Schizophrenia Research Forum - A Catalyst for Creative Thinking

Neuroscience 2009: New Looks at Schizophrenia and Bipolar Disease

4 December 2009. There are many ways to look at a disease as complex as schizophrenia, and a symposium held on 19 October 2009 at the Society for Neuroscience annual meeting covered the bases. Titled “Schizophrenia: Neuropathology and Structural Neuroimaging Studies,” the session offered something for everyone. Most of the talks centered on magnetic resonance imaging, though microscopy and gene expression measures were represented, too. Several of the speakers focused on the overlap between schizophrenia and bipolar disorder, comparing the two diseases to see whether the increasingly apparent genetic links between these diseases (see SRF related news story) can be translated into common changes in brain structure or function. The end result—whether we will have a better understanding of disease pathology and drug action, the ability to predict progression, or the emergence of candidate biomarkers or endophenotypes—remains to be seen, but the studies offer some glimpses of progress.

Automated image analysis shapes up
The anterior cingulate gyrus (ACG) is involved in emotional processing and executive function, and has been a region of interest for schizophrenia studies. Of 10 MRI studies published since 2001 looking at ACG size in people with schizophrenia versus healthy controls, seven reported a reduction, according to James Levitt, VA Boston Healthcare System and Harvard Medical School. These studies all used manual tracing of images to measure size, a tedious and time-consuming procedure. The development of automated image analysis programs promises to make the job easier, and allow more comprehensive and detailed analyses. Levitt described the use of mostly automated techniques to analyze gray matter volume, surface area, and cortical thickness in subregions of the ACG (subgenual, cognitive, and affective) in 28 males with schizophrenia versus 28 healthy men. The results showed a trend toward reduction in volume and surface area of ACG on the right side, driven by right affective and right cognitive regions. A higher number of subjects with schizophrenia had a prominent (deeper) right side paracingulate sulcus, which correlated with a smaller ACG. The exaggerated sulcus and smaller ACG correlated with better performance on cognitive measures, including the Trails B test. The new method led to significant, interesting results with greatly reduced ROI tracing time, Levitt concluded, but he left open the possibility that manual editing may reveal even finer differences.

Daniel Mamah of Washington University in St. Louis, Missouri, deployed his own version of automated image analysis to assess the shape of different brain structures, and to compare shapes between schizophrenia and bipolar disorder. In schizophrenia, some parts of the brain tend to be smaller, with the left hippocampus among the most consistently affected, Mamah said. In bipolar disorder, in contrast, no consistent volume changes have been measured, with the possible exception of the amygdala, which has been reported to either increase or decrease. By looking at shape, Mamah reasoned that he might identify endophenotypes that were not apparent by volumetric measures alone. He and his colleagues had previously reported shape changes in several brain structures in people with schizophrenia and their siblings compared to healthy controls (Wang et al., 2008; Mamah et al., 2008), and the new study extended this analysis by looking at the amygdala, basal ganglia, hippocampus, and thalamus from MRI scans of 12 adults with bipolar disorder, 11 with schizophrenia, and 12 healthy controls. While the scans revealed loss of volume and changes in shape of the structures in people with schizophrenia compared to controls, no such changes were observed in bipolar disorder. Mamah concluded there are no common structural abnormalities in schizophrenia and bipolar disorder. There may be some overlap between schizophrenia and psychotic bipolar disorder, however, as they found that patients with psychotic bipolar disorder showed smaller volumes in the thalamus and caudate compared to non-psychotic bipolar patients.

Taking the measure of bipolar
Another study comparing bipolar and schizophrenia also found divergence when it came to a different morphometric, cortical thickness. People with schizophrenia show a widespread cortical thinning that is present at the start of disease, but determining whether the same occurs in bipolar disorder has been complicated by the fact that the most common treatment, lithium, itself increases cortical thickness, said Lara Foland-Ross of the University of California at Los Angeles. In addition, cortical thickness can also be affected by mood state. Foland-Ross described a study that controlled for these variables by analyzing 35 bipolar patients who were euthymic (of neutral mood) and free of lithium treatment. Cortical thickness was determined on MRI images using a sensitive analysis that controls for sulcal pattern differences. In the bipolar patients, Foland-Ross reported many areas of thinning, and no regions of thickening. Thickness of the Brodmann areas (BA) 8/10/11 in the frontal lobe correlated negatively with the number of prior depressions, and Foland-Ross speculated that could be due to depression-related elevations in cortisol, which can cause cortical thinning. That thinning may explain why prior depression raises risk for recurrence, or alternatively, it may be a marker for more severe illness, Foland-Ross said. In the BA 24/32 region, which includes the dorsolateral prefrontal cortex and left anterior cingulate, cortical thickness negatively correlated with duration of untreated disease, suggesting that loss of cortical tissue is slowed or halted by pharmacological treatment. The study, the first to be done in lithium-free patients, suggests that thinning correlates with course of disease. Because lithium-treated patients in previous studies do not show thinning, the results raise the possibility that lithium treatment may actually prevent thinning.

White matter changes have also been reported in both schizophrenia and bipolar disorder, but in both cases, it is not clear if the cause is the disease or its treatment. Some studies report a smaller corpus callosum, or lower fractional anisotropy (FA, a measure of diffusion of water that indicates structure in white matter tracts) in anterior and posterior brain regions (reviewed in White et al., 2008), but these studies were done mostly in chronic patients. Two studies done in first-episode patients do not show lower FA (Friedman et al., 2008; White et al., 2009), suggesting that the FA abnormality emerges as the disease progresses. To address this question, Lisa Lu of Roosevelt University and colleagues at the University of Illinois, both in Chicago, studied 21 adults with schizophrenia, and 13 with psychotic bipolar disorder at their first episode of psychosis compared to 20 healthy controls. Their results clearly differentiated schizophrenia and bipolar cases: By FA measures, the schizophrenia subjects looked like controls, but bipolar subjects showed lower FA in many brain areas compared to either schizophrenia or control. The finding of normal white matter structure in first-episode schizophrenia makes the third time this has been observed, and suggests that white matter abnormalities in chronic disease are a result of disease progression or medication. The finding of lower FA in bipolar first-episode subjects suggests the changes are specific to early bipolar disease and not to psychosis, consistent with previous findings (Kafantaris et al., 2009). Lu noted that comorbidities reduced the ability to see differences between groups. This makes the results less generalizable, because comorbidities are more common than not.

In schizophrenia, it is not clear why the cortex grows thinner—is the tissue losing neurons, or neuropil? Most evidence to date has suggested no loss of neurons. To examine the connection between reduced width and cellular changes, John Smiley of the Nathan Kline Institute, Orangeburg, New York, and colleagues examined cortex width and volume on high-resolution micrographs of postmortem samples. The technique allows them to look at individual cortical layers (Smiley et al., 2009), which they measured in samples of auditory cortex from 11 males with schizophrenia and 10 age-matched controls. Smiley reported that the upper layer accounted for 61 percent of the cortical thickness in controls and 59 percent in subjects with schizophrenia. That does not sound like much, he said, but it translates into an 8 percent reduction in the thickness of the upper layer, with the lower layer largely unchanged. Counting neurons in layers II, IIIc, and IV of the upper cortex revealed no change in neuron density, consistent with a modest loss of about 7 percent of the neurons in the upper layer. Smiley says they are now trying to identify the missing neurons.

From form to function
Scott Schobel from Columbia University, New York, spoke about his recent work with Scott Small, showing that hippocampal hyperactivity, as detected by elevated baseline blood flow, predicts the transition to psychosis in high-risk patients (see SRF related news story, as well as Hippocampus in Schizophrenia Roundtable). Schobel presented new data on hippocampal size and shape in continuing longitudinal follow-up of those same subjects. Hippocampal size did not predict who would progress to disease, but there was a volume decrease at onset, particularly in the anterior parts of the hippocampus (CA1 and subiculum). The researchers looked at baseline CA1 blood flow, and found that elevation early on was associated with a greater loss of volume later in progressors, where there was an associated emergence of hyperactivity (high cerebral blood flow) in the subiculum and orbitofrontal cortex. Schobel proposed a model where CA1 is differentially targeted in the prodromal state and where, during progression to full-blown disease, structure changes occur that result in loss of hippocampal volume, and later activation of subiculum and orbitofrontal cortex. Changes in the hippocampus did not correlate with changes in dorsomedial prefrontal cortex. Results of experiments in rodents, also recently published by the Small group, suggest that the hippocampal problems may be due to increased glutamate (Gaisler-Salomon et al., 2009). This raises the possibility that decreasing glutamate function in the prodromal stage could ameliorate later effects on the CA1 subregion.

Michael Harms of Washington University in St. Louis, Missouri, combined structural and functional imaging to see if he could draw correlations between anatomical variations and brain function as measured by fMRI. He compared structural MRI with task-related activation on fMRI during the 2-Back test of working memory in 154 subjects, comprising 32 with schizophrenia and 29 unaffected siblings, 46 controls and 47 control siblings. Since this type of analysis has not been done in either normal or disease groups, Harms first looked at the whole sample, and identified regions where working memory task activation correlated with the volume of different brain structures. Interestingly, the structurally correlated regions did not generally coincide with the activated areas, with the exception of the medial frontal gyrus. The researchers also looked for an overlap between the correlated regions and known brain networks. About one-fifth of the voxels that correlated with task activity in the medial frontal gyrus were in the working memory network, but very little correlation was seen in any part of the default mode network. In general, decreased volume correlated with increased activity during the 2-Back test, reminiscent of Schobel and Small’s observation in the hippocampus. The investigators did not see evidence for differential correlation between structure and function in schizophrenia versus control groups, but their results do suggest that structural and functional variability may be related in some regions. The method presents an alternative cross-modal technique to investigate interconnections in the human brain, Harms said.

Alternative views
The end of the session featured three talks on other ways to look at neuropathology. Elizabeth Thomas of the Scripps Research Institute in La Jolla, California, looked at genomewide expression profiles from postmortem frontal cortex during aging in normal and schizophrenic individuals. The study, which was published earlier this year (Tang et al., 2009), revealed that in terms of changing gene expression, early schizophrenia resembles aging, but later in the disease, gene expression diverges widely from that seen in healthy brain. For example, normal aging or early schizophrenia were significantly linked to changes in pathways related to synaptic function, cell cycle/DNA damage, and apoptosis, consistent with previous microarray studies. Aging in schizophrenia was significantly associated with fatty acid and steroid metabolism, but not with those functions associated with normal aging.

Does the divergence of gene expression over time between healthy and affected groups reflect the use of antipsychotic medication? Thomas showed that changes in six genes of the 368 associated with schizophrenia correlated with medication dose, but in mice the same genes were not altered by administration of haloperidol or fluphenazine, the drugs most commonly used by the patients in their sample. The results suggest that schizophrenia onset anticipates the normal aging process, and further, that some symptoms of aging, such as dementia and psychosis, might be explained by these common molecular profiles.

In an additional analysis, Thomas added 45 more subjects from published data on a similar platform and performed cluster analysis. She found two gene modules affected by aging but not schizophrenia. Both are associated with CNS development, neurite outgrowth, differentiation, and neurotransmission and dopamine neurotransmission, and go down with age in normal brain, but stay the same or go up in schizophrenia. The results are consistent with previous work (Harris et al., 2009) showing a downregulation of developmental genes across early life into adulthood. Thomas thinks they are seeing a continuation of that process in normal brains but not in schizophrenia, which raises the possibility that changes in gene expression might occur early in life in people who are destined to develop schizophrenia.

Researchers would like to study gene expression in living cells, but the inaccessibility of brain tissue from living patients hampers their efforts. Shin-ichi Kano from the Johns Hopkins University School of Medicine in Baltimore, Maryland, talked about two possible solutions: olfactory neurons and induced pluripotent stem (iPS) cells. Olfactory neurons are accessible via nasal biopsy, making them a relatively easily obtained source of neurons that resemble CNS neurons. People with schizophrenia have altered odor detection, and a recent report suggests that these neurons, which are continuously regenerated, may themselves be altered in the disease (Turetsky et al., 2008). Kano established olfactory neuronal cells from dissociated olfactory epithelium and found that the gene expression profiles of neurons from people with schizophrenia were significantly different from those of normal controls, and included changes in genes involved in inflammatory stress and immune-related processes. The investigators also generated iPS cells from schizophrenia patients by introducing four defined transcription factors into skin fibroblasts. Comparison of molecular profiles among olfactory neurons, iPS cell-derived neurons, and other patient samples is in progress.

In the final talk, Sanghyeon Kim of the Stanley Medical Research Institute in Rockville, Maryland, described the Stanley Neuropathology Consortium Integrative Database (Kim and Webster, 2009), a publicly accessible Web-based tool that includes both neuropathological traits and microarray RNA expression data for the Stanley Brain Bank samples (15 each of schizophrenia, bipolar disorder, depression, and unaffected controls). Working with the brain bank samples, Kim was able to correlate the decrease in perineuronal oligodendroctyes in schizophrenia, bipolar disease, and depression with changes in RNA expression, including the message for ErbB2 (Kim and Webster, 2008). Kim described other experiments, including correlations of dopamine levels in the prefrontal cortex (elevated in schizophrenia) with expression of genes in pathways involved in stress, cellular metabolism, and organ development. Similarly, they looked at glutamate levels, which were increased in bipolar disorder and depression, and found a correlation with stress response and cell death pathways. The database can steer researchers to the connections between pathological markers and changes in gene expression in these diseases, Kim said. The group is now working to add SNP data and the capability to do association analysis.—Pat McCaffrey.

Comments on Related News

Related News: Large Family Study Links Genetics of Schizophrenia, Bipolar Disorder

Comment by:  Alastair Cardno
Submitted 7 April 2009
Posted 7 April 2009
  I recommend the Primary Papers

The results of the family/adoption study by Lichtenstein et al. (2009) and our twin study (Cardno et al., 2002) are remarkably similar. Using a non-hierarchical diagnostic approach, the genetic correlation between schizophrenia and bipolar/mania was 0.60 in the family/twin study and 0.68 in the twin study. The heritability estimates were somewhat lower in the family/adoption (~60 percent) than twin study (~80 percent), but can still be said to be substantial and similar for both disorders.

When we adopted a hierarchical approach, with schizophrenia above mania, we found no monozygotic twin pairs where one twin had schizophrenia and the other had bipolar/mania, but with their considerably larger sample, Lichtenstein et al. (2009) were able to confirm a significantly elevated risk for bipolar disorder in siblings of probands with schizophrenia (RR = 2.7), even when individuals with co-occurrence of both disorders were excluded.

I think there is a potentially interesting link between the family/adoption and twin studies focusing mainly on non-hierarchical diagnoses: Owen and Craddock’s (2009) commentary on the family/adoption study, where they advocate a dimensional approach, and Will Carpenter’s SRF comment regarding the value of domains of psychopathology. The non-hierarchical approach (where individuals can have a diagnosis of both schizophrenia and bipolar disorder during their lifetime) could be viewed as a form of dimensional/domains of psychopathology approach, with schizophrenia and bipolar disorder each having a dimension of liability, and there is now evidence from family, twin, and adoption analyses that these dimensions are correlated, i.e., that there is some overlap in etiological influences.

If schizophrenia and bipolar disorder share some causal factors in common, what might be the implications for the unresolved status of schizoaffective disorder? Our twin study suggested that the genetic (but not environmental) liability to schizoaffective disorder is entirely shared with schizophrenia and mania, defined non-hierarchically (Cardno et al., 2002). If so, and if schizophrenia and bipolar disorder share some genetic susceptibility loci in common, while other loci are not shared, then risk of schizoaffective disorder (or perhaps the bipolar subtype) could be elevated either by the coincidental co-occurrence of non-shared susceptibility loci, or by the occurrence of loci that are common to both disorders.

In this case, any loci that influence risk of schizoaffective disorder (bipolar subtype?) should also increase risk of schizophrenia and/or bipolar disorder, and this model would be refuted if any relatively specific susceptibility loci for schizoaffective disorder were confidently identified.

Some further outstanding issues:


Cardno AG, Rijsdijk FV, Sham PC, Murray RM, McGuffin P. A twin study of genetic relationships between psychotic symptoms. American Journal of Psychiatry 2002;159:539-545. Abstract

Lichtenstein P, Yip BH, Björk C, Pawitan Y, Cannon TD, Sullivan PF, Hultman CM. Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a population-based study. Lancet 2009;373:234-9. Abstract

Owen MJ, Craddock N. Diagnosis of functional psychoses: time to face the future. Lancet 2009;373:190-191. Abstract

View all comments by Alastair Cardno