To the Editor: Dr. Reich raises useful points in his letter, in particular by drawing attention to the extensive prior research on the interrelationship between axis I disorders, such as depression and panic disorder, and personality dysfunction. As Dr. Reich points out, over a decade of studies have described a number of ways in which such factors interact.

However, we believe our article is not merely a restatement of existing knowledge; several aspects are noteworthy. First, we reported on a large population of dysthymic patients, more than 400 individuals whose low-grade chronic depression had an average duration of over 30 years. The temperamental abnormalities noted in these subjects at baseline with Cloninger’s Tridimensional Personality Questionnaire (1) (including harm avoidance scores that were nearly two standard deviations above community norms) were comparable to those of more severely symptomatic individuals with disorders such as major depression. Second, our study demonstrated that for our dysthymic subjects, elevated levels of harm avoidance correlated with poor social functioning (as measured with the Social Adjustment Scale) at r=0.50 at both pre- and posttreatment, which implied that temperamental variables may be an important component of these patients’ social dysfunction. Third, to our knowledge, our report is the first large study of dysthymic patients to assess the impact of selective serotonin reuptake inhibitors on temperament and has the advantage of comparison groups receiving placebo or a tricyclic antidepressant.

Dr. Reich’s conceptualization of a "stress-induced personality disorder" resulting from anxiety or mood disorders is one possible explanation for the temperamental abnormalities that have been noted in a variety of such patients. It is also possible that preexisting abnormalities could be a substrate on which such axis I disorders later develop. Prospective studies following large groups of subjects from childhood through adulthood are needed to clarify the sequence in which such disorders appear and to elucidate causative factors.

Finally, we feel that our study raised the provocative issue of treating temperamental abnormalities. Dr. Reich refers to Soloff’s summary of pharmacological options in the treatment of personality disorders (1998). There is indeed a growing literature on drug treatment for borderline and other personality disorders. There is less research on the treatment of temperamental abnormalities such as we observed. In groups of patients with chronic and trait-like disorders such as dysthymic disorder and generalized anxiety disorder, future treatments may be targeted not only at symptom remission but at normalization of coexisting temperamental distortions.