`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`British Journal of Clinical Pharmacology
`
`DOI:10.1111/j.1365-2125.2005.02409.x
`
`Pharmacokinetics and pharmacodynamics of single-dose
`triazolam: electroencephalography compared with the
`Digit-Symbol Substitution Test
`
`David J. Greenblatt, Lu Gan, Jerold S. Harmatz & Richard I. Shader
`Department of Pharmacology and Experimental Therapeutics, Tufts University School of Medicine and Tufts-New England Medical Center,
`Boston, MA, USA
`
`Correspondence
`David J. Greenblatt, MD,
` Department
`of Pharmacology and Experimental
`Therapeutics, Tufts University School
`of Medicine, 136 Harrison Avenue,
`Boston, MA 02111, USA.
`+
`Tel:
`
`1 617 636 6997
`+
`Fax:
`
`1 617 636 6738
`E-mail:
` dj.greenblatt@tufts.edu
`
`Keywords
`Triazolam, benzodiazepines,
`pharmacokinetics,
`pharmacodynamics,
`electroencephalography
`
`Received
`2 September 2004
`Accepted
`28 December 2004
`
`Aims
`To investigate whether the electroencephalogram (EEG) directly reflects the CNS
`effects of benzodiazepines by evaluating the relation of the EEG to plasma drug
`concentrations and to Digit-Symbol Substitution Test (DSST) scores after a single dose
`of triazolam, a representative benzodiazepine agonist.
`
`Methods
`Thirteen healthy male subjects were given 0.375 mg triazolam or placebo in a double-
`blind crossover study. Plasma samples were collected during 8 h after dosage.
`Pharmacodynamic effects were measured by DSST and EEG at corresponding times.
`
`Results
`Pharmacokinetic parameters for triazolam were consistent with established values.
`Compared with placebo, triazolam significantly impaired psychomotor performance
`<
`<
`P
`P
`on the DSST (
`
` 0.001) and increased beta amplitude on the EEG (
`
` 0.002).
`DSST and EEG changes both closely tracked changes in plasma concentrations over
`=
`r
`time. The changes for the two measures were highly correlated with each other (
`
`<
`–
`P
`0.94,
`
` 0.001) based on aggregate values at individual time points. However, the
`vs.
`variations in area under the curve of pharmacodynamic effect
` time (AUC
`)
`effect
`measured by either method did not reflect the variations in plasma AUC across
`individuals. The individual variability in AUC
` from the EEG was similar to that
`measured by the DSST.
`
`effect
`
`Conclusions
`Both the EEG and the DSST reflect the central benzodiazepine agonist effects of
`triazolam. Intrinsic variability in both measures is similar.
`
`Introduction
`Benzodiazepines are widely prescribed drugs for the
`treatment of anxiety, insomnia, seizures, alcohol with-
`drawal, and many other disorders [1–4]. Understanding
`of the clinical effects of benzodiazepines requires appro-
`priate measures for quantification of their central ner-
`vous system (CNS) actions. Methods used to assess
`pharmacodynamic response include (i) subjective mea-
`
`Br J Clin Pharmacol
`
`60
`
`:3
`
`244–248
`
`244
`
`sures, through rating of sedative or antianxiety effects
`by the subject or by an observer; (ii) semiobjective mea-
`sures, such as psychomotor tests, memory tests, and
`critical flicker fusion frequency; (iii) objective mea-
`sures, such as the electroencephalogram (EEG), sac-
`cadic eye movements, postural sway, etc. [5–10].
`The subjective and semiobjective measures have
`limitations, in that they are influenced by practice, adap-
`AQUESTIVE EXHIBIT 1135 Page 0001
`
`© 2005 Blackwell Publishing Ltd
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Pharmacokinetics and pharmacodynamics of single-dose triazolam
`
`tation, placebo response, interpretation, fatigue, and
`motivation. One extensively used measure of this type
`is the Digit-Symbol Substitution Test (DSST) [5–11].
`Objective measures do not have these limitations and
`are sensitive, continuous and reproducible. Based on
`quantitative analysis of the EEG, acute administration
`of benzodiazepine derivatives produces replicable EEG
`changes that are dependent on plasma concentration [5–
`7,11]. However, it is not established whether the EEG
`is an indirect measure of drug effect unrelated to the
`primary clinical action, or whether the EEG reflects the
`primary effect of benzodiazepines on the brain.
`The present study evaluated the relation of the time-
`course of the EEG to the DSST after a single dose of a
`benzodiazepine, and the relation of the EEG or the
`DSST to plasma drug concentrations. Triazolam, a ben-
`zodiazepine derivative having a short half-life [12–14],
`was used as a pharmacological model. We did not
`address the question of the mechanism of benzodiaz-
`epine effects on the EEG.
`
`Materials and methods
`The protocol was approved by the Human Investigation
`Review Committee serving Tufts-New England Medical
`Center and Tufts University School of Medicine. Fifteen
`healthy male subjects participated in the study after
`giving written informed consent. Two subjects were
`withdrawn during the trials because of protocol non-
`compliance. Therefore 13 subjects, aged 20–35 years
`(nine Caucasian and two African-American), completed
`the entire study. All were in good health based on med-
`ical history, physical examination and routine laboratory
`tests. The subjects were nonsmokers and not taking any
`medications.
`This was a placebo-controlled, double-blind, single-
`dose, two-way crossover study. Subjects initially under-
`went a nonblind ‘practice’ trial to allow familiarity with
`the testing procedures, thereby minimizing the effects
`of practice. Data from this trial were not used in the
`analyses. The subsequent two trials were under random-
`ized, double-blind conditions. The two medications
`were placebo and 0.375 mg triazolam, which were iden-
`tically packaged. The interval between trials was at least
`one week.
`Subjects fasted overnight and had a light liquid break-
`fast (orange juice) at around 07.00 h on the day of study.
`They arrived at the Clinical Psychopharmacology
`Research Unit at approximately 08.00 h and remained
`fasting until 12.00 h. After 12.00 h, they resumed a nor-
`mal diet (without grapefruit juice or caffeine-containing
`foods). A single dose of the medication was given orally
`with 200 ml of tap water at approximately 09.00 h.
`
`Venous blood samples (8 ml each) were drawn prior
`to and at 10, 20, 30, 40, 50 min, and 1, 1.25, 1.5, 1.75,
`2, 2.5, 3, 4, 5, 6, 7, and 8 h after the drug administration.
`The blood samples were centrifuged and the plasma was
`-
`∞
`separated and stored at
`20
`C until the time of assay.
`The DSST was used to assess psychomotor perfor-
`mance [13, 14]. The DSST was administered twice prior
`to dosing and at times corresponding to blood sampling.
`A worksheet, on which digits (0–9) were arranged ran-
`domly in rows and a code of symbol-for-digit was
`shown on the top, was presented to a subject at each
`time point. Each individual was given a different work-
`sheet at each time point throughout the study. Subjects
`were required to write down as many symbol-for-digit
`substitutions as possible in 2 min. Scores were the total
`number of attempted substitutions.
`The EEG was recorded using a six-electrode mon-
`tage, with instrumentation and methodology described
`previously [5, 13, 15]. At two predosing times and dur-
`ing 8 h postdosing at times corresponding to blood
`sampling, the EEG was quantified in 4-s epochs for as
`long as necessary to ensure at least 2 min of artefact-
`free recording. Subjects were kept awake throughout
`the study. During the EEG recording, subjects were
`instructed to relax with their eyes closed. Data were
`digitized over the power spectrum from 4 to 30 Hz and
`then fast Fourier-transformed to determine amplitude of
`the total spectrum (4–30 Hz) and the beta frequency
`range (13–30 Hz).
`Plasma concentrations of triazolam were determined
`by gas chromatography with electron-capture detection
`-
`1
`[15]. The detection limit was 0.2 ng ml
`. The intra-
`assay variance did not exceed 10%, and the interassay
`variance did not exceed 12%.
`Pharmacokinetic parameters for triazolam were deter-
`mined by nonlinear regression analysis [16]. Data points
`were fitted by weighted nonlinear regression to a linear
`sum of two or three exponential terms, consistent with
`first-order absorption and a one-compartment model
`(nine subjects) or a two-compartment model (one sub-
`ject), with incorporation of a lag time prior to the start
`of first-order absorption [16]. For three subjects, nonlin-
`ear regression did not provide an adequate fit of the data
`points; for these individuals, model-independent analy-
`sis was used. Standard pharmacokinetics were used to
`calculate the elimination half-life (
`), total area under
`t
`1/2
`the plasma concentration curve (AUC), and apparent
`oral clearance (CL). Also determined were the peak
`plasma concentration (
`) and the time of peak con-
`C
`).
`centration (T
`max
`For DSST scores, the two predosing scores were aver-
`aged and used as the baseline value. Post-dosing scores
`AQUESTIVE EXHIBIT 1135 Page 0002
`60
`
`:3
`
`245
`
`max
`
`Br J Clin Pharmacol
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`D.J. Greenblatt et al.
`
`a
`
`3.0
`
`2.5
`
`2.0
`
`1
`
`2
`
`5
`4
`3
`Hours after dose
`
`6
`
`7
`
`8
`
`Placebo
`
`Triazolam
`
`1
`
`2
`
`4
`3
`5
`Hours after dose
`
`6
`
`7
`
`8
`
`Triazolam
`
`Placebo
`
`1
`
`2
`
`3
`4
`5
`6
`Hours after dose
`
`7
`
`8
`
`1.5
`
`1.0
`
`0.5
`
`b
`10
`
`5 0
`
`–5
`
`–10
`
`–15
`
`–20
`
`–25
`
`–30
`
`c
`
`10
`
`8 6 4 2
`
`0
`
`–2
`
`–4
`
`–6
`
`(change over baseline)
`
`Plasma triazolam (ng/ml)
`
`DSST score (change over baseline)
`
`Percent beta EEG amplitude
`
`Figure 1
`(a) Plasma concentrations of triazolam at corresponding times
`±
`=
`n
`(mean
` SE,
`
` 13). Line represents the function of best fit based on a
`linear sum of two exponential terms, modified by a lag time (0.16 h)
`elapsing prior to the start of absorption; (b) Pharmacodynamic effects of
`±
`=
`n
`triazolam and placebo at corresponding times (mean
` SE,
`
` 13) for the
`DSST; (c) Pharmacodynamic effects of triazolam and placebo at
`±
`=
`n
`corresponding times (mean
` SE,
`
` 13) for the EEG
`
`AQUESTIVE EXHIBIT 1135 Page 0003
`
`were expressed as the increment or decrement over the
`mean predose value.
`For each EEG recording session, the ratio (in per
`cent) of beta amplitude divided by total amplitude was
`calculated. The mean value obtained from the two pre-
`dose recordings was used as baseline. All values after
`drug administration were expressed as the increment or
`decrement over the baseline value.
`The area under the effect-
`-time curve (AUC
`)
`vs.
`effect
`during 8 h after triazolam or placebo administration was
`calculated using the linear trapezoidal method for both
`DSST and EEG. The relation of DSST changes and
`EEG changes were evaluated in two ways. First, mean
`values of DSST changes at each observed time point
`were compared with EEG changes at corresponding
`times. The mean values during the triazolam trial were
`normalized by subtracting the values associated with
`placebo at corresponding times. Second, values of
`AUC
` from the two measures (placebo-normalized)
`effect
`were compared across the 13 subjects.
`To evaluate the concentration–effect relation, the
`mean placebo-normalized changes in DSST and EEG
`across the 13 subjects at individual time points were
`plotted against plasma triazolam concentrations at cor-
`responding times. Inspection of the plots indicated that
`a ‘maximum’ pharmacodynamic effect was not attained.
`Accordingly, data points were analysed by nonlinear
`regression using an exponential model [5, 17]. The
`=
`+
`A
` B*C
`equation was: E
`
` K, where E is pharmacody-
`namic effect and C is plasma concentration. Iterated
`variables were the coefficient B, the exponent A, and a
`constant K. This model allows inferences regarding the
`relation between plasma concentration and effect within
`the observed range of concentrations. However the
`model does not allow extrapolation to plasma concen-
`trations exceeding this range.
`The relation between the plasma AUC and the
`AUC
` also was evaluated across 13 subjects.
`effect
`-test, linear
`Statistical procedures included Student’s
`t
`and nonlinear regression. Pearson product-moment
`correlation analysis was used to evaluate the relation
`between EEG effects, DSST effects, and plasma
`concentrations.
`
`Results
`±
`Mean (
`SD) pharmacokinetic parameters for triazolam
`±
`-
`1
`(Figure 1) were:
`, 2.8
` 0.9 ng ml
` (95% CI: 2.2–
`C
`max
`±
`±
`, 1.2
` 0.5 h (95% CI: 0.9–1.6);
`, 3.4
` 1.2
`3.6); T
`t
`1/2
`max
`±
`-
`1
` 129 ml min
`h (95% CI: 2.7–4.2); CL, 455
` (95% CI:
`374–536). The coefficient of variation (CV) in plasma
`AUC among the 13 subjects was 28%.
`Triazolam, but not placebo, produced a reduction in
`
`246
`
`60
`
`:3
`
`Br J Clin Pharmacol
`
`

`

`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`
`Pharmacokinetics and pharmacodynamics of single-dose triazolam
`
`0.5
`1.0
`1.5
`2.0
`Plasma triazolam (ng/ml)
`
`2.5
`
`b
`
`5
`
`0
`
`–5
`
`–10
`
`–15
`
`–20
`
`–25
`
`–30
`
`placebo-normalized)
`(change over baseline,
`
`DSST score
`
`10
`8
`6
`4
`2
`Percent beta EEG amplitude
`(change over baseline,
`placebo-normalized)
`
`12
`
`a
`
`5
`
`0
`
`–5
`
`–10
`
`–15
`
`–20
`
`–25
`
`–30
`
`placebo-normalized)
`(change over baseline,
`
`DSST score
`
`20
`15
`10
`5
`Triazolam plasma AUC
`
`25
`
`c
`30
`
`0
`
`–30
`
`–60
`
`–90
`
`–120
`
`–150
`
`–180
`
`(placebo-normalized)
`
`DSST effect AUC
`
`Figure 2
`(a) Correlation between DSST and EEG measures using mean values of EEG change and DSST change at corresponding times (r = –0.94, P < 0.001)
`based on an exponential equation (y = -1.13x1.4 +3.78). Relative asymptotic standard errors (in per cent) for parameter estimates in the fitted function
`were: 1.13 (±103%); 1.4(±29%); 3.78 (±95%); (b) The relation of mean plasma triazolam concentrations to mean DSST changes over baseline
`(placebo-normalized) at the corresponding time (r = -0.99). Solid line represents an exponential equation [E = -2.14C2.93 + 0.33] fitted to data points
`using nonlinear regression. Relative asymptotic standard errors (in percent) for parameter estimates in the fitted function were: 2.14 (±33%); 2.93
`(13%); 0.33 (±288%); (c) Relation between the plasma triazolam AUC and the placebo-normalized AUCeffect for DSST change score among 13 subjects
`(r = 0.04, NS). Units for AUCeffect (y-axis) are: DSST change score ¥ h. Units for plasma AUC (x-axis) are: ng ml-1 ¥ h
`
`DSST scores and an increase in beta amplitude on
`the EEG (Figure 1). The changes returned to baseline
`±
`levels by 4–8 h after dosing. Mean (
`SD) AUC
`effect
`values for triazolam and placebo for DSST score were:
`-
`±
`-
`-
`±
`57
` 58 (95% CI:
`93 to
`21)
` 16
` 34 (95% CI:
`vs.
`-
`<
`±
`6 to 37;
`
` 0.001). For the EEG, values were: 32
` 35
`P
`-
`±
`-
`
`14
` 24 (95% CI:
`29 to 1;
`(95% CI: 10–54)
`vs.
`<
`
` 0.005). CV values among 13 subjects were: 89%
`P
`, and 95% for
`for placebo-normalized DSST AUC
`placebo-normalized EEG AUC
`.
`effect
`Mean values of EEG change and DSST change for
`13 subjects at corresponding times were strongly corre-
`=
`<
`lated (
`
`
`0.94,
`
` 0.001), based on an exponential
`r
`–
`P
`=
`+
`A
`function of the form: y
` Bx
`
` K, as described previ-
`
`effect
`
`Br J Clin Pharmacol
`
` for each
`ously (Figure 2). However, values of AUC
`effect
`=
`individual were not significantly correlated (
`
`0.418,
`
`–
`r
`NS).
`Mean DSST changes and mean plasma triazolam
`concentrations at corresponding times were highly cor-
`=
`<
`related (
`
`
`0.99,
`
` 0.001) based on an exponential
`r
`–
`P
`function (Figure 2). Mean EEG beta amplitude changes
`and mean plasma concentrations were similarly corre-
`r = 0.94, P < 0.001) through an exponential func-
`lated (
`tion. For both DSST and EEG, no evidence of clockwise
`or counterclockwise hysteresis was found.
`Plasma triazolam AUC for the 13 subjects was not
`significantly correlated with AUCeffect for the DSST (r =
`0.04, NS) (Figure 2) or the EEG (r = 0.21, NS).
`AQUESTIVE EXHIBIT 1135 Page 0004
`60
`
`:3
`
`247
`
`

`

`D.J. Greenblatt et al.
`
`Discussion
`The benefits and disadvantages of various subjective and
`objective measurement techniques used to delineate the
`time-course and intensity of benzodiazepine agonist
`compounds are described previously [5–10]. The DSST
`is a classic psychomotor performance test that is well-
`established as a procedurally straightforward, inexpen-
`sive, and sensitive index of benzodiazepine agonist
`effects. However the DSST is also influenced by prac-
`tice and adaptation as may occur both within and
`between testing sessions. The EEG is a fully objective,
`quantitative, and practice-insensitive measure of central
`benzodiazepine action, but requires specialized instru-
`mentation and may be sensitive to artefact [5, 11]. Also,
`the mechanism of EEG changes associated with benzo-
`diazepine against treatments is still not established. In
`the present study we compared these two approaches
`using the benzodiazepine derivative triazolam as a rep-
`resentative full-agonist ligand.
`DSST decrements and increments in EEG beta ampli-
`tude both clearly distinguished triazolam from placebo,
`either at individual time points, or based on integrated
`8-h effect areas. Both measures had a time-course that
`matched plasma triazolam concentrations. Based on
`mean values at corresponding time points, there was a
`highly significant correlation between plasma level and
`DSST or EEG change. Further, DSST and EEG changes
`themselves were highly
`intercorrelated. However,
`plasma AUC values appeared unrelated to placebo-nor-
`malized AUCeffect for both DSST and EEG, and AUCeffect
`for DSST and AUC were poorly correlated with each
`other. Finally, the between-subject coefficient of varia-
`tion for plasma AUC (standard deviation 28% of the
`mean) was much less than the coefficient of variation
`for placebo-normalized values of AUCeffect for DSST
`(89%) or EEG (95%).
`The results of this study indicate that between-subject
`variability in triazolam kinetics is exceeded by variabil-
`ity in response. Variance in response for DSST and EEG
`is similar, suggesting that factors other than intrinsic
`variability should be considered in distinguishing these
`two pharmacodynamic methods. Based on values aggre-
`gated across subjects at individual times, plasma con-
`centrations, DSST changes, and EEG changes had a
`similar time course and were highly intercorrelated.
`However net kinetic and dynamic exposure measures
`between subjects were poorly correlated, indicating high
`variability in individual sensitivity to benzodiazepine
`agonist effects.
`
`Supported by Grants AG-17880, MH-58435, DA-05258,
`DK-58435, DA-13834, DA-13209, AT-01381, and RR-
`
`248
`
`60:3
`
`Br J Clin Pharmacol
`
`00054 from the Department of Health and Human
`Services.
`
`References
`1 Greenblatt DJ, Shader RI, Abernethy DR. Current status of
`benzodiazepines. N Engl J Med 1983; 309 (354–358): 410–6.
`2 Greenblatt DJ, Shader RI. Benzodiazepines in Clinical Practice, New
`York: Raven Press 1974.
`3 Shader RI, Greenblatt DJ. Use of benzodiazepines in anxiety disorders.
`N Engl J Med 1993; 328: 1398–405.
`4 Hollister LE, Müller-Oerlinghausen B, Rickels K, Shader RI. Clinical uses
`of benzodiazepines. J Clin Psychopharmacol 1993; 13(Suppl 1): 1S–
`169S.
`5 Laurijssens BE, Greenblatt DJ. Pharmacokinetic-pharmacodynamic
`relationships for benzodiazepines. Clin Pharmacokinetics 1996; 30:
`52–76.
`6 Greenblatt DJ, von Moltke LL, Harmatz JS, Shader RI. Pharmacokinetics,
`pharmacodynamics, and drug disposition. In:
`Neuropsychopharmacology: the Fifth Generation of Progress, eds
`Davis KL, Charney D, Coyle JT, Nemeroff C. Philadelphia: Lippincott,
`Williams & Wilkins; 2002; pp. 507–24.
`7 Greenblatt DJ, von Moltke LL. Drug–drug interactions: clinical
`perspective. In: Drug–Drug Interactions, ed Rodrigues AD. New York:
`Marcel Dekker; 2002; pp. 565–84.
`8 Johnson LC, Chernik DA. Sedative-hypnotics and human performance.
`Psychopharmacology 1982; 76: 101–13.
`9 Wittenborn JR. Effects of benzodiazepines on psychomotor
`performance. Br J Clin Pharmacol 1979; 7: 61S–67S.
`10 Stone BM. Pencil and paper tests – sensitivity to psychotropic drugs.
`Br J Clin Pharmacol 1984; 18: 15S–20S.
`11 Mandema JW, Danhof M. Electroencephalogram effect measures and
`relationships between pharmacokinetics and pharmacodynamics of
`centrally acting drugs. Clin Pharmacokinetics 1992; 23: 191–215.
`12 Friedman H, Greenblatt DJ, Burstein ES, Harmatz JS, Shader RI.
`Population study of triazolam pharmacokinetics. Br J Clin Pharmacol
`1986; 22: 639–42.
`13 Greenblatt DJ, Harmatz JS, von Moltke LL, Wright CE, Durol AL, Harrel-
`Joseph LM, Shader RI. Comparative kinetics and response to the
`benzodiazepine agonists triazolam and zolpidem: evaluation of sex-
`dependent differences. J Pharmacol Exp Therapeutics 2000; 293:
`435–43.
`14 Greenblatt DJ, Harmatz JS, Shapiro L, Engelhardt N, Gouthro TA, Shader
`RI. Sensitivity to triazolam in the elderly. N Engl J Med 1991; 324:
`1691–8.
`15 von Moltke LL, Greenblatt DJ, Harmatz JS, Duan SX, Harrel LM, Cotreau-
`Bibbo MM, Pritchard GA, Wright CE, Shader RI. Triazolam
`biotransformation by human liver microsomes in vitro: effects of
`metabolic inhibitors and clinical confirmation of a predicted interaction
`with ketoconazole. J Pharmacol Exp Therapeutics 1996; 276:
`370–9.
`16 Harmatz JS, Greenblatt DJ. A SIMPLEX procedure for fitting nonlinear
`pharmacokinetic models. Computers Biol Med 1987; 17: 199–208.
`17 Greenblatt DJ, Harmatz JS. Kinetic-dynamic modeling in clinical
`psychopharmacology. J Clin Psychopharmacol 1993; 13: 231–4.
`
`AQUESTIVE EXHIBIT 1135 Page 0005
`
`