Journal of Pharmacokinetics and Pharmacodynamics, Vol. 33, No. 5, October 2006 (© 2006)
`DOI: 10.1007/s10928-006-9029-x
`
`Potential γ -Hydroxybutyric acid (GHB) Drug
`Interactions Through Blood–Brain Barrier Transport
`Inhibition: A Pharmacokinetic Simulation-Based
`Evaluation
`Indranil Bhattacharya1 and Kathleen M. K. Boje1,2
`
`Received February 07, 2006—Accepted July 18, 2006—Published Online August 29, 2006
`
`Recreational abuse or overdose of γ -hydroxybutyric acid (GHB) results in dose-dependent central
`nervous system (CNS) effects including death. As GHB undergoes monocarboxylic acid transporter
`(MCT)-mediated transport across the blood–brain barrier (BBB), one possible strategy for the
`management of GHB toxicity/overdose involves inhibition of GHB BBB transport. To test this strat-
`egy, interactions between GHB and MCT substrates (salicylic acid or probenecid) were simulated.
`Competitive, noncompetitive and uncompetitive inhibition mechanisms were incorporated into the
`GHB–MCT substrate interaction model for inhibitor dosing either pre-, concurrent or post-GHB
`administration. Simulations suggested that salicylic acid was the better candidate to limit GHB accu-
`mulation in the CNS. A time window of effect (>10% change) was observed for salicylic acid pre- and
`post-administration, with maximal transport inhibition occurring within 12 hr of pre- and 2 hr of post-
`administration. Consistent with the prediction that reduced GHB brain concentrations could translate
`to decreased pharmacodynamic effects, a pilot study in rats showed that the pronounced GHB seda-
`tive/hypnotic effects (24.0 ± 6.51 min; n = 4) in the control group (1.58 mmol/kg GHB plus saline)
`were significantly (p < 0.05) abrogated by salicylic acid (1.25 mmol/kg) coadministration.
`
`KEY WORDS: γ -hydroxybutyric acid; nonlinear pharmacokinetics; pharmacokinetic simu-
`lations; pharmacokinetics; drug interaction; salicylic acid.
`
`INTRODUCTION
`
`γ -Hydroxybutyric acid (sodium oxybate, GHB) is an endogenous
`compound (1) present in brain and peripheral tissues such as liver, heart,
`
`1Department of Pharmaceutical Sciences, School of Pharmacy and Pharmaceutical
`Sciences, University at Buffalo, H517 Cooke-Hochstetter, Buffalo, NY, USA.
`2To whom correspondence should be addressed. e-mail: boje@buffalo.edu
`657
`
`1567-567X/06/1000-0657/0 © 2006 Springer Science+Business Media, Inc.
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 1 of 25
`
`

`
`658
`
`Bhattacharya and Boje
`
`muscles and brown fat (2). GHB, while approved for the treatment of nar-
`colepsy, is widely abused as an anabolic agent, euphoriant and date rape
`drug. Recreational abuse or overdose of GHB (or precursors) results in
`dose-dependent central nervous system (CNS) effects (respiratory depres-
`sion, unconsciousness, coma, death) as well as tolerance and withdrawal
`(3,4). Currently the treatment of GHB overdose includes empirical inter-
`ventions and symptomatic treatments. Although naloxone and physostig-
`mine have been tried as antidotes, their use is controversial (5,6). In addi-
`tion, treatment of GHB toxicity is complicated by nonlinear pharmacoki-
`netics (7).
`There are multiple transport systems at the blood–brain barrier (BBB)
`which are responsible for influx or efflux of molecules from the CNS and
`form the basis of many possible drug–drug interactions (8–10). Using an
`in situ brain perfusion technique, we demonstrated that GHB undergoes
`carrier-mediated transport at the BBB, likely by an isoform of the mono-
`carboxylic acid transporter (MCT1) (11). Competition for carrier-mediated
`transport may lead to GHB–drug interactions. However, competition for
`carrier-mediated transport might be exploited to develop a strategy for
`treatment of GHB intoxication. Theoretically, administration of a trans-
`port inhibitor would diminish additional brain accumulation of GHB dur-
`ing overdose conditions and potentially shorten the duration of associated
`toxic effects.
`From our in situ experiments, MCT substrates (salicylic acid, valp-
`roic acid, and probenecid) significantly inhibited GHB brain influx (11),
`suggesting that MCT substrates may be potential transport inhibitor can-
`didates for GHB toxicity. Each of these drugs is therapeutically used in
`humans and therefore may potentially cause a GHB–drug interaction.
`Salicylic acid is a primary active metabolite of aspirin, a common over-
`the-counter analgesic. Probenecid is administered in conjunction with anti-
`biotics in the treatment of bacterial sexually transmitted disease such as
`gonorrhea (12) and syphilis (13,14), diseases that are commonly found in
`populations of drug abusers (15,16). Valproic acid is prescribed for epi-
`leptic seizures (absence, partial, myoclonic, and tonic-clonic), bipolar dis-
`order, and migraine prophylaxis. Physicians may also prescribe valproic
`acid for non-FDA approved indications for severe behavioral disturbances
`(e.g., agitation, aggression, explosive temper) which may occur second-
`ary to severe head injuries, Alzheimer’s dementia and behavioral dis-
`orders (attention-deficit hyperactivity, oppositional defiant and conduct)
`(17,18).
`Probenecid and salicylic acid appear to be reasonable candidates
`for the management of GHB toxicity via inhibition of GHB trans-
`port across the BBB. Valproic acid’s psychoactive profile diminishes its
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 2 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`659
`
`utility as a potential transport inhibitor. However, it was not clear which
`drug, salicylic acid or probenecid, would be pharmacokinetically opti-
`mal for the management of GHB toxicity. Hence, we wished to utilize
`pharmacokinetic modeling to better appreciate salicylate–GHB and pro-
`benecid–GHB drug interactions. Our objectives were to (A) model and
`simulate GHB plasma and brain concentrations in rats, (B) identify a
`dose of an inhibitor that will produce therapeutic concentrations of the
`inhibitor, (C) test whether a potential
`interaction is possible between
`GHB and each inhibitor using inhibitor concentrations within the inhib-
`itor’s therapeutic window and (D) understand the effect of pre-, con-
`current or post-administration of the inhibitor in relationship to GHB
`administration.
`
`MATERIALS AND METHODS
`
`Pharmacokinetic Models
`
`All computer modeling and simulations were performed using Win-
`Nonlin (WinNonlin Pro version 4.1, Pharsight Corp, Cary, NC). Litera-
`ture data were extracted by Graph Digitizer (Graph Digitizer, version 2.0,
`internally validated). Criteria for judging the quality of the model fit to the
`literature data were: visual inspections, square residual plots and weighted
`sum of residuals. Akaike and Schwartz criteria were used to discriminate
`between different models used to fit the GHB, salicylic acid or probenecid
`data.
`Definitions of the mathematical symbols are provided in Appendix.
`Simulations were performed assuming a standard rat weight (300 g) or the
`literature reported weight when fitting the model to literature data.
`
`Model 1: GHB Pharmacokinetics
`
`One of the objectives of our simulations was to provide insight into
`the brain concentrations of GHB in presence or absence of transport
`inhibitors. This necessitated development of a pharmacokinetic model that
`would simulate GHB brain concentration–time course data.
`We developed a one-compartment model with nonlinear elimination
`using published GHB plasma time course data and limited brain concen-
`tration data (7). Equation 1 was used to refit published plasma data using
`all data points, since the published pharmacokinetic analysis excluded
`early sampling times (t < 0.5 hr). Published parameter estimates (7) were
`used as initial estimates. (Symbols are defined in Appendix.)
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 3 of 25
`
`

`
`660
`
`Bhattacharya and Boje
`
`The next step was to develop a model that would simulate brain
`concentrations of GHB after intravenous dosing. The plasma profile
`generated by Eq. 1 was used as a forcing function to generate GHB brain
`concentrations, using an expanded model (inclusion of Eq. 2) that incor-
`porated GHB carrier-mediated uptake parameters determined by in situ
`brain perfusion (11). Limited published data on GHB brain concentra-
`tions (7) was used to develop this model, as described in detail in the next
`paragraph. A schematic representation of the complete model is provided
`in Fig. 1 (Model 1).
`
`Model 1
`
`Model 2
`
`Plasma
`
`CGP , VGP
`
`VmaxGP,
`KmGP
`
`BSA
`⇔
`FSA
`
`Model 3
`
`BPB
`⇔
`FPB
`
`VmaxPB,
`KmPB
`
`VmaxBr, KmBr,
`CLNS
`
`Brain
`
`CBr , VBr
`
`CLBr
`
`VmaxSA,
`KmSA
`
` PBTIS
`
`K12
`
`K21
`
`Fig. 1. Schematic diagrams of pharmacokinetic mathematical models. Model 1: GHB;
`Model 2: Salicylic acid; Model 3: Probenecid.
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 4 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`VP ∗ dCPGHB
`dt
`
`= − VmaxGP ∗ CPGHB
`(KmGP + CPGHB)
`
`661
`
`(11)
`
`VBr ∗ dCBrGHB
`dt
`
`= VmaxGBr ∗ CPGHB
`(KmGBr + CPGHB)
`
`+ CLNS ∗ CPGHB − CLBr ∗ CBrGHB
`(2)
`
`The initial conditions were CPGHB(0) = Dose
`
`VP
`
`, CBrGHB(0) = 0.
`
`Published whole brain concentrations of GHB at return of right-
`ing reflex following intravenous GHB (7) were used for model devel-
`opment. These concentrations, corrected for cerebrovascularly entrapped
`blood, were fitted using Eq. 2 to obtain estimates of CLBr and VBr. Simul-
`led to high CV% values as these
`taneous estimation of CLBr and VBr
`parameters are correlated. Therefore, CLBr was estimated and VBr was
`fixed at 1.00 ml, assuming GHB distribution into body water, a rat brain
`weight of 7.2 g/kg and a rat body weight of 0.3 kg (19,20).
`
`Model 2: Salicylic Acid Pharmacokinetics
`
`Another of our objectives was to identify a dose of an inhibitor that
`will produce inhibitor plasma concentrations within its therapeutic win-
`dow. A one compartmental model with nonlinear elimination of free drug
`(Eq. 3) (21) was used to fit the published salicylic acid plasma concen-
`trations (22) (Fig. 1—Model 2). Published data for salicylic acid plasma
`protein binding (22) was used to estimate protein binding parameters
`(BmaxSA, KDSA, KnsSA). These salicylic acid protein binding parameters
`were then fixed in Eq. 4 to fit total salicylic acid concentrations. Sali-
`cylic acid free concentrations were simultaneously fitted as the product of
`free fraction and total concentration of salicylic acid. Initial estimates of
`VmaxSA and KmSA for Eq. 3 were obtained from the published plasma pro-
`files (22). Since salicylic acid volume of distribution is dose-dependent, a
`different volume of distribution was estimated per dose (23,24).
`= −Kel ∗ SAF = −Kel ∗ FUP(SA) ∗ SAT
`= − VmaxSA ∗ FUP(SA) ∗ SAT
`VSA(KmSA + FUP(SA) ∗ SAT)
`
`dSAT
`dt
`
`(3)
`
`1When fitting our equations to published data, the units for volume were ml and were
`later converted to ml/kg for unit balancing. This conversion was followed throughout this
`work.
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 5 of 25
`
`

`
`662
`
`where
`
`FUP(SA) = −b + (cid:1)
`b2 − 4 ∗ a ∗ c
`2 ∗ a ∗ SAT
`
`and
`
`Bhattacharya and Boje
`
`(4)
`
`c = −SAT
`
`,
`
`b = 1 + N + BmaxSA − SAT
`a = 1 + N
`, N = KnsSA,
`KDSA
`The initial condition is SAT(0) = Dose
`
`KDSA
`
`.
`
`VSA
`
`Model 3: Probenecid Pharmacokinetics
`
`We remodeled published data on the dose dependency of probene-
`cid pharmacokinetics (25) using a two-compartment model with nonlinear
`elimination of free drug from the central compartment after intravenous
`administration (Eqs. 5–7).
`= −Kel ∗ PBF − K12 ∗ PBF + K21 ∗ PB
`= −Kel ∗ FUP(PB) ∗ PBT − K12 ∗ FUP(PB) ∗ PBT + K21 ∗ PBTIS
`VmaxPB ∗ FUP(PB) ∗ PBT
`= −
`− K12 ∗ FUP(PB) ∗ PBT
`VPB ∗ (KmPB + FUP(PB) ∗ PBT)
`+K21 ∗ PBTIS
`
`dPBT
`dt
`
`(5)
`
`The total probenecid concentration in the tissue is
`= K12 ∗ FUP(PB) ∗ PBT − K21 ∗ PBTIS
`
`dPBTIS
`dt
`
`Where
`
`a = 1 + N
`
`KDPB
`
`FUP(PB) = −b + (cid:1)
`b2 − 4 ∗ a ∗ c
`2 ∗ a ∗ PBT
`b = 1 + N + BmaxPB − PBT
`
`KDPB
`
`, N = KnsPB,
`
`, C = −PBT
`
`(6)
`
`(7)
`
`The initial conditions were PBT(0) = Dose
`VPB
`
`, PBTIS(0) = 0.
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 6 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`663
`
`Emanuelsson’s paper provided initial parameter estimates for most
`model parameters (25) but without model inclusion of protein binding.
`Therefore the protein binding curve (bound vs. free) was generated using
`the total and free concentrations provided by the authors (25). This regen-
`erated protein binding curve was used to obtain the probenecid protein
`binding parameter estimates (BmaxPB, KDPB, KnsPB).
`
`Model 4: GHB and Salicylic Acid Interaction
`
`=
`
`(cid:2)
`
`KISA
`
`Using an in situ brain perfusion technique, salicylic acid, a substrate
`for MCT (26), inhibited BBB influx of GHB (11). The extent of BBB
`transport inhibition following administration of GHB and salicylic acid
`was simulated for competitive, noncompetitive and uncompetitive inhi-
`bition mechanisms. Each mathematical expression was incorporated into
`Model 1 (Eqs. 8, 9 and 10, respectively), assuming that these interactions
`are driven by salicylic acid free plasma concentrations (Eq. 5). The sali-
`cylic acid inhibitory constant (KISA) of GHB BBB transport was assumed
`to be 3.6 mM, based on the reported salicylic acid KI for acetic acid trans-
`port which is a substrate of MCT (26). Simulations were performed for
`interactions between GHB (400, 600 or 800 mg/kg, i.v.) and salicylic acid
`(175 mg/kg, i.v. which produces a maximum concentration of ∼400 µg/ml).
`In addition, simulations were generated for salicylic acid administration
`occurring at up to 24 hr prior to, concurrently or 4 hr post-GHB adminis-
`tration.
`Kaufmann and Nelson demonstrated in vitro that salicylic acid is an
`inhibitor of GHB dehydrogenase (KI = 15.8 µg/ml) which is the rate limit-
`ing enzyme in GHB metabolism (27,28). Thus, additional simulations were
`performed to assess multiple interaction mechanisms (metabolic interac-
`tion only (Eqs. 11 and 12), metabolic and BBB transporter interaction
`(Eqs. 13 and 14)) between GHB and salicylic acid assuming concurrent
`administration.
`VBr ∗ dCBrGHB
`VmaxGBr ∗ CPGHB
`KmGBr ∗ (cid:2)
`(cid:3)(cid:3) + CPGHB
`1 + SAF
`dt
`+CLNS ∗ CPGHB − CLBr ∗ CBrGHB
`∗ CPGHB
`VmaxGBr
`1+ SAF
`KISA
`(KmGBr + CPGHB)
`+CLNS ∗ CPGHB − CLBr ∗ CBrGHB
`
`VBr ∗ dCBrGHB
`dt
`
`=
`
`(8)
`
`(9)
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 7 of 25
`
`

`
`664
`
`Bhattacharya and Boje
`
`VBr ∗ dCBrGHB
`dt
`
`VP ∗ dCPGHB
`dt
`VBr ∗ dCBrGHB
`dt
`VP ∗ dCPGHB
`dt
`VBr ∗ dCBrGHB
`dt
`
`=
`
`KISA
`
`=
`
`(cid:2)
`
`VmaxGBr
`1 + SAF
`KISA
`+ CPGHB
`KmGBr
`1 + SAF
`−CLBr ∗ CBrGHB
`VmaxGP ∗ CPGHB
`= −
`KmGP ∗ (cid:2)
`(cid:2)
`(cid:3)(cid:3) + CPGHB
`1 + SAF
`= VmaxGBr ∗ CPGHB
`+ CLNS ∗ CPGHB
`KmGBr + CPGHB
`−CLBr ∗ CBrGHB
`VmaxGP ∗ CPGHB
`= −
`KmGP ∗ (cid:2)
`(cid:2)
`(cid:3)(cid:3) + CPGHB
`1 + SAF
`VmaxGBr ∗ CPGHB
`+ CLNS ∗ CPGHB
`KmGBr ∗ (cid:2)
`(cid:3)(cid:3) + CPGHB
`1 + SAF
`−CLBr ∗ CBrGHB
`
`∗ CPGHB
`
`+ CLNS ∗ CPGHB
`
`KImSA
`
`KImSA
`
`KISA
`
`(10)
`
`(11)
`
`(12)
`
`(13)
`
`(14)
`
`AUCs served as measures of the exposure of the brain to GHB. The
`percent change in brain AUC was calculated as the difference between the
`respective AUCs in presence and absence of salicylic acid, normalized to
`the brain AUC in the absence of salicylic acid.
`
`Model 5: GHB and Probenecid Interaction
`
`The mechanism of probenecid inhibition of GHB BBB transport
`is unknown, necessitating separate simulations similar to the interac-
`tion between GHB and salicylic acid for competitive, noncompetitive and
`uncompetitive inhibition mechanisms. Mathematical expressions for each
`of the three interactions were incorporated in Eq. 2, assuming that the
`interactions were driven by the free plasma probenecid concentrations.
`Simulations were performed for interactions between GHB (400, 600 or
`800 mg/kg, i.v.) and probenecid (60 mg/kg (0.21 mmol/kg), i.v.). In addi-
`tion, simulations were generated for probenecid administration occurring
`up to 3 hr prior to, concurrently or 4 hr post-GHB administration.
`An extensive literature search did not provide an estimate of a pro-
`benecid KI value for BBB transport inhibition of MCT substrates. How-
`ever, based on rat brain in situ perfusion data (11), 20 mM salicylic acid
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 8 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`665
`
`inhibited [3H]GHB brain influx by ∼55%, while 10 mM probenecid inhib-
`ited [3H]GHB brain influx by ∼72%. Thus half the concentration of pro-
`benecid (compared to salicylic acid) inhibited [3H]GHB brain influx to a
`greater extent than salicylic acid. Based on these observations, it was arbi-
`trarily assumed that the inhibitory constant (KIPB) of probenecid for GHB
`transport was 0.9 mM, which is approximately four times lower than the
`KI of salicylic acid (3.6 mM).
`
`Experimental Protocol for GHB Sedative/Hypnotic Effect Studies
`
`GHB elicits a time- and dose-dependent sedative/hypnotic effect in
`rodents (7). Utilizing the insights provided by our pharmacokinetic model-
`ing and simulations, we designed a pharmacodynamic pilot study to exper-
`imentally examine the interaction between GHB and salicylic acid.
`Male Sprague Dawley rats (275–300 g; Harlan Sprague Dawley, Inc.
`Indianapolis, IN) were randomly assigned to two groups (n = 4 per
`group). The treatment group was dosed intravenously (i.v.) with 165 mg/kg
`(1.58 mol/kg) GHB (Sigma-Aldrich, St. Louis, MO), which was immedi-
`ately followed by 175 mg/kg (1.25 mol/kg) salicylic acid (Sigma-Aldrich, St.
`Louis, MO). The control group was dosed intravenously with 165 mg/kg
`(1.58 mol/kg) GHB, followed by equivolume saline. The loss in righting
`reflex (LRR) and the return of righting reflex (RRR) were measured in
`both the control and treatment groups. The sedative/hypnotic effect time
`was calculated as the absolute difference between the LRR and RRR.
`Data were statistically analyzed using a Mann–Whitney nonparamet-
`ric independent sample test (SPSS version 14.0, Chicago, IL).
`
`RESULTS
`
`GHB Pharmacokinetic Profiles
`
`Figure 2A shows the GHB pharmacokinetic model (Eq. 1) prediction
`of literature GHB plasma concentrations. The initial and final estimates of
`the various parameters are reported in Table I. The CLBr final estimate
`was obtained by using GHB plasma concentrations as a forcing function,
`setting the in situ GHB BBB uptake parameters as constants and fitting
`the corrected rat brain concentrations from Lettieri and Fung (Table I).
`Figure 2B illustrates the simulated GHB brain concentration profiles for
`400 and 800 mg/kg doses.
`This GHB pharmacokinetic model was validated by comparing the sim-
`ulated and observed GHB brain concentrations (29) following an intrave-
`nous dose of 603 mg/kg (732 mg/kg sodium salt of GHB, 5.8 mmol/kg). As
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 9 of 25
`
`

`
`666
`
`Bhattacharya and Boje
`
`3000
`
`A
`
`1000
`
`500
`
`300
`
`100
`
`50
`
`30
`
`0
`
`1
`
`2
`
`3
`Time (hrs)
`
`4
`
`5
`
`6
`
`1000
`
`B
`
`100
`
`10
`
`1
`
`0.1
`
`0
`
`1
`
`2
`
`4
`3
`Time (hrs)
`
`5
`
`6
`
`7
`
`GHB Plasma Concentration (µg/ml)
`
`GHB Brain Concentration (µg/ml)
`
`Fig. 2. Pharmacokinetic model predictions of literature GHB plasma and brain data (400
`and 800 mg/kg, i.v.). (A) GHB plasma concentrations. (B) GHB brain concentrations. Lit-
`erature GHB concentrations are represented by circles (400 mg/kg, 3.92 mmol/kg GHB) and
`inverted triangles (800 mg/kg, 7.84 mmol/kg). Dashed (400 mg/kg) and solid (800 mg/kg) lines
`represent computer estimated GHB concentrations.
`
`Table I. Initial and Final Estimates of Pharmacokinetic Parameters of GHB in Rats
`
`Parameter
`VmaxGP (µg/hr×102)
`KmGP (µg/ml)
`VP (ml/kg)
`CLBr (ml/hr-kg)
`
`Initial estimates
`
`Predicted final estimate (CV%)
`
`737
`338
`575
`139
`
`944 (11.5)
`622 (22.5)
`467 (4.64)
`179 (14.0)
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 10 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`667
`
`0
`
`1
`Time (hrs)
`
`2
`
`3
`
`1000
`
`100
`
`10
`
`1
`
`GHB Brain Concentration (µg/ml)
`
`Fig. 3. Observed and computer estimated GHB brain concentrations following 732 mg/kg
`GHB (sodium salt form; 604 mg/kg free acid; or 5.80 mol/kg GHB). Circles represent the
`observed brain concentrations and the solid line represents simulated brain concentrations.
`
`shown in Fig. 3, our model simulated concentrations reasonably captured
`the observed data of Giarman and Roth (29).
`
`Salicylic Acid Pharmacokinetic Profiles
`
`A saturable and nonsaturable binding model best described the sali-
`cylic binding data (22) and the resultant protein binding parameters (data
`not shown) were used to simultaneously fit the published salicylate total
`and free plasma concentration–time data using a dual binding site (satu-
`rable and nonsaturable) equation (21). Figure 4 shows the observed and
`model predicted salicylic acid plasma profiles for 25 mg/kg and 400 mg/kg
`intraperitoneal doses. The salicylic acid final pharmacokinetic parameter
`estimates are presented in Table II.
`
`Table II. Model Predicted Pharmacokinetic Parameters of Salicylic Acid Derived from Two
`i.v. Doses (25 mg/kg; 0.181 mmol/kg and 400 mg/kg; 2.89 mmol/kg)
`
`Parameter
`VmaxSA × 101 (µg/hr)
`KmSA (µg/ml)
`VSA25 (ml/kg)
`VSA400 (ml/kg)
`
`Initial estimate
`
`Predicted
`
`289
`60
`271
`443
`
`297
`74.1
`215
`693
`
`CV%
`
`21.4
`5.80
`7.97
`7.42
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 11 of 25
`
`

`
`668
`
`Bhattacharya and Boje
`
`10
`
`20
`
`30
`Time (hrs)
`
`40
`
`50
`
`60
`
`1000
`
`100
`
`10
`
`1
`
`Concentration (µg/ml)
`
`0.1
`
`0
`
`Fig. 4. Total and free salicylic acid plasma concentrations. Published total plasma salicylic
`acid concentrations are represented by closed symbols; free plasma salicylic acid concen-
`trations are represented by open symbols. Salicylic acid,
`i.v. doses: diamonds—25 mg/kg
`(0.18 mmol/kg); squares—400 mg/kg (2.89 mmol/kg). Computer model estimates are repre-
`sented by: dashed line—total plasma salicylic acid for 25 mg/kg; solid line—total plasma
`salicylic acid following 400 mg/kg; dash-dot-dot line—free plasma salicylic acid concentra-
`tions for 25 mg/kg; and dot-dash line—free plasma salicylic acid concentrations following
`400 mg/kg.
`
`The volume of distribution of salicylic acid is dose-dependent (23,24).
`As we wished to select a dose of salicylic acid that would produce thera-
`peutic concentrations we needed to estimate the volume of distribution at
`each salicylic acid dose. To obtain the respective volume of distribution,
`the model (Eq. 5) was fitted to literature data at different doses using the
`salicylic acid dual binding site equation (21). The model generated vol-
`umes of distribution for different doses of salicylic acid corresponded well
`with the observed literature values (data not shown), illustrating the dose
`dependency of volume of distribution.
`
`Probenecid Pharmacokinetic Profiles
`
`Probenecid protein binding was best described by a model with sat-
`urable and nonsaturable components. Figures 5A and 5B show the model
`predictions of probenecid total and free plasma concentrations which were
`described using a dual binding site equation (21), based on the assumption
`that only free drug undergoes elimination and distribution. The pharma-
`cokinetic parameter estimates are shown in Table III.
`Our model over predicts published free probenecid concentrations
`for all three doses, particularly at lower probenecid concentrations. In
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 12 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`669
`
`1000
`
`A
`
`100
`
`10
`
`1
`
`0
`
`1
`
`1000
`
`B
`
`100
`
`10
`
`1
`
`2
`
`3
`Time (hrs)
`
`4
`
`5
`
`6
`
`0.1
`
`0
`
`1
`
`2
`
`3
`Time (hrs)
`
`4
`
`5
`
`6
`
`Total Plasma Concentration (µg/ml)
`
`Free Plasma Concentration (µg/ml)
`
`Fig. 5. (A) Total plasma concentrations of probenecid after intravenous doses. Closed cir-
`cles, triangles and diamonds represents observed total plasma probenecid concentrations for
`50, 75 and 100 mg/kg (0.18, 0.26 and 0.35 mmol/kg) doses respectively while dashed, dash-
`dot-dot, and solid lines represents the corresponding computer estimated concentrations. (B)
`Free plasma concentrations of probenecid after intravenous doses. Open circles, triangles and
`diamonds represents the observed free plasma concentrations of probenecid for 50, 75 and
`100 mg/kg doses respectively while dashed, dash-dot-dot, and solid lines represents the corre-
`sponding computer estimated free concentrations.
`
`our modeling, we used binding parameters from the regenerated protein
`binding curve and simultaneously fit the free and total probenecid concen-
`trations using published data. However, the originally published proben-
`ecid protein binding curve (fraction unbound versus total concentration)
`showed significant variability in the lower concentration range; this var-
`iability likely contributes to our model over prediction. In addition, the
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 13 of 25
`
`

`
`670
`
`Bhattacharya and Boje
`
`Table III. Model Predicted Pharmacokinetic Parameters of Probenecid (50, 75 and 100 mg/kg)
`Administered Intravenously
`
`Parameter
`VmaxPB × 102µg/hr
`KmPBµg/ml
`−1
`K12 hr
`−1
`K21 hr
`VPB100 ml/kg
`VPB75 ml/kg
`VPB50 ml/kg
`
`Initial estimate
`
`Predicted
`
`114
`37.0
`7.80
`9.00
`199
`229
`224
`
`89.0
`26.7
`12.9
`3.98
`133
`137
`128
`
`CV%
`
`13.9
`39.5
`19.8
`13.8
`8.35
`7.70
`7.62
`
`published total probenecid concentration data was presented as an aver-
`age of six rats per dose while the corresponding free concentration data
`was based on one rat per dose. Thus the unavailability of an extensive data
`set for free probenecid concentrations along with the variability in the data
`likely contributes to our model over prediction at lower concentrations.
`
`GHB and Salicylic Acid Interaction
`
`In the treatment of rheumatic fever, the upper limit of therapeu-
`tic range for plasma total salicylate is 400 µg/ml. Hence, we wished to
`restrict our simulated salicylate concentrations within the known therapeu-
`tic range for salicylic acid. Simulations suggested that a dose of 175 mg/kg
`salicylic acid (volume of distribution = 445 ml/kg) would produce a maxi-
`mum concentration of 400 µg/ml in the rat.
`GHB (400, 600 and 800 mg/kg) and salicylic acid (175 mg/kg) inter-
`actions were simulated using different interaction models, assuming that
`only free salicylic acid inhibited GHB BBB transport. Table IV presents
`the percent change in GHB brain AUC exposure with concurrent adminis-
`tration of salicylic acid (175 mg/kg) for each type of transport interaction
`(competitive, noncompetitive or uncompetitive). Tables V and VI present
`the effects of salicylic acid on GHB brain AUC for single dose salicylic
`acid pre-administration (1, 3, 6, 12, 24 hr) and post-administration (1, 2
`and 4 hr) for each type of transport interaction. The maximum observed
`decrease in brain GHB exposure was found to be 20% when salicylic acid
`was concomitantly administered with the lowest dose of GHB (400 mg/kg),
`assuming a noncompetitive interaction. The decreases in GHB brain expo-
`sure ranged from ∼0–19% for all other combinations of GHB doses with
`salicylic acid (concurrent, pre- and post-administration) simulated for the
`three different interaction mechanisms. Assuming that a ≥10% change in
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 14 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`671
`
`Table IV. Effects of Concurrent Administration of GHB and Salicylic Acid:
`Percentage Decrease in GHB Brain AUC in the Presence of 175 mg/kg Salicylic
`Acid (1.27 mmol/kg i.v.) Assuming Competitive (C), Noncompetitive (N) and
`Uncompetitive (U) Interaction Mechanisms
`
`Type of interaction
`
`% Decrease in AUCBrain of GHB (salicylic
`acid administered concomitantly)
`
`Competitive
`Noncompetitive
`Uncompetitive
`
`Dose of GHB (mg/kg)
`
`400
`
`15.8
`19.9
`6.32
`
`600
`
`13.6
`19.0
`7.44
`
`800
`
`11.9
`18.1
`8.21
`
`Table V. Effects of Pre-Administration of Salicylic Acid: Percentage Decrease in GHB Brain
`AUC in Presence of 175 mg/kg Salicylic Acid (1.27 mmol/kg i.v.) Assuming Competitive (C),
`Noncompetitive (N) and Uncompetitive (U) Interaction Mechanisms
`
`Hours
`
`% Decrease in AUCBrain of GHB (salicylic acid pre-administered)
`
`Dose of GHB (mg/kg)
`
`400
`
`N
`
`19.3
`18.1
`16.2
`12.6
`6.55
`
`C
`
`15.4
`14.2
`12.6
`9.64
`4.89
`
`U
`
`6.39
`5.88
`5.15
`3.86
`1.91
`
`C
`
`13.1
`12.2
`10.8
`8.20
`4.10
`
`600
`
`N
`
`18.4
`17.2
`15.4
`11.9
`6.19
`
`U
`
`7.67
`7.07
`6.23
`4.66
`2.32
`
`C
`
`11.5
`10.7
`9.4
`7.12
`3.53
`
`800
`
`N
`
`17.5
`16.4
`14.6
`11.3
`5.85
`
`U
`
`8.38
`7.78
`6.87
`5.16
`2.58
`
`1
`3
`6
`12
`24
`
`GHB brain AUC exposure has a significant effect on GHB pharmacody-
`namic effects, the time window for salicylic acid with maximal GHB brain
`exposure inhibition occurs for up to 12 hr of salicylic acid pre-administra-
`tion and 2 hr of salicylic acid post-administration.
`
`GHB and Probenecid Interaction
`
`Probenecid plasma concentrations of 220–571 µg/ml were reported
`in humans (100 mg/kg orally in four divided doses 5 hr apart) (30). We
`wished to select an i.v. dose of probenecid which would produce maximum
`concentrations of 571 µg/ml. Simulations of probenecid plasma concentra-
`tions (Eq. 9) suggested that a 60 mg/kg i.v. probenecid dose would result in
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 15 of 25
`
`

`
`672
`
`Bhattacharya and Boje
`
`Table VI. Effects of Post-Administration of Salicylic Acid: Percentage Decrease in GHB
`Brain AUC in the Presence of 175 mg/kg Salicylic acid (1.27 mmol/kg i.v.) Assuming Compet-
`itive (C), Noncompetitive (N) and Uncompetitive (U) Interaction Mechanisms
`
`Hours
`
`% Decrease in AUCBrain of GHB (salicylic post-administered)
`
`Dose of GHB (mg/kg)
`
`400
`
`N
`
`10.6
`4.84
`0.67
`
`C
`
`9.23
`4.53
`0.67
`
`U
`
`2.37
`0.62
`0.00
`
`C
`
`9.26
`5.51
`1.17
`
`600
`
`N
`
`11.7
`6.32
`2.75
`
`U
`
`3.76
`1.38
`0.030
`
`C
`
`8.97
`6.11
`1.78
`
`800
`
`N
`
`12.3
`7.54
`1.88
`
`U
`
`4.92
`2.29
`0.19
`
`1
`2
`4
`
`total plasma probenecid concentrations of approximately 523 µg/ml. GHB
`(400, 600 and 800 mg/kg) and probenecid (60 mg/kg) interactions were sim-
`ulated using different interaction models, assuming that only free proben-
`ecid inhibits GHB BBB transport.
`The effect of concomitant, pre- and post-administration of 60 mg/kg
`probenecid on the GHB AUC is depicted in Tables VII, VIII and IX,
`respectively. Assuming that a greater than 10% change in GHB brain
`AUC will lead to a significant pharmacodynamic change, the maximum
`decrease in GHB brain exposure due to probenecid was found to be ∼12%
`when probenecid is concomitantly administered (assuming a noncompeti-
`tive interaction mechanism).
`
`GHB–Salicylic Acid Pilot Studies
`
`The pharmacokinetic simulations predicted a reduced brain exposure of
`GHB when administered concurrently with salicylic acid, which may trans-
`late to a decreased pharmacodynamic effect. To test this prediction, we per-
`formed a pilot pharmacokinetic simulation study to select doses, followed by
`a pharmacodynamic study that monitored GHB sedative/hypnotic effects.
`In humans, therapeutic doses of ∼65 mg/kg GHB orally produce
`plasma concentrations of ∼100 µg/ml; GHB blood concentrations from
`post-mortem drug overdoses may reach 330 µg/ml
`(31,32). Based on
`our simulations (Eq. 1), 165 mg/kg GHB i.v. produces concentrations of
`∼350 µg/ml in rats.
`Kaufmann and Nelson demonstrated in vitro that salicylic acid is an
`inhibitor of GHB dehydrogenase (KI = 15.8 µg/ml) which is the rate lim-
`iting enzyme in GHB metabolism (27,28). In order to address the pos-
`sibility of multiple interaction mechanisms (inhibition of transport and
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 16 of 25
`
`

`
`GHB–Drug Interactions Simulation
`
`673
`
`Table VII. Effects of Concurrent Administration of GHB and Probenecid: Percentage
`Decrease in GHB Brain AUC in the Presence of 60 mg/kg Probenecid (0.21 mmol/kg i.v.)
`Assuming Competitive (C), Noncompetitive (N) and Uncompetitive (U) Interaction Mecha-
`nisms
`
`Type of interaction
`
`% Decrease in AUCBrain of GHB (Probenecid
`concomitantly administered)
`
`Competitive
`Noncompetitive
`Uncompetitive
`
`Dose of GHB (mg/kg)
`
`400
`
`8.29
`11.4
`4.45
`
`600
`
`6.04
`9.67
`4.76
`
`800
`
`4.50
`8.23
`4.68
`
`Table VIII. Effects of Pre-Administration of Probenecid: Percentage Decrease in GHB Brain
`AUC in the Presence of 60 mg/kg Probenecid (0.21 mmol/kg i.v.) Assuming Competitive (C),
`Noncompetitive (N) and Uncompetitive (U) Interaction Mechanisms
`
`Pre-dose adminis-
`tration (hr)
`
`% Decrease in AUCBrain of GHB (probenecid
`pre-administered)
`
`Dose of GHB (mg/kg)
`
`400
`
`N
`
`5.93
`1.55
`
`C
`
`4.12
`0.67
`
`U
`
`2.16
`0.36
`
`C
`
`2.94
`0.44
`
`600
`
`N
`
`4.98
`0.85
`
`U
`
`2.35
`0.38
`
`C
`
`2.14
`0.33
`
`800
`
`N
`
`4.23
`0.72
`
`U
`
`2.32
`0.39
`
`1
`3
`
`Table IX. Effects of Post-Administration of Probenecid: Percentage Decrease in GHB Brain
`AUC in the Presence of 60 mg/kg Probenecid (0.21 mmol/kg i.v.) Assuming Competitive (C),
`Noncompetitive (N) and Uncompetitive (U) Interaction Mechanisms
`
`Post-dose adminis-
`tration (hr)
`
`% Decrease in AUCBrain of GHB (probenecid post-administered)
`
`Dose of GHB (mg/kg)
`
`400
`
`N
`
`5.98
`2.83
`0.36
`
`C
`
`4.95
`2.58
`0.36
`
`U
`
`1.44
`0.31
`0.00
`
`C
`
`4.35
`2.84
`0.59
`
`600
`
`N
`
`5.97
`3.44
`0.63
`
`U
`
`2.16
`0.81
`0.03
`
`C
`
`3.62
`2.77
`0.93
`
`800
`
`N
`
`5.67
`3.77
`0.99
`
`U
`
`2.60
`1.30
`0.11
`
`1
`2
`4
`
`PAR1019
`IPR of U.S. Patent No. 8,772,306
`Page 17 of 25
`
`

`
`674
`
`Bhattacharya and Boje
`
`A
`
`B
`
`C
`
`D
`
`1
`
`2
`Time (hrs)
`
`3
`
`4
`
`RRR
`
`60
`
`50
`
`40
`
`30
`
`20
`
`10
`
`Brain Concentration (µg/ml)
`
`0
`
`0
`
`Fig. 6. Simulated GHB brain concentrations
`following intravenous administration of
`165 mg/kg (1.58 mmol/kg) GHB in presence or absence of 175