Typus melancholicus and the Temperament and Character Inventory personality dimensions in patients with major depression
Correspondence address: SatoshiKimuraMD , Department of Psychiatry, Kyoto First Red-Cross Hospital, 15-749 Honmachi Higashiyama-ku, Kyoto 605-0981, Japan. Email: firstname.lastname@example.org
Although many clinical studies have been conducted to determine the etiological role and clinical implications of typus melancholicus for unipolar depression, maladaptive personality features in depressive patients have not been well described. This study explores typus melancholicus, as measured by the rigidity subscale of the Munich Personality Test, and maladaptive personality features, as measured by the Temperament and Character Inventory (TCI), in 131 remitted patients with DSM-IV major depression and 154 normal controls. The patients reported significantly higher scores on rigidity and harm avoidance and significantly lower scores on self-directedness and cooperativeness. Only 23.6% of the variance of the rigidity scale was explained by the variance of the seven TCI scales, in which only persistence was significantly correlated positively to rigidity. Cluster analysis identified four subgroups, two of which were characterized by a high rigidity score. One of these two subgroups showed no maladaptive personality features, as measured by the TCI, while the other showed high harm avoidance and low self-directedness. These results indicate that the personality of depressive patients is characterized not only by typus melancholicus but also by maladaptive personality features, that typus melancholicus is not well represented by any TCI scale, and that typus melancholicus and maladaptive personality features can coexist in some depressive patients.
Typus melancholicus is a type of personality that was originally hypothesized by Tellenbach, 1 a German theorist, to have a specific etiological relationship to unipolar depression. This hypothesis is widely accepted by Japanese psychiatrists. 2–4 According to Tellenbach and other Japanese theorists, 2–4 this type of personality is essentially characterized by a kind of obsessionality that is directed to the pursuit of identification with social norms and much consideration for other people around the individual. Some authors 3,5,6 emphasize that the obsessionality of typus melancholicus should be differentiated from the obsessio-nal personality in the psychoanalytical literature; 7 the latter is typically directed to the domination and manipulation of other people around the individual. Many empirical studies 5,6,8–13 using personality inventories for assessing typus melancholicus have shown that this type of personality may be specifically related to unipolar depression, although some studies have failed to obtain similar results. 14,15 Several studies with family study designs 16–18 have found that first-degree relatives of probands with affective disorders are also characterized by this type of personality; these results suggest that this personality may be a mediator between the genetic factor and the onset of affective disorders.
The relationship between typus melancholicus and unipolar depression has long fascinated Japanese psychiatrists. A number of psychopathological studies in Japan have emphasized a possible etiological role and clinical implications of typus melancholicus in unipolar depression. These studies seem to have not paid sufficient attention to systematic evaluations of personality features other than typus melancholicus in depressive patients. However, is the personality of depressive patients characterized only by typus melancholicus? Contrary to Japanese psychiatry, many studies from Western countries appear to be interested in describing maladaptive personality features as defined by personality disorders. 19–24 These studies have found that at least 50% of depressive patients are diagnosed as having a comorbid personality disorder. Furthermore, some studies 25–33 have suggested that such maladaptive personality features as seen in depressive patients may be of clinical importance in predicting the response to biological treatments and the long-term prognosis, although these maladaptive personality features have been shown to be less important for the etiology of depression. 34 Admittedly, typus melancholicus is an interesting personality feature that is possibly related to the etiology of unipolar depression, but maladaptive personality features may also be important in developing and understanding the complex relationship between personality and depression.
Why have maladaptive personality features in depressive patients been almost ignored in Japanese psychiatry? One possible reason is that in Japanese psychiatry it appears to have been believed that subjects with typus melancholicus are less likely to have significant inner conflicts or daily problems surrounding social adjustments that may induce clinicians to make a diagnosis of a comorbid personality disorders (namely, typus melancholicus might be negatively correlated to maladaptive personality features). However, such a belief has not been tested in empirical investigations. In our own clinical experiences, this belief in the Japanese psychiatry does not seem to be applicable to all depressive patients. Although some depressive patients with typus melancholicus appear to suffer from no daily problems surrounding social adjustments and no significant inner conflicts, other patients appear to be characterized by high levels of inner conflicts surrounding their social adjustments, even though they clearly show personality features of typus melancholicus.
The present study has attempted to investigate the relationship between typus melancholicus and maladaptive personality features in depressive patients (who had been remitted from DSM-IV major depressive episodes). This study explores typus melancholicus, as measured by a self-rating questionnaire, and maladaptive personality features, as measured by the Temperament and Character Inventory (TCI) 35,36 personality dimensions, in 131 remitted patients with major depression and 154 normal controls. The TCI 35,37 is a self-rating personality questionnaire developed by Cloninger’s biosocial theory of personality. This theory is becoming very popular in the psychiatric field because it refers to proposed biological bases for the development of personality, and its personality questionnaires have a strong capability to describe maladaptive personality features. The TCI produces seven personality dimensions, some of which have been shown to be related to maladaptive personality features.
The present study focused on three issues: (i) whether typus melancholicus and the TCI dimensions are related to the diagnosis of major depression (DSM-IV); (ii) to what extent typus melancholicus is explained by the TCI dimensions; and (iii) whether depressive patients with a high typus melancholicus score show maladaptive personality functions.
Three self-rating questionnaires, von Zerssen’s F-list, 38 Kasahara’s scale, 39,40 and the rigidity subscale of the Munich Personality Test (MPT), 41,42 are available in Japan for assessing typus melancholicus. Of the three questionnaires, the rigidity subscale of the MPT was used in the present study because its high internal consistency and high test–retest reliability have been reported as a result of previous studies. Several studies have provided evidence that the score on rigidity is specifically high in patients with major depression, 41,42 and that first-degree relatives of affective probands report a high score on rigidity; these results indicate that the rigidity scale has extremely high validity. 16–18
The TCI 35,36 is a 240-item true–false self-rating personality inventory that was developed to assess the four temperaments as well as the three character dimensions proposed by Cloninger’s biosocial model of personality. The four temperament dimensions are: (i) novelty-seeking (exploratory activity in response to novelty, impulsive decision-making, active avoidance of frustration); (ii) harm avoidance (pessimistic worry in anticipation of future problems, passive– aggressive behaviors, shyness with strangers); (iii) reward dependence (sentimentality, social attachment, dependence on approval of others); and (iv) persistence (ambitious overachieving, industriousness, persistence). Cloninger 35,37 has proposed that some of these temperaments are independently heritable and related to neurotransmitter functions. Descriptions of the three character dimensions are based on three aspects of the development of self-concepts related to identification of self as an autonomous individual, an integral part of humanity, and an integral part of the universe as a whole. The three character dimensions are: (i) self-directedness (self-determination, will-power, and ability to control, regulate, and adapt one’s behavior); (ii) cooperativeness (social tolerance, empathy, helpfulness, and compassion); and (iii) self-transcendence (acceptance of ambiguity and uncertainty, spiritual acceptance, and identification with the wider world). Several studies have shown that some personality dimensions of the TCI are strongly related to maladaptive personality features, as defined by personality disorders. Svrakic et al.43 have found that low self-directedness and low cooperativeness are core features of DSM-III-R 44 personality disorders, and that low reward dependence, high novelty-seeking, and high harm avoidance are significantly related to symptoms of clusters A, B, and C personality disorders, respectively.
It is well known that clinical diagnoses of depressive disorders are much more reliable when using appropriate diagnostic criteria. Therefore, we used the DSM-IV 45 diagnostic criteria for inclusion of the patients in this study. The subjects of the present study were consecutive out-patients with primary major depression who sought treatment at the Psychiatric Department in Fujita Health University Hospi-tal for 2 years, and who participated in a 26-week outcome assessment. As the 26-week outcome assessment, the depression severity was measured by the Hamilton Rating Scale for Depression (HAMD) 46 and the Beck Depression Inventory (BDI). 47 Inclusion criteria required that patients be 20 years of age or older with a primary diagnosis of DSM-IV major depressive disorder, 45 either single episode or recurrent, unaccompanied by psychotic or catatonic features. Patients with a past history of mania or hypomania as well as those with severe physical illness or possible brain-organic etiology were excluded. A total of 140 depressive patients were included in the study.
These patients were treated with antidepressants mainly with a naturalistic design (i.e. without any limitation of the prescription of the psychotropic drugs to the patients). Within a few days after this 26-week outcome assessment, the patients were asked to complete two self-reporting personality questionnaires (TCI and MPT). The depression severity was significantly reduced at the 26-week outcome assessment (the mean HAMD was 4.1 (SD = 4.0) for the patients included). However, a total of nine patients were not remitted (the HAMD score was higher than 7). To avoid a significant effect of depression on the personality assessments, these patients were excluded. In total, 131 remitted patients were included in the present study.
As a normal control group, 154 adults, who were working in an industrial company, completed the two self-reporting personality questionnaires (TCI and MPT) and the BDI. The normal controls included 90 (58%) women and 64 (42%) men, with a mean age of 43.2 years (SD = 12.1; range 20–65 ). The mean BDI score was 6.0 (SD = 6.5) when the controls filled out the personality questionnaires.
Table 1 demonstrates the demographic and clinical variables in the two subject groups in this study. Age, sex ratio, and the BDI score at personality assessments were not significantly different.
Table 1. Demographic and clinical variables of patients with major depression and controls
|Sex, female, n (%) ||78 (60)||90 (58)||χ2 (d.f. = 1) = 0.07, P = 0.79*|
|Age, mean (SD)||43.6 (12.4)||43.2 (12.1)||t (d.f. = 283) = 0.26, P = 0.79†|
|Onset age, mean (SD)||38.5 (14.0)|
|Melancholia, n (%) ||71 (54)|
|Duration, week (SD)||38.3 (72.8)|
|No. previous episodes, n (SD)||0.9 (1.8)|
|Hamilton Rating Scale for Depression at|
baseline, mean (SD)
|Beck Depression Inventory at personality|
assessments, mean (SD)
|7.7 (6.3)||6.0 (6.5)||Z = 0.36, P = 0.36§|
Age, sex, and depression severity are known to affect self-rating personality assessments. In order to compare the rigidity scale and the seven TCI scales between the depressive patients and normal controls, two-way analysis of variance ( ANOVA) was performed with the diagnosis and sex as independent variables and with age and the BDI score at the personality assessments as covariates. In order to clarify the relationship between the rigidity scale and the TCI scales, a multiple linear regression analysis with the rigidity score as the dependent variable was performed for the patient data. The results of the linear regression analysis can also be affected by age and depression severity. The regression analysis was therefore performed in a hierarchical manner in order to eliminate the effects of age and depression severity at personality assessments. Cluster analysis (Ward method) was used to seek depressive subgroups that differed from each other on the rigidity scale and the TCI scales. All analyses in this study were performed using SPSS (Tokyo, Japan) for Windows.
Table 2 shows the results of comparisons of the rigidity scores and TCI scores between the two groups. To minimize the chance of type I error, the significance level for each analysis was set at P < 0.006 (0.05/8) by using Bonferroni’s adjustments. A significant effect of sex was found on reward dependence; across the groups, women reported a significantly higher score on this scale than did men. Significant effects of diagnosis were found in harm avoidance, self-directedness, cooperativeness, and rigidity; the patients reported significantly higher scores on harm avoidance and rigidity and significantly lower scores on self-directedness and cooperativeness than did the normal controls.
Table 2. Observed mean scores (SD) on the Temperament and Character Inventory dimensions and typus melancholicus
|Harm avoidance||25.9 (5.3)||25.5 (6.4)||16.8 (6.1)||18.7 (5.7)||0.40||53.24**|
|Novelty-seeking||18.8 (5.2)||18.9 (4.7)||19.8 (4.5)||19.7 (4.6)||0.94||1.51|
|Reward dependence||13.7 (3.0)||15.5 (2.9)||14.5 (3.0)||15.4 (2.8)||11.38**||0.00|
|Persistence||3.9 (1.7)||4.1 (2.1)||4.5 (1.9)||4.3 (1.8)||0.99||0.75|
|Self-directedness||22.6 (7.1)||22.6 (7.3)||29.4 (6.0)||25.6 (6.3)||3.81||11.82**|
|Cooperativeness||26.7 (5.9)||27.9 (5.9)||29.9 (5.8)||30.9 (4.6)||2.65|| 8.41*|
|Self-transcendence||10.3 (5.4)||9.8 (4.9)||10.0 (6.3)||10.9 (5.8)||0.46||1.01|
|Rigidity||11.3 (4.7)||10.5 (4.9)||8.1 (4.3)||8.0 (4.1)||1.38|| 9.23*|
Table 3 demonstrates the relationship of the rigidity scale to the TCI scales in depressive patients. In order to eliminate the effects of age and depression severity, a hierarchical multiple linear regression analysis was used with age and depression severity being entered at the first step and all TCI scores being entered at the second step. The dependent variable was the rigidity score in the analysis. As shown in Table 3, the model improvement from step 1 to step 2 was highly significant (F change = 5.56, d.f. = 9,121, P < 0.001). R2 change at the second step was 0.236; this indicates that the variance of all TCI scores explains 23.6% of the variance of the rigidity score. In the model at step 2, persistence was positively correlated to rigidity (P < 0.01), while novelty-seeking tended to be negatively correlated to rigidity (P < 0.1).
Table 3. Relationship of TCI dimensions to rigidity after eliminating out the effects of age and depression severity (results of hierarchical linear multiple regression analysis)
|Step 1: R2 = 0.037|
| F(2, 128) = 1.74, P < 0.10|| || || ||(d.f. = 128)|
| ||BDI score||0.001||0.046||0.002||0.02||0.984|
|Step 2: R2 = 0.273 (R2 change = 0.236)|
| F(9, 121) = 3.55, P = 0.001|
| (F change (7, 120) = 5.56, P < 0.001)|| || || ||(d.f. = 121)|
| ||Harm avoidance||− 0.043||0.117||− 0.053||−0.37||0.713|
| ||Novelty seeking||− 0.234||0.120||− 0.240||−1.95*||0.055|
| ||Reward dependence||− 0.115||0.171||− 0.072||−0.67||0.504|
| ||Self-directedness||− 0.038||0.090||− 0.056||−0.42||0.676|
Table 4 outlines the results of cluster analysis. The four personality scores (rigidity, harm avoidance, self-directedness and cooperativeness) were used in the analysis, since these personality scores significantly differed between the patients and normal controls in Table 2 and were therefore thought to mainly characterize the depressive patients. The cluster analysis procedure in SPSS showed that a four-group solution maximized the differences among the groups. Table 4 summarizes the differences among the four subgroups. The overall group differences in the personality dimensions were estimated by using ANOVA, and thereafter multiple comparison procedures with Bonferroni’s adjusted confidence intervals were performed. Of the four subgroups, two (subgroups II and III) were characterized by high rigidity. These two subgroups reported similarly high cooperativeness scores but were distinguishable from one another based on harm avoidance and self-directedness. While the subjects classified in subgroup II reported scores similar to the normal controls on harm avoidance and self-directedness, the subjects classified in subtype III reported a higher score on harm avoidance and a lower score on self-directedness. In other words, two different subgroups were recognized in depressive patients with high rigidity; one subgroup (III) was, in addition to high rigidity, characterized by significant personality deviations from normal controls on the harm avoidance and self-directedness dimensions; and the other subgroup (II) showed no such additional deviations on the TCI dimensions. The personality of the latter subgroup was characterized mainly by high rigidity.
Table 4. Personality features characterizing four subgroups of depressive patients produced by cluster analysis
|Age, mean (SD)||44.6 (11.8)||48.9 (12.2)||50.3 (15.7)||47.7 (13.5)||F (d.f. = 3, 127) = 0.72†|
|Sex, female, n (%) ||14 (50.0)||15 (53.6)||20 (62.5)||29 (69.0)||χ2 (d.f. = 3) = 2.60§|
|Harm avoidance||29.4 (4.2)||18.7 (5.2)||29.6 (4.1)||24.9 (4.0)||F (d.f. = 3, 127) = 31.6†,*||I = III > IV > II|
|Self-directedness||16.5 (4.2)||29.0 (4.7)||16.5 (4.6)||27.5 (3.7)||F (d.f. = 3, 127) = 18.9†,*||II = IV > I = III|
|Cooperativeness||19.3 (3.7)||32.6 (4.1)||29.6 (3.1)||27.6 (3.8)||F (d.f. = 3, 127) = 51.8†,*||II = III = IV > I|
|Rigidity||9.1 (4.3)||14.6 (4.7)||13.4 (3.7)||7.6 (3.0)||F (d.f. = 3, 127) = 58.7†,*||II = III > I = IV|
The other two subgroups (I and IV) identified by cluster analysis were characterized by low rigidity and high harm avoidance. While subgroup IV reported scores similar to the normal controls on self-directedness and cooperativeness, subgroup I was also characterized by low self-directedness and low cooperativeness. There was no significant difference in age and sex ratio among the four subgroups identified by cluster analysis, indicating that the four subgroups were distributed independently of age and sex.
The results of the present study indicate that the self-rated personality of depressive patients is, compared with the self-rated personality of normal controls, characterized on the whole by a higher score on typus melancholicus, as measured by the rigidity subscale of the MPT. Although this study did not explore control subjects with mental disorders other than major depression, the results are consistent with the view in Japanese psychiatry that typus melancholicus may be a premorbid personality feature of depression.
It appears, however, that the personality of depressive patients is characterized not only by a higher score on the typus melancholicus scale but also by deviated scores on some TCI personality dimensions. As shown in Table 2, depressive patients reported a higher score on harm avoidance and lower scores on self-directedness and cooperativeness. Several studies 43,48 have consistently found that low self-directedness and low cooperativeness are significantly related to maladaptive personality functions (i.e. personality disorders as defined by DSM-III and its revised versions), and high harm avoidance is significantly related to symptoms of cluster C personality disorders. Based on these significant relationships between the three TCI dimensions and personality disorders, the results of this study can be interpreted as indicating that the self-rated personality of depressive patients is, on the whole, characterized by both typus melancholicus and some deviated personality features suggestive of the presence of personality disorders.
The concepts that depressive patients are characterized by maladaptive personality features does not seem to have been well accepted in Japanese psychiatry, although many studies including those conducted in Japan have shown at least 50% of depressive patients are diagnosed as having one comorbid personality disorder. 19–25,49 Furthermore, there is some evidence that comorbid personality disorders in depressive patients are predictive of the response to biological treatments, 25–32 although typus melancholicus has not been shown by several studies to be related to antidepressant responses in major depression. 6 Maladaptive personality features, as measured by the TCI, have also been reported to be related to a poorer response to antidepressant treatments. 50 Some studies have found that comorbid personality disorders or maladaptive personality features in depressive patients predict long-term outcome including converts into bipolar disorders. 33 Given the evidence of these previous studies, the results of the present study suggest that much more attention should be paid to maladaptive personality features in order to describe the various aspects of personality in depressive patients and to more thoroughly understand the relationship between personality and the outcome in depression.
It is interesting to determine whether typus melancholicus is related with the TCI scales, and to what extent typus melancholicus is explained by the TCI scales. The multiple linear regression analysis in the present study indicated that only 23.6% of the variance of typus melancholicus, as measured by the rigidity scale of the MPT, was explained by the variance of TCI scores. This means that typus melancholicus is relatively independent of any TCI scale. The highest partial correlation to the rigidity scale was seen for persistence (r = 0.318, P < 0.01); but this TCI dimension did not differentiate depressive patients from normal controls. These results suggest that typus melancholicus is moderately correlated to persistence but is largely different from persistence in the diagnostic performance. Typus melancholicus may be a special personality feature, the constructs of which cannot be represented by any TCI scale.
The finding that typus melancholicus is relatively independent of any TCI scale suggests that in some depressive patients, typus melancholicus and maladaptive personality features, as measured by the TCI, can coexist. This was confirmed by the cluster analysis conducted in the present study. Among the four subgroups identified by the analysis, two subgroups with a high typus melancholicus score were found: one (subgroup II) was characterized only by a high typus melancholicus score, but the other (subgroup III) was characterized not only by a high typus melancholicus score but also by maladaptive personality features such as high harm avoidance and low self-directedness. High harm avoidance and low self-directedness suggest that maladaptive personality features, as seen in subgroup III, are mainly characterized by features of cluster C personality disorders. The results of the cluster analysis suggest that although typus melancholicus can be seen in some patients without maladaptive personality features, it is also seen in other patients together with maladaptive personality features related to cluster C personality disorders.
The results of the cluster analysis are inconsistent with what has generally been believed in Japanese psychiatry, but are consistent with our clinical experiences described in the introduction. A systematic study by one of the authors (TS) 51 has also shown that in another sample of unipolar depression, some depressive patients with a high typus melancholicus score have severe personality disorders. It should be noted that two empirical studies favor the view that typus melancholicus can, in some depressive patients, coexist with maladaptive personality features, indicating that the belief in Japanese psychiatry is not underpinned by scientific evidence.
Some studies 51–53 have suggested the possibility that depressive patients who show both typus melancholicus and maladaptive personality features may be classified into ‘atypical’ subtype of major depression; this subtype has been newly adopted by DSM-IV. It is unknown, however, whether depressive patients who show both typus melancholicus and maladaptive personality features differ from other depressive patients in terms of other variables such as clinical features, treatment responses, long-term prognosis, and family history. These issues need to be clarified by further study; such a study may make a large contribution to identifying the possible importance of typus melancholicus for DSM-IV subtypes of major depression.
The cluster analysis in this study found two subgroups with a low typus melancholicus score. The number of patients classified into these subgroups was not small. One of the two subgroups (subgroup I: n = 28) was characterized by high harm avoidance, low self-directedness, and low cooperativeness, while the other (subgroup IV: n = 42) was characterized only by high harm avoidance. These results indicate that more than a small number of patients with major depression may not show typus melancholicus, and that the TCI scales may have a strong capability of describing the personality of depressive patients who lack the personality features of typus melancholicus.
Because our exclusion of the nine depressive patients who were not remitted at the 26-week outcome assessment may have selection-biased our results, we replicated our analyses using the entire sample with the nine patients included. Both sets of results were similar, indicating that our findings were not significantly distorted by selection bias. It is interesting to speculate that the personality features that were related to major depression reflect premorbid personality function. The personality assessments in the present study were performed when the patients were remitted, assuring us that the results in the present study have not been distorted by the state effects of depression. In addition, recent longitudinal studies 54–56 have indicated that self-rated personality features, such as neuroticism and similar other personality dimensions, are not significantly affected by prior experience of depressive episodes, as based on both epidemiological and clinical samples. These studies favor the view that personality deviations related to major depression strongly reflect the premorbid personality function. However, prospective studies will be necessary in the future in order to replicate the results of this present study.