a
`and B, the complex constants of the two compartment model. The smallest
`index value obtained in the fitting procedure was assumed to be ,8. The value of
`at was equated with the index value closest to the known values of oz for
`doxycycline given by the intravenous route (mean = 0-80h‘1, reference 8). The
`remaining index value (in most cases the largest). was assigned as ka. In the
`with—food treatment,
`the a value used as a starting estimate in the fitting
`procedure was an average of the or values obtained with the same subject after
`fitting the other two treatments. In the fitting procedure, best fits were obtained
`for ka, oz, ,8, FDO/ Va, kzl, and'm'ag, where k21 is a distribution rate constant and V0
`is the volume of the central compartment.
`The constants km and k'lo were calculated from a, B, and k21 using standard
`methods.10
`The values obtained for the maximum concentration and the time at which
`the maximum concentration occurred were estimated as described for the non—
`
`compartmental parameters.
`
`TWO compartment open model with zero-order input (model 3). This model
`assumes that absorption occurs at a constant rate and that the absorption
`
`

`

`7‘.
`
`98
`
`D. B. WILLIAMS ETAL.
`
`process terminates at a definite point followed by an elimination phase.10
`
`(a) During the absorption phase equation 246 in Gibaldi and Perrier,15 was used
`with tlag
`incorporated. The zero—order absorption rate constant, kg, was
`calculated as the dose absorbed (Do) divided by the time of absorption:
`
`Do
`_
`k —______._
`T — [lag
`0
`
`'
`
`(3)
`
`i
`
`where Tis the time at which absorption stops. This is an approximate estimate
`of koz, as the absorption of doxycycline is generally slightly less than 100 per
`cent.
`
`(b) During the post—absorptive phase, equation 257 of Gibaldi and Perrier15 was
`used. In fitting these equations to the data, 0:, k21, Vc/ F and tiag were entered into
`the equations as parameters which were varied until the best fit was obtained.
`The dose of drug administered (Do), and the time for the end of absorption (D
`were entered as constants. The values of the starting parameters were those
`obtained for each subject from the fitting of the two compartment equation
`assuming first order absorption. T was determined as the time of peak
`concentration of doxycycline in plasma since this was the point at which
`absorption ceased and a decline in the concentration of doxycycline in plasma
`began.
`
`Modelfitting and statistical procedures
`
`The equations described above were fitted to the unweighted data by a Monte
`Carlo parameter estimation technique known as REVOL.16 The appropriate
`equations were written into the program and parameter estimates as described
`above were used to start the fitting procedure. In each case, the procedure was
`continued until a satisfactory fit was obtained. A number of criteria were used to
`determine the best fit. First, a fitted curve was required to be as free as possible
`from large regions of systematic error. Second, the smallest values obtainable for
`the error sum (ES) as defined in by16
`
`
`
`i: 1 m
`
`(4)
`
`where n is the number of data points, yr is the ith observation and ym is the
`calculated value for that observation.
`
`The Akaike Information Criterion (AIC) is
`determining the goodness of fit:17
`
`the third guide useful for
`
`AIC=nlnSS+2p
`
`(5)
`
`

`

`
`
`DOXYCYCLINE ABSORPTION
`
`99
`
`where n is the number of data points, SS is the sum of the squared deviation
`between observed and calculated observations and p is the number of parameters
`fitted in the model. The AIC may also be used to compare the relative suitability
`of different models.17 The lowest value of AIC generally indicates the model
`which best fits the data.
`
`Analysis of variance (ANOVA) used for complete crossover design was used
`on the various parameters from the individual treatments. Differences between
`individual treatments were determined using Tukey’s ‘honestly significant
`
`Table 1. Model independent parameters for doxycycline
`
`Parameters*
`
`ANOVAT
`
`Treatment means
`Pellet product
`Pellet
`product with without food
`food
`(A)
`
`
`Reference
`product
`without food
`(C)
`
`
`(B)
`
`27
`(4)
`
`31
`(6)
`
`39
`(10)
`
`0-33
`(0-08)
`
`0-55
`(0-24)
`
`1-8
`(05)
`
`1-2
`(0-5)
`‘
`
`1-7
`(0-4)
`
`22
`(4)
`
`27
`(4)
`
`31
`(5)
`
`38
`(7)
`
`0-33
`(0-09)
`
`0.27
`(0- 17)
`
`1
`
`1-4
`(0-6)
`
`11
`(0-5)
`V
`
`1-8
`(0-4)
`
`22
`(4)
`
`m1
`
`ns
`
`ns
`
`ns
`
`p <0-001
`(A>B>C)“
`
`.. p <0-001
`(A>B=C)
`
`0-002 < p < 0005
`A>B=C
`
`0002 <p <0-005
`(A <C,A=B,B=C)
`
`ns
`
`AUcé8 (mg h 1“)
`
`AUC‘S’ (mg h 1*)
`
`Aui8(mg)
`
`CIR (m1 mm“l kg—l)
`
`nag (h)
`
`rm (h)
`
`tmax — hag (h)
`
`Cp max (mg 1“)
`
`MRT (h)
`
`26
`(3)§
`
`29
`(4)
`
`38
`(8)
`
`0-34
`(0-08)
`
`1-1
`(0-37)
`
`31
`(0-9)
`
`19
`(0-9)
`1w
`
`1-4
`(0-6)
`
`22
`(3)
`
`*See text for definition of symbols.
`TAnalysis of variance.
`iNot significant.
`§Standard deviation.
`llTukey’s test.
`
`
`
`

`

`100
`
`D. B. WILLIAMS ETAL.
`
`(mg/l)
`
`PlasmaDoxycycline
`
`0
`
`1o
`
`20
`
`30 ‘
`
`4o
`
`50
`
`Time (h)
`
`Figure 1. Plasma concentrations of doxycycline following administration of the pellet product with a
`standard meal to 20 subjects. The data are shown as means and standard deviations and the solid line
`was generated by assuming a two compartment open model with first order absorption
`
`difference’ test.18 All computing procedures were run on either a RSTS system or
`a Sirius 1 microcomputer.
`
`RESULTS AND DISCUSSION
`
`Visual examination of the plasma doxycycline concentration versus time plots
`(Figures 1, 2, and 3), area analysis of individual curves (AUC‘iig and AUC, Table
`1) and the total drug recovered in urine (A4u8, Table 1) all support the View that
`the extent of absorption of each product is equivalent. The MRT values (Table
`l) are also consistent with this result.
`As usual in the evaluation of bioavailability by area analysis, it is assumed that
`averaged clearance (and hence volume of distribution) is unchanged between
`treatments.10 The lack of variation in renal clearance between treatments (Table
`
`(I
`
`it
`
`la
`
`

`

`DOXYCYCLINE ABSORPTION
`
`101
`
`(mg/l)
`
`PlasmaDoxycycline
`
`Time (h)
`
`Figure 2. Plasma concentrations of doxycycline following administration of the pellet product 2h
`before a standard breakfast. The data and fitted curve were as described in Figure 1
`
`1) would support the assumption of constancy of total clearance in this study.
`Area analysis is related to the total amount of drug passing through the system
`but gives no information about the rate of absorption of drug from the gut. The
`rate of absorption is related to the time of peak drug concentration in blood
`(tmax) which is estimated from the data plots. The values of tmax indicate that food
`significantly slows down themabsorption of the pellet product while there is no
`marked difference between the absorption rates of the products when fasting.
`The same conclusion is obtained when the tmax figures are corrected for an initial
`delay in drug release (llag) calculated by the polynomial curve fitting program.
`The corrected values (tmax — tlag, Tables 1 and 2) indicate the same differences in
`absorption rate as tmax- The tlag values (Tables 1 and 2) indicate a delay in the
`initiation of absorption by food for the pellet product. In fasting subjects, the tlag
`for the pellet product Was longer than the reference product. There was no
`correlation between either the urinary pH or urinary volume and doxycycline
`elimination.
`
`

`

`102
`
`D. B. WILLIAMS ETAL.
`
`Compartment models and bioavailability
`
`Three compartment models were fitted to the data. The one compartment
`model (model 1) and the two compartment model with zero order input (model
`3) showed a considerably poorer fit
`to the data compared with the two
`compartment model with first order input (model 2). Hence the latter is used in
`this discussion to evaluate bioavailability. Similar bioavailability determinations
`were made with the former models.
`
`Goodness of fit for the three models was determined by three methods. (a)
`Systematic error was least evident with model 2, particularly in the region of
`peak drug concentration and during the post peak distribution phase. (b) The
`error sum estimates (ES) were smallest for model 2. (c) The AIC values were
`smallest for model 2.
`
`(mg/l)
`
`PlasmaDoxycycline
`
`0
`
`10
`
`20
`
`30
`
`40
`
`50
`
`Time (h)
`
`Figure 3. Plasma concentrations of doxycycline following administration of the reference product 2 h
`before a standard breakfast. The data and fitted curve were as described in Figure 1
`
`

`

`DOXYCYCLINE ABSORPTION
`
`103
`
`Table 2. Parameters obtained assuming a two compartment open model with first order
`absorption for docycycline disposition
`
`Parameters*
`
`Treatment means
`7
`Pellet product
`Pellet
`Reference
`product with without food
`product
`food
`without food
`
`(A)
`(B)
`(C)
`
`ANOVAT
`
`ka (h"‘)
`
`oz (h'l)
`
`B (11“)
`
`FD / Vc (mg 1“‘)
`
`km (h“‘)
`
`km (11“)
`
`k12 (h“‘)
`
`as (h)
`
`1m (h)
`
`rm 4 hag (h)
`
`1-4
`(09)::
`
`0-60
`(0-20)
`
`0-046
`(0-007)
`
`2-2
`(0-5)
`
`0075
`(0-021)
`
`0-37
`(012)
`
`022
`(0-13)
`
`1-13
`(0-35)
`
`3-1
`(08)
`
`2-00
`(075)
`
`2-6
`(1-3)
`
`0-79
`(0-32)
`
`0047
`(0-008)
`
`2-6
`(1-0)
`
`0-084
`(0037)
`
`0-44
`(0-11)
`
`0-31
`(0-26)
`
`0-53
`(023)
`
`1-6
`(0-4)
`
`-
`
`1-08
`(0-33)
`
`2-4
`(1-1)
`
`084
`(0-27)
`
`0-002 <p <0~005
`(A <B=C)§
`
`0-01 <p <0-05
`(A <B <C)
`
`0046
`(0006)
`
`ns H
`
`2-7
`(0-6)
`
`0-087
`(0-020)
`
`0-45
`(0-16)
`
`035
`(016)
`
`030
`(014)
`
`1-5
`(07)
`
`1-18
`(071)
`
`ns
`
`ns
`
`ns
`
`ns
`
`p < <0-001
`(A> 13> C)
`
`p < <0-001
`A > B=C
`
`p $0001
`A > B=C
`
`0-002 <p <0-005
`1-9
`18
`1-5
`cpmax (mg 1")
`A <B=C {us—4r
`(0-4)
`(0-4)
`(0-2)
`
`*See text for definition of symbols.
`“(Analysis of variance.
`iStandard deviation.
`fiTukey’s test.
`Not significant.
`
`The parameters obtained using model 2 (Table 2) which were relevant to the
`absorption rate of docycycline (i.e., ka, tlag, tmax, tmax—tlag, and Cm“) all showed the
`same relative differences between treatments as observed with the model
`
`

`

`
`
`104
`
`D. B. WILLIAMS ET AL.
`
`independent analysis (Table 1). Those parameters concerned with distribution
`and elimination processes (i.e.,
`,8, km, klz, and k21) showed no significant
`difference between treatments. The only exception was a which was significantly
`different between treatments in the same order as ka varied. This discrepency
`probably reflects the difficulty of separating oz and ka, particularly where 0: and
`ka approach each other in magnitude. The pharmacokinetic models used in this
`study assumed a continuous elimination phase, however, some subjects showed
`discontinuous jumps in concentration during the elimination phase. It has been
`reported that this observation is (consistent with a discontinuous absorption
`mechanism occurring,
`such as
`from enterohepatic cycling through the
`gall-bladder.”
`.
`....
`
`CONCLUSIONS
`
`Each of the pharmacokinetic models tested suggests that doxycycline is
`absorbed at the same rate and extent from the pelletised product and the non-
`pelletised product when they are taken in the absence of food. Doxycycline is
`absorbed to an equivalent extent but at a slower rate when taken in the form of
`the pelletised product with food.
`'
`The results from this study suggest that different pharmacokinetic models may
`be used to draw similar conclusions about the bioavailability of doxycycline
`following oral administration from different dosage forms. Comparison of the
`systematic errors, error sum estimates, and AIC values suggested that the two
`compartment open model which included a lag time and a first order absorption
`process best fitted the plasma doxycycline data.
`
`ACKNOWLEDGEMENTS
`
`The authors are indebted to Dr R. Don Brown for the ESTRIP program, and to
`the staff at Biodecisions Laboratories, Pittsburg, Pennsylvania,
`for their
`cooperation in performing this study.
`
`REFERENCES
`
`1. M. Schach von Wittenau, Chemotherapy (suppl.), 13, 41 (1968).
`2. B. A. Cunha, C. M. Sibley and A. M. Ristuccia, Ther. Drug Monit., 4, 115 (1982).,
`3. P. G. Welling, P. A. Koch, C. C. Lau and W. A. Craig, Antimicrob. Agents Chemother., 11, 462
`(1977),
`-
`4. J. D. Arcilla, J. L. Fiore, 0. Resnick, J. W. Nadelmann, J. L. Huth and W. M. Traetel, Curr.
`Ther'. Res., 16, 1126 (1974).
`5. R. Kitzes Cohen, F. Grauer and E. Weisenberg, Israel J. Med. Sci” 16, 545 (1980).
`
`

`

`DOXYCYCLINE ABSORPTION
`
`105
`
`6. A.-S. Malmborg, Chemotherapy, 30, 76 (1984).
`7. M. Gibaldi, Chemotherapia, 12, 265 (1967).
`8. G. Ceccarelli, R. Rossoni, F. Ronita and A. Naddeo, Chemotherapy, 16, l (1971).
`9. T. C. Raghuram and K. Krishnaswamy, Br J Clin Pharmacol, 14, 785 (1982).
`10. M. Gibaldi and D. Perrier, Pharmacokinetics, 2nd edn. Marcel Dekker, New York, 1982.
`11. K. Yamaoka, T. Nakagawa and T. U110, J. Pharmacokinet. Biopharm., 6, 547 (1978).
`12. Code of Federal Regulations, U.S. Government Printing Office, Washington, 1984, p. 244.
`13. R. D. Brown and J. E. Manno, J. Pharm. Sci, 67, 1687 (1978).
`14. S. Riegelman and P. Collier, J. Pharmacokinet. Biopharm., 8, 509 (1980).
`15. M. Gibaldi, D. Perrier, Pharmacokinetics, lst edn. Marcel Dekker, New York, 1975.
`16. P. Koeppe and C. Harnann, Comput. Prog. Biomed., 12, 121 (1980).
`17. K. Yamaoka, T. Nakagawa and T. Uno, J. Pharmacokinet. Biopharm., 6, 165 (1978).
`18. R. P. Sokal and F. J. Rohlf, Biometry, 2nd edn. W. H. Freeman, San Francisco, 1981.
`19. J. M. Jaffe, J. L. Colaizzi, R. I. Poust and R. H. McDonald, J. Pharmacokinet. Biopharm., l
`267 (1973).
`20. J. M. Jaffe, R. I. Poust, S. L. Feld and J. L. Colaizzi, J. Pharm. Sci, 63, 1256 (1974).
`21. J. G. Wagner and C. M. Metzler, J. Pharm. Sci, 56, 658 (1967).
`22. P. Veng Pedersen and R. Miller, J. Pharm. Sci., 69, 204 (1980).
`
`a
`
`Its-m
`
`
`
`