Treatment of depression: are we finally on the right track?

Keywords::

• depression
• treatment
• efficacy
• effectiveness
• outcome

The need for a different approach in depression treatment research

Despite considerable advances in our understanding of the causes and treatment of depression, many patients with mood disorders show only a limited response to current treatments or relapse after initial response1. Hence, it is time to step up our efforts in developing more effective treatments for this highly prevalent and debilitating condition. Two papers by Nelson et al.2 and Brnabic et al.3 recently published in this journal are part of what perhaps will become known as a ‘silent revolution’ in research on depression and its treatment.

Both papers report secondary analyses using data pooled from several existing trials stored in drug development databases. This novel approach may lead to the much-needed revitalization of pharmaceutical trials whilst providing new opportunities for research on the mechanisms involved in both the causation of depression and its treatment. Both papers depart markedly from the typical ‘horse race’ strategy that compares groups of patients randomized to different conditions. Instead, both studies use pooled trial data to investigate predictors and putative mechanisms of change in antidepressant treatment across treatment conditions. Rather than asking ‘Does it work?’, the questions addressed are: ‘What works for whom, how, and under what circumstances?’4. This approach was unusual until recently, but an increasing number of recent studies is relying on it5–7. The strengths of this approach are obvious: data can be pooled from well controlled studies with very similar designs and standardized measures. Besides standardization, these studies also have far more statistical power than individual trials. For instance, the total number of participants is 10,927 in the Nelson et al. study and 7984 in the Brnabic study. This contrasts markedly with many existing individual studies, which are often seriously underpowered, and even with meta-analyses that are forced to lump together studies with various designs, measures and overall study quality. Finally, the typical cross-national, multi-site nature of many drug trials opens the door for truly cross-cultural comparisons of treatment effects and mechanisms implicated in depression. Given the increasing evidence for cross-cultural differences in both psychosocial and biological factors in depression8, this may be one of the most important as yet unexplored strengths of pooled data sets from drug trials. For instance, a meta-analysis has found a strong correlation between the relative frequency of genes implicated in social sensitivity (e.g., 5-HTTLPR, MAOA-uVNTR, OPRM1 A118G) and the emphasis on independence (‘individualism’) versus interdependence (‘collectivism’) of individuals within a given culture9. Importantly, this meta-analysis also showed that the prevalence of depression was negatively correlated with the frequency of these alleles and that this association was mediated by collectivism. Hence, a cultural emphasis on collectivist values seems to buffer the effects of stress in populations with a high proportion of social sensitivity alleles. Yet, recent population-based studies have reported a dramatic increase in depression and suicide in these cultures10–13. This may perhaps be explained by the increasing shift towards greater individualism in many of these cultures and thus decreasing levels of social support. Pooled data sets could shed further light on these findings, and moreover could also address potential implications for treatment. For instance, it may well be that people in collectivistic cultures are on average more responsive to psychosocial interventions compared with people in more individualistic cultures given their sensitivity to their environment. With increasing insights into gene–environment transactions, the hunt for genes that might predict treatment response is now in progress14, but these genes might differ considerably between cultures, again emphasizing the need to consider cross-national data sets in the identification of effective treatments.

From disorders and horse races to persons and mechanisms

The field of psychotherapy research has seen a remarkable shift from efficacy and effectiveness research to the study of mechanisms of change15,16. Antidepressant trials seem to be moving in a similar direction, as they should. Nelson and colleagues2 clearly headed this call. They report that over half of depressed patients in their pooled data set of patients treated with duloxetine and selected SSRIs did not achieve remission, thus replicating meta-analytic findings concerning the efficacy of antidepressants. The ingenious aspect of this study, however, is that these authors then divided the total sample into two groups regardless of treatment condition: placebo remitters (i.e., patients who had features previously found to be associated with remission in placebo conditions, such as younger age and shorter duration of depressive episodes) versus placebo nonremitters (the remaining patients). Next, they investigated the response of both groups of patients across experimental and control conditions, rather than comparing participants in the experimental versus control (i.e., placebo) condition. Using this approach, they found that drug–placebo differences were greater in patients with placebo nonremitter features, thus providing a fairer test of the effects of antidepressants than trials that do not take into account interindividual differences between patients.

This strategy illustrates the promise of a person-centered or patient-level approach over a disorder-centered approach. It represents an important step forward in our attempt to develop and identify effective treatments for depression and related conditions, as these authors identified relatively simple patient features such as age and duration of depressive episodes that predict treatment outcome with duloxetine and SSRIs. Interestingly, similar features have often been found to predict treatment response in the psychotherapeutic treatment of depression as well17.

However, more is needed to demonstrate mechanisms of change and to allow clinicians to truly tailor their interventions to specific patient features. Descriptive features such as age and duration of depressive episodes are unlikely to be directly reflective of underlying biological and psychosocial mechanisms involved in depression. These features are more likely to be proxies of biological and/or psychosocial patient features. Currently, theory-driven efforts to identify psychosocial predictors of change in depression are as least as promising if not more promising than research on biological predictors of treatment response. Indeed, as major proponents of biological psychiatry Kapur, Philips and Insel18 recently argued “the most effective ‘stratifiers’ in psychiatry may well come from standardized cognitive and psychological measures” (p.4), rather than from biological factors alone, given their poor specificity in predicting treatment response. For instance, whereas many biological parameters fail to predict treatment response in depression, studies for instance suggest relatively consistently that early trauma19 and cognitive–affective schemas such as high levels of self-critical perfectionism16 negatively predict treatment outcome in both pharmaceutical and various psychotherapeutic treatments for depression. This may come as no surprise, as early trauma and self-critical perfectionism have both been related to dysfunctions in the hypothalamic–pituitary–adrenal axis and dopaminergic reward circuit that have been implicated in impaired stress and emotion regulation in depression and many other disorders19–21. Hence, more theory-driven studies that are rooted in our growing knowledge of the causation of depression may lead to the identification of patient features and their influence on response to pharmacotherapy and psychosocial interventions. This approach contrasts markedly with the current – and, from a scientific stance, rather unsophisticated – strategy of comparing the efficacy of antidepressants in randomized trials involving individuals selected on the basis of ill-defined categorical diagnoses that seem to bear little relation to underlying biological and psychosocial processes. Because this strategy has dominated the field, however, we currently largely lack the knowledge to develop guidelines for tailoring interventions based on patient features.

Importantly, it is likely that mechanisms involved in various disorders cut across psychiatric disorders, congruent with the principles of equifinality and multifinality from developmental psychopathology22. Impairments in mood are implicated in almost any disorder (multifinality), and it is therefore highly improbable that the effects of antidepressants are specific to a consensus-based disorder such as major depression. Conversely, many pathways are implicated in depression (equifinality). In this context, the recently launched Research Domain Criteria (RDoC) initiative by the National Institutes of Mental Health is a prime example of major ongoing changes in our approach towards the understanding and treatment of psychopathology23. Basically, the RDoC initiative, similar to developmental psychopathology approaches, proposes to study underlying mechanisms (such as reward or fear circuitry) across psychiatric disorders rather than assuming that specific causative factors are implicated in specific disorders. This approach promises to allow the development of guidelines for the tailored treatment of depressed patients in the near future. As noted, however, more research is needed, as it is highly unlikely that vulnerability or treatment response markers are invariant across time and cultures as there is clear evidence for gene–environment transactions (including epigenetic effects) and even gene–culture co-evolution effects in depression9.

Findings from the study by Brnabic et al.3 are another case in point. This paper also points to common mechanisms across disorders, as the authors found very high levels of painful physical symptoms (PPS) in depressed patients. Moreover, they note that the “failure to adequately address PPS may be a contributing cause for not reaching remission”. There is indeed increasing evidence indicating that depression and pain- and fatigue-related disorders may be part of a spectrum of affective or stress-related disorders24–26. Again, it is important that future studies aim to understand this association better, as this may increase treatment efficacy. The study by Brnabic, for example, suggests that with increasing severity of depression there is a failure to differentiate adequately between physical and emotional pain, such that psychological pain seems to be increasingly felt as physical pain by depressed patients. This is congruent with findings of similar neural circuits being involved in emotional and physical pain (‘rejection hurts’)27. This ‘desymbolization’ or pain sensitization process may hinder effective treatment as it may decrease patients’ motivation, hamper the development of a working alliance with health professionals, decrease patients’ feelings of controllability of symptoms20, and decrease the efficacy of specific classes of antidepressants, while antidepressants with analgesic features (like duloxetine) may perhaps be more effective in these patients28. These speculations await further research, but this study illustrates the possibilities of this approach.

Potential pitfalls and solutions

There are at least two important threats to studies based on pooled data of drug development databases. First, they raise the issue of potential conflicts of interest to an even greater extent than the original trials on which these pooled studies are based. Antidepressant research has been under fire for some time now by critics and the general public alike. The influence of conflicts of interest has been one of the main issues in this discussion, particularly since the publication of meta-analyses suggesting that there is publication bias in pharmaceutical trials29.

Second, although clear advances have been made in the pharmaceutical treatment of psychiatric disorders, there is growing consensus that biological psychiatry has not lived up to its promise. A recent editorial in the prestigious British Journal of Psychiatry even announced “the end of the psychopharmacological revolution”30. As a result, there have been calls for a shift in the currently dominant approach in pharmaceutical trials – comparing treatments – to an approach focused on underlying mechanisms that are likely to cut across diagnostic categories, such as in the abovementioned RDoC initiative.

Drug development databases could play an important role in these endeavors, but only provided that pharmaceutical companies provide open access to data, congruent with the trend towards open-access policies in scientific publishing. Pharmaceutical companies should take the lead in opening up their results to public scrutiny in order to prevent criticism concerning the potential role of conflicts of interest and to fully realize the potential of pharmaceutical trials. It is therefore time for a paradigm shift consisting of (a) an open access policy, allowing scientists to explore existing drug development databases to minimize explicit and implicit influences of conflicts of interest, and (b) the inclusion in future drug trials of measures that tap into purported biological as well as psychosocial processes and mechanisms involved in the causation and treatment response of depression. This joining of forces between pharmaceutical companies, biological and psychosocial researchers may be the only way forward. With the increasing realization that the brain is fundamentally a social brain, biological psychiatry should follow suit and team up with psychosocial research. Depression is fundamentally determined by (interactions between) psychosocial and biological factors. Drug trials should reflect this and should open up to researchers interested in the mechanisms of change in depression treatment.

In this respect, we may be witnessing the end of an era, which, despite great hopes, has disappointed many; patients, patient organizations, governments and, not least, those active within the field of psychiatry as researchers and clinicians. We can only hope that this new approach toward trials in antidepressant treatment is more fruitful and that we are finally on the right track in identifying what works for whom in depression treatment. We owe it to our patients.

Transparency

Declaration of funding

The author received no payment in preparation of this article.

Declaration of financial/other relationships

P.L. has disclosed that he has no significant relationships with or financial interests in any commercial companies related to this study or article.

CMRO peer reviewers may have received honoraria for their review work. The peer reviewers on this manuscript have disclosed that they have no relevant financial relationships.