The future of child and adolescent clinical psychopharmacology: A systematic review of phase 2, 3, or 4 randomized controlled trials of pharmacologic agents without regulatory approval or for unapproved indications

We

The future of child and adolescent clinical psychopharmacology: A systematic review of phase 2, 3, or 4 randomized controlled trials of pharmacologic agents without regulatory approval or for unapproved indications

Introduction
The treatment of child and adolescent mental health conditions includes pharmacological and non-pharmacological options, such as psychological therapies (Correll et al., 2021).Although not every mental health condition may be amenable to pharmacological treatment, we lack evidence-based effective pharmacological options for the core symptoms of several prominent conditions frequently managed by child and adolescent mental health services, such as autism spectrum disorder (ASD), posttraumatic stress disorder (PTSD) and anorexia nervosa.The most comprehensive umbrella review (Correll et al., 2021) on efficacy and acceptability of medications in child and adolescent mental health found the strongest support [in terms of highest effect sizes (ESs)] for the following compounds: amphetamine and methylphenidate for core symptoms of attention-deficit/hyperactivity disorder (ADHD); aripiprazole and risperidone for irritability in ASD; risperidone for aggression in disruptive behavior disorders; risperidone, olanzapine, paliperidone, and ziprasidone for symptoms of schizophrenia; fluoxetine for depression; aripiprazole for manic symptoms in bipolar disorder; fluoxetine for anxiety; and fluoxetine/other selective serotonin reuptake inhibitors (SSRIs) for obsessive-compulsive disorder (OCD).A related umbrella review (Solmi et al., 2020) focusing on safety found that the best tolerability/safety profile emerged for escitalopram and fluoxetine (for 1 Co-first authors, equally contributed. 2Posthumously. 3Co-senior authors, equally contributed. S. Cortese et al. depression), lurasidone (for schizophrenia), methylphenidate (for ADHD), and lithium (for bipolar disorder, manic episode).Concerns were identified in relation to the safety profile of venlafaxine, olanzapine, atomoxetine, guanfacine, and valproate.
Even for disorders for which effective medications are available for their core symptoms, pharmacological options for associated problems are suboptimal.For instance, psychostimulants for core symptoms of ADHD (Cortese, 2020) are the most efficacious medications, at least in the short-term, in psychiatry and among the most efficacious medications across all the medical disciplines (Leucht et al., 2012).However, the impact of psychostimulants on other aspects related to ADHD is less striking.For instance, their effects on executive dysfunctions (McKenzie et al., 2022) vary according to the type of executive dysfunction and their benefits on emotional dysregulation (Lenzi et al., 2018) are characterized by lower ESs.
There have been concerns that the pace of development of clinical trials and regulatory approval of novel medications in child and adolescent psychiatry is slow, and drug companies are pulling away from the field given the substantial failures in their programmes (Persico et al., 2015).Quantitative evidence is needed to exactly inform how many medications have been successful in phase 2, 3 or 4 RCTs of agents for child and adolescent mental health disorders.Previous reviews have presented medications in the pipeline or not licensed for specific disorders, e.g., ASD (Baribeau et al., 2022) or ADHD (Nageye and Cortese, 2019).However, to our knowledge, no review has systematically assessed novel unlicensed or off-labelled medications across the main mental health disorders in children/adolescents. Here, we aimed to fill this gap by systematically reviewing recent progress and current clinical trial activity, evaluating promising compounds for child and adolescent mental health problems.As dietary or probiotic interventions are chemical substances that may be recommended by some practitioners and are of interest for patients and their families, we also included RCTs focusing on these compounds.

Methods
The study protocol is available in Open Science Framework, OSF (htt ps://bit.ly/3EiEi5h).

Search strategy
We searched https://clinicaltrials.gov/and and https://www.clinicaltrialsregister.eu/from 01/01/2010-08/23/2022 using search terms related to the mental health conditions of interest for this review (see below).The search was conducted independently by two investigators (KMG and SC).The time frame is similar to the one considered in a recent similar review on phase 2/3 RCTs of psychopharmacological agents in adults (Correll, 2023).We also conducted an additional systematic targeted search in PubMed to check if identified RCTs for which results were not available in the clinical trials platforms had been published.The following filters were used for the search in clinicaltrials.gov: 1) study type: interventional studies (Clinical Trials); 2) recruitment: not yet recruiting/recruiting/enrolling by invitation/active, not recruiting/terminated/completed/unknown status; 3) age group: "Child" (birth-17); 4) phase: phase 2/phase 3 or 4; 5) study start: from 01/01/2010, which, as in the systematic review by Correll et al. (2023) in adults, was deemed adequate to reflect recent developments in the field.We assumed that, if a trial initiated ≥ 12 years ago and its results had not been published or no additional studies were ongoing, this trial program had been discontinued.
The following filters were used for the search in https://www.clinicaltrialsregister.eu/: 1) select trial status: completed/ongoing/restarted; 2) age range: children and adolescents; 3) select trial phase: phase two/ phase three/phase four; and 4) select date range: 2010-01-01 (with the same reasoning for the cut-off date as mentioned for clinicaltrials.gov).

Inclusion and exclusion criteria
We included ongoing or completed phase 2 or 3 RCTs, regardless of their level of blinding, assessing pharmacologic agents, dietary supplements or probiotics that had to the best of our knowledge no regulatory approval in the US, Europe (through EMA licensing procedures, not those approved by individual countries through national licensing procedures) or Asia as of 08/23/2022, for mental health conditions in children or adolescents (all participants aged 18 years or less).We also included phase 4 RCTs of agents approved in psychiatry or other areas of medicine but targeting a currently unapproved mental health indication or an age range different from the one in the approval label in children/ adolescents.
We focused on RCTs targeting the following mental health conditions (in alphabetical order): ADHD, Anxiety Disorders, ASD, Bipolar Disorder, Conduct Disorder/Oppositional Defiant Disorder/Disruptive Mood Dysregulation Disorder/Intermittent Explosive Disorder, Depressive Disorder (including Major Depressive Disorder), Eating Disorders, Intellectual Developmental Disorder (Intellectual and Developmental Disability, IDD), OCD, PTSD (the inclusion of this disorder was post hoc in relation to the protocol), Schizophrenia, and Tourette's Syndrome, accepting any diagnostic definition reported by the study investigators.
We excluded the following interventions: brain stimulation, digital app-based, and psychosocial interventions, except when they were combined with novel pharmacological /dietary treatments.We also excluded any trial program that was listed in the clinical trials registries as having been discontinued or stopped, and RCTs of agents that were abandoned and are not being pursued further based on information in the public domain.

Classification of the mechanisms of action of the tested agents
To classify the possible mechanisms of action of the tested agents, we referred whenever possible to the Neuroscience based Nomenclature (NbN) website (https://nbn2r.com/).A version of the NbN is available for medications used in child and adolescent psychiatry, NbN C&A (https://nbnca.com/)(Cortese et al., 2022).

Assessment of study characteristics
We aimed to present the academic-sponsored versus the industryfunded RCTs, and, with reference to the results of the systematic review of recent/ongoing RCTs of psychotropics in adults by Correll et al. (2023), a comparison between findings in child/adolescent and adults mental disorders, respectively.

Evaluation of promising compounds
After summarizing the search results, we highlighted those agents and mechanisms of action in each disease category that were considered to be most promising based on the current level of evidence with regard to i) positive phase 2 and/or phase 3 or 4 clinical trials indicating superiority vs. placebo/other control; ii) magnitude of the observed effect, with reference to the benchmarks suggested by Cohen (Cohen, 1988): 0.2: small; 0.5: medium; 0,8: large ESs; iii) demonstration of minimum requirements for safety/tolerability, in terms of lack of severe adverse events as defined by the Food and Drug Administration (FDA), i.e., those: resulting in death, or life threatening, or requiring inpatient hospitalisation or causing prolongation of existing hospitalisation, or resulting in persistent or significant disability/incapacity, or contributing to a congenital anomaly/birth defect, or requiring intervention to prevent permanent impairment or damage; and iv) consistency of findings within a clinical trials program, i.e., positive results across all the RCTs testing the medication.
RCTs with positive results on at least one primary outcome (n = 26), and those with negative results on every primary outcome (n = 43) are reported in Tables 1-4, grouped by disorder (in alphabetical order).When available, Tables 1-4 report also data on tolerability, in terms of percentage of participants who dropped out due to adverse events or those who experienced adverse events defined as serious by the study authors, in line with the above-mentioned FDA classification.
Fig. 1 shows the number of positive and negative RCTs for each disorder.Fig. 2 reports a bar graph depicting the number of trials for each condition, indicating the portion of academic sponsored versus industry-funded trials.Mechanisms of action of the compounds assessed in at least five RCTs, by conditions and overall, are reported in Fig. 3.A comparison of the number of adult versus child trials by condition is reported in Fig. 4.
The following sections provide a summary of the efficacy and, when available, tolerability results, from the retrieved RCTs, grouped by disorder, in alphabetical order.Availability of trial results refer to the last full check of the databases (08/23/2022).
As for tolerability, in one study of dasotraline (NCT02428088) discontinuation rates due to adverse events were higher in the dasotraline 4 mg/day arm (12.2%) compared with the 2 mg/day arm (6.3%) and placebo (1.7%).There were no serious adverse events or clinically meaningful changes in blood pressure or heart rate with dasotraline and lisdexamfetamine was generally well tolerated.
Furthermore, one study showed that coenzyme Q was effective when added to atomoxetine (decreasing total ADHD symptom severity on the Conners parent-rating scale by about 34%, vs. 18% in the atomoxetine only group, ES not reported).Finally, in another RCT, treatment with micronutrients improved one of the primary outcomes (the clinical Global Impression Scale-CGI, ES not reported) but not the other primary outcome as labelled by the authors (parents' rating of ADHD symptoms).
Two (out of three) RCTs of fasoracetam showed no significant effects on the primary outcome (results were not available, yet in the third RCT).Likewise, omega-3 fatty acids were not superior to placebo in the only RCT that reported results.
Results with statistical analyses from the RCTs of the other agents were not available.

Anxiety disorders
Seven RCTs were retained.Altogether, 29% were funded by drug companies and 71% were sponsored by universities/hospitals.One RCT focused on generalized anxiety disorder exclusively, the others recruited participants with a variety of anxiety disorders (mainly generalized, social and/or separation anxiety disorder).Two mechanisms of action were assessed, including 4 compounds.Mechanisms of action of the compounds assessed in RCTs in anxiety disorders include: The RCT of guanfacine XR showed no significant differences in the scores of the exploratory efficacy measures (Pediatric Anxiety Rating Scale [PARS] and Screen for Child Anxiety Related Emotional Disorders [SCARED]) although at endpoint, more participants assigned to                 Superior at 2 mg dose, not superior at 0.5 mg dose.
Regarding compounds for which results were available, in one RCT (NCT01078714), bumetanide was superior to placebo for the primary outcome [Childhood Autism Rating Scale (CARS) (p = 0.004)].Of note, in another RCT (2013-003259-39), bumetanide was superior to placebo in the secondary outcomes (Social Responsiveness Scale -SRS, CGI-I p = 0.0043, ES not reported) but not in the primary outcome (CARS).Similarly, in another RCT (2014-001560-35), bumetanide was not  superior to placebo in the primary outcome (SRS).Finally, another RCT of bumetanide was terminated as the 6-month efficacy analysis on the primary outcome (CARS-2 scores) showed no separation from placebo.D-cycloserine was not superior to placebo at the first endpoint (11 weeks) in the primary outcome (SRS) but it separated significantly from placebo at the 22-week analysis (p = 0.048) on the SRS.Suramin was significantly better than placebo in one of the two primary outcomes (Autism Diagnostic Observation Schedule; p = 0.0028 -ES not reported) but not on the other primary outcome (expressive language).
In one RCT (NCT01372449), memantine was not superior to placebo in the primary outcome (adaptive behavior, measured with the Vineland Adaptive Behavior scale).In a withdrawal RCT, memantine did not significantly separate from placebo in any outcome, including the primary one: Proportion of Patients Meeting the Criterion for Loss of Therapeutic Response.Oxytocin was not superior to placebo in two RCTs (NCT01944046 and 2018-000769-35) in any of the outcomes.In another RCT (NCT01624194), oxytocin was superior to placebo on the Total SRS score (primary outcome; p = 0.027) and was well tolerated.In terms of non-core symptoms of ASD, in one RCT, guanfacine XR was found better than placebo for the symptom "hyperactivity" in ASD, measured with the Aberrant Behavior Checklist Hyperactivity Subscale (rather than a formal diagnosis of ADHD for which it is approved).
Four other compounds were found not significantly different from placebo on any outcome: sertraline on expressive language scores, adaptive behavior, anxiety sensory processing, and CGI-I; everolimus on IQ, autistic symptoms, motor skills, sleep, behavioral/emotional problems and quality of life; mirtazapine on anxiety; and lurasidone on irritability.Metformin was tested for the management of overweight/ obesity induced by dopamine and serotonin-dopamine antagonist medications in young people with ASD and was superior to placebo (p = 0.003).Melatonin was superior to placebo for sleep onset latency in one RCT (p < 0.0001 in the double-blind treatment phase. Among the non-pharmaceutical agents, folinic acid (for ASD core symptoms) and methylcobalamin (on the CGI-I), were superior to placebo in single RCTs.Omega-3 fatty acids (2 RCTs with reported results) and sulforaphane (1 RCT) were not superior to placebo.

Bipolar disorder
We found six RCTs, four (67%) sponsored by hospitals/university and two (33%) funded by pharmaceutical companies.Five RCTs included pharmaceutical agents, assessing 6 modes of action, and including 6 compounds.Mechanisms of action of the compounds assessed in RCTs in bipolar disorder included: 1. Glutamate channel blockade (carbamazepine, n = 1) 2. Dopaminergic partial agonism and serotoninergic antagonism (cariprazine, n = 1) 3. Inhibition of inositol monophosphatase, adenyl-cyclase, guanosine monophosphate (GMP), glycogen synthase kinase 3, increasing activity of serotonin and acetylcholine; modulator of intracellular signalling cascade (lithium, n = 1; note: lithium has FDA regulatory approval for adolescents aged 12-17 years; here it was tested in children aged 7-17 years), 4. Glutamate receptor antagonism (memantine, n = 1) 5. Dopaminergic and serotoninergic antagonism (perospirone, n = 1) given with lithium Another RCT tested inositol plus omega-3 free fatty acids.All RCTs, except that for perospirone + lithium, were focused on treating manic/mixed symptoms.Results were reported for the RCT of lithium (n = 53), which was superior to placebo (n = 28) on the Young Mania Rating Scale, ES: 0.53 (0.06-0.99) and generally well tolerated, and for the pilot RCT of inositol plus omega-3 fatty acids, which was superior to inositol plus placebo or omega-3 fatty acids plus placebo (ES not reported) and well tolerated.

Conduct disorder/oppositional defiant disorder/disruptive mood dysregulation disorder/intermittent explosive disorder
We found five RCTs, all sponsored by universities/hospitals.Four modes of action were assessed, including four compounds.Mechanisms  of action of the compounds assessed in RCTs in these disorders included: 1. Dopaminergic, norepinephrinergic, and serotoninergic antagonism (risperidone, n = 2; note: risperidone is approved in isolated European countries but not across Europe through European Medicine Agency (EMA) approval or in the US for conduct disorder) 2. Oxytocin receptor agonism (oxytocin, n = 1) 3. Norepinephrinergic (alpha-2) receptor agonism (guanfacine XR, n = 1) Another RCT tested omega-3 fatty acids.
Results were not available from any of these RCTs.
In one proof-of-concept cross-over RCT (n = 17 participants, 16 of which received both treatments) (NCT02579928, Dwyer et al., 2021), a single ketamine infusion significantly reduced depressive symptoms after 24 h compared to midazolam (ES: 0.78, 95% CI not reported), measured on the Montgomery-Åsberg Depression Rating Scale (MADRS) (primary outcome) and improvements remained 14 days after treatment, but no significant differences were found on the Children's Depression Rating Scale-Revised at days 1 and 24.Of note, unblinding for ketamine was 100%.Although ketamine was associated with transient, self-limited dissociative symptoms, there were no serious adverse events.It should be noted that the study was not powered to detect rare events.
Agomelatine 25 mg (but not 10 mg/day) was statistically superior to placebo (ES: 0.29, 95% CI not reported) and comparable to fluoxetine (ES: 0.26 95% CI not reported) in the whole group of children and adolescents.Findings were similar in the adolescent subgroup (ES: agomelatine: 0.36; fluoxetine: 0.27, 95% CI not reported) but not in children; however it should be noted that the study was underpowered in children.Overall, agomelatine was well tolerated.

Eating disorders
Four RCTs were retained, all sponsored by universities/hospitals.Four mechanisms of action were assessed, including 4 different compounds.Mechanisms of action of the compounds assessed in RCTs in eating disorders included: 1. Dopaminergic and serotoninergic partial agonism and antagonism (aripiprazole, n = 1) 2. Partial agonist at the glycine NMDA co-agonist site and antibiotic (Dcycloserine, n = 1) 3. Steroid hormone (megestrol acetate, n = 1) 4. Hormonal activity (somatropin, n = 1) All these RCTs recruited participants with anorexia nervosa, except for the RCT testing D-cycloserine that focused on feeding disorders.The proof-of-concept RCT on somatropin showed that the percentage of patients with a height velocity > 5 cm/year during the study period was greater in the active compared to the placebo group (100% vs. 50%, p = 0.05).Results were not available for the other RCTs.

Intellectual and developmental disability (IDD)
The vast majority of identified trials (49/41) in this section pertain to genetic syndromes associated with IDD, even though the presence of IDD was not always documented in the retrieved RCTs.Nonetheless, we have reported RCTs as they may provide interesting etiopathophysiologybased interventions.
Forty-one RCTs, including 4 RCTs of dietary supplements, were found, 60% of which were sponsored by university/hospitals/public bodies and 40% by drug companies (55% and 45%, respectively for pharmacological compounds).
Additionally, 3 RCTs tested combination vitamin C and E therapy and one trial investigated coenzyme Q10 therapy.
Fourteen RCTs focused on IDD in fragile X syndrome, 10 on participants with Prader-Willi syndrome, 6 on Rett syndrome, 3 on Down syndrome, 2 on tuberous sclerosis complex, and 1 each on Dup15q syndrome, pyruvate dehydrogenase complex deficiency, neuropathic or non-neuropathic mucopolysaccharidosis Type II, and Hunter syndrome.Two trials recruited patients with intellectual and developmental disability, testing the effect of investigational products on ADHD symptoms or serious behavioural problems.
Carbetocin was well tolerated and significantly better than placebo in two RCTs (ES not reported) to decrease hyperphagia scores in children/adolescents measured via the Hyperphagia for Prader-Willi Syndrome Questionnaire.In another RCT, oxytocin (Syntocinon) was superior to placebo for hyperphagia scores on the Hyperphagia Questionnaire (HQ)-Total Factor Score but not on the other primary outcomes (HQ-Behavior Factor Score, HQ-Drive Factor Score, and HQ-Severity Factor Score).However, in another RCT, oxytocin was not superior to placebo for hyperphagia in Prader-Willi syndrome.Negative results concerning the primary outcomes were found in RCTs of everolimus (in one RCT, without available results for the other RCT) for individuals with tuberous sclerosis, AFQ056, cannabidiol (1 RCT; results not available for 3 other RCTs), lovastatin, and ganaxolone in fragile X syndrome, dextromethorphan in Rett syndrome, idursulfase in Hunter syndrome, and thyroxine in Down syndrome.

Obsessive compulsive disorder (OCD)
Nine RCTs were found, all but one (88.8%)sponsored by universities/hospitals.Seven modes of action were assessed, including 8 compounds.All identified pharmacological compounds and focused on the following mechanisms of action: 1. Dopaminergic partial agonism (aripiprazole, n = 1) 2. Selective serotonin reuptake inhibition (fluvoxamine, n = 1; note: fluvoxamine has FDA approval in children 8 years old or older; here, participants were 6-17 years) 3. NMDA receptor agonism (D-cycloserine, n = 2) 4. Enzyme modulation (naproxen sodium, n = 1; celecoxib, n = 1; note: non-selective and selective cyclooxygenase [COX] inhibition) 5. Immunomodulation (gamunex [immunoglobulin], n = 1) 6. Glutathione enhancement (N-acetylcysteine, n = 1) 7. Antibiotic (azithromycin, n = 1) The RCT on fluvoxamine was positive (significant difference between fluvoxamine and placebo on the Japanese version of the Children's Yale-Brown Obsessive-Compulsive Scale (CY-BOCS), p = 0.044, ES not reported).In one RCT (NCT01411774), D-cycloserine was not superior to placebo on the CGI-S (secondary outcome; results not reported for the primary outcome: CY-BOCS).In another RCT recruiting participants with Pediatric Autoimmune Neuropsychiatric Disorders Associated with Streptococcal infections (PANDAS) and OCD, the difference in the mean decrease in the CY-BOCS score was not significant between the intravenous immunoglobulin Gamunex and the placebo group.Results for the other RCTs were not reported.

Post traumatic stress disorder (PTSD)
We found three RCTs: one, testing sertraline (SSRI) funded by a drug company and two, assessing propranolol (beta-blocker) combined with memory consolidation and reactivation, respectively, sponsored by universities/hospitals.The only RCT with reported results, the one on sertraline, failed to find any significant effect of sertraline vs placebo.

Schizophrenia
Five RCTs were identified, recruiting exclusively individuals aged 17 or younger and testing pharmacological compounds.Of these trials, 60% were funded by drug companies and 40% were sponsored by universities/hospitals.Five modes of action were assessed, including 5 compounds.Mechanisms of action of the compounds assessed in RCTs in schizophrenia included: 1. Dopaminergic, norepinephrinergic, and serotoninergic antagonism (asenapine, n = 1) 2. Dopaminergic and serotoninergic partial agonism (cariprazine, n = 1) 3. COX-2 inhibition (celecoxib, n = 1) 4. D-Amino acid oxidase (DAAO) inhibition (sodium benzoate, n = 1; note: increasing levels of the NMDA co-agonist D-serine) 5. Glutathione enhancement (N-acetylcysteine, n = 1) Asenapine failed to separate from placebo on the Positive and Negative Syndrome Scale (PANSS) Total score (primary outcome), as well as on the subscales of the PANSS and CGI-S (with significant improvement in the CGI-S score observed in the 5 mg b.i.d.group).Results from the other RCTs were not available.

Tourette's syndrome
We retained 12 RCTs, including 1 RCT on a Chinese medicine formula.About 67% (73% when limiting to RCTs of pharmacological agents) were funded by drug companies, the rest were sponsored by universities/hospitals.Six modes of action were assessed, including 7 compounds.Mechanisms of action of the compounds assessed in RCTs in Tourette's syndrome included: 1. Vesicular monoamine transporter-2 inhibition (deutetrabenazine, n = 3; valbenazine, n = 2) 2. Dopaminergic receptor antagonism (ecopipam, n = 2; note: selective dopamine D1 receptor antagonist) 3. Norepinephrinergic receptor agonism (guanfacine XR, n = 1) 4. Histaminergic (3) receptor antagonism (AZD5213, n = 1) 5. Cannabinoid receptor agonism (tetrahydrocannabinol + cannabidiol, n = 1) 6. Glutathione enhancement (N-acetylcysteine, n = 1) Positive RCTs with superiority compared to placebo on the primary outcome Yale Global Tic Severity Scale included the one RCT on AZD5213, which was superior to placebo at 2 mg but not 0.5 mg dose (ES, as well as statistical analysis on side effects, not reported), and one of the two RCTs on ecopipam (ES not reported; results not available for the second RCT).Ecopipam was overall well tolerated.Negative RCTs included those on deutetrabenazine (2 out of 3 RCTs, results not reported for the third trial) and both RCTs of valbenazine.

Discussion
This is, to our knowledge, the first systematic review of phase 2-4 RCTs of compounds across mental health conditions in children and adolescents.About 11% (n = 26) of the retrieved RCTs (n = 234) had positive findings on ≥ 1 primary outcome.The only two compounds with evidence of significant effects that were replicated in ≥ 1 additional RCT without any negative RCTs were dasotraline for ADHDwhich program was halted by the manufacturer in 2020 -and carbetocin for hyperphagia in Prader-Willi syndrome.
The number of retrieved RCTs was unevenly spread across the childhood mental disorders.The bulk of the retrieved RCTs (n = 84, 36%) were for ASD, which is likely accounted for by concerns regarding the lack of approved medications for the defining symptoms of this increasingly more recognised and highly impairing condition (Solmi et al., 2022a).A relatively large number of RCTs (n = 41, 18%) was also found for children with a variety of genetic syndromes associated with IDD and the retrieved RCTs focused on associated mental/physical impairments (e.g., hyperphagia in Prader-Willi syndrome), rather than cognitive or functional abilities per se.Of note, in some RCTs, it was not clearly reported if IQ was tested.Another relatively large number of RCTs (n = 40, 17%) was retrieved for ADHD, reflecting the need not only for novel agents, ideally without abuse potential, but also for the approval of licensed agents in pre-schoolers with ADHD, given the increasing attention to this subgroup of children with ADHD (Halperin and Marks, 2019) (Cortese, 2022).By contrast, a limited number of RCTs were found for other conditions in need of additional pharmacological options, such as anxiety disorders, eating disorders, externalizing/disruptive behavior disorders, mood disorders, OCD, and Tourette's syndrome, and data from the majority of the RCTs retrieved for schizophrenia were not available.
The positive findings of the RCTs included in this review should be considered alongside the effect size (ES) and tolerability of the tested compound, and the availability and efficacy of other agents for any specific disorder.Regarding ADHD, given the high effect size of stimulants (in the order of 0.8-1.0)(Cortese, 2020) the moderate effect size reported for dasotraline (0.48), which is comparable to that of atomoxetine (Schwartz and Correll, 2014) and alpha-2 agonists clonidine and guanfacine (Hirota et al., 2014), would position this compound as a possible second-or third-line pharmacological option.Nevertheless, another non-stimulant option could be still valuable for those patients (around 15% in RCTs and probably more in daily clinical practice) where comorbidities such as IDD or ASD may decrease the response rate (Cortese et al., 2021) and/or those who cannot tolerate available medications.Of note, while high dose dasotraline (4 mg/day) was less well tolerated than placebo, at a low dose (2 mg/day),dasotraline did not separate from placebo in terms of tolerability.Further, no serious adverse events were reported in the dasotraline RCTs.However, as mentioned above, the development program for dasotraline was halted by its manufacturer in 2020.
Regarding ASD, while positive individual RCTs focused mainly on associated symptoms and impairment, the search for agents targeting defining symptoms that are supported by replicated evidence continues to be elusive (Barak and Feng, 2016).Since ASD begins very early in life (Solmi et al., 2022b), abnormal biological processes may occur in a time-bound fashion during potentially developmentally vulnerable times that may require specific mechanistic interventions at certain developmental phases (Green et al., 2010).
Regarding depression, positive findings for agomelatine and, partially, for ketamine are promising and welcome, considering the limited range of approved options in children and adolescents (fluoxetine, for youth aged 8-17 years, and escitalopram for those aged 12-17) and the fact that only about 40% of youth have been found to respond to cognitive behavioral therapy (CBT) (March et al., 2004).However, independent replications of the positive findings for agomelatine are required.Regarding ketamine, it should be noted that no significant differences were reported on an additional (i.e., other than the one used as primary outcome) depression scales.Moreover, ketamine was associated with transient, self-limited dissociative symptoms, which calls for further assessment of its safety.
Likewise, the positive findings for AZD5213 and ecopipam in relation to Tourette's syndrome require replication, alongside a better understanding of the specific effect sizes and tolerability.These currently missing data are especially relevant in comparison to alpha-2 agonists, given that the two most common currently used options in clinical care, alpha-2 agonists and D2 antagonists/partial agonists, have been considered to have similar effect sizes for tic severity reduction, but alpha-2 agonists have better tolerability (Whittington et al., 2016), even though a recent network meta-analysis showed superiority of antipsychotic over alpha-2 agonists in terms of efficacy (Farhat et al., 2023).Similarly, for bipolar disorder, more evidence is needed for inositol + omega-3 fatty acids, in particular data on their effect size and tolerability of lithium in preadolescents.
The lack of positive findings for the core symptoms of other disorders reflects several factors including the clinical challenges in conducting RCTs in children and adolescents, possible placebo effects (Huneke et al., 2022), and the theoretical possibility that some disorders might not be treatable with medications.
Indeed, probably the main conceptual/methodological weakness of the body of research retrieved via our search is the fact that the agent was tested as a "one-size-fits-all" treatment.An exception to this was represented by RCTs in children with IDD, the majority of which included children with IDD within the framework of a genetic syndrome.In these RCTs, the physiological consequences provided the rationale to test specific compounds thought to address the specific pathophysiology of the syndrome.In a few cases only, e.g., in a RCT of the glutamate receptor agonist fasoracetam in children with ADHD with and without mGluR mutations, a stratification of the sample based on neurobiological features was implemented.Therefore, we highlight the potential value of the approach proposed by the Research Domain Criteria framework (Sanislow, 2020), as an opportunity for stratificationincluding cognitive stratification -of patients to be recruited in RCTs.An additional advantage of this approach rests in the evidence it can provide for transnosographic outcomes (such as irritability/aggressiveness) that are arguably highly relevant in child and adolescent mental health.More research into diagnostic and predictive biomarkers is needed, as these are currently missing in a well-replicated fashion for mental disorders with onset during childhood and adolescence (Cortese et al., 2023).
The limited number of RCTs for schizophrenia, with no positive findings, could seem disappointing.However, first, several dopamine antagonists/partial agonists are already approved and available for adolescents with schizophrenia (Pagsberg et al., 2017).Moreover, we excluded a number of RCTs where adolescents were recruited alongside adults, following more recent guidance by the FDA that considers the option of extrapolation of more limited adolescent data embedded within a larger adult trial program under certain circumstances (FDA, accessed 2023).Of note, we limited our focus to schizophrenia, rather than other psychoses, as their heterogeneous nature would hamper the consistency of findings across RCTs.
Several reasons may explain why relatively few RCTs targeted certain mental health conditions in children and adolescents compared to programmes in adults (Correll, 2023) and only isolated trial programmes and agents yielded positive results.First, at least for conditions that also occur in adults, the drug development pathway tests novel compounds and mechanisms of action in adult populations first.Thus, only agents that were successful/reached regulatory approval in adults are generally tested in children and/or adolescents.Second, mental health conditions in children and adolescents may be developmentally sensitive (Welsh et al., 2020).This creates the possibility that interventions provided outside a specific neurobiological window may not be efficacious (Díaz-Caneja et al., 2021).Third, due to age or neurobiological impairments that encompass language and communication as well as cognitive skills, young individuals with (certain) mental disorders may have difficulties recognizing, describing, and expressing the targeted psychopathology.Here, information from multiple informants may be helpful but also complicates the assessment process (Kraemer et al., 2003).Fourth, while rising placebo effects have plagued all of psychiatry (Correll, 2022), this problem may be enhanced in paediatric mental health RCTs, even more so in children than in adolescents (Parellada et al., 2012;Siafis et al., 2020) (Faraone et al., 2022).
Our study also informs research governance and reporting practices in the field.We found that 28% of the included RCTs were completed, but their results were not reported.While it is plausible that the effects of the COVID-19 pandemic impacted on the reporting of RCTs, we also found some RCTs for which results were labelled as "not available" in clinicaltrials.govhad indeed been published in articles in peer-reviewed journals.Therefore, we urge authors to promptly update the RCT record in clinicaltrials.gov.Additionally, for some RCTs, mean and standard deviation values for each arm were reported, but not the results of statistical significance tests.Importantly, in the majority of studies, only p values -which are dependent on sample sizewere reported, rather than standardized effect size. and their 95% confidence intervals.We would urge for more consistent reporting of ES in this field as this would facilitate comparison across RCTs of studied or already available treatment options.This would be more clinically meaningful than solely reporting p-values.Finally, discontinuation of clinical trial programmes and abandonment of compounds should be publicly communicated, alongside the rationale for this.
Our study should be considered in light of some limitations.First, we limited the search from 2010, so we might have missed relevant RCTs registered before this date.However, we deemed a 12-year period as appropriate to retrieve novel agents potentially available for regulatory approval and of interest in day-to-day clinical practice.Second, we may have included agents in this review whose further development has been discontinued by the sponsor without making this decision public.Third, we excluded RCTs recruiting both children/adolescents (until the age of 17) and adults, as separate results for children and adolescents are usually not reported in https://clinicaltrials.gov/ or https://www.clinicaltrialsregister.eu/.Fourth, while we covered a broad range of mental health conditions, our selection did not address all conditions that practitioners could be faced with.More specifically, substance use disorder and enuresis were beyond the scope of this review, the former occurring in an age range overlapping with adulthood and the latter being dealt with more frequently by paediatricians than child and adolescent psychiatrists.Fifth, we included RCTs in which investigated agents were combined with psychotherapeutic or other nonpharmacological interventions, where possible synergistic effects between pharmacological and non-pharmacological interventions cannot be ruled out.However, the combination of pharmacotherapy with psychosocial interventions is guideline-consistent for many, if not most, conditions (e.g.depression (NICE, 2019a), ADHD (NICE, 2019b), and schizophrenia (NICE, 2013).Sixth, to be comprehensive and provide information about potentially promising agents, we included information on agents sponsored and studied by universities and hospitals, that may not be subjected to the lengthy and costly trial requirements for regulatory approval and, thus, may not achieve marketing authorisation for clinical use.Seventh, we limited the search to two databases (https://clinicaltrials.gov/and and https://www.clinicaltrialsregister. eu/) as we could not include every national database.Eighth, we endeavoured to identify RCTs funded and not funded by drug companies, but where the listed sponsor was a public body this does not automatically equate with a drug company not being involved.Finally, for a sizeable proportion of RCTs, results of statistical analyses on tolerability were not available so we could report only the % of participants in each study arm that experienced serious adverse events or who discontinued the trial due to adverse events.
Despite these limitations, we believe that this review will inform researchers and funders of future priorities and opportunities in the field, and practitioners, patients and their families of possible future treatment options.Alongside drug manufacturers, we hope these findings will be informative also for public funders, fostering their collaborations with academia and research institutes in the field.We also hope there will be additional, well designed RCTs in anxiety, bipolar disorder, disruptive behavior disorders, eating disorders and schizophrenia, for which the number of RCTs is still limited.In this respect, regulatory efforts to promote extrapolation, i.e., the use of relevant information in adults as a basis for the further development of a medicinal product in children or adolescents, are welcome.Indeed, extrapolation has the potential not only to inform better studies in children and adolescents, but also to avoid unnecessary ones.Whilst this review has focused on RCTs, we deem it essential for funders to also support large scale pharmacovigilance studies with the potential to reduce risk of harm.While such studies will likely be expensive, they should be a priority for research funders, given the relevance and impact of their findings.

Conflict of interest
S Cortese declares honoraria and reimbursement for travel and accommodation expenses for lectures from the following non-profit associations: Association for Child and Adolescent Central Health (ACAMH), Canadian ADHD Alliance Resource (CADDRA), British Association of Pharmacology (BAP), and from Healthcare Convention for educational activity on ADHD.M Højlund has been a consultant to or has received honoraria from the Lundbeck Foundation, H. Lundbeck, and Otsuka.C

Fig. 4 .
Fig. 4. Summary of RCTs in children and adolescents compared to similar conditions in adults.Notes: Data on the number of RCTs in adults from Correll et al. (2023).Legend: ANX=Anxiety disorders (other than OCD); BD=Bipolar disorder; DD=Depressive disorders; OCD=Obsessive-compulsive disorder; RCT= Randomized controlled trial; SCZ=Schizophrenia.

Table 1
Retrieved RCTs with positive or negative findings for attention-deficit/hyperactivity disorder and anxiety disorders.

Table 1
S.Cortese et al.

Table 2
Retrieved RCTs with positive or negative findings for autism spectrum disorder.
Efficacy: No change from baseline to week 11 in Social Responsiveness Scale (SRS) (continued on next page) S.Cortese et al.

Table 2
(continued ) Words Subtests from baseline to weeks 12 and 24, and (B) change in Expressive Vocabulary Test (EVT) from baseline to weeks 12 and 24 (continued on next page) S.Cortese et al.

Table 2
(continued ) (continued on next page) S.Cortese et al.

Table 2
(continued ) (continued on next page) S.Cortese et al.

Table 2
(continued ) (continued on next page) S.Cortese et al.

Table 3
Retrieved RCTs with positive or negative findings for bipolar disorder, depressive disorders and eating disorders.
Results in doi: 10.1542/ (continued on next page) S.Cortese et al.

Table 3
(continued ) (continued on next page) S.Cortese et al.

Table 3
(continued ) (continued on next page) S.Cortese et al.

Table 4
Retrieved RCTs with positive or negative findings for intellectual and developmental disability, obsessive-compulsive disorder, schizophrenia, Tourette's syndrome, and PTSD.

Table 4
(continued ) (continued on next page) S.Cortese et al.