This cross-sectional imaging study including 463 subjects examined brain structures in a relatively young sample of patients with schizophrenia (n
= 155), their nonpsychotic siblings (n
= 186), and healthy control subjects (n
= 122, including 25 sibling pairs), using various imaging techniques. Global brain volumes of nonpsychotic siblings were not different from those of healthy control subjects, nor did siblings and healthy control subjects differ in cortical thickness or GM density measured using a VBM approach. The paucity of cortical and subcortical brain differences in the siblings of patients is consistent with the findings from another large study in nonpsychotic siblings.28–30
The siblings in our study were about 10 years younger compared with the sample of Goldman et al,28
and in these analyses, we were able to take into account relatedness as we included patient-sibling pairs, as well as healthy control sibling pairs. Our findings contrast with those of smaller imaging studies in nonpsychotic siblings (largest study; total n
Our study did find robust structural brain differences in patients with schizophrenia as compared with healthy control subjects. Indeed, we replicate the global volumetric abnormalities in patients with schizophrenia in TB, GM, WM, lateral ventricle, and third ventricle53,54
. Furthermore, the decreases in cortical thickness and GM density, particularly in the frontal and temporal cortex, as well as in the anterior cingulate cortex, are consistent with earlier studies55–57
and with those studies using a VBM approach53,54,58
. These most replicated findings in the inferior frontal, middle temporal, and the cingulate regions have been found to be associated with speech59
Thus, our findings that brain abnormalities are expressed in patients with schizophrenia but not in nonpsychotic siblings suggest that brain abnormalities in schizophrenia mainly reflect processes related to the manifestation and/or treatment of the illness.
That the illness itself causes brain changes in schizophrenia is corroborated by the findings in ultra–high risk subjects and the association between brain changes and illness-related factors in schizophrenia. Only in those subjects who later converted to psychosis cortical GM deficits were found at baseline, but deficits were not found in those subjects who did not become psychotic over time.60,61
Furthermore, a longitudinal study in adolescents at ultra-high risk for psychosis showed that the development of psychosis was associated with progressive abnormalities around time of onset (which was not attributed to antipsychotic medication).62
In addition, studies that examined symptomatology in relation to brain imaging findings reported that reduced GM volume was related to duration of untreated psychosis14
and duration of psychosis.15,16
In addition, various other studies,6,47,63
but not all,64
reported a relationship between clinical outcome and reduced GM volume. Indeed, in the present study, we found that severity of illness (total and positive symptoms) was associated with reduced GM and increased lateral ventricle volume.
There is also evidence that brain abnormalities reported in schizophrenia are related to the effects of antipsychotic treatment. While post hoc analyses failed to show an association, in cross-sectional non-randomized studies such as ours, it is not possible to rule out medication effects on brain structure completely. A study in macaque monkeys treated long term with olanzapine or haloperidol reported that cortical volume was reduced by both these agents.65
In contrast, in a prospective study of Lieberman et al,66
obtaining MRI scans at multiple intervals, brain morphology was found to be differentially affected by olanzapine and haloperidol over time. In addition, other studies in patients with schizophrenia showed that decrements in GM volume over time, particularly in prefrontal regions, were associated with the (cumulative) intake of typical but not of atypical antipsychotic medication7,17
Other nonshared environmental factors, such as obstetric complications, can result in brain abnormalities in patients with schizophrenia.9,67
Unfortunately, in our study, there was not sufficient information of obstetric complications to investigate its effects on structural brain abnormalities.
To date, the neurobiological processes that underlie the brain abnormalities in patients with schizophrenia remain unclear but may reflect anomalies of synaptic plasticity and abnormal brain maturation. Early (prenatal and perinatal) neurodevelopmental trauma may render the brain vulnerable to aberrant late neurodevelopmental processes, which may further interact with other causative factors associated with the onset of psychosis (eg, substance use, stress, and dysregulation of the hypothalamic-pituitary-adrenal axis function).68
Around transition to psychosis, these processes together may disrupt further brain development. Indeed, it has been suggested that the brain changes in the early state of schizophrenia are the result of the “toxic” effect of the psychotic state.63
Another theory was raised which guide neuroimmunology/virology studies of schizophrenia and derives from a general theoretical focus on central nervous system viral reactivation-induced immunological changes leading to psychosis.69
That the structural brain differences are under genetic control cannot be dismissed by the negative findings of our study. MRI studies in twins do report volume decreases in whole brain, GM and WM, or hippocampus in unaffected twins who are discordant for schizophrenia,1,21,70,71
but not all.72
These studies included monozygotic twins, sharing 100% of the genes with their sibling, and dizygotic twins, sharing 50% of the genes. Interestingly, brain volume differences in twins discordant for schizophrenia were more pronounced in the monozygotic than in the dizygotic twins, compared with healthy control twins.1–3,70
This suggests that the genetic contribution to brain volume reductions in schizophrenia may be subtle and is primarily detectable in subjects with high genetic loading, ie, monozygotic discordant twins and not in the healthy siblings of patients with schizophrenia.
The presence of brain volume differences in unaffected twins but not siblings could also be explained by the contribution of environmental factors that are specific for twins, such as intrauterine viral infections,73
and delivery complications.75
These are common environmental factors that patients share with their (monozygotic) co-twins, while they are not shared with a nontwin sibling.8,76
Stress may also be such a common environmental factor.9
The emotional burden of the disease can be considerable in siblings of patients with schizophrenia.77
For twins, who often have a close emotional relationship with each other, the experience of having a co-twin with a severe psychiatric disease like schizophrenia may represent a more pronounced burden.
Furthermore, the heterogeneity produced by the broadly recruited sample of unaffected siblings in our study may have undermined the apparently high statistical power. Some previous computational neuroanatomical studies assessed relatively homogenous groups of unaffected relatives of patients with schizophrenia, which were chosen deliberately to maximize power through “genetic enrichment,” including high-risk familial subject78
and relatives from multiply affected families.79–82
Some limitations need to be addressed. First, a selection bias may have affected our results. This is reflected in that we included only siblings of patients who were willing to participate. Those siblings whom we were not able to include in the study may be of particular interest as they might share more (sub)clinical features with their ill proband. However, based on the FIGS, the included siblings were not different from those who were not included. Earlier studies reported that schizotypy was found to a much higher degree in first-degree relatives compared with healthy control subjects.32,33
As suggested by Diwadkar et al,83
relatives with high levels of schizotypy may define a hypervulnerable subsample among these relatives of patients with schizophrenia. Interestingly, in the present study, siblings and healthy control subjects were similar in schizotypal scores as measured with the SIS-R, suggesting that these siblings were possibly not vulnerable to develop schizophrenia. Second, there was a preponderance of men in the sample of patients compared with siblings and healthy control subjects. The epidemiological design of this study explains these differences. To minimize the effect of gender on brain structures, we controlled for this variable in all analyses. Male gender has been shown to be associated with larger cerebral volumes84,85
that disappears with head size correction.86
Greater decline of GM volume with age in males has also been reported in some87–89
but not other90
studies. Females have also been shown to have thicker cortex across many regions of the brain.85,91
As we know that gender but also age and handedness may influence brain structures, we have included these as covariates in our analyses. Third, it may be that cross-sectional MRI measurement might not be informative enough to find structural brain abnormalities in siblings of patients with schizophrenia. Fourth, it should be noted that the significant areas found in this study are indicative of locations of effects; their spatial extent is influenced by the smoothing of the data.92
In conclusion, our study did not find structural brain abnormalities in nonpsychotic siblings of patients with schizophrenia compared with healthy control subjects, using multiple imaging methods. This suggests that the structural brain abnormalities found in patients are most likely related to the illness itself.