Abstract
Studies utilising structural neuroimaging methodologies have been used extensively over the past 25 years to investigate the underlying neurobiology of schizophrenia. From the initial reports of ventricular enlargement using computed tomography (CT), to the more recent high-resolution magnetic resonance imaging (MRI) examinations of smaller brain regions, a consensus has emerged of consistent, albeit subtle, changes in a variety of brain regions [1],[2]. These include reduced medial temporal lobe volumes, including the hippocampus and amygdala.
While these studies have been important in providing conclusive evidence for neuroanatomical differences in the brains of people meeting the diagnostic criteria for schizophrenia, the implications of these findings for understanding the development of psychosis in young people have been less clear.
There is a general consensus that many variables derived from structural brain imaging differ significantly in patients who have recently been diagnosed with DSM-IV schizophrenia and healthy volunteers [3],[4], indicating that neuroanatomical abnormalities are present from the onset of the disorder. Some studies have also suggested that there may be progressive changes that occur over time, although to date there have been few studies conducted over a time interval of longer than 5 years where state-of-the-art MRI techniques have been used [5],[6]. Evidence is growing that some, although not all, of these variables may also differ in individuals diagnosed with disorders that are genetically linked with schizophrenia (e.g. schizotypal personality disorder) [7],[8] or those who are at increased risk of the development of schizophrenia and related psychoses (e.g. first degree relatives) [09–11]. Such findings would not be predicted if the structural brain differences reported in patients with schizophrenia were sequelae of long-term psychotic illness, or the physical therapies used to treat it. Instead, such findings have been interpreted as evidence that structural brain changes result from neurodevelopmental insults that have been hypothesised as aetiological contributors to the development of schizophrenia [12–14].
A number of problems must be considered in evaluating the utility of structural brain imaging findings in relation to predicting the development of schizophrenia. First, there are few data concerning the normal range of structural MRI-derived volume measures, in particular the development of the normal human brain in adolescence and young adulthood. Several recent studies have begun to address these issues [15–17] but definitive answers await systematic age-stratified studies of large samples and crosslaboratory replication. Second, there is a paucity of evidence concerning the test–retest reliability of volume estimates over time and those changes that might occur as a result of individual differences in physiological parameters, for example, variation in fluid balance that is associated with excessive drinking. Third, in many studies, those patients examined represent samples of convenience. Given evidence that significant gender differences exist in various brain structural measures, males and females should be equally represented in patient groups [18–21]. Cerebral laterality may also be related to functional assessments such as handedness, which has not always been considered when lateralised differences in structural brain measures are examined [22],[23].
Patients with schizophrenia and young people with prodromal symptoms may also meet DSM-IV criteria for other disorders, such as substance use. Although one recent study reported no significant differences in the regional brain volumes of current frequent marijuana users compared with non-using controls [24], our group recently found smaller cerebellar volumes in a group of current marijuana users who volunteered to participate in a treatment intervention study [Ward PB, Solowij N, Grenyer B: unpublished data]. Given the widespread use of cannabis in young people with first episode psychosis, further investigation is required into the possible confounding effects of such use on MRI-derived neuroanatomical measures in these populations, in addition to the potential effects of regular use of other illicit and licit drugs (e.g. alcohol) on brain structures [25],[26].
Advances in the techniques of MRI acquisition and measurement have been substantial in recent years [27–29], resulting in greater replicability of significant findings across laboratories [30]. Despite this, the functional significance of the relatively small volumetric changes observed among patients with schizophrenia and first episode psychosis remain elusive [31–34]. Given that neuropsychological tests were originally developed for use with patients with gross brain lesions and substantial functional impairments, it is not surprising that the more subtle brain abnormalities among people with schizophrenia or first-episode psychosis have not been reflected in impaired performance in such tests. Similarly, clinical rating scales that were originally developed to assess treatment response in clinical trials may not be sufficiently sensitive to the behavioural consequences of subtle neuroanatomical variation. The use of tasks designed to tap specific cognitive impairments and ratings reflecting the multidimensional nature of schizophrenic symptoms will enhance understanding of the significance of the fine-grained neuroanatomical abnormalities.
Recent studies examining first-degree relatives have overcome some of the difficulties associated with generalised cognitive deficits among patients with schizophrenia [35]. However, without longitudinal follow up, it remains difficult to demonstrate a causal association between changes in regional volumes and specific cognitive domains. An alternative strategy is to examine neurologically normal populations in which subtle structural and functional changes may be posited. In adolescents, those who were born prematurely (less than or equal to 30 weeks of gestation) have recently been shown to have smaller hippocampal volumes bilaterally and impairments in memory and numeracy in comparison to those who were born at full term [36].
Structural neuroimaging has been important in establishing reliable differences between patients with schizophrenia and their first-degree relatives. As such, it has provided strong evidence for neuroanatomical concomitants of the genetically determined diathesis for schizophrenia. The challenge for future research is to develop a better understanding of the functional significance of structural brain abnormalities in young people at risk of developing schizophrenia. To do so will require large-scale studies of healthy volunteers as much as fine-grained analysis of clinical or subclinical samples. With increasing standardisation of MRI image acquisition and analysis, the development of multisite collaborative studies becomes a realistic goal. These studies should have sufficient statistical power to take into account many of the variables discussed here and shed further light on the potential utility of neuroanatomical vulnerability makers in future generations.