6 Oct 2015
October 7, 2015. The promise of genomics research, when combined with a host of new techniques from the molecular to the clinical and epidemiological, was the theme of the 2015 Symposium on Severe Mental Illness at the Stanley Center for Psychiatric Research at the Broad Institute in Cambridge, Massachusetts, September 28-29, organized by Guoping Feng of the Massachusetts Institute of Technology and Steve Hyman of the Stanley Center. During the opening remarks, Louis Gerstner Jr., chair of the board of the Broad Institute, offered a simple goal for the meeting: "I hope all of you will leave the meeting energized to bring scientific breakthroughs closer and closer to people around the world suffering from psychiatric disorders."
Here we report on some of the data related to schizophrenia, along with a sampling of the groundbreaking basic neuroscience and methodology advances on display.
Mark Daly of Massachusetts General Hospital (MGH) and the Broad Institute kicked off the meeting with an overview of genomewide association studies (GWAS) and exome sequencing in autism and schizophrenia, with a focus on rare variants, which are hypothesized to be more damaging. However, given the rarity of these, finding them in the population will require huge sample sizes—thus, the current interest in family studies, particularly for finding de novo variants. However, Daly said that it appears that all people, on average, carry one de novo mutation in their exomes, the protein-coding portion of their genomes. It is hoped that resources such as the Exome Aggregation Consortium's database, which harmonizes data from different labs, can help researchers determine which de novo mutations are likely to be relevant to disease.
There is also a line of Broad research to identify genes that are more likely to be hit by deleterious mutations, and Daly's group has evidence that such genes are more likely to generate a signal in schizophrenia GWAS samples that otherwise did not offer up risk loci.
Rare-variant work is much more advanced in autism and other early-life mental disorders, and Daly discussed preliminary work indicating that the de novo mutations in intellectual disability/development disability/autism spectrum disorders overlap with some of the positive schizophrenia GWAS loci, suggesting that genes that contribute a small amount via a common variation to schizophrenia risk have a greater influence when disrupted more severely, leading to developmental disorders. A theme appears to be emerging of some overlap between rare and common risk allele loci.
In the next talk, Ben Neale of MGH/Broad Institute continued the discussion of rare variants, describing work from the Psychiatric Genomics Consortium (PGC) on copy number variants (CNVs). He reported that in their current analyses of bigger GWAS data sets, PGC researchers are finding stronger support for previously identified deletions and duplications.
Neale also gave an overview of the PsychChip, which is already being deployed to genotype subjects for further GWAS, with especially large numbers in Denmark to date, including 11,000 schizophrenia cases. In early results, positive associations are turning up in childhood-onset disorders such as ADHD and autism, with evidence for overlap with some of schizophrenia's risk variants.
Like Daly, Neale was careful to emphasize that these results will be just the first building blocks for finding treatment targets. "A mountain of work is required for translating these findings into biology," he said.
In reply to an audience member who asked whether better phenotyping might not be critical at this juncture, Neale offered that we may see an iterative process in which genetics on current phenotypes provides some leads until we can do further genetic analysis of better phenotypes, which themselves may be informed by genetic findings.
Naomi Wray of the University of Queensland in Australia described her work as focusing on the interpretation rather than the generation of genomic data. As an example, she discussed recent attempts to make sense of the association between older paternal age and schizophrenia using data from the PGC. Contrary to some earlier reports, her analyses find that the age of the father at the birth of his first child, as well as the age of the mother, drives the association. Indeed, the effect of age at first birth has a U-shaped curve, with increased risk of psychiatric disorders for very young and very old parents (see also McGrath et al., 2014). This suggests that psychosocial factors may play a bigger role than previously expected.
Wray also mentioned that by using the current polygenic risk score (PRS) derived from GWAS, it is possible to predict risk with about the same accuracy as family history, suggesting that the score could at some point contribute to clinical risk prediction. One way to further improve prediction is to use information from other disorders, not just psychiatric ones (see Maier et al., 2015), but non-psychiatric disorders as well.
Her talk wrapped up with some political discussion, particularly of the highly touted NIH Precision Medicine Initiative, which unfortunately does not appear to hold much promise for psychiatric disorders given that it is only collecting data on people 18 years or older. Thus, it will miss early data for disorders such as schizophrenia, as well as autism, ADHD, and other early-onset disorders. It may also divert attention—and funding—from the BRAIN Initiative, opined one audience member.
In the next talk, Pamela Sklar of Mt. Sinai Medical Center in New York City gave an overview of the Common Mind Consortium, which is collecting and making available to other researchers standardized data on gene expression, as well as genetics, epigenetics, and other genomic data on psychiatric disorders from postmortem tissue. They now have data from about 700 cases, mostly schizophrenia.
Sklar went on to talk about current work using expression quantitative trait loci (eQTLs) signals in the PGC data. Through much filtering of data, this has generated solid leads on five genetic loci. Most of such eQTLs are expected to have very small effect sizes on disease risk. Nonetheless, early studies in zebrafish of the five candidates have generated some significant differences in morphology, specifically head size (see related SRF related news report).
Sklar also mentioned that they have data suggesting that people with treatment-resistant schizophrenia may have major lesions to genes that only contribute small effects via common variants.
Preben Mortensen of Aarhus University in Denmark wrapped up the genetics talks with an overview of the iPSYCH collaboration, which will combine genetic, clinical, and environmental data in the study of five major psychiatric disorders. What iPSYCH has in Denmark is 2.2 million people born since 1955 with clinical diagnoses and blood spots. Although most of those cannot be contacted for further characterization, the researchers also have ongoing studies of a subset and are recruiting new subjects. To date, they have genotyped more than 80,000 cases of different disorders.
Early data from the project have found that the PRS and family history only partly overlap—it appears that about half of family history is mediated by PRS, and that people with both a high PRS and family history of psychiatric disorders have an even higher risk of developing schizophrenia (Agerbo et al., 2015).
The genetics talks were followed by some examples of current and possible follow-up on GWAS candidates. Steve McCarroll of the Broad Institute gave an update on his group's exploration of the MHC region of chromosome 6. As reported from last year's World Congress on Psychiatric Genetics (see SRF related conference report), his group has found a signal in a subset of PGC cohorts that points to the complement component 4 (C4) gene, which is found in several forms, varying in copy numbers from person to person. The number of copies determines expression levels, which have been linked in postmortem tissue to schizophrenia risk. McCarroll and colleagues have begun to work out the details of which C4 forms are involved.
McCarroll emphasized that this locus makes a very small contribution to schizophrenia risk in the population, but through C4 it does provide immediate entrée to biology, as there is a growing body of work on the role of complement proteins to tag synapses for elimination by microglia. This work was highlighted by the next speaker, Beth Stevens of Boston Children's Hospital and the Broad, who will take her work into the schizophrenia realm to find out whether microglial pruning of synapses, driven in part by altered complement signaling, could play a role in disease risk, perhaps during adolescence, when cortical connections are heavily pruned. Stevens received a big round of applause when it was announced during the meeting that she had been awarded one of the MacArthur Foundation's "genius" 2015 awards that morning.
The schizophrenia prodrome
Raquel Gur of the University of Pennsylvania in Philadelphia kicked off a session that moved far from the genetic and molecular realm and into the clinical and phenotypic. The Philadelphia Neurodevelopmental Cohort, which she co-leads, is a population-based sample of over 9,500 subjects who were recruited during childhood when they received care for a range for things: from well-child visits and minor problems to chronic or life-threatening health problems. All have been genotyped, and many have had detailed psychiatric assessments, neuropsychological testing, and neuroimaging.
In her talk, Gur reviewed data previously published (see SRF related news report) or presented at meetings (see SRF related conference report) that show early signs of abnormal brain function as predictors of later psychiatric disorders and describe cognitive or imaging abnormalities in children with subthreshold schizophrenia spectrum symptoms (Wolf et al., 2015; Satterthwaite et al., 2015). Following up with these patients, Gur and colleagues find that children who have persistent schizophrenia-spectrum features (e.g., hallucinations or delusion-like experiences that do not warrant a psychosis diagnosis) are more likely to have other psychiatric disorders as well.
This was followed by an overview of a stepwise intervention personalized for young people with psychiatric symptoms, presented by Patrick McGorry of Orygen Youth Health in Melbourne, Australia. He noted that people we denote as "at risk" of developing psychosis already have significant disability, as reinforced by a recent meta-analysis by Fusar-Poli and colleagues (Fusar-Poli et al., 2015; also see SRF interview with McGorry). McGorry reviewed the history of his work in Australia to establish both research and treatment programs for this population. The centerpiece was the approximately 400 patients from the Personal Assessment and Crisis Evaluation clinic who were determined to be at "ultra-high risk" (UHR) of developing a psychotic disorder. In the latest data from this group, one-third develop a psychotic disorder in the medium term (median follow-up 5.7 years), and only 7 percent do not develop some mental or substance use disorder, indicating that the UHR phenotype captures a wide range of disorders, and that a transdiagnostic approach is optimal for people with subclinical symptoms.
McGorry went on to describe a clinical staging process that focuses on symptoms less than diagnoses and tries to match interventions to those stages—diet, exercise, and behavioral therapy in earlier stages, with medication reserved for later stages with increasing symptoms and disability. Research conducted during this staging process might also identify biomarkers that, along with clinical variables (and perhaps even genetic profiles), can guide interventions. With a consortium of funders and researchers, McGorry is now starting a study of such an "adaptive" intervention.
In the final talk of the prodrome session, Sophia Vinogradov of the University of California, San Francisco, described the background of her research: the aforementioned deficits that appear long before psychosis is reached. In terms of the cognitive deficits that she focuses on, she assumes there is a connectivity issue, particularly in prefrontotemporal neural systems, and that this can be improved with cognitive training. As inspiration, she cites the famous study of taxi drivers whose hippocampus changes as they learn the complicated maps of London (Maguire et al., 2000), as well as a wealth of other research in animals.
She described a recent randomized clinical trial of cognitive training at home in recent-onset psychosis. Subjects who underwent an auditory training paradigm had significant improvements in cognition compared to a control group that played computer games, an effect that persists for at least six months. These changes were paralleled by changes in psychophysical auditory responses and EEG-measured gamma activation in prefrontal cortex, which in turn are associated with improvements in an executive function task. Vinogradov's group is also looking at reality monitoring, finding that people with schizophrenia do not properly activate medial prefrontal cortex (mPFC) during such tasks. Cognitive training improves task performance and mPFC activation, as well as real-world functioning.—Hakon Heimer.