Subtyping Major Depressive Disorder

The Diagnostic and Statistical Manual, Third Edition (DSM-III) [1], of the American Psychiatric Association, appeared almost three decades ago and revolutionized the way psychiatrists diagnosed depression. Consistent with the DSM approach for other diagnoses, depression was defined by detailed criteria that strove to describe the clinical picture while excluding considerations of context and etiology. Impressive gains in interrater reliability along with descriptive and predictive validity resulted. The DSM spurred empirical research into the phenomenology of depression, its etiology and its treatment. Supporting the DSM approach was the fact that it was difficult to document differences between depression following psychosocial stress and depression without such precipitants. Moreover, in animals, antidepressant drugs are effective in neurochemical, stress-induced and genetic models of depression [2]. An atheoretical definition of depression facilitated recruitment for large clinical trials of the new antidepressants that have appeared almost yearly since the appearance of DSM-III. Moreover, it may have been important for insurance reimbursement of psychiatric treatment that depression be diagnosed in a non-judgmental symptom-based approach without speculative attention to non-medical causes in individual cases. However, recent years have witnessed a growing awareness of problems that call for a reassessment of how best to classify depression [3].The current definition of depression is very broad; over 16% of the general population will have depression during their lifetime in epidemiological studies using current criteria [4]. It is unlikely that a syndrome as polymorphic and widely diagnosed as major depressive disorder (MDD) will reflect a single process. Attempts to delineate different forms of depression by statistically analyzing the symptomatology of large samples of patients without taking into consideration life events or childhood history have been unsuccessful. It may be necessary to consider childhood trauma, marital and employment stress, and medical health in diagnosing subtypes of depression. Epidemiologic data on the effects of childhood trauma, unemployment and divorce on depression incidence are strong [5,6]. DSM-IV does make one allowance for circumstances by including a bereavement exclusion [7]. However, it seems that bereavement is not different from other losses and stresses that are associated with depression [8].The pharmacotherapy of depression offers a plethora of medications. However, depression is often resistant to standard antidepressant medication, and a large percentage of patients respond just as well to placebo [9,10]. The DSM broad diagnosis of MDD does not encourage a search for subtypes of depression that may require specific treatment. The current diagnostic system may be partly responsible for the rising placebo response rates and increasing numbers of subjects required to demonstrate efficacy in ever larger clinical trials. There is little incentive today to initiate small trials of novel or specific treatments. Most studies are commercially sponsored multicenter projects, and lump many possible subgroups under the rubric of MDD. The rating scale information collected in multicenter trials that is sometimes used for subtype analysis lacks the detailed historical data that are elicited in clinical practice. In fact, most clinicians subtype depression as a matter of course when describing patients to colleagues. We do not believe that useful subtypes will emerge from the current practice of collecting clinical data in large pharmaceutical trials. It may be that the era of large commercial studies of antidepressants using MDD as a diagnosis is over [11]. It may be possible to kick start the system by subtyping depression to encourage a period of small investigator-initiated studies of potential new treatments by subtype. Therefore, the following classification is not an evidence-based replacement of the present DSM-IV MDD: it is an intuition-based proposal for heuristically classifying depressions. Type A: Depression with Anxiety. This is characterized by an enduring tendency to experience anxiety and depression, and to show poor resilience under stress [12,13]. Depressive episodes are not clearly demarcated from periods of baseline discontent. Appetite and sleep patterns are relatively undisturbed. Patients are responsive to improvements in their circumstances, until another setback renews their misery. Heritable factors seem related to the genetics of personality, and some may be mediated by polymorphisms of the promoter of the serotonin transporter (5-HTT) gene [14].Type B: Acute Depression. This subtype has episodes that are relatively discreet and develop with no apparent precipitating stress, or the stress may be disproportionate to the intensity and duration of the depression. The severity of the depression can deteriorate into intense psychological pain and psychomotor retardation or agitation. Recurrent acute depressions of early onset, beginning in the patient's twenties, have some reported gene linkages [15]. The human syndrome has similarities to the reserpinized rodent, which suggests that a deficiency in central serotonergic or noradrenergic neurotransmission may contribute to this form of depression. Some evidence suggests that blockade of both serotonin and noradrenalin reuptake, such as with the old tricyclic antidepressants (TCAs) or the newer serotonin-norepinephrine-reuptake inhibitors (SNRIs), may be more effective than selective serotonin-reuptake inhibitors (SSRIs) [16]. Several authors have suggested that this subtype of depression should be termed 'melancholia' [17,18].Type C: Adult Depression after Childhood Trauma. This form of depression may be unique. Individuals suffering early trauma or loss may develop lasting neurobiological changes, which render them vulnerable to stress throughout their life. In particular, sensitization of the hypothalamic-pituitary-adrenal (HPA) axis may remain throughout life. Recent research suggests that the dexamethasone/CRF test may reveal elevated cortisol and ACTH levels only in those depressed individuals who have also suffered severe stress in childhood [19]. Nemeroff et al. [20] reported that such patients respond to cognitive-behavioral therapy better than to an antidepressant. Type D: Depressive Reaction to Separation Stress. This can be precipitated by acute psychosocial trauma such as bereavement, divorce, job loss or forced emigration, and may sometimes be more severe than other forms of depression [21]. For months, and sometimes years afterwards, some may experience sadness, apathy, insomnia and pessimism [8]. Clearly these reactions are very common in the population, but increase in frequency in times of war, rising unemployment or family breakdown. We hypothesize that primary mental health prevention may be most important for this type of depression [6].Type E: Postpartum Depression. This has a typical peak onset in the first 3 months following delivery [22]. There are vast reductions in estradiol and progesterone levels postpartum, but their exact relevance to the development of depression remains unproven. Psychosocial factors, such as an unsupportive partner or an unwanted pregnancy, have also repeatedly been found to be relevant to the development of postpartum depression. Therapy requires a combination of psychopharmacological and psychotherapeutic interventions [23]. Type F: Late-Life Depression. This occurs in elderly people with no prior personal or family history of depression, but often with risk factors for cardiovascular disease, such as hypertension, diabetes mellitus, smoking or hypercholesterolemia. The patient describes a gradual loss of energy and interest, and a diminishing ability to cope. Cognitive testing may show impairment. Subcortical hyperintensities on MRI are frequent [24]. Antidepressants are effective, as is ECT in severe cases; clinical data supports the use of stimulants such as methylphenidate as well [25]. Smoking cessation, blood pressure control, serum cholesterol level reductions and physical activity may be the most important preventive measures.Type G: Psychotic Depression. This form features delusions and severe disturbances in work and social function. Hyperactivity of the HPA axis as measured by dexamethasone nonsuppression is present in at least half of the patients [26]. Antipsychotic medication in addition to antidepressants is indicated [27]. ECT has been shown to be particularly effective for depression with psychosis.Type H: Atypical Depression. This subtype characterizes patients who show hypersomnia and hyperphagia instead of the insomnia and weight loss typical of acute depression. They have more anxiety, including panic disorder and social phobia, and they are more likely to be a suicide risk and to abuse drugs [28]. Many of these patients may be young people in whom romantic disappointment (not divorce after an ongoing relationship) can be pathogenic. Falling in or out of love has powerful effects on the brain [29]. Several studies have suggested that monoamine oxidase inhibitors [30] are most effective in such patients.Type I: Bipolar Depression. This occurs in patients with previous episodes of mania and should also be considered in depressed patients with a strong family history of bipolar disorder [31]. This subtype could potentially be further subdivided, based on the degree of emotional reactivity during the depression [32]. A recent double-blind placebo-controlled study showed that SSRIs or bupropion do not seem to be effective in bipolar depression [33], whereas TCAs may be effective but induce mania [34,35]. Often, adding a second mood stabilizer is effective [36].Type J: Depression Secondary to Substance Abuse or to a Medical Condition. This subtype, which is recognized by DSM-IV, is a diverse group of disorders that can be difficult to treat [37]. It is a striking biological fact that substances and medical conditions as diverse as therapeutic corticosteroids, illicit cocaine use or pancreatic carcinoma can cause depression. The pathophysiological mechanisms of these depressions may differ from each other, as well as from depressions without underlying substance abuse or medical disorder. The high and perhaps increasing prevalence of MDD in recent decades may be due to inclusion of types A, C, D and J (table 1) in DSM-based epidemiological studies of MDD. Many of these subtypes may not be strongly antidepressant-responsive, and their inclusion in DSM MDD could be responsible for the declining drug-placebo difference in large clinical trials. For example, marital separation causes an increase in the prevalence of depressive symptoms, even among individuals who have not suffered depression in the past [38]. Forced geographic displacement, financial loss and economic hardship [6] are other stressors that have been documented to cause significant increases in depression. Many people who develop depression following these stressors are diagnosed with MDD [8,39]. Is there justification for relating to their condition differently from the way one relates to a bereavement-induced depression? There is some reason to regard a passing depressive reaction in the face of adversity as an evolutionarily adaptive response. In these instances, supplying diagnostic labels and administering a course of therapy could be counterproductive [40].Diagnostic theory always faces a tension between lumpers, who would merge diverse clinical phenomena under one diagnostic rubric, and splitters, who prefer to maintain multiple diagnoses. In the case of depressive disorder, DSM-IV cast its lot with the lumpers. This had the beneficial effect of bringing order to the field, and spurred a surge in research. However, the time may have come to consider a shift of the pendulum toward splitting. The validity of our proposed subtypes as distinct forms of depression is an open question. Any clear demarcation of different subtypes of a phenomenon as complex as depression will always be tenuous, so subtypes might actually be better considered archetypical descriptions of standard cases with which the details of any actual patient may more or less closely coincide [41,42]. However, the creation 30 years ago of MDD as a single catchall entity was a consensus decision, and not an empirical result. Using MDD criteria, meta-analyses show only about a 50% response rate to modern antidepressant treatment with about a 40% response rate to placebo [10,43]. No clear gene linkages have emerged, and no common monoamine or cortisol axis pathophysiological pathway has crystallized [2]. FDA-approved treatments have proliferated based on increasingly expensive and increasingly large trials to prove the increasingly small drug-placebo difference. Epidemiological incidence rates have skyrocketed [6]. Perhaps it is time for a paradigm shift. The subtypes discussed above can be operationalized, and new treatment trials powered for the smaller populations available for each subtype. Perhaps we do not need new 'me-too' compounds with a small effect size for 'major depressive disorder'. The large pool of world depression investigators occupied with multicenter clinical trials of diminishing returns might be encouraged to initiate smaller trials in some of the specific subtypes. There is no guarantee of success in such a paradigm shift, but it is time for a change. Subtyping of depression could lead to identification of subtypes that are more responsive to current pharmacological treatment, and aid in separating out the large burden of worldwide depression for which current antidepressants are not a highly effective treatment [6]. This could help resolve the controversy over the appropriateness of current antidepressant education campaigns in the developing world [44]. Grateful thanks are offered to the librarian Iris Arad for her assistance in providing material for the article and to Yehudit Curiel for editing work and the bibliography.P.L. and R.H.B. participated equally in the writing of this paper. R.H.B. is President of the International College of Neuropsychopharmacology, a worldwide organization that receives funding from many pharmaceutical companies as well as membership dues.