This study investigated the volumetric differences in subcortical gray matter volumes, particularly the amygdala and the hippocampus, in early-onset BPD without psychosis, BPD with psychosis, and SZ in order to elucidate the diagnostic specificity of patterns of subcortical neural abnormality. In addition, this study evaluated diagnosis-by-sex interactions to understand the influence of sexual dimorphism on the limbic structures. We found that there were no volumetric differences between SZ and BPD in the amygdala and the hippocampus, which was not consistent with our a priori hypothesis. However, we found that youths with SZ had the smallest thalamic volumes bilaterally, with at least a moderate effect size when compared with the other 3 groups. In contrast, youths with BPD with and without psychosis had increased right and left NA volumes, with a moderate effect size compared with the other groups. Both the thalamic finding and the right NA finding were independent of sex effects. We also found that the youths with BPD with psychosis only shared findings with the BPD without psychosis and not with the primary psychotic disorder group. Taken collectively, these data suggest that specific subcortical brain regions are differentially affected in early-onset BPD and SZ, perhaps suggesting that these are disease-specific abnormalities. When diagnosis-by-sex interactions were evaluated, males with SZ had the smallest left amygdala and females with BPD had the smallest left hippocampus.
During normal brain development, there are clear patterns of growth that take place in limbic as well as other subcortical structures. For example, Giedd et al30
have reported that in healthy children, the left amygdala increases with age in males while the right hippocampus increases with age in females. In another study, school-age females were found to have larger hippocampal and smaller amygdala volumes, as a proportion of total cerebrum, than males.46
In our sample, we found that our HC girls had larger hippocampal volumes, but we did not find differences in the amygdala. The lack of an amygdala finding in our sample may in part be due to the fact that our sample included both children and adolescents. Finally, De Bellis et al47
noted significant sex-by-age interactions for cerebral gray matter volumes (including both cortical and subcortical structures) in a sample of children and adolescents and that males had more prominent age-related gray matter decreases than females. In addition, these investigators found significant sex-by-Tanner stage (a measure of pubertal maturation) interactions in gray matter which supports the idea that hormones influence brain maturation and that they may play a role in abnormal brain structure and function in psychiatric illness.35,36,47
Such normative influences help lay the foundation for a neurodevelopmental model that associates puberty with onset of illness in BPD and SZ.
Anatomic neuroimaging studies of adults and children with both BPD and SZ generally have implicated frontal-limbic network abnormalities in both disorders.12,14,15,18
Although the limbic findings across studies are mixed, the most consistently reported finding in early-onset BPD is a reduced amygdala volume16
and in early-onset SZ, a decreased hippocampal volume.20,48
However, at least one study in early-onset SZ did not find a reduction in the hippocampal volumes but found that duration of illness negatively correlated with this structure, suggesting that a volumetric difference might be revealed with disease progression.49
In addition, there have been studies that have found a reduction in the hippocampus in BPD12,21,50
and a smaller amygdala in SZ.15,18
We did not find hippocampal or amygdalar volumetric differences in our study in either the SZ or the BPD groups. However, we did find that amygdala volumes in the BPD without psychosis group correlated inversely with the MRS, which suggests the involvement of this structure in the clinical phenotype. For example, higher MRS scores were associated with decreased right and left amygdala volumes in the BPD without psychosis group. In addition, we found that the right amygdala volume correlated with GAF scores in the SZ group. The mixed findings in these reports in the amygdala and hippocampus may be due to a number of factors, but given the findings in this study, age range of the sample, age of onset, and sample size need to be considered.
In our exploratory analyses of the other subcortical structure volumes for differences by diagnosis, we found decreased thalamic volumes bilaterally in SZ and not in BPD and an increased NA in BPD and not in SZ. These findings are consistent with prior reports. For example, volumetric studies of early-onset BPD have not found differences in the thalamus,16,51
whereas studies of SZ have reported a reduced size of this structure.24,33
The NA volume has been reported to be abnormal in early-onset BPD.22,23
To our knowledge, these are the only 2 anatomic neuroimaging studies in youths with BPD that have reported on this structure, which is of interest due to its role in the reward and motivation systems. Additionally, in one functional magnetic resonance imaging study, researchers found abnormal activity in a number of brain regions including the NA.52,53
Of note, we did find a correlation between MRS scores and the right NA in the SZ-spectrum disordered youth, which may provide further support for the role of the NA in manic symptoms regardless of diagnosis. In sharp contrast, studies in early-onset SZ have not found any difference in the NA.54
Our finding of a reduced thalamic volume in SZ and not BPD suggests that the thalamus may be more intimately involved in the SZ phenotype. This is of significant interest when exploring the neural circuits of SZ given that the thalamus is a central filter in the brain that helps individuals to appropriately process sensory information and integrate activity among forebrain regions. Finally, the increased NA in both the BPD without psychosis and the BPD with psychosis and not in the group with SZ suggests that this finding may be relatively specific to bipolar illness processes.
We assessed diagnosis-by-sex interactions in the limbic structures due to their involvement in affect and cognition and due to the fact that they are structures with a high concentration of sex hormone receptors. We found that males with SZ had smaller left amygdala volumes relative to other groups and that females with BPD had the smallest left hippocampus. These findings suggest that the proportion of males to females in anatomic neuroimaging studies is of critical importance and highlights the fact that the mixed findings in the limbic structures in prior reports in both SZ and BPD may be in part due to the sex ratio of the subjects included. These diagnosis-by-sex interactions in the amygdala and hippocampus may be secondary to abnormal hormone levels in males with SZ and females with BPD at critical developmental junctures or could be due to abnormal densities or function of the androgen receptors in the amygdala or estrogen receptors in the hippocampus and/or their interaction with stress, given the density of glucocorticoid receptors in both these structures.29
These structural abnormalities may account for some of the clinical differences seen in males and females affected by BPD or SZ.55–57
To our knowledge this is the first study to compare youths who suffer from BPD without psychosis, BPD with psychosis, and those with SZ. Strengths of the study include the well-characterized sample, the state of the art morphometric analysis, and the sufficient numbers of subjects to analyze limbic volume differences between groups and by sex. The limitations of this study include the fact that the NA, a structure with significant differences between psychiatric groups, is a structure that is small and difficult to measure. Our interrater reliabilities for this structure were not as strong as those of the other subcortical structures and thus may have attenuated power to detect reliable group differences. However, it is unlikely that the weaker reliability for this structure is responsible for the observed group differences because the image analysts were blind to diagnosis and any inconsistencies in measuring the NA would be expected to occur in all groups with equal probability. Another limitation of these data is that they are cross-sectional in nature making them vulnerable to cohort effects. We have learned from prior work that the longitudinal design is ultimately the most informative about the evolution of brain changes in healthy children and in early-onset psychiatric illness.29,58–63
However, these data do lay the foundation for future assessments of the trajectory of these brain changes in a longitudinal design. Such assessments are critical to tease out the complex interaction between normal developmental processes and the expression of disease, particularly in networks known to be involved in affect regulation, thought, and behavior.
Just as importantly, future studies should assess for sexually dimorphic abnormalities and their independent and/or interactive trajectories.63
De Bellis et al47
found sex-by-Tanner stage interactions in the brains of healthy children, and our data highlight that boys and girls with BPD and SZ may move to the expression of disease via divergent paths. Although the evaluation of the impact of pubertal status would yield even more information regarding the hormonal influences on brain anatomy peripubertally by sex, we could not pursue this line of inquiry here due to the relatively small numbers in our BPD with psychosis and SZ groups. Future investigations should include assessments of the association of pubertal status (and perhaps sex hormone levels), sex, and diagnosis to elucidate the various brain changes that occur with maturation and hormonal change. Ultimately, these methodologic approaches would highlight the important developmental periods during which brain changes emerge in these disorders and will contribute to determining the degree of continuity and etiological comparability between sex-specific adolescent and adult forms of disorder. Finally, such information may enhance our knowledge about critical periods during which interventions might be most helpful to prevent illness onset and progression for these diagnoses, perhaps in a sex-specific way.