A Pharmacokinetic/Pharmacodynamic Study Comparing
`a Single Morning Dose of Adderall to Twice-Daily
`Dosing in Children With ADHD
`
`LAURENCE L. GREENHILL, M.D., JAMES M. SWANSON, PH.D., KEN STEINHOFF, M.D., JANE FRIED, M.D.,
`KELLY POSNER, PH.D., MARC LERNER, M.D., SHARON WIGAL, PH.D., SUSAN B. CLAUSEN, PH.D.,
`YUXIN ZHANG, PH.D., AND SIMON TULLOCH, M.D.
`
`ABSTRACT
`Objective: To determine the pharmacokinetic and pharmacodynamic properties of once-daily versus twice-daily doses
`of Adderall®. Method: Following a 1-week wash-out, 12 subjects with attention-deficit/hyperactivity disorder (ADHD)
`entered a double-blind crossover study comparing two conditions: QD (10 mg of Adderall at 7:30 A.M. and placebo at
`noon) or BID (10 mg of Adderall at 7:30 A.M. and at noon). At two sites, cohorts of six subjects each were assessed on
`two different days by a 12-hour laboratory school protocol. Plasma concentrations of d- and l-amphetamine, vital signs,
`teacher ratings of classroom behavior on the SKAMP, and 10-minute Math Test performance were measured repeatedly
`over 12 hours. An analysis of variance used center, subject-within-center, condition, and time-after-second-dose as
`independent variables. Results: The pharmacokinetic profiles revealed similar morning concentrations of d- and l-
`amphetamine. However, concentrations were twice as high in the afternoon for BID as QD. The two conditions showed
`similar pharmacodynamic profiles in the morning, although improvement in math performance and behavior was main-
`tained into the afternoon only in the BID condition (p < .05). Conclusions: This study suggests that twice-daily dosing
`of Adderall may be an effective strategy for afternoon control of attention and deportment for children with ADHD. J. Am.
`Acad. Child Adolesc. Psychiatry, 2003, 42(10):1234–1241. Key Words: attention-deficit/hyperactivity disorder, labora-
`tory school protocol, amphetamine, Adderall, pharmacokinetic, pharmacodynamic.
`
`For more than 50 years, stimulant medication has
`served as the first-line treatment for attention-
`deficit/hyperactivity disorder (ADHD), which is esti-
`mated to affect 3% to 5% of the school-age population
`(American Academy of Child and Adolescent Psychia-
`try, 2002; American Psychiatric Association, 1994;
`Greenhill, 1998; Hinshaw, 1994; Swanson, 1992). Al-
`
`Accepted May 13, 2003.
`Drs. Greenhill, Fried, and Posner are with the New York State Psychiatric
`Institute; Drs. Swanson, Steinhoff, Lerner, and Wigal are with the University
`of California at Irvine; Drs. Tulloch, Clausen, and Zhang are with Shire
`Pharmaceutical Development, Inc., Rockville, MD.
`This study was funded by the Shire Pharmaceutical Development, Inc.,
`Rockville, MD. Drs. Greenhill and Swanson are paid consultants for Shire
`Pharmaceutical Development, Inc.
`Correspondence to Dr. Greenhill, Attention Deficit Disorder Research Pro-
`gram, New York State Psychiatric Institute, 1051 Riverside Drive, New York,
`NY 10032; e-mail: LarryLGreenhill@cs.com.
`0890-8567/03/4210–1234©2003 by the American Academy of Child
`and Adolescent Psychiatry.
`DOI: 10.1097/01.CHI.0000081805.08150.34
`
`though amphetamine (AMP) was the first stimulant
`used (Bradley, 1937), in the 1970s methylphenidate
`became the primary medication for the treatment of
`ADHD, and by the mid-1990s approximately 80% of
`prescriptions for stimulants to treat ADHD were for
`methylphenidate (Swanson et al., 1995). Since the late
`1990s there has been an increase in the use of a racemic
`formulation of AMP (75% d-AMP and 25% l-AMP),
`which was reintroduced and marketed as Adderall®.
`Initially, a once-a-day (QD) dosing regimen was rec-
`ommended for Adderall. To our knowledge, there were
`no pharmacodynamic (PD) or pharmacokinetic (PK)
`Adderall data in children previously published to sup-
`port this claim. PK studies of Dexedrine (d-AMP) in
`ADHD children (Brown et al., 1978, 1979, 1980) have
`demonstrated a shorter T1⁄2 (about 7 hours) than ex-
`pected from the PK studies of adults.
`PD studies of Adderall (Pelham et al., 1999; Swan-
`son et al., 1998a) suggested that higher doses of AMP
`might have a longer duration of action. A double-blind
`
`1234
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`Page 1
`
`SHIRE EX. 2044
`KVK v. SHIRE
`IPR2018-00290
`
`

`

`crossover study in a laboratory school protocol (LSP)
`that compared four doses of Adderall (5, 10, 15, and 20
`mg) with an inactive placebo control and an active
`control (a clinical dose of MPH) (Swanson et al.,
`1998a) reported that higher doses of Adderall extended
`the calculated length of action (from 3.5 to 6.4 hours).
`However, this was at the expense of an increase in the
`time of peak effect (from 1.5 to 3.0 hours) in the
`highest-dose condition (20 mg). Even though higher
`doses extended the half-life more than 50%, the total
`coverage period was less than 7 hours, hardly long
`enough for a standard school day.
`We speculated that by using a 10-mg immediate-
`release (IR) BID dosing regimen instead of a 20-mg IR
`QD dose, it would be possible to extend the length of
`action in the afternoon without producing a high AMP
`concentration in the morning, which could produce
`increased side effects in a 6-year-old child with ADHD
`due to the linear relationship between side effects and
`dose for amphetamines. A double-blind trial of QD
`and BID conditions was run to compare their PK and
`PD effects in the controlled LSP setting to address this
`question and to investigate the basic mechanism in-
`volved in response to Adderall.
`
`METHOD
`
`The study was conducted at two academic medical centers (Uni-
`versity of California at Irvine and Columbia University), whose
`institutional review boards approved all procedures prior to subject
`recruitment. All subjects provided written or verbal assent for study
`participation; parents provided written consent for their child’s
`enrollment.
`
`Subjects
`
`Subjects meeting the following inclusion criteria were recruited:
`(1) age 7 to 12 years; (2) meet the diagnostic criteria for DSM-IV
`ADHD (combined or hyperactive-impulsive subtype as determined
`by clinician evaluation and selected modules of the Diagnostic
`Interview Schedule for Children, Version IV-Lifetime) (Shaffer et
`al., 1996); (3) have a history of a clinically significant response to
`typical doses of methylphenidate (5–20 mg, BID or TID); (4) have
`parental confirmation that the child could attend the two full-day
`test sessions, 1 week apart, that were scheduled.
`Exclusion criteria rejected subjects with (1) blood pressure and
`pulse outside the 95th percentiles for age and gender; (2) abnor-
`malities on physical examination, including the presence of an acute
`or chronic disease such as anemia, hypertension, glaucoma, or hy-
`perthyroidism, or a medical history of a nonfebrile seizure disorder;
`(3) family history of suspected substance use disorders (excluding
`nicotine); (4) family history of Tourette’s disorder; (5) use of clo-
`nidine, anticonvulsant medications, or other CNS medications; (6)
`history of an adverse reaction or nonresponse to Adderall or hy-
`persensitivity to any AMP product; (7) excessive fear of needles; (8)
`history of aggressive behavior incompatible with regular classroom
`activities, as shown by a diagnosis of childhood-onset persistent
`
`PK/PD ADDERALL TRIAL
`
`conduct disorder; (9) history of comorbid psychosis, bipolar illness,
`pervasive developmental disorder, tic disorders, severe obsessive-
`compulsive disorder, severe depression, conduct disorder, panic dis-
`order, current suicidal ideation, or severe anxiety; (10) Full Scale IQ
`less than 80 as assessed by the WISC III. Females who had reached
`menarche were also excluded.
`
`Laboratory Classroom Protocol
`
`The LSP setting was designed to control the timing and context
`of repeated observations over an entire day of testing (Swanson et
`al., 1998b; Wigal et al., 1998). For this study, the standard LSP
`developed at the University of California at Irvine (Swanson et al.,
`2000) was transferred to a second site (Columbia University).
`Each site tested a cohort of six subjects in the LSP on two
`consecutive Saturdays. Before the two LSP test days, each child was
`invited to come to the site to become familiar with the LSP pro-
`cedures. On each LSP test day, subjects arrived at the laboratory
`school at approximately 7 A.M., and indwelling catheters were
`placed before capsules were administered at 7:30 A.M. The daily
`schedule consisted of alternating classroom, recess, and other ac-
`tivities. The classroom sessions began after the morning dose ad-
`ministration (8 A.M.) and occurred every hour for the next 3 hours
`(9, 10, and 11 A.M.). Classroom sessions were scheduled in the
`afternoon following the noon dose and began within an average
`time of 0, 1.5, 3.5, and 6 hours after the second dose (i.e., spread
`across the afternoon at noon, 1:30, 3:30, and 6:00 P.M.). Each
`classroom period lasted a total of 30 minutes and was directed by
`two teachers for a cohort of six subjects. In addition, each classroom
`contained two observers (trained to be reliable) who rated classroom
`behavior after each session using a system described previously
`(Swanson et al., 2000). Outside of the classroom period, a separate
`staff (counselors) directed and supervised the nonclassroom activi-
`ties across the day. No behavioral treatments were used during the
`LSP days.
`
`Medication Dosing
`
`Prior to the two test days, subjects underwent a 6-day washout of
`stimulants and other psychotropic medications. On each LSP test
`day, a pharmacist or a physician administered a capsule with an
`initial 10-mg dose of Adderall to each subject at 7:30 A.M. (30
`minutes before breakfast) and a second capsule at noon (30 minutes
`before lunch), which contained either 10 mg of Adderall (the BID
`condition) or placebo (the QD condition). The order of the two
`conditions (BID-QD or QD-BID) was randomized across subjects
`and was established under double-blind conditions.
`
`PK Sampling
`
`On arrival to the initial and final analog classroom sessions,
`indwelling catheters for plasma sampling were inserted into an
`antecubital vein in each subject. Pharmacokinetic sampling was
`conducted predose and 0.5, 1.5, 3, 4.5, 6, 7.5, 9, 10.5, 12, and 24
`hours following dosing. Blood samples were collected in 10-mL
`EDTA Vacutainer tubes and plasma was prepared immediately by
`centrifugation of the blood samples. Plasma was transferred and
`stored at approximately −20°C prior to shipping for analysis.
`
`Analytic Methods
`
`Plasma samples were analyzed for AMP concentrations (d-AMP
`and l-AMP) by high-performance liquid chromatography (HPLC)
`with turbo-ion spray tandem mass spectometry (LC/MS/MS) with
`chiral separation. The assay involved alkalinization of the plasma
`prior to extraction and back-extraction into acid. Following real-
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`1235
`
`Page 2
`
`

`

`GREENHILL ET AL.
`
`kalinization, the benzoyl derivatives were prepared prior to HPLC
`with a chiral column.
`For d- and l-AMP, concentrations were linear over the range of
`0.5 to 50 ng/mL. A weighted [(1/x), where x = the concentration of
`the compound] linear regression was used to determine slopes,
`intercepts, and correlation coefficients for d- and l-AMP levels in
`study samples and quality control samples. PK variables were cal-
`culated for plasma drug concentration-time area under the curve
`(AUC0-24 and AUC0-inf), maximum drug concentration (Cmax), and
`time to Cmax (Tmax).
`
`Dependent Measures
`
`Primary efficacy variables were the classroom ratings of the At-
`tention and Deportment subscales of the Swanson, Alger, M-Flynn,
`and Pelham (SKAMP) rating scale and performance on a 10-
`minute Math Test (number of problems attempted and number of
`problems solved correctly). These measures have been shown to be
`drug-sensitive in prior research (Wigal et al., 1998).
`In addition, to provide secondary measures of efficacy and side
`effects, parents were instructed to complete a behavior rating scale
`at mid-week and a side effect rating scale at the end of each week,
`and to keep a diary to record adverse events, such as delay of sleep
`onset. Each week, teachers also completed the Teacher Side Effect
`Rating Scale. During the analog classroom day, adverse events were
`noted by study physicians and research staff.
`
`Statistical Analyses
`
`Mean d- and l-AMP concentrations were measured repeatedly
`over time for the QD and BID conditions. The PK parameters for
`each child were calculated on the basis of a noncompartmental
`modal. The AUC was computed using trapezoidal rule. Analyses
`were performed to evaluate effects of condition (QD versus BID)
`on levels of the d- and l-isomers, as well as on the derived PK
`parameters. To evaluate PD effects, analysis of variance (ANOVA)
`was performed using a 2 × 4 design with condition (QD and BID)
`and time-after-second-dose (0, 1.5, 3.5, and 6.0 hours) as the in-
`dependent variables and SKAMP ratings of attention and deport-
`ment and number of problems attempted and solved correctly as
`dependent variables.
`
`RESULTS
`
`Patient Demographics
`
`Characteristics of the sample are shown in Table 1.
`Eleven boys and one girl participated in the study. The
`mean age of the total sample was 9.8 years (± 1.9).
`Nine of the patients were Caucasian (75%) and the
`remainder were Hispanic (25%).
`
`PK Results
`
`The mean plasma concentration profiles for d- and
`l-AMP following QD and BID dosing are presented in
`Table 2. For the QD condition, the average AUC was
`about three times greater for the d-isomer (342.1
`ng/h/mL0-24) than the l-isomer (124.2 ng/h/mL0-24),
`which reflects the 3:1 ratio of the isomers in the Adder-
`
`TABLE 1
`Subject Characteristics and Mean Pharmacokinetic Parameters
`Following 10 mg of Adderall Dosed Once-Daily and Twice-Daily
`Site 1
`Site 2
`Total
`
`Subject (n)
`Gender: n (%)
`Male
`Female
`Age (yr)
`Mean (SD)
`Range
`Weight (lb)
`Mean (SD)
`Range
`
`6
`
`6 (100)
`
`9.8 (1.7)
`7.0–12.0
`
`6
`
`12
`
`5 (83.3)
`1 (16.7)
`
`9.8 (2.0)
`8.0–12.0
`
`11 (92)
`1 (8.3)
`
`9.8 (1.0)
`8.0–12.0
`
`90.5 (29.9)
`59.0–144.0
`
`76.0 (26.9)
`55.0–127.0
`
`83.3 (28.2)
`28.0–144.0
`
`all formulation. For both isomers, the values of AUC0-
`24, AUC0-∞, and Cmax were significantly higher
`(approximately twice) for the BID condition compared
`to the QD condition. The elimination constant (ke)
`and the elimination half-life (T1⁄2) did not differ sig-
`nificantly for the two conditions.
`
`PD Properties of Adderall
`
`For each session across the day, mean SKAMP scores
`(attention and deportment) are shown in Figure 1 for
`QD dosing and in Figure 2 for BID dosing. In the BID
`condition, by 1 hour after the morning dose, the mean
`deportment rating decreased by 83% (from a baseline
`of 1.38 to 0.23). The ratings remained low at 2 hours
`after the morning dose (0.27) but then gradually in-
`creased until the second dose was administered at
`noon, when the mean deportment score was 0.58 (still
`a 58% decrease from baseline). Across the remainder of
`the afternoon, the deportment scores showed another
`gradual decrease and were still low (0.29, a 79% de-
`crease from baseline) at 6 hours after the second dose.
`In the QD condition, the mean deportment score
`showed a similar rapid and large (76%) decrease by 1
`hour after the morning dose (from a baseline of 1.46 to
`0.35), and this trend continued for another hour
`(0.31). By 3 hours after the morning dose, the deport-
`ment ratings showed an increase from the minimum
`(to 0.58) and a general trend of increasing across the
`rest of the day, settling at a mean value of 0.77 at 6
`hours after the second (placebo) dose (a 47% decrease
`from baseline).
`Instead of evaluating the relative change from base-
`line, the ANOVA evaluated the difference across the
`four sessions after the second dose between the two
`conditions to gauge the additional efficacy in the after-
`
`1236
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`Page 3
`
`

`

`PK/PD ADDERALL TRIAL
`
`TABLE 2
`Mean Pharmacokinetic Parameters Derived Using a Noncompartmental Model for 10 mg QD Dose of Adderall
`d-Amphetamine
`l-Amphetamine
`
`Tmax (hours)a
`Cmax (ng/mL)a
`AUC0-24 (ng/mL)a
`AUC0-inf (ng/mL)a
`T1/2
`Ke
`
`QD
`
`2.5 ± 1.2
`28.4 ± 6.5
`342.1 ± 108.6
`384.4 ± 108.6
`7.5 ± 1.0
`0.09 ± 0.01
`
`BID
`
`6.5 ± 0.9
`52.7 ± 16.8
`630.6 ± 161.5
`789.3 ± 242.3
`7.8 ± 1.8
`0.09 ± 0.02
`
`QD
`
`2.5 ± 1.2
`9.6 ± 2.4
`124.2 ± 37.2
`146.3 ± 51.6
`8.6 ± 1.6
`0.08 ± 0.02
`
`BID
`
`6.4 ± 0.7
`17.7 ± 5.2
`227.3 ± 65.9
`305.8 ± 112.0
`8.9 ± 2.5
`0.08 ± 0.02
`
`a For both isomers, the differences between QD and BID schedules were statistically significant.
`
`noon of the BID condition over the QD condition. In
`the ANOVA, the average difference for the deportment
`ratings (0.62–0.40) was statistically significant (p =
`.0187), and the condition × time interaction was also
`significant (p = .0038).
`Similar effects were seen for the rating of attention,
`which were expected to be less sensitive for monitoring
`medication effects (Swanson et al., 1998a). In the BID
`condition, by 1 hour after the morning dose, the at-
`tention ratings had improved by 64% (from a baseline
`of 0.75 to 0.27). These ratings remained low across the
`remainder of the morning and the entire afternoon
`after the second dose (from 0.38 to 0.44, for at least a
`45% decrease from baseline). In the QD condition, by
`1 hour after the morning dose, the mean attention
`
`ratings improved by 54% (from a baseline of 0.83 to
`0.38) and continued to improve for the next two
`morning test sessions (to 0.35 and 0.33) and the first
`session after the second dose (to 0.21, for a 75% de-
`crease from baseline). However, starting at 6 hours
`after the morning dose and continuing across the af-
`ternoon, the attention ratings showed an increase from
`the minimum, with mean ratings ranging from 0.52 to
`0.58. In the ANOVA of the difference between the
`QD and BID conditions across the four afternoon ses-
`sions, the same trend reported for the deportment rat-
`ings was present, but it was not statistically significant
`(p = .4557).
`Mean scores on the math test for the two measures
`(number attempted and number solved) were highly
`
`Fig. 1 SKAMP score: pharmacokinetics and pharmacodynamics: 10 mg of Adderall given at 8 A.M.
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`1237
`
`Page 4
`
`

`

`GREENHILL ET AL.
`
`Fig. 2 SKAMP score: pharmacokinetics and pharmacodynamics 10 mg of Adderall given at 8 A.M. and noon.
`
`correlated, so the results for only one (number solved)
`will be reported. In the BID condition, by 1 hour after
`the morning dose, the number solved had increased by
`23% (from a baseline of 125.9 to 155.3), and this trend
`continued and reached a maximum of 172.3 (a 37%
`increase over baseline) at 2 hours after the morning
`dose. Then performance declined slightly to 158.3. Af-
`ter the noon dose, the increasing trend was reinstated
`over the afternoon, with a range of scores from 164.2
`to 169.9. In the QD condition, a 30% increase from
`baseline occurred by 1 hour after the morning dose
`(from 130.1 to 169.2), and this trend continued and
`reached a maximum (180.3) at 2 hours after the morn-
`ing dose. Starting at the last morning session before the
`noon dose and continuing across the remainder of the
`day, a gradual decrease was observed with a score of
`156.8 at 10 hours after the morning dose. In the
`ANOVA of the difference between the BID and QD
`conditions, the average performance across the after-
`noon in the BID condition (166.2) compared to the
`QD condition (156.5) was statistically significant (p =
`.0129), but the condition × time interaction was not
`(p = .7763).
`Pearson correlation coefficients based on the concen-
`trations of AMP and the PD measures of efficacy are
`shown in Table 3. In the QD condition, all correlations
`
`were negative for the SKAMP ratings, and they were
`significant (at p < .05) for Attention (−0.344 for the
`l-isomer and −0.344 for the d-isomer) and for deport-
`ment (−0.526 for the l-isomer and −0.528 for the d-
`isomer). In the BID condition, the correlations for
`attention were negative but not significant (−0.236 for
`the l-isomer and −0.252 for the d-isomer), but the
`correlations for deportment were significant at p < .05
`(−0.555 for the d-isomer and −0.567 for the l-isomer).
`The number of math problems solved increased with
`increasing plasma concentrations, generating positive
`and statistically significant (at p < .05) correlations for
`both the QD (0.548) and BID (0.553) conditions.
`
`Adverse Events
`
`Four subjects reported adverse events in the QD
`condition (one subject had headaches, one had insom-
`nia, one had blurry vision, and one had a bloody nose
`and diarrhea) and two in the BID condition (one sub-
`ject complained of tiredness and another complained of
`stinging around his IV catheter and paresthesias and hit
`his head on the floor). All adverse events were mild in
`severity and resolved within the same day. No serious
`adverse events were reported during the study, and no
`subjects discontinued because of an adverse event.
`
`1238
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`Page 5
`
`

`

`PK/PD ADDERALL TRIAL
`
`TABLE 3
`Pharmacokinetic/Pharmacodynamic Correlations
`Adderall 10 mg QD
`
`ra
`
`−0.344
`−0.526
`0.544
`0.548
`
`−0.344
`−0.528
`0.548
`0.553
`
`95% CI
`
`(−0.589, −0.099)
`(−0.749, −0.303)*
`(0.350, 0.738)*
`(0.214, 0.664)*
`
`(−0.587, −0.101)*
`(−0.744, −0.312)*
`(0.360, 0.736)*
`(0.353, 0.753)*
`
`ra
`
`−0.236
`−0.555
`0.439
`0.426
`
`−0.252
`−0.567
`0.461
`0.450
`
`Adderall 10 mg BID
`
`95% CI
`
`(−0.520, 0.048)
`(−0.720, −0.390)*
`(0.214, 0.664)*
`(0.201, 0.651)*
`
`(−0.528, 0.024)
`(−0.728, −0.406)*
`(0.245, 0.677)*
`(0.234, 0.666)*
`
`l-Amphetamine
`SKAMP-Attention
`SKAMP-Deportment
`Math question-attempted
`Math questions-solved
`d-Amphetamine
`SKAMP-Attention
`SKAMP-Deportment
`Math questions-attempted
`Math questions-solved
`
`Note: CI = confidence interval.
`a Pearson correlation coefficient.
`* p < .05.
`
`DISCUSSION
`
`To our knowledge, this is the first report of the PK
`properties of IR Adderall in ADHD children. A typical
`dose of Adderall (10 mg), administered in the morning
`before breakfast, reached a peak concentration (Tmax)
`2.5 hours after dosing, which fell to half its level by 7.5
`hours (for the d-isomer) to 8.5 hours (for the l-isomer)
`when given once daily. The addition of a second dose
`4 hours after the first nearly doubled the peak concen-
`trations in the afternoon (52.7 ng/mL for the d-isomer
`and 17.7 ng/mL for the l-isomer) compared to the
`morning peaks in the QD condition (28.4 ng/mL for
`the d-isomer and 9.6 ng/mL for the l-isomer), with a
`peak concentration about 2.5 hours after the second
`dose. The T1⁄2 increased slightly in the BID condition
`(7.8 for the d-isomer and 8.9 hours for the l-isomer). In
`both the QD and BID conditions, the relative concen-
`trations of the d- and l-isomers were equivalent to the
`relative administered doses of the isomers (about 3:1).
`These PK data for Adderall (a racemic mixture of the
`d- and l-isomers) are partially consistent with existing
`data from early studies of the pure d-isomer (Dexe-
`drine) performed by Brown et al. (1979). The T1⁄2
`value calculated for the d-isomer here (about 7.5 hours)
`slightly exceeded the value (6.8 hours) reported by
`Brown et al. (1978). The Tmax value also was slightly
`greater (2.5 hours versus 3.5 hours). Brown et al., in
`the study mentioned above, reported a substantially
`higher Cmax than in the QD condition of the present
`study (65.9 versus 28.5 ng/mL), but they used a higher
`average dose of d-AMP (0.5 mg/kg versus 0.22 mg/kg).
`Cmax normalized for mg dose and kg weight [Cmax ×
`
`(kg/mg)] was equivalent for the Brown et al. study (65.9 ×
`2 = 130.8) and the present study (28.5 × 4.55 = 129.7).
`Borcherding et al. (1989) reported an even higher
`average Cmax (140 ng/mL) for a d-AMP dose of 0.6
`mg/kg. However, this was after 3 weeks of chronic
`dosing, compared to the acute dose after a 6-day wash-
`out in the present study. In the Borcherding et al.
`study, the morning predose concentration was high (60
`ng/mL), suggesting substantial carryover for day to day
`and buildup over time. After the morning dose, the
`peak serum level rose from this initial (carryover) level
`by approximately 80 ng/mL. The normalized values for
`the change in concentration [(Cmax − C0) × (kg/mg)]
`are consistent with the values reported above (80 ×
`1.67 = 133.3 ng/mL).
`
`Pharmacodynamics
`
`The pharmacodynamic findings in the present study
`reflected changes in the plasma level concentrations.
`Peak effects on all measures occurred at the time
`plasma level concentrations were maximal (Tmax), fol-
`lowed by a gradual degradation as the concentrations
`declined. Brown et al. (1978) reported a similar pattern
`with a slight delay in Tmax (as discussed above). How-
`ever, the statistical significance of the effect was elimi-
`nated after the time of peak concentration (Tmax) due
`to an improvement in the placebo control condition
`after 4 hours. Based on this, they concluded that re-
`sponse occurs only during the absorption phase. The
`results in the QD condition of the present study (and
`in a prior study by Swanson et al., 1998a) suggest that
`the PD effects of Adderall continue for about 6 hours
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`1239
`
`Page 6
`
`

`

`GREENHILL ET AL.
`
`after dosing, which is well beyond the absorption phase
`marked by the peak concentration at 2.5 hours.
`We did not include a placebo condition in the pres-
`ent study, so a direct comparison to the findings re-
`ported by Brown et al. is not possible. However, in the
`previous study using a similar LSP (Swanson et al.,
`1998a), we showed that in the placebo condition, our
`surrogate measures of efficacy tended to deteriorate
`across the day rather than improve, which would pre-
`dict the opposite adjustment and a longer rather than
`shorter duration of efficacy. Consistent with other re-
`cent studies of Adderall (Pelham et al., 1999; Swanson
`et al., 1998a), the data from our QD condition suggest
`that despite gradual deterioration in the efficacy mea-
`sures after the peak effects, measures of both perfor-
`mance (number of problems solved) and behavior
`(ratings of deportment and attention) continue to be
`improved relative to baseline.
`Our design was based on a comparison of two active
`conditions rather than to an inactive placebo condi-
`tion. In the ANOVA and planned comparison of the
`present study, the observed differences in efficacy were
`significant for a comparison of the BID and QD con-
`ditions, and the effect sizes for these differences were
`moderate (0.4 for the objective measures of academic
`productivity, number of problems solved) to large (0.8
`for the subjective outcome measures, rating of deport-
`ment in the classroom setting). Thus, the basic ques-
`tions addressed in this study, about the additional
`efficacy in the BID condition in the afternoon com-
`pared to the QD condition, do not depend on the use
`of a placebo control. However, the absence of a placebo
`control would make the interpretation of the results
`difficult if not impossible if there turned out to be no
`difference between the two treatments: was it because
`the treatments had not worked, or because they were
`equally effective?
`This proof-of-concept study’s twice-daily dosing
`data served as a model for the PK profile of a new
`formulation of Adderall, Adderall-XR, which was de-
`veloped for once-a-day administration. Adderall-XR is
`based on a double-pulse release from uncoated (first
`pulse) and coated (second pulse) Adderall beads. The
`PK and PD properties of Adderall-XR have been evalu-
`ated in the multisite proof-of-product study (Mc-
`Gough et al., 2003). The findings of that study were
`consistent with the findings of this study. This shows
`the value of early-stage small PK/PD dose-ranging
`studies, which can serve as the basis for the develop-
`ment of new pharmaceutical products.
`
`Hysteresis Analysis
`
`Another approach to the analysis of the PK and PD
`data is based on plotting concentration versus efficacy
`plots, where each successive time point is connected by
`an arrow to show the changing relationship between
`concentration over time as the end receptor’s response
`to the medication changes. If the same PK value (con-
`centration) always produces the same PD effect (num-
`ber of problems solved or rating on the SKAMP), then
`the points would fall on a 45-degree line. Deviations
`from this line over time may suggest different end-
`receptor responses to the same medication concentra-
`tion as time increases. This phenomenon, called
`hysteresis, may represent tolerance (decreased response
`to same dose of drug).
`The concentration–efficacy plots for the once-daily
`Adderall condition show a rapid increase in efficacy in
`the early morning, when the concentrations are increas-
`ing to the peak. However, after the peak concentration,
`when concentrations remain relatively constant, the
`magnitude of efficacy starts to decline, and the number
`of math problems solved and the subjective rating of
`the child’s deportment decline.
`One possible explanation for these effects would be
`the presence of tolerance. This was previously demon-
`strated for methylphenidate based on similar surrogate
`measures of efficacy (Swanson et al., 1999). However,
`without a placebo control and subjective reports from
`the patients, it is difficult to know if the reduced attention
`and deportment responses to the same plasma concentra-
`tions might be better explained by fatigue or boredom.
`
`Limitations
`
`The results of these pharmacokinetic data must be
`considered in light of the study’s limitations. First, the
`study has only 12 subjects, which severely limits the
`ability to address the apparent intersubject variability in
`PK and potential interactions with patient characteris-
`tics such as gender, ethnicity, or comorbid diagnosis. In
`particular, it would have been useful to determine if a
`comorbid diagnosis of conduct disorder might have
`moderated these effects. However, the use of the LSP
`and the crossover design provides enough power to
`make the simple comparisons needed to evaluate the
`PD differences between the QD and BID conditions in
`the afternoon and to relate these to the PK properties
`of Adderall. Second, the study was limited to a narrow
`range of ages of school-aged children, which severely
`limits the ability to address important age differences.
`
`1240
`
`J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 42:10, OCTOBER 2003
`
`Page 7
`
`

`

`Brown et al. (1978) reported that the apparent elimi-
`nation half-life of d-AMP is shorter in children with
`ADHD than adults, and it may be even shorter in
`young children (e.g., preschool children) than older
`children (elementary school children). The PK param-
`eters derived in this sample may not be applicable to
`adult populations or to preschool populations. Third,
`there was no placebo condition, so our interpretation
`of what might have occurred on placebo in the after-
`noon relies on a theoretical PD placebo profile derived
`from prior studies. This imposes a weakness in the
`design and interpretation of PK/PD relationships that
`may exist. For example, we would not have been able to
`interpret the findings if there had been no differences
`between the two conditions. Fourth, we do not know
`if the need for a greater amount of medication to main-
`tain concentration and control of behavior represents
`the natural onset of fatigue or boredom, or represents a
`habituation or tolerance at the receptor level. Fifth, our
`assessment of potential predictors of PK differences was
`limited. Further studies would have to be carried out to
`determine the effect of food (fasted or fed state) or
`gender on PK values. Fifth, this study was conducted in
`the artificial setting of the LSP. Confirmation of the
`basic findings (i.e., the predicted duration of the BID
`condition) will require effectiveness studies to be car-
`ried out in the ecologically valid context of the child’s
`own classroom.
`
`Clinical Implications
`
`Studies of PK and PD effects of Adderall in children
`with ADHD reveal differences between the once-daily
`and twice-daily administration schedule. This study
`provided data that confirmed other data gathered for
`the development of once-daily Adderall (Adderall XR),
`although they do not prove this twice-daily pattern of
`release, using equal amounts of drug, is necessarily the
`best pattern for a new once-daily, extended-release
`drug. The PD data shown here suggest that Adderall
`given twice daily is likely to be more effective in con-
`trolling behavior and improving performance over the
`course of an entire school day than when a single dose
`of 10 mg is given once in the morning. The degree of
`interindividual variability in derived PK variables un-
`derscores the need for individual titration of medica-
`tion in clinical practice.
`This PK/PD proof-of-concept study modeled one
`possible dual-pulse, equal dose release pattern for an
`extended-release drug. The actual extended-release
`product, Adderall XR, required