(CANCER RFSEARCH 54, 1690-1694. April I, 1994)
`
`A Phase I and Phannacokinetic Study of Intravenous Phenylacetate
`in Patients with Cancer1
`
`Alain Thibault/ Michael R. Cooper, William D. Figg, David J. Venzon, A. Oliver Sartor, Anne C. Tompkins,
`Maribeth S. Weinberger, Donna J. Headlee, Natalie A. McCall, Dvorit Samid, and Charles E. Myers
`Clinical Pharmacology Branch {A. T., M. R. C., W. D. F., A. 0. S., A. C. T., M. S. W., D. J. H., N. A. M., D. S., C. E. MJ and Biostatistics and Data Management Section
`{D. J. V.J, National Cancer Institute, NIH, Bethesda, Maryland 20892
`
`ABSTRACf
`
`Phenylacetate bas recently been shown to suppress tumor growth and
`promote differentiation in experimental models. A phase I trial of phe(cid:173)
`nylacetate was conducted in 17 patients with advanced solid tumors. Each
`patient received a single Lv. bolus dose foUowed by a 14-day continuous
`i.v. infusion of the drug. Twenty-one cycles of therapy were administered
`at four dose levels, achieved by increasing the rate of the continuous
`Lv. infusion. Phenylacetate displayed nonlinear pharmacokinetics [K,. =
`105.1 :t 44.5 (SD) I'W'ml, V.....,. = 24.1 :t 5.2 mg{kg/h and V., = 19.2 :t 3.3 L].
`There was also evidence for induction of drug clearance. Ninety-nine % of
`pbenylacetate elimination was accounted for by convenion to phenylac(cid:173)
`etylglutamine, which was excreted in the urine. Continuous Lv. infusion
`rates resulting in serum phenylacetate concentrations exceeding K,. often
`resulted in rapid drug accumulation and dose-limiting toxicity, which
`c:onsisted of revenible central nervous system depression, preceded by
`emesis. 1bree of nine patients with metastatic, hormone-refractory pros(cid:173)
`tate cancer maintained stable prostatic specific antigen levels for more
`than 2 months; another had less bone pain. One of six patients with
`glioblastoma multiforme, whose steroid dosage has remained unchanged
`for the duration of therapy, has sustained functional improvement for
`more than 9 months. The use of adaptive control with feedback for the
`dosing of each patient enabled us to safely maintain stable phenylacetate
`concentrations up to the l"llDge of 200-300 I'Cfml, which resulted in clinical
`improvement in some patients with advanced disease.
`
`INTRODUCfiON
`
`Phenylacetate, a product of phenylalanine metabolism, is a small
`molecule (M, 136) normally present in the mammalian circulation in
`low concentrations (1). It has been administered primarily to children
`with hyperammonemia due to inborn errors of urea synthesis (2, 3)
`and to patients with hyperammonemia resulting from the chemother(cid:173)
`apy of leukemias (4) or from portal systemic encephalopathy (5). In
`humans, phenylacetate is conjugated with glutamine by the hepatic
`enzyme phenylacetyl Coenzyme A: glutamine acyltransferase to yield
`phenylacetylglutamine (6, 7), which is then excreted in the urine. The
`mobilization of glutamine-associated nitrogen is believed to be the
`mechanism whereby hyperammonemia is improved. More recently,
`phenylacetate and related compounds have received attention for their
`ability to induce tumor cytostasis and differentiation in laboratory
`models (8-11) and fetal hemoglobin synthesis in patients (12). Inter(cid:173)
`est in phenylacetate as an anticancer agent was also generated by
`reports that antineoplaston AS2-1, a preparation which by weight is
`80% phenylacetate, displayed clinical antitumor activity (13).
`Preclinical studies documented that phenylacetate modifies the
`biology of various hematopoietic and solid tumors, including prostatic
`carcinoma and glioblastoma (10, 11). To achieve this effect requires
`that cells be exposed to phenylacetate concentrations in excess of 275
`
`Received 9/15/93; accepted 1/28194.
`The cost of publication of the anicle were defrayed in pan by the payment of page
`charges. This anicle must therefore be hereby marked advertisement in accordance with
`18 U.S.C. Section 1734 solely to indicate this fact.
`1 This study was supported in pan by a grant from Elan Pharmaceutical Research Co.
`2 To wbom requests for reprints should be addressed. at Qinical Pharmacology
`Branch, National Cancer Institute, NIH, Building 10, Room 12C103, Bethesda, MD
`20892.
`
`,...g!ml for a minimum of two weeks. It appeared feasible to expose
`adults with solid tumors to similar concentrations of phenylacetate,
`which children with urea cycle disorders had tolerated (2, 3). Hence,
`a phase I trial was designed to deliver a Cive of phenylacetate over
`a 2-week period. We herein report the clinical and pharmacokinetic
`results of this study and discuss an alternative schedule of drug
`administration for future trials.
`
`MATERIALS AND METHODS
`
`Patient Population. Patients were eligible for this study if they had ad(cid:173)
`vanced solid tumors for which conventional therapy had been ineffective, a
`Karnofsky performance status greater than 60%, normal hepatic transaminases
`and total bilirubin, a serum creatinine less than 1.5 mg/dl, and normal leuko(cid:173)
`cyte (>3,500/mm3
`) and platelet counts(> 150,000/mm3
`). All patients signed an
`informed consent document that had been approved by the National Cancer
`Institute Clinical Research Subpanel. Seventeen patients, 16 men and 1
`woman, with a median age of 57 (range: 36-75) were enrolled between
`January and June 1993. Disease distribution included progressive, metastatic,
`hormone-refractory prostate cancer (9 patients), anaplastic astrocytoma or
`glioblastoma multifonne (6 patients), ganglioglioma (1 patient), and malignant
`pleural mesothelioma (1 patient).
`Drug Preparation and Administration. Sodium phenylacetate for injec(cid:173)
`tion was prepared from bulk sodium phenylacetate powder supplied by Elan
`Pharmaceutical Research Co. (Gainesville, GA). The finished injectable stock
`solution was manufactured by the Pharmaceutical Development Service, Phar(cid:173)
`macy Department, Clinical Center, NIH, in vials containing sodium phenyl(cid:173)
`acetate at a concentration of 500 mglml in sterile water for injection, USP, with
`sodium hydroxide and/or hydrochloric acid added to adjust the pH to approx(cid:173)
`imately 8.5. Doses of sodium phenylacetate to be infused over 30 min to 2 h
`were prepared in 150 ml of sterile water for injection, USP. Doses of pheny(cid:173)
`lacetate to be given over 24 h were prepared similarly to yield a total volume
`of 1000 ml and were administered using an infusion pump.
`Clinical ProtocoL The protocol as originally designed delivered an i.v.
`bolus dose of phenylacetate (150 mg!kg over 2 h) on the first day of therapy,
`to allow for the estimation of pharmacokinetic parameters. This was followed
`24 h later by a CIYI of the drug for the next 14 days. Cycles of 2-week drug
`infusions were repeated every 6 weeks. The rate of drug infusion was to be
`increased in sequential cohorts of at least three patients, and individual patients
`could escalate from one dose level to the next with sequential cycles of therapy
`provided they had experienced no drug-related toxicity and their disease was
`stable or improved.
`The protocol underwent several modifications over the 6-month period. (a)
`the size of the initial bolus dose was reduced from 150 to 60 mg!kg i.v. and the
`bolus infusion duration from 2 h to 30 min, after the first three patients were
`treated. This change resulted in drug concentrations optimal for estimating the
`pharmacokinetics of the drug (see below) within a 6-h time period. (b) after the
`nonlinear nature of the pharmacokinetics of phenylacetate was recognized (see
`below), the protocol was changed from a fixed dose escalation (dose levels 1
`and 2, 150 and 250 mg!kg/day, respectively) to a concentration-guided esca(cid:173)
`lation trial (dose levels 3 and 4, 200 and 400 JA.g/ml, respectively). In the latter
`format each patient was given an i.v. bolus dose of phenylacetate (60 mg!kg
`over 30 min) 1 week prior to beginning a 14-day CIYI of the drug. The
`patient-specific pharmacokinetic parameters estimated from the bolus dose
`were used to calculate an infusion rate that would maintain the serum phen-
`
`3 1be abbreviations used are: CIVI, continuous i. v. infusion; CNS, central nervous
`system; CSF, cerebrospinal fluid; HPLC, bigb performance liquid chromatography.
`1690
`
`Downloaded from cancerres.aacrjournals.org on November 9, 2015. © 1994 American Association for Cancer
`Research.
`
`Par Pharmaceutical, Inc. Ex. 1006
`Par v. Horizon, IPR of Patent No. 9,561,197
`Page 1 of 5
`
`

`

`PHARMACOKINETICS OF PHENYLACETATE
`
`and 440 nm following postcolumn color development with ninhydrin-RX
`(Beckman} at 131 •c. Beckman System Gold software was used for data
`acquisition and data management.
`Pbarmac:oklnedc: Methods. Initial estimates of V mu and Km for phenyla(cid:173)
`cetate were obtained by generating Lineweaver-Burk plots from concentration
`versus time curves following i.v. bolus doses. These initial parameter estimates
`were refmed by nonlinear least squares fitting to a single compartment, open
`nonlinear model, using the Neider-Mead iterative algorithm, as implemented in
`the Abbottbase Pharmacokinetic Systems software package (Abbott Labora(cid:173)
`tories, Abbott Park, IL; version 1.0). Each data point was weighted equally.
`Stadstical Methods. Student's t test was used to compare estimates of the
`pharmacokinetic parameters of phenylacetate derived from the Lineweaver(cid:173)
`Burk plots with those obtained using nonlinear least squares regression. Serum
`phenylacetate concentrations observed on day 2 of the CIVI were compared
`with those observed on day 11, using the Wilcoxon signed rank test for paired
`data (14) to assess the significance of time-related changes in drug concentra(cid:173)
`tions. At dose levels 3 and 4, the CIVI rate varied with time and the presence
`of induction of clearance was assessed by comparing a single compartment,
`open nonlinear model with the same model modified to incorporate two
`additional pharmacokinetic parameters which allowed for time-dependent
`changes in the maximum velocity of drug elimination (V max)· The performance
`of each model in describing a given set of dosing and concentration data was
`quantified by calculating the weighted sum of the errors squared following
`nonlinear least squares fitting. The standard deviation of the errors was
`modeled as a function of drug concentration multiplied by the coefficient of
`variation of the assay. The fitting procedure was used to maximize the
`likelihood of normally distributed variates, and the normality of the distribu(cid:173)
`tion of the standardized errors was confirmed by the method of Shapiro and
`Wilk (15). Confidence regions for the parameters were derived from the
`weighted sum of squares in the model incorporating the induction parameters,
`and approximate significance levels for testing between the two models were
`calculated using the F distribution (16); P < 0.05 was considered significant.
`The Spearman rank correlation method was used in an attempt to discern
`whether there was a relationship between the dose of phenylacetate adminis(cid:173)
`tered and the P value derived from the F distribution for each cycle of therapy.
`
`ylacetate concentration at the targeted level during the 14-day infusion. Drug
`concentrations were monitored according to the sampling schedule described
`below. Serum samples were analyzed weekly, prompting weekly reestimation
`of individual pharmacokinetics and dosage adjustment (adaptive control with
`feedback).
`SampUng Schedule. With the initial 150-mg/kg i.v. bolus, blood samples
`were obtained through a central venous catheter at the following time points
`calculated from the beginning of the infusion: 0, 60, 115, 125, 135, 150, 165,
`180, 240, 360, 480, and 600 min. For the 60-mg/kg bolus given over 30 min,
`blood sampling was performed at 0, 30, 60, 75, 90, 105, 120, 150, 180, 270,
`and 390 min from the beginning of the infusion. During CIVI, blood samples
`were obtained by venipuncture. At dose levels 1 and 2, blood samples were
`obtained daily during the ClVI; while at dose levels 3 and 4, blood samples
`were obtained on days 1, 2, 3, 8, 9, and 10 of the infusion. Twenty-four-h urine
`collections for the determination of phenylacetate and phenylacetylglutamine
`excretion were obtained on days 1, 7, and 14 of therapy. Sampling of the CSF
`was performed only if clinically indicated.
`Alla.lytical Method. Determination of sodium phenylacetate and phenyla(cid:173)
`cetylglutamine in serum and urine by HPLC. Blood was drawn into a Vacu(cid:173)
`tainer tube free of anticoagulant and was then refrigerated. It was centrifuged
`at 1200 X g for 10 min in a Sorvall RT 6000D centrifuge (DuPont Co.,
`Wilmington, DE) at 4•c. Serum was then removed and stored in Nunc
`Cryotubes (Nunc Co., Denmark) at -7o•c until the day of analysis.
`A standard curve was generated by adding known amounts of sodium
`phenylacetate (Elan Pharmaceutical Research Co.) and phenylacetylglutamine
`(a gift from Dr. S. W. Brusilow, Johns Hopkins University, Baltimore, MD) to
`a commercial preparation of pooled serum (Baxter Healthcare Corporation,
`Deerfield, IL). The standard values spanned the expected range of serum
`concentrations: 0, 5, 10, 20, 50, 100, 250, 500, 750, and 1500 p.g/ml.
`Two hundred ,._t of serum were pipeted into a 1.7-ml Eppcndorf tube (PGC
`Scientifics, Gaithersburg, MD). Protein extraction was carried out by adding
`100 ,..I of a 10% (v/v) solution of percbloric acid (Aldrich Chemical Co.,
`Milwaukee, Wl). The tube was vortexed and then centrifuged at 4500 X g for
`10 min. Supernatant (150 ILl) was transferred to a new 1.7-ml Eppcndorf tube
`and 25 ,.,.t of 20% KHC03 (w/v) were added to neutralize the solution. This
`was centrifuged at 4500 X g for 10 min and 125 ,.,.I of supernatant were
`transferred to an autosampler vial and maintained at to•c until HPLC injec(cid:173)
`tion. Urine samples were processed in an identical manner after an initial1:10
`dilution with water.
`The HPLC system (Gilson Medical Electronics, Middleton, Wl} was com(cid:173)
`posed of two pumps (305 and 306}, an 805 manometric module, an 811C
`dynamic mixer, a 117 variable wavelength UV detector, and a 231 autosampler
`fitted with a 20-,.,.I injection loop and cooled with a Grey Line model 1200
`cooling device. The column was a Waters (Millipore Corporation, Milford,
`MA) C 18 Nova-Pak, 3.9 x 300 mm, maintained at 60"C with a Waters
`temperature control module. The mobile phase solutions consisted of acetoni(cid:173)
`trile (J. T. Baker Chemical Co., Inc., Phillipsburg, NJ) and water, both
`acidified with phosphoric acid (0.005 M}. An acetonitrile concentration gradi(cid:173)
`ent was used, which increased from 5% to 30% over 20 min.
`Twenty ,._t of the neutralized supernatant were injected onto the column and
`eluted at 1 ml/min. The progress of the elution was followed by monitoring the
`UV absorbance at 208 nm. Characteristic elution times for sodium phenylac(cid:173)
`etate and phenylacetylglutamine under these conditions were 17.1 and 9.8 min,
`respectively.
`Detenninadon of Plasma Glntamlne Concentradons by Ion-Exc:bange
`Chromatography. Glutamine concentration was measured in sodium hepa(cid:173)
`rin-preserved plasma ( -80"C storage) following a 1:2 dilution/deproteinization
`with 15% 5-sulfosalicylic acid/sarcosine hydrochloride (Sigma Chemical Co.,
`St. Louis, MO). A stock solution of L-glutamine (1000 ,.,.gtml) (Sigma) was
`diluted with Li-S buffer (Beckman Instruments, Inc., Palo Alto, CA) to
`generate a standard curve ranging from 0.78 to 100 ,.,.gtml. A pooled plasma
`sample from 250 patients was used to make a quality control solution.
`Fifty-,.,.1 samples were autoinjected onto a 10-cm cation-ion exchange
`column integrated into a Beckman Model 6300 amino acid analyzer (Beck(cid:173)
`man). The solvent flow rate (2:1 water/ninhydrin) was maintained constant at
`0.5 ml/min. Column temperature was maintained at 33"C until glutamine was
`eluted, at which time the column temperature was raised by t.s•c;min to elute
`Fig. 1. Chromatogram of pbeoylacetate and pbenylacetylglutamioe. The peaks at 9.8
`and 17.1 min represent pbenylacetylglutamioe and phenylacetate, respectively. Serum
`sarcosine, the internal standard. The column was regenerated with lithium
`concentration of 250 ,.glml in both instances.
`hydroxide at 70"C following each injection. Absorbance was measured at 570
`1691
`
`RESULTS
`
`Analytical Assay. The reverse phase HPLC assay allowed both
`serum phenylacetate and phenylacetylglutamine concentrations to be
`determined simultaneously from the same sample (see Fig. 1). The
`lower limit of detection for both compounds in serum and urine was
`5 ~J.g/ml, based upon a signal:noise ratio of 5:1. The interassay
`coefficient of variation for serum concentrations was less than 6%
`within the range of 40-1000 ~J.g/ml (Table 1). The lower limit of
`detection for glutamine was 0.5 IJ.g/ml, with an interassay coefficient
`of variation that did not exceed 7%.
`
`e
`~ 0.4
`~
`
`0
`
`5.00
`
`15.00
`10.00
`Time(min)
`
`20.00
`
`25.00
`
`Downloaded from cancerres.aacrjournals.org on November 9, 2015. © 1994 American Association for Cancer
`Research.
`
`Par Pharmaceutical, Inc. Ex. 1006
`Par v. Horizon, IPR of Patent No. 9,561,197
`Page 2 of 5
`
`

`

`PHARMACOKINETICS OF PHENYLACETATE
`
`Model Specification and Initial Parameter Estimation. Fig. 2
`shows representative concentration versus time curves for simulta(cid:173)
`neously measured serum levels of sodium phenylacetate and phenyl(cid:173)
`acetylglutamine and plasma levels of glutamine following both 150-
`and 60-mg/kg bolus doses of sodium phenylacetate. The decline in
`serum phenylacetate concentration following the 150-mglkg bolus is
`linear when plotted on a nonlogarithmic scale and consistent with
`saturable elimination kinetics. While useful for demonstrating a zero(cid:173)
`order process, the magnitude of the bolus was inadequate for param(cid:173)
`eter estimation insofar as most of the phenylacetate concentrations
`obtained over the 6-h sampling period were above Km. In order to
`generate concentrations both above and below Km, the bolus was
`changed to 60 mglkg i.v. over 30 min. Visual inspection of the
`concentration versus time curves following these boluses revealed no
`evidence of an initial distributive phase, suggesting that a single
`compartment, open nonlinear model would be adequate to describe
`the pharmacokinetics of the drug. Initial estimates of Km [90 ± 30
`(SD) ~-tg/ml], V max (26.0 ± 10 mglkg!h), and Vd (22.4 ± 6.8 liters)
`were calculated in 13 patients using the Lineweaver-Burk equation.
`
`Table I PA 0 s/JJnJard curve assay variability
`cv
`(%)
`
`PAG
`{~tg/ml)
`
`PA
`(~tg/ml)
`
`cv
`(%)
`
`40
`40
`4.6
`2.6
`4.3
`400
`1.7
`400
`3.4
`3.1
`1000
`1000
`a PA, phenylacetate; PAG, phenylacetylglutamine; CV, coefficient of variation.
`
`~Or--------------------------------------,
`
`e400 -J c
`0 I c
`
`300
`
`200
`
`100
`
`8
`
`... --·--------·-
`
`,- -
`-~-=:;--'..: _____ ... ________ ... ________ ._____ _ ..
`
`----... _
`-------..
`
`r
`
`220
`
`~ 180
`a
`c:
`0
`c .. u c:
`~ 140
`
`100
`
`0
`(.)
`E
`" C5
`en
`
`60
`
`20
`0
`
`0
`
`Days
`
`14
`
`Fig. 3. Declining phenylacetate concentrations over time during CIVI (250 mglkg!day)
`in one patient~ suggestive of clearance induction. D, measured serum phenylacetate
`concentrations; bars, 95% confidence limits of the model's fit to the data.
`
`Refinement of these initial parameter estimates by nonlinear least
`squares fitting of the entire concentration versus time profile for each
`bolus dose yielded the following estimates: Km = 105.1 ± 44.5
`J.tg/ml; v max = 24.1 ± 5.2 mglkg!h; and vd = 19.2 ± 3.3 liters. The
`differences between the two methods of estimation were not statisti(cid:173)
`cally different, as measured by Student's t test (P = 0.89).
`Induction of Pbenylacetate Clearance. In some patients treated at
`dose levels 1 and 2, we observed a tendency for the serum phenyl(cid:173)
`acetate concentration to decrease with time. An example of this
`phenomenon is shown in Fig. 3. Considering the 12 cycles of therapy
`delivered at these levels, a comparison of the serum drug concentra(cid:173)
`tion measured on day 2 of CIVI to that observed on day 11 demon(cid:173)
`strated a 23% mean decline in concentration over this time period
`(Wilcoxon signed rank test, P = 0.016).
`At dose levels 3 and 4, attempts at maintaining targeted serum
`phenylacetate concentrations using adaptive control with feedback led
`to variable rates of drug infusion over time, which precluded a simple
`comparison of drug concentrations at the beginning and end of ther(cid:173)
`apy. We therefore analyzed all cycles of therapy at all four dose levels
`and compared the performance of the single compartment nonlinear
`model described above with the same model modified to allow V max
`to increase with time. The formula used to describe this increase was
`
`0~----L---~-----L----~----~----~--~
`400
`200
`700
`500
`600
`300
`100
`0
`
`Time (min)
`
`Vnuu
`
`(t) = Vmax
`
`(t = O)X {1.0 +[(IF- 1.0) X (1.0- etRx')]}
`
`8
`
`200r---------------------------------------~
`
`100
`
`0._~~--L-----~L-----~----~~------~
`500
`200
`400
`100
`300
`0
`
`Time (min)
`
`Fig. 2. Serum concentrations of phenylacetate (e) and phenylacetylglutamine (-:>and
`plasma concentrations of glutamine <•> following a 150-mglkg i.v. bolus of phenylacetate
`over 2 h (A), and a 60-mglkg i. v. bolus over 30 min (B).
`
`where tis the time elapsed (in h) since the initiation of therapy, IF is
`an induction factor representing the maximum-fold increase in v max
`at infinite time, and IR is a first order rate constant (h -•) describing the
`rate at which V max increases over time. Each cycle of therapy (n =
`21) was evaluated by comparing the difference in the weighted sum of
`errors squared generated by nonlinear least squares fitting with each
`model. The significance of the difference was evaluated using the F
`distribution. In 9 of the 21 cycles, allowing V max to increase with time
`yielded an improved fit (induction parameters, IF = 1.87 ± 0.37,
`IR = 0.0028 ± 0.003 h- 1). The Spearman rank correlation method
`did not demonstrate a correlation between rate of drug administration
`and the need to incorporate the two induction parameters into the
`model (rank correlation coefficient, -0.39; P = 0.084). The dose rates
`administered ranged from 450 to 1850 mg!h.
`Review of concomitantly administered medications revealed no
`association between specific drugs and the occurrence of a time(cid:173)
`dependent increase in phenylacetate clearance. In the seven patients
`with primary CNS tumors, treatment with anticonvulsants always
`antedated the administration of phenylacetate by months to years.
`1692
`
`Downloaded from cancerres.aacrjournals.org on November 9, 2015. © 1994 American Association for Cancer
`Research.
`
`Par Pharmaceutical, Inc. Ex. 1006
`Par v. Horizon, IPR of Patent No. 9,561,197
`Page 3 of 5
`
`

`

`Dose
`level
`
`PA" dose
`( mglkg/day)
`
`1
`49
`150
`104
`250
`2
`3
`171!
`266 ~ 40
`397
`374:!: 95
`4
`• PA, phenylacetate; PAG, phenylacetylglutamine.
`bMean:!: SD.
`
`PA
`(p.g/ml)
`19h
`40
`1!5
`244
`
`PAG
`(p.g/ml)
`
`91.1
`150
`181!
`3116
`
`34
`63
`55
`51
`
`PHARMACOKINETICS OF PHENYlACETATE
`
`aphasia for more than 9 months. Although no change in the size of the
`tumor mass was noted, reduction in peritumoral edema was docu(cid:173)
`mented by magnetic resonance imaging (see Fig. 4). His steroid
`regimen had not been changed for 2 weeks prior to starting therapy
`with phenylacetate and has been kept unchanged since.
`
`DISCUSSION
`
`Table 2 PA" and PAG concentrations per dose I.-vel during CIV/
`
`Previous descriptions of the pharmacokinetics of phenylacetate
`have been fragmentary. Simell eta/. (3) reported the drug to have first
`order elimination kinetics with a half-life of 4.2 h following bolus
`dose administration (270 mglkg) in children. The failure to recognize
`the nonlinear nature of phenylacetate pharmacokinetics probably re(cid:173)
`sulted from the smaller total doses given to these patients compared to
`those given in our study. The saturable pharmacokinetics of phenyl-
`
`Mechanisms of Phenylacetate Clearance. As shown in Fig. 2,
`phenylacetate underwent rapid conversion to phenylacetylglutamine.
`In the three patients who received 150 mglkg of phenylacetate over 2
`h, the peak serum concentration of phenylacetylglutamine was 224 ::!::
`81 l-£g/ml, 325 :t 72 min postinfusion. After the 60-mglkg boluses, the
`peak serum phenylacetylglutamine concentration was 104 ::!:: 33
`l-£g/ml at 86 :t 33 min.
`The plasma glutamine concentration prior to bolus treatment with
`phenylacetate was 109 :t 29 IJ.g/ml (n = 16), similar to values
`reported in the literature for normal volunteers (2, 3). The largest
`reduction in circulating plasma glutamine levels (46%) was observed
`in a patient receiving a 150-mg/kg bolus. Since phenylacetate is
`conjugated with glutamine to yield phenylacetate, the molar excretion
`of glutamine was found to increase in direct proportion to the dose of
`drug administered.
`The molar excretion of phenylacetylglutamine was determined
`from 24-h urine collections. It accounted for 99 :t 23% (n = 18) of
`the dose of phenylacetate administered over the same period of time.
`The recovery of free, nonmetabolized drug was only 1.5 ::!:: 2.4% of
`the total administered dose. A strong phenylacetate odor was detect(cid:173)
`able on patients' clothes and on examiners' hands after physical
`examination. This suggests that phenylacetate may also be excreted to
`some extent transdermally.
`Distribution of Pbenylacetate and Pbenylacetylglutamine into
`the CSF. Clinical circumstances required evaluation of the cerebro-
`spinal fluid in two patients who had metastatic prostate cancer and
`were free of CNS metastases. The first had reached steady-state
`phenylacetate and phenylacetylglutamine concentrations of 141 and
`199 l-£g/ml, respectively. The corresponding simultaneous CSF con(cid:173)
`centrations were 74 and 5 IJ.g/ml, respectively. At the time of simul(cid:173)
`taneous serum and CSF sampling, the second patient had not received
`further therapy for 6 h after having reached a serum phenylacetate
`concentration of 1044 IJ.g/ml. Measurements in serum and CSF were
`781 and 863 IJ.g/ml for phenylacetate and 374 and 46 IJ.g/ml for
`phenylacetylglutamine, respectively.
`Clinical Toxicities. No toxicity was associated with bolus admin(cid:173)
`istration of the drug. The highest peak serum concentrations were
`measured after the 150-mglkg bolus over 2 h (533 :!:: 94 l-£g/ml). Table
`2 lists the average serum phenylacetate concentrations per dose level.
`Although those achieved at dose levels 3 and 4 are close to their
`target, the large associated standard deviations reflect our inability to
`maintain serum phenylacetate concentrations within the desired range,
`even when using adaptive control with feedback.
`Drug-related toxicity was clearly related to the serum phenylacetate
`concentration. Three episodes of CNS toxicity, limited to confusion
`and lethargy and often preceded by emesis, occurred in patients
`treated at dose levels 3 and 4. They were associated with drug
`concentrations of 906, 1044, and 1285 IJ.g/ml (1078 :t 192 IJ.g/ml),
`respectively. Symptoms resolved within 18 h of terminating the drug
`infusion in all instances.
`Antitumor Activity. Prostatic specific antigen (measured weekly)
`remained stable for more than 2 months in 3 of the 9 patients with
`prostate cancer treated at dose levels 2, 3, and 4. A fourth patient
`taking 180 mg of morphine daily experienced marked improvement in
`bone pain control and was able to substitute a nonsteroidal antiin(cid:173)
`flammatory drug to his narcotic regimen. The mean phenylacetate
`concentration of the four responders was 244 ± 33 IJ.g/ml (186, 197,
`269, and 325 l-£g/ml). These values were not statistically different
`from those achieved in the other six patients with prostate cancer. One
`of six patients with glioblastoma multiforme that recurred after sur(cid:173)
`gery, standard radiation therapy and chemotherapy with bischloroeth-
`Fig. 4. (A) Pretreatment gadolinium-enhanced brain magnetic resonance imaging in a
`ylnitrosourea has maintained improvements in performance status
`patient with glioblastoma muhiforme. (B) Posttreatment gadolinium-enhanced magnetic
`resonance imaging after I cycle of phenylacetate ( 150 mg/kglday) illustrating resolution
`(30% on Kamofsky's scale), intellectual function, and expressive
`of peritumoral edema.
`1693
`
`Downloaded from cancerres.aacrjournals.org on November 9, 2015. © 1994 American Association for Cancer
`Research.
`
`Par Pharmaceutical, Inc. Ex. 1006
`Par v. Horizon, IPR of Patent No. 9,561,197
`Page 4 of 5
`
`

`

`PHARMACOKINETICS OF PHENYlACETATE
`
`1000
`
`900
`
`<:::- 800
`E
`~ 700
`c
`.2 600
`
`e 'E I 500
`
`(.) 400
`E
`2 300
`~
`
`200
`
`100
`
`00
`
`2
`
`4
`Days
`
`6
`
`8
`
`Fig. 5. Simulation of an every-12-h phenylacetate regimen (200 mg/kg/dosc, 1-h
`infusion) in a pbarmacokinetically avenge patient. For simplicity, induction of clearance
`was 001 factored in. Bars, 95% confidence limits expressing tbe anticipated range of
`concentntions in a population of patients.
`
`acetate is consistent with an enzymatic process and our calculations
`from the 24-h urinary excretion of phenylacetylglutamine confirm that
`this is the major route of elimination. Evidence that drug clearance
`may increase with time was derived from the comparison of drug
`levels on days 2 and 11 of the CIVI, adding another layer of com(cid:173)
`plexity to the pharmacokinetics of phenylacetate. To explain this
`phenomenon, we fust considered the potential role of concomitantly
`administered medications but failed to demonstrate any association.
`Our analysis of a possible relationship between an increase in drug
`clearance with time and the rate of drug administration did not reach
`statistical significance but suffered from the small number of cycles of
`therapy available for analysis. It should also be noted that, relative to
`the 14-day period over which it was assessed, V max tended to increase
`slowly, with an average half-time calculated from the induction rate of
`9.6 days. Longer periods of CIVI would allow this process to be more
`thoroughly characterized.
`Phenylacetate was delivered by CIVI in order to mimic the pre(cid:173)
`clinical conditions that had demonstrated antitumor activity, namely,
`continuous exposure to concentrations equal to or higher than 275
`IJ.g/ml for at least 2 weeks (8 -11 ). The results of Table 2 indicate that
`attempting to maintain serum phenylacetate concentrations at 400
`l-£g/ml using adaptive control with feedback was problematic, with
`drug concentrations that often greatly exceeded the level-specific
`targets. Lower concentrations (200-300 l-£g/ml) were safely main(cid:173)
`tained. Phenylacetate serum concentrations in excess of 900 1-£g/ml
`were typically associated with CNS toxicity. As expected for such a small
`and lipophilic molecule, phenylacetate readily penetrates into the CSF
`(this study and Ref. 1 ). While the ability to cross the blood-brain barrier
`may underlie the clinical improvement seen in the patient with glioblas(cid:173)
`toma, it could also explain the dose-limiting side-effects of the drug, i.e.,
`nausea, vomiting, sedation, and confusion.
`In the average patient, the drug must be infused at a rate equal to
`75% of v max• in order to maintain a constant serum phenylacetate
`concentration of 400 1-£g/ml, which is 4 times greater than Km. Thus,
`the slightest error in the estimation of individual pharmacokinetics or
`
`in the rate of drug infusion results in large changes in drug concen(cid:173)
`tration. An alternative strategy is to deliver the drug by repeated short
`infusions. Our limited experience with the 150-mg/kg i.v. boluses
`suggests that serum phenylacetate concentrations occurring tran(cid:173)
`siently above 500 1-£g/ml are well tolerated. In addition, intermittent
`drug infusion should permit some drug washout to occur, thereby
`minimizing drug accumulation. A regimen of 200 mg/kg every 12 h
`(1-h infusion) is simulated in Fig. 5. It assumes that the pharmacoki(cid:173)
`netic parameters determined from our 17 patients are representative of
`the cancer population at large and that V max does not change with
`time. It predicts that a wide range of peak drug concentrations will be
`observed. However, it is possible that these would be sufficiently
`transient so as not to produce CNS toxicity and the troughs not so
`prolonged as to abrogate the antitumor activity of the drug.
`Although dosing alter