Struggle for Subtypes in Primary and Secondary Depression and Their Mode-Specific Treatment or Healing

The criteria published by Feighner et al. [1] from Washington University in St. Louis in 1972 marked a turning point in the modern American attempt to standardize psychiatric diagnoses suitable for current clinical research. The Feighner criteria or ‘St. Louis criteria’ [2] were used by DSM-III [3] in its atheoretical approach to making a purely symptomatic description of the diagnostic categories. However, the attempt to differentiate between ‘primary’ and ‘secondary’ syndromes within depressive disorder was not fully adopted by DSM-III. As discussed by Guze [4], the ‘St. Louis group’ borrowed this issue of speaking about primary and secondary depression from general medicine when dealing with the problem of co-morbidity. Medical disorders are classified on the basis of such issues as symptomatology, pathophysiology and etiology [5]. As regards major depression according to DMS-III or DSM-IV [6], the pathophysiological mechanism is still unknown [7], implying that the validity of the DSM-IV criteria for depression is around zero if the index of validity is the pathophysiological mechanism [8]. As discussed by Frank and Frank [9], the symptomatic approach behind DSM-III was considered to be ‘atheoretical’ but actually constitutes a theory in itself because factors such as the meaning the patients themselves attach to their symptoms or their social and historical antecedents is seen as epiphenomena, i.e. not central to the DSM-III diagnosis. Conversely, in psychotherapy these factors are mode-specific [9].When Greenberg interviewed me about the history of the Hamilton Depression Scale (HAM-D17) [10] which is a major platform in his book [8], we were very much in agreement that the Feighner depression criteria have a strong resemblance to the items on the HAM-D. According to Horwitz and Wakefield [11], one of the ironies of psychiatric history is that the theory-neutral symptoms of depression in the Feighner criteria were apparently adopted by DSM-III due to their empirical support, but a closer look gives rather fragile evidence of their validity [11,12]. The most obvious reference for the Feighner symptoms of depression is the HAM-D. Another irony is, as pointed out by Guze [4], that the concept of co-morbidity was opened up in DSM-IV for scientific studies by eliminating diagnostic hierarchies; however, this was not followed by an introduction of the primary-secondary dichotomy of depression, even though Feighner had indicated that the symptomatology was the same in primary and secondary depression [13].Two decades before Frank’s first edition [14] or the introduction of antidepressive medication, Lewis [15] remarked that the first task of classification in depression is, like in hypertension, to recognize the clinical state, i.e. the state of abnormal functioning. The second task is to differentiate between primary and secondary types as to etiology. In the case of hypertension, we now define this as a measured blood pressure of 140/90 mm Hg or greater [16]. In the case of depression, we now define clinical (major) depression as a HAM-D17 score of 18 or higher, whereas a score between 13 and 17 indicates ‘less than major depression’ [17].In the case of hypertension, secondary conditions such as cardiac disorder, disorders of the kidneys or adrenal glands are taken into account, and if no such conditions are in operation we refer to the state as primary hypertension [16]. Lichtenberg and Belmaker [18] have suggested a subtyping of major depression with associated mode-specific treatments. In this struggle for a subtyping in primary and secondary depression (table 1) the medical disease model is secretly in operation, thereby making a new attack on the ‘atheoretical’ DSM-III/DSM-IV approach [5].Frank and Frank [9] have actually also proposed a subtyping of DSM-III. They found such primary depression subtypes as psychotic depression and bipolar or unipolar depression to be associated with drug therapy. These conditions, as well as neurotic depression, stress-induced secondary depression and posttraumatic stress disorder (PTSD) were considered to belong within the medical disease model. However, Frank and Frank [9] introduced a category of ‘disconcerted behavior’ (the feeling of not getting enough out of life, an existential despair). They had problems in categorizing ‘disconcerted behavior’ as an illness, and they found that persons in this category typically received Freudian or Jungian psychoanalysis.Quite recently the contents of table 1 have been subjected to attacks, not only by Greenberg [8] on the manufacturing of depression as a modern disease but also by Pigott et al. [19] with special focus on the analysis of Turner et al. [20 ]and the STAR*D study [21]. In the following, these struggles with the subtyping of depression to identify common factors and mode-specific factors within the medical disease model will be evaluated.In table 1 the subtypes of depression have Roman numerals whereas Arabic numerals designate the HAM-D [10] or the Beck Depression Inventory (BDI) [22] (implying that it is possible to indicate mean and standard deviation for the total scores on HAM-D or BDI).Psychotic depression (I), bipolar depression (II), and unipolar depression (III) are subtypes of depression not considered to be secondary to either stress-induced disorders or somatic disorders such as post-stroke depression [18]. These three primary types of depression (I, II, III) are seen as examples of classical diseases, also by Greenberg [8], who considered the stress-induced depressions (V, VI, VII in table 1) to be examples of the secret history of a modern disease [8].DSM-IV lists atypical depression both within major depression and within dysthymia. Because the endogenous or melancholic features in DSM-IV are only valid for major depression, the concept of atypical depression is regarded as a non-melancholic subtype. However Vanggaard [23,24] has found atypical depression to be a subtype of endogenous depression. Thase [25] has recently made an attempt to revise the DSM-IV criteria for atypical depression, with greatest emphasis on the personality dimension. According to Vanggaard [23,24], patients with atypical depression have no atypical features as regards melancholia, the atypical features are to be found within the neurovegetative symptoms [23,24].In contrast to atypical depression, in DSM-IV the melancholic features are listed within seasonal disorder and consequently this leaves no room for atypical symptoms in the DSM-IV description of seasonal affective disorder However, Williams et al. have designed a seasonal affective disorder version of HAM-D to capture the atypical symptoms not included in HAM-D17[26].Table 1 shows the three suptypes of depression identified by Lichtenberg and Belmaker [18] as stress-related within major depression. Especially these subtypes (V, VI, VII) are the ones Greenberg considers to form the secret history of a modern disease, used to justify the use of antidepressive medication. As discussed by Casey et al. [27] the diagnostic behavior implies that patients fulfilling the diagnosis of major (moderate) depression with clearly stress-related problems will only be coded for their major depressive condition. The use of subtyping as suggested by Lichtenberg and Belmaker [18] (table 1) might give an opportunity to perform evidence-based studies.Affective disorders such as postnatal depression (VIII), post-stroke depression (IX) and depression after substance abuse disorder (X) are rather common [28]. Use of the HAM-D to define the clinically significant criteria for depression is more valid than use of the BDI as discussed by Greenberg [8].In his treatise, Greenberg [8] confesses that when he learnt about the research activities of the Mood Disorder Unit at Massachusetts General Hospital, he ‘jumped at’ this opportunity to be diagnosed by a highly respected Harvard psychiatrist. As he obtained a HAM-D score of 18 as well as fulfilling the DSM-IV criteria for major depression, he was qualified to participate in a placebo-controlled clinical trial. As a practicing psychotherapist of many years standing, Greenberg himself felt that he was suffering from dysthymia. During the 4 weeks in the trial, his HAM-D dropped from 18 to 14 and later on he learnt that he had received placebo.As discussed by Boyer and Feighner [29] a ‘double standard’ was introduced with the Feighner criteria for major depression as the term ‘definite’ was used in connection with research while the term ‘probable’ was used in connection with daily clinical practice. Thus, dysthymia might be considered in practice as ‘probable’ or in research as ‘less than major depression’. On the other hand, the study on co-morbidity which the DSM-IV made possible might define a case as ‘double depression’, i.e. the co-existence of major depression and dysthymia [30].Adjustment disorder with depressive symptoms is by definition a case of less than major depression according to ICD-10 [31], as are the anxiety disorders, including panic disorder, phobia, and obsessive-compulsive disorder. PTSD seems to fulfill the criteria for a medical illness (based on shared symptomatology and etiology) [32].The patient’s right to effective treatment is the pragmatic consequence of the DSM-III diagnosis of major depression [33]. Thus, DSM-III acknowledged that major depression implies a history of good response to adequate somatic therapy [29]. Treatment response was among the validation criteria for primary depression in the Feighner approach [34].Table 1 shows the treatment modalities (both somatic and psychosocial) in accordance with Lichtenberg and Belmaker [18], although modified. The means ± SD for HAM-D17 and BDI are baseline scores. As it is outside the scope of this paper to give the exact values – not least because they are difficult to obtain – some landmark studies have been used here. The HAM-D scores are most useful, and the listing from I to XI in table 1 showing a HAM-D17 mean score from 14 to 30 is reflected in the outcome of the cluster analysis performed by Paykel [35]. He showed that severity of depression is related to the subtypes, i.e. the primary depressive categories with melancholic features have the highest severity scores, the secondary depression categories with anxiety have intermediate scores, and the category of dysthymia the lowest scores.Support for the somatic therapies identified by Lichtenberg and Belmaker [18] concerning the primary depression with melancholic features landmark studies with electroconvulsive therapy [36], with tricyclic antidepressants [37], and with monoamine oxidase inhibitors [38] are included in table 1. To cover the secret history of depression as a modern disease [8], the meta-analysis of the second-generation antidepressants performed by Turner et al. [20], the meta-analysis of psychotherapy performed by Cuijpers et al. [39] as well as the ‘real-world’ study on antidepressants, the STAR*D study [21], are included in table 1.In the Lauritzen et al. [36] study inpatients with major depression and with a baseline HAM-D17 mean score of 30 were treated with ECT. These patients covered the subtypes of I, II, and III (table 1) and during around 6 weeks of therapy the HAM-D17 dropped below 10 in approximately 90% of the cases [36]. In the Danish University Antidepressant Group study [37], a tricyclic antidepressant (clomipramine) was compared to a serotonin-specific reuptake inhibitor (SSRI) (citalopram). The patients were inpatients without psychotic features but covering both bipolars and unipolars. Over 5 weeks of treatment, HAM-D17 dropped below 7 (remission) in 60% on clomipramine, but only 30% on citalopram (p ≤ 0.01).The second meta-analysis on the treatment of atypical depression [38] showed that when compared to placebo, the monoamine inhibitor phenelzine obtained an effect size of 0.45, indicating a clinically significant outcome [5]. The prevalence of atypical depression in the primary care setting is approximately 30% of those diagnosed as DSM-IV major depressives [38].A meta-analysis of second-generation antidepressants versus placebo in patients with major depression using data on all trials submitted to the US Food and Drug Administration (FDA) has recently been released by Turner et al. [20]. In total, 12 antidepressant agents were analyzed and the HAM-D [10] was used as outcome measure while effect size statistics were used to demonstrate the clinical response of these agents compared with placebo in the acute 6–8 weeks’ therapy. In these trials, involving 12,564 patients, those with psychotic depression (I) or bipolar depression (II) were excluded, as were patients with depression secondary to somatic conditions (VIII, IX, X). However, no information about the number of patients with stress-related depression (V, VI, VII) was available, or, within primary depression, the ratio between unipolar or atypical depression [20].The outcome of the meta-analysis by Turner et al. [20] was an effect size of 0.31 which was very similar to that obtained by Kirsch et al. [40] using a subgroup from the FDA studies. The HAM-D was also used in the analysis of Kirsch et al. [40] but it should be recalled that the HAM-D was originally developed for patients with primary depression, covering types I, II and III in table 1[10].Greenberg [8] interviewed Arvid Carlsson who had actually used clomipramine to identify serotonin reuptake inhibition as an active mechanism of action in antidepressants [41]. Against this background, Carlsson developed the first SSRI compound (zimeldine). A review of the first controlled clinical trials in outpatients with major depression using zimeldine against amitriptyline demonstrated that amitriptyline was superior on non-core depressive symptoms, such as the sleep factor of the HAM-D, while no difference between the two drugs was seen concerning the HAM-D core symptoms of depression, such as depressed mood, guilt, work and interests, psychic anxiety, psychomotor retardation, and general somatic symptoms [42]. When evaluating the specific antidepressive effect, therefore, these core items (HAM-D6) are to be used as the outcome measure and when using this scale the effect size is 0.40 or higher when SSRIs are compared to placebo [43]. In the studies included in the meta-analysis of Turner et al. [20], the HAM-D had to be 18 or more in order for subjects to suffer from major depression. In the STAR*D study [21], the patients needed a baseline HAM-D score of 14 or more to be included; this is a rather low score, as discussed by Pigott et al. [19], but seems to cover the many subtypes of depression as shown in table 1.Beck’s cognitive therapy was compared to imipramine in a controlled clinical trial [44] in patients with a baseline HAM-D mean score of 18 and fulfilling the DSM-II criteria of depressive neurosis corresponding to the DSM-IV criteria for dysthymia. The results showed that both therapies were effective, although no placebo-imipramine arm was included. As discussed by Greenberg [8], Beck’s psychological theory of depression is framed within the medical disease model in the same way as the trials collected by Turner et al. [20] or in the STAR*D study [21]. According to Greenberg [8], the pathological impairment in Beck’s model of depression is that of having an impaired or biased view of the world. Consequently, the essential element of cognitive psychotherapy [44] is to teach the depressed persons, the patients, to realign their thinking with reality. When these patients are cured, they are ‘... smoothly functioning professors of information, resilient navigators of life’s ebbs and flows ...’ [8].As discussed by de Figueiredo [45] the term ‘demoralization’ is the common factor identified in candidates for psychotherapy whatever their diagnostic subgrouping within stress-related secondary depression (table 1). These candidates are conscious of having failed to meet their own expectations or those of others, or of being unable to cope with some pressing problems. They feel powerless to change the situation or themselves. These candidates seem to fulfill Beck’s clinical theory of depression which focuses on a negative evaluation of the past (guilt), the present situation (helplessness) and the future (hopelessness) [46]. Frank and Frank [9] compare the ‘anti-demoralization’ effect of psychotherapy and placebo medicine which work through an unspecific epiphenomenological factor. According to Greenberg [8] the relatively high score on the BDI in patients with stress-related secondary depression (table 1) compared to HAM-D might be explained by the many alternative items of the same three variables: guilt, helplessness and hopelessness.The meta-analysis performed by Cuijpers et al. [39] on the most qualified trials with psychotherapy (table 1) comparable with controlled trials of antidepressants [20], i.e. as to randomization, blinding of assessors, standardized manual of the psychotherapy, intention-to-treat analysis, resulted in an effect size of 0.22. As shown in table 1, patients with dysthymia were included in the meta-analysis [39], but not patients with PTSD. However, as the medical illness model is based on symptomatology, pathophysiological mechanisms and etiology [39], PTSD seems to fulfill these criteria to the same degree as major depression according to DSM-IV where the pathophysiological mechanism is still unknown [7]. Frank et al. [47] recommended antidepressants for patients with PTSD.In the STAR*D study, the group identified as having atypical features constituted approximately 20% of the patients, and their HAM-D17 mean score was 22.6 compared to 21.6 in the non-atypical group [48]. Depressed patients with atypical features were shown to be less likely to remit on citalopram than those without atypical features [48,49].As discussed by Hegerl et al. [50], cognitive therapy does not appear to be superior to pill placebo apart from the Jarrett et al. study [51] in which patients with atypical DSM-IV major depression were treated with cognitive therapy or phenelzine in a double-blind placebo-controlled study. The baseline HAM-D here was approximately 18 and the corresponding BDI was 28. After 10 weeks of therapy, the HAM-D dropped to 9 in the two active therapies and to 14 in the placebo group (p ≤ 0.01).When discussing the Turner et al. [20 ]meta-analysis or the STAR*D study [21], Pigott et al. [19] did not take the subgroup of atypical depression into account. It is however a major issue when measuring outcome of antidepressants. Greenberg’s discussion of atypical depression is also very limited [8]. It constitutes an alternative subgroup to dysthymia.Both the atheoretical approach of the DSM-III and DSM-IV the and the very trials of antidepressants than are by Greenberg [8] and Pigott et al. Greenberg not even take into of a placebo effect in operation during the controlled trials of antidepressants other than the ‘anti-demoralization’ effect Hamilton this the contrast or the the that before therapy is when the Hamilton is in a score should be 1 or on a the highest is because the has to obtain a score of 18 or more to be included in the study. weeks the same Hamilton might be in the same but now he or the lowest of this because it good to that patients in the during might up to a on the Hamilton The on the HAM-D from 18 at baseline to 14 at obtained by Greenberg himself when treated with placebo might be such a and not the effect by Frank and Frank Pigott et al. [19] point out that the of a between and in the STAR*D study might the low into that no placebo arm was recommended that two should with his scale which was the case in validation study with his scale found that was a HAM-D17 score of 7 or less when using as index of made an attempt to use this outcome score in studies measuring the of patients with response or response When evaluating the STAR*D study, Pigott et al. [19] on the term depression’, that if the patients are not to to an antidepressant they should not be by such a diagnosis which is not even listed in the As pointed out by Guze [4], the primary and secondary in depressive disorders, based on a for treatment Thus, it is to anxiety symptoms during an of depression are considered as of the depressive illness or as a anxiety disorder. that is the the anxiety disorder will when the depression has been when and psychotherapy has recently been by et al. They the of psychotherapy after a response to The primary and secondary seems to cover this of of psychotherapy with in depressive The psychotherapy by et al. as the seems much more with reference to Frank’s concept of than the cognitive therapy as discussed by Greenberg primary and secondary is also in with the of Pigott et al. [19] that we the new treatment model by et al. in which the contents of table 1 are in an with a an and – who be clinical or social we the description of primary depression by Hamilton in that these of depression in the of or out of to it is difficult to the against this primary-secondary dichotomy the scientific the approach by Lichtenberg and Belmaker [18] is for the to the right treatment for by trials with of the subtypes in table 1. The trials many patients of the classification in table 1. of these patients as in the struggle for the of a high number of from a point of In struggle to obtain a high between response to treatment and depression classification we have to take into the by Pigott et al. [19], Lichtenberg and Belmaker [18] and Greenberg evidence from a point of view that it is that the subtyping factors from I to (table 1) be to one disease major depression. The subtypes have to be received from and was a or of for with an in the drug treatment of affective disorders and