0022·3565/91/2573·0972$03.00/0
`THE JOURNAL OF PHARMACOLOGY AND EXPERIMENTAL THERAPEUTICS
`Copyright !&)1991 by The American Society for Pharmacology and Experimental Therapeutics
`
`Vol. 257, No.3
`PrinW in U.S.It..
`
`Probenecid Enhances Central Nervous System Uptake of 2' ,3'-
`Dideoxyinosine by Inhibiting Cerebrospinal Fluid Efflux 1
`
`RAYMOND E. GALINSKY, KRISTEN K. FLAHARTY2 , BARBARA L HOESTEREY and BRADLEY D. ANDERSON
`Department of Pharmaceutics, College of Pharmacy, University of Utah, Salt Lake City, Utah
`Accepted for publication February 20, 1991
`
`ABSTRACT
`The effects of probenecid on the pharmacokinetics of 2' ,3'-
`dideoxyinosine (ddl) and on the distribution of ddl to cerebrospi-
`nal fluid (CSF) and brain tissue were determined in rats during
`and after a 2-hr i.v. infusion of ddl, 125 mgfkg/hr. Probenecid-
`treated rats received a loading dose of probenecid followed by
`an i.v. infusion of probenecid initiated 1 hr before and continued
`during and for 2 hr after termination of the ddl infusion. Plasma
`concentrations of probenecid averaged 221 ± 34 J.tg/ml upon
`termination of the ddl infusion and 258 ± 34 J.tg/ml (mean ±
`S.D., n = 4) 1 hr later. In the probenecid-treated animals, ddl
`concentrations were higher in plasma (1.5-fold), brain (1.5-fold)
`and CSF (5.4-fold) at the termination of the ddl infusion and
`postinfusion concentrations declined more slowly compared to
`controls. Postinfusion, the CSF fplasma and brain/plasma ratios
`steadily increased to a greater extent in the probenecid-treated
`rats compared to control animals. The time course of plasma,
`
`CSF and brain tissue concentrations were analyzed by nonlinear
`least-squares regression using two different compartmental
`models, one which neglected the direct exchange of drug be-
`tween the CSF and brain parenchyma, whereas the other allowed
`for such exchange to occur and neglected direct vascular trans-
`fer of drug to brain tissue. Allowing exchange between the CSF
`and brain tissue gave slightly improved fitting of the data from
`both probenecid-treated and control rats. The parameters reflect-
`ing direct transfer of ddl across the blood-brain barrier (kPS and
`keo) approached zero using the exchange model and could be
`set to zero without deterioration in the criteria for goodness of
`fit, suggesting that ddl may enter brain tissue indirectly via the
`CSF. Probenecid treatment decreased the rate constant for
`efflux of ddl from the CSF by 6-fold. The apparent increase in
`penetration of ddl into the CSF upon probenecid treatment could
`be accounted for solely by the decrease in the efflux rate
`constant.
`
`The most common nervous system disorder occurring in
`patients with AIDS, excluding brain tumors or opportunistic
`infections, is a progressive dementia characterized by severe
`cognitive, behavioral and motor impairment (Price et al., 1988).
`The primary neuronal targets appear to be monocytes, macro-
`phages and glial cells although other resident cells in the central
`nervous system, such as endothelial cells, may also be targets
`(Ho, 1989; Koenig et al., 1986; Wigdahl and Kunsch, 1989).
`Viral DNA can be detected in blood and brain tissue of AIDS
`patients at autopsy and a higher proportion of unintegrated
`viral DNA in tissue from patients with AIDS encephalitis has
`been reported recently (Pang et al., 1990).
`The symptoms and metabolic abnormalities produced by this
`unusual encephalopathy can be, in part, ameliorated by treat-
`
`Received for publication September 4, 1990.
`1 This work was supported in part by National Institutes of Health contract
`N01 AI 82680 from the National Institute of Allergy and Infectious Diseases.
`This work was presented in part at the American Association of Pharmaceutical
`Scientists Fifth Annual Meeting, Las Vegas, NV (Pharm. Res. 7: 5226, 1990).
`'Supported by a Fellowship from the American Society of Hospital Pharma·
`cists.
`
`ment with zidovudine, a reverse transcriptase inhibitor cur-
`rently used for the treatment of HIV infections (All worth and
`Kemp, 1989; Brunetti et al., 1989; Matthes et al., 1988; Yar-
`choan et al., 1987). Moreover, introduction of zidovudine treat-
`ment in the Netherlands (Portegies et al., 1989) in May of 1987
`resulted in a dramatic decrease in the incidence of AIDS
`dementia complex. This was also accompanied by a decrease in
`the presence of HIV -1 p24 antigen, a clinical marker that
`appears to correlate with the progression of the dementia.
`To date there are no reports relative to the effectiveness of
`other reverse transcriptase inhibitors in the treatment of AIDS
`dementia complex. However, there is evidence suggesting that
`other dideoxynucleosides, some currently in clinical trials, enter
`the central nervous system less readily than zidovudine. For
`example, compared to zidovudine, dideoxycytidine exhibits a
`10-fold lower CSF /plasma concentration ratio in monkeys (Col-
`lins et al., 1988). Similarly, we have shown recently that the
`rate constant for entry of ddl into CSF of rats is about 20-fold
`lower than that for zidovudine, paralleling the decreased lipo-
`philicity of ddl (Galinsky et al., 1990). Moreover, the rate
`constant for ddl efflux from the CSF is more than 50-fold
`
`ABBREVIATIONS: AIDS, acquired immunodeficiency syndrome; HIV, human immunodeficiency virus; CSF, cerebrospinal fluid; ddl, dideoxyinosine;
`HPLC, high-performance liquid chromatography.
`972
`
`AMN1070
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`

`

`1991
`higher than that for influx (Anderson et al., 1990), resulting in
`only a slight build-up of ddi concentration in the CSF at steady
`state. Thus, ddi is more likely to have limited utility in the
`treatment of AIDS dementia complex.
`The accumulation of dideoxynucleosides into the central
`nervous may be controlled in part by specialized transport
`carrier systems that regulate the flux of exogenous substrates
`such as purines, pyrimidines and their nucleosides across the
`blood brain barrier (Spector, 1990). Collins et al. (1988) found,
`in a series of pyrimidine dideoxyribonucleosides, that entry into
`the CSF was structure-specific but unrelated to lipophilicity,
`suggestive of a carrier-mediated process. Moreover, probenecid,
`an inhibitor of the organic acid anion transport system,
`enhances the distribution of zidovudine into the CSF due to,
`presumably, competitive inhibition of the CSF -to-plasma
`transport of zidovudine (Hedaya et al., 1990; Sawchuk and
`Hedaya, 1990). The fact that the rate constant for the efflux of
`ddi from the CSF exceeds that for influx by more than 50-fold
`is also indicative of an active or carrier-mediated process, as
`the efflux and influx rate constants should be similar if only
`passive transport processes are operating.
`The purpose of this study was to ascertain the probenecid-
`sensitivity of the uptake and release of ddi from both CSF and
`brain tissue of rats, to test the hypothesis that probenecid may
`inhibit the transport of this dideoxynucleoside from the CSF
`and brain. Probenecid was a logical choice in view of its effects
`on zidovudine distribution to CSF and brain tissue (Hedaya
`and Sawchuk, 1989; Sawchuk and Hedaya, 1990) and because
`of recent reports of the dose-sparing effects of probenecid on
`zidovudine in patients with AIDS (de Miranda et al., 1989).
`
`Methods
`Animal preparation and chemicals. Adult male Sprague-Dawley
`rats (Sasco Laboratories, Omaha, NE) weighing 315 ± 10 g were placed
`under light ether-methoxyflurane anesthesia and received indwelling
`cannula implants in the right jugular vein 1 day before study (Weeks
`and Davis, 1964). Animals were housed individually in plastic metab-
`olism cages. Food and water were withdrawn in the morning and
`withheld for the duration of an experiment.
`2',3'-ddl (NSC 612049) was obtained from the National Institute of
`Allergy and Infectious Diseases (NIH, Bethesda, MD) and used as
`received. Probenecid was purchased from Sigma Chemical Co. (St.
`Louis, MO). Rats received an i.v. infusion of ddl, 20 mg/ml at a rate of
`0.034 ml/min for 120 min which translated to a total dose of about 125
`mg/kg/hr. Seven control rats received the ddl infusion in 0.9% saline
`vehicle and 28 treated rats received the ddl as an infusion dissolved in
`a 10-mg/ml probenecid solution (250 mg of probenecid dissolved in 5-
`7 ml of 0.1 N NaOH, pH adjusted to 7.4 with 0.25 N HCI and final
`volume adjusted to 25 ml with 0.9% saline). One hour before the ddl-
`probenecid infusion, treated animals received a loading dose of proben-
`ecid designed to achieve an initial concentration of 200 ,.g/ml (Eman-
`uelsson and Paalzow, 1988, 1989) followed immediately by the proben-
`ecid infusion in an attempt to maintain steady-state plasma concentra-
`tions over 200 ,.g/ml throughout the study (Emanuelsson and Paalzow,
`1988). The probenecid infusion was continued for up to 2 hr after the
`end of the 2-hr ddl-probenecid infusion. The plasma, CSF and brain
`concentration data from control animals generated in this study were
`combined with data from other rats that received previously an iden-
`tical infusion of ddl (Anderson et al., 1990).
`Infusion and sampling protocol. Samples of CSF, plasma and
`brain tissue were obtained from single animals sacrificed at 90 and 120
`min during the ddl infusions and at 10, 20, 30, 45 and 60 min after
`termination of the ddl infusions. Probenecid (treated group) or saline
`(control) vehicle infusions were maintained before, during and after
`
`ddi-Probenecid Interaction
`
`973
`the ddl infusion. Animals were placed under light ether anesthesia for
`collection of CSF, plasma and brain samples, respectively. First, CSF
`was obtained by cisternal puncture (Waynforth, 1980) using a 22-gauge
`needle and silastic tubing. Next, within 1 to 2 min rats were exsanguin-
`ated by blood withdrawal from the abdominal aorta and then decapi-
`tated immediately. The brain was removed within 1 min of decapita-
`tion, rinsed with saline and quick frozen in liquid nitrogen and stored
`at -20"C. Heparinized blood was centrifuged and 0.3 ml of acetonitrile
`was added to 100 ,.I of plasma. Precipitated plasma samples and CSF
`samples were stored at -20"C until they could be assayed for ddl.
`Additional plasma samples were collected and stored at -20"C for
`probenecid analysis. All animal procedures conformed to the guidelines
`for the care and use of laboratory animals as set forth by the National
`Institutes of Health and the University of Utah.
`Sample preparation and HPLC analysis of DDI and proben-
`ecid. Frozen plasma samples containing acetonitrile were thawed and
`0.1 ml of water was added. These samples were centrifuged and the
`supernatant was collected. The precipitate was extracted with 0.3 ml
`of acetonitrile plus 0.1 ml of water. The supernatants were combined
`and evaporated to dryness under nitrogen, reconstituted in 0.5 to 1 ml
`of water and injected onto the chromatograph. Frozen brain tissue was
`thawed, weighed and homogenized in 1.5 ml of water. Acetonitrile (7
`ml) was added and the samples were vortexed, centrifuged and the
`precipitate was re-extracted with water-acetonitrile (0.5:2). The com-
`bined supernatants were evaporated to dryness under nitrogen, recon-
`stituted in 0.5 ml of water and applied to a solid phase C1s extraction
`column (Sep-Pak, Waters Associates, Milford, MA) which was rinsed
`with 5 ml of 2% methanol in 0.9% saline. ddl was eluted with 10 to
`15% methanol in water with the pH adjusted to 10.7 to 11.7 with
`triethylamine. The eluent was evaporated to dryness, reconstituted
`with a small volume of phosphate buffer (0.01 M, pH 7.4) and injected
`onto the chromatograph. CSF samples were thawed and diluted in
`phosphate buffer before HPLC analysis.
`ddl was quantified by UV spectroscopy at 254 nm using a modular
`HPLC system. The mobile phase for brain samples contained 5%
`methanol and 95% water with 0.1% formic acid and 0.02% w/v tetra-
`butylammonium bromide, pumped at a rate of 1.9 ml/min. The reten-
`tion volume of ddl under these conditions by using a reverse-phase
`column (Spheri-5, RP-18, 5 ,., 22 em X 4.6 mm inside diameter,
`Brownlee Laboratories, Santa Clara, CA) was 20.5 mi. ddl from plasma
`and CSF samples was separated using a reverse-phase column (ODS-
`224, Brownlee Laboratories) and a mobile phase of 8% methanol in
`water with 0.1% formic acid and 0.02% tetrabutylammonium bromide
`at a pH of 3.0 and a flow rate of 1.7 ml/min. Under these conditions,
`the retention volume of ddl was 14.5 mi.
`Probenecid in plasma samples was assayed by HPLC using a modi-
`fication of a previously published method (Harle and Cowen, 1978).
`Briefly, samples were quantified by UV spectroscopy at 254 nm after
`separation on a ,.Bondapak C1s column with a mobile phase of 22%
`acetonitrile and 78% 0.01 M KH2PO., pH 6.0, pumped at a rate of 1.5
`ml/min. Acetonitrile ( 100 ,.1) containing sulfamethazine as the internal
`standard was added to 50 ,.I of plasma to precipitate protein. The
`sample was vortexed, centrifuged and the supernatant injected onto
`theHPLC.
`Computer fitting of plasma, CSF and brain concentration
`data. The control ddl plasma concentration us. time data were pooled
`with data obtained in a previous study (Anderson et al., 1990) and
`analyzed by nonlinear regression (MINSQ, Micromath, Inc., Salt Lake
`City, UT) assuming a two-compartment model. A weighting factor of
`reciprocal of the square of the concentration was used in the regression
`analyses. The same model was applied to the ddl plasma concentration
`us. time data from the probenecid-treated rats. The pharmacokinetic
`parameters generated were subsequently treated as constants for the
`simultaneous fitting of the CSF and brain tissue concentration data.
`In two previous communications (Anderson et al., 1990; Galinsky et al.,
`1990), we characterized the pharmacokinetics of CSF and brain uptake
`of ddl and zidovudine in rats using a model which treats CSF and brain
`tissue as independent noninteracting compartments. The data in the
`
`AMN1070
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`

`

`Galinsky et al.
`
`974
`present investigation were fit initially to this model (Anderson et al.,
`1990). However, a more physiologically realistic model would be one
`which allows for drug to exchange between brain tissue and CSF. We
`therefore developed an alternative compartmental model that now
`accounts for transfer of drug between CSF and brain parenchyma (fig.
`1). This model is equivalent to that used previously when the rate
`constants for transfer between CSF and brain tissue (koc and kco) are
`set equal to zero. Nonlinear least-squares regression analysis of the
`CSF and brain concentration vs. time data were performed to compare
`the appropriateness of the two models.
`The equations for the brain tissue (equation 1) and CSF (equation
`2) concentrations of ddl vs. time were derived from the model shown
`in figure 1 using the method of Laplace transforms. The equation for
`plasma concentration of ddl vs. time was reported previously (Anderson
`et al., 1990).
`C
`_ "/V [<A ·ki'H- D)(k21- A )(1- e"r)e-"'
`A·(B-A)(a-A)(fj-A)
`BRAJN-r<o
`'
`
`+~~~~~~~~~~~
`
`(B·kJ·H- DHk2,- B)(l- eBT)e-8 '
`B·(A- B)(a- 8)({1- B)
`(a·k/'8- D)(k21 - a)(l- e"r)e-•'
`+~~~~~~~~--~-­
`a·(A- a)(B- a)(fj- a)
`
`(fj·kp8- D)(k21- {1)(1- e"r)e-d'] V C
`+
`+ s· P
`IHA- {1)(B- f1)(a- {1)
`[<A·kpc- E)(k2, - A)(1- e"r)e-"'
`_
`CcsF- k../V,
`A·(B -A)(a-A)({1-A)
`(B·kpc- E)(k2, - 8)(1- eBT)e-8'
`+~~B~-~(A~-~B~)(~a--~B~)~({1---B~)~
`
`(1)
`
`+~~~~~~~~--~--
`
`(a·kpc- E)(k21- a)(1- e"r)e-·•
`a·(A- a)(B- a)({1- a)
`+ (f1·kpe-E)(k2,-{1)(1-eJ'T)e-"']
`f1·(A- {1)(B- {1)(a- {1)
`
`(2)
`
`kco + kc8 + k8o + k8c
`- .J ( km + ~-8 + k8o + k8cl2 - 4( k8okco + k8okc8 + kcok8c)
`B=
`2
`
`During the infusion T = t, whereas postinfusion, T is a constant equal
`to the time at which the ddl infusion was terminated (120 min). The
`parameters a, {1, k2" and V, represent the plasma pharmacokinetics,
`assuming that ddl undergoes elimination from plasma according to a
`two-compartment model with elimination occurring from from the
`lcsFI-kco....,.
`y
`~-@ krtlka
`~ lsrainl--...
`
`kao
`Fig. 1. Compartmental model showing the relationship between plasma,
`CSF and brain tissue compartments for uptake and distribution of ddl in
`rats. The rate constants from plasma into CSF and brain are kPC and kpe,
`respectively. The rate constants for exchange of ddl between CSF and
`brain are represented by kcs and ksc.
`
`Vol. 257
`central compartment having a volume of V, (Anderson et al., 1990;
`Gibaldi and Perrier, 1975). ko is the infusion rate expressed in milli-
`grams per kilogram per minute. Vs represents the fraction of brain
`tissue volume comprised of vascular space plus extravascular spaces
`which are readily accessible to ddl without crossing the blood-brain
`barrier.
`
`Results
`The mean probenecid concentration in plasma from the
`probenecid-treated rats at the end of the ddl infusion was 221
`± 34 ~tg/ml. One hour after the end of the ddl infusion the
`probenecid concentrations averaged 258 ± 34 ~tg/ml. Although
`the concentration of probenecid in plasma appeared to rise
`slowly during and after the ddl infusion, for the purposes of
`the computer fitting and subsequent data analysis, we assumed
`that these concentrations of probenecid represented steady
`state. However, there is some uncertainty in this assumption.
`Even though the loading dose administered before the proben-
`ecid infusion was calculated to achieve plasma concentrations
`quite close to those desired at steady state, the actual time to
`reach steady state may have been prolonged due to the fact
`that probenecid displays dose-dependent pharmacokinetics
`(Emanuelsson and Paalzow, 1988).
`The time courses of ddl in plasma, CSF and brain paren-
`chyma in probenecid-treated rats and in the control animals
`are shown in figures 2, 3 and 4, respectively. Concentrations of
`ddl in plasma appeared to reach steady state by the end of the
`2-hr infusion and declined in a biphasic manner after the end
`of the infusion. Concentrations of ddl declined more slowly in
`
`' ~
`
`, ...
`•~-----....
`
`100
`
`10
`
`e
`l
`z
`0
`i= C( a::
`1-z ..... u z 8
`
`120
`TIME, minutes
`Fig. 2. Time course of mean ddl concentration in plasma of rats. Rats
`received 125 mgfkgjhr of ddl i.v. for 2 hr. Probenecid-treated animals
`(e, n = 3-4) received probenecid vehicle infusion, 30 mgfkgjhr, beginning
`1 hr before the ddl infusion and continuing for 1 hr after the end of the
`ddl infusion. Control animals (0, n = 5-6) received saline vehicle.
`
`240
`
`o
`
`o----o~
`
`--------·------~· ..... ·-~ ...
`0 ~0
`
`-._
`
`,--
`
`10
`~
`I
`~~~
`
`i··
`
`8
`
`60
`
`""
`180
`
`120
`TIME,mln~tes
`Fig. 3. Mean ddl CSF concentrations in control and probenecid-treated
`animals described in the experiment in figure 2.
`
`AMN1070
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`

`

`1991
`
`ddi-Probenecid Interaction
`
`975
`
`0
`
`0
`,,,----
`' I
`
`I
`I
`
`10
`
`e
`
`' .. ...
`~ 1.0
`~
`"" ...
`~ 0.1
`z 8
`
`60
`
`120
`TIME, minuln
`Fig. 4. Mean ddl brain concentrations in control and probenecid-treated
`rats described in the experiment in figure 2. Concentrations are uncor-
`rected for vascular space contribution.
`
`180
`
`plasma in the probenecid-treated rats than in the controls (fig.
`2). CSF concentrations during the ddl infusion were higher in
`probenecid-treated animals and, after termination of the ddl
`infusion, CSF concentrations declined more slowly in the pro-
`benecid-treated animals compared to the controls (fig. 3). The
`brain tissue concentrations of ddl, which reflect both vascular
`space and parenchymal contributions, declined in a biexponen-
`tial manner after termination of the ddl infusion in both treated
`and control animals and, as with plasma and CSF, the decline
`was slower in the probenecid-treated animals (fig. 4). The
`concentration of ddl in brain did not accumulate to the same
`extent as in the CSF during the ddl infusion in the probenecid-
`treated rats.
`The time course of the CSF to plasma concentration ratio is
`shown in figure 5a. Probenecid-treatment increased this con-
`centration ratio, relative to control, both during and after the
`ddl infusion. Brain parenchyma/plasma ratios were estimated
`by subtracting the model-calculated vascular space contribu-
`tions, shown in table 2, from the fitted tissue/plasma ratios.
`These ratios are shown in figure 5b. The brain/plasma concen-
`tration ratio was less sensitive to probenecid treatment than
`the CSF /plasma ratio.
`Nonlinear least-squares regression analyses of the plasma
`concentration us. time data were performed using a two-com-
`partment model as described previously (Anderson et al., 1990).
`From these analyses, the pharmacokinetic parameters listed in
`table 1 were generated. The solid and dashed lines drawn in
`figure 2 represent the parameters listed in table 1. The CSF
`and brain tissue concentration us. time data shown in figures 3
`and 4, respectively, were analyzed by nonlinear least-squares
`regression to compare the noninteractive and the exchange
`models and to determine which parameter(s) are most affected
`by probenecid treatment. The results of this analysis are shown
`in table 2. Allowing CSF -brain tissue exchange to occur pro-
`duced slightly computer fitting of the data (table 2), relative to
`the model which ignores possible exchange, as determined by
`reduced residual sums of squares and increased model selection
`criterion values (Akaike, 1976; Fox and Lamson, 1987). The
`only parameter that changed significantly by treatment with
`probenecid was the rate constant for efflux of ddl from the
`CSF, kco, which decreased between 4- and 6-fold depending
`upon the model used (table 2).
`
`Discussion
`The flux of endogenous and exogenous substrates into and
`out from brain tissue can occur by direct transfer across the
`
`.2 -Cl ...
`Cl s
`"' Cl
`c.
`......
`'-"' u
`
`Cl -Cl ...
`
`Cl
`
`s
`"'
`c
`c.
`......
`Cl
`E ::n
`... c c.
`c ... =
`
`0
`
`90
`
`150
`140
`130
`120
`Time, minutes
`
`165
`
`180
`
`30
`
`20
`
`b
`
`~
`1.1
`~ 10
`
`c
`
`0
`
`90
`
`150
`140
`130
`120
`Time, minutes
`Fig. 5. Time course of the CSF/plasma ratio (a) and model calculated
`brain/plasma ratio (b) in control and probenecid-treated rats described in
`the experiment in figure 2. Concentrations in brain were corrected for
`vascular space contribution. Solid bars represent probenecid-treated
`rats and hatched lines represent control animals.
`
`165
`
`180
`
`TABLE 1
`Effect of probenecid on pharmacokinetic parameters obtained by
`nonlinear least-squares regression analysis of ddt plasma
`concentration vs. time data
`
`Value
`
`Parameter"
`
`Control
`Probenecid-treated
`a (min-' x 1 02)
`23 ± 4.1
`7.8 ± 1.4°
`tJ (min-' x 1 03)
`5.1 ± 4.0
`30 ± 6.0
`k2, (min·' x 103)
`32 ± 8.0
`6.5 ± 4.8
`440 ± 100
`Vc (mlfkg)
`184 ± 30
`• Parameters represent the best fit of a two-compartment model with elimination
`occurring from a central compartment having a volume of Vc (Gibaldi and Perrier.
`1975) and were obtained in probenecid-treated and control rats during and afer a
`2-hr infusion of ddl at a rate of 125 rng/kg/hr.
`"Data are expressed as mean± S.D.
`
`AMN1070
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`

`

`Galinsky et al.
`
`976
`TABLE2
`Effect of probenecid on parameters for ddl uptake into the central
`nervous system determined according to two pharmacokinetlc
`models
`
`0.036
`0.032
`0.033
`0.031
`Vs (fraction)
`kPCb
`0.170
`0.239
`0.135
`0.148
`2.5*
`2.3*
`15.0
`9.2
`krxl
`kcsb
`0.63
`4.30
`0.0"
`0.0"
`kecb
`11.4
`0.0"
`0.0"
`2.90
`kP8b
`0.0"
`0.0"
`0.097
`0.019
`0.0"
`0.0"
`9.1
`1.02
`k8fl
`1.21
`0.135
`1.42
`0.137
`Residual sums of squares
`M.S.c.•
`2.55
`2.86
`2.39
`2.85
`• Referring to figure 1, the nonlnteractive compartmental model ignores the
`exchange between CSF and brain tissue. The exchange model includes the CSF-
`brain tissue Interchange and sets the rate constants into and out of brain tissue
`from plasma equal to zero.
`• Expressed as minutes·• x 10".
`• Fixed as a constant at zero thereby assuming that there is no CSF-brain tissue
`exchange.
`~ Fixed as a constant at zero. Allowing these parameters to vary resulted in
`their approaching zero with no Improvement in the overall fit of the data.
`'M.S.C., model selection criterion. For a given set of data, the most appropriate
`model is that with the largest M.S.C. (Fox and Lamson, 1987).
`• P < .05 compared to respective control values as determined by lack of overlap
`of 95% CL using the univariate approximation for the bounds of the 95% confidence
`regions (Fox and Lamson, 1987).
`
`blood-brain barrier (i.e., the endothelial cell wall of cerebral
`capillaries) or via transport across the blood-CSF interface at
`the choroid plexus with subsequent diffusion into and out of
`brain tissue (Ohno et al., 1978; Rapoport et al., 1979, 1982). In
`a previous study of the uptake of 2' ,3' -ddi into brain tissue
`and CSF of rats, we treated the CSF and brain tissue as
`separate, noninteracting compartments (Anderson et al., 1990),
`each with independent rate constants for drug entry and exit.
`As shown in table 2 (noninteractive model), the rate constants
`for direct entry of ddl into the brain (k8 p) were smaller than
`those for entry into the CSF (kpe) in both probenecid and
`control animals, suggesting that ddl concentrations build up
`and decline more slowly in brain tissue relative to CSF. Such
`behavior would also be expected, however, if ddl first entered
`the CSF, through the choroid plexus for example, and then
`diffused into brain parenchyma as suggested by Terasaki and
`Pardridge (1982).
`We have therefore developed a more physiologically realistic,
`alternative model which allows for exchange of drug between
`CSF and brain tissue. This model produced improved fitting of
`the data but also contained two additional parameters. How-
`ever, in the course of computer fitting the data with the ex-
`change model, we observed that the rate constant for direct
`entry into the brain (kp8 ) approached values not significantly
`different from zero and it was possible to set the parameters
`for entry and efflux across the blood brain barrier equal to zero
`without significantly altering the computer's ability to reason-
`ably fit the data. This resulted in a model having the same
`number of parameters as the noninteractive model but with
`improved parameter estimates in comparison to that model.
`We caution, however, that although this model gave the best
`fit of the data, it is possible to obtain reasonable computer
`fitting generating a variety of parameter values, if all seven
`parameters in table 2 are allowed to vary. Our exchange model
`differs from a previous model for solute exchange among com-
`partments of the central nervous system, developed by Rapo-
`
`Vol. 257
`port et al. (1982) in that 1) direct transfer of drug from plasma
`into the CSF is considered in the present model and 2) the
`kinetics of exchange between brain extracellular and intracel-
`lular space are ignored as no data are available pertaining to
`this latter exchange process. The model calculations suggest
`that ddl enters brain tissue indirectly via the CSF rather than
`by direct passage across the blood-brain barrier; however, it
`should be noted that this conclusion is model-dependent. More-
`over, the two models proposed differ only modestly in their
`abilities to fit the data, as judged by the residual sums of
`squares in table 2. Conclusive proof of one hypothesis will
`require more sophisticated experimental procedures.
`For some of the parameters (table 2) it is possible to compare
`the results of our regression analyses with values obtained by
`other investigators using different experimental techniques.
`Thus, the vascular space parameter ( Vs), which represents the
`percentage of brain tissue occupied by vascular space, is ex-
`pected to have a value of 1 to 4%, depending upon the region
`probed and the method utilized (Ohno et al., 1978; Preston et
`al., 1983). The values obtained in this study, namely 3.6 and
`3.2% in the control and probenecid-treated animals, respec-
`tively, agree with the literature values. Moreover, if the as-
`sumptions of the exchange model are valid, the transfer rate
`constants between CSF and brain tissue (kcs and ksd represent
`percolation of ddl into and out of brain parenchyma. In the
`absence of tissue binding, the ratio of kcs and ksc should be
`less than unity, reflecting the volume fraction of brain tissue
`in which ddl distributes. The observed ratio is 0.38 in control
`animals and 0.21 in the probenecid-treated rats. Both ratios
`are reasonable and within the range reported previously for
`polar compounds (Ohno et al., 1978).
`For those compounds that enter the brain by passive diffu-
`sion, physicochemical factors such as permeant lipophilicity
`(octanol-water partition coefficient), fraction unionized (which
`depends upon pKa) and molecular weight are generally consid-
`ered to be the principal determinants of the rate and extent of
`uptake (Ohno et al., 1978; Rapoport et al., 1979, 1982). There
`is compelling evidence to suggest that the central nervous
`system penetration of dideoxyribonucleosides, many of which
`are under consideration as anti-HIV agents, is determined in
`large part by carrier-mediated processes (Anderson et al., 1990;
`Collins et al., 1988; Sawchuk and Hedaya, 1990; Spector,
`1986b). Among the key observations in support of a possible
`role for carrier-mediated transport processes are: 1) the finding
`by Collins et al. (1988) that the central nervous system entry
`of pyrimidine dideoxyribonucleosides is structure-specific but
`not lipophilicity dependent; 2) probenecid, an agent that inhib-
`its the active transport of a variety of substrates (Fishman,
`1966; Spector, 1986a,b; Spector and Goetzl, 1986) also decreases
`the clearance of zidovudine from CSF in rabbits (Sawchuk and
`Hedaya, 1990); and 3) a transporter system has been identified
`recently in both cerebral microvessels and in choroid plexus for
`the purine and pyrimidine nucleosides (Kalaria and Harik,
`1986).
`In our previous studies of the uptake of ddl into brain and
`CSF of rats, the steady-state concentration of ddl in the CSF
`was less than 2% of the plasma concentration. Because the
`degree of penetration at steady state is determined by the
`balance of the influx and efflux rates, this low percentage
`indicates that the rate constant for net efflux exceeds that for
`net influx by more than 50-fold. If the overall CSF uptake of
`ddl were governed by passive diffusion alone, one would expect
`
`AMN1070
`Amneal Pharmaceuticals LLC v. Alkermes Pharma Ireland Limited
`IPR2018-00943
`
`

`

`1991
`these rate constants to be similar. An additional process con-
`tributing to the efflux rate constant, kco, however, is the re-
`moval of CSF by bulk flow, which is estimated to be 2 to 2.5
`#£1/min (Bass and Lundborg, 1973; Bums et al., 1976). The rate
`constant for efflux of inulin, a compound that is cleared from
`CSF exclusively by CSF bulk flow, is estimated from the data
`of Whittico and Giacomini (1988) to be 0.015 min- 1• This is
`only 10% of the value of kco found in our study. These obser-
`vations suggest that the value of kco is too large to be accounted
`for by either passive diffusion or by bulk flow of CSF. Thus,
`inhibiting the active transport of ddl from the CSF should be
`an effective means of increasing the overall central nervous
`system/plasma concentration ratio.
`There is now compelling evidence that several carrier-me-
`diated transport systems, with overlapping substrate specifici-
`ties, exist in the plasma-CSF barrier (Oldendorf, 1973; Spector,
`1990). Probenecid, a uricosuric agent regarded as a classical
`competitive inhibitor of organic anion transport in the renal
`tubule, has been shown to also block the transport of a variety
`of substrates from the central nervous system to plasma (Fish-
`man, 1966). For example, active efflux of such chemically
`diverse substances as penicillin (Fishman, 1966), pantothenic
`acid (Spector, 1986a) and leukotriene C4 (Spector and Goetzl,
`1986) occur via a probenecid-sensitive mechanism. Recently,
`probenecid has been shown to decrease the metabolic and renal
`clearance of zidovudine (de Miranda et al., 1989; Hedaya et al.,
`1990; Kornhauser et al., 1989) and to increase the CSF /plasma
`concentration ratio in rabbits (Sawchuk and Hedaya, 1990).
`Thus, the concomitant use of probenecid with dideoxynucleo-
`sides may be an important clinical strategy (de Miranda et al.,
`1989; Kornhauser et