Addressing the unmet needs of current antidepressants: does neuroscience help or hinder clinical psychopharmacology research?

KEYWORDS:

• Depression
• antidepressants
• psychopharmacology
• translational research
• clinical trials
• precision medicine
• pharmacogenetics

1. Introduction

It is simple enough to enumerate the shortcomings of existing pharmacotherapies for depression: the desire for drugs with faster onsets, higher remission rates, fewer side effects, robust anti-suicide properties, breadth of spectrum across subtypes (melancholic, atypical, agitated, chronic, bipolar), cognitive benefits, durable relapse prevention and persistence of efficacy, better functional outcomes, and the like; but it is a far more complex and speculative effort to propose viable strategies for their remediation. The latter of these is especially daunting in light of two recent events: (1) NIMH’s 2014 decision no longer to fund intramural or extramural clinical trials for the sake of symptom relief, but only when a drug can serve mainly as a ‘probe to generate information about the mechanisms underlying a disorder’ (https://www.nimh.nih.gov/about/directors/thomas-insel/blog/2014/a-new-approach-to-clinical-trials.shtml), and (2) growing divestment within the pharmaceutical industry from research and development of new psychotropic compounds, both in the US and abroad [1]. While the NIMH’s quest for ‘precision medicine’ [2] is scientifically laudable, its decision to prioritize the search for suspected and elusive neurobiological causes of mental illnesses – and to link that search inextricably with efforts to advance therapeutics – will not likely reverse the relative standstill of new drug development in psychiatry, or improve delivery of mental health services to those now in need.

2. Biomarkers: elusive, but necessary?

The pursuit of biomarkers and personalized medicine for depression treatment is a high-risk, high-stakes venture that holds no guarantee for ultimately reducing the burden of disease. Even the quest to identify novel drug mechanisms of action, with hopes to advance beyond the decades’ old monoaminergic theory of depression, runs the risk of hitting upon repeated dead ends. Who can forget the arc of enthusiasm followed by despair over randomized trials with NK-1 antagonists such as aprepitant [3], CRF-1 antagonists such as CP-316,311 [4], and metabotropic glutamate receptor modulators such as basimglurant [5]? Moreover, wrong inferences and assumptions about generalizable class effects also have their way of leading the field astray, as happened in the early 2000s in the case of presumed mood stabilizing effects of most anticonvulsants (such as topiramate [6], gabapentin [7], oxcarbazepine [8], or licarbazepine [9] – none of which has ever shown superiority to placebo to treat any phase of bipolar disorder).

More recently, amid the afterglow of excitement about observed antidepressant effects with intravenous ketamine, disappointment has ensued from randomized trials in depression with other NMDA receptor antagonists such as riluzole [10], memantine [11], lanicemine (AZD6765 [12]), and D-cycloserine [13]). And while ketamine itself remains perhaps the most promising recent innovation in depression treatment, the frequent transience of its antidepressant effect poses challenges for relapse prevention, especially when its chronic administration may be neurotoxic [14]. Mechanistically, it is also hard to reconcile possible antidepressant effects from NMDA receptor antagonism or agonism (e.g. Huang et al. [15]), which may require invoking pharmacodynamic explanations other than NMDA receptor function altogether, such as mu opiate receptor modulation [16]).

Infatuation with biomarkers, and their prerequisite status for developing new depression treatments, also extends to pharmacogenetics. Existing genomic panels assay for genes coding for pharmacokinetic, more than pharmacodynamic, enzymes or receptors – yielding information that could potentially help the small minority of individuals who are poor metabolizers anticipate their chances for intolerance to some drugs, or the inability to process pro-drugs, more than information about drug efficacy. Recent efforts have begun to examine whether some metabolic enzyme variants (e.g. CYP2C19) could play a role not only in pharmacokinetics but also in the vulnerability to depression, and as a possible preclinical screening tool for new antidepressants [17]. Nevertheless, it is not well-established whether psychotropic drug response is even heritable and if it is, how the magnitude of its effect compares to that of other known clinical factors that influence antidepressant treatment outcomes. Commercially available pharmacogenetic tests run the risk of Type II errors when the studies upon which they are based are conducted with alpha levels underpowered to detect genome-wide significance [18]. Yet, these limitations have not dissuaded many clinicians from ordering cheek swab genotyping tests despite the dubious assumption that their results predict whether an antidepressant drug will be effective [18].

When we turn to the parallel problem of industry’s exodus from the arena of new antidepressant drug development, the aforementioned scientific challenges are compounded by financial disincentives alongside a general sense of public (and sometimes governmental) mistrust that the for-profit sector is more unscrupulous than beneficent. Realistically, the cost to develop a new drug may exceed $2.5 billion [19], in a marketplace where only about 12% of new chemical entities that enter Phase I testing ultimately achieve regulatory agency approval. From 1990 to 2012, central nervous system (CNS) compounds were 45% less likely than non-CNS drugs to complete Phase III testing and undergo regulatory filing, with most failures occurring due to lack of efficacy [20]. If the respective agendas of both NIMH and industry cause them to withdraw (or markedly reduce) their resources from the pursuit, development, and funding of new therapies for depression, the negative repercussions for devising better treatments will likely pose the field’s greatest unmet need.

3. Needed: a better integration of neuroscience with clinical phenomenology

Some of the more nuanced unmet needs of depression treatment sit at the crossroads of the scientific and economic logjam. For example, depressed patients with comorbid psychiatric, medical, or substance use disorders comprise a majority of real-world patients who seek treatment; yet, because such treatment-seekers are not ‘prototypical’ of major depression in its pristine form, comorbidities render most such patients ineligible for enrollment in industry-sponsored FDA registration trials [21]. In fact, we know little about the effectiveness of existing antidepressants for depressed patients with comorbid psychiatric and/or substance use disorders [22]. Vast phenotypic heterogeneity across such diverse presentations of major depression may account for the nonreplication of findings from genome-wide association studies [23] – although clinical features such as age at onset [24] or chronicity [25] may help to refine depression phenotypes and specificity of antidepressant response.

Another ‘crossroads dilemma’ of unmet needs involves the study of complex combination drug therapy regimens. Because only about one third of ‘real-world’ major depression patients respond well to a single agent, the use of two or more psychotropic agents has become increasingly routine. From 1996 to 2006, the number of adults taking two or more antidepressant drugs increased by 344% [26]. Yet, as seen in the NIMH COMED trial, outcomes are not demonstrably better in depressed patients taking two versus one antidepressant agents [27]. Most existing antidepressants share relatively redundant mechanistic targets, posing obstacles for devising truly novel combination drug therapies. In contrast to other diseases in which extensive polypharmacology is the norm (e.g. HIV, infectious disease, cancer), it is unlikely that adequately powered multi-arm randomized trials of an elaborate polydrug regimen will be conducted for depression, or that the field will see a careful distillation of truly synergistic combinations involving three or more agents at a time – particularly as NIMH no longer supports purely clinical undertakings.

It is obviously essential to develop a better understanding of how the brain works, how some people fail to develop resilience, how early life stresses modify brain response to current stresses, and to identify new drug targets to assist the third or more of depressed patients who respond poorly to current treatments. On the other hand, we are presently failing to deliver what we already have in our medicine cabinet, comprised of remedies discovered largely by serendipity [28]. Over 20% of depressed patients drop out of treatment before adequate trials occur [29]. We scarcely know how to organize, sequence, and optimally deliver the treatments we already have.

Having too few adequately powered clinical trials leads to opinion replacing evidence, and impressionism replacing empiricism. For example, antidepressants are popularly considered anathema in bipolar depression even though the vast majority of modern antidepressants have never even been formally studied for that condition. High variability in prescribing habits and patterns across practitioners reflects the dearth of evidence from which to make clinical decisions. Clinical evidence is no less important than molecular evidence, yet current funding biases favor the molecular over the macroscopic. The National Cancer Institute and National Heart, Lung and Brain Institute provide good examples of National Institutes of Health divisions that invest in efforts aimed at elucidating the biological cascades that cause disease while simultaneously determining whether or not a particular biomarker, within the context of a definitive clinical trial, can deliver more precise outcomes.

One in 10 American adults takes an antidepressant, and antidepressants are the most commonly prescribed type of medication in people ages 18–44 (https://www.cdc.gov/nchs/data/databriefs/db76.pdf). Yet, the broad category of ‘major depression’ conveys little information from which to make more than bare bones pharmacotherapy or psychotherapy decisions. NIMH National Advisory Mental Health Council members, while acknowledging the potential long-term value of neuroscience-based depression research, have called for a rebalancing of funding priorities to align better with patients’ more immediate and urgent mental healthcare needs [30].

4. Expert opinion

Clinical trials remain the cornerstone for determining effective therapeutics. Efforts to relieve morbidity and mortality from depression cannot be contingent solely on the parallel quest for its putative biomarkers. Future progress demands a more practical integration between basic neuroscience and clinically oriented research efforts. Useful steps toward that end might include (1) creating a clinician- and investigator-friendly ‘crosswalk’ between existing DSM-5/ICD-10 categories of depression and observable domains of behavior, cognition, or emotion (e.g. as intended by the NIMH Research Domain Criteria taxonomy); (2) as an organizing principle for pharmacotherapy trials, encouraging a focus on the development and measurement of specific dimensions of psychopathology found in clinical depression (e.g. sensorimotor reactivity, anhedonia, impulsive aggression, social cognition) – rather than on diagnostic entities whose operational definitions are guided by the pursuit of FDA label indications; (3) focusing pharmacotherapy trials on clinically unique and/or difficult-to-treat subpopulations, such as anxious or otherwise comorbid forms of depression [22], or inflammatory versus noninflammatory depression subtypes (e.g. Uher et al. [31]); (4) devoting resources to study management of adverse drug effects and improving treatment adherence; (5) fostering integrative research on pharmacotherapy plus depression-specific psychotherapies; and (6) funding incentives to rediscover existing off-patent medications for key subpopulations (such as depression in ethnic minorities, adolescents, late-life, pregnancy, or individuals with metabolic syndrome) – perhaps via expanding ‘orphan drug’ status to include understudied drug therapies for depression. Lastly, precision medicine requires a colossal clinical infrastructure; some investigators have begun to call for standardization of electronic health records and expansion of President Obama’s ‘Precision Medicine Initiative’ (https://obamawhitehouse.archives.gov/node/333101) beyond its original starting point in oncology [32]. Enrolling broad, heterogeneous groups of depressed patients in bioinformatic databases such as ClinVar (https://www.ncbi.nlm.nih.gov/clinvar/) and the Electronic Medical Records and Genomics Network (eMERGE; https://www.genome.gov/27540473/) could provide greater power to match ‘real-world’ heterogeneous depression subgroups with treatment outcomes, and allow for data-driven phenotypic categorization of neurobiologically and clinically meaningful depressed subgroups [33].

Declaration of interest

JF Goldberg is on the scientific advisory board or a consultant for Otsuka, Sunovion, Supernus and WebMD. He is on the speakers bureau for Merck, Neurocrine, Otsuka, Sunovion, Takeda-Lundbeck, Vanda Pharmaceuticals, and receives royalties from American Psychiatric Publishing, Inc. AJ Rush receives consulting fees from: the American Psychiatric Association, Brain Resource Ltd., Compass Inc., Curbstone Consultant LLC, Eli Lilly, Emmes Corp., Liva-Nova, Lundbeck A/S, National Institute of Drug Abuse, Taj Medical, Santium Inc., Sunovion, and Takeda (USA). He has received speaking fees from Liva-Nova; and royalties from Guildford Publications and the University of Texas Southwestern Medical Center. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

Additional information

Funding

This paper was not funded.