Dr. Heim and Colleagues Reply

To the Editor: One of the major strengths of our recent study on pituitary-adrenal responses to standard pharmacological challenge tests in adult women with a history of childhood abuse was a balanced study design regarding the four groups of abused or nonabused women with or without major depression. Indeed, we demonstrated that ACTH responses to ovine CRF differed markedly between abused women with concurrent major depression and abused women without concurrent major depression. These findings demonstrated that it is imperative to control for concurrent symptoms of psychopathology when studying neuroendocrine function in victims of childhood abuse. According to our reading of the study by Dr. De Bellis et al. (1994), structured interviews for the diagnosis of psychiatric disorders in sexually abused girls had been performed at some point during a longitudinal study. However, it was not clear that current psychiatric symptoms (except for symptoms of depression) were systematically evaluated at the time of the neuroendocrine investigation. Accordingly, the authors subdivided abused girls on the basis of a history of dysthymia and not on the basis of concurrent symptoms or diagnoses. We showed that concurrent, but not past, symptoms or diagnoses of depression affected pituitary reactivity to CRF administration. In addition, Dr. De Bellis et al. did not report findings on the prevalence of PTSD and the effects of PTSD on neuroendocrine responses throughout their article, although 20%–70% of sexually abused children reportedly suffer from PTSD (1).

With respect to the effects of race on our findings, we point out that the analyses of variance (ANOVAs) of the effects of history of childhood abuse and current major depression on hormone profiles throughout the challenges were corrected for the effects of race by introduction of this variable as a covariate. Dunn’s multiple comparison procedures were also adjusted by using pooled mean square errors derived from analyses of covariance. In accordance with the literature (2), we found that race affected ACTH concentrations in the CRF stimulation test. When controlling for the influence of race, we found significant three-way interaction effects of childhood abuse, major depression, and time in all analyses. It is noteworthy that in contrast to our study, previous studies assessing the effects of race on pituitary-adrenal reactivity to challenges did not control for histories of early life stress (2, 3), which may have affected these findings. We considered introducing race as an independent factor in the ANOVAs, resulting in means for different races within the study groups; however, the small number of nonwhite subjects in three of the four groups did not allow for meaningful results. When limiting the group comparisons to white subjects, we were able to support our previous findings (data not shown). We acknowledge that race markedly affects pituitary reactivity and extend Dr. Merskey’s comment by suggesting that race may represent a constitutional factor that interacts with early environmental influences in shaping a phenotype with vulnerability to stress and disease. The role of race in the neurobiology of early-life stress deserves further attention in future studies.