Transcript
MIRI GITIK: Hello, everybody, and welcome to the NIMH Genomic Team’s 75th anniversary webinar, Celebrating Advancements in Psychiatric Genomics. Today, we will be here for more talks that highlight key advancements in genetic research supported by NIMH. Before we begin, a few housekeeping items just for your own reminder. Please submit your questions via the Q&A box any time during this meeting. We will have a dedicated Q&A session at the end of the webinar. Please do not forget to address your questions to the speaker you would like to respond. If you encounter any technical difficulties hearing or viewing this webinar, please reach out via the Q&A box or send an email to the email that appears on the screen. This webinar is being recorded and will be posted on the NIMH website for later viewing.
And now we would like to share with you opening remarks by the NIMH Deputy Director, Dr. Shelli Avenevoli.
SHELLI AVENEVOLI: Good morning, everyone. I want to welcome you to the NIMH Genomics Team’s 75th Anniversary webinar, Celebrating Advancements in Psychiatric Genomics. Since its inception, the National Institute of Mental Health has recognized the importance of genetics in understanding mental disorders. This is seen in the many consortia and initiatives that have been supportive. There has been an emphasis on gene discovery, functional genomics, and translation to clinical practice. In today’s webinar, we will hear about four important areas of genetics and genomics research that have been supported by NIMH.
First, you’ll hear from Dr. Patrick Sullivan, who will describe the Psychiatric Genomics Consortium, a highly impactful initiative that has led to fundamental insights about the genetic basis of mental illness. Next, Dr. Christa Martin will talk about both basic and clinical aspects of psychiatric disorders. Dr. Martin is also a principal investigator of the Genes to Mental Health Network, which focuses on understanding the role of genetic variation in rare neurodevelopmental and psychiatric disorders. Then, Dr. Flora Vaccarino of the PsychENCODE Consortium will describe molecular studies and developmental trajectories of mental illness. And finally, Dr. Frances McMahon from the Human Genetics Branch of the NIMH Intramural Research Program will describe exciting directions of relating genomics findings to patient care.
As you listen to these presentations, you will hear that much of the progress has involved large consortia supported by team science, involving collaborations between large numbers of investigators, and in many cases, collections of vast numbers of participants. These talks will also highlight many future trends, such as incorporating an understanding of individual genetic risks and how these can better inform therapeutic interventions. So again, I want to thank our speakers and all of you for joining us today in this celebration of the NIMH genomics program. Thank you.
MIRI GITIK: Thank you very much. And I think we can move to our first presentation by Dr. Patrick Sullivan, who will highlight the Psychiatric Genomic Consortium, PGC.
PATRICK SULLIVAN: Good. Thank you very much for the kind introduction. Is my screen visible? Awesome. Okay. I will be talking about the Psych Genomics Consortium today. And let me begin with just a little bit of a glossary because we need a couple of pieces of common language. When I say genetic variants, I mean things that are measurably different between individuals. There’s many different types, but that’s the general term. Single nucleotide polymorphisms, or SNPs, are variants at a very specific place, but it just changes one base in the genome.
And then there are the three methods that we’ll use for discovery. Genome-wide association, or GWAS, which looks at common variation across the genome that are associated with psychiatric disorders, more common in cases than controls. Whole-exome sequencing is a newer method by the way we can find genetic variants in the protein-coding parts of the genome. And then finally, copy number variants. These studies are a method to try to find structural variations. Big hunks of the genome that may be deleted or duplicated, containing from one to 50 or so genes. We have a review coming out shortly that covers all this.
So the PGC began, is currently a bit of a thousand scientists from 50 plus countries. It began in 2007 with the foundation for NIH game projects, and the four PIs of those studies actually came together to begin the PGC. From 2008 to now, we’ve had — we were extremely appreciative that the NIMH has seen fit to fund us four times so far. There’s been four grants along with some support from NIDA, National Institute for Drug Abuse. And the fifth grant is being prepared. We currently have 11 disorder groups, ADHD, autism, bipolar disorder, all the way through to schizophrenia and substance use disorders. There’s five cross-cutting groups in particular. There’s a new suicide group and also the copy number variation group. And there’s a new functional genomics group that we’ve just begun.
From 2020 to now, we’ve greatly increased and worked hard on the globalization of the PGC. So instead of being predominantly European, we’re moving toward a more balanced representation of the world in our studies. And from 2008 to 2024, the PGC has published 654 papers plus 32 med archive preprints. These papers have been cited an average of 162 times per paper and the H-index for the PGC, if the PGC were an investigator, is something like 110. And at the lower right is the link or the PGC itself.
This is an image showing a word cloud of the terms in our abstracts. This is a tree map of the different areas that our studies have focused on, from psychiatry, genetics, heredity, all the way through to pharmacology, pharmacy, statistics, cell biology, et cetera. This is a graph of the number of papers the PGC has published per year up to as of yesterday. There’s been, again, 685 publications. And from 2018 on, we’ve averaged well over 75 papers per year with the exception of the COVID downtick in 2020. And then finally, this is from the NIMH’s eyesight. And this shows one dot is for the impact of every single PGC paper. A typical paper weighs in right in the middle at 50 percent. And the typical PGC paper is above the 75 percent. And as you can see at the very top, we’ve had quite a few extremely high-impact papers over the years. And as far as I’m aware, the PGC is the longest, largest, most inclusive, most productive, and most impactful scientific effort in the history of psychiatry.
In terms of just the general discoveries that I can tell you about, first of all, it enables deep dives in the fundamentals. It really is the main method that we have to really unequivocally get at what is the basis of psychiatric disorders. We’re team scientists, as was mentioned in the introduction, and we’ve been known for our absolutely uncompromising rigor and the deep collaborations that we’ve had. And because we are exposed to our genomes from fertilization on, it’s truly fundamental.
And again, we combine scientific efforts across the globe and across the lifespan. So we study childhood-onset disorders, particularly autism, ADHD, eating disorders, disorders that are common in childhood as well as adults, anxiety, bipolar disorder, major depression, et cetera, schizophrenia, late life, Alzheimer’s. And we also look at crucial clinical features, which of course are substance abuse and suicide. And we’ve included both academic industry collaborators. And the results that we have to date underscore that these are deeply complex human conditions. And we embrace complexity because that’s the way nature wrote these conditions. They’re not easy at any single level, and we take that super seriously. And I believe that every year the PGC gets closer to the goals of understanding the fundamental biology because from biology leads to rational diagnosis, prevention, and ultimately treatment.
So I’m going to mention five stories of work the PGC has done. And the first and the biggest one actually is almost taken for granted now, but I need to emphasize that we’ve gone from confusion to secure knowledge. There were tons of papers before 2008 using the best available methods. And unfortunately, as we’ve sort of documented, they did not have a lot of enduring support. In addition, there were a bunch of non-replications, a bunch of high-profile things that ended up not being true in the long term. And so we’re in the cycle where someone would make a claim, lead to excitement, but then immediately non-replication, confusion, and repetitive cycle. And I believe we’ve broken that cycle. In fact, before 2008, there was still debate as to whether psychiatric disorders were genetic in any way. And that clearly is a settled condition now.
And there’s been an orderly progression since then. There’s been no retractions of the major papers. Successfully larger studies have taught us more. And again, there’s been this uncompromising rigor that’s characterized the by PGC investigators as well as others whom we work with. And the initial discoveries remain the strongest. So for schizophrenia, for instance, in 2009, the major histobatic compatibility complex came up from genome-wide association studies, and that remains the strongest hit. The 2016 finding using exome sequencing for SETD1A, again, still the strongest, and the same for 22q11 as a copy number variant.
Indeed, in a paper we published two years ago, we found that 107 of the 108 significant 14 remain significant. That’s an awfully, awfully good percentage of things that are ended up being true in the long term. And what — in addition, because of our liberal data-sharing policies, many groups study psychiatric disorders that would not do otherwise. And that ends up being a multiplier. And I note that multiple biotechs are actually looking using the data that we’ve generated, they’re looking at drug development possibilities.
Second story is to focus on schizophrenia. And this was the paper in 2014 in Nature, was really one of the landmarks. And it’s one of the highest — it is the highest cited paper the PGC has ever produced. And in that year, in fact, it was the third highest — third most cited paper in all of our medicine. So extremely impactful. And I’m going to walk you through this graph, which sort of summarizes what we’ve learned. So across the bottom, we have how common a variant is, all the way from super rare to super common. And then up on the y-axis going up, we have the effect size, how big of a hit this is, how big of a risk this is. And as you can see, there’s several different territories here. So the things that have been discovered from whole-exome sequencing and cognitive variation tend to be rare but a stronger effect. I would note that absolutely none of these is deterministic. Even if you have one of these variants, the chance of actually getting a psychiatric disorder like schizophrenia is not 100 percent, it’s far less usually.
In the lower right-hand corner, we have the variants discovered by genome-wide association, which tend to be common but have much, much smaller effect. And the shaded blue in the upper right is where we had way over 99.99 percent power to find something if it was there. The famous APOE Alzheimer’s Association is given there. And as you can see, we found nothing in that area. And we can’t completely exclude it, but I think it certainly is fairly unlikely. And if we delve deeper into these things, there’s some tantalizing stories here.
Again, the first thing is that schizophrenia, for instance, is super complex. It’s not just the genetic part, it’s also the environmental part. What happens to people, what they choose to do, exposures, et cetera. And the really interesting thing about this is that we currently have no theory upon how to knit these all together in a really compelling way. We certainly have ideas. There’s a very strong synaptic biology signal in this, for instance. But at the same time, it’s really important that we engage biologists to help us look at this in concert with the genomic findings. And I believe most of our speakers will be talking about that as well, after my talk.
The next is — the third story is that these are syndromes and not diseases. So these are things that are defined without respect to measurable, unmeasurable lab markers, unmeasurable biochemical tests, unmeasurable pathological findings, or brain imaging findings. I’m showing here first what we see when we look at the genetic correlations, the fundamental connections of a stack of neurological disorders. And by and large, they’re not connected, except for things like the different subtypes of epilepsy, subtypes of stroke, and subtypes of migraine. So they tend to be generally their diseases, generally they have fundamental biology and pathophysiology. But if we look at psychiatric disorders, we find that there’s a lot of things connected with a lot of other things.
So for example, the basis of bipolar disorder overlaps strongly with major depression, OCD, and schizophrenia. Depression overlaps with a bunch of stuff. OCD overlaps with a whole bunch of stuff as well. This is a bit of an older paper from 2018. And in fact, this pattern has become even clearer. Unlike neurology, these are syndromes. They tend to have connections that we haven’t been able to disentangle based on clinical data.
The next one is — the fourth story, in specific, I’m going to talk about anorexia nervosa. When the researcher back in the 1980s and 90s, there were a bunch of psychologically defined ideas about what it’s caused for. Some of those are listed there. Our colleagues built a very, very careful case. They looked at whether anorexia ran in families, which it does quite strongly, using twin studies to define whether it’s terrible and it is. And then now we’re in the genomics era where there are large samples that are measured. And some really interesting things are coming up. As with many psychiatric disorders, we see that anorexia has a strong overlap with obsessive-compulsive disorder and major depression. No surprise.
But the one thing that was really interesting was that it’s got negative correlations often with a lot of metabolic ideas. So for example, insulin level, leptin, body mass, et cetera. All these different things have expected patterns. And that’s led to a remarkable possibility. Anorexia nervosa is both a psychiatric disorder and a metabolic disorder at the same time. And this has led to a whole bunch of new ideas about what the causation patterns might be and how we can actually treat people with this awful disorder better.
And then finally, from genetics to biology, the PGC is doing its part. And I’m super looking forward to hearing the talks and the speakers that come after me, except that they are touching on this for sure. But this is the Watson and Crick 1953 idea. And this is something which is almost — it’s too simple to be believed because the reality is so much more complex right now. There’s so many things going on, and it’s so intensely intertwined with the way the brain develops over time, which is itself exceptionally complicated. But it’s absolutely fascinating to see how this is going.
And as one example, and I believe Flora will get at this in a bit, were the PsychENCODE papers that came out in the Journal of Science a couple of weeks ago. So I first would like to thank the 802 people listed here, all of whom are a subset of the PGC collaborators. And in addition, I’d love to thank the NIMH for its continued support and happy 75th, and many happy returns. And I would also note that the PGC contributes to every single aspect of the NIH’s mission. And thank you very much for your attention.
MIRI GITIK: Thank you very much, Dr. Sullivan, great presentation. Moving on to the next presentation, Dr. Christa Martin, please.
CHRISTA MARTIN: Can you see my presentation, Miri?
MIRI GITIK: Yes.
CHRISTA MARTIN: [inaudible] Okay. All right. Good morning, everybody. Today I’m going to talk to you about precision health for neurodevelopmental and psychiatric disorders, or I’ll refer to them as NPD. I don’t have any personal conflicts to disclose. And today what I’ll talk to you about is a general overview of NPD, as well as genetic contributions that we’ve studied that can cause NPD. And then I’ll give a brief insight into work that we’ve done as part of the NIMH to Mental Health Network.
So what does precision health for brain disorders mean? I like to use a patient example to explain this to individuals. And this patient example is a four-year-old male who had autism and increased head size. And through genetic testing, we revealed a deletion on chromosome 17. This deletion is known to cause a variety of clinical features, including neurodevelopmental and psychiatric disorders, large head size, abnormalities of kidney and genital tract, and diabetes because of a gene called HNF1B that resides in the deleted region.
There are examples of clinical management that can now be done in this child, including evaluating his kidneys and urinary tract to monitor their function over time, check HbA1c to annually monitor for any diabetes. That’s a marker for diabetics. And we know that these individuals who carry this deletion are also at increased risk for schizophrenia. So there are certain exposures that this individual should avoid. And at some point, just as Dr. Sullivan alluded to, we may learn that certain therapies work better in individuals with these specific genetic causes compared to others.
So when I’m talking about neurodevelopmental and psychiatric disorders, I’m including symptoms that are characterized by impairments in cognition, communication, behavior, and or motor functioning. And these impact about 14 to 18 percent of our nation’s children and adults. And listed here are some of the clinical diagnoses that can be included under this umbrella, including autism, epilepsy, cerebral palsy, bipolar disorder, anxiety, et cetera. And it should be noted that these pediatric neurodevelopmental and adult psychiatric conditions actually overlap in the underlying genetic causes and presenting symptoms. So they’re not limited to either pediatric or adult conditions, but rather they have shared genetic etiologies.
So what do we know about the genetic causes of NPD today? Well, we know that up to 40 percent of individuals who are referred to genetic testing because they have one of these clinical conditions will have an identifiable genetic etiology. And just as you heard, there are different types of genetic etiologies, such as single gene variants or copy number variants that are deletions or duplications of chromosome regions that can cause these types of NPD diagnoses. And individually, these may be rare, but collectively, as you can see, they account for a large portion of what we know about NPD. And we are learning about other causes, including things that are more on our common genetic backgrounds, such as polygenic risk scores or other contributors.
Genetic testing is actually standard of care. Many people aren’t aware of that. They think that this is work that’s embedded in research. But there is genetic testing that is available and should be offered to any individual who has an NPD diagnosis. There are now multiple professional societies and expert consensus groups that have guidelines for clinical genetic testing. They may have different testing algorithms, but all I think are moving now to either exome or genome sequencing with deletion and duplication analysis. So one main takeaway from my talk today is that if your family member has an NPD diagnosis, you should ask your doctor about genetic testing.
And why is it important to know a genetic cause? So by knowing both a genetic cause in addition to any behavioral or other diagnoses, you can really put the full picture of an individual’s condition together. Knowing the medical or ideological cause, like a genetic change, and understanding a behavioral diagnosis such as autism can really allow you to understand the individual’s needs. A behavioral diagnosis alone will tell you what an individual has and how they are presenting, but it doesn’t tell you why, and it doesn’t tell you what else is needed to be done for this individual’s care.
And this is just an example where you can see each of these three children have a behavioral diagnosis of autism. After genetic testing is done, we can see that the little boy on the left has been diagnosed with Fragile X Syndrome, one of the more common causes of inherited intellectual disability. The child in the middle has a diagnosis of 22q11.2 deletion, which is a deletion on chromosome 22 that is one of the more common copy number variants that’s observed in this patient population with NPD. And the little girl on the right has a diagnosis of Angelman Syndrome, a more severe form of autism with lack of speech, intellectual disability, and other conditions.
So it’s really important that these three individuals have the etiology defined so that their individual needs can be met because these three genetic conditions have very different approaches to therapy. It also is the diagnostic odyssey for the family, so a lot of families are looking for answers as to why their child has the conditions that they do, and this can provide an end to that diagnostic odyssey. As well as now, because there are so many families who learn about a genetic diagnosis in their family, there’s a lot of patient-specific support groups popping up through social media, which can offer families the support that they need and help them to take better care of their children. As I alluded to in the first case I shared, it also can inform medical management based on genetic cause and provide more accurate recurrence risk estimates for families.
So as I just showed, one clinical phenotype or diagnosis, such as autism shown here, can be caused by many different genetic changes, and that’s known as genetic heterogeneity. Similarly, we also know that one genetic change, like this deletion on chromosome 22, can cause many different clinical diagnoses. So you can see that individuals with 22q deletions can have autism, intellectual disability, epilepsy, et cetera. And we know that individuals with this 22q deletion, up to 25 percent of them, are at risk for schizophrenia. And so again, learning this information can allow us to monitor for these conditions more appropriately.
Furthermore, even individuals with the same genetic change within a single family can exhibit variable phenotype. So this family has a deletion of 15q13.3. So every individual on this pedigree that is colored red has that deletion. But as you can see from the conditions below each of the individuals shown, females (circles) or males (squares) in red, they present with different clinical features. And so this is actually one of the main questions our research is trying to answer through our Genes to Mental Health Consortium. So G2MH is a research initiative to understand the role of genomic variation in NPD. It’s funded by NIMH and NICHD. And as you can see, we have 14 institutions from seven countries across North America, Europe, and Africa that are working together to collect, share, and analyze large data sets.
Our institution is involved in G2MH Project 3. It’s mainly focused at Geisinger Healthcare System in Pennsylvania, University of Florida, University of Washington in Seattle, and WashU in St. Louis, and Emory in Atlanta. We also have affiliate members at Penn State, Rutgers, and the University of Alberta. And all of us are working on these individual genetic causes to understand their phenotypic variability and help us to better define what appropriate therapies may work better in the future by garnering this information.
Our study had three recruitment sites, one at Geisinger’s in Pennsylvania, one at the University of Washington in Seattle, and then one at Washington University in St. Louis. And really, what this study was trying to do is leverage these rare genetic etiologies to advance knowledge and hopefully one day treatment of neurodevelopmental and psychiatric disorders. So we actually leveraged genetic data that came from having clinical testing done. And when a family was identified to have a child with a genetic diagnosis, they were enrolled in our study for further analysis. So you can see that we enrolled close to 1,300 individuals as part of this process.
And then in addition to the data that was collected as part of their clinical care and diagnoses, we did project-specific phenotyping to do further assessments of NPD domains, look at medical history, clinical diagnoses, et cetera. We took a deeper dive into a few specific genetic etiologies to try and develop more detailed signatures. And then finally, we’re hoping that this work will lay the foundation for future research, where we may have specific risk algorithms to understand shared pathophysiology better. And as I said, be set up for clinical trials to hopefully understand these diseases better.
So from our studies, we’re trying to understand which individuals may have a higher risk for developing these conditions when they have a specific gene or which may be resilient from having more severe phenotypes and end up on a milder end of the spectrum. And we know that these types of rare genetic causes have large primary effects on neuronal pathways and other clinical phenotypes, as I’ve just described. But now we’re trying to understand what else can modulate this phenotypic variability or differences.
We’re looking at family background specifically, that includes both genetic and environmental factors. And we can now measure that using polygenic scores, which are just a way to look at common variation across our entire genome that may contribute to certain phenotypic effects. And so through our G2 project, we now have a cohort of individuals with this structured genotype and phenotype data and are starting to do analyses to understand the differences between the type and the magnitude of NPD risk in these individuals. Again, overall enabling a more precision health-based approach to hopefully identifying targeted interventions.
So what are we learning? This slide is showing the influence of family background on clinical variability of NPDs. So in the figure on the left, the blue curve is showing the normal distribution for a trait IQ, which is a measure of cognition. And the orange curve is the same curve but related to a child who has a particular genetic change. And so in Family A, the genetic change shifts the affected child’s IQ into the range of intellectual disability. In Family B, the child has the same genetic change as the child in Family A, but you can see that the starting point on the normal curve for that family is higher.
And so when that same genetic change occurs, the shift does not reach the defined threshold for intellectual disability and falls within the normal IQ range. So what we’re learning is that the impact of this genetic change is actually the same across families, but you get variable phenotype based on the starting point of the overall family background to begin with. And we’re certainly doing more studies to delve into this to see how we can use this to hopefully predict phenotypes and provide more targeted treatments.
So what I’ve been describing so far is the genotype — phenotype-based approach where you identify an individual who has a genetic change, and then we do genetic testing to identify –sorry, we do phenotype first, where the individual has an NPD diagnosis and we do genetic testing to identify the genetic change. And most studies to date have used that approach, where you investigate clinical or research cohorts who are ascertained for a particular phenotype, like intellectual disability, autism, schizophrenia, et cetera. A lot of these are also pediatric studies because those children who have the most severe phenotype are identified early in life and have analysis of their genomes.
But we need more data on the clinical consequences of genomic variants in general populations, really to get at the broader phenotypes and where we may have more mild manifestations. So now what we’ve done is use the population-based genotype-first ascertainment. So it’s not based on a phenotype that a person may have. We really just take non-genetic causes of NPD and screen large populations to identify individuals who have these treatments.
And so our study used a cohort called MyCode that is part of a biobank at Geisinger Health System. We evaluated over 90,000 patient participants, and we looked for the frequency of genomic changes that were in these individuals. We used curated lists that were done through another NIH-funded project, which I’m part of, the ClinGen project, to identify 31 known pathogenic copy number regions. So these were deletions or duplications, and then 94 high-confidence NPD genes. So again, we know that these genetic changes can cause NPD, and so we wanted to identify individuals who have those changes and then look at the clinical symptoms that they may have.
And we, in this study, have looked at the prevalence of these. So how many individuals have these changes, either copy number changes or single-gene changes? And we also looked at the personal utility of the individuals that we identified these changes in. So we wanted to learn how the participants reacted to learning that they carried some of these genetic changes.
So key takeaways from this population-based study were that recurrent pathogenic CNVs were observed in close to 1 percent of participants. So these are deletions or duplications. Importantly, only about 6 percent of participants were actually aware that they had this genetic change. And most of these were younger individuals because that’s where genetic testing is done most frequently. The NPD gene variants, so the single-gene disorders that we looked at, were identified in about 0.34 percent of participants. So, if we take those numbers together, we see that NPD variants are actually prevalent in the population, over about 1.1 percent, or one in 89 of our participants in our MyCode Biobank. And so, these results show that these genomic variants play an important contributory role in mental health disorders. And I will say that this prevalence is actually an extremely conservative and minimal estimate since we limit our analyses to a very conservative list of 31 pathogenic CNVs and 94 high-confidence single gene disorders.
So, let’s go back to what precision health can mean for brain disorders in this type of population-based work. So, one of the individuals in our study was a 48-year-old man who was found to have a pathogenic 22q11.2 deletion. He currently lives with his parents. He’s still single, but he did graduate from high school. He got a certificate in welding. He drives, and he independently manages his appointments and finances. When we met with this individual, we noted that he had a typical 22q11.2 facial appearance, but this can be subtle, but he didn’t have any other history of chronic medical conditions or surgeries that are often observed with this deletion. He did note one psychotic episode at 35, and with this particular deletion, about 25 percent of individuals will develop schizophrenia.
He required hospitalization at that time. After he stabilized, psychiatry attempted discontinuation of his antipsychotic medication, but he did have a recurrence of psychosis. And so, identifying this 22q deletion in this individual supported continued treatment, and he is currently stable on low-dose antipsychotic medication. And the quote at the bottom right is from this individual saying, “It feels good to know that there’s a medical name for my condition.”
So, in summary, the pace of discovering rare genetic causes of NPD is really surpassing our knowledge of associated phenotypes, and we have a lot to learn. But these types of detailed studies of NPD phenotypic signatures will be essential for precision medicine breakthroughs in both management and treatment of individuals who have these genetic changes. We’re taking a learning health system population-based approach to these studies so that we can do a lot of this work or learn a lot through routine clinical care. And our hope is that discoveries about rare genetic NPDs can inform care and may hold the key to broader treatment breakthroughs for all neurodevelopmental disorders.
And I’ll just end there by thanking our patient participants and the funding that we have.
MIRI GITIK: Thank you very much, Dr. Martin. Very nice presentation. Dr. Vaccarino?
FLORA VACCARINO: Hi. Good morning. Can you see my screen?
MIRI GITIK: Yes.
FLORA VACCARINO: Thank you. Okay. So, I’m going to outline my talk here. I’m going to start talking about the functional genomics in autism spectrum disorders, and then end up with a new method that we’ve developed in the lab for multi-regional organoid research. So, let me start by saying that, as people have already said in these talks that humans are not one single thing. One thing doesn’t fit all. We are a mosaic of germline and somatic variation. And also, the genetic risk for human disease rarely maps to a single gene. So, we need to study and investigate the convergence in biological mechanism. My talk will focus mostly on the biology of neuropsychiatric disorders. This has not been an easy thing because the developing brain presents many challenges, including its inaccessibility and the many challenges of using post-mortem human tissue.
So, in the last 15 years, and this is work that’s been basically performed mostly in my laboratory, but it’s been supported by the NIMH and particularly by the PsychENCODE Consortium. So, in the last 15 years, we have investigated using a particular technique called induced pluripotent stem cells, which is basically a way to make a somatic cell such as skin cells pluripotent, go back in time, and then become like a stem cell. And then we grow these cells into aggregates called organoids, as you see here. And let’s see if I can — okay, organoids. And we subjected organoids to a protocol in vitro here that you see they undergo a very lengthy differentiation process where they become neurons and glial cells. And we test them by various high throughput assays, including assays for their genome, for their transcripts, for their epigenetic changes, and others.
So, one important thing to say here in this context is that the reason why we have embraced this system is because not only the organoids and the pluripotent cells that they come from can sort of mimic brain development in vitro, but they do so in a personal way because they reflect the genetic background. The personal genetic background is maintained in these cells largely. So, what you’re seeing here is like a little snapshot of somebody’s retrospective brain development in a culture dish.
So, let me just illustrate you one recent set of studies that we’ve done with this system. This is an example of 14 families with idiopathic autism spectrum disorder. It’s called idiopathic simply because we haven’t identified a single gene. And this is often the case, as you have heard, with many neuropsychiatric disorders. So, 28 individuals, and we did single-cell RNA sequencing for transcriptomes, and then also epigenetic assays. And one thing to say also is that there is a phenotypic trait, which is a large brain, which is often present in about 15-20 percent of autism, which involves an accelerated rate of brain growth. And so, we took note of that, we measured head circumference, and we introduced that as a potential variable in our studies.
These are the organoids, we grow them in large batches, and they grow in size over time. If you section them, they have these interesting layers, progenitors layers like in pink here and neuron layers in green and various intermediate layers, so they do reflect the characteristics of brain growth and development to some extent. And they also develop neurons according to schedules, and here you see, for example, that in a cortical organoid, you know, lower layers in blue here are generated first at about one month, and then at about five months. Now you see not only the blue lower layer but also the pink or red upper layer. So the schedule is pretty much conserved and this is truly amazing that we can actually do this in a culture dish. And even later on, you have astrocytes and oligodendrocytes. It’s an example of astrocytes at about five months and a half of development.
So, when you analyze the RNA of single cells, the transcript in single cells in these organoids, you do get a panoramic view here of what brain development might be. And here there are three time points at the very beginning, in the middle, and towards later stages of neurogenesis. Each dot here is a single cell and each color represents a group of cells that share similarities among them in their transcript. So, at the bottom, you see what we call radial glial, those are progenitors, different types of progenitors, then according to this arrow, they develop into neurons, but not just one type, different types. You have excitatory neuron, inhibitory neuron, other types of neurons. And these are annotated according to what they expressed in vitro. So, for example, here, excitatory neurons of the preplate mimic very early stages of neurogenesis, and excitatory neurons of the dorsal cortical plate mimic excitatory neurons that actually are typical of the cortex.
So, one interesting aspect of this work is that it actually reflects the variability of individuals that are donors of these pluripotent cells. And here you see this on the left, by color, you see a representation of the different cell types at the three time points. And the colors reflect the type of cells, but each line is one individual library. So, for example, at the top, you see the TD0, even though they’re all beginning development, the percentage of certain progenitor cells actually is quite different among various individuals. And here at the bottom, you see, for example, the excitatory neuron in blue, they’re there, but it’s not exactly the same percentage in every one, okay?
And so, we have spent quite a lot of time to understand what the source of this variation is. And we came to the conclusion after many different correlations with various variables that may impact these cellular models, that actually a large part of the variation is due to the genetic background of the cells. And one of the reasons is that the percentage of, for example, here you see excitatory or inhibitory neurons is actually highly correlated. And you see the correlation here with the transcription factors that are present in the progenitor cells. So, it’s not that the cells randomly variate, but the genetic programs that generate these cells are quite different. To some extent, they differ in different individuals.
So, when we compare autism patients versus their father, and this is — we do this comparison within families, so we analyze these patients in pairs to try to diminish the effect of different genetic background. We saw that macrocephalic ASD in yellow and normocephalic ASD were different with respect to their controls. So, the macrocephalic ASD had an increase in excitatory neurons and a decrease in inhibitory neurons, and the normocephalic ASD had a fairly different phenotype, actually quite the opposite. So, in this case, you see, for example, at the top, excitatory neurons are decreased in normocephalic ASD. And this is reflected in the circumferential difference of these patients.
So, I’m showing you to emphasize the fact that when you start analyzing individual genetic background, you do see that although these individuals have been all labeled as autistic with autism, their biology actually seems to be quite different. And that correlates with the size of the brain. And here is just another slide to show you that there is a fairly high intersection, which is statistically significant, between these transcriptomic changes and genetic risk genes for autism spectrum disorder.
So another aspect of the work that you can do using these models is understanding what’s upstream, meaning what is causing these changes in gene expression. And by doing epigenomic assay by a technique called CHIP-seq, we’ve actually taken a subset of this family and identified what’s called an enhancer element. And these are regulatory elements that sit upstream of genes or downstream of genes and regulate gene expression, trying to understand, get an answer for why there was this differential gene expression across these families.
So we identified gene-length enhancers, and then we correlated the enhancers to gene expression to build what we call a gene regulatory network, which is, okay, there is a gene that is dysregulated, what’s upstream, what’s causing this dysregulation in particular individuals. And when you make that network, you basically get something like this, which is beautiful to see, but very difficult to interpret. So each dot here, each circle, is a transcription factor, and each little dot is a gene. And in blue, you see the repression, and in yellow, you see the activation of the transcription factor for those genes.
But then when you look more in depth into this network, you see things like that. You can actually deconvolute what’s upstream of certain important genes in development. And the reason I’m saying that is that this would allow us to see actually, potentially, to understand whether there are mutations in these regulatory elements that can impact the expression of those genes. So, for example, one way to use these regulomes is to understand the transcription factors that drive differences in genes across time. So this is TD0 versus TD30 and TD60. And you see you can look at the genes that are downstream of the transcription factor and see how many of those are differentially expressed, and so classify these as more or less impactful.
You can look at transcription factors that drive cell type specification by intersecting with single-cell RNA sequencing. And you can look at transcription factors that drive differential gene expression in autism spectrum disorder. And here, you see, for example, on the left, many transcription factors that are excitatory neuron drivers, they drive the development of excitatory neurons. You can see they form like a hub. They kind of cross-relate to each other through enhancers, reciprocal enhancers. And so that might explain why this subset of patients had an increase in excitatory cells.
And we also validated the regulome using different techniques. I’m not going to go into that. But basically, just to show you one way to use this data is to understand this type of convergence. For example, if there is a gene that’s differentially expressed, you could have many different types of changes in the DNA, including enhancers, promoters, polymerases, mediators, other factors that can impact the transcription of that very same gene. So many different genetic changes may have similar effects.
And this is an example of it, where we, as part of another consortium, have been part of the Brain Somatic Mosaicism Consortium. We looked at mosaic mutations in enhancers, both in organoids and fetal brain. And we saw that people with autism had an increase in enrichment in both enhancers from organoids and enhancers from brain in transcription in mutations that were disrupting certain transcription factor binding motifs, in this case, the MEIS1 transcription factors.
So in conclusion, there is an enormous variability in differentiation programs. We think this is very interesting, very real individual susceptibilities, perhaps, and the gene regulatory network can be used as a way to define transcription factors and genes that may be susceptible to cause changes in development and potentially predisposition to disorders. So in the last few minutes, let me just go over, if I have a few minutes, I hope — yeah. Let me just go over a new method that we developed very recently in the lab to make a little bit better the organoid model, because these are just beginning techniques. They’re really bound to be improved over time. So we reason that during brain development, you actually don’t have an issue culture dish with pluripotent cells. You have a neural tube, developing neural tube, and you have things called morphogens, which are factors that are present in the milieu here in gradients.
And one of the very first important decisions that the CNS has to make is patterning, meaning, what’s anterior, what’s posterior, what’s ventral, and what’s dorsal. So we try to mimic that using a special chamber, which you see here on the left, and putting a WNT agonist, which is mimicking a factor present in the developing brain that posteriorizes the brain, and another factor, Sonic Hedgehog, that ventralizes the brain.
When we did that through gradients in this chamber and put organoids in the chamber, and then we took the organoids out after nine days only of patterning, we actually were able to look almost at the snapshot of the whole brain. So we looked, for example, for various transcription factors and genes, characteristic of certain regions. And you can see that here is the chamber. C1 is the most posterior. C5 is the most anterior. On the left, you see the dorsal half. And on the right, you see the ventral half. You can see that forebrain is clearly here. Forebrain genes are expressed in the anterior. And for example, homeobox genes, posterior genes, are expressed in C1. On the most posterior chamber, you see differentiation of genes that are in the cortex, in the dorsal anterior side of the chamber, and genes for the ganglionic eminences in the ventral side.
So basically, and I don’t want to spend a lot of time here, when we look at single cells several weeks later, we can see that this system reproduces pretty much the entire brain, including the cortex, the thalamus, the basal ganglia, the cerebellum, and various portions of the cerebellum, including cerebellar granule cells, astroglia, ventral midbrain progenitors, hypothalamus, and this is also true by immunocytochemistry. And you see here on top various genes that are typical, typically present, for example, here in blue, the cortex, NEUROD6, and in red, basal ganglia in red, expressing the LHX genes.
So we use the datasets that are now available from the human fetal brain to validate these cell types as truly representative of these regions I’m talking about, and this is validation that you see here. This is the fetal brain, and so thalamus, pons, cerebellum, hypothalamus, cortex. And the most interesting thing is that we saw the similar variability as the one I was describing before with cortical organoids, except that this variability now was basically describable as a differential response to the gradient.
So, for example, here you see in red, this is basal ganglia, and you can see that according to the position in the chamber, you see different proportions of basal ganglia neurons. So clearly, one source of the variability is the particular step of a gradient, how much WNT, how much Sonic Hedgehog you have. And here you see, for example, thalamus, it’s present when there is a lot of WNT, so it’s fairly posterior, but the dopaminergic neurons are present, not only they are posterior, but they are also ventral, so they need Sonic Hedgehog. So that’s one source of variability, and we defined the source of variability according to gene expression. And then there is an interpersonal variability in which each individual has his own way to respond, and you see this in different colors here, various individual IPS lines respond similarly but also differently with a different slope to the same gradients.
And so at the end, I’m going to conclude by saying that according to this new model that we’ve developed, only five days of patterning with these two gradients is sufficient to generate most brain lineages. And we have early spatial patterns of gene expression that anticipate lineage specification as well as additional secondary gradients of factors, and that we can see an inter-individual variation in the response to these patterns. And I think I’m just going to end by saying that this is a promising field in our view. We’ve worked on this for a long time, but I think it needs, of course, innovation. And it can give us — beginning to give us, a glimpse into human development. And potentially understand the biological impact of severe drug mutations, not only the penetrant mutations, but also the most common mutations that might impact disorders.
And I wanted to, of course, thank my lab, collaborating labs, and especially NIMH and the PsychENCODE Consortium we are part of, as well as other foundations that have funded our work. Thank you.
MIRI GITIK: Thank you very much, Dr. Vaccarino. Dr. McMahon, please, when you are ready. And if you have any questions, please, you can add them to the Q&A. You do not need to wait until the Q&A session. Thank you.
FRANCIS McMAHON: Good afternoon, everyone. I’m Francis McMahon from the Intramural Research Program of the NIMH. And today, I’d like to talk with you about the ways in which genetics can be put to work in improving care for mental health disorders. Let’s see if I can [unintelligible] my slide here. There we go. You’ve heard about some fantastic research today, and I think it’s giving you a sense of how quickly the field of psychiatric genetics and genomics is moving forward. But if you treat patients with mental health disorders or live with a mental health disorder yourself, you may wonder how relevant this work is for the daily treatment, the long-term prospects, and the understanding at the clinical level of mental health problems.
Now, genetics could be potentially of great use in clinical psychiatry. As you’ve heard a bit in Dr. Sullivan’s talk today, genetics might help with diagnosing mental health problems. Through genetics, psychiatrists are finding ways to break down broad groups of disorders into smaller, perhaps more biologically meaningful subgroups. And as you heard in Dr. Martin’s talk today, in some cases, it’s actually possible to make a genetic diagnosis and improve the quality of care for patients on the basis of identifying a genetic cause. Now, genetics should also help identify people who are at very high risk of developing a mental health disorder later in life. These people may benefit from efforts aimed at preventing the onset of illness or symptoms, what we call primary prevention, in the same way that we can now identify individuals who are at high risk, say, for a heart attack and treat high cholesterol or high blood pressure in order to prevent that heart attack or reduce its risk in the future.
Now, some kinds of variation at the genetic level may also help us to predict who will respond best to particular treatments and who’s more likely to experience side effects. While all of these uses may still be a ways off, today I would like to give you a feel for how far we’ve come and how far we still have to go in putting genetics to work in psychiatric care. I’ll start by talking about some of the methods and findings over the past 10 years, much of which we’ve already heard in the earlier talks today. I’ll highlight, in particular, some of the clinically relevant findings that have come out of the last 10 years of research. I’ll try to give you a sense of how genetic risk influences the dimensions of variation in psychopathology, and the ways in which clinical presentations are influenced by genetics. And I’ll talk a little bit about the implications for psychiatric care and practice in the near future.
As you’ve heard today, there’s been substantial recent progress in psychiatric genetics. Genome-wide association studies have revealed to us the great numbers of genetic variants that contribute to risk and the complexity of the genetic background for every psychiatric disorder. We’ve also learned that structural variation in the genome, known as copy numb variants, while rare, can have a major role in neurodevelopment and in the onset of psychiatric disorders later in life. We’ve also learned that de novo mutations, which are single base pair changes in DNA that arise as the genetic material is passed from parents to offspring and can affect the function of genes, also play an important role, particularly in neurodevelopmental disorders and in schizophrenia. We’re also beginning to learn through whole exome sequencing studies how other very rare variants can confer substantial risk for mental health disorders.
Now, one of the questions that may be in your mind after hearing about all the different kinds of genetic variation that can play a role in psychiatric illness is how we put all this together in understanding an individual’s risk. One of the major tools we use for this is called the polygenic risk score. And the idea here is that risk alleles or genetic variants that confer risks for disease later in life are one measurable component of the liability or risk for disease. And that liability is distributed in a bell curve across the population. Now most of us carry at least some risk variants. And most of these confer only a small risk for disease. But as the number of risk variants accumulates in each of our genomes, so does the risk for disease rise. And that’s shown here in this blue area to the right of the bell curve.
Individuals who have a low dosage of risk alleles may have a lower than average risk of illness, whereas those who have higher risk of genetic variants that confer risk have a higher risk for disease. Polygenic scores just allow us to sum the effects of these risk alleles and to weight them by the individual risks they confer. What we’ve learned so far through polygenic risk studies is that there are strong genetic overlaps in most — among most mental health disorders, as you heard during Dr. Sullivan’s [spelled phonetically] talk. And while this is an important insight, it also means that polygenic risk scores probably cannot be used to distinguish between individual diagnoses, for example, to distinguish a case of bipolar disorder from a case of schizophrenia. For that we’ll need additional information that may vary at the level of individual families or patients.
Now you’ve heard a little bit about copy-number variants as well. These are small deletions and duplications in the genome that typically span about 500 to 5 million base pairs and can often involve several genes. Each of these copy-number variants tends to be rare, but they’re more common in people with mental health disorders, five to ten times more common in fact. And when all of these copy-number variants are taken together they may play a substantial role in risk, particularly for neurodevelopmental disorders, schizophrenia, and mood disorders. And one thing we’ve learned about these copy-number variants is that the genes that they affect tend to play important roles in the development of the brain, with particular enrichment in neurons, the brain cells that are involved in communicating information across the brain, and in synapses, the places in which brain cells talk to one another.
Now I mentioned de novo mutations as another important source of genetic variation that can predispose to disease. By using very high quality, high accuracy sequencing studies, scientists have discovered that several times each time that when the genetic material is passed from father and mother to the offspring a genetic change may occur by chance. This happens three or four times throughout the genome for each individual. When one of those mutations happens to damage a gene that’s important in brain development, then that genetic change can predispose to later mental health disorders. De novo mutations seem to play a particularly strong role in early onset neurodevelopmental disorders such as autism spectrum disorder and intellectual disability, but probably also contribute to disorders that present later in live such as schizophrenia and mood disorders.
Now de novo mutations account for a small proportion of all the cases of these illnesses in the population. But because they tend to hit a single gene they give us unique information about the biological contribution of that single gene to that individual’s illness. And since the genes that often are hit by de novo mutations are hit again and again across the population, we can begin to put together a strong case of causality for particular genes.
You heard in one of the earlier talks today about high-confidence genes that confer risk for neurodevelopmental disorders. A lot of that confidence is based on the fact that these rare events have been observed again and again to affect those genes in individuals who have particular neurodevelopmental or psychiatric disorders. So de novo mutations are a very important window into the biology of mental illness.
Now another area in which genetics impact psychiatry is through pharmacogenetics. Now over the last ten years pharmacogenetics has promised to deliver a world of precision psychiatry where it will be much easier to match patients with the safest and most effective medication. This promise remains unfulfilled so far, partly because genetic variants associated with differences in treatment response or side effects can be very difficult to locate within large genomic variation that occurs across the population of patients receiving treatment. Now there are some places where we have made some advances. Genetic testing of variants associated with drug metabolism, for example, may play a role in some severe side effects and can be helpful in patients in special situations. So far though there’s evidence for clinical utility in only a minority of situations, such as individuals who have particularly unusual responses to antidepressants or antipsychotics.
Another unusual situation that comes up, but can be very important for individuals who are at risk, are genetic markers in the HLA region that have been associated with severe skin reactions when patients are prescribed particular medications and particular the mood stabilizer or anti-epileptic drug carbamazepine. Although rare, these can be life-threatening events and it’s important to be able to know in advance if someone is at increased risk for one of these conditions. We now have genetic tests that can be used in individuals who on the basis of their ancestry are likely to have this genetic risk so they can be protected against this severe adverse event through treatment.
Now there are also various commercial panels that are being made available to psychiatrists, physicians, and sometimes directly to patients, with the claim that they can be used to predict or select the most efficacious and safest psychiatric medication. While some of these look promising, so far they have not established that they can actually change the incidents of side effects or the rates of good response to psychiatric illness in patients who are tested with them. Now I don’t think that this is the final word on this issue. And as we have larger studies of more diverse populations, I think we’ll begin to be able to crystalize our views of the ways in which pharmacogenetic tests can be used to improve clinical psychiatric, both in the selection of medications and in the avoidance of adverse side effects.
In Dr. Vaccarino’s talk you heard a lot about the great advances in stem cell biology over the last several years, and the ways in which we can study brain cells in induced pluripotent stem cells that are derived from cells donated by patients. So far, most of this work has focused on selected targets such as neuropsychiatric copy-number variants, rare genetic mutations that affect the function of brain-relevant genes, and also individuals whose illness has been resistant to typical treatments for their psychiatric illness. These studies provide a unique window into gene expression, the shape and diversity of brain cells, and of the way in which neuronal function is affected both by genetic variation and by treatment. And they’re beginning to reveal ways in which genetic variants also affect the development of the brain, as you heard in Dr. Vaccarino’s talk.
Someday these kinds of tests may provide a way for us to discover new medications that could prove promising treatments or maybe even cures for mental health disorders. It may also be possible someday to try out drugs in the laboratory before trialing them in patients in order to improve the proportion of drugs that make it through to those crucial clinical that allow us to discover the safest and most effective medications.
So I want to leave you with a couple of take-home points today. First of all, many genes modify risk for psychiatric disorders. And there is a substantial overlap across diagnostic groups. This tells us that polygenic scores are a good research tool and allow us to understand the collective effects of many genetic variants, but so far have been difficult to use in the clinical setting because of that same genetic overlap. We also now know that particular high-risk variants, such as copy-number variants, can have a very important role and may help explain illness in about 5 percent of patients with mental health disorders. This is useful in those individuals who carry such variants, but may also shed light on the biology of mental — of mental illness in other individuals since, although the genetic variation may vary from individual to individual, the biology is probably more general.
And finally, there are pharmacogenetic tests available that look quite promising. Existing tests are probably not useful for most patients, but in particular situations with particular drugs, can be very important, particularly for avoiding severe adverse events. Stem cells may provide a powerful new tool in which we can do pharmacogenetics in a dish and understand the ways in which genetic variation interacts with drug metabolism and response, to allow us to come up with more precision treatments for patients in the future.
I want to close by acknowledging our continuous research funding from the Internal Research Program that allows us to carry out our research over the last 20 years, and to say that many of the ideas included here have been previously published in an article in the American Journal of Psychiatry in 2022. Thanks very much.
MIRI GITIK: Thank you very much, Frances [spelled phonetically]. And thank you very much to all the speakers. The Q and A box is open. You can go ahead and answer your questions. Just to start us out while people are putting in the questions, a more general question to all four speakers, how do you see psychiatric genomics developing? What do you think will be the next big things in the 5-10 years to come?
CHRISTA MARTIN: I can jump in first [laughs]. So, I mean, as I alluded to in my talk, I think we’ll use genetics as more of a clue to how we might differentiate treatment for individuals. And by treatment I don’t always mean a pharmaceutical treatment. It could be behavioral interventions. It could be speech and language interventions. But I think we’ll be able to get clues about what the different genes or deletions or duplications, how they impact individuals, and hopefully allow us to do more targeted therapy rather than treating everybody who has a diagnosis like autism the same when we know at the biological level that they are different.
FRANCIS McMAHON: I’ll add to Dr. Martin’s answer, which I agree with, that genetics will also, I hope, begin to give us a tool for understanding the way in which nongenetic influence affect risk. If we can define more genetically homogeneous groups of patients, or people who are at risk for illness, we’ll then be able to look at the ways in which environmental, life experience, or other nongenetic events influence that risk, both in terms of resilience and in terms of additive risk. And so, although we’ve talked about genetics a lot today, we recognize that they never are the whole answer in an individual’s illness. In order to get a better idea of how each person’s illness develops and how potentially it could be prevented, we need to have more powerful ways to study nongenetic effects.
FLORA VACCARINO: Yeah. I might add that the convergence between nongenetic and genetic effect is function. So I think the ultimate goal of course is to understand how brain dysfunction arises. And of course my bias is very early in development, but it could be any time during a person’s life cycle. And there are such things as somatic mutation that happen even after a person is conceived, throughout development and adulthood. So basically I think we need to really make an effort to understand how all kinds of variation, genetic, epigenetic, nongenetic, affect brain function. And so, brain biology. And that’s the difficulty of it because of, you know, brain is not a simple system, so [laughs]. Yeah. But that’s — yeah.
PATRICK SULLIVAN: It’s interesting how much we agree. I think I have — my esteemed colleagues certainly have made points that I absolutely agree with 100 percent. To that I would add that discovery is not over. We’ve — we’re — in a sense we’re only scratching the surface. There’s a — there’s a whole bunch of very large studies with more detailed looks at a lot of different things that are on the way. And I expect that over the next two or three years we’ll have learned a ton more.
One of the things that I think is super important is, especially as we start to look to see how these conditions might be treated, we have to be sure that the people we’re using to generate those discoveries represent humanity and not just one segment of humanity. And that’s certainly a — that — there’s a bunch of big efforts happening in that sphere too. So I’m really excited, but I think whatever we do, whatever we — whichever directions this takes us, it has to be in team science context with clinicians, with geneticists, with statisticians, and without question neurobiologists as well.
MIRI GITIK: Yeah. Great points. Thank you very much. So we have one question about the role of NDA or the NIMH data archive and other NIH-led archives in neuroscience. Anybody would like to comment on that? What role do they play? Or what do you see — how do you see them in the future being used?
FLORA VACCARINO: Well, I mean NDA is valuable in the sense that allows to protect information that is — could be private because it derives from patients and other individuals. So it allows for that information to be properly protected from misuse and abuse and things like — so we have to be very cognizant of that these days when a lot of genomic information is collected from varieties of individuals.
PATRICK SULLIVAN: All right. A question — I wrote something to Ben Hilliard about — he’s a frontline behavioral health provider working in a pediatric integrated behavioral healthcare system. Are there physical presentations that would indicate genetic disorders? I was going to ask the same question. I think there’s an urgent need for clinicians to say, right, “If you have these features, you know, get a relatively straightforward genetic test,” because there might be a chance that this person actually has a syndrome. So, I’m sorry. Christa, can I ask you to comment on that, please?
CHRISTA MARTIN: Yeah. Yeah, no. I mean certainly there are some clues. You know, we tend to think of individuals historically with genetic causes as being on more the severe end of the spectrum, which would mean that they have a constellation of features that they present with. So it’s not only the neurological presentation but perhaps other congenital anomalies like heart defects or kidney defects or things like that.
But I would say what we’re learning is that that is the severe end of the spectrum. And we know there are individuals who have — may have a behavioral diagnosis or autism, for example, and don’t have any of those other features, who now we know have a genetic cause. So I think the more we learn the more complicated maybe it gets. It’s not a, you know, look for the most severe presentations. And in some cases it’s really important to understand the implication to the family. So, for example, the last — you know, when I said there’s — could be a child just with autism, there’s a common duplication on chromosome 15 that can be inherited through families and they can have a 50 percent chance of having another child with neurodevelopmental phenotypes. And so, getting that information to that family is really important.
So I tell people now if you have any diagnosis that is, you know, considered — so autism, intellectual disability, now cerebral palsy is known to fit into that, you know, broader we think about schizophrenia and bipolar disorder. I think we’re moving to doing genetic testing in those individuals. But certainly the suspicion is higher in ones that may have those complex disorders, or the yield is higher in the ones that have more severe presentations. Did that help, Pat? Okay.
PATRICK SULLIVAN: Excellent. Thank you.
MIRI GITIK: Another question to anyone on the panel. Comments about epigenetics on [unintelligible]?
FLORA VACCARINO: Well, I can say one thing about that. I mean it’s very incompletely understood. These are the regulatory elements that I was mentioning in my talk, in [unintelligible] for example, that drive changes in DNA conformation and gene expression, and probably other changes that we don’t know. Methylation is another. So one thing is that the environment that we talked about as a potential causal factor in a variety of disorders including neuropsychiatric disorders can act through epigenetic changes in DNA because the epigenome is susceptible to change over life in response to a variety of conditions, both internal and external. So that’s an interesting thing that we should investigate. And there are now proper techniques these days that allow us to — I talked about ChIP-seq, but there are others that allow us to actually map the epigenome together with the genome so we can see, for example, whether there are mutations that affect dysregulatory elements in addition to other changes.
MIRI GITIK: And there’s a more specific question about CIPIC [sic] guidelines for antidepressants that came out last year and provide [unintelligible] guidelines.
FRANCIS McMAHON: I’d be happy to field that question. And I think this is a great opportunity to point out that there are now several international organizations that are reviewing the available pharmacogenetic evidence and providing unbiased guidance on how physicians and patients can use genetic testing in the selection of the safest and most effective treatments. So I strongly recommend if you’re a psychiatrist who’s potentially interested in incorporating genetic testing before deciding on antidepressants for your patients that you consult guidelines such as the CIPIC guidelines for good information on the best way to do that, and on the patients who are most like to benefit from that kind of testing.
PATRICK SULLIVAN: Francis, do you — question if I could to follow up. Do you think that should be done routinely for somebody coming to — being considered for antidepressant for the first time or is its role when you get into the treatment resistant setting when a person doesn’t respond to multiple medications?
FRANCIS McMAHON: I think you raise an important point, Pat. And so far I think the evidence from clinical trials is unclear on this point. But if there is likely to be benefit, it’s most likely to come in the unusual patient who has had an unusual reaction to an antidepressant or perhaps multiple antidepressants, or whose illness has failed to respond to multiple antidepressants. Because some of those individuals may actually have genetically-mediated differences in drug metabolism that may explain why they have failed to respond well to the treatments that have been used so far. And that information can be helpful in the selection of a different class of antidepressants or an adjustment of dosages for future treatments.
MIRI GITIK: Another question regarding the use of genetics along with behavioral measures to reclassify the standard DMS diagnosis. Any thoughts about that?
PATRICK SULLIVAN: Perhaps that one’s for me because this is kind of a PGC thing in a sense. It’s a great question. And there’s a bunch of things going on exactly along those lines. As I sort of indicated in my talk, especially when we get out to the sharper end of the spectrum the people whose lives are more impaired by having a chronic severe and enduring psychiatric disorder, we tend to find some more similarities and differences. And so, there’s a bunch of studies that are — that are trying to come at this squarely, trying to understand exactly what’s going on. I think most of it really is trying to look at outcomes, to be honest, as classifiers rather than looking at behavioral features. But what you suggest is certainly something that would be quite relevant as well.
MIRI GITIK: Thank you. Any last questions to our speakers from anybody?
FRANCIS McMAHON: If I could just expand on Dr. Sullivan’s answer to that last question, I think it’s also remarkable, and this was a point that Dr. Martin made in her talk, how the same genetic changes can lead to such a broad range of psychiatric and behavioral outcomes. So that’s telling us that most of what we use in classifying psychiatric disorders and in making psychiatric diagnoses does not map cleanly onto particular genetic changes. I think there’s a critical clue there that might help us understand what would map more cleanly onto genetic differences that’ll allow us to use genetic information to do meaningful neurobiological subgroupings. I think differences in outcome may very well be one important variation that way. But I suspect that many of these differences either have not been appreciated or cannot yet be properly measured in patients. And we need something, you know, equivalent of a coronary angiogram or a liver biopsy, or something that can really tell us molecularly about what’s going on in a patient’s brain. Some of the more advanced imaging techniques that are now being applied in large-scale studies might lead us toward that way eventually, although we’re clearly not there yet.
MIRI GITIK: Another question. Wait. I guess this is from a young researcher. I’m very interested in earlier identification of comorbid medical and mental health disorders. Can you speak on how genetics could play a role in this?
CHRISTA MARTIN: I can — I can start. One of the things we’ve talked about is, you know, there are a lot of groups talking about newborn genomic screening — or sequencing, sorry. And, you know, right now diagnoses have to show up before we can appreciate them. But if we — as we learn more about some of these genetic conditions that include medical and mental health disorders, you could then monitor patients to see — or use those as clues for what may develop and be able to identify any deficiencies earlier. With earlier identification, you would hope you could if there are treatments available, start treatments earlier. So I think you could use the underlying genetic etiologies that we know about as clues if you were using something like newborn sequencing, which again there are a lot of groups starting to actually have studies that are evaluating what would that look like if every newborn had their genome sequenced at birth. So I think that’s one that you could use that information.
FRANCIS McMAHON: Now I’ll add to that it’s underappreciated how common psychiatric comorbidity is with medical illnesses. Over 30 percent of people with a chronic medical condition of any kind will develop depression or an anxiety disorder. And yet only a small proportion of those individuals have that depression or anxiety disorder diagnosed or treated. So I think what this question also underscores is the importance for us within the medical field to understand the full range of presentations of medical illnesses in patients and to be attuned to disturbances in the psychiatric realm, particularly in mood and anxiety, that may accompany a broad range of illnesses, particularly those that have a chronic presentation.
MIRI GITIK: Thank you. So this is a more general question, I guess. Can we move away from language that separates medical from psychiatric?
CHRISTA MARTIN: That’s a —
FRANCIS McMAHON: Howard [spelled phonetically], I think we can but we don’t know how to do it yet.
CHRISTA MARTIN: That’s —
FRANCIS McMAHON: And right now the medical and — or physical illness and psychiatric illness distinctions that we use serve in allowing us to define [unintelligible] in our discussion and to use termination that we understand when specialists, psychiatrists, or patients talk among themselves. But we shouldn’t forget, and I think your question is meant to point this out, that ultimately there really isn’t a distinction. The brain is part of our biology, that psychiatric or mental health issues disorders arise from the pain and thus they fundamentally share the same biology of the heart, the liver, and the rest of the body.
FLORA VACCARINO: Yeah. And if I can add to that, that termination exists because of our ignorance of the brain function. And once we’ll understand that then we will be able to describe the brain in biological terms or terms that are similar to the understanding of the function of the heart, for example, which is based on fluidics, right? The brain is based on electricity, but also on other things. It’s very complex. So that’s been a challenge for all of us, continues to be a challenge.
MIRI GITIK: Yeah. Thank you very much. Jonathan [spelled phonetically]?
AUDIENCE MEMBER: I wonder if the panelists have any thoughts about different categories of variation that might become more important to us in the future? For example, understanding rare variation to an extent that we don’t yet, or understanding when we sequence a whole genome, or perhaps what long-read sequencing through Telomere-to-Telomere project might show us about parts of the genome that haven’t been easy to see up until now. Or it could be anything, even circular RNA or any kind of phenomenon that represents a kind of variation that we have not been focusing on.
CHRISTA MARTIN: I definitely think from my pie chart there’s still a big chunk that we have to learn about, right? So I mean our — just with exome and now genome sequencing specifically coming on board and being used more for research and now even clinical testing, that 40 percent is certainly growing to the rare causes. But the other side, the 60 percent, I think Jonathan, what you said, that’s where we’re going to find, you know, non-coding elements, things in promoters, things in RNA, epigenetics, like that will allow us to start understanding more and more about these brain conditions, probably things we haven’t even appreciated yet will be found that we don’t know about [laughs].
PATRICK SULLIVAN: I like that question because the genome is such a huge and complicated place. You know, there’s — as you’re aware, there’s hundreds of mega-bases, hundreds of millions of bases that are found in some people and not in other people. And the more we get into this the more we study, you know, the increased diversity of humanity, the more we’ll find. And I think that’s — it’s a hell of a question. I really want to see what — I’d like to know what the answer is actually.
[laughter]
You know, the technologies we have get us a long way toward that end, but every time we think we understand something and have narrowed things down something else pops up which tells us that we were a little short of the — our knowledge was short of the reality. So yeah, I think it’s certainly something worth looking at. I hope we do. And I’d love to know the answer.
CHRISTA MARTIN: When I — there’s another question in the chat about the difference between sort of looking at polygenic risk scores versus copy-number variants in single gene. I mean that’s a perfect example of sometimes the combination of things also gives us more clarity. So all of these different pieces that we’re talking about might not stand individually, and likely won’t stand individually. It’s more a combination of a person’s entire genome and all of these different changes that make us who we are or what conditions we might have.
FRANCIS McMAHON: I agree. And it’s also worth remembering that polygenic risk scores so far are still a pretty rudimentary tool. Although the math and the statistics are well worked out, they’re calculated in a way that’s ignorant of the underlying biology. And as we get better insight into the ways in which genes interact in networks it will become possible I think for us to come up with polygenic risk scores that describe individual neurobiological networks. And that may actually have a lot more individual power to support diagnoses, to define risk, and maybe even to select more appropriate treatments.
MIRI GITIK: So we have a question which I think is a bit specific. I’ll make it a bit more general. How do you address comorbidity or what is your approach to addressing comorbidity in genetics, specifically psychiatric and medical comorbidities? How can you distinguish between these?
FRANCIS McMAHON: I’ll take a short at that. And I think it’s a really important point. I made the observation earlier that psychiatric comorbidity is commonplace in people with chronic medical conditions. We also know that medical comorbidity is almost universal in people with chronic psychiatric conditions. And if our goal is for patients to live longer, healthier lives, we need to address both ends of that equation. We need to help psychiatric patients who may not have access to good medical care get that and support what’s often a difficult road of addressing and treating chronic underlying medical conditions. Because we know that chronic underlying medical conditions are often the reason that, for example, an antidepressant trial fails or that antipsychotics are no well tolerated by a patient, or that a child whose behavioral problems have necessitated the use of drugs is unable to use the drugs without being too sleepy in class during the day. So understanding how the psychiatric and the medical interact in every patient I think is going to be really important.
PATRICK SULLIVAN: If I could add to what Francis just said, I think it’s a really interesting question. And to some extent there are methods that actually can be used to think about this and to address it in a quantitative way. The background of this is way beyond this conversation because that would be a — pretty much a full lecture of itself. But there’s a technique called Mendelian randomization where you can actually ask and get an answer, at least tentatively, to that question. You know, if this — if people have both major depression and coronary artery disease, which is a common complication for sure, is that due to the depression causing it or the other way around? That question can be asked and at least tentatively addressed. And that’s neat because it’s one of the few ways that we have to get around this recurrent chicken or egg problem. Which comes first? Is it this or is it that?
Many questions in psychiatry are on that exact topic. You know, does heavy alcohol use cause depression or is it the other way around? You know, is anorexia — cannabis and schizophrenia would be another example. And so, there’s a bunch of elegant things you can do to actually say if we’ve got a risk factor, what’s its relationship to the outcome? Is it a cause or is it an effect?
MIRI GITIK: Thank you. [inaudible]
PATRICK SULLIVAN: Can we take the Dr. Google effect question? I think it’s a good one.
MIRI GITIK: Yeah. [unintelligible] answer that. I often have to address the Dr. Google effect. Could you point me to some parent resources that are readable by the general public and are accurate?
CHRISTA MARTIN: I can certainly point a few out. There’s — but I would say they — a lot of these come — that I’m familiar with but have good resources come from the genetic perspective, so not from the brain disorder perspective. But if you have a child with genetic condition or if you’re seeking information about this, but Unique is one that talks about different genetic conditions. They largely started around copy-number variants but are now broader. Combined Brain is a group that has brought a lot of different family support groups together that are dealing with neurodevelopmental conditions under one large umbrella, but they’re connected for resources and for building resources. And then the Simons Foundation has done a lot in this space. And they have a program called Searchlight which is a research study trying to evaluate individuals with known genetic conditions. But they also put out a lot of helpful basic information, but it’s well-vetted for families. I’m sure there’s others. I don’t know if other people want to add anything. But they’re out there for sure [laughs].
FRANCIS McMAHON: Answer highlights, Christa, that there’s a lot of stuff out there on the internet that’s not good quality. And so, it’s really important that we allow patients and family members and parents the kind of insight they need to be informed consumers of the information that’s out there on the internet when it comes particularly to mental health disorders. Sometimes there are things out there that medicine hasn’t yet realized is true. And so, ways in which we can bring patients and families together who say share the same disorder or share a common genetic change can actually allow us to gain from the insight that only patients and family members can have about the fine points and particular features of an illness. But it’s hard to sort out the wheat from the chaff if you sit down and do a Google search.
PATRICK SULLIVAN: I was going to suggest is this something that the NIMH has looked at to become — and maybe there’s some version out there that I’m not aware of.
MIRI GITIK: Yeah. So the NIMH has information on their website. This is credible information, readable information, for most disorders of interest. So I would suggest if you would like to start there. And there’s also a linking and citation of more complex, you know, more in depth papers relevant to the subjects. Jonathan?
MALE SPEAKER: I would add that the National Library of Medicin through NCBI, the National Center for Biotechnology Information, has a wealth of information that is I think very helpful across a wide range of diseases beyond NIMH resources.
PATRICK SULLIVAN: Can I answer the qualitative research? One of my colleagues, Cynthia Bulick [spelled phonetically] has done a study which is on its way to publication where they selected women with anorexia who were unknown to themselves as well as to the reviewer, who were selected at the very high end — for very high pyogenic risk scores for anorexia and very low scores. And then they did a lengthy qualitative interview with each. The reviewers actually guessed badly as — and patients guessed badly as to which group they were in. So what — I don’t think it was biased in that regard. And the results were fascinating. And what a lot of it came down to was that women with high PRS tended — strongly tended to find being fasted as a positive experience. For most of us when we don’t eat enough we get grumpy, hangry, whatever. And for these individuals it was quite the opposite. It was quite reinforcing to be in negative energy balance. So yes, I think there is a role.
MIRI GITIK: There’s another question about the role of the immune system in neuropsychiatric and neurological disorders. Any thoughts about that?
FLORA VACCARINO: If I might say something, it’s one of the interactions among systems that are so difficult to investigate because, again, it’s a question of cause or consequence. Alterations of the immune system have been found in analysis, say postpartum analysis of the brain of most psychiatric disorders. And so, then you wonder, okay, is this a common response to just an unhealthy circuitry or something like that, or is this a consequence of something else versus, you know, being the actual force that drives the disease progression or even onset. So I don’t think we know very well, although there are certain, of course, comorbidities that we know of. But in general there is probably a larger effect of the immune system on CNS, even development of the CNS is being argued, right, that we don’t really understand yet. And it’s probably important when [unintelligible] interactions. I’m sorry, I can’t give a very meaningful answer because it’s just a very intricated topic I think.
PATRICK SULLIVAN: Yeah. And I agree with you. It’s — the immune system is tricky. Frances Collins [spelled phonetically] had a little quip on this where he said what — that back in the day when you were working for Mendelian disorder and you thought you had the sequence nailed and you popped it into the search thing to find what gene it was in, he said you would sit there and do a silent prayer that it wasn’t the immune system.
FLORA VACCARINO: [laughs]
PATRICK SULLIVAN: But I think, you know, for neurological disorders definitely. Multiple sclerosis certainly comes to mind as one aspect. There’s a strong theory for Alzheimer’s disease that has an immune basis as well. In psychiatric disorders, yes. There’s certainly — there’s a — there’s a number of papers that have looked at this. The largest and the hit — the largest hit for schizophrenia, for instance, is in the major histocompatibility complex, which is chockablock full of the core genes that actually mediate the function of the immune system. In addition, there are certain subclasses of brain cells which are of immunological origin which some people think actually have a lot to do with the way in which the brain gets tuned to fit the environment as well as possible. So yes, there’s certainly a lot going on, but boy is it complicated.
MIRI GITIK: Thank you. Think the questions are winding down. I don’t know if any one of the speakers would like to give some ending remarks? Maybe some thoughts [unintelligible] of early career investigators? Any advice you may have?
FRANCIS McMAHON: I’ll offer something along those lines. And there’s a story that may be apocryphal that when the great physicist, Max Planck, went to his professor in college and said that he wanted to study physics the professor told him, “Oh, don’t do that. All of the important stuff’s already been discovered.” And this was around 1900. So my main message to young investigators who are considering a research career in psychiatric genetics is not to be discouraged that we found everything. As Pat said, we’re really in a lot of ways still at the tip of the iceberg. And as someone who’s been in the field now for, gosh, close to 40 years, I can say that when I started out in the field we didn’t have a lot of the tools to work with. We didn’t have large samples to study. And most importantly, we didn’t really have a fundamental roadmap of how genes were arranged, how they were regulated, and how they were expressed in the brain. We are now beginning to have all those things. And so, I think it’s — there could hardly be a better time to jump into this field. And the tools are stronger than they’ve ever been. We have the fundamental roadmaps we need and the genome sequence, and the way in which genes are expressed in the brain. And we still have a lot to discover, particularly when it comes to discoveries that will alter the way in which we provide psychiatric care. I think that’s really the next big challenge.
CHRISTA MARTIN: Yeah. And I would say collaborate, collaborate, collaborate [laughs]. It’s really important, not only from the standpoint of finding good colleagues with varied backgrounds, because I think that adds to the depth of research. I’ve learned a lot from the neurodevelopmental pediatricians I work with, from our speech and language pathologists who allow us to ask questions in a — in a way that as a geneticist I might not have thought out on my own. And then also collaborate to share data. You heard Pat say how big datasets are needed to find some of these clues. And the more that we can — you know, reappropriate a line and share data I think the better. So it used to be — I feel like science, not that we’ve lost our competitive edge, but it used to be very much that people worked in silos. And I think that was to the deficit of the patients and serving the patients. And now these large consortium that encourage us to work together toward helping patients I think is a — is a huge movement, where we all can still have, you know, our unique interests but — and it gives us room to grow, but it just allows us to get there faster I think.
PATRICK SULLIVAN: My take on this would be that we’re finally to the point of having at least technologies that can begin to approximate and dissect the complexity of the nervous system. You know, put in another way when I was in medical school I thought about going into cardiology but, you know, in the end the heart’s just a pump, right? And you can touch it, you can grab it, you can get pieces of it whenever you want practically. This is — psychiatry has all the great questions. And we’re finally having the tools that actually can allow us to ask and answer some of the important questions that we just had philosophy about in the past.
FLORA VACCARINO: Yeah. I agree. I agree. And for me one of the most encouraging things, not being trained as a geneticist but mostly as a neuropsychiatrist, a neurobiologist, and a physician is the fact that I think genetics is broken — is, you know, limits, is now — is now embracing other things, right? It’s not just — first of all, it’s not just a the genes. It’s also the non-genes. It’s also that back part of the genome that we ignored so long. And also neuroscience and biology. And I think geneticists today are much more open to that and don’t think that genes explain everything, right? But they have to integrate with other things to be — you know, the knowledge has to come together. And I think this is really exciting.
MIRI GITIK: All right. I would like to thank again all of our speakers. Thank you very much for the fascinating talks and spending time and answering everybody’s questions. Thank you very much for attending our celebration. And I would also like to thank Rosalee and Associates [spelled phonetically], the contractor that provided the technical support for this meeting. Thank you very much, everybody.