Depressive disorders: Treatment failures and poor prognosis over the last 50 years

Abstract Depression like many diseases is pleiotropic but unlike cancer and Alzheimer's disease for example, is still largely stigmatized and falls into the dark shadows of human illness. The failure of depression to be in the spotlight for successful treatment options is inherent in the complexity of the disease(s), flawed clinical diagnosis, overgeneralization of the illness, inadequate and biased clinical trial design, restrictive and biased inclusion/exclusion criteria, lack of approved/robust biomarkers, expensive imaging technology along with few advances in neurobiological hypotheses in decades. Clinical trial studies summitted to the regulatory agencies (FDA/EMA) for approval, have continually failed to show significant differences between active and placebo. For decades, we have acknowledged this failure, despite vigorous debated by all stakeholders to provide adequate answers to this escalating problem, with only a few new antidepressants approved in the last 20 years with equivocal efficacy, little improvement in side effects or onset of efficacy. It is also clear that funding and initiatives for mental illness lags far behind other life‐treating diseases. Thus, it is no surprise we have not achieved much success in the last 50 years in treating depression, but we are accountable for the many failures and suboptimal treatment. This review will therefore critically address where we have failed and how future advances in medical science offers a glimmer of light for the patient and aid our future understanding of the neurobiology and pathophysiology of the disease, enabling transformative therapies for the treatment of depressive disorders.

from achieving this goal. Be that as it may, a recent publication in the Lancet, entitled "Comparative efficacy and acceptability of 21 antidepressant drugs for the acute treatment of adults with major depressive disorders: a systemic review and network metaanalysis," would both challenge this pronouncement and suggest we already have effective antidepressants. 2 This recent study and several others purport that SSRI's do work in depression (but only perhaps in subtypes? -see later discussion) and that some older second-generation antidepressants (eg, amitriptyline) showed greater efficacy, than many SSRI's. 2,3 Thus, questioning the need for the next generation of "better" antidepressants and the need to step forward from historical dogma, redundant clinical classification, to a new era of neurobiology, neuronal networks of depression and precision pharmacology with its focussed on diagnosis based science not symptoms. 4 Depressive disorders, in particular major depression disorder (MDD) is based on a 50-year-old monoamine hypothesis, questionable animal models and subjective clinical diagnostic criteria, with comorbidity across several neuropsychiatric disorders. 8,9 Thus, the focus of this review is to ask the question(s) again, why are there so many failures and why so few successes and do we need "better" antidepressants? In this review, I want to build on what we've learned from the past and how this may lead us to future clinical successes in the treatment of depressive disorders. First, we need to understand the complexity this neuropsychiatric disorder, the global crisis, unmet medical needs, its current diagnosis, treatment, and future areas of research.

| G LOBAL DEPRE SS ION CRIS IS-THE UNME T MED I C AL NEED FOR B E T TER ANTIDEPRE SSANTS?
Depression is a significant contributor to the global burden of disease and affects people in all communities across the world requir-

A recent World Health Assembly called on the World Health
Organization and its member states to take action in this direction. https://www.cdc.gov/nchs/data/databriefs/db283_table.pd#4).
It is estimated that the prevalence of depression in the US is 15% percent of the population reportedly taking an antidepressant-if not more. MDD is ranked fourth as a disease measured in disability adjusted for life years (DALYS) in 1990. 7,10 Together with the fact that available antidepressant medications are ranked second behind ischemic heart disease as a potential disease burden by 2020. The risk for MDD, especially for females in developed countries, is 1 in 10. And, there is considerable evidence that depression is associated with increased risk for cardiovascular and infectious diseases as well as immunological and endocrine changes. The World Health Organization predicted that depression will become the leading cause of human disability by 2020. 11 It has been estimated that over a lifetime, the global prevalence of depression is 21.7% for females and 12.7% for males who suffer from depression at some point.
Epidemiological studies have estimated that 5%-9% of women and 2%-3% of men in the US suffer from depression at any time. 8 And, a Norwegian study showed that 24% of women suffer major depression at some point in their lives and 13.3% suffer from dysthymia, while 10% of males suffer from major depression at some point, and 6% suffer from dysthymia. 8 Depression in children and adolescents is a cause of substantial morbidity and mortality in this population, being a common disorder that affects 2% of children and up to 6% of adolescents 12,13 . Although antidepressants are frequently | 3 of 20 used in the treatment of this disorder, there has been major controversy about the efficacy and safety of these medications in this population. 7 This led to the US food and Drug Administration (FDA) publishing a list of recommendations from the Psychopharmacologic Drugs and Paediatric Advisory Committees over the years, including many other neurological and psychotropic drugs.
This critical appraisal on the treatment of depression in children and adolescents is still an area of great concern and controversy in relation to the developing brain. Depression is a common condition with up to 8% of all teenagers having met criteria for depression in the last year. 14 In fact, by the age of 21 years, up to 14.8% of individuals have met criteria for a mood disorder. 13,15 Some types of depression are familial, indicating that there is inherited vulnerability. 16 Similarly, in studies of families in which members of each generation develop bipolar affective disorder (BPAD) it has been found that those with the illness have a different genotype from those who do not become ill. Conversely, the reverse is not true: not all individuals with a purported BPAD genotype will develop the illness (epistasis-mutations in one gene masks a phenotype at another locus). That in additional to other factors, stresses at home, work, or school or other coping skills, are involved in the onset of the disease. In some families, major depression also seems cooccur generation after generation. 17

A National Institutes of Mental Health (NIMH) National
Comorbidity Survey of more than 9000 US adults in 2005; using the

Diagnostic and Statistical Manuel of Mental Disorders-DSM-IV-TR
(Text Revision 2001) criteria, found that 6% of those studied had a debilitating mental illness, yet treatment was difficult to obtain, with only one-third or more of those in care receiving minimally adequate treatment, such as appropriate drugs or a few hours of therapy over a period of several months. In general, the investigators found that things had not changed much over the past decades and would argue that the situation has deteriorated further in recent years. In an earlier Massachusetts Institute of technology (MIT) survey the estimated direct and indirect cost of mood disorders in the US to be $43 billion in 1990. In a more recent MIT survey (including a wider range of disorders and costs, plus EU member countries), estimated the total in 2010 to be $780b, of which 60% was attributable to direct costs and 40% to lost productivity. 18,19 Depressed individuals incur twice the medical cost burden as nondepressed patients, the main part (80%) being for medical care rather than psychiatric or psychological services, with the bulk of antidepressant prescriptions (80% worldwide) being written by primary care physicians. (PCP's).
With up to 30% or more of patients with MDD who do not respond to typical antidepressant medications. 20 Alternative effective treatments for moderate-to-severe depression include a combination of somatic therapies (CDT, pharmacotherapy, repetitive transcranial magnetic stimulation (rTMS), transcranial direct-current stimulation (tDCS), and the more established electroconvulsive therapy (ECT). ECT has been rejuvenated for the treatment for the most severe, melancholic depressions, particularly in the elderly (who are more prone to adverse effects of drugs) and in approximately 30% of patients who do not respond to SSRI antidepressants (treatment resistant patients-see later for further discussion). However, patient accesses to alternative treatments are not only totally inadequate but limited to regional availability and cost. Thus, it's clear that a combination of genetic, developmental, psychological, and environmental, socio-economic factors contributes to the onset and suboptimal treatment of depressive disorders.

| CURRENT D IS E A S E S TATE AND D IAG NOS TI C CRITERIA
Depression is a very common medical condition that is associated with a wide range of emotional, cognitive, and physical symptoms.   (Table 1). 21,22 Although these classifications have varying degrees of overlap and distinguishing features, their goal is to try and accurately classify the burden of patients suffering from mental disorders. However, the ferocious rhetoric regarding previous and the more recent DSM-5 and International Classification of Disease (ICD-11) classification-promises and pitfalls is well documented with regard to the many flaws and discrepancies (see DSM-5 Pros and Cons.). 23,24 In spite of the fact of the many changes and improvements in DSM-5 and ICD-11 from their predecessors, they both remain subjective categorical classification systems that are fundamentally descriptive in nature, based primarily on self-reported symptoms, clinically signs with observer bias and few supportive tests (eg, of intellectual functioning). The fact that since the early 1980's, research bodies e.g. NIMH and other funding agencies had virtually mandated the use of DSM or ICD diagnostic categories was argued as a major part of the problem. The DSM "Bible" was seen as dictating US mental health questioning its validity and widely denounced. What was needed was innovative thinking away from symptomatology-based diagnosis to an alternative approach. In 2009 the NIMH initiated the Research Domain Criteria (RDoc) project was deemed necessary, given the nascent state of the science of mental disorders and the conceptual and empirical constraints of research based on current classifications. The call was that research needed to break out from the straitjacket of current diagnosis.
The development of basic translational science applied to depression and other mental disorders responded slowly to the difficulties of the categorical classification system and represented a long-term NIMH endeavour. What the NIMH RDoc initiative brought to the forefront was the idea that to understand mental illness in all its complexity, the neuroscience field needs a research framework that accommodates the study of all causal factors together. This was acknowledged to be a long-haul and there are no right answers that this framework will work. The notion that neural-circuit based framework will ultimately deepen our understanding of the neurological, biological, psychological, social and cultural structures, and processes that underlie depression and mental illness will ultimately lead to a move away from an out-dated, systematic biases clinical trial methodology. 8,25 Accordingly, Thomas Insel in proposing the NIMH's reorientation away from DSM categories stated, "We cannot succeed if we use DSM categories as the gold standard."

| MA JOR DEPRE SS IVE D ISORDER : DS M -5 SYMP TOMS OR ENDOPHENOT YPE S?
The symptom criteria for major depression according to the recent DSM-5 and ICD-11 guidelines are reported to be very similar although the coding systems are different.

The DSM-5 (Diagnostic and Statistical Manual of Mental
Disorders, 5 th Edition) has focussed on more attention genderspecific factors across disorders, cultural and cross-cultural assessments, Thus, the multi-axial system of psychiatric classification (ie, DSM-III, DSM-IV TF and ICD-10 see Table 1) is to be gradually replaced for all psychiatric and mental disorders that are now to be considered on a single axis. For example, in mood disorders the separation of bipolar and related disorders (BPAD) is a major change in diagnostic criteria and clinical descriptions forming a separate chapter for bipolar (affective) disorders (BPAD) in DSM-5 (see comprehensive reviews on BPAD in references 8,26,27 ).
In the case of depression, there are now 8 specific depressive disorders (single-axis) described in the DSM-5 (see below). With the aim of increasing the focus on these individual ("personalized") disorders, their severity, phenotypes/genotypes, and application of numerous specifiers to capture significant advances in clinical research, including advances in neurobiology and genetics. 28 Whereas, DSM-IV comprised of additional subcategories for catatonic, melancholic, and atypical features and for postpartum onset. Both DSM-IV and ICD-10 present affective disorders together in one section, distinguishing bipolar (BPAD) from unipolar disorder (MDD), including dysthymia (see Table 1). Operational antidepressants exert their action. 32 Thus, the hypothesis failures of the past may therefore represent a protracted learning curve resulting from past failures and a naive understanding of complex brain neurochemistry and multi-modal brain neurocircuitry. And, applying this to ill-defined diagnostic criteria within heterogeneous patient populations to unmask the so-called "final common pathway" for depressive disorders.
The hypothesis failures in neurobiological and clinical studies of the past will be reviewed in the next section.
TA B L E 2 Abridged DSM-IV criteria for major depressive episode A. Over the last 2 weeks, of the following features should be present most of the day, or nearly every day (must include 1 or 2): 1. Depressed mood 2. Loss of interest or pleasure in almost all activities 3. Significant weight loss or gain (more than 5% change in 1 month) or an increase or decrease in appetite nearly every day 4. Insomnia or hypersomnia 5. Psychomotor agitation or retardation (observable by others) 6. Fatigue or loss of energy 7. Feelings of worthlessness or excessive or inappropriate guilt (not merely self-reproach about being sick) 8. Diminished ability to think or concentrate, or indecisiveness (either by subjective account or observation of others) 9. Recurrent thoughts of death (not just fear of dying), or suicidal ideation, or a suicide attempt, or a specific plan for committing suicide B. The symptoms cause clinically significant distress or impairment in functioning. C. The symptoms are not due to a physical/organic factor or illness.
The symptoms are not better explained by bereavement (although this can be complicated by major depression)

| HYP OTHE S IS FAILURE: WHAT WE HAVE LE ARNT FROM THE PA S T, IF ANY THING?
The etiology of depression is unknown. Depression is polygenic in nature with both genetic and epigenetic components, making the use of genetically engineered animal as models for drug discovery unrealistic. 8,9 This along with our emerging understanding of the complex biochemical mechanisms is compromised by the fact that most of the drugs used to treat depression and other neuropsychiatric disorders (eg, lithium and antidepressants in general) have ill-defined pleiotropic mechanisms of action with new signaling pathways and neuronal networks being identified, pointing to no "final common pathway" in the mode of action of antidepressant agents

| THE MONOAMINE THEORY OF DEPRE SS ION-OF LIMITED SUCCE SS OR A DIS MAL FAILURE?
The longest-standing theory of depression is based on monoamine dysfunction and drugs acting on monoamine neurotransmission which has dominated the treatment of depression for over 50 years, albeit much maligned in recent times as a too simplistic and may have misguided our understanding of the complexity of the disorder. 8,32 The fact remains, however, that the monoamine reuptake inhibitors and the MAOI's were shown to have antidepressant activity albeit by chance clinical observations and the discoveries of their modes of action were instrumental in developing the monoamine theory. 8 In the days when the monoamine theory of depression was evolving, the focus was more on norepinephrine (NE) than 5-HT Over 4 decades the therapeutic goal was to find, a fast-acting antidepressant. However, this was contended by Duman and a number of groups, that this approach may not be possible based on their neurogenesis hypothesis of antidepressant efficacy. 32,34 To discover an antidepressant that has an effect within days rather than weeks has challenged researchers for decades to understand the reasons for the delay in onset of the antidepressant action. One theory based on the action of SSRI's is that inhibition of 5HT reuptake initially causes activation of the presynaptic 5HT 1A receptors on the cell bodies in the dorsal and median raphé nucleus. 8,9 This inhibits the firing of 5-HT neurons, so reducing rather than increasing the release of 5HT from the terminals. 8,9 According to this hypothesis first proposed as the primary mechanism of action of SSRI's due to an increased ac- as there is no clear evidence that the monoamine deficiency totally accounts for depression and questions the efficacy of monoaminebased agents. 8,32 The question remains is there a single unifying mechanism underlying the complex manifestations of depression (MDD)?
In the case of MDD, genetic factors account for about 30% of the variance and environmental factors play a major role in inducing the illness. 36 The first direct evidence of the importance of variation in drug response was shown in depressed patients with a short form of the SERT promoter, who had a worse response to SSRI's than those with the long isoform. 8,32 Other genes have been associated with antidepressant treatment and undoubtedly the field of pharmacogenomics and its application to the pathophysiological mechanisms of depressive disorders will continue to grow based on vulnerability gene environment interaction and experience-dependent biological systems that act cumulatively (eg, chronic stress) throughout an individual's lifetime. 29 That being said, the impact of genetics on mental disorders over the last few decades have been disappointing, despite the enthusiasm for new era of personalized medicine and an individual's genome. However, emerging results as discussed later may offer hope for future drug therapy based on endophenotype. 29

| THE FAILURE OF E XPERIMENTAL DIS E A S E MODEL S OF DEPRE SS IVE DISORDER S?
Drug discovery in depression has been hampered by the lack of an universally accepted phenotypic screens-animal model(s) that can be used to screen NCEs for antidepressant -like effects. Animal models of depression have provided insights into mechanisms associated with MDD endophenotypes but how these models apply to human mental illness and its treatments remains difficult to assess.
Although there are several animal models that reproduce some features of depression in the context of stress and/or maternal separation, it is questionable as to whether these are relevant to the human disorder MDD or BPAD. The advantages and disadvantages of animal models for depression are summarized in various comprehensive reviews. 8,37 However, in many cases, the behavioral features can be reversed by conventional antidepressant drug treatment. Despite this holistic notion and their intrinsic limitations, the full potential of these models has not yet been realized and they represent an underexplored opportunity. The heuristic value and the knowledge gain from behavioral animal models in psychopharmacology are, explicitly or implicitly, the central preoccupation of psychopharmacologists. 8,9 There are a number of compelling reasons to believe in the legitimacy of animal models in the development of new improved drugs for the treatment of mental disorders; however, these models need to be based on the following criteria. 8,9,33 • Predictive validity: the ability of a model to accurately predict clinical efficacy of a psychoactive pharmacological agent.
• Face validity: the similarity of the model to clinical manifestations of phenomenon/disorder in terms of major behavioral and/or physiological symptoms and etiology.
• Construct validity: the strength of the theoretical rationale upon which the model is based Animal models have been defined as experimental preparations developed in one species for the purpose of studying or understanding a phenomenon occurring in another species (eg, the "5-HT Syndrome" crosses a number of mammalian species).
In the case of animal models of human psychopathology, the aim is to develop syndromes that resemble those in humans in order to study selected aspects of neuropsychopharmacology. The behavioral models are explicitly related to a broader body of theory, as they fulfill a valuable function in forcing the clinician and psychopharmacologists alike to critically examine their assumptions of the manifestations and pathophysiology of depression and bipolar disorders. Importantly, they are still required to provide guidance on optimal dose level selection for clinical regulatory safety, general toxicological, and efficacy studies in humans. 33 To disparage phenotypic animal models of psychiatric disorders seems unwise today when many molecular manipulations (eg,  33 Recent siRNA-mediated knockdown of the SERT in the adult mouse and rat brain would support the concept, although selectivity and side effects remain an issue. 33,38 Other recent methodologies include, antidepressant drug "signatures" using pharmacodynamic EEG measurements in animals and human studies as a measure of "antidepressant efficacy" and more recently with pharmacodynamic changes in EEG gamma oscillations. 39,40 Finally, the rapid progress in mutated mice studies using CRISPR/ Cas9 gene editing technology, 41 has shown that that differentiation of receptor subtypes can now be achieved for example, the delta subtype -547. 42 The FDA recently approved this intravenous drug, as the first treatment for PPD(https://www.fda.gov/NewsEvents/Newsroom/ PressAnnouncements/ucm633919.htm). A second SAGE compound is currently in a Phase III, study which is a more bioavailable inhib-

| Neurogenesis: creation of new neurones critical to antidepressant action?
Antidepressant treatments, such as SSRIs and electroconvulsive shock (equivalent to human electroconvulsive therapy, ECT) increase neurogenesis specifically in the hippocampus. 45 In fact, the maturation period for neurogenesis in the dentate gyrus appears consistent with the delay for the full therapeutic effects of antidepressants, as previous reported in the seminal work of Duman. [46][47][48] Thus, these preclinical findings suggest that adult neurogenesis may be modulated by factors associated with MDD, including chronic stress, 49 and activation of the HPA axis. 50 While the evidence re-

Receptor Antagonists
Corticotropin-releasing factor receptor antagonists have been sought since the stress-secreted peptide was isolated in 1981. 53 Pharmacological and transgenic studies show that brain and pituitary CRF 1 receptors mediate endocrine, behavioral, and autonomic responses to stress. 54 Thus, the therapeutic utility of CRF1 antagonists soon became clear and several small molecules progressed into clinical development for depression and other stress-related indications. 55 However, data with small-molecule CRF 1 antagonists did not consistently shown efficacy in animal models of "antidepressant-like" activity. 55

AV-101&cond=MDD&rank=2
Thus, with other clinical data still awaited the jury is still out for ketamine-like drugs being the "next generation" of rapidly acting antidepressant drugs Table 4. (see., 64 for an excellent review of glutamate signaling in depression).
In summary, given the overall complexity of the underlying neurochemical changes attribute to the pathophysiology of de-   The previous guideline as discussed, were far from optimal and barely cover the nature and detection of depressive disorders, acute treatment with antidepressant drugs, choice of drug versus alternative treatment, practical issues in prescribing, management when initial treatment fails, maintenance treatment to prevent recurrence, and the increasing importance of discontinuation of treatment. 71 A recent report indicated that it is not uncommon that withdrawal

| CLINI C AL TRIAL DE S I G N: THE ROLE OF PL ACEBO RE S P ON S E AND OUT COME ME A SURE S?
Over the last 30 years, the randomized, double-blind, placebo- yields 75% empirical power compared with 50% for LOCF. is simple to use, easy to implement, and to specify a priori. It is also more likely than LOCF to give adequate control of type I (false-positive) and type II (false-negative) errors. In other words, the use of either MMRM or LOCF will lead to the same conclusions but MMRM is likely to yield fewer mis-steps along the way according to some groups. 80 An extension of the MMRM, the novel nonlinear NLMMRM provides a tool for assessing a weighting factor collected from various centres thereby controlling the confounding effect of high placebo response across sites, to increase signal detection and to provide a more reliable estimate of the "true treatment effect" (TE) by controlling false negative results associated with excessively high placebo. 81 To date, few if any, published comparative study of newer antidepressants has enrolled a sufficiently large group of patients to have the power to reliably detect the differences between 2 effective treatments according to a recent critique. 74 One exception to this is for 7 newer antidepressants, only 65-75 patients were included per study arm. 74 Thus, the average study comparing 2 effective antidepressants would have less than 20% power to find a real, albeit modest (ie, 10%), difference in response rates. Put another way, the likelihood of a false-negative finding (ie, a type II error) would be 4 times greater than the chance of observing a statistically significant difference.
It is apparent that specific treatment effects have declined in recent decades. This may be due to selection bias at work that differs from that of a generation ago. The sample size, the number of centers, Contemporary trials, on the other hand, may be enrolling a different population: highly selected ambulatory less severe depressed patients who are often contacted through the mass media.
Thus, these subjects may be less severely depressed and are rarely treatment naïve. 74 Attempts to lessen these problems by restricting enrolment to patients with relatively high levels of pretreatment severity have often, in fact, accentuated them by inadvertently causing an inflation of entry depression scores. 74 Many clinical trials use entry criteria based in part on a minimum score for the same instrument used to evaluate efficacy. Investigators may be motivated, consciously or not, to increase baseline scores slightly in order to enter subjects into the trial. Such scores may then decrease by that same amount once the subject is entered, thus contributing to what appears to be a placebo effect-if not analyzed appropriately. 74 Another factor influencing the apparent effectiveness of antidepressants aside from the placebo-effect is the so-called "file-drawer

| Bridging studies
In a "bridging study," dosage is optimized early in development by determining the maximum tolerated dose of a compound in patients.
Consecutive panels of patients each receive higher doses of study drug until a minimum in-tolerated dose is reached. The dose immediately below this one is then considered the maximum tolerated dose.
Careful subject selection, adequate facilities, and highly qualified,

| Biomarkers for Depressive Disorders?
The main uses of biomarkers for drug development are: • discovery and selection of lead NCEs; • generation of pharmacokinetic (PK) and pharmacodynamic (PD) models; • aid in clinical trial design and expedite drug development; • serving as surrogates for clinical or mortality endpoints; • optimizing drug therapy based on genotypic or phenotypic factors; and • definition of patient enrolment in studies and help with stratification (biosignature development).
A major factor in the development of "better" antidepressants has been the lack of robust biomarkers, which has seriously limited prog- patients and enabling the development of diagnostic classifiers (biomarkers) that most likely will target specific patient phenotypes. 89,91 The validation of these assessments against relevant biomarkers, across large multi-site studies will add to their cogency. However, researchers have found numerous biomarkers associated with depression, but the statistical significance of each of these in isolation has not been strong enough to make a diagnosis but cross-disciple research efforts are being made to combine these various results, measuring many of these genotypes, phenotypes and analytes to create tests that 1 day could potentially real-world pre- and (c) factors controlling cellular plasticity such as brain-derived neurotrophic factor -, 103 18KDa translocator protein-TSPO, 104,105 Trace Amines-TAAR1. 8,9,106,107 One fascinating and potentially major step forward in our understanding the mechanism that contributes to the SSRI's treatmentresistance observed in approximately 30% of MDD patients and may greatly aid patient stratification, was recently published by in contrast to what is seen in healthy and remitter patient-derived neurons. 108 These findings suggest that postsynaptic forebrain hyperactivity downstream of SSRI treatment may play a role in SSRI resistance in MDD. These studies paint a complex and more nuanced picture of the serotonin hypothesis of depression and further highlight a role for serotoninergic dysfunction in the neuropathology of SSRI resistance in MDD that may lead to a further understanding endophenotypes and better treatment of MDD.
Aside from these recent advances, the number one challenge remains to develop novel antidepressants with greater efficacy and rapid action (see section on ketamine, Table 4). To this end, several pharmaceutical companies still continue to bet on the tried and tested monoamine approach and various augmentation strategies, 109

| CON CLUS ION
Although we live amid a game-changing revolution in neuroscience in the last decade, we are still incumbered with flawed biology, suboptimal regulatory models, clinical trial design, and protocols re- Therefore, as we emerge from this conceptual neurobiological revolution, the integrity and validity of epigenetic data, imaging brain neurocircuitry, molecular and structural insights, will become increasingly important in guiding optimal diagnosis, prediction of treatment responses in the discovery, and development of "better" antidepressants.
To quote Samuel Beckett: "Try again. Fail again, Fail Better," and to add to those wise words "To learn from our mistakes."