Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

ICOSR 2015—Can Neuroimaging Help Reduce Risk for Schizophrenia?

30 Apr 2015

As part of our ongoing coverage of the 2015 International Congress on Schizophrenia Research (ICOSR), held March 29 through April 1 in Colorado Springs, we bring you session summaries from some of the participants in the Young Investigator program. We are, as always, grateful for the gracious assistance of conference directors Carol Tamminga and Chuck Schulz, as well as meeting staff Cristan Tamminga and Dorothy Denton. For this report, we thank Mette Odegaard Nielsen of the Center for Neuropsychiatric Schizophrenia Research in Copenhagen, Denmark.

May 1, 2015. "Translating neuroimaging findings in people at high risk for psychosis into clinical practice" was the challenge posed to speakers at a morning session on Monday, March 30, at the International Congress on Schizophrenia Research in Colorado Springs.

In the first presentation, Matthijs Bossong of King's College London, UK, described a study of how hippocampal glutamate levels were related to the development of psychosis in a group of 68 ultra-high-risk (UHR) patients fulfilling the CAARMS (Comprehensive Assessment of At-Risk Mental States) criteria. Alterations in the glutamatergic system have been suggested to play a major role in the development of schizophrenia. Likewise, animal models have pointed toward an altered function of the hippocampus being important in the pathophysiology of schizophrenia.

The UHR subjects were followed for 18 months, during which nine had a transition to psychosis (T-UHR), and 59 subjects had no transition (NT-UHR). Comparing the levels of glutamate between the whole group of UHR patients and healthy controls, the researchers found no group difference. However, after subgrouping patients depending on their transition, significantly higher levels of glutamate were found in the T-UHR group compared to the NT-UHR and healthy controls. Bossong and colleagues also found elevated levels of myoinositol, a marker of glial activity in the T-UHR patients. Relating the findings to psychopathology, the investigators found a correlation between the CAARMS positive symptom score and increased myoinositol, as well as glutamate.

The finding of selectively elevated levels of glutamate and myoinositol in the UHR subjects who later had a transition to psychosis suggests that this could help predict the outcomes of UHR patients.

Dopamine plays a central role in the development of psychosis, said the next speaker, Romina Mizrahi of the Centre for Addiction and Mental Health in Toronto, Canada, and an increase of dopamine in UHR patients has been related to transition to psychosis. Stress is known to increase the level of dopamine, and as increased levels of experienced stress have been found in schizophrenia patients prior to psychotic relapse, this may also be relevant for UHR patients developing psychosis.

In a sample of UHR, schizophrenia, and healthy control subjects, dopamine activity was examined using 11C-PHNO PET before and after a stress test (MIST-task). Behavioral data showed that all participants found the test stressful, and the patients increased slightly in PANSS positive score after the stress test. Mizrahi and her colleagues found increased stress-induced dopamine levels in UHR and schizophrenia subjects compared to controls. Furthermore, this was related to an increase in psychotic symptoms, indicating that stress-induced dopamine release could play an important role in developing psychosis. The implication is that one could target stress regulation in UHR patients.

Mizrahi's group also examined whether cannabis could induce psychotic symptoms through increasing stress-induced dopamine levels. They found the same behavioral stress and a small increase in the level of positive symptoms in patients who used cannabis as well those who did not. A dopamine increase was observed in the non-cannabis-using patients, but there was a reduction in stress-induced dopamine release in the cannabis users. These results, then, do not seem to bear on the mechanism for the hypothesized cannabis-induced transition to psychosis in UHR patients, although they may indicate why patients find cannabis useful for self-medication.

Disturbances related to schizophrenia have a widespread anatomical distribution in the brain, which supports the idea that it is not a disease of separate brain regions, but rather a problem of network interaction, said the next speaker, Nicolas Crossley, who recently moved from King's College London to the Catholic University of Chile in Santiago. This has led to the hypothesis of disconnectivity, a term covering hypo- as well as hyperconnectivity.

In his talk, Crossley presented the framework of the network perspective of the brain, which describes the brain as composed of interacting elements. By determining the interactions between different nodes, researchers have discovered that the number of connections of the nodes does not follow a normal distribution, but rather a heavy tail distribution, with a few nodes having a lot of connections. The distribution of brain connections is interesting, since networks with similarly asymmetrical distribution of links are, in theory, highly resilient to damage or dysfunction in peripheral (less connected) nodes, but highly vulnerable to damage in central (highly connected) hub nodes. Crossley and colleagues hypothesized that brain disorders across the spectrum, which by definition cause an important dysfunction of the system, should tend to lesion central areas. This theory applies to many different neurological disorders, including schizophrenia and Alzheimer's disease.

To understand the disconnectivity of schizophrenia, Crossley and colleagues analyzed 314 functional and 51 structural MR studies including more than 10,000 individuals. The meta-analysis found that the more connections a region had, the higher the probability of it being abnormal/dysfunctional in schizophrenia. In a much smaller study, Crossley found the same is true in UHR subjects. Based on these observations, Crossley concluded that this issue of centrality should be incorporated in future approaches when using connectivity disturbance to predict outcome in UHR subjects, as abnormalities in highly connected regions may lead to a higher risk of transition to psychosis.

The final speaker was Philip McGuire, also of King's College London. He mentioned the well-replicated finding that approximately one-third of UHR patients will transition to psychosis, one-third will develop a nonpsychotic psychiatric disorder, and one-third will develop no psychiatric diagnosis. The current clinical problem is that the outcome is unpredictable based on the clinical profiles.

Until now, several group-level predictors of transition have been found using neuroimaging, such as structural changes in temporal regions, increased dopamine levels in striatum, and elevated hippocampal glutamate levels. The next step will be to translate these group-level findings to individual predictors. One promising method for doing this is machine learning. However, in building a clinical tool, practical feasibility must be taken into account: for example, availability, costs, and acceptability. This underlines the importance of finding proxy markers: for instance, PET scanning for striatal dopamine levels is not likely to be universally available and would be expensive in a clinical setting.

The aim, therefore, will be to develop a clinical tool based on proxy markers or low-cost, readily available brain biomarkers combined with other information such as family history. The future challenge, McGuire said, is not to find more markers, but to translate the ones we know into a useful tool.—Mette Odegaard Nielsen.