- Email: [email protected]
Bioavailability and Disposition of Terodiline in Man
Bioavailability and Disposition of Terodiline in Man B. HALL EN'^, M. 0. bRLSSONS, s. STROMBERGS,
AND
B. NORENS
Received June 24, 1993, from the Kabi Pharmacia A6, Departments of *Pharmacokinetics and §Structural Analysis, R&D, 5-112 87 Stockholm. Sweden. and 'Deoartment of 6io~harmaceufics and Pharmacokinefics, University of Uppsala, $751 23 Uppsala, Sweden. ' Accepted for publication May' 23, 1994? unlabeled terodiline, was given orally in study A and intravenously in study B. The solution for study A contained 0.565 mg/mL (ratio [3Hlterodiline/[1Hlterodiline ca. 0.008%) or 0.182 MBq/mL. The solution with L3H1terodilinefor study B contained 2.48 mg/mL (ratio [3Hlterodiline/[1Hlterodilineca. 0.003%)or 0.666 MBq/mL. Deuteriumlabeled terodiline (N-tert-butyl-l-([2H~lmethyl)-3,3-diphenylpropylamine hydrochloride) was prepared using [2H31methyliodide as the deuterium-containing reagent.l0 The identity and isotope purity (>99.9%) were verified before use. The aqueous solutions of [2H]terodiline contained 2.5 mg/mL. Solutions with unlabeled terodiline for intravenous administration contained 2.49 mgfL. All solutions for intravenous and oral administrations were specifically prepared for this study by Kabi Pharmacia. Quality-control procedures of all terodiline formulations were carried out to assure that the formulations met the requirements for human use. Protocol-The subjects were normal healthy volunteers. The mean age and body weight were similar (Student's t-test, not significant) in studies A and B (50 2= 11y, 75 5 kg and 54 d~8 y, 76 f 9 kg). No pharmaceutical products were used 48 h before dosage and during the following study period. The subjects were fasting from 22.00 h on the evening before the study until 4 h after the dosage, when they were given lunch. Study A-The terodiline solution (20 mL) with a tracer dose of [3H]terodiline was given orally followed by 200 mL of water to six subjects (nos. 1-6; nos. 2 and 6 were females). The exact volume was calculated by weighing the bottles before and after intake and was found to correspond to a dose of 11.1 mg (3.59 MBq). The blood sampling schedule was a s follows: 0 (predose), 20, and 40 min, 1, 2, 3, 4, 6, and 8 h, and 1, 2, and 3 days. Serum was, after coagulation of the blood at room temperature, separated by centrifugation and was kept deep frozen (-20 "C) until the time of analysis of terodiline. Introduction and 12Urine was quantitatively collected at 0-2,2-4,4-8,8-12, 24 h and then in 24-h intervals up to 24 days. Prior to dosage, blank Terodiline is an anticholinergic and calcium antagonistic samples of urine and feces were collected. drug used to treat urge incontinence.1-6 The oral pharmacoStudy B-[2HlTerodiline, 12.5 mg in a 5-mL water solution, was kinetic profile has been investigated in healthy volunteers7 given orally to 10 subjects (nos. 1-3, 5, 7-12; nos. 2,6,8, and 9 were and in geriatric patient^.^,^ Due to its low clearance and long females). Terodiline was given intravenously with a tracer dose of 3H-labeled drug to four subjects (nos. 1-3,5) and solely as unlabeled half-life, conventional comparative studies (e.g. oral us intraterodiline to six subjects (nos. 7-12). The intravenous infusion venous) have to be awkwardly extended in time for adequate (Terufusion syringe pump STC-521; 1mumin for 5 min, giving a total determination of the profile (long wash-out period). The dose of 12.5 mg) was started at the same time as [2Hlterodiline was design of the present study has utilized the advantages of the ingested orally. The blood sampling schedule was as follows: 0 stable-isotope approach to minimize possible variability in(predose); 1, 2, 3, 4,and 5 min (during infusion); and 5, 10, 15, 30, troduced by treatment periods being widely separated. The and 45 min, 1, 2, 3 , 4 , 6 , and 8 h, and 1, 2 , 3 , 4 , 7 , 9, 11,and 14 days present study is composed of two parts: A, oral mass balance, (after infusion). Urine (subject nos. 1-3, 5, 7-12) and feces (subject and B, oral bioavailability and intravenous mass balance. The nos. 1-3, 5) were collected in 24-h portions quantitatively up to 20 days. objectives of the study are to (i)determine the bioavailability The study was performed according to the "Declaration of Helsinki" of oral terodiline using the stable-isotope technique, (ii) and later revisions. The general study plan and the written informainvestigate the kinetics of terodiline after intravenous injection given to the volunteers was approved by the Ethics Committee tion, (iii) quantify the amounts of terodiline-derived radioacat Uppsala University and the Swedish National Board of Health and tivity excreted via urine and feces after oral and intravenous the Isotope Committee at the Karolinska Hospital. administration, and (iv) identify major metabolites in feces. Terodiline Analyses in Serum and Urine-The serum and urine samples were analyzed for the content of [lHIterodiline and [2H]terodiline with a gas chromatographidmass spectrometric method. Experimental Section The method comprises a two-step partition with acid and base followed by acylation with trifluoroacetic anhydride (TFAA). The Materials-Tritium-labeled terodiline (N-terf-butyl-l-([3H31methyl)- method is essentially the same as previously described.12 However, 3,3-diphenylpropylamine hydrochloride) was prepared with 3H-labeled capillary column gas chromatography was used instead of packedmethyl iodide as the tritium-containing reagent.1° The labeling column GC. A Hewlett-Packard 5890 gas chromatograph equipped position is known to be metabolically stable. l1 The radiochemical with a Hewlett-Packard 5970 mass selective detector was used: SIM purity was > 99%. Tritium-labeled terodiline, supplemented with mode; ionization energy, 70 eV; transfer line temperature, 275 "C. The analyses were performed with 1-min splitless injection onto a H P Ultra capillary column (methyl silicon gum, 25 m x 0.32 mm, Mailing address: Kabi Pharmacia, S-751 82 Uppsala, Sweden. Abstract published in Advance ACS Abstracts, July 1, 1994. 0.11-pm phase thickness; injector temperature, 250 "C; column
Abstract 0 Terodiline was concomitantly administered intravenously (12.5 mg) and orally ([2H]terodiline, 12.5 mg) to 10 healthy volunteers. In four of the subjects, a tracer dose of the intravenously given terodiline was 3H-labeled. In a separate study, six subjects were given [3H]terodiline orally. Estimated pharmacokinetic parameters were as follows: systemic clearance, 93 mumin; renal clearance, 14 mumin; volume of distribution at steady-state, 407 L; terminal half-life, 54 h; and mean residence time, 77 h. After intravenous infusion, a rapid distribution phase (half-life, 4.5 min) could be observed. The maximum serum concentration after the oral dose was 29 vg/L and the time to maximum concentration was 5 h (estimated by noncompartmental analysis). Absorption commenced within the first hour and by deconvolution the maximum rate of absorption was determined to occur between 1 and 3 h, and by 3.4 h 90% of the available dose had been absorbed. Calculation of bioavailability by noncompartmental AUC, two-compartmental analysis, urinary excretion, and 24-h oralhtravenous concentration ratio gave similar results (ANOVA test, not significant). About 75% and 25% of administered radioactivity could be recovered in urine and feces, respectively. Intact terodiline in feces accounted for about 1% of the dose. pHydroxyterodiline was quantitated in feces and accounted for about 5% of the dose. Another metabolite, 3,4-dihydroxyterodiline, which has not previously been detected in urine or serum, was also identified.
*
@
0 1994, American Chemical Society and American Pharmaceutical Association
0022-3549/94/1200-1241$04.50/0
Journal of Pharmaceutical Sciences / 1241 Vol. 83, No. 9,September 1994
temperature program, 85 "C (1min), 40 "C/min to 300 "C, program stops after 10 min; flow rate of the helium carrier gas, 0.9 mumin). The detection was performed in the SIM acquisition mode a t the 321 ion for terodiline and 330 for internal standard, N-tert-butyl-3,3-bis([2Hs]phenyl)-1-methylpropylamine . The linearity expressed as the slope of standard curves shows a coefficient of variation of 4.9% with a mean regression coefficient of 0.9996 (CV% = 0.05%). The limit of quantitation, LO&, is 1pg/L (signal to noise ratio 5). The accuracy for serum analysis varies between 97 and 117%. The precision is 22% a t LOQ, about 7% at 3-50 pgL, and 3% at 50-200 pg/L. The system reproducibility is 3.3% (CV%). The accuracy for urine analysis is 106-119% and the precision is 3-5% in the interval 500-5000 p g L . Terodiline and Metabolites in Feces-Recovery of [3HlTerodiline after Addition to Homogenates of Blank Fece~-[~H]Terodilinewas added to 1-5 g of blank feces homogenate (study B, subject no. 5, day 0) and extracted according t o the bioanalytical method for terodiline. The effect of addition of 3 M trifluoroacetic acid the day before extraction was also studied. Radioactivity in Supernatant of Nonhydrolyzed Feces HomogenateWater (25 mL) was added to feces homogenate (study B, subject no. 5, day 3) and was sonicated for 1h. After 1day at room temperature, the homogenate was centrifuged and the clear supernatant was evaporated in uacuo at 50 "C. The radioactivity was measured by liquid scintillation counting. Extractable Radioactivity from Supernatant of Hydrolyzed Feces Homogenate-Homogenates (study B, subject no. 5 , day 3) were hydrolyzed with 3 M trifluoroacetic acid overnight a t 100 "C. After sonication for 1 h and centrifugation, aliquots of supernatant were evaporated in uucuo at 50 "C. The residue was washed with water and then combusted in Packard's sample oxidizer for measurement of radioactivity. To a n aliquot of the supernatant were added sodium hydroxide and sodium hydrogen carbonate until pH 12, and extractions were performed with diethyl ether (3 x 10 mL). Identification of Metabolites in the Supernatant of Hydrolyzed Feces Homogenate-The supernatant from feces homogenates (study B, subject no. 5, day 3), which had been hydrolyzed and sonicated for several hours, was evaporated in uacuo and dissolved in methanol and water (1.5 + 1.5 mL). The supernatant was run through a n HPLC system (methanol, water, trifluoroacetic acid, 1 1 0.1%) with a preparative C18 column UV 279 p with radioactive detection. Fractions were collected and GLCMS analyses were performed. Quantitative Determinations of Terodiline and p-hydroxyterodiline were made with homogenates (study B, subject no. 5, days 0-4) after hydrolysis with trifluoroacetic acid (3-6 M) at 100 "C by applying the bioanalytical method for terodiline. Terodiline and p-hydroxyterodiline were stabile in this medium for at least 12 h. Radioactivity Measurements-The 3H-activity was determined in the urine and feces samples. Urine (1mL) was added to 15 mL of Monophase (Packard Instrument) and the radioactivity was determined with a liquid scintillation spectrometer (Packard Tri-Carb 300) using the external standard for correction of quenching. Feces was homogenized in deionized water (1:5) for 2 min using a Stomacher Lab-Blender 3500 or 400 (Seward Medical). Samples of the suspension (2 mL) were allowed to dry in combustion cones and were oxidized together with some cellulose powder in a Packard Oxidizer 306, whereupon the radioactivity in the samples was determined. At regular intervals, tritium standards for sample oxidizers (study A, Amersham, VK; code TRR.lO1; study B, S~ec-Chec-~H, Packard) were combusted to check the combustion efficiency, which was found to be 90% (study A) and 93% (study B). PharmacokineticAnalyses-Noncompartmental Analysis (Studies A and B)-Maximum serum concentration of terodiline and the time of its appearance were those actually observed. log-linear regression analyses of the slopes of the serum concentration time curve from the 24-h sample and onward were used to calculate the half-life of elimination (tliz). The area under the serum concentration time curve from zero to the last measurable concentration was calculated by the linear trapezoidal method.13 The residual area from time t to infinity was calculated from (regression-line-estimated concentration at time of last measurement)/(ln 2/t1/2). For the purpose of estimation of the radioactivity in urine and feces after the period of quantitative collection, the extrapolated amount was estimated as (regression-line-estimated excretion rate on last day of collection)/(ln 2/t1/2). The determination of the half-life of the
+ +
1242 / Journal of Pharmaceufical Sciences Vo/.83, No. 9, Sepfember 1994
terminal monoexponential part of the rate plots was performed by log-linear regression analysis. The bioavailability in study B was calculated from the ratio of AUC,JAUCi, or from the amount excreted in urine (AepdAei,). These two methods are both based on the amounts eliminated. The bioavailability was also estimated from the 24-h oraVintravenous serum concentration ratio, C24,JC24iV. This single-time method is based on the amount in the body (Wekt)at a time (t) when absorption and distribution have ceased and the amount eliminated relative to the final amount eliminated is close enough to be regarded as equal after oral and intravenous dosage. A time point with sufficient precision that approximates these prerequisites can be found for a drug with absorption over a few hours and a long half-life and low clearance, like terodiline. The time point was chosen by visual inspection of the semi-log serum concentration time curves (see also the first paragraph in Noncompartmental Analyses). The oral/ intravenous concentration ratio method would for may drugs give a falsely high estimate of the bioavailability, since more drug, relative to the amount finally eliminated, is generally eliminated following intravenous administration compared to after oral administration at a time when the absorption phase is finished. However, if the absorption and distribution are very rapid relative to elimination, the error would be small. Compartmental Analysis (Study B)-Two approaches were tested to estimate the intravenous pharmacokinetic parameters (A, a ; B, p) used in the subsequent deconvolution analysis: (i) only postinfusion data with correction for the time of infusion and (ii) the complete infusion profile. Mono-, bi, and triexponential functions were fitted to the data by the program PCNONLIN, version 3.0,14 and using standard e q ~ a t i 0 n s . l The ~ Nelder-Mead algorithm was used to minimize the sum of squared residuals. Discrimination between the structural models and weighting schemes (1, lly, and l/y2, where y is the calculated concentration) was based on the goodness of fit, residual analysis, and the standard deviation of estimated parameters. Simultaneous fitting of intravenous and oral serum concentration time profiles were performed for estimation of the bioavailability with a model-dependent approach using a first-order absorption model. Parameters used in this fit were lag time for absorption, bioavailability, first-order absorption rate constant, disposition rate constants, fractional intercept, and initial volume of dilution. The following secondary parameters were ca1culated:l3 clearance, volume of distribution at steady state, mean residence time, and half-lives corresponding to the absorption phases a-phase and p-phase. The renal clearance was determined by using the estimated serum concentrations obtained from the simultaneous fit of the intravenous and oral models as constants in the following equation: urinary excretion rate = renal clearance x serum Concentration. Renal clearance was estimated from simultaneous fitting to the intravenous and oral urine data, which were weighted by the reciprocal of the excretion rates. Deconvolution (Studies A and B)-Decon~olution~~ was performed with data from studies A and B. The input rate function was determined from the convolution integral c(t) = flt)*Cs(t), where c(t) is the observed oral serum concentration profile, flt) is the input rate function, and C&) is the unit impulse function described by parameter estimates from PCNONLIN curve fits of the intravenous postinfusion profile. The bioavailability estimated from deconvolution was calculated as a mean value of the estimated cumulative amount absorbed from the time point when apparent plateau was reached. Statistical Analysis-The methods used for calculation of bioavailability were compared with one-way ANOVA. Model discriminations were made by F-test of the full and reduced m o d e P for each subject besides the standard procedures described in Compartmental Analysis. Other estimates were evaluated with Student's paired or unpaired t-test. Differences were regarded as statistically significant for p < 0.05. Descriptive statistical analyses have been made by calculations of means and standard deviations.
Results Study A-Pharmacokinetics and Radioactivity Measurements- The maximum serum concentration of terodiline was 38 f 9 ,ugh and the corresponding time point 6 f 3 h. The percentages of the given oral dose recovered as radioactivity (extrapolated t o infinity) in urine and feces were 69.9 f 6.3
Table 1-Oral Bioavailability (%) of Terodiline Assessed by Different Methods from Serum or Urine Data' % Bioavailability
Subject No.
AUC
Ae
Dcon
c24
2cpt
1 2 3 5 7 8 9 10 11 12
93 76 77 76 94 101 91 85 99 113
99 79 81 88 105 103 104 87 99 115
100 79 78 78 104 108 100 93 104 113
91 72 74 76 97 98 99 84 102 102
98 79 78 75 99 101 106 90 105 103
Mean rt SD Median Range
90k12 92 76-1 13
96 k 12 99 79-1 15
96k13 100 78-1 13
89+12 94 72-1 02
93+12 98 75-1 06
aANOVA test of AUC, Ae, C24, Dcon, and 2cpt. p = 0.673. AUC = noncompartmental determination of area under serum concentration time curve. Ae = noncompartmental determination of urinary excretion. c24 = serum concentration 24 h after administration. Dcon = deconvolution of serum concentrations (mean value). 2cpt = compartmental analysis of serum concentrations.
and 26.1 f 7.3, respectively. The half-life of radioactivity in urine and feces was 65.1 f 10.1 and 84.8 f 10.3 h, respectively, which means that the extrapolated amount beyond the last sampling interval is minimal. The total recovery from urine and feces was 96.2 f 3.3%. Study B-Pharmacokinetics: Noncompartmental Analysis- After intravenous infusion the observed maximum serum concentration was 211 f 161pg/L (range, 38-448 pg/L). After oral administration, the peak serum concentration was 29 f 5 p g L after 5 f 2 h. The extrapolated area under the serum concentration time curve beyond the last measurable concentration constituted 8 and 7% after intravenous and oral administration, respectively. The oral bioavailability estimated from AUC ratios was 90 f 12%(Table 1). The urinary excretion was 1.95 f 0.64 mg following the intravenous dose and 1.88 f 0.66 mg after the oral dose. The bioavailability estimated from urinary excretion ratios was 96 f 12%. Pharmacokinetics: Compartmental Analysis-The intravenous data were best described by fitting a biexponential function t o postinfusion data (F-test, p < 0.05). The fitted parameters were fractional intercept, initial volume of dilution, and disposition rate constants, and the weight used was lly. The disposition parameters obtained when simultaneously fitting the intravenous and oral models were unaltered (paired t-test, not significant or deviation < 0.3%)
compared to those estimated when fitting separately (weight, Uy). The parameters are given in Table 2. Examples of model-fitted serum and urine data are given in Figure 1. The point estimates are as follows: lag time for absorption (Tlag), 0.77 h; absorption half-life (tl/2&), 1.3 h; initial volume of dilution (Vc),91 L; volume of distribution a t steady-state (Vss), 407 L; half-life of a-phase (tl/zm),4.5 min; half-life of B-phase (tlizp), 54 h; mean residence time (MRT), 77 h; systemic clearance (CL), 93 mumin; and renal clearance (CL,), 14 m u min. The bioavailability was 93% (Table 1). Comparison of serum concentrations during infusion (using parameters obtained from postinfusion data) shows a discrepancy between observed and predicted concentrations (Figure 2a). Figure 2b shows that the time course to achieve the predicted concentration (the mixing time) can be approximated to an exponential process with an average half-life (t1lzrnk)of 0.8 min (Table 3). A lag time of 1.1min between the start of infusion and appearance in serum was found. The time course to reach the predicted concentrations was also analyzed by estimating the time (2'0.9) when observed and predicted serum concentrations converged, i.e. was within 90%. T0.9 was found to have a mean value of 3.4 min. The estimations of 2'0.9 and t l ~ are ~ kgenerally in agreement (2'0.9 = 3.4 min corresponds to a tl/Zmk= 1.0 min). The observed AUC during infusion (0-5 min) is 74% of the predicted AUC. However, AUC during infusion is only a minute part of AUCo-,. (about 0.3%). The fraction of the intravenous dose excreted in urine (CL,./ CL) was 16.6 f 5.2%, which when considering the excretion of radioactivity, indicates a substantial excretion of metabolites. It can be noted that the observed excretion rates occasionally deviated from the model calculated rates, e.g. 12 h excretion rate for subject no. 3 (Figure lc). A possible explanation may be spontaneous changes in urinary pH, which will influence the renal clearance of terodiline. However, the design of the study (concomitant administration of iv and oral drug) will compensate for these kinds of effects on, for example, bioavailability estimates. Pharmacokinetics: Deconvolution-The deconvolution of the oral serum concentration time profiles with the unit impulse response (intravenous bolus dose) resulted in a staircase input function. In all subjects the absorption rate decreased after 1-3 h. The maximum absorption rate was reached after 1.5 h (Figure 3a). The final average cumulative amount absorbed was 89 f 12%. Within 3.4 h, 90% of this amount had been absorbed (Figure 3b). Inspection of the complete profiles (until day 14) revealed unphysiological negative "absorption rates" in some cases when absorption could be expected to have ceased. However, no differences in the estimates were obtained (paired t-test, not significant) when using 24 h or
Table 2-PCNONLIN Estimates of Pharmacokinetic Parameters of Terodiline Subject No.
1 2 3 5 7 8 9 10 11 12 Mean SD CV% Mean RSD% of point estimate
7, (h)
tl/Zabs
(h)
V, (L)
V,, (L)
ha(min)
fil2p (h)
MRT (h)
CL (mUmin)
C L (mumin)
0.56 0.89 0.57 0.86 0.95 0.77 0.61 0.75 1.oo 0.74
1.45 1.47 2.03 1.30 0.90 1.28 1.19 1.71 0.95 0.76
97.1 13.1 15.2 12.5 190.0 32.3 267.0 169.0 89.3 28.2
397.0 429.0 315.0 294.0 308.0 354.0 613.0 513.0 370.0 476.0
5.6 1.6 1.7 1.7 16.0 3.7 3.8 5.5 2.9 2.6
55.3 50.3 54.7 56.4 36.8 65.5 38.8 45.5 52.1 84.0
79.4 71.4 78.1 80.5 52.9 93.5 55.9 65.3 74.9 120.2
83.3 100.3 67.3 60.8 97.1 63.1 182.7 130.9 82.4 66.0
16.0 13.9 12.6 10.7 8.4 13.8 11.8 26.3 16.5 12.5
0.77 0.15 20 17
1.30 0.39 31 31
91 90 98
407 102 25
11
6
4.5 4.3 94 20
53.9 13.6 23 9
77 19 25 9
93 38 41 8
14 4.9 3 5
Journal of Pharmaceutical Sciences / 1243 Vol. 83, No. 9, September 1994
0
2
1
3
6
5
4
Minutes
'oool
10
100
7: B
y = 3 6808
.
H"2 = 0 893
10A10 4 4 3 4 8 x 1
1:
10
1
-
'
n
0
'
50
l
.
-
I
100
Time
150
200
250
300
!
'
1
1
'
2
1
'
3
I
'
I
'
l
6
5
4
Minutes
(hours)
Figure 2-(A) Lag time of serum concentration during infusion (0-5 min) in subject no. 3. By linear regression between observed consecutive time points, the time to achieve 90% of the predicted concentration can be estimated to be 3.1 min (the mean value for nine subjects is 3.4 min). The serum profile predicted from postinfusion pharmacokinetic parameters is indicated (- - - -). (B) Semiiog plot of fraction remaining to predicted concentration (1 - CobJCpred)during infusion (0-5 min) in subject no. 3. From the regression line, a half-life of the mixing process can be estimated to be 0.7 rnin (mean value for nine subjects is 0.8 min, 12 = 0.95).
C
1000
. 0 1 i . 0
100
Table 3-Discrepancies between Observed and Predicted Serum Concentrations of Terodiline during Infusion (0-5 miny
10
0
100
200 Time
300
400
500
(hours)
Figure 1-Simultaneous tit of a two-compartment model to intravenous (A)and oral (0)serum concentrations in the time scales 0-8 hours (A) and 0-264 hours (11 days) (B) for subject no. 3. (C) Simultaneous fit of a two-compartment model to intravenous (A)and oral (0)urinary excretion rates in Subject no. 3.
the complete sampling period, i.e. up to 14 days for the deconvolution. Radioactivity Measurements-Radioactivity determinations in urine and feces showed that the percentages recovered (extrapolated to infinity) after intravenous administration were 74.2 f 6.8 and 19.4 f 4.2, respectively. The extrapolation from day 20 to infinity (percent of dose) was <1%for urine as well as for feces data. The total recovery from urine and feces was 93.5 f 8.1%. The half-life of radioactivity in urine and feces was 62.5 f 3.2 and 85.3 f 14.5 h, respectively. The fraction of administered dose absorbed into the hepatoportal system (Fa)was estimated from the oral/iv ratio of urinary excretion of radioactivity for subject nos. 1, 2, 3, and 5 to be 1.11, 0.92, 0.87, and 0.91 (mean f SD, 0.95 f O . l l ) , respec1244 / Journal of Pharmaceutical Sciences Vd. 83, No. 9, September 1994
Mean SD Range
Tag(min)
To.9 (min)
1.1 0.7
3.4 0.7 2.1-4.9
0.0-2.2
hmix
(min)
0.8 0.4 0.4-1.5
12
AUCobdprd
0.85 0.12 0.59-0.98
73.8 11.3 55.7-92.0
~~~
a
Tag= calculated time when 1 - CObJCpred = 1.00 (cf. Figure 2b). To,9=
time when observed concentration (cobs) is 90% of predicted concentration (Cprd) and is calculated from linear regression analysis of two consecutive time/ concentration data pairs. tinmix is estimated from a semilog plot of 1 - cob$c@ vs time, and 12 is the coefficient of determination of the regression line. AUCobd p r d = the ratio of observed to predicted AUC during 0-5 min.
tively, i.e. marginally higher ( p < 0.05)than the corresponding bioavailability estimates (Table 1). Terodiline and Metabolites in Feces-About 80% [3H]terodiline could be recovered from homogenates of blank feces. Addition of 3 M trifluoroacetic acid the day before extraction resulted in about 90% recovery of terodiline. Measurement of radioactivity in supernatant of nonhydrolyzed feces homogenate resulted in 30-45% recovery. Base extraction showed that about 13%of the radioactivity could be recovered in the ether phase from supernatant of hydrolyzed feces homogenate. A total of 47% was left in the
0
2
4
Time
5
D
m
!
5
u
604
6
a
6
5 8
(hours)
/
401 P 0 0
2
4
Time
(hours)
Figure 3-(A) Mean absorption rate profile in the 10 subjects calculated by deconvolution of the oral serum concentration-time profile. The midpoint of each staircase profile has been linearly connected. Vertical bars indicate the standard error of the mean. (B) Mean cumulative absorption profile in the 10 subjects calculated by deconvolution of the oral serum conceiitration-time profile. Vertical bars indicate the standard error of the mean.
water phase, 6% in the rinse water, and 40% in the solid residue. The possibility of extracting radioactivity from the feces homogenates was also studied under other experimental conditions: acid extraction or ion-pair extraction with 2 M perchloric acid resulted in negligible radioactivity recovery. After HPLC separation and radioactive detection one double peak of a polar compound and another double peak representing a somewhat less polar compound were seen. After silylation and GLC/MS analysis, the latter peak was identified as p-hydroxyterodiline and the former peak as 3,4-dihydroxyterodiline. Quantitative determination of 3,4-dihydroxyterodiline was not possible since this proved to be very unstable (black precipitate) in aqueous solution at the high pH used for extractions. The terodiline concentration was 1 5 nglg in the feces homogenate. p-Hydroxyterodiline had a peak concentration a t day 3 of about 15 nglg. None of the other known metabolites of terodilinell could be detected. The total amount of unchanged drug and metabolites per gram of the homogenate, estimated from radioactivity measurements, was about 350 ng. This means that the major part, about 335 nglg, could not be accounted for.
Discussion In the present investigations the initial radioactive mass balance study with [3H]terodilinegiven orally to the healthy volunteers showed that a relatively high proportion, about 26%, was recovered in feces. Incomplete oral absorption is less probable for a compound with lipophilic characteristics like terodiline-octanollphosphate buffer partition ratio of 29.7
at pH 7.3-and it would also be in contrast t o the complete A more absorption by the oral route when given to the feasible explanation to the recovery of terodiline-derived radioactivity in feces after oral dosing would be a significant excretion into the gastrointestinal tract via bile or through the mucosa of the gastrointestinal (GI) tract. The subsequently performed mass balance study with r3H]terodi1ine given parenterally demonstrated that the fraction of dose absorbed was virtually complete. Substantial amounts must evidently be excreted into the GI tract. Bulky, amphipathic molecules with molecular weights above 200-300 are generally good substrates for hepatobiliary transport. l8 Feces constitutes a complex matrix from which the terodilinederived radioactivity could only partly be extracted. Investigations of the composition of the terodiline-derived radioactivity revealed that intact terodiline represented < 1%of the radioactivity excreted in feces. p-Hydroxyterodiline, which is a major metabolite in urine,ll was also excreted in feces. The amount recovered was about 5%. Moreover, a previously unidentified metabolite, 3,4-dihydroxyterodiline,could also be found. This metabolite could not be quantitated due to methodological problems, i.e. instability of this kind of compound a t high pH (cf. refs 19 and 20). This metabolite, or its conjugate, may constitute a quantitatively important metabolite in feces. However, since it was not possible to extract more than 13%of the radioactivity in feces, the existence of other major metabolites cannot be excluded. Unextractable radioactivity may represent decomposed 3,4-dihydroq&rodiline or other unidentified metabolites. Following the termination of the intravenous infusion, the serum concentration declined in a biexponential manner with a marked a-phase which was terminated within about 30 min. The early concentration measurements of terodiline during infusion did not seem to be described by a conventional twocompartment body model. This model assumes that the intravascular mixing of the drug is instantaneous and can consequently in some cases lead to a lower estimate of clearance and volume of distribution compared with estimates obtained from actually measured concentrations (if available). A mean transit time for circulation through the vascular system can theoretically be estimated to about 1 min (blood volumelcardiac output). Experimental observations of a lag time between peripheral vein t o peripheral artery of about 0.5 min (Evans blue and indocyanine green) and a second peak due to recirculation after 1-2 min have been made.21 Moreover, the time for intravascular mixing, with ICG as the model substance, has been estimated t o be about 3.5 min.22 Regarding terodiline, observed and predicted serum concentrations agree after 3.4 min of infusion. Not only the fraction of the dose entering and passing through the gut wall into the hepatoportal system (Fa) but also the bioavailability (F$h) of terodiline assessed from serum or urine was high, about 0.90, which means that the first pass effect (1- F h ) is low. The different approaches used to estimate the bioavailability (Table 1) showed that deconvolution gave similar (ANOVA test, p = 0.673) estimates as the other methods tested. The advantage of the deconvolution algorithm used is that no assumptions besides linearity, timeinvariance, and zero initial state regarding the input functions have to be made. The input rate is calculated directly from the orally determined serum concentrations so that convolution of the unit impulse response and the input rate gives the exact response function. The disadvantage is that “negative absorption rates” and oscillations in the cumulative amount absorbed often occur. Observations like that were, as expected, observed with the presently used data. To deal with this issue the oral data may either be deconvolved only for the time period when the absorption rate is considered to be of significance (e.g. 8 h) or a mean value during the time of Journal of Pharmaceutical Sciences / 1245 Vol. 83, No. 9, September 1994
oscillations may be calculated. Problems like these have been addressedZ3 by putting constraints on the input function (nonnegative and piecewise monotonic) and applying leastsquare fitting, thereby obtaining a n alternative and more stabile deconvolution method. On the other hand, assumptions regarding the form of the input function thus have to be imposed. The possibility of determining bioavailability by using truncated areas or oralliv concentration ratio a t oral T,, has been discussed e l s e ~ h e r e . ~In * the present investigation, a further step of simplification has been tested. The very simple model using the independent 24-h oraYintravenous serum concentration ratio proved to be a reliable first estimate of the bioavailability; this method resulted in estimates that did not differ (ANOVA test, not significant) from calculations obtained from noncompartmental AUC, two-compartmental analysis, and urinary excretion. The single-time method assumes that negligible drug has been eliminated by the time absorption is completed and that the measurement has been made. This is likely to be true for drugs rapidly absorbed and having extremely long half-life, such as terodiline. It should be realized that the method is subject to some variability. Any variability in assay measurements could be overcome by undertaking replicate measurements of each serum sample. The problem of inherent biological variability in pharmacokinetics with time may perhaps be resolved by extending the method t o more than one time measurement. Then the statistical issues have t o be dealth with.
References and Notes 1. Sole, G. M.; Arkell, D. G. Scand. J . Urol. Nephrol. 1984, Suppl 87, 21-23. 2. Macfarlane, J. R.; Tolley, D. A. Scand. J . Urol. Nephrol. 1984, SUDD~ 87. 51-54. 3. Ekman, G.; Andersson, K.-E.; Rud, T.; Ulmsten, U. Acta Pharmacol. Toxicol. 1980, 46 Suppl, 39-43. 4. Klarskov, P.; Gerstenberg, T. C.; Hald, T. Scand. J . Urol. Nephrol. 1986, 20, 41-46.
1246 / Journal of Pharmaceutical Sciences Vo/. 83, No. 9, September 1994
5. Andersen, J. R.; Lose, G.; Nergaard, M.; Stimpel, H.; Andersen, J. T. Br. J . Urol. 1988, 61, 310-313. 6. Beisland, H. 0.;Fossberg, E. J . Am . Geriatr. SOC.1985,33,2932. 7. Hallen, B.; Guilbaud, 0.;Stromberg, S.; Lindeke, B. Biopharm. Drug Disoos. 1988.9.229-250. 8. Halren, B.; Bogentoft; S.; Magnusson, A.; Ekelund, P. Eur. J . Clin. Pharmacol. 1988,34, 291-297. 9. Hallen, B.; Bogentoft, S.; Sandquist, S.; Stromberg, S.; Setterberg, G.; Ryd-Kjellen, E. Eur. J . Clin. Pharmacol. 1989,36,487493. 10. Werner, T.; Olsson, L.-I. J . Labelled Compds. Radiopharm. 1987, 27, 29-39. 11. Noren, B.; Stromberg, S.; Ericsson, 0.;Lindeke, B. Acta Pharm. Suec. 1988,25, 281-292. 12. KarlBn, B.; Andersson, K.-E.; Ekman, G.; Stromberg, S.; Ulmsten, U. Eur. J . Clin. Pharmacol. 1982,23, 267-270. 13. Gibaldi, M.; Perrier, D. Pharmacokinetics. Marcel Dekker Inc.: New York and Basel, 1982. 14. Statistical Consultants, Inc. The American Statistician. 1986, 40,52. 15. Iga, K.; Ogawa, Y.; Yashiki, T.; Shimamoto, T. J . Pharmacokin. Biooharm. 1986.14.213-225. 16. Landaw, E. M.; Di'Stefano, 111, J. J. A m . J . Physiol. 1984,246, R665-R677. 17. Hallen, B.; Gralls, M.; Brotell, H.; Sandquist, S.; Stromberg, S. Pharmacol. Toxicol. 1990, 66, 373-381. 18. Gregus, Z.; Klaassen, C. D. J . Clin. Pharmacol. 1987,27,537541. 19. Heacock, R. A.; Powell, W. S. Progr. Med. Chem. 1973,9,275339. 20. Eriksson, B.-M.; Persson, B.-A. J . Cromatogr. 1987,386, 1-9. 21. Chiou, W. L. J . Pharmacokin. Bionharm. 1979. 7. 527-536. 22. Henthorn, T. K.; Avram, M. J.; %.ejcie, T. C . CZin; Pharmacol. Ther. 1989, 45,56-65. 23. Verotta, D. J . Pharmacokin. Biopharm. 1989, 17, 269-289. 24. Urso, R.; Aarons, L. Eur. J . Clin. Pharmacol. 1983, 25, 68993.
Acknowledgments We thank Tom Werner for synthesis of [2Hlterodiline and L3H]terodiline and Mona Gralls, Ursula Multan, and Niclas Brynne for technical assistance.
AND
B. NORENS
Received June 24, 1993, from the Kabi Pharmacia A6, Departments of *Pharmacokinetics and §Structural Analysis, R&D, 5-112 87 Stockholm. Sweden. and 'Deoartment of 6io~harmaceufics and Pharmacokinefics, University of Uppsala, $751 23 Uppsala, Sweden. ' Accepted for publication May' 23, 1994? unlabeled terodiline, was given orally in study A and intravenously in study B. The solution for study A contained 0.565 mg/mL (ratio [3Hlterodiline/[1Hlterodiline ca. 0.008%) or 0.182 MBq/mL. The solution with L3H1terodilinefor study B contained 2.48 mg/mL (ratio [3Hlterodiline/[1Hlterodilineca. 0.003%)or 0.666 MBq/mL. Deuteriumlabeled terodiline (N-tert-butyl-l-([2H~lmethyl)-3,3-diphenylpropylamine hydrochloride) was prepared using [2H31methyliodide as the deuterium-containing reagent.l0 The identity and isotope purity (>99.9%) were verified before use. The aqueous solutions of [2H]terodiline contained 2.5 mg/mL. Solutions with unlabeled terodiline for intravenous administration contained 2.49 mgfL. All solutions for intravenous and oral administrations were specifically prepared for this study by Kabi Pharmacia. Quality-control procedures of all terodiline formulations were carried out to assure that the formulations met the requirements for human use. Protocol-The subjects were normal healthy volunteers. The mean age and body weight were similar (Student's t-test, not significant) in studies A and B (50 2= 11y, 75 5 kg and 54 d~8 y, 76 f 9 kg). No pharmaceutical products were used 48 h before dosage and during the following study period. The subjects were fasting from 22.00 h on the evening before the study until 4 h after the dosage, when they were given lunch. Study A-The terodiline solution (20 mL) with a tracer dose of [3H]terodiline was given orally followed by 200 mL of water to six subjects (nos. 1-6; nos. 2 and 6 were females). The exact volume was calculated by weighing the bottles before and after intake and was found to correspond to a dose of 11.1 mg (3.59 MBq). The blood sampling schedule was a s follows: 0 (predose), 20, and 40 min, 1, 2, 3, 4, 6, and 8 h, and 1, 2, and 3 days. Serum was, after coagulation of the blood at room temperature, separated by centrifugation and was kept deep frozen (-20 "C) until the time of analysis of terodiline. Introduction and 12Urine was quantitatively collected at 0-2,2-4,4-8,8-12, 24 h and then in 24-h intervals up to 24 days. Prior to dosage, blank Terodiline is an anticholinergic and calcium antagonistic samples of urine and feces were collected. drug used to treat urge incontinence.1-6 The oral pharmacoStudy B-[2HlTerodiline, 12.5 mg in a 5-mL water solution, was kinetic profile has been investigated in healthy volunteers7 given orally to 10 subjects (nos. 1-3, 5, 7-12; nos. 2,6,8, and 9 were and in geriatric patient^.^,^ Due to its low clearance and long females). Terodiline was given intravenously with a tracer dose of 3H-labeled drug to four subjects (nos. 1-3,5) and solely as unlabeled half-life, conventional comparative studies (e.g. oral us intraterodiline to six subjects (nos. 7-12). The intravenous infusion venous) have to be awkwardly extended in time for adequate (Terufusion syringe pump STC-521; 1mumin for 5 min, giving a total determination of the profile (long wash-out period). The dose of 12.5 mg) was started at the same time as [2Hlterodiline was design of the present study has utilized the advantages of the ingested orally. The blood sampling schedule was as follows: 0 stable-isotope approach to minimize possible variability in(predose); 1, 2, 3, 4,and 5 min (during infusion); and 5, 10, 15, 30, troduced by treatment periods being widely separated. The and 45 min, 1, 2, 3 , 4 , 6 , and 8 h, and 1, 2 , 3 , 4 , 7 , 9, 11,and 14 days present study is composed of two parts: A, oral mass balance, (after infusion). Urine (subject nos. 1-3, 5, 7-12) and feces (subject and B, oral bioavailability and intravenous mass balance. The nos. 1-3, 5) were collected in 24-h portions quantitatively up to 20 days. objectives of the study are to (i)determine the bioavailability The study was performed according to the "Declaration of Helsinki" of oral terodiline using the stable-isotope technique, (ii) and later revisions. The general study plan and the written informainvestigate the kinetics of terodiline after intravenous injection given to the volunteers was approved by the Ethics Committee tion, (iii) quantify the amounts of terodiline-derived radioacat Uppsala University and the Swedish National Board of Health and tivity excreted via urine and feces after oral and intravenous the Isotope Committee at the Karolinska Hospital. administration, and (iv) identify major metabolites in feces. Terodiline Analyses in Serum and Urine-The serum and urine samples were analyzed for the content of [lHIterodiline and [2H]terodiline with a gas chromatographidmass spectrometric method. Experimental Section The method comprises a two-step partition with acid and base followed by acylation with trifluoroacetic anhydride (TFAA). The Materials-Tritium-labeled terodiline (N-terf-butyl-l-([3H31methyl)- method is essentially the same as previously described.12 However, 3,3-diphenylpropylamine hydrochloride) was prepared with 3H-labeled capillary column gas chromatography was used instead of packedmethyl iodide as the tritium-containing reagent.1° The labeling column GC. A Hewlett-Packard 5890 gas chromatograph equipped position is known to be metabolically stable. l1 The radiochemical with a Hewlett-Packard 5970 mass selective detector was used: SIM purity was > 99%. Tritium-labeled terodiline, supplemented with mode; ionization energy, 70 eV; transfer line temperature, 275 "C. The analyses were performed with 1-min splitless injection onto a H P Ultra capillary column (methyl silicon gum, 25 m x 0.32 mm, Mailing address: Kabi Pharmacia, S-751 82 Uppsala, Sweden. Abstract published in Advance ACS Abstracts, July 1, 1994. 0.11-pm phase thickness; injector temperature, 250 "C; column
Abstract 0 Terodiline was concomitantly administered intravenously (12.5 mg) and orally ([2H]terodiline, 12.5 mg) to 10 healthy volunteers. In four of the subjects, a tracer dose of the intravenously given terodiline was 3H-labeled. In a separate study, six subjects were given [3H]terodiline orally. Estimated pharmacokinetic parameters were as follows: systemic clearance, 93 mumin; renal clearance, 14 mumin; volume of distribution at steady-state, 407 L; terminal half-life, 54 h; and mean residence time, 77 h. After intravenous infusion, a rapid distribution phase (half-life, 4.5 min) could be observed. The maximum serum concentration after the oral dose was 29 vg/L and the time to maximum concentration was 5 h (estimated by noncompartmental analysis). Absorption commenced within the first hour and by deconvolution the maximum rate of absorption was determined to occur between 1 and 3 h, and by 3.4 h 90% of the available dose had been absorbed. Calculation of bioavailability by noncompartmental AUC, two-compartmental analysis, urinary excretion, and 24-h oralhtravenous concentration ratio gave similar results (ANOVA test, not significant). About 75% and 25% of administered radioactivity could be recovered in urine and feces, respectively. Intact terodiline in feces accounted for about 1% of the dose. pHydroxyterodiline was quantitated in feces and accounted for about 5% of the dose. Another metabolite, 3,4-dihydroxyterodiline, which has not previously been detected in urine or serum, was also identified.
*
@
0 1994, American Chemical Society and American Pharmaceutical Association
0022-3549/94/1200-1241$04.50/0
Journal of Pharmaceutical Sciences / 1241 Vol. 83, No. 9,September 1994
temperature program, 85 "C (1min), 40 "C/min to 300 "C, program stops after 10 min; flow rate of the helium carrier gas, 0.9 mumin). The detection was performed in the SIM acquisition mode a t the 321 ion for terodiline and 330 for internal standard, N-tert-butyl-3,3-bis([2Hs]phenyl)-1-methylpropylamine . The linearity expressed as the slope of standard curves shows a coefficient of variation of 4.9% with a mean regression coefficient of 0.9996 (CV% = 0.05%). The limit of quantitation, LO&, is 1pg/L (signal to noise ratio 5). The accuracy for serum analysis varies between 97 and 117%. The precision is 22% a t LOQ, about 7% at 3-50 pgL, and 3% at 50-200 pg/L. The system reproducibility is 3.3% (CV%). The accuracy for urine analysis is 106-119% and the precision is 3-5% in the interval 500-5000 p g L . Terodiline and Metabolites in Feces-Recovery of [3HlTerodiline after Addition to Homogenates of Blank Fece~-[~H]Terodilinewas added to 1-5 g of blank feces homogenate (study B, subject no. 5, day 0) and extracted according t o the bioanalytical method for terodiline. The effect of addition of 3 M trifluoroacetic acid the day before extraction was also studied. Radioactivity in Supernatant of Nonhydrolyzed Feces HomogenateWater (25 mL) was added to feces homogenate (study B, subject no. 5, day 3) and was sonicated for 1h. After 1day at room temperature, the homogenate was centrifuged and the clear supernatant was evaporated in uacuo at 50 "C. The radioactivity was measured by liquid scintillation counting. Extractable Radioactivity from Supernatant of Hydrolyzed Feces Homogenate-Homogenates (study B, subject no. 5 , day 3) were hydrolyzed with 3 M trifluoroacetic acid overnight a t 100 "C. After sonication for 1 h and centrifugation, aliquots of supernatant were evaporated in uucuo at 50 "C. The residue was washed with water and then combusted in Packard's sample oxidizer for measurement of radioactivity. To a n aliquot of the supernatant were added sodium hydroxide and sodium hydrogen carbonate until pH 12, and extractions were performed with diethyl ether (3 x 10 mL). Identification of Metabolites in the Supernatant of Hydrolyzed Feces Homogenate-The supernatant from feces homogenates (study B, subject no. 5, day 3), which had been hydrolyzed and sonicated for several hours, was evaporated in uacuo and dissolved in methanol and water (1.5 + 1.5 mL). The supernatant was run through a n HPLC system (methanol, water, trifluoroacetic acid, 1 1 0.1%) with a preparative C18 column UV 279 p with radioactive detection. Fractions were collected and GLCMS analyses were performed. Quantitative Determinations of Terodiline and p-hydroxyterodiline were made with homogenates (study B, subject no. 5, days 0-4) after hydrolysis with trifluoroacetic acid (3-6 M) at 100 "C by applying the bioanalytical method for terodiline. Terodiline and p-hydroxyterodiline were stabile in this medium for at least 12 h. Radioactivity Measurements-The 3H-activity was determined in the urine and feces samples. Urine (1mL) was added to 15 mL of Monophase (Packard Instrument) and the radioactivity was determined with a liquid scintillation spectrometer (Packard Tri-Carb 300) using the external standard for correction of quenching. Feces was homogenized in deionized water (1:5) for 2 min using a Stomacher Lab-Blender 3500 or 400 (Seward Medical). Samples of the suspension (2 mL) were allowed to dry in combustion cones and were oxidized together with some cellulose powder in a Packard Oxidizer 306, whereupon the radioactivity in the samples was determined. At regular intervals, tritium standards for sample oxidizers (study A, Amersham, VK; code TRR.lO1; study B, S~ec-Chec-~H, Packard) were combusted to check the combustion efficiency, which was found to be 90% (study A) and 93% (study B). PharmacokineticAnalyses-Noncompartmental Analysis (Studies A and B)-Maximum serum concentration of terodiline and the time of its appearance were those actually observed. log-linear regression analyses of the slopes of the serum concentration time curve from the 24-h sample and onward were used to calculate the half-life of elimination (tliz). The area under the serum concentration time curve from zero to the last measurable concentration was calculated by the linear trapezoidal method.13 The residual area from time t to infinity was calculated from (regression-line-estimated concentration at time of last measurement)/(ln 2/t1/2). For the purpose of estimation of the radioactivity in urine and feces after the period of quantitative collection, the extrapolated amount was estimated as (regression-line-estimated excretion rate on last day of collection)/(ln 2/t1/2). The determination of the half-life of the
+ +
1242 / Journal of Pharmaceufical Sciences Vo/.83, No. 9, Sepfember 1994
terminal monoexponential part of the rate plots was performed by log-linear regression analysis. The bioavailability in study B was calculated from the ratio of AUC,JAUCi, or from the amount excreted in urine (AepdAei,). These two methods are both based on the amounts eliminated. The bioavailability was also estimated from the 24-h oraVintravenous serum concentration ratio, C24,JC24iV. This single-time method is based on the amount in the body (Wekt)at a time (t) when absorption and distribution have ceased and the amount eliminated relative to the final amount eliminated is close enough to be regarded as equal after oral and intravenous dosage. A time point with sufficient precision that approximates these prerequisites can be found for a drug with absorption over a few hours and a long half-life and low clearance, like terodiline. The time point was chosen by visual inspection of the semi-log serum concentration time curves (see also the first paragraph in Noncompartmental Analyses). The oral/ intravenous concentration ratio method would for may drugs give a falsely high estimate of the bioavailability, since more drug, relative to the amount finally eliminated, is generally eliminated following intravenous administration compared to after oral administration at a time when the absorption phase is finished. However, if the absorption and distribution are very rapid relative to elimination, the error would be small. Compartmental Analysis (Study B)-Two approaches were tested to estimate the intravenous pharmacokinetic parameters (A, a ; B, p) used in the subsequent deconvolution analysis: (i) only postinfusion data with correction for the time of infusion and (ii) the complete infusion profile. Mono-, bi, and triexponential functions were fitted to the data by the program PCNONLIN, version 3.0,14 and using standard e q ~ a t i 0 n s . l The ~ Nelder-Mead algorithm was used to minimize the sum of squared residuals. Discrimination between the structural models and weighting schemes (1, lly, and l/y2, where y is the calculated concentration) was based on the goodness of fit, residual analysis, and the standard deviation of estimated parameters. Simultaneous fitting of intravenous and oral serum concentration time profiles were performed for estimation of the bioavailability with a model-dependent approach using a first-order absorption model. Parameters used in this fit were lag time for absorption, bioavailability, first-order absorption rate constant, disposition rate constants, fractional intercept, and initial volume of dilution. The following secondary parameters were ca1culated:l3 clearance, volume of distribution at steady state, mean residence time, and half-lives corresponding to the absorption phases a-phase and p-phase. The renal clearance was determined by using the estimated serum concentrations obtained from the simultaneous fit of the intravenous and oral models as constants in the following equation: urinary excretion rate = renal clearance x serum Concentration. Renal clearance was estimated from simultaneous fitting to the intravenous and oral urine data, which were weighted by the reciprocal of the excretion rates. Deconvolution (Studies A and B)-Decon~olution~~ was performed with data from studies A and B. The input rate function was determined from the convolution integral c(t) = flt)*Cs(t), where c(t) is the observed oral serum concentration profile, flt) is the input rate function, and C&) is the unit impulse function described by parameter estimates from PCNONLIN curve fits of the intravenous postinfusion profile. The bioavailability estimated from deconvolution was calculated as a mean value of the estimated cumulative amount absorbed from the time point when apparent plateau was reached. Statistical Analysis-The methods used for calculation of bioavailability were compared with one-way ANOVA. Model discriminations were made by F-test of the full and reduced m o d e P for each subject besides the standard procedures described in Compartmental Analysis. Other estimates were evaluated with Student's paired or unpaired t-test. Differences were regarded as statistically significant for p < 0.05. Descriptive statistical analyses have been made by calculations of means and standard deviations.
Results Study A-Pharmacokinetics and Radioactivity Measurements- The maximum serum concentration of terodiline was 38 f 9 ,ugh and the corresponding time point 6 f 3 h. The percentages of the given oral dose recovered as radioactivity (extrapolated t o infinity) in urine and feces were 69.9 f 6.3
Table 1-Oral Bioavailability (%) of Terodiline Assessed by Different Methods from Serum or Urine Data' % Bioavailability
Subject No.
AUC
Ae
Dcon
c24
2cpt
1 2 3 5 7 8 9 10 11 12
93 76 77 76 94 101 91 85 99 113
99 79 81 88 105 103 104 87 99 115
100 79 78 78 104 108 100 93 104 113
91 72 74 76 97 98 99 84 102 102
98 79 78 75 99 101 106 90 105 103
Mean rt SD Median Range
90k12 92 76-1 13
96 k 12 99 79-1 15
96k13 100 78-1 13
89+12 94 72-1 02
93+12 98 75-1 06
aANOVA test of AUC, Ae, C24, Dcon, and 2cpt. p = 0.673. AUC = noncompartmental determination of area under serum concentration time curve. Ae = noncompartmental determination of urinary excretion. c24 = serum concentration 24 h after administration. Dcon = deconvolution of serum concentrations (mean value). 2cpt = compartmental analysis of serum concentrations.
and 26.1 f 7.3, respectively. The half-life of radioactivity in urine and feces was 65.1 f 10.1 and 84.8 f 10.3 h, respectively, which means that the extrapolated amount beyond the last sampling interval is minimal. The total recovery from urine and feces was 96.2 f 3.3%. Study B-Pharmacokinetics: Noncompartmental Analysis- After intravenous infusion the observed maximum serum concentration was 211 f 161pg/L (range, 38-448 pg/L). After oral administration, the peak serum concentration was 29 f 5 p g L after 5 f 2 h. The extrapolated area under the serum concentration time curve beyond the last measurable concentration constituted 8 and 7% after intravenous and oral administration, respectively. The oral bioavailability estimated from AUC ratios was 90 f 12%(Table 1). The urinary excretion was 1.95 f 0.64 mg following the intravenous dose and 1.88 f 0.66 mg after the oral dose. The bioavailability estimated from urinary excretion ratios was 96 f 12%. Pharmacokinetics: Compartmental Analysis-The intravenous data were best described by fitting a biexponential function t o postinfusion data (F-test, p < 0.05). The fitted parameters were fractional intercept, initial volume of dilution, and disposition rate constants, and the weight used was lly. The disposition parameters obtained when simultaneously fitting the intravenous and oral models were unaltered (paired t-test, not significant or deviation < 0.3%)
compared to those estimated when fitting separately (weight, Uy). The parameters are given in Table 2. Examples of model-fitted serum and urine data are given in Figure 1. The point estimates are as follows: lag time for absorption (Tlag), 0.77 h; absorption half-life (tl/2&), 1.3 h; initial volume of dilution (Vc),91 L; volume of distribution a t steady-state (Vss), 407 L; half-life of a-phase (tl/zm),4.5 min; half-life of B-phase (tlizp), 54 h; mean residence time (MRT), 77 h; systemic clearance (CL), 93 mumin; and renal clearance (CL,), 14 m u min. The bioavailability was 93% (Table 1). Comparison of serum concentrations during infusion (using parameters obtained from postinfusion data) shows a discrepancy between observed and predicted concentrations (Figure 2a). Figure 2b shows that the time course to achieve the predicted concentration (the mixing time) can be approximated to an exponential process with an average half-life (t1lzrnk)of 0.8 min (Table 3). A lag time of 1.1min between the start of infusion and appearance in serum was found. The time course to reach the predicted concentrations was also analyzed by estimating the time (2'0.9) when observed and predicted serum concentrations converged, i.e. was within 90%. T0.9 was found to have a mean value of 3.4 min. The estimations of 2'0.9 and t l ~ are ~ kgenerally in agreement (2'0.9 = 3.4 min corresponds to a tl/Zmk= 1.0 min). The observed AUC during infusion (0-5 min) is 74% of the predicted AUC. However, AUC during infusion is only a minute part of AUCo-,. (about 0.3%). The fraction of the intravenous dose excreted in urine (CL,./ CL) was 16.6 f 5.2%, which when considering the excretion of radioactivity, indicates a substantial excretion of metabolites. It can be noted that the observed excretion rates occasionally deviated from the model calculated rates, e.g. 12 h excretion rate for subject no. 3 (Figure lc). A possible explanation may be spontaneous changes in urinary pH, which will influence the renal clearance of terodiline. However, the design of the study (concomitant administration of iv and oral drug) will compensate for these kinds of effects on, for example, bioavailability estimates. Pharmacokinetics: Deconvolution-The deconvolution of the oral serum concentration time profiles with the unit impulse response (intravenous bolus dose) resulted in a staircase input function. In all subjects the absorption rate decreased after 1-3 h. The maximum absorption rate was reached after 1.5 h (Figure 3a). The final average cumulative amount absorbed was 89 f 12%. Within 3.4 h, 90% of this amount had been absorbed (Figure 3b). Inspection of the complete profiles (until day 14) revealed unphysiological negative "absorption rates" in some cases when absorption could be expected to have ceased. However, no differences in the estimates were obtained (paired t-test, not significant) when using 24 h or
Table 2-PCNONLIN Estimates of Pharmacokinetic Parameters of Terodiline Subject No.
1 2 3 5 7 8 9 10 11 12 Mean SD CV% Mean RSD% of point estimate
7, (h)
tl/Zabs
(h)
V, (L)
V,, (L)
ha(min)
fil2p (h)
MRT (h)
CL (mUmin)
C L (mumin)
0.56 0.89 0.57 0.86 0.95 0.77 0.61 0.75 1.oo 0.74
1.45 1.47 2.03 1.30 0.90 1.28 1.19 1.71 0.95 0.76
97.1 13.1 15.2 12.5 190.0 32.3 267.0 169.0 89.3 28.2
397.0 429.0 315.0 294.0 308.0 354.0 613.0 513.0 370.0 476.0
5.6 1.6 1.7 1.7 16.0 3.7 3.8 5.5 2.9 2.6
55.3 50.3 54.7 56.4 36.8 65.5 38.8 45.5 52.1 84.0
79.4 71.4 78.1 80.5 52.9 93.5 55.9 65.3 74.9 120.2
83.3 100.3 67.3 60.8 97.1 63.1 182.7 130.9 82.4 66.0
16.0 13.9 12.6 10.7 8.4 13.8 11.8 26.3 16.5 12.5
0.77 0.15 20 17
1.30 0.39 31 31
91 90 98
407 102 25
11
6
4.5 4.3 94 20
53.9 13.6 23 9
77 19 25 9
93 38 41 8
14 4.9 3 5
Journal of Pharmaceutical Sciences / 1243 Vol. 83, No. 9, September 1994
0
2
1
3
6
5
4
Minutes
'oool
10
100
7: B
y = 3 6808
.
H"2 = 0 893
10A10 4 4 3 4 8 x 1
1:
10
1
-
'
n
0
'
50
l
.
-
I
100
Time
150
200
250
300
!
'
1
1
'
2
1
'
3
I
'
I
'
l
6
5
4
Minutes
(hours)
Figure 2-(A) Lag time of serum concentration during infusion (0-5 min) in subject no. 3. By linear regression between observed consecutive time points, the time to achieve 90% of the predicted concentration can be estimated to be 3.1 min (the mean value for nine subjects is 3.4 min). The serum profile predicted from postinfusion pharmacokinetic parameters is indicated (- - - -). (B) Semiiog plot of fraction remaining to predicted concentration (1 - CobJCpred)during infusion (0-5 min) in subject no. 3. From the regression line, a half-life of the mixing process can be estimated to be 0.7 rnin (mean value for nine subjects is 0.8 min, 12 = 0.95).
C
1000
. 0 1 i . 0
100
Table 3-Discrepancies between Observed and Predicted Serum Concentrations of Terodiline during Infusion (0-5 miny
10
0
100
200 Time
300
400
500
(hours)
Figure 1-Simultaneous tit of a two-compartment model to intravenous (A)and oral (0)serum concentrations in the time scales 0-8 hours (A) and 0-264 hours (11 days) (B) for subject no. 3. (C) Simultaneous fit of a two-compartment model to intravenous (A)and oral (0)urinary excretion rates in Subject no. 3.
the complete sampling period, i.e. up to 14 days for the deconvolution. Radioactivity Measurements-Radioactivity determinations in urine and feces showed that the percentages recovered (extrapolated to infinity) after intravenous administration were 74.2 f 6.8 and 19.4 f 4.2, respectively. The extrapolation from day 20 to infinity (percent of dose) was <1%for urine as well as for feces data. The total recovery from urine and feces was 93.5 f 8.1%. The half-life of radioactivity in urine and feces was 62.5 f 3.2 and 85.3 f 14.5 h, respectively. The fraction of administered dose absorbed into the hepatoportal system (Fa)was estimated from the oral/iv ratio of urinary excretion of radioactivity for subject nos. 1, 2, 3, and 5 to be 1.11, 0.92, 0.87, and 0.91 (mean f SD, 0.95 f O . l l ) , respec1244 / Journal of Pharmaceutical Sciences Vd. 83, No. 9, September 1994
Mean SD Range
Tag(min)
To.9 (min)
1.1 0.7
3.4 0.7 2.1-4.9
0.0-2.2
hmix
(min)
0.8 0.4 0.4-1.5
12
AUCobdprd
0.85 0.12 0.59-0.98
73.8 11.3 55.7-92.0
~~~
a
Tag= calculated time when 1 - CObJCpred = 1.00 (cf. Figure 2b). To,9=
time when observed concentration (cobs) is 90% of predicted concentration (Cprd) and is calculated from linear regression analysis of two consecutive time/ concentration data pairs. tinmix is estimated from a semilog plot of 1 - cob$c@ vs time, and 12 is the coefficient of determination of the regression line. AUCobd p r d = the ratio of observed to predicted AUC during 0-5 min.
tively, i.e. marginally higher ( p < 0.05)than the corresponding bioavailability estimates (Table 1). Terodiline and Metabolites in Feces-About 80% [3H]terodiline could be recovered from homogenates of blank feces. Addition of 3 M trifluoroacetic acid the day before extraction resulted in about 90% recovery of terodiline. Measurement of radioactivity in supernatant of nonhydrolyzed feces homogenate resulted in 30-45% recovery. Base extraction showed that about 13%of the radioactivity could be recovered in the ether phase from supernatant of hydrolyzed feces homogenate. A total of 47% was left in the
0
2
4
Time
5
D
m
!
5
u
604
6
a
6
5 8
(hours)
/
401 P 0 0
2
4
Time
(hours)
Figure 3-(A) Mean absorption rate profile in the 10 subjects calculated by deconvolution of the oral serum concentration-time profile. The midpoint of each staircase profile has been linearly connected. Vertical bars indicate the standard error of the mean. (B) Mean cumulative absorption profile in the 10 subjects calculated by deconvolution of the oral serum conceiitration-time profile. Vertical bars indicate the standard error of the mean.
water phase, 6% in the rinse water, and 40% in the solid residue. The possibility of extracting radioactivity from the feces homogenates was also studied under other experimental conditions: acid extraction or ion-pair extraction with 2 M perchloric acid resulted in negligible radioactivity recovery. After HPLC separation and radioactive detection one double peak of a polar compound and another double peak representing a somewhat less polar compound were seen. After silylation and GLC/MS analysis, the latter peak was identified as p-hydroxyterodiline and the former peak as 3,4-dihydroxyterodiline. Quantitative determination of 3,4-dihydroxyterodiline was not possible since this proved to be very unstable (black precipitate) in aqueous solution at the high pH used for extractions. The terodiline concentration was 1 5 nglg in the feces homogenate. p-Hydroxyterodiline had a peak concentration a t day 3 of about 15 nglg. None of the other known metabolites of terodilinell could be detected. The total amount of unchanged drug and metabolites per gram of the homogenate, estimated from radioactivity measurements, was about 350 ng. This means that the major part, about 335 nglg, could not be accounted for.
Discussion In the present investigations the initial radioactive mass balance study with [3H]terodilinegiven orally to the healthy volunteers showed that a relatively high proportion, about 26%, was recovered in feces. Incomplete oral absorption is less probable for a compound with lipophilic characteristics like terodiline-octanollphosphate buffer partition ratio of 29.7
at pH 7.3-and it would also be in contrast t o the complete A more absorption by the oral route when given to the feasible explanation to the recovery of terodiline-derived radioactivity in feces after oral dosing would be a significant excretion into the gastrointestinal tract via bile or through the mucosa of the gastrointestinal (GI) tract. The subsequently performed mass balance study with r3H]terodi1ine given parenterally demonstrated that the fraction of dose absorbed was virtually complete. Substantial amounts must evidently be excreted into the GI tract. Bulky, amphipathic molecules with molecular weights above 200-300 are generally good substrates for hepatobiliary transport. l8 Feces constitutes a complex matrix from which the terodilinederived radioactivity could only partly be extracted. Investigations of the composition of the terodiline-derived radioactivity revealed that intact terodiline represented < 1%of the radioactivity excreted in feces. p-Hydroxyterodiline, which is a major metabolite in urine,ll was also excreted in feces. The amount recovered was about 5%. Moreover, a previously unidentified metabolite, 3,4-dihydroxyterodiline,could also be found. This metabolite could not be quantitated due to methodological problems, i.e. instability of this kind of compound a t high pH (cf. refs 19 and 20). This metabolite, or its conjugate, may constitute a quantitatively important metabolite in feces. However, since it was not possible to extract more than 13%of the radioactivity in feces, the existence of other major metabolites cannot be excluded. Unextractable radioactivity may represent decomposed 3,4-dihydroq&rodiline or other unidentified metabolites. Following the termination of the intravenous infusion, the serum concentration declined in a biexponential manner with a marked a-phase which was terminated within about 30 min. The early concentration measurements of terodiline during infusion did not seem to be described by a conventional twocompartment body model. This model assumes that the intravascular mixing of the drug is instantaneous and can consequently in some cases lead to a lower estimate of clearance and volume of distribution compared with estimates obtained from actually measured concentrations (if available). A mean transit time for circulation through the vascular system can theoretically be estimated to about 1 min (blood volumelcardiac output). Experimental observations of a lag time between peripheral vein t o peripheral artery of about 0.5 min (Evans blue and indocyanine green) and a second peak due to recirculation after 1-2 min have been made.21 Moreover, the time for intravascular mixing, with ICG as the model substance, has been estimated t o be about 3.5 min.22 Regarding terodiline, observed and predicted serum concentrations agree after 3.4 min of infusion. Not only the fraction of the dose entering and passing through the gut wall into the hepatoportal system (Fa) but also the bioavailability (F$h) of terodiline assessed from serum or urine was high, about 0.90, which means that the first pass effect (1- F h ) is low. The different approaches used to estimate the bioavailability (Table 1) showed that deconvolution gave similar (ANOVA test, p = 0.673) estimates as the other methods tested. The advantage of the deconvolution algorithm used is that no assumptions besides linearity, timeinvariance, and zero initial state regarding the input functions have to be made. The input rate is calculated directly from the orally determined serum concentrations so that convolution of the unit impulse response and the input rate gives the exact response function. The disadvantage is that “negative absorption rates” and oscillations in the cumulative amount absorbed often occur. Observations like that were, as expected, observed with the presently used data. To deal with this issue the oral data may either be deconvolved only for the time period when the absorption rate is considered to be of significance (e.g. 8 h) or a mean value during the time of Journal of Pharmaceutical Sciences / 1245 Vol. 83, No. 9, September 1994
oscillations may be calculated. Problems like these have been addressedZ3 by putting constraints on the input function (nonnegative and piecewise monotonic) and applying leastsquare fitting, thereby obtaining a n alternative and more stabile deconvolution method. On the other hand, assumptions regarding the form of the input function thus have to be imposed. The possibility of determining bioavailability by using truncated areas or oralliv concentration ratio a t oral T,, has been discussed e l s e ~ h e r e . ~In * the present investigation, a further step of simplification has been tested. The very simple model using the independent 24-h oraYintravenous serum concentration ratio proved to be a reliable first estimate of the bioavailability; this method resulted in estimates that did not differ (ANOVA test, not significant) from calculations obtained from noncompartmental AUC, two-compartmental analysis, and urinary excretion. The single-time method assumes that negligible drug has been eliminated by the time absorption is completed and that the measurement has been made. This is likely to be true for drugs rapidly absorbed and having extremely long half-life, such as terodiline. It should be realized that the method is subject to some variability. Any variability in assay measurements could be overcome by undertaking replicate measurements of each serum sample. The problem of inherent biological variability in pharmacokinetics with time may perhaps be resolved by extending the method t o more than one time measurement. Then the statistical issues have t o be dealth with.
References and Notes 1. Sole, G. M.; Arkell, D. G. Scand. J . Urol. Nephrol. 1984, Suppl 87, 21-23. 2. Macfarlane, J. R.; Tolley, D. A. Scand. J . Urol. Nephrol. 1984, SUDD~ 87. 51-54. 3. Ekman, G.; Andersson, K.-E.; Rud, T.; Ulmsten, U. Acta Pharmacol. Toxicol. 1980, 46 Suppl, 39-43. 4. Klarskov, P.; Gerstenberg, T. C.; Hald, T. Scand. J . Urol. Nephrol. 1986, 20, 41-46.
1246 / Journal of Pharmaceutical Sciences Vo/. 83, No. 9, September 1994
5. Andersen, J. R.; Lose, G.; Nergaard, M.; Stimpel, H.; Andersen, J. T. Br. J . Urol. 1988, 61, 310-313. 6. Beisland, H. 0.;Fossberg, E. J . Am . Geriatr. SOC.1985,33,2932. 7. Hallen, B.; Guilbaud, 0.;Stromberg, S.; Lindeke, B. Biopharm. Drug Disoos. 1988.9.229-250. 8. Halren, B.; Bogentoft; S.; Magnusson, A.; Ekelund, P. Eur. J . Clin. Pharmacol. 1988,34, 291-297. 9. Hallen, B.; Bogentoft, S.; Sandquist, S.; Stromberg, S.; Setterberg, G.; Ryd-Kjellen, E. Eur. J . Clin. Pharmacol. 1989,36,487493. 10. Werner, T.; Olsson, L.-I. J . Labelled Compds. Radiopharm. 1987, 27, 29-39. 11. Noren, B.; Stromberg, S.; Ericsson, 0.;Lindeke, B. Acta Pharm. Suec. 1988,25, 281-292. 12. KarlBn, B.; Andersson, K.-E.; Ekman, G.; Stromberg, S.; Ulmsten, U. Eur. J . Clin. Pharmacol. 1982,23, 267-270. 13. Gibaldi, M.; Perrier, D. Pharmacokinetics. Marcel Dekker Inc.: New York and Basel, 1982. 14. Statistical Consultants, Inc. The American Statistician. 1986, 40,52. 15. Iga, K.; Ogawa, Y.; Yashiki, T.; Shimamoto, T. J . Pharmacokin. Biooharm. 1986.14.213-225. 16. Landaw, E. M.; Di'Stefano, 111, J. J. A m . J . Physiol. 1984,246, R665-R677. 17. Hallen, B.; Gralls, M.; Brotell, H.; Sandquist, S.; Stromberg, S. Pharmacol. Toxicol. 1990, 66, 373-381. 18. Gregus, Z.; Klaassen, C. D. J . Clin. Pharmacol. 1987,27,537541. 19. Heacock, R. A.; Powell, W. S. Progr. Med. Chem. 1973,9,275339. 20. Eriksson, B.-M.; Persson, B.-A. J . Cromatogr. 1987,386, 1-9. 21. Chiou, W. L. J . Pharmacokin. Bionharm. 1979. 7. 527-536. 22. Henthorn, T. K.; Avram, M. J.; %.ejcie, T. C . CZin; Pharmacol. Ther. 1989, 45,56-65. 23. Verotta, D. J . Pharmacokin. Biopharm. 1989, 17, 269-289. 24. Urso, R.; Aarons, L. Eur. J . Clin. Pharmacol. 1983, 25, 68993.
Acknowledgments We thank Tom Werner for synthesis of [2Hlterodiline and L3H]terodiline and Mona Gralls, Ursula Multan, and Niclas Brynne for technical assistance.