Major depression is characterized by a high lifetime prevalence of 16–17% in the general population (1) with the possibility of evolution into a recurrent or chronic course (2) and a negative impact on the general functioning and quality of life (3). Major depression also has a high level of comorbidity with various psychiatric and non-psychiatric disorders (1, 4), with a high occurrence of completed suicide (4), and significant burden associated with the heavy personal, societal, and economic costs (4). Treatment-resistant depression (TRD) has been estimated to represent half of the overall treatment costs for major depression (5, 6). Although not clearly established in France, it can be assumed that approximately 20–30% of depressed patients experience TRD, as reported in Anglo-Saxon countries (7), and up to one half of patients responding only partially (8). TRD is currently defined by the failure of at least two attempts of antidepressant treatments administered sequentially at adequate dose and duration (8). Patients may be unresponsive to usual therapeutic strategies because of clinical features of the condition itself, factors interfering with the proper delivery of an optimal treatment such as under-dosing or poor adherence related to side effects, cognitions, comorbid disorders (psychiatric and non-psychiatric), etc. (5, 6, 8–12). Although controversial (13), one of the major issue relies on the misdiagnosis of TRD instead of BP-II because hypomanic symptoms are difficult to detect, especially on a retrospective basis (14). Indeed, as reported by Judd et al. (15), patients with BP-II were symptomatic more than half of all follow-up weeks while depressive symptoms were largely more preponderant during the course as compared with hypomanic and cycling/mixed symptoms. Unrecognized bipolar spectrum illness increases substantially the risk for the inappropriate prescription of antidepressant treatment known to be associated with the occurrence of short-term resistance, poor tolerance, cycle acceleration, manic switch as well as late response loss after several months of recovery (13, 14, 16, 17). Thus, an inaccurate diagnosis almost always means a significant delay in appropriate and effective treatment based on mood stabilizers for bipolar patients. However, in this study, the adopted procedure tends to minimize hidden bipolar disorders behind TRD cases by using the structured Mini International Neuropsychiatric Interview (MINI) for the diagnostic evaluations coupled with the Mood Disorder Questionnaire (MDQ) useful for screening signs and symptoms of bipolar spectrum disorders.

Also, there is a heterogeneous etiology of TRD, including different environmental risk factors like childhood adversities coupled with multiple genetic determinants related to numerous genetic loci, and various epigenetic contributors involving a wide range of biological systems including the hypothalamic–pituitary–adrenal axis, immune function, monoamines, neurotrophic factors, etc. (5, 6, 8, 9, 18). These factors can significantly modulate the therapeutic response while remission remains the aim of treatment for depression, given its implications for recovery of daily functioning and better long-term prognosis (7, 19–22). The absence of appropriate treatment increases the burden of depression and the resulting costs. This burden manifests itself in many ways including residual symptoms, cognitive impairment, relapse and recurrence, decreased quality of life, suicide, cardio- and cerebrovascular morbimortality, psychosocial alteration, work days lost, family impact, and economic costs (4, 23). Taken together, all these considerations underlie the particular relevance for the determination of contributing factors by a large and exhaustive assessment of patients suffering from TRD.

Many recommendations for clinical practice have been developed by various psychiatric societies around the world (24–28). These recommendations are primarily based on the analysis and classification of the scientific literature. They consider randomized, controlled and double-blind studies on large samples as offering the highest level of evidence. In spite of their rigor from a methodological and scientific point of view, the adaptability of these recommendations in everyday clinical practice can raise questions (29). High-level evidence studies, in fact, refer to selected samples of subjects, often free from comorbidities. They exclude characteristics such as advanced age, increased suicidal risk or high level of pharmacological resistance. These clinical situations are, however, frequently observed in daily clinical practice in the patient population suffering from TRD (12, 30).

In France, the need for improving the diagnosis, assessment, and management of psychiatric pathologies has led both the Ministry of Research and the Ministry of Health to support jointly the development of a national network of expert centers for Bipolar disorder (31), schizophrenia, TRD, and autism under the leadership of the foundation of scientific cooperation named “FondaMental” created in 2007. This article describes a new model of clinical collaboration between the expert centers and the regional community of clinicians, especially those general practitioners and psychiatrists who provide the first point of contact with health services for most TRD patients. Thus, within this network, we propose a systematic, comprehensive, longitudinal, and multidimensional assessment for patients who have failed to achieve remission after at least two adequate trials using different classes of antidepressants. The psychiatric, non-psychiatric, socioenvironmental, metabolic, biological, and cognitive dimensions of the current depressive episode are explored. A personalized evaluation and care program is delivered by a clinical expert highly specialized in the field of TRD. It also intends to narrow the gap between research knowledge and clinical practice. This represents an additional and locally accessible approach to the patients and their referent physician. The objective of this article is to disseminate information about this highly relevant project and to describe the standardized evaluation as proposed within the French network of centers specifically dedicated to TRD.

By offering a systematic, complete, longitudinal, and multidimensional evaluation of each TRD case, using shared medical records containing common assessment tools exploring separately the psychiatric, non-psychiatric dimensions, metabolic functions, biological, and cognitive impairments, we aim at making accessible to all patients with a diagnosis of TRD, the most personalized and adapted treatment strategies possible. Indeed, it seems important to be able to apply the therapeutic guidelines to improve the prognosis of TRD. Guo et al. (65) showed an improvement in the management using measurement-based care for outpatients with moderate to severe major depression. Indeed, the responses as well as remission rates are higher, and adjustment of therapeutic doses of antidepressant is more straightforward in the measurement-based care group. As proposed in the expert centers, the administration of a wide range of clinical instruments and neuropsychological tests for the comprehensive assessment of depressed patients could be perceived negatively, especially with regard to the long time, attentional demand and sustained effort required for completion. However, clear information about the study design is delivered before final acceptance of participation. All these assessments are scheduled over three half-days on two consecutive days but interleaved systematically with frequent rest periods to prevent sufficiently a gradual decline in concentration and excessive fatigability. Phone reminders are even scheduled at least one week before each visit to limit dropouts. As supported in the study by Rush (66), measurement-based care also offers the advantage of better preparing and engaging the patients in shared decision making and adherence to long-term care needs, thereby familiarizing themselves with relevant tools currently used for monitoring their clinical symptoms and side effects caused by standard treatments.

Moreover, diagnostic and clinical heterogeneity appears to be a major obstacle for an exact understanding of the pathophysiology of mental illnesses and, in particular, major depression. Although there is a widely demonstrated hereditary component (67) the identification of genetic risk factors has proven difficult even in meta-analyses of genome-wide association studies (68).

Similarly, efforts to develop new treatments have stagnated, partly due to a lack of new biological targets and the choice of the people most likely to benefit from them (69). All these challenges have been attributed in part to the fact that our diagnostic system assigns a unique label to a syndrome that is not unitary but could be caused by distinct pathological processes, which would require different treatments. Indeed, major depression is a widespread disease presumably represented by a large number of patient subgroups that must be identified precisely to propose adapted therapeutic strategies.

Our national network must be part of the existing regional health care system. Although adherence to treatment guidelines has been shown to improve clinical outcomes (65), there are barriers to their application, including the fact that guidelines are often viewed as “top-down” measures. One aim of the network is to make available the standard recommendations derived from the currently used guidelines for each patient. It should bridge the gap between evidence-based medicine and routine practice.

In addition, clinical research is an important issue within the network. The electronic health-care record system for each expert center is connected to a national database and data on this large clinically representative cohort of TRD cases will allow prospective follow-up and comparative-effectiveness studies. Combination of clinical, biological, genetic, and epigenetic data could contribute to a more precise definition of separate clusters of major depression for a more specific and tailored management of this pathology. To refine the diagnostic and therapeutic approach, by focusing on more precise clusters, the main lines of future research are focused on: (i) the role of comorbidities, beyond what are now defined as risk factors for resistance [cardiovascular, neurological, metabolic pathologies, etc. (70–72)]; (ii) the role and impact of traumatic factors; (iii) the presence of suicidal ideation or behaviors; (iv) the functional repercussions; (v) resistance to treatments; (vi) neurocognitive profiles; (vii) genetic, epigenetic, immuno-inflammatory, neuroendocrine, and neurochemical markers, etc. In this context, patients benefiting from the evaluation within the expert centers, are asked to participate in various national and European research studies implemented in partnership with the “Fondation FondaMental” addressing jointly clinical, pathophysiological, and therapeutic issues (73).

The adopted integrative approach shared within the network of expert centers for resistant depression is a compelling model for the precise characterization of sociodemographic, clinical, neuropsychological, biological, and metabolic factors of treatment resistance and to further develop innovative and personalized strategies through the multi-faceted questions of assessment, management, and treatment of major depression. This is one of the best options for changing the poor prognosis of TRD, reducing the deleterious impact on daily functioning and quality of life and thereby lowering the serious societal and economic burden related to high direct and indirect costs of TRD.

Ethics approval and consent to participate: the database (clinical data and biobanking) was validated by an ethics committee: CNIL (French CNIL: DR-2015-673). Consent for publication: all patients in the expert centers receive oral and written information on the possible use of their data for research purposes. Availability of data and material: all the material used for the assessment is validated in French.

AY, DB, EH, WE–H, and BA realized the initial frame of the manuscript. The authors have made corrections after proofreading. All the authors participated in the recruitment of patients.

AY: I acted in advisory capacities, carried out clinical studies in relation to the development of a medicine (Janssen), received travel allowance, gave presentations at meetings, and received remuneration for my input from the following pharmaceutical organizations: AstraZeneca, Janssen, Lundbeck, Otsuka, and Servier. DB has received honoraria from Lundbeck. EH: I acted in advisory capacities, carried out clinical studies in relation to the development of a medicine, received personal researches, studies or travel allowance, gave presentations at meetings, and received remuneration for my input from the following pharmaceutical organizations: AstraZeneca, BMS, Cellgene, Euthérapie—Servier, Janssen, Elli Lilly, Lundbeck, Otsuka, Pfizer, and Sanofi. And, I held a managerial position in Fondation FondaMental, Créteil, and the French Association of Biological Psychiatry. MG has received honoraria from Janssen—Cilag and Lundbeck. TD, PD, LM, and LS: no competing interest. OD has received honoraria from Lilly, AstraZeneca, Jansen, Servier, and Lundbeck. FH has received honoraria from AstraZeneca, a grant from Institut Servier and is currently supported by Fonds de recherche du Québec—Nature et technologies (grant number #200123.). FS has received honoraria from Otsuka. GV has received speaker honoraria from Otsuka/Lundbeck (annual intervention at the reception day for new professors of psychiatry). WE-H has received speaker honoraria from Janssen, Lundbeck, Otsuka, and UCB. BA: as received speaker honoraria from Lundbeck, Janssen, and Eli Lilly. All other authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.