Moreover, low SD is the principal TCI correlate of Axis II pathology. Because we included remitted patients without Axis I or II comorbidity, our results suggest that low SD is related to suicidality and not to the severity of depressive symptoms and personality problems. This result is in accordance with previous studies that reported had lower SD among suicide attempters compared with non-attempters. Moreover, the relatively high ST and low SD scores of suicide attempters in the present study should be viewed in the context of previous studies’ reports of an association between high ST and suicidality, as high ST is associated with low SD and is characterized by illogical, immature, and suspicious behavior. It is unclear why suicidal ideation was observed so frequently in remitted depressive patients in the present study. However, the widely accepted relationship between impulsivity and suicidality may provide an explanation. In recent studies, Ekinci et al. investigated impulsivity, temperament, and character in euthymic patients with major depressive disorder. They reported that impulsivity scores were higher in remitted depressive patients compared with a healthy control group, and also that elevated impulsivity scores were associated with a history of suicide attempts. Accordingly, trait impulsivity may continue to influence suicidal ideation in remitted depressive patients. Moreover, there is evidence of a close relationship between trait anxiety and trait impulsivity, which could also be an important risk factor for suicide ideation. The higher HAMA score at week 12 in the suicidal-ideation and suicide-attempt groups, compared with the non-suicidalideation group, may reflect group differences in impulsivity. The present study had a number of strengths. First, the study used data from the CRESCEND study, which is the largest-scale depression cohort study ever conducted in Asia and includes a wide range of data pertaining to the socio-demographic and clinical characteristics of BEZ235 depressed Korean patients. Second, we excluded depressed patients with psychiatric comorbidities to control for the confounding effects of personality disorders or other Axis I disorders, such as anxiety disorders, eating disorders, and substance misuse-related disorders. Third, we included only patients who were in remission from depressive episodes, and assessed TCI at week 12 to control for the possible effects of depressive symptom severity. The present study also has several limitations. First, we included only Korean patients because cultural issues may lead to variation when assessing dimensions of temperament or character. In a meta-analytic study comparing Cloninger’s temperament dimensions across 20 countries, differences were particularly apparent between Asian and Western countries. The lowest mean score for RD was observed in the Japanese sample, and the highest mean score in PE was observed in the US sample. Moreover, the Asian sample had lower scores compared with other countries for NS, RD, and PE, and higher scores for HA.