8 June 2011. The push to intervene early to lessen the burden of psychosis faces a challenge: many people who are at ultra-high risk for psychosis will never become psychotic, and interventions could harm them. Offering interventions only to those who need them when they need them requires predicting who will become ill and when. In lieu of a crystal ball, researchers have been searching for neurobiological, clinical, and cognitive markers of psychosis risk and progression.
This was the subject of a symposium in Colorado Springs, Colorado, on 2 April 2011, organized by the International Prodromal Research Network. The symposium formed part of the Prodromal Satellite Meeting, which took place in conjunction with the 2011 International Congress on Schizophrenia Research. The day had begun with a session on interventions to keep psychosis from taking its toll (see SRF related news story) and continued with updates on the hunt for markers—the subject of this story.
Less invasive than colonoscopy
Neuroimaging research to identify brain changes that could flag subjects who are on the edge of psychosis has taken its cue from studies of diagnosed schizophrenia. In the first talk on biomarkers, Christos Pantelis, University of Melbourne, Australia, said that voxel-based morphometry studies have found reduced gray matter in frontal and temporal regions in established schizophrenia. These losses appear particularly large in the anterior cingulate/medial prefrontal cortex and insula. In contrast, Pantelis said researchers have had less luck finding reliable markers of impending psychosis.
Given that the brain is still maturing at the typical ages of psychosis onset, Pantelis thinks trajectories of structural and functional change in the brain will reveal more than static markers for the transition to psychosis. Therefore, he said, researchers should “search for a moving target” by conducting longitudinal studies. In new data, he and his colleagues found ventricular differences among subject groups that were defined by illness stage. Ventricular enlargement correlated with illness duration in those with established schizophrenia but did not appear in clinical high-risk subjects before psychosis; rather, it occurred after schizophreniform first-episode psychosis appeared.
Previously, Pantelis and colleagues found that thinning of the anterior cingulate cortex predicted which ultra-high-risk subjects would develop schizophrenia-spectrum psychosis (Fornito et al., 2008). In a recent longitudinal study, they used voxel-based morphometry to compare regional gray matter volumes in ultra-high-risk subjects. Those analyses revealed a smaller subcallosal cingulate gyrus in subjects who transitioned to affective psychosis than in those who converted to schizophrenia (Dazzan et al., 2011), hinting that the brain may change differently depending on the kind of psychosis brewing.
The next speaker, Philip McGuire, Institute of Psychiatry, London, U.K., has been seeking structural, functional, and chemical clues in the brain that would predict transition to psychosis. In a five-site study (Mechelli et al., 2011), he and his colleagues paired magnetic resonance imaging with voxel-based morphometry to assess the volume of gray matter in the inferior frontal, parahippocampal, and superior temporal cortices. Reduced parahippocampal volume predicted which at-risk subjects would develop psychosis during the follow-up period, which lasted an average of two years. In contrast, volumes of the inferior frontal and superior temporal gyri failed to foretell who would become ill.
As for neurochemical markers, McGuire's group had previously found striatal dopamine abnormalities in subjects with first-episode schizophrenia and, to a lesser extent, in ultra-high-risk subjects (Howes et al., 2009). More recently, he and his colleagues tied dysfunction of the prefrontal and medial temporal lobes, as reflected in the blood-oxygen-level-dependent signal, to high striatal dopamine in high-risk subjects. Those who went on to develop psychosis showed the most marked abnormalities. Using positron emission tomography, the researchers also found that striatal dopamine production increased in prodromal patients around the time that psychosis developed (Howes et al., 2011).
Other work explored whether changes in glutamate would reflect the transition to psychosis in ultra-high-risk subjects. According to McGuire, thalamic glutamate levels at baseline failed to flag those who would develop psychosis later on; however, at the time of transition, these levels decreased. Furthermore, using both positron emission tomography and magnetic resonance spectroscopy, the study found interactions between striatal dopamine and hippocampal glutamate in high-risk subjects, especially those who transitioned, but not in healthy controls.
Taking a different tack, the group led by René Kahn, University Medical Centre, Utrecht, The Netherlands, has been weighing the effects of genetic factors versus illness progression on brain changes in schizophrenia. For instance, he and others in the STAR (Schizophrenia Twins and Relatives) Consortium performed a large, international study in which twins and their siblings underwent magnetic resonance imaging. The study found that genetic effects explained most of the correlation between schizophrenia and reduced cerebral volume.
In another study, Kahn and colleagues used fiber-tract diffusion tensor imaging to evaluate white matter integrity in patients with schizophrenia, their non-psychotic siblings, and healthy subjects. Cross-sectional analyses showed decreased fractional anisotropy in the genu of the corpus callosum in patients as they aged. Neither siblings nor controls showed an age effect or differed from each other. According to Kahn, the results show a possible effect of illness progression in the genu.
Kahn also presented findings from a recent study that compared brain changes in youngsters who were at ultra-high risk for psychosis and their healthy peers (Ziermans et al., 2010). Subjects received magnetic resonance imaging at baseline and two years later. Compared to control subjects, at-risk young people who became psychotic lost more total brain volume and white matter volume; they also lost more total brain volume than those at similar risk who did not become psychotic. Furthermore, those who developed psychosis showed more thinning than control subjects in the left anterior cingulate, precuneus, and temporo-parietal-occipital regions. These findings, which Kahn said did not result from antipsychotic drugs, suggest that brain structure changes as psychosis appears.
As for the implications of the neuroimaging findings, discussant Charles Schulz, University of Minnesota, Minneapolis, said that biomarkers could be crucial for personalizing treatment. He noted that most of the scans are noninvasive and painless, but their risks must be weighed against potential benefits. Drawing parallels to heart disease treatment, which takes into account multiple risk factors such as cholesterol levels, chest pain, and smoking history, he cautioned against relying on just one measure of psychosis risk.
Back to behavior
The last session of the day discussed clinical and cognitive markers of psychosis risk and progression. Barbara Cornblatt, Zucker Hillside Hospital, Glen Oaks, New York, focused on functional measures, particularly the ability to carry out roles at work or school, and to interact with other people. She said that researchers typically study social and role functioning as outcomes, but she sees them also as long-standing risk states that arise independently of psychosis and affect illness progression. Accordingly, she has been studying their determinants and consequences in clinical high-risk young people.
In the model that guides Cornblatt’s work (Cornblatt et al., 2003), the road to psychosis goes from negative symptoms alone to positive symptoms of increasing severity, then to psychosis-like symptoms that stop short of diagnosable psychosis, and finally to frank psychosis. In new work, Cornblatt and colleagues grouped prodromal subjects from the RAP (Recognition and Prevention) program according to their prodromal stage vis-à-vis the model and examined their subsequent outcomes. Results showed that groups further along in the model showed higher rates of conversion to psychosis. Furthermore, at the three-year follow-up, nearly 30 percent of those who had started with only negative symptoms now reported positive symptoms, and 16 percent of those with moderate positive symptoms had them worsen.
Regression analyses to identify predictors of psychosis found that disorganized communication, a higher number of symptoms, poor verbal memory, and worsening social functioning upped the risk, whereas mood disorder lessened it. In further analyses, subjects with positive symptoms showed deficits in social and role functioning that stayed stable over three years. Cornblatt sees both social and role functioning as ripe for early intervention.
After Cornblatt’s talk, Larry Seidman, Harvard Medical School, talked about the usefulness of neuropsychological measures during the prodrome. He asked whether they could help with defining “caseness” and, in particular, with identifying subgroups of prodromal subjects with different treatment needs. Noting that some people show no neuropsychological deficits before they become ill, he said that combining disparate groups might yield misleading conclusions.
Seidman and researchers at eight sites have been conducting the North American Prodrome Longitudinal Study (NAPLS), which previously found below-normal cognitive performance in clinical high-risk subjects, especially those who later developed psychosis (Seidman et al., 2010). Their new study classified subjects at baseline as having a normal, borderline, or abnormal neuropsychological profile, based on a cognitive test battery that spanned eight domains.
In data hot off the computer, Seidman and colleagues found that less than half of prodromal subjects who developed psychosis had an abnormal neuropsychological profile; even so, profile severity doubled the odds of psychosis. Subjects with poor profiles showed greater deficits in social, role, and global functioning at baseline. More importantly, the profiles predicted all three kinds of functioning a year later, over and above conversion status. According to Seidman, the results show the usefulness of neuropsychological measures for predicting conversion, social function, and role function.
Longer-term outcomes concern Alison Yung, University of Melbourne, Australia. She noted that the longest published follow-up of at-risk subjects has been about five years, during which nearly four of every 10 subjects developed psychosis. In contrast, her new study looks at outcomes of up to 15 years. She and her colleagues followed up subjects who had participated in studies at the PACE (Personal Assessment and Crisis Evaluation) clinic years ago. In a process that she said exhausted the research assistants, they tracked them not only through traditional sources, such as death records, but also newer ones like Facebook.
During the follow-up period, 27 percent of the subjects became psychotic. Most did so during the first two years, and no one transitioned after a decade. In adjusted regression models, long duration of symptoms, low Global Assessment of Functioning scores, and year of joining the study predicted who would develop psychosis. For each 10-point gain in function, the likelihood of transitioning to psychosis declined by 61 percent. Later cohorts were less likely than earlier ones to transition, but the gap between cohorts was narrowing as time went on and could have been an artifact.
To Yung, transition seems arbitrary, so she has been exploring how well transition status predicts long-term functional outcome. In preliminary findings, transitioning quadrupled the odds of poor global function at follow-up. However, only about 40 percent of those who transitioned had a poor functional outcome, and nearly half of those who functioned poorly had not transitioned. The results back Yung’s view that not all transitions matter in real life (see Yung et al., 2010).
After Yung’s talk, discussant Deanna Barch, Washington University, St. Louis, Missouri, said that most prodromal studies have focused on symptomatic help-seeking patients with poor role functioning and cognitive deficits, but she is glad to see population-based studies that aim to disentangle functioning from psychosis. In addition, she hopes prodromal researchers will think about how social and role deficits might raise psychosis risk.
Rather than pit behavioral and biological markers against each other, Barch said the two kinds of information complement each other; they might point to different mechanisms and different levels of analysis for the same mechanisms. She noted that brain measures might be less clinically feasible than behavioral measures in some settings; social and role function may be further from neurobiology, but could help clarify outcomes. She encouraged researchers to pursue both behavioral and biological markers of emerging psychosis.—Victoria L. Wilcox.