Neonatal Hypoglycemia: Updates on Diagnosis and Management Guidelines

Timely, effective and safe treatment of mental disorders is a key focus in medicine, due to their severity, chronicity and effects on multiple biopsychosocial aspects of life. Physicians, patients, families, and society have a substantial interest in the availability of new therapeutic options with greater efficacy, specificity, and tolerability compared to already available agents.

Psychopharmacology is a symptom-based discipline. Systems for classifying mental illnesses consist of often concurrent and/or connected patterns of symptoms, which are elevated to disorders when they cause distress or dysfunction and are not due to the effects of a substance or medical condition. This classification is not related to the underlying biology of the identified disorders. Comorbidities are common and medications often do not work in a substantial number of people and/or have pleiotropic and nonspecific effects, working for more than one disorder.

Pharmacological nomenclature has remained arcane, being only rarely or incompletely related to the mechanisms of action of medications. For example, antipsychotics are approved for diverse indications such as schizophrenia, bipolar mania, bipolar depression, major depressive disorder (MDD), tics, autism, anxiety, insomnia, agitation/aggression, obsessive-compulsive disorder (OCD), and post-traumatic stress disorder. (PTSD). Likewise, antidepressants are used for depressive disorders, anxiety disorders, OCD, and insomnia, among other conditions.

At the same time, medications often do not affect a single biological system, but instead have a variety of biological effects that may depend on the dose. A drug initially marketed for a given disorder may automatically receive a black box warning when it becomes indicated for another disorder, even though the safety data motivating that warning apply to a pharmacologically different class of drug, and have not been described the risk for that medication. This carryover effect occurred, for example, for dopamine receptor blockers and dopamine partial agonists with respect to suicide risk, when they received regulatory approval from the FDA for major depressive disorders, although the relevant data (possibly medication-related) in adolescents and young adults were restricted to traditional "antidepressants." Neuroscience-based nomenclature has been to some extent useful in trying to refine pharmacological terminology.

At the heart of state-of-the-art testing of the risks and benefits of a new molecular entity in psychopharmacology are parallel-group randomized controlled clinical trials (RCTs). However, multiple obstacles in design and conduct can interfere with the development of promising drugs in phase 1 and 2 trials, when they are tested in increasingly larger phase 3 studies. These failures may be related to the true ineffectiveness of a drug, its toxicity profile, poorly understood dose-response relationships, patient factors, and limited knowledge of the biological mechanisms underlying mental disorders, which prevent the identification of potentially relevant subgroups. An additional factor is the increasing placebo response in multiple mental disorders, the reasons for which are not sufficiently understood.

The most recently approved drugs target the melatonin, orexin, GABA-A, opioid, and N-methyl-D-aspartate (NMDA) receptor systems, vesicular monoamine transporter-2 (VMAT-2), and inverse agonism. of 5-HT2A receptors. Additionally, there is currently a renaissance of psychedelics, attempting to isolate their beneficial effects from short- or long-term risks or addictive potential. However, there is great concern because many of the drugs under study with new mechanisms of action cannot pass the “valley of death” of phase 2 and, especially, phase 3 of development.

This paper provides an overview of medications with innovative mechanisms of action that are in phase 2 or 3 testing for the treatment of a mental disorder in adults, highlighting the most promising (by superiority over placebo, magnitude of effect observed, safety and tolerability). The methodology, design and conduct of clinical trials considered when developing and testing pharmacological agents for the treatment of mental disorders in humans are also discussed.

> Schizophrenia . Of 176 phase 2 or 3 trials identified for the treatment of schizophrenia, only 12 molecules evaluated in 42 trials have so far outperformed placebo.

For total symptoms, a 5-week phase 2 trial showed that KarXT (a fixed combination of the muscarinic M1/M4 agonist xanomeline plus the non-centrally acting anticholinergic trospium chloride), administered twice/day, outperformed placebo, without relevant cardiometabolic or neuromotor adverse effects, but with some moderate and mostly limited anticholinergic adverse events. A second phase 3 trial of KarXT monotherapy versus placebo is ongoing, as well as a 6-week dose-escalating trial in patients with residual positive symptoms.

In a small 6-week phase 1B study, emraclidine, a positive M4 allosteric modulator, was also separated from placebo at doses of 20 mg bid and 30 mg qd. Results are being followed in two 6-week phase 2 trials testing 10 mg and 30 mg/day, as well as 15 mg and 30 mg qd versus placebo.

Ulotaront, an agonist of the TAAR-1 (trace amine-associated) and 5-HT1A receptor system, outperformed placebo in a 4-week phase 2 study in patients ≤ 40 years of age with schizophrenia and with no more than two previous hospitalizations due to its exacerbation, without risk of relevant neuromotor or cardiometabolic adverse effects. Three additional placebo-controlled trials with patients ≤65 years of age are ongoing. Additionally, ralmitaront, a TAAR-1 partial agonist, is in phase 2 testing.

Brilaroxazine, a D2, D3, D4, 5-HT1A, 5HT2A partial agonist and 5-HT2B, 5-HT6, and 5-HT7 antagonist, was superior to placebo in a 4-week phase 2 trial, and was recently initiated phase 3. Two phase 3 trials were completed for a new monthly and bimonthly extended-acting subcutaneous formulation of risperidone, TV-46000, confirming the efficacy of this drug in the acute treatment and relapse prevention of schizophrenia.

Raloxifene, an estrogen receptor modulator, improved the Positivity and Negativity Syndrome Scale (PANSS) total score and overall symptoms in a phase 3 trial in postmenopausal women with schizophrenia, but another phase 3 trial showed lower efficacy compared to placebo. Melatonin also improved PANSS total scores more than placebo in a phase 2 trial.

For positive symptoms (co-primary outcome), a phase 2 trial showed that cannabidiol outperformed placebo after 6 weeks of treatment. While there was a significant difference for Clinical Global Impression-Severity (CGI-G), cannabidiol was not superior to placebo with respect to total symptoms. Estradiol outperformed placebo on PANSS positive symptoms after 8 weeks of treatment in a phase 2 trial.

For negative symptoms of schizophrenia, the 5-HT2A inverse agonist/antagonist pimavanserin, approved for Parkinson’s psychosis and under review for dementia psychosis, showed positivity in a phase 2 study with respect to the primary outcome (total change in the Negative Symptom Assessment-16 [NSA-16]), but no improvement versus placebo on ICG-G and other negative symptom assessment scales. A phase 3 trial in schizophrenics with psychotic symptoms did not report improvement in total symptoms with adjuvant pimavanserin, but in negative symptoms and CGI-S of the total sample.

Roluperidone, a 5-HT2A and sigma-2 receptor antagonist, was successful in a phase 2 trial for negative symptoms. The subsequent phase 3 trial was suggestive of efficacy, but statistical significance was lost vs. placebo in the intention-to-treat analysis. A possible complication of this drug is that it has been tested only as monotherapy, with no documentation that it is effective on total and positive symptoms.

Regarding cognitive dysfunction in schizophrenia, a phase 3 clinical trial followed up on a successful phase 2 study with BI 425809, a glycine transporter-1 inhibitor, which outperformed placebo at week 12, but not in SCoRS, which measures the functional impact of cognitive improvement, is a requirement for regulatory approval.

Currently the most promising targets for schizophrenia appear to be muscarinic M1/M4 receptor agonism, muscarinic-positive M4 allosteric agonism, TAAR-1 agonism, and dopamine-serotonin partial agonism/serotonin antagonism. Due to mixed/inconclusive findings, questions remain regarding 5-HT2A agonism/reverse antagonism for negative and residual psychotic symptoms, 5-HT2A/sigma-2 antagonism for negative symptoms, and insulin transporter-1 inhibition. glycine to improve cognitive dysfunction and obtain regulatory approval.

> Bipolar disorder. For the treatment of bipolar disorder, only six molecules tested in 11 trials outperformed placebo in primary outcomes.

For bipolar depression , N-acetylcysteine ​​(a glutathione precursor) plus aspirin, added to usual care, outperformed placebo in a phase 2 trial. Additionally, nonracemic amisulpride (SEP4199) was superior to placebo at 6 weeks on the Montgomery Asberg Depression Rating Scale (MADRS). Adjuvant armodafinil, a renantiomer of modafinil, was associated with a significantly greater reduction in the Inventory of Depressive Symptomatology (IDS-C) total score at week 8 in a phase 3 trial vs. placebo, but two other trials did not confirm this superiority.

D-cycloserine (an NMDA antagonist) plus lurasidone outperformed lurasidone plus placebo after an initial infusion of ketamine in reducing depressive symptoms in patients with bipolar disorder. Additionally, adjuvant infliximab – a tumor necrosis factor alpha (TNF-α) inhibitor – was superior to placebo with respect to depressive symptoms in a phase 2 trial, but with no difference in treatment response. Ketamine outperformed placebo in a phase 2 trial targeting suicidal ideation.

> Major depressive disorder. 19 molecules were tested in 43 trials that outperformed placebo in results.

Cariprazine, a dopamine D3/D2 partial agonist with antagonist activity at 5-HT2B and 5-HT2A receptors, is currently under FDA review as a potentiator in major depressive disorder, following a positive phase 3 trial. Lurasidone, a 5-HT2A/D2 antagonist with 5-HT7 antagonism, was superior to placebo in a phase 3 trial in subjects with MDD and mixed features. The extended-release (LP) formulation of levomilnacipran, a serotonin and norepinephrine reuptake inhibitor, outperformed placebo in a phase 3 trial, although it was not superior to quetiapine plus antidepressants in another trial. Pimavanserin, a 5-HT2A antagonist/inverse agonist, had a positive phase 2 design as an enhancer in major depressive disorder, followed by a negative standard phase 3 study compared to placebo.

With the approval of intranasal esketamine and the off-label use of IV or intranasal racemic ketamine for resistant depression, psychopharmacology has renewed the development focus on antidepressant therapies that modulate the glutamatergic system. One such agent is AXS-05, a combination of dextromethorphan with low doses of bupropion, with non-competitive NMDA receptor antagonism, sigma-1 receptor agonism, nicotinic acetylcholine receptor antagonism, and inhibition of serotonin, noradrenaline, and serotonin transporters. dopamine.

In two phase 2 trials, AXS-05 was superior to low-dose bupropion or placebo in MADRS at week 6, leading to FDA approval for MDD. For the treatment of resistant depression, AXS-05 significantly delayed time to relapse (primary outcome) and decreased the relapse rate (secondary outcome) in a 12-week study. Another antiglutamatergic agent is esmethadone, an NMDA receptor antagonist with very weak opioid agonism, which is being developed as an potentiating agent in treatment-resistant depression. The Phase 3 program is ongoing, with three 4-week placebo-controlled studies. A single dose of rapastinel, an NMDA partial agonist, was superior to placebo when given at 5 or 10 mg in a phase 2 trial, but phase 3 trials were negative.

There has also been great interest in GABAergic modulation for the treatment of depression. Orally administered zuranolone, a neuroactive steroid that binds to GABA-A receptors, is being developed for both postpartum depression and MDD in phase 3 studies, with encouraging results.

Pioglitazone, a gamma receptor agonist, together with citalopram plus chlordiazepoxide was superior to placebo in a phase 2/3 study. Naltrexone, an opioid receptor antagonist, was superior to placebo in a phase 2 trial for preventing relapse or recurrence of symptoms based on MADRS. The combination of buprenorphine, a kappa opioid agonist, with the mu opioid antagonist samidorphan as an adjuvant treatment in MDD was superior to placebo in two trials, but not in three other phase 3 trials.

Ezogabine, which induces the opening of neuronal voltage-gated potassium channels, was superior to placebo in MADRS in a phase 2 trial. Botulinum toxin type A neurotoxin complex, an inhibitor of acetylcholine release, was superior to placebo in a phase 2 trial. The anesthetic nitrous oxide was superior to placebo at 24 hours in a phase 2 study, and at 2 hours, 24 hours, and 1 week in another phase 2 trial.

Psychedelics are also being investigated, with positive findings in phase 2 trials for ayahuasca (partial 5-HT2A agonism, affinity for other 5-HT receptors, TAAR-1 and sigma-1 agonism) and psilocybin (5-HT2A agonism).

The combination of metformin and fluoxetine (selective serotonin reuptake inhibitor) was superior to placebo plus fluoxetine in a phase 1/2 trial. Finally, transdermal estradiol plus intermittent micronized progesterone was more effective than placebo in preventing the development of significant depressive symptoms in initially euthymic perimenopausal women and in postmenopausal women in a phase 2/3 study.

> Disorders related to anxiety and trauma . Only nine molecules tested in 31 trials outperformed placebo in results.

In PTSD, intranasal oxytocin was more effective than placebo on amygdala connectivity in a phase 2 trial, and 3,4-methylenedioxy-methamphetamine (MDMA)-assisted psychotherapy (via serotonin and norepinephrine release ) was superior to placebo on symptoms in four phase 2 trials and one phase 3 trial.

In panic disorder, d-cycloserine (NMDA co-agonist) as a therapy enhancer outperformed placebo in neurocognitive processing in a phase 2 trial. In social anxiety disorder, a phase 2 trial showed that d-cycloserine as a cognitive behavioral therapy (CBT) enhancer outperformed placebo, although two other studies were negative.

In generalized anxiety disorder, ABIO 08/01 (a selective inhibitor of GABA- and glutamate-activated chloride channels) outperformed placebo in a phase 3 study. Agomelatine (melatonin receptor agonist) was superior to placebo in relapse rate and in the control of anxiety symptoms in phase 3 studies. Pregabalin (voltage-gated calcium channel modulator) was more effective than placebo on anxiety symptoms in two phase 3 trials.

Quetiapine extended-release was superior to placebo in two phase 3 trials on anxiety symptoms and relapse rate. Finally, SR58611A (selective beta-3 adrenergic receptor agonist) reduced anxiety symptoms more than placebo in a phase 3 trial, and vortioxetine (multimodal serotonergic modulator) prevented relapse in a phase 3 trial. In particular , no promising treatment was identified for OCD.

> Substance use disorders. Many agents that outperform placebo in Phase 2/3 clinical trials are drugs already approved for another indication.

For alcohol use disorder , these include baclofen (GABA agonist) in a phase 3 trial on time to decline and relapse and percentage of abstinent participants; gabapentin (voltage-gated calcium channel modulator) in a phase 2 trial in relation to excessive alcohol consumption; ibudilast (phosphodiesterase 4 inhibitor and toll-like receptor-4 antagonist) in a phase 2 trial similar to the previous one; and ketamine (NMDA antagonist) in a phase 2 trial on days of abstinence.

For methamphetamine use disorder, mirtazapine (alpha-2-adrenergic antagonist, histamine-1, 5-HT2A/C, and 5-HT3), and the combination of naltrexone (opioid antagonist) and extended-release bupropion (inhibitor of norepinephrine-dopamine reuptake, nicotinic receptor antagonist, non-selective serotonin reuptake inhibitor and sigma-1 receptor agonist), were positive in phase 2 trials vs. placebo on the number of urine samples positive for substances. In amphetamine use disorder, sustained-release methylphenidate (DA and NA reuptake inhibitor) reduced the number of substance-positive urine samples in dependent individuals with comorbid attention-deficit/hyperactivity disorder in a phase trial. 2.

For cocaine use disorder , drugs that outperform controls include AFQ056 (metabotropic glutamate receptor antagonist) in the proportion of days of cocaine use in a phase 2 study; ketamine (NMDA antagonist) on motivation to quit cocaine and cue-induced craving in a phase 2 trial; and zonisamide (voltage-sensitive sodium channel blocker and allosteric GABA receptor agonist) in a visual analogue questionnaire in a phase 1/2 trial.

For nicotine use disorder , the combination of zonisamide plus varenicline was superior in self-reported smoking and nicotine withdrawal, but not biochemically verified smoking, in a phase 1/2 trial. For opioid use disorder, positive results are available for cortisol in users with low, but not medium or high, daily heroin use in a phase 2 trial.

It should be considered that the positive results mainly involved drugs already marketed for other disorders, while new mechanisms of action have yielded much less positive results, despite continued efforts.

> Dementia. Fourteen molecules outperformed placebo in the primary outcomes of 15 trials.

Among trials targeting cognition or disease-modifying markers, positive phase 2 trials included those investigating acitretin (retinoid of MAPkinase), ORM12741 (selective alpha-2C adrenoceptor antagonist), sargramostim (granulocyte-macrophage colony-stimulating factor), and rasagiline (monoamine oxidase-B inhibitor).

Among trials aiming to improve behavioral and psychiatric symptoms in dementia, brexpiprazole, a dopamine partial agonist; dextromethorphan/quinidine, a sigma-1 agonist/NMDA antagonist/multimodal agent; and CB1/2, a nabilone partial agonist improved agitation. Additionally, AVP-786 (deuterated form of dextromethorphan/quinidine) improved agitation in a phase 3 trial.

Two orexin 1 and 2 receptor antagonists – lemborexant and suvorexant – improved restlessness and sleep. AXS-05, a combination of dextromethorphan with low-dose bupropion, was superior to placebo in agitation in a phase 2/3 trial. Pimavanserin, a 5-HT2A receptor antagonist/inverse agonist, with lower activity as a 5-HT2C antagonist/inverse agonist, outperformed placebo for relapse of dementia-related psychosis in a phase 2 and a phase 3 study .

Most promising drug targets for the treatment of the core features of dementia, and in particular Alzheimer’s disease, are difficult to predict. There is little evidence available to suggest that ongoing trials will be successful.   

The above summary of trials of new pharmacotherapies indicates that a large number of potentially useful chemical entities and mechanisms of action are being tested. Many, if not most, of these programs will not produce an approved drug that can be used in clinical care. The following is a discussion of the challenges, opportunities and possible solutions regarding methodology, conduct and interpretation of clinical trials. This should help future research de-risk trials of new or known agents for new psychiatric indications, increasing the chance of success.

The randomized clinical trial (RCT) is the cornerstone of research aimed at obtaining regulatory approval for pharmacological agents.

It provides an accurate assessment of the efficacy/effectiveness and safety of a treatment in a given patient population.

The primary consideration in RCTs is validity , that is, minimizing the likelihood of misleading approval of an ineffective or unsafe treatment. Power is also important, that is, the likelihood of support if the treatment is truly effective and safe enough in that population to justify its clinical use.

The foundation on which each RCT is built is based on an a priori exploration , through a review of the literature in relation to the disorder or objective symptom of interest, and the treatments already available and their effectiveness and safety. Relevant results in animals, pre-post and case-control studies in patients, and post-hoc exploration of studies carried out are included. Finally, pilot studies can be conducted to evaluate the feasibility or viability of the strategies considered for the proposed RCT.

This sequential process is necessary for three reasons. Firstly, it allows the formulation of the a priori hypothesis , that is, the statement of what exactly is expected from the RCT, which, if true, would lead to the approval of the drug. Second, it is unethical to randomize patients unless researchers are in "clinical equilibrium," that is, there must be empirical justification for thinking that the hypothesis may be true and important, but also reasonable doubt as to whether it is true. or not. Third, the best option for each of the decisions depends on what is known from an a priori exploration . The more information that guides the design of the RCT, the greater the validity and power of the RCT.

Several aspects of the trial design may affect the chances of finding significant differences between the active and control arms. Trials must be “adaptive by design.” Early learning trials (e.g., minimally effective dose or toxicity) are often necessary before confirmatory trials, which are used for regulatory approval of drugs. Early trials need greater control for type II errors (false negatives), and less control for type I errors (false positives), which are crucial in phases 2 and 3 trials.

One aspect that can be adapted in terms of design is the dose of the drug. It is usually established a priori, and tested in different arms, with many patients exposed to doses of drugs that are not effective, and not necessarily safe. It is important to identify the optimal dose of a drug as early as possible in an RCT. The continuous reassessment method is a Bayesian approach that leverages dose-response curves to identify the maximum tolerated dose (MTD), allowing the dose to be rapidly established around the MTD during the early stages of the study.

A second aspect that can be adapted is randomization. Covariate adaptive randomization can be applied, which randomizes allocation within matched levels of putative prognostic factors.

A potentially key adaptive element is sample size. This should be as large as possible to ensure sufficient statistical power to avoid type II error, and has to consider dropout rates, as well as associated costs and duration, which increase linearly with the number of people to be recruited.

A fourth aspect that can be adapted by design is to reduce population characteristics, to identify subgroups of patients who benefit from a treatment. This “enrichment” design has great value in the later stages of learning, and is consistent with the concept of precision medicine. Its main disadvantage is that it produces findings that are not generalizable and can hinder their replicability in the real world.

Enriched sample selection can also be useful for rapid trials where the data are used to make a decision on whether and how or in whom to continue the drug development process of a given molecule.  

However, it is unclear to what extent sample and effect size calculations need to be adjusted by expanding the population to be more inclusive and less enriched. Enrichment designs may base their randomization on prior response, as occurs in trials conducted in stabilized patients who are randomized to continue the study drug or switch to placebo.

The duration and degree of stability and relapse rates with placebo are important considerations when designing such trials, as shorter durations and less effective remissions increase the likelihood of relapse, particularly in the placebo arm. An additional strategy for randomizing patients is to have an initial phase with single-blind placebo, open-label medication, or double-blind placebo, basing randomization on response during this initial phase.

In the pre-inclusion stage, patients are treated with placebo, and then only those who do not respond to placebo are randomly assigned to placebo or active treatment. However, considering costs, sample size, and trial duration, the sequential parallel design may be more effective for phase 3 trials seeking regulatory approval.

Sequential parallel comparison is a study design that attempts to overcome limitations of the initial placebo stages. The trials are structured in two stages, and can be carried out with one randomization, if the trial has two arms, or two randomizations if three arms are used (one active, two placebos). Participants are first randomized to placebo (stage 1). Non-responders are then rerandomized to the two treatment options (stage 2), in case of two-arm trials, or to the active arm or the two placebo arms in 3-arm trials. The data from the first and second randomization are analyzed and grouped in the same analysis generating a p value. With this design it is possible to maintain the same power level with 20% to 50% fewer people.

Finally, “seamless adaptive designs” are trial designs that attempt a multi-phase trial. This design can reduce the time from phase 1 to phase 3 toward regulatory approval, implementing continuous recruitment, intense monitoring, and data analysis that can inform adaptive dosing, randomization, and sample size. However, there may be a risk of type 1 error in this type of design.

Despite adaptive designs, trials often fail. The worst case scenario, which is far from rare, is to recruit a large number of participants, exposing them to experimental drugs, with potential safety issues and significant costs, without observing significant differences between medication and placebo. Stopping for futility is an important design that allows trials to be terminated prematurely as soon as there is no evidence of any significant effect of the interventions versus the control. 

While factors driving the placebo effect have the potential to identify safe therapeutic elements that can be exported to clinical care, high placebo response is a plague that plagues RCTs of different mental disorders. In fact, it has been suggested that some pharmaceutical companies have decreased their investment in developing drugs for mental disorders due to challenges in detecting signals from higher-than-expected placebo responses.

This practice has been questioned by the growing reluctance of doctors and patients to participate in such studies. Furthermore, ethics committees in many countries are making it increasingly difficult to conduct placebo-controlled clinical trials. In recent years, large attrition rates have been observed in trials using placebo controls, as well as a decrease in the drug-placebo difference, driven largely by the increase of the placebo effect to similar degrees as the effects of drugs.

The magnitude of the placebo effect is largest in trials on depressive disorder, bipolar depression and mania, and smallest in trials on schizophrenia. However, the placebo effect has been increasing over the last 24 years, and is a major obstacle to the development of new medications. In fact, placebo response is particularly high in industry-sponsored trials. Having a large placebo response fatally reduces the chances of finding important differences with the experimental arm.

Several variables have been identified as consistently associated with an increase in the drug-placebo difference in different mental disorders. An open-label lead-in phase before double-blind randomization increases the placebo effect. A second factor is poor recruitment with invalid baseline assessment and case mix verification. On the other hand, more severe symptoms at the beginning of the study are associated with a lower response to placebo and a greater drug-placebo difference. 

Factors that have been shown to moderate the largest differences between drug and placebo in schizophrenia trials were smaller sample size, fewer study sites, less active study arms, more patients assigned to placebo, use of the Abbreviated Rating Scale Psychiatric (BPRS) instead of PANSS, longer previous period, longer duration of the study, shorter duration of illness and younger age.

The results of each clinical trial apply to the population represented by the sample, and not beyond. For ethical reasons, we cannot include those who are unwilling to give consent, or patients who are likely to be harmed by participation. 

Furthermore, the results of RCTs do not necessarily apply to all subgroups of the sampled population. If a highly effective treatment is shown in the sampled population, there may still be a minority subgroup in which the treatment is ineffective or toxic. If an RCT detects little or no difference between treatment and control, the population can be divided into two subgroups; In one, the treatment is more effective and safe, while in the other, the control is more effective and safe, canceling each other out in the entire population.

Patients included in schizophrenia trials are often not representative of the real-world population seen in practice. Furthermore, trial and population characteristics have changed over time; Patients with schizophrenia who are typically eligible have fewer physical and psychiatric comorbidities and fewer suicidal behaviors.  

This limited representativeness of phase 2 and 3 placebo-controlled trials also applies to other conditions, such as mood and substance use disorders, due to similarly restricted inclusion criteria and the fact that patients must be capable of giving informed consent. This limited representativeness emphasizes the importance of well-designed phase 4 studies that aim to test not if, but in whom and under what circumstances a drug acts.

Another relevant problem is the inflation of symptoms at the beginning. Symptoms vary over the course of an illness, and may be reactive to stressful stimuli, such as disruption of routine or new scenarios. Another explanation may be sites that recruit patients, which may produce, even unintentionally, higher baseline symptom scores.

Several strategies can be implemented to optimize patient representativeness, and reduce symptom inflation at baseline.

First, it is good to have medical records that document the recent past of patients who were not recruited from regular clinical care settings.

Second, relax the inclusion criteria to some extent, allowing participants with a certain set of physical or psychiatric comorbidities. This would facilitate recruitment, make the trial more pragmatic and clinically useful, potentially decrease placebo response, and allow for greater adherence to the principles of equity, diversity, and inclusion. Retention is also part of recruitment, meaning patients remain in the study. It is important to balance the desire for low attrition with the need for patient safety by allowing more rescue strategies within the study that are transitory and/or do not compromise the outcome.  

Trials are typically conducted at multiple sites, to allow timely recruitment of sufficiently large samples. However, this also has its issues. First, sites are frequently incentivized to recruit, which may lead to inclusion of inappropriate patients. The more sites participating in a trial, the greater the heterogeneity, the greater the likelihood of poor quality compliance with the trial procedure, and the more difficult quality control.

Early elimination of sites with poor recruitment or showing abnormal placebo response may mitigate the impact of this heterogeneity. Sites must be certified and strictly supervised, and it is preferable to have fewer sites, but with high efficiency and quality. Reliability training is almost always performed for interview ratings conducted by an expert, allowing for efficient rater training.

The introduction of new technologies holds great promise to make processes more reliable, continuous, applicable, and profitable. For example, language processing and analysis of voice and facial expressiveness could be very informative in conditions such as schizophrenia, mania and depression. This can minimize recall bias and maximize validity. The use of smartphones and other devices can provide objective information on geolocation, activity levels, frequency and timing of social interactions, sleep, and other measures of interest to clinical researchers.

Evaluators must administer scales and measures that are clinically relevant and widely used in the field. They should pay special attention at the time of evaluation, particularly, but not only, in relation to cognitive symptoms. The evaluation should ideally be repeated over time, to feed the analyzes with richer data. The availability of measures repeated over time also improves imputation, better protecting validity. However, mandating measures can sometimes erase these advantages by encouraging attrition and missing data.

It is desirable to have more than one result in a trial, since a single result can hardly provide a complete clinical picture. For secondary and exploratory, hypothesis-generating outcomes, and for those requiring large amounts of multidimensional data, modern tools including digital phenotyping and ecological momentary assessment can be of great value and should be progressively introduced into trial evaluation. These tools can be used multiple times and can even be continuous in cases of passive surveillance. It is unclear to what degree interactive digital phenotyping may affect placebo response.

Beyond secondary and exploratory outcomes that may be multiple, the most important issue, however, is multiplicity for primary outcome measures in an RCT. The objective of an RCT is to recommend one treatment over another in the sampled population: that is, a decision. With a primary outcome, the chance of a false positive with standard approaches is less than 5%. With two independent primary outcomes, the chance of one or more false positives is 10%; with three it is 14%, increasingly increasing the possibility of a misleading conclusion. If there is adjustment for multiple tests, using a significance level low enough for each result so that the chance of a false positive result is less than 5%, there is a loss of power, an increased risk of a failed assay, and yet, contradictory results in multiple tests.

An RCT should have only one primary outcome measure, but it can be composite. For example, a decrease in symptoms during treatment could be an acceptable outcome measure. However, if patients develop health problems due to treatment or monitoring, it is not a sufficient primary outcome measure. Ideally, an appropriate outcome measure should reflect a balance between benefit and harm. If there are several independent benefits and harms of interest, the outcome of the treatment is the cumulative effect on the patient of any of the benefits and harms experienced.

Finally, dichotomizing an ordinal outcome is always a poor choice. For example, if “success” were defined by a ≥50% decrease in symptoms, a patient with a 51% decrease in symptoms would have the same outcome as one with a 100% decrease, while a patient with a 49% decrease would be considered the same as one with 0% decrease. There is a significant risk of misclassification and a large loss of power with dichotomization.

The success of a trial, and the approval of a drug to treat a given disease, also depends largely on the results of statistical analyses. Therefore, it is essential to adopt appropriate statistical approaches that minimize the possibilities of type I and II errors.

Clinical trials are the cornerstone of today’s evidence-based medicine. Designs range from effectiveness trials with maximized internal validity but limited external generalizability, to large single trials that maximize validity but have reduced precision.

In the case of non-randomized trials , studies of large national databases can help generate hypotheses, but are insufficient to allow causal inferences to be made.

Data analyzes have evolved, and it has become increasingly important to choose the most appropriate statistical analysis plan for a given design, research question, or attempt to minimize type I and/or type II error.

In drug development for regulatory approval purposes, randomized, placebo-controlled, and parallel-design trials are primarily used. Placebo-controlled trials are included for the approval of treatments, as well as placebo substitution trials for the approval of maintenance interventions. Increasingly, an active control arm is included to test the integrity of the study, allowing us to distinguish between negative trials (the established medication separates from the placebo, while the experimental drug does not) and failed trials (neither the medication neither the experimental nor the established medication are separated from the placebo).  

Sample sizes have increased, especially in phase 3 trials, because of the disproportionate increase in placebo response with relatively little increase in drug response in recent decades. By focusing on outcomes beyond symptoms, such as quality of life and functioning in multiple relevant domains, medications mostly “only” prepare the brains of people with mental disorders to have the potential to function better, without putting their “capacity” increased or restored in action.

Digital tools can more reliably and objectively provide assessments of cognitive, academic, behavioral, and social functioning. 

In conclusion , the process of developing and approving new drugs is complex, and this complexity and the dangers related to its failure can even be improved when it comes to mental disorders. The information contained in this work aims to provide practical knowledge on topics related to the methodology and implementation of clinical trials that must be considered and weighed, with their relative pros and cons, serving as a roadmap that aims for the successful approval of new agents for the treatment of mental disorders.