Atomoxetine Versus Placebo in Children and Adolescents with Attention-Deficit/Hyperactivity Disorder and Comorbid Oppositional Defiant Disorder: A Double-Blind,Randomized,Multicenter Trial in Germany

Abstract

Objectives:

The primary objective of this study was to evaluate the efficacy of atomoxetine (ATX, target dose 1.2 mg/kg daily) on symptoms of oppositional defiant disorder (ODD) in children and adolescents with attention-deficit/hyperactivity disorder (ADHD). A secondary objective was to compare fast versus slow up-titration of ATX.

Methods:

This was a 3-arm, 9-week, randomized, placebo-controlled, double-blind study in ADHD patients (6–17 years) with comorbid ODD (Diagnostic and Statistical Manual of Mental Disorders, 4th edition [DSM-IV] criteria A–C) or conduct disorder (CD). ATX-treatment arms were as follows—ATX-fast: 7 days 0.5 mg/kg, then 1.2 mg/kg; ATX-slow: 7 days each at 0.5 and 0.8 mg/kg, then 1.2 mg/kg. Primary outcome was the Swanson, Nolan, and Pelham Rating Scale-Revised (SNAP-IV) ODD-score after 9 weeks (Mixed Effects Model for Repeated Measures, ATX-up-titration groups pooled).

Results:

In total, 181 patients were randomized, and 180 evaluated (ATX-fast/ATX-slow/placebo: 60/61/59). Baseline characteristics were comparable (84.4% boys; mean age 11.0 years; DSM-IV: 100% ADHD, 75.6% with combined type, 74.4% ODD, 24.4% CD; SNAP-IV ODD-scores, mean ± standard deviation 15.5 ± 4.35). At week 9, SNAP-IV ODD scores were significantly lower versus placebo in both ATX-groups (least square mean [95% confidence interval]: ATX-fast 8.6 [7.2;9.9]; ATX-slow 9.0 [7.7;10.3]; placebo 12.0 [10.6;13.5]; least square mean, ATX-pooled minus placebo: −3.2 [−5.0, −1.5], effect size: −0.69, p < 0.001). SNAP-IV ADHD-scores, CD symptoms (investigator-rated Attention-Deficit and Disruptive Behavior Disorders Instrument, disruptive behavior), Clinical Global Impressions-Severity, and individual treatment behaviors showed corresponding results. Post-hoc analyses indicated interrelationships between the medication effects on ADHD, ODD, and CD symptom scores. For ATX-slow, time to early dropout was significantly longer versus placebo (Hazard Ratio [95% confidence interval]: 3.57 [1.42;8.94]; p = 0.007). Clinically relevant adverse effects (fatigue, sleep disorders, nausea, and gastrointestinal complaints; weeks 1–3) were reported in 60.0% of ATX-fast, 44.3% of ATX-slow, and 18.6% of placebo group patients.

Conclusions:

ATX for 9 weeks significantly reduced symptoms of ODD/CD and ADHD; slower ATX-up-titration may be better tolerated.

Introduction

A ttention-Deficit/Hyperactivity Disorder (ADHD) is frequently accompanied by additional psychiatric disorders such as mood disorders, anxiety disorders, learning disabilities, and other disruptive behavior disorders. Oppositional defiant disorder (ODD) is the most common comorbid psychiatric disorder in ADHD patients, affecting 40%–60% of children with ADHD (August et al. 1996; Turgay 2005; Biederman et al. 2007). According to Diagnostic and Statistical Manual of Mental Disorders, 4th edition (DSM-IV) criteria (American Psychiatric Association 2000), ODD is characterized by a pattern of developmentally inappropriate negativistic, hostile, and defiant behaviors causing clinically significant impairment in social, family, or academic functioning. Children and adolescents with ADHD and comorbid ODD tend to present with more severe ADHD symptoms, peer problems, and family distress, which in turn are associated with increased severity of the disorder and a poorer prognosis (Kuhne et al. 1997). According to DSM-IV, conduct disorder (CD) is considered a separate exclusionary disease entity, characterized by a more severe pattern of behaviors than ODD; a patient who meets DSM-IV criteria for CD is no longer declared as having ODD. Differently, in the International Classification of Diseases (ICD)-10 criteria set (WHO 2004), conduct disorders (F91) comprise ODD as a less severe form (F91.3).

Separate pathophysiological mechanisms and etiologies for ADHD and ODD/CD have been suggested based on genetic (Kirley et al. 2004), family (Satake et al. 2004), electrophysiological (Banaschewski et al. 2003), and psychometric (Oosterlaan et al. 2005; Luman et al. 2009) studies.

Pharmacologic treatment options for ADHD have been extensively and systematically studied. Psychostimulants are known to be effective in the treatment of ADHD, as reported in the large multimodal treatment study conducted by the MTA cooperative group (1999a). This study has also shown that psychostimulant treatment improved symptoms of both ADHD and oppositionality, independent of comorbid ODD/CD status (MTA cooperative group 1999b). Several other studies have shown that short- and long-acting methylphenidate (MPH) formulations are effective in ADHD patients with comorbid ODD or related aggressive symptomatology (Klein et al. 1997; Turgay 2005; Spencer et al. 2006; Sinzig et al. 2007; Gadow et al. 2008). In a meta-analysis including 28 studies, Connor et al. (2002) reported that psychostimulants exert effects on overt and covert aggression-related behaviors in ADHD with effect sizes similar to those for ADHD core symptoms; comorbid CD was associated with lower effect sizes for overt aggression.

In contrast, only few studies have prospectively addressed the effect of nonstimulant treatment with atomoxetine (ATX) on ODD/CD comorbidity, and with conflicting results. ATX is a centrally acting, norepinephrine reuptake inhibitor with little affinity for other transporters or neurotransmitter receptors (Bymaster et al. 2002). In the United States, ATX has been approved for ADHD treatment in children, adolescents, and adults since 2002, and in Germany for ADHD treatment in children and adolescents since 2004. Two double-blind, placebo-controlled studies had been specifically designed to evaluate the efficacy of ATX in treating ODD symptoms in children with ADHD and comorbid ODD/CD, as measured by the ODD subscale of the investigator-rated Swanson, Nolan, and Pelham Rating Scale-Revised (SNAP-IV) (Swanson et al. 2001). In one study, ATX was superior in reducing ODD symptoms over time, significantly better at week 2 and week 5, but no longer at week 8 (Bangs et al. 2008). Thus, it remained uncertain whether ATX exerts a specific and enduring effect on ODD symptoms. Of note, the patient population in this study was restricted to patients with severe ADHD and ODD symptoms and poor Clinical Global Impressions (CGI) ratings. The second study evaluated the efficacy of ATX in improving ADHD and ODD symptoms in pediatric patients with ADHD and comorbid ODD who were nonresponders to a previous parent training intervention (Dell'Agnello et al. 2009). In this 8-week study, both ADHD and ODD symptoms had significantly improved compared with placebo at the end of the 8-week double-blind treatment period.

In addition, post-hoc analyses have also been conducted on data of the placebo-controlled core registration studies of ATX conducted in children and adolescents with ADHD (Michelson et al. 2001, 2002; Spencer et al. 2001, 2002; Kelsey et al. 2004). In these studies, comorbid ODD was typically present in 30%–40% of patients, and ODD symptoms had been measured by the oppositional subscale of the Conners' Parent Rating Scale-Revised Short Form (CPRS-R:S). These post-hoc analyses had also given inconclusive results: A first subset analysis of 98 children with ADHD and comorbid ODD from two early placebo-controlled trials revealed significant improvement of ADHD symptoms, but no significant reduction of ODD symptoms versus placebo (Kaplan et al. 2004). A second, more detailed subgroup analysis based on a larger placebo-controlled ATX study in 226 children and adolescents with ADHD with and without comorbid ODD (Michelson et al. 2002) concluded that ATX significantly reduced both ADHD and ODD symptoms and improved quality-of-life measures compared with placebo (Newcorn et al. 2005). Data also suggested that the comorbid group may require higher ATX doses (1.8 mg/kg per day).

In an additional post-hoc meta-analysis, acute-phase data were analyzed from three double-blind placebo-controlled ATX studies in 512 children and adolescents with ADHD (age 6–16 years); 158 had comorbid ODD. ATX significantly reduced ADHD symptoms in patients with and without ODD to a similar extent. However, the reduction of ODD symptoms in patients with comorbid ODD was not significantly higher than with placebo (Biederman et al. 2007). Finally, Cheng et al. (2007) performed a systematic meta-analysis on nine placebo-controlled ATX trials in children and adolescents (ATX 1,150, placebo 678 patients). In the smaller subgroup of comorbid ADHD/ODD patients, both ADHD and ODD symptoms were significantly reduced with ATX compared with placebo. Comorbid ODD status was significantly associated with smaller treatment-related reductions in ADHD symptoms. Cheng et al. concluded that ATX may have a role in treating comorbid ODD.

In the light of these inconclusive results, additional prospective studies evaluating the efficacy and safety of ATX in pediatric patients with comorbid ADHD and ODD/CD were suggested, in particular in patients with less severe symptoms compared to the Bangs et al. (2008) study. The primary objective of this double-blind, placebo-controlled study was to assess the efficacy of ATX in treating symptoms of ODD in children and adolescents with ADHD and comorbid ODD: all patients were to meet both DSM-IV criteria (American Psychiatric Association 2000) of ADHD and DSM-IV criteria A–C of ODD. The presence of DSM-IV diagnostic criteria for CD was not exclusionary, but there was no inclusionary minimum severity score on a structured ODD symptom scale. Applying these criteria, it was intended to enroll a majority of patients suffering from moderate rather than high severity of ODD/CD symptoms.

An important secondary objective of this study was to compare the tolerability of ATX when applying two different up-titration schemes in this pediatric population. In early clinical practice and based on the German Summary of Product Characteristics (SPC), ATX treatment was commonly up-titrated from 0.5 to 1.2 mg/kg per day in one single step. Based on subsequent clinical observation, it was hypothesized that a slower, more careful approach, such as including an additional up-titration step at 0.8 mg/kg per day, might improve tolerability.

Methods

Participants

Patients were aged 6 to 17 years and had to meet DSM-IV, Text Revision (DSM-IV-TR) (American Psychiatric Association 2000) criteria for ADHD (any subtype), and DSM-IV-TR criteria A–C of ODD. The presence of DSM-IV-TR diagnostic criteria for CD was not exclusionary. Diagnoses were validated using the “Diagnose-Checkliste Hyperkinetische Störungen” (Diagnostic Checklist for Hyperkinetic Disorders), and the “Diagnose-Checkliste Störungen des Sozialverhaltens” (Diagnostic Checklist for ODD/CD, Diagnostic Checklist for SSV) as part of the DISYPS-KJ standard instrument (Döpfner and Lehmkuhl 2000). Based on patient and parent interview, behavioral observation, plus patient, parent, and teacher questionnaires, it allows the investigator to derive diagnoses according to both the DSM-IV and the ICD-10 criteria (algorithms). The use of this German language instrument for diagnosing ADHD and ODD/CD is common clinical practice in Germany.

Patients who had a history of bipolar I or II disorder, psychosis, pervasive developmental disorder, or seizure disorder (other than febrile seizures) were excluded. Patients were excluded if they were at serious suicidal risk, as determined by the investigator, or if they were likely to require psychotropic medications other than study drug or a structured psychotherapy. Psychotherapy initiated before study participation was acceptable.

Patients were recruited from November 2006 until November 2008 at 20 child and adolescent psychiatric and pediatric practices and hospitals throughout Germany; only outpatients were enrolled. Written informed consent was obtained from all legal representatives (for example, both parents). In addition, written informed consent or assent was obtained from the child/adolescent. The study was approved by a central ethics review board and conducted in accordance with the ethical standards of the Declaration of Helsinki.

Efficacy and safety measures

The primary efficacy measure was the investigator-rated SNAP-IV scale. The SNAP IV is 26-item rating scale, which includes 1 item for each of the 18 symptoms contained in the DSM-IV diagnosis of ADHD and 1 item for each of the 8 symptoms contained in the DSM-IV diagnosis of ODD (Swanson et al. 2001). The intensity of each item during the last 7 days is scored on a 0-to-3 scale (0 = “not at all,” 1 = “just a little,” 2 = “pretty much,” 3 = “very much”). The SNAP-IV scale has been validated and normed in a sample of school-aged children from the United States (Swanson 1992; Gaub and Carlson 1997). The SNAP-IV ODD score (sum of items 19–26) was the primary efficacy outcome; the SNAP-IV ADHD score (sum of items 1–18) was considered a co-primary outcome. The SNAP-IV ADHD domain scores for inattention (sum of items 1–9) and hyperactivity/impulsivity (items 10–18) were regarded as secondary outcomes.

All SNAP-IV items relating to disruptive behavior refer to the DSM-IV criteria of ODD; there was no standardized investigator-rated instrument available that addresses the DSM-IV criteria for CD/disruptive behavior in addition to ADHD and ODD criteria. Therefore, we derived a new instrument as secondary outcome measure, the investigator-rated Attention-Deficit and Disruptive Behavior Disorders Instrument (ADDB-Inv), which included the same items as SNAP-IV, plus 1 item each for the following 7 less severe criteria selected from the 15 primary criteria for the diagnosis of CD according to DSM-IV (American Psychiatric Association 2000)—A1: often bullies, threatens, or initimidates others; A2: often initiates physical fights; A5: has been physically cruel to animals; A11: often lies to obtain goods or favors or to avoid obligations; A12: has stolen items of nontrivial value without confronting victim; A13: often stays out at night despite parental prohibition, beginning before age 13 years; A15: is often truant from school, beginning before age 13 years. For each ADDB-Inv item, investigators rated symptom severity during the last 7 days using the same scoring as in SNAP-IV. In addition, investigators rated the symptom-related impairment during the last 7 days for each item on the same 0 to 3 scale. The following ADDB-Inv scores were calculated (separately for severity and impairment): ADDB-Inv ODD (sum of SNAP-IV-based items 19–26), ADDB-Inv ADHD (sum of SNAP-IV based items 1–18), and ADDB-Inv disruptive behavior reflecting the CD symptoms (sum of the 7 DSM-IV based items relating to CD: A1, A2, A5, A11, A12, A13, A15).

Further, the course of three individually defined key behavior problems (individual target behaviors [ITBs]) was assessed using an investigator-scored instrument (ITB-Inv), which was adopted from the target behavior assessment based on the German Treatment Program for Hyperkinetic and Oppositional Problem Behavior (Döpfner et al. 1998, 2004). At baseline, the investigator defined three ITBs to be considered the main treatment targets for each individual patient, based upon parent/primary caregiver interviews and any additional information available to him. Those specific behaviors considered as most impairing for the child or most stressful for the parents were chosen. The investigator rated the frequency of each ITB during the last 7 days on a 6-point rating scale (from 0 = never, to 5 = always), and the intensity of each ITB on a 10-point rating scale (from 0 = no problems, to 9 = most severe problems) at baseline and the following study visits. Separate scores (sum of ratings for target behaviors 1–3) were calculated for intensity and frequency of these target behaviors.

Finally, the CGI-Severity (CGI-S; Guy 1976; National Institute of Mental Health 1985) was used in this study to measure the severity of the patient's ADHD, ODD, and overall ADHD + ODD symptoms (CGI-S ADHD, CGI-S ODD, and CGI-S ADHD + ODD).

Safety and tolerability were assessed by open-ended questioning for adverse events (AEs) and collection of vital signs and body weight. AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA), Version 9.1. Following a blinded database review of all MedDRA preferred terms reported in the study, four categories of clinically relevant AEs were predefined in the statistical analysis plan. Clinical relevance was defined based on AEs associated with ATX treatment in previous studies (Spencer et al. 1998, 2001, 2002; Michelson et al. 2001, 2002; Kelsey et al. 2004). These predefined clinically relevant AEs included the following: (1) fatigue or related symptoms (included fatigue, somnolence, and sedation), (2) nausea or related symptoms (included vomiting, nausea, anorexia, and decreased appetite), (3) gastrointestinal (GI) complaints (included upper abdominal pain, abdominal pain, diarrhea, and constipation), and (4) sleep disturbance (included insomnia, sleep disorder, early morning awakening, middle insomnia, and poor quality sleep).

Study design

This was a randomized, double-blind, placebo-controlled, three-arm, parallel group, multicenter study (B4Z-SB-LYDW). After a 3- to 28-day screening and washout period, eligible patients were randomized to 9-week double-blind treatment with (1) ATX 0.5 mg/kg per day for 7 days, followed by the target dose of 1.2 mg/kg (ATX-fast titration group); (2) ATX 0.5 mg/kg once daily for 7 days, followed by 0.8 mg/kg for 7 days, followed by target dose of 1.2 mg/kg (ATX-slow titration group); or (3) placebo. Study medication was given once daily in the morning. Patients who were unable to tolerate the target dose were discontinued from the study; down-titration was not allowed. The concomitant use of other psychotropic medications was not allowed. Randomization was based on a computer-generated random sequence using an interactive voice response system and was stratified by patients' age (<12 vs. ≥12 years). Efficacy and safety assessments were performed at weeks 0 (baseline), 1, 2, 3, 5, 7, and 9.

Study objectives

The primary objective was to test the hypothesis that ATX, given once daily for 9 weeks (target dose of 1.2 mg/kg per day), using either fast or slow titration, was superior to placebo in the outpatient treatment of ODD symptoms in children and adolescents with ADHD and comorbid ODD. ODD symptoms were primarily assessed using the SNAP-IV ODD score. Co-primary and secondary objectives were to evaluate (1) the effect of ATX on ADHD, ODD, and CD symptoms as measured by SNAP-IV ADHD, ADDB-Inv, ITB-Inv, and CGI-S scores; (2) time to early dropout; (3) safety and tolerability of ATX fast and slow titration; and (4) the effects of ATX on health-related quality of life and family burden (to be published elsewhere).

Statistical methods

The sample size calculation was based on data from the previous study by Bangs et al. (2008): Assuming standard deviations (SDs) for the change in SNAP-IV ODD score from baseline of 5.31 with ATX and 4.29 with placebo, a sample size of 144 evaluable patients was needed to demonstrate an effect size of 0.5 on the SNAP-IV ODD scale at the 5% significance level with a power of 80%. An effect size of 0.5 is above the minimal clinically relevant difference. Assuming a 20% dropout rate, 180 patients were planned to be enrolled.

ODD symptoms were measured primarily using the SNAP-IV ODD score. For the primary efficacy analysis, the pooled ATX data (fast and slow titration arm) were tested versus placebo. A Mixed Effects Model for Repeated Measures (MMRM) was used, including treatment, visit, age strata, baseline SNAP-IV ODD score, baseline-by-visit interaction, and treatment-by-visit interaction as fixed effects and assuming an unstructured covariance matrix during the 9-week treatment period. Week-9 estimates were compared. Corresponding MMRM analyses were applied for co-primary and secondary outcomes. Further MMRM analyses included additional covariates as fixed effects (such as gender, age, and pretreatment with psychostimulants) to explore the robustness of primary results. In case of convergence problems, an autoregressive covariance matrix (AR[1]) was used instead of the unstructured type. A supportive analysis of covariance (ANCOVA) was conducted comparing the SNAP-IV ODD scores at the end of study (last observation carried forward [LOCF] analysis), using fixed effects for baseline, age strata, and treatment.

Further, path analyses were used post-hoc to separate the total effects of ATX treatment on physician-rated detailed scores (rating the dimensions ADHD, ODD, and CD using SNAP-IV and ADDB) and physician-rated global scores (rating ADHD and ODD using CGI-S) each into direct and indirect effects (Bollen 1989). The path analyses were based on ANCOVA modeling using change in the ADHD or ODD or CD symptoms at endpoint (LOCF) as dependent variable. The ANCOVAs included treatment, age, respective baseline score, and, if needed, changes from baseline to study end (LOCF) as fixed effects. The direct and indirect effects of treatment on outcome were all obtained in units of the respective endpoint and as percentages of the overall effect. The direct effect of the respective symptom dimension (e.g., ADHD) was defined as the treatment effect remaining after controlling for changes in the other dimensions (e.g., ODD/CD). To further assess the correlation between the dimensions, Pearson correlation coefficients with 95% confidence interval (CI) were calculated post-hoc for the baseline values as well as for the changes to endpoint (LOCF).

In addition, continuous variables (LOCF) at week 9 were evaluated by ANCOVA and logistic regression post-hoc to assess whether there was an interaction between pretreatment (treatment with psychostimulants: yes vs. no) and study treatment, using fixed main effects for treatment, pretreatment, age, and baseline score, and, additionally, the pretreatment by treatment interaction. Parameter estimates together with p values and CIs were calculated as well as effect sizes for treatment by subgroup.

Time to early dropout was displayed using Kaplan-Meier plots. Tests to compare treatment groups were based on the Cox proportional hazards model with supportive Log-rank and Wilcoxon tests. Adverse event rates were evaluated descriptively for each treatment arm. Derived incidence rates for clinically relevant categories of AEs related to study drug were compared post-hoc using Fisher's exact test separately for the initial 3 weeks of treatment and for the complete 9-week treatment period.

All efficacy and safety analyses were based on the full analysis population, including all patients randomized who received at least one dose of study drug, according to the intent-to-treat principle. The primary analysis was additionally performed based on a per protocol population, excluding all patients with major protocol violations regarding the diagnosis of ADHD and ODD, assigned treatment, titration and dose, patient compliance, prior and concomitant medications, and maintenance of blinding.

Results

Patient disposition

Twenty investigational sites in Germany participated in this study; they enrolled between 1 and 23 patients per site in the enrollment period from November 2006 to November 2008. All patients were studied in the same trial design.

Of 186 patients initially assessed, 181 patients were randomly assigned to treatment and 180 patients received at least one dose of study drug (ATX-fast 60, ATX-slow 61, and placebo 59; Full Analysis Population). Of the latter, 129 (71.7%) patients completed the study (Fig. 1). The most common reason for early dropout was lack of efficacy (15.6% overall), followed by AEs (5.0%) and parent/caregiver decision (5.0%). The dropout rate was highest in the placebo group (37.3%), caused by the high rate of dropouts due to lack of efficacy (n = 17, 28.8%), compared to 24.0% for any reason with ATX treatment (with ATX-fast: 26.7%, and ATX-slow: 21.3%), most commonly due to lack of efficacy (ATX-pooled: 9.1%), AE (ATX-pooled: 6.6%), and parent/caregiver decision (ATX-pooled: 6.6%).

FIG. 1.

Patient disposition. ATX = atomoxetine; ATX-fast = atomoxetine fast titration group: 0.5 mg/kg per day for 7 days, then increase to the target dose of 1.2 mg/kg per day; ATX-slow = atomoxetine slow titration group: 0.5 mg/kg per day for 7 days, followed by 0.8 mg/kg for 7 days, then increase to the target dose of 1.2 mg/kg per day.

Major protocol violations leading to the exclusion of patients from the Per Protocol Population occurred in seven patients only (three ATX-fast, two ATX-slow, and two placebo).

Baseline characteristics

The three treatment groups were similar in terms of baseline characteristics and baseline disease severity (Table 1; full analysis set, N = 180). All patients met DSM-IV criteria of ADHD, the combined subtype was most frequent (n = 136, 75.6%), followed by the predominantly inattentive (n = 35, 19.4%) and hyperactive/impulsive (n = 9, 5.0%) subtypes. All but two patients had a DSM-IV comorbid diagnosis of either ODD (n = 134, 74.4%) or CD (n = 44, 24.4%). Mean CGI-S ratings showed patients were on average rated as “markedly ill,” with mean overall SNAP-IV ADHD scores (±SD) of 37.3 (±9.28) and mean overall SNAP-IV ODD scores of 15.5 (±4.35). Overall, 44.4% of patients (n = 80) had been pretreated with stimulant medication (mainly MPH); of those, most patients switched treatment due to inadequate response (n = 56, 70.0%), AE (n = 19, 23.8%), noncompliance (n = 8, 10.0%), and patient decision (n = 8, 10.0%; multiple responses possible). No patient received concomitant psychostimulant or other psychotropic medication.

Table 1.

Baseline Characteristics

	ATX-fast, N = 60	ATX-slow, N = 61	ATX-pooled, N = 121	Placebo, N = 59
Demographics
Age (years), mean (SD)	11.1 (2.9)	10.8 (3.4)	10.9 (3.1)	11.1 (2.8)
Age ≥12 years, n (%)	21 (35.0)	21 (34.4)	42 (34.7)	21 (35.6)
Gender: male, n (%)	51 (85.0)	53 (86.9)	104 (86.0)	48 (81.4)
BMI (mg/m²), mean (SD)	18.7 (3.1)	19.9 (4.3)	19.3 (3.8)	18.8 (3.1)
ADHD subtype, n (%)
Combined	44 (73.3)	45 (73.8)	89 (73.6)	47 (79.7)
Predominantly inattentive	13 (21.7)	14 (23.0)	27 (22.3)	8 (13.6)
Predominantly hyperactive-impulsive	3 (5.0)	2 (3.3)	5 (4.1)	4 (6.8)
Diagnoses related to ODD/CD, n (%)
ODD	44 (73.3)	45 (73.8)	89 (73.6)	45 (76.3)
CD	14 (23.3)	16 (26.2)	30 (24.8)	14 (23.7)
Disruptive behavior disorder not otherwise specified	1 (1.7)	0 (0.0)	1 (0.8)	0 (0.0)
Adjustment disorder with mixed disturbance of emotions and conduct	1 (1.7)	0 (0.0)	1 (0.8)	0 (0.0)
Previous treatment
Previous stimulant exposure, n (%)	23 (38.3)	28 (45.9)	51 (42.1)	29 (49.2)
Disease severity, mean score (SD)
SNAP-IV ADHD	37.6 (9,70)	38.0 (8.91)	37.8 (9.27)	36.4 (9.31)
SNAP-IV ADHD, Inattention	17.7 (5.23)	18.3 (4.91)	18.0 (15.06)	17.5 (4.94)
SNAP-IV ADHD, Hyperactivity/Impulsivity	19.9 (4.98)	19.7 (4.76)	19.8 (4.85)	18.9 (4.85)
SNAP-IV ODD	15.5 (4.07)	15.6 (3.82)	15.5 (3.93	15.6 (5.13)
CGI-S ADHD	5.0 (0.77)	5.2 (0.79)	5.1 (0.78)	5.1 (0.74)
CGI-S ODD	4.9 (0.92)	5.3 (0.83)	5.1 (0.89)	5.0 (0.86)

Data presented for Full Analysis Population.

ATX = atomoxetine; ATX-fast = atomoxetine fast titration group; ATX-slow = atomoxetine slow titration group; SD = standard deviation; BMI = body mass index; ADHD = attention-deficit/hyperactivity disorder; ODD = oppositional defiant disorder; CD = conduct disorder; CGI-S = Clinical Global Impressions, Severity; SNAP-IV = Swanson, Nolan, and Pelham Rating Scale-Revised (SNAP-IV) scale.

Demographics showed the expected distribution for a population of childhood and adolescent ADHD patients, with 84.4% being male, and a mean age (±SD) of 11 (± 3) years (range 6–17.9 years; 65.0% <12 years). Almost half of the patients (n = 85, 47.2%) were attending elementary school (Grundschule) at baseline. The proportion of patients visiting higher-level secondary schools (Realschule, Gymnasium) was 18.9%; <3% of patients attended the highest level school type (Gymnasium; n = 5, 2.8%). A considerable proportion of patients (22.8% overall) lived with a single mother.

Primary efficacy outcome: severity of ODD symptoms

The primary efficacy analysis (MMRM) revealed that 9-week treatment with ATX once daily, using either fast or slow titration up to a target dose of 1.2 mg/kg per day, was significantly superior to placebo in reducing ODD symptoms as measured by the SNAP-IV ODD score (Fig. 2; least square [LS] mean treatment group difference at week 9, ATX-pooled minus placebo [95% CI]: −3.2 [−5.0 to −1.5], effect size: −0.69, p < 0.001).

FIG. 2.

Symptoms of ODD (SNAP-IV ODD), ADHD (SNAP-IV ADHD), and conduct disorder (ADDB-Inv, Disruptive Behavior) over time and MMRM analyses at week 9. Graphs show LS means and 95% confidence intervals, and the number of patients per group with data available at baseline and week 9, respectively. PBO = Placebo; ADHD = attention-deficit/hyperactivity disorder; ODD = oppositional defiant disorder; ADDB-Inv = Investigator-rated Attention-Deficit and Disruptive Behavior Disorders Instrument; ADDB-Inv Disruptive Behavior = ADDB-Inv subscale rating symptoms of disruptive behavior; LS mean = least square mean; MMRM = Mixed Model Repeated Measures Analysis; SNAP-IV = Swanson, Nolan and Pelham Rating Scale-Revised; SNAP-IV ADHD = SNAP-IV subscale rating ADHD symptoms; SNAP-IV ODD = SNAP-IV subscale rating ODD symptoms.

Looking at the fast and slow titration groups separately, the decrease in ODD symptom severity was significant for both individual titration groups as well (Fig. 2; LS mean scores at week 9 [95% CI]: ATX-fast 8.6 [7.2, 9.9]; ATX-slow 9.0 [7.7, 10.3]; placebo 12.0 [10.6, 13.5]; MMRM versus placebo: p < 0.001, effect size: −0.74 and p = 0.003, effect size: −0.65). The direct comparison of the ATX-fast versus ATX-slow titration groups was not significant (p = 0.669, effect size: −0.09). The corresponding MMRM analysis based on the per protocol population, as well as modified MMRM analyses including additional covariates (such as gender, age, or pretreatment with psychostimulants), and the supportive ANCOVA analysis comparing LOCF rather than week 9 estimates, yielded corresponding results (data not shown).

An improvement in SNAP-IV ODD scores of at least 30% (at least 50%) was achieved by 48.3% (35.0%) of patients in the ATX-fast titration group, compared with 55.7% (47.5%) in the ATX-slow titration group, and 35.6% (16.9%) in the placebo group (Fig. 3).

FIG. 3.

Proportion of patients with decreased SNAP-IV ODD subscale scores at the end of study (last observation carried forward). Patients with increased SNAP-IV ODD scales not shown. SNAP-IV = Swanson, Nolan, and Pelham Rating Scale-Revised; ODD = oppositional defiant disorder.

ANCOVA modeling of the SNAP-IV ODD subscore showed that a one-point increase at baseline lead to an 0.4-point increase at endpoint (p < 0.001 for the regression coefficient of the slope). However, no significant interaction with treatment was found for the baseline value.

Severity of ADHD symptoms and disruptive behavior

ATX was also significantly superior to placebo in reducing the severity of ADHD symptoms as measured by the SNAP-IV ADHD scores (Fig. 2, co-primary analysis, LS mean treatment group difference at week 9, ATX-pooled minus placebo [95% CI]: −7.4 [−11.0, −3.8], effect size −0.72, p < 0.001). Again, each individual titration group was significantly superior to placebo (Fig. 2; LS mean scores at week 9 [95% CI]: ATX-fast 22.9 [20.1, 25.8]; ATX-slow 21.3 [18.5, 24.1]; placebo 29.6 [26.6, 32.5]; p = 0.002 and p < 0.001, respectively). The direct comparison of ATX-fast versus slow titration was not significant (p = 0.416). The respective SNAP-IV ADHD domain scores reflecting the severity of inattention and hyperactivity/impulsivity symptoms improved correspondingly (data not shown).

Finally, the score reflecting the severity of patients' CD symptoms (ADDB-Inv disruptive behavior) was also significantly reduced after 9 weeks of ATX treatment (Fig. 2; LS mean treatment group difference at week 9, ATX-pooled minus placebo [95% CI]: −1.4 [−2.1, −0.7], effect size −0.62, p < 0.001). Each titration group was significantly superior to placebo as well (Fig. 2; p < 0.001 and p = 0.002, respectively), whereas the direct comparison of ATX-fast versus slow titration groups was not significant (p = 0.607).

The ADDB-Inv scores reflecting the patients' impairment by ODD symptoms, ADHD symptoms, and disruptive behavior all showed a corresponding pattern of improvement (data not shown).

Individual target behaviors

At baseline, three ITBs were defined based on parent/caregiver information, to be considered as main treatment targets for each individual patient. After 9 weeks of treatment, both intensity and frequency of these ITBs were significantly lower with ATX than placebo (Fig. 4; LS mean treatment group difference at week 9, ATX-pooled minus placebo [95% CI]: for ITB intensity: −3.5 [−6.2, −0.9], effect size −0.52, p = 0.01; for ITB frequency: −1.8 [−3.2, −0.4], effect size −0.53, p = 0.01).

FIG. 4.

Individual target behaviors over time and MMRM analyses at week 9. Graphs show LS means and 95% confidence intervals, and the number of patients per group with data available at baseline and week 9, respectively. Intensity and frequency of three ITBs, as identified by the patients' parents, were rated by physicians from 0 to 9 (intensity), or 0 to 5 (frequency), individual intensity and frequency scores were added up. ITB = individual target behaviors; ATX = atomoxetine.

CGI-severity

For CGI-S, ATX was significantly superior to placebo after 9 weeks of treatment (LS mean treatment group difference at week 9, ATX-pooled minus placebo [95% CI]: for CGI-ODD: −0.8 [−1.1, −0.4], effect size −0.22, p < 0.001; for CGI-ADHD: −0.7 [−1.1, −0.4], effect size −0.21, p < 0.001; corresponding results for the CGI-S ODD + ADHD). Each titration group was significantly superior to placebo (all p < 0.01), whereas the direct comparison of ATX-fast versus slow titration group was not significant (all p > 0.4).

Dependency of treatment effects on ADHD, ODD, and CD symptoms (path analysis)

Path analysis can be used to explain treatment effects based on theoretical models about the relationship of different symptoms. First, it was evaluated if treatment effects on ODD symptoms were influenced through treatment effect on ADHD and/or CD symptoms. In this case, the path-analyses based on SNAP-IV and ADDB ratings revealed nonadditive effects. This methodological approach implied a negative direct effect of ATX on ODD (adjusted difference versus placebo at endpoint: −0.282). Consequently, the sum of the indirect effects (ADHD: 2.204, CD: 0.896) was higher than the overall effect.

Assuming that treatment effects on ADHD symptoms were influenced by effects on ODD and CD symptoms, a path analyses showed a direct effect of ATX on ADHD symptoms of 3.268 points (43% of the overall effect). The pathways via ODD and CD symptoms showed indirect effects on ADHD symptoms of 4.165 (54.8%) and of 0.174 (2.3%), respectively. Alternatively, it might be assumed that treatment effects on ADHD and ODD symptoms influence the treatment effect on CD symptoms. In this case, ATX had a direct effect of 0.346 (30.3% of the overall effect) on CD symptoms. The indirect effects via ADHD and ODD were 0.074 (6.5%) and 0.721 (63.2%) on CD symptoms, respectively.

ADHD and ODD symptoms were also assessed using the CGI-S. Path analysis based on these scores revealed a direct ATX treatment effect on ODD symptoms of 0.042 (6.4%) and an indirect ATX treatment effect on ODD symptoms via the effect on ADHD symptoms of 0.606 (93.6%). Using an alternative model, the direct effect of ATX on ADHD symptoms was 0.215 (32.3%) and the indirect effect on ADHD symptoms via the effect on ODD symptoms was 0.451 (67.7%).

The correlations between the ratings of the symptoms underlying the different path analyses revealed moderate to large correlations depending on the instruments used as well as on the time points assessed (Table 2).

Table 2.

Correlation Analysis of Baseline Values and Changes from Baseline (Last Observation Carried Forward) to Physician-Rated Scores Pearson's Correlation Coefficients with 95% Confidence Intervals

	ATX-pooled, N = 121		Placebo, N = 59
	Baseline	Change to LOCF	Baseline	Change to LOCF
ADDB-Inv Disruptive Behavior Subscore vs. SNAP-IV ODD	0.58 [0.44, 0.68]	0.58 [0.45, 0.69]	0.65 [0.47, 0.78]	0.66 [0.49, 0.78]
ADDB-Inv Disruptive Behavior Subscore vs. SNAP-IV ADHD	0.48 [0.33, 0.61]	0.49 [0.34, 0.62]	0.53 [0.31, 0.69]	0.52 [0.31, 0.69]
SNAP-IV ADHD vs. SNAP-IV ODD	0.52 [0.38, 0.64]	0.71 [0.61, 0.79]	0.73 [0.58, 0.83]	0.61 [0.42, 0.75]
CGI-S ADHD vs. CGI-S ODD	0.44 [0.29, 0.58]	0.72 [0.62, 0.80]	0.56 [0.36, 0.71]	0.74 [0.60, 0.84]

Data presented for Full Analysis Population.

LOCF = last observation carried forward; ADDB-Inv = Investigator-rated Attention-Deficit and Disruptive Behavior Disorders Instrument; ADDB-Inv Disruptive Behavior = ADDB-Inv subscale rating symptoms of disruptive behavior; SNAP-IV ADHD = SNAP-IV subscale rating ADHD symptoms; SNAP-IV ODD = SNAP-IV subscale rating ODD symptoms.

Effect of pretreatment on efficacy and tolerability

The analysis evaluating the interaction between pretreatment status (treatment with psychostimulants: yes/no) and outcome at week 9 revealed that only for the SNAP-IV ODD score was there a significant interaction (p = 0.032) between pretreatment status and study treatment (ATX vs. PBO). The treatment effect was 3.6 points [95% CI 0.3, 6.9] higher for the pretreated patients (effect size: 0.860) than for the nonpretreated patients (effect size: 0.165). Analyses with the other scales revealed no significant interaction with pretreatment and study treatment. Logistic regression revealed no significant interaction for ADHD or ODD response.

Time to early dropout

An analysis of the time to early dropout from the study, based on the Cox proportional hazards model, showed that patients in the ATX-slow titration group stayed on treatment significantly longer than patients on placebo (hazard ratio or HR [95% CI]: 3.57 [1.42, 8.94], p = 0.007). The ATX-fast titration group did not differ from placebo in this analysis (HR 1.57 [0.78, 3.19], p = 0.208). However, the direct comparison between fast and slow titration groups did not reach statistical significance (HR 2.24 [0.85, 5.89], p = 0.103). Log-rank and Wilcoxon tests supported these results (data not shown). A subgroup analysis revealed corresponding results for children aged <12 years (Fig. 5), whereas no notable treatment group differences were seen in the subgroup of adolescent patients, probably due to the smaller sample size (35.0% aged ≥12 years).

FIG. 5.

Time to early dropout from the study, in pediatric ADHD patients (Kaplan-Meier plot): (a) For children <12 years and (b) for adolescents ≥12 years. HR = hazard ratio; 95% CI = 95% confidence interval; ATX = atomoxetine.

Tolerability

The most commonly reported treatment-emergent AEs during ATX treatment were fatigue (ATX-fast/slow 35.0%/21.3%; vs. placebo 10.2%), nausea (21.7/19.7% vs. 5.1%), headache (25.0/14.8% vs. 15.3%), vomiting (15.0/18.0% vs. 5.1%), upper abdominal pain (15.0/13.1% vs. 0.0%), and anorexia (15.0/11.5% vs. 1.7%). Any treatment-related AEs were reported for 70.0% of patients in the fast titration group, 57.4% of patients in the slow titration group, and 30.5% of patients in the placebo group. Figure 6 compares the rates of clinically relevant adverse effects, predefined as treatment-related AEs relating to fatigue, nausea, GI complaints, or sleep disturbances. In both titration groups, the rates of such events were significantly higher than with placebo (60.0%/44.3% vs. 18.6% during the initial 3 weeks of treatment, p < 0.001 and 0.003, respectively). However, the difference between the fast and slow-titration groups did not reach statistical significance (p = 0.102). In the ATX-fast titration group, fatigue or related symptoms, GI complaints, and nausea or related symptoms were all significantly more frequent than with placebo (each p < 0.01), whereas in the ATX-slow titration group, the statistical test versus placebo reached statistical significance only for nausea or related symptoms (Fig. 6). No patient died during the study. Serious AEs were reported for three patients (one each per treatment group). They were considered as treatment-related in one case (ATX-fast titration; severe stomach cramps and abdominal pain). The two other cases (1 ATX-slow titration: severe upper abdominal pain and chest pain for 1 day; one placebo: mild chest pain after sports) were not related to study drug.

FIG. 6.

Proportion of patients with clinically relevant adverse drug reactions. ATX = atomoxetine.

Early dropout rates due to AEs were low—there were eight patients who discontinued ATX (6.6%), six in the fast and two in the slow titration group (placebo one patient). Three cases were related to nausea or vomiting, one to aggression, one to fatigue, one to headache, and one to tachycardia. One patient of the ATX-slow titration group discontinued due to a nonserious event of suicidal ideation (moderate severity, 3 weeks after start of ATX). No other AEs relating to suicidal ideation were reported. Of the six ATX-fast titration patients who stopped ATX early due to AEs, two discontinued during the initial dose titration period, the other four patients discontinued after 20–42 days of treatment.

The analysis evaluating the interaction between pretreatment status (treatment with psychostimulants: yes/no) and treatment emergent AEs up to week 9 revealed no significant interactions for the grouped preferred terms of fatigue and related symptoms, nausea and related symptoms, GI complaints, and sleep disturbance. Also for the single preferred terms fatigue, nausea, headache, vomiting, upper abdominal pain, and anorexia, no significant interactions were found.

Discussion

The primary repeated-measures analysis (MMRM) demonstrated that 9-week treatment with ATX once daily, using either fast or slow titration up to a target dose of 1.2 mg/kg per day, was significantly superior to placebo in reducing ODD symptoms as measured by the investigator-rated SNAP-IV ODD subscale score. The direct comparison of the ATX-fast versus ATX-slow titration groups showed no significant result. ATX was also significantly superior to placebo in reducing ADHD symptoms as measured by the SNAP-IV ADHD subscale score (co-primary outcome). Several supportive MMRM and ANCOVA (LOCF) analyses all yielded corresponding results, thus illustrating the robustness of our findings. In alignment with the primary outcome measure, all investigator-rated instruments used in this study showed a consistent pattern of a significant improvement in the pooled ATX group versus placebo at week 9, with no differences between ATX-fast and ATX-slow up-titration groups. This included the ratings for all SNAP-IV ADHD subscale scores, all CGI-severity ratings, and the severity of and impairment by CD symptoms. CD symptoms were measured based on the newly derived, investigator-scored instrument ADDB-Inv, which included a subscore rating of several of the DSM-IV criteria for CD (ADDB-Inv disruptive behavior). CD symptoms have not specifically been studied in previous ATX trials; thus, our study provides unique new data regarding the effect of ATX on CD symptoms.

Further, as a post-hoc path analysis, we tried to specifically evaluate to what extent the treatment effect of ATX on ODD symptoms could be explained by its effect on ADHD core symptoms or vice versa. The results of these analyses did not reveal a consistent picture, as some contained nonadditive effects. Also, the amount of direct and indirect effects varied across instruments. The findings were also strongly dependent on the underlying hypothetical concept that explains how effects on symptoms influence each other. In addition, moderate to high correlations were found between the respective scores. This was in line with the ADHD/ODD correlation result (r = 0.63) from the Newcorn et al. (2005) study, which compared ADHD patients with or without comorbid ODD. Also a previous meta-analysis including three ATX studies (Biederman et al. 2007) had indicated that there may be a strong relationship of the medication effect on ODD to the magnitude of ADHD response. In summary, this suggests that ATX treatment effects on ADHD, ODD, and CD symptoms should be considered interrelated in this ADHD and ODD/CD comorbid population. Further research in this respect involving other pharmacological and nonpharmacological treatments and various populations is recommended.

Other post hoc analyses explored the influence of the pretreatment status on the effect of ATX on the various efficacy and tolerability endpoints. The treatment effect was only significant in terms of ODD symptoms: a higher treatment effect was observed for the pretreated patients than for the nonpretreated patients. This unexpected finding from our post-hoc analyses needs to be interpreted with caution. Replication in other studies is needed.

Further, individual treatment target behaviors (ITBs) that had been defined for each individual patient based on parent interviews also improved during ATX treatment compared to placebo. A post-hoc review of the ITBs defined at patient level revealed that almost all behaviors were ADHD and/or ODD related. Thus, the specific improvement of these behaviors, which the parents considered a particular burden, may possibly reflect a reduction of the core-symptom-related family burden. Additional information on quality of life and family burden studied in this trial will be published separately.

The primary analysis indicates that ATX may exert a specific effect on ODD symptoms in addition to the improvement of ADHD symptoms. Our results are well in line with the previous systematic meta-analysis by Cheng et al. (2007), who included nine different placebo-controlled ATX studies, which, however, had used another instrument to evaluate ODD symptoms (CPRS-R:S). In the subgroup analysis of patients with ADHD and comorbid ODD (ATX; N = 137, placebo: N = 76; included data from Kaplan et al. 2004; Newcorn et al. 2005), there had been a significant reduction of both ADHD and ODD symptoms with ATX compared to placebo. The lack of significance in the corresponding subgroup analysis (patients with CPRS-R:S data) of the meta-analysis by Biederman et al. (2007) may be related to the small sample size (ATX: N = 46, placebo: N = 32).

Our results need to be discussed in the light of the inconclusive results of the previous study by Bangs et al. (2008), performed in children aged 6–12 years. In the Bangs et al. study, ATX had been superior to placebo in reducing SNAP-IV-ODD subscale scores only at the initial weeks 2 and 5, but no longer at week 8. Mean changes in SNAP-IV ODD subscale scores (±SD) at the end of study were −3.7 ± 5.3 with ATX and −2.9 ± 4.3 with placebo (p = 0.252). The Bangs et al. study had specifically been designed to evaluate improvement of ODD symptoms, and had particularly looked at children with high baseline symptom severity for ADHD and ODD. Only patients with baseline SNAP-IV ODD subscale scores ≥15, and with SNAP-IV ADHD subscale scores at least +1.5 SDs above average age- and gender-specific normative scores, and with CGI-S ≥4 were to be enrolled. Actual mean baseline scores (±SD) on SNAP-IV ODD and SNAP-IV ADHD subscales were 18.9 ± 2.3 and 44.7 ± 6.4 in the Bangs et al. study, compared with 15.5 ± 4.4 and 37.3 ± 9.3 in our study. CGI-S baseline scores were similar in both trials. The ATX target dose was the same in both studies (1.2 mg/kg per day), but the Bangs et al. study had used a faster up-titration, starting with 0.8 mg/kg per day for 3 days only, followed by an immediate increase up to the target dose. Actual mean days on ATX treatment had been 79 days in the Bangs study and 55 days in our study. Treatment periods, sample size, and randomization rates were similar in both studies. It may be hypothesized that the higher baseline severity of ADHD and ODD symptoms, the differences in age distribution, and the fast up-titration used in the Bangs trial have possibly contributed to the differences in primary results, that is, a transient significant decrease of ODD symptoms during the initial weeks no longer present at week 8 in the Bangs et al. study, as opposed to a continuous decrease in ODD symptoms in the current study that was significant versus placebo by week 9.

Additional support for our results comes from the second placebo-controlled, double-blind study from Italy (Dell'Agnello et al. 2009), which evaluated the efficacy of ATX in improving ADHD and ODD symptoms in pediatric patients with ADHD and comorbid ODD who had been nonresponders to a previous parent training intervention. A total of 156 children and adolescents (aged 6–15 years) were enrolled; 137 were available for the efficacy analysis. Actual mean baseline scores on SNAP-IV ODD and SNAP-IV ADHD subscales were 17.2 ± 3.3 and 42.1 ± 6.9, which is lower than in the Bangs et al. (2008) study and higher than in our study. ATX up-titration and target dose were the same as in the fast-titration arm of our study. At the end of the 8-week double-blind treatment period, both ADHD and ODD symptoms had been found significantly improved compared to placebo: mean changes in SNAP-IV ODD subscale scores (±SD) were −2.7 ± 4.1 with ATX and −0.3 ± 2.6 with placebo (p < 0.001, ANCOVA) and thus in the same range as we had observed in our study.

Finally, the dropout rate due to lack of efficacy in the pooled ATX group in this study was 9.1%. In the MTA study, a similar dropout rate due to lack of response of 10% was found during the initial titration period (MTA 1999a). In the recent landmark study by Newcorn et al., which compared ATX versus long-acting MPH, discontinuation rates due to lack of efficacy were 0% with ATX, 2.2% with long-acting MPH, and 5.1% with placebo (Newcorn et al. 2008) during the initial acute-phase treatment period.

In terms of tolerability, AEs reported for ATX in our study were very similar to those described in previous placebo-controlled studies. The most commonly reported AEs (fatigue, nausea, headache, vomiting, and upper updominal pain) are well in line with the current SPC, where decreased appetite, headache, somnolence, abdominal pain, vomiting, and nausea are noted as “very frequent” AEs (reported in >10% of patients).

In addition, several findings suggested that the slow up-titration schedule may be associated with a more beneficial safety profile compared to the fast up-titration. First, clinically relevant AEs during the initial 3 weeks of treatment were reported for 60.0% of patients on ATX-fast and 44.3% of patients on ATX-slow titration. The statistical test did not reach significance, potentially because this study was not powered for the comparison between titration groups for such tolerability differences. This is in accordance with the results from a recent meta-analysis by Greenhill et al. (2007), who found a lower risk for AEs within the first few weeks of treatment associated with slow/twice daily titration (twice-daily dosing and titration to a dose of 1.2 mg/kg per day over at least 2 weeks) compared to fast/once daily titration (once-daily dosing and titration to 1.2 mg/kg per day over 3 days). In the ATX-fast titration group, when grouped, fatigue or related symptoms, GI complaints, and nausea or related symptoms were all significantly more frequent than with placebo, whereas in the ATX-slow titration group, the statistical test versus placebo reached statistical significance only for nausea or related symptoms. Further, descriptive data for rates of treatment-related AEs overall point toward the same direction. Finally, patients in the ATX-slow titration group stayed on treatment significantly longer than patients on placebo. Specific early dropouts due to AEs were too rare for statistical analysis, but the numerical data indicated that they were less frequent in the slow titration group.

Our study has several limitations that should be considered. First, the treatment period of 9 weeks may have been too short to conclude that there is a long-term ATX treatment effect on ODD symptoms. Second, we decided to include patients with less pronounced disease severity in our study compared to the populations in the Bangs et al. (2008) study. This was done since these authors had not found a long-term effect of ATX on ODD symptoms. In the recent study by Dell'Agnello et al. (2009) patients showed intermediate baseline ADHD and ODD symptom severity scores. The ODD symptom reduction in this study was similar to the findings in our study.

Also, our study does not provide any information whether the ATX target dose commonly used for ADHD treatment (1.2 mg/kg per day) also exerts a maximum treatment effect on ODD symptoms. However, preliminary evidence is available from Bangs et al. (2008). They previously reported that ADHD and comorbid ODD patients with low ATX plasma concentrations who did not achieve adequate ODD symptom reduction after 8 weeks on ATX 1.2 mg/kg per day did not benefit from 4 more weeks on ATX at a doubled dose. Recent post-hoc analyses based on the same study further suggested that ATX plasma concentration measurements cannot be used to predict ODD or ADHD symptom improvement, although some patients may benefit from higher plasma concentrations (Hazell et al. 2009).

Conclusions

Our study confirmed that patients with ADHD and comorbid ODD showed statistically and clinically significant improvement of ADHD as well as ODD symptoms after 9 weeks of treatment with ATX once daily (target dose 1.2 mg/kg per day). Taken together with the results of the recent study by Dell'Agnello et al. (2009), our data indicate that ATX may exert a specific effect on ODD symptoms lasting for at least 2 months. Our findings also suggest that ATX may have a treatment effect with respect to the CD symptom complex, which warrants further investigation in additional study populations. Post-hoc analyses indicated interrelationships between the medication effects on ADHD, ODD, and CD symptom scores. Further, the significant improvement seen in individual treatment target behaviors may reflect a decrease in core-symptom-related family burden. Improvements did not differ significantly between ATX-fast and ATX-slow up-titration groups. Therefore, there was no indication that a slower ATX up-titration might be associated with a loss in efficacy when compared to the faster up-titration scheme as recommended by the current SPC. However, the patterns seen for AEs and early dropouts suggest that ATX may be better tolerated when applying a slow up-titration schedule compared to a faster up-titration. Further research is needed to confirm this finding.

Footnotes

Acknowledgments

We would like to thank all investigators and subinvestigators, all patients, and parents who contributed to the conduct of this clinical trial. Statistical Consultant: Alexander Schacht, Lilly Deutschland GmbH, Bad Homburg, Germany.

Disclosures

The study was funded by Lilly Deutschland, the German affiliate of Eli Lilly and Company. Data were analyzed by PSI Contract Drug Development Company, St. Petersburg, Russia. The article was drafted by R.W. Dittmann (first author, drafted introduction and discussion) and K. Helsberg (co-author, medical writer, drafted methods and results). R.W. Dittmann is a former employee of Lilly Deutschland and now holds the Eli Lilly Endowed Chair of Pediatric Psychopharmacology at the Central Institute of Mental Health, Medical Faculty Mannheim, University of Heidelberg. R.W. Dittmann also received research grants from Eli Lilly & Co., and is a member of a Lilly Advisory Board. K. Helsberg, A. Schacht, C. Schneider-Fresenius, Martin Lehmann, and P.M. Wehmeier are full-time employees of Lilly Deutschland. G. Lehmkuhl has received research grants and speaker honoraria from Eli Lilly & Co and Bristol Myers-Squibb, and he is member of a Lilly Advisory Board. R.W. Dittmann, K. Helsberg, C. Schneider-Fresenius, and P.M. Wehmeier own Eli Lilly & Co. stock.

The study was funded by Lilly Deutschland GmbH, Bad Homburg Germany.

References

American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders. 4th Text Revision (DSM-IV-TR) Washington, DC: American Psychiatric Association, 2000.

August

, Realmuto

, MacDonald

3rd , Nugent

, Crosby

. Prevalence of ADHD and comorbid disorders among elementary school children screened for disruptive behavior. J Abnorm Child Psychol, 24:571–595. 1996.

Banaschewski

, Brandeis

, Heinrich

, Albrecht

, Brunner

, Rothenberger

. Association of ADHD and conduct disorder—brain electrical evidence for the existence of a distinct subtype. J Child Psychol Psychiatry, 44:356–376. 2003.

Bangs

, Hazell

, Danckaerts

, Hoare

, Coghill

, Wehmeier

, Williams

, Moore

, Levine

. Atomoxetine ADHD/ODD Study Group. Atomoxetine for the treatment of attention-deficit/hyperactivity disorder and oppositional defiant disorder. Pediatrics, 121:e314–e20. 2008.

Biederman

, Spencer

, Newcorn

, Gao

, Milton

, Feldman

, Witte

. Effect of comorbid symptoms of oppositional defiant disorder on responses to atomoxetine in children with ADHD: A meta-analysis of controlled clinical trial data. Psychopharmacology (Berl), 190:31–41. 2007.

Bollen

. Structural Equations with Latent Variables. Hoboken NJ: Wiley, 1989.

Bymaster

, Katner

, Nelson

, Luecke

, Threlkeld

, Heiligenstein

, Morin

, Gehlert

, Perry

. Atomoxetine increases extracellular levels of norepinephrine and dopamine in prefrontal cortex of rat: A potential mechanism for efficacy in attention deficit/hyperactivity disorder. Neuropsychopharmacology, 27:699–711. 2002.

Cheng

, Chen

, Ko

, Ng

. Efficacy and safety of atomoxetine for attention-deficit/hyperactivity disorder in children and adolescents-meta-analysis and meta-regression analysis. Psychopharmacology (Berl), 194:197–209. 2007.

Connor

, Glatt

, Lopez

, Jackson

, Melloni

Jr.

Psychopharmacology and aggression. I: A meta-analysis of stimulant effects on overt/covert aggression-related behaviors in ADHD. J Am Acad Child Adolesc Psychiatry, 41:253–261. 2002.

10.

Dell'Agnello

, Maschietto

, Pancotto

, Calamoneri

, Masi

, Curatolo

, Besana

, Mancini

, Rossi

, Escobar

, Zuddas

. for the LYCY Study Group. Atomoxetine hydrochloride in the treatment of children and adolescents with attention-deficit/hyperactivity disorder and comorbid oppositional defiant disorder: Results from a placebo-controlled Italian study. Eur Neuropsychopharmacol, 19:822–834. 2009.

11.

Döpfner

, Bruer

, Schürmann

, Wolff Metternich

, Rademacher

, Lehmkuhl

. Effectiveness of an adaptive multimodal treatment in children with attention-deficit hyperactivity disorder—global outcome. Eur Child Adoles Psychiatry, 13,Suppl 1:117–129. 2004.

12.

Döpfner

, Lehmkuhl

, DISYPS

. Diagnostik-System für psychische Störungen im Kindes- und Jugendalter nach ICD-10 und DSM-IV, Second. Bern, Switzerland: Hans Huber, 2000.

13.

Döpfner

, Schürmann

, Frölich

. Therapieprogramm für Kinder mit hyperkinetischem und oppositionellem Trotzverhalten, Second. Weinheim, Germany: Beltz, Psychologie Verlags Union, 1998.

14.

Gadow

, Nolan

, Sverd

, Sprafkin

, Schneider

. Methylphenidate in children with oppositional defiant disorder and both comorbid chronic multiple tic disorder and ADHD. J Child Neurol, 23:981–990. 2008.

15.

Gaub

, Carlson

. Behavioral characteristics of DSM-IV ADHD subtypes in a school-based population. J Abnorm Child Psychol, 25:103–111. 1997.

16.

Greenhill

, Newcorn

, Gao

, Feldman

. Effect of two different methods of initiating atomoxetine on the adverse event profile of atomoxetine. J Am Acad Child Adolesc Psychiatry, 46:566Y572. 2007.

17.

Guy

. ECDEU Assessment Manual for Psychopharmacology. Washington, DC: US Department of health, Education and WelfareNo. 338534–537. 1976.

18.

Hazell

, Becker

, Nikkanen

, Trzepacz

, Tanaka

, Tabas

, D'Souza

, Witcher

, Long

, Posner

, Dittmann

. Relationship between atomoxetine plasma concentration, treatment response and tolerability in attention-deficit/hyperactivity disorder and comorbid oppositional defiant disorder. Atten Defic Hyperact Disord, 1:201–210. 2009.

19.

Kaplan

, Heiligenstein

, West

, Busner

, Harder

, Dittmann

, Casat

, Wernicke

. Efficacy and safety of atomoxetine in childhood attention-deficit/hyperactivity disorder with comorbid oppositional defiant disorder. J Atten Disord, 8:45–52. 2004.

20.

Kelsey

, Sumner

, Casat

, Coury

, Quintana

, Saylor

, Sutton

, Gonzales

, Malcolm

, Schuh

, Allen

. Once-daily atomoxetine treatment for children with attention-deficit/hyperactivity disorder, including an assessment of evening and morning behavior: A double-blind, placebo-controlled trial. Pediatrics, 114:e1–e8. 2004.

21.

Kirley

, Lowe

, Mullins

, McCarron

, Daly

, Waldman

, Fitzgerald

, Gill

, Hawi

. Phenotype studies of the DRD4 gene polymorphisms in ADHD. Association with oppositional defiant disorder, positive family history. Am J Med Genet B Neuropsychiatr Genet, 131B:38–42. 2004.

22.

Klein

, Abikoff

, Klass

, Ganeles

, Seese

, Pollack

. Clinical efficacy of methylphenidate in conduct disorder with and without attention deficit hyperactivity disorder. Arch Gen Psychiatry, 54:1073–1080. 1997.

23.

Kuhne

, Schachar

, Tannock

. Impact of comorbid oppositional or conduct problems on attention-deficit hyperactivity disorder. J Am Acad Child Adolesc Psychiatry, 36:1715–1725. 1997.

24.

Luman

, van Noesel

, Papanikolau

, Van Oostenbruggen-Scheffer

, Veugelers

, Sergeant

, Oosterlaan

. Inhibition, Reinforcement Sensitivity and Temporal Information Processing in ADHD and ADHD + ODD: Evidence of a Separate Entity? J Abnorm Child Psychol, 37:1123–1135. 2009.

25.

Michelson

, Allen

, Busner

, Casat

, Dunn

, Kratochvil

, Newcorn

, Sallee

, Sangal

, Saylor

, West

, Kelsey

, Wernicke

, Trapp

, Harder

. Once-daily atomoxetine treatment for children and adolescents with attention deficit hyperactivity disorder: A randomized, placebo-controlled study. Am J Psychiatry, 159:1896–1901. 2002.

26.

Michelson

, Fries

, Wernicke

, Kelsey

, Kendrick

, Sallee

, Spencer

. Atomoxetine ADHD Study Group. Atomoxetine in the treatment of children, adolescents with attention deficit/hyperactivity disorder: A randomized, placebo-controlled dose-response study. Pediatrics, 108:e83. 2001.

27.

MTA Cooperative Group. A 14-month randomized clinical trial of treatment strategies for attention-deficit/hyperactivity disorder. The MTA Cooperative Group. Multimodal Treatment Study of Children with ADHD. Arch Gen Psychiatry, 56:1073–1186. 1999a.

28.

MTA Cooperative Group. Moderators and mediators of treatment response for children with attention-deficit/hyperactivity disorder: The multimodal treatment study of children with attention-deficit/hyperactivity disorder. Arch Gen Psychiatry, 56:1088–1096. 1999b.

29.

National Institutes of Mental Health. Clinical global impressions. Psychopharmacol Bull, 21:839–943. 1985.

30.

Newcorn

, Kratochvil

, Allen

, Casat

, Ruff

, Moore

, Michelson

. Atomoxetine/Methylphenidate Comparative Study Group. Atomoxetine and osmotically released methylphenidate for the treatment of attention deficit hyperactivity disorder: Acute comparison and differential response. Am J Psychiatry, 165:721–730. 2008.

31.

Newcorn

, Spencer

, Biederman

, Milton

, Michelson

. Atomoxetine treatment in children and adolescents with attention-deficit/hyperactivity disorder and comorbid oppositional defiant disorder. J Am Acad Child Adolesc Psychiatry, 44:240–248. 2005.

32.

Oosterlaan

, Scheres

, Sergeant

. Which executive functioning deficits are associated with AD/HD, ODD/CD and comorbid AD/HD + ODD/CD? J Abnorm Child Psychol, 33:69–85. 2005.

33.

Satake

, Yamashita

, Yoshida

. The family psychosocial characteristics of children with attention-deficit hyperactivity disorder with or without oppositional or conduct problems in Japan. Child Psychiatry Hum Dev, 34:219–235. 2004.

34.

Sinzig

, Döpfner

, Lehmkuhl G; German Methylphenidate Study

Group

, Uebel

, Schmeck

, Poustka

, Gerber

, Günter

, Knölker

, Gehrke

, Hässler

, Resch

, Brünger

, Ose

, Fischer

. Long-acting methylphenidate has an effect on aggressive behavior in children with attention-deficit/hyperactivity disorder. J Child Adolesc Psychopharmacol, 17:421–432. 2007.

35.

Spencer

, Biederman

, Heiligenstein

, Wilens

, Faries

, Prince

, Faraone

, Rea

, Witcher

, Zervas

. An open-label, dose-ranging study of atomoxetine in children with attention deficit hyperactivity disorder. J Child Adolesc Psychopharmacol, 11:251–265. 2001.

36.

Spencer

, Biederman

, Wilens

, Prince

, Hatch

, Jones

, Harding

, Farone

, Seidman

. Effectiveness and tolerability of tomoxetine in adults with attention deficit hyperactivity disorder. Am J Psychiatry, 155:693–695. 1998.

37.

Spencer

, Heiligenstein

, Biederman

, Faries

, Kratochvil

, Conners

, Potter

. Results from 2 proof-of-concept, placebo-controlled studies of atomoxetine in children with attention-deficit/hyperactivity disorder. J Clin Psychiatry, 63:1140–1147. 2002.

38.

Spencer

, Abikoff

, Connor

, Biederman

, Pliszka

, Boellner

, Read

, Pratt

. Efficacy and safety of mixed amphetamine salts extended release (adderall XR) in the management of oppositional defiant disorder with or without comorbid attention-deficit/hyperactivity disorder in school-aged children and adolescents: A 4-week, multicenter, randomized, double-blind, parallel-group, placebo-controlled, forced-dose-escalation study. Clin Ther, 28:402–418. 2006.

39.

Swanson

. School-based assessments and interventions for ADD. Irvine, CA: KC Publishing, 184. 1992.

40.

Swanson

, Kraemer

, Hinshaw

, Arnold

, Conners

, Abikoff

, Clevenger

, Davies

, Elliott

, Greenhill

, Hechtman

, Hoza

, Jensen

, March

, Newcorn

, Owens

, Pelham

, Schiller

, Severe

, Simpson

, Vitiello

, Wells

, Wigal

, Wu

. Clinical relevance of the primary findings of the MTA: Success rates based on severity of ADHD and ODD symptoms at the end of treatment. J Am Acad Child Adolesc Psychiatry, 40:168–179. 2001.

41.

Turgay

. Treatment of comorbidity in conduct disorder with attention-deficit hyperactivity disorder (ADHD) Essent Psychopharmacol, 6:277–290. 2005.

42.

World Health Organization (WHO). Multiaxial Classification of Child and Adolescent Psychiatric Disorders: The ICD-10 Classification of Mental and Behavioural Disorders in Children and Adolescents. Cambridge: Cambridge University Press, 2004.