Email Icon Facebook icon Twitter Icon GooglePlus Icon Contact

User Top Menu

Schizophrenia Genetics 2015—Part 5, Plan of Action

24 Aug 2015

In SRF's five-part 2015 schizophrenia genetics update, reporter Michele Solis surveys leaders in the field about milestones, challenges, and current research.

See Part 1, Renaissance; Part 2, From Discovery to Understanding; Part 3, Rare Allure; and Part 4, Rethinking Diagnoses.

Download a PDF of the entire series.

August 25, 2015. Between schizophrenia's landmark genomewide association study (GWAS) last year and reports of rare genomic variants that increase risk for the disorder, there is no shortage of genetic clues to follow up. As the gene discovery phase continues apace, it will continue to add to this line-up of hundreds of suspect genes. Researchers and drug companies must wrestle with how to prioritize these clues to find therapeutic targets efficiently, a decision that should not be taken lightly.

"We've seen examples where labs have launched major efforts on genes where the evidence simply wasn't unequivocal," said Patrick Sullivan of the University of North Carolina in Chapel Hill and a leader of the Psychiatric Genomics Consortium (PGC). "I think neuroscientists need to be far more cautious than they sometimes are. We want people to work on the best clues."

Along with the choice of genes comes a choice of directions. One path is an attempt to relate the genetic findings to how schizophrenia takes hold in the brain, which potentially leads to new therapeutic targets. Another path looks for subtypes of the disorder, which may involve a combination of genetic information and other biomarkers. This may eventually prove useful in the clinic to predict risk or outcomes. Matching a person's genetic and/or biomarker profile to a customized treatment is the hope of precision medicine, a fledgling idea for psychiatric disorders (Insel et al., 2015).

Moving toward personalized medicine will depend on the efforts of many. The deluge of genetic data, and the scale of the follow up it requires, is reshaping the way science is done both in academia and industry. With a collaborative framework becoming the norm, researchers will have to tackle questions about how to assign credit, how to rigorously evaluate their work when every known expert is a collaborator, and how to maintain individual research programs.

"It's a big-data science now, which changes the whole experience pretty significantly," said Jonathan Sebat of the University of California, San Diego. "But we haven't abandoned those really rewarding solo projects that we love doing."

He added, "We could easily devote all of our time to big science projects because there's that much richness in the datasets. But of course we need to maintain a balance between contemplation and conference calls."

The expectation is that giant collaborations will set the stage for productive smaller-scale research. "Anyone who wants to solve the problem has to work in large groups, both the genetics and the stem cell biology must occur at scale, and then the useful transfer of information requires interdisciplinarity," said Steven Hyman, director of the Stanley Center for Psychiatric Research at the Broad Institute in Cambridge, Massachusetts. "Once all these initial observations are made, there will be many hypotheses for small labs to test."

Finding targets

Despite the tiny effects of common variants found by GWAS, drug companies are sifting through them for new drug target gems. Pfizer, a member of the PGC, has developed a pipeline to sort through the latest GWAS hits to come up with targets that may generalize to the schizophrenia population as a whole (described in Schubert et al., 2014). This approach banks on the fact that while the small but significant GWAS hits may say little about biological effect size, they do have an unequivocal relationship to disease.

For Pfizer, high-priority targets include calcium channels, glutamatergic signaling, and intracellular signaling molecules along the Akt/mTOR pathway, which may be selectively perturbed in the service of dopamine signaling.

For prioritizing other genes, the company sees gene expression data from the brain as critical, prompting Pfizer to collaborate with the Lieber Institute for Brain Development's RNA-Seq Consortium (see SRF Genetics Series, Part 2). Integrating other veins of information will also delineate worthy targets; for example, knowing how druggable a molecule is, or with what molecules it interacts, could lead to targets that don't belong to the GWAS hit pantheon.

"We are trying to escape from this perception that a clear GWAS hit has to be your target," said Patricio O'Donnell of Pfizer in Cambridge, Massachusetts. "I think we can use the current information to point to a network of genes, ask what are the biological processes that these genes are serving, and then find the way to intervene there."

AstraZeneca in Cambridge, Massachusetts, has made collaboration their mainstay. In 2012 the company downsized drastically, closing its labs but preserving a very small drug development team that formed partnerships with academics or other companies. They also are working with the Lieber Institute to integrate data from genetics, brain transcriptomes, and induced pluripotent stem cells (iPSCs).

"With so much information breaking in neuroscience in the last five to 10 years, we saw that we had to be more flexible and work externally in a much more complete way," said Nick Brandon of AstraZeneca. "Now, if things don't work out for a drug candidate, we can change very quickly to another."

Brandon expects that this approach will deliver drug candidates for schizophrenia in five years.

Subdividing schizophrenia

Both company research leaders stressed the importance of identifying subgroups of people with schizophrenia who may respond differently to treatment. Drawing from biomarkers and genetics alike, researchers might be able to verify that a particular target is indeed dysregulated in a subgroup before advancing to clinical trials.

Pfizer is involved in analyzing a well-phenotyped cohort collected at the University of Maryland to search for subgroups. With genotype, brain scans, brain electrophysiology, cognitive testing, and blood samples collected from 1,100 people, 400 with schizophrenia, the team hopes that combining the information will reveal distinct subgroups of people with different causes of their illness.

"This would give us a more personalized medicine," O'Donnell said. "It's going to take a few years to get there, but I think we now have the tools to do it."

One of the first subtypes to be delineated may be treatment-resistant schizophrenia (TRS). About one-third of people with schizophrenia have TRS, meaning that symptoms are not alleviated by antipsychotic drugs that block dopamine signaling. Brain imaging finds that those with TRS do not have the excess dopamine signals found in others with schizophrenia (see SRF related news report), and genetic differences are also falling out of GWAS that compare those with TRS to others with schizophrenia (see SRF related conference report). If a gene test or series of biomarkers could be developed for TRS, this would spare patients the lengthy antipsychotic trial and error. It might also explain why clinical trials of glutamate pharmacology, which did not select patients for glutamate disturbances, failed (see SRF related news report).

Although people develop schizophrenia for different genetic and environmental reasons, whether genes alone can extract subtypes remains unclear. In 2014, a controversial study purported to identify eight different subtypes by analyzing the patterns of SNPs loosely associated with schizophrenia by early GWAS. The study was pilloried, with critics pointing to more mundane reasons for the segmentation (see SRF related news report).

Yet purely genetic information can be put together to estimate a person's risk for schizophrenia, such as the polygenic risk score (see SRF Genetics Series, Part 1). This index can partition people into groups with different levels of risk, which may be useful clinically, though it lacks the selectivity and specificity needed for an actual predictive test for an individual.

"You're never going to get an accurate estimate of risk for an individual from genetic data alone," said Naomi Wray of the University of Queensland in Brisbane, Australia. Wray and colleagues have been investigating ways of combining genetic data to improve risk measures (e.g., Maier et al., 2015).

This type of approach might also guide treatment for those in the earliest stages of mental illness. For example, of young adults reporting mild delusions and hallucinations, between one-quarter and one-third later develop full-blown psychosis (Fusar-Poli et al., 2013). A genetic tool could identify a sector of people with these symptoms who are at high genetic risk, and that group might be enriched for those who eventually transition to schizophrenia, Wray said.


As in the rest of medicine, pharmacogenomics is a hope in psychiatry. It is assumed that people's genetic makeup will influence how they respond to different drugs, and so far the only traction for schizophrenia has been in side effects. A GWAS exploring antipsychotic-induced weight gain found a common variant near the MC4R gene, which encodes a melanocortin receptor with links to obesity (see SRF related news report).

MC4R is not among the PGC's hits, and probably has nothing to do with schizophrenia's origins. This means that genetic variants important for treatment response, such as those controlling drug absorption or metabolism, need not be among the disorder-related hits.

"Our position on this is really that the genes for the disorder may not necessarily have any pharmacogenetic impact," said Anil Malhotra of Zucker Hillside Hospital in Glen Oaks, New York, who led the weight gain study, and who founded and organizes the Pharmacogenetics in Psychiatry conference.

Another study last year linked two rare variants in HLA that confer high risk for agranulocytosis, a life-threatening side effect of the antipsychotic clozapine. Half the cases of agranulocytosis did not carry either variant, however, limiting the findings' clinical utility (Goldstein et al., 2014).

"Drug response may not be as complex as schizophrenia, but it's still pretty complex," Malhotra said, noting that ideally a study would treat subjects with the same drug at the same dose for the same duration. "So I think we need larger sample sizes with these parameters held pretty steady to start to be able to discern those genes."

But this complexity leaves funding agencies skittish.

"Pharmacogenomics is a tricky and difficult field," said Thomas Lehner, branch chief of genomics research at the National Institute of Mental Health (NIMH) in Bethesda, Maryland. Noting that drug metabolism can be a slippery phenotype, with multiple interacting and sometimes opposing variants involved, he added, "We should only make limited investments in it at the moment."

Dawn of the team science era

Lehner stressed the need to push research in all directions, with an eye toward collaboration. For example, the NIMH started a partnership with the Stanley Foundation to form the Whole Genome Sequencing for Psychiatric Disorders (WGSPD) Consortium to comprehensively address the function of the genome's mysterious regulatory regions (see SRF Genetics Series, Part 3). He also called for work geared toward understanding the biology behind the genetic clues rolling in.

"Would I say that the PGC has done its duty and should be disbanded? I don't think so. The PGC serves an important function," Lehner said, noting the PGC's Psych Chip experiments, which will allow more gene discovery and replication at a fraction of the cost of previous GWAS.

As consortia continue to mushroom, the field seems to grow more comfortable with collaboration as a necessary reality, though some worry the drive for large datasets will create a gap between knowledge and understanding.

"I see the field of schizophrenia genetics, and more generally the origins of mental illness, to be in a state of chaos induced by too much information," said Irv Gottesman of the University of Minnesota. "It may be generating too many data for our current comprehension to go forward to biology."

The shift toward collaboration will also reshape how funding agencies and academic departments judge an individual's work. While doing science in a big group might seem to make it hard for an individual researcher to make a mark, there are also benefits.

"It's a great opportunity to people starting out in the field," Francis McMahon of NIMH said. "There's always a role to play in these consortia. Just get involved, learn a lot, and get access to datasets you couldn't dream of getting on your own."

Even the Stanley Center, with its recent windfall of hundreds of millions of dollars, sees collaboration as critical. "I think that our good fortune really demands of us that we support not only our own efforts, but other peoples' efforts, to understand schizophrenia and to facilitate a path to new treatments," said Hyman.

"Team science is never going to go away," he added. "But it will coexist with small science as well."—Michele Solis.