To the editor:

I read the news feature “Mood Swings” (Nat. Med. 10, 1010–1012; 2004) with some disappointment. There are some points in the article that I would like to clarify for the benefit of your readers.

The author rightly notes that the research community does not currently understand the pathology of depression to a degree that readily allows us to identify specific molecular targets not linked to symptoms. But the story does not mention the research being performed to help us better define the disease. For example, in recent years, some exciting work is beginning to suggest that degeneration of certain populations of neurons in the hippocampus may be linked to depression, and this may introduce new opportunities for target identification (Mol. Psychiatry 5, 262–269; 2000; Science 8, 805–809; 2003).

Even with new targets, however, the greatest challenge in depression remains identifying mechanisms and compounds that will provide a meaningful therapeutic benefit to patients. Efforts to develop substance P antagonists that block NK-1 receptors are an example of this challenge. As with many depression trials, high placebo effects often confound our ability to determine whether treatments are effective. Even in studies that show efficacy, we must be convinced that the therapeutic profile will be sufficiently different from available agents, as we are not interested in introducing new therapies that simply recapitulate the efficacy of selective serotonin reuptake inhibitors or dual reuptake inhibitors.

As a result of these clinical hurdles, unless there is real promise for yielding a therapy that will provide clinically meaningful benefit, we are very selective about promoting a mechanistically novel compound into advanced development. This was the point of my comment that we no longer simply invest in expensive trials and “hope for the best.” The comment does not relate to the issues of safety and disclosure of data from clinical trials.

Finally, I do not want your readers to be misled by the recommendation to “go ahead anyway” with a development compound in the absence of activity in animal models. The predictive validity of available models may only hold for mechanisms associated with available agents (for example, inhibition of serotonin and/or norepinephrine reuptake). Therefore, if we believe strongly in the scientific rationale, we would not necessarily abandon a mechanistically novel compound simply because it was not active in one of the existing models. It must also be noted that scientists in both industry and academia are diligently working to develop new models that could offer more robust predictive validity, particularly with regard to the issue of differentiation mentioned above.