To the Editor: In our recent article on the mood spectrum, we showed that in patients with carefully diagnosed recurrent unipolar depression, there is variability in the lifetime experience of manic-hypomanic symptoms and that increased scores on this manic-hypomanic component of our measure of mood spectrum were associated with a higher likelihood of suicidality and paranoia.

It is undeniable, as Dr. Mattes asserts, that the presence of irritability does not necessarily imply mania. Indeed, our conclusions were not based on individual symptoms but on a dimension that includes 60 items, of which only three could be construed to assess irritability. Therefore, although irritability is frequent, it is not the most prominent aspect of the manic-hypomanic component, which includes a range of mood, energy, and cognitive features.

Regarding the attribution of diagnostic significance to "nonspecific symptoms," our intention was not to purport that unipolar patients who have a high number of manic-hypomanic features should be relabeled "bipolar." Still, the linear relationship found between the depressive and the manic-hypomanic components in patients with both unipolar and bipolar disorder when we used a dimensional approach suggests continuity between these disorders. Moreover, we found an association between the manic-hypomanic component and suicidality and paranoia both in unipolar and bipolar patients. In our view, this finding has important clinical implications. The question of whether this dimensional spectrum approach will eventually lead to the identification of a distinct phenotype of unipolar patients presenting similarities with bipolar patients is still open. We are currently conducting a clinical trial that we hope will shed some light on this issue.