What Turns Schizophrenia into Psychosis? A Clue Hidden in the Brain’s Circuitry

James Han | james.han@yale.edu March 24, 2019

What Turns Schizophrenia into Psychosis? A Clue Hidden in the Brain’s Circuitry

Seeing, hearing, or feeling things that aren’t there; mistrusting others for no reason; or thinking that you’re someone other than yourself: these can be some of the debilitating symptoms of psychosis. In certain cases, schizophrenic patients can become psychotic, but scientists are still unsure why some convert to psychosis while others do not. Using brain imaging and statistics, Tyrone Cannon, the Clark L. Hull Professor of Psychology and Psychiatry at Yale, and his team have found a clue in the circuitry of the brain which predicts whether patients will convert to psychosis.


What is psychosis?


Psychosis is a mental condition in which a patient has lost touch with reality. It is estimated that around three percent of people will experience psychosis in their lives. Usually, psychosis is accompanied by hallucinations, paranoia, and delusions. It can be caused by a variety of factors, such as medications, alcohol or drug abuse, or mental illnesses, such as schizophrenia. During a period called the prodromal period, a schizophrenic patient who will become psychotic will first experience a period of decreased mental function, such as memory problems. In an effort to treat psychosis before it sets in in patients with schizophrenia, doctors and scientists have adopted a system of classifying patients as “high-risk.” These classifications rely on a variety of factors, such as genetic risks of schizophrenia or a history of psychotic symptoms. However, only a small percentage of clinical high-risk patients ever develop psychosis. This begs the question of why some high-risk patients develop psychosis while others do not.


In a paper published in Nature Communications, Cannon and Hengyi Cao, a postdoctoral fellow, describe differences in the circuitry of the brains of patients who convert to psychosis compared to patients who do not. Cao notes that the paper is a continuation of a long-term project, called the North American Prodromal Longitudinal Study (NAPLS-2), whose goal is to search for “biomarkers”—characteristics specific to converters that would allow researchers to “mark” them—to predict the onset of psychosis. The Nature Communications paper is a new statistical perspective at the data that has been collected for NAPLS-2.


The data from NAPLS-2 are Blood Oxygen-Level Dependent (BOLD) signals from functional Magnetic Resonance Imaging (fMRI). When neurons in the brain perform an operation, they use up oxygen, which is replenished from blood. Using magnets, scientists can see which areas of the brain are receiving more blood and therefore are active for certain tasks. Cao notes two key differences in this study compared to previous studies: the use of multiple “paradigms” and a focus on the entire brain. In fMRI studies, a “paradigm” indicates a certain task a patient performs while their brain is recorded. While looking at brain activity for a single paradigm, or task, is useful for some purposes, such as looking at the brain region involved with moving your toe, combining multiple paradigms allowed the researchers to look for universal characteristics that defined conversion to psychosis. Further, Cao added that while previous studies focused on certain regions of the brain, this study looked at the brain as a whole, which provided greater insight into identifying possible risk factors for psychosis.


How did they do it?


The first step of the study was a statistical technique called principal component analysis, which Cannon described as a way to figure out which regions in the brain are connected with each other. “Just think of cause and effect relationships in the environment,” said Cannon, “when you have things that correlate with each other, like the sun coming up, the light source leads to downstream effects on, for example, plant growth”. Once the network was identified through this method, the second step was network-based statistics, which Cannon explained was a way to check if the identified network varied between those who developed psychosis and those who did not. The results from these methods suggested that the network that the team had identified was hyperactive, or more active, in those who developed psychosis than those who did not.


The next step was to check if this pattern emerged in a different dataset, which included a different set of fMRI data for patients with schizophrenia, bipolar disorder, ADHD, and a healthy control. After looking at the network that had been previously identified in this new dataset, they found that the only statistically significant differences in brain circuitry was between the schizophrenia group and the healthy control group, which further supported the network’s role in the development of psychosis, and that this pattern is independent of fMRI paradigms.


What can we learn from this?


The network that Cao and Cannon identified is a network connecting the cerebellum, thalamus, and the cerebral cortex. This brain region in which the network is found is an error processing network, suggesting that the network might be hyperconnected to compensate for higher-than-average incoming error signals. Cannon likens the abnormal circuitry to a faulty timing belt in an engine— “the engine system can muffle along if the timing is off a little bit, but not well, and it doesn’t have the same ability to accelerate and decelerate: the performance is jumbled.”


Both Cannon and Cao did not find the network itself to be very surprising: previous studies had suggested that this area of the brain might be an indicator of whether a high-risk patient might become psychotic. However, Cao noted that he was primarily surprised with how well the results were replicated in the different datasets. “Psychotic disorders are very heterogeneous, but we only had to use a single network to not only predict psychosis but also to categorize psychosis,” said Cao.


Cannon believes that the primary implication of this study is the possibility of using information about these networks in risk prediction. This would help increase the accuracy of the “high-risk” categorization that physicians currently use with schizophrenia patients. Cannon also hopes to see this over-connected network as a target for drugs or behavioral treatments to replace current anti-psychotic drugs, which don’t attack psychosis at its source and have a variety of side effects. Cao also notes the potential for the network identified in the study as a biomarker or predictor for psychosis. He added that these findings help support previous studies and theories about the mechanism of psychosis.


What’s next?


Cao envisions two goals for the future of psychosis research. “We found that the connectivity in the circuitry is increased in psychotic patients and we know this is related to error processing deficits, but exactly how this is related to error processing deficits is not clear,” Cao noted. He hopes to better understand how the over-connected network found in the study relates to error-processing deficits in the brain. He also plans to test the power of this network as a biomarker and risk indicator for psychosis onset in other independent samples. Cannon plans to test whether patients who experience reduced psychotic symptoms end up with a less-connected network, as opposed to the highly connected network characteristic of psychosis. “The older drugs that we currently have available were discovered many years ago, have negative side effects that are problematic, aren’t curative of the disorders, and aren’t targeting the primary issue in the brain,” Cannon added. He hopes to see the development of drugs that have a better safety profile and an alternative mechanism that directly affects the over-excited networks identified in the study.